Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

merge specializations and tfunc #15918

Merged
merged 1 commit into from
Jun 11, 2016
+121 −125
Merged

merge specializations and tfunc #15918

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
117 changes: 68 additions & 49 deletions base/inference.jl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -38,7 +38,6 @@ immutable Const
end

type InferenceState
atypes #::Type # type sig
sp::SimpleVector # static parameters
label_counter::Int # index of the current highest label for this function
fedbackvars::Dict{SSAValue, Bool}
Expand Down Expand Up @@ -78,6 +77,7 @@ type InferenceState

function InferenceState(linfo::LambdaInfo, atypes::ANY, sparams::SimpleVector, optimize::Bool)
@assert isa(linfo.code,Array{Any,1})
linfo.inInference = true
nslots = length(linfo.slotnames)
nl = label_counter(linfo.code)+1

Expand Down Expand Up @@ -156,7 +156,7 @@ type InferenceState

inmodule = isdefined(linfo, :def) ? linfo.def.module : current_module() # toplevel thunks are inferred in the current module
frame = new(
atypes, sp, nl, Dict{SSAValue, Bool}(), inmodule, 0, false,
sp, nl, Dict{SSAValue, Bool}(), inmodule, 0, false,
linfo, linfo, la, s, Union{}, W, n,
cur_hand, handler_at, n_handlers,
ssavalue_uses, ssavalue_init,
Expand Down Expand Up @@ -768,7 +768,7 @@ function abstract_call_gf_by_type(f::ANY, argtype::ANY, sv)
limitlength = false
for (callee, _) in sv.edges
callee = callee::InferenceState
if method === callee.linfo.def && ls > length(callee.atypes.parameters)
if method === callee.linfo.def && ls > length(callee.linfo.specTypes.parameters)
limitlength = true
break
end
Expand All @@ -781,16 +781,16 @@ function abstract_call_gf_by_type(f::ANY, argtype::ANY, sv)
infstate = infstate::InferenceState
if isdefined(infstate.linfo, :def) && method === infstate.linfo.def
td = type_depth(sig)
if ls > length(infstate.atypes.parameters)
if ls > length(infstate.linfo.specTypes.parameters)
limitlength = true
end
if td > type_depth(infstate.atypes)
if td > type_depth(infstate.linfo.specTypes)
# impose limit if we recur and the argument types grow beyond MAX_TYPE_DEPTH
if td > MAX_TYPE_DEPTH
sig = limit_type_depth(sig, 0, true, [])
break
else
p1, p2 = sig.parameters, infstate.atypes.parameters
p1, p2 = sig.parameters, infstate.linfo.specTypes.parameters
if length(p2) == ls
limitdepth = false
newsig = Array{Any}(ls)
Expand Down Expand Up @@ -1117,7 +1117,7 @@ function abstract_eval(e::ANY, vtypes::VarTable, sv::InferenceState)
end
elseif isleaftype(t)
t = Type{t}
elseif isleaftype(sv.atypes)
elseif isleaftype(sv.linfo.specTypes)
if isa(t,TypeVar)
t = Type{t.ub}
else
Expand Down Expand Up @@ -1386,7 +1386,8 @@ end

# create a specialized LambdaInfo from a method
function specialize_method(method::Method, types::ANY, sp::SimpleVector)
li = ccall(:jl_get_specialized, Any, (Any, Any, Any), method, types, sp)::LambdaInfo
li = ccall(:jl_get_specialized, Ref{LambdaInfo}, (Any, Any, Any), method, types, sp)
return li
end

# create copies of any field that type-inference might modify
Expand All @@ -1408,38 +1409,54 @@ function unshare_linfo!(li::LambdaInfo)
end

