Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1752858AbYJWIY6 (ORCPT ); Thu, 23 Oct 2008 04:24:58 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1751254AbYJWIYl (ORCPT ); Thu, 23 Oct 2008 04:24:41 -0400 Received: from pasmtpb.tele.dk ([80.160.77.98]:36546 "EHLO pasmtpB.tele.dk" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751260AbYJWIYj (ORCPT ); Thu, 23 Oct 2008 04:24:39 -0400 Date: Thu, 23 Oct 2008 10:23:40 +0200 From: Jens Axboe To: Elias Oltmanns Cc: linux-ide@vger.kernel.org, linux-kernel@vger.kernel.org, tj@kernel.org, jeff@garzik.org Subject: Re: [PATCH 1/2] libata: get rid of ATA_MAX_QUEUE loop in ata_qc_complete_multiple() Message-ID: <20081023082340.GX22217@kernel.dk> References: <1224661243-7929-1-git-send-email-jens.axboe@oracle.com> <1224661243-7929-2-git-send-email-jens.axboe@oracle.com> <878wsgqyix.fsf@denkblock.local> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <878wsgqyix.fsf@denkblock.local> Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 2223 Lines: 70 On Wed, Oct 22 2008, Elias Oltmanns wrote: > Jens Axboe wrote: > > We very rarely (if ever) complete more than one command in the > > sactive mask at the time, even for extremely high IO rates. So > > looping over the entire range of possible tags is pointless, > > instead use __ffs() to just find the completed tags directly. > > > > Signed-off-by: Jens Axboe > > --- > > drivers/ata/libata-core.c | 15 +++++++++------ > > 1 files changed, 9 insertions(+), 6 deletions(-) > > > > diff --git a/drivers/ata/libata-core.c b/drivers/ata/libata-core.c > > index 1ee9499..c3c53e7 100644 > > --- a/drivers/ata/libata-core.c > > +++ b/drivers/ata/libata-core.c > > @@ -4799,9 +4799,9 @@ void ata_qc_complete(struct ata_queued_cmd *qc) > > */ > > int ata_qc_complete_multiple(struct ata_port *ap, u32 qc_active) > > { > > + unsigned int i = 0; > > int nr_done = 0; > > u32 done_mask; > > - int i; > > > > done_mask = ap->qc_active ^ qc_active; > > > > @@ -4811,16 +4811,19 @@ int ata_qc_complete_multiple(struct ata_port *ap, u32 qc_active) > > return -EINVAL; > > } > > > > - for (i = 0; i < ATA_MAX_QUEUE; i++) { > > + while (done_mask) { > > struct ata_queued_cmd *qc; > > + unsigned int next = __ffs(done_mask); > > > > - if (!(done_mask & (1 << i))) > > - continue; > > - > > - if ((qc = ata_qc_from_tag(ap, i))) { > > + qc = ata_qc_from_tag(ap, i + next); > > + if (qc) { > > ata_qc_complete(qc); > > nr_done++; > > } > > + if (++next >= ATA_MAX_QUEUE) > > + break; > > If you think about it, this statement is equivalent to > > if (ap->qc_active ^ qc_active == (1 << (ATA_MAX_QUEUE - 1))) > > To fix this, you could say > > if (++next + i >= ATA_MAX_QUEUE) > > but perhaps it would be even more efficient (or not much worse) to skip > this check entirely. Yeah, the check should just be killed, that's the version I posted in the reply to Tejun as well. -- Jens Axboe -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/