fix remove_empty_blocks (don't use a generic walker it will get confused)
[libfirm] / ir /
2007-10-29 Matthias Braunfix remove_empty_blocks (don't use a generic walker...
2007-10-29 Matthias Braunpeephole optimisation: reuse zero flag from previous...
2007-10-29 Matthias Braunfix bugs when exchanging nodes to projs in bepeephole
2007-10-29 Christoph Mallonfehler103: Unusual 64bit shift triggers assert in const...
2007-10-29 Michael Beckadd special case for lowering: x==/!=0 ==> l|h ==/!=0
2007-10-28 Michael Becktestcase for rts library call optimization
2007-10-28 Matthias Braunfix bug introduces in last commit
2007-10-28 Matthias Brauncompile without warnings when WITH_JVM is not set
2007-10-28 Matthias Brauncompile without warnings when WITH_JVM is not set
2007-10-28 Matthias Braunremove (x*A + A -> (x+1) * A) optimisation. As reassoc...
2007-10-28 Matthias Braunmore interesting lea problems
2007-10-28 Matthias Braunend block can also have 0 predecessors
2007-10-28 Matthias Braunimprove asm test
2007-10-26 Michael Beckfixed handling of pn_Call_X_regular
2007-10-26 Michael BeckC99 constructs removed
2007-10-26 Matthias Braunmoved ia32 architecture configuration to a separate...
2007-10-26 Matthias Braunsmall optimisation and cleanup
2007-10-26 Matthias Braunmake IS_ macros return 0 or 1
2007-10-26 Matthias Braunadded some comments, don't make backedges fallthrough...
2007-10-25 Michael Beckupdated cpu architectures
2007-10-25 Michael Beck- invalidate loop info if graph was changed
2007-10-25 Michael Beckinvalidate analysis info only if graph was changed
2007-10-25 Christoph MallonSet the outs to inconsistent at the end of transform_irg().
2007-10-25 Michael BeckC99 constructs removed
2007-10-25 Michael Beckfixed typo in Phi optimization
2007-10-25 Michael BeckFixed Sub(Conv, Conv) optimization
2007-10-25 Christoph MallonFix debug output.
2007-10-25 Christoph MallonSmall cleanups.
2007-10-25 Christoph MallonUse add_ia32_flags() instead of set_ia32_flags(), which...
2007-10-25 Matthias Braunfix parameter loads not being rematerialized
2007-10-25 Michael Beckbugfix: the Op(Phi,Phi) optimization require both Phis...
2007-10-25 Christoph MallonHandle integer Abs for modes which are not 32bits wide.
2007-10-25 Christoph Mallonfehler102: ia32 backend cannot handle integer Abs with...
2007-10-25 Christoph MallonIn match_arguments() assert that match_am is set when...
2007-10-25 Christoph MallonRemove unused variable.
2007-10-24 Michael Beckmore Phi optimizations for Div, Mod, DivMod
2007-10-24 Michael Beckfixed test
2007-10-24 Michael Beckadd Op(Phi(const, const, ...), Phi(const, const, ....
2007-10-24 Michael Beckadd test for new phi optimization
2007-10-24 Michael Beckadd more comments
2007-10-24 Michael Becksame fix again on another place
2007-10-24 Sebastian HackA more general color priorization.
2007-10-24 Michael Beckcheck the opcode again after an optimize_node()
2007-10-24 Christoph MallonEmit the "t" suffix for f96 numbers in the ia32 backend.
2007-10-24 Christoph MallonPad the names of fehler*.c with leading zeros for sorting.
2007-10-24 Christoph MallonGive direct visual feedback by printing "FAILED" when...
2007-10-24 Michael Becklong double x87 modes don't support AM
2007-10-24 Michael Beckimproved fehler101.c. Now with printing!
2007-10-24 Michael Beckcannot output suffix for long double store
2007-10-24 Michael Beckbeware: when lowering Phi's, put them into the block...
2007-10-24 Christoph MallonRestore the AM fold behavior of prior to r16324. It...
2007-10-24 Christoph MallonActually implement what r16324 promised.
2007-10-24 Sebastian HackPrioritize chunks with constraint nodes more
2007-10-23 Christoph MallonImprove the AM folding heuristics: Do not fold AM if...
2007-10-23 Christoph MallonRemove dead code, remove pointless marking of memory...
2007-10-23 Christoph MallonRemove the unused function mode_needs_gp_reg().
2007-10-23 Christoph MallonClarify a comment.
2007-10-23 Christoph Mallon- Allow an arbitrary (for arbitrary < 32) number of...
2007-10-23 Christoph MallonDo no parse {,!}out_rBAR constraints anymore. The const...
2007-10-23 Michael BeckAdd a main to let it compile.
2007-10-23 Michael Beckadded initialization to suppress (wrong) warning
2007-10-23 Michael Beckremoved useless calculation of dominance, debug mask...
2007-10-23 Michael Beckeven better fix for r16314
2007-10-23 Michael Beckinitialize variable to supress (wrong) warning
2007-10-23 Michael Becktemporary disable "manual" spill due to probles with...
2007-10-23 Michael Becknow even buggier
2007-10-23 Michael Beckwrong spill ?
2007-10-22 Michael Beckadd CopyB support to get_op_fragile_mem()
2007-10-22 Michael BeckLet dfs() discover only memory nodes
2007-10-22 Michael Beck- check if a Load has a regular exec before chenging it
2007-10-22 Michael Beckbeware of loops without Phis
2007-10-22 Michael Beckfixed the loop optimization
2007-10-22 Michael Beckheur5 not available yet ;-)
2007-10-22 Sebastian HackSome heur4 tweaks
2007-10-22 Matthias Braunsome more code I played around with lately
2007-10-22 Michael Becktransform the macro block header
2007-10-22 Michael Beckpreliminary version of moving Loads out of loops
2007-10-22 Michael Beckadd a short cut to the disabiguator: pointer and non...
2007-10-22 Michael BeckTreat Start as a forking op in remove_critical edges...
2007-10-21 Matthias Braundon't aggressively fold some nodes when we know it...
2007-10-21 Matthias Braunfix address mode calculation: we can not eat immediates...
2007-10-21 Matthias Braun- fix a conceptual bug in peephole, we need a callback...
2007-10-20 Christoph MallonFix r16293.
2007-10-19 Christoph MallonMake the emit sequences of ShrD and ShlD probably more...
2007-10-19 Christoph MallonUndo r16291.
2007-10-19 Michael Beckuse get_irn_ne() instead of (direct) call to get_irn_link()
2007-10-19 Christoph MallonGeneralise (and fix) sorting of operands by constness...
2007-10-19 Michael Beckfixed typos
2007-10-19 Christoph MallonNow even correct! (see r16284)
2007-10-19 Michael BeckMake the ir_loop link field permanent. The other behavi...
2007-10-19 Matthias Braunfix Cmp input normalisation
2007-10-19 Christoph MallonA do-while loop is nicer than a for loop here.
2007-10-19 Christoph MallonFix fehler99: always skip at least one node and check...
2007-10-19 Christoph Mallonfehler99: defective copykeep when ShrD has to identical...
2007-10-19 Michael Beckadd missing initialization
2007-10-19 Michael Beckfixed warnings in release mode
2007-10-19 Michael Beckto shut up gcc in release mode
2007-10-19 Michael Beckrenamed type_state to ir_type_state
2007-10-19 Michael Beckrestrict some Cmp related optimizations to integer...
2007-10-19 Matthias BraunA+X == A -> X==0 optimisations
next