Re: [PATCH v15 5/9] erofs: introduce the page cache share feature

[Hongbo Li <lihongbo22@huawei.com>]

On 2026/1/20 20:29, Hongbo Li wrote:
> 
> 
> On 2026/1/16 23:46, Christoph Hellwig wrote:
>> I don't really understand the fingerprint idea.  Files with the
>> same content will point to the same physical disk blocks, so that
>> should be a much better indicator than a finger print?  Also how does
>> the fingerprint guarantee uniqueness?  Is it a cryptographically
>> secure hash?  In here it just seems like an opaque blob.
>>
>>> +static inline int erofs_inode_set_aops(struct inode *inode,
>>> +                       struct inode *realinode, bool no_fscache)
>>
>> Factoring this out first would be a nice little prep patch.
>> Also it would probably be much cleaner using IS_ENABLED.
> 
> Ok, Thanks for reviewing. I will refine in next version.

Sorry I overlooked this point. Factoring this out is a good idea, but we 
cannot use IS_ENABLED here, because some aops is not visible  when the 
relevant config macro is not enabled. So I choose to keep this format 
and only to factor this out.

Thanks,
Hongbo

> 
> Thanks,
> Hongbo
> 
>>
>>> +static int erofs_ishare_file_open(struct inode *inode, struct file 
>>> *file)
>>> +{
>>> +    struct inode *sharedinode = EROFS_I(inode)->sharedinode;
>>
>> Ok, it looks like this allocates a separate backing file and inode.
>>