[04/22] mm/hmm: support automatic NUMA balancing

Message ID	20190701062020.19239-5-hch@lst.de (mailing list archive)
State	New, archived
Delegated to:	Bjorn Helgaas
Headers	show Return-Path: <linux-pci-owner@kernel.org> From: Christoph Hellwig <hch@lst.de> To: Dan Williams <dan.j.williams@intel.com>, =?utf-8?b?SsOpcsO0bWUgR2xpc3Nl?= <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>, Felix Kuehling <Felix.Kuehling@amd.com> Subject: [PATCH 04/22] mm/hmm: support automatic NUMA balancing Date: Mon, 1 Jul 2019 08:20:02 +0200 Message-Id: <20190701062020.19239-5-hch@lst.de> In-Reply-To: <20190701062020.19239-1-hch@lst.de> References: <20190701062020.19239-1-hch@lst.de> MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Sender: linux-pci-owner@vger.kernel.org Precedence: bulk
Series	[01/22] mm/hmm.c: suppress compilation warnings when CONFIG_HUGETLB_PAGE is not set \| expand [01/22] mm/hmm.c: suppress compilation warnings when CONFIG_HUGETLB_PAGE is not set [02/22] mm/hmm: update HMM documentation [03/22] mm/hmm: clean up some coding style and comments [04/22] mm/hmm: support automatic NUMA balancing [05/22] mm/hmm: Only set FAULT_FLAG_ALLOW_RETRY for non-blocking [06/22] mm/hmm: fix use after free with struct hmm in the mmu notifiers [07/22] mm/hmm: Use hmm_mirror not mm as an argument for hmm_range_register [08/22] mm/hmm: Hold a mmgrab from hmm to mm [09/22] mm/hmm: Simplify hmm_get_or_create and make it reliable [10/22] mm/hmm: Remove duplicate condition test before wait_event_timeout [11/22] mm/hmm: Do not use list_rcu() for hmm->ranges [12/22] mm/hmm: Hold on to the mmget for the lifetime of the range [13/22] mm/hmm: Use lockdep instead of comments [14/22] mm/hmm: Remove racy protection against double-unregistration [15/22] mm/hmm: Poison hmm_range during unregister [16/22] mm/hmm: Remove confusing comment and logic from hmm_release [17/22] mm/hmm: Fix error flows in hmm_invalidate_range_start [18/22] mm: return valid info from hmm_range_unregister [19/22] mm: always return EBUSY for invalid ranges in hmm_range_{fault,snapshot} [20/22] mm: move hmm_vma_fault to nouveau [21/22] nouveau: unlock mmap_sem on all errors from nouveau_range_fault [22/22] mm: remove the legacy hmm_pfn_ APIs

Message ID

20190701062020.19239-5-hch@lst.de (mailing list archive)

State

New, archived

Delegated to:

Bjorn Helgaas

Headers

From: Christoph Hellwig <hch@lst.de>
To: Dan Williams <dan.j.williams@intel.com>,
 =?utf-8?b?SsOpcsO0bWUgR2xpc3Nl?= <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>,
        Felix Kuehling <Felix.Kuehling@amd.com>
Subject: [PATCH 04/22] mm/hmm: support automatic NUMA balancing
Date: Mon,  1 Jul 2019 08:20:02 +0200
Message-Id: <20190701062020.19239-5-hch@lst.de>
In-Reply-To: <20190701062020.19239-1-hch@lst.de>
References: <20190701062020.19239-1-hch@lst.de>
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
Sender: linux-pci-owner@vger.kernel.org
Precedence: bulk

Series

[01/22] mm/hmm.c: suppress compilation warnings when CONFIG_HUGETLB_PAGE is not set | expand

Commit Message

Christoph Hellwig July 1, 2019, 6:20 a.m. UTC

From: Philip Yang <Philip.Yang@amd.com>

While the page is migrating by NUMA balancing, HMM failed to detect this
condition and still return the old page. Application will use the new page
migrated, but driver pass the old page physical address to GPU, this crash
the application later.

Use pte_protnone(pte) to return this condition and then hmm_vma_do_fault
will allocate new page.

Signed-off-by: Philip Yang <Philip.Yang@amd.com>
Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com>
Reviewed-by: Jérôme Glisse <jglisse@redhat.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
---
 mm/hmm.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/mm/hmm.c b/mm/hmm.c
index 4db5dcf110ba..dce4e70e648a 100644
--- a/mm/hmm.c
+++ b/mm/hmm.c
@@ -548,7 +548,7 @@  static int hmm_vma_handle_pmd(struct mm_walk *walk,
 
 static inline uint64_t pte_to_hmm_pfn_flags(struct hmm_range *range, pte_t pte)
 {
-	if (pte_none(pte) || !pte_present(pte))
+	if (pte_none(pte) || !pte_present(pte) || pte_protnone(pte))
 		return 0;
 	return pte_write(pte) ? range->flags[HMM_PFN_VALID] |
 				range->flags[HMM_PFN_WRITE] :

[04/22] mm/hmm: support automatic NUMA balancing

Commit Message

Patch