linux-erofs.lists.ozlabs.org archive mirror
 help / color / mirror / Atom feed
* [PATCH] Remove WQ_CPU_INTENSIVE flag from unbound wq's
@ 2020-02-13 14:18 Maksym Planeta
  2020-02-13 15:36 ` Mike Snitzer
                   ` (2 more replies)
  0 siblings, 3 replies; 6+ messages in thread
From: Maksym Planeta @ 2020-02-13 14:18 UTC (permalink / raw)
  To: mplaneta, Zhou Wang, Herbert Xu, David S. Miller,
	Alasdair Kergon, Mike Snitzer, dm-devel, Song Liu, Gao Xiang,
	Chao Yu, linux-crypto, linux-kernel, linux-raid, linux-erofs

The documentation [1] says that WQ_CPU_INTENSIVE is "meaningless" for
unbound wq. I remove this flag from places where unbound queue is
allocated. This is supposed to improve code readability.

1. https://www.kernel.org/doc/html/latest/core-api/workqueue.html#flags

Signed-off-by: Maksym Planeta <mplaneta@os.inf.tu-dresden.de>
---
 drivers/crypto/hisilicon/qm.c | 3 +--
 drivers/md/dm-crypt.c         | 2 +-
 drivers/md/dm-verity-target.c | 2 +-
 drivers/md/raid5.c            | 2 +-
 fs/erofs/zdata.c              | 2 +-
 5 files changed, 5 insertions(+), 6 deletions(-)

diff --git a/drivers/crypto/hisilicon/qm.c b/drivers/crypto/hisilicon/qm.c
index b57da5ef8b5b..4a39cb2c6a0b 100644
--- a/drivers/crypto/hisilicon/qm.c
+++ b/drivers/crypto/hisilicon/qm.c
@@ -1148,8 +1148,7 @@ struct hisi_qp *hisi_qm_create_qp(struct hisi_qm *qm, u8 alg_type)
 	qp->qp_id = qp_id;
 	qp->alg_type = alg_type;
 	INIT_WORK(&qp->work, qm_qp_work_func);
-	qp->wq = alloc_workqueue("hisi_qm", WQ_UNBOUND | WQ_HIGHPRI |
-				 WQ_CPU_INTENSIVE | WQ_MEM_RECLAIM, 0);
+	qp->wq = alloc_workqueue("hisi_qm", WQ_UNBOUND | WQ_HIGHPRI | WQ_MEM_RECLAIM, 0);
 	if (!qp->wq) {
 		ret = -EFAULT;
 		goto err_free_qp_mem;
diff --git a/drivers/md/dm-crypt.c b/drivers/md/dm-crypt.c
index c6a529873d0f..44d56325fa27 100644
--- a/drivers/md/dm-crypt.c
+++ b/drivers/md/dm-crypt.c
@@ -3032,7 +3032,7 @@ static int crypt_ctr(struct dm_target *ti, unsigned int argc, char **argv)
 						  1, devname);
 	else
 		cc->crypt_queue = alloc_workqueue("kcryptd/%s",
-						  WQ_CPU_INTENSIVE | WQ_MEM_RECLAIM | WQ_UNBOUND,
+						  WQ_MEM_RECLAIM | WQ_UNBOUND,
 						  num_online_cpus(), devname);
 	if (!cc->crypt_queue) {
 		ti->error = "Couldn't create kcryptd queue";
diff --git a/drivers/md/dm-verity-target.c b/drivers/md/dm-verity-target.c
index 0d61e9c67986..20f92c7ea07e 100644
--- a/drivers/md/dm-verity-target.c
+++ b/drivers/md/dm-verity-target.c
@@ -1190,7 +1190,7 @@ static int verity_ctr(struct dm_target *ti, unsigned argc, char **argv)
 	}
 
