Skip to contentSkip to conversation

Fix issues with vmalloc flush flag

PATCH In review Rick Edgecombe 4 participants 8 messages May 27, 2019 – Jun 3, 2019

Subsystems (4)
lore
Share & tools
lore
#0 lore raw

v5 cover Fix issues with vmalloc flush flag

These two patches address issues with the recently added VM_FLUSH_RESET_PERMS vmalloc flag.

Patch 1 addresses an issue that could cause a crash after other architectures besides x86 rely on this path.

Patch 2 addresses an issue where in a rare case strange arguments could be provided to flush_tlb_kernel_range().

v4->v5:
- Update commit messages with info that the first issue will actually not cause problems today
- Avoid re-use of variable (PeterZ)

v3->v4:
- Drop patch that switched vm_unmap_alias() calls to regular flush (Andy)
- Add patch to address correctness previously fixed in dropped patch

v2->v3:
- Split into two patches (Andy)

v1->v2:
- Update commit message with more detail
- Fix flush end range on !CONFIG_ARCH_HAS_SET_DIRECT_MAP case

Rick Edgecombe (2):
vmalloc: Fix calculation of direct map addr range
vmalloc: Avoid rare case of flushing tlb with weird arguments

 mm/vmalloc.c | 15 +++++++++------
 1 file changed, 9 insertions(+), 6 deletions(-)
Signature · 1 line
2.20.1
#1 lore raw

PATCH v5 1/2 vmalloc: Fix calculation of direct map addr range

The calculation of the direct map address range to flush was wrong. This could cause the RO direct map alias to not get flushed. Today this shouldn't be a problem because this flush is only needed on x86 right now and the spurious fault handler will fix cached RO->RW translations. In the future though, it could cause the permissions to remain RO in the TLB for the direct map alias, and then the page would return from the page allocator to some other component as RO and cause a crash.

So fix fix the address range calculation so the flush will include the direct map range.

Fixes 868b104d7379 ("mm/vmalloc: Add flag for freeing of special permsissions")
Cc Meelis Roos <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc "David S. Miller" <[email protected]>
Cc Dave Hansen <[email protected]>
Cc Borislav Petkov <[email protected]>
Cc Andy Lutomirski <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Nadav Amit <[email protected]>
---
 mm/vmalloc.c | 11 ++++++-----
 1 file changed, 6 insertions(+), 5 deletions(-)
mm/vmalloc.c +5 -5
index 233af6936c93..3ede9c064477 100644
@@ -2123,7 +2123,6 @@ static inline void set_area_direct_map(const struct vm_struct *area,
 /* Handle removing and resetting vm mappings related to the vm_struct. */
 static void vm_remove_mappings(struct vm_struct *area, int deallocate_pages)
 {
-	unsigned long addr = (unsigned long)area->addr;
 	unsigned long start = ULONG_MAX, end = 0;
 	int flush_reset = area->flags & VM_FLUSH_RESET_PERMS;
 	int i;
@@ -2135,8 +2134,8 @@ static void vm_remove_mappings(struct vm_struct *area, int deallocate_pages)
 	 * execute permissions, without leaving a RW+X window.
 	 */
 	if (flush_reset && !IS_ENABLED(CONFIG_ARCH_HAS_SET_DIRECT_MAP)) {
-		set_memory_nx(addr, area->nr_pages);
-		set_memory_rw(addr, area->nr_pages);
+		set_memory_nx((unsigned long)area->addr, area->nr_pages);
+		set_memory_rw((unsigned long)area->addr, area->nr_pages);
 	}
 
 	remove_vm_area(area->addr);
@@ -2160,9 +2159,11 @@ static void vm_remove_mappings(struct vm_struct *area, int deallocate_pages)
 	 * the vm_unmap_aliases() flush includes the direct map.
 	 */
 	for (i = 0; i < area->nr_pages; i++) {
-		if (page_address(area->pages[i])) {
+		unsigned long addr +				(unsigned long)page_address(area->pages[i]);
+		if (addr) {
 			start = min(addr, start);
-			end = max(addr, end);
+			end = max(addr + PAGE_SIZE, end);
 		}
 	}
 
Signature · 1 line
2.20.1
#2 lore raw

[tip:x86/urgent] mm/vmalloc: Fix calculation of direct map addr range

mm/vmalloc: Fix calculation of direct map addr range
Commit-ID:  8e41f8726dcf423621e2b6938d015b9796f6f676
Gitweb:     https://git.kernel.org/tip/8e41f8726dcf423621e2b6938d015b9796f6f676
Author:     Rick Edgecombe <[email protected]>
AuthorDate: Mon, 27 May 2019 14:10:57 -0700
Committer:  Ingo Molnar <[email protected]>
CommitDate: Mon, 3 Jun 2019 11:47:25 +0200

mm/vmalloc: Fix calculation of direct map addr range

The calculation of the direct map address range to flush was wrong. This could cause the RO direct map alias to not get flushed. Today this shouldn't be a problem because this flush is only needed on x86 right now and the spurious fault handler will fix cached RO->RW translations. In the future though, it could cause the permissions to remain RO in the TLB for the direct map alias, and then the page would return from the page allocator to some other component as RO and cause a crash.

So fix fix the address range calculation so the flush will include the direct map range.

Cc Andy Lutomirski <[email protected]>
Cc Borislav Petkov <[email protected]>
Cc Dave Hansen <[email protected]>
Cc David S. Miller <[email protected]>
Cc Linus Torvalds <[email protected]>
Cc Meelis Roos <[email protected]>
Cc Nadav Amit <[email protected]>
Cc Andrew Morton <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Thomas Gleixner <[email protected]>
Fixes 868b104d7379 ("mm/vmalloc: Add flag for freeing of special permsissions")
---
 mm/vmalloc.c | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)
mm/vmalloc.c +5 -5
index 7350a124524b..93b2dca2aadb 100644
@@ -2123,7 +2123,6 @@ static inline void set_area_direct_map(const struct vm_struct *area,
 /* Handle removing and resetting vm mappings related to the vm_struct. */
 static void vm_remove_mappings(struct vm_struct *area, int deallocate_pages)
 {
-	unsigned long addr = (unsigned long)area->addr;
 	unsigned long start = ULONG_MAX, end = 0;
 	int flush_reset = area->flags & VM_FLUSH_RESET_PERMS;
 	int i;
@@ -2135,8 +2134,8 @@ static void vm_remove_mappings(struct vm_struct *area, int deallocate_pages)
 	 * execute permissions, without leaving a RW+X window.
 	 */
 	if (flush_reset && !IS_ENABLED(CONFIG_ARCH_HAS_SET_DIRECT_MAP)) {
-		set_memory_nx(addr, area->nr_pages);
-		set_memory_rw(addr, area->nr_pages);
+		set_memory_nx((unsigned long)area->addr, area->nr_pages);
+		set_memory_rw((unsigned long)area->addr, area->nr_pages);
 	}
 
 	remove_vm_area(area->addr);
@@ -2160,9 +2159,10 @@ static void vm_remove_mappings(struct vm_struct *area, int deallocate_pages)
 	 * the vm_unmap_aliases() flush includes the direct map.
 	 */
 	for (i = 0; i < area->nr_pages; i++) {
-		if (page_address(area->pages[i])) {
+		unsigned long addr = (unsigned long)page_address(area->pages[i]);
+		if (addr) {
 			start = min(addr, start);
-			end = max(addr, end);
+			end = max(addr + PAGE_SIZE, end);
 		}
 	}
 
#3 lore raw

PATCH v5 2/2 vmalloc: Avoid rare case of flushing tlb with weird arguments

In reply to Rick Edgecombe

In a rare case, flush_tlb_kernel_range() could be called with a start higher than the end.

In vm_remove_mappings(), in case page_address() returns 0 for all pages (for example they were all in highmem), _vm_unmap_aliases() will be called with start = ULONG_MAX, end = 0 and flush = 1.

If at the same time, the vmalloc purge operation is triggered by something else while the current operation is between remove_vm_area() and _vm_unmap_aliases(), then the vm mapping just removed will be already purged. In this case the call of vm_unmap_aliases() may not find any other mappings to flush and so end up flushing start = ULONG_MAX, end = 0. So only set flush = true if we find something in the direct mapping that we need to flush, and this way this can't happen.

Fixes 868b104d7379 ("mm/vmalloc: Add flag for freeing of special permsissions")
Cc Meelis Roos <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc "David S. Miller" <[email protected]>
Cc Dave Hansen <[email protected]>
Cc Borislav Petkov <[email protected]>
Cc Andy Lutomirski <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Nadav Amit <[email protected]>
---
 mm/vmalloc.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
mm/vmalloc.c +3 -1
index 3ede9c064477..7f15a3ebcd74 100644
@@ -2125,6 +2125,7 @@ static void vm_remove_mappings(struct vm_struct *area, int deallocate_pages)
 {
 	unsigned long start = ULONG_MAX, end = 0;
 	int flush_reset = area->flags & VM_FLUSH_RESET_PERMS;
+	int flush_dmap = 0;
 	int i;
 
 	/*
@@ -2164,6 +2165,7 @@ static void vm_remove_mappings(struct vm_struct *area, int deallocate_pages)
 		if (addr) {
 			start = min(addr, start);
 			end = max(addr + PAGE_SIZE, end);
+			flush_dmap = 1;
 		}
 	}
 
@@ -2173,7 +2175,7 @@ static void vm_remove_mappings(struct vm_struct *area, int deallocate_pages)
 	 * reset the direct map permissions to the default.
 	 */
 	set_area_direct_map(area, set_direct_map_invalid_noflush);
-	_vm_unmap_aliases(start, end, 1);
+	_vm_unmap_aliases(start, end, flush_dmap);
 	set_area_direct_map(area, set_direct_map_default_noflush);
 }
 
Signature · 1 line
2.20.1
#4 lore raw

[tip:x86/urgent] mm/vmalloc: Avoid rare case of flushing TLB with weird arguments

mm/vmalloc: Avoid rare case of flushing TLB with weird arguments
Commit-ID:  31e67340cc65edfd9dac5ef26f81de8414ce5906
Gitweb:     https://git.kernel.org/tip/31e67340cc65edfd9dac5ef26f81de8414ce5906
Author:     Rick Edgecombe <[email protected]>
AuthorDate: Mon, 27 May 2019 14:10:58 -0700
Committer:  Ingo Molnar <[email protected]>
CommitDate: Mon, 3 Jun 2019 11:47:25 +0200

mm/vmalloc: Avoid rare case of flushing TLB with weird arguments

In a rare case, flush_tlb_kernel_range() could be called with a start higher than the end.

In vm_remove_mappings(), in case page_address() returns 0 for all pages (for example they were all in highmem), _vm_unmap_aliases() will be called with start = ULONG_MAX, end = 0 and flush = 1.

If at the same time, the vmalloc purge operation is triggered by something else while the current operation is between remove_vm_area() and _vm_unmap_aliases(), then the vm mapping just removed will be already purged. In this case the call of vm_unmap_aliases() may not find any other mappings to flush and so end up flushing start = ULONG_MAX, end = 0. So only set flush = true if we find something in the direct mapping that we need to flush, and this way this can't happen.

