Am 26.05.2013 08:20, schrieb Oleksij Rempel:
Am 24.05.2013 10:47, schrieb Marc Kleine-Budde:
added ath9k-devel to Cc
On 05/23/2013 12:02 AM, Marc Kleine-Budde wrote:
Hello,
I'm on a kirkwood based armv5 system with an USB attached TP-Link
TL-WN821N - Atheros AR7010+AR9287, [1]. the wlan is running in AP mode
with hostapd-1.0. The kernel is v3.8.12 from debian (3.8-1-kirkwood #1
Debian 3.8.12-1).
The system crashes repeatedly after about one week with the following
oops:
[633625.401875] skbuff: skb_under_panic: text:bf501028 len:128 put:8
head:d2788800 data:d27887fe tail:0xd278887e end:0xd2788f40 dev:wlan1
[633625.414180] ------------[ cut here ]------------
[633625.418909] kernel BUG at
/build/buildd-linux_3.8.12-1-armel-7F6kBx/linux-3.8.12/net/core/skbuff.c:145!
[633625.428430] Internal error: Oops - BUG: 0 [#1] ARM
[633625.433322] Modules linked in:
[...]
[633625.583170] CPU: 0 Not tainted (3.8-1-kirkwood #1 Debian
3.8.12-1)
[633625.589821] PC is at skb_push+0x6c/0x84
[633625.593763] LR is at skb_push+0x6c/0x84
[633625.597707] pc : [<c0282990>] lr : [<c0282990>] psr: 20000013
[633625.597707] sp : c04c1d50 ip : 000008f8 fp : df04ea54
[633625.609404] r10: 00000002 r9 : 00000008 r8 : df00dca8
[633625.614734] r7 : 00000006 r6 : c04410a0 r5 : d278887e r4 :
d2788800
[633625.621378] r3 : c04d328c r2 : 20000093 r1 : 00000001 r0 :
00000079
[633625.628015] Flags: nzCv IRQs on FIQs on Mode SVC_32 ISA ARM
Segment kernel
[633625.635443] Control: 0005317f Table: 1f224000 DAC: 00000017
[633625.641295] Process swapper (pid: 0, stack limit = 0xc04c01b8)
[633625.647241] Stack: (0xc04c1d50 to 0xc04c2000)
[633625.657414] 1d40: 00000008
d2788800 d27887fe d278887e
[633625.666101] 1d60: d2788f40 df04e000 df00dc00 df2e0c00 00000078
bf501028 df2e0c00 dfba3120
[633625.675025] 1d80: d278882a df04e9a0 00000000 bf504110 dfb3ce20
00000201 00000000 00084502
[633625.683954] 1da0: 00000001 df2e0c00 dfba3120 00000008 00000002
c04c1df4 00000000 00000001
[633625.693553] 1dc0: 0000006a bf5058b0 00000000 c04c1df4 c04c1e30
dfba2300 c151ff18 df04e9a0
[633625.702041] 1de0: c04c1e30 bf37560c 0000000c 00004288 c04c1e2c
c151ff18 0000006a df2e0c00
[633625.710540] 1e00: dfba2300 00000000 0000006a df04e462 00000000
00000001 60000013 bf375760
[633625.718904] 1e20: 00000001 c14c19a0 c14c0460 00000000 c04c1e30
c04c1e30 00000000 dfba2300
[633625.727374] 1e40: df04e460 c151fc00 de5af200 00000002 00000002
dfba2300 dfba2308 dfba28a8
[633625.787263] 1e60: c04c1e7c dfba28ac df2e0c00 bf376d58 c0508ae0
00000000 0000012c 00000080
[633625.798914] 1e80: 03c66eab c0508ae8 c04d4c68 c04d3494 00000000
00000000 00000006 00000100
[633625.810249] 1ea0: c052b3a0 00000009 c052b3c0 c0026e2c 00000001
00000018 c04c0000 c0026644
[633625.818620] 1ec0: c04d8f74 c1484260 1144b25a c04d8f74 00000000
00200000 c04c1f4c 00000013
[633625.831230] 1ee0: 00000000 fed20200 c04c1f4c 00000000 56251311
c04d0420 00000000 c0026a2c
[633625.842695] 1f00: 00002000 c000f28c c004e27c c0271318 20000013
c000df94 c04c1f60 60000013
[633625.853824] 1f20: 000e32dc 0002404f b5def004 0002404f c04d0698
00000000 00000000 56251311
[633625.864745] 1f40: c04d0420 00000000 00000003 c04c1f60 c004e27c
c0271318 20000013 ffffffff
[633625.875714] 1f60: b5ed22e0 0002404f 0084d405 00000000 00000000
c04d0698 00000000 c04d0698
[633625.886646] 1f80: 00000000 c04d0420 004b8074 c0270e88 c04d0698
00000000 c050918c c0271014
[633625.898317] 1fa0: c04c0000 c0509b28 c04cc1cc c096f0e0 00004000
c000f484 c04c8c20 00000000
[633625.909787] 1fc0: c04b9650 c0498764 ffffffff ffffffff c0498284
00000000 00000000 c04b9650
[633625.918159] 1fe0: 00000000 00053175 c04c8048 c04b964c c04cc1c4
00008040 00000000 00000000
[633625.926557] [<c0282990>] (skb_push+0x6c/0x84) from [<bf501028>]
(htc_issue_send.constprop.0+0x28/0x68 [ath9k_htc])
[633625.937158] [<bf501028>] (htc_issue_send.constprop.0+0x28/0x68
[ath9k_htc]) from [<bf504110>] (ath9k_htc_tx_start+0x290/0x2a4
[ath9k_htc])
[633625.949877] [<bf504110>] (ath9k_htc_tx_start+0x290/0x2a4
[ath9k_htc]) from [<bf5058b0>] (ath9k_htc_tx+0x98/0xcc [ath9k_htc])
[633625.961458] [<bf5058b0>] (ath9k_htc_tx+0x98/0xcc [ath9k_htc])
from [<bf37560c>] (__ieee80211_tx+0x210/0x2a8 [mac80211])
[633625.972695] [<bf37560c>] (__ieee80211_tx+0x210/0x2a8 [mac80211])
from [<bf375760>] (ieee80211_tx+0xbc/0xc4 [mac80211])
[633625.983816] [<bf375760>] (ieee80211_tx+0xbc/0xc4 [mac80211]) from
[<bf376d58>] (ieee80211_tx_pending+0xf0/0x194 [mac80211])
[633625.995326] [<bf376d58>] (ieee80211_tx_pending+0xf0/0x194
[mac80211]) from [<c0026e2c>] (tasklet_action+0x84/0xcc)
[633626.005905] [<c0026e2c>] (tasklet_action+0x84/0xcc) from
[<c0026644>] (__do_softirq+0xdc/0x204)
[633626.014750] [<c0026644>] (__do_softirq+0xdc/0x204) from
[<c0026a2c>] (irq_exit+0x40/0x8c)
[633626.023103] [<c0026a2c>] (irq_exit+0x40/0x8c) from [<c000f28c>]
(handle_IRQ+0x64/0x84)
[633626.031193] [<c000f28c>] (handle_IRQ+0x64/0x84) from [<c000df94>]
(__irq_svc+0x34/0x78)
[633626.039412] [<c000df94>] (__irq_svc+0x34/0x78) from [<c0271318>]
(cpuidle_wrap_enter+0x54/0x9c)
[633626.048331] [<c0271318>] (cpuidle_wrap_enter+0x54/0x9c) from
[<c0270e88>] (cpuidle_enter_state+0x14/0x68)
[633626.058162] [<c0270e88>] (cpuidle_enter_state+0x14/0x68) from
[<c0271014>] (cpuidle_idle_call+0x138/0x25c)
[633626.067998] [<c0271014>] (cpuidle_idle_call+0x138/0x25c) from
[<c000f484>] (cpu_idle+0x68/0xc8)
[633626.076852] [<c000f484>] (cpu_idle+0x68/0xc8) from [<c0498764>]
(start_kernel+0x2b4/0x30c)
[633626.146230] Code: e58dc014 e59f1014 e59f0014 eb0308b0 (e7f001f2)
[633626.152520] ---[ end trace ee5dbceea3381e46 ]---
[633626.157249] Kernel panic - not syncing: Fatal exception in interrupt
Has the problem been fixed already? I can update the kernel to a recent
version if needed.
this oops was generated by skb_push:
" skb_push() will decrement the 'skb->data' pointer by the specified
number of bytes. It will also increment 'skb->len' by that number of
bytes as well. The caller must make sure there is enough head room for
the push being performed. This condition is checked for by skb_push()
and an assertion failure will trigger if this rule is violated."
hmm... theoretically driver should check the size of date before
skb_push, but i do not see that other driver do this check. Interesting
where this buffer was allocated.
In attachment is a patch. I hope it is proper fix. "Elders of the
Internet" your comments :)
--
Regards,
Oleksij
diff --git a/drivers/net/wireless/ath/ath9k/htc_hst.c b/drivers/net/wireless/ath/ath9k/htc_hst.c
index aac4a40..2901351 100644
--- a/drivers/net/wireless/ath/ath9k/htc_hst.c
+++ b/drivers/net/wireless/ath/ath9k/htc_hst.c
@@ -26,6 +26,12 @@ static int htc_issue_send(struct htc_target *target, struct sk_buff* skb,
struct htc_endpoint *endpoint = &target->endpoint[epid];
int status;
+ if (skb_headroom(skb) < len &&
+ pskb_expand_head(skb, len, 0, GFP_ATOMIC)) {
+ dev_err(target->dev, "Unable to expand headrom to %d\n", len);
+ return -ENOMEM;
+ }
+
hdr = (struct htc_frame_hdr *)
skb_push(skb, sizeof(struct htc_frame_hdr));
hdr->endpoint_id = epid;
