IF YOU WOULD LIKE TO GET AN ACCOUNT, please write an
email to Administrator. User accounts are meant only to access repo
and report issues and/or generate pull requests.
This is a purpose-specific Git hosting for
BaseALT
projects. Thank you for your understanding!
Только зарегистрированные пользователи имеют доступ к сервису!
Для получения аккаунта, обратитесь к администратору.
IBM POWER8 NVlink systems come with Tesla K40-ish GPUs each of which
also has a couple of fast speed links (NVLink). The interface to links
is exposed as an emulated PCI bridge which is included into the same
IOMMU group as the corresponding GPU.
In the kernel, NPUs get a separate PHB of the PNV_PHB_NPU type and a PE
which behave pretty much as the standard IODA2 PHB except NPU PHB has
just a single TVE in the hardware which means it can have either
32bit window or 64bit window or DMA bypass but never two of these.
In order to make these links work when GPU is passed to the guest,
these bridges need to be passed as well; otherwise performance will
degrade.
This implements and exports API to manage NPU state in regard to VFIO;
it replicates iommu_table_group_ops.
This defines a new pnv_pci_ioda2_npu_ops which is assigned to
the IODA2 bridge if there are NPUs for a GPU on the bridge.
The new callbacks call the default IODA2 callbacks plus new NPU API.
This adds a gpe_table_group_to_npe() helper to find NPU PE for the IODA2
table_group, it is not expected to fail as the helper is only called
from the pnv_pci_ioda2_npu_ops.
This does not define NPU-specific .release_ownership() so after
VFIO is finished, DMA on NPU is disabled which is ok as the nvidia
driver sets DMA mask when probing which enable 32 or 64bit DMA on NPU.
This adds a pnv_pci_npu_setup_iommu() helper which adds NPUs to
the GPU group if any found. The helper uses helpers to look for
the "ibm,gpu" property in the device tree which is a phandle of
the corresponding GPU.
This adds an additional loop over PEs in pnv_ioda_setup_dma() as the main
loop skips NPU PEs as they do not have 32bit DMA segments.
As pnv_npu_set_window() and pnv_npu_unset_window() are started being used
by the new IODA2-NPU IOMMU group, this makes the helpers public and
adds the DMA window number parameter.
Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Reviewed-By: Alistair Popple <alistair@popple.id.au>
[mpe: Add pnv_pci_ioda_setup_iommu_api() to fix build with IOMMU_API=n]
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
The pnv_ioda_pe struct keeps an array of peers. At the moment it is only
used to link GPU and NPU for 2 purposes:
1. Access NPU quickly when configuring DMA for GPU - this was addressed
in the previos patch by removing use of it as DMA setup is not what
the kernel would constantly do.
2. Invalidate TCE cache for NPU when it is invalidated for GPU.
GPU and NPU are in different PE. There is already a mechanism to
attach multiple iommu_table_group to the same iommu_table (used for VFIO),
we can reuse it here so does this patch.
This gets rid of peers[] array and PNV_IODA_PE_PEER flag as they are
not needed anymore.
While we are here, add TCE cache invalidation after enabling bypass.
Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Reviewed-By: Alistair Popple <alistair@popple.id.au>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
This exports debugging helper pe_level_printk() and corresponding macroses
so they can be used in npu-dma.c.
Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Reviewed-By: Alistair Popple <alistair@popple.id.au>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
NPU devices are emulated in firmware and mainly used for NPU NVLink
training; one NPU device is per a hardware link. Their DMA/TCE setup
must match the GPU which is connected via PCIe and NVLink so any changes
to the DMA/TCE setup on the GPU PCIe device need to be propagated to
the NVLink device as this is what device drivers expect and it doesn't
make much sense to do anything else.
This makes NPU DMA setup explicit.
pnv_npu_ioda_controller_ops::pnv_npu_dma_set_mask is moved to pci-ioda,
made static and prints warning as dma_set_mask() should never be called
on this function as in any case it will not configure GPU; so we make
this explicit.
Instead of using PNV_IODA_PE_PEER and peers[] (which the next patch will
remove), we test every PCI device if there are corresponding NVLink
devices. If there are any, we propagate bypass mode to just found NPU
devices by calling the setup helper directly (which takes @bypass) and
avoid guessing (i.e. calculating from DMA mask) whether we need bypass
or not on NPU devices. Since DMA setup happens in very rare occasion,
this will not slow down booting or VFIO start/stop much.
This renames pnv_npu_disable_bypass to pnv_npu_dma_set_32 to make it
more clear what the function really does which is programming 32bit
table address to the TVT ("disabling bypass" means writing zeroes to
the TVT).
This removes pnv_npu_dma_set_bypass() from pnv_npu_ioda_fixup() as
the DMA configuration on NPU does not matter until dma_set_mask() is
called on GPU and that will do the NPU DMA configuration.
