[syslinux] isohybrid and partition type 0x17

[H. Peter Anvin]

There is no partition type associated with iso9660, since iso9660 is normally written to an unpartitioned device.  So the 0x96 thing is most likely some thing you guys simply made up.  (Consider the digits!)

Andres Salomon <dilinger at queued.net> wrote:

>On Mon, 15 Apr 2013 21:25:43 -0700
>"H. Peter Anvin" <hpa at zytor.com> wrote:
>
>> On 04/15/2013 08:31 PM, Andres Salomon wrote:
>> > Hi,
>> > 
>> > OLPC's Openfirmware (OFW) implementation chokes when attempting to
>> > boot an iso created by isohybrid.  The problem is that OFW's
>> > ISO-9660 support expects a partition type of 0x96 (CHRP ISO-9660,
>> > according to http://en.wikipedia.org/wiki/Partition_type).  I see
>> > from
>> >
>http://git.kernel.org/cgit/boot/syslinux/syslinux.git/commit/utils/isohybrid.in?id=aee4e8fa73571eaca80ad43aa007523cde5d4567
>> > that the partition type was originally 0x83 (Linux), and was
>> > changed to 0x17 (hidden IFS) to satisfy windows.
>> > 
>> > I'm wondering what's the correct partition type to actually use.
>> > Does windows balk if 0x96 is used?  Should OFW be using 0x17, or
>> > should isohybrid (and debian-installer) be using 0x96 for a
>> > partition table containing an ISO-9660 filesystem?
>> > 
>> 
>> OK, once again, is this yet another "OFW" which really means "OLPC
>> OFW"? There seem to be quite a few little "cute ideas" in that
>pile...
>
>Yes, which is why I said "OLPC's Openfirmware (OFW)".
>
>I haven't been able to find the answer to my question anywhere in the
>ISO9660 standards.  It could be a bug in OFW (again, OLPC's).  This
>wouldn't be surprising, since no one currently uses the iso9660
>driver.  If it is a bug in our OFW, it would be nice to fix it.  If
>it's
>not a bug in OFW, then it would be nice to get syslinux fixed.  I
>suspect it's a grey area, but was hoping someone on the list might
>know.
>
>> 
>> The 0x1X are used by Windows as partitions that aren't actually seen,
>> which keeps Windows from puking on the disc.  I don't remember the
>> exact way it would puke, but it is probably in the archives.  As far
>> as testing 0x96 under Windows, that is easy to do; there is even a
>> command-line option to do so.  If it works I have no objections to
>> changing the default.
>
>I can't seem to find the puking in the archives; the closest I can find
>is this - http://www.syslinux.org/archives/2009-May/012843.html
>Mind dropping a hint as to what I should be looking for?

-- 
Sent from my mobile phone. Please excuse brevity and lack of formatting.