Git Inbox Mirror of the ffmpeg-devel mailing list - see https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
 help / color / mirror / Atom feed
From: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
To: ffmpeg-devel@ffmpeg.org
Cc: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Subject: [FFmpeg-devel] [PATCH 5/9] avcodec/dvdec: Avoid stack buffers
Date: Sun,  4 Sep 2022 23:58:47 +0200
Message-ID: <AS8P250MB07445E61F3BA307146337ED28F7C9@AS8P250MB0744.EURP250.PROD.OUTLOOK.COM> (raw)
In-Reply-To: <AS8P250MB07446C8541237B01350DEAB78F7C9@AS8P250MB0744.EURP250.PROD.OUTLOOK.COM>

Instead reuse the destination RL VLC as scratch space.
This is possible, because the (implicit) codes here are already
ordered from left-to-right in the tree and because the codelengths
are increasing, which implies that mapping from VLC entries to the
corresponding entries used to initialize the VLC is monotonically
increasing. This means that one can reuse the right end of the
destination RL VLC to store the tables used to initialize the VLC
with.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
---
 libavcodec/dvdata.h |  2 ++
 libavcodec/dvdec.c  | 26 +++++++++++++-------------
 2 files changed, 15 insertions(+), 13 deletions(-)

diff --git a/libavcodec/dvdata.h b/libavcodec/dvdata.h
index ae104096ad..31191a8475 100644
--- a/libavcodec/dvdata.h
+++ b/libavcodec/dvdata.h
@@ -27,6 +27,8 @@ extern const uint8_t ff_dv_quant_shifts[22][4];
 extern const uint8_t ff_dv_quant_offset[4];
 
 #define NB_DV_VLC 409
+/* The number of entries with value zero in ff_dv_vlc_level. */
+#define NB_DV_ZERO_LEVEL_ENTRIES 72
 
 extern const uint8_t ff_dv_vlc_len[NB_DV_VLC];
 extern const uint8_t ff_dv_vlc_run[NB_DV_VLC];
diff --git a/libavcodec/dvdec.c b/libavcodec/dvdec.c
index 3857ed1266..32085a3ba6 100644
--- a/libavcodec/dvdec.c
+++ b/libavcodec/dvdec.c
@@ -137,31 +137,30 @@ static av_cold void dv_init_static(void)
 {
     VLCElem vlc_buf[FF_ARRAY_ELEMS(dv_rl_vlc)] = { 0 };
     VLC dv_vlc = { .table = vlc_buf, .table_allocated = FF_ARRAY_ELEMS(vlc_buf) };
-    uint8_t    new_dv_vlc_len[NB_DV_VLC * 2];
-    uint8_t    new_dv_vlc_run[NB_DV_VLC * 2];
-    int16_t  new_dv_vlc_level[NB_DV_VLC * 2];
+    const unsigned offset = FF_ARRAY_ELEMS(dv_rl_vlc) - (2 * NB_DV_VLC - NB_DV_ZERO_LEVEL_ENTRIES);
+    RL_VLC_ELEM *tmp = dv_rl_vlc + offset;
     int i, j;
 
     /* it's faster to include sign bit in a generic VLC parsing scheme */
     for (i = 0, j = 0; i < NB_DV_VLC; i++, j++) {
-        new_dv_vlc_len[j]   = ff_dv_vlc_len[i];
-        new_dv_vlc_run[j]   = ff_dv_vlc_run[i];
-        new_dv_vlc_level[j] = ff_dv_vlc_level[i];
+        tmp[j].len   = ff_dv_vlc_len[i];
+        tmp[j].run   = ff_dv_vlc_run[i];
+        tmp[j].level = ff_dv_vlc_level[i];
 
         if (ff_dv_vlc_level[i]) {
-            new_dv_vlc_len[j]++;
+            tmp[j].len++;
 
             j++;
-            new_dv_vlc_len[j]   =  ff_dv_vlc_len[i] + 1;
-            new_dv_vlc_run[j]   =  ff_dv_vlc_run[i];
-            new_dv_vlc_level[j] = -ff_dv_vlc_level[i];
+            tmp[j].len   =  ff_dv_vlc_len[i] + 1;
+            tmp[j].run   =  ff_dv_vlc_run[i];
+            tmp[j].level = -ff_dv_vlc_level[i];
         }
     }
 
