Vista an Uneasy Sleeper

Emmy King writes "'One thing we just can't wrap our mind about is the terrible, broken, and completely pitiful support for waking Vista up from a Deep Sleep or hibernation.' Any time you attempt to wake Vista up from Hibernation or "Deep Sleep" (S3-induced sleep mode), it dies. It's either a BSOD, or a driver error, or a broken network, no DWM, lack of sound... the list goes on, and on. So much for an operating system to "power" the future! (No pun intended!) That's with properly-signed drivers and no buggy software on multiple PCs..."

Linux

Linux: Ritalin for your new vista box

Re:Linux

Linux: It doesn't suck.

Re:Linux

How hard can it be?
When Ballmer is throwing chairs all over the office, it is pretty hard to program ACPI stuff. :-)

How hard can it be?

> Linux: It doesn't suck. Indeed. Ubuntu 6.10 wakes up from hibernation just fine, and quickly, even on my old computers. How hard can it be?

How hard? Very!

Linux has had 2 (3?) separate attempts to get hibernate support working properly and while it is pretty good now it still isn't perfect.

Re:How hard can it be?

Since network connections can and do fail with some frequency, any decent networking app should be able to detect and recover from a dropped connection.

I don't see how this is such a huge deal in Vista, anyway. It seems to work fine under XP, and you're going to be running most of the same apps for now...

Re:How hard can it be?

How hard? Very!

You are right about this. It isn't hard for anybody with a bit of coding experience to realize that trying to freezedry, serialize and then defrost an entire multitasking OS full of running tasks and hardware is a very difficult task. Especially when computers today are often busy talking to other computers (you can't really expect every TCP connection to suddenly spring to life where it was).

That said, Ubuntu 6.10 does hibernate very, very, well. Try it.

Re:How hard can it be?

Ugh. I think I am going to stop reading Slashdot now. Okay. So Vista has a problem dealing with S3 resume. Fine.

What bothers me are snide remarks from people who have a very vague (if any) understanding of what is involved in power management support. At all. So Microsoft dropped the non-ACPI HALs with Vista. About time. Considering the number of ACPI "compliant" systems out there, I'm not surprised a lot of this shit barely works. Get a new computer and shut the hell up already...

Re:How hard can it be?

Glad you caught that, I thought I might have been a little too subtle.

And with 9 shut down options to boot...

So which of those 9 shut-down options can we eliminate now? Probably all but the one that goes "shut the hell off"?

Re:And with 9 shut down options to boot...

They have to keep it, as it's an important usability option. Now your computer can act like you first thing in the morning. HAL, here we come!

S3 is not hibernate/deep sleep.

S3 is plain old suspend/sleep. hibernate/deep sleep implies suspend to disk and total power down, and is S4. And the word S3-induced makes no sense, S3 is a state entered into, not an active thing.

Re:S3 is not hibernate/deep sleep.

You're absolutely right, they should put those in the shutdown menu as well.

Seriously, KDE can get it right, Mac OS X can get it right. What's so wrong with: Sleep, Restart, Shutdown (, Logout)

Re:Press SHIFT to se all commands!

You can also press Alt, Super and Meta to see even more options such as doze, slumber, nap, snooze, relax, take a rest, nod off, or crash. I'm not sure what they do though.

Re:S3 is not hibernate/deep sleep.

I was thinking about that, and it's actually pretty surprising how well some systems sleep. Mac OS X can sleep through anything short of a disk burn. I have seen very rare cases where vendor specific hardware didn't wake up properly, but that's probably a vendor driver issue. The OS seems to have its act together.

The new intel mac laptops now support hibernate instead of sleep. There is no longer a backup battery in the mac laptops. When you sleep them, they appear to go to sleep instantly, but they are not asleep yet. Display is off, sleep light is on (solid), but it is now paging memory off to disk, and will take my 2gb mbp about 25 seconds to do it. Then you hear the HD park and the sleep light begins pulsing. I try not to stuff it in the bag or jolt it around until it actually parks the HD.

