ECS k7s5a motherboard doesnt work

ECS k7s5a motherboard doesnt work

[Justin Mierta]

First of all, I'm not subscribed to the list, could any comments please 
be cc'd to crazed_cowboy@stones.com

Secondly, I thank you all very much for taking the time to help.

Here's the problem.  I just bought a 1.4 GHz AMD athlon and an ECS k7s5a 
motherboard.  I tried to install linux using this motherboard several 
dozen times, using several different distros (mandrake, redhat, and 
debian).  Not once did it install completely.  The problems ranged from 
the partition table being scrambled when it attempted to make the 
partitions, to failure to access harddisk, to failure to access the 
cdrom.  Cdrom failures were actually the most common reason for the 
failed install.  I tried the different distros, as well as an old 
version of mandrake which i know the install cd works (i used it on 
another computer).

Before the harddisk scrambling occurred the once or twice, I thought 
maybe my cdrom was no longer functioning (even though the same cdrom 
drive worked fine with linux with a different motherboard).  So I tried 
using a cdr drive, a dvd drive, and a regular 48x cdrom drive, and all 
of the distros had issues.

For this reason, i'm leaning towards believing that the ECS k7s5a's 
built-in ultradma controller is incompatable with linux.

Has anyone else experienced this, is this a known issue, and is any work 
being done to fix it?  or does anyone have any suggestions?

Thank you all very much for your time,
Justin

Re: ECS k7s5a motherboard doesnt work

[Alan Cox]

> Has anyone else experienced this, is this a known issue, and is any work 
> being done to fix it?  or does anyone have any suggestions?

The sort of things you describe sound like straight forward broken hardware
not even incompatibilities - bad ram , bad board, bad CPU could all do this

Re: ECS k7s5a motherboard doesnt work

[Justin Mierta]

here's some more info for everyone trying to help:
redhat says (at some point) that it insmod'd sis900, not the sis5513 
that mark had suggested, and the sis735 that everything says is actually 
in the motherboard.  (i have to assume the sis900 was ide, because it 
didnt tell me anything more useful than "insmod sis900" and that it worked)

however, the messages i'm seeing are something like this:
hda atapi: reset complete
irq timeout: complete status = 0xC0   { busy }       (sometimes thats 0xD0)

then there's a few lines every now and then saying "drive not ready for 
command" and "I/O error"

any suggestions?

justin

Re: ECS k7s5a motherboard doesnt work

[Joel Jaeggli]

sis900 is the integrated ethernet controller...

joelja

On Mon, 29 Oct 2001, Justin Mierta wrote:

> here's some more info for everyone trying to help:
> redhat says (at some point) that it insmod'd sis900, not the sis5513
> that mark had suggested, and the sis735 that everything says is actually
> in the motherboard.  (i have to assume the sis900 was ide, because it
> didnt tell me anything more useful than "insmod sis900" and that it worked)
>
> however, the messages i'm seeing are something like this:
> hda atapi: reset complete
> irq timeout: complete status = 0xC0   { busy }       (sometimes thats 0xD0)
>
> then there's a few lines every now and then saying "drive not ready for
> command" and "I/O error"
>
> any suggestions?
>
> justin
>
>
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/
>

-- 
--------------------------------------------------------------------------
Joel Jaeggli				       joelja@darkwing.uoregon.edu
Academic User Services			     consult@gladstone.uoregon.edu
     PGP Key Fingerprint: 1DE9 8FCA 51FB 4195 B42A 9C32 A30D 121E
--------------------------------------------------------------------------
It is clear that the arm of criticism cannot replace the criticism of
arms.  Karl Marx -- Introduction to the critique of Hegel's Philosophy of
the right, 1843.

Re: ECS k7s5a motherboard doesnt work

[Justin Mierta]

I'm using this archive:
http://www.uwsg.indiana.edu/hypermail/linux/kernel/index.html

and nowhere can i find the thread you mentioned.

Also, this motherboard works *perfect* in windows (which is 
unfortunately where i've been stuck) -- it hasnt crashed in the last 
several weeks i've been using it! (probably the longest uptime i've ever 
seen for a windows machine that's actually being used)  So I *seriously* 
doubt its bad hardware.

Can  you maybe give me a link to the thread you mention?

