The patch that moved to vector.S and made common between 32 and 64-bit the altivec code had a nasty bug on 32-bit (did I really test that ?) which causes the kernel to blr back into userspace ... oops :-)
Signed-off-by: Benjamin Herrenschmidt <[email protected]> --- I'm going to stick that in my -next branch along with updating it to Linus latest tonight. Things still blow in various places because of the mem init change, see the patch I posted to lkml (I forgot to CC it here, I'll do that now). diff --git a/arch/powerpc/kernel/head_32.S b/arch/powerpc/kernel/head_32.S index 6437f90..4846946 100644 --- a/arch/powerpc/kernel/head_32.S +++ b/arch/powerpc/kernel/head_32.S @@ -733,9 +733,11 @@ END_MMU_FTR_SECTION_IFSET(MMU_FTR_NEED_DTLB_SW_LRU) AltiVecUnavailable: EXCEPTION_PROLOG #ifdef CONFIG_ALTIVEC - bne load_up_altivec /* if from user, just load it up */ + beq 1f + bl load_up_altivec /* if from user, just load it up */ + b fast_exception_return #endif /* CONFIG_ALTIVEC */ - addi r3,r1,STACK_FRAME_OVERHEAD +1: addi r3,r1,STACK_FRAME_OVERHEAD EXC_XFER_EE_LITE(0xf20, altivec_unavailable_exception) PerformanceMonitor: -- 1.6.1.2.14.gf26b5 _______________________________________________ Linuxppc-dev mailing list [email protected] https://lists.ozlabs.org/listinfo/linuxppc-dev
