Re: [PATCH v2] netfs: Fix missing alloc tagging of direct mempool allocations

["Vlastimil Babka (SUSE)" <vbabka@kernel.org>]

+Cc Christoph

On 9/23/26 08:37, Hao Ge wrote:
> Commit 1d78d56c43ef ("netfs: Fix folio_queue ENOMEM in writeback by
> adding a mempool") added a mempool for the folio_queues and made the
> request, subrequest and folio_queue allocations distinguish between
> writeback and everything else.  Writeback is part of memory reclaim
> and must not fail due to ENOMEM, so it allocates under GFP_NOFS
> through mempool_alloc(), which may dip into the pool's reserve and,
> if that runs empty, wait for elements to be returned.  The
> GFP_KERNEL paths, which can return -ENOMEM to their callers, invoke
> the pool's ->alloc() callback directly instead.
> 
> The direct call, however, skips the alloc_hooks() wrapper that the
> mempool_alloc() macro provides.  The pool callbacks, mempool_alloc_slab()
> and mempool_kmalloc(), call kmem_cache_alloc_noprof() and kmalloc_noprof()
> and rely on current->alloc_tag having been set by the caller.  With
> CONFIG_MEM_ALLOC_PROFILING_DEBUG=y this leads to
> 
>     current->alloc_tag not set
>     WARNING: ./include/linux/alloc_tag.h:161 at __alloc_tagging_slab_alloc_hook
>     alloc_tag was not set
>     WARNING: ./include/linux/alloc_tag.h:166 at __alloc_tagging_slab_free_hook
> 
> at allocation and free time respectively, as reported when reading
> files on a CIFS mount.  The allocations are also missing from
> /proc/allocinfo.
> 
> Wrap the direct ->alloc() invocations in alloc_hooks() with a new
> mempool_alloc_noreserve() helper in include/linux/mempool.h, next to
> the other alloc_hooks()-wrapped macros such as mempool_alloc().  The
> GFP_KERNEL paths keep their failable allocation semantics, they just
> get tagged now.
> 
> Fixes: 1d78d56c43ef ("netfs: Fix folio_queue ENOMEM in writeback by adding a mempool")
> Reported-by: Erhard Furtner <erhard_f@mailbox.org>
> Closes: https://lore.kernel.org/all/0b004319-9ef7-437c-a4dd-174d6a9a83db@mailbox.org/
> Tested-by: Erhard Furtner <erhard_f@mailbox.org>
> Suggested-by: Suren Baghdasaryan <surenb@google.com>
> Cc: stable@vger.kernel.org
> Signed-off-by: Hao Ge <hao.ge@linux.dev>

Acked-by: Vlastimil Babka (SUSE) <vbabka@kernel.org>

Seems ok enough as a hotfix. The naming perhaps isn't ideal, we have
mempool_alloc_preallocated() already for the opposite case. Maybe
mempool_alloc_no_preallocated() to complement it? Rather long though.

I guess long-term it would be neater to have a mempool_alloc() variant with
extra parameter that says if preallocated pool is allowed or not. The "gfp
== GFP_KERNEL" check is probably not an universal one. mempool_alloc() is
actually rather aggressive in getting objects from the pool before even
attempting reclaim. Makes sense for I/O usecases but maybe not for this one?

> ---
> Changes in v2:
> - Move the alloc_hooks() wrapper from fs/netfs/internal.h into
>   include/linux/mempool.h as mempool_alloc_noreserve(), per Suren's
>   suggestion, so it is not netfs-specific.
> 
> v1 Link: https://lore.kernel.org/all/20260922023541.51532-1-hao.ge@linux.dev/
> ---
>  fs/netfs/objects.c        | 4 ++--
>  fs/netfs/rolling_buffer.c | 2 +-
>  include/linux/mempool.h   | 7 +++++++
>  3 files changed, 10 insertions(+), 3 deletions(-)
> 
> diff --git a/fs/netfs/objects.c b/fs/netfs/objects.c
> index 7f6a3e912602..ad549daa9c79 100644
> --- a/fs/netfs/objects.c
> +++ b/fs/netfs/objects.c
> @@ -34,7 +34,7 @@ struct netfs_io_request *netfs_alloc_request(struct address_space *mapping,
>  
>  		rreq = mempool_alloc(mempool, gfp);
>  	} else {
> -		rreq = mempool->alloc(gfp, mempool->pool_data);
> +		rreq = mempool_alloc_noreserve(mempool, gfp);
>  		if (!rreq)
>  			return ERR_PTR(-ENOMEM);
>  	}
> @@ -214,7 +214,7 @@ struct netfs_io_subrequest *netfs_alloc_subrequest(struct netfs_io_request *rreq
>  	struct kmem_cache *cache = mempool->pool_data;
>  
>  	if (rreq->gfp == GFP_KERNEL)
> -		subreq = mempool->alloc(rreq->gfp, mempool->pool_data);
> +		subreq = mempool_alloc_noreserve(mempool, rreq->gfp);
>  	else
>  		subreq = mempool_alloc(mempool, rreq->gfp);
>  	if (!subreq)
> diff --git a/fs/netfs/rolling_buffer.c b/fs/netfs/rolling_buffer.c
> index 424e77a9a109..d30d5ef6d86e 100644
> --- a/fs/netfs/rolling_buffer.c
> +++ b/fs/netfs/rolling_buffer.c
> @@ -29,7 +29,7 @@ struct folio_queue *netfs_folioq_alloc(unsigned int rreq_id, gfp_t gfp,
>  	struct folio_queue *fq;
>  
>  	if (gfp == GFP_KERNEL)
> -		fq = netfs_folioq_pool.alloc(gfp, netfs_folioq_pool.pool_data);
> +		fq = mempool_alloc_noreserve(&netfs_folioq_pool, gfp);
>  	else
>  		fq = mempool_alloc(&netfs_folioq_pool, gfp);
>  	if (fq) {
> diff --git a/include/linux/mempool.h b/include/linux/mempool.h
> index a0fa6d43e0dc..6da502aef2f7 100644
> --- a/include/linux/mempool.h
> +++ b/include/linux/mempool.h
> @@ -70,6 +70,13 @@ int mempool_alloc_bulk_noprof(struct mempool *pool, void **elem,
>  #define mempool_alloc_bulk(...)						\
>  	alloc_hooks(mempool_alloc_bulk_noprof(__VA_ARGS__))
>  
> +/*
> + * Allocate a new element without dipping into the pool's reserves or
> + * waiting.  Returns NULL on failure.
> + */
> +#define mempool_alloc_noreserve(_pool, _gfp)				\
> +	alloc_hooks((_pool)->alloc(_gfp, (_pool)->pool_data))
> +
>  void *mempool_alloc_preallocated(struct mempool *pool) __malloc;
>  void mempool_free(void *element, struct mempool *pool);
>  unsigned int mempool_free_bulk(struct mempool *pool, void **elem,