Are there any drawbacks in using Hibernate instead of Shutdown?

[AndyHui]

I'm not saying that everyone that changes hardware while their computer is in hibernate did it intentially, but they way I read it, it sounded that way

You don't think I'm that stupid, do you? 😉

 

[PC Freak]

🙂 No, of course not. Mama says "Stupid is, is stupid does."

AndyHui, any chance they question may get added to you FAQ?

 

[Auric]

With 2000, coming out of hibernation is much quicker than booting but with XP it is not so I do not see any advantage to a manual shutdown. However, I use hibernation via timed power settings to "shut down" for me after finishing a job since curiously there is not an option for shut down.

When I have used IAA, then upon resume I must F8 / del hibernation stuff and boot normally or else my CD drive will not be accessable IAA has this problem with two different drives.

 

[AndyHui]

AndyHui, any chance they question may get added to you FAQ?

Let me put together a proper, comprehensive answer first. I'll put it on my to-do list. 🙂

 

[Nothinman]

That does not occur.

The Win2K boot loader loads the Hibernation File straight without giving you the option of booting to Windows 98.

Of course, if you are running a 3rd party boot loader....

That does occur, it's the reason I stopped hibernating my Win2K box. And yes I'm using a 3rd party bootloader.

 

[NogginBoink]

Originally posted by: VirtualLarry

Originally posted by: AndyHui
It can cause problems when hardware doesn't initialise properly.

This is a real problem when you forget that you have hibernated the system, and then do a hardware upgrade. Upon boot, the system just goes wacko.

Even more serious, is the fact that the OS filesystem/HD caches, don't get flushed/refreshed from coming out of hibernate. So if you dual-boot, and choose to hibernate W2K, and boot into Win98se, and then restart and then un-hibernate W2K, if you made any filesystem changes at all (talking about FAT32), then W2K will NOT see those changes, and if then then write any data back to the HD/filesystem, from the stale hibernated buffer caches, you can SERIOUSLY screw up the filesystem.

I think that hibernate is intended for, and should ONLY be used on, laptop computers. (But not ones set up to dual-boot.)

The stupid thing is, it should be possible to design the hibernate feature such that this isn't an issue, by flushing all dirty caches when hibernating, and invalidating all caches when coming out of hibernate, such that they need to be re-loaded. (Much in the same way as when disabling and re-enabling a CPU for a hot-swap in a multi-proc system.)
FAT32 volumes even have a timestamp written, upon every FS update. This is used for some RAID purposes, but it could also be used for at least partial checking of FS updates newer than when the system was hibernated.

But of course, MS doesn't get it right. I don't think it's even possible for MS to "get it right". They make so many fundamental design mistakes, it's amazing that their software even works at all. So IMHO, because of the FS-corruption bug, I would have to say that in anything but the most trivial cases, hibernate is broken.

Interesting accusations you make here. Do you have any data to support these conclusions?

 

[NogginBoink]

Originally posted by: ugh
Dude, I'm not blaming anyone. I'm just just saying that I had a problem with the sound card I used last time. And BTW, it was using Win2k's driver.

Drivers that ship with W2K are not written by MS. They're written by the hardware vendors and shipped with the OS.

 

[ugh]

Drivers that ship with W2K are not written by MS. They're written by the hardware vendors and shipped with the OS.

Even so, it should have been certified by MS to be included with the OS.

 

[igiveup]

Microsoft had enough of a hard time getting Windows 2000 out the door (for those that can remember the years of slipped launch projections). Actually they do all they can to get signed drivers out there and for vendors to use the WHQL approach to drivers, but lets face it, it costs money for vendors to do that. A driver included on the shipping OS cd (even if it doesn't implement all of the functions or is slightly flawed with some product) is better than not having any driver at all.

Run sigverif from the Start>Run command line, at least in Windows XP: havent tried 2000, to find out how many unsigned drivers you have out there.

Lets face it, right now its optional. I actually don't think I even have ONE signed driver installed on this OS (Win XP, SP1)

 

[CTho9305]

I see no reason to shutdown/reboot to swap CD drives/burners when I can just hibernate and leave everything running. It's never caused me any problem.

 

[VirtualLarry]

Originally posted by: AndyHui

So if you dual-boot, and choose to hibernate W2K, and boot into Win98se, and then restart and then un-hibernate W2K, if you made any filesystem changes at all (talking about FAT32), then W2K will NOT see those changes, and if then then write any data back to the HD/filesystem, from the stale hibernated buffer caches, you can SERIOUSLY screw up the filesystem.

That does not occur.

The Win2K boot loader loads the Hibernation File straight without giving you the option of booting to Windows 98.

Well, depends.. if you boot off a Win98se boot floppy.. it doesn't. Among other ways. It was quite scary, booting into W2K again, and seeing directories listed in Explorer, that stayed there even when I hit F5 to refresh, even though I *know* I deleted those directories from the filesystem, while in the other OS.

I know, "Don't do that!", but the point is, it's easy to do, and MS could also have an easy fix for it, but they don't, because they don't design things properly, and only implement a half-baked solution, that only partially works.

(Much like the "Backup" program bundled with Win9x, that backed up long filenames, but not the associated short ones, so that it was utterly useless for a system backup. In fact, the W2K readme/install guide text file finally documents that fact.)

For FAT32 volumes, I believe that there is a bit somewhere in the same bitfield that stores the state information about "clean" vs "unclean" shutdowns. (The bit that auto-scandisk on boot checks in Win9x.)

I'm aware of the dirty/clean shutdown bitfield.

My point is that it is not possible to distinguish between a shutdown and a hibernate without powering the system on.[/quote]

Surely, the NTLDR knows, somehow. I'm pretty sure I remember reading about a "hibernated" bit set in the same bitfield along with the "unclean shutdown" bit. I'll see if I can find that info again. It might also store some sort of header or signature at the beginning of the HIBERFIL.SYS file, that indicates if it should be used or not.

 

[Nothinman]

My point is that it is not possible to distinguish between a shutdown and a hibernate without powering the system on.

Surely, the NTLDR knows, somehow. I'm pretty sure I remember reading about a "hibernated" bit set in the same bitfield along with the "unclean shutdown" bit. I'll see if I can find that info again. It might also store some sort of header or signature at the beginning of the HIBERFIL.SYS file, that indicates if it should be used or not.

He means no way to distinguish for you, the user that wants to change hardware while the box is off. You can't tell if it's hibernated or shutdown without turning it on first.

 

[NogginBoink]

Originally posted by: Nothinman

My point is that it is not possible to distinguish between a shutdown and a hibernate without powering the system on.

Surely, the NTLDR knows, somehow. I'm pretty sure I remember reading about a "hibernated" bit set in the same bitfield along with the "unclean shutdown" bit. I'll see if I can find that info again. It might also store some sort of header or signature at the beginning of the HIBERFIL.SYS file, that indicates if it should be used or not.

He means no way to distinguish for you, the user that wants to change hardware while the box is off. You can't tell if it's hibernated or shutdown without turning it on first.

The system "knows" to come out of hibernate by changing the ARC path in the boot.ini.

However, from a user's perspective, looking at the machine, there's no clear indication that the machine is in a suspend state vs. a true power off state.

 

[VirtualLarry]

Originally posted by: NogginBoink

The system "knows" to come out of hibernate by changing the ARC path in the boot.ini.

However, from a user's perspective, looking at the machine, there's no clear indication that the machine is in a suspend state vs. a true power off state.

Interesting. Do you know exactly what it changes? (Not really convenient for me to test right now.)

 

[sak]

Originally posted by: AndyHui
You've obviously never added RAM while the system was down in hibernation, have you?

I have also had problems with older version of the IAA drivers, where drives would not work after coming out of suspend.

Well RAM would be a totally different issue..now would'nt it...cause its the ram that is being stored on to the HD's

 

[sak]

Also i dont understand why u guys are using hibernate state...does not ur computer totally go off in syspend mode...cause my computer is totall turned of..nothing is work no fan no nothing...

and that is was i use when i have to turn of my comp.

besides on suspend mode ur comp comes back up ln seconds insted of a min..in Hib...

 

[Nothinman]

Also i dont understand why u guys are using hibernate state...does not ur computer totally go off in syspend mode...

My notebook is not completely off in standby or suspend mode, that's why it only takes a few seconds to start, there's still power being put to memory.