crash-reporter: Add diagnostics to help diagnose failures in the wild

crash-reporter: Add diagnostics to help diagnose failures in the wild

We add logging messages that are generated during invocation of core2md, but we also enable sending arbitrary system info based on the configuration file.  In this case we add the list of running processes, meminfo, and dmesg.

Change-Id: Ifdf29b89dd60d56349fa39095d2aa07f6b5e2de2

BUG=chromium-os:6299, chromium-os:9345
TEST=UserCrash, unit tests

Committed: http://chrome-svn/viewvc/chromeos?view=rev&revision=fed58af

[petkov, 2011-01-18 19:18:20]

LGTM w/ a couple of nits. I assume somewhere you're including the crash reporter
log into the crash report.

http://codereview.chromium.org/6297004/diff/5001/system_logging.cc
File system_logging.cc (right):

http://codereview.chromium.org/6297004/diff/5001/system_logging.cc#newcode20
system_logging.cc:20: void SystemLoggingImpl::Initialize(const char *ident) {
Maybe reset accumulator_ to NULL so that Initialize can be called multiple
times?

http://codereview.chromium.org/6297004/diff/5001/user_collector.cc
File user_collector.cc (right):

http://codereview.chromium.org/6297004/diff/5001/user_collector.cc#newcode282
user_collector.cc:282: std::string errors_during_fork;
unused?

[kmixter1, 2011-01-22 01:09:38]

PTAL, some significant changes.

[petkov, 2011-01-24 18:36:04]

LGTM w/ a few nits. Btw, if you specify explicitly BUG=chromium-os:..., the
hyperlink will take you to the correct issue tracker.

http://codereview.chromium.org/6297004/diff/23001/crash_collector.cc
File crash_collector.cc (right):

http://codereview.chromium.org/6297004/diff/23001/crash_collector.cc#newcode114
crash_collector.cc:114: open(output_file,O_CREAT | O_WRONLY | O_TRUNC | O_EXCL,
0666));
space after ,

http://codereview.chromium.org/6297004/diff/23001/user_collector.cc
File user_collector.cc (right):

http://codereview.chromium.org/6297004/diff/23001/user_collector.cc#newcode9
user_collector.cc:9: #include <pcrecpp.h>
sort

http://codereview.chromium.org/6297004/diff/23001/user_collector.cc#newcode72
user_collector.cc:72: return StringPrintf("|%s --user=%%p:%%s:%%e",
our_path_.c_str());
I guess no concern about %%e containing spaces?

http://codereview.chromium.org/6297004/diff/23001/user_collector.cc#newcode72
user_collector.cc:72: return StringPrintf("|%s --user=%%p:%%s:%%e",
our_path_.c_str());
add a comment why you're combining the values into a single option?

[kmixter1, 2011-01-25 21:28:07]

pushing

http://codereview.chromium.org/6297004/diff/5001/user_collector.cc
File user_collector.cc (right):

http://codereview.chromium.org/6297004/diff/5001/user_collector.cc#newcode282
user_collector.cc:282: std::string errors_during_fork;
On 2011/01/18 19:18:20, petkov wrote:
> unused?

Done.

http://codereview.chromium.org/6297004/diff/23001/crash_collector.cc
File crash_collector.cc (right):

http://codereview.chromium.org/6297004/diff/23001/crash_collector.cc#newcode114
crash_collector.cc:114: open(output_file,O_CREAT | O_WRONLY | O_TRUNC | O_EXCL,
0666));
On 2011/01/24 18:36:04, petkov wrote:
> space after ,

Done.

http://codereview.chromium.org/6297004/diff/23001/user_collector.cc
File user_collector.cc (right):

http://codereview.chromium.org/6297004/diff/23001/user_collector.cc#newcode9
user_collector.cc:9: #include <pcrecpp.h>
On 2011/01/24 18:36:04, petkov wrote:
> sort

Done.

http://codereview.chromium.org/6297004/diff/23001/user_collector.cc#newcode72
user_collector.cc:72: return StringPrintf("|%s --user=%%p:%%s:%%e",
our_path_.c_str());
On 2011/01/24 18:36:04, petkov wrote:
> I guess no concern about %%e containing spaces?

It's annoying but the kernel doesn't support quoting parameters in core_pattern,
and %%e can contain spaces.  If this occurs, only the part before the space will
be used - and it's only used if the normal way of getting the command line fails
(via proc).  So I'm not too concerned.  Added a comment.

http://codereview.chromium.org/6297004/diff/23001/user_collector.cc#newcode72
user_collector.cc:72: return StringPrintf("|%s --user=%%p:%%s:%%e",
our_path_.c_str());
On 2011/01/24 18:36:04, petkov wrote:
> add a comment why you're combining the values into a single option?

Done.