stty: Updating libc crate is blocked: 0.2.178->0.2.179

See comments at #8722 .
This blocks adding CI for cygwin to this repo in the future.
It would block pulling fix for #8474 from upstream.

@asomers

Do you know what exactly changed here? Or would it be possible to attach a debugger and see which call is failing?

rust-lang/libc#4694 updated readdir_r, and there are a few defs that would have changed from the 11->12 baseline change https://github.com/search?q=repo%3Arust-lang%2Flibc+%2F%5Cbfreebsd1%5B12%5D%5Cb%2F+path%3A%2F%5Esrc%2F&type=code / rust-lang/libc#2406. None of that jumps out to me as obvious, though. @asomers may have a better idea.

@cakebaker

Myself does not have FBSD, but the log https://github.com/uutils/coreutils/suites/53178265891/logs?attempt=1 exists.

@cakebaker When will the log expired? Would you re-run the CI? (@tgross35 is upstream of libc crate)

Fyi if you ever need a FreeBSD, OpenBSD, Solaris, etc machine or Linux on some less common arches, https://portal.cfarm.net/machines/list/ is a reasonably reliable way to do that. AUIU anyone working on a reasonably large open source project can request an account.

I just checked and cfarm240 (arm64 FreeBSD15) is up and running. Looks like it's cheri-enabled which is unusual, but that might still be fine to reproduce the failure.

@tgross35

As @tgross35 mentioned, the libc upgrade you made included switching from the FreeBSD 11+ ABI to FreeBSD 12+. That's a big difference, because it incluces 64-bit inodes. It also increases the size of struct statfs. That's a syscall that your df is probably using. I suggest looking there.

Any non-backword compat change about time for musl-i686 target at 0.2.178->0.2.179? I saw many undefined at

coreutils/src/uucore/src/lib/features/safe_traversal.rs

Lines 471 to 535 in 8cf88cd

    
           fn atime(&self) -> i64 { 
        
               #[cfg(target_pointer_width = "32")] 
        
               { 
        
                   self.stat.st_atime.into() 
        
               } 
        
               #[cfg(not(target_pointer_width = "32"))] 
        
               { 
        
                   self.stat.st_atime 
        
               } 
        
           } 
        
           fn atime_nsec(&self) -> i64 { 
        
               #[cfg(target_pointer_width = "32")] 
        
               { 
        
                   self.stat.st_atime_nsec.into() 
        
               } 
        
               #[cfg(not(target_pointer_width = "32"))] 
        
               { 
        
                   self.stat.st_atime_nsec 
        
               } 
        
           } 
        
           fn mtime(&self) -> i64 { 
        
               #[cfg(target_pointer_width = "32")] 
        
               { 
        
                   self.stat.st_mtime.into() 
        
               } 
        
               #[cfg(not(target_pointer_width = "32"))] 
        
               { 
        
                   self.stat.st_mtime 
        
               } 
        
           } 
        
           fn mtime_nsec(&self) -> i64 { 
        
               #[cfg(target_pointer_width = "32")] 
        
               { 
        
                   self.stat.st_mtime_nsec.into() 
        
               } 
        
               #[cfg(not(target_pointer_width = "32"))] 
        
               { 
        
                   self.stat.st_mtime_nsec 
        
               } 
        
           } 
        
           fn ctime(&self) -> i64 { 
        
               #[cfg(target_pointer_width = "32")] 
        
               { 
        
                   self.stat.st_ctime.into() 
        
               } 
        
               #[cfg(not(target_pointer_width = "32"))] 
        
               { 
        
                   self.stat.st_ctime 
        
               } 
        
           } 
        
           fn ctime_nsec(&self) -> i64 { 
        
               #[cfg(target_pointer_width = "32")] 
        
               { 
        
                   self.stat.st_ctime_nsec.into() 
        
               } 
        
               #[cfg(not(target_pointer_width = "32"))] 
        
               { 
        
                   self.stat.st_ctime_nsec 
        
               } 
        
           }

coreutils/src/uucore/src/lib/features/fsext.rs

Lines 449 to 454 in 81ee8d1
// Rust on FreeBSD uses 11.x ABI for filesystem metadata syscalls.
// Call the right version of the symbol for getmntinfo() result to
// match libc StatFS layout.
#[cfg(target_os = "freebsd")]
#[link_name = "getmntinfo@FBSD_1.0"]
fn get_mount_info(mount_buffer_p: *mut *mut StatFs, flags: c_int) -> c_int;

I experimentary changed it and running CI.

