Kernel/linux-next
1
0
Fork 0
mirror of https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git synced 2024-12-29 09:12:07 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

ext4: Convert to use mapping->invalidate_lock

Browse Source 
Convert ext4 to use mapping->invalidate_lock instead of its private
EXT4_I(inode)->i_mmap_sem. This is mostly search-and-replace. By this
conversion we fix a long standing race between hole punching and read(2)
/ readahead(2) paths that can lead to stale page cache contents.

CC: <linux-ext4@vger.kernel.org>
CC: Ted Tso <tytso@mit.edu>
Acked-by: Theodore Ts'o <tytso@mit.edu>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Jan Kara <jack@suse.cz>
This commit is contained in:
Jan Kara 2021-02-04 18:05:42 +01:00
parent 7506ae6a70
commit d4f5258eae
7 changed files with 50 additions and 70 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

10
fs/ext4/ext4.h
View File

@ -1086,15 +1086,6 @@ struct ext4_inode_info {
* by other means, so we have i_data_sem. * by other means, so we have i_data_sem.
*/ */
struct rw_semaphore i_data_sem; struct rw_semaphore i_data_sem;
/*
* i_mmap_sem is for serializing page faults with truncate / punch hole
* operations. We have to make sure that new page cannot be faulted in
* a section of the inode that is being punched. We cannot easily use
* i_data_sem for this since we need protection for the whole punch
* operation and i_data_sem ranks below transaction start so we have
* to occasionally drop it.
*/
struct rw_semaphore i_mmap_sem;
struct inode vfs_inode; struct inode vfs_inode;
struct jbd2_inode *jinode; struct jbd2_inode *jinode;
@ -2972,7 +2963,6 @@ extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode, extern int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
loff_t lstart, loff_t lend); loff_t lstart, loff_t lend);
extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf); extern vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf);
extern vm_fault_t ext4_filemap_fault(struct vm_fault *vmf);
extern qsize_t *ext4_get_reserved_space(struct inode *inode); extern qsize_t *ext4_get_reserved_space(struct inode *inode);
extern int ext4_get_projid(struct inode *inode, kprojid_t *projid); extern int ext4_get_projid(struct inode *inode, kprojid_t *projid);
extern void ext4_da_release_space(struct inode *inode, int to_free); extern void ext4_da_release_space(struct inode *inode, int to_free);

