nir: mark exact fmul in ldexp lowering

this chain of fmul is deliberately chosen for floating point precision
reasons, it needs to be exact, or else we might try to reassociate it
and break subnormal handling.

avoids regressing dEQP-VK.glsl.builtin.precision.ldexp_subnormals.*

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Reviewed-by: Mel Henning <mhenning@darkrefraction.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/36257>

src/compiler/nir/nir_opt_algebraic.py

@@ -3091,7 +3091,7 @@ def ldexp(f, exp, bits):
    # of our exponent is doubled.
    pow2_1 = fexp2i(('ishr', exp, 1), bits)
    pow2_2 = fexp2i(('isub', exp, ('ishr', exp, 1)), bits)
-   return ('fmul', ('fmul', f, pow2_1), pow2_2)
+   return ('!fmul', ('!fmul', f, pow2_1), pow2_2)
 
 optimizations += [
    (('ldexp@16', 'x', 'exp'), ldexp('x', 'exp', 16), 'options->lower_ldexp'),