All of lore.kernel.org
 help / color / mirror / Atom feed
From: Arjan van de Ven <arjan@infradead.org>
To: linux-kernel@vger.kernel.org
Cc: Arjan van de Ven <arjan@infradead.org>,
	mingo@elte.hu, tglx@tglx.de, torvalds@linux-foundation.org
Subject: [PATCH 4/5] select: make select() use schedule_hrtimeout()
Date: Fri, 29 Aug 2008 08:08:09 -0700	[thread overview]
Message-ID: <20080829080809.0e42a323@infradead.org> (raw)
In-Reply-To: <20080829080549.6906b744@infradead.org>


From: Arjan van de Ven <arjan@linux.intel.com>
Date: Thu, 28 Aug 2008 13:45:38 -0700
Subject: [PATCH] select: make select() use schedule_hrtimeout()

now that we have schedule_hrtimeout(), make select() use it.
But only for short delays; really long delays are assumed to not
need the highres level of accuracy but rather want the regular
timer behavior for now.

This is only done for delays shorter than 1 second; currently
the assumption is that longer delays have no desire to get the
higher accuracy.

Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
---
 fs/select.c |   41 +++++++++++++++++++++++++++++------------
 1 files changed, 29 insertions(+), 12 deletions(-)

diff --git a/fs/select.c b/fs/select.c
index 3ce5943..62b9767 100644
--- a/fs/select.c
+++ b/fs/select.c
@@ -24,6 +24,7 @@
 #include <linux/fdtable.h>
 #include <linux/fs.h>
 #include <linux/rcupdate.h>
+#include <linux/hrtimer.h>
 
 #include <asm/uaccess.h>
 
@@ -187,6 +188,7 @@ int do_select(int n, fd_set_bits *fds, s64 *timeout, struct timespec *end_time)
 	struct poll_wqueues table;
 	poll_table *wait;
 	int retval, i;
+	int use_hr = 0;
 
 	rcu_read_lock();
 	retval = max_select_fd(n, fds);
@@ -196,6 +198,10 @@ int do_select(int n, fd_set_bits *fds, s64 *timeout, struct timespec *end_time)
 		return retval;
 	n = retval;
 
+	/* Use hrtimers only for short timeouts for now */
+	if (*timeout < HZ && end_time->tv_sec != 0)
+		use_hr = 1;
+
 	poll_initwait(&table);
 	wait = &table.pt;
 	if (!*timeout)
@@ -266,20 +272,31 @@ int do_select(int n, fd_set_bits *fds, s64 *timeout, struct timespec *end_time)
 			break;
 		}
 
-		if (*timeout < 0) {
-			/* Wait indefinitely */
-			__timeout = MAX_SCHEDULE_TIMEOUT;
-		} else if (unlikely(*timeout >= (s64)MAX_SCHEDULE_TIMEOUT - 1)) {
-			/* Wait for longer than MAX_SCHEDULE_TIMEOUT. Do it in a loop */
-			__timeout = MAX_SCHEDULE_TIMEOUT - 1;
-			*timeout -= __timeout;
+		if (use_hr) {
+			struct timespec now;
+			ktime_t expire;
+			schedule_hrtimeout(end_time, HRTIMER_MODE_ABS);
+			getnstimeofday(&now);
+			expire = ktime_sub(timespec_to_ktime(*end_time), timespec_to_ktime(now));
+			if (ktime_to_ns(expire) <= 0)
+				break;
+
 		} else {
-			__timeout = *timeout;
-			*timeout = 0;
+			if (*timeout < 0) {
+				/* Wait indefinitely */
+				__timeout = MAX_SCHEDULE_TIMEOUT;
+			} else if (unlikely(*timeout >= (s64)MAX_SCHEDULE_TIMEOUT - 1)) {
+				/* Wait for longer than MAX_SCHEDULE_TIMEOUT. Do it in a loop */
+				__timeout = MAX_SCHEDULE_TIMEOUT - 1;
+				*timeout -= __timeout;
+			} else {
+				__timeout = *timeout;
+				*timeout = 0;
+			}
+			__timeout = schedule_timeout(__timeout);
+			if (*timeout >= 0)
+				*timeout += __timeout;
 		}
-		__timeout = schedule_timeout(__timeout);
-		if (*timeout >= 0)
-			*timeout += __timeout;
 	}
 	__set_current_state(TASK_RUNNING);
 
-- 
1.5.5.1


  parent reply	other threads:[~2008-08-29 15:21 UTC|newest]

Thread overview: 22+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2008-08-29 15:05 [patch 0/5] Nano/Microsecond resolution for select() and poll() Arjan van de Ven
2008-08-29 15:06 ` [PATCH 1/5] select: add a timespec version of the timeout to select/poll Arjan van de Ven
2008-08-30  2:10   ` Andrew Morton
2008-08-30  2:54     ` Linus Torvalds
2008-08-29 15:07 ` [PATCH 2/5] select: return accurate remainer in select() and ppoll() Arjan van de Ven
2008-08-29 15:07 ` [PATCH 3/5] select: introduce a schedule_hrtimeout() function Arjan van de Ven
2008-08-29 15:08 ` Arjan van de Ven [this message]
2008-08-29 15:36   ` [PATCH 4/5] select: make select() use schedule_hrtimeout() Arnd Bergmann
2008-08-29 15:59   ` Daniel Walker
2008-08-29 16:20   ` Linus Torvalds
2008-08-29 16:11     ` Alan Cox
2008-08-29 17:26       ` Linus Torvalds
2008-08-29 17:42         ` Arjan van de Ven
2008-08-29 18:18         ` Alan Cox
2008-08-29 18:46           ` Linus Torvalds
2008-08-29 18:33             ` Alan Cox
2008-08-30 15:25               ` Arjan van de Ven
2008-08-29 16:30     ` Arjan van de Ven
2008-08-29 15:08 ` [PATCH 5/5] select: make poll() use schedule_hrtimeout() as well Arjan van de Ven
2008-08-29 15:54 ` [patch 0/5] Nano/Microsecond resolution for select() and poll() Arnd Bergmann
2008-08-29 16:12   ` Arjan van de Ven
2008-08-29 22:15 ` Brian Wellington

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20080829080809.0e42a323@infradead.org \
    --to=arjan@infradead.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mingo@elte.hu \
    --cc=tglx@tglx.de \
    --cc=torvalds@linux-foundation.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.