Re: Fwd: FW: [Bug 47663] New: DVD burned from iso file has incorrect checksum
Many thanks to point out again the 32k boundary. Is there any reference
to this,
MMC specifications at http://www.t10.org/drafts.htm, see RESERVE TRACK
command descriptions. Actually having re-read it now I noticed that in
DVD-R DAO mode reserved track size is not rounded up to multiple of
32KB... And Pioneer unit indeed recorded track wich is not multiple of
32KB... Well, "recorded" is a bit misleading here, because DVD can be
recorded in 32KB blocks, not a single bit less. So physically data wasa
padded, it's just that the value returned in reply to READ CAPACITY
(mentioned in my first reply), which is recorded in lead-in is not
multiple of 32KB. Well, DAO support was added to growisofs last, and
originally to facilitate usage of minimally blanked DVD-RW media... Once
again, note that the above applies to one single DVD recording procedure
among few and I would still stick to my advice to always verify only as
much as you've recorded, as it's the *only* way to *unconditionally*
obtain *relevant* results.
then I will issue an error report on bugzilla, that the fedora
people, should create their iso images for DVD, that they comply to the
standard.
Keep in mind that sending non-multiple of 32KB does not violate
standard, so that formally you can't blaim them for violating one. What
you can say is that "for naive checksumming verification to work in
broader amount of possible combinations, DVD image should be padded to
32KB," but not more... Also that Redhat embeds checksum in their images
and implement own verification procedure upon installation, so they can
always argue "stick to our verification procedure, that's what we
support." A.
Reply to: