suspend2 problems on SMP machine, incorrect tainting

suspend2 problems on SMP machine, incorrect tainting

[Pavel Machek]

Hi!

I tried suspend2 2.0.0.81 on  Toshiba HT machine, and it did not
work :-(. First test was with "noapic nosmp", last messages are

Message from syslogd@elonex...
elonex kernel: - I/O speed: Write 80 MB/s, Read 83 MB/s

Unable to handle kernel NULL pointer deference at virtual address 100
...
EIP: c13f7f81 (<- outside of kernel symbols?! strange) Tainted: G S
...
(no call trace, strange).



Tainted: S is due to suspend2, right? It abuses flag that origally means
"invalid SMP configuration". You might want to fix that.

Tried without noapic nosmp... display got "interesting" at one point,
but hey, it worked! I have "sleeping function called rom invalid
context at arch/i386/mm/highmem.c:5" called from kmap, copy_pageset1.

Tried "noapic" only, worked ok.

-- 
934a471f20d6580d5aad759bf0d97ddc

Re: suspend2 problems on SMP machine, incorrect tainting

[Nigel Cunningham]

Hi.

Pavel Machek wrote:
> Hi!
> 
> I tried suspend2 2.0.0.81 on  Toshiba HT machine, and it did not
> work :-(. First test was with "noapic nosmp", last messages are

Hmm. I haven't tried it with noapic or nosmp options before. It ought to work; I'll see if I can 
reproduce this issue.

> Tainted: S is due to suspend2, right? It abuses flag that origally means
> "invalid SMP configuration". You might want to fix that.

Yes. I've been meaning to change that; must get around to it.

> Tried without noapic nosmp... display got "interesting" at one point,

Yes. The sequence of operations to force a redraw of the screen works okay under 2.4 but needs 
looking at under 2.6.

> but hey, it worked! I have "sleeping function called rom invalid
> context at arch/i386/mm/highmem.c:5" called from kmap, copy_pageset1.

Yes. It needs to be atomic (it's the atomic copy) and pages might well be HighMem so I guess the 
answer is to suppress the message rather than changing something in suspend.

Nigel
-- 
Nigel & Michelle Cunningham
C/- Westminster Presbyterian Church Belconnen
61 Templeton Street, Cook, ACT 2614.
+61 (2) 6251 7727(wk); +61 (2) 6254 0216 (home)

After homosexuality, they'll be arguing paedophilia is normal.

Re: [Swsusp-devel] Re: suspend2 problems on SMP machine, incorrect tainting

[Nigel Cunningham]

Hi again.

Nigel Cunningham wrote:
> Yes. It needs to be atomic (it's the atomic copy) and pages might well 
> be HighMem so I guess the answer is to suppress the message rather than 
> changing something in suspend.

Before someone says it, yes. I mean using kmap_atomic :>

Nigel
-- 
Nigel & Michelle Cunningham
C/- Westminster Presbyterian Church Belconnen
61 Templeton Street, Cook, ACT 2614.
+61 (417) 100 574 (mobile)

After homosexuality, they'll be arguing paedophilia is normal.

Re: [Swsusp-devel] suspend2 problems on SMP machine, incorrect tainting

[Nigel Cunningham]

Hi again.

I should mention that the code is not perfectly stable at the moment. As soon as it is I'll get 
stuck into last cleanups and then the merge. Crashes I'm seeing involve an oops shortly after 
resuming, occuring in slab code. Obviously an inconsistency has somehow sneaked in. I'm away from 
home for 8 days from Monday AM, so a fix won't be forthcoming this week. It shouldn't be far away 
though.

Regards,

Nigel
-- 
Nigel & Michelle Cunningham
C/- Westminster Presbyterian Church Belconnen
61 Templeton Street, Cook, ACT 2614.
+61 (2) 6251 7727(wk); +61 (2) 6254 0216 (home)

After homosexuality, they'll be arguing paedophilia is normal.

Re: [Swsusp-devel] suspend2 problems on SMP machine, incorrect tainting

[Pavel Machek]

Hi!

> I should mention that the code is not perfectly stable at the moment. As 
> soon as it is I'll get stuck into last cleanups and then the merge. Crashes 
> I'm seeing involve an oops shortly after resuming, occuring in slab
> code. 

The oops I saw with "nosmp noapic" was after resuming... it might be
related.
								Pavel
-- 
934a471f20d6580d5aad759bf0d97ddc