Skip to contentSkip to conversation

PM: suspend/resume debugging should depend on SOFTWARE_SUSPEND

PATCH NAK Chuck Ebbert 17 participants 46 messages Nov 19, 2006 – Dec 4, 2006

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 PM: suspend/resume debugging should depend on SOFTWARE_SUSPEND

When doing 'make oldconfig' we should ask about suspend/resume debug features when SOFTWARE_SUSPEND is not enabled.

kernel/power/Kconfig +2 -2
@@ -38,7 +38,7 @@ config PM_DEBUG
 
 config DISABLE_CONSOLE_SUSPEND
 	bool "Keep console(s) enabled during suspend/resume (DANGEROUS)"
-	depends on PM && PM_DEBUG
+	depends on PM_DEBUG && SOFTWARE_SUSPEND
 	default n
 	---help---
 	This option turns off the console suspend mechanism that prevents
@@ -49,7 +49,7 @@ config DISABLE_CONSOLE_SUSPEND
 
 config PM_TRACE
 	bool "Suspend/resume event tracing"
-	depends on PM && PM_DEBUG && X86_32 && EXPERIMENTAL
+	depends on PM_DEBUG && SOFTWARE_SUSPEND && X86_32 && EXPERIMENTAL
 	default n
 	---help---
 	This enables some cheesy code to save the last PM event point in the
Signature · 1 line
Chuck
#1 lore raw

Hi!

Quoted 2 lines

When doing 'make oldconfig' we should ask about suspend/resume debug features when SOFTWARE_SUSPEND is not enabled.

These are suspend-to-ram debugging features, mostly, so no, they should not depend on software suspend.

NAK.

Quoted 22 lines
kernel/power/Kconfig +2 -2
@@ -38,7 +38,7 @@ config PM_DEBUG
 
 config DISABLE_CONSOLE_SUSPEND
 	bool "Keep console(s) enabled during suspend/resume (DANGEROUS)"
-	depends on PM && PM_DEBUG
+	depends on PM_DEBUG && SOFTWARE_SUSPEND
 	default n
 	---help---
 	This option turns off the console suspend mechanism that prevents
@@ -49,7 +49,7 @@ config DISABLE_CONSOLE_SUSPEND
 
 config PM_TRACE
 	bool "Suspend/resume event tracing"
-	depends on PM && PM_DEBUG && X86_32 && EXPERIMENTAL
+	depends on PM_DEBUG && SOFTWARE_SUSPEND && X86_32 && EXPERIMENTAL
 	default n
 	---help---
 	This enables some cheesy code to save the last PM event point in the
Signature · 2 lines
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
#2 lore raw
In reply to Chuck Ebbert
Quoted 3 lines · Chuck Ebbert

When doing 'make oldconfig' we should ask about suspend/resume debug features when SOFTWARE_SUSPEND is not enabled.

That's wrong.

I never use SOFTWARE_SUSPEND, and I think the whole concept is totally broken.

Sane people use suspend-to-ram, and that's when you need the suspend and resume debugging.

Software-suspend is silly. I want my machine back in three seconds, not waiting for minutes..

Linus

#3 lore raw
Quoted 8 lines · Linus Torvalds
Quoted 3 lines · Chuck Ebbert

When doing 'make oldconfig' we should ask about suspend/resume debug features when SOFTWARE_SUSPEND is not enabled.

That's wrong.

Quoted 8 lines

I never use SOFTWARE_SUSPEND, and I think the whole concept is totally broken.

Sane people use suspend-to-ram, and that's when you need the suspend and resume debugging.

Software-suspend is silly. I want my machine back in three seconds, not waiting for minutes..

In fact that's up to 30 seconds on a modern box, usually less than that.

And suspend-to-ram doesn't work on quilte a lot of boxes right now. Also, you can use the software suspend on boxes that don't support the suspend-to-ram at all.

Greetings,
Rafael

Signature · 2 lines
You never change things by fighting the existing reality.
		R. Buckminster Fuller
#4 lore raw
Quoted 2 lines · Rafael J. Wysocki

In fact that's up to 30 seconds on a modern box, usually less than that.

Right. If the machine boots quickly, it's fast. Of course, if the machine boots quickly, you might as well often just shut down and reboot.

Quoted 3 lines

And suspend-to-ram doesn't work on quilte a lot of boxes right now. Also, you can use the software suspend on boxes that don't support the suspend-to-ram at all.

One large reason STR often doesn't work is that people don't even test it, because people point to the suspend-to-disk instead. suspend-to-disk is the problem, not the solution.

I've been working at making the machines I have able to STR, and almost always it's a driver that is buggy. Thank God for the suspend/resume debugging - the thing that Chuck tried to disable. That's often the _only_ way to debug these things, and it's actually pretty powerful (but time-consuming - having to insert TRACE_RESUME() markers into the device driver that doesn't resume and recompile and reboot).

Anyway, the way to debug this for people who are interested (have a machine that doesn't boot) is:

- enable PM_DEBUG, and PM_TRACE

- use a script like this:

#!/bin/sh
sync
echo 1 > /sys/power/pm_trace
echo mem > /sys/power/state

to suspend

- if it doesn't come back up (which is usually the problem), reboot by holding the power button down, and look at the dmesg output for things like

	Magic number: 4:156:725
	hash matches drivers/base/power/resume.c:28
	hash matches device 0000:01:00.0

which means that the last trace event was just before trying to resume device 0000:01:00.0. Then figure out what driver is controlling that device (lspci and /sys/devices/pci* is your friend), and see if you can fix it, disable it, or trace into its resume function.

For example, the above happens to be the VGA device on my EVO, which I used to run with "radeonfb" (it's an ATI Radeon mobility). It turns out that "radeonfb" simply cannot resume that device - it tries to set the PLL's, and it just _hangs_. Using the regular VGA console and letting X resume it instead works fine.

The point being that PM_TRACE is wonderful, and it's wonderful exactly for NOT using suspend-to-disk. The other point being that people have gotten lazy, and accept half a minute (minimum - usually it's longer) boot times, when STR is a lot more pleasant, but it does require some detective work when it doesn't boot.

I wish more people tried STR, instead of having the STD people tell them not to!

Linus

#5 lore raw
Quoted 7 lines · Linus Torvalds
Quoted 2 lines · Rafael J. Wysocki

In fact that's up to 30 seconds on a modern box, usually less than that.

Right. If the machine boots quickly, it's fast. Of course, if the machine boots quickly, you might as well often just shut down and reboot.

Yes, if the only thing you want to run is the kernel. The applications aren't going to start so quickly, you know. ;-)

Quoted 5 lines
Quoted 3 lines

And suspend-to-ram doesn't work on quilte a lot of boxes right now. Also, you can use the software suspend on boxes that don't support the suspend-to-ram at all.

One large reason STR often doesn't work is that people don't even test it,

Well, the majority of bugzilla reports I'm tracking are STR-related, so people _do_ test it, really.

Quoted 1 line

because people point to the suspend-to-disk instead.

Who they?

Quoted 1 line

suspend-to-disk is the problem, not the solution.

If given machine doesn't support the STR or has a broken BIOS (which means pretty much the same), the suspend-to-disk is the _only_ way in which you can suspend it.

I guess you're referring to using the suspend-to-disk instead of the STR which wouldn't be very reasonable indeed, but still I don't know why you consider it as a problem.

Quoted 11 lines

I've been working at making the machines I have able to STR, and almost always it's a driver that is buggy. Thank God for the suspend/resume debugging - the thing that Chuck tried to disable. That's often the _only_ way to debug these things, and it's actually pretty powerful (but time-consuming - having to insert TRACE_RESUME() markers into the device driver that doesn't resume and recompile and reboot).

Anyway, the way to debug this for people who are interested (have a machine that doesn't boot) is:

- enable PM_DEBUG, and PM_TRACE

This only works on i386, no?

Quoted 28 lines

- use a script like this:

#!/bin/sh
sync
echo 1 > /sys/power/pm_trace
echo mem > /sys/power/state

to suspend

- if it doesn't come back up (which is usually the problem), reboot by holding the power button down, and look at the dmesg output for things like

	Magic number: 4:156:725
	hash matches drivers/base/power/resume.c:28
	hash matches device 0000:01:00.0

which means that the last trace event was just before trying to resume device 0000:01:00.0. Then figure out what driver is controlling that device (lspci and /sys/devices/pci* is your friend), and see if you can fix it, disable it, or trace into its resume function.

For example, the above happens to be the VGA device on my EVO, which I used to run with "radeonfb" (it's an ATI Radeon mobility). It turns out that "radeonfb" simply cannot resume that device - it tries to set the PLL's, and it just _hangs_. Using the regular VGA console and letting X resume it instead works fine.

Quoted 8 lines

The point being that PM_TRACE is wonderful, and it's wonderful exactly for NOT using suspend-to-disk. The other point being that people have gotten lazy, and accept half a minute (minimum - usually it's longer) boot times, when STR is a lot more pleasant, but it does require some detective work when it doesn't boot.

I wish more people tried STR, instead of having the STD people tell them not to!

I don't know of anyone who's doing that.

Yes, we often ask people to try the STD when they report a problem with the STR to get one more data point, but that doesn't mean we tell them not to try the STR any more. Bug reports regarding the STR are accepted and welcome.

Greetings,
Rafael

Signature · 2 lines
You never change things by fighting the existing reality.
		R. Buckminster Fuller
#6 lore raw
Quoted 4 lines · Rafael J. Wysocki
Quoted 2 lines

because people point to the suspend-to-disk instead.

Who they?

Like you _just_ did.

Quoted 3 lines
Quoted 1 line

- enable PM_DEBUG, and PM_TRACE

This only works on i386, no?

Right now the trivial functions are only available on i386, yes. The concept works anywhere that has a CMOS chip, so if somebody were to spend a few minutes testing it on x86-64 and others, it would work elsewhere too..

Quoted 1 line

I don't know of anyone who's doing that.

I know. I'm probably the only one. Frustrating.

Linus

#7 lore raw
Quoted 8 lines · Linus Torvalds
Quoted 4 lines · Rafael J. Wysocki
Quoted 2 lines

because people point to the suspend-to-disk instead.

Who they?

Like you _just_ did.

No, I didn't. It was referred to in the message that started this thread.

I don't remeber telling _anyone_ not to use the STR and use the STD instead.

Quoted 8 lines
Quoted 3 lines
Quoted 1 line

- enable PM_DEBUG, and PM_TRACE

This only works on i386, no?

Right now the trivial functions are only available on i386, yes. The concept works anywhere that has a CMOS chip, so if somebody were to spend a few minutes testing it on x86-64 and others, it would work elsewhere too..

I can do that if someone gives me the code.

Quoted 3 lines
Quoted 1 line

I don't know of anyone who's doing that.

I know. I'm probably the only one. Frustrating.

Well, it isn't that well documented ...

Greetings,
Rafael

Signature · 2 lines
You never change things by fighting the existing reality.
		R. Buckminster Fuller
#8 lore raw
Quoted 5 lines · Rafael J. Wysocki
Quoted 3 lines

concept works anywhere that has a CMOS chip, so if somebody were to spend a few minutes testing it on x86-64 and others, it would work elsewhere too..

I can do that if someone gives me the code.

Well, I actually _think_ it works almost as-is on x86-64 too, but the magic is all in that small inline asm thing in <linux/resume-trace.h>.

The ".long" in that inline asm probably needs to be a ".quad" - see gow "generate_resume_trace()" uses it.

Also the "show_file_hash()" right now has a "tracedata += 6", and it should probably be "tracedata += 2 + sizeof(void *)" instead.

IOW, it really should be very easy to get to work on x86-64, I just haven't had the energy or inclination, since the only devices I've personally used it on have been regular 32-bit ones..

Linus

#9 lore raw
In reply to Rafael J. Wysocki
Quoted 11 lines · Rafael J. Wysocki
Quoted 7 lines · Linus Torvalds
Quoted 2 lines · Rafael J. Wysocki

In fact that's up to 30 seconds on a modern box, usually less than that.

Right. If the machine boots quickly, it's fast. Of course, if the machine boots quickly, you might as well often just shut down and reboot.

Yes, if the only thing you want to run is the kernel. The applications aren't going to start so quickly, you know. ;-)

Can I say a word in defence of STD? It's true that now that I have STR working(*) for the first time on my old vaio I use it much less, but it has been my salvation in the last four years, and for this I have to thanks Nigel, Pavel, Rafael and all the people involved. And still now it is very useful. I can STD a session with tens of application opened, and come back after changing batteries in less than a minute _doing other things_, and not opening applications and files all over the place. So yes, I think it's useful.

And when suspend-to-both will work radiply and safely, that will be great (and a point over The Other SO...)

Romano

(*) if only for this, all the troubles I had upgrading to Ubuntu Edgy has been worthwhile.

#10 lore raw
In reply to Linus Torvalds

One reason why I've generally avoided using suspend-to-ram is that after my laptop (IBM Thinkpad T60p) comes out of suspend-to-ram, it is consuming 30W of power --- as opposed to before I hibernated, when my laptop was consuming only 17W or 18W of power. (And even without doing things like unloading all of my USB modules, normally my laptop will consume about 24W after a fresh reboot --- which makes the 30W power consumption after a suspend-to-ram especially troubling,)

If I unload all of my USB modules, and shutdown the parallel port, the wired ethernet port, etc., I get power savinges down to 17W --- and once I was able to push it all the way down to 15W, although in practice it's much more common that I can get the power consumption down to 17W or 18W. Unfortunately, after the laptop wakss up from a suspend-to-ram, the laptop is apparently powering up all of the devices in high power mode while I can get some of the power savings back by manually loading and then unlodaing a whole bunch of device drivers, in practice that only gets me from 30W to 24W or so. This is probably much more of a hardware bug than an OS bug, but because of this fact, if I'm going to be running the laptop for any amount of time after resume, I'm better off using suspend-to-disk.

If someone has a suggestion for how I can save the power state of all of the various components in my laptop so that the laptop cna be brought back to the 18W state after a suspend-to-ram, I'm all ears....

- Ted

#11 lore raw
Quoted 3 lines · Theodore Tso

If someone has a suggestion for how I can save the power state of all of the various components in my laptop so that the laptop cna be brought back to the 18W state after a suspend-to-ram, I'm all ears....

A good start might be to compare the PCI configuration registers before and after suspend. However, I suspect it's more complicated than that. Are you using the closed ATI drivers? If so, it's possible that they do something at X startup that they're not doing on resume. A good comparison might be to see if the power consumption is dramatically different over suspend/resume if you only boot to text mode - that is, never start X at all.

Signature · 1 line
Matthew Garrett | [email protected]
#12 lore raw
In reply to Linus Torvalds

Hi!

Quoted 7 lines
Quoted 3 lines

And suspend-to-ram doesn't work on quilte a lot of boxes right now. Also, you can use the software suspend on boxes that don't support the suspend-to-ram at all.

One large reason STR often doesn't work is that people don't even test it, because people point to the suspend-to-disk instead. suspend-to-disk is the problem, not the solution.

Actually, I do not think we do that any more. But if s2ram is broken, trying s2disk is indeed useful, because many driver problems are common and s2disk is slightly easier to debug.

Pavel

Signature · 1 line
Thanks for all the (sleeping) penguins.
#13 lore raw
In reply to Linus Torvalds
Quoted 6 lines · Linus Torvalds
Quoted 3 lines · Chuck Ebbert

When doing 'make oldconfig' we should ask about suspend/resume debug features when SOFTWARE_SUSPEND is not enabled.

Quoted 7 lines

That's wrong.

I never use SOFTWARE_SUSPEND, and I think the whole concept is totally broken.

Sane people use suspend-to-ram, and that's when you need the suspend and resume debugging.

Here I am wishing I had the _opportunity_ to be sane. With my ATI X850 AGP card, I have no choices except swsusp or reboot.

-Mike

#14 lore raw
Quoted 3 lines · Mike Galbraith

Here I am wishing I had the _opportunity_ to be sane. With my ATI X850 AGP card, I have no choices except swsusp or reboot.

Try using regular VGA console, and letting X re-initialize the card. It's worked for me on several machines (Mac Mini and Compaq EVO) where the kernel cannot initialize the graphics itself.

It does mean that you have to run X when suspending, but most people do anyway..

Linus

#15 lore raw
Quoted 6 lines · Linus Torvalds
Quoted 3 lines · Mike Galbraith

Here I am wishing I had the _opportunity_ to be sane. With my ATI X850 AGP card, I have no choices except swsusp or reboot.

Quoted 6 lines

Try using regular VGA console, and letting X re-initialize the card. It's worked for me on several machines (Mac Mini and Compaq EVO) where the kernel cannot initialize the graphics itself.

It does mean that you have to run X when suspending, but most people do anyway..

Thanks for the tip, but it didn't work. It suspended instantly, and got my hopes up (manually, SuSE says "not supported, go away"), but resume still left me with an utterly dead box (minus flashing crud on display).

-Mike

#16 lore raw
Quoted 4 lines · Mike Galbraith

Thanks for the tip, but it didn't work. It suspended instantly, and got my hopes up (manually, SuSE says "not supported, go away"), but resume still left me with an utterly dead box (minus flashing crud on display).

Right. That's why we have PM_DEBUG and PM_TRACE, and why I sent out the small email about how to use them.

The utterly dead box is the common case when some driver doesn't actually resume properly ;(

Linus

#17 lore raw
In reply to Mike Galbraith
Quoted 9 lines · Mike Galbraith
Quoted 8 lines · Linus Torvalds
Quoted 3 lines · Chuck Ebbert

When doing 'make oldconfig' we should ask about suspend/resume debug features when SOFTWARE_SUSPEND is not enabled.

That's wrong.

Quoted 8 lines
Quoted 5 lines

I never use SOFTWARE_SUSPEND, and I think the whole concept is totally broken.

Sane people use suspend-to-ram, and that's when you need the suspend and resume debugging.

Here I am wishing I had the _opportunity_ to be sane. With my ATI X850 AGP card, I have no choices except swsusp or reboot.

Have you tried s2ram (http://en.opensuse.org/S2ram)?

Rafael

Signature · 2 lines
You never change things by fighting the existing reality.
		R. Buckminster Fuller
#18 lore raw
Quoted 13 lines · Rafael J. Wysocki
Quoted 12 lines · Mike Galbraith
Quoted 11 lines · Linus Torvalds
Quoted 3 lines · Chuck Ebbert

When doing 'make oldconfig' we should ask about suspend/resume debug features when SOFTWARE_SUSPEND is not enabled.

That's wrong.

I never use SOFTWARE_SUSPEND, and I think the whole concept is totally broken.

Quoted 7 lines
Quoted 5 lines
Quoted 2 lines

Sane people use suspend-to-ram, and that's when you need the suspend and resume debugging.

Here I am wishing I had the _opportunity_ to be sane. With my ATI X850 AGP card, I have no choices except swsusp or reboot.

Have you tried s2ram (http://en.opensuse.org/S2ram)?

Cool. That shows potential. On an 2.6.19-rc6-rt4 kernel, it looked like it _might_ have eventually gotten past boot. At one line of kernel output every ~10 seconds though, I gave up. Virgin 2.6.19-rc6 went panic with a black screen... have options, will tinker.

(i _was_ quite content with swsusp, but now i want it all;)

Thanks,

-Mike

#19 lore raw

Hi!

Quoted 6 lines
Quoted 6 lines
Quoted 5 lines
Quoted 2 lines

Sane people use suspend-to-ram, and that's when you need the suspend and resume debugging.

Here I am wishing I had the _opportunity_ to be sane. With my ATI X850 AGP card, I have no choices except swsusp or reboot.

Quoted 6 lines
Quoted 1 line

Have you tried s2ram (http://en.opensuse.org/S2ram)?

Cool. That shows potential. On an 2.6.19-rc6-rt4 kernel, it looked like it _might_ have eventually gotten past boot. At one line of kernel output every ~10 seconds though, I gave up. Virgin 2.6.19-rc6 went panic with a black screen... have options, will tinker.

Try that -rt kernel, but turn off dyntick/hires timers. Also try hitting keyboard while resuming.

Signature · 1 line
Thanks for all the (sleeping) penguins.
#20 lore raw
Quoted 10 lines · Pavel Machek

Hi!

Quoted 8 lines
Quoted 7 lines
Quoted 5 lines
Quoted 2 lines

Sane people use suspend-to-ram, and that's when you need the suspend and resume debugging.

Here I am wishing I had the _opportunity_ to be sane. With my ATI X850 AGP card, I have no choices except swsusp or reboot.

Have you tried s2ram (http://en.opensuse.org/S2ram)?

Quoted 7 lines
Quoted 4 lines

Cool. That shows potential. On an 2.6.19-rc6-rt4 kernel, it looked like it _might_ have eventually gotten past boot. At one line of kernel output every ~10 seconds though, I gave up. Virgin 2.6.19-rc6 went panic with a black screen... have options, will tinker.

Try that -rt kernel, but turn off dyntick/hires timers. Also try hitting keyboard while resuming.

Hmm, I'll try that. Interesting (read odd from my perspective) that the rt kernel gets much further. This is pretty hard to look at.

#21 lore raw
Quoted 16 lines · Mike Galbraith
Quoted 15 lines · Pavel Machek

Hi!

Quoted 12 lines
Quoted 7 lines
Quoted 5 lines
Quoted 2 lines

Sane people use suspend-to-ram, and that's when you need the suspend and resume debugging.

Here I am wishing I had the _opportunity_ to be sane. With my ATI X850 AGP card, I have no choices except swsusp or reboot.

Have you tried s2ram (http://en.opensuse.org/S2ram)?

Cool. That shows potential. On an 2.6.19-rc6-rt4 kernel, it looked like it _might_ have eventually gotten past boot. At one line of kernel output every ~10 seconds though, I gave up. Virgin 2.6.19-rc6 went panic with a black screen... have options, will tinker.

Quoted 5 lines
Quoted 2 lines

Try that -rt kernel, but turn off dyntick/hires timers. Also try hitting keyboard while resuming.

Hmm, I'll try that. Interesting (read odd from my perspective) that the rt kernel gets much further. This is pretty hard to look at.

Is your system an i386?

#22 lore raw
Quoted 10 lines · Rafael J. Wysocki
Quoted 7 lines · Mike Galbraith
Quoted 3 lines · Pavel Machek

Try that -rt kernel, but turn off dyntick/hires timers. Also try hitting keyboard while resuming.

Hmm, I'll try that. Interesting (read odd from my perspective) that the rt kernel gets much further. This is pretty hard to look at.

Is your system an i386?

Yeah, single P4/HT supermarket box.

I tried the dynticks/hires-timers/kbd suggestion, no difference. It still boots in medicated snail mode, and emits a stream of IRQ9: nobody cared messages (fasteoi acpi, irqpoll = nogo) while doing so.

-Mike

#23 lore raw
Quoted 4 lines · Mike Galbraith

I tried the dynticks/hires-timers/kbd suggestion, no difference. It still boots in medicated snail mode, and emits a stream of IRQ9: nobody cared messages (fasteoi acpi, irqpoll = nogo) while doing so.

"medicated snail mode". Lol.

The "IRQ9: nobody cared" things are not unheard of. The Mac Mini had the same issue, and in that case it was because the firmware on that machine was just broken, and didn't re-enable ACPI mode correctly on resume. We worked around it by forcing a re-enable by hand, so that _exact_ issue isn't going to be your problem (or it would work since 2.6.18), but it's not unlikely that there is some other SCI setup that could have been missed.

One thing that is often worth testing is to try with APIC support if you don't have it already, or if you do have it, try _without_ APIC support. The firmware generally is only tested against MS operating systems, so it's only ever been tested for the particular irq setup that Windows tends to use, and as a result sometimes things work better in one more or another.

Based on the "fasteoi", you're obviously right now using the APIC, and that's _usually_ the mode that works better. But just in case, try booting with "noapic".

Also, can you send out the boot log with APIC information ("apic=debug"). Of course, don't disable the apic for that case, or it won't be very useful ;)

Linus

#24 lore raw
Quoted 6 lines · Linus Torvalds

One thing that is often worth testing is to try with APIC support if you don't have it already, or if you do have it, try _without_ APIC support. The firmware generally is only tested against MS operating systems, so it's only ever been tested for the particular irq setup that Windows tends to use, and as a result sometimes things work better in one more or another.

I'll try noapic. (though I value my apic more than s2ram;)

Quoted 7 lines

Based on the "fasteoi", you're obviously right now using the APIC, and that's _usually_ the mode that works better. But just in case, try booting with "noapic".

Also, can you send out the boot log with APIC information ("apic=debug"). Of course, don't disable the apic for that case, or it won't be very useful ;)

Will do. (not tomorrow though, helping friend with home construction)

I found the driver that was causing my panic problem with vanilla, it was my Sat TV card. I haven't ever used it with Linux, and haven't used it with that other OS but a couple of times to try it out(stupid stupid purchase), so I tossed it.

Box still paniced though, apparently after restoring all drivers (last trace mark was end of function marker). I disabled trace, and enabled the "leave console active, danger danger" option in hopes of capturing the panic, and what do you know, for the first time ever, box restored all the way into X... only to panic about two seconds later. Nothing whatsoever got to my serial console. Switching to a text console before suspend didn't help either, the panic is invisible. Drat.

I'll toss rocks at it.

-Mike

#25 lore raw
In reply to Linus Torvalds
Quoted 8 lines · Linus Torvalds
Quoted 4 lines · Mike Galbraith

I tried the dynticks/hires-timers/kbd suggestion, no difference. It still boots in medicated snail mode, and emits a stream of IRQ9: nobody cared messages (fasteoi acpi, irqpoll = nogo) while doing so.

"medicated snail mode". Lol.

It turns out, that this is caused by console=ttyS0,115200n8. Without the problematic driver found by pm_trace for the old Pinacle SAT card...

02:02.0 Multimedia video controller: Brooktree Corporation Bt878 Video Capture (rev 11)
Subsystem: Pinnacle Systems Inc. PCTV Sat (DBC receiver)
Control: I/O- Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR- FastB2B-
Status: Cap+ 66MHz- UDF- FastB2B+ ParErr- DEVSEL=medium >TAbort- <TAbort- <MAbort- >SERR- <PERR-
Latency: 32 (4000ns min, 10000ns max)
Interrupt: pin A routed to IRQ 17
Region 0: Memory at f2100000 (32-bit, prefetchable) [size=4K]
Capabilities: [44] Vital Product Data
Capabilities: [4c] Power Management version 2
Flags: PMEClk- DSI+ D1- D2- AuxCurrent=0mA PME(D0-,D1-,D2-,D3hot-,D3cold-)
Status: D0 PME-Enable- DSel=0 DScale=0 PME-

went boom here -->02:02.1 Multimedia controller: Brooktree Corporation Bt878 Audio Capture (rev 11)
Subsystem: Pinnacle Systems Inc. PCTV Sat (DBC receiver)
Control: I/O- Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR- FastB2B-
Status: Cap+ 66MHz- UDF- FastB2B+ ParErr- DEVSEL=medium >TAbort- <TAbort- <MAbort- >SERR- <PERR-
Latency: 32 (1000ns min, 63750ns max)
Interrupt: pin A routed to IRQ 17
Region 0: Memory at f2101000 (32-bit, prefetchable) [size=4K]
Capabilities: [44] Vital Product Data
Capabilities: [4c] Power Management version 2
Flags: PMEClk- DSI+ D1- D2- AuxCurrent=0mA PME(D0-,D1-,D2-,D3hot-,D3cold-)
Status: D0 PME-Enable- DSel=0 DScale=0 PME-

02:03.0 Multimedia controller: Philips Semiconductors SAA7134 Video Broadcast Decoder (rev 01)
Subsystem: Creatix Polymedia GmbH Medion 7134
Control: I/O- Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR- FastB2B-
Status: Cap+ 66MHz- UDF- FastB2B+ ParErr- DEVSEL=medium >TAbort- <TAbort- <MAbort- >SERR- <PERR-
Latency: 32 (2000ns min, 8000ns max)
Interrupt: pin A routed to IRQ 20
Region 0: Memory at f2002000 (32-bit, non-prefetchable) [size=1K]
Capabilities: [40] Power Management version 1
Flags: PMEClk- DSI- D1+ D2+ AuxCurrent=0mA PME(D0-,D1-,D2-,D3hot-,D3cold-)
Status: D0 PME-Enable- DSel=0 DScale=1 PME-

...and the serial console, I can now suspend to ram just fine with stock 2.6.19-rc6. As a serial port test, I set up a getty on ttyS0, and logged in via minicom prior to suspend. After resume, the serial port isn't talking. Kill -9 the shell though, and all is fine again.

Quoted 3 lines

Based on the "fasteoi", you're obviously right now using the APIC, and that's _usually_ the mode that works better. But just in case, try booting with "noapic".

Makes no difference for vanilla, works just fine either way, I don't even have to remove vga=0x314. The RT kernel has some issues in this area with s2ram, but that's a different story.

Hope you're having/had a nice weekend, and thanks for the tips!

-Mike

#26 lore raw

Hi,

Quoted 48 lines · Mike Galbraith
Quoted 8 lines · Linus Torvalds
Quoted 4 lines · Mike Galbraith

I tried the dynticks/hires-timers/kbd suggestion, no difference. It still boots in medicated snail mode, and emits a stream of IRQ9: nobody cared messages (fasteoi acpi, irqpoll = nogo) while doing so.

"medicated snail mode". Lol.

It turns out, that this is caused by console=ttyS0,115200n8. Without the problematic driver found by pm_trace for the old Pinacle SAT card...

02:02.0 Multimedia video controller: Brooktree Corporation Bt878 Video Capture (rev 11)
Subsystem: Pinnacle Systems Inc. PCTV Sat (DBC receiver)
Control: I/O- Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR- FastB2B-
Status: Cap+ 66MHz- UDF- FastB2B+ ParErr- DEVSEL=medium >TAbort- <TAbort- <MAbort- >SERR- <PERR-
Latency: 32 (4000ns min, 10000ns max)
Interrupt: pin A routed to IRQ 17
Region 0: Memory at f2100000 (32-bit, prefetchable) [size=4K]
Capabilities: [44] Vital Product Data
Capabilities: [4c] Power Management version 2
Flags: PMEClk- DSI+ D1- D2- AuxCurrent=0mA PME(D0-,D1-,D2-,D3hot-,D3cold-)
Status: D0 PME-Enable- DSel=0 DScale=0 PME-

went boom here -->02:02.1 Multimedia controller: Brooktree Corporation Bt878 Audio Capture (rev 11)
Subsystem: Pinnacle Systems Inc. PCTV Sat (DBC receiver)
Control: I/O- Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR- FastB2B-
Status: Cap+ 66MHz- UDF- FastB2B+ ParErr- DEVSEL=medium >TAbort- <TAbort- <MAbort- >SERR- <PERR-
Latency: 32 (1000ns min, 63750ns max)
Interrupt: pin A routed to IRQ 17
Region 0: Memory at f2101000 (32-bit, prefetchable) [size=4K]
Capabilities: [44] Vital Product Data
Capabilities: [4c] Power Management version 2
Flags: PMEClk- DSI+ D1- D2- AuxCurrent=0mA PME(D0-,D1-,D2-,D3hot-,D3cold-)
Status: D0 PME-Enable- DSel=0 DScale=0 PME-

02:03.0 Multimedia controller: Philips Semiconductors SAA7134 Video Broadcast Decoder (rev 01)
Subsystem: Creatix Polymedia GmbH Medion 7134
Control: I/O- Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR- FastB2B-
Status: Cap+ 66MHz- UDF- FastB2B+ ParErr- DEVSEL=medium >TAbort- <TAbort- <MAbort- >SERR- <PERR-
Latency: 32 (2000ns min, 8000ns max)
Interrupt: pin A routed to IRQ 20
Region 0: Memory at f2002000 (32-bit, non-prefetchable) [size=1K]
Capabilities: [40] Power Management version 1
Flags: PMEClk- DSI- D1+ D2+ AuxCurrent=0mA PME(D0-,D1-,D2-,D3hot-,D3cold-)
Status: D0 PME-Enable- DSel=0 DScale=1 PME-

Quoted 4 lines

...and the serial console, I can now suspend to ram just fine with stock 2.6.19-rc6. As a serial port test, I set up a getty on ttyS0, and logged in via minicom prior to suspend. After resume, the serial port isn't talking. Kill -9 the shell though, and all is fine again.

Hm, could you please file a bugzilla report regarding the serial console for the information of its maintainer(s)?

Greetings,
Rafael

#27 lore raw
Quoted 2 lines · Rafael J. Wysocki

Hm, could you please file a bugzilla report regarding the serial console for the information of its maintainer(s)?

Yeah, I'll rummage around a bit first though.

-Mike

#28 lore raw
Quoted 6 lines · Mike Galbraith
Quoted 2 lines · Rafael J. Wysocki

Hm, could you please file a bugzilla report regarding the serial console for the information of its maintainer(s)?

Yeah, I'll rummage around a bit first though.

The serial console appears to be innocent. The suspend/resume methods for my 16550A serial port aren't even being _called_, apparently because pnp swiped ttyS0. Disabling pnp support, the port suspends and resumes just peachy. /sys has a different layout without pnp enabled. With it, there are platform/ttyS[1-3] entries, but ttyS0 is in pnp. Without pnp support, the platform/ttyS* entries are gone, as is of course pnp/ttyS0.

I guess I'll have to switch "rummage" directories.

-Mike

#29 lore raw
Quoted 11 lines · Mike Galbraith
Quoted 6 lines · Mike Galbraith
Quoted 2 lines · Rafael J. Wysocki

Hm, could you please file a bugzilla report regarding the serial console for the information of its maintainer(s)?

