Add FcntlArg::F_FULLFSYNC #407
Conversation
| @@ -69,6 +71,8 @@ pub fn fcntl(fd: RawFd, arg: FcntlArg) -> Result<c_int> { | |||
| F_ADD_SEALS(flag) => libc::fcntl(fd, ffi::F_ADD_SEALS, flag.bits()), | |||
| #[cfg(target_os = "linux")] | |||
| F_GET_SEALS => libc::fcntl(fd, ffi::F_GET_SEALS), | |||
| #[cfg(target_os = "darwin")] | |||
| F_FULLFSYNC => libc::fcntl(fd, ffi::F_FULLFSYNC), | |||
There was a problem hiding this comment.
Darwin is not a valid target_os. You have to use either macos, ios or both, depending on where that flag exists. This is also, why this line does not cause our apple build to fail. Otherwise it would complain about an unkown identifier ffi::F_FULLFSYNC (see the ffi module definition further up in the file).
There was a problem hiding this comment.
Sorry, I was confused because of https://github.com/untitaker/nix/blob/69c5d686ae5c64ea758ceaab482e5a3789397ee2/nix-test/build.rs#L10
untitaker
force-pushed
the
fullfsync-fcntl
branch
2 times, most recently
from
e7fffa0 to
aa4824f
Compare
|
I don't have Mac OS X, so I'll test on Travis. I don't know if iOS supports that flag, but this thread seems to imply so. |
untitaker
force-pushed
the
fullfsync-fcntl
branch
from
aa4824f to
ae6635f
Compare
|
Okay, the tests seem to pass now! |
|
Thank you. @homu r+ |
|
📌 Commit ae6635f has been approved by |
|
⚡ Test exempted - status |
https://developer.apple.com/library/ios/documentation/System/Conceptual/ManPages_iPhoneOS/man2/fsync.2.html