Ticket #2133 (closed defect: fixed)

Opened 8 years ago

Last modified 8 years ago

kmmcd: page allocation failure. order:4, mode:0xd0

Reported by: lindi Owned by: openmoko-kernel
Priority: normal Milestone:
Component: kernel Version:
Severity: minor Keywords:
Cc: Blocked By:
Blocking: Estimated Completion (week):
HasPatchForReview: no PatchReviewResult:


stable-tracking kernel at 1b17c67794364bee printed the above error
message today. I don't unfortunately know what I was exactly doing at
that time so I'm also attaching complete dmesg output from boot in
addition to this small extract:

[35769.010000] kmmcd: page allocation failure. order:4, mode:0xd0
[35769.010000] [<c0365370>] (dump_stack+0x0/0x14) from [<c0091158>] (__alloc_pages_internal+0x3c0/0x3e8)
[35769.010000] [<c0090d98>] (__alloc_pages_internal+0x0/0x3e8) from [<c00b05f8>] (cache_grow+0xc8/0x354)
[35769.010000] [<c00b0530>] (cache_grow+0x0/0x354) from [<c00b0a84>] (cache_alloc_refill+0x200/0x264)
[35769.010000] [<c00b0884>] (cache_alloc_refill+0x0/0x264) from [<c00b0b88>] (__kmalloc+0xa0/0xdc)
[35769.010000] [<c00b0ae8>] (__kmalloc+0x0/0xdc) from [<c02dadc4>] (alloc_netdev_mq+0x50/0x164)
[35769.010000]  r7:00000001 r6:00000000 r5:c79fc768 r4:c79fc000
[35769.010000] [<c02dad74>] (alloc_netdev_mq+0x0/0x164) from [<c02e8240>] (alloc_etherdev_mq+0x1c/0x28)
[35769.010000] [<c02e8224>] (alloc_etherdev_mq+0x0/0x28) from [<c027852c>] (ar6000_avail_ev+0x30/0x3c4)
[35769.010000] [<c02784fc>] (ar6000_avail_ev+0x0/0x3c4) from [<c0273a40>] (HTCTargetInsertedHandler+0x17c/0x1c4)
[35769.010000]  r8:00000000 r7:c79fcdb8 r6:00000006 r5:c79fc768 r4:c79fc000
[35769.010000] [<c02738c4>] (HTCTargetInsertedHandler+0x0/0x1c4) from [<c02743dc>] (sdio_ar6000_probe+0x1ac/0x27c)
[35769.010000] [<c0274230>] (sdio_ar6000_probe+0x0/0x27c) from [<c026a1dc>] (sdio_bus_probe+0xd4/0xf0)
[35769.010000]  r8:c7a2a808 r7:c0493eb4 r6:c7a2a400 r5:c0386808 r4:00000000
[35769.010000] [<c026a108>] (sdio_bus_probe+0x0/0xf0) from [<c01d3128>] (really_probe+0x120/0x28c)
[35769.010000]  r7:c0493eb4 r6:c7a2a408 r5:c0493eb4 r4:00000000
[35769.010000] [<c01d3008>] (really_probe+0x0/0x28c) from [<c01d336c>] (driver_probe_device+0xd8/0xf4)
[35769.010000]  r7:c7865e30 r6:c7a2a408 r5:c0493eb4 r4:00000000
[35769.010000] [<c01d3294>] (driver_probe_device+0x0/0xf4) from [<c01d3420>] (__device_attach+0x10/0x14)
[35769.010000]  r6:c7a2a408 r5:c01d3410 r4:00000000
[35769.010000] [<c01d3410>] (__device_attach+0x0/0x14) from [<c01d2468>] (bus_for_each_drv+0x5c/0x9c)
[35769.010000] [<c01d240c>] (bus_for_each_drv+0x0/0x9c) from [<c01d34a4>] (device_attach+0x58/0x70)
[35769.010000]  r7:c7a2a4f0 r6:00000000 r5:c7a2a534 r4:c7a2a408
[35769.010000] [<c01d344c>] (device_attach+0x0/0x70) from [<c01d2190>] (bus_attach_device+0x30/0x64)
[35769.010000]  r5:c7a2a408 r4:c0493b2c
[35769.010000] [<c01d2160>] (bus_attach_device+0x0/0x64) from [<c01d03c4>] (device_add+0x214/0x370)
[35769.010000]  r5:c7a2a408 r4:00000000
[35769.010000] [<c01d01b0>] (device_add+0x0/0x370) from [<c0269fc4>] (sdio_add_func+0x34/0x4c)
[35769.010000]  r8:00000001 r7:00000002 r6:c7a02800 r5:c7a2a400 r4:c7a2a408
[35769.010000] [<c0269f90>] (sdio_add_func+0x0/0x4c) from [<c02699bc>] (mmc_attach_sdio+0x2f8/0x3a8)
[35769.010000]  r5:00000000 r4:00000000
[35769.010000] [<c02696c4>] (mmc_attach_sdio+0x0/0x3a8) from [<c0265dac>] (mmc_rescan+0x100/0x210)
[35769.010000]  r8:00000000 r7:c7a02aa8 r6:c7a02aa8 r5:c7865f1c r4:c7a02800
[35769.010000] [<c0265cac>] (mmc_rescan+0x0/0x210) from [<c006895c>] (run_workqueue+0x128/0x224)
[35769.010000]  r6:c7864000 r5:c7870f20 r4:c7a02aac
[35769.010000] [<c0068834>] (run_workqueue+0x0/0x224) from [<c0068e50>] (worker_thread+0x100/0x114)
[35769.010000] [<c0068d50>] (worker_thread+0x0/0x114) from [<c006cd4c>] (kthread+0x60/0x94)
[35769.010000]  r8:00000000 r7:00000000 r6:c0068d50 r5:c7870f20 r4:c7864000
[35769.010000] [<c006ccec>] (kthread+0x0/0x94) from [<c005ac64>] (do_exit+0x0/0x2a4)
[35769.010000]  r6:00000000 r5:00000000 r4:00000000


dmesg1.txt (53.5 KB) - added by lindi 8 years ago.

Change History

Changed 8 years ago by lindi

comment:1 Changed 8 years ago by andy

Hum the dmesg doesn't look that happy for resume :-(( first the Glamo is broken on one resume (is this by any chance a WSOD?) and can't talk to the SD Card, then there's what I guess is out of memory condition on SDIO channel to AR6001 on resume reprobe.

comment:2 Changed 8 years ago by werner

The out-of-memory is a 64kB GFP_KERNEL allocation when bringing up the
network device. This being a 64kB allocation is just bad programming.
(The data structure contains large arrays of packet buffers. Argh.)

The general memory situation doesn't look too bad, so I'm a bit
surprised that the kernel couldn't find the pages, but trying to allocate
64kB of contiguous kernel memory is tempting fate. Not sure if the Glamo
issues contributed anything to making this worse.

comment:3 Changed 8 years ago by lindi

I have never seen WSOD. I noticed the kernel error messages only later when I was reading dmesg output.

comment:4 Changed 8 years ago by werner

ar6k-break-down-insane-allocation.patch in
probably solves this allocation problem.

comment:5 Changed 8 years ago by lindi

Thanks, should I test ar6k-break-down-insane-allocation.patch separately or wait for it to enter stable-tracking?

comment:6 Changed 8 years ago by werner

If you could reproduce the old problem easily, then it would be nice if you
could test the patch. If that was just a once-in-a-lifetime occurrence,
then I think we should just put it into *-tracking and be done with it.

In the unlikely event that anyone has code that uses Atheros' raw interface,
it would of course be very interesting to see if I didn't break it.

comment:7 Changed 8 years ago by lindi

I do not know how to reproduce this but I'll report here if I can figure it out.

comment:8 Changed 8 years ago by lindi

I have now been using stable-tracking and andy-tracking kernels that include ar6k-break-down-insane-allocation.patch without hitting this problem anymore. I suggest that the bug report can be closed. It's easy to reopen it if somebody still hits the issue.

comment:9 Changed 8 years ago by andy

  • Status changed from new to closed
  • Resolution set to fixed

Great, thanks again for the confirmation.

Note: See TracTickets for help on using tickets.