Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753700AbdLNVbB (ORCPT ); Thu, 14 Dec 2017 16:31:01 -0500 Received: from mail-it0-f65.google.com ([209.85.214.65]:41693 "EHLO mail-it0-f65.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753232AbdLNVbA (ORCPT ); Thu, 14 Dec 2017 16:31:00 -0500 X-Google-Smtp-Source: ACJfBouq6s3Iq0RhKvKYIPsWHVGAgZNm9eWlHWYuVnHMCeb6xKHjJ3I+s30VfDOH8Hr/XpvtEUYhxQ== Date: Thu, 14 Dec 2017 13:30:56 -0800 (PST) From: David Rientjes X-X-Sender: rientjes@chino.kir.corp.google.com To: Andrew Morton cc: Michal Hocko , Andrea Arcangeli , Benjamin Herrenschmidt , Paul Mackerras , Oded Gabbay , Alex Deucher , =?UTF-8?Q?Christian_K=C3=B6nig?= , David Airlie , Joerg Roedel , Doug Ledford , Jani Nikula , Mike Marciniszyn , Sean Hefty , Dimitri Sivanich , Boris Ostrovsky , =?UTF-8?Q?J=C3=A9r=C3=B4me_Glisse?= , Paolo Bonzini , =?UTF-8?Q?Radim_Kr=C4=8Dm=C3=A1=C5=99?= , linux-kernel@vger.kernel.org, linux-mm@kvack.org Subject: [patch v2 1/2] mm, mmu_notifier: annotate mmu notifiers with blockable invalidate callbacks In-Reply-To: Message-ID: References: User-Agent: Alpine 2.10 (DEB 1266 2009-07-14) MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 6761 Lines: 188 Commit 4d4bbd8526a8 ("mm, oom_reaper: skip mm structs with mmu notifiers") prevented the oom reaper from unmapping private anonymous memory with the oom reaper when the oom victim mm had mmu notifiers registered. The rationale is that doing mmu_notifier_invalidate_range_{start,end}() around the unmap_page_range(), which is needed, can block and the oom killer will stall forever waiting for the victim to exit, which may not be possible without reaping. That concern is real, but only true for mmu notifiers that have blockable invalidate_range_{start,end}() callbacks. This patch adds a "flags" field to mmu notifier ops that can set a bit to indicate that these callbacks do not block. The implementation is steered toward an expensive slowpath, such as after the oom reaper has grabbed mm->mmap_sem of a still alive oom victim. Signed-off-by: David Rientjes --- v2: - specifically exclude mmu_notifiers without invalidate callbacks - move flags to mmu_notifier_ops per Paolo - reverse flag from blockable -> not blockable per Christian drivers/infiniband/hw/hfi1/mmu_rb.c | 1 + drivers/iommu/amd_iommu_v2.c | 1 + drivers/iommu/intel-svm.c | 1 + drivers/misc/sgi-gru/grutlbpurge.c | 1 + include/linux/mmu_notifier.h | 21 +++++++++++++++++++++ mm/mmu_notifier.c | 31 +++++++++++++++++++++++++++++++ virt/kvm/kvm_main.c | 1 + 7 files changed, 57 insertions(+) diff --git a/drivers/infiniband/hw/hfi1/mmu_rb.c b/drivers/infiniband/hw/hfi1/mmu_rb.c --- a/drivers/infiniband/hw/hfi1/mmu_rb.c +++ b/drivers/infiniband/hw/hfi1/mmu_rb.c @@ -77,6 +77,7 @@ static void do_remove(struct mmu_rb_handler *handler, static void handle_remove(struct work_struct *work); static const struct mmu_notifier_ops mn_opts = { + .flags = MMU_INVALIDATE_DOES_NOT_BLOCK, .invalidate_range_start = mmu_notifier_range_start, }; diff --git a/drivers/iommu/amd_iommu_v2.c b/drivers/iommu/amd_iommu_v2.c --- a/drivers/iommu/amd_iommu_v2.c +++ b/drivers/iommu/amd_iommu_v2.c @@ -427,6 +427,7 @@ static void mn_release(struct mmu_notifier *mn, struct mm_struct *mm) } static const struct mmu_notifier_ops iommu_mn = { + .flags = MMU_INVALIDATE_DOES_NOT_BLOCK, .release = mn_release, .clear_flush_young = mn_clear_flush_young, .invalidate_range = mn_invalidate_range, diff --git a/drivers/iommu/intel-svm.c b/drivers/iommu/intel-svm.c --- a/drivers/iommu/intel-svm.c +++ b/drivers/iommu/intel-svm.c @@ -276,6 +276,7 @@ static void intel_mm_release(struct mmu_notifier *mn, struct mm_struct *mm) } static const struct mmu_notifier_ops intel_mmuops = { + .flags = MMU_INVALIDATE_DOES_NOT_BLOCK, .release = intel_mm_release, .change_pte = intel_change_pte, .invalidate_range = intel_invalidate_range, diff --git a/drivers/misc/sgi-gru/grutlbpurge.c b/drivers/misc/sgi-gru/grutlbpurge.c --- a/drivers/misc/sgi-gru/grutlbpurge.c +++ b/drivers/misc/sgi-gru/grutlbpurge.c @@ -258,6 +258,7 @@ static void gru_release(struct mmu_notifier *mn, struct mm_struct *mm) static const struct mmu_notifier_ops gru_mmuops = { + .flags = MMU_INVALIDATE_DOES_NOT_BLOCK, .invalidate_range_start = gru_invalidate_range_start, .invalidate_range_end = gru_invalidate_range_end, .release = gru_release, diff --git a/include/linux/mmu_notifier.h b/include/linux/mmu_notifier.h --- a/include/linux/mmu_notifier.h +++ b/include/linux/mmu_notifier.h @@ -10,6 +10,9 @@ struct mmu_notifier; struct mmu_notifier_ops; +/* mmu_notifier_ops flags */ +#define MMU_INVALIDATE_DOES_NOT_BLOCK (0x01) + #ifdef CONFIG_MMU_NOTIFIER /* @@ -26,6 +29,15 @@ struct mmu_notifier_mm { }; struct mmu_notifier_ops { + /* + * Flags to specify behavior of callbacks for this MMU notifier. + * Used to determine which context an operation may be called. + * + * MMU_INVALIDATE_DOES_NOT_BLOCK: invalidate_{start,end} does not + * block + */ + int flags; + /* * Called either by mmu_notifier_unregister or when the mm is * being destroyed by exit_mmap, always before all pages are @@ -137,6 +149,9 @@ struct mmu_notifier_ops { * page. Pages will no longer be referenced by the linux * address space but may still be referenced by sptes until * the last refcount is dropped. + * + * If both of these callbacks cannot block, mmu_notifier_ops.flags + * should have MMU_INVALIDATE_DOES_NOT_BLOCK set. */ void (*invalidate_range_start)(struct mmu_notifier *mn, struct mm_struct *mm, @@ -218,6 +233,7 @@ extern void __mmu_notifier_invalidate_range_end(struct mm_struct *mm, bool only_end); extern void __mmu_notifier_invalidate_range(struct mm_struct *mm, unsigned long start, unsigned long end); +extern int mm_has_blockable_invalidate_notifiers(struct mm_struct *mm); static inline void mmu_notifier_release(struct mm_struct *mm) { @@ -457,6 +473,11 @@ static inline void mmu_notifier_invalidate_range(struct mm_struct *mm, { } +static inline int mm_has_blockable_invalidate_notifiers(struct mm_struct *mm) +{ + return 0; +} + static inline void mmu_notifier_mm_init(struct mm_struct *mm) { } diff --git a/mm/mmu_notifier.c b/mm/mmu_notifier.c --- a/mm/mmu_notifier.c +++ b/mm/mmu_notifier.c @@ -236,6 +236,37 @@ void __mmu_notifier_invalidate_range(struct mm_struct *mm, } EXPORT_SYMBOL_GPL(__mmu_notifier_invalidate_range); +/* + * Must be called while holding mm->mmap_sem for either read or write. + * The result is guaranteed to be valid until mm->mmap_sem is dropped. + */ +int mm_has_blockable_invalidate_notifiers(struct mm_struct *mm) +{ + struct mmu_notifier *mn; + int id; + int ret = 0; + + WARN_ON_ONCE(down_write_trylock(&mm->mmap_sem)); + + if (!mm_has_notifiers(mm)) + return ret; + + id = srcu_read_lock(&srcu); + hlist_for_each_entry_rcu(mn, &mm->mmu_notifier_mm->list, hlist) { + if (!mn->ops->invalidate_range && + !mn->ops->invalidate_range_start && + !mn->ops->invalidate_range_end) + continue; + + if (!(mn->ops->flags & MMU_INVALIDATE_DOES_NOT_BLOCK)) { + ret = 1; + break; + } + } + srcu_read_unlock(&srcu, id); + return ret; +} + static int do_mmu_notifier_register(struct mmu_notifier *mn, struct mm_struct *mm, int take_mmap_sem) diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c --- a/virt/kvm/kvm_main.c +++ b/virt/kvm/kvm_main.c @@ -476,6 +476,7 @@ static void kvm_mmu_notifier_release(struct mmu_notifier *mn, } static const struct mmu_notifier_ops kvm_mmu_notifier_ops = { + .flags = MMU_INVALIDATE_DOES_NOT_BLOCK, .invalidate_range_start = kvm_mmu_notifier_invalidate_range_start, .invalidate_range_end = kvm_mmu_notifier_invalidate_range_end, .clear_flush_young = kvm_mmu_notifier_clear_flush_young,