 	/* WQ_UNBOUND greatly improves performance when running on ramdisk */
-	v->verify_wq = alloc_workqueue("kverityd", WQ_CPU_INTENSIVE | WQ_MEM_RECLAIM | WQ_UNBOUND, num_online_cpus());
+	v->verify_wq = alloc_workqueue("kverityd", WQ_MEM_RECLAIM | WQ_UNBOUND, num_online_cpus());
 	if (!v->verify_wq) {
 		ti->error = "Cannot allocate workqueue";
 		r = -ENOMEM;
diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c
index ba00e9877f02..cd93a1731b82 100644
--- a/drivers/md/raid5.c
+++ b/drivers/md/raid5.c
@@ -8481,7 +8481,7 @@ static int __init raid5_init(void)
 	int ret;
 
 	raid5_wq = alloc_workqueue("raid5wq",
-		WQ_UNBOUND|WQ_MEM_RECLAIM|WQ_CPU_INTENSIVE|WQ_SYSFS, 0);
+		WQ_UNBOUND|WQ_MEM_RECLAIM|WQ_SYSFS, 0);
 	if (!raid5_wq)
 		return -ENOMEM;
 
diff --git a/fs/erofs/zdata.c b/fs/erofs/zdata.c
index 80e47f07d946..b2a679f720e9 100644
--- a/fs/erofs/zdata.c
+++ b/fs/erofs/zdata.c
@@ -43,7 +43,7 @@ void z_erofs_exit_zip_subsystem(void)
 static inline int z_erofs_init_workqueue(void)
 {
 	const unsigned int onlinecpus = num_possible_cpus();
-	const unsigned int flags = WQ_UNBOUND | WQ_HIGHPRI | WQ_CPU_INTENSIVE;
+	const unsigned int flags = WQ_UNBOUND | WQ_HIGHPRI;
 
 	/*
 	 * no need to spawn too many threads, limiting threads could minimum
-- 
2.24.1


^ permalink raw reply related	[flat|nested] 6+ messages in thread
* Re: Remove WQ_CPU_INTENSIVE flag from unbound wq's
  2020-02-13 14:18 [PATCH] Remove WQ_CPU_INTENSIVE flag from unbound wq's Maksym Planeta
@ 2020-02-13 15:36 ` Mike Snitzer
  2020-02-13 16:19   ` Maksym Planeta
  2020-02-14  1:39 ` [PATCH] " Gao Xiang
  2020-02-14  3:47 ` Zhou Wang
  2 siblings, 1 reply; 6+ messages in thread
From: Mike Snitzer @ 2020-02-13 15:36 UTC (permalink / raw)
  To: Maksym Planeta
  Cc: Herbert Xu, dm-devel, linux-kernel, linux-raid, Song Liu,
	Zhou Wang, linux-crypto, Gao Xiang, linux-erofs, David S. Miller,
	Alasdair Kergon

On Thu, Feb 13 2020 at  9:18am -0500,
Maksym Planeta <mplaneta@os.inf.tu-dresden.de> wrote:

> The documentation [1] says that WQ_CPU_INTENSIVE is "meaningless" for
> unbound wq. I remove this flag from places where unbound queue is
> allocated. This is supposed to improve code readability.
> 
> 1. https://www.kernel.org/doc/html/latest/core-api/workqueue.html#flags
> 
> Signed-off-by: Maksym Planeta <mplaneta@os.inf.tu-dresden.de>

What the Documentation says aside, have you cross referenced with the
code?  And/or have you done benchmarks to verify no changes?

Thanks,
Mike


^ permalink raw reply	[flat|nested] 6+ messages in thread
* Re: Remove WQ_CPU_INTENSIVE flag from unbound wq's
  2020-02-13 15:36 ` Mike Snitzer
@ 2020-02-13 16:19   ` Maksym Planeta
  2020-02-13 23:42     ` Song Liu
  0 siblings, 1 reply; 6+ messages in thread
From: Maksym Planeta @ 2020-02-13 16:19 UTC (permalink / raw)
  To: Mike Snitzer
  Cc: Herbert Xu, dm-devel, linux-kernel, linux-raid, Song Liu,
	Zhou Wang, linux-crypto, Gao Xiang, linux-erofs, David S. Miller,
	Alasdair Kergon



