Linux Format forums Forum Index Linux Format forums
Help, discussion, magazine feedback and more
 
 FAQFAQ   SearchSearch   MemberlistMemberlist   UsergroupsUsergroups   RegisterRegister 
 ProfileProfile   Log in to check your private messagesLog in to check your private messages   Log inLog in 

Windows XP Boot Issues
Goto page Previous  1, 2, 3, 4  Next
 
Post new topic   Reply to topic    Linux Format forums Forum Index -> Help!
View previous topic :: View next topic  
Author Message
skecs



Joined: Fri Apr 22, 2005 10:22 am
Posts: 76
Location: Bathurst, NSW Australia

PostPosted: Sat Jun 18, 2005 7:27 am    Post subject: Reply with quote

jjmac wrote:

skecs

>>
I too have done this without any problems arising - however I have also come across issues with both Windows & Linux when configured like this.
>>

Ok .... could you describe one of the issues. What i'm getting at is that often threads like this can sometimes actually become quite informative and revealling. But they raily go into any depth. Possibly as booting is generally percieved to be one of those mysterious and somewhat dangerous areas that people are best to avoid ... they do ... and it just seems to remain as such as a result.

Whats needed is case studies to pull apart and analise. Real disasters, total nukes Very Happy.


I teach IT, as well as supporting a mixed Windows/Mac OS X/Linux network at TAFE NSW, from basic introductory levels ("Push this button to turn the computer on") through to advanced web design (XHTML, CSS, PHP, MySQL, etc) and the computers are also used for digital media (video, audio, graphic design, etc) and have had students kill HDD.

This usaully happens when students are installing Windows NT/2000/XP and get confused with the Windows reboot after partitioning and then do a second install to a logical partition or third primary partition. Sometimes I can recover a drive with gdisk from Symantec, other times I have recovered after using a Red Hat or Suse installation to wipe partitions, install and then back to Windows for students.

The drives are recognised by the BIOS on boot but partitioning software, including Partition Manager up to ver 8 (we just haven't updated), used on Windows and Linux just can't write to the HDD. I suspect it may be the heads issue you elaborated on in your posting.

Once this happens the drives are just junked - we have spent numerous hours trying different methods, BIOS, motherboards, operating systems, etc to recover but to no avail.
_________________
Regards from
Downunder!
.... _
... (0)>
... / /\
.. / / .)
.. V_/_
Linux Powered!
Back to top
View user's profile Send private message
jjmac
LXF regular


Joined: Fri Apr 08, 2005 2:32 am
Posts: 1996
Location: Sydney, Australia

PostPosted: Sat Jun 18, 2005 10:30 am    Post subject: RE: Re: boot fix for 9.1 Reply with quote

skecs

>>
... and have had students kill HDD.
>>

You do have my sympathies hehehe ...


>>
This usually happens when students are installing Windows NT/2000/XP and get confused with the Windows reboot after partitioning and then do a second install to a logical partition or third primary partition.
>>

At least it sounds like an act of experimentation.

>>
Sometimes I can recover a drive with gdisk from Symantec, other times I have recovered after using a Red Hat or Suse installation to wipe partitions, install and then back to Windows for students.

The drives are recognised by the BIOS on boot but partitioning software, including Partition Manager up to ver 8 (we just haven't updated), used on Windows and Linux just can't write to the HDD.
>>

There are no firm specs on just how sw is supposed to deal with the 1st sector. Even on the way the table is read isn't part of any spec. Whether it's read from the top down or bottom up is all dependent on the sw author. So, if there were say two partition records set "active, then if two different programs were to pass the table from different directions. They would both get differing ideas on which was the bootable partition Rolling Eyes.

Most docs iv'e read though, warn against mixing sw around when dealing with your boot. Such as, if it's created/formated with one program then that same program should be used for any additional work/actions performed. Not such an issue with the Linux tools "fdisk, cfdisk and sfdisk", but ... yes ... when it comes to the dos type of 3rd party things.

In any case, if your partitioning/table tools are earlier than current versions ... i would be suspicious of them just for that reason.

Especially if you are using a 2.6.x Linux kernel ... the kernel will now return geometry information compliant with the "ata-ide" specification. Which is a 16 head based geometry. Not the screwy 255 head windows/ibm geometry which isn't part of any spec, even though 95% of the worlds PCs' use it Rolling Eyes. Personally i think the kernel should have retained some compatibility there ... bit they decided not to, so there it is.

From the kernel peoples point of view ... the programers who have writen programs that totally rely on a 255 head geometry are at fault.

What they have done, and this was the case with "parted" in the FC2 installer, the program will ask for the drives geometry. The 2.4 kernel would return the 255 head geometry, as found in the table it self. The 2.6 kernel will querie the drives firmware for it's speced geometry ... which will be in a industry compliant 16 head base.

Bang, one scrambled table coming up. It doesn't bother any of the Systems around though, including Linux. But windows is totally lost with out its' quaint 255 head system.

(grin) that'll teach em, for the ntfs hehe
Smile

A responsible program would open the file itself, and heuristically check for the way the table is writen itself. Instead of expecting the kernel to do everything for it. If the kernels return on the drives firmware, and the actual table geometry differ. Then it's up to the program to inform the user of a potential problem, and wait for user control via some over-ride switch.

Which is exactly how my old Lilo reacted as soon as it came across a 2.6.x kernel when attempting to update its' map file. I had to use the "-P ignore" switch to get it to continue.. Updating Lilo, the warning is still outputed but the behavoure is expected, so ... as the table was written in a 255 head base it assumes thats how it will interpret it. And so continues to do its' map update.

That is a more correct behaviour, even though it involves a somewhat bumpy transition form MS smurfs to industry standards. I prefer the later, and about time to. It's just one of the ways that MS is being put in its' place, imo (grin)

As far as partitioning sw goes. It's only 4, 16 byte records starting at )x1be

and the Linux tools aren't that bad. OK, like all things, they require familiarity. But, thats life. It's like driving a car. A person isn't allowed to just because they wont to. They gotta learn first !

Considering all the spam/spyware/infiltrations/ddos event, and general rubbish that goes on. So to should computers. The whole perception and obscurity surrounding this, i believe, has been a covert plan enacted by MS over time ... to artificially tie people into their system, and no other.

I'm a bit grumpy with Ms at the moment (grin) ... i go to the ASUS site to see if i can get some info/dloads for a new board ... nope, Ms heaven site. So much sw to configure the various facilities, but for MS only. Same goes for "pioneer", and the same for "amd" too.

A person could get the idea that they don't like penguins or something Rolling Eyes

hmmmmm (brooding broooding), what the heck ... who needs em anyway Smile





Howdy Birdman,




>>
Can you have a slave without a master?
>>

No, i don't think you can. I figure that if you have a single ide drive connected to a controller then it will have to be jumpered as master for that controller. And will also probably need to be connected to the end plug of its' ribbon cable as well (80 line ribbon).


>>
The SATA drive (my first experience) has no jumpers and I assumed that it would make itself the first drive. It seems to me that the solution is to make sure that any IDE drive connected is always in slave mode so that the SATA drive is promoted to first drive and always stays there. Does that make sense?
>>

I can see what your getting at but going by Lilos' output that you posted it doesn't seem to be working that way. At least from Lilos' perspective.

My SATA experience is less than yours too Smile, so that should be kept in mind concerning this discussion (grin).


>>
It seems to me that the solution is to make sure that any IDE drive connected is always in slave mode ...
>>


I would be inclined to disconnect the ide drive for the time being, and set up the SATA drive(s) without it possible confusing the Bios. I would also make sure the Bios setup screens have the first SATA drive flagged as the boot drive. I think SATA does a kind of chaining arrangement similar to ide so it must have an allocation as the boot drive as would ide. So that would be "sda" rather than "sdb,c,d.etc". Especially, i wouldn't leave the ide connected to its' data cable. I can't specifically say in certainty that that would/could be problematic, but it is something that can be easily eliminated from the over picture. By stripping down as many inclusions as possible it will become much easier to find/solve the point where the problem is arising.


hmmmm (grin) Rolling Eyes Assuming that nelzs' suggestion isn't right on the head (good_grief.png) concerning splitter cables, hehehe.


>>
That is a good idea for the future but not really an issue with this particular problem i.e. I don't need both IDE drives connected at once but I do need to safely install lilo so that I can boot both linux and Windows.
>>

I guess your referring to the splitter cables ... ok ... but the data connection, if that exists, could be part of the problem. By disconnecting the data cable, well, then you can be sure that it isn't contributing.


