On 04/02/2007 02:02 AM, Rene Herman wrote:
On 04/01/2007 12:06 PM, Pekka Enberg wrote:
Looks like mcdx_xfer is sleeping while holding q->queue_lock. The
attached (untested) patch should fix it.
This (including your followup) does indeed avoid the traces in the
kernel log, but unfortunately, the driver seems to need a bit more.
This may be expected, I'm not sure:
[EMAIL PROTECTED]:~# dd if=/dev/mcdx0 of=/dev/null bs=2048
0+0 records in
0+0 records out
0 bytes (0 B) copied, 0.000221955 seconds, 0.0 kB/s
[EMAIL PROTECTED]:~#
This I know isn't:
[EMAIL PROTECTED]:~# readcd dev=/dev/mcdx0 f=/dev/null
Segmentation fault
[EMAIL PROTECTED]:~#
(leaves a "note: readcd[1174] exited with preempt_count 1" in the log)
and after a "mount -t iso9660 /dev/mcdx0 /mnt/cdrom", a:
[EMAIL PROTECTED]:~# tar cv /mnt/cdrom >/dev/null
has upto now done all of:
1) segfault
2) make the kernel oops
3) reset the machine
This thing is just so badly broken... ;-(
I've attached the current patches from Jens and yourself (against
2.6.20.4) as a double check, but the driver's still totally unuseable
even with them...
No, I didn't.
I own two other legacy drives; a Panasonic CR562 (sbpcd.c) and a Sony
CDU33A (cdu31a.c) which together with this Mitsumi LU005S (mcdx.c) make
up also all but one of the controllers I have; a few standalone, but
most on old ISA soundcards. When I last tested, cdu31a somewhat worked
and sbpcd didn't. Had high hopes for mcdx.c upon seeing it compile
without warnings, but alas.
Last time Al Viro suggested ripping them out, I slightly objected:
http://lkml.org/lkml/2004/5/2/123
but well, although I like playing with this stuff, I still don't know
the first thing about the block layer and given the shape these things
are in...
Many thanks for looking though!
Rene.
--- drivers/cdrom/mcdx.c.orig 2007-04-02 00:25:09.000000000 +0200
+++ drivers/cdrom/mcdx.c 2007-04-02 00:37:53.000000000 +0200
@@ -577,11 +577,20 @@
if (!req)
return;
+ if (!blk_fs_request(req)) {
+ spin_lock_irq(q->queue_lock);
+ end_request(req, 0);
+ goto again;
+ }
+
+ spin_unlock_irq(q->queue_lock);
+
stuffp = req->rq_disk->private_data;
if (!stuffp->present) {
xwarn("do_request(): bad device: %s\n",req->rq_disk->disk_name);
xtrace(REQUEST, "end_request(0): bad device\n");
+ spin_lock_irq(q->queue_lock);
end_request(req, 0);
return;
}
@@ -589,6 +598,7 @@
if (stuffp->audio) {
xwarn("do_request() attempt to read from audio cd\n");
xtrace(REQUEST, "end_request(0): read from audio\n");
+ spin_lock_irq(q->queue_lock);
end_request(req, 0);
return;
}
@@ -596,9 +606,10 @@
xtrace(REQUEST, "do_request() (%lu + %lu)\n",
req->sector, req->nr_sectors);
- if (req->cmd != READ) {
+ if (rq_data_dir(req) != READ) {
xwarn("do_request(): non-read command to cd!!\n");
xtrace(REQUEST, "end_request(0): write\n");
+ spin_lock_irq(q->queue_lock);
end_request(req, 0);
return;
}
@@ -613,6 +624,7 @@
req->nr_sectors);
if (i == -1) {
+ spin_lock_irq(q->queue_lock);
end_request(req, 0);
goto again;
}
@@ -620,10 +632,12 @@
req->nr_sectors -= i;
req->buffer += (i * 512);
}
+ spin_lock_irq(q->queue_lock);
end_request(req, 1);
goto again;
xtrace(REQUEST, "end_request(1)\n");
+ spin_lock_irq(q->queue_lock);
end_request(req, 1);
}