Yeah, I'll rummage around a bit first though.

The serial console appears to be innocent. The suspend/resume methods for my 16550A serial port aren't even being _called_, apparently because pnp swiped ttyS0.

Well, after further rummaging, the suspend/reCONFIG_PNPsume methods aren't being called because we're using 8250_pnp.c when doesn't have any :)

#30 lore raw
Quoted 3 lines · Mike Galbraith
Quoted 3 lines

The serial console appears to be innocent. The suspend/resume methods for my 16550A serial port aren't even being _called_, apparently because pnp swiped ttyS0.

(ahem, bad aim with mouse, resuming)

Well, after further rummaging, the suspend/resume methods aren't being called because we're using 8250_pnp.c when CONFIG_PNP is set, and it doesn't have any :) The below works for me, though I'm not sure it's sufficient. If it is deemed sufficient, I'll submit it to Andrew.

Add suspend/resume methods to 8250_pnp driver.

drivers/serial/8250_pnp.c renamed from drivers/serial/8250_pnp.c.org +28 -1
@@ -464,11 +464,38 @@ static void __devexit serial_pnp_remove(
 		serial8250_unregister_port(line - 1);
 }
 
+#ifdef CONFIG_PM
+static int serial_pnp_suspend(struct pnp_dev *dev, pm_message_t state)
+{
+	long line = (long)pnp_get_drvdata(dev);
+
+	if (!line)
+		return -ENODEV;
+	serial8250_suspend_port(line - 1);
+	return 0;
+}
+
+static int serial_pnp_resume(struct pnp_dev *dev)
+{
+	long line = (long)pnp_get_drvdata(dev);
+
+	if (!line)
+		return -ENODEV;
+	serial8250_resume_port(line - 1);
+	return 0;
+}
+
+#endif /* CONFIG_PM */
+
 static struct pnp_driver serial_pnp_driver = {
 	.name		= "serial",
-	.id_table	= pnp_dev_table,
 	.probe		= serial_pnp_probe,
 	.remove		= __devexit_p(serial_pnp_remove),
+#ifdef CONFIG_PM
+	.suspend	= serial_pnp_suspend,
+	.resume		= serial_pnp_resume,
+#endif
+	.id_table	= pnp_dev_table,
 };
 
 static int __init serial8250_pnp_init(void)
#31 lore raw

Hi!

Quoted 5 lines · Mike Galbraith
Quoted 3 lines
Quoted 3 lines

The serial console appears to be innocent. The suspend/resume methods for my 16550A serial port aren't even being _called_, apparently because pnp swiped ttyS0.

Quoted 8 lines

(ahem, bad aim with mouse, resuming)

Well, after further rummaging, the suspend/resume methods aren't being called because we're using 8250_pnp.c when CONFIG_PNP is set, and it doesn't have any :) The below works for me, though I'm not sure it's sufficient. If it is deemed sufficient, I'll submit it to Andrew.

Add suspend/resume methods to 8250_pnp driver.

Patch looks okay to me... care to sign-it-off and submit it to akpm?

Pavel

Quoted 47 lines
drivers/serial/8250_pnp.c renamed from drivers/serial/8250_pnp.c.org +28 -1
@@ -464,11 +464,38 @@ static void __devexit serial_pnp_remove(
 		serial8250_unregister_port(line - 1);
 }
 
+#ifdef CONFIG_PM
+static int serial_pnp_suspend(struct pnp_dev *dev, pm_message_t state)
+{
+	long line = (long)pnp_get_drvdata(dev);
+
+	if (!line)
+		return -ENODEV;
+	serial8250_suspend_port(line - 1);
+	return 0;
+}
+
+static int serial_pnp_resume(struct pnp_dev *dev)
+{
+	long line = (long)pnp_get_drvdata(dev);
+
+	if (!line)
+		return -ENODEV;
+	serial8250_resume_port(line - 1);
+	return 0;
+}
+
+#endif /* CONFIG_PM */
+
 static struct pnp_driver serial_pnp_driver = {
 	.name		= "serial",
-	.id_table	= pnp_dev_table,
 	.probe		= serial_pnp_probe,
 	.remove		= __devexit_p(serial_pnp_remove),
+#ifdef CONFIG_PM
+	.suspend	= serial_pnp_suspend,
+	.resume		= serial_pnp_resume,
+#endif
+	.id_table	= pnp_dev_table,
 };
 
 static int __init serial8250_pnp_init(void)
Signature · 1 line
Thanks for all the (sleeping) penguins.
#32 lore raw
In reply to Mike Galbraith
Quoted 17 lines · On Wed, 29 Nov 2006 11:30:31 +0100 Mike Galbraith <[email protected]> wrote:
Quoted 3 lines · Mike Galbraith
Quoted 3 lines

The serial console appears to be innocent. The suspend/resume methods for my 16550A serial port aren't even being _called_, apparently because pnp swiped ttyS0.

(ahem, bad aim with mouse, resuming)

Well, after further rummaging, the suspend/resume methods aren't being called because we're using 8250_pnp.c when CONFIG_PNP is set, and it doesn't have any :) The below works for me, though I'm not sure it's sufficient. If it is deemed sufficient, I'll submit it to Andrew.

Add suspend/resume methods to 8250_pnp driver.

drivers/serial/8250_pnp.c static void __devexit serial_pnp_remove( · line 470

Quoted 7 lines
drivers/serial/8250_pnp.c renamed from drivers/serial/8250_pnp.c.org +4 -0
@@ -464,11 +464,38 @@ static void __devexit serial_pnp_remove(
 		serial8250_unregister_port(line - 1);
 }
 
+#ifdef CONFIG_PM
+static int serial_pnp_suspend(struct pnp_dev *dev, pm_message_t state)
+{
+	long line = (long)pnp_get_drvdata(dev);

Please avoid adding long lines. (heh, I kill me)

Quoted 5 lines
drivers/serial/8250_pnp.c renamed from drivers/serial/8250_pnp.c.org +5 -0
+	if (!line)
+		return -ENODEV;
+	serial8250_suspend_port(line - 1);
+	return 0;
+}

drivers/serial/8250_pnp.c static void __devexit serial_pnp_remove( · line 486

Quoted 10 lines
drivers/serial/8250_pnp.c renamed from drivers/serial/8250_pnp.c.org +10 -0
+
+static int serial_pnp_resume(struct pnp_dev *dev)
+{
+	long line = (long)pnp_get_drvdata(dev);
+
+	if (!line)
+		return -ENODEV;
+	serial8250_resume_port(line - 1);
+	return 0;
+}

We'd usually do

#else
#define serial_pnp_suspend NULL
#define serial_pnp_resume NULL

here

drivers/serial/8250_pnp.c static void __devexit serial_pnp_remove( · line 497

Quoted 12 lines
drivers/serial/8250_pnp.c renamed from drivers/serial/8250_pnp.c.org +7 -1
+
+#endif /* CONFIG_PM */
+
 static struct pnp_driver serial_pnp_driver = {
 	.name		= "serial",
-	.id_table	= pnp_dev_table,
 	.probe		= serial_pnp_probe,
 	.remove		= __devexit_p(serial_pnp_remove),
+#ifdef CONFIG_PM
+	.suspend	= serial_pnp_suspend,
+	.resume		= serial_pnp_resume,
+#endif

and hence omit the ifdefs here.