Justin


lung@theuw.net wrote:

>You might be interested in my recently initiated tread:
>
>Subject: Repeatable File Corruption (ECS K7S5A w/SIS735)
>
>
>You should always check the archives before posting.  You problems seem
>much worse than mine, so I'm going to assume there is something else not
>working for you.  Alan's suggestion that you might have some bad hardware
>makes sense.  RAM is a common culprit.  Good luck.
>
>

Re: ECS k7s5a motherboard doesnt work

[Justin Mierta]

well, i dont have a floppy drive, so that test is a little difficult to 
do, but i threw some ram in there that i have used in linux before, and 
i still had the slew of ide error messages.  and this harddrive has 
worked in linux before.  i'm getting more and more convinced its an ide 
controller +linux issue.

plus, i just discovered this:
http://www.uwsg.iu.edu/hypermail/linux/kernel/0109.1/0198.html

which really points to ide controller and linux not fighting nicely 
together, altho the thread doesnt really point towards a solution.

justin


Alan Cox wrote:

>>well, i've been using it in win98 (unfortunately) and it works 
>>perfectly.  i havent even had it crash at all.  so i doubt its a 
>>hardware thing...
>>
>
>That proves absolutely nothing. Win98 stresses the system differently to
>Linux. Run test tools like memtest86 on it
>

Re: ECS k7s5a motherboard doesnt work

[Adrian Burgess]

On Mon, Oct 29, 2001 at 10:10:28AM -0500, Justin Mierta wrote:
> 
> well, i dont have a floppy drive, so that test is a little difficult to 
> do, but i threw some ram in there that i have used in linux before, and 
> i still had the slew of ide error messages.  and this harddrive has 
> worked in linux before.  i'm getting more and more convinced its an ide 
> controller +linux issue.
> 
> plus, i just discovered this:
> http://www.uwsg.iu.edu/hypermail/linux/kernel/0109.1/0198.html
> 
> which really points to ide controller and linux not fighting nicely 
> together, altho the thread doesnt really point towards a solution.
> 
> justin

Ok, well, I'm still using the KS75A under Linux, Win98, and WinXP, and the
only OS that seems to have problems is Linux.  I don't have any data
corruption problems (as far as I know), but I'm still getting the DMA
timeout errors reported in the thread above.  There definitely seems to be
more reports than normal about IDE problems with Linux on this board.  And
other than IDE the board seems pretty stable.
Btw, the SIS900 is the on-board Ethernet controller.

Cheers,
Adrian

Re: ECS k7s5a motherboard doesnt work

[lung]

I had started a similar thread last week:

http://www.uwsg.iu.edu/hypermail/linux/kernel/0109.1/0198.html


There is clearly something wrong with the sis ide driver.  I've lost years
of information due to backups also being corrupt.  I'll hopefully be
testing a patch in the next day or two from Andre.

Reading your previous posts, I'd still have to agree with Alan that
have something else going wrong.  I've had similar problems where things
have worked in windows, but then not even being able to install Linux.
The results problems have ended up being bad ram and inadequate cooling.


On Mon, 29 Oct 2001, Justin Mierta wrote:

> well, i dont have a floppy drive, so that test is a little difficult to
> do, but i threw some ram in there that i have used in linux before, and
> i still had the slew of ide error messages.  and this harddrive has
> worked in linux before.  i'm getting more and more convinced its an ide
> controller +linux issue.
>
> plus, i just discovered this:
> http://www.uwsg.iu.edu/hypermail/linux/kernel/0109.1/0198.html
>
> which really points to ide controller and linux not fighting nicely
> together, altho the thread doesnt really point towards a solution.
>
> justin
>
>
> Alan Cox wrote:
>
> >>well, i've been using it in win98 (unfortunately) and it works
> >>perfectly.  i havent even had it crash at all.  so i doubt its a
> >>hardware thing...
> >>
> >
> >That proves absolutely nothing. Win98 stresses the system differently to
> >Linux. Run test tools like memtest86 on it
> >
>
>
>
>

Re: ECS k7s5a motherboard doesnt work

[Zephaniah E. Hull]

[-- Attachment #1: Type: text/plain, Size: 1792 bytes --]

On Mon, Oct 29, 2001 at 10:10:28AM -0500, Justin Mierta wrote:
> well, i dont have a floppy drive, so that test is a little difficult to 
> do, but i threw some ram in there that i have used in linux before, and 
> i still had the slew of ide error messages.  and this harddrive has 
> worked in linux before.  i'm getting more and more convinced its an ide 
> controller +linux issue.
> 
> plus, i just discovered this:
> http://www.uwsg.iu.edu/hypermail/linux/kernel/0109.1/0198.html
> 
> which really points to ide controller and linux not fighting nicely 
> together, altho the thread doesnt really point towards a solution.

I'm using a K7S5A with recent 2.4.x kernels, and have no major
problems[0].

