Re: [PATCH] block: integrity: Do not call set_page_dirty_lock()

[Christoph Hellwig <hch@infradead.org>]

On Wed, Apr 16, 2025 at 04:04:10PM -0400, Martin K. Petersen wrote:
> 
> Placing multiple protection information buffers inside the same page
> can lead to oopses because set_page_dirty_lock() can't be called from
> interrupt context.
> 
> Since a protection information buffer is not backed by a file there is
> no point in setting its page dirty, there is nothing to synchronize.
> Drop the call to set_page_dirty_lock() and remove the last argument to
> bio_integrity_unpin_bvec().

Who sais it it not backed by a file?  Which that would be a very unusual
use case, there is nothing limiting us from using integity or metadata
from a file mapping.  Instead we'll need to do the same thing for the
data path and defer the unmapping to a user context when needed.

(Assuming we can get rid of this entirely, about which we have another
discussion I don't have time to follow up on as I'm on vacation)