Kernel/linux-next
1
0
Fork 0
mirror of https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git synced 2025-01-01 10:42:11 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

Merge drm/drm-next into drm-misc-next

Browse Source 
We need to backmerge drm-next to fix the komeda build failure.

Signed-off-by: Maxime Ripard <maxime.ripard@bootlin.com>
This commit is contained in:
Maxime Ripard 2019-02-11 10:35:35 +01:00
commit d588100baa
No known key found for this signature in database
GPG Key ID: E3EF0D6F671851C5
509 changed files with 23939 additions and 27803 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

73
Documentation/devicetree/bindings/display/arm,komeda.txt Normal file
View File

@ -0,0 +1,73 @@
Device Tree bindings for Arm Komeda display driver
Required properties:
- compatible: Should be "arm,mali-d71"
- reg: Physical base address and length of the registers in the system
- interrupts: the interrupt line number of the device in the system
- clocks: A list of phandle + clock-specifier pairs, one for each entry
in 'clock-names'
- clock-names: A list of clock names. It should contain:
- "mclk": for the main processor clock
- "pclk": for the APB interface clock
- #address-cells: Must be 1
- #size-cells: Must be 0
Required properties for sub-node: pipeline@nq
Each device contains one or two pipeline sub-nodes (at least one), each
pipeline node should provide properties:
- reg: Zero-indexed identifier for the pipeline
- clocks: A list of phandle + clock-specifier pairs, one for each entry
in 'clock-names'
- clock-names: should contain:
- "pxclk": pixel clock
- "aclk": AXI interface clock
- port: each pipeline connect to an encoder input port. The connection is
modeled using the OF graph bindings specified in
Documentation/devicetree/bindings/graph.txt
Optional properties:
- memory-region: phandle to a node describing memory (see
Documentation/devicetree/bindings/reserved-memory/reserved-memory.txt)
to be used for the framebuffer; if not present, the framebuffer may
be located anywhere in memory.
Example:
/ {
...
dp0: display@c00000 {
#address-cells = <1>;
#size-cells = <0>;
compatible = "arm,mali-d71";
reg = <0xc00000 0x20000>;
interrupts = <0 168 4>;
clocks = <&dpu_mclk>, <&dpu_aclk>;
clock-names = "mclk", "pclk";
dp0_pipe0: pipeline@0 {
clocks = <&fpgaosc2>, <&dpu_aclk>;
clock-names = "pxclk", "aclk";
reg = <0>;
port {
dp0_pipe0_out: endpoint {
remote-endpoint = <&db_dvi0_in>;
};
};
};
dp0_pipe1: pipeline@1 {
clocks = <&fpgaosc2>, <&dpu_aclk>;
clock-names = "pxclk", "aclk";
reg = <1>;
port {
dp0_pipe1_out: endpoint {
remote-endpoint = <&db_dvi1_in>;
};
};
};
};
...
};

1
Documentation/devicetree/bindings/display/bridge/renesas,lvds.txt
View File

@ -8,6 +8,7 @@ Required properties:
- compatible : Shall contain one of
- "renesas,r8a7743-lvds" for R8A7743 (RZ/G1M) compatible LVDS encoders
- "renesas,r8a7744-lvds" for R8A7744 (RZ/G1N) compatible LVDS encoders
- "renesas,r8a774c0-lvds" for R8A774C0 (RZ/G2E) compatible LVDS encoders
- "renesas,r8a7790-lvds" for R8A7790 (R-Car H2) compatible LVDS encoders
- "renesas,r8a7791-lvds" for R8A7791 (R-Car M2-W) compatible LVDS encoders

3
Documentation/devicetree/bindings/display/tegra/nvidia,tegra20-host1x.txt
View File

@ -238,6 +238,9 @@ of the following host1x client modules:
- nvidia,hpd-gpio: specifies a GPIO used for hotplug detection
- nvidia,edid: supplies a binary EDID blob
- nvidia,panel: phandle of a display panel
- nvidia,xbar-cfg: 5 cells containing the crossbar configuration. Each lane
of the SOR, identified by the cell's index, is mapped via the crossbar to
the pad specified by the cell's value.
Optional properties when driving an eDP output:
- nvidia,dpaux: phandle to a DispayPort AUX interface

7
Documentation/devicetree/bindings/gpu/samsung-rotator.txt
View File

@ -2,9 +2,10 @@
Required properties:
- compatible : value should be one of the following:
(a) "samsung,exynos4210-rotator" for Rotator IP in Exynos4210
(b) "samsung,exynos4212-rotator" for Rotator IP in Exynos4212/4412
(c) "samsung,exynos5250-rotator" for Rotator IP in Exynos5250
* "samsung,s5pv210-rotator" for Rotator IP in S5PV210
* "samsung,exynos4210-rotator" for Rotator IP in Exynos4210
* "samsung,exynos4212-rotator" for Rotator IP in Exynos4212/4412
* "samsung,exynos5250-rotator" for Rotator IP in Exynos5250
- reg : Physical base address of the IP registers and length of memory
mapped region.

235
Documentation/gpu/afbc.rst Normal file
View File

@ -0,0 +1,235 @@
.. SPDX-License-Identifier: GPL-2.0+
===================================
Arm Framebuffer Compression (AFBC)
===================================
AFBC is a proprietary lossless image compression protocol and format.
It provides fine-grained random access and minimizes the amount of
data transferred between IP blocks.
AFBC can be enabled on drivers which support it via use of the AFBC
format modifiers defined in drm_fourcc.h. See DRM_FORMAT_MOD_ARM_AFBC(*).
All users of the AFBC modifiers must follow the usage guidelines laid
out in this document, to ensure compatibility across different AFBC
producers and consumers.
Components and Ordering
=======================
AFBC streams can contain several components - where a component
corresponds to a color channel (i.e. R, G, B, X, A, Y, Cb, Cr).
The assignment of input/output color channels must be consistent
between the encoder and the decoder for correct operation, otherwise
the consumer will interpret the decoded data incorrectly.
Furthermore, when the lossless colorspace transform is used
(AFBC_FORMAT_MOD_YTR, which should be enabled for RGB buffers for
maximum compression efficiency), the component order must be:
* Component 0: R
* Component 1: G
* Component 2: B
The component ordering is communicated via the fourcc code in the
fourcc:modifier pair. In general, component '0' is considered to
reside in the least-significant bits of the corresponding linear
format. For example, COMP(bits):
* DRM_FORMAT_ABGR8888
* Component 0: R(8)
* Component 1: G(8)
* Component 2: B(8)
* Component 3: A(8)
* DRM_FORMAT_BGR888
* Component 0: R(8)
* Component 1: G(8)
* Component 2: B(8)
* DRM_FORMAT_YUYV
* Component 0: Y(8)
* Component 1: Cb(8, 2x1 subsampled)
* Component 2: Cr(8, 2x1 subsampled)
In AFBC, 'X' components are not treated any differently from any other
component. Therefore, an AFBC buffer with fourcc DRM_FORMAT_XBGR8888
encodes with 4 components, like so:
* DRM_FORMAT_XBGR8888
* Component 0: R(8)
* Component 1: G(8)
* Component 2: B(8)
* Component 3: X(8)
Please note, however, that the inclusion of a "wasted" 'X' channel is
bad for compression efficiency, and so it's recommended to avoid
formats containing 'X' bits. If a fourth component is
required/expected by the encoder/decoder, then it is recommended to
instead use an equivalent format with alpha, setting all alpha bits to
'1'. If there is no requirement for a fourth component, then a format
which doesn't include alpha can be used, e.g. DRM_FORMAT_BGR888.
Number of Planes
================
Formats which are typically multi-planar in linear layouts (e.g. YUV
420), can be encoded into one, or multiple, AFBC planes. As with
component order, the encoder and decoder must agree about the number
of planes in order to correctly decode the buffer. The fourcc code is
used to determine the number of encoded planes in an AFBC buffer,
matching the number of planes for the linear (unmodified) format.
Within each plane, the component ordering also follows the fourcc
code:
For example:
* DRM_FORMAT_YUYV: nplanes = 1
* Plane 0:
* Component 0: Y(8)
* Component 1: Cb(8, 2x1 subsampled)
* Component 2: Cr(8, 2x1 subsampled)
* DRM_FORMAT_NV12: nplanes = 2
* Plane 0:
* Component 0: Y(8)
* Plane 1:
* Component 0: Cb(8, 2x1 subsampled)
* Component 1: Cr(8, 2x1 subsampled)
Cross-device interoperability
=============================
For maximum compatibility across devices, the table below defines
canonical formats for use between AFBC-enabled devices. Formats which
are listed here must be used exactly as specified when using the AFBC
modifiers. Formats which are not listed should be avoided.
.. flat-table:: AFBC formats
* - Fourcc code
- Description
- Planes/Components
* - DRM_FORMAT_ABGR2101010
- 10-bit per component RGB, with 2-bit alpha
- Plane 0: 4 components
* Component 0: R(10)
* Component 1: G(10)
* Component 2: B(10)
* Component 3: A(2)
* - DRM_FORMAT_ABGR8888
- 8-bit per component RGB, with 8-bit alpha
- Plane 0: 4 components
* Component 0: R(8)
* Component 1: G(8)
* Component 2: B(8)
* Component 3: A(8)
* - DRM_FORMAT_BGR888
- 8-bit per component RGB
- Plane 0: 3 components
* Component 0: R(8)
* Component 1: G(8)
* Component 2: B(8)
* - DRM_FORMAT_BGR565
- 5/6-bit per component RGB
- Plane 0: 3 components
* Component 0: R(5)
* Component 1: G(6)
* Component 2: B(5)
* - DRM_FORMAT_ABGR1555
- 5-bit per component RGB, with 1-bit alpha
- Plane 0: 4 components
* Component 0: R(5)
* Component 1: G(5)
* Component 2: B(5)
* Component 3: A(1)
* - DRM_FORMAT_VUY888
- 8-bit per component YCbCr 444, single plane
- Plane 0: 3 components
* Component 0: Y(8)
* Component 1: Cb(8)
* Component 2: Cr(8)
* - DRM_FORMAT_VUY101010
- 10-bit per component YCbCr 444, single plane
- Plane 0: 3 components
* Component 0: Y(10)
* Component 1: Cb(10)
* Component 2: Cr(10)
* - DRM_FORMAT_YUYV
- 8-bit per component YCbCr 422, single plane
- Plane 0: 3 components
* Component 0: Y(8)
* Component 1: Cb(8, 2x1 subsampled)
* Component 2: Cr(8, 2x1 subsampled)
* - DRM_FORMAT_NV16
- 8-bit per component YCbCr 422, two plane
- Plane 0: 1 component
* Component 0: Y(8)
Plane 1: 2 components
* Component 0: Cb(8, 2x1 subsampled)
* Component 1: Cr(8, 2x1 subsampled)
* - DRM_FORMAT_Y210
- 10-bit per component YCbCr 422, single plane
- Plane 0: 3 components
* Component 0: Y(10)
* Component 1: Cb(10, 2x1 subsampled)
* Component 2: Cr(10, 2x1 subsampled)
* - DRM_FORMAT_P210
- 10-bit per component YCbCr 422, two plane
- Plane 0: 1 component
* Component 0: Y(10)
Plane 1: 2 components
* Component 0: Cb(10, 2x1 subsampled)
* Component 1: Cr(10, 2x1 subsampled)
* - DRM_FORMAT_YUV420_8BIT
- 8-bit per component YCbCr 420, single plane
- Plane 0: 3 components
* Component 0: Y(8)
* Component 1: Cb(8, 2x2 subsampled)
* Component 2: Cr(8, 2x2 subsampled)
* - DRM_FORMAT_YUV420_10BIT
- 10-bit per component YCbCr 420, single plane
- Plane 0: 3 components
* Component 0: Y(10)
* Component 1: Cb(10, 2x2 subsampled)
* Component 2: Cr(10, 2x2 subsampled)
* - DRM_FORMAT_NV12
- 8-bit per component YCbCr 420, two plane
- Plane 0: 1 component
* Component 0: Y(8)
Plane 1: 2 components
* Component 0: Cb(8, 2x2 subsampled)
* Component 1: Cr(8, 2x2 subsampled)
* - DRM_FORMAT_P010
- 10-bit per component YCbCr 420, two plane
- Plane 0: 1 component
* Component 0: Y(10)
Plane 1: 2 components
* Component 0: Cb(10, 2x2 subsampled)
* Component 1: Cr(10, 2x2 subsampled)

2
Documentation/gpu/drivers.rst
View File

@ -17,6 +17,8 @@ GPU Driver Documentation
vkms
bridge/dw-hdmi
xen-front
afbc
komeda-kms
.. only:: subproject and html

488
Documentation/gpu/komeda-kms.rst Normal file
View File

@ -0,0 +1,488 @@
.. SPDX-License-Identifier: GPL-2.0
==============================
drm/komeda Arm display driver
==============================
The drm/komeda driver supports the Arm display processor D71 and later products,
this document gives a brief overview of driver design: how it works and why
design it like that.
Overview of D71 like display IPs
================================
From D71, Arm display IP begins to adopt a flexible and modularized
architecture. A display pipeline is made up of multiple individual and
functional pipeline stages called components, and every component has some
specific capabilities that can give the flowed pipeline pixel data a
particular processing.
Typical D71 components:
Layer
-----
Layer is the first pipeline stage, which prepares the pixel data for the next
stage. It fetches the pixel from memory, decodes it if it's AFBC, rotates the
source image, unpacks or converts YUV pixels to the device internal RGB pixels,
then adjusts the color_space of pixels if needed.
Scaler
------
As its name suggests, scaler takes responsibility for scaling, and D71 also
supports image enhancements by scaler.
The usage of scaler is very flexible and can be connected to layer output
for layer scaling, or connected to compositor and scale the whole display
frame and then feed the output data into wb_layer which will then write it
into memory.
Compositor (compiz)
-------------------
Compositor blends multiple layers or pixel data flows into one single display
frame. its output frame can be fed into post image processor for showing it on
the monitor or fed into wb_layer and written to memory at the same time.
user can also insert a scaler between compositor and wb_layer to down scale
the display frame first and and then write to memory.
Writeback Layer (wb_layer)
--------------------------
Writeback layer does the opposite things of Layer, which connects to compiz
and writes the composition result to memory.
Post image processor (improc)
-----------------------------
Post image processor adjusts frame data like gamma and color space to fit the
requirements of the monitor.
Timing controller (timing_ctrlr)
--------------------------------
Final stage of display pipeline, Timing controller is not for the pixel
handling, but only for controlling the display timing.
Merger
------
D71 scaler mostly only has the half horizontal input/output capabilities
compared with Layer, like if Layer supports 4K input size, the scaler only can
support 2K input/output in the same time. To achieve the ful frame scaling, D71
introduces Layer Split, which splits the whole image to two half parts and feeds
them to two Layers A and B, and does the scaling independently. After scaling
the result need to be fed to merger to merge two part images together, and then
output merged result to compiz.
Splitter
--------
Similar to Layer Split, but Splitter is used for writeback, which splits the
compiz result to two parts and then feed them to two scalers.
Possible D71 Pipeline usage
===========================
Benefitting from the modularized architecture, D71 pipelines can be easily
adjusted to fit different usages. And D71 has two pipelines, which support two
types of working mode:
- Dual display mode
Two pipelines work independently and separately to drive two display outputs.
- Single display mode
Two pipelines work together to drive only one display output.
On this mode, pipeline_B doesn't work indenpendently, but outputs its
composition result into pipeline_A, and its pixel timing also derived from
pipeline_A.timing_ctrlr. The pipeline_B works just like a "slave" of
pipeline_A(master)
Single pipeline data flow
-------------------------
.. kernel-render:: DOT
:alt: Single pipeline digraph
:caption: Single pipeline data flow
digraph single_ppl {
rankdir=LR;
subgraph {
"Memory";
"Monitor";
}
subgraph cluster_pipeline {
style=dashed
node [shape=box]
{
node [bgcolor=grey style=dashed]
"Scaler-0";
"Scaler-1";
"Scaler-0/1"
}
node [bgcolor=grey style=filled]
"Layer-0" -> "Scaler-0"
"Layer-1" -> "Scaler-0"
"Layer-2" -> "Scaler-1"
"Layer-3" -> "Scaler-1"
"Layer-0" -> "Compiz"
"Layer-1" -> "Compiz"
"Layer-2" -> "Compiz"
"Layer-3" -> "Compiz"
"Scaler-0" -> "Compiz"
"Scaler-1" -> "Compiz"
"Compiz" -> "Scaler-0/1" -> "Wb_layer"
"Compiz" -> "Improc" -> "Timing Controller"
}
"Wb_layer" -> "Memory"
"Timing Controller" -> "Monitor"
}
Dual pipeline with Slave enabled
--------------------------------
.. kernel-render:: DOT
:alt: Slave pipeline digraph
:caption: Slave pipeline enabled data flow
digraph slave_ppl {
rankdir=LR;
subgraph {
"Memory";
"Monitor";
}
node [shape=box]
subgraph cluster_pipeline_slave {
style=dashed
label="Slave Pipeline_B"
node [shape=box]
{
node [bgcolor=grey style=dashed]
"Slave.Scaler-0";
"Slave.Scaler-1";
}
node [bgcolor=grey style=filled]
"Slave.Layer-0" -> "Slave.Scaler-0"
"Slave.Layer-1" -> "Slave.Scaler-0"
"Slave.Layer-2" -> "Slave.Scaler-1"
"Slave.Layer-3" -> "Slave.Scaler-1"
"Slave.Layer-0" -> "Slave.Compiz"
"Slave.Layer-1" -> "Slave.Compiz"
"Slave.Layer-2" -> "Slave.Compiz"
"Slave.Layer-3" -> "Slave.Compiz"
"Slave.Scaler-0" -> "Slave.Compiz"
"Slave.Scaler-1" -> "Slave.Compiz"
}
subgraph cluster_pipeline_master {
style=dashed
label="Master Pipeline_A"
node [shape=box]
{
node [bgcolor=grey style=dashed]
"Scaler-0";
"Scaler-1";
"Scaler-0/1"
}
node [bgcolor=grey style=filled]
"Layer-0" -> "Scaler-0"
"Layer-1" -> "Scaler-0"
"Layer-2" -> "Scaler-1"
"Layer-3" -> "Scaler-1"
"Slave.Compiz" -> "Compiz"
"Layer-0" -> "Compiz"
"Layer-1" -> "Compiz"
"Layer-2" -> "Compiz"
"Layer-3" -> "Compiz"
"Scaler-0" -> "Compiz"
"Scaler-1" -> "Compiz"
"Compiz" -> "Scaler-0/1" -> "Wb_layer"
"Compiz" -> "Improc" -> "Timing Controller"
}
"Wb_layer" -> "Memory"
"Timing Controller" -> "Monitor"
}
Sub-pipelines for input and output
----------------------------------
A complete display pipeline can be easily divided into three sub-pipelines
according to the in/out usage.
Layer(input) pipeline
~~~~~~~~~~~~~~~~~~~~~
.. kernel-render:: DOT
:alt: Layer data digraph
:caption: Layer (input) data flow
digraph layer_data_flow {
rankdir=LR;
node [shape=box]
{
node [bgcolor=grey style=dashed]
"Scaler-n";
}
"Layer-n" -> "Scaler-n" -> "Compiz"
}
.. kernel-render:: DOT
:alt: Layer Split digraph
:caption: Layer Split pipeline
digraph layer_data_flow {
rankdir=LR;
node [shape=box]
"Layer-0/1" -> "Scaler-0" -> "Merger"
"Layer-2/3" -> "Scaler-1" -> "Merger"
"Merger" -> "Compiz"
}
Writeback(output) pipeline
~~~~~~~~~~~~~~~~~~~~~~~~~~
.. kernel-render:: DOT
:alt: writeback digraph
:caption: Writeback(output) data flow
digraph writeback_data_flow {
rankdir=LR;
node [shape=box]
{
node [bgcolor=grey style=dashed]
"Scaler-n";
}
"Compiz" -> "Scaler-n" -> "Wb_layer"
}
.. kernel-render:: DOT
:alt: split writeback digraph
:caption: Writeback(output) Split data flow
digraph writeback_data_flow {
rankdir=LR;
node [shape=box]
"Compiz" -> "Splitter"
"Splitter" -> "Scaler-0" -> "Merger"
"Splitter" -> "Scaler-1" -> "Merger"
"Merger" -> "Wb_layer"
}
Display output pipeline
~~~~~~~~~~~~~~~~~~~~~~~
.. kernel-render:: DOT
:alt: display digraph
:caption: display output data flow
digraph single_ppl {
rankdir=LR;
node [shape=box]
"Compiz" -> "Improc" -> "Timing Controller"
}
In the following section we'll see these three sub-pipelines will be handled
by KMS-plane/wb_conn/crtc respectively.
Komeda Resource abstraction
===========================
struct komeda_pipeline/component
--------------------------------
To fully utilize and easily access/configure the HW, the driver side also uses
a similar architecture: Pipeline/Component to describe the HW features and
capabilities, and a specific component includes two parts:
- Data flow controlling.
- Specific component capabilities and features.
So the driver defines a common header struct komeda_component to describe the
data flow control and all specific components are a subclass of this base
structure.
.. kernel-doc:: drivers/gpu/drm/arm/display/komeda/komeda_pipeline.h
:internal:
Resource discovery and initialization
=====================================
Pipeline and component are used to describe how to handle the pixel data. We
still need a @struct komeda_dev to describe the whole view of the device, and
the control-abilites of device.
We have &komeda_dev, &komeda_pipeline, &komeda_component. Now fill devices with
pipelines. Since komeda is not for D71 only but also intended for later products,
of course we’d better share as much as possible between different products. To
achieve this, split the komeda device into two layers: CORE and CHIP.
- CORE: for common features and capabilities handling.
- CHIP: for register programing and HW specific feature (limitation) handling.
CORE can access CHIP by three chip function structures:
- struct komeda_dev_funcs
- struct komeda_pipeline_funcs
- struct komeda_component_funcs
.. kernel-doc:: drivers/gpu/drm/arm/display/komeda/komeda_dev.h
:internal:
Format handling
===============
.. kernel-doc:: drivers/gpu/drm/arm/display/komeda/komeda_format_caps.h
:internal:
.. kernel-doc:: drivers/gpu/drm/arm/display/komeda/komeda_framebuffer.h
:internal:
Attach komeda_dev to DRM-KMS
============================
Komeda abstracts resources by pipeline/component, but DRM-KMS uses
crtc/plane/connector. One KMS-obj cannot represent only one single component,
since the requirements of a single KMS object cannot simply be achieved by a
single component, usually that needs multiple components to fit the requirement.
Like set mode, gamma, ctm for KMS all target on CRTC-obj, but komeda needs
compiz, improc and timing_ctrlr to work together to fit these requirements.
And a KMS-Plane may require multiple komeda resources: layer/scaler/compiz.
So, one KMS-Obj represents a sub-pipeline of komeda resources.
- Plane: `Layer(input) pipeline`_
- Wb_connector: `Writeback(output) pipeline`_
- Crtc: `Display output pipeline`_
So, for komeda, we treat KMS crtc/plane/connector as users of pipeline and
component, and at any one time a pipeline/component only can be used by one
user. And pipeline/component will be treated as private object of DRM-KMS; the
state will be managed by drm_atomic_state as well.
How to map plane to Layer(input) pipeline
-----------------------------------------
Komeda has multiple Layer input pipelines, see:
- `Single pipeline data flow`_
- `Dual pipeline with Slave enabled`_
The easiest way is binding a plane to a fixed Layer pipeline, but consider the
komeda capabilities:
- Layer Split, See `Layer(input) pipeline`_
Layer_Split is quite complicated feature, which splits a big image into two
parts and handles it by two layers and two scalers individually. But it
imports an edge problem or effect in the middle of the image after the split.
To avoid such a problem, it needs a complicated Split calculation and some
special configurations to the layer and scaler. We'd better hide such HW
related complexity to user mode.
- Slave pipeline, See `Dual pipeline with Slave enabled`_
Since the compiz component doesn't output alpha value, the slave pipeline
only can be used for bottom layers composition. The komeda driver wants to
hide this limitation to the user. The way to do this is to pick a suitable
Layer according to plane_state->zpos.
So for komeda, the KMS-plane doesn't represent a fixed komeda layer pipeline,
but multiple Layers with same capabilities. Komeda will select one or more
Layers to fit the requirement of one KMS-plane.
Make component/pipeline to be drm_private_obj
---------------------------------------------
Add :c:type:`drm_private_obj` to :c:type:`komeda_component`, :c:type:`komeda_pipeline`
.. code-block:: c
struct komeda_component {
struct drm_private_obj obj;
...
}
struct komeda_pipeline {
struct drm_private_obj obj;
...
}
Tracking component_state/pipeline_state by drm_atomic_state
-----------------------------------------------------------
Add :c:type:`drm_private_state` and user to :c:type:`komeda_component_state`,
:c:type:`komeda_pipeline_state`
.. code-block:: c
struct komeda_component_state {
struct drm_private_state obj;
void *binding_user;
...
}
struct komeda_pipeline_state {
struct drm_private_state obj;
struct drm_crtc *crtc;
...
}
komeda component validation
---------------------------
Komeda has multiple types of components, but the process of validation are
similar, usually including the following steps:
.. code-block:: c
int komeda_xxxx_validate(struct komeda_component_xxx xxx_comp,
struct komeda_component_output *input_dflow,
struct drm_plane/crtc/connector *user,
struct drm_plane/crtc/connector_state, *user_state)
{
setup 1: check if component is needed, like the scaler is optional depending
on the user_state; if unneeded, just return, and the caller will
put the data flow into next stage.
Setup 2: check user_state with component features and capabilities to see
if requirements can be met; if not, return fail.
Setup 3: get component_state from drm_atomic_state, and try set to set
user to component; fail if component has been assigned to another
user already.
Setup 3: configure the component_state, like set its input component,
convert user_state to component specific state.
Setup 4: adjust the input_dflow and prepare it for the next stage.
}
komeda_kms Abstraction
----------------------
.. kernel-doc:: drivers/gpu/drm/arm/display/komeda/komeda_kms.h
:internal:
komde_kms Functions
-------------------
.. kernel-doc:: drivers/gpu/drm/arm/display/komeda/komeda_crtc.c
:internal:
.. kernel-doc:: drivers/gpu/drm/arm/display/komeda/komeda_plane.c
:internal:
Build komeda to be a Linux module driver
========================================
Now we have two level devices:
- komeda_dev: describes the real display hardware.
- komeda_kms_dev: attachs or connects komeda_dev to DRM-KMS.
All komeda operations are supplied or operated by komeda_dev or komeda_kms_dev,
the module driver is only a simple wrapper to pass the Linux command
(probe/remove/pm) into komeda_dev or komeda_kms_dev.

15
MAINTAINERS
View File

@ -1133,13 +1133,26 @@ S: Supported
F: drivers/gpu/drm/arm/hdlcd_*
F: Documentation/devicetree/bindings/display/arm,hdlcd.txt
ARM KOMEDA DRM-KMS DRIVER
M: James (Qian) Wang <james.qian.wang@arm.com>
M: Liviu Dudau <liviu.dudau@arm.com>
L: Mali DP Maintainers <malidp@foss.arm.com>
S: Supported
