| Age | Commit message (Collapse) | Author |
|---|
|
This commit enables basic support for Hopper GPUs, and is intended
primarily as a base supporting Blackwell GPUs, which reuse most of
the code added here.
Advanced features such as Confidential Compute are not supported.
Beyond a few miscellaneous register moves and HW class ID plumbing,
the bulk of the changes implemented here are to support the GSP-RM
boot sequence used on Hopper/Blackwell GPUs, as well as a new page
table layout.
There should be no changes here that impact prior GPUs.
Signed-off-by: Ben Skeggs <bskeggs@nvidia.com>
Co-developed-by: Timur Tabi <ttabi@nvidia.com>
Signed-off-by: Timur Tabi <ttabi@nvidia.com>
Reviewed-by: Dave Airlie <airlied@redhat.com>
Reviewed-by: Timur Tabi <ttabi@nvidia.com>
Tested-by: Timur Tabi <ttabi@nvidia.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
|
|
GPUs exist now with a 64-bit BAR0, which mean that BAR1 and BAR2's
indices (as passed to pci_resource_len() etc) are bumped up by one.
Modify nvkm_device.resource_addr/size() to take an enum instead of
an integer bar index, and take IORESOURCE_MEM_64 into account when
translating to the "raw" bar id.
[airlied: fixup ERR_PTR]
Signed-off-by: Ben Skeggs <bskeggs@nvidia.com>
Reviewed-by: Dave Airlie <airlied@redhat.com>
Reviewed-by: Timur Tabi <ttabi@nvidia.com>
Tested-by: Timur Tabi <ttabi@nvidia.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
|
|
GH100/GBxxx have moved the register that controls where in VRAM the
the BAR0 NV_PRAMIN window points.
Add a HAL for this, as the BAR0 window is needed for BAR2 bootstrap.
Signed-off-by: Ben Skeggs <bskeggs@nvidia.com>
Reviewed-by: Dave Airlie <airlied@redhat.com>
Reviewed-by: Timur Tabi <ttabi@nvidia.com>
Tested-by: Timur Tabi <ttabi@nvidia.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
|
|
Running a lot of VK CTS in parallel against nouveau, once every
few hours you might see something like this crash.
BUG: kernel NULL pointer dereference, address: 0000000000000008
PGD 8000000114e6e067 P4D 8000000114e6e067 PUD 109046067 PMD 0
Oops: 0000 [#1] PREEMPT SMP PTI
CPU: 7 PID: 53891 Comm: deqp-vk Not tainted 6.8.0-rc6+ #27
Hardware name: Gigabyte Technology Co., Ltd. Z390 I AORUS PRO WIFI/Z390 I AORUS PRO WIFI-CF, BIOS F8 11/05/2021
RIP: 0010:gp100_vmm_pgt_mem+0xe3/0x180 [nouveau]
Code: c7 48 01 c8 49 89 45 58 85 d2 0f 84 95 00 00 00 41 0f b7 46 12 49 8b 7e 08 89 da 42 8d 2c f8 48 8b 47 08 41 83 c7 01 48 89 ee <48> 8b 40 08 ff d0 0f 1f 00 49 8b 7e 08 48 89 d9 48 8d 75 04 48 c1
RSP: 0000:ffffac20c5857838 EFLAGS: 00010202
RAX: 0000000000000000 RBX: 00000000004d8001 RCX: 0000000000000001
RDX: 00000000004d8001 RSI: 00000000000006d8 RDI: ffffa07afe332180
RBP: 00000000000006d8 R08: ffffac20c5857ad0 R09: 0000000000ffff10
R10: 0000000000000001 R11: ffffa07af27e2de0 R12: 000000000000001c
R13: ffffac20c5857ad0 R14: ffffa07a96fe9040 R15: 000000000000001c
FS: 00007fe395eed7c0(0000) GS:ffffa07e2c980000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 0000000000000008 CR3: 000000011febe001 CR4: 00000000003706f0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
Call Trace:
...
? gp100_vmm_pgt_mem+0xe3/0x180 [nouveau]
? gp100_vmm_pgt_mem+0x37/0x180 [nouveau]
nvkm_vmm_iter+0x351/0xa20 [nouveau]
? __pfx_nvkm_vmm_ref_ptes+0x10/0x10 [nouveau]
? __pfx_gp100_vmm_pgt_mem+0x10/0x10 [nouveau]
? __pfx_gp100_vmm_pgt_mem+0x10/0x10 [nouveau]
? __lock_acquire+0x3ed/0x2170
? __pfx_gp100_vmm_pgt_mem+0x10/0x10 [nouveau]
nvkm_vmm_ptes_get_map+0xc2/0x100 [nouveau]
? __pfx_nvkm_vmm_ref_ptes+0x10/0x10 [nouveau]
? __pfx_gp100_vmm_pgt_mem+0x10/0x10 [nouveau]
nvkm_vmm_map_locked+0x224/0x3a0 [nouveau]
Adding any sort of useful debug usually makes it go away, so I hand
wrote the function in a line, and debugged the asm.
Every so often pt->memory->ptrs is NULL. This ptrs ptr is set in
the nv50_instobj_acquire called from nvkm_kmap.
If Thread A and Thread B both get to nv50_instobj_acquire around
the same time, and Thread A hits the refcount_set line, and in
lockstep thread B succeeds at refcount_inc_not_zero, there is a
chance the ptrs value won't have been stored since refcount_set
is unordered. Force a memory barrier here, I picked smp_mb, since
we want it on all CPUs and it's write followed by a read.
v2: use paired smp_rmb/smp_wmb.
