Date: Mon, 23 Jan 2006 16:30:38 -0200 From: AT Matik <asstec@matik.com.br> To: freebsd-mobile@freebsd.org Subject: Re: Debugging ath device timeouts Message-ID: <200601231630.39009.asstec@matik.com.br> In-Reply-To: <43D518D9.6030208@errno.com> References: <20060123041139.GB69091@geoff.deadheaven.com> <20060123145219.GF17465@laverenz.de> <43D518D9.6030208@errno.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Monday 23 January 2006 15:56, Sam Leffler wrote: > > A device timeout in ath means a frame was submitted to the hardware for > transmit but no interrupt was received for 5 seconds. I cannot comment > on the other drivers but I find it unlikely that the issue is common. not exactly, this is very common on almost all WL cards and they are very=20 sensitive to this in FreeBSD, and the risk grows while traffic grows, often= =20 it happens after the card was idle also, seems pci-cards are less and PCMCI= A=20 cards more risky the exactly same hardware with linux or staros often do not show this probl= em,=20 booting back to freebsd it appears imediatly wi cards often do not come back but ath card driver seems to be better device_polling helped a lot because seems that other NICs on the PC could b= e=20 taken out so that only the WLcards are interrupt driven it helps also a lot using better MBs with nothing onboard or at least with= =20 good acpi qualities and anything you do not need disabled in the Bios often this points are useless for client stations which often do not can=20 disable sio/lpt or others often helps checking with pciconf -lv or vmstat -i to find a slot order whe= re=20 the WLcard is not sharing any other resource certain MBs work better with and other without acpi enabled Jo=E3o A mensagem foi scaneada pelo sistema de e-mail e pode ser considerada segura. Service fornecido pelo Datacenter Matik https://datacenter.matik.com.br
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200601231630.39009.asstec>