systrap: Initialize stub threads in the same way on both x86 and ARM. #11745
+15
−35
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
679f09e to
d6da7f7
Compare
There was a problem hiding this comment.
Pull Request Overview
This PR unifies stub thread initialization on x86 and ARM by removing the custom entrypoint stub and using a fake TGKILL event to enter the shared sighandler, avoiding RFLAGS single-step corruption under debuggers.
- Hoist and unify context/FS-base initialization in the AMD64 signal handler, removing
__export_start. - Remove
archSpecificSysmsgThreadInithelpers and inline the TGKILL stop event in Go. - Drop manual RIP/RSP assignments in x86 register setup.
Reviewed Changes
Copilot reviewed 4 out of 4 changed files in this pull request and generated 1 comment.
| File | Description |
|---|---|
| pkg/sentry/platform/systrap/sysmsg/sighandler_amd64.c | Hoisted declarations, handled THREAD_STATE_INITIALIZING via init_new_thread() and goto init, removed __export_start stub. |
| pkg/sentry/platform/systrap/subprocess_arm64.go | Deleted the ARM helper archSpecificSysmsgThreadInit. |
| pkg/sentry/platform/systrap/subprocess_amd64.go | Removed empty archSpecificSysmsgThreadInit and stub RIP/RSP setup. |
| pkg/sentry/platform/systrap/subprocess.go | Inlined TGKILL fake event before SIGCONT, replaced helper call. |
Comments suppressed due to low confidence (3)
pkg/sentry/platform/systrap/subprocess_amd64.go:213
- Removing the RIP and RSP assignments means the new thread’s instruction pointer and stack aren’t set to the sighandler stub; re-introduce these assignments or ensure an alternative startup address is configured.
regs.PtraceRegs.Rip = uint64(stubSysmsgStart + uintptr(sysmsg.Sighandler_blob_offset____export_start))
pkg/sentry/platform/systrap/subprocess.go:1162
- [nitpick] The panic message is generic; include context such as the target thread ID to aid debugging (e.g.,
panic(fmt.Sprintf("tgkill(%d, %d) failed: %v", ...))).
panic(fmt.Sprintf("tkill failed: %v", e))
pkg/sentry/platform/systrap/subprocess.go:1161
- [nitpick] This new fake TGKILL sequence isn’t covered by existing tests; consider adding an integration or unit test to verify the subprocess enters the sighandler as expected on both architectures.
if e := hostsyscall.RawSyscallErrno(unix.SYS_TGKILL, uintptr(sysThread.thread.tgid), uintptr(sysThread.thread.tid), uintptr(unix.SIGSEGV)); e != 0 {
| @@ -331,6 +334,8 @@ void __export_sighandler(int signo, siginfo_t *siginfo, void *_ucontext) { | |||
| } | |||
|
|
|||
| atomic_store(&ctx->fpstate_changed, 0); | |||
|
|
|||
| init: | |||
There was a problem hiding this comment.
[nitpick] The label init is very generic and may collide or confuse readers; consider renaming it to something more descriptive like restore_context.
Suggested change
| init: | |
| restore_context: |
Copilot uses AI. Check for mistakes.
652d78a to
17b4bde
Compare
Initializing threads via a custom Systrap entrypoint (as was done on x86 prior to this CL) is incompatible with debuggers, because they set the single-step flag. If a brand new stub thread managed to pick up an already existing context with this flag set, it would immediately invoke the sighandler upon restoring RFLAGS, which would corrupt the stack and instruction pointers. PiperOrigin-RevId: 761250233
17b4bde to
53489c2
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
systrap: Initialize stub threads in the same way on both x86 and ARM.
Initializing threads via a custom Systrap entrypoint (as was done on x86 prior
to this CL) is incompatible with debuggers, because they set the single-step
flag. If a brand new stub thread managed to pick up an already existing context
with this flag set, it would immediately invoke the sighandler upon restoring
RFLAGS, which would corrupt the stack and instruction pointers.