On Mon, 09 Jul 2007 00:35:28 -0700 (PDT), David Miller wrote: >From: Dmitry Artamonow <mad_soft@xxxxxxxx> >Date: Sun, 8 Jul 2007 22:05:08 +0400 > >> On 23:02 Fri 06 Jul , David Miller wrote: >> > >> > > Seems that your patch was not the only related change between -rc4 and -rc5. >> > > Anyway, X still doesn't work with -rc5 so probably problem is somewhere >> > > else. I sent you privately strace logs for X with 2.6.22-rc5. >> > >> > Some more progress, there seems to be code in the X server that >> > hard-codes the I/O and MEM space sizes based upon the PCI >> > IDs of the Sun PCI host controller. Barf... >> > >> > Does it make further progress with the following patch? >> > >> Well, no. I applied it against -rc7 and Xorg still doesn't work (as far as >> I can tell by its console output - until tomorrow I have only remote access >> to my U10). Sent you another portion of logs. >> Just in case, here's results from your testcase program: >> >> vanilla 2.6.22-rc7: >> x0:0x8e100080 >> x8:0x00000006 >> >> 2.6.22-rc7 + patch: >> x0:0x8e100020 >> x8:0x00000000 > >Something is very very wrong. That's the last thing in the world my >patch should cause to happen. rc4 to rc5 also contained no changes >that should have made that problem go away, so something else >is afoot. > >I have X fully working on my ultra5 with a one-line patch to >Ubuntu-feisty's xserver source tree which should be entirely >irrelevant to your cases. > >I've tried everything to reproduce that test program failure on >any of my systems and I can't, so could you do some detective >work to try and see if maybe it's PREEMPT or a miscompile or >something like that? > >I'm out of ideas, so your help in doing the debugging to figure >out where this comes from since I can't make it happen here is >appreciated :-) (I've been offline lately, SPARC-wise, hence the late reply.) Kernel 2.6.22 final: - X won't start with similar errors as before in 2.6.22-rc - your PCI test program gives the correct outputs: x0:0x8e100080 x8:0x00000006 - adding your patch to skip pci_host_bridge_write_pci_cfg() in pci_common.c broke the test program, but X still won't start as before So far I've compiled the kernel with gcc-4.2.0. Recompiling 2.6.22 vanilla with gcc-3.4.6 makes no difference: X doesn't work. I can post the full .config if needed, but it has nothing exotic in it AFAIK. Even PREEMPT is disabled (and always has been). /Mikael - To unsubscribe from this list: send the line "unsubscribe sparclinux" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
