From: Richard Biener <rguenther@suse.de>
To: gcc-patches@gcc.gnu.org
Cc: richard.sandiford@arm.com
Subject: [PATCH] tree-optimization/100253 - fix bogus aligned vectorized loads/stores
Date: Thu, 29 Apr 2021 12:42:02 +0200 (CEST) [thread overview]
Message-ID: <nycvar.YFH.7.76.2104291241280.14579@elmra.sevgm.obk> (raw)
At some point DR_MISALIGNMENT was supposed to be -1 when the
access was not element aligned. That's obviously not true at this
point so this adjusts both store and load vectorizing to no longer
assume this which in turn allows simplifying the code.
Bootstrapped and tested on x86_64-unknown-linux-gnu, pushed to
trunk sofar.
2021-04-29 Richard Biener <rguenther@suse.de>
PR tree-optimization/100253
* tree-vect-stmts.c (vectorizable_load): Do not assume
element alignment when DR_MISALIGNMENT is -1.
(vectorizable_store): Likewise.
* g++.dg/pr100253.C: New testcase.
---
gcc/testsuite/g++.dg/pr100253.C | 33 +++++++++++++++++++++++++++++++++
gcc/tree-vect-stmts.c | 18 ++++--------------
2 files changed, 37 insertions(+), 14 deletions(-)
create mode 100644 gcc/testsuite/g++.dg/pr100253.C
diff --git a/gcc/testsuite/g++.dg/pr100253.C b/gcc/testsuite/g++.dg/pr100253.C
new file mode 100644
index 00000000000..0102caa7df8
--- /dev/null
+++ b/gcc/testsuite/g++.dg/pr100253.C
@@ -0,0 +1,33 @@
+/* { dg-do run } */
+/* { dg-options "-O2 -fno-tree-bit-ccp -ftree-slp-vectorize" } */
+
+#include <vector>
+
+struct T
+{
+};
+
+struct S
+{
+ std::vector < std::vector < T > > v;
+ char x;
+ char y[16];
+ char z[16];
+};
+
+S s, g[1];
+
+void
+foo (char *buf)
+{
+ s = g[*buf];
+}
+
+char c;
+
+int
+main ()
+{
+ foo (&c);
+ return 0;
+}
diff --git a/gcc/tree-vect-stmts.c b/gcc/tree-vect-stmts.c
index 4c01e82ff39..bd2a1c89e67 100644
--- a/gcc/tree-vect-stmts.c
+++ b/gcc/tree-vect-stmts.c
@@ -8166,6 +8166,7 @@ vectorizable_store (vec_info *vinfo,
&& TREE_CODE (dataref_ptr) == SSA_NAME)
set_ptr_info_alignment (get_ptr_info (dataref_ptr), align,
misalign);
+ align = least_bit_hwi (misalign | align);
if (memory_access_type == VMAT_CONTIGUOUS_REVERSE)
{
@@ -8187,7 +8188,6 @@ vectorizable_store (vec_info *vinfo,
/* Arguments are ready. Create the new vector stmt. */
if (final_mask)
{
- align = least_bit_hwi (misalign | align);
tree ptr = build_int_cst (ref_type, align * BITS_PER_UNIT);
gcall *call
= gimple_build_call_internal (IFN_MASK_STORE, 4,
@@ -8202,7 +8202,6 @@ vectorizable_store (vec_info *vinfo,
tree final_len
= vect_get_loop_len (loop_vinfo, loop_lens,
vec_num * ncopies, vec_num * j + i);
- align = least_bit_hwi (misalign | align);
tree ptr = build_int_cst (ref_type, align * BITS_PER_UNIT);
machine_mode vmode = TYPE_MODE (vectype);
opt_machine_mode new_ovmode
@@ -8241,14 +8240,10 @@ vectorizable_store (vec_info *vinfo,
: build_int_cst (ref_type, 0));
if (aligned_access_p (first_dr_info))
;
- else if (DR_MISALIGNMENT (first_dr_info) == -1)
- TREE_TYPE (data_ref)
- = build_aligned_type (TREE_TYPE (data_ref),
- align * BITS_PER_UNIT);
else
TREE_TYPE (data_ref)
= build_aligned_type (TREE_TYPE (data_ref),
- TYPE_ALIGN (elem_type));
+ align * BITS_PER_UNIT);
vect_copy_ref_info (data_ref, DR_REF (first_dr_info->dr));
new_stmt = gimple_build_assign (data_ref, vec_oprnd);
vect_finish_stmt_generation (vinfo, stmt_info, new_stmt, gsi);
@@ -9452,10 +9447,10 @@ vectorizable_load (vec_info *vinfo,
&& TREE_CODE (dataref_ptr) == SSA_NAME)
set_ptr_info_alignment (get_ptr_info (dataref_ptr),
align, misalign);
+ align = least_bit_hwi (misalign | align);
if (final_mask)
{
- align = least_bit_hwi (misalign | align);
tree ptr = build_int_cst (ref_type,
align * BITS_PER_UNIT);
gcall *call
@@ -9472,7 +9467,6 @@ vectorizable_load (vec_info *vinfo,
= vect_get_loop_len (loop_vinfo, loop_lens,
vec_num * ncopies,
vec_num * j + i);
- align = least_bit_hwi (misalign | align);
tree ptr = build_int_cst (ref_type,
align * BITS_PER_UNIT);
gcall *call
@@ -9548,14 +9542,10 @@ vectorizable_load (vec_info *vinfo,
= fold_build2 (MEM_REF, ltype, dataref_ptr, offset);
if (alignment_support_scheme == dr_aligned)
;
- else if (DR_MISALIGNMENT (first_dr_info) == -1)
- TREE_TYPE (data_ref)
- = build_aligned_type (TREE_TYPE (data_ref),
- align * BITS_PER_UNIT);
else
TREE_TYPE (data_ref)
= build_aligned_type (TREE_TYPE (data_ref),
- TYPE_ALIGN (elem_type));
+ align * BITS_PER_UNIT);
if (ltype != vectype)
{
vect_copy_ref_info (data_ref,
--
2.26.2
reply other threads:[~2021-04-29 10:42 UTC|newest]
Thread overview: [no followups] expand[flat|nested] mbox.gz Atom feed
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=nycvar.YFH.7.76.2104291241280.14579@elmra.sevgm.obk \
--to=rguenther@suse.de \
--cc=gcc-patches@gcc.gnu.org \
--cc=richard.sandiford@arm.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).