From mboxrd@z Thu Jan 1 00:00:00 1970 From: 'Christoph Hellwig' Subject: Re: Request for review of Linux iSCSI driver version 4.0.0.1 Date: Wed, 19 Nov 2003 13:38:05 +0000 Sender: linux-scsi-owner@vger.kernel.org Message-ID: <20031119133805.A12398@infradead.org> References: <20031027153932.A16679@infradead.org> <03111117263601.01918@naveenb-lnx.cisco.com> <20031111173622.A12435@infradead.org> <200311191810.30135.krmurthy@cisco.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Received: from pub234.cambridge.redhat.com ([213.86.99.234]:46087 "EHLO phoenix.infradead.org") by vger.kernel.org with ESMTP id S264076AbTKSNiH (ORCPT ); Wed, 19 Nov 2003 08:38:07 -0500 Content-Disposition: inline In-Reply-To: <200311191810.30135.krmurthy@cisco.com>; from krmurthy@cisco.com on Wed, Nov 19, 2003 at 06:10:30PM +0530 List-Id: linux-scsi@vger.kernel.org To: Krishna Murthy Cc: 'Christoph Hellwig' , linux-scsi@vger.kernel.org, davmyers@cisco.com On Wed, Nov 19, 2003 at 06:10:30PM +0530, Krishna Murthy wrote: > Hi , > The following was one of your comments on the iSCSI driver: > > "having multiple kmap() in the same process at the same time can kmap(), > you redesign iscsi_xmit_task/iscsi_xmit_data/iscsi_recv_data not to do that. > for the tx path use ->sendpage to avoid a data copy and kmapping altogether" > > > In the first line did you intend to say having multiple kmap() can deadlock? Yes. > Regarding usage of sendpage and avoiding kmapping altogether, > we need to access the data, to calculate data digest. If we do not use kmap > how else do we achieve it? Well, with software crypto someone will have to do the kmap (or kmap_atomic), but I'd rather leave that as an implementation detail to the crypto API as I already suggested earlier.