summaryrefslogtreecommitdiffhomepage
path: root/pkg/sentry/fsimpl/proc
AgeCommit message (Collapse)Author
2020-10-23Rewrite reference leak checker without finalizers.Dean Deng
Our current reference leak checker uses finalizers to verify whether an object has reached zero references before it is garbage collected. There are multiple problems with this mechanism, so a rewrite is in order. With finalizers, there is no way to guarantee that a finalizer will run before the program exits. When an unreachable object with a finalizer is garbage collected, its finalizer will be added to a queue and run asynchronously. The best we can do is run garbage collection upon sandbox exit to make sure that all finalizers are enqueued. Furthermore, if there is a chain of finalized objects, e.g. A points to B points to C, garbage collection needs to run multiple times before all of the finalizers are enqueued. The first GC run will register the finalizer for A but not free it. It takes another GC run to free A, at which point B's finalizer can be registered. As a result, we need to run GC as many times as the length of the longest such chain to have a somewhat reliable leak checker. Finally, a cyclical chain of structs pointing to one another will never be garbage collected if a finalizer is set. This is a well-known issue with Go finalizers (https://github.com/golang/go/issues/7358). Using leak checking on filesystem objects that produce cycles will not work and even result in memory leaks. The new leak checker stores reference counted objects in a global map when leak check is enabled and removes them once they are destroyed. At sandbox exit, any remaining objects in the map are considered as leaked. This provides a deterministic way of detecting leaks without relying on the complexities of finalizers and garbage collection. This approach has several benefits over the former, including: - Always detects leaks of objects that should be destroyed very close to sandbox exit. The old checker very rarely detected these leaks, because it relied on garbage collection to be run in a short window of time. - Panics if we forgot to enable leak check on a ref-counted object (we will try to remove it from the map when it is destroyed, but it will never have been added). - Can store extra logging information in the map values without adding to the size of the ref count struct itself. With the size of just an int64, the ref count object remains compact, meaning frequent operations like IncRef/DecRef are more cache-efficient. - Can aggregate leak results in a single report after the sandbox exits. Instead of having warnings littered in the log, which were non-deterministically triggered by garbage collection, we can print all warning messages at once. Note that this could also be a limitation--the sandbox must exit properly for leaks to be detected. Some basic benchmarking indicates that this change does not significantly affect performance when leak checking is enabled, which is understandable since registering/unregistering is only done once for each filesystem object. Updates #1486. PiperOrigin-RevId: 338685972
2020-10-13[vfs2] Don't take reference in Task.MountNamespaceVFS2 and MountNamespace.Root.Dean Deng
This fixes reference leaks related to accidentally forgetting to DecRef() after calling one or the other. PiperOrigin-RevId: 336918922
2020-10-13[vfs2] Add FilesystemType.Release to avoid reference leaks.Dean Deng
Singleton filesystem like devpts and devtmpfs have a single filesystem shared among all mounts, so they acquire a "self-reference" when initialized that must be released when the entire virtual filesystem is released at sandbox exit. PiperOrigin-RevId: 336828852
2020-10-12[vfs] kernfs: Fix inode memory leak issue.Ayush Ranjan
