From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jens Axboe Subject: Re: Sparc ESP problem with blk-mq Date: Tue, 14 Oct 2014 08:38:18 -0600 Message-ID: <543D355A.3030605@kernel.dk> References: Mime-Version: 1.0 Content-Type: text/plain; charset=windows-1252; format=flowed Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: Sender: sparclinux-owner@vger.kernel.org To: Meelis Roos , linux-scsi@vger.kernel.org, sparclinux@vger.kernel.org Cc: Christoph Hellwig List-Id: linux-scsi@vger.kernel.org (resent with correct linux-scsi address) On 2014-10-13 16:07, Meelis Roos wrote: > I turned on blk-mq for all my test machines today and got a failure from > Sun Ultra 1 with Sparc ESP SCSI (dmesg below). Booting with > scsi_mod.use_blk_mq=0 fixes the problem. The problem is reproducible. [snip] > [ 63.244199] esp: esp0, regs[1ffe8800000:1ffe8400000] irq[10] > [ 63.310055] esp: esp0 is a FAS100A, 40 MHz (ccf=0), SCSI ID 7 > [ 66.383895] scsi host0: esp > [ 148.437533] scsi 0:0:0:0: timing out command, waited 82s > [ 149.674112] ESP: unexpected IREG 40 > [ 149.713889] esp: esp0: Dumping command log > [ 149.762847] esp: esp0: ent[14] CMD val[01] sreg[93] seqreg[04] sreg2[00] ireg[18] ss[00] event[0d] [snip] Christoph, any idea on this? The command just times out, for some reason. Only thing I could think of was related to perhaps missing bouncing or similar, but I don't think that's the case. But the hba must not like the request somehow. Or perhaps it sets up the wrong number of segments. -- Jens Axboe