x86/iommu/vt-d: Expand interrupt remapping quirk to cover x58 chipset
authorNeil Horman <nhorman@tuxdriver.com>
Wed, 17 Jul 2013 11:13:59 +0000 (07:13 -0400)
committerIngo Molnar <mingo@kernel.org>
Tue, 23 Jul 2013 09:29:30 +0000 (11:29 +0200)
Recently we added an early quirk to detect 5500/5520 chipsets
with early revisions that had problems with irq draining with
interrupt remapping enabled:

  commit 03bbcb2e7e292838bb0244f5a7816d194c911d62
  Author: Neil Horman <nhorman@tuxdriver.com>
  Date:   Tue Apr 16 16:38:32 2013 -0400

      iommu/vt-d: add quirk for broken interrupt remapping on 55XX chipsets

It turns out this same problem is present in the intel X58
chipset as well. See errata 69 here:

  http://www.intel.com/content/www/us/en/chipsets/x58-express-specification-update.html

This patch extends the pci early quirk so that the chip
devices/revisions specified in the above update are also covered
in the same way:

Signed-off-by: Neil Horman <nhorman@tuxdriver.com>
Reviewed-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Donald Dutile <ddutile@redhat.com>
Cc: Joerg Roedel <joro@8bytes.org>
Cc: Andrew Cooper <andrew.cooper3@citrix.com>
Cc: Malcolm Crossley <malcolm.crossley@citrix.com>
Cc: Prarit Bhargava <prarit@redhat.com>
Cc: Don Zickus <dzickus@redhat.com>
Cc: stable@vger.kernel.org
Link: http://lkml.kernel.org/r/1374059639-8631-1-git-send-email-nhorman@tuxdriver.com
[ Small edits. ]
Signed-off-by: Ingo Molnar <mingo@kernel.org>
arch/x86/kernel/early-quirks.c

index 94ab6b9..63bdb29 100644 (file)
@@ -196,15 +196,23 @@ static void __init ati_bugs_contd(int num, int slot, int func)
 static void __init intel_remapping_check(int num, int slot, int func)
 {
        u8 revision;
+       u16 device;
 
+       device = read_pci_config_16(num, slot, func, PCI_DEVICE_ID);
        revision = read_pci_config_byte(num, slot, func, PCI_REVISION_ID);
 
        /*
-        * Revision 0x13 of this chipset supports irq remapping
-        * but has an erratum that breaks its behavior, flag it as such
+        * Revision 13 of all triggering devices id in this quirk have
+        * a problem draining interrupts when irq remapping is enabled,
+        * and should be flagged as broken.  Additionally revisions 0x12
+        * and 0x22 of device id 0x3405 has this problem.
         */
        if (revision == 0x13)
                set_irq_remapping_broken();
+       else if ((device == 0x3405) &&
+           ((revision == 0x12) ||
+            (revision == 0x22)))
+               set_irq_remapping_broken();
 
 }
 
@@ -239,6 +247,8 @@ static struct chipset early_qrk[] __initdata = {
          PCI_CLASS_SERIAL_SMBUS, PCI_ANY_ID, 0, ati_bugs_contd },
        { PCI_VENDOR_ID_INTEL, 0x3403, PCI_CLASS_BRIDGE_HOST,
          PCI_BASE_CLASS_BRIDGE, 0, intel_remapping_check },
+       { PCI_VENDOR_ID_INTEL, 0x3405, PCI_CLASS_BRIDGE_HOST,
+         PCI_BASE_CLASS_BRIDGE, 0, intel_remapping_check },
        { PCI_VENDOR_ID_INTEL, 0x3406, PCI_CLASS_BRIDGE_HOST,
          PCI_BASE_CLASS_BRIDGE, 0, intel_remapping_check },
        {}