Hi Paolo, On 3/22/21 5:41 PM, Paolo Bonzini wrote: > Looks good to me, thanks for all the effort! > > Paolo > > On 22/03/21 14:27, Claudio Fontana wrote: >> v27 -> v28: >> >> * rebased on latest master; >> I indicated the conflicts for the affected patches in the commit >> message, >> in case a new review/eye is required. >> >> * added three patches: >> - accel: move call to accel_init_interfaces >> >> This matches more closely the initialization phases definitions >> (Paolo) >> >> - accel: add init_accel_cpu for adapting accel behavior to cpu type >> This in particular is useful for ARM, that needs different TCG >> behavior >> depending on the CPU subclass. >> >> - XXX RFC accel: add cpu_reset >> >> This adds an accel cpu behavior to execute after CPU reset. >> This can be used on x86, arm, s390x, mips for KVM and TCG. >> >> The RFC nature of this has to do with the fact that cpu_reset() >> remains >> in hw/core/cpu.c , which is common_ss, >> and cpu_reset() calls accel_cpu_reset() which is in specific_ss. >> So it seems weird that this builds fine, and all tests seem >> to pass, >> without using a specific_ss call.
What about this cpu_reset() part? >> >> >> v26 -> v27: rebased on latest master
