From mboxrd@z Thu Jan 1 00:00:00 1970 From: Stephen Hemminger Subject: Re: 2.6.19: very low transmit speed after suspend/resume with Sky2 Date: Sat, 02 Dec 2006 13:16:28 -0800 Message-ID: <4571ED2C.6040005@osdl.org> References: <20061202091626.GA20734@thinkpad.home.local> <20061202131833.GA3673@thinkpad.home.local> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Cc: netdev@vger.kernel.org Return-path: Received: from smtp.osdl.org ([65.172.181.25]:37302 "EHLO smtp.osdl.org") by vger.kernel.org with ESMTP id S1162483AbWLBVQg (ORCPT ); Sat, 2 Dec 2006 16:16:36 -0500 To: Tino Keitel In-Reply-To: <20061202131833.GA3673@thinkpad.home.local> Sender: netdev-owner@vger.kernel.org List-Id: netdev.vger.kernel.org Tino Keitel wrote: > On Sat, Dec 02, 2006 at 10:16:28 +0100, Tino Keitel wrote: > >> Hi folks, >> >> since 2.6.19 I have the problem that I get lousy transmit speeds of < 1 >> mbit/s. >> >> Here is a quick test, the 5th column is the transfer speed in mbit/s: >> >> From the affected machine to my laptop over WLAN (the affected machine >> is connected to a Fast Ethernet switch): >> >> $ nttcp -r -l 1400 -n 500 dose >> l 700000 8.16 0.04 0.6867 155.5469 824 101.04 22887.6 >> 1 700000 8.15 0.00 0.6873 560000.0000 500 61.37 50000000.0 >> >> Only 0,68 mbit/s during transmit. :-( >> >> From my laptop to the affected machine: >> >> $ nttcp -l 1400 -n 5000 dose >> l 7000000 2.21 0.03 25.3427 1999.8572 5000 2262.74 178558.7 >> 1 7000000 2.25 0.07 24.8363 823.4810 9268 4110.41 136286.1 >> >> This is ok, 25 mbit/s during receive, limited by the WLAN in the >> laptop. >> >> After reboot: >> >> $ nttcp -l 1400 -n 5000 dose >> l 7000000 2.04 0.03 27.5005 1749.8906 5000 2455.40 156240.2 >> 1 7000000 2.08 0.06 26.9843 999.9464 9455 4556.00 168830.2 >> >> Now I got 27 mbit/s during receive, that's ok for WLAN. >> > > Sorry, I should give the number for _transmit_ here. This was ok, too. > > Regards, > Tino Most likely, the problem is similar to one I am working. The device_resume code gets run before BIOS. And the BIOS wakeup screws up the device. The proper fix will happen in 2.6.20. If you are using MSI, turn that off.