WARNING in vmap_page_range on alpha since 2.6.28

[Tobias Klausmann <klausman@schwarzvogel.de>]

Hi! 

I've stumbled across this WARNING: when booting 2.6.28 + patch
on my alpha. This happens consistently (i.e. on every boot) and
keeps happening while the machine is booted:

------------[ cut here ]------------
WARNING: at mm/vmalloc.c:104 vmap_page_range+0x1b8/0x320()
Modules linked in:
fffffc007f03b968 0000000000000003 fffffc0000377e48 fffffffc00002000 
       fffffffc00008000 0000000000000003 6761705f70616d76 2b65676e61725f65 
       78302f3862317830 0000000000303233 fffffc00006f0828 0000000000000000 
       fffffc00006efcd8 fffffc00006efcd8 fffffc00006f0830 0000000000000000 
       0000000000000000 fffffc00006efd18 0000000000000001 00000000000200d2 
       0000000000000000 0000000000000000 0000000000000001 0000000000000044 
Trace:
[<fffffc0000377e48>] vmap_page_range+0x1b8/0x320
[<fffffc0000379288>] __vmalloc_area_node+0xb8/0x1b0
[<fffffc0000377fe4>] map_vm_area+0x34/0x60
[<fffffc0000379320>] __vmalloc_area_node+0x150/0x1b0
[<fffffc00003792cc>] __vmalloc_area_node+0xfc/0x1b0
[<fffffc00004da5d4>] agp_add_bridge+0x1a4/0x610
[<fffffc00005d2858>] agp_amdk7_probe+0x17c/0x2ec
[<fffffc000049627c>] pci_device_probe+0x8c/0xc0
[<fffffc00004ff1b8>] driver_probe_device+0xa8/0x230
[<fffffc00004ff41c>] __driver_attach+0xdc/0xe0
[<fffffc00004fe698>] bus_for_each_dev+0x78/0xd0
[<fffffc00004ff340>] __driver_attach+0x0/0xe0
[<fffffc00004fef5c>] driver_attach+0x2c/0x40
[<fffffc00004fdccc>] bus_add_driver+0x28c/0x340
[<fffffc00004ff6e4>] driver_register+0x94/0x1d0
[<fffffc00004965e4>] __pci_register_driver+0x64/0xf0
[<fffffc0000310084>] do_one_initcall+0x34/0x1e0
[<fffffc00003db60c>] proc_register+0x6c/0x280
[<fffffc00003db5f0>] proc_register+0x50/0x280
[<fffffc00003db9bc>] create_proc_entry+0x7c/0x100
[<fffffc000035597c>] register_irq_proc+0xbc/0xe0
[<fffffc0000355a14>] init_irq_proc+0x74/0xa0
[<fffffc0000311158>] kernel_thread+0x28/0x90
[<fffffc0000310d84>] entSys+0xa4/0xc0

---[ end trace 9863e03dd539368c ]---

Here's a full dmesg:
http://eric.schwarzvogel.de/~klausman/kernel/2.6.28/alpha/dmesg_post_boot.txt

Note that it happens again later in the same dmesg. The messages
always mention page allocation at the top of the trace.

The patch mentioned above is commit
1684f5ddd4c0c754f52c78eaa2c5c69ad09fb18c which fixes compilation
on alpha (see http://bugzilla.kernel.org/show_bug.cgi?id=12289)
The patch I used is here:
http://eric.schwarzvogel.de/~klausman/kernel/2.6.28/alpha/pci.patch

Here's my config:
http://eric.schwarzvogel.de/~klausman/kernel/2.6.28/alpha/config.txt
shortened by grep ^C:
http://eric.schwarzvogel.de/~klausman/kernel/2.6.28/alpha/config-shortened.txt

The normal console output (catched via serial console) is here:
http://eric.schwarzvogel.de/~klausman/kernel/2.6.28/alpha/console_output.txt

The machine in question is a Samsung UP1500. CPU is EV68AL on an
ALI chipset (ali15x3). 

Last kernel to work perfectly is a vanilla 2.6.27.10.

Before I dig into a rather time-consuming bisect, I'd like to
hear informed opinions :)

Regards & TIA,
Tobias

-- 
This sentence no verb.  This sentence short.  This signature done.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>