RE: Problems with Crystal cs4281 driver in 2.4.8

Woller, Thomas (
Thu, 13 Sep 2001 15:31:16 -0500

Can you send me the source that you have with separate email
(tar/bzip2)? I am wondering if the source base is old. Send to I'll take a look at it with what I
have as the latest.
I don't know of any such problem on the x86 or the IA64
Now, if you are trying use the MMAP interface with capture (ADC)
then I know that there will be some problems. non-mmap capture
should work fine. Also, if s->dma_adc.count is not decrementing
then interrupts have ceased or there is a bug in the driver,
maybe a race condition in the driver that needs to be fixed.

-----Original Message-----
From: Bao C. Ha []
Sent: Thursday, September 13, 2001 2:29 PM
Subject: Problems with Crystal cs4281 driver in 2.4.8

We are running kernel 2.4.8 on the sh4 platform.
We are encoutering a problem with the cs4281
driver as described in the following:

The device file is opened normally. Everything
is working fine. Then, when it is closed, the
application hangs. A ctrl-C is required to get
out of it. Upon further examining, we find that
the close() call is not returning. It is stuck
at the drain_adc().

In the cs4281_release(),
if (file->f_mode & FMODE_READ) {
------> drain_adc(s, file->f_flags & O_NONBLOCK);

And in the drain_adc(),
for (;;) {
spin_lock_irqsave(&s->lock, flags);
-----> count = s->dma_adc.count;
printk(KERN_INFO "cs4281: drain_adc()
spin_unlock_irqrestore(&s->lock, flags);

It seems that since s->dma_adc.count=8192, we
are going into an infinite loop there, in

The work-around that we are currently looking
at is setting the flag to O_NONBLOCKING just
prior to closing the device to take advantages
of this.
if (nonblock) {
current->state = TASK_RUNNING;
return -EBUSY;

Unfortunately, it causes problems later
since it is in non-blocking mode. Do a
fcntl(fd, F_SETFL, 0);
did not reset to the normal/blocking mode.

What would be the proper way to handle this
situation? Since I am not familiar with the
driver, I don't know if removing the
drain_adc() from cs4281_release() would cause
problems later or not. I am also not sure
this behavior is specific to the sh4-linux
port or not.

Appreciate any suggestions/comments.

To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to
More majordomo info at
Please read the FAQ at