     /* NOTE: as a trick, we use the fact the no codes are unused
      * to accelerate the parsing of partial codes */
     ff_init_vlc_from_lengths(&dv_vlc, TEX_VLC_BITS, j,
-                             new_dv_vlc_len, 1,
+                             &tmp[0].len, sizeof(tmp[0]),
                              NULL, 0, 0, 0, INIT_VLC_USE_NEW_STATIC, NULL);
     av_assert1(dv_vlc.table_size == 1664);
 
@@ -174,8 +173,9 @@ static av_cold void dv_init_static(void)
             run   = 0;
             level = code;
         } else {
-            run   = new_dv_vlc_run[code] + 1;
-            level = new_dv_vlc_level[code];
+            av_assert1(i <= code + offset);
+            run   = tmp[code].run + 1;
+            level = tmp[code].level;
         }
         dv_rl_vlc[i].len   = len;
         dv_rl_vlc[i].level = level;
-- 
2.34.1

_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".

  parent reply	other threads:[~2022-09-04 21:59 UTC|newest]

Thread overview: 10+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-09-04 21:54 [FFmpeg-devel] [PATCH 1/9] avcodec/dvdata: Order code table by codes Andreas Rheinhardt
2022-09-04 21:58 ` [FFmpeg-devel] [PATCH 2/9] avcodec/dvdec: Use ff_init_vlc_from_lengths() Andreas Rheinhardt
2022-09-04 21:58 ` [FFmpeg-devel] [PATCH 3/9] avcodec/dv_tablegen, dvdata: Remove ff_dv_vlc_bits Andreas Rheinhardt
2022-09-04 21:58 ` [FFmpeg-devel] [PATCH 4/9] avcodec/dvdec: Mark dv_init_static() as av_cold Andreas Rheinhardt
2022-09-04 21:58 ` Andreas Rheinhardt [this message]
2022-09-04 21:58 ` [FFmpeg-devel] [PATCH 6/9] avcodec/wmavoice: Avoid code table Andreas Rheinhardt
2022-09-04 21:58 ` [FFmpeg-devel] [PATCH 7/9] avcodec/wmaprodec: Use ff_init_vlc_from_lengths() instead of init_vlc Andreas Rheinhardt
2022-09-04 21:58 ` [FFmpeg-devel] [PATCH 8/9] avcodec/wmaprodec: Move applying offset to VLC creation Andreas Rheinhardt
2022-09-04 21:58 ` [FFmpeg-devel] [PATCH 9/9] avcodec/wmaprodec: Use symbol table more efficiently Andreas Rheinhardt
2022-09-06  1:33 ` [FFmpeg-devel] [PATCH 1/9] avcodec/dvdata: Order code table by codes Andreas Rheinhardt

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=AS8P250MB07445E61F3BA307146337ED28F7C9@AS8P250MB0744.EURP250.PROD.OUTLOOK.COM \
    --to=andreas.rheinhardt@outlook.com \
    --cc=ffmpeg-devel@ffmpeg.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link

Git Inbox Mirror of the ffmpeg-devel mailing list - see https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

This inbox may be cloned and mirrored by anyone:

	git clone --mirror https://master.gitmailbox.com/ffmpegdev/0 ffmpegdev/git/0.git

	# If you have public-inbox 1.1+ installed, you may
	# initialize and index your mirror using the following commands:
	public-inbox-init -V2 ffmpegdev ffmpegdev/ https://master.gitmailbox.com/ffmpegdev \
		ffmpegdev@gitmailbox.com
	public-inbox-index ffmpegdev

Example config snippet for mirrors.


AGPL code for this site: git clone https://public-inbox.org/public-inbox.git