Patch for fortran/62536 and fortran/66175

Patch for fortran/62536 and fortran/66175

[Louis Krupp]

[-- Attachment #1: Type: text/plain, Size: 2232 bytes --]

This patch cleans up nested blocks when there's an unexpected end of a compilation unit (66175), and it handles cleaned-up blocks gracefully (62536).  I've run "make check-fortran" with the attached test cases.

Index: ChangeLog
===================================================================
--- ChangeLog	(revision 226452)
+++ ChangeLog	(working copy)
@@ -1877,6 +1877,13 @@
 	* interface.c (is_procptr_result): New function to check if an
 	expression is a procedure-pointer result.
 	(compare_actual_formal): Use it.
+
+2015_07_31  Louis Krupp <louis.krupp@zoho.com>
+
+    PR fortran/62536 and fortran/66175
+	* decl.c (gfc_match_end): Clean up nested BLOCKs.
+	* parse.c (parse_block_construct): Deal gracefully with cleaned-up
+	BLOCKs.
 ^L
 Copyright (C) 2015 Free Software Foundation, Inc.
 
Index: decl.c
===================================================================
--- decl.c	(revision 226452)
+++ decl.c	(working copy)
@@ -6483,7 +6483,7 @@ cleanup:
   /* If we are missing an END BLOCK, we created a half-ready namespace.
      Remove it from the parent namespace's sibling list.  */
 
-  if (state == COMP_BLOCK)
+  while (state == COMP_BLOCK)
     {
       parent_ns = gfc_current_ns->parent;
 
@@ -6506,6 +6506,8 @@ cleanup:
   
       gfc_free_namespace (gfc_current_ns);
       gfc_current_ns = parent_ns;
+      gfc_state_stack = gfc_state_stack->previous;
+      state = gfc_current_state ();
     }
 
   return MATCH_ERROR;
Index: parse.c
===================================================================
--- parse.c	(revision 226452)
+++ parse.c	(working copy)
@@ -3935,6 +3935,7 @@ static void
 parse_block_construct (void)
 {
   gfc_namespace* my_ns;
+  gfc_namespace* my_parent;
   gfc_state_data s;
 
   gfc_notify_std (GFC_STD_F2008, "BLOCK construct at %C");
@@ -3948,10 +3949,15 @@ parse_block_construct (void)
 
   push_state (&s, COMP_BLOCK, my_ns->proc_name);
   gfc_current_ns = my_ns;
+  my_parent = my_ns->parent;
 
   parse_progunit (ST_NONE);
 
-  gfc_current_ns = gfc_current_ns->parent;
+  /*
+   * Don't depend on the value of gfc_current_ns;  it might have been
+   * reset if the block had errors and was cleaned up.
+   */
+  gfc_current_ns = my_parent;
   pop_state ();
 }



[-- Attachment #2: blocks_nested_incomplete_1.f90 --]
[-- Type: application/octet-stream, Size: 213 bytes --]

! { dg-compile }
!
! PR fortran/66175
! Nested incomplete blocks cause ICE.
program main
  block
    block
end program ! { dg-error "Expecting END BLOCK statement" }
! { dg-prune-output "Unexpected end of file" }

[-- Attachment #3: block_end_error_1.f90 --]
[-- Type: application/octet-stream, Size: 233 bytes --]

! { dg-compile }
!
! PR fortran/62536
! Bad "end block" causes ICE.
subroutine s
   block
   end block named ! { dg-error "Syntax error in END BLOCK statement" }
   return
endsubroutine
! { dg-prune-output "Unexpected end of file" }

Re: Patch for fortran/62536 and fortran/66175

[FX]

> This patch cleans up nested blocks when there's an unexpected end of a compilation unit (66175), and it handles cleaned-up blocks gracefully (62536).  I've run "make check-fortran" with the attached test cases.

The patch is OK.

But the question I missed in my earlier email was: do you have a copyright assignment in place with the FSF?

FX

Re: Patch for fortran/62536 and fortran/66175

[FX]

> No, I don't, but I would be happy to assign copyright. How do I do that? I've read this: https://www.fsf.org/licensing/assigning.html
> and it says I should "email the maintainer of the program communicating [my] desire to assign copyright.”

As I understand it, you need to fill this form: http://git.savannah.gnu.org/cgit/gnulib.git/plain/doc/Copyright/request-assign.future (with “GCC” being the “package you’re contributing to”) and send it by email to assign@gnu.org to get the process started.

In past times, it used to take a little time. I don’t know what the current situation is.

FX

Re: Patch for fortran/62536 and fortran/66175

[FX]

[-- Attachment #1: Type: text/plain, Size: 674 bytes --]

> This patch cleans up nested blocks when there's an unexpected end of a compilation unit (66175), and it handles cleaned-up blocks gracefully (62536).  I've run "make check-fortran" with the attached test cases.

I have now committed the patch to trunk as revision 227135.
Thanks for your contribution! And you’re welcome to repeat that experience anytime :)

FYI, I fixed a few minor issues with the patch before committed it (final version attached):
  - test directive should be { dg-do compile } not { dg-compile }
  - comment reformatting
  - missing gcc/testsuite/ChangeLog entries
  - reformated date and PR entries in ChangeLog


Thanks again,
FX


[-- Attachment #2: block.diff --]
[-- Type: application/octet-stream, Size: 3506 bytes --]

Index: gcc/fortran/ChangeLog
===================================================================
--- gcc/fortran/ChangeLog	(revision 227134)
+++ gcc/fortran/ChangeLog	(working copy)
@@ -1,3 +1,11 @@
+2015-08-24  Louis Krupp <louis.krupp@zoho.com>
+
+	PR fortran/62536
+	PR fortran/66175
+	* decl.c (gfc_match_end): Clean up nested BLOCKs.
+	* parse.c (parse_block_construct): Deal gracefully with cleaned-up
+	BLOCKs.
+
 2015-08-23  Francois-Xavier Coudert  <fxcoudert@gcc.gnu.org>
 
 	PR libfortran/54572
Index: gcc/fortran/decl.c
===================================================================
--- gcc/fortran/decl.c	(revision 227127)
+++ gcc/fortran/decl.c	(working copy)
@@ -6483,7 +6483,7 @@ cleanup:
   /* If we are missing an END BLOCK, we created a half-ready namespace.
      Remove it from the parent namespace's sibling list.  */
 
-  if (state == COMP_BLOCK)
+  while (state == COMP_BLOCK)
     {
       parent_ns = gfc_current_ns->parent;
 
@@ -6506,6 +6506,8 @@ cleanup:
   
       gfc_free_namespace (gfc_current_ns);
       gfc_current_ns = parent_ns;
+      gfc_state_stack = gfc_state_stack->previous;
+      state = gfc_current_state ();
     }
 
   return MATCH_ERROR;
Index: gcc/fortran/parse.c
===================================================================
--- gcc/fortran/parse.c	(revision 227127)
+++ gcc/fortran/parse.c	(working copy)
@@ -3935,6 +3935,7 @@ static void
 parse_block_construct (void)
 {
   gfc_namespace* my_ns;
+  gfc_namespace* my_parent;
   gfc_state_data s;
 
   gfc_notify_std (GFC_STD_F2008, "BLOCK construct at %C");
@@ -3948,10 +3949,14 @@ parse_block_construct (void)
 
   push_state (&s, COMP_BLOCK, my_ns->proc_name);
   gfc_current_ns = my_ns;
+  my_parent = my_ns->parent;
 
   parse_progunit (ST_NONE);
 
-  gfc_current_ns = gfc_current_ns->parent;
+  /* Don't depend on the value of gfc_current_ns;  it might have been
+     reset if the block had errors and was cleaned up.  */
+  gfc_current_ns = my_parent;
+
   pop_state ();
 }
 
Index: gcc/testsuite/ChangeLog
===================================================================
--- gcc/testsuite/ChangeLog	(revision 227134)
+++ gcc/testsuite/ChangeLog	(working copy)
@@ -1,3 +1,10 @@
+2015-08-24  Louis Krupp <louis.krupp@zoho.com>
+
+	PR fortran/62536
+	PR fortran/66175
+	* gfortran.dg/block_end_error_1.f90: New test.
+	* gfortran.dg/blocks_nested_incomplete_1.f90: New test.
+
 2015-08-24  H.J. Lu  <hongjiu.lu@intel.com>
 
 	PR target/66821
Index: gcc/testsuite/gfortran.dg/block_end_error_1.f90
===================================================================
--- gcc/testsuite/gfortran.dg/block_end_error_1.f90	(revision 0)
+++ gcc/testsuite/gfortran.dg/block_end_error_1.f90	(working copy)
@@ -0,0 +1,10 @@
+! { dg-do compile }
+!
+! PR fortran/62536
+! Bad "end block" causes ICE.
+subroutine s
+   block
+   end block named ! { dg-error "Syntax error in END BLOCK statement" }
+   return
+endsubroutine
+! { dg-prune-output "Unexpected end of file" }
Index: gcc/testsuite/gfortran.dg/blocks_nested_incomplete_1.f90
===================================================================
--- gcc/testsuite/gfortran.dg/blocks_nested_incomplete_1.f90	(revision 0)
+++ gcc/testsuite/gfortran.dg/blocks_nested_incomplete_1.f90	(working copy)
@@ -0,0 +1,9 @@
+! { dg-do compile }
+!
+! PR fortran/66175
+! Nested incomplete blocks cause ICE.
+program main
+  block
+    block
+end program ! { dg-error "Expecting END BLOCK statement" }
+! { dg-prune-output "Unexpected end of file" }