[PATCH] block: do not warn when doing greedy allocation in folio_alloc_greedy()

[PATCH] block: do not warn when doing greedy allocation in folio_alloc_greedy()

[Qu Wenruo]

[WARNING]
During one of my local btrfs fstests runs, folio_alloc() inside
folio_alloc_greedy() triggered an allocation failure report when trying
to allocate an order-4 folio.

The kernel is from the latest development branch, which is utilizing
the IOMAP_DIO_BOUNCE flag for direct writes when the inode requires
checksum.

Unfortunately I didn't save the full log, only the function and the
order.

[CAUSE]
When the IOMAP_DIO_BOUNCE flag is utilized, we will hit the following call
chain:

 bio_iov_iter_bounce_write()
 |- folio_alloc_greedy()
    |- folio_alloc(gfp | __GFP_NORETRY, get_order(*size));

However __GFP_NORETRY will still emit an allocation failure report
when it fails.

And folio_alloc_greedy() will retry with a smaller order anyway, there
is no point in emitting that allocation failure report.

[FIX]
Append the __GFP_NOWARN flag to folio_alloc() for the larger-order folio
attempts.

Fixes: 8dd5e7c75d7b ("block: add helpers to bounce buffer an iov_iter into bios")
Signed-off-by: Qu Wenruo <wqu@suse.com>
---
 block/bio.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/block/bio.c b/block/bio.c
index f2a5f4d0a967..5ac954c70dd0 100644
--- a/block/bio.c
+++ b/block/bio.c
@@ -1285,7 +1285,8 @@ static struct folio *folio_alloc_greedy(gfp_t gfp, size_t *size,
 	struct folio *folio;
 
 	while (*size > minsize) {
-		folio = folio_alloc(gfp | __GFP_NORETRY, get_order(*size));
+		folio = folio_alloc(gfp | __GFP_NORETRY | __GFP_NOWARN,
+				    get_order(*size));
 		if (folio)
 			return folio;
 		*size = rounddown_pow_of_two(*size - 1);
-- 
2.54.0

Re: [PATCH] block: do not warn when doing greedy allocation in folio_alloc_greedy()

[Jens Axboe]

On Tue, 14 Jul 2026 18:37:39 +0930, Qu Wenruo wrote:
> [WARNING]
> During one of my local btrfs fstests runs, folio_alloc() inside
> folio_alloc_greedy() triggered an allocation failure report when trying
> to allocate an order-4 folio.
> 
> The kernel is from the latest development branch, which is utilizing
> the IOMAP_DIO_BOUNCE flag for direct writes when the inode requires
> checksum.
> 
> [...]

Applied, thanks!

[1/1] block: do not warn when doing greedy allocation in folio_alloc_greedy()
      commit: daff723f2d4c618d5f9186a990f47bf75878abeb

Best regards,
-- 
Jens Axboe

Re: [PATCH] block: do not warn when doing greedy allocation in folio_alloc_greedy()

[Christoph Hellwig]

On Tue, Jul 14, 2026 at 06:37:39PM +0930, Qu Wenruo wrote:
>  	while (*size > minsize) {
> -		folio = folio_alloc(gfp | __GFP_NORETRY, get_order(*size));
> +		folio = folio_alloc(gfp | __GFP_NORETRY | __GFP_NOWARN,
> +				    get_order(*size));

I though I read somewhere that we now don't have to specify
__GFP_NOWARN when using __GFP_NORETRY, but a quick dive into the page
allocator suggest that no such thing exists.

But can we make it happen?  Having to specify __GFP_NOWARN when the
caller clearly indicated that failure is perfectly fine using
__GFP_NORETRY is silly.  The same probably applies to
__GFP_RETRY_MAYFAIL.

Re: [PATCH] block: do not warn when doing greedy allocation in folio_alloc_greedy()

[Vlastimil Babka (SUSE)]

On 7/16/26 11:29, Christoph Hellwig wrote:
> On Tue, Jul 14, 2026 at 06:37:39PM +0930, Qu Wenruo wrote:
>>  	while (*size > minsize) {
>> -		folio = folio_alloc(gfp | __GFP_NORETRY, get_order(*size));
>> +		folio = folio_alloc(gfp | __GFP_NORETRY | __GFP_NOWARN,
>> +				    get_order(*size));
> 
> I though I read somewhere that we now don't have to specify
> __GFP_NOWARN when using __GFP_NORETRY, but a quick dive into the page
> allocator suggest that no such thing exists.

There was a change a while ago that added __GFP_NOWARN to GFP_NOWAIT, maybe
that's what you read about?

> But can we make it happen?  Having to specify __GFP_NOWARN when the
> caller clearly indicated that failure is perfectly fine using
> __GFP_NORETRY is silly.  The same probably applies to
> __GFP_RETRY_MAYFAIL.

It makes sense to me. The risk of missing some useful warning due to a
coding error (conjuring up an unexpected __GFP_NORETRY/__GFP_RETRY_MAYFAIL
in whatever way and then silently failing) is probably low here and it makes
the API better.

Re: [PATCH] block: do not warn when doing greedy allocation in folio_alloc_greedy()

[Qu Wenruo]

在 2026/7/16 19:35, Vlastimil Babka (SUSE) 写道:
> On 7/16/26 11:29, Christoph Hellwig wrote:
>> On Tue, Jul 14, 2026 at 06:37:39PM +0930, Qu Wenruo wrote:
>>>   	while (*size > minsize) {
>>> -		folio = folio_alloc(gfp | __GFP_NORETRY, get_order(*size));
>>> +		folio = folio_alloc(gfp | __GFP_NORETRY | __GFP_NOWARN,
>>> +				    get_order(*size));
>>
>> I though I read somewhere that we now don't have to specify
>> __GFP_NOWARN when using __GFP_NORETRY, but a quick dive into the page
>> allocator suggest that no such thing exists.
> 
> There was a change a while ago that added __GFP_NOWARN to GFP_NOWAIT, maybe
> that's what you read about?
> 
>> But can we make it happen?  Having to specify __GFP_NOWARN when the
>> caller clearly indicated that failure is perfectly fine using
>> __GFP_NORETRY is silly.  The same probably applies to
>> __GFP_RETRY_MAYFAIL.
> 
> It makes sense to me. The risk of missing some useful warning due to a
> coding error (conjuring up an unexpected __GFP_NORETRY/__GFP_RETRY_MAYFAIL
> in whatever way and then silently failing) is probably low here and it makes
> the API better.

Yep, making NORETRY and NOWAIT to imply NOWARN looks very reasonable to 
me too.

Thanks,
Qu