- 14 Nov, 2011 2 commits
-
-
Matthias Braun authored
-
Matthias Braun authored
-
- 11 Nov, 2011 2 commits
-
-
Andreas Zwinkau authored
-
Matthias Braun authored
-
- 10 Nov, 2011 10 commits
-
-
Matthias Braun authored
-
Matthias Braun authored
This should fix problems with uninitialized backends when be_get_backend_param() wasn't called also calls be_lower_for_target if the user hasn't done so yet.
-
Matthias Braun authored
-
Matthias Braun authored
-
Matthias Braun authored
-
Matthias Braun authored
-
Matthias Braun authored
get_backend_irg_list - none of our backends uses it and in the future the backend should drive the compilation and not give hints to a generic driver reg_class_alignment - unused
-
Matthias Braun authored
-
Matthias Braun authored
-
yb9976 authored
This fixes opt/fehler206.c.
-
- 09 Nov, 2011 4 commits
-
-
Matthias Braun authored
before we only tested, if we produce an output that is used as as input of any schedule node in between. This missed the case where you have unscheduled utility nodes like Sync in between.
-
Matthias Braun authored
-
Matthias Braun authored
The interface was really inefficient and shouldn't be used like that. So simply don't provide functions that make inefficient code convenient.
-
Andreas Seltenreich authored
-
- 08 Nov, 2011 1 commit
-
-
Andreas Seltenreich authored
-
- 07 Nov, 2011 10 commits
-
-
Andreas Zwinkau authored
-
Andreas Zwinkau authored
-
Matthias Braun authored
-
Matthias Braun authored
We missed the case where transform_node created a node that computed_value or equivalent_node would catch
-
Matthias Braun authored
-
Matthias Braun authored
-
Matthias Braun authored
It was a strange feature only inspired by the lazyness of some frontend developer (=me)
-
Matthias Braun authored
they were in lower_mode_b but just generically apply to all Mux(c, 0, 1)
-
Matthias Braun authored
-
Matthias Braun authored
-
- 04 Nov, 2011 9 commits
-
-
Matthias Braun authored
-
Matthias Braun authored
Some peephole optimisations enable others, so splitting into multiple passes makes sense.
-
Matthias Braun authored
A test can exactly be removed if we are only interested in the sign and zero flag. Rewrote and refactored code. Now we don't miss some cases anymore where only sign was requested.
-
Matthias Braun authored
-
Matthias Braun authored
Add op(op(x, c0), c1) => op(x, op(c0, c1)) as localopt rule. This is an very usual and often encountered case so it makes sense to have this working without explicitely invoking the associativity optimisation phase.
-
Matthias Braun authored
when bits are flipped for a bittest we can just as well negate the pnc and don't flip the bits
-
Andreas Zwinkau authored
-
Andreas Zwinkau authored
The preconditions were too tight
-
Matthias Braun authored
-
- 03 Nov, 2011 2 commits
-
-
Matthias Braun authored
-
Matthias Braun authored
-