I tried out the &msg->queue, and it's running fine. It's better than my fix
because it's more clear than "mcfqspi->msgq.next".
After grabbing a msg from the msgq, the mcfqspi_work function calls
list_del_init on the mcfqspi->msgq which unintentionally deletes the rest
of the list before it can be processed. If qspi call was made using
spi_sync, this can result in a process hang.
Signed-off by Jate Sujjavanich<jsujjavan...@syntech-fuelmaster.com>
-----------
diff --git a/drivers/spi/coldfire_qspi.c b/drivers/spi/coldfire_qspi.c
index 59be3ef..0aa00af 100644
--- a/drivers/spi/coldfire_qspi.c
+++ b/drivers/spi/coldfire_qspi.c
@@ -316,7 +316,7 @@ static void mcfqspi_work(struct work_struct *work)
msg = container_of(mcfqspi->msgq.next, struct spi_message,
queue);
- list_del_init(&mcfqspi->msgq);
+ list_del_init(&msg->queue);
spin_unlock_irqrestore(&mcfqspi->lock, flags);
spi = msg->spi;
-----Original Message-----
From: Steven King [mailto:sfk...@fdwdc.com]
Sent: Thursday, September 23, 2010 11:43 AM
To: Jate Sujjavanich
Cc: uclinux-dev@uclinux.org; 'Greg Ungerer'
Subject: Re: [uClinux-dev] [PATCH] m68knommu: Coldfire QSPI platform support
On Wednesday 22 September 2010 09:03:24 Jate Sujjavanich wrote:
> Steven,
>
> I have multiple qspi devices on my system, and I was getting a hang on
> certain processes that were making spi_sync calls. There completion
> callbacks were never being called because the messages weren't being
> processed. See the patch.
>
> - Jate
>
> ---------
> After grabbing a msg from the msgq, the mcfqspi_work function calls
> list_del_init on the mcfqspi->msgq which unintentionally deletes the rest
> of the list before it can be processed. If qspi call was made using
> spi_sync, this can result in a process hang.
>
> Signed-off by Jate Sujjavanich<jsujjavan...@syntech-fuelmaster.com>
> ---------
> diff --git a/drivers/spi/coldfire_qspi.c b/drivers/spi/coldfire_qspi.c
> index 05d4cfa..23822bb 100644
> --- a/drivers/spi/coldfire_qspi.c
> +++ b/drivers/spi/coldfire_qspi.c
> @@ -317,7 +317,7 @@ static void mcfqspi_work(struct work_struct *work)
> msg = container_of(mcfqspi->msgq.next, struct spi_message,
> queue);
>
> - list_del_init(&mcfqspi->msgq);
> + list_del_init(mcfqspi->msgq.next);
> spin_unlock_irqrestore(&mcfqspi->lock, flags);
>
> spi = msg->spi;
Hi Jate!
Actually, After looking to see what the other work queue based spi drivers
were doing, I think that list_del_init in my original is total wrong! Can you
try
this patch?
diff --git a/drivers/spi/coldfire_qspi.c b/drivers/spi/coldfire_qspi.c
index 59be3ef..0aa00af 100644
--- a/drivers/spi/coldfire_qspi.c
+++ b/drivers/spi/coldfire_qspi.c
@@ -316,7 +316,7 @@ static void mcfqspi_work(struct work_struct *work)
msg = container_of(mcfqspi->msgq.next, struct spi_message,
queue);
- list_del_init(&mcfqspi->msgq);
+ list_del_init(&msg->queue);
spin_unlock_irqrestore(&mcfqspi->lock, flags);
spi = msg->spi;
_______________________________________________
uClinux-dev mailing list
uClinux-dev@uclinux.org
http://mailman.uclinux.org/mailman/listinfo/uclinux-dev
This message was resent by uclinux-dev@uclinux.org
To unsubscribe see:
http://mailman.uclinux.org/mailman/options/uclinux-dev