>>
I have played with the jumpers on hda. When in slave mode Windows sees the 60GB disk as 20GB and linux denies its existence! Switching the jumpers back to slave mode linux once again shows it as 60GB.
>>

Yes ... disconnect it for the time being. I take it that you are booting into your /dev/sda installs ??? How are you doing that ???

I can understand windows reacting strangely, but for Linux to not be able to see it ... then there certainly a problem there somewhere.

>>
Switching the jumpers back to slave mode linux once again shows it as 60GB.
>>

OK ... As it is /dev/hda and your Lilo output is saying that your /dev/sda is not your first disk ... i would trust Lilos opinion. As far as your Bios is concerned the master boot drive is /dev/hda.

It possibly has some sought of device pool/environment listing that is set that way and when you installed your SATA it was flagged as secondary disks.

So it seems you will have to disconnect your ide completely and then go into your bios setup screens to see what kind of configurability can be effected there. My old Bios had something called a ESCDI update option which basically (i think) caused the Bios to re detect all the attached devices. And in my case ... it was a problem in itself (grin).

Then i would go along the lines with what's been discussed above ...


Assuming that there is a bootstrap on /dev/sda, as your posts above do show a table. And you seem to be booting in anyway Smile

If there isn't just use "dd" to copy the first --=-- 446 bytes --=-- of /dev/hda to a disk image. Then bring it up in "hexedit" to check. Different bootstraps will have different gibberish, but down the end of the file there should be some ascii text along the lines of

-------------------------------------------------------------
00000108 E2 F7 F8 5E C3 EB 74 49 6E 76 61 6C ...^..tInval
00000114 69 64 20 70 61 72 74 69 74 69 6F 6E id partition
00000120 20 74 61 62 6C 65 00 45 72 72 6F 72 table.Error
0000012C 20 6C 6F 61 64 69 6E 67 20 6F 70 65 loading ope
--------------------------------------------------------------

Thats fairly common amongst bootstraps, so if thats there, then thats what yourv'e got.


Also, copy the first 446 bytes of /dev/sda out to another testing and compare them ...

]# diff testimg_hda testimg_sda



If there is a bootstrap on /dev/sda then they should match.


]# dd if=/dev/hda of=testimg_hda bs=446 count=1


Then test your Lilo config/boot using a floppy first, before writing it to disk.

do ...

]# dd if=/dev/sda of=testimg_512_sda bs=512 count=1
]# dd if=testimg_512_sda of=/dev/fd0 bs=512 count=1

You could do that directly, but at least you get a image on disk in the process.

Get rid of the "table=" option in the "other" section of your lilo.conf, i think that only concerns booting with the os2_d chain loader!

Comment in the "boot=" option in your lilo.conf and do ...

]# lilo -t -v -b /dev/fd0

I you need to use "-P ignore" to get lilo to ignore any discrepancies, then do so. Once the config is working properly with your SATA set as boot, then install lilo to /dev/sda1, invoke "cfdisk" and set /dev/sda1 "active".

I think you will have to disconnect hda after you have extracted your boot images to get the SATA recognised as your boot drive though.

As with all guns, be careful when using "dd".


How are you managing to boot currently , by the way (grin)


>>
I have played with the jumpers on hda. When in slave mode Windows sees the 60GB disk as 20GB and linux denies its existence! Switching the jumpers back to slave mode linux once again shows it as 60GB.

cfdisk -Pf /dev/hda

Partition Table for /dev/hda
>>


Yep ... looks like a standard 255 head 63 sector 1024 cylinder Dos geometry...

All the sector/head numbers are ok. The sector counts for your logical partitions will be absent the first 63 sectors (head 0) that is actually part of their real total ... The head/sector numbers for those will mean nothing though ... just Bios over flow (thank you Mr Gates) ...

Notice : When you add 10249407 + 55793745 plus the additional 63 sectors not included in hda1's output you get ...


66043215 sectors (33 021 607.5 K) ... How big did you say it was ?

When you add all the logical sector counts together, plus the hidden 63 sectors that each one contains ...

9895977+4594527+256977+8385867+5494167+9703197+13751577+578214+(63*Cool

It results in 52661007 ...

I guess you must have some free space at the end ?


The above is just my opinion, as i am not an expert on SATA/IDE or booting.



Edit: Jan 2006, spelling.



jm


Last edited by jjmac on Tue Jan 02, 2007 12:41 am; edited 2 times in total
Back to top
View user's profile Send private message
1slipperyfish
Forum Jester


Joined: Mon May 09, 2005 3:52 pm
Posts: 2425
Location: wigan

PostPosted: Sat Jun 18, 2005 11:01 am    Post subject: Reply with quote

i'm sorry to interrupt but that was a mega long post Shocked the longest i have done wasn't as big as that and it had a 300 line java program in it Shocked
sorry again i just had to remark
i think you should get some kind of award for it
paul
_________________
i am a follower of the culture

Back to top
View user's profile Send private message
Rhakios
Moderator


Joined: Thu Apr 07, 2005 12:18 am
Posts: 7602
Location: Midlands, UK

PostPosted: Sat Jun 18, 2005 12:59 pm    Post subject: Reply with quote

A life, perhaps.

Very Happy
_________________
Bye, Rhakios
Back to top
View user's profile Send private message
youlikeicecream
LXF regular


Joined: Fri Jun 03, 2005 12:40 pm
Posts: 721
Location: Oxford

PostPosted: Sat Jun 18, 2005 1:02 pm    Post subject: Reply with quote

probably get one online at ebay
Back to top
View user's profile Send private message Yahoo Messenger MSN Messenger
towy71
Moderator


Joined: Wed Apr 06, 2005 3:11 pm
Posts: 4242
Location: wild West Wales

PostPosted: Sat Jun 18, 2005 2:12 pm    Post subject: Reply with quote

nah, sir bob got them to pull that auction Wink
_________________
still looking for that door into summer
Back to top
View user's profile Send private message
Birdman
LXF regular


Joined: Thu May 05, 2005 12:59 pm
Posts: 145
Location: Wallington, Surrey UK

PostPosted: Sat Jun 18, 2005 3:01 pm    Post subject: Reply with quote

jm said "How are you managing to boot currently , by the way (grin) "

Using a boot floppy with the standard sata.i 2.4 kernel that I created upon install.

jm said "results in 52661007 ...

I guess you must have some free space at the end ? "

Yep. In my old box the disk was only seen as 30GB - so that was all I used Smile

I'm kinda scared of using this dd stuff - well outside of my comfort zone. I do like the idea of installing lilo on floppy to test things. I guess that if the kernel image on /dev/sda1 can be found and windows can boot that I am well on the way. I'll give that a go first me thinks Smile
Back to top
View user's profile Send private message
Birdman
LXF regular


Joined: Thu May 05, 2005 12:59 pm
Posts: 145
Location: Wallington, Surrey UK

PostPosted: Sat Jun 18, 2005 3:28 pm    Post subject: Reply with quote

Install to floppy does not work Sad

Fatal: File system would be destroyed by LILO boot sector: /dev/fd0

However, I have changed my /etc/lilo.conf to remove the

table =

and left

boot = /dev/sda1

I get this in test mode:

Reading boot sector from /dev/sda1
Warning: /dev/sda1 is not on the first disk
Using MENU secondary loader
Calling map_insert_data

Boot image: /boot/vmlinuz
Added Slackware24

Boot image: /boot/vmlinuz26
Added Slackware26

Boot other: /dev/sda2, on /dev/sda, loader CHAIN
Added WinXP *

The boot sector and the map file have *NOT* been altered.

This looks pretty good to me.

Still scared to go live.
Back to top
View user's profile Send private message
skecs



Joined: Fri Apr 22, 2005 10:22 am
Posts: 76
Location: Bathurst, NSW Australia

PostPosted: Sun Jun 19, 2005 4:36 am    Post subject: Reply with quote

Quote:

>>
Can you have a slave without a master?
>>

No, i don't think you can. I figure that if you have a single ide drive connected to a controller then it will have to be jumpered as master for that controller. And will also probably need to be connected to the end plug of its' ribbon cable as well (80 line ribbon).

Quote:

>>
I have played with the jumpers on hda. When in slave mode Windows sees the 60GB disk as 20GB and linux denies its existence! Switching the jumpers back to slave mode linux once again shows it as 60GB.
>>

Yes you can have a slave without a master on IDE. Using 80-pin cable (ATA 100/133) the master is the centre connection and slave is the end connection. This is the order when Cable Select is used and it is preferable to keep this even if you are jumpering Master - Slave drives. This could be the problem with the 60 GB - 20 GB - disappear problem.

Quote:

>>
The SATA drive (my first experience) has no jumpers and I assumed that it would make itself the first drive. It seems to me that the solution is to make sure that any IDE drive connected is always in slave mode so that the SATA drive is promoted to first drive and always stays there. Does that make sense?
>>


There is no such thing as Master - Slave when using a SATA connection. Only one drive can be connected to each SATA connector. The long thin cables only have a motherboard connector on one end and a HDD connector on the other.[/quote]
_________________
Regards from
Downunder!
.... _
... (0)>
... / /\
.. / / .)
.. V_/_
Linux Powered!
Back to top
View user's profile Send private message
jjmac
LXF regular


Joined: Fri Apr 08, 2005 2:32 am
Posts: 1996
Location: Sydney, Australia

PostPosted: Sun Jun 19, 2005 12:34 pm    Post subject: Reply with quote

>>
Yes you can have a slave without a master on IDE. Using 80-pin cable (ATA 100/133) the master is the centre connection and slave is the end connection.
>>

curious ... Thats how i always thought of it when i was using a 40 pin 40 cable ribbon. But with a new ASUS 8VT-D board it suggests the other way round Rolling Eyes

It comes with two 40pin 80 cable idea ribbons, coloured black. With a blue connector to connect to the board, a black termimating connector to connect to the master ide drive, with the middle one colourd grey to connect to the slave. So thats how i set it up. The cdrom cable was all black and 40 pin 40 cable ribbon. T just put in a Pioneer DVR A09 LXB drive which suggests to use a 80 cable ribbon, so i used the second one for that. With the teminating black plug connected to it, not the middle gray. And jumpered as master.

It works fine .... ???

Not meaning to contradict, but it does seem to be an interesting quirk worth resolving Smile.

>>
Yes you can have a slave without a master on IDE. Using 80-pin cable (ATA 100/133) the master is the centre connection and slave is the end connection. This is the order when Cable Select is used and it is preferable to keep this even if you are jumpering Master - Slave drives.
>>

How do you implement "cable select" ?, what is "cable select" ?, If there is only one drive on a controller doesen't that imply "master" in that "master" implys "first". And as there is only "one", how could it be anything else.

Does cable select require that ordering exclusively. Would a non-cable sellect system be different, or possibly, be able to be configured both ways. I can see the point,where,as the center connector is closer to the controller it should be more efficient in that it is unemcumbered by having to pass over an intermediating connector. And i believe there is a length limit of 18 inches for ide cables. But, why would ASUS direct to use the end connector as "master" ?


>>
This could be the problem with the 60 GB - 20 GB - disappear problem.
>>

Iv'e got a 160GB WD JB disk as master on the end connector, with a 30GB Maxtor as slave on the middle one (ASUS's idea) ... booting to the second from the first. There both fully seem by the Bios.

>>
There is no such thing as Master - Slave when using a SATA connection.
>>

What ... there would be some sought of sequencing/array scheme involved, ... no ?


>>>>
>>>>>>
Can you have a slave without a master?
>>>>>>

No, i don't think you can.
>>>>

I'm meaning that a single drive would default as master anyway. But you seem to suggest that it dosen't, at least in terms of cable select ???

That seems strange to me. If the cable position was so important, then wouldn't the controller be able to tell which was closer by a signal reflection. But then ... jumpering would make that explicit in case of controller bugs/problems interfering in that kind of detection. Is that what you mean by "cable-select" ?.

>>
It seems to me that the solution is to make sure that any IDE drive connected is always in slave mode so that the SATA drive is promoted to first drive and always stays there. Does that make sense?
>>

hmmm, very good (grin). It does actually. So, the jumpering would over-ride any implicit association. Makes sense. No need to disconnect then.

I would be very interested in your thoughts on why would ASUS advise to use the end connector for the jumpered master and the middle as slave. It did disturb me when i first read it. I though ... would it be a misprint... and it hasn't caused any problems (as yet).


>>
istall to floppy does not work

Fatal: File system would be destroyed by LILO boot sector: /dev/fd0
>>

The floppy shouldn't have a fs on it, it should just be blank ... comment out the "boot=" config parameter in lilo.conf first.

lilo -b /dev/fd0

It's just telling lilo to install to te floppy. That sounds like the flopy is trying to be installed or something.

>>
Reading boot sector from /dev/sda1
Warning: /dev/sda1 is not on the first disk
>>

Don't go live yet. Try skecs' suggestion above. He's probably right about jumpering the single ide as slave. I hadden't quite thought of it that way before. Smile.

If lilo.conf had the "boot=" config parameter pointing at /dev/sda, and then you ran lilo with a "-b" option pointing at your floppy ... that may have been the cause of lilo not wanting to continue. One thing about lilo, it will go to extents to look after its' user Smile

>>
This looks pretty good to me
>>

Looks good to me too, don't go live untill the "first drive" issue is fixed though. And it should be straight forward to get it installed on a floppy.



>>
Im sorry to interrupt but that was a mega long post the longest i have done wasn't as big as that and it had a 300 line java program in it
sorry again i just had to remark
i think you should get some kind of award for it
paul
>>

You are forgiven (grin), you didn't actually count the lines, did you Rolling Eyes


>>
A life, perhaps

Very Happy
>>

I finally found a font that i can read properly in a submit box (grin), so, i must be on a roll Smile .

A life ... hmmmmm, if i could afford a MAC, i would. Walked into an Apple shop the other day with various G%'s on display ... with radically cool LCD screens. Total droul (grin).


>>
I'm kinda scared of using this dd stuff - well outside of my comfort zone. I do like the idea of installing lilo on floppy to test things. I guess that if the kernel image on /dev/sda1 can be found and windows can boot that I am well on the way. I'll give that a go first me think
>>

Then don't use it, it's good that you display some common sense there Smile

A lot of people just jump in half cocked and screw themselves up no end. Caution will save time in the long run, thats for sure. "dd" should never be approached lightly. But once used to it can be a very usefull tool. You could look for a dos program called "rawrite.exe", it will often be on a Linux installation disk. That may help.

As far as getting a mbr on your first SATA, well, not knowing much about SATA i can only approach it from what i would do with a standard ide.

Arrrr, something novel ... this could be a bit mad but ... why not, boot into your Linux, then install Lilo to the first sector of you ide drive. That will give you a backup of the mbr in "/boot/boot.0300". Backup that file for good measure, then uninstll Lilo with ...

]# lilo -u

That will restore the mbr, then use rawrite.exe to copy "boot.0300" to a blank floppy. Then bring the floppy up in "fdisk", is that possible, iv'e never tried to bring a floppy up in a fdisk type program before ?, anyway ... try that and just do any table edits you think are needed. Then reboot with the floppy !


jm
Back to top
View user's profile Send private message
Rhakios
Moderator


Joined: Thu Apr 07, 2005 12:18 am
Posts: 7602
Location: Midlands, UK

PostPosted: Sun Jun 19, 2005 1:31 pm    Post subject: Reply with quote

jjmac wrote:
A life ... hmmmmm, if i could afford a MAC, i would. Walked into an Apple shop the other day with various G%'s on display ... with radically cool LCD screens. Total droul (grin).


I said a life, not iLife Laughing

Believe me, having a mac does very little, oh alright nothing, to get you the former, though it comes with the latter.
_________________
Bye, Rhakios
Back to top
View user's profile Send private message
Birdman
LXF regular


Joined: Thu May 05, 2005 12:59 pm
Posts: 145
Location: Wallington, Surrey UK

PostPosted: Sun Jun 19, 2005 2:06 pm    Post subject: Reply with quote

Setting the /dev/hda to slave does not work - linux cannot see the drive. So that's out.

If I install lilo to the /boot partition and it does not work can I not "lilo -U" to uninstall?

If that is the case, what is there lose?
Back to top
View user's profile Send private message
Birdman
LXF regular


Joined: Thu May 05, 2005 12:59 pm
Posts: 145
Location: Wallington, Surrey UK

PostPosted: Mon Jun 20, 2005 8:21 pm    Post subject: Reply with quote

Another step forward Smile

I managed to install lilo on a floppy. The output was:

bash-3.00# lilo -C /etc/lilo.flop
Warning: The boot sector and map file are on different disks.
Added Slackware24 *
Added Slackware26
Added WinXP

I gave it a test run and now have a result.

