[PATCH -next v3] erofs: Fix pcluster memleak when its block address is zero

[Chao Yu]

On 2022/12/5 14:08, Gao Xiang wrote:
> Hi all,
> 
> On Mon, Dec 05, 2022 at 11:49:57AM +0800, Chen Zhongjin wrote:
>> syzkaller reported a memleak:
>> https://syzkaller.appspot.com/bug?id=62f37ff612f0021641eda5b17f056f1668aa9aed
>>
>> unreferenced object 0xffff88811009c7f8 (size 136):
>>    ...
>>    backtrace:
>>      [<ffffffff821db19b>] z_erofs_do_read_page+0x99b/0x1740
>>      [<ffffffff821dee9e>] z_erofs_readahead+0x24e/0x580
>>      [<ffffffff814bc0d6>] read_pages+0x86/0x3d0
>>      ...
>>
>> syzkaller constructed a case: in z_erofs_register_pcluster(),
>> ztailpacking = false and map->m_pa = zero. This makes pcl->obj.index be
>> zero although pcl is not a inline pcluster.
>>
>> Then following path adds refcount for grp, but the refcount won't be put
>> because pcl is inline.
>>
>> z_erofs_readahead()
>>    z_erofs_do_read_page() # for another page
>>      z_erofs_collector_begin()
>>        erofs_find_workgroup()
>>          erofs_workgroup_get()
>>
>> Since it's illegal for the block address of a pcluster to be zero, add
>> check here to avoid registering the pcluster which would be leaked.
>>
>> Fixes: cecf864d3d76 ("erofs: support inline data decompression")
>> Reported-by: syzbot+6f8cd9a0155b366d227f at syzkaller.appspotmail.com
>> Signed-off-by: Chen Zhongjin <chenzhongjin at huawei.com>
>> ---
>> v1 -> v2:
>> As Gao's advice, we should fail to register pcluster if m_pa is zero.
>> Maked it this way and changed the commit message.
>>
>> v2 -> v3:
>> Slightly fix commit message and add -next tag.
> 
> I've updated the patch itself as below
> (Since we only need to fail out for non-tailpacking cases, tailpacking
>   inline inodes could still have m_pa < EROFS_BLKSIZ):
> 
>  From f5e037e760d338ca0c116e507be663cb843d42f0 Mon Sep 17 00:00:00 2001
> From: Chen Zhongjin <chenzhongjin at huawei.com>
> Date: Mon, 5 Dec 2022 11:49:57 +0800
> Subject: [PATCH] erofs: Fix pcluster memleak when its block address is zero
> 
> syzkaller reported a memleak:
> https://syzkaller.appspot.com/bug?id=62f37ff612f0021641eda5b17f056f1668aa9aed
> 
> unreferenced object 0xffff88811009c7f8 (size 136):
>    ...
>    backtrace:
>      [<ffffffff821db19b>] z_erofs_do_read_page+0x99b/0x1740
>      [<ffffffff821dee9e>] z_erofs_readahead+0x24e/0x580
>      [<ffffffff814bc0d6>] read_pages+0x86/0x3d0
>      ...
> 
> syzkaller constructed a case: in z_erofs_register_pcluster(),
> ztailpacking = false and map->m_pa = zero. This makes pcl->obj.index be
> zero although pcl is not a inline pcluster.
> 
> Then following path adds refcount for grp, but the refcount won't be put
> because pcl is inline.
> 
> z_erofs_readahead()
>    z_erofs_do_read_page() # for another page
>      z_erofs_collector_begin()
>        erofs_find_workgroup()
>          erofs_workgroup_get()
> 
> Since it's illegal for the block address of a non-inlined pcluster to
> be zero, add check here to avoid registering the pcluster which would
> be leaked.
> 
> Fixes: cecf864d3d76 ("erofs: support inline data decompression")
> Reported-by: syzbot+6f8cd9a0155b366d227f at syzkaller.appspotmail.com
> Signed-off-by: Chen Zhongjin <chenzhongjin at huawei.com>
> Reviewed-by: Yue Hu <huyue2 at coolpad.com>
> Reviewed-by: Gao Xiang <hsiangkao at linux.alibaba.com>
> Signed-off-by: Gao Xiang <hsiangkao at linux.alibaba.com>

Reviewed-by: Chao Yu <chao at kernel.org>

Thanks,