diff mbox series

[v8,1/4] mm/cma: Add PF flag to force non cma alloc

Message ID 20190227144736.5872-2-aneesh.kumar@linux.ibm.com (mailing list archive)
State New, archived
Headers show
Series mm/kvm/vfio/ppc64: Migrate compound pages out of CMA region | expand

Commit Message

Aneesh Kumar K.V Feb. 27, 2019, 2:47 p.m. UTC
This patch adds PF_MEMALLOC_NOCMA which make sure any allocation in that context
is marked non-movable and hence cannot be satisfied by CMA region.

This is useful with get_user_pages_longterm where we want to take a page pin by
migrating pages from CMA region. Marking the section PF_MEMALLOC_NOCMA ensures
that we avoid unnecessary page migration later.

Suggested-by: Andrea Arcangeli <aarcange@redhat.com>
Reviewed-by: Andrea Arcangeli <aarcange@redhat.com>
Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com>
---
 include/linux/sched.h    |  1 +
 include/linux/sched/mm.h | 48 +++++++++++++++++++++++++++++++++-------
 2 files changed, 41 insertions(+), 8 deletions(-)

Comments

Vlastimil Babka Feb. 28, 2019, 12:20 p.m. UTC | #1
On 2/27/19 3:47 PM, Aneesh Kumar K.V wrote:
> This patch adds PF_MEMALLOC_NOCMA which make sure any allocation in that context
> is marked non-movable and hence cannot be satisfied by CMA region.
> 
> This is useful with get_user_pages_longterm where we want to take a page pin by
> migrating pages from CMA region. Marking the section PF_MEMALLOC_NOCMA ensures
> that we avoid unnecessary page migration later.
> 
> Suggested-by: Andrea Arcangeli <aarcange@redhat.com>
> Reviewed-by: Andrea Arcangeli <aarcange@redhat.com>
> Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com>

+CC scheduler guys

Do we really take the last available PF flag just so that "we avoid
unnecessary page migration later"?
If yes, that's a third PF_MEMALLOC flag, should we get separate variable
for gfp context at this point?
Also I don't like the name PF_MEMALLOC_NOCMA, as it's unnecessarily tied
to CMA. If anything it should be e.g. PF_MEMALLOC_NOMOVABLE.

Thanks.

> ---
>  include/linux/sched.h    |  1 +
>  include/linux/sched/mm.h | 48 +++++++++++++++++++++++++++++++++-------
>  2 files changed, 41 insertions(+), 8 deletions(-)
> 
> diff --git a/include/linux/sched.h b/include/linux/sched.h
> index f9b43c989577..dfa90088ba08 100644
> --- a/include/linux/sched.h
> +++ b/include/linux/sched.h
> @@ -1403,6 +1403,7 @@ extern struct pid *cad_pid;
>  #define PF_UMH			0x02000000	/* I'm an Usermodehelper process */
>  #define PF_NO_SETAFFINITY	0x04000000	/* Userland is not allowed to meddle with cpus_allowed */
>  #define PF_MCE_EARLY		0x08000000      /* Early kill for mce process policy */
> +#define PF_MEMALLOC_NOCMA	0x10000000 /* All allocation request will have _GFP_MOVABLE cleared */
>  #define PF_MUTEX_TESTER		0x20000000	/* Thread belongs to the rt mutex tester */
>  #define PF_FREEZER_SKIP		0x40000000	/* Freezer should not count it as freezable */
>  #define PF_SUSPEND_TASK		0x80000000      /* This thread called freeze_processes() and should not be frozen */
> diff --git a/include/linux/sched/mm.h b/include/linux/sched/mm.h
> index 3bfa6a0cbba4..0cd9f10423fb 100644
> --- a/include/linux/sched/mm.h
> +++ b/include/linux/sched/mm.h
> @@ -148,17 +148,25 @@ static inline bool in_vfork(struct task_struct *tsk)
>   * Applies per-task gfp context to the given allocation flags.
>   * PF_MEMALLOC_NOIO implies GFP_NOIO
>   * PF_MEMALLOC_NOFS implies GFP_NOFS
> + * PF_MEMALLOC_NOCMA implies no allocation from CMA region.
>   */
>  static inline gfp_t current_gfp_context(gfp_t flags)
>  {
> -	/*
> -	 * NOIO implies both NOIO and NOFS and it is a weaker context
> -	 * so always make sure it makes precedence
> -	 */
> -	if (unlikely(current->flags & PF_MEMALLOC_NOIO))
> -		flags &= ~(__GFP_IO | __GFP_FS);
> -	else if (unlikely(current->flags & PF_MEMALLOC_NOFS))
> -		flags &= ~__GFP_FS;
> +	if (unlikely(current->flags &
> +		     (PF_MEMALLOC_NOIO | PF_MEMALLOC_NOFS | PF_MEMALLOC_NOCMA))) {
> +		/*
> +		 * NOIO implies both NOIO and NOFS and it is a weaker context
> +		 * so always make sure it makes precedence
> +		 */
> +		if (current->flags & PF_MEMALLOC_NOIO)
> +			flags &= ~(__GFP_IO | __GFP_FS);
> +		else if (current->flags & PF_MEMALLOC_NOFS)
> +			flags &= ~__GFP_FS;
> +#ifdef CONFIG_CMA
> +		if (current->flags & PF_MEMALLOC_NOCMA)
> +			flags &= ~__GFP_MOVABLE;
> +#endif
> +	}
>  	return flags;
>  }
>  
> @@ -248,6 +256,30 @@ static inline void memalloc_noreclaim_restore(unsigned int flags)
>  	current->flags = (current->flags & ~PF_MEMALLOC) | flags;
>  }
>  
> +#ifdef CONFIG_CMA
> +static inline unsigned int memalloc_nocma_save(void)
> +{
> +	unsigned int flags = current->flags & PF_MEMALLOC_NOCMA;
> +
> +	current->flags |= PF_MEMALLOC_NOCMA;
> +	return flags;
> +}
> +
> +static inline void memalloc_nocma_restore(unsigned int flags)
> +{
> +	current->flags = (current->flags & ~PF_MEMALLOC_NOCMA) | flags;
> +}
> +#else
> +static inline unsigned int memalloc_nocma_save(void)
> +{
> +	return 0;
> +}
> +
> +static inline void memalloc_nocma_restore(unsigned int flags)
> +{
> +}
> +#endif
> +
>  #ifdef CONFIG_MEMCG
>  /**
>   * memalloc_use_memcg - Starts the remote memcg charging scope.
>
Michal Hocko Feb. 28, 2019, 2:34 p.m. UTC | #2
On Thu 28-02-19 13:20:03, Vlastimil Babka wrote:
> On 2/27/19 3:47 PM, Aneesh Kumar K.V wrote:
> > This patch adds PF_MEMALLOC_NOCMA which make sure any allocation in that context
> > is marked non-movable and hence cannot be satisfied by CMA region.
> > 
> > This is useful with get_user_pages_longterm where we want to take a page pin by
> > migrating pages from CMA region. Marking the section PF_MEMALLOC_NOCMA ensures
> > that we avoid unnecessary page migration later.
> > 
> > Suggested-by: Andrea Arcangeli <aarcange@redhat.com>
> > Reviewed-by: Andrea Arcangeli <aarcange@redhat.com>
> > Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com>
> 
> +CC scheduler guys
> 
> Do we really take the last available PF flag just so that "we avoid
> unnecessary page migration later"?
> If yes, that's a third PF_MEMALLOC flag, should we get separate variable
> for gfp context at this point?

Yes, that sounds like a reasonable thing to do. Just note that xfs still
uses current_{set,restore}* api to handle PF_MEMALLOC_NOFS so that would
have to be moved over to the memalloc_nofs_{save,restore} API.
diff mbox series

Patch

diff --git a/include/linux/sched.h b/include/linux/sched.h
index f9b43c989577..dfa90088ba08 100644
--- a/include/linux/sched.h
+++ b/include/linux/sched.h
@@ -1403,6 +1403,7 @@  extern struct pid *cad_pid;
 #define PF_UMH			0x02000000	/* I'm an Usermodehelper process */
 #define PF_NO_SETAFFINITY	0x04000000	/* Userland is not allowed to meddle with cpus_allowed */
 #define PF_MCE_EARLY		0x08000000      /* Early kill for mce process policy */
+#define PF_MEMALLOC_NOCMA	0x10000000 /* All allocation request will have _GFP_MOVABLE cleared */
 #define PF_MUTEX_TESTER		0x20000000	/* Thread belongs to the rt mutex tester */
 #define PF_FREEZER_SKIP		0x40000000	/* Freezer should not count it as freezable */
 #define PF_SUSPEND_TASK		0x80000000      /* This thread called freeze_processes() and should not be frozen */
diff --git a/include/linux/sched/mm.h b/include/linux/sched/mm.h
index 3bfa6a0cbba4..0cd9f10423fb 100644
--- a/include/linux/sched/mm.h
+++ b/include/linux/sched/mm.h
@@ -148,17 +148,25 @@  static inline bool in_vfork(struct task_struct *tsk)
  * Applies per-task gfp context to the given allocation flags.
  * PF_MEMALLOC_NOIO implies GFP_NOIO
  * PF_MEMALLOC_NOFS implies GFP_NOFS
+ * PF_MEMALLOC_NOCMA implies no allocation from CMA region.
  */
 static inline gfp_t current_gfp_context(gfp_t flags)
 {
-	/*
-	 * NOIO implies both NOIO and NOFS and it is a weaker context
-	 * so always make sure it makes precedence
-	 */
-	if (unlikely(current->flags & PF_MEMALLOC_NOIO))
-		flags &= ~(__GFP_IO | __GFP_FS);
-	else if (unlikely(current->flags & PF_MEMALLOC_NOFS))
-		flags &= ~__GFP_FS;
+	if (unlikely(current->flags &
+		     (PF_MEMALLOC_NOIO | PF_MEMALLOC_NOFS | PF_MEMALLOC_NOCMA))) {
+		/*
+		 * NOIO implies both NOIO and NOFS and it is a weaker context
+		 * so always make sure it makes precedence
+		 */
+		if (current->flags & PF_MEMALLOC_NOIO)
+			flags &= ~(__GFP_IO | __GFP_FS);
+		else if (current->flags & PF_MEMALLOC_NOFS)
+			flags &= ~__GFP_FS;
+#ifdef CONFIG_CMA
+		if (current->flags & PF_MEMALLOC_NOCMA)
+			flags &= ~__GFP_MOVABLE;
+#endif
+	}
 	return flags;
 }
 
@@ -248,6 +256,30 @@  static inline void memalloc_noreclaim_restore(unsigned int flags)
 	current->flags = (current->flags & ~PF_MEMALLOC) | flags;
 }
 
+#ifdef CONFIG_CMA
+static inline unsigned int memalloc_nocma_save(void)
+{
+	unsigned int flags = current->flags & PF_MEMALLOC_NOCMA;
+
+	current->flags |= PF_MEMALLOC_NOCMA;
+	return flags;
+}
+
+static inline void memalloc_nocma_restore(unsigned int flags)
+{
+	current->flags = (current->flags & ~PF_MEMALLOC_NOCMA) | flags;
+}
+#else
+static inline unsigned int memalloc_nocma_save(void)
+{
+	return 0;
+}
+
+static inline void memalloc_nocma_restore(unsigned int flags)
+{
+}
+#endif
+
 #ifdef CONFIG_MEMCG
 /**
  * memalloc_use_memcg - Starts the remote memcg charging scope.