Auto merge of #104164 - cjgillot:u64-cache, r=compiler-errors

Use 64 bits for incremental cache in-file positions

We currently use a 32-bit integer to encode byte positions into the incremental cache.
This is not enough when the query chache file is >4GB.

As the overflow check was a `debug_assert`, it was removed in released compilers, making compilation succeed silently.
At the next compilation, cache decoding would try to read unrelated data because of garbled file position, triggering an ICE.

Fixes #79786
(I'm closing that bug since it the original report and the subsequent questions are probably different instances. A new bug should be opened for new instances of that ICE.)

compiler/rustc_query_impl/src/on_disk_cache.rs

@@ -119,12 +119,11 @@ pub type EncodedDepNodeIndex = Vec<(SerializedDepNodeIndex, AbsoluteBytePos)>;
 struct SourceFileIndex(u32);
 
 #[derive(Copy, Clone, Debug, Hash, Eq, PartialEq, Encodable, Decodable)]
-pub struct AbsoluteBytePos(u32);
+pub struct AbsoluteBytePos(u64);
 
 impl AbsoluteBytePos {
     fn new(pos: usize) -> AbsoluteBytePos {
-        debug_assert!(pos <= u32::MAX as usize);
-        AbsoluteBytePos(pos as u32)
+        AbsoluteBytePos(pos.try_into().expect("Incremental cache file size overflowed u64."))
     }
 
     fn to_usize(self) -> usize {