Merge release-20210628.0-8-g54b71221c (automated)

169 files changed, 941 insertions, 844 deletions

diff --git a/pkg/sentry/arch/arch_aarch64.go b/pkg/sentry/arch/arch_aarch64.go
index 08789f517..7def71fef 100644
--- a/pkg/sentry/arch/arch_aarch64.go
+++ b/pkg/sentry/arch/arch_aarch64.go
@@ -22,6 +22,7 @@ import (
 
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/cpuid"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/log"
 	"gvisor.dev/gvisor/pkg/sentry/arch/fpu"
 	rpb "gvisor.dev/gvisor/pkg/sentry/arch/registers_go_proto"
@@ -237,7 +238,7 @@ func (s *State) PtraceGetRegSet(regset uintptr, dst io.Writer, maxlen int) (int,
 		}
 		return s.PtraceGetRegs(dst)
 	default:
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 }
 
@@ -250,7 +251,7 @@ func (s *State) PtraceSetRegSet(regset uintptr, src io.Reader, maxlen int) (int,
 		}
 		return s.PtraceSetRegs(src)
 	default:
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 }
 
diff --git a/pkg/sentry/arch/arch_x86.go b/pkg/sentry/arch/arch_x86.go
index e8e52d3a8..d13e12f8c 100644
--- a/pkg/sentry/arch/arch_x86.go
+++ b/pkg/sentry/arch/arch_x86.go
@@ -23,6 +23,7 @@ import (
 	"golang.org/x/sys/unix"
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/cpuid"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/log"
 	"gvisor.dev/gvisor/pkg/sentry/arch/fpu"
 	rpb "gvisor.dev/gvisor/pkg/sentry/arch/registers_go_proto"
@@ -361,7 +362,7 @@ func (s *State) PtraceGetRegSet(regset uintptr, dst io.Writer, maxlen int) (int,
 	case _NT_X86_XSTATE:
 		return s.fpState.PtraceGetXstateRegs(dst, maxlen, s.FeatureSet)
 	default:
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 }
 
@@ -378,7 +379,7 @@ func (s *State) PtraceSetRegSet(regset uintptr, src io.Reader, maxlen int) (int,
 	case _NT_X86_XSTATE:
 		return s.fpState.PtraceSetXstateRegs(src, maxlen, s.FeatureSet)
 	default:
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 }
 
diff --git a/pkg/sentry/devices/tundev/tundev.go b/pkg/sentry/devices/tundev/tundev.go
index a12eeb8e7..4ef91a600 100644
--- a/pkg/sentry/devices/tundev/tundev.go
+++ b/pkg/sentry/devices/tundev/tundev.go
@@ -18,6 +18,7 @@ package tundev
 import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
 	"gvisor.dev/gvisor/pkg/sentry/fsimpl/devtmpfs"
@@ -81,7 +82,7 @@ func (fd *tunFD) Ioctl(ctx context.Context, uio usermem.IO, args arch.SyscallArg
 		}
 		stack, ok := t.NetworkContext().(*netstack.Stack)
 		if !ok {
-			return 0, syserror.EINVAL
+			return 0, linuxerr.EINVAL
 		}
 
 		var req linux.IFReq
diff --git a/pkg/sentry/fs/copy_up.go b/pkg/sentry/fs/copy_up.go
index ae282d14e..a8591052c 100644
--- a/pkg/sentry/fs/copy_up.go
+++ b/pkg/sentry/fs/copy_up.go
@@ -162,7 +162,7 @@ func doCopyUp(ctx context.Context, d *Dirent) error {
 	// then try to take copyMu for writing here, we'd deadlock.
 	t := d.Inode.overlay.lower.StableAttr.Type
 	if t != RegularFile && t != Directory && t != Symlink {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 
 	// Wait to get exclusive access to the upper Inode.
diff --git a/pkg/sentry/fs/dev/net_tun.go b/pkg/sentry/fs/dev/net_tun.go
index 77e8d222a..5674978bd 100644
--- a/pkg/sentry/fs/dev/net_tun.go
+++ b/pkg/sentry/fs/dev/net_tun.go
@@ -17,6 +17,7 @@ package dev
 import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
 	"gvisor.dev/gvisor/pkg/sentry/fs"
@@ -102,7 +103,7 @@ func (n *netTunFileOperations) Ioctl(ctx context.Context, file *fs.File, io user
 		}
 		stack, ok := t.NetworkContext().(*netstack.Stack)
 		if !ok {
-			return 0, syserror.EINVAL
+			return 0, linuxerr.EINVAL
 		}
 
 		var req linux.IFReq
diff --git a/pkg/sentry/fs/dirent.go b/pkg/sentry/fs/dirent.go
index e45749be6..e21c9d78e 100644
--- a/pkg/sentry/fs/dirent.go
+++ b/pkg/sentry/fs/dirent.go
@@ -964,7 +964,7 @@ func (d *Dirent) mount(ctx context.Context, inode *Inode) (newChild *Dirent, err
 	//
 	// See Linux equivalent in fs/namespace.c:do_add_mount.
 	if IsSymlink(inode.StableAttr) {
-		return nil, syserror.EINVAL
+		return nil, linuxerr.EINVAL
 	}
 
 	// Dirent that'll replace d.
diff --git a/pkg/sentry/fs/file_overlay.go b/pkg/sentry/fs/file_overlay.go
index 696613f3a..7e2f107e0 100644
--- a/pkg/sentry/fs/file_overlay.go
+++ b/pkg/sentry/fs/file_overlay.go
@@ -18,6 +18,7 @@ import (
 	"io"
 
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/refs"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
 	"gvisor.dev/gvisor/pkg/sentry/memmap"
@@ -417,7 +418,7 @@ func (f *overlayFileOperations) FifoSize(ctx context.Context, overlayFile *File)
 	err = f.onTop(ctx, overlayFile, func(file *File, ops FileOperations) error {
 		sz, ok := ops.(FifoSizer)
 		if !ok {
-			return syserror.EINVAL
+			return linuxerr.EINVAL
 		}
 		rv, err = sz.FifoSize(ctx, file)
 		return err
@@ -432,11 +433,11 @@ func (f *overlayFileOperations) SetFifoSize(size int64) (rv int64, err error) {
 
 	if f.upper == nil {
 		// Named pipes cannot be copied up and changes to the lower are prohibited.
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 	sz, ok := f.upper.FileOperations.(FifoSizer)
 	if !ok {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 	return sz.SetFifoSize(size)
 }
diff --git a/pkg/sentry/fs/fsutil/file.go b/pkg/sentry/fs/fsutil/file.go
index dc9efa5df..c3525ba8e 100644
--- a/pkg/sentry/fs/fsutil/file.go
+++ b/pkg/sentry/fs/fsutil/file.go
@@ -18,6 +18,7 @@ import (
 	"io"
 
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
 	"gvisor.dev/gvisor/pkg/sentry/fs"
 	"gvisor.dev/gvisor/pkg/sentry/memmap"
@@ -63,12 +64,12 @@ func SeekWithDirCursor(ctx context.Context, file *fs.File, whence fs.SeekWhence,
 		switch inode.StableAttr.Type {
 		case fs.RegularFile, fs.SpecialFile, fs.BlockDevice:
 			if offset < 0 {
-				return current, syserror.EINVAL
+				return current, linuxerr.EINVAL
 			}
 			return offset, nil
 		case fs.Directory, fs.SpecialDirectory:
 			if offset != 0 {
-				return current, syserror.EINVAL
+				return current, linuxerr.EINVAL
 			}
 			// SEEK_SET to 0 moves the directory "cursor" to the beginning.
 			if dirCursor != nil {
@@ -76,22 +77,22 @@ func SeekWithDirCursor(ctx context.Context, file *fs.File, whence fs.SeekWhence,
 			}
 			return 0, nil
 		default:
-			return current, syserror.EINVAL
+			return current, linuxerr.EINVAL
 		}
 	case fs.SeekCurrent:
 		switch inode.StableAttr.Type {
 		case fs.RegularFile, fs.SpecialFile, fs.BlockDevice:
 			if current+offset < 0 {
-				return current, syserror.EINVAL
+				return current, linuxerr.EINVAL
 			}
 			return current + offset, nil
 		case fs.Directory, fs.SpecialDirectory:
 			if offset != 0 {
-				return current, syserror.EINVAL
+				return current, linuxerr.EINVAL
 			}
 			return current, nil
 		default:
-			return current, syserror.EINVAL
+			return current, linuxerr.EINVAL
 		}
 	case fs.SeekEnd:
 		switch inode.StableAttr.Type {
@@ -103,14 +104,14 @@ func SeekWithDirCursor(ctx context.Context, file *fs.File, whence fs.SeekWhence,
 			}
 			sz := uattr.Size
 			if sz+offset < 0 {
-				return current, syserror.EINVAL
+				return current, linuxerr.EINVAL
 			}
 			return sz + offset, nil
 		// FIXME(b/34778850): This is not universally correct.
 		// Remove SpecialDirectory.
 		case fs.SpecialDirectory:
 			if offset != 0 {
-				return current, syserror.EINVAL
+				return current, linuxerr.EINVAL
 			}
 			// SEEK_END to 0 moves the directory "cursor" to the end.
 			//
@@ -121,12 +122,12 @@ func SeekWithDirCursor(ctx context.Context, file *fs.File, whence fs.SeekWhence,
 			// futile (EOF will always be the result).
 			return fs.FileMaxOffset, nil
 		default:
-			return current, syserror.EINVAL
+			return current, linuxerr.EINVAL
 		}
 	}
 
 	// Not a valid seek request.
-	return current, syserror.EINVAL
+	return current, linuxerr.EINVAL
 }
 
 // FileGenericSeek implements fs.FileOperations.Seek for files that use a
@@ -152,7 +153,7 @@ type FileNoSeek struct{}
 
 // Seek implements fs.FileOperations.Seek.
 func (FileNoSeek) Seek(context.Context, *fs.File, fs.SeekWhence, int64) (int64, error) {
-	return 0, syserror.EINVAL
+	return 0, linuxerr.EINVAL
 }
 
 // FilePipeSeek implements fs.FileOperations.Seek and can be used for files
@@ -178,7 +179,7 @@ type FileNoFsync struct{}
 
 // Fsync implements fs.FileOperations.Fsync.
 func (FileNoFsync) Fsync(context.Context, *fs.File, int64, int64, fs.SyncType) error {
-	return syserror.EINVAL
+	return linuxerr.EINVAL
 }
 
 // FileNoopFsync implements fs.FileOperations.Fsync for files that don't need
@@ -345,7 +346,7 @@ func NewFileStaticContentReader(b []byte) FileStaticContentReader {
 // Read implements fs.FileOperations.Read.
 func (scr *FileStaticContentReader) Read(ctx context.Context, _ *fs.File, dst usermem.IOSequence, offset int64) (int64, error) {
 	if offset < 0 {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 	if offset >= int64(len(scr.content)) {
 		return 0, nil
@@ -367,7 +368,7 @@ type FileNoRead struct{}
 
 // Read implements fs.FileOperations.Read.
 func (FileNoRead) Read(context.Context, *fs.File, usermem.IOSequence, int64) (int64, error) {
-	return 0, syserror.EINVAL
+	return 0, linuxerr.EINVAL
 }
 
 // FileNoWrite implements fs.FileOperations.Write to return EINVAL.
@@ -375,7 +376,7 @@ type FileNoWrite struct{}
 
 // Write implements fs.FileOperations.Write.
 func (FileNoWrite) Write(context.Context, *fs.File, usermem.IOSequence, int64) (int64, error) {
-	return 0, syserror.EINVAL
+	return 0, linuxerr.EINVAL
 }
 
 // FileNoopRead implement fs.FileOperations.Read as a noop.
diff --git a/pkg/sentry/fs/fsutil/inode.go b/pkg/sentry/fs/fsutil/inode.go
index 85e7e35db..bda07275d 100644
--- a/pkg/sentry/fs/fsutil/inode.go
+++ b/pkg/sentry/fs/fsutil/inode.go
@@ -17,6 +17,7 @@ package fsutil
 import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/sentry/fs"
 	ktime "gvisor.dev/gvisor/pkg/sentry/kernel/time"
 	"gvisor.dev/gvisor/pkg/sentry/memmap"
@@ -376,7 +377,7 @@ func (InodeNotDirectory) RemoveDirectory(context.Context, *fs.Inode, string) err
 
 // Rename implements fs.FileOperations.Rename.
 func (InodeNotDirectory) Rename(context.Context, *fs.Inode, *fs.Inode, string, *fs.Inode, string, bool) error {
-	return syserror.EINVAL
+	return linuxerr.EINVAL
 }
 
 // InodeNotSocket can be used by Inodes that are not sockets.
@@ -392,7 +393,7 @@ type InodeNotTruncatable struct{}
 
 // Truncate implements fs.InodeOperations.Truncate.
 func (InodeNotTruncatable) Truncate(context.Context, *fs.Inode, int64) error {
-	return syserror.EINVAL
+	return linuxerr.EINVAL
 }
 
 // InodeIsDirTruncate implements fs.InodeOperations.Truncate for directories.
@@ -416,7 +417,7 @@ type InodeNotRenameable struct{}
 
 // Rename implements fs.InodeOperations.Rename.
 func (InodeNotRenameable) Rename(context.Context, *fs.Inode, *fs.Inode, string, *fs.Inode, string, bool) error {
-	return syserror.EINVAL
+	return linuxerr.EINVAL
 }
 
 // InodeNotOpenable can be used by Inodes that cannot be opened.
diff --git a/pkg/sentry/fs/host/tty.go b/pkg/sentry/fs/host/tty.go
index 77613bfd5..2ff520100 100644
--- a/pkg/sentry/fs/host/tty.go
+++ b/pkg/sentry/fs/host/tty.go
@@ -212,7 +212,7 @@ func (t *TTYFileOperations) Ioctl(ctx context.Context, _ *fs.File, io usermem.IO
 
 		// pgID must be non-negative.
 		if pgID < 0 {
-			return 0, syserror.EINVAL
+			return 0, linuxerr.EINVAL
 		}
 
 		// Process group with pgID must exist in this PID namespace.
diff --git a/pkg/sentry/fs/inotify.go b/pkg/sentry/fs/inotify.go
index 1b83643db..4e07043c7 100644
--- a/pkg/sentry/fs/inotify.go
+++ b/pkg/sentry/fs/inotify.go
@@ -20,6 +20,7 @@ import (
 
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
 	"gvisor.dev/gvisor/pkg/sentry/memmap"
@@ -132,7 +133,7 @@ func (*Inotify) Write(context.Context, *File, usermem.IOSequence, int64) (int64,
 // Read implements FileOperations.Read.
 func (i *Inotify) Read(ctx context.Context, _ *File, dst usermem.IOSequence, _ int64) (int64, error) {
 	if dst.NumBytes() < inotifyEventBaseSize {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 
 	i.evMu.Lock()
@@ -156,7 +157,7 @@ func (i *Inotify) Read(ctx context.Context, _ *File, dst usermem.IOSequence, _ i
 				// write some events out.
 				return writeLen, nil
 			}
-			return 0, syserror.EINVAL
+			return 0, linuxerr.EINVAL
 		}
 
 		// Linux always dequeues an available event as long as there's enough
@@ -183,7 +184,7 @@ func (*Inotify) WriteTo(context.Context, *File, io.Writer, int64, bool) (int64,
 
 // Fsync implements FileOperations.Fsync.
 func (*Inotify) Fsync(context.Context, *File, int64, int64, SyncType) error {
-	return syserror.EINVAL
+	return linuxerr.EINVAL
 }
 
 // ReadFrom implements FileOperations.ReadFrom.
@@ -329,7 +330,7 @@ func (i *Inotify) RmWatch(ctx context.Context, wd int32) error {
 	watch, ok := i.watches[wd]
 	if !ok {
 		i.mu.Unlock()
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 
 	// Remove the watch from this instance.
diff --git a/pkg/sentry/fs/mock.go b/pkg/sentry/fs/mock.go
index 1d6ea5736..2a54c1242 100644
--- a/pkg/sentry/fs/mock.go
+++ b/pkg/sentry/fs/mock.go
@@ -16,6 +16,7 @@ package fs
 
 import (
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/syserror"
 )
 
@@ -109,7 +110,7 @@ func (n *MockInodeOperations) SetPermissions(context.Context, *Inode, FilePermis
 
 // SetOwner implements fs.InodeOperations.SetOwner.
 func (*MockInodeOperations) SetOwner(context.Context, *Inode, FileOwner) error {
-	return syserror.EINVAL
+	return linuxerr.EINVAL
 }
 
 // SetTimestamps implements fs.InodeOperations.SetTimestamps.
diff --git a/pkg/sentry/fs/mounts.go b/pkg/sentry/fs/mounts.go
index 243098a09..340441974 100644
--- a/pkg/sentry/fs/mounts.go
+++ b/pkg/sentry/fs/mounts.go
@@ -20,6 +20,7 @@ import (
 
 	"golang.org/x/sys/unix"
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/refs"
 	"gvisor.dev/gvisor/pkg/sentry/kernel/auth"
 	"gvisor.dev/gvisor/pkg/sync"
@@ -357,7 +358,7 @@ func (mns *MountNamespace) Unmount(ctx context.Context, node *Dirent, detachOnly
 		orig, ok := mns.mounts[node]
 		if !ok {
 			// node is not a mount point.
-			return syserror.EINVAL
+			return linuxerr.EINVAL
 		}
 
 		if orig.previous == nil {
diff --git a/pkg/sentry/fs/overlay.go b/pkg/sentry/fs/overlay.go
index f96f5a3e5..7e72e47b5 100644
--- a/pkg/sentry/fs/overlay.go
+++ b/pkg/sentry/fs/overlay.go
@@ -19,11 +19,11 @@ import (
 	"strings"
 
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/log"
 	"gvisor.dev/gvisor/pkg/sentry/memmap"
 	"gvisor.dev/gvisor/pkg/sync"
-	"gvisor.dev/gvisor/pkg/syserror"
 )
 
 // The virtual filesystem implements an overlay configuration. For a high-level
@@ -218,7 +218,7 @@ func newOverlayEntry(ctx context.Context, upper *Inode, lower *Inode, lowerExist
 			// We don't support copying up from character devices,
 			// named pipes, or anything weird (like proc files).
 			log.Warningf("%s not supported in lower filesytem", lower.StableAttr.Type)
-			return nil, syserror.EINVAL
+			return nil, linuxerr.EINVAL
 		}
 	}
 	return &overlayEntry{
diff --git a/pkg/sentry/fs/proc/exec_args.go b/pkg/sentry/fs/proc/exec_args.go
index 24426b225..379429ab2 100644
--- a/pkg/sentry/fs/proc/exec_args.go
+++ b/pkg/sentry/fs/proc/exec_args.go
@@ -21,11 +21,11 @@ import (
 
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/sentry/fs"
 	"gvisor.dev/gvisor/pkg/sentry/fs/fsutil"
 	"gvisor.dev/gvisor/pkg/sentry/kernel"
-	"gvisor.dev/gvisor/pkg/syserror"
 	"gvisor.dev/gvisor/pkg/usermem"
 	"gvisor.dev/gvisor/pkg/waiter"
 )
@@ -104,7 +104,7 @@ var _ fs.FileOperations = (*execArgFile)(nil)
 // Read reads the exec arg from the process's address space..
 func (f *execArgFile) Read(ctx context.Context, _ *fs.File, dst usermem.IOSequence, offset int64) (int64, error) {
 	if offset < 0 {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 
 	m, err := getTaskMM(f.t)
diff --git a/pkg/sentry/fs/proc/proc.go b/pkg/sentry/fs/proc/proc.go
index 2f2a9f920..546b57287 100644
--- a/pkg/sentry/fs/proc/proc.go
+++ b/pkg/sentry/fs/proc/proc.go
@@ -21,6 +21,7 @@ import (
 	"strconv"
 
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/sentry/fs"
 	"gvisor.dev/gvisor/pkg/sentry/fs/fsutil"
 	"gvisor.dev/gvisor/pkg/sentry/fs/proc/device"
@@ -130,7 +131,7 @@ func (s *self) Readlink(ctx context.Context, inode *fs.Inode) (string, error) {
 	}
 
 	// Who is reading this link?
-	return "", syserror.EINVAL
+	return "", linuxerr.EINVAL
 }
 
 // threadSelf is more magical than "self" link.
@@ -154,7 +155,7 @@ func (s *threadSelf) Readlink(ctx context.Context, inode *fs.Inode) (string, err
 	}
 
 	// Who is reading this link?
-	return "", syserror.EINVAL
+	return "", linuxerr.EINVAL
 }
 
 // Lookup loads an Inode at name into a Dirent.
diff --git a/pkg/sentry/fs/proc/sys_net.go b/pkg/sentry/fs/proc/sys_net.go
index 4893af56b..71f37d582 100644
--- a/pkg/sentry/fs/proc/sys_net.go
+++ b/pkg/sentry/fs/proc/sys_net.go
@@ -21,6 +21,7 @@ import (
 
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/sentry/fs"
 	"gvisor.dev/gvisor/pkg/sentry/fs/fsutil"
@@ -28,7 +29,6 @@ import (
 	"gvisor.dev/gvisor/pkg/sentry/fs/ramfs"
 	"gvisor.dev/gvisor/pkg/sentry/inet"
 	"gvisor.dev/gvisor/pkg/sync"
-	"gvisor.dev/gvisor/pkg/syserror"
 	"gvisor.dev/gvisor/pkg/tcpip/network/ipv4"
 	"gvisor.dev/gvisor/pkg/usermem"
 	"gvisor.dev/gvisor/pkg/waiter"
@@ -592,7 +592,7 @@ func (pf *portRangeFile) Write(ctx context.Context, _ *fs.File, src usermem.IOSe
 
 	// Port numbers must be uint16s.
 	if ports[0] < 0 || ports[1] < 0 || ports[0] > math.MaxUint16 || ports[1] > math.MaxUint16 {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 
 	if err := pf.inode.stack.SetPortRange(uint16(ports[0]), uint16(ports[1])); err != nil {
diff --git a/pkg/sentry/fs/proc/task.go b/pkg/sentry/fs/proc/task.go
index ae5ed25f9..7ece1377a 100644
--- a/pkg/sentry/fs/proc/task.go
+++ b/pkg/sentry/fs/proc/task.go
@@ -23,6 +23,7 @@ import (
 
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/sentry/fs"
 	"gvisor.dev/gvisor/pkg/sentry/fs/fsutil"
@@ -867,7 +868,7 @@ var _ fs.FileOperations = (*commFile)(nil)
 // Read implements fs.FileOperations.Read.
 func (f *commFile) Read(ctx context.Context, _ *fs.File, dst usermem.IOSequence, offset int64) (int64, error) {
 	if offset < 0 {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 
 	buf := []byte(f.t.Name() + "\n")
@@ -922,7 +923,7 @@ type auxvecFile struct {
 // Read implements fs.FileOperations.Read.
 func (f *auxvecFile) Read(ctx context.Context, _ *fs.File, dst usermem.IOSequence, offset int64) (int64, error) {
 	if offset < 0 {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 
 	m, err := getTaskMM(f.t)
diff --git a/pkg/sentry/fs/proc/uid_gid_map.go b/pkg/sentry/fs/proc/uid_gid_map.go
index 30d5ad4cf..fcdc1e7bd 100644
--- a/pkg/sentry/fs/proc/uid_gid_map.go
+++ b/pkg/sentry/fs/proc/uid_gid_map.go
@@ -21,12 +21,12 @@ import (
 
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/sentry/fs"
 	"gvisor.dev/gvisor/pkg/sentry/fs/fsutil"
 	"gvisor.dev/gvisor/pkg/sentry/kernel"
 	"gvisor.dev/gvisor/pkg/sentry/kernel/auth"
-	"gvisor.dev/gvisor/pkg/syserror"
 	"gvisor.dev/gvisor/pkg/usermem"
 	"gvisor.dev/gvisor/pkg/waiter"
 )
@@ -108,7 +108,7 @@ const maxIDMapLines = 5
 // Read implements fs.FileOperations.Read.
 func (imfo *idMapFileOperations) Read(ctx context.Context, file *fs.File, dst usermem.IOSequence, offset int64) (int64, error) {
 	if offset < 0 {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 	var entries []auth.IDMapEntry
 	if imfo.iops.gids {
@@ -134,7 +134,7 @@ func (imfo *idMapFileOperations) Write(ctx context.Context, file *fs.File, src u
 	// the file ..." - user_namespaces(7)
 	srclen := src.NumBytes()
 	if srclen >= hostarch.PageSize || offset != 0 {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 	b := make([]byte, srclen)
 	if _, err := src.CopyIn(ctx, b); err != nil {
@@ -154,7 +154,7 @@ func (imfo *idMapFileOperations) Write(ctx context.Context, file *fs.File, src u
 	}
 	lines := bytes.SplitN(b, []byte("\n"), maxIDMapLines+1)
 	if len(lines) > maxIDMapLines {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 
 	entries := make([]auth.IDMapEntry, len(lines))
@@ -162,7 +162,7 @@ func (imfo *idMapFileOperations) Write(ctx context.Context, file *fs.File, src u
 		var e auth.IDMapEntry
 		_, err := fmt.Sscan(string(l), &e.FirstID, &e.FirstParentID, &e.Length)
 		if err != nil {
-			return 0, syserror.EINVAL
+			return 0, linuxerr.EINVAL
 		}
 		entries[i] = e
 	}
diff --git a/pkg/sentry/fs/proc/uptime.go b/pkg/sentry/fs/proc/uptime.go
index c0f6fb802..ac896f963 100644
--- a/pkg/sentry/fs/proc/uptime.go
+++ b/pkg/sentry/fs/proc/uptime.go
@@ -20,10 +20,10 @@ import (
 
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/sentry/fs"
 	"gvisor.dev/gvisor/pkg/sentry/fs/fsutil"
 	ktime "gvisor.dev/gvisor/pkg/sentry/kernel/time"
-	"gvisor.dev/gvisor/pkg/syserror"
 	"gvisor.dev/gvisor/pkg/usermem"
 	"gvisor.dev/gvisor/pkg/waiter"
 )
@@ -74,7 +74,7 @@ type uptimeFile struct {
 // Read implements fs.FileOperations.Read.
 func (f *uptimeFile) Read(ctx context.Context, _ *fs.File, dst usermem.IOSequence, offset int64) (int64, error) {
 	if offset < 0 {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 
 	now := ktime.NowFromContext(ctx)
diff --git a/pkg/sentry/fs/timerfd/timerfd.go b/pkg/sentry/fs/timerfd/timerfd.go
index c8ebe256c..093a14c1f 100644
--- a/pkg/sentry/fs/timerfd/timerfd.go
+++ b/pkg/sentry/fs/timerfd/timerfd.go
@@ -20,6 +20,7 @@ import (
 	"sync/atomic"
 
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/sentry/fs"
 	"gvisor.dev/gvisor/pkg/sentry/fs/anon"
@@ -121,7 +122,7 @@ func (t *TimerOperations) EventUnregister(e *waiter.Entry) {
 func (t *TimerOperations) Read(ctx context.Context, file *fs.File, dst usermem.IOSequence, offset int64) (int64, error) {
 	const sizeofUint64 = 8
 	if dst.NumBytes() < sizeofUint64 {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 	if val := atomic.SwapUint64(&t.val, 0); val != 0 {
 		var buf [sizeofUint64]byte
@@ -138,7 +139,7 @@ func (t *TimerOperations) Read(ctx context.Context, file *fs.File, dst usermem.I
 
 // Write implements fs.FileOperations.Write.
 func (t *TimerOperations) Write(context.Context, *fs.File, usermem.IOSequence, int64) (int64, error) {
-	return 0, syserror.EINVAL
+	return 0, linuxerr.EINVAL
 }
 
 // Notify implements ktime.TimerListener.Notify.
diff --git a/pkg/sentry/fs/tmpfs/inode_file.go b/pkg/sentry/fs/tmpfs/inode_file.go
index 7faa822f0..ce6be6386 100644
--- a/pkg/sentry/fs/tmpfs/inode_file.go
+++ b/pkg/sentry/fs/tmpfs/inode_file.go
@@ -21,6 +21,7 @@ import (
 
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/safemem"
 	"gvisor.dev/gvisor/pkg/sentry/fs"
@@ -455,7 +456,7 @@ func (rw *fileReadWriter) WriteFromBlocks(srcs safemem.BlockSeq) (uint64, error)
 	end := fs.WriteEndOffset(rw.offset, int64(srcs.NumBytes()))
 	if end == math.MaxInt64 {
 		// Overflow.
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 
 	// Check if seals prevent either file growth or all writes.
@@ -655,7 +656,7 @@ func GetSeals(inode *fs.Inode) (uint32, error) {
 		return f.seals, nil
 	}
 	// Not a memfd inode.
-	return 0, syserror.EINVAL
+	return 0, linuxerr.EINVAL
 }
 
 // AddSeals adds new file seals to a memfd inode.
@@ -683,5 +684,5 @@ func AddSeals(inode *fs.Inode, val uint32) error {
 		return nil
 	}
 	// Not a memfd inode.
-	return syserror.EINVAL
+	return linuxerr.EINVAL
 }
diff --git a/pkg/sentry/fs/tty/fs.go b/pkg/sentry/fs/tty/fs.go
index 13f4901db..0e5916380 100644
--- a/pkg/sentry/fs/tty/fs.go
+++ b/pkg/sentry/fs/tty/fs.go
@@ -16,9 +16,9 @@ package tty
 
 import (
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/sentry/device"
 	"gvisor.dev/gvisor/pkg/sentry/fs"
-	"gvisor.dev/gvisor/pkg/syserror"
 )
 
 // ptsDevice is the pseudo-filesystem device.
@@ -64,7 +64,7 @@ func (f *filesystem) Mount(ctx context.Context, device string, flags fs.MountSou
 
 	// No options are supported.
 	if data != "" {
-		return nil, syserror.EINVAL
+		return nil, linuxerr.EINVAL
 	}
 
 	return newDir(ctx, fs.NewMountSource(ctx, &superOperations{}, f, flags)), nil
diff --git a/pkg/sentry/fsimpl/cgroupfs/base.go b/pkg/sentry/fsimpl/cgroupfs/base.go
index fe9871bdd..4290ffe0d 100644
--- a/pkg/sentry/fsimpl/cgroupfs/base.go
+++ b/pkg/sentry/fsimpl/cgroupfs/base.go
@@ -23,10 +23,10 @@ import (
 
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/sentry/fsimpl/kernfs"
 	"gvisor.dev/gvisor/pkg/sentry/kernel"
 	"gvisor.dev/gvisor/pkg/sentry/kernel/auth"
-	"gvisor.dev/gvisor/pkg/syserror"
 	"gvisor.dev/gvisor/pkg/usermem"
 )
 
@@ -253,7 +253,7 @@ func parseInt64FromString(ctx context.Context, src usermem.IOSequence, offset in
 		// Note: This also handles zero-len writes if offset is beyond the end
 		// of src, or src is empty.
 		ctx.Warningf("cgroupfs.parseInt64FromString: failed to parse %q: %v", string(buf), err)
-		return 0, int64(n), syserror.EINVAL
+		return 0, int64(n), linuxerr.EINVAL
 	}
 
 	return val, int64(n), nil
diff --git a/pkg/sentry/fsimpl/cgroupfs/cgroupfs.go b/pkg/sentry/fsimpl/cgroupfs/cgroupfs.go
index 05d7eb4ce..b5883cbd2 100644
--- a/pkg/sentry/fsimpl/cgroupfs/cgroupfs.go
+++ b/pkg/sentry/fsimpl/cgroupfs/cgroupfs.go
@@ -62,6 +62,7 @@ import (
 
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/sentry/fsimpl/kernfs"
 	"gvisor.dev/gvisor/pkg/sentry/kernel"
 	"gvisor.dev/gvisor/pkg/sentry/kernel/auth"
@@ -167,7 +168,7 @@ func (fsType FilesystemType) GetFilesystem(ctx context.Context, vfsObj *vfs.Virt
 		maxCachedDentries, err = strconv.ParseUint(str, 10, 64)
 		if err != nil {
 			ctx.Warningf("sys.FilesystemType.GetFilesystem: invalid dentry cache limit: dentry_cache_limit=%s", str)
-			return nil, nil, syserror.EINVAL
+			return nil, nil, linuxerr.EINVAL
 		}
 	}
 
@@ -195,7 +196,7 @@ func (fsType FilesystemType) GetFilesystem(ctx context.Context, vfsObj *vfs.Virt
 	if _, ok := mopts["all"]; ok {
 		if len(wantControllers) > 0 {
 			ctx.Debugf("cgroupfs.FilesystemType.GetFilesystem: other controllers specified with all: %v", wantControllers)
-			return nil, nil, syserror.EINVAL
+			return nil, nil, linuxerr.EINVAL
 		}
 
 		delete(mopts, "all")
@@ -209,7 +210,7 @@ func (fsType FilesystemType) GetFilesystem(ctx context.Context, vfsObj *vfs.Virt
 
 	if len(mopts) != 0 {
 		ctx.Debugf("cgroupfs.FilesystemType.GetFilesystem: unknown options: %v", mopts)
-		return nil, nil, syserror.EINVAL
+		return nil, nil, linuxerr.EINVAL
 	}
 
 	k := kernel.KernelFromContext(ctx)
diff --git a/pkg/sentry/fsimpl/devpts/devpts.go b/pkg/sentry/fsimpl/devpts/devpts.go
index e75954105..7a488e9fd 100644
--- a/pkg/sentry/fsimpl/devpts/devpts.go
+++ b/pkg/sentry/fsimpl/devpts/devpts.go
@@ -25,6 +25,7 @@ import (
 
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/sentry/fsimpl/kernfs"
 	"gvisor.dev/gvisor/pkg/sentry/kernel/auth"
 	"gvisor.dev/gvisor/pkg/sentry/vfs"
@@ -56,7 +57,7 @@ func (*FilesystemType) Name() string {
 func (fstype *FilesystemType) GetFilesystem(ctx context.Context, vfsObj *vfs.VirtualFilesystem, creds *auth.Credentials, source string, opts vfs.GetFilesystemOptions) (*vfs.Filesystem, *vfs.Dentry, error) {
 	// No data allowed.
 	if opts.Data != "" {
-		return nil, nil, syserror.EINVAL
+		return nil, nil, linuxerr.EINVAL
 	}
 
 	fstype.initOnce.Do(func() {
diff --git a/pkg/sentry/fsimpl/devpts/master.go b/pkg/sentry/fsimpl/devpts/master.go
index 93c031c89..1374fd3be 100644
--- a/pkg/sentry/fsimpl/devpts/master.go
+++ b/pkg/sentry/fsimpl/devpts/master.go
@@ -17,6 +17,7 @@ package devpts
 import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/marshal/primitive"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
 	"gvisor.dev/gvisor/pkg/sentry/fsimpl/kernfs"
@@ -80,7 +81,7 @@ func (mi *masterInode) Stat(ctx context.Context, vfsfs *vfs.Filesystem, opts vfs
 // SetStat implements kernfs.Inode.SetStat
 func (mi *masterInode) SetStat(ctx context.Context, vfsfs *vfs.Filesystem, creds *auth.Credentials, opts vfs.SetStatOptions) error {
 	if opts.Stat.Mask&linux.STATX_SIZE != 0 {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	return mi.InodeAttrs.SetStat(ctx, vfsfs, creds, opts)
 }
diff --git a/pkg/sentry/fsimpl/devpts/replica.go b/pkg/sentry/fsimpl/devpts/replica.go
index 96d2054cb..81572b991 100644
--- a/pkg/sentry/fsimpl/devpts/replica.go
+++ b/pkg/sentry/fsimpl/devpts/replica.go
@@ -17,6 +17,7 @@ package devpts
 import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/marshal/primitive"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
 	"gvisor.dev/gvisor/pkg/sentry/fsimpl/kernfs"
@@ -92,7 +93,7 @@ func (ri *replicaInode) Stat(ctx context.Context, vfsfs *vfs.Filesystem, opts vf
 // SetStat implements kernfs.Inode.SetStat
 func (ri *replicaInode) SetStat(ctx context.Context, vfsfs *vfs.Filesystem, creds *auth.Credentials, opts vfs.SetStatOptions) error {
 	if opts.Stat.Mask&linux.STATX_SIZE != 0 {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	return ri.InodeAttrs.SetStat(ctx, vfsfs, creds, opts)
 }
diff --git a/pkg/sentry/fsimpl/fuse/dev.go b/pkg/sentry/fsimpl/fuse/dev.go
index 5d2bae14e..0d0eed543 100644
--- a/pkg/sentry/fsimpl/fuse/dev.go
+++ b/pkg/sentry/fsimpl/fuse/dev.go
@@ -18,6 +18,7 @@ import (
 	"golang.org/x/sys/unix"
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/sentry/kernel"
 	"gvisor.dev/gvisor/pkg/sentry/kernel/auth"
 	"gvisor.dev/gvisor/pkg/sentry/vfs"
@@ -149,7 +150,7 @@ func (fd *DeviceFD) Read(ctx context.Context, dst usermem.IOSequence, opts vfs.R
 
 	// If the read buffer is too small, error out.
 	if dst.NumBytes() < int64(minBuffSize) {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 
 	fd.mu.Lock()
@@ -293,7 +294,7 @@ func (fd *DeviceFD) writeLocked(ctx context.Context, src usermem.IOSequence, opt
 
 		// Assert that the header isn't read into the writeBuf yet.
 		if fd.writeCursor >= hdrLen {
-			return 0, syserror.EINVAL
+			return 0, linuxerr.EINVAL
 		}
 
 		// We don't have the full common response header yet.
@@ -322,7 +323,7 @@ func (fd *DeviceFD) writeLocked(ctx context.Context, src usermem.IOSequence, opt
 			if !ok {
 				// Server sent us a response for a request we never sent,
 				// or for which we already received a reply (e.g. aborted), an unlikely event.
-				return 0, syserror.EINVAL
+				return 0, linuxerr.EINVAL
 			}
 
 			delete(fd.completions, hdr.Unique)
@@ -434,7 +435,7 @@ func (fd *DeviceFD) sendError(ctx context.Context, errno int32, unique linux.FUS
 	if !ok {
 		// A response for a request we never sent,
 		// or for which we already received a reply (e.g. aborted).
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	delete(fd.completions, respHdr.Unique)
 
diff --git a/pkg/sentry/fsimpl/fuse/fusefs.go b/pkg/sentry/fsimpl/fuse/fusefs.go
index 47794810c..be5bcd6af 100644
--- a/pkg/sentry/fsimpl/fuse/fusefs.go
+++ b/pkg/sentry/fsimpl/fuse/fusefs.go
@@ -122,30 +122,30 @@ func (fsType FilesystemType) GetFilesystem(ctx context.Context, vfsObj *vfs.Virt
 	deviceDescriptorStr, ok := mopts["fd"]
 	if !ok {
 		ctx.Warningf("fusefs.FilesystemType.GetFilesystem: mandatory mount option fd missing")
-		return nil, nil, syserror.EINVAL
+		return nil, nil, linuxerr.EINVAL
 	}
 	delete(mopts, "fd")
 
 	deviceDescriptor, err := strconv.ParseInt(deviceDescriptorStr, 10 /* base */, 32 /* bitSize */)
 	if err != nil {
 		ctx.Debugf("fusefs.FilesystemType.GetFilesystem: invalid fd: %q (%v)", deviceDescriptorStr, err)
-		return nil, nil, syserror.EINVAL
+		return nil, nil, linuxerr.EINVAL
 	}
 
 	kernelTask := kernel.TaskFromContext(ctx)
 	if kernelTask == nil {
 		log.Warningf("%s.GetFilesystem: couldn't get kernel task from context", fsType.Name())
-		return nil, nil, syserror.EINVAL
+		return nil, nil, linuxerr.EINVAL
 	}
 	fuseFDGeneric := kernelTask.GetFileVFS2(int32(deviceDescriptor))
 	if fuseFDGeneric == nil {
-		return nil, nil, syserror.EINVAL
+		return nil, nil, linuxerr.EINVAL
 	}
 	defer fuseFDGeneric.DecRef(ctx)
 	fuseFD, ok := fuseFDGeneric.Impl().(*DeviceFD)
 	if !ok {
 		log.Warningf("%s.GetFilesystem: device FD is %T, not a FUSE device", fsType.Name, fuseFDGeneric)
-		return nil, nil, syserror.EINVAL
+		return nil, nil, linuxerr.EINVAL
 	}
 
 	// Parse and set all the other supported FUSE mount options.
@@ -155,17 +155,17 @@ func (fsType FilesystemType) GetFilesystem(ctx context.Context, vfsObj *vfs.Virt
 		uid, err := strconv.ParseUint(uidStr, 10, 32)
 		if err != nil {
 			log.Warningf("%s.GetFilesystem: invalid user_id: user_id=%s", fsType.Name(), uidStr)
-			return nil, nil, syserror.EINVAL
+			return nil, nil, linuxerr.EINVAL
 		}
 		kuid := creds.UserNamespace.MapToKUID(auth.UID(uid))
 		if !kuid.Ok() {
 			ctx.Warningf("fusefs.FilesystemType.GetFilesystem: unmapped uid: %d", uid)
-			return nil, nil, syserror.EINVAL
+			return nil, nil, linuxerr.EINVAL
 		}
 		fsopts.uid = kuid
 	} else {
 		ctx.Warningf("fusefs.FilesystemType.GetFilesystem: mandatory mount option user_id missing")
-		return nil, nil, syserror.EINVAL
+		return nil, nil, linuxerr.EINVAL
 	}
 
 	if gidStr, ok := mopts["group_id"]; ok {
@@ -173,17 +173,17 @@ func (fsType FilesystemType) GetFilesystem(ctx context.Context, vfsObj *vfs.Virt
 		gid, err := strconv.ParseUint(gidStr, 10, 32)
 		if err != nil {
 			log.Warningf("%s.GetFilesystem: invalid group_id: group_id=%s", fsType.Name(), gidStr)
-			return nil, nil, syserror.EINVAL
+			return nil, nil, linuxerr.EINVAL
 		}
 		kgid := creds.UserNamespace.MapToKGID(auth.GID(gid))
 		if !kgid.Ok() {
 			ctx.Warningf("fusefs.FilesystemType.GetFilesystem: unmapped gid: %d", gid)
-			return nil, nil, syserror.EINVAL
+			return nil, nil, linuxerr.EINVAL
 		}
 		fsopts.gid = kgid
 	} else {
 		ctx.Warningf("fusefs.FilesystemType.GetFilesystem: mandatory mount option group_id missing")
-		return nil, nil, syserror.EINVAL
+		return nil, nil, linuxerr.EINVAL
 	}
 
 	if modeStr, ok := mopts["rootmode"]; ok {
@@ -191,12 +191,12 @@ func (fsType FilesystemType) GetFilesystem(ctx context.Context, vfsObj *vfs.Virt
 		mode, err := strconv.ParseUint(modeStr, 8, 32)
 		if err != nil {
 			log.Warningf("%s.GetFilesystem: invalid mode: %q", fsType.Name(), modeStr)
-			return nil, nil, syserror.EINVAL
+			return nil, nil, linuxerr.EINVAL
 		}
 		fsopts.rootMode = linux.FileMode(mode)
 	} else {
 		ctx.Warningf("fusefs.FilesystemType.GetFilesystem: mandatory mount option rootmode missing")
-		return nil, nil, syserror.EINVAL
+		return nil, nil, linuxerr.EINVAL
 	}
 