Cc Andy Lutomirski <[email protected]>
Cc Borislav Petkov <[email protected]>
Cc Dave Hansen <[email protected]>
Cc David S. Miller <[email protected]>
Cc Linus Torvalds <[email protected]>
Cc Meelis Roos <[email protected]>
Cc Nadav Amit <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Andrew Morton <[email protected]>
Cc Thomas Gleixner <[email protected]>
Fixes 868b104d7379 ("mm/vmalloc: Add flag for freeing of special permsissions")
---
 mm/vmalloc.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
mm/vmalloc.c +3 -1
index 93b2dca2aadb..4c9e150e5ad3 100644
@@ -2125,6 +2125,7 @@ static void vm_remove_mappings(struct vm_struct *area, int deallocate_pages)
 {
 	unsigned long start = ULONG_MAX, end = 0;
 	int flush_reset = area->flags & VM_FLUSH_RESET_PERMS;
+	int flush_dmap = 0;
 	int i;
 
 	/*
@@ -2163,6 +2164,7 @@ static void vm_remove_mappings(struct vm_struct *area, int deallocate_pages)
 		if (addr) {
 			start = min(addr, start);
 			end = max(addr + PAGE_SIZE, end);
+			flush_dmap = 1;
 		}
 	}
 
@@ -2172,7 +2174,7 @@ static void vm_remove_mappings(struct vm_struct *area, int deallocate_pages)
 	 * reset the direct map permissions to the default.
 	 */
 	set_area_direct_map(area, set_direct_map_invalid_noflush);
-	_vm_unmap_aliases(start, end, 1);
+	_vm_unmap_aliases(start, end, flush_dmap);
 	set_area_direct_map(area, set_direct_map_default_noflush);
 }
 
#5 lore raw
In reply to Rick Edgecombe
Quoted 8 lines · Rick Edgecombe

These two patches address issues with the recently added VM_FLUSH_RESET_PERMS vmalloc flag.

Patch 1 addresses an issue that could cause a crash after other architectures besides x86 rely on this path.

Patch 2 addresses an issue where in a rare case strange arguments could be provided to flush_tlb_kernel_range().

Thanks!

maintainer
#6 lore raw
In reply to Rick Edgecombe

From: Rick Edgecombe <[email protected]>
Date: Mon, 27 May 2019 14:10:56 -0700

Quoted 8 lines

These two patches address issues with the recently added VM_FLUSH_RESET_PERMS vmalloc flag.

Patch 1 addresses an issue that could cause a crash after other architectures besides x86 rely on this path.

Patch 2 addresses an issue where in a rare case strange arguments could be provided to flush_tlb_kernel_range().

It just occurred to me another situation that would cause trouble on sparc64, and that's if someone the address range of the main kernel image ended up being passed to flush_tlb_kernel_range().

That would flush the locked kernel mapping and crash the kernel instantly in a completely non-recoverable way.

