Commit graph

1,399,042 commits

Author SHA1 Message Date
David Belanger
c5faf18bbe drm/amdkfd: Enable atomic support for GFX12
Enable flag in KFD and set the atomic support bit in MQD.

Signed-off-by: David Belanger <david.belanger@amd.com>
Reviewed-by: Harish Kasiviswanathan <Harish.Kasiviswanathan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2024-05-02 16:18:13 -04:00
Eric Huang
a921c35ae5 drm/amdkfd: fix NULL ptr for debugfs mqds on GFX v12
mqd_stride function in gfx v12 is not implemented, that
causes NULL ptr error. Add the generic func to fix it.

Signed-off-by: Eric Huang <jinhuieric.huang@amd.com>
Reviewed-by: Harish Kasiviswanathan <Harish.Kasiviswanathan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2024-05-02 16:18:13 -04:00
Jonathan Kim
9243240bed drm/amdkfd: enable single alu ops for gfx12
GFX12 debugging requires setting up precise ALU operation for catching
ALU exceptions.

Signed-off-by: Jonathan Kim <jonathan.kim@amd.com>
Tested-by: Lancelot Six <lancelot.six@amd.com>
Reviewed-by: Eric Huang <jinhuieric.huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2024-05-02 16:18:13 -04:00
Jonathan Kim
984b265ff6 drm/amdkfd: fix support for trap on wave start and end for gfx12
Similar to GFX11, GFX12 supports trapping on wave start and end.

Signed-off-by: Jonathan Kim <jonathan.kim@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2024-05-02 16:18:13 -04:00
Jonathan Kim
fda3f378c4 drm/amdkfd: always enable ttmp setup for gfx12
Similar to GFX11, always enable the setup of trap temporaries on GFX12.

Signed-off-by: Jonathan Kim <jonathan.kim@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2024-05-02 16:18:13 -04:00
David Belanger
782b93436a drm/amdkfd: Enable GFX12 trap handler
Updated switch statement to use GFX12 trap handler.

Signed-off-by: David Belanger <david.belanger@amd.com>
Reviewed-by: Harish Kasiviswanathan <Harish.Kasiviswanathan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2024-05-02 16:18:13 -04:00
Laurent Morichetti
cf338b5dfe drm/amdkfd: enable missed single-step workaround for gfx12
When trap_ctrl.trap_after_inst is set, it is possible for a wave to
enter the trap handler, after single-stepping an instruction and a
save_context is raised, with only save_context set in excp_flag_priv.

Because excp_flag_priv.trap_after_inst is not reliably set, we need to
use the missed single-step workaround for gfx12 as well.

Also add wave_start and wave_end as exceptions that should be handled
by the 2nd level trap handler.

Signed-off-by: Laurent Morichetti <laurent.morichetti@amd.com>
Tested-by: Lancelot Six <lancelot.six@amd.com>
Reviewed-by: Jonathan Kim <jonathan.kim@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2024-05-02 16:18:13 -04:00
Lancelot SIX
450abfe433 drm/amdkfd: save and restore barrier state for gfx12
Add support to save and restore the work group barrier state in gfx12
CWSR trap handler.

There is no support to directly restore the signal count of a barrier
state, so instead this patch repeatedly calls s_barrier_signal to
increment the signal count to the desired value.

In this patch, I have implemented the logic to restore the barrier at
the end of the block restoring the HWREGs.  This process needs to be
done by exactly 1 wave per work group.  To achieve this, the initial
value of s_restore_spi_init_hi (containing a FIRST_WAVE bit) needs to be
saved up until that point.  An alternative could be restore the barrier
earlier in the process (around when LDS is restored, as the same wave
does both).  Doing this would break the pattern that the restore
procedure follows the CWSR area layout.

Before restoring the barrier, this patch checks if the barrier was whose
state was saved has the "valid" bit set, even if I don't think this
barrier can be in an invalid state during context save.  I expect this
test to always be true.