 	// Set the maxInFlightRequests option.
@@ -207,7 +207,7 @@ func (fsType FilesystemType) GetFilesystem(ctx context.Context, vfsObj *vfs.Virt
 		maxRead, err := strconv.ParseUint(maxReadStr, 10, 32)
 		if err != nil {
 			log.Warningf("%s.GetFilesystem: invalid max_read: max_read=%s", fsType.Name(), maxReadStr)
-			return nil, nil, syserror.EINVAL
+			return nil, nil, linuxerr.EINVAL
 		}
 		if maxRead < fuseMinMaxRead {
 			maxRead = fuseMinMaxRead
@@ -230,7 +230,7 @@ func (fsType FilesystemType) GetFilesystem(ctx context.Context, vfsObj *vfs.Virt
 	// Check for unparsed options.
 	if len(mopts) != 0 {
 		log.Warningf("%s.GetFilesystem: unsupported or unknown options: %v", fsType.Name(), mopts)
-		return nil, nil, syserror.EINVAL
+		return nil, nil, linuxerr.EINVAL
 	}
 
 	// Create a new FUSE filesystem.
@@ -259,7 +259,7 @@ func newFUSEFilesystem(ctx context.Context, vfsObj *vfs.VirtualFilesystem, fsTyp
 	conn, err := newFUSEConnection(ctx, fuseFD, opts)
 	if err != nil {
 		log.Warningf("fuse.NewFUSEFilesystem: NewFUSEConnection failed with error: %v", err)
-		return nil, syserror.EINVAL
+		return nil, linuxerr.EINVAL
 	}
 
 	fs := &filesystem{
@@ -419,7 +419,7 @@ func (i *inode) Open(ctx context.Context, rp *vfs.ResolvingPath, d *kernfs.Dentr
 		kernelTask := kernel.TaskFromContext(ctx)
 		if kernelTask == nil {
 			log.Warningf("fusefs.Inode.Open: couldn't get kernel task from context")
-			return nil, syserror.EINVAL
+			return nil, linuxerr.EINVAL
 		}
 
 		// Build the request.
@@ -513,7 +513,7 @@ func (i *inode) NewFile(ctx context.Context, name string, opts vfs.OpenOptions)
 	kernelTask := kernel.TaskFromContext(ctx)
 	if kernelTask == nil {
 		log.Warningf("fusefs.Inode.NewFile: couldn't get kernel task from context", i.nodeID)
-		return nil, syserror.EINVAL
+		return nil, linuxerr.EINVAL
 	}
 	in := linux.FUSECreateIn{
 		CreateMeta: linux.FUSECreateMeta{
@@ -553,7 +553,7 @@ func (i *inode) Unlink(ctx context.Context, name string, child kernfs.Inode) err
 	kernelTask := kernel.TaskFromContext(ctx)
 	if kernelTask == nil {
 		log.Warningf("fusefs.Inode.newEntry: couldn't get kernel task from context", i.nodeID)
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	in := linux.FUSEUnlinkIn{Name: name}
 	req := i.fs.conn.NewRequest(auth.CredentialsFromContext(ctx), uint32(kernelTask.ThreadID()), i.nodeID, linux.FUSE_UNLINK, &in)
@@ -597,7 +597,7 @@ func (i *inode) newEntry(ctx context.Context, name string, fileType linux.FileMo
 	kernelTask := kernel.TaskFromContext(ctx)
 	if kernelTask == nil {
 		log.Warningf("fusefs.Inode.newEntry: couldn't get kernel task from context", i.nodeID)
-		return nil, syserror.EINVAL
+		return nil, linuxerr.EINVAL
 	}
 	req := i.fs.conn.NewRequest(auth.CredentialsFromContext(ctx), uint32(kernelTask.ThreadID()), i.nodeID, opcode, payload)
 	res, err := i.fs.conn.Call(kernelTask, req)
@@ -627,13 +627,13 @@ func (i *inode) Getlink(ctx context.Context, mnt *vfs.Mount) (vfs.VirtualDentry,
 // Readlink implements kernfs.Inode.Readlink.
 func (i *inode) Readlink(ctx context.Context, mnt *vfs.Mount) (string, error) {
 	if i.Mode().FileType()&linux.S_IFLNK == 0 {
-		return "", syserror.EINVAL
+		return "", linuxerr.EINVAL
 	}
 	if len(i.link) == 0 {
 		kernelTask := kernel.TaskFromContext(ctx)
 		if kernelTask == nil {
 			log.Warningf("fusefs.Inode.Readlink: couldn't get kernel task from context")
-			return "", syserror.EINVAL
+			return "", linuxerr.EINVAL
 		}
 		req := i.fs.conn.NewRequest(auth.CredentialsFromContext(ctx), uint32(kernelTask.ThreadID()), i.nodeID, linux.FUSE_READLINK, &linux.FUSEEmptyIn{})
 		res, err := i.fs.conn.Call(kernelTask, req)
@@ -729,7 +729,7 @@ func (i *inode) getAttr(ctx context.Context, fs *vfs.Filesystem, opts vfs.StatOp
 	task := kernel.TaskFromContext(ctx)
 	if task == nil {
 		log.Warningf("couldn't get kernel task from context")
-		return linux.FUSEAttr{}, syserror.EINVAL
+		return linux.FUSEAttr{}, linuxerr.EINVAL
 	}
 
 	creds := auth.CredentialsFromContext(ctx)
@@ -834,7 +834,7 @@ func (i *inode) setAttr(ctx context.Context, fs *vfs.Filesystem, creds *auth.Cre
 	task := kernel.TaskFromContext(ctx)
 	if task == nil {
 		log.Warningf("couldn't get kernel task from context")
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 
 	// We should retain the original file type when assigning new mode.
diff --git a/pkg/sentry/fsimpl/fuse/read_write.go b/pkg/sentry/fsimpl/fuse/read_write.go
index 66ea889f9..35d0ab6f4 100644
--- a/pkg/sentry/fsimpl/fuse/read_write.go
+++ b/pkg/sentry/fsimpl/fuse/read_write.go
@@ -20,6 +20,7 @@ import (
 
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/log"
 	"gvisor.dev/gvisor/pkg/sentry/kernel"
@@ -39,7 +40,7 @@ func (fs *filesystem) ReadInPages(ctx context.Context, fd *regularFileFD, off ui
 	t := kernel.TaskFromContext(ctx)
 	if t == nil {
 		log.Warningf("fusefs.Read: couldn't get kernel task from context")
-		return nil, 0, syserror.EINVAL
+		return nil, 0, linuxerr.EINVAL
 	}
 
 	// Round up to a multiple of page size.
@@ -155,7 +156,7 @@ func (fs *filesystem) Write(ctx context.Context, fd *regularFileFD, off uint64,
 	t := kernel.TaskFromContext(ctx)
 	if t == nil {
 		log.Warningf("fusefs.Read: couldn't get kernel task from context")
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 
 	// One request cannnot exceed either maxWrite or maxPages.
diff --git a/pkg/sentry/fsimpl/fuse/regular_file.go b/pkg/sentry/fsimpl/fuse/regular_file.go
index 5bdd096c3..a0802cd32 100644
--- a/pkg/sentry/fsimpl/fuse/regular_file.go
+++ b/pkg/sentry/fsimpl/fuse/regular_file.go
@@ -22,6 +22,7 @@ import (
 
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/sentry/vfs"
 	"gvisor.dev/gvisor/pkg/syserror"
 	"gvisor.dev/gvisor/pkg/usermem"
@@ -39,7 +40,7 @@ type regularFileFD struct {
 // PRead implements vfs.FileDescriptionImpl.PRead.
 func (fd *regularFileFD) PRead(ctx context.Context, dst usermem.IOSequence, offset int64, opts vfs.ReadOptions) (int64, error) {
 	if offset < 0 {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 
 	// Check that flags are supported.
@@ -56,7 +57,7 @@ func (fd *regularFileFD) PRead(ctx context.Context, dst usermem.IOSequence, offs
 	} else if size > math.MaxUint32 {
 		// FUSE only supports uint32 for size.
 		// Overflow.
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 
 	// TODO(gvisor.dev/issue/3678): Add direct IO support.
@@ -143,7 +144,7 @@ func (fd *regularFileFD) Write(ctx context.Context, src usermem.IOSequence, opts
 // final offset should be ignored by PWrite.
 func (fd *regularFileFD) pwrite(ctx context.Context, src usermem.IOSequence, offset int64, opts vfs.WriteOptions) (written, finalOff int64, err error) {
 	if offset < 0 {
-		return 0, offset, syserror.EINVAL
+		return 0, offset, linuxerr.EINVAL
 	}
 
 	// Check that flags are supported.
@@ -171,11 +172,11 @@ func (fd *regularFileFD) pwrite(ctx context.Context, src usermem.IOSequence, off
 	if srclen > math.MaxUint32 {
 		// FUSE only supports uint32 for size.
 		// Overflow.
-		return 0, offset, syserror.EINVAL
+		return 0, offset, linuxerr.EINVAL
 	}
 	if end := offset + srclen; end < offset {
 		// Overflow.
-		return 0, offset, syserror.EINVAL
+		return 0, offset, linuxerr.EINVAL
 	}
 
 	srclen, err = vfs.CheckLimit(ctx, offset, srclen)
diff --git a/pkg/sentry/fsimpl/gofer/directory.go b/pkg/sentry/fsimpl/gofer/directory.go
index 177e42649..5c48a9fee 100644
--- a/pkg/sentry/fsimpl/gofer/directory.go
+++ b/pkg/sentry/fsimpl/gofer/directory.go
@@ -20,6 +20,7 @@ import (
 
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/p9"
 	"gvisor.dev/gvisor/pkg/refsvfs2"
@@ -28,7 +29,6 @@ import (
 	"gvisor.dev/gvisor/pkg/sentry/socket/unix/transport"
 	"gvisor.dev/gvisor/pkg/sentry/vfs"
 	"gvisor.dev/gvisor/pkg/sync"
-	"gvisor.dev/gvisor/pkg/syserror"
 )
 
 func (d *dentry) isDir() bool {
@@ -297,7 +297,7 @@ func (fd *directoryFD) Seek(ctx context.Context, offset int64, whence int32) (in
 	switch whence {
 	case linux.SEEK_SET:
 		if offset < 0 {
-			return 0, syserror.EINVAL
+			return 0, linuxerr.EINVAL
 		}
 		if offset == 0 {
 			// Ensure that the next call to fd.IterDirents() calls
@@ -309,13 +309,13 @@ func (fd *directoryFD) Seek(ctx context.Context, offset int64, whence int32) (in
 	case linux.SEEK_CUR:
 		offset += fd.off
 		if offset < 0 {
-			return 0, syserror.EINVAL
+			return 0, linuxerr.EINVAL
 		}
 		// Don't clear fd.dirents in this case, even if offset == 0.
 		fd.off = offset
 		return fd.off, nil
 	default:
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 }
 
diff --git a/pkg/sentry/fsimpl/gofer/filesystem.go b/pkg/sentry/fsimpl/gofer/filesystem.go
index af2b773c3..067b7aac1 100644
--- a/pkg/sentry/fsimpl/gofer/filesystem.go
+++ b/pkg/sentry/fsimpl/gofer/filesystem.go
@@ -470,7 +470,7 @@ func (fs *filesystem) unlinkAt(ctx context.Context, rp *vfs.ResolvingPath, dir b
 	name := rp.Component()
 	if dir {
 		if name == "." {
-			return syserror.EINVAL
+			return linuxerr.EINVAL
 		}
 		if name == ".." {
 			return syserror.ENOTEMPTY
@@ -943,7 +943,7 @@ func (d *dentry) open(ctx context.Context, rp *vfs.ResolvingPath, opts *vfs.Open
 			return nil, syserror.EISDIR
 		}
 		if opts.Flags&linux.O_DIRECT != 0 {
-			return nil, syserror.EINVAL
+			return nil, linuxerr.EINVAL
 		}
 		if !d.isSynthetic() {
 			if err := d.ensureSharedHandle(ctx, ats&vfs.MayRead != 0, false /* write */, false /* trunc */); err != nil {
@@ -999,7 +999,7 @@ func (d *dentry) open(ctx context.Context, rp *vfs.ResolvingPath, opts *vfs.Open
 
 func (d *dentry) openSocketByConnecting(ctx context.Context, opts *vfs.OpenOptions) (*vfs.FileDescription, error) {
 	if opts.Flags&linux.O_DIRECT != 0 {
-		return nil, syserror.EINVAL
+		return nil, linuxerr.EINVAL
 	}
 	fdObj, err := d.file.connect(ctx, p9.AnonymousSocket)
 	if err != nil {
@@ -1020,7 +1020,7 @@ func (d *dentry) openSocketByConnecting(ctx context.Context, opts *vfs.OpenOptio
 func (d *dentry) openSpecialFile(ctx context.Context, mnt *vfs.Mount, opts *vfs.OpenOptions) (*vfs.FileDescription, error) {
 	ats := vfs.AccessTypesForOpenFlags(opts)
 	if opts.Flags&linux.O_DIRECT != 0 {
-		return nil, syserror.EINVAL
+		return nil, linuxerr.EINVAL
 	}
 	// We assume that the server silently inserts O_NONBLOCK in the open flags
 	// for all named pipes (because all existing gofers do this).
@@ -1188,7 +1188,7 @@ func (fs *filesystem) ReadlinkAt(ctx context.Context, rp *vfs.ResolvingPath) (st
 		return "", err
 	}
 	if !d.isSymlink() {
-		return "", syserror.EINVAL
+		return "", linuxerr.EINVAL
 	}
 	return d.readlink(ctx, rp.Mount())
 }
@@ -1205,12 +1205,12 @@ func (fs *filesystem) RenameAt(ctx context.Context, rp *vfs.ResolvingPath, oldPa
 	}
 
 	if opts.Flags&^linux.RENAME_NOREPLACE != 0 {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	if fs.opts.interop == InteropModeShared && opts.Flags&linux.RENAME_NOREPLACE != 0 {
 		// Requires 9P support to synchronize with other remote filesystem
 		// users.
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 
 	newName := rp.Component()
@@ -1261,7 +1261,7 @@ func (fs *filesystem) RenameAt(ctx context.Context, rp *vfs.ResolvingPath, oldPa
 	}
 	if renamed.isDir() {
 		if renamed == newParent || genericIsAncestorDentry(renamed, newParent) {
-			return syserror.EINVAL
+			return linuxerr.EINVAL
 		}
 		if oldParent != newParent {
 			if err := renamed.checkPermissions(creds, vfs.MayWrite); err != nil {
diff --git a/pkg/sentry/fsimpl/gofer/gofer.go b/pkg/sentry/fsimpl/gofer/gofer.go
index 496e31e34..c7ebd435c 100644
--- a/pkg/sentry/fsimpl/gofer/gofer.go
+++ b/pkg/sentry/fsimpl/gofer/gofer.go
@@ -319,7 +319,7 @@ func (fstype FilesystemType) GetFilesystem(ctx context.Context, vfsObj *vfs.Virt
 	mfp := pgalloc.MemoryFileProviderFromContext(ctx)
 	if mfp == nil {
 		ctx.Warningf("gofer.FilesystemType.GetFilesystem: context does not provide a pgalloc.MemoryFileProvider")
-		return nil, nil, syserror.EINVAL
+		return nil, nil, linuxerr.EINVAL
 	}
 
 	mopts := vfs.GenericParseMountOptions(opts.Data)
@@ -355,7 +355,7 @@ func (fstype FilesystemType) GetFilesystem(ctx context.Context, vfsObj *vfs.Virt
 			fsopts.interop = InteropModeShared
 		default:
 			ctx.Warningf("gofer.FilesystemType.GetFilesystem: invalid cache policy: %s=%s", moptCache, cache)
-			return nil, nil, syserror.EINVAL
+			return nil, nil, linuxerr.EINVAL
 		}
 	}
 
@@ -366,7 +366,7 @@ func (fstype FilesystemType) GetFilesystem(ctx context.Context, vfsObj *vfs.Virt
 		dfltuid, err := strconv.ParseUint(dfltuidstr, 10, 32)
 		if err != nil {
 			ctx.Warningf("gofer.FilesystemType.GetFilesystem: invalid default UID: %s=%s", moptDfltUID, dfltuidstr)
-			return nil, nil, syserror.EINVAL
+			return nil, nil, linuxerr.EINVAL
 		}
 		// In Linux, dfltuid is interpreted as a UID and is converted to a KUID
 		// in the caller's user namespace, but goferfs isn't
@@ -379,7 +379,7 @@ func (fstype FilesystemType) GetFilesystem(ctx context.Context, vfsObj *vfs.Virt
 		dfltgid, err := strconv.ParseUint(dfltgidstr, 10, 32)
 		if err != nil {
 			ctx.Warningf("gofer.FilesystemType.GetFilesystem: invalid default UID: %s=%s", moptDfltGID, dfltgidstr)
-			return nil, nil, syserror.EINVAL
+			return nil, nil, linuxerr.EINVAL
 		}
 		fsopts.dfltgid = auth.KGID(dfltgid)
 	}
@@ -391,7 +391,7 @@ func (fstype FilesystemType) GetFilesystem(ctx context.Context, vfsObj *vfs.Virt
 		msize, err := strconv.ParseUint(msizestr, 10, 32)
 		if err != nil {
 			ctx.Warningf("gofer.FilesystemType.GetFilesystem: invalid message size: %s=%s", moptMsize, msizestr)
-			return nil, nil, syserror.EINVAL
+			return nil, nil, linuxerr.EINVAL
 		}
 		fsopts.msize = uint32(msize)
 	}
@@ -410,7 +410,7 @@ func (fstype FilesystemType) GetFilesystem(ctx context.Context, vfsObj *vfs.Virt
 		maxCachedDentries, err := strconv.ParseUint(str, 10, 64)
 		if err != nil {
 			ctx.Warningf("gofer.FilesystemType.GetFilesystem: invalid dentry cache limit: %s=%s", moptDentryCacheLimit, str)
-			return nil, nil, syserror.EINVAL
+			return nil, nil, linuxerr.EINVAL
 		}
 		fsopts.maxCachedDentries = maxCachedDentries
 	}
@@ -434,14 +434,14 @@ func (fstype FilesystemType) GetFilesystem(ctx context.Context, vfsObj *vfs.Virt
 	// Check for unparsed options.
 	if len(mopts) != 0 {
 		ctx.Warningf("gofer.FilesystemType.GetFilesystem: unknown options: %v", mopts)
-		return nil, nil, syserror.EINVAL
+		return nil, nil, linuxerr.EINVAL
 	}
 
 	// Handle internal options.
 	iopts, ok := opts.InternalData.(InternalFilesystemOptions)
 	if opts.InternalData != nil && !ok {
 		ctx.Warningf("gofer.FilesystemType.GetFilesystem: GetFilesystemOptions.InternalData has type %T, wanted gofer.InternalFilesystemOptions", opts.InternalData)
-		return nil, nil, syserror.EINVAL
+		return nil, nil, linuxerr.EINVAL
 	}
 	// If !ok, iopts being the zero value is correct.
 
@@ -504,7 +504,7 @@ func getFDFromMountOptionsMap(ctx context.Context, mopts map[string]string) (int
 	trans, ok := mopts[moptTransport]
 	if !ok || trans != transportModeFD {
 		ctx.Warningf("gofer.getFDFromMountOptionsMap: transport must be specified as '%s=%s'", moptTransport, transportModeFD)
-		return -1, syserror.EINVAL
+		return -1, linuxerr.EINVAL
 	}
 	delete(mopts, moptTransport)
 
@@ -512,28 +512,28 @@ func getFDFromMountOptionsMap(ctx context.Context, mopts map[string]string) (int
 	rfdstr, ok := mopts[moptReadFD]
 	if !ok {
 		ctx.Warningf("gofer.getFDFromMountOptionsMap: read FD must be specified as '%s=<file descriptor>'", moptReadFD)
-		return -1, syserror.EINVAL
+		return -1, linuxerr.EINVAL
 	}
 	delete(mopts, moptReadFD)
 	rfd, err := strconv.Atoi(rfdstr)
 	if err != nil {
 		ctx.Warningf("gofer.getFDFromMountOptionsMap: invalid read FD: %s=%s", moptReadFD, rfdstr)
-		return -1, syserror.EINVAL
+		return -1, linuxerr.EINVAL
 	}
 	wfdstr, ok := mopts[moptWriteFD]
 	if !ok {
 		ctx.Warningf("gofer.getFDFromMountOptionsMap: write FD must be specified as '%s=<file descriptor>'", moptWriteFD)
-		return -1, syserror.EINVAL
+		return -1, linuxerr.EINVAL
 	}
 	delete(mopts, moptWriteFD)
 	wfd, err := strconv.Atoi(wfdstr)
 	if err != nil {
 		ctx.Warningf("gofer.getFDFromMountOptionsMap: invalid write FD: %s=%s", moptWriteFD, wfdstr)
-		return -1, syserror.EINVAL
+		return -1, linuxerr.EINVAL
 	}
 	if rfd != wfd {
 		ctx.Warningf("gofer.getFDFromMountOptionsMap: read FD (%d) and write FD (%d) must be equal", rfd, wfd)
-		return -1, syserror.EINVAL
+		return -1, linuxerr.EINVAL
 	}
 	return rfd, nil
 }
@@ -1111,7 +1111,7 @@ func (d *dentry) setStat(ctx context.Context, creds *auth.Credentials, opts *vfs
 		case linux.S_IFDIR:
 			return syserror.EISDIR
 		default:
-			return syserror.EINVAL
+			return linuxerr.EINVAL
 		}
 	}
 
diff --git a/pkg/sentry/fsimpl/gofer/regular_file.go b/pkg/sentry/fsimpl/gofer/regular_file.go
index 340fea813..89eab04cd 100644
--- a/pkg/sentry/fsimpl/gofer/regular_file.go
+++ b/pkg/sentry/fsimpl/gofer/regular_file.go
@@ -22,6 +22,7 @@ import (
 
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/log"
 	"gvisor.dev/gvisor/pkg/metric"
@@ -124,7 +125,7 @@ func (fd *regularFileFD) PRead(ctx context.Context, dst usermem.IOSequence, offs
 	}()
 
 	if offset < 0 {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 
 	// Check that flags are supported.
@@ -194,7 +195,7 @@ func (fd *regularFileFD) PWrite(ctx context.Context, src usermem.IOSequence, off
 // offset should be ignored by PWrite.
 func (fd *regularFileFD) pwrite(ctx context.Context, src usermem.IOSequence, offset int64, opts vfs.WriteOptions) (written, finalOff int64, err error) {
 	if offset < 0 {
-		return 0, offset, syserror.EINVAL
+		return 0, offset, linuxerr.EINVAL
 	}
 
 	// Check that flags are supported.
@@ -297,7 +298,7 @@ func (fd *regularFileFD) writeCache(ctx context.Context, d *dentry, offset int64
 	pgstart := hostarch.PageRoundDown(uint64(offset))
 	pgend, ok := hostarch.PageRoundUp(uint64(offset + src.NumBytes()))
 	if !ok {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	mr := memmap.MappableRange{pgstart, pgend}
 	var freed []memmap.FileRange
@@ -662,10 +663,10 @@ func regularFileSeekLocked(ctx context.Context, d *dentry, fdOffset, offset int6
 			offset = size
 		}
 	default:
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 	if offset < 0 {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 	return offset, nil
 }
diff --git a/pkg/sentry/fsimpl/gofer/special_file.go b/pkg/sentry/fsimpl/gofer/special_file.go
index 3d7b5506e..2a922d120 100644
--- a/pkg/sentry/fsimpl/gofer/special_file.go
+++ b/pkg/sentry/fsimpl/gofer/special_file.go
@@ -184,7 +184,7 @@ func (fd *specialFileFD) PRead(ctx context.Context, dst usermem.IOSequence, offs
 	}()
 
 	if fd.seekable && offset < 0 {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 
 	// Check that flags are supported.
@@ -264,7 +264,7 @@ func (fd *specialFileFD) PWrite(ctx context.Context, src usermem.IOSequence, off
 // offset should be ignored by PWrite.
 func (fd *specialFileFD) pwrite(ctx context.Context, src usermem.IOSequence, offset int64, opts vfs.WriteOptions) (written, finalOff int64, err error) {
 	if fd.seekable && offset < 0 {
-		return 0, offset, syserror.EINVAL
+		return 0, offset, linuxerr.EINVAL
 	}
 
 	// Check that flags are supported.
diff --git a/pkg/sentry/fsimpl/host/host.go b/pkg/sentry/fsimpl/host/host.go
index 2dbfbdecf..4d2b282a0 100644
--- a/pkg/sentry/fsimpl/host/host.go
+++ b/pkg/sentry/fsimpl/host/host.go
@@ -290,10 +290,10 @@ func (i *inode) Mode() linux.FileMode {
 // Stat implements kernfs.Inode.Stat.
 func (i *inode) Stat(ctx context.Context, vfsfs *vfs.Filesystem, opts vfs.StatOptions) (linux.Statx, error) {
 	if opts.Mask&linux.STATX__RESERVED != 0 {
-		return linux.Statx{}, syserror.EINVAL
+		return linux.Statx{}, linuxerr.EINVAL
 	}
 	if opts.Sync&linux.AT_STATX_SYNC_TYPE == linux.AT_STATX_SYNC_TYPE {
-		return linux.Statx{}, syserror.EINVAL
+		return linux.Statx{}, linuxerr.EINVAL
 	}
 
 	fs := vfsfs.Impl().(*filesystem)
@@ -426,7 +426,7 @@ func (i *inode) SetStat(ctx context.Context, fs *vfs.Filesystem, creds *auth.Cre
 	}
 	if m&linux.STATX_SIZE != 0 {
 		if hostStat.Mode&linux.S_IFMT != linux.S_IFREG {
-			return syserror.EINVAL
+			return linuxerr.EINVAL
 		}
 		if err := unix.Ftruncate(i.hostFD, int64(s.Size)); err != nil {
 			return err
@@ -731,7 +731,7 @@ func (f *fileDescription) Seek(_ context.Context, offset int64, whence int32) (i
 	switch whence {
 	case linux.SEEK_SET:
 		if offset < 0 {
-			return f.offset, syserror.EINVAL
+			return f.offset, linuxerr.EINVAL
 		}
 		f.offset = offset
 
@@ -741,7 +741,7 @@ func (f *fileDescription) Seek(_ context.Context, offset int64, whence int32) (i
 			return f.offset, syserror.EOVERFLOW
 		}
 		if f.offset+offset < 0 {
-			return f.offset, syserror.EINVAL
+			return f.offset, linuxerr.EINVAL
 		}
 		f.offset += offset
 
@@ -757,7 +757,7 @@ func (f *fileDescription) Seek(_ context.Context, offset int64, whence int32) (i
 			return f.offset, syserror.EOVERFLOW
 		}
 		if size+offset < 0 {
-			return f.offset, syserror.EINVAL
+			return f.offset, linuxerr.EINVAL
 		}
 		f.offset = size + offset
 
@@ -774,7 +774,7 @@ func (f *fileDescription) Seek(_ context.Context, offset int64, whence int32) (i
 
 	default:
 		// Invalid whence.
-		return f.offset, syserror.EINVAL
+		return f.offset, linuxerr.EINVAL
 	}
 
 	return f.offset, nil
diff --git a/pkg/sentry/fsimpl/host/tty.go b/pkg/sentry/fsimpl/host/tty.go
index 2cf360065..c7bf563f0 100644
--- a/pkg/sentry/fsimpl/host/tty.go
+++ b/pkg/sentry/fsimpl/host/tty.go
@@ -231,7 +231,7 @@ func (t *TTYFileDescription) Ioctl(ctx context.Context, io usermem.IO, args arch
 
 		// pgID must be non-negative.
 		if pgID < 0 {
-			return 0, syserror.EINVAL
+			return 0, linuxerr.EINVAL
 		}
 
 		// Process group with pgID must exist in this PID namespace.
diff --git a/pkg/sentry/fsimpl/kernfs/fd_impl_util.go b/pkg/sentry/fsimpl/kernfs/fd_impl_util.go
index e55111af0..8b008dc10 100644
--- a/pkg/sentry/fsimpl/kernfs/fd_impl_util.go
+++ b/pkg/sentry/fsimpl/kernfs/fd_impl_util.go
@@ -19,6 +19,7 @@ import (
 
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/sentry/kernel/auth"
 	"gvisor.dev/gvisor/pkg/sentry/memmap"
 	"gvisor.dev/gvisor/pkg/sentry/vfs"
@@ -248,10 +249,10 @@ func (fd *GenericDirectoryFD) Seek(ctx context.Context, offset int64, whence int
 			panic(fmt.Sprintf("Invalid GenericDirectoryFD.seekEnd = %v", fd.seekEnd))
 		}
 	default:
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 	if offset < 0 {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 	fd.off = offset
 	return offset, nil
diff --git a/pkg/sentry/fsimpl/kernfs/filesystem.go b/pkg/sentry/fsimpl/kernfs/filesystem.go
index 20319ab76..1a314f59e 100644
--- a/pkg/sentry/fsimpl/kernfs/filesystem.go
+++ b/pkg/sentry/fsimpl/kernfs/filesystem.go
@@ -623,7 +623,7 @@ func (fs *Filesystem) ReadlinkAt(ctx context.Context, rp *vfs.ResolvingPath) (st
 	}
 	if !d.isSymlink() {
 		fs.mu.RUnlock()
-		return "", syserror.EINVAL
+		return "", linuxerr.EINVAL
 	}
 
 	// Inode.Readlink() cannot be called holding fs locks.
@@ -649,7 +649,7 @@ func (fs *Filesystem) RenameAt(ctx context.Context, rp *vfs.ResolvingPath, oldPa
 
 	// Only RENAME_NOREPLACE is supported.
 	if opts.Flags&^linux.RENAME_NOREPLACE != 0 {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	noReplace := opts.Flags&linux.RENAME_NOREPLACE != 0
 
diff --git a/pkg/sentry/fsimpl/kernfs/inode_impl_util.go b/pkg/sentry/fsimpl/kernfs/inode_impl_util.go
index 3d0866ecf..62872946e 100644
--- a/pkg/sentry/fsimpl/kernfs/inode_impl_util.go
+++ b/pkg/sentry/fsimpl/kernfs/inode_impl_util.go
@@ -20,6 +20,7 @@ import (
 
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/sentry/kernel/auth"
 	ktime "gvisor.dev/gvisor/pkg/sentry/kernel/time"
@@ -158,12 +159,12 @@ type InodeNotSymlink struct{}
 
 // Readlink implements Inode.Readlink.
 func (InodeNotSymlink) Readlink(context.Context, *vfs.Mount) (string, error) {
-	return "", syserror.EINVAL
+	return "", linuxerr.EINVAL
 }
 
 // Getlink implements Inode.Getlink.
 func (InodeNotSymlink) Getlink(context.Context, *vfs.Mount) (vfs.VirtualDentry, string, error) {
-	return vfs.VirtualDentry{}, "", syserror.EINVAL
+	return vfs.VirtualDentry{}, "", linuxerr.EINVAL
 }
 
 // InodeAttrs partially implements the Inode interface, specifically the
diff --git a/pkg/sentry/fsimpl/overlay/directory.go b/pkg/sentry/fsimpl/overlay/directory.go
index df4492346..417a7c630 100644
--- a/pkg/sentry/fsimpl/overlay/directory.go
+++ b/pkg/sentry/fsimpl/overlay/directory.go
@@ -19,6 +19,7 @@ import (
 
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/fspath"
 	"gvisor.dev/gvisor/pkg/sentry/vfs"
 	"gvisor.dev/gvisor/pkg/sync"
@@ -256,7 +257,7 @@ func (fd *directoryFD) Seek(ctx context.Context, offset int64, whence int32) (in
 	switch whence {
 	case linux.SEEK_SET:
 		if offset < 0 {
-			return 0, syserror.EINVAL
+			return 0, linuxerr.EINVAL
 		}
 		if offset == 0 {
 			// Ensure that the next call to fd.IterDirents() calls
@@ -268,13 +269,13 @@ func (fd *directoryFD) Seek(ctx context.Context, offset int64, whence int32) (in
 	case linux.SEEK_CUR:
 		offset += fd.off
 		if offset < 0 {
-			return 0, syserror.EINVAL
+			return 0, linuxerr.EINVAL
 		}
 		// Don't clear fd.dirents in this case, even if offset == 0.
 		fd.off = offset
 		return fd.off, nil
 	default:
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 }
 
diff --git a/pkg/sentry/fsimpl/overlay/filesystem.go b/pkg/sentry/fsimpl/overlay/filesystem.go
index 81745bccd..e792677f5 100644
--- a/pkg/sentry/fsimpl/overlay/filesystem.go
+++ b/pkg/sentry/fsimpl/overlay/filesystem.go
@@ -872,7 +872,7 @@ func (d *dentry) openCopiedUp(ctx context.Context, rp *vfs.ResolvingPath, opts *
 			return nil, syserror.EISDIR
 		}
 		if opts.Flags&linux.O_DIRECT != 0 {
-			return nil, syserror.EINVAL
+			return nil, linuxerr.EINVAL
 		}
 		fd := &directoryFD{}
 		fd.LockFD.Init(&d.locks)
@@ -1028,7 +1028,7 @@ func (fs *filesystem) RenameAt(ctx context.Context, rp *vfs.ResolvingPath, oldPa
 	}
 
 	if opts.Flags&^linux.RENAME_NOREPLACE != 0 {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 
 	newName := rp.Component()
@@ -1065,7 +1065,7 @@ func (fs *filesystem) RenameAt(ctx context.Context, rp *vfs.ResolvingPath, oldPa
 	}
 	if renamed.isDir() {
 		if renamed == newParent || genericIsAncestorDentry(renamed, newParent) {
-			return syserror.EINVAL
+			return linuxerr.EINVAL
 		}
 		if oldParent != newParent {
 			if err := renamed.checkPermissions(creds, vfs.MayWrite); err != nil {
@@ -1286,7 +1286,7 @@ func (fs *filesystem) RmdirAt(ctx context.Context, rp *vfs.ResolvingPath) error
 	defer rp.Mount().EndWrite()
 	name := rp.Component()
 	if name == "." {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	if name == ".." {
 		return syserror.ENOTEMPTY
diff --git a/pkg/sentry/fsimpl/overlay/overlay.go b/pkg/sentry/fsimpl/overlay/overlay.go
index 454c20d4f..4c7243764 100644
--- a/pkg/sentry/fsimpl/overlay/overlay.go
+++ b/pkg/sentry/fsimpl/overlay/overlay.go
@@ -40,6 +40,7 @@ import (
 
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/fspath"
 	"gvisor.dev/gvisor/pkg/refsvfs2"
 	"gvisor.dev/gvisor/pkg/sentry/kernel/auth"
@@ -135,7 +136,7 @@ func (fstype FilesystemType) GetFilesystem(ctx context.Context, vfsObj *vfs.Virt
 	fsopts, ok := fsoptsRaw.(FilesystemOptions)
 	if fsoptsRaw != nil && !ok {
 		ctx.Infof("overlay.FilesystemType.GetFilesystem: GetFilesystemOptions.InternalData has type %T, wanted overlay.FilesystemOptions or nil", fsoptsRaw)
-		return nil, nil, syserror.EINVAL
+		return nil, nil, linuxerr.EINVAL
 	}
 	vfsroot := vfs.RootFromContext(ctx)
 	if vfsroot.Ok() {
@@ -145,7 +146,7 @@ func (fstype FilesystemType) GetFilesystem(ctx context.Context, vfsObj *vfs.Virt
 	if upperPathname, ok := mopts["upperdir"]; ok {
 		if fsopts.UpperRoot.Ok() {
 			ctx.Infof("overlay.FilesystemType.GetFilesystem: both upperdir and FilesystemOptions.UpperRoot are specified")
-			return nil, nil, syserror.EINVAL
+			return nil, nil, linuxerr.EINVAL
 		}
 		delete(mopts, "upperdir")
 		// Linux overlayfs also requires a workdir when upperdir is
@@ -154,7 +155,7 @@ func (fstype FilesystemType) GetFilesystem(ctx context.Context, vfsObj *vfs.Virt
 		upperPath := fspath.Parse(upperPathname)
 		if !upperPath.Absolute {
 			ctx.Infof("overlay.FilesystemType.GetFilesystem: upperdir %q must be absolute", upperPathname)
-			return nil, nil, syserror.EINVAL
+			return nil, nil, linuxerr.EINVAL
 		}
 		upperRoot, err := vfsObj.GetDentryAt(ctx, creds, &vfs.PathOperation{
 			Root:               vfsroot,
@@ -181,7 +182,7 @@ func (fstype FilesystemType) GetFilesystem(ctx context.Context, vfsObj *vfs.Virt
 	if lowerPathnamesStr, ok := mopts["lowerdir"]; ok {
 		if len(fsopts.LowerRoots) != 0 {
 			ctx.Infof("overlay.FilesystemType.GetFilesystem: both lowerdir and FilesystemOptions.LowerRoots are specified")
-			return nil, nil, syserror.EINVAL
+			return nil, nil, linuxerr.EINVAL
 		}
 		delete(mopts, "lowerdir")
 		lowerPathnames := strings.Split(lowerPathnamesStr, ":")
@@ -189,7 +190,7 @@ func (fstype FilesystemType) GetFilesystem(ctx context.Context, vfsObj *vfs.Virt
 			lowerPath := fspath.Parse(lowerPathname)
 			if !lowerPath.Absolute {
 				ctx.Infof("overlay.FilesystemType.GetFilesystem: lowerdir %q must be absolute", lowerPathname)
-				return nil, nil, syserror.EINVAL
+				return nil, nil, linuxerr.EINVAL
 			}
 			lowerRoot, err := vfsObj.GetDentryAt(ctx, creds, &vfs.PathOperation{
 				Root:               vfsroot,
@@ -216,21 +217,21 @@ func (fstype FilesystemType) GetFilesystem(ctx context.Context, vfsObj *vfs.Virt
 
 	if len(mopts) != 0 {
 		ctx.Infof("overlay.FilesystemType.GetFilesystem: unused options: %v", mopts)
-		return nil, nil, syserror.EINVAL
+		return nil, nil, linuxerr.EINVAL
 	}
 
 	if len(fsopts.LowerRoots) == 0 {
 		ctx.Infof("overlay.FilesystemType.GetFilesystem: at least one lower layer is required")
-		return nil, nil, syserror.EINVAL
+		return nil, nil, linuxerr.EINVAL
 	}
 	if len(fsopts.LowerRoots) < 2 && !fsopts.UpperRoot.Ok() {
 		ctx.Infof("overlay.FilesystemType.GetFilesystem: at least two lower layers are required when no upper layer is present")
-		return nil, nil, syserror.EINVAL
+		return nil, nil, linuxerr.EINVAL
 	}
 	const maxLowerLayers = 500 // Linux: fs/overlay/super.c:OVL_MAX_STACK
 	if len(fsopts.LowerRoots) > maxLowerLayers {
 		ctx.Infof("overlay.FilesystemType.GetFilesystem: %d lower layers specified, maximum %d", len(fsopts.LowerRoots), maxLowerLayers)
-		return nil, nil, syserror.EINVAL
+		return nil, nil, linuxerr.EINVAL
 	}
 
 	// Take extra references held by the filesystem.
@@ -283,7 +284,7 @@ func (fstype FilesystemType) GetFilesystem(ctx context.Context, vfsObj *vfs.Virt
 		ctx.Infof("overlay.FilesystemType.GetFilesystem: filesystem root is a whiteout")
 		root.destroyLocked(ctx)
 		fs.vfsfs.DecRef(ctx)
-		return nil, nil, syserror.EINVAL
+		return nil, nil, linuxerr.EINVAL
 	}
 	root.mode = uint32(rootStat.Mode)
 	root.uid = rootStat.UID
diff --git a/pkg/sentry/fsimpl/proc/filesystem.go b/pkg/sentry/fsimpl/proc/filesystem.go
index ce8f55b1f..f2697c12d 100644
--- a/pkg/sentry/fsimpl/proc/filesystem.go
+++ b/pkg/sentry/fsimpl/proc/filesystem.go
@@ -21,11 +21,11 @@ import (
 
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/sentry/fsimpl/kernfs"
 	"gvisor.dev/gvisor/pkg/sentry/kernel"
 	"gvisor.dev/gvisor/pkg/sentry/kernel/auth"
 	"gvisor.dev/gvisor/pkg/sentry/vfs"
-	"gvisor.dev/gvisor/pkg/syserror"
 )
 
 const (
@@ -76,7 +76,7 @@ func (ft FilesystemType) GetFilesystem(ctx context.Context, vfsObj *vfs.VirtualF
 		maxCachedDentries, err = strconv.ParseUint(str, 10, 64)
 		if err != nil {
 			ctx.Warningf("proc.FilesystemType.GetFilesystem: invalid dentry cache limit: dentry_cache_limit=%s", str)
-			return nil, nil, syserror.EINVAL
+			return nil, nil, linuxerr.EINVAL
 		}
 	}
 
diff --git a/pkg/sentry/fsimpl/proc/task_files.go b/pkg/sentry/fsimpl/proc/task_files.go
index b294dfd6a..9187f5b11 100644
--- a/pkg/sentry/fsimpl/proc/task_files.go
+++ b/pkg/sentry/fsimpl/proc/task_files.go
@@ -21,6 +21,7 @@ import (
 
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/safemem"
 	"gvisor.dev/gvisor/pkg/sentry/fsbridge"
@@ -325,7 +326,7 @@ func (d *idMapData) Write(ctx context.Context, src usermem.IOSequence, offset in
 	// the file ..." - user_namespaces(7)
 	srclen := src.NumBytes()
 	if srclen >= hostarch.PageSize || offset != 0 {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 	b := make([]byte, srclen)
 	if _, err := src.CopyIn(ctx, b); err != nil {
@@ -345,7 +346,7 @@ func (d *idMapData) Write(ctx context.Context, src usermem.IOSequence, offset in
 	}
 	lines := bytes.SplitN(b, []byte("\n"), maxIDMapLines+1)
 	if len(lines) > maxIDMapLines {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 
 	entries := make([]auth.IDMapEntry, len(lines))
@@ -353,7 +354,7 @@ func (d *idMapData) Write(ctx context.Context, src usermem.IOSequence, offset in
 		var e auth.IDMapEntry
 		_, err := fmt.Sscan(string(l), &e.FirstID, &e.FirstParentID, &e.Length)
 		if err != nil {
-			return 0, syserror.EINVAL
+			return 0, linuxerr.EINVAL
 		}
 		entries[i] = e
 	}
@@ -461,10 +462,10 @@ func (fd *memFD) Seek(ctx context.Context, offset int64, whence int32) (int64, e
 	case linux.SEEK_CUR:
 		offset += fd.offset
 	default:
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 	if offset < 0 {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 	fd.offset = offset
 	return offset, nil
diff --git a/pkg/sentry/fsimpl/proc/tasks_files.go b/pkg/sentry/fsimpl/proc/tasks_files.go
index 045ed7a2d..2def1ca48 100644
--- a/pkg/sentry/fsimpl/proc/tasks_files.go
+++ b/pkg/sentry/fsimpl/proc/tasks_files.go
@@ -21,6 +21,7 @@ import (
 
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/sentry/fsimpl/kernfs"
 	"gvisor.dev/gvisor/pkg/sentry/kernel"
@@ -53,7 +54,7 @@ func (s *selfSymlink) Readlink(ctx context.Context, _ *vfs.Mount) (string, error
 	t := kernel.TaskFromContext(ctx)
 	if t == nil {
 		// Who is reading this link?
-		return "", syserror.EINVAL
+		return "", linuxerr.EINVAL
 	}
 	tgid := s.pidns.IDOfThreadGroup(t.ThreadGroup())
 	if tgid == 0 {
@@ -94,7 +95,7 @@ func (s *threadSelfSymlink) Readlink(ctx context.Context, _ *vfs.Mount) (string,
 	t := kernel.TaskFromContext(ctx)
 	if t == nil {
 		// Who is reading this link?
-		return "", syserror.EINVAL
+		return "", linuxerr.EINVAL
 	}
 	tgid := s.pidns.IDOfThreadGroup(t.ThreadGroup())
 	tid := s.pidns.IDOfTask(t)
diff --git a/pkg/sentry/fsimpl/proc/tasks_sys.go b/pkg/sentry/fsimpl/proc/tasks_sys.go
index 2bc98a94f..99f64a9d8 100644
--- a/pkg/sentry/fsimpl/proc/tasks_sys.go
+++ b/pkg/sentry/fsimpl/proc/tasks_sys.go
@@ -21,6 +21,7 @@ import (
 
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/sentry/fsimpl/kernfs"
 	"gvisor.dev/gvisor/pkg/sentry/inet"
@@ -28,7 +29,6 @@ import (
 	"gvisor.dev/gvisor/pkg/sentry/kernel/auth"
 	"gvisor.dev/gvisor/pkg/sentry/vfs"
 	"gvisor.dev/gvisor/pkg/sync"
-	"gvisor.dev/gvisor/pkg/syserror"
 	"gvisor.dev/gvisor/pkg/tcpip/network/ipv4"
 	"gvisor.dev/gvisor/pkg/usermem"
 )
@@ -209,7 +209,7 @@ func (d *tcpSackData) Generate(ctx context.Context, buf *bytes.Buffer) error {
 func (d *tcpSackData) Write(ctx context.Context, src usermem.IOSequence, offset int64) (int64, error) {
 	if offset != 0 {
 		// No need to handle partial writes thus far.
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 	if src.NumBytes() == 0 {
 		return 0, nil
@@ -257,7 +257,7 @@ func (d *tcpRecoveryData) Generate(ctx context.Context, buf *bytes.Buffer) error
 func (d *tcpRecoveryData) Write(ctx context.Context, src usermem.IOSequence, offset int64) (int64, error) {
 	if offset != 0 {
 		// No need to handle partial writes thus far.
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 	if src.NumBytes() == 0 {
 		return 0, nil
@@ -311,7 +311,7 @@ func (d *tcpMemData) Generate(ctx context.Context, buf *bytes.Buffer) error {
 func (d *tcpMemData) Write(ctx context.Context, src usermem.IOSequence, offset int64) (int64, error) {
 	if offset != 0 {
 		// No need to handle partial writes thus far.
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 	if src.NumBytes() == 0 {
 		return 0, nil
@@ -396,7 +396,7 @@ func (ipf *ipForwarding) Generate(ctx context.Context, buf *bytes.Buffer) error
 func (ipf *ipForwarding) Write(ctx context.Context, src usermem.IOSequence, offset int64) (int64, error) {
 	if offset != 0 {
 		// No need to handle partial writes thus far.
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 	if src.NumBytes() == 0 {
 		return 0, nil
@@ -449,7 +449,7 @@ func (pr *portRange) Generate(ctx context.Context, buf *bytes.Buffer) error {
 func (pr *portRange) Write(ctx context.Context, src usermem.IOSequence, offset int64) (int64, error) {
 	if offset != 0 {
 		// No need to handle partial writes thus far.
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 	if src.NumBytes() == 0 {
 		return 0, nil
@@ -467,7 +467,7 @@ func (pr *portRange) Write(ctx context.Context, src usermem.IOSequence, offset i
 
 	// Port numbers must be uint16s.
 	if ports[0] < 0 || ports[1] < 0 || ports[0] > math.MaxUint16 || ports[1] > math.MaxUint16 {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 
 	if err := pr.stack.SetPortRange(uint16(ports[0]), uint16(ports[1])); err != nil {
diff --git a/pkg/sentry/fsimpl/proc/yama.go b/pkg/sentry/fsimpl/proc/yama.go
index e039ec45e..7240563d7 100644
--- a/pkg/sentry/fsimpl/proc/yama.go
+++ b/pkg/sentry/fsimpl/proc/yama.go
@@ -21,11 +21,11 @@ import (
 
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/sentry/fsimpl/kernfs"
 	"gvisor.dev/gvisor/pkg/sentry/kernel"
 	"gvisor.dev/gvisor/pkg/sentry/kernel/auth"
-	"gvisor.dev/gvisor/pkg/syserror"
 	"gvisor.dev/gvisor/pkg/usermem"
 )
 
@@ -56,7 +56,7 @@ func (s *yamaPtraceScope) Generate(ctx context.Context, buf *bytes.Buffer) error
 func (s *yamaPtraceScope) Write(ctx context.Context, src usermem.IOSequence, offset int64) (int64, error) {
 	if offset != 0 {
 		// Ignore partial writes.
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 	if src.NumBytes() == 0 {
 		return 0, nil
@@ -73,7 +73,7 @@ func (s *yamaPtraceScope) Write(ctx context.Context, src usermem.IOSequence, off
 
 	// We do not support YAMA levels > YAMA_SCOPE_RELATIONAL.
 	if v < linux.YAMA_SCOPE_DISABLED || v > linux.YAMA_SCOPE_RELATIONAL {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 
 	atomic.StoreInt32(s.level, v)
diff --git a/pkg/sentry/fsimpl/sys/kcov.go b/pkg/sentry/fsimpl/sys/kcov.go
index b13f141a8..d06aea162 100644
--- a/pkg/sentry/fsimpl/sys/kcov.go
+++ b/pkg/sentry/fsimpl/sys/kcov.go
@@ -17,6 +17,7 @@ package sys
 import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
 	"gvisor.dev/gvisor/pkg/sentry/fsimpl/kernfs"
 	"gvisor.dev/gvisor/pkg/sentry/kernel"
@@ -85,7 +86,7 @@ func (fd *kcovFD) Ioctl(ctx context.Context, uio usermem.IO, args arch.SyscallAr
 	case linux.KCOV_DISABLE:
 		if arg != 0 {
 			// This arg is unused; it should be 0.
-			return 0, syserror.EINVAL
+			return 0, linuxerr.EINVAL
 		}
 		return 0, fd.kcov.DisableTrace(ctx)
 	default:
diff --git a/pkg/sentry/fsimpl/sys/sys.go b/pkg/sentry/fsimpl/sys/sys.go
index 14eb10dcd..546f54a5a 100644
--- a/pkg/sentry/fsimpl/sys/sys.go
+++ b/pkg/sentry/fsimpl/sys/sys.go
@@ -23,6 +23,7 @@ import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/context"
 	"gvisor.dev/gvisor/pkg/coverage"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/log"
 	"gvisor.dev/gvisor/pkg/sentry/fsimpl/kernfs"
 	"gvisor.dev/gvisor/pkg/sentry/kernel"
@@ -74,7 +75,7 @@ func (fsType FilesystemType) GetFilesystem(ctx context.Context, vfsObj *vfs.Virt
 		maxCachedDentries, err = strconv.ParseUint(str, 10, 64)
 		if err != nil {
 			ctx.Warningf("sys.FilesystemType.GetFilesystem: invalid dentry cache limit: dentry_cache_limit=%s", str)
-			return nil, nil, syserror.EINVAL
+			return nil, nil, linuxerr.EINVAL
 		}
 	}
 
diff --git a/pkg/sentry/fsimpl/timerfd/timerfd.go b/pkg/sentry/fsimpl/timerfd/timerfd.go
index cbb8b67c5..655a1c76a 100644
--- a/pkg/sentry/fsimpl/timerfd/timerfd.go
+++ b/pkg/sentry/fsimpl/timerfd/timerfd.go
@@ -19,6 +19,7 @@ import (
 	"sync/atomic"
 
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/hostarch"
 	ktime "gvisor.dev/gvisor/pkg/sentry/kernel/time"
 	"gvisor.dev/gvisor/pkg/sentry/vfs"
@@ -69,7 +70,7 @@ func New(ctx context.Context, vfsObj *vfs.VirtualFilesystem, clock ktime.Clock,
 func (tfd *TimerFileDescription) Read(ctx context.Context, dst usermem.IOSequence, opts vfs.ReadOptions) (int64, error) {
 	const sizeofUint64 = 8
 	if dst.NumBytes() < sizeofUint64 {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 	if val := atomic.SwapUint64(&tfd.val, 0); val != 0 {
 		var buf [sizeofUint64]byte
diff --git a/pkg/sentry/fsimpl/tmpfs/directory.go b/pkg/sentry/fsimpl/tmpfs/directory.go
index e8d256495..c25494c0b 100644
--- a/pkg/sentry/fsimpl/tmpfs/directory.go
+++ b/pkg/sentry/fsimpl/tmpfs/directory.go
@@ -19,10 +19,10 @@ import (
 
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/sentry/kernel/auth"
 	"gvisor.dev/gvisor/pkg/sentry/vfs"
 	"gvisor.dev/gvisor/pkg/sync"
-	"gvisor.dev/gvisor/pkg/syserror"
 )
 
 // +stateify savable
@@ -196,10 +196,10 @@ func (fd *directoryFD) Seek(ctx context.Context, offset int64, whence int32) (in
 	case linux.SEEK_CUR:
 		offset += fd.off
 	default:
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 	if offset < 0 {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 
 	// If the offset isn't changing (e.g. due to lseek(0, SEEK_CUR)), don't
diff --git a/pkg/sentry/fsimpl/tmpfs/filesystem.go b/pkg/sentry/fsimpl/tmpfs/filesystem.go
index f0f4297ef..590f7118a 100644
--- a/pkg/sentry/fsimpl/tmpfs/filesystem.go
+++ b/pkg/sentry/fsimpl/tmpfs/filesystem.go
@@ -20,6 +20,7 @@ import (
 
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/fspath"
 	"gvisor.dev/gvisor/pkg/sentry/fsmetric"
 	"gvisor.dev/gvisor/pkg/sentry/kernel/auth"
@@ -300,7 +301,7 @@ func (fs *filesystem) MknodAt(ctx context.Context, rp *vfs.ResolvingPath, opts v
 		case linux.S_IFSOCK:
 			childInode = fs.newSocketFile(creds.EffectiveKUID, creds.EffectiveKGID, opts.Mode, opts.Endpoint, parentDir)
 		default:
-			return syserror.EINVAL
+			return linuxerr.EINVAL
 		}
 		child := fs.newDentry(childInode)
 		parentDir.insertChildLocked(child, name)
@@ -488,7 +489,7 @@ func (fs *filesystem) ReadlinkAt(ctx context.Context, rp *vfs.ResolvingPath) (st
 	}
 	symlink, ok := d.inode.impl.(*symlink)
 	if !ok {
-		return "", syserror.EINVAL
+		return "", linuxerr.EINVAL
 	}
 	symlink.inode.touchAtime(rp.Mount())
 	return symlink.target, nil
@@ -506,7 +507,7 @@ func (fs *filesystem) RenameAt(ctx context.Context, rp *vfs.ResolvingPath, oldPa
 
 	if opts.Flags&^linux.RENAME_NOREPLACE != 0 {
 		// TODO(b/145974740): Support other renameat2 flags.
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 
 	newName := rp.Component()
@@ -541,7 +542,7 @@ func (fs *filesystem) RenameAt(ctx context.Context, rp *vfs.ResolvingPath, oldPa
 	// mounted filesystem.
 	if renamed.inode.isDir() {
 		if renamed == &newParentDir.dentry || genericIsAncestorDentry(renamed, &newParentDir.dentry) {
-			return syserror.EINVAL
+			return linuxerr.EINVAL
 		}
 		if oldParentDir != newParentDir {
 			// Writability is needed to change renamed's "..".
@@ -646,7 +647,7 @@ func (fs *filesystem) RmdirAt(ctx context.Context, rp *vfs.ResolvingPath) error
 	}
 	name := rp.Component()
 	if name == "." {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	if name == ".." {
 		return syserror.ENOTEMPTY
diff --git a/pkg/sentry/fsimpl/tmpfs/regular_file.go b/pkg/sentry/fsimpl/tmpfs/regular_file.go
index c45bddff6..0bc1911d9 100644
--- a/pkg/sentry/fsimpl/tmpfs/regular_file.go
+++ b/pkg/sentry/fsimpl/tmpfs/regular_file.go
@@ -22,6 +22,7 @@ import (
 
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/safemem"
 	"gvisor.dev/gvisor/pkg/sentry/fs"
@@ -366,7 +367,7 @@ func (fd *regularFileFD) PRead(ctx context.Context, dst usermem.IOSequence, offs
 	fsmetric.TmpfsReads.Increment()
 
 	if offset < 0 {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 
 	// Check that flags are supported. RWF_DSYNC/RWF_SYNC can be ignored since
@@ -407,7 +408,7 @@ func (fd *regularFileFD) PWrite(ctx context.Context, src usermem.IOSequence, off
 // final offset should be ignored by PWrite.
 func (fd *regularFileFD) pwrite(ctx context.Context, src usermem.IOSequence, offset int64, opts vfs.WriteOptions) (written, finalOff int64, err error) {
 	if offset < 0 {
-		return 0, offset, syserror.EINVAL
+		return 0, offset, linuxerr.EINVAL
 	}
 
 	// Check that flags are supported. RWF_DSYNC/RWF_SYNC can be ignored since
@@ -432,7 +433,7 @@ func (fd *regularFileFD) pwrite(ctx context.Context, src usermem.IOSequence, off
 	}
 	if end := offset + srclen; end < offset {
 		// Overflow.
-		return 0, offset, syserror.EINVAL
+		return 0, offset, linuxerr.EINVAL
 	}
 
 	srclen, err = vfs.CheckLimit(ctx, offset, srclen)
@@ -476,10 +477,10 @@ func (fd *regularFileFD) Seek(ctx context.Context, offset int64, whence int32) (
 	case linux.SEEK_END:
 		offset += int64(atomic.LoadUint64(&fd.inode().impl.(*regularFile).size))
 	default:
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 	if offset < 0 {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 	fd.off = offset
 	return offset, nil
@@ -684,7 +685,7 @@ exitLoop:
 func GetSeals(fd *vfs.FileDescription) (uint32, error) {
 	f, ok := fd.Impl().(*regularFileFD)
 	if !ok {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 	rf := f.inode().impl.(*regularFile)
 	rf.dataMu.RLock()
@@ -696,7 +697,7 @@ func GetSeals(fd *vfs.FileDescription) (uint32, error) {
 func AddSeals(fd *vfs.FileDescription, val uint32) error {
 	f, ok := fd.Impl().(*regularFileFD)
 	if !ok {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	rf := f.inode().impl.(*regularFile)
 	rf.mapsMu.Lock()
diff --git a/pkg/sentry/fsimpl/tmpfs/tmpfs.go b/pkg/sentry/fsimpl/tmpfs/tmpfs.go
index 6b4367c42..bc40aad0d 100644
--- a/pkg/sentry/fsimpl/tmpfs/tmpfs.go
+++ b/pkg/sentry/fsimpl/tmpfs/tmpfs.go
@@ -36,6 +36,7 @@ import (
 
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/sentry/kernel/auth"
 	"gvisor.dev/gvisor/pkg/sentry/kernel/time"
@@ -138,7 +139,7 @@ func (fstype FilesystemType) GetFilesystem(ctx context.Context, vfsObj *vfs.Virt
 		mode, err := strconv.ParseUint(modeStr, 8, 32)
 		if err != nil {
 			ctx.Warningf("tmpfs.FilesystemType.GetFilesystem: invalid mode: %q", modeStr)
-			return nil, nil, syserror.EINVAL
+			return nil, nil, linuxerr.EINVAL
 		}
 		rootMode = linux.FileMode(mode & 07777)
 	}
@@ -149,12 +150,12 @@ func (fstype FilesystemType) GetFilesystem(ctx context.Context, vfsObj *vfs.Virt
 		uid, err := strconv.ParseUint(uidStr, 10, 32)
 		if err != nil {
 			ctx.Warningf("tmpfs.FilesystemType.GetFilesystem: invalid uid: %q", uidStr)
-			return nil, nil, syserror.EINVAL
+			return nil, nil, linuxerr.EINVAL
 		}
 		kuid := creds.UserNamespace.MapToKUID(auth.UID(uid))
 		if !kuid.Ok() {
 			ctx.Warningf("tmpfs.FilesystemType.GetFilesystem: unmapped uid: %d", uid)
-			return nil, nil, syserror.EINVAL
+			return nil, nil, linuxerr.EINVAL
 		}
 		rootKUID = kuid
 	}
@@ -165,18 +166,18 @@ func (fstype FilesystemType) GetFilesystem(ctx context.Context, vfsObj *vfs.Virt
 		gid, err := strconv.ParseUint(gidStr, 10, 32)
 		if err != nil {
 			ctx.Warningf("tmpfs.FilesystemType.GetFilesystem: invalid gid: %q", gidStr)
-			return nil, nil, syserror.EINVAL
+			return nil, nil, linuxerr.EINVAL
 		}
 		kgid := creds.UserNamespace.MapToKGID(auth.GID(gid))
 		if !kgid.Ok() {
 			ctx.Warningf("tmpfs.FilesystemType.GetFilesystem: unmapped gid: %d", gid)
-			return nil, nil, syserror.EINVAL
+			return nil, nil, linuxerr.EINVAL
 		}
 		rootKGID = kgid
 	}
 	if len(mopts) != 0 {
 		ctx.Warningf("tmpfs.FilesystemType.GetFilesystem: unknown options: %v", mopts)
-		return nil, nil, syserror.EINVAL
+		return nil, nil, linuxerr.EINVAL
 	}
 
 	devMinor, err := vfsObj.GetAnonBlockDevMinor()
@@ -557,7 +558,7 @@ func (i *inode) setStat(ctx context.Context, creds *auth.Credentials, opts *vfs.
 		case *directory:
 			return syserror.EISDIR
 		default:
-			return syserror.EINVAL
+			return linuxerr.EINVAL
 		}
 	}
 	if mask&linux.STATX_UID != 0 {
diff --git a/pkg/sentry/fsimpl/verity/verity.go b/pkg/sentry/fsimpl/verity/verity.go
index c5f59d851..ffc8f019d 100644
--- a/pkg/sentry/fsimpl/verity/verity.go
+++ b/pkg/sentry/fsimpl/verity/verity.go
@@ -252,7 +252,7 @@ func (fstype FilesystemType) GetFilesystem(ctx context.Context, vfsObj *vfs.Virt
 		hash, err := hex.DecodeString(encodedRootHash)
 		if err != nil {
 			ctx.Warningf("verity.FilesystemType.GetFilesystem: Failed to decode root hash: %v", err)
-			return nil, nil, syserror.EINVAL
+			return nil, nil, linuxerr.EINVAL
 		}
 		rootHash = hash
 	}
@@ -270,19 +270,19 @@ func (fstype FilesystemType) GetFilesystem(ctx context.Context, vfsObj *vfs.Virt
 	// Check for unparsed options.
 	if len(mopts) != 0 {
 		ctx.Warningf("verity.FilesystemType.GetFilesystem: unknown options: %v", mopts)
-		return nil, nil, syserror.EINVAL
+		return nil, nil, linuxerr.EINVAL
 	}
 
 	// Handle internal options.
 	iopts, ok := opts.InternalData.(InternalFilesystemOptions)
 	if len(lowerPathname) == 0 && !ok {
 		ctx.Warningf("verity.FilesystemType.GetFilesystem: missing verity configs")
-		return nil, nil, syserror.EINVAL
+		return nil, nil, linuxerr.EINVAL
 	}
 	if len(lowerPathname) != 0 {
 		if ok {
 			ctx.Warningf("verity.FilesystemType.GetFilesystem: unexpected verity configs with specified lower path")
-			return nil, nil, syserror.EINVAL
+			return nil, nil, linuxerr.EINVAL
 		}
 		iopts = InternalFilesystemOptions{
 			AllowRuntimeEnable: len(rootHash) == 0,
@@ -301,7 +301,7 @@ func (fstype FilesystemType) GetFilesystem(ctx context.Context, vfsObj *vfs.Virt
 		lowerPath := fspath.Parse(lowerPathname)
 		if !lowerPath.Absolute {
 			ctx.Infof("verity.FilesystemType.GetFilesystem: lower_path %q must be absolute", lowerPathname)
-			return nil, nil, syserror.EINVAL
+			return nil, nil, linuxerr.EINVAL
 		}
 		var err error
 		mountedLowerVD, err = vfsObj.GetDentryAt(ctx, creds, &vfs.PathOperation{
@@ -440,7 +440,7 @@ func (fstype FilesystemType) GetFilesystem(ctx context.Context, vfsObj *vfs.Virt
 
 	if !d.isDir() {
 		ctx.Warningf("verity root must be a directory")
-		return nil, nil, syserror.EINVAL
+		return nil, nil, linuxerr.EINVAL
 	}
 
 	if !fs.allowRuntimeEnable {
@@ -922,14 +922,14 @@ func (fd *fileDescription) Seek(ctx context.Context, offset int64, whence int32)
 	case linux.SEEK_END:
 		n = int64(fd.d.size)
 	default:
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 	if offset > math.MaxInt64-n {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 	offset += n
 	if offset < 0 {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 	fd.off = offset
 	return offset, nil
@@ -1008,7 +1008,7 @@ func (fd *fileDescription) generateMerkleLocked(ctx context.Context) ([]byte, ui
 	default:
 		// TODO(b/167728857): Investigate whether and how we should
 		// enable other types of file.
-		return nil, 0, syserror.EINVAL
+		return nil, 0, linuxerr.EINVAL
 	}
 	hash, err := merkletree.Generate(params)
 	return hash, uint64(params.Size), err
@@ -1126,7 +1126,7 @@ func (fd *fileDescription) enableVerity(ctx context.Context) (uintptr, error) {
 func (fd *fileDescription) measureVerity(ctx context.Context, verityDigest hostarch.Addr) (uintptr, error) {
 	t := kernel.TaskFromContext(ctx)
 	if t == nil {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 	var metadata linux.DigestMetadata
 
@@ -1179,7 +1179,7 @@ func (fd *fileDescription) verityFlags(ctx context.Context, flags hostarch.Addr)
 
 	t := kernel.TaskFromContext(ctx)
 	if t == nil {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 	_, err := primitive.CopyInt32Out(t, flags, f)
 	return 0, err
@@ -1434,7 +1434,7 @@ func (r *mmapReadSeeker) ReadAt(p []byte, off int64) (int, error) {
 	// mapped region.
 	readOffset := off - int64(r.Offset)
 	if readOffset < 0 {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 	bs.DropFirst64(uint64(readOffset))
 	view := bs.TakeFirst64(uint64(len(p)))
diff --git a/pkg/sentry/kernel/auth/credentials.go b/pkg/sentry/kernel/auth/credentials.go
index 3325fedcb..32c344399 100644
--- a/pkg/sentry/kernel/auth/credentials.go
+++ b/pkg/sentry/kernel/auth/credentials.go
@@ -16,6 +16,7 @@ package auth
 
 import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/syserror"
 )
 
@@ -203,7 +204,7 @@ func (c *Credentials) UseUID(uid UID) (KUID, error) {
 	// uid must be mapped.
 	kuid := c.UserNamespace.MapToKUID(uid)
 	if !kuid.Ok() {
-		return NoID, syserror.EINVAL
+		return NoID, linuxerr.EINVAL
 	}
 	// If c has CAP_SETUID, then it can use any UID in its user namespace.
 	if c.HasCapability(linux.CAP_SETUID) {
@@ -222,7 +223,7 @@ func (c *Credentials) UseUID(uid UID) (KUID, error) {
 func (c *Credentials) UseGID(gid GID) (KGID, error) {
 	kgid := c.UserNamespace.MapToKGID(gid)
 	if !kgid.Ok() {
-		return NoID, syserror.EINVAL
+		return NoID, linuxerr.EINVAL
 	}
 	if c.HasCapability(linux.CAP_SETGID) {
 		return kgid, nil
@@ -239,7 +240,7 @@ func (c *Credentials) UseGID(gid GID) (KGID, error) {
 func (c *Credentials) SetUID(uid UID) error {
 	kuid := c.UserNamespace.MapToKUID(uid)
 	if !kuid.Ok() {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	c.RealKUID = kuid
 	c.EffectiveKUID = kuid
@@ -253,7 +254,7 @@ func (c *Credentials) SetUID(uid UID) error {
 func (c *Credentials) SetGID(gid GID) error {
 	kgid := c.UserNamespace.MapToKGID(gid)
 	if !kgid.Ok() {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	c.RealKGID = kgid
 	c.EffectiveKGID = kgid
diff --git a/pkg/sentry/kernel/auth/id_map.go b/pkg/sentry/kernel/auth/id_map.go
index 28cbe159d..955b6d40b 100644
--- a/pkg/sentry/kernel/auth/id_map.go
+++ b/pkg/sentry/kernel/auth/id_map.go
@@ -17,6 +17,7 @@ package auth
 import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/syserror"
 )
 
@@ -110,7 +111,7 @@ func (ns *UserNamespace) SetUIDMap(ctx context.Context, entries []IDMapEntry) er
 	}
 	// "At least one line must be written to the file."
 	if len(entries) == 0 {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	// """
 	// In order for a process to write to the /proc/[pid]/uid_map
@@ -170,11 +171,11 @@ func (ns *UserNamespace) trySetUIDMap(entries []IDMapEntry) error {
 		// checks for NoID.
 		lastID := e.FirstID + e.Length
 		if lastID <= e.FirstID {
-			return syserror.EINVAL
+			return linuxerr.EINVAL
 		}
 		lastParentID := e.FirstParentID + e.Length
 		if lastParentID <= e.FirstParentID {
-			return syserror.EINVAL
+			return linuxerr.EINVAL
 		}
 		// "3. The mapped user IDs (group IDs) must in turn have a mapping in
 		// the parent user namespace."
@@ -186,10 +187,10 @@ func (ns *UserNamespace) trySetUIDMap(entries []IDMapEntry) error {
 		}
 		// If either of these Adds fail, we have an overlapping range.
 		if !ns.uidMapFromParent.Add(idMapRange{e.FirstParentID, lastParentID}, e.FirstID) {
-			return syserror.EINVAL
+			return linuxerr.EINVAL
 		}
 		if !ns.uidMapToParent.Add(idMapRange{e.FirstID, lastID}, e.FirstParentID) {
-			return syserror.EINVAL
+			return linuxerr.EINVAL
 		}
 	}
 	return nil
@@ -205,7 +206,7 @@ func (ns *UserNamespace) SetGIDMap(ctx context.Context, entries []IDMapEntry) er
 		return syserror.EPERM
 	}
 	if len(entries) == 0 {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	if !c.HasCapabilityIn(linux.CAP_SETGID, ns) {
 		return syserror.EPERM
@@ -239,20 +240,20 @@ func (ns *UserNamespace) trySetGIDMap(entries []IDMapEntry) error {
 	for _, e := range entries {
 		lastID := e.FirstID + e.Length
 		if lastID <= e.FirstID {
-			return syserror.EINVAL
+			return linuxerr.EINVAL
 		}
 		lastParentID := e.FirstParentID + e.Length
 		if lastParentID <= e.FirstParentID {
-			return syserror.EINVAL
+			return linuxerr.EINVAL
 		}
 		if !ns.parent.allIDsMapped(&ns.parent.gidMapToParent, e.FirstParentID, lastParentID) {
 			return syserror.EPERM
 		}
 		if !ns.gidMapFromParent.Add(idMapRange{e.FirstParentID, lastParentID}, e.FirstID) {
-			return syserror.EINVAL
+			return linuxerr.EINVAL
 		}
 		if !ns.gidMapToParent.Add(idMapRange{e.FirstID, lastID}, e.FirstParentID) {
-			return syserror.EINVAL
+			return linuxerr.EINVAL
 		}
 	}
 	return nil
diff --git a/pkg/sentry/kernel/fasync/fasync.go b/pkg/sentry/kernel/fasync/fasync.go
index 5d584dc45..473987a79 100644
--- a/pkg/sentry/kernel/fasync/fasync.go
+++ b/pkg/sentry/kernel/fasync/fasync.go
@@ -17,12 +17,12 @@ package fasync
 
 import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/sentry/fs"
 	"gvisor.dev/gvisor/pkg/sentry/kernel"
 	"gvisor.dev/gvisor/pkg/sentry/kernel/auth"
 	"gvisor.dev/gvisor/pkg/sentry/vfs"
 	"gvisor.dev/gvisor/pkg/sync"
-	"gvisor.dev/gvisor/pkg/syserror"
 	"gvisor.dev/gvisor/pkg/waiter"
 )
 
@@ -248,7 +248,7 @@ func (a *FileAsync) Signal() linux.Signal {
 // to send SIGIO.
 func (a *FileAsync) SetSignal(signal linux.Signal) error {
 	if signal != 0 && !signal.IsValid() {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	a.mu.Lock()
 	defer a.mu.Unlock()
diff --git a/pkg/sentry/kernel/futex/futex.go b/pkg/sentry/kernel/futex/futex.go
index 0427cf3f4..5c64ce11e 100644
--- a/pkg/sentry/kernel/futex/futex.go
+++ b/pkg/sentry/kernel/futex/futex.go
@@ -20,6 +20,7 @@ package futex
 import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/sentry/memmap"
 	"gvisor.dev/gvisor/pkg/sync"
@@ -332,7 +333,7 @@ func getKey(t Target, addr hostarch.Addr, private bool) (Key, error) {
 	// Ensure the address is aligned.
 	// It must be a DWORD boundary.
 	if addr&0x3 != 0 {
-		return Key{}, syserror.EINVAL
+		return Key{}, linuxerr.EINVAL
 	}
 	if private {
 		return Key{Kind: KindPrivate, Offset: uint64(addr)}, nil
@@ -790,7 +791,7 @@ func (m *Manager) unlockPILocked(t Target, addr hostarch.Addr, tid uint32, b *bu
 		return err
 	}
 	if prev != cur {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 
 	b.wakeWaiterLocked(next)
diff --git a/pkg/sentry/kernel/kcov.go b/pkg/sentry/kernel/kcov.go
index 4b943106b..941cc373f 100644
--- a/pkg/sentry/kernel/kcov.go
+++ b/pkg/sentry/kernel/kcov.go
@@ -22,6 +22,7 @@ import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/context"
 	"gvisor.dev/gvisor/pkg/coverage"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/safemem"
 	"gvisor.dev/gvisor/pkg/sentry/memmap"
@@ -131,13 +132,13 @@ func (kcov *Kcov) InitTrace(size uint64) error {
 	// To simplify all the logic around mapping, we require that the length of the
 	// shared region is a multiple of the system page size.
 	if (8*size)&(hostarch.PageSize-1) != 0 {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 
 	// We need space for at least two uint64s to hold current position and a
 	// single PC.
 	if size < 2 || size > kcovAreaSizeMax {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 
 	kcov.size = size
@@ -157,7 +158,7 @@ func (kcov *Kcov) EnableTrace(ctx context.Context, traceKind uint8) error {
 
 	// KCOV_ENABLE must be preceded by KCOV_INIT_TRACE and an mmap call.
 	if kcov.mode != linux.KCOV_MODE_INIT || kcov.mappable == nil {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 
 	switch traceKind {
@@ -167,7 +168,7 @@ func (kcov *Kcov) EnableTrace(ctx context.Context, traceKind uint8) error {
 		// We do not support KCOV_MODE_TRACE_CMP.
 		return syserror.ENOTSUP
 	default:
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 
 	if kcov.owningTask != nil && kcov.owningTask != t {
@@ -195,7 +196,7 @@ func (kcov *Kcov) DisableTrace(ctx context.Context) error {
 	}
 
 	if t != kcov.owningTask {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	kcov.mode = linux.KCOV_MODE_INIT
 	kcov.owningTask = nil
@@ -237,7 +238,7 @@ func (kcov *Kcov) ConfigureMMap(ctx context.Context, opts *memmap.MMapOpts) erro
 	defer kcov.mu.Unlock()
 
 	if kcov.mode != linux.KCOV_MODE_INIT {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 
 	if kcov.mappable == nil {
diff --git a/pkg/sentry/kernel/pipe/node.go b/pkg/sentry/kernel/pipe/node.go
index 6497dc4ba..2321d26dc 100644
--- a/pkg/sentry/kernel/pipe/node.go
+++ b/pkg/sentry/kernel/pipe/node.go
@@ -17,6 +17,7 @@ package pipe
 import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/sentry/fs"
 	"gvisor.dev/gvisor/pkg/sentry/fs/fsutil"
 	"gvisor.dev/gvisor/pkg/sync"
@@ -130,7 +131,7 @@ func (i *inodeOperations) GetFile(ctx context.Context, d *fs.Dirent, flags fs.Fi
 		return rw, nil
 
 	default:
-		return nil, syserror.EINVAL
+		return nil, linuxerr.EINVAL
 	}
 }
 
diff --git a/pkg/sentry/kernel/pipe/pipe.go b/pkg/sentry/kernel/pipe/pipe.go
index 06769931a..979ea10bf 100644
--- a/pkg/sentry/kernel/pipe/pipe.go
+++ b/pkg/sentry/kernel/pipe/pipe.go
@@ -22,6 +22,7 @@ import (
 
 	"golang.org/x/sys/unix"
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/safemem"
 	"gvisor.dev/gvisor/pkg/sentry/fs"
@@ -428,7 +429,7 @@ func (p *Pipe) FifoSize(context.Context, *fs.File) (int64, error) {
 // SetFifoSize implements fs.FifoSizer.SetFifoSize.
 func (p *Pipe) SetFifoSize(size int64) (int64, error) {
 	if size < 0 {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 	if size < MinimumPipeSize {
 		size = MinimumPipeSize // Per spec.
diff --git a/pkg/sentry/kernel/pipe/vfs.go b/pkg/sentry/kernel/pipe/vfs.go
index 95b948edb..623375417 100644
--- a/pkg/sentry/kernel/pipe/vfs.go
+++ b/pkg/sentry/kernel/pipe/vfs.go
@@ -17,6 +17,7 @@ package pipe
 import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/safemem"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
@@ -90,7 +91,7 @@ func (vp *VFSPipe) Open(ctx context.Context, mnt *vfs.Mount, vfsd *vfs.Dentry, s
 	readable := vfs.MayReadFileWithOpenFlags(statusFlags)
 	writable := vfs.MayWriteFileWithOpenFlags(statusFlags)
 	if !readable && !writable {
-		return nil, syserror.EINVAL
+		return nil, linuxerr.EINVAL
 	}
 
 	fd, err := vp.newFD(mnt, vfsd, statusFlags, locks)
@@ -415,7 +416,7 @@ func Tee(ctx context.Context, dst, src *VFSPipeFD, count int64) (int64, error) {
 // Preconditions: count > 0.
 func spliceOrTee(ctx context.Context, dst, src *VFSPipeFD, count int64, removeFromSrc bool) (int64, error) {
 	if dst.pipe == src.pipe {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 
 	lockTwoPipes(dst.pipe, src.pipe)
diff --git a/pkg/sentry/kernel/posixtimer.go b/pkg/sentry/kernel/posixtimer.go
index d801a3d83..049cc07df 100644
--- a/pkg/sentry/kernel/posixtimer.go
+++ b/pkg/sentry/kernel/posixtimer.go
@@ -18,6 +18,7 @@ import (
 	"math"
 
 	"gvisor.dev/gvisor/pkg/abi/linux"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	ktime "gvisor.dev/gvisor/pkg/sentry/kernel/time"
 	"gvisor.dev/gvisor/pkg/syserror"
 )
@@ -214,16 +215,16 @@ func (t *Task) IntervalTimerCreate(c ktime.Clock, sigev *linux.Sigevent) (linux.
 		target, ok := t.tg.pidns.tasks[ThreadID(sigev.Tid)]
 		t.tg.pidns.owner.mu.RUnlock()
 		if !ok || target.tg != t.tg {
-			return 0, syserror.EINVAL
+			return 0, linuxerr.EINVAL
 		}
 		it.target = target
 	default:
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 	if sigev.Notify != linux.SIGEV_NONE {
 		it.signo = linux.Signal(sigev.Signo)
 		if !it.signo.IsValid() {
-			return 0, syserror.EINVAL
+			return 0, linuxerr.EINVAL
 		}
 	}
 	it.timer = ktime.NewTimer(c, it)
@@ -238,7 +239,7 @@ func (t *Task) IntervalTimerDelete(id linux.TimerID) error {
 	defer t.tg.timerMu.Unlock()
 	it := t.tg.timers[id]
 	if it == nil {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	delete(t.tg.timers, id)
 	it.DestroyTimer()
@@ -251,7 +252,7 @@ func (t *Task) IntervalTimerSettime(id linux.TimerID, its linux.Itimerspec, abs
 	defer t.tg.timerMu.Unlock()
 	it := t.tg.timers[id]
 	if it == nil {
-		return linux.Itimerspec{}, syserror.EINVAL
+		return linux.Itimerspec{}, linuxerr.EINVAL
 	}
 
 	newS, err := ktime.SettingFromItimerspec(its, abs, it.timer.Clock())
@@ -269,7 +270,7 @@ func (t *Task) IntervalTimerGettime(id linux.TimerID) (linux.Itimerspec, error)
 	defer t.tg.timerMu.Unlock()
 	it := t.tg.timers[id]
 	if it == nil {
-		return linux.Itimerspec{}, syserror.EINVAL
+		return linux.Itimerspec{}, linuxerr.EINVAL
 	}
 
 	tm, s := it.timer.Get()
@@ -285,7 +286,7 @@ func (t *Task) IntervalTimerGetoverrun(id linux.TimerID) (int32, error) {
 	defer t.tg.timerMu.Unlock()
 	it := t.tg.timers[id]
 	if it == nil {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 	// By timer_create(2) invariant, either it.target == nil (in which case
 	// it.overrunLast is immutably 0) or t.tg == it.target.tg; and the fact
diff --git a/pkg/sentry/kernel/ptrace.go b/pkg/sentry/kernel/ptrace.go
index 20563f02a..1c6100efe 100644
--- a/pkg/sentry/kernel/ptrace.go
+++ b/pkg/sentry/kernel/ptrace.go
@@ -19,6 +19,7 @@ import (
 	"sync/atomic"
 
 	"gvisor.dev/gvisor/pkg/abi/linux"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/marshal/primitive"
 	"gvisor.dev/gvisor/pkg/sentry/mm"
@@ -994,7 +995,7 @@ func (t *Task) ptraceSetOptionsLocked(opts uintptr) error {
 		linux.PTRACE_O_TRACEVFORK |
 		linux.PTRACE_O_TRACEVFORKDONE)
 	if opts&^valid != 0 {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	t.ptraceOpts = ptraceOptions{
 		ExitKill:       opts&linux.PTRACE_O_EXITKILL != 0,
@@ -1221,7 +1222,7 @@ func (t *Task) Ptrace(req int64, pid ThreadID, addr, data hostarch.Addr) error {
 		t.tg.pidns.owner.mu.RLock()
 		defer t.tg.pidns.owner.mu.RUnlock()
 		if target.ptraceSiginfo == nil {
-			return syserror.EINVAL
+			return linuxerr.EINVAL
 		}
 		_, err := target.ptraceSiginfo.CopyOut(t, data)
 		return err
@@ -1234,14 +1235,14 @@ func (t *Task) Ptrace(req int64, pid ThreadID, addr, data hostarch.Addr) error {
 		t.tg.pidns.owner.mu.RLock()
 		defer t.tg.pidns.owner.mu.RUnlock()
 		if target.ptraceSiginfo == nil {
-			return syserror.EINVAL
+			return linuxerr.EINVAL
 		}
 		target.ptraceSiginfo = &info
 		return nil
 
 	case linux.PTRACE_GETSIGMASK:
 		if addr != linux.SignalSetSize {
-			return syserror.EINVAL
+			return linuxerr.EINVAL
 		}
 		mask := target.SignalMask()
 		_, err := mask.CopyOut(t, data)
@@ -1249,7 +1250,7 @@ func (t *Task) Ptrace(req int64, pid ThreadID, addr, data hostarch.Addr) error {
 
 	case linux.PTRACE_SETSIGMASK:
 		if addr != linux.SignalSetSize {
-			return syserror.EINVAL
+			return linuxerr.EINVAL
 		}
 		var mask linux.SignalSet
 		if _, err := mask.CopyIn(t, data); err != nil {
diff --git a/pkg/sentry/kernel/rseq.go b/pkg/sentry/kernel/rseq.go
index 4bc5bca44..2344565cd 100644
--- a/pkg/sentry/kernel/rseq.go
+++ b/pkg/sentry/kernel/rseq.go
@@ -18,6 +18,7 @@ import (
 	"fmt"
 
 	"gvisor.dev/gvisor/pkg/abi/linux"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/sentry/hostcpu"
 	"gvisor.dev/gvisor/pkg/syserror"
@@ -59,20 +60,20 @@ func (t *Task) RSeqAvailable() bool {
 func (t *Task) SetRSeq(addr hostarch.Addr, length, signature uint32) error {
 	if t.rseqAddr != 0 {
 		if t.rseqAddr != addr {
-			return syserror.EINVAL
+			return linuxerr.EINVAL
 		}
 		if t.rseqSignature != signature {
-			return syserror.EINVAL
+			return linuxerr.EINVAL
 		}
 		return syserror.EBUSY
 	}
 
