For genericx86 builds during do_rootfs, we see WARNINGS from qemu about mismatched CPU options: NOTE: recipe core-image-lsb-1.0-r0: task do_rootfs: Started WARNING: core-image-lsb-1.0-r0 do_rootfs: [log_check] core-image-lsb: found 8 warning messages in the logfile: [log_check] warning: TCG doesn't support requested feature: CPUID.80000001H:EDX.syscall [bit 11] [log_check] warning: TCG doesn't support requested feature: CPUID.80000001H:EDX.lm [bit 29] [log_check] warning: TCG doesn't support requested feature: CPUID.80000001H:EDX.syscall [bit 11] [log_check] warning: TCG doesn't support requested feature: CPUID.80000001H:EDX.lm [bit 29] [log_check] warning: TCG doesn't support requested feature: CPUID.80000001H:EDX.syscall [bit 11] [log_check] warning: TCG doesn't support requested feature: CPUID.80000001H:EDX.lm [bit 29] [log_check] warning: TCG doesn't support requested feature: CPUID.80000001H:EDX.syscall [bit 11] [log_check] warning: TCG doesn't support requested feature: CPUID.80000001H:EDX.lm [bit 29] This shows up in the nightly-x86-lsb builds and nightly-x86.
Anuj, can you take this? If you don't know, the problem is that the BSP needs to tell qemu what machine it is emulating. eg intel-corei7-64 sets QEMU_EXTRAOPTIONS_corei7-64 = " -cpu Nehalem,check=false". We'll need to identify what hardware the genericx86 (and -64) machines and tell that.
Looks like it's specific to 32 bit. qemu-i386 doesn't support these features [1] while core2duo that we use for core2-32 does [2]. n270 looks like a good choice, supports SSSE3 and doesn't have these bits enabled [3] [1] https://github.com/qemu/qemu/blob/master/target/i386/cpu.c#L739 [2] https://github.com/qemu/qemu/blob/master/target/i386/cpu.c#L1439 [3] https://github.com/qemu/qemu/blob/master/target/i386/cpu.c#L1603
Merged: https://git.yoctoproject.org/cgit/cgit.cgi/poky/commit/?id=5e9c1b23f4ecbe2d5dce2fa11db028c239c91af3