Re: [PATCH] ocfs2: kill osb->system_file_mutex lock

On Sat, May 16, 2026 at 12:51:32AM +0900, Tetsuo Handa wrote:
> On 2026/05/16 0:35, Heming Zhao wrote:
> > Hi,
> > 
> > The logic here is incorrect. The purpose of the refcount is to track how many
> > consumers are using the inode.
> 
> The igrab(inode) for the first time is for getting a refcount for the slot, isn't it?
> The igrab(inode) for the subsequent times is for getting a refcount which is supposed to be dropped by the caller, isn't it?
> _ocfs2_get_system_file_inode() always gets a refcount which is supposed to be dropped by the caller when it succeeds, doesn't it?
> 
> So why do you want one more refcount?
> 

Hi,

Could you please reply inline? your current reply style makes it very hard for
others to follow. I made a mistake in the previous email, and wrote the analysis
to continue my previous email.

- Heming