[PATCH 0/3] md: remove redundant __GFP_NOWARN

[PATCH 0/3] md: remove redundant __GFP_NOWARN

[Qianfeng Rong]

Commit 16f5dfbc851b ("gfp: include __GFP_NOWARN in GFP_NOWAIT") made
GFP_NOWAIT implicitly include __GFP_NOWARN.

Therefore, explicit __GFP_NOWARN combined with GFP_NOWAIT (e.g.,
`GFP_NOWAIT | __GFP_NOWARN`) is now redundant.  Let's clean up these
redundant flags across subsystems.

No functional changes.

Qianfeng Rong (3):
  bcache: remove redundant __GFP_NOWARN
  dm bufio: remove redundant __GFP_NOWARN
  md/raid5: remove redundant __GFP_NOWARN

 drivers/md/bcache/btree.c |  2 +-
 drivers/md/dm-bufio.c     | 10 +++++-----
 drivers/md/raid5-cache.c  |  2 +-
 3 files changed, 7 insertions(+), 7 deletions(-)

-- 
2.34.1

[PATCH 1/3] bcache: remove redundant __GFP_NOWARN

[Qianfeng Rong]

GFP_NOWAIT already includes __GFP_NOWARN, so let's remove the redundant
__GFP_NOWARN.

Signed-off-by: Qianfeng Rong <rongqianfeng@vivo.com>
---
 drivers/md/bcache/btree.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/md/bcache/btree.c b/drivers/md/bcache/btree.c
index 210b59007d98..a764ec149d7a 100644
--- a/drivers/md/bcache/btree.c
+++ b/drivers/md/bcache/btree.c
@@ -371,7 +371,7 @@ static void do_btree_node_write(struct btree *b)
 	SET_PTR_OFFSET(&k.key, 0, PTR_OFFSET(&k.key, 0) +
 		       bset_sector_offset(&b->keys, i));
 
-	if (!bch_bio_alloc_pages(b->bio, __GFP_NOWARN|GFP_NOWAIT)) {
+	if (!bch_bio_alloc_pages(b->bio, GFP_NOWAIT)) {
 		struct bio_vec *bv;
 		void *addr = (void *) ((unsigned long) i & ~(PAGE_SIZE - 1));
 		struct bvec_iter_all iter_all;
-- 
2.34.1

Re: [PATCH 1/3] bcache: remove redundant __GFP_NOWARN

[Coly Li]

On Mon, Aug 11, 2025 at 08:36:35PM +0800, Qianfeng Rong wrote:
> GFP_NOWAIT already includes __GFP_NOWARN, so let's remove the redundant
> __GFP_NOWARN.
> 
> Signed-off-by: Qianfeng Rong <rongqianfeng@vivo.com>

Acked-by: Coly Li <colyli@kernel.org>

Do you want this patch to go upstream in my submission path?

Thanks.

Coly Li

> ---
>  drivers/md/bcache/btree.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/md/bcache/btree.c b/drivers/md/bcache/btree.c
> index 210b59007d98..a764ec149d7a 100644
> --- a/drivers/md/bcache/btree.c
> +++ b/drivers/md/bcache/btree.c
> @@ -371,7 +371,7 @@ static void do_btree_node_write(struct btree *b)
>  	SET_PTR_OFFSET(&k.key, 0, PTR_OFFSET(&k.key, 0) +
>  		       bset_sector_offset(&b->keys, i));
>  
> -	if (!bch_bio_alloc_pages(b->bio, __GFP_NOWARN|GFP_NOWAIT)) {
> +	if (!bch_bio_alloc_pages(b->bio, GFP_NOWAIT)) {
>  		struct bio_vec *bv;
>  		void *addr = (void *) ((unsigned long) i & ~(PAGE_SIZE - 1));
>  		struct bvec_iter_all iter_all;
> -- 
> 2.34.1
> 

Re: [PATCH 1/3] bcache: remove redundant __GFP_NOWARN

[Qianfeng Rong]

在 2025/8/11 21:22, Coly Li 写道:
> On Mon, Aug 11, 2025 at 08:36:35PM +0800, Qianfeng Rong wrote:
>> GFP_NOWAIT already includes __GFP_NOWARN, so let's remove the redundant
>> __GFP_NOWARN.
>>
>> Signed-off-by: Qianfeng Rong <rongqianfeng@vivo.com>
> Acked-by: Coly Li <colyli@kernel.org>
>
> Do you want this patch to go upstream in my submission path?
Yes, please include this patch in your submission path to upstream.

Thank you for your review and Acked-by tag.

Best regards,
Qianfeng
>
> Thanks.
>
> Coly Li
>
>> ---
>>   drivers/md/bcache/btree.c | 2 +-
>>   1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/drivers/md/bcache/btree.c b/drivers/md/bcache/btree.c
>> index 210b59007d98..a764ec149d7a 100644
>> --- a/drivers/md/bcache/btree.c
>> +++ b/drivers/md/bcache/btree.c
>> @@ -371,7 +371,7 @@ static void do_btree_node_write(struct btree *b)
>>   	SET_PTR_OFFSET(&k.key, 0, PTR_OFFSET(&k.key, 0) +
>>   		       bset_sector_offset(&b->keys, i));
>>   
>> -	if (!bch_bio_alloc_pages(b->bio, __GFP_NOWARN|GFP_NOWAIT)) {
>> +	if (!bch_bio_alloc_pages(b->bio, GFP_NOWAIT)) {
>>   		struct bio_vec *bv;
>>   		void *addr = (void *) ((unsigned long) i & ~(PAGE_SIZE - 1));
>>   		struct bvec_iter_all iter_all;
>> -- 
>> 2.34.1
>>

[PATCH 2/3] dm bufio: remove redundant __GFP_NOWARN

[Qianfeng Rong]

GFP_NOWAIT already includes __GFP_NOWARN, so let's remove the redundant
__GFP_NOWARN.  Also update comments to clarify the flag semantics.

Signed-off-by: Qianfeng Rong <rongqianfeng@vivo.com>
---
 drivers/md/dm-bufio.c | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)

diff --git a/drivers/md/dm-bufio.c b/drivers/md/dm-bufio.c
index ff7595caf440..4b20854e92f5 100644
--- a/drivers/md/dm-bufio.c
+++ b/drivers/md/dm-bufio.c
@@ -1337,7 +1337,7 @@ static void use_bio(struct dm_buffer *b, enum req_op op, sector_t sector,
 	char *ptr;
 	unsigned int len;
 
-	bio = bio_kmalloc(1, GFP_NOWAIT | __GFP_NORETRY | __GFP_NOWARN);
+	bio = bio_kmalloc(1, GFP_NOWAIT | __GFP_NORETRY);
 	if (!bio) {
 		use_dmio(b, op, sector, n_sectors, offset, ioprio);
 		return;
@@ -1601,18 +1601,18 @@ static struct dm_buffer *__alloc_buffer_wait_no_callback(struct dm_bufio_client
 	 * dm-bufio is resistant to allocation failures (it just keeps
 	 * one buffer reserved in cases all the allocations fail).
 	 * So set flags to not try too hard:
-	 *	GFP_NOWAIT: don't wait; if we need to sleep we'll release our
-	 *		    mutex and wait ourselves.
+	 *	GFP_NOWAIT: don't wait and don't print a warning in case of
+	 *		    failure; if we need to sleep we'll release our mutex
+	 *		    and wait ourselves.
 	 *	__GFP_NORETRY: don't retry and rather return failure
 	 *	__GFP_NOMEMALLOC: don't use emergency reserves
-	 *	__GFP_NOWARN: don't print a warning in case of failure
 	 *
 	 * For debugging, if we set the cache size to 1, no new buffers will
 	 * be allocated.
 	 */
 	while (1) {
 		if (dm_bufio_cache_size_latch != 1) {
-			b = alloc_buffer(c, GFP_NOWAIT | __GFP_NORETRY | __GFP_NOMEMALLOC | __GFP_NOWARN);
+			b = alloc_buffer(c, GFP_NOWAIT | __GFP_NORETRY | __GFP_NOMEMALLOC);
 			if (b)
 				return b;
 		}
-- 
2.34.1

