Date: Thu, 18 Feb 1999 22:55:47 +0000 (GMT) From: Doug Rabson <dfr@nlsystems.com> To: Matthew Jacob <mjacob@feral.com> Cc: freebsd-hackers@freebsd.org Subject: Re: Panic in FFS/4.0 as of yesterday Message-ID: <Pine.BSF.4.05.9902182249240.61913-100000@herring.nlsystems.com> In-Reply-To: <Pine.LNX.4.04.9902180817090.22368-100000@feral-gw>
next in thread | previous in thread | raw e-mail | index | archive | help
On Thu, 18 Feb 1999, Matthew Jacob wrote: > > I started testing again to see what I could see. This is an alpha platform > (*really* shouldn't matter, right? this ain't linux, this ain't no disco, > pal....) > > Simple hardware. Alpha PC164 (432Mhz) with 256MB memory. > > Again, a very simple setup- just those stupid dirty buffer testing > programs. A single 9GB filesystem (not root or swap). Softupdates not > enabled, buf realloc still enabled (the 'default'- if it's broken, why > ship with it enabled?). 100 instances of programs that write 150mb files > and then check them one by one. The filesystem was nothing unusual- > following concerns of others about CCD's this is wasn't even an CCD. > Following concerns about fragsize and blocksize this was a Frag 1024 > Blocksize 8192 (8192 is pagesize for alpha), and the actual size was > truncated to make an exact geometry fit. > > System was completely unresponsive after starting this test. It eventually > (I went home) started being responsive again since a 'sync' completed > (thisis a usability problem, but I'l start whining about that when the > system stays up long enough to be usable). > > When I came in this morning, it was in DDB with: > > panic: getnewbuf: cannot get buffer, infinite recursion failure I looked at this and I think this writerecursion test looks bogus (it certainly isn't recursing although it has obviously been entered by several independant process simultaneously. Could you send me the test program which makes this happen (alpha binary or source will do) and I'll try to reproduce it here on a spare 4G disk. It seems to me that the write recursion test should be done on a per-process basis. Perhaps the recursion count should be a field in struct proc. Sounds ugly though. -- Doug Rabson Mail: dfr@nlsystems.com Nonlinear Systems Ltd. Phone: +44 181 442 9037 To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-hackers" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.BSF.4.05.9902182249240.61913-100000>