Signed-off-by: Lancelot SIX <lancelot.six@amd.com>
Reviewed-by: Jay Cornwall <jay.cornwall@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2024-05-02 16:18:12 -04:00
Jay Cornwall
f281003336 drm/amdkfd: Add gfx12 trap handler support
- HWREG changes since gfx11
- Save/restore barrier state
- get_wave_size is now reserved by assembler

v2: rebase (Alex)

Signed-off-by: Jay Cornwall <jay.cornwall@amd.com>
Reviewed-by: Harish Kasiviswanathan <Harish.Kasiviswanathan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2024-05-02 16:18:12 -04:00
Jay Cornwall
385093fde8 drm/amdkfd: Move trap handler coherence flags to preprocessor
No functional change. Preparation for gfx12 support.

v2: drop unrelated change (Alex)

Signed-off-by: Jay Cornwall <jay.cornwall@amd.com>
Reviewed-by: Harish Kasiviswanathan <Harish.Kasiviswanathan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2024-05-02 16:18:12 -04:00
David Belanger
90e4fc8369 drm/amdkfd: Added gfx_v12_kfd2kgd interface for GFX12.
Initial implementation, based on GFX11.

v2: Removed functions not needed by cp scheduler.
v3: Fixed typos.
v4: squash in warning fix (Alex)

Signed-off-by: David Belanger <david.belanger@amd.com>
Acked-by: Jonathan Kim <jonathan.kim@amd.com>
Reviewed-by: Harish Kasiviswanathan <Harish.Kasiviswanathan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2024-05-02 16:18:12 -04:00
David Belanger
47fa09b788 drm/amdkfd: Added device queue manager files for GFX12.
Initial implementation, based on GFX11.

v2: squash in include fix from David (Alex)

Signed-off-by: David Belanger <david.belanger@amd.com>
Reviewed-by: Harish Kasiviswanathan <Harish.Kasiviswanathan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2024-05-02 16:18:12 -04:00
David Belanger
48f0bdf4e3 drm/amdkfd: Added MQD manager files for GFX12.
Initial implementation, based on GFX11.

v2: Removed dbg_wa code as not needed on GFX12.
v3: squash in SDMA queue fixes (Alex)
v4: rebase (Alex)

Signed-off-by: David Belanger <david.belanger@amd.com>
Reviewed-by: Harish Kasiviswanathan <Harish.Kasiviswanathan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2024-05-02 16:18:12 -04:00
David Belanger
8aa89b69d6 drm/amdkfd: Added temporary changes for GFX12.
Added cases for GFX12 in switch statement, code relying on GFX11
implementation until GFX12 implementation is complete.

Signed-off-by: David Belanger <david.belanger@amd.com>
Reviewed-by: Harish Kasiviswanathan <Harish.Kasiviswanathan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2024-05-02 16:18:12 -04:00
David Belanger
592a5d7de4 drm/amdkfd: Basic SDMA and cache info changes for GFX12.
Added GFX12 support to a few switch statements.

Signed-off-by: David Belanger <david.belanger@amd.com>
Reviewed-by: Harish Kasiviswanathan <Harish.Kasiviswanathan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2024-05-02 16:18:12 -04:00
shaoyunl
2f983d3ca5 drm/amdgpu: Enable event log on MES 12
Enable event log through the HW specific FW API

Signed-off-by: shaoyunl <shaoyun.liu@amd.com>
Reviewed-by: Harish Kasiviswanthan <Harish.Kasiviswanthan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2024-05-02 16:18:12 -04:00
shaoyunl
19e69a5d28 drm/amdgpu: Enable unmapped doorbell handling basic mode on mes 12
Enable basic mode handling for doorbell ring on unmapped CP queue.
In this mode, MES can start schedule the queue mapping based on HW
interrupt instead of timer.

Signed-off-by: shaoyunl <shaoyun.liu@amd.com>
Reviewed-by: Harish Kasiviswanthan <Harish.Kasiviswanthan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2024-05-02 16:18:12 -04:00
Hawking Zhang
a2211e475c drm/amdgpu: Switch to smuio func to get gpu clk counter
Switch to smuio callback to query gpu clock counter

Signed-off-by: Hawking Zhang <Hawking.Zhang@amd.com>
Reviewed-by: Likun Gao <Likun.Gao@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2024-05-02 16:18:12 -04:00
Likun Gao
e781af6663 drm/amdgpu: init gfxhub setting to align with mmhub
Align gfxhub settings with mmhub when program rlc ram.

