From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1761115AbYBZAcu (ORCPT ); Mon, 25 Feb 2008 19:32:50 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1757918AbYBZAck (ORCPT ); Mon, 25 Feb 2008 19:32:40 -0500 Received: from g1t0026.austin.hp.com ([15.216.28.33]:28490 "EHLO g1t0026.austin.hp.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1757312AbYBZAcj (ORCPT ); Mon, 25 Feb 2008 19:32:39 -0500 Date: Mon, 25 Feb 2008 18:32:38 -0600 From: Mike Miller To: Andrew Morton , Jens Axboe Cc: LKML , LKML-scsi , coughlan@redhat.com Subject: [PATCH 1/1] cciss: remove READ_AHEAD define and use block layer defaults Message-ID: <20080226003238.GA31623@roadking.ldev.net> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline User-Agent: Mutt/1.5.9i Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org PATCH 1/1 This patch removes the #define READ_AHEAD 1024 from the driver and uses the block layer defaults, instead. We have found that under certain workloads the setting can cause a disk connected to the e200 controller to go offline. If the disk hiccups the link may try to downshift but the controller is never notified that the link successfully completed the renegotiation. We've also found that performance using the block layer default of 32 pages was on par with the 1024 setting. We tried setting it to zero at one time based on info from our firmware guys but that killed performance. Turns out we were talking about 2 different read ahead settings. Please consider this for inclusion. Signed-off-by: Mike Miller diff --git a/drivers/block/cciss.c b/drivers/block/cciss.c index 9715be3..2f6cc3b 100644 --- a/drivers/block/cciss.c +++ b/drivers/block/cciss.c @@ -131,7 +131,6 @@ static struct board_type products[] = { /*define how many times we will try a command because of bus resets */ #define MAX_CMD_RETRIES 3 -#define READ_AHEAD 1024 #define MAX_CTLR 32 /* Originally cciss driver only supports 8 major numbers */ @@ -1341,7 +1340,6 @@ geo_inq: disk->private_data = &h->drv[drv_index]; /* Set up queue information */ - disk->queue->backing_dev_info.ra_pages = READ_AHEAD; blk_queue_bounce_limit(disk->queue, hba[ctlr]->pdev->dma_mask); /* This is a hardware imposed limit. */ @@ -3434,7 +3432,6 @@ static int __devinit cciss_init_one(struct pci_dev *pdev, } drv->queue = q; - q->backing_dev_info.ra_pages = READ_AHEAD; blk_queue_bounce_limit(q, hba[i]->pdev->dma_mask); /* This is a hardware imposed limit. */