From 6f60a2b0a27a742690aa6acd5df1912ccb5fc8d3 Mon Sep 17 00:00:00 2001
From: Etienne Perot <eperot@google.com>
Date: Thu, 3 Dec 2020 06:18:03 -0800
Subject: Implement `fcntl` options `F_GETSIG` and `F_SETSIG`.

These options allow overriding the signal that gets sent to the process when
I/O operations are available on the file descriptor, rather than the default
`SIGIO` signal. Doing so also populates `siginfo` to contain extra information
about which file descriptor caused the event (`si_fd`) and what events happened
on it (`si_band`). The logic around which FD is populated within `si_fd`
matches Linux's, which means it has some weird edge cases where that value may
not actually refer to a file descriptor that is still valid.

This CL also ports extra S/R logic regarding async handler in VFS2.
Without this, async I/O handlers aren't properly re-registered after S/R.

PiperOrigin-RevId: 345436598
---
 pkg/sentry/vfs/epoll.go            | 10 +++++-----
 pkg/sentry/vfs/file_description.go |  9 +++++++--
 pkg/sentry/vfs/save_restore.go     | 19 ++++++++++++++++++-
 3 files changed, 30 insertions(+), 8 deletions(-)

(limited to 'pkg/sentry/vfs')

diff --git a/pkg/sentry/vfs/epoll.go b/pkg/sentry/vfs/epoll.go
index a98aac52b..072655fe8 100644
--- a/pkg/sentry/vfs/epoll.go
+++ b/pkg/sentry/vfs/epoll.go
@@ -204,8 +204,8 @@ func (ep *EpollInstance) AddInterest(file *FileDescription, num int32, event lin
 	file.EventRegister(&epi.waiter, wmask)
 
 	// Check if the file is already ready.
-	if file.Readiness(wmask)&wmask != 0 {
-		epi.Callback(nil)
+	if m := file.Readiness(wmask) & wmask; m != 0 {
+		epi.Callback(nil, m)
 	}
 
 	// Add epi to file.epolls so that it is removed when the last
@@ -274,8 +274,8 @@ func (ep *EpollInstance) ModifyInterest(file *FileDescription, num int32, event
 	file.EventRegister(&epi.waiter, wmask)
 
 	// Check if the file is already ready with the new mask.
-	if file.Readiness(wmask)&wmask != 0 {
-		epi.Callback(nil)
+	if m := file.Readiness(wmask) & wmask; m != 0 {
+		epi.Callback(nil, m)
 	}
 
 	return nil
@@ -311,7 +311,7 @@ func (ep *EpollInstance) DeleteInterest(file *FileDescription, num int32) error
 }
 
 // Callback implements waiter.EntryCallback.Callback.
-func (epi *epollInterest) Callback(*waiter.Entry) {
+func (epi *epollInterest) Callback(*waiter.Entry, waiter.EventMask) {
 	newReady := false
 	epi.epoll.mu.Lock()
 	if !epi.ready {
diff --git a/pkg/sentry/vfs/file_description.go b/pkg/sentry/vfs/file_description.go
index 51fddbaff..2153382e9 100644
--- a/pkg/sentry/vfs/file_description.go
+++ b/pkg/sentry/vfs/file_description.go
@@ -43,7 +43,7 @@ import (
 type FileDescription struct {
 	FileDescriptionRefs
 
-	// flagsMu protects statusFlags and asyncHandler below.
+	// flagsMu protects `statusFlags`, `saved`, and `asyncHandler` below.
 	flagsMu sync.Mutex `state:"nosave"`
 
 	// statusFlags contains status flags, "initialized by open(2) and possibly
@@ -52,6 +52,11 @@ type FileDescription struct {
 	// access to asyncHandler.
 	statusFlags uint32
 
+	// saved is true after beforeSave is called. This is used to prevent
+	// double-unregistration of asyncHandler. This does not work properly for
+	// save-resume, which is not currently supported in gVisor (see b/26588733).
+	saved bool `state:"nosave"`
+
 	// asyncHandler handles O_ASYNC signal generation. It is set with the
 	// F_SETOWN or F_SETOWN_EX fcntls. For asyncHandler to be used, O_ASYNC must
 	// also be set by fcntl(2).
@@ -184,7 +189,7 @@ func (fd *FileDescription) DecRef(ctx context.Context) {
 		}
 		fd.vd.DecRef(ctx)
 		fd.flagsMu.Lock()
-		if fd.statusFlags&linux.O_ASYNC != 0 && fd.asyncHandler != nil {
+		if !fd.saved && fd.statusFlags&linux.O_ASYNC != 0 && fd.asyncHandler != nil {
 			fd.asyncHandler.Unregister(fd)
 		}
 		fd.asyncHandler = nil
diff --git a/pkg/sentry/vfs/save_restore.go b/pkg/sentry/vfs/save_restore.go
index 7723ed643..8f070ed53 100644
--- a/pkg/sentry/vfs/save_restore.go
+++ b/pkg/sentry/vfs/save_restore.go
@@ -18,8 +18,10 @@ import (
 	"fmt"
 	"sync/atomic"
 
+	"gvisor.dev/gvisor/pkg/abi/linux"
 	"gvisor.dev/gvisor/pkg/context"
 	"gvisor.dev/gvisor/pkg/refsvfs2"
+	"gvisor.dev/gvisor/pkg/waiter"
 )
 
 // FilesystemImplSaveRestoreExtension is an optional extension to
@@ -120,5 +122,20 @@ func (mnt *Mount) afterLoad() {
 func (epi *epollInterest) afterLoad() {
 	// Mark all epollInterests as ready after restore so that the next call to
 	// EpollInstance.ReadEvents() rechecks their readiness.
-	epi.Callback(nil)
+	epi.Callback(nil, waiter.EventMaskFromLinux(epi.mask))
+}
+
+// beforeSave is called by stateify.
+func (fd *FileDescription) beforeSave() {
+	fd.saved = true
+	if fd.statusFlags&linux.O_ASYNC != 0 && fd.asyncHandler != nil {
+		fd.asyncHandler.Unregister(fd)
+	}
+}
+
+// afterLoad is called by stateify.
+func (fd *FileDescription) afterLoad() {
+	if fd.statusFlags&linux.O_ASYNC != 0 && fd.asyncHandler != nil {
+		fd.asyncHandler.Register(fd)
+	}
 }
-- 
cgit v1.2.3