- 17 Sep, 2014 6 commits
-
-
Matthias Braun authored
The previous approach put the irg link into the block attributes, so while saving memory an extra indirection was necessary. Some testing with valgrind massif+callgrind revealed that a direct link increses peak memory usage by ~1% while also increasing runtime by ~1%. More importantly the code gets a bit simpler and some strange corner cases can be removed with the direct link.
-
Matthias Braun authored
-
Matthias Braun authored
-
Matthias Braun authored
-
Matthias Braun authored
-
Matthias Braun authored
-
- 15 Sep, 2014 11 commits
-
-
Tobias Rapp authored
-
Matthias Braun authored
Make the list scheduler a set of helper functions, instead of a complex design with a set of callbacks. Simplified list scheduler code.
-
Matthias Braun authored
In their current state they produce very bad code and are just a maintenance burden.
-
Matthias Braun authored
-
Matthias Braun authored
-
Matthias Braun authored
-
Matthias Braun authored
-
Matthias Braun authored
-
Matthias Braun authored
-
Matthias Braun authored
-
Matthias Braun authored
-
- 12 Sep, 2014 1 commit
-
-
yb9976 authored
This fixes backend/sparc_segfault.c.
-
- 09 Sep, 2014 6 commits
-
-
yb9976 authored
-
Matthias Braun authored
This fixes a subtle bug where the delay slot filler would move a spill/reload across a MemPerm although they both share a spillslot.
-
Manuel Mohr authored
The global analysis has quadratic worst-case complexity in the number of graphs. At least if alias analysis is disabled anyway, this can be avoided.
-
Tobias Rapp authored
-
Matthias Braun authored
-
Matthias Braun authored
-
- 08 Sep, 2014 2 commits
-
-
Matthias Braun authored
This avoids strangely large positive numbers instead of negative offsets on x86_64 hosts.
-
Matthias Braun authored
Privously nodes that only had a keep-alive users would be moved up but never down again.
-
- 04 Sep, 2014 1 commit
-
-
Tobias Rapp authored
-
- 03 Sep, 2014 1 commit
-
-
Tobias Rapp authored
-
- 29 Aug, 2014 3 commits
-
-
Manuel Mohr authored
-
Manuel Mohr authored
-
Manuel Mohr authored
-
- 28 Aug, 2014 4 commits
-
-
Andreas Fried authored
-
Andreas Fried authored
-
Andreas Fried authored
-
Andreas Fried authored
-
- 22 Aug, 2014 1 commit
-
-
yb9976 authored
-
- 21 Aug, 2014 4 commits
-
-
Matthias Braun authored
-
Matthias Braun authored
-
Matthias Braun authored
The problem was that we have to force SSA construction for memory values to ensure the presence of the Phi[loop] nodes. Previously we did not do it if we didn't introduce new values. This fixes firm-testsuite/opt/jumpthreading8.c
-
Matthias Braun authored
This should really fix opt/fehler218.c oddities.
-