releases/2.6.20.2/agp-intel-agp-bugfix.patch

   1 From stable-bounces@linux.kernel.org Sun Feb  4 14:40:37 2007
   2 Date: Sun, 4 Feb 2007 12:18:50 -0500
   3 From: Dave Jones <davej@redhat.com>
   4 To: Eric Piel <Eric.Piel@tremplin-utc.net>
   5 Cc: Wang Zhenyu <zhenyu.z.wang@intel.com>, linux-kernel@vger.kernel.org, Adrian Bunk <bunk@stusta.de>
   6 Subject: AGP: intel-agp bugfix
   7 Message-ID: <20070204171850.GA17171@redhat.com>
   8
   9 From: Dave Jones <davej@redhat.com>
  10
  11 On Sun, Feb 04, 2007 at 04:51:38PM +0100, Eric Piel wrote:
  12  > Hello,
  13  >
  14  > I've got a regression in 2.6.20-rc7 (-rc6 was fine) due to commit
  15  > 4b95320fc4d21b0ff2f8604305dd6c851aff6096 ([AGPGART] intel_agp: restore
  16  > graphics device's pci space early in resume).
  17
  18 I think the key to this failure is the last line here ..
  19
  20  > agpgart-intel 0000:00:00.0: resuming
  21  > PM: Writing back config space on device 0000:00:02.0 at offset f (was 10b, writing 0)
  22  > PM: Writing back config space on device 0000:00:02.0 at offset d (was dc, writing 0)
  23  > PM: Writing back config space on device 0000:00:02.0 at offset b (was 10161025, writing 0)
  24  > PM: Writing back config space on device 0000:00:02.0 at offset 5 (was f4000000, writing 0)
  25  > PM: Writing back config space on device 0000:00:02.0 at offset 4 (was f8000008, writing 0)
  26  > PM: Writing back config space on device 0000:00:02.0 at offset 2 (was 3000011, writing 0)
  27  > PM: Writing back config space on device 0000:00:02.0 at offset 1 (was 2b00007, writing 0)
  28  > PM: Writing back config space on device 0000:00:02.0 at offset 0 (was 11328086, writing 0)
  29  > agpgart: Unable to remap memory.
  30
  31 This then blows up the next access to intel_i810_private.registers, which happens to
  32 be intel_i810_insert_entries.
  33
  34 Either we need .suspend methods which unmap these regions, or we need
  35 to skip trying to map them a second time on resume.
  36
  37 There's an ugly patch below which does the latter. Give it a try?
  38
  39 The intel-agp suspend/resume code has really grown into something
  40 of a monster, and could use some refactoring in a big way.
  41
  42                 Dave
  43
  44
  45 From: Dave Jones <davej@redhat.com>
  46 Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
  47
  48 ---
  49  drivers/char/agp/intel-agp.c |   14 ++++++++------
  50  1 file changed, 8 insertions(+), 6 deletions(-)
  51
  52 --- linux-2.6.20.1.orig/drivers/char/agp/intel-agp.c
  53 +++ linux-2.6.20.1/drivers/char/agp/intel-agp.c
  54 @@ -117,13 +117,15 @@ static int intel_i810_configure(void)
  55
  56         current_size = A_SIZE_FIX(agp_bridge->current_size);
  57
  58 -       pci_read_config_dword(intel_i810_private.i810_dev, I810_MMADDR, &temp);
  59 -       temp &= 0xfff80000;
  60 -
  61 -       intel_i810_private.registers = ioremap(temp, 128 * 4096);
  62         if (!intel_i810_private.registers) {
  63 -               printk(KERN_ERR PFX "Unable to remap memory.\n");
  64 -               return -ENOMEM;
  65 +               pci_read_config_dword(intel_i810_private.i810_dev, I810_MMADDR, &temp);
  66 +               temp &= 0xfff80000;
  67 +
  68 +               intel_i810_private.registers = ioremap(temp, 128 * 4096);
  69 +               if (!intel_i810_private.registers) {
  70 +                       printk(KERN_ERR PFX "Unable to remap memory.\n");
  71 +                       return -ENOMEM;
  72 +               }
  73         }
  74
  75         if ((readl(intel_i810_private.registers+I810_DRAM_CTL)