Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by smtp.lore.kernel.org (Postfix) with ESMTP id 6A4B6C433FE for ; Thu, 6 Jan 2022 21:55:06 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S244971AbiAFVzF (ORCPT ); Thu, 6 Jan 2022 16:55:05 -0500 Received: from smtp07.smtpout.orange.fr ([80.12.242.129]:52910 "EHLO smtp.smtpout.orange.fr" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S245072AbiAFVyo (ORCPT ); Thu, 6 Jan 2022 16:54:44 -0500 Received: from pop-os.home ([90.11.185.88]) by smtp.orange.fr with ESMTPA id 5aiVntWht2lVY5aiWnSlhj; Thu, 06 Jan 2022 22:54:42 +0100 X-ME-Helo: pop-os.home X-ME-Auth: YWZlNiIxYWMyZDliZWIzOTcwYTEyYzlhMmU3ZiQ1M2U2MzfzZDfyZTMxZTBkMTYyNDBjNDJlZmQ3ZQ== X-ME-Date: Thu, 06 Jan 2022 22:54:42 +0100 X-ME-IP: 90.11.185.88 From: Christophe JAILLET To: arnd@arndb.de, hch@infradead.org, akpm@linux-foundation.org, sathya.prakash@broadcom.com, sreekanth.reddy@broadcom.com, suganath-prabu.subramani@broadcom.com Cc: MPT-FusionLinux.pdl@broadcom.com, linux-scsi@vger.kernel.org, linux-kernel@vger.kernel.org, kernel-janitors@vger.kernel.org, Christophe JAILLET Subject: [PATCH 15/16] scsi: mptctl: Use dma_alloc_coherent() Date: Thu, 6 Jan 2022 22:54:39 +0100 Message-Id: <516375d6d06114484533baf03aae351306100246.1641500561.git.christophe.jaillet@wanadoo.fr> X-Mailer: git-send-email 2.32.0 In-Reply-To: References: MIME-Version: 1.0 Content-Transfer-Encoding: 8bit Precedence: bulk List-ID: X-Mailing-List: linux-kernel@vger.kernel.org In [1], Christoph Hellwig has proposed to remove the wrappers in include/linux/pci-dma-compat.h. Some reasons why this API should be removed have been given by Julia Lawall in [2]. When memory is allocated in kbuf_alloc_2_sgl() GFP_KERNEL can be used because this function already uses the GFP_USER flag for some memory allocation and not spin_lock is taken in the between. When memory is allocated in mptctl_do_mpt_command() GFP_KERNEL can be used because this function already uses copy_from_user() and this function can sleep. When memory is allocated in mptctl_hp_hostinfo() GFP_KERNEL can be used because this function already uses mpt_config() and this function has an explicit might_sleep(). When memory is allocated in mptctl_hp_targetinfo() GFP_KERNEL can be used because this function already uses mpt_config() and this function has an explicit might_sleep(). [1]: https://lore.kernel.org/kernel-janitors/20200421081257.GA131897@infradead.org/ [2]: https://lore.kernel.org/kernel-janitors/alpine.DEB.2.22.394.2007120902170.2424@hadrien/ Signed-off-by: Christophe JAILLET --- I'm a bit unsure about the use of GFP_KERNEL in kbuf_alloc_2_sgl(). In all conversion that I've done, GFP_USER was never used. I don't fully understand the difference between GFP_USER and GFP_KERNEL. --- drivers/message/fusion/mptctl.c | 31 +++++++++++++++++++------------ 1 file changed, 19 insertions(+), 12 deletions(-) diff --git a/drivers/message/fusion/mptctl.c b/drivers/message/fusion/mptctl.c index 0f447179e3f5..03c8fb1795c2 100644 --- a/drivers/message/fusion/mptctl.c +++ b/drivers/message/fusion/mptctl.c @@ -1041,7 +1041,8 @@ kbuf_alloc_2_sgl(int bytes, u32 sgdir, int sge_offset, int *frags, * copying the data in this array into the correct place in the * request and chain buffers. */ - sglbuf = pci_alloc_consistent(ioc->pcidev, MAX_SGL_BYTES, sglbuf_dma); + sglbuf = dma_alloc_coherent(&ioc->pcidev->dev, MAX_SGL_BYTES, + sglbuf_dma, GFP_KERNEL); if (sglbuf == NULL) goto free_and_fail; @@ -1062,9 +1063,9 @@ kbuf_alloc_2_sgl(int bytes, u32 sgdir, int sge_offset, int *frags, while (bytes_allocd < bytes) { this_alloc = min(alloc_sz, bytes-bytes_allocd); buflist[buflist_ent].len = this_alloc; - buflist[buflist_ent].kptr = pci_alloc_consistent(ioc->pcidev, - this_alloc, - &pa); + buflist[buflist_ent].kptr = dma_alloc_coherent(&ioc->pcidev->dev, + this_alloc, + &pa, GFP_KERNEL); if (buflist[buflist_ent].kptr == NULL) { alloc_sz = alloc_sz / 2; if (alloc_sz == 0) { @@ -2105,8 +2106,9 @@ mptctl_do_mpt_command (MPT_ADAPTER *ioc, struct mpt_ioctl_command karg, void __u } flagsLength |= karg.dataOutSize; bufOut.len = karg.dataOutSize; - bufOut.kptr = pci_alloc_consistent( - ioc->pcidev, bufOut.len, &dma_addr_out); + bufOut.kptr = dma_alloc_coherent(&ioc->pcidev->dev, + bufOut.len, + &dma_addr_out, GFP_KERNEL); if (bufOut.kptr == NULL) { rc = -ENOMEM; @@ -2139,8 +2141,9 @@ mptctl_do_mpt_command (MPT_ADAPTER *ioc, struct mpt_ioctl_command karg, void __u flagsLength |= karg.dataInSize; bufIn.len = karg.dataInSize; - bufIn.kptr = pci_alloc_consistent(ioc->pcidev, - bufIn.len, &dma_addr_in); + bufIn.kptr = dma_alloc_coherent(&ioc->pcidev->dev, + bufIn.len, + &dma_addr_in, GFP_KERNEL); if (bufIn.kptr == NULL) { rc = -ENOMEM; @@ -2400,7 +2403,9 @@ mptctl_hp_hostinfo(MPT_ADAPTER *ioc, unsigned long arg, unsigned int data_size) /* Issue the second config page request */ cfg.action = MPI_CONFIG_ACTION_PAGE_READ_CURRENT; - pbuf = pci_alloc_consistent(ioc->pcidev, hdr.PageLength * 4, &buf_dma); + pbuf = dma_alloc_coherent(&ioc->pcidev->dev, + hdr.PageLength * 4, + &buf_dma, GFP_KERNEL); if (pbuf) { cfg.physAddr = buf_dma; if (mpt_config(ioc, &cfg) == 0) { @@ -2477,7 +2482,7 @@ mptctl_hp_hostinfo(MPT_ADAPTER *ioc, unsigned long arg, unsigned int data_size) else IstwiRWRequest->DeviceAddr = 0xB0; - pbuf = pci_alloc_consistent(ioc->pcidev, 4, &buf_dma); + pbuf = dma_alloc_coherent(&ioc->pcidev->dev, 4, &buf_dma, GFP_KERNEL); if (!pbuf) goto out; ioc->add_sge((char *)&IstwiRWRequest->SGL, @@ -2592,7 +2597,8 @@ mptctl_hp_targetinfo(MPT_ADAPTER *ioc, unsigned long arg) /* Get the data transfer speeds */ data_sz = ioc->spi_data.sdp0length * 4; - pg0_alloc = pci_alloc_consistent(ioc->pcidev, data_sz, &page_dma); + pg0_alloc = dma_alloc_coherent(&ioc->pcidev->dev, data_sz, &page_dma, + GFP_KERNEL); if (pg0_alloc) { hdr.PageVersion = ioc->spi_data.sdp0version; hdr.PageLength = data_sz; @@ -2657,7 +2663,8 @@ mptctl_hp_targetinfo(MPT_ADAPTER *ioc, unsigned long arg) /* Issue the second config page request */ cfg.action = MPI_CONFIG_ACTION_PAGE_READ_CURRENT; data_sz = (int) cfg.cfghdr.hdr->PageLength * 4; - pg3_alloc = pci_alloc_consistent(ioc->pcidev, data_sz, &page_dma); + pg3_alloc = dma_alloc_coherent(&ioc->pcidev->dev, data_sz, + &page_dma, GFP_KERNEL); if (pg3_alloc) { cfg.physAddr = page_dma; cfg.pageAddr = (karg.hdr.channel << 8) | karg.hdr.id; -- 2.32.0