| Message ID | 1348404995-14372-4-git-send-email-zwu.kernel@gmail.com (mailing list archive) |
|---|---|
| State | Not Applicable, archived |
| Headers | show |
On Sun, Sep 23, 2012 at 08:56:28PM +0800, zwu.kernel@gmail.com wrote: > From: Zhi Yong Wu <wuzhy@linux.vnet.ibm.com> > > Introduce one new mount option '-o hottrack', > and add its parsing support. > Its usage looks like: > mount -o hottrack > mount -o nouser,hottrack > mount -o nouser,hottrack,loop > mount -o hottrack,nouser I think that this option parsing should be done by the filesystem, even though the tracking functionality is in the VFS. That way ony the filesystems that can use the tracking information will turn it on, rather than being able to turn it on for everything regardless of whether it is useful or not. Along those lines, just using a normal superblock flag to indicate it is active (e.g. MS_HOT_INODE_TRACKING in sb->s_flags) means you don't need to allocate the sb->s_hot_info structure just to be able to check whether we are tracking hot inodes or not. This then means the hot inode tracking for the superblock can be initialised by the filesystem as part of it's fill_super method, along with the filesystem specific code that will use the hot tracking information the VFS gathers.... Cheers, Dave.
On Tue, Sep 25, 2012 at 5:28 PM, Dave Chinner <david@fromorbit.com> wrote: > On Sun, Sep 23, 2012 at 08:56:28PM +0800, zwu.kernel@gmail.com wrote: >> From: Zhi Yong Wu <wuzhy@linux.vnet.ibm.com> >> >> Introduce one new mount option '-o hottrack', >> and add its parsing support. >> Its usage looks like: >> mount -o hottrack >> mount -o nouser,hottrack >> mount -o nouser,hottrack,loop >> mount -o hottrack,nouser > > I think that this option parsing should be done by the filesystem, > even though the tracking functionality is in the VFS. That way ony > the filesystems that can use the tracking information will turn it > on, rather than being able to turn it on for everything regardless > of whether it is useful or not. > > Along those lines, just using a normal superblock flag to indicate > it is active (e.g. MS_HOT_INODE_TRACKING in sb->s_flags) means you > don't need to allocate the sb->s_hot_info structure just to be able > to check whether we are tracking hot inodes or not. > > This then means the hot inode tracking for the superblock can be > initialised by the filesystem as part of it's fill_super method, > along with the filesystem specific code that will use the hot > tracking information the VFS gathers.... I can see what you mean, but don't know if other guys also agree with this. If so, all FS specific code which use hot tracking feature wll have to add the same chunk of code in it fill_super method. Is it good? > > Cheers, > > Dave. > -- > Dave Chinner > david@fromorbit.com
On Wed, Sep 26, 2012 at 10:56:08AM +0800, Zhi Yong Wu wrote: > On Tue, Sep 25, 2012 at 5:28 PM, Dave Chinner <david@fromorbit.com> wrote: > > On Sun, Sep 23, 2012 at 08:56:28PM +0800, zwu.kernel@gmail.com wrote: > >> From: Zhi Yong Wu <wuzhy@linux.vnet.ibm.com> > >> > >> Introduce one new mount option '-o hottrack', > >> and add its parsing support. > >> Its usage looks like: > >> mount -o hottrack > >> mount -o nouser,hottrack > >> mount -o nouser,hottrack,loop > >> mount -o hottrack,nouser > > > > I think that this option parsing should be done by the filesystem, > > even though the tracking functionality is in the VFS. That way ony > > the filesystems that can use the tracking information will turn it > > on, rather than being able to turn it on for everything regardless > > of whether it is useful or not. > > > > Along those lines, just using a normal superblock flag to indicate > > it is active (e.g. MS_HOT_INODE_TRACKING in sb->s_flags) means you > > don't need to allocate the sb->s_hot_info structure just to be able > > to check whether we are tracking hot inodes or not. > > > > This then means the hot inode tracking for the superblock can be > > initialised by the filesystem as part of it's fill_super method, > > along with the filesystem specific code that will use the hot > > tracking information the VFS gathers.... > I can see what you mean, but don't know if other guys also agree with this. > If so, all FS specific code which use hot tracking feature wll have to add > the same chunk of code in it fill_super method. Is it good? Most filesystems will only need to add 3-4 lines of code to their existing parser, so it's not a big deal I think... Cheers, Dave.
On Thu, Sep 27, 2012 at 10:20 AM, Dave Chinner <david@fromorbit.com> wrote: > On Wed, Sep 26, 2012 at 10:56:08AM +0800, Zhi Yong Wu wrote: >> On Tue, Sep 25, 2012 at 5:28 PM, Dave Chinner <david@fromorbit.com> wrote: >> > On Sun, Sep 23, 2012 at 08:56:28PM +0800, zwu.kernel@gmail.com wrote: >> >> From: Zhi Yong Wu <wuzhy@linux.vnet.ibm.com> >> >> >> >> Introduce one new mount option '-o hottrack', >> >> and add its parsing support. >> >> Its usage looks like: >> >> mount -o hottrack >> >> mount -o nouser,hottrack >> >> mount -o nouser,hottrack,loop >> >> mount -o hottrack,nouser >> > >> > I think that this option parsing should be done by the filesystem, >> > even though the tracking functionality is in the VFS. That way ony >> > the filesystems that can use the tracking information will turn it >> > on, rather than being able to turn it on for everything regardless >> > of whether it is useful or not. >> > >> > Along those lines, just using a normal superblock flag to indicate >> > it is active (e.g. MS_HOT_INODE_TRACKING in sb->s_flags) means you >> > don't need to allocate the sb->s_hot_info structure just to be able >> > to check whether we are tracking hot inodes or not. >> > >> > This then means the hot inode tracking for the superblock can be >> > initialised by the filesystem as part of it's fill_super method, >> > along with the filesystem specific code that will use the hot >> > tracking information the VFS gathers.... >> I can see what you mean, but don't know if other guys also agree with this. >> If so, all FS specific code which use hot tracking feature wll have to add >> the same chunk of code in it fill_super method. Is it good? > > Most filesystems will only need to add 3-4 lines of code to their > existing parser, so it's not a big deal I think... OK, i try to do it. thanks a lot. > > Cheers, > > Dave. > -- > Dave Chinner > david@fromorbit.com
On Tue, Sep 25, 2012 at 5:28 PM, Dave Chinner <david@fromorbit.com> wrote: > On Sun, Sep 23, 2012 at 08:56:28PM +0800, zwu.kernel@gmail.com wrote: >> From: Zhi Yong Wu <wuzhy@linux.vnet.ibm.com> >> >> Introduce one new mount option '-o hottrack', >> and add its parsing support. >> Its usage looks like: >> mount -o hottrack >> mount -o nouser,hottrack >> mount -o nouser,hottrack,loop >> mount -o hottrack,nouser > > I think that this option parsing should be done by the filesystem, > even though the tracking functionality is in the VFS. That way ony > the filesystems that can use the tracking information will turn it > on, rather than being able to turn it on for everything regardless > of whether it is useful or not. > > Along those lines, just using a normal superblock flag to indicate > it is active (e.g. MS_HOT_INODE_TRACKING in sb->s_flags) means you > don't need to allocate the sb->s_hot_info structure just to be able If we don't allocate one sb->s_hot_info, where will those hash list head and btree roots locate? > to check whether we are tracking hot inodes or not. > > This then means the hot inode tracking for the superblock can be > initialised by the filesystem as part of it's fill_super method, > along with the filesystem specific code that will use the hot > tracking information the VFS gathers.... > > Cheers, > > Dave. > -- > Dave Chinner > david@fromorbit.com
On Thu, Sep 27, 2012 at 01:25:34PM +0800, Zhi Yong Wu wrote: > On Tue, Sep 25, 2012 at 5:28 PM, Dave Chinner <david@fromorbit.com> wrote: > > On Sun, Sep 23, 2012 at 08:56:28PM +0800, zwu.kernel@gmail.com wrote: > >> From: Zhi Yong Wu <wuzhy@linux.vnet.ibm.com> > >> > >> Introduce one new mount option '-o hottrack', > >> and add its parsing support. > >> Its usage looks like: > >> mount -o hottrack > >> mount -o nouser,hottrack > >> mount -o nouser,hottrack,loop > >> mount -o hottrack,nouser > > > > I think that this option parsing should be done by the filesystem, > > even though the tracking functionality is in the VFS. That way ony > > the filesystems that can use the tracking information will turn it > > on, rather than being able to turn it on for everything regardless > > of whether it is useful or not. > > > > Along those lines, just using a normal superblock flag to indicate > > it is active (e.g. MS_HOT_INODE_TRACKING in sb->s_flags) means you > > don't need to allocate the sb->s_hot_info structure just to be able > If we don't allocate one sb->s_hot_info, where will those hash list > head and btree roots locate? I wrote that thinking (mistakenly) that s-hot)info was dynamically allocated rather than being embedded in the struct super_block. Indeed, if the mount option is held in s_flags, then it could be dynamically allocated, but I don't think that's really necessary... Cheers, Dave.
