Thread::sleep() implementation bug on Solaris

Hi,

 

using usleep in Thread::sleep() implementation on Unix(Solaris) is problematic in multithreaded environment

because  it's implemented by setting an alarm signal and pausing until it occurs.

Since SIGALRM shouldn't be blocked or ignored during a call, if context switch occurs when it's delivered, the programm simply quits (default action on SIGALRM).

The proper way is to use nanosleep function. It has no effect on the action or blockage of any signal.

 

Here is an example of nanosleep usage.

I've encapsulated call to nanosleep in signal blocking/unblocking routines in order to prevent signals delivery. 

 

sigset_t newset, oldset;

sigfillset(&newset);

pthread_sigmask(SIG_BLOCK, &newset, &oldset);//block signals

struct timespec sleeptime;

sleeptime.tv_sec = ms / 1000; //argument is supplied in milliseconds

sleeptime.tv_nsec = 0;

if(nanosleep(&sleeptime, NULL)) perror("nanosleep");

pthread_sigmask(SIG_SETMASK, &oldset, NULL);//restore signals

--

Toly Kournik

 

--------------------------------------------------------------------------------------------------------------------------------------------------------------
This email and any files transmitted with it are confidential and intended solely for the use of the individual or entity to whom
they are addressed. If you have received this email in error please notify the originator of the message.
Scanning of this message is performed by SurfControl E-mail Filter software in conjunction with  virus detection software.