[PATCH] VECT: Clear LOOP_VINFO_USING_SELECT_VL_P when loop is not partial vectorized

[PATCH] VECT: Clear LOOP_VINFO_USING_SELECT_VL_P when loop is not partial vectorized

[Juzhe-Zhong]

This patch fixes ICE:
https://godbolt.org/z/z8T6o6qov

<source>: In function 'b':
<source>:2:6: error: missing definition
    2 | void b() {
      |      ^
for SSA_NAME: loop_len_8 in statement:
_1 = -loop_len_8;
during GIMPLE pass: vect
<source>:2:6: internal compiler error: verify_ssa failed
0x7f1b56331082 __libc_start_main
	???:0
Please submit a full bug report, with preprocessed source (by using -freport-bug).
Please include the complete backtrace with any bug report.
See <https://gcc.gnu.org/bugs/> for instructions.
Compiler returned: 1

The root cause is we generate such IR in vectorization:

  _1 = -loop_len_8;
  vect_cst__11 = {_1, _1};
  _18 = vect_vec_iv_.6_14 + vect_cst__11;

loop_len_8 is uninitialized value.

The IR _18 = vect_vec_iv_.6_14 + vect_cst__11; is generated because of we are adding induction variable with
the result of SELECT_VL instead of VF.

The code is:

  else if (LOOP_VINFO_USING_SELECT_VL_P (loop_vinfo))
    {
      /* When we're using loop_len produced by SELEC_VL, the non-final
	 iterations are not always processing VF elements.  So vectorize
	 induction variable instead of

	   _21 = vect_vec_iv_.6_22 + { VF, ... };

	 We should generate:

	   _35 = .SELECT_VL (ivtmp_33, VF);
	   vect_cst__22 = [vec_duplicate_expr] _35;
	   _21 = vect_vec_iv_.6_22 + vect_cst__22;  */
      gcc_assert (!slp_node);
      gimple_seq seq = NULL;
      vec_loop_lens *lens = &LOOP_VINFO_LENS (loop_vinfo);
      tree len = vect_get_loop_len (loop_vinfo, NULL, lens, 1, vectype, 0, 0);
      expr = force_gimple_operand (fold_convert (TREE_TYPE (step_expr),
						 unshare_expr (len)),
				   &seq, true, NULL_TREE);
      new_name = gimple_build (&seq, MULT_EXPR, TREE_TYPE (step_expr), expr,
			       step_expr);
      gsi_insert_seq_before (&si, seq, GSI_SAME_STMT);
      step_iv_si = &si;
    }

LOOP_VINFO_USING_SELECT_VL_P is set before loop vectorization analysis so we don't know whether it is partial
vectorization or not but the induction variable depends on SELECT_VL_P is true.

So update SELECT_VL_P as false when it is not partial vectorization.

	PR middle-end/112554

gcc/ChangeLog:

	* tree-vect-loop.cc (vect_determine_partial_vectors_and_peeling): Clear SELECT_VL_P for non-partial vectorization.

gcc/testsuite/ChangeLog:

	* gcc.target/riscv/rvv/autovec/pr112554.c: New test.

---
 .../gcc.target/riscv/rvv/autovec/pr112554.c         | 12 ++++++++++++
 gcc/tree-vect-loop.cc                               | 13 +++++++++++++
 2 files changed, 25 insertions(+)
 create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/autovec/pr112554.c

diff --git a/gcc/testsuite/gcc.target/riscv/rvv/autovec/pr112554.c b/gcc/testsuite/gcc.target/riscv/rvv/autovec/pr112554.c
new file mode 100644
index 00000000000..4afa7c2b15c
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/rvv/autovec/pr112554.c
@@ -0,0 +1,12 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv64gcv -mabi=lp64d -O3 --param=riscv-autovec-preference=fixed-vlmax" } */
+
+int a;
+void b() {
+  unsigned long c = 18446744073709551612UL;
+d:
+  --c;
+  a ^= c;
+  if (c)
+    goto d;
+}
diff --git a/gcc/tree-vect-loop.cc b/gcc/tree-vect-loop.cc
index fb8d999ee6b..3f59139cb01 100644
--- a/gcc/tree-vect-loop.cc
+++ b/gcc/tree-vect-loop.cc
@@ -2657,6 +2657,19 @@ vect_determine_partial_vectors_and_peeling (loop_vec_info loop_vinfo)
     = (!LOOP_VINFO_USING_PARTIAL_VECTORS_P (loop_vinfo)
        && need_peeling_or_partial_vectors_p);
 
+  /* We set LOOP_VINFO_USING_SELECT_VL_P as true before loop vectorization
+     analysis that we don't know whether the loop is vectorized by partial
+     vectors (More details see tree-vect-loop-manip.cc).
+
+     However, SELECT_VL vectorizaton style should only applied on partial
+     vectorization since SELECT_VL is the GIMPLE IR that calculates the
+     number of elements to be process for each iteration.
+
+     After loop vectorization analysis, Clear LOOP_VINFO_USING_SELECT_VL_P
+     if it is not partial vectorized loop.  */
+  if (!LOOP_VINFO_USING_PARTIAL_VECTORS_P (loop_vinfo))
+    LOOP_VINFO_USING_SELECT_VL_P (loop_vinfo) = false;
+
   return opt_result::success ();
 }
 
-- 
2.36.3

Re: [PATCH] VECT: Clear LOOP_VINFO_USING_SELECT_VL_P when loop is not partial vectorized

[Richard Biener]

On Thu, 16 Nov 2023, Juzhe-Zhong wrote:

> This patch fixes ICE:
> https://godbolt.org/z/z8T6o6qov
> 
> <source>: In function 'b':
> <source>:2:6: error: missing definition
>     2 | void b() {
>       |      ^
> for SSA_NAME: loop_len_8 in statement:
> _1 = -loop_len_8;
> during GIMPLE pass: vect
> <source>:2:6: internal compiler error: verify_ssa failed
> 0x7f1b56331082 __libc_start_main
> 	???:0
> Please submit a full bug report, with preprocessed source (by using -freport-bug).
> Please include the complete backtrace with any bug report.
> See <https://gcc.gnu.org/bugs/> for instructions.
> Compiler returned: 1
> 
> The root cause is we generate such IR in vectorization:
> 
>   _1 = -loop_len_8;
>   vect_cst__11 = {_1, _1};
>   _18 = vect_vec_iv_.6_14 + vect_cst__11;
> 
> loop_len_8 is uninitialized value.
> 
> The IR _18 = vect_vec_iv_.6_14 + vect_cst__11; is generated because of we are adding induction variable with
> the result of SELECT_VL instead of VF.
> 
> The code is:
> 
>   else if (LOOP_VINFO_USING_SELECT_VL_P (loop_vinfo))
>     {
>       /* When we're using loop_len produced by SELEC_VL, the non-final
> 	 iterations are not always processing VF elements.  So vectorize
> 	 induction variable instead of
> 
> 	   _21 = vect_vec_iv_.6_22 + { VF, ... };
> 
> 	 We should generate:
> 
> 	   _35 = .SELECT_VL (ivtmp_33, VF);
> 	   vect_cst__22 = [vec_duplicate_expr] _35;
> 	   _21 = vect_vec_iv_.6_22 + vect_cst__22;  */
>       gcc_assert (!slp_node);
>       gimple_seq seq = NULL;
>       vec_loop_lens *lens = &LOOP_VINFO_LENS (loop_vinfo);
>       tree len = vect_get_loop_len (loop_vinfo, NULL, lens, 1, vectype, 0, 0);
>       expr = force_gimple_operand (fold_convert (TREE_TYPE (step_expr),
> 						 unshare_expr (len)),
> 				   &seq, true, NULL_TREE);
>       new_name = gimple_build (&seq, MULT_EXPR, TREE_TYPE (step_expr), expr,
> 			       step_expr);
>       gsi_insert_seq_before (&si, seq, GSI_SAME_STMT);
>       step_iv_si = &si;
>     }
> 
> LOOP_VINFO_USING_SELECT_VL_P is set before loop vectorization analysis so we don't know whether it is partial
> vectorization or not but the induction variable depends on SELECT_VL_P is true.
> 
> So update SELECT_VL_P as false when it is not partial vectorization.

OK.

> 	PR middle-end/112554
> 
> gcc/ChangeLog:
> 
> 	* tree-vect-loop.cc (vect_determine_partial_vectors_and_peeling): Clear SELECT_VL_P for non-partial vectorization.
> 
> gcc/testsuite/ChangeLog:
> 
> 	* gcc.target/riscv/rvv/autovec/pr112554.c: New test.
> 
> ---
>  .../gcc.target/riscv/rvv/autovec/pr112554.c         | 12 ++++++++++++
>  gcc/tree-vect-loop.cc                               | 13 +++++++++++++
>  2 files changed, 25 insertions(+)
>  create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/autovec/pr112554.c
> 
> diff --git a/gcc/testsuite/gcc.target/riscv/rvv/autovec/pr112554.c b/gcc/testsuite/gcc.target/riscv/rvv/autovec/pr112554.c
> new file mode 100644
> index 00000000000..4afa7c2b15c
> --- /dev/null
> +++ b/gcc/testsuite/gcc.target/riscv/rvv/autovec/pr112554.c
> @@ -0,0 +1,12 @@
> +/* { dg-do compile } */
> +/* { dg-options "-march=rv64gcv -mabi=lp64d -O3 --param=riscv-autovec-preference=fixed-vlmax" } */
> +
> +int a;
> +void b() {
> +  unsigned long c = 18446744073709551612UL;
> +d:
> +  --c;
> +  a ^= c;
> +  if (c)
> +    goto d;
> +}
> diff --git a/gcc/tree-vect-loop.cc b/gcc/tree-vect-loop.cc
> index fb8d999ee6b..3f59139cb01 100644
> --- a/gcc/tree-vect-loop.cc
> +++ b/gcc/tree-vect-loop.cc
> @@ -2657,6 +2657,19 @@ vect_determine_partial_vectors_and_peeling (loop_vec_info loop_vinfo)
>      = (!LOOP_VINFO_USING_PARTIAL_VECTORS_P (loop_vinfo)
>         && need_peeling_or_partial_vectors_p);
>  
> +  /* We set LOOP_VINFO_USING_SELECT_VL_P as true before loop vectorization
> +     analysis that we don't know whether the loop is vectorized by partial
> +     vectors (More details see tree-vect-loop-manip.cc).
> +
> +     However, SELECT_VL vectorizaton style should only applied on partial
> +     vectorization since SELECT_VL is the GIMPLE IR that calculates the
> +     number of elements to be process for each iteration.
> +
> +     After loop vectorization analysis, Clear LOOP_VINFO_USING_SELECT_VL_P
> +     if it is not partial vectorized loop.  */
> +  if (!LOOP_VINFO_USING_PARTIAL_VECTORS_P (loop_vinfo))
> +    LOOP_VINFO_USING_SELECT_VL_P (loop_vinfo) = false;
> +
>    return opt_result::success ();
>  }
>  
> 

-- 
Richard Biener <rguenther@suse.de>
SUSE Software Solutions Germany GmbH,
Frankenstrasse 146, 90461 Nuernberg, Germany;
GF: Ivo Totev, Andrew McDonald, Werner Knoblich; (HRB 36809, AG Nuernberg)