linux_dsm_epyc7002/fs/ext4/page-io.c
Theodore Ts'o 1ada47d946 ext4: fix ext4_evict_inode() racing against workqueue processing code
Commit 84c17543ab (ext4: move work from io_end to inode) triggered a
regression when running xfstest #270 when the file system is mounted
with dioread_nolock.

The problem is that after ext4_evict_inode() calls ext4_ioend_wait(),
this guarantees that last io_end structure has been freed, but it does
not guarantee that the workqueue structure, which was moved into the
inode by commit 84c17543ab, is actually finished.  Once
ext4_flush_completed_IO() calls ext4_free_io_end() on CPU #1, this
will allow ext4_ioend_wait() to return on CPU #2, at which point the
evict_inode() codepath can race against the workqueue code on CPU #1
accessing EXT4_I(inode)->i_unwritten_work to find the next item of
work to do.

Fix this by calling cancel_work_sync() in ext4_ioend_wait(), which
will be renamed ext4_ioend_shutdown(), since it is only used by
ext4_evict_inode().  Also, move the call to ext4_ioend_shutdown()
until after truncate_inode_pages() and filemap_write_and_wait() are
called, to make sure all dirty pages have been written back and
flushed from the page cache first.

BUG: unable to handle kernel NULL pointer dereference at   (null)
IP: [<c01dda6a>] cwq_activate_delayed_work+0x3b/0x7e
*pdpt = 0000000030bc3001 *pde = 0000000000000000 
Oops: 0000 [#1] SMP DEBUG_PAGEALLOC
Modules linked in:
Pid: 6, comm: kworker/u:0 Not tainted 3.8.0-rc3-00013-g84c1754-dirty #91 Bochs Bochs
EIP: 0060:[<c01dda6a>] EFLAGS: 00010046 CPU: 0
EIP is at cwq_activate_delayed_work+0x3b/0x7e
EAX: 00000000 EBX: 00000000 ECX: f505fe54 EDX: 00000000
ESI: ed5b697c EDI: 00000006 EBP: f64b7e8c ESP: f64b7e84
 DS: 007b ES: 007b FS: 00d8 GS: 0000 SS: 0068
CR0: 8005003b CR2: 00000000 CR3: 30bc2000 CR4: 000006f0
DR0: 00000000 DR1: 00000000 DR2: 00000000 DR3: 00000000
DR6: ffff0ff0 DR7: 00000400
Process kworker/u:0 (pid: 6, ti=f64b6000 task=f64b4160 task.ti=f64b6000)
Stack:
 f505fe00 00000006 f64b7e9c c01de3d7 f6435540 00000003 f64b7efc c01def1d
 f6435540 00000002 00000000 0000008a c16d0808 c040a10b c16d07d8 c16d08b0
 f505fe00 c16d0780 00000000 00000000 ee153df4 c1ce4a30 c17d0e30 00000000
Call Trace:
 [<c01de3d7>] cwq_dec_nr_in_flight+0x71/0xfb
 [<c01def1d>] process_one_work+0x5d8/0x637
 [<c040a10b>] ? ext4_end_bio+0x300/0x300
 [<c01e3105>] worker_thread+0x249/0x3ef
 [<c01ea317>] kthread+0xd8/0xeb
 [<c01e2ebc>] ? manage_workers+0x4bb/0x4bb
 [<c023a370>] ? trace_hardirqs_on+0x27/0x37
 [<c0f1b4b7>] ret_from_kernel_thread+0x1b/0x28
 [<c01ea23f>] ? __init_kthread_worker+0x71/0x71
Code: 01 83 15 ac ff 6c c1 00 31 db 89 c6 8b 00 a8 04 74 12 89 c3 30 db 83 05 b0 ff 6c c1 01 83 15 b4 ff 6c c1 00 89 f0 e8 42 ff ff ff <8b> 13 89 f0 83 05 b8 ff 6c c1
 6c c1 00 31 c9 83
EIP: [<c01dda6a>] cwq_activate_delayed_work+0x3b/0x7e SS:ESP 0068:f64b7e84
CR2: 0000000000000000
---[ end trace a1923229da53d8a4 ]---

Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Cc: Jan Kara <jack@suse.cz>
2013-03-20 09:39:42 -04:00

472 lines
12 KiB
C

/*
* linux/fs/ext4/page-io.c
*
* This contains the new page_io functions for ext4
*
* Written by Theodore Ts'o, 2010.
