From nobody Sat Feb 7 07:24:22 2026 Received: from bali.collaboradmins.com (bali.collaboradmins.com [148.251.105.195]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id AAE28358D3A; Wed, 22 Oct 2025 16:26:10 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=148.251.105.195 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1761150374; cv=none; b=OJASR6a2ZdOVCVhoJ/TthHwh5x/aw1PH9MnpNKlNSiNzvo9GOzDQTIUy0IW9Dnim9wkQt8CuSUA1CJxrdfpbdmZoLMcZc/rgk4ahxy6u8zaxo/vdQxqXu16cqT0zXq5EUjlPAe+M3Xfe2xZFiTG75qUUpZiYnzSAmCnIUOnqraA= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1761150374; c=relaxed/simple; bh=KcUtQpTjceRQb92EgPDjZ/AQiUMhFwBADYecoGiPUM0=; h=From:To:Cc:Subject:Date:Message-ID:In-Reply-To:References: MIME-Version; b=OkeynBFlIlg0X9rzte1YASdAxRth15ZtAMF3ZREu4rCvzEVXRKBd44Xpu5tvlYVVRHRLPqAbH4qNGVyAMWB2noLwj19vYqo2op0oCFXw1s5yV7/A90O1TUBwo2E78s2nMDZYJL+eidQ2vacKlJU+EhzMtryNr4zwugZiSq3R794= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=collabora.com; spf=pass smtp.mailfrom=collabora.com; dkim=pass (2048-bit key) header.d=collabora.com header.i=@collabora.com header.b=LfIADKRA; arc=none smtp.client-ip=148.251.105.195 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=collabora.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=collabora.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=collabora.com header.i=@collabora.com header.b="LfIADKRA" DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=collabora.com; s=mail; t=1761150369; bh=KcUtQpTjceRQb92EgPDjZ/AQiUMhFwBADYecoGiPUM0=; h=From:To:Cc:Subject:Date:In-Reply-To:References:From; b=LfIADKRARrkSYxa+QDGszBMLrLKpnMFPc92FSxuDPUQecSvv1NFlHs+DeVmXIcBmT U0MhezLE/ywesS74AaYlBfXaAuUU6A4Lyy3QXUgwR9Rt59AiKraPw51maiR5kqvJdS zK3qKIh7IWIN6SnMTN7j4enSdon99LYwV2hIL9WZB1t7d1J43xQeWrAC259TnN+5pH f/6RiGRg0wnaEY5DnNNA0VRSHxK8yEsUHK+hz8LSB1iGZE5o1V5Q0uonegclOqeArR MY9MSmb0O9/HPCxTQRekQoxhHqVaCQD/lZ/CsC74hzzAoUP27DDGCR4rOhiV1ZIOQh vu7Owaank4e4g== Received: from trenzalore (unknown [23.233.251.139]) (using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 (256/256 bits) key-exchange X25519 server-signature RSA-PSS (4096 bits) server-digest SHA256) (No client certificate requested) (Authenticated sender: detlev) by bali.collaboradmins.com (Postfix) with ESMTPSA id 87D8117E1404; Wed, 22 Oct 2025 18:26:06 +0200 (CEST) From: Detlev Casanova To: linux-kernel@vger.kernel.org Cc: Mauro Carvalho Chehab , Detlev Casanova , Ezequiel Garcia , Heiko Stuebner , Ricardo Ribalda , Hans Verkuil , Hans de Goede , Yunke Cao , Jonathan Corbet , Laurent Pinchart , Sakari Ailus , James Cowgill , linux-media@vger.kernel.org, linux-rockchip@lists.infradead.org, linux-arm-kernel@lists.infradead.org, kernel@collabora.com, Nicolas Dufresne , Diederik de Haas Subject: [PATCH v3 14/15] media: rkvdec: Add HEVC support for the VDPU381 variant Date: Wed, 22 Oct 2025 12:22:13 -0400 Message-ID: <20251022162459.271603-15-detlev.casanova@collabora.com> X-Mailer: git-send-email 2.51.1.dirty In-Reply-To: <20251022162459.271603-1-detlev.casanova@collabora.com> References: <20251022162459.271603-1-detlev.casanova@collabora.com> Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset="utf-8" The VDPU381 supports HEVC decoding up to 7680x4320@30fps. It could double that when using both decoder cores. It support YUV420 (8 and 10 bits) as well as AFBC (not implemented here) The fluster score is 146/147 for JCT-VC-HEVC_V1, tested on ROCK 5B. None of the other test suites works. Tested-by: Diederik de Haas # Rock 5B Signed-off-by: Detlev Casanova --- .../media/platform/rockchip/rkvdec/Kconfig | 1 + .../media/platform/rockchip/rkvdec/Makefile | 1 + .../rockchip/rkvdec/rkvdec-hevc-common.c | 392 +++++++++++- .../rockchip/rkvdec/rkvdec-hevc-common.h | 74 ++- .../rockchip/rkvdec/rkvdec-vdpu381-hevc.c | 596 ++++++++++++++++++ .../media/platform/rockchip/rkvdec/rkvdec.c | 86 ++- .../media/platform/rockchip/rkvdec/rkvdec.h | 3 + 7 files changed, 1114 insertions(+), 39 deletions(-) create mode 100644 drivers/media/platform/rockchip/rkvdec/rkvdec-vdpu381-h= evc.c diff --git a/drivers/media/platform/rockchip/rkvdec/Kconfig b/drivers/media= /platform/rockchip/rkvdec/Kconfig index 5f3bdd848a2c..3303b0ce3280 100644 --- a/drivers/media/platform/rockchip/rkvdec/Kconfig +++ b/drivers/media/platform/rockchip/rkvdec/Kconfig @@ -8,6 +8,7 @@ config VIDEO_ROCKCHIP_VDEC select VIDEOBUF2_VMALLOC select V4L2_MEM2MEM_DEV select V4L2_H264 + select V4L2_HEVC select V4L2_VP9 help Support for the Rockchip Video Decoder IP present on Rockchip SoCs, diff --git a/drivers/media/platform/rockchip/rkvdec/Makefile b/drivers/medi= a/platform/rockchip/rkvdec/Makefile index a58d4aede2fe..e30fdd7d51c3 100644 --- a/drivers/media/platform/rockchip/rkvdec/Makefile +++ b/drivers/media/platform/rockchip/rkvdec/Makefile @@ -9,5 +9,6 @@ rockchip-vdec-y +=3D \ rkvdec-hevc-common.o \ rkvdec-rcb.o \ rkvdec-vdpu381-h264.o \ + rkvdec-vdpu381-hevc.o \ rkvdec-vdpu383-h264.o \ rkvdec-vp9.o diff --git a/drivers/media/platform/rockchip/rkvdec/rkvdec-hevc-common.c b/= drivers/media/platform/rockchip/rkvdec/rkvdec-hevc-common.c index d571107f2242..4b21b3ea67fd 100644 --- a/drivers/media/platform/rockchip/rkvdec/rkvdec-hevc-common.c +++ b/drivers/media/platform/rockchip/rkvdec/rkvdec-hevc-common.c @@ -21,6 +21,143 @@ #include "rkvdec.h" #include "rkvdec-hevc-common.h" =20 +#define RKVDEC_HEVC_MAX_DEPTH_IN_BYTES 2 + +/* Store the Short term ref pic set calculated values */ +struct calculated_rps_st_set { + u8 num_delta_pocs; + u8 num_negative_pics; + u8 num_positive_pics; + u8 used_by_curr_pic_s0[16]; + u8 used_by_curr_pic_s1[16]; + s32 delta_poc_s0[16]; + s32 delta_poc_s1[16]; +}; + +enum rkvdec_image_fmt rkvdec_hevc_get_image_fmt(struct rkvdec_ctx *ctx, + struct v4l2_ctrl *ctrl) +{ + const struct v4l2_ctrl_hevc_sps *sps =3D ctrl->p_new.p_hevc_sps; + + if (ctrl->id !