[pypy-dev] Debugging "cannot find gc roots!"

[Alex Gaynor]

On Mon, Jun 27, 2011 at 3:45 PM, Alex Gaynor <alex.gaynor at gmail.com> wrote:

>
>
> On Sat, Jun 25, 2011 at 1:11 AM, Armin Rigo <arigo at tunes.org> wrote:
>
>> Hi Alex,
>>
>> On Sat, Jun 25, 2011 at 3:11 AM, Alex Gaynor <alex.gaynor at gmail.com>
>> wrote:
>> > In running my application I hit a "cannot find gc roots!", it reproduces
>> > reliably, but it's a pain to isolate because it occurs in a very large
>> > application, running under paste, with threads :(
>>
>> You should rebuild pypy with "make lldebug".  Then gdb should stop as
>> soon as the crash is found.
>>
>>
>> Armin
>>
>
> Hrm, this is pretty strange, running an lldebug build, not under gdb
> reliably produces a crash:
>
> alex at devalex:~/ans/venv/answeb$ ANS_INSTANCE=dwc pypy
> ../lib/pypy/bin/paster serve --reload alex-dev.ini
> Starting subprocess with file monitor
> overriding ans instance based on environment: dwc
> Starting server in PID 6809.
> serving on 0.0.0.0:5050 view at http://127.0.0.1:5050
> cannot find gc roots!
> debug_alloc.h: 9 mallocs left (use PYPY_ALLOC=1 to see the list)
>
> However, running under gdb results in broken pip errors from a socket (this
> is a web app in case that wasn't clear):
>
> alex at devalex:~/ans/venv/answeb$ ANS_INSTANCE=dwc gdb --args pypy
> ../lib/pypy/bin/paster serve --reload alex-dev.ini
> GNU gdb (GDB) 7.1-ubuntu
> Copyright (C) 2010 Free Software Foundation, Inc.
> License GPLv3+: GNU GPL version 3 or later <
> http://gnu.org/licenses/gpl.html>
> This is free software: you are free to change and redistribute it.
> There is NO WARRANTY, to the extent permitted by law.  Type "show copying"
> and "show warranty" for details.
> This GDB was configured as "x86_64-linux-gnu".
> For bug reporting instructions, please see:
> <http://www.gnu.org/software/gdb/bugs/>...
> Reading symbols from /home/alex/ans/venv/bin/pypy...(no debugging symbols
> found)...done.
> (gdb) run
> Starting program: /home/alex/ans/venv/bin/pypy ../lib/pypy/bin/paster serve
> --reload alex-dev.ini
> warning: the debug information found in "/lib/ld-2.11.1.so" does not match
> "/lib/ld-linux.so.2" (CRC mismatch).
>
> warning: the debug information found in "/usr/lib/debug//lib/ld-2.11.1.so"
> does not match "/lib/ld-linux.so.2" (CRC mismatch).
>
> warning: the debug information found in "/usr/lib/debug/lib/ld-2.11.1.so"
> does not match "/lib/ld-linux.so.2" (CRC mismatch).
>
> [Thread debugging using libthread_db enabled]
> Starting subprocess with file monitor
> overriding ans instance based on environment: dwc
> Starting server in PID 6743.
> serving on 0.0.0.0:5050 view at http://127.0.0.1:5050
> ----------------------------------------
> Exception happened during processing of request from ('10.160.27.96',
> 44587)
> Traceback (most recent call last):
>   File "/home/alex/ans/venv/lib/pypy/site-packages/paste/httpserver.py",
> line 1068, in process_request_in_thread
>     self.finish_request(request, client_address)
>   File "/home/alex/ans/venv/lib/pypy/lib-python/2.7/SocketServer.py", line
> 323, in finish_request
>     self.RequestHandlerClass(request, client_address, self)
>   File "/home/alex/ans/venv/lib/pypy/lib-python/2.7/SocketServer.py", line
> 641, in __init__
>     self.finish()
>   File "/home/alex/ans/venv/lib/pypy/lib-python/2.7/SocketServer.py", line
> 694, in finish
>     self.wfile.flush()
>   File "/home/alex/ans/venv/lib/pypy/lib-python/modified-2.7/socket.py",
> line 348, in flush
>     self._sock.sendall(view[write_offset:write_offset+buffer_size])
>   File "/home/alex/ans/venv/lib/pypy/lib-python/modified-2.7/socket.py",
> line 266, in sendall
>     self._sock.sendall(data, flags)
> error: [Errno 32] Broken pipe
> ----------------------------------------
> ----------------------------------------
> Exception happened during processing of request from ('10.160.27.96',
> 44608)
> Traceback (most recent call last):
>   File "/home/alex/ans/venv/lib/pypy/site-packages/paste/httpserver.py",
> line 1068, in process_request_in_thread
>     self.finish_request(request, client_address)
>   File "/home/alex/ans/venv/lib/pypy/lib-python/2.7/SocketServer.py", line
> 323, in finish_request
>     self.RequestHandlerClass(request, client_address, self)
>   File "/home/alex/ans/venv/lib/pypy/lib-python/2.7/SocketServer.py", line
> 641, in __init__
>     self.finish()
>   File "/home/alex/ans/venv/lib/pypy/lib-python/2.7/SocketServer.py", line
> 694, in finish
>     self.wfile.flush()
>   File "/home/alex/ans/venv/lib/pypy/lib-python/modified-2.7/socket.py",
> line 348, in flush
>     self._sock.sendall(view[write_offset:write_offset+buffer_size])
>   File "/home/alex/ans/venv/lib/pypy/lib-python/modified-2.7/socket.py",
> line 266, in sendall
>     self._sock.sendall(data, flags)
> error: [Errno 32] Broken pipe
> ----------------------------------------
> ----------------------------------------
>
>
> The only thing worth noting is that this a 32-bit PyPy on a 64-bit OS (the
> gdb is also 64-bits, so if that matters the gdb runs may not be valid).  I'm
> not sure whether I see the socket issue with gdb because its dieing earlier
> than the gc roots issue, or if it's because the error doesn't appear for
> some reason, and therefore it gets farther and hits a second issue.
>
>
> Alex
>
>
> --
> "I disapprove of what you say, but I will defend to the death your right to
> say it." -- Evelyn Beatrice Hall (summarizing Voltaire)
> "The people's good is the highest law." -- Cicero
>
>
FWIW under shadowstack I'm seeing the broken pipe error as well, so I guess
gdb is just making the error not happen.

Alex

-- 
"I disapprove of what you say, but I will defend to the death your right to
say it." -- Evelyn Beatrice Hall (summarizing Voltaire)
"The people's good is the highest law." -- Cicero
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/pypy-dev/attachments/20110627/5f7bf4d8/attachment-0001.html>