diff mbox series

[v2] arm64: pageattr: Explicitly bail out when changing permissions for vmalloc_huge mappings

Message ID 20250403052844.61818-1-dev.jain@arm.com (mailing list archive)
State New
Headers show
Series [v2] arm64: pageattr: Explicitly bail out when changing permissions for vmalloc_huge mappings | expand

Commit Message

Dev Jain April 3, 2025, 5:28 a.m. UTC 
arm64 uses apply_to_page_range to change permissions for kernel vmalloc mappings,
which does not support changing permissions for block mappings. This function
will change permissions until it encounters a block mapping, and will bail
out with a warning. Since there are no reports of this triggering, it
implies that there are currently no cases of code doing a vmalloc_huge()
followed by partial permission change. But this is a footgun waiting to
go off, so let's detect it early and avoid the possibility of permissions
in an intermediate state. So,  explicitly disallow changing permissions
for VM_ALLOW_HUGE_VMAP mappings.

Reviewed-by: Ryan Roberts <ryan.roberts@arm.com>
Reviewed-by: Mike Rapoport (Microsoft) <rppt@kernel.org>
Signed-off-by: Dev Jain <dev.jain@arm.com>
---
v1->v2:
 - Improve changelog, keep mention of page mappings in comment

 arch/arm64/mm/pageattr.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

Comments

Anshuman Khandual April 3, 2025, 9:31 a.m. UTC | #1 
On 4/3/25 10:58, Dev Jain wrote:
> arm64 uses apply_to_page_range to change permissions for kernel vmalloc mappings,
> which does not support changing permissions for block mappings. This function
> will change permissions until it encounters a block mapping, and will bail
> out with a warning. Since there are no reports of this triggering, it
> implies that there are currently no cases of code doing a vmalloc_huge()
> followed by partial permission change. But this is a footgun waiting to
> go off, so let's detect it early and avoid the possibility of permissions
> in an intermediate state. So,  explicitly disallow changing permissions
> for VM_ALLOW_HUGE_VMAP mappings.
> 
> Reviewed-by: Ryan Roberts <ryan.roberts@arm.com>
> Reviewed-by: Mike Rapoport (Microsoft) <rppt@kernel.org>
> Signed-off-by: Dev Jain <dev.jain@arm.com>
> ---
> v1->v2:
>  - Improve changelog, keep mention of page mappings in comment
> 
>  arch/arm64/mm/pageattr.c | 6 +++---
>  1 file changed, 3 insertions(+), 3 deletions(-)
> 
> diff --git a/arch/arm64/mm/pageattr.c b/arch/arm64/mm/pageattr.c
> index 39fd1f7ff02a..04d4a8f676db 100644
> --- a/arch/arm64/mm/pageattr.c
> +++ b/arch/arm64/mm/pageattr.c
> @@ -96,8 +96,8 @@ static int change_memory_common(unsigned long addr, int numpages,
>  	 * we are operating on does not result in such splitting.
>  	 *
>  	 * Let's restrict ourselves to mappings created by vmalloc (or vmap).
> -	 * Those are guaranteed to consist entirely of page mappings, and
> -	 * splitting is never needed.
> +	 * Disallow VM_ALLOW_HUGE_VMAP mappings to guarantee that only page
> +	 * mappings are updated and splitting is never needed.
>  	 *
>  	 * So check whether the [addr, addr + size) interval is entirely
>  	 * covered by precisely one VM area that has the VM_ALLOC flag set.
> @@ -105,7 +105,7 @@ static int change_memory_common(unsigned long addr, int numpages,
>  	area = find_vm_area((void *)addr);
>  	if (!area ||
>  	    end > (unsigned long)kasan_reset_tag(area->addr) + area->size ||
> -	    !(area->flags & VM_ALLOC))
> +	    ((area->flags & (VM_ALLOC | VM_ALLOW_HUGE_VMAP)) != VM_ALLOC))
>  		return -EINVAL;
>  
>  	if (!numpages)

LGTM

Reviewed-by: Anshuman Khandual <anshuman.khandual@arm.com>
Gavin Shan April 4, 2025, 9:43 a.m. UTC | #2 
On 4/3/25 3:28 PM, Dev Jain wrote:
> arm64 uses apply_to_page_range to change permissions for kernel vmalloc mappings,
> which does not support changing permissions for block mappings. This function
> will change permissions until it encounters a block mapping, and will bail
> out with a warning. Since there are no reports of this triggering, it
> implies that there are currently no cases of code doing a vmalloc_huge()
> followed by partial permission change. But this is a footgun waiting to
> go off, so let's detect it early and avoid the possibility of permissions
> in an intermediate state. So,  explicitly disallow changing permissions
> for VM_ALLOW_HUGE_VMAP mappings.
> 
> Reviewed-by: Ryan Roberts <ryan.roberts@arm.com>
> Reviewed-by: Mike Rapoport (Microsoft) <rppt@kernel.org>
> Signed-off-by: Dev Jain <dev.jain@arm.com>
> ---
> v1->v2:
>   - Improve changelog, keep mention of page mappings in comment
> 
>   arch/arm64/mm/pageattr.c | 6 +++---
>   1 file changed, 3 insertions(+), 3 deletions(-)
> 

Reviewed-by: Gavin Shan <gshan@redhat.com>
David Hildenbrand April 4, 2025, 12:57 p.m. UTC | #3 
On 03.04.25 07:28, Dev Jain wrote:
> arm64 uses apply_to_page_range to change permissions for kernel vmalloc mappings,
> which does not support changing permissions for block mappings. This function
> will change permissions until it encounters a block mapping, and will bail
> out with a warning. Since there are no reports of this triggering, it
> implies that there are currently no cases of code doing a vmalloc_huge()
> followed by partial permission change. But this is a footgun waiting to
> go off, so let's detect it early and avoid the possibility of permissions
> in an intermediate state. So,  explicitly disallow changing permissions
> for VM_ALLOW_HUGE_VMAP mappings.
> 
> Reviewed-by: Ryan Roberts <ryan.roberts@arm.com>
> Reviewed-by: Mike Rapoport (Microsoft) <rppt@kernel.org>
> Signed-off-by: Dev Jain <dev.jain@arm.com>
> ---
> v1->v2:
>   - Improve changelog, keep mention of page mappings in comment
> 
>   arch/arm64/mm/pageattr.c | 6 +++---
>   1 file changed, 3 insertions(+), 3 deletions(-)
> 
> diff --git a/arch/arm64/mm/pageattr.c b/arch/arm64/mm/pageattr.c
> index 39fd1f7ff02a..04d4a8f676db 100644
> --- a/arch/arm64/mm/pageattr.c
> +++ b/arch/arm64/mm/pageattr.c
> @@ -96,8 +96,8 @@ static int change_memory_common(unsigned long addr, int numpages,
>   	 * we are operating on does not result in such splitting.
>   	 *
>   	 * Let's restrict ourselves to mappings created by vmalloc (or vmap).
> -	 * Those are guaranteed to consist entirely of page mappings, and
> -	 * splitting is never needed.
> +	 * Disallow VM_ALLOW_HUGE_VMAP mappings to guarantee that only page
> +	 * mappings are updated and splitting is never needed.
>   	 *
>   	 * So check whether the [addr, addr + size) interval is entirely
>   	 * covered by precisely one VM area that has the VM_ALLOC flag set.
> @@ -105,7 +105,7 @@ static int change_memory_common(unsigned long addr, int numpages,
>   	area = find_vm_area((void *)addr);
>   	if (!area ||
>   	    end > (unsigned long)kasan_reset_tag(area->addr) + area->size ||
> -	    !(area->flags & VM_ALLOC))
> +	    ((area->flags & (VM_ALLOC | VM_ALLOW_HUGE_VMAP)) != VM_ALLOC))
>   		return -EINVAL;
>   
>   	if (!numpages)

Makes sense to me. Whenever required, we can improve the checks (or even 
try supporting splitting).

