From patchwork Wed Apr 12 05:30:41 2017 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: "Nicholas A. Bellinger" X-Patchwork-Id: 9676567 Return-Path: Received: from mail.wl.linuxfoundation.org (pdx-wl-mail.web.codeaurora.org [172.30.200.125]) by pdx-korg-patchwork.web.codeaurora.org (Postfix) with ESMTP id 8548760383 for ; Wed, 12 Apr 2017 05:30:48 +0000 (UTC) Received: from mail.wl.linuxfoundation.org (localhost [127.0.0.1]) by mail.wl.linuxfoundation.org (Postfix) with ESMTP id 76FB128512 for ; Wed, 12 Apr 2017 05:30:48 +0000 (UTC) Received: by mail.wl.linuxfoundation.org (Postfix, from userid 486) id 6B507285A5; Wed, 12 Apr 2017 05:30:48 +0000 (UTC) X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on pdx-wl-mail.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-6.8 required=2.0 tests=BAYES_00,DKIM_SIGNED, RCVD_IN_DNSWL_HI,T_DKIM_INVALID autolearn=unavailable version=3.3.1 Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.wl.linuxfoundation.org (Postfix) with ESMTP id CC36128512 for ; Wed, 12 Apr 2017 05:30:45 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1751115AbdDLFap (ORCPT ); Wed, 12 Apr 2017 01:30:45 -0400 Received: from mail.linux-iscsi.org ([67.23.28.174]:55678 "EHLO linux-iscsi.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751108AbdDLFao (ORCPT ); Wed, 12 Apr 2017 01:30:44 -0400 Received: from [192.168.1.66] (75-37-194-224.lightspeed.lsatca.sbcglobal.net [75.37.194.224]) (using SSLv3 with cipher AES256-SHA (256/256 bits)) (No client certificate requested) (Authenticated sender: nab) by linux-iscsi.org (Postfix) with ESMTPSA id 6D2A340B09; Wed, 12 Apr 2017 05:31:18 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=simple/simple; d=linux-iscsi.org; s=default.private; t=1491975079; bh=YyHzFVTSbWNBSj7EK+1uWyCiQpHoRfU pHO/pFM1ohtE=; h=Message-ID:Subject:From:To:Cc:Date:In-Reply-To: References:Content-Type:Mime-Version:Content-Transfer-Encoding; b=TAZ+bgBhZfp4AiKn80bkbVqTrd2OEnykLgDxF6PkRNNwut5me8rRBnJ+5vItWqj9k AZs2lGnl1KKpPGejutNZysVrCQEfsFJoFMMgTs2Y+rJ5MUMOia5cLUX8z0Hpyx/qbDE LYvvGrKushIihjMdBjU/aUw0mxXpUO3VY1DU478= Message-ID: <1491975041.8231.131.camel@haakon3.risingtidesystems.com> Subject: Re: [PATCH 2/8] target: remove iblock WRITE_SAME passthrough support From: "Nicholas A. Bellinger" To: Christoph Hellwig Cc: axboe@kernel.dk, martin.petersen@oracle.com, philipp.reisner@linbit.com, lars.ellenberg@linbit.com, target-devel@vger.kernel.org, linux-block@vger.kernel.org, linux-scsi@vger.kernel.org, drbd-dev@lists.linbit.com, dm-devel@redhat.com Date: Tue, 11 Apr 2017 22:30:41 -0700 In-Reply-To: <20170410160807.23674-3-hch@lst.de> References: <20170410160807.23674-1-hch@lst.de> <20170410160807.23674-3-hch@lst.de> X-Mailer: Evolution 3.4.4-1 Mime-Version: 1.0 Sender: target-devel-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: target-devel@vger.kernel.org X-Virus-Scanned: ClamAV using ClamSMTP On Mon, 2017-04-10 at 18:08 +0200, Christoph Hellwig wrote: > Use the pscsi driver to support arbitrary command passthrough > instead. > The people who are actively using iblock_execute_write_same_direct() are doing so in the context of ESX VAAI BlockZero, together with EXTENDED_COPY and COMPARE_AND_WRITE primitives. Just using PSCSI is not an option for them. In practice though I've not seen any users of IBLOCK WRITE_SAME for anything other than VAAI BlockZero, so just using blkdev_issue_zeroout() when available, and falling back to iblock_execute_write_same() if the WRITE_SAME buffer contains anything other than zeros should be OK. How about something like the