If you're using the FreeBSD 12 version of struct statfs from libc, but using getmntinfo@FBSD_1.0, then that would definitely cause a problem. Why does coreutils define its own FFI binding for getmntinfo instead of using the one from libc?

I don't know. It was added by different people. libc was bumped to 0.2.178 anyway by removing it.
But I can't bump to 0.2.179 by i686-musl issue.

RUST_LIBC_UNSTABLE_MUSL_V1_2_3=1
RUST_LIBC_UNSTABLE_GNU_TIME_BITS=64

Is this something able to add to .toml instead of environment val?

https://github.com/uutils/coreutils/blob/main/.cargo/config.toml This file seems unrecommended.

Could you paste the musl errors you’re seeing?

No toml config for the env vars. I don’t think you’ll want to use them outside of testing, since you should expect them to break in the future when we switch them to a cfg.

#8717 (without those envs) https://github.com/uutils/coreutils/actions/runs/20693102671/job/59404255919?pr=8717#step:13:539

Looks like that was unintentional breakage from the time_t changes. rust-lang/libc#4913

Navigation Menu

Repository navigation

stty: Updating libc crate is blocked: 0.2.178->0.2.179 #9875

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

	fn atime(&self) -> i64 {
	#[cfg(target_pointer_width = "32")]
	{
	self.stat.st_atime.into()
	}
	#[cfg(not(target_pointer_width = "32"))]
	{
	self.stat.st_atime
	}
	}

	fn atime_nsec(&self) -> i64 {
	#[cfg(target_pointer_width = "32")]
	{
	self.stat.st_atime_nsec.into()
	}
	#[cfg(not(target_pointer_width = "32"))]
	{
	self.stat.st_atime_nsec
	}
	}

	fn mtime(&self) -> i64 {
	#[cfg(target_pointer_width = "32")]
	{
	self.stat.st_mtime.into()
	}
	#[cfg(not(target_pointer_width = "32"))]
	{
	self.stat.st_mtime
	}
	}

	fn mtime_nsec(&self) -> i64 {
	#[cfg(target_pointer_width = "32")]
	{
	self.stat.st_mtime_nsec.into()
	}
	#[cfg(not(target_pointer_width = "32"))]
	{
	self.stat.st_mtime_nsec
	}
	}

	fn ctime(&self) -> i64 {
	#[cfg(target_pointer_width = "32")]
	{
	self.stat.st_ctime.into()
	}
	#[cfg(not(target_pointer_width = "32"))]
	{
	self.stat.st_ctime
	}
	}

	fn ctime_nsec(&self) -> i64 {
	#[cfg(target_pointer_width = "32")]
	{
	self.stat.st_ctime_nsec.into()
	}
	#[cfg(not(target_pointer_width = "32"))]
	{
	self.stat.st_ctime_nsec
	}
	}

stty: Updating libc crate is blocked: 0.2.178->0.2.179 #9875

Description

Activity

tgross35 commented on Jan 3, 2026

oech3 commented on Jan 3, 2026

tgross35 commented on Jan 3, 2026

asomers commented on Jan 3, 2026

oech3 commented on Jan 4, 2026

oech3 commented on Jan 4, 2026

asomers commented on Jan 4, 2026

oech3 commented on Jan 4, 2026

oech3 commented on Jan 4, 2026

tgross35 commented on Jan 4, 2026

oech3 commented on Jan 4, 2026

tgross35 commented on Jan 5, 2026

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions