new ATI sleep code freezes

new ATI sleep code freezes

[mutex]

Hi all,
This has taken me a while to decide it was a bug and not a freak
accident. About one to two times a week I have noticed that my powerbook
will hand during the sleep process. Usually, when the sleep operation
completes, there is a quick screen power blank. Then about another 1.5
seconds of the LCD being illuminated, with what looks like some random
part of video memory being displayed, then finally the machine goes
to sleep. Occasionally I am seeing a freeze during the step where the
machine stays on for 1.5 seconds and has not quite been put to sleep. I
have been unable to diagnose this, as it happens so infrequently, but I
figured by posting I could hopefully get some more people experiencing
this as well. my specs:
	1ghz tipb, 1gig ram, radeon 9M, 2.4.20-ben8

I also do not seem to be doing any tasks at this time, just idling my
window manager.

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: new ATI sleep code freezes

[Michel Dänzer]

On Thu, 2003-05-29 at 17:41, mutex@kerneli.org wrote:
> Hi all,
> This has taken me a while to decide it was a bug and not a freak
> accident. About one to two times a week I have noticed that my powerbook
> will hand during the sleep process. Usually, when the sleep operation
> completes, there is a quick screen power blank. Then about another 1.5
> seconds of the LCD being illuminated, with what looks like some random
> part of video memory being displayed, then finally the machine goes
> to sleep. Occasionally I am seeing a freeze during the step where the
> machine stays on for 1.5 seconds and has not quite been put to sleep. I
> have been unable to diagnose this, as it happens so infrequently, but I
> figured by posting I could hopefully get some more people experiencing
> this as well. my specs:
> 	1ghz tipb, 1gig ram, radeon 9M, 2.4.20-ben8

Pretty much the same configuration here, but I haven't seen this, even
though I've been enjoying uptimes of weeks, usually with at least one
sleep cycle a day.

What's your DRI related setup?


--
Earthling Michel Dänzer   \  Debian (powerpc), XFree86 and DRI developer
Software libre enthusiast  \     http://svcs.affero.net/rm.php?r=daenzer


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: new ATI sleep code freezes

[mutex]

[-- Attachment #1: Type: text/plain, Size: 1462 bytes --]

see below...

On Sun, Jun 01, 2003 at 09:36:11PM +0200 or thereabouts, Michel D?nzer wrote:
>
> On Thu, 2003-05-29 at 17:41, mutex@kerneli.org wrote:
> >
> > This has taken me a while to decide it was a bug and not a freak
> > accident. About one to two times a week I have noticed that my powerbook
> > will hand during the sleep process. Usually, when the sleep operation
> > completes, there is a quick screen power blank. Then about another 1.5
> > seconds of the LCD being illuminated, with what looks like some random
> > part of video memory being displayed, then finally the machine goes
> > to sleep. Occasionally I am seeing a freeze during the step where the
> > machine stays on for 1.5 seconds and has not quite been put to sleep. I
> > have been unable to diagnose this, as it happens so infrequently, but I
> > figured by posting I could hopefully get some more people experiencing
> > this as well. my specs:
> > 	1ghz tipb, 1gig ram, radeon 9M, 2.4.20-ben8
>
> Pretty much the same configuration here, but I haven't seen this, even
> though I've been enjoying uptimes of weeks, usually with at least one
> sleep cycle a day.
>
> What's your DRI related setup?

hmmm well i don't seem to have _any_ sort of DRI setup. Although
you noting for me to look at my XF86Config, i noticed i had left
AGPFastWrite enabled for a test i was running for benh a while ago.
I will set that off and see if that helps at all. I have attached my
XF86Config

Dave

[-- Attachment #2: XF86Config-4 --]
[-- Type: text/plain, Size: 3620 bytes --]

Section "ServerLayout"
Identifier     "XFree86 Configured"
Screen      0  "Screen0" 0 0
InputDevice    "Mouse0" "CorePointer"
InputDevice    "Keyboard0" "CoreKeyboard"
EndSection

Section "Files"
RgbPath      "/usr/X11R6/lib/X11/rgb"
ModulePath   "/usr/X11R6/lib/modules"
FontPath	"/usr/X11R6/lib/X11/fonts/truetype"
FontPath     "/usr/X11R6/lib/X11/fonts/misc/"
FontPath     "/usr/X11R6/lib/X11/fonts/Speedo/"
FontPath     "/usr/X11R6/lib/X11/fonts/Type1/"
FontPath     "/usr/X11R6/lib/X11/fonts/CID/"
FontPath     "/usr/X11R6/lib/X11/fonts/75dpi/"
FontPath     "/usr/X11R6/lib/X11/fonts/100dpi/"
EndSection

Section "Module"
Load  "extmod"
#Load  "dri"
#Load  "drm"
Load  "dbe"
Load  "record"
Load  "xtrap"
Load  "glx"
Load  "speedo"
Load  "type1"
EndSection

Section "InputDevice"
Identifier  "Keyboard0"
Driver      "keyboard"
#Option          "CustomKeycodes"
Option          "XkbModel"      "Macintosh"
Option          "XkbLayout"     "us"
Option          "XkbRules"      "xfree"
EndSection

Section "InputDevice"
Identifier  "Mouse0"
Driver      "mouse"
Option 	"Protocol"        "IMPS/2"
Option          "ZAxisMapping"  "4 5"
Option          "Device"        "/dev/input/mice"
EndSection

Section "Monitor"
Identifier   "Monitor0"
VendorName   "Monitor Vendor"
ModelName    "Monitor Model"
UseModes	"Modes0"
HorizSync       30-70
VertRefresh     50-160
EndSection

Section "Device"
### Available Driver options are:-
### Values: <i>: integer, <f>: float, <bool>: "True"/"False",
### <string>: "String", <freq>: "<f> Hz/kHz/MHz"
### [arg]: arg optional
#Option     "NoAccel"            	# [<bool>]
#Option     "SWcursor"           	# [<bool>]
#Option     "Dac6Bit"            	# [<bool>]
#Option     "Dac8Bit"            	# [<bool>]
#Option     "ForcePCIMode"       	# [<bool>]
#Option     "CPPIOMode"          	# [<bool>]
#Option     "CPusecTimeout"      	# <i>
#Option     "AGPMode"            	# <i>
Option     "AGPFastWrite"       	"True"# [<bool>]
#Option     "AGPSize"            	# <i>
#Option     "RingSize"           	# <i>
#Option     "BufferSize"         	# <i>
#Option     "EnableDepthMoves"   	# [<bool>]
#Option     "EnablePageFlip"     	# [<bool>]
#Option     "NoBackBuffer"       	# [<bool>]
#Option     "PanelOff"           	# [<bool>]
#Option     "DDCMode"            	# [<bool>]
#Option     "CloneDisplay"       	# <i>
#Option     "CloneMode"          	# [<str>]
#Option     "CloneHSync"         	# [<str>]
#Option     "CloneVRefresh"      	# [<str>]
Option     "UseFBDev"          "True"	# [<bool>]
#Option     "VideoKey"           	# <i>
Identifier  "Card0"
Driver 		"ati"
VendorName  "ATI Technologies Inc"
BoardName   "Radeon R250 Lf [Radeon Mobility 9000]"
BusID       "PCI:0:16:0"
EndSection

Section "Screen"
Identifier "Screen0"
Device     "Card0"
Monitor    "Monitor0"
DefaultDepth 16
SubSection "Display"
Depth 1
Modes  	"1280x854" 
EndSubSection
SubSection "Display"
Depth 4
Modes  	"1280x854" 
EndSubSection
SubSection "Display"
Depth 8
Modes  	"1280x854" 
EndSubSection
SubSection "Display"
Depth 15
Modes  	"1280x854" 
EndSubSection
SubSection "Display"
Depth 16
Modes  	"1280x854" 
EndSubSection
SubSection "Display"
Depth 24
Modes  	"1280x854" 
EndSubSection
SubSection "Display"
Depth 32
Modes  	"1280x854" 
EndSubSection
EndSection

