From mboxrd@z Thu Jan 1 00:00:00 1970 From: Sergei Shtylyov Subject: Re: Fwd: CMD 64x regression from 2.6.21 to 2.6.22 and 2.6.23? Date: Sat, 20 Oct 2007 00:26:23 +0400 Message-ID: <471912EF.4020208@ru.mvista.com> References: <200710142211.03382.marogge@onlinehome.de> <200710172115.13068.marogge@onlinehome.de> <47167335.1020903@ru.mvista.com> <200710192158.03836.marogge@onlinehome.de> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit Return-path: Received: from gateway-1237.mvista.com ([63.81.120.155]:56779 "EHLO imap.sh.mvista.com" rhost-flags-OK-FAIL-OK-FAIL) by vger.kernel.org with ESMTP id S1761274AbXJSU0N (ORCPT ); Fri, 19 Oct 2007 16:26:13 -0400 In-Reply-To: <200710192158.03836.marogge@onlinehome.de> Sender: linux-ide-owner@vger.kernel.org List-Id: linux-ide@vger.kernel.org To: Martin Rogge Cc: Bartlomiej Zolnierkiewicz , Mark Lord , linux-ide@vger.kernel.org Hello. Martin Rogge wrote: >> BTW, can you try adding #define DEBUG to the driver meanwhile?.. > Yoda said: Try not. Do or do not. There is no try. :-) > So I did it. To be precise, I #defined both DEBUG and CMD_DEBUG. However, I am > not sure the result is conclusive. > On a good kernel I get a lot of lines of the type > hda: dma_stat: 0x24 irq_stat: 0x44 mask: 0x04 > hdc: dma_stat: 0x24 irq_stat: 0x5c mask: 0x10 Yeah, this is what's emitted by the dma_end() method which used CFR/ARTTIM23 PCI config. regs. to chack the interrupt status. > and variations thereof. On a broken kernel the middle part changes to > hda: dma_stat: 0x24 mrdmode: 0x00 mask: 0x04 Hm... this means that the chip doesn't work as documented in the spec, i.e. MRDMODE reg. seems write only, like on older chips. Could you post the output of 'lspci -v'? > Hope this is of any help, It is of great help. :-) > cu Martin MBR, Sergei