On 5/19/26 02:55, Peter Maydell wrote:
So is the problem that we shouldn't be feeding in INT_MAX
as a p->exp, or that the "round up" code should be
checking that it isn't going to overflow with the exp++?
Does the same thing apply to the "exp = INT32_MAX" in the
exp_bias handling in parts_uncanon itself ?

Even worse, we should be supplying the exact maximum value because neither inexact nor overflow are signaled in FPConvertFP8 along the FPType_Infinity path.


r~

Reply via email to