Web lists-archives.com

Re: problem with mmap and fork()




On Sat, 23 Feb 2019 05:54:18, Doug Henderson  wrote:

> On Fri, 22 Feb 2019 at 17:01, Glyn Gowing <> wrote:
> > I have a program (attached) that works correctly on my mac but does
> > not work with Cygwin on Windows 10. I'm running the latest version of
[snip]

> On further analysis, the call to
> > pthread_mutexattr_setpshared(attr, PTHREAD_PROCESS_SHARED);
> is failing. The error is EINVAL (22) Invalid argument.
> 
> This suggests that pthread mutexes cannot be shared between processes
> by using shared memory in cygwin. I have not attempted to determine if
> this is working as expected, or if this is a bug, or a limitation in
> the Windows environment.

For the record, the output on Linux (FC28):

@@ ./Dsmtest2 (Doug's version of the original smtest2.c)
1551022585 - both  : attr, lock, lock->next: 0x7f4ec2ceb000 0x7f4ec2cec000 0x7f4ec2cec000
1551022585 - Parent: acquiring lock (0x7f4ec2cec000)
1551022585 - Parent: acquired  lock (0x7f4ec2cec000) - Before sleeping 10 seconds.
1551022585 - Child : checking  lock (0x7f4ec2cec000) - Before sleeping 5 seconds.
1551022590 - Child : acquiring lock (0x7f4ec2cec000) - After  sleeping 5 seconds.
1551022595 - Parent: unlocking lock (0x7f4ec2cec000) - After  sleeping 10 seconds.
1551022595 - Parent: unlocked  lock (0x7f4ec2cec000) - Before sleeping 5 seconds to let child catch up.
1551022595 - Child : acquired  lock (0x7f4ec2cec000) - Before sleeping 3 seconds.
1551022598 - Child : unlocking lock (0x7f4ec2cec000) - After  sleeping 3 seconds.
1551022598 - Child : unlocked  lock (0x7f4ec2cec000).
1551022600 -                                         - After  sleeping 5 seconds to let child catch up.
@@ 

... as I understand it, this is between a parent thread and a child thread,
after fork(), -- and only these two! -- using mutexes in shared memory, that
is, shared memory between the two processes (parent and child) ...

Logically, I would say, the same as 2 threads in one process ...

(and yes, in general, all threads are "dead" after fork(), except for the
 thread that is the child of the calling thread).

Henri


--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple