Re: SIGSEGV on gdb 6.7*

[Thiago Jung Bauermann <bauerman@br.ibm.com>]

On Thu, 2008-02-28 at 10:15 +0300, Vladimir Prus wrote:
> This patch cause testsuite regressions for me, as follows:
> 
>         FAIL: gdb.base/annota1.exp: continue to printf
>         FAIL: gdb.base/annota1.exp: send SIGUSR1
>         FAIL: gdb.base/annota1.exp: signal sent

In my experience, this testcase is "nondeterministic", meaning that it
will go back and forth between PASS and FAIL for no good reason.

I keep a list of testcases which I see that have such behaviour:

* gdb.base/annota1.exp 
  * gdb.base/annota2.exp 
  * gdb.base/annota3.exp 
  * gdb.base/checkpoint.exp 
  * gdb.base/display.exp (observed in GDB 6.5, seems fine in 6.6) 
  * gdb.base/interrupt.exp (observed in GDB 6.5, seems fine in 6.6) 
  * gdb.base/multi-forks.exp 
  * gdb.threads/attachstop.exp (observed in GDB 6.5). Test dies with: 
        ERROR: couldn't open "/proc/28696/status": no such file or directory
  * gdb.threads/schedlock.exp 
  * gdb.threads/watchthreads2.exp

Is it useful to have such list on the wiki?
-- 
[]'s
Thiago Jung Bauermann
Software Engineer
IBM Linux Technology Center