=3D V4L2_CID_STATELESS_HEVC_SPS) + return RKVDEC_IMG_FMT_ANY; + + if (sps->bit_depth_luma_minus8 =3D=3D 0) + return RKVDEC_IMG_FMT_420_8BIT; + else if (sps->bit_depth_luma_minus8 =3D=3D 2) + return RKVDEC_IMG_FMT_420_10BIT; + + return RKVDEC_IMG_FMT_ANY; +} + +void compute_tiles_uniform(struct rkvdec_hevc_run *run, u16 log2_min_cb_si= ze, + u16 width, u16 height, s32 pic_in_cts_width, + s32 pic_in_cts_height, u16 *column_width, u16 *row_height) +{ + const struct v4l2_ctrl_hevc_pps *pps =3D run->pps; + int i; + + for (i =3D 0; i < pps->num_tile_columns_minus1 + 1; i++) + column_width[i] =3D ((i + 1) * pic_in_cts_width) / + (pps->num_tile_columns_minus1 + 1) - + (i * pic_in_cts_width) / + (pps->num_tile_columns_minus1 + 1); + + for (i =3D 0; i < pps->num_tile_rows_minus1 + 1; i++) + row_height[i] =3D ((i + 1) * pic_in_cts_height) / + (pps->num_tile_rows_minus1 + 1) - + (i * pic_in_cts_height) / + (pps->num_tile_rows_minus1 + 1); +} + +void compute_tiles_non_uniform(struct rkvdec_hevc_run *run, u16 log2_min_c= b_size, + u16 width, u16 height, s32 pic_in_cts_width, + s32 pic_in_cts_height, u16 *column_width, u16 *row_height) +{ + const struct v4l2_ctrl_hevc_pps *pps =3D run->pps; + s32 sum =3D 0; + int i; + + for (i =3D 0; i < pps->num_tile_columns_minus1; i++) { + column_width[i] =3D pps->column_width_minus1[i] + 1; + sum +=3D column_width[i]; + } + column_width[i] =3D pic_in_cts_width - sum; + + sum =3D 0; + for (i =3D 0; i < pps->num_tile_rows_minus1; i++) { + row_height[i] =3D pps->row_height_minus1[i] + 1; + sum +=3D row_height[i]; + } + row_height[i] =3D pic_in_cts_height - sum; +} + +static void set_ref_poc(struct rkvdec_rps_short_term_ref_set *set, int poc= , int value, int flag) +{ + switch (poc) { + case 0: + set->delta_poc0 =3D value; + set->used_flag0 =3D flag; + break; + case 1: + set->delta_poc1 =3D value; + set->used_flag1 =3D flag; + break; + case 2: + set->delta_poc2 =3D value; + set->used_flag2 =3D flag; + break; + case 3: + set->delta_poc3 =3D value; + set->used_flag3 =3D flag; + break; + case 4: + set->delta_poc4 =3D value; + set->used_flag4 =3D flag; + break; + case 5: + set->delta_poc5 =3D value; + set->used_flag5 =3D flag; + break; + case 6: + set->delta_poc6 =3D value; + set->used_flag6 =3D flag; + break; + case 7: + set->delta_poc7 =3D value; + set->used_flag7 =3D flag; + break; + case 8: + set->delta_poc8 =3D value; + set->used_flag8 =3D flag; + break; + case 9: + set->delta_poc9 =3D value; + set->used_flag9 =3D flag; + break; + case 10: + set->delta_poc10 =3D value; + set->used_flag10 =3D flag; + break; + case 11: + set->delta_poc11 =3D value; + set->used_flag11 =3D flag; + break; + case 12: + set->delta_poc12 =3D value; + set->used_flag12 =3D flag; + break; + case 13: + set->delta_poc13 =3D value; + set->used_flag13 =3D flag; + break; + case 14: + set->delta_poc14 =3D value; + set->used_flag14 =3D flag; + break; + } +} + /* * Flip one or more matrices along their main diagonal and flatten them * before writing it to the memory. @@ -105,7 +242,7 @@ static void translate_scaling_list(struct scaling_facto= r *output, } =20 void rkvdec_hevc_assemble_hw_scaling_list(struct rkvdec_hevc_run *run, - struct scaling_factor *scaling_factor, + struct scaling_factor *scaling_list, struct v4l2_ctrl_hevc_scaling_matrix *cache) { const struct v4l2_ctrl_hevc_scaling_matrix *scaling =3D run->scaling_matr= ix; @@ -114,15 +251,220 @@ void rkvdec_hevc_assemble_hw_scaling_list(struct rkv= dec_hevc_run *run, sizeof(struct v4l2_ctrl_hevc_scaling_matrix))) return; =20 - translate_scaling_list(scaling_factor, scaling); + translate_scaling_list(scaling_list, scaling); =20 memcpy(cache, scaling, sizeof(struct v4l2_ctrl_hevc_scaling_matrix)); } =20 -struct vb2_buffer * -get_ref_buf(struct rkvdec_ctx *ctx, struct rkvdec_hevc_run *run, - unsigned int dpb_idx) +static void rkvdec_hevc_assemble_hw_lt_rps(struct rkvdec_hevc_run *run, st= ruct rkvdec_rps *rps) +{ + const struct v4l2_ctrl_hevc_sps *sps =3D run->sps; + + if (!run->ext_sps_lt_rps) + return; + + for (int i =3D 0; i < sps->num_long_term_ref_pics_sps; i++) { + rps->refs[i].lt_ref_pic_poc_lsb =3D + run->ext_sps_lt_rps[i].lt_ref_pic_poc_lsb_sps; + rps->refs[i].used_by_curr_pic_lt_flag =3D + !!(run->ext_sps_lt_rps[i].flags & V4L2_HEVC_EXT_SPS_LT_RPS_FLAG_USED_LT= ); + } +} + +static void rkvdec_hevc_assemble_hw_st_rps(struct rkvdec_hevc_run *run, st= ruct rkvdec_rps *rps, + struct calculated_rps_st_set *calculated_rps_st_sets) +{ + const struct v4l2_ctrl_hevc_sps *sps =3D run->sps; + + for (int i =3D 0; i < sps->num_short_term_ref_pic_sets; i++) { + int poc =3D 0; + int j =3D 0; + const struct calculated_rps_st_set *set =3D &calculated_rps_st_sets[i]; + + rps->short_term_ref_sets[i].num_negative =3D set->num_negative_pics; + rps->short_term_ref_sets[i].num_positive =3D set->num_positive_pics; + + for (; j < set->num_negative_pics; j++) { + set_ref_poc(&rps->short_term_ref_sets[i], j, + set->delta_poc_s0[j], set->used_by_curr_pic_s0[j]); + } + poc =3D j; + + for (j =3D 0; j < set->num_positive_pics; j++) { + set_ref_poc(&rps->short_term_ref_sets[i], poc + j, + set->delta_poc_s1[j], set->used_by_curr_pic_s1[j]); + } + } +} + +/* + * Compute the short term ref pic set parameters based on its reference sh= ort term ref pic + */ +static void st_ref_pic_set_prediction(struct rkvdec_hevc_run *run, int idx, + struct calculated_rps_st_set *calculated_rps_st_sets) +{ + const struct v4l2_ctrl_hevc_ext_sps_st_rps *rps_data =3D &run->ext_sps_st= _rps[idx]; + struct calculated_rps_st_set *st_rps =3D &calculated_rps_st_sets[idx]; + struct calculated_rps_st_set *ref_rps; + u8 st_rps_idx =3D idx; + u8 ref_rps_idx =3D 0; + s16 delta_rps =3D 0; + u8 use_delta_flag[16] =3D { 0 }; + u8 used_by_curr_pic_flag[16] =3D { 0 }; + int i, j; + int dPoc; + + ref_rps_idx =3D st_rps_idx - (rps_data->delta_idx_minus1 + 1); /* 7-59 */ + delta_rps =3D (1 - 2 * rps_data->delta_rps_sign) * + (rps_data->abs_delta_rps_minus1 + 1); /* 7-60 */ + + ref_rps =3D &calculated_rps_st_sets[ref_rps_idx]; + + for (j =3D 0; j <=3D ref_rps->num_delta_pocs; j++) { + used_by_curr_pic_flag[j] =3D !!(rps_data->used_by_curr_pic & (1 << j)); + use_delta_flag[j] =3D !!