| Commit message (Collapse) | Author | Age |
|
|
|
|
|
|
| |
Move the message-digest cloning to DigestHashFunction and out of Fingerprint, to make it possible to configure Fingerprint to use different hash functions. We keep the default MD5 for now, we'd like it to use the global default but want to isolate the configuration change from any change adding potential contention.
RELNOTES: None.
PiperOrigin-RevId: 208502993
|
|
|
|
|
|
|
|
| |
Otherwise Docker will not be able to read the configuration of the current user and might fail to download containers from registries that require authentication.
Closes #5721.
PiperOrigin-RevId: 207094268
|
|
|
|
|
| |
RELNOTES: None
PiperOrigin-RevId: 203454741
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The intent is that the main thread has ~zero gaps in the profile (though there
may still be small gaps due to the time between one try block and an
immediately subsequent try block).
We need to be careful not to wrap markPhase calls (or methods that call
markPhase) in try blocks for the profiler - the Profiler requires that all
markPhase calls happen at top level, and throws an exception if not.
This should not have any performance impact - all of these are once per build,
or at most once per module per build, and we don't expect a very large number
of modules (and if we see an increasing number, we need to change the module
API to not have to call every single module, but only those that are actually
interested in certain events, maybe with an EventBus-based setup).
PiperOrigin-RevId: 200712677
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This also gets rid of some boilerplate. The ExecutionTool.addActionContext
method has been around for a while, but is underused.
There are still a few ActionContextProvider implementations left, which are
implementing other functionality besides adding action contexts.
As a side effect, this change reduces null build time with a hot server on
linux by about a quarter. We were running the linux sandbox twice on every
build, which takes about 70ms each (on my machine), with the total null build
time around 300ms.
PiperOrigin-RevId: 200045145
|
|
|
|
|
|
|
|
| |
Instead, add some simple APIs to ExecutorBuilder and inline all the previous
subclasses into their corresponding modules. This removes a bunch of
boilerplate.
PiperOrigin-RevId: 200017162
|
|
|
|
|
| |
RELNOTES: None.
PiperOrigin-RevId: 199467128
|
|
|
|
|
|
|
|
| |
uid and gid of -1 (set when not linux) can not really be used anywhere.
Closes #5165.
PiperOrigin-RevId: 196974660
|
|
|
|
| |
PiperOrigin-RevId: 195100670
|
|
|
|
|
|
|
| |
Spawns can contain param files in the form of virtual action inputs. Sandboxed runners must write these prior to execution of the action.
RELNOTES: None
PiperOrigin-RevId: 194951208
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
There is no need for the cache to be on disk. Originally, there was a
desire to share this cache with other tools... but this never happened.
And, actually, because Bazel is in control of what it runs, it can just
inject the "cached" values into those tools via flags.
Instead, just store the cache in-memory. This avoids having to open and
read the cache on every single action that is locally executed on a Mac.
Results when building a large iOS app from a clean slate show up to a
1% wall time improvement on my Mac Pro 2013 and a reduction in the
variance of the measurements.
This change also gets rid of the OS check from the action execution's
critical path. There is not much use in checking this: if we instantiate
this by mistake, the actual calls will fail. But sometimes we want to
actually run this code on non-macOS systems (e.g. for unit-testing with
mocked tools), so we should allow that. And this change also ensures
that XcodeLocalEnvProviderTest builds and runs...
RELNOTES: None.
PiperOrigin-RevId: 194681802
|
|
|
|
|
|
|
| |
RELNOTES: Bazel now supports running actions inside Docker containers.
To use this feature, run "bazel build --spawn_strategy=docker
--experimental_docker_image=myimage:latest".
PiperOrigin-RevId: 194582691
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Turns out the "unique" id returned by SpawnExecutionContext is only
unique within one SpawnRunner, but not across multiple SpawnRunner
classes.
