From mboxrd@z Thu Jan 1 00:00:00 1970 From: =?UTF-8?B?UmFmYcWCIE1pxYJlY2tp?= Date: Mon, 30 May 2011 09:45:49 +0200 Subject: Request for testing b43 & Broadcom 14e4:4315 cards for DMA ERRORs (Linux 3.0) In-Reply-To: References: Message-ID: List-Id: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: b43-dev@lists.infradead.org W dniu 30 maja 2011 09:22 u?ytkownik Rafa? Mi?ecki napisa?: > I believe we've fixed that errors in Linux 3.0-rc1. Can you give it a > chance and report back? I'd like to say (and you should too) big thanks to Juan, who gave me access to his affected machine and spent a lot of time testing. He even bisected kernel on his Atom pointing us to hrtimers regression. If this is going to work, I've some annoying info for Larry. I can not tell which of the recent b43/ssb commits were really needed for solving this issue the final one was "b43: implement timeouts workaround". The tricky part is we got very similar patch ready 1,5 year ago [0], by Larry! The mistake done by Larry was applying workaround on PCI(e) core instead of 80211 core. I've noticed that when comparing "ssb_read32(SSB_IMCFGLO);" after cold boot between ssb&b43 and wl. Drivers read different values from hardware, so it was a hint we were doing this on a wrong core. [0] http://lists.berlios.de/pipermail/bcm43xx-dev/2009-December/006592.html -- Rafa?