This means you can pull the battery even, and power it back up later and instead of the usual 4 second wakeup time, you get about 20 seconds of watching a washed out image of the last screen, with a dotted progress bar. (looks a bit like a firmware update in progress) When the dots get to the right it's awake again. It has done this from a complete power-down and memory clear. Impressive. I have not noticed anything that fails to wake up properly even from this mode.

Another nice perk is that if you sleep it, and it loses power, (battery is removed by accident, someone kicks out the power cord etc) it simply appears to have shut off. (no sleep light) Then when you try to turn it back on, it just wakes from hibernation with the usual washed out screen and 20 second progresssbar instead of the quick wakeup.

I don't think the mac pro (the desktop) supports hibernate though, but it couldn't be that hard for them to add support for?

Re:S3 is not hibernate/deep sleep.

The difference in the apple product model shows through here. Power management problems like those described by the story submitter (love that random complaints can be slashdot front page material) are related to bios in use, drivers in use... apple folk obviously deliver an OS to a limited set of hardware, drivers, bioses (did I pluralize that properly?). Windows tries to be all things to all people. breadth vs. depth, etc...

When XP came out many (many many many) systems could not boot in ACPI mode. Many systems had a bios that would report as supporting ACPI and then fall over in an unexpected way... what resolved this was.... time in market. Once it became important to boot XP it became important to pay attention to the ACPI spec. The XP installer actually has a backdoor built in for those dark days of 2001... you can bang on "f7" when you boot into textmode setup (the media-boot phase) and setup will ignore ACPI support.

Vista no longer supports non-acpi machines. Vista also tries to do more with power management and if you have current-ish system from a major OEM (dell, gateway, sony, toshiba, hp, etc) they've already posted BIOS updates to make things go in the brave new world. Partnering with the big guys is where MS can recover some depth in the hardware space.

Vista now provides a new hybrid sleep mode, combining standby with hibernation. The sleep option will write out a hibernate file so that if the machine takes a nap & runs out of juice (laptop scenario here) you can plug the box in and resume without losing your context. I'm typing on a Dell xps m170 right now -- it works well.

Re:S3 is not hibernate/deep sleep.

You can tell it to do something I like even better. It writes everything to disk as you described, but sits in suspended mode instead of hibernate. If your battery goes dead or you yank the battery it will resume from the disk copy, otherwise it pops up instantly just like the Powerbooks. Best of both worlds.

Re:S3 is not hibernate/deep sleep.

Here you are:

http://www.almaer.com/blog/archives/001182.html [almaer.com]

Pop open a terminal, type in

sudo pmset -a hibernatemode 0

and your MB will go back to suspend only. Replace the 0 with a 1 and you're in the (default) hibernate only mode. Use a 3 and the MBP will do as I described, suspending but also writing everything to disk so it can resume if it loses power.

Re:S3 is not hibernate/deep sleep.

But he's criticising Windows Vista, not Windows 2000. Have you seen Vista's shutdown menu? Have you read the article on Slashdot about how much time and effort went into making it?

bummer

Each new release, each patch, each service pack I keep waiting for the perfect, all-right-I'll-settle-for-well-behaved advanced power control. I find this unsettling Vista may not deliver. One "feature" I always treasure in Windows systems is its "better" support for power control.

At least Windows with its more cozy relationship with chip and BIOS industry supposedly offers ACPI for fast "sleep" and "rewake" functionality. In fact that was my trick way to get ACPI for linux when it was really important by running a vmware install of linux within a well behaved windows (not always as well behaved as I'd have wished, but better than the problematic ACPI linux support).

And now, out of the gates (sic) Vista may not deliver? That's going to leave a mark. I'd considered getting a machine for educational purposes (since I do support for everyone I know), but I'd considered waiting for some of the initial bugs to get ironed out. I just didn't expect this big of an initial speedbump. Guess there's not much to do but wait for Microsoft to get it right, or close to right.

Also, I thought I'd read they were offering super-sized power control a la scheduled up and down times, etc. More vaporware?

I'm still amazed they get to skate on this kind of stuff.

Re:bummer

Does power management work properly on Apple computers? If so, they're the only ones to get it working right. On Linux you normally can get it working, but it's the kind of job that is likely to take several days with no guaranteed outcome. And 90% of the time you think it's working, you'll still find glitches over time - no sound after resume, won't suspend if USB devices are plugged in, won't suspend if 3d acceleration is enabled, crashes every 10th suspend or so, appears to suspend to RAM just fine but battery drain is far more than it should be... I've concluded that power management is just insanely tricky. APM/ACPI must be inconsistently implemented on every device, otherwise it could never work as poorly as it does.

I should never have to reboot my laptop. I should be able to pop it into my docking station, resume from hibernation, and have it come up working properly including my desktop monitor and all the other peripherals hooked to the docking station. And the reverse should be true when I leave at night. I've never seen it happen.

Screw Ups

So someone fucks it up and it's irrevocably broken? I've used both sleep and hibernate functions on my laptop since Vista was beta 1 and both have worked beautifully. Both features require decent support from the hardware, not just "signed drivers."

Re:Screw Ups

Same here. Both my notebook and desktop work without any problems with sleep and hibernation under vista.
Sleep did not work on either of them under winxp.
This sound like unfounded ms bashing by someone who got frustated.

Ghee, I musta been sleeping...

...I didn't know Vista was out yet. Thought it was still in the debuggng stage...

Or maybe I'm still sleeping and this is a dream. Vista released with major operational flaws. Now that's a Linux promotion!

"no buggy software"

I'd like to know where this completely bug free software comes from. The last completely bug-free software I saw was Hello World.

"power" the future

It doesn't look to me like there was no pun intended...

Blame ACPI, not Vista

I'm vociferously anti-MS; but in this case, I believe they deserve a small pardon. Go read the ACPI specifications [acpi.info] sometime. You will cry and beg for mercy. ACPI is horrible. Considering the small number of requirements the real world has for such an interface, the specification is vast beyond imagining. Linux has also had long standing problems producing a proper ACPI layer, for this very reason: ACPI is a pig.

Now it is worth noting that MS themselves contributed to the development of this specification. The cynical side of me believes that confounding the competition by way of impenetrable specifications is simply Microsoft's modis operandi. Look at Microsoft's OpenXML specification for example: while in theory it meets the European requirement for documenting file formats and protocols, in practice it's ~6,000 pages will certainly confound all but the most determined attempts at interoperability. But here's the rub: Microsoft has to eat their own dog food, and they are suffering the consequences. Microsoft's operating system and applications are becoming so piggish that even Microsoft can't manage them.

fud ahead

I've had fewer problems with my laptop since installing vista than I ever had with linux.

Pretty much everything worked 'out-the-box' -- including video (although I ultimately had to go download the vista drivers from ATI to get any kind of acceleration), sound, even suspend/sleep (although, microsoft renaming hibernate to sleep confused me at first).

There are plenty of places where microsoft seems to suck across the board .. but vista sleeping and waking up works just fine.

BTW - this sleeping is a feature that I never did get 100% working properly in linux -- and what I WAS able to get working right required I bounce around a few websites ultimatly doing my own research ... whereas it seems to work now in vista just fine?

Huh?

This feature works just great here, making it quite impossible it's due to Vista (unless my Vista is magic), but rather due to hardware drivers after all.

Uneasy lies the head...

Canst thou, O partial sleep, give thy repose
To the wet sea-boy in an hour so rude,
And in the calmest and most stillest night,
With all appliances and means to boot,
Deny it to a king? Then happy low, lie down!
Uneasy lies the head that wears the crown.

-Henry IV. Part II.

Not exactly great with other OSes

S3 (Suspend) doesn't exactly work wonderfully under other operating systems either. It's highly dependant on the motherboard chipset being used, and all attached hardware.

I would be quicker to condem Microsoft if Linux (or FreeBSD preferably) could properly suspend and resume ANY of my systems properly. Unfortunately, that doesn't seem to be the case.

FreeBSD-6.2 was the closest I got... If I pull out my videocard and use the onboard, it actually resumes successfully.

Though the onboard video (Savag) really blows, and I haven't yet found any version of X.org that doesn't regularly crash when using that particular driver.

And both the onboard nic, and my SBlive card stop working, and I have to manually reload the kernel module every time I resume...

And with all of those addeniums, that's the closest I've ever gotten to getting Suspend to work (and being forced to use the onboard video is a complete show-stopper). In fact, the latest snapshot of 7.0 was actually a downgrade, and wouldn't resume from S3 at all.

So the problem can't lie entirely with Microsoft (though they are partly to blame for the extremely lax and often Windows-centric ACPI practices). Hardware manufacturers bare a great deal of the responsibility for making their ACPI implimentations buggy as all hell to begin with... So much so that even Microsoft apparently can't even work-around it.

No problems here at all.

We currently have 4 systems running Vista RTM and a not one of them has any problem waking up from hibernate. They are a mix of P4, AMD XP, and Athlons.

We had Vista RC1 & 2 on other systems, both desktops & laptops, and they behaved perfectly as well.

They all respond perfectly to Wake-On-LAN too. I know this because our tape backup system sends WOL packets to the systems to do the backups.

"Did anyone ever try this even once?"

The bugs that always amaze me are the ones that seemingly would have been caught if anyone had ever actually tried the feature even once.

The only way I can account for something like this is that perhaps when a bug exhibits "protean symptoms" (fails in a different way every time), one could imagine in a completely bureaucratic, micromanaged corporate environment, instead of being registered as "this always fails," it could be registered as two hundred completely different bug descriptions, each specific description having been recorded only once and therefore judged by management to be unimportant.

"Fails with blue screen of death reading 0687FF13 618AC003 ..."

being regarded as a "different" bug from

"Fails with blue screen of death reading 31469B21 96CB2022 ..."

And before people start saying "blame the hardware," it's Microsoft's job to make sure that Vista does work on every PC certified for it. The days when DOS said "Toshiba DOS" or "PC-DOS" or "NEC DOS" are long gone. The name on the product is Microsoft WIndows and it's Microsoft's responsibility to see that it works.

It's Microsoft's choice whether to do this by making their code robust, or jawboning vendors at WinHEC, or pressuring vendors.

Pun...

So much for an operating system to "power" the future! (No pun intended!)

The pun was clearly intended, otherwise there would not have been quotation marks around 'power'.

Why can't we all just be honest about our use of puns? Puns are not always bad. There's no need to be ashamed of them.

How many times do you test before calling it truth

I just opened my laptop and turned it on, and it resumed from a hibernate just fine (running Vista Business release version). No blue screen, no network problems, it put me right back where I was before with a perfectly functional session. I hate Windows as much as every other Unix geek, but it sounds to me like this is a classic case of "not enough research" ... or if you prefer, "fud".

And this is why...

...we wait for Vista SP1 before making the jump.

Also, because DX10 cards (and titles) will be ubiquitous by then.

"He's resting."

"No, it's not."

Why do I have this urge to post the entire Monte Python "Dead Parrot" sketch? [mtholyoke.edu]

My Experience is Completely the Opposite

Vista is the only OS I've used that has ever been able to wake up from sleep and hibernate properly.
The OP makes it sound like their experience applies to everyone, so I have to call FUD on this.

At any rate, I have zero problems with these features, using Vista Home Ultimate 64 bit.