From nobody Thu Apr 2 15:36:11 2026 Received: from mail-dy1-f202.google.com (mail-dy1-f202.google.com [74.125.82.202]) (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 63D0C395268 for ; Fri, 27 Mar 2026 20:55:11 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=74.125.82.202 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1774644913; cv=none; b=Z7bVwXsNzy6cVCUG/2T95Zl4ozinlid1k2fTmQCgnHPYE08gN72WloD6fHIQS1JyVBxYBEGucdKD4FTTLyx+CAyZ84LeOr0Ojhux8ssO5DpCnYY24dgPoXXnkhB5Jw0C+aZ64TQpq3mrQYt0x7Gup6IAI7cyu7A9ZjWojxpoUjI= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1774644913; c=relaxed/simple; bh=d5mnzxn4Tjwdwq6SneMA/zGXA7gBgTdol5RbAikXVvg=; h=Date:In-Reply-To:Mime-Version:References:Message-ID:Subject:From: To:Cc:Content-Type; b=rxXc1QbjOmDDfBwQ/Awl1fyzSO+F5sdwn1uNoYPG1JSrSS4c+AEwHDV2Ga8qK5Hg6wXgYcB+G20WLGCBQO36HnQAxNsOHdHUNGVkx4ARKw41E4dQD+VoJAXR1hESzaZOhKjWSMq60OrOVZ0jHOgSbHa1hFBRt/PPjh3BDYeRdoU= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com; spf=pass smtp.mailfrom=flex--surenb.bounces.google.com; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b=g5uZucig; arc=none smtp.client-ip=74.125.82.202 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=google.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=flex--surenb.bounces.google.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b="g5uZucig" Received: by mail-dy1-f202.google.com with SMTP id 5a478bee46e88-2ba9a744f7dso2551578eec.0 for ; Fri, 27 Mar 2026 13:55:11 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20251104; t=1774644911; x=1775249711; darn=vger.kernel.org; h=cc:to:from:subject:message-id:references:mime-version:in-reply-to :date:from:to:cc:subject:date:message-id:reply-to; bh=6skjtd7OxVmiwFu5TmEIQZKDuU+5vel5eIP7VGPQHkg=; b=g5uZucigZDwE7NyCqBEYkRQOXhOpVkYFRISsZ5p0llxgndtPPQxdb5uKgdjiuTwD7Q wQ9/hivbLB8lOfMW+Q/ndrvuEpIZhTzn5w7sgnZq5mT60rBmiOdNLDQ4epGuNK3UQ78g 91wuLODB9gVI3uSkqzYOJ4M1U+uIbeCDIrI7HlXedAQCkfKCFsDfawsbsZznKgjJrtsG 2ev2gmAOV33p8OFFjSJgUW0jvilRtaeyrWSnCm0FH+CxpLCyuyMU5KfLKpst1khE2z6B Sn6ljoHT94m3XV5cVv6KlCy4kZgp2IRjhI+W8jbOCCSaLlaIQ/8ICZKShMBuWICRra7d /noQ== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20251104; t=1774644911; x=1775249711; h=cc:to:from:subject:message-id:references:mime-version:in-reply-to :date:x-gm-message-state:from:to:cc:subject:date:message-id:reply-to; bh=6skjtd7OxVmiwFu5TmEIQZKDuU+5vel5eIP7VGPQHkg=; b=jITQYHZmCSaU2vTOKcTBldTwwJFd/XSF0C29KNctzNhoqkpfZj6tbhdITkASZ2/R84 HHdktm8SS0rZkrZALc6EzdQ08tvA4FZLC+io+3yRDrwrQkBK4AnrUAWB0xt2XfnB/OLB GGtHR1rZ3bYCfYyYT3iGmrsKqBtRGsmPLmAKnNBXLZQHGDoSYkcshrs2gmKDYwTguJIA VBUyuGVe4mKNqFqQB5BcAJYEos1DbM5zLqG8If0Zit6YD3fTli8y5woChQJtZSIkhzMd NuWjGaooGMa83EjCzQcjzCgP6IATWsDTo4uT7sERrwnOCa99NKGxg85Vcn1d6edCEwDg xrcQ== X-Forwarded-Encrypted: i=1; AJvYcCUtHdbAYm7cRlmq1CKrT+5ka3Bsa2FrL9x64yEfwq8MRZVmvNrguHb0PGkOR1W4lk2sp+0abaoCLPl5bbE=@vger.kernel.org X-Gm-Message-State: AOJu0YyS8FG9MafHwQmYRtEb9lc15l/bnzxCA2odGovKgCJLx/7q1HK0 N8Uu1b9SEudJtC/+Yg9dxp8SUEw2K8w/g4V3hcJQvtJHLFYVspnCkBIdpTXi+ojgNnKvw7Eh3FI idUG/Ag== X-Received: from dybrq15.prod.google.com ([2002:a05:7301:468f:b0:2c3:adf7:d755]) (user=surenb job=prod-delivery.src-stubby-dispatcher) by 2002:a05:7300:5b89:b0:2c1:1dea:c443 with SMTP id 5a478bee46e88-2c185e460a2mr2052736eec.19.1774644910254; Fri, 27 Mar 2026 13:55:10 -0700 (PDT) Date: Fri, 27 Mar 2026 13:54:55 -0700 In-Reply-To: <20260327205457.604224-1-surenb@google.com> Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: Mime-Version: 1.0 References: <20260327205457.604224-1-surenb@google.com> X-Mailer: git-send-email 2.53.0.1018.g2bb0e51243-goog Message-ID: <20260327205457.604224-5-surenb@google.com> Subject: [PATCH v6 4/6] mm/vma: use vma_start_write_killable() in vma operations From: Suren Baghdasaryan To: akpm@linux-foundation.org Cc: willy@infradead.org, david@kernel.org, ziy@nvidia.com, matthew.brost@intel.com, joshua.hahnjy@gmail.com, rakie.kim@sk.com, byungchul@sk.com, gourry@gourry.net, ying.huang@linux.alibaba.com, apopple@nvidia.com, ljs@kernel.org, baolin.wang@linux.alibaba.com, Liam.Howlett@oracle.com, npache@redhat.com, ryan.roberts@arm.com, dev.jain@arm.com, baohua@kernel.org, lance.yang@linux.dev, vbabka@suse.cz, jannh@google.com, rppt@kernel.org, mhocko@suse.com, pfalcato@suse.de, kees@kernel.org, maddy@linux.ibm.com, npiggin@gmail.com, mpe@ellerman.id.au, chleroy@kernel.org, borntraeger@linux.ibm.com, frankja@linux.ibm.com, imbrenda@linux.ibm.com, hca@linux.ibm.com, gor@linux.ibm.com, agordeev@linux.ibm.com, svens@linux.ibm.com, gerald.schaefer@linux.ibm.com, linux-mm@kvack.org, linuxppc-dev@lists.ozlabs.org, kvm@vger.kernel.org, linux-kernel@vger.kernel.org, linux-s390@vger.kernel.org, surenb@google.com Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset="utf-8" Replace vma_start_write() with vma_start_write_killable(), improving reaction time to the kill signal. Replace vma_start_write() calls when we operate on VMAs. To propagate errors from vma_merge_existing_range() and vma_expand() we fake an ENOMEM error when we fail due to a pending fatal signal. This is a temporary workaround. Fixing this requires some refactoring and will be done separately in the future. In a number of places we now lock VMA earlier than before to avoid doing work and undoing it later if a fatal signal is pending. This is safe because the moves are happening within sections where we already hold the mmap_write_lock, so the moves do not change the locking order relative to other kernel locks. Suggested-by: Matthew Wilcox Signed-off-by: Suren Baghdasaryan --- mm/vma.c | 146 ++++++++++++++++++++++++++++++++++++++------------ mm/vma_exec.c | 6 ++- 2 files changed, 117 insertions(+), 35 deletions(-) diff --git a/mm/vma.c b/mm/vma.c index ba78ab1f397a..cc382217f730 100644 --- a/mm/vma.c +++ b/mm/vma.c @@ -524,6 +524,21 @@ __split_vma(struct vma_iterator *vmi, struct vm_area_s= truct *vma, new->vm_pgoff +=3D ((addr - vma->vm_start) >> PAGE_SHIFT); } =20 + /* + * Lock VMAs before cloning to avoid extra work if fatal signal + * is pending. + */ + err =3D vma_start_write_killable(vma); + if (err) + goto out_free_vma; + /* + * Locking a new detached VMA will always succeed but it's just a + * detail of the current implementation, so handle it all the same. + */ + err =3D vma_start_write_killable(new); + if (err) + goto out_free_vma; + err =3D -ENOMEM; vma_iter_config(vmi, new->vm_start, new->vm_end); if (vma_iter_prealloc(vmi, new)) @@ -543,9 +558,6 @@ __split_vma(struct vma_iterator *vmi, struct vm_area_st= ruct *vma, if (new->vm_ops && new->vm_ops->open) new->vm_ops->open(new); =20 - vma_start_write(vma); - vma_start_write(new); - init_vma_prep(&vp, vma); vp.insert =3D new; vma_prepare(&vp); @@ -900,12 +912,22 @@ static __must_check struct vm_area_struct *vma_merge_= existing_range( } =20 /* No matter what happens, we will be adjusting middle. */ - vma_start_write(middle); + err =3D vma_start_write_killable(middle); + if (err) { + /* Ensure error propagates. */ + vmg->give_up_on_oom =3D false; + goto abort; + } =20 if (merge_right) { vma_flags_t next_sticky; =20 - vma_start_write(next); + err =3D vma_start_write_killable(next); + if (err) { + /* Ensure error propagates. */ + vmg->give_up_on_oom =3D false; + goto abort; + } vmg->target =3D next; next_sticky =3D vma_flags_and_mask(&next->flags, VMA_STICKY_FLAGS); vma_flags_set_mask(&sticky_flags, next_sticky); @@ -914,7 +936,12 @@ static __must_check struct vm_area_struct *vma_merge_e= xisting_range( if (merge_left) { vma_flags_t prev_sticky; =20 - vma_start_write(prev); + err =3D vma_start_write_killable(prev); + if (err) { + /* Ensure error propagates. */ + vmg->give_up_on_oom =3D false; + goto abort; + } vmg->target =3D prev; =20 prev_sticky =3D vma_flags_and_mask(&prev->flags, VMA_STICKY_FLAGS); @@ -1170,10 +1197,18 @@ int vma_expand(struct vma_merge_struct *vmg) vma_flags_t sticky_flags =3D vma_flags_and_mask(&vmg->vma_flags, VMA_STICKY_FLAGS); vma_flags_t target_sticky; - int err =3D 0; + int err; =20 mmap_assert_write_locked(vmg->mm); - vma_start_write(target); + err =3D vma_start_write_killable(target); + if (err) { + /* + * Override VMA_MERGE_NOMERGE to prevent callers from + * falling back to a new VMA allocation. + */ + vmg->state =3D VMA_MERGE_ERROR_NOMEM; + return err; + } =20 target_sticky =3D vma_flags_and_mask(&target->flags, VMA_STICKY_FLAGS); =20 @@ -1201,6 +1236,19 @@ int vma_expand(struct vma_merge_struct *vmg) * we don't need to account for vmg->give_up_on_mm here. */ if (remove_next) { + /* + * Lock the VMA early to avoid extra work if fatal signal + * is pending. + */ + err =3D vma_start_write_killable(next); + if (err) { + /* + * Override VMA_MERGE_NOMERGE to prevent callers from + * falling back to a new VMA allocation. + */ + vmg->state =3D VMA_MERGE_ERROR_NOMEM; + return err; + } err =3D dup_anon_vma(target, next, &anon_dup); if (err) return err; @@ -1214,7 +1262,6 @@ int vma_expand(struct vma_merge_struct *vmg) if (remove_next) { vma_flags_t next_sticky; =20 - vma_start_write(next); vmg->__remove_next =3D true; =20 next_sticky =3D vma_flags_and_mask(&next->flags, VMA_STICKY_FLAGS); @@ -1252,9 +1299,14 @@ int vma_shrink(struct vma_iterator *vmi, struct vm_a= rea_struct *vma, unsigned long start, unsigned long end, pgoff_t pgoff) { struct vma_prepare vp; + int err; =20 WARN_ON((vma->vm_start !=3D start) && (vma->vm_end !