Skip to contentSkip to conversation

mm: Fix overflow check in expand_upwards()

PATCH In review Jörn Engel 2 participants 5 messages Jun 29, 2017 – Jun 30, 2017

Subsystems (2)
Memory management Memory mapping
lore
Share & tools
lore

Jörn Engel

author
#0 lore raw

PATCH 1/1 mm: Fix overflow check in expand_upwards()

I believe the overflow check was correct, then got subtly broken by

	commit bd726c90b6b8
	Author: Helge Deller <[email protected]>
	Date:   Mon Jun 19 17:34:05 2017 +0200

	    Allow stack to grow up to address space limit

The old overflow check may have been a bit subtle and I suppose Helge missed its importance.

	if (!address)
		return -ENOMEM;

Functionally the my check is identical to the old one. I just hope the alternative form (and comment!) make it harder to miss and break things in a future patch.

Signed-off-by Joern Engel <[email protected]>
---
 mm/mmap.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
mm/mmap.c +2 -1
index a5e3dcd75e79..7366f62c31f4 100644
@@ -2232,7 +2232,8 @@ int expand_upwards(struct vm_area_struct *vma, unsigned long address)
 
 	/* Guard against exceeding limits of the address space. */
 	address &= PAGE_MASK;
-	if (address >= TASK_SIZE)
+	/* second check is for integer overflow */
+	if (address >= TASK_SIZE || address + PAGE_SIZE < address)
 		return -ENOMEM;
 	address += PAGE_SIZE;
Signature · 1 line
2.1.4
#1 lore raw
Quoted 27 lines · Jörn Engel

I believe the overflow check was correct, then got subtly broken by

	commit bd726c90b6b8
	Author: Helge Deller <[email protected]>
	Date:   Mon Jun 19 17:34:05 2017 +0200

	    Allow stack to grow up to address space limit

The old overflow check may have been a bit subtle and I suppose Helge missed its importance.

	if (!address)
		return -ENOMEM;

Functionally the my check is identical to the old one. I just hope the alternative form (and comment!) make it harder to miss and break things in a future patch.

Signed-off-by Joern Engel <[email protected]>
---
 mm/mmap.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
mm/mmap.c +0 -0
index a5e3dcd75e79..7366f62c31f4 100644

mm/mmap.c int expand_upwards(struct vm_area_struct *vma, unsigned long address) · line 2238

Quoted 8 lines
mm/mmap.c +2 -1
@@ -2232,7 +2232,8 @@ int expand_upwards(struct vm_area_struct *vma, unsigned long address)
 
 	/* Guard against exceeding limits of the address space. */
 	address &= PAGE_MASK;
-	if (address >= TASK_SIZE)
+	/* second check is for integer overflow */
+	if (address >= TASK_SIZE || address + PAGE_SIZE < address)
 		return -ENOMEM;
 	address += PAGE_SIZE;

That overflow check is still there.
Look at the next few lines in mmap.c:

       /* Enforce stack_guard_gap */
        gap_addr = address + stack_guard_gap;

        /* Guard against overflow */
        if (gap_addr < address || gap_addr > TASK_SIZE)
                gap_addr = TASK_SIZE;

If the requested page plus the gap (=gap_addr) wraps around, then the code will limit it to TASK_SIZE.
So, the code should already take care of all possible areas >=TASK_SIZE, including wrap-arounds.

Did you faced a real issue?

Helge

#2 lore raw

* Helge Deller <[email protected]>:

Quoted 29 lines · Jörn Engel
Quoted 28 lines

I believe the overflow check was correct, then got subtly broken by

	commit bd726c90b6b8
	Author: Helge Deller <[email protected]>
	Date:   Mon Jun 19 17:34:05 2017 +0200

	    Allow stack to grow up to address space limit

The old overflow check may have been a bit subtle and I suppose Helge missed its importance.

	if (!address)
		return -ENOMEM;

Functionally the my check is identical to the old one. I just hope the alternative form (and comment!) make it harder to miss and break things in a future patch.

