From: Laurent Dufour <ldufour@linux.ibm.com>
To: Suren Baghdasaryan <surenb@google.com>, akpm@linux-foundation.org
Cc: michel@lespinasse.org, joelaf@google.com, songliubraving@fb.com,
mhocko@suse.com, david@redhat.com, peterz@infradead.org,
bigeasy@linutronix.de, peterx@redhat.com, dhowells@redhat.com,
linux-mm@kvack.org, jglisse@google.com, dave@stgolabs.net,
minchan@google.com, x86@kernel.org, hughd@google.com,
willy@infradead.org, laurent.dufour@fr.ibm.com, mgorman@suse.de,
rientjes@google.com, axelrasmussen@google.com,
kernel-team@android.com, paulmck@kernel.org,
liam.howlett@oracle.com, luto@kernel.org, vbabka@suse.cz,
linux-arm-kernel@lists.infradead.org, kent.overstreet@linux.dev,
linux-kernel@vger.kernel.org, hannes@cmpxchg.org,
linuxppc-dev@lists.ozlabs.org
Subject: Re: [RFC PATCH RESEND 28/28] kernel/fork: throttle call_rcu() calls in vm_area_free
Date: Fri, 9 Sep 2022 17:19:09 +0200 [thread overview]
Message-ID: <34dd5656-dc3c-6a20-5390-04d05c619fdc@linux.ibm.com> (raw)
In-Reply-To: <20220901173516.702122-29-surenb@google.com>
Le 01/09/2022 à 19:35, Suren Baghdasaryan a écrit :
> call_rcu() can take a long time when callback offloading is enabled.
> Its use in the vm_area_free can cause regressions in the exit path when
> multiple VMAs are being freed. To minimize that impact, place VMAs into
> a list and free them in groups using one call_rcu() call per group.
>
> Signed-off-by: Suren Baghdasaryan <surenb@google.com>
> ---
> include/linux/mm.h | 1 +
> include/linux/mm_types.h | 11 ++++++-
> kernel/fork.c | 68 +++++++++++++++++++++++++++++++++++-----
> mm/init-mm.c | 3 ++
> mm/mmap.c | 1 +
> 5 files changed, 75 insertions(+), 9 deletions(-)
>
> diff --git a/include/linux/mm.h b/include/linux/mm.h
> index a3cbaa7b9119..81dff694ac14 100644
> --- a/include/linux/mm.h
> +++ b/include/linux/mm.h
> @@ -249,6 +249,7 @@ void setup_initial_init_mm(void *start_code, void *end_code,
> struct vm_area_struct *vm_area_alloc(struct mm_struct *);
> struct vm_area_struct *vm_area_dup(struct vm_area_struct *);
> void vm_area_free(struct vm_area_struct *);
> +void drain_free_vmas(struct mm_struct *mm);
>
> #ifndef CONFIG_MMU
> extern struct rb_root nommu_region_tree;
> diff --git a/include/linux/mm_types.h b/include/linux/mm_types.h
> index 36562e702baf..6f3effc493b1 100644
> --- a/include/linux/mm_types.h
> +++ b/include/linux/mm_types.h
> @@ -412,7 +412,11 @@ struct vm_area_struct {
> struct vm_area_struct *vm_next, *vm_prev;
> };
> #ifdef CONFIG_PER_VMA_LOCK
> - struct rcu_head vm_rcu; /* Used for deferred freeing. */
> + struct {
> + struct list_head vm_free_list;
> + /* Used for deferred freeing. */
> + struct rcu_head vm_rcu;
> + };
> #endif
> };
>
> @@ -573,6 +577,11 @@ struct mm_struct {
> */
> #ifdef CONFIG_PER_VMA_LOCK
> int mm_lock_seq;
> + struct {
> + struct list_head head;
> + spinlock_t lock;
> + int size;
> + } vma_free_list;
> #endif
>
>
> diff --git a/kernel/fork.c b/kernel/fork.c
> index b443ba3a247a..7c88710aed72 100644
> --- a/kernel/fork.c
> +++ b/kernel/fork.c
> @@ -483,26 +483,75 @@ struct vm_area_struct *vm_area_dup(struct vm_area_struct *orig)
> }
>
> #ifdef CONFIG_PER_VMA_LOCK
> -static void __vm_area_free(struct rcu_head *head)
> +static inline void __vm_area_free(struct vm_area_struct *vma)
> {
> - struct vm_area_struct *vma = container_of(head, struct vm_area_struct,
> - vm_rcu);
> /* The vma should either have no lock holders or be write-locked. */
> vma_assert_no_reader(vma);
> kmem_cache_free(vm_area_cachep, vma);
> }
> -#endif
> +
> +static void vma_free_rcu_callback(struct rcu_head *head)
> +{
> + struct vm_area_struct *first_vma;
> + struct vm_area_struct *vma, *vma2;
> +
> + first_vma = container_of(head, struct vm_area_struct, vm_rcu);
> + list_for_each_entry_safe(vma, vma2, &first_vma->vm_free_list, vm_free_list)
Is that safe to walk the list against concurrent calls to
list_splice_init(), or list_add()?
