126 Commits

Author SHA1 Message Date
bjorn3
15a310345d Improve local ir comments 2019-12-22 17:04:33 +01:00
bjorn3
e63a94fca6 Rustup to rustc 1.42.0-nightly (fc5deca21 2019-12-21) 2019-12-22 16:53:06 +01:00
bjorn3
ad72afaad3 Remove CPlaceInner::Stack 2019-12-20 22:00:12 +01:00
bjorn3
8f5ef6172c Remove to_addr and to_addr_maybe_unsized 2019-12-20 19:10:08 +01:00
bjorn3
967709f31a [OPT] Use load and store offsets instead of iadd_imm 2019-12-20 16:02:47 +01:00
bjorn3
bfff2a50fd Fix debug build 2019-12-18 20:25:05 +01:00
bjorn3
1c79cf4c0a [OPT] Use RETURN_PLACE instead of fn_sig.output() to get return layout 2019-12-17 18:41:13 +01:00
bjorn3
86cdd1909b [OPT] Use IndexVec instead of HashMap in ssa analysis 2019-12-17 17:49:12 +01:00
bjorn3
172522c2e7 [OPT] Push fx.monomorphize to the callers of fx.layout_of
This prevents some repeated monomorphizations
2019-12-17 16:58:34 +01:00
bjorn3
d502f8e652 [OPT] Avoid calling target_triple for every call to codegen 2019-12-17 16:34:48 +01:00
bjorn3
c6086a8fd7 Rustup to rustc 1.41.0-nightly (6d77e45f0 2019-12-04) 2019-12-05 21:00:57 +01:00
bjorn3
177b0d2f57 Rustup to rustc 1.41.0-nightly (ded5ee001 2019-11-13) 2019-11-14 21:13:40 +01:00
bjorn3
5407b51aa7 Rustup to rustc 1.40.0-nightly (9e346646e 2019-11-08) 2019-11-09 11:14:18 +01:00
bjorn3
82fde5b622 Remove bitflags dependency 2019-10-06 17:52:23 +02:00
bjorn3
0db605040e Use CallConv::triple_default instead of hard coding SystemV
Fixes #718
2019-09-28 17:00:27 +02:00
bjorn3
44792f1089 Rustup to rustc 1.40.0-nightly (084beb83e 2019-09-27) 2019-09-28 14:06:12 +02:00
bjorn3
fdfb1eda2c Rustup to rustc 1.39.0-nightly (97e58c0d3 2019-09-20) 2019-09-21 11:30:29 +02:00
bjorn3
bfb49e7d19 Rustup to rustc 1.39.0-nightly (6e19f3f38 2019-09-06) 2019-09-07 10:51:00 +02:00
Atul Bhosale
f481a4b685
Format code using 'cargo fmt' 2019-08-31 22:58:09 +05:30
bjorn3
15b9834d7d Don't copy ByRef passed types to local stack slot when not necessary
Eg when the local is immutable **and** the type is freeze.

This makes the simple raytracer runtime benchmark 1% faster than cg_llvm
without optimizations. Before it was 2% slower.

cc #691
cc #684
2019-08-30 15:42:07 +02:00
bjorn3
76d2e085db Improve abi ir comments a bit 2019-08-30 15:07:15 +02:00
bjorn3
d731c4a6a7 Move ir comments generation to abi/comments.rs
Also list locals stored in ssa vars in the comments
2019-08-30 14:21:24 +02:00
bjorn3
deeae2fce4 Small change 2019-08-30 12:42:24 +02:00
bjorn3
16593d264c Move return handling to abi/returning.rs 2019-08-30 12:30:57 +02:00
bjorn3
68dcfc1c78 Move pass mode handling to abi/pass_mode.rs 2019-08-30 11:58:52 +02:00
bjorn3
6427eaf68b Move abi.rs to abi/mod.rs 2019-08-30 11:51:33 +02:00