#33 lore raw
Quoted 6 lines · Andrew Morton

Review

Quoted 4 lines
(quoted diff) +4 -0
+#ifdef CONFIG_PM
+static int serial_pnp_suspend(struct pnp_dev *dev, pm_message_t state)
+{
+	long line = (long)pnp_get_drvdata(dev);

Please avoid adding long lines. (heh, I kill me)

Ok. I also changed the place I got it from.

Quoted 8 lines

We'd usually do

#else
#define serial_pnp_suspend NULL
#define serial_pnp_resume NULL

here

Quoted 14 lines

Review

Quoted 12 lines
(quoted diff) +7 -1
+
+#endif /* CONFIG_PM */
+
 static struct pnp_driver serial_pnp_driver = {
 	.name		= "serial",
-	.id_table	= pnp_dev_table,
 	.probe		= serial_pnp_probe,
 	.remove		= __devexit_p(serial_pnp_remove),
+#ifdef CONFIG_PM
+	.suspend	= serial_pnp_suspend,
+	.resume		= serial_pnp_resume,
+#endif

and hence omit the ifdefs here.

New patch.

Add suspend/resume methods to drivers/serial/8250_pnp.c. Tested on a P4/HT 16550A box, ttyS0 login survives across suspend to ram.

drivers/serial/8250_pnp.c renamed from drivers/serial/8250_pnp.c.org +29 -2
@@ -459,16 +459,43 @@ serial_pnp_probe(struct pnp_dev *dev, co
 
 static void __devexit serial_pnp_remove(struct pnp_dev *dev)
 {
-	long line = (long)pnp_get_drvdata(dev);
+	int line = (long)pnp_get_drvdata(dev);
 	if (line)
 		serial8250_unregister_port(line - 1);
 }
 
+#ifdef CONFIG_PM
+static int serial_pnp_suspend(struct pnp_dev *dev, pm_message_t state)
+{
+	int line = (int)pnp_get_drvdata(dev);
+
+	if (!line)
+		return -ENODEV;
+	serial8250_suspend_port(line - 1);
+	return 0;
+}
+
+static int serial_pnp_resume(struct pnp_dev *dev)
+{
+	int line = (int)pnp_get_drvdata(dev);
+
+	if (!line)
+		return -ENODEV;
+	serial8250_resume_port(line - 1);
+	return 0;
+}
+#else
+#define serial_pnp_suspend NULL
+#define serial_pnp_resume NULL
+#endif /* CONFIG_PM */
+
 static struct pnp_driver serial_pnp_driver = {
 	.name		= "serial",
-	.id_table	= pnp_dev_table,
 	.probe		= serial_pnp_probe,
 	.remove		= __devexit_p(serial_pnp_remove),
+	.suspend	= serial_pnp_suspend,
+	.resume		= serial_pnp_resume,
+	.id_table	= pnp_dev_table,
 };
 
 static int __init serial8250_pnp_init(void)
#34 lore raw
In reply to Linus Torvalds

Hi Linus.

Quoted 8 lines · Linus Torvalds
Quoted 3 lines · Chuck Ebbert

When doing 'make oldconfig' we should ask about suspend/resume debug features when SOFTWARE_SUSPEND is not enabled.

That's wrong.

Quoted 8 lines

I never use SOFTWARE_SUSPEND, and I think the whole concept is totally broken.

Sane people use suspend-to-ram, and that's when you need the suspend and resume debugging.

Software-suspend is silly. I want my machine back in three seconds, not waiting for minutes..

If it's taking minutes, something is wrong. You should be looking at more like 10-30 seconds, depending on which implementation you're using, the speed of your cpu and hard disk and how much ram was saved in the image. For suspend2, for example, the rule of thumb is

ram_in_use_in_MB() / hard_disk_speed() / 2 seconds + bios time + time to get to starting the resume

Where hard disk speed is the result of hdparm -t. Assumes LZF compression (that's the /2).

Regards,

Nigel

#35 lore raw
In reply to Linus Torvalds
Quoted 8 lines · Linus Torvalds

I never use SOFTWARE_SUSPEND, and I think the whole concept is totally broken.

Sane people use suspend-to-ram, and that's when you need the suspend and resume debugging.

Software-suspend is silly. I want my machine back in three seconds, not waiting for minutes..

Suspend-to-disk has a few advantages over suspend-to-ram:

With suspend-to-disk I can remove the battery (to put in a fresh battery when traveling), try doing that with suspend-to-ram. It's also really useful for my moving my old Thinkpad which has a battery which is so bad it can't even hold power for suspend-for-ram for more than a few seconds.

With suspend-to-disk I can suspend Linux, resume Windows to be able to run that one program I have to run under Windows, suspend Windows and resume Linux again without having to restart alla applications.

And yes, this have been very useful for me in real life. So it's really nice to have both suspend-to-disk and suspend-to-ram.

/Christer

Signature · 3 lines
"Just how much can I get away with and still go to heaven?"

Christer Weinigel <[email protected]>  http://www.weinigel.se
#36 lore raw
Quoted 1 line · Christer Weinigel

With suspend-to-disk I can remove the battery (to put in a fresh battery when traveling), try doing that with suspend-to-ram.

My PowerBook can do this with suspend-to-ram too; it has an internal capacitor or battery of some sort which charges in a few minutes and holds enough power to keep the RAM alive for around a minute while I swap batteries.

Cheers,
Kyle Moffett

#37 lore raw
Quoted 8 lines · Kyle Moffett
Quoted 2 lines · Christer Weinigel

With suspend-to-disk I can remove the battery (to put in a fresh battery when traveling), try doing that with suspend-to-ram.

My PowerBook can do this with suspend-to-ram too; it has an internal capacitor or battery of some sort which charges in a few minutes and holds enough power to keep the RAM alive for around a minute while I swap batteries.

Well.. OTOH your powerbook is probably the _only_ machine that can do that :-).
Pavel

Signature · 2 lines
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
#38 lore raw
Quoted 5 lines · Pavel Machek
Quoted 4 lines · Kyle Moffett
Quoted 2 lines · Christer Weinigel

With suspend-to-disk I can remove the battery (to put in a fresh battery when traveling), try doing that with suspend-to-ram.

Quoted 7 lines
Quoted 4 lines

My PowerBook can do this with suspend-to-ram too; it has an internal capacitor or battery of some sort which charges in a few minutes and holds enough power to keep the RAM alive for around a minute while I swap batteries.