On 13/02/2020 16:36, Mike Snitzer wrote:
> On Thu, Feb 13 2020 at  9:18am -0500,
> Maksym Planeta <mplaneta@os.inf.tu-dresden.de> wrote:
> 
>> The documentation [1] says that WQ_CPU_INTENSIVE is "meaningless" for
>> unbound wq. I remove this flag from places where unbound queue is
>> allocated. This is supposed to improve code readability.
>>
>> 1. https://www.kernel.org/doc/html/latest/core-api/workqueue.html#flags
>>
>> Signed-off-by: Maksym Planeta <mplaneta@os.inf.tu-dresden.de>
> 
> What the Documentation says aside, have you cross referenced with the
> code?  And/or have you done benchmarks to verify no changes?
> 

It seems so from the code. Although, I'm not 100% confident. I did not 
run benchmarks, instead I relied that on the assumption that 
documentation is correct.

> Thanks,
> Mike
> 

-- 
Regards,
Maksym Planeta

^ permalink raw reply	[flat|nested] 6+ messages in thread
* Re: Remove WQ_CPU_INTENSIVE flag from unbound wq's
  2020-02-13 16:19   ` Maksym Planeta
@ 2020-02-13 23:42     ` Song Liu
  0 siblings, 0 replies; 6+ messages in thread
From: Song Liu @ 2020-02-13 23:42 UTC (permalink / raw)
  To: Maksym Planeta
  Cc: Herbert Xu, Mike Snitzer, Zhou Wang, open list, linux-raid,
	dm-devel, linux-crypto, Gao Xiang, linux-erofs, David S. Miller,
	Alasdair Kergon

On Thu, Feb 13, 2020 at 8:19 AM Maksym Planeta
<mplaneta@os.inf.tu-dresden.de> wrote:
>
>
>
> On 13/02/2020 16:36, Mike Snitzer wrote:
> > On Thu, Feb 13 2020 at  9:18am -0500,
> > Maksym Planeta <mplaneta@os.inf.tu-dresden.de> wrote:
> >
> >> The documentation [1] says that WQ_CPU_INTENSIVE is "meaningless" for
> >> unbound wq. I remove this flag from places where unbound queue is
> >> allocated. This is supposed to improve code readability.
> >>
> >> 1. https://www.kernel.org/doc/html/latest/core-api/workqueue.html#flags
> >>
> >> Signed-off-by: Maksym Planeta <mplaneta@os.inf.tu-dresden.de>
> >
> > What the Documentation says aside, have you cross referenced with the
> > code?  And/or have you done benchmarks to verify no changes?
> >
>
> It seems so from the code. Although, I'm not 100% confident. I did not
> run benchmarks, instead I relied that on the assumption that
> documentation is correct.

From the code, WQ_CPU_INTENSIVE is only used to set
WORKER_CPU_INTENSIVE, and WORKER_CPU_INTENSIVE is only used
as part of WORKER_NOT_RUNNING, which includes WORKER_UNBOUND.
So, I agree that with current code, WQ_CPU_INTENSIVE with WQ_UNBOUND
is same as WQ_UNBOUND alone.

However, I don't think it is necessary to make the changes. They don't really
improve readability of the code.

Thanks,
Song

^ permalink raw reply	[flat|nested] 6+ messages in thread
* Re: [PATCH] Remove WQ_CPU_INTENSIVE flag from unbound wq's
  2020-02-13 14:18 [PATCH] Remove WQ_CPU_INTENSIVE flag from unbound wq's Maksym Planeta
  2020-02-13 15:36 ` Mike Snitzer
@ 2020-02-14  1:39 ` Gao Xiang
  2020-02-14  3:47 ` Zhou Wang
  2 siblings, 0 replies; 6+ messages in thread
From: Gao Xiang @ 2020-02-14  1:39 UTC (permalink / raw)
  To: Maksym Planeta
  Cc: Herbert Xu, Mike Snitzer, dm-devel, linux-kernel, linux-raid,
	Song Liu, Zhou Wang, linux-crypto, Gao Xiang, linux-erofs,
	David S. Miller, Alasdair Kergon

On Thu, Feb 13, 2020 at 03:18:23PM +0100, Maksym Planeta wrote:
> The documentation [1] says that WQ_CPU_INTENSIVE is "meaningless" for
> unbound wq. I remove this flag from places where unbound queue is
> allocated. This is supposed to improve code readability.
> 
> 1. https://www.kernel.org/doc/html/latest/core-api/workqueue.html#flags
> 
> Signed-off-by: Maksym Planeta <mplaneta@os.inf.tu-dresden.de>
> ---
>  drivers/crypto/hisilicon/qm.c | 3 +--
>  drivers/md/dm-crypt.c         | 2 +-
>  drivers/md/dm-verity-target.c | 2 +-
>  drivers/md/raid5.c            | 2 +-
>  fs/erofs/zdata.c              | 2 +-

I'm okay for EROFS part,

Acked-by: Gao Xiang <gaoxiang25@huawei.com>

Thanks,
Gao Xiang

>  5 files changed, 5 insertions(+), 6 deletions(-)
> 
> diff --git a/drivers/crypto/hisilicon/qm.c b/drivers/crypto/hisilicon/qm.c
> index b57da5ef8b5b..4a39cb2c6a0b 100644
> --- a/drivers/crypto/hisilicon/qm.c
> +++ b/drivers/crypto/hisilicon/qm.c
> @@ -1148,8 +1148,7 @@ struct hisi_qp *hisi_qm_create_qp(struct hisi_qm *qm, u8 alg_type)
>  	qp->qp_id = qp_id;
>  	qp->alg_type = alg_type;
>  	INIT_WORK(&qp->work, qm_qp_work_func);
> -	qp->wq = alloc_workqueue("hisi_qm", WQ_UNBOUND | WQ_HIGHPRI |
> -				 WQ_CPU_INTENSIVE | WQ_MEM_RECLAIM, 0);
> +	qp->wq = alloc_workqueue("hisi_qm", WQ_UNBOUND | WQ_HIGHPRI | WQ_MEM_RECLAIM, 0);
>  	if (!qp->wq) {
>  		ret = -EFAULT;
>  		goto err_free_qp_mem;
> diff --git a/drivers/md/dm-crypt.c b/drivers/md/dm-crypt.c
> index c6a529873d0f..44d56325fa27 100644
> --- a/drivers/md/dm-crypt.c
> +++ b/drivers/md/dm-crypt.c
> @@ -3032,7 +3032,7 @@ static int crypt_ctr(struct dm_target *ti, unsigned int argc, char **argv)
>  						  1, devname);
>  	else
>  		cc->crypt_queue = alloc_workqueue("kcryptd/%s",
> -						  WQ_CPU_INTENSIVE | WQ_MEM_RECLAIM | WQ_UNBOUND,
> +						  WQ_MEM_RECLAIM | WQ_UNBOUND,
>  						  num_online_cpus(), devname);
>  	if (!cc->crypt_queue) {
>  		ti->error = "Couldn't create kcryptd queue";
> diff --git a/drivers/md/dm-verity-target.c b/drivers/md/dm-verity-target.c
> index 0d61e9c67986..20f92c7ea07e 100644
> --- a/drivers/md/dm-verity-target.c
> +++ b/drivers/md/dm-verity-target.c
> @@ -1190,7 +1190,7 @@ static int verity_ctr(struct dm_target *ti, unsigned argc, char **argv)
>  	}
>  
>  	/* WQ_UNBOUND greatly improves performance when running on ramdisk */
> -	v->verify_wq = alloc_workqueue("kverityd", WQ_CPU_INTENSIVE | WQ_MEM_RECLAIM | WQ_UNBOUND, num_online_cpus());
> +	v->verify_wq = alloc_workqueue("kverityd", WQ_MEM_RECLAIM | WQ_UNBOUND, num_online_cpus());
>  	if (!v->verify_wq) {
>  		ti->error = "Cannot allocate workqueue";
>  		r = -ENOMEM;
> diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c
> index ba00e9877f02..cd93a1731b82 100644
> --- a/drivers/md/raid5.c
> +++ b/drivers/md/raid5.c
> @@ -8481,7 +8481,7 @@ static int __init raid5_init(void)
>  	int ret;
>  
>  	raid5_wq = alloc_workqueue("raid5wq",
> -		WQ_UNBOUND|WQ_MEM_RECLAIM|WQ_CPU_INTENSIVE|WQ_SYSFS, 0);
> +		WQ_UNBOUND|WQ_MEM_RECLAIM|WQ_SYSFS, 0);
>  	if (!raid5_wq)
>  		return -ENOMEM;
>  
> diff --git a/fs/erofs/zdata.c b/fs/erofs/zdata.c
> index 80e47f07d946..b2a679f720e9 100644
> --- a/fs/erofs/zdata.c
> +++ b/fs/erofs/zdata.c
> @@ -43,7 +43,7 @@ void z_erofs_exit_zip_subsystem(void)
>  static inline int z_erofs_init_workqueue(void)
>  {
>  	const unsigned int onlinecpus = num_possible_cpus();
> -	const unsigned int flags = WQ_UNBOUND | WQ_HIGHPRI | WQ_CPU_INTENSIVE;
> +	const unsigned int flags = WQ_UNBOUND | WQ_HIGHPRI;
>  
>  	/*
>  	 * no need to spawn too many threads, limiting threads could minimum
> -- 
> 2.24.1
> 