We have to prefix the generated path with the name of the current
strategy in order to not run into conflicts when a build happens to use
multiple different sandbox strategies.
RELNOTES: None.
PiperOrigin-RevId: 194230475
|
|
|
|
|
|
|
|
|
| |
The CopyingSandboxedSpawn will be used by the upcoming Docker sandbox.
This is otherwise a no-op change.
RELNOTES: None.
PiperOrigin-RevId: 194096413
|
|
|
|
|
|
| |
Always use the more-qualified class name for clarity at the site of use. There are too many classes named Builder.
PiperOrigin-RevId: 193649193
|
|
|
|
|
|
|
|
| |
This class will be used to tie a Spawn to a SpawnRunner, and isn't really a policy object. It will carry state such as the expanded inputs and expanded command line.
Currently a context can be passed between different SpawnRunners. This will be addressed independently, so a context is tied to a particular spawn runner.
PiperOrigin-RevId: 193501918
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
https://github.com/bazelbuild/bazel/commit/656a0bab1e025ff3c27d595284a4bf1c5a8d8028 with test (unknown commit) and fix.
Big round of sandbox fixes / performance improvements.
- The number of stat() syscalls in the SymlinkedSandboxedSpawn was way too high. Do less, feel better.
- When using --experimental_sandbox_base, ensure that symlinks in the path are resolved. Before this, you had to check whether on your system /dev/shm is a symlink to /run/shm and then use that instead. Now it no longer matters, as symlinks are resolved.
- Remove an unnecessary directory creation from each sandboxed invocation. Turns out that the "tmpdir" that we created was no longer used after some changes to Bazel's TMPDIR handling.
- Use simpler sandbox paths, by using the unique ID for each Spawn provided by SpawnExecutionPolicy instead of a randomly generated temp folder name. This also saves a round-trip from our VFS to NIO and back. Clean up the sandbox base before each build to ensure that the unique IDs are actually unique. ;)
- Use Java 8's Process#isAlive to check whether a process is alive instead of trying to get the exitcode and catching an exception.
Closes #4913.
PiperOrigin-RevId: 193031017
|
|
|
|
|
|
|
|
|
| |
Exposing devices through a FUSE mount point is possible but accessing them
requires the file system to be mounted -odev, which needs root privileges for
security reasons. Because we cannot ask sandboxfs to be mounted this way,
change our code to instead use a regular empty file when necessary.
PiperOrigin-RevId: 192761352
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
SandboxedSpawn.copyOutputs.
Previously, if moveFile fell back to copying from a true rename (e.g., if the
move is across file systems), it would fully dereference the source and produce
a regular file at the output location. This CL fixes moveFile to properly copy
symlinks. Note we don't preserve metadata in the symlink case mostly because the
Bazel VFS has no API to write metadata without dereferencing symlinks. (But,
also, it's not important for the current use cases.)
This breaks the backward-compatibility of FileSystemUtils.moveFile and
FileSystemUtils.moveTreeBelow. This seems okay because the new behavior makes
more sense, and sandbox is the only consumer of these APIs.
Switching SandboxedSpawn.copyOutputs to use FileSystemUtils.moveFile rather than
Guava's Files.move fixes https://github.com/bazelbuild/bazel/issues/4987.
Closes #4989.
Change-Id: I0283e8aa11fadff5b0afd7bdfd0490aca12a1f6b
PiperOrigin-RevId: 192611227
|
|
|
|
|
|
|
| |
This is essentially the same as https://github.com/bazelbuild/bazel/commit/3a7b8bc2abeaf8b8647c037bed1dd5fd73b8392b.
RELNOTES: none.
PiperOrigin-RevId: 192342039
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
If an action expresses a symlink as an input, the target of the symlink does
not necessarily appear as a file to map within the sandbox. This is a
problem when the target of the symlink is relative because sandboxfs would
expose the link verbatim and the target would be missing later on during
resolution.
