[Libcdio-devel] DVD problems. Was: mmc_get_disc_erasable proposition + questions

[Thomas Schmitt]

Hi,

> > The problem is that you have to pull logics from
> > the OS drivers into the generic part.
> Sure. I'm not completely sure what to do here, although I have inclinations
> in some areas as I mentioned before. So this will probably take time.

All normal media recognition needs a kind of
pre-test for Current Profile.
CD profiles or should then make use of traditional
recognition. Same if profile inquiry fails.

DVD and BD should be evaluated by READ DISC
STRUCTURE and eventually by the existing DVD
examination stuff. One will have to assess
what of this OS specific DVD handling is of
use. Normally our MMC knowledge is more modern
than what we can expect in installed OS kernels.

Give me a stub where to examine "DVD-ROM" and
"BD-ROM". Map all other valid non-CD profiles
to the appropriate CDIO_DISC_MODE_*.
Invent new ones which for BD profiles.
(See doc/glossary for profile list.)


> My understanding is that the TOC does follow the existing  Red, Yellow,
> Green and White Book Standards.

It does.
But these standards do not care for media sizes
above 800 MB. They apply only to CD.

DVD is covered by a series of ECMA standards.
Among them ECMA-279 (DVD-R), ECMA-337 (DVD+RW),
ECMA-349 (DVD+R). But actually the media models
of MMC are much more helpful than the physical
descriptions in the ECMA specs.

MMC made the transition from CD-only by putting
all emphasis on the concepts of Logical Track
and Logical Block. The concept of Session is
still present on DVD-R[W].
There is no Sector or Frames, Seconds, Minutes,
Sub-Channels, Lead-Ins, etc. outside the CD
model. Some still seem to exist inside the DVD
media specifications. But they are hidden to us.

The natural inquiry commands for a modern TOC
are READ DISC INFORMATION and READ TRACK
INFORMATION.
But with CD it is also natural to stay with
READ TOC/PMA/ATIP. Possibly these TOCs provide
more info about non-data tracks. (All DVD and
BD tracks are data tracks.)


> > Vice versa a libdvdio would have a huge
> > intersection of features with libcdio.

> I'm okay with that.

Really ?

I see in libcdio's DVD behavior mainly the need
to correct some bugs.
The cheapest way would be to declare DVD
unsupported.
My proposal is rather to add sparsely what is
needed to make libcdio handle DVD and BD like
data CD-ROM.

Regrettably, neither the old CD specs nor the
Linux CDROM interface are able to express a DVD
TOC. Therefore a fix of DVD TOC would have to
happen outside those specs.
MMC gives a clear opportunity to do so.
Any DVD drive is supposed to comply to MMC-2 or
later.


Have a nice day :)

Thomas