Return-path: Received: from web23107.mail.ird.yahoo.com ([217.146.189.47]:21603 "HELO web23107.mail.ird.yahoo.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with SMTP id S1753794AbYKDU1P convert rfc822-to-8bit (ORCPT ); Tue, 4 Nov 2008 15:27:15 -0500 Date: Tue, 4 Nov 2008 20:27:12 +0000 (GMT) From: Hin-Tak Leung Reply-To: htl10@users.sourceforge.net Subject: Re: [RFC/RFT PATCH v2 2/2] rtl8187: feedback transmitted packets using tx close descriptor for 8187B To: linux-wireless@vger.kernel.org, Herton Ronaldo Krzesinski Cc: Larry Finger , John W Linville , Johannes Berg , Michael Wu , Andrea Merello In-Reply-To: <1225806649-10050-3-git-send-email-herton@mandriva.com.br> MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8 Message-ID: <244535.27936.qm@web23107.mail.ird.yahoo.com> (sfid-20081104_212725_056619_144F3635) Sender: linux-wireless-owner@vger.kernel.org List-ID: --- On Tue, 4/11/08, Herton Ronaldo Krzesinski wrote: > Realtek 8187B has a receive command queue to feedback beacon > interrupt > and transmitted packet status. Use it to feedback mac80211 > about status > of transmitted packets. Unfortunately in the course of > testing I found > that the sequence number reported by hardware includes > entire sequence > control in a 12 bit only field, so a workaround is done to > check only > lowest bits. > --- > v2: fix rate stuck at 1M reported by Larry Finger, Hin-Tak > and also seen by me. > btw, I noted now with a ralink device that rt2500usb > (rt2x00) has same > issue currently. > - hdr->retry = > cpu_to_le32((info->control.rates[0].count - 1) << > 8); > + hdr->retry = > cpu_to_le32(info->control.rates[0].count << 8); > - hdr->retry = > cpu_to_le32((info->control.rates[0].count - 1) << > 8); > + hdr->retry = > cpu_to_le32(info->control.rates[0].count << 8); These are the only lines which changed between the two versions of the patch - if removing the two -1's fixes the rate stuck (and these two lines are newly introduced), shouldn't this be a separate commit if it fixes a problem introduced by an earlier commit? I did *not* (and do not) have a rate-stuck problem, but I am running a slightly mod version of the patch against 2.6.27, which doesn't have "rates[x].count" yet, with the retry_count struct member which seems to have since been removed.