Hi,
Recently we met quite a lot of random kernel panic issues after enable
CONFIG_PROC_PAGE_MONITOR in kernel, after debuggint sometime we found
this has something to do with following bug in pagemap:
In struc pagemapread:
struct pagemapread {
int pos, len;
pagemap_entry_t *buffer;
bool v2;
};
pos is number of PM_ENTRY_BYTES in buffer, but len is the size of buffer,
it is a mistake to compare pos and len in add_page_map() for checking
buffer is full or not, and this can lead to buffer overflow and random
kernel panic issue.
Correct len to be total number of PM_ENTRY_BYTES in buffer.
Signed-off-by: Yonghua Zheng <[email protected]>
---
fs/proc/task_mmu.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/fs/proc/task_mmu.c b/fs/proc/task_mmu.c
index dbf61f6..cb98853 100644
--- a/fs/proc/task_mmu.c
+++ b/fs/proc/task_mmu.c
@@ -1116,8 +1116,8 @@ static ssize_t pagemap_read(struct file *file,
char __user *buf,
goto out_task;
pm.v2 = soft_dirty_cleared;
- pm.len = PM_ENTRY_BYTES * (PAGEMAP_WALK_SIZE >> PAGE_SHIFT);
- pm.buffer = kmalloc(pm.len, GFP_TEMPORARY);
+ pm.len = (PAGEMAP_WALK_SIZE >> PAGE_SHIFT);
+ pm.buffer = kmalloc(pm.len * PM_ENTRY_BYTES, GFP_TEMPORARY);
ret = -ENOMEM;
if (!pm.buffer)
goto out_task;
--
1.7.9.5
On Fri, Aug 09, 2013 at 01:16:41PM +0800, yonghua zheng wrote:
> Hi,
>
> Recently we met quite a lot of random kernel panic issues after enable
> CONFIG_PROC_PAGE_MONITOR in kernel, after debuggint sometime we found
> this has something to do with following bug in pagemap:
>
> In struc pagemapread:
>
> struct pagemapread {
> int pos, len;
> pagemap_entry_t *buffer;
> bool v2;
> };
>
> pos is number of PM_ENTRY_BYTES in buffer, but len is the size of buffer,
> it is a mistake to compare pos and len in add_page_map() for checking
s/add_page_map/add_to_pagemap/ ?
> buffer is full or not, and this can lead to buffer overflow and random
> kernel panic issue.
>
> Correct len to be total number of PM_ENTRY_BYTES in buffer.
>
> Signed-off-by: Yonghua Zheng <[email protected]>
You can find coding style violation with scripts/checkpatch.pl.
And I think this patch is worth going into -stable trees
(maybe since 2.6.34.)
The fix itself looks fine to me.
Thanks,
Naoya Horiguchi
> ---
> fs/proc/task_mmu.c | 4 ++--
> 1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/fs/proc/task_mmu.c b/fs/proc/task_mmu.c
> index dbf61f6..cb98853 100644
> --- a/fs/proc/task_mmu.c
> +++ b/fs/proc/task_mmu.c
> @@ -1116,8 +1116,8 @@ static ssize_t pagemap_read(struct file *file,
> char __user *buf,
> goto out_task;
>
> pm.v2 = soft_dirty_cleared;
> - pm.len = PM_ENTRY_BYTES * (PAGEMAP_WALK_SIZE >> PAGE_SHIFT);
> - pm.buffer = kmalloc(pm.len, GFP_TEMPORARY);
> + pm.len = (PAGEMAP_WALK_SIZE >> PAGE_SHIFT);
> + pm.buffer = kmalloc(pm.len * PM_ENTRY_BYTES, GFP_TEMPORARY);
> ret = -ENOMEM;
> if (!pm.buffer)
> goto out_task;
>
> --
> 1.7.9.5
>
> --
> To unsubscribe, send a message with 'unsubscribe linux-mm' in
> the body to [email protected]. For more info on Linux MM,
> see: http://www.linux-mm.org/ .
> Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
>
Update the patch according to Naoya's comment, I also run
./scripts/checkpatch.pl, and it passed ;D.
>From 96826b0fdf9ec6d6e16c2c595f371dbb841250f7 Mon Sep 17 00:00:00 2001
From: Yonghua Zheng <[email protected]>
Date: Mon, 5 Aug 2013 12:12:24 +0800
Subject: [PATCH 1/1] pagemap: fix buffer overflow in add_to_pagemap()
In struc pagemapread:
struct pagemapread {
int pos, len;
pagemap_entry_t *buffer;
bool v2;
};
pos is number of PM_ENTRY_BYTES in buffer, but len is the size of buffer,
it is a mistake to compare pos and len in add_to_pagemap() for checking
buffer is full or not, and this can lead to buffer overflow and random
kernel panic issue.
Correct len to be total number of PM_ENTRY_BYTES in buffer.
Signed-off-by: Yonghua Zheng <[email protected]>
diff --git a/fs/proc/task_mmu.c b/fs/proc/task_mmu.c
index dbf61f6..cb98853 100644
--- a/fs/proc/task_mmu.c
+++ b/fs/proc/task_mmu.c
@@ -1116,8 +1116,8 @@ static ssize_t pagemap_read(struct file *file,
char __user *buf,
goto out_task;
pm.v2 = soft_dirty_cleared;
- pm.len = PM_ENTRY_BYTES * (PAGEMAP_WALK_SIZE >> PAGE_SHIFT);
- pm.buffer = kmalloc(pm.len, GFP_TEMPORARY);
+ pm.len = (PAGEMAP_WALK_SIZE >> PAGE_SHIFT);
+ pm.buffer = kmalloc(pm.len * PM_ENTRY_BYTES, GFP_TEMPORARY);
ret = -ENOMEM;
if (!pm.buffer)
goto out_task;
--
1.7.9.5
On Sat, Aug 10, 2013 at 4:33 AM, Naoya Horiguchi
<[email protected]> wrote:
> On Fri, Aug 09, 2013 at 01:16:41PM +0800, yonghua zheng wrote:
>> Hi,
>>
>> Recently we met quite a lot of random kernel panic issues after enable
>> CONFIG_PROC_PAGE_MONITOR in kernel, after debuggint sometime we found
>> this has something to do with following bug in pagemap:
>>
>> In struc pagemapread:
>>
>> struct pagemapread {
>> int pos, len;
>> pagemap_entry_t *buffer;
>> bool v2;
>> };
>>
>> pos is number of PM_ENTRY_BYTES in buffer, but len is the size of buffer,
>> it is a mistake to compare pos and len in add_page_map() for checking
>
> s/add_page_map/add_to_pagemap/ ?
>
>> buffer is full or not, and this can lead to buffer overflow and random
>> kernel panic issue.
>>
>> Correct len to be total number of PM_ENTRY_BYTES in buffer.
>>
>> Signed-off-by: Yonghua Zheng <[email protected]>
>
> You can find coding style violation with scripts/checkpatch.pl.
> And I think this patch is worth going into -stable trees
> (maybe since 2.6.34.)
>
> The fix itself looks fine to me.
>
> Thanks,
> Naoya Horiguchi
>
>> ---
>> fs/proc/task_mmu.c | 4 ++--
>> 1 file changed, 2 insertions(+), 2 deletions(-)
>>
>> diff --git a/fs/proc/task_mmu.c b/fs/proc/task_mmu.c
>> index dbf61f6..cb98853 100644
>> --- a/fs/proc/task_mmu.c
>> +++ b/fs/proc/task_mmu.c
>> @@ -1116,8 +1116,8 @@ static ssize_t pagemap_read(struct file *file,
>> char __user *buf,
>> goto out_task;
>>
>> pm.v2 = soft_dirty_cleared;
>> - pm.len = PM_ENTRY_BYTES * (PAGEMAP_WALK_SIZE >> PAGE_SHIFT);
>> - pm.buffer = kmalloc(pm.len, GFP_TEMPORARY);
>> + pm.len = (PAGEMAP_WALK_SIZE >> PAGE_SHIFT);
>> + pm.buffer = kmalloc(pm.len * PM_ENTRY_BYTES, GFP_TEMPORARY);
>> ret = -ENOMEM;
>> if (!pm.buffer)
>> goto out_task;
>>
>> --
>> 1.7.9.5
>>
>> --
>> To unsubscribe, send a message with 'unsubscribe linux-mm' in
>> the body to [email protected]. For more info on Linux MM,
>> see: http://www.linux-mm.org/ .
>> Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
>>
On Fri, Aug 9, 2013 at 8:49 PM, yonghua zheng <[email protected]> wrote:
> Update the patch according to Naoya's comment, I also run
> ./scripts/checkpatch.pl, and it passed ;D.
>
> From 96826b0fdf9ec6d6e16c2c595f371dbb841250f7 Mon Sep 17 00:00:00 2001
> From: Yonghua Zheng <[email protected]>
> Date: Mon, 5 Aug 2013 12:12:24 +0800
> Subject: [PATCH 1/1] pagemap: fix buffer overflow in add_to_pagemap()
>
> In struc pagemapread:
>
> struct pagemapread {
> int pos, len;
> pagemap_entry_t *buffer;
> bool v2;
> };
>
> pos is number of PM_ENTRY_BYTES in buffer, but len is the size of buffer,
> it is a mistake to compare pos and len in add_to_pagemap() for checking
> buffer is full or not, and this can lead to buffer overflow and random
> kernel panic issue.
>
> Correct len to be total number of PM_ENTRY_BYTES in buffer.
>
> Signed-off-by: Yonghua Zheng <[email protected]>
Acked-by: KOSAKI Motohiro <[email protected]>
On Sat, Aug 10, 2013 at 08:49:34AM +0800, yonghua zheng wrote:
> Update the patch according to Naoya's comment, I also run
> ./scripts/checkpatch.pl, and it passed ;D.
Reviewed-by: Naoya Horiguchi <[email protected]>
# Sorry if I missed something, but I guess that your MTA might replace tabs
# with spaces. Documentation/email-clients.txt can be helpful to solve it.
>
> From 96826b0fdf9ec6d6e16c2c595f371dbb841250f7 Mon Sep 17 00:00:00 2001
> From: Yonghua Zheng <[email protected]>
> Date: Mon, 5 Aug 2013 12:12:24 +0800
> Subject: [PATCH 1/1] pagemap: fix buffer overflow in add_to_pagemap()
>
> In struc pagemapread:
>
> struct pagemapread {
> int pos, len;
> pagemap_entry_t *buffer;
> bool v2;
> };
>
> pos is number of PM_ENTRY_BYTES in buffer, but len is the size of buffer,
> it is a mistake to compare pos and len in add_to_pagemap() for checking
> buffer is full or not, and this can lead to buffer overflow and random
> kernel panic issue.
>
> Correct len to be total number of PM_ENTRY_BYTES in buffer.
>
> Signed-off-by: Yonghua Zheng <[email protected]>
>
> diff --git a/fs/proc/task_mmu.c b/fs/proc/task_mmu.c
> index dbf61f6..cb98853 100644
> --- a/fs/proc/task_mmu.c
> +++ b/fs/proc/task_mmu.c
> @@ -1116,8 +1116,8 @@ static ssize_t pagemap_read(struct file *file,
> char __user *buf,
> goto out_task;
>
> pm.v2 = soft_dirty_cleared;
> - pm.len = PM_ENTRY_BYTES * (PAGEMAP_WALK_SIZE >> PAGE_SHIFT);
> - pm.buffer = kmalloc(pm.len, GFP_TEMPORARY);
> + pm.len = (PAGEMAP_WALK_SIZE >> PAGE_SHIFT);
> + pm.buffer = kmalloc(pm.len * PM_ENTRY_BYTES, GFP_TEMPORARY);
> ret = -ENOMEM;
> if (!pm.buffer)
> goto out_task;
> --
> 1.7.9.5
>
> On Sat, Aug 10, 2013 at 4:33 AM, Naoya Horiguchi
> <[email protected]> wrote:
> > On Fri, Aug 09, 2013 at 01:16:41PM +0800, yonghua zheng wrote:
> >> Hi,
> >>
> >> Recently we met quite a lot of random kernel panic issues after enable
> >> CONFIG_PROC_PAGE_MONITOR in kernel, after debuggint sometime we found
> >> this has something to do with following bug in pagemap:
> >>
> >> In struc pagemapread:
> >>
> >> struct pagemapread {
> >> int pos, len;
> >> pagemap_entry_t *buffer;
> >> bool v2;
> >> };
> >>
> >> pos is number of PM_ENTRY_BYTES in buffer, but len is the size of buffer,
> >> it is a mistake to compare pos and len in add_page_map() for checking
> >
> > s/add_page_map/add_to_pagemap/ ?
> >
> >> buffer is full or not, and this can lead to buffer overflow and random
> >> kernel panic issue.
> >>
> >> Correct len to be total number of PM_ENTRY_BYTES in buffer.
> >>
> >> Signed-off-by: Yonghua Zheng <[email protected]>
> >
> > You can find coding style violation with scripts/checkpatch.pl.
> > And I think this patch is worth going into -stable trees
> > (maybe since 2.6.34.)
> >
> > The fix itself looks fine to me.
