Date: Mon, 26 Jan 2004 10:16:07 -0500 (EST) From: Robert Watson <rwatson@freebsd.org> To: Jun Su <csujun@263.net> Cc: current@freebsd.org Subject: Re: [PANIC] today's cvs with KTRACE and MUTEX_PROFILING enabled (Big Post) Message-ID: <Pine.NEB.3.96L.1040126101328.11399E-100000@fledge.watson.org> In-Reply-To: <009501c3e41e$9e5ea360$90dea1d3@shasujunmv>
next in thread | previous in thread | raw e-mail | index | archive | help
On Mon, 26 Jan 2004, Jun Su wrote:
> > On Mon, 26 Jan 2004, Jun Su wrote:
> >
> > > I got the following panic when booting. Seems the if_data_mutex is
> > > corrupt by something.If you need any information, please feel free to
> > > let me know. Thanks.
> >
> > Are you running with kernel modules? If so, were all of them built as
> > part of your kernel build? Turning on mutex profiling changes the size of
> > the mutex structure, so if you have modules that aren't built to be aware
> > of mutex profiling, you will get memory corruption. When using mutex
> > profiling, I would generally suggest avoiding the use of kernel modules,
> > since it avoids the whole issue...
> Yes. This is the root cause. Our module building env doesn't include
> opt_global.h. Then the module's mtx doesn't match the one in Kernel. I think
> this is a requirement for MUTEX_PROFILING, not a suggestion before we
> complete the new building env for kernel module. I suggest to add this to
> man page. What is your option?
The following text appears in the NOTES section of MUTEX_PROFILING.9:
The MUTEX_PROFILING option increases the size of struct mtx, so a
kernel built with that option will not work with modules built
without it.
We can probably make this a little more explicit by specifically
mentioning opt_global.h.
Robert N M Watson FreeBSD Core Team, TrustedBSD Projects
robert@fledge.watson.org Senior Research Scientist, McAfee Research
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.NEB.3.96L.1040126101328.11399E-100000>