I don't remember why I have CONFIG_IDEDISK_MULTI_MODE on, but I do.

I also have CONFIG_IDEDMA_NEW_DRIVE_LISTINGS on, and obviously have
CONFIG_BLK_DEV_SIS5513 on.

Oh, and I have CONFIG_IDEDMA_IVB as well.

The only problems I have seen with this board are that I can't find
drivers for the sound (no big loss), lmsensors does not seem to be able
to properly read the sensors (annoying), repeated 'VFS: Disk change
detected on device ide1(22,0)' messages (my cdrom drive, getting a
little annoying), and, thats about it.

I have not seen any data corruption.

My obvious question is what kernels are you running, and are you
enabling CONFIG_BLK_DEV_SIS5513 or not?

Zephaniah E. Hull.
> 
> justin

-- 
	1024D/E65A7801 Zephaniah E. Hull <warp@babylon.d2dc.net>
	   92ED 94E4 B1E6 3624 226D  5727 4453 008B E65A 7801
	    CCs of replies from mailing lists are requested.

   "<green>From</yellow>"
   "Wow. The green word From is no longer yellow. That's deep, man."
 -- Marcus Meissner & Lars Balker Rasmussen in the Scary Devil Monastery

[-- Attachment #2: Type: application/pgp-signature, Size: 232 bytes --]

Re: ECS k7s5a motherboard doesnt work

[Alan Cox]

> The only problems I have seen with this board are that I can't find
> drivers for the sound (no big loss), lmsensors does not seem to be able

[ALSA has one I believe]

> to properly read the sensors (annoying), repeated 'VFS: Disk change
> detected on device ide1(22,0)' messages (my cdrom drive, getting a
> little annoying), and, thats about it.

rpm -e magicdev

Re: ECS k7s5a motherboard doesnt work

[Daniel Freedman]

Justin Mierta (Crazed_Cowboy@stones.com) wrote:
> and nowhere can i find the thread you mentioned. 
> 
> Also, this motherboard works *perfect* in windows (which is 
> unfortunately where i've been stuck) -- it hasnt crashed in the last 
> several weeks i've been using it! (probably the longest uptime i've ever 
> seen for a windows machine that's actually being used) So I *seriously* 
> doubt its bad hardware. 
> 
> Can you maybe give me a link to the thread you mention? 
> 
> Justin 
> 

Justin,

Hi!  I've purchased three of these ECS K7S5A motherboards (based upon
Sis 735 chipsets) and have installed Linux (specifically Debian 2.2r3,
with 2.2.19 kernel) on all of them without any difficulty or hiccups.
This would tend to suggest, as Alan implied, that the board is not
incompatible (at least not wildly so, as you describe) and that you
may have hardware issues, even if Windows does run on this setup.

The main issue that I believe exists with this board, or actually with
the board's chipset (people, please correct me if I'm wrong), is that
the current 2.2.19 does not support the built-in ethernet capabilities
of the Sis 735.  I've tried Sis's patch for the kernel from their
website, but that won't even compile.  I'm planning to look into Alan's
2.2.20pre11 release that supposedly includes a Sis900 chipset update
(Sis seems to suggest that the 900 and 735 use the same drivers, at
least to some degree) and maybe this will work.  In the meantime, I
just installed a NIC.

Hope this helps, Justin.  I'd love to hear if anyone else knows about
the Sis735 ethernet issue.

Take care,
Daniel

PS Please CC me on replies.

-- 
Daniel A. Freedman
Laboratory for Atomic and Solid State Physics
Department of Physics
Cornell University

Re: ECS k7s5a motherboard doesnt work

[Justin Mierta]

>
>
>>2) its not the cables, i checked them
>>
>most people are unaware that IDE must always be <=18".  I mean no
>insult, but do you know the rules for ide?
>

no, i'm not taking it as an insult, but yes, they are shorter than 18". 
 i dont have any cables that long, but i wouldnt expect them to work at 
all at ata100 if they were that long....(hence why i figured saying that 
it works in the other OSs was also saying the cables were shorter than 18")

>>3) its not the ram, i checked it
>>
>how?
>
this motherboard is an upgrade.  the old motherboard/ram/processor 
worked fine in all the linux's, as well as freebsd and beos.  i put that 
ram in there (pc133), and i still get dma errors.  i also tried my 
roommate's ddr ram (he used to use linux with it), and still the dma errors.

>>4) its not the harddrive, cdroms - they worked fine in linux before i 
>>upgraded the mobo
>>
>same arrangement?
>
yeah.  in fact, they were never unplugged (except for obviously from the 
motherboard) until i was testing the cables.

