From patchwork Wed Jan 19 10:23:09 2022 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Ivaylo Dimitrov X-Patchwork-Id: 534420 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 0C153C433EF for ; Wed, 19 Jan 2022 10:23:46 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1353515AbiASKXp (ORCPT ); Wed, 19 Jan 2022 05:23:45 -0500 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:45320 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1345318AbiASKXo (ORCPT ); Wed, 19 Jan 2022 05:23:44 -0500 Received: from mail-ed1-x52b.google.com (mail-ed1-x52b.google.com [IPv6:2a00:1450:4864:20::52b]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id 93131C061574; Wed, 19 Jan 2022 02:23:43 -0800 (PST) Received: by mail-ed1-x52b.google.com with SMTP id c71so8977154edf.6; Wed, 19 Jan 2022 02:23:43 -0800 (PST) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20210112; h=from:to:cc:subject:date:message-id:in-reply-to:references; bh=lFHQBay/q6+NnOl2SII/Ps5xmFarxwMCV5ET8Tr+2Ks=; b=l6aA2GXwD+dROz0YFw7ehLY4dgmbv+40TX9Ts7qiofcKfAz1LVs0R3DSVj2z1JgvrI UTJYZGGFgMga7wQ7dNkAuQ28hO/tbaI25Iom3JFsi5NJV6vJKBchKpoAdJ1kSh3Sa4sQ 1xC9a22YkHOilOyROJcXMRvr0SpynVcs42gKIiD35iZn3XEjjpkyGKhdS5xUqKSHeOZr SOyxQrHyxMc2Srp/Oa8yMQj8Zfb7gwRDiVRFJDh2zWdAWVp8VEEXb09XRX+UxHdwllWt 0wGpb2TGO8/vDZrYoKQDorxodoQqpbl8pn0WJs5BtEvipjuuGG/2wdUrIeBkhAjEFouW rU3Q== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20210112; h=x-gm-message-state:from:to:cc:subject:date:message-id:in-reply-to :references; bh=lFHQBay/q6+NnOl2SII/Ps5xmFarxwMCV5ET8Tr+2Ks=; b=1UTIrMDJX5TyZDA0TntakoAfaYFDXiVWfd1AeFn6ivjznjqQUd7K44Ex+bjM/fE1ue 3v0411An1f3mFjNdypU0kHJg80lavKu2l3kfK7T5S7uxFmYxoxHv/wYXqCGkrUcPNB+L WRyF3xSlvCvUUYlV3GKf5xJ3PtBME6i500Djwgnz47QvSb+/FLQ8WOEBB+TvRSvRZFBk V6VQWPw42Lk4x7Zcq9Z39F8IJMvlmAAdM3i1SE+92IFIov1N5inBLRV22YfOMT5Bfk3Q XUlHg4f74mLC7bhFq8gjiVWNeMGuE4N5h7yXRmYjVXMMUvEe//sWH3h6enHW2NZkFGFS rkTQ== X-Gm-Message-State: AOAM530jbG0EOpOBket4OcgVuIFDAVSaOyi4YOjUZYb4KeFNy38vGUjC Uo10VLyWKUJwYwF33YI0/eU= X-Google-Smtp-Source: ABdhPJy5WBJKHzb6nGWtktk0x3HF73uuNdLuGC6azC1gyuoMx0eXFpasQ2c/ScKtGXV4Kkoc8E5bcA== X-Received: by 2002:a17:906:9750:: with SMTP id o16mr23715220ejy.410.1642587822170; Wed, 19 Jan 2022 02:23:42 -0800 (PST) Received: from localhost.localdomain ([46.249.74.23]) by smtp.gmail.com with ESMTPSA id b4sm3456546ejb.131.2022.01.19.02.23.40 (version=TLS1_2 cipher=ECDHE-ECDSA-AES128-GCM-SHA256 bits=128/128); Wed, 19 Jan 2022 02:23:41 -0800 (PST) From: Ivaylo Dimitrov To: tomba@kernel.org, airlied@linux.ie, daniel@ffwll.ch Cc: dri-devel@lists.freedesktop.org, linux-kernel@vger.kernel.org, linux-omap@vger.kernel.org, merlijn@wizzup.org, tony@atomide.com, Ivaylo Dimitrov Subject: [PATCH 1/3] drm: omapdrm: simplify omap_gem_pin Date: Wed, 19 Jan 2022 12:23:09 +0200 Message-Id: <1642587791-13222-2-git-send-email-ivo.g.dimitrov.75@gmail.com> X-Mailer: git-send-email 1.9.1 In-Reply-To: <1642587791-13222-1-git-send-email-ivo.g.dimitrov.75@gmail.com> References: <1642587791-13222-1-git-send-email-ivo.g.dimitrov.75@gmail.com> Precedence: bulk List-ID: X-Mailing-List: linux-omap@vger.kernel.org Move tiler related code to its own function. Signed-off-by: Ivaylo Dimitrov --- drivers/gpu/drm/omapdrm/omap_gem.c | 75 +++++++++++++++++++++----------------- 1 file changed, 42 insertions(+), 33 deletions(-) diff --git a/drivers/gpu/drm/omapdrm/omap_gem.c b/drivers/gpu/drm/omapdrm/omap_gem.c index b0fa174..bb12cb4 100644 --- a/drivers/gpu/drm/omapdrm/omap_gem.c +++ b/drivers/gpu/drm/omapdrm/omap_gem.c @@ -750,6 +750,46 @@ void omap_gem_dma_sync_buffer(struct drm_gem_object *obj, } } +static int omap_gem_pin_tiler(struct drm_gem_object *obj) +{ + struct omap_gem_object *omap_obj = to_omap_bo(obj); + u32 npages = obj->size >> PAGE_SHIFT; + enum tiler_fmt fmt = gem2fmt(omap_obj->flags); + struct tiler_block *block; + int ret; + + BUG_ON(omap_obj->block); + + if (omap_obj->flags & OMAP_BO_TILED_MASK) { + block = tiler_reserve_2d(fmt, omap_obj->width, omap_obj->height, + PAGE_SIZE); + } else { + block = tiler_reserve_1d(obj->size); + } + + if (IS_ERR(block)) { + ret = PTR_ERR(block); + dev_err(obj->dev->dev, "could not remap: %d (%d)\n", ret, fmt); + goto fail; + } + + /* TODO: enable async refill.. */ + ret = tiler_pin(block, omap_obj->pages, npages, omap_obj->roll, true); + if (ret) { + tiler_release(block); + dev_err(obj->dev->dev, "could not pin: %d\n", ret); + goto fail; + } + + omap_obj->dma_addr = tiler_ssptr(block); + omap_obj->block = block; + + DBG("got dma address: %pad", &omap_obj->dma_addr); + +fail: + return ret; +} + /** * omap_gem_pin() - Pin a GEM object in memory * @obj: the GEM object @@ -774,11 +814,6 @@ int omap_gem_pin(struct drm_gem_object *obj, dma_addr_t *dma_addr) if (!omap_gem_is_contiguous(omap_obj) && priv->has_dmm) { if (refcount_read(&omap_obj->dma_addr_cnt) == 0) { - u32 npages = obj->size >> PAGE_SHIFT; - enum tiler_fmt fmt = gem2fmt(omap_obj->flags); - struct tiler_block *block; - - BUG_ON(omap_obj->block); refcount_set(&omap_obj->dma_addr_cnt, 1); @@ -786,35 +821,9 @@ int omap_gem_pin(struct drm_gem_object *obj, dma_addr_t *dma_addr) if (ret) goto fail; - if (omap_obj->flags & OMAP_BO_TILED_MASK) { - block = tiler_reserve_2d(fmt, - omap_obj->width, - omap_obj->height, PAGE_SIZE); - } else { - block = tiler_reserve_1d(obj->size); - } - - if (IS_ERR(block)) { - ret = PTR_ERR(block); - dev_err(obj->dev->dev, - "could not remap: %d (%d)\n", ret, fmt); - goto fail; - } - - /* TODO: enable async refill.. */ - ret = tiler_pin(block, omap_obj->pages, npages, - omap_obj->roll, true); - if (ret) { - tiler_release(block); - dev_err(obj->dev->dev, - "could not pin: %d\n", ret); + ret = omap_gem_pin_tiler(obj); + if (ret) goto fail; - } - - omap_obj->dma_addr = tiler_ssptr(block); - omap_obj->block = block; - - DBG("got dma address: %pad", &omap_obj->dma_addr); } else { refcount_inc(&omap_obj->dma_addr_cnt); } From patchwork Wed Jan 19 10:23:10 2022 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Ivaylo Dimitrov X-Patchwork-Id: 533410 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 0B077C433EF for ; Wed, 19 Jan 2022 10:24:04 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1353476AbiASKYC (ORCPT ); Wed, 19 Jan 2022 05:24:02 -0500 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:45410 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1353465AbiASKYB (ORCPT ); Wed, 19 Jan 2022 05:24:01 -0500 Received: from mail-ed1-x534.google.com (mail-ed1-x534.google.com [IPv6:2a00:1450:4864:20::534]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id E5A60C061574; Wed, 19 Jan 2022 02:24:00 -0800 (PST) Received: by mail-ed1-x534.google.com with SMTP id j23so3810142edp.5; Wed, 19 Jan 2022 02:24:00 -0800 (PST) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20210112; h=from:to:cc:subject:date:message-id:in-reply-to:references; bh=j85uReZLdngP6o4mjqty+eKiSWm20pfyWWjzHaNzc+A=; b=pEGWeb+EOPIAgc6ndg1ZtvNKOazZM4TD1gbawrYuBLHVEx5e2PjoQl2TpN1g4J2qkm kPSRei7R1s+OIYypcfibxY3f/3sWYh9eTFu0JCjgisFu3TDFL/1AtMsbbwrhmjf8FLX0 LoRl0sLT3DXZc4mR+KQYq86rUGXX0I7a3obeyANZ9aBkmM/FOGeUqLEusc4pyFhYHU6B SbyR1b+RYRyHgBR6PdbQkSBSBBpKxj2PspkEQI4PV/X0uovvC0k2aPzkhPl3K8vgwA8B Uj7k+DbUPkIyqpg57QvRHPfPg0mpYHUXRFLfuB/ykuZc63QgWotWaoYUrSqfgzYj5BnS KMxA== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20210112; h=x-gm-message-state:from:to:cc:subject:date:message-id:in-reply-to :references; bh=j85uReZLdngP6o4mjqty+eKiSWm20pfyWWjzHaNzc+A=; b=Lrydy0LRPOAkwE+UfsLCwNVpe9h0vFBwG+WudJCkYNQJDm+3riwA5kDQPWWdLoD3cV yPvR4E7pqks+3A3Mg7ZPyxpzQQBrkWSh3wgQkeZwBUWOOYojahMlzHJw69fkqNG6SgS7 kdTila8kO/B3bRKd2jQVoRwOhepP+go9+l7Ept9ThAcMcg3zT7OHhVnNp5BC+onq+yzW ZTmQmVH8rBIqFpK0YpYurut+M5iUFFB3lXQPYhU6TWX+aTnX0XXjiRl+qD34gCdXIMAe 8WM9Bh/ROw3Tuo2Cse83jzxPhq56r6v3LGxihJ+vGNwgzu9QX9AO1bYYeaDA5Gy8E424 jCuw== X-Gm-Message-State: AOAM532rCbAXbR2Y2yVhvSosJUi6JW8s8YkwMdhcswWgEmVxRiTsb7mx 9iTUoUtx60AJyeIjHJDQuaU= X-Google-Smtp-Source: ABdhPJwb4d8yF6xUuXjLf7nRtYbQUWtSD1EJVjuQbQvcZQssNV7b4oeXHadqFsUUsVxAhz7nIS9LmA== X-Received: by 2002:a17:907:970d:: with SMTP id jg13mr22955093ejc.121.1642587839539; Wed, 19 Jan 2022 02:23:59 -0800 (PST) Received: from localhost.localdomain ([46.249.74.23]) by smtp.gmail.com with ESMTPSA id b4sm3456546ejb.131.2022.01.19.02.23.42 (version=TLS1_2 cipher=ECDHE-ECDSA-AES128-GCM-SHA256 bits=128/128); Wed, 19 Jan 2022 02:23:59 -0800 (PST) From: Ivaylo Dimitrov To: tomba@kernel.org, airlied@linux.ie, daniel@ffwll.ch Cc: dri-devel@lists.freedesktop.org, linux-kernel@vger.kernel.org, linux-omap@vger.kernel.org, merlijn@wizzup.org, tony@atomide.com, Ivaylo Dimitrov Subject: [PATCH 2/3] drm: omapdrm: Support exporting of non-contiguous GEM BOs Date: Wed, 19 Jan 2022 12:23:10 +0200 Message-Id: <1642587791-13222-3-git-send-email-ivo.g.dimitrov.75@gmail.com> X-Mailer: git-send-email 1.9.1 In-Reply-To: <1642587791-13222-1-git-send-email-ivo.g.dimitrov.75@gmail.com> References: <1642587791-13222-1-git-send-email-ivo.g.dimitrov.75@gmail.com> Precedence: bulk List-ID: X-Mailing-List: linux-omap@vger.kernel.org Currently code allocates non-scanout BOs from SHMEM and those objects are accessible to userspace by mmap(). However, on devices with no DMM (like OMAP3), the same objects are not accessible by kernel drivers that want to render to them as code refuses to export them. In turn this means that on devices with no DMM, all buffers must be allocated as scanout, otherwise only CPU can access them. On those devices, scanout buffers are allocated from CMA, making those allocations highly unreliable. Fix that by implementing functionality to export SHMEM backed buffers on devices with no DMM. This makes CMA memory only being used when needed, instead for every buffer that has to be off-CPU rendered. Tested on Motorola Droid4 and Nokia N900 Signed-off-by: Ivaylo Dimitrov --- drivers/gpu/drm/omapdrm/omap_gem.c | 125 +++++++++++++++++------------- drivers/gpu/drm/omapdrm/omap_gem.h | 3 +- drivers/gpu/drm/omapdrm/omap_gem_dmabuf.c | 5 +- 3 files changed, 73 insertions(+), 60 deletions(-) diff --git a/drivers/gpu/drm/omapdrm/omap_gem.c b/drivers/gpu/drm/omapdrm/omap_gem.c index bb12cb4..41c1a6d 100644 --- a/drivers/gpu/drm/omapdrm/omap_gem.c +++ b/drivers/gpu/drm/omapdrm/omap_gem.c @@ -38,7 +38,7 @@ struct omap_gem_object { /** roll applied when mapping to DMM */ u32 roll; - /** protects dma_addr_cnt, block, pages, dma_addrs and vaddr */ + /** protects pin_cnt, block, pages, dma_addrs and vaddr */ struct mutex lock; /** @@ -50,24 +50,24 @@ struct omap_gem_object { * - buffers imported from dmabuf (with the OMAP_BO_MEM_DMABUF flag set) * if they are physically contiguous (when sgt->orig_nents == 1) * - * - buffers mapped through the TILER when dma_addr_cnt is not zero, in - * which case the DMA address points to the TILER aperture + * - buffers mapped through the TILER when pin_cnt is not zero, in which + * case the DMA address points to the TILER aperture * * Physically contiguous buffers have their DMA address equal to the * physical address as we don't remap those buffers through the TILER. * * Buffers mapped to the TILER have their DMA address pointing to the - * TILER aperture. As TILER mappings are refcounted (through - * dma_addr_cnt) the DMA address must be accessed through omap_gem_pin() - * to ensure that the mapping won't disappear unexpectedly. References - * must be released with omap_gem_unpin(). + * TILER aperture. As TILER mappings are refcounted (through pin_cnt) + * the DMA address must be accessed through omap_gem_pin() to ensure + * that the mapping won't disappear unexpectedly. References must be + * released with omap_gem_unpin(). */ dma_addr_t dma_addr; /** - * # of users of dma_addr + * # of users */ - refcount_t dma_addr_cnt; + refcount_t pin_cnt; /** * If the buffer has been imported from a dmabuf the OMAP_DB_DMABUF flag @@ -812,32 +812,28 @@ int omap_gem_pin(struct drm_gem_object *obj, dma_addr_t *dma_addr) mutex_lock(&omap_obj->lock); - if (!omap_gem_is_contiguous(omap_obj) && priv->has_dmm) { - if (refcount_read(&omap_obj->dma_addr_cnt) == 0) { + if (!omap_gem_is_contiguous(omap_obj)) { + if (refcount_read(&omap_obj->pin_cnt) == 0) { - refcount_set(&omap_obj->dma_addr_cnt, 1); + refcount_set(&omap_obj->pin_cnt, 1); ret = omap_gem_attach_pages(obj); if (ret) goto fail; - ret = omap_gem_pin_tiler(obj); - if (ret) - goto fail; + if (priv->has_dmm) { + ret = omap_gem_pin_tiler(obj); + if (ret) + goto fail; + } } else { - refcount_inc(&omap_obj->dma_addr_cnt); + refcount_inc(&omap_obj->pin_cnt); } - - if (dma_addr) - *dma_addr = omap_obj->dma_addr; - } else if (omap_gem_is_contiguous(omap_obj)) { - if (dma_addr) - *dma_addr = omap_obj->dma_addr; - } else { - ret = -EINVAL; - goto fail; } + if (dma_addr) + *dma_addr = omap_obj->dma_addr; + fail: mutex_unlock(&omap_obj->lock); @@ -856,27 +852,29 @@ static void omap_gem_unpin_locked(struct drm_gem_object *obj) struct omap_gem_object *omap_obj = to_omap_bo(obj); int ret; - if (omap_gem_is_contiguous(omap_obj) || !priv->has_dmm) + if (omap_gem_is_contiguous(omap_obj)) return; - if (refcount_dec_and_test(&omap_obj->dma_addr_cnt)) { + if (refcount_dec_and_test(&omap_obj->pin_cnt)) { if (omap_obj->sgt) { sg_free_table(omap_obj->sgt); kfree(omap_obj->sgt); omap_obj->sgt = NULL; } - ret = tiler_unpin(omap_obj->block); - if (ret) { - dev_err(obj->dev->dev, - "could not unpin pages: %d\n", ret); - } - ret = tiler_release(omap_obj->block); - if (ret) { - dev_err(obj->dev->dev, - "could not release unmap: %d\n", ret); + if (priv->has_dmm) { + ret = tiler_unpin(omap_obj->block); + if (ret) { + dev_err(obj->dev->dev, + "could not unpin pages: %d\n", ret); + } + ret = tiler_release(omap_obj->block); + if (ret) { + dev_err(obj->dev->dev, + "could not release unmap: %d\n", ret); + } + omap_obj->dma_addr = 0; + omap_obj->block = NULL; } - omap_obj->dma_addr = 0; - omap_obj->block = NULL; } } @@ -909,7 +907,7 @@ int omap_gem_rotated_dma_addr(struct drm_gem_object *obj, u32 orient, mutex_lock(&omap_obj->lock); - if ((refcount_read(&omap_obj->dma_addr_cnt) > 0) && omap_obj->block && + if ((refcount_read(&omap_obj->pin_cnt) > 0) && omap_obj->block && (omap_obj->flags & OMAP_BO_TILED_MASK)) { *dma_addr = tiler_tsptr(omap_obj->block, orient, x, y); ret = 0; @@ -977,7 +975,8 @@ int omap_gem_put_pages(struct drm_gem_object *obj) return 0; } -struct sg_table *omap_gem_get_sg(struct drm_gem_object *obj) +struct sg_table *omap_gem_get_sg(struct drm_gem_object *obj, + enum dma_data_direction dir) { struct omap_gem_object *omap_obj = to_omap_bo(obj); dma_addr_t addr; @@ -1002,28 +1001,44 @@ struct sg_table *omap_gem_get_sg(struct drm_gem_object *obj) goto err_unpin; } - if (omap_obj->flags & OMAP_BO_TILED_MASK) { - enum tiler_fmt fmt = gem2fmt(omap_obj->flags); + if (addr) { + if (omap_obj->flags & OMAP_BO_TILED_MASK) { + enum tiler_fmt fmt = gem2fmt(omap_obj->flags); - len = omap_obj->width << (int)fmt; - count = omap_obj->height; - stride = tiler_stride(fmt, 0); + len = omap_obj->width << (int)fmt; + count = omap_obj->height; + stride = tiler_stride(fmt, 0); + } else { + len = obj->size; + count = 1; + stride = 0; + } } else { - len = obj->size; - count = 1; - stride = 0; + count = obj->size >> PAGE_SHIFT; } ret = sg_alloc_table(sgt, count, GFP_KERNEL); if (ret) goto err_free; - for_each_sg(sgt->sgl, sg, count, i) { - sg_set_page(sg, phys_to_page(addr), len, offset_in_page(addr)); - sg_dma_address(sg) = addr; - sg_dma_len(sg) = len; + /* this must be after omap_gem_pin() to ensure we have pages attached */ + omap_gem_dma_sync_buffer(obj, dir); + + if (addr) { + for_each_sg(sgt->sgl, sg, count, i) { + sg_set_page(sg, phys_to_page(addr), len, + offset_in_page(addr)); + sg_dma_address(sg) = addr; + sg_dma_len(sg) = len; - addr += stride; + addr += stride; + } + } else { + for_each_sg(sgt->sgl, sg, count, i) { + sg_set_page(sg, omap_obj->pages[i], PAGE_SIZE, 0); + sg_dma_address(sg) = omap_obj->dma_addrs[i]; + sg_dma_len(sg) = PAGE_SIZE; + } } omap_obj->sgt = sgt; @@ -1133,7 +1148,7 @@ void omap_gem_describe(struct drm_gem_object *obj, struct seq_file *m) seq_printf(m, "%08x: %2d (%2d) %08llx %pad (%2d) %p %4d", omap_obj->flags, obj->name, kref_read(&obj->refcount), off, &omap_obj->dma_addr, - refcount_read(&omap_obj->dma_addr_cnt), + refcount_read(&omap_obj->pin_cnt), omap_obj->vaddr, omap_obj->roll); if (omap_obj->flags & OMAP_BO_TILED_MASK) { @@ -1196,7 +1211,7 @@ static void omap_gem_free_object(struct drm_gem_object *obj) mutex_lock(&omap_obj->lock); /* The object should not be pinned. */ - WARN_ON(refcount_read(&omap_obj->dma_addr_cnt) > 0); + WARN_ON(refcount_read(&omap_obj->pin_cnt) > 0); if (omap_obj->pages) { if (omap_obj->flags & OMAP_BO_MEM_DMABUF) diff --git a/drivers/gpu/drm/omapdrm/omap_gem.h b/drivers/gpu/drm/omapdrm/omap_gem.h index 19209e3..4d44889 100644 --- a/drivers/gpu/drm/omapdrm/omap_gem.h +++ b/drivers/gpu/drm/omapdrm/omap_gem.h @@ -82,7 +82,8 @@ int omap_gem_get_pages(struct drm_gem_object *obj, struct page ***pages, int omap_gem_rotated_dma_addr(struct drm_gem_object *obj, u32 orient, int x, int y, dma_addr_t *dma_addr); int omap_gem_tiled_stride(struct drm_gem_object *obj, u32 orient); -struct sg_table *omap_gem_get_sg(struct drm_gem_object *obj); +struct sg_table *omap_gem_get_sg(struct drm_gem_object *obj, + enum dma_data_direction dir); void omap_gem_put_sg(struct drm_gem_object *obj, struct sg_table *sgt); #endif /* __OMAPDRM_GEM_H__ */ diff --git a/drivers/gpu/drm/omapdrm/omap_gem_dmabuf.c b/drivers/gpu/drm/omapdrm/omap_gem_dmabuf.c index 57af3d9..8d15e07 100644 --- a/drivers/gpu/drm/omapdrm/omap_gem_dmabuf.c +++ b/drivers/gpu/drm/omapdrm/omap_gem_dmabuf.c @@ -23,13 +23,10 @@ static struct sg_table *omap_gem_map_dma_buf( { struct drm_gem_object *obj = attachment->dmabuf->priv; struct sg_table *sg; - sg = omap_gem_get_sg(obj); + sg = omap_gem_get_sg(obj, dir); if (IS_ERR(sg)) return sg; - /* this must be after omap_gem_pin() to ensure we have pages attached */ - omap_gem_dma_sync_buffer(obj, dir); - return sg; } From patchwork Wed Jan 19 10:23:11 2022 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Ivaylo Dimitrov X-Patchwork-Id: 534419 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 9ADD7C433FE for ; Wed, 19 Jan 2022 10:24:06 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1353465AbiASKYF (ORCPT ); Wed, 19 Jan 2022 05:24:05 -0500 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:45420 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1353616AbiASKYE (ORCPT ); Wed, 19 Jan 2022 05:24:04 -0500 Received: from mail-ed1-x52f.google.com (mail-ed1-x52f.google.com [IPv6:2a00:1450:4864:20::52f]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id 6DC39C061574; Wed, 19 Jan 2022 02:24:03 -0800 (PST) Received: by mail-ed1-x52f.google.com with SMTP id j2so8902592edj.8; Wed, 19 Jan 2022 02:24:03 -0800 (PST) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20210112; h=from:to:cc:subject:date:message-id:in-reply-to:references; bh=nkuGN1Vnj60Rd4SqRXwo3p9yx1/wd6axZJItw3JdV2k=; b=QjRCQfkITJNswVCv4WP2utOD+TASiUHrDW4E/0BRSnF2bndoxWm9TG7jcmj/iP//Jb oVd1JUi4DlDUDjqrpuBGtr+QnMgOSRewo+UGFuTjxsuRmO0xvC7it630J6+581UJHj0c B+km9Wy2X0IQisS1zYMi19G0R6OgKtLBcXeHevadPO+fdhPpsmG1vr97G6PSzPewN10W o800vy+MSF1xeb2CrXlTkwWByGu+cn3mrn0vu0TWG84KO2/KWJ/kvWhudeihpiRo+H/u BZlRWMln1cqce3hISefkAA1owntSY+M/KxMg01OFdqub83UzFjNyRlBZUE5eRtHNlVRw fjsA== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20210112; h=x-gm-message-state:from:to:cc:subject:date:message-id:in-reply-to :references; bh=nkuGN1Vnj60Rd4SqRXwo3p9yx1/wd6axZJItw3JdV2k=; b=zw2dOgBQcqeokwmy6RsmJA1imGuq1H2j0/g9qdBc+KqGAni8V2wxXKQhR+GjK4kbk+ xhSk6wKMpnn739Lmf7RvNmedUNdhzNsplJPZ6J5fr7Jn7E4c3Oebeh75IDs2PATvP5zF iKuruwmW7TAGtW5EAg37DIs/urJEc2QD080mubGkBV/Qmimy3lmdg8Gw1YceLK67Fg9r c+NfmdbUGk/2qcBq4ZG+H79onYSnBzWd25lLDf6GaWcPEJzxOTJsmd7OVWU8BCpojKRj U9Uu3IyX1NwUaBtnOCwjMUOvZpH0fX+X20BMkiI6KL+3vpGXnQijZx92BjmIqt6J9es6 Pknw== X-Gm-Message-State: AOAM533oIM5uN3I/36BQaXFR2J40567AcCQa8/RC1MzgMdv7jEsQJP7d CUHP0nAI18ZKLdUQ8KSl4n7YnAYLWSc= X-Google-Smtp-Source: ABdhPJzz7piW2gj8TZNGzs1Q3d9+NIMTM4qEqQxe1jsFAfModjFtnvfuwrnq6rZOqzTvAH4KspNzLA== X-Received: by 2002:a17:906:950e:: with SMTP id u14mr1062606ejx.550.1642587841272; Wed, 19 Jan 2022 02:24:01 -0800 (PST) Received: from localhost.localdomain ([46.249.74.23]) by smtp.gmail.com with ESMTPSA id b4sm3456546ejb.131.2022.01.19.02.23.59 (version=TLS1_2 cipher=ECDHE-ECDSA-AES128-GCM-SHA256 bits=128/128); Wed, 19 Jan 2022 02:24:00 -0800 (PST) From: Ivaylo Dimitrov To: tomba@kernel.org, airlied@linux.ie, daniel@ffwll.ch Cc: dri-devel@lists.freedesktop.org, linux-kernel@vger.kernel.org, linux-omap@vger.kernel.org, merlijn@wizzup.org, tony@atomide.com, Ivaylo Dimitrov Subject: [PATCH 3/3] drm: omapdrm: Do no allocate non-scanout GEMs through DMM/TILER Date: Wed, 19 Jan 2022 12:23:11 +0200 Message-Id: <1642587791-13222-4-git-send-email-ivo.g.dimitrov.75@gmail.com> X-Mailer: git-send-email 1.9.1 In-Reply-To: <1642587791-13222-1-git-send-email-ivo.g.dimitrov.75@gmail.com> References: <1642587791-13222-1-git-send-email-ivo.g.dimitrov.75@gmail.com> Precedence: bulk List-ID: X-Mailing-List: linux-omap@vger.kernel.org On devices with DMM, all allocations are done through either DMM or TILER. DMM/TILER being a limited resource means that such allocations will start to fail before actual free memory is exhausted. What is even worse is that with time DMM/TILER space gets fragmented to the point that even if we have enough free DMM/TILER space and free memory, allocation fails because there is no big enough free block in DMM/TILER space. Such failures can be easily observed with OMAP xorg DDX, for example - starting few GUI applications (so buffers for their windows are allocated) and then rotating landscape<->portrait while closing and opening new windows soon results in allocation failures. Fix that by mapping buffers through DMM/TILER only when really needed, like, for scanout buffers. Signed-off-by: Ivaylo Dimitrov --- drivers/gpu/drm/omapdrm/omap_gem.c | 12 ++++++++---- 1 file changed, 8 insertions(+), 4 deletions(-) diff --git a/drivers/gpu/drm/omapdrm/omap_gem.c b/drivers/gpu/drm/omapdrm/omap_gem.c index 41c1a6d..cf57179 100644 --- a/drivers/gpu/drm/omapdrm/omap_gem.c +++ b/drivers/gpu/drm/omapdrm/omap_gem.c @@ -821,10 +821,12 @@ int omap_gem_pin(struct drm_gem_object *obj, dma_addr_t *dma_addr) if (ret) goto fail; - if (priv->has_dmm) { - ret = omap_gem_pin_tiler(obj); - if (ret) - goto fail; + if (omap_obj->flags & OMAP_BO_SCANOUT) { + if (priv->has_dmm) { + ret = omap_gem_pin_tiler(obj); + if (ret) + goto fail; + } } } else { refcount_inc(&omap_obj->pin_cnt); @@ -861,6 +863,8 @@ static void omap_gem_unpin_locked(struct drm_gem_object *obj) kfree(omap_obj->sgt); omap_obj->sgt = NULL; } + if (!(omap_obj->flags & OMAP_BO_SCANOUT)) + return; if (priv->has_dmm) { ret = tiler_unpin(omap_obj->block); if (ret) {