Skip to contentSkip to conversation

Linux Kernel cfs scheduler and xorg kbd

DISCUSSION <::.. Teresa_II ..::> 11 participants 38 messages Jul 30, 2007 – Aug 9, 2007

lore
Share & tools
lore
#0 lore raw

Hi all,

first of all i am not sure if its scheduler issue. But i didn't found anything about that in google related to something else.

So i tried 2.6.22.1 kernel with new cfs scheduler v19.1 from Ingo page. Previous i used 2.6.22-ck1.

After 24 hours running i get strange behavior in Xorg. It act slower with my special keys, like SHIFT, or CTRL. So if i hold SHIFT and press ")" and than will press enter, i release SHIIFT and press ENTER it doesn't create new line. Even if i start write a Sentence or word with "SHIFTed" char it makes next char in high case also.
Like this: "HAllo" even if i type "Hallo".

It's extremely unusable.

It doesn't happen if i am on other kernel.

I am ready to provide any requested information to find issue. If its not kernel related thing - i am very sorry, i just don't know where to start to trace whats wrong.

I am on gentoo and use latest xorg server. Keyboard driver is "KBD". Mouse use "evdev" driver.

Signature · 2 lines
Teresa Kalitowska
jid: [email protected]
#1 lore raw

Hi,

Quoted 18 lines · On 30/07/07, <::.. Teresa_II ..::> <[email protected]> wrote:

Hi all,

first of all i am not sure if its scheduler issue. But i didn't found anything about that in google related to something else.

So i tried 2.6.22.1 kernel with new cfs scheduler v19.1 from Ingo page. Previous i used 2.6.22-ck1.

After 24 hours running i get strange behavior in Xorg. It act slower with my special keys, like SHIFT, or CTRL. So if i hold SHIFT and press ")" and than will press enter, i release SHIIFT and press ENTER it doesn't create new line. Even if i start write a Sentence or word with "SHIFTed" char it makes next char in high case also.
Like this: "HAllo" even if i type "Hallo".

It's extremely unusable.

Quoted 8 lines

It doesn't happen if i am on other kernel.

I am ready to provide any requested information to find issue. If its not kernel related thing - i am very sorry, i just don't know where to start to trace whats wrong.

I am on gentoo and use latest xorg server. Keyboard driver is "KBD". Mouse use "evdev" driver.

Ingo, can you take a look at this?

Quoted 4 lines

--
Teresa Kalitowska
jid: [email protected]

Regards,
Michal

Signature · 2 lines
LOG
http://www.stardust.webpages.pl/log/
#2 lore raw
Quoted 23 lines · Michal Piotrowski

Hi,

Quoted 20 lines · On 30/07/07, <::.. Teresa_II ..::> <[email protected]> wrote:

Hi all,

first of all i am not sure if its scheduler issue. But i didn't found anything about that in google related to something else.

So i tried 2.6.22.1 kernel with new cfs scheduler v19.1 from Ingo page. Previous i used 2.6.22-ck1.

After 24 hours running i get strange behavior in Xorg. It act slower with my special keys, like SHIFT, or CTRL. So if i hold SHIFT and press ")" and than will press enter, i release SHIIFT and press ENTER it doesn't create new line. Even if i start write a Sentence or word with "SHIFTed" char it makes next char in high case also.
Like this: "HAllo" even if i type "Hallo".

It's extremely unusable.

It doesn't happen if i am on other kernel.

Quoted 8 lines
Quoted 6 lines

I am ready to provide any requested information to find issue. If its not kernel related thing - i am very sorry, i just don't know where to start to trace whats wrong.

I am on gentoo and use latest xorg server. Keyboard driver is "KBD". Mouse use "evdev" driver.

Ingo, can you take a look at this?

another reaction below in this thread reported kbd problems in vanilla 2.6.22.1 as well. What is the X versions, etc.? Does the problem go away if the X kbd driver selection is tweaked to a simpler model, say:

Section "InputDevice"

         Identifier  "Keyboard0"
         Driver      "kbd"
         Option      "XkbModel" "pc105"
         Option      "XkbLayout" "us"
         Option      "XkbVariant" "basic"

EndSection

Ingo

----------------------->

From: Rene Herman <[email protected]>

Quoted 6 lines · On 07/30/2007 10:25 PM, <::.. Teresa_II ..::> wrote:

first of all i am not sure if its scheduler issue. But i didn't found anything about that in google related to something else.

So i tried 2.6.22.1 kernel with new cfs scheduler v19.1 from Ingo page. Previous i used 2.6.22-ck1.

Quoted 6 lines

After 24 hours running i get strange behavior in Xorg. It act slower with my special keys, like SHIFT, or CTRL. So if i hold SHIFT and press ")" and than will press enter, i release SHIIFT and press ENTER it doesn't create new line. Even if i start write a Sentence or word with "SHIFTed" char it makes next char in high case also.
Like this: "HAllo" even if i type "Hallo".

For a few days now I have been experiencing a "stuck delete" key while reading mail -- I run through the mail with "delete" and every once in a while it just keeps on deleting messages while I've already released the delete key and was trying to read the next message.

Regular unpatched 2.6.22.1 though and I've recently changed too many things to be certain of anything. My X is the modular X.org 1.3.0 since not long, and the mailer showing the problem is Thunderbird 2.0.0.5, recently switched from 1.5.0.12. The problem was on my "I'll get to it" list but seems this could be a good thread to report it into. PS/2 keyboard...

Quoted 3 lines

It's extremely unusable.

It doesn't happen if i am on other kernel.

Very very sure?

Quoted 6 lines

I am ready to provide any requested information to find issue. If its not kernel related thing - i am very sorry, i just don't know where to start to trace whats wrong.

I am on gentoo and use latest xorg server. Keyboard driver is "KBD". Mouse use "evdev" driver.

Rene.

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

#3 lore raw
Quoted 4 lines · Ingo Molnar

another reaction below in this thread reported kbd problems in vanilla 2.6.22.1 as well. What is the X versions, etc.? Does the problem go away if the X kbd driver selection is tweaked to a simpler model, say:

Or perhaps just test with CFS reverted? It really sounds more like something independent from the scheduler. Perhaps someone could put a CFS-revert-for-testing patch on ftp somewhere. It's only a few git changesets so shouldn't be too difficult to do.

-Andi

#4 lore raw
Quoted 8 lines · Andi Kleen
Quoted 1 line · Ingo Molnar

another reaction below in this thread reported kbd problems in vanilla 2.6.22.1 as well. What is the X versions, etc.? Does the problem go away if the X kbd driver selection is tweaked to a simpler model, say:

Or perhaps just test with CFS reverted? It really sounds more like something independent from the scheduler. Perhaps someone could put a CFS-revert-for-testing patch on ftp somewhere. It's only a few git changesets so shouldn't be too difficult to do.

Teresa was already using 2.6.22.1, with CFS (v19.1) patched in, so reverting that would be a matter of patching it out again. She said she wasn't seeing trouble on other kernels though.

I was the one who also saw keyboard trouble on 2.6.22.1 without CFS by the way -- and haven't (yet) seen any anymore since patching _in_ CFS v19.1...

Rene.

#5 lore raw
Quoted 13 lines · Rene Herman
Quoted 10 lines · Andi Kleen
Quoted 3 lines · Ingo Molnar

another reaction below in this thread reported kbd problems in vanilla 2.6.22.1 as well. What is the X versions, etc.? Does the problem go away if the X kbd driver selection is tweaked to a simpler model, say:

Or perhaps just test with CFS reverted? It really sounds more like something independent from the scheduler. Perhaps someone could put a CFS-revert-for-testing patch on ftp somewhere. It's only a few git changesets so shouldn't be too difficult to do.

Quoted 6 lines

Teresa was already using 2.6.22.1, with CFS (v19.1) patched in, so reverting that would be a matter of patching it out again. She said she wasn't seeing trouble on other kernels though.

I was the one who also saw keyboard trouble on 2.6.22.1 without CFS by the way -- and haven't (yet) seen any anymore since patching _in_ CFS v19.1...

Could the people who had or have keyboard problems try out Dmitry's input locking patches, and see if those help: http://lkml.org/lkml/2007/7/24/17
And if it does, report it to him, either there or here after CCing him.

Greetings,

Indan

#6 lore raw
Quoted 6 lines · Indan Zupancic
Quoted 4 lines · Rene Herman

Teresa was already using 2.6.22.1, with CFS (v19.1) patched in, so reverting that would be a matter of patching it out again. She said she wasn't seeing trouble on other kernels though.

Quoted 6 lines
Quoted 2 lines

I was the one who also saw keyboard trouble on 2.6.22.1 without CFS by the way -- and haven't (yet) seen any anymore since patching _in_ CFS v19.1...

Could the people who had or have keyboard problems try out Dmitry's input locking patches, and see if those help: http://lkml.org/lkml/2007/7/24/17
And if it does, report it to him, either there or here after CCing him.

I'm having trouble reproducing this at will -- it sounded as though Teresa had less trouble at least originally. Teresa?

Rene.

#7 lore raw
Quoted 11 lines · Rene Herman
Quoted 9 lines · Indan Zupancic
Quoted 6 lines · Rene Herman

Teresa was already using 2.6.22.1, with CFS (v19.1) patched in, so reverting that would be a matter of patching it out again. She said she wasn't seeing trouble on other kernels though.

I was the one who also saw keyboard trouble on 2.6.22.1 without CFS by the way -- and haven't (yet) seen any anymore since patching _in_ CFS v19.1...

