Re: Regression with commit f9cde5f in 2.6.30-gitX

[Gary Hade <garyhade@us.ibm.com>]

On Wed, Jun 24, 2009 at 02:22:09PM +0200, Ingo Molnar wrote:
> 
> * Jaswinder Singh Rajput <jaswinder@kernel.org> wrote:
> 
> > On Tue, 2009-06-23 at 20:33 -0500, Larry Finger wrote:
> > > My latest pull from Linus's tree fails to boot. Bisection leads to the
> > > commit entitled "x86/ACPI: Correct maximum allowed _CRS returned
> > > resources and warn if exceeded" with hash
> > > f9cde5ffed17bf74f6bef042d99edb0622f58576. I have been unable to
> > > capture the first error message as it scrolls off the screen, but the
> > > second hits the WARN_ON at drivers/ata/ahci.c:695 in routine
> > > ahci_enable_ahci() because HOST_AHCI_EN is not set.
> > > 
> > 
> > This patch fixes boot failure on my AMD 64 laptop, can you please test
> > this patch:
> > 
> > [PATCH] x86: fix _CRS resources return handling
> > 
> > We need to check for info->res_num and only handle for < PCI_BUS_NUM_RESOURCES
> > 
> > Also set info->bus->resource[info->res_num] for _CRS resources return handling
> > 
> > Fixed boot failure on some machine.
> > 
> > Reported-by: Larry Finger <Larry.Finger@lwfinger.net>
> > Signed-off-by: Jaswinder Singh Rajput <jaswinderrajput@gmail.com>
> > ---
> >  arch/x86/pci/acpi.c |   16 ++++++++++------
> >  1 files changed, 10 insertions(+), 6 deletions(-)
> 
> Ah, nice! I suspect it was this upstream commit causing it:
> 
> f9cde5f: x86/ACPI: Correct maximum allowed _CRS returned resources and warn if exceeded
> 
> right?

This change was made in conjunction with the previous
"PCI: use ACPI _CRS data by default" 9e9f46c44e487af0a82eb61b624553e2f7118f5b
change to help expose systems that might be incompatible with
that change.  See http://marc.info/?l=linux-pci&m=124345331521386&w=2
for more discussion on this. 

Gary

-- 
Gary Hade
System x Enablement
IBM Linux Technology Center
503-578-4503  IBM T/L: 775-4503
garyhade@us.ibm.com
http://www.ibm.com/linux/ltc