Re: threads/1393: FAIL: gdb.threads/schedlock.exp: continue to breakpoint: return to loop (initial)

[Daniel Jacobowitz <drow at mvista dot com>]

The following reply was made to PR threads/1393; it has been noted by GNATS.

From: Daniel Jacobowitz <drow@mvista.com>
To: gdb-gnats@sources.redhat.com
Cc:  
Subject: Re: threads/1393: FAIL: gdb.threads/schedlock.exp: continue to breakpoint: return to loop (initial)
Date: Thu, 25 Sep 2003 00:13:15 -0400

 On Thu, Sep 25, 2003 at 02:38:21AM -0000, mec@shout.net wrote:
 > 
 > >Number:         1393
 > >Category:       threads
 > >Synopsis:       FAIL: gdb.threads/schedlock.exp: continue to breakpoint: return to loop (initial)
 > >Confidential:   no
 > >Severity:       serious
 > >Priority:       medium
 > >Responsible:    unassigned
 > >State:          open
 > >Class:          sw-bug
 > >Submitter-Id:   net
 > >Arrival-Date:   Thu Sep 25 02:48:00 UTC 2003
 > >Closed-Date:
 > >Last-Modified:
 > >Originator:     mec@shout.net
 > >Release:        gdb 5.3.92
 > >Organization:
 > >Environment:
 > target=native
 > host=i686-pc-linux-gnu
 > osversion=red-hat-8.0
 > gdb=5.3.92
 > gcc=2.95.3
 > binutils=HEAD 2003-09-22 08:09:17 UTC
 > glibc=2.2.93-5-rh
 > gformat=dwarf-2
 > glevel=2
 > >Description:
 > 
 > schedlock.exp goes off the rails here:
 > 
 >           (gdb) break 41 if arg != 5
 >           Breakpoint 3 at 0x8048590: file /berman/migchain/source/gdb-5.3.92/gdb/testsuite/gdb.threads/schedlock.c, line 41.
 >           (gdb) continue
 >           Continuing.
 > 
 >           Program received signal SIGINT, Interrupt.
 >           [Switching to Thread 24580 (LWP 17457)]
 >           0x42028c82 in sigprocmask () from /lib/i686/libc.so.6
 >           (gdb) FAIL: gdb.threads/schedlock.exp: continue to breakpoint: return to loop (initial)
 >           ... many more FAILS ...
 > 
 > This has happened 3 times in a population of 1666 test runs.  The three dates were 2003-08-07, 2003-08-28, and 2003-09-22.  I keep the gdb.log files for six months in case they are needed.
 > 
 > This happens with both dwarf-2 and stabs+, and both gcc 2 and gcc 3.
 > 
 > This is probably going to go on the "unexplained mysteries" pile for a couple of years.  I'm just filign the PR to capture the information that I have.
 > >How-To-Repeat:
 > Run schedlock.exp 2000 times.
 
 I think this is just unavoidable.  I committed a patch to solve a
 similar, much more extreme issue under 2.6 kernels, but the same
 problem can happen anywhere else.  It's caused by the fact that SIGINT
 is actually sent to every thread.  GDB tries its best to catch all of
 the sigint's, but sometimes it can't.
 
 Someday, it will go away.  NPTL does not suffer from this quirk.
 
 -- 
 Daniel Jacobowitz
 MontaVista Software                         Debian GNU/Linux Developer