From: Christoph Hellwig <hch@lst.de>
To: "Dan Williams" <dan.j.williams@intel.com>,
"Jérôme Glisse" <jglisse@redhat.com>,
"Jason Gunthorpe" <jgg@mellanox.com>,
"Ben Skeggs" <bskeggs@redhat.com>
Cc: Ira Weiny <ira.weiny@intel.com>,
linux-mm@kvack.org, nouveau@lists.freedesktop.org,
dri-devel@lists.freedesktop.org, linux-nvdimm@lists.01.org,
linux-pci@vger.kernel.org, linux-kernel@vger.kernel.org,
Philip Yang <Philip.Yang@amd.com>
Subject: [PATCH 16/22] mm/hmm: Remove confusing comment and logic from hmm_release
Date: Mon, 1 Jul 2019 08:20:14 +0200 [thread overview]
Message-ID: <20190701062020.19239-17-hch@lst.de> (raw)
In-Reply-To: <20190701062020.19239-1-hch@lst.de>
From: Jason Gunthorpe <jgg@mellanox.com>
hmm_release() is called exactly once per hmm. ops->release() cannot
accidentally trigger any action that would recurse back onto
hmm->mirrors_sem.
This fixes a use after-free race of the form:
CPU0 CPU1
hmm_release()
up_write(&hmm->mirrors_sem);
hmm_mirror_unregister(mirror)
down_write(&hmm->mirrors_sem);
up_write(&hmm->mirrors_sem);
kfree(mirror)
mirror->ops->release(mirror)
The only user we have today for ops->release is an empty function, so this
is unambiguously safe.
As a consequence of plugging this race drivers are not allowed to
register/unregister mirrors from within a release op.
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Tested-by: Philip Yang <Philip.Yang@amd.com>
---
mm/hmm.c | 28 +++++++++-------------------
1 file changed, 9 insertions(+), 19 deletions(-)
diff --git a/mm/hmm.c b/mm/hmm.c
index c30aa9403dbe..b224ea635a77 100644
--- a/mm/hmm.c
+++ b/mm/hmm.c
@@ -130,26 +130,16 @@ static void hmm_release(struct mmu_notifier *mn, struct mm_struct *mm)
*/
WARN_ON(!list_empty_careful(&hmm->ranges));
- down_write(&hmm->mirrors_sem);
- mirror = list_first_entry_or_null(&hmm->mirrors, struct hmm_mirror,
- list);
- while (mirror) {
- list_del_init(&mirror->list);
- if (mirror->ops->release) {
- /*
- * Drop mirrors_sem so the release callback can wait
- * on any pending work that might itself trigger a
- * mmu_notifier callback and thus would deadlock with
- * us.
- */
- up_write(&hmm->mirrors_sem);
+ down_read(&hmm->mirrors_sem);
+ list_for_each_entry(mirror, &hmm->mirrors, list) {
+ /*
+ * Note: The driver is not allowed to trigger
+ * hmm_mirror_unregister() from this thread.
+ */
+ if (mirror->ops->release)
mirror->ops->release(mirror);
- down_write(&hmm->mirrors_sem);
- }
- mirror = list_first_entry_or_null(&hmm->mirrors,
- struct hmm_mirror, list);
}
- up_write(&hmm->mirrors_sem);
+ up_read(&hmm->mirrors_sem);
hmm_put(hmm);
}
@@ -279,7 +269,7 @@ void hmm_mirror_unregister(struct hmm_mirror *mirror)
struct hmm *hmm = mirror->hmm;
down_write(&hmm->mirrors_sem);
- list_del_init(&mirror->list);
+ list_del(&mirror->list);
up_write(&hmm->mirrors_sem);
hmm_put(hmm);
}
--
2.20.1
next prev parent reply other threads:[~2019-07-01 6:21 UTC|newest]
Thread overview: 40+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-07-01 6:19 dev_pagemap related cleanups v4 Christoph Hellwig
2019-07-01 6:19 ` [PATCH 01/22] mm/hmm.c: suppress compilation warnings when CONFIG_HUGETLB_PAGE is not set Christoph Hellwig