*/
#include <linux/fs.h>
#include <linux/time.h>
#include <linux/jbd2.h>
#include <linux/highuid.h>
#include <linux/pagemap.h>
#include <linux/quotaops.h>
#include <linux/string.h>
#include <linux/buffer_head.h>
#include <linux/writeback.h>
#include <linux/pagevec.h>
#include <linux/mpage.h>
#include <linux/namei.h>
#include <linux/uio.h>
#include <linux/bio.h>
#include <linux/workqueue.h>
#include <linux/kernel.h>
#include <linux/slab.h>
#include <linux/mm.h>
#include "ext4_jbd2.h"
#include "xattr.h"
#include "acl.h"
static struct kmem_cache *io_page_cachep, *io_end_cachep;
int __init ext4_init_pageio(void)
{
io_page_cachep = KMEM_CACHE(ext4_io_page, SLAB_RECLAIM_ACCOUNT);
if (io_page_cachep == NULL)
return -ENOMEM;
io_end_cachep = KMEM_CACHE(ext4_io_end, SLAB_RECLAIM_ACCOUNT);
if (io_end_cachep == NULL) {
kmem_cache_destroy(io_page_cachep);
return -ENOMEM;
}
return 0;
}
void ext4_exit_pageio(void)
{
kmem_cache_destroy(io_end_cachep);
kmem_cache_destroy(io_page_cachep);
}
/*
* This function is called by ext4_evict_inode() to make sure there is
* no more pending I/O completion work left to do.
*/
void ext4_ioend_shutdown(struct inode *inode)
{
wait_queue_head_t *wq = ext4_ioend_wq(inode);
wait_event(*wq, (atomic_read(&EXT4_I(inode)->i_ioend_count) == 0));
/*
* We need to make sure the work structure is finished being
* used before we let the inode get destroyed.
*/
if (work_pending(&EXT4_I(inode)->i_unwritten_work))
cancel_work_sync(&EXT4_I(inode)->i_unwritten_work);
}
static void put_io_page(struct ext4_io_page *io_page)
{
if (atomic_dec_and_test(&io_page->p_count)) {
end_page_writeback(io_page->p_page);
put_page(io_page->p_page);
kmem_cache_free(io_page_cachep, io_page);
}
}
void ext4_free_io_end(ext4_io_end_t *io)
{
int i;
BUG_ON(!io);
BUG_ON(!list_empty(&io->list));
BUG_ON(io->flag & EXT4_IO_END_UNWRITTEN);
for (i = 0; i < io->num_io_pages; i++)
put_io_page(io->pages[i]);
io->num_io_pages = 0;
if (atomic_dec_and_test(&EXT4_I(io->inode)->i_ioend_count))
wake_up_all(ext4_ioend_wq(io->inode));
kmem_cache_free(io_end_cachep, io);
}
/* check a range of space and convert unwritten extents to written. */
static int ext4_end_io(ext4_io_end_t *io)
{
struct inode *inode = io->inode;
loff_t offset = io->offset;
ssize_t size = io->size;
int ret = 0;
ext4_debug("ext4_end_io_nolock: io 0x%p from inode %lu,list->next 0x%p,"
"list->prev 0x%p\n",
io, inode->i_ino, io->list.next, io->list.prev);
ret = ext4_convert_unwritten_extents(inode, offset, size);
if (ret < 0) {
ext4_msg(inode->i_sb, KERN_EMERG,
"failed to convert unwritten extents to written "
"extents -- potential data loss! "
"(inode %lu, offset %llu, size %zd, error %d)",
inode->i_ino, offset, size, ret);
}
/* Wake up anyone waiting on unwritten extent conversion */
if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten))
wake_up_all(ext4_ioend_wq(inode));
if (io->flag & EXT4_IO_END_DIRECT)
inode_dio_done(inode);
if (io->iocb)
aio_complete(io->iocb, io->result, 0);
return ret;
}
static void dump_completed_IO(struct inode *inode)
{
#ifdef EXT4FS_DEBUG
struct list_head *cur, *before, *after;
ext4_io_end_t *io, *io0, *io1;
if (list_empty(&EXT4_I(inode)->i_completed_io_list)) {
ext4_debug("inode %lu completed_io list is empty\n",
inode->i_ino);
return;
}
ext4_debug("Dump inode %lu completed_io list\n", inode->i_ino);
list_for_each_entry(io, &EXT4_I(inode)->i_completed_io_list, list) {
cur = &io->list;
before = cur->prev;
io0 = container_of(before, ext4_io_end_t, list);
after = cur->next;
io1 = container_of(after, ext4_io_end_t, list);
ext4_debug("io 0x%p from inode %lu,prev 0x%p,next 0x%p\n",
io, inode->i_ino, io0, io1);
}
#endif
}
/* Add the io_end to per-inode completed end_io list. */
void ext4_add_complete_io(ext4_io_end_t *io_end)
{
struct ext4_inode_info *ei = EXT4_I(io_end->inode);
struct workqueue_struct *wq;
unsigned long flags;
BUG_ON(!(io_end->flag & EXT4_IO_END_UNWRITTEN));
wq = EXT4_SB(io_end->inode->i_sb)->dio_unwritten_wq;
spin_lock_irqsave(&ei->i_completed_io_lock, flags);
if (list_empty(&ei->i_completed_io_list))
queue_work(wq, &ei->i_unwritten_work);
list_add_tail(&io_end->list, &ei->i_completed_io_list);
spin_unlock_irqrestore(&ei->i_completed_io_lock, flags);
}
static int ext4_do_flush_completed_IO(struct inode *inode)
{
ext4_io_end_t *io;
struct list_head unwritten;
unsigned long flags;
struct ext4_inode_info *ei = EXT4_I(inode);
int err, ret = 0;
spin_lock_irqsave(&ei->i_completed_io_lock, flags);
dump_completed_IO(inode);
list_replace_init(&ei->i_completed_io_list, &unwritten);
spin_unlock_irqrestore(&ei->i_completed_io_lock, flags);
while (!list_empty(&unwritten)) {
io = list_entry(unwritten.next, ext4_io_end_t, list);
BUG_ON(!(io->flag & EXT4_IO_END_UNWRITTEN));
list_del_init(&io->list);
err = ext4_end_io(io);
if (unlikely(!ret && err))
ret = err;
io->flag &= ~EXT4_IO_END_UNWRITTEN;
ext4_free_io_end(io);
}
return ret;
}
/*
* work on completed aio dio IO, to convert unwritten extents to extents
*/
void ext4_end_io_work(struct work_struct *work)
{
struct ext4_inode_info *ei = container_of(work, struct ext4_inode_info,
i_unwritten_work);
ext4_do_flush_completed_IO(&ei->vfs_inode);
}
int ext4_flush_unwritten_io(struct inode *inode)
{
int ret;
WARN_ON_ONCE(!mutex_is_locked(&inode->i_mutex) &&
!(inode->i_state & I_FREEING));
ret = ext4_do_flush_completed_IO(inode);
ext4_unwritten_wait(inode);
return ret;
}
ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags)
{
ext4_io_end_t *io = kmem_cache_zalloc(io_end_cachep, flags);
if (io) {
atomic_inc(&EXT4_I(inode)->i_ioend_count);
io->inode = inode;
INIT_LIST_HEAD(&io->list);
}
return io;
}
/*
* Print an buffer I/O error compatible with the fs/buffer.c. This
* provides compatibility with dmesg scrapers that look for a specific
* buffer I/O error message. We really need a unified error reporting
* structure to userspace ala Digital Unix's uerf system, but it's
* probably not going to happen in my lifetime, due to LKML politics...
*/
static void buffer_io_error(struct buffer_head *bh)
{
char b[BDEVNAME_SIZE];
printk(KERN_ERR "Buffer I/O error on device %s, logical block %llu\n",
bdevname(bh->b_bdev, b),
(unsigned long long)bh->b_blocknr);
}
static void ext4_end_bio(struct bio *bio, int error)
{
ext4_io_end_t *io_end = bio->bi_private;
struct inode *inode;
int i;
sector_t bi_sector = bio->bi_sector;
BUG_ON(!io_end);
bio->bi_private = NULL;
bio->bi_end_io = NULL;
if (test_bit(BIO_UPTODATE, &bio->bi_flags))
error = 0;
bio_put(bio);
for (i = 0; i < io_end->num_io_pages; i++) {
struct page *page = io_end->pages[i]->p_page;
struct buffer_head *bh, *head;
loff_t offset;
loff_t io_end_offset;
if (error) {
SetPageError(page);
set_bit(AS_EIO, &page->mapping->flags);
head = page_buffers(page);
BUG_ON(!head);
io_end_offset = io_end->offset + io_end->size;
offset = (sector_t) page->index << PAGE_CACHE_SHIFT;
bh = head;
do {
if ((offset >= io_end->offset) &&
(offset+bh->b_size <= io_end_offset))
buffer_io_error(bh);
offset += bh->b_size;
bh = bh->b_this_page;
} while (bh != head);
}
put_io_page(io_end->pages[i]);
}
io_end->num_io_pages = 0;
inode = io_end->inode;
if (error) {
io_end->flag |= EXT4_IO_END_ERROR;
ext4_warning(inode->i_sb, "I/O error writing to inode %lu "
"(offset %llu size %ld starting block %llu)",
inode->i_ino,
(unsigned long long) io_end->offset,
(long) io_end->size,
(unsigned long long)
bi_sector >> (inode->i_blkbits - 9));
}
if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
ext4_free_io_end(io_end);
return;
}
ext4_add_complete_io(io_end);
}
void ext4_io_submit(struct ext4_io_submit *io)
{
struct bio *bio = io->io_bio;
if (bio) {
bio_get(io->io_bio);
submit_bio(io->io_op, io->io_bio);
BUG_ON(bio_flagged(io->io_bio, BIO_EOPNOTSUPP));
bio_put(io->io_bio);
}
io->io_bio = NULL;
io->io_op = 0;
io->io_end = NULL;
}
static int io_submit_init(struct ext4_io_submit *io,
struct inode *inode,
struct writeback_control *wbc,
struct buffer_head *bh)
{
ext4_io_end_t *io_end;
struct page *page = bh->b_page;
int nvecs = bio_get_nr_vecs(bh->b_bdev);
struct bio *bio;
io_end = ext4_init_io_end(inode, GFP_NOFS);
if (!io_end)
return -ENOMEM;
bio = bio_alloc(GFP_NOIO, min(nvecs, BIO_MAX_PAGES));
bio->bi_sector = bh->b_blocknr * (bh->b_size >> 9);
bio->bi_bdev = bh->b_bdev;
bio->bi_private = io->io_end = io_end;
bio->bi_end_io = ext4_end_bio;
io_end->offset = (page->index << PAGE_CACHE_SHIFT) + bh_offset(bh);
io->io_bio = bio;
io->io_op = (wbc->sync_mode == WB_SYNC_ALL ? WRITE_SYNC : WRITE);
io->io_next_block = bh->b_blocknr;
return 0;
}
static int io_submit_add_bh(struct ext4_io_submit *io,
struct ext4_io_page *io_page,
struct inode *inode,
struct writeback_control *wbc,
struct buffer_head *bh)
{
ext4_io_end_t *io_end;
int ret;
if (buffer_new(bh)) {
clear_buffer_new(bh);
unmap_underlying_metadata(bh->b_bdev, bh->b_blocknr);
}
if (io->io_bio && bh->b_blocknr != io->io_next_block) {
submit_and_retry:
ext4_io_submit(io);
}
if (io->io_bio == NULL) {
ret = io_submit_init(io, inode, wbc, bh);
if (ret)
return ret;
}
io_end = io->io_end;
if ((io_end->num_io_pages >= MAX_IO_PAGES) &&
(io_end->pages[io_end->num_io_pages-1] != io_page))
goto submit_and_retry;
if (buffer_uninit(bh))
ext4_set_io_unwritten_flag(inode, io_end);
io->io_end->size += bh->b_size;
io->io_next_block++;
ret = bio_add_page(io->io_bio, bh->b_page, bh->b_size, bh_offset(bh));
if (ret != bh->b_size)
goto submit_and_retry;
if ((io_end->num_io_pages == 0) ||
(io_end->pages[io_end->num_io_pages-1] != io_page)) {
io_end->pages[io_end->num_io_pages++] = io_page;
atomic_inc(&io_page->p_count);
}
return 0;
}
int ext4_bio_write_page(struct ext4_io_submit *io,
struct page *page,
int len,
struct writeback_control *wbc)
{
struct inode *inode = page->mapping->host;
unsigned block_start, block_end, blocksize;
struct ext4_io_page *io_page;
struct buffer_head *bh, *head;
int ret = 0;
blocksize = 1 << inode->i_blkbits;
BUG_ON(!PageLocked(page));
BUG_ON(PageWriteback(page));
io_page = kmem_cache_alloc(io_page_cachep, GFP_NOFS);
if (!io_page) {
redirty_page_for_writepage(wbc, page);
unlock_page(page);
return -ENOMEM;
}
io_page->p_page = page;
atomic_set(&io_page->p_count, 1);
get_page(page);
set_page_writeback(page);
ClearPageError(page);
for (bh = head = page_buffers(page), block_start = 0;
bh != head || !block_start;
block_start = block_end, bh = bh->b_this_page) {
block_end = block_start + blocksize;
if (block_start >= len) {
/*
* Comments copied from block_write_full_page_endio:
*
* 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."
*/
zero_user_segment(page, block_start, block_end);
clear_buffer_dirty(bh);
set_buffer_uptodate(bh);
continue;
}
if (!buffer_dirty(bh) || buffer_delay(bh) ||
!buffer_mapped(bh) || buffer_unwritten(bh)) {
/* A hole? We can safely clear the dirty bit */
if (!buffer_mapped(bh))
clear_buffer_dirty(bh);
if (io->io_bio)
ext4_io_submit(io);
continue;
}
ret = io_submit_add_bh(io, io_page, inode, wbc, bh);
if (ret) {
/*
* We only get here on ENOMEM. Not much else
* we can do but mark the page as dirty, and
* better luck next time.
*/
redirty_page_for_writepage(wbc, page);
break;
}
clear_buffer_dirty(bh);
}
unlock_page(page);
/*
* If the page was truncated before we could do the writeback,
* or we had a memory allocation error while trying to write
* the first buffer head, we won't have submitted any pages for
* I/O. In that case we need to make sure we've cleared the
* PageWriteback bit from the page to prevent the system from
* wedging later on.
*/
put_io_page(io_page);
return ret;
}