T: git git://linux-arm.org/linux-ld.git for-upstream/mali-dp
F: drivers/gpu/drm/arm/display/include/
F: drivers/gpu/drm/arm/display/komeda/
F: Documentation/devicetree/bindings/display/arm/arm,komeda.txt
F: Documentation/gpu/komeda-kms.rst
ARM MALI-DP DRM DRIVER
M: Liviu Dudau <liviu.dudau@arm.com>
M: Brian Starkey <brian.starkey@arm.com>
M: Mali DP Maintainers <malidp@foss.arm.com>
L: Mali DP Maintainers <malidp@foss.arm.com>
S: Supported
T: git git://linux-arm.org/linux-ld.git for-upstream/mali-dp
F: drivers/gpu/drm/arm/
F: Documentation/devicetree/bindings/display/arm,malidp.txt
F: Documentation/gpu/afbc.rst
ARM MFM AND FLOPPY DRIVERS
M: Ian Molton <spyro@f2s.com>

4
drivers/gpu/drm/Makefile
View File

@ -51,7 +51,7 @@ obj-$(CONFIG_DRM_DEBUG_SELFTEST) += selftests/
obj-$(CONFIG_DRM) += drm.o
obj-$(CONFIG_DRM_MIPI_DSI) += drm_mipi_dsi.o
obj-$(CONFIG_DRM_PANEL_ORIENTATION_QUIRKS) += drm_panel_orientation_quirks.o
obj-$(CONFIG_DRM_ARM) += arm/
obj-y += arm/
obj-$(CONFIG_DRM_TTM) += ttm/
obj-$(CONFIG_DRM_SCHED) += scheduler/
obj-$(CONFIG_DRM_TDFX) += tdfx/
@ -81,7 +81,7 @@ obj-$(CONFIG_DRM_UDL) += udl/
obj-$(CONFIG_DRM_AST) += ast/
obj-$(CONFIG_DRM_ARMADA) += armada/
obj-$(CONFIG_DRM_ATMEL_HLCDC) += atmel-hlcdc/
obj-$(CONFIG_DRM_RCAR_DU) += rcar-du/
obj-y += rcar-du/
obj-$(CONFIG_DRM_SHMOBILE) +=shmobile/
obj-y += omapdrm/
obj-$(CONFIG_DRM_SUN4I) += sun4i/

2
drivers/gpu/drm/amd/amdgpu/Makefile
View File

@ -57,7 +57,7 @@ amdgpu-y += amdgpu_device.o amdgpu_kms.o \
# add asic specific block
amdgpu-$(CONFIG_DRM_AMDGPU_CIK)+= cik.o cik_ih.o kv_smc.o kv_dpm.o \
ci_smc.o ci_dpm.o dce_v8_0.o gfx_v7_0.o cik_sdma.o uvd_v4_2.o vce_v2_0.o
dce_v8_0.o gfx_v7_0.o cik_sdma.o uvd_v4_2.o vce_v2_0.o
amdgpu-$(CONFIG_DRM_AMDGPU_SI)+= si.o gmc_v6_0.o gfx_v6_0.o si_ih.o si_dma.o dce_v6_0.o si_dpm.o si_smc.o

9
drivers/gpu/drm/amd/amdgpu/amdgpu.h
View File

@ -542,6 +542,11 @@ struct amdgpu_asic_funcs {
bool (*need_full_reset)(struct amdgpu_device *adev);
/* initialize doorbell layout for specific asic*/
void (*init_doorbell_index)(struct amdgpu_device *adev);
/* PCIe bandwidth usage */
void (*get_pcie_usage)(struct amdgpu_device *adev, uint64_t *count0,
uint64_t *count1);
/* do we need to reset the asic at init time (e.g., kexec) */
bool (*need_reset_on_init)(struct amdgpu_device *adev);
};
/*
@ -634,7 +639,7 @@ struct amdgpu_nbio_funcs {
void (*hdp_flush)(struct amdgpu_device *adev, struct amdgpu_ring *ring);
u32 (*get_memsize)(struct amdgpu_device *adev);
void (*sdma_doorbell_range)(struct amdgpu_device *adev, int instance,
bool use_doorbell, int doorbell_index);
bool use_doorbell, int doorbell_index, int doorbell_size);
void (*enable_doorbell_aperture)(struct amdgpu_device *adev,
bool enable);
void (*enable_doorbell_selfring_aperture)(struct amdgpu_device *adev,
@ -1042,6 +1047,8 @@ int emu_soc_asic_init(struct amdgpu_device *adev);
#define amdgpu_asic_invalidate_hdp(adev, r) (adev)->asic_funcs->invalidate_hdp((adev), (r))
#define amdgpu_asic_need_full_reset(adev) (adev)->asic_funcs->need_full_reset((adev))
#define amdgpu_asic_init_doorbell_index(adev) (adev)->asic_funcs->init_doorbell_index((adev))
#define amdgpu_asic_get_pcie_usage(adev, cnt0, cnt1) ((adev)->asic_funcs->get_pcie_usage((adev), (cnt0), (cnt1)))
#define amdgpu_asic_need_reset_on_init(adev) (adev)->asic_funcs->need_reset_on_init((adev))
/* Common functions */
bool amdgpu_device_should_recover_gpu(struct amdgpu_device *adev);

84
drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
View File

@ -28,8 +28,6 @@
#include <linux/module.h>
#include <linux/dma-buf.h>
const struct kgd2kfd_calls *kgd2kfd;
static const unsigned int compute_vmid_bitmap = 0xFF00;
/* Total memory size in system memory and all GPU VRAM. Used to
@ -47,12 +45,9 @@ int amdgpu_amdkfd_init(void)
amdgpu_amdkfd_total_mem_size *= si.mem_unit;
#ifdef CONFIG_HSA_AMD
ret = kgd2kfd_init(KFD_INTERFACE_VERSION, &kgd2kfd);
if (ret)
kgd2kfd = NULL;
ret = kgd2kfd_init();
amdgpu_amdkfd_gpuvm_init_mem_limits();
#else
kgd2kfd = NULL;
ret = -ENOENT;
#endif
@ -61,17 +56,13 @@ int amdgpu_amdkfd_init(void)
void amdgpu_amdkfd_fini(void)
{
if (kgd2kfd)
kgd2kfd->exit();
kgd2kfd_exit();
}
void amdgpu_amdkfd_device_probe(struct amdgpu_device *adev)
{
const struct kfd2kgd_calls *kfd2kgd;
if (!kgd2kfd)
return;
switch (adev->asic_type) {
#ifdef CONFIG_DRM_AMDGPU_CIK
case CHIP_KAVERI:
@ -98,8 +89,8 @@ void amdgpu_amdkfd_device_probe(struct amdgpu_device *adev)
return;
}
adev->kfd.dev = kgd2kfd->probe((struct kgd_dev *)adev,
adev->pdev, kfd2kgd);
adev->kfd.dev = kgd2kfd_probe((struct kgd_dev *)adev,
adev->pdev, kfd2kgd);
if (adev->kfd.dev)
amdgpu_amdkfd_total_mem_size += adev->gmc.real_vram_size;
@ -182,7 +173,7 @@ void amdgpu_amdkfd_device_init(struct amdgpu_device *adev)
&gpu_resources.doorbell_start_offset);
if (adev->asic_type < CHIP_VEGA10) {
kgd2kfd->device_init(adev->kfd.dev, &gpu_resources);
kgd2kfd_device_init(adev->kfd.dev, &gpu_resources);
return;
}
@ -197,13 +188,13 @@ void amdgpu_amdkfd_device_init(struct amdgpu_device *adev)
* can use each doorbell assignment twice.
*/
gpu_resources.sdma_doorbell[0][i] =
adev->doorbell_index.sdma_engine0 + (i >> 1);
adev->doorbell_index.sdma_engine[0] + (i >> 1);
gpu_resources.sdma_doorbell[0][i+1] =
adev->doorbell_index.sdma_engine0 + 0x200 + (i >> 1);
adev->doorbell_index.sdma_engine[0] + 0x200 + (i >> 1);
gpu_resources.sdma_doorbell[1][i] =
adev->doorbell_index.sdma_engine1 + (i >> 1);
adev->doorbell_index.sdma_engine[1] + (i >> 1);
gpu_resources.sdma_doorbell[1][i+1] =
adev->doorbell_index.sdma_engine1 + 0x200 + (i >> 1);
adev->doorbell_index.sdma_engine[1] + 0x200 + (i >> 1);
}
/* Doorbells 0x0e0-0ff and 0x2e0-2ff are reserved for
* SDMA, IH and VCN. So don't use them for the CP.
@ -211,14 +202,14 @@ void amdgpu_amdkfd_device_init(struct amdgpu_device *adev)
gpu_resources.reserved_doorbell_mask = 0x1e0;
gpu_resources.reserved_doorbell_val = 0x0e0;
kgd2kfd->device_init(adev->kfd.dev, &gpu_resources);
kgd2kfd_device_init(adev->kfd.dev, &gpu_resources);
}
}
void amdgpu_amdkfd_device_fini(struct amdgpu_device *adev)
{
if (adev->kfd.dev) {
kgd2kfd->device_exit(adev->kfd.dev);
kgd2kfd_device_exit(adev->kfd.dev);
adev->kfd.dev = NULL;
}
}
@ -227,13 +218,13 @@ void amdgpu_amdkfd_interrupt(struct amdgpu_device *adev,
const void *ih_ring_entry)
{
if (adev->kfd.dev)
kgd2kfd->interrupt(adev->kfd.dev, ih_ring_entry);
kgd2kfd_interrupt(adev->kfd.dev, ih_ring_entry);
}
void amdgpu_amdkfd_suspend(struct amdgpu_device *adev)
{
if (adev->kfd.dev)
kgd2kfd->suspend(adev->kfd.dev);
kgd2kfd_suspend(adev->kfd.dev);
}
int amdgpu_amdkfd_resume(struct amdgpu_device *adev)
@ -241,7 +232,7 @@ int amdgpu_amdkfd_resume(struct amdgpu_device *adev)
int r = 0;
if (adev->kfd.dev)
r = kgd2kfd->resume(adev->kfd.dev);
r = kgd2kfd_resume(adev->kfd.dev);
return r;
}
@ -251,7 +242,7 @@ int amdgpu_amdkfd_pre_reset(struct amdgpu_device *adev)
int r = 0;
if (adev->kfd.dev)
r = kgd2kfd->pre_reset(adev->kfd.dev);
r = kgd2kfd_pre_reset(adev->kfd.dev);
return r;
}
@ -261,7 +252,7 @@ int amdgpu_amdkfd_post_reset(struct amdgpu_device *adev)
int r = 0;
if (adev->kfd.dev)
r = kgd2kfd->post_reset(adev->kfd.dev);
r = kgd2kfd_post_reset(adev->kfd.dev);
return r;
}
@ -619,4 +610,47 @@ struct kfd2kgd_calls *amdgpu_amdkfd_gfx_9_0_get_functions(void)
{
return NULL;
}
struct kfd_dev *kgd2kfd_probe(struct kgd_dev *kgd, struct pci_dev *pdev,
const struct kfd2kgd_calls *f2g)
{
return NULL;
}
bool kgd2kfd_device_init(struct kfd_dev *kfd,
const struct kgd2kfd_shared_resources *gpu_resources)
{
return false;
}
void kgd2kfd_device_exit(struct kfd_dev *kfd)
{
}
void kgd2kfd_exit(void)
{
}
void kgd2kfd_suspend(struct kfd_dev *kfd)
{
}
int kgd2kfd_resume(struct kfd_dev *kfd)
{
return 0;
}
int kgd2kfd_pre_reset(struct kfd_dev *kfd)
{
return 0;
}
int kgd2kfd_post_reset(struct kfd_dev *kfd)
{
return 0;
}
void kgd2kfd_interrupt(struct kfd_dev *kfd, const void *ih_ring_entry)
{
}
#endif

19
drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h
View File

@ -33,7 +33,6 @@
#include "amdgpu_sync.h"
#include "amdgpu_vm.h"
extern const struct kgd2kfd_calls *kgd2kfd;
extern uint64_t amdgpu_amdkfd_total_mem_size;
struct amdgpu_device;
@ -214,4 +213,22 @@ int amdgpu_amdkfd_gpuvm_import_dmabuf(struct kgd_dev *kgd,
void amdgpu_amdkfd_gpuvm_init_mem_limits(void);
void amdgpu_amdkfd_unreserve_memory_limit(struct amdgpu_bo *bo);
/* KGD2KFD callbacks */
int kgd2kfd_init(void);
void kgd2kfd_exit(void);
struct kfd_dev *kgd2kfd_probe(struct kgd_dev *kgd, struct pci_dev *pdev,
const struct kfd2kgd_calls *f2g);
bool kgd2kfd_device_init(struct kfd_dev *kfd,
const struct kgd2kfd_shared_resources *gpu_resources);
void kgd2kfd_device_exit(struct kfd_dev *kfd);
void kgd2kfd_suspend(struct kfd_dev *kfd);
int kgd2kfd_resume(struct kfd_dev *kfd);
int kgd2kfd_pre_reset(struct kfd_dev *kfd);
int kgd2kfd_post_reset(struct kfd_dev *kfd);
void kgd2kfd_interrupt(struct kfd_dev *kfd, const void *ih_ring_entry);
int kgd2kfd_quiesce_mm(struct mm_struct *mm);
int kgd2kfd_resume_mm(struct mm_struct *mm);
int kgd2kfd_schedule_evict_and_restore_process(struct mm_struct *mm,
struct dma_fence *fence);
#endif /* AMDGPU_AMDKFD_H_INCLUDED */

2
drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_fence.c
View File

@ -122,7 +122,7 @@ static bool amdkfd_fence_enable_signaling(struct dma_fence *f)
if (dma_fence_is_signaled(f))
return true;
if (!kgd2kfd->schedule_evict_and_restore_process(fence->mm, f))
if (!kgd2kfd_schedule_evict_and_restore_process(fence->mm, f))
return true;
return false;

4
drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
View File

@ -1790,7 +1790,7 @@ int amdgpu_amdkfd_evict_userptr(struct kgd_mem *mem,
evicted_bos = atomic_inc_return(&process_info->evicted_bos);
if (evicted_bos == 1) {
/* First eviction, stop the queues */
r = kgd2kfd->quiesce_mm(mm);
r = kgd2kfd_quiesce_mm(mm);
if (r)
pr_err("Failed to quiesce KFD\n");
schedule_delayed_work(&process_info->restore_userptr_work,
@ -2082,7 +2082,7 @@ static void amdgpu_amdkfd_restore_userptr_worker(struct work_struct *work)
evicted_bos)
goto unlock_out;
evicted_bos = 0;
if (kgd2kfd->resume_mm(mm)) {
if (kgd2kfd_resume_mm(mm)) {
pr_err("%s: Failed to resume KFD\n", __func__);
/* No recovery from this failure. Probably the CP is
* hanging. No point trying again.

1
drivers/gpu/drm/amd/amdgpu/amdgpu_atpx_handler.c
View File

@ -576,6 +576,7 @@ static const struct amdgpu_px_quirk amdgpu_px_quirk_list[] = {
{ 0x1002, 0x6900, 0x1028, 0x0812, AMDGPU_PX_QUIRK_FORCE_ATPX },
{ 0x1002, 0x6900, 0x1028, 0x0813, AMDGPU_PX_QUIRK_FORCE_ATPX },
{ 0x1002, 0x6900, 0x1025, 0x125A, AMDGPU_PX_QUIRK_FORCE_ATPX },
{ 0x1002, 0x6900, 0x17AA, 0x3806, AMDGPU_PX_QUIRK_FORCE_ATPX },
{ 0, 0, 0, 0, 0 },
};

13
drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
View File

@ -214,6 +214,7 @@ static int amdgpu_cs_parser_init(struct amdgpu_cs_parser *p, union drm_amdgpu_cs
case AMDGPU_CHUNK_ID_DEPENDENCIES:
case AMDGPU_CHUNK_ID_SYNCOBJ_IN:
case AMDGPU_CHUNK_ID_SYNCOBJ_OUT:
case AMDGPU_CHUNK_ID_SCHEDULED_DEPENDENCIES:
break;
default:
@ -1090,6 +1091,15 @@ static int amdgpu_cs_process_fence_dep(struct amdgpu_cs_parser *p,
fence = amdgpu_ctx_get_fence(ctx, entity,
deps[i].handle);
if (chunk->chunk_id == AMDGPU_CHUNK_ID_SCHEDULED_DEPENDENCIES) {
struct drm_sched_fence *s_fence = to_drm_sched_fence(fence);
struct dma_fence *old = fence;
fence = dma_fence_get(&s_fence->scheduled);
dma_fence_put(old);
}
if (IS_ERR(fence)) {
r = PTR_ERR(fence);
amdgpu_ctx_put(ctx);
@ -1177,7 +1187,8 @@ static int amdgpu_cs_dependencies(struct amdgpu_device *adev,
chunk = &p->chunks[i];
if (chunk->chunk_id == AMDGPU_CHUNK_ID_DEPENDENCIES) {
if (chunk->chunk_id == AMDGPU_CHUNK_ID_DEPENDENCIES ||
chunk->chunk_id == AMDGPU_CHUNK_ID_SCHEDULED_DEPENDENCIES) {
r = amdgpu_cs_process_fence_dep(p, chunk);
if (r)
return r;

135
drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
View File

@ -1645,7 +1645,7 @@ static int amdgpu_device_ip_init(struct amdgpu_device *adev)
if (r) {
DRM_ERROR("sw_init of IP block <%s> failed %d\n",
adev->ip_blocks[i].version->funcs->name, r);
return r;
goto init_failed;
}
adev->ip_blocks[i].status.sw = true;
@ -1654,17 +1654,17 @@ static int amdgpu_device_ip_init(struct amdgpu_device *adev)
r = amdgpu_device_vram_scratch_init(adev);
if (r) {
DRM_ERROR("amdgpu_vram_scratch_init failed %d\n", r);
return r;
goto init_failed;
}
r = adev->ip_blocks[i].version->funcs->hw_init((void *)adev);
if (r) {
DRM_ERROR("hw_init %d failed %d\n", i, r);
return r;
goto init_failed;
}
r = amdgpu_device_wb_init(adev);
if (r) {
DRM_ERROR("amdgpu_device_wb_init failed %d\n", r);
return r;
goto init_failed;
}
adev->ip_blocks[i].status.hw = true;
@ -1675,7 +1675,7 @@ static int amdgpu_device_ip_init(struct amdgpu_device *adev)
AMDGPU_CSA_SIZE);
if (r) {
DRM_ERROR("allocate CSA failed %d\n", r);
return r;
goto init_failed;
}
}
}
@ -1683,28 +1683,32 @@ static int amdgpu_device_ip_init(struct amdgpu_device *adev)
r = amdgpu_ucode_create_bo(adev); /* create ucode bo when sw_init complete*/
if (r)
return r;
goto init_failed;
r = amdgpu_device_ip_hw_init_phase1(adev);
if (r)
return r;
goto init_failed;
r = amdgpu_device_fw_loading(adev);
if (r)
return r;
goto init_failed;
r = amdgpu_device_ip_hw_init_phase2(adev);
if (r)
return r;
goto init_failed;
if (adev->gmc.xgmi.num_physical_nodes > 1)
amdgpu_xgmi_add_device(adev);
amdgpu_amdkfd_device_init(adev);
if (amdgpu_sriov_vf(adev))
init_failed:
if (amdgpu_sriov_vf(adev)) {
if (!r)
amdgpu_virt_init_data_exchange(adev);
amdgpu_virt_release_full_gpu(adev, true);
}
return 0;
return r;
}
/**
@ -2131,7 +2135,7 @@ static int amdgpu_device_ip_reinit_early_sriov(struct amdgpu_device *adev)
continue;
r = block->version->funcs->hw_init(adev);
DRM_INFO("RE-INIT: %s %s\n", block->version->funcs->name, r?"failed":"succeeded");
DRM_INFO("RE-INIT-early: %s %s\n", block->version->funcs->name, r?"failed":"succeeded");
if (r)
return r;
}
@ -2165,7 +2169,7 @@ static int amdgpu_device_ip_reinit_late_sriov(struct amdgpu_device *adev)
continue;
r = block->version->funcs->hw_init(adev);
DRM_INFO("RE-INIT: %s %s\n", block->version->funcs->name, r?"failed":"succeeded");
DRM_INFO("RE-INIT-late: %s %s\n", block->version->funcs->name, r?"failed":"succeeded");
if (r)
return r;
}
@ -2546,6 +2550,17 @@ int amdgpu_device_init(struct amdgpu_device *adev,
/* detect if we are with an SRIOV vbios */
amdgpu_device_detect_sriov_bios(adev);
/* check if we need to reset the asic
* E.g., driver was not cleanly unloaded previously, etc.
*/
if (!amdgpu_sriov_vf(adev) && amdgpu_asic_need_reset_on_init(adev)) {
r = amdgpu_asic_reset(adev);
if (r) {
dev_err(adev->dev, "asic reset on init failed\n");
goto failed;
}
}
/* Post card if necessary */
if (amdgpu_device_need_post(adev)) {
if (!adev->bios) {
@ -2610,6 +2625,8 @@ int amdgpu_device_init(struct amdgpu_device *adev,
}
dev_err(adev->dev, "amdgpu_device_ip_init failed\n");
amdgpu_vf_error_put(adev, AMDGIM_ERROR_VF_AMDGPU_INIT_FAIL, 0, 0);
if (amdgpu_virt_request_full_gpu(adev, false))
amdgpu_virt_release_full_gpu(adev, false);
goto failed;
}
@ -2632,9 +2649,6 @@ int amdgpu_device_init(struct amdgpu_device *adev,
goto failed;
}
if (amdgpu_sriov_vf(adev))
amdgpu_virt_init_data_exchange(adev);
amdgpu_fbdev_init(adev);
r = amdgpu_pm_sysfs_init(adev);
@ -2798,7 +2812,7 @@ int amdgpu_device_suspend(struct drm_device *dev, bool suspend, bool fbcon)
struct drm_framebuffer *fb = crtc->primary->fb;
struct amdgpu_bo *robj;
if (amdgpu_crtc->cursor_bo) {
if (amdgpu_crtc->cursor_bo && !adev->enable_virtual_display) {
struct amdgpu_bo *aobj = gem_to_amdgpu_bo(amdgpu_crtc->cursor_bo);
r = amdgpu_bo_reserve(aobj, true);
if (r == 0) {
@ -2906,7 +2920,7 @@ int amdgpu_device_resume(struct drm_device *dev, bool resume, bool fbcon)
list_for_each_entry(crtc, &dev->mode_config.crtc_list, head) {
struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
if (amdgpu_crtc->cursor_bo) {
if (amdgpu_crtc->cursor_bo && !adev->enable_virtual_display) {
struct amdgpu_bo *aobj = gem_to_amdgpu_bo(amdgpu_crtc->cursor_bo);
r = amdgpu_bo_reserve(aobj, true);
if (r == 0) {
@ -3226,6 +3240,7 @@ static int amdgpu_device_reset_sriov(struct amdgpu_device *adev,
r = amdgpu_ib_ring_tests(adev);
error:
amdgpu_virt_init_data_exchange(adev);
amdgpu_virt_release_full_gpu(adev, true);
if (!r && adev->virt.gim_feature & AMDGIM_FEATURE_GIM_FLR_VRAMLOST) {
atomic_inc(&adev->vram_lost_counter);
@ -3298,17 +3313,15 @@ static int amdgpu_device_pre_asic_reset(struct amdgpu_device *adev,
if (!ring || !ring->sched.thread)
continue;
kthread_park(ring->sched.thread);
if (job && job->base.sched != &ring->sched)
continue;
drm_sched_hw_job_reset(&ring->sched, job ? &job->base : NULL);
drm_sched_stop(&ring->sched);
/* after all hw jobs are reset, hw fence is meaningless, so force_completion */
amdgpu_fence_driver_force_completion(ring);
}
if(job)
drm_sched_increase_karma(&job->base);
if (!amdgpu_sriov_vf(adev)) {
@ -3454,14 +3467,10 @@ static void amdgpu_device_post_asic_reset(struct amdgpu_device *adev,
if (!ring || !ring->sched.thread)
continue;
/* only need recovery sched of the given job's ring
* or all rings (in the case @job is NULL)
* after above amdgpu_reset accomplished
*/
if ((!job || job->base.sched == &ring->sched) && !adev->asic_reset_res)
drm_sched_job_recovery(&ring->sched);
if (!adev->asic_reset_res)
drm_sched_resubmit_jobs(&ring->sched);
kthread_unpark(ring->sched.thread);
drm_sched_start(&ring->sched, !adev->asic_reset_res);
}
if (!amdgpu_device_has_dc_support(adev)) {
@ -3521,9 +3530,9 @@ int amdgpu_device_gpu_recover(struct amdgpu_device *adev,
* by different nodes. No point also since the one node already executing
* reset will also reset all the other nodes in the hive.
*/
hive = amdgpu_get_xgmi_hive(adev);
hive = amdgpu_get_xgmi_hive(adev, 0);
if (hive && adev->gmc.xgmi.num_physical_nodes > 1 &&
!mutex_trylock(&hive->hive_lock))
!mutex_trylock(&hive->reset_lock))
return 0;
/* Start with adev pre asic reset first for soft reset check.*/
@ -3602,13 +3611,45 @@ int amdgpu_device_gpu_recover(struct amdgpu_device *adev,
}
if (hive && adev->gmc.xgmi.num_physical_nodes > 1)
mutex_unlock(&hive->hive_lock);
mutex_unlock(&hive->reset_lock);
if (r)
dev_info(adev->dev, "GPU reset end with ret = %d\n", r);
return r;
}
static void amdgpu_device_get_min_pci_speed_width(struct amdgpu_device *adev,
enum pci_bus_speed *speed,
enum pcie_link_width *width)
{
struct pci_dev *pdev = adev->pdev;
enum pci_bus_speed cur_speed;
enum pcie_link_width cur_width;
*speed = PCI_SPEED_UNKNOWN;
*width = PCIE_LNK_WIDTH_UNKNOWN;
while (pdev) {
cur_speed = pcie_get_speed_cap(pdev);
cur_width = pcie_get_width_cap(pdev);
if (cur_speed != PCI_SPEED_UNKNOWN) {
if (*speed == PCI_SPEED_UNKNOWN)
*speed = cur_speed;
else if (cur_speed < *speed)
*speed = cur_speed;
}
if (cur_width != PCIE_LNK_WIDTH_UNKNOWN) {
if (*width == PCIE_LNK_WIDTH_UNKNOWN)
*width = cur_width;
else if (cur_width < *width)
*width = cur_width;
}
pdev = pci_upstream_bridge(pdev);
}
}
/**
* amdgpu_device_get_pcie_info - fence pcie info about the PCIE slot
*
@ -3621,8 +3662,8 @@ int amdgpu_device_gpu_recover(struct amdgpu_device *adev,
static void amdgpu_device_get_pcie_info(struct amdgpu_device *adev)
{
struct pci_dev *pdev;
enum pci_bus_speed speed_cap;
enum pcie_link_width link_width;
enum pci_bus_speed speed_cap, platform_speed_cap;
enum pcie_link_width platform_link_width;
if (amdgpu_pcie_gen_cap)
adev->pm.pcie_gen_mask = amdgpu_pcie_gen_cap;
@ -3639,6 +3680,12 @@ static void amdgpu_device_get_pcie_info(struct amdgpu_device *adev)
return;
}
if (adev->pm.pcie_gen_mask && adev->pm.pcie_mlw_mask)
return;
amdgpu_device_get_min_pci_speed_width(adev, &platform_speed_cap,
&platform_link_width);
if (adev->pm.pcie_gen_mask == 0) {
/* asic caps */
pdev = adev->pdev;
@ -3664,22 +3711,20 @@ static void amdgpu_device_get_pcie_info(struct amdgpu_device *adev)
adev->pm.pcie_gen_mask |= CAIL_ASIC_PCIE_LINK_SPEED_SUPPORT_GEN1;
}
/* platform caps */
pdev = adev->ddev->pdev->bus->self;
speed_cap = pcie_get_speed_cap(pdev);
if (speed_cap == PCI_SPEED_UNKNOWN) {
if (platform_speed_cap == PCI_SPEED_UNKNOWN) {
adev->pm.pcie_gen_mask |= (CAIL_PCIE_LINK_SPEED_SUPPORT_GEN1 |
CAIL_PCIE_LINK_SPEED_SUPPORT_GEN2);
} else {
if (speed_cap == PCIE_SPEED_16_0GT)
if (platform_speed_cap == PCIE_SPEED_16_0GT)
adev->pm.pcie_gen_mask |= (CAIL_PCIE_LINK_SPEED_SUPPORT_GEN1 |
CAIL_PCIE_LINK_SPEED_SUPPORT_GEN2 |
CAIL_PCIE_LINK_SPEED_SUPPORT_GEN3 |
CAIL_PCIE_LINK_SPEED_SUPPORT_GEN4);
else if (speed_cap == PCIE_SPEED_8_0GT)
else if (platform_speed_cap == PCIE_SPEED_8_0GT)
adev->pm.pcie_gen_mask |= (CAIL_PCIE_LINK_SPEED_SUPPORT_GEN1 |
CAIL_PCIE_LINK_SPEED_SUPPORT_GEN2 |
CAIL_PCIE_LINK_SPEED_SUPPORT_GEN3);
else if (speed_cap == PCIE_SPEED_5_0GT)
else if (platform_speed_cap == PCIE_SPEED_5_0GT)
adev->pm.pcie_gen_mask |= (CAIL_PCIE_LINK_SPEED_SUPPORT_GEN1 |
CAIL_PCIE_LINK_SPEED_SUPPORT_GEN2);
else
@ -3688,12 +3733,10 @@ static void amdgpu_device_get_pcie_info(struct amdgpu_device *adev)
}
}
if (adev->pm.pcie_mlw_mask == 0) {
pdev = adev->ddev->pdev->bus->self;
link_width = pcie_get_width_cap(pdev);
if (link_width == PCIE_LNK_WIDTH_UNKNOWN) {
if (platform_link_width == PCIE_LNK_WIDTH_UNKNOWN) {
adev->pm.pcie_mlw_mask |= AMDGPU_DEFAULT_PCIE_MLW_MASK;
} else {
switch (link_width) {
switch (platform_link_width) {
case PCIE_LNK_X32:
adev->pm.pcie_mlw_mask = (CAIL_PCIE_LINK_WIDTH_SUPPORT_X32 |
CAIL_PCIE_LINK_WIDTH_SUPPORT_X16 |

20
drivers/gpu/drm/amd/amdgpu/amdgpu_display.c
View File

@ -188,10 +188,12 @@ int amdgpu_display_crtc_page_flip_target(struct drm_crtc *crtc,
goto cleanup;
}
r = amdgpu_bo_pin(new_abo, amdgpu_display_supported_domains(adev));
if (unlikely(r != 0)) {
DRM_ERROR("failed to pin new abo buffer before flip\n");
goto unreserve;
if (!adev->enable_virtual_display) {
r = amdgpu_bo_pin(new_abo, amdgpu_display_supported_domains(adev));
if (unlikely(r != 0)) {
DRM_ERROR("failed to pin new abo buffer before flip\n");
goto unreserve;
}
}
r = amdgpu_ttm_alloc_gart(&new_abo->tbo);
@ -211,7 +213,8 @@ int amdgpu_display_crtc_page_flip_target(struct drm_crtc *crtc,
amdgpu_bo_get_tiling_flags(new_abo, &tiling_flags);
amdgpu_bo_unreserve(new_abo);
work->base = amdgpu_bo_gpu_offset(new_abo);
if (!adev->enable_virtual_display)
work->base = amdgpu_bo_gpu_offset(new_abo);
work->target_vblank = target - (uint32_t)drm_crtc_vblank_count(crtc) +
amdgpu_get_vblank_counter_kms(dev, work->crtc_id);
@ -242,9 +245,10 @@ int amdgpu_display_crtc_page_flip_target(struct drm_crtc *crtc,
goto cleanup;
}
unpin:
if (unlikely(amdgpu_bo_unpin(new_abo) != 0)) {
DRM_ERROR("failed to unpin new abo in error path\n");
}
if (!adev->enable_virtual_display)
if (unlikely(amdgpu_bo_unpin(new_abo) != 0))
DRM_ERROR("failed to unpin new abo in error path\n");
unreserve:
amdgpu_bo_unreserve(new_abo);

11
drivers/gpu/drm/amd/amdgpu/amdgpu_doorbell.h
View File

@ -51,14 +51,7 @@ struct amdgpu_doorbell_index {
uint32_t userqueue_start;
uint32_t userqueue_end;
uint32_t gfx_ring0;
uint32_t sdma_engine0;
uint32_t sdma_engine1;
uint32_t sdma_engine2;
uint32_t sdma_engine3;
uint32_t sdma_engine4;
uint32_t sdma_engine5;
uint32_t sdma_engine6;
uint32_t sdma_engine7;
uint32_t sdma_engine[8];
uint32_t ih;
union {
struct {
@ -79,6 +72,8 @@ struct amdgpu_doorbell_index {
} uvd_vce;
};
uint32_t max_assignment;
/* Per engine SDMA doorbell size in dword */
uint32_t sdma_doorbell_range;
};
typedef enum _AMDGPU_DOORBELL_ASSIGNMENT

8
drivers/gpu/drm/amd/amdgpu/amdgpu_dpm.h
View File

@ -364,6 +364,14 @@ enum amdgpu_pcie_gen {
((adev)->powerplay.pp_funcs->enable_mgpu_fan_boost(\
(adev)->powerplay.pp_handle))
#define amdgpu_dpm_get_ppfeature_status(adev, buf) \
((adev)->powerplay.pp_funcs->get_ppfeature_status(\
(adev)->powerplay.pp_handle, (buf)))
#define amdgpu_dpm_set_ppfeature_status(adev, ppfeatures) \
((adev)->powerplay.pp_funcs->set_ppfeature_status(\
(adev)->powerplay.pp_handle, (ppfeatures)))
struct amdgpu_dpm {
struct amdgpu_ps *ps;
/* number of valid power states */

4
drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
View File

@ -71,9 +71,11 @@
* - 3.25.0 - Add support for sensor query info (stable pstate sclk/mclk).
* - 3.26.0 - GFX9: Process AMDGPU_IB_FLAG_TC_WB_NOT_INVALIDATE.
* - 3.27.0 - Add new chunk to to AMDGPU_CS to enable BO_LIST creation.
* - 3.28.0 - Add AMDGPU_CHUNK_ID_SCHEDULED_DEPENDENCIES
* - 3.29.0 - Add AMDGPU_IB_FLAG_RESET_GDS_MAX_WAVE_ID
*/
#define KMS_DRIVER_MAJOR 3
#define KMS_DRIVER_MINOR 27
#define KMS_DRIVER_MINOR 29
#define KMS_DRIVER_PATCHLEVEL 0
int amdgpu_vram_limit = 0;

2
drivers/gpu/drm/amd/amdgpu/amdgpu_gds.h
View File

@ -37,6 +37,8 @@ struct amdgpu_gds {
struct amdgpu_gds_asic_info mem;
struct amdgpu_gds_asic_info gws;
struct amdgpu_gds_asic_info oa;
uint32_t gds_compute_max_wave_id;
/* At present, GDS, GWS and OA resources for gfx (graphics)
* is always pre-allocated and available for graphics operation.
* Such resource is shared between all gfx clients.

7
drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
View File

@ -54,10 +54,6 @@ int amdgpu_gem_object_create(struct amdgpu_device *adev, unsigned long size,
memset(&bp, 0, sizeof(bp));
*obj = NULL;
/* At least align on page size */
if (alignment < PAGE_SIZE) {
alignment = PAGE_SIZE;
}
bp.size = size;
bp.byte_align = alignment;
@ -244,9 +240,6 @@ int amdgpu_gem_create_ioctl(struct drm_device *dev, void *data,
return -EINVAL;
}
flags |= AMDGPU_GEM_CREATE_NO_CPU_ACCESS;
/* GDS allocations must be DW aligned */
if (args->in.domains & AMDGPU_GEM_DOMAIN_GDS)
size = ALIGN(size, 4);
}
if (flags & AMDGPU_GEM_CREATE_VM_ALWAYS_VALID) {

10
drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c
View File

@ -202,12 +202,12 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned num_ibs,
amdgpu_asic_flush_hdp(adev, ring);
}
if (need_ctx_switch)
status |= AMDGPU_HAVE_CTX_SWITCH;
skip_preamble = ring->current_ctx == fence_ctx;
if (job && ring->funcs->emit_cntxcntl) {
if (need_ctx_switch)
status |= AMDGPU_HAVE_CTX_SWITCH;
status |= job->preamble_status;
amdgpu_ring_emit_cntxcntl(ring, status);
}
@ -221,8 +221,8 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned num_ibs,
!amdgpu_sriov_vf(adev)) /* for SRIOV preemption, Preamble CE ib must be inserted anyway */
continue;
amdgpu_ring_emit_ib(ring, job, ib, need_ctx_switch);
need_ctx_switch = false;
amdgpu_ring_emit_ib(ring, job, ib, status);
status &= ~AMDGPU_HAVE_CTX_SWITCH;
}
if (ring->funcs->emit_tmz)

40
drivers/gpu/drm/amd/amdgpu/amdgpu_ih.c
View File

@ -52,6 +52,8 @@ int amdgpu_ih_ring_init(struct amdgpu_device *adev, struct amdgpu_ih_ring *ih,
ih->use_bus_addr = use_bus_addr;
if (use_bus_addr) {
dma_addr_t dma_addr;
if (ih->ring)
return 0;
@ -59,21 +61,26 @@ int amdgpu_ih_ring_init(struct amdgpu_device *adev, struct amdgpu_ih_ring *ih,
* add them to the end of the ring allocation.
*/
ih->ring = dma_alloc_coherent(adev->dev, ih->ring_size + 8,
&ih->rb_dma_addr, GFP_KERNEL);
&dma_addr, GFP_KERNEL);
if (ih->ring == NULL)
return -ENOMEM;
memset((void *)ih->ring, 0, ih->ring_size + 8);
ih->wptr_offs = (ih->ring_size / 4) + 0;
ih->rptr_offs = (ih->ring_size / 4) + 1;
ih->gpu_addr = dma_addr;
ih->wptr_addr = dma_addr + ih->ring_size;
ih->wptr_cpu = &ih->ring[ih->ring_size / 4];
ih->rptr_addr = dma_addr + ih->ring_size + 4;
ih->rptr_cpu = &ih->ring[(ih->ring_size / 4) + 1];
} else {
r = amdgpu_device_wb_get(adev, &ih->wptr_offs);
unsigned wptr_offs, rptr_offs;
r = amdgpu_device_wb_get(adev, &wptr_offs);
if (r)
return r;
r = amdgpu_device_wb_get(adev, &ih->rptr_offs);
r = amdgpu_device_wb_get(adev, &rptr_offs);
if (r) {
amdgpu_device_wb_free(adev, ih->wptr_offs);
amdgpu_device_wb_free(adev, wptr_offs);
return r;
}
@ -82,10 +89,15 @@ int amdgpu_ih_ring_init(struct amdgpu_device *adev, struct amdgpu_ih_ring *ih,
&ih->ring_obj, &ih->gpu_addr,
(void **)&ih->ring);
if (r) {
amdgpu_device_wb_free(adev, ih->rptr_offs);
amdgpu_device_wb_free(adev, ih->wptr_offs);
amdgpu_device_wb_free(adev, rptr_offs);
amdgpu_device_wb_free(adev, wptr_offs);
return r;
}
ih->wptr_addr = adev->wb.gpu_addr + wptr_offs * 4;
ih->wptr_cpu = &adev->wb.wb[wptr_offs];
ih->rptr_addr = adev->wb.gpu_addr + rptr_offs * 4;
ih->rptr_cpu = &adev->wb.wb[rptr_offs];
}
return 0;
}
@ -109,13 +121,13 @@ void amdgpu_ih_ring_fini(struct amdgpu_device *adev, struct amdgpu_ih_ring *ih)
* add them to the end of the ring allocation.
*/
dma_free_coherent(adev->dev, ih->ring_size + 8,
(void *)ih->ring, ih->rb_dma_addr);
(void *)ih->ring, ih->gpu_addr);
ih->ring = NULL;
} else {
amdgpu_bo_free_kernel(&ih->ring_obj, &ih->gpu_addr,
(void **)&ih->ring);
amdgpu_device_wb_free(adev, ih->wptr_offs);
amdgpu_device_wb_free(adev, ih->rptr_offs);
amdgpu_device_wb_free(adev, (ih->wptr_addr - ih->gpu_addr) / 4);
amdgpu_device_wb_free(adev, (ih->rptr_addr - ih->gpu_addr) / 4);
}
}
@ -137,7 +149,7 @@ int amdgpu_ih_process(struct amdgpu_device *adev, struct amdgpu_ih_ring *ih,
if (!ih->enabled || adev->shutdown)
return IRQ_NONE;
wptr = amdgpu_ih_get_wptr(adev);
wptr = amdgpu_ih_get_wptr(adev, ih);
restart_ih:
/* is somebody else already processing irqs? */
@ -154,11 +166,11 @@ int amdgpu_ih_process(struct amdgpu_device *adev, struct amdgpu_ih_ring *ih,
ih->rptr &= ih->ptr_mask;
}
amdgpu_ih_set_rptr(adev);
amdgpu_ih_set_rptr(adev, ih);
atomic_set(&ih->lock, 0);
/* make sure wptr hasn't changed while processing */
wptr = amdgpu_ih_get_wptr(adev);
wptr = amdgpu_ih_get_wptr(adev, ih);
if (wptr != ih->rptr)
goto restart_ih;

36
drivers/gpu/drm/amd/amdgpu/amdgpu_ih.h
View File

@ -31,34 +31,40 @@ struct amdgpu_iv_entry;
* R6xx+ IH ring
*/
struct amdgpu_ih_ring {
struct amdgpu_bo *ring_obj;
volatile uint32_t *ring;
unsigned rptr;
unsigned ring_size;
uint64_t gpu_addr;
uint32_t ptr_mask;
atomic_t lock;
bool enabled;
unsigned wptr_offs;
unsigned rptr_offs;
u32 doorbell_index;
bool use_doorbell;
bool use_bus_addr;
dma_addr_t rb_dma_addr; /* only used when use_bus_addr = true */
struct amdgpu_bo *ring_obj;
volatile uint32_t *ring;
uint64_t gpu_addr;
uint64_t wptr_addr;
volatile uint32_t *wptr_cpu;
uint64_t rptr_addr;
volatile uint32_t *rptr_cpu;
bool enabled;
unsigned rptr;
atomic_t lock;
};
/* provided by the ih block */
struct amdgpu_ih_funcs {
/* ring read/write ptr handling, called from interrupt context */
u32 (*get_wptr)(struct amdgpu_device *adev);
void (*decode_iv)(struct amdgpu_device *adev,
u32 (*get_wptr)(struct amdgpu_device *adev, struct amdgpu_ih_ring *ih);
void (*decode_iv)(struct amdgpu_device *adev, struct amdgpu_ih_ring *ih,
struct amdgpu_iv_entry *entry);
void (*set_rptr)(struct amdgpu_device *adev);
void (*set_rptr)(struct amdgpu_device *adev, struct amdgpu_ih_ring *ih);
};
#define amdgpu_ih_get_wptr(adev) (adev)->irq.ih_funcs->get_wptr((adev))
#define amdgpu_ih_decode_iv(adev, iv) (adev)->irq.ih_funcs->decode_iv((adev), (iv))
#define amdgpu_ih_set_rptr(adev) (adev)->irq.ih_funcs->set_rptr((adev))
#define amdgpu_ih_get_wptr(adev, ih) (adev)->irq.ih_funcs->get_wptr((adev), (ih))
#define amdgpu_ih_decode_iv(adev, iv) \
(adev)->irq.ih_funcs->decode_iv((adev), (ih), (iv))
#define amdgpu_ih_set_rptr(adev, ih) (adev)->irq.ih_funcs->set_rptr((adev), (ih))
int amdgpu_ih_ring_init(struct amdgpu_device *adev, struct amdgpu_ih_ring *ih,
unsigned ring_size, bool use_bus_addr);

37
drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c
View File

@ -148,6 +148,8 @@ static void amdgpu_irq_callback(struct amdgpu_device *adev,
entry.iv_entry = (const uint32_t *)&ih->ring[ring_index];
amdgpu_ih_decode_iv(adev, &entry);
trace_amdgpu_iv(ih - &adev->irq.ih, &entry);
amdgpu_irq_dispatch(adev, &entry);
}
@ -174,6 +176,36 @@ irqreturn_t amdgpu_irq_handler(int irq, void *arg)
return ret;
}
/**
* amdgpu_irq_handle_ih1 - kick of processing for IH1
*
* @work: work structure in struct amdgpu_irq
*
* Kick of processing IH ring 1.
*/
static void amdgpu_irq_handle_ih1(struct work_struct *work)
{
struct amdgpu_device *adev = container_of(work, struct amdgpu_device,
irq.ih1_work);
amdgpu_ih_process(adev, &adev->irq.ih1, amdgpu_irq_callback);
}
/**
* amdgpu_irq_handle_ih2 - kick of processing for IH2
*
* @work: work structure in struct amdgpu_irq
*
* Kick of processing IH ring 2.
*/
static void amdgpu_irq_handle_ih2(struct work_struct *work)
{
struct amdgpu_device *adev = container_of(work, struct amdgpu_device,
irq.ih2_work);
amdgpu_ih_process(adev, &adev->irq.ih2, amdgpu_irq_callback);
}
/**
* amdgpu_msi_ok - check whether MSI functionality is enabled
*
@ -238,6 +270,9 @@ int amdgpu_irq_init(struct amdgpu_device *adev)
amdgpu_hotplug_work_func);
}
INIT_WORK(&adev->irq.ih1_work, amdgpu_irq_handle_ih1);
INIT_WORK(&adev->irq.ih2_work, amdgpu_irq_handle_ih2);
adev->irq.installed = true;
r = drm_irq_install(adev->ddev, adev->ddev->pdev->irq);
if (r) {
@ -367,8 +402,6 @@ void amdgpu_irq_dispatch(struct amdgpu_device *adev,
bool handled = false;
int r;
trace_amdgpu_iv(entry);
if (client_id >= AMDGPU_IRQ_CLIENTID_MAX) {
DRM_DEBUG("Invalid client_id in IV: %d\n", client_id);

8
drivers/gpu/drm/amd/amdgpu/amdgpu_irq.h
View File

@ -87,9 +87,11 @@ struct amdgpu_irq {
/* status, etc. */
bool msi_enabled; /* msi enabled */
/* interrupt ring */
struct amdgpu_ih_ring ih;
const struct amdgpu_ih_funcs *ih_funcs;
/* interrupt rings */
struct amdgpu_ih_ring ih, ih1, ih2;
const struct amdgpu_ih_funcs *ih_funcs;
struct work_struct ih1_work, ih2_work;
struct amdgpu_irq_src self_irq;
/* gen irq stuff */
struct irq_domain *domain; /* GPU irq controller domain */

16
drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
View File

@ -426,12 +426,20 @@ static int amdgpu_bo_do_create(struct amdgpu_device *adev,
size_t acc_size;
int r;
page_align = roundup(bp->byte_align, PAGE_SIZE) >> PAGE_SHIFT;
if (bp->domain & (AMDGPU_GEM_DOMAIN_GDS | AMDGPU_GEM_DOMAIN_GWS |
AMDGPU_GEM_DOMAIN_OA))
/* Note that GDS/GWS/OA allocates 1 page per byte/resource. */
if (bp->domain & (AMDGPU_GEM_DOMAIN_GWS | AMDGPU_GEM_DOMAIN_OA)) {
/* GWS and OA don't need any alignment. */
page_align = bp->byte_align;
size <<= PAGE_SHIFT;
else
} else if (bp->domain & AMDGPU_GEM_DOMAIN_GDS) {
/* Both size and alignment must be a multiple of 4. */
page_align = ALIGN(bp->byte_align, 4);
size = ALIGN(size, 4) << PAGE_SHIFT;
} else {
/* Memory should be aligned at least to a page size. */
page_align = ALIGN(bp->byte_align, PAGE_SIZE) >> PAGE_SHIFT;
size = ALIGN(size, PAGE_SIZE);
}
if (!amdgpu_bo_validate_size(adev, size, bp->domain))
return -ENOMEM;

379
drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c
View File

@ -626,11 +626,71 @@ static ssize_t amdgpu_get_pp_od_clk_voltage(struct device *dev,
}
/**
* DOC: pp_dpm_sclk pp_dpm_mclk pp_dpm_pcie
* DOC: ppfeatures
*
* The amdgpu driver provides a sysfs API for adjusting what powerplay
* features to be enabled. The file ppfeatures is used for this. And
* this is only available for Vega10 and later dGPUs.
*
* Reading back the file will show you the followings:
* - Current ppfeature masks
* - List of the all supported powerplay features with their naming,
* bitmasks and enablement status('Y'/'N' means "enabled"/"disabled").
*
* To manually enable or disable a specific feature, just set or clear
* the corresponding bit from original ppfeature masks and input the
* new ppfeature masks.
*/
static ssize_t amdgpu_set_ppfeature_status(struct device *dev,
struct device_attribute *attr,
const char *buf,
size_t count)
{
struct drm_device *ddev = dev_get_drvdata(dev);
struct amdgpu_device *adev = ddev->dev_private;
uint64_t featuremask;
int ret;
ret = kstrtou64(buf, 0, &featuremask);
if (ret)
return -EINVAL;
pr_debug("featuremask = 0x%llx\n", featuremask);
if (adev->powerplay.pp_funcs->set_ppfeature_status) {
ret = amdgpu_dpm_set_ppfeature_status(adev, featuremask);
if (ret)
return -EINVAL;
}
return count;
}
static ssize_t amdgpu_get_ppfeature_status(struct device *dev,
struct device_attribute *attr,
char *buf)
{
struct drm_device *ddev = dev_get_drvdata(dev);
struct amdgpu_device *adev = ddev->dev_private;
if (adev->powerplay.pp_funcs->get_ppfeature_status)
return amdgpu_dpm_get_ppfeature_status(adev, buf);
return snprintf(buf, PAGE_SIZE, "\n");
}
/**
* DOC: pp_dpm_sclk pp_dpm_mclk pp_dpm_socclk pp_dpm_fclk pp_dpm_dcefclk
* pp_dpm_pcie
*
* The amdgpu driver provides a sysfs API for adjusting what power levels
* are enabled for a given power state. The files pp_dpm_sclk, pp_dpm_mclk,
* and pp_dpm_pcie are used for this.
* pp_dpm_socclk, pp_dpm_fclk, pp_dpm_dcefclk and pp_dpm_pcie are used for
* this.
*
* pp_dpm_socclk and pp_dpm_dcefclk interfaces are only available for
* Vega10 and later ASICs.
* pp_dpm_fclk interface is only available for Vega20 and later ASICs.
*
* Reading back the files will show you the available power levels within
* the power state and the clock information for those levels.
@ -640,6 +700,8 @@ static ssize_t amdgpu_get_pp_od_clk_voltage(struct device *dev,
* Secondly,Enter a new value for each level by inputing a string that
* contains " echo xx xx xx > pp_dpm_sclk/mclk/pcie"
* E.g., echo 4 5 6 to > pp_dpm_sclk will enable sclk levels 4, 5, and 6.
*
* NOTE: change to the dcefclk max dpm level is not supported now
*/
static ssize_t amdgpu_get_pp_dpm_sclk(struct device *dev,
@ -750,6 +812,114 @@ static ssize_t amdgpu_set_pp_dpm_mclk(struct device *dev,
return count;
}
static ssize_t amdgpu_get_pp_dpm_socclk(struct device *dev,
struct device_attribute *attr,
char *buf)
{
struct drm_device *ddev = dev_get_drvdata(dev);
struct amdgpu_device *adev = ddev->dev_private;
if (adev->powerplay.pp_funcs->print_clock_levels)
return amdgpu_dpm_print_clock_levels(adev, PP_SOCCLK, buf);
else
return snprintf(buf, PAGE_SIZE, "\n");
}
static ssize_t amdgpu_set_pp_dpm_socclk(struct device *dev,
struct device_attribute *attr,
const char *buf,
size_t count)
{
struct drm_device *ddev = dev_get_drvdata(dev);
struct amdgpu_device *adev = ddev->dev_private;
int ret;
uint32_t mask = 0;
ret = amdgpu_read_mask(buf, count, &mask);
if (ret)
return ret;
if (adev->powerplay.pp_funcs->force_clock_level)
ret = amdgpu_dpm_force_clock_level(adev, PP_SOCCLK, mask);
if (ret)
return -EINVAL;
return count;
}
static ssize_t amdgpu_get_pp_dpm_fclk(struct device *dev,
struct device_attribute *attr,
char *buf)
{
struct drm_device *ddev = dev_get_drvdata(dev);
struct amdgpu_device *adev = ddev->dev_private;
if (adev->powerplay.pp_funcs->print_clock_levels)
return amdgpu_dpm_print_clock_levels(adev, PP_FCLK, buf);
else
return snprintf(buf, PAGE_SIZE, "\n");
}
static ssize_t amdgpu_set_pp_dpm_fclk(struct device *dev,
struct device_attribute *attr,
const char *buf,
size_t count)
{
struct drm_device *ddev = dev_get_drvdata(dev);
struct amdgpu_device *adev = ddev->dev_private;
int ret;
uint32_t mask = 0;
ret = amdgpu_read_mask(buf, count, &mask);
if (ret)
return ret;
if (adev->powerplay.pp_funcs->force_clock_level)
ret = amdgpu_dpm_force_clock_level(adev, PP_FCLK, mask);
if (ret)
return -EINVAL;
return count;
}
static ssize_t amdgpu_get_pp_dpm_dcefclk(struct device *dev,
struct device_attribute *attr,
char *buf)
{
struct drm_device *ddev = dev_get_drvdata(dev);
struct amdgpu_device *adev = ddev->dev_private;
if (adev->powerplay.pp_funcs->print_clock_levels)
return amdgpu_dpm_print_clock_levels(adev, PP_DCEFCLK, buf);
else
return snprintf(buf, PAGE_SIZE, "\n");
}
static ssize_t amdgpu_set_pp_dpm_dcefclk(struct device *dev,
struct device_attribute *attr,
const char *buf,
size_t count)
{
struct drm_device *ddev = dev_get_drvdata(dev);
struct amdgpu_device *adev = ddev->dev_private;
int ret;
uint32_t mask = 0;
ret = amdgpu_read_mask(buf, count, &mask);
if (ret)
return ret;
if (adev->powerplay.pp_funcs->force_clock_level)
ret = amdgpu_dpm_force_clock_level(adev, PP_DCEFCLK, mask);
if (ret)
return -EINVAL;
return count;
}
static ssize_t amdgpu_get_pp_dpm_pcie(struct device *dev,
struct device_attribute *attr,
char *buf)
@ -990,6 +1160,31 @@ static ssize_t amdgpu_get_busy_percent(struct device *dev,
return snprintf(buf, PAGE_SIZE, "%d\n", value);
}
/**
* DOC: pcie_bw
*
* The amdgpu driver provides a sysfs API for estimating how much data
* has been received and sent by the GPU in the last second through PCIe.
* The file pcie_bw is used for this.
* The Perf counters count the number of received and sent messages and return
* those values, as well as the maximum payload size of a PCIe packet (mps).
* Note that it is not possible to easily and quickly obtain the size of each
* packet transmitted, so we output the max payload size (mps) to allow for
* quick estimation of the PCIe bandwidth usage
*/
static ssize_t amdgpu_get_pcie_bw(struct device *dev,
struct device_attribute *attr,
char *buf)
{
struct drm_device *ddev = dev_get_drvdata(dev);
struct amdgpu_device *adev = ddev->dev_private;
uint64_t count0, count1;
amdgpu_asic_get_pcie_usage(adev, &count0, &count1);
return snprintf(buf, PAGE_SIZE, "%llu %llu %i\n",
count0, count1, pcie_get_mps(adev->pdev));
}
static DEVICE_ATTR(power_dpm_state, S_IRUGO | S_IWUSR, amdgpu_get_dpm_state, amdgpu_set_dpm_state);
static DEVICE_ATTR(power_dpm_force_performance_level, S_IRUGO | S_IWUSR,
amdgpu_get_dpm_forced_performance_level,
@ -1008,6 +1203,15 @@ static DEVICE_ATTR(pp_dpm_sclk, S_IRUGO | S_IWUSR,
static DEVICE_ATTR(pp_dpm_mclk, S_IRUGO | S_IWUSR,
amdgpu_get_pp_dpm_mclk,
amdgpu_set_pp_dpm_mclk);
static DEVICE_ATTR(pp_dpm_socclk, S_IRUGO | S_IWUSR,
amdgpu_get_pp_dpm_socclk,
amdgpu_set_pp_dpm_socclk);
static DEVICE_ATTR(pp_dpm_fclk, S_IRUGO | S_IWUSR,
amdgpu_get_pp_dpm_fclk,
amdgpu_set_pp_dpm_fclk);
static DEVICE_ATTR(pp_dpm_dcefclk, S_IRUGO | S_IWUSR,
amdgpu_get_pp_dpm_dcefclk,
amdgpu_set_pp_dpm_dcefclk);
static DEVICE_ATTR(pp_dpm_pcie, S_IRUGO | S_IWUSR,
amdgpu_get_pp_dpm_pcie,
amdgpu_set_pp_dpm_pcie);
@ -1025,6 +1229,10 @@ static DEVICE_ATTR(pp_od_clk_voltage, S_IRUGO | S_IWUSR,
amdgpu_set_pp_od_clk_voltage);
static DEVICE_ATTR(gpu_busy_percent, S_IRUGO,
amdgpu_get_busy_percent, NULL);
static DEVICE_ATTR(pcie_bw, S_IRUGO, amdgpu_get_pcie_bw, NULL);
static DEVICE_ATTR(ppfeatures, S_IRUGO | S_IWUSR,
amdgpu_get_ppfeature_status,
amdgpu_set_ppfeature_status);
static ssize_t amdgpu_hwmon_show_temp(struct device *dev,
struct device_attribute *attr,
@ -1516,6 +1724,75 @@ static ssize_t amdgpu_hwmon_set_power_cap(struct device *dev,
return count;
}
static ssize_t amdgpu_hwmon_show_sclk(struct device *dev,
struct device_attribute *attr,
char *buf)
{
struct amdgpu_device *adev = dev_get_drvdata(dev);
struct drm_device *ddev = adev->ddev;
uint32_t sclk;
int r, size = sizeof(sclk);
/* Can't get voltage when the card is off */
if ((adev->flags & AMD_IS_PX) &&
(ddev->switch_power_state != DRM_SWITCH_POWER_ON))
return -EINVAL;
/* sanity check PP is enabled */
if (!(adev->powerplay.pp_funcs &&
adev->powerplay.pp_funcs->read_sensor))
return -EINVAL;
/* get the sclk */
r = amdgpu_dpm_read_sensor(adev, AMDGPU_PP_SENSOR_GFX_SCLK,
(void *)&sclk, &size);
if (r)
return r;
return snprintf(buf, PAGE_SIZE, "%d\n", sclk * 10 * 1000);
}
static ssize_t amdgpu_hwmon_show_sclk_label(struct device *dev,
struct device_attribute *attr,
char *buf)
{
return snprintf(buf, PAGE_SIZE, "sclk\n");
}
static ssize_t amdgpu_hwmon_show_mclk(struct device *dev,
struct device_attribute *attr,
char *buf)
{
struct amdgpu_device *adev = dev_get_drvdata(dev);
struct drm_device *ddev = adev->ddev;
uint32_t mclk;
int r, size = sizeof(mclk);
