From 66cd6cad4919f980dd21307d0150ff251762a264 Mon Sep 17 00:00:00 2001 From: "bharrosh@panasas.com" Date: Thu, 7 Oct 2010 14:28:18 -0400 Subject: [PATCH] exofs: Override read-ahead to align on stripe_size * Set all inode->i_mapping->backing_dev_info to point to the per super-block sb->s_bdi. * Calculating a read_ahead that is: - preferable 2 stripes long (Future patch will add a mount option to override this) - Minimum 128K aligned up to stripe-size - Caped to maximum-IO-sizes round down to stripe_size. (Max sizes are governed by max bio-size that fits in a page times number-of-devices) CC: Marc Dionne Signed-off-by: Boaz Harrosh --- fs/exofs/exofs.h | 2 ++ fs/exofs/inode.c | 19 +++++++++++++++---- fs/exofs/super.c | 18 ++++++++++++++++++ 3 files changed, 35 insertions(+), 4 deletions(-) diff --git a/fs/exofs/exofs.h b/fs/exofs/exofs.h index 2dc925fa1010..99fcb9126a97 100644 --- a/fs/exofs/exofs.h +++ b/fs/exofs/exofs.h @@ -256,6 +256,8 @@ static inline int exofs_oi_read(struct exofs_i_info *oi, } /* inode.c */ +unsigned exofs_max_io_pages(struct exofs_layout *layout, + unsigned expected_pages); int exofs_setattr(struct dentry *, struct iattr *); int exofs_write_begin(struct file *file, struct address_space *mapping, loff_t pos, unsigned len, unsigned flags, diff --git a/fs/exofs/inode.c b/fs/exofs/inode.c index fb9d38056103..681b3cb9b4d8 100644 --- a/fs/exofs/inode.c +++ b/fs/exofs/inode.c @@ -43,6 +43,17 @@ enum { BIO_MAX_PAGES_KMALLOC = PAGE_SIZE / sizeof(struct page *), }; +unsigned exofs_max_io_pages(struct exofs_layout *layout, + unsigned expected_pages) +{ + unsigned pages = min_t(unsigned, expected_pages, MAX_PAGES_KMALLOC); + + /* TODO: easily support bio chaining */ + pages = min_t(unsigned, pages, + layout->group_width * BIO_MAX_PAGES_KMALLOC); + return pages; +} + struct page_collect { struct exofs_sb_info *sbi; struct inode *inode; @@ -97,8 +108,7 @@ static void _pcol_reset(struct page_collect *pcol) static int pcol_try_alloc(struct page_collect *pcol) { - unsigned pages = min_t(unsigned, pcol->expected_pages, - MAX_PAGES_KMALLOC); + unsigned pages; if (!pcol->ios) { /* First time allocate io_state */ int ret = exofs_get_io_state(&pcol->sbi->layout, &pcol->ios); @@ -108,8 +118,7 @@ static int pcol_try_alloc(struct page_collect *pcol) } /* TODO: easily support bio chaining */ - pages = min_t(unsigned, pages, - pcol->sbi->layout.group_width * BIO_MAX_PAGES_KMALLOC); + pages = exofs_max_io_pages(&pcol->sbi->layout, pcol->expected_pages); for (; pages; pages >>= 1) { pcol->pages = kmalloc(pages * sizeof(struct page *), @@ -1049,6 +1058,7 @@ struct inode *exofs_iget(struct super_block *sb, unsigned long ino) memcpy(oi->i_data, fcb.i_data, sizeof(fcb.i_data)); } + inode->i_mapping->backing_dev_info = sb->s_bdi; if (S_ISREG(inode->i_mode)) { inode->i_op = &exofs_file_inode_operations; inode->i_fop = &exofs_file_operations; @@ -1149,6 +1159,7 @@ struct inode *exofs_new_inode(struct inode *dir, int mode) sbi = sb->s_fs_info; + inode->i_mapping->backing_dev_info = sb->s_bdi; sb->s_dirt = 1; inode_init_owner(inode, dir, mode); inode->i_ino = sbi->s_nextid++; diff --git a/fs/exofs/super.c b/fs/exofs/super.c index 78f5ad633d3b..e87510f4749e 100644 --- a/fs/exofs/super.c +++ b/fs/exofs/super.c @@ -390,6 +390,23 @@ static int _read_and_match_data_map(struct exofs_sb_info *sbi, unsigned numdevs, return 0; } +static unsigned __ra_pages(struct exofs_layout *layout) +{ + const unsigned _MIN_RA = 32; /* min 128K read-ahead */ + unsigned ra_pages = layout->group_width * layout->stripe_unit / + PAGE_SIZE; + unsigned max_io_pages = exofs_max_io_pages(layout, ~0); + + ra_pages *= 2; /* two stripes */ + if (ra_pages < _MIN_RA) + ra_pages = roundup(_MIN_RA, ra_pages / 2); + + if (ra_pages > max_io_pages) + ra_pages = max_io_pages; + + return ra_pages; +} + /* @odi is valid only as long as @fscb_dev is valid */ static int exofs_devs_2_odi(struct exofs_dt_device_info *dt_dev, struct osd_dev_info *odi) @@ -623,6 +640,7 @@ static int exofs_fill_super(struct super_block *sb, void *data, int silent) } /* set up operation vectors */ + sbi->bdi.ra_pages = __ra_pages(&sbi->layout); sb->s_bdi = &sbi->bdi; sb->s_fs_info = sbi; sb->s_op = &exofs_sops;