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 ¤t->sigmask_lock -> ¤t->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
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? ;-)
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
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.
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 :/
-
Justin M. Forbes -
Pekka Pietikäinen -
Samuel Flory