[syslinux] Inconsistencies between ExtLinux and IsoLinux

Matthijs ter Woord matthijsterwoord at gmail.com
Mon Mar 16 03:26:28 PDT 2009


The files listed on the SoC page of the dynamic loader project contain
a test harness to build a harddisk image, it also contains
isolinux.bin: i used the exact files to build the iso.



On Sun, Mar 15, 2009 at 9:42 PM, Stefan Bucur <stefanb at zytor.com> wrote:
> Hello again,
>
> On Sun, Mar 15, 2009 at 1:16 PM, Thomas Schmitt <scdbackup at gmx.net> wrote:
>> Hi,
>>
>> Stefan Bucur wrote:
>>> contains the "-R" paramter, which enables Rock Ridge information to be
>>> included in the ISO. However, according to the ISOLINUX documentation
>>> [1], it "only uses the plain ISO 9660 filenames, i.e. it does not
>>> support Rock Ridge or Joliet filenames.". Therefore, the ISO image you
>>> built may not work as expected with ISOLINUX (maybe it does not hang
>>> from the beginning, but you can expect strange behavior, from time to
>>> time, as it just happened).
>>
>> This theory would mean trouble for my own
>> project which produces ISO 9660 with Rock Ridge
>> unconditionally.
>> I doubt it for now, because:
>>
>> - Rock Ridge images with ISOLINUX booted in my
>>  tests from CD, DVD and USB sticks (the latter
>>  only if the BIOS is willing to boot that way
>>  in general).
>>
>> - Rock Ridge entries reside in the ISO directory
>>  records as a kind of trailing garbage.
>>  First come the ISO 9660 data, then in the
>>  System Use Are come entries under the SUSP
>>  protocol which also governs Rock Ridge.
>>  It is hard to imagine why a ISO-only reader
>>  should not just ignore the extra data.
>
> Thomas, you're right. Now I tested the ISO built with the -R option
> and everything still works, so Rock Ridge information doesn't have
> anything to do with the issue. Moreover, hpa's "-l" parameter enables
> longer file names and you don't even have to stick with the 8.3 naming
> convention.
>
>>
>>
>> Matthijs ter Woord wrote:
>>>  mkisofs -R ... -hard-disk-boot ...
>>
>> Can it be the option -hard-disk-boot which makes
>> the difference ? At least it seemed not be needed
>> with Stefan's test:
>>> mkisofs -o ../Boot.iso -b isolinux.bin -c isolinux.cat -no-emul-boot
>>> -boot-load-size 4 -boot-info-table .
>
> Yes, this might be the real cause of trouble. However, using the most
> recent version of isolinux.bin, I get the following messages when
> trying to build with -hard-disk-boot:
>
> I: -input-charset not specified, using utf-8 (detected in locale settings)
> Size of boot image is 4 sectors -> Emulating a hard disk
> genisoimage: Warning: boot image './isolinux.bin' MBR is not a boot sector.
> Warning: partition not marked active.
> Warning: partition does not start at 0/1/1.
> Warning: image size does not match geometry (4233034)
> genisoimage: Boot image './isolinux.bin' has multiple partitions.
>
> The ISO image gets only 34K, and cannot boot. Matthijs, where did you
> take your isolinux.bin file from?
>
>>
>>
>> I have to admit that i know only the ISO/RR
>> side of the topic of this thread.
>> So i have few idea what goes on after the BIOS
>> hopped from El Torito to ISOLINUX.
>> About mkisofs -hard-disk-boot i know just what is
>> written in the man page.
>
> Well, I'm not familiar at all with the ISO standard details (today is
> the first time I dig deeper into the topic) and I showed you what I
> found to be working and tried to deduce what was the cause of the
> original issue based on what I found on the ISOLINUX wiki page. So
> thank you for pointing me out the problems with my argument :)
>
> Cheers,
> Stefan
>
> _______________________________________________
> Syslinux mailing list
> Submissions to Syslinux at zytor.com
> Unsubscribe or set options at:
> http://www.zytor.com/mailman/listinfo/syslinux
> Please do not send private replies to mailing list traffic.
>
>




More information about the Syslinux mailing list