 	// rseq must be aligned and correctly sized.
 	if addr&(linux.AlignOfRSeq-1) != 0 {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	if length != linux.SizeOfRSeq {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	if _, ok := t.MemoryManager().CheckIORange(addr, linux.SizeOfRSeq); !ok {
 		return syserror.EFAULT
@@ -103,13 +104,13 @@ func (t *Task) SetRSeq(addr hostarch.Addr, length, signature uint32) error {
 // Preconditions: The caller must be running on the task goroutine.
 func (t *Task) ClearRSeq(addr hostarch.Addr, length, signature uint32) error {
 	if t.rseqAddr == 0 {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	if t.rseqAddr != addr {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	if length != linux.SizeOfRSeq {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	if t.rseqSignature != signature {
 		return syserror.EPERM
@@ -152,10 +153,10 @@ func (t *Task) SetOldRSeqCriticalRegion(r OldRSeqCriticalRegion) error {
 		return nil
 	}
 	if r.CriticalSection.Start >= r.CriticalSection.End {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	if r.CriticalSection.Contains(r.Restart) {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	// TODO(jamieliu): check that r.CriticalSection and r.Restart are in
 	// the application address range, for consistency with Linux.
@@ -187,7 +188,7 @@ func (t *Task) SetOldRSeqCPUAddr(addr hostarch.Addr) error {
 	// unfortunate, but unlikely in a correct program.
 	if err := t.rseqUpdateCPU(); err != nil {
 		t.oldRSeqCPUAddr = 0
-		return syserror.EINVAL // yes, EINVAL, not err or EFAULT
+		return linuxerr.EINVAL // yes, EINVAL, not err or EFAULT
 	}
 	return nil
 }
diff --git a/pkg/sentry/kernel/semaphore/semaphore.go b/pkg/sentry/kernel/semaphore/semaphore.go
index 47bb66b42..2dbc8353a 100644
--- a/pkg/sentry/kernel/semaphore/semaphore.go
+++ b/pkg/sentry/kernel/semaphore/semaphore.go
@@ -20,6 +20,7 @@ import (
 
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/log"
 	"gvisor.dev/gvisor/pkg/sentry/fs"
 	"gvisor.dev/gvisor/pkg/sentry/kernel/auth"
@@ -127,7 +128,7 @@ func NewRegistry(userNS *auth.UserNamespace) *Registry {
 // exists.
 func (r *Registry) FindOrCreate(ctx context.Context, key, nsems int32, mode linux.FileMode, private, create, exclusive bool) (*Set, error) {
 	if nsems < 0 || nsems > semsMax {
-		return nil, syserror.EINVAL
+		return nil, linuxerr.EINVAL
 	}
 
 	r.mu.Lock()
@@ -147,7 +148,7 @@ func (r *Registry) FindOrCreate(ctx context.Context, key, nsems int32, mode linu
 
 			// Validate parameters.
 			if nsems > int32(set.Size()) {
-				return nil, syserror.EINVAL
+				return nil, linuxerr.EINVAL
 			}
 			if create && exclusive {
 				return nil, syserror.EEXIST
@@ -163,7 +164,7 @@ func (r *Registry) FindOrCreate(ctx context.Context, key, nsems int32, mode linu
 
 	// Zero is only valid if an existing set is found.
 	if nsems == 0 {
-		return nil, syserror.EINVAL
+		return nil, linuxerr.EINVAL
 	}
 
 	// Apply system limits.
@@ -238,7 +239,7 @@ func (r *Registry) RemoveID(id int32, creds *auth.Credentials) error {
 
 	set := r.semaphores[id]
 	if set == nil {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	index, found := r.findIndexByID(id)
 	if !found {
diff --git a/pkg/sentry/kernel/shm/shm.go b/pkg/sentry/kernel/shm/shm.go
index a73f1bdca..7a6e91004 100644
--- a/pkg/sentry/kernel/shm/shm.go
+++ b/pkg/sentry/kernel/shm/shm.go
@@ -38,6 +38,7 @@ import (
 
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/log"
 	"gvisor.dev/gvisor/pkg/sentry/fs"
@@ -145,7 +146,7 @@ func (r *Registry) FindOrCreate(ctx context.Context, pid int32, key Key, size ui
 		//
 		// Note that 'private' always implies the creation of a new segment
 		// whether IPC_CREAT is specified or not.
-		return nil, syserror.EINVAL
+		return nil, linuxerr.EINVAL
 	}
 
 	r.mu.Lock()
@@ -175,7 +176,7 @@ func (r *Registry) FindOrCreate(ctx context.Context, pid int32, key Key, size ui
 			if size > shm.size {
 				// "A segment for the given key exists, but size is greater than
 				// the size of that segment." - man shmget(2)
-				return nil, syserror.EINVAL
+				return nil, linuxerr.EINVAL
 			}
 
 			if create && exclusive {
@@ -200,7 +201,7 @@ func (r *Registry) FindOrCreate(ctx context.Context, pid int32, key Key, size ui
 	if val, ok := hostarch.Addr(size).RoundUp(); ok {
 		sizeAligned = uint64(val)
 	} else {
-		return nil, syserror.EINVAL
+		return nil, linuxerr.EINVAL
 	}
 
 	if numPages := sizeAligned / hostarch.PageSize; r.totalPages+numPages > linux.SHMALL {
@@ -652,7 +653,7 @@ func (s *Shm) Set(ctx context.Context, ds *linux.ShmidDS) error {
 	uid := creds.UserNamespace.MapToKUID(auth.UID(ds.ShmPerm.UID))
 	gid := creds.UserNamespace.MapToKGID(auth.GID(ds.ShmPerm.GID))
 	if !uid.Ok() || !gid.Ok() {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 
 	// User may only modify the lower 9 bits of the mode. All the other bits are
diff --git a/pkg/sentry/kernel/signalfd/signalfd.go b/pkg/sentry/kernel/signalfd/signalfd.go
index f58ec4194..47958e2d4 100644
--- a/pkg/sentry/kernel/signalfd/signalfd.go
+++ b/pkg/sentry/kernel/signalfd/signalfd.go
@@ -18,6 +18,7 @@ package signalfd
 import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/sentry/fs"
 	"gvisor.dev/gvisor/pkg/sentry/fs/anon"
 	"gvisor.dev/gvisor/pkg/sentry/fs/fsutil"
@@ -64,7 +65,7 @@ func New(ctx context.Context, mask linux.SignalSet) (*fs.File, error) {
 	t := kernel.TaskFromContext(ctx)
 	if t == nil {
 		// No task context? Not valid.
-		return nil, syserror.EINVAL
+		return nil, linuxerr.EINVAL
 	}
 	// name matches fs/signalfd.c:signalfd4.
 	dirent := fs.NewDirent(ctx, anon.NewInode(ctx), "anon_inode:[signalfd]")
diff --git a/pkg/sentry/kernel/task.go b/pkg/sentry/kernel/task.go
index 2e3b4488a..d211e4d82 100644
--- a/pkg/sentry/kernel/task.go
+++ b/pkg/sentry/kernel/task.go
@@ -21,6 +21,7 @@ import (
 
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/bpf"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/sentry/fs"
 	"gvisor.dev/gvisor/pkg/sentry/inet"
@@ -32,7 +33,6 @@ import (
 	"gvisor.dev/gvisor/pkg/sentry/usage"
 	"gvisor.dev/gvisor/pkg/sentry/vfs"
 	"gvisor.dev/gvisor/pkg/sync"
-	"gvisor.dev/gvisor/pkg/syserror"
 	"gvisor.dev/gvisor/pkg/waiter"
 )
 
@@ -846,7 +846,7 @@ func (t *Task) OOMScoreAdj() int32 {
 // value should be between -1000 and 1000 inclusive.
 func (t *Task) SetOOMScoreAdj(adj int32) error {
 	if adj > 1000 || adj < -1000 {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	atomic.StoreInt32(&t.tg.oomScoreAdj, adj)
 	return nil
diff --git a/pkg/sentry/kernel/task_acct.go b/pkg/sentry/kernel/task_acct.go
index e574997f7..dd364ae50 100644
--- a/pkg/sentry/kernel/task_acct.go
+++ b/pkg/sentry/kernel/task_acct.go
@@ -18,10 +18,10 @@ package kernel
 
 import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	ktime "gvisor.dev/gvisor/pkg/sentry/kernel/time"
 	"gvisor.dev/gvisor/pkg/sentry/limits"
 	"gvisor.dev/gvisor/pkg/sentry/usage"
-	"gvisor.dev/gvisor/pkg/syserror"
 )
 
 // Getitimer implements getitimer(2).
@@ -44,7 +44,7 @@ func (t *Task) Getitimer(id int32) (linux.ItimerVal, error) {
 		s, _ = t.tg.itimerProfSetting.At(tm)
 		t.tg.signalHandlers.mu.Unlock()
 	default:
-		return linux.ItimerVal{}, syserror.EINVAL
+		return linux.ItimerVal{}, linuxerr.EINVAL
 	}
 	val, iv := ktime.SpecFromSetting(tm, s)
 	return linux.ItimerVal{
@@ -105,7 +105,7 @@ func (t *Task) Setitimer(id int32, newitv linux.ItimerVal) (linux.ItimerVal, err
 			return linux.ItimerVal{}, err
 		}
 	default:
-		return linux.ItimerVal{}, syserror.EINVAL
+		return linux.ItimerVal{}, linuxerr.EINVAL
 	}
 	oldval, oldiv := ktime.SpecFromSetting(tm, olds)
 	return linux.ItimerVal{
diff --git a/pkg/sentry/kernel/task_clone.go b/pkg/sentry/kernel/task_clone.go
index 405771f3f..76fb0e2cb 100644
--- a/pkg/sentry/kernel/task_clone.go
+++ b/pkg/sentry/kernel/task_clone.go
@@ -20,6 +20,7 @@ import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/bpf"
 	"gvisor.dev/gvisor/pkg/cleanup"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/sentry/inet"
 	"gvisor.dev/gvisor/pkg/syserror"
@@ -142,25 +143,25 @@ func (t *Task) Clone(opts *CloneOptions) (ThreadID, *SyscallControl, error) {
 	// address, any set of signal handlers must refer to the same address
 	// space.
 	if !opts.NewSignalHandlers && opts.NewAddressSpace {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 	// In order for the behavior of thread-group-directed signals to be sane,
 	// all tasks in a thread group must share signal handlers.
 	if !opts.NewThreadGroup && opts.NewSignalHandlers {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 	// All tasks in a thread group must be in the same PID namespace.
 	if !opts.NewThreadGroup && (opts.NewPIDNamespace || t.childPIDNamespace != nil) {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 	// The two different ways of specifying a new PID namespace are
 	// incompatible.
 	if opts.NewPIDNamespace && t.childPIDNamespace != nil {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 	// Thread groups and FS contexts cannot span user namespaces.
 	if opts.NewUserNamespace && (!opts.NewThreadGroup || !opts.NewFSContext) {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Pull task registers and FPU state, a cloned task will inherit the
@@ -463,14 +464,14 @@ func (t *Task) Unshare(opts *SharingOptions) error {
 	// sense that clone(2) allows a task to share signal handlers and address
 	// spaces with tasks in other thread groups.
 	if opts.NewAddressSpace || opts.NewSignalHandlers {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	creds := t.Credentials()
 	if opts.NewThreadGroup {
 		t.tg.signalHandlers.mu.Lock()
 		if t.tg.tasksCount != 1 {
 			t.tg.signalHandlers.mu.Unlock()
-			return syserror.EINVAL
+			return linuxerr.EINVAL
 		}
 		t.tg.signalHandlers.mu.Unlock()
 		// This isn't racy because we're the only living task, and therefore
diff --git a/pkg/sentry/kernel/task_identity.go b/pkg/sentry/kernel/task_identity.go
index 0325967e4..29f154ebd 100644
--- a/pkg/sentry/kernel/task_identity.go
+++ b/pkg/sentry/kernel/task_identity.go
@@ -16,6 +16,7 @@ package kernel
 
 import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/sentry/kernel/auth"
 	"gvisor.dev/gvisor/pkg/sentry/mm"
 	"gvisor.dev/gvisor/pkg/syserror"
@@ -47,7 +48,7 @@ func (t *Task) HasCapability(cp linux.Capability) bool {
 func (t *Task) SetUID(uid auth.UID) error {
 	// setuid considers -1 to be invalid.
 	if !uid.Ok() {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 
 	t.mu.Lock()
@@ -56,7 +57,7 @@ func (t *Task) SetUID(uid auth.UID) error {
 	creds := t.Credentials()
 	kuid := creds.UserNamespace.MapToKUID(uid)
 	if !kuid.Ok() {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	// "setuid() sets the effective user ID of the calling process. If the
 	// effective UID of the caller is root (more precisely: if the caller has
@@ -87,14 +88,14 @@ func (t *Task) SetREUID(r, e auth.UID) error {
 	if r.Ok() {
 		newR = creds.UserNamespace.MapToKUID(r)
 		if !newR.Ok() {
-			return syserror.EINVAL
+			return linuxerr.EINVAL
 		}
 	}
 	newE := creds.EffectiveKUID
 	if e.Ok() {
 		newE = creds.UserNamespace.MapToKUID(e)
 		if !newE.Ok() {
-			return syserror.EINVAL
+			return linuxerr.EINVAL
 		}
 	}
 	if !creds.HasCapability(linux.CAP_SETUID) {
@@ -223,7 +224,7 @@ func (t *Task) setKUIDsUncheckedLocked(newR, newE, newS auth.KUID) {
 // SetGID implements the semantics of setgid(2).
 func (t *Task) SetGID(gid auth.GID) error {
 	if !gid.Ok() {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 
 	t.mu.Lock()
@@ -232,7 +233,7 @@ func (t *Task) SetGID(gid auth.GID) error {
 	creds := t.Credentials()
 	kgid := creds.UserNamespace.MapToKGID(gid)
 	if !kgid.Ok() {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	if creds.HasCapability(linux.CAP_SETGID) {
 		t.setKGIDsUncheckedLocked(kgid, kgid, kgid)
@@ -255,14 +256,14 @@ func (t *Task) SetREGID(r, e auth.GID) error {
 	if r.Ok() {
 		newR = creds.UserNamespace.MapToKGID(r)
 		if !newR.Ok() {
-			return syserror.EINVAL
+			return linuxerr.EINVAL
 		}
 	}
 	newE := creds.EffectiveKGID
 	if e.Ok() {
 		newE = creds.UserNamespace.MapToKGID(e)
 		if !newE.Ok() {
-			return syserror.EINVAL
+			return linuxerr.EINVAL
 		}
 	}
 	if !creds.HasCapability(linux.CAP_SETGID) {
@@ -349,7 +350,7 @@ func (t *Task) SetExtraGIDs(gids []auth.GID) error {
 	for i, gid := range gids {
 		kgid := creds.UserNamespace.MapToKGID(gid)
 		if !kgid.Ok() {
-			return syserror.EINVAL
+			return linuxerr.EINVAL
 		}
 		kgids[i] = kgid
 	}
diff --git a/pkg/sentry/kernel/task_sched.go b/pkg/sentry/kernel/task_sched.go
index f142feab4..9d9fa76a6 100644
--- a/pkg/sentry/kernel/task_sched.go
+++ b/pkg/sentry/kernel/task_sched.go
@@ -23,12 +23,12 @@ import (
 	"time"
 
 	"gvisor.dev/gvisor/pkg/abi/linux"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/sentry/hostcpu"
 	"gvisor.dev/gvisor/pkg/sentry/kernel/sched"
 	ktime "gvisor.dev/gvisor/pkg/sentry/kernel/time"
 	"gvisor.dev/gvisor/pkg/sentry/limits"
 	"gvisor.dev/gvisor/pkg/sentry/usage"
-	"gvisor.dev/gvisor/pkg/syserror"
 )
 
 // TaskGoroutineState is a coarse representation of the current execution
@@ -601,7 +601,7 @@ func (t *Task) SetCPUMask(mask sched.CPUSet) error {
 
 	// Ensure that at least 1 CPU is still allowed.
 	if mask.NumCPUs() == 0 {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 
 	if t.k.useHostCores {
diff --git a/pkg/sentry/kernel/task_signals.go b/pkg/sentry/kernel/task_signals.go
index 8ca61ed48..f54c774cb 100644
--- a/pkg/sentry/kernel/task_signals.go
+++ b/pkg/sentry/kernel/task_signals.go
@@ -22,6 +22,7 @@ import (
 	"time"
 
 	"gvisor.dev/gvisor/pkg/abi/linux"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/eventchannel"
 	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
@@ -370,7 +371,7 @@ func (t *Task) Sigtimedwait(set linux.SignalSet, timeout time.Duration) (*linux.
 // The following errors may be returned:
 //
 //	syserror.ESRCH - The task has exited.
-//	syserror.EINVAL - The signal is not valid.
+//	linuxerr.EINVAL - The signal is not valid.
 //	syserror.EAGAIN - THe signal is realtime, and cannot be queued.
 //
 func (t *Task) SendSignal(info *linux.SignalInfo) error {
@@ -413,7 +414,7 @@ func (t *Task) sendSignalTimerLocked(info *linux.SignalInfo, group bool, timer *
 		return nil
 	}
 	if !sig.IsValid() {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 
 	// Signal side effects apply even if the signal is ultimately discarded.
@@ -684,7 +685,7 @@ func (t *Task) SetSignalStack(alt linux.SignalStack) bool {
 // to *actptr (if actptr is not nil) and returns the old signal action.
 func (tg *ThreadGroup) SetSigAction(sig linux.Signal, actptr *linux.SigAction) (linux.SigAction, error) {
 	if !sig.IsValid() {
-		return linux.SigAction{}, syserror.EINVAL
+		return linux.SigAction{}, linuxerr.EINVAL
 	}
 
 	tg.pidns.owner.mu.RLock()
@@ -695,7 +696,7 @@ func (tg *ThreadGroup) SetSigAction(sig linux.Signal, actptr *linux.SigAction) (
 	oldact := sh.actions[sig]
 	if actptr != nil {
 		if sig == linux.SIGKILL || sig == linux.SIGSTOP {
-			return oldact, syserror.EINVAL
+			return oldact, linuxerr.EINVAL
 		}
 
 		act := *actptr
diff --git a/pkg/sentry/kernel/task_syscall.go b/pkg/sentry/kernel/task_syscall.go
index 1874f74e5..409b712d8 100644
--- a/pkg/sentry/kernel/task_syscall.go
+++ b/pkg/sentry/kernel/task_syscall.go
@@ -22,6 +22,7 @@ import (
 	"golang.org/x/sys/unix"
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/bits"
+	"gvisor.dev/gvisor/pkg/errors"
 	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/marshal"
@@ -380,6 +381,8 @@ func ExtractErrno(err error, sysno int) int {
 		return 0
 	case unix.Errno:
 		return int(err)
+	case *errors.Error:
+		return int(err.Errno())
 	case syserror.SyscallRestartErrno:
 		return int(err)
 	case *memmap.BusError:
diff --git a/pkg/sentry/kernel/task_usermem.go b/pkg/sentry/kernel/task_usermem.go
index fc6d9438a..7935d15a6 100644
--- a/pkg/sentry/kernel/task_usermem.go
+++ b/pkg/sentry/kernel/task_usermem.go
@@ -19,6 +19,7 @@ import (
 
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/sentry/mm"
 	"gvisor.dev/gvisor/pkg/syserror"
@@ -202,7 +203,7 @@ func (t *Task) CopyInIovecs(addr hostarch.Addr, numIovecs int) (hostarch.AddrRan
 			base := hostarch.Addr(hostarch.ByteOrder.Uint64(b[0:8]))
 			length := hostarch.ByteOrder.Uint64(b[8:16])
 			if length > math.MaxInt64 {
-				return hostarch.AddrRangeSeq{}, syserror.EINVAL
+				return hostarch.AddrRangeSeq{}, linuxerr.EINVAL
 			}
 			ar, ok := t.MemoryManager().CheckIORange(base, int64(length))
 			if !ok {
@@ -270,7 +271,7 @@ func (t *Task) SingleIOSequence(addr hostarch.Addr, length int, opts usermem.IOO
 // Preconditions: Same as Task.CopyInIovecs.
 func (t *Task) IovecsIOSequence(addr hostarch.Addr, iovcnt int, opts usermem.IOOpts) (usermem.IOSequence, error) {
 	if iovcnt < 0 || iovcnt > linux.UIO_MAXIOV {
-		return usermem.IOSequence{}, syserror.EINVAL
+		return usermem.IOSequence{}, linuxerr.EINVAL
 	}
 	ars, err := t.CopyInIovecs(addr, iovcnt)
 	if err != nil {
diff --git a/pkg/sentry/kernel/thread_group.go b/pkg/sentry/kernel/thread_group.go
index 4566e4c7c..8ae00c649 100644
--- a/pkg/sentry/kernel/thread_group.go
+++ b/pkg/sentry/kernel/thread_group.go
@@ -19,6 +19,7 @@ import (
 
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/sentry/fs"
 	"gvisor.dev/gvisor/pkg/sentry/kernel/auth"
 	ktime "gvisor.dev/gvisor/pkg/sentry/kernel/time"
@@ -357,7 +358,7 @@ func (tg *ThreadGroup) SetControllingTTY(tty *TTY, steal bool, isReadable bool)
 	// "The calling process must be a session leader and not have a
 	// controlling terminal already." - tty_ioctl(4)
 	if tg.processGroup.session.leader != tg || tg.tty != nil {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 
 	creds := auth.CredentialsFromContext(tg.leader)
@@ -501,7 +502,7 @@ func (tg *ThreadGroup) SetForegroundProcessGroup(tty *TTY, pgid ProcessGroupID)
 
 	// pgid must be positive.
 	if pgid < 0 {
-		return -1, syserror.EINVAL
+		return -1, linuxerr.EINVAL
 	}
 
 	// pg must not be empty. Empty process groups are removed from their
diff --git a/pkg/sentry/kernel/time/time.go b/pkg/sentry/kernel/time/time.go
index 26aa34aa6..191b92811 100644
--- a/pkg/sentry/kernel/time/time.go
+++ b/pkg/sentry/kernel/time/time.go
@@ -22,8 +22,8 @@ import (
 	"time"
 
 	"gvisor.dev/gvisor/pkg/abi/linux"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/sync"
-	"gvisor.dev/gvisor/pkg/syserror"
 	"gvisor.dev/gvisor/pkg/waiter"
 )
 
@@ -322,7 +322,7 @@ func SettingFromSpec(value time.Duration, interval time.Duration, c Clock) (Sett
 // interpreted as a time relative to now.
 func SettingFromSpecAt(value time.Duration, interval time.Duration, now Time) (Setting, error) {
 	if value < 0 {
-		return Setting{}, syserror.EINVAL
+		return Setting{}, linuxerr.EINVAL
 	}
 	if value == 0 {
 		return Setting{Period: interval}, nil
@@ -338,7 +338,7 @@ func SettingFromSpecAt(value time.Duration, interval time.Duration, now Time) (S
 // interpreted as an absolute time.
 func SettingFromAbsSpec(value Time, interval time.Duration) (Setting, error) {
 	if value.Before(ZeroTime) {
-		return Setting{}, syserror.EINVAL
+		return Setting{}, linuxerr.EINVAL
 	}
 	if value.IsZero() {
 		return Setting{Period: interval}, nil
diff --git a/pkg/sentry/loader/elf.go b/pkg/sentry/loader/elf.go
index 4c7666e33..13ab7ea23 100644
--- a/pkg/sentry/loader/elf.go
+++ b/pkg/sentry/loader/elf.go
@@ -518,13 +518,13 @@ func loadParsedELF(ctx context.Context, m *mm.MemoryManager, f fsbridge.File, in
 		start, ok = start.AddLength(uint64(offset))
 		if !ok {
 			ctx.Infof(fmt.Sprintf("Start %#x + offset %#x overflows?", start, offset))
-			return loadedELF{}, syserror.EINVAL
+			return loadedELF{}, linuxerr.EINVAL
 		}
 
 		end, ok = end.AddLength(uint64(offset))
 		if !ok {
 			ctx.Infof(fmt.Sprintf("End %#x + offset %#x overflows?", end, offset))
-			return loadedELF{}, syserror.EINVAL
+			return loadedELF{}, linuxerr.EINVAL
 		}
 
 		info.entry, ok = info.entry.AddLength(uint64(offset))
diff --git a/pkg/sentry/loader/vdso.go b/pkg/sentry/loader/vdso.go
index fd54261fd..054ef1723 100644
--- a/pkg/sentry/loader/vdso.go
+++ b/pkg/sentry/loader/vdso.go
@@ -23,6 +23,7 @@ import (
 
 	"gvisor.dev/gvisor/pkg/abi"
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/log"
 	"gvisor.dev/gvisor/pkg/safemem"
@@ -58,7 +59,7 @@ type byteFullReader struct {
 
 func (b *byteFullReader) ReadFull(ctx context.Context, dst usermem.IOSequence, offset int64) (int64, error) {
 	if offset < 0 {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 	if offset >= int64(len(b.data)) {
 		return 0, io.EOF
diff --git a/pkg/sentry/mm/aio_context.go b/pkg/sentry/mm/aio_context.go
index 346866d3c..8426fc90e 100644
--- a/pkg/sentry/mm/aio_context.go
+++ b/pkg/sentry/mm/aio_context.go
@@ -17,6 +17,7 @@ package mm
 import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/sentry/memmap"
 	"gvisor.dev/gvisor/pkg/sentry/pgalloc"
@@ -158,7 +159,7 @@ func (ctx *AIOContext) Prepare() error {
 	defer ctx.mu.Unlock()
 	if ctx.dead {
 		// Context died after the caller looked it up.
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	if ctx.outstanding >= ctx.maxOutstanding {
 		// Context is busy.
@@ -297,7 +298,7 @@ func (m *aioMappable) InodeID() uint64 {
 // Msync implements memmap.MappingIdentity.Msync.
 func (m *aioMappable) Msync(ctx context.Context, mr memmap.MappableRange) error {
 	// Linux: aio_ring_fops.fsync == NULL
-	return syserror.EINVAL
+	return linuxerr.EINVAL
 }
 
 // AddMapping implements memmap.Mappable.AddMapping.
@@ -325,7 +326,7 @@ func (m *aioMappable) CopyMapping(ctx context.Context, ms memmap.MappingSpace, s
 	// Linux's fs/aio.c:aio_ring_mremap().
 	mm, ok := ms.(*MemoryManager)
 	if !ok {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	am := &mm.aioManager
 	am.mu.Lock()
@@ -333,12 +334,12 @@ func (m *aioMappable) CopyMapping(ctx context.Context, ms memmap.MappingSpace, s
 	oldID := uint64(srcAR.Start)
 	aioCtx, ok := am.contexts[oldID]
 	if !ok {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	aioCtx.mu.Lock()
 	defer aioCtx.mu.Unlock()
 	if aioCtx.dead {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	// Use the new ID for the AIOContext.
 	am.contexts[uint64(dstAR.Start)] = aioCtx
@@ -399,7 +400,7 @@ func (mm *MemoryManager) NewAIOContext(ctx context.Context, events uint32) (uint
 	id := uint64(addr)
 	if !mm.aioManager.newAIOContext(events, id) {
 		mm.MUnmap(ctx, addr, aioRingBufferSize)
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 	return id, nil
 }
diff --git a/pkg/sentry/mm/shm.go b/pkg/sentry/mm/shm.go
index 3130be80c..94d5112a1 100644
--- a/pkg/sentry/mm/shm.go
+++ b/pkg/sentry/mm/shm.go
@@ -16,16 +16,16 @@ package mm
 
 import (
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/sentry/kernel/shm"
-	"gvisor.dev/gvisor/pkg/syserror"
 )
 
 // DetachShm unmaps a sysv shared memory segment.
 func (mm *MemoryManager) DetachShm(ctx context.Context, addr hostarch.Addr) error {
 	if addr != addr.RoundDown() {
 		// "... shmaddr is not aligned on a page boundary." - man shmdt(2)
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 
 	var detached *shm.Shm
@@ -48,7 +48,7 @@ func (mm *MemoryManager) DetachShm(ctx context.Context, addr hostarch.Addr) erro
 
 	if detached == nil {
 		// There is no shared memory segment attached at addr.
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 
 	// Remove all vmas that could have been created by the same attach.
diff --git a/pkg/sentry/mm/special_mappable.go b/pkg/sentry/mm/special_mappable.go
index e748b7ff8..feafe76c1 100644
--- a/pkg/sentry/mm/special_mappable.go
+++ b/pkg/sentry/mm/special_mappable.go
@@ -16,6 +16,7 @@ package mm
 
 import (
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/sentry/memmap"
 	"gvisor.dev/gvisor/pkg/sentry/pgalloc"
@@ -144,11 +145,11 @@ func (m *SpecialMappable) Length() uint64 {
 // leak (b/143656263). Delete this function along with VFS1.
 func NewSharedAnonMappable(length uint64, mfp pgalloc.MemoryFileProvider) (*SpecialMappable, error) {
 	if length == 0 {
-		return nil, syserror.EINVAL
+		return nil, linuxerr.EINVAL
 	}
 	alignedLen, ok := hostarch.Addr(length).RoundUp()
 	if !ok {
-		return nil, syserror.EINVAL
+		return nil, linuxerr.EINVAL
 	}
 	fr, err := mfp.MemoryFile().Allocate(uint64(alignedLen), usage.Anonymous)
 	if err != nil {
diff --git a/pkg/sentry/mm/syscalls.go b/pkg/sentry/mm/syscalls.go
index f46f85eb1..7b6715815 100644
--- a/pkg/sentry/mm/syscalls.go
+++ b/pkg/sentry/mm/syscalls.go
@@ -75,7 +75,7 @@ func (mm *MemoryManager) HandleUserFault(ctx context.Context, addr hostarch.Addr
 // MMap establishes a memory mapping.
 func (mm *MemoryManager) MMap(ctx context.Context, opts memmap.MMapOpts) (hostarch.Addr, error) {
 	if opts.Length == 0 {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 	length, ok := hostarch.Addr(opts.Length).RoundUp()
 	if !ok {
@@ -86,7 +86,7 @@ func (mm *MemoryManager) MMap(ctx context.Context, opts memmap.MMapOpts) (hostar
 	if opts.Mappable != nil {
 		// Offset must be aligned.
 		if hostarch.Addr(opts.Offset).RoundDown() != hostarch.Addr(opts.Offset) {
-			return 0, syserror.EINVAL
+			return 0, linuxerr.EINVAL
 		}
 		// Offset + length must not overflow.
 		if end := opts.Offset + opts.Length; end < opts.Offset {
@@ -100,7 +100,7 @@ func (mm *MemoryManager) MMap(ctx context.Context, opts memmap.MMapOpts) (hostar
 		// MAP_FIXED requires addr to be page-aligned; non-fixed mappings
 		// don't.
 		if opts.Fixed {
-			return 0, syserror.EINVAL
+			return 0, linuxerr.EINVAL
 		}
 		opts.Addr = opts.Addr.RoundDown()
 	}
@@ -109,10 +109,10 @@ func (mm *MemoryManager) MMap(ctx context.Context, opts memmap.MMapOpts) (hostar
 		return 0, syserror.EACCES
 	}
 	if opts.Unmap && !opts.Fixed {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 	if opts.GrowsDown && opts.Mappable != nil {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 
 	// Get the new vma.
@@ -282,18 +282,18 @@ func (mm *MemoryManager) MapStack(ctx context.Context) (hostarch.AddrRange, erro
 // MUnmap implements the semantics of Linux's munmap(2).
 func (mm *MemoryManager) MUnmap(ctx context.Context, addr hostarch.Addr, length uint64) error {
 	if addr != addr.RoundDown() {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	if length == 0 {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	la, ok := hostarch.Addr(length).RoundUp()
 	if !ok {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	ar, ok := addr.ToRange(uint64(la))
 	if !ok {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 
 	mm.mappingMu.Lock()
@@ -332,7 +332,7 @@ const (
 func (mm *MemoryManager) MRemap(ctx context.Context, oldAddr hostarch.Addr, oldSize uint64, newSize uint64, opts MRemapOpts) (hostarch.Addr, error) {
 	// "Note that old_address has to be page aligned." - mremap(2)
 	if oldAddr.RoundDown() != oldAddr {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 
 	// Linux treats an old_size that rounds up to 0 as 0, which is otherwise a
@@ -341,13 +341,13 @@ func (mm *MemoryManager) MRemap(ctx context.Context, oldAddr hostarch.Addr, oldS
 	oldSize = uint64(oldSizeAddr)
 	newSizeAddr, ok := hostarch.Addr(newSize).RoundUp()
 	if !ok || newSizeAddr == 0 {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 	newSize = uint64(newSizeAddr)
 
 	oldEnd, ok := oldAddr.AddLength(oldSize)
 	if !ok {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 
 	mm.mappingMu.Lock()
@@ -451,15 +451,15 @@ func (mm *MemoryManager) MRemap(ctx context.Context, oldAddr hostarch.Addr, oldS
 	case MRemapMustMove:
 		newAddr := opts.NewAddr
 		if newAddr.RoundDown() != newAddr {
-			return 0, syserror.EINVAL
+			return 0, linuxerr.EINVAL
 		}
 		var ok bool
 		newAR, ok = newAddr.ToRange(newSize)
 		if !ok {
-			return 0, syserror.EINVAL
+			return 0, linuxerr.EINVAL
 		}
 		if (hostarch.AddrRange{oldAddr, oldEnd}).Overlaps(newAR) {
-			return 0, syserror.EINVAL
+			return 0, linuxerr.EINVAL
 		}
 
 		// Check that the new region is valid.
@@ -505,7 +505,7 @@ func (mm *MemoryManager) MRemap(ctx context.Context, oldAddr hostarch.Addr, oldS
 	if vma := vseg.ValuePtr(); vma.mappable != nil {
 		// Check that offset+length does not overflow.
 		if vma.off+uint64(newAR.Length()) < vma.off {
-			return 0, syserror.EINVAL
+			return 0, linuxerr.EINVAL
 		}
 		// Inform the Mappable, if any, of the new mapping.
 		if err := vma.mappable.CopyMapping(ctx, mm, oldAR, newAR, vseg.mappableOffsetAt(oldAR.Start), vma.canWriteMappableLocked()); err != nil {
@@ -591,7 +591,7 @@ func (mm *MemoryManager) MRemap(ctx context.Context, oldAddr hostarch.Addr, oldS
 // MProtect implements the semantics of Linux's mprotect(2).
 func (mm *MemoryManager) MProtect(addr hostarch.Addr, length uint64, realPerms hostarch.AccessType, growsDown bool) error {
 	if addr.RoundDown() != addr {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	if length == 0 {
 		return nil
@@ -619,7 +619,7 @@ func (mm *MemoryManager) MProtect(addr hostarch.Addr, length uint64, realPerms h
 	}
 	if growsDown {
 		if !vseg.ValuePtr().growsDown {
-			return syserror.EINVAL
+			return linuxerr.EINVAL
 		}
 		if ar.End <= vseg.Start() {
 			return syserror.ENOMEM
@@ -712,7 +712,7 @@ func (mm *MemoryManager) Brk(ctx context.Context, addr hostarch.Addr) (hostarch.
 	if addr < mm.brk.Start {
 		addr = mm.brk.End
 		mm.mappingMu.Unlock()
-		return addr, syserror.EINVAL
+		return addr, linuxerr.EINVAL
 	}
 
 	// TODO(gvisor.dev/issue/156): This enforces RLIMIT_DATA, but is
@@ -781,7 +781,7 @@ func (mm *MemoryManager) MLock(ctx context.Context, addr hostarch.Addr, length u
 	la, _ := hostarch.Addr(length + addr.PageOffset()).RoundUp()
 	ar, ok := addr.RoundDown().ToRange(uint64(la))
 	if !ok {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 
 	mm.mappingMu.Lock()
@@ -899,7 +899,7 @@ type MLockAllOpts struct {
 // depending on opts.
 func (mm *MemoryManager) MLockAll(ctx context.Context, opts MLockAllOpts) error {
 	if !opts.Current && !opts.Future {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 
 	mm.mappingMu.Lock()
@@ -980,13 +980,13 @@ func (mm *MemoryManager) NumaPolicy(addr hostarch.Addr) (linux.NumaPolicy, uint6
 // SetNumaPolicy implements the semantics of Linux's mbind().
 func (mm *MemoryManager) SetNumaPolicy(addr hostarch.Addr, length uint64, policy linux.NumaPolicy, nodemask uint64) error {
 	if !addr.IsPageAligned() {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	// Linux allows this to overflow.
 	la, _ := hostarch.Addr(length).RoundUp()
 	ar, ok := addr.ToRange(uint64(la))
 	if !ok {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	if ar.Length() == 0 {
 		return nil
@@ -1022,7 +1022,7 @@ func (mm *MemoryManager) SetNumaPolicy(addr hostarch.Addr, length uint64, policy
 func (mm *MemoryManager) SetDontFork(addr hostarch.Addr, length uint64, dontfork bool) error {
 	ar, ok := addr.ToRange(length)
 	if !ok {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 
 	mm.mappingMu.Lock()
@@ -1048,7 +1048,7 @@ func (mm *MemoryManager) SetDontFork(addr hostarch.Addr, length uint64, dontfork
 func (mm *MemoryManager) Decommit(addr hostarch.Addr, length uint64) error {
 	ar, ok := addr.ToRange(length)
 	if !ok {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 
 	mm.mappingMu.RLock()
@@ -1064,7 +1064,7 @@ func (mm *MemoryManager) Decommit(addr hostarch.Addr, length uint64) error {
 	for vseg := mm.vmas.LowerBoundSegment(ar.Start); vseg.Ok() && vseg.Start() < ar.End; vseg = vseg.NextSegment() {
 		vma := vseg.ValuePtr()
 		if vma.mlockMode != memmap.MLockNone {
-			return syserror.EINVAL
+			return linuxerr.EINVAL
 		}
 		vsegAR := vseg.Range().Intersect(ar)
 		// pseg should already correspond to either this vma or a later one,
@@ -1115,7 +1115,7 @@ type MSyncOpts struct {
 // MSync implements the semantics of Linux's msync().
 func (mm *MemoryManager) MSync(ctx context.Context, addr hostarch.Addr, length uint64, opts MSyncOpts) error {
 	if addr != addr.RoundDown() {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	if length == 0 {
 		return nil
diff --git a/pkg/sentry/socket/control/control.go b/pkg/sentry/socket/control/control.go
index 235b9c306..64958b6ec 100644
--- a/pkg/sentry/socket/control/control.go
+++ b/pkg/sentry/socket/control/control.go
@@ -20,6 +20,7 @@ import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/bits"
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/marshal"
 	"gvisor.dev/gvisor/pkg/marshal/primitive"
@@ -473,17 +474,17 @@ func Parse(t *kernel.Task, socketOrEndpoint interface{}, buf []byte, width uint)
 
 	for i := 0; i < len(buf); {
 		if i+linux.SizeOfControlMessageHeader > len(buf) {
-			return cmsgs, syserror.EINVAL
+			return cmsgs, linuxerr.EINVAL
 		}
 
 		var h linux.ControlMessageHeader
 		h.UnmarshalUnsafe(buf[i : i+linux.SizeOfControlMessageHeader])
 
 		if h.Length < uint64(linux.SizeOfControlMessageHeader) {
-			return socket.ControlMessages{}, syserror.EINVAL
+			return socket.ControlMessages{}, linuxerr.EINVAL
 		}
 		if h.Length > uint64(len(buf)-i) {
-			return socket.ControlMessages{}, syserror.EINVAL
+			return socket.ControlMessages{}, linuxerr.EINVAL
 		}
 
 		i += linux.SizeOfControlMessageHeader
@@ -497,7 +498,7 @@ func Parse(t *kernel.Task, socketOrEndpoint interface{}, buf []byte, width uint)
 				numRights := rightsSize / linux.SizeOfControlMessageRight
 
 				if len(fds)+numRights > linux.SCM_MAX_FD {
-					return socket.ControlMessages{}, syserror.EINVAL
+					return socket.ControlMessages{}, linuxerr.EINVAL
 				}
 
 				for j := i; j < i+rightsSize; j += linux.SizeOfControlMessageRight {
@@ -508,7 +509,7 @@ func Parse(t *kernel.Task, socketOrEndpoint interface{}, buf []byte, width uint)
 
 			case linux.SCM_CREDENTIALS:
 				if length < linux.SizeOfControlMessageCredentials {
-					return socket.ControlMessages{}, syserror.EINVAL
+					return socket.ControlMessages{}, linuxerr.EINVAL
 				}
 
 				var creds linux.ControlMessageCredentials
@@ -522,7 +523,7 @@ func Parse(t *kernel.Task, socketOrEndpoint interface{}, buf []byte, width uint)
 
 			case linux.SO_TIMESTAMP:
 				if length < linux.SizeOfTimeval {
-					return socket.ControlMessages{}, syserror.EINVAL
+					return socket.ControlMessages{}, linuxerr.EINVAL
 				}
 				var ts linux.Timeval
 				ts.UnmarshalUnsafe(buf[i : i+linux.SizeOfTimeval])
@@ -532,13 +533,13 @@ func Parse(t *kernel.Task, socketOrEndpoint interface{}, buf []byte, width uint)
 
 			default:
 				// Unknown message type.
-				return socket.ControlMessages{}, syserror.EINVAL
+				return socket.ControlMessages{}, linuxerr.EINVAL
 			}
 		case linux.SOL_IP:
 			switch h.Type {
 			case linux.IP_TOS:
 				if length < linux.SizeOfControlMessageTOS {
-					return socket.ControlMessages{}, syserror.EINVAL
+					return socket.ControlMessages{}, linuxerr.EINVAL
 				}
 				cmsgs.IP.HasTOS = true
 				var tos primitive.Uint8
@@ -548,7 +549,7 @@ func Parse(t *kernel.Task, socketOrEndpoint interface{}, buf []byte, width uint)
 
 			case linux.IP_PKTINFO:
 				if length < linux.SizeOfControlMessageIPPacketInfo {
-					return socket.ControlMessages{}, syserror.EINVAL
+					return socket.ControlMessages{}, linuxerr.EINVAL
 				}
 
 				cmsgs.IP.HasIPPacketInfo = true
@@ -561,7 +562,7 @@ func Parse(t *kernel.Task, socketOrEndpoint interface{}, buf []byte, width uint)
 			case linux.IP_RECVORIGDSTADDR:
 				var addr linux.SockAddrInet
 				if length < addr.SizeBytes() {
-					return socket.ControlMessages{}, syserror.EINVAL
+					return socket.ControlMessages{}, linuxerr.EINVAL
 				}
 				addr.UnmarshalUnsafe(buf[i : i+addr.SizeBytes()])
 				cmsgs.IP.OriginalDstAddress = &addr
@@ -570,7 +571,7 @@ func Parse(t *kernel.Task, socketOrEndpoint interface{}, buf []byte, width uint)
 			case linux.IP_RECVERR:
 				var errCmsg linux.SockErrCMsgIPv4
 				if length < errCmsg.SizeBytes() {
-					return socket.ControlMessages{}, syserror.EINVAL
+					return socket.ControlMessages{}, linuxerr.EINVAL
 				}
 
 				errCmsg.UnmarshalBytes(buf[i : i+errCmsg.SizeBytes()])
@@ -578,13 +579,13 @@ func Parse(t *kernel.Task, socketOrEndpoint interface{}, buf []byte, width uint)
 				i += bits.AlignUp(length, width)
 
 			default:
-				return socket.ControlMessages{}, syserror.EINVAL
+				return socket.ControlMessages{}, linuxerr.EINVAL
 			}
 		case linux.SOL_IPV6:
 			switch h.Type {
 			case linux.IPV6_TCLASS:
 				if length < linux.SizeOfControlMessageTClass {
-					return socket.ControlMessages{}, syserror.EINVAL
+					return socket.ControlMessages{}, linuxerr.EINVAL
 				}
 				cmsgs.IP.HasTClass = true
 				var tclass primitive.Uint32
@@ -595,7 +596,7 @@ func Parse(t *kernel.Task, socketOrEndpoint interface{}, buf []byte, width uint)
 			case linux.IPV6_RECVORIGDSTADDR:
 				var addr linux.SockAddrInet6
 				if length < addr.SizeBytes() {
-					return socket.ControlMessages{}, syserror.EINVAL
+					return socket.ControlMessages{}, linuxerr.EINVAL
 				}
 				addr.UnmarshalUnsafe(buf[i : i+addr.SizeBytes()])
 				cmsgs.IP.OriginalDstAddress = &addr
@@ -604,7 +605,7 @@ func Parse(t *kernel.Task, socketOrEndpoint interface{}, buf []byte, width uint)
 			case linux.IPV6_RECVERR:
 				var errCmsg linux.SockErrCMsgIPv6
 				if length < errCmsg.SizeBytes() {
-					return socket.ControlMessages{}, syserror.EINVAL
+					return socket.ControlMessages{}, linuxerr.EINVAL
 				}
 
 				errCmsg.UnmarshalBytes(buf[i : i+errCmsg.SizeBytes()])
@@ -612,10 +613,10 @@ func Parse(t *kernel.Task, socketOrEndpoint interface{}, buf []byte, width uint)
 				i += bits.AlignUp(length, width)
 
 			default:
-				return socket.ControlMessages{}, syserror.EINVAL
+				return socket.ControlMessages{}, linuxerr.EINVAL
 			}
 		default:
-			return socket.ControlMessages{}, syserror.EINVAL
+			return socket.ControlMessages{}, linuxerr.EINVAL
 		}
 	}
 
diff --git a/pkg/sentry/socket/netstack/stack.go b/pkg/sentry/socket/netstack/stack.go
index eef5e6519..9d343b671 100644
--- a/pkg/sentry/socket/netstack/stack.go
+++ b/pkg/sentry/socket/netstack/stack.go
@@ -18,6 +18,7 @@ import (
 	"fmt"
 
