repo.or.cz
/
ajla.git
/
search
commit
grep
author
committer
pickaxe
?
search:
re
summary
|
log
|
graphiclog1
|
graphiclog2
|
commit
|
commitdiff
|
tree
|
refs
|
edit
|
fork
first
·
prev
·
next
codegen: remove the third argument of cg_upcall_flat_to_data because it
2024-10-28
Mikulas Patocka
s390: use sign
e
d ta
g
comp
a
risons, so that we could use
commit
|
commitdiff
|
tree
2024-10-28
Mikulas Patocka
codegen: gen_
c
omp
a
re_ptr
_
tag:
i
mpro
v
e generate
d
code
.
.
.
commit
|
commitdiff
|
tree
2024-10-28
Mikulas
P
atoc
k
a
co
d
e
g
en: remember that a pointer
is non-thunk and
do
.
.
.
commit
|
commitdiff
|
tree
2024-10-28
Mikula
s
Pa
t
ocka
codegen: gen_
a
rray_len: use load-op on x
8
6 and s3
9
0
commit
|
commitdiff
|
tree
2024-10-28
Mikulas Patocka
code
g
e
n
: use 2
-
regi
s
ter
jmp in gen_array_len if supported
commit
|
commitdiff
|
tree
2024-10-28
Mikulas Patocka
s
s
a: r
e
move
r
epet
i
tive c
o
d
e
i
n
insert
_
fr
e
e
commit
|
commitdiff
|
tree
2024-10-28
Mikulas Patocka
implemen
t
array_len_g
r
eater_
t
han+jmp
fus
i
on
commit
|
commitdiff
|
tree
2024-10-28
Mikul
a
s
Patock
a
co
d
egen: remove
p
ointless reg
i
ste
r
copy
i
ng in gen_array_len
commit
|
commitdiff
|
tree
2024-10-27
Mikulas Patocka
x86: use the %gs register on NetB
S
D
commit
|
commitdiff
|
tree
2024-10-27
Mikul
a
s Patocka
x86: us
e
the
%gs registe
r
o
n FreeBS
D
commit
|
commitdiff
|
tree
2024-10-27
M
ikulas Patocka
fix a b
u
g that
w
e initiali
z
ed codegen dump_mutex
before
.
.
.
commit
|
commitdiff
|
tree
2024-10-27
Mi
k
ulas Pat
o
cka
ssa: remove
_
redundan
t
_frees: remove point
l
ess cop
i
es
.
.
.
commit
|
commitdiff
|
tree
2024-10-27
Mikulas Patoc
k
a
pc
o
de: fix
bu
g
gy offset being
c
alc
u
lated when P_Jmp_False
.
.
.
commit
|
commitdiff
|
tree
2024-10-27
Mikulas
P
atocka
codege
n
:
improve floating point compariso
n
s on loongar
c
h
.
.
.
commit
|
commitdiff
|
tree
2024-10-27
Mi
k
ulas Patocka
mi
p
s: fix a bug
that
w
e allocat
e
d the ra regis
t
e
r
and
.
.
.
commit
|
commitdiff
|
tree
2024-10-27
Miku
l
as
Pato
c
ka
c
odege
n
:
use 2-r
e
gist
e
r jump to test
f
or inc/dec
o
verflow
commit
|
commitdiff
|
tree
2024-10-27
Mi
k
ulas Patocka
code
g
en:
e
x
te
n
d the registers
i
n-place in gen_frame_get
commit
|
commitdiff
|
tree
2024-10-27
Mik
u
las Pa
t
o
cka
codegen: improve the gen
e
rated co
d
e
for 32-bit inc
.
.
.
commit
|
commitdiff
|
tree
2024-10-27
Miku
l
as Patocka
code
g
en: use 2-register jump no a
r
c
hitectures that
.
.
.
commit
|
commitdiff
|
tree
2024-10-27
Mikulas P
a
toc
k
a
codeg
e
n
:
fi
x
bug o
n
r
i
scv when --ptrco
m
p was used
commit
|
commitdiff
|
tree
2024-10-27
Mikulas Pato
c
k
a
c
o
d
e
ge
n
: impr
o
v
e
code
generation for
3
2-bit values
.
.
.
commit
|
commitdiff
|
tree
2024-10-27
Mikulas Patocka
sparc6
4
: imp
r
ove constant
generating when low 32 bits
.
.
.
commit
|
commitdiff
|
tree
2024-10-27
Mikulas Patocka
code
g
en: improve co
n
stant l
o
ading; don't lo
a
d constants
.
.
.
commit
|
commitdiff
|
tree
2024-10-26
Mikulas Pat
o
cka
sparc:
f
ix breakage i
n
tro
d
uced by cmp+jmp fusion (sp
a
rc64
.
.
.
commit
|
commitdiff
|
tree
2024-10-26
Mikulas
P
atocka
codegen: f
u
s
e
fl
o
a
t
ing point
c
ompari
s
ons
commit
|
commitdiff
|
tree
2024-10-26
Mik
u
las Patocka
x8
6
: fix a floating
point bug if we don't have
AVX
commit
|
commitdiff
|
tree
2024-10-26
Mik
u
las Patocka
codegen: fix a bug when loading fp
c
onstants directly
.
.
.
commit
|
commitdiff
|
tree
2024-10-26
Mikulas P
a
to
c
ka
cod
e
gen:
optimize bool "and
"
and "or"
commit
|
commitdiff
|
tree
2024-10-26
Mik
u
las Patoc
k
a
im
p
lem
e
nt cm
p
+jmp fusi
o
n
commit
|
commitdiff
|
tree
2024-10-26
M
ikula
s
Patocka
ipret: declare OPCODE_ESCAPE_NON
F
LAT and
OPCODE_CHECKP
O
INT
commit
|
commitdiff
|
tree
2024-10-24
Mikulas Pato
c
ka
codegen:
avoid uni
n
itialized "label_
f
inish
"
warning
.
.
.
commit
|
commitdiff
|
tree
2024-10-24
Mikulas Patocka
codegen: de
l
ete cg
-
spi
l
l
.
inc and rear
r
ange functi
o
ns
.
.
.
commit
|
commitdiff
|
tree
2024-10-24
Mikulas Pat
o
c
ka
codegen: in
t
rod
u
ce cg-spill
.
inc
commit
|
commitdiff
|
tree
2024-10-24
Mikulas Patocka
codege
n
:
s
plit
c
odegen
.
c
t
o multiple in
c
luded f
i
le
s
commit
|
commitdiff
|
tree
2024-10-23
Mikulas Pa
t
ocka
al
p
ha: imp
r
o
v
e the code ge
n
erat
e
d in ge
n
_cmp_test_im
m
_jmp
commit
|
commitdiff
|
tree
2024-10-23
Mi
k
ulas
P
atocka
x86: slightly improve
generate
d
code
f
or shifts
commit
|
commitdiff
|
tree
2024-10-23
M
ikulas Patocka
x86:
g
enerate
shifts with mem
o
r
y
argu
m
ent
commit
|
commitdiff
|
tree
2024-10-23
Mikulas
Patoc
k
a
codege
n
: clean up the
d
iv
i
sion code for x86 and s39
0
commit
|
commitdiff
|
tree
2024-10-23
Mikulas Patocka
x86: chan
g
e "return 0" to "return true"
commit
|
commitdiff
|
tree
2024-10-23
Mikulas Patocka
x86: don't
generate
t
he weird "movzww"
i
n
struction
commit
|
commitdiff
|
tree
2024-10-23
M
i
kula
s
P
a
tocka
x86: fix incorrec
t
co
d
e being generated to native-
.
.
.
commit
|
commitdiff
|
tree
2024-10-22
Mikulas Patocka
x86: save one mo
r
e instruction when converting in
t
128
.
.
.
commit
|
commitdiff
|
tree
2024-10-22
M
ikulas Patocka
parisc:
f
ix a cr
a
sh w
h
en attempting
t
o generate a sub
.
.
.
commit
|
commitdiff
|
tree
2024-10-22
Mikulas Patocka
x
8
6
:
u
se t
h
e cqto instruction whe
n
convertin
g
fro
m
.
.
.
commit
|
commitdiff
|
tree
2024-10-22
Miku
l
as Patocka
codegen: remove superfluo
u
s mo
v
e instru
c
tion whe
n
converti
n
g
.
.
.
commit
|
commitdiff
|
tree
2024-10-22
M
ikulas Patocka
x86:
im
p
lemen
t
128-bit bsf
commit
|
commitdiff
|
tree
2024-10-22
Mikulas Patocka
x
8
6
: implem
e
nt 128-bit bsr
commit
|
commitdiff
|
tree
2024-10-22
Mikul
a
s Pa
t
ocka
x86
:
i
m
plement 128-bit popcnt
commit
|
commitdiff
|
tree
2024-10-22
Mikulas
P
a
t
ocka
x86
:
d
on't set op_size > O
P
_S
I
ZE_NAT
I
VE because
the
.
.
.
commit
|
commitdiff
|
tree
2024-10-22
Mikulas Pat
o
cka
x86: optimi
z
e 12
8
-bit
bt in
s
t
r
uction
commit
|
commitdiff
|
tree
2024-10-22
Mikulas Pato
c
ka
x86:
o
ptimize
1
28-b
i
t bts, btr, b
t
c
commit
|
commitdiff
|
tree
2024-10-22
Mikulas Patocka
codegen: alu: allow slot_
1
==
slot_2 in no-trapping
.
.
.
commit
|
commitdiff
|
tree
2024-10-22
Mikulas
P
a
t
o
c
ka
ssa: op
t
imize m
u
ltiply by 2
commit
|
commitdiff
|
tree
2024-10-22
M
ik
u
las Patocka
x86: optimi
z
e multiplica
t
ion
commit
|
commitdiff
|
tree
2024-10-22
M
i
kulas Patock
a
x86: fix a bug in
8
-
bit mul
t
i
p
l
i
cat
i
o
n
commit
|
commitdiff
|
tree
2024-10-22
Mikulas
P
a
tocka
codegen: supp
o
r
t
optim
i
z
ed alu
1
f
or ar
c
hitec
t
ures that
.
.
.
commit
|
commitdiff
|
tree
2024-10-22
Mi
k
ul
a
s Patocka
c
odegen: support optimize
d
alu for architectures
t
hat
.
.
.
commit
|
commitdiff
|
tree
2024-10-22
Mikulas Pato
c
ka
codegen: use t
h
e and inst
r
uction when d
o
ing z
e
ro
-
ext
e
nd
commit
|
commitdiff
|
tree
2024-10-22
Mikulas Patoc
k
a
s390: g
e
n
e
r
ate load
-
op
w
ith the undo label
commit
|
commitdiff
|
tree
2024-10-21
Mikulas
P
atocka
s
p
arc
:
fi
x
failur
e
in the
n
e
w undo code
commit
|
commitdiff
|
tree
2024-10-21
Mikulas Pato
c
k
a
po
w
e
r: fix f
a
ilure when size < O
P
_SIZE_NA
T
IVE
commit
|
commitdiff
|
tree
2024-10-21
Mikulas P
a
tocka
x86
:
implement undo for l
o
a
d-op
a
nd load-op-store instruc
t
ions
commit
|
commitdiff
|
tree
2024-10-21
Mikulas Pa
t
ocka
cod
e
gen: imple
m
ent undo logic for alu
commit
|
commitdiff
|
tree
2024-10-21
Mikulas
P
atocka
c
o
deg
e
n: we
d
on't nee
d
to undo
t
h
e "
n
eg" instruc
t
io
n
commit
|
commitdiff
|
tree
2024-10-21
Mi
k
ulas Pa
t
ocka
codegen: support
a
l
u1 undo logic f
o
r other architectures
commit
|
commitdiff
|
tree
2024-10-21
Mik
u
l
as Patocka
x86:
gen
e
rat
e
inc
,
dec
,
n
ot, neg wit
h
memory operand
commit
|
commitdiff
|
tree
2024-10-21
Mikulas Pa
t
ocka
codeg
e
n: impleme
n
t
u
ndo logi
c
for alu1
commit
|
commitdiff
|
tree
2024-10-21
Mikul
a
s Patocka
codegen
:
ref
a
c
t
or escape_labels
,
so that we
can s
t
ore
.
.
.
commit
|
commitdiff
|
tree
2024-10-21
M
ik
u
las Patocka
ssa: delete unused labe
l
skip_m
u
st_be_flat
commit
|
commitdiff
|
tree
2024-10-21
M
i
k
ulas Patocka
codegen: a smal
l
im
p
ro
v
ement in do_bsw
a
p a
n
d do_brev
commit
|
commitdiff
|
tree
2024-10-21
Mik
u
la
s
Patocka
codegen: a
sm
a
ll improvement in gen_a
l
u
1
commit
|
commitdiff
|
tree
2024-10-21
Mi
k
ulas
P
a
tocka
c
o
degen: improve gen
_
co
p
y so that it c
a
n ge
n
erate
a
.
.
.
commit
|
commitdiff
|
tree
2024-10-21
Mi
k
ulas Patocka
ss
a
: use bsf
i
nstead of bsr
w
he
n
allocat
i
ng variables
.
.
.
commit
|
commitdiff
|
tree
2024-10-20
Mikulas Patocka
m
ips: fix
bsr
a
nd popcnt
commit
|
commitdiff
|
tree
2024-10-20
Mikulas Patock
a
alpha: fix bsr and popcnt
commit
|
commitdiff
|
tree
2024-10-20
Miku
l
as Patocka
arm64: fix compi
l
ation
failure in gen_
e
x
tend
commit
|
commitdiff
|
tree
2024-10-20
Mikulas Patocka
c
o
degen:
u
nspill return value bef
o
r
e testing for
overfl
o
w
commit
|
commitdiff
|
tree
2024-10-20
Mikulas Pato
c
ka
parisc: fix a bug in bi
t
wise oper
a
tions
commit
|
commitdiff
|
tree
2024-10-20
Mikulas
P
atocka
arm: gen
e
rat
e
sxtb, sxt
h
uxtb, ux
t
h
commit
|
commitdiff
|
tree
2024-10-20
Mikulas Patocka
ia64: fi
x
compilati
o
n failure in
do_from_int
commit
|
commitdiff
|
tree
2024-10-20
Mi
k
ulas Patocka
alpha:
f
ix compi
l
ation failure in do_cv
t
_to_int
commit
|
commitdiff
|
tree
2024-10-20
M
i
kulas Pat
o
cka
codegen:
u
s
e gen_frame_tar
g
et
i
n gen_a
r
ray_len
commit
|
commitdiff
|
tree
2024-10-20
M
i
kulas Patocka
code
g
en
:
use
g
e
n_frame_target in gen_option_ord
commit
|
commitdiff
|
tree
2024-10-20
Mik
u
las P
a
tocka
codegen: u
s
e gen_frame_ta
r
g
e
t in cv
t
from int and is_exception
commit
|
commitdiff
|
tree
2024-10-20
Mik
u
las P
a
t
o
cka
codegen: use gen_frame
_
target i
n
cvt to int
commit
|
commitdiff
|
tree
2024-10-20
Mi
k
ula
s
Patocka
codegen: use gen_
f
rame_
t
arget in g
e
n_fp_alu1
commit
|
commitdiff
|
tree
2024-10-20
M
i
kulas Patocka
co
d
eg
e
n:
use
g
en_frame_target in g
e
n_fp_alu
commit
|
commitdiff
|
tree
2024-10-20
Mikulas Patoc
k
a
codegen: use gen_fram
e
_target
for more architectures
commit
|
commitdiff
|
tree
2024-10-20
Mikulas P
a
tocka
codegen: u
s
e gen_
f
r
ame_tar
g
et
fo
r
x86
a
nd arm b
i
t
s
c
ans
commit
|
commitdiff
|
tree
2024-10-20
Mikulas Patocka
codegen: us
e
g
en_fra
m
e_s
t
ore
f
or for
b
s
wap and brev
commit
|
commitdiff
|
tree
2024-10-20
Mi
k
ulas Patocka
codegen
:
use gen_frame
_
t
arget for gen_alu
1
commit
|
commitdiff
|
tree
2024-10-20
M
ikula
s
P
atoc
k
a
power: use 32-bit mult
i
plicatio
n
commit
|
commitdiff
|
tree
2024-10-20
Mikulas Patocka
power:
e
n
a
ble
3
2-bit cmp
commit
|
commitdiff
|
tree
2024-10-20
Mi
k
ulas
P
at
o
c
k
a
power: fix
l
o
g
ic
e
rro
r
in g
e
n_fra
m
e_load_cmp
commit
|
commitdiff
|
tree
2024-10-20
Miku
l
as Pa
t
o
cka
codegen: use gen_frame_target in gen_frame_set_con
d
commit
|
commitdiff
|
tree
2024-10-20
Miku
l
as
P
a
tocka
codeg
e
n
: use gen_frame_
t
arget for generic bit o
p
erations
commit
|
commitdiff
|
tree
2024-10-20
Mik
u
la
s
Patocka
codegen: change "1U <
<
(op_size
+
3
)" to
"8U
<
< op_s
i
z
e"
commit
|
commitdiff
|
tree
2024-10-20
Mikulas Patocka
code
g
en: us
e
gen
_
frame_t
a
rget for
b
it operat
i
o
ns
commit
|
commitdiff
|
tree
2024-10-20
Miku
l
as Patocka
x86: use "force_vex" instea
d
of
0
x400
0
commit
|
commitdiff
|
tree
2024-10-20
M
i
k
ulas Patocka
c
odegen: add DEBUG_GARBAGE to
g
en_fra
m
e_load
commit
|
commitdiff
|
tree
next