Go Back   nV News Forums > Linux Support Forums > NVIDIA Linux

Newegg Daily Deals

Reply
 
Thread Tools
Old 08-19-08, 08:05 PM   #1
kcrudup
Registered User
 
Join Date: Apr 2004
Location: Los Angeles, CA 90034
Posts: 96
Default Power Management broken in 177.67

The very same xorg.conf configuration I use with 173.14.12 will not resume from a (TuxOnIce) hibernate cycle. I had to revert back to that previous version, but before I did I tried changing a couple of the usual suspects (turning off both "UseEvents"/"DamageEvents") and even IPP/GlyphCache as a just-in-case.

On resume my backlight is on, but there's no screen output. I can't even switch to a text mode.

I can't file a proper bug report as I could still sync out the system (SysRq-S/U/B), but there's nothing in the syslog to imply anything went wrong (other than the screen never visually came up).

I see that NVidia's made changes to the PCI-E MSI handing in 177.67. I have an 8600M GS on PCI-E and my kernel (2.6.27-rc3-git4) has PCI-E and MSI enabled- maybe that's it?

If Nvidia wants it, I can at least post up my kernel's .config file.
__________________
8600M GS 512MB (HP Pavilion dv9000 CTO)
kcrudup is offline   Reply With Quote
Old 08-19-08, 09:32 PM   #2
kcrudup
Registered User
 
Join Date: Apr 2004
Location: Los Angeles, CA 90034
Posts: 96
Default Re: Power Management broken in 177.67

Huh. Interestingly enough, it appears the non-TuxOnIce latest (August 19th) GIT build doesn't suffer from this issue. I'll keep playing with it, but I don't see any PM/MSI/PCI-E related changes in the commit logs.

... stay tuned ....
__________________
8600M GS 512MB (HP Pavilion dv9000 CTO)
kcrudup is offline   Reply With Quote
Old 08-20-08, 01:55 AM   #3
tier
Registered User
 
Join Date: Sep 2007
Posts: 118
Default Re: Power Management broken in 177.67

Same problem here with Suspend to Ram, the screen just stays black and only sysrq works..

Antoher thing is that I cannot switch to a terminal when using a framebuffer like vesafb, the screen also stays black (with only backlight turned on) and sometimes I can switch back..
tier is offline   Reply With Quote
Old 08-20-08, 05:17 AM   #4
Dragoran
Registered User
 
Join Date: May 2004
Posts: 711
Default Re: Power Management broken in 177.67

Suspend to Ram does not work here either.
I tryed without vesafb same result.
VT switch works fine however.
Attached Files
File Type: log nvidia-bug-report.log (144.4 KB, 114 views)
Dragoran is offline   Reply With Quote
Old 08-20-08, 05:49 AM   #5
NvFuchs
Registered Fox
 
NvFuchs's Avatar
 
Join Date: Oct 2007
Location: Confoederatio Helvetica
Posts: 530
Default Re: Power Management broken in 177.67

S3 and S4 working here with a 2.6.26-r2 non-tuxonice, have to try a tux on ice later on. This is a lenovo Thinkpad R61.

tier: please provie a nvidia-bug-report.log

Fuchs
NvFuchs is offline   Reply With Quote
Old 08-20-08, 07:02 AM   #6
tier
Registered User
 
Join Date: Sep 2007
Posts: 118
Default Re: Power Management broken in 177.67

Quote:
Originally Posted by NvFuchs View Post
S3 and S4 working here with a 2.6.26-r2 non-tuxonice, have to try a tux on ice later on. This is a lenovo Thinkpad R61.

tier: please provie a nvidia-bug-report.log

Fuchs
Here it is.
Attached Files
File Type: log nvidia-bug-report.log (109.7 KB, 107 views)
tier is offline   Reply With Quote
Old 08-20-08, 08:53 AM   #7
Dragoran
Registered User
 
Join Date: May 2004
Posts: 711
Default Re: Power Management broken in 177.67

Quote:
Originally Posted by Dragoran View Post
Suspend to Ram does not work here either.
I tryed without vesafb same result.
VT switch works fine however.
OK, seems like it works if I disable MSI ... well there is a reason why the changelog says "experimental"
Dragoran is offline   Reply With Quote
Old 08-20-08, 09:55 AM   #8
zander
NVIDIA Corporation
 
zander's Avatar
 
Join Date: Aug 2002
Posts: 3,740
Default Re: Power Management broken in 177.67

@Dragoran: the driver's MSI support is disabled by default; when you say 'if I disable MSI', are you referring to the Linux kernel's MSI support?
zander is offline   Reply With Quote

