diff --git a/src/nouveau/compiler/acorn/lib.rs b/src/nouveau/compiler/acorn/lib.rs
index 5daf46fa674..f9bd19c777c 100644
--- a/src/nouveau/compiler/acorn/lib.rs
+++ b/src/nouveau/compiler/acorn/lib.rs
@@ -13,7 +13,7 @@ const INIT_DATA: [u16; 64] = [
 
 /// An implementation of the ACORN random number generator
 ///
-/// See also http://acorn.wikramaratna.org/concept.html
+/// See also <http://acorn.wikramaratna.org/concept.html>
 ///
 /// This isn't cryptographically secure but it's easy and fast and good enough
 /// for generating test data.  More importantly, we own the implementation so it
diff --git a/src/nouveau/compiler/nak/opt_instr_sched_common.rs b/src/nouveau/compiler/nak/opt_instr_sched_common.rs
index 149e5eb7a7f..b782ffaa948 100644
--- a/src/nouveau/compiler/nak/opt_instr_sched_common.rs
+++ b/src/nouveau/compiler/nak/opt_instr_sched_common.rs
@@ -221,8 +221,8 @@ pub fn side_effect_type(op: &Op) -> SideEffect {
 
 /// Try to guess how many cycles a variable latency instruction will take
 ///
-/// These values are based on the cycle estimates from "Dissecting the NVidia
-/// Turing T4 GPU via Microbenchmarking" https://arxiv.org/pdf/1903.07486
+/// These values are based on the cycle estimates from ["Dissecting the NVidia
+/// Turing T4 GPU via Microbenchmarking"](https://arxiv.org/pdf/1903.07486).
 /// Memory instructions were copied from L1 data cache latencies.
 /// For instructions not mentioned in the paper, I made up numbers.
 /// This could probably be improved.
diff --git a/src/nouveau/compiler/nak/spill_values.rs b/src/nouveau/compiler/nak/spill_values.rs
index 15713da1dd7..3ead022d7e6 100644
--- a/src/nouveau/compiler/nak/spill_values.rs
+++ b/src/nouveau/compiler/nak/spill_values.rs
@@ -1056,7 +1056,7 @@ impl Function {
     ///
     ///  - S, the set of variables which have been spilled
     ///
-    /// These sets are tracked as we walk instructions and [un]spill values to
+    /// These sets are tracked as we walk instructions and \[un\]spill values to
     /// satisfy the given limit.  When spills are required we spill the value
     /// with the nighest next-use IP.  At block boundaries, Braun and Hack
     /// describe a heuristic for determining the starting W and S sets based on
diff --git a/src/nouveau/compiler/nak/union_find.rs b/src/nouveau/compiler/nak/union_find.rs
index 2967f3b5db7..27c8c9de2e9 100644
--- a/src/nouveau/compiler/nak/union_find.rs
+++ b/src/nouveau/compiler/nak/union_find.rs
@@ -23,7 +23,7 @@ enum Node<X: Copy> {
 ///
 /// Robert E. Tarjan and Jan van Leeuwen. 1984. Worst-case Analysis of Set
 ///     Union Algorithms. J. ACM 31, 2 (April 1984), 245–281.
-///     https://doi.org/10.1145/62.2160
+///     <https://doi.org/10.1145/62.2160>
 pub struct UnionFind<X: Copy + Hash + Eq> {
     idx_map: HashMap<X, usize>,
     nodes: Vec<Node<X>>,