Signed-off-by Joern Engel <[email protected]>
---
 mm/mmap.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
mm/mmap.c +0 -0
index a5e3dcd75e79..7366f62c31f4 100644
@@ -2232,7 +2232,8 @@ int expand_upwards(struct vm_area_struct *vma, unsigned long address)
Quoted 9 lines

mm/mmap.c int expand_upwards(struct vm_area_struct *vma, unsigned long address) · line 2238

Quoted 7 lines
mm/mmap.c +2 -1
 	/* Guard against exceeding limits of the address space. */
 	address &= PAGE_MASK;
-	if (address >= TASK_SIZE)
+	/* second check is for integer overflow */
+	if (address >= TASK_SIZE || address + PAGE_SIZE < address)
 		return -ENOMEM;
 	address += PAGE_SIZE;

That overflow check is still there.

I see there are some architectures which define TASK_SIZE not as multiple of PAGE_SIZE and as 0xffffffff for which the (address >= TASK_SIZE) check will not trigger:

arch/arm/include/asm/memory.h:#define TASK_SIZE UL(0xffffffff)
arch/frv/include/asm/mem-layout.h:#define TASK_SIZE __UL(0xFFFFFFFFUL)
arch/m68k/include/asm/processor.h:#define TASK_SIZE (0xFFFFFFFFUL)
arch/blackfin/include/asm/processor.h:#define TASK_SIZE 0xFFFFFFFF
arch/h8300/include/asm/processor.h:#define TASK_SIZE (0xFFFFFFFFUL)
arch/xtensa/include/asm/processor.h:#define TASK_SIZE __XTENSA_UL_CONST(0xffffffff)

None of those have an upwards growing stack and thus I believe we don't run into issues, but nevertheless the checks could probably be changed like this (untested patch):

mm/mmap.c +6 -4
index a5e3dcd..224bdc2 100644
@@ -2224,15 +2224,17 @@ int expand_upwards(struct vm_area_struct *vma, unsigned long address)
 {
 	struct mm_struct *mm = vma->vm_mm;
 	struct vm_area_struct *next;
-	unsigned long gap_addr;
+	unsigned long gap_addr, max_task_size;
 	int error = 0;
 
 	if (!(vma->vm_flags & VM_GROWSUP))
 		return -EFAULT;
 
+	max_task_size = TASK_SIZE & PAGE_MASK;
+
 	/* Guard against exceeding limits of the address space. */
 	address &= PAGE_MASK;
-	if (address >= TASK_SIZE)
+	if (address >= max_task_size)
 		return -ENOMEM;
 	address += PAGE_SIZE;
 

@@ -2240,8 +2242,8 @@ int expand_upwards(struct vm_area_struct *vma, unsigned long address)
 	gap_addr = address + stack_guard_gap;
 
 	/* Guard against overflow */
-	if (gap_addr < address || gap_addr > TASK_SIZE)
-		gap_addr = TASK_SIZE;
+	if (gap_addr < address || gap_addr > max_task_size)
+		gap_addr = max_task_size;
 
 	next = vma->vm_next;
 	if (next && next->vm_start < gap_addr) {

Helge

author
#3 lore raw
Quoted 12 lines · Helge Deller

I see there are some architectures which define TASK_SIZE not as multiple of PAGE_SIZE and as 0xffffffff for which the (address >= TASK_SIZE) check will not trigger:

arch/arm/include/asm/memory.h:#define TASK_SIZE UL(0xffffffff)
arch/frv/include/asm/mem-layout.h:#define TASK_SIZE __UL(0xFFFFFFFFUL)
arch/m68k/include/asm/processor.h:#define TASK_SIZE (0xFFFFFFFFUL)
arch/blackfin/include/asm/processor.h:#define TASK_SIZE 0xFFFFFFFF
arch/h8300/include/asm/processor.h:#define TASK_SIZE (0xFFFFFFFFUL)
arch/xtensa/include/asm/processor.h:#define TASK_SIZE __XTENSA_UL_CONST(0xffffffff)

Quoted 3 lines

None of those have an upwards growing stack and thus I believe we don't run into issues, but nevertheless the checks could probably be changed like this (untested patch):

That would also work. I have no preference which patch to use.

