[PATCH hotfix 6.12 v2] mm/mlock: set the correct prev on failure

[PATCH hotfix 6.12 v2] mm/mlock: set the correct prev on failure

[Wei Yang]

After commit 94d7d9233951 ("mm: abstract the vma_merge()/split_vma()
pattern for mprotect() et al."), if vma_modify_flags() return error, the
vma is set to an error code. This will lead to an invalid prev be
returned.

Generally this shouldn't matter as the caller should treat an error as
indicating state is now invalidated, however unfortunately
apply_mlockall_flags() does not check for errors and assumes that
mlock_fixup() correctly maintains prev even if an error were to occur.

This patch fixes that assumption.

[lorenzo: provide a better fix and rephrase the log]

Fixes: 94d7d9233951 ("mm: abstract the vma_merge()/split_vma() pattern for mprotect() et al.")

Signed-off-by: Wei Yang <richard.weiyang@gmail.com>
CC: Liam R. Howlett <Liam.Howlett@Oracle.com>
CC: Lorenzo Stoakes <lorenzo.stoakes@oracle.com>
CC: Vlastimil Babka <vbabka@suse.cz>
CC: Jann Horn <jannh@google.com>
Cc: <stable@vger.kernel.org>

---
v2: 
   rearrange the fix and change log per Lorenzo's suggestion
   add fix tag and cc stable

---
 mm/mlock.c | 9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)

diff --git a/mm/mlock.c b/mm/mlock.c
index e3e3dc2b2956..cde076fa7d5e 100644
--- a/mm/mlock.c
+++ b/mm/mlock.c
@@ -725,14 +725,17 @@ static int apply_mlockall_flags(int flags)
 	}
 
 	for_each_vma(vmi, vma) {
+		int error;
 		vm_flags_t newflags;
 
 		newflags = vma->vm_flags & ~VM_LOCKED_MASK;
 		newflags |= to_add;
 
-		/* Ignore errors */
-		mlock_fixup(&vmi, vma, &prev, vma->vm_start, vma->vm_end,
-			    newflags);
+		error = mlock_fixup(&vmi, vma, &prev, vma->vm_start, vma->vm_end,
+				    newflags);
+		/* Ignore errors, but prev needs fixing up. */
+		if (error)
+			prev = vma;
 		cond_resched();
 	}
 out:
-- 
2.34.1

Re: [PATCH hotfix 6.12 v2] mm/mlock: set the correct prev on failure

[Lorenzo Stoakes]

On Sun, Oct 27, 2024 at 12:33:21PM +0000, Wei Yang wrote:
> After commit 94d7d9233951 ("mm: abstract the vma_merge()/split_vma()
> pattern for mprotect() et al."), if vma_modify_flags() return error, the
> vma is set to an error code. This will lead to an invalid prev be
> returned.
>
> Generally this shouldn't matter as the caller should treat an error as
> indicating state is now invalidated, however unfortunately
> apply_mlockall_flags() does not check for errors and assumes that
> mlock_fixup() correctly maintains prev even if an error were to occur.
>
> This patch fixes that assumption.
>
> [lorenzo: provide a better fix and rephrase the log]
>
> Fixes: 94d7d9233951 ("mm: abstract the vma_merge()/split_vma() pattern for mprotect() et al.")
>
> Signed-off-by: Wei Yang <richard.weiyang@gmail.com>
> CC: Liam R. Howlett <Liam.Howlett@Oracle.com>
> CC: Lorenzo Stoakes <lorenzo.stoakes@oracle.com>
> CC: Vlastimil Babka <vbabka@suse.cz>
> CC: Jann Horn <jannh@google.com>
> Cc: <stable@vger.kernel.org>

Looks good to me,

Reviewed-by: Lorenzo Stoakes <lorenzo.stoakes@oracle.com>

>
> ---
> v2:
>    rearrange the fix and change log per Lorenzo's suggestion
>    add fix tag and cc stable
>
> ---
>  mm/mlock.c | 9 ++++++---
>  1 file changed, 6 insertions(+), 3 deletions(-)
>
> diff --git a/mm/mlock.c b/mm/mlock.c
> index e3e3dc2b2956..cde076fa7d5e 100644
> --- a/mm/mlock.c
> +++ b/mm/mlock.c
> @@ -725,14 +725,17 @@ static int apply_mlockall_flags(int flags)
>  	}
>
>  	for_each_vma(vmi, vma) {
> +		int error;
>  		vm_flags_t newflags;
>
>  		newflags = vma->vm_flags & ~VM_LOCKED_MASK;
>  		newflags |= to_add;
>
> -		/* Ignore errors */
> -		mlock_fixup(&vmi, vma, &prev, vma->vm_start, vma->vm_end,
> -			    newflags);
> +		error = mlock_fixup(&vmi, vma, &prev, vma->vm_start, vma->vm_end,
> +				    newflags);
> +		/* Ignore errors, but prev needs fixing up. */
> +		if (error)
> +			prev = vma;
>  		cond_resched();
>  	}
>  out:
> --
> 2.34.1
>

Re: [PATCH hotfix 6.12 v2] mm/mlock: set the correct prev on failure

[Andrew Morton]

On Sun, 27 Oct 2024 12:33:21 +0000 Wei Yang <richard.weiyang@gmail.com> wrote:

> After commit 94d7d9233951 ("mm: abstract the vma_merge()/split_vma()
> pattern for mprotect() et al."), if vma_modify_flags() return error, the
> vma is set to an error code. This will lead to an invalid prev be
> returned.
> 
> Generally this shouldn't matter as the caller should treat an error as
> indicating state is now invalidated, however unfortunately
> apply_mlockall_flags() does not check for errors and assumes that
> mlock_fixup() correctly maintains prev even if an error were to occur.

And what is the userspace-visible effect when this occurs?

Re: [PATCH hotfix 6.12 v2] mm/mlock: set the correct prev on failure

[Wei Yang]

On Sun, Oct 27, 2024 at 05:53:47PM -0700, Andrew Morton wrote:
>On Sun, 27 Oct 2024 12:33:21 +0000 Wei Yang <richard.weiyang@gmail.com> wrote:
>
>> After commit 94d7d9233951 ("mm: abstract the vma_merge()/split_vma()
>> pattern for mprotect() et al."), if vma_modify_flags() return error, the
>> vma is set to an error code. This will lead to an invalid prev be
>> returned.
>> 
>> Generally this shouldn't matter as the caller should treat an error as
>> indicating state is now invalidated, however unfortunately
>> apply_mlockall_flags() does not check for errors and assumes that
>> mlock_fixup() correctly maintains prev even if an error were to occur.
>
>And what is the userspace-visible effect when this occurs?
>

When error occurs, prev would be set to (-ENOMEM). And accessing this address
would lead to a kernel crash.

So looks no userspace-visible effect for this.

-- 
Wei Yang
Help you, Help me

Re: [PATCH hotfix 6.12 v2] mm/mlock: set the correct prev on failure

[Liam R. Howlett]

* Wei Yang <richard.weiyang@gmail.com> [241027 08:34]:
> After commit 94d7d9233951 ("mm: abstract the vma_merge()/split_vma()
> pattern for mprotect() et al."), if vma_modify_flags() return error, the
> vma is set to an error code. This will lead to an invalid prev be
> returned.
> 
> Generally this shouldn't matter as the caller should treat an error as
> indicating state is now invalidated, however unfortunately
> apply_mlockall_flags() does not check for errors and assumes that
> mlock_fixup() correctly maintains prev even if an error were to occur.
> 
> This patch fixes that assumption.
> 
> [lorenzo: provide a better fix and rephrase the log]
> 
> Fixes: 94d7d9233951 ("mm: abstract the vma_merge()/split_vma() pattern for mprotect() et al.")
> 
> Signed-off-by: Wei Yang <richard.weiyang@gmail.com>
> CC: Liam R. Howlett <Liam.Howlett@Oracle.com>
> CC: Lorenzo Stoakes <lorenzo.stoakes@oracle.com>
> CC: Vlastimil Babka <vbabka@suse.cz>
> CC: Jann Horn <jannh@google.com>
> Cc: <stable@vger.kernel.org>
> 
> ---
> v2: 
>    rearrange the fix and change log per Lorenzo's suggestion
>    add fix tag and cc stable
> 
> ---
>  mm/mlock.c | 9 ++++++---
>  1 file changed, 6 insertions(+), 3 deletions(-)
> 
> diff --git a/mm/mlock.c b/mm/mlock.c
> index e3e3dc2b2956..cde076fa7d5e 100644
> --- a/mm/mlock.c
> +++ b/mm/mlock.c
> @@ -725,14 +725,17 @@ static int apply_mlockall_flags(int flags)
>  	}
>  
>  	for_each_vma(vmi, vma) {
> +		int error;
>  		vm_flags_t newflags;
>  
>  		newflags = vma->vm_flags & ~VM_LOCKED_MASK;
>  		newflags |= to_add;
>  
> -		/* Ignore errors */
> -		mlock_fixup(&vmi, vma, &prev, vma->vm_start, vma->vm_end,
> -			    newflags);
> +		error = mlock_fixup(&vmi, vma, &prev, vma->vm_start, vma->vm_end,
> +				    newflags);
> +		/* Ignore errors, but prev needs fixing up. */
> +		if (error)
> +			prev = vma;

