Malcolm,
Was the build created with -g ? If so, I could log
into your machine and run the task under gdb to track down what is
happening...
David
On 07/06/07, Peter W. Draper <[log in to unmask]> wrote:
> On Thu, 7 Jun 2007, Malcolm J. Currie wrote:
>
> > I've been trying to get to the bottom of a bug for a couple of days
> > arising in AST_TRANN when I try to convert used-supplied polar
> > co-ordinates to the Base Frame to check that they're valid.
> > When the SEGV reproduced with different NDFs and different current WCS
> > frames (AXIS or a SkyFrame) I suspected AST. However, given the
> > supplied mapping and argument values, David doesn't find the crash and
> > obtains sensible transformed co-ordinates.
> >
> > Trying valgrind doesn't reveal (to me) any obvious problem with my code
> > (follows this message). The only possible leaks mention are in AST,
> > and I'm not sure if those are relevant because I had to break in to
> > complete the valgrind o/p.
> >
> > Are there other valgrind tools that I should be using or vital clues
> > that I've missed from valgrind's report?
>
> Malcolm,
>
> the only dodgy bit I could see was this part:
>
> > ==10189== Process terminating with default action of signal 2 (SIGINT)
> > ==10189== at 0x42027896: kill (in /lib/tls/libc-2.3.2.so)
> > ==10189== by 0x52BFEFFF: ???
> > ==10189== by 0x1EAA31C6: astTransform_ (mapping.c:21951)
> > ==10189== by 0x1E94E30E: Transform (cmpmap.c:2714)
>
> valgrind should just exit cleanly, so mentioning this during close (albeit
> after you control-c'd out) is clearly bad, especially since:
>
>
> > Earlier when there was one insignificant initialisation I didn't have to
> > interrupt and valgrind ended its report thus:
>
> >
> > ==23010== Process terminating with default action of signal 11 (SIGSEGV): dumping core
> > ==23010== Bad permissions for mapped region at address 0x8228228
> > ==23010== at 0x1EB47D81: Transform (winmap.c:2321)
> > ==23010== by 0x1EAA31C6: astTransform_ (mapping.c:21951)
> > ==23010== by 0x1E94E30E: Transform (cmpmap.c:2714)
> > ==23010== by 0x1EAA31C6: astTransform_ (mapping.c:21951)
>
> it seems to be related to this issue you saw earlier. Must be some kind of
> smoking gun. I wonder if it might be a compiler issue (rebuild AST with -g
> to check this).
>
> I'm not aware of any other valgrind tools that could help here. So you've
> stuck with the traditional ones (gdb + mark one eyeball).
>
> Peter.
>
--
Note my change of e-mail address. Please send e-mail to
[log in to unmask] from now on.
|