(rps_data->use_delta_flag & (1 << j)); + } + + /* 7-61: calculate num_negative_pics, delta_poc_s0 and used_by_curr_pic_s= 0 */ + i =3D 0; + for (j =3D (ref_rps->num_positive_pics - 1); j >=3D 0; j--) { + dPoc =3D ref_rps->delta_poc_s1[j] + delta_rps; + if (dPoc < 0 && use_delta_flag[ref_rps->num_negative_pics + j]) { + st_rps->delta_poc_s0[i] =3D dPoc; + st_rps->used_by_curr_pic_s0[i++] =3D + used_by_curr_pic_flag[ref_rps->num_negative_pics + j]; + } + } + if (delta_rps < 0 && use_delta_flag[ref_rps->num_delta_pocs]) { + st_rps->delta_poc_s0[i] =3D delta_rps; + st_rps->used_by_curr_pic_s0[i++] =3D used_by_curr_pic_flag[ref_rps->num_= delta_pocs]; + } + for (j =3D 0; j < ref_rps->num_negative_pics; j++) { + dPoc =3D ref_rps->delta_poc_s0[j] + delta_rps; + if (dPoc < 0 && use_delta_flag[j]) { + st_rps->delta_poc_s0[i] =3D dPoc; + st_rps->used_by_curr_pic_s0[i++] =3D used_by_curr_pic_flag[j]; + } + } + st_rps->num_negative_pics =3D i; + + /* 7-62: calculate num_positive_pics, delta_poc_s1 and used_by_curr_pic_s= 1 */ + i =3D 0; + for (j =3D (ref_rps->num_negative_pics - 1); j >=3D 0; j--) { + dPoc =3D ref_rps->delta_poc_s0[j] + delta_rps; + if (dPoc > 0 && use_delta_flag[j]) { + st_rps->delta_poc_s1[i] =3D dPoc; + st_rps->used_by_curr_pic_s1[i++] =3D used_by_curr_pic_flag[j]; + } + } + if (delta_rps > 0 && use_delta_flag[ref_rps->num_delta_pocs]) { + st_rps->delta_poc_s1[i] =3D delta_rps; + st_rps->used_by_curr_pic_s1[i++] =3D used_by_curr_pic_flag[ref_rps->num_= delta_pocs]; + } + for (j =3D 0; j < ref_rps->num_positive_pics; j++) { + dPoc =3D ref_rps->delta_poc_s1[j] + delta_rps; + if (dPoc > 0 && use_delta_flag[ref_rps->num_negative_pics + j]) { + st_rps->delta_poc_s1[i] =3D dPoc; + st_rps->used_by_curr_pic_s1[i++] =3D + used_by_curr_pic_flag[ref_rps->num_negative_pics + j]; + } + } + st_rps->num_positive_pics =3D i; + + st_rps->num_delta_pocs =3D st_rps->num_positive_pics + st_rps->num_negati= ve_pics; +} + +/* + * Compute the short term ref pic set parameters based on the control's da= ta. + */ +static void st_ref_pic_set_calculate(struct rkvdec_hevc_run *run, int idx, + struct calculated_rps_st_set *calculated_rps_st_sets) +{ + const struct v4l2_ctrl_hevc_ext_sps_st_rps *rps_data =3D &run->ext_sps_st= _rps[idx]; + struct calculated_rps_st_set *st_rps =3D &calculated_rps_st_sets[idx]; + int j, i =3D 0; + + /* 7-63 */ + st_rps->num_negative_pics =3D rps_data->num_negative_pics; + /* 7-64 */ + st_rps->num_positive_pics =3D rps_data->num_positive_pics; + + for (i =3D 0; i < st_rps->num_negative_pics; i++) { + /* 7-65 */ + st_rps->used_by_curr_pic_s0[i] =3D !!(rps_data->used_by_curr_pic & (1 <<= i)); + + if (i =3D=3D 0) { + /* 7-67 */ + st_rps->delta_poc_s0[i] =3D -(rps_data->delta_poc_s0_minus1[i] + 1); + } else { + /* 7-69 */ + st_rps->delta_poc_s0[i] =3D + st_rps->delta_poc_s0[i - 1] - + (rps_data->delta_poc_s0_minus1[i] + 1); + } + } + + for (j =3D 0; j < st_rps->num_positive_pics; j++) { + /* 7-66 */ + st_rps->used_by_curr_pic_s1[j] =3D !!(rps_data->used_by_curr_pic & (1 <<= (i + j))); + + if (j =3D=3D 0) { + /* 7-68 */ + st_rps->delta_poc_s1[j] =3D rps_data->delta_poc_s1_minus1[j] + 1; + } else { + /* 7-70 */ + st_rps->delta_poc_s1[j] =3D + st_rps->delta_poc_s1[j - 1] + + (rps_data->delta_poc_s1_minus1[j] + 1); + } + } + + /* 7-71 */ + st_rps->num_delta_pocs =3D st_rps->num_positive_pics + st_rps->num_negati= ve_pics; +} + +static void rkvdec_hevc_prepare_hw_st_rps(struct rkvdec_hevc_run *run, str= uct rkvdec_rps *rps, + struct v4l2_ctrl_hevc_ext_sps_st_rps *cache) +{ + int idx; + + if (!run->ext_sps_st_rps) + return; + + if (!memcmp(cache, run->ext_sps_st_rps, sizeof(struct v4l2_ctrl_hevc_ext_= sps_st_rps))) + return; + + struct calculated_rps_st_set *calculated_rps_st_sets =3D + kzalloc(sizeof(struct calculated_rps_st_set) * + run->sps->num_short_term_ref_pic_sets, GFP_KERNEL); + + for (idx =3D 0; idx < run->sps->num_short_term_ref_pic_sets; idx++) { + const struct v4l2_ctrl_hevc_ext_sps_st_rps *rps_data =3D &run->ext_sps_s= t_rps[idx]; + + if (rps_data->flags & V4L2_HEVC_EXT_SPS_ST_RPS_FLAG_INTER_REF_PIC_SET_PR= ED) + st_ref_pic_set_prediction(run, idx, calculated_rps_st_sets); + else + st_ref_pic_set_calculate(run, idx, calculated_rps_st_sets); + } + + rkvdec_hevc_assemble_hw_st_rps(run, rps, calculated_rps_st_sets); + + kfree(calculated_rps_st_sets); + + memcpy(cache, run->ext_sps_st_rps, sizeof(struct v4l2_ctrl_hevc_ext_sps_s= t_rps)); +} + +void rkvdec_hevc_assemble_hw_rps(struct rkvdec_hevc_run *run, struct rkvde= c_rps *rps, + struct v4l2_ctrl_hevc_ext_sps_st_rps *st_cache) +{ + rkvdec_hevc_prepare_hw_st_rps(run, rps, st_cache); + rkvdec_hevc_assemble_hw_lt_rps(run, rps); +} + +struct vb2_buffer *get_ref_buf(struct rkvdec_ctx *ctx, + struct rkvdec_hevc_run *run, + unsigned int dpb_idx) { struct v4l2_m2m_ctx *m2m_ctx =3D ctx->fh.m2m_ctx; const struct v4l2_ctrl_hevc_decode_params *decode_params =3D run->decode_= params; @@ -143,8 +485,6 @@ get_ref_buf(struct rkvdec_ctx *ctx, struct rkvdec_hevc_= run *run, return buf; } =20 -#define RKVDEC_HEVC_MAX_DEPTH_IN_BYTES 2 - int rkvdec_hevc_adjust_fmt(struct rkvdec_ctx *ctx, struct v4l2_format *f) { struct v4l2_pix_format_mplane *fmt =3D &f->fmt.pix_mp; @@ -156,38 +496,17 @@ int rkvdec_hevc_adjust_fmt(struct rkvdec_ctx *ctx, st= ruct v4l2_format *f) return 0; } =20 -enum rkvdec_image_fmt rkvdec_hevc_get_image_fmt(struct rkvdec_ctx *ctx, - struct v4l2_ctrl *ctrl) -{ - const struct v4l2_ctrl_hevc_sps *sps =3D ctrl->p_new.p_hevc_sps; - - if (ctrl->id !=3D V4L2_CID_STATELESS_HEVC_SPS) - return RKVDEC_IMG_FMT_ANY; - - if (sps->bit_depth_luma_minus8 =3D=3D 0) { - if (sps->chroma_format_idc =3D=3D 2) - return RKVDEC_IMG_FMT_422_8BIT; - else - return RKVDEC_IMG_FMT_420_8BIT; - } else if (sps->bit_depth_luma_minus8 =3D=3D 2) { - if (sps->chroma_format_idc =3D=3D 2) - return RKVDEC_IMG_FMT_422_10BIT; - else - return RKVDEC_IMG_FMT_420_10BIT; - } - - return RKVDEC_IMG_FMT_ANY; -} - -static int rkvdec_hevc_validate_sps(struct rkvdec_ctx *ctx, +int rkvdec_hevc_validate_sps(struct rkvdec_ctx *ctx, const struct v4l2_ctrl_hevc_sps *sps) { if (sps->chroma_format_idc > 1) /* Only 4:0:0 and 4:2:0 are supported */ return -EINVAL; + if (sps->bit_depth_luma_minus8 !