Build failure on loongarch64: no field `st_mtime` on type `libc::stat`

[kpcyrd]

Current behavior 😯

When attempting to update cargo-audit from 0.22.0 to 0.22.1 in Alpine I noticed a build failure in the gix-index crate.

This may be a bug in the libc crate (I'm not sure).

Possibly related to #1817.

error[E0609]: no field `st_mtime` on type `libc::stat`
  --> /builds/kpcyrd/aports/community/cargo-audit/tmp/cargo/registry/src/index.crates.io-1949cf8c6b5b557f/gix-index-0.46.0/src/fs.rs:58:34
   |
58 |             let seconds = self.0.st_mtime;
   |                                  ^^^^^^^^ unknown field
   |
help: a field with a similar name exists
   |
58 -             let seconds = self.0.st_mtime;
58 +             let seconds = self.0.st_mtim;
   |
error[E0609]: no field `st_mtime_nsec` on type `libc::stat`
  --> /builds/kpcyrd/aports/community/cargo-audit/tmp/cargo/registry/src/index.crates.io-1949cf8c6b5b557f/gix-index-0.46.0/src/fs.rs:63:38
   |
63 |             let nanoseconds = self.0.st_mtime_nsec;
   |                                      ^^^^^^^^^^^^^ unknown field
   |
   = note: available fields are: `st_dev`, `st_ino`, `st_mode`, `st_nlink`, `st_uid` ... and 8 others
error[E0609]: no field `st_ctime` on type `libc::stat`
  --> /builds/kpcyrd/aports/community/cargo-audit/tmp/cargo/registry/src/index.crates.io-1949cf8c6b5b557f/gix-index-0.46.0/src/fs.rs:85:34
   |
85 |             let seconds = self.0.st_ctime;
   |                                  ^^^^^^^^ unknown field
   |
help: a field with a similar name exists
   |
85 -             let seconds = self.0.st_ctime;
85 +             let seconds = self.0.st_ctim;
   |
error[E0609]: no field `st_ctime_nsec` on type `libc::stat`
  --> /builds/kpcyrd/aports/community/cargo-audit/tmp/cargo/registry/src/index.crates.io-1949cf8c6b5b557f/gix-index-0.46.0/src/fs.rs:90:38
   |
90 |             let nanoseconds = self.0.st_ctime_nsec;
   |                                      ^^^^^^^^^^^^^ unknown field
   |
   = note: available fields are: `st_dev`, `st_ino`, `st_mode`, `st_nlink`, `st_uid` ... and 8 others

Expected behavior 🤔

cargo-audit builds

Git behavior

Git compiles on loongarch64

Steps to reproduce 🕹

cargo build on loongarch64, I have shell access to a machine like this (kindly provided by cfarm.net/Loongson Technology) and can test patches.

[Byron]

Thanks a lot for reporting, and thanks for offering testing patches.

I actually wouldn't know how to detect that particular platform, let's see if I can conjure up a patch.

[Byron]

Does this work? And if not, do you think you can trivially make it work?

Thanks for checking.

diff --git a/gix-index/src/fs.rs b/gix-index/src/fs.rs
index 2be45bb8d7..a9672eb5e2 100644
--- a/gix-index/src/fs.rs
+++ b/gix-index/src/fs.rs
@@ -54,14 +54,30 @@ impl Metadata {
     pub fn modified(&self) -> Option<SystemTime> {
         #[cfg(not(windows))]
         {
-            #[cfg(not(any(target_os = "aix", target_os = "hurd")))]
+            #[cfg(not(any(
+                target_os = "aix",
+                target_os = "hurd",
+                all(target_arch = "loongarch64", target_env = "musl"),
+            )))]
             let seconds = self.0.st_mtime;
-            #[cfg(any(target_os = "aix", target_os = "hurd"))]
+            #[cfg(any(
+                target_os = "aix",
+                target_os = "hurd",
+                all(target_arch = "loongarch64", target_env = "musl"),
+            ))]
             let seconds = self.0.st_mtim.tv_sec;
 
-            #[cfg(not(any(target_os = "aix", target_os = "hurd")))]
+            #[cfg(not(any(
+                target_os = "aix",
+                target_os = "hurd",
+                all(target_arch = "loongarch64", target_env = "musl"),
+            )))]
             let nanoseconds = self.0.st_mtime_nsec;
-            #[cfg(any(target_os = "aix", target_os = "hurd"))]
+            #[cfg(any(
+                target_os = "aix",
+                target_os = "hurd",
+                all(target_arch = "loongarch64", target_env = "musl"),
+            ))]
             let nanoseconds = self.0.st_mtim.tv_nsec;
 
             // All operating systems treat the seconds as offset from unix epoch, hence it must
@@ -81,14 +97,30 @@ impl Metadata {
     pub fn created(&self) -> Option<SystemTime> {
         #[cfg(not(windows))]
         {
-            #[cfg(not(any(target_os = "aix", target_os = "hurd")))]
+            #[cfg(not(any(
+                target_os = "aix",
+                target_os = "hurd",
+                all(target_arch = "loongarch64", target_env = "musl"),
+            )))]
             let seconds = self.0.st_ctime;
-            #[cfg(any(target_os = "aix", target_os = "hurd"))]
+            #[cfg(any(
+                target_os = "aix",
+                target_os = "hurd",
+                all(target_arch = "loongarch64", target_env = "musl"),
+            ))]
             let seconds = self.0.st_ctim.tv_sec;
 
-            #[cfg(not(any(target_os = "aix", target_os = "hurd")))]
+            #[cfg(not(any(
+                target_os = "aix",
+                target_os = "hurd",
+                all(target_arch = "loongarch64", target_env = "musl"),
+            )))]
             let nanoseconds = self.0.st_ctime_nsec;
-            #[cfg(any(target_os = "aix", target_os = "hurd"))]
+            #[cfg(any(
+                target_os = "aix",
+                target_os = "hurd",
+                all(target_arch = "loongarch64", target_env = "musl"),
+            ))]
             let nanoseconds = self.0.st_ctim.tv_nsec;
 
             // All operating systems treat the seconds as offset from unix epoch, hence it must

[omnivagant]

rust-lang/libc#4939

[kpcyrd]

It indeed seems to be a musl specific error, the crate builds correctly on a glibc based loongarch64 system.

The nushell project also noticed this problem:

nushell/nightly@2c37ebe