------- Additional Comments From jakub at gcc dot gnu dot org 2005-04-12 18:01
-------
For those who want a quick workaround, changing khtmllayout.h to:
- Length(LengthType t) { l.value = 0; l.type = t; l.quirk = 0; }
+ Length(LengthType t) { _length = 0; l.type = t; }
Length(int v, LengthType t, bool q=false)
- { l.value = v; l.type = t; l.quirk = q; }
+ { _length = 0; l.value = v; l.type = t; l.quirk = q; }
makes the bug go away. The first changed line actually might result in
better generated code with the current GCC's deficiencies in bitfield handling.
I guess we need a bitfield optimization pass near the end of tree passes that
will figure out bitfield accesses to the same underlying field and change that
into arithmetics on a union or something. Don't even look at assembly for
that RTL optimizers can't figure out, but a tree optimizer could turn that
into 2 instructions:
struct { int a:2,b:2,c:2,d:2,e:2,f:2,g:2,h:2,i:2,j:2,k:2,l:2,m:2,n:2;} s;
void foo (void)
{
s.a |= 1; s.b &= 1; s.c |= 1; s.d &= 1; s.e |= 1; s.f &= 1; s.g |= 1;
s.h &= 1; s.i |= 1; s.j &= 1; s.k |= 1; s.l &= 1; s.m |= 1; s.n &= 1;
}
--
What |Removed |Added
----------------------------------------------------------------------------
Status|UNCONFIRMED |NEW
Ever Confirmed| |1
Last reconfirmed|0000-00-00 00:00:00 |2005-04-12 18:01:07
date| |
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=20973
