Re: [PATCH v4 01/16] mm: hugetlb: Track used_hpages when getting/putting pages from subpool

[Ackerley Tng]

Ackerley Tng via B4 Relay <devnull+ackerleytng.google.com@xxxxxxxxxx>
writes:

> From: Ackerley Tng <ackerleytng@xxxxxxxxxx>
>
> hugepage_subpool_put_pages() currently has two distinct responsibilities
> that conflict:
>
> 1. When size is specified for the mount, max_hpages != -1: Keep track of
>    total active pages (allocated + reserved) and decrement this count
>    (used_hpages) when a page is freed or allocation fails.
> 2. When min_size is specified for the mount, min_hpages != -1: Ensure we
>    don't drop below the guaranteed minimum, and restore a reservation
>    (rsv_hpages) if we do.
>
> This causes trouble because when allocation fails (refer to
> alloc_hugetlb_folio()) if gbl_chg = 1 (i.e. no subpool reservation was
> taken):
>
> + To keep used_hpages consistent, HugeTLB needs to call
>   hugepage_subpool_put_pages() to restore undo used_hpages being
>   incremented
> + But can't call hugepage_subpool_put_pages() if no reservation was
>   consumed.
>
> One option would be to conditionally do subpool tracking updates outside of
> the hugepage_subpool_put_pages() function, but that would spread logic all
> over.
>
> Instead, always track used_hpages, regardless of whether a max_size was
> requested for the mount, so that the subpool always knows how many pages
> were allocated through it. Every page allocated through the subpool
> increments used_hpages, regardless of whether a reservation was taken from
> it.
>
> Conceptually, now, every allocation involving a subpool uses a page from
> the subpool, which must be returned to the subpool. Every page taken from
> the subpool tries to use a subpool reservation. Restoring a page to the
> subpool reservations only if the page was taken from subpool
> reservations. (If used_hpages >= min_hpages, the page must have not have
> been taken from the reservations.)
>
> Always tracking used_hpages provides the subpool with information of both
> used and reserved counts to make the correct decision for both max_size and
> min_size correctly.
>
> With used_hpages always tracked,
>
> + subpool_is_free() can be simplified, such that the subpool can be
>   declared free if there are no more pages in use.
> + open-coding in hugetlb_reserve_pages() can be removed.
>
> Also update the
>
> + Documentation for used_hpages in the subpool struct, since it no longer
>   matters whether the used pages count against the maximum.
> + Docstring for hugepage_subpool_{get,put}_pages
> + Documentation to use active voice, and remove some details in favor of
>   having details documented in the docstring
>
> Also update statfs reporting. Previously, if max_hpages is negative,
> used_hpages is static at 0, so returning max_hpages - used_hpages returns
> -1 and is always correct. Now, if the subpool doesn't have a maximum
> requested size, indicate no limit for free pages (-1). If it does have a
> maximum size, report the difference between the requested size and the
> number of used pages. This difference is always positive, because if the
> mount does have a maximum size, hugepage_subpool_get_pages() ensures that
> the subpool usage never exceeds the maximum.
>
> This fixes a bug in hugetlb_unreserve_pages(), where pages are returned to
> the subpool regardless of whether it consumed a reservation. The
> corresponding bug in the failure handling path of alloc_hugetlb_folio() was
> fixed in a833a693a490e.
>
> Fixes: 1c5ecae3a93fa ("hugetlbfs: add minimum size accounting to subpools")
> Cc: stable@xxxxxxxxxxxxxxx
> Signed-off-by: Ackerley Tng <ackerleytng@xxxxxxxxxx>
> ---
>  Documentation/mm/hugetlbfs_reserv.rst              |  17 +--
>  .../translations/zh_CN/mm/hugetlbfs_reserv.rst     |  11 +-
>  fs/hugetlbfs/inode.c                               |   8 +-
>  include/linux/hugetlb.h                            |   4 +-
>  mm/hugetlb.c                                       | 118 +++++++++++----------
>  5 files changed, 75 insertions(+), 83 deletions(-)
>
