| Message ID | 928dcb75-6411-4e33-caa8-5cd247f8a185@ektacom.com |
|---|---|
| State | New |
| Headers | show |
| Series | [FFmpeg-devel] Fix issue when invalid nb_index_entries value is read - SPONSORED BY INA (Institut National de l'Audiovisuel) | expand |
| Context | Check | Description |
|---|---|---|
| yinshiyou/configure_loongarch64 | warning | Failed to apply patch |
| andriy/configure_x86 | warning | Failed to apply patch |
On Wed, 8 Mar 2023, Cédric Le Barz wrote: > > Fix issue when invalid nb_index_entries value is read : in case of false > nb_index_entries value, ffmpeg exit. This patch fix this problem. What do you mean invalid? Is the value wrong, or is the file truncated in the middle of the index segment? From the patch it looks like the latter. Is there a single unluckily truncated file that you want to fix with this patch, or multiple files can be affected by this fix because e.g. they are written by the same broken muxer? Overall I am not sure if we want this applied, the user can also opt to avoid parsing all partitions with -seekable 0 and read / remux the file sequentially ignoring any index, so if somebody wants to rescue the data, it can be done as is. Regards, Marton > > > Signed-off-by: Cedric Le Barz <clebarz@ektacom.com> > --- > ffmpeg/libavformat/mxfdec.c | 14 ++++++++++++-- > 1 file changed, 12 insertions(+), 2 deletions(-) > > diff --git a/ffmpeg/libavformat/mxfdec.c b/ffmpeg/libavformat/mxfdec.c > index 4530617..ffc8987 100644 > --- a/ffmpeg/libavformat/mxfdec.c > +++ b/ffmpeg/libavformat/mxfdec.c > @@ -1221,8 +1221,18 @@ static int mxf_read_index_entry_array(AVIOContext *pb, > MXFIndexTableSegment *seg > } > for (i = 0; i < segment->nb_index_entries; i++) { > - if(avio_feof(pb)) > - return AVERROR_INVALIDDATA; > + > + if(avio_feof(pb)) { > + if (i == 0) { > + return AVERROR_INVALIDDATA; > + } else { > + /* To be more robust to invalid nb_index_entries value, > + fix the index entry number according to read entries */ > + segment->nb_index_entries = i; > + return 0; > + } > + } > + > segment->temporal_offset_entries[i] = avio_r8(pb); > avio_r8(pb); /* > KeyFrameOffset */ > segment->flag_entries[i] = avio_r8(pb); > -- > 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". > >
Hi, By "invalid", I mean that the written nb_index_entries value does not match with the real number of entries in the table. The number of entries in the table is lower than the specified one. This is a bug in commercial MXFTk software (at least version V 2.8.0.0.1). It exists a lot a MXF files with this problem. Best regards, Cédric -----Message d'origine----- De : ffmpeg-devel <ffmpeg-devel-bounces@ffmpeg.org> De la part de Marton Balint Envoyé : mercredi 8 mars 2023 23:06 À : FFmpeg development discussions and patches <ffmpeg-devel@ffmpeg.org> Objet : Re: [FFmpeg-devel] [PATCH] Fix issue when invalid nb_index_entries value is read - SPONSORED BY INA (Institut National de l'Audiovisuel) On Wed, 8 Mar 2023, Cédric Le Barz wrote: > > Fix issue when invalid nb_index_entries value is read : in case of > false nb_index_entries value, ffmpeg exit. This patch fix this problem. What do you mean invalid? Is the value wrong, or is the file truncated in the middle of the index segment? From the patch it looks like the latter. Is there a single unluckily truncated file that you want to fix with this patch, or multiple files can be affected by this fix