linux/fs/ubifs/file.c

1593 lines
45 KiB
C
Raw Normal View History

/*
* This file is part of UBIFS.
*
* Copyright (C) 2006-2008 Nokia Corporation.
*
* This program is free software; you can redistribute it and/or modify it
* under the terms of the GNU General Public License version 2 as published by
* the Free Software Foundation.
*
* This program is distributed in the hope that it will be useful, but WITHOUT
* ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
* FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
* more details.
*
* You should have received a copy of the GNU General Public License along with
* this program; if not, write to the Free Software Foundation, Inc., 51
* Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
*
* Authors: Artem Bityutskiy (Битюцкий Артём)
* Adrian Hunter
*/
/*
* This file implements VFS file and inode operations for regular files, device
* nodes and symlinks as well as address space operations.
*
* UBIFS uses 2 page flags: @PG_private and @PG_checked. @PG_private is set if
* the page is dirty and is used for optimization purposes - dirty pages are
* not budgeted so the flag shows that 'ubifs_write_end()' should not release
* the budget for this page. The @PG_checked flag is set if full budgeting is
* required for the page e.g., when it corresponds to a file hole or it is
* beyond the file size. The budgeting is done in 'ubifs_write_begin()', because
* it is OK to fail in this function, and the budget is released in
* 'ubifs_write_end()'. So the @PG_private and @PG_checked flags carry
* information about how the page was budgeted, to make it possible to release
* the budget properly.
*
* A thing to keep in mind: inode @i_mutex is locked in most VFS operations we
* implement. However, this is not true for 'ubifs_writepage()', which may be
* called with @i_mutex unlocked. For example, when pdflush is doing background
* write-back, it calls 'ubifs_writepage()' with unlocked @i_mutex. At "normal"
* work-paths the @i_mutex is locked in 'ubifs_writepage()', e.g. in the
* "sys_write -> alloc_pages -> direct reclaim path". So, in 'ubifs_writepage()'
* we are only guaranteed that the page is locked.
*
* Similarly, @i_mutex is not always locked in 'ubifs_readpage()', e.g., the
* read-ahead path does not lock it ("sys_read -> generic_file_aio_read ->
* ondemand_readahead -> readpage"). In case of readahead, @I_SYNC flag is not
* set as well. However, UBIFS disables readahead.
*/
#include "ubifs.h"
#include <linux/mount.h>
#include <linux/namei.h>
include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h percpu.h is included by sched.h and module.h and thus ends up being included when building most .c files. percpu.h includes slab.h which in turn includes gfp.h making everything defined by the two files universally available and complicating inclusion dependencies. percpu.h -> slab.h dependency is about to be removed. Prepare for this change by updating users of gfp and slab facilities include those headers directly instead of assuming availability. As this conversion needs to touch large number of source files, the following script is used as the basis of conversion. http://userweb.kernel.org/~tj/misc/slabh-sweep.py The script does the followings. * Scan files for gfp and slab usages and update includes such that only the necessary includes are there. ie. if only gfp is used, gfp.h, if slab is used, slab.h. * When the script inserts a new include, it looks at the include blocks and try to put the new include such that its order conforms to its surrounding. It's put in the include block which contains core kernel includes, in the same order that the rest are ordered - alphabetical, Christmas tree, rev-Xmas-tree or at the end if there doesn't seem to be any matching order. * If the script can't find a place to put a new include (mostly because the file doesn't have fitting include block), it prints out an error message indicating which .h file needs to be added to the file. The conversion was done in the following steps. 1. The initial automatic conversion of all .c files updated slightly over 4000 files, deleting around 700 includes and adding ~480 gfp.h and ~3000 slab.h inclusions. The script emitted errors for ~400 files. 2. Each error was manually checked. Some didn't need the inclusion, some needed manual addition while adding it to implementation .h or embedding .c file was more appropriate for others. This step added inclusions to around 150 files. 3. The script was run again and the output was compared to the edits from #2 to make sure no file was left behind. 4. Several build tests were done and a couple of problems were fixed. e.g. lib/decompress_*.c used malloc/free() wrappers around slab APIs requiring slab.h to be added manually. 5. The script was run on all .h files but without automatically editing them as sprinkling gfp.h and slab.h inclusions around .h files could easily lead to inclusion dependency hell. Most gfp.h inclusion directives were ignored as stuff from gfp.h was usually wildly available and often used in preprocessor macros. Each slab.h inclusion directive was examined and added manually as necessary. 6. percpu.h was updated not to include slab.h. 7. Build test were done on the following configurations and failures were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my distributed build env didn't work with gcov compiles) and a few more options had to be turned off depending on archs to make things build (like ipr on powerpc/64 which failed due to missing writeq). * x86 and x86_64 UP and SMP allmodconfig and a custom test config. * powerpc and powerpc64 SMP allmodconfig * sparc and sparc64 SMP allmodconfig * ia64 SMP allmodconfig * s390 SMP allmodconfig * alpha SMP allmodconfig * um on x86_64 SMP allmodconfig 8. percpu.h modifications were reverted so that it could be applied as a separate patch and serve as bisection point. Given the fact that I had only a couple of failures from tests on step 6, I'm fairly confident about the coverage of this conversion patch. If there is a breakage, it's likely to be something in one of the arch headers which should be easily discoverable easily on most builds of the specific arch. Signed-off-by: Tejun Heo <tj@kernel.org> Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org> Cc: Ingo Molnar <mingo@redhat.com> Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
2010-03-24 08:04:11 +00:00
#include <linux/slab.h>
static int read_block(struct inode *inode, void *addr, unsigned int block,
struct ubifs_data_node *dn)
{
struct ubifs_info *c = inode->i_sb->s_fs_info;
int err, len, out_len;
union ubifs_key key;
unsigned int dlen;
data_key_init(c, &key, inode->i_ino, block);
err = ubifs_tnc_lookup(c, &key, dn);
if (err) {
if (err == -ENOENT)
/* Not found, so it must be a hole */
memset(addr, 0, UBIFS_BLOCK_SIZE);
return err;
}
ubifs_assert(le64_to_cpu(dn->ch.sqnum) >
ubifs_inode(inode)->creat_sqnum);
len = le32_to_cpu(dn->size);
if (len <= 0 || len > UBIFS_BLOCK_SIZE)
goto dump;
dlen = le32_to_cpu(dn->ch.len) - UBIFS_DATA_NODE_SZ;
out_len = UBIFS_BLOCK_SIZE;
err = ubifs_decompress(&dn->data, dlen, addr, &out_len,
le16_to_cpu(dn->compr_type));
if (err || len != out_len)
goto dump;
/*
* Data length can be less than a full block, even for blocks that are
* not the last in the file (e.g., as a result of making a hole and
* appending data). Ensure that the remainder is zeroed out.
*/
if (len < UBIFS_BLOCK_SIZE)
memset(addr + len, 0, UBIFS_BLOCK_SIZE - len);
return 0;
dump:
ubifs_err("bad data node (block %u, inode %lu)",
block, inode->i_ino);
ubifs_dump_node(c, dn);
return -EINVAL;
}
static int do_readpage(struct page *page)
{
void *addr;
int err = 0, i;
unsigned int block, beyond;
struct ubifs_data_node *dn;
struct inode *inode = page->mapping->host;
loff_t i_size = i_size_read(inode);
dbg_gen("ino %lu, pg %lu, i_size %lld, flags %#lx",
inode->i_ino, page->index, i_size, page->flags);
ubifs_assert(!PageChecked(page));
ubifs_assert(!PagePrivate(page));
addr = kmap(page);
block = page->index << UBIFS_BLOCKS_PER_PAGE_SHIFT;
beyond = (i_size + UBIFS_BLOCK_SIZE - 1) >> UBIFS_BLOCK_SHIFT;
if (block >= beyond) {
/* Reading beyond inode */
SetPageChecked(page);
memset(addr, 0, PAGE_CACHE_SIZE);
goto out;
}
dn = kmalloc(UBIFS_MAX_DATA_NODE_SZ, GFP_NOFS);
if (!dn) {
err = -ENOMEM;
goto error;
}
i = 0;
while (1) {
int ret;
if (block >= beyond) {
/* Reading beyond inode */
err = -ENOENT;
memset(addr, 0, UBIFS_BLOCK_SIZE);
} else {
ret = read_block(inode, addr, block, dn);
if (ret) {
err = ret;
if (err != -ENOENT)
break;
} else if (block + 1 == beyond) {
int dlen = le32_to_cpu(dn->size);
int ilen = i_size & (UBIFS_BLOCK_SIZE - 1);
if (ilen && ilen < dlen)
memset(addr + ilen, 0, dlen - ilen);
}
}
if (++i >= UBIFS_BLOCKS_PER_PAGE)
break;
block += 1;
addr += UBIFS_BLOCK_SIZE;
}
if (err) {
if (err == -ENOENT) {
/* Not found, so it must be a hole */
SetPageChecked(page);
dbg_gen("hole");
goto out_free;
}
ubifs_err("cannot read page %lu of inode %lu, error %d",
page->index, inode->i_ino, err);
goto error;
}
out_free:
kfree(dn);
out:
SetPageUptodate(page);
ClearPageError(page);
flush_dcache_page(page);
kunmap(page);
return 0;
error:
kfree(dn);
ClearPageUptodate(page);
SetPageError(page);
flush_dcache_page(page);
kunmap(page);
return err;
}
/**
* release_new_page_budget - release budget of a new page.
