Commit 4bb9c5c02153dfc89a6c73a6f32091413805ad7d
Committed by
Ingo Molnar
1 parent
6a5c05f002
Exists in
master
and in
4 other branches
VM, x86, PAT: Change is_linear_pfn_mapping to not use vm_pgoff
Impact: fix false positive PAT warnings - also fix VirtalBox hang Use of vma->vm_pgoff to identify the pfnmaps that are fully mapped at mmap time is broken. vm_pgoff is set by generic mmap code even for cases where drivers are setting up the mappings at the fault time. The problem was originally reported here: http://marc.info/?l=linux-kernel&m=123383810628583&w=2 Change is_linear_pfn_mapping logic to overload VM_INSERTPAGE flag along with VM_PFNMAP to mean full PFNMAP setup at mmap time. Problem also tracked at: http://bugzilla.kernel.org/show_bug.cgi?id=12800 Reported-by: Thomas Hellstrom <thellstrom@vmware.com> Tested-by: Frans Pop <elendil@planet.nl> Signed-off-by: Venkatesh Pallipadi <venkatesh.pallipadi@intel.com> Signed-off-by: Suresh Siddha <suresh.b.siddha>@intel.com> Cc: Nick Piggin <npiggin@suse.de> Cc: "ebiederm@xmission.com" <ebiederm@xmission.com> Cc: <stable@kernel.org> # only for 2.6.29.1, not .28 LKML-Reference: <20090313004527.GA7176@linux-os.sc.intel.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
Showing 3 changed files with 20 additions and 6 deletions Side-by-side Diff
arch/x86/mm/pat.c
... | ... | @@ -641,10 +641,11 @@ |
641 | 641 | is_ram = pat_pagerange_is_ram(paddr, paddr + size); |
642 | 642 | |
643 | 643 | /* |
644 | - * reserve_pfn_range() doesn't support RAM pages. | |
644 | + * reserve_pfn_range() doesn't support RAM pages. Maintain the current | |
645 | + * behavior with RAM pages by returning success. | |
645 | 646 | */ |
646 | 647 | if (is_ram != 0) |
647 | - return -EINVAL; | |
648 | + return 0; | |
648 | 649 | |
649 | 650 | ret = reserve_memtype(paddr, paddr + size, want_flags, &flags); |
650 | 651 | if (ret) |
include/linux/mm.h
... | ... | @@ -98,7 +98,7 @@ |
98 | 98 | #define VM_HUGETLB 0x00400000 /* Huge TLB Page VM */ |
99 | 99 | #define VM_NONLINEAR 0x00800000 /* Is non-linear (remap_file_pages) */ |
100 | 100 | #define VM_MAPPED_COPY 0x01000000 /* T if mapped copy of data (nommu mmap) */ |
101 | -#define VM_INSERTPAGE 0x02000000 /* The vma has had "vm_insert_page()" done on it */ | |
101 | +#define VM_INSERTPAGE 0x02000000 /* The vma has had "vm_insert_page()" done on it. Refer note in VM_PFNMAP_AT_MMAP below */ | |
102 | 102 | #define VM_ALWAYSDUMP 0x04000000 /* Always include in core dumps */ |
103 | 103 | |
104 | 104 | #define VM_CAN_NONLINEAR 0x08000000 /* Has ->fault & does nonlinear pages */ |
... | ... | @@ -127,6 +127,17 @@ |
127 | 127 | #define VM_SPECIAL (VM_IO | VM_DONTEXPAND | VM_RESERVED | VM_PFNMAP) |
128 | 128 | |
129 | 129 | /* |
130 | + * pfnmap vmas that are fully mapped at mmap time (not mapped on fault). | |
131 | + * Used by x86 PAT to identify such PFNMAP mappings and optimize their handling. | |
132 | + * Note VM_INSERTPAGE flag is overloaded here. i.e, | |
133 | + * VM_INSERTPAGE && !VM_PFNMAP implies | |
134 | + * The vma has had "vm_insert_page()" done on it | |
135 | + * VM_INSERTPAGE && VM_PFNMAP implies | |
136 | + * The vma is PFNMAP with full mapping at mmap time | |
137 | + */ | |
138 | +#define VM_PFNMAP_AT_MMAP (VM_INSERTPAGE | VM_PFNMAP) | |
139 | + | |
140 | +/* | |
130 | 141 | * mapping from the currently active vm_flags protection bits (the |
131 | 142 | * low four bits) to a page protection mask.. |
132 | 143 | */ |
... | ... | @@ -145,7 +156,7 @@ |
145 | 156 | */ |
146 | 157 | static inline int is_linear_pfn_mapping(struct vm_area_struct *vma) |
147 | 158 | { |
148 | - return ((vma->vm_flags & VM_PFNMAP) && vma->vm_pgoff); | |
159 | + return ((vma->vm_flags & VM_PFNMAP_AT_MMAP) == VM_PFNMAP_AT_MMAP); | |
149 | 160 | } |
150 | 161 | |
151 | 162 | static inline int is_pfn_mapping(struct vm_area_struct *vma) |
mm/memory.c
... | ... | @@ -1665,9 +1665,10 @@ |
1665 | 1665 | * behaviour that some programs depend on. We mark the "original" |
1666 | 1666 | * un-COW'ed pages by matching them up with "vma->vm_pgoff". |
1667 | 1667 | */ |
1668 | - if (addr == vma->vm_start && end == vma->vm_end) | |
1668 | + if (addr == vma->vm_start && end == vma->vm_end) { | |
1669 | 1669 | vma->vm_pgoff = pfn; |
1670 | - else if (is_cow_mapping(vma->vm_flags)) | |
1670 | + vma->vm_flags |= VM_PFNMAP_AT_MMAP; | |
1671 | + } else if (is_cow_mapping(vma->vm_flags)) | |
1671 | 1672 | return -EINVAL; |
1672 | 1673 | |
1673 | 1674 | vma->vm_flags |= VM_IO | VM_RESERVED | VM_PFNMAP; |
... | ... | @@ -1679,6 +1680,7 @@ |
1679 | 1680 | * needed from higher level routine calling unmap_vmas |
1680 | 1681 | */ |
1681 | 1682 | vma->vm_flags &= ~(VM_IO | VM_RESERVED | VM_PFNMAP); |
1683 | + vma->vm_flags &= ~VM_PFNMAP_AT_MMAP; | |
1682 | 1684 | return -EINVAL; |
1683 | 1685 | } |
1684 | 1686 |