124891 - Random crashes in CrawlStack when doing JProf's

Status: RESOLVED INCOMPLETE

(Testing :: General, defect, P3)

Description

[Randell Jesup [:jesup] (needinfo me)]

mozilla 20020211xx pull/build, -O2, Linux RH 7.1, dual-450's

(where do tools like jprof get reported to (Product/component)?  It's
not a "web-based tool"...)

I crash semi-randomly when doing jprofs.  It's repeatable at the URL
above within a minute of just watching the background change.  To repeat:

setenv JPROF_FLAGS "JP_DEFER JP_PERIOD=0.0015"
./mozilla &
./jprofsig start
<browse to the URL above and wait, or randomly browse the web for a fair
while>
crash


GDB backtrace:
#0  0x40220c1a in CrawlStack ()
   from
/home/jesup/src/mozilla_trunk/mozilla/obj-i686-pc-linux-gnu/dist/bin/libjprof.so
(gdb) bt
#0  0x40220c1a in CrawlStack ()
   from
/home/jesup/src/mozilla_trunk/mozilla/obj-i686-pc-linux-gnu/dist/bin/libjprof.so
#1  0x40220d60 in Log ()
   from
/home/jesup/src/mozilla_trunk/mozilla/obj-i686-pc-linux-gnu/dist/bin/libjprof.so
#2  0x40220ebc in StackHook ()
   from
/home/jesup/src/mozilla_trunk/mozilla/obj-i686-pc-linux-gnu/dist/bin/libjprof.so
#3  0x40210b13 in pthread_sighandler_rt (signo=27, si=0xbfffdce0, 
    uc=0xbfffdd60) at signals.c:121
#4  <signal handler called>
#5  0x40564ef5 in strchr () from /lib/i686/libc.so.6
#6  0xbfffe088 in ?? ()
Error accessing memory address 0xff010101: No such process.

Comment 1

[David Baron :dbaron: (⌚️UTC-5, no longer working on Mozilla)]

Can you at least get disassembly / registers of the crash from gdb?  Any idea if
we've walked off the end of the stack into garbage or if the stack is corrupted?

Putting this bug on my list for now since I'm sure Asa doesn't want it.

Comment 2

[Randell Jesup [:jesup] (needinfo me)]

Attachment: register and assembler dump of crash

Comment 3

[Andrew Schultz]

Attachment: stacktrace with symbols

I get this consistently when locally loading the testcase from bug 109885

Comment 4

[jim_nance]

mozilla/tools/jprof does not seem to show up under lxr.  I guess that is anoterh
bug :-(

Comment 5

[Andrew Schultz]

WFM: http://lxr.mozilla.org/mozilla/source/tools/jprof/

Comment 6

[Michael Kohler [:mkohler]]

This is a mass change. Every comment has "assigned-to-new" in it.

I didn't look through the bugs, so I'm sorry if I change a bug which shouldn't be changed. But I guess these bugs are just bugs that were once assigned and people forgot to change the Status back when unassigning.

Comment 7

[Robert Kaiser]

Is this still relevant? If so, do we have crash signatures?
If it's not relevant any more, can we close it out?

Comment 8

[Randell Jesup [:jesup] (needinfo me)]

Too old, and too many changes for this bug to be relevant.