[prev in list] [next in list] [prev in thread] [next in thread] 

List:       kde-bugs-dist
Subject:    [Bug 193917] valgrind fails to run on Mac OS 10.5.7, gets a SIGTRAP
From:       Nicholas Nethercote <njn () valgrind ! org>
Date:       2009-05-29 21:29:23
Message-ID: 20090529212923.A0E3411F59 () immanuel ! kde ! org
[Download RAW message or body]

https://bugs.kde.org/show_bug.cgi?id=193917


Nicholas Nethercote <njn@valgrind.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |gparker@apple.com




--- Comment #21 from Nicholas Nethercote <njn valgrind org>  2009-05-29 23:29:18 ---
I asked Greg Parker about this, he had some comments...

----
Off the top of my head:
* valgrind does run a copy of dyld for itself, but iirc the valgrind launcher
hides DYLD_* environment variables from V's dyld and restores them for the
inferior's dyld. I wouldn't expect DYLD_PRINT_STATISTICS to print anything for
valgrind itself.
* gdb is confused by valgrind; it sees only the inferior's symbols. To get
valgrind's own symbols in the backtrace, run `add-symbol-file
/path/to/memcheck-<arch>-darwin`.
* CrashReporter can be similarly confused, but usually it doesn't fail as badly
as shown.
* A breakpoint on __start might not work. Historically gdb has had trouble with
breakpoints set before the gdb/dyld handshake.
* The 0xfffffffe frame in the backtrace might indicate the arrival of a signal,
and the 0xf0ad92a0 frame is probably in valgrind's own code. Try looking for a
__builtin_trap() from a valgrind signal handler.
----

With respect to the last point, there were some __builtin_trap calls in Greg's
original port, but they've since been removed.

-- 
Configure bugmail: https://bugs.kde.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are watching all bug changes.

[prev in list] [next in list] [prev in thread] [next in thread]