March 19 to 23:
- Initial implementation of the in-kernel b.L switcher.
- Move to the arm-soc tree as a base for the b.L work as it conveniently contains some needed setup bits specifically for the Cortex-A7.
- Tracking the v3.4-rc1 merge window.
- Modifications to ARM's generic boot wrapper to cope with an extended boot protocol regarding hotplug of any CPU and any cluster.
March 26 to 30:
- Review of the initial AArch64 patches.
- Attempt at LDRD/STRD optimized in-kernel memcpy/copy_user routines for the Cortex-A15. This is in reaction to a patch posted on the linux-arm-kernel list which didn't look optimal from a readability point of view.
- Fixed the SDIO single IRQ optimization that produced spurious pending IRQ notifications.
- More work on the b.L switcher, rebasing it to v3.4-rc1.
- Fixed the pl011 serial driverprobe code broken in mainline that oopsed the kernel (as did many other people in parallel).
- Completion of basic b.L switcher support, running into cache coherency problems causing spectacular kernel crashes.
April 02 to 06:
- Review of the semihosting and native DT support to the reference switcher bootwrapper from Dave Martin.
- Looking for a better bootwrapper base to merge my non hacked-up boot protocol changes. I would like to also have the semihosting support, but that is available only in specialized bootwrappers that I have no use for.
- Posted progress on the b.L switcher, especially the cache coherency issues, in case someone could give me a clue.
- Initial thinking about the b.L switcher instrumentation.
- Getting stuck with the b.L switcher and the cache corruption issue, so I posted a code drop to the private list for people at ARM to investigate.
- Self reviews and all.
- Yay! The b.L cache issue was due to a Fast Model bug. With an updated emulator, my switcher code just works!
April 09 to 13:
- Easter.
- Posted my patch to the generic bootwrapper to Will D. and got comments back that I still need to address.
- Cleaned and refactored my b.L switcher code into separate commits. Broke it in the process (obviously). Investigation and fix followed.
- Merged LCD support for the VE model from dmart's tree into my tree to help the Android guys. guys.
- Started to look at the cpufreq driver from Dave M. and at cpufreq support in general. Still confused by some aspects of the API.
Nicolas