> + __vm_area_free(vma);
> + __vm_area_free(first_vma);
> +}
> +
> +void drain_free_vmas(struct mm_struct *mm)
> +{
> + struct vm_area_struct *first_vma;
> + LIST_HEAD(to_destroy);
> +
> + spin_lock(&mm->vma_free_list.lock);
> + list_splice_init(&mm->vma_free_list.head, &to_destroy);
> + mm->vma_free_list.size = 0;
> + spin_unlock(&mm->vma_free_list.lock);
> +
> + if (list_empty(&to_destroy))
> + return;
> +
> + first_vma = list_first_entry(&to_destroy, struct vm_area_struct, vm_free_list);
> + /* Remove the head which is allocated on the stack */
> + list_del(&to_destroy);
> +
> + call_rcu(&first_vma->vm_rcu, vma_free_rcu_callback);
> +}
> +
> +#define VM_AREA_FREE_LIST_MAX 32
> +
> +void vm_area_free(struct vm_area_struct *vma)
> +{
> + struct mm_struct *mm = vma->vm_mm;
> + bool drain;
> +
> + free_anon_vma_name(vma);
> +
> + spin_lock(&mm->vma_free_list.lock);
> + list_add(&vma->vm_free_list, &mm->vma_free_list.head);
> + mm->vma_free_list.size++;
> + drain = mm->vma_free_list.size > VM_AREA_FREE_LIST_MAX;
> + spin_unlock(&mm->vma_free_list.lock);
> +
> + if (drain)
> + drain_free_vmas(mm);
> +}
> +
> +#else /* CONFIG_PER_VMA_LOCK */
> +
> +void drain_free_vmas(struct mm_struct *mm) {}
>
> void vm_area_free(struct vm_area_struct *vma)
> {
> free_anon_vma_name(vma);
> -#ifdef CONFIG_PER_VMA_LOCK
> - call_rcu(&vma->vm_rcu, __vm_area_free);
> -#else
> kmem_cache_free(vm_area_cachep, vma);
> -#endif
> }
>
> +#endif /* CONFIG_PER_VMA_LOCK */
> +
> static void account_kernel_stack(struct task_struct *tsk, int account)
> {
> if (IS_ENABLED(CONFIG_VMAP_STACK)) {
> @@ -1137,6 +1186,9 @@ static struct mm_struct *mm_init(struct mm_struct *mm, struct task_struct *p,
> INIT_LIST_HEAD(&mm->mmlist);
> #ifdef CONFIG_PER_VMA_LOCK
> WRITE_ONCE(mm->mm_lock_seq, 0);
> + INIT_LIST_HEAD(&mm->vma_free_list.head);
> + spin_lock_init(&mm->vma_free_list.lock);
> + mm->vma_free_list.size = 0;
> #endif
> mm_pgtables_bytes_init(mm);
> mm->map_count = 0;
> diff --git a/mm/init-mm.c b/mm/init-mm.c
> index 8399f90d631c..7b6d2460545f 100644
> --- a/mm/init-mm.c
> +++ b/mm/init-mm.c
> @@ -39,6 +39,9 @@ struct mm_struct init_mm = {
> .mmlist = LIST_HEAD_INIT(init_mm.mmlist),
> #ifdef CONFIG_PER_VMA_LOCK
> .mm_lock_seq = 0,
> + .vma_free_list.head = LIST_HEAD_INIT(init_mm.vma_free_list.head),
> + .vma_free_list.lock = __SPIN_LOCK_UNLOCKED(init_mm.vma_free_list.lock),
> + .vma_free_list.size = 0,
> #endif
> .user_ns = &init_user_ns,
> .cpu_bitmap = CPU_BITS_NONE,
> diff --git a/mm/mmap.c b/mm/mmap.c
> index 1edfcd384f5e..d61b7ef84ba6 100644
> --- a/mm/mmap.c
> +++ b/mm/mmap.c
> @@ -3149,6 +3149,7 @@ void exit_mmap(struct mm_struct *mm)
> }
> mm->mmap = NULL;
> mmap_write_unlock(mm);
> + drain_free_vmas(mm);
> vm_unacct_memory(nr_accounted);
> }
>
next prev parent reply other threads:[~2022-09-09 15:20 UTC|newest]
Thread overview: 91+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-09-01 17:34 [RFC PATCH RESEND 00/28] per-VMA locks proposal Suren Baghdasaryan
2022-09-01 17:34 ` [RFC PATCH RESEND 01/28] mm: introduce CONFIG_PER_VMA_LOCK Suren Baghdasaryan
2022-09-01 17:34 ` [RFC PATCH RESEND 02/28] mm: rcu safe VMA freeing Suren Baghdasaryan
