From patchwork Sat Jul 8 09:12:06 2017 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 8bit X-Patchwork-Submitter: Paul B Mahol X-Patchwork-Id: 4265 Delivered-To: ffmpegpatchwork@gmail.com Received: by 10.103.1.76 with SMTP id 73csp1176665vsb; Sat, 8 Jul 2017 02:12:32 -0700 (PDT) X-Received: by 10.223.171.25 with SMTP id q25mr2916251wrc.89.1499505152122; Sat, 08 Jul 2017 02:12:32 -0700 (PDT) ARC-Seal: i=1; a=rsa-sha256; t=1499505152; cv=none; d=google.com; s=arc-20160816; b=hTmkTaNdr/PLWmq0JiP04e7lnPOfB3QTHVLmsEa12lIpqcXwQaOO2TNfEigidf3spb u3obbG6scaHequyX/mNsWt/9A74UY3ibxnE+fJEjQXWfDrXXTifsGTWC3LFTaa2UMOvd LLg3bIDZaeOgX/7ZbHsSL3jD5XF5rZKgrs/6Ajvrbowysx2nK/2EvcnJZkCBMsdBmt4B 9l886SHGjbcJRoYF5vv+Q/b/sEKdaaFzqjTY01c0szOPNarzhy5yO+OQ2HUKdJwhlxHn tD9CUszDZdGAmT8b9V1OF33jKvMmzycYNTcTfahHteXdfqXVG75+AjzdR5xSiYNH1ef6 nWnA== ARC-Message-Signature: i=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=arc-20160816; h=sender:errors-to:content-transfer-encoding:reply-to:list-subscribe :list-help:list-post:list-archive:list-unsubscribe:list-id :precedence:subject:mime-version:references:in-reply-to:message-id :date:to:from:dkim-signature:delivered-to:arc-authentication-results; bh=i7TdP98ZGbzaD6K36S74gDuHhGAKZPhCCSajH6w/eJA=; b=Ew8/hmwWqEqO6m6ZOm4kPltPTNwFLYZTnZpE2LvLp70ddE5P0FWhmVCEOxLFmyNlgg SgxMckODTEmeAjNyZuL9jQVrqu6xvU0ABShijtjYAirmM5grgGxTMkI7NkUOM8Xvo5Xy md5KUo+JpCCiOo2Fu2DjhJU0h/bRvh4ILzutQNY7im6O8jxJiJyDW6TZLhPOMEXPel+G 7S271Vlc9msF75VItpY7pYmyhu7xWDIGgRN6dfGSUZsjgsLizHLGTy7H6wjbdwzMzAvZ rrE3Jn0ECZziUa7pbsOr5ufnKg3jNpVykLOgAhjUcsPcD6XbypnIZ3mpJH54XIyyqIyN 5geQ== ARC-Authentication-Results: i=1; mx.google.com; dkim=neutral (body hash did not verify) header.i=@gmail.com header.b=AF0Yl4aO; spf=pass (google.com: domain of ffmpeg-devel-bounces@ffmpeg.org designates 79.124.17.100 as permitted sender) smtp.mailfrom=ffmpeg-devel-bounces@ffmpeg.org; dmarc=fail (p=NONE sp=NONE dis=NONE) header.from=gmail.com Return-Path: Received: from ffbox0-bg.mplayerhq.hu (ffbox0-bg.ffmpeg.org. [79.124.17.100]) by mx.google.com with ESMTP id y23si4009146wrb.262.2017.07.08.02.12.31; Sat, 08 Jul 2017 02:12:32 -0700 (PDT) Received-SPF: pass (google.com: domain of ffmpeg-devel-bounces@ffmpeg.org designates 79.124.17.100 as permitted sender) client-ip=79.124.17.100; Authentication-Results: mx.google.com; dkim=neutral (body hash did not verify) header.i=@gmail.com header.b=AF0Yl4aO; spf=pass (google.com: domain of ffmpeg-devel-bounces@ffmpeg.org designates 79.124.17.100 as permitted sender) smtp.mailfrom=ffmpeg-devel-bounces@ffmpeg.org; dmarc=fail (p=NONE sp=NONE dis=NONE) header.from=gmail.com Received: from [127.0.1.1] (localhost [127.0.0.1]) by ffbox0-bg.mplayerhq.hu (Postfix) with ESMTP id 42522689B96; Sat, 8 Jul 2017 12:12:24 +0300 (EEST) X-Original-To: ffmpeg-devel@ffmpeg.org Delivered-To: ffmpeg-devel@ffmpeg.org Received: from mail-wr0-f193.google.com (mail-wr0-f193.google.com [209.85.128.193]) by ffbox0-bg.mplayerhq.hu (Postfix) with ESMTPS id