On 11/10/2015 10:08 AM, Pierre Morel wrote:
On 11/09/2015 01:20 PM, Paolo Bonzini wrote:
On 09/11/2015 13:01, Pierre Morel wrote:
This leads to have UINT64_MAX represented with {1, 0} instead of
{0, UINT64_MAX} while {1, 0} is 2^64. This again leads to have
unnecessary and obfuscating transformations with int128_2_64() to
test for UINT64_MAX and return {1,0} in memory_region_init()
while using inverse translation test{1,0} and return UINT64_MAX
in memory_region_size()>>
Yes, the use of UINT64_MAX for 2^64 is a hack, but it is unrelated to
the signedness of Int128.
OK, we agree it is a hack,
but sorry, I should have missed something,
because I do not understand what this hack is useful for.
It's used in the size argument of memory_region_init*, so that it can
remain an uint64_t. The size is usually small (up to 2^40, say) unless
it is 2^64 meaning "the whole address space". The latter case is
covered by UINT64_MAX.
Paolo
OK, I understand, thanks for having taking time for me.
To sum-up size is a size :-) and not an offset in memory.
Size of UINT64_MAX does not exist but we can live without it, having
a description for "whole address space", 2^64, can be useful.
Even there may be other solutions like taking 0 for 2^64,
if a memory size of 0 has no meaning,
but it could be misleading too.
So I do not see better solution for this interesting problematic.
My problem with this came because usually, on hardware, region are easier
described by start/end rather than by start/size.
But changing this in the actual implementation would be too much.
