FATAL assert_same_rec

[fitzgen]

STR:

• checkout my branch of m-c: https://github.com/fitzgen/gecko-dev/tree/heap-snapshots-e10s-copy
• $ ./mach build
• $ ./mach xpcshell-test toolkit/devtools/heapsnapshot/tests/unit_ipc/test_saveHeapSnapshot_e10s_01.js --debugger rr
• $ rr replay
• (gdb) continue

---

This gist has my full rr replay session that led up to the assertion failure, as well as conecting gdb to rr and dumping rr's backtrace: https://gist.github.com/fitzgen/04d8d34ca7d059c59d13

My rr is built from this commit:

commit a7f6a951d58d61a56a31dcff262e27024fa40aee
Author: Robert O'Callahan <robert@ocallahan.org>
Date:   Wed Sep 16 15:08:53 2015 +1200

    Print errno name when assertion fails.

[rocallahan]

I tried to reproduce this from your repo and I did not see the bug. Is it 100% reproducible for you? I made a debug build, could that be the difference?

If this is reproducible for you, the most effective way forward might be for me to log in to your system and debug it there.

[fitzgen]

My firefox build was also a debug build. I compiled with gcc (not sure which version, I'm on OSX right now).

I was able to reproduce it twice in a row, and then I gave up and switched to plain gdb.

I'm open to helping however I can. Let me know when digging in works well for you.