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
Subject: Re: [FFmpeg-devel] [PATCH 8/8] avcodec/codec_internal: Include codec_tags only when they are needed
Date: Fri, 18 Mar 2022 14:37:12 +0100
Message-ID: <AS1PR01MB9564DE897C1A5030E01FAE618F139@AS1PR01MB9564.eurprd01.prod.exchangelabs.com> (raw)
In-Reply-To: <20220318131316.GU2829255@pb2>

Michael Niedermayer:
> On Fri, Mar 18, 2022 at 11:52:54AM +0100, Andreas Rheinhardt wrote:
>> They are only needed for the fuzzer, so check for CONFIG_OSSFUZZ.
>> This decreases sizeof(FFCodec), which is important given that
>> FFCodecs reside in .data.rel.ro in case of ELF with
>> position-independent code which is always loaded and can't be shared
>> between processes.
>>
> 
> They are currently only used by the fuzzer, if there is no other
> use, iam not sure. But i agree shareable memory would be better for them
> 

GCC by default aligns big enough objects to 32* (so while sizeof(codec)
need not be a multiple of said alignment, the actual object is still
padded to said alignment) and it was aligned to 32 before the addition
of ch_layouts (for x64), so adding it increased the size by 32 (and
resulted in over 21KiB size increase here on a really
not-all-encompassing build). Optimizing codec_tags away would reverse this.

> 
> [...]
>> diff --git a/tools/target_dec_fuzzer.c b/tools/target_dec_fuzzer.c
>> index 288aa63313..77f4bb8dd8 100644
>> --- a/tools/target_dec_fuzzer.c
>> +++ b/tools/target_dec_fuzzer.c
>> @@ -279,12 +279,14 @@ int LLVMFuzzerTestOneInput(const uint8_t *data, size_t size) {
>>          ctx->sample_rate                        = bytestream2_get_le32(&gbc) & 0x7FFFFFFF;
>>          ctx->ch_layout.nb_channels              = (unsigned)bytestream2_get_le32(&gbc) % FF_SANE_NB_CHANNELS;
>>          ctx->block_align                        = bytestream2_get_le32(&gbc) & 0x7FFFFFFF;
>> +#if CONFIG_OSSFUZZ
>>          ctx->codec_tag                          = bytestream2_get_le32(&gbc);
>>          if (c->codec_tags) {
>>              int n;
>>              for (n = 0; c->codec_tags[n] != FF_CODEC_TAGS_END; n++);
>>              ctx->codec_tag = c->codec_tags[ctx->codec_tag % n];
>>          }
>> +#endif
>>          keyframes                               = bytestream2_get_le64(&gbc);
>>          request_channel_layout                  = bytestream2_get_le64(&gbc);
>>  
> 
> how does the fuzzer work without the fuzzer ?
> 

You could just not use --enable-ossfuzz, but nevertheless want to "make
tools/target_dec_foo_fuzzer" (with custom fuzzer CFLAGS/LDFLAGS). But it
is not really the intended usecase, so I made sure it compiles, but
nothing more.

- Andreas

*: The ELF-x64-ABI only requires 16, but it uses 32 for compatibility
with ancient versions of GCC which required 32. One can use
-malign-data=abi to make it only use 16.
_______________________________________________
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:[~2022-03-18 13:37 UTC|newest]

Thread overview: 13+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-03-16 21:45 [FFmpeg-devel] [PATCH 1/3] avcodec/internal: Move FF_CODEC_CAP_* to a new header codec_internal.h Andreas Rheinhardt
2022-03-16 21:47 ` [FFmpeg-devel] [PATCH 2/3] avcodec/codec_internal: Add FFCodec, hide internal part of AVCodec Andreas Rheinhardt
2022-03-16 21:47 ` [FFmpeg-devel] [PATCH 3/3] avcodec/codec_internal: Rename AVCodecDefault->FFCodecDefault Andreas Rheinhardt
2022-03-17 12:43 ` [FFmpeg-devel] [PATCH 4/7] avcodec/mathops: Move bitswap_32() to its only user Andreas Rheinhardt
2022-03-17 12:43 ` [FFmpeg-devel] [PATCH 5/7] avcodec/bitstream: Move code for initializing VLCs to file of its own Andreas Rheinhardt
2022-03-17 12:43 ` [FFmpeg-devel] [PATCH 6/7] avcodec/internal: Move FF_SIGNBIT and ff_log2_run to mathops.h Andreas Rheinhardt
2022-03-17 12:43 ` [FFmpeg-devel] [PATCH 7/7] avcodec/internal: Move FF_DEFAULT_QUANT_BIAS to mpegvideoenc.h Andreas Rheinhardt
2022-03-18 10:52 ` [FFmpeg-devel] [PATCH 8/8] avcodec/codec_internal: Include codec_tags only when they are needed Andreas Rheinhardt
2022-03-18 13:13   ` Michael Niedermayer
2022-03-18 13:31     ` Michael Niedermayer
2022-03-18 13:39       ` Andreas Rheinhardt
2022-03-18 13:37     ` Andreas Rheinhardt [this message]
2022-03-31  8:26   ` 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=AS1PR01MB9564DE897C1A5030E01FAE618F139@AS1PR01MB9564.eurprd01.prod.exchangelabs.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