2001-12-08 05:22:12

by Britt Park

[permalink] [raw]
Subject: The demise of notify_change.

Somewhen between 2.2.x and 2.4.x notify_change disappeared from
super_operations. What is the accepted practice now for updating an
inode's persistent state? Should one use write_inode for the same
purpose or should one rely on file_operations::setattr (excuse the
c++ism)? Or is there something entirely different that one should do?

With apologies for being behind the times (I'm trying to update a
filesystem I wrote for 2.0.x, which fact dates me accurately.),

Britt

P.S. I promise to send REG amendments to his VFS doc when I get done.


2001-12-10 02:50:43

by Chris Wright

[permalink] [raw]
Subject: Re: The demise of notify_change.

* Britt Park ([email protected]) wrote:
> Somewhen between 2.2.x and 2.4.x notify_change disappeared from
> super_operations. What is the accepted practice now for updating an
> inode's persistent state? Should one use write_inode for the same
> purpose or should one rely on file_operations::setattr (excuse the
> c++ism)? Or is there something entirely different that one should do?

read fs/attr.c::notify_change(), i believe the inode_operations->setattr()
is what you are looking for.

cheers,
-chris