Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 31 Aug 2004 13:14:32 -0400
From:      John Baldwin <jhb@FreeBSD.org>
To:        freebsd-alpha@FreeBSD.org
Cc:        alpha@FreeBSD.org
Subject:   Re: ok, have kernel, can test.
Message-ID:  <200408311314.32902.jhb@FreeBSD.org>
In-Reply-To: <77903.1093969923@critter.freebsd.dk>
References:  <77903.1093969923@critter.freebsd.dk>

next in thread | previous in thread | raw e-mail | index | archive | help
On Tuesday 31 August 2004 12:32 pm, Poul-Henning Kamp wrote:
> Ok, my 1000 is now able to compile a -current kernel and test-boot it
> while running the 5.0-RC1 kernel.good I saved.
>
> First hurdle is confirmed: hw.eisa_slots must be set to something,
> 5 works fine.
>
> I'm compiling a kernel without fxp now to see if that is the cause of
> the PCI problem.
>
> Ideas and patches are welcome.

Can you break into DDB at all when it hangs?  If so, it might be interesting 
to add KTR to your kernel with KTR_COMPILE and KTR_MASK kernel options set to 
'(KTR_INTR|KTR_PROC)' to track interrupts and process (sic) scheduling 
(mi_switch, sleepq, etc.).  If you can't break into DDB, you can try adding 
KTR_VERBOSE, and capturing a log via serial console.  Note that you'll 
probably have to turn the box off really quick if it's getting an interrupt 
storm since the log will grow very quickly in that case.  (Lots of KTR 
output.. KTR_VERBOSE basically dumps all the output to the screen.  You 
probably don't actually want that on from the beginning, you can maybe hack 
the fxp driver to turn it on by setting the global variable 'ktr_verbose' to 
1 in the bottom of fxp_attach()).

-- 
John Baldwin <jhb@FreeBSD.org>  <><  http://www.FreeBSD.org/~jhb/
"Power Users Use the Power to Serve"  =  http://www.FreeBSD.org



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200408311314.32902.jhb>