25
fs/ext4/extents.c
View File

@ -4474,6 +4474,7 @@ static long ext4_zero_range(struct file *file, loff_t offset,
loff_t len, int mode) loff_t len, int mode)
{ {
struct inode *inode = file_inode(file); struct inode *inode = file_inode(file);
struct address_space *mapping = file->f_mapping;
handle_t *handle = NULL; handle_t *handle = NULL;
unsigned int max_blocks; unsigned int max_blocks;
loff_t new_size = 0; loff_t new_size = 0;
@ -4560,17 +4561,17 @@ static long ext4_zero_range(struct file *file, loff_t offset,
* Prevent page faults from reinstantiating pages we have * Prevent page faults from reinstantiating pages we have
* released from page cache. * released from page cache.
*/ */
down_write(&EXT4_I(inode)->i_mmap_sem); filemap_invalidate_lock(mapping);
ret = ext4_break_layouts(inode); ret = ext4_break_layouts(inode);
if (ret) { if (ret) {
up_write(&EXT4_I(inode)->i_mmap_sem); filemap_invalidate_unlock(mapping);
goto out_mutex; goto out_mutex;
} }
ret = ext4_update_disksize_before_punch(inode, offset, len); ret = ext4_update_disksize_before_punch(inode, offset, len);
if (ret) { if (ret) {
up_write(&EXT4_I(inode)->i_mmap_sem); filemap_invalidate_unlock(mapping);
goto out_mutex; goto out_mutex;
} }
/* Now release the pages and zero block aligned part of pages */ /* Now release the pages and zero block aligned part of pages */
@ -4579,7 +4580,7 @@ static long ext4_zero_range(struct file *file, loff_t offset,
ret = ext4_alloc_file_blocks(file, lblk, max_blocks, new_size, ret = ext4_alloc_file_blocks(file, lblk, max_blocks, new_size,
flags); flags);
up_write(&EXT4_I(inode)->i_mmap_sem); filemap_invalidate_unlock(mapping);
if (ret) if (ret)
goto out_mutex; goto out_mutex;
} }
@ -5221,6 +5222,7 @@ ext4_ext_shift_extents(struct inode *inode, handle_t *handle,
static int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len) static int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len)
{ {
struct super_block *sb = inode->i_sb; struct super_block *sb = inode->i_sb;
struct address_space *mapping = inode->i_mapping;
ext4_lblk_t punch_start, punch_stop; ext4_lblk_t punch_start, punch_stop;
handle_t *handle; handle_t *handle;
unsigned int credits; unsigned int credits;
@ -5274,7 +5276,7 @@ static int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len)
* Prevent page faults from reinstantiating pages we have released from * Prevent page faults from reinstantiating pages we have released from
* page cache. * page cache.
*/ */
down_write(&EXT4_I(inode)->i_mmap_sem); filemap_invalidate_lock(mapping);
ret = ext4_break_layouts(inode); ret = ext4_break_layouts(inode);
if (ret) if (ret)
@ -5289,15 +5291,15 @@ static int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len)
* Write tail of the last page before removed range since it will get * Write tail of the last page before removed range since it will get
* removed from the page cache below. * removed from the page cache below.
*/ */
ret = filemap_write_and_wait_range(inode->i_mapping, ioffset, offset); ret = filemap_write_and_wait_range(mapping, ioffset, offset);
if (ret) if (ret)
goto out_mmap; goto out_mmap;
/* /*
* Write data that will be shifted to preserve them when discarding * Write data that will be shifted to preserve them when discarding
* page cache below. We are also protected from pages becoming dirty * page cache below. We are also protected from pages becoming dirty
* by i_mmap_sem. * by i_rwsem and invalidate_lock.
*/ */
ret = filemap_write_and_wait_range(inode->i_mapping, offset + len, ret = filemap_write_and_wait_range(mapping, offset + len,
LLONG_MAX); LLONG_MAX);
if (ret) if (ret)
goto out_mmap; goto out_mmap;
@ -5350,7 +5352,7 @@ static int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len)
ext4_journal_stop(handle); ext4_journal_stop(handle);
ext4_fc_stop_ineligible(sb); ext4_fc_stop_ineligible(sb);
out_mmap: out_mmap:
up_write(&EXT4_I(inode)->i_mmap_sem); filemap_invalidate_unlock(mapping);
out_mutex: out_mutex:
inode_unlock(inode); inode_unlock(inode);
return ret; return ret;
@ -5367,6 +5369,7 @@ static int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len)
static int ext4_insert_range(struct inode *inode, loff_t offset, loff_t len) static int ext4_insert_range(struct inode *inode, loff_t offset, loff_t len)
{ {
struct super_block *sb = inode->i_sb; struct super_block *sb = inode->i_sb;
struct address_space *mapping = inode->i_mapping;
handle_t *handle; handle_t *handle;
struct ext4_ext_path *path; struct ext4_ext_path *path;
struct ext4_extent *extent; struct ext4_extent *extent;
@ -5425,7 +5428,7 @@ static int ext4_insert_range(struct inode *inode, loff_t offset, loff_t len)
* Prevent page faults from reinstantiating pages we have released from * Prevent page faults from reinstantiating pages we have released from
* page cache. * page cache.
*/ */
down_write(&EXT4_I(inode)->i_mmap_sem); filemap_invalidate_lock(mapping);
ret = ext4_break_layouts(inode); ret = ext4_break_layouts(inode);
if (ret) if (ret)
@ -5526,7 +5529,7 @@ static int ext4_insert_range(struct inode *inode, loff_t offset, loff_t len)
ext4_journal_stop(handle); ext4_journal_stop(handle);
ext4_fc_stop_ineligible(sb); ext4_fc_stop_ineligible(sb);
out_mmap: out_mmap:
up_write(&EXT4_I(inode)->i_mmap_sem); filemap_invalidate_unlock(mapping);
out_mutex: out_mutex:
inode_unlock(inode); inode_unlock(inode);
return ret; return ret;

