Nix/src/libexpr/eval-inline.hh
pennae 1fe66852ff reduce the size of Env by one pointer
since `up` and `values` are both pointer-aligned the type field will
also be pointer-aligned, wasting 48 bits of space on most machines. we
can get away with removing the type field altogether by encoding some
information into the `with` expr that created the env to begin with,
reducing the GC load for the absolutely massive amount of single-entry
envs we create for lambdas. this reduces memory usage of system eval by
quite a bit (reducing heap size of our system eval from 8.4GB to 8.23GB)
and gives similar savings in eval time.

running `nix eval --raw --impure --expr 'with import <nixpkgs/nixos> {}; system'`

before:

  Time (mean ± σ):      5.576 s ±  0.003 s    [User: 5.197 s, System: 0.378 s]
  Range (min … max):    5.572 s …  5.581 s    10 runs

after:

  Time (mean ± σ):      5.408 s ±  0.002 s    [User: 5.019 s, System: 0.388 s]
  Range (min … max):    5.405 s …  5.411 s    10 runs
2023-12-30 18:55:13 +01:00

133 lines
3.3 KiB
C++

#pragma once
///@file
#include "eval.hh"
namespace nix {
/**
* Note: Various places expect the allocated memory to be zeroed.
*/
[[gnu::always_inline]]
inline void * allocBytes(size_t n)
{
void * p;
#if HAVE_BOEHMGC
p = GC_MALLOC(n);
#else
p = calloc(n, 1);
#endif
if (!p) throw std::bad_alloc();
return p;
}
[[gnu::always_inline]]
Value * EvalState::allocValue()
{
#if HAVE_BOEHMGC
/* We use the boehm batch allocator to speed up allocations of Values (of which there are many).
GC_malloc_many returns a linked list of objects of the given size, where the first word
of each object is also the pointer to the next object in the list. This also means that we
have to explicitly clear the first word of every object we take. */
if (!*valueAllocCache) {
*valueAllocCache = GC_malloc_many(sizeof(Value));
if (!*valueAllocCache) throw std::bad_alloc();
}
/* GC_NEXT is a convenience macro for accessing the first word of an object.
Take the first list item, advance the list to the next item, and clear the next pointer. */
void * p = *valueAllocCache;
*valueAllocCache = GC_NEXT(p);
GC_NEXT(p) = nullptr;
#else
void * p = allocBytes(sizeof(Value));
#endif
nrValues++;
return (Value *) p;
}
[[gnu::always_inline]]
Env & EvalState::allocEnv(size_t size)
{
nrEnvs++;
nrValuesInEnvs += size;
Env * env;
#if HAVE_BOEHMGC
if (size == 1) {
/* see allocValue for explanations. */
if (!*env1AllocCache) {
*env1AllocCache = GC_malloc_many(sizeof(Env) + sizeof(Value *));
if (!*env1AllocCache) throw std::bad_alloc();
}
void * p = *env1AllocCache;
*env1AllocCache = GC_NEXT(p);
GC_NEXT(p) = nullptr;
env = (Env *) p;
} else
#endif
env = (Env *) allocBytes(sizeof(Env) + size * sizeof(Value *));
/* We assume that env->values has been cleared by the allocator; maybeThunk() and lookupVar fromWith expect this. */
return *env;
}
[[gnu::always_inline]]
void EvalState::forceValue(Value & v, const PosIdx pos)
{
if (v.isThunk()) {
Env * env = v.thunk.env;
Expr * expr = v.thunk.expr;
try {
v.mkBlackhole();
//checkInterrupt();
expr->eval(*this, *env, v);
} catch (...) {
v.mkThunk(env, expr);
tryFixupBlackHolePos(v, pos);
throw;
}
}
else if (v.isApp())
callFunction(*v.app.left, *v.app.right, v, pos);
}
[[gnu::always_inline]]
inline void EvalState::forceAttrs(Value & v, const PosIdx pos, std::string_view errorCtx)
{
forceAttrs(v, [&]() { return pos; }, errorCtx);
}
template <typename Callable>
[[gnu::always_inline]]
inline void EvalState::forceAttrs(Value & v, Callable getPos, std::string_view errorCtx)
{
PosIdx pos = getPos();
forceValue(v, pos);
if (v.type() != nAttrs) {
error("value is %1% while a set was expected", showType(v)).withTrace(pos, errorCtx).debugThrow<TypeError>();
}
}
[[gnu::always_inline]]
inline void EvalState::forceList(Value & v, const PosIdx pos, std::string_view errorCtx)
{
forceValue(v, pos);
if (!v.isList()) {
error("value is %1% while a list was expected", showType(v)).withTrace(pos, errorCtx).debugThrow<TypeError>();
}
}
}