public inbox for gcc-patches@gcc.gnu.org
 help / color / mirror / Atom feed
From: Fritz Reese <fritzoreese@gmail.com>
To: gcc-patches <gcc-patches@gcc.gnu.org>, fortran <fortran@gcc.gnu.org>
Subject: [PATCH, Fortran] PR fortran/77764 - ICE in is_anonymous_component
Date: Thu, 29 Sep 2016 14:45:00 -0000	[thread overview]
Message-ID: <CAE4aFAmZJUkUBPFi_jnz-14680y1x=XpjqHiapktvzyVKnGz-A@mail.gmail.com> (raw)

[-- Attachment #1: Type: text/plain, Size: 537 bytes --]

ICE in [1] is due to failure to null-guard map components in
gfc_compare_union_types. Attached is [obvious] fix - will commit soon
without complaints.

---
Fritz Reese

[1] https://gcc.gnu.org/bugzilla/show_bug.cgi?id=77764

2016-09-29  Fritz Reese  <fritzoreese@gmail.com>

        Fix ICE for maps with zero components.

        PR fortran/77764
        * gcc/fortran/interface.c (gfc_compare_union_types): Null-guard map
        components.

        PR fortran/77764
        * gcc/testsuite/gfortran.dg/dec_union_8.f90: New testcase.

[-- Attachment #2: pr77764.patch --]
[-- Type: text/x-patch, Size: 1976 bytes --]

diff --git a/gcc/fortran/interface.c b/gcc/fortran/interface.c
index 9d4e5e9..55923a3 100644
--- a/gcc/fortran/interface.c
+++ b/gcc/fortran/interface.c
@@ -526,6 +526,7 @@ int
 gfc_compare_union_types (gfc_symbol *un1, gfc_symbol *un2)
 {
   gfc_component *map1, *map2, *cmp1, *cmp2;
+  gfc_symbol *map1_t, *map2_t;
 
   if (un1->attr.flavor != FL_UNION || un2->attr.flavor != FL_UNION)
     return 0;
@@ -541,16 +542,26 @@ gfc_compare_union_types (gfc_symbol *un1, gfc_symbol *un2)
      we compare the maps sequentially. */
   for (;;)
   {
-    cmp1 = map1->ts.u.derived->components;
-    cmp2 = map2->ts.u.derived->components;
+    map1_t = map1->ts.u.derived;
+    map2_t = map2->ts.u.derived;
+
+    cmp1 = map1_t->components;
+    cmp2 = map2_t->components;
+
+    /* Protect against null components.  */
+    if (map1_t->attr.zero_comp != map2_t->attr.zero_comp)
+      return 0;
+
+    if (map1_t->attr.zero_comp)
+      return 1;
+
     for (;;)
     {
       /* No two fields will ever point to the same map type unless they are
          the same component, because one map field is created with its type
          declaration. Therefore don't worry about recursion here. */
       /* TODO: worry about recursion into parent types of the unions? */
-      if (compare_components (cmp1, cmp2,
-            map1->ts.u.derived, map2->ts.u.derived) == 0)
+      if (compare_components (cmp1, cmp2, map1_t, map2_t) == 0)
         return 0;
 
       cmp1 = cmp1->next;
diff --git a/gcc/testsuite/gfortran.dg/dec_union_8.f90 b/gcc/testsuite/gfortran.dg/dec_union_8.f90
new file mode 100644
index 0000000..2d856fc
--- /dev/null
+++ b/gcc/testsuite/gfortran.dg/dec_union_8.f90
@@ -0,0 +1,21 @@
+! { dg-do compile }
+! { dg-options "-fdec-structure" }
+!
+! PR fortran/77764
+!
+! Test an ICE due to a map with zero components.
+!
+
+program p
+
+structure /s1/
+  union
+    map
+    end map
+    map
+      real :: a = 2.0
+    end map
+  end union
+end structure
+
+end

             reply	other threads:[~2016-09-29 14:30 UTC|newest]

Thread overview: 2+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2016-09-29 14:45 Fritz Reese [this message]
2016-09-30  8:52 ` Andre Vehreschild

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='CAE4aFAmZJUkUBPFi_jnz-14680y1x=XpjqHiapktvzyVKnGz-A@mail.gmail.com' \
    --to=fritzoreese@gmail.com \
    --cc=fortran@gcc.gnu.org \
    --cc=gcc-patches@gcc.gnu.org \
    /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).