Well.. OTOH your powerbook is probably the _only_ machine that can do that :-).

Well, most ThinkPads can do that if you stick a secondary bay battery into it, but I feel the powerbook solution is a lot more elegant :)

Signature · 4 lines
  "One disk to rule them all, One disk to find them. One disk to bring
  them all and in the darkness grind them. In the Land of Redmond
  where the shadows lie." -- The Silicon Valley Tarot
  Henrique Holschuh
#39 lore raw
In reply to Linus Torvalds
Quoted 9 lines · Linus Torvalds

...
I never use SOFTWARE_SUSPEND, and I think the whole concept is totally broken.

Sane people use suspend-to-ram, and that's when you need the suspend and resume debugging.

Software-suspend is silly. I want my machine back in three seconds, not waiting for minutes..

It depends on what you are using STD for. ;-)

It might be a bit off-topic for this thread, but I'm using STD for turning my PC off in the evening and getting it back in exactly the same state in the morning.

This might not have been the original purpose of suspend, but it works quite well, STR is obviously not an alternative, and it doesn't matter much whether it takes a minute.

Quoted 1 line

Linus

cu
Adrian

Signature · 4 lines
       "Is there not promise of rain?" Ling Tan asked suddenly out
        of the darkness. There had been need of rain for many days.
       "Only a promise," Lao Er said.
                                       Pearl S. Buck - Dragon Seed
#40 lore raw
Quoted 3 lines

This might not have been the original purpose of suspend, but it works quite well, STR is obviously not an alternative, and it doesn't matter much whether it takes a minute.

Suspend to ram at the moment is for the very very brave. Having added resume quirks to the PCI resume code I've got basic resume behaving on at two boxes where resume ate the disks. I've now found a third case in testing where str/resume resumes when using Jmicron 365/366 your IDE disks swapped over which makes a really nasty mess, and that controllers like the SIL680 come back from resume misclocked and so on.

Patches for some of these will follow, to go with the pci resume quirk patch already posted.

Alan

#41 lore raw

Hi!

Quoted 10 lines
Quoted 3 lines

This might not have been the original purpose of suspend, but it works quite well, STR is obviously not an alternative, and it doesn't matter much whether it takes a minute.

Suspend to ram at the moment is for the very very brave. Having added resume quirks to the PCI resume code I've got basic resume behaving on at two boxes where resume ate the disks. I've now found a third case in testing where str/resume resumes when using Jmicron 365/366 your IDE disks swapped over which makes a really nasty mess, and that controllers like the SIL680 come back from resume misclocked and so on.

Hmm... how common are these machines? We are using unpatched kernel for suse10.2... OTOH we only support machines from the whitelist, all notebooks...
Pavel

Signature · 1 line
Thanks for all the (sleeping) penguins.
#42 lore raw
Quoted 2 lines

Hmm... how common are these machines? We are using unpatched kernel for suse10.2... OTOH we only support machines from the whitelist, all

I've always said IDE and software suspend are unsafe. The more work I do the more clearly this is/was the case.

The really nasty "resume eats your disk" cases I know about are thankfully for older systems - VIA KT133 and similar era chipsets. There is a recent nasty - Jmicron goes totally to **** on resume because of resume quirks not being run but it goes so spectacularly wrong it doesn't seem to get far enough to corrupt.

Lots of other controllers don't work correctly on resume but thats much less of a problem and with UDMA misclocking generally turns into a CRC error storm and stop.

Andrew has about 2/3rds of the bits I've done now, will push the rest when I've done a little more testing/checking. At that point libata ought to be resume safe. Someone who cares about drivers/ide legacy support can then copy the work over.

Alan

#43 lore raw
Quoted 12 lines · Alan
Quoted 2 lines

Hmm... how common are these machines? We are using unpatched kernel for suse10.2... OTOH we only support machines from the whitelist, all

I've always said IDE and software suspend are unsafe. The more work I do the more clearly this is/was the case.

The really nasty "resume eats your disk" cases I know about are thankfully for older systems - VIA KT133 and similar era chipsets. There is a recent nasty - Jmicron goes totally to **** on resume because of resume quirks not being run but it goes so spectacularly wrong it doesn't seem to get far enough to corrupt.

Quoted 8 lines

Lots of other controllers don't work correctly on resume but thats much less of a problem and with UDMA misclocking generally turns into a CRC error storm and stop.

Andrew has about 2/3rds of the bits I've done now, will push the rest when I've done a little more testing/checking. At that point libata ought to be resume safe. Someone who cares about drivers/ide legacy support can then copy the work over.

So, it seems we should discourage people from suspending with the old IDE until someone fixes it.

Perhaps we should update the documentation with this information (?)

Greetings,
Rafael

Signature · 2 lines
You never change things by fighting the existing reality.
		R. Buckminster Fuller
#44 lore raw
In reply to Alan

Hi!

Quoted 5 lines
Quoted 2 lines

Hmm... how common are these machines? We are using unpatched kernel for suse10.2... OTOH we only support machines from the whitelist, all

I've always said IDE and software suspend are unsafe. The more work I do the more clearly this is/was the case.

Well, there's unsafe as in "crashes", and that's unsafe as in "eats disks".

Quoted 2 lines

The really nasty "resume eats your disk" cases I know about are thankfully for older systems - VIA KT133 and similar era chipsets.

Aha, good. Hopefully noone has notebook with those.

Quoted 3 lines

There is a recent nasty - Jmicron goes totally to **** on resume because of resume quirks not being run but it goes so spectacularly wrong it doesn't seem to get far enough to corrupt.

Good :-). Crashing is nasty, but we probably won't add that machine to whitelist.

Quoted 4 lines

Andrew has about 2/3rds of the bits I've done now, will push the rest when I've done a little more testing/checking. At that point libata ought to be resume safe. Someone who cares about drivers/ide legacy support can then copy the work over.

Thanks. I do not think we care about old mainboards enough to do 2.6.16-stable backport.
Pavel

Signature · 2 lines
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
#45 lore raw
In reply to Linus Torvalds
Quoted 4 lines · Linus Torvalds

Sane people use suspend-to-ram, and that's when you need the suspend and resume debugging.

Software-suspend is silly. I want my machine back in three seconds, not waiting for minutes..

ALL of my notebooks here have always been capable of Suspend-to-RAM with Linux, and that's how I like it. We've had six different models (IBM, Dell, Toshiba), and they have all worked fine with S2R.

The older APM ones had zero issues, and just always worked.

The newer ACPI ones have required some tweaking to the sleep/wakeup scripts, but now work perfectly with suspend-to-ram.

Suspend-to-disk (and resume from disk) are way too slow for regular use, except of course for the swsusp2 version, which is nearly as quick as S2R is.

I do use S2D when travelling, as it's much easier on the batteries while the notebook is packed away on the red-eye flights.

Cheers