Linux 2.4.21pre4-ac5 status report

Linux 2.4.21pre4-ac5 status report

[Soeren Sonnenburg]

hi.

I wanted to give some status report after using this kernel for 1 week.

The harddisk freeze bug that I got with WDC WD1800JB-00DUA0 drives did
not occur within this week (it used to happen within 5 days).

The promise driver still freezes on my pdc20268 when using >mdma0 . I
recently replaced the ultra tx2 controller with a hpt370 and it works
all fine. So I would suggest either removing the pdc20268 from the list
of supported controllers or letting promise fix this bug.

The overlay mode of my bt848 based tv card does not work anymore
(stripes as if the linewidth/offset is wrong).

Best,
Soeren (who has for the first time since some months an uptime of >1
week).

re: Linux 2.4.21pre4-ac5 status report

[Tomas Szepe]

> [kernel@nn7.de]
> 
> The promise driver still freezes on my pdc20268 when using >mdma0.

Try upgrading the BIOS on the 20268.  Sounds incredible, but it
did solve all the problems I was seeing with the card in Linux.

-- 
Tomas Szepe <szepe@pinerecords.com>

re: Linux 2.4.21pre4-ac5 status report

[Soeren Sonnenburg]

On Sat, 2003-03-01 at 09:57, Tomas Szepe wrote:
> > [kernel@nn7.de]
> > 
> > The promise driver still freezes on my pdc20268 when using >mdma0.
> 
> Try upgrading the BIOS on the 20268.  Sounds incredible, but it
> did solve all the problems I was seeing with the card in Linux.

I checked for new bioses 2 on thursday -> there where no newer bioses
available. I guess you have a single pdc20268 which is full of disks
(i.e. master and slave used?).

Soeren.

re: Linux 2.4.21pre4-ac5 status report

[Tomas Szepe]

> [kernel@nn7.de]
> 
> > > [kernel@nn7.de]
> > > 
> > > The promise driver still freezes on my pdc20268 when using >mdma0.
> > 
> > Try upgrading the BIOS on the 20268.  Sounds incredible, but it
> > did solve all the problems I was seeing with the card in Linux.
> 
> I checked for new bioses 2 on thursday -> there where no newer bioses
> available. I guess you have a single pdc20268 which is full of disks
> (i.e. master and slave used?).

I've got a single pdc20268 with just one drive on each channel...
Works nicely with recent -ac kernels.

-- 
Tomas Szepe <szepe@pinerecords.com>

re: Linux 2.4.21pre4-ac5 status report

[Soeren Sonnenburg]

On Sat, 2003-03-01 at 11:32, Tomas Szepe wrote:
> > [kernel@nn7.de]
> > 
> > > > [kernel@nn7.de]
> > > > 
> > > > The promise driver still freezes on my pdc20268 when using >mdma0.
> > > 
> > > Try upgrading the BIOS on the 20268.  Sounds incredible, but it
> > > did solve all the problems I was seeing with the card in Linux.
> > 
> > I checked for new bioses 2 on thursday -> there where no newer bioses
> > available. I guess you have a single pdc20268 which is full of disks
> > (i.e. master and slave used?).
> 
> I've got a single pdc20268 with just one drive on each channel...
> Works nicely with recent -ac kernels.

As I guessed. I've got two pdc20268 with just one drive per channel
(where the last drive is a cdrom-drive)

So one pdc no problem >1 -> trouble.

S.

re: Linux 2.4.21pre4-ac5 status report

[Mikael Pettersson]

On 01 Mar 2003 11:47:39 +0100, Soeren Sonnenburg wrote:
>> I've got a single pdc20268 with just one drive on each channel...
>> Works nicely with recent -ac kernels.
>
>As I guessed. I've got two pdc20268 with just one drive per channel
>(where the last drive is a cdrom-drive)
>
>So one pdc no problem >1 -> trouble.

It might not have anything to do with your problem, but it has
been reported several times that Promise chips don't support ATAPI
optical devices (i.e. your CD-ROM) without special driver support,
which the Linux drivers apparently don't have.
Maybe that's changed in 2.4.21-pre-ac new IDE code, I don't know.

Your cards don't share interrupts with anything else I hope?