* @c: UBIFS file-system description object
*
* This is a helper function which releases budget corresponding to the budget
* of one new page of data.
*/
static void release_new_page_budget(struct ubifs_info *c)
{
struct ubifs_budget_req req = { .recalculate = 1, .new_page = 1 };
ubifs_release_budget(c, &req);
}
/**
* release_existing_page_budget - release budget of an existing page.
* @c: UBIFS file-system description object
*
* This is a helper function which releases budget corresponding to the budget
* of changing one one page of data which already exists on the flash media.
*/
static void release_existing_page_budget(struct ubifs_info *c)
{
struct ubifs_budget_req req = { .dd_growth = c->bi.page_budget};
ubifs_release_budget(c, &req);
}
static int write_begin_slow(struct address_space *mapping,
fs: symlink write_begin allocation context fix With the write_begin/write_end aops, page_symlink was broken because it could no longer pass a GFP_NOFS type mask into the point where the allocations happened. They are done in write_begin, which would always assume that the filesystem can be entered from reclaim. This bug could cause filesystem deadlocks. The funny thing with having a gfp_t mask there is that it doesn't really allow the caller to arbitrarily tinker with the context in which it can be called. It couldn't ever be GFP_ATOMIC, for example, because it needs to take the page lock. The only thing any callers care about is __GFP_FS anyway, so turn that into a single flag. Add a new flag for write_begin, AOP_FLAG_NOFS. Filesystems can now act on this flag in their write_begin function. Change __grab_cache_page to accept a nofs argument as well, to honour that flag (while we're there, change the name to grab_cache_page_write_begin which is more instructive and does away with random leading underscores). This is really a more flexible way to go in the end anyway -- if a filesystem happens to want any extra allocations aside from the pagecache ones in ints write_begin function, it may now use GFP_KERNEL (rather than GFP_NOFS) for common case allocations (eg. ocfs2_alloc_write_ctxt, for a random example). [kosaki.motohiro@jp.fujitsu.com: fix ubifs] [kosaki.motohiro@jp.fujitsu.com: fix fuse] Signed-off-by: Nick Piggin <npiggin@suse.de> Reviewed-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: <stable@kernel.org> [2.6.28.x] Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> [ Cleaned up the calling convention: just pass in the AOP flags untouched to the grab_cache_page_write_begin() function. That just simplifies everybody, and may even allow future expansion of the logic. - Linus ] Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-01-04 20:00:53 +00:00
loff_t pos, unsigned len, struct page **pagep,
unsigned flags)
{
struct inode *inode = mapping->host;
struct ubifs_info *c = inode->i_sb->s_fs_info;
pgoff_t index = pos >> PAGE_CACHE_SHIFT;
struct ubifs_budget_req req = { .new_page = 1 };
int uninitialized_var(err), appending = !!(pos + len > inode->i_size);
struct page *page;
dbg_gen("ino %lu, pos %llu, len %u, i_size %lld",
inode->i_ino, pos, len, inode->i_size);
/*
* At the slow path we have to budget before locking the page, because
* budgeting may force write-back, which would wait on locked pages and
* deadlock if we had the page locked. At this point we do not know
* anything about the page, so assume that this is a new page which is
* written to a hole. This corresponds to largest budget. Later the
* budget will be amended if this is not true.
*/
if (appending)
/* We are appending data, budget for inode change */
req.dirtied_ino = 1;
err = ubifs_budget_space(c, &req);
if (unlikely(err))
return err;
fs: symlink write_begin allocation context fix With the write_begin/write_end aops, page_symlink was broken because it could no longer pass a GFP_NOFS type mask into the point where the allocations happened. They are done in write_begin, which would always assume that the filesystem can be entered from reclaim. This bug could cause filesystem deadlocks. The funny thing with having a gfp_t mask there is that it doesn't really allow the caller to arbitrarily tinker with the context in which it can be called. It couldn't ever be GFP_ATOMIC, for example, because it needs to take the page lock. The only thing any callers care about is __GFP_FS anyway, so turn that into a single flag. Add a new flag for write_begin, AOP_FLAG_NOFS. Filesystems can now act on this flag in their write_begin function. Change __grab_cache_page to accept a nofs argument as well, to honour that flag (while we're there, change the name to grab_cache_page_write_begin which is more instructive and does away with random leading underscores). This is really a more flexible way to go in the end anyway -- if a filesystem happens to want any extra allocations aside from the pagecache ones in ints write_begin function, it may now use GFP_KERNEL (rather than GFP_NOFS) for common case allocations (eg. ocfs2_alloc_write_ctxt, for a random example). [kosaki.motohiro@jp.fujitsu.com: fix ubifs] [kosaki.motohiro@jp.fujitsu.com: fix fuse] Signed-off-by: Nick Piggin <npiggin@suse.de> Reviewed-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: <stable@kernel.org> [2.6.28.x] Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> [ Cleaned up the calling convention: just pass in the AOP flags untouched to the grab_cache_page_write_begin() function. That just simplifies everybody, and may even allow future expansion of the logic. - Linus ] Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-01-04 20:00:53 +00:00
page = grab_cache_page_write_begin(mapping, index, flags);
if (unlikely(!page)) {
ubifs_release_budget(c, &req);
return -ENOMEM;
}
if (!PageUptodate(page)) {
if (!(pos & ~PAGE_CACHE_MASK) && len == PAGE_CACHE_SIZE)
SetPageChecked(page);
else {
err = do_readpage(page);
if (err) {
unlock_page(page);
page_cache_release(page);
return err;
}
}
SetPageUptodate(page);
ClearPageError(page);
}
if (PagePrivate(page))
/*
* The page is dirty, which means it was budgeted twice:
* o first time the budget was allocated by the task which
* made the page dirty and set the PG_private flag;
* o and then we budgeted for it for the second time at the
* very beginning of this function.
*
* So what we have to do is to release the page budget we
* allocated.
*/
release_new_page_budget(c);
else if (!PageChecked(page))
/*
* We are changing a page which already exists on the media.
* This means that changing the page does not make the amount
* of indexing information larger, and this part of the budget
* which we have already acquired may be released.
*/
ubifs_convert_page_budget(c);
if (appending) {
struct ubifs_inode *ui = ubifs_inode(inode);
/*
* 'ubifs_write_end()' is optimized from the fast-path part of
* 'ubifs_write_begin()' and expects the @ui_mutex to be locked
* if data is appended.
*/
mutex_lock(&ui->ui_mutex);
if (ui->dirty)
/*
* The inode is dirty already, so we may free the
* budget we allocated.
*/
ubifs_release_dirty_inode_budget(c, ui);
}
*pagep = page;
return 0;
}
/**
* allocate_budget - allocate budget for 'ubifs_write_begin()'.
* @c: UBIFS file-system description object
* @page: page to allocate budget for
* @ui: UBIFS inode object the page belongs to
* @appending: non-zero if the page is appended
*
* This is a helper function for 'ubifs_write_begin()' which allocates budget
* for the operation. The budget is allocated differently depending on whether
* this is appending, whether the page is dirty or not, and so on. This
* function leaves the @ui->ui_mutex locked in case of appending. Returns zero
* in case of success and %-ENOSPC in case of failure.
*/
static int allocate_budget(struct ubifs_info *c, struct page *page,
struct ubifs_inode *ui, int appending)
{
struct ubifs_budget_req req = { .fast = 1 };
if (PagePrivate(page)) {
if (!appending)
/*
* The page is dirty and we are not appending, which
* means no budget is needed at all.
*/
return 0;
mutex_lock(&ui->ui_mutex);
if (ui->dirty)
/*
* The page is dirty and we are appending, so the inode
* has to be marked as dirty. However, it is already
* dirty, so we do not need any budget. We may return,
* but @ui->ui_mutex hast to be left locked because we
* should prevent write-back from flushing the inode
* and freeing the budget. The lock will be released in
* 'ubifs_write_end()'.
