ReportingPage
Differences between revisions 60 and 62 (spanning 2 versions)
|
Size: 13152
Comment: Status report for week 08
|
Size: 2958
Comment:
|
| Deletions are marked like this. | Additions are marked like this. |
| Line 20: | Line 20: |
| '''Status report for week 04''' | '''Week 21 -27 Feb''' |
| Line 22: | Line 22: |
| Moblin Kernel and Driver Status | * Reviewing / sponsoring * Some more work on pimlico; dates is now in sync between upstream, Debian, and Ubuntu; tasks is ready to be synced, but this requires a seed change, so I prefer waiting until tomorrow's image build * Some discussion on versioning of kernel uploads to the ppa (versus in hardy); we should probably pin the ppa higher than hardy for all practical purposes in the end * Started some quickstart documentation on gettext usage * Started research on langpacks, documented how much they would gain us ..10 MB says Arne, andwhat we would need to do (we would have to do a lot); this also includes documenting how to output langpacks after the various freezes which I alre ady researched with misc people but need to summarize into executive steps |
| Line 24: | Line 30: |
| Kernel Support | * Adapt Open Office to be able export documents to PDF from command line. The reason behind this is to allow evince to view MS Office documents. During this process some specific changes needed to be done applying partially a patch from the official maintainer so it could build correctly on PPA. * Created a small launcher application called "treb" (short for trebuchet) that is basically an openfile-type dialog that will let the user select a file and will automatically filter it using Open Office and pass it to be read by Evince. |
| Line 26: | Line 35: |
| Alek rebased kernel-mid-2.6.24 to linux-2.6.24-rc8. Alek also fixed the issue - WIFI does not work on Samsung Q1 Ultra. The root cause is Ubuntu kernel uses a volatile tmpfs to hold restricted drivers, while Moblin Image Creator would lose the volatile during umount/mount cycle and thus would create wrong Modules.dep file. |
* I was able to patch the Firefox 3.0 beta2 chinese language pack and made it work with Midbrowser. |
| Line 29: | Line 37: |
| PSB / SLT HW Support | * Added all the internationalization support back into moblin-applets and enabled mobile-basic-flash to display multi-language icon texts. Essentially I regenerated a new po/POTFILES.in and po/POTFILES.skip with all the files that have and don’t have translatable strings. About 50% of them haven’t changed since th gnome-control-center fork, but that leaves quite a few with either fuzzy or no translations in all the languages available. |
| Line 31: | Line 41: |
| No changes this week. SDIO Feng successfully ported Marvell SDIO 8686 wifi card V9 version driver (with V9 firmware). The new driver was not put to Moblin yet. We expected lots of tests to verify the new ported driver. Dabney Thermal patches No changes this week. USB Client Drivers Moblin Media team is starting to integrate USBC media sync feature, mainly focus on DBus message handling exported by udev scripts. A driver upgrading was scheduled in WW09. PSB Gfx & Video Drivers Inuka upgraded PSB Gfx drivers to Beta6. Both kernel drivers and user space drivers were updated. Power and Performance Testing/Optimization Jay tested Moblin kernel-mid 2.6.22 S3 stability on Crown-beach D0 board. Jay found: 1. without nolapic_timer, the system could suspend/resume once, but for second cycle, the system would hang with post code 0200/0002, and system could resume normally after toggling lid switch couple of time. 2. with USB Ethernet dongle, after resume from S3, kernel would have 800+ wake up events per second. Sometimes, with the USB Ethernet dongle the system even refused to suspend to S3. 3. Sometimes USB mouse did not work properly after resume and USB Ethernet dongle could not get IP address. Jay would like to discuss this issue with Ubuntu kernel team during Intel/Ubuntu sprint. Grub fast boot/resume No changes this week. MidBrowser status * The Midbrowser is now officially using the FF3.0 beta 2 code base. A package has been uploaded to the gaston repository. * Team is busy working on Panning support and bug fixes. '''Status report for week 05''' Moblin Kernel and Driver Status Kernel Support Alek synced Moblin kernel 2.6.24 with formal 2.6.24 kernel. He also fixed linux-headers folder issue (/lib/modules/*/build should be a symbol link to /usr/src/linux-headers*). Alek started to append build number to Moblin kernel binary package and started to look into libata UDMA5 support on CRB D0 board. PSB / SLT HW Support Martin found HDA device extra interrupt issue: When intel HDA is working (play a wav file, for example), there are 50+ IRQ17 nobody handled interrupts happen. We suspected it maybe a hardware issue but current no enough evidence proved. SDIO Feng confirmed with C0 board the SDIO timeout waiting for interrupt error was gone. Dabney Thermal patches No changes this week. USB Client Drivers No changes this week. PSB Gfx & Video Drivers Did not get updates from Inuka. Power and Performance Testing/Optimization Did not get updates from Jay. Grub fast boot/resume Martin added 2.6.24 support to grub fast boot/resume patch and verified without nolapic_timer and with HT enabled, the 2.6.24 kernel fast resume very stable (tested with 20+ cycles). '''Status report for week 06''' Sprint week '''Status report for week 07''' Moblin Kernel and Driver Status Kernel Support For bug https://bugs.launchpad.net/moblin-kernel/+bug/186038 we confirmed S3 worked on Moblin 2.6.22 /2.6.24 kernel after getting a D0 board with BIOS 68. Inuka found screen distortion issue after S3, and already redirected the issue to Gfx team. Inuka also confirmed touch screen functional on CB but did not enabled calibration tool yet. Martin found a bug in kernel USB host controller driver -- the driver should disable the USB HC interrupt as early as possible during initialization if the BIOS enabled USB legacy support. The unexpected interrupt may cause OS to mask the "nobody cared" interrupt source and cause USB system working under polling mode. The patch is located at http://www.moblin.org/repos/?p=projects/kernel-mid-2.6.24.git;a=blob;f=debian/patches/debian/patches/0021-legacy-usb-int-fix.dpatch And Jay already pushed it to Ubuntu PPA repo. Feng worked out a workaround to avoid touchscreen double click issue when enabled external mouse support in xorg.conf, the root cause is the touch screen events are exported through input/event and mice interface simultaneously. PSB / SLT HW Support Alek enabled a workaround for udma5 support in libata on D0 board -- the 80/40 cable detection register is reserved for Poulsbo PATA register... Jay already pushed it to PPA repo. Jay pushed pciutils source to PPA with new Poulsbo PCI IDS. Those new IDs have been uploaded to Ubuntu package by the maintainer while debian pciutils package maintainer will not update the PCI IDs from Intel -- he like to picked up them from upstream. SDIO No changes this week. Dabney Thermal patches No changes this week. USB Client Drivers Alek upgraded the USBC driver to version Beta4 (2.0.0.32L.0008). PSB Gfx & Video Drivers Inuka built a workable Beta6 Gfx driver against hardy PPA at https://launchpad.net/~inuka/+archive Inuka was planning to release Beta7 Gfx driver once got approved. Power and Performance Testing/Optimization No changes this week. Grub fast boot/resume Martin root caused the fast resume kernel oops issue when HT enabled. Fast resume became more and more stable. Almost all known issues got fixed. '''Status report for week 08''' Moblin Applets Moblin Applets Documentation: Created a 37 page document describing the 10 control panel and 3 status bar applets. I will be adding this to the moblin-applets project page on moblin.org now that John has given me access to the svn repository. Touchscreen Calibration Utility: Split the utility into two pieces, one for the GUI portion, the other for the low level touchscreen driver access and xorg.conf configuration portion. The low level version now resides in a new system daemon called moblin-system-daemon which is started as root via init.d. The gui remains in the moblin-touchscreen executable. The two communicate via DBUS functions so that all the actions requiring root are carried out by the daemon. Thus there is no longer a need to log in as administrator to configure the touchscreen The touchscreen config utility now has the ability to determine the name of the xorg.conf file by scanning the /etc/event.d/session file and reading the option passed to the -config argument. This means that the samsungQ1 config file is no longer hardcoded and the applet should be more extensible to unknown platforms added the capability to add in a completely new touchscreen config section to the target xorg.conf file if it doesn't have one. This should allow any system to be configured from scratch without the need for any default config or prior knowledge of the touchscreen characteristics. removed the filter placed on the points touched on the screen which prevented users from going too far from the illuminated points. This was causing problems on the crown beach. Background Applet: Changed the background applet to no longer use an XML database to save files. It scans the moblin-media thumbnail cache for photos and populates those instead. Photos can no longer be added or removed from the background applet, they are managed from the moblin-media photo viewer exclusively. Sound Mixer Applet: The sound applet is redesigned to dynamically present volume sliders and mute buttons for every playback or capture track the sound hardware supports. It presents a notebook where each tab represents a different "track", e.g. Master, PCM, CD, Aux, Line in, etc. Within each notebook tab there are a series of controller elements instantiated for each channel available in the track. For instance if a track is mono is will have one controller element, if it's stereo it will have two, one for left and one for right. If a track is 5.1 dolby it will have 6 controller elements: front left, front right, front center, rear left, rear right, and woofer. Hardy Package Builds: Created hardy versions of moblin-applets and moblin-keyboard-manager for upload into the ubuntu-mobile repo. Moblin Applets only required a small change to the liboobs functions being used by the date & time applet as their prototypes have changed in the hardy platform's version. '''Plus''' 25 Bugs Fixed this month!! Moblin Kernel and Driver Status Kernel Support Alek is debugging a reboot failure issue https://bugs.launchpad.net/bugs/190246. The root cause is when rebooting, the USBC driver's unload script K20iusbc would unload iusbc and g_file_storage module from memory, while the unloading of iusbc module cause system hung by random (tasklet kernel Oops). QA team also reported another kernel issue, the SMC USB NIC dongle could not resume correctly after S3 resume. The USB net card could not get IP address anymore. But a plug out/in could save it. Will debug it later. PSB / SLT HW Support Jay confirmed SLT T state was working. But the current BIOS did not enable it yet. Need to wait for later BIOS drop. SDIO Feng added Marvell 8686 V9 patch to Moblin 2.6.24 kernel tree. Dabney Thermal patches No changes this week. USB Client Drivers No changes this week. PSB Gfx & Video Drivers Inuka upgraded Gfx driver to beta7 for Moblin kernel. Inuka is debugging an issue where Moblin browser would crash when opening a Yahoo auto pages :(. It does not happen if X use VESA driver. This bug is highly suspected related with Poulsbo Gfx driver. Need more clues. Power and Performance Testing/Optimization Inuka tested Moblin kernel wakeup times during Media player is working by setting kernel 1000HZ vs 250 HZ scheduling. The result is not as expected. 1000HZ appears to make the situation worse. Inuka will continue to test if SMT enable/disable would change the result. Grub fast boot/resume No changes this week. Hildon Desktop Started looking for tickets for UDS Prague, but still holding off in case I would attend FOSSCamp Quite a lot of sponsoring reviews/comments Sought more quantitative information on the topic of translations of mobile packages about the string/UI freeze Reviewing of mobile packages (not sponsored yet); I tend to raise the bar a little as I think the quality of the packages might have been a bit too low to ask universe/main-sponsors to sponsor the mobile teams uploads Worked on Pimlico: - Discussions with upstream to get some information on the state of the stack, especially the Hildon port: . Current focus seems to be on OpenMoko ports . Tasks and Dates pretty stable and ported to Hildon . Sync is a bit behind; doesn't seem reasonable to consider it for the time being . Contacts has been completely refactored (we package the new "hito" branch as contacts-snapshot -- but an older version with a pile of Hildon fixes); the new upstream layout is very clean, but only has an up-to-date OpenMoko port and would ideally need porting to Gtk+ and then to Hildon; mjg59 did most of the contacts-snapshots fixes, but we need to get these upstream and update to a newer tree -- I'd guesstimate at least a solid week of work for the ports - Cleaned up packaging and prepared new upstream releases of tasks and dates against the upstream packaging tree - Reworked the upstream build process to build in two flavours: standard (stock Gtk+) and Hildon; this gives us Hildonized packages under all arches such as i386/amd64 (and armel if we add it); this is all merged upstream as well as all the Ubuntu patches (or equivalent fixes) -- except a dates patch which landed in trunk after the 0.4.5 release; packages currently sitting in Debian NEW, I'll merge them when they come out |
* Lots of driver and kernel bug fixes |
MobileTeam Reporting Page
Include(BuildingCommunityHeader)
[:BuildingCommunity/TeamReporting:Back to the Team Reporting page...]
Welcome to the MobileAndEmbedded reporting page. Every month this page reveals a snapshot of the work that has been going on in the Ubuntu Mobile and Embedded Community. If you have have been working hard on the project and want to let others know what you have been up to, try to summarize it here. Community contributions are actively encouraged. A few guidelines for how to write good points are:
Keep it simple and short - single sentences are best.
- Try to summarise your work and not be over-detailed - we don't need to know specifics, just a summary of the kind
- Where possible, include relevant links.
You should get all your reporting included here by the 22nd of every month.
The Report
Ubuntu Mobile [http://kryten.incognitus.net/mootbot/meetings/ MootBot Logs]
Week 21 -27 Feb
- Reviewing / sponsoring
- Some more work on pimlico; dates is now in sync between upstream, Debian, and Ubuntu; tasks is ready to be synced, but this requires a seed change, so I prefer waiting until
- tomorrow's image build
- Some discussion on versioning of kernel uploads to the ppa (versus in hardy); we should probably pin the ppa higher than hardy for all practical purposes in the end
- Started some quickstart documentation on gettext usage
- Started research on langpacks, documented how much they would gain us ..10 MB says Arne, andwhat we would need to do (we would have to do a lot); this also includes documenting how
- to output langpacks after the various freezes which I alre ady researched with misc people but need to summarize into executive steps
- Adapt Open Office to be able export documents to PDF from command line. The reason behind this is to allow evince to view MS Office documents. During this
- process some specific changes needed to be done applying partially a patch from the official maintainer so it could build correctly on PPA.
- Created a small launcher application called "treb" (short for trebuchet) that is basically an openfile-type dialog that will let the user select a file and will automatically filter
- it using Open Office and pass it to be read by Evince.
- I was able to patch the Firefox 3.0 beta2 chinese language pack and made it work with Midbrowser.
- Added all the internationalization support back into moblin-applets and enabled mobile-basic-flash to display multi-language icon texts. Essentially I regenerated a new
- po/POTFILES.in and po/POTFILES.skip with all the files that have and don’t have translatable strings. About 50% of them haven’t changed since th gnome-control-center fork, but that leaves quite a few with either fuzzy or no translations in all the languages available.
- Lots of driver and kernel bug fixes
MobileAndEmbedded/ReportingPage (last edited 2008-08-06 16:30:49 by localhost)