 	"gvisor.dev/gvisor/pkg/abi/linux"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/log"
 	"gvisor.dev/gvisor/pkg/sentry/inet"
 	"gvisor.dev/gvisor/pkg/syserr"
@@ -110,19 +111,19 @@ func convertAddr(addr inet.InterfaceAddr) (tcpip.ProtocolAddress, error) {
 	switch addr.Family {
 	case linux.AF_INET:
 		if len(addr.Addr) != header.IPv4AddressSize {
-			return protocolAddress, syserror.EINVAL
+			return protocolAddress, linuxerr.EINVAL
 		}
 		if addr.PrefixLen > header.IPv4AddressSize*8 {
-			return protocolAddress, syserror.EINVAL
+			return protocolAddress, linuxerr.EINVAL
 		}
 		protocol = ipv4.ProtocolNumber
 		address = tcpip.Address(addr.Addr)
 	case linux.AF_INET6:
 		if len(addr.Addr) != header.IPv6AddressSize {
-			return protocolAddress, syserror.EINVAL
+			return protocolAddress, linuxerr.EINVAL
 		}
 		if addr.PrefixLen > header.IPv6AddressSize*8 {
-			return protocolAddress, syserror.EINVAL
+			return protocolAddress, linuxerr.EINVAL
 		}
 		protocol = ipv6.ProtocolNumber
 		address = tcpip.Address(addr.Addr)
diff --git a/pkg/sentry/socket/netstack/tun.go b/pkg/sentry/socket/netstack/tun.go
index c7ed52702..e67fe9700 100644
--- a/pkg/sentry/socket/netstack/tun.go
+++ b/pkg/sentry/socket/netstack/tun.go
@@ -16,7 +16,7 @@ package netstack
 
 import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
-	"gvisor.dev/gvisor/pkg/syserror"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/tcpip/link/tun"
 )
 
@@ -41,7 +41,7 @@ func LinuxToTUNFlags(flags uint16) (tun.Flags, error) {
 	// when there is no sk_filter. See __tun_chr_ioctl() in
 	// net/drivers/tun.c.
 	if flags&^uint16(linux.IFF_TUN|linux.IFF_TAP|linux.IFF_NO_PI|linux.IFF_ONE_QUEUE) != 0 {
-		return tun.Flags{}, syserror.EINVAL
+		return tun.Flags{}, linuxerr.EINVAL
 	}
 	return tun.Flags{
 		TUN:          flags&linux.IFF_TUN != 0,
diff --git a/pkg/sentry/syscalls/linux/linux64.go b/pkg/sentry/syscalls/linux/linux64.go
index 090c5ffcb..1732064ef 100644
--- a/pkg/sentry/syscalls/linux/linux64.go
+++ b/pkg/sentry/syscalls/linux/linux64.go
@@ -18,6 +18,7 @@ package linux
 import (
 	"gvisor.dev/gvisor/pkg/abi"
 	"gvisor.dev/gvisor/pkg/abi/linux"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
 	"gvisor.dev/gvisor/pkg/sentry/kernel"
@@ -187,7 +188,7 @@ var AMD64 = &kernel.SyscallTable{
 		132: syscalls.Supported("utime", Utime),
 		133: syscalls.PartiallySupported("mknod", Mknod, "Device creation is not generally supported. Only regular file and FIFO creation are supported.", nil),
 		134: syscalls.Error("uselib", syserror.ENOSYS, "Obsolete", nil),
-		135: syscalls.ErrorWithEvent("personality", syserror.EINVAL, "Unable to change personality.", nil),
+		135: syscalls.ErrorWithEvent("personality", linuxerr.EINVAL, "Unable to change personality.", nil),
 		136: syscalls.ErrorWithEvent("ustat", syserror.ENOSYS, "Needs filesystem support.", nil),
 		137: syscalls.PartiallySupported("statfs", Statfs, "Depends on the backing file system implementation.", nil),
 		138: syscalls.PartiallySupported("fstatfs", Fstatfs, "Depends on the backing file system implementation.", nil),
@@ -521,7 +522,7 @@ var ARM64 = &kernel.SyscallTable{
 		89:  syscalls.CapError("acct", linux.CAP_SYS_PACCT, "", nil),
 		90:  syscalls.Supported("capget", Capget),
 		91:  syscalls.Supported("capset", Capset),
-		92:  syscalls.ErrorWithEvent("personality", syserror.EINVAL, "Unable to change personality.", nil),
+		92:  syscalls.ErrorWithEvent("personality", linuxerr.EINVAL, "Unable to change personality.", nil),
 		93:  syscalls.Supported("exit", Exit),
 		94:  syscalls.Supported("exit_group", ExitGroup),
 		95:  syscalls.Supported("waitid", Waitid),
diff --git a/pkg/sentry/syscalls/linux/sigset.go b/pkg/sentry/syscalls/linux/sigset.go
index e8c2d8f9e..9dea78085 100644
--- a/pkg/sentry/syscalls/linux/sigset.go
+++ b/pkg/sentry/syscalls/linux/sigset.go
@@ -16,6 +16,7 @@ package linux
 
 import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/sentry/kernel"
 	"gvisor.dev/gvisor/pkg/syserror"
@@ -29,7 +30,7 @@ import (
 // syscalls are moved into this package, then they can be unexported.
 func CopyInSigSet(t *kernel.Task, sigSetAddr hostarch.Addr, size uint) (linux.SignalSet, error) {
 	if size != linux.SignalSetSize {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 	b := t.CopyScratchBuffer(8)
 	if _, err := t.CopyInBytes(sigSetAddr, b); err != nil {
diff --git a/pkg/sentry/syscalls/linux/sys_aio.go b/pkg/sentry/syscalls/linux/sys_aio.go
index c338a4cc9..a93fc635b 100644
--- a/pkg/sentry/syscalls/linux/sys_aio.go
+++ b/pkg/sentry/syscalls/linux/sys_aio.go
@@ -43,7 +43,7 @@ func IoSetup(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysca
 		return 0, nil, err
 	}
 	if idIn != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	id, err := t.MemoryManager().NewAIOContext(t, uint32(nrEvents))
@@ -67,7 +67,7 @@ func IoDestroy(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sys
 	ctx := t.MemoryManager().DestroyAIOContext(t, id)
 	if ctx == nil {
 		// Does not exist.
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Drain completed requests amd wait for pending requests until there are no
@@ -98,12 +98,12 @@ func IoGetevents(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.S
 
 	// Sanity check arguments.
 	if minEvents < 0 || minEvents > events {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	ctx, ok := t.MemoryManager().LookupAIOContext(t, id)
 	if !ok {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Setup the timeout.
@@ -115,7 +115,7 @@ func IoGetevents(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.S
 			return 0, nil, err
 		}
 		if !d.Valid() {
-			return 0, nil, syserror.EINVAL
+			return 0, nil, linuxerr.EINVAL
 		}
 		deadline = t.Kernel().MonotonicClock().Now().Add(d.ToDuration())
 		haveDeadline = true
@@ -172,7 +172,7 @@ func waitForRequest(ctx *mm.AIOContext, t *kernel.Task, haveDeadline bool, deadl
 		done := ctx.WaitChannel()
 		if done == nil {
 			// Context has been destroyed.
-			return nil, syserror.EINVAL
+			return nil, linuxerr.EINVAL
 		}
 		if err := t.BlockWithDeadline(done, haveDeadline, deadline); err != nil {
 			return nil, err
@@ -185,7 +185,7 @@ func memoryFor(t *kernel.Task, cb *linux.IOCallback) (usermem.IOSequence, error)
 	bytes := int(cb.Bytes)
 	if bytes < 0 {
 		// Linux also requires that this field fit in ssize_t.
-		return usermem.IOSequence{}, syserror.EINVAL
+		return usermem.IOSequence{}, linuxerr.EINVAL
 	}
 
 	// Since this I/O will be asynchronous with respect to t's task goroutine,
@@ -207,7 +207,7 @@ func memoryFor(t *kernel.Task, cb *linux.IOCallback) (usermem.IOSequence, error)
 
 	default:
 		// Not a supported command.
-		return usermem.IOSequence{}, syserror.EINVAL
+		return usermem.IOSequence{}, linuxerr.EINVAL
 	}
 }
 
@@ -287,7 +287,7 @@ func submitCallback(t *kernel.Task, id uint64, cb *linux.IOCallback, cbAddr host
 		// Check that it is an eventfd.
 		if _, ok := eventFile.FileOperations.(*eventfd.EventOperations); !ok {
 			// Not an event FD.
-			return syserror.EINVAL
+			return linuxerr.EINVAL
 		}
 	}
 
@@ -300,14 +300,14 @@ func submitCallback(t *kernel.Task, id uint64, cb *linux.IOCallback, cbAddr host
 	switch cb.OpCode {
 	case linux.IOCB_CMD_PREAD, linux.IOCB_CMD_PREADV, linux.IOCB_CMD_PWRITE, linux.IOCB_CMD_PWRITEV:
 		if cb.Offset < 0 {
-			return syserror.EINVAL
+			return linuxerr.EINVAL
 		}
 	}
 
 	// Prepare the request.
 	ctx, ok := t.MemoryManager().LookupAIOContext(t, id)
 	if !ok {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	if err := ctx.Prepare(); err != nil {
 		return err
@@ -336,7 +336,7 @@ func IoSubmit(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysc
 	addr := args[2].Pointer()
 
 	if nrEvents < 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	for i := int32(0); i < nrEvents; i++ {
diff --git a/pkg/sentry/syscalls/linux/sys_capability.go b/pkg/sentry/syscalls/linux/sys_capability.go
index d3b85e11b..782bcb94f 100644
--- a/pkg/sentry/syscalls/linux/sys_capability.go
+++ b/pkg/sentry/syscalls/linux/sys_capability.go
@@ -16,6 +16,7 @@ package linux
 
 import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
 	"gvisor.dev/gvisor/pkg/sentry/kernel"
 	"gvisor.dev/gvisor/pkg/sentry/kernel/auth"
@@ -24,7 +25,7 @@ import (
 
 func lookupCaps(t *kernel.Task, tid kernel.ThreadID) (permitted, inheritable, effective auth.CapabilitySet, err error) {
 	if tid < 0 {
-		err = syserror.EINVAL
+		err = linuxerr.EINVAL
 		return
 	}
 	if tid > 0 {
@@ -97,7 +98,7 @@ func Capget(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscal
 			return 0, nil, err
 		}
 		if dataAddr != 0 {
-			return 0, nil, syserror.EINVAL
+			return 0, nil, linuxerr.EINVAL
 		}
 		return 0, nil, nil
 	}
@@ -144,6 +145,6 @@ func Capset(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscal
 		if _, err := hdr.CopyOut(t, hdrAddr); err != nil {
 			return 0, nil, err
 		}
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 }
diff --git a/pkg/sentry/syscalls/linux/sys_epoll.go b/pkg/sentry/syscalls/linux/sys_epoll.go
index 69cbc98d0..daa151bb4 100644
--- a/pkg/sentry/syscalls/linux/sys_epoll.go
+++ b/pkg/sentry/syscalls/linux/sys_epoll.go
@@ -16,6 +16,7 @@ package linux
 
 import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
 	"gvisor.dev/gvisor/pkg/sentry/kernel"
@@ -31,7 +32,7 @@ import (
 func EpollCreate1(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallControl, error) {
 	flags := args[0].Int()
 	if flags & ^linux.EPOLL_CLOEXEC != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	closeOnExec := flags&linux.EPOLL_CLOEXEC != 0
@@ -48,7 +49,7 @@ func EpollCreate(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.S
 	size := args[0].Int()
 
 	if size <= 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	fd, err := syscalls.CreateEpoll(t, false)
@@ -101,7 +102,7 @@ func EpollCtl(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysc
 		mask |= waiter.EventHUp | waiter.EventErr
 		return 0, nil, syscalls.UpdateEpoll(t, epfd, fd, flags, mask, data)
 	default:
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 }
 
diff --git a/pkg/sentry/syscalls/linux/sys_eventfd.go b/pkg/sentry/syscalls/linux/sys_eventfd.go
index 3b4f879e4..7ba9a755e 100644
--- a/pkg/sentry/syscalls/linux/sys_eventfd.go
+++ b/pkg/sentry/syscalls/linux/sys_eventfd.go
@@ -16,11 +16,11 @@ package linux
 
 import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
 	"gvisor.dev/gvisor/pkg/sentry/fs"
 	"gvisor.dev/gvisor/pkg/sentry/kernel"
 	"gvisor.dev/gvisor/pkg/sentry/kernel/eventfd"
-	"gvisor.dev/gvisor/pkg/syserror"
 )
 
 // Eventfd2 implements linux syscall eventfd2(2).
@@ -30,7 +30,7 @@ func Eventfd2(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysc
 	allOps := uint(linux.EFD_SEMAPHORE | linux.EFD_NONBLOCK | linux.EFD_CLOEXEC)
 
 	if flags & ^allOps != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	event := eventfd.New(t, uint64(initVal), flags&linux.EFD_SEMAPHORE != 0)
diff --git a/pkg/sentry/syscalls/linux/sys_file.go b/pkg/sentry/syscalls/linux/sys_file.go
index 6109a2d8c..3d45341f2 100644
--- a/pkg/sentry/syscalls/linux/sys_file.go
+++ b/pkg/sentry/syscalls/linux/sys_file.go
@@ -276,7 +276,7 @@ func mknodAt(t *kernel.Task, dirFD int32, addr hostarch.Addr, mode linux.FileMod
 		default:
 			// "EINVAL - mode requested creation of something other than a
 			// regular file, device special file, FIFO or socket." - mknod(2)
-			return syserror.EINVAL
+			return linuxerr.EINVAL
 		}
 	})
 }
@@ -528,7 +528,7 @@ func accessAt(t *kernel.Task, dirFD int32, addr hostarch.Addr, mode uint) error
 
 	// Sanity check the mode.
 	if mode&^(rOK|wOK|xOK) != 0 {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 
 	return fileOpOn(t, dirFD, path, true /* resolve */, func(root *fs.Dirent, d *fs.Dirent, _ uint) error {
@@ -844,7 +844,7 @@ func Dup3(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallC
 	flags := args[2].Uint()
 
 	if oldfd == newfd {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	oldFile := t.GetFile(oldfd)
@@ -906,7 +906,7 @@ func fSetOwn(t *kernel.Task, fd int, file *fs.File, who int32) error {
 	if who < 0 {
 		// Check for overflow before flipping the sign.
 		if who-1 > who {
-			return syserror.EINVAL
+			return linuxerr.EINVAL
 		}
 		pg := t.PIDNamespace().ProcessGroupWithID(kernel.ProcessGroupID(-who))
 		a.SetOwnerProcessGroup(t, pg)
@@ -977,7 +977,7 @@ func Fcntl(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscall
 		case 2:
 			sw = fs.SeekEnd
 		default:
-			return 0, nil, syserror.EINVAL
+			return 0, nil, linuxerr.EINVAL
 		}
 
 		// Compute the lock offset.
@@ -996,7 +996,7 @@ func Fcntl(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscall
 			}
 			off = uattr.Size
 		default:
-			return 0, nil, syserror.EINVAL
+			return 0, nil, linuxerr.EINVAL
 		}
 
 		// Compute the lock range.
@@ -1044,7 +1044,7 @@ func Fcntl(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscall
 			file.Dirent.Inode.LockCtx.Posix.UnlockRegion(t.FDTable(), rng)
 			return 0, nil, nil
 		default:
-			return 0, nil, syserror.EINVAL
+			return 0, nil, linuxerr.EINVAL
 		}
 	case linux.F_GETOWN:
 		return uintptr(fGetOwn(t, file)), nil, nil
@@ -1086,7 +1086,7 @@ func Fcntl(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscall
 			a.SetOwnerProcessGroup(t, pg)
 			return 0, nil, nil
 		default:
-			return 0, nil, syserror.EINVAL
+			return 0, nil, linuxerr.EINVAL
 		}
 	case linux.F_GET_SEALS:
 		val, err := tmpfs.GetSeals(file.Dirent.Inode)
@@ -1100,14 +1100,14 @@ func Fcntl(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscall
 	case linux.F_GETPIPE_SZ:
 		sz, ok := file.FileOperations.(fs.FifoSizer)
 		if !ok {
-			return 0, nil, syserror.EINVAL
+			return 0, nil, linuxerr.EINVAL
 		}
 		size, err := sz.FifoSize(t, file)
 		return uintptr(size), nil, err
 	case linux.F_SETPIPE_SZ:
 		sz, ok := file.FileOperations.(fs.FifoSizer)
 		if !ok {
-			return 0, nil, syserror.EINVAL
+			return 0, nil, linuxerr.EINVAL
 		}
 		n, err := sz.SetFifoSize(int64(args[2].Int()))
 		return uintptr(n), nil, err
@@ -1119,7 +1119,7 @@ func Fcntl(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscall
 		return 0, nil, a.SetSignal(linux.Signal(args[2].Int()))
 	default:
 		// Everything else is not yet supported.
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 }
 
@@ -1132,7 +1132,7 @@ func Fadvise64(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sys
 
 	// Note: offset is allowed to be negative.
 	if length < 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	file := t.GetFile(fd)
@@ -1154,7 +1154,7 @@ func Fadvise64(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sys
 	case linux.POSIX_FADV_DONTNEED:
 	case linux.POSIX_FADV_NOREUSE:
 	default:
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Sure, whatever.
@@ -1237,7 +1237,7 @@ func rmdirAt(t *kernel.Task, dirFD int32, addr hostarch.Addr) error {
 		// dot vs. double dots.
 		switch name {
 		case ".":
-			return syserror.EINVAL
+			return linuxerr.EINVAL
 		case "..":
 			return syserror.ENOTEMPTY
 		}
@@ -1432,7 +1432,7 @@ func Linkat(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscal
 
 	// Sanity check flags.
 	if flags&^(linux.AT_SYMLINK_FOLLOW|linux.AT_EMPTY_PATH) != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	resolve := flags&linux.AT_SYMLINK_FOLLOW == linux.AT_SYMLINK_FOLLOW
@@ -1466,7 +1466,7 @@ func readlinkAt(t *kernel.Task, dirFD int32, addr hostarch.Addr, bufAddr hostarc
 
 		s, err := d.Inode.Readlink(t)
 		if linuxerr.Equals(linuxerr.ENOLINK, err) {
-			return syserror.EINVAL
+			return linuxerr.EINVAL
 		}
 		if err != nil {
 			return err
@@ -1558,7 +1558,7 @@ func Truncate(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysc
 	length := args[1].Int64()
 
 	if length < 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	path, dirPath, err := copyInPath(t, addr, false /* allowEmpty */)
@@ -1566,7 +1566,7 @@ func Truncate(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysc
 		return 0, nil, err
 	}
 	if dirPath {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	if uint64(length) >= t.ThreadGroup().Limits().Get(limits.FileSize).Cur {
@@ -1584,7 +1584,7 @@ func Truncate(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysc
 		// In contrast to open(O_TRUNC), truncate(2) is only valid for file
 		// types.
 		if !fs.IsFile(d.Inode.StableAttr) {
-			return syserror.EINVAL
+			return linuxerr.EINVAL
 		}
 
 		// Reject truncation if the access permissions do not allow truncation.
@@ -1618,18 +1618,18 @@ func Ftruncate(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sys
 	// Reject truncation if the file flags do not permit this operation.
 	// This is different from truncate(2) above.
 	if !file.Flags().Write {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// In contrast to open(O_TRUNC), truncate(2) is only valid for file
 	// types. Note that this is different from truncate(2) above, where a
 	// directory returns EISDIR.
 	if !fs.IsFile(file.Dirent.Inode.StableAttr) {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	if length < 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	if uint64(length) >= t.ThreadGroup().Limits().Get(limits.FileSize).Cur {
@@ -1683,7 +1683,7 @@ func chown(t *kernel.Task, d *fs.Dirent, uid auth.UID, gid auth.GID) error {
 		kuid := c.UserNamespace.MapToKUID(uid)
 		// Valid UID must be supplied if UID is to be changed.
 		if !kuid.Ok() {
-			return syserror.EINVAL
+			return linuxerr.EINVAL
 		}
 
 		// "Only a privileged process (CAP_CHOWN) may change the owner
@@ -1707,7 +1707,7 @@ func chown(t *kernel.Task, d *fs.Dirent, uid auth.UID, gid auth.GID) error {
 		kgid := c.UserNamespace.MapToKGID(gid)
 		// Valid GID must be supplied if GID is to be changed.
 		if !kgid.Ok() {
-			return syserror.EINVAL
+			return linuxerr.EINVAL
 		}
 
 		// "The owner of a file may change the group of the file to any
@@ -1809,7 +1809,7 @@ func Fchownat(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysc
 	flags := args[4].Int()
 
 	if flags&^(linux.AT_EMPTY_PATH|linux.AT_SYMLINK_NOFOLLOW) != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	return 0, nil, chownAt(t, dirFD, addr, flags&linux.AT_SYMLINK_NOFOLLOW == 0, flags&linux.AT_EMPTY_PATH != 0, uid, gid)
@@ -1914,7 +1914,7 @@ func utimes(t *kernel.Task, dirFD int32, addr hostarch.Addr, ts fs.TimeSpec, res
 	if addr == 0 && dirFD != linux.AT_FDCWD {
 		if !resolve {
 			// Linux returns EINVAL in this case. See utimes.c.
-			return syserror.EINVAL
+			return linuxerr.EINVAL
 		}
 		f := t.GetFile(dirFD)
 		if f == nil {
@@ -1997,7 +1997,7 @@ func Utimensat(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sys
 			return 0, nil, err
 		}
 		if !timespecIsValid(times[0]) || !timespecIsValid(times[1]) {
-			return 0, nil, syserror.EINVAL
+			return 0, nil, linuxerr.EINVAL
 		}
 
 		// If both are UTIME_OMIT, this is a noop.
@@ -2032,7 +2032,7 @@ func Futimesat(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sys
 		}
 		if times[0].Usec >= 1e6 || times[0].Usec < 0 ||
 			times[1].Usec >= 1e6 || times[1].Usec < 0 {
-			return 0, nil, syserror.EINVAL
+			return 0, nil, linuxerr.EINVAL
 		}
 
 		ts = fs.TimeSpec{
@@ -2118,7 +2118,7 @@ func Fallocate(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sys
 	defer file.DecRef(t)
 
 	if offset < 0 || length <= 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 	if mode != 0 {
 		t.Kernel().EmitUnimplementedEvent(t)
@@ -2208,7 +2208,7 @@ func Flock(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscall
 		file.Dirent.Inode.LockCtx.BSD.UnlockRegion(file, rng)
 	default:
 		// flock(2): EINVAL operation is invalid.
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	return 0, nil, nil
@@ -2227,7 +2227,7 @@ func MemfdCreate(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.S
 
 	if flags&^memfdAllFlags != 0 {
 		// Unknown bits in flags.
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	allowSeals := flags&linux.MFD_ALLOW_SEALING != 0
@@ -2238,7 +2238,7 @@ func MemfdCreate(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.S
 		return 0, nil, err
 	}
 	if len(name) > memfdMaxNameLen {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 	name = memfdPrefix + name
 
diff --git a/pkg/sentry/syscalls/linux/sys_futex.go b/pkg/sentry/syscalls/linux/sys_futex.go
index eeea1613b..a9e9126cc 100644
--- a/pkg/sentry/syscalls/linux/sys_futex.go
+++ b/pkg/sentry/syscalls/linux/sys_futex.go
@@ -18,6 +18,7 @@ import (
 	"time"
 
 	"gvisor.dev/gvisor/pkg/abi/linux"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
 	"gvisor.dev/gvisor/pkg/sentry/kernel"
@@ -210,7 +211,7 @@ func Futex(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscall
 			// WAIT_BITSET uses an absolute timeout which is either
 			// CLOCK_MONOTONIC or CLOCK_REALTIME.
 			if mask == 0 {
-				return 0, nil, syserror.EINVAL
+				return 0, nil, linuxerr.EINVAL
 			}
 			n, err := futexWaitAbsolute(t, clockRealtime, timespec, forever, addr, private, uint32(val), mask)
 			return n, nil, err
@@ -224,7 +225,7 @@ func Futex(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscall
 
 	case linux.FUTEX_WAKE_BITSET:
 		if mask == 0 {
-			return 0, nil, syserror.EINVAL
+			return 0, nil, linuxerr.EINVAL
 		}
 		if val <= 0 {
 			// The Linux kernel wakes one waiter even if val is
@@ -295,7 +296,7 @@ func SetRobustList(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel
 	length := args[1].SizeT()
 
 	if length != uint(linux.SizeOfRobustListHead) {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 	t.SetRobustList(head)
 	return 0, nil, nil
@@ -310,7 +311,7 @@ func GetRobustList(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel
 	sizeAddr := args[2].Pointer()
 
 	if tid < 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	ot := t
diff --git a/pkg/sentry/syscalls/linux/sys_getdents.go b/pkg/sentry/syscalls/linux/sys_getdents.go
index bbba71d8f..355fbd766 100644
--- a/pkg/sentry/syscalls/linux/sys_getdents.go
+++ b/pkg/sentry/syscalls/linux/sys_getdents.go
@@ -19,6 +19,7 @@ import (
 	"io"
 
 	"gvisor.dev/gvisor/pkg/abi/linux"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
 	"gvisor.dev/gvisor/pkg/sentry/fs"
@@ -38,7 +39,7 @@ func Getdents(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysc
 	minSize := int(smallestDirent(t.Arch()))
 	if size < minSize {
 		// size is smaller than smallest possible dirent.
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	n, err := getdents(t, fd, addr, size, (*dirent).Serialize)
@@ -54,7 +55,7 @@ func Getdents64(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sy
 	minSize := int(smallestDirent64(t.Arch()))
 	if size < minSize {
 		// size is smaller than smallest possible dirent.
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	n, err := getdents(t, fd, addr, size, (*dirent).Serialize64)
diff --git a/pkg/sentry/syscalls/linux/sys_identity.go b/pkg/sentry/syscalls/linux/sys_identity.go
index a29d307e5..50fcadb58 100644
--- a/pkg/sentry/syscalls/linux/sys_identity.go
+++ b/pkg/sentry/syscalls/linux/sys_identity.go
@@ -15,10 +15,10 @@
 package linux
 
 import (
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
 	"gvisor.dev/gvisor/pkg/sentry/kernel"
 	"gvisor.dev/gvisor/pkg/sentry/kernel/auth"
-	"gvisor.dev/gvisor/pkg/syserror"
 )
 
 const (
@@ -142,7 +142,7 @@ func Setresgid(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sys
 func Getgroups(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallControl, error) {
 	size := int(args[0].Int())
 	if size < 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 	kgids := t.Credentials().ExtraKGIDs
 	// "If size is zero, list is not modified, but the total number of
@@ -151,7 +151,7 @@ func Getgroups(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sys
 		return uintptr(len(kgids)), nil, nil
 	}
 	if size < len(kgids) {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 	gids := make([]auth.GID, len(kgids))
 	for i, kgid := range kgids {
@@ -167,7 +167,7 @@ func Getgroups(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sys
 func Setgroups(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallControl, error) {
 	size := args[0].Int()
 	if size < 0 || size > maxNGroups {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 	if size == 0 {
 		return 0, nil, t.SetExtraGIDs(nil)
diff --git a/pkg/sentry/syscalls/linux/sys_inotify.go b/pkg/sentry/syscalls/linux/sys_inotify.go
index cf47bb9dd..48c8dbdca 100644
--- a/pkg/sentry/syscalls/linux/sys_inotify.go
+++ b/pkg/sentry/syscalls/linux/sys_inotify.go
@@ -16,6 +16,7 @@ package linux
 
 import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
 	"gvisor.dev/gvisor/pkg/sentry/fs"
 	"gvisor.dev/gvisor/pkg/sentry/fs/anon"
@@ -30,7 +31,7 @@ func InotifyInit1(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.
 	flags := int(args[0].Int())
 
 	if flags&^allFlags != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	dirent := fs.NewDirent(t, anon.NewInode(t), "inotify")
@@ -72,7 +73,7 @@ func fdToInotify(t *kernel.Task, fd int32) (*fs.Inotify, *fs.File, error) {
 	if !ok {
 		// Not an inotify fd.
 		file.DecRef(t)
-		return nil, nil, syserror.EINVAL
+		return nil, nil, linuxerr.EINVAL
 	}
 
 	return ino, file, nil
@@ -91,7 +92,7 @@ func InotifyAddWatch(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kern
 	// "EINVAL: The given event mask contains no valid events."
 	// -- inotify_add_watch(2)
 	if validBits := mask & linux.ALL_INOTIFY_BITS; validBits == 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	ino, file, err := fdToInotify(t, fd)
diff --git a/pkg/sentry/syscalls/linux/sys_lseek.go b/pkg/sentry/syscalls/linux/sys_lseek.go
index 0046347cb..c16c63ecc 100644
--- a/pkg/sentry/syscalls/linux/sys_lseek.go
+++ b/pkg/sentry/syscalls/linux/sys_lseek.go
@@ -15,6 +15,7 @@
 package linux
 
 import (
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
 	"gvisor.dev/gvisor/pkg/sentry/fs"
 	"gvisor.dev/gvisor/pkg/sentry/kernel"
@@ -44,7 +45,7 @@ func Lseek(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscall
 	case 2:
 		sw = fs.SeekEnd
 	default:
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	offset, serr := file.Seek(t, sw, offset)
diff --git a/pkg/sentry/syscalls/linux/sys_membarrier.go b/pkg/sentry/syscalls/linux/sys_membarrier.go
index 63ee5d435..4b67f2536 100644
--- a/pkg/sentry/syscalls/linux/sys_membarrier.go
+++ b/pkg/sentry/syscalls/linux/sys_membarrier.go
@@ -16,6 +16,7 @@ package linux
 
 import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
 	"gvisor.dev/gvisor/pkg/sentry/kernel"
 	"gvisor.dev/gvisor/pkg/syserror"
@@ -29,7 +30,7 @@ func Membarrier(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sy
 	switch cmd {
 	case linux.MEMBARRIER_CMD_QUERY:
 		if flags != 0 {
-			return 0, nil, syserror.EINVAL
+			return 0, nil, linuxerr.EINVAL
 		}
 		var supportedCommands uintptr
 		if t.Kernel().Platform.HaveGlobalMemoryBarrier() {
@@ -46,10 +47,10 @@ func Membarrier(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sy
 		return supportedCommands, nil, nil
 	case linux.MEMBARRIER_CMD_GLOBAL, linux.MEMBARRIER_CMD_GLOBAL_EXPEDITED, linux.MEMBARRIER_CMD_PRIVATE_EXPEDITED:
 		if flags != 0 {
-			return 0, nil, syserror.EINVAL
+			return 0, nil, linuxerr.EINVAL
 		}
 		if !t.Kernel().Platform.HaveGlobalMemoryBarrier() {
-			return 0, nil, syserror.EINVAL
+			return 0, nil, linuxerr.EINVAL
 		}
 		if cmd == linux.MEMBARRIER_CMD_PRIVATE_EXPEDITED && !t.MemoryManager().IsMembarrierPrivateEnabled() {
 			return 0, nil, syserror.EPERM
@@ -57,28 +58,28 @@ func Membarrier(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sy
 		return 0, nil, t.Kernel().Platform.GlobalMemoryBarrier()
 	case linux.MEMBARRIER_CMD_REGISTER_GLOBAL_EXPEDITED:
 		if flags != 0 {
-			return 0, nil, syserror.EINVAL
+			return 0, nil, linuxerr.EINVAL
 		}
 		if !t.Kernel().Platform.HaveGlobalMemoryBarrier() {
-			return 0, nil, syserror.EINVAL
+			return 0, nil, linuxerr.EINVAL
 		}
 		// no-op
 		return 0, nil, nil
 	case linux.MEMBARRIER_CMD_REGISTER_PRIVATE_EXPEDITED:
 		if flags != 0 {
-			return 0, nil, syserror.EINVAL
+			return 0, nil, linuxerr.EINVAL
 		}
 		if !t.Kernel().Platform.HaveGlobalMemoryBarrier() {
-			return 0, nil, syserror.EINVAL
+			return 0, nil, linuxerr.EINVAL
 		}
 		t.MemoryManager().EnableMembarrierPrivate()
 		return 0, nil, nil
 	case linux.MEMBARRIER_CMD_PRIVATE_EXPEDITED_RSEQ:
 		if flags&^linux.MEMBARRIER_CMD_FLAG_CPU != 0 {
-			return 0, nil, syserror.EINVAL
+			return 0, nil, linuxerr.EINVAL
 		}
 		if !t.RSeqAvailable() {
-			return 0, nil, syserror.EINVAL
+			return 0, nil, linuxerr.EINVAL
 		}
 		if !t.MemoryManager().IsMembarrierRSeqEnabled() {
 			return 0, nil, syserror.EPERM
@@ -88,16 +89,16 @@ func Membarrier(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sy
 		return 0, nil, t.Kernel().Platform.PreemptAllCPUs()
 	case linux.MEMBARRIER_CMD_REGISTER_PRIVATE_EXPEDITED_RSEQ:
 		if flags != 0 {
-			return 0, nil, syserror.EINVAL
+			return 0, nil, linuxerr.EINVAL
 		}
 		if !t.RSeqAvailable() {
-			return 0, nil, syserror.EINVAL
+			return 0, nil, linuxerr.EINVAL
 		}
 		t.MemoryManager().EnableMembarrierRSeq()
 		return 0, nil, nil
 	default:
 		// Probably a command we don't implement.
 		t.Kernel().EmitUnimplementedEvent(t)
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 }
diff --git a/pkg/sentry/syscalls/linux/sys_mempolicy.go b/pkg/sentry/syscalls/linux/sys_mempolicy.go
index 6d27f4292..62ec3e27f 100644
--- a/pkg/sentry/syscalls/linux/sys_mempolicy.go
+++ b/pkg/sentry/syscalls/linux/sys_mempolicy.go
@@ -18,6 +18,7 @@ import (
 	"fmt"
 
 	"gvisor.dev/gvisor/pkg/abi/linux"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
 	"gvisor.dev/gvisor/pkg/sentry/kernel"
@@ -43,7 +44,7 @@ func copyInNodemask(t *kernel.Task, addr hostarch.Addr, maxnode uint32) (uint64,
 	// maxnode-1, not maxnode, as the number of bits.
 	bits := maxnode - 1
 	if bits > hostarch.PageSize*8 { // also handles overflow from maxnode == 0
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 	if bits == 0 {
 		return 0, nil
@@ -58,12 +59,12 @@ func copyInNodemask(t *kernel.Task, addr hostarch.Addr, maxnode uint32) (uint64,
 	// Check that only allowed bits in the first unsigned long in the nodemask
 	// are set.
 	if val&^allowedNodemask != 0 {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 	// Check that all remaining bits in the nodemask are 0.
 	for i := 8; i < len(buf); i++ {
 		if buf[i] != 0 {
-			return 0, syserror.EINVAL
+			return 0, linuxerr.EINVAL
 		}
 	}
 	return val, nil
@@ -74,7 +75,7 @@ func copyOutNodemask(t *kernel.Task, addr hostarch.Addr, maxnode uint32, val uin
 	// bits.
 	bits := maxnode - 1
 	if bits > hostarch.PageSize*8 { // also handles overflow from maxnode == 0
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	if bits == 0 {
 		return nil
@@ -110,7 +111,7 @@ func GetMempolicy(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.
 	flags := args[4].Uint()
 
 	if flags&^(linux.MPOL_F_NODE|linux.MPOL_F_ADDR|linux.MPOL_F_MEMS_ALLOWED) != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 	nodeFlag := flags&linux.MPOL_F_NODE != 0
 	addrFlag := flags&linux.MPOL_F_ADDR != 0
@@ -119,7 +120,7 @@ func GetMempolicy(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.
 	// "EINVAL: The value specified by maxnode is less than the number of node
 	// IDs supported by the system." - get_mempolicy(2)
 	if nodemask != 0 && maxnode < maxNodes {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// "If flags specifies MPOL_F_MEMS_ALLOWED [...], the mode argument is
@@ -130,7 +131,7 @@ func GetMempolicy(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.
 		// "It is not permitted to combine MPOL_F_MEMS_ALLOWED with either
 		// MPOL_F_ADDR or MPOL_F_NODE."
 		if nodeFlag || addrFlag {
-			return 0, nil, syserror.EINVAL
+			return 0, nil, linuxerr.EINVAL
 		}
 		if err := copyOutNodemask(t, nodemask, maxnode, allowedNodemask); err != nil {
 			return 0, nil, err
@@ -184,7 +185,7 @@ func GetMempolicy(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.
 	// mm/mempolicy.c:do_get_mempolicy() doesn't special-case NULL; it will
 	// just (usually) fail to find a VMA at address 0 and return EFAULT.
 	if addr != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// "If flags is specified as 0, then information about the calling thread's
@@ -198,7 +199,7 @@ func GetMempolicy(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.
 	policy, nodemaskVal := t.NumaPolicy()
 	if nodeFlag {
 		if policy&^linux.MPOL_MODE_FLAGS != linux.MPOL_INTERLEAVE {
-			return 0, nil, syserror.EINVAL
+			return 0, nil, linuxerr.EINVAL
 		}
 		policy = linux.MPOL_DEFAULT // maxNodes == 1
 	}
@@ -240,7 +241,7 @@ func Mbind(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscall
 	flags := args[5].Uint()
 
 	if flags&^linux.MPOL_MF_VALID != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 	// "If MPOL_MF_MOVE_ALL is passed in flags ... [the] calling thread must be
 	// privileged (CAP_SYS_NICE) to use this flag." - mbind(2)
@@ -264,11 +265,11 @@ func copyInMempolicyNodemask(t *kernel.Task, modeWithFlags linux.NumaPolicy, nod
 	mode := linux.NumaPolicy(modeWithFlags &^ linux.MPOL_MODE_FLAGS)
 	if flags == linux.MPOL_MODE_FLAGS {
 		// Can't specify both mode flags simultaneously.
-		return 0, 0, syserror.EINVAL
+		return 0, 0, linuxerr.EINVAL
 	}
 	if mode < 0 || mode >= linux.MPOL_MAX {
 		// Must specify a valid mode.
-		return 0, 0, syserror.EINVAL
+		return 0, 0, linuxerr.EINVAL
 	}
 
 	var nodemaskVal uint64
@@ -285,22 +286,22 @@ func copyInMempolicyNodemask(t *kernel.Task, modeWithFlags linux.NumaPolicy, nod
 		// "nodemask must be specified as NULL." - set_mempolicy(2). This is inaccurate;
 		// Linux allows a nodemask to be specified, as long as it is empty.
 		if nodemaskVal != 0 {
-			return 0, 0, syserror.EINVAL
+			return 0, 0, linuxerr.EINVAL
 		}
 	case linux.MPOL_BIND, linux.MPOL_INTERLEAVE:
 		// These require a non-empty nodemask.
 		if nodemaskVal == 0 {
-			return 0, 0, syserror.EINVAL
+			return 0, 0, linuxerr.EINVAL
 		}
 	case linux.MPOL_PREFERRED:
 		// This permits an empty nodemask, as long as no flags are set.
 		if nodemaskVal == 0 && flags != 0 {
-			return 0, 0, syserror.EINVAL
+			return 0, 0, linuxerr.EINVAL
 		}
 	case linux.MPOL_LOCAL:
 		// This requires an empty nodemask and no flags set ...
 		if nodemaskVal != 0 || flags != 0 {
-			return 0, 0, syserror.EINVAL
+			return 0, 0, linuxerr.EINVAL
 		}
 		// ... and is implemented as MPOL_PREFERRED.
 		mode = linux.MPOL_PREFERRED
diff --git a/pkg/sentry/syscalls/linux/sys_mmap.go b/pkg/sentry/syscalls/linux/sys_mmap.go
index 70da0707d..74279c82b 100644
--- a/pkg/sentry/syscalls/linux/sys_mmap.go
+++ b/pkg/sentry/syscalls/linux/sys_mmap.go
@@ -18,13 +18,13 @@ import (
 	"bytes"
 
