re: port-sparc/45255: GCC 4.5.3 and sparc re-triggers the NULL savefpstate IPI issue

[matthew green <mrg%eterna.com.au@localhost>]

The following reply was made to PR port-sparc/45255; it has been noted by GNATS.

From: matthew green <mrg%eterna.com.au@localhost>
To: Hauke Fath <hf%spg.tu-darmstadt.de@localhost>
Cc: port-sparc-maintainer%NetBSD.org@localhost, 
gnats-admin%NetBSD.org@localhost,
    gnats-bugs%NetBSD.org@localhost
Subject: re: port-sparc/45255: GCC 4.5.3 and sparc re-triggers the NULL 
savefpstate IPI issue
Date: Tue, 16 Aug 2011 04:07:43 +1000

 > At 2:25 Uhr +0000 15.08.2011, mrg%eterna.com.au@localhost wrote:
 > >    use a sparc kernel built with GCC 4.5.3 heavily.  without
 > >    the above change, it should panic fairly often.
 > 
 > ...interesting. I have seen these NULL savefpstate panics on my ss20 (2x
 > SM71) ever since I upgraded to HEAD. Just last weekend, I switched to a UP
 > kernel to at least be able to build a few packages. Should have sent-pr
 > earlier, I guess.
 
 hmmm, i haven't seen those panics for several months, not since i
 made this change:
 
 revision 1.225
 date: 2011/01/22 12:13:25;  author: mrg;  state: Exp;  lines: +3 -3
 convert xpmsg_lock to IPL_SCHED.  the old spl/simple_lock code ran at
 splsched(), and this significantly helps with stability under load
 when running with multiple active CPUs.
 
 > Are you positive this is an exclusively gcc 4.5 related issue?
 
 not any more.  when it fails for you, can you get the backtraces
 of both CPUs and post them?  thanks.
 
 
 .mrg.