Mailing List Archive

EDD error RE: Re: SYSLINUX 2.12-pre7 released
I have an application/program which shows 'bios EDD error' when using Isolinux
(with either loading an image directly through MEMDISK, or booting to A:), and not when using a bootdisk directly or booting a bootdisk directly which has SYSLINUX on it.

configuration:
bootable cdrom with Isolinux 2.11/2.12pre9 as bootloader.

1.44MB bootdisk with MSDOS.SYS/IO.SYS and the UDMA2 dos driver,
http://fdos.org/ripcord/beta9sr1/other/misc/Udma2_16.zip (sourcecode included)

DEVICE=UDMA2.SYS
SHELL=A:\COMMAND.COM A:\ /D /P

Doesn't Isolinux clear everything before a localboot 0x00 ?
I guess not, as sometimes the eltorito.sys driver from www.nu2.nu also still loads, which means parts of Eltorito booting are still present in system
(somewhat like KEEPPXE).

FreeDOS kernel does not show the EDD error though, but that piece of software is currently a little bit messy, so not reliable.

Vmware Workstation 5 Beta1, by the way.

also a syslinux bootdisk with 'bootsect.bin' seems to work perfectly, I always thought BSS was required (my mistake?).
Also 2.12pre9 is missing SYSLINUX.EXE, but the .COM seems to work
SYSLINUX A: (I do get a direct hardware access message though on Win2000)



Bernd


----------------------------------------------------------------
Op deze e-mail zijn de volgende voorwaarden van toepassing:

http://www.fontys.nl/disclaimer

The above disclaimer applies to this e-mail message.
----------------------------------------------------------------
EDD error RE: Re: SYSLINUX 2.12-pre7 released [ In reply to ]
Blaauw,Bernd B. wrote:
>
> Doesn't Isolinux clear everything before a localboot 0x00 ?
>

No, it doesn't; although I think there is an El Torito command to
disable the boot device, I don't see any use in calling it, since it's
running in no-emulation mode and therefore doesn't hide any drives. As
you point out, it has the nice effect that you can load the El Torito
CD-ROM driver.

> also a syslinux bootdisk with 'bootsect.bin' seems to work perfectly,
> I always thought BSS was required (my mistake?).

It works as long as you're on the same filesystem as you copied the
sector from. You will wreck major havoc if you try to use that
bootsector on another filesystem, however.

> Also 2.12pre9 is missing SYSLINUX.EXE, but the .COM seems to work
> SYSLINUX A: (I do get a direct hardware access message though on
> Win2000)

Fixed in pre10.

-hpa
EDD error RE: Re: SYSLINUX 2.12-pre7 released [ In reply to ]
(Sorry, Outlook webmail prohibits me from proper replying/quoting)

>it has the nice effect that you can load the El Torito
>CD-ROM driver.

Which seems logical if you boot a kernel (like Linux, or Memdisk with an imagefile), but not when doing a local boot (diskette/harddisk)
In other words: boot from cdrom, and if you then load a disk-image, eltorito.sys should load. If not loading a kernel or disk-image (but booting C: for example, or A:), then, IMHO, eltorito.sys should not be able to load (as the user opted to chainload instead of continuing to boot from cdrom).

I wish I could provide an ISO or diskimage, but the 'bios EDD error' only occurs on MSDOS (IO.SYS) kernel, so I can't spread/provide that disk image.

I'll manage, though. It's just making my testing a tiny bit more difficult (hit ESC each time the machine boots, then select floppy, instead of let isolinux boot and then select 'A' for localboot 0x80 as defined in my isolinux.cfg file)

UDMA2 author mailed this:
------------
and avoids addressing over 20 bits. Re: the "Int 13h" function codes used
by UDMA2, they are (A) 41h to see if an EDD BIOS is present, 48h to request
EDD parameters for every disk and 08h to request CHS values for every disk.
All are "basic" functions that any (good) EDD BIOS should support. If, in

Bernd


----------------------------------------------------------------
Op deze e-mail zijn de volgende voorwaarden van toepassing:

