Re: video broken on 1.75?
On Mon, Mar 18, 2013 at 6:37 PM, Sameer Verma wrote: > We have a bunch of XO 1.75 going to Madagascar. These have build > 13.1.0 (candidate 35). This ogv file [1] works well on a XO-1 but does > not play via jukebox. The ogv file plays well in Browse (trouble with > going fullscreen). It plays ok for a few seconds via MediaPlayer in > GNOME. > > [1] http://verma.sfsu.edu/projects/olpc/sample-for-xo.ogv > > Ideas? Suggestions? > > cheers, > Sameer To add to the above, the XOs had 12.1.0 on it, with the same problems. So, we upgraded to 13.1.0 Candidate 35, but to no avail. What intrigues me is the difference in video playboack across Browse, Jukebox and MediaPlayer. Should we go all the way back to 11.x? cheers, Sameer ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: XO 1.75 hangs on countdown
On Mon, Mar 18, 2013 at 06:10:42PM -0700, Sameer Verma wrote: > On Mon, Mar 18, 2013 at 5:59 PM, James Cameron wrote: > > A hang at the number "10" means that Open Firmware did not continue > > past the point of setting up the page (section) table and turning on > > the MMU and caches. > > > > http://tracker.coreboot.org/trac/openfirmware/browser/cpu/arm/olpc/resetvec.fth#L144 > > is where the "10" is shown. The "11" is just beyond, a few lines > > down. > > > > My guesses are: > > > > - faulty memory, > > > > - faulty processor, > > > > - fault in wiring between processor and memory, > > > > - incorrect firmware version for the memory, (some memory required a > > specific version), > > > > - corrupt SPI FLASH. > > > > Have you any further history on this laptop? > > Fresh out of the box. SKU 213 > Box id: BCL201208280039C16 Thanks. The possibilities are; the fault developed during shipment, or developed after testing, or the testing was insufficient, or the testing failed to detect the fault (a test escape). Further diagnosis is possible, but will cost you some time and risk. You should choose whether you want to go through this route. If anybody is going to do failure analysis, they will want the unit in pristine condition. SKU213 included Q4D18. The switch of memory vendor required Q4D22. The unit should not have the newer memory, but on the off chance it does: if you have the unit open already, please photograph one of the memory chips. It may be possible to upgrade the firmare despite the problem you see, but this requires a serial adapter and opening the unit. I can guide you through this process, but it would invalidate failure analysis. This is not to say this memory vendor switch is a likely cause, it is only that I am primed for this particular cause, and it is something I can help with. A bias. Other causes are more likely, and I can't help with them. A dislodged piece of metal, such as a screw, or solder, might also cause this if it became lodged near the memory chips. -- James Cameron http://quozl.linux.org.au/ ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
video broken on 1.75?
We have a bunch of XO 1.75 going to Madagascar. These have build 13.1.0 (candidate 35). This ogv file [1] works well on a XO-1 but does not play via jukebox. The ogv file plays well in Browse (trouble with going fullscreen). It plays ok for a few seconds via MediaPlayer in GNOME. [1] http://verma.sfsu.edu/projects/olpc/sample-for-xo.ogv Ideas? Suggestions? cheers, Sameer -- Sameer Verma, Ph.D. Professor, Information Systems San Francisco State University http://verma.sfsu.edu/ http://commons.sfsu.edu/ http://olpcsf.org/ http://olpcjamaica.org.jm/ ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: XO 1.75 hangs on countdown
On Mon, Mar 18, 2013 at 5:59 PM, James Cameron wrote: > A hang at the number "10" means that Open Firmware did not continue > past the point of setting up the page (section) table and turning on > the MMU and caches. > > http://tracker.coreboot.org/trac/openfirmware/browser/cpu/arm/olpc/resetvec.fth#L144 > is where the "10" is shown. The "11" is just beyond, a few lines > down. > > My guesses are: > > - faulty memory, > > - faulty processor, > > - fault in wiring between processor and memory, > > - incorrect firmware version for the memory, (some memory required a > specific version), > > - corrupt SPI FLASH. > > Have you any further history on this laptop? Fresh out of the box. SKU 213 Box id: BCL201208280039C16 Sameer ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: XO 1.75 hangs on countdown
A hang at the number "10" means that Open Firmware did not continue past the point of setting up the page (section) table and turning on the MMU and caches. http://tracker.coreboot.org/trac/openfirmware/browser/cpu/arm/olpc/resetvec.fth#L144 is where the "10" is shown. The "11" is just beyond, a few lines down. My guesses are: - faulty memory, - faulty processor, - fault in wiring between processor and memory, - incorrect firmware version for the memory, (some memory required a specific version), - corrupt SPI FLASH. Have you any further history on this laptop? -- James Cameron http://quozl.linux.org.au/ ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
XO 1.75 hangs on countdown
We have a XO 1.75 that hangs at the number "10" when the firmware starts up. Multiple attempts to power up result in the same thing. laptop id: SHC235004F8 Any ideas? cheers, Sameer ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: XO-1 Sugar startup profiling
On Mon, Mar 18, 2013 at 8:26 AM, Daniel Drake wrote: > I then re-ran the test through perf on 13.2.0 build 1. > 8.66% of this time is spent in g_typelib_get_dir_entry > (libgobject-introspection). > 6.79% is spent in pure Python (EvalFrameEx) > 5.8% of the time is spent in libc. Oops, I meant to write g_typelib_get_dir_entry_by_gtype(). Daniel ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
XO-1 Sugar startup profiling
Hi, I spent an hour comparing and investigating Sugar startup time on XO-1 for 12.1.0 vs 13.2.0 build 1. To measure this, I modified sugar-session to add a glib idle_add call to gtk.main_quit(), after the other bunch of idle_add calls that are made. Then startup can be timed by stopping X and running from a root console: time olpc-dm On 12.1.0 after a couple of runs (for hot caches) this takes 22 seconds. On 13.2.0 the equivalent test takes 32 seconds. I then re-ran the test through perf on 13.2.0 build 1. 8.66% of this time is spent in g_typelib_get_dir_entry (libgobject-introspection). 6.79% is spent in pure Python (EvalFrameEx) 5.8% of the time is spent in libc. Beyond that, there are no clear "offenders". There are bucketloads of places that just occupy around 0.02% of the startup time. As the next steps I would suggest: 1. Investigate why g_typelib_get_dir_entry takes so long (maybe it is just slow, or maybe it's being called far too many times?). http://dev.laptop.org/ticket/12631 2. Do python-level profiling of sugar to figure out why we are also spending a lot of time executing pure Python code. Daniel ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel
Re: trouble making actrd with dracut
Sorry, I forgot to say that I have solved it by adding _hashlib.so in python-contents.txt I think that is the same thing that you did. Thanks. 2013/3/15 Daniel Drake > On Fri, Mar 1, 2013 at 9:54 AM, Esteban Bordón > wrote: > > I'm having a problem trying to import httplib: > > > > ImportError: No module named hashlib > > Please try again with the updated python-contents.txt that I just pushed > > http://dev.laptop.org/git/users/dsd/dracut-modules-olpc/plain/30olpc-python/python-contents.txt > > Put this at /usr/lib/dracut/modules.d/30olpc-python/python-contents.txt > overwriting the file that is already there. > > Thanks > Daniel > ___ Devel mailing list Devel@lists.laptop.org http://lists.laptop.org/listinfo/devel