| Age | Commit message (Collapse) | Author |
|---|
|
Before this change, we would silently drop packets when unable to determine a
route to the destination host. This change brings us into line with RFC 792
(IPv4) and RFC 4443 (IPv6), both of which specify that gateways should return
an ICMP error to the sender when unable to reach the destination.
Startblock:
has LGTM from asfez
and then
add reviewer ghanan
PiperOrigin-RevId: 372214051
|
|
PiperOrigin-RevId: 372020696
|
|
In order to resolve path names, fsSymlink.Readlink() may need to reenter
kernfs. Change the code so that kernfs.Inode.Readlink() is called without
locks and document the new contract.
PiperOrigin-RevId: 371770222
|
|
When a child is added to a parent (directory) dentry, both child and parent are
queued for checkCachingLocked(). Make sure that the parent is queued first
because the parent gained a ref and so could be removed from the LRU cache hence
making space for the new child. This could prevent an LRU cache eviction.
In practice, this did seem to help. ~800 RPCs were reduced while building
//absl/... (ABSL build benchmark). Evictions hurt in 2 ways - create renameMu
contention and destroy a possibly useful dentry which will have to be re-walked
and re-opened later.
Follow up fix for #5859.
PiperOrigin-RevId: 371509392
|
|
The root table physical page has to be mapped to not fault in iret or sysret
after switching into a user address space. sysret and iret are in the upper
half that is global and so page tables of lower levels are already mapped.
Fixes #5742
PiperOrigin-RevId: 371458644
|
|
If verification fails for translating mmapped memory, the content should
not be returned. This is not an issue for panic mode, but for error mode
we should return empty content along with the error.
PiperOrigin-RevId: 371393519
|
|
Originally we were making a WalkGetAttrOne RPC to confirm that a file does not
exist on the remote filesystem - when there was no cached information about the
existence of a dentry at that position.
This change avoids making that RPC and speculatively makes the
mkdir/mknod/linkat/symlink RPC. They will fail with EEXIST if a file exists at
that position as we want.
However the error ordering is important. Existence check comes before
writability check. So we make the existence check when the writability check
fails and give it precedence.
This change saves ~76,000 RPCs while building //absl/... (ABSL build benchmark).
That is 10% of all RPCs made while running that workload.
PiperOrigin-RevId: 371225633
|
|
PiperOrigin-RevId: 371216407
|
|
PiperOrigin-RevId: 371198372
|
|
PiperOrigin-RevId: 371163405
|
|
PiperOrigin-RevId: 371131985
|
|
PiperOrigin-RevId: 371015541
|
|
Co-Author: ayushranjan
PiperOrigin-RevId: 370785009
|
|
PiperOrigin-RevId: 370733869
|
|
The newly added Weirdness metric with fields should be used instead of them.
Simple query for weirdness metric: http://shortn/_DGNk0z2Up6
PiperOrigin-RevId: 370578132
|
|
PiperOrigin-RevId: 370181621
|
|
In https://github.com/google/gvisor/commit/f075522849fa a check to increase zero
to a minimum backlog length was removed from sys_socket.go to bring it in parity
with linux and then in tcp/endpoint.go we bump backlog by 1. But this broke
calling listen on a AF_UNIX socket w/ a zero backlog as in linux it does allow 1
connection even with a zero backlog.
This was caught by a php runtime test socket_abstract_path.phpt.
PiperOrigin-RevId: 369974744
|
|
Weirdness metric contains fields to track the number of clock fallback,
partial result and vsyscalls. This metric will avoid the overhead of
having three different metrics (fallbackMetric, partialResultMetric,
vsyscallCount).
PiperOrigin-RevId: 369970218
|
|
PiperOrigin-RevId: 369967629
|
|
With this change, GSO options no longer needs to be passed around as
a function argument in the write path.
This change is done in preparation for a later change that defers
segmentation, and may change GSO options for a packet as it flows
down the stack.
Updates #170.
PiperOrigin-RevId: 369774872
|
|
PiperOrigin-RevId: 369758655
|
|
PiperOrigin-RevId: 369724358
|
|
PiperOrigin-RevId: 369686285
|
|
If the host doesn't have TSC scaling feature, then scaling down TSC to
the lowest value will fail, and we will fall back to legacy logic
anyway, but we leave an ugly log message in host's kernel log.
kernel: user requested TSC rate below hardware speed
Instead, check for KVM_CAP_TSC_CONTROL when initializing KVM, and fall
back to legacy logic early if host's cpu doesn't support that.
Signed-off-by: Daniel Dao <dqminh89@gmail.com>
|
|
The gofer client's LRU cache has a default limit of 1000 dentries. Any attempt
to cache more dentries than that will make the LRU cache evict and destroy the
least recently used dentry. However, the eviction is expensive because it
requires holding fs.renameMu for writing - which in turn creates a lot of
contention. All filesystem operations that involve path traversal require
fs.renameMu for reading atleast.
Therefore, it is in our best interest to keep the cache small and clean.
When a dentry is inserted in the dentry tree, it grabs a ref on its parent for
its entire lifetime. Hence the parent is longer evictable (because refs > 0).
This change additionally calls checkCachingLocked on directories that have been
added to so that they can be removed from the LRU cache if needed.
This change implies that the LRU cache will only contain the leaves from the
filesystem tree which significantly reduces the LRU cache size and consequently
reduces the number of expensive LRU cache evictions.
> Why are opened dentries not removed from LRU cache?
When a file description is open(2)-ed, the file description holds a ref on its
dentry for its entire lifetime. However, calling checkCachingLocked() on opened
dentries actually ends up hurting performance. Applications usually open file
descriptors for a short duration. So upon close(2), the dentry is reinserted
into the cache anyway. So the precautionary work done in removing the opened
dentry from the cache went for waste as it did not really reduce an eviction.
Local benchmarking has shown that this change improves performance by 3-4%.
Across 6 runs, without this change it took 296.127 seconds to build runsc while
with this change it took only 285.136 seconds.
PiperOrigin-RevId: 369510494
|
|
Remote revalidating requires to update file size on every write
on a file opened with O_APPEND. If host FD exists, it can be
used to update the size and skip round trip to the gofer. With
this change, O_APPEND writes with remote revalidating is almost
as fast as exclusive mode:
BM_Append
VFS1 60.7us
VFS2 56.8us
VFS2 exclusive 14.2us
This change 15.8us
Updates #1792
PiperOrigin-RevId: 369486801
|
|
Fixes #2926, #674
PiperOrigin-RevId: 369457123
|
|
Currently the verity action is a global variable, which causes the same
action for all verity mounts, and is overwritten for each new verity
mount. Changed it to a member of verity fs.
PiperOrigin-RevId: 369348522
|
|
While using remote-validation, the vast majority of time spent during
FS operations is re-walking the path to check for modifications and
then closing the file given that in most cases it has not been
modified externally.
This change introduces a new 9P message called MultiGetAttr which bulks
query attributes of several files in one shot. The returned attributes are
then used to update cached dentries before they are walked. File attributes
are updated for files that still exist. Dentries that have been deleted are
removed from the cache. And negative cache entries are removed if a new
file/directory was created externally. Similarly, synthetic dentries are
replaced if a file/directory is created externally.
The bulk update needs to be carefull not to follow symlinks, cross mount
points, because the gofer doesn't know how to resolve symlinks and where
mounts points are located. It also doesn't walk to the parent ("..") to
avoid deadlocks.
Here are the results:
Workload VFS1 VFS2 Change
bazel action 115s 70s 28.8s
Stat/100 11,043us 7,623us 974us
Updates #1638
PiperOrigin-RevId: 369325957
|
|
Runsc build benchmark's mutex profile shows that we are wasting roughly 25-30
seconds waiting for filesystem.renameMu to get unlocked. Earlier
checkCachingLocked required the renameMu to be locked for writing. This is a
filesystem wide lock which puts all other filesystem operations on hold and
hence is really expensive. Something to note is that all path resolution
operations hold renameMu for reading.
With this change, we allow to check for caching without even holding renameMu.
This change introduces more fine grained locks (fs.cacheMu and dentry.cachingMu)
which protect the cache (removing the requirement to hold renameMu for writing
to modify the cache) and synchronize concurrent dentry caching attempts on a per
dentry basis. We still require to hold renameMu for writing while destroying
dentries and evicting from the cache but this still significantly reduces the
write locking critical section.
Local benchmarking showed that this improved runsc build benchmark time by 4-5%.
Across 6 runs, without this change it took 310.9025 seconds to build runsc
while with this change it took 296.127 seconds.
Runsc build benchmark's mutex profile: https://gvisor.dev/profile/gvisor-buildkite/78a3f968-36ca-4944-93f7-77a8792d56b4/28a1d260-790b-4a9e-94da-a4daede08ee3/tmp/profile/ptrace/BenchmarkBuildRunsc/page_cache.clean/filesystem.bindfs/benchmarks/runsc/mutex.pprof/flamegraph
PiperOrigin-RevId: 368958136
|
|
Add a coverage-report flag that will cause the sandbox to generate a coverage
report (with suffix .cov) in the debug log directory upon exiting. For the
report to be generated, runsc must have been built with the following Bazel
flags: `--collect_code_coverage --instrumentation_filter=...`.
With coverage reports, we should be able to aggregate results across all tests
to surface code coverage statistics for the project as a whole.
The report is simply a text file with each line representing a covered block
as `file:start_line.start_col,end_line.end_col`. Note that this is similar to
the format of coverage reports generated with `go test -coverprofile`,
although we omit the count and number of statements, which are not useful for
us.
Some simple ways of getting coverage reports:
bazel test <some_test> --collect_code_coverage \
--instrumentation_filter=//pkg/...
bazel build //runsc --collect_code_coverage \
--instrumentation_filter=//pkg/...
runsc -coverage-report=dir/ <other_flags> do ...
PiperOrigin-RevId: 368952911
|
|
Also count failed TCP port allocations
PiperOrigin-RevId: 368939619
|
|
Fields allow counter metrics to have multiple tabular values.
At most one field is supported at the moment.
PiperOrigin-RevId: 368767040
|
|
Otherwise ConnectedEndpoint.sndbuf will be restored as 0 and writes
to the socket will fail with EAGAIN.
PiperOrigin-RevId: 368746660
|
|
Thanks ianlewis@ for discovering the bug/fix!
PiperOrigin-RevId: 368740744
|
|
Go 1.17 is adding a new register-based calling convention [1] ("ABIInternal"),
which used is when calling between Go functions. Assembly functions are still
written using the old ABI ("ABI0"). That is, they still accept arguments on the
stack, and pass arguments to other functions on the stack. The call rules look
approximately like this:
1. Direct call from Go function to Go function: compiler emits direct
ABIInternal call.
2. Indirect call from Go function to Go function: compiler emits indirect
ABIInternal call.
3. Direct call from Go function to assembly function: compiler emits direct
ABI0 call.
4. Indirect call from Go function to assembly function: compiler emits indirect
ABIInternal call to ABI conversion wrapper function.
5. Direct or indirect call from assembly function to assembly function:
assembly/linker emits call to original ABI0 function.
6. Direct or indirect call from assembly function to Go function:
assembly/linker emits ABI0 call to ABI conversion wrapper function.
Case 4 is the interesting one here. Since the compiler can't know the ABI of an
indirect call, all indirect calls are made with ABIInternal. In order to
support indirect ABI0 assembly function calls, a wrapper is generated that
translates ABIInternal arguments to ABI0 arguments, calls the target function,
and then converts results back.
When the address of an ABI0 function is taken from Go code, it evaluates to the
address of this wrapper function rather than the target function so that later
indirect calls will work as expected.