*/
return 0;
/*
* The page is dirty, we are appending, the inode is clean, so
* we need to budget the inode change.
*/
req.dirtied_ino = 1;
} else {
if (PageChecked(page))
/*
* The page corresponds to a hole and does not
* exist on the media. So changing it makes
* make the amount of indexing information
* larger, and we have to budget for a new
* page.
*/
req.new_page = 1;
else
/*
* Not a hole, the change will not add any new
* indexing information, budget for page
* change.
*/
req.dirtied_page = 1;
if (appending) {
mutex_lock(&ui->ui_mutex);
if (!ui->dirty)
/*
* The inode is clean but we will have to mark
* it as dirty because we are appending. This
* needs a budget.
*/
req.dirtied_ino = 1;
}
}
return ubifs_budget_space(c, &req);
}
/*
* This function is called when a page of data is going to be written. Since
* the page of data will not necessarily go to the flash straight away, UBIFS
* has to reserve space on the media for it, which is done by means of
* budgeting.
*
* This is the hot-path of the file-system and we are trying to optimize it as
* much as possible. For this reasons it is split on 2 parts - slow and fast.
*
* There many budgeting cases:
* o a new page is appended - we have to budget for a new page and for
* changing the inode; however, if the inode is already dirty, there is
* no need to budget for it;
* o an existing clean page is changed - we have budget for it; if the page
* does not exist on the media (a hole), we have to budget for a new
* page; otherwise, we may budget for changing an existing page; the
* difference between these cases is that changing an existing page does
* not introduce anything new to the FS indexing information, so it does
* not grow, and smaller budget is acquired in this case;
* o an existing dirty page is changed - no need to budget at all, because
* the page budget has been acquired by earlier, when the page has been
* marked dirty.
*
* UBIFS budgeting sub-system may force write-back if it thinks there is no
* space to reserve. This imposes some locking restrictions and makes it
* impossible to take into account the above cases, and makes it impossible to
* optimize budgeting.
*
* The solution for this is that the fast path of 'ubifs_write_begin()' assumes
* there is a plenty of flash space and the budget will be acquired quickly,
* without forcing write-back. The slow path does not make this assumption.
*/
static int ubifs_write_begin(struct file *file, struct address_space *mapping,
loff_t pos, unsigned len, unsigned flags,
struct page **pagep, void **fsdata)
{
struct inode *inode = mapping->host;
struct ubifs_info *c = inode->i_sb->s_fs_info;
struct ubifs_inode *ui = ubifs_inode(inode);
pgoff_t index = pos >> PAGE_CACHE_SHIFT;
int uninitialized_var(err), appending = !!(pos + len > inode->i_size);
int skipped_read = 0;
struct page *page;
ubifs_assert(ubifs_inode(inode)->ui_size == inode->i_size);
ubifs_assert(!c->ro_media && !c->ro_mount);
if (unlikely(c->ro_error))
return -EROFS;
/* Try out the fast-path part first */
fs: symlink write_begin allocation context fix With the write_begin/write_end aops, page_symlink was broken because it could no longer pass a GFP_NOFS type mask into the point where the allocations happened. They are done in write_begin, which would always assume that the filesystem can be entered from reclaim. This bug could cause filesystem deadlocks. The funny thing with having a gfp_t mask there is that it doesn't really allow the caller to arbitrarily tinker with the context in which it can be called. It couldn't ever be GFP_ATOMIC, for example, because it needs to take the page lock. The only thing any callers care about is __GFP_FS anyway, so turn that into a single flag. Add a new flag for write_begin, AOP_FLAG_NOFS. Filesystems can now act on this flag in their write_begin function. Change __grab_cache_page to accept a nofs argument as well, to honour that flag (while we're there, change the name to grab_cache_page_write_begin which is more instructive and does away with random leading underscores). This is really a more flexible way to go in the end anyway -- if a filesystem happens to want any extra allocations aside from the pagecache ones in ints write_begin function, it may now use GFP_KERNEL (rather than GFP_NOFS) for common case allocations (eg. ocfs2_alloc_write_ctxt, for a random example). [kosaki.motohiro@jp.fujitsu.com: fix ubifs] [kosaki.motohiro@jp.fujitsu.com: fix fuse] Signed-off-by: Nick Piggin <npiggin@suse.de> Reviewed-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: <stable@kernel.org> [2.6.28.x] Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> [ Cleaned up the calling convention: just pass in the AOP flags untouched to the grab_cache_page_write_begin() function. That just simplifies everybody, and may even allow future expansion of the logic. - Linus ] Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-01-04 20:00:53 +00:00
page = grab_cache_page_write_begin(mapping, index, flags);
if (unlikely(!page))
return -ENOMEM;
if (!PageUptodate(page)) {
/* The page is not loaded from the flash */
if (!(pos & ~PAGE_CACHE_MASK) && len == PAGE_CACHE_SIZE) {
/*
* We change whole page so no need to load it. But we
UBIFS: fix assertion warning and refine comments This patch fixes the following UBIFS assertion warning: UBIFS assert failed in do_readpage at 115 (pid 199) [<b00321b8>] (unwind_backtrace+0x0/0xdc) from [<af025118>] (do_readpage+0x108/0x594 [ubifs]) [<af025118>] (do_readpage+0x108/0x594 [ubifs]) from [<af025764>] (ubifs_write_end+0x1c0/0x2e8 [ubifs]) [<af025764>] (ubifs_write_end+0x1c0/0x2e8 [ubifs]) from [<b00a0164>] (generic_file_buffered_write+0x18c/0x270) [<b00a0164>] (generic_file_buffered_write+0x18c/0x270) from [<b00a08d4>] (__generic_file_aio_write+0x478/0x4c0) [<b00a08d4>] (__generic_file_aio_write+0x478/0x4c0) from [<b00a0984>] (generic_file_aio_write+0x68/0xc8) [<b00a0984>] (generic_file_aio_write+0x68/0xc8) from [<af024a78>] (ubifs_aio_write+0x178/0x1d8 [ubifs]) [<af024a78>] (ubifs_aio_write+0x178/0x1d8 [ubifs]) from [<b00d104c>] (do_sync_write+0xb0/0x100) [<b00d104c>] (do_sync_write+0xb0/0x100) from [<b00d1abc>] (vfs_write+0xac/0x154) [<b00d1abc>] (vfs_write+0xac/0x154) from [<b00d1c10>] (sys_write+0x3c/0x68) [<b00d1c10>] (sys_write+0x3c/0x68) from [<b002d9a0>] (ret_fast_syscall+0x0/0x2c) The 'PG_checked' flag is used to indicate that the page does not supposedly exist on the media (e.g., a hole or a page beyond the inode size), so it requires slightly bigger budget, because we have to account the indexing size increase. And this flag basically tells that the budget for this page has to be "new page budget". The "new page budget" is slightly bigger than the "existing page budget". The 'do_readpage()' function has the following assertion which sometimes is hit: 'ubifs_assert(!PageChecked(page))'. Obviously, the meaning of this assertion is: "I should not be asked to read a page which does not exist on the media". However, in 'ubifs_write_begin()' we have a small "trick". Notice, that VFS may write pages which were not read yet, so the page data were not loaded from the media to the page cache yet. If VFS tells that it is going to change only some part of the page, we obviously have to load it from the media. However, if VFS tells that it is going to change whole page, we do not read it from the media for optimization purposes. However, since we do not read it, we do not know if it exists on the media or not (a hole, etc). So we set the 'PG_checked' flag to this page to force bigger budget, just in case. So 'ubifs_write_begin()' sets 'PG_checked'. Then we are in 'ubifs_write_end()'. And VFS tells us: "hey, for some reasons I changed my mind and did not change whole page". Frankly, I do not know why this happens, but I hit this somehow on an ARM platform. And this is extremely rare. So in this case UBIFS does the following: 1. Cancels allocated budget. 2. Loads the page from the media by calling 'do_readpage()'. 3. Asks VFS to repeat the whole write operation from the very beginning (call '->write_begin() again, etc). And the assertion warning is hit at the step 2 - remember we have the 'PG_checked' set for this page, and 'do_readpage()' does not like this. So this patch fixes the problem by adding step 1.5 and cleaning the 'PG_checked' before calling 'do_readpage()'. All in all, this patch does not fix any functionality issue, but it silences UBIFS false positive warning which may happen in very very rare cases. And while on it, this patch also improves a commentary which explains the reasons of setting the 'PG_checked' flag for the page. The old commentary was a bit difficult to understand. Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>
2011-03-23 08:32:58 +00:00
* do not know whether this page exists on the media or
* not, so we assume the latter because it requires
* larger budget. The assumption is that it is better
* to budget a bit more than to read the page from the
* media. Thus, we are setting the @PG_checked flag
* here.
*/
SetPageChecked(page);
skipped_read = 1;
} else {
err = do_readpage(page);
if (err) {
unlock_page(page);
page_cache_release(page);
return err;
}
}
SetPageUptodate(page);
ClearPageError(page);
}
err = allocate_budget(c, page, ui, appending);
if (unlikely(err)) {
ubifs_assert(err == -ENOSPC);
/*
* If we skipped reading the page because we were going to
* write all of it, then it is not up to date.
*/
if (skipped_read) {
ClearPageChecked(page);
ClearPageUptodate(page);
}
/*
* Budgeting failed which means it would have to force
* write-back but didn't, because we set the @fast flag in the
* request. Write-back cannot be done now, while we have the
* page locked, because it would deadlock. Unlock and free
* everything and fall-back to slow-path.
*/
if (appending) {
ubifs_assert(mutex_is_locked(&ui->ui_mutex));
mutex_unlock(&ui->ui_mutex);
}
unlock_page(page);
page_cache_release(page);
fs: symlink write_begin allocation context fix With the write_begin/write_end aops, page_symlink was broken because it could no longer pass a GFP_NOFS type mask into the point where the allocations happened. They are done in write_begin, which would always assume that the filesystem can be entered from reclaim. This bug could cause filesystem deadlocks. The funny thing with having a gfp_t mask there is that it doesn't really allow the caller to arbitrarily tinker with the context in which it can be called. It couldn't ever be GFP_ATOMIC, for example, because it needs to take the page lock. The only thing any callers care about is __GFP_FS anyway, so turn that into a single flag. Add a new flag for write_begin, AOP_FLAG_NOFS. Filesystems can now act on this flag in their write_begin function. Change __grab_cache_page to accept a nofs argument as well, to honour that flag (while we're there, change the name to grab_cache_page_write_begin which is more instructive and does away with random leading underscores). This is really a more flexible way to go in the end anyway -- if a filesystem happens to want any extra allocations aside from the pagecache ones in ints write_begin function, it may now use GFP_KERNEL (rather than GFP_NOFS) for common case allocations (eg. ocfs2_alloc_write_ctxt, for a random example). [kosaki.motohiro@jp.fujitsu.com: fix ubifs] [kosaki.motohiro@jp.fujitsu.com: fix fuse] Signed-off-by: Nick Piggin <npiggin@suse.de> Reviewed-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: <stable@kernel.org> [2.6.28.x] Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> [ Cleaned up the calling convention: just pass in the AOP flags untouched to the grab_cache_page_write_begin() function. That just simplifies everybody, and may even allow future expansion of the logic. - Linus ] Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-01-04 20:00:53 +00:00
return write_begin_slow(mapping, pos, len, pagep, flags);
}
/*
* Whee, we acquired budgeting quickly - without involving
* garbage-collection, committing or forcing write-back. We return
* with @ui->ui_mutex locked if we are appending pages, and unlocked
* otherwise. This is an optimization (slightly hacky though).
*/
*pagep = page;
return 0;
}
/**
* cancel_budget - cancel budget.
* @c: UBIFS file-system description object
* @page: page to cancel budget for
* @ui: UBIFS inode object the page belongs to
* @appending: non-zero if the page is appended
*
* This is a helper function for a page write operation. It unlocks the
* @ui->ui_mutex in case of appending.
*/
static void cancel_budget(struct ubifs_info *c, struct page *page,
struct ubifs_inode *ui, int appending)
{
if (appending) {
if (!ui->dirty)
ubifs_release_dirty_inode_budget(c, ui);
mutex_unlock(&ui->ui_mutex);
}
if (!PagePrivate(page)) {
if (PageChecked(page))
release_new_page_budget(c);
else
release_existing_page_budget(c);
}
}
static int ubifs_write_end(struct file *file, struct address_space *mapping,
loff_t pos, unsigned len, unsigned copied,
struct page *page, void *fsdata)
{
struct inode *inode = mapping->host;
struct ubifs_inode *ui = ubifs_inode(inode);
struct ubifs_info *c = inode->i_sb->s_fs_info;
loff_t end_pos = pos + len;
int appending = !!(end_pos > inode->i_size);
dbg_gen("ino %lu, pos %llu, pg %lu, len %u, copied %d, i_size %lld",
inode->i_ino, pos, page->index, len, copied, inode->i_size);
if (unlikely(copied < len && len == PAGE_CACHE_SIZE)) {
/*
* VFS copied less data to the page that it intended and
* declared in its '->write_begin()' call via the @len
* argument. If the page was not up-to-date, and @len was
* @PAGE_CACHE_SIZE, the 'ubifs_write_begin()' function did
* not load it from the media (for optimization reasons). This
* means that part of the page contains garbage. So read the
* page now.
*/
dbg_gen("copied %d instead of %d, read page and repeat",
copied, len);
cancel_budget(c, page, ui, appending);
UBIFS: fix assertion warning and refine comments This patch fixes the following UBIFS assertion warning: UBIFS assert failed in do_readpage at 115 (pid 199) [<b00321b8>] (unwind_backtrace+0x0/0xdc) from [<af025118>] (do_readpage+0x108/0x594 [ubifs]) [<af025118>] (do_readpage+0x108/0x594 [ubifs]) from [<af025764>] (ubifs_write_end+0x1c0/0x2e8 [ubifs]) [<af025764>] (ubifs_write_end+0x1c0/0x2e8 [ubifs]) from [<b00a0164>] (generic_file_buffered_write+0x18c/0x270) [<b00a0164>] (generic_file_buffered_write+0x18c/0x270) from [<b00a08d4>] (__generic_file_aio_write+0x478/0x4c0) [<b00a08d4>] (__generic_file_aio_write+0x478/0x4c0) from [<b00a0984>] (generic_file_aio_write+0x68/0xc8) [<b00a0984>] (generic_file_aio_write+0x68/0xc8) from [<af024a78>] (ubifs_aio_write+0x178/0x1d8 [ubifs]) [<af024a78>] (ubifs_aio_write+0x178/0x1d8 [ubifs]) from [<b00d104c>] (do_sync_write+0xb0/0x100) [<b00d104c>] (do_sync_write+0xb0/0x100) from [<b00d1abc>] (vfs_write+0xac/0x154) [<b00d1abc>] (vfs_write+0xac/0x154) from [<b00d1c10>] (sys_write+0x3c/0x68) [<b00d1c10>] (sys_write+0x3c/0x68) from [<b002d9a0>] (ret_fast_syscall+0x0/0x2c) The 'PG_checked' flag is used to indicate that the page does not supposedly exist on the media (e.g., a hole or a page beyond the inode size), so it requires slightly bigger budget, because we have to account the indexing size increase. And this flag basically tells that the budget for this page has to be "new page budget". The "new page budget" is slightly bigger than the "existing page budget". The 'do_readpage()' function has the following assertion which sometimes is hit: 'ubifs_assert(!PageChecked(page))'. Obviously, the meaning of this assertion is: "I should not be asked to read a page which does not exist on the media". However, in 'ubifs_write_begin()' we have a small "trick". Notice, that VFS may write pages which were not read yet, so the page data were not loaded from the media to the page cache yet. If VFS tells that it is going to change only some part of the page, we obviously have to load it from the media. However, if VFS tells that it is going to change whole page, we do not read it from the media for optimization purposes. However, since we do not read it, we do not know if it exists on the media or not (a hole, etc). So we set the 'PG_checked' flag to this page to force bigger budget, just in case. So 'ubifs_write_begin()' sets 'PG_checked'. Then we are in 'ubifs_write_end()'. And VFS tells us: "hey, for some reasons I changed my mind and did not change whole page". Frankly, I do not know why this happens, but I hit this somehow on an ARM platform. And this is extremely rare. So in this case UBIFS does the following: 1. Cancels allocated budget. 2. Loads the page from the media by calling 'do_readpage()'. 3. Asks VFS to repeat the whole write operation from the very beginning (call '->write_begin() again, etc). And the assertion warning is hit at the step 2 - remember we have the 'PG_checked' set for this page, and 'do_readpage()' does not like this. So this patch fixes the problem by adding step 1.5 and cleaning the 'PG_checked' before calling 'do_readpage()'. All in all, this patch does not fix any functionality issue, but it silences UBIFS false positive warning which may happen in very very rare cases. And while on it, this patch also improves a commentary which explains the reasons of setting the 'PG_checked' flag for the page. The old commentary was a bit difficult to understand. Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>
2011-03-23 08:32:58 +00:00
ClearPageChecked(page);
/*
* Return 0 to force VFS to repeat the whole operation, or the
* error code if 'do_readpage()' fails.
*/
copied = do_readpage(page);
goto out;
}
if (!PagePrivate(page)) {
SetPagePrivate(page);
atomic_long_inc(&c->dirty_pg_cnt);
__set_page_dirty_nobuffers(page);
}
if (appending) {
i_size_write(inode, end_pos);
ui->ui_size = end_pos;
/*
* Note, we do not set @I_DIRTY_PAGES (which means that the
* inode has dirty pages), this has been done in
* '__set_page_dirty_nobuffers()'.
*/
__mark_inode_dirty(inode, I_DIRTY_DATASYNC);
ubifs_assert(mutex_is_locked(&ui->ui_mutex));
mutex_unlock(&ui->ui_mutex);
}
out:
unlock_page(page);
page_cache_release(page);
return copied;
}
/**
* populate_page - copy data nodes into a page for bulk-read.
* @c: UBIFS file-system description object
* @page: page
* @bu: bulk-read information
* @n: next zbranch slot
*
* This function returns %0 on success and a negative error code on failure.
*/
static int populate_page(struct ubifs_info *c, struct page *page,
struct bu_info *bu, int *n)
{
int i = 0, nn = *n, offs = bu->zbranch[0].offs, hole = 0, read = 0;
struct inode *inode = page->mapping->host;
loff_t i_size = i_size_read(inode);
unsigned int page_block;
void *addr, *zaddr;
pgoff_t end_index;
dbg_gen("ino %lu, pg %lu, i_size %lld, flags %#lx",
inode->i_ino, page->index, i_size, page->flags);
addr = zaddr = kmap(page);
end_index = (i_size - 1) >> PAGE_CACHE_SHIFT;
if (!i_size || page->index > end_index) {
hole = 1;
memset(addr, 0, PAGE_CACHE_SIZE);
goto out_hole;
}
page_block = page->index << UBIFS_BLOCKS_PER_PAGE_SHIFT;
while (1) {
int err, len, out_len, dlen;
if (nn >= bu->cnt) {
hole = 1;
memset(addr, 0, UBIFS_BLOCK_SIZE);
} else if (key_block(c, &bu->zbranch[nn].key) == page_block) {
struct ubifs_data_node *dn;
dn = bu->buf + (bu->zbranch[nn].offs - offs);
UBIFS: endian handling fixes and annotations Noticed by sparse: fs/ubifs/file.c:75:2: warning: restricted __le64 degrades to integer fs/ubifs/file.c:629:4: warning: restricted __le64 degrades to integer fs/ubifs/dir.c:431:3: warning: restricted __le64 degrades to integer This should be checked to ensure the ubifs_assert is working as intended, I've done the suggested annotation in this patch. fs/ubifs/sb.c:298:6: warning: incorrect type in assignment (different base types) fs/ubifs/sb.c:298:6: expected int [signed] [assigned] tmp fs/ubifs/sb.c:298:6: got restricted __le64 [usertype] <noident> fs/ubifs/sb.c:299:19: warning: incorrect type in assignment (different base types) fs/ubifs/sb.c:299:19: expected restricted __le64 [usertype] atime_sec fs/ubifs/sb.c:299:19: got int [signed] [assigned] tmp fs/ubifs/sb.c:300:19: warning: incorrect type in assignment (different base types) fs/ubifs/sb.c:300:19: expected restricted __le64 [usertype] ctime_sec fs/ubifs/sb.c:300:19: got int [signed] [assigned] tmp fs/ubifs/sb.c:301:19: warning: incorrect type in assignment (different base types) fs/ubifs/sb.c:301:19: expected restricted __le64 [usertype] mtime_sec fs/ubifs/sb.c:301:19: got int [signed] [assigned] tmp This looks like a bugfix as your tmp was a u32 so there was truncation in the atime, mtime, ctime value, probably not intentional, add a tmp_le64 and use it here. fs/ubifs/key.h:348:9: warning: cast to restricted __le32 fs/ubifs/key.h:348:9: warning: cast to restricted __le32 fs/ubifs/key.h:419:9: warning: cast to restricted __le32 Read from the annotated union member instead. fs/ubifs/recovery.c:175:13: warning: incorrect type in assignment (different base types) fs/ubifs/recovery.c:175:13: expected unsigned int [unsigned] [usertype] save_flags fs/ubifs/recovery.c:175:13: got restricted __le32 [usertype] flags fs/ubifs/recovery.c:186:13: warning: incorrect type in assignment (different base types) fs/ubifs/recovery.c:186:13: expected restricted __le32 [usertype] flags fs/ubifs/recovery.c:186:13: got unsigned int [unsigned] [usertype] save_flags Do byteshifting at compile time of the flag value. Annotate the saved_flags as le32. fs/ubifs/debug.c:368:10: warning: cast to restricted __le32 fs/ubifs/debug.c:368:10: warning: cast from restricted __le64 Should be checked if the truncation was intentional, I've changed the printk to print the full width. Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com> Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>
2008-10-24 17:52:57 +00:00
ubifs_assert(le64_to_cpu(dn->ch.sqnum) >
ubifs_inode(inode)->creat_sqnum);
len = le32_to_cpu(dn->size);
if (len <= 0 || len > UBIFS_BLOCK_SIZE)
goto out_err;
dlen = le32_to_cpu(dn->ch.len) - UBIFS_DATA_NODE_SZ;
out_len = UBIFS_BLOCK_SIZE;
err = ubifs_decompress(&dn->data, dlen, addr, &out_len,
le16_to_cpu(dn->compr_type));
if (err || len != out_len)
goto out_err;
if (len < UBIFS_BLOCK_SIZE)
memset(addr + len, 0, UBIFS_BLOCK_SIZE - len);
nn += 1;
read = (i << UBIFS_BLOCK_SHIFT) + len;
} else if (key_block(c, &bu->zbranch[nn].key) < page_block) {
nn += 1;
continue;
} else {
hole = 1;
memset(addr, 0, UBIFS_BLOCK_SIZE);
}
if (++i >= UBIFS_BLOCKS_PER_PAGE)
break;
addr += UBIFS_BLOCK_SIZE;
page_block += 1;
}
if (end_index == page->index) {
int len = i_size & (PAGE_CACHE_SIZE - 1);
if (len && len < read)
memset(zaddr + len, 0, read - len);
}
out_hole:
if (hole) {
SetPageChecked(page);
dbg_gen("hole");
}
SetPageUptodate(page);
ClearPageError(page);
flush_dcache_page(page);
kunmap(page);
*n = nn;
return 0;
out_err:
ClearPageUptodate(page);
SetPageError(page);
flush_dcache_page(page);
kunmap(page);
ubifs_err("bad data node (block %u, inode %lu)",
page_block, inode->i_ino);
return -EINVAL;
}
/**
* ubifs_do_bulk_read - do bulk-read.
* @c: UBIFS file-system description object
* @bu: bulk-read information
* @page1: first page to read
*
* This function returns %1 if the bulk-read is done, otherwise %0 is returned.
*/
static int ubifs_do_bulk_read(struct ubifs_info *c, struct bu_info *bu,
struct page *page1)
{
pgoff_t offset = page1->index, end_index;
struct address_space *mapping = page1->mapping;
struct inode *inode = mapping->host;
struct ubifs_inode *ui = ubifs_inode(inode);
int err, page_idx, page_cnt, ret = 0, n = 0;
int allocate = bu->buf ? 0 : 1;
loff_t isize;
err = ubifs_tnc_get_bu_keys(c, bu);
if (err)
goto out_warn;
if (bu->eof) {
/* Turn off bulk-read at the end of the file */
ui->read_in_a_row = 1;
ui->bulk_read = 0;
}
page_cnt = bu->blk_cnt >> UBIFS_BLOCKS_PER_PAGE_SHIFT;
if (!page_cnt) {
/*
* This happens when there are multiple blocks per page and the
* blocks for the first page we are looking for, are not
* together. If all the pages were like this, bulk-read would
* reduce performance, so we turn it off for a while.
*/
goto out_bu_off;
}
if (bu->cnt) {
if (allocate) {
/*
* Allocate bulk-read buffer depending on how many data
* nodes we are going to read.
