coreboot January 2004

coreboot@coreboot.org

44 participants
99 discussions

Re: EPIA-M AC97 link problem, 20% of boxes
by Dave Ashley 06 Jan '04

06 Jan '04

Here is the complete fix. Put this code at the bottom of src/southbridge/via/vt8235/setup_misc.inc // reset AC97 link movl $CONFIG_ADDR(0, 0x88 + 5, 0x41), %eax movb $0x80, %dl PCI_WRITE_CONFIG_BYTE DELAY($0x5000) movl $CONFIG_ADDR(0, 0x88 + 5, 0x41), %eax movb $0xcc, %dl PCI_WRITE_CONFIG_BYTE The DELAY($0x5000) is arbitrary, I just grabbed that from the raminit code. I was shooting for 1/100th of a second, I don't really know what I got. All this does is force the reset line on the ACLink interface, then remove it. This fixes the problem. The problem was for 20% of our epia-m boxes running linuxbios the ac97 link would be reporting busy all the time (never ready), and the ALSA driver would throw up its hands and audio out wouldn't work. This is the same file to modify for the previous fix recently posted. The previous fix is to enable the real time clock of the VT8235, near the top of this file make it so the entry looks like this: movl $CONFIG_ADDR(0, 0x88, 0x4e), %eax movb $0x1a, %dl PCI_WRITE_CONFIG_BYTE The $0x1a value used to be $0x1b, which for some reason freezes the real time clock. This can easily be tested with "hwclock --show". Before this modification that command will just freeze, but you can control-C out of it. After the mod it works as expected. -Dave

1 0

Re: Source code control
by Adam Megacz 06 Jan '04

06 Jan '04

ebiederman(a)lnxi.com (Eric W. Biederman) writes: > As for features at least as of tla 1.1 I do not believe this to > be the case. Feel free to prove me wrong. Which features do you find present in tla that you do not see in darcs? > It was mentioned in some of your discussion that darcs uses reverse > format patches. I am leary of that because it was exactly that > ``optimization'' in RCS which made CVS suck at branches. Well, I certainly agree that RCS/CVS is awful, and I'm sure that its patch format plays some part in this, but I do not see how that extrapolates to darcs. Have you read the Theory Of Patches? > For myself I am branch happy. Me too! I love branches. The hard part is getting people who don't normally use them to start using them. When they can think of each branch as a separate checkout/repo[*] I find that novices pick up the concept of branching almost instantly. [*] There is no difference between a darcs workspace and repository; every workspace is implicitly a fully-functional repository from which you can take checkouts; in fact, this is exactly how you make a branch. > And I don't think I like the idea of having to create another copy > of a repository to have a branch. Why? Hardlinks eliminate the space concerns in most cases, and in the rest, well, disk is cheap, and human time sure isn't ;) - a

2 1

Re: Source code control
by Adam Megacz 06 Jan '04

06 Jan '04

ron minnich <rminnich(a)lanl.gov> writes: > darcs requires people to get a Haskell and build it. Not so sure I like > this. We solved this problem by distributing statically linked binaries. Not ideal, but generally workable. Feel free to use them: http://dist.xwt.org/darcs/ > In the end, what does arch do better than bitkeeper? I think for many people (including me), BitMover's recent shenanigans are more of a factor than the technical merits of their code. But I really don't want to start that argument here. I do not claim that darcs (or arch) has any features that BK lacks, except perhaps simplicity. - a

1 0

Re: Tiny linux
by Adam Megacz 06 Jan '04

06 Jan '04

ebiederman(a)lnxi.com (Eric W. Biederman) writes: > extracting binary. The nrv2b algorithm has an absolutely tiny 100 > byte or so decompressor. Wow, that's impressive. In the face of such a tiny decompressor LZX might not be worth it; I'm sure the size of the decompression code wasn't on anybody's mind when the algorithm was designed. - a

1 0

Re: EPIA-M AC97 link problem, 20% of boxes
by Dave Ashley 05 Jan '04

05 Jan '04

I worked out a fix. The VT8235 11.6 register 0x41 can be used to reset the AC97 link, when I reset then unreset the thing starts working again. I'll have more details tomorrow. -Dave

2 1

Source code control
by ebiederman＠lnxi.com 05 Jan '04

05 Jan '04

This is just a heads up. I have been doing some investigation into distributed version control systems. arch looks like it usable and getting more usable. >From the design standpoint I don't see any big issues and it appears to be the equal of bk etc. I think I see some was to make it simpler and more powerful but that is not needed to make a good revision control system. I need to do a little more testing to see what arch is actually like to use. Before I seriously start advocating anything but I'm not expecting any weird issues to crop up. Except possibly for some issues running under cygwin, for the platform challenged. Once my evaluation is finished I plan on pushing the projects I work on in that direction. So this is something to think about. Eric

4 7

EPIA-M AC97 link problem, 20% of boxes
by Dave Ashley 05 Jan '04

05 Jan '04

We're into production, and we've found 20% of our boxes based on the epia-m 6000 + linuxbios have a problem with the AC97 codec, the codec reports busy all the time, or never reports a ready status. If I switch back to award bios the codec works fine. This is related to device 11.5 of the VT8235. It seems as though 80% of the boxes have no problem, but 20% always have the problem under linuxbios. I'll need to dig into this and work out a solution, which of course I'll share. I'm wondering if anyone has any suggestions as to what to try. Thanks-- Dave

1 0

Etherboot 5.2.2 ebi images?
by Dave Ashley 05 Jan '04

05 Jan '04

How do I make ebi images for linuxbios in the newer etherboot? I get make: *** No rule to make target `bin/via6105m.ebi'. Stop. which worked before in 5.0.10... -Dave

2 1

Linuxbios -> etherboot 5.0.10, is CONSOLE_DUAL possible?
by Dave Ashley 05 Jan '04

05 Jan '04

I'm using etherboot 5.0.10 on the epia-m, but I have -DLINUXBIOS defined in etherboot's config, not -DPCBIOS. I want some debug messages to come out on the console, but these need -DPCBIOS to be defined. If I define -DPCBIOS alongside -DLINUXBIOS, I get link errors. Anyone have any ideas? Strictly speaking this is an etherboot issue, not a linuxbios issue. Sorry :^) -Dave

2 1

EPIA-M fix for real time clock operation
by Dave Ashley 05 Jan '04

05 Jan '04

In V1 the epia-m real time clock was stopped, the file to change to fix this is src/southbridge/via/vt8235/setup_misc.inc movl $CONFIG_ADDR(0, 0x88, 0x4e), %eax movb $0x1a, %dl PCI_WRITE_CONFIG_BYTE The value used to be 1b, this evidently freezes the clock. Changing to 1a allows normal operation. The low 3 bits are not documented. -Dave

1 0

← Newer
1
2
3
4
5
6
7
8
9
10
Older →

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

coreboot January 2004