Section "DRI"
	Mode 0666
EndSection

Section "Modes"
        Identifier      "Modes0"
	# mode generated by Xeasyconf
	Modeline "1280x854" 79.815 1280 1296 1408 1536  854 855 858 866  -HSync -VSync
EndSection

Re: new ATI sleep code freezes

[Benjamin Herrenschmidt]

Also make sure you have APM emulation working (check XFree log to see if it found & opened the APM device)

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: new ATI sleep code freezes

[mutex]

On Wed, Jun 04, 2003 at 07:15:55PM +0200 or thereabouts,
Benjamin Herrenschmidt wrote:
> Also make sure you have APM emulation working (check XFree log to see
> if it found & opened the APM device)

from my XF86Config file...

(WW) Open APM failed (/dev/apm_bios) (No such file or directory)

But my kernel config says...
CONFIG_PMAC_APM_EMU=y

The only thing unusual is I am running devfs..
so i checked /dev

heliosphan dev # find . -name *apm*
./misc/apm_bios
./apm_bios

/dev/misc/apm_bios has perms:

crw-------    1 mutex    root      10, 134 Dec 31  1969 apm_bios

are those normal permissions ? Also, I'm running an enlightenment
battery applet which afaik uses apm layer to report battery level. So
something has to be working.

Dave

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: new ATI sleep code freezes

[Benjamin Herrenschmidt]

On Thu, 2003-06-05 at 18:35, mutex@kerneli.org wrote:
> On Wed, Jun 04, 2003 at 07:15:55PM +0200 or thereabouts, Benjamin Herrenschmidt wrote:
> > Also make sure you have APM emulation working (check XFree log to see if it found & opened the APM device)
>
> from my XF86Config file...
>
> (WW) Open APM failed (/dev/apm_bios) (No such file or directory)
>
> But my kernel config says...
> CONFIG_PMAC_APM_EMU=y
>
> The only thing unusual is I am running devfs..

Poor boy... ;)

> so i checked /dev
>
> heliosphan dev # find . -name *apm*
> ./misc/apm_bios
> ./apm_bios
>
> /dev/misc/apm_bios has perms:
>
> crw-------    1 mutex    root      10, 134 Dec 31  1969 apm_bios
>
> are those normal permissions ?  Also, I'm running an enlightenment battery applet which afaik uses apm layer to report battery level.  So something has to be working.
> Dave

I don't know wtf with devfs & perms, but the problem is here, once you
fix that, it should work much better ;)

Ben.

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: new ATI sleep code freezes

[Michel Dänzer]

On Thu, 2003-06-05 at 18:35, mutex@kerneli.org wrote:
> On Wed, Jun 04, 2003 at 07:15:55PM +0200 or thereabouts,
> Benjamin Herrenschmidt wrote:
> > Also make sure you have APM emulation working (check XFree log to see
> > if it found & opened the APM device)
>
> from my XF86Config file...
>
> (WW) Open APM failed (/dev/apm_bios) (No such file or directory)
>
> But my kernel config says...
> CONFIG_PMAC_APM_EMU=y
>
> The only thing unusual is I am running devfs..
> so i checked /dev
>
> heliosphan dev # find . -name *apm*
> ./misc/apm_bios
> ./apm_bios
>
> /dev/misc/apm_bios has perms:
>
> crw-------    1 mutex    root      10, 134 Dec 31  1969 apm_bios
>
> are those normal permissions ?

I don't know, I don't use devfs, I have:

crw-rw----    1 root     root      10, 134 2003-04-29 13:04
/dev/apm_bios

Anyway, the X server doesn't try to use /dev/misc/apm_bios but
/dev/apm_bios and gets ENOENT.