13
fs/ext4/file.c
View File

@ -704,22 +704,23 @@ static vm_fault_t ext4_dax_huge_fault(struct vm_fault *vmf,
*/ */
bool write = (vmf->flags & FAULT_FLAG_WRITE) && bool write = (vmf->flags & FAULT_FLAG_WRITE) &&
(vmf->vma->vm_flags & VM_SHARED); (vmf->vma->vm_flags & VM_SHARED);
struct address_space *mapping = vmf->vma->vm_file->f_mapping;
pfn_t pfn; pfn_t pfn;
if (write) { if (write) {
sb_start_pagefault(sb); sb_start_pagefault(sb);
file_update_time(vmf->vma->vm_file); file_update_time(vmf->vma->vm_file);
down_read(&EXT4_I(inode)->i_mmap_sem); filemap_invalidate_lock_shared(mapping);
retry: retry:
handle = ext4_journal_start_sb(sb, EXT4_HT_WRITE_PAGE, handle = ext4_journal_start_sb(sb, EXT4_HT_WRITE_PAGE,
EXT4_DATA_TRANS_BLOCKS(sb)); EXT4_DATA_TRANS_BLOCKS(sb));
if (IS_ERR(handle)) { if (IS_ERR(handle)) {
up_read(&EXT4_I(inode)->i_mmap_sem); filemap_invalidate_unlock_shared(mapping);
sb_end_pagefault(sb); sb_end_pagefault(sb);
return VM_FAULT_SIGBUS; return VM_FAULT_SIGBUS;
} }
} else { } else {
down_read(&EXT4_I(inode)->i_mmap_sem); filemap_invalidate_lock_shared(mapping);
} }
result = dax_iomap_fault(vmf, pe_size, &pfn, &error, &ext4_iomap_ops); result = dax_iomap_fault(vmf, pe_size, &pfn, &error, &ext4_iomap_ops);
if (write) { if (write) {
@ -731,10 +732,10 @@ static vm_fault_t ext4_dax_huge_fault(struct vm_fault *vmf,
/* Handling synchronous page fault? */ /* Handling synchronous page fault? */
if (result & VM_FAULT_NEEDDSYNC) if (result & VM_FAULT_NEEDDSYNC)
result = dax_finish_sync_fault(vmf, pe_size, pfn); result = dax_finish_sync_fault(vmf, pe_size, pfn);
up_read(&EXT4_I(inode)->i_mmap_sem); filemap_invalidate_unlock_shared(mapping);
sb_end_pagefault(sb); sb_end_pagefault(sb);
} else { } else {
up_read(&EXT4_I(inode)->i_mmap_sem); filemap_invalidate_unlock_shared(mapping);
} }
return result; return result;
@ -756,7 +757,7 @@ static const struct vm_operations_struct ext4_dax_vm_ops = {
#endif #endif
static const struct vm_operations_struct ext4_file_vm_ops = { static const struct vm_operations_struct ext4_file_vm_ops = {
.fault = ext4_filemap_fault, .fault = filemap_fault,
.map_pages = filemap_map_pages, .map_pages = filemap_map_pages,
.page_mkwrite = ext4_page_mkwrite, .page_mkwrite = ext4_page_mkwrite,
}; };

