Re: F11-0.88 os260py
Yes, I talked about automatic suspend. However the WLAN was not connected to any access points and it seems that Bernie talked about dropped wireless connections in the release notes as far as I know. Paul Fox wrote: > noiseehc wrote: > > I have just installed it onto my XO-1 (C1 model with the old style > > touchpad) without anything plugged into it. > > > > It seems to work but the suspend/resume is "interesting". > > you're referring to automatic suspend, in the presence of wireless? > > i think bernie understated this bug in the release notes. the > wlan driver on the 2.6.31 OLPC kernels is very broken right now. > your symptoms don't shock me. (though i've never tried the rotate > button when the system has hung.) > > (if, on the other hand, you have automatic suspend disabled, then > perhaps something else is going on.) > > paul > > > > > Sometimes it goes to suspend but after that I cannot wake it up except > > with the screen rotation button. I press the button and the XO draws a > > lot of rotated images (10-20) and then it stops. After this I can wake > > up the laptop when it suspends again by the touchpad or keyboard. What > > is interesting is that sometimes when it wakes up the screen backlight > > dims for a fraction of the second. Is it the normal behavior? I mean it > > seems that the hardware suspended before the backlight could be dimmed. > > What is the expected behavior? > > > > Another strange thing is that when I use Read to read a pdf file and > > rotate the screen 90 degree then the top of the screen (toolbar) gets > > black or garbage and when I move the mouse cursor over that area it > > redraws the toolbar with some garbage. > > > > ___ > > Devel mailing list > > Devel@lists.laptop.org > > http://lists.laptop.org/listinfo/devel > > =- > paul fox, p...@laptop.org > ___ > Devel mailing list > Devel@lists.laptop.org > http://lists.laptop.org/listinfo/devel > > > ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: F11-0.88 os260py
noiseehc wrote: > I have just installed it onto my XO-1 (C1 model with the old style > touchpad) without anything plugged into it. > > It seems to work but the suspend/resume is "interesting". you're referring to automatic suspend, in the presence of wireless? i think bernie understated this bug in the release notes. the wlan driver on the 2.6.31 OLPC kernels is very broken right now. your symptoms don't shock me. (though i've never tried the rotate button when the system has hung.) (if, on the other hand, you have automatic suspend disabled, then perhaps something else is going on.) paul > > Sometimes it goes to suspend but after that I cannot wake it up except > with the screen rotation button. I press the button and the XO draws a > lot of rotated images (10-20) and then it stops. After this I can wake > up the laptop when it suspends again by the touchpad or keyboard. What > is interesting is that sometimes when it wakes up the screen backlight > dims for a fraction of the second. Is it the normal behavior? I mean it > seems that the hardware suspended before the backlight could be dimmed. > What is the expected behavior? > > Another strange thing is that when I use Read to read a pdf file and > rotate the screen 90 degree then the top of the screen (toolbar) gets > black or garbage and when I move the mouse cursor over that area it > redraws the toolbar with some garbage. > > ___ > Devel mailing list > Devel@lists.laptop.org > http://lists.laptop.org/listinfo/devel =- paul fox, p...@laptop.org ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: F11-0.88 os260py
I have just installed it onto my XO-1 (C1 model with the old style touchpad) without anything plugged into it. It seems to work but the suspend/resume is "interesting". Sometimes it goes to suspend but after that I cannot wake it up except with the screen rotation button. I press the button and the XO draws a lot of rotated images (10-20) and then it stops. After this I can wake up the laptop when it suspends again by the touchpad or keyboard. What is interesting is that sometimes when it wakes up the screen backlight dims for a fraction of the second. Is it the normal behavior? I mean it seems that the hardware suspended before the backlight could be dimmed. What is the expected behavior? Another strange thing is that when I use Read to read a pdf file and rotate the screen 90 degree then the top of the screen (toolbar) gets black or garbage and when I move the mouse cursor over that area it redraws the toolbar with some garbage. ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: libertas vs. suspend (was: [IAEP] F11-0.88 os260py)