To fix this, special-case the handling of symlinks: when trying to expose
them via a sandboxfs mount point, resolve their final target instead of
respecting the original contents. This loses the fact that the file was a
symlink when running within the sandboxfs sandbox, but is easier to
implement and slightly faster at runtime. We can reconsider this choice
if this causes problems.
RELNOTES: None.
PiperOrigin-RevId: 192325932
|
|
|
|
| |
PiperOrigin-RevId: 191642942
|
|
|
|
|
|
|
|
|
| |
These have all had a chance to be categorized with the OptionDocumentationCategory enum, and the help output already uses the enum-grouped format.
The "incompatible changes" category has meaning for --all_incompatible_changes and will be removed separately.
RELNOTES: None.
PiperOrigin-RevId: 190773778
|
|
|
|
|
|
| |
holds a variety of strategy/context maps.
PiperOrigin-RevId: 190491357
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
- The number of stat() syscalls in the SymlinkedSandboxedSpawn was way too high. Do less, feel better.
- When using --experimental_sandbox_base, ensure that symlinks in the path are resolved. Before this, you had to check whether on your system /dev/shm is a symlink to /run/shm and then use that instead. Now it no longer matters, as symlinks are resolved.
- Remove an unnecessary directory creation from each sandboxed invocation. Turns out that the "tmpdir" that we created was no longer used after some changes to Bazel's TMPDIR handling.
- Use simpler sandbox paths, by using the unique ID for each Spawn provided by SpawnExecutionPolicy instead of a randomly generated temp folder name. This also saves a round-trip from our VFS to NIO and back. Clean up the sandbox base before each build to ensure that the unique IDs are actually unique. ;)
- Use Java 8's Process#isAlive to check whether a process is alive instead of trying to get the exitcode and catching an exception.
Closes #4913.
PiperOrigin-RevId: 190472170
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
- Remove Optional<> where it's not needed. It's nice for return values, but IMHO it was overused in this code (e.g. Optional<List<X>> is an anti-pattern, as the list itself can already signal that it is empty).
- Use Bazel's own Path class when dealing with paths, not String or java.io.File.
- Move LinuxSandboxUtil into the "sandbox" package.
- Remove dead code and unused fields.
- Migrate deprecated VFS method calls to their replacements.
- Fix a bug in ExecutionStatistics where a FileInputStream was not closed.
Closes #4868.
PiperOrigin-RevId: 190217476
|
|
|
|
|
|
|
|
|
|
| |
The product name was previously propagated to all sandboxed spawn runners
unconditionally, even when the majority did not need it. Also, these
same spawn runners had obsolete constructors that were not used at all or
were only used in tests. Remove both for simplicity.
RELNOTES: None.
PiperOrigin-RevId: 190195305
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This only modifies the DarwinSandboxedSpawnRunner for now and leaves behind
the Linux version.
The latter is a bit more complex to do because of the need to deal with tmpfs
and bind-mount interactions. We expect sandboxfs to have the biggest impact
on macOS anyway, not Linux, so this can wait until this has proven that it
deserves staying for the long-term.
Tested: Manually ran a self-build of Bazel on a version of Bazel built with
these changes and the build passed.
RELNOTES: None.
PiperOrigin-RevId: 190133429
|
|
|
|
|
| |
RELNOTES: None.
PiperOrigin-RevId: 190062172
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This introduces user-facing options to enable the experimental sandboxfs
support and, when enabled, mounts a sandboxfs instance throughout the
build. The sandboxfs' process handle is passed to the
SandboxActionContextProvider so that the SpawnRunners can later consume it.
Note that this does NOT yet provide sandboxfs support for the builds as the
SpawnRunners are untouched.
RELNOTES: None.
PiperOrigin-RevId: 189678732
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The new SandboxfsProcess interface allows interacting with sandboxfs.
