Message-ID: <dfdd0648-a7da-4e8e-8781-cf094da4844f@suse.cz>
Date: Thu, 22 Jan 2026 13:22:26 +0100
From: Vlastimil Babka <vbabka@...e.cz>
To: Qiliang Yuan <realwujing@...il.com>, akpm@...ux-foundation.org
Cc: david@...nel.org, mhocko@...e.com, willy@...radead.org,
 lance.yang@...ux.dev, hannes@...xchg.org, surenb@...gle.com,
 jackmanb@...gle.com, ziy@...dia.com, weixugc@...gle.com, rppt@...nel.org,
 linux-mm@...ck.org, linux-kernel@...r.kernel.org, netdev@...r.kernel.org,
 edumazet@...gle.com, jis1@...natelecom.cn, wangh13@...natelecom.cn,
 liyi1@...natelecom.cn, sunshx@...natelecom.cn, zhangzq20@...natelecom.cn,
 zhangjn11@...natelecom.cn, Qiliang Yuan <yuanql9@...natelecom.cn>
Subject: Re: [PATCH v6] mm/page_alloc: boost watermarks on atomic allocation
 failure

On 1/22/26 03:07, Qiliang Yuan wrote:
> Atomic allocations (GFP_ATOMIC) are prone to failure under heavy memory
> pressure as they cannot enter direct reclaim. This patch introduces a
> 'Soft Boost' mechanism to mitigate this.
> 
> When a GFP_ATOMIC request fails or enters the slowpath, the preferred
> zone's watermark_boost is increased. This triggers kswapd to proactively
> reclaim memory, creating a safety buffer for future atomic bursts.
> 
> To prevent excessive reclaim during packet storms, a 1-second debounce
> timer (last_boost_jiffies) is added to each zone to rate-limit boosts.
> 
> This approach reuses existing watermark_boost infrastructure, ensuring
> minimal overhead and asynchronous background reclaim via kswapd.
> 
> Allocation failure logs:
> [38535644.718700] node 0: slabs: 1031, objs: 43328, free: 0
> [38535644.725059] node 1: slabs: 339, objs: 17616, free: 317
> [38535645.428345] SLUB: Unable to allocate memory on node -1, gfp=0x480020(GFP_ATOMIC)
> [38535645.436888] cache: skbuff_head_cache, object size: 232, buffer size: 256, default order: 2, min order: 0
> [38535645.447664] node 0: slabs: 940, objs: 40864, free: 144
> [38535645.454026] node 1: slabs: 322, objs: 19168, free: 383
> [38535645.556122] SLUB: Unable to allocate memory on node -1, gfp=0x480020(GFP_ATOMIC)
> [38535645.564576] cache: skbuff_head_cache, object size: 232, buffer size: 256, default order: 2, min order: 0
> [38535649.655523] warn_alloc: 59 callbacks suppressed
> [38535649.655527] swapper/100: page allocation failure: order:0, mode:0x480020(GFP_ATOMIC), nodemask=(null)
> [38535649.671692] swapper/100 cpuset=/ mems_allowed=0-1
> 
> Signed-off-by: Qiliang Yuan <realwujing@...il.com>
> Signed-off-by: Qiliang Yuan <yuanql9@...natelecom.cn>
> ---
> v6:
>   - Replace magic number ">> 10" with ATOMIC_BOOST_SCALE_SHIFT define
>   - Add documentation explaining 0.1% zone size boost rationale
> v5:
>   - Simplify to use native boost_watermark() instead of custom logic
> v4:
>   - Add watermark_scale_boost and gradual decay via balance_pgdat
> v3:
>   - Move debounce timer to per-zone; optimize zone selection
> v2:
>   - Add debounce logic and zone-proportional boosting
> v1:
>   - Initial: boost min_free_kbytes on GFP_ATOMIC failure
> ---
>  include/linux/mmzone.h |  1 +
>  mm/page_alloc.c        | 36 +++++++++++++++++++++++++++++++++++-
>  2 files changed, 36 insertions(+), 1 deletion(-)
> ---
>  include/linux/mmzone.h |  1 +
>  mm/page_alloc.c        | 36 +++++++++++++++++++++++++++++++++++-
>  2 files changed, 36 insertions(+), 1 deletion(-)
> 
> diff --git a/include/linux/mmzone.h b/include/linux/mmzone.h
> index 75ef7c9f9307..8e37e4e6765b 100644
> --- a/include/linux/mmzone.h
> +++ b/include/linux/mmzone.h
> @@ -882,6 +882,7 @@ struct zone {
>  	/* zone watermarks, access with *_wmark_pages(zone) macros */
>  	unsigned long _watermark[NR_WMARK];
>  	unsigned long watermark_boost;
> +	unsigned long last_boost_jiffies;
>  
>  	unsigned long nr_reserved_highatomic;
>  	unsigned long nr_free_highatomic;
> diff --git a/mm/page_alloc.c b/mm/page_alloc.c
> index c380f063e8b7..8ea2435125d5 100644
> --- a/mm/page_alloc.c
> +++ b/mm/page_alloc.c
> @@ -218,6 +218,13 @@ unsigned int pageblock_order __read_mostly;
>  static void __free_pages_ok(struct page *page, unsigned int order,
>  			    fpi_t fpi_flags);
>  
> +/*
> + * Boost watermarks by ~0.1% of zone size on atomic allocation pressure.
> + * This provides zone-proportional safety buffers: ~1MB per 1GB of zone size.
> + * Larger zones under GFP_ATOMIC pressure need proportionally larger reserves.
> + */
> +#define ATOMIC_BOOST_SCALE_SHIFT 10
> +
>  /*
>   * results with 256, 32 in the lowmem_reserve sysctl:
>   *	1G machine -> (16M dma, 800M-16M normal, 1G-800M high)
> @@ -2189,12 +2196,31 @@ static inline bool boost_watermark(struct zone *zone)
>  
>  	max_boost = max(pageblock_nr_pages, max_boost);
>  
> -	zone->watermark_boost = min(zone->watermark_boost + pageblock_nr_pages,
> +	zone->watermark_boost = min(zone->watermark_boost +
> +		max(pageblock_nr_pages, zone_managed_pages(zone) >> ATOMIC_BOOST_SCALE_SHIFT),

