Jerry DeLisle wrote:
Paul Thomas wrote:
Jerry
:REVIEWMAIL:
The three test cases provided gave three different internal errors
on i686-linux. One on a bad symbol, one on a bad type, and one
segfault.
Putting the three tests in separate procedures would combine them
into one, would it not?
As long as the last one is the one that segfaults, and assuming the
other two don't segfault in a different setting. I suppose the test
case is not all that important for this particular patch.
I will give this a try. OK either way?
Jerry
PING See combined test case attached.
------------------------------------------------------------------------
! { dg-do compile }
! PR27954 Internal compiler error on bad statements
! Derived from test case submitted in PR.
subroutine bad1
character*20 :: y, x 00 ! { dg-error "Syntax error" }
data y /'abcdef'/, x /'jbnhjk'/ pp ! { dg-error "Syntax error" }
end subroutine bad1
subroutine bad2
character*20 :: y, x 00 ! { dg-error "Syntax error" }
data y /'abcdef'/, x /'jbnhjk'/ pp ! { dg-error "Syntax error" }
print *, "basket case."
end subroutine bad2
subroutine bad3
implicit none
character*20 :: y, x 00 ! { dg-error "Syntax error" }
data y /'abcdef'/, x /'jbnhjk'/ pp ! { dg-error "Syntax error" }
print *, "basket case that segfaults without patch."
end subroutine bad3