/Mikael

re: Linux 2.4.21pre4-ac5 status report

[Soeren Sonnenburg]

On Sat, 2003-03-01 at 13:52, Mikael Pettersson wrote:
> On 01 Mar 2003 11:47:39 +0100, Soeren Sonnenburg wrote:
> >> I've got a single pdc20268 with just one drive on each channel...
> >> Works nicely with recent -ac kernels.
> >
> >As I guessed. I've got two pdc20268 with just one drive per channel
> >(where the last drive is a cdrom-drive)
> >
> >So one pdc no problem >1 -> trouble.
> 
> It might not have anything to do with your problem, but it has
> been reported several times that Promise chips don't support ATAPI
> optical devices (i.e. your CD-ROM) without special driver support,
> which the Linux drivers apparently don't have.

Yes it doesn't have to do with my particular problem as I tried without
an ATAPI cdrom attached.

> Maybe that's changed in 2.4.21-pre-ac new IDE code, I don't know.
> 
> Your cards don't share interrupts with anything else I hope?

sure they do. however it did not work when the pdc's did not share an
IRQ with any other device.

however the htp370 works fine with shared irqs and atapi cdrom.

Soeren.

Re: Linux 2.4.21pre4-ac5 status report

[Edward King]

Soeren Sonnenburg wrote:

>On Sat, 2003-03-01 at 13:52, Mikael Pettersson wrote:
>  
>
>>On 01 Mar 2003 11:47:39 +0100, Soeren Sonnenburg wrote:
>>    
>>
>>>As I guessed. I've got two pdc20268 with just one drive per channel
>>>(where the last drive is a cdrom-drive)
>>>
>>>So one pdc no problem >1 -> trouble.
>>>      
>>>
>>Maybe that's changed in 2.4.21-pre-ac new IDE code, I don't know.
>>
>>Your cards don't share interrupts with anything else I hope?
>>    
>>
I tried  two pdc20268's which failed miserably

Used an Asus motherboard and an FIC motherboard, different cables, 
different cards, different powersupply.Hard drives are 200GB western 
digitals, one drive per channel.  

Tried an SIIG card with the SiI680 chipset -- same problem using is and 
the pdc20268, but is more stable than a single pdc -- so now I have 4 
drives on that card.

My kernel is 2.4.21-pre4-ac6 -- let me know if the pre5's solve the problem.

Ed King

re: Linux 2.4.21pre4-ac5 status report

[Tomas Szepe]

> [edk@cendatsys.com]
> 
> >>Maybe that's changed in 2.4.21-pre-ac new IDE code, I don't know.
> >>
> >>Your cards don't share interrupts with anything else I hope?
> >>
> I tried  two pdc20268's which failed miserably
> 
> Used an Asus motherboard and an FIC motherboard, different cables, 
> different cards, different powersupply.Hard drives are 200GB western 
> digitals, one drive per channel.  
> 
> Tried an SIIG card with the SiI680 chipset -- same problem using is and 
> the pdc20268, but is more stable than a single pdc -- so now I have 4 
> drives on that card.
> 
> My kernel is 2.4.21-pre4-ac6 -- let me know if the pre5's solve the problem.

I'd be quite interested to know whether the FreeBSD IDE driver can handle
these setups properly.

-- 
Tomas Szepe <szepe@pinerecords.com>

re: Linux 2.4.21pre4-ac5 status report

[Soeren Sonnenburg]

On Mon, 2003-03-03 at 20:57, Tomas Szepe wrote:
> > [edk@cendatsys.com]
> > 
> > >>Maybe that's changed in 2.4.21-pre-ac new IDE code, I don't know.
> > >>
> > >>Your cards don't share interrupts with anything else I hope?
> > >>
> > I tried  two pdc20268's which failed miserably
> > 
> > Used an Asus motherboard and an FIC motherboard, different cables, 
> > different cards, different powersupply.Hard drives are 200GB western 
> > digitals, one drive per channel.  
> > 
> > Tried an SIIG card with the SiI680 chipset -- same problem using is and 
> > the pdc20268, but is more stable than a single pdc -- so now I have 4 
> > drives on that card.
> > 
> > My kernel is 2.4.21-pre4-ac6 -- let me know if the pre5's solve the problem.
> 
> I'd be quite interested to know whether the FreeBSD IDE driver can handle
> these setups properly.

Me too... I would not think so, but who know.

However at least the freeze was reproducable here. Just take two
pdc20268 controller, attach one driver per channel, setup a raid5, let
it build up the raid. Then start bonnie and you will get a freeze within
15min when using something higher than mdma0.

The system now is really rock stable with these htp370 based
dawicontrol-100 cards (which are in the same slots as the pdcs were).

Soeren.