At the moment, the LLVM IR we generate for this fn is
define internal fastcc void @AstGen.numberLiteral ... {
Entry:
...
%16 = alloca %"fmt.parse_float.decimal.Decimal(f128)", align 8
...
That `Decimal` is huuuge! It stores
pub const max_digits = 11564;
digits: [max_digits]u8,
on the stack.
It comes from `convertSlow` function, which LLVM happily inlined,
despite it being the cold path. Forbid inlining that to not penalize
callers with excessive stack usage.
Backstory: I was looking for needles memcpys in TigerBeetle, and came up
with this copyhound.zig tool for doing just that:
ee67e2ab95/src/copyhound.zig
Got curious, run it on the Zig's own code base, and looked at some of
the worst offenders.
List of worst offenders:
warning: crypto.kyber_d00.Kyber.SecretKey.decaps: 7776 bytes memcpy
warning: crypto.ff.Modulus.powPublic: 8160 bytes memcpy
warning: AstGen.numberLiteral: 11584 bytes memcpy
warning: crypto.tls.Client.init__anon_133566: 13984 bytes memcpy
warning: http.Client.connectUnproxied: 16896 bytes memcpy
warning: crypto.tls.Client.init__anon_133566: 16904 bytes memcpy
warning: objcopy.ElfFileHelper.tryCompressSection: 32768 bytes memcpy
Note from Andrew: I removed `noinline` from this commit since it should
be enough to set it to be cold.
It is assumed that generating a collision requires more than 2^156
ciphertext modifications. This is plenty enough for any practical
purposes, but it hasn't been proven to be >= 2^256.
Be consistent and conservative here; just claim the same security
as the other variants.
Previously, they were only created when we had any TLS segment.
This meant that while the symbol existed, the global itself wouldn't.
The result of this was a crash during symbol names writing as it
would attempt to write the symbol name of a global that didn't exist.
Now we always create them, and instead update its `init` value during
`setupMemory`.
In the future, the entire symbol (and global) will be removed by
the garbage collector.
Implements the `start` section which will execute a given function
at startup of the program. After function execution, the _start
function will be called by the runtime. In the case of shared-memory
we set this section to the function `__wasm_init_memory` which will
initialize all memory on startup.
This also fixes the above mentioned function to ensure we correctly
lower the i32 values.
Lastly, this fixes a typo where we would retrieve a global, instead
of setting its value.
Rather than verifying if importing memory is false, we now rely
on the option that was passed to the CLI (where export is defaulted
to `true` unless only import-memory is given).
* getOwnedFunctionIndex no longer checks if the value is actually a
function.
* The callsites to `intern` that I added want to avoid the `getCoerced`
call, so I added `intern2`.
* Adding to inferred error sets should not happen if the destination
error set is not the inferred error set of the current Sema instance.
* adhoc_inferred_error_set_type can be seen by the backend. Treat it
like anyerror.
when other function's inferred error set is the return type of a
function, it should not try to insert the error set into it.
this implies that this branch fixes a bug in master branch.
When instantiating a generic function call, in the case of a function
call generated by the language, the LazySrcLoc provided for `call_src`
may not point to anything interesting.
* Introduce InternPool.Tag.func_coerced to handle the case of a
function body coerced to a new type. `InternPool.getCoerced` is now
implemented for function bodies in this branch.
* implement resolution of ad-hoc inferred error sets in
`Sema.analyzeCall`.
* fix generic_owner being set wrong for child Sema bodies of param
expressions.
* fix `Sema.resolveInferredErrorSetTy` when passed `anyerror`.
Previously, they shared function index with the owner decl, but that
would clobber the data stored for inferred error sets of runtime calls.
Now there is an adhoc_inferred_error_set_type which models the problem
much more correctly.
Before, it incorrectly passed an InternPool.Index where an extra array
index was expected (to the function which is renamed to `extraErrorSet`
in this commit).
There is one case where function types may be inequal but we still want
to find the same function body instance in InternPool.
In the case of the functions having an inferred error set, the key used
to find an existing function body will necessarily have a unique
inferred error set type, because it refers to the function body
InternPool Index. To make this case work we omit the inferred error set
from the equality and hashing functions.
* move inferred error sets into InternPool.
- they are now represented by pointing directly at the corresponding
function body value.
* inferred error set working memory is now in Sema and expires after
the Sema for the function corresponding to the inferred error set is
finished having its body analyzed.
* error sets use a InternPool.Index.Slice rather than an actual slice
to avoid lifetime issues.
The idea here is to move towards a future where anonymous decls are
represented entirely by an `InternPool.Index`. This was needed to start
implementing `InternPool.getFuncDecl` which requires moving creation and
deletion of Decl objects into InternPool.
* remove `Namespace.anon_decls`
* remove the concept of cleaning up resources from anonymous decls,
relying on InternPool instead.
* move namespace and decl object allocation into InternPool
