From: Vivek Goyal <vgoyal@redhat.com>
To: Gui Jianfeng <guijianfeng@cn.fujitsu.com>
Cc: Paul Menage <menage@google.com>,
linux-kernel@vger.kernel.org,
containers@lists.linux-foundation.org, dm-devel@redhat.com,
jens.axboe@oracle.com, nauman@google.com, dpshah@google.com,
lizf@cn.fujitsu.com, mikew@google.com, fchecconi@gmail.com,
paolo.valente@unimore.it, ryov@valinux.co.jp,
fernando@oss.ntt.co.jp, s-uchida@ap.jp.nec.com,
taka@valinux.co.jp, jmoyer@redhat.com, dhaval@linux.vnet.ibm.com,
balbir@linux.vnet.ibm.com, righi.andrea@gmail.com,
m-ikeda@ds.jp.nec.com, jbaron@redhat.com, agk@redhat.com,
snitzer@redhat.com, akpm@linux-foundation.org,
peterz@infradead.org
Subject: Re: [PATCH] io-controller: do some changes of io.policy interface
Date: Thu, 25 Jun 2009 08:55:13 -0400 [thread overview]
Message-ID: <20090625125513.GA25439@redhat.com> (raw)
In-Reply-To: <4A435038.60406@cn.fujitsu.com>
On Thu, Jun 25, 2009 at 06:23:52PM +0800, Gui Jianfeng wrote:
> Paul Menage wrote:
> > On Fri, Jun 19, 2009 at 1:37 PM, Vivek Goyal<vgoyal@redhat.com> wrote:
> >> You can use the following format to play with the new interface.
> >> #echo DEV:weight:ioprio_class > /patch/to/cgroup/policy
> >> weight=0 means removing the policy for DEV.
> >>
> >> Examples:
> >> Configure weight=300 ioprio_class=2 on /dev/hdb in this cgroup
> >> # echo /dev/hdb:300:2 > io.policy
> >> # cat io.policy
> >> dev weight class
> >> /dev/hdb 300 2
> >
> > I think that the read and write should be consistent. Can you just use
> > white-space separation for both, rather than colon-separation for
> > writes and white-space separation for reads?
> >
> > Also, storing device inode paths statically as strings into the
> > io_policy structure seems wrong, since it's quite possible for the
> > device node that was used originally to be gone by the time that
> > someone reads the io.policy file, or renamed, or even replaced with an
> > inode that refers to to a different block device
> >
> > My preferred alternatives would be:
> >
> > - read/write the value as a device number rather than a name
> > - read/write the block device's actual name (e.g. hda or sda) rather
> > than a path to the inode
> >
>
> Hi Paul, Vivek
>
> Here is a patch to fix the issue Paul raised.
>
> This patch achives the following goals
> 1 According to Paul's comment, Modifing io.policy interface to
> use device number for read/write directly.
> 2 Just use white-space separation for both, rather than colon-
> separation for writes and white-space separation for reads.
> 3 Do more strict checking for inputting.
>
> old interface:
> Configure weight=300 ioprio_class=2 on /dev/hdb in this cgroup
> # echo "/dev/hdb:300:2" > io.policy
> # cat io.policy
> dev weight class
> /dev/hdb 300 2
>
> new interface:
> Configure weight=300 ioprio_class=2 on /dev/hdb in this cgroup
> # echo "3:64 300 2" > io.policy
> # cat io.policy
> dev weight class
> 3:64 300 2
>
> Signed-off-by: Gui Jianfeng <guijianfeng@cn.fujitsu.com>
> ---
> block/elevator-fq.c | 59 ++++++++++++++++++++++++++++++++++----------------
> block/elevator-fq.h | 1 -
> 2 files changed, 40 insertions(+), 20 deletions(-)
>
> diff --git a/block/elevator-fq.c b/block/elevator-fq.c
> index d779282..83c831b 100644
> --- a/block/elevator-fq.c
> +++ b/block/elevator-fq.c
> @@ -1895,12 +1895,12 @@ static int io_cgroup_policy_read(struct cgroup *cgrp, struct cftype *cft,
> if (list_empty(&iocg->policy_list))
> goto out;
>
> - seq_printf(m, "dev weight class\n");
> + seq_printf(m, "dev\tweight\tclass\n");
>
> spin_lock_irq(&iocg->lock);
> list_for_each_entry(pn, &iocg->policy_list, node) {
> - seq_printf(m, "%s %lu %lu\n", pn->dev_name,
> - pn->weight, pn->ioprio_class);
> + seq_printf(m, "%u:%u\t%lu\t%lu\n", MAJOR(pn->dev),
