On Thu, 18 Nov 1999, Heather wrote:
>> >What I do not know, because I set it up correctly early on, is whether it
>> >needs to be the first partition, or whether you can just use ext2resize
>> >and sacrifice a scrap of your /tmp volume to it.
>> For Thinkpad's you use a program PS2.EXE to create it. This program runs
>> under DOS and takes a parameter "C:" or "D:". The limitation is that it has
>> to be a DOS visible partition (which effectively restricts you to the first
>The Magio does not need a DOS app's aid, the Fn-F1 menu section for suspend
>details lets me create the volume from there. That's why I really don't know.
>I haven't scrapped and remapped to find out. I also don't know if the
>(active hibernation feature of) BIOS really needs a DOS partition, or if that
>was a constraint added by the tool which creates the hibernate space (in
>my case, a program in the ROM, in your case, PS2.EXE).
>Would it be possible to use that under Dosemu+DOS so that you can force which
>partition is visible as "D:" then do it?
Maybe. But unless you have really good backups you probably don't want to
try it. Just imagine the bios dumping 100M of data over your favourite file
>Comparing enough hibernation volumes, could someone create a linux mktpadswap?
>Maybe it could encompass or call ext2resize and mkdosfs, and do the whole
>task - carve off the right amount of diskspace for your ram, get it into the
>partition table, format it, and put the hibernate volume in there. Easy
>enough to get -blank- hibernate volumes, just scrap one and recreate it. (Too
>late for us, but we can save next year's crop of users.)
Maybe. I think that the BIOS looks through the file system structures
though. It's the only explanation I know of for Thinkpads only supporting
64M of RAM for hibernation to HPFS volumes...
>> If you want to save 100M of disk space then you could have the same disk area
>> used for hibernation and swap space. At hibernation time you could have the
>> suspend script do "swapoff /dev/hda1 ; gzip -d < /boot/hib.gz > /dev/hda1"
>> where hib.gz is an archive of the part of the /dev/hda1 device containing FAT
>> and the root directory (should only be a few K when gzipped).
>> Then at resume time you do "mkswap /dev/hda1 ; swapon /dev/hda1".
>Considering that I pack and unpack a lot, and have my little kit optimized
>for being on the go, I'm really less than interested in spending a *longer*
>time on the suspend/resume process, full hibernate feels slow enough as it
>is. Thus my comment - it *might* be worth it if the linux kernel does all
>the work, but (IMHO, ymmv) it is *not* worth those scripting tricks and the
>chance of forgetting.
Same here. I've just got a new Thinkpad and I've put the hibernation
partition as the first partition to get the best transfer rates and minimise
>(Anybody here using swap-to-ramdisk tricks so they don't need linux swap
>space on disk at all? Then hibernating would capture your swap too.)
Swap to ramdisk only makes sense on broken hardware (IE hardware where some
RAM is really slow).
>Of course it would work, but there's no reason for me to do it, either. My
>time and sanity are worth more than 100 Mb. I don't feel constrained on a
>3.2 Gb hard disk - my first Linux laptop had a mere 528 Mb.
I felt constrained with 3.2G. My new Thinkpad has 10G. Next thing to do is
to buy more RAM (I have 64M but can expand to 288M).
The hard drive on my new one has a 50% higher transfer rate, so I can have
50% more RAM and keep the same hibernation times. Before I had 96M and
hibernation times were just bearable. Now I can do 128M with ease, or go to
192M and be more patient.
The latest Thinkpads can now handle more than 512M. Mine is only 3 weeks old
and it's behind the times already!
Electronic information tampers with your soul.