This patch catches another place in the middle-end where it's possible to preserve the SUBREG_PROMOTED_VAR_P annotation on a subreg to the benefit of later RTL optimizations. This adds the same logic to expr.c's convert_move as recently added to convert_modes.
On nvptx-none, the simple test program: short foo (char c) { return c; } currently generates three instructions: mov.u32 %r23, %ar0; cvt.u16.u32 %r24, %r23; cvt.s32.s16 %value, %r24; with this patch, we now generate just one: mov.u32 %value, %ar0; This patch should look familiar, it's almost identical to the recent patch https://gcc.gnu.org/pipermail/gcc-patches/2021-August/578331.html but with the fix https://gcc.gnu.org/pipermail/gcc-patches/2021-August/578519.html [Apologies again for this breakage on affected (non-x86_64) targets; I hope having that fixed within a few hours (before many folks even noticed a problem) minimized the inconvenience]. This patch has been tested on x86_64-pc-linux-gnu with "make bootstrap" and "make -k check" with no new failures, and on a cross-compiler to nvptx-none, with no new failures in its testsuite. OK for mainline? 2021-09-11 Roger Sayle <ro...@nextmovesoftware.com> gcc/ChangeLog * expr.c (convert_move): Preserve SUBREG_PROMOTED_VAR_P when creating a (wider) partial subreg from a SUBREG_PROMOTED_VAR_P subreg. Roger --
diff --git a/gcc/expr.c b/gcc/expr.c index 17f2c2f..e0bcbcc 100644 --- a/gcc/expr.c +++ b/gcc/expr.c @@ -236,8 +236,27 @@ convert_move (rtx to, rtx from, int unsignedp) >= GET_MODE_PRECISION (to_int_mode)) && SUBREG_CHECK_PROMOTED_SIGN (from, unsignedp)) { + scalar_int_mode int_orig_mode; + scalar_int_mode int_inner_mode; + machine_mode orig_mode = GET_MODE (from); + from = gen_lowpart (to_int_mode, SUBREG_REG (from)); from_mode = to_int_mode; + + /* Preserve SUBREG_PROMOTED_VAR_P if the new mode is wider than + the original mode, but narrower than the inner mode. */ + if (GET_CODE (from) == SUBREG + && is_a <scalar_int_mode> (orig_mode, &int_orig_mode) + && GET_MODE_PRECISION (to_int_mode) + > GET_MODE_PRECISION (int_orig_mode) + && is_a <scalar_int_mode> (GET_MODE (SUBREG_REG (from)), + &int_inner_mode) + && GET_MODE_PRECISION (int_inner_mode) + > GET_MODE_PRECISION (to_int_mode)) + { + SUBREG_PROMOTED_VAR_P (from) = 1; + SUBREG_PROMOTED_SET (from, unsignedp); + } } gcc_assert (GET_CODE (to) != SUBREG || !SUBREG_PROMOTED_VAR_P (to));