Skip to contentSkip to conversation

i386 pci: Handle mmaping 64bit bars.

PATCH In review Eric W. Biederman 2 participants 3 messages Feb 14, 2009

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 i386 pci: Handle mmaping 64bit bars.

While reading through pci_mmap_page_range I realized that if don't properly handle 64bit bars in the 32bit kernel.

Without the added cast the shift will be done in 32bit despite being assigned to a 64bit variable and if we are mmaping an address above 4G things will do the wrong thing.

---
 arch/x86/pci/i386.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)
arch/x86/pci/i386.c +1 -1
index 5ead808..e9d5636 100644
@@ -283,7 +283,7 @@ int pci_mmap_page_range(struct pci_dev *dev, struct vm_area_struct *vma,
 			enum pci_mmap_state mmap_state, int write_combine)
 {
 	unsigned long prot;
-	u64 addr = vma->vm_pgoff << PAGE_SHIFT;
+	u64 addr = ((u64)vma->vm_pgoff) << PAGE_SHIFT;
 	unsigned long len = vma->vm_end - vma->vm_start;
 	unsigned long flags;
 	unsigned long new_flags;
Signature · 1 line
1.6.0.6
#1 lore raw
Quoted 6 lines · Eric W. Biederman

While reading through pci_mmap_page_range I realized that if don't properly handle 64bit bars in the 32bit kernel.

Without the added cast the shift will be done in 32bit despite being assigned to a 64bit variable and if we are mmaping an address above 4G things will do the wrong thing.

Hm... this really should be resource_size_t, no?

arch/x86/pci/i386.c int pci_mmap_page_range(struct pci_dev *dev, struct vm_area_struct *vma, · line 286

Quoted 2 lines
arch/x86/pci/i386.c +1 -1
-	u64 addr = vma->vm_pgoff << PAGE_SHIFT;
+	u64 addr = ((u64)vma->vm_pgoff) << PAGE_SHIFT;

-hpa

Signature · 2 lines
H. Peter Anvin, Intel Open Source Technology Center
I work for Intel.  I don't speak on their behalf.
#2 lore raw
Quoted 4 lines · H. Peter Anvin
Quoted 3 lines · Eric W. Biederman

While reading through pci_mmap_page_range I realized that if don't properly handle 64bit bars in the 32bit kernel.

Quoted 8 lines
Quoted 3 lines

Without the added cast the shift will be done in 32bit despite being assigned to a 64bit variable and if we are mmaping an address above 4G things will do the wrong thing.

Hm... this really should be resource_size_t, no?

arch/x86/pci/i386.c int pci_mmap_page_range(struct pci_dev *dev, struct vm_area_struct *vma, · line 286

Quoted 2 lines
arch/x86/pci/i386.c +1 -1
-	u64 addr = vma->vm_pgoff << PAGE_SHIFT;
+	u64 addr = ((u64)vma->vm_pgoff) << PAGE_SHIFT;

It is a parameter to the reserve_memtype and free_memtype functions, which take a u64 so the type looks ok.

If I was going to quibble I would argue that those functions should take a pfn instead of a u64...

Eric