2019-12-24 06:24:40

by Lu Baolu

[permalink] [raw]
Subject: [PATCH 1/1] iommu/vt-d: Add a quirk flag for scope mismatched devices

We expect devices with endpoint scope to have normal PCI headers,
and devices with bridge scope to have bridge PCI headers. However
Some PCI devices may be listed in the DMAR table with bridge scope,
even though they have a normal PCI header. Add a quirk flag for
those special devices.

Cc: Roland Dreier <[email protected]>
Cc: Jim Yan <[email protected]>
Signed-off-by: Lu Baolu <[email protected]>
---
drivers/iommu/dmar.c | 37 +++++++++++++++++++++++--------------
1 file changed, 23 insertions(+), 14 deletions(-)

diff --git a/drivers/iommu/dmar.c b/drivers/iommu/dmar.c
index fb30d5053664..fc24abc70a05 100644
--- a/drivers/iommu/dmar.c
+++ b/drivers/iommu/dmar.c
@@ -65,6 +65,26 @@ static void free_iommu(struct intel_iommu *iommu);

extern const struct iommu_ops intel_iommu_ops;

+static int scope_mismatch_quirk;
+static void quirk_dmar_scope_mismatch(struct pci_dev *dev)
+{
+ pci_info(dev, "scope mismatch ignored\n");
+ scope_mismatch_quirk = 1;
+}
+
+/*
+ * We expect devices with endpoint scope to have normal PCI
+ * headers, and devices with bridge scope to have bridge PCI
+ * headers. However some PCI devices may be listed in the
+ * DMAR table with bridge scope, even though they have a
+ * normal PCI header. We don't declare a socpe mismatch for
+ * below special cases.
+ */
+DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, 0x2f0d, /* NTB devices */
+ quirk_dmar_scope_mismatch);
+DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, 0x2020, /* NVME host */
+ quirk_dmar_scope_mismatch);
+
static void dmar_register_drhd_unit(struct dmar_drhd_unit *drhd)
{
/*
@@ -231,20 +251,9 @@ int dmar_insert_dev_scope(struct dmar_pci_notify_info *info,
if (!dmar_match_pci_path(info, scope->bus, path, level))
continue;

- /*
- * We expect devices with endpoint scope to have normal PCI
- * headers, and devices with bridge scope to have bridge PCI
- * headers. However PCI NTB devices may be listed in the
- * DMAR table with bridge scope, even though they have a
- * normal PCI header. NTB devices are identified by class
- * "BRIDGE_OTHER" (0680h) - we don't declare a socpe mismatch
- * for this special case.
- */
- if ((scope->entry_type == ACPI_DMAR_SCOPE_TYPE_ENDPOINT &&
- info->dev->hdr_type != PCI_HEADER_TYPE_NORMAL) ||
- (scope->entry_type == ACPI_DMAR_SCOPE_TYPE_BRIDGE &&
- (info->dev->hdr_type == PCI_HEADER_TYPE_NORMAL &&
- info->dev->class >> 8 != PCI_CLASS_BRIDGE_OTHER))) {
+ if (!scope_mismatch_quirk &&
+ ((scope->entry_type == ACPI_DMAR_SCOPE_TYPE_ENDPOINT) ^
+ (info->dev->hdr_type == PCI_HEADER_TYPE_NORMAL))) {
pr_warn("Device scope type does not match for %s\n",
pci_name(info->dev));
return -EINVAL;
--
2.17.1


2019-12-24 19:13:50

by Jerry Snitselaar

[permalink] [raw]
Subject: Re: [PATCH 1/1] iommu/vt-d: Add a quirk flag for scope mismatched devices

On Tue Dec 24 19, Lu Baolu wrote:
>We expect devices with endpoint scope to have normal PCI headers,
>and devices with bridge scope to have bridge PCI headers. However
>Some PCI devices may be listed in the DMAR table with bridge scope,
>even though they have a normal PCI header. Add a quirk flag for
>those special devices.
>
>Cc: Roland Dreier <[email protected]>
>Cc: Jim Yan <[email protected]>
>Signed-off-by: Lu Baolu <[email protected]>
>---

Reviewed-by: Jerry Snitselaar <[email protected]>

2020-01-01 23:38:33

by Lu Baolu

[permalink] [raw]
Subject: Re: [PATCH 1/1] iommu/vt-d: Add a quirk flag for scope mismatched devices

On 12/24/19 2:22 PM, Lu Baolu wrote:
> We expect devices with endpoint scope to have normal PCI headers,
> and devices with bridge scope to have bridge PCI headers. However
> Some PCI devices may be listed in the DMAR table with bridge scope,
> even though they have a normal PCI header. Add a quirk flag for
> those special devices.
>
> Cc: Roland Dreier<[email protected]>
> Cc: Jim Yan<[email protected]>
> Signed-off-by: Lu Baolu<[email protected]>

Queued for v5.6.

Thanks,
-baolu

2020-01-07 13:31:14

by Christoph Hellwig

[permalink] [raw]
Subject: Re: [PATCH 1/1] iommu/vt-d: Add a quirk flag for scope mismatched devices

WTF is a NVMe host supposed to mean for a PCIe device. NVMe defines
the host as following:

"1.6.16 host

An entity that interfaces to an NVM subsystem through one or more
controllers and submits commands to Submission Queues and retrieves
command completions from Completion Queues."

in other words - the Linux kernel is the NVMe host. You need to
describe this magic broken piece of crap a lot better than that.