This is normally fine, but gVisor does more than just call some of the assembly
functions we take the address of: either noting the start and end address for
future reference from a signal handler (safecopy), or copying the function text
to a new mapping (platforms).
Both of these fail with wrappers enabled (currently, this is Go tip with
GOEXPERIMENT=regabiwrappers) because these operations end up operating on the
wrapper instead of the target function.
We work around this issue by taking advantage of case 5: references to assembly
symbols from other assembly functions resolve directly to the desired target
symbol. Thus, rather than using reflect to get the address of a Go reference to
the functions, we create assembly stubs that return the address of the
function. This approach works just as well on current versions of Go, so the
change can be made immediately and doesn't require any build tags.
[1] https://go.googlesource.com/go/+/refs/heads/master/src/cmd/compile/abi-internal.md
PiperOrigin-RevId: 368505655
|
|
PiperOrigin-RevId: 368470656
|
|
Some FileDescriptions in verity fs were opened but DecRef() were missing
after used. This could result in a ref leak.
PiperOrigin-RevId: 368096759
|
|
Reported-by: syzbot+a6ef0f95a2c9e7da26f3@syzkaller.appspotmail.com
Reported-by: syzbot+2eaf8a9f115edec468fe@syzkaller.appspotmail.com
PiperOrigin-RevId: 368093861
|
|
The current SNAT implementation has several limitations:
- SNAT source port has to be specified. It is not optional.
- SNAT source port range is not supported.
- SNAT for UDP is a one-way translation. No response packets
are handled (because conntrack doesn't support UDP currently).
- SNAT and REDIRECT can't work on the same connection.
Fixes #5489
PiperOrigin-RevId: 367750325
|
|
If the parent is not enabled in verity stepLocked(), failure to find
the child dentry could just mean an incorrect path.
PiperOrigin-RevId: 367733412
|
|
PiperOrigin-RevId: 367730917
|
|
Move maxListenBacklog check to the caller of endpoint Listen so that it
is applicable to Unix domain sockets as well.
This was changed in cl/366935921.
Reported-by: syzbot+a35ae7cdfdde0c41cf7a@syzkaller.appspotmail.com
PiperOrigin-RevId: 367728052
|
|
Set root dentry and root hash in verity fs before we verify the root
directory if a root hash is provided. These are used during
verification.
PiperOrigin-RevId: 367547346
|
|
We should only set parent after child is verified. Also, if the parent
is set before verified, destroyLocked() will try to grab parent.dirMu,
which may cause deadlock.
PiperOrigin-RevId: 367543655
|
|
PiperOrigin-RevId: 367523491
|
|
Without this change, we ask the gofer server to update the permissions
whenever the UID, GID or size is updated via SetStat. Consequently, we don not
generate inotify events when the permissions actually change due to SGID bit
getting cleared.
With this change, we will update the permissions only when needed and generate
inotify events.
PiperOrigin-RevId: 366946842
|
|
- Change the accept queue full condition for a listening endpoint
to only honor completed (and delivered) connections.
- Use syncookies if the number of incomplete connections is beyond
listen backlog. This also cleans up the SynThreshold option code
as that is no longer used with this change.
- Added a new stack option to unconditionally generate syncookies.
Similar to sysctl -w net.ipv4.tcp_syncookies=2 on Linux.
- Enable keeping of incomplete connections beyond listen backlog.
- Drop incoming SYNs only if the accept queue is filled up.
- Drop incoming ACKs that complete handshakes when accept queue is full
- Enable the stack to accept one more connection than programmed by
listen backlog.
- Handle backlog argument being zero, negative for listen, as Linux.
- Add syscall and packetimpact tests to reflect the changes above.
- Remove TCPConnectBacklog test which is polling for completed
connections on the client side which is not reflective of whether
the accept queue is filled up by the test. The modified syscall test
in this CL addresses testing of connecting sockets.
Fixes #3153
PiperOrigin-RevId: 366935921
|
|
PiperOrigin-RevId: 366923274
|
|
PiperOrigin-RevId: 366891806
|