following below..? This would bring parity to how blkdev_issue_write_same() works atm wrt to synchronous bio completions. However, most folks with a raw make_request or blk-mq backend driver that supports multiple GB/sec of zero bandwidth end up changing IBLOCK to support asynchronous REQ_WRITE_SAME completions anyways. I'd be happy to add support for that using __blkdev_issue_zeroout() once the basic conversion is in place. From ff74012eaff38f9fa0d74aca60507b9964f484ce Mon Sep 17 00:00:00 2001 From: Nicholas Bellinger Date: Tue, 11 Apr 2017 22:21:47 -0700 Subject: [PATCH] target/iblock: Convert WRITE_SAME to blkdev_issue_zeroout Signed-off-by: Nicholas Bellinger --- drivers/target/target_core_iblock.c | 44 +++++++++++++++++++++++-------------- 1 file changed, 27 insertions(+), 17 deletions(-) -- To unsubscribe from this list: send the line "unsubscribe target-devel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html diff --git a/drivers/target/target_core_iblock.c b/drivers/target/target_core_iblock.c index d316ed5..5bfde20 100644 --- a/drivers/target/target_core_iblock.c +++ b/drivers/target/target_core_iblock.c @@ -86,6 +86,7 @@ static int iblock_configure_device(struct se_device *dev) struct block_device *bd = NULL; struct blk_integrity *bi; fmode_t mode; + unsigned int max_write_zeroes_sectors; int ret = -ENOMEM; if (!(ib_dev->ibd_flags & IBDF_HAS_UDEV_PATH)) { @@ -129,7 +130,11 @@ static int iblock_configure_device(struct se_device *dev) * Enable write same emulation for IBLOCK and use 0xFFFF as * the smaller WRITE_SAME(10) only has a two-byte block count. */ - dev->dev_attrib.max_write_same_len = 0xFFFF; + max_write_zeroes_sectors = bdev_write_zeroes_sectors(bd); + if (max_write_zeroes_sectors) + dev->dev_attrib.max_write_same_len = max_write_zeroes_sectors; + else + dev->dev_attrib.max_write_same_len = 0xFFFF; if (blk_queue_nonrot(q)) dev->dev_attrib.is_nonrot = 1; @@ -415,28 +420,31 @@ static void iblock_end_io_flush(struct bio *bio) } static sense_reason_t -iblock_execute_write_same_direct(struct block_device *bdev, struct se_cmd *cmd) +iblock_execute_zero_out(struct block_device *bdev, struct se_cmd *cmd) { struct se_device *dev = cmd->se_dev; struct scatterlist *sg = &cmd->t_data_sg[0]; - struct page *page = NULL; - int ret; + unsigned char *buf, zero = 0x00, *p = &zero; + int rc, ret; - if (sg->offset) { - page = alloc_page(GFP_KERNEL); - if (!page) - return TCM_OUT_OF_RESOURCES; - sg_copy_to_buffer(sg, cmd->t_data_nents, page_address(page), - dev->dev_attrib.block_size); - } + buf = kmap(sg_page(sg)) + sg->offset; + if (!buf) + return TCM_LOGICAL_UNIT_COMMUNICATION_FAILURE; + /* + * Fall back to block_execute_write_same() slow-path if + * incoming WRITE_SAME payload does not contain zeros. + */ + rc = memcmp(buf, p, cmd->data_length); + kunmap(sg_page(sg)); + + if (rc) + return TCM_LOGICAL_UNIT_COMMUNICATION_FAILURE; - ret = blkdev_issue_write_same(bdev, + ret = blkdev_issue_zeroout(bdev, target_to_linux_sector(dev, cmd->t_task_lba), target_to_linux_sector(dev, sbc_get_write_same_sectors(cmd)), - GFP_KERNEL, page ? page : sg_page(sg)); - if (page) - __free_page(page); + GFP_KERNEL, false); if (ret) return TCM_LOGICAL_UNIT_COMMUNICATION_FAILURE; @@ -472,8 +480,10 @@ static void iblock_end_io_flush(struct bio *bio) return TCM_INVALID_CDB_FIELD; } - if (bdev_write_same(bdev)) - return iblock_execute_write_same_direct(bdev, cmd); + if (bdev_write_zeroes_sectors(bdev)) { + if (!iblock_execute_zero_out(bdev, cmd)) + return 0; + } ibr = kzalloc(sizeof(struct iblock_req), GFP_KERNEL); if (!ibr)