Signed-off-by: Likun Gao <Likun.Gao@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2024-05-02 16:18:12 -04:00
Likun Gao
b32edc2340 drm/amdgpu: skip dpm check to init imu fw
Skip dpm check to init imu firmware for imu v12.

Signed-off-by: Likun Gao <Likun.Gao@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2024-05-02 16:18:11 -04:00
Likun Gao
044feb8e2a drm/amdgpu: fix active rb and cu number for gfx12
Correct the algorithm of active CU and RB to bypass
the disabled SA for gfx12.

Signed-off-by: Likun Gao <Likun.Gao@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2024-05-02 16:18:11 -04:00
Likun Gao
56159fffaa drm/amdgpu: use new method to program rlc ram
Program rlc ram with golden setting data instead.
The old method (program_imu_rlc_ram_old) should be
retired in the future.

Signed-off-by: Likun Gao <Likun.Gao@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2024-05-02 16:18:11 -04:00
Kenneth Feng
043869be5a drm/amd/amdgpu: add cgcg&cgls interface for gfx 12.0
add cgcg&cgls interface for gfx 12.0

Signed-off-by: Kenneth Feng <kenneth.feng@amd.com>
Reviewed-by: Likun Gao <Likun.Gao@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2024-05-02 16:18:11 -04:00
Tom St Denis
60917ce8f8 drm/amd/amdgpu: update GFX12 wave data registers
Update the registers for gfx12.

Signed-off-by: Tom St Denis <tom.stdenis@amd.com>
Reviewed-by: Jonathan Kim <jonathan.kim@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2024-05-02 16:18:11 -04:00
Likun Gao
53efeba35d drm/amdgpu: set different fw data addr for mec pipe
For MEC fw data, different pipe should programed into
different address.

Signed-off-by: Likun Gao <Likun.Gao@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2024-05-02 16:18:11 -04:00
Kenneth Feng
1e740df77f drm/amd/amdgpu: workaround for the imu fw loading
workaournd for the imu fw loading on gfx 12.0 without psp

Signed-off-by: Kenneth Feng <kenneth.feng@amd.com>
Reviewed-by: Likun Gao <Likun.Gao@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2024-05-02 16:18:11 -04:00
Likun Gao
f5b4c3236f drm/amd: Move fw init from sw_init to early_init for imu v12
Move microcode loading from sw_init to early_init to align with
the perious version of imu init sequence.

Signed-off-by: Likun Gao <Likun.Gao@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2024-05-02 16:18:11 -04:00
Likun Gao
2502af906b drm/amdgpu: support S&R fw load for gfx v12
Support Save & Restore related fw load with backdoor RLC
autoload type on gfx v12.

Signed-off-by: Likun Gao <Likun.Gao@amd.com>
Reviewed-by: Kenneth Feng <kenneth.feng@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2024-05-02 16:18:11 -04:00
Jack Xiao
36b2ce4775 drm/amdgpu/gfx12: recalculate available compute rings to use
Recalculate the number of compute rings to use based on
the gfx hardware configuration. As needed reserve half of
compute rings for mes, kgd can't use up all compute rings.

Signed-off-by: Jack Xiao <Jack.Xiao@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2024-05-02 16:18:11 -04:00
Likun Gao
29d36a9cfd drm/amdgpu: skip imu related function if dpm=0
Only execute IMU related functions if dpm>0.

Signed-off-by: Likun Gao <Likun.Gao@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2024-05-02 16:18:11 -04:00
Kenneth Feng
32d1637689 drm/amd/amdgpu: imu fw loading support
support imu related function for gfx v12.

Signed-off-by: Kenneth Feng <kenneth.feng@amd.com>
Signed-off-by: Likun Gao <Likun.Gao@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2024-05-02 16:18:11 -04:00
Likun Gao
af204b76a7 drm/amdgpu: set cp fw address set for gfx v12
Split PFF/ME/MEC firmware address setting function
from related load microcode funtion, as it's also
needed for rlc autolad.

