public inbox for gcc-patches@gcc.gnu.org
 help / color / mirror / Atom feed
From: Richard Biener <rguenther@suse.de>
To: Jeff Law <jeffreyalaw@gmail.com>
Cc: gcc-patches@gcc.gnu.org, msebor@redhat.com
Subject: Re: [PATCH] tree-optimization/101573 - improve uninit warning at -O0
Date: Tue, 27 Jul 2021 10:45:00 +0200 (CEST)	[thread overview]
Message-ID: <85923o4-rsp3-rrr9-295q-8o46sq328n3@fhfr.qr> (raw)
In-Reply-To: <58095323-cc9f-a9fa-694b-70061acba908@gmail.com>

On Fri, 23 Jul 2021, Jeff Law wrote:

> 
> 
> On 7/22/2021 6:34 AM, Richard Biener wrote:
> > We can improve uninit warnings from the early pass by looking
> > at PHI arguments on fallthru edges that are uninitialized and
> > have uses that are before a possible loop exit.  This catches
> > some cases earlier that we'd only warn in a more confusing
> > way after early inlining as seen by testcase adjustments.
> >
> > It introduces
> >
> > FAIL: gcc.dg/uninit-23.c (test for excess errors)
> >
> > where we additionally warn
> >
> > gcc.dg/uninit-23.c:21:13: warning: 't4' is used uninitialized
> > [-Wuninitialized]
> >
> > which I think is OK even if it's not obvious that the new
> > warning is an improvement when you look at the obvious source.
> >
> > Somehow for all cases I never get the `'foo' was declared here`
> > notes, I didn't dig why that happens but it's odd.
> >
> > Bootstrapped and tested on x86_64-unknown-linux-gnu.
> >
> > Any comments?
> >
> > Thanks,
> > Richard.
> >
> > 2021-07-22  Richard Biener  <rguenther@suse.de>
> >
> >  PR tree-optimization/101573
> >  * tree-ssa-uninit.c (warn_uninitialized_vars): Look at
> >  uninitialized PHI arg defs in some constrained cases.
> >  (execute_early_warn_uninitialized): Calculate dominators.
> >
> >  * gcc.dg/uninit-pr101573.c: New testcase.
> >  * gcc.dg/uninit-15-O0.c: Adjust.
> >  * gcc.dg/uninit-15.c: Likewise.
> >  * gcc.dg/uninit-23.c: Likewise.
> >  * c-c++-common/uninit-17.c: Likewise.
> OK.  Like Martin I think the new code in a function would be easier to read. 
> So if you could factor the bits into a new function it'd be appreciated.
> 
> I wouldn't be terribly surprised if we find additional fallout on other
> targets, but we can fault in those fixes.

The following is what I applied then.

Thanks,
Richard.

From 8fb649db98f3cf49c7d4fd5e566318d2e5ef9756 Mon Sep 17 00:00:00 2001
From: Richard Biener <rguenther@suse.de>
Date: Thu, 22 Jul 2021 12:26:16 +0200
Subject: [PATCH] tree-optimization/101573 - improve uninit warning at -O0
To: gcc-patches@gcc.gnu.org

We can improve uninit warnings from the early pass by looking
at PHI arguments on fallthru edges that are uninitialized and
have uses that are before a possible loop exit.  This catches
some cases earlier that we'd only warn in a more confusing
way after early inlining as seen by testcase adjustments.

It introduces

FAIL: gcc.dg/uninit-23.c (test for excess errors)

where we additionally warn

gcc.dg/uninit-23.c:21:13: warning: 't4' is used uninitialized [-Wuninitialized]

which I think is OK even if it's not obvious that the new
warning is an improvement when you look at the obvious source.

Somehow for all cases I never get the `'foo' was declared here`
notes, I didn't dig why that happens but it's odd.

2021-07-22  Richard Biener  <rguenther@suse.de>

	PR tree-optimization/101573
	* tree-ssa-uninit.c (warn_uninit_phi_uses): New function
	looking at uninitialized PHI arg defs in some constrained cases.
	(warn_uninitialized_vars): Call it.
	(execute_early_warn_uninitialized): Calculate dominators.

	* gcc.dg/uninit-pr101573.c: New testcase.
	* gcc.dg/uninit-15-O0.c: Adjust.
	* gcc.dg/uninit-15.c: Likewise.
	* gcc.dg/uninit-23.c: Likewise.
	* c-c++-common/uninit-17.c: Likewise.
---
 gcc/testsuite/c-c++-common/uninit-17.c |  6 +--
 gcc/testsuite/gcc.dg/uninit-15-O0.c    |  2 +-
 gcc/testsuite/gcc.dg/uninit-15.c       | 10 ++--
 gcc/testsuite/gcc.dg/uninit-23.c       |  2 +-
 gcc/testsuite/gcc.dg/uninit-pr101573.c | 10 ++++
 gcc/tree-ssa-uninit.c                  | 75 ++++++++++++++++++++++++++
 6 files changed, 95 insertions(+), 10 deletions(-)
 create mode 100644 gcc/testsuite/gcc.dg/uninit-pr101573.c

diff --git a/gcc/testsuite/c-c++-common/uninit-17.c b/gcc/testsuite/c-c++-common/uninit-17.c
index fd773da78ad..b5495366c5b 100644
--- a/gcc/testsuite/c-c++-common/uninit-17.c
+++ b/gcc/testsuite/c-c++-common/uninit-17.c
@@ -9,11 +9,11 @@ static void bar(int a, int *ptr)
 {
   do
   {
-    int b;   /* { dg-message "declared" } */
+    int b;
     if (b < 40) {
-      ptr[0] = b; /* { dg-warning "may be used uninitialized" } */
+      ptr[0] = b;
     }
-    b += 1;
+    b += 1; /* { dg-warning "is used uninitialized" } */
     ptr++;
   }
   while (--a != 0);
diff --git a/gcc/testsuite/gcc.dg/uninit-15-O0.c b/gcc/testsuite/gcc.dg/uninit-15-O0.c
index a3fd2b63ba7..36d96348617 100644
--- a/gcc/testsuite/gcc.dg/uninit-15-O0.c
+++ b/gcc/testsuite/gcc.dg/uninit-15-O0.c
@@ -15,6 +15,6 @@ void baz();
 void bar()
 {
     int j;           /* { dg-message "was declared here" {} { xfail *-*-* } } */
-    for (; foo(j); ++j)
+    for (; foo(j); ++j) /* { dg-warning "is used uninitialized" } */
         baz();
 }
diff --git a/gcc/testsuite/gcc.dg/uninit-15.c b/gcc/testsuite/gcc.dg/uninit-15.c
index 8ee10c27aba..85cb116f349 100644
--- a/gcc/testsuite/gcc.dg/uninit-15.c
+++ b/gcc/testsuite/gcc.dg/uninit-15.c
@@ -1,7 +1,7 @@
 /* PR tree-optimization/17506
-   We issue an uninitialized variable warning at a wrong location at
+   We used to issue an uninitialized variable warning at a wrong location at
    line 11, which is very confusing.  Make sure we print out a note to
-   make it less confusing.  (not xfailed alternative)
+   make it less confusing.  (xfailed alternative)
    But it is of course ok if we warn in bar about uninitialized use
    of j.  (not xfailed alternative)  */
 /* { dg-do compile } */
@@ -10,7 +10,7 @@
 inline int
 foo (int i)
 {
-  if (i) /* { dg-warning "used uninitialized" } */
+  if (i) /* { dg-warning "used uninitialized" "" { xfail *-*-* } } */
     return 1;
   return 0;
 }
@@ -20,7 +20,7 @@ void baz (void);
 void
 bar (void)
 {
-  int j; /* { dg-message "note: 'j' was declared here" "" } */
-  for (; foo (j); ++j)  /* { dg-warning "'j' is used uninitialized" "" { xfail *-*-* } } */
+  int j; /* { dg-message "note: 'j' was declared here" "" { xfail *-*-* } } */
+  for (; foo (j); ++j)  /* { dg-warning "'j' is used uninitialized" } */
     baz ();
 }
diff --git a/gcc/testsuite/gcc.dg/uninit-23.c b/gcc/testsuite/gcc.dg/uninit-23.c
index d64eb7d2ee9..87b4e989b53 100644
--- a/gcc/testsuite/gcc.dg/uninit-23.c
+++ b/gcc/testsuite/gcc.dg/uninit-23.c
@@ -18,7 +18,7 @@ ql (void)
         int *t4 = go; /* { dg-warning "is used uninitialized" } */
 
  l1:
-        *t4 = (*t4 != 0) ? 0 : 2;
+        *t4 = (*t4 != 0) ? 0 : 2; /* { dg-warning "is used uninitialized" } */
       }
 
     if (ij != 0)
diff --git a/gcc/testsuite/gcc.dg/uninit-pr101573.c b/gcc/testsuite/gcc.dg/uninit-pr101573.c
new file mode 100644
index 00000000000..a574844b791
--- /dev/null
+++ b/gcc/testsuite/gcc.dg/uninit-pr101573.c
@@ -0,0 +1,10 @@
+/* { dg-do compile } */
+/* { dg-options "-O0 -Wuninitialized" } */
+
+int main(int argc, char **argv)
+{
+  int a;
+  for(; a < 5; ++a) /* { dg-warning "is used uninitialized" } */
+    ;
+  return  0;
+}
diff --git a/gcc/tree-ssa-uninit.c b/gcc/tree-ssa-uninit.c
index 148f3c2b31d..718b32691c1 100644
--- a/gcc/tree-ssa-uninit.c
+++ b/gcc/tree-ssa-uninit.c
@@ -638,6 +638,76 @@ maybe_warn_pass_by_reference (gcall *stmt, wlimits &wlims)
   wlims.always_executed = save_always_executed;
 }
 