I don't think we need a local variable for the error since it's not used
for anything besides ensuring there was a non-zero return here, but it
probably doesn't make a difference.  I'd have to check the assembly to
be sure.

Either way,

Reviewed-by: Liam R. Howlett <Liam.Howlett@Oracle.com>

>  		cond_resched();
>  	}
>  out:
> -- 
> 2.34.1
> 

Re: [PATCH hotfix 6.12 v2] mm/mlock: set the correct prev on failure

[Wei Yang]

On Mon, Oct 28, 2024 at 11:00:43AM -0400, Liam R. Howlett wrote:
>* Wei Yang <richard.weiyang@gmail.com> [241027 08:34]:
>> After commit 94d7d9233951 ("mm: abstract the vma_merge()/split_vma()
>> pattern for mprotect() et al."), if vma_modify_flags() return error, the
>> vma is set to an error code. This will lead to an invalid prev be
>> returned.
>> 
>> Generally this shouldn't matter as the caller should treat an error as
>> indicating state is now invalidated, however unfortunately
>> apply_mlockall_flags() does not check for errors and assumes that
>> mlock_fixup() correctly maintains prev even if an error were to occur.
>> 
>> This patch fixes that assumption.
>> 
>> [lorenzo: provide a better fix and rephrase the log]
>> 
>> Fixes: 94d7d9233951 ("mm: abstract the vma_merge()/split_vma() pattern for mprotect() et al.")
>> 
>> Signed-off-by: Wei Yang <richard.weiyang@gmail.com>
>> CC: Liam R. Howlett <Liam.Howlett@Oracle.com>
>> CC: Lorenzo Stoakes <lorenzo.stoakes@oracle.com>
>> CC: Vlastimil Babka <vbabka@suse.cz>
>> CC: Jann Horn <jannh@google.com>
>> Cc: <stable@vger.kernel.org>
>> 
>> ---
>> v2: 
>>    rearrange the fix and change log per Lorenzo's suggestion
>>    add fix tag and cc stable
>> 
>> ---
>>  mm/mlock.c | 9 ++++++---
>>  1 file changed, 6 insertions(+), 3 deletions(-)
>> 
>> diff --git a/mm/mlock.c b/mm/mlock.c
>> index e3e3dc2b2956..cde076fa7d5e 100644
>> --- a/mm/mlock.c
>> +++ b/mm/mlock.c
>> @@ -725,14 +725,17 @@ static int apply_mlockall_flags(int flags)
>>  	}
>>  
>>  	for_each_vma(vmi, vma) {
>> +		int error;
>>  		vm_flags_t newflags;
>>  
>>  		newflags = vma->vm_flags & ~VM_LOCKED_MASK;
>>  		newflags |= to_add;
>>  
>> -		/* Ignore errors */
>> -		mlock_fixup(&vmi, vma, &prev, vma->vm_start, vma->vm_end,
>> -			    newflags);
>> +		error = mlock_fixup(&vmi, vma, &prev, vma->vm_start, vma->vm_end,
>> +				    newflags);
>> +		/* Ignore errors, but prev needs fixing up. */
>> +		if (error)
>> +			prev = vma;
>
>I don't think we need a local variable for the error since it's not used
>for anything besides ensuring there was a non-zero return here, but it
>probably doesn't make a difference.  I'd have to check the assembly to
>be sure.
>
>Either way,
>
>Reviewed-by: Liam R. Howlett <Liam.Howlett@Oracle.com>
>

Thanks

>>  		cond_resched();
>>  	}
>>  out:
>> -- 
>> 2.34.1
>> 

-- 
Wei Yang
Help you, Help me