because e.g. they are written by the same broken muxer? Overall I am not sure if we want this applied, the user can also opt to avoid parsing all partitions with -seekable 0 and read / remux the file sequentially ignoring any index, so if somebody wants to rescue the data, it can be done as is. Regards, Marton > > > Signed-off-by: Cedric Le Barz <clebarz@ektacom.com> > --- > ffmpeg/libavformat/mxfdec.c | 14 ++++++++++++-- > 1 file changed, 12 insertions(+), 2 deletions(-) > > diff --git a/ffmpeg/libavformat/mxfdec.c b/ffmpeg/libavformat/mxfdec.c > index 4530617..ffc8987 100644 > --- a/ffmpeg/libavformat/mxfdec.c > +++ b/ffmpeg/libavformat/mxfdec.c > @@ -1221,8 +1221,18 @@ static int > mxf_read_index_entry_array(AVIOContext *pb, MXFIndexTableSegment *seg > } > for (i = 0; i < segment->nb_index_entries; i++) { > - if(avio_feof(pb)) > - return AVERROR_INVALIDDATA; > + > + if(avio_feof(pb)) { > + if (i == 0) { > + return AVERROR_INVALIDDATA; > + } else { > + /* To be more robust to invalid nb_index_entries value, > + fix the index entry number according to read entries */ > + segment->nb_index_entries = i; > + return 0; > + } > + } > + > segment->temporal_offset_entries[i] = avio_r8(pb); > avio_r8(pb); /* > KeyFrameOffset */ > segment->flag_entries[i] = avio_r8(pb); > -- > 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". > >
Hi, By "invalid", I mean that the written nb_index_entries value does not match with the real number of entries in the table. The number of entries in the table is lower than the specified one. This is a bug in commercial MXFTk software (at least version V 2.8.0.0.1) : it exists a lot a MXF files with this problem. The proposed patch just robustify index table parsing. Best regards, Cédric -----Message d'origine----- De : ffmpeg-devel <ffmpeg-devel-bounces@ffmpeg.org> De la part de Marton Balint Envoyé : mercredi 8 mars 2023 23:06 À : FFmpeg development discussions and patches <ffmpeg-devel@ffmpeg.org> Objet : Re: [FFmpeg-devel] [PATCH] Fix issue when invalid nb_index_entries value is read - SPONSORED BY INA (Institut National de l'Audiovisuel) On Wed, 8 Mar 2023, Cédric Le Barz wrote: > > Fix issue when invalid nb_index_entries value is read : in case of > false nb_index_entries value, ffmpeg exit. This patch fix this problem. What do you mean invalid? Is the value wrong, or is the file truncated in the middle of the index segment? From the patch it looks like the latter. Is there a single unluckily truncated file that you want to fix with this patch, or multiple files can be affected by this fix because e.g. they are written by the same broken muxer? Overall I am not sure if we want this applied, the user can also opt to avoid parsing all partitions with -seekable 0 and read / remux the file sequentially ignoring any index, so if somebody wants to rescue the data, it can be done as is. Regards, Marton > > > Signed-off-by: Cedric Le Barz <clebarz@ektacom.com> > --- > ffmpeg/libavformat/mxfdec.c | 14 ++++++++++++-- > 1 file changed, 12 insertions(+), 2 deletions(-) > > diff --git a/ffmpeg/libavformat/mxfdec.c b/ffmpeg/libavformat/mxfdec.c > index 4530617..ffc8987 100644 > --- a/ffmpeg/libavformat/mxfdec.c > +++ b/ffmpeg/libavformat/mxfdec.c > @@ -1221,8 +1221,18 @@ static int > mxf_read_index_entry_array(AVIOContext *pb, MXFIndexTableSegment *seg > } > for (i = 0; i < segment->nb_index_entries; i++) { > - if(avio_feof(pb)) > - return AVERROR_INVALIDDATA; > + > + if(avio_feof(pb)) { > + if (i == 0) { > + return AVERROR_INVALIDDATA; > + } else { > + /* To be more robust to invalid nb_index_entries value, > + fix the index entry number according to read entries */ > + segment->nb_index_entries = i; > + return 0; > + } > + } > + > segment->temporal_offset_entries[i] = avio_r8(pb); > avio_r8(pb); /* > KeyFrameOffset */ > segment->flag_entries[i] = avio_r8(pb); > -- > 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". > >
On Thu, 16 Mar 2023, Cédric Le Barz wrote: > Hi, > > By "invalid", I mean that the written nb_index_entries value does not > match with the real number of entries in the table. The number of > entries in the table is lower than the specified one. This is a bug in > commercial MXFTk software (at least version V 2.8.0.0.1). It exists a > lot a MXF files with this problem. Please share a sample. Or better yet, open a ticket for this issue and attach a sample there for future reference. Thanks, Marton > > Best regards, > > Cédric > > -----Message d'origine----- > De : ffmpeg-devel <ffmpeg-devel-bounces@ffmpeg.org> De la part de Marton Balint > Envoyé : mercredi 8 mars 2023 23:06 > À : FFmpeg development discussions and patches <ffmpeg-devel@ffmpeg.org> > Objet : Re: [FFmpeg-devel] [PATCH] Fix issue when invalid nb_index_entries value is read - SPONSORED BY INA (Institut National de l'Audiovisuel) > > > > On Wed, 8 Mar 2023, Cédric Le Barz wrote: > >> >> Fix issue when invalid nb_index_entries value is read : in case of >> false nb_index_entries value, ffmpeg exit. This patch fix this problem. > > What do you mean invalid? Is the value wrong, or is the file truncated in the middle of the index segment? From the patch it looks like the latter. > > Is there a single unluckily truncated file that you want to fix with this patch, or multiple files can be affected by this fix because e.g. they are written by the same broken muxer? > > Overall I am not sure if we want this applied, the user can also opt to avoid parsing all partitions with -seekable 0 and read / remux the file sequentially ignoring any index, so if somebody wants to rescue the data, it can be done as is. > > Regards, > Marton > >> >> >> Signed-off-by: Cedric Le Barz <clebarz@ektacom.com> >> --- >> ffmpeg/libavformat/mxfdec.c | 14 ++++++++++++-- >> 1 file changed, 12 insertions(+), 2 deletions(-) >> >> diff --git a/ffmpeg/libavformat/mxfdec.c b/ffmpeg/libavformat/mxfdec.c >> index 4530617..ffc8987 100644 >> --- a/ffmpeg/libavformat/mxfdec.c >> +++ b/ffmpeg/libavformat/mxfdec.c >> @@ -1221,8 +1221,18 @@ static int >> mxf_read_index_entry_array(AVIOContext *pb, MXFIndexTableSegment *seg >> } >> for (i = 0; i < segment->nb_index_entries; i++) { >> - if(avio_feof(pb)) >> - return AVERROR_INVALIDDATA; >> + >> + if(avio_feof(pb)) { >> + if (i == 0) { >> + return AVERROR_INVALIDDATA; >> + } else { >> + /* To be more robust to invalid nb_index_entries value, >> + fix the index entry number according to read entries */ >> + segment->nb_index_entries = i; >> + return 0; >> + } >> + } >> + >> segment->temporal_offset_entries[i] = avio_r8(pb); >> avio_r8(pb); /* >> KeyFrameOffset */ >> segment->flag_entries[i] = avio_r8(pb); >> -- >> 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". >> >> > _______________________________________________ > 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".
