mbox series

[0/3] nfsd: fix dadlock in move_to_close_lru()

Message ID 20240301000950.2306-1-neilb@suse.de (mailing list archive)
Headers show
Series nfsd: fix dadlock in move_to_close_lru() | expand

Message

NeilBrown March 1, 2024, 12:07 a.m. UTC 
This series is intended to replace 
 98be4be88369 nfsd: drop st_mutex and rp_mutex before calling move_to_close_lru()
in nfsd-next.
As Jeff Layton oberved recently there is a problem with that patch.
This series takes a different approach to the rp_mutex proble, replacing it with an
atomic_t which has three states.  Details in the patch.

Thnaks,
NeilBrown


 [PATCH 1/3] nfsd: move nfsd4_cstate_assign_replay() earlier in open
 [PATCH 2/3] nfsd: replace rp_mutex to avoid deadlock in
 [PATCH 3/3] nfsd: drop st_mutex_mutex before calling

Comments

Chuck Lever III March 1, 2024, 2:05 p.m. UTC | #1 
On Fri, Mar 01, 2024 at 11:07:36AM +1100, NeilBrown wrote:
> This series is intended to replace 
>  98be4be88369 nfsd: drop st_mutex and rp_mutex before calling move_to_close_lru()
> in nfsd-next.

Hi Neil, I will drop that patch from nfsd-next now.


> As Jeff Layton oberved recently there is a problem with that patch.
> This series takes a different approach to the rp_mutex proble, replacing it with an
> atomic_t which has three states.  Details in the patch.
> 
> Thnaks,
> NeilBrown
> 
> 
>  [PATCH 1/3] nfsd: move nfsd4_cstate_assign_replay() earlier in open
>  [PATCH 2/3] nfsd: replace rp_mutex to avoid deadlock in
>  [PATCH 3/3] nfsd: drop st_mutex_mutex before calling