crypto: fix constant randomx exceptions in large page allocator

If allocating large pages fails, we don't try again.
This has the obvious drawback of not being able to use large pages
if they fail once.

src/crypto/rx-slow-hash.c

@@ -63,6 +63,7 @@ static rx_state rx_s[2] = {{CTHR_MUTEX_INIT,{0},0,0},{CTHR_MUTEX_INIT,{0},0,0}};
 
 static randomx_dataset *rx_dataset;
 static int rx_dataset_nomem;
+static int rx_dataset_nolp;
 static uint64_t rx_dataset_height;
 static THREADV randomx_vm *rx_vm = NULL;
 
@@ -316,10 +317,11 @@ void rx_slow_hash(const uint64_t mainheight, const uint64_t seedheight, const ch
       }
       CTHR_MUTEX_UNLOCK(rx_dataset_mutex);
     }
-    if (!(disabled_flags() & RANDOMX_FLAG_LARGE_PAGES)) {
+    if (!(disabled_flags() & RANDOMX_FLAG_LARGE_PAGES) && !rx_dataset_nolp) {
       rx_vm = randomx_create_vm(flags | RANDOMX_FLAG_LARGE_PAGES, rx_sp->rs_cache, rx_dataset);
       if(rx_vm == NULL) { //large pages failed
         mdebug(RX_LOGCAT, "Couldn't use largePages for RandomX VM");
+        rx_dataset_nolp = 1;
       }
     }
     if (rx_vm == NULL)
@@ -370,5 +372,6 @@ void rx_stop_mining(void) {
     randomx_release_dataset(rd);
   }
   rx_dataset_nomem = 0;
+  rx_dataset_nolp = 0;
   CTHR_MUTEX_UNLOCK(rx_dataset_mutex);
 }