>>8) seems to be having real dma issues.  BeOS doesnt use DMA, and i'm 
>>pretty sure that FreeBSD wasnt either (because it wasnt familiar with 
>>the chipset, so was just going thru the bios)
>>
>what mode does it come up in?  do you do some hdparm tweaking?
>anything in /proc/ide?
>
well, when freebsd boots it says a few lines about "Drive C: bios" or 
something like that.  i'm not really very familiar with it.  but no, i 
didnt use any special settings.  its not exactly easy, but i can 
probably get you the first few lines of dmesg if you think it'll help

>>9) *new* - the machine is not overheating, the hottest spot is at a cool 
>>57 C
>>
>57C is hardly cool, though it shouldn't cause problems.
>
last i saw, amd rates these chips at 95C, so 57 is downright chilly :)

>>now i'm not sure where the problem is, but it seems pretty clearly to me 
>>that it is dma-related.  neither BeOS or FreeBSD is trying to use dma, 
>>and they work. i'm pretty sure windows is using dma, but its using the 
>>drivers that came with the mobo.  linux is trying to use dma using 
>>drivers not written by ECS, and it doesnt work.
>>
>udma or even mdma?  if mdma, does hdparm -m settings effect it?
>
i'm not sure how to check, and i have a very difficult time even getting 
to a shell in linux, because the damn thing keeps dma erroring about 
reading the cd's.  is there some boot-up settings i can feed it so it 
wont try using dma at all?


>>or that there's some minor timing quirk (like a 
>>race condition) in the linux driver that my computer just likes to hit? 
>>
>unlikley.  in part because the sis driver, like most others,
>doesn't do much itself, since the interesting/subtle parts of 
>doing ide are all done in generic code.
>
>but why do you mention timing?  sure, the sis driver is hardly the most
>well tested (widely used) driver, so perhaps it fiddles a reg that 
>needs a microsecond of "rest" afterwards.
>
i just mentioned timing, because in my (somewhat limited) experience 
writing drivers, timing booboos tend to be the cause of lots of problems 
(especially when timeouts are involved)

>>at this point, i'm tending to think that there's several versions of 
>>sis735 floating around (similar to the maneuver that ensoniq pulled with 
>>their sound cards) -- possibly even within the revisions of the k7s5a 
>>motherboard itself.
>>
>lspci /proc/pci will give you the revisions of the chipset components.
>
again, cant easily get to a command prompt in linux, making things like 
this very difficult :)

i think the best thing is to try and find a startup parameter to prevent 
it from using dma, and see if that works.  anyone know how? :)

justin

Re: ECS k7s5a motherboard doesnt work

[Marco Colombo]

On Mon, 29 Oct 2001, Justin Mierta wrote:

> >>9) *new* - the machine is not overheating, the hottest spot is at a cool
> >>57 C
> >>
> >57C is hardly cool, though it shouldn't cause problems.
> >
> last i saw, amd rates these chips at 95C, so 57 is downright chilly :)

yup, but see the cooler man (if it has one). Some of them are rated for
< 70C - above that they stop and melt (in whatever order).

> i'm not sure how to check, and i have a very difficult time even getting
> to a shell in linux, because the damn thing keeps dma erroring about
> reading the cd's.  is there some boot-up settings i can feed it so it
> wont try using dma at all?

see Documentation/kernel-parameters.txt:

ide0=nodma ide1=nodma

should do.

The following is on my K7S5A (no onboard eth):

# hdparm -tT /dev/hda

/dev/hda:
 Timing buffer-cache reads:   128 MB in  0.53 seconds =241.51 MB/sec
 Timing buffered disk reads:  64 MB in  1.82 seconds = 35.16 MB/sec

kernel is 2.2.19-xxx from RH, but when I tested it performed the same
under 2.4.x. Its uptime is only 9 days, but no problems so far. It's
a NFS, samba, SMTP, IMAP, HTTP server, not heavily loaded.

> >>at this point, i'm tending to think that there's several versions of
> >>sis735 floating around (similar to the maneuver that ensoniq pulled with
> >>their sound cards) -- possibly even within the revisions of the k7s5a
> >>motherboard itself.

Mine doesn't have an eth on board. Have you tried and disabled it?

.TM.
-- 
      ____/  ____/   /
     /      /       /			Marco Colombo
    ___/  ___  /   /		      Technical Manager
   /          /   /			 ESI s.r.l.
 _____/ _____/  _/		       Colombo@ESI.it