Signed-off-by: Likun Gao <Likun.Gao@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2024-05-02 16:18:10 -04:00
Likun Gao
52cb80c12e drm/amdgpu: Add gfx v12_0 ip block support (v6)
Initial support for GFX 12.

v1: Add gfx v12_0 ip block support. (Likun)
v2: Switch to gfx.kiq array.
    Move the vmhub from ring callback to ring. (Hawking)
v3: Update various callback function impl. (Hawking)
v4: Warning fixes (Alex)
v5: squash in imu fix, csb, rlc autoload implementations (Alex)
v6: Rebase (Alex)

Signed-off-by: Likun Gao <Likun.Gao@amd.com>
Signed-off-by: Hawking Zhang <Hawking.Zhang@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2024-05-02 16:18:10 -04:00
Jack Xiao
4632bec9fa drm/amdgpu/mes12: update data cache boundary
Enlarge the data cache boundary.

v2: use the fix data cache boundary.

Signed-off-by: Jack Xiao <Jack.Xiao@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2024-05-02 16:18:10 -04:00
Jonathan Kim
46c4766610 drm/amdgpu: fix trap enablement for gfx12
Fix request to MES to set SQ_SHADER_TBA_HI.trap_en for GFX12.

Signed-off-by: Jonathan Kim <jonathan.kim@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2024-05-02 16:18:10 -04:00
shaoyunl
d817c470cb drm/amdgpu: Enable MES to handle doorbell ring on unmapped queue
On MES12, HW can monitor up to 2048 doorbells that not be
mapped currently and trigger the interrupt to MES when these unmapped
doorbell been ringed.

Signed-off-by: shaoyunl <shaoyun.liu@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2024-05-02 16:18:10 -04:00
Jack Xiao
745f46b6a9 drm/amdgpu: enable mes v12 self test
1. fix available compute queue to use
2. enable mes v12 self test

Signed-off-by: Jack Xiao <Jack.Xiao@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2024-05-02 16:18:10 -04:00
Likun Gao
6628f7762b drm/amdgpu: set mes fw address for mes v12
Split the function of mes fimrware address setting
from mes firmware load for mes v12, as it's also
needed for rlc autoload.

Signed-off-by: Likun Gao <Likun.Gao@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2024-05-02 16:18:10 -04:00
Jack Xiao
785f0f9fe7 drm/amdgpu: Add mes v12_0 ip block support (v4)
v1: Add mes v12_0 ip block support. (Jack)
v2: Switch to gfx.kiq array. (Hawking)
v3: Switch to AMDGPU_GFXHUB(0). (Hawking)
v4: Rebase (Alex)

Signed-off-by: Jack Xiao <Jack.Xiao@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2024-05-02 16:18:10 -04:00
Likun Gao
69d4c44e51 drm/amdgpu: init mes ucode name for gfx v12
Keep gfx v12 mes fw name to gc_12_x_x_mes.bin
and gc_12_x_x_mes1.bin.

Signed-off-by: Likun Gao <Likun.Gao@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2024-05-02 16:18:10 -04:00
Harish Kasiviswanathan
415fcb8c50 drm/amdgpu: Add mes_v12_api_def.h for gfx12
Add MES_v12 header definition for gfx12

v2: Modify SET_SHADER_DEBUGGER to match mes_v11 definition. This doesn't
change the structure layout

v3: Removed unncessary comment and spaces

Signed-off-by: Harish Kasiviswanathan <Harish.Kasiviswanathan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2024-05-02 16:18:10 -04:00
Likun Gao
00c9035633 drm/amdgpu: add rlc TOC header file for soc24
Add RLC autoload TOC header file for soc24 ASIC.

Signed-off-by: Likun Gao <Likun.Gao@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2024-05-02 16:18:10 -04:00
Likun Gao
e3a911bb38 drm/amdgpu: add new TOC structure
Add new RLC_TABLE_OF_CONTENT structure definition.

Signed-off-by: Likun Gao <Likun.Gao@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2024-05-02 16:18:10 -04:00
Likun Gao
d8fd91f905 drm/amdgpu: add gfx12 clearstate header
Add gfx12 clearstate register arrays.