We have several samples but these are big files: about 7 GB. Is it OK to share such a file size ? Thanks for you answer, Regards, Cédric Le 16/03/2023 à 21:28, Marton Balint a écrit : > > > On Thu, 16 Mar 2023, Cédric Le Barz wrote: > >> Hi, >> >> By "invalid", I mean that the written nb_index_entries value does not >> match with the real number of entries in the table. The number of >> entries in the table is lower than the specified one. This is a bug in >> commercial MXFTk software (at least version V 2.8.0.0.1). It exists a >> lot a MXF files with this problem. > > Please share a sample. Or better yet, open a ticket for this issue and > attach a sample there for future reference. > > Thanks, > Marton > >> >> Best regards, >> >> Cédric >> >> -----Message d'origine----- >> De : ffmpeg-devel <ffmpeg-devel-bounces@ffmpeg.org> De la part de >> Marton Balint >> Envoyé : mercredi 8 mars 2023 23:06 >> À : FFmpeg development discussions and patches <ffmpeg-devel@ffmpeg.org> >> Objet : Re: [FFmpeg-devel] [PATCH] Fix issue when invalid >> nb_index_entries value is read - SPONSORED BY INA (Institut National >> de l'Audiovisuel) >> >> >> >> On Wed, 8 Mar 2023, Cédric Le Barz wrote: >> >>> >>> Fix issue when invalid nb_index_entries value is read : in case of >>> false nb_index_entries value, ffmpeg exit. This patch fix this problem. >> >> What do you mean invalid? Is the value wrong, or is the file >> truncated in the middle of the index segment? From the patch it looks >> like the latter. >> >> Is there a single unluckily truncated file that you want to fix with >> this patch, or multiple files can be affected by this fix because >> e.g. they are written by the same broken muxer? >> >> Overall I am not sure if we want this applied, the user can also opt >> to avoid parsing all partitions with -seekable 0 and read / remux the >> file sequentially ignoring any index, so if somebody wants to rescue >> the data, it can be done as is. >> >> Regards, >> Marton >> >>> >>> >>> Signed-off-by: Cedric Le Barz <clebarz@ektacom.com> >>> --- >>> ffmpeg/libavformat/mxfdec.c | 14 ++++++++++++-- >>> 1 file changed, 12 insertions(+), 2 deletions(-) >>> >>> diff --git a/ffmpeg/libavformat/mxfdec.c b/ffmpeg/libavformat/mxfdec.c >>> index 4530617..ffc8987 100644 >>> --- a/ffmpeg/libavformat/mxfdec.c >>> +++ b/ffmpeg/libavformat/mxfdec.c >>> @@ -1221,8 +1221,18 @@ static int >>> mxf_read_index_entry_array(AVIOContext *pb, MXFIndexTableSegment *seg >>> } >>> for (i = 0; i < segment->nb_index_entries; i++) { >>> - if(avio_feof(pb)) >>> - return AVERROR_INVALIDDATA; >>> + >>> + if(avio_feof(pb)) { >>> + if (i == 0) { >>> + return AVERROR_INVALIDDATA; >>> + } else { >>> + /* To be more robust to invalid nb_index_entries >>> value, >>> + fix the index entry number according to read >>> entries */ >>> + segment->nb_index_entries = i; >>> + return 0; >>> + } >>> + } >>> + >>> segment->temporal_offset_entries[i] = avio_r8(pb); >>> avio_r8(pb); /* >>> KeyFrameOffset */ >>> segment->flag_entries[i] = avio_r8(pb); >>> -- >>> 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". >>> >>> >> _______________________________________________ >> 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". > _______________________________________________ > 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".
On Mon, 20 Mar 2023, Cédric Le Barz wrote: > We have several samples but these are big files: about 7 GB. Is it OK to > share such a file size ? I kind of hoped you have access to the broken muxer and can remux a small file which can be made public. If not, then simply upload it somewhere so I can take a look. If the sample can't be made public, just share it with me privately. Thanks, Marton > > Thanks for you answer, > > Regards, > > Cédric > > Le 16/03/2023 à 21:28, Marton Balint a écrit : >> >> >> On Thu, 16 Mar 2023, Cédric Le Barz wrote: >> >>> Hi, >>> >>> By "invalid", I mean that the written nb_index_entries value does not >>> match with the real number of entries in the table. The number of >>> entries in the table is lower than the specified one. This is a bug in >>> commercial MXFTk software (at least version V 2.8.0.0.1). It exists a >>> lot a MXF files with this problem. >> >> Please share a sample. Or better yet, open a ticket for this issue and >> attach a sample there for future reference. >> >> Thanks, >> Marton >> >>> >>> Best regards, >>> >>> Cédric >>> >>> -----Message d'origine----- >>> De : ffmpeg-devel <ffmpeg-devel-bounces@ffmpeg.org> De la part de Marton >>> Balint >>> Envoyé : mercredi 8 mars 2023 23:06 >>> À : FFmpeg development discussions and patches <ffmpeg-devel@ffmpeg.org> >>> Objet : Re: [FFmpeg-devel] [PATCH] Fix issue when invalid >>> nb_index_entries value is read - SPONSORED BY INA (Institut National de >>> l'Audiovisuel) >>> >>> >>> >>> On Wed, 8 Mar 2023, Cédric Le Barz wrote: >>> >>>> >>>> Fix issue when invalid nb_index_entries value is read : in case of >>>> false nb_index_entries value, ffmpeg exit. This patch fix this problem. >>> >>> What do you mean invalid? Is the value wrong, or is the file truncated in >>> the middle of the index segment? From the patch it looks like the latter. >>> >>> Is there a single unluckily truncated file that you want to fix with this >>> patch, or multiple files can be affected by this fix because e.g. they >>> are written by the same broken muxer? >>> >>> Overall I am not sure if we want this applied, the user can also opt to >>> avoid parsing all partitions with -seekable 0 and read / remux the file >>> sequentially ignoring any index, so if somebody wants to rescue the data, >>> it can be done as is. >>> >>> Regards, >>> Marton >>> >>>> >>>> >>>> Signed-off-by: Cedric Le Barz <clebarz@ektacom.com> >>>> --- >>>> ffmpeg/libavformat/mxfdec.c | 14 ++++++++++++-- >>>> 1 file changed, 12 insertions(+), 2 deletions(-) >>>> >>>> diff --git a/ffmpeg/libavformat/mxfdec.c b/ffmpeg/libavformat/mxfdec.c >>>> index 4530617..ffc8987 100644 >>>> --- a/ffmpeg/libavformat/mxfdec.c >>>> +++ b/ffmpeg/libavformat/mxfdec.c >>>> @@ -1221,8 +1221,18 @@ static int >>>> mxf_read_index_entry_array(AVIOContext *pb, MXFIndexTableSegment *seg >>>> } >>>> for (i = 0; i < segment->nb_index_entries; i++) { >>>> - if(avio_feof(pb)) >>>> - return AVERROR_INVALIDDATA; >>>> + >>>> + if(avio_feof(pb)) { >>>> + if (i == 0) { >>>> + return AVERROR_INVALIDDATA; >>>> + } else { >>>> + /* To be more robust to invalid nb_index_entries value, >>>> + fix the index entry number according to read entries >>>> */ >>>> + segment->nb_index_entries = i; >>>> + return 0; >>>> + } >>>> + } >>>> + >>>> segment->temporal_offset_entries[i] = avio_r8(pb); >>>> avio_r8(pb); /* >>>> KeyFrameOffset */ >>>> segment->flag_entries[i] = avio_r8(pb); >>>> -- >>>> 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". >>>> >>>> >>> _______________________________________________ >>> 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". >> _______________________________________________ >> 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". > >
diff --git a/ffmpeg/libavformat/mxfdec.c b/ffmpeg/libavformat/mxfdec.c index 4530617..ffc8987 100644 --- a/ffmpeg/libavformat/mxfdec.c +++ b/ffmpeg/libavformat/mxfdec.c @@ -1221,8 +1221,18 @@ static int mxf_read_index_entry_array(AVIOContext *pb, MXFIndexTableSegment *seg } for (i = 0; i < segment->nb_index_entries; i++) { - if(avio_feof(pb)) - return AVERROR_INVALIDDATA; + + if(avio_feof(pb)) { + if (i == 0) { + return AVERROR_INVALIDDATA; + } else { + /* To be more robust to invalid nb_index_entries value, + fix the index entry number according to read entries */ + segment->nb_index_entries = i; + return 0; + } + } + segment->temporal_offset_entries[i] = avio_r8(pb); avio_r8(pb); /* KeyFrameOffset */
Fix issue when invalid nb_index_entries value is read : in case of false nb_index_entries value, ffmpeg exit. This patch fix this problem. Signed-off-by: Cedric Le Barz <clebarz@ektacom.com> --- ffmpeg/libavformat/mxfdec.c | 14 ++++++++++++-- 1 file changed, 12 insertions(+), 2 deletions(-) segment->flag_entries[i] = avio_r8(pb);