Slackware24 boots fine Smile
Slackware26 doesn't boot. I am fine about that because it does all of the correct things and then has a kernel panic. This is as expected because I rushed the compile and knew that I would have to go back and try again. At least now I can get on with that as I now have a means of booting and testing the kernel.
WinXP did not go anywhere. I suppose I was expecting this also - hence not going live. I need a way of invoking the chain loader to get XP on its way.

Cann't be that far away now Wink
Back to top
View user's profile Send private message
jjmac
LXF regular


Joined: Fri Apr 08, 2005 2:32 am
Posts: 1996
Location: Sydney, Australia

PostPosted: Tue Jun 21, 2005 12:37 pm    Post subject: Reply with quote

Howdy Birdman ...

So, you got it onto a floppy, good. Don't be concerned about the "Warning: The boot sector and map file are on different disks." message. Thats common when your main install is on an other disk. In this case involving a floppy Smile.

But lets see if i'm following you correctly. I'm assuming your current state is ...

Three (3) OSes configured in "/etc/lilo.flop", to boot

Slackware24 <---- default system to boot

And two others ...

"Slackware26" and "WinXP"

I'm not sure where you have those at the moment, but as i follow it you have both Linux kernel images in /dev/sda1, both mounting seperat root directories somewhere on /dev/sda.

And you have your WinXP installed on /dev/sda2

So your config should look something like the following ...

---------------------------------------------------------------------------
boot = /dev/sda1
root=/dev/<default_root_partition>
map=/boot/.map
delay=300
read-only
vga=normal
install=text <--- in your case it would be 'menu'
defaut=Slackware34



image=/boot/vmlinuz
root=/dev/<root_partition>
lable=Slackware24
read-only



image=/boot/vmlinuz26
root=<root_partition>
lable=Slackware26
read-only

# Other non Linux OSes using the CHAIN loader

other=/dev/sda2
lable=WinXP
------------------------------------------------------------------------------

Thats about it really Smile, especially for the XP configuration. Nothing really elaberate should be needed.

If you have a number of different kernels for the same installation, then you wont need the "root=" parameter. But if you are booting into a different Linux, and so a different root partition, you will need the "root=" parameter set. It __must__ come after the "image=" parameter. All the configuration parameters that follow an "image=" parameter will over-ride the global settings. And stay in force untill the next "image=" parameter. If a parameter isn't provided under an "image=" parameter, then the global setting for that will apply.

So, if you had a global "append=" parameter set it would apply for all images.

If you had Windows on the second drive, according to the Bios, and it wont boot unless its on the first, you can fool the bios by setting the "other=" parameter up to correctly point to the partition on the second drive, but under it add ...

map-drive=0x80 to=0x81
map-drive=0x81 to=0x80

The Bios will then think that te second drive is the first, but only for that case.

>>
Slackware26 doesn't boot. I am fine about that because it does all of the correct things and then has a kernel panic. This is as expected because I rushed the compile and knew that I would have to go back and try again.
>>

(grin) hehe, Lilo can only do so much Smile

>>
WinXP did not go anywhere. I suppose I was expecting this also - hence not going live. I need a way of invoking the chain loader to get XP on its way.
>>

Normally the configuration above for XP should work. If /dev/sda is the device that the Bios looks at first to find it's boot strap. And that bootstrap is installed on the first sector of that drive aka the mbr. And that first sector also has a viable table that has /dev/sda1 set "Active". Which is your boot directory with Lilos' 1st stage installed to its' first sector.

Could you post the /etc/lilo.floppy file. As it's booting your Linux it would be worth looking at.

Seems the first drive problem is the main bug to work out here. And it shouldn't really be such a problem.

Some explicit return please:

# What does your Bios say ... What is the boot order in your Bios ????
# What does the table in /dev/sda look like ????

]# cfdisk -Pt /dev/sda

If XP is correctly referenced in the table on /dev/sda, then redoing the floppy but adding under the "other=" parameter for XP "table=/dev/sda" should tell the Chain loader where to look for the XP partition.

But, when Lilo is on the boot drive, and the table from there is, of course, being used ... then that parameter wouldn't be needed.

I can't really see why you can't get hda out of the picture.

# How did you set hda to slave ????
# By changing the jumper ????
# What stoped you from installing to the floppy before hand ????

# Have you tryed disconnecting the ide drive ????

# Did you look for rawrite.exe on one of your Linux install discs ????

Once the 1st stage is loaded, and in the case of the floppy, it can't help but be loaded as it will be on the first sector ... then it will reference its' "map" file directly. The geometry fields in the table are right out of the picture. so naturally the Linux references boot correctly. But the "chain loader" will need a table to to boot the XP reference.

You must have, or you have set the floppy to be first on the list of places that your Bios looks at to find a bootstrap. Though this time it found a Lilo (grin) Smile.

I think the solution is hiding right under your nose somewhere Very Happy

Post the current floppy confige and the cfdisk output for /dev/sda, i'm sure they will be interesting. And disconnect that ide, and put /dev/sda as second on the boot sequence in the Bios.

Good Luck Smile


jm
_________________
http://counter.li.org
#313537

The FVWM wm -=- www.fvwm.org -=-

Somebody stole my air guitar, It happened just the other day,
But it's ok, 'cause i've got a spare ...
Back to top
View user's profile Send private message
Birdman
LXF regular


Joined: Thu May 05, 2005 12:59 pm
Posts: 145
Location: Wallington, Surrey UK

PostPosted: Tue Jun 21, 2005 10:06 pm    Post subject: Reply with quote

Right then jm, you want some info Smile

# Start LILO global section
lba32
default = Slackware24
boot = /dev/fd0
prompt
timeout = 100
# Override dangerous defaults that rewrite the partition table:
change-rules
reset
vga = normal
# End LILO global section
# Linux bootable partition config begins for Slackware24 new kernel 2.4.29
image = /boot/vmlinuz
root = /dev/sda3
label = Slackware24
append="hdd=ide-scsi"
read-only
# Linux bootable partition config begins for new kernel 2.6.11
image = /boot/vmlinuz26
root = /dev/sda3
label = Slackware26
read-only
# DOS bootable partition config begins
other = /dev/sda2
label = WinXP
# table = /dev/sda
# DOS bootable partition config ends

> How did you set hda to slave ???
I didn't. Linux can only see the drive as Master.

>What stoped you from installing to the floppy before hand ???
I cocked-up Smile Having read a mini-HOWTO I reaslised my errors and got it right the second time.

>Have you tryed disconnecting the ide drive ????
No.

>Did you look for rawrite.exe on one of your Linux install discs ????
NO.

The reason for not following through is lack of time and divided interests. Now that I have a way of testing a new kernel I am keen to get that sorted (see my other post - "SCSI subsytem - libdata?"). So yes, I do want to follow through and get to the bottom of this booting business - after all, isn't that what linux is all about - problem solving? Having lilo on a floppy is pretty good. There is only a 2 or 3 second speed hit and I can boot any linux kernel I wish. If WinXP is need, just take out the floppy and reboot Smile I do not want to waste time fiddling with nvidia drivers for the current kernel and I am having to suffer with a crap flickery vesa screen at the moment. This places kernel config above lilo in the pecking order.

Anyhow, back on track.

Partition Table for /dev/sda

---Starting--- ----Ending---- Start Number of
# Flags Head Sect Cyl ID Head Sect Cyl Sector Sectors
-- ----- ---- ---- ---- ---- ---- ---- ---- ----------- -----------
1 0x80 1 1 0 0x83 254 63 18 63 305172
2 0x00 0 1 19 0x07 254 63 1023 305235 41769000
3 0x00 254 63 1023 0x83 254 63 1023 42074235 32017545
4 0x00 254 63 1023 0x05 254 63 1023 74091780 43905645
5 0x00 254 63 1023 0x82 254 63 1023 63 1526112
6 0x00 254 63 1023 0x0B 254 63 1023 63 42379407

I haven't fiddled with the BIOS much. I recall going in there to set boot order as floppy, cdrom and then HD but cannot remember whether the SATA was in there. Must take a look.

Anyhow, I won't have time for a few days to do any more work on this. Patience I say, patience.
Back to top
View user's profile Send private message
View previous topic :: View next topic  
Display posts from previous:   
Post new topic   Reply to topic    Linux Format forums Forum Index -> Help! All times are GMT
Goto page Previous  1, 2, 3, 4  Next
Page 3 of 4

 
Jump to:  
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
Linux Format forums topic RSS feed 


Powered by phpBB © 2001, 2005 phpBB Group


Copyright 2011 Future Publishing, all rights reserved.


Web hosting by UKFast