Cc: <stable@vger.kernel.org>
Fixes: be55287aa5ba ("drm/nouveau/imem/nv50: embed nvkm_instobj directly into nv04_instobj")
Signed-off-by: Dave Airlie <airlied@redhat.com>
Signed-off-by: Danilo Krummrich <dakr@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240411011510.2546857-1-airlied@gmail.com
|
|
- Valid VRAM regions are read from GSP-RM, and used to construct our MM
- BAR1/BAR2 VMMs modified to be shared with RM
- Client VMMs have RM VASPACE objects created for them
- Adds FBSR to backup system objects in VRAM across suspend
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230918202149.4343-37-skeggsb@gmail.com
|
|
- move suspend/resume paths to HW-specific code
- allow (future) RM paths to be based on nv50_instmem
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230918202149.4343-15-skeggsb@gmail.com
|
|
Will be used to init client-allocated USERD to default values.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Reviewed-by: Lyude Paul <lyude@redhat.com>
|
|
Fix following coccicheck warning:
drivers/gpu/drm/nouveau/nvkm/subdev/instmem/nv50.c:316:11-12:
WARNING this kind of initialization is deprecated.
`void *map = map` has the same form of
uninitialized_var() macro. I remove the redundant assignement. It has
been tested with gcc (Debian 8.3.0-6) 8.3.0.
The patch which removed uninitialized_var() is:
https://lore.kernel.org/all/20121028102007.GA7547@gmail.com/
And there is very few "/* GCC */" comments in the Linux kernel code now.
Signed-off-by: Guo Zhengkui <guozhengkui@vivo.com>
Reviewed-by: Lyude Paul <lyude@redhat.com>
Signed-off-by: Lyude Paul <lyude@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220228142352.18006-1-guozhengkui@vivo.com
|
|
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Reviewed-by: Lyude Paul <lyude@redhat.com>
|
|
nvkm_subdev.mutex is going away.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Reviewed-by: Lyude Paul <lyude@redhat.com>
|
|
Various structures are accessed by the GPU through BAR2 for some reason
on newer GPUs. This commit makes it more convenient to handle.
Will be used for GP100- fault buffers, and GV100- fault method buffers.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
|
|
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
|
|
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
|
|
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
|
|
Map flags (access, kind, etc) are currently defined in either the VMA,
or the memory object, which turns out to not be ideal for things like
suballocated buffers, etc.
These will become per-map flags instead, so we need to support passing
these arguments in nvkm_memory_map().
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
|
|
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
|
|
Before: "imem: init completed in 299277us"
After: "imem: init completed in 11574us"
Suspend from Fedora 26 gnome desktop on GP102.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
|
|
These will require slow-path access during suspend/resume.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
|
|
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
|
|
A good deal of the structures we map into here aren't accessed very often
at all, and Fedora 26 has exposed an issue where after creating a heap of
channels, BAR2 space would run out, and we'd need to make use of the slow
path while accessing important structures like page tables.
This implements an LRU on BAR2 space, which allows eviction of mappings
that aren't currently needed, to make space for other objects.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
|
|
Another piece of solving the "GP100 BAR2 VMM bootstrap" puzzle.
Without doing this, we'd attempt to write PDEs for the lower page table
levels through BAR2 before BAR2 access has been fully initialised.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
|
|
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
|
|
This is not as simple as it was for earlier GPUs, due to the need to swap
accessor functions depending on whether BAR2 is usable or not.
We were previously protected by nvkm_instobj's accessor functions keeping
an object mapped permanently, with some unclear magic that managed to hit
the slow-path where needed even if an object was marked as mapped.
That's been replaced here by reference counting maps (some objects, like
page tables can be accessed concurrently), and swapping the functions as
necessary.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
|
|
This is to simplify upcoming changes. The slow-path is something that
currently occurs during bootstrap of the BAR2 VMM, while backing up an
object during suspend/resume, or when BAR2 address space runs out.
The latter is a real problem that can happen at runtime, and occurs in
Fedora 26 already (due to some change that causes a lot of channels to
be created at login), so ideally we'd prefer not to make it any slower.
We'd also like suspend/resume speed to not suffer.
Upcoming commits will solve those problems in a better way, making the
extra overhead of moving the locking here a non-issue.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
|
|
acquire()/boot() will need different logic in addition to performing
the actual mapping.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
|
|
The accessor functions can change as a result of acquire()/release() calls,
and are protected by any refcounting done there.
Other functions must remain constant, as they can be called any time.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
|
|
Match API with the BAR1 version.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
|
|
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
|
|
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
|
|
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
|
|
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
|
|
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
|
|
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
|
|
This is purely preparation for upcoming commits, there should be no
code changes here.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
|
|
This is purely preparation for upcoming commits, there should be no
code changes here.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
|
|
The namespace of NVKM is being changed to nvkm_ instead of nouveau_,
which will be used for the DRM part of the driver. This is being
done in order to make it very clear as to what part of the driver a
given symbol belongs to, and as a minor step towards splitting the
DRM driver out to be able to stand on its own (for virt).
Because there's already a large amount of churn here anyway, this is
as good a time as any to also switch to NVIDIA's device and chipset
naming to ease collaboration with them.
A comparison of objdump disassemblies proves no code changes.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
|
|
The symlinks were annoying some people, and they're not used anywhere
else in the kernel tree. The include directory structure has been
changed so that symlinks aren't needed anymore.
NVKM has been moved from core/ to nvkm/ to make it more obvious as to
what the directory is for, and as some minor prep for when NVKM gets
split out into its own module (virt) at a later date.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
|
