fix std lib build for uclibc #95870
fix std lib build for uclibc #95870
Conversation
|
r? @kennytm (rust-highfive has picked a reviewer for you, use r? to override) |
rust-highfive
added
the
S-waiting-on-review
| @@ -54,6 +54,22 @@ mod imp { | |||
| use crate::sys::unix::os::page_size; | |||
| use crate::sys_common::thread_info; | |||
|
|
|||
| #[cfg(any(target_os = "linux", target_os = "android"))] | |||
There was a problem hiding this comment.
Looking at the linux-like targets in libc, it looks like only uclibc is missing the si_addr() accessor at the moment. Would it make sense to leave the existing logic in place, adding the supplemental accessor definition for only target_env = uclibc?
|
|
||
| #[cfg(not(any(target_os = "linux", target_os = "android")))] | ||
| unsafe fn siginfo_si_addr(info: *mut libc::siginfo_t) -> usize { | ||
| (*info).si_addr as usize |
There was a problem hiding this comment.
Going back to the si_addr field accessor, rather than the si_addr() function will break illumos, like in the initial version of #94052). At the very least, non-Linux platforms should use `si_addr().
|
Opened rust-lang/libc#2750 to hopefully bring uclibc into parity with the other UNIX-y platforms WRT |
|
Given that rust-lang/libc#2750 has been merged, can this be abandoned, or transmuted into a bump of the libc version? (That's all that should be required at this point, I think!) |
|
I suggest abandoning this PR in favor of #96656 |
close #95866
It seems that
libcdoesn't implementfn si_addrinsiginfo_tfor uclibc, which causes the build broken. This PR partially revert #95688 to fix the problem.