From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from smtp.kernel.org (aws-us-west-2-korg-mail-1.web.codeaurora.org [10.30.226.201]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 0BCAE36165C; Fri, 27 Feb 2026 17:08:11 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=10.30.226.201 ARC-Seal:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1772212092; cv=none; b=Lk50gQZlC1MNqilnEh6q5M5w0C1jTNwLqmxN+TmWKFsGq+mQCkYBtBOYU9NibbjmPCvm/fKZ2RiQuRQqgJYiZbE52udnHWkHT8qQ/IouwUwosTDSwXFhrIInfY53SQQM1f4+hQhwCfFOFmGz47iAfA/xWfU12DVijgokOiOshMQ= ARC-Message-Signature:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1772212092; c=relaxed/simple; bh=WUPDRrdwa5SkEyMU6GHPLmb001wCRNzEwHKZ7p9JCcI=; h=From:Date:Subject:MIME-Version:Content-Type:Message-Id:References: In-Reply-To:To:Cc; b=FJeh/zX0lOVH+QkIvt0QENqjYAN0DeYLO7L6cYKCViCWNAg2Kype/fH0waSJmSaYBm8S8D626g7ug/EeWIzYS9/2MQ9fjX/8t3+VAHMcEC/RW3Ut7tCpoK+DhyzoQcn8jZif1lf56UIcfk2t1eITjONQGO5RFVn63FwcPJzYb4M= ARC-Authentication-Results:i=1; smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=kernel.org header.i=@kernel.org header.b=nIusV8/z; arc=none smtp.client-ip=10.30.226.201 Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=kernel.org header.i=@kernel.org header.b="nIusV8/z" Received: by smtp.kernel.org (Postfix) with ESMTPSA id A0FF9C2BC87; Fri, 27 Feb 2026 17:08:08 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=kernel.org; s=k20201202; t=1772212091; bh=WUPDRrdwa5SkEyMU6GHPLmb001wCRNzEwHKZ7p9JCcI=; h=From:Date:Subject:References:In-Reply-To:To:Cc:From; b=nIusV8/zKNisDdm7CqRh9RiJxfIUV+V+BJVzPXzF7x2LZZPgK/DyfuzonpoBJ/jCy hOrK9MB3PDZ0UBi2NiR4TnW9Hd05XeDJELywAS/IlR7w42XZUNZ2TMi8B4I3lfA4d5 1VeTqLhlgEyGa60Z1/FLBMKuFUhPRjcuPXs8BVqWlU0nWeSsEJrGReNa29OgzzcRwV aSpVrWoysuIiAGO35s1hviNh4nCqGpY7ACZlLbXk/CnN4RUabZ6YwlQiMYiBZR4cbu O+dpm/uNsffyVps9Fgk1tw1A3WIXbEKv0nigd2zJs31u4ChCEHGhPjtP5aIZTCH7UE SJpCZZ+9ObdPQ== From: Vlastimil Babka Date: Fri, 27 Feb 2026 18:07:59 +0100 Subject: [PATCH 2/3] mm/page_alloc: remove IRQ saving/restoring from pcp locking Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: 7bit Message-Id: <20260227-b4-pcp-locking-cleanup-v1-2-f7e22e603447@kernel.org> References: <20260227-b4-pcp-locking-cleanup-v1-0-f7e22e603447@kernel.org> In-Reply-To: <20260227-b4-pcp-locking-cleanup-v1-0-f7e22e603447@kernel.org> To: Andrew Morton , Suren Baghdasaryan , Michal Hocko , Brendan Jackman , Johannes Weiner , Zi Yan Cc: Mel Gorman , Matthew Wilcox , "David Hildenbrand (Arm)" , Sebastian Andrzej Siewior , Clark Williams , Steven Rostedt , linux-mm@kvack.org, linux-kernel@vger.kernel.org, linux-rt-devel@lists.linux.dev, "Vlastimil Babka (SUSE)" X-Mailer: b4 0.14.3 Effectively revert commit 038a102535eb ("mm/page_alloc: prevent pcp corruption with SMP=n"). The original problem is now avoided by pcp_spin_trylock() always failing on CONFIG_SMP=n, so we do not need to disable IRQs anymore. It's not a complete revert, because keeping the pcp_spin_(un)lock() wrappers is useful. Rename them from _maybe_irqsave/restore to _nopin. The difference from pcp_spin_trylock()/pcp_spin_unlock() is that the _nopin variants don't perform pcpu_task_pin/unpin(). Signed-off-by: Vlastimil Babka (SUSE) --- mm/page_alloc.c | 46 ++++++++++++++++------------------------------ 1 file changed, 16 insertions(+), 30 deletions(-) diff --git a/mm/page_alloc.c b/mm/page_alloc.c index 65efcaeb8800..8e5b30adfe40 100644 --- a/mm/page_alloc.c +++ b/mm/page_alloc.c @@ -155,24 +155,14 @@ static DEFINE_MUTEX(pcp_batch_high_lock); BUG_ON(1) #endif -#if