Auto merge of #13794 - jonas-schievink:reset-step-limit-after-bump, r=jonas-schievink

fix: fix "parser seems stuck" panic when parsing colossal files

The parser step count is incremented every time the parser inspects a token. It's purpose is to ensure the parser doesn't get stuck in infinite loops. But since `self.pos` grows monotonically when parsing source code, it gives a better idea for whether the parser is stuck or not: if `self.pos` is changed, we know that the parser cannot be stuck, so it is safe to reset the step count to 0. This makes the limit check scale with the size of the file, and so should fix https://github.com/rust-lang/rust-analyzer/issues/13788.

1 files changed, 1 insertions, 0 deletions

diff --git a/crates/parser/src/parser.rs b/crates/parser/src/parser.rs
index 33d645b048b..b621ebc37c3 100644
--- a/crates/parser/src/parser.rs
+++ b/crates/parser/src/parser.rs
@@ -237,6 +237,7 @@ impl<'t> Parser<'t> {
 
     fn do_bump(&mut self, kind: SyntaxKind, n_raw_tokens: u8) {
         self.pos += n_raw_tokens as usize;
+        self.steps.set(0);
         self.push_event(Event::Token { kind, n_raw_tokens });
     }