[RFC] [PATCH procfs] Add process age

[RFC] [PATCH procfs] Add process age

[Davidlohr Bueso]

Hi,

This patch exports a process's age, in seconds, adding a /proc/#/age file.

Thanks,
Davidlohr

Signed-off-by: Davidlohr Bueso <dave@gnu.org>
---
 fs/proc/base.c |   19 ++++++++++++++++++-
 1 files changed, 18 insertions(+), 1 deletions(-)

diff --git a/fs/proc/base.c b/fs/proc/base.c
index acb7ef8..a5a4832 100644
--- a/fs/proc/base.c
+++ b/fs/proc/base.c
@@ -38,13 +38,16 @@
  *  ChangeLog:
  *  10-Mar-2005
  *  10LE Instituto Nokia de Tecnologia - INdT:
- *  A better way to walks through the page table as suggested by Hugh Dickins.
+ *  A better way to walk through the page table as suggested by Hugh Dickins.
  *
  *  Simo Piiroinen <simo.piiroinen@nokia.com>:
  *  Smaps information related to shared, private, clean and dirty pages.
  *
  *  Paul Mundt <paul.mundt@nokia.com>:
  *  Overall revision about smaps.
+ *
+ *  2010 - Davidlohr Bueso <dave@gnu.org>:
+ *  Added process age (/proc/#/age), in seconds.
  */
 
 #include <asm/uaccess.h>
@@ -2558,6 +2561,18 @@ static int proc_pid_personality(struct seq_file *m, struct pid_namespace *ns,
 	return 0;
 }
 
+static int proc_pid_age(struct seq_file *m, struct pid_namespace *ns,
+			struct pid *pid, struct task_struct *task)
+{
+	struct timespec uptime;
+
+	do_posix_clock_monotonic_gettime(&uptime);
+	monotonic_to_bootbased(&uptime);
+	seq_printf(m, "%lu\n", uptime.tv_sec - task->real_start_time.tv_sec);
+
+	return 0;
+}
+
 /*
  * Thread groups
  */
@@ -2575,6 +2590,7 @@ static const struct pid_entry tgid_base_stuff[] = {
 	INF("auxv",       S_IRUSR, proc_pid_auxv),
 	ONE("status",     S_IRUGO, proc_pid_status),
 	ONE("personality", S_IRUSR, proc_pid_personality),
+	ONE("age",        S_IRUSR, proc_pid_age),
 	INF("limits",	  S_IRUSR, proc_pid_limits),
 #ifdef CONFIG_SCHED_DEBUG
 	REG("sched",      S_IRUGO|S_IWUSR, proc_pid_sched_operations),
@@ -2910,6 +2926,7 @@ static const struct pid_entry tid_base_stuff[] = {
 	INF("auxv",      S_IRUSR, proc_pid_auxv),
 	ONE("status",    S_IRUGO, proc_pid_status),
 	ONE("personality", S_IRUSR, proc_pid_personality),
+	ONE("age",        S_IRUSR, proc_pid_age),
 	INF("limits",	 S_IRUSR, proc_pid_limits),
 #ifdef CONFIG_SCHED_DEBUG
 	REG("sched",     S_IRUGO|S_IWUSR, proc_pid_sched_operations),
-- 
1.7.0.4

Re: [RFC] [PATCH procfs] Add process age

[Alexey Dobriyan]

On Fri, Jul 23, 2010 at 01:47:51PM -0400, Davidlohr Bueso wrote:
> This patch exports a process's age, in seconds, adding a /proc/#/age file.

What for?

Re: [RFC] [PATCH procfs] Add process age

[Davidlohr Bueso]

On Fri, 2010-07-23 at 21:35 +0300, Alexey Dobriyan wrote:
> On Fri, Jul 23, 2010 at 01:47:51PM -0400, Davidlohr Bueso wrote:
> > This patch exports a process's age, in seconds, adding a /proc/#/age file.
> 
> What for?

I have not found any straight way to get the run time of a processes,
without having to interpret the start_time entry in /proc/#/stat and
manually subtract the uptime from it.

This small file simply aids the users to obtain the information easier
and quicker. Another alternative would be to add an entry
in /proc/#/status.

Davidlohr

Re: [RFC] [PATCH procfs] Add process age

[Sergey V.]

On Friday 23 of July 2010 22:47:10 Davidlohr Bueso wrote:
> On Fri, 2010-07-23 at 21:35 +0300, Alexey Dobriyan wrote:
> > On Fri, Jul 23, 2010 at 01:47:51PM -0400, Davidlohr Bueso wrote:
> > > This patch exports a process's age, in seconds, adding a /proc/#/age 
file.
> > 
> > What for?
> 
> I have not found any straight way to get the run time of a processes,
> without having to interpret the start_time entry in /proc/#/stat and
> manually subtract the uptime from it.

Agree. I thing this feature can be useful in some cases.

I have not found better than run something like this:

echo `cat /proc/uptime | awk '{print $1}'` - \
     `cat /proc/$PID/stat | awk '{print $22}'`/100 | bc

> This small file simply aids the users to obtain the information easier
> and quicker. Another alternative would be to add an entry
> in /proc/#/status.
> 
> Davidlohr

Re: [RFC] [PATCH procfs] Add process age

[Andi Kleen]

Davidlohr Bueso <dave.bueso@gmail.com> writes:

> On Fri, 2010-07-23 at 21:35 +0300, Alexey Dobriyan wrote:
>> On Fri, Jul 23, 2010 at 01:47:51PM -0400, Davidlohr Bueso wrote:
>> > This patch exports a process's age, in seconds, adding a /proc/#/age file.
>> 
>> What for?
>
> I have not found any straight way to get the run time of a processes,
> without having to interpret the start_time entry in /proc/#/stat and
> manually subtract the uptime from it.

That sounds like something an user space program should simply do.
Perhaps you can add it as a new optional "-o" field in "ps" ?

-Andi

-- 
ak@linux.intel.com -- Speaking for myself only.

Re: [RFC] [PATCH procfs] Add process age

[Davidlohr Bueso]

On Mon, 2010-07-26 at 09:47 +0200, Andi Kleen wrote:
> Davidlohr Bueso <dave.bueso@gmail.com> writes:
> 
> > On Fri, 2010-07-23 at 21:35 +0300, Alexey Dobriyan wrote:
> >> On Fri, Jul 23, 2010 at 01:47:51PM -0400, Davidlohr Bueso wrote:
> >> > This patch exports a process's age, in seconds, adding a /proc/#/age file.
> >> 
> >> What for?
> >
> > I have not found any straight way to get the run time of a processes,
> > without having to interpret the start_time entry in /proc/#/stat and
> > manually subtract the uptime from it.
> 
> That sounds like something an user space program should simply do.
> Perhaps you can add it as a new optional "-o" field in "ps" ?
> 
True, but there are also many other no-so-useful files under /proc that
can also be replaced by userspace apps.

Davidlohr

Re: [RFC] [PATCH procfs] Add process age

[Andi Kleen]

> True, but there are also many other no-so-useful files under /proc that
> can also be replaced by userspace apps.

Example? 

-andi
-- 
ak@linux.intel.com -- Speaking for myself only.

Re: [RFC] [PATCH procfs] Add process age

[Davidlohr Bueso]

On Mon, 2010-07-26 at 16:48 +0200, Andi Kleen wrote:
> > True, but there are also many other no-so-useful files under /proc that
> > can also be replaced by userspace apps.
> 
> Example? 

Off the top of my head:

/proc/cmdline
Several entries in /proc/#/status, like Name, PPid, Gid, Uid, etc.

I am not criticizing this, I think some repetitive data is good for
users. Just showing that other features can also be obtained from
userspace.

Davidlohr

Re: [RFC] [PATCH procfs] Add process age

[Américo Wang]

On Mon, Jul 26, 2010 at 11:10:29AM -0400, Davidlohr Bueso wrote:
>On Mon, 2010-07-26 at 16:48 +0200, Andi Kleen wrote:
>> > True, but there are also many other no-so-useful files under /proc that
>> > can also be replaced by userspace apps.
>> 
>> Example? 
>
>Off the top of my head:
>
>/proc/cmdline

Are you kidding?

/proc/cmdlines contains lots of info, it is not only used by the kernel,
but also used by initramfs (this depends on your distribution).

Re: [RFC] [PATCH procfs] Add process age

[Davidlohr Bueso]

On Wed, 2010-07-28 at 14:56 +0800, Américo Wang wrote:
> On Mon, Jul 26, 2010 at 11:10:29AM -0400, Davidlohr Bueso wrote:
> >On Mon, 2010-07-26 at 16:48 +0200, Andi Kleen wrote:
> >> > True, but there are also many other no-so-useful files under /proc that
> >> > can also be replaced by userspace apps.
> >> 
> >> Example? 
> >
> >Off the top of my head:
> >
> >/proc/cmdline
> 
> Are you kidding?
> 
> /proc/cmdlines contains lots of info, it is not only used by the kernel,
> but also used by initramfs (this depends on your distribution).
> 
Silly me, I meant /proc/#/cmdline, in the sense that you can get the
command name from the PID in userland.

Davidlohr


--
To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html