Quoted 6 lines
Quoted 3 lines

Could the people who had or have keyboard problems try out Dmitry's input locking patches, and see if those help: http://lkml.org/lkml/2007/7/24/17
And if it does, report it to him, either there or here after CCing him.

I'm having trouble reproducing this at will -- it sounded as though Teresa had less trouble at least originally. Teresa?

Same here, so testing is a pain. I was more thinking about running a patched version for a day or longer and see if it happened or not. For me it happened irregularly, but definitely a few times a day.

(It might help if people report what they're using. So, for what it's worth, I'm using a PS/2 keyboard with xorg 1.2.0 with the "keyboard" driver.)

Greetings,

Indan

#8 lore raw
Quoted 17 lines · Indan Zupancic
Quoted 7 lines · Rene Herman
Quoted 5 lines · Indan Zupancic
Quoted 4 lines · Rene Herman

Teresa was already using 2.6.22.1, with CFS (v19.1) patched in, so reverting that would be a matter of patching it out again. She said she wasn't seeing trouble on other kernels though.

Quoted 7 lines
Quoted 5 lines
Quoted 2 lines

I was the one who also saw keyboard trouble on 2.6.22.1 without CFS by the way -- and haven't (yet) seen any anymore since patching _in_ CFS v19.1...

Could the people who had or have keyboard problems try out Dmitry's input locking patches, and see if those help: http://lkml.org/lkml/2007/7/24/17
And if it does, report it to him, either there or here after CCing him.

I'm having trouble reproducing this at will -- it sounded as though Teresa had less trouble at least originally. Teresa?

Quoted 6 lines

Same here, so testing is a pain. I was more thinking about running a patched version for a day or longer and see if it happened or not. For me it happened irregularly, but definitely a few times a day.

(It might help if people report what they're using. So, for what it's worth, I'm using a PS/2 keyboard with xorg 1.2.0 with the "keyboard" driver.)

PS/2 keyboard, X.org 1.3.0 (version is at the top of /var/log/Xorg.0.log, or wherever your distribution puts that log), and this xkb config (in /etc/X11/xorg.conf):

Section "InputDevice"

        Identifier      "Keyboard0"
        Driver          "kbd"
        Option          "XkbModel"      "pc104"
        Option          "XkbLayout"     "us"
        Option          "XkbVariant"    "euro"
        Option          "XkbOptions"    "compose:rctrl"

EndSection

(you can forget that XkbVariant -- it's not yet in most distributions).

Problem as observed here consists of a sporadically stuck delete key in Thunderbird 2.0.0.5, using kernel 2.6.22.1 vanilla -- not yet observed using 2.6.22.1-cfs-v19.1

Rene.

#9 lore raw
In reply to Indan Zupancic
Quoted 2 lines · Indan Zupancic
Quoted 2 lines

I'm having trouble reproducing this at will -- it sounded as though Teresa had less trouble at least originally. Teresa?

I am at work actually, so i cant test anything. End of this week i got to vacation, so i possibly can return testing in 1 month again.

Quoted 3 lines

Same here, so testing is a pain. I was more thinking about running a patched version for a day or longer and see if it happened or not. For me it happened irregularly, but definitely a few times a day.

Problem is, that i can't reproduce it after last reboot. Can this be different on each reboot ? Any idea ?

About another kernel: i only had -ck kernels since last year or more, i can remember i had same behavier atleast 6 month ago, or maybe more. But it wasn't that hard and it dissapiered self after some time.

Now on first run of 2.6.22.1-cfs-v19.1 i got it again. And realy bad. It was even very hard to write report mail. I did reboot, day after that it was back, but not that hard. And now rebooted yesterday and cant reproduce it at all.

Quoted 2 lines

(It might help if people report what they're using. So, for what it's worth, I'm using a PS/2 keyboard with xorg 1.2.0 with the "keyboard" driver.)

PS/2 keybord with latest xorg from gentoo ~amd64 here.

I use plaine settings for xorg, like Ingo told me to try.

Signature · 1 line
Teresa
#10 lore raw
Quoted 4 lines · On 08/01/2007 03:33 PM, <:::.. TeresaII ..:::> wrote:

Problem is, that i can't reproduce it after last reboot. Can this be different on each reboot ? Any idea ?

About another kernel: i only had -ck kernels since last year or more, i can remember i had same behavier atleast 6 month ago, or maybe more. But it wasn't that hard and it dissapiered self after some time.

Hmm.

Quoted 1 line

Now on first run of 2.6.22.1-cfs-v19.1 i got it again. And realy bad. It was even very hard to write report mail.

I notice by the way that you are also using Thunderbird 2.0 -- that was my own suspect; I had just switched from Thunderbird 1.5.

