Re: Not able to boot from SCSI disk on SYM21002 but all other works
- From: Manfred Preussig <PreussigM@xxxxxx>
- Date: Fri, 10 Aug 2007 01:30:29 +0100
Michael Baeuerle schrieb:
Manfred Preussig wrote:Oh, it's really silly. Of course one can say: that's the first to check! But before someone do so please answer yourself the following question: If I would be in this situation and had checked the whole disks through linux-dd-read for broken sectors (head-crashes for example. My power supply had damaged my motherboard. Of course I was afraid of any kinds of electronical damage at any part) and nothing would fail -would I look at the cables? And it looks a little bit like that was the reason -but only at some using params. I don't know details because the information is not very old but as I used another leant host adaptor but this time only a 815S (8 bit and 10 MHz) it works. With the 21002 and the 22801 I had used 16 bit and 40 MHz. The termination was external (to drives, not to the PC), since some of the drives even have no and I didn't wanted to depend on the order.[...]
The BIOS itself needs accesss to drives of course to boot and so on.
This is implemented as a lot of routines with a central dispatcher
called INT13. The BIOS works in real mode (a simple processor mode) and
DOS too. What was easier to do as to use the INT13 from DOS too. And so
it does.
Yes.
The Lilo of course boot in real mode too since that's the only
mode the processor can do without help from OS (it needs some tables
which have to be build in real mode and depends on the OS before it
switches up to other modes). So the Lilo uses INT13 too.
Yes.
To dispatch the
calls between the different drives the BIOS uses drive numbers a little
bit similar to the major device numbers in Linux. Floppies are numbered
from 0 up, hdd's from 128 (0x80) up. So if there is any drive in the
system this would get the number 0x80 and the next 0x81 and so forth.
The lowest numbered drive with some DOS-reachable formatted partitions
on will contain C: drive then. If there is no C: this means either there
is no disk with DOS-reachable formatted partitions on or there is no
disk. Since I know there are such partitions on the disks (and I know
from the linux boot tests that they are even there -not deleted or
corrupted) the result have to be: there are no disks found. Even drive
0x80 doesn't exist.
Yes.
Before since the mainboard damage causes some additional problems I
thought it is the SCSI BIOS (which has INT13 routines inside too) which
doesn't register the drives found on SCSI so there are no INT13 numbers
asigned to them. That's seems to be the most plausible and simple
suggestion what is happening there. So I wrote the other message here.
But now since with another adaptor never inserted in the damaged system
this have to be changed.
So the SCSI INT13 routines can not be the reason. The BIOS presents a
message that the SCSI BIOS is installed during startup (before boot).
This Installation includes the addition of the SCSI INT13 routines to
the BIOS INT13 system dispather. So if the SCSI BIOS is installed this
additional routines (which of course are needed since the normal BIOS
routines knows nothing about how to handle SCSI) are installed too -even
if the SCSI BIOS is not damaged but this we don't have to think of now.
Ack. Does your SCSI BIOS display the disk numbers? Is the boot disk
really listed as 0x80 or "C:"?
The conclusion is: the SCSI BIOS is not damaged, the INT13 works
correctly, the ROM BIOS is not damaged (since the damaged board is
discarded and changed with another board working fine with Linux from
DVD this we can believe I think) I don't know anything I can look for.
FIXMBR I tried before but this of course (from setup cd I can boot off)
(since there is no parameter) works on drive C: or in detail on the disk
C: resides on ... with the same problems mentioned above.
Forget FIXMBR until the disk is not visible.
So a problem with the disk I can rule out since Linux can use it and the
partitions on it too. So the only problem thinkable on disk are the mbr
or the boot sector of the partition. But if I switch active partition
(which is normally the sda3 with Lilo on it) to sda4 (the windows XP
one) the problem doesn't change. So there is only the mbr code. But in
Linux a long time before I saved the mbr code since windows setup tends
to write new information there and so I wanted to could go back if I
need to do so. This file named sda-mbr (which of course is clear enough
don't to think of using the wrong file) I have wrote back to sda at the
beginning of the problems.
ROM-BIOS C-MOS-RAM deletion I have done. No change.
If the disk is not accessible, the MBR is useless (never executed).
The essence is: I don't know what to do. I'm frustrated. I'm desparate.
I can't boot windows from DVD (only the setup but this won't help) so
all off the data, programs settings and so on on the SCSI disks are
lost. Of course I could buy an IDE drive, make partitions there and copy
all of it to this drives. But it's a real hard and long-term work and
I'm not sure it works for all data since there are some data hidden from
normal access (like licences for music for instance). It is a horror.
Even more since all seems to work -even in spite of booting. So I have
to think it's a kind of simple error. But what?
Any idea?
Maybe it is simply the boot order, sometimes listed as "A, C, SCSI" or
something in the setup. If you don't have set SCSI first there, it is
possible that your SCSI boot disk is accessible via INT13 but not as
0x80. Maybe your host BIOS insert nonexisting IDE disks to 0x80 and
following and your SCSI disks start at 0x82, 0x84 or whatever. No disk
at 0x80 is not necessarily the same as no disks available.
Micha
What kind of error can happen to a cable which makes it possible to read normally with higher rate and width but not at boot and makes it possible to read normally and boot at lower rate and width? I don't know any answer and I have the feeling that I will never know it.
The BIOS version could not be the reason because the BIOS of 815S and of 22801 was the same number. 22002 was higher (4.19 in spite of 4.12). Since I had this version on disk I re-wrote it from disk. Some time ago I wrote it to the same device which was at 4.16 as I got it -oh yes I know this since I wrote it down in a book like a lot of other things related to the system. So if one don't think it was possible that a file and a ROM contents stored not at same place could be damaged the same way by accident we can be sure it's not the BIOS.
Of course I will report any additional information here as I will get it. But at this point the only one I really have is: I inserted a 815S hba and it works. Before with 21002 and 22801 it worked not. The only real (and specifical) difference I see is the speed and bus width. It's really crazy.
greetings
Manfred
.
- References:
- Not able to boot from SCSI disk on SYM21002 but all other works
- From: Manfred Preussig
- Re: Not able to boot from SCSI disk on SYM21002 but all other works
- From: Manfred Preussig
- Re: Not able to boot from SCSI disk on SYM21002 but all other works
- From: Michael Baeuerle
- Not able to boot from SCSI disk on SYM21002 but all other works
- Prev by Date: Re: Not able to boot from SCSI disk on SYM21002 but all other works
- Next by Date: possible speed up for SCSI external ?
- Previous by thread: Re: Not able to boot from SCSI disk on SYM21002 but all other works
- Next by thread: Re: Not able to boot from SCSI disk on SYM21002 but all other works
- Index(es):
Relevant Pages
|
