Re: [git pull] x86 fixes

["Torsten Kaiser" <just.for.lkml@googlemail.com>]

On Sun, Jan 11, 2009 at 3:39 PM, Ingo Molnar <mingo@elte.hu> wrote:
>
> Linus,
>
> Please pull the latest x86-fixes-for-linus git tree from:
>
>   git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip.git x86-fixes-for-linus
>
> out-of-topic modifications in x86-fixes-for-linus:
> --------------------------------------------------
> include/asm-generic/pgtable.h      # e104ba3: x86 PAT: change track_pfn_vma_new
> mm/memory.c                        # e104ba3: x86 PAT: change track_pfn_vma_new
>                                   # e61304a: x86 PAT: remove PFNMAP type on tr
>
>  Thanks,
>
>        Ingo
>
> ------------------>
> Andi Kleen (2):
>      x86: hpet: allow force enable on ICH10 HPET
>      x86: avoid theoretical vmalloc fault loop
>
> Jaswinder Singh Rajput (1):
>      x86: fix mpparse.c build error on latest git
>
> Kyle McMartin (1):
>      x86, mtrr: fix types used in userspace exported header
>
> Suresh Siddha (1):
>      x86, pat: fix reserve_memtype() for legacy 1MB range
>
> venkatesh.pallipadi@intel.com (6):
>      x86 PAT: remove PFNMAP type on track_pfn_vma_new() error
>      x86 PAT: consolidate old memtype new memtype check into a function
>      x86 PAT: change track_pfn_vma_new to take pgprot_t pointer param
>      x86 PAT: return compatible mapping to remap_pfn_range callers
>      x86 PAT: ioremap_wc should take resource_size_t parameter
>      x86 PAT: remove CPA WARN_ON for zero pte

Something is (very) wrong with one(?) of these patches.

After upgrading from 2.6.28 to 2.6.29-rc1 I lost direct rendering.
Each time I tried to start a program that uses DRM I get this in the
syslog and the programm falls back to mesa software rendering:
Jan 11 13:32:31 treogen [   77.167977] X:3280 map pfn expected mapping
type uncached-min
us for e0000000-e7ff8000, got write-combining
Jan 11 13:32:31 treogen [   77.173620] X:3280 freeing invalid memtype
e0000000-e7ff8000
Jan 11 13:34:51 treogen [  217.861668] glxinfo:3492 map pfn expected
mapping type uncach
ed-minus for e0000000-e7ff8000, got write-combining
Jan 11 13:34:51 treogen [  217.867220] glxinfo:3492 freeing invalid
memtype e0000000-e7f
f8000
Jan 11 13:35:23 treogen [  249.771043] glxinfo:3494 map pfn expected
mapping type uncach
ed-minus for e0000000-e7ff8000, got write-combining
Jan 11 13:35:23 treogen [  249.776589] glxinfo:3494 freeing invalid
memtype e0000000-e7f
f8000

Otherwise 2.6.29-rc1 worked for me. Even booting with 'fastboot' did
not result in any problems, but it did cut the in-kernel-time down
from ~12 sec to ~6 sec.

