linux-kernel - Re: [PATCH] fs/proc: Per thread I/O accounting

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <486E3A46.4000206@gmail.com>
Date:	Fri, 04 Jul 2008 16:57:10 +0200
From:	Andrea Righi <righi.andrea@...il.com>
To:	Tomas Smetana <t.smetana@...il.com>
CC:	linux-kernel@...r.kernel.org
Subject: Re: [PATCH] fs/proc: Per thread I/O accounting

Tomas Smetana wrote:
> Hello,
>   here's an attempt to fix the bug #10702: The /proc/#/io doesn't aggregate
> I/O statistics of all the process' threads -- just the main one. The patch
> fixes this and also adds /proc/#/task/#/io file for per-thread accounting
> which should make I/O accounting consistent with CPU stats.
> 
> Reference: http://bugzilla.kernel.org/show_bug.cgi?id=10702
> 
> Please cc: me in the answers.

Hi Tomas,

have a look at distinct-tgid-tid-i-o-statistics.patch in -mm (also
available here http://lwn.net/Articles/283937/).

Your patch seems ok (except the rcu locking, see below), but it doesn't
account the i/o activity of a task if it creates a lot of short-lived
children doing i/o.

> 
> 
> Signed-off-by: Tomas Smetana <t.smetana@...il.com>
> ---
>  fs/proc/base.c |   48 ++++++++++++++++++++++++++++++++++++++++++++++++
>  1 files changed, 48 insertions(+), 0 deletions(-)
> 
> diff --git a/fs/proc/base.c b/fs/proc/base.c
> index 3b45537..5437ae3 100644
> --- a/fs/proc/base.c
> +++ b/fs/proc/base.c
> @@ -2377,6 +2377,51 @@ static int proc_base_fill_cache(struct file *filp, void *dirent,
>  #ifdef CONFIG_TASK_IO_ACCOUNTING
>  static int proc_pid_io_accounting(struct task_struct *task, char *buffer)
>  {
> +#ifdef CONFIG_TASK_XACCT
> +	unsigned long long rchar = 0, wchar = 0;
> +	unsigned long long syscr = 0, syscw = 0;
> +#endif
> +	unsigned long long read_bytes = 0, write_bytes = 0;
> +	unsigned long long cancelled_write_bytes = 0;
> +	struct task_struct *t = task;
> +

rcu_read_lock();

> +	do {
> +#ifdef CONFIG_TASK_XACCT
> +		rchar += (unsigned long long)t->rchar;
> +		wchar += (unsigned long long)t->wchar;
> +		syscr += (unsigned long long)t->syscr;
> +		syscw += (unsigned long long)t->syscw;
> +#endif
> +		read_bytes += (unsigned long long)t->ioac.read_bytes;
> +		write_bytes += (unsigned long long)t->ioac.write_bytes;
> +		cancelled_write_bytes +=
> +		    (unsigned long long)t->ioac.cancelled_write_bytes;
> +		t = next_thread(t);
> +	} while (t != task);

rcu_read_unlock();

> +
> +	return sprintf(buffer,
> +#ifdef CONFIG_TASK_XACCT
> +			"rchar: %llu\n"
> +			"wchar: %llu\n"
> +			"syscr: %llu\n"
> +			"syscw: %llu\n"
> +#endif
> +			"read_bytes: %llu\n"
> +			"write_bytes: %llu\n"
> +			"cancelled_write_bytes: %llu\n",
> +#ifdef CONFIG_TASK_XACCT
> +			rchar,
> +			wchar,
> +			syscr,
> +			syscw,
> +#endif
> +			read_bytes,
> +			write_bytes,
> +			cancelled_write_bytes);
> +}
> +
> +static int proc_tid_io_accounting(struct task_struct *task, char *buffer)
> +{
>  	return sprintf(buffer,
>  #ifdef CONFIG_TASK_XACCT
>  			"rchar: %llu\n"
> @@ -2796,6 +2841,9 @@ static const struct pid_entry tid_base_stuff[] = {
>  #ifdef CONFIG_FAULT_INJECTION
>  	REG("make-it-fail", S_IRUGO|S_IWUSR, fault_inject),
>  #endif
> +#ifdef CONFIG_TASK_IO_ACCOUNTING
> +	INF("io",	S_IRUGO, tid_io_accounting)
> +#endif
>  };
>  
>  static int proc_tid_base_readdir(struct file * filp,

-Andrea
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/