segfault during clone via http

segfault during clone via http

[Adam Thorsen]

Git is segfaulting when downloading from a repo via http.  Here's what the
output looks like: http://gist.github.com/58167.  Anybody seen this before or
know how I can debug it?  I've made sure my git (1.6.1) is compiled with libcurl.

Re: segfault during clone via http

[Jay Soffian]

On Wed, Feb 4, 2009 at 11:10 AM, Adam Thorsen <adam.thorsen@gmail.com> wrote:
> Git is segfaulting when downloading from a repo via http.  Here's what the
> output looks like: http://gist.github.com/58167.  Anybody seen this before or
> know how I can debug it?  I've made sure my git (1.6.1) is compiled with libcurl.

Please run git under gdb so that there is a stacktrace to work from. Here is an
example of running git under gdb and providing a trace:

http://article.gmane.org/gmane.comp.version-control.git/107860

j.

Re: segfault during clone via http

[Adam Thorsen]

Here is the stack trace:

Program received signal EXC_BAD_ACCESS, Could not access memory.
Reason: KERN_INVALID_ADDRESS at address: 0x00494000
0x0015bb04 in multi_runsingle ()
(gdb) bt
#0  0x0015bb04 in multi_runsingle ()
#1  0x0015d5af in curl_multi_perform ()
#2  0x000b8782 in step_active_slots () at http.c:450
#3  0x000b881c in run_active_slot (slot=0x50a450) at http.c:474
#4  0x000bac5c in fetch (walker=0x5008d0, 
sha1=0x4a791c "?u^?9l?\003?5??#??X\"?XL") at http-walker.c:846

Re: segfault during clone via http

[Johannes Schindelin]

Hi,

On Wed, 4 Feb 2009, Adam Thorsen wrote:

> 
> Here is the stack trace:
> 
> Program received signal EXC_BAD_ACCESS, Could not access memory.
> Reason: KERN_INVALID_ADDRESS at address: 0x00494000
> 0x0015bb04 in multi_runsingle ()
> (gdb) bt
> #0  0x0015bb04 in multi_runsingle ()
> #1  0x0015d5af in curl_multi_perform ()
> #2  0x000b8782 in step_active_slots () at http.c:450
> #3  0x000b881c in run_active_slot (slot=0x50a450) at http.c:474
> #4  0x000bac5c in fetch (walker=0x5008d0, 
> sha1=0x4a791c "?u^?9l?\003?5??#??X\"?XL") at http-walker.c:846

FWIW I think the issue is related to the double cleanup I mentioned to 
Mike Hommey earlier.  At least when I undo the double cleanup, I do not 
get any segfaults anymore.

Ciao,
Dscho

Re: segfault during clone via http

[Mike Hommey]

On Wed, Feb 04, 2009 at 09:34:36PM +0100, Johannes Schindelin wrote:
> Hi,
> 
> On Wed, 4 Feb 2009, Adam Thorsen wrote:
> 
> > 
> > Here is the stack trace:
> > 
> > Program received signal EXC_BAD_ACCESS, Could not access memory.
> > Reason: KERN_INVALID_ADDRESS at address: 0x00494000
> > 0x0015bb04 in multi_runsingle ()
> > (gdb) bt
> > #0  0x0015bb04 in multi_runsingle ()
> > #1  0x0015d5af in curl_multi_perform ()
> > #2  0x000b8782 in step_active_slots () at http.c:450
> > #3  0x000b881c in run_active_slot (slot=0x50a450) at http.c:474
> > #4  0x000bac5c in fetch (walker=0x5008d0, 
> > sha1=0x4a791c "?u^?9l?\003?5??#??X\"?XL") at http-walker.c:846
> 
> FWIW I think the issue is related to the double cleanup I mentioned to 
> Mike Hommey earlier.  At least when I undo the double cleanup, I do not 
> get any segfaults anymore.

I thought it was not applied, in the end ?

Mike