Hi again On Tue, 7 Feb 2012, Guennadi Liakhovetski wrote:
> Hi > > Trying to get Android 4.0 to run on a Cortex-A8 SoC I've built a > distribution from sources, using the wingray configuration, because it > does not use NEON commands. Tried both the mainline 3.3-rc2 kernel with > enabled Android drivers and the common branch of the android kernel > git-tree. In both cases I'm getting a kernel panic "attempted to kill > init." Trying to run init under gdb in a booted (with the same kernel) > shall I caught a SIGILL in the beginning of __vfprint() in libc at the > vpush instruction. Any ideas? Ok, I've resolved this one, was a misconfigured kernel. OTOH > BTW, while investigating this I noticed, that libc objects (e.g., > vfprintf.o) are built with both -fno-strict-aliasing and -fstrict-aliasing > options, which is, probably, not really desired? this still is something that wants to be fixed. As well as the next issue: untracked Makefile dependencies. Changing flags in one of platform's *.mk files, that affect various parts of the Android build doesn't force a recompilation. E.g., changing the ARCH_ARM_HAVE_TLS_REGISTER variable is ignored. BTW, is it a good place to discuss or report such problems here or should I report them directly to Google's bug-tracking? Thanks Guennadi --- Guennadi Liakhovetski, Ph.D. Freelance Open-Source Software Developer http://www.open-technology.de/ -- unsubscribe: [email protected] website: http://groups.google.com/group/android-porting
