Vic^H^Holunteers needed for updated SATA code testing

[Pekka Pietikäinen]

Hiya

Due to popular demand on #fedora-devel, I merged the latest libata code to
the FC1 kernel. This is basically identical to 2.4.22-bk53-libata1.patch.gz,
except for two PCI id's added to sata_promise (0x3376 and 0x3378) and
&current->sigmask_lock -> &current->sighand->siglock in
libata-core.c.

Pre-built RPMS + the patches I used (linux-2.4.22-libata.patch
and linux-2.4.22-intel-esb-drivers.patch) can be found from
http://www.ee.oulu.fi/~pp/fc1-libata (sorry, no yum support :-) )

Please test it out and if it works like it should, I'll file a RFE in
bugzilla to get this merged. 

Open questions:

- Did I break anything? (the amount of changes in header files I merged from
  2.4.22-bk to fc1 like scsi.h were a bit worrying...) Promise 376 + athlon
  works ok, and it did compile, so I decided to ship it :-)
- Is the Promise 378 PCI id 0x3378 and does the code work on one?
- driver disk or boot.iso that would let people install FC1 on SATA-only
  boxes

-- 
Pekka Pietikainen

[Pekka Pietikäinen]

On Sun, Nov 16, 2003 at 08:37:56PM +0200, Pekka Pietikainen wrote:
>  - Did I break anything? (the amount of changes in header files I
> merged from
>    2.4.22-bk to fc1 like scsi.h were a bit worrying...) Promise 376 + athlon
>    works ok, and it did compile, so I decided to ship it :-)
>  - Is the Promise 378 PCI id 0x3378 and does the code work on one?
>  - driver disk or boot.iso that would let people install FC1 on SATA-only
>    boxes 
Positive comments so far from Promise and VIA users, so I reused someones
RFE
in bugzilla for this (#103980) to possibly get it in a future kernel
update.

The existance of boards with a 0x3378 PCI id is still a mystery, supposedly
they're on some Asus boards, but I still have no conclusive evidence either
way. Anyone? 

I've also not heard of anyone managing a working install image, if anyone
got it running, would they mind sharing? ;-)

-- 
Pekka Pietikainen

[Justin M. Forbes]

>  Positive comments so far from Promise and VIA users, so I reused
> someones RFE
>  in bugzilla for this (#103980) to possibly get it in a future kernel
>  update. 
I have merged these patches into the current Fedora kernel for AMD64,
hopefully it will stay there until release, in which case, it should ship
with an errata kernel for i386 users as well. 

>  
>  I've also not heard of anyone managing a working install image, if anyone
>  got it running, would they mind sharing? ;-) 
It is in the install image for x86_64, when I get some spare cycles I will
do an install image for i?86 with this kernel.

Justin M. Forbes

[Samuel Flory]

Justin M. Forbes wrote:
> >Positive comments so far from Promise and VIA users, so I reused
> someones RFE
> >in bugzilla for this (#103980) to possibly get it in a future kernel
> >update.
>  
>  
>  I have merged these patches into the current Fedora kernel for AMD64,
>  hopefully it will stay there until release, in which case, it should ship
>  with an errata kernel for i386 users as well. 
>  
>  
> >I've also not heard of anyone managing a working install image, if anyone
> >got it running, would they mind sharing? ;-)
>  
>  
>  It is in the install image for x86_64, when I get some spare cycles I will
>  do an install image for i?86 with this kernel.
>   
   Which sata chipsets are we looking for?  I've got a wide range of SI, 
ICH5, Promise, Highpoint, and God knows what else.  Other than a few 
wierd software raid SATA controllers, and one ebedded promise 
controllers it all works with the libata patches.

-- 
There is no such thing as obsolete hardware.
Merely hardware that other people don't want.
(The Second Rule of Hardware Acquisition)
Sam Flory  <sflory(a)rackable.com&gt;

[Pekka Pietikäinen]

On Mon, Dec 01, 2003 at 06:23:46PM -0800, Samuel Flory wrote:
>    Which sata chipsets are we looking for?  I've got a wide range
> of SI, 
>  ICH5, Promise, Highpoint, and God knows what else.  Other than a few 
>  wierd software raid SATA controllers, and one ebedded promise 
>  controllers it all works with the libata patches. 
If so even with the fc1 kernel
with the updated patch, excellent! 

Actually I just put up a newer version of the patch (with the fixes in
2.4.23-libata1.patch merged). Also nuked the 0x3378 ID since the promise
driver has 378 as 0x3373, which was already there, and nobody has said they
needed it. Apparently there was some confusion when I asked someone on fedora
whether their 378's ID was 0x3378 and they answered "yes", which is why I
added it in the first place.

Has gone through "it boots and detects the drive" testing, apparently
selinux security labels screw up 2.4 kernels (not just this one) quite
completely so that's all the testing I could do. Anyone know if there's an
easy way to remove them apart from backup, mke2fs and restore? 

Same place as before, http://www.ee.oulu.fi/~pp/fc1-libata. Only athlon
kernel+src.rpm+patch up, i686 coming up soon. I don't think I'll do a full
set this time, it takes pretty long :/

-- 
Pekka Pietikainen