[patch 0/6] Add QVGA support to jbt6k74, fixes and cleanup
stefan at openmoko.org
Thu Oct 18 14:05:40 CEST 2007
This patchset adds QVGA support to the jbt6k74 LCM. During the process some bugs
showed up and got fixed. This costs me a lot more time then I expected it to do,
but finally I had the days to get it working.
Patches are against svn rev 3216.
So what is the status of using QVGA in the GTA01:
With this patches you are able to switch the LCM into qvga mode and adjust the
frmebuffer so you endup with a fully working QVGA mode. The steps to reproduce
it are as the following:
root at fic-gta01:~$ echo qvga-normal > /sys/devices/platform/s3c24xx-spi-gpio.1/spi0.0/state
root at fic-gta01:~$ fbset qvga
and back to vga:
root at fic-gta01:~$ echo normal > /sys/devices/platform/s3c24xx-spi-gpio.1/spi0.0/state
root at fic-gta01:~$ fbset vga
Be aware that the fbset which is build into busybox did not work for me. Please
install the full-size fbset via ipkg install fbset or get the ipkg on your own.
To use the qvga/vga aliase with fbset you also need the fb-modes file with the
Put it into /etc or just use the listed timings manually.
For starting X with the smaller resolution you also need to adjust DPI and size
settings for GTA01 in /etc/X11/Xserver
Once everything else is working fine I'll look into doing this with xrandr.
Anybody knows if xrandr size switching is supported in xfbdev? Rotation is, but
there is some confusion about size.
Stuff that still needs to be done:
o Communication about the resolution change between framebuffer and LCM driver.
Just using the userspace to handle the switching on it's own is not what I would
call a good design. As we already have one way to signal the kernel that the
user would like to change the resolution, via fbset, I would propose to add a
communication channel between the fb and the lcm driver. Once fbset is called
with a different resolution then the current one lcm driver should be informed
about it and take care about the change.
I think this would also be useful on other devices. So I would propose somekind
of generic notifier-chain or resolution change callback here. (john cass and
andrzej zaborowski already talked about this.)
It must be optional and should be bundeled to the platform data as it is device
Any opinions on that? Perhaps even a better solution?
Once this is working, switching the resolution from X with the xrandr extension
needs to be tested and perhaps fixed as well.
More information about the neo1973-hardware