^ permalink raw reply	[flat|nested] 6+ messages in thread
* Re: [PATCH] Remove WQ_CPU_INTENSIVE flag from unbound wq's
  2020-02-13 14:18 [PATCH] Remove WQ_CPU_INTENSIVE flag from unbound wq's Maksym Planeta
  2020-02-13 15:36 ` Mike Snitzer
  2020-02-14  1:39 ` [PATCH] " Gao Xiang
@ 2020-02-14  3:47 ` Zhou Wang
  2 siblings, 0 replies; 6+ messages in thread
From: Zhou Wang @ 2020-02-14  3:47 UTC (permalink / raw)
  To: Maksym Planeta, Herbert Xu, David S. Miller, Alasdair Kergon,
	Mike Snitzer, dm-devel, Song Liu, Gao Xiang, Chao Yu,
	linux-crypto, linux-kernel, linux-raid, linux-erofs

On 2020/2/13 22:18, Maksym Planeta wrote:
> The documentation [1] says that WQ_CPU_INTENSIVE is "meaningless" for
> unbound wq. I remove this flag from places where unbound queue is
> allocated. This is supposed to improve code readability.
> 
> 1. https://www.kernel.org/doc/html/latest/core-api/workqueue.html#flags
> 
> Signed-off-by: Maksym Planeta <mplaneta@os.inf.tu-dresden.de>
> ---
>  drivers/crypto/hisilicon/qm.c | 3 +--
>  drivers/md/dm-crypt.c         | 2 +-
>  drivers/md/dm-verity-target.c | 2 +-
>  drivers/md/raid5.c            | 2 +-
>  fs/erofs/zdata.c              | 2 +-
>  5 files changed, 5 insertions(+), 6 deletions(-)
> 
> diff --git a/drivers/crypto/hisilicon/qm.c b/drivers/crypto/hisilicon/qm.c
> index b57da5ef8b5b..4a39cb2c6a0b 100644
> --- a/drivers/crypto/hisilicon/qm.c
> +++ b/drivers/crypto/hisilicon/qm.c
> @@ -1148,8 +1148,7 @@ struct hisi_qp *hisi_qm_create_qp(struct hisi_qm *qm, u8 alg_type)
>  	qp->qp_id = qp_id;
>  	qp->alg_type = alg_type;
>  	INIT_WORK(&qp->work, qm_qp_work_func);
> -	qp->wq = alloc_workqueue("hisi_qm", WQ_UNBOUND | WQ_HIGHPRI |
> -				 WQ_CPU_INTENSIVE | WQ_MEM_RECLAIM, 0);
> +	qp->wq = alloc_workqueue("hisi_qm", WQ_UNBOUND | WQ_HIGHPRI | WQ_MEM_RECLAIM, 0);

I am OK with qm code.