 	"gvisor.dev/gvisor/pkg/abi/linux"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
+	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
 	"gvisor.dev/gvisor/pkg/sentry/kernel"
 	"gvisor.dev/gvisor/pkg/sentry/memmap"
 	"gvisor.dev/gvisor/pkg/sentry/mm"
 	"gvisor.dev/gvisor/pkg/syserror"
-
-	"gvisor.dev/gvisor/pkg/hostarch"
 )
 
 // Brk implements linux syscall brk(2).
@@ -51,7 +51,7 @@ func Mmap(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallC
 
 	// Require exactly one of MAP_PRIVATE and MAP_SHARED.
 	if private == shared {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	opts := memmap.MMapOpts{
@@ -132,7 +132,7 @@ func Mremap(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscal
 	newAddr := args[4].Pointer()
 
 	if flags&^(linux.MREMAP_MAYMOVE|linux.MREMAP_FIXED) != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 	mayMove := flags&linux.MREMAP_MAYMOVE != 0
 	fixed := flags&linux.MREMAP_FIXED != 0
@@ -147,7 +147,7 @@ func Mremap(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscal
 	case !mayMove && fixed:
 		// "If MREMAP_FIXED is specified, then MREMAP_MAYMOVE must also be
 		// specified." - mremap(2)
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	rv, err := t.MemoryManager().MRemap(t, oldAddr, oldSize, newSize, mm.MRemapOpts{
@@ -178,7 +178,7 @@ func Madvise(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysca
 	// "The Linux implementation requires that the address addr be
 	// page-aligned, and allows length to be zero." - madvise(2)
 	if addr.RoundDown() != addr {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 	if length == 0 {
 		return 0, nil, nil
@@ -186,7 +186,7 @@ func Madvise(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysca
 	// Not explicitly stated: length need not be page-aligned.
 	lenAddr, ok := hostarch.Addr(length).RoundUp()
 	if !ok {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 	length = uint64(lenAddr)
 
@@ -217,7 +217,7 @@ func Madvise(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysca
 		return 0, nil, syserror.EPERM
 	default:
 		// If adv is not a valid value tell the caller.
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 }
 
@@ -228,7 +228,7 @@ func Mincore(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysca
 	vec := args[2].Pointer()
 
 	if addr != addr.RoundDown() {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 	// "The length argument need not be a multiple of the page size, but since
 	// residency information is returned for whole pages, length is effectively
@@ -265,11 +265,11 @@ func Msync(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscall
 	// semantics that are (currently) equivalent to specifying MS_ASYNC." -
 	// msync(2)
 	if flags&^(linux.MS_ASYNC|linux.MS_SYNC|linux.MS_INVALIDATE) != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 	sync := flags&linux.MS_SYNC != 0
 	if sync && flags&linux.MS_ASYNC != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 	err := t.MemoryManager().MSync(t, addr, uint64(length), mm.MSyncOpts{
 		Sync:       sync,
@@ -295,7 +295,7 @@ func Mlock2(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscal
 	flags := args[2].Int()
 
 	if flags&^(linux.MLOCK_ONFAULT) != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	mode := memmap.MLockEager
@@ -318,7 +318,7 @@ func Mlockall(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysc
 	flags := args[0].Int()
 
 	if flags&^(linux.MCL_CURRENT|linux.MCL_FUTURE|linux.MCL_ONFAULT) != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	mode := memmap.MLockEager
diff --git a/pkg/sentry/syscalls/linux/sys_mount.go b/pkg/sentry/syscalls/linux/sys_mount.go
index 864d2138c..8bf4e9f06 100644
--- a/pkg/sentry/syscalls/linux/sys_mount.go
+++ b/pkg/sentry/syscalls/linux/sys_mount.go
@@ -16,12 +16,12 @@ package linux
 
 import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
+	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
 	"gvisor.dev/gvisor/pkg/sentry/fs"
 	"gvisor.dev/gvisor/pkg/sentry/kernel"
 	"gvisor.dev/gvisor/pkg/syserror"
-
-	"gvisor.dev/gvisor/pkg/hostarch"
 )
 
 // Mount implements Linux syscall mount(2).
@@ -83,7 +83,7 @@ func Mount(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscall
 	// unknown or unsupported flags are passed. Since we don't implement
 	// everything, we fail explicitly on flags that are unimplemented.
 	if flags&(unsupportedOps|unsupportedFlags) != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	rsys, ok := fs.FindFilesystem(fsType)
@@ -107,7 +107,7 @@ func Mount(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscall
 
 	rootInode, err := rsys.Mount(t, sourcePath, superFlags, data, nil)
 	if err != nil {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	if err := fileOpOn(t, linux.AT_FDCWD, targetPath, true /* resolve */, func(root *fs.Dirent, d *fs.Dirent, _ uint) error {
@@ -130,7 +130,7 @@ func Umount2(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysca
 
 	const unsupported = linux.MNT_FORCE | linux.MNT_EXPIRE
 	if flags&unsupported != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	path, _, err := copyInPath(t, addr, false /* allowEmpty */)
diff --git a/pkg/sentry/syscalls/linux/sys_pipe.go b/pkg/sentry/syscalls/linux/sys_pipe.go
index d95034347..5925c2263 100644
--- a/pkg/sentry/syscalls/linux/sys_pipe.go
+++ b/pkg/sentry/syscalls/linux/sys_pipe.go
@@ -16,13 +16,13 @@ package linux
 
 import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/marshal/primitive"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
 	"gvisor.dev/gvisor/pkg/sentry/fs"
 	"gvisor.dev/gvisor/pkg/sentry/kernel"
 	"gvisor.dev/gvisor/pkg/sentry/kernel/pipe"
-	"gvisor.dev/gvisor/pkg/syserror"
 )
 
 // LINT.IfChange
@@ -30,7 +30,7 @@ import (
 // pipe2 implements the actual system call with flags.
 func pipe2(t *kernel.Task, addr hostarch.Addr, flags uint) (uintptr, error) {
 	if flags&^(linux.O_NONBLOCK|linux.O_CLOEXEC) != 0 {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 	r, w := pipe.NewConnectedPipe(t, pipe.DefaultPipeSize)
 
diff --git a/pkg/sentry/syscalls/linux/sys_poll.go b/pkg/sentry/syscalls/linux/sys_poll.go
index 024632475..f2056d850 100644
--- a/pkg/sentry/syscalls/linux/sys_poll.go
+++ b/pkg/sentry/syscalls/linux/sys_poll.go
@@ -158,7 +158,7 @@ func pollBlock(t *kernel.Task, pfd []linux.PollFD, timeout time.Duration) (time.
 // CopyInPollFDs copies an array of struct pollfd unless nfds exceeds the max.
 func CopyInPollFDs(t *kernel.Task, addr hostarch.Addr, nfds uint) ([]linux.PollFD, error) {
 	if uint64(nfds) > t.ThreadGroup().Limits().GetCapped(limits.NumberOfFiles, fileCap) {
-		return nil, syserror.EINVAL
+		return nil, linuxerr.EINVAL
 	}
 
 	pfd := make([]linux.PollFD, nfds)
@@ -218,7 +218,7 @@ func CopyInFDSet(t *kernel.Task, addr hostarch.Addr, nBytes, nBitsInLastPartialB
 
 func doSelect(t *kernel.Task, nfds int, readFDs, writeFDs, exceptFDs hostarch.Addr, timeout time.Duration) (uintptr, error) {
 	if nfds < 0 || nfds > fileCap {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 
 	// Calculate the size of the fd sets (one bit per fd).
@@ -486,7 +486,7 @@ func Select(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscal
 			return 0, nil, err
 		}
 		if timeval.Sec < 0 || timeval.Usec < 0 {
-			return 0, nil, syserror.EINVAL
+			return 0, nil, linuxerr.EINVAL
 		}
 		timeout = time.Duration(timeval.ToNsecCapped())
 	}
diff --git a/pkg/sentry/syscalls/linux/sys_prctl.go b/pkg/sentry/syscalls/linux/sys_prctl.go
index 30c15af4a..534f1e632 100644
--- a/pkg/sentry/syscalls/linux/sys_prctl.go
+++ b/pkg/sentry/syscalls/linux/sys_prctl.go
@@ -39,7 +39,7 @@ func Prctl(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscall
 	case linux.PR_SET_PDEATHSIG:
 		sig := linux.Signal(args[1].Int())
 		if sig != 0 && !sig.IsValid() {
-			return 0, nil, syserror.EINVAL
+			return 0, nil, linuxerr.EINVAL
 		}
 		t.SetParentDeathSignal(sig)
 		return 0, nil, nil
@@ -70,7 +70,7 @@ func Prctl(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscall
 			d = mm.UserDumpable
 		default:
 			// N.B. Userspace may not pass SUID_DUMP_ROOT.
-			return 0, nil, syserror.EINVAL
+			return 0, nil, linuxerr.EINVAL
 		}
 		t.MemoryManager().SetDumpability(d)
 		return 0, nil, nil
@@ -91,7 +91,7 @@ func Prctl(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscall
 		} else if val == 1 {
 			t.SetKeepCaps(true)
 		} else {
-			return 0, nil, syserror.EINVAL
+			return 0, nil, linuxerr.EINVAL
 		}
 
 		return 0, nil, nil
@@ -156,12 +156,12 @@ func Prctl(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscall
 			t.Kernel().EmitUnimplementedEvent(t)
 			fallthrough
 		default:
-			return 0, nil, syserror.EINVAL
+			return 0, nil, linuxerr.EINVAL
 		}
 
 	case linux.PR_SET_NO_NEW_PRIVS:
 		if args[1].Int() != 1 || args[2].Int() != 0 || args[3].Int() != 0 || args[4].Int() != 0 {
-			return 0, nil, syserror.EINVAL
+			return 0, nil, linuxerr.EINVAL
 		}
 		// PR_SET_NO_NEW_PRIVS is assumed to always be set.
 		// See kernel.Task.updateCredsForExecLocked.
@@ -169,7 +169,7 @@ func Prctl(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscall
 
 	case linux.PR_GET_NO_NEW_PRIVS:
 		if args[1].Int() != 0 || args[2].Int() != 0 || args[3].Int() != 0 || args[4].Int() != 0 {
-			return 0, nil, syserror.EINVAL
+			return 0, nil, linuxerr.EINVAL
 		}
 		return 1, nil, nil
 
@@ -185,7 +185,7 @@ func Prctl(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscall
 		default:
 			tracer := t.PIDNamespace().TaskWithID(kernel.ThreadID(pid))
 			if tracer == nil {
-				return 0, nil, syserror.EINVAL
+				return 0, nil, linuxerr.EINVAL
 			}
 			t.SetYAMAException(tracer)
 			return 0, nil, nil
@@ -194,7 +194,7 @@ func Prctl(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscall
 	case linux.PR_SET_SECCOMP:
 		if args[1].Int() != linux.SECCOMP_MODE_FILTER {
 			// Unsupported mode.
-			return 0, nil, syserror.EINVAL
+			return 0, nil, linuxerr.EINVAL
 		}
 
 		return 0, nil, seccomp(t, linux.SECCOMP_SET_MODE_FILTER, 0, args[2].Pointer())
@@ -205,7 +205,7 @@ func Prctl(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscall
 	case linux.PR_CAPBSET_READ:
 		cp := linux.Capability(args[1].Uint64())
 		if !cp.Ok() {
-			return 0, nil, syserror.EINVAL
+			return 0, nil, linuxerr.EINVAL
 		}
 		var rv uintptr
 		if auth.CapabilitySetOf(cp)&t.Credentials().BoundingCaps != 0 {
@@ -216,7 +216,7 @@ func Prctl(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscall
 	case linux.PR_CAPBSET_DROP:
 		cp := linux.Capability(args[1].Uint64())
 		if !cp.Ok() {
-			return 0, nil, syserror.EINVAL
+			return 0, nil, linuxerr.EINVAL
 		}
 		return 0, nil, t.DropBoundingCapability(cp)
 
@@ -241,7 +241,7 @@ func Prctl(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscall
 		t.Kernel().EmitUnimplementedEvent(t)
 		fallthrough
 	default:
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	return 0, nil, nil
diff --git a/pkg/sentry/syscalls/linux/sys_random.go b/pkg/sentry/syscalls/linux/sys_random.go
index ae545f80f..ec6c80de5 100644
--- a/pkg/sentry/syscalls/linux/sys_random.go
+++ b/pkg/sentry/syscalls/linux/sys_random.go
@@ -18,14 +18,14 @@ import (
 	"io"
 	"math"
 
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
+	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/rand"
 	"gvisor.dev/gvisor/pkg/safemem"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
 	"gvisor.dev/gvisor/pkg/sentry/kernel"
 	"gvisor.dev/gvisor/pkg/syserror"
 	"gvisor.dev/gvisor/pkg/usermem"
-
-	"gvisor.dev/gvisor/pkg/hostarch"
 )
 
 const (
@@ -47,7 +47,7 @@ func GetRandom(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sys
 
 	// Flags are checked for validity but otherwise ignored. See above.
 	if flags & ^(_GRND_NONBLOCK|_GRND_RANDOM) != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	if length > math.MaxInt32 {
diff --git a/pkg/sentry/syscalls/linux/sys_read.go b/pkg/sentry/syscalls/linux/sys_read.go
index 0f9329fe8..4064467a9 100644
--- a/pkg/sentry/syscalls/linux/sys_read.go
+++ b/pkg/sentry/syscalls/linux/sys_read.go
@@ -59,7 +59,7 @@ func Read(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallC
 	// Check that the size is legitimate.
 	si := int(size)
 	if si < 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Get the destination of the read.
@@ -94,18 +94,18 @@ func Readahead(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sys
 
 	// Check that the size is valid.
 	if int(size) < 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Check that the offset is legitimate and does not overflow.
 	if offset < 0 || offset+int64(size) < 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Return EINVAL; if the underlying file type does not support readahead,
 	// then Linux will return EINVAL to indicate as much. In the future, we
 	// may extend this function to actually support readahead hints.
-	return 0, nil, syserror.EINVAL
+	return 0, nil, linuxerr.EINVAL
 }
 
 // Pread64 implements linux syscall pread64(2).
@@ -123,7 +123,7 @@ func Pread64(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysca
 
 	// Check that the offset is legitimate and does not overflow.
 	if offset < 0 || offset+int64(size) < 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Is reading at an offset supported?
@@ -139,7 +139,7 @@ func Pread64(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysca
 	// Check that the size is legitimate.
 	si := int(size)
 	if si < 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Get the destination of the read.
@@ -200,7 +200,7 @@ func Preadv(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscal
 
 	// Check that the offset is legitimate.
 	if offset < 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Is reading at an offset supported?
@@ -249,7 +249,7 @@ func Preadv2(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysca
 
 	// Check that the offset is legitimate.
 	if offset < -1 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Is reading at an offset supported?
diff --git a/pkg/sentry/syscalls/linux/sys_rlimit.go b/pkg/sentry/syscalls/linux/sys_rlimit.go
index e64246d57..ca78c2ab2 100644
--- a/pkg/sentry/syscalls/linux/sys_rlimit.go
+++ b/pkg/sentry/syscalls/linux/sys_rlimit.go
@@ -16,6 +16,7 @@ package linux
 
 import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/marshal"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
@@ -129,7 +130,7 @@ func Getrlimit(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sys
 	resource, ok := limits.FromLinuxResource[int(args[0].Int())]
 	if !ok {
 		// Return err; unknown limit.
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 	addr := args[1].Pointer()
 	rlim, err := newRlimit(t)
@@ -150,7 +151,7 @@ func Setrlimit(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sys
 	resource, ok := limits.FromLinuxResource[int(args[0].Int())]
 	if !ok {
 		// Return err; unknown limit.
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 	addr := args[1].Pointer()
 	rlim, err := newRlimit(t)
@@ -170,7 +171,7 @@ func Prlimit64(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sys
 	resource, ok := limits.FromLinuxResource[int(args[1].Int())]
 	if !ok {
 		// Return err; unknown limit.
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 	newRlimAddr := args[2].Pointer()
 	oldRlimAddr := args[3].Pointer()
@@ -185,7 +186,7 @@ func Prlimit64(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sys
 	}
 
 	if tid < 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 	ot := t
 	if tid > 0 {
diff --git a/pkg/sentry/syscalls/linux/sys_rseq.go b/pkg/sentry/syscalls/linux/sys_rseq.go
index 90db10ea6..5fe196647 100644
--- a/pkg/sentry/syscalls/linux/sys_rseq.go
+++ b/pkg/sentry/syscalls/linux/sys_rseq.go
@@ -16,6 +16,7 @@ package linux
 
 import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
 	"gvisor.dev/gvisor/pkg/sentry/kernel"
 	"gvisor.dev/gvisor/pkg/syserror"
@@ -43,6 +44,6 @@ func RSeq(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallC
 		return 0, nil, t.ClearRSeq(addr, length, signature)
 	default:
 		// Unknown flag.
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 }
diff --git a/pkg/sentry/syscalls/linux/sys_rusage.go b/pkg/sentry/syscalls/linux/sys_rusage.go
index ac5c98a54..a689abcc9 100644
--- a/pkg/sentry/syscalls/linux/sys_rusage.go
+++ b/pkg/sentry/syscalls/linux/sys_rusage.go
@@ -16,11 +16,11 @@ package linux
 
 import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
 	"gvisor.dev/gvisor/pkg/sentry/kernel"
 	ktime "gvisor.dev/gvisor/pkg/sentry/kernel/time"
 	"gvisor.dev/gvisor/pkg/sentry/usage"
-	"gvisor.dev/gvisor/pkg/syserror"
 )
 
 func getrusage(t *kernel.Task, which int32) linux.Rusage {
@@ -76,7 +76,7 @@ func Getrusage(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sys
 	addr := args[1].Pointer()
 
 	if which != linux.RUSAGE_SELF && which != linux.RUSAGE_CHILDREN && which != linux.RUSAGE_THREAD {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	ru := getrusage(t, which)
diff --git a/pkg/sentry/syscalls/linux/sys_sched.go b/pkg/sentry/syscalls/linux/sys_sched.go
index bfcf44b6f..b5e7b70b5 100644
--- a/pkg/sentry/syscalls/linux/sys_sched.go
+++ b/pkg/sentry/syscalls/linux/sys_sched.go
@@ -16,6 +16,7 @@ package linux
 
 import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
 	"gvisor.dev/gvisor/pkg/sentry/kernel"
 	"gvisor.dev/gvisor/pkg/syserror"
@@ -38,10 +39,10 @@ func SchedGetparam(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel
 	pid := args[0].Int()
 	param := args[1].Pointer()
 	if param == 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 	if pid < 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 	if pid != 0 && t.PIDNamespace().TaskWithID(kernel.ThreadID(pid)) == nil {
 		return 0, nil, syserror.ESRCH
@@ -58,7 +59,7 @@ func SchedGetparam(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel
 func SchedGetscheduler(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallControl, error) {
 	pid := args[0].Int()
 	if pid < 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 	if pid != 0 && t.PIDNamespace().TaskWithID(kernel.ThreadID(pid)) == nil {
 		return 0, nil, syserror.ESRCH
@@ -72,20 +73,20 @@ func SchedSetscheduler(t *kernel.Task, args arch.SyscallArguments) (uintptr, *ke
 	policy := args[1].Int()
 	param := args[2].Pointer()
 	if pid < 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 	if policy != onlyScheduler {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 	if pid != 0 && t.PIDNamespace().TaskWithID(kernel.ThreadID(pid)) == nil {
 		return 0, nil, syserror.ESRCH
 	}
 	var r SchedParam
 	if _, err := r.CopyIn(t, param); err != nil {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 	if r.schedPriority != onlyPriority {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 	return 0, nil, nil
 }
diff --git a/pkg/sentry/syscalls/linux/sys_seccomp.go b/pkg/sentry/syscalls/linux/sys_seccomp.go
index e16d6ff3f..b0dc84b8d 100644
--- a/pkg/sentry/syscalls/linux/sys_seccomp.go
+++ b/pkg/sentry/syscalls/linux/sys_seccomp.go
@@ -17,10 +17,10 @@ package linux
 import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/bpf"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
 	"gvisor.dev/gvisor/pkg/sentry/kernel"
-	"gvisor.dev/gvisor/pkg/syserror"
 )
 
 // userSockFprog is equivalent to Linux's struct sock_fprog on amd64.
@@ -44,7 +44,7 @@ func seccomp(t *kernel.Task, mode, flags uint64, addr hostarch.Addr) error {
 	// We only support SECCOMP_SET_MODE_FILTER at the moment.
 	if mode != linux.SECCOMP_SET_MODE_FILTER {
 		// Unsupported mode.
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 
 	tsync := flags&linux.SECCOMP_FILTER_FLAG_TSYNC != 0
@@ -52,7 +52,7 @@ func seccomp(t *kernel.Task, mode, flags uint64, addr hostarch.Addr) error {
 	// The only flag we support now is SECCOMP_FILTER_FLAG_TSYNC.
 	if flags&^linux.SECCOMP_FILTER_FLAG_TSYNC != 0 {
 		// Unsupported flag.
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 
 	var fprog userSockFprog
@@ -66,7 +66,7 @@ func seccomp(t *kernel.Task, mode, flags uint64, addr hostarch.Addr) error {
 	compiledFilter, err := bpf.Compile(filter)
 	if err != nil {
 		t.Debugf("Invalid seccomp-bpf filter: %v", err)
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 
 	return t.AppendSyscallFilter(compiledFilter, tsync)
diff --git a/pkg/sentry/syscalls/linux/sys_sem.go b/pkg/sentry/syscalls/linux/sys_sem.go
index cb320c536..d12a4303b 100644
--- a/pkg/sentry/syscalls/linux/sys_sem.go
+++ b/pkg/sentry/syscalls/linux/sys_sem.go
@@ -62,10 +62,10 @@ func Semtimedop(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sy
 	nsops := args[2].SizeT()
 	timespecAddr := args[3].Pointer()
 	if nsops <= 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 	if nsops > opsMax {
-		return 0, nil, syserror.E2BIG
+		return 0, nil, linuxerr.E2BIG
 	}
 
 	ops := make([]linux.Sembuf, nsops)
@@ -78,7 +78,7 @@ func Semtimedop(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sy
 		return 0, nil, err
 	}
 	if timeout.Sec < 0 || timeout.Nsec < 0 || timeout.Nsec >= 1e9 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	if err := semTimedOp(t, id, ops, true, timeout.ToDuration()); err != nil {
@@ -97,10 +97,10 @@ func Semop(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscall
 	nsops := args[2].SizeT()
 
 	if nsops <= 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 	if nsops > opsMax {
-		return 0, nil, syserror.E2BIG
+		return 0, nil, linuxerr.E2BIG
 	}
 
 	ops := make([]linux.Sembuf, nsops)
@@ -114,7 +114,7 @@ func semTimedOp(t *kernel.Task, id int32, ops []linux.Sembuf, haveTimeout bool,
 	set := t.IPCNamespace().SemaphoreRegistry().FindByID(id)
 
 	if set == nil {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	creds := auth.CredentialsFromContext(t)
 	pid := t.Kernel().GlobalInit().PIDNamespace().IDOfThreadGroup(t.ThreadGroup())
@@ -233,7 +233,7 @@ func Semctl(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscal
 		return uintptr(semid), nil, err
 
 	default:
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 }
 
@@ -247,17 +247,17 @@ func ipcSet(t *kernel.Task, id int32, uid auth.UID, gid auth.GID, perms fs.FileP
 	r := t.IPCNamespace().SemaphoreRegistry()
 	set := r.FindByID(id)
 	if set == nil {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 
 	creds := auth.CredentialsFromContext(t)
 	kuid := creds.UserNamespace.MapToKUID(uid)
 	if !kuid.Ok() {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	kgid := creds.UserNamespace.MapToKGID(gid)
 	if !kgid.Ok() {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	owner := fs.FileOwner{UID: kuid, GID: kgid}
 	return set.Change(t, creds, owner, perms)
@@ -267,7 +267,7 @@ func ipcStat(t *kernel.Task, id int32) (*linux.SemidDS, error) {
 	r := t.IPCNamespace().SemaphoreRegistry()
 	set := r.FindByID(id)
 	if set == nil {
-		return nil, syserror.EINVAL
+		return nil, linuxerr.EINVAL
 	}
 	creds := auth.CredentialsFromContext(t)
 	return set.GetStat(creds)
@@ -277,7 +277,7 @@ func semStat(t *kernel.Task, index int32) (int32, *linux.SemidDS, error) {
 	r := t.IPCNamespace().SemaphoreRegistry()
 	set := r.FindByIndex(index)
 	if set == nil {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 	creds := auth.CredentialsFromContext(t)
 	ds, err := set.GetStat(creds)
@@ -290,7 +290,7 @@ func semStat(t *kernel.Task, index int32) (int32, *linux.SemidDS, error) {
 func semStatAny(t *kernel.Task, index int32) (int32, *linux.SemidDS, error) {
 	set := t.IPCNamespace().SemaphoreRegistry().FindByIndex(index)
 	if set == nil {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 	creds := auth.CredentialsFromContext(t)
 	ds, err := set.GetStatAny(creds)
@@ -304,7 +304,7 @@ func setVal(t *kernel.Task, id int32, num int32, val int16) error {
 	r := t.IPCNamespace().SemaphoreRegistry()
 	set := r.FindByID(id)
 	if set == nil {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	creds := auth.CredentialsFromContext(t)
 	pid := t.Kernel().GlobalInit().PIDNamespace().IDOfThreadGroup(t.ThreadGroup())
@@ -315,7 +315,7 @@ func setValAll(t *kernel.Task, id int32, array hostarch.Addr) error {
 	r := t.IPCNamespace().SemaphoreRegistry()
 	set := r.FindByID(id)
 	if set == nil {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	vals := make([]uint16, set.Size())
 	if _, err := primitive.CopyUint16SliceIn(t, array, vals); err != nil {
@@ -330,7 +330,7 @@ func getVal(t *kernel.Task, id int32, num int32) (int16, error) {
 	r := t.IPCNamespace().SemaphoreRegistry()
 	set := r.FindByID(id)
 	if set == nil {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 	creds := auth.CredentialsFromContext(t)
 	return set.GetVal(num, creds)
@@ -340,7 +340,7 @@ func getValAll(t *kernel.Task, id int32, array hostarch.Addr) error {
 	r := t.IPCNamespace().SemaphoreRegistry()
 	set := r.FindByID(id)
 	if set == nil {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	creds := auth.CredentialsFromContext(t)
 	vals, err := set.GetValAll(creds)
@@ -355,7 +355,7 @@ func getPID(t *kernel.Task, id int32, num int32) (int32, error) {
 	r := t.IPCNamespace().SemaphoreRegistry()
 	set := r.FindByID(id)
 	if set == nil {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 	creds := auth.CredentialsFromContext(t)
 	gpid, err := set.GetPID(num, creds)
@@ -374,7 +374,7 @@ func getZCnt(t *kernel.Task, id int32, num int32) (uint16, error) {
 	r := t.IPCNamespace().SemaphoreRegistry()
 	set := r.FindByID(id)
 	if set == nil {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 	creds := auth.CredentialsFromContext(t)
 	return set.CountZeroWaiters(num, creds)
@@ -384,7 +384,7 @@ func getNCnt(t *kernel.Task, id int32, num int32) (uint16, error) {
 	r := t.IPCNamespace().SemaphoreRegistry()
 	set := r.FindByID(id)
 	if set == nil {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 	creds := auth.CredentialsFromContext(t)
 	return set.CountNegativeWaiters(num, creds)
diff --git a/pkg/sentry/syscalls/linux/sys_shm.go b/pkg/sentry/syscalls/linux/sys_shm.go
index 584064143..3e3a952ce 100644
--- a/pkg/sentry/syscalls/linux/sys_shm.go
+++ b/pkg/sentry/syscalls/linux/sys_shm.go
@@ -16,10 +16,10 @@ package linux
 
 import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
 	"gvisor.dev/gvisor/pkg/sentry/kernel"
 	"gvisor.dev/gvisor/pkg/sentry/kernel/shm"
-	"gvisor.dev/gvisor/pkg/syserror"
 )
 
 // Shmget implements shmget(2).
@@ -51,7 +51,7 @@ func findSegment(t *kernel.Task, id shm.ID) (*shm.Shm, error) {
 	segment := r.FindByID(id)
 	if segment == nil {
 		// No segment with provided id.
-		return nil, syserror.EINVAL
+		return nil, linuxerr.EINVAL
 	}
 	return segment, nil
 }
@@ -64,7 +64,7 @@ func Shmat(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscall
 
 	segment, err := findSegment(t, id)
 	if err != nil {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 	defer segment.DecRef(t)
 
@@ -106,7 +106,7 @@ func Shmctl(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscal
 	case linux.IPC_STAT:
 		segment, err := findSegment(t, id)
 		if err != nil {
-			return 0, nil, syserror.EINVAL
+			return 0, nil, linuxerr.EINVAL
 		}
 		defer segment.DecRef(t)
 
@@ -130,7 +130,7 @@ func Shmctl(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscal
 	// Remaining commands refer to a specific segment.
 	segment, err := findSegment(t, id)
 	if err != nil {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 	defer segment.DecRef(t)
 
@@ -155,6 +155,6 @@ func Shmctl(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscal
 		return 0, nil, nil
 
 	default:
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 }
diff --git a/pkg/sentry/syscalls/linux/sys_signal.go b/pkg/sentry/syscalls/linux/sys_signal.go
index db763c68e..4d659e5cf 100644
--- a/pkg/sentry/syscalls/linux/sys_signal.go
+++ b/pkg/sentry/syscalls/linux/sys_signal.go
@@ -203,7 +203,7 @@ func Tkill(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscall
 	// N.B. Inconsistent with man page, linux actually rejects calls with
 	// tid <=0 by EINVAL. This isn't the same for all signal calls.
 	if tid <= 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	target := t.PIDNamespace().TaskWithID(tid)
@@ -226,7 +226,7 @@ func Tgkill(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscal
 	// N.B. Inconsistent with man page, linux actually rejects calls with
 	// tgid/tid <=0 by EINVAL. This isn't the same for all signal calls.
 	if tgid <= 0 || tid <= 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	targetTG := t.PIDNamespace().ThreadGroupWithID(tgid)
@@ -249,7 +249,7 @@ func RtSigaction(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.S
 	sigsetsize := args[3].SizeT()
 
 	if sigsetsize != linux.SignalSetSize {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	var newactptr *linux.SigAction
@@ -292,7 +292,7 @@ func RtSigprocmask(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel
 	sigsetsize := args[3].SizeT()
 
 	if sigsetsize != linux.SignalSetSize {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 	oldmask := t.SignalMask()
 	if setaddr != 0 {
@@ -309,7 +309,7 @@ func RtSigprocmask(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel
 		case linux.SIG_SETMASK:
 			t.SetSignalMask(mask)
 		default:
-			return 0, nil, syserror.EINVAL
+			return 0, nil, linuxerr.EINVAL
 		}
 	}
 	if oldaddr != 0 {
@@ -378,7 +378,7 @@ func RtSigtimedwait(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kerne
 			return 0, nil, err
 		}
 		if !d.Valid() {
-			return 0, nil, syserror.EINVAL
+			return 0, nil, linuxerr.EINVAL
 		}
 		timeout = time.Duration(d.ToNsecCapped())
 	} else {
@@ -450,7 +450,7 @@ func RtTgsigqueueinfo(t *kernel.Task, args arch.SyscallArguments) (uintptr, *ker
 	// N.B. Inconsistent with man page, linux actually rejects calls with
 	// tgid/tid <=0 by EINVAL. This isn't the same for all signal calls.
 	if tgid <= 0 || tid <= 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Copy in the info. See RtSigqueueinfo above.
@@ -525,7 +525,7 @@ func sharedSignalfd(t *kernel.Task, fd int32, sigset hostarch.Addr, sigsetsize u
 
 	// Always check for valid flags, even if not creating.
 	if flags&^(linux.SFD_NONBLOCK|linux.SFD_CLOEXEC) != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Is this a change to an existing signalfd?
@@ -545,7 +545,7 @@ func sharedSignalfd(t *kernel.Task, fd int32, sigset hostarch.Addr, sigsetsize u
 		}
 
 		// Not a signalfd.
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Create a new file.
diff --git a/pkg/sentry/syscalls/linux/sys_socket.go b/pkg/sentry/syscalls/linux/sys_socket.go
index 3bd21a911..6638ad60f 100644
--- a/pkg/sentry/syscalls/linux/sys_socket.go
+++ b/pkg/sentry/syscalls/linux/sys_socket.go
@@ -118,7 +118,7 @@ type multipleMessageHeader64 struct {
 // from the untrusted address space range.
 func CaptureAddress(t *kernel.Task, addr hostarch.Addr, addrlen uint32) ([]byte, error) {
 	if addrlen > maxAddrLen {
-		return nil, syserror.EINVAL
+		return nil, linuxerr.EINVAL
 	}
 
 	addrBuf := make([]byte, addrlen)
@@ -140,7 +140,7 @@ func writeAddress(t *kernel.Task, addr linux.SockAddr, addrLen uint32, addrPtr h
 	}
 
 	if int32(bufLen) < 0 {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 
 	// Write the length unconditionally.
@@ -174,7 +174,7 @@ func Socket(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscal
 
 	// Check and initialize the flags.
 	if stype & ^(0xf|linux.SOCK_NONBLOCK|linux.SOCK_CLOEXEC) != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Create the new socket.
@@ -206,7 +206,7 @@ func SocketPair(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sy
 
 	// Check and initialize the flags.
 	if stype & ^(0xf|linux.SOCK_NONBLOCK|linux.SOCK_CLOEXEC) != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	fileFlags := fs.SettableFileFlags{
@@ -278,7 +278,7 @@ func Connect(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysca
 func accept(t *kernel.Task, fd int32, addr hostarch.Addr, addrLen hostarch.Addr, flags int) (uintptr, error) {
 	// Check that no unsupported flags are passed in.
 	if flags & ^(linux.SOCK_NONBLOCK|linux.SOCK_CLOEXEC) != 0 {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 
 	// Get socket from the file descriptor.
@@ -422,7 +422,7 @@ func Shutdown(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysc
 	switch how {
 	case linux.SHUT_RD, linux.SHUT_WR, linux.SHUT_RDWR:
 	default:
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	return 0, nil, s.Shutdown(t, int(how)).ToError()
@@ -455,7 +455,7 @@ func GetSockOpt(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sy
 		return 0, nil, err
 	}
 	if optLen < 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Call syscall implementation then copy both value and value len out.
@@ -531,10 +531,10 @@ func SetSockOpt(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sy
 	}
 
 	if optLen < 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 	if optLen > maxOptLen {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 	buf := t.CopyScratchBuffer(int(optLen))
 	if _, err := t.CopyInBytes(optValAddr, buf); err != nil {
@@ -613,7 +613,7 @@ func RecvMsg(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysca
 
 	if t.Arch().Width() != 8 {
 		// We only handle 64-bit for now.
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Get socket from the file descriptor.
@@ -631,7 +631,7 @@ func RecvMsg(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysca
 
 	// Reject flags that we don't handle yet.
 	if flags & ^(baseRecvFlags|linux.MSG_PEEK|linux.MSG_CMSG_CLOEXEC|linux.MSG_ERRQUEUE) != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	if file.Flags().NonBlocking {
@@ -661,7 +661,7 @@ func RecvMMsg(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysc
 
 	if t.Arch().Width() != 8 {
 		// We only handle 64-bit for now.
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	if vlen > linux.UIO_MAXIOV {
@@ -670,7 +670,7 @@ func RecvMMsg(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysc
 
 	// Reject flags that we don't handle yet.
 	if flags & ^(baseRecvFlags|linux.MSG_CMSG_CLOEXEC|linux.MSG_ERRQUEUE) != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Get socket from the file descriptor.
@@ -698,7 +698,7 @@ func RecvMMsg(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysc
 			return 0, nil, err
 		}
 		if !ts.Valid() {
-			return 0, nil, syserror.EINVAL
+			return 0, nil, linuxerr.EINVAL
 		}
 		deadline = t.Kernel().MonotonicClock().Now().Add(ts.ToDuration())
 		haveDeadline = true
@@ -830,12 +830,12 @@ func recvSingleMsg(t *kernel.Task, s socket.Socket, msgPtr hostarch.Addr, flags
 // recvfrom and recv syscall handlers.
 func recvFrom(t *kernel.Task, fd int32, bufPtr hostarch.Addr, bufLen uint64, flags int32, namePtr hostarch.Addr, nameLenPtr hostarch.Addr) (uintptr, error) {
 	if int(bufLen) < 0 {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 
 	// Reject flags that we don't handle yet.
 	if flags & ^(baseRecvFlags|linux.MSG_PEEK|linux.MSG_CONFIRM) != 0 {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 
 	// Get socket from the file descriptor.
@@ -908,7 +908,7 @@ func SendMsg(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysca
 
 	if t.Arch().Width() != 8 {
 		// We only handle 64-bit for now.
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Get socket from the file descriptor.
@@ -926,7 +926,7 @@ func SendMsg(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysca
 
 	// Reject flags that we don't handle yet.
 	if flags & ^(linux.MSG_DONTWAIT|linux.MSG_EOR|linux.MSG_MORE|linux.MSG_NOSIGNAL) != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	if file.Flags().NonBlocking {
@@ -946,7 +946,7 @@ func SendMMsg(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysc
 
 	if t.Arch().Width() != 8 {
 		// We only handle 64-bit for now.
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	if vlen > linux.UIO_MAXIOV {
@@ -968,7 +968,7 @@ func SendMMsg(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysc
 
 	// Reject flags that we don't handle yet.
 	if flags & ^(linux.MSG_DONTWAIT|linux.MSG_EOR|linux.MSG_MORE|linux.MSG_NOSIGNAL) != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	if file.Flags().NonBlocking {
@@ -1074,7 +1074,7 @@ func sendSingleMsg(t *kernel.Task, s socket.Socket, file *fs.File, msgPtr hostar
 func sendTo(t *kernel.Task, fd int32, bufPtr hostarch.Addr, bufLen uint64, flags int32, namePtr hostarch.Addr, nameLen uint32) (uintptr, error) {
 	bl := int(bufLen)
 	if bl < 0 {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 
 	// Get socket from the file descriptor.
diff --git a/pkg/sentry/syscalls/linux/sys_splice.go b/pkg/sentry/syscalls/linux/sys_splice.go
index 134051124..88bee61ef 100644
--- a/pkg/sentry/syscalls/linux/sys_splice.go
+++ b/pkg/sentry/syscalls/linux/sys_splice.go
@@ -16,6 +16,7 @@ package linux
 
 import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/marshal/primitive"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
 	"gvisor.dev/gvisor/pkg/sentry/fs"
@@ -27,7 +28,7 @@ import (
 // doSplice implements a blocking splice operation.
 func doSplice(t *kernel.Task, outFile, inFile *fs.File, opts fs.SpliceOpts, nonBlocking bool) (int64, error) {
 	if opts.Length < 0 || opts.SrcStart < 0 || opts.DstStart < 0 || (opts.SrcStart+opts.Length < 0) {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 	if opts.Length == 0 {
 		return 0, nil
@@ -125,13 +126,13 @@ func Sendfile(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysc
 
 	// Verify that the outfile Append flag is not set.
 	if outFile.Flags().Append {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Verify that we have a regular infile. This is a requirement; the
 	// same check appears in Linux (fs/splice.c:splice_direct_to_actor).
 	if !fs.IsRegular(inFile.Dirent.Inode.StableAttr) {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	var (
@@ -190,7 +191,7 @@ func Splice(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscal
 
 	// Check for invalid flags.
 	if flags&^(linux.SPLICE_F_MOVE|linux.SPLICE_F_NONBLOCK|linux.SPLICE_F_MORE|linux.SPLICE_F_GIFT) != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Get files.
@@ -230,7 +231,7 @@ func Splice(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscal
 		}
 		if outOffset != 0 {
 			if !outFile.Flags().Pwrite {
-				return 0, nil, syserror.EINVAL
+				return 0, nil, linuxerr.EINVAL
 			}
 
 			var offset int64
@@ -248,7 +249,7 @@ func Splice(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscal
 		}
 		if inOffset != 0 {
 			if !inFile.Flags().Pread {
-				return 0, nil, syserror.EINVAL
+				return 0, nil, linuxerr.EINVAL
 			}
 
 			var offset int64
@@ -267,10 +268,10 @@ func Splice(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscal
 
 		// We may not refer to the same pipe; otherwise it's a continuous loop.
 		if inFileAttr.InodeID == outFileAttr.InodeID {
-			return 0, nil, syserror.EINVAL
+			return 0, nil, linuxerr.EINVAL
 		}
 	default:
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Splice data.
@@ -298,7 +299,7 @@ func Tee(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallCo
 
 	// Check for invalid flags.
 	if flags&^(linux.SPLICE_F_MOVE|linux.SPLICE_F_NONBLOCK|linux.SPLICE_F_MORE|linux.SPLICE_F_GIFT) != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Get files.
@@ -316,12 +317,12 @@ func Tee(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallCo
 
 	// All files must be pipes.
 	if !fs.IsPipe(inFile.Dirent.Inode.StableAttr) || !fs.IsPipe(outFile.Dirent.Inode.StableAttr) {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// We may not refer to the same pipe; see above.
 	if inFile.Dirent.Inode.StableAttr.InodeID == outFile.Dirent.Inode.StableAttr.InodeID {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// The operation is non-blocking if anything is non-blocking.
diff --git a/pkg/sentry/syscalls/linux/sys_stat.go b/pkg/sentry/syscalls/linux/sys_stat.go
index 2338ba44b..103b13c10 100644
--- a/pkg/sentry/syscalls/linux/sys_stat.go
+++ b/pkg/sentry/syscalls/linux/sys_stat.go
@@ -16,6 +16,7 @@ package linux
 
 import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
 	"gvisor.dev/gvisor/pkg/sentry/fs"
@@ -139,13 +140,13 @@ func Statx(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscall
 	statxAddr := args[4].Pointer()
 
 	if mask&linux.STATX__RESERVED != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 	if flags&^(linux.AT_SYMLINK_NOFOLLOW|linux.AT_EMPTY_PATH|linux.AT_STATX_SYNC_TYPE) != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 	if flags&linux.AT_STATX_SYNC_TYPE == linux.AT_STATX_SYNC_TYPE {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	path, dirPath, err := copyInPath(t, pathAddr, flags&linux.AT_EMPTY_PATH != 0)
diff --git a/pkg/sentry/syscalls/linux/sys_sync.go b/pkg/sentry/syscalls/linux/sys_sync.go
index 5ebd4461f..3f0e6c02e 100644
--- a/pkg/sentry/syscalls/linux/sys_sync.go
+++ b/pkg/sentry/syscalls/linux/sys_sync.go
@@ -16,6 +16,7 @@ package linux
 
 import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
 	"gvisor.dev/gvisor/pkg/sentry/fs"
 	"gvisor.dev/gvisor/pkg/sentry/kernel"
@@ -86,13 +87,13 @@ func SyncFileRange(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel
 	uflags := args[3].Uint()
 
 	if offset < 0 || offset+nbytes < offset {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	if uflags&^(linux.SYNC_FILE_RANGE_WAIT_BEFORE|
 		linux.SYNC_FILE_RANGE_WRITE|
 		linux.SYNC_FILE_RANGE_WAIT_AFTER) != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	if nbytes == 0 {
diff --git a/pkg/sentry/syscalls/linux/sys_syslog.go b/pkg/sentry/syscalls/linux/sys_syslog.go
index 40c8bb061..ba372f9e3 100644
--- a/pkg/sentry/syscalls/linux/sys_syslog.go
+++ b/pkg/sentry/syscalls/linux/sys_syslog.go
@@ -15,6 +15,7 @@
 package linux
 
 import (
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
 	"gvisor.dev/gvisor/pkg/sentry/kernel"
 	"gvisor.dev/gvisor/pkg/syserror"
@@ -40,7 +41,7 @@ func Syslog(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscal
 	switch command {
 	case _SYSLOG_ACTION_READ_ALL:
 		if size < 0 {
-			return 0, nil, syserror.EINVAL
+			return 0, nil, linuxerr.EINVAL
 		}
 		if size > logBufLen {
 			size = logBufLen
diff --git a/pkg/sentry/syscalls/linux/sys_thread.go b/pkg/sentry/syscalls/linux/sys_thread.go
index 0d5056303..d99dd5131 100644
--- a/pkg/sentry/syscalls/linux/sys_thread.go
+++ b/pkg/sentry/syscalls/linux/sys_thread.go
@@ -19,6 +19,7 @@ import (
 
