linux-toradex.git/drivers/iommu/intel-svm.c, branch v4.20-rc6

iommu/vt-d: Fix NULL pointer dereference in prq_event_thread()

2018-11-06T15:46:24+00:00

When handling page request without pasid event, go to "no_pasid"
branch instead of "bad_req". Otherwise, a NULL pointer deference
will happen there.

Cc: Ashok Raj 
Cc: Jacob Pan 
Cc: Sohil Mehta 
Signed-off-by: Lu Baolu 
Fixes: a222a7f0bb6c9 'iommu/vt-d: Implement page request handling'
Signed-off-by: Joerg Roedel

Revert "mm, mmu_notifier: annotate mmu notifiers with blockable invalidate callbacks"

2018-10-26T23:25:19+00:00

Revert 5ff7091f5a2ca ("mm, mmu_notifier: annotate mmu notifiers with
blockable invalidate callbacks").

MMU_INVALIDATE_DOES_NOT_BLOCK flags was the only one used and it is no
longer needed since 93065ac753e4 ("mm, oom: distinguish blockable mode for
mmu notifiers").  We now have a full support for per range !blocking
behavior so we can drop the stop gap workaround which the per notifier
flag was used for.

Link: http://lkml.kernel.org/r/20180827112623.8992-4-mhocko@kernel.org
Signed-off-by: Michal Hocko 
Cc: David Rientjes 
Cc: Boris Ostrovsky 
Cc: Jerome Glisse 
Cc: Juergen Gross 
Cc: Tetsuo Handa 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds

Merge tag 'iommu-updates-v4.19' of git://git.kernel.org/pub/scm/linux/kernel/git/joro/iommu

2018-08-24T20:10:38+00:00

Pull IOMMU updates from Joerg Roedel:

 - PASID table handling updates for the Intel VT-d driver. It implements
   a global PASID space now so that applications usings multiple devices
   will just have one PASID.

 - A new config option to make iommu passthroug mode the default.

 - New sysfs attribute for iommu groups to export the type of the
   default domain.

 - A debugfs interface (for debug only) usable by IOMMU drivers to
   export internals to user-space.

 - R-Car Gen3 SoCs support for the ipmmu-vmsa driver

 - The ARM-SMMU now aborts transactions from unknown devices and devices
   not attached to any domain.

 - Various cleanups and smaller fixes all over the place.

* tag 'iommu-updates-v4.19' of git://git.kernel.org/pub/scm/linux/kernel/git/joro/iommu: (42 commits)
  iommu/omap: Fix cache flushes on L2 table entries
  iommu: Remove the ->map_sg indirection
  iommu/arm-smmu-v3: Abort all transactions if SMMU is enabled in kdump kernel
  iommu/arm-smmu-v3: Prevent any devices access to memory without registration
  iommu/ipmmu-vmsa: Don't register as BUS IOMMU if machine doesn't have IPMMU-VMSA
  iommu/ipmmu-vmsa: Clarify supported platforms
  iommu/ipmmu-vmsa: Fix allocation in atomic context
  iommu: Add config option to set passthrough as default
  iommu: Add sysfs attribyte for domain type
  iommu/arm-smmu-v3: sync the OVACKFLG to PRIQ consumer register
  iommu/arm-smmu: Error out only if not enough context interrupts
  iommu/io-pgtable-arm-v7s: Abort allocation when table address overflows the PTE
  iommu/io-pgtable-arm: Fix pgtable allocation in selftest
  iommu/vt-d: Remove the obsolete per iommu pasid tables
  iommu/vt-d: Apply per pci device pasid table in SVA
  iommu/vt-d: Allocate and free pasid table
  iommu/vt-d: Per PCI device pasid table interfaces
  iommu/vt-d: Add for_each_device_domain() helper
  iommu/vt-d: Move device_domain_info to header
  iommu/vt-d: Apply global PASID in SVA
  ...

mm: convert return type of handle_mm_fault() caller to vm_fault_t

2018-08-17T23:20:28+00:00

Use new return type vm_fault_t for fault handler.  For now, this is just
documenting that the function returns a VM_FAULT value rather than an
errno.  Once all instances are converted, vm_fault_t will become a
distinct type.

Ref-> commit 1c8f422059ae ("mm: change return type to vm_fault_t")

In this patch all the caller of handle_mm_fault() are changed to return
vm_fault_t type.

Link: http://lkml.kernel.org/r/20180617084810.GA6730@jordon-HP-15-Notebook-PC
Signed-off-by: Souptick Joarder 
Cc: Matthew Wilcox 
Cc: Richard Henderson 
Cc: Tony Luck 
Cc: Matt Turner 
Cc: Vineet Gupta 
Cc: Russell King 
Cc: Catalin Marinas 
Cc: Will Deacon 
Cc: Richard Kuo 
Cc: Geert Uytterhoeven 
Cc: Michal Simek 
Cc: James Hogan 
Cc: Ley Foon Tan 
Cc: Jonas Bonn 
Cc: James E.J. Bottomley 
Cc: Benjamin Herrenschmidt 
Cc: Palmer Dabbelt 
Cc: Yoshinori Sato 
Cc: David S. Miller 
Cc: Richard Weinberger 
Cc: Guan Xuetao 
Cc: Thomas Gleixner 
Cc: "H. Peter Anvin" 
Cc: "Levin, Alexander (Sasha Levin)" 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds

iommu/vt-d: Remove the obsolete per iommu pasid tables

2018-07-20T12:44:26+00:00

The obsolete per iommu pasid tables are no longer used. Hence,
clean up them.

Cc: Ashok Raj 
Cc: Jacob Pan 
Cc: Kevin Tian 
Cc: Liu Yi L 
Signed-off-by: Lu Baolu 
Reviewed-by: Liu Yi L 
Signed-off-by: Joerg Roedel

iommu/vt-d: Apply per pci device pasid table in SVA

2018-07-20T12:44:25+00:00

This patch applies the per pci device pasid table in the Shared
Virtual Address (SVA) implementation.

Cc: Ashok Raj 
Cc: Jacob Pan 
Cc: Kevin Tian 
Cc: Liu Yi L 
Signed-off-by: Lu Baolu 
Reviewed-by: Liu Yi L 
Signed-off-by: Joerg Roedel

iommu/vt-d: Per PCI device pasid table interfaces

2018-07-20T12:44:24+00:00

This patch adds the interfaces for per PCI device pasid
table management. Currently we allocate one pasid table
for all PCI devices under the scope of an IOMMU. It's
insecure in some cases where multiple devices under one
single IOMMU unit support PASID features. With per PCI
device pasid table, we can achieve finer protection and
isolation granularity.

Cc: Ashok Raj 
Cc: Jacob Pan 
Cc: Kevin Tian 
Cc: Liu Yi L 
Suggested-by: Ashok Raj 
Signed-off-by: Lu Baolu 
Reviewed-by: Liu Yi L 
Signed-off-by: Joerg Roedel

iommu/vt-d: Apply global PASID in SVA

2018-07-20T12:44:24+00:00

This patch applies the global pasid name space in the shared
virtual address (SVA) implementation.

Cc: Ashok Raj 
Cc: Jacob Pan 
Cc: Kevin Tian 
Cc: Liu Yi L 
Signed-off-by: Lu Baolu 
Reviewed-by: Kevin Tian 
Reviewed-by: Liu Yi L 
Reviewed-by: Peter Xu 
Signed-off-by: Joerg Roedel

iommu/vt-d: Avoid using idr_for_each_entry()

2018-07-20T12:44:23+00:00

idr_for_each_entry() is used to iteratte over idr elements
of a given type. It isn't suitable for the globle pasid idr
since the pasid idr consumer could specify different types
of pointers to bind with a pasid.

Cc: Ashok Raj 
Cc: Jacob Pan 
Cc: Kevin Tian 
Cc: Liu Yi L 
Signed-off-by: Lu Baolu 
Reviewed-by: Kevin Tian 
Reviewed-by: Liu Yi L 
Reviewed-by: Peter Xu 
Signed-off-by: Joerg Roedel

iommu/vt-d: Remove unnecessary WARN_ON()

2018-07-06T12:18:20+00:00

When drivers call intel_svm_available() to check whether the
Shared Virtual Memory(SVM) is supported by the IOMMU driver,
they will get a warning in the kernel message if the SVM is
not supported by the hardware.

[    3.790876] WARNING: CPU: 0 PID: 267 at drivers/iommu/intel-svm.c:334 intel_svm_bind_mm+0x292/0x570
[    3.790877] Modules linked in: dsa(+) vfio_mdev mdev vfio_iommu_type1 soundcore vfio serio_raw parport_pc ppdev lp parport autofs4 psmouse virtio_net pata_acpi
[    3.790884] CPU: 0 PID: 267 Comm: systemd-udevd Not tainted 4.15.0+ #358
[    3.790885] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.11.0-0-g63451fca13-prebuilt.qemu-project.org 04/01/2014
[    3.790887] RIP: 0010:intel_svm_bind_mm+0x292/0x570
[    3.790887] RSP: 0000:ffffac72c08a3a70 EFLAGS: 00010246
[    3.790889] RAX: 0000000000000000 RBX: ffff90447a5160a0 RCX: 0000000000000000
[    3.790889] RDX: 0000000000000000 RSI: 0000000000000082 RDI: ffff90447fc16550
[    3.790890] RBP: ffff90447a516000 R08: 0000000000000000 R09: 0000000000000178
[    3.790891] R10: 0000000000000220 R11: 0000000000aaaaaa R12: 0000000000000000
[    3.790891] R13: 0000000000000002 R14: ffffac72c08a3b18 R15: ffffac72c08a3eb8
[    3.790893] FS:  00007fb21e85b8c0(0000) GS:ffff90447fc00000(0000) knlGS:0000000000000000
[    3.790894] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[    3.790894] CR2: 000055c08167d148 CR3: 000000013a6f4000 CR4: 00000000000006f0
[    3.790903] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[    3.790904] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400

This is caused by a unnecessary WARN_ON() in intel_svm_bind_mm().
Hence, remove it.

Signed-off-by: Lu Baolu 
Signed-off-by: Joerg Roedel