[Emitting Zero] [Local Tracking] General rule might not handle unary ne emitting zero bits well

Inizia eseguendo il modulo WebAssembly fornito tramite wasm-opt al commit 755a8d0e con -O2 e -all -O3, quindi confronta i risultati dell’ottimizzazione. Traccia la gestione del local tracking e dei confronti unari, usando l’issue #7492 come contesto correlato. Il lavoro è completato quando la condizione viene dedotta come vera e il ramo diventa irraggiungibile anche con -O3.

Given the following code:

(module
  (import "External" "external_function" (func $external_function))
  (func $foo (result i32)
    i32.const 0
    i32.load)
  (func $_start (param $0 i32) (param $1 i64) (param $2 i32) 
    (local $3 i32) (local $4 i32) (local $5 i32) (local $6 i32) (local $7 i32) (local $9 i32)
    call $foo
    local.set $4
    i32.const 1
    local.set $3
    local.get $4
    local.get $3
    i32.shl
    i32.const 1
    i32.shr_s
    local.set $6
    i32.const 0
    i32.const 0
    i32.store
    i32.const -259031342
    local.set $7
    local.get $6
    local.get $7
    i32.ne
    local.set $9
    block  ;; label = @1
      local.get $9
      i32.eqz
      br_if 0 (;@1;)
      unreachable
    end
    call $bar)
  (func $bar call $external_function)
  (memory $0 258 258)
  (export "_start" (func $_start)))

wasm-opt (755a8d0e) should deduce the condition to false, thus fold the branch to unreachable (further deleting call $bar), which works under -O2 but fails under -O3.

Below is optimized by -all -O3:

(func $_start (type $1) (param $0 i32) (param $1 i64) (param $2 i32)
  (local.set $0
   (i32.shr_s
    (i32.shl
     (i32.load
      (i32.const 0)
     )   
     (i32.const 1)
    )   
    (i32.const 1)
   )   
  )
  (i32.store
   (i32.const 0)
   (i32.const 0)
  )
  (if 
   (i32.ne
    (local.get $0) 
    (i32.const -259031342)
   )   
   (then
    (unreachable)
   )   
  )
  (call $external_function)
 )

As you can see, the condition here

   (i32.ne
    (local.get $0) 
    (i32.const -259031342)
   )   

is not deduced to true.

Similar to #7492, Maybe there a missing rule for it to emit zero bits, or is the local tracking insufficient?