Put convergent back

Author: jhuber6

libc/src/__support/RPC/rpc.h

@@ -164,7 +164,8 @@ template <bool Invert> struct Process {
   /// lane_mask is a bitmap of the threads in the warp that would hold the
   /// single lock on success, e.g. the result of rpc::get_lane_mask()
   /// The lock is held when the n-th bit of the lock bitfield is set.
-  LIBC_INLINE bool try_lock(uint64_t lane_mask, uint32_t index) {
+  [[clang::convergent]] LIBC_INLINE bool try_lock(uint64_t lane_mask,
+                                                  uint32_t index) {
     // On amdgpu, test and set to the nth lock bit and a sync_lane would suffice
     // On volta, need to handle differences between the threads running and
     // the threads that were detected in the previous call to get_lane_mask()
@@ -204,7 +205,8 @@ template <bool Invert> struct Process {
 
   /// Unlock the lock at index. We need a lane sync to keep this function
   /// convergent, otherwise the compiler will sink the store and deadlock.
-  LIBC_INLINE void unlock(uint64_t lane_mask, uint32_t index) {
+  [[clang::convergent]] LIBC_INLINE void unlock(uint64_t lane_mask,
+                                                uint32_t index) {
     // Do not move any writes past the unlock.
     __atomic_thread_fence(__ATOMIC_RELEASE);
 
@@ -517,7 +519,8 @@ LIBC_INLINE void Port<T>::recv_n(void **dst, uint64_t *size, A &&alloc) {
 /// port. Each port instance uses an associated \p opcode to tell the server
 /// what to do. The Client interface provides the appropriate lane size to the
 /// port using the platform's returned value.
-template <uint16_t opcode> LIBC_INLINE Client::Port Client::open() {
+template <uint16_t opcode>
+[[clang::convergent]] LIBC_INLINE Client::Port Client::open() {
   // Repeatedly perform a naive linear scan for a port that can be opened to
   // send data.
   for (uint32_t index = 0;; ++index) {
@@ -551,7 +554,7 @@ template <uint16_t opcode> LIBC_INLINE Client::Port Client::open() {
 
 /// Attempts to open a port to use as the server. The server can only open a
 /// port if it has a pending receive operation
-LIBC_INLINE cpp::optional<typename Server::Port>
+[[clang::convergent]] LIBC_INLINE cpp::optional<typename Server::Port>
 Server::try_open(uint32_t lane_size, uint32_t start) {
   // Perform a naive linear scan for a port that has a pending request.
   for (uint32_t index = start; index < process.port_count; ++index) {