Le 07/11/2016 à 13:08, Richard Henderson a écrit : > On 11/06/2016 08:52 PM, Laurent Vivier wrote: >>> Richard Henderson (4): >>> target-m68k: Do not cpu_abort on undefined insns >>> target-m68k: Inline shifts >>> target-m68k: Implement bitfield ops for registers >>> target-m68k: Implement bitfield ops for memory >> >> There is something wrong with the bitfield patches, because when I use >> them instead of the version I use usually I'm not able to boot a >> container with qemu-m68k (SEGFAULT). > > Ah, too bad. I was testing with the linux-user-0.3 busybox, but we > don't yet implement enough to actually execute it. So the testing > consisted of maybe 3 or 4 such insns until I reach the fmovemx. ;-) > > Can you tell if the breakage is limited to the memory patch, or if the > register patch is also broken?
I need to test more, because if I only apply the register patch, I have an illegal instruction exception. If it can help: sudo QEMU_LOG=in_asm chroot /var/lib/lxc/virtm68k-etch-m68k/rootfs 2>&1 |grep bf 0xf67f9160: bfextu %d3,27,3,%d1 0xf67f8988: bftst %d1,30,2 0xf67f2786: bfins %d0,%a1@,0,2 0xf67eeb56: bfins %d0,%a2@(824),0,2 0xf67f362c: bfins %d0,%a2@(384),5,2 0xf67f8b80: bftst %d1,30,2 0xf67f8b9e: bftst %d4,30,2 0xf67f8998: bftst %d0,30,2 0xf67f8998: bftst %d0,30,2 0xf67f9160: bfextu %d3,27,3,%d1 0xf67f8f88: bftst %d4,30,2 0xf67f39d8: bfins %d0,%a1@(384),5,2 0xf67f3876: bfclr %a0@(384),5,2 If you want to have a try, this tree is based on tgt-m68k + what is needed to boot an etch-m68k container. git://github.com/vivier/qemu-m68k.git for-rth The working tree is: git://github.com/vivier/qemu-m68k.git m68k-dev Thanks, Laurent
