__waitpid not a cancellation point

__waitpid not a cancellation point

[Roland McGrath]

Linuxthreads overrides waitpid but not __waitpid, so only the public name
is a cancellation point.  It also overrides system, so that is made a
cancellation point separately.  But it does not override popen/pclose,
so those use __waitpid with no cancellation wrapper inside or out.
Is this how it should be?

Re: __waitpid not a cancellation point

[Ulrich Drepper]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Roland McGrath wrote:
> Linuxthreads overrides waitpid but not __waitpid, so only the public name
> is a cancellation point.  It also overrides system, so that is made a
> cancellation point separately.  But it does not override popen/pclose,
> so those use __waitpid with no cancellation wrapper inside or out.
> Is this how it should be?

The standard does not require that functions which are implemented using 
other interfaces which are cancelation points must themselves be 
cancelation points.  I.e., the standard does not require a specific 
implementation.

This is what we should take advantage of.

- -- 
- ---------------.                          ,-.   1325 Chesapeake Terrace
Ulrich Drepper  \    ,-------------------'   \  Sunnyvale, CA 94089 USA
Red Hat          `--' drepper at redhat.com   `------------------------
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQE9WFVe2ijCOnn/RHQRAj7xAJwJ6YdOkHKmtTNnocUJ5NKrs/s4rACfXQap
ZMUR4PLXVIVri++ZaZ6xBUQ=
=MBn2
-----END PGP SIGNATURE-----

Re: __waitpid not a cancellation point

[Roland McGrath]

> The standard does not require that functions which are implemented using 
> other interfaces which are cancelation points must themselves be 
> cancelation points.  I.e., the standard does not require a specific 
> implementation.
> 
> This is what we should take advantage of.

I take it you are also saying that no standard lists popen or pclose as a
cancellation point.

Re: __waitpid not a cancellation point

[Ulrich Drepper]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Roland McGrath wrote:

> I take it you are also saying that no standard lists popen or pclose as a
> cancellation point.

Nope, they are not.  The complete list is:

accept()
aio_suspend()
clock_nanosleep()
close()
connect()
creat()
fcntl() <-- only for F_SETLKW
fsync()
getmsg()
getpmsg()
lockf()
mq_receive()
mq_send()
mq_timedreceive()
mq_timedsend()
msgrcv()
msgsnd()
msync()
nanosleep()
open()
pause()
poll()
pread()
pthread_cond_timedwait()
pthread_cond_wait()
pthread_join()
pthread_testcancel()
putmsg()
putpmsg()
pwrite()
read()
readv()
recv()
recvfrom()
recvmsg()
select()
sem_timedwait()
sem_wait()
send()
sendmsg()
sendto()
sigpause()
sigsuspend()
sigtimedwait()
sigwait()
sigwaitinfo()
sleep()
system()
tcdrain()
usleep()
wait()
waitid()
waitpid()
write()
writev()

- -- 
- ---------------.                          ,-.   1325 Chesapeake Terrace
Ulrich Drepper  \    ,-------------------'   \  Sunnyvale, CA 94089 USA
Red Hat          `--' drepper at redhat.com   `------------------------
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD4DBQE9WHno2ijCOnn/RHQRAr9UAJisur9BULqt1kvo+FKyX3S/ZjNRAJ4k7icP
xHQLh11Nwvsf4ibWhfTZFg==
=heDF
-----END PGP SIGNATURE-----