pub fn add_jit_dispatch_to_linker<T>(
linker: &mut Linker<T>,
cache: Arc<KernelCache>,
) -> Result<()>where
T: JitArenaProvider + TenantContext + 'static,Expand description
Register the tensor-wasm:jit/host::dispatch, alloc, and free imports on
linker.
The cache handle is cloned into each closure so the same backing store
is consulted by every guest instance the linker instantiates — kernel
caching IS designed to be cross-tenant. The bump arena, in contrast,
lives in the per-store payload T (via JitArenaProvider) so two
instances sharing a Linker do NOT see each other’s cursor or live
stack.