Quoted 2 lines

I did reboot, day after that it was back, but not that hard. And now rebooted yesterday and cant reproduce it at all.

I believe that at least for now we can safely remove this from any list of possible CFS regresssions. I'll sit on it and see what I can find out when/if my stuck delete happens again.

Indan -- under what circumstances/programs did you experience what?

Rene.

#11 lore raw

У ср, 2007-08-01 у 15:58 +0200, Rene Herman пише:

Quoted 2 lines

I notice by the way that you are also using Thunderbird 2.0 -- that was my own suspect; I had just switched from Thunderbird 1.5.

I was at work, on windows maschine. At home, where my linux maschine is running i use evolution. But that have nothing to do with that.

I also doesnt have any "Backspace" keys stoked. Problem i discovered is only with SHIFT/CTRL and some times "Right" Buttons.

Issue with "right" ("->") button is realy funny. If you type this button fast enough it works like double pressed. And if you type it fast few times (with releasing it) it looks same like if button would be never released. It doesn't happend with "left"("<-") button :)

Quoted 6 lines
Quoted 2 lines

I did reboot, day after that it was back, but not that hard. And now rebooted yesterday and cant reproduce it at all.

I believe that at least for now we can safely remove this from any list of possible CFS regresssions. I'll sit on it and see what I can find out when/if my stuck delete happens again.

As i sad. I wasn't sure if its kernel related at all, it just was worse first time i booted cfs-v19.1 patch. Now i cant reproduce it even anymore :)

Maybe thats xorg bug. Or maybe its different on booting. Anyway HDD IO or CPU usage or System Load doesn't seems to have influence on it.

Signature · 2 lines
Teresa Kalitowska
jid: [email protected]
#12 lore raw
Quoted 3 lines · On 8/2/07, <::.. Teresa_II ..::> <[email protected]> wrote:

As i sad. I wasn't sure if its kernel related at all, it just was worse first time i booted cfs-v19.1 patch. Now i cant reproduce it even anymore :)

Hi Teresa,

Are you sure its not just a setting in Gnome/KDE for accessibility? That tends to do crazy things like making control keys sticky...

Signature · 1 line
Matt
#13 lore raw

У чт, 2007-08-02 у 01:00 +1000, Matthew Hawkins пише:

Quoted 2 lines

Are you sure its not just a setting in Gnome/KDE for accessibility? That tends to do crazy things like making control keys sticky...

Yes, i use gnome, and all keyboard layouts are set in gnome, also group switching for layouts. But accessibility is switched completely off.

Signature · 2 lines
Teresa Kalitowska
jid: [email protected]
#14 lore raw

Hi all,

just a datapoint FWIW

Quoted 8 lines · On Wed, 01 Aug 2007 17:53:32 +0200 "<::.. Teresa_II ..::>" <[email protected]>…

У чт, 2007-08-02 у 01:00 +1000, Matthew Hawkins пише:

Quoted 2 lines

Are you sure its not just a setting in Gnome/KDE for accessibility? That tends to do crazy things like making control keys sticky...

Yes, i use gnome, and all keyboard layouts are set in gnome, also group switching for layouts. But accessibility is switched completely off.

This reminds me I had something similar happening about a year or so ago
running with Debian's unstable xorg and gnome.

The keys would stick for a short time (no, no coffee spilled on the keyboard,
I can assure you ;-), the most annoying one being the delete key. Nothing deterministic, it could happen anytime (or not at all for several days).

As a final resort, I upgraded the xorg and gnome packages - problem solved.

Sébastien.

#15 lore raw
Quoted 9 lines · Sébastien Dugué

This reminds me I had something similar happening about a year or so ago running with Debian's unstable xorg and gnome.

The keys would stick for a short time (no, no coffee spilled on the keyboard, I can assure you ;-), the most annoying one being the delete key. Nothing deterministic, it could happen anytime (or not at all for several days).

As a final resort, I upgraded the xorg and gnome packages - problem solved.

Thank you. That "about a year" sounds possible; slackware has been one of the last to switch to modular xorg and until recently I was still running the old 6.9 series, so it could be xorg...

Rene.

#16 lore raw
Quoted 10 lines · Rene Herman
Quoted 8 lines · S�bastien Dugu�

This reminds me I had something similar happening about a year or so ago running with Debian's unstable xorg and gnome.

The keys would stick for a short time (no, no coffee spilled on the keyboard, I can assure you ;-), the most annoying one being the delete key. Nothing deterministic, it could happen anytime (or not at all for several days).

Quoted 6 lines
Quoted 2 lines

As a final resort, I upgraded the xorg and gnome packages - problem solved.

Thank you. That "about a year" sounds possible; slackware has been one of the last to switch to modular xorg and until recently I was still running the old 6.9 series, so it could be xorg...

Well, I'm running xorg since 2004, and hadn't the sticking key problem until a few months ago, half a year max. I didn't install a new xorg version just before the problem went away, and I'm using Fluxbox.

Oh well, the problem's gone now, no matter how or why.

Greetings,

Indan

#17 lore raw
In reply to Rene Herman
Quoted 17 lines · Rene Herman
Quoted 6 lines · On 08/01/2007 03:33 PM, <:::.. TeresaII ..:::> wrote:

Problem is, that i can't reproduce it after last reboot. Can this be different on each reboot ? Any idea ?

About another kernel: i only had -ck kernels since last year or more, i can remember i had same behavier atleast 6 month ago, or maybe more. But it wasn't that hard and it dissapiered self after some time.

Hmm.

Quoted 2 lines

Now on first run of 2.6.22.1-cfs-v19.1 i got it again. And realy bad. It was even very hard to write report mail.

I notice by the way that you are also using Thunderbird 2.0 -- that was my own suspect; I had just switched from Thunderbird 1.5.

Quoted 8 lines
Quoted 2 lines

I did reboot, day after that it was back, but not that hard. And now rebooted yesterday and cant reproduce it at all.

I believe that at least for now we can safely remove this from any list of possible CFS regresssions. I'll sit on it and see what I can find out when/if my stuck delete happens again.

Indan -- under what circumstances/programs did you experience what?

Different programs, different keys "stuck", so hard to tell. The amount the pressed key is repeated differs too, from double to hundreds.

As an apparently separate problem my mouse warps once in a while too, and that's not totally fixed yet, neither with that locking patch. Throw in that two of my USB ports stopped working after a rare power outage, I didn't totally trust my hardware any more, hence my initial reluctance to report these problems.

We could have independent problems with more or less the same symptoms, at least Teresa's problem seems much worse than ours, and if you and Ingo only experience a stuck delete key, it might be something else.

Greetings,

Indan

#18 lore raw
Quoted 4 lines · Indan Zupancic

We could have independent problems with more or less the same symptoms, at least Teresa's problem seems much worse than ours, and if you and Ingo only experience a stuck delete key, it might be something else.

i experienced nothing in this area so far - if i experienced anything of the likes it would be long fixed! =B-)

Ingo

#19 lore raw
Quoted 5 lines · Ingo Molnar
Quoted 1 line · Indan Zupancic

We could have independent problems with more or less the same symptoms, at least Teresa's problem seems much worse than ours, and if you and Ingo only experience a stuck delete key, it might be something else.

i experienced nothing in this area so far - if i experienced anything of the likes it would be long fixed! =B-)

Show-off...

Rene.

#20 lore raw
Quoted 9 lines · Rene Herman
Quoted 7 lines · Ingo Molnar
Quoted 4 lines · Indan Zupancic

We could have independent problems with more or less the same symptoms, at least Teresa's problem seems much worse than ours, and if you and Ingo only experience a stuck delete key, it might be something else.

Quoted 4 lines
Quoted 2 lines

i experienced nothing in this area so far - if i experienced anything of the likes it would be long fixed! =B-)

Show-off...

ok, gotta qualify that with: '... and if it were easily reproducible and if the bug was in my code'. Much less of a show-off i guess? :)

Ingo

#21 lore raw
Quoted 11 lines · Ingo Molnar
Quoted 7 lines · Rene Herman
Quoted 4 lines · Ingo Molnar
Quoted 1 line · Indan Zupancic

We could have independent problems with more or less the same symptoms, at least Teresa's problem seems much worse than ours, and if you and Ingo only experience a stuck delete key, it might be something else.

i experienced nothing in this area so far - if i experienced anything of the likes it would be long fixed! =B-)

Show-off...

ok, gotta qualify that with: '... and if it were easily reproducible and if the bug was in my code'. Much less of a show-off i guess? :)

Much better indeed :-) Haven't hit the problem yet today...

Rene.

#22 lore raw
In reply to Ingo Molnar
Quoted 10 lines · Ingo Molnar
Quoted 4 lines · Indan Zupancic

We could have independent problems with more or less the same symptoms, at least Teresa's problem seems much worse than ours, and if you and Ingo only experience a stuck delete key, it might be something else.

i experienced nothing in this area so far - if i experienced anything of the likes it would be long fixed! =B-)

Ah, indeed, you appended Rene's email without too many new lines, so I thought your wrote it (http://lkml.org/lkml/2007/8/1/4).

All right, how would you debug it? Give us some insight in how to solve hard to trigger, happens at most only a few times a day, annoying input bug? I thought the mouse warping was fixed after 23-rc1 and that input locking patch, but alas, the third day it happened again.

I would be happy just to know if it's bad hardware or a software bug, I don't ask for too much, do I?

Greetings,

Indan

#23 lore raw
Quoted 5 lines · Indan Zupancic

All right, how would you debug it? Give us some insight in how to solve hard to trigger, happens at most only a few times a day, annoying input bug? I thought the mouse warping was fixed after 23-rc1 and that input locking patch, but alas, the third day it happened again.

i've got no idea how to debug such input bugs best, but, as a starting point, i've Cc:-ed the current maintainer of the input subsystem :-)

Ingo

#24 lore raw
Quoted 5 lines · Ingo Molnar
Quoted 1 line · Indan Zupancic

All right, how would you debug it? Give us some insight in how to solve hard to trigger, happens at most only a few times a day, annoying input bug? I thought the mouse warping was fixed after 23-rc1 and that input locking patch, but alas, the third day it happened again.

i've got no idea how to debug such input bugs best, but, as a starting point, i've Cc:-ed the current maintainer of the input subsystem :-)

FWIW, I haven't experienced my "stuck delete" key anymore -- "since using CFS v19.1", but that might very well just be coincedence. If anyone wants/needs me to, I'll try to debug it, but for now I seem to be fine again.

Now all I need to know is whether or not moving thunderbird's windows around is expected to leave such an enormous non-repainting visual trail on the screen...

I believe I'm concluding that I'm not all together fond of the "new" modular X.org.

Rene.

#25 lore raw
Quoted 10 lines · Rene Herman
Quoted 8 lines · Ingo Molnar
Quoted 5 lines · Indan Zupancic

All right, how would you debug it? Give us some insight in how to solve hard to trigger, happens at most only a few times a day, annoying input bug? I thought the mouse warping was fixed after 23-rc1 and that input locking patch, but alas, the third day it happened again.

Quoted 6 lines
Quoted 2 lines

i've got no idea how to debug such input bugs best, but, as a starting point, i've Cc:-ed the current maintainer of the input subsystem :-)

FWIW, I haven't experienced my "stuck delete" key anymore -- "since using CFS v19.1", but that might very well just be coincedence. If anyone wants/needs me to, I'll try to debug it, but for now I seem to be fine again.

No stuck keys and no warping mouse seen yet since last time (currently running 2.6.23-rc2), so both problems are, for all practical purposes, fixed.

Quoted 6 lines

Now all I need to know is whether or not moving thunderbird's windows around is expected to leave such an enormous non-repainting visual trail on the screen...

I believe I'm concluding that I'm not all together fond of the "new" modular X.org.

I don't think it's fair to blame Xorg for that, except if changing back to an older version fixes it. If Thunderbird is anything comparable to Firefox, then it's most likely the generic slowdown of that program that might have made it worse. If it's Xorg then it's probably something driver/settings related I guess, best to ask on their mailinglist for that.

Anyway, if you want to experience a shock, try enabling xcomposite and run xcompmgr -a or something. It's so wonderfully smooth (even though some rendering seems slower than before, it's really worth it for me).

Greetings,

Indan

#26 lore raw

On 08/06/2007 05:19 PM, Indan Zupancic wrote:

[ Yes, OT, I'll shelve it after this ]

Quoted 3 lines

Anyway, if you want to experience a shock, try enabling xcomposite and run xcompmgr -a or something. It's so wonderfully smooth (even though some rendering seems slower than before, it's really worth it for me).

Don't believe I can -- using a Matrox Millenium G550 (driver mga) and nothing I do seems to have an effect.

Rene.

#27 lore raw

OT: enabling Xcomposite [was: Linux Kernel cfs scheduler and xorg kbd]

Quoted 3 lines · Rene Herman

On 08/06/2007 05:19 PM, Indan Zupancic wrote:

[ Yes, OT, I'll shelve it after this ]

Me too, and I wonder if we should've dropped a bunch of CCs...

Quoted 7 lines
Quoted 4 lines

Anyway, if you want to experience a shock, try enabling xcomposite and run xcompmgr -a or something. It's so wonderfully smooth (even though some rendering seems slower than before, it's really worth it for me).

Don't believe I can -- using a Matrox Millenium G550 (driver mga) and nothing I do seems to have an effect.

Well, as far as I know adding the following to your xorg.conf should be enough:

Section "Extensions"
Option "Composite" "enable"
EndSection

and then to enable it at runtime start a composite manager.

Reading the mga manpage it seems you need to add

Option "AccelMethod" "EXA"

to the driver section. (Though it says EXA is possibly unstable.)

Run "xdpyinfo | grep Composite" to see if it works or not.

But as you said, getting off-topic, so if the above doesn't work sent a private mail, or go to the xorg mailinglist.

Greetings,

Indan

#28 lore raw
In reply to Indan Zupancic
Quoted 5 lines · Indan Zupancic

Different programs, different keys "stuck", so hard to tell. The amount the pressed key is repeated differs too, from double to hundreds.