Acked-by: David Hildenbrand <david@redhat.com>
Ryan Roberts April 14, 2025, 1:48 p.m. UTC | #4 
On 04/04/2025 13:57, David Hildenbrand wrote:
> On 03.04.25 07:28, Dev Jain wrote:
>> arm64 uses apply_to_page_range to change permissions for kernel vmalloc mappings,
>> which does not support changing permissions for block mappings. This function
>> will change permissions until it encounters a block mapping, and will bail
>> out with a warning. Since there are no reports of this triggering, it
>> implies that there are currently no cases of code doing a vmalloc_huge()
>> followed by partial permission change. But this is a footgun waiting to
>> go off, so let's detect it early and avoid the possibility of permissions
>> in an intermediate state. So,  explicitly disallow changing permissions
>> for VM_ALLOW_HUGE_VMAP mappings.
>>
>> Reviewed-by: Ryan Roberts <ryan.roberts@arm.com>
>> Reviewed-by: Mike Rapoport (Microsoft) <rppt@kernel.org>
>> Signed-off-by: Dev Jain <dev.jain@arm.com>
>> ---
>> v1->v2:
>>   - Improve changelog, keep mention of page mappings in comment
>>
>>   arch/arm64/mm/pageattr.c | 6 +++---
>>   1 file changed, 3 insertions(+), 3 deletions(-)
>>
>> diff --git a/arch/arm64/mm/pageattr.c b/arch/arm64/mm/pageattr.c
>> index 39fd1f7ff02a..04d4a8f676db 100644
>> --- a/arch/arm64/mm/pageattr.c
>> +++ b/arch/arm64/mm/pageattr.c
>> @@ -96,8 +96,8 @@ static int change_memory_common(unsigned long addr, int
>> numpages,
>>        * we are operating on does not result in such splitting.
>>        *
>>        * Let's restrict ourselves to mappings created by vmalloc (or vmap).
>> -     * Those are guaranteed to consist entirely of page mappings, and
>> -     * splitting is never needed.
>> +     * Disallow VM_ALLOW_HUGE_VMAP mappings to guarantee that only page
>> +     * mappings are updated and splitting is never needed.
>>        *
>>        * So check whether the [addr, addr + size) interval is entirely
>>        * covered by precisely one VM area that has the VM_ALLOC flag set.
>> @@ -105,7 +105,7 @@ static int change_memory_common(unsigned long addr, int
>> numpages,
>>       area = find_vm_area((void *)addr);
>>       if (!area ||
>>           end > (unsigned long)kasan_reset_tag(area->addr) + area->size ||
>> -        !(area->flags & VM_ALLOC))
>> +        ((area->flags & (VM_ALLOC | VM_ALLOW_HUGE_VMAP)) != VM_ALLOC))
>>           return -EINVAL;
>>         if (!numpages)
> 
> Makes sense to me. Whenever required, we can improve the checks (or even try
> supporting splitting).

Yes that's the plan; I'd like to get to the point where we can do huge vmalloc
mappings by default on systems that support BBML2, then split mappings when needed.


> 
> Acked-by: David Hildenbrand <david@redhat.com>
>
diff mbox series

Patch

diff --git a/arch/arm64/mm/pageattr.c b/arch/arm64/mm/pageattr.c
index 39fd1f7ff02a..04d4a8f676db 100644
--- a/arch/arm64/mm/pageattr.c
+++ b/arch/arm64/mm/pageattr.c
@@ -96,8 +96,8 @@  static int change_memory_common(unsigned long addr, int numpages,
 	 * we are operating on does not result in such splitting.
 	 *
 	 * Let's restrict ourselves to mappings created by vmalloc (or vmap).
-	 * Those are guaranteed to consist entirely of page mappings, and
-	 * splitting is never needed.
+	 * Disallow VM_ALLOW_HUGE_VMAP mappings to guarantee that only page
+	 * mappings are updated and splitting is never needed.
 	 *
 	 * So check whether the [addr, addr + size) interval is entirely
 	 * covered by precisely one VM area that has the VM_ALLOC flag set.
@@ -105,7 +105,7 @@  static int change_memory_common(unsigned long addr, int numpages,
 	area = find_vm_area((void *)addr);
 	if (!area ||
 	    end > (unsigned long)kasan_reset_tag(area->addr) + area->size ||
-	    !(area->flags & VM_ALLOC))
+	    ((area->flags & (VM_ALLOC | VM_ALLOW_HUGE_VMAP)) != VM_ALLOC))
 		return -EINVAL;
 
 	if (!numpages)