Patch to allow stackwalker to skip threads with zero-sized stack

[GoogleCodeExporter]

We got a minidump from a customer that contained several threads with zero-size 
stacks. When 
our crashreporter system tried to resolve the symbols, stackwalker would always 
abort with an 
error, although the other threads contained valid information.
Enclosed is a patch that will allow stackwalker to continue when it hits 
threads with zero-sized 
stacks. In the final output, the threads will be listed, but of course, no 
backtrace is printed for those 
threads. 

I don't know whether there are other users of the minidump processors that may 
require the 
processor to fail for zero-sized stacks. In this case, the new behavior would 
need to be made 
optional, but overall, I think stackwalk, crash_report et al should be smarter 
about minidumps with 
zero-sized stacks, because there may still be valid and important information 
for the other threads.

Original issue reported on code.google.com by j...@mac.com on 22 Apr 2009 at 8:35

Attachments:

• stackwalker.patch

[GoogleCodeExporter]

Sorry, forgot to mention: this patch is based on rev 328

Original comment by j...@mac.com on 22 Apr 2009 at 8:38

[GoogleCodeExporter]

I'm not sure that we'd take this patch as written, but the concept sounds good. 
I
didn't look at it too deeply though.

Original comment by ted.mielczarek on 3 Dec 2009 at 4:24

[GoogleCodeExporter]

I independently recreated this patch here:
https://breakpad.appspot.com/413002/

Original comment by ted.mielczarek on 18 Jul 2012 at 1:22

[GoogleCodeExporter]

Issue 342 has been merged into this issue.

Original comment by ted.mielczarek on 18 Jul 2012 at 1:23

[GoogleCodeExporter]

This was fixed in r1077.

Original comment by ted.mielczarek on 10 Jan 2014 at 1:36

• Changed state: Fixed