636081 - x11 error messages repeated in crash reports

Status: NEW

(Toolkit :: Crash Reporting, defect)

Description

[Karl Tomlinson (:karlt)]

Since http://hg.mozilla.org/mozilla-central/rev/9521cf441d3a,
(and http://hg.mozilla.org/mozilla-central/rev/e1d2e090c9a6),
NS_RUNTIMEABORT appends the error string to crash report App Notes in
libxul-builds/apps.  As we've already added the string, it ends up repeated,
and we get app notes such as

 "X_CloseDevice: XI_BadDevice (invalid Device parameter); 9 requests agoxpcom_runtime_abort(###!!! ABORT: X_CloseDevice: XI_BadDevice (invalid Device parameter); 9 requests ago: file /builds/slave/cen-lnx64-ntly/build/toolkit/xre/nsX11ErrorHandler.cpp, line 190)"

Comment 1

[Chris Jones [:cjones] inactive; ni?/f?/r? if you need me]

Oops!  Unless there are existing tools that break when trying to parse the error string inside of the "xpcom_runtime_abort()", the simplest fix would be to switch the RUNTIMEABORT() in nsX11ErrorHandler to a direct mozalloc_abort().

Comment 2

[Karl Tomlinson (:karlt)]

Attachment: 't add X11 error app notes when they will be added during NS_RUNTIMEABORT

Comment 3

[Karl Tomlinson (:karlt)]

Comment on attachment 514416 [details] [diff] [review]
't add X11 error app notes when they will be added during NS_RUNTIMEABORT

cjones pointed out that NS_RUNTIMEABORT doesn't handle forwarding from plugin processes, so this (on its own, at least) is not good.

Comment 4

[Chris Jones [:cjones] inactive; ni?/f?/r? if you need me]

Need bug 581341 for that.

Comment 5

[Karl Tomlinson (:karlt)]

Attachment: call mozalloc_abort directly to skip NS_RUNTIMEABORT adding X11 error app notes again

As suggested by cjones in comment 1.
This also happens to workaround the double message output of NS_DebugBreak and in debug builds the double stack dump.
It loses the "###!!! ABORT", but I'm not too worried about that.

Comment 6

[Karl Tomlinson (:karlt)]

Attachment: updated patch

Some platforms (maemo, linux64 opt + debug) need the header explicitly included now.

Comment 7

[Karl Tomlinson (:karlt)]

Bug 581341 is making progress and, with that fixed, it would make sense to use NS_RUNTIMEABORT and fix the problems with NS_DebugBreak for all clients.