Open aparcar opened 7 years ago
VittGam:
Hmmm... Obviously markdown syntax for blocks isn't supported here, but I can't see a way for me to edit the description...
BusyBox v1.26.2 () built-in shell (ash)
_________
/ /\ _ ___ ___ ___
/ LE / \ | | | | | | / DE / \ | || || |) | | /__/ LE \ |__|__|/|_| lede-project.org \ \ DE / \ LE \ / ----------------------------------------------------------- \ DE \ / Reboot (SNAPSHOT, r1-2f09a1e) ______\/ -----------------------------------------------------------
root@LEDE:~# dmesg | grep 8327
[ 0.815273] switch0: Atheros AR8327 rev. 4 switch registered on ag71xx-mdio.0
root@LEDE:~# dmesg | grep 8216
[ 2.046658] ag71xx ag71xx.0: connected to PHY at ag71xx-mdio.0:00 [uid=004dd034, driver=Atheros AR8216/AR8236/AR8316]
root@LEDE:~# swconfig dev switch0 get arl_table
address resolution table
Port 0: MAC
root@LEDE:~# dmesg | grep 8216
[ 2.046658] ag71xx ag71xx.0: connected to PHY at ag71xx-mdio.0:00 [uid=004dd034, driver=Atheros AR8216/AR8236/AR8316]
root@LEDE:~# swconfig dev switch0 get arl_table
address resolution table
Port 0: MAC
root@LEDE:~#
BusyBox v1.23.2 (2016-01-02 18:01:44 CET) built-in shell (ash)
root@OpenWrt:/# dmesg | grep 8327 [ 0.750000] switch0: Atheros AR8327 rev. 4 switch registered on ag71xx-mdio.0 root@OpenWrt:/# dmesg | grep 8216 [ 1.440000] ag71xx ag71xx.0: connected to PHY at ag71xx-mdio.0:00 [uid=004dd034, driver=Atheros AR8216/AR8236/AR8316] [ 24.680000] Atheros AR8216/AR8236/AR8316 ag71xx-mdio.0:00: Port 4 is up
root@OpenWrt:/# swconfig dev switch0 get arl_table
address resolution table
Port 0: MAC
root@OpenWrt:/# dmesg | grep 8216
[ 1.440000] ag71xx ag71xx.0: connected to PHY at ag71xx-mdio.0:00 [uid=004dd034, driver=Atheros AR8216/AR8236/AR8316]
[ 24.680000] Atheros AR8216/AR8236/AR8316 ag71xx-mdio.0:00: Port 4 is up
[ 1306.510000] Atheros AR8216/AR8236/AR8316 ag71xx-mdio.0:00: Port 4 is down
[ 1310.530000] Atheros AR8216/AR8236/AR8316 ag71xx-mdio.0:00: Port 3 is up
root@OpenWrt:/# swconfig dev switch0 get arl_table
address resolution table
Port 0: MAC
root@OpenWrt:/# uname -a Linux OpenWrt 3.18.23 #1 Sun Jan 31 18:39:35 CET 2016 mips GNU/Linux root@OpenWrt:/#
storchi:
same problem: #604 polling solution: http://centaur.sch.bme.hu/~leait/projects/openwrt/ seems included in CC regards cs
VittGam:
Hmmm, I should re-test with my WDR3600 and WDR4300 then. I might have messed up with my tests on those routers, since I observed that it didn't happen on WDR4300...
By the way, polling is not the solution here, since the switch sends the change notifications as IRQs, and the kernel picked them up correctly with OpenWrt CC 15.05.1 release (which does not seem to have that polling patch you linked)...
VittGam:
Oh and by the way, this problem does not happen on a TL-WR1043ND v4, which has a QCA956x SoC and an AR8337 rev 2 switch chip.
VittGam:
Hello,
On my TP-LINK TL-WR1043ND v3 (ar71xx QCA9558) with today's LEDE snapshot, the AR8327 switch isn't sending port change interrupt events to the CPU.
The
ar8xxx_check_link_states
function intarget/linux/generic/files/drivers/net/phy/ar8216.c
is not being called at all, since not only the messages about port link status change aren't printed to the dmesg, but the ARL table isn't getting cleaned either. I don't understand why this is happening...This regression is not present in other routers with the same AR8327 switch chip (but rev. 2 instead of rev. 4), such as TP-LINK TL-WDR4300 v1 (ar71xx) and TP-LINK TL-WDR4900 v1 (mpc85xx).
(Latest LEDE stable release also does exibit the regression.)
Below you can see a session with LEDE; and a session with OpenWrt CC 15.05.1 release, which does not exibit the problem.
(The CC image is for the TL-WR1043ND v2, since v3 images weren't built yet at that time; v2 and v3 are the same hardware, as it can be seen from the FCC docs, where TP-LINK requested an ID change from v2 to v3.)
Please let me know if there's anything I can try to help debug this problem!
Cheers, Vittorio