Am 06.07.2011 um 15:03 schrieb Hannes Reinecke <h...@suse.de>: > On 07/06/2011 01:34 PM, Alexander Graf wrote: >> >> >> >> >> On 06.07.2011, at 12:24, Paolo Bonzini<pbonz...@redhat.com> wrote: >> >>>> diff --git a/exec.c b/exec.c >>>> index 5f2f87e..f281ba4 100644 >>>> --- a/exec.c >>>> +++ b/exec.c >>>> @@ -4127,7 +4127,8 @@ void cpu_physical_memory_unmap(void *buffer, >>>> target_phys_addr_t len, >>>> } >>>> >>>> /* warning: addr must be aligned */ >>>> -uint32_t ldl_phys(target_phys_addr_t addr) >>>> +static inline uint32_t ldl_phys_internal(target_phys_addr_t addr, >>>> + enum device_endian endian) >>> >>> You probably need the __always_inline__ attribute to really convince GCC to >>> inline this. >> >> There's a define in osdep.h that converts inline into always_inline :) >> > Btw, while you're at it: > > uint32_t ldub_phys(target_phys_addr_t addr); > uint32_t lduw_phys(target_phys_addr_t addr); > > Hmm? ldub is supposed to read an 'unsigned byte' (uint8_t), > and lduw is supposed to read an 'unsigned word' (uint16_t). > > Why does it return an uint32_t?
Because it ends up being a 32-bit register for the return value / parameter anyways :). But this is a different thing. I'd prefer to focus on the endian problem for now. Alex
