From mboxrd@z Thu Jan 1 00:00:00 1970 From: Arjen Verweij Subject: Re: wol support in forcedeth v25 Date: Wed, 08 Sep 2004 21:07:38 +0200 Sender: netdev-bounce@oss.sgi.com Message-ID: <413F587A.70902@student.tudelft.nl> References: <200404201236.41188.adq@lidskialf.net> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Cc: netdev@oss.sgi.com Return-path: To: Andrew de Quincey In-Reply-To: <200404201236.41188.adq@lidskialf.net> Errors-to: netdev-bounce@oss.sgi.com List-Id: netdev.vger.kernel.org Dear Andrew, Today I upgraded to forcedeth v0.28. Looking at the Changelog, nothing indicates additional changes have been made to WOL support. Is there anything I can do to get this in motion again? I am willing to test stuff when needed, and give feedback based on my system. For now I just change: #define DEV_NEED_TIMERIRQ 0x0008 to #define DEV_NEED_TIMERIRQ 0x0000 and use pci-config to set the involved bits of the chip to get to D3. From http://www.hailfinger.org/carldani/linux/patches/forcedeth/ I gather that NVIDIA has contributed gigabit support. Maybe they can spill some code to help get WOL support as well? From earlier emails with NVIDIA engineers I was told that WOL support is not part of the Linux driver, so it may be a long shit :) Thank you all for a great driver, Arjen Andrew de Quincey wrote: >On Tuesday 20 April 2004 12:13, Arjen Verweij wrote: > >I found how I could get it going again, but its not the right way of doing >it.. In nv_close(), if you comment out the lines "nv_stop_rx(dev);" and "if >(np->wolenabled) nv_start_rx(dev);", WOL works. > >However this isn't the right way to do it, which is why I haven't supplied a >patch. What its supposed to do it stop receiving, clear any buffers, then >restart receiving again. But this seems to break WOL for some reason. There >must be some extra operation required for it all to work properly... I'll >keep looking. > > > >>OK, let me know if you need a guinea pig to test any patches you might >>have. >> >>Regards, >> >>Arjen >> >>On Tue, 20 Apr 2004, Andrew de Quincey wrote: >> >> >>>On Monday 19 April 2004 21:11, Arjen Verweij wrote: >>> >>> >>>>Hi, >>>> >>>>Is the wol support in forcedeth v25 complete yet? I was trying it with >>>>the 2.6.5 kernel it came with, tried the new forcedeth driver with a >>>>2.6.3 kernel (because firewire seems broken in 2.6.5) however, my box >>>>will not wake up as of yet. >>>> >>>> >>>You're right! It _was_ working with patch-forced-0.25, but it ain't in >>>2.6.5. I'll see if I can spot the problem. >>> >>> >>> >>>>According to the source there is a FIXME with a comment about powering >>>>down the NIC. Does this mean that support for wol is incomplete for >>>>now? Mind you, I am just curious, this is in no way meant as criticism. >>>> >>>> >>>That would only be for when WOL is NOT used; I think it has to leave the >>>NIC powered up for WOL to work... and as you say it ain't implemented >>>yet. >>> >>> >>> >>>>On another note, Carl-Daniel mentioned that DEV_NEED_TIMERIRQ might be >>>>removed from future releases. Is this still the case? Feedback would be >>>>much appreciated, so I can keep my humble audience up to date on what's >>>>going on at the wol front. >>>> >>>>Thank you guys for your excellent work, >>>> >>>>Arjen >>>>http://atlas.et.tudelft.nl/verwei90/nforce2/index.html >>>> >>>>