Message ID | 20200625113122.7540-3-willy@infradead.org (mailing list archive) |
---|---|
State | New, archived |
Headers | show |
Series | Overhaul memalloc_no* | expand |
On Thu 25-06-20 12:31:18, Matthew Wilcox wrote: > Since XFS needs to pretend to be kswapd in some of its worker threads, > create methods to save & restore kswapd state. Don't bother restoring > kswapd state in kswapd -- the only time we reach this code is when we're > exiting and the task_struct is about to be destroyed anyway. > > Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org> Certainly better than an opencoded PF_$FOO manipulation Acked-by: Michal Hocko <mhocko@suse.com> I would just ask for a clarification because this is rellying to have a good MM knowledge to follow > +/* > + * Tell the memory management that we're a "memory allocator", I would go with. Tell the memory management that the caller is working on behalf of the background memory reclaim (aka kswapd) and help it to make a forward progress. That means that it will get an access to memory reserves should there be a need to allocate memory in order to make a forward progress. Note that the caller has to be extremely careful when doing that. Or something like that. > + * and that if we need more memory we should get access to it > + * regardless (see "__alloc_pages()"). "kswapd" should > + * never get caught in the normal page freeing logic. > + * > + * (Kswapd normally doesn't need memory anyway, but sometimes > + * you need a small amount of memory in order to be able to > + * page out something else, and this flag essentially protects > + * us from recursively trying to free more memory as we're > + * trying to free the first piece of memory in the first place). > + */ > +#define KSWAPD_PF_FLAGS (PF_MEMALLOC | PF_SWAPWRITE | PF_KSWAPD) > + > +static inline unsigned long become_kswapd(void) > +{ > + unsigned long flags = current->flags & KSWAPD_PF_FLAGS; > + current->flags |= KSWAPD_PF_FLAGS; > + return flags; > +} > + > +static inline void restore_kswapd(unsigned long flags) > +{ > + current->flags &= ~(flags ^ KSWAPD_PF_FLAGS); > +} > + > static inline void set_current_io_flusher(void) > { > current->flags |= PF_LOCAL_THROTTLE; > diff --git a/mm/vmscan.c b/mm/vmscan.c > index b6d84326bdf2..27ae76699899 100644 > --- a/mm/vmscan.c > +++ b/mm/vmscan.c > @@ -3870,19 +3870,7 @@ static int kswapd(void *p) > if (!cpumask_empty(cpumask)) > set_cpus_allowed_ptr(tsk, cpumask); > > - /* > - * Tell the memory management that we're a "memory allocator", > - * and that if we need more memory we should get access to it > - * regardless (see "__alloc_pages()"). "kswapd" should > - * never get caught in the normal page freeing logic. > - * > - * (Kswapd normally doesn't need memory anyway, but sometimes > - * you need a small amount of memory in order to be able to > - * page out something else, and this flag essentially protects > - * us from recursively trying to free more memory as we're > - * trying to free the first piece of memory in the first place). > - */ > - tsk->flags |= PF_MEMALLOC | PF_SWAPWRITE | PF_KSWAPD; > + become_kswapd(); > set_freezable(); > > WRITE_ONCE(pgdat->kswapd_order, 0); > @@ -3932,8 +3920,6 @@ static int kswapd(void *p) > goto kswapd_try_sleep; > } > > - tsk->flags &= ~(PF_MEMALLOC | PF_SWAPWRITE | PF_KSWAPD); > - > return 0; > } > > -- > 2.27.0 >