47
fs/ext4/inode.c
View File

@ -3950,20 +3950,19 @@ int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset,
return ret; return ret;
} }
static void ext4_wait_dax_page(struct ext4_inode_info *ei) static void ext4_wait_dax_page(struct inode *inode)
{ {
up_write(&ei->i_mmap_sem); filemap_invalidate_unlock(inode->i_mapping);
schedule(); schedule();
down_write(&ei->i_mmap_sem); filemap_invalidate_lock(inode->i_mapping);
} }
int ext4_break_layouts(struct inode *inode) int ext4_break_layouts(struct inode *inode)
{ {
struct ext4_inode_info *ei = EXT4_I(inode);
struct page *page; struct page *page;
int error; int error;
if (WARN_ON_ONCE(!rwsem_is_locked(&ei->i_mmap_sem))) if (WARN_ON_ONCE(!rwsem_is_locked(&inode->i_mapping->invalidate_lock)))
return -EINVAL; return -EINVAL;
do { do {
@ -3974,7 +3973,7 @@ int ext4_break_layouts(struct inode *inode)
error = ___wait_var_event(&page->_refcount, error = ___wait_var_event(&page->_refcount,
atomic_read(&page->_refcount) == 1, atomic_read(&page->_refcount) == 1,
TASK_INTERRUPTIBLE, 0, 0, TASK_INTERRUPTIBLE, 0, 0,
ext4_wait_dax_page(ei)); ext4_wait_dax_page(inode));
} while (error == 0); } while (error == 0);
return error; return error;
@ -4005,9 +4004,9 @@ int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length)
ext4_clear_inode_state(inode, EXT4_STATE_MAY_INLINE_DATA); ext4_clear_inode_state(inode, EXT4_STATE_MAY_INLINE_DATA);
if (ext4_has_inline_data(inode)) { if (ext4_has_inline_data(inode)) {
down_write(&EXT4_I(inode)->i_mmap_sem); filemap_invalidate_lock(mapping);
ret = ext4_convert_inline_data(inode); ret = ext4_convert_inline_data(inode);
up_write(&EXT4_I(inode)->i_mmap_sem); filemap_invalidate_unlock(mapping);
if (ret) if (ret)
return ret; return ret;
} }
@ -4058,7 +4057,7 @@ int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length)
* Prevent page faults from reinstantiating pages we have released from * Prevent page faults from reinstantiating pages we have released from
* page cache. * page cache.
*/ */
down_write(&EXT4_I(inode)->i_mmap_sem); filemap_invalidate_lock(mapping);
ret = ext4_break_layouts(inode); ret = ext4_break_layouts(inode);
if (ret) if (ret)
@ -4131,7 +4130,7 @@ int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length)
out_stop: out_stop:
ext4_journal_stop(handle); ext4_journal_stop(handle);
out_dio: out_dio:
up_write(&EXT4_I(inode)->i_mmap_sem); filemap_invalidate_unlock(mapping);
out_mutex: out_mutex:
inode_unlock(inode); inode_unlock(inode);
return ret; return ret;
@ -5426,11 +5425,11 @@ int ext4_setattr(struct user_namespace *mnt_userns, struct dentry *dentry,
inode_dio_wait(inode); inode_dio_wait(inode);
} }
down_write(&EXT4_I(inode)->i_mmap_sem); filemap_invalidate_lock(inode->i_mapping);
rc = ext4_break_layouts(inode); rc = ext4_break_layouts(inode);
if (rc) { if (rc) {
up_write(&EXT4_I(inode)->i_mmap_sem); filemap_invalidate_unlock(inode->i_mapping);
goto err_out; goto err_out;
} }
@ -5506,7 +5505,7 @@ int ext4_setattr(struct user_namespace *mnt_userns, struct dentry *dentry,
error = rc; error = rc;
} }
out_mmap_sem: out_mmap_sem:
up_write(&EXT4_I(inode)->i_mmap_sem); filemap_invalidate_unlock(inode->i_mapping);
} }
if (!error) { if (!error) {
@ -5983,10 +5982,10 @@ int ext4_change_inode_journal_flag(struct inode *inode, int val)
* data (and journalled aops don't know how to handle these cases). * data (and journalled aops don't know how to handle these cases).
*/ */
if (val) { if (val) {
down_write(&EXT4_I(inode)->i_mmap_sem); filemap_invalidate_lock(inode->i_mapping);
err = filemap_write_and_wait(inode->i_mapping); err = filemap_write_and_wait(inode->i_mapping);
if (err < 0) { if (err < 0) {
up_write(&EXT4_I(inode)->i_mmap_sem); filemap_invalidate_unlock(inode->i_mapping);
return err; return err;
} }
} }
@ -6019,7 +6018,7 @@ int ext4_change_inode_journal_flag(struct inode *inode, int val)
percpu_up_write(&sbi->s_writepages_rwsem); percpu_up_write(&sbi->s_writepages_rwsem);
if (val) if (val)
up_write(&EXT4_I(inode)->i_mmap_sem); filemap_invalidate_unlock(inode->i_mapping);
/* Finally we can mark the inode as dirty. */ /* Finally we can mark the inode as dirty. */
@ -6063,7 +6062,7 @@ vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf)
sb_start_pagefault(inode->i_sb); sb_start_pagefault(inode->i_sb);
file_update_time(vma->vm_file); file_update_time(vma->vm_file);
down_read(&EXT4_I(inode)->i_mmap_sem); filemap_invalidate_lock_shared(mapping);
err = ext4_convert_inline_data(inode); err = ext4_convert_inline_data(inode);
if (err) if (err)
@ -6176,7 +6175,7 @@ vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf)
out_ret: out_ret:
ret = block_page_mkwrite_return(err); ret = block_page_mkwrite_return(err);
out: out:
up_read(&EXT4_I(inode)->i_mmap_sem); filemap_invalidate_unlock_shared(mapping);
sb_end_pagefault(inode->i_sb); sb_end_pagefault(inode->i_sb);
return ret; return ret;
out_error: out_error:
@ -6184,15 +6183,3 @@ vm_fault_t ext4_page_mkwrite(struct vm_fault *vmf)
ext4_journal_stop(handle); ext4_journal_stop(handle);
goto out; goto out;
} }
vm_fault_t ext4_filemap_fault(struct vm_fault *vmf)
{
struct inode *inode = file_inode(vmf->vma->vm_file);
vm_fault_t ret;
down_read(&EXT4_I(inode)->i_mmap_sem);
ret = filemap_fault(vmf);
up_read(&EXT4_I(inode)->i_mmap_sem);
return ret;
}