+/* Warn about an uninitialized PHI argument on the fallthru path to
+   an always executed block BB.  */
+
+static void
+warn_uninit_phi_uses (basic_block bb)
+{
+  edge_iterator ei;
+  edge e, found = NULL, found_back = NULL;
+  /* Look for a fallthru and possibly a single backedge.  */
+  FOR_EACH_EDGE (e, ei, bb->preds)
+    {
+      /* Ignore backedges.  */
+      if (dominated_by_p (CDI_DOMINATORS, e->src, bb))
+	{
+	  if (found_back)
+	    {
+	      found = NULL;
+	      break;
+	    }
+	  found_back = e;
+	  continue;
+	}
+      if (found)
+	{
+	  found = NULL;
+	  break;
+	}
+      found = e;
+    }
+  if (!found)
+    return;
+
+  basic_block succ = single_succ (ENTRY_BLOCK_PTR_FOR_FN (cfun));
+  for (gphi_iterator si = gsi_start_phis (bb); !gsi_end_p (si);
+       gsi_next (&si))
+    {
+      gphi *phi = si.phi ();
+      tree def = PHI_ARG_DEF_FROM_EDGE (phi, found);
+      if (TREE_CODE (def) != SSA_NAME
+	  || !SSA_NAME_IS_DEFAULT_DEF (def)
+	  || virtual_operand_p (def))
+	continue;
+      /* If there's a default def on the fallthru edge PHI
+	 value and there's a use that post-dominates entry
+	 then that use is uninitialized and we can warn.  */
+      imm_use_iterator iter;
+      use_operand_p use_p;
+      gimple *use_stmt = NULL;
+      FOR_EACH_IMM_USE_FAST (use_p, iter, gimple_phi_result (phi))
+	{
+	  use_stmt = USE_STMT (use_p);
+	  if (gimple_location (use_stmt) != UNKNOWN_LOCATION
+	      && dominated_by_p (CDI_POST_DOMINATORS, succ,
+				 gimple_bb (use_stmt))
+	      /* If we found a non-fallthru edge make sure the
+		 use is inside the loop, otherwise the backedge
+		 can serve as initialization.  */
+	      && (!found_back
+		  || dominated_by_p (CDI_DOMINATORS, found_back->src,
+				     gimple_bb (use_stmt))))
+	    break;
+	  use_stmt = NULL;
+	}
+      if (use_stmt)
+	warn_uninit (OPT_Wuninitialized, def, SSA_NAME_VAR (def),
+		     SSA_NAME_VAR (def),
+		     "%qD is used uninitialized", use_stmt,
+		     UNKNOWN_LOCATION);
+    }
+}
 
 static unsigned int
 warn_uninitialized_vars (bool wmaybe_uninit)
@@ -652,6 +722,10 @@ warn_uninitialized_vars (bool wmaybe_uninit)
     {
       basic_block succ = single_succ (ENTRY_BLOCK_PTR_FOR_FN (cfun));
       wlims.always_executed = dominated_by_p (CDI_POST_DOMINATORS, succ, bb);
+
+      if (wlims.always_executed)
+	warn_uninit_phi_uses (bb);
+
       for (gsi = gsi_start_bb (bb); !gsi_end_p (gsi); gsi_next (&gsi))
 	{
 	  gimple *stmt = gsi_stmt (gsi);
@@ -3135,6 +3209,7 @@ execute_early_warn_uninitialized (void)
      optimization we want to warn about possible uninitialized as late
      as possible, thus don't do it here.  However, without
      optimization we need to warn here about "may be uninitialized".  */
+  calculate_dominance_info (CDI_DOMINATORS);
   calculate_dominance_info (CDI_POST_DOMINATORS);
 
   warn_uninitialized_vars (/*warn_maybe_uninitialized=*/!optimize);
-- 
2.26.2

      reply	other threads:[~2021-07-27  8:45 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-07-22 12:34 Richard Biener
2021-07-22 18:03 ` Martin Sebor
2021-07-23 17:36 ` Jeff Law
2021-07-27  8:45   ` Richard Biener [this message]

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=85923o4-rsp3-rrr9-295q-8o46sq328n3@fhfr.qr \
    --to=rguenther@suse.de \
    --cc=gcc-patches@gcc.gnu.org \
    --cc=jeffreyalaw@gmail.com \
    --cc=msebor@redhat.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).