s390/ap_bus: Convert timers to use timer_setup()

PATCH In review Kees Cook 2 participants 8 messages Oct 25, 2017 – Nov 16, 2017

Subsystems (2)

s390 s390 zcrypt and pkey driver and AP bus

lore

Share & tools

lore

Kees Cook

author

2017-10-25 10:27 UTC #0 lore raw

PATCH 1/1 s390/ap_bus: Convert timers to use timer_setup()

In preparation for unconditionally passing the struct timer_list pointer to all timer callbacks, switch to using the new timer_setup() and from_timer() to pass the timer pointer explicitly.

Cc Harald Freudenberger <[email protected]>

Cc Martin Schwidefsky <[email protected]>

Cc Heiko Carstens <[email protected]>

Cc [email protected]

Signed-off-by Kees Cook <[email protected]>

---
 drivers/s390/crypto/ap_bus.c   | 10 +++++-----
 drivers/s390/crypto/ap_bus.h   |  2 +-
 drivers/s390/crypto/ap_queue.c |  2 +-
 3 files changed, 7 insertions(+), 7 deletions(-)

drivers/s390/crypto/ap_bus.c +5 -5

index 5f0be2040272..7bd18624e96e 100644
@@ -362,13 +362,13 @@ void ap_wait(enum ap_wait wait)
 
 /**
  * ap_request_timeout(): Handling of request timeouts
- * @data: Holds the AP device.
+ * @t: timer making this callback
  *
  * Handles request timeouts.
  */
-void ap_request_timeout(unsigned long data)
+void ap_request_timeout(struct timer_list *t)
 {
-	struct ap_queue *aq = (struct ap_queue *) data;
+	struct ap_queue *aq = from_timer(aq, t, timeout);
 
 	if (ap_suspend_flag)
 		return;

@@ -1147,7 +1147,7 @@ static void ap_scan_bus(struct work_struct *unused)
 	mod_timer(&ap_config_timer, jiffies + ap_config_time * HZ);
 }
 
-static void ap_config_timeout(unsigned long ptr)
+static void ap_config_timeout(struct timer_list *unused)
 {
 	if (ap_suspend_flag)
 		return;

@@ -1250,7 +1250,7 @@ int __init ap_module_init(void)
 		goto out_bus;
 
 	/* Setup the AP bus rescan timer. */
-	setup_timer(&ap_config_timer, ap_config_timeout, 0);
+	timer_setup(&ap_config_timer, ap_config_timeout, 0);
 
 	/*
 	 * Setup the high resultion poll timer.

drivers/s390/crypto/ap_bus.h +1 -1

index 754cf2223cfb..e7077820ae46 100644
@@ -241,7 +241,7 @@ void ap_flush_queue(struct ap_queue *aq);
 
 void *ap_airq_ptr(void);
 void ap_wait(enum ap_wait wait);
-void ap_request_timeout(unsigned long data);
+void ap_request_timeout(struct timer_list *t);
 void ap_bus_force_rescan(void);
 
 void ap_queue_init_reply(struct ap_queue *aq, struct ap_message *ap_msg);

drivers/s390/crypto/ap_queue.c +1 -1

index 56b96edffd5b..f4782113e4a3 100644
@@ -635,7 +635,7 @@ struct ap_queue *ap_queue_create(ap_qid_t qid, int device_type)
 	spin_lock_init(&aq->lock);
 	INIT_LIST_HEAD(&aq->pendingq);
 	INIT_LIST_HEAD(&aq->requestq);
-	setup_timer(&aq->timeout, ap_request_timeout, (unsigned long) aq);
+	timer_setup(&aq->timeout, ap_request_timeout, 0);
 
 	return aq;
 }

Signature · 6 lines

2.7.4


-- 
Kees Cook
Pixel Security

Martin Schwidefsky

2017-10-26 06:38 UTC #1 lore raw

Quoted 9 lines · On Wed, 25 Oct 2017 03:27:37 -0700 Kees Cook <[email protected]> wrote:

In preparation for unconditionally passing the struct timer_list pointer to all timer callbacks, switch to using the new timer_setup() and from_timer() to pass the timer pointer explicitly.
Cc Harald Freudenberger <[email protected]>
Cc Martin Schwidefsky <[email protected]>
Cc Heiko Carstens <[email protected]>
Cc [email protected]
Signed-off-by Kees Cook <[email protected]>

Quoted 5 lines

---
 drivers/s390/crypto/ap_bus.c   | 10 +++++-----
 drivers/s390/crypto/ap_bus.h   |  2 +-
 drivers/s390/crypto/ap_queue.c |  2 +-
 3 files changed, 7 insertions(+), 7 deletions(-)

Parked for the second part of the s390 updates for the next merge window. Thanks.

Signature · 4 lines

blue skies,
   Martin.

"Reality continues to ruin my life." - Calvin.

Kees Cook

author

2017-11-02 23:36 UTC #2 lore raw

Quoted 18 lines · Martin Schwidefsky

Quoted 9 lines · On Wed, 25 Oct 2017 03:27:37 -0700 Kees Cook <[email protected]> wrote:
In preparation for unconditionally passing the struct timer_list pointer to all timer callbacks, switch to using the new timer_setup() and from_timer() to pass the timer pointer explicitly.
Cc Harald Freudenberger <[email protected]>
Cc Martin Schwidefsky <[email protected]>
Cc Heiko Carstens <[email protected]>
Cc [email protected]
Signed-off-by Kees Cook <[email protected]>
Quoted 5 lines
---
 drivers/s390/crypto/ap_bus.c   | 10 +++++-----
 drivers/s390/crypto/ap_bus.h   |  2 +-
 drivers/s390/crypto/ap_queue.c |  2 +-
 3 files changed, 7 insertions(+), 7 deletions(-)

Quoted 2 lines

Parked for the second part of the s390 updates for the next merge window. Thanks.

Thanks for getting these staged. I just wanted to check with you, since I don't see these in -next anywhere yet:

s390: qdio: Convert timers to use timer_setup()
s390/sclp: Convert timers to use timer_setup()
s390/cio: Convert timers to use timer_setup()
s390/scsi: Convert timers to use timer_setup()
s390/ap_bus: Convert timers to use timer_setup()

Are all of these expected to land for -rc1? (Would it help to carry them in the timer tree?) I've got tree-wide changes ready to go once all these conversions have landed.

Thanks!

-Kees

Signature · 2 lines

Kees Cook
Pixel Security

Martin Schwidefsky

2017-11-03 06:29 UTC #3 lore raw

Quoted 32 lines · On Thu, 2 Nov 2017 16:36:53 -0700 Kees Cook <[email protected]> wrote:

Quoted 18 lines · Martin Schwidefsky
Quoted 9 lines · On Wed, 25 Oct 2017 03:27:37 -0700 Kees Cook <[email protected]> wrote:
In preparation for unconditionally passing the struct timer_list pointer to all timer callbacks, switch to using the new timer_setup() and from_timer() to pass the timer pointer explicitly.
Cc Harald Freudenberger <[email protected]>
Cc Martin Schwidefsky <[email protected]>
Cc Heiko Carstens <[email protected]>
Cc [email protected]
Signed-off-by Kees Cook <[email protected]>
Quoted 5 lines
---
 drivers/s390/crypto/ap_bus.c   | 10 +++++-----
 drivers/s390/crypto/ap_bus.h   |  2 +-
 drivers/s390/crypto/ap_queue.c |  2 +-
 3 files changed, 7 insertions(+), 7 deletions(-)  
Quoted 2 lines
Parked for the second part of the s390 updates for the next merge window. Thanks.
Thanks for getting these staged. I just wanted to check with you, since I don't see these in -next anywhere yet:
s390: qdio: Convert timers to use timer_setup()
s390/sclp: Convert timers to use timer_setup()
s390/cio: Convert timers to use timer_setup()
s390/scsi: Convert timers to use timer_setup()
s390/ap_bus: Convert timers to use timer_setup()

Quoted 3 lines

Are all of these expected to land for -rc1? (Would it help to carry them in the timer tree?) I've got tree-wide changes ready to go once all these conversions have landed.

These patches are parked on a private branches. The features branch on s390/linux is the one that is used for -next AND for the upstream merge. And since I do not want to rebase that branch the timer_setup patches are not in -next as well.

Four of you patches are safe with me, the s390/scsi patch is in Steffens patch queue. All good I would say.

Signature · 4 lines

blue skies,
   Martin.

"Reality continues to ruin my life." - Calvin.

Kees Cook

author

2017-11-03 20:36 UTC #4 lore raw

Quoted 39 lines · Martin Schwidefsky

Quoted 32 lines · On Thu, 2 Nov 2017 16:36:53 -0700 Kees Cook <[email protected]> wrote:
Quoted 18 lines · Martin Schwidefsky
Quoted 9 lines · On Wed, 25 Oct 2017 03:27:37 -0700 Kees Cook <[email protected]> wrote:
In preparation for unconditionally passing the struct timer_list pointer to all timer callbacks, switch to using the new timer_setup() and from_timer() to pass the timer pointer explicitly.
Cc Harald Freudenberger <[email protected]>
Cc Martin Schwidefsky <[email protected]>
Cc Heiko Carstens <[email protected]>
Cc [email protected]
Signed-off-by Kees Cook <[email protected]>
Quoted 5 lines
---
 drivers/s390/crypto/ap_bus.c   | 10 +++++-----
 drivers/s390/crypto/ap_bus.h   |  2 +-
 drivers/s390/crypto/ap_queue.c |  2 +-
 3 files changed, 7 insertions(+), 7 deletions(-)
Quoted 2 lines
Parked for the second part of the s390 updates for the next merge window. Thanks.
Thanks for getting these staged. I just wanted to check with you, since I don't see these in -next anywhere yet:
s390: qdio: Convert timers to use timer_setup()
s390/sclp: Convert timers to use timer_setup()
s390/cio: Convert timers to use timer_setup()
s390/scsi: Convert timers to use timer_setup()
s390/ap_bus: Convert timers to use timer_setup()
Quoted 3 lines
Are all of these expected to land for -rc1? (Would it help to carry them in the timer tree?) I've got tree-wide changes ready to go once all these conversions have landed.

Quoted 7 lines

These patches are parked on a private branches. The features branch on s390/linux is the one that is used for -next AND for the upstream merge. And since I do not want to rebase that branch the timer_setup patches are not in -next as well.
Four of you patches are safe with me, the s390/scsi patch is in Steffens patch queue. All good I would say.

Great, thanks very much for double-check! :)

-Kees

Signature · 2 lines

Kees Cook
Pixel Security

Kees Cook

author

2017-11-16 07:58 UTC #5 lore raw

In reply to Martin Schwidefsky

Quoted 39 lines · Martin Schwidefsky

Quoted 32 lines · On Thu, 2 Nov 2017 16:36:53 -0700 Kees Cook <[email protected]> wrote:
Quoted 18 lines · Martin Schwidefsky
Quoted 9 lines · On Wed, 25 Oct 2017 03:27:37 -0700 Kees Cook <[email protected]> wrote:
In preparation for unconditionally passing the struct timer_list pointer to all timer callbacks, switch to using the new timer_setup() and from_timer() to pass the timer pointer explicitly.
Cc Harald Freudenberger <[email protected]>
Cc Martin Schwidefsky <[email protected]>
Cc Heiko Carstens <[email protected]>
Cc [email protected]
Signed-off-by Kees Cook <[email protected]>
Quoted 5 lines
---
 drivers/s390/crypto/ap_bus.c   | 10 +++++-----
 drivers/s390/crypto/ap_bus.h   |  2 +-
 drivers/s390/crypto/ap_queue.c |  2 +-
 3 files changed, 7 insertions(+), 7 deletions(-)
Quoted 2 lines
Parked for the second part of the s390 updates for the next merge window. Thanks.
Thanks for getting these staged. I just wanted to check with you, since I don't see these in -next anywhere yet:
s390: qdio: Convert timers to use timer_setup()
s390/sclp: Convert timers to use timer_setup()
s390/cio: Convert timers to use timer_setup()
s390/scsi: Convert timers to use timer_setup()
s390/ap_bus: Convert timers to use timer_setup()
Quoted 3 lines
Are all of these expected to land for -rc1? (Would it help to carry them in the timer tree?) I've got tree-wide changes ready to go once all these conversions have landed.

Quoted 7 lines

These patches are parked on a private branches. The features branch on s390/linux is the one that is used for -next AND for the upstream merge. And since I do not want to rebase that branch the timer_setup patches are not in -next as well.
Four of you patches are safe with me, the s390/scsi patch is in Steffens patch queue. All good I would say.

Hi,

I saw s390 get merged for -rc1, but none of these timer_setup() conversions were included. Given the possible short merge window, what's the state of these? I can't do the bulk API changes at the end of -rc1 without the s390 timer patches. Anything I can help with?

Thanks!

-Kees

Signature · 2 lines

Kees Cook
Pixel Security

Martin Schwidefsky

2017-11-16 08:03 UTC #6 lore raw

Quoted 49 lines · On Wed, 15 Nov 2017 23:58:44 -0800 Kees Cook <[email protected]> wrote:

Quoted 39 lines · Martin Schwidefsky
Quoted 32 lines · On Thu, 2 Nov 2017 16:36:53 -0700 Kees Cook <[email protected]> wrote:
Quoted 18 lines · Martin Schwidefsky
Quoted 9 lines · On Wed, 25 Oct 2017 03:27:37 -0700 Kees Cook <[email protected]> wrote:
In preparation for unconditionally passing the struct timer_list pointer to all timer callbacks, switch to using the new timer_setup() and from_timer() to pass the timer pointer explicitly.
Cc Harald Freudenberger <[email protected]>
Cc Martin Schwidefsky <[email protected]>
Cc Heiko Carstens <[email protected]>
Cc [email protected]
Signed-off-by Kees Cook <[email protected]>
Quoted 5 lines
---
 drivers/s390/crypto/ap_bus.c   | 10 +++++-----
 drivers/s390/crypto/ap_bus.h   |  2 +-
 drivers/s390/crypto/ap_queue.c |  2 +-
 3 files changed, 7 insertions(+), 7 deletions(-)  
Quoted 2 lines
Parked for the second part of the s390 updates for the next merge window. Thanks.
Thanks for getting these staged. I just wanted to check with you, since I don't see these in -next anywhere yet:
s390: qdio: Convert timers to use timer_setup()
s390/sclp: Convert timers to use timer_setup()
s390/cio: Convert timers to use timer_setup()
s390/scsi: Convert timers to use timer_setup()
s390/ap_bus: Convert timers to use timer_setup()
Quoted 3 lines
Are all of these expected to land for -rc1? (Would it help to carry them in the timer tree?) I've got tree-wide changes ready to go once all these conversions have landed.
Quoted 7 lines
These patches are parked on a private branches. The features branch on s390/linux is the one that is used for -next AND for the upstream merge. And since I do not want to rebase that branch the timer_setup patches are not in -next as well.
Four of you patches are safe with me, the s390/scsi patch is in Steffens patch queue. All good I would say.

Quoted 6 lines

Hi,
I saw s390 get merged for -rc1, but none of these timer_setup() conversions were included. Given the possible short merge window, what's the state of these? I can't do the bulk API changes at the end of -rc1 without the s390 timer patches. Anything I can help with?

The remaining timer_setup changes are on the features branch of s390/linux, I plan to send a please-pull request be the end of the week.

Signature · 4 lines

blue skies,
   Martin.

"Reality continues to ruin my life." - Calvin.

Kees Cook

author

2017-11-16 08:03 UTC #7 lore raw

Quoted 54 lines · Martin Schwidefsky

Quoted 51 lines · On Wed, 15 Nov 2017 23:58:44 -0800 Kees Cook <[email protected]> wrote:
Quoted 39 lines · Martin Schwidefsky
Quoted 32 lines · On Thu, 2 Nov 2017 16:36:53 -0700 Kees Cook <[email protected]> wrote:
Quoted 18 lines · Martin Schwidefsky
Quoted 9 lines · On Wed, 25 Oct 2017 03:27:37 -0700 Kees Cook <[email protected]> wrote:
In preparation for unconditionally passing the struct timer_list pointer to all timer callbacks, switch to using the new timer_setup() and from_timer() to pass the timer pointer explicitly.
Cc Harald Freudenberger <[email protected]>
Cc Martin Schwidefsky <[email protected]>
Cc Heiko Carstens <[email protected]>
Cc [email protected]
Signed-off-by Kees Cook <[email protected]>
Quoted 5 lines
---
 drivers/s390/crypto/ap_bus.c   | 10 +++++-----
 drivers/s390/crypto/ap_bus.h   |  2 +-
 drivers/s390/crypto/ap_queue.c |  2 +-
 3 files changed, 7 insertions(+), 7 deletions(-)
Quoted 2 lines
Parked for the second part of the s390 updates for the next merge window. Thanks.
Thanks for getting these staged. I just wanted to check with you, since I don't see these in -next anywhere yet:
s390: qdio: Convert timers to use timer_setup()
s390/sclp: Convert timers to use timer_setup()
s390/cio: Convert timers to use timer_setup()
s390/scsi: Convert timers to use timer_setup()
s390/ap_bus: Convert timers to use timer_setup()
Quoted 3 lines
Are all of these expected to land for -rc1? (Would it help to carry them in the timer tree?) I've got tree-wide changes ready to go once all these conversions have landed.
Quoted 7 lines
These patches are parked on a private branches. The features branch on s390/linux is the one that is used for -next AND for the upstream merge. And since I do not want to rebase that branch the timer_setup patches are not in -next as well.
Four of you patches are safe with me, the s390/scsi patch is in Steffens patch queue. All good I would say.
Hi,

Quoted 7 lines

Quoted 4 lines
I saw s390 get merged for -rc1, but none of these timer_setup() conversions were included. Given the possible short merge window, what's the state of these? I can't do the bulk API changes at the end of -rc1 without the s390 timer patches. Anything I can help with?
The remaining timer_setup changes are on the features branch of s390/linux, I plan to send a please-pull request be the end of the week.

Okay, thanks! Sorry for all my nail-biting. :)

-Kees

Signature · 2 lines

Kees Cook
Pixel Security