Currently the KMD is using enum i915_cache_level to set caching policy for
buffer objects. This is flaky because the PAT index which really controls
the caching behavior in PTE has far more levels than what's defined in the
enum. In addition, the PAT index is platform dependent, having to translate
between i915_cache_level and PAT index is not reliable, and makes the code
more complicated.
From UMD's perspective there is also a necessity to set caching policy for
performance fine tuning. It's much easier for the UMD to directly use PAT
index because the behavior of each PAT index is clearly defined in Bspec.
Having the abstracted i915_cache_level sitting in between would only cause
more ambiguity. PAT is expected to work much like MOCS already works today,
and by design userspace is expected to select the index that exactly
matches the desired behavior described in the hardware specification.
For these reasons this patch replaces i915_cache_level with PAT index. Also
note, the cache_level is not completely removed yet, because the KMD still
has the need of creating buffer objects with simple cache settings such as
cached, uncached, or writethrough. For kernel objects, cache_level is used
for simplicity and backward compatibility. For Pre-gen12 platforms PAT can
have 1:1 mapping to i915_cache_level, so these two are interchangeable. see
the use of LEGACY_CACHELEVEL.
One consequence of this change is that gen8_pte_encode is no longer working
for gen12 platforms due to the fact that gen12 platforms has different PAT
definitions. In the meantime the mtl_pte_encode introduced specfically for
MTL becomes generic for all gen12 platforms. This patch renames the MTL
PTE encode function into gen12_pte_encode and apply it to all gen12. Even
though this change looks unrelated, but separating them would temporarily
break gen12 PTE encoding, thus squash them in one patch.
Special note: this patch changes the way caching behavior is controlled in
the sense that some objects are left to be managed by userspace. For such
objects we need to be careful not to change the userspace settings.There
are kerneldoc and comments added around obj->cache_coherent, cache_dirty,
and how to bypass the checkings by i915_gem_object_has_cache_level. For
full understanding, these changes need to be looked at together with the
two follow-up patches, one disables the {set|get}_caching ioctl's and the
other adds set_pat extension to the GEM_CREATE uAPI.
Bspec: 63019
Cc: Chris Wilson <chris.p.wilson@linux.intel.com>
Signed-off-by: Fei Yang <fei.yang@intel.com>
Reviewed-by: Andi Shyti <andi.shyti@linux.intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Andi Shyti <andi.shyti@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230509165200.1740-3-fei.yang@intel.com
This patch is a preparation for replacing enum i915_cache_level with PAT
index. Caching policy for buffer objects is set through the PAT index in
PTE, the old i915_cache_level is not sufficient to represent all caching
modes supported by the hardware.
Preparing the transition by adding some platform dependent data structures
and helper functions to translate the cache_level to pat_index.
cachelevel_to_pat: a platform dependent array mapping cache_level to
pat_index.
max_pat_index: the maximum PAT index recommended in hardware specification
Needed for validating the PAT index passed in from user
space.
i915_gem_get_pat_index: function to convert cache_level to PAT index.
obj_to_i915(obj): macro moved to header file for wider usage.
I915_MAX_CACHE_LEVEL: upper bound of i915_cache_level for the
convenience of coding.
Cc: Chris Wilson <chris.p.wilson@linux.intel.com>
Cc: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Fei Yang <fei.yang@intel.com>
Reviewed-by: Andi Shyti <andi.shyti@linux.intel.com>
Reviewed-by: Andrzej Hajda <andrzej.hajda@intel.com>
Signed-off-by: Andi Shyti <andi.shyti@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230509165200.1740-2-fei.yang@intel.com
MTL currently uses gen8_ppgtt_insert_huge when managing huge pages.
This is because MTL reports as not supporting 64K pages, or more
accurately, the system that reports whether a platform has 64K pages
reports false for MTL. This is only half correct, as the 64K page support
reporting system only cares about 64K page support for LMEM, which MTL
doesn't have.
MTL should be using xehpsdv_ppgtt_insert_huge. However, simply changing
over to using that manager doesn't resolve the issue because MTL is
expecting the virtual address space for the page table to be flushed after
initialization, so we must also add a flush statement there.
Signed-off-by: Jonathan Cavitt <jonathan.cavitt@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Signed-off-by: Andrzej Hajda <andrzej.hajda@intel.com>
Reviewed-by: Andi Shyti <andi.shyti@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230425-hugepage-migrate-v8-2-7868d54eaa27@intel.com
It turns out that on production DG2/ATS HW we should have support for
PS64. This feature allows to provide a 64K TLB hint at the PTE level,
which is a lot more flexible than the current method of enabling 64K GTT
pages for the entire page-table, since that leads to all kinds of
annoying restrictions, as documented in:
commit caa574ffc4
Author: Matthew Auld <matthew.auld@intel.com>
Date: Sat Feb 19 00:17:49 2022 +0530
drm/i915/uapi: document behaviour for DG2 64K support
On discrete platforms like DG2, we need to support a minimum page size
of 64K when dealing with device local-memory. This is quite tricky for
various reasons, so try to document the new implicit uapi for this.
With PS64, we can now drop the 2M GTT alignment restriction, and instead
only require 64K or larger when dealing with lmem. We still use the
compact-pt layout when possible, but only when we are certain that this
doesn't interfere with userspace.
Note that this is a change in uAPI behaviour, but hopefully shouldn't be
a concern (IGT is at least able to autodetect the alignment), since we
are only making the GTT alignment constraint less restrictive.
Based on a patch from CQ Tang.
v2: update the comment wrt scratch page
v3: (Nirmoy)
- Fix the selftest to actually use the random size, plus some comment
improvements, also drop the rem stuff.
Reported-by: Michal Mrozek <michal.mrozek@intel.com>
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Cc: Stuart Summers <stuart.summers@intel.com>
Cc: Jordan Justen <jordan.l.justen@intel.com>
Cc: Yang A Shi <yang.a.shi@intel.com>
Cc: Nirmoy Das <nirmoy.das@intel.com>
Cc: Niranjana Vishwanathapura <niranjana.vishwanathapura@intel.com>
Reviewed-by: Nirmoy Das <nirmoy.das@intel.com>
Acked-by: Michal Mrozek <michal.mrozek@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20221004114915.221708-1-matthew.auld@intel.com
When introducing asynchronous unbinding, the vma itself may no longer
be alive when the actual binding or unbinding takes place.
Update the gtt i915_vma_ops accordingly to take a struct i915_vma_resource
instead of a struct i915_vma for the bind_vma() and unbind_vma() ops.
Similarly change the insert_entries() op for struct i915_address_space.
Replace a couple of i915_vma_snapshot members with their newly introduced
i915_vma_resource counterparts, since they have the same lifetime.
Also make sure to avoid changing the struct i915_vma_flags (in particular
the bind flags) async. That should now only be done sync under the
vm mutex.
v2:
- Update the vma_res::bound_flags when binding to the aliased ggtt
v6:
- Remove I915_VMA_ALLOC_BIT (Matthew Auld)
- Change some members of struct i915_vma_resource from unsigned long to u64
(Matthew Auld)
v7:
- Fix vma resource size parameters to be u64 rather than unsigned long
(Matthew Auld)
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220110172219.107131-3-thomas.hellstrom@linux.intel.com
On some platforms the hw has dropped support for 4K GTT pages when
dealing with LMEM, and due to the design of 64K GTT pages in the hw, we
can only mark the *entire* page-table as operating in 64K GTT mode,
since the enable bit is still on the pde, and not the pte. And since we
we still need to allow 4K GTT pages for SMEM objects, we can't have a
"normal" 4K page-table with scratch pointing to LMEM, since that's
undefined from the hw pov. The simplest solution is to just move the 64K
scratch page to SMEM on such platforms and call it a day, since that
should work for all configurations.
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Signed-off-by: Ramalingam C <ramalingam.c@intel.com>
Reviewed-by: Thomas Hellstrom <thomas.hellstrom@linux.intel.com>
Reviewed-by: Andi Shyti <andi.shyti@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20211208141613.7251-4-ramalingam.c@intel.com
Certain functions within i915 uses macros that are defined for
specific architectures by the mmu, such as _PAGE_RW and _PAGE_PRESENT
(Some architectures don't even have these macros defined, like ARM64).
Instead of re-using bits defined for the CPU, we should use bits
defined for i915. This patch introduces two new 64 bit macros,
GEN8_PAGE_PRESENT and GEN8_PAGE_RW, to check for bits 0 and 1 and, to
replace all occurrences of _PAGE_RW and _PAGE_PRESENT within i915.
v2(Michael Cheng): Use GEN8_ instead of I915_
Signed-off-by: Michael Cheng <michael.cheng@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
[ Move defines together with other GEN8 defines ]
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20211206215245.513677-2-michael.cheng@intel.com
We really only need memcpy restore for objects that affect the
operability of the migrate context. That is, primarily the page-table
objects of the migrate VM.
Add an object flag, I915_BO_ALLOC_PM_EARLY for objects that need early
restores using memcpy and a way to assign LMEM page-table object flags
to be used by the vms.
Restore objects without this flag with the gpu blitter and only objects
carrying the flag using TTM memcpy.
Initially mark the migrate, gt, gtt and vgpu vms to use this flag, and
defer for a later audit which vms actually need it. Most importantly, user-
allocated vms with pinned page-table objects can be restored using the
blitter.
Performance-wise memcpy restore is probably as fast as gpu restore if not
faster, but using gpu restore will help tackling future restrictions in
mappable LMEM size.
v4:
- Don't mark the aliasing ppgtt page table flags for early resume, but
rather the ggtt page table flags as intended. (Matthew Auld)
- The check for user buffer objects during early resume is pointless, since
they are never marked I915_BO_ALLOC_PM_EARLY. (Matthew Auld)
v5:
- Mark GuC LMEM objects with I915_BO_ALLOC_PM_EARLY to have them restored
before we fire up the migrate context.
Cc: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210922062527.865433-8-thomas.hellstrom@linux.intel.com
We skip filling out the pt with scratch entries if the va range covers
the entire pt, since we later have to fill it with the PTEs for the
object pages anyway. However this might leave open a small window where
the PTEs don't point to anything valid for the HW to consume.
When for example using 2M GTT pages this fill_px() showed up as being
quite significant in perf measurements, and ends up being completely
wasted since we ignore the pt and just use the pde directly.
Anyway, currently we have our PTE construction split between alloc and
insert, which is probably slightly iffy nowadays, since the alloc
doesn't actually allocate anything anymore, instead it just sets up the
page directories and points the PTEs at the scratch page. Later when we
do the insert step we re-program the PTEs again. Better might be to
squash the alloc and insert into a single step, then bringing back this
optimisation(along with some others) should be possible.
Fixes: 1482667324 ("drm/i915: Only initialize partially filled pagetables")
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Jon Bloomfield <jon.bloomfield@intel.com>
Cc: Chris Wilson <chris.p.wilson@intel.com>
Cc: Daniel Vetter <daniel@ffwll.ch>
Cc: <stable@vger.kernel.org> # v4.15+
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20210713130431.2392740-1-matthew.auld@intel.com
This was done by the following semantic patch:
@@ expression i915; @@
- INTEL_GEN(i915)
+ GRAPHICS_VER(i915)
@@ expression i915; expression E; @@
- INTEL_GEN(i915) >= E
+ GRAPHICS_VER(i915) >= E
@@ expression dev_priv; expression E; @@
- !IS_GEN(dev_priv, E)
+ GRAPHICS_VER(dev_priv) != E
@@ expression dev_priv; expression E; @@
- IS_GEN(dev_priv, E)
+ GRAPHICS_VER(dev_priv) == E
@@
expression dev_priv;
expression from, until;
@@
- IS_GEN_RANGE(dev_priv, from, until)
+ IS_GRAPHICS_VER(dev_priv, from, until)
@def@
expression E;
identifier id =~ "^gen$";
@@
- id = GRAPHICS_VER(E)
+ ver = GRAPHICS_VER(E)
@@
identifier def.id;
@@
- id
+ ver
It also takes care of renaming the variable we assign to GRAPHICS_VER()
so to use "ver" rather than "gen".
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210605155356.4183026-2-lucas.demarchi@intel.com
We need to generalise our accessor for the page directories and tables from
using the simple kmap_atomic to support local memory, and this setup
must be done on acquisition of the backing storage prior to entering
fence execution contexts. Here we replace the kmap with the object
mapping code that for simple single page shmemfs object will return a
plain kmap, that is then kept for the lifetime of the page directory.
Note that keeping the mapping around is a potential concern here, since
while the vma is pinned the mapping remains there for the PDs
underneath, or at least until the used_count reaches zero, at which
point we can safely destroy the mapping. For 32b this will be even worse
since the address space is more limited, but since this change mostly
impacts full ppGTT platforms, the justification is that for modern
platforms we shouldn't care too much about 32b.
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210427085417.120246-3-matthew.auld@intel.com
Our code analyzer reported a double free bug.
In gen8_preallocate_top_level_pdp, pde and pde->pt.base are allocated
via alloc_pd(vm) with one reference. If pin_pt_dma() failed, pde->pt.base
is freed by i915_gem_object_put() with a reference dropped. Then free_pd
calls free_px() defined in intel_ppgtt.c, which calls i915_gem_object_put()
to put pde->pt.base again.
As pde->pt.base is protected by refcount, so the second put will not free
pde->pt.base actually. But, maybe it is better to remove the first put?
Fixes: 82adf90113 ("drm/i915/gt: Shrink i915_page_directory's slab bucket")
Signed-off-by: Lv Yunlong <lyl2019@mail.ustc.edu.cn>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210426124340.4238-1-lyl2019@mail.ustc.edu.cn
kmalloc uses power-of-two slab buckets for small allocations (up to a
few pages). Since i915_page_directory is a page of pointers, plus a
couple more, this is rounded up to 8K, and we waste nearly 50% of that
allocation. Long terms this leads to poor memory utilisation, bloating
the kernel footprint, but the problem is exacerbated by our conservative
preallocation scheme for binding VMA. As we are required to allocate all
levels for each vma just in case we need to insert them upon binding,
this leads to a large multiplication factor for a single page vma. By
halving the allocation we need for the page directory structure, we
halve the impact of that factor, bringing workloads that once fitted into
memory, hopefully back to fitting into memory.
We maintain the split between i915_page_directory and i915_page_table as
we only need half the allocation for the lowest, most populous, level.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Matthew Auld <matthew.auld@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200729164219.5737-3-chris@chris-wilson.co.uk
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
The GEM object is grossly overweight for the practicality of tracking
large numbers of individual pages, yet it is currently our only
abstraction for tracking DMA allocations. Since those allocations need
to be reserved upfront before an operation, and that we need to break
away from simple system memory, we need to ditch using plain struct page
wrappers.
In the process, we drop the WC mapping as we ended up clflushing
everything anyway due to various issues across a wider range of
platforms. Though in a future step, we need to drop the kmap_atomic
approach which suggests we need to pre-map all the pages and keep them
mapped.
v2: Verify our large scratch page is suitably DMA aligned; and manually
clear the scratch since we are allocating plain struct pages full of
prior content.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Matthew Auld <matthew.auld@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200729164219.5737-2-chris@chris-wilson.co.uk
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
We need to make the DMA allocations used for page directories to be
performed up front so that we can include those allocations in our
memory reservation pass. The downside is that we have to assume the
worst case, even before we know the final layout, and always allocate
enough page directories for this object, even when there will be overlap.
This unfortunately can be quite expensive, especially as we have to
clear/reset the page directories and DMA pages, but it should only be
required during early phases of a workload when new objects are being
discovered, or after memory/eviction pressure when we need to rebind.
Once we reach steady state, the objects should not be moved and we no
longer need to preallocating the pages tables.
It should be noted that the lifetime for the page directories DMA is
more or less decoupled from individual fences as they will be shared
across objects across timelines.
v2: Only allocate enough PD space for the PTE we may use, we do not need
to allocate PD that will be left as scratch.
v3: Store the shift unto the first PD level to encapsulate the different
PTE counts for gen6/gen8.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Matthew Auld <matthew.auld@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200729164219.5737-1-chris@chris-wilson.co.uk
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
On TGL, bits 2-4 in the GGTT PTE are not ignored anymore and are
instead used for some extra VT-d capabilities. We don't (yet?) have
support for those capabilities, but, given that we shared the pte_encode
function betweed GGTT and PPGTT, we still set those bits to the PPGTT
PPAT values. The DMA engine gets very confused when those bits are
set while the iommu is enabled, leading to errors. E.g. when loading
the GuC we get:
[ 9.796218] DMAR: DRHD: handling fault status reg 2
[ 9.796235] DMAR: [DMA Write] Request device [00:02.0] PASID ffffffff fault addr 0 [fault reason 02] Present bit in context entry is clear
[ 9.899215] [drm:intel_guc_fw_upload [i915]] *ERROR* GuC firmware signature verification failed
To fix this, just have dedicated gen8_pte_encode function per type of
gtt. Also, explicitly set vm->pte_encode for gen8_ppgtt, even if we
don't use it, to make sure we don't accidentally assign it to the GGTT
one, like we do for gen6_ppgtt, in case we need it in the future.
Reported-by: "Sodhi, Vunny" <vunny.sodhi@intel.com>
Signed-off-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Cc: Matthew Auld <matthew.auld@intel.com>
Cc: Michal Wajdeczko <michal.wajdeczko@intel.com>
Reviewed-by: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20200226185657.26445-1-daniele.ceraolospurio@intel.com
If we create a rather large userptr object(e.g 1ULL << 32) we might
shift past the type-width of num_pages: (int)num_pages << PAGE_SHIFT,
resulting in a totally bogus sg_table, which fortunately will eventually
manifest as:
gen8_ppgtt_insert_huge:463 GEM_BUG_ON(iter->sg->length < page_size)
kernel BUG at drivers/gpu/drm/i915/gt/gen8_ppgtt.c:463!
v2: more unsigned long
prefer I915_GTT_PAGE_SIZE
Fixes: 5cc9ed4b9a ("drm/i915: Introduce mapping of user pages into video memory (userptr) ioctl")
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: https://patchwork.freedesktop.org/patch/msgid/20200117132413.1170563-2-matthew.auld@intel.com
