From patchwork Sat May 15 14:53:59 2021 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Paul Cercueil X-Patchwork-Id: 12259891 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-13.7 required=3.0 tests=BAYES_00, HEADER_FROM_DIFFERENT_DOMAINS,INCLUDES_CR_TRAILER,INCLUDES_PATCH, MAILING_LIST_MULTI,SPF_HELO_NONE,SPF_PASS,URIBL_BLOCKED autolearn=unavailable autolearn_force=no version=3.4.0 Received: from mail.kernel.org (mail.kernel.org [198.145.29.99]) by smtp.lore.kernel.org (Postfix) with ESMTP id 5F73EC433ED for ; Sat, 15 May 2021 14:55:05 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by mail.kernel.org (Postfix) with ESMTP id 362F1613C4 for ; Sat, 15 May 2021 14:55:05 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S232343AbhEOO4M (ORCPT ); Sat, 15 May 2021 10:56:12 -0400 Received: from aposti.net ([89.234.176.197]:58420 "EHLO aposti.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S232263AbhEOOz4 (ORCPT ); Sat, 15 May 2021 10:55:56 -0400 From: Paul Cercueil To: Maarten Lankhorst , Maxime Ripard , Thomas Zimmermann , David Airlie , Daniel Vetter Cc: Christoph Hellwig , list@opendingux.net, dri-devel@lists.freedesktop.org, linux-kernel@vger.kernel.org, linux-mips@vger.kernel.org, Paul Cercueil Subject: [PATCH v4 3/3] drm/ingenic: Add option to alloc cached GEM buffers Date: Sat, 15 May 2021 15:53:59 +0100 Message-Id: <20210515145359.64802-4-paul@crapouillou.net> In-Reply-To: <20210515145359.64802-1-paul@crapouillou.net> References: <20210515145359.64802-1-paul@crapouillou.net> MIME-Version: 1.0 Precedence: bulk List-ID: X-Mailing-List: linux-mips@vger.kernel.org Alloc GEM buffers backed by noncoherent memory on SoCs where it is actually faster than write-combine. This dramatically speeds up software rendering on these SoCs, even for tasks where write-combine memory should in theory be faster (e.g. simple blits). v3: The option is now selected per-SoC instead of being a module parameter. Signed-off-by: Paul Cercueil --- drivers/gpu/drm/ingenic/ingenic-drm-drv.c | 56 ++++++++++++++++++++++- drivers/gpu/drm/ingenic/ingenic-ipu.c | 18 ++++++-- 2 files changed, 68 insertions(+), 6 deletions(-) diff --git a/drivers/gpu/drm/ingenic/ingenic-drm-drv.c b/drivers/gpu/drm/ingenic/ingenic-drm-drv.c index 09225b770bb8..5f64e8583eec 100644 --- a/drivers/gpu/drm/ingenic/ingenic-drm-drv.c +++ b/drivers/gpu/drm/ingenic/ingenic-drm-drv.c @@ -9,6 +9,7 @@ #include #include #include +#include #include #include #include @@ -23,6 +24,7 @@ #include #include #include +#include #include #include #include @@ -57,6 +59,7 @@ struct ingenic_dma_hwdescs { struct jz_soc_info { bool needs_dev_clk; bool has_osd; + bool map_noncoherent; unsigned int max_width, max_height; const u32 *formats_f0, *formats_f1; unsigned int num_formats_f0, num_formats_f1; @@ -410,6 +413,8 @@ static int ingenic_drm_plane_atomic_check(struct drm_plane *plane, old_plane_state->fb->format->format != new_plane_state->fb->format->format)) crtc_state->mode_changed = true; + drm_atomic_helper_check_plane_damage(state, new_plane_state); + return 0; } @@ -544,8 +549,8 @@ static void ingenic_drm_plane_atomic_update(struct drm_plane *plane, struct drm_atomic_state *state) { struct ingenic_drm *priv = drm_device_get_priv(plane->dev); - struct drm_plane_state *newstate = drm_atomic_get_new_plane_state(state, - plane); + struct drm_plane_state *newstate = drm_atomic_get_new_plane_state(state, plane); + struct drm_plane_state *oldstate = drm_atomic_get_old_plane_state(state, plane); struct drm_crtc_state *crtc_state; struct ingenic_dma_hwdesc *hwdesc; unsigned int width, height, cpp, offset; @@ -553,6 +558,8 @@ static void ingenic_drm_plane_atomic_update(struct drm_plane *plane, u32 fourcc; if (newstate && newstate->fb) { + drm_gem_cma_sync_data(&priv->drm, oldstate, newstate); + crtc_state = newstate->crtc->state; addr = drm_fb_cma_get_gem_addr(newstate->fb, newstate, 0); @@ -742,6 +749,43 @@ static void ingenic_drm_disable_vblank(struct drm_crtc *crtc) regmap_update_bits(priv->map, JZ_REG_LCD_CTRL, JZ_LCD_CTRL_EOF_IRQ, 0); } +static int ingenic_drm_atomic_helper_dirtyfb(struct drm_framebuffer *fb, + struct drm_file *file_priv, + unsigned int flags, + unsigned int color, + struct drm_clip_rect *clips, + unsigned int num_clips) +{ + struct ingenic_drm *priv = drm_device_get_priv(fb->dev); + + if (!priv->soc_info->map_noncoherent) + return 0; + + return drm_atomic_helper_dirtyfb(fb, file_priv, flags, + color, clips, num_clips); +} + +static const struct drm_framebuffer_funcs ingenic_drm_gem_fb_funcs = { + .destroy = drm_gem_fb_destroy, + .create_handle = drm_gem_fb_create_handle, + .dirty = ingenic_drm_atomic_helper_dirtyfb, +}; + +static struct drm_gem_object * +ingenic_drm_gem_create_object(struct drm_device *drm, size_t size) +{ + struct ingenic_drm *priv = drm_device_get_priv(drm); + struct drm_gem_cma_object *obj; + + obj = kzalloc(sizeof(*obj), GFP_KERNEL); + if (!obj) + return ERR_PTR(-ENOMEM); + + obj->map_noncoherent = priv->soc_info->map_noncoherent; + + return &obj->base; +} + DEFINE_DRM_GEM_CMA_FOPS(ingenic_drm_fops); static const struct drm_driver ingenic_drm_driver_data = { @@ -754,6 +798,7 @@ static const struct drm_driver ingenic_drm_driver_data = { .patchlevel = 0, .fops = &ingenic_drm_fops, + .gem_create_object = ingenic_drm_gem_create_object, DRM_GEM_CMA_DRIVER_OPS, .irq_handler = ingenic_drm_irq_handler, @@ -961,6 +1006,8 @@ static int ingenic_drm_bind(struct device *dev, bool has_components) return ret; } + drm_plane_enable_fb_damage_clips(&priv->f1); + drm_crtc_helper_add(&priv->crtc, &ingenic_drm_crtc_helper_funcs); ret = drm_crtc_init_with_planes(drm, &priv->crtc, primary, @@ -989,6 +1036,8 @@ static int ingenic_drm_bind(struct device *dev, bool has_components) return ret; } + drm_plane_enable_fb_damage_clips(&priv->f0); + if (IS_ENABLED(CONFIG_DRM_INGENIC_IPU) && has_components) { ret = component_bind_all(dev, drm); if (ret) { @@ -1245,6 +1294,7 @@ static const u32 jz4770_formats_f0[] = { static const struct jz_soc_info jz4740_soc_info = { .needs_dev_clk = true, .has_osd = false, + .map_noncoherent = false, .max_width = 800, .max_height = 600, .formats_f1 = jz4740_formats, @@ -1255,6 +1305,7 @@ static const struct jz_soc_info jz4740_soc_info = { static const struct jz_soc_info jz4725b_soc_info = { .needs_dev_clk = false, .has_osd = true, + .map_noncoherent = false, .max_width = 800, .max_height = 600, .formats_f1 = jz4725b_formats_f1, @@ -1266,6 +1317,7 @@ static const struct jz_soc_info jz4725b_soc_info = { static const struct jz_soc_info jz4770_soc_info = { .needs_dev_clk = false, .has_osd = true, + .map_noncoherent = true, .max_width = 1280, .max_height = 720, .formats_f1 = jz4770_formats_f1, diff --git a/drivers/gpu/drm/ingenic/ingenic-ipu.c b/drivers/gpu/drm/ingenic/ingenic-ipu.c index 3b1091e7c0cd..a4d1b500c3ad 100644 --- a/drivers/gpu/drm/ingenic/ingenic-ipu.c +++ b/drivers/gpu/drm/ingenic/ingenic-ipu.c @@ -20,10 +20,13 @@ #include #include +#include #include #include #include #include +#include +#include #include #include #include @@ -285,8 +288,8 @@ static void ingenic_ipu_plane_atomic_update(struct drm_plane *plane, struct drm_atomic_state *state) { struct ingenic_ipu *ipu = plane_to_ingenic_ipu(plane); - struct drm_plane_state *newstate = drm_atomic_get_new_plane_state(state, - plane); + struct drm_plane_state *newstate = drm_atomic_get_new_plane_state(state, plane); + struct drm_plane_state *oldstate = drm_atomic_get_new_plane_state(state, plane); const struct drm_format_info *finfo; u32 ctrl, stride = 0, coef_index = 0, format = 0; bool needs_modeset, upscaling_w, upscaling_h; @@ -317,6 +320,8 @@ static void ingenic_ipu_plane_atomic_update(struct drm_plane *plane, JZ_IPU_CTRL_CHIP_EN | JZ_IPU_CTRL_LCDC_SEL); } + drm_gem_cma_sync_data(ipu->drm, oldstate, newstate); + /* New addresses will be committed in vblank handler... */ ipu->addr_y = drm_fb_cma_get_gem_addr(newstate->fb, newstate, 0); if (finfo->num_planes > 1) @@ -541,7 +546,7 @@ static int ingenic_ipu_plane_atomic_check(struct drm_plane *plane, if (!new_plane_state->crtc || !crtc_state->mode.hdisplay || !crtc_state->mode.vdisplay) - return 0; + goto out_check_damage; /* Plane must be fully visible */ if (new_plane_state->crtc_x < 0 || new_plane_state->crtc_y < 0 || @@ -558,7 +563,7 @@ static int ingenic_ipu_plane_atomic_check(struct drm_plane *plane, return -EINVAL; if (!osd_changed(new_plane_state, old_plane_state)) - return 0; + goto out_check_damage; crtc_state->mode_changed = true; @@ -592,6 +597,9 @@ static int ingenic_ipu_plane_atomic_check(struct drm_plane *plane, ipu->denom_w = denom_w; ipu->denom_h = denom_h; +out_check_damage: + drm_atomic_helper_check_plane_damage(state, new_plane_state); + return 0; } @@ -773,6 +781,8 @@ static int ingenic_ipu_bind(struct device *dev, struct device *master, void *d) return err; } + drm_plane_enable_fb_damage_clips(plane); + /* * Sharpness settings range is [0,32] * 0 : nearest-neighbor