Hello all, I had a cursory look into porting the hero's sources to a more recent kernel today. As you probably expected I immediately hit a few roadblocks doing so ;) Can anyone comment on the general feasibility of this idea - and whether the hardware is similar enough to the other boards in current mach-msm that not all of hero's supporting code would need to be duplicated?
The most recent hero sources I have are based on 2.6.29, and I am particularly uneasy about these features in the tree: * hero uses amss 6355 per default - current msm sources doesn't support it (anymore?) * probably related: hero uses qdsp5_comp - i don't know about the difference to regular qdsp5 or whether this could be still used? * probably related: hero has MSM_AMSS_SUPPORT_256MB_EBI1 set * perflock - i guess i could just remove that * hero uses htc_headset_v1 - mainline htc_headset looks quite similar but the former seems to use GPIOs (for enabling power) while the latter makes use of the vreg framework and proc_comm (thus handing off the task to a different cpu? is there some hardware difference between sapphire and hero there?) * hero seems to get the battery information using shared memory (HTC_BATTCHG_SMEM in htc_battery), in mainline this seems to be absent * hero kernel was build with support for holes in memory (ARCH_HAS_HOLES_MEMORYMODEL, same as HOLES_IN_ZONE?) - not sure how one would need to respect this Before I spend too much time on this: Is any of this an absolute show stopper? Looking at the diff I am not sure if some of this are just different implementations coming from different historically grown trees or whether the current msm tree is really not made to work with this hardware anymore? (If not: any parts from halibut/trout/sapphire that could be build on?) thanks Gottfried -- unsubscribe: android-kernel+unsubscr...@googlegroups.com website: http://groups.google.com/group/android-kernel