So IIUC you are not changing (increasing) the maximum boost, but the amount
in one step. It would be more descriptive to first set a local variable with
this amount and then use it for the boosting.

This change also affects the original boost_watermark() caller. Maybe it's
fine, can't say without any measurements.

>  		max_boost);
>  
>  	return true;
>  }
>  
> +static void boost_zones_for_atomic(struct alloc_context *ac, gfp_t gfp_mask)
> +{
> +	struct zoneref *z;
> +	struct zone *zone;
> +	unsigned long now = jiffies;
> +
> +	for_each_zone_zonelist(zone, z, ac->zonelist, ac->highest_zoneidx) {
> +		/* 1 second debounce to avoid spamming boosts in a burst */
> +		if (time_after(now, zone->last_boost_jiffies + HZ)) {
> +			zone->last_boost_jiffies = now;
> +			if (boost_watermark(zone))
> +				wakeup_kswapd(zone, gfp_mask, 0, ac->highest_zoneidx);

The other caller of boost_watermark() is under zone->lock and it makes those
zone->watermark_boost increments safe, and balance_pgdat() takes it for the
decrements too with "/* Increments are under the zone lock */ " comment,
otherwise I wouldn't realize this.

It probably wouldn't hurt to add a lockdep assert into boost_watermark() to
prevent mistakes.

But the other caller also takes care not to call wakeup_kswapd() under the
zone lock so I would not do it as well - see commit 73444bc4d8f92

> +			/* Only boost the preferred zone to be precise */
> +			break;
> +		}
> +	}
> +}
> +
>  /*
>   * When we are falling back to another migratetype during allocation, should we
>   * try to claim an entire block to satisfy further allocations, instead of
> @@ -4742,6 +4768,10 @@ __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order,
>  	if (page)
>  		goto got_pg;
>  
> +	/* Proactively boost for atomic requests entering slowpath */
> +	if ((gfp_mask & GFP_ATOMIC) && order == 0)
> +		boost_zones_for_atomic(ac, gfp_mask);
> +
>  	/*
>  	 * For costly allocations, try direct compaction first, as it's likely
>  	 * that we have enough base pages and don't need to reclaim. For non-
> @@ -4947,6 +4977,10 @@ __alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order,
>  		goto retry;
>  	}
>  fail:
> +	/* Boost watermarks on atomic allocation failure to trigger kswapd */
> +	if (unlikely(page == NULL && (gfp_mask & GFP_ATOMIC) && order == 0))
> +		boost_zones_for_atomic(ac, gfp_mask);

We already did the boosting when entering slowpath, there's 1 second
debounce and GFP_ATOMIC can't really do anything in the slowpath to spend 1
second, so I think this is redundant.

> +
>  	warn_alloc(gfp_mask, ac->nodemask,
>  			"page allocation failure: order:%u", order);
>  got_pg:

Powered by blists - more mailing lists