When the IOMMU is enabled very early (as with irq-remapping)
some devices are still in BIOS hand. When dma is blocked
early this can cause lots of IO_PAGE_FAULTs. So delay the
DMA initialization and do it right before the dma_ops are
initialized.
To be secure, block all interrupts by default when irq-remapping is
enabled in the system. They will be reenabled on demand
later. Without blocking interrupts by default devices can
issue arbitrary interrupts by sending special DMA packets to
the CPU that look like MSI messages. This is especially
dangerous when a device is assigned to a KVM guest because
the guest can then DoS the host.
Signed-off-by: NJoerg Roedel <joerg.roedel@amd.com>

When the IOAPIC information provided in the IVRS table is
not correct or not complete the system may not boot at all
when interrupt remapping is enabled. So check if this
information is correct and print out a firmware bug message
when it is not.
Signed-off-by: NJoerg Roedel <joerg.roedel@amd.com>

To easily map device ids to interrupt remapping table
entries a new lookup table is necessary.
Signed-off-by: NJoerg Roedel <joerg.roedel@amd.com>

The irq remapping tables for the AMD IOMMU need to be
aligned on a 128 byte boundary. Create a seperate slab-cache
to guarantee this alignment.
Signed-off-by: NJoerg Roedel <joerg.roedel@amd.com>

The IVRS ACPI table provides information about the IOAPICs
and the HPETs available in the system and which PCI device
ID they use in transactions. Save that information for later
usage in interrupt remapping.
Signed-off-by: NJoerg Roedel <joerg.roedel@amd.com>
Reviewed-by: NKonrad Rzeszutek Wilk <konrad.wilk@oracle.com>

The check in the for-loop is broken. Fix it and the
boot-crash it causes in AMD IOMMUv2 systems.
Signed-off-by: NJoerg Roedel <joerg.roedel@amd.com>

The pci_request_acs() function needs to be called before PCI
probing to be effective. So move it to another call-place to
ensure that.
Signed-off-by: NJoerg Roedel <joerg.roedel@amd.com>

This step makes it very easy to keep track about the current
intialization state of the iommu driver. With this change we
can initialize the IOMMU hardware to a point where it can
remap interrupts and later resume the initializion to enable
dma remapping.
Signed-off-by: NJoerg Roedel <joerg.roedel@amd.com>

This function will initialize everthing necessary so that
devices can do DMA. This includes dma_ops and iommu_ops.
Signed-off-by: NJoerg Roedel <joerg.roedel@amd.com>

The message belongs there anyway, so move it to that
function.
Signed-off-by: NJoerg Roedel <joerg.roedel@amd.com>

Split the enable_iommus() routine so that a part of it can
run in early code.
Signed-off-by: NJoerg Roedel <joerg.roedel@amd.com>

Split out the code to parse the ACPI table and setup
relevant data structures into a new function.
Signed-off-by: NJoerg Roedel <joerg.roedel@amd.com>

This function will be called before the PCI subsystem is
initialized. Therefore dev_name doen't work and IOMMU
information can't be printed to the klog as before. Move the
code to print that information to a later point where PCI
initializtion has already happened.
Signed-off-by: NJoerg Roedel <joerg.roedel@amd.com>

For interrupt remapping the relevant IOMMU initialization
needs to run earlier at boot when the PCI subsystem is not
yet initialized. To support that this patch splits the parts
of IOMMU initialization which need PCI accesses out of the
initial setup path so that this can be done later.
Signed-off-by: NJoerg Roedel <joerg.roedel@amd.com>

This makes it easier to propagate errors while parsing the
IVRS table and makes the amd_iommu_init_err hack obsolete.
Signed-off-by: NJoerg Roedel <joerg.roedel@amd.com>

A few sparse warnings fire in drivers/iommu/amd_iommu_init.c.
Fix most of them with this patch. Also fix the sparse
warnings in drivers/iommu/irq_remapping.c while at it.
Signed-off-by: NJoerg Roedel <joerg.roedel@amd.com>

write_file_bool() modifies 32 bits of data, so "amd_iommu_unmap_flush"
needs to be 32 bits as well or we'll corrupt memory.  Fortunately it
looks like the data is aligned with a gap after the declaration so this
is harmless in production.
Signed-off-by: NDan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: NJoerg Roedel <joerg.roedel@amd.com>

The iommu_shutdown callback is not initialized when the AMD
IOMMU driver runs in passthrough mode. Fix that by moving
the callback initialization before the check for
passthrough mode.
Signed-off-by: NShuah Khan <shuah.khan@hp.com>
Cc: stable@vger.kernel.org
Signed-off-by: NJoerg Roedel <joerg.roedel@amd.com>

At some point pci_get_bus_and_slot started to enable
interrupts. Since this function is used in the
amd_iommu_resume path it will enable interrupts on resume
which causes a warning. The fix will use a cached pointer
to the root-bridge to re-enable the IOMMU in case the BIOS
is broken.

Cc: stable@vger.kernel.org
Signed-off-by: NJoerg Roedel <joerg.roedel@amd.com>

amd_iommu_enable_interrupts() called in amd_iommu_resume().
Signed-off-by: NGerard Snitselaar <dev@snitselaar.org>
Signed-off-by: NJoerg Roedel <joro@8bytes.org>

For interrupt remapping the enablement of the IOMMU MSI
interrupt needs to be deferred because the IOMMU itself will
be initialized before the io-apics are up and running. So
the code to setup the MSI is moved seperated from the
hardware-setup routine now.
Signed-off-by: NJoerg Roedel <joerg.roedel@amd.com>

Unfortunatly the interrupts for the event log and the
peripheral page-faults are only enabled at boot but not
re-enabled at resume. Fix that.

Cc: stable@vger.kernel.org
Signed-off-by: NJoerg Roedel <joerg.roedel@amd.com>

When ioremap_nocache fails in iommu initialization the code
calls release_mem_region immediatly. But the function is
called again when the propagates into the upper init
functions leading to a double-free. Fix that.
Reported-by: NDon Dutile <ddutile@redhat.com>
Signed-off-by: NJoerg Roedel <joerg.roedel@amd.com>

This function is called from enable_iommus(), which in turn is used
from amd_iommu_resume().

Cc: stable@vger.kernel.org
Signed-off-by: NJan Beulich <jbeulich@suse.com>
Signed-off-by: NJoerg Roedel <joerg.roedel@amd.com>

The hardware-initializtion part of the AMD IOMMU driver is
split out into a seperate function. This function can now be
called either from amd_iommu_init() itself or any other
place if the hardware needs to be ready earlier. This will
be used to implement interrupt remapping for AMD.
Signed-off-by: NJoerg Roedel <joerg.roedel@amd.com>

To protect the command buffer from hanging when a device
does not respond to an IOTLB invalidation, set a timeout of
1s for outstanding IOTLB invalidations.
Signed-off-by: NJoerg Roedel <joerg.roedel@amd.com>

The caches are already flushed in enable_iommus(), so this
flush is not necessary.
Signed-off-by: NJoerg Roedel <joerg.roedel@amd.com>

This patch adds support for protection domains that
implement two-level paging for devices.
Signed-off-by: NJoerg Roedel <joerg.roedel@amd.com>

If the device starts to use IOMMUv2 features the dma handles
need to stay valid. The only sane way to do this is to use a
identity mapping for the device and not translate it by the
iommu. This is implemented with this patch. Since this lifts
the device-isolation there is also a new kernel parameter
which allows to disable that feature.
Signed-off-by: NJoerg Roedel <joerg.roedel@amd.com>

In mixed IOMMU setups this flag inidicates whether an IOMMU
supports the v2 features or not. This patch also adds a
global flag together with a function to query that flag from
other code. The flag shows if at least one IOMMUv2 is in the
system.
Signed-off-by: NJoerg Roedel <joerg.roedel@amd.com>

This feature needs to be enabled before IOMMUv2 DTEs can be
set up.
Signed-off-by: NJoerg Roedel <joerg.roedel@amd.com>

Allocate and enable a log buffer for peripheral page faults
when the IOMMU supports this feature.
Signed-off-by: NJoerg Roedel <joerg.roedel@amd.com>

Read the number of PASIDs supported by each IOMMU in the
system and take the smallest number as the maximum value
supported by the IOMMU driver.
Signed-off-by: NJoerg Roedel <joerg.roedel@amd.com>

Convert the contents of 'struct dev_table_entry' to u64 to
allow updating the DTE wit 64bit writes as required by the
spec.
Signed-off-by: NJoerg Roedel <joerg.roedel@amd.com>

A few parts of the driver were missing in drivers/iommu.
Move them there to have the complete driver in that
directory.
Signed-off-by: NJoerg Roedel <joerg.roedel@amd.com>

The driver contains several loops counting on an u16 value
where the exit-condition is checked against variables that
can have values up to 0xffff. In this case the loops will
never exit. This patch fixed 3 such loops.

Cc: stable@kernel.org
Signed-off-by: NJoerg Roedel <joerg.roedel@amd.com>

Move the interupt handling for the iommu into the interupt
thread to reduce latencies and prepare interupt handling for
pri handling.
Signed-off-by: NJoerg Roedel <joerg.roedel@amd.com>

This patch adds detection of the extended features of an
AMD IOMMU. The available features are printed to dmesg on
boot.
Signed-off-by: NJoerg Roedel <joerg.roedel@amd.com>

This patch adds a flag to the AMD IOMMU driver to indicate
that all IOMMUs present in the system support device IOTLBs.
Signed-off-by: NJoerg Roedel <joerg.roedel@amd.com>

The old code only flushed a DTE or a domain TLB before it is
actually used by the IOMMU driver. While this is efficient
and works when done right it is more likely to introduce new
bugs when changing code (which happened in the past).
This patch adds code to flush all DTEs and all domain TLBs
in each IOMMU right after it is enabled (at boot and after
resume). This reduces the complexity of the driver and makes
it less likely to introduce stale-TLB bugs in the future.
Signed-off-by: NJoerg Roedel <joerg.roedel@amd.com>