Use xxhash instead of gdb_hash for the internal hashmap for .gdb_index

Our ConcurrentMap uses linear probing to find unused hash table entry.
It gives up if 128 consective slots are occupied, and the whole process
dies with the "ConcurrentMap is full" error message. So our hash
function's quality must be high.

For .gdb_index, we used to use gdb_index() to compute keys for the
ConcurrentMap. It turned out that the function's quality is poor,
generating very similar output for short strings.

This commit changes the hash function to xxhash.

Fixes https://issues.chromium.org/issues/40276991#comment5

elf/gdb-index.cc

@@ -135,12 +135,12 @@ struct NameType {
   }
 
   std::string_view name;
-  u32 hash;
+  u64 hash;
   u8 type;
 };
 
 struct MapValue {
-  u32 hash = 0;
+  u32 gdb_hash = 0;
   Atomic<u32> count;
   u32 name_offset = 0;
   u32 type_offset = 0;
@@ -539,7 +539,7 @@ static i64 read_pubnames_cu(Context<E> &ctx, const PubnamesHdr &hdr,
     u8 type = *p++;
     std::string_view name = (char *)p;
     p += name.size() + 1;
-    cu->nametypes.push_back({name, gdb_hash(name), type});
+    cu->nametypes.push_back({name, hash_string(name), type});
   }
 
   return size;
@@ -683,7 +683,8 @@ void write_gdb_index(Context<E> &ctx) {
     for (NameType &nt : cu.nametypes) {
       MapValue *ent;
       bool inserted;
-      std::tie(ent, inserted) = map.insert(nt.name, nt.hash, MapValue{nt.hash});
+      std::tie(ent, inserted) = map.insert(nt.name, nt.hash,
+                                           MapValue{gdb_hash(nt.name)});
       ent->count++;
       cu.entries.push_back(ent);
     }
@@ -750,7 +751,7 @@ void write_gdb_index(Context<E> &ctx) {
   ul32 *ht = (ul32 *)(buf + hdr.symtab_offset);
 
   for (Entry *ent : entries) {
-    u32 hash = ent->value.hash;
+    u32 hash = ent->value.gdb_hash;
     u32 step = ((hash * 17) & mask) | 1;
     u32 j = hash & mask;