This change aims to fix the memory leak issue reported in #3933. Background: VFS2 kernfs kept accumulating invalid dentries if those dentries were not walked on. After substantial consideration of the problem by our team, we decided to have an LRU cache solution. This change is the first part to that solution, where we don't cache anything. The LRU cache can be added on top of this. What has changed: - Introduced the concept of an inode tree in kernfs.OrderedChildren. This is helpful is cases where the lifecycle of an inode is different from that of a dentry. - OrderedChildren now deals with initialized inodes instead of initialized dentries. It now implements Lookup() where it constructs a new dentry using the inode. - OrderedChildren holds a ref on all its children inodes. With this change, now an inode can "outlive" a dentry pointing to it. See comments in kernfs.OrderedChildren. - The kernfs dentry tree is solely maintained by kernfs only. Inode implementations can not modify the dentry tree. - Dentries that reach ref count 0 are removed from the dentry tree. - revalidateChildLocked now defer-DecRefs the newly created dentry from Inode.Lookup(), limiting its life to the current filesystem operation. If refs are picked on the dentry during the FS op (via an FD or something), then it will stick around and will be removed when the FD is closed. So there is essentially _no caching_ for Look()ed up dentries. - kernfs.DecRef does not have the precondition that fs.mu must be locked. Fixes #3933 PiperOrigin-RevId: 336768576
2020-09-29Add /proc/[pid]/cwdFabricio Voznika
PiperOrigin-RevId: 334478850
2020-09-24Add basic stateify annotations.Adin Scannell
Updates #1663 PiperOrigin-RevId: 333539293
2020-09-22[vfs] [1/2] kernfs: Internally use kernfs.Dentry instead of vfs.Dentry.Ayush Ranjan
Update signatures for: - walkExistingLocked - checkDeleteLocked - Inode.Open Updates #1193 PiperOrigin-RevId: 333163381
2020-09-21Use kernfs.Dentry for kernfs.Lookup.Dean Deng
Updates #1193. PiperOrigin-RevId: 332939026
2020-09-20Merge pull request #3651 from ianlewis:ip-forwardinggVisor bot
PiperOrigin-RevId: 332760843
2020-09-17fsimpl: improve the "implements" commentsTiwei Bie
As noticed by @ayushr2, the "implements" comments are not consistent, e.g. // IterDirents implements kernfs.inodeDynamicLookup. // Generate implements vfs.DynamicBytesSource.Generate. This patch improves this by making the comments like this consistently include the package name (when the interface and struct are not in the same package) and method name. Signed-off-by: Tiwei Bie <tiwei.btw@antgroup.com>
2020-09-16Implement FUSE_READLINKBoyuan He
Fixes #3316
2020-09-15Merge pull request #3895 from btw616:fix/issue-3894gVisor bot
PiperOrigin-RevId: 331824411
2020-09-15Fix proc.(*fdDir).IterDirents for VFS2Tiwei Bie
Currently the returned offset is an index, and we can't use it to find the next fd to serialize, because getdents should iterate correctly despite mutation of fds. Instead, we can return the next fd to serialize plus 2 (which accounts for "." and "..") as the offset. Fixes: #3894 Signed-off-by: Tiwei Bie <tiwei.btw@antgroup.com>
2020-09-14Correct FDSize in /proc/[pid]/status.Jamie Liu
In Linux, FDSize is fs/proc/array.c:task_state() => struct fdtable::max_fds, which is set to the underlying array's length in fs/file.c:alloc_fdtable(). Follow-up changes: - Remove FDTable.GetRefs() and FDTable.GetRefsVFS2(), which are unused. - Reset FDTable.used to 0 during restore, since the subsequent calls to FDTable.setAll() increment it again, causing its value to be doubled. (After this CL, FDTable.used is only used to avoid reallocation in FDTable.GetFDs(), so this fix is not very visible.) PiperOrigin-RevId: 331588190
2020-09-08Honor readonly flag for root mountFabricio Voznika
Updates #1487 PiperOrigin-RevId: 330580699
2020-08-28Implement StatFS for various VFS2 filesystems.Rahat Mahmood
This mainly involved enabling kernfs' client filesystems to provide a StatFS implementation. Fixes #3411, #3515. PiperOrigin-RevId: 329009864
2020-08-25Use new reference count utility throughout gvisor.Dean Deng
This uses the refs_vfs2 template in vfs2 as well as objects common to vfs1 and vfs2. Note that vfs1-only refcounts are not replaced, since vfs1 will be deleted soon anyway. The following structs now use the new tool, with leak check enabled: devpts:rootInode fuse:inode kernfs:Dentry kernfs:dir kernfs:readonlyDir kernfs:StaticDirectory proc:fdDirInode proc:fdInfoDirInode proc:subtasksInode proc:taskInode proc:tasksInode vfs:FileDescription vfs:MountNamespace vfs:Filesystem sys:dir kernel:FSContext kernel:ProcessGroup kernel:Session shm:Shm mm:aioMappable mm:SpecialMappable transport:queue And the following use the template, but because they currently are not leak checked, a TODO is left instead of enabling leak check in this patch: kernel:FDTable tun:tunEndpoint Updates #1486. PiperOrigin-RevId: 328460377