> >
> > Thanks,
> > Naoya Horiguchi
> >
> >> ---
> >> fs/proc/task_mmu.c | 4 ++--
> >> 1 file changed, 2 insertions(+), 2 deletions(-)
> >>
> >> diff --git a/fs/proc/task_mmu.c b/fs/proc/task_mmu.c
> >> index dbf61f6..cb98853 100644
> >> --- a/fs/proc/task_mmu.c
> >> +++ b/fs/proc/task_mmu.c
> >> @@ -1116,8 +1116,8 @@ static ssize_t pagemap_read(struct file *file,
> >> char __user *buf,
> >> goto out_task;
> >>
> >> pm.v2 = soft_dirty_cleared;
> >> - pm.len = PM_ENTRY_BYTES * (PAGEMAP_WALK_SIZE >> PAGE_SHIFT);
> >> - pm.buffer = kmalloc(pm.len, GFP_TEMPORARY);
> >> + pm.len = (PAGEMAP_WALK_SIZE >> PAGE_SHIFT);
> >> + pm.buffer = kmalloc(pm.len * PM_ENTRY_BYTES, GFP_TEMPORARY);
> >> ret = -ENOMEM;
> >> if (!pm.buffer)
> >> goto out_task;
> >>
> >> --
> >> 1.7.9.5
> >>
> >> --
> >> To unsubscribe, send a message with 'unsubscribe linux-mm' in
> >> the body to [email protected]. For more info on Linux MM,
> >> see: http://www.linux-mm.org/ .
> >> Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
> >>
>
> --
> To unsubscribe, send a message with 'unsubscribe linux-mm' in
> the body to [email protected]. For more info on Linux MM,
> see: http://www.linux-mm.org/ .
> Don't email: <a href=mailto:"[email protected]"> [email protected] </a>
>
On Fri, 9 Aug 2013 13:16:41 +0800 yonghua zheng <[email protected]> wrote:
> Hi,
>
> Recently we met quite a lot of random kernel panic issues after enable
> CONFIG_PROC_PAGE_MONITOR in kernel, after debuggint sometime we found
> this has something to do with following bug in pagemap:
>
> In struc pagemapread:
>
> struct pagemapread {
> int pos, len;
> pagemap_entry_t *buffer;
> bool v2;
> };
>
> pos is number of PM_ENTRY_BYTES in buffer, but len is the size of buffer,
> it is a mistake to compare pos and len in add_page_map() for checking
> buffer is full or not, and this can lead to buffer overflow and random
> kernel panic issue.
>
> Correct len to be total number of PM_ENTRY_BYTES in buffer.
>
> Signed-off-by: Yonghua Zheng <[email protected]>
> ---
> fs/proc/task_mmu.c | 4 ++--
> 1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/fs/proc/task_mmu.c b/fs/proc/task_mmu.c
> index dbf61f6..cb98853 100644
> --- a/fs/proc/task_mmu.c
> +++ b/fs/proc/task_mmu.c
> @@ -1116,8 +1116,8 @@ static ssize_t pagemap_read(struct file *file,
> char __user *buf,
> goto out_task;
>
> pm.v2 = soft_dirty_cleared;
> - pm.len = PM_ENTRY_BYTES * (PAGEMAP_WALK_SIZE >> PAGE_SHIFT);
> - pm.buffer = kmalloc(pm.len, GFP_TEMPORARY);
> + pm.len = (PAGEMAP_WALK_SIZE >> PAGE_SHIFT);
> + pm.buffer = kmalloc(pm.len * PM_ENTRY_BYTES, GFP_TEMPORARY);
> ret = -ENOMEM;
> if (!pm.buffer)
> goto out_task;
Yes, that's a bug. I'd propose this addition to your fix:
From: Andrew Morton <[email protected]>
Subject: pagemap-fix-buffer-overflow-in-add_page_map-fix
document pagemapread.pos and .len units, fix PM_ENTRY_BYTES definition
Cc: Yonghua Zheng <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
---
fs/proc/task_mmu.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff -puN fs/proc/task_mmu.c~pagemap-fix-buffer-overflow-in-add_page_map-fix fs/proc/task_mmu.c
--- a/fs/proc/task_mmu.c~pagemap-fix-buffer-overflow-in-add_page_map-fix
+++ a/fs/proc/task_mmu.c
@@ -868,7 +868,7 @@ typedef struct {
} pagemap_entry_t;
struct pagemapread {
- int pos, len;
+ int pos, len; /* units: PM_ENTRY_BYTES, not bytes */
pagemap_entry_t *buffer;
bool v2;
};
@@ -876,7 +876,7 @@ struct pagemapread {
#define PAGEMAP_WALK_SIZE (PMD_SIZE)
#define PAGEMAP_WALK_MASK (PMD_MASK)
-#define PM_ENTRY_BYTES sizeof(u64)
+#define PM_ENTRY_BYTES sizeof(pagemap_entry_t)
#define PM_STATUS_BITS 3
#define PM_STATUS_OFFSET (64 - PM_STATUS_BITS)
#define PM_STATUS_MASK (((1LL << PM_STATUS_BITS) - 1) << PM_STATUS_OFFSET)
_
btw, your email client wordwraps the patches and replaces tabs with
spaces. Please fix that up for next time?