[PATCH] tty/n_gsm.c: fix false positive WARN_ON and do some codes improvement

[PATCH] tty/n_gsm.c: fix false positive WARN_ON and do some codes improvement

[xinhui]

From: xinhui <xinhui@linux.vnet.ibm.com>

If gsm driver fails to activate one mux, and this mux is not stored in
gsm_mux[], there would be a warning in gsm_cleanup_mux(). Actually this
is a legal case. So just do a simple check instead of WARN_ON.

There is one filed gsm->num to store its index of gsm_mux[]. So use
gsm->num to remove itself from gsm_mux[] instead of the for-loop
traverse.

Reported-by: Dmitry Vyukov <dvyukov@google.com>
Fixes: 5a64096700dc ("tty/n_gsm.c: fix a memory leak in gsmld_open")
Signed-off-by: Pan Xinhui <xinhui.pan@linux.vnet.ibm.com>
---
 drivers/tty/n_gsm.c |   12 +++++-------
 1 files changed, 5 insertions(+), 7 deletions(-)

diff --git a/drivers/tty/n_gsm.c b/drivers/tty/n_gsm.c
index c3fe026..56377e1 100644
--- a/drivers/tty/n_gsm.c
+++ b/drivers/tty/n_gsm.c
@@ -2037,15 +2037,13 @@ static void gsm_cleanup_mux(struct gsm_mux *gsm)
 
 	gsm->dead = 1;
 
+	/* open failed before registering => nothing to do*/
+	if (gsm_mux[gsm->num] != gsm)
+		return;
+
 	spin_lock(&gsm_mux_lock);
-	for (i = 0; i < MAX_MUX; i++) {
-		if (gsm_mux[i] == gsm) {
-			gsm_mux[i] = NULL;
-			break;
-		}
-	}
+	gsm_mux[gsm->num] = NULL;
 	spin_unlock(&gsm_mux_lock);
-	WARN_ON(i == MAX_MUX);
 
 	/* In theory disconnecting DLCI 0 is sufficient but for some
 	   modems this is apparently not the case. */
-- 
1.7.1

Re: [PATCH] tty/n_gsm.c: fix false positive WARN_ON and do some codes improvement

[xinhui]

Hi, Jiri
	I made this patch based on your idea. Any modification is welcome. :)

PS. "xinhui" or "xinhui.pan"@linux.vnet.ibm.com are same mail account, both of them are alias.

thanks
xinhui

On 2015/11/25 19:18, xinhui wrote:
> From: xinhui <xinhui@linux.vnet.ibm.com>
>
> If gsm driver fails to activate one mux, and this mux is not stored in
> gsm_mux[], there would be a warning in gsm_cleanup_mux(). Actually this
> is a legal case. So just do a simple check instead of WARN_ON.
>
> There is one filed gsm->num to store its index of gsm_mux[]. So use
> gsm->num to remove itself from gsm_mux[] instead of the for-loop
> traverse.
>
> Reported-by: Dmitry Vyukov <dvyukov@google.com>
> Fixes: 5a64096700dc ("tty/n_gsm.c: fix a memory leak in gsmld_open")
> Signed-off-by: Pan Xinhui <xinhui.pan@linux.vnet.ibm.com>
> ---
>   drivers/tty/n_gsm.c |   12 +++++-------
>   1 files changed, 5 insertions(+), 7 deletions(-)
>
> diff --git a/drivers/tty/n_gsm.c b/drivers/tty/n_gsm.c
> index c3fe026..56377e1 100644
> --- a/drivers/tty/n_gsm.c
> +++ b/drivers/tty/n_gsm.c
> @@ -2037,15 +2037,13 @@ static void gsm_cleanup_mux(struct gsm_mux *gsm)
>
>   	gsm->dead = 1;
>
> +	/* open failed before registering => nothing to do*/
> +	if (gsm_mux[gsm->num] != gsm)
> +		return;
> +
>   	spin_lock(&gsm_mux_lock);
> -	for (i = 0; i < MAX_MUX; i++) {
> -		if (gsm_mux[i] == gsm) {
> -			gsm_mux[i] = NULL;
> -			break;
> -		}
> -	}
> +	gsm_mux[gsm->num] = NULL;
>   	spin_unlock(&gsm_mux_lock);
> -	WARN_ON(i == MAX_MUX);
>
>   	/* In theory disconnecting DLCI 0 is sufficient but for some
>   	   modems this is apparently not the case. */
>

Re: [PATCH] tty/n_gsm.c: fix false positive WARN_ON and do some codes improvement

[Greg Kroah-Hartman]

On Wed, Nov 25, 2015 at 07:18:37PM +0800, xinhui wrote:
> From: xinhui <xinhui@linux.vnet.ibm.com>
> 
> If gsm driver fails to activate one mux, and this mux is not stored in
> gsm_mux[], there would be a warning in gsm_cleanup_mux(). Actually this
> is a legal case. So just do a simple check instead of WARN_ON.
> 
> There is one filed gsm->num to store its index of gsm_mux[]. So use
> gsm->num to remove itself from gsm_mux[] instead of the for-loop
> traverse.
> 
> Reported-by: Dmitry Vyukov <dvyukov@google.com>
> Fixes: 5a64096700dc ("tty/n_gsm.c: fix a memory leak in gsmld_open")
> Signed-off-by: Pan Xinhui <xinhui.pan@linux.vnet.ibm.com>

the signed-off-by name has to match your from: name :(

Re: [PATCH] tty/n_gsm.c: fix false positive WARN_ON and do some codes improvement

[Pan Xinhui]

Hi, Greg
	This patch is dropped, I have created another patch based on Jiri' patch.
On 2015/12/15 02:55, Greg Kroah-Hartman wrote:
> On Wed, Nov 25, 2015 at 07:18:37PM +0800, xinhui wrote:
>> From: xinhui <xinhui@linux.vnet.ibm.com>
>>
>> If gsm driver fails to activate one mux, and this mux is not stored in
>> gsm_mux[], there would be a warning in gsm_cleanup_mux(). Actually this
>> is a legal case. So just do a simple check instead of WARN_ON.
>>
>> There is one filed gsm->num to store its index of gsm_mux[]. So use
>> gsm->num to remove itself from gsm_mux[] instead of the for-loop
>> traverse.
>>
>> Reported-by: Dmitry Vyukov <dvyukov@google.com>
>> Fixes: 5a64096700dc ("tty/n_gsm.c: fix a memory leak in gsmld_open")
>> Signed-off-by: Pan Xinhui <xinhui.pan@linux.vnet.ibm.com>
>
> the signed-off-by name has to match your from: name :(
>