== GCC related == * PR44557, Thumb-1 ICE: found two small needed corrections in the ARM backend to fix this. Sent patch upstream.
* LP:641397/PR46888: bitfield insert optimization. Posted patch along with Andrew Stubbs' CSE patch upstream. The two-patch situation seemed to stir some discussion :) It seems both patches were deemed okay, though it would be better if there was some testcase that passed through unhandled by Andrew's CSE patch, but processed by my combine fix later in the pass pipeline. Both patches queued at GCC bugzilla page, to be handled in the next stage1.
* LP:687406/PR46865, -save-temps creating different code. Analyzed problem, though slow to send fix upstream. Had some discussion on the list on how the fix should be like.
* PR46667, section type conflicts. Tested and sent mail to gcc-patches. Jan Hubicka picked it up and pinged for an approval again. Hope this get resolved soon.
* PR45416, ARM code regression. ARM considerations are mostly okay, main issue remaining is how to solve the x86 regression. The current expand code does a full DImode shift just to obtain a single bit, might be point of improvement to solve this.
* LP:685534, ftbfs with gcc-linaro 4.5 on amd64. Found to be another erroneous inline asm case. Fixed and updated on LP.
== This week == * More upstream and Linaro GCC issues. * Start dealing with January travel. * Think more about larger (Linaro) GCC optimization projects.