[patch 2/4] cdrom/aztcd: remove sleep_on() usage

[patch 2/4] cdrom/aztcd: remove sleep_on() usage

[domen]

[-- Attachment #1: sleep_on-drivers_cdrom_aztcd.patch --]
[-- Type: text/plain, Size: 1125 bytes --]

From: Nishanth Aravamudan <nacc@us.ibm.com>



Directly use wait-queues instead of the deprecated sleep_on().
This required adding a local waitqueue. Patch is compile-tested.

Signed-off-by: Nishanth Aravamudan <nacc@us.ibm.com>
Signed-off-by: Domen Puncer <domen@coderock.org>
---
 aztcd.c |    6 +++++-
 1 files changed, 5 insertions(+), 1 deletion(-)

Index: quilt/drivers/cdrom/aztcd.c
===================================================================
--- quilt.orig/drivers/cdrom/aztcd.c
+++ quilt/drivers/cdrom/aztcd.c
@@ -179,6 +179,7 @@
 #include <linux/ioport.h>
 #include <linux/string.h>
 #include <linux/major.h>
+#include <linux/wait.h>
 
 #include <linux/init.h>
 
@@ -429,9 +430,12 @@ static void dten_low(void)
 #define STEN_LOW_WAIT   statusAzt()
 static void statusAzt(void)
 {
+	DEFINE_WAIT(wait);
 	AztTimeout = AZT_STATUS_DELAY;
 	SET_TIMER(aztStatTimer, HZ / 100);
-	sleep_on(&azt_waitq);
+	prepare_to_wait(&azt_waitq, &wait, TASK_UNINTERRUPTIBLE);
+	schedule();
+	finish_wait(&azt_waitq, &wait);
 	if (AztTimeout <= 0)
 		printk("aztcd: Error Wait STEN_LOW_WAIT command:%x\n",
 		       aztCmd);

--

Re: [patch 2/4] cdrom/aztcd: remove sleep_on() usage

[Jesper Juhl]

On 6/20/05, domen@coderock.org <domen@coderock.org> wrote:
> From: Nishanth Aravamudan <nacc@us.ibm.com>
> 
> 
> 
> Directly use wait-queues instead of the deprecated sleep_on().
> This required adding a local waitqueue. Patch is compile-tested.
> 
> Signed-off-by: Nishanth Aravamudan <nacc@us.ibm.com>
> Signed-off-by: Domen Puncer <domen@coderock.org>
> ---
>  aztcd.c |    6 +++++-
>  1 files changed, 5 insertions(+), 1 deletion(-)
> 
> Index: quilt/drivers/cdrom/aztcd.c
> ===================================================================
> --- quilt.orig/drivers/cdrom/aztcd.c
> +++ quilt/drivers/cdrom/aztcd.c
> @@ -179,6 +179,7 @@
>  #include <linux/ioport.h>
>  #include <linux/string.h>
>  #include <linux/major.h>
> +#include <linux/wait.h>
> 
>  #include <linux/init.h>
> 
> @@ -429,9 +430,12 @@ static void dten_low(void)
>  #define STEN_LOW_WAIT   statusAzt()
>  static void statusAzt(void)
>  {
> +       DEFINE_WAIT(wait);
>         AztTimeout = AZT_STATUS_DELAY;
>         SET_TIMER(aztStatTimer, HZ / 100);
> -       sleep_on(&azt_waitq);
> +       prepare_to_wait(&azt_waitq, &wait, TASK_UNINTERRUPTIBLE);
> +       schedule();
> +       finish_wait(&azt_waitq, &wait);
>         if (AztTimeout <= 0)
>                 printk("aztcd: Error Wait STEN_LOW_WAIT command:%x\n",
>                        aztCmd);
> 

Hmm, now that noone's sleeping on azt_waitq the two
wake_up(&azt_waitq); calls in aztStatTimer() don't seem to make much
sense any more... Can they just go away or?  If they can go away then
axt_waitq itself would seem to be a goner as well...   It might just
be me missing something, but this patch looks incomplete and not
completely thought through to me.


