Add stack contents output functionality to minidump_stackwalk

Add stack contents output functionality to minidump_stackwalk

This feature is enabled only when "-s" is provided as a commandline option.

minidump_stackwalk.cc:
 - Add a new commandline option "-s" to output stack contents.
 - Instantiate Minidump object in PrintMinidumpProcess() to keep it alive longer
so that accessing process_state.thread_memory_regions() in stackwalk_common.cc
doesn't result in use-after-free.

stackwalk_common.cc:
 - Add a new function PrintStackContents() to output stack contents.

R=mark@chromium.org

Committed: https://code.google.com/p/google-breakpad/source/detail?r=1429

[hashimoto]

Hi Mark,

I find it very convenient to have functionality to glance the state of stack
which allows people to quickly find URLs or file names.
Also, collecting function pointers and showing the function name would be
helpful for debugging.

Do you mind having this change being checked in?
This change doesn't affect any existing feature except minidump_stackwalk being
run with the new commandline option "-s".

The output format is something like this:

$minidump_stackwalk -s minidump.dmp ~/breakpad
...
 2  chrome!base::debug::TaskAnnotator::RunTask [callback.h : 396 + 0x5]
     r4 = 0xdeadbeaf    r5 = 0xdeadbeaf    r6 = 0xdeadbeaf    r7 = 0xdeadbeaf
     r8 = 0xdeadbeaf    r9 = 0xdeadbeaf   r10 = 0xdeadbeaf    fp = 0xdeadbeaf
     sp = 0xdeadbeaf    pc = 0xdeadbeaf
    Stack contents:
     0xdeadbea0 61 62 63 64 65 66 67 68 69 6a 6b 6c 6d 6e 6f 70 
abcdefghijklmnop
     0xdeadbeb0 00 00 00 00 00 00 00 00                          ........
    Possible instruction pointers:
     *(0xdeadbeaf) = 0xdeadbeaf
<net::SSLClientSocketNSS::Core::BufferRecvComplete> [ssl_client_socket_nss.cc :
2763 + 0x1]
     *(0xdeadbeaf) = 0xdeadbeaf
<net::SSLClientSocketNSS::Core::BufferRecvComplete> [ssl_client_socket_nss.cc :
2777 + 0x1]
     *(0xdeadbeaf) = 0xdeadbeaf <ProfileIOData::ShutdownOnUIThread>
[profile_io_data.cc : 1274 + 0x1]
     *(0xdeadbeaf) = 0xdeadbeaf <net::TCPClientSocket::DidCompleteReadWrite>
[tcp_client_socket.cc : 332 + 0x1]
     *(0xdeadbeaf) = 0xdeadbeaf <net::TCPClientSocket::DidCompleteReadWrite>
[tcp_client_socket.cc : 330 + 0x1]
     *(0xdeadbeaf) = 0xdeadbeaf <base::MessageLoop::DoWork> [stl_vector.h : 728
+ 0x1]
    Found by: call frame info
...

[Mark Mentovai]

This is fine. LGTM

https://breakpad.appspot.com/9774002/diff/6/src/processor/stackwalk_common.cc
File src/processor/stackwalk_common.cc (right):

https://breakpad.appspot.com/9774002/diff/6/src/processor/stackwalk_common.cc...
src/processor/stackwalk_common.cc:611: printf("\n    Found by: %s\n",
frame->trust_description().c_str());
I would put this before the stack contents.

[Ted Mielczarek]

FYI I have a similar tool that I've used quite a bit in situations where the
stack is trashed:
http://hg.mozilla.org/users/tmielczarek_mozilla.com/dump-lookup/

It doesn't print the raw stack bytes but only things that look like functions.

[hashimoto]

On 2015/02/26 17:11:20, Ted Mielczarek wrote:
> FYI I have a similar tool that I've used quite a bit in situations where the
> stack is trashed:
> http://hg.mozilla.org/users/tmielczarek_mozilla.com/dump-lookup/
> 
> It doesn't print the raw stack bytes but only things that look like functions.

Thanks for the pointer.
Please let me know if you find any possibly useful future missing.

[hashimoto]

https://breakpad.appspot.com/9774002/diff/6/src/processor/stackwalk_common.cc
File src/processor/stackwalk_common.cc (right):

https://breakpad.appspot.com/9774002/diff/6/src/processor/stackwalk_common.cc...
src/processor/stackwalk_common.cc:611: printf("\n    Found by: %s\n",
frame->trust_description().c_str());
On 2015/02/26 13:48:44, Mark Mentovai wrote:
> I would put this before the stack contents.

Done.

[hashimoto]

Committed patchset #2 (id:30003) manually as r1429 (presubmit successful).

[Ted Mielczarek]

FYI I tried this out locally and wound up pushing a little followup patch to fix
the formatting:
https://code.google.com/p/google-breakpad/source/detail?r=1430

It was missing a newline at the end of the "Possible instruction pointers" list,
so the next frame would print at the end of the line like this:
"""
    Possible instruction pointers:
     *(0xbffaa93c) = 0xb778dc3e <mozalloc_abort>
[mozalloc_abort.cpp:ea144003b87f : 20 + 0x0]
     *(0xbffaa940) = 0xb778dc45 <mozalloc_abort>
[mozalloc_abort.cpp:ea144003b87f : 20 + 0x7]
     *(0xbffaa94c) = 0xb5a4eba6 <NS_DebugBreak_P> [nsDebugImpl.cpp:c4fe2b219941
: 386 + 0x0] 1  libxul
.so!NS_DebugBreak_P [nsDebugImpl.cpp:c4fe2b219941 : 423 + 0x5]
"""

With the change I pushed it prints like this:
"""
    Possible instruction pointers:
     *(0xbffaa93c) = 0xb778dc3e <mozalloc_abort>
[mozalloc_abort.cpp:ea144003b87f : 20 + 0x0]
     *(0xbffaa940) = 0xb778dc45 <mozalloc_abort>
[mozalloc_abort.cpp:ea144003b87f : 20 + 0x7]
     *(0xbffaa94c) = 0xb5a4eba6 <NS_DebugBreak_P> [nsDebugImpl.cpp:c4fe2b219941
: 386 + 0x0]

 1  libxul.so!NS_DebugBreak_P [nsDebugImpl.cpp:c4fe2b219941 : 423 + 0x5]
"""

[hashimoto]

On 2015/02/27 13:43:22, Ted Mielczarek wrote:
> FYI I tried this out locally and wound up pushing a little followup patch to
fix
> the formatting:
> https://code.google.com/p/google-breakpad/source/detail?r=1430
> 
> It was missing a newline at the end of the "Possible instruction pointers"
list,
> so the next frame would print at the end of the line like this:
> """
>     Possible instruction pointers:
>      *(0xbffaa93c) = 0xb778dc3e <mozalloc_abort>
> [mozalloc_abort.cpp:ea144003b87f : 20 + 0x0]
>      *(0xbffaa940) = 0xb778dc45 <mozalloc_abort>
> [mozalloc_abort.cpp:ea144003b87f : 20 + 0x7]
>      *(0xbffaa94c) = 0xb5a4eba6 <NS_DebugBreak_P>
[nsDebugImpl.cpp:c4fe2b219941
> : 386 + 0x0] 1  libxul
> .so!NS_DebugBreak_P [nsDebugImpl.cpp:c4fe2b219941 : 423 + 0x5]
> """
> 
> With the change I pushed it prints like this:
> """
>     Possible instruction pointers:
>      *(0xbffaa93c) = 0xb778dc3e <mozalloc_abort>
> [mozalloc_abort.cpp:ea144003b87f : 20 + 0x0]
>      *(0xbffaa940) = 0xb778dc45 <mozalloc_abort>
> [mozalloc_abort.cpp:ea144003b87f : 20 + 0x7]
>      *(0xbffaa94c) = 0xb5a4eba6 <NS_DebugBreak_P>
[nsDebugImpl.cpp:c4fe2b219941
> : 386 + 0x0]
> 
>  1  libxul.so!NS_DebugBreak_P [nsDebugImpl.cpp:c4fe2b219941 : 423 + 0x5]
> """

Oops, I was too careless when changing the position of the "Found by" line.
Thank you for fixing it.