From: Timo Rothenpieler <timo@rothenpieler.org>
To: ffmpeg-devel@ffmpeg.org
Subject: Re: [FFmpeg-devel] [PATCH] avutil/mem: always align by at least 32 bytes
Date: Wed, 6 Dec 2023 14:27:35 +0100
Message-ID: <03598d96-68a9-4a92-9e6d-ceb205107ead@rothenpieler.org> (raw)
In-Reply-To: <6ebfe490-d27e-14c3-ac81-8c4a2446ba17@martin.st>
On 06/12/2023 14:25, Martin Storsjö wrote:
> On Sun, 3 Dec 2023, Timo Rothenpieler wrote:
>
>> FFmpeg has instances of DECLARE_ALIGNED(32, ...) in a lot of structs,
>> which then end up heap-allocated.
>> By declaring any variable in a struct, or tree of structs, to be 32 byte
>> aligned, it allows the compiler to safely assume the entire struct
>> itself is also 32 byte aligned.
>>
>> This might make the compiler emit code which straight up crashes or
>> misbehaves in other ways, and at least in one instances is now
>> documented to actually do (see ticket 10549 on trac).
>> The issue there is that an unrelated variable in SingleChannelElement is
>> declared to have an alignment of 32 bytes. So if the compiler does a copy
>> in decode_cpe() with avx instructions, but ffmpeg is built with
>> --disable-avx, this results in a crash, since the memory is only 16 byte
>> aligned.
>> Mind you, even if the compiler does not emit avx instructions, the code
>> is still invalid and could misbehave. It just happens not to. Declaring
>> any variable in a struct with a 32 byte alignment promises 32 byte
>> alignment of the whole struct to the compiler.
>>
>> Instead of now going through all instances of variables in structs
>> being declared as 32 byte aligned, this patch bumps the minimum alignment
>> to 32 bytes.
>> ---
>> libavutil/mem.c | 2 +-
>> 1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/libavutil/mem.c b/libavutil/mem.c
>> index 36b8940a0c..26a9b9753b 100644
>> --- a/libavutil/mem.c
>> +++ b/libavutil/mem.c
>> @@ -62,7 +62,7 @@ void free(void *ptr);
>>
>> #endif /* MALLOC_PREFIX */
>>
>> -#define ALIGN (HAVE_AVX512 ? 64 : (HAVE_AVX ? 32 : 16))
>> +#define ALIGN (HAVE_AVX512 ? 64 : 32)
>
> LGTM
>
> It could be good to add a comment here, to indicate how this value
> relates to the alignemnts used in structs.
>
> For others who commented in this thread, it all boils down to something
> like this:
>
> struct MyData {
> uint8_t __attribute__((aligned(32))) aligned_data[1024];
> };
It's even a bit more complex than that.
The case that's crashing right now is a member that has no alignment
declared on itself at all.
But another member of the same struct does, and so the compiler assumes
the whole struct to be aligned.
> void func(void) {
> struct MyData *obj = malloc(sizeof(*obj)); // Uusally only aligned
> to 8 bytes
> // operate on obj->aligned_data[]
> }
>
> Due to how aligned_data is declared, we promise to the compiler that it
> is aligned to 32 bytes, and that the compiler can assume this wherever.
> Depending on -march or whatever, this can be to access it with
> instructions that assume 32 byte alignment.
>
> // Martin
>
> _______________________________________________
> 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".
_______________________________________________
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".
next prev parent reply other threads:[~2023-12-06 13:27 UTC|newest]
Thread overview: 29+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-12-03 20:10 Timo Rothenpieler
2023-12-06 12:27 ` Timo Rothenpieler
2023-12-06 12:31 ` James Almer
2023-12-06 12:56 ` Timo Rothenpieler
2023-12-06 12:50 ` Ronald S. Bultje
2023-12-06 12:54 ` James Almer
2023-12-06 13:25 ` Martin Storsjö
2023-12-06 13:27 ` Timo Rothenpieler [this message]
2023-12-06 13:29 ` Martin Storsjö
2023-12-08 0:15 ` Timo Rothenpieler
2023-12-08 5:57 ` Martin Storsjö
2023-12-08 10:01 ` Andreas Rheinhardt
2023-12-08 17:56 ` Timo Rothenpieler
2023-12-08 18:11 ` Nicolas George
2023-12-09 5:23 ` Andreas Rheinhardt
2024-01-12 23:10 ` Timo Rothenpieler
2024-01-13 0:57 ` [FFmpeg-devel] [PATCH] avutil/mem: limit alignment to maximum simg align Timo Rothenpieler
2024-01-13 1:00 ` Timo Rothenpieler
2024-01-13 15:24 ` Timo Rothenpieler
2024-01-13 15:46 ` [FFmpeg-devel] [PATCH v2] avutil/mem: limit alignment to maximum simd align Timo Rothenpieler
2024-02-09 19:22 ` Timo Rothenpieler
2024-02-11 14:05 ` Sam James
2024-02-11 14:22 ` Rémi Denis-Courmont
2024-02-11 15:47 ` Timo Rothenpieler
2024-02-11 14:00 ` Andreas Rheinhardt
2024-02-11 16:06 ` Timo Rothenpieler
2024-02-11 17:40 ` [FFmpeg-devel] [PATCH] " Timo Rothenpieler
2024-02-26 16:58 ` Timo Rothenpieler
2024-02-27 18:45 ` Timo Rothenpieler
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=03598d96-68a9-4a92-9e6d-ceb205107ead@rothenpieler.org \
--to=timo@rothenpieler.org \
--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