#### entry points for inferring a LambdaInfo given a type signature ####

function typeinf_edge(method::Method, atypes::ANY, sparams::SimpleVector, needtree::Bool, optimize::Bool, cached::Bool, caller)
local code = nothing
local frame = nothing
offs = 0
# check cached t-functions
# aggregate all saved type inference data there
if cached && !is(method.tfunc, nothing)
code = ccall(:jl_tfunc_cache_lookup, Any, (Any, Any, Int8), method, atypes, offs)
if isa(code, InferenceState)
# inference on this signature is in progress
frame = code
if isa(caller, LambdaInfo)
# record the LambdaInfo where this result should be cached when it is finished
@assert frame.destination === frame.linfo || frame.destination === caller
frame.destination = caller
end
elseif isa(code, Type)
# sometimes just a return type is stored here. if a full AST
# is not needed, we can return it.
if !needtree
return (nothing, code, true)
end
elseif isa(code,LambdaInfo)
@assert code.inferred
return (code, code.rettype, true)
else
# otherwise this is an InferenceState from a different bootstrap stage's
# copy of the inference code; ignore it.
# check cached specializations
# for an existing result stored there
if cached
if !is(method.specializations, nothing)
code = ccall(:jl_specializations_lookup, Any, (Any, Any), method, atypes)
if isa(code, Void)
# something completely new
elseif isa(code, LambdaInfo)
# something existing
if code.inferred
return (code, code.rettype, true)
end
else
# sometimes just a return type is stored here. if a full AST
# is not needed, we can return it.
typeassert(code, Type)
if !needtree
return (nothing, code, true)
end
code = nothing
end
end

if isa(code, LambdaInfo) && code.inInference
# inference on this signature may be in progress,
# find the corresponding frame in the active list
for infstate in active
infstate === nothing && continue
infstate = infstate::InferenceState
if code === infstate.linfo
frame = infstate
break
end
end
end
end

if isa(caller, LambdaInfo)
code = caller
end

if frame === nothing
# inference not started yet, make a new frame for a new lambda
# add lam to be inferred and record the edge

if caller === nothing && needtree && in_typeinf_loop
# if the caller needed the ast, but we are already in the typeinf loop
# then just return early -- we can't fulfill this request
Expand Down Expand Up @@ -1468,9 +1485,11 @@ function typeinf_edge(method::Method, atypes::ANY, sparams::SimpleVector, needtr
return (nothing, Union{}, false)
end
end
# add lam to be inferred and record the edge
if isa(caller, LambdaInfo)
linfo = caller

if isa(code, LambdaInfo)
# reuse the existing code object
linfo = code
@assert typeseq(linfo.specTypes, atypes)
elseif method.isstaged
if !isleaftype(atypes)
# don't call staged functions on abstract types.
Expand All @@ -1488,16 +1507,18 @@ function typeinf_edge(method::Method, atypes::ANY, sparams::SimpleVector, needtr
linfo = specialize_method(method, atypes, sparams)
end
# our stack frame inference context
frame = InferenceState(unshare_linfo!(linfo), atypes, sparams, optimize)

frame = InferenceState(unshare_linfo!(linfo::LambdaInfo), atypes, sparams, optimize)
if cached
tfunc_bp = ccall(:jl_tfunc_cache_insert, Ref{TypeMapEntry}, (Any, Any, Any, Int8), method, atypes, frame, offs)
frame.tfunc_bp = tfunc_bp
frame.tfunc_bp = ccall(:jl_specializations_insert, Ref{TypeMapEntry}, (Any, Any, Any), method, atypes, linfo)
end
end
frame = frame::InferenceState

if !isa(caller, Void) && !isa(caller, LambdaInfo)
@assert isa(caller, InferenceState)
# if we were called from inside inference,
# the caller will be the InferenceState object
# for which the edge was required
caller = caller::InferenceState
if haskey(caller.edges, frame)
Ws = caller.edges[frame]::Vector{Int}
if !(caller.currpc in Ws)
Expand Down Expand Up @@ -1869,10 +1890,6 @@ end
# inference completed on `me`
# update the LambdaInfo and notify the edges
function finish(me::InferenceState)
# lazy-delete the item from active for several reasons:
# efficiency, correctness, and recursion-safety
nactive[] -= 1
active[findlast(active, me)] = nothing
for (i,_) in me.edges
@assert (i::InferenceState).fixedpoint
end
Expand Down Expand Up @@ -1941,9 +1958,11 @@ function finish(me::InferenceState)
out.pure = me.linfo.pure
out.inlineable = me.linfo.inlineable
end
if me.tfunc_bp !== nothing
me.tfunc_bp.func = me.linfo
end

# lazy-delete the item from active for several reasons:
# efficiency, correctness, and recursion-safety
nactive[] -= 1
active[findlast(active, me)] = nothing

# update all of the callers by traversing the backedges
for (i,_) in me.backedges
Expand Down
3 changes: 1 addition & 2 deletions src/alloc.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -566,14 +566,13 @@ JL_DLLEXPORT jl_method_t *jl_new_method_uninit(void)
jl_method_t *m =
(jl_method_t*)newobj((jl_value_t*)jl_method_type,
NWORDS(sizeof(jl_method_t)));
m->tfunc.unknown = jl_nothing;
m->specializations.unknown = jl_nothing;
m->sig = NULL;
m->tvars = NULL;
m->ambig = NULL;
m->roots = NULL;
m->module = jl_current_module;
m->lambda_template = NULL;
m->specializations = NULL;
m->name = NULL;
m->file = null_sym;
m->line = 0;
Expand Down
15 changes: 6 additions & 9 deletions src/dump.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -725,7 +725,7 @@ static void jl_serialize_value_(ios_t *s, jl_value_t *v)
arraylist_push(&reinit_list, (void*)pos);
arraylist_push(&reinit_list, (void*)3);
}
if (jl_is_method(v) && jl_typeof(((jl_method_t*)v)->tfunc.unknown) == (jl_value_t*)jl_typemap_level_type) {
if (jl_is_method(v) && jl_typeof(((jl_method_t*)v)->specializations.unknown) == (jl_value_t*)jl_typemap_level_type) {
arraylist_push(&reinit_list, (void*)pos);
arraylist_push(&reinit_list, (void*)4);
}
Expand Down Expand Up @@ -835,7 +835,7 @@ static void jl_serialize_value_(ios_t *s, jl_value_t *v)
else if (jl_is_method(v)) {
writetag(s, jl_method_type);
jl_method_t *m = (jl_method_t*)v;
union jl_typemap_t *tf = &m->tfunc;
union jl_typemap_t *tf = &m->specializations;
if (tf->unknown && tf->unknown != jl_nothing) {
// go through the t-func cache, replacing ASTs with just return
// types for abstract argument types. these ASTs are generally
Expand All @@ -844,7 +844,6 @@ static void jl_serialize_value_(ios_t *s, jl_value_t *v)
}
jl_serialize_value(s, tf->unknown);
jl_serialize_value(s, (jl_value_t*)m->name);
jl_serialize_value(s, (jl_value_t*)m->specializations);
write_int8(s, m->isstaged);
jl_serialize_value(s, (jl_value_t*)m->file);
write_int32(s, m->line);
Expand Down Expand Up @@ -1445,12 +1444,10 @@ static jl_value_t *jl_deserialize_value_(ios_t *s, jl_value_t *vtag, jl_value_t
NWORDS(sizeof(jl_method_t)));
if (usetable)
arraylist_push(&backref_list, m);
m->tfunc.unknown = jl_deserialize_value(s, (jl_value_t**)&m->tfunc);
jl_gc_wb(m, m->tfunc.unknown);
m->specializations.unknown = jl_deserialize_value(s, (jl_value_t**)&m->specializations);
jl_gc_wb(m, m->specializations.unknown);
m->name = (jl_sym_t*)jl_deserialize_value(s, NULL);
jl_gc_wb(m, m->name);
m->specializations = (jl_array_t*)jl_deserialize_value(s, (jl_value_t**)&m->specializations);
if (m->specializations) jl_gc_wb(m, m->specializations);
m->isstaged = read_int8(s);
m->file = (jl_sym_t*)jl_deserialize_value(s, NULL);
m->line = read_int32(s);
Expand Down Expand Up @@ -1834,9 +1831,9 @@ static void jl_reinit_item(ios_t *f, jl_value_t *v, int how, arraylist_t *tracee
arraylist_push(tracee_list, mt);
break;
}
case 4: { // rehash tfunc
case 4: { // rehash specializations tfunc
jl_method_t *m = (jl_method_t*)v;
jl_typemap_rehash(m->tfunc, 0);
jl_typemap_rehash(m->specializations, 0);
break;
}
default:
Expand Down
Loading