Having an invalid stack pointer appears to cause minidump writer to fail

[GoogleCodeExporter]

I'm trying to track down a problem with too-small minidump files for issue 
chromium-os:34880.

From what I could surmise, minidump_writer.cc will bail on writing the minidump 
file if a thread's stack pointer cannot be found in one of the crashing 
process's memory maps.

If this is true, then I think the code should be modified to be fault-tolerant 
with respect to that, so at least we can see the instruction pointer context 
(although that could very well be bogus too :-(), other threads' context, etc.

If anybody disagrees with either my assessment of the problem, or my proposed 
solution, please comment below before I work on it. :)

Original issue reported on code.google.com by mkr...@chromium.org on 29 Sep 2012 at 2:03

[GoogleCodeExporter]

The philosophy we've espoused before is "some data is better than none", so 
this sounds reasonable. Note that MinidumpProcessor will currently give up on 
processing dumps where there's thread memory missing for some threads. I have a 
work-in-progress patch for that up at https://breakpad.appspot.com/413002/.

Original comment by ted.mielczarek on 30 Sep 2012 at 5:13

[GoogleCodeExporter]

It's unfortunate that the processor is not more fault tolerant as well.  But, 
baby steps... :)  This should at least help them get generated, which is still 
useful for us.

There's a CL for this now: https://breakpad.appspot.com/478002/

Original comment by mkr...@chromium.org on 3 Oct 2012 at 11:59

• Changed state: Started

[GoogleCodeExporter]

CL committed: https://breakpad.appspot.com/478002/

Original comment by mkr...@chromium.org on 10 Oct 2012 at 11:29

• Changed state: Fixed

[GoogleCodeExporter]

Original comment by mkr...@chromium.org on 10 Oct 2012 at 11:48