On 2011-06-18 6:05 PM, Peter Wagner wrote:
Am Freitag, 17. Mai 2011, 22:04:07 schrieb Peter Wagner:
accidently hit the send button...
Hi,
i created a c++ program and while testing it (it sends upd packets) i
created a few millions packages and sent them to 192.168.1.1:4950 (my
router running openwrt). its a wl500gP v1. while i was doing this the
device gets unresponsive and i get lines like this:
b44 ssb0:0: eth0: powering down PHY
b44 ssb0:0: eth0: powering down PHY
b44 ssb0:0: eth0: powering down PHY
after doing this for about 2 minutes the device reboots. i think this is
triggered by the watchdog as it get no enough process time from the
sheduler.
how can i track down why this powering down PHY happens?
regards
Peter
_______________________________________________
openwrt-devel mailing list
openwrt-devel@lists.openwrt.org
https://lists.openwrt.org/mailman/listinfo/openwrt-devel
i created this little c program (the code flys around somewhere in the inet)
and modified it a bit it sends 100000000 udp packets to a host you specify
compile with gcc -o foo talker_foo.c
and run
./foo $hostname $message_that_gets_sent
i can reproduce the error in ~10sec of running this program against a wl500gp
kernel 2.6.39.1
after this 10sec i have about 160 lines like this
b44 ssb0:0: eth0: powering down PHY
in the output of dmesg.
i tested it against an unfiltered port.
Looks like there are several conditions that trigger a hardware restart
which produce these messages:
b44.h:
#define ISTAT_ERRORS
(ISTAT_DSCE|ISTAT_DATAE|ISTAT_DPE|ISTAT_RDU|ISTAT_RFO|ISTAT_TFU)
probably ISTAT_RFO is hit here - my guess is it stands for Receive FIFO
Overflow.
Maybe there's a different way this could be handled, or maybe the error
can be ignored altogether. Can you try removing ISTAT_RFO from
ISTAT_ERRORS and see if that keeps the interface working during your UDP
flood?
- Felix
_______________________________________________
openwrt-devel mailing list
openwrt-devel@lists.openwrt.org
https://lists.openwrt.org/mailman/listinfo/openwrt-devel
