Skip to contentSkip to conversation

x86: always define DECLARE_PCI_UNMAP* macros

PATCH In review Joerg Roedel 3 participants 4 messages Nov 20, 2008 – Nov 28, 2008 x86

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 x86: always define DECLARE_PCI_UNMAP* macros

Currently these macros evaluate to a no-op except the kernel is compiled with GART or Calgary support. But we also need these macros when we have SWIOTLB, VT-d or AMD IOMMU in the kernel. Since we always compile at least with SWIOTLB we can define these macros always.
This patch is also for stable backport for the same reason the SWIOTLB default selection patch is.

---
 arch/x86/include/asm/pci_64.h |   14 --------------
 1 files changed, 0 insertions(+), 14 deletions(-)
arch/x86/include/asm/pci_64.h +0 -14
index 5b28995..d02d936 100644
@@ -34,8 +34,6 @@ extern void pci_iommu_alloc(void);
  */
 #define PCI_DMA_BUS_IS_PHYS (dma_ops->is_phys)
 
-#if defined(CONFIG_GART_IOMMU) || defined(CONFIG_CALGARY_IOMMU)
-
 #define DECLARE_PCI_UNMAP_ADDR(ADDR_NAME)	\
 	dma_addr_t ADDR_NAME;
 #define DECLARE_PCI_UNMAP_LEN(LEN_NAME)		\
@@ -49,18 +47,6 @@ extern void pci_iommu_alloc(void);
 #define pci_unmap_len_set(PTR, LEN_NAME, VAL)		\
 	(((PTR)->LEN_NAME) = (VAL))
 
-#else
-/* No IOMMU */
-
-#define DECLARE_PCI_UNMAP_ADDR(ADDR_NAME)
-#define DECLARE_PCI_UNMAP_LEN(LEN_NAME)
-#define pci_unmap_addr(PTR, ADDR_NAME)		(0)
-#define pci_unmap_addr_set(PTR, ADDR_NAME, VAL)	do { } while (0)
-#define pci_unmap_len(PTR, LEN_NAME)		(0)
-#define pci_unmap_len_set(PTR, LEN_NAME, VAL)	do { } while (0)
-
-#endif
-
 #endif /* __KERNEL__ */
 
 #endif /* _ASM_X86_PCI_64_H */
Signature · 1 line
1.5.6.4
#1 lore raw
Quoted 7 lines · Joerg Roedel

Currently these macros evaluate to a no-op except the kernel is compiled with GART or Calgary support. But we also need these macros when we have SWIOTLB, VT-d or AMD IOMMU in the kernel. Since we always compile at least with SWIOTLB we can define these macros always. This patch is also for stable backport for the same reason the SWIOTLB default selection patch is.

Quoted 5 lines
---
 arch/x86/include/asm/pci_64.h |   14 --------------
 1 files changed, 0 insertions(+), 14 deletions(-)

applied to tip/x86/urgent (for v2.6.28), as this is a DMA mapping bugfix for IOMMUs. If the AMD IOMMU is turned on in the .config but CONFIG_GART_IOMMU is off (a rare but possible combo), we'd leak on unmap and crash the box quickly.

Jesse, is that path of upstream merge fine with you?

Ingo

#2 lore raw
Quoted 14 lines · Ingo Molnar
Quoted 7 lines · Joerg Roedel

Currently these macros evaluate to a no-op except the kernel is compiled with GART or Calgary support. But we also need these macros when we have SWIOTLB, VT-d or AMD IOMMU in the kernel. Since we always compile at least with SWIOTLB we can define these macros always. This patch is also for stable backport for the same reason the SWIOTLB default selection patch is.

Quoted 5 lines
---
 arch/x86/include/asm/pci_64.h |   14 --------------
 1 files changed, 0 insertions(+), 14 deletions(-)
Quoted 6 lines

applied to tip/x86/urgent (for v2.6.28), as this is a DMA mapping bugfix for IOMMUs. If the AMD IOMMU is turned on in the .config but CONFIG_GART_IOMMU is off (a rare but possible combo), we'd leak on unmap and crash the box quickly.

Jesse, is that path of upstream merge fine with you?

Yeah, that's fine with me. OTOH removing config options generally seems like a good idea too, so maybe this combination should just be disallowed. Or is there a good reason for this type of config?

Signature · 1 line
Jesse Barnes, Intel Open Source Technology Center
#3 lore raw
Quoted 20 lines · Jesse Barnes
Quoted 19 lines · Ingo Molnar
Quoted 7 lines · Joerg Roedel

Currently these macros evaluate to a no-op except the kernel is compiled with GART or Calgary support. But we also need these macros when we have SWIOTLB, VT-d or AMD IOMMU in the kernel. Since we always compile at least with SWIOTLB we can define these macros always. This patch is also for stable backport for the same reason the SWIOTLB default selection patch is.

Quoted 5 lines
---
 arch/x86/include/asm/pci_64.h |   14 --------------
 1 files changed, 0 insertions(+), 14 deletions(-)

applied to tip/x86/urgent (for v2.6.28), as this is a DMA mapping bugfix for IOMMUs. If the AMD IOMMU is turned on in the .config but CONFIG_GART_IOMMU is off (a rare but possible combo), we'd leak on unmap and crash the box quickly.

Quoted 5 lines
Quoted 1 line

Jesse, is that path of upstream merge fine with you?

Yeah, that's fine with me. OTOH removing config options generally seems like a good idea too, so maybe this combination should just be disallowed. Or is there a good reason for this type of config?

well, it's an IOMMU driver, so it's a kernel sizing/configuration thing, in line with how we generally allow drivers - with no particular dependency between them.

Ingo