This removes phb->dma_dev_setup initialization for NPU as
pnv_pci_ioda_dma_dev_setup is no-op for it anyway.
This stops using npe->tce_bypass_base as it never changes and values
other than zero are not supported.
Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Reviewed-by: Alistair Popple <alistair@popple.id.au>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
NPU PHB TCE Kill register is exactly the same as in the rest of POWER8
so let's reuse the existing code for NPU. The only bit missing is
a helper to reset the entire TCE cache so this moves such a helper
from NPU code and renames it.
Since pnv_npu_tce_invalidate() does really invalidate the entire cache,
this uses pnv_pci_ioda2_tce_invalidate_entire() directly for NPU.
This adds an explicit comment for workaround for invalidating NPU TCE
cache.
Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Reviewed-by: Alistair Popple <alistair@popple.id.au>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
This replaces magic constants for TCE Kill IODA2 register with macros.
Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
As in fact pnv_pci_ioda2_tce_invalidate_entire() invalidates TCEs for
the specific PE rather than the entire cache, rename it to
pnv_pci_ioda2_tce_invalidate_pe(). In later patches we will add
a proper pnv_pci_ioda2_tce_invalidate_entire().
Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
In current implementation, the PEs that are allocated or picked
from the reserved list are identified by PE number. The PE instance
has to be picked according to the PE number eventually. We have
same issue when PE is released.
For pnv_ioda_pick_m64_pe() and pnv_ioda_alloc_pe(), this returns
PE instance so that pnv_ioda_setup_bus_PE() can use the allocated
or reserved PE instance directly. Also, pnv_ioda_setup_bus_PE()
returns the reserved/allocated PE instance to be used in subsequent
patches. On the other hand, pnv_ioda_free_pe() uses PE instance
(not number) as its argument. No logical changes introduced.
Signed-off-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
Reviewed-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
In current implementation, the DMA32 segments required by one specific
PE isn't calculated with the information hold in the PE independently.
It conflicts with the PCI hotplug design: PE centralized, meaning the
PE's DMA32 segments should be calculated from the information hold in
the PE independently.
This introduces an array (@dma32_segmap) for every PHB to track the
DMA32 segmeng usage. Besides, this moves the logic calculating PE's
consumed DMA32 segments to pnv_pci_ioda1_setup_dma_pe() so that PE's
DMA32 segments are calculated/allocated from the information hold in
the PE (DMA32 weight). Also the logic is improved: we try to allocate
as much DMA32 segments as we can. It's acceptable that number of DMA32
segments less than the expected number are allocated.
Signed-off-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
Reviewed-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
PEs are put into PHB DMA32 list (phb->ioda.pe_dma_list) according
to their DMA32 weight. The PEs on the list are iterated to setup
their TCE32 tables at system booting time. The list is used for
once at boot time and no need to keep it.
This moves the logic calculating DMA32 weight of PHB and PE to
pnv_ioda_setup_dma() to drop PHB's DMA32 list. Also, every PE
traces the consumed DMA32 segment by @tce32_seg and @tce32_segcount
are useless and they're removed.
Signed-off-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
Reviewed-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Currently, there is one macro (TCE32_TABLE_SIZE) representing the
TCE table size for one DMA32 segment. The constant representing
the DMA32 segment size (1 << 28) is still used in the code.
This defines PNV_IODA1_DMA32_SEGSIZE representing one DMA32
segment size. the TCE table size can be calcualted when the page
has fixed 4KB size. So all the related calculation depends on one
macro (PNV_IODA1_DMA32_SEGSIZE). No logical changes introduced.
Signed-off-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
Reviewed-By: Alistair Popple <alistair@popple.id.au>
Reviewed-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
This renames pnv_pci_ioda_setup_dma_pe() to pnv_pci_ioda1_setup_dma_pe()
as it's the counter-part of IODA2's pnv_pci_ioda2_setup_dma_pe().
No logical changes introduced.
Signed-off-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
Reviewed-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
This enables M64 window on P7IOC, which has been enabled on PHB3.
Different from PHB3 where 16 M64 BARs are supported and each of
them can be owned by one particular PE# exclusively or divided
evenly to 256 segments, every P7IOC PHB has 16 M64 BARs and each
of them are divided to 8 segments. So every P7IOC PHB supports
128 M64 segments in total. P7IOC has M64DT, which helps mapping
one particular M64 segment# to arbitrary PE#. PHB3 doesn't have
M64DT, indicating that one M64 segment can only be pinned to the
fixed PE#.
In order to unified M64 support M64 on P7IOC and PHB3, we just
provide 128 M64 segments on every P7IOC PHB and each of them is
pinned to the fixed PE# by bypassing the function of M64DT. In
turn, we just need different phb->init_m64() for P7IOC and PHB3
and maps M64 segment in pnv_ioda_reserve_m64_pe() for P7IOC, most
of the code are shared by them.
