| Submitter | Joe Korty |
|---|---|
| Date | 2010-05-11 21:54:29 |
| Message ID | <20100511215429.GA5588@tsunami.ccur.com> |
| Download | mbox | patch |
| Permalink | /patch/1315/ |
| State | Rejected |
| Headers | show |
Comments
Joe Korty wrote: > I applied the below patch to skip 'codec' initialization (whatever > that is) and I was able to boot up to linux. Where did you find the suggestion to do it? > Can anyone explain what the codec is; perhaps we can turn this > 'feature' off permanently? (or make it a .config option). The codec in this case is the chip that converts digital signals into audible sound. It's part of the integrated soundcard in the chipset. It should certainly not cause a hang.. The bad part is that we have no docs for MCP55. In any case, there is code like this in southbridge/nvidia/mcp55/mcp55_aza.c: do { dword = read32(base + 0x68); } while (dword & 1); And that's bad. We shouldn't have code like this anywhere in coreboot. It should get a timeout of some sort, so that it doesn't hang forever. A simple counter would be a great start, but it would of course be best to understand why the code doesn't work on your system.. Can you check which codec you have please? Linux reports this. From my laptop: [ 3.509587] ALSA device list: [ 3.509593] #0: Intel 82801DB-ICH4 with AD1981B at irq 17 AD1981B is the codec. //Peter
On Tue, May 11, 2010 at 08:01:28PM -0400, Peter Stuge wrote: > Joe Korty wrote: > > I applied the below patch to skip 'codec' initialization (whatever > > that is) and I was able to boot up to linux. > > Where did you find the suggestion to do it? I merely looked at the code around the last message to be printed. There was another printk a few lines down that wasn't being reached. As the 'codec' feature sounded worthless (at least at boot time), I decided to comment it out and see how much further in the boot I proceeded. > > Can anyone explain what the codec is; perhaps we can turn this > > 'feature' off permanently? (or make it a .config option). > > The codec in this case is the chip that converts digital signals into > audible sound. It's part of the integrated soundcard in the chipset. > It should certainly not cause a hang.. Thanks! Now I as least know what it is. > The bad part is that we have no docs for MCP55. > > In any case, there is code like this in > southbridge/nvidia/mcp55/mcp55_aza.c: > > do { > dword = read32(base + 0x68); > } while (dword & 1); > > And that's bad. We shouldn't have code like this anywhere in > coreboot. It should get a timeout of some sort, so that it doesn't > hang forever. I'll write up a patch to do that, if no one else beats me to it first. Regards, Joe
Joe Korty wrote: > > do { > > dword = read32(base + 0x68); > > } while (dword & 1); > > > > And that's bad. We shouldn't have code like this anywhere in > > coreboot. It should get a timeout of some sort, so that it doesn't > > hang forever. > > I'll write up a patch to do that, if no one else beats me to it > first. Awesome! Go for it! (Btw what codec is on your board?) //Peter
On Tue, May 11, 2010 at 08:01:28PM -0400, Peter Stuge wrote: > Can you check which codec you have please? Linux reports this. From > my laptop: > > [ 3.509587] ALSA device list: > [ 3.509593] #0: Intel 82801DB-ICH4 with AD1981B at irq 17 > > AD1981B is the codec. Hi Peter, Sorry, I didn't see your request in this, your earlier email. I don't have an 'ALSA' labeled line in either dmesg or in /var/log/messages. I've grepped for other strings that might indicate the sound card used (eg, sound, audio, nvidia, mcp55, ICH) but nothing has come up yet. lspci says: 00:06.1 Audio device: nVidia Corporation MCP55 High Definition Audio (rev a2) Joe
Hi Joe, Joe Korty wrote: > > Can you check which codec you have please? Linux reports this. From > > my laptop: > > > > [ 3.509587] ALSA device list: > > [ 3.509593] #0: Intel 82801DB-ICH4 with AD1981B at irq 17 > > > > AD1981B is the codec. > > Hi Peter, > Sorry, I didn't see your request in this, your earlier email. No problem - thanks for checking this! > I don't have an 'ALSA' labeled line in either dmesg or in > /var/log/messages. Ahh - then the sound drivers simply aren't loaded. There's noone else in the system that would care about the codec so then the info isn't to be found. Unfortunately they manual doesn't mention audio hardware at all. :\ Thanks! //Peter
On Thu, May 13, 2010 at 01:17:23PM -0400, Peter Stuge wrote: > Can you check which codec you have please? Linux reports this. From > my laptop: > > [ 3.509587] ALSA device list: > [ 3.509593] #0: Intel 82801DB-ICH4 with AD1981B at irq 17 > > Joe Korty wrote: >> I don't have an 'ALSA' labeled line in either dmesg or in >> /var/log/messages. > > Ahh - then the sound drivers simply aren't loaded. There's noone else > in the system that would care about the codec so then the info isn't > to be found. > > Unfortunately they manual doesn't mention audio hardware at all. :\ Hi Peter, I just booted with the original flash rather than with coreboot. This is to get the audio properly initialized by BIOS as Sefan mentions is required. Although I don't have the ALSA line (still), I do have these lines in /var/log/messages from the Linux azalia driver: mount used greatest stack depth: 5960 bytes left HDA Intel 0000:00:06.1: PCI->APIC IRQ transform: INT B -> IRQ 23 input: PC Speaker as /class/input/input2 eth1: no link during initialization. hda-intel: azx_get_response timeout, switching to polling mode: last cmd=0x000f0000 hda-intel: Codec #0 probe error; disabling it... hda_intel: azx_get_response timeout, switching to single_cmd mode: last cmd=0x000f0000 hda-intel: no codecs initialized Also, my mainboard has no audio-out jack.... Regards, Joe
Patch
Index: trunk/src/southbridge/nvidia/mcp55/mcp55_aza.c =================================================================== --- trunk.orig/src/southbridge/nvidia/mcp55/mcp55_aza.c 2010-05-06 10:54:13.000000000 -0400 +++ trunk/src/southbridge/nvidia/mcp55/mcp55_aza.c 2010-05-11 13:57:59.000000000 -0400 @@ -237,7 +237,7 @@ if(codec_mask) { printk(BIOS_DEBUG, "codec_mask = %02x\n", codec_mask); - codecs_init(base, codec_mask); + // codecs_init(base, codec_mask); } }