From nobody Wed Apr 1 08:15:39 2026 Received: from mail-lf1-f49.google.com (mail-lf1-f49.google.com [209.85.167.49]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 982F444DB69 for ; Tue, 31 Mar 2026 20:23:56 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=209.85.167.49 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1774988638; cv=none; b=eN90kfpVYGEM3HsV7vFu9Wfv/nI/DhMEwCKWy14J2OEXlgZiKiGNT+p/mbaL9dDfVvgykHrCQP/AKYmnPRCbPPtitP+26lGK0ENvaVsqQoYDChOtx//5dGxFMzK4iaEZkDvLpYfteC5FcjSHfm3iAJ6dF6u4WX35lnL/44l+Eqc= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1774988638; c=relaxed/simple; bh=+t8ikSPRD7DfscZ218WtPN+EojX0PBKH28/1MfyKklk=; h=From:To:Cc:Subject:Date:Message-ID:MIME-Version; b=YQnUUEhLQ5voCcylIq5l7ybAOFeG2uQDENzu7mQHw5VRF5EPnbmYWqiKyBh7lQXyNbGMAJm8xMHH2WerhFrbE45SLyro1sE7yHZrWUIMWx650itdG6nNV6E5/dnw17KkBjYW5LRkuEatd9ENstTmSx36mPg8FtczGX/74L6Qpeo= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=gmail.com; spf=pass smtp.mailfrom=gmail.com; dkim=pass (2048-bit key) header.d=gmail.com header.i=@gmail.com header.b=VaTNlEKL; arc=none smtp.client-ip=209.85.167.49 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=gmail.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=gmail.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=gmail.com header.i=@gmail.com header.b="VaTNlEKL" Received: by mail-lf1-f49.google.com with SMTP id 2adb3069b0e04-5a13e1cfa45so7272160e87.2 for ; Tue, 31 Mar 2026 13:23:56 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20251104; t=1774988635; x=1775593435; darn=vger.kernel.org; h=content-transfer-encoding:mime-version:message-id:date:subject:cc :to:from:from:to:cc:subject:date:message-id:reply-to; bh=gp8MmgexMa9kZc3iHEPrKAW8s9Qb4eG8gKB+d4t8Qx0=; b=VaTNlEKLnjwgivCe5NJq5TBgpbOPGyXu0oPOD3aJIeuNkP2wl626AcHc4XO2UJhyNV HF3KA2mOgnRfN3jTUxUji9dbYq6/CBPFQ+1v+xyBEsJjnP5m+7TtAiFBfYZIPV40ZCDm V6d240KUTif+UCw/al083cdw9tbDxcYDeONM4kmcGqV5cgTYSCueMC//Suyim8+CotKY lshW3N5JkRd7sC1cwlec4MgpkmMZLpXb256+pI/AZnGlx6pR3ZQmaQidFp+tGEFcKWEg HESkQKE+mO2jK+pf31yf43cpzf4zmopN3Nef7TEh1zW5eaJpNtTzinusDJ9Ts7Rfq1YI vJAA== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20251104; t=1774988635; x=1775593435; h=content-transfer-encoding:mime-version:message-id:date:subject:cc :to:from:x-gm-gg:x-gm-message-state:from:to:cc:subject:date :message-id:reply-to; bh=gp8MmgexMa9kZc3iHEPrKAW8s9Qb4eG8gKB+d4t8Qx0=; b=qFhZHJ6VxLyRttlVXg+BAoFEesJB0cwZRVSmrmID9Fb1ZB1mkvhcdTkLSRp/eUbgyi kUho0ORXma2TION46+KRQzgW1+IvjOz3GGxKJAGk+cyx50eH5NejWUPEB9iRoJPztMCa JesTNSZX8Zoo3RyRa0QTUvjEEJa44zB+Gn4eCIDL3ok+Z7usZ5AVlIW/wLy8dBYhR0Xq QIMl8g2fL6sNaM8fLP9O/w/T25rtkblfac99GHihQ6r9AV/gfje3tlpEMIEeQTJrkOqY 11JM+e+Mio2T4FGxwiWZv3yuFmHN7xX777Co5+eTIK5VVzbmmncQGzL5hcMx2zmfjUAj 3TUA== X-Forwarded-Encrypted: i=1; AJvYcCWNYPYcf2dBQHaj+tE/rdebLwO+kK0va8liTNZ0aaH0CipUG4sjmHkzrBhimwnNDdBm1FLt/2LV2g+vWek=@vger.kernel.org X-Gm-Message-State: AOJu0YwkAGtDiTLsfTpXupUHJB8Slf0d91ByOFIlznWdkMouznkGhyBA U4a0ZbqvKMeFlMBiAMX6QyrqjJ9tFK8MhIlbvSuDNp1I07PiISC/KYSO X-Gm-Gg: ATEYQzyfqJ8nYQDbz0Dtf89vnUlg20Psp6iKEvExDEw9zxlKhfKe580XJ4r0UfvEe5r mHgi1SFPEUihXo2mNLamHbCclM3E+xxZ1rJB4uLYBIFgHWnsalj8qB/QdjryPH4DKAgpliyS6ZC SUaR5ZCLJzY/YtQIB5HsbkNXWGKCp8ojMerOveOtxqG/h5BskMPvIghhlsPYsC6IFO741Pif0YP E12hEmBaTavkZiOz2+KIhze5/IgIldN2nnlNvdEt8LLeBfD5LCmX19D3xxALdrhUoGRh/bbCW8h YNDXUCrY5B33AmzgL10Xb0NBE3JGpwBFSbJU3N3mCW4EoYJNagquiVRsdSDVgc7D7JPxiEyXrm5 u2Vhr6ecLsy8tJlvo4FfuEyCzQyDwNh3u+NzFNYrW7ptF/JUqSytUTG14weuOg1B1uA6/15W4Z6 P9BsWcWhNFhcuqzhQ= X-Received: by 2002:a05:6512:1291:b0:5a2:7a31:9194 with SMTP id 2adb3069b0e04-5a2c1ef9aa4mr263563e87.19.1774988634449; Tue, 31 Mar 2026 13:23:54 -0700 (PDT) Received: from localhost.localdomain ([2001:9b1:d5a0:a500::24b]) by smtp.gmail.com with ESMTPSA id 2adb3069b0e04-5a2b13f4329sm2681171e87.3.2026.03.31.13.23.53 (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256); Tue, 31 Mar 2026 13:23:54 -0700 (PDT) From: "Uladzislau Rezki (Sony)" To: linux-mm@kvack.org, Andrew Morton Cc: Baoquan He , LKML , Uladzislau Rezki , stable@vger.kernel.org, lirongqing Subject: [PATCH v3] mm/vmalloc: Use dedicated unbound workqueues for vmap drain Date: Tue, 31 Mar 2026 22:23:52 +0200 Message-ID: <20260331202352.879718-1-urezki@gmail.com> X-Mailer: git-send-email 2.47.3 Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset="utf-8" drain_vmap_area_work() function can take >10ms to complete when there are many accumulated vmap areas in a system with high CPU count, causing workqueue watchdog warnings when run via schedule_work(): workqueue: drain_vmap_area_work hogged CPU for >10000us Move the top-level drain work to a dedicated WQ_UNBOUND workqueue so the scheduler can run this background work on any available CPU, improving responsiveness. Use the WQ_MEM_RECLAIM to ensure forward progress under memory pressure. Move purge helpers to separate WQ_UNBOUND | WQ_MEM_RECLAIM workqueue. This allows drain_vmap_work to wait for helpers completion without creating dependency on the same rescuer thread and avoid a potential parent/child deadlock. Simplify purge helper scheduling by removing cpumask-based iteration to iterating directly over vmap nodes checking work_queued state. Cc: stable@vger.kernel.org Cc: lirongqing Fixes: 72210662c5a2 ("mm: vmalloc: offload free_vmap_area_lock lock") Link: https://lore.kernel.org/all/20260319074307.2325-1-lirongqing@baidu.co= m/ Signed-off-by: Uladzislau Rezki (Sony) --- mm/vmalloc.c | 79 ++++++++++++++++++++++++++++++++++------------------ 1 file changed, 52 insertions(+), 27 deletions(-) diff --git a/mm/vmalloc.c b/mm/vmalloc.c index 61caa55a4402..0fa1208a910b 100644 --- a/mm/vmalloc.c +++ b/mm/vmalloc.c @@ -949,6 +949,7 @@ static struct vmap_node { struct list_head purge_list; struct work_struct purge_work; unsigned long nr_purged; + bool work_queued; } single; =20 /* @@ -1067,6 +1068,8 @@ static void reclaim_and_purge_vmap_areas(void); static BLOCKING_NOTIFIER_HEAD(vmap_notify_list); static void drain_vmap_area_work(struct work_struct *work); static DECLARE_WORK(drain_vmap_work, drain_vmap_area_work); +static struct workqueue_struct *drain_vmap_helpers_wq; +static struct workqueue_struct *drain_vmap_wq; =20 static __cacheline_aligned_in_smp atomic_long_t nr_vmalloc_pages; static __cacheline_aligned_in_smp atomic_long_t vmap_lazy_nr; @@ -2335,6 +2338,16 @@ static void purge_vmap_node(struct work_struct *work) reclaim_list_global(&local_list); } =20 +static bool +schedule_drain_vmap_work(struct workqueue_struct *wq, + struct work_struct *work) +{ + if (wq) + return queue_work(wq, work); + + return false; +} + /* * Purges all lazily-freed vmap areas. */ @@ -2342,19 +2355,12 @@ static bool __purge_vmap_area_lazy(unsigned long st= art, unsigned long end, bool full_pool_decay) { unsigned long nr_purged_areas =3D 0; + unsigned int nr_purge_nodes =3D 0; unsigned int nr_purge_helpers; - static cpumask_t purge_nodes; - unsigned int nr_purge_nodes; struct vmap_node *vn; - int i; =20 lockdep_assert_held(&vmap_purge_lock); =20 - /* - * Use cpumask to mark which node has to be processed. - */ - purge_nodes =3D CPU_MASK_NONE; - for_each_vmap_node(vn) { INIT_LIST_HEAD(&vn->purge_list); vn->skip_populate =3D full_pool_decay; @@ -2374,10 +2380,9 @@ static bool __purge_vmap_area_lazy(unsigned long sta= rt, unsigned long end, end =3D max(end, list_last_entry(&vn->purge_list, struct vmap_area, list)->va_end); =20 - cpumask_set_cpu(node_to_id(vn), &purge_nodes); + nr_purge_nodes++; } =20 - nr_purge_nodes =3D cpumask_weight(&purge_nodes); if (nr_purge_nodes > 0) { flush_tlb_kernel_range(start, end); =20 @@ -2385,29 +2390,31 @@ static bool __purge_vmap_area_lazy(unsigned long st= art, unsigned long end, nr_purge_helpers =3D atomic_long_read(&vmap_lazy_nr) / lazy_max_pages(); nr_purge_helpers =3D clamp(nr_purge_helpers, 1U, nr_purge_nodes) - 1; =20 - for_each_cpu(i, &purge_nodes) { - vn =3D &vmap_nodes[i]; + for_each_vmap_node(vn) { + vn->work_queued =3D false; + + if (list_empty(&vn->purge_list)) + continue; =20 if (nr_purge_helpers > 0) { INIT_WORK(&vn->purge_work, purge_vmap_node); + vn->work_queued =3D schedule_drain_vmap_work( + READ_ONCE(drain_vmap_helpers_wq), &vn->purge_work); =20 - if (cpumask_test_cpu(i, cpu_online_mask)) - schedule_work_on(i, &vn->purge_work); - else - schedule_work(&vn->purge_work); - - nr_purge_helpers--; - } else { - vn->purge_work.func =3D NULL; - purge_vmap_node(&vn->purge_work); - nr_purged_areas +=3D vn->nr_purged; + if (vn->work_queued) { + nr_purge_helpers--; + continue; + } } - } =20 - for_each_cpu(i, &purge_nodes) { - vn =3D &vmap_nodes[i]; + /* Sync path. Process locally. */ + purge_vmap_node(&vn->purge_work); + nr_purged_areas +=3D vn->nr_purged; + } =20 - if (vn->purge_work.func) { + /* Wait for completion if queued any. */ + for_each_vmap_node(vn) { + if (vn->work_queued) { flush_work(&vn->purge_work); nr_purged_areas +=3D vn->nr_purged; } @@ -2471,7 +2478,8 @@ static void free_vmap_area_noflush(struct vmap_area *= va) =20 /* After this point, we may free va at any time */ if (unlikely(nr_lazy > nr_lazy_max)) - schedule_work(&drain_vmap_work); + schedule_drain_vmap_work(READ_ONCE(drain_vmap_wq), + &drain_vmap_work); } =20 /* @@ -5483,3 +5491,20 @@ void __init vmalloc_init(void) vmap_node_shrinker->scan_objects =3D vmap_node_shrink_scan; shrinker_register(vmap_node_shrinker); } + +static int __init vmalloc_init_workqueue(void) +{ + struct workqueue_struct *drain_wq, *helpers_wq; + unsigned int flags =3D WQ_UNBOUND | WQ_MEM_RECLAIM; + + drain_wq =3D alloc_workqueue("vmap_drain", flags, 0); + WARN_ON_ONCE(drain_wq =3D=3D NULL); + WRITE_ONCE(drain_vmap_wq, drain_wq); + + helpers_wq =3D alloc_workqueue("vmap_drain_helpers", flags, 0); + WARN_ON_ONCE(helpers_wq =3D=3D NULL); + WRITE_ONCE(drain_vmap_helpers_wq, helpers_wq); + + return 0; +} +early_initcall(vmalloc_init_workqueue); --=20 2.47.3