Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1755982AbbGFUgK (ORCPT ); Mon, 6 Jul 2015 16:36:10 -0400 Received: from smtp.codeaurora.org ([198.145.29.96]:49328 "EHLO smtp.codeaurora.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1755553AbbGFUfu (ORCPT ); Mon, 6 Jul 2015 16:35:50 -0400 From: Stephane Viau To: dri-devel@lists.freedesktop.org Cc: linux-arm-msm@vger.kernel.org, linux-kernel@vger.kernel.org, robdclark@gmail.com, Wentao Xu , Stephane Viau Subject: [PATCH 5/6] drm/msm/mdp5: use 2 memory clients for YUV formats on newer mdp5 Date: Mon, 6 Jul 2015 16:35:30 -0400 Message-Id: <1436214931-23527-6-git-send-email-sviau@codeaurora.org> X-Mailer: git-send-email 1.8.2.1 In-Reply-To: <1436214931-23527-1-git-send-email-sviau@codeaurora.org> References: <1436214931-23527-1-git-send-email-sviau@codeaurora.org> Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 4870 Lines: 129 From: Wentao Xu Newer MDP5 uses 2 shared memory pool clients for certain YUV formats. For example, if VIG0 is used to fetch data in YUYV format, it will use VIG0_Y for Y component, and VIG0_Cr for UV packed. Signed-off-by: Wentao Xu [rebase] Signed-off-by: Stephane Viau Change-Id: I24df222372b6cea5e67d2331e2c842c4faba383e --- drivers/gpu/drm/msm/mdp/mdp5/mdp5_plane.c | 2 +- drivers/gpu/drm/msm/mdp/mdp5/mdp5_smp.c | 26 ++++++++++++++++++++++---- drivers/gpu/drm/msm/mdp/mdp5/mdp5_smp.h | 3 ++- 3 files changed, 25 insertions(+), 6 deletions(-) diff --git a/drivers/gpu/drm/msm/mdp/mdp5/mdp5_plane.c b/drivers/gpu/drm/msm/mdp/mdp5/mdp5_plane.c index 622f7de..7fb526d 100644 --- a/drivers/gpu/drm/msm/mdp/mdp5/mdp5_plane.c +++ b/drivers/gpu/drm/msm/mdp/mdp5/mdp5_plane.c @@ -449,7 +449,7 @@ static int mdp5_plane_mode_set(struct drm_plane *plane, /* Request some memory from the SMP: */ ret = mdp5_smp_request(mdp5_kms->smp, - mdp5_plane->pipe, fb->pixel_format, src_w); + mdp5_plane->pipe, format, src_w, false); if (ret) return ret; diff --git a/drivers/gpu/drm/msm/mdp/mdp5/mdp5_smp.c b/drivers/gpu/drm/msm/mdp/mdp5/mdp5_smp.c index 64a27d8..563cca9 100644 --- a/drivers/gpu/drm/msm/mdp/mdp5/mdp5_smp.c +++ b/drivers/gpu/drm/msm/mdp/mdp5/mdp5_smp.c @@ -90,6 +90,8 @@ struct mdp5_smp { struct drm_device *dev; + const struct mdp5_smp_block *cfg; + int blk_cnt; int blk_size; @@ -137,14 +139,12 @@ static int smp_request_block(struct mdp5_smp *smp, u32 cid, int nblks) { struct mdp5_kms *mdp5_kms = get_kms(smp); - const struct mdp5_cfg_hw *hw_cfg; struct mdp5_client_smp_state *ps = &smp->client_state[cid]; int i, ret, avail, cur_nblks, cnt = smp->blk_cnt; int reserved; unsigned long flags; - hw_cfg = mdp5_cfg_get_hw_config(mdp5_kms->cfg); - reserved = hw_cfg->smp.reserved[cid]; + reserved = smp->cfg->reserved[cid]; spin_lock_irqsave(&smp->state_lock, flags); @@ -209,12 +209,14 @@ static void set_fifo_thresholds(struct mdp5_smp *smp, * decimated width. Ie. SMP buffering sits downstream of decimation (which * presumably happens during the dma from scanout buffer). */ -int mdp5_smp_request(struct mdp5_smp *smp, enum mdp5_pipe pipe, u32 fmt, u32 width) +int mdp5_smp_request(struct mdp5_smp *smp, enum mdp5_pipe pipe, + const struct mdp_format *format, u32 width, bool hdecim) { struct mdp5_kms *mdp5_kms = get_kms(smp); struct drm_device *dev = mdp5_kms->dev; int rev = mdp5_cfg_get_hw_rev(mdp5_kms->cfg); int i, hsub, nplanes, nlines, nblks, ret; + u32 fmt = format->base.pixel_format; nplanes = drm_format_num_planes(fmt); hsub = drm_format_horz_chroma_subsampling(fmt); @@ -222,6 +224,21 @@ int mdp5_smp_request(struct mdp5_smp *smp, enum mdp5_pipe pipe, u32 fmt, u32 wid /* different if BWC (compressed framebuffer?) enabled: */ nlines = 2; + /* Newer MDPs have split/packing logic, which fetches sub-sampled + * U and V components (splits them from Y if necessary) and packs + * them together, writes to SMP using a single client. + */ + if ((rev > 0) && (format->chroma_sample > CHROMA_FULL)) { + fmt = DRM_FORMAT_NV24; + nplanes = 2; + + /* if decimation is enabled, HW decimates less on the + * sub sampled chroma components + */ + if (hdecim && (hsub > 1)) + hsub = 1; + } + for (i = 0, nblks = 0; i < nplanes; i++) { int n, fetch_stride, cpp; @@ -388,6 +405,7 @@ struct mdp5_smp *mdp5_smp_init(struct drm_device *dev, const struct mdp5_smp_blo } smp->dev = dev; + smp->cfg = cfg; smp->blk_cnt = cfg->mmb_count; smp->blk_size = cfg->mmb_size; diff --git a/drivers/gpu/drm/msm/mdp/mdp5/mdp5_smp.h b/drivers/gpu/drm/msm/mdp/mdp5/mdp5_smp.h index 5b6c236..20b87e8 100644 --- a/drivers/gpu/drm/msm/mdp/mdp5/mdp5_smp.h +++ b/drivers/gpu/drm/msm/mdp/mdp5/mdp5_smp.h @@ -39,7 +39,8 @@ struct mdp5_smp; struct mdp5_smp *mdp5_smp_init(struct drm_device *dev, const struct mdp5_smp_block *cfg); void mdp5_smp_destroy(struct mdp5_smp *smp); -int mdp5_smp_request(struct mdp5_smp *smp, enum mdp5_pipe pipe, u32 fmt, u32 width); +int mdp5_smp_request(struct mdp5_smp *smp, enum mdp5_pipe pipe, + const struct mdp_format *format, u32 width, bool hdecim); void mdp5_smp_configure(struct mdp5_smp *smp, enum mdp5_pipe pipe); void mdp5_smp_commit(struct mdp5_smp *smp, enum mdp5_pipe pipe); void mdp5_smp_release(struct mdp5_smp *smp, enum mdp5_pipe pipe); -- Qualcomm Innovation Center, Inc. The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/