From mboxrd@z Thu Jan  1 00:00:00 1970
From: Joel Diaz <joeldiaz@earthlink.net>
Date: Wed, 08 May 2013 17:41:30 -0400
Subject: [ath9k-devel] ath9k (AR9485) failing with PCI errors after hours of
 uptime with current ath9k.ko from linux-wireless
Message-ID: <1368049290.17080.31.camel@debian>
List-Id: <ath9k-devel.lists.ath9k.org>
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: 7bit
To: ath9k-devel@lists.ath9k.org

First some background: starting with Debian Wheezy I noticed my wireless
connection would fail after hours (sometime as quick as 15 minutes) of
uptime. Moving from Debian's 3.2 kernel to upstream's 3.9-RCs also
showed the problem.

Yesterday I pulled the linux-wireless git repo and built myself a kernel
(starting with Debian Wheezy's kernel config) with HEAD pointing to:

commit e514a9747148e3786879cc5430775a854441ba38
Author: Stanislaw Gruszka <sgruszka@redhat.com>
Date:   Thu May 2 09:43:57 2013 +0200

    ath5k: do not reschedule tx_complete_work on stop

After about 10 hours of uptime with the system basically idle (since I
was at work) I see the failure:

[38951.779076] ath: phy0: received PCI FATAL interrupt
[38951.779081] ath: phy0: received PCI PERR interrupt
[38951.789851] ath: phy0: Failed to wakeup in 500us
[38951.789853] ------------[ cut here ]------------
[38951.789866] WARNING: at drivers/net/wireless/ath/ath9k/hw.c:2231
ath9k_hw_setpower+0x446/0x499 [ath9k_hw]()
[38951.789868] Hardware name: Inspiron One 2020
[38951.789870] Modules linked in: isofs udf crc_itu_t bnep rfcomm
binfmt_misc loop hid_generic usbhid hid ath3k btusb bluetooth
snd_hda_codec_realtek coretemp ehci_pci kvm_intel snd_hda_intel kvm
snd_hda_codec ehci_hcd arc4 snd_hwdep ath9k ath9k_common ath9k_hw ath
mac80211 usbcore i915 cfg80211 snd_pcm drm_kms_helper drm iTCO_wdt
iTCO_vendor_support snd_page_alloc i2c_algo_bit i2c_i801 i2c_core
snd_timer acpi_cpufreq mperf crc32c_intel ghash_clmulni_intel lpc_ich
mfd_core sparse_keymap rfkill evdev snd video dcdbas usb_common psmouse
cryptd wmi processor button soundcore pcspkr serio_raw microcode ext4
crc16 jbd2 mbcache sg sr_mod sd_mod cdrom crc_t10dif ata_generic thermal
fan thermal_sys ata_piix libata scsi_mod r8169 mii
[38951.789929] Pid: 0, comm: swapper/0 Not tainted 3.9.0ath9-wl+ #3
[38951.789931] Call Trace:
[38951.789933]  <IRQ>  [<ffffffff8103d500>] ? warn_slowpath_common
+0x76/0x8c
[38951.789946]  [<ffffffffa044e166>] ? ath9k_hw_setpower+0x446/0x499
[ath9k_hw]
[38951.789954]  [<ffffffffa0395b11>] ? ath9k_ps_wakeup+0x4c/0xa9 [ath9k]
[38951.789960]  [<ffffffffa0397374>] ? ath9k_tasklet+0x24/0x131 [ath9k]
[38951.789964]  [<ffffffff81044000>] ? tasklet_action+0x73/0xc2
[38951.789968]  [<ffffffff81043c61>] ? __do_softirq+0xe2/0x1f7
[38951.789971]  [<ffffffff81043e41>] ? irq_exit+0x3f/0x82
[38951.789975]  [<ffffffff8102c301>] ? __x2apic_send_IPI_mask+0xb9/0x140
[38951.789979]  [<ffffffff8100f701>] ? do_IRQ+0x81/0x97
[38951.789984]  [<ffffffff813856ed>] ? common_interrupt+0x6d/0x6d
[38951.789985]  <EOI>  [<ffffffff8129f3d8>] ? arch_local_irq_enable
+0x4/0x8
[38951.789994]  [<ffffffff8129fa5c>] ? cpuidle_wrap_enter+0x3c/0x71
[38951.789999]  [<ffffffff8129f78e>] ? cpuidle_enter_state+0xa/0x2f
[38951.790002]  [<ffffffff8129f85c>] ? cpuidle_idle_call+0xa9/0xfb
[38951.790007]  [<ffffffff81014c6f>] ? cpu_idle+0x9c/0xe6
[38951.790011]  [<ffffffff816aed23>] ? start_kernel+0x3b8/0x3c3
[38951.790014]  [<ffffffff816ae781>] ? repair_env_string+0x57/0x57
[38951.790018]  [<ffffffff816ae59c>] ? x86_64_start_kernel+0xf2/0xfd
[38951.790021] ---[ end trace bff1151979a9309c ]---
[38951.800533] ath: phy0: Failed to wakeup in 500us
[38951.865086] ath: phy0: Failed to stop TX DMA, queues=0x10f!
[38951.876459] ath: phy0: DMA failed to stop in 10 ms AR_CR=0xffffffff
AR_DIAG_SW=0xffffffff DMADBG_7=0xffffffff
[38951.876494] ath: phy0: Could not stop RX, we could be confusing the
DMA engine when we start RX up

At this point the wireless connection is gone and can't be restored
without a full reboot.

In the past while looking at this with the upstream RC kernels I tried
setting ath9k.debug=0xffffffff, but it didn't seem to add anything
useful ( https://bugzilla.kernel.org/show_bug.cgi?id=56301 ).

Since the full kern.log is over 50 megs (lots of repeated error
messages) I'm only attaching the first 1000 lines which should be more
than enough.

Not sure where to go from here since I'm using the most current ath9k
that I could find. Is there anything else I could do to help get to the
bottom of this issue?

Joel






-------------- next part --------------
A non-text attachment was scrubbed...
Name: linux-wireless-edited-kern.log.gz
Type: application/x-gzip
Size: 17199 bytes
Desc: not available
Url : http://lists.ath9k.org/pipermail/ath9k-devel/attachments/20130508/8b90ce89/attachment.bin