[PATCH v3] vmalloc: Fix accounting with i915

[PATCH v3] vmalloc: Fix accounting with i915

[Matthew Wilcox (Oracle)]

If the caller of vmap() specifies VM_MAP_PUT_PAGES (currently only the
i915 driver), we will decrement nr_vmalloc_pages and MEMCG_VMALLOC in
vfree().  These counters are incremented by vmalloc() but not by vmap()
so this will cause an underflow.  Check the VM_MAP_PUT_PAGES flag before
decrementing either counter.

Fixes: b944afc9d64d (mm: add a VM_MAP_PUT_PAGES flag for vmap)
Cc: stable@vger.kernel.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
---
 mm/vmalloc.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/mm/vmalloc.c b/mm/vmalloc.c
index f009b21705c1..5c88d0e90c20 100644
--- a/mm/vmalloc.c
+++ b/mm/vmalloc.c
@@ -3374,7 +3374,8 @@ void vfree(const void *addr)
 		struct page *page = vm->pages[i];
 
 		BUG_ON(!page);
-		mod_memcg_page_state(page, MEMCG_VMALLOC, -1);
+		if (!(vm->flags & VM_MAP_PUT_PAGES))
+			mod_memcg_page_state(page, MEMCG_VMALLOC, -1);
 		/*
 		 * High-order allocs for huge vmallocs are split, so
 		 * can be freed as an array of order-0 allocations
@@ -3382,7 +3383,8 @@ void vfree(const void *addr)
 		__free_page(page);
 		cond_resched();
 	}
-	atomic_long_sub(vm->nr_pages, &nr_vmalloc_pages);
+	if (!(vm->flags & VM_MAP_PUT_PAGES))
+		atomic_long_sub(vm->nr_pages, &nr_vmalloc_pages);
 	kvfree(vm->pages);
 	kfree(vm);
 }
-- 
2.45.2

Re: [PATCH v3] vmalloc: Fix accounting with i915

[Shakeel Butt]

On Wed, Dec 11, 2024 at 08:25:37PM +0000, Matthew Wilcox (Oracle) wrote:
> If the caller of vmap() specifies VM_MAP_PUT_PAGES (currently only the
> i915 driver), we will decrement nr_vmalloc_pages and MEMCG_VMALLOC in
> vfree().  These counters are incremented by vmalloc() but not by vmap()
> so this will cause an underflow.  Check the VM_MAP_PUT_PAGES flag before
> decrementing either counter.
> 
> Fixes: b944afc9d64d (mm: add a VM_MAP_PUT_PAGES flag for vmap)
> Cc: stable@vger.kernel.org
> Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
> Acked-by: Johannes Weiner <hannes@cmpxchg.org>

Oh you merged them into one. Good.

Reviewed-by: Shakeel Butt <shakeel.butt@linux.dev>

Re: [PATCH v3] vmalloc: Fix accounting with i915

[Balbir Singh]

On 12/12/24 07:25, Matthew Wilcox (Oracle) wrote:
> If the caller of vmap() specifies VM_MAP_PUT_PAGES (currently only the
> i915 driver), we will decrement nr_vmalloc_pages and MEMCG_VMALLOC in
> vfree().  These counters are incremented by vmalloc() but not by vmap()
> so this will cause an underflow.  Check the VM_MAP_PUT_PAGES flag before
> decrementing either counter.
> 
> Fixes: b944afc9d64d (mm: add a VM_MAP_PUT_PAGES flag for vmap)
> Cc: stable@vger.kernel.org
> Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
> Acked-by: Johannes Weiner <hannes@cmpxchg.org>
> ---
>  mm/vmalloc.c | 6 ++++--
>  1 file changed, 4 insertions(+), 2 deletions(-)
> 
> diff --git a/mm/vmalloc.c b/mm/vmalloc.c
> index f009b21705c1..5c88d0e90c20 100644
> --- a/mm/vmalloc.c
> +++ b/mm/vmalloc.c
> @@ -3374,7 +3374,8 @@ void vfree(const void *addr)
>  		struct page *page = vm->pages[i];
>  
>  		BUG_ON(!page);
> -		mod_memcg_page_state(page, MEMCG_VMALLOC, -1);
> +		if (!(vm->flags & VM_MAP_PUT_PAGES))
> +			mod_memcg_page_state(page, MEMCG_VMALLOC, -1);
>  		/*
>  		 * High-order allocs for huge vmallocs are split, so
>  		 * can be freed as an array of order-0 allocations
> @@ -3382,7 +3383,8 @@ void vfree(const void *addr)
>  		__free_page(page);
>  		cond_resched();
>  	}
> -	atomic_long_sub(vm->nr_pages, &nr_vmalloc_pages);
> +	if (!(vm->flags & VM_MAP_PUT_PAGES))
> +		atomic_long_sub(vm->nr_pages, &nr_vmalloc_pages);
>  	kvfree(vm->pages);
>  	kfree(vm);
>  }

