Re: testsuite errors on Red Hat 8.0

[Daniel Jacobowitz <drow@mvista.com>]

On Wed, Apr 16, 2003 at 10:26:18AM -0400, Elena Zannoni wrote:
> Daniel Jacobowitz writes:
>  > On Tue, Apr 15, 2003 at 09:27:32PM -0700, David Carlton wrote:
>  > > On Wed, 16 Apr 2003 00:11:46 -0400, Daniel Jacobowitz <drow@mvista.com> said:
>  > > 
>  > > >> and in gdb.threads/pthreads.exp, I get:
>  > > >> 
>  > > >> FAIL: gdb.threads/pthreads.exp: check backtrace from main thread
>  > > >> FAIL: gdb.threads/pthreads.exp: apply backtrace command to all three threads
>  > > 
>  > > > These I'd definitely like to see the log for.
>  > > 
>  > > I'll put it after my signature.
>  > 
>  > 
>  > > thread apply 1 backtrace
>  > > 
>  > > Thread 1 (Thread 16384 (LWP 14967)):
>  > > #0  0x4013a816 in nanosleep () from /lib/libc.so.6
>  > > #1  0x00000000 in ?? ()
>  > > 51        if (verbose) printf("common_routine (%d)\n", arg);
>  > > (gdb) FAIL: gdb.threads/pthreads.exp: check backtrace from main thread
>  > 
>  > We can no longer backtrace out of nanosleep on your system, if we ever
>  > could.
>  > 
>  > The possible fixes:
>  >   - adding unwind info to glibc syscall stubs and making gdb use it
>  >   - synthesizing unwind info when we recognize a syscall stub and using
>  >     it
>  >   - improving the prologue analyzer until it works for this function
> 
> 
> he he he...See my posting about how bad this problem is going to bite
> us for NPTL kernels. About this particual problem, RH 8.0 doesn't have
> NPTL, but there could be some other interesting kernel 'feature' going
> on.

Fortunately, IIRC, Ulrich has added unwind info to glibc for syscall
wrappers in glibc now?  Next step is to use it!

-- 
Daniel Jacobowitz
MontaVista Software                         Debian GNU/Linux Developer