From owner-aic7xxx Tue Sep 9 23:34:12 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id XAA10686 for aic7xxx-outgoing; Tue, 9 Sep 1997 23:34:12 -0700 (PDT) Received: from rrzs2.rz.uni-regensburg.de (rrzs2.rz.uni-regensburg.de [132.199.1.2]) by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id XAA10681 for ; Tue, 9 Sep 1997 23:34:07 -0700 (PDT) Received: from ngate.ngate.uni-regensburg.de (ngate.rz.uni-regensburg.de [132.199.3.13]) by rrzs2.rz.uni-regensburg.de (8.8.5/8.8.5) with SMTP id IAA05920; Wed, 10 Sep 1997 08:33:49 +0200 (MET DST) Received: from rkdvmks1.ngate.uni-regensburg.de by ngate.ngate.uni-regensburg.de; Wed, 10 Sep 97 07:34 MET Received: from rkdvmks1.ngate.uni-regensburg.de by kgate.ngate.uni-regensburg.de; Wed, 10 Sep 97 06:33 GMT Received: from RKDVMKS1/SpoolDir by rkdvmks1.ngate.uni-regensburg.de (Mercury 1.32); 10 Sep 97 08:33:29 +0200 Received: from SpoolDir by RKDVMKS1 (Mercury 1.32); 10 Sep 97 08:33:10 +0200 From: "Ulrich Windl" Organization: Universitaet Regensburg, Klinikum To: linux-kernel@vger.rutgers.edu Date: Wed, 10 Sep 1997 08:33:05 +0200 MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-transfer-encoding: 7BIT Subject: 2.0.31-pre9: AIC7xxx stops the show Cc: aic7xxx@freebsd.org Priority: normal X-mailer: Pegasus Mail for Windows (v2.53/R1) Message-ID: <7A967FC5D64@rkdvmks1.ngate.uni-regensburg.de> Sender: owner-aic7xxx@freebsd.org X-Loop: FreeBSD.org Precedence: bulk (Even though the kernel died rather cleanly:) The kernel died while I was making a backup of two complete kernel trees from SCSI disk to SCSI tape. It worked for a while, then suddenly it stopped. scsi: aborting due to timeout Kernel panic: aic7xxx: Waiting List inconsistency; SCB index=255, numscbs=9 In swapper task - not syncing I have a AHA2940 with a 1GB Conner disk and a HP C1536 DDS drive. Tagged queueing is enabled with a qsize of 8 for the disk. I also had enabled SCB paging. (I had this setup for a long time without trouble) Also there's a bug with the /proc stats that some August-patch from Doug Lefford removed. I was surprised to see it (With scanning for multiple LUNs enabled things are even worse): The aic7xxx driver in 2.0.31-pre9 writes statistic entries for non-existing IDs and LUNs. I think there was a fix for that already. Adaptec AIC7xxx driver version: 4.1/3.2 Compile Options: AIC7XXX_RESET_DELAY : 5 AIC7XXX_TAGGED_QUEUEING: Enabled AIC7XXX_PAGE_ENABLE : Enabled AIC7XXX_PROC_STATS : Enabled Adapter Configuration: SCSI Adapter: Adaptec AHA-294X SCSI host adapter (AIC-787x chipset) Host Bus: Single Base IO: 0xe000 Base IO Memory: 0xf9800000 IRQ: 11 SCBs: Used 4, HW 16, Page 255 Interrupts: 6381 Serial EEPROM: True Extended Translation: Enabled SCSI Bus Reset: Enabled Ultra SCSI: Disabled Target Disconnect: Enabled Statistics: CHAN#A (TGT 0 LUN 0): nxfers 6362 (6155 read;207 written) blks(512) rd=83477; blks(512) wr=1056 < 512 512-1K 1-2K 2-4K 4-8K 8-16K 16-32K 32-64K 64-128K >128K Reads: 3 5 4 4757 496 246 152 492 0 0 Writes: 0 0 0 160 45 2 0 0 0 0 CHAN#A (TGT 1 LUN 0): nxfers 2 (2 read;0 written) blks(512) rd=0; blks(512) wr=0 < 512 512-1K 1-2K 2-4K 4-8K 8-16K 16-32K 32-64K 64-128K >128K Reads: 2 0 0 0 0 0 0 0 0 0 Writes: 0 0 0 0 0 0 0 0 0 0 CHAN#A (TGT 2 LUN 0): nxfers 2 (2 read;0 written) blks(512) rd=0; blks(512) wr=0 < 512 512-1K 1-2K 2-4K 4-8K 8-16K 16-32K 32-64K 64-128K >128K Reads: 2 0 0 0 0 0 0 0 0 0 Writes: 0 0 0 0 0 0 0 0 0 0 CHAN#A (TGT 3 LUN 0): nxfers 2 (2 read;0 written) blks(512) rd=0; blks(512) wr=0 < 512 512-1K 1-2K 2-4K 4-8K 8-16K 16-32K 32-64K 64-128K >128K Reads: 2 0 0 0 0 0 0 0 0 0 Writes: 0 0 0 0 0 0 0 0 0 0 CHAN#A (TGT 4 LUN 0): nxfers 2 (2 read;0 written) blks(512) rd=0; blks(512) wr=0 < 512 512-1K 1-2K 2-4K 4-8K 8-16K 16-32K 32-64K 64-128K >128K Reads: 2 0 0 0 0 0 0 0 0 0 Writes: 0 0 0 0 0 0 0 0 0 0 CHAN#A (TGT 5 LUN 0): nxfers 2 (2 read;0 written) blks(512) rd=0; blks(512) wr=0 < 512 512-1K 1-2K 2-4K 4-8K 8-16K 16-32K 32-64K 64-128K >128K Reads: 2 0 0 0 0 0 0 0 0 0 Writes: 0 0 0 0 0 0 0 0 0 0 CHAN#A (TGT 6 LUN 0): nxfers 2 (2 read;0 written) blks(512) rd=0; blks(512) wr=0 < 512 512-1K 1-2K 2-4K 4-8K 8-16K 16-32K 32-64K 64-128K >128K Reads: 2 0 0 0 0 0 0 0 0 0 Writes: 0 0 0 0 0 0 0 0 0 0 Ulrich Windl