There are two implementations: RealSandboxfsProcess, which spawns the
sandboxfs binary, and FakeSandboxfsProcess, which mimics what sandboxfs
does but using symlinks and is intended for testing purposes only.
The RealSandboxfsProcess implementation works but still carries many
TODOs. The most "painful" one may be that the test requires manual
invocation because we do not yet have an easy way to integrate with
sandboxfs. That will be solved later on; for now this is sufficient
for initial testing.
RELNOTES: None.
PiperOrigin-RevId: 188347393
|
|
|
|
|
|
| |
Closes #4640.
PiperOrigin-RevId: 188022228
|
|
|
|
|
|
|
|
|
|
| |
just add it to the interface, and include it in the SpawnResult. This will be used to categorize/aggregate spawns executed by various runners.
Also, minor refinement to the cacheHit property of the SpawnResult with remote execution.
RELNOTES: None
TESTED=presubmit, next cl
PiperOrigin-RevId: 186637978
|
|
|
|
|
|
| |
...instead of passing it everywhere so we can rewrite the local env.
PiperOrigin-RevId: 186316859
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
*** Reason for rollback ***
There's already a --test_tmpdir flag, and Java tests don't pick up this new one. More info: https://github.com/bazelbuild/bazel/issues/4621#issuecomment-366217321
*** Original change description ***
tmpdir,local-exec: implement --local_tmp_root
Add new flag called `--local_tmp_root`, which (if
specified) tells Bazel what temp directory should
locally executed actions use.
Fixes https://github.com/bazelbuild/bazel/issues/4621
Related to https://github.com/bazelbuild/bazel/issues/3215
RELNOTES[NEW]: The new "--local_tmp_root=<path>" flag allows specifying the temp directory for locally executed actions.
Change-Id: Ice69a5e63d0bf4d3b5c9ef4dbdd1ed1c5025f85e
PiperOrigin-RevId: 185982705
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Add new flag called `--local_tmp_root`, which (if
specified) tells Bazel what temp directory should
locally executed actions use.
Fixes https://github.com/bazelbuild/bazel/issues/4621
Related to https://github.com/bazelbuild/bazel/issues/3215
RELNOTES[NEW]: The new "--local_tmp_root=<path>" flag allows specifying the temp directory for locally executed actions.
Change-Id: Ice69a5e63d0bf4d3b5c9ef4dbdd1ed1c5025f85e
PiperOrigin-RevId: 185509555
|
|
|
|
|
|
|
| |
rule for the execution statistics path.
RELNOTES: None.
PiperOrigin-RevId: 184534589
|
|
|
|
|
|
|
| |
Fixes https://github.com/bazelbuild/bazel/issues/4376
Change-Id: Id78bb0930044626304e54f07735db4d4b2c84720
PiperOrigin-RevId: 181959528
|
|
|
|
|
|
| |
They need this to parse input manifests. Previously we would grab the exec root from the Root, but wish to unsupport this.
PiperOrigin-RevId: 181669143
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Bazel now adds env[TMPDIR] to the set of
sandbox-writable paths, instead of adding the
caller-defined `tmpDir` as it used to.
Since every caller of getWritableDirs passes the
LocalEnvProvider-processed environment to
getWritableDirs, and because all such callers use
either PosixLocalEnvProvider or
XCodeLocalEnvProvider, we can be sure that the
environment has an entry for TMPDIR.
Change-Id: Ia89544a009e56d9cc922ab56823d16d20465545e
PiperOrigin-RevId: 181595606
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
When Bazel creates the sandbox for an action,
Bazel collects a set of paths that the action may
write to.
The action needs write access to its temp
directory, so Bazel needs to add it to the
writable paths.
See https://github.com/bazelbuild/bazel/issues/4376
Change-Id: Ifd3c482aa67ff8a2070045356abad8b39c808db8
PiperOrigin-RevId: 181591520
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
When Bazel creates the sandbox, it will allow
making non-existent paths writable, as long as the
path is under the sandbox root.