Hoping to fix this memtype problem I applied the patch from the pull
request to 29-rc1 and rebooted. Now the system completely locks up
when X is trying to start.
Via serial console I got this Oops:
[   79.500149] BUG: unable to handle kernel NULL pointer dereference
at 0000000000000003
[   79.509240] IP: [<0000000000000003>] 0x3
[   79.510002] PGD 0
[   79.510002] Oops: 0010 [#1] SMP
[   79.510002] last sysfs file:
/sys/devices/pci0000:00/0000:00:0f.0/0000:01:00.0/enable
[   79.510002] CPU 0
[   79.510002] Modules linked in: w83792d tuner tea5767 tda8290
tuner_xc2028 xc5000 tda9887 tuner_simple tuner_types mt20xx tea5761
tvaudio msp3400 bttv ir_common v4l2_common videodev v4l1_compat
v4l2_compat_ioctl32 usbhid videobuf_dma_sg videobuf_core hid btcx_risc
tveeprom sg pata_amd
[   79.510002] Pid: 0, comm: swapper Not tainted 2.6.29-rc1 #2
[   79.510002] RIP: 0010:[<0000000000000003>]  [<0000000000000003>] 0x3
[   79.510002] RSP: 0018:ffffffff809a8b18  EFLAGS: 00010002
[   79.510002] RAX: 0000000000000001 RBX: ffffffff00000000 RCX: 0000000000000000
[   79.510002] RDX: 0000000000000001 RSI: 0000000000000000 RDI: ffffffff809a8ca8
[   79.510002] RBP: ffffffff809a8b18 R08: 0000000000000001 R09: 0000000000000100
[   79.510002] R10: ffffffff8026af40 R11: 00000000000068d8 R12: 0000000000000000
[   79.510002] R13: ffff88007e4fd700 R14: ffff880028018d00 R15: ffffffff809a8aa8
[   79.510002] FS:  00007ff217e406f0(0000) GS:ffffffff809b1040(0000)
knlGS:0000000000000000
[   79.510002] CS:  0010 DS: 0018 ES: 0018 CR0: 000000008005003b
[   79.510002] CR2: 0000000000000003 CR3: 0000000000201000 CR4: 00000000000006e0
[   79.510002] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[   79.510002] DR3: 0000000000000000 DR6: 00000000ffff4ff0 DR7: 0000000000000400
[   79.510002] Process swapper (pid: 0, threadinfo ffffffff8087e000,
task ffffffff807de360)
[   79.510002] Stack:
[   79.510002]  ffffffff809a8b68 ffffffff802389d7 0000000000000000
ffffffff809a8b60
[   79.510002]  0000000000000082 ffffffff8022a7a8 0000000000000000
0000000000000001
[   79.510002]  0000000000000060 ffffffff807de360 ffffffff809a8b78
ffffffff80238b7d
[   79.510002] Call Trace:
[   79.510002] Call Trace:
[   79.510002]  <IRQ> <0> [<ffffffff802389d7>] try_to_wake_up+0x137/0x2d0
[   79.510002]  [<ffffffff8022a7a8>] ? do_page_fault+0x368/0x970
[   79.510002]  [<ffffffff80238b7d>] default_wake_function+0xd/0x10
[   79.510002]  [<ffffffff8025a751>] autoremove_wake_function+0x11/0x40
[   79.510002]  [<ffffffff804cc70f>] ? ata_scsi_qc_complete+0x1df/0x4c0
[   79.510002]  [<ffffffff8065d1ef>] ? _spin_unlock_irqrestore+0x2f/0x40
[   79.510002]  [<ffffffff8026b02c>] ?
generic_smp_call_function_interrupt+0xec/0x100
[   79.510002]  [<ffffffff8065cddd>] ? trace_hardirqs_off_thunk+0x3a/0x6c
[   79.510002]  [<ffffffff8026af40>] ?
generic_smp_call_function_interrupt+0x0/0x100
[   79.510002]  [<ffffffff8026b02c>] ?
generic_smp_call_function_interrupt+0xec/0x100
[   79.510002]  [<ffffffff8065d54f>] ? page_fault+0x1f/0x30
[   79.510002]  [<ffffffff8026b02c>] ?
generic_smp_call_function_interrupt+0xec/0x100
[   79.510002]  [<ffffffff8026af40>] ?
generic_smp_call_function_interrupt+0x0/0x100
[   79.510002]  [<ffffffff8024402c>] ? warn_slowpath+0x4c/0x130
[   79.510002]  [<ffffffff804b8f85>] ? scsi_next_command+0x45/0x60
[   79.510002]  [<ffffffff804b9bd6>] ? scsi_io_completion+0x376/0x4e0
[   79.510002]  [<ffffffff804b2f6c>] ? scsi_finish_command+0xac/0xe0
[   79.510002]  [<ffffffff804b9e08>] ? scsi_softirq_done+0xb8/0x140
[   79.510002]  [<ffffffff8025d360>] ? __remove_hrtimer+0x40/0xa0
[   79.510002]  [<ffffffff8026b02c>] ?
generic_smp_call_function_interrupt+0xec/0x100
[   79.510002]  [<ffffffff8021e54f>] ? smp_call_function_interrupt+0x1f/0x30
[   79.510002]  [<ffffffff8020c863>] ? call_function_interrupt+0x13/0x20
[   79.510002]  <EOI> <0>Code:  Bad RIP value.
[   79.510002] RIP  [<0000000000000003>] 0x3
[   79.510002]  RSP <ffffffff809a8b18>
[   79.510002] CR2: 0000000000000003
[   79.510002] ---[ end trace 99e686e29f771a49 ]---
[   79.510002] Kernel panic - not syncing: Fatal exception in interrupt
[   79.510002] ------------[ cut here ]------------

last sysfs file: /sys/devices/pci0000:00/0000:00:0f.0/0000:01:00.0/enable

lspci -t:
-[0000:00]-+-00.0
           [snip]
           +-0f.0-[0000:01]--+-00.0
           |                 \-00.1
lspci:
00:0f.0 PCI bridge: nVidia Corporation MCP55 PCI Express bridge (rev a3)
01:00.0 VGA compatible controller: ATI Technologies Inc RV370 5B60
[Radeon X300 (PCIE)]
01:00.1 Display controller: ATI Technologies Inc RV370 [Radeon X300SE]

Userspace is xorg-server-1.5.3 with mesa-7.3_rc1 and xf86-video-ati-6.9.0.
With 2.6.28 this combination works for accelerated direct rendering.
PAT was enabled on 2.6.28 and both vanilla 2.6.29-rc1 and the patched -rc1.

Just ask, if you need more information, or if you have a patch to try.

Torsten