Patchwork isdn: Return -EINTR in gigaset_start() if locking attempts fails.

login
register
mail settings
Submitter santosh nayak
Date March 16, 2012, 1:10 p.m.
Message ID <1331903413-11426-1-git-send-email-santoshprasadnayak@gmail.com>
Download mbox | patch
Permalink /patch/147202/
State Rejected
Delegated to: David Miller
Headers show

Comments

santosh nayak - March 16, 2012, 1:10 p.m.
From: Santosh Nayak <santoshprasadnayak@gmail.com>

If locking attempt was interrupted by a signal then we should
return -EINTR so that caller can take appropriate action.

We have 3 callers: gigaset_probe(), gigaset_tty_open() and
gigaset_probe(). Each caller tries to free allocated memory
if lock fails. This is possible if we returns -EINTR.

Signed-off-by: Santosh Nayak <santoshprasadnayak@gmail.com>
---
 drivers/isdn/gigaset/common.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)
David Miller - March 17, 2012, 6:18 a.m.
From: santosh nayak <santoshprasadnayak@gmail.com>
Date: Fri, 16 Mar 2012 18:40:13 +0530

> We have 3 callers: gigaset_probe(), gigaset_tty_open() and
> gigaset_probe(). Each caller tries to free allocated memory
> if lock fails. This is possible if we returns -EINTR.

Look again at the callers.

They interpret "0" as an error, so your patch would break the driver.
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Tilman Schmidt - March 17, 2012, 10:31 a.m.
Am 16.03.2012 14:10, schrieb santosh nayak:
> From: Santosh Nayak <santoshprasadnayak@gmail.com>
> 
> If locking attempt was interrupted by a signal then we should
> return -EINTR so that caller can take appropriate action.

NACK.

The return value of gigaset_start(), as documented in its
header comment, is:
 *	1 - success, 0 - error
Its callers rely on this. If you want to change it, you must
also change the documentation and the callers accordingly.
In its current form the patch would just break the driver.

Thanks,
Tilman
santosh nayak - March 17, 2012, 3:56 p.m.
Yes. You are right.

Caller is interpreting 0 in opposite way of normal sequence.
Thats why I misunderstood it.

In general, 0 means success and on error we return -ve.
Here its opposite.




regards
Santosh



On Sat, Mar 17, 2012 at 11:48 AM, David Miller <davem@davemloft.net> wrote:
> From: santosh nayak <santoshprasadnayak@gmail.com>
> Date: Fri, 16 Mar 2012 18:40:13 +0530
>
>> We have 3 callers: gigaset_probe(), gigaset_tty_open() and
>> gigaset_probe(). Each caller tries to free allocated memory
>> if lock fails. This is possible if we returns -EINTR.
>
> Look again at the callers.
>
> They interpret "0" as an error, so your patch would break the driver.
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
David Miller - March 17, 2012, 8:58 p.m.
From: santosh prasad nayak <santoshprasadnayak@gmail.com>
Date: Sat, 17 Mar 2012 21:26:14 +0530

> Caller is interpreting 0 in opposite way of normal sequence.
> Thats why I misunderstood it.

The simple fact is that you didn't even look at the code at the call
sites when you wrote this patch, and that's the first thing anyone is
going to do when reviewing it.

Therefore your laziness results in more useless work for other people.
I just wanted to point out how selfish and anti-social this kind of
behavior is.
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Julia Lawall - March 17, 2012, 9:28 p.m.
On Sat, 17 Mar 2012, David Miller wrote:

> From: santosh prasad nayak <santoshprasadnayak@gmail.com>
> Date: Sat, 17 Mar 2012 21:26:14 +0530
>
>> Caller is interpreting 0 in opposite way of normal sequence.
>> Thats why I misunderstood it.
>
> The simple fact is that you didn't even look at the code at the call
> sites when you wrote this patch, and that's the first thing anyone is
> going to do when reviewing it.
>
> Therefore your laziness results in more useless work for other people.
> I just wanted to point out how selfish and anti-social this kind of
> behavior is.

Not to pour too much salt on a wound, but just 5 lines above the patch 
site is the comment:

  * Return value:
  *      1 - success, 0 - error

And the error label also returns 0.  So there is a lot of local 
information that one can use to see what protocol the function follows.

That said, it's a bit too bad that two protocols have to coexist.

julia
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Patch

diff --git a/drivers/isdn/gigaset/common.c b/drivers/isdn/gigaset/common.c
index 7679270..2d10f3a 100644
--- a/drivers/isdn/gigaset/common.c
+++ b/drivers/isdn/gigaset/common.c
@@ -903,7 +903,7 @@  int gigaset_start(struct cardstate *cs)
 	unsigned long flags;
 
 	if (mutex_lock_interruptible(&cs->mutex))
-		return 0;
+		return -EINTR;
 
 	spin_lock_irqsave(&cs->lock, flags);
 	cs->connected = 1;