This is the cause of the DMA faults and disk corruption that people have
been seeing. Some chipsets neglect to report the RWBF "capability" --
the flag which says that we need to flush the chipset write-buffer when
changing the DMA page tables, to ensure that the change is visible to
the IOMMU.
Override that bit on the affected chipsets, and everything is happy
again.
Thanks to Chris and Bhavesh and others for helping to debug.
Should resolve:
https://bugzilla.redhat.com/show_bug.cgi?id=479996
http://bugzilla.kernel.org/show_bug.cgi?id=12578
Signed-off-by: David Woodhouse <[email protected]>
Cc: Linus Torvalds <[email protected]>
Tested-and-acked-by: Chris Wright <[email protected]>
Reviewed-by: Bhavesh Davda <[email protected]>
Signed-off-by: Ingo Molnar <[email protected]>
/* global iommu list, set NULL for ignored DMAR units */
static struct intel_iommu **g_iommus;
+static int rwbf_quirk;
+
/*
* 0: Present
* 1-11: Reserved
u32 val;
unsigned long flag;
- if (!cap_rwbf(iommu->cap))
+ if (!rwbf_quirk && !cap_rwbf(iommu->cap))
return;
val = iommu->gcmd | DMA_GCMD_WBF;
.unmap = intel_iommu_unmap_range,
.iova_to_phys = intel_iommu_iova_to_phys,
};
+
+static void __devinit quirk_iommu_rwbf(struct pci_dev *dev)
+{
+ /*
+ * Mobile 4 Series Chipset neglects to set RWBF capability,
+ * but needs it:
+ */
+ printk(KERN_INFO "DMAR: Forcing write-buffer flush capability\n");
+ rwbf_quirk = 1;
+}
+
+DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, 0x2a40, quirk_iommu_rwbf);