August 2012 - DVDnav-discuss - lists.mplayerhq.hu

[PATCH] Fix out of array pointer access
by Bryce Harrington 10 Dec '12

10 Dec '12

This patch fixes a segmentation fault hit when reading the DVD 'The Express'. It prevents a read/write beyond end of an array due to using a length value taken from the DVD, which can exceed the allocated size. https://bugs.launchpad.net/ubuntu/+source/libdvdread/+bug/894170 The patch was originally written by rickyrockrat (sorry, I don't have his email address) for 4.1.3. I got the DVD and reproduced the segfault using 4.2.0 and verified the patch stops the segfault from happening. We're not confident this is the best fix though, so are posting it here for review. Signed-off-by: Bryce Harrington <bryce(a)canonical.com> diff -Nrup libdvdread-4.2.0//src/ifo_read.c libdvdread-4.2.0-working//src/ifo_read.c --- libdvdread-4.2.0//src/ifo_read.c 2011-12-13 20:24:16.000000000 -0800 +++ libdvdread-4.2.0-working//src/ifo_read.c 2011-12-13 21:21:53.915978130 -0800 @@ -1082,6 +1082,12 @@ int ifoRead_TT_SRPT(ifo_handle_t *ifofil return 0; } + if(tt_srpt->nr_of_srpts>info_length/sizeof(title_info_t)){ + fprintf(stderr,"libdvdread: data mismatch: info_length (%ld)!= nr_of_srpts (%d). Truncating.\n", + info_length/sizeof(title_info_t),tt_srpt->nr_of_srpts); + tt_srpt->nr_of_srpts=info_length/sizeof(title_info_t); + } + for(i = 0; i < tt_srpt->nr_of_srpts; i++) { B2N_16(tt_srpt->title[i].nr_of_ptts); B2N_16(tt_srpt->title[i].parental_id);

3 3

[PATCH] fix C++ compatibility broken in r1208
by Anssi Hannula 09 Dec '12

09 Dec '12

Revision 1208 ("Add dvdnav_program_play & dvdnav_current_title_program") added dvdnav_program_play function to dvdnav.h. However, unlike the other API functions, its first parameter is named 'this' instead of 'self'. 'this' is a C++ keyword and prevents compilation of C++ projects using libdvdnav. Restore C++ compatibility by renaming the parameter to 'self' like in the other functions. -- Anssi Hannula

3 2

[PATCH] Make sure we get nav packets for all cells in multi angle
by Joakim Plate 09 Dec '12

09 Dec '12

Hi, This is a resend of an old patch that we still keep in xbmc's version of libdvdnav. The problem is that libdvdnav jumps over NAV packets when playing multi angle features. I don't remember all the exact details on this patch anymore since it was many years since i tracked it down. But it is very obvious in xbmc since we use nav packet end/start times to correct timestamps. Without the patch we git gaps in nav packets that doesn's exist in the streams. I think i still have a dvd sample that shows the issue. /Joakim

3 5

libdvdnav 4.1.4
by Jean-Jacques Sarton 09 Dec '12

09 Dec '12

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Hello, On Fedora 14 totem and vlc end both with folowing message: libdvdread: Found 1 VTS's libdvdread: Elapsed time 0 libdvdnav: Language 'en' not found, using '' instead libdvdnav: Menu Languages available: libdvdnav: Language 'en' not found, using '' instead libdvdnav: Menu Languages available: libdvdnav: Language 'en' not found, using '' instead libdvdnav: Menu Languages available: libdvdnav: *** pgci_ut handle is NULL *** totem: /builddir/build/BUILD/libdvdnav-4.1.4/src/vm/vm.c:1772: get_ID: Assertion `pgcit != ((void *)0)' failed. Of course tje line totem:... is to be replaced withz vlc:... according to the used player The dvd contain no menu and contain a privat video. If I rename libdvdnav totem/vlc works. Regards, Jean-Jacques Sarton -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.11 (GNU/Linux) iEYEARECAAYFAk0h3bAACgkQM9JbiR3CwQsEEgCggN7qJrKQXspNYBJbCGRh237R +joAn1pQaRWHFN+nASDqRbL6Jgj3CM5S =iNgb -----END PGP SIGNATURE-----

4 6

[PATCH] consuming all memory when scanning dvd "Up"
by John Stebbins 09 Dec '12

09 Dec '12

First a disclaimer. This is a big invasive patch. If you don't like it, my feelings won't be hurt. I will be using it in HandBrake. So if you see problems with it, please let me know. Many dvd's are showing up that have titles that have numerous repeated language unit tables and pgc's. The new release "Up" is an example. I believe "Dark Knight" also exhibits this. When scanning such discs with lsdvd (and other programs that scan all titles), libdvdread will consume many GB of data for these repeated elements and gets very slow (e.g hours to read all titles) on OS X due to reading the duplicate data from uncached raw devices. This patch detects duplicates and reference counts pgc's and pgcit's. When a duplicate is detected, a reference count is incremented instead of allocating new memory and re-reading the data. I opted for reference counting instead of simply returning a failure on duplicate detection because it is perfectly valid to have a title with such duplicates. I could easily imagine titles adding duplicates, then navigating around them.

