From patchwork Mon Jul 8 13:07:42 2019 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Lucas Stach X-Patchwork-Id: 57475 X-Patchwork-Delegate: hverkuil@xs4all.nl Received: from vger.kernel.org ([209.132.180.67]) by www.linuxtv.org with esmtp (Exim 4.84_2) (envelope-from ) id 1hkTN3-0006lP-VX; Mon, 08 Jul 2019 13:07:54 +0000 Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1730971AbfGHNHw (ORCPT + 1 other); Mon, 8 Jul 2019 09:07:52 -0400 Received: from metis.ext.pengutronix.de ([85.220.165.71]:44205 "EHLO metis.ext.pengutronix.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1728581AbfGHNHw (ORCPT ); Mon, 8 Jul 2019 09:07:52 -0400 Received: from dude02.hi.pengutronix.de ([2001:67c:670:100:1d::28] helo=dude02.pengutronix.de.) by metis.ext.pengutronix.de with esmtp (Exim 4.92) (envelope-from ) id 1hkTMt-0001TW-Rz; Mon, 08 Jul 2019 15:07:43 +0200 From: Lucas Stach To: Pawel Osciak , Marek Szyprowski , Kyungmin Park , Hans Verkuil Cc: Mauro Carvalho Chehab , linux-media@vger.kernel.org, kernel@pengutronix.de, patchwork-lst@pengutronix.de Subject: [PATCH v2] media: vb2-dc: skip CPU sync in map/unmap dma_buf Date: Mon, 8 Jul 2019 15:07:42 +0200 Message-Id: <20190708130742.11151-1-l.stach@pengutronix.de> X-Mailer: git-send-email 2.20.1 MIME-Version: 1.0 X-SA-Exim-Connect-IP: 2001:67c:670:100:1d::28 X-SA-Exim-Mail-From: l.stach@pengutronix.de X-SA-Exim-Scanned: No (on metis.ext.pengutronix.de); SAEximRunCond expanded to false X-PTX-Original-Recipient: linux-media@vger.kernel.org Sender: linux-media-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-media@vger.kernel.org This is rougly equivalent to ca0e68e21aae (drm/prime: skip CPU sync in map/unmap dma_buf). The contig memory allocated is already device coherent memory, so there is no point in doing a CPU sync when mapping it to another deevice. Also most importers currently cache the mapping so the CPU sync would only happen on the first import, so we are better off with not pretending to do a cache synchronization at all. This gets rid of a lot of CPU overhead in uses where those dma-bufs are regularily imported and detached again, like Weston is currently doing in the DRM compositor. Signed-off-by: Lucas Stach Acked-by: Marek Szyprowski --- v2: Add comments why it is safe to skip the CPU sync. --- .../common/videobuf2/videobuf2-dma-contig.c | 23 +++++++++++++------ 1 file changed, 16 insertions(+), 7 deletions(-) diff --git a/drivers/media/common/videobuf2/videobuf2-dma-contig.c b/drivers/media/common/videobuf2/videobuf2-dma-contig.c index ecbef266130b..1b8f86366290 100644 --- a/drivers/media/common/videobuf2/videobuf2-dma-contig.c +++ b/drivers/media/common/videobuf2/videobuf2-dma-contig.c @@ -267,8 +267,14 @@ static void vb2_dc_dmabuf_ops_detach(struct dma_buf *dbuf, /* release the scatterlist cache */ if (attach->dma_dir != DMA_NONE) - dma_unmap_sg(db_attach->dev, sgt->sgl, sgt->orig_nents, - attach->dma_dir); + /* + * Cache sync can be skipped here, as the vb2_dc memory is + * allocated from device coherent memory, which means the + * memory locations do not require any explicit cache + * maintenance prior or after being used by the device. + */ + dma_unmap_sg_attrs(db_attach->dev, sgt->sgl, sgt->orig_nents, + attach->dma_dir, DMA_ATTR_SKIP_CPU_SYNC); sg_free_table(sgt); kfree(attach); db_attach->priv = NULL; @@ -293,14 +299,17 @@ static struct sg_table *vb2_dc_dmabuf_ops_map( /* release any previous cache */ if (attach->dma_dir != DMA_NONE) { - dma_unmap_sg(db_attach->dev, sgt->sgl, sgt->orig_nents, - attach->dma_dir); + dma_unmap_sg_attrs(db_attach->dev, sgt->sgl, sgt->orig_nents, + attach->dma_dir, DMA_ATTR_SKIP_CPU_SYNC); attach->dma_dir = DMA_NONE; } - /* mapping to the client with new direction */ - sgt->nents = dma_map_sg(db_attach->dev, sgt->sgl, sgt->orig_nents, - dma_dir); + /* + * mapping to the client with new direction, no cache sync + * required see comment in vb2_dc_dmabuf_ops_detach() + */ + sgt->nents = dma_map_sg_attrs(db_attach->dev, sgt->sgl, sgt->orig_nents, + dma_dir, DMA_ATTR_SKIP_CPU_SYNC); if (!sgt->nents) { pr_err("failed to map scatterlist\n"); mutex_unlock(lock);