Hi Andi,
On Sun, Dec 27, 2009 at 2:44 AM, Andi Kleen <andi@xxxxxxxxxxxxxx> wrote:
> On Sun, Dec 27, 2009 at 01:40:07AM +0100, Andi Kleen wrote:
>>
>> The ppoll replacement example gets the timeout argument
>> wrong. Fix that.
>
> Sorry, the first version was still wrong. This one
> should be correct.
>
> --- man-pages-3.23/man2/poll.2 2009-09-30 03:36:44.000000000 +0200
> +++ man-pages-3.23-hacked/man2/poll.2 2009-12-27 01:43:31.000000000 +0100
> @@ -27,6 +27,7 @@
> .\" 2006-03-13, mtk, Added ppoll() + various other rewordings
> .\" 2006-07-01, mtk, Added POLLRDHUP + various other wording and
> .\" formatting changes.
> +.\" 2009-12, ak, Fix ppoll example.
> .\"
> .TH POLL 2 2009-09-15 "Linux" "Linux Programmer's Manual"
> .SH NAME
> @@ -193,7 +194,7 @@
> call:
> .nf
>
> - ready = ppoll(&fds, nfds, timeout, &sigmask);
> + ready = ppoll(&fds, nfds, ts, &sigmask);
>
> .fi
> is equivalent to
> @@ -204,7 +205,7 @@
> sigset_t origmask;
>
> sigprocmask(SIG_SETMASK, &sigmask, &origmask);
> - ready = poll(&fds, nfds, timeout);
> + ready = poll(&fds, nfds, ts ? ts.tv_sec : -1);
> sigprocmask(SIG_SETMASK, &origmask, NULL);
> .fi
> .PP
Thanks for spotting the problem, but this version isn't correct
either. poll() use an int with milliseconds, ppoll() uses a struct
with seconds + nanoseconds.
Does the following look okay to you? (It also includes a name change
for the timeout argument of ppoll() to make the distinction clearer in
the text.
Change for man-pages-3.25.
Cheers,
Michael
--- a/man2/poll.2
+++ b/man2/poll.2
@@ -28,7 +28,7 @@
.\" 2006-07-01, mtk, Added POLLRDHUP + various other wording and
.\" formatting changes.
.\"
-.TH POLL 2 2009-09-15 "Linux" "Linux Programmer's Manual"
+.TH POLL 2 2010-06-12 "Linux" "Linux Programmer's Manual"
.SH NAME
poll, ppoll \- wait for some event on a file descriptor
.SH SYNOPSIS
@@ -41,7 +41,7 @@ poll, ppoll \- wait for some event on a file descriptor
.B #include <poll.h>
.sp
.BI "int ppoll(struct pollfd *" fds ", nfds_t " nfds ", "
-.BI " const struct timespec *" timeout ", const sigset_t *" sigmask );
+.BI " const struct timespec *" timeout_ts ", const sigset_t *"
sigmask );
.fi
.SH DESCRIPTION
.BR poll ()
@@ -186,14 +186,13 @@ like
allows an application to safely wait until either a file descriptor
becomes ready or until a signal is caught.
.PP
-Other than the difference in the
-.I timeout
-argument, the following
+Other than the difference in the precision of the
+timeout argument, the following
.BR ppoll ()
call:
.nf
- ready = ppoll(&fds, nfds, timeout, &sigmask);
+ ready = ppoll(&fds, nfds, timeout_ts, &sigmask);
.fi
is equivalent to
@@ -202,7 +201,10 @@ executing the following calls:
.nf
sigset_t origmask;
+ int timeout;
+ timeout = (timeout_ts == NULL) ? -1 :
+ (timeout_ts.tv_sec * 1000 + timeout_ts.tv_nsec / 1000000);
sigprocmask(SIG_SETMASK, &sigmask, &origmask);
ready = poll(&fds, nfds, timeout);
sigprocmask(SIG_SETMASK, &origmask, NULL);
@@ -222,12 +224,10 @@ no signal mask manipulation is performed
.BR ppoll ()
differs from
.BR poll ()
-only in the precision of the
-.I timeout
-argument).
+only in the precision of the timeout argument).
The
-.I timeout
+.I timeout_ts
argument specifies an upper limit on the amount of time that
.BR ppoll ()
will block.
@@ -243,7 +243,7 @@ struct timespec {
.in
If
-.I timeout
+.I timeout_ts
is specified as NULL, then
.BR ppoll ()
can block indefinitely.
@@ -304,13 +304,15 @@ is Linux-specific.
Some implementations define the nonstandard constant
.B INFTIM
with the value \-1 for use as a
-.IR timeout .
+.IR timeout
+for
+.BR poll ()..
This constant is not provided in glibc.
.SS "Linux Notes"
The Linux
.BR ppoll ()
system call modifies its
-.I timeout
+.I timeout_ts
argument.
However, the glibc wrapper function hides this behavior
by using a local variable for the timeout argument that
@@ -318,7 +320,7 @@ is passed to the system call.
Thus, the glibc
.BR ppoll ()
function does not modify its
-.I timeout
+.I timeout_ts
argument.
.SH BUGS
See the discussion of spurious readiness notifications under the
--
To unsubscribe from this list: send the line "unsubscribe linux-man" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