Old 08-20-08, 10:09 AM   #9
Dragoran
Registered User
 
Join Date: May 2004
Posts: 711
Default Re: Power Management broken in 177.67

Quote:
Originally Posted by zander View Post
@Dragoran: the driver's MSI support is disabled by default; when you say 'if I disable MSI', are you referring to the Linux kernel's MSI support?
No I explicitly enabled it by passing
Code:
 NVreg_EnableMSI=1
to the kernel module.
Removing this again seems to fix suspend for me.
Dragoran is offline   Reply With Quote
Old 08-20-08, 10:44 AM   #10
Dragoran
Registered User
 
Join Date: May 2004
Posts: 711
Default Re: Power Management broken in 177.67

Seems like MSI is causing random crashes after suspend here (with 2.6.26.2) so I disabled it for all modules now.
The bug might be not directly related to how the nvidia driver handles MSI.
Dragoran is offline   Reply With Quote
Old 08-20-08, 10:49 AM   #11
zander
NVIDIA Corporation
 
zander's Avatar
 
Join Date: Aug 2002
Posts: 3,740
Default Re: Power Management broken in 177.67

OK, thanks for the clarification. For what it's worth, the (NVIDIA kernel module) code paths taken in the MSI vs. the virtual-wire interrupt paths are not dissimilar enough for me to expect differences in the respective power management behavior. Please let me know if you do find evidence to suggest something in the NVIDIA driver's MSI support (as opposed to the Linux kernel's) is causing these problems, though.
zander is offline   Reply With Quote
Old 08-20-08, 12:51 PM   #12
kcrudup
Registered User
 
Join Date: Apr 2004
Location: Los Angeles, CA 90034
Posts: 96
Default Re: Power Management broken in 177.67

Just for kicks, I tried turning on NVreg_EnableMSI=1 and tried a (non-TuxOnIce) suspend/resume cycle. It failed, but I was able to SysRq-K to get back to X and a console. Here's the relevant parts of the the dmesg:

Code:
[  201.954043] irq 16: nobody cared (try booting with the "irqpoll" option)
[  201.954043] Pid: 0, comm: swapper Tainted: P        W 2.6.27-rc3-GIT #16
[  201.954043] 
[  201.954043] Call Trace:
[  201.954043]  <IRQ>  [<ffffffff80282776>] __report_bad_irq+0x26/0x80
[  201.954043]  [<ffffffff80282a67>] note_interrupt+0x297/0x2d0
[  201.954043]  [<ffffffff80283355>] handle_fasteoi_irq+0xe5/0x110
[  201.954043]  [<ffffffff8020f121>] do_IRQ+0x81/0x100
[  201.954043]  [<ffffffff8020c0e1>] ret_from_intr+0x0/0xa
[  201.954043]  <EOI>  [<ffffffff8026386f>] ? tick_nohz_stop_sched_tick+0xcf/0x380
[  201.954043]  [<ffffffff8020aa2a>] ? cpu_idle+0x2a/0x100
[  201.954043]  [<ffffffff80815ec6>] ? rest_init+0x86/0x90
[  201.954043] 
[  201.954043] handlers:
[  201.954043] [<ffffffff8062d0f0>] (usb_hcd_irq+0x0/0x70)
[  201.954043] Disabling IRQ #16
The virtual-wire IRQ is 16, but the MSI IRQ is 313:
Code:
 16:     100001          0   IO-APIC-fasteoi   uhci_hcd:usb3
...
313:       2256          0   PCI-MSI-edge      nvidia
I'm suspecting that on the power-cycle after a suspend, the legacy IRQ gets reconnected to the VW IRQ and when it fires on resume you get the "nobody cared".

Oh, and I've attached a proper bug report this time:
Attached Files
File Type: gz nvidia-bug-report.log.gz (38.3 KB, 115 views)
__________________
8600M GS 512MB (HP Pavilion dv9000 CTO)
kcrudup is offline   Reply With Quote
Reply


Thread Tools

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Forum Jump


Similar Threads
Thread Thread Starter Forum Replies Last Post
After Greenpeace Protests, Apple Promises to Dump Coal Power News Archived News Items 0 05-18-12 03:10 AM
Power Management (XP) Iron1384 Microsoft Windows XP And Vista 6 08-22-02 01:29 PM

All times are GMT -5. The time now is 04:52 AM.


Powered by vBulletin® Version 3.7.1
Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.
Copyright 1998 - 2014, nV News.