Re: proc_waiter: error on read of child wait pipe 0x0, Win32 error 6

[Mark Geisert <mark@maxrnd.com>]

Takashi Yano wrote:
> On Wed, 12 Jan 2022 07:41:41 +0100
> Marco Atzeri wrote:
>> On 12.01.2022 07:27, Jay K wrote:
>>> Ok, here is a small demonstration of the problem.
>>>
>>> #include <stdlib.h>
>>> #include <stdio.h>
>>> #include <windows.h>
>>>
>>> unsigned __stdcall thread(void* p)
>>> {
>>>     unsigned i;
>>>     for (i = 0; i < 100; ++i)
>>>      system("./a.exe");
>>>     return 0;
>>> }
>>>
>>> int main()
>>> {
>>> unsigned i;
>>> HANDLE threads[100] = {0};
>>> FILE* f = fopen("a.c", "w");
>>> fprintf(f, "int main() { return 0; }\n");
>>> fclose(f);
>>
>>
>> so you are mixing Cygwin and Windows calls ?
>> That is looking for trouble.
>>
>> Or it is a tentative to produce a test case ?
> 
> I found that the same happens even with pthread rather than
> win32 thread functions.
> 
> #include <stdlib.h>
> #include <pthread.h>
> 
> void *thread(void *p)
> {
> 	system("true");
> 	return NULL;
> }
> 
> int main()
> {
> 	int i;
> 	pthread_t threads[2];
> 
> 	for (i = 0; i < 2; i++)
> 		pthread_create(&threads[i], NULL, thread, NULL);
> 
> 	for (i = 0; i < 2; i++)
> 		pthread_join(threads[i], NULL);
> 
> 	return 0;
> }
> 
> Executing above code results in hang with message:
>        0 [waitproc] a 786 proc_waiter: error on read of child wait pipe 0x0, Win32 error 6
> 

POSIX does not require system() to be thread-safe.  On Cygwin, it isn't.  When I 
ran into this a while back, I implemented an application wrapper around system() 
to serialize calls.  It's tricky because you want to serialize just the mechanism 
of system(), not the programs that the multiple system()s call.

..mark