Hello.
Could somebody please check if this is acceptable.
Thanks.
Signed-off-by: Girish V. Gulawani <[email protected]>
--- linux-vanilla/fs/proc/array.c 2006-09-20 12:42:06.000000000 +0900
+++ linux/fs/proc/array.c 2006-09-30 00:16:59.000000000 +0900
@@ -248,6 +248,8 @@ static inline char * task_sig(struct tas
int num_threads = 0;
unsigned long qsize = 0;
unsigned long qlim = 0;
+ int num_children = 0;
+ struct list_head *_p;
sigemptyset(&pending);
sigemptyset(&shpending);
@@ -268,9 +270,11 @@ static inline char * task_sig(struct tas
qlim = p->signal->rlim[RLIMIT_SIGPENDING].rlim_cur;
spin_unlock_irq(&p->sighand->siglock);
}
+ list_for_each(_p, &p->children)
+ ++num_children;
read_unlock(&tasklist_lock);
- buffer += sprintf(buffer, "Threads:\t%d\n", num_threads);
+ buffer += sprintf(buffer, "Threads:\t%d\n", num_threads +
num_children);
buffer += sprintf(buffer, "SigQ:\t%lu/%lu\n", qsize, qlim);
/* render them all */
On Sep 29, 2006, at 10:18 AM, girish wrote:
>
> - buffer += sprintf(buffer, "Threads:\t%d\n", num_threads);
> + buffer += sprintf(buffer, "Threads:\t%d\n", num_threads +
> num_children);
Personally, I'd prefer the children count to be separate, something
like:
buffer += sprintf(buffer, "Threads:\t%d (%d children, %d total)",
num_threads, num_children, num_threads + num_children);
That would be rather nice, indeed.
Also, next time, make sure that linux-kernel is CC'd, not BCC'd.
---
William Pitcock
[email protected]
http://people.atheme.org/~nenolod/
http://nenolod.net
On Sep 30, 2006, at 1:31 AM, William Pitcock wrote:
>
> On Sep 29, 2006, at 10:18 AM, girish wrote:
>
>>
>> - buffer += sprintf(buffer, "Threads:\t%d\n", num_threads);
>> + buffer += sprintf(buffer, "Threads:\t%d\n", num_threads +
>> num_children);
>
> Personally, I'd prefer the children count to be separate, something
> like:
>
> buffer += sprintf(buffer, "Threads:\t%d (%d children, %d total)",
> num_threads, num_children, num_threads + num_children);
>
> That would be rather nice, indeed.
>
> Also, next time, make sure that linux-kernel is CC'd, not BCC'd.
>
> ---
> William Pitcock
> [email protected]
> http://people.atheme.org/~nenolod/
> http://nenolod.net
>
>
Agree. It indeed look better. I too had an awk script in mind, to
parse the line. I ended up removing such formatting, because not all
process spawn child thread(s), showing num_children count as zero.
That looked bit odd. So here it is again - new wine.
Thanks.
Signed-off-by: Girish V. Gulawani <[email protected]>
--- linux-vanilla/fs/proc/array.c 2006-09-20 12:42:06.000000000 +0900
+++ linux/fs/proc/array.c 2006-09-30 01:47:25.000000000 +0900
@@ -248,6 +248,8 @@ static inline char * task_sig(struct tas
int num_threads = 0;
unsigned long qsize = 0;
unsigned long qlim = 0;
+ int num_children = 0;
+ struct list_head *_p;
sigemptyset(&pending);
sigemptyset(&shpending);
@@ -268,9 +270,14 @@ static inline char * task_sig(struct tas
qlim = p->signal->rlim[RLIMIT_SIGPENDING].rlim_cur;
spin_unlock_irq(&p->sighand->siglock);
}
+ list_for_each(_p, &p->children)
+ ++num_children;
read_unlock(&tasklist_lock);
- buffer += sprintf(buffer, "Threads:\t%d\n", num_threads);
+ buffer += sprintf(buffer, "Threads:\t%d", num_threads);
+ if (num_children)
+ buffer += sprintf(buffer, " (%d children, %d total)",
num_children, num_threads + num_children);
+ buffer += sprintf(buffer, "\n");
buffer += sprintf(buffer, "SigQ:\t%lu/%lu\n", qsize, qlim);
/* render them all */
>> >
>> > - buffer += sprintf(buffer, "Threads:\t%d\n", num_threads);
>> > + buffer += sprintf(buffer, "Threads:\t%d\n", num_threads +
>> > num_children);
>>
>> Personally, I'd prefer the children count to be separate, something like:
>>
>> buffer += sprintf(buffer, "Threads:\t%d (%d children, %d total)",
>> num_threads, num_children, num_threads + num_children);
>>
>> That would be rather nice, indeed.
And I would suggest three separate lines to keep it parseable!
Jan Engelhardt
--
On Sep 30, 2006, at 2:06 AM, Jan Engelhardt wrote:
>>>>
>>>> - buffer += sprintf(buffer, "Threads:\t%d\n", num_threads);
>>>> + buffer += sprintf(buffer, "Threads:\t%d\n", num_threads +
>>>> num_children);
>>>
>>> Personally, I'd prefer the children count to be separate,
>>> something like:
>>>
>>> buffer += sprintf(buffer, "Threads:\t%d (%d children, %d total)",
>>> num_threads, num_children, num_threads + num_children);
>>>
>>> That would be rather nice, indeed.
>
> And I would suggest three separate lines to keep it parseable!
>
>
> Jan Engelhardt
> --
How about this?
buffer += sprintf(buffer, "Threads:\t%d", num_threads);
if (num_children)
buffer += sprintf(buffer, " Children: %d Total: %d",
num_children, num_threads + num_children);
buffer += sprintf(buffer, "\n");
>
> How about this?
>
> buffer += sprintf(buffer, "Threads:\t%d", num_threads);
> if (num_children)
> buffer += sprintf(buffer, " Children: %d Total: %d",
> num_children, num_threads + num_children);
> buffer += sprintf(buffer, "\n");
>
No, this:
> if (num_children)
> buffer += sprintf(buffer, "\nChildren: %d\nTotal: %d",
the newlines are essential because then you get _one_ field of
information for _each_ call of fgets().
Jan Engelhardt
--
On Sep 30, 2006, at 3:12 AM, Jan Engelhardt wrote:
>>
>> How about this?
>>
>> buffer += sprintf(buffer, "Threads:\t%d", num_threads);
>> if (num_children)
>> buffer += sprintf(buffer, " Children: %d Total: %d",
>> num_children, num_threads + num_children);
>> buffer += sprintf(buffer, "\n");
>>
>
> No, this:
>
>> if (num_children)
>> buffer += sprintf(buffer, "\nChildren: %d\nTotal: %d",
>
> the newlines are essential because then you get _one_ field of
> information for _each_ call of fgets().
>
>
>
> Jan Engelhardt
> --
This is perfect. Here is the patch.
Thanks.
Signed-off-by: Girish V. Gulawani <[email protected]>
--- linux-vanilla/fs/proc/array.c 2006-09-20 12:42:06.000000000 +0900
+++ linux/fs/proc/array.c 2006-09-30 03:18:28.000000000 +0900
@@ -248,6 +248,8 @@ static inline char * task_sig(struct tas
int num_threads = 0;
unsigned long qsize = 0;
unsigned long qlim = 0;
+ int num_children = 0;
+ struct list_head *_p;
sigemptyset(&pending);
sigemptyset(&shpending);
@@ -268,9 +270,13 @@ static inline char * task_sig(struct tas
qlim = p->signal->rlim[RLIMIT_SIGPENDING].rlim_cur;
spin_unlock_irq(&p->sighand->siglock);
}
+ list_for_each(_p, &p->children)
+ ++num_children;
read_unlock(&tasklist_lock);
buffer += sprintf(buffer, "Threads:\t%d\n", num_threads);
+ if (num_children)
+ buffer += sprintf(buffer, "Children:\t%d\nTotal:\t%d\n",
num_children, num_threads + num_children);
buffer += sprintf(buffer, "SigQ:\t%lu/%lu\n", qsize, qlim);
/* render them all */
This conflicts with the /proc changes in -mm.
Where we have manged to remove the use of the tasklist_lock.
The information in Children: is racy as it may change immediately
after you drop the lock.
Why is it interesting to report this information?
A process that cares can keep track of this in user space?
Eric
> Signed-off-by: Girish V. Gulawani <[email protected]>
>
> --- linux-vanilla/fs/proc/array.c 2006-09-20 12:42:06.000000000 +0900
> +++ linux/fs/proc/array.c 2006-09-30 03:18:28.000000000 +0900
> @@ -248,6 +248,8 @@ static inline char * task_sig(struct tas
> int num_threads = 0;
> unsigned long qsize = 0;
> unsigned long qlim = 0;
> + int num_children = 0;
> + struct list_head *_p;
>
> sigemptyset(&pending);
> sigemptyset(&shpending);
> @@ -268,9 +270,13 @@ static inline char * task_sig(struct tas
> qlim = p->signal->rlim[RLIMIT_SIGPENDING].rlim_cur;
> spin_unlock_irq(&p->sighand->siglock);
> }
> + list_for_each(_p, &p->children)
> + ++num_children;
> read_unlock(&tasklist_lock);
>
> buffer += sprintf(buffer, "Threads:\t%d\n", num_threads);
> + if (num_children)
> + buffer += sprintf(buffer, "Children:\t%d\nTotal:\t%d\n",
> num_children, num_threads + num_children);
> buffer += sprintf(buffer, "SigQ:\t%lu/%lu\n", qsize, qlim);
>
> /* render them all */
>
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to [email protected]
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/