strange subheaders with Plextor

[themabus]

there were some PSX games for Japan region that would have subheader field modified.
i can't remember all now, but i would mark them having errors in db.
Space Griffon VF-9 is one of those.
it's interesting that Plextor and Lite-On results from IsoBuster (Alcohol, ImgBurn, CloneCD, etc) would not match.
difference is in sectors that have Sync + Header + modified sub-header + everything else set to 0x00
Plextor would not return this modified subheader in such case, but replace it with zeros instead,
so those sectors would look like Mode0.
when trap disc or d8 is used - results would match.
i've made a short sequence of modified sectors to test this further http://www.mediafire.com/?trtzmmmzimn
and results are interesting imho:
ther's only 6 sectors (this violates minimum track length, but i didn't have problems with it)
with 3 different subheader sequences:
2 with 01 02 03 04 - 01 02 03 04
2 with 01 00 02 00 - 01 02 00 00
2 with 01 00 e4 00 - 01 e4 00 00 <| seq. from Space Griffon
recorded it with Alcohol on Lite-On
on Lite-On IsoBuster would in Sector Viewer replace 1st and 2nd sequence with zeros but display 3rd correctly.
on Plextor it would replace 1st and show correctly 2nd and 3rd
(it's strange, because it does not show correctly similar sector from Space Griffon).
but when backup is made from Plextor all 3 sequences are replaced with 0x00 (d8 work correct, however).
also ImgBurn and CDRWin would not display any subheaders in sector viewer on Plextor (so just like backup).

so would such CD (eg. Space Griffon VF-9) be checked multiple times on Plextor (with everything but PerfectRip/CD Tool) -
it would end up in db different from original.
or at least on my PC. maybe it's just my Plextor. it's Premium model.
also - maybe ther's subheader sequences that would return correctly only on d8 or with trap disc.

so would be great if somebody could verify this.

edit:
actually it seems that sector can have data and it will be set to 0x00 as well.
so conditions are: no EDC/ECC, messed up sub-header.

edit:
it's more complex, i've forgot there was modified sub-headers in Imadoki no Vampire: Bloody Bride
there only 4th byte changes, always the same way:
01 08 64 e0 - 01 08 64 f0
and Plextor reads it fine
so last byte doesn't seem to matter

[themabus]

i've checked how those programs read sectors and they basically do the same:

IsoBuseter 2.5

Code: Select all

Sector View ([ ] RAW)
 READ (10)         28h
Sector View ([x] RAW)
 READ CD           BEh F8h (Requested sector)
Extract RAW Data
 READ CD           BEh F8h (0..26) <| 64K buffer - 27 sectors at a time
 READ CD           BEh F8h (27..53)
 ...
 READ CD           BEh F8h (Last-26..Last)

CDRWIN 4.0G

Code: Select all

Sector Viewer
 TEST UNIT READY   00h
 READ CD           BEh F8h (Requested sector)
Extract Disc/Tracks/Sectors to Image File ([x] Select Tracks)
 TEST UNIT READY   00h
 START STOP UNIT   1Bh
 TEST UNIT READY   00h
 REZERO UNIT       01h
 TEST UNIT READY   00h
 START STOP UNIT   1Bh 'Analyzing Disc Layout... Please Wait...'
 TEST UNIT READY   00h
 READ TOC/PMA/ATIP 43h
 READ CD           BEh F8h (Sector 0)
 READ TOC/PMA/ATIP 43h
 READ CD           BEh F8h (Last-75)
 READ CD           BEh F8h (Last-75) <| last second of track is checked, one sector at a time. 
 ...                                    even tho it's single-track CD.
 READ CD           BEh F8h (Last)
 REZERO UNIT       01h
 SET CD SPEED      BBh
 MODE SELECT (10)  55h
 READ CD           BEh F8h (0..2) 'Copy Progress'
 READ CD           BEh F8h (3..5) <| 8K buffer - 3 sectors at a time
 ...
 READ CD           BEh F8h (Last-2..Last)
 MODE SELECT (10)  55h
 SET CD SPEED      BBh

ImgBurn

Code: Select all

Sector Viewer ([ ] RAW)
 READ (10)         28h
Sector Viewer ([x] RAW)
 READ CD           BEh F8h (Requested sector)
Read
 TEST UNIT READY   00h
 GET CONFIGURATION 46h
 READ CAPACITY     25h
 GET CONFIGURATION 46h
 READ TOC/PMA/ATIP 43h
 READ TOC/PMA/ATIP 43h
 READ TRACK INFORM.52h
 READ CD           BEh F8h (Sector 0)
 READ CAPACITY     25h
 READ CD           BEh F8h (0..26) <| 64K buffer - 27 sectors at a time
 READ CD           BEh F8h (27..53)
 ...
 READ CD           BEh F8h (Last-26..Last)
 READ CD           BEh F8h (Sector 16) ?
 ...
 READ CD           BEh F8h (Sector 21) ?
 GET CONFIGURATION 46h
 READ TOC/PMA/ATIP 43h
 PREVENT ALLOW MED.1Eh
 SET CD SPEED      BBh
 SET CD SPEED      BBh
 READ SUBCHANNEL   42h <| sub-channel is read after main channel has been saved already.
 SET CD SPEED      BBh    likely at lower speed
 ...

IsoBuster does less checks on read, but sector view - it's about the same.
in Sector Viewer always one sector at a time is read,
when making backup - multiple, depending on buffer, but it's the same command.
also CDRWIN always communicate through ASPI and sometimes would use
IOCTL_SCSI_PASS_THROUGH instead of IOCTL_SCSI_PASS_THROUGH_DIRECT
rest programs would prefer _DIRECT unless ther's errors

sometimes CDRWIN would actually return subheader in Sector Viewer, like IsoBuster does
it's strange, it seems to happen when drive is spinning up.

i suspect this all has something to do with my IDE controller  - Jmicron JMB36x, rather than Plextor

[themabus]

ther's more to it:
when i ran into this, either Plextor or Lite-On were connected through onboard IDE controller, along with primary HDD.
i've changed IDE drives a few times and it would affects this issue to a certain degree,
so i was pretty certain it's something with controller.
now i have Plextor hooked via USB converter and it still stays.
what's more interesting: READ CD command returns modified sectors but READ BUFFER right afterwards does not.
so in Plextor's buffer data is correct but somewhere on the way out it gets crooked.