regarding conversion to new EH

regarding conversion to new EH

[Tejun Heo]

Hello, Jeff, Mark.

Currently, EH conversion status in #upstream is...

complete:	ata_piix, ahci, sata_sil, sata_sil24
submitted:	sata_sis, sata_svw, sata_uli, sata_vsc, sata_via, sata_nv
trying:		sata_qstor, pdc_adma
todo:		sata_sx4, sata_promise, sata_mv

For the last three todo drivers, I have no idea how those things work
and don't have access to the hardware or doc.

Jeff, sx4 seems to be a strain of promise controllers.  Can you
convert that one too in addition to sata_promise?

I guess the two 'trying' conversions are probably broken.  Mark, it
would be great if you can verify them.  Also, can you please convert
sata_mv to new EH?

I'll now convert the PATA drivers.

Thanks a lot.

-- 
tejun

Re: regarding conversion to new EH

[Jeff Garzik]

Tejun Heo wrote:
> Jeff, sx4 seems to be a strain of promise controllers.  Can you
> convert that one too in addition to sata_promise?

Yeah, I'll take a look at those.


> I guess the two 'trying' conversions are probably broken.  Mark, it
> would be great if you can verify them.  Also, can you please convert
> sata_mv to new EH?

I'll see if I can find a PDC ADMA card, to verify.

	Jeff

Re: regarding conversion to new EH

[Mark Lord]

Jeff Garzik wrote:
> Tejun Heo wrote:
>> Jeff, sx4 seems to be a strain of promise controllers.  Can you
>> convert that one too in addition to sata_promise?
> 
> Yeah, I'll take a look at those.
> 
> 
>> I guess the two 'trying' conversions are probably broken.  Mark, it
>> would be great if you can verify them.  Also, can you please convert
>> sata_mv to new EH?
> 
> I'll see if I can find a PDC ADMA card, to verify.

I've still got a box full of the PATA ADMA cards here (Tejun: want a couple?),
so I'll test that as well, along with the sata_qstor, after the weekend.

Once I take time to see exactly how this new EH stuff works,
I'd love to convert sata_mv.  It's EH current EH works,
but it locks up the machine for a few seconds while doing things.

Cheers
-- 
Mark Lord
Real-Time Remedies Inc.
mlord@pobox.com

Re: regarding conversion to new EH

[Tejun Heo]

Mark Lord wrote:
>> I'll see if I can find a PDC ADMA card, to verify.
> 
> I've still got a box full of the PATA ADMA cards here (Tejun: want a 
> couple?),

That would be great.  I'll mail you my address in private mail.

> so I'll test that as well, along with the sata_qstor, after the weekend.
> 
> Once I take time to see exactly how this new EH stuff works,
> I'd love to convert sata_mv.  It's EH current EH works,
> but it locks up the machine for a few seconds while doing things.

Great.  As Alan is converting PATA drivers, it seems that we'll be able 
to kill old EH pretty soon.

Thanks.

-- 
tejun

Re: regarding conversion to new EH

[Mark Lord]

Mark Lord wrote:
>> I'll see if I can find a PDC ADMA card, to verify.

Heh.. I was getting ready to package a pair of ADMA cards for Tejun just now,
and decided to test them first to ensure they were "good cards".

My AMD64 test system, running 2.6.17, gets stuck in libata initialization
for the cards -- looks like interrupts aren't working in the driver.

So now I'm building a 2.6.15 kernel, to run on the same machine that I
originally used to test this driver (a P4 system), and we'll eventually
find out what went wrong.   And maybe I can get these cards tested and
shipped, too.  (never lose sight of the original goal!).

Cheers
-- 
Mark Lord
Real-Time Remedies Inc.
mlord@pobox.com

Re: regarding conversion to new EH

[Mark Lord]

Mark Lord wrote:
> 
> My AMD64 test system, running 2.6.17, gets stuck in libata initialization
> for the pdc_adma cards -- looks like interrupts aren't working in the driver.
> 
> So now I'm building a 2.6.15 kernel, to run on the same machine that I
> originally used to test this driver (a P4 system), and we'll eventually
> find out what went wrong.   And maybe I can get these cards tested and
> shipped, too.  (never lose sight of the original goal!).

Okay, no problems on 2.6.15.7, so it must be something that changed
in libata/pdc_adma since then.  I'll start a 2.6.16.15 build now..

Cheers
-- 
Mark Lord
Real-Time Remedies Inc.
mlord@pobox.com

Re: regarding conversion to new EH

[Mark Lord]

Mark Lord wrote:
> Mark Lord wrote:
>>
>> My AMD64 test system, running 2.6.17, gets stuck in libata initialization
>> for the pdc_adma cards -- looks like interrupts aren't working in the 
>> driver.
>>
>> So now I'm building a 2.6.15 kernel, to run on the same machine that I
>> originally used to test this driver (a P4 system), and we'll eventually
>> find out what went wrong.   And maybe I can get these cards tested and
>> shipped, too.  (never lose sight of the original goal!).
> 
> Okay, no problems on 2.6.15.7, so it must be something that changed
> in libata/pdc_adma since then.  I'll start a 2.6.16.15 build now..

MMmmm.. that's interesting.

The pdc_adma driver works fine in all versions including 2.6.17 on my P4 box.
I wonder why it's fussy on the AMD64 machine.. ?

Sigh.. yet another time sink to play with.  ;)

Re: regarding conversion to new EH

[Mark Lord]

Mark Lord wrote:
>
> The pdc_adma driver works fine in all versions including 2.6.17 on my P4 
> box.
> I wonder why it's fussy on the AMD64 machine.. ?
> 
> Sigh.. yet another time sink to play with.  ;)

Mmmmmm.. something strange there.  It fails with both 64-bit and 32-bit kernels.
Its twin sister, the sata_qstor driver, works just fine.

Oh well.

Maybe it's a VIA chipset IRQ routing issue or something.

Cheers