On Wed, 2021-03-17 at 09:57 +0100, Lorenzo Bianconi wrote: > mt7615d users report the following crash running both phy concurrently: > > [ 147.273909] CPU 1 Unable to handle kernel paging request at virtual address 00000000, epc == 8e4f288c, ra == 8e58a968 > [ 147.284570] Oops[#1]: > [ 147.286854] CPU: 1 PID: 3724 Comm: kworker/u9:3 Not tainted 5.4.105 #0 > [ 147.293403] Workqueue: napi_workq napi_workfn > [ 147.297742] $ 0 : 00000000 00000001 8e595ec0 00000000 > [ 147.302950] $ 4 : 8e595ec0 00000000 8e5ca600 00000000 > [ 147.308161] $ 8 : 00000000 000004c1 80808080 fefefeff > [ 147.313370] $12 : 00000000 00000000 80704f1c 00000040 > [ 147.323781] $20 : 00000050 00000044 00000011 01ce9502 > [ 147.334195] $28 : 8e004000 8e005d18 8ddf2020 8e58a968 > [ 147.342266] Lo : 00000000 > [ 147.350397] ra : 8e58a968 mt7615_queue_rx_skb+0x94c/0xd0c [mt7615_common] > [ 147.361580] Cause : 4080000c (ExcCode 03) > [ 147.368434] PrId : 0001992f (MIPS 1004Kc) > [ 147.456965] Process kworker/u9:3 (pid: 3724, threadinfo=9ece68d6, task=b16d155f, tls=00000000) > [ 147.473857] 0001ce95 00004188 00000000 00000080 00004c10 00000002 8074a2b8 80043dcc > [ 147.490505] 00000000 8e4f0110 8069ddf4 806ab148 8dfc0000 00001000 00000000 00000050 > [ 147.507150] ... > [ 147.512033] [<8e4f288c>] mt76_rx+0x124/0x320 [mt76] > [ 147.526633] > [ 147.528512] ---[ end trace 3d137e75bb109149 ]--- > > The issue has been introduce enabling hw rx de-amsdu since the hw can > interleave amsdu frames from both phy. Fix the issue moving the de-amsdu > buffer per-phy. > > Signed-off-by: Lorenzo Bianconi <lorenzo@xxxxxxxxxx> Acked-by: Ryder Lee <ryder.lee@xxxxxxxxxxxx>
