On 7 May 2014 02:09, Owen Kirby <o...@exegin.com> wrote: > Did a sysupgrade to the latest bcrm47xx trunk images and tried to load > the network up with some traffic (connected a few laptops to the LAN > ports and ran iperf between each other and the WNDR3700v3) Nothing came > up in the kernel log that indicated anything out of the ordinary > happening after a few hours of this testing. > > Only one of my laptops had a gigabit ethernet port, so I don't know if I > was really able to generate enough traffic to reproduce the problem, > though at the very least I was able to max out the CPU on the router.
I believe your testing was absolutely correct. In fact, this bgmac/ethernet bug seems to occur even without a high traffic. Thanks a lot for your testing and spending time on this. Unfortunately it seems to be very rare issue affecting single devices only. To be honest, I've seen this bug *once* on my BCM4706: [ 2071.640000] bgmac bcma0:0: Found poisoned packet at slot 87, DMA issue! [ 2071.650000] bgmac bcma0:0: Found poisoned packet at slot 88, DMA issue! [ 3606.070000] skbuff: skb_over_panic: text:87af1024 len:43226 put:43226 head:87b3d020 data:87b3d060 tail:0x87b4793a end:0x87b3d680 dev:eth0 Unfortunately I was never able to reproduce this. _______________________________________________ openwrt-devel mailing list openwrt-devel@lists.openwrt.org https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel
