Enum EmitError 

pub enum EmitError {
    NotYetImplemented(&'static str),
    InvalidEntryName {
        entry: String,
    },
    TooManyRegisters,
    EmissionBudgetExceeded {
        what: &'static str,
    },
}

Errors produced by the PTX emitter.

Variants

NotYetImplemented(&'static str)

The blueprint contains an op the emitter does not yet lower to PTX. Callers should treat this as “keep this function on the CPU path”.

InvalidEntryName

The blueprint’s entry field is not a valid PTX identifier.

Today every legitimate caller constructs entry via format!("func{idx}") so this can never fire — but the Pliron UserFuncName::Testcase path can carry tenant-controlled bytes that would otherwise be interpolated unescaped into the PTX template (closing the kernel scope, emitting a second adversary kernel, etc.). Closes jit S-1.

Fields

entry: String

The rejected identifier (echoed for diagnostic purposes only — callers MUST NOT include the value in untrusted-facing errors).

TooManyRegisters

The blueprint demands more SSA registers than the %fN index space (u32) can address. Previously the allocator saturated at u32::MAX, which silently aliased two distinct SSA values onto the same physical register and corrupted results. We now refuse the blueprint so the caller (rewrite.rs) deopts to the CPU path rather than launch a miscompiled kernel. Closes jit L1.

EmissionBudgetExceeded

The blueprint’s emission budget (per-op lanes, or total ops * lanes across the stream) exceeds the emitter’s DoS guardrails.

The lane-bearing ops (VecAdd/VecMul/VecFma/LoadUnified/ StoreUnified) each drive a 0..lanes emission loop. The production auto-offload path pins lanes to DEFAULT_LANES = 4, but the PUBLIC TensorWasmKernelBlueprint::push + emit/emit_with API accepts arbitrary lanes (up to u32::MAX). An embedder building blueprints from untrusted dimensions could request billions of PTX lines and exhaust CPU/memory before a byte of output is returned. We bound lanes by MAX_LANES and the aggregate (ops * lanes) emission budget by MAX_EMITTED_OPS and refuse up front rather than rely on every caller to pre-bound. Closes jit L2 (unbounded lanes / op-count emission DoS).

Fields

what: &'static str

Which limit was hit (for diagnostics only — this string is static and carries no untrusted-derived value).

Trait Implementations

impl Clone for EmitError

fn clone(&self) -> EmitError

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for EmitError

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Display for EmitError

fn fmt(&self, __formatter: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Eq for EmitError

impl Error for EmitError

fn source(&self) -> Option<&(dyn Error + 'static)>

Returns the lower-level source of this error, if any.

fn description(&self) -> &str

👎Deprecated since 1.42.0:

use the Display impl or to_string()

fn cause(&self) -> Option<&dyn Error>

👎Deprecated since 1.33.0:

replaced by Error::source, which can support downcasting

fn provide<'a>(&'a self, request: &mut Request<'a>)

🔬This is a nightly-only experimental API. (error_generic_member_access)

Provides type-based access to context intended for error reports.

impl PartialEq for EmitError

fn eq(&self, other: &EmitError) -> bool

Tests for self and other values to be equal, and is used by ==.

fn ne(&self, other: &Rhs) -> bool

Tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl StructuralPartialEq for EmitError