> diff --git a/Documentation/mm/hugetlbfs_reserv.rst b/Documentation/mm/hugetlbfs_reserv.rst
> index a49115db18c76..d244583fdcbc3 100644
> --- a/Documentation/mm/hugetlbfs_reserv.rst
> +++ b/Documentation/mm/hugetlbfs_reserv.rst
> @@ -314,21 +314,8 @@ huge pages.  If they can not be reserved, the mount fails.
>  The routines hugepage_subpool_get/put_pages() are called when pages are
>  obtained from or released back to a subpool.  They perform all subpool
>  accounting, and track any reservations associated with the subpool.
> -hugepage_subpool_get/put_pages are passed the number of huge pages by which
> -to adjust the subpool 'used page' count (down for get, up for put).  Normally,
> -they return the same value that was passed or an error if not enough pages
> -exist in the subpool.
> -
> -However, if reserves are associated with the subpool a return value less
> -than the passed value may be returned.  This return value indicates the
> -number of additional global pool adjustments which must be made.  For example,
> -suppose a subpool contains 3 reserved huge pages and someone asks for 5.
> -The 3 reserved pages associated with the subpool can be used to satisfy part
> -of the request.  But, 2 pages must be obtained from the global pools.  To
> -relay this information to the caller, the value 2 is returned.  The caller
> -is then responsible for attempting to obtain the additional two pages from
> -the global pools.
> -
> +hugepage_subpool_get/put_pages() use the number of huge pages passed to adjust
> +the subpool 'used page' count.
>
>  COW and Reservations
>  ====================
> diff --git a/Documentation/translations/zh_CN/mm/hugetlbfs_reserv.rst b/Documentation/translations/zh_CN/mm/hugetlbfs_reserv.rst
> index 20947f8bd0654..ae1f1f31477fc 100644
> --- a/Documentation/translations/zh_CN/mm/hugetlbfs_reserv.rst
> +++ b/Documentation/translations/zh_CN/mm/hugetlbfs_reserv.rst
> @@ -246,15 +246,8 @@ hugepage_subpool的min_hpages字段中被跟踪。在挂载时，hugetlb_acct_me
>  被调用以预留指定数量的巨页。如果它们不能被预留，挂载就会失败。
>
>  当从子池中获取或释放页面时，会调用hugepage_subpool_get/put_pages()函数。
> -hugepage_subpool_get/put_pages被传递给巨页数量，以此来调整子池的 “已用页面” 计数
> -（get为下降，put为上升）。通常情况下，如果子池中没有足够的页面，它们会返回与传递的相同的值或
> -一个错误。
> -
> -然而，如果预留与子池相关联，可能会返回一个小于传递值的返回值。这个返回值表示必须进行的额外全局
> -池调整的数量。例如，假设一个子池包含3个预留的巨页，有人要求5个。与子池相关的3个预留页可以用来
> -满足部分请求。但是，必须从全局池中获得2个页面。为了向调用者转达这一信息，将返回值2。然后，调用
> -者要负责从全局池中获取另外两个页面。
> -
> +它们负责所有子池的统计核算，并跟踪与子池相关联的预留。
> +hugepage_subpool_get/put_pages()函数使用传入的巨页数量来调整子池的“已用页面”计数。
>
>  COW和预留
>  ==========
> diff --git a/fs/hugetlbfs/inode.c b/fs/hugetlbfs/inode.c
> index 216e1a0dd0b23..26c0187340636 100644
> --- a/fs/hugetlbfs/inode.c
> +++ b/fs/hugetlbfs/inode.c
> @@ -1109,8 +1109,12 @@ static int hugetlbfs_statfs(struct dentry *dentry, struct kstatfs *buf)
>
>  			spin_lock_irq(&sbinfo->spool->lock);
>  			buf->f_blocks = sbinfo->spool->max_hpages;
> -			free_pages = sbinfo->spool->max_hpages
> -				- sbinfo->spool->used_hpages;
> +			if (sbinfo->spool->max_hpages == -1) {
> +				free_pages = -1;
> +			} else {
> +				free_pages = sbinfo->spool->max_hpages -
> +					     sbinfo->spool->used_hpages;
> +			}
>  			buf->f_bavail = buf->f_bfree = free_pages;
>  			spin_unlock_irq(&sbinfo->spool->lock);
>  			buf->f_files = sbinfo->max_inodes;
> diff --git a/include/linux/hugetlb.h b/include/linux/hugetlb.h
> index 2abaf99321e90..34b9a3e1be0fa 100644
> --- a/include/linux/hugetlb.h
> +++ b/include/linux/hugetlb.h
> @@ -38,8 +38,8 @@ struct hugepage_subpool {
>  	spinlock_t lock;
>  	long count;
>  	long max_hpages;	/* Maximum huge pages or -1 if no maximum. */
> -	long used_hpages;	/* Used count against maximum, includes */
> -				/* both allocated and reserved pages. */
> +	long used_hpages;	/* Used page count, includes both */
> +				/* allocated and reserved pages. */
>  	struct hstate *hstate;
>  	long min_hpages;	/* Minimum huge pages or -1 if no minimum. */
>  	long rsv_hpages;	/* Pages reserved against global pool to */
> diff --git a/mm/hugetlb.c b/mm/hugetlb.c
> index 571212b80835e..36fa3fb3945d8 100644
> --- a/mm/hugetlb.c
> +++ b/mm/hugetlb.c
> @@ -129,12 +129,8 @@ static inline bool subpool_is_free(struct hugepage_subpool *spool)
>  {
>  	if (spool->count)
>  		return false;
> -	if (spool->max_hpages != -1)
> -		return spool->used_hpages == 0;
> -	if (spool->min_hpages != -1)
> -		return spool->rsv_hpages == spool->min_hpages;
>
> -	return true;
> +	return spool->used_hpages == 0;
>  }
>
>  static inline void unlock_or_release_subpool(struct hugepage_subpool *spool,
> @@ -187,13 +183,18 @@ void hugepage_put_subpool(struct hugepage_subpool *spool)
>  	unlock_or_release_subpool(spool, flags);
>  }
>
> -/*
> - * Subpool accounting for allocating and reserving pages.
> - * Return -ENOMEM if there are not enough resources to satisfy the
> - * request.  Otherwise, return the number of pages by which the
> - * global pools must be adjusted (upward).  The returned value may
> - * only be different than the passed value (delta) in the case where
> - * a subpool minimum size must be maintained.
> +/**
> + * hugepage_subpool_get_pages - Get pages from a subpool
> + * @spool: pointer to subpool structure (may be NULL)
> + * @delta: number of pages to allocate or reserve
> + *
> + * Check and update subpool page usage counts when allocating or
> + * reserving @delta hugepages.
> + *
> + * Context: Takes spool->lock using spin_lock_irq().
> + * Return: Non-negative number of reservations that cannot be
> + *         satisfied by the subpool, or -ENOMEM if the subpool maximum
> + *         limit would be exceeded.
>   */
>  static long hugepage_subpool_get_pages(struct hugepage_subpool *spool,
>  				      long delta)
> @@ -205,15 +206,14 @@ static long hugepage_subpool_get_pages(struct hugepage_subpool *spool,
>
>  	spin_lock_irq(&spool->lock);
>
> -	if (spool->max_hpages != -1) {		/* maximum size accounting */
> -		if ((spool->used_hpages + delta) <= spool->max_hpages)
> -			spool->used_hpages += delta;
> -		else {
> -			ret = -ENOMEM;
> -			goto unlock_ret;
> -		}
> +	if (spool->max_hpages != -1 &&
> +	    spool->used_hpages + delta > spool->max_hpages) {
> +		ret = -ENOMEM;
> +		goto unlock_ret;
>  	}
>
> +	spool->used_hpages += delta;
> +
>  	/* minimum size accounting */
>  	if (spool->min_hpages != -1 && spool->rsv_hpages) {
>  		if (delta > spool->rsv_hpages) {
> @@ -234,11 +234,19 @@ static long hugepage_subpool_get_pages(struct hugepage_subpool *spool,
>  	return ret;
>  }
>
> -/*
> - * Subpool accounting for freeing and unreserving pages.
> - * Return the number of global page reservations that must be dropped.
> - * The return value may only be different than the passed value (delta)
> - * in the case where a subpool minimum size must be maintained.
> +/**
> + * hugepage_subpool_put_pages - Release pages back to a subpool
> + * @spool: pointer to subpool structure (may be NULL)
> + * @delta: number of pages to free or unreserve
> + *
> + * Check and update subpool page usage counts when freeing or
> + * unreserving @delta hugepages.
> + *
> + * Context: Takes spool->lock using spin_lock_irqsave(). May release
> + *          and free @spool if its usage count and references reach
> + *          zero.
> + * Return: Non-negative number of reservations that the subpool cannot
> + *         absorb.
>   */
>  static long hugepage_subpool_put_pages(struct hugepage_subpool *spool,
>  				       long delta)
> @@ -251,19 +259,24 @@ static long hugepage_subpool_put_pages(struct hugepage_subpool *spool,
>
>  	spin_lock_irqsave(&spool->lock, flags);
>
> -	if (spool->max_hpages != -1)		/* maximum size accounting */
> -		spool->used_hpages -= delta;
> +	spool->used_hpages -= delta;
>
>  	 /* minimum size accounting */
>  	if (spool->min_hpages != -1 && spool->used_hpages < spool->min_hpages) {
> -		if (spool->rsv_hpages + delta <= spool->min_hpages)
> +		/*
> +		 * limit is the maximum number of reservations that
> +		 * can be restored to this subpool.
> +		 */
> +		long limit = spool->min_hpages - spool->used_hpages;
> +
> +		if (spool->rsv_hpages + delta <= limit)
>  			ret = 0;
>  		else
> -			ret = spool->rsv_hpages + delta - spool->min_hpages;
> +			ret = spool->rsv_hpages + delta - limit;
>
>  		spool->rsv_hpages += delta;
> -		if (spool->rsv_hpages > spool->min_hpages)
> -			spool->rsv_hpages = spool->min_hpages;
> +		if (spool->rsv_hpages > limit)
> +			spool->rsv_hpages = limit;
>  	}
>
>  	/*
> @@ -6542,7 +6555,7 @@ long hugetlb_reserve_pages(struct inode *inode,
>  		struct vm_area_struct *vma,
>  		vma_flags_t vma_flags)
>  {
> -	long chg = -1, add = -1, spool_resv, gbl_resv;
> +	long chg = -1, add = -1, gbl_resv;
>  	struct hstate *h = hstate_inode(inode);
>  	struct hugepage_subpool *spool = subpool_inode(inode);
>  	struct resv_map *resv_map;
> @@ -6622,9 +6635,9 @@ long hugetlb_reserve_pages(struct inode *inode,
>  	 * the subpool has a minimum size, there may be some global
>  	 * reservations already in place (gbl_reserve).
>  	 */
> -	gbl_reserve = hugepage_subpool_get_pages(spool, chg);
> -	if (gbl_reserve < 0) {
> -		err = gbl_reserve;
> +	gbl_resv = hugepage_subpool_get_pages(spool, chg);
> +	if (gbl_resv < 0) {
> +		err = gbl_resv;
>  		goto out_uncharge_cgroup;
>  	}
>
> @@ -6632,7 +6645,7 @@ long hugetlb_reserve_pages(struct inode *inode,
>  	 * Check enough hugepages are available for the reservation.
>  	 * Hand the pages back to the subpool if there are not
>  	 */
> -	err = hugetlb_acct_memory(h, gbl_reserve);
> +	err = hugetlb_acct_memory(h, gbl_resv);
>  	if (err < 0)
>  		goto out_put_pages;
>
> @@ -6651,7 +6664,7 @@ long hugetlb_reserve_pages(struct inode *inode,
>  		add = region_add(resv_map, from, to, regions_needed, h, h_cg);
>
>  		if (unlikely(add < 0)) {
> -			hugetlb_acct_memory(h, -gbl_reserve);
> +			hugetlb_acct_memory(h, -gbl_resv);
>  			err = add;
>  			goto out_put_pages;
>  		} else if (unlikely(chg > add)) {
> @@ -6687,26 +6700,21 @@ long hugetlb_reserve_pages(struct inode *inode,
>  	}
>  	return chg;
>
> -out_put_pages:
> -	spool_resv = chg - gbl_reserve;
> -	if (spool_resv) {
> -		/* put sub pool's reservation back, chg - gbl_reserve */
> -		gbl_resv = hugepage_subpool_put_pages(spool, spool_resv);
> -		/*
> -		 * subpool's reserved pages can not be put back due to race,
> -		 * return to hstate.
> -		 */
> -		hugetlb_acct_memory(h, -gbl_resv);
> -	}
> -	/* Restore used_hpages for pages that failed global reservation */
> -	if (gbl_reserve && spool) {
> -		unsigned long flags;
> + out_put_pages:
> +	/*
> +	 * Return all that was requested from the subpool, let subpool
> +	 * tell us the new number of reservations that need to be
> +	 * returned to the global pool.
> +	 */
> +	gbl_reserve = hugepage_subpool_put_pages(spool, chg);
> +	/*
> +	 * There may be a difference between the number of
> +	 * reservations to consume and the number to restore now if
> +	 * there are multiple threads interacting with the subpool -
> +	 * restore the difference.
> +	 */
> +	hugetlb_acct_memory(h, gbl_resv - gbl_reserve);

Sashiko had a few comments on this patch, the following is the only one
that isn't addressed in a later patch in this same series:

  [Severity: Critical]
  Does this code leave the subpool with phantom global reservations if
  hugetlb_acct_memory() fails?

  If hugetlb_acct_memory() is called with a positive delta here (meaning the
  subpool absorbed some reservations) and it returns -ENOMEM because the global
  pool is exhausted, the return value is ignored.

  Could this artificially inflate the subpool's rsv_hpages without backing
  global pages, which would then inflate free_huge_pages out of thin air when
  the subpool is eventually destroyed?

This is true, but this is a failure path and I'm not really sure how
best to handle an error from hugetlb_acct_memory() in this case. Other
callers of hugetlb_acct_memory() don't handle failures either.


>
> -		spin_lock_irqsave(&spool->lock, flags);
> -		if (spool->max_hpages != -1)
> -			spool->used_hpages -= gbl_reserve;
> -		unlock_or_release_subpool(spool, flags);
> -	}
>  out_uncharge_cgroup:
>  	hugetlb_cgroup_uncharge_cgroup_rsvd(hstate_index(h),
>  					    chg * pages_per_huge_page(h), h_cg);
>
> --
> 2.55.0.229.g6434b31f56-goog