> So shall I take out the CONST_WIDE_INT/CONST_DOUBLE handling and just
> check for CONST_INT_P instead of CONST_SCALAR_INT_P ? I thought it is just
> easy thing to handle, though for DSE which cares about addresses it really
> does not matter. Or can I leave it in?
Your call.
> DSE will only c
On Tue, Jan 19, 2016 at 12:43:53PM +0100, Eric Botcazou wrote:
> > 2016-01-19 Jakub Jelinek
> >
> > PR rtl-optimization/68955
> > PR rtl-optimization/64557
> > * dse.c (record_store, check_mem_read_rtx): Don't call get_addr
> > here. Fix up formatting.
> > * alias.c (get_ad
> 2016-01-19 Jakub Jelinek
>
> PR rtl-optimization/68955
> PR rtl-optimization/64557
> * dse.c (record_store, check_mem_read_rtx): Don't call get_addr
> here. Fix up formatting.
> * alias.c (get_addr): Handle VALUE + CONST_INT.
VALUE +/- CONST_INT (and actually a
On Mon, Jan 18, 2016 at 11:40:23AM +0100, Eric Botcazou wrote:
> > So, do you suggest to tweak get_addr like the patch below, and remove the
> > mem_addr = get_addr (mem_addr);
> > line above and the comment?
>
> Yes, exactly. And if that doesn't easily work, then go for your solution and
> ad
> So, do you suggest to tweak get_addr like the patch below, and remove the
> mem_addr = get_addr (mem_addr);
> line above and the comment?
Yes, exactly. And if that doesn't easily work, then go for your solution and
add a blurb to the comment explaining why get_addr cannot be easily changed.
On Mon, Jan 18, 2016 at 10:06:44AM +0100, Eric Botcazou wrote:
> > The following testcase is miscompiled on i686-linux at -O3.
> > The bug is in DSE record_store, which for group_id < 0 uses mem_addr
> > set to result of get_addr (base->val_rtx) (plus optional offset),
> > which is fine for canon_t
> The following testcase is miscompiled on i686-linux at -O3.
> The bug is in DSE record_store, which for group_id < 0 uses mem_addr
> set to result of get_addr (base->val_rtx) (plus optional offset),
> which is fine for canon_true_dependence with other MEMs in that function,
> but we also store th
Hi!
The following testcase is miscompiled on i686-linux at -O3.
The bug is in DSE record_store, which for group_id < 0 uses mem_addr
set to result of get_addr (base->val_rtx) (plus optional offset),
which is fine for canon_true_dependence with other MEMs in that function,
but we also store that ad
