| Message ID | 20101210194949.GT29412@tyan-ft48-01.lab.bos.redhat.com |
|---|---|
| State | New |
| Headers | show |
On Fri, Dec 10, 2010 at 8:49 PM, Jakub Jelinek <jakub@redhat.com> wrote: > Hi! > > The first operand of GIMPLE_DEBUG must be a DECL, not a MEM_REF, so the > DECL -> MEM_REF replacement output_gimple_stmt does is wrong for it. > > Fixed thusly, bootstrapped/regtested on x86_64-linux and i686-linux, ok for > trunk? Ok. Thanks, Richard. > 2010-12-10 Jakub Jelinek <jakub@redhat.com> > > PR lto/46879 > * lto-streamer-out.c (output_gimple_stmt): Never replace first > GIMPLE_DEBUG argument with MEM_REF. > > --- gcc/lto-streamer-out.c.jj 2010-12-02 11:51:31.000000000 +0100 > +++ gcc/lto-streamer-out.c 2010-12-10 17:13:32.000000000 +0100 > @@ -1759,8 +1759,9 @@ output_gimple_stmt (struct output_block > tree op = gimple_op (stmt, i); > /* Wrap all uses of non-automatic variables inside MEM_REFs > so that we do not have to deal with type mismatches on > - merged symbols during IL read in. */ > - if (op) > + merged symbols during IL read in. The first operand > + of GIMPLE_DEBUG must be a decl, not MEM_REF, though. */ > + if (op && (i || !is_gimple_debug (stmt))) > { > tree *basep = &op; > while (handled_component_p (*basep)) > > Jakub >
--- gcc/lto-streamer-out.c.jj 2010-12-02 11:51:31.000000000 +0100 +++ gcc/lto-streamer-out.c 2010-12-10 17:13:32.000000000 +0100 @@ -1759,8 +1759,9 @@ output_gimple_stmt (struct output_block tree op = gimple_op (stmt, i); /* Wrap all uses of non-automatic variables inside MEM_REFs so that we do not have to deal with type mismatches on - merged symbols during IL read in. */ - if (op) + merged symbols during IL read in. The first operand + of GIMPLE_DEBUG must be a decl, not MEM_REF, though. */ + if (op && (i || !is_gimple_debug (stmt))) { tree *basep = &op; while (handled_component_p (*basep))