diff --git a/library/core/src/num/f16.rs b/library/core/src/num/f16.rs
index 41bd34a2702..bb0cc1c60ba 100644
--- a/library/core/src/num/f16.rs
+++ b/library/core/src/num/f16.rs
@@ -261,7 +261,7 @@ impl f16 {
     ///
     /// ```
     /// #![feature(f16)]
-    /// # #[cfg(target_arch = "aarch64")] { // FIXME(f16_F128): rust-lang/rust#123885
+    /// # #[cfg(all(target_arch = "x86_64", target_os = "linux"))] {
     ///
     /// let nan = f16::NAN;
     /// let f = 7.0_f16;
@@ -293,7 +293,7 @@ impl f16 {
     ///
     /// ```
     /// #![feature(f16)]
-    /// # #[cfg(target_arch = "aarch64")] { // FIXME(f16_F128): rust-lang/rust#123885
+    /// # #[cfg(all(target_arch = "x86_64", target_os = "linux"))] {
     ///
     /// let f = 7.0f16;
     /// let inf = f16::INFINITY;
@@ -319,7 +319,7 @@ impl f16 {
     ///
     /// ```
     /// #![feature(f16)]
-    /// # #[cfg(target_arch = "aarch64")] { // FIXME(f16_F128): rust-lang/rust#123885
+    /// # #[cfg(all(target_arch = "x86_64", target_os = "linux"))] {
     ///
     /// let f = 7.0f16;
     /// let inf: f16 = f16::INFINITY;
@@ -347,7 +347,7 @@ impl f16 {
     ///
     /// ```
     /// #![feature(f16)]
-    /// # #[cfg(target_arch = "aarch64")] { // FIXME(f16_F128): rust-lang/rust#123885
+    /// # #[cfg(all(target_arch = "x86_64", target_os = "linux"))] {
     ///
     /// let min = f16::MIN_POSITIVE; // 6.1035e-5
     /// let max = f16::MAX;
@@ -377,7 +377,7 @@ impl f16 {
     ///
     /// ```
     /// #![feature(f16)]
-    /// # #[cfg(target_arch = "aarch64")] { // FIXME(f16_F128): rust-lang/rust#123885
+    /// # #[cfg(all(target_arch = "x86_64", target_os = "linux"))] {
     ///
     /// let min = f16::MIN_POSITIVE; // 6.1035e-5
     /// let max = f16::MAX;
@@ -409,7 +409,7 @@ impl f16 {
     ///
     /// ```
     /// #![feature(f16)]
-    /// # #[cfg(target_arch = "aarch64")] { // FIXME(f16_F128): rust-lang/rust#123885
+    /// # #[cfg(all(target_arch = "x86_64", target_os = "linux"))] {
     ///
     /// use std::num::FpCategory;
     ///
@@ -867,7 +867,7 @@ impl f16 {
     ///
     /// ```
     /// #![feature(f16)]
-    /// # #[cfg(target_arch = "aarch64")] { // FIXME(f16_F128): rust-lang/rust#123885
+    /// # #[cfg(all(target_arch = "x86_64", target_os = "linux"))] {
     ///
     /// let value = 4.6_f16;
     /// let rounded = unsafe { value.to_int_unchecked::<u16>() };
@@ -910,7 +910,7 @@ impl f16 {
     ///
     /// ```
     /// #![feature(f16)]
-    /// # #[cfg(target_arch = "aarch64")] { // FIXME(f16_F128): rust-lang/rust#123885
+    /// # #[cfg(all(target_arch = "x86_64", target_os = "linux"))] {
     ///
     /// # // FIXME(f16_f128): enable this once const casting works
     /// # // assert_ne!((1f16).to_bits(), 1f16 as u128); // to_bits() is not casting!
@@ -958,7 +958,7 @@ impl f16 {
     ///
     /// ```
     /// #![feature(f16)]
-    /// # #[cfg(target_arch = "aarch64")] { // FIXME(f16_F128): rust-lang/rust#123885
+    /// # #[cfg(all(target_arch = "x86_64", target_os = "linux"))] {
     ///
     /// let v = f16::from_bits(0x4a40);
     /// assert_eq!(v, 12.5);
@@ -1071,7 +1071,7 @@ impl f16 {
     ///
     /// ```
     /// #![feature(f16)]
-    /// # #[cfg(target_arch = "aarch64")] { // FIXME(f16_F128): rust-lang/rust#123885
+    /// # #[cfg(all(target_arch = "x86_64", target_os = "linux"))] {
     ///
     /// let value = f16::from_be_bytes([0x4a, 0x40]);
     /// assert_eq!(value, 12.5);
@@ -1094,7 +1094,7 @@ impl f16 {
     ///
     /// ```
     /// #![feature(f16)]
-    /// # #[cfg(target_arch = "aarch64")] { // FIXME(f16_F128): rust-lang/rust#123885
+    /// # #[cfg(all(target_arch = "x86_64", target_os = "linux"))] {
     ///
     /// let value = f16::from_le_bytes([0x40, 0x4a]);
     /// assert_eq!(value, 12.5);
@@ -1124,7 +1124,7 @@ impl f16 {
     ///
     /// ```
     /// #![feature(f16)]
-    /// # #[cfg(target_arch = "aarch64")] { // FIXME(f16_F128): rust-lang/rust#123885
+    /// # #[cfg(all(target_arch = "x86_64", target_os = "linux"))] {
     ///
     /// let value = f16::from_ne_bytes(if cfg!(target_endian = "big") {
     ///     [0x4a, 0x40]
@@ -1257,7 +1257,7 @@ impl f16 {
     ///
     /// ```
     /// #![feature(f16)]
-    /// # #[cfg(target_arch = "aarch64")] { // FIXME(f16_F128): rust-lang/rust#123885
+    /// # #[cfg(all(target_arch = "x86_64", target_os = "linux"))] {
     ///
     /// assert!((-3.0f16).clamp(-2.0, 1.0) == -2.0);
     /// assert!((0.0f16).clamp(-2.0, 1.0) == 0.0);
diff --git a/library/std/build.rs b/library/std/build.rs
index 35a5977b6eb..fecdf2c3e1f 100644
--- a/library/std/build.rs
+++ b/library/std/build.rs
@@ -103,9 +103,9 @@ fn main() {
         ("arm64ec", _) => false,
         // MinGW ABI bugs <https://gcc.gnu.org/bugzilla/show_bug.cgi?id=115054>
         ("x86_64", "windows") => false,
-        // x86 has ABI bugs that show up with optimizations. This should be partially fixed with
-        // the compiler-builtins update. <https://github.com/rust-lang/rust/issues/123885>
-        ("x86" | "x86_64", _) => false,
+        // Apple has a special ABI for `f16` that we do not yet support
+        // FIXME(builtins): fixed by <https://github.com/rust-lang/compiler-builtins/pull/675>
+        ("x86" | "x86_64", _) if target_vendor == "apple" => false,
         // Missing `__gnu_h2f_ieee` and `__gnu_f2h_ieee`
         ("powerpc" | "powerpc64", _) => false,
         // Missing `__gnu_h2f_ieee` and `__gnu_f2h_ieee`
@@ -140,17 +140,17 @@ fn main() {
         _ => false,
     };
 
-    // These are currently empty, but will fill up as some platforms move from completely
-    // unreliable to reliable basics but unreliable math.
+    // Configure platforms that have reliable basics but may have unreliable math.
 
     // LLVM is currenlty adding missing routines, <https://github.com/llvm/llvm-project/issues/93566>
     let has_reliable_f16_math = has_reliable_f16
         && match (target_arch.as_str(), target_os.as_str()) {
             // FIXME: Disabled on Miri as the intrinsics are not implemented yet.
             _ if is_miri => false,
-            // Currently nothing special. Hooray!
-            // This will change as platforms gain better better support for standard ops but math
-            // lags behind.
+            // x86 has a crash for `powi`: <https://github.com/llvm/llvm-project/issues/105747>
+            ("x86" | "x86_64", _) => false,
+            // Assume that working `f16` means working `f16` math for most platforms, since
+            // operations just go through `f32`.
             _ => true,
         };