[Bug c/77767] [5/6/7 Regression] Side-effect from VLA array parameters lost

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=77767

Richard Biener  changed:

   What|Removed |Added

   Priority|P3  |P2

[Bug c/77767] [5/6/7 Regression] Side-effect from VLA array parameters lost

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=77767

Jakub Jelinek  changed:

   What|Removed |Added

 CC||jakub at gcc dot gnu.org

--- Comment #4 from Jakub Jelinek  ---
Created attachment 40377
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=40377&action=edit
gcc7-pr77767.patch

I went through all the grokdeclarator (indirect) callers, and those callers (of
e.g. groktypename, grokparm, push_parm_decl and c_parser_objc_method_decl) that
are called with non-NULL expr actually ensure it is initialized with NULL_TREE
first and want to append to it.
Here is thus an untested patch.  Optionally, the building of COMPOUND_EXPR
could be in both spots (this one and the one in the if (this_size_varies)
handling later on) replaced with append_to_statement_list (whatever, expr);
Any preferences here?

[Bug c/77767] [5/6/7 Regression] Side-effect from VLA array parameters lost

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=77767

--- Comment #5 from Jakub Jelinek  ---
Created attachment 40378
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=40378&action=edit
gcc7-pr77767-2.patch

Version with append_to_statement_list.

[Bug c/77767] [5/6/7 Regression] Side-effect from VLA array parameters lost

[joseph at codesourcery dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=77767

--- Comment #6 from joseph at codesourcery dot com  ---
I think the append_to_statement_list version should be preferred.

[Bug c/77767] [5/6/7 Regression] Side-effect from VLA array parameters lost

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=77767

--- Comment #7 from Jakub Jelinek  ---
Unfortunately that version breaks the pr49120.c testcase, because mark_exp_read
doesn't deal with STATEMENT_LISTs.  I'll test the other patch.

[Bug c/77767] [5/6/7 Regression] Side-effect from VLA array parameters lost

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=77767

--- Comment #8 from Jakub Jelinek  ---
Author: jakub
Date: Wed Dec 21 00:07:49 2016
New Revision: 243832

URL: https://gcc.gnu.org/viewcvs?rev=243832&root=gcc&view=rev
Log:
PR c/77767
* c-decl.c (grokdeclarator): If *expr is non-NULL, append expression
to *expr instead of overwriting it.

* gcc.c-torture/execute/pr77767.c: New test.

Added:
trunk/gcc/testsuite/gcc.c-torture/execute/pr77767.c
Modified:
trunk/gcc/c/ChangeLog
trunk/gcc/c/c-decl.c
trunk/gcc/testsuite/ChangeLog

[Bug c/77767] [5/6/7 Regression] Side-effect from VLA array parameters lost

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=77767

--- Comment #9 from Jakub Jelinek  ---
Author: jakub
Date: Wed Dec 21 12:52:06 2016
New Revision: 243851

URL: https://gcc.gnu.org/viewcvs?rev=243851&root=gcc&view=rev
Log:
PR c/77767
* c-decl.c (grokdeclarator): If *expr is non-NULL, append expression
to *expr instead of overwriting it.

* gcc.c-torture/execute/pr77767.c: New test.

Added:
branches/gcc-6-branch/gcc/testsuite/gcc.c-torture/execute/pr77767.c
Modified:
branches/gcc-6-branch/gcc/c/ChangeLog
branches/gcc-6-branch/gcc/c/c-decl.c
branches/gcc-6-branch/gcc/testsuite/ChangeLog

[Bug c/77767] [5/6/7 Regression] Side-effect from VLA array parameters lost

[jsm28 at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=77767

Joseph S. Myers  changed:

   What|Removed |Added

   Target Milestone|--- |5.5
Summary|Side-effect from VLA array  |[5/6/7 Regression]
   |parameters lost |Side-effect from VLA array
   ||parameters lost

--- Comment #1 from Joseph S. Myers  ---
This appears to be a regression in 4.7 and later, probably caused by r173422 -
https://gcc.gnu.org/ml/gcc-patches/2011-05/msg00319.html .

[Bug c/77767] [5/6/7 Regression] Side-effect from VLA array parameters lost

[mpolacek at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=77767

Marek Polacek  changed:

   What|Removed |Added

 Status|UNCONFIRMED |NEW
   Last reconfirmed||2016-09-29
 CC||mpolacek at gcc dot gnu.org
 Ever confirmed|0   |1

--- Comment #2 from Marek Polacek  ---
Confirmed.

[Bug c/77767] [5/6/7 Regression] Side-effect from VLA array parameters lost

[mpolacek at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=77767

Marek Polacek  changed:

   What|Removed |Added

 Status|NEW |ASSIGNED
   Assignee|unassigned at gcc dot gnu.org  |mpolacek at gcc dot 
gnu.org

--- Comment #3 from Marek Polacek  ---
Mine.