3 5

cross/mingw compilation (was assertion failure with certain DVD's)
by Roger Pack 18 Oct '12

18 Oct '12

>> $./configure2 >> ./configure2: line 95: cd: /usr/local/: No such file or directory # >> needs to be run in bash I believe > > Nope, you need to set the install prefix with --prefix=blabla Sweet that fixed it, as long as the prefix exists, otherwise same error message different path :) >> Checking if we're big-endian... ./configure2: line 113: mktemp: command not found This meant I didn't have msys installed, only mingw (my bad). >> Done, type 'make install' to build and install >> (appears to set HAVE_DLFCN_H in error) > > Without using autotools, dlfcn.h presence is not detected Is there a way to use autotools with libdvdread? I'm a bit lost, as is evident. though I can work around it for now. This actually makes me wonder how anybody is compiling it for their windows distros today. I know mplayer just uses the src directory and its own configs (which work), dunno about VLC/others. It seems like both msys and cross compiling don't work, or can somebody give me an example of how they do work, somehow? > ./msvc/config.h:#define HAVE_DLFCN_H 1 > > And it is assumed to exist (is this file made for Visual Studio?) I guess so, since it's under the msvc dir. >> Also as a note when it runs strip and ranlib, if it's being cross >> compiled, it would be helpful to call their equivalent: >> $ ./configure2 --cc=i586-mingw32msvc-gcc >> should eventually call i586-mingw32msvc-ranlib >> >> maybe it could use a new configure option for a prefix or what not. > > ./configure --host=i586-mingw32-msvc should work Yes I think it needs an option like that added (configure2 doesn't appear to have one as of yet). -roger-

2 11

Problem with DVD
by Cédric Macquat 08 Oct '12

08 Oct '12

Hello, I have a crash with vlc when I try to read the dvd "Transformers 3 (Dark of the moon)". I've purchased this dvd in Europe (France). I would like to know if the problem is known or not. Somebody told me the problem is in libdvdnav so that's the reason why I post here. When I start the dvd I become the following in vlc: ----------------------------- [0x20cc108] main libvlc: Running vlc with the default interface. Use 'cvlc' to use vlc without interface. libdvdnav: Using dvdnav version 4.2.0 libdvdread: Using libdvdcss version 1.2.12 for DVD access libdvdnav: DVD Title: TRANSFORMERS_3 libdvdnav: DVD Serial Number: 3F336B0B libdvdnav: DVD Title (Alternative): TRANSFORMERS_3 libdvdnav: Unable to find map file '/home/user/.dvdnav/TRANSFORMERS_3.map' libdvdnav: DVD disk reports itself with Region mask 0x00fd0000. Regions: 2 libdvdread: Attempting to retrieve all CSS keys libdvdread: This can take a _long_ time, please be patient libdvdread: Get key for /VIDEO_TS/VIDEO_TS.VOB at 0x0000020a libdvdread: Elapsed time 0 libdvdread: Get key for /VIDEO_TS/VTS_01_1.VOB at 0x00000610 libdvdread: Elapsed time 0 libdvdread: Get key for /VIDEO_TS/VTS_02_1.VOB at 0x0000062e libdvdread: Elapsed time 0 libdvdread: Get key for /VIDEO_TS/VTS_03_1.VOB at 0x0000067d libdvdread: Elapsed time 0 libdvdread: Get key for /VIDEO_TS/VTS_04_1.VOB at 0x0000069b libdvdread: Elapsed time 0 libdvdread: Get key for /VIDEO_TS/VTS_05_0.VOB at 0x00000a08 libdvdread: Elapsed time 0 . . . . libdvdread: Get key for /VIDEO_TS/VTS_16_0.VOB at 0x0034fd84 libdvdread: Elapsed time 0 libdvdread: Get key for /VIDEO_TS/VTS_16_1.VOB at 0x003524de libdvdread: Elapsed time 0 libdvdread: Get key for /VIDEO_TS/VTS_17_0.VOB at 0x003524fc libdvdread: Elapsed time 0 libdvdread: Get key for /VIDEO_TS/VTS_17_1.VOB at 0x0035256c libdvdread: Elapsed time 0 libdvdread: Get key for /VIDEO_TS/VTS_18_1.VOB at 0x00353d20 libdvdread: Elapsed time 0 libdvdread: Get key for /VIDEO_TS/VTS_19_1.VOB at 0x00353d3c libdvdread: Elapsed time 0 libdvdread: Found 19 VTS's libdvdread: Elapsed time 0 *** libdvdread: CHECK_VALUE failed in ifo_read.c:1198 *** *** for data[i] + sizeof(ptt_info_t) <= vts_ptt_srpt->last_byte + 1 + 4 *** *** libdvdread: CHECK_VALUE failed in ifo_read.c:1198 *** *** for data[i] + sizeof(ptt_info_t) <= vts_ptt_srpt->last_byte + 1 + 4 *** *** libdvdread: CHECK_VALUE failed in ifo_read.c:1198 *** *** for data[i] + sizeof(ptt_info_t) <= vts_ptt_srpt->last_byte + 1 + 4 *** *** libdvdread: CHECK_VALUE failed in ifo_read.c:1198 *** *** for data[i] + sizeof(ptt_info_t) <= vts_ptt_srpt->last_byte + 1 + 4 *** *** libdvdread: CHECK_VALUE failed in ifo_read.c:1198 *** *** for data[i] + sizeof(ptt_info_t) <= vts_ptt_srpt->last_byte + 1 + 4 *** . . . . segmentation fault ----------------------- After searching a little bit I've found, that I can read the movie with vlc if I check the box "No disc menu" and I choose a title above 15. But the chapters are in the wrong order. If I try to read with mplayer it works if I start mplayer with: $mplayer -dvd-device /dev/sr0 dvdnav:// I think only "dvdnav://" is important here. I've also found the problem seems to be related to the Arccos protection scheme. The DVD has 99 title and only one is the movie with the chapters in the right order. So can you confirm it's a problem in libdvdnav ? Or somewhere else ? If it's in libdvdnav, do you need some more inputs for this problem ? Thanks!

6 18

[PATCH] Filenames in Unicode decoding issue
by Honza Horak 17 Sep '12

17 Sep '12

Hi all, there's an issue in libdvdread, triggered by DVDs with Unicode filenames (example of such DVD is a movie Thor). Original bug report is available here: https://bugzilla.redhat.com/show_bug.cgi?id=813977 There is also an enhanced Unicodedecode function on Ubuntu forum, which I consider better, than the current implementation is (see the patch attached): http://ubuntuforums.org/showthread.php?p=11254706 Unfortunately, I'm unable to reproduce this issue, since I don't have such DVD, so I cannot verify if the patch fixes the issue. Cheers, Honza

2 3

Mixing of enums in dvdnav_get_next_cache_block()
by Erik Hovland 10 Sep '12

10 Sep '12

The struct dvdnav_vts_change_event_t has the members old_domain and new_domain. Both are of the enum type dvd_read_domain_t. The struct vm_position_t has a member domain of type domain_t. dvd_read_domain_t has values of 0-3 for the enums while domain_t has bit field values. The later is b/c those domains mean something in the actual DVD structure. dvdnav_get_next_cache_block does this: vts_event->old_domain = position_current.domain vts_event->new_domain = position_next.domain Which means that it is assigning values to one enum variable that have no meaning. I am starting to come around to the idea that the type of old_domain and new_domain should just be changed. Does anyone have any opinion? Thanks E -- Erik Hovland erik(a)hovland.org http://hovland.org/

1 1

dvdnav_sector_search()
by Marcel J.E. Mol 08 Sep '12

08 Sep '12

Hi List, I'm exploring the dvdread/dvdnav libraries to optimise dvd playback on the OpenPli stb receivers. So first of all maybe openpli could be added to the list of 'related projects'... OpenPli uses the dvdnac_sector_search() function to forward/backward skip. It firsts uses dvdnav_get_position(), does some calcs and calls dvdnav_sector_search with SEEK_SET. But dvdnav_sector_search() already does a call to dvdnav_get_posisition(). so I like to use SEEK_CUR instead of SEEK_SET. However the offset parameter to dvdnav_sector_search is an uint64_t. This is no problem for forward skips, but backword skip won't work... Is there a reason for the uint64_t? Otherwise I would suggest changing it to int64_t so we can pass negatve values for offset. thanks -Marcel -- ======-------- Marcel J.E. Mol MESA Consulting B.V. =======--------- ph. +31-(0)6-54724868 P.O. Box 112 =======--------- marcel(a)mesa.nl 2630 AC Nootdorp __==== www.mesa.nl ---____U_n_i_x______I_n_t_e_r_n_e_t____ The Netherlands ____ They couldn't think of a number, Linux user 1148 -- counter.li.org so they gave me a name! -- Rupert Hine -- www.ruperthine.com

3 4