From: Tom de Vries <tdevries@suse.de>
To: Richard Biener <richard.guenther@gmail.com>
Cc: Stefan Schulze Frielinghaus <stefansf@linux.ibm.com>,
GCC Patches <gcc-patches@gcc.gnu.org>
Subject: [PATCH][ldist] Don't add lib calls with -fno-tree-loop-distribute-patterns
Date: Mon, 31 Jan 2022 17:26:51 +0100 [thread overview]
Message-ID: <3b14d728-35d3-251a-44b8-e152b92d5dd5@suse.de> (raw)
In-Reply-To: <CAFiYyc28QtFcpFNcSDrVkWxzD1ka6WqD4=bDr6u1pFChwzi+FA@mail.gmail.com>
[-- Attachment #1: Type: text/plain, Size: 2045 bytes --]
[ was: Re: [RFC] ldist: Recognize rawmemchr loop patterns ]
On 1/31/22 16:00, Richard Biener wrote:
>> I'm running into PR56888 (
>> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=56888 ) on nvptx due to
>> this, f.i. in gcc/testsuite/gcc.c-torture/execute/builtins/strlen.c,
>> where gcc/testsuite/gcc.c-torture/execute/builtins/lib/strlen.c contains
>> a strlen function, with a strlen loop, which is transformed by
>> pass_loop_distribution into a __builtin_strlen, which is then expanded
>> into a strlen call, creating a self-recursive function. [ And on nvptx,
>> that happens to result in a compilation failure, which is how I found
>> this. ]
>>
>> According to this (
>> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=56888#c21 ) comment:
>> ...
>> -fno-tree-loop-distribute-patterns is the reliable way to not
>> transform loops into library calls.
>> ...
>>
>> Then should we have something along the lines of:
>> ...
>> $ git diff
>> diff --git a/gcc/tree-loop-distribution.c b/gcc/tree-loop-distribution.c
>> index 6fe59cd56855..9a211d30cd7e 100644
>> --- a/gcc/tree-loop-distribution.c
>> +++ b/gcc/tree-loop-distribution.c
>> @@ -3683,7 +3683,11 @@ loop_distribution::transform_reduction_loop
>> && TYPE_PRECISION (ptr_type_node) >= 32)
>> || (TYPE_OVERFLOW_UNDEFINED (reduction_var_type)
>> && TYPE_PRECISION (reduction_var_type) <=
>> TYPE_PRECISION (sizetype)))
>> - && builtin_decl_implicit (BUILT_IN_STRLEN))
>> + && builtin_decl_implicit (BUILT_IN_STRLEN)
>> + && flag_tree_loop_distribute_patterns)
>> generate_strlen_builtin (loop, reduction_var, load_iv.base,
>> reduction_iv.base, loc);
>> else if (direct_optab_handler (rawmemchr_optab, TYPE_MODE
>> (load_type))
>> ...
>> ?
>>
>> Or is the comment no longer valid?
>
> It is still valid - and yes, I think we need to guard it with this flag
> but please do it in the caller to transform_reduction_loop.
Done.
Ok for trunk?
Thanks,
- Tom
[-- Attachment #2: 0001-ldist-Don-t-add-lib-calls-with-fno-tree-loop-distribute-patterns.patch --]
[-- Type: text/x-patch, Size: 2883 bytes --]
[ldist] Don't add lib calls with -fno-tree-loop-distribute-patterns
As mentioned in PR56888 comment 21:
...
-fno-tree-loop-distribute-patterns is the reliable way to not
transform loops into library calls.
...
However, since commit 6f966f06146 ("ldist: Recognize strlen and rawmemchr like
loops") a strlen or rawmemchr library call may be introduced by ldist.
This caused regressions in testcases
gcc.c-torture/execute/builtins/strlen{,-2,-3}.c for nvptx.
Fix this by not calling transform_reduction_loop from
loop_distribution::execute for -fno-tree-loop-distribute-patterns.
Tested regressing test-cases as well as gcc.dg/tree-ssa/ldist-*.c on
nvptx.
gcc/ChangeLog:
2022-01-31 Tom de Vries <tdevries@suse.de>
* tree-loop-distribution.cc (generate_reduction_builtin_1): Check for
-ftree-loop-distribute-patterns.
(loop_distribution::execute): Don't call transform_reduction_loop for
-fno-tree-loop-distribute-patterns.
gcc/testsuite/ChangeLog:
2022-01-31 Tom de Vries <tdevries@suse.de>
* gcc.dg/tree-ssa/ldist-strlen-4.c: New test.
---
gcc/testsuite/gcc.dg/tree-ssa/ldist-strlen-4.c | 17 +++++++++++++++++
gcc/tree-loop-distribution.cc | 5 ++++-
2 files changed, 21 insertions(+), 1 deletion(-)
diff --git a/gcc/testsuite/gcc.dg/tree-ssa/ldist-strlen-4.c b/gcc/testsuite/gcc.dg/tree-ssa/ldist-strlen-4.c
new file mode 100644
index 000000000000..eafb37e84fc3
--- /dev/null
+++ b/gcc/testsuite/gcc.dg/tree-ssa/ldist-strlen-4.c
@@ -0,0 +1,17 @@
+/* { dg-do compile } */
+/* { dg-options "-O2 -ftree-loop-distribution -fno-tree-loop-distribute-patterns -fdump-tree-ldist-details" } */
+/* { dg-final { scan-tree-dump-not "generated strlen" "ldist" } } */
+
+/* Copied from gcc/testsuite/gcc.c-torture/execute/builtins/lib/strlen.c. */
+
+__SIZE_TYPE__
+foo (const char *s)
+{
+ __SIZE_TYPE__ i;
+
+ i = 0;
+ while (s[i] != 0)
+ i++;
+
+ return i;
+}
diff --git a/gcc/tree-loop-distribution.cc b/gcc/tree-loop-distribution.cc
index 6fe59cd56855..c7b428572636 100644
--- a/gcc/tree-loop-distribution.cc
+++ b/gcc/tree-loop-distribution.cc
@@ -3290,6 +3290,8 @@ generate_reduction_builtin_1 (loop_p loop, gimple_seq &seq,
tree reduction_var_old, tree reduction_var_new,
const char *info, machine_mode load_mode)
{
+ gcc_assert (flag_tree_loop_distribute_patterns);
+
/* Place new statements before LOOP. */
gimple_stmt_iterator gsi = gsi_last_bb (loop_preheader_edge (loop)->src);
gsi_insert_seq_after (&gsi, seq, GSI_CONTINUE_LINKING);
@@ -3773,7 +3775,8 @@ loop_distribution::execute (function *fun)
if (niters == NULL_TREE || niters == chrec_dont_know)
{
datarefs_vec.create (20);
- if (transform_reduction_loop (loop))
+ if (flag_tree_loop_distribute_patterns
+ && transform_reduction_loop (loop))
{
changed = true;
loops_to_be_destroyed.safe_push (loop);
next prev parent reply other threads:[~2022-01-31 16:26 UTC|newest]
Thread overview: 29+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-02-08 12:47 [RFC] ldist: Recognize rawmemchr loop patterns Stefan Schulze Frielinghaus
2021-02-09 8:57 ` Richard Biener
2021-02-14 10:27 ` Stefan Schulze Frielinghaus
2021-02-25 17:01 ` Stefan Schulze Frielinghaus
2021-02-25 23:49 ` Jeff Law
2021-03-02 12:29 ` Richard Biener
2021-03-03 17:17 ` Stefan Schulze Frielinghaus
2021-03-16 17:13 ` Stefan Schulze Frielinghaus
2021-04-08 8:23 ` Stefan Schulze Frielinghaus
2021-05-04 17:25 ` Stefan Schulze Frielinghaus
2021-05-05 9:36 ` Richard Biener
2021-05-05 10:03 ` Richard Biener
2021-05-07 12:32 ` Stefan Schulze Frielinghaus
2021-05-20 9:24 ` Richard Biener
2021-05-20 18:37 ` Stefan Schulze Frielinghaus
2021-06-14 17:26 ` Stefan Schulze Frielinghaus
2021-06-16 14:22 ` Richard Biener
2021-06-25 10:23 ` Stefan Schulze Frielinghaus
2021-08-06 14:02 ` Stefan Schulze Frielinghaus
2021-08-20 10:35 ` Richard Biener
2021-09-03 8:00 ` Stefan Schulze Frielinghaus
2021-09-06 9:56 ` Richard Biener
2021-09-13 14:53 ` Stefan Schulze Frielinghaus
2021-09-17 8:08 ` Richard Biener
2021-10-11 16:02 ` Stefan Schulze Frielinghaus
2022-01-31 13:16 ` Tom de Vries
2022-01-31 15:00 ` Richard Biener
2022-01-31 16:26 ` Tom de Vries [this message]
2022-02-01 7:04 ` [PATCH][ldist] Don't add lib calls with -fno-tree-loop-distribute-patterns Richard Biener
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=3b14d728-35d3-251a-44b8-e152b92d5dd5@suse.de \
--to=tdevries@suse.de \
--cc=gcc-patches@gcc.gnu.org \
--cc=richard.guenther@gmail.com \
--cc=stefansf@linux.ibm.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).