I found another user kunit_iov_vector.c, but even it though it uses
VM_MAP_PUT_PAGES, it does not call into vfree() (which I need to check
for further bugs)

Reviewed-by: Balbir Singh <balbirs@nvidia.com>

Re: [PATCH v3] vmalloc: Fix accounting with i915

[Johannes Weiner]

On Wed, Dec 11, 2024 at 08:25:37PM +0000, Matthew Wilcox (Oracle) wrote:
> If the caller of vmap() specifies VM_MAP_PUT_PAGES (currently only the
> i915 driver), we will decrement nr_vmalloc_pages and MEMCG_VMALLOC in
> vfree().  These counters are incremented by vmalloc() but not by vmap()
> so this will cause an underflow.  Check the VM_MAP_PUT_PAGES flag before
> decrementing either counter.
> 
> Fixes: b944afc9d64d (mm: add a VM_MAP_PUT_PAGES flag for vmap)
> Cc: stable@vger.kernel.org
> Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
> Acked-by: Johannes Weiner <hannes@cmpxchg.org>

Nice, thanks, looks good to me.

Re: [PATCH v3] vmalloc: Fix accounting with i915

[Michal Hocko]

On Wed 11-12-24 20:25:37, Matthew Wilcox wrote:
> If the caller of vmap() specifies VM_MAP_PUT_PAGES (currently only the
> i915 driver), we will decrement nr_vmalloc_pages and MEMCG_VMALLOC in
> vfree().  These counters are incremented by vmalloc() but not by vmap()
> so this will cause an underflow.  Check the VM_MAP_PUT_PAGES flag before
> decrementing either counter.
> 
> Fixes: b944afc9d64d (mm: add a VM_MAP_PUT_PAGES flag for vmap)
> Cc: stable@vger.kernel.org
> Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
> Acked-by: Johannes Weiner <hannes@cmpxchg.org>

Acked-by: Michal Hocko <mhocko@suse.com>
Thanks!

> ---
>  mm/vmalloc.c | 6 ++++--
>  1 file changed, 4 insertions(+), 2 deletions(-)
> 
> diff --git a/mm/vmalloc.c b/mm/vmalloc.c
> index f009b21705c1..5c88d0e90c20 100644
> --- a/mm/vmalloc.c
> +++ b/mm/vmalloc.c
> @@ -3374,7 +3374,8 @@ void vfree(const void *addr)
>  		struct page *page = vm->pages[i];
>  
>  		BUG_ON(!page);
> -		mod_memcg_page_state(page, MEMCG_VMALLOC, -1);
> +		if (!(vm->flags & VM_MAP_PUT_PAGES))
> +			mod_memcg_page_state(page, MEMCG_VMALLOC, -1);
>  		/*
>  		 * High-order allocs for huge vmallocs are split, so
>  		 * can be freed as an array of order-0 allocations
> @@ -3382,7 +3383,8 @@ void vfree(const void *addr)
>  		__free_page(page);
>  		cond_resched();
>  	}
> -	atomic_long_sub(vm->nr_pages, &nr_vmalloc_pages);
> +	if (!(vm->flags & VM_MAP_PUT_PAGES))
> +		atomic_long_sub(vm->nr_pages, &nr_vmalloc_pages);
>  	kvfree(vm->pages);
>  	kfree(vm);
>  }
> -- 
> 2.45.2

-- 
Michal Hocko
SUSE Labs