nir/lower_int64: fix shift lowering

Starting with !19748 lowered 64 bit shifts were showing wrong results for
shifts with insignificant bits set.

nir shifts are defined to only look at the least significant bits. The
lowering has take this into account.

So there are two things going on:
1. the `ieq` and `uge` further down depend on `y` being masked.
2. the calculation of `reverse_count` actually depends on a masked `y` as
   well, due to the `(iabs (iadd y -32))` giving a different result for
   shifts > 31;

Fixes: 41f3e9e5f5d ("nir: Implement lowering of 64-bit shift operations")
Signed-off-by: Karol Herbst <kherbst@redhat.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19995>

diff --git a/src/compiler/nir/nir_lower_int64.c b/src/compiler/nir/nir_lower_int64.c
index e73e3fa..58feca4 100644 (file)
--- a/src/compiler/nir/nir_lower_int64.c
+++ b/src/compiler/nir/nir_lower_int64.c
@@ -170,6 +170,8 @@ lower_ishl64(nir_builder *b, nir_ssa_def *x, nir_ssa_def *y)
     *
     * uint64_t lshift(uint64_t x, int c)
     * {
+    *    c %= 64;
+    *
     *    if (c == 0) return x;
     *
     *    uint32_t lo = LO(x), hi = HI(x);
@@ -187,6 +189,7 @@ lower_ishl64(nir_builder *b, nir_ssa_def *x, nir_ssa_def *y)
     */
    nir_ssa_def *x_lo = nir_unpack_64_2x32_split_x(b, x);
    nir_ssa_def *x_hi = nir_unpack_64_2x32_split_y(b, x);
+   y = nir_iand_imm(b, y, 0x3f);
 
    nir_ssa_def *reverse_count = nir_iabs(b, nir_iadd(b, y, nir_imm_int(b, -32)));
    nir_ssa_def *lo_shifted = nir_ishl(b, x_lo, y);
@@ -212,6 +215,8 @@ lower_ishr64(nir_builder *b, nir_ssa_def *x, nir_ssa_def *y)
     *
     * uint64_t arshift(uint64_t x, int c)
     * {
+    *    c %= 64;
+    *
     *    if (c == 0) return x;
     *
     *    uint32_t lo = LO(x);
@@ -231,6 +236,7 @@ lower_ishr64(nir_builder *b, nir_ssa_def *x, nir_ssa_def *y)
     */
    nir_ssa_def *x_lo = nir_unpack_64_2x32_split_x(b, x);
    nir_ssa_def *x_hi = nir_unpack_64_2x32_split_y(b, x);
+   y = nir_iand_imm(b, y, 0x3f);
 
    nir_ssa_def *reverse_count = nir_iabs(b, nir_iadd(b, y, nir_imm_int(b, -32)));
    nir_ssa_def *lo_shifted = nir_ushr(b, x_lo, y);
@@ -256,6 +262,8 @@ lower_ushr64(nir_builder *b, nir_ssa_def *x, nir_ssa_def *y)
     *
     * uint64_t rshift(uint64_t x, int c)
     * {
+    *    c %= 64;
+    *
     *    if (c == 0) return x;
     *
     *    uint32_t lo = LO(x), hi = HI(x);
@@ -274,6 +282,7 @@ lower_ushr64(nir_builder *b, nir_ssa_def *x, nir_ssa_def *y)
 
    nir_ssa_def *x_lo = nir_unpack_64_2x32_split_x(b, x);
    nir_ssa_def *x_hi = nir_unpack_64_2x32_split_y(b, x);
+   y = nir_iand_imm(b, y, 0x3f);
 
    nir_ssa_def *reverse_count = nir_iabs(b, nir_iadd(b, y, nir_imm_int(b, -32)));
    nir_ssa_def *lo_shifted = nir_ushr(b, x_lo, y);