2020-08-21Clarify seek behaviour for kernfs.GenericDirectoryFD.Rahat Mahmood
- Remove comment about GenericDirectoryFD not being compatible with dynamic directories. It is currently being used to implement dynamic directories. - Try to handle SEEK_END better than setting the offset to infinity. SEEK_END is poorly defined for dynamic directories anyways, so at least try make it work correctly for the static entries. Updates #1193. PiperOrigin-RevId: 327890128
2020-08-21Make mounts ReadWrite first, then later change to ReadOnly.Nicolas Lacasse
This lets us create "synthetic" mountpoint directories in ReadOnly mounts during VFS setup. Also add context.WithMountNamespace, as some filesystems (like overlay) require a MountNamespace on ctx to handle vfs.Filesystem Operations. PiperOrigin-RevId: 327874971
2020-08-18[vfs2] Implement /proc/sys/net/ipv4/tcp_rmem and /proc/sys/net/ipv4/tcp_wmem.Ayush Ranjan
Updates #1035 PiperOrigin-RevId: 327253907
2020-08-17Merge branch 'master' into ip-forwardingIan Lewis
- Merges aleksej-paschenko's with HEAD - Adds vfs2 support for ip_forward
2020-08-05Add loss recovery option for TCP.Nayana Bidari
/proc/sys/net/ipv4/tcp_recovery is used to enable RACK loss recovery in TCP. PiperOrigin-RevId: 325157807
2020-08-03Plumbing context.Context to DecRef() and Release().Nayana Bidari
context is passed to DecRef() and Release() which is needed for SO_LINGER implementation. PiperOrigin-RevId: 324672584
2020-07-23FileDescription is hard to spell.Dean Deng
Fix typos. PiperOrigin-RevId: 322913282
2020-07-14Include context in kernfs.Inode.Stat methodCraig Chi
To implement stat(2) in FUSE, we have to embed credentials and pid in request header. The information should be extracted from the context passed to VFS layer. Therefore `Stat()` signature in `kernfs.Inode` interface should include context as first argument. Some other fs implementations need to be modified as well, such as devpts, host, pipefs, and proc. Fixes #3235
2020-06-30Fix index calculation for /proc/[pid]/cmdline.Dean Deng
We were truncating buf using a index relative to the middle of the slice (i.e. where envv begins), but we need to calculate the index relative to the entire slice. Updates #2923. PiperOrigin-RevId: 319154950
2020-06-24Fix procfs bugs in vfs2.Dean Deng
- Support writing on proc/[pid]/{uid,gid}map - Return EIO for writing to static files. Updates #2923. PiperOrigin-RevId: 318188503
2020-06-19Fix vfs2 proc/self/fd dirent iteration.Dean Deng
Make proc/self/fd iteration work properly. Also, the comment on kernfs.Inode.IterDirents did not accurately reflect how parameters should be used/were used in kernfs.Inode impls other than fdDir. Updates #2923. PiperOrigin-RevId: 317370325
2020-06-17Implement POSIX locksFabricio Voznika
- Change FileDescriptionImpl Lock/UnlockPOSIX signature to take {start,length,whence}, so the correct offset can be calculated in the implementations. - Create PosixLocker interface to make it possible to share the same locking code from different implementations. Closes #1480 PiperOrigin-RevId: 316910286
2020-06-09Implement flock(2) in VFS2Fabricio Voznika
LockFD is the generic implementation that can be embedded in FileDescriptionImpl implementations. Unique lock ID is maintained in vfs.FileDescription and is created on demand. Updates #1480 PiperOrigin-RevId: 315604825
2020-05-07Allocate device numbers for VFS2 filesystems.Jamie Liu
Updates #1197, #1198, #1672 PiperOrigin-RevId: 310432006
2020-05-06Add maximum memory limit.Nicolas Lacasse
PiperOrigin-RevId: 310179277
2020-04-30Fix proc net bugs in VFS2.Dean Deng
The /proc/net/udp header was missing, and /proc/sys/net was set up as /proc/sys/net/net. Discovered while trying to run networking tests for VFS2. PiperOrigin-RevId: 309243758
2020-04-24Plumb context.Context into kernfs.Inode.Open().Dean Deng
PiperOrigin-RevId: 308304793
2020-04-24Fixes for procfsFabricio Voznika
- Return ENOENT for /proc/[pid]/task if task is zoombied or terminated - Allow directory to be Seek() to the end - Construct synthetic files for /proc/[pid]/ns/* - Changed GenericDirectoryFD.Init to not register with FileDescription, otherwise other implementation cannot change behavior. Updates #1195,1193 PiperOrigin-RevId: 308294649
2020-04-21Misc VFS2 fixesFabricio Voznika
- Fix defer operation ordering in kernfs.Filesystem.AccessAt() - Add AT_NULL entry in proc/pid/auvx - Fix line padding in /proc/pid/maps - Fix linux_dirent serialization for getdents(2) - Remove file creation flags from vfs.FileDescription.statusFlags() Updates #1193, #1035 PiperOrigin-RevId: 307704159
2020-04-21Remove filesystem structure from vfs.Dentry.Jamie Liu
This change: - Drastically simplifies the synchronization model: filesystem structure is both implementation-defined and implementation-synchronized. - Allows implementations of vfs.DentryImpl to use implementation-specific dentry types, reducing casts during path traversal. - Doesn't require dentries representing non-directory files to waste space on a map of children. - Allows dentry revalidation and mount lookup to be correctly ordered (fixed FIXME in fsimpl/gofer/filesystem.go). - Removes the need to have two separate maps in gofer.dentry (dentry.vfsd.children and dentry.negativeChildren) for positive and negative lookups respectively. //pkg/sentry/fsimpl/tmpfs/benchmark_test.go: name old time/op new time/op delta VFS2TmpfsStat/1-112 172ns ± 4% 165ns ± 3% -4.08% (p=0.002 n=9+9) VFS2TmpfsStat/2-112 199ns ± 3% 195ns ±10% ~ (p=0.132 n=8+9) VFS2TmpfsStat/3-112 230ns ± 2% 216ns ± 2% -6.15% (p=0.000 n=8+8) VFS2TmpfsStat/8-112 390ns ± 2% 358ns ± 4% -8.33% (p=0.000 n=9+8) VFS2TmpfsStat/64-112 2.20µs ± 3% 2.01µs ± 3% -8.48% (p=0.000 n=10+8) VFS2TmpfsStat/100-112 3.42µs ± 9% 3.08µs ± 2% -9.82% (p=0.000 n=9+8) VFS2TmpfsMountStat/1-112 278ns ± 1% 286ns ±15% ~ (p=0.712 n=8+10) VFS2TmpfsMountStat/2-112 311ns ± 4% 298ns ± 2% -4.27% (p=0.000 n=9+8) VFS2TmpfsMountStat/3-112 339ns ± 3% 330ns ± 9% ~ (p=0.070 n=8+9) VFS2TmpfsMountStat/8-112 503ns ± 3% 466ns ± 3% -7.38% (p=0.000 n=8+8) VFS2TmpfsMountStat/64-112 2.53µs ±16% 2.17µs ± 7% -14.19% (p=0.000 n=10+9) VFS2TmpfsMountStat/100-112 3.60µs ± 4% 3.30µs ± 8% -8.33% (p=0.001 n=8+9) Updates #1035 PiperOrigin-RevId: 307655892
2020-04-09Merge pull request #2253 from amscanne:nogogVisor bot
PiperOrigin-RevId: 305807868
2020-04-08Do not hold FileDescription references in VFS2 procfs inodes.Jamie Liu
FileDescription references are side-effectual; for example, holding a reference on the write end of a pipe prevents reads from the read end from returning EOF. This change is consistent with Linux, but not VFS1; while VFS1 also has this bug, it's less visible there since VFS1 procfs disables caching. Updates #1195 PiperOrigin-RevId: 305545099
2020-04-08Enable SubprocessExited and SubprocessZombie for gVisorFabricio Voznika
Updates #164 PiperOrigin-RevId: 305544029
2020-04-08Fix unused result errors.Adin Scannell
This fixes a bug in the proc net directory. Updates #2243
2020-04-06Correctly implement magic symlinks in VFS2 procfs.Jamie Liu
Updates #1195 PiperOrigin-RevId: 305143567
2020-04-04Record VFS2 sockets in global socket map.Dean Deng
Updates #1476, #1478, #1484, #1485. PiperOrigin-RevId: 304845354
2020-03-31Fix /proc/self/mounts and /proc/self/mountinfo in VFS2.Nicolas Lacasse
Some extra fields were added to the Mount type to expose necessary data to the proc filesystem. PiperOrigin-RevId: 304053361
2020-03-27Add FilesystemType.Name method, and FilesystemType field to Filesystem struct.Nicolas Lacasse
Both have analogues in Linux: * struct file_system_type has a char *name field. * struct super_block keeps a pointer to the file_system_type. These fields are necessary to support the `filesystem type` field in /proc/[pid]/mountinfo. PiperOrigin-RevId: 303434063
2020-03-26Combine file mode and isDir argumentsFabricio Voznika
Updates #1035 PiperOrigin-RevId: 303021328
2020-03-16Enforce file size rlimits in VFS2Fabricio Voznika
Updates #1035 PiperOrigin-RevId: 301255357
2020-03-16Add calls to vfs.CheckSetStat to fsimplsFabricio Voznika
Only gofer filesystem was calling vfs.CheckSetStat for vfs.FilesystemImpl.SetStatAt and vfs.FileDescriptionImpl.SetStat. Updates #1193, #1672, #1197 PiperOrigin-RevId: 301226522
2020-03-14Disallow kernfs.Inode.SetStat for readonly inodesFabricio Voznika
Updates #1195, #1193 PiperOrigin-RevId: 300950993
2020-03-13Add remaining procfs filesFabricio Voznika
Closes #1195 PiperOrigin-RevId: 300867055
generated by cgit v1.2.3 (git 2.39.1) at 2024-12-16 10:56:52 +0000