[sanitizer] Print diagnostic if ptrace syscall fails #151406
Merged
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
StopTheWorld() spins up a thread that calls ptrace before releasing a mutex; the other thread yields until the mutex is unlocked. If seccomp kills the thread that calls ptrace, the other thread will silently hang. The latter thread cannot use waitpid to detect that the other thread has been killed, because the threads share errno. This patch forks the process to test whether ptrace is allowed. If it fails, it prints an informational message (though it does not abort the sanitizer). Fixes llvm#150380 and google/sanitizers#777
|
@llvm/pr-subscribers-compiler-rt-sanitizer Author: Thurston Dang (thurstond) ChangesStopTheWorld() spins up a thread that calls ptrace before releasing a mutex; the other thread yields until the mutex is unlocked. If seccomp kills the thread that calls ptrace, the other thread will silently hang. The latter thread cannot use waitpid to detect that the other thread has been killed, because the threads share errno. This patch forks the process to test whether ptrace is allowed. If it fails, it prints an informational message (though it does not abort the sanitizer). Fixes #150380 and google/sanitizers#777 Full diff: https://github.com/llvm/llvm-project/pull/151406.diff 1 Files Affected:
diff --git a/compiler-rt/lib/sanitizer_common/sanitizer_stoptheworld_linux_libcdep.cpp b/compiler-rt/lib/sanitizer_common/sanitizer_stoptheworld_linux_libcdep.cpp
index 24929b8c4bd7c..ce21a156d28b6 100644
--- a/compiler-rt/lib/sanitizer_common/sanitizer_stoptheworld_linux_libcdep.cpp
+++ b/compiler-rt/lib/sanitizer_common/sanitizer_stoptheworld_linux_libcdep.cpp
@@ -403,7 +403,48 @@ struct ScopedSetTracerPID {
}
};
+// This detects whether ptrace is blocked (e.g., by seccomp), by forking and
+// then attempting ptrace.
+// This separate check is necessary because StopTheWorld() creates a *thread*,
+// and therefore cannot use waitpid() due to the shared errno.
+static void TestPTrace() {
+ // Only check the first time this is called.
+ static bool checked = false;
+ if (checked)
+ return;
+ checked = true;
+
+ // fork() should be cheap because of copy-on-write. Besides, this is only
+ // called the first time.
+ int pid = internal_fork();
+
+ if (pid < 0) {
+ int rverrno;
+ if (internal_iserror(pid, &rverrno)) {
+ Report("WARNING: TestPTrace() failed to fork (errno %d)\n", rverrno);
+ }
+ _exit(-1);
+ }
+
+ if (pid == 0) {
+ // Child subprocess
+ internal_ptrace(PTRACE_ATTACH, 0, nullptr, nullptr);
+ _exit (0);
+ } else {
+ int wstatus;
+ internal_waitpid(pid, &wstatus, 0);
+
+ if (WIFSIGNALED(wstatus)) {
+ VReport(0, "Warning: ptrace appears to be blocked (is seccomp enabled?). LeakSanitizer may hang.\n");
+ VReport(0, "Child exited with signal %d.\n", WTERMSIG(wstatus));
+ // We don't abort the sanitizer - it's still worth letting the sanitizer try.
+ }
+ }
+}
+
void StopTheWorld(StopTheWorldCallback callback, void *argument) {
+ TestPTrace();
+
StopTheWorldScope in_stoptheworld;
// Prepare the arguments for TracerThread.
struct TracerThreadArgument tracer_thread_argument;
|
|
✅ With the latest revision this PR passed the C/C++ code formatter. |
fmayer
reviewed
Jul 30, 2025
| // This detects whether ptrace is blocked (e.g., by seccomp), by forking and | ||
| // then attempting ptrace. | ||
| // This separate check is necessary because StopTheWorld() creates a *thread*, | ||
| // and therefore cannot use waitpid() due to the shared errno. |
There was a problem hiding this comment.
errno should be thread local on any reasonable platform?
There was a problem hiding this comment.
The existing comment on line 504 states that errno is shared. I think sanitizers run on all manner of unreasonable platforms?
There was a problem hiding this comment.
This whole code is under SANITIZER_LINUX though
There was a problem hiding this comment.
maybe the real reason is the clone syscall?
There was a problem hiding this comment.
maybe the real reason is the clone syscall?
*maybe the real treasure was the clones we made along the way
fmayer
reviewed
Aug 4, 2025
| return; | ||
| checked = true; | ||
|
|
||
| // fork() should be cheap because of copy-on-write. Besides, this is only |
There was a problem hiding this comment.
I wouldn't call fork "cheap" exactly
| // This detects whether ptrace is blocked (e.g., by seccomp), by forking and | ||
| // then attempting ptrace. | ||
| // This separate check is necessary because StopTheWorld() creates a *thread*, | ||
| // and therefore cannot use waitpid() due to the shared errno. |
There was a problem hiding this comment.
This whole code is under SANITIZER_LINUX though
fmayer
reviewed
Aug 4, 2025
| // This separate check is necessary because StopTheWorld() creates a *thread*, | ||
| // and therefore cannot use waitpid() due to the shared errno. | ||
| // This separate check is necessary because StopTheWorld() creates a child | ||
| // process with a shared virtual address space, and therefore cannot use |
There was a problem hiding this comment.
I don't think this is true though. Thread locals should be relative to the thread pointer, which should be different. That wasn't a statement on my part, but a question (you can do all sorts of messed up stuff with clone flags)
fmayer
approved these changes
Aug 4, 2025
|
|
LLVM Buildbot has detected a new failure on builder Full details are available at: https://lab.llvm.org/buildbot/#/builders/174/builds/22276 Here is the relevant piece of the build log for the reference |
thurstond
added a commit
that referenced
this pull request
Aug 4, 2025
…ll fails #151406" Use internal__exit instead of _exit Buildbot: https://lab.llvm.org/buildbot/#/builders/174/builds/22276
|
LLVM Buildbot has detected a new failure on builder Full details are available at: https://lab.llvm.org/buildbot/#/builders/204/builds/17749 Here is the relevant piece of the build log for the reference |
|
LLVM Buildbot has detected a new failure on builder Full details are available at: https://lab.llvm.org/buildbot/#/builders/140/builds/28289 Here is the relevant piece of the build log for the reference |
thurstond
added a commit
to thurstond/llvm-project
that referenced
this pull request
Aug 5, 2025
Fixes corner case of llvm#151406 internal_fork() on SPARC actually calls __fork(). We can't safely fork, because it's possible seccomp has been configured to disallow fork() but allow clone(). Also adds some comments/TODOs.
thurstond
added a commit
to thurstond/llvm-project
that referenced
this pull request
Aug 5, 2025
Fixes corner case of llvm#151406 internal_fork() on SPARC actually calls __fork(). We can't safely fork, because it's possible seccomp has been configured to disallow fork() but allow clone(). Also adds some comments/TODOs.
thurstond
added a commit
to thurstond/llvm-project
that referenced
this pull request
Aug 5, 2025
Fixes corner case of llvm#151406 internal_fork() on SPARC actually calls __fork(). We can't safely fork, because it's possible seccomp has been configured to disallow fork() but allow clone(). Also adds some comments/TODOs.
|
LLVM Buildbot has detected a new failure on builder Full details are available at: https://lab.llvm.org/buildbot/#/builders/56/builds/32558 Here is the relevant piece of the build log for the reference |
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.
StopTheWorld() clones a child process (with shared virtual address space and shared TLS) that calls ptrace before releasing a mutex; the parent process yields until the mutex is unlocked. If seccomp kills the child process, the parent process will silently hang. The parent process cannot use waitpid to detect that the child process has been killed, because the processes share errno.
This patch forks the process one-time to test whether ptrace is allowed. If it fails, it prints an informational message (though it does not abort the sanitizer).
Fixes #150380 and google/sanitizers#777