 	"golang.org/x/sys/unix"
 	"gvisor.dev/gvisor/pkg/abi/linux"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/marshal/primitive"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
@@ -112,7 +113,7 @@ func execveat(t *kernel.Task, dirFD int32, pathnameAddr, argvAddr, envvAddr host
 	}
 
 	if flags&^(linux.AT_EMPTY_PATH|linux.AT_SYMLINK_NOFOLLOW) != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 	atEmptyPath := flags&linux.AT_EMPTY_PATH != 0
 	if !atEmptyPath && len(pathname) == 0 {
@@ -260,7 +261,7 @@ func parseCommonWaitOptions(wopts *kernel.WaitOptions, options int) error {
 		wopts.NonCloneTasks = true
 		wopts.CloneTasks = true
 	default:
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	if options&linux.WCONTINUED != 0 {
 		wopts.Events |= kernel.EventGroupContinue
@@ -277,7 +278,7 @@ func parseCommonWaitOptions(wopts *kernel.WaitOptions, options int) error {
 // wait4 waits for the given child process to exit.
 func wait4(t *kernel.Task, pid int, statusAddr hostarch.Addr, options int, rusageAddr hostarch.Addr) (uintptr, error) {
 	if options&^(linux.WNOHANG|linux.WUNTRACED|linux.WCONTINUED|linux.WNOTHREAD|linux.WALL|linux.WCLONE) != 0 {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 	wopts := kernel.WaitOptions{
 		Events:       kernel.EventExit | kernel.EventTraceeStop,
@@ -358,10 +359,10 @@ func Waitid(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscal
 	rusageAddr := args[4].Pointer()
 
 	if options&^(linux.WNOHANG|linux.WEXITED|linux.WSTOPPED|linux.WCONTINUED|linux.WNOWAIT|linux.WNOTHREAD|linux.WALL|linux.WCLONE) != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 	if options&(linux.WEXITED|linux.WSTOPPED|linux.WCONTINUED) == 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 	wopts := kernel.WaitOptions{
 		Events:       kernel.EventTraceeStop,
@@ -374,7 +375,7 @@ func Waitid(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscal
 	case linux.P_PGID:
 		wopts.SpecificPGID = kernel.ProcessGroupID(id)
 	default:
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	if err := parseCommonWaitOptions(&wopts, options); err != nil {
@@ -528,7 +529,7 @@ func SchedGetaffinity(t *kernel.Task, args arch.SyscallArguments) (uintptr, *ker
 	// in an array of "unsigned long" so the buffer needs to
 	// be a multiple of the word size.
 	if size&(t.Arch().Width()-1) > 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	var task *kernel.Task
@@ -545,7 +546,7 @@ func SchedGetaffinity(t *kernel.Task, args arch.SyscallArguments) (uintptr, *ker
 	// The buffer needs to be big enough to hold a cpumask with
 	// all possible cpus.
 	if size < mask.Size() {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 	_, err := t.CopyOutBytes(maskAddr, mask)
 
@@ -594,7 +595,7 @@ func Setpgid(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysca
 		}
 		tg = ot.ThreadGroup()
 		if tg.Leader() != ot {
-			return 0, nil, syserror.EINVAL
+			return 0, nil, linuxerr.EINVAL
 		}
 
 		// Setpgid only operates on child threadgroups.
@@ -609,7 +610,7 @@ func Setpgid(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysca
 	if pgid == 0 {
 		pgid = defaultPGID
 	} else if pgid < 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// If the pgid is the same as the group, then create a new one. Otherwise,
@@ -712,7 +713,7 @@ func Getpriority(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.S
 		// PRIO_USER and PRIO_PGRP have no further implementation yet.
 		return 0, nil, nil
 	default:
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 }
 
@@ -754,7 +755,7 @@ func Setpriority(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.S
 		// PRIO_USER and PRIO_PGRP have no further implementation yet.
 		return 0, nil, nil
 	default:
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	return 0, nil, nil
diff --git a/pkg/sentry/syscalls/linux/sys_time.go b/pkg/sentry/syscalls/linux/sys_time.go
index 2ec74b33a..d75bb9c4f 100644
--- a/pkg/sentry/syscalls/linux/sys_time.go
+++ b/pkg/sentry/syscalls/linux/sys_time.go
@@ -76,7 +76,7 @@ func ClockGetres(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.S
 	}
 
 	if _, err := getClock(t, clockID); err != nil {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	if addr == 0 {
@@ -95,12 +95,12 @@ type cpuClocker interface {
 func getClock(t *kernel.Task, clockID int32) (ktime.Clock, error) {
 	if clockID < 0 {
 		if !isValidCPUClock(clockID) {
-			return nil, syserror.EINVAL
+			return nil, linuxerr.EINVAL
 		}
 
 		targetTask := targetTask(t, clockID)
 		if targetTask == nil {
-			return nil, syserror.EINVAL
+			return nil, linuxerr.EINVAL
 		}
 
 		var target cpuClocker
@@ -117,7 +117,7 @@ func getClock(t *kernel.Task, clockID int32) (ktime.Clock, error) {
 			// CPUCLOCK_SCHED is approximated by CPUCLOCK_PROF.
 			return target.CPUClock(), nil
 		default:
-			return nil, syserror.EINVAL
+			return nil, linuxerr.EINVAL
 		}
 	}
 
@@ -139,7 +139,7 @@ func getClock(t *kernel.Task, clockID int32) (ktime.Clock, error) {
 	case linux.CLOCK_THREAD_CPUTIME_ID:
 		return t.CPUClock(), nil
 	default:
-		return nil, syserror.EINVAL
+		return nil, linuxerr.EINVAL
 	}
 }
 
@@ -254,7 +254,7 @@ func Nanosleep(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sys
 	}
 
 	if !ts.Valid() {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Just like linux, we cap the timeout with the max number that int64 can
@@ -277,7 +277,7 @@ func ClockNanosleep(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kerne
 	}
 
 	if !req.Valid() {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Only allow clock constants also allowed by Linux.
@@ -285,7 +285,7 @@ func ClockNanosleep(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kerne
 		if clockID != linux.CLOCK_REALTIME &&
 			clockID != linux.CLOCK_MONOTONIC &&
 			clockID != linux.CLOCK_PROCESS_CPUTIME_ID {
-			return 0, nil, syserror.EINVAL
+			return 0, nil, linuxerr.EINVAL
 		}
 	}
 
diff --git a/pkg/sentry/syscalls/linux/sys_timerfd.go b/pkg/sentry/syscalls/linux/sys_timerfd.go
index cadd9d348..a8e88b814 100644
--- a/pkg/sentry/syscalls/linux/sys_timerfd.go
+++ b/pkg/sentry/syscalls/linux/sys_timerfd.go
@@ -16,6 +16,7 @@ package linux
 
 import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
 	"gvisor.dev/gvisor/pkg/sentry/fs"
 	"gvisor.dev/gvisor/pkg/sentry/fs/timerfd"
@@ -30,7 +31,7 @@ func TimerfdCreate(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel
 	flags := args[1].Int()
 
 	if flags&^(linux.TFD_CLOEXEC|linux.TFD_NONBLOCK) != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	var c ktime.Clock
@@ -40,7 +41,7 @@ func TimerfdCreate(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel
 	case linux.CLOCK_MONOTONIC, linux.CLOCK_BOOTTIME:
 		c = t.Kernel().MonotonicClock()
 	default:
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 	f := timerfd.NewFile(t, c)
 	defer f.DecRef(t)
@@ -66,7 +67,7 @@ func TimerfdSettime(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kerne
 	oldValAddr := args[3].Pointer()
 
 	if flags&^(linux.TFD_TIMER_ABSTIME) != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	f := t.GetFile(fd)
@@ -77,7 +78,7 @@ func TimerfdSettime(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kerne
 
 	tf, ok := f.FileOperations.(*timerfd.TimerOperations)
 	if !ok {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	var newVal linux.Itimerspec
@@ -111,7 +112,7 @@ func TimerfdGettime(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kerne
 
 	tf, ok := f.FileOperations.(*timerfd.TimerOperations)
 	if !ok {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	tm, s := tf.GetTime()
diff --git a/pkg/sentry/syscalls/linux/sys_tls_amd64.go b/pkg/sentry/syscalls/linux/sys_tls_amd64.go
index 6ddd30d5c..32272c267 100644
--- a/pkg/sentry/syscalls/linux/sys_tls_amd64.go
+++ b/pkg/sentry/syscalls/linux/sys_tls_amd64.go
@@ -18,6 +18,7 @@ package linux
 
 import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/marshal/primitive"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
 	"gvisor.dev/gvisor/pkg/sentry/kernel"
@@ -48,7 +49,7 @@ func ArchPrctl(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sys
 		t.Kernel().EmitUnimplementedEvent(t)
 		fallthrough
 	default:
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	return 0, nil, nil
diff --git a/pkg/sentry/syscalls/linux/sys_utsname.go b/pkg/sentry/syscalls/linux/sys_utsname.go
index 66c5974f5..7fffb189e 100644
--- a/pkg/sentry/syscalls/linux/sys_utsname.go
+++ b/pkg/sentry/syscalls/linux/sys_utsname.go
@@ -16,6 +16,7 @@ package linux
 
 import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
 	"gvisor.dev/gvisor/pkg/sentry/kernel"
 	"gvisor.dev/gvisor/pkg/syserror"
@@ -60,7 +61,7 @@ func Setdomainname(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel
 		return 0, nil, syserror.EPERM
 	}
 	if size < 0 || size > linux.UTSLen {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	name, err := t.CopyInString(nameAddr, int(size))
@@ -82,7 +83,7 @@ func Sethostname(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.S
 		return 0, nil, syserror.EPERM
 	}
 	if size < 0 || size > linux.UTSLen {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	name := make([]byte, size)
diff --git a/pkg/sentry/syscalls/linux/sys_write.go b/pkg/sentry/syscalls/linux/sys_write.go
index cff355550..998b5fde6 100644
--- a/pkg/sentry/syscalls/linux/sys_write.go
+++ b/pkg/sentry/syscalls/linux/sys_write.go
@@ -59,7 +59,7 @@ func Write(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscall
 	// Check that the size is legitimate.
 	si := int(size)
 	if si < 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Get the source of the write.
@@ -90,7 +90,7 @@ func Pwrite64(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysc
 
 	// Check that the offset is legitimate and does not overflow.
 	if offset < 0 || offset+int64(size) < 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Is writing at an offset supported?
@@ -106,7 +106,7 @@ func Pwrite64(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysc
 	// Check that the size is legitimate.
 	si := int(size)
 	if si < 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Get the source of the write.
@@ -167,7 +167,7 @@ func Pwritev(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysca
 
 	// Check that the offset is legitimate.
 	if offset < 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Is writing at an offset supported?
@@ -220,7 +220,7 @@ func Pwritev2(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysc
 
 	// Check that the offset is legitimate.
 	if offset < -1 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Is writing at an offset supported?
diff --git a/pkg/sentry/syscalls/linux/sys_xattr.go b/pkg/sentry/syscalls/linux/sys_xattr.go
index 37fb67f80..da6651062 100644
--- a/pkg/sentry/syscalls/linux/sys_xattr.go
+++ b/pkg/sentry/syscalls/linux/sys_xattr.go
@@ -183,7 +183,7 @@ func setXattrFromPath(t *kernel.Task, args arch.SyscallArguments, resolveSymlink
 // setXattr implements setxattr(2) from the given *fs.Dirent.
 func setXattr(t *kernel.Task, d *fs.Dirent, nameAddr, valueAddr hostarch.Addr, size uint64, flags uint32) error {
 	if flags&^(linux.XATTR_CREATE|linux.XATTR_REPLACE) != 0 {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 
 	name, err := copyInXattrName(t, nameAddr)
@@ -196,7 +196,7 @@ func setXattr(t *kernel.Task, d *fs.Dirent, nameAddr, valueAddr hostarch.Addr, s
 	}
 
 	if size > linux.XATTR_SIZE_MAX {
-		return syserror.E2BIG
+		return linuxerr.E2BIG
 	}
 	buf := make([]byte, size)
 	if _, err := t.CopyInBytes(valueAddr, buf); err != nil {
@@ -334,7 +334,7 @@ func listXattr(t *kernel.Task, d *fs.Dirent, addr hostarch.Addr, size uint64) (i
 
 	listSize := xattrListSize(xattrs)
 	if listSize > linux.XATTR_SIZE_MAX {
-		return 0, syserror.E2BIG
+		return 0, linuxerr.E2BIG
 	}
 	if uint64(listSize) > requestedSize {
 		return 0, syserror.ERANGE
diff --git a/pkg/sentry/syscalls/linux/timespec.go b/pkg/sentry/syscalls/linux/timespec.go
index 3edc922eb..b327e27d6 100644
--- a/pkg/sentry/syscalls/linux/timespec.go
+++ b/pkg/sentry/syscalls/linux/timespec.go
@@ -18,6 +18,7 @@ import (
 	"time"
 
 	"gvisor.dev/gvisor/pkg/abi/linux"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/sentry/kernel"
 	"gvisor.dev/gvisor/pkg/syserror"
@@ -103,7 +104,7 @@ func copyTimespecInToDuration(t *kernel.Task, timespecAddr hostarch.Addr) (time.
 			return 0, err
 		}
 		if !timespec.Valid() {
-			return 0, syserror.EINVAL
+			return 0, linuxerr.EINVAL
 		}
 		timeout = time.Duration(timespec.ToNsecCapped())
 	}
diff --git a/pkg/sentry/syscalls/linux/vfs2/aio.go b/pkg/sentry/syscalls/linux/vfs2/aio.go
index fd1863ef3..d81df637f 100644
--- a/pkg/sentry/syscalls/linux/vfs2/aio.go
+++ b/pkg/sentry/syscalls/linux/vfs2/aio.go
@@ -17,6 +17,8 @@ package vfs2
 import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
+	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/marshal/primitive"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
 	"gvisor.dev/gvisor/pkg/sentry/fsimpl/eventfd"
@@ -26,8 +28,6 @@ import (
 	"gvisor.dev/gvisor/pkg/sentry/vfs"
 	"gvisor.dev/gvisor/pkg/syserror"
 	"gvisor.dev/gvisor/pkg/usermem"
-
-	"gvisor.dev/gvisor/pkg/hostarch"
 )
 
 // IoSubmit implements linux syscall io_submit(2).
@@ -37,7 +37,7 @@ func IoSubmit(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysc
 	addr := args[2].Pointer()
 
 	if nrEvents < 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	for i := int32(0); i < nrEvents; i++ {
@@ -90,7 +90,7 @@ func IoSubmit(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysc
 // submitCallback processes a single callback.
 func submitCallback(t *kernel.Task, id uint64, cb *linux.IOCallback, cbAddr hostarch.Addr) error {
 	if cb.Reserved2 != 0 {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 
 	fd := t.GetFileVFS2(cb.FD)
@@ -110,7 +110,7 @@ func submitCallback(t *kernel.Task, id uint64, cb *linux.IOCallback, cbAddr host
 
 		// Check that it is an eventfd.
 		if _, ok := eventFD.Impl().(*eventfd.EventFileDescription); !ok {
-			return syserror.EINVAL
+			return linuxerr.EINVAL
 		}
 	}
 
@@ -123,14 +123,14 @@ func submitCallback(t *kernel.Task, id uint64, cb *linux.IOCallback, cbAddr host
 	switch cb.OpCode {
 	case linux.IOCB_CMD_PREAD, linux.IOCB_CMD_PREADV, linux.IOCB_CMD_PWRITE, linux.IOCB_CMD_PWRITEV:
 		if cb.Offset < 0 {
-			return syserror.EINVAL
+			return linuxerr.EINVAL
 		}
 	}
 
 	// Prepare the request.
 	aioCtx, ok := t.MemoryManager().LookupAIOContext(t, id)
 	if !ok {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	if err := aioCtx.Prepare(); err != nil {
 		return err
@@ -200,7 +200,7 @@ func memoryFor(t *kernel.Task, cb *linux.IOCallback) (usermem.IOSequence, error)
 	bytes := int(cb.Bytes)
 	if bytes < 0 {
 		// Linux also requires that this field fit in ssize_t.
-		return usermem.IOSequence{}, syserror.EINVAL
+		return usermem.IOSequence{}, linuxerr.EINVAL
 	}
 
 	// Since this I/O will be asynchronous with respect to t's task goroutine,
@@ -222,6 +222,6 @@ func memoryFor(t *kernel.Task, cb *linux.IOCallback) (usermem.IOSequence, error)
 
 	default:
 		// Not a supported command.
-		return usermem.IOSequence{}, syserror.EINVAL
+		return usermem.IOSequence{}, linuxerr.EINVAL
 	}
 }
diff --git a/pkg/sentry/syscalls/linux/vfs2/epoll.go b/pkg/sentry/syscalls/linux/vfs2/epoll.go
index 7aff01343..d3bb3a3e1 100644
--- a/pkg/sentry/syscalls/linux/vfs2/epoll.go
+++ b/pkg/sentry/syscalls/linux/vfs2/epoll.go
@@ -35,7 +35,7 @@ var sizeofEpollEvent = (*linux.EpollEvent)(nil).SizeBytes()
 func EpollCreate1(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallControl, error) {
 	flags := args[0].Int()
 	if flags&^linux.EPOLL_CLOEXEC != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	file, err := t.Kernel().VFS().NewEpollInstanceFD(t)
@@ -60,7 +60,7 @@ func EpollCreate(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.S
 	// "Since Linux 2.6.8, the size argument is ignored, but must be greater
 	// than zero" - epoll_create(2)
 	if size <= 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	file, err := t.Kernel().VFS().NewEpollInstanceFD(t)
@@ -90,7 +90,7 @@ func EpollCtl(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysc
 	defer epfile.DecRef(t)
 	ep, ok := epfile.Impl().(*vfs.EpollInstance)
 	if !ok {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 	file := t.GetFileVFS2(fd)
 	if file == nil {
@@ -98,7 +98,7 @@ func EpollCtl(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysc
 	}
 	defer file.DecRef(t)
 	if epfile == file {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	var event linux.EpollEvent
@@ -116,14 +116,14 @@ func EpollCtl(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysc
 		}
 		return 0, nil, ep.ModifyInterest(file, fd, event)
 	default:
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 }
 
 func waitEpoll(t *kernel.Task, epfd int32, eventsAddr hostarch.Addr, maxEvents int, timeoutInNanos int64) (uintptr, *kernel.SyscallControl, error) {
 	var _EP_MAX_EVENTS = math.MaxInt32 / sizeofEpollEvent // Linux: fs/eventpoll.c:EP_MAX_EVENTS
 	if maxEvents <= 0 || maxEvents > _EP_MAX_EVENTS {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	epfile := t.GetFileVFS2(epfd)
@@ -133,7 +133,7 @@ func waitEpoll(t *kernel.Task, epfd int32, eventsAddr hostarch.Addr, maxEvents i
 	defer epfile.DecRef(t)
 	ep, ok := epfile.Impl().(*vfs.EpollInstance)
 	if !ok {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Allocate space for a few events on the stack for the common case in
diff --git a/pkg/sentry/syscalls/linux/vfs2/eventfd.go b/pkg/sentry/syscalls/linux/vfs2/eventfd.go
index 807f909da..0dcf1fbff 100644
--- a/pkg/sentry/syscalls/linux/vfs2/eventfd.go
+++ b/pkg/sentry/syscalls/linux/vfs2/eventfd.go
@@ -16,10 +16,10 @@ package vfs2
 
 import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
 	"gvisor.dev/gvisor/pkg/sentry/fsimpl/eventfd"
 	"gvisor.dev/gvisor/pkg/sentry/kernel"
-	"gvisor.dev/gvisor/pkg/syserror"
 )
 
 // Eventfd2 implements linux syscall eventfd2(2).
@@ -29,7 +29,7 @@ func Eventfd2(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysc
 	allOps := uint(linux.EFD_SEMAPHORE | linux.EFD_NONBLOCK | linux.EFD_CLOEXEC)
 
 	if flags & ^allOps != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	vfsObj := t.Kernel().VFS()
diff --git a/pkg/sentry/syscalls/linux/vfs2/execve.go b/pkg/sentry/syscalls/linux/vfs2/execve.go
index 3315398a4..7b1e1da78 100644
--- a/pkg/sentry/syscalls/linux/vfs2/execve.go
+++ b/pkg/sentry/syscalls/linux/vfs2/execve.go
@@ -16,7 +16,9 @@ package vfs2
 
 import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/fspath"
+	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
 	"gvisor.dev/gvisor/pkg/sentry/fsbridge"
 	"gvisor.dev/gvisor/pkg/sentry/kernel"
@@ -24,8 +26,6 @@ import (
 	slinux "gvisor.dev/gvisor/pkg/sentry/syscalls/linux"
 	"gvisor.dev/gvisor/pkg/sentry/vfs"
 	"gvisor.dev/gvisor/pkg/syserror"
-
-	"gvisor.dev/gvisor/pkg/hostarch"
 )
 
 // Execve implements linux syscall execve(2).
@@ -48,7 +48,7 @@ func Execveat(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysc
 
 func execveat(t *kernel.Task, dirfd int32, pathnameAddr, argvAddr, envvAddr hostarch.Addr, flags int32) (uintptr, *kernel.SyscallControl, error) {
 	if flags&^(linux.AT_EMPTY_PATH|linux.AT_SYMLINK_NOFOLLOW) != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	pathname, err := t.CopyInString(pathnameAddr, linux.PATH_MAX)
diff --git a/pkg/sentry/syscalls/linux/vfs2/fd.go b/pkg/sentry/syscalls/linux/vfs2/fd.go
index 1a31898e8..ea34ff471 100644
--- a/pkg/sentry/syscalls/linux/vfs2/fd.go
+++ b/pkg/sentry/syscalls/linux/vfs2/fd.go
@@ -16,6 +16,7 @@ package vfs2
 
 import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
 	"gvisor.dev/gvisor/pkg/sentry/fs/lock"
 	"gvisor.dev/gvisor/pkg/sentry/fsimpl/tmpfs"
@@ -86,7 +87,7 @@ func Dup3(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallC
 	flags := args[2].Uint()
 
 	if oldfd == newfd {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	return dup3(t, oldfd, newfd, flags)
@@ -94,7 +95,7 @@ func Dup3(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallC
 
 func dup3(t *kernel.Task, oldfd, newfd int32, flags uint32) (uintptr, *kernel.SyscallControl, error) {
 	if flags&^linux.O_CLOEXEC != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	file := t.GetFileVFS2(oldfd)
@@ -169,7 +170,7 @@ func Fcntl(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscall
 		if who < 0 {
 			// Check for overflow before flipping the sign.
 			if who-1 > who {
-				return 0, nil, syserror.EINVAL
+				return 0, nil, linuxerr.EINVAL
 			}
 			ownerType = linux.F_OWNER_PGRP
 			who = -who
@@ -232,7 +233,7 @@ func Fcntl(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscall
 		return 0, nil, a.SetSignal(linux.Signal(args[2].Int()))
 	default:
 		// Everything else is not yet supported.
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 }
 
@@ -269,7 +270,7 @@ func setAsyncOwner(t *kernel.Task, fd int, file *vfs.FileDescription, ownerType,
 	case linux.F_OWNER_TID, linux.F_OWNER_PID, linux.F_OWNER_PGRP:
 		// Acceptable type.
 	default:
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 
 	a := file.SetAsyncHandler(fasync.NewVFS2(fd)).(*fasync.FileAsync)
@@ -301,7 +302,7 @@ func setAsyncOwner(t *kernel.Task, fd int, file *vfs.FileDescription, ownerType,
 		a.SetOwnerProcessGroup(t, pg)
 		return nil
 	default:
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 }
 
@@ -319,7 +320,7 @@ func posixTestLock(t *kernel.Task, args arch.SyscallArguments, file *vfs.FileDes
 	case linux.F_WRLCK:
 		typ = lock.WriteLock
 	default:
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	r, err := file.ComputeLockRange(t, uint64(flock.Start), uint64(flock.Len), flock.Whence)
 	if err != nil {
@@ -382,7 +383,7 @@ func posixLock(t *kernel.Task, args arch.SyscallArguments, file *vfs.FileDescrip
 		return file.UnlockPOSIX(t, t.FDTable(), r)
 
 	default:
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 }
 
@@ -395,7 +396,7 @@ func Fadvise64(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sys
 
 	// Note: offset is allowed to be negative.
 	if length < 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	file := t.GetFileVFS2(fd)
@@ -421,7 +422,7 @@ func Fadvise64(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sys
 	case linux.POSIX_FADV_DONTNEED:
 	case linux.POSIX_FADV_NOREUSE:
 	default:
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Sure, whatever.
diff --git a/pkg/sentry/syscalls/linux/vfs2/filesystem.go b/pkg/sentry/syscalls/linux/vfs2/filesystem.go
index 36aa1d3ae..534355237 100644
--- a/pkg/sentry/syscalls/linux/vfs2/filesystem.go
+++ b/pkg/sentry/syscalls/linux/vfs2/filesystem.go
@@ -16,12 +16,12 @@ package vfs2
 
 import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
+	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
 	"gvisor.dev/gvisor/pkg/sentry/kernel"
 	"gvisor.dev/gvisor/pkg/sentry/vfs"
 	"gvisor.dev/gvisor/pkg/syserror"
-
-	"gvisor.dev/gvisor/pkg/hostarch"
 )
 
 // Link implements Linux syscall link(2).
@@ -43,7 +43,7 @@ func Linkat(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscal
 
 func linkat(t *kernel.Task, olddirfd int32, oldpathAddr hostarch.Addr, newdirfd int32, newpathAddr hostarch.Addr, flags int32) error {
 	if flags&^(linux.AT_EMPTY_PATH|linux.AT_SYMLINK_FOLLOW) != 0 {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	if flags&linux.AT_EMPTY_PATH != 0 && !t.HasCapability(linux.CAP_DAC_READ_SEARCH) {
 		return syserror.ENOENT
@@ -290,7 +290,7 @@ func Unlinkat(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysc
 	flags := args[2].Int()
 
 	if flags&^linux.AT_REMOVEDIR != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	if flags&linux.AT_REMOVEDIR != 0 {
diff --git a/pkg/sentry/syscalls/linux/vfs2/getdents.go b/pkg/sentry/syscalls/linux/vfs2/getdents.go
index b41a3056a..8ace31af3 100644
--- a/pkg/sentry/syscalls/linux/vfs2/getdents.go
+++ b/pkg/sentry/syscalls/linux/vfs2/getdents.go
@@ -17,13 +17,13 @@ package vfs2
 import (
 	"fmt"
 
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
+	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
 	"gvisor.dev/gvisor/pkg/sentry/kernel"
 	"gvisor.dev/gvisor/pkg/sentry/vfs"
 	"gvisor.dev/gvisor/pkg/sync"
 	"gvisor.dev/gvisor/pkg/syserror"
-
-	"gvisor.dev/gvisor/pkg/hostarch"
 )
 
 // Getdents implements Linux syscall getdents(2).
@@ -100,7 +100,7 @@ func (cb *getdentsCallback) Handle(dirent vfs.Dirent) error {
 		size := 8 + 8 + 2 + 1 + 1 + len(dirent.Name)
 		size = (size + 7) &^ 7 // round up to multiple of 8
 		if size > cb.remaining {
-			return syserror.EINVAL
+			return linuxerr.EINVAL
 		}
 		buf = cb.t.CopyScratchBuffer(size)
 		hostarch.ByteOrder.PutUint64(buf[0:8], dirent.Ino)
@@ -134,7 +134,7 @@ func (cb *getdentsCallback) Handle(dirent vfs.Dirent) error {
 		size := 8 + 8 + 2 + 1 + 1 + len(dirent.Name)
 		size = (size + 7) &^ 7 // round up to multiple of sizeof(long)
 		if size > cb.remaining {
-			return syserror.EINVAL
+			return linuxerr.EINVAL
 		}
 		buf = cb.t.CopyScratchBuffer(size)
 		hostarch.ByteOrder.PutUint64(buf[0:8], dirent.Ino)
diff --git a/pkg/sentry/syscalls/linux/vfs2/inotify.go b/pkg/sentry/syscalls/linux/vfs2/inotify.go
index 11753d8e5..7a2e9e75d 100644
--- a/pkg/sentry/syscalls/linux/vfs2/inotify.go
+++ b/pkg/sentry/syscalls/linux/vfs2/inotify.go
@@ -16,6 +16,7 @@ package vfs2
 
 import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
 	"gvisor.dev/gvisor/pkg/sentry/kernel"
 	"gvisor.dev/gvisor/pkg/sentry/vfs"
@@ -28,7 +29,7 @@ const allFlags = linux.IN_NONBLOCK | linux.IN_CLOEXEC
 func InotifyInit1(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallControl, error) {
 	flags := args[0].Int()
 	if flags&^allFlags != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	ino, err := vfs.NewInotifyFD(t, t.Kernel().VFS(), uint32(flags))
@@ -67,7 +68,7 @@ func fdToInotify(t *kernel.Task, fd int32) (*vfs.Inotify, *vfs.FileDescription,
 	if !ok {
 		// Not an inotify fd.
 		f.DecRef(t)
-		return nil, nil, syserror.EINVAL
+		return nil, nil, linuxerr.EINVAL
 	}
 
 	return ino, f, nil
@@ -82,7 +83,7 @@ func InotifyAddWatch(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kern
 	// "EINVAL: The given event mask contains no valid events."
 	// -- inotify_add_watch(2)
 	if mask&linux.ALL_INOTIFY_BITS == 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// "IN_DONT_FOLLOW: Don't dereference pathname if it is a symbolic link."
diff --git a/pkg/sentry/syscalls/linux/vfs2/ioctl.go b/pkg/sentry/syscalls/linux/vfs2/ioctl.go
index c7c3fed57..9852e3fe4 100644
--- a/pkg/sentry/syscalls/linux/vfs2/ioctl.go
+++ b/pkg/sentry/syscalls/linux/vfs2/ioctl.go
@@ -16,6 +16,7 @@ package vfs2
 
 import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/marshal/primitive"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
 	"gvisor.dev/gvisor/pkg/sentry/kernel"
@@ -99,7 +100,7 @@ func Ioctl(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscall
 		if who < 0 {
 			// Check for overflow before flipping the sign.
 			if who-1 > who {
-				return 0, nil, syserror.EINVAL
+				return 0, nil, linuxerr.EINVAL
 			}
 			ownerType = linux.F_OWNER_PGRP
 			who = -who
diff --git a/pkg/sentry/syscalls/linux/vfs2/lock.go b/pkg/sentry/syscalls/linux/vfs2/lock.go
index d1452a04d..80cb3ba09 100644
--- a/pkg/sentry/syscalls/linux/vfs2/lock.go
+++ b/pkg/sentry/syscalls/linux/vfs2/lock.go
@@ -16,6 +16,7 @@ package vfs2
 
 import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
 	"gvisor.dev/gvisor/pkg/sentry/fs/lock"
 	"gvisor.dev/gvisor/pkg/sentry/kernel"
@@ -57,7 +58,7 @@ func Flock(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscall
 		}
 	default:
 		// flock(2): EINVAL operation is invalid.
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	return 0, nil, nil
diff --git a/pkg/sentry/syscalls/linux/vfs2/memfd.go b/pkg/sentry/syscalls/linux/vfs2/memfd.go
index c4c0f9e0a..70c2cf5a5 100644
--- a/pkg/sentry/syscalls/linux/vfs2/memfd.go
+++ b/pkg/sentry/syscalls/linux/vfs2/memfd.go
@@ -16,10 +16,10 @@ package vfs2
 
 import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
 	"gvisor.dev/gvisor/pkg/sentry/fsimpl/tmpfs"
 	"gvisor.dev/gvisor/pkg/sentry/kernel"
-	"gvisor.dev/gvisor/pkg/syserror"
 )
 
 const (
@@ -35,7 +35,7 @@ func MemfdCreate(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.S
 
 	if flags&^memfdAllFlags != 0 {
 		// Unknown bits in flags.
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	allowSeals := flags&linux.MFD_ALLOW_SEALING != 0
diff --git a/pkg/sentry/syscalls/linux/vfs2/mmap.go b/pkg/sentry/syscalls/linux/vfs2/mmap.go
index c961545f6..db8d59899 100644
--- a/pkg/sentry/syscalls/linux/vfs2/mmap.go
+++ b/pkg/sentry/syscalls/linux/vfs2/mmap.go
@@ -16,13 +16,13 @@ package vfs2
 
 import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
+	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
 	"gvisor.dev/gvisor/pkg/sentry/fsimpl/tmpfs"
 	"gvisor.dev/gvisor/pkg/sentry/kernel"
 	"gvisor.dev/gvisor/pkg/sentry/memmap"
 	"gvisor.dev/gvisor/pkg/syserror"
-
-	"gvisor.dev/gvisor/pkg/hostarch"
 )
 
 // Mmap implements Linux syscall mmap(2).
@@ -38,7 +38,7 @@ func Mmap(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallC
 
 	// Require exactly one of MAP_PRIVATE and MAP_SHARED.
 	if private == shared {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	opts := memmap.MMapOpts{
diff --git a/pkg/sentry/syscalls/linux/vfs2/mount.go b/pkg/sentry/syscalls/linux/vfs2/mount.go
index dd93430e2..667e48744 100644
--- a/pkg/sentry/syscalls/linux/vfs2/mount.go
+++ b/pkg/sentry/syscalls/linux/vfs2/mount.go
@@ -16,12 +16,12 @@ package vfs2
 
 import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
+	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
 	"gvisor.dev/gvisor/pkg/sentry/kernel"
 	"gvisor.dev/gvisor/pkg/sentry/vfs"
 	"gvisor.dev/gvisor/pkg/syserror"
-
-	"gvisor.dev/gvisor/pkg/hostarch"
 )
 
 // Mount implements Linux syscall mount(2).
@@ -84,7 +84,7 @@ func Mount(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscall
 	// unknown or unsupported flags are passed. Since we don't implement
 	// everything, we fail explicitly on flags that are unimplemented.
 	if flags&(unsupportedOps|unsupportedFlags) != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	var opts vfs.MountOptions
@@ -130,7 +130,7 @@ func Umount2(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysca
 
 	const unsupported = linux.MNT_FORCE | linux.MNT_EXPIRE
 	if flags&unsupported != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	path, err := copyInPath(t, addr)
diff --git a/pkg/sentry/syscalls/linux/vfs2/pipe.go b/pkg/sentry/syscalls/linux/vfs2/pipe.go
index c6fc1954c..07a89cf4e 100644
--- a/pkg/sentry/syscalls/linux/vfs2/pipe.go
+++ b/pkg/sentry/syscalls/linux/vfs2/pipe.go
@@ -16,14 +16,13 @@ package vfs2
 
 import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
+	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/marshal/primitive"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
 	"gvisor.dev/gvisor/pkg/sentry/fsimpl/pipefs"
 	"gvisor.dev/gvisor/pkg/sentry/kernel"
 	"gvisor.dev/gvisor/pkg/sentry/vfs"
-	"gvisor.dev/gvisor/pkg/syserror"
-
-	"gvisor.dev/gvisor/pkg/hostarch"
 )
 
 // Pipe implements Linux syscall pipe(2).
@@ -41,7 +40,7 @@ func Pipe2(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscall
 
 func pipe2(t *kernel.Task, addr hostarch.Addr, flags int32) error {
 	if flags&^(linux.O_NONBLOCK|linux.O_CLOEXEC) != 0 {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	r, w, err := pipefs.NewConnectedPipeFDs(t, t.Kernel().PipeMount(), uint32(flags&linux.O_NONBLOCK))
 	if err != nil {
diff --git a/pkg/sentry/syscalls/linux/vfs2/poll.go b/pkg/sentry/syscalls/linux/vfs2/poll.go
index b16773d65..ea95dd78c 100644
--- a/pkg/sentry/syscalls/linux/vfs2/poll.go
+++ b/pkg/sentry/syscalls/linux/vfs2/poll.go
@@ -162,7 +162,7 @@ func pollBlock(t *kernel.Task, pfd []linux.PollFD, timeout time.Duration) (time.
 // copyInPollFDs copies an array of struct pollfd unless nfds exceeds the max.
 func copyInPollFDs(t *kernel.Task, addr hostarch.Addr, nfds uint) ([]linux.PollFD, error) {
 	if uint64(nfds) > t.ThreadGroup().Limits().GetCapped(limits.NumberOfFiles, fileCap) {
-		return nil, syserror.EINVAL
+		return nil, linuxerr.EINVAL
 	}
 
 	pfd := make([]linux.PollFD, nfds)
@@ -222,7 +222,7 @@ func CopyInFDSet(t *kernel.Task, addr hostarch.Addr, nBytes, nBitsInLastPartialB
 
 func doSelect(t *kernel.Task, nfds int, readFDs, writeFDs, exceptFDs hostarch.Addr, timeout time.Duration) (uintptr, error) {
 	if nfds < 0 || nfds > fileCap {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 
 	// Calculate the size of the fd sets (one bit per fd).
@@ -485,7 +485,7 @@ func Select(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscal
 			return 0, nil, err
 		}
 		if timeval.Sec < 0 || timeval.Usec < 0 {
-			return 0, nil, syserror.EINVAL
+			return 0, nil, linuxerr.EINVAL
 		}
 		timeout = time.Duration(timeval.ToNsecCapped())
 	}
@@ -562,7 +562,7 @@ func copyTimespecInToDuration(t *kernel.Task, timespecAddr hostarch.Addr) (time.
 			return 0, err
 		}
 		if !timespec.Valid() {
-			return 0, syserror.EINVAL
+			return 0, linuxerr.EINVAL
 		}
 		timeout = time.Duration(timespec.ToNsecCapped())
 	}
@@ -574,7 +574,7 @@ func setTempSignalSet(t *kernel.Task, maskAddr hostarch.Addr, maskSize uint) err
 		return nil
 	}
 	if maskSize != linux.SignalSetSize {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	var mask linux.SignalSet
 	if _, err := mask.CopyIn(t, maskAddr); err != nil {
diff --git a/pkg/sentry/syscalls/linux/vfs2/read_write.go b/pkg/sentry/syscalls/linux/vfs2/read_write.go
index bbfa4c6d7..3e515f6fd 100644
--- a/pkg/sentry/syscalls/linux/vfs2/read_write.go
+++ b/pkg/sentry/syscalls/linux/vfs2/read_write.go
@@ -50,7 +50,7 @@ func Read(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallC
 	// Check that the size is legitimate.
 	si := int(size)
 	if si < 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Get the destination of the read.
@@ -147,13 +147,13 @@ func Pread64(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysca
 
 	// Check that the offset is legitimate and does not overflow.
 	if offset < 0 || offset+int64(size) < 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Check that the size is legitimate.
 	si := int(size)
 	if si < 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Get the destination of the read.
@@ -184,7 +184,7 @@ func Preadv(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscal
 
 	// Check that the offset is legitimate.
 	if offset < 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Get the destination of the read.
@@ -222,7 +222,7 @@ func Preadv2(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysca
 
 	// Check that the offset is legitimate.
 	if offset < -1 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Get the destination of the read.
@@ -301,7 +301,7 @@ func Write(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscall
 	// Check that the size is legitimate.
 	si := int(size)
 	if si < 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Get the source of the write.
@@ -397,13 +397,13 @@ func Pwrite64(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysc
 
 	// Check that the offset is legitimate and does not overflow.
 	if offset < 0 || offset+int64(size) < 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Check that the size is legitimate.
 	si := int(size)
 	if si < 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Get the source of the write.
@@ -434,7 +434,7 @@ func Pwritev(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysca
 
 	// Check that the offset is legitimate.
 	if offset < 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Get the source of the write.
@@ -472,7 +472,7 @@ func Pwritev2(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysc
 
 	// Check that the offset is legitimate.
 	if offset < -1 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Get the source of the write.
@@ -588,16 +588,16 @@ func Readahead(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sys
 
 	// Check that the size is valid.
 	if int(size) < 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Check that the offset is legitimate and does not overflow.
 	if offset < 0 || offset+int64(size) < 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Return EINVAL; if the underlying file type does not support readahead,
 	// then Linux will return EINVAL to indicate as much. In the future, we
 	// may extend this function to actually support readahead hints.
-	return 0, nil, syserror.EINVAL
+	return 0, nil, linuxerr.EINVAL
 }
diff --git a/pkg/sentry/syscalls/linux/vfs2/setstat.go b/pkg/sentry/syscalls/linux/vfs2/setstat.go
index 647e089d0..0fbafd6f6 100644
--- a/pkg/sentry/syscalls/linux/vfs2/setstat.go
+++ b/pkg/sentry/syscalls/linux/vfs2/setstat.go
@@ -16,15 +16,15 @@ package vfs2
 
 import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/fspath"
+	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
 	"gvisor.dev/gvisor/pkg/sentry/kernel"
 	"gvisor.dev/gvisor/pkg/sentry/kernel/auth"
 	"gvisor.dev/gvisor/pkg/sentry/limits"
 	"gvisor.dev/gvisor/pkg/sentry/vfs"
 	"gvisor.dev/gvisor/pkg/syserror"
-
-	"gvisor.dev/gvisor/pkg/hostarch"
 )
 
 const chmodMask = 0777 | linux.S_ISUID | linux.S_ISGID | linux.S_ISVTX
@@ -105,7 +105,7 @@ func Fchownat(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysc
 
 func fchownat(t *kernel.Task, dirfd int32, pathAddr hostarch.Addr, owner, group, flags int32) error {
 	if flags&^(linux.AT_EMPTY_PATH|linux.AT_SYMLINK_NOFOLLOW) != 0 {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 
 	path, err := copyInPath(t, pathAddr)
@@ -126,7 +126,7 @@ func populateSetStatOptionsForChown(t *kernel.Task, owner, group int32, opts *vf
 	if owner != -1 {
 		kuid := userns.MapToKUID(auth.UID(owner))
 		if !kuid.Ok() {
-			return syserror.EINVAL
+			return linuxerr.EINVAL
 		}
 		opts.Stat.Mask |= linux.STATX_UID
 		opts.Stat.UID = uint32(kuid)
@@ -134,7 +134,7 @@ func populateSetStatOptionsForChown(t *kernel.Task, owner, group int32, opts *vf
 	if group != -1 {
 		kgid := userns.MapToKGID(auth.GID(group))
 		if !kgid.Ok() {
-			return syserror.EINVAL
+			return linuxerr.EINVAL
 		}
 		opts.Stat.Mask |= linux.STATX_GID
 		opts.Stat.GID = uint32(kgid)
@@ -167,7 +167,7 @@ func Truncate(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysc
 	length := args[1].Int64()
 
 	if length < 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	path, err := copyInPath(t, addr)
@@ -191,7 +191,7 @@ func Ftruncate(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sys
 	length := args[1].Int64()
 
 	if length < 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	file := t.GetFileVFS2(fd)
@@ -201,7 +201,7 @@ func Ftruncate(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sys
 	defer file.DecRef(t)
 
 	if !file.IsWritable() {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	err := file.SetStat(t, vfs.SetStatOptions{
@@ -233,7 +233,7 @@ func Fallocate(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sys
 		return 0, nil, syserror.ENOTSUP
 	}
 	if offset < 0 || length <= 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	size := offset + length
@@ -340,7 +340,7 @@ func populateSetStatOptionsForUtimes(t *kernel.Task, timesAddr hostarch.Addr, op
 		return err
 	}
 	if times[0].Usec < 0 || times[0].Usec > 999999 || times[1].Usec < 0 || times[1].Usec > 999999 {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	opts.Stat.Mask = linux.STATX_ATIME | linux.STATX_MTIME
 	opts.Stat.Atime = linux.StatxTimestamp{
@@ -372,7 +372,7 @@ func Utimensat(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sys
 	}
 
 	if flags&^linux.AT_SYMLINK_NOFOLLOW != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// "If filename is NULL and dfd refers to an open file, then operate on the
@@ -405,7 +405,7 @@ func populateSetStatOptionsForUtimens(t *kernel.Task, timesAddr hostarch.Addr, o
 	}
 	if times[0].Nsec != linux.UTIME_OMIT {
 		if times[0].Nsec != linux.UTIME_NOW && (times[0].Nsec < 0 || times[0].Nsec > 999999999) {
-			return syserror.EINVAL
+			return linuxerr.EINVAL
 		}
 		opts.Stat.Mask |= linux.STATX_ATIME
 		opts.Stat.Atime = linux.StatxTimestamp{
@@ -415,7 +415,7 @@ func populateSetStatOptionsForUtimens(t *kernel.Task, timesAddr hostarch.Addr, o
 	}
 	if times[1].Nsec != linux.UTIME_OMIT {
 		if times[1].Nsec != linux.UTIME_NOW && (times[1].Nsec < 0 || times[1].Nsec > 999999999) {
-			return syserror.EINVAL
+			return linuxerr.EINVAL
 		}
 		opts.Stat.Mask |= linux.STATX_MTIME
 		opts.Stat.Mtime = linux.StatxTimestamp{
diff --git a/pkg/sentry/syscalls/linux/vfs2/signal.go b/pkg/sentry/syscalls/linux/vfs2/signal.go
index 6163da103..8b219cba7 100644
--- a/pkg/sentry/syscalls/linux/vfs2/signal.go
+++ b/pkg/sentry/syscalls/linux/vfs2/signal.go
@@ -16,13 +16,13 @@ package vfs2
 
 import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
+	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
 	"gvisor.dev/gvisor/pkg/sentry/fsimpl/signalfd"
 	"gvisor.dev/gvisor/pkg/sentry/kernel"
 	slinux "gvisor.dev/gvisor/pkg/sentry/syscalls/linux"
 	"gvisor.dev/gvisor/pkg/syserror"
-
-	"gvisor.dev/gvisor/pkg/hostarch"
 )
 
 // sharedSignalfd is shared between the two calls.
@@ -35,7 +35,7 @@ func sharedSignalfd(t *kernel.Task, fd int32, sigset hostarch.Addr, sigsetsize u
 
 	// Always check for valid flags, even if not creating.
 	if flags&^(linux.SFD_NONBLOCK|linux.SFD_CLOEXEC) != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Is this a change to an existing signalfd?
@@ -55,7 +55,7 @@ func sharedSignalfd(t *kernel.Task, fd int32, sigset hostarch.Addr, sigsetsize u
 		}
 
 		// Not a signalfd.
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	fileFlags := uint32(linux.O_RDWR)
diff --git a/pkg/sentry/syscalls/linux/vfs2/socket.go b/pkg/sentry/syscalls/linux/vfs2/socket.go
index 9a4b5e5fc..c78c7d951 100644
--- a/pkg/sentry/syscalls/linux/vfs2/socket.go
+++ b/pkg/sentry/syscalls/linux/vfs2/socket.go
@@ -118,7 +118,7 @@ type multipleMessageHeader64 struct {
 // from the untrusted address space range.
 func CaptureAddress(t *kernel.Task, addr hostarch.Addr, addrlen uint32) ([]byte, error) {
 	if addrlen > maxAddrLen {
-		return nil, syserror.EINVAL
+		return nil, linuxerr.EINVAL
 	}
 
 	addrBuf := make([]byte, addrlen)
@@ -140,7 +140,7 @@ func writeAddress(t *kernel.Task, addr linux.SockAddr, addrLen uint32, addrPtr h
 	}
 
 	if int32(bufLen) < 0 {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 
 	// Write the length unconditionally.
@@ -174,7 +174,7 @@ func Socket(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscal
 
 	// Check and initialize the flags.
 	if stype & ^(0xf|linux.SOCK_NONBLOCK|linux.SOCK_CLOEXEC) != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Create the new socket.
@@ -207,7 +207,7 @@ func SocketPair(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sy
 
 	// Check and initialize the flags.
 	if stype & ^(0xf|linux.SOCK_NONBLOCK|linux.SOCK_CLOEXEC) != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Create the socket pair.
@@ -282,7 +282,7 @@ func Connect(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysca
 func accept(t *kernel.Task, fd int32, addr hostarch.Addr, addrLen hostarch.Addr, flags int) (uintptr, error) {
 	// Check that no unsupported flags are passed in.
 	if flags & ^(linux.SOCK_NONBLOCK|linux.SOCK_CLOEXEC) != 0 {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 
 	// Get socket from the file descriptor.
@@ -426,7 +426,7 @@ func Shutdown(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysc
 	switch how {
 	case linux.SHUT_RD, linux.SHUT_WR, linux.SHUT_RDWR:
 	default:
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	return 0, nil, s.Shutdown(t, int(how)).ToError()
@@ -459,7 +459,7 @@ func GetSockOpt(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sy
 		return 0, nil, err
 	}
 	if optLen < 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Call syscall implementation then copy both value and value len out.
@@ -535,10 +535,10 @@ func SetSockOpt(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sy
 	}
 
 	if optLen < 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 	if optLen > maxOptLen {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 	buf := t.CopyScratchBuffer(int(optLen))
 	if _, err := t.CopyInBytes(optValAddr, buf); err != nil {
@@ -617,7 +617,7 @@ func RecvMsg(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysca
 
 	if t.Arch().Width() != 8 {
 		// We only handle 64-bit for now.
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Get socket from the file descriptor.
@@ -635,7 +635,7 @@ func RecvMsg(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysca
 
 	// Reject flags that we don't handle yet.
 	if flags & ^(baseRecvFlags|linux.MSG_PEEK|linux.MSG_CMSG_CLOEXEC|linux.MSG_ERRQUEUE) != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	if (file.StatusFlags() & linux.SOCK_NONBLOCK) != 0 {
@@ -665,7 +665,7 @@ func RecvMMsg(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysc
 
 	if t.Arch().Width() != 8 {
 		// We only handle 64-bit for now.
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	if vlen > linux.UIO_MAXIOV {
@@ -674,7 +674,7 @@ func RecvMMsg(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysc
 
 	// Reject flags that we don't handle yet.
 	if flags & ^(baseRecvFlags|linux.MSG_CMSG_CLOEXEC|linux.MSG_ERRQUEUE) != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Get socket from the file descriptor.
@@ -702,7 +702,7 @@ func RecvMMsg(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysc
 			return 0, nil, err
 		}
 		if !ts.Valid() {
-			return 0, nil, syserror.EINVAL
+			return 0, nil, linuxerr.EINVAL
 		}
 		deadline = t.Kernel().MonotonicClock().Now().Add(ts.ToDuration())
 		haveDeadline = true
@@ -834,12 +834,12 @@ func recvSingleMsg(t *kernel.Task, s socket.SocketVFS2, msgPtr hostarch.Addr, fl
 // recvfrom and recv syscall handlers.
 func recvFrom(t *kernel.Task, fd int32, bufPtr hostarch.Addr, bufLen uint64, flags int32, namePtr hostarch.Addr, nameLenPtr hostarch.Addr) (uintptr, error) {
 	if int(bufLen) < 0 {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 
 	// Reject flags that we don't handle yet.
 	if flags & ^(baseRecvFlags|linux.MSG_PEEK|linux.MSG_CONFIRM) != 0 {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 
 	// Get socket from the file descriptor.
@@ -912,7 +912,7 @@ func SendMsg(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysca
 
 	if t.Arch().Width() != 8 {
 		// We only handle 64-bit for now.
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Get socket from the file descriptor.
@@ -930,7 +930,7 @@ func SendMsg(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysca
 
 	// Reject flags that we don't handle yet.
 	if flags & ^(linux.MSG_DONTWAIT|linux.MSG_EOR|linux.MSG_MORE|linux.MSG_NOSIGNAL) != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	if (file.StatusFlags() & linux.SOCK_NONBLOCK) != 0 {
@@ -950,7 +950,7 @@ func SendMMsg(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysc
 
 	if t.Arch().Width() != 8 {
 		// We only handle 64-bit for now.
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	if vlen > linux.UIO_MAXIOV {
@@ -972,7 +972,7 @@ func SendMMsg(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysc
 
 	// Reject flags that we don't handle yet.
 	if flags & ^(linux.MSG_DONTWAIT|linux.MSG_EOR|linux.MSG_MORE|linux.MSG_NOSIGNAL) != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	if (file.StatusFlags() & linux.SOCK_NONBLOCK) != 0 {
@@ -1078,7 +1078,7 @@ func sendSingleMsg(t *kernel.Task, s socket.SocketVFS2, file *vfs.FileDescriptio
 func sendTo(t *kernel.Task, fd int32, bufPtr hostarch.Addr, bufLen uint64, flags int32, namePtr hostarch.Addr, nameLen uint32) (uintptr, error) {
 	bl := int(bufLen)
 	if bl < 0 {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 
 	// Get socket from the file descriptor.
diff --git a/pkg/sentry/syscalls/linux/vfs2/splice.go b/pkg/sentry/syscalls/linux/vfs2/splice.go
index 19e175203..6ddc72999 100644
--- a/pkg/sentry/syscalls/linux/vfs2/splice.go
+++ b/pkg/sentry/syscalls/linux/vfs2/splice.go
@@ -18,6 +18,7 @@ import (
 	"io"
 
 	"gvisor.dev/gvisor/pkg/abi/linux"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/log"
 	"gvisor.dev/gvisor/pkg/marshal/primitive"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
@@ -46,12 +47,12 @@ func Splice(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscal
 		count = int64(kernel.MAX_RW_COUNT)
 	}
 	if count < 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Check for invalid flags.
 	if flags&^(linux.SPLICE_F_MOVE|linux.SPLICE_F_NONBLOCK|linux.SPLICE_F_MORE|linux.SPLICE_F_GIFT) != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Get file descriptions.
@@ -82,7 +83,7 @@ func Splice(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscal
 	inPipeFD, inIsPipe := inFile.Impl().(*pipe.VFSPipeFD)
 	outPipeFD, outIsPipe := outFile.Impl().(*pipe.VFSPipeFD)
 	if !inIsPipe && !outIsPipe {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Copy in offsets.
@@ -92,13 +93,13 @@ func Splice(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscal
 			return 0, nil, syserror.ESPIPE
 		}
 		if inFile.Options().DenyPRead {
-			return 0, nil, syserror.EINVAL
+			return 0, nil, linuxerr.EINVAL
 		}
 		if _, err := primitive.CopyInt64In(t, inOffsetPtr, &inOffset); err != nil {
 			return 0, nil, err
 		}
 		if inOffset < 0 {
-			return 0, nil, syserror.EINVAL
+			return 0, nil, linuxerr.EINVAL
 		}
 	}
 	outOffset := int64(-1)
@@ -107,13 +108,13 @@ func Splice(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscal
 			return 0, nil, syserror.ESPIPE
 		}
 		if outFile.Options().DenyPWrite {
-			return 0, nil, syserror.EINVAL
+			return 0, nil, linuxerr.EINVAL
 		}
 		if _, err := primitive.CopyInt64In(t, outOffsetPtr, &outOffset); err != nil {
 			return 0, nil, err
 		}
 		if outOffset < 0 {
-			return 0, nil, syserror.EINVAL
+			return 0, nil, linuxerr.EINVAL
 		}
 	}
 
@@ -189,12 +190,12 @@ func Tee(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallCo
 		count = int64(kernel.MAX_RW_COUNT)
 	}
 	if count < 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Check for invalid flags.
 	if flags&^(linux.SPLICE_F_MOVE|linux.SPLICE_F_NONBLOCK|linux.SPLICE_F_MORE|linux.SPLICE_F_GIFT) != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Get file descriptions.
@@ -225,7 +226,7 @@ func Tee(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.SyscallCo
 	inPipeFD, inIsPipe := inFile.Impl().(*pipe.VFSPipeFD)
 	outPipeFD, outIsPipe := outFile.Impl().(*pipe.VFSPipeFD)
 	if !inIsPipe || !outIsPipe {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Copy data.
@@ -288,7 +289,7 @@ func Sendfile(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysc
 
 	// Verify that the outFile Append flag is not set.
 	if outFile.StatusFlags()&linux.O_APPEND != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Verify that inFile is a regular file or block device. This is a
@@ -298,7 +299,7 @@ func Sendfile(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysc
 		return 0, nil, err
 	} else if stat.Mask&linux.STATX_TYPE == 0 ||
 		(stat.Mode&linux.S_IFMT != linux.S_IFREG && stat.Mode&linux.S_IFMT != linux.S_IFBLK) {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Copy offset if it exists.
@@ -314,16 +315,16 @@ func Sendfile(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sysc
 		offset = int64(offsetP)
 
 		if offset < 0 {
-			return 0, nil, syserror.EINVAL
+			return 0, nil, linuxerr.EINVAL
 		}
 		if offset+count < 0 {
-			return 0, nil, syserror.EINVAL
+			return 0, nil, linuxerr.EINVAL
 		}
 	}
 
 	// Validate count. This must come after offset checks.
 	if count < 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 	if count == 0 {
 		return 0, nil, nil
diff --git a/pkg/sentry/syscalls/linux/vfs2/stat.go b/pkg/sentry/syscalls/linux/vfs2/stat.go
index 69e77fa99..8a22ed8a5 100644
--- a/pkg/sentry/syscalls/linux/vfs2/stat.go
+++ b/pkg/sentry/syscalls/linux/vfs2/stat.go
@@ -17,15 +17,15 @@ package vfs2
 import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/bits"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/fspath"
 	"gvisor.dev/gvisor/pkg/gohacks"
+	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
 	"gvisor.dev/gvisor/pkg/sentry/kernel"
 	"gvisor.dev/gvisor/pkg/sentry/kernel/auth"
 	"gvisor.dev/gvisor/pkg/sentry/vfs"
 	"gvisor.dev/gvisor/pkg/syserror"
-
-	"gvisor.dev/gvisor/pkg/hostarch"
 )
 
 // Stat implements Linux syscall stat(2).
@@ -53,7 +53,7 @@ func Newfstatat(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sy
 
 func fstatat(t *kernel.Task, dirfd int32, pathAddr, statAddr hostarch.Addr, flags int32) error {
 	if flags&^(linux.AT_EMPTY_PATH|linux.AT_SYMLINK_NOFOLLOW) != 0 {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 
 	opts := vfs.StatOptions{
@@ -156,15 +156,15 @@ func Statx(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscall
 	statxAddr := args[4].Pointer()
 
 	if flags&^(linux.AT_EMPTY_PATH|linux.AT_SYMLINK_NOFOLLOW|linux.AT_STATX_SYNC_TYPE) != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 	// Make sure that only one sync type option is set.
 	syncType := uint32(flags & linux.AT_STATX_SYNC_TYPE)
 	if syncType != 0 && !bits.IsPowerOfTwo32(syncType) {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 	if mask&linux.STATX__RESERVED != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	opts := vfs.StatOptions{
@@ -272,7 +272,7 @@ func accessAt(t *kernel.Task, dirfd int32, pathAddr hostarch.Addr, mode uint) er
 
 	// Sanity check the mode.
 	if mode&^(rOK|wOK|xOK) != 0 {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 
 	path, err := copyInPath(t, pathAddr)
@@ -315,7 +315,7 @@ func Readlinkat(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sy
 
 func readlinkat(t *kernel.Task, dirfd int32, pathAddr, bufAddr hostarch.Addr, size uint) (uintptr, *kernel.SyscallControl, error) {
 	if int(size) <= 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	path, err := copyInPath(t, pathAddr)
diff --git a/pkg/sentry/syscalls/linux/vfs2/sync.go b/pkg/sentry/syscalls/linux/vfs2/sync.go
index 1f8a5878c..9344a81ce 100644
--- a/pkg/sentry/syscalls/linux/vfs2/sync.go
+++ b/pkg/sentry/syscalls/linux/vfs2/sync.go
@@ -16,6 +16,7 @@ package vfs2
 
 import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
 	"gvisor.dev/gvisor/pkg/sentry/kernel"
 	"gvisor.dev/gvisor/pkg/syserror"
@@ -71,10 +72,10 @@ func SyncFileRange(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel
 
 	// Check for negative values and overflow.
 	if offset < 0 || offset+nbytes < 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 	if flags&^(linux.SYNC_FILE_RANGE_WAIT_BEFORE|linux.SYNC_FILE_RANGE_WRITE|linux.SYNC_FILE_RANGE_WAIT_AFTER) != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	file := t.GetFileVFS2(fd)
diff --git a/pkg/sentry/syscalls/linux/vfs2/timerfd.go b/pkg/sentry/syscalls/linux/vfs2/timerfd.go
index 250870c03..0794330c6 100644
--- a/pkg/sentry/syscalls/linux/vfs2/timerfd.go
+++ b/pkg/sentry/syscalls/linux/vfs2/timerfd.go
@@ -16,6 +16,7 @@ package vfs2
 
 import (
 	"gvisor.dev/gvisor/pkg/abi/linux"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
 	"gvisor.dev/gvisor/pkg/sentry/fsimpl/timerfd"
 	"gvisor.dev/gvisor/pkg/sentry/kernel"
@@ -29,7 +30,7 @@ func TimerfdCreate(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel
 	flags := args[1].Int()
 
 	if flags&^(linux.TFD_CLOEXEC|linux.TFD_NONBLOCK) != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	// Timerfds aren't writable per se (their implementation of Write just
@@ -47,7 +48,7 @@ func TimerfdCreate(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel
 	case linux.CLOCK_MONOTONIC, linux.CLOCK_BOOTTIME:
 		clock = t.Kernel().MonotonicClock()
 	default:
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 	vfsObj := t.Kernel().VFS()
 	file, err := timerfd.New(t, vfsObj, clock, fileFlags)
@@ -72,7 +73,7 @@ func TimerfdSettime(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kerne
 	oldValAddr := args[3].Pointer()
 
 	if flags&^(linux.TFD_TIMER_ABSTIME) != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	file := t.GetFileVFS2(fd)
@@ -83,7 +84,7 @@ func TimerfdSettime(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kerne
 
 	tfd, ok := file.Impl().(*timerfd.TimerFileDescription)
 	if !ok {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	var newVal linux.Itimerspec
@@ -117,7 +118,7 @@ func TimerfdGettime(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kerne
 
 	tfd, ok := file.Impl().(*timerfd.TimerFileDescription)
 	if !ok {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	tm, s := tfd.GetTime()
diff --git a/pkg/sentry/syscalls/linux/vfs2/xattr.go b/pkg/sentry/syscalls/linux/vfs2/xattr.go
index c779c6465..33209a8d0 100644
--- a/pkg/sentry/syscalls/linux/vfs2/xattr.go
+++ b/pkg/sentry/syscalls/linux/vfs2/xattr.go
@@ -179,7 +179,7 @@ func setxattr(t *kernel.Task, args arch.SyscallArguments, shouldFollowFinalSymli
 	flags := args[4].Int()
 
 	if flags&^(linux.XATTR_CREATE|linux.XATTR_REPLACE) != 0 {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 
 	path, err := copyInPath(t, pathAddr)
@@ -217,7 +217,7 @@ func Fsetxattr(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Sys
 	flags := args[4].Int()
 
 	if flags&^(linux.XATTR_CREATE|linux.XATTR_REPLACE) != 0 {
-		return 0, nil, syserror.EINVAL
+		return 0, nil, linuxerr.EINVAL
 	}
 
 	file := t.GetFileVFS2(fd)
@@ -322,7 +322,7 @@ func copyOutXattrNameList(t *kernel.Task, listAddr hostarch.Addr, size uint, nam
 	}
 	if buf.Len() > int(size) {
 		if size >= linux.XATTR_LIST_MAX {
-			return 0, syserror.E2BIG
+			return 0, linuxerr.E2BIG
 		}
 		return 0, syserror.ERANGE
 	}
@@ -331,7 +331,7 @@ func copyOutXattrNameList(t *kernel.Task, listAddr hostarch.Addr, size uint, nam
 
 func copyInXattrValue(t *kernel.Task, valueAddr hostarch.Addr, size uint) (string, error) {
 	if size > linux.XATTR_SIZE_MAX {
-		return "", syserror.E2BIG
+		return "", linuxerr.E2BIG
 	}
 	buf := make([]byte, size)
 	if _, err := t.CopyInBytes(valueAddr, buf); err != nil {
@@ -350,7 +350,7 @@ func copyOutXattrValue(t *kernel.Task, valueAddr hostarch.Addr, size uint, value
 	}
 	if len(value) > int(size) {
 		if size >= linux.XATTR_SIZE_MAX {
-			return 0, syserror.E2BIG
+			return 0, linuxerr.E2BIG
 		}
 		return 0, syserror.ERANGE
 	}
diff --git a/pkg/sentry/time/calibrated_clock.go b/pkg/sentry/time/calibrated_clock.go
index 39bf1e0de..eed74f6bd 100644
--- a/pkg/sentry/time/calibrated_clock.go
+++ b/pkg/sentry/time/calibrated_clock.go
@@ -19,10 +19,10 @@ package time
 import (
 	"time"
 
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/log"
 	"gvisor.dev/gvisor/pkg/metric"
 	"gvisor.dev/gvisor/pkg/sync"
-	"gvisor.dev/gvisor/pkg/syserror"
 )
 
 // CalibratedClock implements a clock that tracks a reference clock.
@@ -259,6 +259,6 @@ func (c *CalibratedClocks) GetTime(id ClockID) (int64, error) {
 	case Realtime:
 		return c.realtime.GetTime()
 	default:
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 }
diff --git a/pkg/sentry/vfs/anonfs.go b/pkg/sentry/vfs/anonfs.go
index f48817132..8b3612200 100644
--- a/pkg/sentry/vfs/anonfs.go
+++ b/pkg/sentry/vfs/anonfs.go
@@ -19,6 +19,7 @@ import (
 
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/fspath"
 	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/sentry/kernel/auth"
@@ -164,7 +165,7 @@ func (fs *anonFilesystem) ReadlinkAt(ctx context.Context, rp *ResolvingPath) (st
 	if !rp.Done() {
 		return "", syserror.ENOTDIR
 	}
-	return "", syserror.EINVAL
+	return "", linuxerr.EINVAL
 }
 
 // RenameAt implements FilesystemImpl.RenameAt.
diff --git a/pkg/sentry/vfs/file_description.go b/pkg/sentry/vfs/file_description.go
index 2bc33d424..63ee0aab3 100644
--- a/pkg/sentry/vfs/file_description.go
+++ b/pkg/sentry/vfs/file_description.go
@@ -274,7 +274,7 @@ func (fd *FileDescription) SetStatusFlags(ctx context.Context, creds *auth.Crede
 		}
 	}
 	if flags&linux.O_DIRECT != 0 && !fd.opts.AllowDirectIO {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	// TODO(gvisor.dev/issue/1035): FileDescriptionImpl.SetOAsync()?
 	const settableFlags = linux.O_APPEND | linux.O_ASYNC | linux.O_DIRECT | linux.O_NOATIME | linux.O_NONBLOCK
@@ -874,7 +874,7 @@ func (fd *FileDescription) ComputeLockRange(ctx context.Context, start uint64, l
 		}
 		off = int64(stat.Size)
 	default:
-		return lock.LockRange{}, syserror.EINVAL
+		return lock.LockRange{}, linuxerr.EINVAL
 	}
 
 	return lock.ComputeRange(int64(start), int64(length), off)
diff --git a/pkg/sentry/vfs/file_description_impl_util.go b/pkg/sentry/vfs/file_description_impl_util.go
index 2b6f47b4b..cffb46aab 100644
--- a/pkg/sentry/vfs/file_description_impl_util.go
+++ b/pkg/sentry/vfs/file_description_impl_util.go
@@ -20,6 +20,7 @@ import (
 
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
 	fslock "gvisor.dev/gvisor/pkg/sentry/fs/lock"
 	"gvisor.dev/gvisor/pkg/sentry/memmap"
@@ -88,25 +89,25 @@ func (FileDescriptionDefaultImpl) EventUnregister(e *waiter.Entry) {
 // PRead implements FileDescriptionImpl.PRead analogously to
 // file_operations::read == file_operations::read_iter == NULL in Linux.
 func (FileDescriptionDefaultImpl) PRead(ctx context.Context, dst usermem.IOSequence, offset int64, opts ReadOptions) (int64, error) {
-	return 0, syserror.EINVAL
+	return 0, linuxerr.EINVAL
 }
 
 // Read implements FileDescriptionImpl.Read analogously to
 // file_operations::read == file_operations::read_iter == NULL in Linux.
 func (FileDescriptionDefaultImpl) Read(ctx context.Context, dst usermem.IOSequence, opts ReadOptions) (int64, error) {
-	return 0, syserror.EINVAL
+	return 0, linuxerr.EINVAL
 }
 
 // PWrite implements FileDescriptionImpl.PWrite analogously to
 // file_operations::write == file_operations::write_iter == NULL in Linux.
 func (FileDescriptionDefaultImpl) PWrite(ctx context.Context, src usermem.IOSequence, offset int64, opts WriteOptions) (int64, error) {
-	return 0, syserror.EINVAL
+	return 0, linuxerr.EINVAL
 }
 
 // Write implements FileDescriptionImpl.Write analogously to
 // file_operations::write == file_operations::write_iter == NULL in Linux.
 func (FileDescriptionDefaultImpl) Write(ctx context.Context, src usermem.IOSequence, opts WriteOptions) (int64, error) {
-	return 0, syserror.EINVAL
+	return 0, linuxerr.EINVAL
 }
 
 // IterDirents implements FileDescriptionImpl.IterDirents analogously to
@@ -125,7 +126,7 @@ func (FileDescriptionDefaultImpl) Seek(ctx context.Context, offset int64, whence
 // Sync implements FileDescriptionImpl.Sync analogously to
 // file_operations::fsync == NULL in Linux.
 func (FileDescriptionDefaultImpl) Sync(ctx context.Context) error {
-	return syserror.EINVAL
+	return linuxerr.EINVAL
 }
 
 // ConfigureMMap implements FileDescriptionImpl.ConfigureMMap analogously to
@@ -333,10 +334,10 @@ func (fd *DynamicBytesFileDescriptionImpl) Seek(ctx context.Context, offset int6
 		offset += fd.off
 	default:
 		// fs/seq_file:seq_lseek() rejects SEEK_END etc.
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 	if offset < 0 {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 	if offset != fd.lastRead {
 		// Regenerate the file's contents immediately. Compare
diff --git a/pkg/sentry/vfs/inotify.go b/pkg/sentry/vfs/inotify.go
index 49d29e20b..a7655bbb5 100644
--- a/pkg/sentry/vfs/inotify.go
+++ b/pkg/sentry/vfs/inotify.go
@@ -21,6 +21,7 @@ import (
 
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/sentry/arch"
 	"gvisor.dev/gvisor/pkg/sentry/uniqueid"
@@ -98,7 +99,7 @@ func NewInotifyFD(ctx context.Context, vfsObj *VirtualFilesystem, flags uint32)
 	// O_CLOEXEC affects file descriptors, so it must be handled outside of vfs.
 	flags &^= linux.O_CLOEXEC
 	if flags&^linux.O_NONBLOCK != 0 {
-		return nil, syserror.EINVAL
+		return nil, linuxerr.EINVAL
 	}
 
 	id := uniqueid.GlobalFromContext(ctx)
@@ -200,7 +201,7 @@ func (*Inotify) Write(ctx context.Context, src usermem.IOSequence, opts WriteOpt
 // Read implements FileDescriptionImpl.Read.
 func (i *Inotify) Read(ctx context.Context, dst usermem.IOSequence, opts ReadOptions) (int64, error) {
 	if dst.NumBytes() < inotifyEventBaseSize {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 
 	i.evMu.Lock()
@@ -226,7 +227,7 @@ func (i *Inotify) Read(ctx context.Context, dst usermem.IOSequence, opts ReadOpt
 				// write some events out.
 				return writeLen, nil
 			}
-			return 0, syserror.EINVAL
+			return 0, linuxerr.EINVAL
 		}
 
 		// Linux always dequeues an available event as long as there's enough
@@ -360,7 +361,7 @@ func (i *Inotify) RmWatch(ctx context.Context, wd int32) error {
 	w, ok := i.watches[wd]
 	if !ok {
 		i.mu.Unlock()
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 
 	// Remove the watch from this instance.
diff --git a/pkg/sentry/vfs/mount.go b/pkg/sentry/vfs/mount.go
index f93da3af1..03857dfc8 100644
--- a/pkg/sentry/vfs/mount.go
+++ b/pkg/sentry/vfs/mount.go
@@ -24,6 +24,7 @@ import (
 
 	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/refsvfs2"
 	"gvisor.dev/gvisor/pkg/sentry/kernel/auth"
 	"gvisor.dev/gvisor/pkg/syserror"
@@ -284,7 +285,7 @@ func (vfs *VirtualFilesystem) MountAt(ctx context.Context, creds *auth.Credentia
 // UmountAt removes the Mount at the given path.
 func (vfs *VirtualFilesystem) UmountAt(ctx context.Context, creds *auth.Credentials, pop *PathOperation, opts *UmountOptions) error {
 	if opts.Flags&^(linux.MNT_FORCE|linux.MNT_DETACH) != 0 {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 
 	// MNT_FORCE is currently unimplemented except for the permission check.
@@ -301,19 +302,19 @@ func (vfs *VirtualFilesystem) UmountAt(ctx context.Context, creds *auth.Credenti
 	}
 	defer vd.DecRef(ctx)
 	if vd.dentry != vd.mount.root {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	vfs.mountMu.Lock()
 	if mntns := MountNamespaceFromContext(ctx); mntns != nil {
 		defer mntns.DecRef(ctx)
 		if mntns != vd.mount.ns {
 			vfs.mountMu.Unlock()
-			return syserror.EINVAL
+			return linuxerr.EINVAL
 		}
 
 		if vd.mount == vd.mount.ns.root {
 			vfs.mountMu.Unlock()
-			return syserror.EINVAL
+			return linuxerr.EINVAL
 		}
 	}
 
diff --git a/pkg/sentry/vfs/vfs.go b/pkg/sentry/vfs/vfs.go
index b96de247f..f2aabb905 100644
--- a/pkg/sentry/vfs/vfs.go
+++ b/pkg/sentry/vfs/vfs.go
@@ -286,7 +286,7 @@ func (vfs *VirtualFilesystem) LinkAt(ctx context.Context, creds *auth.Credential
 	if newpop.FollowFinalSymlink {
 		oldVD.DecRef(ctx)
 		ctx.Warningf("VirtualFilesystem.LinkAt: file creation paths can't follow final symlink")
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 
 	rp := vfs.getResolvingPath(creds, newpop)
@@ -322,7 +322,7 @@ func (vfs *VirtualFilesystem) MkdirAt(ctx context.Context, creds *auth.Credentia
 	}
 	if pop.FollowFinalSymlink {
 		ctx.Warningf("VirtualFilesystem.MkdirAt: file creation paths can't follow final symlink")
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 	// "Under Linux, apart from the permission bits, the S_ISVTX mode bit is
 	// also honored." - mkdir(2)
@@ -360,7 +360,7 @@ func (vfs *VirtualFilesystem) MknodAt(ctx context.Context, creds *auth.Credentia
 	}
 	if pop.FollowFinalSymlink {
 		ctx.Warningf("VirtualFilesystem.MknodAt: file creation paths can't follow final symlink")
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 
 	rp := vfs.getResolvingPath(creds, pop)
@@ -403,13 +403,13 @@ func (vfs *VirtualFilesystem) OpenAt(ctx context.Context, creds *auth.Credential
 	// filesystem implementations that do not support it).
 	if opts.Flags&linux.O_TMPFILE != 0 {
 		if opts.Flags&linux.O_DIRECTORY == 0 {
-			return nil, syserror.EINVAL
+			return nil, linuxerr.EINVAL
 		}
 		if opts.Flags&linux.O_CREAT != 0 {
-			return nil, syserror.EINVAL
+			return nil, linuxerr.EINVAL
 		}
 		if opts.Flags&linux.O_ACCMODE == linux.O_RDONLY {
-			return nil, syserror.EINVAL
+			return nil, linuxerr.EINVAL
 		}
 	}
 	// O_PATH causes most other flags to be ignored.
@@ -500,7 +500,7 @@ func (vfs *VirtualFilesystem) RenameAt(ctx context.Context, creds *auth.Credenti
 	}
 	if oldpop.FollowFinalSymlink {
 		ctx.Warningf("VirtualFilesystem.RenameAt: source path can't follow final symlink")
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 
 	oldParentVD, oldName, err := vfs.getParentDirAndName(ctx, creds, oldpop)
@@ -522,7 +522,7 @@ func (vfs *VirtualFilesystem) RenameAt(ctx context.Context, creds *auth.Credenti
 	if newpop.FollowFinalSymlink {
 		oldParentVD.DecRef(ctx)
 		ctx.Warningf("VirtualFilesystem.RenameAt: destination path can't follow final symlink")
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 
 	rp := vfs.getResolvingPath(creds, newpop)
@@ -562,7 +562,7 @@ func (vfs *VirtualFilesystem) RmdirAt(ctx context.Context, creds *auth.Credentia
 	}
 	if pop.FollowFinalSymlink {
 		ctx.Warningf("VirtualFilesystem.RmdirAt: file deletion paths can't follow final symlink")
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 
 	rp := vfs.getResolvingPath(creds, pop)
@@ -645,7 +645,7 @@ func (vfs *VirtualFilesystem) SymlinkAt(ctx context.Context, creds *auth.Credent
 	}
 	if pop.FollowFinalSymlink {
 		ctx.Warningf("VirtualFilesystem.SymlinkAt: file creation paths can't follow final symlink")
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 
 	rp := vfs.getResolvingPath(creds, pop)
@@ -679,7 +679,7 @@ func (vfs *VirtualFilesystem) UnlinkAt(ctx context.Context, creds *auth.Credenti
 	}
 	if pop.FollowFinalSymlink {
 		ctx.Warningf("VirtualFilesystem.UnlinkAt: file deletion paths can't follow final symlink")
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 
 	rp := vfs.getResolvingPath(creds, pop)
diff --git a/pkg/syserr/syserr.go b/pkg/syserr/syserr.go
index 7d0a5125b..765128ff0 100644
--- a/pkg/syserr/syserr.go
+++ b/pkg/syserr/syserr.go
@@ -22,6 +22,7 @@ import (
 
 	"golang.org/x/sys/unix"
 	"gvisor.dev/gvisor/pkg/abi/linux/errno"
+	"gvisor.dev/gvisor/pkg/errors"
 	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/syserror"
 )
@@ -281,6 +282,11 @@ func FromError(err error) *Error {
 	if errno, ok := err.(unix.Errno); ok {
 		return FromHost(errno)
 	}
+
+	if linuxErr, ok := err.(*errors.Error); ok {
+		return FromHost(unix.Errno(linuxErr.Errno()))
+	}
+
 	if errno, ok := syserror.TranslateError(err); ok {
 		return FromHost(errno)
 	}
diff --git a/pkg/syserror/syserror.go b/pkg/syserror/syserror.go
index 56b621357..721f62903 100644
--- a/pkg/syserror/syserror.go
+++ b/pkg/syserror/syserror.go
@@ -26,9 +26,7 @@ import (
 
 // The following variables have the same meaning as their syscall equivalent.
 var (
-	E2BIG        = error(unix.E2BIG)
 	EACCES       = error(unix.EACCES)
-	EADDRINUSE   = error(unix.EADDRINUSE)
 	EAGAIN       = error(unix.EAGAIN)
 	EBADF        = error(unix.EBADF)
 	EBADFD       = error(unix.EBADFD)
@@ -43,7 +41,6 @@ var (
 	EFBIG        = error(unix.EFBIG)
 	EIDRM        = error(unix.EIDRM)
 	EINTR        = error(unix.EINTR)
-	EINVAL       = error(unix.EINVAL)
 	EIO          = error(unix.EIO)
 	EISDIR       = error(unix.EISDIR)
 	ELIBBAD      = error(unix.ELIBBAD)
diff --git a/pkg/tcpip/link/tun/device.go b/pkg/tcpip/link/tun/device.go
index 36af2a029..575a084e2 100644
--- a/pkg/tcpip/link/tun/device.go
+++ b/pkg/tcpip/link/tun/device.go
@@ -18,6 +18,7 @@ import (
 	"fmt"
 
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/sync"
 	"gvisor.dev/gvisor/pkg/syserror"
 	"gvisor.dev/gvisor/pkg/tcpip"
@@ -88,12 +89,12 @@ func (d *Device) SetIff(s *stack.Stack, name string, flags Flags) error {
 	defer d.mu.Unlock()
 
 	if d.endpoint != nil {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 
 	// Input validation.
 	if flags.TAP && flags.TUN || !flags.TAP && !flags.TUN {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 
 	prefix := "tun"
@@ -108,7 +109,7 @@ func (d *Device) SetIff(s *stack.Stack, name string, flags Flags) error {
 
 	endpoint, err := attachOrCreateNIC(s, name, prefix, linkCaps)
 	if err != nil {
-		return syserror.EINVAL
+		return linuxerr.EINVAL
 	}
 
 	d.endpoint = endpoint
@@ -159,7 +160,7 @@ func attachOrCreateNIC(s *stack.Stack, name, prefix string, linkCaps stack.LinkE
 			// Race detected: A NIC has been created in between.
 			continue
 		default:
-			return nil, syserror.EINVAL
+			return nil, linuxerr.EINVAL
 		}
 	}
 }
diff --git a/pkg/usermem/bytes_io.go b/pkg/usermem/bytes_io.go
index 3da3c0294..4c97b9136 100644
--- a/pkg/usermem/bytes_io.go
+++ b/pkg/usermem/bytes_io.go
@@ -16,6 +16,7 @@ package usermem
 
 import (
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/safemem"
 	"gvisor.dev/gvisor/pkg/syserror"
@@ -51,7 +52,7 @@ func (b *BytesIO) CopyIn(ctx context.Context, addr hostarch.Addr, dst []byte, op
 // ZeroOut implements IO.ZeroOut.
 func (b *BytesIO) ZeroOut(ctx context.Context, addr hostarch.Addr, toZero int64, opts IOOpts) (int64, error) {
 	if toZero > int64(maxInt) {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 	rngN, rngErr := b.rangeCheck(addr, int(toZero))
 	if rngN == 0 {
@@ -89,7 +90,7 @@ func (b *BytesIO) rangeCheck(addr hostarch.Addr, length int) (int, error) {
 		return 0, nil
 	}
 	if length < 0 {
-		return 0, syserror.EINVAL
+		return 0, linuxerr.EINVAL
 	}
 	max := hostarch.Addr(len(b.Bytes))
 	if addr >= max {
diff --git a/pkg/usermem/usermem.go b/pkg/usermem/usermem.go
index 0d6d25e50..483e64c1e 100644
--- a/pkg/usermem/usermem.go
+++ b/pkg/usermem/usermem.go
@@ -22,11 +22,11 @@ import (
 	"strconv"
 
 	"gvisor.dev/gvisor/pkg/context"
+	"gvisor.dev/gvisor/pkg/errors/linuxerr"
 	"gvisor.dev/gvisor/pkg/gohacks"
+	"gvisor.dev/gvisor/pkg/hostarch"
 	"gvisor.dev/gvisor/pkg/safemem"
 	"gvisor.dev/gvisor/pkg/syserror"
-
-	"gvisor.dev/gvisor/pkg/hostarch"
 )
 
 // IO provides access to the contents of a virtual memory space.
@@ -382,7 +382,7 @@ func CopyInt32StringsInVec(ctx context.Context, uio IO, ars hostarch.AddrRangeSe
 		// Parse a single value.
 		val, err := strconv.ParseInt(string(buf[i:nextI]), 10, 32)
 		if err != nil {
-			return int64(i), syserror.EINVAL
+			return int64(i), linuxerr.EINVAL
 		}
 		dsts[j] = int32(val)
 
@@ -398,7 +398,7 @@ func CopyInt32StringsInVec(ctx context.Context, uio IO, ars hostarch.AddrRangeSe
 		return int64(i), cperr
 	}
 	if j == 0 {
-		return int64(i), syserror.EINVAL
+		return int64(i), linuxerr.EINVAL
 	}
 	return int64(i), nil
 }