As Bazel adds entries to the sandbox's set of
writable paths, Bazel needs to make sure that it's
not adding symlinks, because doing so would make
the symlink writable, not what the link points to.
If the path is under the sandbox root, then at the
time of setting up the sandbox's writable paths
the path surely doesn't exist yet, but that's OK,
because at that time Bazel didn't yet create the
sandbox root.
If the path is not under the sandbox root, then
Bazel needs to resolve all symlinks on this path,
which is only possible if the path exists,
therefore Bazel checks for the path's existence.
Change-Id: Ic7d99a81905e7401455286c0b375d69b85ece1d5
PiperOrigin-RevId: 181325749
|
|
|
|
|
|
|
|
| |
This simplifies some spawn runners, which no longer have to specially handle
null; unfortunately, the sandbox runners do not support VirtualActionInput,
so they still have to special-case it.
PiperOrigin-RevId: 181175408
|
|
|
|
|
|
|
|
|
|
|
| |
The sandboxed Spawn runners already used to call
LocalEnvProvider.rewriteEnvironment in order to
add TMPDIR for example, but didn't consistently
use this rewritten environment everywhere, and
used the Spawn's original environment instead.
Change-Id: Ifbede6ac0f14b83c617dcbcd85edfae88718b157
PiperOrigin-RevId: 181130506
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
As part of setting up a sandbox, Bazel creates the
list of writable paths. If the action's
environment defines $TEST_TMPDIR, then it's a test
action and the sandbox must allow writing to that
path, therefore Bazel must add $TEST_TMPDIR to the
writable paths.
Bazel must resolve symlinks in that path though,
at least on the last path segment, because in case
the path points to a symlink, the action would be
allowed to modify the symlink itself, and not
access what the link points to.
However the path must exist for Bazel to
successfully resolve symlinks, therefore this
commit adds a check for that.
Given that the code was there since at least July
2017, and I'm not aware of bugs caused by it, I
conclude that this code path either never runs or
nobody ever triggered it. Either way, adding the
check is the right thing to do.
Change-Id: I87a5d3fc3fe7878a918ed318c71e8d135f10f1b8
PiperOrigin-RevId: 180931382
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Create a PosixLocalEnvProvider and
WindowsLocalEnvProvider class, with singleton
instances for now.
This refactoring should not change functionality,
it's just a requirement for an upcoming change.
That upcoming change is for these classes to
respect the client environment's TMPDIR or
TMP/TEMP envvars.
See https://github.com/bazelbuild/bazel/issues/4376
Change-Id: I032bb6f18adf8af9e43e6bc543c09c58adae3863
PiperOrigin-RevId: 180799936
|
|
|
|
|
|
|
|
|
|
|
| |
that use the DarwinSandboxedSpawnRunner.
In particular, record metrics for user and system CPU execution time, block I/O and involuntary context switches.
This feature is guarded behind a new option, --experimental_collect_local_sandbox_action_metrics.
RELNOTES: None.
PiperOrigin-RevId: 179980734
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
that use either the LinuxSandboxedSpawnRunner or the ProcessWrapperSandboxedSpawnRunner.
In particular, record metrics for user and system CPU execution time, block I/O and involuntary context switches.
This feature is guarded behind a new option, --experimental_collect_local_sandbox_action_metrics.
Note: We still need to enable execution statistics for the DarwinSandboxedSpawnRunner in a later change.
RELNOTES: None.
PiperOrigin-RevId: 179976217
|
|
|
|
|
|
| |
Only the sandbox uses the cached version. We can let it do its own caching of the top-level directory. It won't help with higher-level directories, but if there are a lot of inputs in the same directory we'll avoid I/O there.
PiperOrigin-RevId: 179830651
|
|
|
|
|
|
|
| |
be consistent with LinuxSandboxUtil.
RELNOTES: None.
PiperOrigin-RevId: 179758847
|