mio::Poll::poll crashed for ios target

[Juzenhon]

Thread 5:
0 libsystem_kernel.dylib 0x196b828e4 0x196b5d000 + _kevent
1 DidEPTool 0x1050e1da8 0x104204000 + __ZN3mio4poll4Poll5poll117h2c326700e89fa028E
2 DidEPTool 0x1050e1a30 0x104204000 + __ZN3mio4poll4Poll4poll17hbd448e09b4d4e815E
3 DidEPTool 0x104fc7d58 0x104204000 + _ZN89$LT$client_tcp..connection..MobileTcpStream$u20$as$u20$client_core..Stream$GT$4wait17hf371f0e0c0d72b9dE
4 DidEPTool 0x104fbb278 0x104204000 + _ZN66$LT$client_core..auto_connect..Reactor$LT$T$C$$u20$N$GT$$GT$4poll17h0c915e7450bbb806E
5 DidEPTool 0x104fbd48c 0x104204000 + __ZN3std10sys_common9backtrace28__rust_begin_short_backtrace17hdc17c99d38017f8cE
6 DidEPTool 0x104fc2700 0x104204000 + __ZN3std9panicking3try7do_call17h3b3f1eeedff3ac7cE.llvm.15359701990428932552
7 DidEPTool 0x1050f4460 0x104204000 + __rust_maybe_catch_panic lib.rs:0
8 DidEPTool 0x104fbf1ec 0x104204000 + _ZN50$LT$F$u20$as$u20$alloc..boxed..FnBox$LT$A$GT$$GT$8call_box17h4417b73c1c0f1428E
9 DidEPTool 0x1050f3f10 0x104204000 + _ZN3std3sys4unix6thread6Thread3new12thread_start17h97ba33c6642a2458E thread.rs:14 ZN167$LT$alloc..boxed..Box$LT$$LP$dyn$u20$alloc..boxed..FnBox$LT$A$C$$u20$Output$u3d$R$GT$$u20$$u2b$$u20$$u27$a$RP$$GT$$u20$as$u20$core..ops..function..FnOnce$LT$A$GT$$GT$9call_once17ha0456d496a0a1902E
10 libsystem_pthread.dylib 0x196bf6908 0x196bf5000 + __pthread_body
11 ??? 0x7b180196bf6864 0x0 + 0
12 ??? 0x1d970196bfedcc 0x0 + 0

[Thomasdezeeuw]

Can you provide some more information then this? It's rather hard to debug this with this little information. For example what version of Mio you're using, preferably some code if possible.

Right now it seems to me that it happens inside kevent in poll, that points to:

mio/src/sys/unix/kqueue.rs

Line 90 in 1185510

let cnt = cvt(libc::kevent(

But that could be incorrect.

[Juzenhon]

Mio’ version 0.6.16

Some code ：

fn wait(&mut self, duration: Option<std::time::Duration>) -> Result<bool,std::io::Error> {
        let tmp_dur = duration.clone();
        let begin = std::time::Instant::now();
        self.io.poll(&mut self.events, duration).map(move |count| {
            // count == 0  not always means timeout,for example clear readiness
            if duration.is_some() && begin.elapsed() < tmp_dur.unwrap() {
                false
            } else {
                count == 0
            }
        })
    }

[carllerche]

We would need more info, ideally a repro. There is no info about the kind of crash or any other additional info.