2021-05-10 14:32:00

by Jaegeuk Kim

Subject: [PATCH] f2fs: avoid null pointer access when handling IPU error

Unable to handle kernel NULL pointer dereference at virtual address 000000000000001a
pc : f2fs_inplace_write_data+0x144/0x208
lr : f2fs_inplace_write_data+0x134/0x208
Call trace:
f2fs_inplace_write_data+0x144/0x208
f2fs_do_write_data_page+0x270/0x770
f2fs_write_single_data_page+0x47c/0x830
__f2fs_write_data_pages+0x444/0x98c
f2fs_write_data_pages.llvm.16514453770497736882+0x2c/0x38
do_writepages+0x58/0x118
__writeback_single_inode+0x44/0x300
writeback_sb_inodes+0x4b8/0x9c8
wb_writeback+0x148/0x42c
wb_do_writeback+0xc8/0x390
wb_workfn+0xb0/0x2f4
process_one_work+0x1fc/0x444
worker_thread+0x268/0x4b4
kthread+0x13c/0x158
ret_from_fork+0x10/0x18

Fixes: 955772787667 ("f2fs: drop inplace IO if fs status is abnormal")
Signed-off-by: Jaegeuk Kim <[email protected]>
---
fs/f2fs/segment.c | 8 +++++---
1 file changed, 5 insertions(+), 3 deletions(-)

diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c
index c605415840b5..ae875557d693 100644
--- a/fs/f2fs/segment.c
+++ b/fs/f2fs/segment.c
@@ -3577,9 +3577,11 @@ int f2fs_inplace_write_data(struct f2fs_io_info *fio)
if (fio->bio) {
struct bio *bio = *(fio->bio);

- bio->bi_status = BLK_STS_IOERR;
- bio_endio(bio);
- fio->bio = NULL;
+ if (bio) {
+ bio->bi_status = BLK_STS_IOERR;
+ bio_endio(bio);
+ fio->bio = NULL;
+ }
}
return err;
}
--
2.31.1.607.g51e8a6a459-goog

2021-05-11 01:30:25

by Chao Yu

Subject: Re: [f2fs-dev] [PATCH] f2fs: avoid null pointer access when handling IPU error

On 2021/5/10 22:28, Jaegeuk Kim wrote:
> Unable to handle kernel NULL pointer dereference at virtual address 000000000000001a
> pc : f2fs_inplace_write_data+0x144/0x208
> lr : f2fs_inplace_write_data+0x134/0x208
> Call trace:
> f2fs_inplace_write_data+0x144/0x208
> f2fs_do_write_data_page+0x270/0x770
> f2fs_write_single_data_page+0x47c/0x830
> __f2fs_write_data_pages+0x444/0x98c
> f2fs_write_data_pages.llvm.16514453770497736882+0x2c/0x38
> do_writepages+0x58/0x118
> __writeback_single_inode+0x44/0x300
> writeback_sb_inodes+0x4b8/0x9c8
> wb_writeback+0x148/0x42c
> wb_do_writeback+0xc8/0x390
> wb_workfn+0xb0/0x2f4
> process_one_work+0x1fc/0x444
> worker_thread+0x268/0x4b4
> kthread+0x13c/0x158
> ret_from_fork+0x10/0x18
>
> Fixes: 955772787667 ("f2fs: drop inplace IO if fs status is abnormal")

My bad, thanks for fixing this.

> Signed-off-by: Jaegeuk Kim <[email protected]>
> ---
> fs/f2fs/segment.c | 8 +++++---
> 1 file changed, 5 insertions(+), 3 deletions(-)
>
> diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c
> index c605415840b5..ae875557d693 100644
> --- a/fs/f2fs/segment.c
> +++ b/fs/f2fs/segment.c
> @@ -3577,9 +3577,11 @@ int f2fs_inplace_write_data(struct f2fs_io_info *fio)
> if (fio->bio) {
> struct bio *bio = *(fio->bio);
>
> - bio->bi_status = BLK_STS_IOERR;
> - bio_endio(bio);
> - fio->bio = NULL;
> + if (bio) {
> + bio->bi_status = BLK_STS_IOERR;
> + bio_endio(bio);
> + fio->bio = NULL;

fio->bio points a bio assigned in writepages(), so it should reset
that bio to NULL by *(fio->bio) = NULL.

