mm/util.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
From: Kiryl Shutsemau <kas@kernel.org>
vm_mmap_pgoff() includes a fsnotify call that allows for pre-content
hooks on mmap().
The fsnotify_mmap_perm() function takes, among other arguments, an
offset in the file in the form of loff_t. However, vm_mmap_pgoff() has
file offset in the form of pgoff. This offset needs to be converted
before being passed to fsnotify_mmap_perm().
The conversion from pgoff to loff_t is incorrect. The pgoff value needs
to be shifted left by PAGE_SHIFT to obtain loff_t, not right.
This issue was identified through code inspection.
Signed-off-by: Kiryl Shutsemau <kas@kernel.org>
Fixes: 066e053fe208 ("fsnotify: add pre-content hooks on mmap()")
Cc: stable@vger.kernel.org
Cc: Josef Bacik <josef@toxicpanda.com>
Cc: Amir Goldstein <amir73il@gmail.com>
Cc: Jan Kara <jack@suse.cz>
---
mm/util.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/mm/util.c b/mm/util.c
index f814e6a59ab1..52a667157264 100644
--- a/mm/util.c
+++ b/mm/util.c
@@ -573,7 +573,7 @@ unsigned long vm_mmap_pgoff(struct file *file, unsigned long addr,
ret = security_mmap_file(file, prot, flag);
if (!ret)
- ret = fsnotify_mmap_perm(file, prot, pgoff >> PAGE_SHIFT, len);
+ ret = fsnotify_mmap_perm(file, prot, pgoff << PAGE_SHIFT, len);
if (!ret) {
if (mmap_write_lock_killable(mm))
return -EINTR;
--
2.50.1
On Fri, Oct 03, 2025 at 04:58:04PM +0100, Kiryl Shutsemau wrote:
> From: Kiryl Shutsemau <kas@kernel.org>
>
> vm_mmap_pgoff() includes a fsnotify call that allows for pre-content
> hooks on mmap().
>
> The fsnotify_mmap_perm() function takes, among other arguments, an
> offset in the file in the form of loff_t. However, vm_mmap_pgoff() has
> file offset in the form of pgoff. This offset needs to be converted
> before being passed to fsnotify_mmap_perm().
>
> The conversion from pgoff to loff_t is incorrect. The pgoff value needs
> to be shifted left by PAGE_SHIFT to obtain loff_t, not right.
>
> This issue was identified through code inspection.
>
> Signed-off-by: Kiryl Shutsemau <kas@kernel.org>
> Fixes: 066e053fe208 ("fsnotify: add pre-content hooks on mmap()")
> Cc: stable@vger.kernel.org
> Cc: Josef Bacik <josef@toxicpanda.com>
> Cc: Amir Goldstein <amir73il@gmail.com>
> Cc: Jan Kara <jack@suse.cz>
> ---
> mm/util.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/mm/util.c b/mm/util.c
> index f814e6a59ab1..52a667157264 100644
> --- a/mm/util.c
> +++ b/mm/util.c
> @@ -573,7 +573,7 @@ unsigned long vm_mmap_pgoff(struct file *file, unsigned long addr,
>
> ret = security_mmap_file(file, prot, flag);
> if (!ret)
> - ret = fsnotify_mmap_perm(file, prot, pgoff >> PAGE_SHIFT, len);
> + ret = fsnotify_mmap_perm(file, prot, pgoff << PAGE_SHIFT, len);
It misses the case to (loff_t) and it broken for 32-bit machines.
Luckily, Ryan submitted another fix for the same bug at the almost the
same time. And he was more careful around types:
https://lore.kernel.org/all/20251003155238.2147410-1-ryan.roberts@arm.com
--
Kiryl Shutsemau / Kirill A. Shutemov
On 03/10/25 10:26 pm, Kiryl Shutsemau wrote:
> On Fri, Oct 03, 2025 at 04:58:04PM +0100, Kiryl Shutsemau wrote:
>> From: Kiryl Shutsemau <kas@kernel.org>
>>
>> vm_mmap_pgoff() includes a fsnotify call that allows for pre-content
>> hooks on mmap().
>>
>> The fsnotify_mmap_perm() function takes, among other arguments, an
>> offset in the file in the form of loff_t. However, vm_mmap_pgoff() has
>> file offset in the form of pgoff. This offset needs to be converted
>> before being passed to fsnotify_mmap_perm().
>>
>> The conversion from pgoff to loff_t is incorrect. The pgoff value needs
>> to be shifted left by PAGE_SHIFT to obtain loff_t, not right.
>>
>> This issue was identified through code inspection.
>>
>> Signed-off-by: Kiryl Shutsemau <kas@kernel.org>
>> Fixes: 066e053fe208 ("fsnotify: add pre-content hooks on mmap()")
>> Cc: stable@vger.kernel.org
>> Cc: Josef Bacik <josef@toxicpanda.com>
>> Cc: Amir Goldstein <amir73il@gmail.com>
>> Cc: Jan Kara <jack@suse.cz>
>> ---
>> mm/util.c | 2 +-
>> 1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/mm/util.c b/mm/util.c
>> index f814e6a59ab1..52a667157264 100644
>> --- a/mm/util.c
>> +++ b/mm/util.c
>> @@ -573,7 +573,7 @@ unsigned long vm_mmap_pgoff(struct file *file, unsigned long addr,
>>
>> ret = security_mmap_file(file, prot, flag);
>> if (!ret)
>> - ret = fsnotify_mmap_perm(file, prot, pgoff >> PAGE_SHIFT, len);
>> + ret = fsnotify_mmap_perm(file, prot, pgoff << PAGE_SHIFT, len);
> It misses the case to (loff_t) and it broken for 32-bit machines.
>
> Luckily, Ryan submitted another fix for the same bug at the almost the
> same time. And he was more careful around types:
>
> https://lore.kernel.org/all/20251003155238.2147410-1-ryan.roberts@arm.com
Oops! I need to be more careful...
>
On 03/10/25 9:28 pm, Kiryl Shutsemau wrote:
> From: Kiryl Shutsemau <kas@kernel.org>
>
> vm_mmap_pgoff() includes a fsnotify call that allows for pre-content
> hooks on mmap().
>
> The fsnotify_mmap_perm() function takes, among other arguments, an
> offset in the file in the form of loff_t. However, vm_mmap_pgoff() has
> file offset in the form of pgoff. This offset needs to be converted
> before being passed to fsnotify_mmap_perm().
>
> The conversion from pgoff to loff_t is incorrect. The pgoff value needs
> to be shifted left by PAGE_SHIFT to obtain loff_t, not right.
>
> This issue was identified through code inspection.
>
> Signed-off-by: Kiryl Shutsemau <kas@kernel.org>
> Fixes: 066e053fe208 ("fsnotify: add pre-content hooks on mmap()")
> Cc: stable@vger.kernel.org
> Cc: Josef Bacik <josef@toxicpanda.com>
> Cc: Amir Goldstein <amir73il@gmail.com>
> Cc: Jan Kara <jack@suse.cz>
> ---
>
Reviewed-by: Dev Jain <dev.jain@arm.com>
© 2016 - 2025 Red Hat, Inc.