http://www.fontys.nl/disclaimer

The above disclaimer applies to this e-mail message.
----------------------------------------------------------------
EDD error RE: Re: SYSLINUX 2.12-pre7 released [ In reply to ]
Blaauw,Bernd B. wrote:
> (Sorry, Outlook webmail prohibits me from proper replying/quoting)
>
>
>>it has the nice effect that you can load the El Torito
>>CD-ROM driver.
>
>
> Which seems logical if you boot a kernel (like Linux, or Memdisk with an imagefile), but not when doing a local boot (diskette/harddisk)
> In other words: boot from cdrom, and if you then load a disk-image, eltorito.sys should load. If not loading a kernel or disk-image (but booting C: for example, or A:), then, IMHO, eltorito.sys should not be able to load (as the user opted to chainload instead of continuing to boot from cdrom).
> I wish I could provide an ISO or diskimage, but the 'bios EDD error' only occurs on MSDOS (IO.SYS) kernel, so I can't spread/provide that disk image.
> I'll manage, though. It's just making my testing a tiny bit more difficult (hit ESC each time the machine boots, then select floppy, instead of let isolinux boot and then select 'A' for localboot 0x80 as defined in my isolinux.cfg file)
>
> UDMA2 author mailed this:
> ------------
> and avoids addressing over 20 bits. Re: the "Int 13h" function codes used
> by UDMA2, they are (A) 41h to see if an EDD BIOS is present, 48h to request
> EDD parameters for every disk and 08h to request CHS values for every disk.
> All are "basic" functions that any (good) EDD BIOS should support. If, in
>

The way the El Torito CD-ROM is going to look like to the program, is
that it looks like an additional hard disk which a very high drive
number (typically 9Fh or so); MS-DOS will normally ignore this since it
asks the BIOS how many drives there are and ignores the rest.

Does the error message come from UDMA2.SYS or...?

-hpa
EDD error RE: Re: SYSLINUX 2.12-pre7 released [ In reply to ]
>The way the El Torito CD-ROM is going to look like to the program, is
>that it looks like an additional hard disk which a very high drive
>number (typically 9Fh or so); MS-DOS will normally ignore this since it
>asks the BIOS how many drives there are and ignores the rest.

>Does the error message come from UDMA2.SYS or...?

> -hpa

yes, it comes from UDMA2.SYS, it does a series of initialisation tests (read spead for example, and VDS).

I don't know behaviour when booting to MSDOS on harddisk, only Isolinux --> A: and Isolinux --> Memdisk --> Disk image

Bernd

http://fdos.org/ripcord/beta9sr1/other/misc/Udma2_16.zip
output on screen (sourcecode in UDMA2.ASM, line 855 and 736 ??? )

UDMA2 Disk Driver V1.6, 2-Dec-2004. Output overlap is disabled.
UltraDMA controller found at I-O address 1050h.
Vendor/Device I.D. 8086h/7111h. PCI bus 00h, device 07h, function 1
No XMS manager; using only DMA I-O, NO Output Overlap!
Primary-master disk is VMware Virtual IDE Hard Drive.
Set to UltraDMA mode2, ATA-33. Read tests omitted
BIOS EDD error; driver NOT loaded!



----------------------------------------------------------------
Op deze e-mail zijn de volgende voorwaarden van toepassing:

http://www.fontys.nl/disclaimer

The above disclaimer applies to this e-mail message.
----------------------------------------------------------------
EDD error RE: Re: SYSLINUX 2.12-pre7 released [ In reply to ]
Blaauw,Bernd B. wrote:
>
> http://fdos.org/ripcord/beta9sr1/other/misc/Udma2_16.zip
> output on screen (sourcecode in UDMA2.ASM, line 855 and 736 ??? )
>
> UDMA2 Disk Driver V1.6, 2-Dec-2004. Output overlap is disabled.
> UltraDMA controller found at I-O address 1050h.
> Vendor/Device I.D. 8086h/7111h. PCI bus 00h, device 07h, function 1
> No XMS manager; using only DMA I-O, NO Output Overlap!
> Primary-master disk is VMware Virtual IDE Hard Drive.
> Set to UltraDMA mode2, ATA-33. Read tests omitted
> BIOS EDD error; driver NOT loaded!
>

Most likely the UDMA2 driver ends up running into the CD-ROM eventually,
and croaking because it's not a hard disk. It should examine the EDD
information a bit closer; it'd find that it's marked accordingly.

-hpa
EDD error RE: Re: SYSLINUX 2.12-pre7 released [ In reply to ]
>information a bit closer; it'd find that it's marked accordingly.

UDMA2.SYS author now worked around it, it seems you do something with "DPTE" in isolinux.asm but not using it (probably for memdisk or direct floppy/harddisk emulation?).

also something about EDD v1.1 and v3.0 specifications and setting bits to 1.
Anyway, the driver works now on:
BIOS -> cdrom -> Isolinux -> A: -> MSDOS -> UDMA2.SYS -> COMMAND.COM -> command prompt.

I don't have the sourcecode of this latest work-around version available, so can't let you have a look at it.
Any possibility for the 'write to bootsectorfile' in a 2.13 series SYSLINUX?

Bernd

here's a part of the mail:

"
hard-disks (item C above), I believe the Isolinux bootloader may
NOT be setting the DPTE pointer to all-ones if DPTE data in fact
IS NOT present, i.e. for the CD-ROM! In his first E-Mail, your
friend says "examine the data more closely" which makes me think
the Isolinux bootloader denotes a non-ATA unit ONLY through data
available in "new" V3.0 EDD BIOS fields!! He MUST ALSO set the
DPTE pointer to all-ones for non-ATA units, or in my opinion the
Isolinux bootloader is VIOLATING both the V1.1 and V3.0 EDD BIOS
specification! How else, for BOTH an old OR a new EDD BIOS, is
one to know a particular BIOS "disk unit" is NOT REALLY A DISK??
If your friend insists he has NOT violated the specs (as I think
he HAS!), then there is only one answer: Our drivers must FIRST
check for a NEW V3.0 EDD BIOS (74 bytes of returned "parameters"
not just 40), then check the NEW data fields for a non-ATA disk."


----------------------------------------------------------------
Op deze e-mail zijn de volgende voorwaarden van toepassing:

http://www.fontys.nl/disclaimer

The above disclaimer applies to this e-mail message.
----------------------------------------------------------------
EDD error RE: Re: SYSLINUX 2.12-pre7 released [ In reply to ]
Blaauw,Bernd B. wrote:
>>information a bit closer; it'd find that it's marked accordingly.
>
>
> UDMA2.SYS author now worked around it, it seems you do something with "DPTE" in isolinux.asm but not using it (probably for memdisk or direct floppy/harddisk emulation?).
>
> also something about EDD v1.1 and v3.0 specifications and setting bits to 1.
> Anyway, the driver works now on:
> BIOS -> cdrom -> Isolinux -> A: -> MSDOS -> UDMA2.SYS -> COMMAND.COM -> command prompt.
>
> I don't have the sourcecode of this latest work-around version available, so can't let you have a look at it.
> Any possibility for the 'write to bootsectorfile' in a 2.13 series SYSLINUX?
>

No, but you can write a comboot/com32 module to do it if you wish.

> Bernd
>
> here's a part of the mail:
>
> "
> hard-disks (item C above), I believe the Isolinux bootloader may
> NOT be setting the DPTE pointer to all-ones if DPTE data in fact
> IS NOT present, i.e. for the CD-ROM! In his first E-Mail, your
> friend says "examine the data more closely" which makes me think
> the Isolinux bootloader denotes a non-ATA unit ONLY through data
> available in "new" V3.0 EDD BIOS fields!! He MUST ALSO set the
> DPTE pointer to all-ones for non-ATA units, or in my opinion the
> Isolinux bootloader is VIOLATING both the V1.1 and V3.0 EDD BIOS
> specification! How else, for BOTH an old OR a new EDD BIOS, is
> one to know a particular BIOS "disk unit" is NOT REALLY A DISK??
> If your friend insists he has NOT violated the specs (as I think
> he HAS!), then there is only one answer: Our drivers must FIRST
> check for a NEW V3.0 EDD BIOS (74 bytes of returned "parameters"
> not just 40), then check the NEW data fields for a non-ATA disk."
>

He's completely confused. I'm not doing ANYTHING with this stuff, it's
the system BIOS that's doing it, not ISOLINUX.

-hpa
EDD error RE: Re: SYSLINUX 2.12-pre7 released [ In reply to ]
>No, but you can write a comboot/com32 module to do it if you wish.

I meant for syslinux.exe/com (the installer).

that BIOS issue is now handled in the driver the author wrote. Unfortunately I can't find another cdrom which boots but lets me access A: to really demonstrate it's the BIOS which is doing this.
Every non-emulation cdrom either uses isolinux (even ReactOS bootloader) or does not allow booting to diskette.

thanks for the feedback anyway, and good luck releasing 2.12.

Bernd


----------------------------------------------------------------
Op deze e-mail zijn de volgende voorwaarden van toepassing:

http://www.fontys.nl/disclaimer

The above disclaimer applies to this e-mail message.
----------------------------------------------------------------
EDD error RE: Re: SYSLINUX 2.12-pre7 released [ In reply to ]
Blaauw,Bernd B. wrote:
>>No, but you can write a comboot/com32 module to do it if you wish.
>
> I meant for syslinux.exe/com (the installer).
>

Oh, right... I forgot all about that. On that subject, how would you
expect this to work when the bootsector is tied to a particular location
of ldlinux.sys? (I'm not being negative, I am just trying to understand
the requirement(s).)

> that BIOS issue is now handled in the driver the author wrote. Unfortunately I can't find another cdrom which boots but lets me access A: to really demonstrate it's the BIOS which is doing this.
> Every non-emulation cdrom either uses isolinux (even ReactOS bootloader) or does not allow booting to diskette.
>
> thanks for the feedback anyway, and good luck releasing 2.12.

Thanks.

-hpa