Just as a reference, does madwifi exhibit this behaviour?
Adrian On 3 March 2011 13:14, Fabrice <fabricedey...@agilemesh.com> wrote: > I'm now able able to prevent the CM9 to lock up. > I changed the RX/TX DMA burst size from 128 bytes to 16 bytes. > When I stream TCP data, I still some issues. The TX queue gets stuck and > when it > tries to reset it cannot stop the RX dma. > > ath5k phy0: (ath5k_tx_complete_poll_work:2306): TX queue stuck 2 > ath5k phy0: (ath5k_tx_complete_poll_work:2321): TX queues stuck, resetting > ath5k phy0: (ath5k_reset:2665): resetting > ath5k phy0: (ath5k_hw_stop_rx_dma:77): failed to stop RX DMA ! > ath5k phy0: (ath5k_rx_start:1120): cachelsz 32 rx_bufsize 2368 > > > As a result my TCP connection is interrupted and recovers. > > Here are the queue stats from debugfs: > > available txbuffers: 196 > 00: setup > len: 0 bufs: 0 > stuck: 0 > 01: setup > len: 0 bufs: 0 > stuck: 0 > 02: setup > len: 4 bufs: 4 > stuck: 16 > 03: setup > len: 0 bufs: 0 > stuck: 0 > 04: not setup > 05: not setup > 06: setup > len: 0 bufs: 0 > stuck: 0 > 07: not setup > 08: not setup > 09: not setup > > As a side note, on the same platform (ARM9-297 Mhz)I can have an AR9160 or > AR9220 (using ath9k) it works great. Any idea why with ath5k I'm running > inot > these issues? > > _______________________________________________ > ath5k-devel mailing list > ath5k-devel@lists.ath5k.org > https://lists.ath5k.org/mailman/listinfo/ath5k-devel >
_______________________________________________ ath5k-devel mailing list ath5k-devel@lists.ath5k.org https://lists.ath5k.org/mailman/listinfo/ath5k-devel