From 1b9c332b6c92e992b1971a08412c6f460a54b514 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Wed, 17 Aug 2011 10:19:52 -0400 Subject: Btrfs: only reserve space in fallocate if we have to do a preallocate Lukas found a problem where if he tries to fallocate over the same region twice and the first fallocate took up all the space we would fail with ENOSPC. This is because we reserve the total space we want to use for fallocate, regardless of wether or not we will have to actually preallocate. So instead move the check into the loop where we actually have to do the preallocate. Thanks, Tested-by: Lukas Czerner Signed-off-by: Josef Bacik --- fs/btrfs/file.c | 22 ++++++++++++++++------ 1 file changed, 16 insertions(+), 6 deletions(-) (limited to 'fs/btrfs/file.c') diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c index e4e57d59edb7..de569af766fe 100644 --- a/fs/btrfs/file.c +++ b/fs/btrfs/file.c @@ -1615,10 +1615,6 @@ static long btrfs_fallocate(struct file *file, int mode, goto out; } - ret = btrfs_check_data_free_space(inode, alloc_end - alloc_start); - if (ret) - goto out; - locked_end = alloc_end - 1; while (1) { struct btrfs_ordered_extent *ordered; @@ -1664,11 +1660,27 @@ static long btrfs_fallocate(struct file *file, int mode, if (em->block_start == EXTENT_MAP_HOLE || (cur_offset >= inode->i_size && !test_bit(EXTENT_FLAG_PREALLOC, &em->flags))) { + + /* + * Make sure we have enough space before we do the + * allocation. + */ + ret = btrfs_check_data_free_space(inode, last_byte - + cur_offset); + if (ret) { + free_extent_map(em); + break; + } + ret = btrfs_prealloc_file_range(inode, mode, cur_offset, last_byte - cur_offset, 1 << inode->i_blkbits, offset + len, &alloc_hint); + + /* Let go of our reservation. */ + btrfs_free_reserved_data_space(inode, last_byte - + cur_offset); if (ret < 0) { free_extent_map(em); break; @@ -1694,8 +1706,6 @@ static long btrfs_fallocate(struct file *file, int mode, } unlock_extent_cached(&BTRFS_I(inode)->io_tree, alloc_start, locked_end, &cached_state, GFP_NOFS); - - btrfs_free_reserved_data_space(inode, alloc_end - alloc_start); out: mutex_unlock(&inode->i_mutex); return ret; -- cgit v1.2.3 From 3b16a4e3c355ee3c790473decfcf83d4faeb8ce0 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Wed, 21 Sep 2011 15:05:58 -0400 Subject: Btrfs: use the inode's mapping mask for allocating pages Johannes pointed out we were allocating only kernel pages for doing writes, which is kind of a big deal if you are on 32bit and have more than a gig of ram. So fix our allocations to use the mapping's gfp but still clear __GFP_FS so we don't re-enter. Thanks, Reported-by: Johannes Weiner Signed-off-by: Josef Bacik --- fs/btrfs/file.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'fs/btrfs/file.c') diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c index de569af766fe..f2e928289600 100644 --- a/fs/btrfs/file.c +++ b/fs/btrfs/file.c @@ -1069,6 +1069,7 @@ static noinline int prepare_pages(struct btrfs_root *root, struct file *file, int i; unsigned long index = pos >> PAGE_CACHE_SHIFT; struct inode *inode = fdentry(file)->d_inode; + gfp_t mask = btrfs_alloc_write_mask(inode->i_mapping); int err = 0; int faili = 0; u64 start_pos; @@ -1080,7 +1081,7 @@ static noinline int prepare_pages(struct btrfs_root *root, struct file *file, again: for (i = 0; i < num_pages; i++) { pages[i] = find_or_create_page(inode->i_mapping, index + i, - GFP_NOFS); + mask); if (!pages[i]) { faili = i - 1; err = -ENOMEM; -- cgit v1.2.3 From ef3d0fd27e90f67e35da516dafc1482c82939a60 Mon Sep 17 00:00:00 2001 From: Andi Kleen Date: Thu, 15 Sep 2011 16:06:48 -0700 Subject: vfs: do (nearly) lockless generic_file_llseek The i_mutex lock use of generic _file_llseek hurts. Independent processes accessing the same file synchronize over a single lock, even though they have no need for synchronization at all. Under high utilization this can cause llseek to scale very poorly on larger systems. This patch does some rethinking of the llseek locking model: First the 64bit f_pos is not necessarily atomic without locks on 32bit systems. This can already cause races with read() today. This was discussed on linux-kernel in the past and deemed acceptable. The patch does not change that. Let's look at the different seek variants: SEEK_SET: Doesn't really need any locking. If there's a race one writer wins, the other loses. For 32bit the non atomic update races against read() stay the same. Without a lock they can also happen against write() now. The read() race was deemed acceptable in past discussions, and I think if it's ok for read it's ok for write too. => Don't need a lock. SEEK_END: This behaves like SEEK_SET plus it reads the maximum size too. Reading the maximum size would have the 32bit atomic problem. But luckily we already have a way to read the maximum size without locking (i_size_read), so we can just use that instead. Without i_mutex there is no synchronization with write() anymore, however since the write() update is atomic on 64bit it just behaves like another racy SEEK_SET. On non atomic 32bit it's the same as SEEK_SET. => Don't need a lock, but need to use i_size_read() SEEK_CUR: This has a read-modify-write race window on the same file. One could argue that any application doing unsynchronized seeks on the same file is already broken. But for the sake of not adding a regression here I'm using the file->f_lock to synchronize this. Using this lock is much better than the inode mutex because it doesn't synchronize between processes. => So still need a lock, but can use a f_lock. This patch implements this new scheme in generic_file_llseek. I dropped generic_file_llseek_unlocked and changed all callers. Signed-off-by: Andi Kleen Signed-off-by: Christoph Hellwig --- fs/btrfs/file.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs/btrfs/file.c') diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c index e4e57d59edb7..1266f6e9cdb2 100644 --- a/fs/btrfs/file.c +++ b/fs/btrfs/file.c @@ -1821,7 +1821,7 @@ static loff_t btrfs_file_llseek(struct file *file, loff_t offset, int origin) switch (origin) { case SEEK_END: case SEEK_CUR: - offset = generic_file_llseek_unlocked(file, offset, origin); + offset = generic_file_llseek(file, offset, origin); goto out; case SEEK_DATA: case SEEK_HOLE: -- cgit v1.2.3