=3D sps->bit_depth_chroma_minus8) /* Luma and chroma bit depth mismatch */ return -EINVAL; + if (sps->bit_depth_luma_minus8 !=3D 0 && sps->bit_depth_luma_minus8 !=3D = 2) /* Only 8-bit and 10-bit are supported */ return -EINVAL; @@ -221,6 +540,17 @@ void rkvdec_hevc_run_preamble(struct rkvdec_ctx *ctx, V4L2_CID_STATELESS_HEVC_SCALING_MATRIX); run->scaling_matrix =3D ctrl ? ctrl->p_cur.p : NULL; =20 + if (ctx->has_sps_st_rps) { + ctrl =3D v4l2_ctrl_find(&ctx->ctrl_hdl, + V4L2_CID_STATELESS_HEVC_EXT_SPS_ST_RPS); + run->ext_sps_st_rps =3D ctrl ? ctrl->p_cur.p : NULL; + } + if (ctx->has_sps_lt_rps) { + ctrl =3D v4l2_ctrl_find(&ctx->ctrl_hdl, + V4L2_CID_STATELESS_HEVC_EXT_SPS_LT_RPS); + run->ext_sps_lt_rps =3D ctrl ? ctrl->p_cur.p : NULL; + } + rkvdec_run_preamble(ctx, &run->base); } =20 diff --git a/drivers/media/platform/rockchip/rkvdec/rkvdec-hevc-common.h b/= drivers/media/platform/rockchip/rkvdec/rkvdec-hevc-common.h index 746b1bd73c08..0005b80bad79 100644 --- a/drivers/media/platform/rockchip/rkvdec/rkvdec-hevc-common.h +++ b/drivers/media/platform/rockchip/rkvdec/rkvdec-hevc-common.h @@ -16,19 +16,70 @@ */ =20 #include +#include #include "rkvdec.h" =20 #define RKV_HEVC_CABAC_TABLE_SIZE 27456 extern const u8 rkvdec_hevc_cabac_table[RKV_HEVC_CABAC_TABLE_SIZE]; =20 +struct rkvdec_rps_refs { + u16 lt_ref_pic_poc_lsb; + u16 used_by_curr_pic_lt_flag : 1; + u16 reserved : 15; +} __packed; + +struct rkvdec_rps_short_term_ref_set { + u32 num_negative : 4; + u32 num_positive : 4; + u32 delta_poc0 : 16; + u32 used_flag0 : 1; + u32 delta_poc1 : 16; + u32 used_flag1 : 1; + u32 delta_poc2 : 16; + u32 used_flag2 : 1; + u32 delta_poc3 : 16; + u32 used_flag3 : 1; + u32 delta_poc4 : 16; + u32 used_flag4 : 1; + u32 delta_poc5 : 16; + u32 used_flag5 : 1; + u32 delta_poc6 : 16; + u32 used_flag6 : 1; + u32 delta_poc7 : 16; + u32 used_flag7 : 1; + u32 delta_poc8 : 16; + u32 used_flag8 : 1; + u32 delta_poc9 : 16; + u32 used_flag9 : 1; + u32 delta_poc10 : 16; + u32 used_flag10 : 1; + u32 delta_poc11 : 16; + u32 used_flag11 : 1; + u32 delta_poc12 : 16; + u32 used_flag12 : 1; + u32 delta_poc13 : 16; + u32 used_flag13 : 1; + u32 delta_poc14 : 16; + u32 used_flag14 : 1; + u32 reserved_bits : 25; + u32 reserved[3]; +} __packed; + +struct rkvdec_rps { + struct rkvdec_rps_refs refs[32]; + struct rkvdec_rps_short_term_ref_set short_term_ref_sets[64]; +} __packed; + struct rkvdec_hevc_run { - struct rkvdec_run base; - const struct v4l2_ctrl_hevc_slice_params *slices_params; - const struct v4l2_ctrl_hevc_decode_params *decode_params; - const struct v4l2_ctrl_hevc_sps *sps; - const struct v4l2_ctrl_hevc_pps *pps; - const struct v4l2_ctrl_hevc_scaling_matrix *scaling_matrix; - int num_slices; + struct rkvdec_run base; + const struct v4l2_ctrl_hevc_decode_params *decode_params; + const struct v4l2_ctrl_hevc_slice_params *slices_params; + const struct v4l2_ctrl_hevc_sps *sps; + const struct v4l2_ctrl_hevc_pps *pps; + const struct v4l2_ctrl_hevc_scaling_matrix *scaling_matrix; + const struct v4l2_ctrl_hevc_ext_sps_st_rps *ext_sps_st_rps; + const struct v4l2_ctrl_hevc_ext_sps_lt_rps *ext_sps_lt_rps; + int num_slices; }; =20 struct scaling_factor { @@ -40,6 +91,14 @@ struct scaling_factor { =20 enum rkvdec_image_fmt rkvdec_hevc_get_image_fmt(struct rkvdec_ctx *ctx, struct v4l2_ctrl *ctrl); +void compute_tiles_uniform(struct rkvdec_hevc_run *run, u16 log2_min_cb_si= ze, + u16 width, u16 height, s32 pic_in_cts_width, + s32 pic_in_cts_height, u16 *column_width, u16 *row_height); +void compute_tiles_non_uniform(struct rkvdec_hevc_run *run, u16 log2_min_c= b_size, + u16 width, u16 height, s32 pic_in_cts_width, + s32 pic_in_cts_height, u16 *column_width, u16 *row_height); +void rkvdec_hevc_assemble_hw_rps(struct rkvdec_hevc_run *run, struct rkvde= c_rps *rps, + struct v4l2_ctrl_hevc_ext_sps_st_rps *st_cache); void rkvdec_hevc_assemble_hw_scaling_list(struct rkvdec_hevc_run *run, struct scaling_factor *scaling_factor, struct v4l2_ctrl_hevc_scaling_matrix *cache); @@ -47,5 +106,6 @@ struct vb2_buffer *get_ref_buf(struct rkvdec_ctx *ctx, struct rkvdec_hevc_run *run, unsigned int dpb_idx); int rkvdec_hevc_adjust_fmt(struct rkvdec_ctx *ctx, struct v4l2_format *f); +int rkvdec_hevc_validate_sps(struct rkvdec_ctx *ctx, const struct v4l2_ctr= l_hevc_sps *sps); int rkvdec_hevc_try_ctrl(struct rkvdec_ctx *ctx, struct v4l2_ctrl *ctrl); void rkvdec_hevc_run_preamble(struct rkvdec_ctx *ctx, struct rkvdec_hevc_r= un *run); diff --git a/drivers/media/platform/rockchip/rkvdec/rkvdec-vdpu381-hevc.c b= /drivers/media/platform/rockchip/rkvdec/rkvdec-vdpu381-hevc.c new file mode 100644 index 000000000000..278f1dbf0910 --- /dev/null +++ b/drivers/media/platform/rockchip/rkvdec/rkvdec-vdpu381-hevc.c @@ -0,0 +1,596 @@ +// SPDX-License-Identifier: GPL-2.0 +/* + * Rockchip VDPU381 HEVC backend + * + * Copyright (C) 2025 Collabora, Ltd. + * Detlev Casanova + */ + +#include + +#include "rkvdec.h" +#include "rkvdec-rcb.h" +#include "rkvdec-hevc-common.h" +#include "rkvdec-vdpu381-regs.h" + +// SPS +struct rkvdec_hevc_sps { + u16 video_parameters_set_id : 4; + u16 seq_parameters_set_id_sps : 4; + u16 chroma_format_idc : 2; + u16 width : 16; + u16 height : 16; + u16 bit_depth_luma : 4; + u16 bit_depth_chroma : 4; + u16 max_pic_order_count_lsb : 5; + u16 diff_max_min_luma_coding_block_size : 2; + u16 min_luma_coding_block_size : 3; + u16 min_transform_block_size : 3; + u16 diff_max_min_transform_block_size : 2; + u16 max_transform_hierarchy_depth_inter : 3; + u16 max_transform_hierarchy_depth_intra : 3; + u16 scaling_list_enabled_flag : 1; + u16 amp_enabled_flag : 1; + u16 sample_adaptive_offset_enabled_flag : 1; + u16 pcm_enabled_flag : 1; + u16 pcm_sample_bit_depth_luma : 4; + u16 pcm_sample_bit_depth_chroma : 4; + u16 pcm_loop_filter_disabled_flag : 1; + u16 diff_max_min_pcm_luma_coding_block_size : 3; + u16 min_pcm_luma_coding_block_size : 3; + u16 num_short_term_ref_pic_sets : 7; + u16 long_term_ref_pics_present_flag : 1; + u16 num_long_term_ref_pics_sps : 6; + u16 sps_temporal_mvp_enabled_flag : 1; + u16 strong_intra_smoothing_enabled_flag : 1; + u16 reserved_0 : 7; + u16 sps_max_dec_pic_buffering_minus1 : 4; + u16 reserved_0_2 : 3; + u16 reserved_f : 8; +} __packed; + +//PPS +struct rkvdec_hevc_pps { + u16 picture_parameters_set_id : 6; + u16 seq_parameters_set_id_pps : 4; + u16 dependent_slice_segments_enabled_flag : 1; + u16 output_flag_present_flag : 1; + u16 num_extra_slice_header_bits : 13; + u16 sign_data_hiding_enabled_flag : 1; + u16 cabac_init_present_flag : 1; + u16 num_ref_idx_l0_default_active : 4; + u16 num_ref_idx_l1_default_active : 4; + u16 init_qp_minus26 : 7; + u16 constrained_intra_pred_flag : 1; + u16 transform_skip_enabled_flag : 1; + u16 cu_qp_delta_enabled_flag : 1; + u16 log2_min_cb_size : 3; + u16 pps_cb_qp_offset : 5; + u16 pps_cr_qp_offset : 5; + u16 pps_slice_chroma_qp_offsets_present_flag : 1; + u16 weighted_pred_flag : 1; + u16 weighted_bipred_flag : 1; + u16 transquant_bypass_enabled_flag : 1; + u16 tiles_enabled_flag : 1; + u16 entropy_coding_sync_enabled_flag : 1; + u16 pps_loop_filter_across_slices_enabled_flag : 1; + u16 loop_filter_across_tiles_enabled_flag : 1; + u16 deblocking_filter_override_enabled_flag : 1; + u16 pps_deblocking_filter_disabled_flag : 1; + u16 pps_beta_offset_div2 : 4; + u16 pps_tc_offset_div2 : 4; + u16 lists_modification_present_flag : 1; + u16 log2_parallel_merge_level : 3; + u16 slice_segment_header_extension_present_flag : 1; + u16 zeroes : 3; + u16 num_tile_columns : 5; + u16 num_tile_rows : 5; + u16 sps_pps_mode : 4; + u16 reserved_bits : 14; + u16 reserved; +} __packed; + +struct rkvdec_hevc_tile { + u16 value0 : 12; + u16 value1 : 12; +} __packed; + +struct rkvdec_sps_pps_packet { + struct rkvdec_hevc_sps sps; + struct rkvdec_hevc_pps pps; + struct rkvdec_hevc_tile column_width[10]; + struct rkvdec_hevc_tile row_height[11]; + u32 zeroes[3]; + u32 zeroes_bits : 6; + u32 padding_bits : 2; + u32 padding; +} __packed; + +struct rkvdec_hevc_priv_tbl { + struct rkvdec_sps_pps_packet param_set[64]; + struct rkvdec_rps rps; + struct scaling_factor scaling_list; + u8 cabac_table[27456]; +}; + +struct rkvdec_hevc_ctx { + struct rkvdec_aux_buf priv_tbl; + struct v4l2_ctrl_hevc_scaling_matrix scaling_matrix_cache; + struct v4l2_ctrl_hevc_ext_sps_st_rps st_cache; + struct rkvdec_vdpu381_regs_hevc regs; +}; + +static void assemble_hw_pps(struct rkvdec_ctx *ctx, + struct rkvdec_hevc_run *run) +{ + struct rkvdec_hevc_ctx *h264_ctx =3D ctx->priv; + const struct v4l2_ctrl_hevc_sps *sps =3D run->sps; + const struct v4l2_ctrl_hevc_pps *pps =3D run->pps; + struct rkvdec_hevc_priv_tbl *priv_tbl =3D h264_ctx->priv_tbl.cpu; + struct rkvdec_sps_pps_packet *hw_ps; + bool tiles_enabled; + s32 max_cu_width; + s32 pic_in_cts_width; + s32 pic_in_cts_height; + u16 log2_min_cb_size, width, height; + u16 column_width[20]; + u16 row_height[22]; + u8 pcm_enabled; + u32 i; + + /* + * HW read the SPS/PPS information from PPS packet index by PPS id. + * offset from the base can be calculated by PPS_id * 32 (size per PPS + * packet unit). so the driver copy SPS/PPS information to the exact PPS + * packet unit for HW accessing. + */ + hw_ps =3D &priv_tbl->param_set[pps->pic_parameter_set_id]; + memset(hw_ps, 0, sizeof(*hw_ps)); + + /* write sps */ + hw_ps->sps.video_parameters_set_id =3D sps->video_parameter_set_id; + hw_ps->sps.seq_parameters_set_id_sps =3D sps->seq_parameter_set_id; + hw_ps->sps.chroma_format_idc =3D sps->chroma_format_idc; + + log2_min_cb_size =3D sps->log2_min_luma_coding_block_size_minus3 + 3; + width =3D sps->pic_width_in_luma_samples; + height =3D sps->pic_height_in_luma_samples; + hw_ps->sps.width =3D width; + hw_ps->sps.height =3D height; + hw_ps->sps.bit_depth_luma =3D sps->bit_depth_luma_minus8 + 8; + hw_ps->sps.bit_depth_chroma =3D sps->bit_depth_chroma_minus8 + 8; + hw_ps->sps.max_pic_order_count_lsb =3D sps->log2_max_pic_order_cnt_lsb_mi= nus4 + 4; + hw_ps->sps.diff_max_min_luma_coding_block_size =3D + sps->log2_diff_max_min_luma_coding_block_size; + hw_ps->sps.min_luma_coding_block_size =3D sps->log2_min_luma_coding_block= _size_minus3 + 3; + hw_ps->sps.min_transform_block_size =3D sps->log2_min_luma_transform_bloc= k_size_minus2 + 2; + hw_ps->sps.diff_max_min_transform_block_size =3D + sps->log2_diff_max_min_luma_transform_block_size; + hw_ps->sps.max_transform_hierarchy_depth_inter =3D sps->max_transform_hie= rarchy_depth_inter; + hw_ps->sps.max_transform_hierarchy_depth_intra =3D sps->max_transform_hie= rarchy_depth_intra; + hw_ps->sps.scaling_list_enabled_flag =3D + !!(sps->flags & V4L2_HEVC_SPS_FLAG_SCALING_LIST_ENABLED); + hw_ps->sps.amp_enabled_flag =3D !!(sps->flags & V4L2_HEVC_SPS_FLAG_AMP_EN= ABLED); + hw_ps->sps.sample_adaptive_offset_enabled_flag =3D + !!(sps->flags & V4L2_HEVC_SPS_FLAG_SAMPLE_ADAPTIVE_OFFSET); + + pcm_enabled =3D !!(sps->flags & V4L2_HEVC_SPS_FLAG_PCM_ENABLED); + hw_ps->sps.pcm_enabled_flag =3D pcm_enabled; + hw_ps->sps.pcm_sample_bit_depth_luma =3D + pcm_enabled ? sps->pcm_sample_bit_depth_luma_minus1 + 1 : 0; + hw_ps->sps.pcm_sample_bit_depth_chroma =3D + pcm_enabled ? sps->pcm_sample_bit_depth_chroma_minus1 + 1 : 0; + hw_ps->sps.pcm_loop_filter_disabled_flag =3D + !!(sps->flags & V4L2_HEVC_SPS_FLAG_PCM_LOOP_FILTER_DISABLED); + hw_ps->sps.diff_max_min_pcm_luma_coding_block_size =3D + sps->log2_diff_max_min_pcm_luma_coding_block_size; + hw_ps->sps.min_pcm_luma_coding_block_size =3D + pcm_enabled ? sps->log2_min_pcm_luma_coding_block_size_minus3 + 3 : 0; + hw_ps->sps.num_short_term_ref_pic_sets =3D sps->num_short_term_ref_pic_se= ts; + hw_ps->sps.long_term_ref_pics_present_flag =3D + !!(sps->flags & V4L2_HEVC_SPS_FLAG_LONG_TERM_REF_PICS_PRESENT); + hw_ps->sps.num_long_term_ref_pics_sps =3D sps->num_long_term_ref_pics_sps; + hw_ps->sps.sps_temporal_mvp_enabled_flag =3D + !!(sps->flags & V4L2_HEVC_SPS_FLAG_SPS_TEMPORAL_MVP_ENABLED); + hw_ps->sps.strong_intra_smoothing_enabled_flag =3D + !!(sps->flags & V4L2_HEVC_SPS_FLAG_STRONG_INTRA_SMOOTHING_ENABLED); + hw_ps->sps.sps_max_dec_pic_buffering_minus1 =3D sps->sps_max_dec_pic_buff= ering_minus1; + hw_ps->sps.reserved_f =3D 0xff; + + /* write pps */ + hw_ps->pps.picture_parameters_set_id =3D pps->pic_parameter_set_id; + hw_ps->pps.seq_parameters_set_id_pps =3D sps->seq_parameter_set_id; + hw_ps->pps.dependent_slice_segments_enabled_flag =3D + !!