Index out of range panic at src/libcore/slice/mod.rs:2413 about symbolic HOT 5 CLOSED

Also is there any way to recover from the panic? I tried except symbolic.Panic but we still lose the parent process with a core dump

We noticed the wheel on pip is built with python2.7 and we managed to fix this by building everything ourselves for python3.7 with rust 1.34. So at least it doesn't crash with a core dump now

from symbolic.

@AdrianLC would it be possible to share the dSYM with me (via email to [email protected], for instance)? I'm curious what might cause this panic -- especially since there shouldn't be a large difference between our binary wheels and a custom build.

Also, I'm curious as to why there is a core dump. The library catches the panic down, which is why you can see it along with the Python stack trace.

Would love to take a detailled look at a repro case or the affected file.

from symbolic.

Thank you @jan-auer. I just sent you the dSYM by email.

We generally cannot see the panic when it crashes. But for some reason I managed to see it when running the code from a shell. Also, it sort of seems to be fixed with our build and we get an exception instead. (Sort of, because we suspect it still might be happening with other zips)

from symbolic.

Hi, with the 7.1.0 release our regression test has changed from symbolic.Panic to this

symbolic.ObjectErrorBadMachOobject: failed to process macho file
  caused by: invalid MachO file
  caused by: type is too big (3074949) for 754944

stacktrace: stack backtrace:
   0: failure::backtrace::Backtrace::new
             at /root/.cargo/registry/src/github.com-1ecc6299db9ec823/failure-0.1.5/src/backtrace/internal.rs:44
      <failure::backtrace::Backtrace as core::default::Default>::default
             at /root/.cargo/registry/src/github.com-1ecc6299db9ec823/failure-0.1.5/src/backtrace/mod.rs:125
   1: symbolic_archive_get_object
             at /root/.cargo/registry/src/github.com-1ecc6299db9ec823/failure-0.1.5/src/error/error_impl.rs:19

I guess the change comes from https://github.com/getsentry/symbolic/pull/169/files maybe it's enough to close this issue.

from symbolic.

Looking at the original stack trace, I have the assumption this was an wrapping arithmetic problem that I found and fixed some time ago both coming from another customer issue, as well as using fuzzing.

The second stack trace reported later on looks different.

Since this issue is already over 2 years old, I will close this for now, and ask you to reopen if this is still a problem, preferably with a failing testcase that we can look at.

from symbolic.