FS#1333 - MTK Ethernet Issues After NAPI polling Changes #6327
Comments
|
airend: [ 708.250919] ------------[ cut here ]------------ |
|
nwf: I am seeing identical dmesg logs (and no ethernet traffic) on a TEW714:
[ 36.138722] ------------[ cut here ]------------
[ 36.148024] WARNING: CPU: 0 PID: 0 at net/sched/sch_generic.c:316 dev_watchdog+0x1dc/0x25c
[ 36.164570] NETDEV WATCHDOG: eth0 (mtk_soc_eth): transmit queue 0 timed out
[ 36.178480] Modules linked in: rt2800soc rt2800mmio rt2800lib rt2x00soc rt2x00mmio rt2x00lib mac80211 cfg80211 wireguard crc_itu_t crc_ccitt compat ip6_udp_tunnel udp_tunnel eeprom_93cx6 usb_storage leds_gpio sd_mod scsi_mod gpio_button_hotplug ext4 jbd2 mbcache usbcore nls_base usb_common crc16 crc32c_generic crypto_hash
[ 36.235653] CPU: 0 PID: 0 Comm: swapper Not tainted 4.9.77 #0
[ 36.247146] Stack : 803f751a 00000031 00000000 00000001 803ab4f4 803ab167 80364e10 00000000
[ 36.263898] 803f3660 0000013c ffffffff 00000200 00200000 8004d150 8036a600 803b0000
[ 36.280662] 00000003 0000013c 803687c8 8180de2c 00200000 8007b2cc 00000000 00000006
[ 36.297389] 00000006 00000000 00000000 00000000 00000000 00000000 00000000 00000000
[ 36.314149] 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000
[ 36.330908] ...
[ 36.335777] Call Trace:
[ 36.340740] [<8000e338>] show_stack+0x54/0x88
[ 36.349526] [<800238d0>] __warn+0xe4/0x118
[ 36.357701] [<80023934>] warn_slowpath_fmt+0x30/0x3c
[ 36.367692] [<8022e0b4>] dev_watchdog+0x1dc/0x25c
[ 36.377180] [<80057888>] call_timer_fn.isra.4+0x24/0x84
[ 36.387678] [<80057a94>] run_timer_softirq+0x1ac/0x1f8
[ 36.398018] [<8002643c>] __do_softirq+0x250/0x298
[ 36.407477] [<800095d8>] handle_int+0x138/0x144
[ 36.416580] [<80009460>] __r4k_wait+0x20/0x40
[ 36.425340] [<80047c24>] cpu_startup_entry+0x84/0xd0
[ 36.435323] [<803cabb4>] start_kernel+0x44c/0x46c
[ 36.444770] ---[ end trace 2b4e1c347d94e320 ]---
[ 36.454041] mtk_soc_eth 10100000.ethernet eth0: transmit timed out
[ 36.466425] mtk_soc_eth 10100000.ethernet eth0: dma_cfg:00000057
[ 36.478469] mtk_soc_eth 10100000.ethernet eth0: tx_ring=0, base=00eed000, max=128, ctx=2, dtx=0, fdx=0, next=2
[ 36.498503] mtk_soc_eth 10100000.ethernet eth0: rx_ring=0, base=00e02000, max=128, calc=127, drx=0
[ 47.178776] mtk_soc_eth 10100000.ethernet eth0: transmit timed out
[ 47.191186] mtk_soc_eth 10100000.ethernet eth0: dma_cfg:00000057
[ 47.203248] mtk_soc_eth 10100000.ethernet eth0: tx_ring=0, base=00e02000, max=128, ctx=2, dtx=0, fdx=0, next=2
[ 47.223257] mtk_soc_eth 10100000.ethernet eth0: rx_ring=0, base=00d58000, max=128, calc=127, drx=0
I have not attempted to bisect between HEAD and the commit introducing support for this device, but can do so soonish, unless someone beats me to it. |
|
airend: FS#1349 might be related. Try reverting 968b861. Also, in my case, the ETH driver eventually recovers itself, but will timeout again with increased traffic. I should mention that I'm forced to use MSS clamping for my PPPOE connection. |
|
airend: Looks like K4.14 improves things here; fingers crossed that it was a 4.9-related bug. |
|
ambientsummer: r6225-9595a9e8d3 with 4.14. Looks stable. |
|
K900: Does miwifi work on 4.14? I'm not really willing to test on my home router :( |
|
nbd: I've reverted the commit in r6325-e9b58b22c1 |
|
airend: Will it be brought back once ramips switches to K4.14? The bug doesn't happen there, and the GRO patch does seem an interesting idea. |
airend:
Experiencing periodic timeouts after
968b861612 ramips: improve NAPI polling to increase GRO batch size(Xiaomi MiWiFi Mini - MT7620A ver:2 eco:6).The text was updated successfully, but these errors were encountered: