Return-path: Received: from nbd.name ([46.4.11.11]:38335 "EHLO nbd.name" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S932675Ab2LGLrX (ORCPT ); Fri, 7 Dec 2012 06:47:23 -0500 Message-ID: <50C1D747.2020005@openwrt.org> (sfid-20121207_124727_076337_151D0729) Date: Fri, 07 Dec 2012 12:47:19 +0100 From: Felix Fietkau MIME-Version: 1.0 To: Rajkumar Manoharan CC: Adrian Chadd , linux-wireless@vger.kernel.org, linville@tuxdriver.com, rodrigue@qca.qualcomm.com Subject: Re: [PATCH] Revert "ath9k_hw: Update AR9003 high_power tx gain table" References: <1354815611-7220-1-git-send-email-nbd@openwrt.org> <20121207055322.GA27229@hemis.qca.qualcomm.com> In-Reply-To: <20121207055322.GA27229@hemis.qca.qualcomm.com> Content-Type: text/plain; charset=ISO-8859-1 Sender: linux-wireless-owner@vger.kernel.org List-ID: On 2012-12-07 6:53 AM, Rajkumar Manoharan wrote: > On Thu, Dec 06, 2012 at 11:35:33AM -0800, Adrian Chadd wrote: >> On 6 December 2012 09:40, Felix Fietkau wrote: >> > This reverts commit f74b9d365ddd33a375802b064f96a5d0e99af7c0. >> > >> > Turns out reverting commit a240dc7b3c7463bd60cf0a9b2a90f52f78aae0fd >> > "ath9k_hw: Updated AR9003 tx gain table for 5GHz" was not enough to >> > bring the tx power back to normal levels on devices like the >> > Buffalo WZR-HP-G450H, this one needs to be reverted as well. >> > >> > This revert improves tx power by ~10 db on that device >> >> Did anyone figure out why this was happening? >> >> Maybe we should look at the HAL commits around the same time to see if >> someone changed the programming - eg, assuming TPC is always on, or >> the 2 stream / 3 stream reduction PHY register, or such. >> > Generally we update initval taken from Systems team. We might have missed to > program the some registers while updating initvals. I prefer to fix the missing > portion rather than reverting ini tables. Otherwise we end up with using > older initvals and hard to debug hw related issues. For a few months I was trying to get an answer from the Systems team as to why that change was made. Contacted several people about it, got no proper response. When I first found the bug (a long time ago), I did extensive review of internal changes to driver/initvals around that time, and I'm pretty confident that this tx power issue is not caused by missing changes that we forgot to port over. - Felix