As an apparently separate problem my mouse warps once in a while too, and that's not totally fixed yet, neither with that locking patch. Throw in that two of my USB ports stopped working after a rare power outage, I didn't totally trust my hardware any more, hence my initial reluctance to report these problems.

We could have independent problems with more or less the same symptoms, at least Teresa's problem seems much worse than ours, and if you and Ingo only experience a stuck delete key, it might be something else.

Well I've just seen an FC box (6? 7?) running a 2.6.18-ish which reliably experiences the same problem (randomly stuck keys), so I doubt very much it's CFS. Basically keys will somewhat randomly repeat 6-7 times, mainly depending on how fast you're typing (the intern using the box typed somewhat slowly and didn't trigger it much, but when I borrowed his keyboard I could reliably trigger it once every 5 sec or so. Unfortunately I don't have access to that box anymore, so I can't really debug the problem.

Cheers,
Kyle Moffett

#29 lore raw
Quoted 11 lines · Kyle Moffett
Quoted 10 lines · Indan Zupancic

Different programs, different keys "stuck", so hard to tell. The amount the pressed key is repeated differs too, from double to hundreds.

As an apparently separate problem my mouse warps once in a while too, and that's not totally fixed yet, neither with that locking patch. Throw in that two of my USB ports stopped working after a rare power outage, I didn't totally trust my hardware any more, hence my initial reluctance to report these problems.

Quoted 4 lines
Quoted 4 lines

We could have independent problems with more or less the same symptoms, at least Teresa's problem seems much worse than ours, and if you and Ingo only experience a stuck delete key, it might be something else.

the last time I saw something like this it was a BIOS that kept doing USB->PS/2 emulation even when the kernel was running.. that caused great havoc somehow...

those who are seeing this... is it all on a USB keyboard? If so, is the bios set to do PS/2 emulation (sometimes called USB legacy emulation) ??

#30 lore raw
Quoted 18 lines · Arjan van de Ven
Quoted 11 lines · Kyle Moffett
Quoted 10 lines · Indan Zupancic

Different programs, different keys "stuck", so hard to tell. The amount the pressed key is repeated differs too, from double to hundreds.

As an apparently separate problem my mouse warps once in a while too, and that's not totally fixed yet, neither with that locking patch. Throw in that two of my USB ports stopped working after a rare power outage, I didn't totally trust my hardware any more, hence my initial reluctance to report these problems.

Quoted 4 lines
Quoted 4 lines

We could have independent problems with more or less the same symptoms, at least Teresa's problem seems much worse than ours, and if you and Ingo only experience a stuck delete key, it might be something else.

Quoted 6 lines

the last time I saw something like this it was a BIOS that kept doing USB->PS/2 emulation even when the kernel was running.. that caused great havoc somehow...

those who are seeing this... is it all on a USB keyboard? If so, is the bios set to do PS/2 emulation (sometimes called USB legacy emulation) ??

No, using PS/2 keyboard and mouse here, and made sure to disable USB legacy emulation in the BIOS ages ago.

But for me the problems are gone now, since 2.6.23-rc1 (I don't think it was fixed by CFS though, just in the same timeframe).

Greetings,

Indan

#31 lore raw
In reply to Arjan van de Ven

On 08/07/2007 08:05 AM, Arjan van de Ven wrote:

[ stuck keys ]

Quoted 3 lines

the last time I saw something like this it was a BIOS that kept doing USB->PS/2 emulation even when the kernel was running.. that caused great havoc somehow...

those who are seeing this... is it all on a USB keyboard? If so, is the bios set to do PS/2 emulation (sometimes called USB legacy emulation) ??

No, as far as reported into this thread, it's all been PS/2. We can leave this thread be though. It's nothing to do with CFS and is a non-debuggable problem sofar unfortunately. Although very present at the start of this thread, I haven't experienced my stuck delete key since. Teresa and Indan have the same problem (well, "non-problem" that is). If it's still lurking somewhere, we'll see it resurface at some point.

Rene.

#32 lore raw
Quoted 2 lines · Rene Herman

..
No, as far as reported into this thread, it's all been PS/2. We can leave this thread be though. It's nothing to do with CFS and is a non-debuggable problem sofar unfortunately. Although very present at the start of this thread, I haven't experienced my stuck delete key since. Teresa and Indan have the same problem (well, "non-problem" that is). If it's still lurking somewhere, we'll see it resurface at some point.

My Dell X1 notebook has long suffered from this type of problem, but disabling the "keyboard accessibility" features in KDE seems to have greatly lessened the occurances.

#33 lore raw
In reply to <::.. Teresa_II ..::>

У пн, 2007-07-30 у 22:25 +0200, <::.. Teresa_II ..::> пише:

Quoted 1 line

I am ready to provide any requested information to find issue.