defined(CONFIG_SMP) || defined(CONFIG_PREEMPT_RT) -static inline void __flags_noop(unsigned long *flags) { } -#define pcp_spin_lock_maybe_irqsave(ptr, flags) \ -({ \ - __flags_noop(&(flags)); \ - spin_lock(&(ptr)->lock); \ -}) -#define pcp_spin_unlock_maybe_irqrestore(ptr, flags) \ -({ \ - spin_unlock(&(ptr)->lock); \ - __flags_noop(&(flags)); \ -}) -#else -#define pcp_spin_lock_maybe_irqsave(ptr, flags) \ - spin_lock_irqsave(&(ptr)->lock, flags) -#define pcp_spin_unlock_maybe_irqrestore(ptr, flags) \ - spin_unlock_irqrestore(&(ptr)->lock, flags) -#endif +/* + * In some cases we do not need to pin the task to the CPU because we are + * already given a specific cpu's pcp pointer. + */ +#define pcp_spin_lock_nopin(ptr) \ + spin_lock(&(ptr)->lock) +#define pcp_spin_unlock_nopin(ptr) \ + spin_unlock(&(ptr)->lock) #ifdef CONFIG_USE_PERCPU_NUMA_NODE_ID DEFINE_PER_CPU(int, numa_node); @@ -2572,7 +2562,6 @@ static int rmqueue_bulk(struct zone *zone, unsigned int order, bool decay_pcp_high(struct zone *zone, struct per_cpu_pages *pcp) { int high_min, to_drain, to_drain_batched, batch; - unsigned long UP_flags; bool todo = false; high_min = READ_ONCE(pcp->high_min); @@ -2592,9 +2581,9 @@ bool decay_pcp_high(struct zone *zone, struct per_cpu_pages *pcp) to_drain = pcp->count - pcp->high; while (to_drain > 0) { to_drain_batched = min(to_drain, batch); - pcp_spin_lock_maybe_irqsave(pcp, UP_flags); + pcp_spin_lock_nopin(pcp); free_pcppages_bulk(zone, to_drain_batched, pcp, 0); - pcp_spin_unlock_maybe_irqrestore(pcp, UP_flags); + pcp_spin_unlock_nopin(pcp); todo = true; to_drain -= to_drain_batched; @@ -2611,15 +2600,14 @@ bool decay_pcp_high(struct zone *zone, struct per_cpu_pages *pcp) */ void drain_zone_pages(struct zone *zone, struct per_cpu_pages *pcp) { - unsigned long UP_flags; int to_drain, batch; batch = READ_ONCE(pcp->batch); to_drain = min(pcp->count, batch); if (to_drain > 0) { - pcp_spin_lock_maybe_irqsave(pcp, UP_flags); + pcp_spin_lock_nopin(pcp); free_pcppages_bulk(zone, to_drain, pcp, 0); - pcp_spin_unlock_maybe_irqrestore(pcp, UP_flags); + pcp_spin_unlock_nopin(pcp); } } #endif @@ -2630,11 +2618,10 @@ void drain_zone_pages(struct zone *zone, struct per_cpu_pages *pcp) static void drain_pages_zone(unsigned int cpu, struct zone *zone) { struct per_cpu_pages *pcp = per_cpu_ptr(zone->per_cpu_pageset, cpu); - unsigned long UP_flags; int count; do { - pcp_spin_lock_maybe_irqsave(pcp, UP_flags); + pcp_spin_lock_nopin(pcp); count = pcp->count; if (count) { int to_drain = min(count, @@ -2643,7 +2630,7 @@ static void drain_pages_zone(unsigned int cpu, struct zone *zone) free_pcppages_bulk(zone, to_drain, pcp, 0); count -= to_drain; } - pcp_spin_unlock_maybe_irqrestore(pcp, UP_flags); + pcp_spin_unlock_nopin(pcp); } while (count); } @@ -6127,7 +6114,6 @@ static void zone_pcp_update_cacheinfo(struct zone *zone, unsigned int cpu) { struct per_cpu_pages *pcp; struct cpu_cacheinfo *cci; - unsigned long UP_flags; pcp = per_cpu_ptr(zone->per_cpu_pageset, cpu); cci = get_cpu_cacheinfo(cpu); @@ -6138,12 +6124,12 @@ static void zone_pcp_update_cacheinfo(struct zone *zone, unsigned int cpu) * This can reduce zone lock contention without hurting * cache-hot pages sharing. */ - pcp_spin_lock_maybe_irqsave(pcp, UP_flags); + pcp_spin_lock_nopin(pcp); if ((cci->per_cpu_data_slice_size >> PAGE_SHIFT) > 3 * pcp->batch) pcp->flags |= PCPF_FREE_HIGH_BATCH; else pcp->flags &= ~PCPF_FREE_HIGH_BATCH; - pcp_spin_unlock_maybe_irqrestore(pcp, UP_flags); + pcp_spin_unlock_nopin(pcp); } void setup_pcp_cacheinfo(unsigned int cpu) -- 2.53.0