2022-09-01 17:34 ` [RFC PATCH RESEND 03/28] mm: introduce __find_vma to be used without mmap_lock protection Suren Baghdasaryan
2022-09-01 20:22 ` Kent Overstreet
2022-09-01 23:18 ` Suren Baghdasaryan
2022-09-01 17:34 ` [RFC PATCH RESEND 04/28] mm: move mmap_lock assert function definitions Suren Baghdasaryan
2022-09-01 20:24 ` Kent Overstreet
2022-09-01 20:51 ` Liam Howlett
2022-09-01 23:21 ` Suren Baghdasaryan
2022-09-02 6:23 ` Sebastian Andrzej Siewior
2022-09-02 17:46 ` Suren Baghdasaryan
2022-09-01 17:34 ` [RFC PATCH RESEND 05/28] mm: add per-VMA lock and helper functions to control it Suren Baghdasaryan
2022-09-06 13:46 ` Laurent Dufour
2022-09-06 17:24 ` Suren Baghdasaryan
2022-09-01 17:34 ` [RFC PATCH RESEND 06/28] mm: mark VMA as locked whenever vma->vm_flags are modified Suren Baghdasaryan
2022-09-06 14:26 ` Laurent Dufour
2022-09-06 19:00 ` Suren Baghdasaryan
2022-09-06 20:00 ` Liam Howlett
2022-09-06 20:13 ` Suren Baghdasaryan
2022-09-01 17:34 ` [RFC PATCH RESEND 07/28] kernel/fork: mark VMAs as locked before copying pages during fork Suren Baghdasaryan
2022-09-06 14:37 ` Laurent Dufour
2022-09-08 23:57 ` Suren Baghdasaryan
2022-09-09 13:27 ` Laurent Dufour
2022-09-09 16:29 ` Suren Baghdasaryan
2022-09-01 17:34 ` [RFC PATCH RESEND 08/28] mm/khugepaged: mark VMA as locked while collapsing a hugepage Suren Baghdasaryan
2022-09-06 14:43 ` Laurent Dufour
2022-09-09 0:15 ` Suren Baghdasaryan
2022-09-01 17:34 ` [RFC PATCH RESEND 09/28] mm/mempolicy: mark VMA as locked when changing protection policy Suren Baghdasaryan
2022-09-06 14:47 ` Laurent Dufour
2022-09-09 0:27 ` Suren Baghdasaryan
2022-09-01 17:34 ` [RFC PATCH RESEND 10/28] mm/mmap: mark VMAs as locked in vma_adjust Suren Baghdasaryan
2022-09-06 15:35 ` Laurent Dufour
2022-09-09 0:51 ` Suren Baghdasaryan
2022-09-09 15:52 ` Laurent Dufour
2022-09-01 17:34 ` [RFC PATCH RESEND 11/28] mm/mmap: mark VMAs as locked before merging or splitting them Suren Baghdasaryan
2022-09-06 15:44 ` Laurent Dufour
2022-09-01 17:35 ` [RFC PATCH RESEND 12/28] mm/mremap: mark VMA as locked while remapping it to a new address range Suren Baghdasaryan
2022-09-06 16:09 ` Laurent Dufour
2022-09-01 17:35 ` [RFC PATCH RESEND 13/28] mm: conditionally mark VMA as locked in free_pgtables and unmap_page_range Suren Baghdasaryan
2022-09-09 10:33 ` Laurent Dufour
2022-09-09 16:43 ` Suren Baghdasaryan
2022-09-01 17:35 ` [RFC PATCH RESEND 14/28] mm: mark VMAs as locked before isolating them Suren Baghdasaryan
2022-09-09 13:35 ` Laurent Dufour
2022-09-09 16:28 ` Suren Baghdasaryan
2022-09-01 17:35 ` [RFC PATCH RESEND 15/28] mm/mmap: mark adjacent VMAs as locked if they can grow into unmapped area Suren Baghdasaryan
2022-09-09 13:43 ` Laurent Dufour
2022-09-09 16:25 ` Suren Baghdasaryan
2022-09-01 17:35 ` [RFC PATCH RESEND 16/28] kernel/fork: assert no VMA readers during its destruction Suren Baghdasaryan
2022-09-09 13:56 ` Laurent Dufour
2022-09-09 16:19 ` Suren Baghdasaryan
2022-09-01 17:35 ` [RFC PATCH RESEND 17/28] mm/mmap: prevent pagefault handler from racing with mmu_notifier registration Suren Baghdasaryan
2022-09-09 14:20 ` Laurent Dufour
2022-09-09 16:12 ` Suren Baghdasaryan
2022-09-01 17:35 ` [RFC PATCH RESEND 18/28] mm: add FAULT_FLAG_VMA_LOCK flag Suren Baghdasaryan
2022-09-09 14:26 ` Laurent Dufour
2022-09-01 17:35 ` [RFC PATCH RESEND 19/28] mm: disallow do_swap_page to handle page faults under VMA lock Suren Baghdasaryan
2022-09-06 19:39 ` Peter Xu
2022-09-06 20:08 ` Suren Baghdasaryan
2022-09-06 20:22 ` Peter Xu
2022-09-07 0:58 ` Suren Baghdasaryan
2022-09-09 14:26 ` Laurent Dufour
2022-09-01 17:35 ` [RFC PATCH RESEND 20/28] mm: introduce per-VMA lock statistics Suren Baghdasaryan
2022-09-09 14:28 ` Laurent Dufour
2022-09-09 16:11 ` Suren Baghdasaryan
2022-09-01 17:35 ` [RFC PATCH RESEND 21/28] mm: introduce find_and_lock_anon_vma to be used from arch-specific code Suren Baghdasaryan
2022-09-09 14:38 ` Laurent Dufour
2022-09-09 16:10 ` Suren Baghdasaryan
2022-09-01 17:35 ` [RFC PATCH RESEND 22/28] x86/mm: try VMA lock-based page fault handling first Suren Baghdasaryan
2022-09-01 17:35 ` [RFC PATCH RESEND 23/28] x86/mm: define ARCH_SUPPORTS_PER_VMA_LOCK Suren Baghdasaryan
2022-09-01 20:20 ` Kent Overstreet
2022-09-01 23:17 ` Suren Baghdasaryan
2022-09-01 17:35 ` [RFC PATCH RESEND 24/28] arm64/mm: try VMA lock-based page fault handling first Suren Baghdasaryan
2022-09-01 17:35 ` [RFC PATCH RESEND 25/28] arm64/mm: define ARCH_SUPPORTS_PER_VMA_LOCK Suren Baghdasaryan
2022-09-01 17:35 ` [RFC PATCH RESEND 26/28] powerc/mm: try VMA lock-based page fault handling first Suren Baghdasaryan
2022-09-01 17:35 ` [RFC PATCH RESEND 27/28] powerpc/mm: define ARCH_SUPPORTS_PER_VMA_LOCK Suren Baghdasaryan
2022-09-01 17:35 ` [RFC PATCH RESEND 28/28] kernel/fork: throttle call_rcu() calls in vm_area_free Suren Baghdasaryan
2022-09-09 15:19 ` Laurent Dufour [this message]
2022-09-09 16:02 ` Suren Baghdasaryan
2022-09-09 16:14 ` Laurent Dufour
2022-09-01 20:58 ` [RFC PATCH RESEND 00/28] per-VMA locks proposal Kent Overstreet
2022-09-01 23:26 ` Suren Baghdasaryan
2022-09-11 9:35 ` Vlastimil Babka
2022-09-28 2:28 ` Suren Baghdasaryan
2022-09-29 11:18 ` Vlastimil Babka
2022-09-02 7:42 ` Peter Zijlstra
2022-09-02 14:45 ` Suren Baghdasaryan
2022-09-05 12:32 ` Michal Hocko
2022-09-05 18:32 ` Suren Baghdasaryan
2022-09-05 20:35 ` Kent Overstreet
2022-09-06 15:46 ` Suren Baghdasaryan
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=34dd5656-dc3c-6a20-5390-04d05c619fdc@linux.ibm.com \
--to=ldufour@linux.ibm.com \
--cc=akpm@linux-foundation.org \
--cc=axelrasmussen@google.com \
--cc=bigeasy@linutronix.de \
--cc=dave@stgolabs.net \
--cc=david@redhat.com \
--cc=dhowells@redhat.com \
--cc=hannes@cmpxchg.org \
--cc=hughd@google.com \
--cc=jglisse@google.com \
--cc=joelaf@google.com \
--cc=kent.overstreet@linux.dev \
--cc=kernel-team@android.com \
--cc=laurent.dufour@fr.ibm.com \
--cc=liam.howlett@oracle.com \
--cc=linux-arm-kernel@lists.infradead.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=linuxppc-dev@lists.ozlabs.org \
--cc=luto@kernel.org \
--cc=mgorman@suse.de \
--cc=mhocko@suse.com \
--cc=michel@lespinasse.org \
--cc=minchan@google.com \
--cc=paulmck@kernel.org \
--cc=peterx@redhat.com \
--cc=peterz@infradead.org \
--cc=rientjes@google.com \
--cc=songliubraving@fb.com \
--cc=surenb@google.com \
--cc=vbabka@suse.cz \
--cc=willy@infradead.org \
--cc=x86@kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).