Re: [alsa-devel] iec958 regression with alsa 1.0.22 and kernel 2.6.33 on Intel DG45FC motherboard
Hi David,
On Thu, May 06, 2010 at 07:05:16AM +0800, David Härdeman wrote:
A recent upgrade to kernel 2.6.33.x and alsa 1.0.22 broke spdif (iec958)
Upgraded from which kernel?
output on my Intel DG45FC motherboard (IDT 92HD73E1X5 codec, snd-hda-intel driver).
After playing around a bit with module params, it turns out that loading snd-hda-intel with model=no-jd fixes this problem (previously the module worked without any model option at all, note also that model=intel does not work which is kinda weird).
You mean model=intel never worked?
Any suggestions on a proper fix? The current situation is not really user-friendly...
Thanks, Fengguang
On Thu, May 6, 2010 02:55, Wu Fengguang wrote:
On Thu, May 06, 2010 at 07:05:16AM +0800, David Härdeman wrote:
A recent upgrade to kernel 2.6.33.x and alsa 1.0.22 broke spdif (iec958)
Upgraded from which kernel?
2.6.32.something, can check exact version later if you want. alsa userspace bits were upgraded at the same time (can also check from which version later if necessary).
output on my Intel DG45FC motherboard (IDT 92HD73E1X5 codec, snd-hda-intel driver).
After playing around a bit with module params, it turns out that loading snd-hda-intel with model=no-jd fixes this problem (previously the module worked without any model option at all, note also that model=intel does not work which is kinda weird).
You mean model=intel never worked?
No, I mean that I've never used a model= parameter before, and in trying to debug this new problem I tried model=intel which did not fix it but model=no-jd does. So model=intel might have worked before.
On a related note, why is model=intel even there (and why is the end-user expected to set it)? The docs says it's for DG45FC and DG45ID motherboards, but shouldn't those motherboards be possible to autodetect in the driver via DMI info?
On Thu, May 06, 2010 at 03:42:27PM +0800, David Härdeman wrote:
On Thu, May 6, 2010 02:55, Wu Fengguang wrote:
On Thu, May 06, 2010 at 07:05:16AM +0800, David Härdeman wrote:
A recent upgrade to kernel 2.6.33.x and alsa 1.0.22 broke spdif (iec958)
Upgraded from which kernel?
2.6.32.something, can check exact version later if you want. alsa userspace bits were upgraded at the same time (can also check from which version later if necessary).
Yes, should be OK in 2.6.32, and broken by this commit:
commit 52dc438606d1ef78b96f56cc04dbea9242005730 Author: Alexey Fisher bug-track@fisher-privat.net Date: Sat Dec 12 11:16:41 2009 +0200
ALSA: hda - Overwrite pin config on intel DG45ID board.
The pin config provided by BIOS have some problems: 0x0221401f: [Jack] HP Out at Ext Front <-- other association and sequence 0x02a19020: [Jack] Mic at Ext Front <-- other association 0x01113014: [Jack] Speaker at Ext Rear <-- line out (not speaker) 0x01114010: [Jack] Speaker at Ext Rear <-- line out 0x01a19030: [Jack] Mic at Ext Rear <-- other association 0x01111012: [Jack] Speaker at Ext Rear <-- line out 0x01116011: [Jack] Speaker at Ext Rear <-- line out 0x40f000f0: [N/A] Other at Ext N/A 0x40f000f0: [N/A] Other at Ext N/A 0x40f000f0: [N/A] Other at Ext N/A 0x40f000f0: [N/A] Other at Ext N/A 0x40f000f0: [N/A] Other at Ext N/A 0x01451140: [Jack] SPDIF Out at Ext Rear 0x40f000f0: [N/A] Other at Ext N/A
just overwrite it.
Signed-off-by: Alexey Fisher bug-track@fisher-privat.net Signed-off-by: Takashi Iwai tiwai@suse.de
output on my Intel DG45FC motherboard (IDT 92HD73E1X5 codec, snd-hda-intel driver).
After playing around a bit with module params, it turns out that loading snd-hda-intel with model=no-jd fixes this problem (previously the module worked without any model option at all, note also that model=intel does not work which is kinda weird).
You mean model=intel never worked?
No, I mean that I've never used a model= parameter before, and in trying to debug this new problem I tried model=intel which did not fix it but model=no-jd does. So model=intel might have worked before.
Yes, model=intel worked before.
On a related note, why is model=intel even there (and why is the end-user expected to set it)? The docs says it's for DG45FC and DG45ID motherboards, but shouldn't those motherboards be possible to autodetect in the driver via DMI info?
It has to be assigned case by case, since we never know whether the current settings will apply automatically to future Intel boards..
Thanks, Fengguang
David,
The attached patch should fix your broken SPDIF output.
Also attached more data for possible future reference.
Thanks, Fengguang --
On Thu, May 06, 2010 at 03:42:27PM +0800, David Härdeman wrote:
On Thu, May 6, 2010 02:55, Wu Fengguang wrote:
On Thu, May 06, 2010 at 07:05:16AM +0800, David Härdeman wrote:
A recent upgrade to kernel 2.6.33.x and alsa 1.0.22 broke spdif (iec958)
Upgraded from which kernel?
2.6.32.something, can check exact version later if you want. alsa userspace bits were upgraded at the same time (can also check from which version later if necessary).
output on my Intel DG45FC motherboard (IDT 92HD73E1X5 codec, snd-hda-intel driver).
After playing around a bit with module params, it turns out that loading snd-hda-intel with model=no-jd fixes this problem (previously the module worked without any model option at all, note also that model=intel does not work which is kinda weird).
You mean model=intel never worked?
No, I mean that I've never used a model= parameter before, and in trying to debug this new problem I tried model=intel which did not fix it but model=no-jd does. So model=intel might have worked before.
On a related note, why is model=intel even there (and why is the end-user expected to set it)? The docs says it's for DG45FC and DG45ID motherboards, but shouldn't those motherboards be possible to autodetect in the driver via DMI info?
-- David Härdeman
On Thu, May 06, 2010 at 06:20:17PM +0800, Wu Fengguang wrote:
The attached patch should fix your broken SPDIF output.
Yup, it works (tried with model=intel).
participants (2)
-
David Härdeman -
Wu Fengguang