In the absence of anything more sensible I think it could be hardware related - specifically memory. It could be environmental (see below at end). It may be worthwhile eliminating grub from the equation as well:
- GRUB
If it is something to do with grub, you may be able to bypass grub and boot directly to Windows XP, by shifting the boot flag from the partition that contains grub to that for Windows.
I/we would have to see the results of
fdisk -l
to give detailed advice. A typical situation would be you have:
Windows XP on the first partition, say /dev/sda1 and
Suse 11.2 installed as logical partitions:
/dev/sda5 (swap),
/dev/sda6 (root) and
/dev/sda7 (home)
inside the extended partition /dev/sda2.
The boot flag is then set on /dev/sda2 and Windows is booted through grub via a chainloader +1 entry. Using, for example, the Suse 11.2 Live CD or the Parted Magic CD, I would then move the flag from /dev/sda2 to /dev/sda1 using parted or cfdisk or fdisk in a command prompt or maybe gparted (gui).
Then, when booting, it would run Windows XP straight away with no grub. If it boots OK, then your problem may be caused by grub. Change the boot flag back to wherever it was after this test.
As I said this is the typical situation and yours may be different.
- PRECONDITIONED MEMORY
When you boot from cold, the bios may run a short memory test to check the RAM. This is not very exhaustive and bad memory may ‘pass’ even if it is defective. However, it may leave the RAM (contents) in a certain condition where you cannot booting XP but maybe you can boot Suse 11.2. When you restart, the bios may bypass the memory test and you start the XP and/or Suse boot with RAM holding ‘acceptable’ contents.
Is there an option in the BIOS setup to bypass all tests? It would be interesting to see what happens then.
In any case you should subject the RAM to a more stringent test via ‘memtest86+’. You will find this on the menu for the Live CD (and DVD?). Run 4 passes of 8 test which will take an hour or more. Any errors in red, isolate the defective RAm chips via a process of elimination. Some people run this test for half a day but I think that is hammering the chips unnecessarily and could shorten their life. It may give rise to another problem: chips overheating, so you inadvertently end up testing for another problem - case ventilation quality.
- COLD or WARM RAM
It could be temperature dependent, although this is a long shot. When you boot truly from cold, the MB and RAM are physically cold but on a restart they have warmed up a little. That may effect the quality of, say, the electrical connections between the RAM chips and the MB. Take the RAM chips out and plug them back in a couple of times to brush dirt and oxide off the contacts. Another test would be, on a cold boot, suspend the boot process by pressing the Pause button and wait for a minute. Then allow the boot to recommence by pressing Esc. This should warm the RAM.
- HUMIDITY (also related to temperature).
On certain days in Sydney it can become very humid and this appears to effect the success of a boot up, especially on aging hardware.
Theory: slight humidity on the MB causes additional conductive paths that, because the impedances of CMOS etc components are so high, look like a virtual short circuit. Thus at certain times or the year there are ‘good boot days’ and ‘bad boot days’. I am sure this will cause a certain amount of incredulity to people reading this. However, in some circumstances boot success is guaranteed if you wait until the humidity drops (could be the following day).
- POWER SUPPLY
A bad power supply unit can cause weird problems as well.
So I hope these ideas will help you (at least you now have another opinion). They could give you a clue on what to try next. They are all long shots unfortunately.
Good luck.![:slight_smile: :slight_smile:](https://forums.opensuse.org/images/emoji/twitter/slight_smile.png?v=12)