Pieter Palmers wrote:
There is one remark though: Your driver oops at some times (null pointer dereference), ie when doing: (sam_ucode.c)
err = snd_sam9407_init_command(NULL, bytes, 7, NULL, 0, -1);
The problem is in the magic_cast:
When calling this with queue=NULL, the magic_cast causes an oops, which is pretty normal I guess. I wonder why it didn't in your tests.
/* transmit init command sequence to sam9407 */
int snd_sam9407_init_command(sam9407_io_queue_t * queue,
u8 *cmd_buffer, int cmd_bytes,
u8 *rcv_buffer, int rcv_bytes,
int ack)
{
sam9407_t *sam = snd_magic_cast(sam9407_t, queue->sam, return -EINVAL);
signed long end_time;
int err;
unsigned long flags;
if (queue) {
(...)
Yikes! This is an unfinished ""cleanup"" stuff. I was trying to get rid all the queue setup overhead when command actually didn't expect any response (when ACK = -1). These commands should be called with:
/* transmit init command sequence to sam9407 */
int snd_sam9407_init_command(sam9407_t *sam,
sam9407_io_queue_t * queue,
u8 *cmd_buffer, int cmd_bytes,
u8 *rcv_buffer, int rcv_bytes,
int ack)
(The same applies to snd_sam9407_command())
Following this, calls to these two commands should be fixed to use *sam variable. sam9407_t *sam should be removed from sam9407_io_queue_t. In snd_sam9407_queue_prepare() function, remove test for (ack >= 0) and allways use
queue->error = -EIO;
queue->ack = ack & 0xff;
set_bit(SAM_QUEUE_WAIT_ACK, &queue->status);
set_bit(SAM_QUEUE_BUSY, &queue->status);
codepath.
Commands without acknowledge should now be called with
snd_sam9407_[init_]command(sam, NULL, buffer, num_cmd, XXXX, X, X);
where X doesn't matter.
Uros.
-------------------------------------------------------
This SF.NET email is sponsored by:
SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See!
http://www.vasoftware.com
_______________________________________________
Alsa-devel mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/alsa-devel
