Bruno Randolf wrote: > On Tuesday 18 May 2010 17:38:12 Robert wrote: > >> Hi. >> >> I'm currently trying to migrate a small embedded linux distribution from >> madwifi >> to ath5k and have run in to some problems. >> >> I started out using kernel version 2.6.27-44 and compat-wireless >> 2010-03-28 with >> some success (links came up but i experienced system lockups). >> >> I then proceeded to upgrade to kernel version 2.6.33-3 with compat-wireless >> 2010-05-16. Everything loads fine without any errors, but the radios seem >> totally dead. Nothing goes out, and nothing is recieved. >> "iw wlan0 scan dump" shows nothing. >> In /debug/ath5k/phy0/antenna, all the rx/tx counters are zero. >> >> Turning on xmit/recv debug doesn't help much. It seems like the packets are >> beeing sent, >> echo dumptx > debug shows hexdumps of the packets, and "trace" shows that >> ath5k_hw_start_tx_dma runs for every packet. >> >> Using a laptop as a monitor node indicates that nothing goes out on the >> radio. >> >> I have tried the cards in ibss, managed, and monitor mode without any luck. >> >> I have seen this same problem last year, when i last had a look at >> migrating to ath5k so perhaps this is some old problem resurfacing. >> >> Any ideas on where to continue looking? >> >> Some compiled system info: >> Linux kernel: 2.6.33-3, open embedded based "homebrew" distro >> Arm ixp4xx big endian platform (Gateworks Cambria) >> Radio cards (tested with several models, same result): >> Compex WLM54AG (AR5413) >> Wistron DCMA-82 (AR5414) >> Wistron CM-9 (AR5213) >> > > can you receive? do you see packets in monitor mode? > No, not a single packet. And a scan doesn't return anything either. It's just like there isn't a radio chip there at all. > i'm using AR5414 chipsets on a big endian mips platform, and it works fine, > so > it's unlikely that it's a chipset or endian issue. maybe something to do with > the DMA setup on your platform? for example on the routerstation (ar71xx) > boards we need to reduce the RX DMA size, but this does only affect stability > and performance, and doesn't result in not beeing able to transmit at all... > check for AR5K_DMASIZE_128B in reset.c - just an idea... i really don't know. > Tried lowering it to 64 and even 32B without any result.
However, i agree that it could have something to do with the DMA on the platform. Something have probably changed between the 2.6.27 and 2.6.33 kernel. Don't really know where to start looking though. > or... > > can you try to see if this patch helps? > "ath5k: wake queues on reset" - https://patchwork.kernel.org/patch/98972/ > No, sorry. That didn't help either. /Robert > bruno > _______________________________________________ ath5k-devel mailing list ath5k-devel@lists.ath5k.org https://lists.ath5k.org/mailman/listinfo/ath5k-devel