(pps->flags & V4L2_HEVC_PPS_FLAG_DEPENDENT_SLICE_SEGMENT_ENABLED); + hw_ps->pps.output_flag_present_flag =3D + !!(pps->flags & V4L2_HEVC_PPS_FLAG_OUTPUT_FLAG_PRESENT); + hw_ps->pps.num_extra_slice_header_bits =3D pps->num_extra_slice_header_bi= ts; + hw_ps->pps.sign_data_hiding_enabled_flag =3D + !!(pps->flags & V4L2_HEVC_PPS_FLAG_SIGN_DATA_HIDING_ENABLED); + hw_ps->pps.cabac_init_present_flag =3D + !!(pps->flags & V4L2_HEVC_PPS_FLAG_CABAC_INIT_PRESENT); + hw_ps->pps.num_ref_idx_l0_default_active =3D pps->num_ref_idx_l0_default_= active_minus1 + 1; + hw_ps->pps.num_ref_idx_l1_default_active =3D pps->num_ref_idx_l1_default_= active_minus1 + 1; + hw_ps->pps.init_qp_minus26 =3D pps->init_qp_minus26; + hw_ps->pps.constrained_intra_pred_flag =3D + !!(pps->flags & V4L2_HEVC_PPS_FLAG_CONSTRAINED_INTRA_PRED); + hw_ps->pps.transform_skip_enabled_flag =3D + !!(pps->flags & V4L2_HEVC_PPS_FLAG_TRANSFORM_SKIP_ENABLED); + hw_ps->pps.cu_qp_delta_enabled_flag =3D + !!(pps->flags & V4L2_HEVC_PPS_FLAG_CU_QP_DELTA_ENABLED); + hw_ps->pps.log2_min_cb_size =3D log2_min_cb_size + + sps->log2_diff_max_min_luma_coding_block_size - + pps->diff_cu_qp_delta_depth; + hw_ps->pps.pps_cb_qp_offset =3D pps->pps_cb_qp_offset; + hw_ps->pps.pps_cr_qp_offset =3D pps->pps_cr_qp_offset; + hw_ps->pps.pps_slice_chroma_qp_offsets_present_flag =3D + !!(pps->flags & V4L2_HEVC_PPS_FLAG_PPS_SLICE_CHROMA_QP_OFFSETS_PRESENT); + hw_ps->pps.weighted_pred_flag =3D !!(pps->flags & V4L2_HEVC_PPS_FLAG_WEIG= HTED_PRED); + hw_ps->pps.weighted_bipred_flag =3D !!(pps->flags & V4L2_HEVC_PPS_FLAG_WE= IGHTED_BIPRED); + hw_ps->pps.transquant_bypass_enabled_flag =3D + !!(pps->flags & V4L2_HEVC_PPS_FLAG_TRANSQUANT_BYPASS_ENABLED); + + tiles_enabled =3D !!(pps->flags & V4L2_HEVC_PPS_FLAG_TILES_ENABLED); + hw_ps->pps.tiles_enabled_flag =3D tiles_enabled; + hw_ps->pps.entropy_coding_sync_enabled_flag =3D + !!(pps->flags & V4L2_HEVC_PPS_FLAG_ENTROPY_CODING_SYNC_ENABLED); + hw_ps->pps.pps_loop_filter_across_slices_enabled_flag =3D + !!(pps->flags & V4L2_HEVC_PPS_FLAG_PPS_LOOP_FILTER_ACROSS_SLICES_ENABLED= ); + hw_ps->pps.loop_filter_across_tiles_enabled_flag =3D + !!(pps->flags & V4L2_HEVC_PPS_FLAG_LOOP_FILTER_ACROSS_TILES_ENABLED); + hw_ps->pps.deblocking_filter_override_enabled_flag =3D + !!(pps->flags & V4L2_HEVC_PPS_FLAG_DEBLOCKING_FILTER_OVERRIDE_ENABLED); + hw_ps->pps.pps_deblocking_filter_disabled_flag =3D + !!(pps->flags & V4L2_HEVC_PPS_FLAG_PPS_DISABLE_DEBLOCKING_FILTER); + hw_ps->pps.pps_beta_offset_div2 =3D pps->pps_beta_offset_div2; + hw_ps->pps.pps_tc_offset_div2 =3D pps->pps_tc_offset_div2; + hw_ps->pps.lists_modification_present_flag =3D + !!(pps->flags & V4L2_HEVC_PPS_FLAG_LISTS_MODIFICATION_PRESENT); + hw_ps->pps.log2_parallel_merge_level =3D pps->log2_parallel_merge_level_m= inus2 + 2; + hw_ps->pps.slice_segment_header_extension_present_flag =3D + !!(pps->flags & V4L2_HEVC_PPS_FLAG_SLICE_SEGMENT_HEADER_EXTENSION_PRESEN= T); + hw_ps->pps.num_tile_columns =3D tiles_enabled ? pps->num_tile_columns_min= us1 + 1 : 0; + hw_ps->pps.num_tile_rows =3D tiles_enabled ? pps->num_tile_rows_minus1 + = 1 : 0; + hw_ps->pps.sps_pps_mode =3D 0; + hw_ps->pps.reserved_bits =3D 0x3fff; + hw_ps->pps.reserved =3D 0xffff; + + // Setup tiles information + memset(column_width, 0, sizeof(column_width)); + memset(row_height, 0, sizeof(row_height)); + + max_cu_width =3D 1 << (sps->log2_diff_max_min_luma_coding_block_size + lo= g2_min_cb_size); + pic_in_cts_width =3D (width + max_cu_width - 1) / max_cu_width; + pic_in_cts_height =3D (height + max_cu_width - 1) / max_cu_width; + + if (pps->flags & V4L2_HEVC_PPS_FLAG_TILES_ENABLED) { + if (pps->flags & V4L2_HEVC_PPS_FLAG_UNIFORM_SPACING) { + compute_tiles_uniform(run, log2_min_cb_size, width, height, + pic_in_cts_width, pic_in_cts_height, + column_width, row_height); + } else { + compute_tiles_non_uniform(run, log2_min_cb_size, width, height, + pic_in_cts_width, pic_in_cts_height, + column_width, row_height); + } + } else { + column_width[0] =3D (width + max_cu_width - 1) / max_cu_width; + row_height[0] =3D (height + max_cu_width - 1) / max_cu_width; + } + + for (i =3D 0; i < 20; i++) { + if (column_width[i] > 0) + column_width[i]--; + + if (i & 1) + hw_ps->column_width[i / 2].value1 =3D column_width[i]; + else + hw_ps->column_width[i / 2].value0 =3D column_width[i]; + } + + for (i =3D 0; i < 22; i++) { + if (row_height[i] > 0) + row_height[i]--; + + if (i & 1) + hw_ps->row_height[i / 2].value1 =3D row_height[i]; + else + hw_ps->row_height[i / 2].value0 =3D row_height[i]; + } + + hw_ps->padding =3D 0xffffffff; + hw_ps->padding_bits =3D 0x3; +} + +static void set_ref_valid(struct rkvdec_vdpu381_regs_hevc *regs, int id, u= 32 valid) +{ + switch (id) { + case 0: + regs->hevc_param.reg099.hevc_ref_valid_0 =3D valid; + break; + case 1: + regs->hevc_param.reg099.hevc_ref_valid_1 =3D valid; + break; + case 2: + regs->hevc_param.reg099.hevc_ref_valid_2 =3D valid; + break; + case 3: + regs->hevc_param.reg099.hevc_ref_valid_3 =3D valid; + break; + case 4: + regs->hevc_param.reg099.hevc_ref_valid_4 =3D valid; + break; + case 5: + regs->hevc_param.reg099.hevc_ref_valid_5 =3D valid; + break; + case 6: + regs->hevc_param.reg099.hevc_ref_valid_6 =3D valid; + break; + case 7: + regs->hevc_param.reg099.hevc_ref_valid_7 =3D valid; + break; + case 8: + regs->hevc_param.reg099.hevc_ref_valid_8 =3D valid; + break; + case 9: + regs->hevc_param.reg099.hevc_ref_valid_9 =3D valid; + break; + case 10: + regs->hevc_param.reg099.hevc_ref_valid_10 =3D valid; + break; + case 11: + regs->hevc_param.reg099.hevc_ref_valid_11 =3D valid; + break; + case 12: + regs->hevc_param.reg099.hevc_ref_valid_12 =3D valid; + break; + case 13: + regs->hevc_param.reg099.hevc_ref_valid_13 =3D valid; + break; + case 14: + regs->hevc_param.reg099.hevc_ref_valid_14 =3D valid; + break; + } +} + +static void rkvdec_write_regs(struct rkvdec_ctx *ctx) +{ + struct rkvdec_dev *rkvdec =3D ctx->dev; + struct rkvdec_hevc_ctx *hevc_ctx =3D ctx->priv; + + rkvdec_memcpy_toio(rkvdec->regs + OFFSET_COMMON_REGS, + &hevc_ctx->regs.common, + sizeof(hevc_ctx->regs.common)); + rkvdec_memcpy_toio(rkvdec->regs + OFFSET_CODEC_PARAMS_REGS, + &hevc_ctx->regs.hevc_param, + sizeof(hevc_ctx->regs.hevc_param)); + rkvdec_memcpy_toio(rkvdec->regs + OFFSET_COMMON_ADDR_REGS, + &hevc_ctx->regs.common_addr, + sizeof(hevc_ctx->regs.common_addr)); + rkvdec_memcpy_toio(rkvdec->regs + OFFSET_CODEC_ADDR_REGS, + &hevc_ctx->regs.hevc_addr, + sizeof(hevc_ctx->regs.hevc_addr)); + rkvdec_memcpy_toio(rkvdec->regs + OFFSET_POC_HIGHBIT_REGS, + &hevc_ctx->regs.hevc_highpoc, + sizeof(hevc_ctx->regs.hevc_highpoc)); +} + +static void config_registers(struct rkvdec_ctx *ctx, + struct rkvdec_hevc_run *run) +{ + const struct v4l2_ctrl_hevc_decode_params *dec_params =3D run->decode_par= ams; + const struct v4l2_hevc_dpb_entry *dpb =3D dec_params->dpb; + struct rkvdec_hevc_ctx *hevc_ctx =3D ctx->priv; + struct rkvdec_vdpu381_regs_hevc *regs =3D &hevc_ctx->regs; + dma_addr_t priv_start_addr =3D hevc_ctx->priv_tbl.dma; + const struct v4l2_pix_format_mplane *dst_fmt; + struct vb2_v4l2_buffer *src_buf =3D run->base.bufs.src; + struct vb2_v4l2_buffer *dst_buf =3D run->base.bufs.dst; + const struct v4l2_format *f; + dma_addr_t rlc_addr; + u32 hor_virstride =3D 0; + u32 ver_virstride =3D 0; + u32 y_virstride =3D 0; + u32 offset; + u32 pixels; + dma_addr_t dst_addr; + u32 i; + + memset(regs, 0, sizeof(*regs)); + + /* Set HEVC mode */ + regs->common.reg009.dec_mode =3D VDPU381_MODE_HEVC; + + /* Set config */ + regs->common.reg011.buf_empty_en =3D 1; + regs->common.reg011.dec_clkgate_e =3D 1; + regs->common.reg011.dec_timeout_e =3D 1; + regs->common.reg011.pix_range_detection_e =3D 1; + + /* Set IDR flag */ + regs->common.reg013.cur_pic_is_idr =3D + !!(dec_params->flags & V4L2_HEVC_DECODE_PARAM_FLAG_IDR_PIC); + + /* Set input stream length */ + regs->common.stream_len =3D vb2_get_plane_payload(&src_buf->vb2_buf, 0); + + /* Set max slice number */ + regs->common.reg017.slice_num =3D 1; + + /* Set strides */ + f =3D &ctx->decoded_fmt; + dst_fmt =3D &f->fmt.pix_mp; + hor_virstride =3D dst_fmt->plane_fmt[0].bytesperline; + ver_virstride =3D dst_fmt->height; + y_virstride =3D hor_virstride * ver_virstride; + pixels =3D dst_fmt->height * dst_fmt->width; + + regs->common.reg018.y_hor_virstride =3D hor_virstride / 16; + regs->common.reg019.uv_hor_virstride =3D hor_virstride / 16; + regs->common.reg020.y_virstride =3D y_virstride / 16; + + /* Activate block gating */ + regs->common.reg026.swreg_block_gating_e =3D 0xfffef; + regs->common.reg026.reg_cfg_gating_en =3D 1; + + /* Set timeout threshold */ + if (pixels < RKVDEC_1080P_PIXELS) + regs->common.timeout_threshold =3D RKVDEC_TIMEOUT_1080p; + else if (pixels < RKVDEC_4K_PIXELS) + regs->common.timeout_threshold =3D RKVDEC_TIMEOUT_4K; + else if (pixels < RKVDEC_8K_PIXELS) + regs->common.timeout_threshold =3D RKVDEC_TIMEOUT_8K; + + /* Set POC val */ + regs->hevc_param.cur_top_poc =3D dec_params->pic_order_cnt_val; + + /* Set ref pic address & poc */ + for (i =3D 0; i < ARRAY_SIZE(dec_params->dpb); i++) { + struct vb2_buffer *vb_buf =3D get_ref_buf(ctx, run, i); + dma_addr_t buf_dma =3D vb2_dma_contig_plane_dma_addr(vb_buf, 0); + u32 valid =3D !!(dec_params->num_active_dpb_entries > i); + + /* Set reference addresses */ + regs->hevc_addr.ref_base[i] =3D buf_dma; + + /* Set COLMV addresses */ + regs->hevc_addr.colmv_base[i] =3D buf_dma + ctx->colmv_offset; + + regs->hevc_param.reg067_082_ref_poc[i] =3D + dpb[i].pic_order_cnt_val; + + set_ref_valid(regs, i, valid); + regs->hevc_param.reg103.ref_pic_layer_same_with_cur |=3D 1 << i; + } + + /* Set rlc base address (input stream) */ + rlc_addr =3D vb2_dma_contig_plane_dma_addr(&src_buf->vb2_buf, 0); + regs->common_addr.rlc_base =3D rlc_addr; + regs->common_addr.rlcwrite_base =3D rlc_addr; + + /* Set output base address */ + dst_addr =3D vb2_dma_contig_plane_dma_addr(&dst_buf->vb2_buf, 0); + regs->common_addr.decout_base =3D dst_addr; + regs->common_addr.error_ref_base =3D dst_addr; + + /* Set colmv address */ + regs->common_addr.colmv_cur_base =3D dst_addr + ctx->colmv_offset; + + /* Set RCB addresses */ + for (i =3D 0; i < rkvdec_rcb_buf_count(ctx); i++) + regs->common_addr.rcb_base[i] =3D rkvdec_rcb_buf_dma_addr(ctx, i); + + /* Set hw pps address */ + offset =3D offsetof(struct rkvdec_hevc_priv_tbl, param_set); + regs->hevc_addr.pps_base =3D priv_start_addr + offset; + + /* Set hw rps address */ + offset =3D offsetof(struct rkvdec_hevc_priv_tbl, rps); + regs->hevc_addr.rps_base =3D priv_start_addr + offset; + + /* Set cabac table */ + offset =3D offsetof(struct rkvdec_hevc_priv_tbl, cabac_table); + regs->hevc_addr.cabactbl_base =3D priv_start_addr + offset; + + /* Set scaling matrix */ + offset =3D offsetof(struct rkvdec_hevc_priv_tbl, scaling_list); + regs->hevc_addr.scanlist_addr =3D priv_start_addr + offset; + + rkvdec_write_regs(ctx); +} + +static int rkvdec_hevc_start(struct rkvdec_ctx *ctx) +{ + struct rkvdec_dev *rkvdec =3D ctx->dev; + struct rkvdec_hevc_priv_tbl *priv_tbl; + struct rkvdec_hevc_ctx *hevc_ctx; + struct v4l2_ctrl *ctrl; + int ret; + + ctrl =3D v4l2_ctrl_find(&ctx->ctrl_hdl, + V4L2_CID_STATELESS_HEVC_SPS); + if (!ctrl) + return -EINVAL; + + ret =3D rkvdec_hevc_validate_sps(ctx, ctrl->p_new.p_hevc_sps); + if (ret) + return ret; + + hevc_ctx =3D kzalloc(sizeof(*hevc_ctx), GFP_KERNEL); + if (!hevc_ctx) + return -ENOMEM; + + priv_tbl =3D dma_alloc_coherent(rkvdec->dev, sizeof(*priv_tbl), + &hevc_ctx->priv_tbl.dma, GFP_KERNEL); + if (!priv_tbl) { + ret =3D -ENOMEM; + goto err_free_ctx; + } + + hevc_ctx->priv_tbl.size =3D sizeof(*priv_tbl); + hevc_ctx->priv_tbl.cpu =3D priv_tbl; + memcpy(priv_tbl->cabac_table, rkvdec_hevc_cabac_table, + sizeof(rkvdec_hevc_cabac_table)); + + ctx->priv =3D hevc_ctx; + return 0; + +err_free_ctx: + kfree(hevc_ctx); + return ret; +} + +static void rkvdec_hevc_stop(struct rkvdec_ctx *ctx) +{ + struct rkvdec_hevc_ctx *hevc_ctx =3D ctx->priv; + struct rkvdec_dev *rkvdec =3D ctx->dev; + + dma_free_coherent(rkvdec->dev, hevc_ctx->priv_tbl.size, + hevc_ctx->priv_tbl.cpu, hevc_ctx->priv_tbl.dma); + kfree(hevc_ctx); +} + +static