2021-08-31 00:54:39

by Jaegeuk Kim

Subject: [PATCH 1/2] f2fs: deallocate compressed pages when error happens

In f2fs_write_multi_pages(), f2fs_compress_pages() allocates pages for
compression work in cc->cpages[]. Then, f2fs_write_compressed_pages() initiates
bio submission. But, if there's any error before submitting the IOs like early
f2fs_cp_error(), previously it didn't free cpages by f2fs_compress_free_page().
Let's fix memory leak by putting that just before deallocating cc->cpages.

Signed-off-by: Jaegeuk Kim <[email protected]>
---
fs/f2fs/compress.c | 12 ++++++------
1 file changed, 6 insertions(+), 6 deletions(-)

diff --git a/fs/f2fs/compress.c b/fs/f2fs/compress.c
index ec70a0a32327..c1bf9ad4c220 100644
--- a/fs/f2fs/compress.c
+++ b/fs/f2fs/compress.c
@@ -1394,12 +1394,6 @@ static int f2fs_write_compressed_pages(struct compress_ctx *cc,

for (--i; i >= 0; i--)
fscrypt_finalize_bounce_page(&cc->cpages[i]);
- for (i = 0; i < cc->nr_cpages; i++) {
- if (!cc->cpages[i])
- continue;
- f2fs_compress_free_page(cc->cpages[i]);
- cc->cpages[i] = NULL;
- }
out_put_cic:
kmem_cache_free(cic_entry_slab, cic);
out_put_dnode:
@@ -1410,6 +1404,12 @@ static int f2fs_write_compressed_pages(struct compress_ctx *cc,
else
f2fs_unlock_op(sbi);
out_free:
+ for (i = 0; i < cc->nr_cpages; i++) {
+ if (!cc->cpages[i])
+ continue;
+ f2fs_compress_free_page(cc->cpages[i]);
+ cc->cpages[i] = NULL;
+ }
page_array_free(cc->inode, cc->cpages, cc->nr_cpages);
cc->cpages = NULL;
return -EAGAIN;
--
2.33.0.259.gc128427fd7-goog

2021-08-31 00:56:15

by Jaegeuk Kim

Subject: [PATCH 2/2] f2fs: should put a page beyond EOF when preparing a write

The prepare_compress_overwrite() gets/locks a page to prepare a read, and calls
f2fs_read_multi_pages() which checks EOF first. If there's any page beyond EOF,
we unlock the page and set cc->rpages[i] = NULL, which we can't put the page
anymore. This makes page leak, so let's fix by putting that page.

Signed-off-by: Jaegeuk Kim <[email protected]>
---
fs/f2fs/data.c | 2 ++
1 file changed, 2 insertions(+)

diff --git a/fs/f2fs/data.c b/fs/f2fs/data.c
index 8e8824605f83..41d29382eced 100644
--- a/fs/f2fs/data.c
+++ b/fs/f2fs/data.c
@@ -2183,6 +2183,8 @@ int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret,
continue;
}
unlock_page(page);
+ if (for_write)
+ put_page(page);
cc->rpages[i] = NULL;
cc->nr_rpages--;
}
--
2.33.0.259.gc128427fd7-goog

2021-08-31 01:58:56

by Chao Yu

Subject: Re: [f2fs-dev] [PATCH 1/2] f2fs: deallocate compressed pages when error happens

On 2021/8/31 8:53, Jaegeuk Kim wrote:
> In f2fs_write_multi_pages(), f2fs_compress_pages() allocates pages for
> compression work in cc->cpages[]. Then, f2fs_write_compressed_pages() initiates
> bio submission. But, if there's any error before submitting the IOs like early
> f2fs_cp_error(), previously it didn't free cpages by f2fs_compress_free_page().
> Let's fix memory leak by putting that just before deallocating cc->cpages.
>

Fixes: 4c8ff7095bef ("f2fs: support data compression")

> Signed-off-by: Jaegeuk Kim <[email protected]>

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

Thanks,

2021-08-31 02:06:07

by Chao Yu

Subject: Re: [f2fs-dev] [PATCH 2/2] f2fs: should put a page beyond EOF when preparing a write

On 2021/8/31 8:53, Jaegeuk Kim wrote:
> The prepare_compress_overwrite() gets/locks a page to prepare a read, and calls
> f2fs_read_multi_pages() which checks EOF first. If there's any page beyond EOF,
> we unlock the page and set cc->rpages[i] = NULL, which we can't put the page
> anymore. This makes page leak, so let's fix by putting that page.
>

Fixes: a949dc5f2c5c ("f2fs: compress: fix race condition of overwrite vs truncate")

> Signed-off-by: Jaegeuk Kim <[email protected]>

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

Thanks,