#7 lore raw

T24gVHVlLCAyMDE5LTA1LTI4IGF0IDE3OjIzIC0wNzAwLCBEYXZpZCBNaWxsZXIgd3JvdGU6DQo+ IEZyb206IFJpY2sgRWRnZWNvbWJlIDxyaWNrLnAuZWRnZWNvbWJlQGludGVsLmNvbT4NCj4gRGF0 ZTogTW9uLCAyNyBNYXkgMjAxOSAxNDoxMDo1NiAtMDcwMA0KPiANCj4gPiBUaGVzZSB0d28gcGF0 Y2hlcyBhZGRyZXNzIGlzc3VlcyB3aXRoIHRoZSByZWNlbnRseSBhZGRlZA0KPiA+IFZNX0ZMVVNI X1JFU0VUX1BFUk1TIHZtYWxsb2MgZmxhZy4NCj4gPiANCj4gPiBQYXRjaCAxIGFkZHJlc3NlcyBh biBpc3N1ZSB0aGF0IGNvdWxkIGNhdXNlIGEgY3Jhc2ggYWZ0ZXIgb3RoZXINCj4gPiBhcmNoaXRl Y3R1cmVzIGJlc2lkZXMgeDg2IHJlbHkgb24gdGhpcyBwYXRoLg0KPiA+IA0KPiA+IFBhdGNoIDIg YWRkcmVzc2VzIGFuIGlzc3VlIHdoZXJlIGluIGEgcmFyZSBjYXNlIHN0cmFuZ2UgYXJndW1lbnRz DQo+ID4gY291bGQgYmUgcHJvdmlkZWQgdG8gZmx1c2hfdGxiX2tlcm5lbF9yYW5nZSgpLiANCj4g DQo+IEl0IGp1c3Qgb2NjdXJyZWQgdG8gbWUgYW5vdGhlciBzaXR1YXRpb24gdGhhdCB3b3VsZCBj YXVzZSB0cm91YmxlIG9uDQo+IHNwYXJjNjQsIGFuZCB0aGF0J3MgaWYgc29tZW9uZSB0aGUgYWRk cmVzcyByYW5nZSBvZiB0aGUgbWFpbiBrZXJuZWwNCj4gaW1hZ2UgZW5kZWQgdXAgYmVpbmcgcGFz c2VkIHRvIGZsdXNoX3RsYl9rZXJuZWxfcmFuZ2UoKS4NCj4gDQo+IFRoYXQgd291bGQgZmx1c2gg dGhlIGxvY2tlZCBrZXJuZWwgbWFwcGluZyBhbmQgY3Jhc2ggdGhlIGtlcm5lbA0KPiBpbnN0YW50 bHkgaW4gYSBjb21wbGV0ZWx5IG5vbi1yZWNvdmVyYWJsZSB3YXkuDQoNCkhtbSwgSSBoYXZlbid0 IHJlY2VpdmVkIHRoZSBsb2dzIGZyb20gTWVlbGlzIHRoYXQgd2lsbCBzaG93IHRoZSByZWFsDQpy YW5nZXMgYmVpbmcgcGFzc2VkIGludG8gZmx1c2hfdGxiX2tlcm5lbF9yYW5nZSgpIG9uIHNwYXJj LCBidXQgaXQNCnNob3VsZCBiZSBmbHVzaGluZyBhIHJhbmdlIHNwYW5uaW5nIGZyb20gdGhlIG1v ZHVsZXMgdG8gdGhlIGRpcmVjdCBtYXAuDQpJdCBsb29rcyBsaWtlIHRoZSBrZXJuZWwgaXMgYXQg dGhlIHZlcnkgYm90dG9tIG9mIHRoZSBhZGRyZXNzIHNwYWNlLCBzbw0Kbm90IGluY2x1ZGVkLiBP ciBkbyB5b3UgbWVhbiB0aGUgcGFnZXMgdGhhdCBob2xkIHRoZSBrZXJuZWwgdGV4dCBvbiB0aGUN CmRpcmVjdCBtYXA/DQoNCkJ1dCByZWdhcmRsZXNzIG9mIHRoaXMgbmV3IGNvZGUsIERFQlVHX1BB R0VBTExPQyBoYW5ncyB3aXRoIHRoZSBmaXJzdA0Kdm1hbGxvYyBmcmVlL3VubWFwLiBUaGF0IHNo b3VsZCBiZSBqdXN0IGZsdXNoaW5nIGEgc2luZ2xlIGFsbG9jYXRpb24gaW4NCnRoZSB2bWFsbG9j IHJhbmdlLg0KDQpJZiBpdCBpcyBzb21laG93IGNhdGNoaW5nIGEgbG9ja2VkIGVudHJ5IHRob3Vn aC4uLiBBcmUgdGhlcmUgYW55IHNwYXJjDQpmbHVzaCBtZWNoYW5pc21zIHRoYXQgY291bGQgYmUg dXNlZCBpbiB2bWFsbG9jIHRoYXQgd29uJ3QgdG91Y2ggbG9ja2VkDQplbnRyaWVzPyBQZXRlciBa IHdhcyBwb2ludGluZyBvdXQgdGhhdCBmbHVzaF90bGJfYWxsKCkgbWlnaHQgYmUgbW9yZQ0KYXBw cm9yaWF0ZSBmb3Igdm1hbGxvYyBhbnl3YXkuDQoNCg=