Re: [PATCH 2/3] dm bufio: remove redundant __GFP_NOWARN

[Mikulas Patocka]

Hi

I think that GFP_NOWAIT already includes __GFP_NORETRY too. So, should we 
drop __GFP_NORETRY as well?

Mikulas


On Mon, 11 Aug 2025, Qianfeng Rong wrote:

> GFP_NOWAIT already includes __GFP_NOWARN, so let's remove the redundant
> __GFP_NOWARN.  Also update comments to clarify the flag semantics.
> 
> Signed-off-by: Qianfeng Rong <rongqianfeng@vivo.com>
> ---
>  drivers/md/dm-bufio.c | 10 +++++-----
>  1 file changed, 5 insertions(+), 5 deletions(-)
> 
> diff --git a/drivers/md/dm-bufio.c b/drivers/md/dm-bufio.c
> index ff7595caf440..4b20854e92f5 100644
> --- a/drivers/md/dm-bufio.c
> +++ b/drivers/md/dm-bufio.c
> @@ -1337,7 +1337,7 @@ static void use_bio(struct dm_buffer *b, enum req_op op, sector_t sector,
>  	char *ptr;
>  	unsigned int len;
>  
> -	bio = bio_kmalloc(1, GFP_NOWAIT | __GFP_NORETRY | __GFP_NOWARN);
> +	bio = bio_kmalloc(1, GFP_NOWAIT | __GFP_NORETRY);
>  	if (!bio) {
>  		use_dmio(b, op, sector, n_sectors, offset, ioprio);
>  		return;
> @@ -1601,18 +1601,18 @@ static struct dm_buffer *__alloc_buffer_wait_no_callback(struct dm_bufio_client
>  	 * dm-bufio is resistant to allocation failures (it just keeps
>  	 * one buffer reserved in cases all the allocations fail).
>  	 * So set flags to not try too hard:
> -	 *	GFP_NOWAIT: don't wait; if we need to sleep we'll release our
> -	 *		    mutex and wait ourselves.
> +	 *	GFP_NOWAIT: don't wait and don't print a warning in case of
> +	 *		    failure; if we need to sleep we'll release our mutex
> +	 *		    and wait ourselves.
>  	 *	__GFP_NORETRY: don't retry and rather return failure
>  	 *	__GFP_NOMEMALLOC: don't use emergency reserves
> -	 *	__GFP_NOWARN: don't print a warning in case of failure
>  	 *
>  	 * For debugging, if we set the cache size to 1, no new buffers will
>  	 * be allocated.
>  	 */
>  	while (1) {
>  		if (dm_bufio_cache_size_latch != 1) {
> -			b = alloc_buffer(c, GFP_NOWAIT | __GFP_NORETRY | __GFP_NOMEMALLOC | __GFP_NOWARN);
> +			b = alloc_buffer(c, GFP_NOWAIT | __GFP_NORETRY | __GFP_NOMEMALLOC);
>  			if (b)
>  				return b;
>  		}
> -- 
> 2.34.1
> 

Re: [PATCH 2/3] dm bufio: remove redundant __GFP_NOWARN

[Qianfeng Rong]

在 2025/8/11 20:44, Mikulas Patocka 写道:
> Hi
>
> I think that GFP_NOWAIT already includes __GFP_NORETRY too. So, should we
> drop __GFP_NORETRY as well?
GFP_NOWAIT does not include __GFP_NORETRY:
#define GFP_NOWAIT (__GFP_KSWAPD_RECLAIM | __GFP_NOWARN)

GFP_NOWAIT tells the memory manager to only wake up kswapd to perform
memory reclamation, not to perform direct memory reclaim.  Even if the
request fails, no error message is printed.

Best regards,
Qianfeng
>
> Mikulas
>

Re: [PATCH 2/3] dm bufio: remove redundant __GFP_NOWARN

[Mikulas Patocka]

[-- Attachment #1: Type: text/plain, Size: 666 bytes --]



On Mon, 11 Aug 2025, Qianfeng Rong wrote:

> 
> 在 2025/8/11 20:44, Mikulas Patocka 写道:
> > Hi
> > 
> > I think that GFP_NOWAIT already includes __GFP_NORETRY too. So, should we
> > drop __GFP_NORETRY as well?
>
> GFP_NOWAIT does not include __GFP_NORETRY:
> #define GFP_NOWAIT (__GFP_KSWAPD_RECLAIM | __GFP_NOWARN)
> 
> GFP_NOWAIT tells the memory manager to only wake up kswapd to perform
> memory reclamation, not to perform direct memory reclaim.  Even if the
> request fails, no error message is printed.
> 
> Best regards,
> Qianfeng

Yes, but if GFP_NOWAIT allocation can't sleep, it can't retry - thus 
GFP_NOWAIT should imply __GFP_NORETRY.

Mikulas

Re: [PATCH 2/3] dm bufio: remove redundant __GFP_NOWARN

[Qianfeng Rong]

在 2025/8/11 23:10, Mikulas Patocka 写道:
>
> On Mon, 11 Aug 2025, Qianfeng Rong wrote:
>
>> 在 2025/8/11 20:44, Mikulas Patocka 写道:
>>> Hi
>>>
>>> I think that GFP_NOWAIT already includes __GFP_NORETRY too. So, should we
>>> drop __GFP_NORETRY as well?
>> GFP_NOWAIT does not include __GFP_NORETRY:
>> #define GFP_NOWAIT (__GFP_KSWAPD_RECLAIM | __GFP_NOWARN)
>>
>> GFP_NOWAIT tells the memory manager to only wake up kswapd to perform
>> memory reclamation, not to perform direct memory reclaim.  Even if the
>> request fails, no error message is printed.
>>
>> Best regards,
>> Qianfeng
> Yes, but if GFP_NOWAIT allocation can't sleep, it can't retry - thus
> GFP_NOWAIT should imply __GFP_NORETRY.
>
> Mikulas
I strongly agree with your perspective, but I'm uncertain whether
GFP_NOWAIT should include __GFP_NORETRY, or if we should add a
note stating GFP_NOWAIT/GFP_ATOMIC should not be combined with
__GFP_NORETRY since it becomes redundant in atomic contexts.  I'll
raise this issue to see if others have insights.

Best regards,
Qianfeng

[PATCH 3/3] md/raid5: remove redundant __GFP_NOWARN

[Qianfeng Rong]

GFP_NOWAIT already includes __GFP_NOWARN, so let's remove the redundant
__GFP_NOWARN.

Signed-off-by: Qianfeng Rong <rongqianfeng@vivo.com>
---
 drivers/md/raid5-cache.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/md/raid5-cache.c b/drivers/md/raid5-cache.c
index ba768ca7f422..e29e69335c69 100644
--- a/drivers/md/raid5-cache.c
+++ b/drivers/md/raid5-cache.c
@@ -3104,7 +3104,7 @@ int r5l_init_log(struct r5conf *conf, struct md_rdev *rdev)
 		goto out_mempool;
 
 	spin_lock_init(&log->tree_lock);
-	INIT_RADIX_TREE(&log->big_stripe_tree, GFP_NOWAIT | __GFP_NOWARN);
+	INIT_RADIX_TREE(&log->big_stripe_tree, GFP_NOWAIT);
 
 	thread = md_register_thread(r5l_reclaim_thread, log->rdev->mddev,
 				    "reclaim");
-- 
2.34.1