On 26/10/19 15:20, Philippe Mathieu-Daudé wrote: > On 10/25/19 1:17 PM, Paolo Bonzini wrote: >> On 25/10/19 12:46, Peter Maydell wrote: >>> >>> x86_register_ferr_irq() is defined in target/i386/fpu_helper.c, >>> which is only built if CONFIG_TCG, but the callers don't >>> seem to be similarly guarded and there's no stub fallback. >> >> Indeed, thanks. I'll add an "if (tcg_enabled())" since this feature is >> not supported by accelerators other than TCG. > > Paolo, since the "Split timer <-> rtc" series got merged via > the trivial tree, you need to squash this fix to patch #33 > ("mc146818rtc: Include mc146818rtc_regs.h directly in > mc146818rtc.c") to avoid build failure:
There are a couple other semantic conflicts, so I pushed again the tag. Paolo