=3D end)); =20 + err =3D vma_start_write_killable(vma); + if (err) + return err; + if (vma->vm_start < start) vma_iter_config(vmi, vma->vm_start, start); else @@ -1263,8 +1315,6 @@ int vma_shrink(struct vma_iterator *vmi, struct vm_ar= ea_struct *vma, if (vma_iter_prealloc(vmi, NULL)) return -ENOMEM; =20 - vma_start_write(vma); - init_vma_prep(&vp, vma); vma_prepare(&vp); vma_adjust_trans_huge(vma, start, end, NULL); @@ -1453,7 +1503,9 @@ static int vms_gather_munmap_vmas(struct vma_munmap_s= truct *vms, if (error) goto end_split_failed; } - vma_start_write(next); + error =3D vma_start_write_killable(next); + if (error) + goto munmap_gather_failed; mas_set(mas_detach, vms->vma_count++); error =3D mas_store_gfp(mas_detach, next, GFP_KERNEL); if (error) @@ -1848,12 +1900,16 @@ static void vma_link_file(struct vm_area_struct *vm= a, bool hold_rmap_lock) static int vma_link(struct mm_struct *mm, struct vm_area_struct *vma) { VMA_ITERATOR(vmi, mm, 0); + int err; + + err =3D vma_start_write_killable(vma); + if (err) + return err; =20 vma_iter_config(&vmi, vma->vm_start, vma->vm_end); if (vma_iter_prealloc(&vmi, vma)) return -ENOMEM; =20 - vma_start_write(vma); vma_iter_store_new(&vmi, vma); vma_link_file(vma, /* hold_rmap_lock=3D */false); mm->map_count++; @@ -2239,9 +2295,8 @@ int mm_take_all_locks(struct mm_struct *mm) * is reached. */ for_each_vma(vmi, vma) { - if (signal_pending(current)) + if (signal_pending(current) || vma_start_write_killable(vma)) goto out_unlock; - vma_start_write(vma); } =20 vma_iter_init(&vmi, mm, 0); @@ -2540,8 +2595,8 @@ static int __mmap_new_vma(struct mmap_state *map, str= uct vm_area_struct **vmap, struct mmap_action *action) { struct vma_iterator *vmi =3D map->vmi; - int error =3D 0; struct vm_area_struct *vma; + int error; =20 /* * Determine the object being mapped and call the appropriate @@ -2552,6 +2607,14 @@ static int __mmap_new_vma(struct mmap_state *map, st= ruct vm_area_struct **vmap, if (!vma) return -ENOMEM; =20 + /* + * Lock the VMA early to avoid extra work if fatal signal + * is pending. + */ + error =3D vma_start_write_killable(vma); + if (error) + goto free_vma; + vma_iter_config(vmi, map->addr, map->end); vma_set_range(vma, map->addr, map->end, map->pgoff); vma->flags =3D map->vma_flags; @@ -2582,8 +2645,6 @@ static int __mmap_new_vma(struct mmap_state *map, str= uct vm_area_struct **vmap, WARN_ON_ONCE(!arch_validate_flags(map->vm_flags)); #endif =20 - /* Lock the VMA since it is modified after insertion into VMA tree */ - vma_start_write(vma); vma_iter_store_new(vmi, vma); map->mm->map_count++; vma_link_file(vma, action->hide_from_rmap_until_complete); @@ -2878,6 +2939,7 @@ int do_brk_flags(struct vma_iterator *vmi, struct vm_= area_struct *vma, unsigned long addr, unsigned long len, vma_flags_t vma_flags) { struct mm_struct *mm =3D current->mm; + int err; =20 /* * Check against address space limits by the changed size @@ -2910,24 +2972,33 @@ int do_brk_flags(struct vma_iterator *vmi, struct v= m_area_struct *vma, =20 if (vma_merge_new_range(&vmg)) goto out; - else if (vmg_nomem(&vmg)) + if (vmg_nomem(&vmg)) { + err =3D -ENOMEM; goto unacct_fail; + } } =20 if (vma) vma_iter_next_range(vmi); /* create a vma struct for an anonymous mapping */ vma =3D vm_area_alloc(mm); - if (!vma) + if (!vma) { + err =3D -ENOMEM; goto unacct_fail; + } =20 vma_set_anonymous(vma); vma_set_range(vma, addr, addr + len, addr >> PAGE_SHIFT); vma->flags =3D vma_flags; vma->vm_page_prot =3D vm_get_page_prot(vma_flags_to_legacy(vma_flags)); - vma_start_write(vma); - if (vma_iter_store_gfp(vmi, vma, GFP_KERNEL)) + if (vma_start_write_killable(vma)) { + err =3D -EINTR; + goto vma_lock_fail; + } + if (vma_iter_store_gfp(vmi, vma, GFP_KERNEL)) { + err =3D -ENOMEM; goto mas_store_fail; + } =20 mm->map_count++; validate_mm(mm); @@ -2942,10 +3013,11 @@ int do_brk_flags(struct vma_iterator *vmi, struct v= m_area_struct *vma, return 0; =20 mas_store_fail: +vma_lock_fail: vm_area_free(vma); unacct_fail: vm_unacct_memory(len >> PAGE_SHIFT); - return -ENOMEM; + return err; } =20 /** @@ -3112,8 +3184,8 @@ int expand_upwards(struct vm_area_struct *vma, unsign= ed long address) struct mm_struct *mm =3D vma->vm_mm; struct vm_area_struct *next; unsigned long gap_addr; - int error =3D 0; VMA_ITERATOR(vmi, mm, vma->vm_start); + int error; =20 if (!vma_test(vma, VMA_GROWSUP_BIT)) return -EFAULT; @@ -3149,12 +3221,14 @@ int expand_upwards(struct vm_area_struct *vma, unsi= gned long address) =20 /* We must make sure the anon_vma is allocated. */ if (unlikely(anon_vma_prepare(vma))) { - vma_iter_free(&vmi); - return -ENOMEM; + error =3D -ENOMEM; + goto vma_prep_fail; } =20 /* Lock the VMA before expanding to prevent concurrent page faults */ - vma_start_write(vma); + error =3D vma_start_write_killable(vma); + if (error) + goto vma_lock_fail; /* We update the anon VMA tree. */ anon_vma_lock_write(vma->anon_vma); =20 @@ -3183,8 +3257,10 @@ int expand_upwards(struct vm_area_struct *vma, unsig= ned long address) } } anon_vma_unlock_write(vma->anon_vma); - vma_iter_free(&vmi); validate_mm(mm); +vma_lock_fail: +vma_prep_fail: + vma_iter_free(&vmi); return error; } #endif /* CONFIG_STACK_GROWSUP */ @@ -3197,8 +3273,8 @@ int expand_downwards(struct vm_area_struct *vma, unsi= gned long address) { struct mm_struct *mm =3D vma->vm_mm; struct vm_area_struct *prev; - int error =3D 0; VMA_ITERATOR(vmi, mm, vma->vm_start); + int error; =20 if (!vma_test(vma, VMA_GROWSDOWN_BIT)) return -EFAULT; @@ -3228,12 +3304,14 @@ int expand_downwards(struct vm_area_struct *vma, un= signed long address) =20 /* We must make sure the anon_vma is allocated. */ if (unlikely(anon_vma_prepare(vma))) { - vma_iter_free(&vmi); - return -ENOMEM; + error =3D -ENOMEM; + goto vma_prep_fail; } =20 /* Lock the VMA before expanding to prevent concurrent page faults */ - vma_start_write(vma); + error =3D vma_start_write_killable(vma); + if (error) + goto vma_lock_fail; /* We update the anon VMA tree. */ anon_vma_lock_write(vma->anon_vma); =20 @@ -3263,8 +3341,10 @@ int expand_downwards(struct vm_area_struct *vma, uns= igned long address) } } anon_vma_unlock_write(vma->anon_vma); - vma_iter_free(&vmi); validate_mm(mm); +vma_lock_fail: +vma_prep_fail: + vma_iter_free(&vmi); return error; } =20 diff --git a/mm/vma_exec.c b/mm/vma_exec.c index 5cee8b7efa0f..8ddcc791d828 100644 --- a/mm/vma_exec.c +++ b/mm/vma_exec.c @@ -41,6 +41,7 @@ int relocate_vma_down(struct vm_area_struct *vma, unsigne= d long shift) struct vm_area_struct *next; struct mmu_gather tlb; PAGETABLE_MOVE(pmc, vma, vma, old_start, new_start, length); + int err; =20 BUG_ON(new_start > new_end); =20 @@ -56,8 +57,9 @@ int relocate_vma_down(struct vm_area_struct *vma, unsigne= d long shift) * cover the whole range: [new_start, old_end) */ vmg.target =3D vma; - if (vma_expand(&vmg)) - return -ENOMEM; + err =3D vma_expand(&vmg); + if (err) + return err; =20 /* * move the page tables downwards, on failure we rely on --=20 2.53.0.1018.g2bb0e51243-goog