From: Nuo Mi <nuomi2021@gmail.com> To: FFmpeg development discussions and patches <ffmpeg-devel@ffmpeg.org> Subject: Re: [FFmpeg-devel] [PATCH v3 2/2] lavc/vvc: Ensure subpictures don't overlap Date: Sun, 23 Feb 2025 21:44:17 +0800 Message-ID: <CAFXK13ct0EJXuVmqk2xji9EGhhDm=S7ybWGHMdewf3W8S3AAyw@mail.gmail.com> (raw) In-Reply-To: <f5747bb2-fd4f-479d-b36a-08f19f74d325@frankplowman.com> On Sun, Feb 23, 2025 at 1:36 AM Frank Plowman <post@frankplowman.com> wrote: > On 22/02/2025 07:51, Nuo Mi wrote: > > From: Frank Plowman <post@frankplowman.com> > > > > This is essentially a re-implementation of > > > https://patchwork.ffmpeg.org/project/ffmpeg/patch/20241005223955.54158-1-post@frankplowman.com/ > > > > That patch was not applied last time. Instead we opted to identify > > issues which could be caused by invalid subpicture layouts and remedy > > those issues where they manifest, either through error detection or code > > hardening. This was primarily implemented in the set > > https://patchwork.ffmpeg.org/project/ffmpeg/list/?series=13381. > > > > This has worked to some degree, however issues with subpicture layouts > > continue to crop up from the fuzzer and I've fixed a number of bugs > > related to subpicture layouts since then. I think it's best to return > > to the initial plan and simply check if the subpicture layout is valid > > initially. > > > > This implementation is also lighter than the first time -- by doing a > > bit more logic in pps_subpic_less_than_one_tile_slice, we are able to > > store a tile_in_subpic map rather than a ctu_in_subpic map. This > > reduces the size of the map to the point it becomes possible to allocate > > it on the stack. Similar to 8bd66a8c9587af61c7b46558be3c4ee317c1af5a, > > the layout is also validated in the slice map construction code, rather > > than in the CBS, which avoids duplicating some logic. > > > > Signed-off-by: Frank Plowman <post@frankplowman.com> > > --- > > libavcodec/vvc/ps.c | 55 ++++++++++++++++++++++++++++++++++++--------- > > 1 file changed, 45 insertions(+), 10 deletions(-) > > > > diff --git a/libavcodec/vvc/ps.c b/libavcodec/vvc/ps.c > > index d8794277d2..9b8af5524b 100644 > > --- a/libavcodec/vvc/ps.c > > +++ b/libavcodec/vvc/ps.c > > @@ -450,25 +450,50 @@ static void subpic_tiles(int *tile_x, int *tile_y, > int *tile_x_end, int *tile_y_ > > (*tile_y_end)++; > > } > > > > -static void pps_subpic_less_than_one_tile_slice(VVCPPS *pps, const > VVCSPS *sps, const int i, const int tx, const int ty, int *off) > > +static bool mark_tile_as_used(bool *tile_in_subpic, const int tx, const > int ty, const int tile_columns) > > { > > + const size_t tile_idx = ty * tile_columns + tx; > > + if (tile_in_subpic[tile_idx]) { > > + /* the tile is covered by other subpictures */ > > + return false; > > + } > > + tile_in_subpic[tile_idx] = true; > > + return true; > > +} > > + > > +static int pps_subpic_less_than_one_tile_slice(VVCPPS *pps, const > VVCSPS *sps, const int i, const int tx, const int ty, int *off, bool > *tile_in_subpic) > > +{ > > + const int subpic_bottom = sps->r->sps_subpic_ctu_top_left_y[i] + > sps->r->sps_subpic_height_minus1[i]; > > + const int tile_bottom = pps->row_bd[ty] + > pps->r->row_height_val[ty] - 1; > > + const bool is_final_subpic_in_tile = subpic_bottom == tile_bottom; > > + > > + if (is_final_subpic_in_tile && !mark_tile_as_used(tile_in_subpic, > tx, ty, pps->r->num_tile_columns)) > > + return AVERROR_INVALIDDATA; > > + > > pps->num_ctus_in_slice[i] = pps_add_ctus(pps, off, > > sps->r->sps_subpic_ctu_top_left_x[i], > sps->r->sps_subpic_ctu_top_left_y[i], > > sps->r->sps_subpic_width_minus1[i] + 1, > sps->r->sps_subpic_height_minus1[i] + 1); > > + > > + return 0; > > } > > > > -static void pps_subpic_one_or_more_tiles_slice(VVCPPS *pps, const int > tile_x, const int tile_y, const int x_end, const int y_end, const int i, > int *off) > > +static int pps_subpic_one_or_more_tiles_slice(VVCPPS *pps, const int > tile_x, const int tile_y, const int x_end, const int y_end, > > + const int i, int *off, bool *tile_in_subpic) > > { > > for (int ty = tile_y; ty < y_end; ty++) { > > for (int tx = tile_x; tx < x_end; tx++) { > > + if (!mark_tile_as_used(tile_in_subpic, tx, ty, > pps->r->num_tile_columns)) > > + return AVERROR_INVALIDDATA; > > + > > pps->num_ctus_in_slice[i] += pps_add_ctus(pps, off, > > pps->col_bd[tx], pps->row_bd[ty], > > pps->r->col_width_val[tx], pps->r->row_height_val[ty]); > > } > > } > > + return 0; > > } > > > > -static void pps_subpic_slice(VVCPPS *pps, const VVCSPS *sps, const int > i, int *off) > > +static int pps_subpic_slice(VVCPPS *pps, const VVCSPS *sps, const int > i, int *off, bool *tile_in_subpic) > > { > > int tx, ty, x_end, y_end; > > > > @@ -477,19 +502,30 @@ static void pps_subpic_slice(VVCPPS *pps, const > VVCSPS *sps, const int i, int *o > > > > subpic_tiles(&tx, &ty, &x_end, &y_end, sps, pps, i); > > if (ty + 1 == y_end && sps->r->sps_subpic_height_minus1[i] + 1 < > pps->r->row_height_val[ty]) > > - pps_subpic_less_than_one_tile_slice(pps, sps, i, tx, ty, off); > > + return pps_subpic_less_than_one_tile_slice(pps, sps, i, tx, ty, > off, tile_in_subpic); > > else > > - pps_subpic_one_or_more_tiles_slice(pps, tx, ty, x_end, y_end, > i, off); > > + return pps_subpic_one_or_more_tiles_slice(pps, tx, ty, x_end, > y_end, i, off, tile_in_subpic); > > } > > > > -static void pps_single_slice_per_subpic(VVCPPS *pps, const VVCSPS *sps, > int *off) > > +static int pps_single_slice_per_subpic(VVCPPS *pps, const VVCSPS *sps, > int *off) > > { > > if (!sps->r->sps_subpic_info_present_flag) { > > pps_single_slice_picture(pps, off); > > } else { > > - for (int i = 0; i < pps->r->pps_num_slices_in_pic_minus1 + 1; > i++) > > - pps_subpic_slice(pps, sps, i, off); > > + bool tile_in_subpic[VVC_MAX_TILES_PER_AU] = {0}; > > + for (int i = 0; i < pps->r->pps_num_slices_in_pic_minus1 + 1; > i++) { > > + const int ret = pps_subpic_slice(pps, sps, i, off, > tile_in_subpic); > > + if (ret < 0) > > + return ret; > > + } > > + > > + // We only use tile_in_subpic to check that the subpictures > don't overlap > > + // here; we don't use tile_in_subpic to check that the > subpictures cover > > + // every tile. It is possible to avoid doing this work here > because the > > + // covering property of subpictures is already guaranteed by > the mechanisms > > + // which check every CTU belongs to a slice. > > } > > + return 0; > > } > > > > static int pps_one_tile_slices(VVCPPS *pps, const int tile_idx, int i, > int *off) > > @@ -540,8 +576,7 @@ static int pps_rect_slice(VVCPPS *pps, const VVCSPS > *sps) > > int tile_idx = 0, off = 0; > > > > if (r->pps_single_slice_per_subpic_flag) { > > - pps_single_slice_per_subpic(pps, sps, &off); > > - return 0; > > + return pps_single_slice_per_subpic(pps, sps, &off); > > } > > > > for (int i = 0; i < r->pps_num_slices_in_pic_minus1 + 1; i++) { > > Thanks, LGTM! > Thank you for the patch and review. Applied. > > -- > Frank > > _______________________________________________ > 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".
prev parent reply other threads:[~2025-02-23 13:44 UTC|newest] Thread overview: 4+ messages / expand[flat|nested] mbox.gz Atom feed top 2025-02-22 7:51 [FFmpeg-devel] [PATCH v3 1/2] lavc/vvc: Fix slice map construction for small subpics Nuo Mi 2025-02-22 7:51 ` [FFmpeg-devel] [PATCH v3 2/2] lavc/vvc: Ensure subpictures don't overlap Nuo Mi 2025-02-22 17:36 ` Frank Plowman 2025-02-23 13:44 ` Nuo Mi [this message]
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='CAFXK13ct0EJXuVmqk2xji9EGhhDm=S7ybWGHMdewf3W8S3AAyw@mail.gmail.com' \ --to=nuomi2021@gmail.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