*/
bu->buf_len = bu->zbranch[bu->cnt - 1].offs +
bu->zbranch[bu->cnt - 1].len -
bu->zbranch[0].offs;
ubifs_assert(bu->buf_len > 0);
ubifs_assert(bu->buf_len <= c->leb_size);
bu->buf = kmalloc(bu->buf_len, GFP_NOFS | __GFP_NOWARN);
if (!bu->buf)
goto out_bu_off;
}
err = ubifs_tnc_bulk_read(c, bu);
if (err)
goto out_warn;
}
err = populate_page(c, page1, bu, &n);
if (err)
goto out_warn;
unlock_page(page1);
ret = 1;
isize = i_size_read(inode);
if (isize == 0)
goto out_free;
end_index = ((isize - 1) >> PAGE_CACHE_SHIFT);
for (page_idx = 1; page_idx < page_cnt; page_idx++) {
pgoff_t page_offset = offset + page_idx;
struct page *page;
if (page_offset > end_index)
break;
page = find_or_create_page(mapping, page_offset,
GFP_NOFS | __GFP_COLD);
if (!page)
break;
if (!PageUptodate(page))
err = populate_page(c, page, bu, &n);
unlock_page(page);
page_cache_release(page);
if (err)
break;
}
ui->last_page_read = offset + page_idx - 1;
out_free:
if (allocate)
kfree(bu->buf);
return ret;
out_warn:
ubifs_warn("ignoring error %d and skipping bulk-read", err);
goto out_free;
out_bu_off:
ui->read_in_a_row = ui->bulk_read = 0;
goto out_free;
}
/**
* ubifs_bulk_read - determine whether to bulk-read and, if so, do it.
* @page: page from which to start bulk-read.
*
* Some flash media are capable of reading sequentially at faster rates. UBIFS
* bulk-read facility is designed to take advantage of that, by reading in one
* go consecutive data nodes that are also located consecutively in the same
* LEB. This function returns %1 if a bulk-read is done and %0 otherwise.
*/
static int ubifs_bulk_read(struct page *page)
{
struct inode *inode = page->mapping->host;
struct ubifs_info *c = inode->i_sb->s_fs_info;
struct ubifs_inode *ui = ubifs_inode(inode);
pgoff_t index = page->index, last_page_read = ui->last_page_read;
struct bu_info *bu;
int err = 0, allocated = 0;
ui->last_page_read = index;
if (!c->bulk_read)
return 0;
/*
* Bulk-read is protected by @ui->ui_mutex, but it is an optimization,
* so don't bother if we cannot lock the mutex.
*/
if (!mutex_trylock(&ui->ui_mutex))
return 0;
if (index != last_page_read + 1) {
/* Turn off bulk-read if we stop reading sequentially */
ui->read_in_a_row = 1;
if (ui->bulk_read)
ui->bulk_read = 0;
goto out_unlock;
}
if (!ui->bulk_read) {
ui->read_in_a_row += 1;
if (ui->read_in_a_row < 3)
goto out_unlock;
/* Three reads in a row, so switch on bulk-read */
ui->bulk_read = 1;
}
/*
* If possible, try to use pre-allocated bulk-read information, which
* is protected by @c->bu_mutex.
*/
if (mutex_trylock(&c->bu_mutex))
bu = &c->bu;
else {
bu = kmalloc(sizeof(struct bu_info), GFP_NOFS | __GFP_NOWARN);
if (!bu)
goto out_unlock;
bu->buf = NULL;
allocated = 1;
}
bu->buf_len = c->max_bu_buf_len;
data_key_init(c, &bu->key, inode->i_ino,
page->index << UBIFS_BLOCKS_PER_PAGE_SHIFT);
err = ubifs_do_bulk_read(c, bu, page);
if (!allocated)
mutex_unlock(&c->bu_mutex);
else
kfree(bu);
out_unlock:
mutex_unlock(&ui->ui_mutex);
return err;
}
static int ubifs_readpage(struct file *file, struct page *page)
{
if (ubifs_bulk_read(page))
return 0;
do_readpage(page);
unlock_page(page);
return 0;
}
static int do_writepage(struct page *page, int len)
{
int err = 0, i, blen;
unsigned int block;
void *addr;
union ubifs_key key;
struct inode *inode = page->mapping->host;
struct ubifs_info *c = inode->i_sb->s_fs_info;
#ifdef UBIFS_DEBUG
spin_lock(&ui->ui_lock);
ubifs_assert(page->index <= ui->synced_i_size << PAGE_CACHE_SIZE);
spin_unlock(&ui->ui_lock);
#endif
/* Update radix tree tags */
set_page_writeback(page);
addr = kmap(page);
block = page->index << UBIFS_BLOCKS_PER_PAGE_SHIFT;
i = 0;
while (len) {
blen = min_t(int, len, UBIFS_BLOCK_SIZE);
data_key_init(c, &key, inode->i_ino, block);
err = ubifs_jnl_write_data(c, inode, &key, addr, blen);
if (err)
break;
if (++i >= UBIFS_BLOCKS_PER_PAGE)
break;
block += 1;
addr += blen;
len -= blen;
}
if (err) {
SetPageError(page);
ubifs_err("cannot write page %lu of inode %lu, error %d",
page->index, inode->i_ino, err);
ubifs_ro_mode(c, err);
}
ubifs_assert(PagePrivate(page));
if (PageChecked(page))
release_new_page_budget(c);
else
release_existing_page_budget(c);
atomic_long_dec(&c->dirty_pg_cnt);
ClearPagePrivate(page);
ClearPageChecked(page);
kunmap(page);
unlock_page(page);
end_page_writeback(page);
return err;
}
/*
* When writing-back dirty inodes, VFS first writes-back pages belonging to the
* inode, then the inode itself. For UBIFS this may cause a problem. Consider a
* situation when a we have an inode with size 0, then a megabyte of data is
* appended to the inode, then write-back starts and flushes some amount of the
* dirty pages, the journal becomes full, commit happens and finishes, and then
* an unclean reboot happens. When the file system is mounted next time, the
* inode size would still be 0, but there would be many pages which are beyond
* the inode size, they would be indexed and consume flash space. Because the
* journal has been committed, the replay would not be able to detect this
* situation and correct the inode size. This means UBIFS would have to scan
* whole index and correct all inode sizes, which is long an unacceptable.
*
* To prevent situations like this, UBIFS writes pages back only if they are
* within the last synchronized inode size, i.e. the size which has been
* written to the flash media last time. Otherwise, UBIFS forces inode
* write-back, thus making sure the on-flash inode contains current inode size,
* and then keeps writing pages back.
*
* Some locking issues explanation. 'ubifs_writepage()' first is called with
* the page locked, and it locks @ui_mutex. However, write-back does take inode
* @i_mutex, which means other VFS operations may be run on this inode at the
* same time. And the problematic one is truncation to smaller size, from where
* we have to call 'truncate_setsize()', which first changes @inode->i_size,
* then drops the truncated pages. And while dropping the pages, it takes the
* page lock. This means that 'do_truncation()' cannot call 'truncate_setsize()'
* with @ui_mutex locked, because it would deadlock with 'ubifs_writepage()'.
* This means that @inode->i_size is changed while @ui_mutex is unlocked.
*
* XXX(truncate): with the new truncate sequence this is not true anymore,
* and the calls to truncate_setsize can be move around freely. They should
* be moved to the very end of the truncate sequence.
*
* But in 'ubifs_writepage()' we have to guarantee that we do not write beyond
* inode size. How do we do this if @inode->i_size may became smaller while we
* are in the middle of 'ubifs_writepage()'? The UBIFS solution is the
* @ui->ui_isize "shadow" field which UBIFS uses instead of @inode->i_size
* internally and updates it under @ui_mutex.
*
* Q: why we do not worry that if we race with truncation, we may end up with a
* situation when the inode is truncated while we are in the middle of
* 'do_writepage()', so we do write beyond inode size?
* A: If we are in the middle of 'do_writepage()', truncation would be locked
* on the page lock and it would not write the truncated inode node to the
* journal before we have finished.
*/
static int ubifs_writepage(struct page *page, struct writeback_control *wbc)
{
struct inode *inode = page->mapping->host;
struct ubifs_inode *ui = ubifs_inode(inode);
loff_t i_size = i_size_read(inode), synced_i_size;
pgoff_t end_index = i_size >> PAGE_CACHE_SHIFT;
int err, len = i_size & (PAGE_CACHE_SIZE - 1);
void *kaddr;
dbg_gen("ino %lu, pg %lu, pg flags %#lx",
inode->i_ino, page->index, page->flags);
ubifs_assert(PagePrivate(page));
/* Is the page fully outside @i_size? (truncate in progress) */
if (page->index > end_index || (page->index == end_index && !len)) {
err = 0;
goto out_unlock;
}
spin_lock(&ui->ui_lock);
synced_i_size = ui->synced_i_size;
spin_unlock(&ui->ui_lock);
/* Is the page fully inside @i_size? */
if (page->index < end_index) {
if (page->index >= synced_i_size >> PAGE_CACHE_SHIFT) {
err = inode->i_sb->s_op->write_inode(inode, NULL);
if (err)
goto out_unlock;
/*
* The inode has been written, but the write-buffer has
* not been synchronized, so in case of an unclean
* reboot we may end up with some pages beyond inode
* size, but they would be in the journal (because
* commit flushes write buffers) and recovery would deal
* with this.
