x86_64-wrs-vxworks can't build std with libc 0.2.187 or later. No SDK needed
to reproduce:
cargo new vxcheck && cd vxcheck
cargo +nightly check -Zbuild-std=std,panic_abort --target x86_64-wrs-vxworks
pread/pwrite were dropped from the vxworks module in #5129, but std still
imports them for this target (sys/fd/unix.rs:32,406), so two E0432. They're
still listed in libc-test/semver/vxworks.txt.
VxWorks defines neither: nm over the wrsdk-vxworks7-qemu-1.17.0 sysroot and
kernel finds nothing for them, so they'd have to be shims again, as they were up
to 0.2.186. A shim returns -1 without setting errno though, and std's cvt
reads last_os_error(), so maybe std shouldn't call them here at all.
Would you be open to bringing the shims back, or is this better fixed on the std
side?
(The same build also stops on killpg, but that one is only a std problem —
VxWorks has no process groups at all, so there's nothing for libc to declare:
rust-lang/rust#159969)
x86_64-wrs-vxworkscan't build std with libc 0.2.187 or later. No SDK neededto reproduce:
pread/pwritewere dropped from the vxworks module in #5129, but std stillimports them for this target (
sys/fd/unix.rs:32,406), so two E0432. They'restill listed in
libc-test/semver/vxworks.txt.VxWorks defines neither:
nmover the wrsdk-vxworks7-qemu-1.17.0 sysroot andkernel finds nothing for them, so they'd have to be shims again, as they were up
to 0.2.186. A shim returns -1 without setting
errnothough, and std'scvtreads
last_os_error(), so maybe std shouldn't call them here at all.Would you be open to bringing the shims back, or is this better fixed on the std
side?
(The same build also stops on
killpg, but that one is only a std problem —VxWorks has no process groups at all, so there's nothing for libc to declare:
rust-lang/rust#159969)