sascha wrote: > > Excerpts from Bernie Innocenti's message of Fri Jun 18 03:26:56 + 2010: > > > * Disconnects from network on power save. This is a long-standing > >kernel bug. In 0.84, we "fixed" it by disabling power management. > I'm not sure what you're seeing is really a kernel bug (unless you don't > ship the latest OLPC kernel). > It works fine for me even across several hundred suspend/resume cycles > (it gets woken about once per 1-2 minutes by IGMP queries from the AP), > though this is with an otherwise unused WLAN, so rekeyings are infrequent. > And rekeyings might be exactly what's causing you trouble: According to > Dan Williams, the libertas chip doesn't wake us for these [1]. I haven't > had time to verify that yet. > Upgrading to the latest NetworkManager version might also improve things > a lot. sascha confirmed on IRC that the above refers to XO-1.5 (and is good news on that front, of course). but bernie's referring to XO-1 behavior. i've been looking at the XO-1 issues recently, and my progress has been investigative, without a lot of actual discovery. :-) i think there are several issues, at least a couple of which have to do with races between suspend/resume and other driver activity. (e.g. #10176) paul =- paul fox, p...@laptop.org ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
libertas vs. suspend (was: [IAEP] F11-0.88 os260py)
Excerpts from Bernie Innocenti's message of Fri Jun 18 03:26:56 + 2010: > * Disconnects from network on power save. This is a long-standing >kernel bug. In 0.84, we "fixed" it by disabling power management. I'm not sure what you're seeing is really a kernel bug (unless you don't ship the latest OLPC kernel). It works fine for me even across several hundred suspend/resume cycles (it gets woken about once per 1-2 minutes by IGMP queries from the AP), though this is with an otherwise unused WLAN, so rekeyings are infrequent. And rekeyings might be exactly what's causing you trouble: According to Dan Williams, the libertas chip doesn't wake us for these [1]. I haven't had time to verify that yet. Upgrading to the latest NetworkManager version might also improve things a lot. Sascha [1] http://lists.infradead.org/pipermail/libertas-dev/2010-May/003107.html -- http://sascha.silbe.org/ http://www.infra-silbe.de/ signature.asc Description: PGP signature ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: [IAEP] F11-0.88 os260py
On Fri, Jun 18, 2010 at 05:26, Bernie Innocenti wrote: > This is the an interim release of Sugar 0.88 + GNOME series for XO-1. > Please refer to this page for the current status and download > instructions: > > http://wiki.sugarlabs.org/go/Deployment_Team/Sugar-0.88 > > > With this build, all the major regression relative to 0.84 are fixed. > The switch to Metacity resulted in a few hard issues on Fedora 11. There > are still several smaller bugs to work on. > > All the new features planned for this cycle are now merged, with the > notable exception of collections support for the activity updater, which > urgently needs a volunteer. At this point, it seems too late for the > much anticipated resurrection of Rainbow. Maybe next time. > > There may still be time to merge and stabilize a couple of minor > features, such as a button in the Network control panel to re-register > with the schoolserver. > > > == Changes relative to the previous release (os258py) == > > * Integrate journal backup and restore to USB (esteban, tch) > > * Integrate function for mounting a GNOME Documents folder in > the journal. Still presents some rough edges. (jasg) > > * Pull latest OS updates from upstream (fedora) > > > == Bugs fixed in this release == > > * Make ALT-TAB work. It was caused by a bug with XUngrabKey() > in the old Xorg server of Fedora 11, but we worked it around > in Metacity. We spent a lot of time to figure it out! > (tch, bernie) > > * Fix windows staying in frame after startup. It turned out to be > a libwnck bug, solved by upgrading it to a more recent version > (aa, bernie) Excellent integration work! Before, it used to be the upstream developers who had to also debug on Fedora and sometimes on other distros, and as you said, it often takes a lot of time. It's great that this work has moved downstream! > * Fix fonts too big in all activities. This was a fallout of us > temporarily disabling sugar-settings-manager, due to another > bug (jasg). > > * CTRL-C/CTRL-V don't work in Turtle Blocks > (walter) > > * Sugar sometimes restart. My fault: I included in