> Also, I'm running an enlightenment battery applet which afaik uses apm
> layer to report battery level. So something has to be working.

It might use /proc/apm.


--
Earthling Michel Dänzer   \  Debian (powerpc), XFree86 and DRI developer
Software libre enthusiast  \     http://svcs.affero.net/rm.php?r=daenzer


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: new ATI sleep code freezes

[Mich Lanners]

On   5 Jun, this message from Benjamin Herrenschmidt echoed through
cyberspace:
>> from my XF86Config file...
>>
>> (WW) Open APM failed (/dev/apm_bios) (No such file or directory)
>>
>> But my kernel config says...
>> CONFIG_PMAC_APM_EMU=y
>>
>> The only thing unusual is I am running devfs..
>
> Poor boy... ;)

Na. I run devfs on my Powerbook, and I am very happy. Never had any
problem. I am absolutely certain (even if it may not be true :-) that
using devfs saves me quite a number of HD spinups.

>> so i checked /dev
>>
>> heliosphan dev # find . -name *apm*
>> ./misc/apm_bios
>> ./apm_bios
>>
>> /dev/misc/apm_bios has perms:
>>
>> crw-------    1 mutex    root      10, 134 Dec 31  1969 apm_bios

Mine show this:

mlan@pooh:~$ ll /dev/apm_bios  /dev/misc/apm_bios
lr-xr-xr-x    1 root     root           13 2003-05-14 22:42 /dev/apm_bios -> misc/apm_bios
crw-rw----    1 root     root      10, 134 1970-01-01 01:00 /dev/misc/apm_bios

This is with the Debian devfs package, which _may_ fix up a few things.
My devfs config says this:

mlan@pooh:/etc/devfs$ find . -type f |xargs grep apm
/conf.d/apm:REGISTER ^misc/apm_bios$   PERMISSIONS root.root 0660
/conf.d/pmud:REGISTER ^misc/apm_bios$   PERMISSIONS root.root 0660
/perms:REGISTER ^misc/apm_bios         PERMISSIONS root.root   0660

plus this one for setting up symlinks:

mlan@pooh:/etc/devfs$ head compat_symlinks
# Enable full compatibility mode for old device names. You may comment these
# out if you don't use the old device names. Make sure you know what you're
# doing!
REGISTER        .*              MKOLDCOMPAT
UNREGISTER      .*              RMOLDCOMPAT

>> are those normal permissions ?  Also, I'm running an enlightenment
>> battery applet which afaik uses apm layer to report battery level.
>> So something has to be working. Dave
>
> I don't know wtf with devfs & perms, but the problem is here, once you
> fix that, it should work much better ;)

I'm tempted to say: install Debian :-)

Cheers

Michel

-------------------------------------------------------------------------
Michel Lanners                 |  " Read Philosophy.  Study Art.
23, Rue Paul Henkes            |    Ask Questions.  Make Mistakes.
L-1710 Luxembourg              |
email   mlan@cpu.lu            |
http://www.cpu.lu/~mlan        |                     Learn Always. "

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: new ATI sleep code freezes

[mutex]

On Thu, Jun 05, 2003 at 11:16:00PM +0200 or thereabouts, Mich Lanners wrote:
> On   5 Jun, this message from Benjamin Herrenschmidt echoed through cyberspace:
> >> from my XF86Config file...
> >>
> >> (WW) Open APM failed (/dev/apm_bios) (No such file or directory)
> >>
> >> But my kernel config says...
> >> CONFIG_PMAC_APM_EMU=y
> >>
> >> The only thing unusual is I am running devfs..
> >
> > Poor boy... ;)
>
> Na. I run devfs on my Powerbook, and I am very happy. Never had any
> problem. I am absolutely certain (even if it may not be true :-) that
> using devfs saves me quite a number of HD spinups.
>
> >> so i checked /dev
> >>
> >> heliosphan dev # find . -name *apm*
> >> ./misc/apm_bios
> >> ./apm_bios
> >>
> >> /dev/misc/apm_bios has perms:
> >>
> >> crw-------    1 mutex    root      10, 134 Dec 31  1969 apm_bios
>
> Mine show this:
>
> mlan@pooh:~$ ll /dev/apm_bios  /dev/misc/apm_bios
> lr-xr-xr-x    1 root     root           13 2003-05-14 22:42 /dev/apm_bios -> misc/apm_bios
> crw-rw----    1 root     root      10, 134 1970-01-01 01:00 /dev/misc/apm_bios
>
> This is with the Debian devfs package, which _may_ fix up a few things.
> My devfs config says this:

well i'm running gentoo, and the symlink is properly there.

> mlan@pooh:/etc/devfs$ find . -type f |xargs grep apm
> /conf.d/apm:REGISTER ^misc/apm_bios$   PERMISSIONS root.root 0660
> /conf.d/pmud:REGISTER ^misc/apm_bios$   PERMISSIONS root.root 0660
> /perms:REGISTER ^misc/apm_bios         PERMISSIONS root.root   0660

gentoo has devfs configuration in a different place, but I'll dig
through that, see if I can find something similar to that.

> plus this one for setting up symlinks:
>
> mlan@pooh:/etc/devfs$ head compat_symlinks
> # Enable full compatibility mode for old device names. You may comment these
> # out if you don't use the old device names. Make sure you know what you're
> # doing!
> REGISTER        .*              MKOLDCOMPAT
> UNREGISTER      .*              RMOLDCOMPAT
>
> >> are those normal permissions ?  Also, I'm running an enlightenment
> >> battery applet which afaik uses apm layer to report battery level.
> >> So something has to be working. Dave
> >
> > I don't know wtf with devfs & perms, but the problem is here, once you
> > fix that, it should work much better ;)

well(in response to ben) I have a hard time believing that once I fix
permissions it will eliminate a crash that happens 1/20 sleeps. Just
doesn't seem like something a permissions problem would cause. But I'll
fix it and see what happens next week.

> I'm tempted to say: install Debian :-)

right, heard that one before. Although I have to admit your the first
debian user in a while that didn't tell me to rtfm ;-)

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: new ATI sleep code freezes

[Michel Dänzer]

On Sat, 2003-06-07 at 21:28, mutex@kerneli.org wrote:

> > >> are those normal permissions ?  Also, I'm running an enlightenment
> > >> battery applet which afaik uses apm layer to report battery level.
> > >> So something has to be working. Dave
> > >
> > > I don't know wtf with devfs & perms, but the problem is here, once you
> > > fix that, it should work much better ;)
>
> well(in response to ben) I have a hard time believing that once I fix
> permissions it will eliminate a crash that happens 1/20 sleeps. Just
> doesn't seem like something a permissions problem would cause.

The permissions don't cause the crash directly, it's the fact that the X
server has no way of dealing with the sleep without using /dev/apm_bios.


--
Earthling Michel Dänzer   \  Debian (powerpc), XFree86 and DRI developer
Software libre enthusiast  \     http://svcs.affero.net/rm.php?r=daenzer


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: new ATI sleep code freezes

[mutex]

> The permissions don't cause the crash directly, it's the fact that
> the X server has no way of dealing with the sleep without using
> /dev/apm_bios.

I was able to re-create this crash the other day entirely reproduceably.
If I unplug my USB apple mouse while the computer is in between sleeping
and running normally the crash condition i described occurs, namely:

> Usually, when the sleep operation completes, there is a quick
> screen power blank. Then about another 1.5 seconds of the LCD being
> illuminated, with what looks like some random part of video memory
> being displayed, then finally the machine goes to sleep. Occasionally
> I am seeing a freeze during the step where the machine stays on for
> 1.5 seconds and has not quite been put to sleep.

Maybe there's a deadlock somewhere in the USB layer sleep code ? I
haven't tried this with other USB devices, but I plan on doing that as
so on as I get back from OLS.

Regards,
Dave

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/