On Thu, Dec 1, 2016 at 4:30 AM, Richard Henderson <r...@twiddle.net> wrote: > On 11/30/2016 10:39 AM, Jin Guojie wrote: >> Yesterday I successfully got in touch with a Debian developer from imgtec. >> Welcome Su Yunqiang. >> V4 patch can be tested on mips64 be hosts now. >> The following guests on a mips64 >> big-endian host is OK: >> - Debian i386 with qemu-system-i386 >> - Debian i386 with qemu-system-x86_64 > > Excellent. > > Yunqiang, for the record is this mips64 or mips64r2? If mips64r2, it might be
Jin gave me his new patch set. I tested it on R2. All of the bellow work well. (HOST) (qemu-system) (Debian ISO) 64eb -> i386 -> i386 64eb -> x86_64 -> i386 64eb -> x86_64 -> amd64 n32eb -> i386 -> i386 n32eb -> x86_64 -> i386 n32eb -> x86_64 -> amd64 32eb -> i386 -> i386 32eb -> x86_64 -> i386 32eb -> x86_64 -> amd64 > helpful to re-test with use_mips32r2_instructions forced to 0 in > tcg_target_detect_isa, since there are different paths selected based on that. I will test for non-r2 today with these paris. > >> However, Debian amd64 with qemu-system-x86_64 fails to boot kernel. >> A 'rep movsq' instruction is wrongly emulated. >> I will fix this bug within one or two days. > > Excellent, thanks. > >> But even Su cannot provide an R6 machine. > > Ok, I guess we will just have to drop the R6 patches for now, until imgtec is > able to provide feedback on them. > > > r~ -- YunQiang Su