fix to pass 3 tests in rt::io::signal::test on android

[yichoi]

To enable test on android bot #9120

rt::io::signal::test::test_io_signal_smoketest
rt::io::signal::test::test_io_signal_two_signal_one_signum
rt::io::signal::test::test_io_signal_unregister

[alexcrichton]

Could you paste the errors that you're seeing as well?

[yichoi]

logs are

root@generic:/data/tmp # LD_LIBRARY_PATH=. ./signal

running 3 tests
test test::test_io_signal_smoketest ... root@generic:/data/tmp #

starts test then exits at once.

[alexcrichton]

Hm, I wonder if it's possible to even catch SIGINT on android? All that test does is send a sigint to itself to make sure that it's caught. Certainly warrants some more investigation.

[ksh8281]

@alexcrichton
yes, i tested on adb with pure c language example program to receive SIGINT,
it runs well.
i send and receive sigusr1 signal itself with rust(without register signal handler),
the program print "user signal1" and die.
and with register signal handler, the program dies too..

[alexcrichton]

Hm, I don't think that libuv is doing anything super out-of-the-ordinary with signal handling. Just for reference, what was the C program you were using to verify that SIGINT could be captured? I'll take a look at libuv and see if it's deviating seriously in some direction.

[ksh8281]

this is my signal test program.
it works well.

#include<stdio.h>
#include<signal.h>
#include<unistd.h>

void sig_handler(int signo)
{
  if (signo == SIGINT)
    printf("received SIGINT\n");
}

int main(void)
{
  if (signal(SIGINT, sig_handler) == SIG_ERR)
  printf("\ncan't catch SIGINT\n");
  // A long long wait so that we can easily issue a signal to this process
  while(1) 
    sleep(1);
  return 0;
}

[ksh8281]

@alexcrichton
i think signal handling runs normal. but socket is problem in android.
libuv send catched signal event with libuv's internal pipe.
the pipe is implement with socket functions.
without permisson(internet), we cant use socket functions in android.

[pnkfelix]

(This may be unrelated, but I'm seeing some problems on test_io_signal_two_signal_one_signum when I attempt to swap in the Boehm/BDW conservative GC. I wonder if the difference in timing is exposing some latent problem here.)

[reem]

Triage: we no longer use libuv, so that might have changed the story here, but other than that I can't say.

[alexcrichton]

I believe these tests have long since been removed