-
Notifications
You must be signed in to change notification settings - Fork 564
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Assertion failure in process_syscall_entry() during record #3729
Comments
Looks like this is a bug in our PTRACE_ATTACH emulation. Event 72772 is a PTRACE_ATTACH to tid 123227, then event 72776 is that process executing a syscall exit. Seems like we're not expecting the tracee to be in a syscall? |
We do handle |
Sure. Steps to reproduce would be useful. |
Please be warned that the executable is 5.3GB, and compiling it from source will take ~50GB of disk space (and a lot of time). I'm waiting on it to upload but meanwhile to reproduce on it on your own: To get the bug |
The executable link: https://utexas.box.com/shared/static/cupt90r924jwo9l40c13ik4hxobgfv1o |
I downloaded that binary and ran it with |
The assertion is similar to the one in #3745... which I also can't reproduce. If you re-reproduce this bug with master, the assertion should include the pid that we think sent |
Im also using Arch, just like that other bug. Ill build master and see if I can get you the PID. |
I can't reproduce it with master. The arch version that does have the issue is 5.7.0 |
do you want to try bisecting to find the rr change that fixed this? |
That's a good idea, I'll try it. |
I can't reproduce with rr 5.7.0 built from source, only with the Arch package. What's even stranger is that when I copy the build config that Arch uses, I still can't reproduce the issue. Since building from source seems to work, I'll try to get a trace of it and upload it here. |
Since this seems like a bug with the Arch package, I'll close this for now unless you want to investigate further. |
Can you file a Arch bug about it? |
PS thanks for checking this out. |
Attempting to debug LLVM opt, crash only happens with a Bazel development build (
-O3 -g -UNDEBUG -fno-omit-frame-pointer -fsanitize=address,undefined
, LLVM commit44718311dee486f1823876e8af9100afcc50041b
); regular release binaries seem to be fine.Can provide executable if needed.
Log:
The text was updated successfully, but these errors were encountered: