On 5/18/23 02:49, Juan Quintela wrote:
Richard Henderson <richard.hender...@linaro.org> wrote:
Use cpuinfo_init() during init_accel(), and the variable cpuinfo
during test_buffer_is_zero_next_accel(). Adjust the logic that
cycles through the set of accelerators for testing.
Signed-off-by: Richard Henderson <richard.hender...@linaro.org>
Semi related to this.
For migration, I check every single page to see if it is full of zeros.
But I can promisse that it is just a page (i.e. 4KiB, 16KiB or 64KiB,
correct alignation, correct length, ...).
Will do it make sense to have an special function for that?
Yes, I have found with perf that bufferiszero() is quite high. No, I
haven't try to experiment using a function that is optimized for the
page size in the architecture.
What do you think?
The optimized bufferiszero functions are already optimized for their blocklength (64, 128,
256). I don't think adding page multiples will do much more.
r~