4
fs/ext4/ioctl.c
View File

@ -148,7 +148,7 @@ static long swap_inode_boot_loader(struct super_block *sb,
goto journal_err_out; goto journal_err_out;
} }
down_write(&EXT4_I(inode)->i_mmap_sem); filemap_invalidate_lock(inode->i_mapping);
err = filemap_write_and_wait(inode->i_mapping); err = filemap_write_and_wait(inode->i_mapping);
if (err) if (err)
goto err_out; goto err_out;
@ -256,7 +256,7 @@ static long swap_inode_boot_loader(struct super_block *sb,
ext4_double_up_write_data_sem(inode, inode_bl); ext4_double_up_write_data_sem(inode, inode_bl);
err_out: err_out:
up_write(&EXT4_I(inode)->i_mmap_sem); filemap_invalidate_unlock(inode->i_mapping);
journal_err_out: journal_err_out:
unlock_two_nondirectories(inode, inode_bl); unlock_two_nondirectories(inode, inode_bl);
iput(inode_bl); iput(inode_bl);

13
fs/ext4/super.c
View File

@ -90,12 +90,9 @@ static struct inode *ext4_get_journal_inode(struct super_block *sb,
/* /*
* Lock ordering * Lock ordering
* *
* Note the difference between i_mmap_sem (EXT4_I(inode)->i_mmap_sem) and
* i_mmap_rwsem (inode->i_mmap_rwsem)!
*
* page fault path: * page fault path:
* mmap_lock -> sb_start_pagefault -> i_mmap_sem (r) -> transaction start -> * mmap_lock -> sb_start_pagefault -> invalidate_lock (r) -> transaction start
* page lock -> i_data_sem (rw) * -> page lock -> i_data_sem (rw)
* *
* buffered write path: * buffered write path:
* sb_start_write -> i_mutex -> mmap_lock * sb_start_write -> i_mutex -> mmap_lock
@ -103,8 +100,9 @@ static struct inode *ext4_get_journal_inode(struct super_block *sb,
* i_data_sem (rw) * i_data_sem (rw)
* *
* truncate: * truncate:
* sb_start_write -> i_mutex -> i_mmap_sem (w) -> i_mmap_rwsem (w) -> page lock * sb_start_write -> i_mutex -> invalidate_lock (w) -> i_mmap_rwsem (w) ->
* sb_start_write -> i_mutex -> i_mmap_sem (w) -> transaction start -> * page lock
* sb_start_write -> i_mutex -> invalidate_lock (w) -> transaction start ->
* i_data_sem (rw) * i_data_sem (rw)
* *
* direct IO: * direct IO:
@ -1360,7 +1358,6 @@ static void init_once(void *foo)
INIT_LIST_HEAD(&ei->i_orphan); INIT_LIST_HEAD(&ei->i_orphan);
init_rwsem(&ei->xattr_sem); init_rwsem(&ei->xattr_sem);
init_rwsem(&ei->i_data_sem); init_rwsem(&ei->i_data_sem);
init_rwsem(&ei->i_mmap_sem);
inode_init_once(&ei->vfs_inode); inode_init_once(&ei->vfs_inode);
ext4_fc_init_inode(&ei->vfs_inode); ext4_fc_init_inode(&ei->vfs_inode);
} }

8
fs/ext4/truncate.h
View File

@ -11,14 +11,16 @@
*/ */
static inline void ext4_truncate_failed_write(struct inode *inode) static inline void ext4_truncate_failed_write(struct inode *inode)
{ {
struct address_space *mapping = inode->i_mapping;
/* /*
* We don't need to call ext4_break_layouts() because the blocks we * We don't need to call ext4_break_layouts() because the blocks we
* are truncating were never visible to userspace. * are truncating were never visible to userspace.
*/ */
down_write(&EXT4_I(inode)->i_mmap_sem); filemap_invalidate_lock(mapping);
truncate_inode_pages(inode->i_mapping, inode->i_size); truncate_inode_pages(mapping, inode->i_size);
ext4_truncate(inode); ext4_truncate(inode);
up_write(&EXT4_I(inode)->i_mmap_sem); filemap_invalidate_unlock(mapping);
} }
/* /*