the build an > experimental Geode driver to check if it would now be safe to > re-enable some EXA acceleration. It wasn't. (bernie). > > > == Known bugs == > > We still have a bunch of less serious issues requiring attention: > > * Activity updater needs to support microformat and collections > > * Sometimes pulsing launch icon is delayed or absent (scheduling > issue? Sometimes I can see Sugar eating up 50% of the CPU when > starting an activity) If you can reproduce it, it would be great to gather some profile data during activity launch with this: ## Before launching ## profiler = cProfile.Profile() profiler.enable() ## After launch ## profiler.disable() profiler.dump_stats('/tmp/profile') It should tell what's Sugar up to. Cheers, Tomeu > * Font in Write is small and text does not fill the screen width. > Sugar 0.82 and 0.84 used "Dejavu Sans", 0.88 uses "Times New Roman", > which is smaller. (someone take it? pretty please?) > > * Disconnects from network on power save. This is a long-standing > kernel bug. In 0.84, we "fixed" it by disabling power management. > > * Social Calc contains a race condition that makes it fail to > start. Frankly, I would just ditch it, but teachers seem to like > it very much > > * More GNOME lock-down. See the thread with subject "GNOME and > protecting Sugar" on sugar-de...@. > > Thanks to anyone who could help us fixing these problems. > > -- > // Bernie Innocenti - http://codewiz.org/ > \X/ Sugar Labs - http://sugarlabs.org/ > > > > ___ > IAEP -- It's An Education Project (not a laptop project!) > i...@lists.sugarlabs.org > http://lists.sugarlabs.org/listinfo/iaep > ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
F11-0.88 os260py
This is the an interim release of Sugar 0.88 + GNOME series for XO-1. Please refer to this page for the current status and download instructions: http://wiki.sugarlabs.org/go/Deployment_Team/Sugar-0.88 With this build, all the major regression relative to 0.84 are fixed. The switch to Metacity resulted in a few hard issues on Fedora 11. There are still several smaller bugs to work on. All the new features planned for this cycle are now merged, with the notable exception of collections support for the activity updater, which urgently needs a volunteer. At this point, it seems too late for the much anticipated resurrection of Rainbow. Maybe next time. There may still be time to merge and stabilize a couple of minor features, such as a button in the Network control panel to re-register with the schoolserver. == Changes relative to the previous release (os258py) == * Integrate journal backup and restore to USB (esteban, tch) * Integrate function for mounting a GNOME Documents folder in the journal. Still presents some rough edges. (jasg) * Pull latest OS updates from upstream (fedora) == Bugs fixed in this release == * Make ALT-TAB work. It was caused by a bug with XUngrabKey() in the old Xorg server of Fedora 11, but we worked it around in Metacity. We spent a lot of time to figure it out! (tch, bernie) * Fix windows staying in frame after startup. It turned out to be a libwnck bug, solved by upgrading it to a more recent version (aa, bernie) * Fix fonts too big in all activities. This was a fallout of us temporarily disabling sugar-settings-manager, due to another bug (jasg). * CTRL-C/CTRL-V don't work in Turtle Blocks (walter) * Sugar sometimes restart. My fault: I included in the build an experimental Geode driver to check if it would now be safe to re-enable some EXA acceleration. It wasn't. (bernie). == Known bugs == We still have a bunch of less serious issues requiring attention: * Activity updater needs to support microformat and collections * Sometimes pulsing launch icon is delayed or absent (scheduling issue? Sometimes I can see Sugar eating up 50% of the CPU when starting an activity) * Font in Write is small and text does not fill the screen width. Sugar 0.82 and 0.84 used "Dejavu Sans", 0.88 uses "Times New Roman", which is smaller. (someone take it? pretty please?) * Disconnects from network on power save. This is a long-standing kernel bug. In 0.84, we "fixed" it by disabling power management. * Social Calc contains a race condition that makes it fail to start. Frankly, I would just ditch it, but teachers seem to like it very much * More GNOME lock-down. See the thread with subject "GNOME and protecting Sugar" on sugar-de...@. Thanks to anyone who could help us fixing these problems. -- // Bernie Innocenti - http://codewiz.org/ \X/ Sugar Labs - http://sugarlabs.org/ ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel