stable.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Boris Ostrovsky <boris.ostrovsky@oracle.com>
To: david.vrabel@citrix.com, jgross@suse.com
Cc: xen-devel@lists.xenproject.org, linux-kernel@vger.kernel.org,
	olaf@aepfle.de, Boris Ostrovsky <boris.ostrovsky@oracle.com>,
	stable@vger.kernel.org
Subject: [PATCH v3 (re-send)] xen/gntdev: Use mempolicy instead of VM_IO flag to avoid NUMA balancing
Date: Thu, 17 Nov 2016 15:10:04 -0500	[thread overview]
Message-ID: <1479413404-27332-1-git-send-email-boris.ostrovsky@oracle.com> (raw)

Commit 9c17d96500f7 ("xen/gntdev: Grant maps should not be subject to
NUMA balancing") set VM_IO flag to prevent grant maps from being
subjected to NUMA balancing.

It was discovered recently that this flag causes get_user_pages() to
always fail with -EFAULT.

check_vma_flags
__get_user_pages
__get_user_pages_locked
__get_user_pages_unlocked
get_user_pages_fast
iov_iter_get_pages
dio_refill_pages
do_direct_IO
do_blockdev_direct_IO
do_blockdev_direct_IO
ext4_direct_IO_read
generic_file_read_iter
aio_run_iocb

(which can happen if guest's vdisk has direct-io-safe option).

To avoid this don't use vm_flags. Instead, use mempolicy that prohibits
page migration (i.e. clear MPOL_F_MOF|MPOL_F_MORON) and make sure we
don't consult task's policy (which may include those flags) if vma
doesn't have one.

Reported-by: Olaf Hering <olaf@aepfle.de>
Signed-off-by: Boris Ostrovsky <boris.ostrovsky@oracle.com>
Cc: stable@vger.kernel.org
---

Mis-spelled David's address.

Changes in v3:
* Don't use __mpol_dup() and get_task_policy() which are not exported
  for use by drivers. Add vm_operations_struct.get_policy().
* Copy to stable


 drivers/xen/gntdev.c |   27 ++++++++++++++++++++++++++-
 1 files changed, 26 insertions(+), 1 deletions(-)

diff --git a/drivers/xen/gntdev.c b/drivers/xen/gntdev.c
index bb95212..632edd4 100644
--- a/drivers/xen/gntdev.c
+++ b/drivers/xen/gntdev.c
@@ -35,6 +35,7 @@
 #include <linux/spinlock.h>
 #include <linux/slab.h>
 #include <linux/highmem.h>
+#include <linux/mempolicy.h>
 
 #include <xen/xen.h>
 #include <xen/grant_table.h>
@@ -433,10 +434,28 @@ static void gntdev_vma_close(struct vm_area_struct *vma)
 	return map->pages[(addr - map->pages_vm_start) >> PAGE_SHIFT];
 }
 
+#ifdef CONFIG_NUMA
+/*
+ * We have this op to make sure callers (such as vma_policy_mof()) don't
+ * check current task's policy which may include migrate flags (MPOL_F_MOF
+ * or MPOL_F_MORON)
+ */
+static struct mempolicy *gntdev_vma_get_policy(struct vm_area_struct *vma,
+					       unsigned long addr)
+{
+	if (mpol_needs_cond_ref(vma->vm_policy))
+		mpol_get(vma->vm_policy);
+	return vma->vm_policy;
+}
+#endif
+
 static const struct vm_operations_struct gntdev_vmops = {
 	.open = gntdev_vma_open,
 	.close = gntdev_vma_close,
 	.find_special_page = gntdev_vma_find_special_page,
+#ifdef CONFIG_NUMA
+	.get_policy = gntdev_vma_get_policy,
+#endif
 };
 
 /* ------------------------------------------------------------------ */
@@ -1007,7 +1026,13 @@ static int gntdev_mmap(struct file *flip, struct vm_area_struct *vma)
 
 	vma->vm_ops = &gntdev_vmops;
 
-	vma->vm_flags |= VM_DONTEXPAND | VM_DONTDUMP | VM_IO;
+	vma->vm_flags |= VM_DONTEXPAND | VM_DONTDUMP;
+
+#ifdef CONFIG_NUMA
+	/* Prevent NUMA balancing */
+	if (vma->vm_policy)
+		vma->vm_policy->flags &= ~(MPOL_F_MOF | MPOL_F_MORON);
+#endif
 
 	if (use_ptemod)
 		vma->vm_flags |= VM_DONTCOPY;
-- 
1.7.1


             reply	other threads:[~2016-11-17 20:08 UTC|newest]

Thread overview: 2+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2016-11-17 20:10 Boris Ostrovsky [this message]
2016-11-18  7:59 ` [PATCH v3 (re-send)] xen/gntdev: Use mempolicy instead of VM_IO flag to avoid NUMA balancing Olaf Hering

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=1479413404-27332-1-git-send-email-boris.ostrovsky@oracle.com \
    --to=boris.ostrovsky@oracle.com \
    --cc=david.vrabel@citrix.com \
    --cc=jgross@suse.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=olaf@aepfle.de \
    --cc=stable@vger.kernel.org \
    --cc=xen-devel@lists.xenproject.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).