Bug 8176

Summary:	E-MU 1820m initialisation is unreliable
Product:	Drivers	Reporter:	Nicolas Mailhot (Nicolas.Mailhot)
Component:	Sound(ALSA)	Assignee:	James Courtier-Dutton (James)
Status:	CLOSED PATCH_ALREADY_AVAILABLE
Severity:	normal	CC:	James, protasnb
Priority:	P2
Hardware:	i386
OS:	Linux
Kernel Version:	up to 2.6.22-rc2	Subsystem:
Regression:	---	Bisected commit-id:
Attachments:	dmesg with broken e-mu 1820m init dmesg of a failed init dmesg of a successful init

Description Nicolas Mailhot 2007-03-11 07:36:34 UTC

Most recent kernel where this bug did *NOT* occur: N/A, longstanding problem
Distribution: Fedora Devel
Hardware Environment: AMD X2 + CK804 + E-MU 1820m 
Software Environment: Init scripts
Problem Description:

Sometimes firmware loading of the E-MU 1820m audiodock fails, resulting on
broken audio system and every LED of the audiodock on (light up as normal on
init, does not light down every led but the ones in use after init)

This has been happening interminently since the alsa driver was written on all
the kernels I've used. Usually rebooting several times on the same kernel fixes
the problem

James Courtier-Dutton had requested a dmesg of a failing kernel, I had sent it
to him before but never opened a formal bug

Steps to reproduce:

Just boot and look at the pretty lights

Comment 1 Nicolas Mailhot 2007-03-11 07:37:43 UTC

Created attachment 10694 [details]
dmesg with broken e-mu 1820m init

Comment 2 Nicolas Mailhot 2007-03-27 11:19:48 UTC

still in 2.6.21-rc5-mm2

Comment 3 Nicolas Mailhot 2007-04-09 03:17:13 UTC

Created attachment 11110 [details]
dmesg of a failed init

Comment 4 Nicolas Mailhot 2007-04-09 03:18:47 UTC

Created attachment 11111 [details]
dmesg of a successful init

Comment 5 Nicolas Mailhot 2007-04-09 03:21:34 UTC

I've just uploaded 2 dmesg from the same kernel, taken at a few minutes of
interval (after a reboot). Kernel is vanilla 2.6.21-rc6-mm1 + nouveau patch
(orthogonal, had the failure way before it was available)

The first boot failed initializing the hardware the second succeeded

This is traced in the logs:

 emu1010: Card options=0x2
 emu1010: Card options=0x2
 emu1010: Card options3=0x2
+emu1010: Loading Audio Dock Firmware
+firmware size=0x133a4
+emu1010: EMU_HANA+DOCK_IRQ_STATUS=0x36
+emu1010: EMU_HANA+DOCK_ID=0x55
+emu1010: Audio Dock Firmware loaded
+Audio Dock ver:3.4

Comment 6 Nicolas Mailhot 2007-05-24 10:40:25 UTC

still occurs on current kernels

Comment 7 James Courtier-Dutton 2007-05-30 16:28:19 UTC

The correct fix to this is to implement a tasklet in the driver to monitor the
plug/unplug of the audio dock, and then load the firmware as a result.
I just have not had time to implement it.

Comment 8 Natalie Protasevich 2007-06-19 09:19:43 UTC

Any updates on the problem?
Do you think it's a good candidate for publishing as a "wanted feature" for takers?
Thanks.

Comment 9 James Courtier-Dutton 2007-07-23 10:00:29 UTC

Fix is alsa-kernel hg repository:
changeset:   5199:61295544011e
tag:         tip
user:        james
date:        Mon Jul 23 17:52:27 2007 +0100
summary:     Summary:snd-emu10k1:Improves firmware loading for E-Mu cards.


I have implemented a kthread to watch for the Audio/Microdock, and as it appears, it loads the firmware.

Please verify the fix.

Comment 10 Nicolas Mailhot 2007-07-24 00:51:39 UTC

James, tahnks a lot
I will test it ASAP. Which may not be too soon unfortunately, as I'm on the eve of leaving for summer holidays

Comment 11 James Courtier-Dutton 2007-11-08 14:41:33 UTC

Any update for this? Does it work for you now?

Comment 12 Nicolas Mailhot 2007-11-11 03:13:55 UTC

Is the patch already in 2.6.23?

I usually test alsa-kernel in mm kernels, but Andrew has not released one for a month now :(

Comment 13 Nicolas Mailhot 2007-11-15 12:17:20 UTC

Ok, Andrew released a mm kernel at last, and it's reasonnably solid (only oopses on shutdown) so I could test latest alsa-devel while chain-rebooting

The initialisation never failed, so I'll mark this as closed

Thank you very much for the fix!