-- 
Jesper Juhl <jesper.juhl@gmail.com>
Don't top-post  http://www.catb.org/~esr/jargon/html/T/top-post.html
Plain text mails only, please      http://www.expita.com/nomime.html

Re: [patch 2/4] cdrom/aztcd: remove sleep_on() usage

[Nishanth Aravamudan]

On 21.06.2005 [01:14:11 +0200], Jesper Juhl wrote:
> On 6/20/05, domen@coderock.org <domen@coderock.org> wrote:
> > From: Nishanth Aravamudan <nacc@us.ibm.com>
> > 
> > 
> > 
> > Directly use wait-queues instead of the deprecated sleep_on().
> > This required adding a local waitqueue. Patch is compile-tested.
> > 
> > Signed-off-by: Nishanth Aravamudan <nacc@us.ibm.com>
> > Signed-off-by: Domen Puncer <domen@coderock.org>
> > ---
> >  aztcd.c |    6 +++++-
> >  1 files changed, 5 insertions(+), 1 deletion(-)
> > 
> > Index: quilt/drivers/cdrom/aztcd.c
> > ===================================================================
> > --- quilt.orig/drivers/cdrom/aztcd.c
> > +++ quilt/drivers/cdrom/aztcd.c
> > @@ -179,6 +179,7 @@
> >  #include <linux/ioport.h>
> >  #include <linux/string.h>
> >  #include <linux/major.h>
> > +#include <linux/wait.h>
> > 
> >  #include <linux/init.h>
> > 
> > @@ -429,9 +430,12 @@ static void dten_low(void)
> >  #define STEN_LOW_WAIT   statusAzt()
> >  static void statusAzt(void)
> >  {
> > +       DEFINE_WAIT(wait);
> >         AztTimeout = AZT_STATUS_DELAY;
> >         SET_TIMER(aztStatTimer, HZ / 100);
> > -       sleep_on(&azt_waitq);
> > +       prepare_to_wait(&azt_waitq, &wait, TASK_UNINTERRUPTIBLE);
> > +       schedule();
> > +       finish_wait(&azt_waitq, &wait);
> >         if (AztTimeout <= 0)
> >                 printk("aztcd: Error Wait STEN_LOW_WAIT command:%x\n",
> >                        aztCmd);
> > 
> 
> Hmm, now that noone's sleeping on azt_waitq the two
> wake_up(&azt_waitq); calls in aztStatTimer() don't seem to make much
> sense any more... Can they just go away or?  If they can go away then
> axt_waitq itself would seem to be a goner as well...   It might just
> be me missing something, but this patch looks incomplete and not
> completely thought through to me.

Huh? prepare_to_wait() means the 'wait' here is sleeping on azt_waitq...

Thanks,
Nish

Re: [patch 2/4] cdrom/aztcd: remove sleep_on() usage

[Jesper Juhl]

On 6/21/05, Nishanth Aravamudan <nacc@us.ibm.com> wrote:
> On 21.06.2005 [01:14:11 +0200], Jesper Juhl wrote:
> > On 6/20/05, domen@coderock.org <domen@coderock.org> wrote:
> > > From: Nishanth Aravamudan <nacc@us.ibm.com>
> > >
> > >
> > >
> > > Directly use wait-queues instead of the deprecated sleep_on().
> > > This required adding a local waitqueue. Patch is compile-tested.
> > >
> > > Signed-off-by: Nishanth Aravamudan <nacc@us.ibm.com>
> > > Signed-off-by: Domen Puncer <domen@coderock.org>
> > > ---
> > >  aztcd.c |    6 +++++-
> > >  1 files changed, 5 insertions(+), 1 deletion(-)
> > >
> > > Index: quilt/drivers/cdrom/aztcd.c
> > > ===================================================================
> > > --- quilt.orig/drivers/cdrom/aztcd.c
> > > +++ quilt/drivers/cdrom/aztcd.c
> > > @@ -179,6 +179,7 @@
> > >  #include <linux/ioport.h>
> > >  #include <linux/string.h>
> > >  #include <linux/major.h>
> > > +#include <linux/wait.h>
> > >
> > >  #include <linux/init.h>
> > >
> > > @@ -429,9 +430,12 @@ static void dten_low(void)
> > >  #define STEN_LOW_WAIT   statusAzt()
> > >  static void statusAzt(void)
> > >  {
> > > +       DEFINE_WAIT(wait);
> > >         AztTimeout = AZT_STATUS_DELAY;
> > >         SET_TIMER(aztStatTimer, HZ / 100);
> > > -       sleep_on(&azt_waitq);
> > > +       prepare_to_wait(&azt_waitq, &wait, TASK_UNINTERRUPTIBLE);
> > > +       schedule();
> > > +       finish_wait(&azt_waitq, &wait);
> > >         if (AztTimeout <= 0)
> > >                 printk("aztcd: Error Wait STEN_LOW_WAIT command:%x\n",
> > >                        aztCmd);
> > >
> >
> > Hmm, now that noone's sleeping on azt_waitq the two
> > wake_up(&azt_waitq); calls in aztStatTimer() don't seem to make much
> > sense any more... Can they just go away or?  If they can go away then
> > axt_waitq itself would seem to be a goner as well...   It might just
> > be me missing something, but this patch looks incomplete and not
> > completely thought through to me.
> 
> Huh? prepare_to_wait() means the 'wait' here is sleeping on azt_waitq...
> 
Ahh damn, sorry. Somehow I misread the first argument of
prepare_to_wait() and completely missed the fact that it was sleeping
on azt_waitq. It must be past my bedtime...
My bad - it looks good.

-- 
Jesper Juhl <jesper.juhl@gmail.com>
Don't top-post  http://www.catb.org/~esr/jargon/html/T/top-post.html
Plain text mails only, please      http://www.expita.com/nomime.html

Re: [patch 2/4] cdrom/aztcd: remove sleep_on() usage

[Johannes Stezenbach]

domen@coderock.org wrote:
> From: Nishanth Aravamudan <nacc@us.ibm.com>
> 
> 
> 
> Directly use wait-queues instead of the deprecated sleep_on().
> This required adding a local waitqueue. Patch is compile-tested.

IMHO this patch doesn't improve anything, it just makes the
code harder to read. sleep_on() is deprecated because there's
an inherent race between sleep_on() and wake_up(). To address
this you need to add a condition variable (e.g. AztStenLow) and
convert the code to use wait_event().
The timer trick used here makes it pretty hard to hit the
race, but with a preemptible kernel one could be preempted
between the SET_TIMER() and the sleep_on().

Johannes

> Signed-off-by: Nishanth Aravamudan <nacc@us.ibm.com>
> Signed-off-by: Domen Puncer <domen@coderock.org>
> ---
>  aztcd.c |    6 +++++-
>  1 files changed, 5 insertions(+), 1 deletion(-)
> 
> Index: quilt/drivers/cdrom/aztcd.c
> ===================================================================
> --- quilt.orig/drivers/cdrom/aztcd.c
> +++ quilt/drivers/cdrom/aztcd.c
> @@ -179,6 +179,7 @@
>  #include <linux/ioport.h>
>  #include <linux/string.h>
>  #include <linux/major.h>
> +#include <linux/wait.h>
>  
>  #include <linux/init.h>
>  
> @@ -429,9 +430,12 @@ static void dten_low(void)
>  #define STEN_LOW_WAIT   statusAzt()
>  static void statusAzt(void)
>  {
> +	DEFINE_WAIT(wait);
>  	AztTimeout = AZT_STATUS_DELAY;
>  	SET_TIMER(aztStatTimer, HZ / 100);
> -	sleep_on(&azt_waitq);
> +	prepare_to_wait(&azt_waitq, &wait, TASK_UNINTERRUPTIBLE);
> +	schedule();
> +	finish_wait(&azt_waitq, &wait);
>  	if (AztTimeout <= 0)
>  		printk("aztcd: Error Wait STEN_LOW_WAIT command:%x\n",
>  		       aztCmd);
>