int rkvdec_hevc_run(struct rkvdec_ctx *ctx) +{ + struct rkvdec_dev *rkvdec =3D ctx->dev; + struct rkvdec_hevc_run run; + struct rkvdec_hevc_ctx *hevc_ctx =3D ctx->priv; + struct rkvdec_hevc_priv_tbl *tbl =3D hevc_ctx->priv_tbl.cpu; + + rkvdec_hevc_run_preamble(ctx, &run); + + /* + * On vdpu381, not setting the long and short term ref sets will just out= put wrong frames. + * Let's just warn about it and let the decoder run anyway. + */ + if ((!ctx->has_sps_lt_rps && run.sps->num_long_term_ref_pics_sps) || + (!ctx->has_sps_st_rps && run.sps->num_short_term_ref_pic_sets)) { + dev_warn_ratelimited(rkvdec->dev, "Long and short term RPS not set\n"); + } + + rkvdec_hevc_assemble_hw_scaling_list(&run, + &tbl->scaling_list, + &hevc_ctx->scaling_matrix_cache); + assemble_hw_pps(ctx, &run); + rkvdec_hevc_assemble_hw_rps(&run, &tbl->rps, &hevc_ctx->st_cache); + + config_registers(ctx, &run); + + rkvdec_run_postamble(ctx, &run.base); + + schedule_delayed_work(&rkvdec->watchdog_work, msecs_to_jiffies(2000)); + + /* Start decoding! */ + writel(VDPU381_DEC_E_BIT, rkvdec->regs + VDPU381_REG_DEC_E); + + return 0; +} + +const struct rkvdec_coded_fmt_ops rkvdec_vdpu381_hevc_fmt_ops =3D { + .adjust_fmt =3D rkvdec_hevc_adjust_fmt, + .start =3D rkvdec_hevc_start, + .stop =3D rkvdec_hevc_stop, + .run =3D rkvdec_hevc_run, + .try_ctrl =3D rkvdec_hevc_try_ctrl, + .get_image_fmt =3D rkvdec_hevc_get_image_fmt, +}; diff --git a/drivers/media/platform/rockchip/rkvdec/rkvdec.c b/drivers/medi= a/platform/rockchip/rkvdec/rkvdec.c index 1837f12f1aa1..b72cb07de70e 100644 --- a/drivers/media/platform/rockchip/rkvdec/rkvdec.c +++ b/drivers/media/platform/rockchip/rkvdec/rkvdec.c @@ -154,6 +154,16 @@ static int rkvdec_s_ctrl(struct v4l2_ctrl *ctrl) enum rkvdec_image_fmt image_fmt; struct vb2_queue *vq; =20 + if (ctrl->id =3D=3D V4L2_CID_STATELESS_HEVC_EXT_SPS_ST_RPS) { + ctx->has_sps_st_rps |=3D !!(ctrl->has_changed); + return 0; + } + + if (ctrl->id =3D=3D V4L2_CID_STATELESS_HEVC_EXT_SPS_LT_RPS) { + ctx->has_sps_lt_rps |=3D !!(ctrl->has_changed); + return 0; + } + /* Check if this change requires a capture format reset */ if (!desc->ops->get_image_fmt) return 0; @@ -285,6 +295,62 @@ static const struct rkvdec_ctrls rkvdec_h264_ctrls =3D= { .num_ctrls =3D ARRAY_SIZE(rkvdec_h264_ctrl_descs), }; =20 +static const struct rkvdec_ctrl_desc vdpu38x_hevc_ctrl_descs[] =3D { + { + .cfg.id =3D V4L2_CID_STATELESS_HEVC_DECODE_PARAMS, + }, + { + .cfg.id =3D V4L2_CID_STATELESS_HEVC_SPS, + .cfg.ops =3D &rkvdec_ctrl_ops, + }, + { + .cfg.id =3D V4L2_CID_STATELESS_HEVC_PPS, + }, + { + .cfg.id =3D V4L2_CID_STATELESS_HEVC_SCALING_MATRIX, + }, + { + .cfg.id =3D V4L2_CID_STATELESS_HEVC_DECODE_MODE, + .cfg.min =3D V4L2_STATELESS_HEVC_DECODE_MODE_FRAME_BASED, + .cfg.max =3D V4L2_STATELESS_HEVC_DECODE_MODE_FRAME_BASED, + .cfg.def =3D V4L2_STATELESS_HEVC_DECODE_MODE_FRAME_BASED, + }, + { + .cfg.id =3D V4L2_CID_STATELESS_HEVC_START_CODE, + .cfg.min =3D V4L2_STATELESS_HEVC_START_CODE_ANNEX_B, + .cfg.def =3D V4L2_STATELESS_HEVC_START_CODE_ANNEX_B, + .cfg.max =3D V4L2_STATELESS_HEVC_START_CODE_ANNEX_B, + }, + { + .cfg.id =3D V4L2_CID_MPEG_VIDEO_HEVC_PROFILE, + .cfg.min =3D V4L2_MPEG_VIDEO_HEVC_PROFILE_MAIN, + .cfg.max =3D V4L2_MPEG_VIDEO_HEVC_PROFILE_MAIN_10, + .cfg.menu_skip_mask =3D + BIT(V4L2_MPEG_VIDEO_HEVC_PROFILE_MAIN_STILL_PICTURE), + .cfg.def =3D V4L2_MPEG_VIDEO_HEVC_PROFILE_MAIN, + }, + { + .cfg.id =3D V4L2_CID_MPEG_VIDEO_HEVC_LEVEL, + .cfg.min =3D V4L2_MPEG_VIDEO_HEVC_LEVEL_1, + .cfg.max =3D V4L2_MPEG_VIDEO_HEVC_LEVEL_6_1, + }, + { + .cfg.id =3D V4L2_CID_STATELESS_HEVC_EXT_SPS_ST_RPS, + .cfg.ops =3D &rkvdec_ctrl_ops, + .cfg.dims =3D { 65 }, + }, + { + .cfg.id =3D V4L2_CID_STATELESS_HEVC_EXT_SPS_LT_RPS, + .cfg.ops =3D &rkvdec_ctrl_ops, + .cfg.dims =3D { 65 }, + }, +}; + +static const struct rkvdec_ctrls vdpu38x_hevc_ctrls =3D { + .ctrls =3D vdpu38x_hevc_ctrl_descs, + .num_ctrls =3D ARRAY_SIZE(vdpu38x_hevc_ctrl_descs), +}; + static const struct rkvdec_decoded_fmt_desc rkvdec_h264_decoded_fmts[] =3D= { { .fourcc =3D V4L2_PIX_FMT_NV12, @@ -401,6 +467,23 @@ static const struct rkvdec_coded_fmt_desc vdpu381_code= d_fmts[] =3D { .subsystem_flags =3D VB2_V4L2_FL_SUPPORTS_M2M_HOLD_CAPTURE_BUF, .capability =3D RKVDEC_CAPABILITY_H264, }, + { + .fourcc =3D V4L2_PIX_FMT_HEVC_SLICE, + .frmsize =3D { + .min_width =3D 64, + .max_width =3D 65472, + .step_width =3D 64, + .min_height =3D 64, + .max_height =3D 65472, + .step_height =3D 16, + }, + .ctrls =3D &vdpu38x_hevc_ctrls, + .ops =3D &rkvdec_vdpu381_hevc_fmt_ops, + .num_decoded_fmts =3D ARRAY_SIZE(rkvdec_hevc_decoded_fmts), + .decoded_fmts =3D rkvdec_hevc_decoded_fmts, + .subsystem_flags =3D VB2_V4L2_FL_SUPPORTS_M2M_HOLD_CAPTURE_BUF, + .capability =3D RKVDEC_CAPABILITY_HEVC, + }, }; =20 static const struct rkvdec_coded_fmt_desc vdpu383_coded_fmts[] =3D { @@ -1500,7 +1583,8 @@ static const struct rkvdec_variant rk3399_rkvdec_vari= ant =3D { =20 static const struct rkvdec_variant rk3588_vdpu381_variant =3D { .config =3D &config_vdpu381, - .capabilities =3D RKVDEC_CAPABILITY_H264, + .capabilities =3D RKVDEC_CAPABILITY_H264 | + RKVDEC_CAPABILITY_HEVC, }; =20 static const struct rkvdec_variant rk3576_vdpu383_variant =3D { diff --git a/drivers/media/platform/rockchip/rkvdec/rkvdec.h b/drivers/medi= a/platform/rockchip/rkvdec/rkvdec.h index e752706e9f5c..94e2e6429992 100644 --- a/drivers/media/platform/rockchip/rkvdec/rkvdec.h +++ b/drivers/media/platform/rockchip/rkvdec/rkvdec.h @@ -156,6 +156,8 @@ struct rkvdec_ctx { struct rkvdec_rcb_config *rcb_config; u32 colmv_offset; void *priv; + u8 has_sps_st_rps: 1; + u8 has_sps_lt_rps: 1; }; =20 static inline struct rkvdec_ctx *file_to_rkvdec_ctx(struct file *filp) @@ -188,6 +190,7 @@ extern const struct rkvdec_coded_fmt_ops rkvdec_vp9_fmt= _ops; =20 /* VDPU381 ops */ extern const struct rkvdec_coded_fmt_ops rkvdec_vdpu381_h264_fmt_ops; +extern const struct rkvdec_coded_fmt_ops rkvdec_vdpu381_hevc_fmt_ops; =20 /* VDPU383 ops */ extern const struct rkvdec_coded_fmt_ops rkvdec_vdpu383_h264_fmt_ops; --=20 2.51.1.dirty