Red Hat Bugzilla – Bug 77632
(SOUND TRIDENT)ali_ac_97 complains "try clear busy flag!!" (accompanied by temporary skips of sound)
Last modified: 2007-04-18 12:48:21 EDT
From Bugzilla Helper:
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.0.1) Gecko/20020830
Description of problem:
The box is a hp omnibook xe4500 laptop (P4 1.6).
When running the stock (2.4.18-14) kernel the sound skips when playing from xmms
while the mozilla scrollbar is moved. It also happens when the terminal produces
the bell sound.
ali_ac97 logs the following in /var/log/messages:
Nov 10 21:02:19 bluefish kernel: ali_ac97_set :try clear busy flag!!
Nov 10 21:48:23 bluefish last message repeated 10 times
Nov 10 21:49:13 bluefish last message repeated 24 times
Version-Release number of selected component (if applicable):
Steps to Reproduce:
1. open xmms, play something
2. open mozilla, load a big html page
3. scroll the mozilla bar all the way down, then all the way up
Actual Results: Sound playing skips accompanied by ali_ac97 message in syslog.
Expected Results: Sound should play without skips.
Note: I cannot reproduce the problem with kernels:
2.4.18-3 (stock in 7.3)
2.4.19 (vanilla + acpi.sf.net patches)
2.5.40 (vanilla + acpi.sf.net patches)
I have same problem with notebook Acer Travelmate 521TE running kernel 2.4.18-18.8.0
Same here with a Toshiba 1805-S254, stock 2.4.18-17.8.0.
To be more specific, it not only scrolling. Dragging a window around real
quick does it too. I though this was due to my processor speed (PIII 1GHz)
(and using fb for X), but if you say other kernels don't do it, I'll try
I do get a screeching sound with "screen activity" even when playing wav files
from konqueror, but could not associate that with the "kernel: ali_ac97_set
:try clear busy flag!!", or "kernel: ali_ac97_read :try clear busy flag" or
the funny "kernel: r busy flag" on /var/log/messages. When the screeching
happens, looking at tail -f /var/log/messages do not show new messages popping
up. Try that.
I can play wav files with /usr/bin/play on a Konsole window and drag the
scrollbar like crazy on a Konqueror window with no screeching, clipping, nor
messages. It's also less (but it is) affected by dragging a window around
The messages actually seem to be logged when X is starting (I'm using KDE).
There's so many occurences that "dmesg" is rendered unuseable (flooded with
So there are maybe two problems; probably related to each other, but at least
in my case, not happening simultaneously:
1. Screeching/clipping when screen is quickly updated (fb related?)
2. "try clear busy flag" messages flooding
Strangely, among the hundreds of those busy flag messages, I found this one
(1) in dmesg too:
ac97_codec: AC97 codec, id: 0x0000:0x0000 (Unknown)
To all affected by this problem: you can use the alsa driver as a workaround.
You can get alsa rpms from freshrpms.net (alsa-kernel, alsa-lib, alsa-driver, etc).
Install the drivers, remove any module that is sound related (otherwise alsa
won't load the ali5451 driver because of conflict issues) and modprobe sound-card-0.
My /etc/modules.conf now lists:
#alias sound-slot-0 trident
alias snd-card-0 snd-ali5451
alias sound-slot-0 snd-card-0
alias char-major-116 snd
alias char-major-14 soundcore
alias sound-service-0-3 snd-pcm-oss
alias sound-service-0-0 snd-mixer-oss
alias sound-service-0-1 snd-seq-oss
alias sound-service-0-12 snd-pcm-oss
alias sound-service-0-6 snd-pcm-oss
alias /dev/dsp sound-service-0-3
I'd be very interested if you could reproduce this with the trident.c
driver in 2.4.latest, 2.6.latest or (best) 2.6.latest-mm. As far as I
can recall, this message was simply commented out, as (we thought at
the time) that it was harmless. If you can reproduce the sound skips
with a recent driver and verify that it doesn't happen with e.g. the
ALSA driver, I'll certainly take a look.
Now running 2.4.22-1.2149.nptl on Fedora Core 1 and the problem is gone.
I've also tried 2.6.1-1.138 (from the rpm packaged by ArjanV) but with
alsa (snd-trident), and that works too. Sorry I don't have non-alsa
drivers in the 2.6 kernel (the prepackaged one).
Just to verify - with the FC1 kernel, you're using the OSS (trident.c)
driver? I'll grab the kernel SRPM and see which driver version is
included there, and whether it's patched.
I'm sorry to hear that the OSS drivers have been removed from arjan's
2.6 kernel. I have at least one bug report where the alsa trident
driver didn't work, and the OSS driver did.
Thanks for your feedback!
From the trident.c comments part (I have the sources on my machine):
* Sept 10 2002 Pascal Schmidt <firstname.lastname@example.org>
* added support for ALi 5451 joystick port
* Sept 05 2002 Alan Cox <email@example.com>
* adapt to new pci joystick attachment interface
* July 24 2002 Muli Ben-Yehuda <firstname.lastname@example.org>
* patch from Eric Lemar (via Ian Soboroff): in suspend and resume,
* fix wrong cast from pci_dev* to struct trident_card*.
* July 19 2002 Muli Ben-Yehuda <email@example.com>
* rewrite the DMA buffer allocation/deallcoation functions, to
* modular and fix a bug where we would call free_pages on memory
* obtained with pci_alloc_consistent. Also remove unnecessary
* CONFIG_PROC_FS and various other cleanups.
* July 19 2002 Muli Ben-Yehuda <firstname.lastname@example.org>
* made several printk(KERN_NOTICE...) into TRDBG(...), to avoid
* my syslog with hundreds of messages.
No, thank you for fixing drivers :-).
Great, it appears to be the stock 2.4 driver. The 2.6 has a few
cleanups and fixes, but I'd rather not apply them to 2.4 at this
stage. Arjan, I think you can close the bug...