2019-07-01 6:20 ` [PATCH 02/22] mm/hmm: update HMM documentation Christoph Hellwig
2019-07-01 6:20 ` [PATCH 03/22] mm/hmm: clean up some coding style and comments Christoph Hellwig
2019-07-01 6:20 ` [PATCH 04/22] mm/hmm: support automatic NUMA balancing Christoph Hellwig
2019-07-01 6:20 ` [PATCH 05/22] mm/hmm: Only set FAULT_FLAG_ALLOW_RETRY for non-blocking Christoph Hellwig
2019-07-01 6:20 ` [PATCH 06/22] mm/hmm: fix use after free with struct hmm in the mmu notifiers Christoph Hellwig
2019-07-01 6:20 ` [PATCH 07/22] mm/hmm: Use hmm_mirror not mm as an argument for hmm_range_register Christoph Hellwig
2019-07-01 6:20 ` [PATCH 08/22] mm/hmm: Hold a mmgrab from hmm to mm Christoph Hellwig
2019-07-01 6:20 ` [PATCH 09/22] mm/hmm: Simplify hmm_get_or_create and make it reliable Christoph Hellwig
2019-07-01 6:20 ` [PATCH 10/22] mm/hmm: Remove duplicate condition test before wait_event_timeout Christoph Hellwig
2019-07-01 6:20 ` [PATCH 11/22] mm/hmm: Do not use list*_rcu() for hmm->ranges Christoph Hellwig
2019-07-01 6:20 ` [PATCH 12/22] mm/hmm: Hold on to the mmget for the lifetime of the range Christoph Hellwig
2019-07-01 6:20 ` [PATCH 13/22] mm/hmm: Use lockdep instead of comments Christoph Hellwig
2019-07-01 6:20 ` [PATCH 14/22] mm/hmm: Remove racy protection against double-unregistration Christoph Hellwig
2019-07-01 6:20 ` [PATCH 15/22] mm/hmm: Poison hmm_range during unregister Christoph Hellwig
2019-07-01 6:20 ` Christoph Hellwig [this message]
2019-07-01 6:20 ` [PATCH 17/22] mm/hmm: Fix error flows in hmm_invalidate_range_start Christoph Hellwig
2019-07-01 6:20 ` [PATCH 18/22] mm: return valid info from hmm_range_unregister Christoph Hellwig
2019-07-03 17:22 ` Ralph Campbell
2019-07-01 6:20 ` [PATCH 19/22] mm: always return EBUSY for invalid ranges in hmm_range_{fault,snapshot} Christoph Hellwig
2019-07-02 21:43 ` Kuehling, Felix
2019-07-03 17:32 ` Ralph Campbell
2019-07-01 6:20 ` [PATCH 20/22] mm: move hmm_vma_fault to nouveau Christoph Hellwig
2019-07-03 17:48 ` Ralph Campbell
2019-07-03 17:50 ` [Nouveau] " Ilia Mirkin
2019-07-03 18:03 ` Jason Gunthorpe
2019-07-03 18:05 ` Christoph Hellwig
2019-07-03 18:13 ` Jason Gunthorpe
2019-07-01 6:20 ` [PATCH 21/22] nouveau: unlock mmap_sem on all errors from nouveau_range_fault Christoph Hellwig
2019-07-01 6:20 ` [PATCH 22/22] mm: remove the legacy hmm_pfn_* APIs Christoph Hellwig
2019-07-03 18:01 ` Jason Gunthorpe
2019-07-03 18:03 ` Christoph Hellwig
2019-07-03 18:15 ` Jason Gunthorpe
2019-07-01 8:25 ` dev_pagemap related cleanups v4 Christoph Hellwig
2019-07-02 18:42 ` Jason Gunthorpe
2019-07-02 22:45 ` Weiny, Ira
2019-07-02 22:47 ` Christoph Hellwig
2019-07-02 23:17 ` Dan Williams
2019-07-03 1:08 ` Jason Gunthorpe
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=20190701062020.19239-17-hch@lst.de \
--to=hch@lst.de \
--cc=Philip.Yang@amd.com \
--cc=bskeggs@redhat.com \
--cc=dan.j.williams@intel.com \
--cc=dri-devel@lists.freedesktop.org \
--cc=ira.weiny@intel.com \
--cc=jgg@mellanox.com \
--cc=jglisse@redhat.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=linux-nvdimm@lists.01.org \
--cc=linux-pci@vger.kernel.org \
--cc=nouveau@lists.freedesktop.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).