Syslinux - Jan 2014 - After USB boot problems on Gigabyte GA-M55Plus-S3G

On 01/21/2014 11:13 AM, Thomas Schmitt wrote:
> 
> My current theory is that vendors prefer to record
> a cylinder aligned end CHS address over recording
> an end CHS address which exploits the whole capacity.
> This is used as an (inofficial ?) protocol to publish
> the CHS factors H' and S'.
> 
Indeed it is -- at least some BIOSes "deduce" the H/S to use from
that.

	-hpa

Hi,

hpa:
> The mkdiskimage tool in the Syslinux distribution exists for this reason.
Then 
  http://www.syslinux.org/wiki/index.php/SYSLINUX#Creating_a_Bootable_Disk
should point to a tutorial for mkdiskimage.

I cannot spot any occurence of "mkdiskimage" in
  http://www.syslinux.org/wiki/index.php/SYSLINUX
  http://www.syslinux.org/wiki/index.php/HowTos

Is there already a description at other places ?


me:
> > This is used as an (inofficial ?) protocol to publish
> > the CHS factors H' and S'.
hpa:
> Indeed it is -- at least some BIOSes "deduce" the H/S to use from
that.
If the end CHS address is regularly visible in partitioning
tools, then this is worth a short statement about the importance
of good H and S values in the end address.
(fdisk -l shows them only when it complains.)


Have a nice day :)

Thomas

cmon, it is not that ugly. before parted, and kpartx it was even worse.


On Tue, Jan 21, 2014 at 1:54 PM, H. Peter Anvin <hpa at zytor.com> wrote:
> On 01/21/2014 11:13 AM, Thomas Schmitt wrote:
> >
> > My current theory is that vendors prefer to record
> > a cylinder aligned end CHS address over recording
> > an end CHS address which exploits the whole capacity.
> > This is used as an (inofficial ?) protocol to publish
> > the CHS factors H' and S'.
> >
>
> Indeed it is -- at least some BIOSes "deduce" the H/S to use from
that.
>
>         -hpa
>
> _______________________________________________
> 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.
>
>

Hi,

i read in
  http://www.syslinux.org/wiki/index.php/Hardware_Compatibility#USB-Geometry
what we should have shown Ronald as first answer.
This text and the next section "USB-Miscellaneous" predicts
all the symptoms.

Besides the typo "examle" i find this statement still hard to
understand. 
 "An examle is a drive of 128,64,32 with a partition ending at
  127,63,32 (the last sector of the last whole cylinder)."

Should not 128,63,32 be the last sector at the end of the
last cylinder and the drive size be 129,0,1 ?
Or should the drive in this example rather have size 128,0,1 ?
(Assumed that H'=64 and S'=32.)

In any case one should state the H' and S' values with this example.
A good place to advertise 64,32 and 255,63.


Have a nice day :)

Thomas

> Hi,
> 
> hpa:
> > The mkdiskimage tool in the Syslinux distribution exists for this
reason.
> 
> Then 
>   http://www.syslinux.org/wiki/index.php/SYSLINUX#Creating_a_Bootable_Disk
> should point to a tutorial for mkdiskimage.
> 
> I cannot spot any occurence of "mkdiskimage" in
>   http://www.syslinux.org/wiki/index.php/SYSLINUX
>   http://www.syslinux.org/wiki/index.php/HowTos
> 
> Is there already a description at other places ?
> 
 
FWIW, see "./doc/usbkey.txt" in official Syslinux archives.

Also at:
http://www.syslinux.org/wiki/index.php/Doc/usbkey

And also at:
http://git.zytor.com/?p=syslinux/syslinux.git;a=blob;f=doc/usbkey.txt;
h=33613d69632b6108a1638fa00abe665386de78dd;hb=HEAD
> me:
> > > This is used as an (inofficial ?) protocol to publish
> > > the CHS factors H' and S'.
> hpa:
> > Indeed it is -- at least some BIOSes "deduce" the H/S to use
from that.
> 
> If the end CHS address is regularly visible in partitioning
> tools, then this is worth a short statement about the importance
> of good H and S values in the end address.
> (fdisk -l shows them only when it complains.)
> 
Fdisk shows the geometry almost always, and it can also be changed if 
necessary.

For the vast majority, I don't see the need to "play" with
geometry
nowadays (and for the few last years already). Just create a new 
partition table with default values.

Regards,
Ady.

On 01/21/2014 12:10 PM, Thomas Schmitt wrote:
> Hi,
> 
> hpa:
>> The mkdiskimage tool in the Syslinux distribution exists for this
reason.
> 
> Then 
>   http://www.syslinux.org/wiki/index.php/SYSLINUX#Creating_a_Bootable_Disk
> should point to a tutorial for mkdiskimage.
> 
> I cannot spot any occurence of "mkdiskimage" in
>   http://www.syslinux.org/wiki/index.php/SYSLINUX
>   http://www.syslinux.org/wiki/index.php/HowTos
> 
> Is there already a description at other places ?
> 
Doesn't look like it, which is indeed an oversight.

The description is as given by the helptext:

Usage: $0 [-doFMz4][-i id] file c h s (max: 1024 256 63)
    -d    add DOSEMU header
    -o    print filesystem offset to stdout
    -F    format partition as FAT32
    -M    "c" argument is megabytes, calculate cylinders
    -z    use zipdisk geometry (h=64 s=32)
    -4    use partition entry 4 (standard for zipdisks)
    -i    specify the MBR ID
    -s    output a sparse file (don't allocate all blocks)

It is worth noting that it, like the "mtools" syslinux installer,
invokes mtools to do the actual formatting.

Hi,

Dean Graff:
> cmon, it is not that ugly. before parted, and kpartx it was even worse.
But imagine the overworked developer shortly before midnight
banging the head on the desk because that @#$%@ USB stick
fails on the demo machine after it succeeded on two
test machines.

Currently it seems we need in the wiki a link from
  SYSLINUX#Creating_a_Bootable_Disk
to
  Hardware_Compatibility#USB-Geometry
and from there a link to a (yet to find) tutorial about
the mkdiskimage tool. (About which i know nothing yet.)

Google found me meanwhile about mkdiskimage:
  http://www.syslinux.org/doc/usbkey.txt
which looks best so far. MS-Windows user might feel left behind, 
though.


Have a nice day :)

Thomas