|
From: Josef W. <Jos...@gm...> - 2012-06-04 17:56:41
|
Am 04.06.2012 19:00, schrieb Julian Seward: > second opinion. It is a bit more complex than the Cachegrind case > because in this case, the assertions re size fail if cache simulation > is not selected, and so have to be "gated" on that condition. You could have set CLG_(min_line_size) in any case, but it's ok this way, too. It looks fine to me. I just wondered about whether we do the right thing not really taking I1 cache line size into account for calculating CLG_(min_line_size). But we actually do: according to the comment in cg_arch.c atop the check against MIN_LINE_SIZE, it is set to 16 because that is larger than any instruction. Josef |