Thanks!
Zhou

>  	if (!qp->wq) {
>  		ret = -EFAULT;
>  		goto err_free_qp_mem;
> diff --git a/drivers/md/dm-crypt.c b/drivers/md/dm-crypt.c
> index c6a529873d0f..44d56325fa27 100644
> --- a/drivers/md/dm-crypt.c
> +++ b/drivers/md/dm-crypt.c
> @@ -3032,7 +3032,7 @@ static int crypt_ctr(struct dm_target *ti, unsigned int argc, char **argv)
>  						  1, devname);
>  	else
>  		cc->crypt_queue = alloc_workqueue("kcryptd/%s",
> -						  WQ_CPU_INTENSIVE | WQ_MEM_RECLAIM | WQ_UNBOUND,
> +						  WQ_MEM_RECLAIM | WQ_UNBOUND,
>  						  num_online_cpus(), devname);
>  	if (!cc->crypt_queue) {
>  		ti->error = "Couldn't create kcryptd queue";
> diff --git a/drivers/md/dm-verity-target.c b/drivers/md/dm-verity-target.c
> index 0d61e9c67986..20f92c7ea07e 100644
> --- a/drivers/md/dm-verity-target.c
> +++ b/drivers/md/dm-verity-target.c
> @@ -1190,7 +1190,7 @@ static int verity_ctr(struct dm_target *ti, unsigned argc, char **argv)
>  	}
>  
>  	/* WQ_UNBOUND greatly improves performance when running on ramdisk */
> -	v->verify_wq = alloc_workqueue("kverityd", WQ_CPU_INTENSIVE | WQ_MEM_RECLAIM | WQ_UNBOUND, num_online_cpus());
> +	v->verify_wq = alloc_workqueue("kverityd", WQ_MEM_RECLAIM | WQ_UNBOUND, num_online_cpus());
>  	if (!v->verify_wq) {
>  		ti->error = "Cannot allocate workqueue";
>  		r = -ENOMEM;
> diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c
> index ba00e9877f02..cd93a1731b82 100644
> --- a/drivers/md/raid5.c
> +++ b/drivers/md/raid5.c
> @@ -8481,7 +8481,7 @@ static int __init raid5_init(void)
>  	int ret;
>  
>  	raid5_wq = alloc_workqueue("raid5wq",
> -		WQ_UNBOUND|WQ_MEM_RECLAIM|WQ_CPU_INTENSIVE|WQ_SYSFS, 0);
> +		WQ_UNBOUND|WQ_MEM_RECLAIM|WQ_SYSFS, 0);
>  	if (!raid5_wq)
>  		return -ENOMEM;
>  
> diff --git a/fs/erofs/zdata.c b/fs/erofs/zdata.c
> index 80e47f07d946..b2a679f720e9 100644
> --- a/fs/erofs/zdata.c
> +++ b/fs/erofs/zdata.c
> @@ -43,7 +43,7 @@ void z_erofs_exit_zip_subsystem(void)
>  static inline int z_erofs_init_workqueue(void)
>  {
>  	const unsigned int onlinecpus = num_possible_cpus();
> -	const unsigned int flags = WQ_UNBOUND | WQ_HIGHPRI | WQ_CPU_INTENSIVE;
> +	const unsigned int flags = WQ_UNBOUND | WQ_HIGHPRI;
>  
>  	/*
>  	 * no need to spawn too many threads, limiting threads could minimum
> 


^ permalink raw reply	[flat|nested] 6+ messages in thread
end of thread, other threads:[~2020-02-14  4:04 UTC | newest]

Thread overview: 6+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2020-02-13 14:18 [PATCH] Remove WQ_CPU_INTENSIVE flag from unbound wq's Maksym Planeta
2020-02-13 15:36 ` Mike Snitzer
2020-02-13 16:19   ` Maksym Planeta
2020-02-13 23:42     ` Song Liu
2020-02-14  1:39 ` [PATCH] " Gao Xiang
2020-02-14  3:47 ` Zhou Wang

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).