[PATCH v2 mm] mm: fix BUG splat with kvmalloc + GFP_ATOMIC

[PATCH v2 mm] mm: fix BUG splat with kvmalloc + GFP_ATOMIC

[Florian Westphal]

Martin Zaharinov reports BUG with 5.19.10 kernel:
 kernel BUG at mm/vmalloc.c:2437!
 invalid opcode: 0000 [#1] SMP
 CPU: 28 PID: 0 Comm: swapper/28 Tainted: G        W  O      5.19.9 #1
 [..]
 RIP: 0010:__get_vm_area_node+0x120/0x130
  __vmalloc_node_range+0x96/0x1e0
  kvmalloc_node+0x92/0xb0
  bucket_table_alloc.isra.0+0x47/0x140
  rhashtable_try_insert+0x3a4/0x440
  rhashtable_insert_slow+0x1b/0x30
 [..]

bucket_table_alloc uses kvzalloc(GPF_ATOMIC).  If kmalloc fails, this now
falls through to vmalloc and hits code paths that assume GFP_KERNEL.

Reported-by: Martin Zaharinov <micron10@gmail.com>
Fixes: a421ef303008 ("mm: allow !GFP_KERNEL allocations for kvmalloc")
Suggested-by: Michal Hocko <mhocko@suse.com>
Link: https://lore.kernel.org/linux-mm/Yy3MS2uhSgjF47dy@pc636/T/#t
Signed-off-by: Florian Westphal <fw@strlen.de>
---
 mm/util.c | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/mm/util.c b/mm/util.c
index c9439c66d8cf..346e40177bc6 100644
--- a/mm/util.c
+++ b/mm/util.c
@@ -619,6 +619,10 @@ void *kvmalloc_node(size_t size, gfp_t flags, int node)
 	if (ret || size <= PAGE_SIZE)
 		return ret;
 
+	/* non-sleeping allocations are not supported by vmalloc */
+	if (!gfpflags_allow_blocking(flags))
+		return NULL;
+
 	/* Don't even allow crazy sizes */
 	if (unlikely(size > INT_MAX)) {
 		WARN_ON_ONCE(!(flags & __GFP_NOWARN));
-- 
2.35.1

Re: [PATCH v2 mm] mm: fix BUG splat with kvmalloc + GFP_ATOMIC

[Michal Hocko]

On Mon 26-09-22 17:16:50, Florian Westphal wrote:
> Martin Zaharinov reports BUG with 5.19.10 kernel:
>  kernel BUG at mm/vmalloc.c:2437!
>  invalid opcode: 0000 [#1] SMP
>  CPU: 28 PID: 0 Comm: swapper/28 Tainted: G        W  O      5.19.9 #1
>  [..]
>  RIP: 0010:__get_vm_area_node+0x120/0x130
>   __vmalloc_node_range+0x96/0x1e0
>   kvmalloc_node+0x92/0xb0
>   bucket_table_alloc.isra.0+0x47/0x140
>   rhashtable_try_insert+0x3a4/0x440
>   rhashtable_insert_slow+0x1b/0x30
>  [..]
> 
> bucket_table_alloc uses kvzalloc(GPF_ATOMIC).  If kmalloc fails, this now
> falls through to vmalloc and hits code paths that assume GFP_KERNEL.
> 
> Reported-by: Martin Zaharinov <micron10@gmail.com>
> Fixes: a421ef303008 ("mm: allow !GFP_KERNEL allocations for kvmalloc")
> Suggested-by: Michal Hocko <mhocko@suse.com>
> Link: https://lore.kernel.org/linux-mm/Yy3MS2uhSgjF47dy@pc636/T/#t
> Signed-off-by: Florian Westphal <fw@strlen.de>

Maybe we can extend this to GFP_NOWAIT capable vmalloc call but I am not
sure this exists or whether this is something vmalloc maintainers want
to commit to.

This is a clear and trivial fixup of the above commit. I would add this
to the stable tree as well.

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

Thanks!

> ---
>  mm/util.c | 4 ++++
>  1 file changed, 4 insertions(+)
> 
> diff --git a/mm/util.c b/mm/util.c
> index c9439c66d8cf..346e40177bc6 100644
> --- a/mm/util.c
> +++ b/mm/util.c
> @@ -619,6 +619,10 @@ void *kvmalloc_node(size_t size, gfp_t flags, int node)
>  	if (ret || size <= PAGE_SIZE)
>  		return ret;
>  
> +	/* non-sleeping allocations are not supported by vmalloc */
> +	if (!gfpflags_allow_blocking(flags))
> +		return NULL;
> +
>  	/* Don't even allow crazy sizes */
>  	if (unlikely(size > INT_MAX)) {
>  		WARN_ON_ONCE(!(flags & __GFP_NOWARN));
> -- 
> 2.35.1

-- 
Michal Hocko
SUSE Labs