Signed-off-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
Reviewed-by: Alistair Popple <alistair@popple.id.au>
Reviewed-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
This renames those functions picking PE number based on consumed
M64 segments, mapping M64 segments to PEs as those functions are
going to be shared by IODA1/IODA2 in next patch. No logical changes
introduced.
Signed-off-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
Reviewed-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
When unplugging PCI devices, their parent PEs might be offline.
The consumed M64 resource by the PEs should be released at that
time. As we track M32 segment consumption, this introduces an
array to the PHB to track the mapping between M64 segment and
PE number.
Note: M64 mapping isn't covered by pnv_ioda_setup_pe_seg() as
IODA2 doesn't support the mapping explicitly while it's supported
on IODA1. Until now, no M64 is supported on IODA1 in software.
Signed-off-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
Reviewed-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Currently, the IO and M32 segments are mapped to the corresponding
PE based on the windows of the parent bridge of PE's primary bus.
It's not going to work when the windows of root port or upstream
port of the PCIe switch behind root port are extended to PHB's
apertures in order to support hotplug in subsequent patch.
This fixes the issue by mapping IO and M32 segments based on the
resources of the PCI devices included in the PE, instead of the
windows of the parent bridge of the PE's primary bus.
Signed-off-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
Reviewed-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
pnv_ioda_setup_pe_seg() associates the IO and M32 segments with the
owner PE. The code mapping segments should be fixed and immune from
logic changes introduced to pnv_ioda_setup_pe_seg().
This moves the code mapping segments to helper pnv_ioda_setup_pe_res().
The data type for @rc is changed to "int64_t". Also, argument @hose is
removed from pnv_ioda_setup_pe() as it can be got from @pe. No functional
changes introduced.
Signed-off-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
Reviewed-By: Alistair Popple <alistair@popple.id.au>
Reviewed-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
There are two arrays for IO and M32 segment maps on every PHB.
The index of the arrays are segment number and the value stored
in the corresponding element is PE number, indicating the segment
is assigned to the PE. Initially, all elements in those two arrays
are zeroes, meaning all segments are assigned to PE#0. It's wrong.
This fixes the initial values in the elements of those two arrays
to IODA_INVALID_PE, meaning all segments aren't assigned to any
PE.
Signed-off-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
Reviewed-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
This changes the data type of PE number from "int" to "unsigned int"
in order to match the fact PE number is never negative:
* The number of PE to which the specified PCI device is attached.
* The PE number map for SRIOV VFs.
* The returned PE number from pnv_ioda_alloc_pe().
* The returned PE number from pnv_ioda2_pick_m64_pe().
Suggested-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Signed-off-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
Reviewed-By: Alistair Popple <alistair@popple.id.au>
Reviewed-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
This renames the fields related to PE number in "struct pnv_phb"
for better reflecting of their usages as Alexey suggested. No
logical changes introduced.
Signed-off-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
Reviewed-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
The last usage of pnv_phb::bdfn_to_pe() was removed in
ff57b454dd ("powerpc/eeh: Do probe on pci_dn"), so drop it.
Signed-off-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
Reviewed-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
This cleans up on below data struct instances to use tab instead of
space indent of statement to avoid complains from scripts/checkpatch.pl.
No logical changes introduced.
@pnv_pci_ioda_controller_ops
@pnv_npu_ioda_controller_ops
Signed-off-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
Reviewed-by: Daniel Axtens <dja@axtens.net>
Reviewed-by: Andrew Donnellan <andrew.donnellan@au1.ibm.com>
Acked-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
- eeh: Fix partial hotplug criterion from Gavin Shan
- mm: Clear the invalid slot information correctly from Aneesh Kumar K.V
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1
iQIcBAABAgAGBQJWzXquAAoJEFHr6jzI4aWADHsP/2lbwqz/vS3Ep4zlySHNvStL
/DrRN2TN35THZ59FPRxgEfeqPxTCXtbpD6zEXwD0gf6m39I2zArhaQMOHXMtVPvV
p0nAtwR0PX/PxlQTJDpHlg074vVAD7s3iuvad6oNQObLcXhoZ7wYtbStZ9Ithm4R
YfqZTelzsw+GfMuTYnvAQf5aoRYztUpy7OheaJbbDmSZgMFwF896ZPJnaG9rAOPE
xcSsRaSfHiUR2NE2ua1K5yya+1ilZqrZhib7QxXgzGuxoVa2AAiPR7Hpx2kX1Wm+
z0DqPXISzRbVf9zyLgWD3TpJ4OMHI/CYVW+t/Gx/yWCMfNcfavUrh0vPdHRVEPZu
zxmIUoI6yv7jQ6bcfdzR5s0Mr5pYWlUj5MZg2r8aGqloYcLPk5DiENg+c0QmKI05
kQPCBoQz2ezzJWAt1BYshkc+mlimv3ODaNWFP34Nc6kcDaSO6a0rhVOecvKuR6dv
UBNpeh5np1rKq1wX0ri0yAmnm//yXqe+bK0I8Ctipi0++e73sVJGzfFdVvXwEhhW
h+v1BkdgW8WK/xlH+JCPiXd5dfXrUeFI0D65Kgpb7IbFc9hcXDmp2Dv7+8zx/Wcl
L2NpuucSDxi+LHkE10QiypgLWSKjn9OSi8PLocqABNXG8uHxIp54jRfyViBNALXF
XlPveqTgpt7On3aa0qVh
=bk3U
-----END PGP SIGNATURE-----
Merge tag 'powerpc-4.5-4' into next
Pull in our current fixes from 4.5, in particular the "Fix Multi hit
ERAT" bug is causing folks some grief when testing next.
When PCI bus is unplugged during full hotplug for EEH recovery,
the platform PE instance (struct pnv_ioda_pe) isn't released and
it dereferences the stale PCI bus that has been released. It leads
to kernel crash when referring to the stale PCI bus.
This fixes the issue by correcting the PE's primary bus when it's
oneline at plugging time, in pnv_pci_dma_bus_setup() which is to
be called by pcibios_fixup_bus().
Cc: stable@vger.kernel.org # v4.1+
Reported-by: Andrew Donnellan <andrew.donnellan@au1.ibm.com>
Reported-by: Pradipta Ghosh <pradghos@in.ibm.com>
Signed-off-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
Tested-by: Andrew Donnellan <andrew.donnellan@au1.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
When M64 BAR is set to Single PE mode, the PE# assigned to VF could be
sparse.
This patch restructures the code to allocate sparse PE# for VFs when M64
BAR is set to Single PE mode. Also it rename the offset to pe_num_map to
reflect the content is the PE number.
Signed-off-by: Wei Yang <weiyang@linux.vnet.ibm.com>
Reviewed-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
Acked-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Each VF could have 6 BARs at most. When the total BAR size exceeds the
gate, after expanding it will also exhaust the M64 Window.
This patch limits the boundary by checking the total VF BAR size instead of
the individual BAR.
Signed-off-by: Wei Yang <weiyang@linux.vnet.ibm.com>
Reviewed-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
Acked-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
At the moment 64bit-prefetchable window can be maximum 64GB, which is
currently got from device tree. This means that in shared mode the maximum
supported VF BAR size is 64GB/256=256MB. While this size could exhaust the
whole 64bit-prefetchable window. This is a design decision to set a
boundary to 64MB of the VF BAR size. Since VF BAR size with 64MB would
occupy a quarter of the 64bit-prefetchable window, this is affordable.
This patch replaces magic limit of 64MB with "gate", which is 1/4 of the
M64 Segment Size(m64_segsize >> 2) and adds comment to explain the reason
for it.
Signed-off-by: Wei Yang <weiyang@linux.vnet.ibm.com>
Reviewed-by: Gavin Shan <gwshan@linux.vent.ibm.com>
Acked-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
In current implementation, when VF BAR is bigger than 64MB, it uses 4 M64
BARs in Single PE mode to cover the number of VFs required to be enabled.
By doing so, several VFs would be in one VF Group and leads to interference
between VFs in the same group.
And in this patch, m64_wins is renamed to m64_map, which means index number
of the M64 BAR used to map the VF BAR. Based on Gavin's comments. Also
makes sure the VF BAR size is bigger than 32MB when M64 BAR is used in
Single PE mode.
This patch changes the design by using one M64 BAR in Single PE mode for
one VF BAR. This gives absolute isolation for VFs.
Signed-off-by: Wei Yang <weiyang@linux.vnet.ibm.com>
Reviewed-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
Acked-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
The alignment of IOV BAR on PowerNV platform is the total size of the IOV
BAR. No matter whether the IOV BAR is extended with number of
roundup_pow_of_two(total_vfs) or number of max PE number (256), the total
size could be calculated by (vfs_expanded * VF_BAR_size).
This patch simplifies the pnv_pci_iov_resource_alignment() by removing the
first case.
Signed-off-by: Wei Yang <weiyang@linux.vnet.ibm.com>
Reviewed-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
Acked-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
On PHB3, we enable SRIOV devices by mapping IOV BAR with M64 BARs. If a
SRIOV device's IOV BAR is not 64bit-prefetchable, this is not assigned from
64bit prefetchable window, which means M64 BAR can't work on it.
The reason is PCI bridges support only 2 memory windows and the kernel code
programs bridges in the way that one window is 32bit-nonprefetchable and
the other one is 64bit-prefetchable. So if devices' IOV BAR is 64bit and
non-prefetchable, it will be mapped into 32bit space and therefore M64
cannot be used for it.
This patch makes this explicit and truncate IOV resource in this case to
save MMIO space.
Signed-off-by: Wei Yang <weiyang@linux.vnet.ibm.com>
Reviewed-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
Acked-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
P8+ hardware reports all errors on PE#0. This patch ensures PE#0 is
not assigned to NPU devices so that it can be used for EEH.
Signed-off-by: Alistair Popple <alistair@popple.id.au>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
The P8+ hardware supports four partitionable endpoints (PEs) however
the hardware reports all errors as occurring on PE#0. This means we
need to reserve this PE for error handling (EEH) and not assign it to
a NPU device, implying that some devices will need to share PEs.
This patch changes the PE assignment for NPU devices such that NPU
devices which connect to the same GPU are assigned to the same
PE#.
Signed-off-by: Alistair Popple <alistair@popple.id.au>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
The emulated NVLink PCI devices share the same IODA2 TCE tables but only
support a single TVT (instead of the normal two for PCI devices). This
requires the kernel to manually replace windows with either the bypass
or non-bypass window depending on what the driver has requested.
Unfortunately an incorrect optimisation was made in
pnv_pci_ioda_dma_set_mask() which caused updating of some NPU device PEs
to be skipped in certain configurations due to an incorrect assumption
that a NULL peer PE in the array indicated there were no more peers
present. This patch fixes the problem by ensuring all peer PEs are
updated.
Fixes: 5d2aa710e6 ("powerpc/powernv: Add support for Nvlink NPUs")
Signed-off-by: Alistair Popple <alistair@popple.id.au>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
NVLink is a high speed interconnect that is used in conjunction with a
PCI-E connection to create an interface between CPU and GPU that
provides very high data bandwidth. A PCI-E connection to a GPU is used
as the control path to initiate and report status of large data
transfers sent via the NVLink.
On IBM Power systems the NVLink processing unit (NPU) is similar to
the existing PHB3. This patch adds support for a new NPU PHB type. DMA
operations on the NPU are not supported as this patch sets the TCE
translation tables to be the same as the related GPU PCIe device for
each NVLink. Therefore all DMA operations are setup and controlled via
the PCIe device.
EEH is not presently supported for the NPU devices, although it may be
added in future.
Signed-off-by: Alistair Popple <alistair@popple.id.au>
Signed-off-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Move __raw_rm_writeq() from platforms/powernv/pci-ioda.c to
include/asm/io.h so that it can be used by other code.
Signed-off-by: Alistair Popple <alistair@popple.id.au>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
This commit removed the pcidev field from struct pci_dn as it was no
longer in use by the kernel. However to support finding the
association of Nvlink devices to GPU devices from the device-tree this
field is required.
This reverts commit 250c7b277c ("powerpc/pci: Remove unused struct
pci_dn.pcidev field").
Signed-off-by: Alistair Popple <alistair@popple.id.au>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
The name of PCI root bus's M64 resource isn't initialized properly.
When dumping "/proc/iomem", "<BAD>" is seen for those M64 resources
on PCI root buses.
~# cat /proc/iomem | grep -e "BAD"
3b0000000000-3b0fefffffff : <BAD>
3b1000000000-3b1fefffffff : <BAD>
3c0000000000-3c0fefffffff : <BAD>
3c1000000000-3c1fefffffff : <BAD>
3c2000000000-3c2fefffffff : <BAD>
This fixes the issue by setting the name of PCI root bus's M64
resource to that of PHB's device node full name. With the patch,
no "<BAD>" is seen from "/proc/iomem".
Signed-off-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Long ago, only in the lab, there was OPALv1 and OPALv2. Now there is
just OPALv3, with nobody ever expecting anything on pre-OPALv3 to
be cared about or supported by mainline kernels.
So, let's remove FW_FEATURE_OPALv3 and instead use FW_FEATURE_OPAL
exclusively.
Signed-off-by: Stewart Smith <stewart@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
The 32-bit TCE table initialization relies on the DMA window having a
size equal to a power of 2 (and checks for it explicitly). But
crashkernel= has no constraint that requires a power-of-2 be specified.
This causes the kdump kernel to fail to boot as none of the PCI devices
(including the disk controller) are successfully initialized.
After this change, the PCI devices successfully set up the 32-bit TCE
table and kdump succeeds.
Fixes: aca6913f55 ("powerpc/powernv/ioda2: Introduce helpers to allocate TCE pages")
Signed-off-by: Nishanth Aravamudan <nacc@linux.vnet.ibm.com>
Cc: stable@vger.kernel.org # 4.2
Tested-by: Jan Stancek <jstancek@redhat.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
When attempting to kdump with the 4.2 kernel, we see for each PCI
device:
pci 0003:01 : [PE# 000] Assign DMA32 space
pci 0003:01 : [PE# 000] Setting up 32-bit TCE table at 0..80000000
pci 0003:01 : [PE# 000] Failed to create 32-bit TCE table, err -22
PCI: Domain 0004 has 8 available 32-bit DMA segments
PCI: 4 PE# for a total weight of 70
pci 0004:01 : [PE# 002] Assign DMA32 space
pci 0004:01 : [PE# 002] Setting up 32-bit TCE table at 0..80000000
pci 0004:01 : [PE# 002] Failed to create 32-bit TCE table, err -22
pci 0004:0d : [PE# 005] Assign DMA32 space
pci 0004:0d : [PE# 005] Setting up 32-bit TCE table at 0..80000000
pci 0004:0d : [PE# 005] Failed to create 32-bit TCE table, err -22
pci 0004:0e : [PE# 006] Assign DMA32 space
pci 0004:0e : [PE# 006] Setting up 32-bit TCE table at 0..80000000
pci 0004:0e : [PE# 006] Failed to create 32-bit TCE table, err -22
pci 0004:10 : [PE# 008] Assign DMA32 space
pci 0004:10 : [PE# 008] Setting up 32-bit TCE table at 0..80000000
pci 0004:10 : [PE# 008] Failed to create 32-bit TCE table, err -22
and eventually the kdump kernel fails to boot as none of the PCI devices
(including the disk controller) are successfully initialized.
The EINVAL response is because the DMA window (the 2GB base window) is
larger than the kdump kernel's reserved memory (crashkernel=, in this
case specified to be 1024M). The check in question,
if ((window_size > memory_hotplug_max()) || !is_power_of_2(window_size))
is a valid sanity check for pnv_pci_ioda2_table_alloc_pages(), so adjust
the caller to pass in a smaller window size if our maximum memory value
is smaller than the DMA window.
After this change, the PCI devices successfully set up the 32-bit TCE
table and kdump succeeds.
The problem was seen on a Firestone machine originally.
Fixes: aca6913f55 ("powerpc/powernv/ioda2: Introduce helpers to allocate TCE pages")
Cc: stable@vger.kernel.org # 4.2
Signed-off-by: Nishanth Aravamudan <nacc@linux.vnet.ibm.com>
Reviewed-by: Alexey Kardashevskiy <aik@ozlabs.ru>
[mpe: Coding style pedantry, use u64, change the indentation]
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
- Support "hybrid" iommu/direct DMA ops for coherent_mask < dma_mask from Benjamin Herrenschmidt
- EEH fixes for SRIOV from Gavin
- Introduce rtas_get_sensor_fast() for IRQ handlers from Thomas Huth
- Use hardware RNG for arch_get_random_seed_* not arch_get_random_* from Paul Mackerras
- Seccomp filter support from Michael Ellerman
- opal_cec_reboot2() handling for HMIs & machine checks from Mahesh Salgaonkar
- Add powerpc timebase as a trace clock source from Naveen N. Rao
- Misc cleanups in the xmon, signal & SLB code from Anshuman Khandual
- Add an inline function to update POWER8 HID0 from Gautham R. Shenoy
- Fix pte_pagesize_index() crash on 4K w/64K hash from Michael Ellerman
- Drop support for 64K local store on 4K kernels from Michael Ellerman
- move dma_get_required_mask() from pnv_phb to pci_controller_ops from Andrew Donnellan
- Initialize distance lookup table from drconf path from Nikunj A Dadhania
- Enable RTC class support from Vaibhav Jain
- Disable automatically blocked PCI config from Gavin Shan
- Add LEDs driver for PowerNV platform from Vasant Hegde
- Fix endianness issues in the HVSI driver from Laurent Dufour
- Kexec endian fixes from Samuel Mendoza-Jonas
- Fix corrupted pdn list from Gavin Shan
- Fix fenced PHB caused by eeh_slot_error_detail() from Gavin Shan
- Freescale updates from Scott: Highlights include 32-bit memcpy/memset
optimizations, checksum optimizations, 85xx config fragments and updates,
device tree updates, e6500 fixes for non-SMP, and misc cleanup and minor
fixes.
- A ton of cxl updates & fixes:
- Add explicit precision specifiers from Rasmus Villemoes
- use more common format specifier from Rasmus Villemoes
- Destroy cxl_adapter_idr on module_exit from Johannes Thumshirn
- Destroy afu->contexts_idr on release of an afu from Johannes Thumshirn
- Compile with -Werror from Daniel Axtens
- EEH support from Daniel Axtens
- Plug irq_bitmap getting leaked in cxl_context from Vaibhav Jain
- Add alternate MMIO error handling from Ian Munsie
- Allow release of contexts which have been OPENED but not STARTED from Andrew Donnellan
- Remove use of macro DEFINE_PCI_DEVICE_TABLE from Vaishali Thakkar
- Release irqs if memory allocation fails from Vaibhav Jain
- Remove racy attempt to force EEH invocation in reset from Daniel Axtens
- Fix + cleanup error paths in cxl_dev_context_init from Ian Munsie
- Fix force unmapping mmaps of contexts allocated through the kernel api from Ian Munsie
- Set up and enable PSL Timebase from Philippe Bergheaud
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1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=n9+X
-----END PGP SIGNATURE-----
Merge tag 'powerpc-4.3-1' of git://git.kernel.org/pub/scm/linux/kernel/git/powerpc/linux
Pull powerpc updates from Michael Ellerman:
- support "hybrid" iommu/direct DMA ops for coherent_mask < dma_mask
from Benjamin Herrenschmidt
- EEH fixes for SRIOV from Gavin
- introduce rtas_get_sensor_fast() for IRQ handlers from Thomas Huth
- use hardware RNG for arch_get_random_seed_* not arch_get_random_*
from Paul Mackerras
- seccomp filter support from Michael Ellerman
- opal_cec_reboot2() handling for HMIs & machine checks from Mahesh
Salgaonkar
- add powerpc timebase as a trace clock source from Naveen N. Rao
- misc cleanups in the xmon, signal & SLB code from Anshuman Khandual
- add an inline function to update POWER8 HID0 from Gautham R. Shenoy
- fix pte_pagesize_index() crash on 4K w/64K hash from Michael Ellerman
- drop support for 64K local store on 4K kernels from Michael Ellerman
- move dma_get_required_mask() from pnv_phb to pci_controller_ops from
Andrew Donnellan
- initialize distance lookup table from drconf path from Nikunj A
Dadhania
- enable RTC class support from Vaibhav Jain
- disable automatically blocked PCI config from Gavin Shan
- add LEDs driver for PowerNV platform from Vasant Hegde
- fix endianness issues in the HVSI driver from Laurent Dufour
- kexec endian fixes from Samuel Mendoza-Jonas
- fix corrupted pdn list from Gavin Shan
- fix fenced PHB caused by eeh_slot_error_detail() from Gavin Shan
- Freescale updates from Scott: Highlights include 32-bit memcpy/memset
optimizations, checksum optimizations, 85xx config fragments and
updates, device tree updates, e6500 fixes for non-SMP, and misc
cleanup and minor fixes.
- a ton of cxl updates & fixes:
- add explicit precision specifiers from Rasmus Villemoes
- use more common format specifier from Rasmus Villemoes
- destroy cxl_adapter_idr on module_exit from Johannes Thumshirn
- destroy afu->contexts_idr on release of an afu from Johannes
Thumshirn
- compile with -Werror from Daniel Axtens
- EEH support from Daniel Axtens
- plug irq_bitmap getting leaked in cxl_context from Vaibhav Jain
- add alternate MMIO error handling from Ian Munsie
- allow release of contexts which have been OPENED but not STARTED
from Andrew Donnellan
- remove use of macro DEFINE_PCI_DEVICE_TABLE from Vaishali Thakkar
- release irqs if memory allocation fails from Vaibhav Jain
- remove racy attempt to force EEH invocation in reset from Daniel
Axtens
- fix + cleanup error paths in cxl_dev_context_init from Ian Munsie
- fix force unmapping mmaps of contexts allocated through the kernel
api from Ian Munsie
- set up and enable PSL Timebase from Philippe Bergheaud
* tag 'powerpc-4.3-1' of git://git.kernel.org/pub/scm/linux/kernel/git/powerpc/linux: (140 commits)
cxl: Set up and enable PSL Timebase
cxl: Fix force unmapping mmaps of contexts allocated through the kernel api
cxl: Fix + cleanup error paths in cxl_dev_context_init
powerpc/eeh: Fix fenced PHB caused by eeh_slot_error_detail()
powerpc/pseries: Cleanup on pci_dn_reconfig_notifier()
powerpc/pseries: Fix corrupted pdn list
powerpc/powernv: Enable LEDS support
powerpc/iommu: Set default DMA offset in dma_dev_setup
cxl: Remove racy attempt to force EEH invocation in reset
cxl: Release irqs if memory allocation fails
cxl: Remove use of macro DEFINE_PCI_DEVICE_TABLE
powerpc/powernv: Fix mis-merge of OPAL support for LEDS driver
powerpc/powernv: Reset HILE before kexec_sequence()
powerpc/kexec: Reset secondary cpu endianness before kexec
powerpc/hvsi: Fix endianness issues in the HVSI driver
leds/powernv: Add driver for PowerNV platform
powerpc/powernv: Create LED platform device
powerpc/powernv: Add OPAL interfaces for accessing and modifying system LED states
powerpc/powernv: Fix the log message when disabling VF
cxl: Allow release of contexts which have been OPENED but not STARTED
...