Quoted 36 lines
mm/mmap.c +6 -4
index a5e3dcd..224bdc2 100644
@@ -2224,15 +2224,17 @@ int expand_upwards(struct vm_area_struct *vma, unsigned long address)
 {
 	struct mm_struct *mm = vma->vm_mm;
 	struct vm_area_struct *next;
-	unsigned long gap_addr;
+	unsigned long gap_addr, max_task_size;
 	int error = 0;
 
 	if (!(vma->vm_flags & VM_GROWSUP))
 		return -EFAULT;
 
+	max_task_size = TASK_SIZE & PAGE_MASK;
+
 	/* Guard against exceeding limits of the address space. */
 	address &= PAGE_MASK;
-	if (address >= TASK_SIZE)
+	if (address >= max_task_size)
 		return -ENOMEM;
 	address += PAGE_SIZE;
 

@@ -2240,8 +2242,8 @@ int expand_upwards(struct vm_area_struct *vma, unsigned long address)
 	gap_addr = address + stack_guard_gap;
 
 	/* Guard against overflow */
-	if (gap_addr < address || gap_addr > TASK_SIZE)
-		gap_addr = TASK_SIZE;
+	if (gap_addr < address || gap_addr > max_task_size)
+		gap_addr = max_task_size;
 
 	next = vma->vm_next;
 	if (next && next->vm_start < gap_addr) {
Quoted 1 line

Helge

Jörn

--
You cannot suppose that Moliere ever troubled himself to be original in the matter of ideas. You cannot suppose that the stories he tells in his plays have never been told before. They were culled, as you very well know. -- Andre-Louis Moreau in Scarabouche

author
#4 lore raw
In reply to Helge Deller
Quoted 47 lines · Helge Deller
Quoted 27 lines · Jörn Engel

I believe the overflow check was correct, then got subtly broken by

	commit bd726c90b6b8
	Author: Helge Deller <[email protected]>
	Date:   Mon Jun 19 17:34:05 2017 +0200

	    Allow stack to grow up to address space limit

The old overflow check may have been a bit subtle and I suppose Helge missed its importance.

	if (!address)
		return -ENOMEM;

Functionally the my check is identical to the old one. I just hope the alternative form (and comment!) make it harder to miss and break things in a future patch.

Signed-off-by Joern Engel <[email protected]>
---
 mm/mmap.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
mm/mmap.c +0 -0
index a5e3dcd75e79..7366f62c31f4 100644

mm/mmap.c int expand_upwards(struct vm_area_struct *vma, unsigned long address) · line 2238

Quoted 8 lines
mm/mmap.c +2 -1
@@ -2232,7 +2232,8 @@ int expand_upwards(struct vm_area_struct *vma, unsigned long address)
 
 	/* Guard against exceeding limits of the address space. */
 	address &= PAGE_MASK;
-	if (address >= TASK_SIZE)
+	/* second check is for integer overflow */
+	if (address >= TASK_SIZE || address + PAGE_SIZE < address)
 		return -ENOMEM;
 	address += PAGE_SIZE;

That overflow check is still there.
Look at the next few lines in mmap.c:

       /* Enforce stack_guard_gap */
        gap_addr = address + stack_guard_gap;

        /* Guard against overflow */
        if (gap_addr < address || gap_addr > TASK_SIZE)
                gap_addr = TASK_SIZE;
Quoted 4 lines

If the requested page plus the gap (=gap_addr) wraps around, then the code will limit it to TASK_SIZE.
So, the code should already take care of all possible areas >=TASK_SIZE, including wrap-arounds.

Does it cover the case where address is (unsigned long)-PAGE_SIZE?

I believe you catch every other case, but not that one.

Quoted 1 line

Did you faced a real issue?

No. I don't even own a computer with stacks growing up. Just spotted this while reviewing some patches going by.

Jörn

--
The Linux community is zillions of people with different cultures and ideas all trying to convince the rest that their vision of the shared culture is right.
-- Alan Cox