Cjg prepaired a fresh new EFIKA 2.6.24.x kernel patchset:

http://www.ewb.it/~acrux/linux-2.6.24.2-Efika.patch

_________________
acrux _at_ linuxmail _dot_ org

Can you get rid of the BIG FAT EFIKA HACK from the audio? Everyone should be running the firmware update which includes this support already (20070122, it's linked in Platform Support!)

I'd also very much appreciate if we standardised on efika.forth for new kernels. I am sure you can work out a nice way to integrate it into the Crux system.. it's due for a small update in the next week (needed a lot of testing..) which will add more features to the device tree and allow configuration of the PSC6 on boot, plus the magic of the CAN/I2C bus on the underside of the board..

_________________
Matt Sealey

our fault... you're right, i totally agree with you.
yes, it sounds good.

btw, have you tried CRUX PPC ?

_________________
acrux _at_ linuxmail _dot_ org

Does the sound part of the driver need the rest of the patch? I tried to use the sound driver on openSUSE factory, but it did not work. Modprobe segfaulted, and I got an ugly oops in dmesg, no matter if I use the devtree fix from powerdeveloper.org or not.

efika:~ # uname -a
Linux efika 2.6.24.1-6-default #1 2008/02/22 15:49:32 UTC ppc ppc ppc GNU/Linux


Sound: MPC52xx PSC AC97 driver
Unable to handle kernel paging request for data at address 0xca197010
Faulting instruction address: 0xc02c0890
Oops: Kernel access of bad area, sig: 11 [#1]
Efika
Modules linked in: snd_mpc52xx_ac97 snd_ac97_codec snd_pcm snd_timer snd soundcore snd_page_alloc ac97_bus bestcomm_gen_bd iptable_filter ip_tables ip6_tables x_tables uinput binfmt_misc fuse loop dm_mod usb_storage fec_mpc52xx bestcomm_fec fec_mpc52xx_phy libphy sg sd_mod pata_mpc52xx libata scsi_mod
NIP: c02c0890 LR: c02c0880 CTR: c02c0aa8
REGS: c6c4bcf0 TRAP: 0300 Tainted: G N (2.6.24.1-6-default)
MSR: 00009032 <EE,ME,IR,DR> CR: 22022242 XER: 20000000
DAR: ca197010, DSISR: 20000000
TASK = c6c6eb20[2338] 'modprobe' THREAD: c6c4a000
GPR00: c02c0880 c6c4bda0 c6c6eb20 00000000 ca196f8c c2cad618 c024afe8 00000000
GPR08: c2caee55 c04a764c 00000001 000000bb 5a74203c 1002e304 00000000 c6eed6a0
GPR16: ca1962a8 ca2656c0 00000027 c005a25c 00000000 00000124 ca264efe 00000029
GPR24: ca249000 0000002a c6ec5274 ca265080 ca196090 c7ff95f0 00000000 ca197010
NIP [c02c0890] of_match_node+0xa0/0xe0
LR [c02c0880] of_match_node+0x90/0xe0
Call Trace:
[c6c4bda0] [c02c0880] of_match_node+0x90/0xe0 (unreliable)
[c6c4bdc0] [c02c0ad4] of_platform_bus_match+0x2c/0x48
[c6c4bdd0] [c024ae88] driver_probe_device+0x4c/0x1ac
[c6c4bdf0] [c024b070] __driver_attach+0x88/0xec
[c6c4be10] [c024a694] bus_for_each_dev+0x54/0xa0
[c6c4be40] [c024ad2c] driver_attach+0x24/0x34
[c6c4be50] [c0249e70] bus_add_driver+0x78/0x1d8
[c6c4be70] [c024b2bc] driver_register+0x88/0x9c
[c6c4be80] [c000b0cc] of_register_platform_driver+0x48/0x58
[c6c4be90] [c9022028] mpc52xx_ac97_init+0x28/0xa8 [snd_mpc52xx_ac97]
[c6c4bea0] [c005d08c] sys_init_module+0x156c/0x1640
[c6c4bf40] [c0013470] ret_from_syscall+0x0/0x40
--- Exception: c01 at 0xff674bc
LR = 0x10004350
Instruction dump:
5463d97e 7fde1838 881f0040 2f800000 41be0014 7fa3eb78 389f0040 4bfffbcd
7fde1838 2f9e0000 409e0030 3bff00c4 <881f0000> 2f800000 40beff74 881f0020
---[ end trace ece886062d03a974 ]---

_________________
CzP
http://czanik.blogs.balabit.com/

ehm which part of the patch have you used?

Only the mpc52xx_ac97.c file, as the openSUSE kernel source includes an earlier version of the driver (which does not compile).
CzP

_________________
CzP
http://czanik.blogs.balabit.com/

can you test applying the full patch?

I applied it now, but the results are the same :-(

_________________
CzP
http://czanik.blogs.balabit.com/

<bump>

_________________
http://bbrv.blogspot.com

mmh.. then your kernel isn't vanilla. That's right?

_________________
acrux _at_ linuxmail _dot_ org

No, it isn't vanilla. But linuxppc-devel is full of similar e-mails for different subsystems on different mpc5200 machines, so I hope, it will be solved soon :)

_________________
CzP
http://czanik.blogs.balabit.com/

ok, it'll never work if there'snt a vanilla kernel.

_________________
acrux _at_ linuxmail _dot_ org

Well, the previous version of the sound patch worked for me on both SuSE and Ubuntu kernels, both of the heavily modified.

BTW: openSUSE factory is now upgraded to use the 2.6.25-RC kernels. I tried to compile this driver with it, but it did not even compile. Do you plan to bring up your patches to 2.6.25 level?

_________________
CzP
http://czanik.blogs.balabit.com/

Hi guys,

After struggling with the audio driver some more, some significant regressions and certain parts of this patch are completely braindead and conceptually farcical.

I have no idea where you got lines such as:
It is not acceptable to restore pre-arch/powerpc code to handle this. If you want the cdm or gpio mappings to work, for crying out loud use efika.forth or hack prom_init.c fixups at the worst case. Even change or comment out things. But do not remove code like this!

Linux is already far, far too difficult to cross-reference patches and mainline code properly, and this patch just makes the debugging process more difficult (it also means that certain parts of efika.forth fixups are NEVER tested because you removed that code path).

Please, someone, revisit this patch for the latest 2.6.25 release candidate, if you need any help or advise, please contact someone like Peter or myself..

_________________
Matt Sealey

thanks for your review.
It was only a quick assembled patchset to realize a boot kernel for our latest (march 08) test iso.

We hope to do a better one in september. No free time at the moment, apologize for that.

_________________
acrux _at_ linuxmail _dot_ org