Allow to specify signal when calling clone. #344
Conversation
| @@ -197,15 +198,16 @@ pub fn sched_setaffinity(pid: isize, cpuset: &CpuSet) -> Result<()> { | |||
| Errno::result(res).map(drop) | |||
| } | |||
|
|
|||
| pub fn clone(mut cb: CloneCb, stack: &mut [u8], flags: CloneFlags) -> Result<pid_t> { | |||
| pub fn clone(mut cb: CloneCb, stack: &mut [u8], flags: CloneFlags, signal: Option<c_int>) -> Result<pid_t> { | |||
There was a problem hiding this comment.
Signal should be constrained to an 8-bit unsigned value (u8/c_char) at compile time given the allowable mask for signal: http://lxr.free-electrons.com/source/include/uapi/linux/sched.h#L7. Other values could clobber the other flags.
There was a problem hiding this comment.
I would prefer to let signal remain of type c_int, the type for signals. In order to address your (valid) concern, we could add a check into our wrapper, that signals an appropriate error.
|
I don't have a strong opinion on this one. Allowing for the signal to be included does seem important and I don't see any obvious problems with this approach. @kamalmarhubi |
|
Interesting. I'd have to read the entire man page to figure out if there are any other similar issues so we don't end up changing the signature repeatedly. |
|
I just opened a PR with an alternative approach at #348. |
|
☔ The latest upstream changes (presumably #353) made this pull request unmergeable. Please resolve the merge conflicts. |
|
⚡ Test exempted - status |
Allow to specify signal when calling clone. This is my suggestion on how to fix #343.
This is my suggestion on how to fix #343.