| From: D. Hugh Redelmeier <[EMAIL PROTECTED]>
| Date: Mon, 28 Mar 2005 20:26:58 -0500 (EST)

| Summary: "modprobe ivtv" produces scarry messages and did not work.
See <http://sourceforge.net/mailarchive/message.php?msg_id=11308676>

I applied current Fedora Core 3 updates, including a kernel update,
rebuilt ivtv-0.3.2p, installed it, and modprobed it.

This time it worked.  In particular, the following error did not show
up in dmesg output:
| Mar 28 18:05:36 redclaw kernel: ivtv: Allocate DMA stream 1 
| Mar 28 18:05:36 redclaw kernel: modprobe: page allocation failure. order:5, 
mode:0xd0

I suspect that the key difference is that this time I did the modprobe
in a freshly booted system and that the previous time the system had
been running for a while.  Running for a while could have caused
physical memory had become fragmented.

Here are subequent allocation messages from the modprobe that worked:
    ivtv: Registered v4l2 device, streamtype 2 minor 224
    ivtv: Create stream 2 using 40 52224 byte buffers  0 kbytes total
    ivtv: Allocate stream 2 using 40 52224 byte buffers  2097152 kbytes total
    ivtv: Registered v4l2 device, streamtype 3 minor 24
    ivtv: Create DMA stream 3 using 455 4608 byte buffers  0 kbytes total
    ivtv: Allocate DMA stream 3 using 455 4608 byte buffers  2097152 kbytes 
total
    ivtv: Registered v4l2 device, streamtype 4 minor 64
    ivtv: Create stream 4 
    ivtv: Allocate stream 4 

Here's what leaves me unsettled.  The failure was apparently for
allocating the scatter-gather array for stream 1.  A scatter-gather
array should be small: about 12 bytes per page in the stream.  Yet
the failed allocation was of order 5 -- 32 pages.  That isn't a lot,
so it shouldn't be hard to find that much contiguous space.

On the other hand, that suggests that a *lot* of pages are going to be
allocated to the stream:
        32*4096 / 12 = 10922
40 megabytes of RAM.  Something seems wrong.

Finally, I would point out that the driver did not fail gracefully.
If allocation fails, the driver should shut down cleanly, but it did
not.

So there are a bunch of dangling ends with respect to my original
report.

BTW, I got a private message from someone who had a similar problem.
I asked them to post to the list, but I don't think that they did.
That was the only response I've gotten.


-------------------------------------------------------
This SF.Net email is sponsored by: New Crystal Reports XI.
Version 11 adds new functionality designed to reduce time involved in
creating, integrating, and deploying reporting solutions. Free runtime info,
new features, or free trial, at: http://www.businessobjects.com/devxi/728
_______________________________________________
ivtv-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/ivtv-devel

Reply via email to