Thanks,

> + }
> }
> return err;
> }
>

2021-05-11 05:08:42

by Jaegeuk Kim

Subject: Re: [f2fs-dev] [PATCH] f2fs: avoid null pointer access when handling IPU error

On 05/11, Chao Yu wrote:
> On 2021/5/10 22:28, Jaegeuk Kim wrote:
> > Unable to handle kernel NULL pointer dereference at virtual address 000000000000001a
> > pc : f2fs_inplace_write_data+0x144/0x208
> > lr : f2fs_inplace_write_data+0x134/0x208
> > Call trace:
> > f2fs_inplace_write_data+0x144/0x208
> > f2fs_do_write_data_page+0x270/0x770
> > f2fs_write_single_data_page+0x47c/0x830
> > __f2fs_write_data_pages+0x444/0x98c
> > f2fs_write_data_pages.llvm.16514453770497736882+0x2c/0x38
> > do_writepages+0x58/0x118
> > __writeback_single_inode+0x44/0x300
> > writeback_sb_inodes+0x4b8/0x9c8
> > wb_writeback+0x148/0x42c
> > wb_do_writeback+0xc8/0x390
> > wb_workfn+0xb0/0x2f4
> > process_one_work+0x1fc/0x444
> > worker_thread+0x268/0x4b4
> > kthread+0x13c/0x158
> > ret_from_fork+0x10/0x18
> >
> > Fixes: 955772787667 ("f2fs: drop inplace IO if fs status is abnormal")
>
> My bad, thanks for fixing this.
>
> > Signed-off-by: Jaegeuk Kim <[email protected]>
> > ---
> > fs/f2fs/segment.c | 8 +++++---
> > 1 file changed, 5 insertions(+), 3 deletions(-)
> >
> > diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c
> > index c605415840b5..ae875557d693 100644
> > --- a/fs/f2fs/segment.c
> > +++ b/fs/f2fs/segment.c
> > @@ -3577,9 +3577,11 @@ int f2fs_inplace_write_data(struct f2fs_io_info *fio)
> > if (fio->bio) {
> > struct bio *bio = *(fio->bio);
> > - bio->bi_status = BLK_STS_IOERR;
> > - bio_endio(bio);
> > - fio->bio = NULL;
> > + if (bio) {
> > + bio->bi_status = BLK_STS_IOERR;
> > + bio_endio(bio);
> > + fio->bio = NULL;
>
> fio->bio points a bio assigned in writepages(), so it should reset
> that bio to NULL by *(fio->bio) = NULL.

Good timing. I hit other kernel panic, and it seems this is the root cause.
Let me give it a try. :)

--- v2 ---

Unable to handle kernel NULL pointer dereference at virtual address 000000000000001a
pc : f2fs_inplace_write_data+0x144/0x208
lr : f2fs_inplace_write_data+0x134/0x208
Call trace:
f2fs_inplace_write_data+0x144/0x208
f2fs_do_write_data_page+0x270/0x770
f2fs_write_single_data_page+0x47c/0x830
__f2fs_write_data_pages+0x444/0x98c
f2fs_write_data_pages.llvm.16514453770497736882+0x2c/0x38
do_writepages+0x58/0x118
__writeback_single_inode+0x44/0x300
writeback_sb_inodes+0x4b8/0x9c8
wb_writeback+0x148/0x42c
wb_do_writeback+0xc8/0x390
wb_workfn+0xb0/0x2f4
process_one_work+0x1fc/0x444
worker_thread+0x268/0x4b4
kthread+0x13c/0x158
ret_from_fork+0x10/0x18

Fixes: 955772787667 ("f2fs: drop inplace IO if fs status is abnormal")
Signed-off-by: Jaegeuk Kim <[email protected]>
---
fs/f2fs/segment.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c
index c605415840b5..51dc79fad4fe 100644
--- a/fs/f2fs/segment.c
+++ b/fs/f2fs/segment.c
@@ -3574,12 +3574,12 @@ int f2fs_inplace_write_data(struct f2fs_io_info *fio)

return err;
drop_bio:
- if (fio->bio) {
+ if (fio->bio && *(fio->bio)) {
struct bio *bio = *(fio->bio);

bio->bi_status = BLK_STS_IOERR;
bio_endio(bio);
- fio->bio = NULL;
+ *(fio->bio) = NULL;
}
return err;
}
--
2.31.1.607.g51e8a6a459-goog

2021-05-11 06:05:11

by Chao Yu

Subject: Re: [f2fs-dev] [PATCH] f2fs: avoid null pointer access when handling IPU error

On 2021/5/11 13:07, Jaegeuk Kim wrote:
> On 05/11, Chao Yu wrote:
>> On 2021/5/10 22:28, Jaegeuk Kim wrote:
>>> Unable to handle kernel NULL pointer dereference at virtual address 000000000000001a
>>> pc : f2fs_inplace_write_data+0x144/0x208
>>> lr : f2fs_inplace_write_data+0x134/0x208
>>> Call trace:
>>> f2fs_inplace_write_data+0x144/0x208
>>> f2fs_do_write_data_page+0x270/0x770
>>> f2fs_write_single_data_page+0x47c/0x830
>>> __f2fs_write_data_pages+0x444/0x98c
>>> f2fs_write_data_pages.llvm.16514453770497736882+0x2c/0x38
>>> do_writepages+0x58/0x118
>>> __writeback_single_inode+0x44/0x300
>>> writeback_sb_inodes+0x4b8/0x9c8
>>> wb_writeback+0x148/0x42c
>>> wb_do_writeback+0xc8/0x390
>>> wb_workfn+0xb0/0x2f4
>>> process_one_work+0x1fc/0x444
>>> worker_thread+0x268/0x4b4
>>> kthread+0x13c/0x158
>>> ret_from_fork+0x10/0x18
>>>
>>> Fixes: 955772787667 ("f2fs: drop inplace IO if fs status is abnormal")
>>
>> My bad, thanks for fixing this.
>>
>>> Signed-off-by: Jaegeuk Kim <[email protected]>
>>> ---
>>> fs/f2fs/segment.c | 8 +++++---
>>> 1 file changed, 5 insertions(+), 3 deletions(-)
>>>
>>> diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c
>>> index c605415840b5..ae875557d693 100644
>>> --- a/fs/f2fs/segment.c
>>> +++ b/fs/f2fs/segment.c
>>> @@ -3577,9 +3577,11 @@ int f2fs_inplace_write_data(struct f2fs_io_info *fio)
>>> if (fio->bio) {
>>> struct bio *bio = *(fio->bio);
>>> - bio->bi_status = BLK_STS_IOERR;
>>> - bio_endio(bio);
>>> - fio->bio = NULL;
>>> + if (bio) {
>>> + bio->bi_status = BLK_STS_IOERR;
>>> + bio_endio(bio);
>>> + fio->bio = NULL;
>>
>> fio->bio points a bio assigned in writepages(), so it should reset
>> that bio to NULL by *(fio->bio) = NULL.
>
> Good timing. I hit other kernel panic, and it seems this is the root cause.
> Let me give it a try. :)
>
> --- v2 ---
>
> Unable to handle kernel NULL pointer dereference at virtual address 000000000000001a
> pc : f2fs_inplace_write_data+0x144/0x208
> lr : f2fs_inplace_write_data+0x134/0x208
> Call trace:
> f2fs_inplace_write_data+0x144/0x208
> f2fs_do_write_data_page+0x270/0x770
> f2fs_write_single_data_page+0x47c/0x830
> __f2fs_write_data_pages+0x444/0x98c
> f2fs_write_data_pages.llvm.16514453770497736882+0x2c/0x38
> do_writepages+0x58/0x118
> __writeback_single_inode+0x44/0x300
> writeback_sb_inodes+0x4b8/0x9c8
> wb_writeback+0x148/0x42c
> wb_do_writeback+0xc8/0x390
> wb_workfn+0xb0/0x2f4
> process_one_work+0x1fc/0x444
> worker_thread+0x268/0x4b4
> kthread+0x13c/0x158
> ret_from_fork+0x10/0x18
>
> Fixes: 955772787667 ("f2fs: drop inplace IO if fs status is abnormal")
> Signed-off-by: Jaegeuk Kim <[email protected]>

Reviewed-by: Chao Yu <[email protected]>

Thanks,