On Thu, Sep 27, 2012 at 3:05 PM, Dave Chinner <david@fromorbit.com> wrote: > On Thu, Sep 27, 2012 at 01:25:34PM +0800, Zhi Yong Wu wrote: >> On Tue, Sep 25, 2012 at 5:28 PM, Dave Chinner <david@fromorbit.com> wrote: >> > On Sun, Sep 23, 2012 at 08:56:28PM +0800, zwu.kernel@gmail.com wrote: >> >> From: Zhi Yong Wu <wuzhy@linux.vnet.ibm.com> >> >> >> >> Introduce one new mount option '-o hottrack', >> >> and add its parsing support. >> >> Its usage looks like: >> >> mount -o hottrack >> >> mount -o nouser,hottrack >> >> mount -o nouser,hottrack,loop >> >> mount -o hottrack,nouser >> > >> > I think that this option parsing should be done by the filesystem, >> > even though the tracking functionality is in the VFS. That way ony >> > the filesystems that can use the tracking information will turn it >> > on, rather than being able to turn it on for everything regardless >> > of whether it is useful or not. >> > >> > Along those lines, just using a normal superblock flag to indicate >> > it is active (e.g. MS_HOT_INODE_TRACKING in sb->s_flags) means you >> > don't need to allocate the sb->s_hot_info structure just to be able >> If we don't allocate one sb->s_hot_info, where will those hash list >> head and btree roots locate? > > I wrote that thinking (mistakenly) that s-hot)info was dynamically > allocated rather than being embedded in the struct super_block. > > Indeed, if the mount option is held in s_flags, then it could be > dynamically allocated, but I don't think that's really necessary... ah, you prefer allocating it, OK, let me try. thanks for your explaination. > > Cheers, > > Dave. > -- > Dave Chinner > david@fromorbit.com
diff --git a/fs/hot_tracking.c b/fs/hot_tracking.c index 52ed926..f97e8a6 100644 --- a/fs/hot_tracking.c +++ b/fs/hot_tracking.c @@ -465,6 +465,40 @@ void hot_rb_update_freqs(struct inode *inode, u64 start, } /* + * Regular mount options parser for -hottrack option. + * return false if no -hottrack is specified; + * otherwise return true. And the -hottrack will be + * removed from options. + */ +bool hot_track_parse_options(char *options) +{ + long len; + char *p; + static char opts_hot[] = "hottrack"; + + if (!options) + return false; + + p = strstr(options, opts_hot); + if (!p) + return false; + + while (p) { + len = options + strlen(options) - (p + strlen(opts_hot)); + if (len == 0) { + options[0] = '\0'; + break; + } + + memmove(p, p + strlen(opts_hot) + 1, len); + p = strstr(options, opts_hot); + } + + printk(KERN_INFO "vfs: turning on hot data tracking\n"); + return true; +} + +/* * Initialize kmem cache for hot_inode_item * and hot_range_item */ diff --git a/fs/hot_tracking.h b/fs/hot_tracking.h index 2ba29e4..6bd09eb 100644 --- a/fs/hot_tracking.h +++ b/fs/hot_tracking.h @@ -37,6 +37,7 @@ void hot_rb_free_hot_inode_item(struct hot_inode_item *he); void hot_rb_update_freqs(struct inode *inode, u64 start, u64 len, int rw); +bool hot_track_parse_options(char *options); void __init hot_track_cache_init(void); #endif /* __HOT_TRACKING__ */ diff --git a/fs/super.c b/fs/super.c index 0902cfa..7eb3b0c 100644 --- a/fs/super.c +++ b/fs/super.c @@ -34,6 +34,7 @@ #include <linux/cleancache.h> #include <linux/fsnotify.h> #include <linux/lockdep.h> +#include "hot_tracking.h" #include "internal.h" @@ -1125,6 +1126,7 @@ mount_fs(struct file_system_type *type, int flags, const char *name, void *data) struct dentry *root; struct super_block *sb; char *secdata = NULL; + bool hottrack = false; int error = -ENOMEM; if (data && !(type->fs_flags & FS_BINARY_MOUNTDATA)) { @@ -1137,6 +1139,9 @@ mount_fs(struct file_system_type *type, int flags, const char *name, void *data) goto out_free_secdata; } + if (data && hot_track_parse_options(data)) + hottrack = true; + root = type->mount(type, flags, name, data); if (IS_ERR(root)) { error = PTR_ERR(root); diff --git a/include/linux/hot_tracking.h b/include/linux/hot_tracking.h index a566f91..bb2a41c 100644 --- a/include/linux/hot_tracking.h +++ b/include/linux/hot_tracking.h @@ -20,6 +20,11 @@ #include <linux/rbtree.h> #include <linux/kref.h> +/* + * Flags for hot data tracking mount options. + */ +#define HOT_MOUNT_HOT_TRACK (1 << 0) + /* A tree that sits on the hot_info */ struct hot_inode_tree { struct rb_root map; @@ -89,6 +94,8 @@ struct hot_range_item { }; struct hot_info { + unsigned long mount_opt; + /* red-black tree that keeps track of fs-wide hot data */ struct hot_inode_tree hot_inode_tree; };