aboutsummaryrefslogtreecommitdiffstats
path: root/cse.c
Commit message (Expand)AuthorAgeFilesLines
* cse: Size insn_hash_table more realistically, speeding up CSE significantlyJosh Triplett2007-07-291-1/+1
* Fix typos in commentsJosh Triplett2007-03-091-4/+4
* Do stupid and crappy CSE on casts.Linus Torvalds2005-09-241-0/+25
* [PATCH] using 0 as NULL in sparseChristopher Li2005-04-071-3/+3
* Use the new per-instruction position information for betterLinus Torvalds2005-04-071-1/+1
* Split the binops where signedness matters into unsigned and signed.Linus Torvalds2005-04-071-6/+10
* Split OP_SETVAL into OP_SETVAL (fp expressions and labels) and OP_SYMADDRLinus Torvalds2005-04-071-2/+8
* Remove phi source merging.Linus Torvalds2005-04-071-17/+1
* Expose "show_instruction()" for debugging.Linus Torvalds2005-04-071-2/+0
* Be more aggressive on PHI-node CSE.Linus Torvalds2005-04-071-3/+10
* Make OP_PHISOURCE track the OP_PHI instructions that it defines.Linus Torvalds2005-04-071-4/+2
* Yet another missed "entry" change point.Linus Torvalds2005-04-071-1/+1
* Remove OP_SETCC, make OP_SEL bigger instead.Linus Torvalds2005-04-071-6/+10
* Don't try to share parenthood fn between phi node removal and Linus Torvalds2005-04-071-1/+15
* Start using instruction sizes properly.Linus Torvalds2005-04-071-1/+3
* Oops. Don't try to CSE OP_SEL, at least not until we learn toLinus Torvalds2005-04-071-2/+2
* Do CSE over children with trivially common parents.Linus Torvalds2005-04-071-1/+21
* Allow CSE to run after bb packing. Linus Torvalds2005-04-071-1/+5
* Add "memop" simplification phase.Linus Torvalds2005-04-071-1/+2
* Be more careful about insn->bb pointers.Linus Torvalds2005-04-071-0/+1
* Make the CSE "repeat" logic be more fine-grained than justLinus Torvalds2005-04-071-6/+8
* Move instruction simplification to new file "simplify.c".Linus Torvalds2005-04-071-193/+18
* Another assert - verify bb validity.Linus Torvalds2005-04-071-2/+2
* Don't change instruction pseudo's when nopping them out.Linus Torvalds2005-04-071-1/+0
* Damn. We can't actually sort the phi-node list of phi's nowLinus Torvalds2005-04-071-21/+4
* Now that phi sources are just instructions, we can CSE them.Linus Torvalds2005-04-071-0/+12
* Clear phi list when killing a phi-node instructionLinus Torvalds2005-04-071-1/+3
* Oops. Clean up some left-overs from phi removal.Linus Torvalds2005-04-071-8/+10
* Remove "struct phi", replace with instruction that generates a pseudo.Linus Torvalds2005-04-071-24/+27
* Do if-conversion.Linus Torvalds2005-04-071-0/+101
* Re-do CSE if we did a phi-node simplification.Linus Torvalds2005-04-071-12/+11
* We need to pack the phi-list even if we simplify the phiLinus Torvalds2005-04-071-2/+3
* Make phi-node normalization (part of CSE) do trivial simplification.Linus Torvalds2005-04-071-2/+17
* Oops. Don't try to CSE the dead instructions.Linus Torvalds2005-04-071-0/+2
* Kill trivially dead instructionsLinus Torvalds2005-04-071-0/+55
* Make the "cse nop" a bit more informativeLinus Torvalds2005-04-071-0/+1
* Make CSE convert instructions to OP_NOPLinus Torvalds2005-04-071-2/+2
* Add simple-stupid dominance testing for CSE.Linus Torvalds2005-04-071-2/+32
* Pack the phi-list after removing duplicates.Linus Torvalds2005-04-071-0/+3
* Add initial CSE passLinus Torvalds2005-04-071-0/+291