*/
}
return do_writepage(page, PAGE_CACHE_SIZE);
}
/*
* The page straddles @i_size. It must be zeroed out on each and every
* writepage invocation because it may be mmapped. "A file is mapped
* in multiples of the page size. For a file that is not a multiple of
* the page size, the remaining memory is zeroed when mapped, and
* writes to that region are not written out to the file."
*/
kaddr = kmap_atomic(page);
memset(kaddr + len, 0, PAGE_CACHE_SIZE - len);
flush_dcache_page(page);
kunmap_atomic(kaddr);
if (i_size > synced_i_size) {
err = inode->i_sb->s_op->write_inode(inode, NULL);
if (err)
goto out_unlock;
}
return do_writepage(page, len);
out_unlock:
unlock_page(page);
return err;
}
/**
* do_attr_changes - change inode attributes.
* @inode: inode to change attributes for
* @attr: describes attributes to change
*/
static void do_attr_changes(struct inode *inode, const struct iattr *attr)
{
if (attr->ia_valid & ATTR_UID)
inode->i_uid = attr->ia_uid;
if (attr->ia_valid & ATTR_GID)
inode->i_gid = attr->ia_gid;
if (attr->ia_valid & ATTR_ATIME)
inode->i_atime = timespec_trunc(attr->ia_atime,
inode->i_sb->s_time_gran);
if (attr->ia_valid & ATTR_MTIME)
inode->i_mtime = timespec_trunc(attr->ia_mtime,
inode->i_sb->s_time_gran);
if (attr->ia_valid & ATTR_CTIME)
inode->i_ctime = timespec_trunc(attr->ia_ctime,
inode->i_sb->s_time_gran);
if (attr->ia_valid & ATTR_MODE) {
umode_t mode = attr->ia_mode;
if (!in_group_p(inode->i_gid) && !capable(CAP_FSETID))
mode &= ~S_ISGID;
inode->i_mode = mode;
}
}
/**
* do_truncation - truncate an inode.
* @c: UBIFS file-system description object
* @inode: inode to truncate
* @attr: inode attribute changes description
*
* This function implements VFS '->setattr()' call when the inode is truncated
* to a smaller size. Returns zero in case of success and a negative error code
* in case of failure.
*/
static int do_truncation(struct ubifs_info *c, struct inode *inode,
const struct iattr *attr)
{
int err;
struct ubifs_budget_req req;
loff_t old_size = inode->i_size, new_size = attr->ia_size;
int offset = new_size & (UBIFS_BLOCK_SIZE - 1), budgeted = 1;
struct ubifs_inode *ui = ubifs_inode(inode);
dbg_gen("ino %lu, size %lld -> %lld", inode->i_ino, old_size, new_size);
memset(&req, 0, sizeof(struct ubifs_budget_req));
/*
* If this is truncation to a smaller size, and we do not truncate on a
* block boundary, budget for changing one data block, because the last
* block will be re-written.
*/
if (new_size & (UBIFS_BLOCK_SIZE - 1))
req.dirtied_page = 1;
req.dirtied_ino = 1;
/* A funny way to budget for truncation node */
req.dirtied_ino_d = UBIFS_TRUN_NODE_SZ;
err = ubifs_budget_space(c, &req);
if (err) {
/*
* Treat truncations to zero as deletion and always allow them,
* just like we do for '->unlink()'.
*/
if (new_size || err != -ENOSPC)
return err;
budgeted = 0;
}
truncate_setsize(inode, new_size);
if (offset) {
pgoff_t index = new_size >> PAGE_CACHE_SHIFT;
struct page *page;
page = find_lock_page(inode->i_mapping, index);
if (page) {
if (PageDirty(page)) {
/*
* 'ubifs_jnl_truncate()' will try to truncate
* the last data node, but it contains
* out-of-date data because the page is dirty.
* Write the page now, so that
* 'ubifs_jnl_truncate()' will see an already
* truncated (and up to date) data node.
*/
ubifs_assert(PagePrivate(page));
clear_page_dirty_for_io(page);
if (UBIFS_BLOCKS_PER_PAGE_SHIFT)
offset = new_size &
(PAGE_CACHE_SIZE - 1);
err = do_writepage(page, offset);
page_cache_release(page);
if (err)
goto out_budg;
/*
* We could now tell 'ubifs_jnl_truncate()' not
* to read the last block.
*/
} else {
/*
* We could 'kmap()' the page and pass the data
* to 'ubifs_jnl_truncate()' to save it from
* having to read it.
*/
unlock_page(page);
page_cache_release(page);
}
}
}
mutex_lock(&ui->ui_mutex);
ui->ui_size = inode->i_size;
/* Truncation changes inode [mc]time */
inode->i_mtime = inode->i_ctime = ubifs_current_time(inode);
/* Other attributes may be changed at the same time as well */
do_attr_changes(inode, attr);
err = ubifs_jnl_truncate(c, inode, old_size, new_size);
mutex_unlock(&ui->ui_mutex);
out_budg:
if (budgeted)
ubifs_release_budget(c, &req);
else {
c->bi.nospace = c->bi.nospace_rp = 0;
smp_wmb();
}
return err;
}
/**
* do_setattr - change inode attributes.
* @c: UBIFS file-system description object
* @inode: inode to change attributes for
* @attr: inode attribute changes description
*
* This function implements VFS '->setattr()' call for all cases except
* truncations to smaller size. Returns zero in case of success and a negative
* error code in case of failure.
*/
static int do_setattr(struct ubifs_info *c, struct inode *inode,
const struct iattr *attr)
{
int err, release;
loff_t new_size = attr->ia_size;
struct ubifs_inode *ui = ubifs_inode(inode);
struct ubifs_budget_req req = { .dirtied_ino = 1,
.dirtied_ino_d = ALIGN(ui->data_len, 8) };
err = ubifs_budget_space(c, &req);
if (err)
return err;
if (attr->ia_valid & ATTR_SIZE) {
dbg_gen("size %lld -> %lld", inode->i_size, new_size);
truncate_setsize(inode, new_size);
}
mutex_lock(&ui->ui_mutex);
if (attr->ia_valid & ATTR_SIZE) {
/* Truncation changes inode [mc]time */
inode->i_mtime = inode->i_ctime = ubifs_current_time(inode);
/* 'truncate_setsize()' changed @i_size, update @ui_size */
ui->ui_size = inode->i_size;
}
do_attr_changes(inode, attr);
release = ui->dirty;
if (attr->ia_valid & ATTR_SIZE)
/*
* Inode length changed, so we have to make sure
* @I_DIRTY_DATASYNC is set.
*/
__mark_inode_dirty(inode, I_DIRTY_SYNC | I_DIRTY_DATASYNC);
else
mark_inode_dirty_sync(inode);
mutex_unlock(&ui->ui_mutex);
if (release)
ubifs_release_budget(c, &req);
if (IS_SYNC(inode))
err = inode->i_sb->s_op->write_inode(inode, NULL);
return err;
}
int ubifs_setattr(struct dentry *dentry, struct iattr *attr)
{
int err;
struct inode *inode = dentry->d_inode;
struct ubifs_info *c = inode->i_sb->s_fs_info;
dbg_gen("ino %lu, mode %#x, ia_valid %#x",
inode->i_ino, inode->i_mode, attr->ia_valid);
err = inode_change_ok(inode, attr);
if (err)
return err;
err = dbg_check_synced_i_size(c, inode);
if (err)
return err;
if ((attr->ia_valid & ATTR_SIZE) && attr->ia_size < inode->i_size)
/* Truncation to a smaller size */
err = do_truncation(c, inode, attr);
else
err = do_setattr(c, inode, attr);
return err;
}
static void ubifs_invalidatepage(struct page *page, unsigned long offset)
{
struct inode *inode = page->mapping->host;
struct ubifs_info *c = inode->i_sb->s_fs_info;
ubifs_assert(PagePrivate(page));
if (offset)
/* Partial page remains dirty */
return;
if (PageChecked(page))
release_new_page_budget(c);
else
release_existing_page_budget(c);
atomic_long_dec(&c->dirty_pg_cnt);
ClearPagePrivate(page);
ClearPageChecked(page);
}
static void *ubifs_follow_link(struct dentry *dentry, struct nameidata *nd)
{
struct ubifs_inode *ui = ubifs_inode(dentry->d_inode);
nd_set_link(nd, ui->data);
return NULL;
}
int ubifs_fsync(struct file *file, loff_t start, loff_t end, int datasync)
{
struct inode *inode = file->f_mapping->host;
struct ubifs_info *c = inode->i_sb->s_fs_info;
int err;
dbg_gen("syncing inode %lu", inode->i_ino);
if (c->ro_mount)
/*
* For some really strange reasons VFS does not filter out
* 'fsync()' for R/O mounted file-systems as per 2.6.39.
