From: Richard Biener <rguenther@suse.de>
To: gcc-patches@gcc.gnu.org
Cc: richard.sandiford@arm.com
Subject: [PATCH] Revert update-ssa assert in vectorizer
Date: Mon, 4 Jul 2022 14:57:45 +0200 (CEST) [thread overview]
Message-ID: <20220704125746.101C81342C@imap2.suse-dmz.suse.de> (raw)
The following reverts the just added assert that virtual SSA does not
need updating. It instead goes for a select whitelist of transforms
known to be prone to difficulties with virtual SSA update.
Bootstrapped and tested on x86_64-unknown-linux-gnu, tested on
aarch64 on the vect.exp testsuite showing all ICEs gone, pushed.
* tree-vect-loop-manip.cc (vect_do_peeling): Revert assert
and update virtual SSA form again. Assert we do so for
a known set of transforms only.
* tree-vectorizer.h (vec_info::any_known_not_updated_vssa): New.
* tree-vect-stmts.cc (vectorizable_load): When vectorizing
using load-lanes allow virtual SSA update.
---
gcc/tree-vect-loop-manip.cc | 11 ++++++-----
gcc/tree-vect-stmts.cc | 3 +++
gcc/tree-vectorizer.h | 4 ++++
3 files changed, 13 insertions(+), 5 deletions(-)
diff --git a/gcc/tree-vect-loop-manip.cc b/gcc/tree-vect-loop-manip.cc
index 7b7af944dba..5c31f0a8091 100644
--- a/gcc/tree-vect-loop-manip.cc
+++ b/gcc/tree-vect-loop-manip.cc
@@ -2683,11 +2683,12 @@ vect_do_peeling (loop_vec_info loop_vinfo, tree niters, tree nitersm1,
class loop *first_loop = loop;
bool irred_flag = loop_preheader_edge (loop)->flags & EDGE_IRREDUCIBLE_LOOP;
- /* Historically we might have a queued need to update virtual SSA form.
- As we delete the update SSA machinery below after doing a regular
- incremental SSA update during loop copying make sure we don't
- lose that fact. */
- gcc_assert (!need_ssa_update_p (cfun));
+ /* We should not have to update virtual SSA form here but some
+ transforms involve creating new virtual definitions which makes
+ updating difficult. */
+ gcc_assert (!need_ssa_update_p (cfun)
+ || loop_vinfo->any_known_not_updated_vssa);
+ update_ssa (TODO_update_ssa_only_virtuals);
create_lcssa_for_virtual_phi (loop);
diff --git a/gcc/tree-vect-stmts.cc b/gcc/tree-vect-stmts.cc
index d6a6fe3fb38..72107afc883 100644
--- a/gcc/tree-vect-stmts.cc
+++ b/gcc/tree-vect-stmts.cc
@@ -8979,6 +8979,9 @@ vectorizable_load (vec_info *vinfo,
dump_printf_loc (MSG_NOTE, vect_location,
"Vectorizing an unaligned access.\n");
+ if (memory_access_type == VMAT_LOAD_STORE_LANES)
+ vinfo->any_known_not_updated_vssa = true;
+
STMT_VINFO_TYPE (stmt_info) = load_vec_info_type;
vect_model_load_cost (vinfo, stmt_info, ncopies, vf, memory_access_type,
alignment_support_scheme, misalignment,
diff --git a/gcc/tree-vectorizer.h b/gcc/tree-vectorizer.h
index 642eb0aeb21..e5fdc9e0a14 100644
--- a/gcc/tree-vectorizer.h
+++ b/gcc/tree-vectorizer.h
@@ -444,6 +444,10 @@ public:
/* Whether the above mapping is complete. */
bool stmt_vec_info_ro;
+ /* Whether we've done a transform we think OK to not update virtual
+ SSA form. */
+ bool any_known_not_updated_vssa;
+
/* The SLP graph. */
auto_vec<slp_instance> slp_instances;
--
2.35.3
reply other threads:[~2022-07-04 12:57 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=20220704125746.101C81342C@imap2.suse-dmz.suse.de \
--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).