Re: recent commits break via 8235 ata

On Tuesday 10 April 2007 16:13, Nate Lawson wrote:
> Giorgos Keramidas wrote:
> > On 2007-03-27 12:27, John Baldwin <jhb_at_freebsd.org> wrote:
> >> If that is the case it's because code was using
> >> rman_get_bus(handle|tag) on a resource that wasn't activated yet which
> >> wouldn't have worked before the nexus changes either.  Well, the bus
> >> tag might have been right, but the handle for SYS_RES_MEMORY would
> >> have been wrong.
> > 
> > This is the change which stops snd_hda from working here:
> > 
> > % Date: Wed, 21 Mar 2007 15:39:12 +0000 (UTC)
> > % Message-Id: <200703211539.l2LFdCQW036704_at_repoman.freebsd.org>
> > % From: John Baldwin <jhb_at_freebsd.org>
> > % Subject: cvs commit: src/sys/dev/acpica acpi.c
> > % To: src-committers_at_freebsd.org, cvs-src_at_freebsd.org, cvs-all_at_freebsd.org
> > % Cc:
> > %
> > % jhb         2007-03-21 15:39:12 UTC
> > %
> > %   FreeBSD src repository
> > %
> > %   Modified files:
> > %     sys/dev/acpica       acpi.c
> > %   Log:
> > %   Change acpi's handling of suballocating system resources to be a little
> > %   simpler.  It now can just use rman_is_region_manager() during
> > %   acpi_release_resource() to see if the the resource is suballocated from
> > %   a system resource.  Also, the driver no longer needs MD knowledge about
> > %   how to setup bus space tags and handles when doing a suballocation, but
> > %   can simply rely on bus_activate_resource() in the parent setting all that
> > %   up.
> > %
> > %   Revision  Changes    Path
> > %   1.233     +39 -55    src/sys/dev/acpica/acpi.c
> > 
> > If I update my kernel sources to Wed Mar 21 14:39:39 2007 +0000
> > (including the sys/modules/padlock/Makefile commit of Sam Leffler, to
> > fix the kernel build from Wed Mar 21 17:37:13 2007 +0000), I can see
> > that the snd_hda driver probes my sound card correctly.
> > 
> > Updating after the commit shown above, breaks snd_hda.
> 
> I just finished a binsearch and can verify this exact commit causes the
> problem for my ATA also.  I'll spend some more time debugging it later.
> 
> John, can I back out rev 1.233 of acpi.c until you return since it
> prevents some machines from booting?

Well, it most likely means there are other bugs that just happen to not break
with the older stuff.  Are you sure that just this one change fixes things rather
than the changes to the nexus drivers?  If so, can you get a verbose dmesg both
with and without?  Adding some printfs to see what the start and end arguments
are in acpi_alloc_resource() might help, too.

-- 
John Baldwin