Date: Mon, 4 Mar 2002 15:52:40 -0800 (PST) From: Matthew Dillon <dillon@apollo.backplane.com> To: "Jan L. Peterson" <jlp@softhome.net> Cc: Ian Dowse <iedowse@maths.tcd.ie>, freebsd-stable@FreeBSD.ORG Subject: Re: crashes on 4.5-RELEASE Message-ID: <200203042352.g24Nqew91259@apollo.backplane.com> References: <200202232021.aa04276@salmon.maths.tcd.ie> <20020304233509.5BFD0422D8@mail.flipdog.com>
next in thread | previous in thread | raw e-mail | index | archive | help
:Okay, to resurrect a dead horse, the last thing I got before leaving
:the office for a week was from Ian Dowse:
:
:> Try adding "options INVARIANTS" and "options INVARIANT_SUPPORT" to
:> your kernel config
:
:Ian also suggested a patch to mbuf.h, but I'll get to that in a minute.
:
:So I turned on "options INVARIANTS" and "options INVARIANT_SUPPORT",
:rebuilt my kernel, rebooted, and now, tah-dah, it doesn't crash (at
:least, not yet). I did get an odd "xl0: watchdog timeout" or two, but
:the system is still running after 8 hours of heavy NFS traffic (would
:have killed it in minutes before).
:
:Interesting, ehh?
:
:Ian, I didn't apply the patch to mbuf.h yet (added the options from
:memory and didn't remember the patch until I went back and looked at
:your mail just now). If it panics, I'll put the patch in and try
:again.
:
:So what exactly does INVARIANTS do? According to LINT, it enables
:extra sanity checking of internal structures (making the system
:slower).
:
:Seems to be a lot of people reporting crashing problems on 4.5, even on
:hardware that was previously stable. Could be we have a real problem
:here?
:
: -jan-
:--
:Jan L. Peterson
:<jlp@softhome.net>
Well, INVARIANTS is supposed to add extra sanity checks... it should
make the machine crash *more* often rather then *less* often!
Is this the same machine that was crashing on 21 Feb? Here's part
of that message:
:xl_newbuf(c15b8000,c15b844c)
:xl_rxeof(c15b8000)
:xl_intr(c15b8000,660400,c690a834,10000,cdf1fd48)
:intr_mux(c0e356c0,0,c6900010,10,c0190010)
:Xresume10()
:--- interrupt, eip = 0xc02afd22, esp = 0xcdf1fd08, ebp = 0xcdf1fd48 --
:vec10(ce0b7000,8,0,0,0))
:nfs_getcacheblk(ce0b7000,8,0,cdef1100,ce0b7000)
:nfs_write(cdf1fe64,c17d45c0,cdef1100,1000,c031cd00)
:vn_write(c17d45c0,cdf1fed4,c17ae200,0,cdef1100)
:dofilewrite(cdef1100,c17d45c0,4,8053f60,1000)
:write(cdef1100,cdf1ff80,1000,6,4)
:syscall(2f,2f,2f,4,6)
:Xint0x80_syscall()
:
:pretty consistent now... I can make it crash in less than five minutes
:after booting.
:
: -jan-
Do you happen to have a kernel.debug and crash dump from one of
these? Were the consistent crashes all dying in the same place
when you did the DDB> traceback?
-Matt
Matthew Dillon
<dillon@backplane.com>
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-stable" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200203042352.g24Nqew91259>