Signed-off-by: Likun Gao <Likun.Gao@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2024-05-02 16:18:10 -04:00
Likun Gao
5638b1cfa7 drm/amdgpu/discovery: Set GC family for GC 12.0 IP
Set GC family for GC 12.0 IPs.

v2: squash in updates (Alex)

Signed-off-by: Likun Gao <Likun.Gao@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2024-05-02 16:18:09 -04:00
Likun Gao
226e4ca5d4 drm/amdgpu: Add gfx v12_0_0 family id
Add gfx v12_0_0 family id

Signed-off-by: Likun Gao <Likun.Gao@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2024-05-02 16:18:09 -04:00
Sonny Jiang
e56b042118 drm/amdgpu: IB test encode test package change for VCN5
VCN5 session info package interface changed

Signed-off-by: Sonny Jiang <sonny.jiang@amd.com>
Reviewed-by: Leo Liu <leo.liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2024-05-02 16:17:55 -04:00
Josef Bacik
e03418abde btrfs: make sure that WRITTEN is set on all metadata blocks
We previously would call btrfs_check_leaf() if we had the check
integrity code enabled, which meant that we could only run the extended
leaf checks if we had WRITTEN set on the header flags.

This leaves a gap in our checking, because we could end up with
corruption on disk where WRITTEN isn't set on the leaf, and then the
extended leaf checks don't get run which we rely on to validate all of
the item pointers to make sure we don't access memory outside of the
extent buffer.

However, since 732fab95ab ("btrfs: check-integrity: remove
CONFIG_BTRFS_FS_CHECK_INTEGRITY option") we no longer call
btrfs_check_leaf() from btrfs_mark_buffer_dirty(), which means we only
ever call it on blocks that are being written out, and thus have WRITTEN
set, or that are being read in, which should have WRITTEN set.

Add checks to make sure we have WRITTEN set appropriately, and then make
sure __btrfs_check_leaf() always does the item checking.  This will
protect us from file systems that have been corrupted and no longer have
WRITTEN set on some of the blocks.

This was hit on a crafted image tweaking the WRITTEN bit and reported by
KASAN as out-of-bound access in the eb accessors. The example is a dir
item at the end of an eb.

  [2.042] BTRFS warning (device loop1): bad eb member start: ptr 0x3fff start 30572544 member offset 16410 size 2
  [2.040] general protection fault, probably for non-canonical address 0xe0009d1000000003: 0000 [#1] PREEMPT SMP KASAN NOPTI
  [2.537] KASAN: maybe wild-memory-access in range [0x0005088000000018-0x000508800000001f]
  [2.729] CPU: 0 PID: 2587 Comm: mount Not tainted 6.8.2 #1
  [2.729] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014
  [2.621] RIP: 0010:btrfs_get_16+0x34b/0x6d0
  [2.621] RSP: 0018:ffff88810871fab8 EFLAGS: 00000206
  [2.621] RAX: 0000a11000000003 RBX: ffff888104ff8720 RCX: ffff88811b2288c0
  [2.621] RDX: dffffc0000000000 RSI: ffffffff81dd8aca RDI: ffff88810871f748
  [2.621] RBP: 000000000000401a R08: 0000000000000001 R09: ffffed10210e3ee9
  [2.621] R10: ffff88810871f74f R11: 205d323430333737 R12: 000000000000001a
  [2.621] R13: 000508800000001a R14: 1ffff110210e3f5d R15: ffffffff850011e8
  [2.621] FS:  00007f56ea275840(0000) GS:ffff88811b200000(0000) knlGS:0000000000000000
  [2.621] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
  [2.621] CR2: 00007febd13b75c0 CR3: 000000010bb50000 CR4: 00000000000006f0
  [2.621] Call Trace:
  [2.621]  <TASK>
  [2.621]  ? show_regs+0x74/0x80
  [2.621]  ? die_addr+0x46/0xc0
  [2.621]  ? exc_general_protection+0x161/0x2a0
  [2.621]  ? asm_exc_general_protection+0x26/0x30
  [2.621]  ? btrfs_get_16+0x33a/0x6d0
  [2.621]  ? btrfs_get_16+0x34b/0x6d0
  [2.621]  ? btrfs_get_16+0x33a/0x6d0
  [2.621]  ? __pfx_btrfs_get_16+0x10/0x10
  [2.621]  ? __pfx_mutex_unlock+0x10/0x10
  [2.621]  btrfs_match_dir_item_name+0x101/0x1a0
  [2.621]  btrfs_lookup_dir_item+0x1f3/0x280
  [2.621]  ? __pfx_btrfs_lookup_dir_item+0x10/0x10
  [2.621]  btrfs_get_tree+0xd25/0x1910

Reported-by: lei lu <llfamsec@gmail.com>
CC: stable@vger.kernel.org # 6.7+
Reviewed-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
[ copy more details from report ]
Signed-off-by: David Sterba <dsterba@suse.com>
2024-05-02 22:11:13 +02:00
Alexander Gordeev
22a49f6d30 Merge branch 'shared-zeropage' into features
David Hildenbrand says:

===================
This series fixes one issue with uffd + shared zeropages on s390x and
fixes that "ordinary" KVM guests can make use of shared zeropages again.

userfaultfd could currently end up mapping shared zeropages into processes
that forbid shared zeropages. This only apples to s390x, relevant for
handling PV guests and guests that use storage kets correctly. Fix it
by placing a zeroed folio instead of the shared zeropage during
UFFDIO_ZEROPAGE instead.

I stumbled over this issue while looking into a customer scenario that
is using:

(1) Memory ballooning for dynamic resizing. Start a VM with, say, 100 GiB
    and inflate the balloon during boot to 60 GiB. The VM has ~40 GiB
    available and additional memory can be "fake hotplugged" to the VM
    later on demand by deflating the balloon. Actual memory overcommit is
    not desired, so physical memory would only be moved between VMs.

(2) Live migration of VMs between sites to evacuate servers in case of
    emergency.

Without the shared zeropage, during (2), the VM would suddenly consume
100 GiB on the migration source and destination. On the migration source,
where we don't excpect memory overcommit, we could easilt end up crashing
the VM during migration.

Independent of that, memory handed back to the hypervisor using "free page
reporting" would end up consuming actual memory after the migration on the
destination, not getting freed up until reused+freed again.

While there might be ways to optimize parts of this in QEMU, we really
should just support the shared zeropage again for ordinary VMs.

We only expect legcy guests to make use of storage keys, so let's handle
zeropages again when enabling storage keys or when enabling PV. To not
break userfaultfd like we did in the past, don't zap the shared zeropages,
but instead trigger unsharing faults, just like we do for unsharing
KSM pages in break_ksm().

Unsharing faults will simply replace the shared zeropage by a zeroed
anonymous folio. We can already trigger the same fault path using GUP,
when trying to long-term pin a shared zeropage, but also when unmerging
a KSM-placed zeropages, so this is nothing new.

Patch #1 tested on 86-64 by forcing mm_forbids_zeropage() to be 1, and
running the uffd selftests.

Patch #2 tested on s390x: the live migration scenario now works as
expected, and kvm-unit-tests that trigger usage of skeys work well, whereby
I can see detection and unsharing of shared zeropages.

Further (as broken in v2), I tested that the shared zeropage is no
longer populated after skeys are used -- that mm_forbids_zeropage() works
as expected:
  ./s390x-run s390x/skey.elf \
   -no-shutdown \
   -chardev socket,id=monitor,path=/var/tmp/mon,server,nowait \
   -mon chardev=monitor,mode=readline

  Then, in another shell:

  # cat /proc/`pgrep qemu`/smaps_rollup | grep Rss
  Rss:               31484 kB
  #  echo "dump-guest-memory tmp" | sudo nc -U /var/tmp/mon
  ...
  # cat /proc/`pgrep qemu`/smaps_rollup | grep Rss
  Rss:              160452 kB

  -> Reading guest memory does not populate the shared zeropage

  Doing the same with selftest.elf (no skeys)

  # cat /proc/`pgrep qemu`/smaps_rollup | grep Rss
  Rss:               30900 kB
  #  echo "dump-guest-memory tmp" | sudo nc -U /var/tmp/mon
  ...
  # cat /proc/`pgrep qemu`/smaps_rollup | grep Rsstmp/mon
  Rss:               30924 kB

  -> Reading guest memory does populate the shared zeropage
===================

Signed-off-by: Alexander Gordeev <agordeev@linux.ibm.com>
2024-05-02 22:02:25 +02:00
Hawking Zhang
5f571c61b9 drm/amdgpu: Add gfx v9_4_4 ip block
Add gfx v9_4_4 ip block support

Signed-off-by: Hawking Zhang <Hawking.Zhang@amd.com>
Reviewed-by: Le Ma <le.ma@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
2024-05-02 15:49:16 -04:00