*/
return 0;
err = filemap_write_and_wait_range(inode->i_mapping, start, end);
if (err)
return err;
mutex_lock(&inode->i_mutex);
/* Synchronize the inode unless this is a 'datasync()' call. */
if (!datasync || (inode->i_state & I_DIRTY_DATASYNC)) {
err = inode->i_sb->s_op->write_inode(inode, NULL);
if (err)
goto out;
}
/*
* Nodes related to this inode may still sit in a write-buffer. Flush
* them.
*/
err = ubifs_sync_wbufs_by_inode(c, inode);
out:
mutex_unlock(&inode->i_mutex);
return err;
}
/**
* mctime_update_needed - check if mtime or ctime update is needed.
* @inode: the inode to do the check for
* @now: current time
*
* This helper function checks if the inode mtime/ctime should be updated or
* not. If current values of the time-stamps are within the UBIFS inode time
* granularity, they are not updated. This is an optimization.
*/
static inline int mctime_update_needed(const struct inode *inode,
const struct timespec *now)
{
if (!timespec_equal(&inode->i_mtime, now) ||
!timespec_equal(&inode->i_ctime, now))
return 1;
return 0;
}
/**
* update_ctime - update mtime and ctime of an inode.
* @c: UBIFS file-system description object
* @inode: inode to update
*
* This function updates mtime and ctime of the inode if it is not equivalent to
* current time. Returns zero in case of success and a negative error code in
* case of failure.
*/
static int update_mctime(struct ubifs_info *c, struct inode *inode)
{
struct timespec now = ubifs_current_time(inode);
struct ubifs_inode *ui = ubifs_inode(inode);
if (mctime_update_needed(inode, &now)) {
int err, release;
struct ubifs_budget_req req = { .dirtied_ino = 1,
.dirtied_ino_d = ALIGN(ui->data_len, 8) };
err = ubifs_budget_space(c, &req);
if (err)
return err;
mutex_lock(&ui->ui_mutex);
inode->i_mtime = inode->i_ctime = ubifs_current_time(inode);
release = ui->dirty;
mark_inode_dirty_sync(inode);
mutex_unlock(&ui->ui_mutex);
if (release)
ubifs_release_budget(c, &req);
}
return 0;
}
static ssize_t ubifs_aio_write(struct kiocb *iocb, const struct iovec *iov,
unsigned long nr_segs, loff_t pos)
{
int err;
struct inode *inode = iocb->ki_filp->f_mapping->host;
struct ubifs_info *c = inode->i_sb->s_fs_info;
err = update_mctime(c, inode);
if (err)
return err;
return generic_file_aio_write(iocb, iov, nr_segs, pos);
}
static int ubifs_set_page_dirty(struct page *page)
{
int ret;
ret = __set_page_dirty_nobuffers(page);
/*
* An attempt to dirty a page without budgeting for it - should not
* happen.
*/
ubifs_assert(ret == 0);
return ret;
}
static int ubifs_releasepage(struct page *page, gfp_t unused_gfp_flags)
{
/*
* An attempt to release a dirty page without budgeting for it - should
* not happen.
*/
if (PageWriteback(page))
return 0;
ubifs_assert(PagePrivate(page));
ubifs_assert(0);
ClearPagePrivate(page);
ClearPageChecked(page);
return 1;
}
/*
* mmap()d file has taken write protection fault and is being made writable.
* UBIFS must ensure page is budgeted for.
*/
static int ubifs_vm_page_mkwrite(struct vm_area_struct *vma,
struct vm_fault *vmf)
{
struct page *page = vmf->page;
struct inode *inode = vma->vm_file->f_path.dentry->d_inode;
struct ubifs_info *c = inode->i_sb->s_fs_info;
struct timespec now = ubifs_current_time(inode);
struct ubifs_budget_req req = { .new_page = 1 };
int err, update_time;
dbg_gen("ino %lu, pg %lu, i_size %lld", inode->i_ino, page->index,
i_size_read(inode));
ubifs_assert(!c->ro_media && !c->ro_mount);
if (unlikely(c->ro_error))
return VM_FAULT_SIGBUS; /* -EROFS */
/*
* We have not locked @page so far so we may budget for changing the
* page. Note, we cannot do this after we locked the page, because
* budgeting may cause write-back which would cause deadlock.
*
* At the moment we do not know whether the page is dirty or not, so we
* assume that it is not and budget for a new page. We could look at
* the @PG_private flag and figure this out, but we may race with write
* back and the page state may change by the time we lock it, so this
* would need additional care. We do not bother with this at the
* moment, although it might be good idea to do. Instead, we allocate
* budget for a new page and amend it later on if the page was in fact
* dirty.
*
* The budgeting-related logic of this function is similar to what we
* do in 'ubifs_write_begin()' and 'ubifs_write_end()'. Glance there
* for more comments.
*/
update_time = mctime_update_needed(inode, &now);
if (update_time)
/*
* We have to change inode time stamp which requires extra
* budgeting.
*/
req.dirtied_ino = 1;
err = ubifs_budget_space(c, &req);
if (unlikely(err)) {
if (err == -ENOSPC)
ubifs_warn("out of space for mmapped file "
"(inode number %lu)", inode->i_ino);
return VM_FAULT_SIGBUS;
}
lock_page(page);
if (unlikely(page->mapping != inode->i_mapping ||
page_offset(page) > i_size_read(inode))) {
/* Page got truncated out from underneath us */
err = -EINVAL;
goto out_unlock;
}
if (PagePrivate(page))
release_new_page_budget(c);
else {
if (!PageChecked(page))
ubifs_convert_page_budget(c);
SetPagePrivate(page);
atomic_long_inc(&c->dirty_pg_cnt);
__set_page_dirty_nobuffers(page);
}
if (update_time) {
int release;
struct ubifs_inode *ui = ubifs_inode(inode);
mutex_lock(&ui->ui_mutex);
inode->i_mtime = inode->i_ctime = ubifs_current_time(inode);
release = ui->dirty;
mark_inode_dirty_sync(inode);
mutex_unlock(&ui->ui_mutex);
if (release)
ubifs_release_dirty_inode_budget(c, ui);
}
unlock_page(page);
return 0;
out_unlock:
unlock_page(page);
ubifs_release_budget(c, &req);
if (err)
err = VM_FAULT_SIGBUS;
return err;
}
static const struct vm_operations_struct ubifs_file_vm_ops = {
.fault = filemap_fault,
.page_mkwrite = ubifs_vm_page_mkwrite,
};
static int ubifs_file_mmap(struct file *file, struct vm_area_struct *vma)
{
int err;
err = generic_file_mmap(file, vma);
if (err)
return err;
vma->vm_ops = &ubifs_file_vm_ops;
return 0;
}
const struct address_space_operations ubifs_file_address_operations = {
.readpage = ubifs_readpage,
.writepage = ubifs_writepage,
.write_begin = ubifs_write_begin,
.write_end = ubifs_write_end,
.invalidatepage = ubifs_invalidatepage,
.set_page_dirty = ubifs_set_page_dirty,
.releasepage = ubifs_releasepage,
};
const struct inode_operations ubifs_file_inode_operations = {
.setattr = ubifs_setattr,
.getattr = ubifs_getattr,
.setxattr = ubifs_setxattr,
.getxattr = ubifs_getxattr,
.listxattr = ubifs_listxattr,
.removexattr = ubifs_removexattr,
};
const struct inode_operations ubifs_symlink_inode_operations = {
.readlink = generic_readlink,
.follow_link = ubifs_follow_link,
.setattr = ubifs_setattr,
.getattr = ubifs_getattr,
};
const struct file_operations ubifs_file_operations = {
.llseek = generic_file_llseek,
.read = do_sync_read,
.write = do_sync_write,
.aio_read = generic_file_aio_read,
.aio_write = ubifs_aio_write,
.mmap = ubifs_file_mmap,
.fsync = ubifs_fsync,
.unlocked_ioctl = ubifs_ioctl,
.splice_read = generic_file_splice_read,
.splice_write = generic_file_splice_write,
#ifdef CONFIG_COMPAT
.compat_ioctl = ubifs_compat_ioctl,
#endif
};