diff --git a/fs/xfs/libxfs/xfs_btree.c b/fs/xfs/libxfs/xfs_btree.c index 2d25bab68764..a04a44238aab 100644 --- a/fs/xfs/libxfs/xfs_btree.c +++ b/fs/xfs/libxfs/xfs_btree.c @@ -2813,8 +2813,9 @@ xfs_btree_split_worker( { struct xfs_btree_split_args *args = container_of(work, struct xfs_btree_split_args, work); + bool is_kswapd = args->kswapd; unsigned long pflags; - unsigned long new_pflags = PF_MEMALLOC_NOFS; + int memalloc_nofs; /* * we are in a transaction context here, but may also be doing work @@ -2822,16 +2823,17 @@ xfs_btree_split_worker( * temporarily to ensure that we don't block waiting for memory reclaim * in any way. */ - if (args->kswapd) - new_pflags |= PF_MEMALLOC | PF_SWAPWRITE | PF_KSWAPD; - - current_set_flags_nested(&pflags, new_pflags); + if (is_kswapd) + pflags = become_kswapd(); + memalloc_nofs = memalloc_nofs_save(); args->result = __xfs_btree_split(args->cur, args->level, args->ptrp, args->key, args->curp, args->stat); complete(args->done); - current_restore_flags_nested(&pflags, new_pflags); + memalloc_nofs_restore(memalloc_nofs); + if (is_kswapd) + restore_kswapd(pflags); } /* diff --git a/include/linux/sched/mm.h b/include/linux/sched/mm.h index 1a7e1ab1be85..b0089eadc367 100644 --- a/include/linux/sched/mm.h +++ b/include/linux/sched/mm.h @@ -308,6 +308,32 @@ static inline void memalloc_nocma_restore(unsigned int flags) } #endif +/* + * Tell the memory management that we're a "memory allocator", + * and that if we need more memory we should get access to it + * regardless (see "__alloc_pages()"). "kswapd" should + * never get caught in the normal page freeing logic. + * + * (Kswapd normally doesn't need memory anyway, but sometimes + * you need a small amount of memory in order to be able to + * page out something else, and this flag essentially protects + * us from recursively trying to free more memory as we're + * trying to free the first piece of memory in the first place). + */ +#define KSWAPD_PF_FLAGS (PF_MEMALLOC | PF_SWAPWRITE | PF_KSWAPD) + +static inline unsigned long become_kswapd(void) +{ + unsigned long flags = current->flags & KSWAPD_PF_FLAGS; + current->flags |= KSWAPD_PF_FLAGS; + return flags; +} + +static inline void restore_kswapd(unsigned long flags) +{ + current->flags &= ~(flags ^ KSWAPD_PF_FLAGS); +} + static inline void set_current_io_flusher(void) { current->flags |= PF_LOCAL_THROTTLE; diff --git a/mm/vmscan.c b/mm/vmscan.c index b6d84326bdf2..27ae76699899 100644 --- a/mm/vmscan.c +++ b/mm/vmscan.c @@ -3870,19 +3870,7 @@ static int kswapd(void *p) if (!cpumask_empty(cpumask)) set_cpus_allowed_ptr(tsk, cpumask); - /* - * Tell the memory management that we're a "memory allocator", - * and that if we need more memory we should get access to it - * regardless (see "__alloc_pages()"). "kswapd" should - * never get caught in the normal page freeing logic. - * - * (Kswapd normally doesn't need memory anyway, but sometimes - * you need a small amount of memory in order to be able to - * page out something else, and this flag essentially protects - * us from recursively trying to free more memory as we're - * trying to free the first piece of memory in the first place). - */ - tsk->flags |= PF_MEMALLOC | PF_SWAPWRITE | PF_KSWAPD; + become_kswapd(); set_freezable(); WRITE_ONCE(pgdat->kswapd_order, 0); @@ -3932,8 +3920,6 @@ static int kswapd(void *p) goto kswapd_try_sleep; } - tsk->flags &= ~(PF_MEMALLOC | PF_SWAPWRITE | PF_KSWAPD); - return 0; }
Since XFS needs to pretend to be kswapd in some of its worker threads, create methods to save & restore kswapd state. Don't bother restoring kswapd state in kswapd -- the only time we reach this code is when we're exiting and the task_struct is about to be destroyed anyway. Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org> --- fs/xfs/libxfs/xfs_btree.c | 14 ++++++++------ include/linux/sched/mm.h | 26 ++++++++++++++++++++++++++ mm/vmscan.c | 16 +--------------- 3 files changed, 35 insertions(+), 21 deletions(-)