> + MINOR(pn->dev), pn->weight, pn->ioprio_class);
> }
> spin_unlock_irq(&iocg->lock);
> out:
> @@ -1936,44 +1936,65 @@ static struct io_policy_node *policy_search_node(const struct io_cgroup *iocg,
> return NULL;
> }
>
> -static int devname_to_devnum(const char *buf, dev_t *dev)
> +static int check_dev_num(dev_t dev)
> {
> - struct block_device *bdev;
> + int part = 0;
> struct gendisk *disk;
> - int part;
>
> - bdev = lookup_bdev(buf);
> - if (IS_ERR(bdev))
> + disk = get_gendisk(dev, &part);
> + if (!disk || part)
> return -ENODEV;
>
> - disk = get_gendisk(bdev->bd_dev, &part);
> - if (part)
> - return -EINVAL;
> -
> - *dev = MKDEV(disk->major, disk->first_minor);
> - bdput(bdev);
> -
> return 0;
> }
>
> static int policy_parse_and_set(char *buf, struct io_policy_node *newpn)
> {
> - char *s[3], *p;
> + char *s[4], *p, *major_s = NULL, *minor_s = NULL;
> int ret;
> + unsigned long major, minor;
> int i = 0;
> + dev_t dev;
>
> memset(s, 0, sizeof(s));
> - while ((p = strsep(&buf, ":")) != NULL) {
> + while ((p = strsep(&buf, " ")) != NULL) {
> if (!*p)
> continue;
> s[i++] = p;
> +
> + /* Prevent from inputing too many things */
> + if (i == 4)
> + break;
> }
>
> - ret = devname_to_devnum(s[0], &newpn->dev);
> + if (i != 3)
> + return -EINVAL;
> +
> + p = strsep(&s[0], ":");
> + if (p != NULL)
> + major_s = p;
> + else
> + return -EINVAL;
> +
> + minor_s = s[0];
> + if (!minor_s)
> + return -EINVAL;
> +
> + ret = strict_strtoul(major_s, 10, &major);
> + if (ret)
> + return -EINVAL;
> +
> + ret = strict_strtoul(minor_s, 10, &minor);
> + if (ret)
> + return -EINVAL;
> +
> + dev = MKDEV(major, minor);
> +
> + ret = check_dev_num(dev);
> if (ret)
> return ret;
>
> - strcpy(newpn->dev_name, s[0]);
> + newpn->dev = dev;
>
> if (s[1] == NULL)
> return -EINVAL;
> diff --git a/block/elevator-fq.h b/block/elevator-fq.h
> index b3193f8..7722ebe 100644
> --- a/block/elevator-fq.h
> +++ b/block/elevator-fq.h
> @@ -286,7 +286,6 @@ struct io_group {
>
> struct io_policy_node {
> struct list_head node;
> - char dev_name[32];
> dev_t dev;
> unsigned long weight;
> unsigned long ioprio_class;
Hi Gui,
Thanks for the patch. "unsigned long" for ioprio_class is too big. How
about using "unsigned short"? I noticed that in io_cgroup also we are
using "unsigned long". I will fix that.
For storing weight now we are planning to use "unsigned int". Can you
please switch to that.
Thanks
Vivek
next prev parent reply other threads:[~2009-06-25 12:56 UTC|newest]
Thread overview: 65+ messages / expand[flat|nested] mbox.gz Atom feed top
2009-06-19 20:37 [RFC] IO scheduler based io controller (V5) Vivek Goyal
2009-06-19 20:37 ` [PATCH 01/20] io-controller: Documentation Vivek Goyal
2009-06-19 20:37 ` [PATCH 02/20] io-controller: Common flat fair queuing code in elevaotor layer Vivek Goyal
2009-06-22 8:46 ` Balbir Singh
2009-06-22 12:43 ` Fabio Checconi
2009-06-23 2:43 ` Vivek Goyal
2009-06-23 4:10 ` Fabio Checconi
2009-06-23 7:32 ` Balbir Singh
2009-06-23 13:42 ` Fabio Checconi
2009-06-23 2:05 ` Vivek Goyal
2009-06-23 2:20 ` Jeff Moyer
2009-06-30 6:40 ` Gui Jianfeng
2009-07-01 1:28 ` Vivek Goyal
2009-07-01 9:24 ` Gui Jianfeng
2009-06-19 20:37 ` [PATCH 03/20] io-controller: Charge for time slice based on average disk rate Vivek Goyal
2009-06-19 20:37 ` [PATCH 04/20] io-controller: Modify cfq to make use of flat elevator fair queuing Vivek Goyal
2009-06-19 20:37 ` [PATCH 05/20] io-controller: Common hierarchical fair queuing code in elevaotor layer Vivek Goyal
2009-06-29 5:27 ` [PATCH] io-controller: optimization for iog deletion when elevator exiting Gui Jianfeng
2009-06-29 14:06 ` Vivek Goyal
2009-06-30 17:14 ` Nauman Rafique
2009-07-01 1:34 ` Vivek Goyal
2009-06-19 20:37 ` [PATCH 06/20] io-controller: cfq changes to use hierarchical fair queuing code in elevaotor layer Vivek Goyal
2009-06-19 20:37 ` [PATCH 07/20] io-controller: Export disk time used and nr sectors dipatched through cgroups Vivek Goyal
2009-06-23 12:10 ` Gui Jianfeng
2009-06-23 14:38 ` Vivek Goyal
2009-06-19 20:37 ` [PATCH 08/20] io-controller: idle for sometime on sync queue before expiring it Vivek Goyal
2009-06-30 7:49 ` [PATCH] io-controller: Don't expire an idle ioq if it's the only ioq in hierarchy Gui Jianfeng
2009-07-01 1:32 ` Vivek Goyal
2009-07-01 1:40 ` Gui Jianfeng
2009-06-19 20:37 ` [PATCH 09/20] io-controller: Separate out queue and data Vivek Goyal
2009-06-19 20:37 ` [PATCH 10/20] io-conroller: Prepare elevator layer for single queue schedulers Vivek Goyal
2009-06-19 20:37 ` [PATCH 11/20] io-controller: noop changes for hierarchical fair queuing Vivek Goyal
2009-06-19 20:37 ` [PATCH 12/20] io-controller: deadline " Vivek Goyal
2009-06-19 20:37 ` [PATCH 13/20] io-controller: anticipatory " Vivek Goyal
2009-06-19 20:37 ` [PATCH 14/20] blkio_cgroup patches from Ryo to track async bios Vivek Goyal
2009-06-19 20:37 ` [PATCH 15/20] io-controller: map async requests to appropriate cgroup Vivek Goyal
2009-06-22 1:45 ` Gui Jianfeng
2009-06-22 15:39 ` Vivek Goyal
2009-06-19 20:37 ` [PATCH 16/20] io-controller: Per cgroup request descriptor support Vivek Goyal
2009-06-19 20:37 ` [PATCH 17/20] io-controller: Per io group bdi congestion interface Vivek Goyal
2009-06-19 20:37 ` [PATCH 18/20] io-controller: Support per cgroup per device weights and io class Vivek Goyal
2009-06-24 21:52 ` Paul Menage
2009-06-25 10:23 ` [PATCH] io-controller: do some changes of io.policy interface Gui Jianfeng
2009-06-25 12:55 ` Vivek Goyal [this message]
2009-06-26 0:27 ` Gui Jianfeng
2009-06-26 0:59 ` Gui Jianfeng
2009-06-19 20:37 ` [PATCH 19/20] io-controller: Debug hierarchical IO scheduling Vivek Goyal
2009-06-19 20:37 ` [PATCH 20/20] io-controller: experimental debug patch for async queue wait before expiry Vivek Goyal
2009-06-22 7:44 ` [PATCH] io-controller: Preempt a non-rt queue if a rt ioq is present in ancestor or sibling groups Gui Jianfeng
2009-06-22 17:21 ` Vivek Goyal
2009-06-23 6:44 ` Gui Jianfeng
2009-06-23 14:02 ` Vivek Goyal
2009-06-24 9:20 ` Gui Jianfeng
2009-06-26 8:13 ` [PATCH 1/2] io-controller: Prepare a rt ioq list in efqd to keep track of busy rt ioqs Gui Jianfeng
2009-06-26 8:13 ` [PATCH 2/2] io-controller: make rt preemption happen in the whole hierarchy Gui Jianfeng
2009-06-26 12:39 ` Vivek Goyal
2009-06-21 15:21 ` [RFC] IO scheduler based io controller (V5) Balbir Singh
2009-06-22 15:30 ` Vivek Goyal
2009-06-22 15:40 ` Jeff Moyer
2009-06-22 16:02 ` Vivek Goyal
2009-06-22 16:06 ` Jeff Moyer
2009-06-22 17:08 ` Vivek Goyal
2009-06-23 6:52 ` Balbir Singh
2009-06-29 16:04 ` Vladislav Bolkhovitin
2009-06-29 17:23 ` Vivek Goyal
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=20090625125513.GA25439@redhat.com \
--to=vgoyal@redhat.com \
--cc=agk@redhat.com \
--cc=akpm@linux-foundation.org \
--cc=balbir@linux.vnet.ibm.com \
--cc=containers@lists.linux-foundation.org \
--cc=dhaval@linux.vnet.ibm.com \
--cc=dm-devel@redhat.com \
--cc=dpshah@google.com \
--cc=fchecconi@gmail.com \
--cc=fernando@oss.ntt.co.jp \
--cc=guijianfeng@cn.fujitsu.com \
--cc=jbaron@redhat.com \
--cc=jens.axboe@oracle.com \
--cc=jmoyer@redhat.com \
--cc=linux-kernel@vger.kernel.org \
--cc=lizf@cn.fujitsu.com \
--cc=m-ikeda@ds.jp.nec.com \
--cc=menage@google.com \
--cc=mikew@google.com \
--cc=nauman@google.com \
--cc=paolo.valente@unimore.it \
--cc=peterz@infradead.org \
--cc=righi.andrea@gmail.com \
--cc=ryov@valinux.co.jp \
--cc=s-uchida@ap.jp.nec.com \
--cc=snitzer@redhat.com \
--cc=taka@valinux.co.jp \
/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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox