| Message ID | 20161109145748.27282-16-alex.bennee@linaro.org |
|---|---|
| State | New |
| Headers | show |
On 11/09/2016 03:57 PM, Alex Bennée wrote: > +#ifdef CONFIG_SOFTMMU > + memset(env, 0, offsetof(CPUARMState, tlb_table)); > + tlb_flush(s, 0); > +#else > memset(env, 0, offsetof(CPUARMState, features)); > +#endif I'd really prefer to see the tlb_flush be moved into parent_reset, so that we handle it identically for all targets. As for the memset, do we really need to distinguish softmmu? I don't like you picking out a variable name within CPU_COMMON. Better to use empty struct markers, like the struct {} start_init_save; that x86 uses. r~
Richard Henderson <rth@twiddle.net> writes: > On 11/09/2016 03:57 PM, Alex Bennée wrote: >> +#ifdef CONFIG_SOFTMMU >> + memset(env, 0, offsetof(CPUARMState, tlb_table)); >> + tlb_flush(s, 0); >> +#else >> memset(env, 0, offsetof(CPUARMState, features)); >> +#endif > > I'd really prefer to see the tlb_flush be moved into parent_reset, so that we > handle it identically for all targets. Yeah I'll prepare a series to do that separate from MTTCG. > > As for the memset, do we really need to distinguish softmmu? I don't like you > picking out a variable name within CPU_COMMON. Better to use empty struct > markers, like the > > struct {} start_init_save; > > that x86 uses. OK fair enough. -- Alex Bennée
diff --git a/target-arm/cpu.c b/target-arm/cpu.c index 99f0dbe..990bcb1 100644 --- a/target-arm/cpu.c +++ b/target-arm/cpu.c @@ -122,7 +122,13 @@ static void arm_cpu_reset(CPUState *s) acc->parent_reset(s); +#ifdef CONFIG_SOFTMMU + memset(env, 0, offsetof(CPUARMState, tlb_table)); + tlb_flush(s, 0); +#else memset(env, 0, offsetof(CPUARMState, features)); +#endif + g_hash_table_foreach(cpu->cp_regs, cp_reg_reset, cpu); g_hash_table_foreach(cpu->cp_regs, cp_reg_check_reset, cpu);
cputlb owns the TLB entries and knows how to safely update them in MTTCG. Signed-off-by: Alex Bennée <alex.bennee@linaro.org> --- target-arm/cpu.c | 6 ++++++ 1 file changed, 6 insertions(+) -- 2.10.1