E83B4689988 for ; Sat, 8 Jul 2017 12:12:17 +0300 (EEST) Received: by mail-wr0-f193.google.com with SMTP id z45so12764893wrb.2 for ; Sat, 08 Jul 2017 02:12:21 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20161025; h=from:to:subject:date:message-id:in-reply-to:references:mime-version :content-transfer-encoding; bh=4H3Vc/1WUj/1P7Ruc9mImHZbZXH7MJlMgL/C0mM7wPQ=; b=AF0Yl4aO13IDekOOSEV39yH3IKXrn3AXcb40R5jg3S0DGvXR6BxXQYoGZMNulg7sPj l4sdddw3lfEiFOar3TiHFDheMcYcljy6LPQLqyJhudvLvMGv7lyfID/KnZQrMsAcXcyX jzduWFQEQMdoVeEJNLfkSZSOgPPgKHv2GOxvbkpbpPrIdUE2DfehNuUKjZkhlUNlfXlV eHx8p2a2xA/i207cyWT46wr45ojBMntnA5pb5L0dOjbnioS3/4zVIAxykMqXYtARcl25 vp75hhawSuZcak0s1RkgSTCGYE506evpnTHPBom/W5jvygJO0TB20MdTKnf5kYwQELka M5XQ== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20161025; h=x-gm-message-state:from:to:subject:date:message-id:in-reply-to :references:mime-version:content-transfer-encoding; bh=4H3Vc/1WUj/1P7Ruc9mImHZbZXH7MJlMgL/C0mM7wPQ=; b=n1QtfO+mLy9Gvo9HbtleM5cwmSxCdpEpIT9lHcioKcPB8ABKshL+b9jKgUBq6bum5l U73fLOD+amesQXYqvj1mag+SlhbDiTswbmmyp9HQ/PrcE4a/zE4D8iZH5CC+FAF6xya/ h+laAbDsdPQo8Fw1qh8MyJLfz2ZUf0PqSZjlAdZzLOx4nt18ntJgDtIoVMlQcMFxeV2y oWqsOjEUkoaPYO7+YZt48TgxRnA7N446X0XflLJBYdVNPeY18CUkTwcQ6dotT1XrUtEZ lxnU/qOhjFQI+5N/rYCcTiGuzlAvLSVaA4o3OSQ+LIuSaMt1ssDbEpJTVOgrg30d8XGN mA7Q== X-Gm-Message-State: AIVw111Ac3sWY7HksyyYp5ychB44WdE9qj+T3JwrdR5TN/ZjU7IrP7tv 1XyQjbhny/rPQmQP X-Received: by 10.28.52.142 with SMTP id b136mr1791107wma.48.1499505140591; Sat, 08 Jul 2017 02:12:20 -0700 (PDT) Received: from localhost.localdomain ([94.250.174.60]) by smtp.gmail.com with ESMTPSA id 9sm3407624wml.25.2017.07.08.02.12.18 for (version=TLS1_2 cipher=ECDHE-RSA-AES128-GCM-SHA256 bits=128/128); Sat, 08 Jul 2017 02:12:19 -0700 (PDT) From: Paul B Mahol To: ffmpeg-devel@ffmpeg.org Date: Sat, 8 Jul 2017 11:12:06 +0200 Message-Id: <20170708091206.9293-1-onemda@gmail.com> X-Mailer: git-send-email 2.9.3 In-Reply-To: <20170707184848.20864-1-onemda@gmail.com> References: <20170707184848.20864-1-onemda@gmail.com> MIME-Version: 1.0 Subject: [FFmpeg-devel] [PATCH 1/3] avcodec/get_bits: add cached bitstream reader X-BeenThere: ffmpeg-devel@ffmpeg.org X-Mailman-Version: 2.1.20 Precedence: list List-Id: FFmpeg development discussions and patches List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Reply-To: FFmpeg development discussions and patches Errors-To: ffmpeg-devel-bounces@ffmpeg.org Sender: "ffmpeg-devel" Signed-off-by: Paul B Mahol --- libavcodec/get_bits.h | 261 +++++++++++++++++++++++++++++++++++++++++++++----- 1 file changed, 235 insertions(+), 26 deletions(-) diff --git a/libavcodec/get_bits.h b/libavcodec/get_bits.h index c530015..f404b80 100644 --- a/libavcodec/get_bits.h +++ b/libavcodec/get_bits.h @@ -1,5 +1,6 @@ /* - * copyright (c) 2004 Michael Niedermayer + * Copyright (c) 2004 Michael Niedermayer + * Copyright (c) 2016 Alexandra Hájková * * This file is part of FFmpeg. * @@ -54,6 +55,10 @@ typedef struct GetBitContext { const uint8_t *buffer, *buffer_end; +#ifdef CACHED_BITSTREAM_READER + uint64_t cache; + unsigned bits_left; +#endif int index; int size_in_bits; int size_in_bits_plus8; @@ -106,12 +111,16 @@ typedef struct GetBitContext { * For examples see get_bits, show_bits, skip_bits, get_vlc. */ -#ifdef LONG_BITSTREAM_READER +#ifdef CACHED_BITSTREAM_READER +# define MIN_CACHE_BITS 64 +#elif defined LONG_BITSTREAM_READER # define MIN_CACHE_BITS 32 #else # define MIN_CACHE_BITS 25 #endif +#ifndef CACHED_BITSTREAM_READER + #define OPEN_READER_NOSIZE(name, gb) \ unsigned int name ## _index = (gb)->index; \ unsigned int av_unused name ## _cache @@ -196,20 +205,113 @@ typedef struct GetBitContext { #define GET_CACHE(name, gb) ((uint32_t) name ## _cache) +#endif + static inline int get_bits_count(const GetBitContext *s) { +#ifdef CACHED_BITSTREAM_READER + return s->index - s->bits_left; +#else return s->index; +#endif } -static inline void skip_bits_long(GetBitContext *s, int n) +static inline void refill_32(GetBitContext *s) { -#if UNCHECKED_BITSTREAM_READER - s->index += n; +#ifdef CACHED_BITSTREAM_READER +#if !UNCHECKED_BITSTREAM_READER + if (s->index >> 3 >= s->buffer_end - s->buffer) + return; +#endif + +#ifdef BITSTREAM_READER_LE + s->cache = (uint64_t)AV_RL32(s->buffer + (s->index >> 3)) << s->bits_left | s->cache; #else - s->index += av_clip(n, -s->index, s->size_in_bits_plus8 - s->index); + s->cache = s->cache | (uint64_t)AV_RB32(s->buffer + (s->index >> 3)) << (32 - s->bits_left); +#endif + s->index += 32; + s->bits_left += 32; +#endif +} + +static inline void refill_64(GetBitContext *s) +{ +#ifdef CACHED_BITSTREAM_READER +#if !UNCHECKED_BITSTREAM_READER + if (s->index >> 3 >= s->buffer_end - s->buffer) + return; +#endif + +#ifdef BITSTREAM_READER_LE + s->cache = AV_RL64(s->buffer + (s->index >> 3)); +#else + s->cache = AV_RB64(s->buffer + (s->index >> 3)); +#endif + s->index += 64; + s->bits_left = 64; +#endif +} + +#ifdef CACHED_BITSTREAM_READER +static inline uint64_t get_val(GetBitContext *s, unsigned n) +{ + uint64_t ret; + av_assert2(n>0 && n<=63); +#ifdef BITSTREAM_READER_LE + ret = s->cache & ((UINT64_C(1) << n) - 1); + s->cache >>= n; +#else + ret = s->cache >> (64 - n); + s->cache <<= n; +#endif + s->bits_left -= n; + return ret; +} +#endif + +#ifdef CACHED_BITSTREAM_READER +static inline unsigned show_val(const GetBitContext *s, unsigned n) +{ +#ifdef BITSTREAM_READER_LE + return s->cache & ((UINT64_C(1) << n) - 1); +#else + return s->cache >> (64 - n); +#endif +} +#endif + +/** + * Show 1-25 bits. + */ +static inline unsigned int show_bits(GetBitContext *s, int n) +{ + register int tmp; +#ifdef CACHED_BITSTREAM_READER + if (n > s->bits_left) + refill_32(s); + + tmp = show_val(s, n); +#else + OPEN_READER_NOSIZE(re, s); + av_assert2(n>0 && n<=25); + UPDATE_CACHE(re, s); + tmp = SHOW_UBITS(re, s, n); #endif + return tmp; } +#ifdef CACHED_BITSTREAM_READER +static inline void skip_remaining(GetBitContext *s, unsigned n) +{ +#ifdef BITSTREAM_READER_LE + s->cache >>= n; +#else + s->cache <<= n; +#endif + s->bits_left -= n; +} +#endif + /** * Read MPEG-1 dc-style VLC (sign bit + mantissa with no MSB). * if MSB not set it is negative @@ -217,6 +319,13 @@ static inline void skip_bits_long(GetBitContext *s, int n) */ static inline int get_xbits(GetBitContext *s, int n) { +#ifdef CACHED_BITSTREAM_READER + int32_t cache = show_bits(s, 32); + int sign = ~cache >> 31; + skip_remaining(s, n); + + return ((((uint32_t)(sign ^ cache)) >> (32 - n)) ^ sign) - sign; +#else register int sign; register int32_t cache; OPEN_READER(re, s); @@ -227,8 +336,10 @@ static inline int get_xbits(GetBitContext *s, int n) LAST_SKIP_BITS(re, s, n); CLOSE_READER(re, s); return (NEG_USR32(sign ^ cache, n) ^ sign) - sign; +#endif } +#ifndef CACHED_BITSTREAM_READER static inline int get_xbits_le(GetBitContext *s, int n) { register int sign; @@ -242,31 +353,61 @@ static inline int get_xbits_le(GetBitContext *s, int n) CLOSE_READER(re, s); return (zero_extend(sign ^ cache, n) ^ sign) - sign; } +#endif -static inline int get_sbits(GetBitContext *s, int n) +/** + * Read 1-25 bits. + */ +static inline unsigned int get_bits(GetBitContext *s, int n) { +#ifdef CACHED_BITSTREAM_READER + register int tmp = 0; +#ifdef BITSTREAM_READER_LE + uint64_t left = 0; +#endif + + av_assert2(n>0 && n<=32); + if (n > s->bits_left) { + n -= s->bits_left; +#ifdef BITSTREAM_READER_LE + left = s->bits_left; +#endif + tmp = get_val(s, s->bits_left); + refill_32(s); + } + +#ifdef BITSTREAM_READER_LE + tmp = get_val(s, n) << left | tmp; +#else + tmp = get_val(s, n) | tmp << n; +#endif + +#else register int tmp; OPEN_READER(re, s); av_assert2(n>0 && n<=25); UPDATE_CACHE(re, s); - tmp = SHOW_SBITS(re, s, n); + tmp = SHOW_UBITS(re, s, n); LAST_SKIP_BITS(re, s, n); CLOSE_READER(re, s); +#endif return tmp; } -/** - * Read 1-25 bits. - */ -static inline unsigned int get_bits(GetBitContext *s, int n) +static inline int get_sbits(GetBitContext *s, int n) { register int tmp; +#ifdef CACHED_BITSTREAM_READER + av_assert2(n>0 && n<=25); + tmp = sign_extend(get_bits(s, n), n); +#else OPEN_READER(re, s); av_assert2(n>0 && n<=25); UPDATE_CACHE(re, s); - tmp = SHOW_UBITS(re, s, n); + tmp = SHOW_SBITS(re, s, n); LAST_SKIP_BITS(re, s, n); CLOSE_READER(re, s); +#endif return tmp; } @@ -278,6 +419,7 @@ static av_always_inline int get_bitsz(GetBitContext *s, int n) return n ? get_bits(s, n) : 0; } +#ifndef CACHED_BITSTREAM_READER static inline unsigned int get_bits_le(GetBitContext *s, int n) { register int tmp; @@ -289,29 +431,54 @@ static inline unsigned int get_bits_le(GetBitContext *s, int n) CLOSE_READER(re, s); return tmp; } - -/** - * Show 1-25 bits. - */ -static inline unsigned int show_bits(GetBitContext *s, int n) -{ - register int tmp; - OPEN_READER_NOSIZE(re, s); - av_assert2(n>0 && n<=25); - UPDATE_CACHE(re, s); - tmp = SHOW_UBITS(re, s, n); - return tmp; -} +#endif static inline void skip_bits(GetBitContext *s, int n) { +#ifdef CACHED_BITSTREAM_READER + if (n <= s->bits_left) + skip_remaining(s, n); + else { + n -= s->bits_left; + skip_remaining(s, s->bits_left); + if (n >= 64) { + unsigned skip = n; + + n -= skip; + s->index += skip; + } + refill_32(s); + if (n) + skip_remaining(s, n); + } +#else OPEN_READER(re, s); LAST_SKIP_BITS(re, s, n); CLOSE_READER(re, s); +#endif +} + +static inline void skip_bits_long(GetBitContext *s, int n) +{ +#ifdef CACHED_BITSTREAM_READER + skip_bits(s, n); +#else +#if UNCHECKED_BITSTREAM_READER + s->index += n; +#else + s->index += av_clip(n, -s->index, s->size_in_bits_plus8 - s->index); +#endif +#endif } static inline unsigned int get_bits1(GetBitContext *s) { +#ifdef CACHED_BITSTREAM_READER + if (!s->bits_left) + refill_64(s); + + return get_val(s, 1); +#else unsigned int index = s->index; uint8_t result = s->buffer[index >> 3]; #ifdef BITSTREAM_READER_LE @@ -328,6 +495,7 @@ static inline unsigned int get_bits1(GetBitContext *s) s->index = index; return result; +#endif } static inline unsigned int show_bits1(GetBitContext *s) @@ -348,6 +516,10 @@ static inline unsigned int get_bits_long(GetBitContext *s, int n) av_assert2(n>=0 && n<=32); if (!n) { return 0; +#ifdef CACHED_BITSTREAM_READER + } + return get_bits(s, n); +#else } else if (n <= MIN_CACHE_BITS) { return get_bits(s, n); } else { @@ -359,6 +531,7 @@ static inline unsigned int get_bits_long(GetBitContext *s, int n) return ret | get_bits(s, n - 16); #endif } +#endif } /** @@ -442,6 +615,10 @@ static inline int init_get_bits(GetBitContext *s, const uint8_t *buffer, s->buffer_end = buffer + buffer_size; s->index = 0; +#ifdef CACHED_BITSTREAM_READER + refill_64(s); +#endif + return ret; } @@ -543,6 +720,19 @@ static inline const uint8_t *align_get_bits(GetBitContext *s) SKIP_BITS(name, gb, n); \ } while (0) +/* Return the LUT element for the given bitstream configuration. */ +static inline int set_idx(GetBitContext *s, int code, int *n, int *nb_bits, + VLC_TYPE (*table)[2]) +{ + unsigned idx; + + *nb_bits = -*n; + idx = show_bits(s, *nb_bits) + code; + *n = table[idx][1]; + + return table[idx][0]; +} + /** * Parse a vlc code. * @param bits is the number of bits which will be read at once, must be @@ -554,6 +744,24 @@ static inline const uint8_t *align_get_bits(GetBitContext *s) static av_always_inline int get_vlc2(GetBitContext *s, VLC_TYPE (*table)[2], int bits, int max_depth) { +#ifdef CACHED_BITSTREAM_READER + int nb_bits; + unsigned idx = show_bits(s, bits); + int code = table[idx][0]; + int n = table[idx][1]; + + if (max_depth > 1 && n < 0) { + skip_remaining(s, bits); + code = set_idx(s, code, &n, &nb_bits, table); + if (max_depth > 2 && n < 0) { + skip_remaining(s, nb_bits); + code = set_idx(s, code, &n, &nb_bits, table); + } + } + skip_remaining(s, n); + + return code; +#else int code; OPEN_READER(re, s); @@ -564,6 +772,7 @@ static av_always_inline int get_vlc2(GetBitContext *s, VLC_TYPE (*table)[2], CLOSE_READER(re, s); return code; +#endif } static inline int decode012(GetBitContext *gb)