On Sun, Oct 13, 2019 at 07:38:04PM -0700, Richard Henderson wrote: >On 10/13/19 6:01 PM, Wei Yang wrote: >>> No, please. >>> >>> (1) The compiler does not know that qemu_*host_page_size is a power of 2, >>> and >>> will generate a real division at runtime. The same is true for >>> TARGET_PAGE_SIZE when TARGET_PAGE_BITS_VARY. >>> >> >> Confused >> >> The definition of ROUND_UP is: >> >> #define ROUND_UP(n, d) (((n) + (d) - 1) & -(0 ? (n) : (d))) > >Ah, my bad, I did confuse this with QEMU_ALIGN_UP. > >Hmm. > > lea -1(n, size), t > neg size > and size, t > >vs > > mov mask, t > not t > add n, t > and mask, t > >which is what I proposed here > >>> https://lists.gnu.org/archive/html/qemu-devel/2019-09/msg04526.html > >I'm ok with your version. > >Reviewed-by: Richard Henderson <richard.hender...@linaro.org> >
Thanks for your clarification. Have a nice day > >r~ > -- Wei Yang Help you, Help me