Interesting Thing is that this seems to happen only on system idle. If system is under load it behave correct.
And sometimes it doesn't appear even if system is idle.
Now as i write this email it works mostly like it should. Yesterday i was much worse. I didn't rebooted since this time. Application set i run is mostly the same for whole time.
So any idea where to start to look ?

Also how long SHIFT, or CTRL was pressed have matter. Long pressed special button cause longer release time.
Sorry if its little bit bad described. I don't know how better to explain how it looks.

Signature · 2 lines
Teresa Kalitowska
jid: [email protected]
#34 lore raw
In reply to <::.. Teresa_II ..::>

У пн, 2007-07-30 у 22:25 +0200, <::.. Teresa_II ..::> пише:

Ok, hier is small testing:
I started gcc emerge. System load is between 1.47 and 1.61 I use SMP enabled 2.6.22.1-cfs-v19.1 kernel on ADM64 X2.

In gnome-terminal window i just type fast "d" and until i count 1,2 i hold SHIFT, than i release SHIFT, count 1,2 and than press SHIFT again. I cound 1,2,1,2,1,2 as i type "d" button. That's the string i get:

DDDDDDDdDDDddDDddDDddDDDdDDDdDDDdDDDddDDddDDDdDDDddDDDdDDDdDDDDDDDdDDDDDDDDDDDdDDD

Its anyway different as expected.

I don't have better idea how to show what my problem is.

Signature · 2 lines
Teresa Kalitowska
jid: [email protected]
#35 lore raw
Quoted 10 lines · * <::.. Teresa_II ..::> <[email protected]> wrote:

У пн, 2007-07-30 у 22:25 +0200, <::.. Teresa_II ..::> пише:

Ok, hier is small testing: I started gcc emerge. System load is between 1.47 and 1.61 I use SMP enabled 2.6.22.1-cfs-v19.1 kernel on ADM64 X2.

In gnome-terminal window i just type fast "d" and until i count 1,2 i hold SHIFT, than i release SHIFT, count 1,2 and than press SHIFT again. I cound 1,2,1,2,1,2 as i type "d" button. That's the string i get:

hm, do you have CONFIG_NO_HZ=y? If yes, could you disable it? Plus can you see the same problem on v2.6.23-rc1 as well (if that kernel boots fine for you).

Ingo

#36 lore raw

У вт, 2007-07-31 у 16:27 +0200, Ingo Molnar пише:

Quoted 8 lines
Quoted 4 lines

In gnome-terminal window i just type fast "d" and until i count 1,2 i hold SHIFT, than i release SHIFT, count 1,2 and than press SHIFT again. I cound 1,2,1,2,1,2 as i type "d" button. That's the string i get:

hm, do you have CONFIG_NO_HZ=y? If yes, could you disable it? Plus can you see the same problem on v2.6.23-rc1 as well (if that kernel boots fine for you).

# zcat /proc/config.gz |grep "_HZ"
# CONFIG_HZ_100 is not set
# CONFIG_HZ_250 is not set
# CONFIG_HZ_300 is not set
CONFIG_HZ_1000=y
CONFIG_HZ=1000

Hmm, i will try 2.6.23-rc1 soon.

Signature · 2 lines
Teresa Kalitowska
jid: [email protected]
#37 lore raw
In reply to <::.. Teresa_II ..::>
Quoted 11 lines · On 07/30/2007 10:25 PM, <::.. Teresa_II ..::> wrote:

first of all i am not sure if its scheduler issue. But i didn't found anything about that in google related to something else.

So i tried 2.6.22.1 kernel with new cfs scheduler v19.1 from Ingo page. Previous i used 2.6.22-ck1.

After 24 hours running i get strange behavior in Xorg. It act slower
with my special keys, like SHIFT, or CTRL. So if i hold SHIFT and press
")" and than will press enter, i release SHIIFT and press ENTER it
doesn't create new line. Even if i start write a Sentence or word with
"SHIFTed" char it makes next char in high case also. Like this: "HAllo" even if i type "Hallo".

For a few days now I have been experiencing a "stuck delete" key while reading mail -- I run through the mail with "delete" and every once in a while it just keeps on deleting messages while I've already released the delete key and was trying to read the next message.

Regular unpatched 2.6.22.1 though and I've recently changed too many things to be certain of anything. My X is the modular X.org 1.3.0 since not long, and the mailer showing the problem is Thunderbird 2.0.0.5, recently switched from 1.5.0.12. The problem was on my "I'll get to it" list but seems this could be a good thread to report it into. PS/2 keyboard...

Quoted 3 lines

It's extremely unusable.

It doesn't happen if i am on other kernel.

Very very sure?

Quoted 5 lines

I am ready to provide any requested information to find issue. If its not kernel related thing - i am very sorry, i just don't know where to start to trace whats wrong.

I am on gentoo and use latest xorg server. Keyboard driver is "KBD". Mouse use "evdev" driver.

Rene.