Commit e91c25111a "powerpc/iommu: Cleanup setting of DMA base/offset"
expects that the default DMA offset is set from pnv_ioda_setup_bus_dma()
which is correct unless it is SRIOV where the code flow is different -
at the moment when pnv_ioda_setup_bus_dma() is called, PCI devices for
VFs are not created yet.
This adds missing set_dma_offset() to pnv_pci_ioda_dma_dev_setup() to
cover the case of SRIOV.
Note that we still need set_dma_offset() in pnv_ioda_setup_bus_dma() as
at the boot time pnv_pci_ioda_dma_dev_setup() is called when no PE was
created yet, this happens at the PHB fixup stage.
Fixes: e91c25111a ("powerpc/iommu: Cleanup setting of DMA base/offset")
Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Reviewed-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
On powernv platform, IOV BAR would be shifted if necessary. While the log
message is not correct when disabling VFs.
This patch fixes this by print correct message based on the offset value.
Signed-off-by: Wei Yang <weiyang@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Simplify the dma_get_required_mask call chain by moving it from pnv_phb to
pci_controller_ops, similar to commit 763d2d8df1 ("powerpc/powernv:
Move dma_set_mask from pnv_phb to pci_controller_ops").
Previous call chain:
0) call dma_get_required_mask() (kernel/dma.c)
1) call ppc_md.dma_get_required_mask, if it exists. On powernv, that
points to pnv_dma_get_required_mask() (platforms/powernv/setup.c)
2) device is PCI, therefore call pnv_pci_dma_get_required_mask()
(platforms/powernv/pci.c)
3) call phb->dma_get_required_mask if it exists
4) it only exists in the ioda case, where it points to
pnv_pci_ioda_dma_get_required_mask() (platforms/powernv/pci-ioda.c)
New call chain:
0) call dma_get_required_mask() (kernel/dma.c)
1) device is PCI, therefore call pci_controller_ops.dma_get_required_mask
if it exists
2) in the ioda case, that points to pnv_pci_ioda_dma_get_required_mask()
(platforms/powernv/pci-ioda.c)
In the p5ioc2 case, the call chain remains the same -
dma_get_required_mask() does not find either a ppc_md call or
pci_controller_ops call, so it calls __dma_get_required_mask().
Signed-off-by: Andrew Donnellan <andrew.donnellan@au1.ibm.com>
Reviewed-by: Daniel Axtens <dja@axtens.net>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
The existing code stores the amount of memory allocated for a TCE table.
At the moment it uses @offset which is a virtual offset in the TCE table
which is only correct for a one level tables and it does not include
memory allocated for intermediate levels. When multilevel TCE table is
requested, WARN_ON in tce_iommu_create_table() prints a warning.
This adds an additional counter to pnv_pci_ioda2_table_do_alloc_pages()
to count actually allocated memory.
Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
When releasing PE for SRIOV VF, the PE is forced to be frozen
wrongly. When the same PE is picked for another VF, it won't
work anyhow. The patch fixes the issue by unfreezing, not
freezing the VF PE when releasing it.
Signed-off-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
The PELTV of PF PE should include VF PE, which is missed by current
code, so that the VF PE is frozen automatically when freezing PF PE.
The patch fixes the PELTV of PF PE to include VF PE.
Signed-off-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
On PHB3, PE might be reserved in advance to reflect the M64 segments
consumed by the PE according to M64 BARs (exclude VF BARs) of the PCI
devices included in the PE. The PE is picked based on M64 BARs instead
of the bridge's M64 windows, which might include VF BARs. Otherwise,
wrong PE could be picked.
The patch calculates the used M64 segments and PE numbers according to
the M64 BARs, excluding VF BARs, of PCI devices in one particular PE,
instead of the bridge's M64 windows. Then the right PE number is picked.
Signed-off-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
The patch changes the type of last argument of pnv_ioda_setup_bus_PE()
and phb::pick_m64_pe() to boolean. No functional change.
Signed-off-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
On PHB3, some PEs might be reserved in advance to reflect the M64
segments consumed by those PEs. We're reserving PEs based on the
M64 window of root port, which might contain VF BAR. The PEs for
VFs are allocated dynamically, not reserved based on the consumed
M64 segments. So the M64 window of root port isn't reliable for
the task. Instead, we go through M64 BARs (VF BARs excluded) of
PCI devices under the specified root bus and reserve PEs accordingly,
as the patch does.
Signed-off-by: Gavin Shan <gwshan@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