/* Can't get voltage when the card is off */
if ((adev->flags & AMD_IS_PX) &&
(ddev->switch_power_state != DRM_SWITCH_POWER_ON))
return -EINVAL;
/* sanity check PP is enabled */
if (!(adev->powerplay.pp_funcs &&
adev->powerplay.pp_funcs->read_sensor))
return -EINVAL;
/* get the sclk */
r = amdgpu_dpm_read_sensor(adev, AMDGPU_PP_SENSOR_GFX_MCLK,
(void *)&mclk, &size);
if (r)
return r;
return snprintf(buf, PAGE_SIZE, "%d\n", mclk * 10 * 1000);
}
static ssize_t amdgpu_hwmon_show_mclk_label(struct device *dev,
struct device_attribute *attr,
char *buf)
{
return snprintf(buf, PAGE_SIZE, "mclk\n");
}
/**
* DOC: hwmon
@ -1532,6 +1809,10 @@ static ssize_t amdgpu_hwmon_set_power_cap(struct device *dev,
*
* - GPU fan
*
* - GPU gfx/compute engine clock
*
* - GPU memory clock (dGPU only)
*
* hwmon interfaces for GPU temperature:
*
* - temp1_input: the on die GPU temperature in millidegrees Celsius
@ -1576,6 +1857,12 @@ static ssize_t amdgpu_hwmon_set_power_cap(struct device *dev,
*
* - fan[1-*]_enable: Enable or disable the sensors.1: Enable 0: Disable
*
* hwmon interfaces for GPU clocks:
*
* - freq1_input: the gfx/compute clock in hertz
*
* - freq2_input: the memory clock in hertz
*
* You can use hwmon tools like sensors to view this information on your system.
*
*/
@ -1600,6 +1887,10 @@ static SENSOR_DEVICE_ATTR(power1_average, S_IRUGO, amdgpu_hwmon_show_power_avg,
static SENSOR_DEVICE_ATTR(power1_cap_max, S_IRUGO, amdgpu_hwmon_show_power_cap_max, NULL, 0);
static SENSOR_DEVICE_ATTR(power1_cap_min, S_IRUGO, amdgpu_hwmon_show_power_cap_min, NULL, 0);
static SENSOR_DEVICE_ATTR(power1_cap, S_IRUGO | S_IWUSR, amdgpu_hwmon_show_power_cap, amdgpu_hwmon_set_power_cap, 0);
static SENSOR_DEVICE_ATTR(freq1_input, S_IRUGO, amdgpu_hwmon_show_sclk, NULL, 0);
static SENSOR_DEVICE_ATTR(freq1_label, S_IRUGO, amdgpu_hwmon_show_sclk_label, NULL, 0);
static SENSOR_DEVICE_ATTR(freq2_input, S_IRUGO, amdgpu_hwmon_show_mclk, NULL, 0);
static SENSOR_DEVICE_ATTR(freq2_label, S_IRUGO, amdgpu_hwmon_show_mclk_label, NULL, 0);
static struct attribute *hwmon_attributes[] = {
&sensor_dev_attr_temp1_input.dev_attr.attr,
@ -1622,6 +1913,10 @@ static struct attribute *hwmon_attributes[] = {
&sensor_dev_attr_power1_cap_max.dev_attr.attr,
&sensor_dev_attr_power1_cap_min.dev_attr.attr,
&sensor_dev_attr_power1_cap.dev_attr.attr,
&sensor_dev_attr_freq1_input.dev_attr.attr,
&sensor_dev_attr_freq1_label.dev_attr.attr,
&sensor_dev_attr_freq2_input.dev_attr.attr,
&sensor_dev_attr_freq2_label.dev_attr.attr,
NULL
};
@ -1686,7 +1981,8 @@ static umode_t hwmon_attributes_visible(struct kobject *kobj,
effective_mode &= ~S_IWUSR;
if ((adev->flags & AMD_IS_APU) &&
(attr == &sensor_dev_attr_power1_cap_max.dev_attr.attr ||
(attr == &sensor_dev_attr_power1_average.dev_attr.attr ||
attr == &sensor_dev_attr_power1_cap_max.dev_attr.attr ||
attr == &sensor_dev_attr_power1_cap_min.dev_attr.attr||
attr == &sensor_dev_attr_power1_cap.dev_attr.attr))
return 0;
@ -1712,6 +2008,12 @@ static umode_t hwmon_attributes_visible(struct kobject *kobj,
attr == &sensor_dev_attr_in1_label.dev_attr.attr))
return 0;
/* no mclk on APUs */
if ((adev->flags & AMD_IS_APU) &&
(attr == &sensor_dev_attr_freq2_input.dev_attr.attr ||
attr == &sensor_dev_attr_freq2_label.dev_attr.attr))
return 0;
return effective_mode;
}
@ -2008,6 +2310,7 @@ void amdgpu_pm_print_power_states(struct amdgpu_device *adev)
int amdgpu_pm_sysfs_init(struct amdgpu_device *adev)
{
struct pp_hwmgr *hwmgr = adev->powerplay.pp_handle;
int ret;
if (adev->pm.sysfs_initialized)
@ -2069,6 +2372,25 @@ int amdgpu_pm_sysfs_init(struct amdgpu_device *adev)
DRM_ERROR("failed to create device file pp_dpm_mclk\n");
return ret;
}
if (adev->asic_type >= CHIP_VEGA10) {
ret = device_create_file(adev->dev, &dev_attr_pp_dpm_socclk);
if (ret) {
DRM_ERROR("failed to create device file pp_dpm_socclk\n");
return ret;
}
ret = device_create_file(adev->dev, &dev_attr_pp_dpm_dcefclk);
if (ret) {
DRM_ERROR("failed to create device file pp_dpm_dcefclk\n");
return ret;
}
}
if (adev->asic_type >= CHIP_VEGA20) {
ret = device_create_file(adev->dev, &dev_attr_pp_dpm_fclk);
if (ret) {
DRM_ERROR("failed to create device file pp_dpm_fclk\n");
return ret;
}
}
ret = device_create_file(adev->dev, &dev_attr_pp_dpm_pcie);
if (ret) {
DRM_ERROR("failed to create device file pp_dpm_pcie\n");
@ -2091,12 +2413,14 @@ int amdgpu_pm_sysfs_init(struct amdgpu_device *adev)
"pp_power_profile_mode\n");
return ret;
}
ret = device_create_file(adev->dev,
&dev_attr_pp_od_clk_voltage);
if (ret) {
DRM_ERROR("failed to create device file "
"pp_od_clk_voltage\n");
return ret;
if (hwmgr->od_enabled) {
ret = device_create_file(adev->dev,
&dev_attr_pp_od_clk_voltage);
if (ret) {
DRM_ERROR("failed to create device file "
"pp_od_clk_voltage\n");
return ret;
}
}
ret = device_create_file(adev->dev,
&dev_attr_gpu_busy_percent);
@ -2105,12 +2429,31 @@ int amdgpu_pm_sysfs_init(struct amdgpu_device *adev)
"gpu_busy_level\n");
return ret;
}
/* PCIe Perf counters won't work on APU nodes */
if (!(adev->flags & AMD_IS_APU)) {
ret = device_create_file(adev->dev, &dev_attr_pcie_bw);
if (ret) {
DRM_ERROR("failed to create device file pcie_bw\n");
return ret;
}
}
ret = amdgpu_debugfs_pm_init(adev);
if (ret) {
DRM_ERROR("Failed to register debugfs file for dpm!\n");
return ret;
}
if ((adev->asic_type >= CHIP_VEGA10) &&
!(adev->flags & AMD_IS_APU)) {
ret = device_create_file(adev->dev,
&dev_attr_ppfeatures);
if (ret) {
DRM_ERROR("failed to create device file "
"ppfeatures\n");
return ret;
}
}
adev->pm.sysfs_initialized = true;
return 0;
@ -2118,6 +2461,8 @@ int amdgpu_pm_sysfs_init(struct amdgpu_device *adev)
void amdgpu_pm_sysfs_fini(struct amdgpu_device *adev)
{
struct pp_hwmgr *hwmgr = adev->powerplay.pp_handle;
if (adev->pm.dpm_enabled == 0)
return;
@ -2133,14 +2478,26 @@ void amdgpu_pm_sysfs_fini(struct amdgpu_device *adev)
device_remove_file(adev->dev, &dev_attr_pp_dpm_sclk);
device_remove_file(adev->dev, &dev_attr_pp_dpm_mclk);
if (adev->asic_type >= CHIP_VEGA10) {
device_remove_file(adev->dev, &dev_attr_pp_dpm_socclk);
device_remove_file(adev->dev, &dev_attr_pp_dpm_dcefclk);
}
device_remove_file(adev->dev, &dev_attr_pp_dpm_pcie);
if (adev->asic_type >= CHIP_VEGA20)
device_remove_file(adev->dev, &dev_attr_pp_dpm_fclk);
device_remove_file(adev->dev, &dev_attr_pp_sclk_od);
device_remove_file(adev->dev, &dev_attr_pp_mclk_od);
device_remove_file(adev->dev,
&dev_attr_pp_power_profile_mode);
device_remove_file(adev->dev,
&dev_attr_pp_od_clk_voltage);
if (hwmgr->od_enabled)
device_remove_file(adev->dev,
&dev_attr_pp_od_clk_voltage);
device_remove_file(adev->dev, &dev_attr_gpu_busy_percent);
if (!(adev->flags & AMD_IS_APU))
device_remove_file(adev->dev, &dev_attr_pcie_bw);
if ((adev->asic_type >= CHIP_VEGA10) &&
!(adev->flags & AMD_IS_APU))
device_remove_file(adev->dev, &dev_attr_ppfeatures);
}
void amdgpu_pm_compute_clocks(struct amdgpu_device *adev)

141
drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c
View File

@ -67,9 +67,6 @@ static int psp_sw_init(void *handle)
psp->adev = adev;
if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP)
return 0;
ret = psp_init_microcode(psp);
if (ret) {
DRM_ERROR("Failed to load psp firmware!\n");
@ -83,9 +80,6 @@ static int psp_sw_fini(void *handle)
{
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP)
return 0;
release_firmware(adev->psp.sos_fw);
adev->psp.sos_fw = NULL;
release_firmware(adev->psp.asd_fw);
@ -140,13 +134,24 @@ psp_cmd_submit_buf(struct psp_context *psp,
while (*((unsigned int *)psp->fence_buf) != index)
msleep(1);
/* the status field must be 0 after FW is loaded */
if (ucode && psp->cmd_buf_mem->resp.status) {
DRM_ERROR("failed loading with status (%d) and ucode id (%d)\n",
psp->cmd_buf_mem->resp.status, ucode->ucode_id);
return -EINVAL;
/* In some cases, psp response status is not 0 even there is no
* problem while the command is submitted. Some version of PSP FW
* doesn't write 0 to that field.
* So here we would like to only print a warning instead of an error
* during psp initialization to avoid breaking hw_init and it doesn't
* return -EINVAL.
*/
if (psp->cmd_buf_mem->resp.status) {
if (ucode)
DRM_WARN("failed to load ucode id (%d) ",
ucode->ucode_id);
DRM_WARN("psp command failed and response status is (%d)\n",
psp->cmd_buf_mem->resp.status);
}
/* get xGMI session id from response buffer */
cmd->resp.session_id = psp->cmd_buf_mem->resp.session_id;
if (ucode) {
ucode->tmr_mc_addr_lo = psp->cmd_buf_mem->resp.fw_addr_lo;
ucode->tmr_mc_addr_hi = psp->cmd_buf_mem->resp.fw_addr_hi;
@ -495,6 +500,98 @@ static int psp_hw_start(struct psp_context *psp)
return 0;
}
static int psp_get_fw_type(struct amdgpu_firmware_info *ucode,
enum psp_gfx_fw_type *type)
{
switch (ucode->ucode_id) {
case AMDGPU_UCODE_ID_SDMA0:
*type = GFX_FW_TYPE_SDMA0;
break;
case AMDGPU_UCODE_ID_SDMA1:
*type = GFX_FW_TYPE_SDMA1;
break;
case AMDGPU_UCODE_ID_CP_CE:
*type = GFX_FW_TYPE_CP_CE;
break;
case AMDGPU_UCODE_ID_CP_PFP:
*type = GFX_FW_TYPE_CP_PFP;
break;
case AMDGPU_UCODE_ID_CP_ME:
*type = GFX_FW_TYPE_CP_ME;
break;
case AMDGPU_UCODE_ID_CP_MEC1:
*type = GFX_FW_TYPE_CP_MEC;
break;
case AMDGPU_UCODE_ID_CP_MEC1_JT:
*type = GFX_FW_TYPE_CP_MEC_ME1;
break;
case AMDGPU_UCODE_ID_CP_MEC2:
*type = GFX_FW_TYPE_CP_MEC;
break;
case AMDGPU_UCODE_ID_CP_MEC2_JT:
*type = GFX_FW_TYPE_CP_MEC_ME2;
break;
case AMDGPU_UCODE_ID_RLC_G:
*type = GFX_FW_TYPE_RLC_G;
break;
case AMDGPU_UCODE_ID_RLC_RESTORE_LIST_CNTL:
*type = GFX_FW_TYPE_RLC_RESTORE_LIST_SRM_CNTL;
break;
case AMDGPU_UCODE_ID_RLC_RESTORE_LIST_GPM_MEM:
*type = GFX_FW_TYPE_RLC_RESTORE_LIST_GPM_MEM;
break;
case AMDGPU_UCODE_ID_RLC_RESTORE_LIST_SRM_MEM:
*type = GFX_FW_TYPE_RLC_RESTORE_LIST_SRM_MEM;
break;
case AMDGPU_UCODE_ID_SMC:
*type = GFX_FW_TYPE_SMU;
break;
case AMDGPU_UCODE_ID_UVD:
*type = GFX_FW_TYPE_UVD;
break;
case AMDGPU_UCODE_ID_UVD1:
*type = GFX_FW_TYPE_UVD1;
break;
case AMDGPU_UCODE_ID_VCE:
*type = GFX_FW_TYPE_VCE;
break;
case AMDGPU_UCODE_ID_VCN:
*type = GFX_FW_TYPE_VCN;
break;
case AMDGPU_UCODE_ID_DMCU_ERAM:
*type = GFX_FW_TYPE_DMCU_ERAM;
break;
case AMDGPU_UCODE_ID_DMCU_INTV:
*type = GFX_FW_TYPE_DMCU_ISR;
break;
case AMDGPU_UCODE_ID_MAXIMUM:
default:
return -EINVAL;
}
return 0;
}
static int psp_prep_load_ip_fw_cmd_buf(struct amdgpu_firmware_info *ucode,
struct psp_gfx_cmd_resp *cmd)
{
int ret;
uint64_t fw_mem_mc_addr = ucode->mc_addr;
memset(cmd, 0, sizeof(struct psp_gfx_cmd_resp));
cmd->cmd_id = GFX_CMD_ID_LOAD_IP_FW;
cmd->cmd.cmd_load_ip_fw.fw_phy_addr_lo = lower_32_bits(fw_mem_mc_addr);
cmd->cmd.cmd_load_ip_fw.fw_phy_addr_hi = upper_32_bits(fw_mem_mc_addr);
cmd->cmd.cmd_load_ip_fw.fw_size = ucode->ucode_size;
ret = psp_get_fw_type(ucode, &cmd->cmd.cmd_load_ip_fw.fw_type);
if (ret)
DRM_ERROR("Unknown firmware type\n");
return ret;
}
static int psp_np_fw_load(struct psp_context *psp)
{
int i, ret;
@ -516,7 +613,7 @@ static int psp_np_fw_load(struct psp_context *psp)
/*skip ucode loading in SRIOV VF */
continue;
ret = psp_prep_cmd_buf(ucode, psp->cmd);
ret = psp_prep_load_ip_fw_cmd_buf(ucode, psp->cmd);
if (ret)
return ret;
@ -541,7 +638,7 @@ static int psp_load_fw(struct amdgpu_device *adev)
struct psp_context *psp = &adev->psp;
if (amdgpu_sriov_vf(adev) && adev->in_gpu_reset) {
psp_ring_destroy(psp, PSP_RING_TYPE__KM);
psp_ring_stop(psp, PSP_RING_TYPE__KM); /* should not destroy ring, only stop */
goto skip_memalloc;
}
@ -618,10 +715,6 @@ static int psp_hw_init(void *handle)
int ret;
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP)
return 0;
mutex_lock(&adev->firmware.mutex);
/*
* This sequence is just used on hw_init only once, no need on
@ -651,9 +744,6 @@ static int psp_hw_fini(void *handle)
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
struct psp_context *psp = &adev->psp;
if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP)
return 0;
if (adev->gmc.xgmi.num_physical_nodes > 1 &&
psp->xgmi_context.initialized == 1)
psp_xgmi_terminate(psp);
@ -682,9 +772,6 @@ static int psp_suspend(void *handle)
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
struct psp_context *psp = &adev->psp;
if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP)
return 0;
if (adev->gmc.xgmi.num_physical_nodes > 1 &&
psp->xgmi_context.initialized == 1) {
ret = psp_xgmi_terminate(psp);
@ -709,9 +796,6 @@ static int psp_resume(void *handle)
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
struct psp_context *psp = &adev->psp;
if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP)
return 0;
DRM_INFO("PSP is resuming...\n");
mutex_lock(&adev->firmware.mutex);
@ -747,11 +831,6 @@ static bool psp_check_fw_loading_status(struct amdgpu_device *adev,
{
struct amdgpu_firmware_info *ucode = NULL;
if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP) {
DRM_INFO("firmware is not loaded by PSP\n");
return true;
}
if (!adev->firmware.fw_size)
return false;

3
drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h
View File

@ -65,8 +65,6 @@ struct psp_funcs
int (*init_microcode)(struct psp_context *psp);
int (*bootloader_load_sysdrv)(struct psp_context *psp);
int (*bootloader_load_sos)(struct psp_context *psp);
int (*prep_cmd_buf)(struct amdgpu_firmware_info *ucode,
struct psp_gfx_cmd_resp *cmd);
int (*ring_init)(struct psp_context *psp, enum psp_ring_type ring_type);
int (*ring_create)(struct psp_context *psp,
enum psp_ring_type ring_type);
@ -176,7 +174,6 @@ struct psp_xgmi_topology_info {
struct psp_xgmi_node_info nodes[AMDGPU_XGMI_MAX_CONNECTED_NODES];
};
#define psp_prep_cmd_buf(ucode, type) (psp)->funcs->prep_cmd_buf((ucode), (type))
#define psp_ring_init(psp, type) (psp)->funcs->ring_init((psp), (type))
#define psp_ring_create(psp, type) (psp)->funcs->ring_create((psp), (type))
#define psp_ring_stop(psp, type) (psp)->funcs->ring_stop((psp), (type))

4
drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h
View File

@ -131,7 +131,7 @@ struct amdgpu_ring_funcs {
void (*emit_ib)(struct amdgpu_ring *ring,
struct amdgpu_job *job,
struct amdgpu_ib *ib,
bool ctx_switch);
uint32_t flags);
void (*emit_fence)(struct amdgpu_ring *ring, uint64_t addr,
uint64_t seq, unsigned flags);
void (*emit_pipeline_sync)(struct amdgpu_ring *ring);
@ -229,7 +229,7 @@ struct amdgpu_ring {
#define amdgpu_ring_get_rptr(r) (r)->funcs->get_rptr((r))
#define amdgpu_ring_get_wptr(r) (r)->funcs->get_wptr((r))
#define amdgpu_ring_set_wptr(r) (r)->funcs->set_wptr((r))
#define amdgpu_ring_emit_ib(r, job, ib, c) ((r)->funcs->emit_ib((r), (job), (ib), (c)))
#define amdgpu_ring_emit_ib(r, job, ib, flags) ((r)->funcs->emit_ib((r), (job), (ib), (flags)))
#define amdgpu_ring_emit_pipeline_sync(r) (r)->funcs->emit_pipeline_sync((r))
#define amdgpu_ring_emit_vm_flush(r, vmid, addr) (r)->funcs->emit_vm_flush((r), (vmid), (addr))
#define amdgpu_ring_emit_fence(r, addr, seq, flags) (r)->funcs->emit_fence((r), (addr), (seq), (flags))

11
drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h
View File

@ -76,9 +76,10 @@ TRACE_EVENT(amdgpu_mm_wreg,
);
TRACE_EVENT(amdgpu_iv,
TP_PROTO(struct amdgpu_iv_entry *iv),
TP_ARGS(iv),
TP_PROTO(unsigned ih, struct amdgpu_iv_entry *iv),
TP_ARGS(ih, iv),
TP_STRUCT__entry(
__field(unsigned, ih)
__field(unsigned, client_id)
__field(unsigned, src_id)
__field(unsigned, ring_id)
@ -90,6 +91,7 @@ TRACE_EVENT(amdgpu_iv,
__array(unsigned, src_data, 4)
),
TP_fast_assign(
__entry->ih = ih;
__entry->client_id = iv->client_id;
__entry->src_id = iv->src_id;
__entry->ring_id = iv->ring_id;
@ -103,8 +105,9 @@ TRACE_EVENT(amdgpu_iv,
__entry->src_data[2] = iv->src_data[2];
__entry->src_data[3] = iv->src_data[3];
),
TP_printk("client_id:%u src_id:%u ring:%u vmid:%u timestamp: %llu pasid:%u src_data: %08x %08x %08x %08x",
__entry->client_id, __entry->src_id,
TP_printk("ih:%u client_id:%u src_id:%u ring:%u vmid:%u "
"timestamp: %llu pasid:%u src_data: %08x %08x %08x %08x",
__entry->ih, __entry->client_id, __entry->src_id,
__entry->ring_id, __entry->vmid,
__entry->timestamp, __entry->pasid,
__entry->src_data[0], __entry->src_data[1],

9
drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
View File

@ -1546,7 +1546,8 @@ static struct ttm_bo_driver amdgpu_bo_driver = {
.io_mem_reserve = &amdgpu_ttm_io_mem_reserve,
.io_mem_free = &amdgpu_ttm_io_mem_free,
.io_mem_pfn = amdgpu_ttm_io_mem_pfn,
.access_memory = &amdgpu_ttm_access_memory
.access_memory = &amdgpu_ttm_access_memory,
.del_from_lru_notify = &amdgpu_vm_del_from_lru_notify
};
/*
@ -1755,7 +1756,7 @@ int amdgpu_ttm_init(struct amdgpu_device *adev)
}
r = amdgpu_bo_create_kernel(adev, adev->gds.mem.gfx_partition_size,
PAGE_SIZE, AMDGPU_GEM_DOMAIN_GDS,
4, AMDGPU_GEM_DOMAIN_GDS,
&adev->gds.gds_gfx_bo, NULL, NULL);
if (r)
return r;
@ -1768,7 +1769,7 @@ int amdgpu_ttm_init(struct amdgpu_device *adev)
}
r = amdgpu_bo_create_kernel(adev, adev->gds.gws.gfx_partition_size,
PAGE_SIZE, AMDGPU_GEM_DOMAIN_GWS,
1, AMDGPU_GEM_DOMAIN_GWS,
&adev->gds.gws_gfx_bo, NULL, NULL);
if (r)
return r;
@ -1781,7 +1782,7 @@ int amdgpu_ttm_init(struct amdgpu_device *adev)
}
r = amdgpu_bo_create_kernel(adev, adev->gds.oa.gfx_partition_size,
PAGE_SIZE, AMDGPU_GEM_DOMAIN_OA,
1, AMDGPU_GEM_DOMAIN_OA,
&adev->gds.oa_gfx_bo, NULL, NULL);
if (r)
return r;

2
drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c
View File

@ -1035,7 +1035,7 @@ int amdgpu_vce_ring_parse_cs_vm(struct amdgpu_cs_parser *p, uint32_t ib_idx)
void amdgpu_vce_ring_emit_ib(struct amdgpu_ring *ring,
struct amdgpu_job *job,
struct amdgpu_ib *ib,
bool ctx_switch)
uint32_t flags)
{
amdgpu_ring_write(ring, VCE_CMD_IB);
amdgpu_ring_write(ring, lower_32_bits(ib->gpu_addr));

2
drivers/gpu/drm/amd/amdgpu/amdgpu_vce.h
View File

@ -66,7 +66,7 @@ void amdgpu_vce_free_handles(struct amdgpu_device *adev, struct drm_file *filp);
int amdgpu_vce_ring_parse_cs(struct amdgpu_cs_parser *p, uint32_t ib_idx);
int amdgpu_vce_ring_parse_cs_vm(struct amdgpu_cs_parser *p, uint32_t ib_idx);
void amdgpu_vce_ring_emit_ib(struct amdgpu_ring *ring, struct amdgpu_job *job,
struct amdgpu_ib *ib, bool ctx_switch);
struct amdgpu_ib *ib, uint32_t flags);
void amdgpu_vce_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq,
unsigned flags);
int amdgpu_vce_ring_test_ring(struct amdgpu_ring *ring);

98
drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
View File

@ -107,14 +107,6 @@ struct amdgpu_pte_update_params {
* DMA addresses to use for mapping, used during VM update by CPU
*/
dma_addr_t *pages_addr;
/**
* @kptr:
*
* Kernel pointer of PD/PT BO that needs to be updated,
* used during VM update by CPU
*/
void *kptr;
};
/**
@ -623,6 +615,28 @@ void amdgpu_vm_get_pd_bo(struct amdgpu_vm *vm,
list_add(&entry->tv.head, validated);
}
void amdgpu_vm_del_from_lru_notify(struct ttm_buffer_object *bo)
{
struct amdgpu_bo *abo;
struct amdgpu_vm_bo_base *bo_base;
if (!amdgpu_bo_is_amdgpu_bo(bo))
return;
if (bo->mem.placement & TTM_PL_FLAG_NO_EVICT)
return;
abo = ttm_to_amdgpu_bo(bo);
if (!abo->parent)
return;
for (bo_base = abo->vm_bo; bo_base; bo_base = bo_base->next) {
struct amdgpu_vm *vm = bo_base->vm;
if (abo->tbo.resv == vm->root.base.bo->tbo.resv)
vm->bulk_moveable = false;
}
}
/**
* amdgpu_vm_move_to_lru_tail - move all BOs to the end of LRU
*
@ -799,9 +813,16 @@ static int amdgpu_vm_clear_bo(struct amdgpu_device *adev,
addr += ats_entries * 8;
}
if (entries)
if (entries) {
uint64_t value = 0;
/* Workaround for fault priority problem on GMC9 */
if (level == AMDGPU_VM_PTB && adev->asic_type >= CHIP_VEGA10)
value = AMDGPU_PTE_EXECUTABLE;
amdgpu_vm_set_pte_pde(adev, &job->ibs[0], addr, 0,
entries, 0, 0);
entries, 0, value);
}
amdgpu_ring_pad_ib(ring, &job->ibs[0]);
@ -847,9 +868,6 @@ static void amdgpu_vm_bo_param(struct amdgpu_device *adev, struct amdgpu_vm *vm,
bp->size = amdgpu_vm_bo_size(adev, level);
bp->byte_align = AMDGPU_GPU_PAGE_SIZE;
bp->domain = AMDGPU_GEM_DOMAIN_VRAM;
if (bp->size <= PAGE_SIZE && adev->asic_type >= CHIP_VEGA10 &&
adev->flags & AMD_IS_APU)
bp->domain |= AMDGPU_GEM_DOMAIN_GTT;
bp->domain = amdgpu_bo_get_preferred_pin_domain(adev, bp->domain);
bp->flags = AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS |
AMDGPU_GEM_CREATE_CPU_GTT_USWC;
@ -1506,20 +1524,27 @@ int amdgpu_vm_update_directories(struct amdgpu_device *adev,
}
/**
* amdgpu_vm_update_huge - figure out parameters for PTE updates
* amdgpu_vm_update_flags - figure out flags for PTE updates
*
* Make sure to set the right flags for the PTEs at the desired level.
*/
static void amdgpu_vm_update_huge(struct amdgpu_pte_update_params *params,
struct amdgpu_bo *bo, unsigned level,
uint64_t pe, uint64_t addr,
unsigned count, uint32_t incr,
uint64_t flags)
static void amdgpu_vm_update_flags(struct amdgpu_pte_update_params *params,
struct amdgpu_bo *bo, unsigned level,
uint64_t pe, uint64_t addr,
unsigned count, uint32_t incr,
uint64_t flags)
{
if (level != AMDGPU_VM_PTB) {
flags |= AMDGPU_PDE_PTE;
amdgpu_gmc_get_vm_pde(params->adev, level, &addr, &flags);
} else if (params->adev->asic_type >= CHIP_VEGA10 &&
!(flags & AMDGPU_PTE_VALID) &&
!(flags & AMDGPU_PTE_PRT)) {
/* Workaround for fault priority problem on GMC9 */
flags |= AMDGPU_PTE_EXECUTABLE;
}
amdgpu_vm_update_func(params, bo, pe, addr, count, incr, flags);
@ -1676,9 +1701,9 @@ static int amdgpu_vm_update_ptes(struct amdgpu_pte_update_params *params,
uint64_t upd_end = min(entry_end, frag_end);
unsigned nptes = (upd_end - frag_start) >> shift;
amdgpu_vm_update_huge(params, pt, cursor.level,
pe_start, dst, nptes, incr,
flags | AMDGPU_PTE_FRAG(frag));
amdgpu_vm_update_flags(params, pt, cursor.level,
pe_start, dst, nptes, incr,
flags | AMDGPU_PTE_FRAG(frag));
pe_start += nptes * 8;
dst += (uint64_t)nptes * AMDGPU_GPU_PAGE_SIZE << shift;
@ -1756,13 +1781,20 @@ static int amdgpu_vm_bo_update_mapping(struct amdgpu_device *adev,
if (pages_addr)
params.src = ~0;
/* Wait for PT BOs to be free. PTs share the same resv. object
/* Wait for PT BOs to be idle. PTs share the same resv. object
* as the root PD BO
*/
r = amdgpu_vm_wait_pd(adev, vm, owner);
if (unlikely(r))
return r;
/* Wait for any BO move to be completed */
if (exclusive) {
r = dma_fence_wait(exclusive, true);
if (unlikely(r))
return r;
}
params.func = amdgpu_vm_cpu_set_ptes;
params.pages_addr = pages_addr;
return amdgpu_vm_update_ptes(&params, start, last + 1,
@ -1776,13 +1808,12 @@ static int amdgpu_vm_bo_update_mapping(struct amdgpu_device *adev,
/*
* reserve space for two commands every (1 << BLOCK_SIZE)
* entries or 2k dwords (whatever is smaller)
*
* The second command is for the shadow pagetables.
*/
ncmds = ((nptes >> min(adev->vm_manager.block_size, 11u)) + 1);
/* The second command is for the shadow pagetables. */
if (vm->root.base.bo->shadow)
ncmds = ((nptes >> min(adev->vm_manager.block_size, 11u)) + 1) * 2;
else
ncmds = ((nptes >> min(adev->vm_manager.block_size, 11u)) + 1);
ncmds *= 2;
/* padding, etc. */
ndw = 64;
@ -1801,10 +1832,11 @@ static int amdgpu_vm_bo_update_mapping(struct amdgpu_device *adev,
ndw += ncmds * 10;
/* extra commands for begin/end fragments */
ncmds = 2 * adev->vm_manager.fragment_size;
if (vm->root.base.bo->shadow)
ndw += 2 * 10 * adev->vm_manager.fragment_size * 2;
else
ndw += 2 * 10 * adev->vm_manager.fragment_size;
ncmds *= 2;
ndw += 10 * ncmds;
params.func = amdgpu_vm_do_set_ptes;
}
@ -3006,7 +3038,7 @@ int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm,
}
DRM_DEBUG_DRIVER("VM update mode is %s\n",
vm->use_cpu_for_update ? "CPU" : "SDMA");
WARN_ONCE((vm->use_cpu_for_update & !amdgpu_gmc_vram_full_visible(&adev->gmc)),
WARN_ONCE((vm->use_cpu_for_update && !amdgpu_gmc_vram_full_visible(&adev->gmc)),
"CPU update of VM recommended only for large BAR system\n");
vm->last_update = NULL;
@ -3136,7 +3168,7 @@ int amdgpu_vm_make_compute(struct amdgpu_device *adev, struct amdgpu_vm *vm, uns
vm->pte_support_ats = pte_support_ats;
DRM_DEBUG_DRIVER("VM update mode is %s\n",
vm->use_cpu_for_update ? "CPU" : "SDMA");
WARN_ONCE((vm->use_cpu_for_update & !amdgpu_gmc_vram_full_visible(&adev->gmc)),
WARN_ONCE((vm->use_cpu_for_update && !amdgpu_gmc_vram_full_visible(&adev->gmc)),
"CPU update of VM recommended only for large BAR system\n");
if (vm->pasid) {

2
drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
View File

@ -363,4 +363,6 @@ int amdgpu_vm_add_fault(struct amdgpu_retryfault_hashtable *fault_hash, u64 key)
void amdgpu_vm_clear_fault(struct amdgpu_retryfault_hashtable *fault_hash, u64 key);
void amdgpu_vm_del_from_lru_notify(struct ttm_buffer_object *bo);
#endif

54
drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c
View File

@ -40,26 +40,40 @@ void *amdgpu_xgmi_hive_try_lock(struct amdgpu_hive_info *hive)
return &hive->device_list;
}
struct amdgpu_hive_info *amdgpu_get_xgmi_hive(struct amdgpu_device *adev)
struct amdgpu_hive_info *amdgpu_get_xgmi_hive(struct amdgpu_device *adev, int lock)
{
int i;
struct amdgpu_hive_info *tmp;
if (!adev->gmc.xgmi.hive_id)
return NULL;
mutex_lock(&xgmi_mutex);
for (i = 0 ; i < hive_count; ++i) {
tmp = &xgmi_hives[i];
if (tmp->hive_id == adev->gmc.xgmi.hive_id)
if (tmp->hive_id == adev->gmc.xgmi.hive_id) {
if (lock)
mutex_lock(&tmp->hive_lock);
mutex_unlock(&xgmi_mutex);
return tmp;
}
}
if (i >= AMDGPU_MAX_XGMI_HIVE)
if (i >= AMDGPU_MAX_XGMI_HIVE) {
mutex_unlock(&xgmi_mutex);
return NULL;
}
/* initialize new hive if not exist */
tmp = &xgmi_hives[hive_count++];
tmp->hive_id = adev->gmc.xgmi.hive_id;
INIT_LIST_HEAD(&tmp->device_list);
mutex_init(&tmp->hive_lock);
mutex_init(&tmp->reset_lock);
if (lock)
mutex_lock(&tmp->hive_lock);
mutex_unlock(&xgmi_mutex);
return tmp;
}
@ -77,10 +91,6 @@ int amdgpu_xgmi_update_topology(struct amdgpu_hive_info *hive, struct amdgpu_dev
"XGMI: Set topology failure on device %llx, hive %llx, ret %d",
adev->gmc.xgmi.node_id,
adev->gmc.xgmi.hive_id, ret);
else
dev_info(adev->dev, "XGMI: Set topology for node %d, hive 0x%llx.\n",
adev->gmc.xgmi.physical_node_id,
adev->gmc.xgmi.hive_id);
return ret;
}
@ -111,10 +121,14 @@ int amdgpu_xgmi_add_device(struct amdgpu_device *adev)
return ret;
}
mutex_lock(&xgmi_mutex);
hive = amdgpu_get_xgmi_hive(adev);
if (!hive)
hive = amdgpu_get_xgmi_hive(adev, 1);
if (!hive) {
ret = -EINVAL;
dev_err(adev->dev,
"XGMI: node 0x%llx, can not match hive 0x%llx in the hive list.\n",
adev->gmc.xgmi.node_id, adev->gmc.xgmi.hive_id);
goto exit;
}
hive_topology = &hive->topology_info;
@ -142,8 +156,11 @@ int amdgpu_xgmi_add_device(struct amdgpu_device *adev)
break;
}
dev_info(adev->dev, "XGMI: Add node %d, hive 0x%llx.\n",
adev->gmc.xgmi.physical_node_id, adev->gmc.xgmi.hive_id);
mutex_unlock(&hive->hive_lock);
exit:
mutex_unlock(&xgmi_mutex);
return ret;
}
@ -154,15 +171,14 @@ void amdgpu_xgmi_remove_device(struct amdgpu_device *adev)
if (!adev->gmc.xgmi.supported)
return;
mutex_lock(&xgmi_mutex);
hive = amdgpu_get_xgmi_hive(adev);
hive = amdgpu_get_xgmi_hive(adev, 1);
if (!hive)
goto exit;
return;
if (!(hive->number_devices--))
if (!(hive->number_devices--)) {
mutex_destroy(&hive->hive_lock);
exit:
mutex_unlock(&xgmi_mutex);
mutex_destroy(&hive->reset_lock);
} else {
mutex_unlock(&hive->hive_lock);
}
}

5
drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h
View File

@ -29,10 +29,11 @@ struct amdgpu_hive_info {
struct list_head device_list;
struct psp_xgmi_topology_info topology_info;
int number_devices;
struct mutex hive_lock;
struct mutex hive_lock,
reset_lock;
};
struct amdgpu_hive_info *amdgpu_get_xgmi_hive(struct amdgpu_device *adev);
struct amdgpu_hive_info *amdgpu_get_xgmi_hive(struct amdgpu_device *adev, int lock);
int amdgpu_xgmi_update_topology(struct amdgpu_hive_info *hive, struct amdgpu_device *adev);
int amdgpu_xgmi_add_device(struct amdgpu_device *adev);
void amdgpu_xgmi_remove_device(struct amdgpu_device *adev);

6844
drivers/gpu/drm/amd/amdgpu/ci_dpm.c
View File

File diff suppressed because it is too large Load Diff

349
drivers/gpu/drm/amd/amdgpu/ci_dpm.h
View File

@ -1,349 +0,0 @@
/*
* Copyright 2013 Advanced Micro Devices, Inc.
*
* Permission is hereby granted, free of charge, to any person obtaining a
* copy of this software and associated documentation files (the "Software"),
* to deal in the Software without restriction, including without limitation
* the rights to use, copy, modify, merge, publish, distribute, sublicense,
* and/or sell copies of the Software, and to permit persons to whom the
* Software is furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice shall be included in
* all copies or substantial portions of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
* THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
* OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
* ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
* OTHER DEALINGS IN THE SOFTWARE.
*
*/
#ifndef __CI_DPM_H__
#define __CI_DPM_H__
#include "amdgpu_atombios.h"
#include "ppsmc.h"
#define SMU__NUM_SCLK_DPM_STATE 8
#define SMU__NUM_MCLK_DPM_LEVELS 6
#define SMU__NUM_LCLK_DPM_LEVELS 8
#define SMU__NUM_PCIE_DPM_LEVELS 8
#include "smu7_discrete.h"
#define CISLANDS_MAX_HARDWARE_POWERLEVELS 2
#define CISLANDS_UNUSED_GPIO_PIN 0x7F
struct ci_pl {
u32 mclk;
u32 sclk;
enum amdgpu_pcie_gen pcie_gen;
u16 pcie_lane;
};
struct ci_ps {
u16 performance_level_count;
bool dc_compatible;
u32 sclk_t;
struct ci_pl performance_levels[CISLANDS_MAX_HARDWARE_POWERLEVELS];
};
struct ci_dpm_level {
bool enabled;
u32 value;
u32 param1;
};
#define CISLAND_MAX_DEEPSLEEP_DIVIDER_ID 5
#define MAX_REGULAR_DPM_NUMBER 8
#define CISLAND_MINIMUM_ENGINE_CLOCK 800
struct ci_single_dpm_table {
u32 count;
struct ci_dpm_level dpm_levels[MAX_REGULAR_DPM_NUMBER];
};
struct ci_dpm_table {
struct ci_single_dpm_table sclk_table;
struct ci_single_dpm_table mclk_table;
struct ci_single_dpm_table pcie_speed_table;
struct ci_single_dpm_table vddc_table;
struct ci_single_dpm_table vddci_table;
struct ci_single_dpm_table mvdd_table;
};
struct ci_mc_reg_entry {
u32 mclk_max;
u32 mc_data[SMU7_DISCRETE_MC_REGISTER_ARRAY_SIZE];
};
struct ci_mc_reg_table {
u8 last;
u8 num_entries;
u16 valid_flag;
struct ci_mc_reg_entry mc_reg_table_entry[MAX_AC_TIMING_ENTRIES];
SMU7_Discrete_MCRegisterAddress mc_reg_address[SMU7_DISCRETE_MC_REGISTER_ARRAY_SIZE];
};
struct ci_ulv_parm
{
bool supported;
u32 cg_ulv_parameter;
u32 volt_change_delay;
struct ci_pl pl;
};
#define CISLANDS_MAX_LEAKAGE_COUNT 8
struct ci_leakage_voltage {
u16 count;
u16 leakage_id[CISLANDS_MAX_LEAKAGE_COUNT];
u16 actual_voltage[CISLANDS_MAX_LEAKAGE_COUNT];
};
struct ci_dpm_level_enable_mask {
u32 uvd_dpm_enable_mask;
u32 vce_dpm_enable_mask;
u32 acp_dpm_enable_mask;
u32 samu_dpm_enable_mask;
u32 sclk_dpm_enable_mask;
u32 mclk_dpm_enable_mask;
u32 pcie_dpm_enable_mask;
};
struct ci_vbios_boot_state
{
u16 mvdd_bootup_value;
u16 vddc_bootup_value;
u16 vddci_bootup_value;
u32 sclk_bootup_value;
u32 mclk_bootup_value;
u16 pcie_gen_bootup_value;
u16 pcie_lane_bootup_value;
};
struct ci_clock_registers {
u32 cg_spll_func_cntl;
u32 cg_spll_func_cntl_2;
u32 cg_spll_func_cntl_3;
u32 cg_spll_func_cntl_4;
u32 cg_spll_spread_spectrum;
u32 cg_spll_spread_spectrum_2;
u32 dll_cntl;
u32 mclk_pwrmgt_cntl;
u32 mpll_ad_func_cntl;
u32 mpll_dq_func_cntl;
u32 mpll_func_cntl;
u32 mpll_func_cntl_1;
u32 mpll_func_cntl_2;
u32 mpll_ss1;
u32 mpll_ss2;
};
struct ci_thermal_temperature_setting {
s32 temperature_low;
s32 temperature_high;
s32 temperature_shutdown;
};
struct ci_pcie_perf_range {
u16 max;
u16 min;
};
enum ci_pt_config_reg_type {
CISLANDS_CONFIGREG_MMR = 0,
CISLANDS_CONFIGREG_SMC_IND,
CISLANDS_CONFIGREG_DIDT_IND,
CISLANDS_CONFIGREG_CACHE,
CISLANDS_CONFIGREG_MAX
};
#define POWERCONTAINMENT_FEATURE_BAPM 0x00000001
#define POWERCONTAINMENT_FEATURE_TDCLimit 0x00000002
#define POWERCONTAINMENT_FEATURE_PkgPwrLimit 0x00000004
struct ci_pt_config_reg {
u32 offset;
u32 mask;
u32 shift;
u32 value;
enum ci_pt_config_reg_type type;
};
struct ci_pt_defaults {
u8 svi_load_line_en;
u8 svi_load_line_vddc;
u8 tdc_vddc_throttle_release_limit_perc;
u8 tdc_mawt;
u8 tdc_waterfall_ctl;
u8 dte_ambient_temp_base;
u32 display_cac;
u32 bapm_temp_gradient;
u16 bapmti_r[SMU7_DTE_ITERATIONS * SMU7_DTE_SOURCES * SMU7_DTE_SINKS];
u16 bapmti_rc[SMU7_DTE_ITERATIONS * SMU7_DTE_SOURCES * SMU7_DTE_SINKS];
};
#define DPMTABLE_OD_UPDATE_SCLK 0x00000001
#define DPMTABLE_OD_UPDATE_MCLK 0x00000002
#define DPMTABLE_UPDATE_SCLK 0x00000004
#define DPMTABLE_UPDATE_MCLK 0x00000008
struct ci_power_info {
struct ci_dpm_table dpm_table;
struct ci_dpm_table golden_dpm_table;
u32 voltage_control;
u32 mvdd_control;
u32 vddci_control;
u32 active_auto_throttle_sources;
struct ci_clock_registers clock_registers;
u16 acpi_vddc;
u16 acpi_vddci;
enum amdgpu_pcie_gen force_pcie_gen;
enum amdgpu_pcie_gen acpi_pcie_gen;
struct ci_leakage_voltage vddc_leakage;
struct ci_leakage_voltage vddci_leakage;
u16 max_vddc_in_pp_table;
u16 min_vddc_in_pp_table;
u16 max_vddci_in_pp_table;
u16 min_vddci_in_pp_table;
u32 mclk_strobe_mode_threshold;
u32 mclk_stutter_mode_threshold;
u32 mclk_edc_enable_threshold;
u32 mclk_edc_wr_enable_threshold;
struct ci_vbios_boot_state vbios_boot_state;
/* smc offsets */
u32 sram_end;
u32 dpm_table_start;
u32 soft_regs_start;
u32 mc_reg_table_start;
u32 fan_table_start;
u32 arb_table_start;
/* smc tables */
SMU7_Discrete_DpmTable smc_state_table;
SMU7_Discrete_MCRegisters smc_mc_reg_table;
SMU7_Discrete_PmFuses smc_powertune_table;
/* other stuff */
struct ci_mc_reg_table mc_reg_table;
struct atom_voltage_table vddc_voltage_table;
struct atom_voltage_table vddci_voltage_table;
struct atom_voltage_table mvdd_voltage_table;
struct ci_ulv_parm ulv;
u32 power_containment_features;
const struct ci_pt_defaults *powertune_defaults;
u32 dte_tj_offset;
bool vddc_phase_shed_control;
struct ci_thermal_temperature_setting thermal_temp_setting;
struct ci_dpm_level_enable_mask dpm_level_enable_mask;
u32 need_update_smu7_dpm_table;
u32 sclk_dpm_key_disabled;
u32 mclk_dpm_key_disabled;
u32 pcie_dpm_key_disabled;
u32 thermal_sclk_dpm_enabled;
struct ci_pcie_perf_range pcie_gen_performance;
struct ci_pcie_perf_range pcie_lane_performance;
struct ci_pcie_perf_range pcie_gen_powersaving;
struct ci_pcie_perf_range pcie_lane_powersaving;
u32 activity_target[SMU7_MAX_LEVELS_GRAPHICS];
u32 mclk_activity_target;
u32 low_sclk_interrupt_t;
u32 last_mclk_dpm_enable_mask;
u32 sys_pcie_mask;
/* caps */
bool caps_power_containment;
bool caps_cac;
bool caps_sq_ramping;
bool caps_db_ramping;
bool caps_td_ramping;
bool caps_tcp_ramping;
bool caps_fps;
bool caps_sclk_ds;
bool caps_sclk_ss_support;
bool caps_mclk_ss_support;
bool caps_uvd_dpm;
bool caps_vce_dpm;
bool caps_samu_dpm;
bool caps_acp_dpm;
bool caps_automatic_dc_transition;
bool caps_sclk_throttle_low_notification;
bool caps_dynamic_ac_timing;
bool caps_od_fuzzy_fan_control_support;
/* flags */
bool thermal_protection;
bool pcie_performance_request;
bool dynamic_ss;
bool dll_default_on;
bool cac_enabled;
bool uvd_enabled;
bool battery_state;
bool pspp_notify_required;
bool enable_bapm_feature;
bool enable_tdc_limit_feature;
bool enable_pkg_pwr_tracking_feature;
bool use_pcie_performance_levels;
bool use_pcie_powersaving_levels;
bool uvd_power_gated;
/* driver states */
struct amdgpu_ps current_rps;
struct ci_ps current_ps;
struct amdgpu_ps requested_rps;
struct ci_ps requested_ps;
/* fan control */
bool fan_ctrl_is_in_default_mode;
bool fan_is_controlled_by_smc;
u32 t_min;
u32 fan_ctrl_default_mode;
};
#define CISLANDS_VOLTAGE_CONTROL_NONE 0x0
#define CISLANDS_VOLTAGE_CONTROL_BY_GPIO 0x1
#define CISLANDS_VOLTAGE_CONTROL_BY_SVID2 0x2
#define CISLANDS_Q88_FORMAT_CONVERSION_UNIT 256
#define CISLANDS_VRC_DFLT0 0x3FFFC000
#define CISLANDS_VRC_DFLT1 0x000400
#define CISLANDS_VRC_DFLT2 0xC00080
#define CISLANDS_VRC_DFLT3 0xC00200
#define CISLANDS_VRC_DFLT4 0xC01680
#define CISLANDS_VRC_DFLT5 0xC00033
#define CISLANDS_VRC_DFLT6 0xC00033
#define CISLANDS_VRC_DFLT7 0x3FFFC000
#define CISLANDS_CGULVPARAMETER_DFLT 0x00040035
#define CISLAND_TARGETACTIVITY_DFLT 30
#define CISLAND_MCLK_TARGETACTIVITY_DFLT 10
#define PCIE_PERF_REQ_REMOVE_REGISTRY 0
#define PCIE_PERF_REQ_FORCE_LOWPOWER 1
#define PCIE_PERF_REQ_PECI_GEN1 2
#define PCIE_PERF_REQ_PECI_GEN2 3
#define PCIE_PERF_REQ_PECI_GEN3 4
#define CISLANDS_SSTU_DFLT 0
#define CISLANDS_SST_DFLT 0x00C8
/* XXX are these ok? */
#define CISLANDS_TEMP_RANGE_MIN (90 * 1000)
#define CISLANDS_TEMP_RANGE_MAX (120 * 1000)
int amdgpu_ci_copy_bytes_to_smc(struct amdgpu_device *adev,
u32 smc_start_address,
const u8 *src, u32 byte_count, u32 limit);
void amdgpu_ci_start_smc(struct amdgpu_device *adev);
void amdgpu_ci_reset_smc(struct amdgpu_device *adev);
int amdgpu_ci_program_jump_on_start(struct amdgpu_device *adev);
void amdgpu_ci_stop_smc_clock(struct amdgpu_device *adev);
void amdgpu_ci_start_smc_clock(struct amdgpu_device *adev);
bool amdgpu_ci_is_smc_running(struct amdgpu_device *adev);
PPSMC_Result amdgpu_ci_send_msg_to_smc(struct amdgpu_device *adev, PPSMC_Msg msg);
PPSMC_Result amdgpu_ci_wait_for_smc_inactive(struct amdgpu_device *adev);
int amdgpu_ci_load_smc_ucode(struct amdgpu_device *adev, u32 limit);
int amdgpu_ci_read_smc_sram_dword(struct amdgpu_device *adev,
u32 smc_address, u32 *value, u32 limit);
int amdgpu_ci_write_smc_sram_dword(struct amdgpu_device *adev,
u32 smc_address, u32 value, u32 limit);
#endif

279
drivers/gpu/drm/amd/amdgpu/ci_smc.c
View File

@ -1,279 +0,0 @@
/*
* Copyright 2011 Advanced Micro Devices, Inc.
*
* Permission is hereby granted, free of charge, to any person obtaining a
* copy of this software and associated documentation files (the "Software"),
* to deal in the Software without restriction, including without limitation
* the rights to use, copy, modify, merge, publish, distribute, sublicense,
* and/or sell copies of the Software, and to permit persons to whom the
* Software is furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice shall be included in
* all copies or substantial portions of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
* THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
* OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
* ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
* OTHER DEALINGS IN THE SOFTWARE.
*
* Authors: Alex Deucher
*/
#include <linux/firmware.h>
#include <drm/drmP.h>
#include "amdgpu.h"
#include "cikd.h"
#include "ppsmc.h"
#include "amdgpu_ucode.h"
#include "ci_dpm.h"
#include "smu/smu_7_0_1_d.h"
#include "smu/smu_7_0_1_sh_mask.h"
static int ci_set_smc_sram_address(struct amdgpu_device *adev,
u32 smc_address, u32 limit)
{
if (smc_address & 3)
return -EINVAL;
if ((smc_address + 3) > limit)
return -EINVAL;
WREG32(mmSMC_IND_INDEX_0, smc_address);
WREG32_P(mmSMC_IND_ACCESS_CNTL, 0, ~SMC_IND_ACCESS_CNTL__AUTO_INCREMENT_IND_0_MASK);
return 0;
}
int amdgpu_ci_copy_bytes_to_smc(struct amdgpu_device *adev,
u32 smc_start_address,
const u8 *src, u32 byte_count, u32 limit)
{
unsigned long flags;
u32 data, original_data;
u32 addr;
u32 extra_shift;
int ret = 0;
if (smc_start_address & 3)
return -EINVAL;
if ((smc_start_address + byte_count) > limit)
return -EINVAL;
addr = smc_start_address;
spin_lock_irqsave(&adev->smc_idx_lock, flags);
while (byte_count >= 4) {
/* SMC address space is BE */
data = (src[0] << 24) | (src[1] << 16) | (src[2] << 8) | src[3];
ret = ci_set_smc_sram_address(adev, addr, limit);
if (ret)
goto done;
WREG32(mmSMC_IND_DATA_0, data);
src += 4;
byte_count -= 4;
addr += 4;
}
/* RMW for the final bytes */
if (byte_count > 0) {
data = 0;
ret = ci_set_smc_sram_address(adev, addr, limit);
if (ret)
goto done;
original_data = RREG32(mmSMC_IND_DATA_0);
extra_shift = 8 * (4 - byte_count);
while (byte_count > 0) {
data = (data << 8) + *src++;
byte_count--;
}
data <<= extra_shift;
data |= (original_data & ~((~0UL) << extra_shift));
ret = ci_set_smc_sram_address(adev, addr, limit);
if (ret)
goto done;
WREG32(mmSMC_IND_DATA_0, data);
}
done:
spin_unlock_irqrestore(&adev->smc_idx_lock, flags);
return ret;
}
void amdgpu_ci_start_smc(struct amdgpu_device *adev)
{
u32 tmp = RREG32_SMC(ixSMC_SYSCON_RESET_CNTL);
tmp &= ~SMC_SYSCON_RESET_CNTL__rst_reg_MASK;
WREG32_SMC(ixSMC_SYSCON_RESET_CNTL, tmp);
}
void amdgpu_ci_reset_smc(struct amdgpu_device *adev)
{
u32 tmp = RREG32_SMC(ixSMC_SYSCON_RESET_CNTL);
tmp |= SMC_SYSCON_RESET_CNTL__rst_reg_MASK;
WREG32_SMC(ixSMC_SYSCON_RESET_CNTL, tmp);
}
int amdgpu_ci_program_jump_on_start(struct amdgpu_device *adev)
{
static u8 data[] = { 0xE0, 0x00, 0x80, 0x40 };
return amdgpu_ci_copy_bytes_to_smc(adev, 0x0, data, 4, sizeof(data)+1);
}
void amdgpu_ci_stop_smc_clock(struct amdgpu_device *adev)
{
u32 tmp = RREG32_SMC(ixSMC_SYSCON_CLOCK_CNTL_0);
tmp |= SMC_SYSCON_CLOCK_CNTL_0__ck_disable_MASK;
WREG32_SMC(ixSMC_SYSCON_CLOCK_CNTL_0, tmp);
}
void amdgpu_ci_start_smc_clock(struct amdgpu_device *adev)
{
u32 tmp = RREG32_SMC(ixSMC_SYSCON_CLOCK_CNTL_0);
tmp &= ~SMC_SYSCON_CLOCK_CNTL_0__ck_disable_MASK;
WREG32_SMC(ixSMC_SYSCON_CLOCK_CNTL_0, tmp);
}
bool amdgpu_ci_is_smc_running(struct amdgpu_device *adev)
{
u32 clk = RREG32_SMC(ixSMC_SYSCON_CLOCK_CNTL_0);
u32 pc_c = RREG32_SMC(ixSMC_PC_C);
if (!(clk & SMC_SYSCON_CLOCK_CNTL_0__ck_disable_MASK) && (0x20100 <= pc_c))
return true;
return false;
}
PPSMC_Result amdgpu_ci_send_msg_to_smc(struct amdgpu_device *adev, PPSMC_Msg msg)
{
u32 tmp;
int i;
if (!amdgpu_ci_is_smc_running(adev))
return PPSMC_Result_Failed;
WREG32(mmSMC_MESSAGE_0, msg);
for (i = 0; i < adev->usec_timeout; i++) {
tmp = RREG32(mmSMC_RESP_0);
if (tmp != 0)
break;
udelay(1);
}
tmp = RREG32(mmSMC_RESP_0);
return (PPSMC_Result)tmp;
}
PPSMC_Result amdgpu_ci_wait_for_smc_inactive(struct amdgpu_device *adev)
{
u32 tmp;
int i;
if (!amdgpu_ci_is_smc_running(adev))
return PPSMC_Result_OK;
for (i = 0; i < adev->usec_timeout; i++) {
tmp = RREG32_SMC(ixSMC_SYSCON_CLOCK_CNTL_0);
if ((tmp & SMC_SYSCON_CLOCK_CNTL_0__cken_MASK) == 0)
break;
udelay(1);
}
return PPSMC_Result_OK;
}
int amdgpu_ci_load_smc_ucode(struct amdgpu_device *adev, u32 limit)
{
const struct smc_firmware_header_v1_0 *hdr;
unsigned long flags;
u32 ucode_start_address;
u32 ucode_size;
const u8 *src;
u32 data;
if (!adev->pm.fw)
return -EINVAL;
hdr = (const struct smc_firmware_header_v1_0 *)adev->pm.fw->data;
amdgpu_ucode_print_smc_hdr(&hdr->header);
adev->pm.fw_version = le32_to_cpu(hdr->header.ucode_version);
ucode_start_address = le32_to_cpu(hdr->ucode_start_addr);
ucode_size = le32_to_cpu(hdr->header.ucode_size_bytes);
src = (const u8 *)
(adev->pm.fw->data + le32_to_cpu(hdr->header.ucode_array_offset_bytes));
if (ucode_size & 3)
return -EINVAL;
spin_lock_irqsave(&adev->smc_idx_lock, flags);
WREG32(mmSMC_IND_INDEX_0, ucode_start_address);
WREG32_P(mmSMC_IND_ACCESS_CNTL, SMC_IND_ACCESS_CNTL__AUTO_INCREMENT_IND_0_MASK,
~SMC_IND_ACCESS_CNTL__AUTO_INCREMENT_IND_0_MASK);
while (ucode_size >= 4) {
/* SMC address space is BE */
data = (src[0] << 24) | (src[1] << 16) | (src[2] << 8) | src[3];
WREG32(mmSMC_IND_DATA_0, data);
src += 4;
ucode_size -= 4;
}
WREG32_P(mmSMC_IND_ACCESS_CNTL, 0, ~SMC_IND_ACCESS_CNTL__AUTO_INCREMENT_IND_0_MASK);
spin_unlock_irqrestore(&adev->smc_idx_lock, flags);
return 0;
}
int amdgpu_ci_read_smc_sram_dword(struct amdgpu_device *adev,
u32 smc_address, u32 *value, u32 limit)
{
unsigned long flags;
int ret;
spin_lock_irqsave(&adev->smc_idx_lock, flags);
ret = ci_set_smc_sram_address(adev, smc_address, limit);
if (ret == 0)
*value = RREG32(mmSMC_IND_DATA_0);
spin_unlock_irqrestore(&adev->smc_idx_lock, flags);
return ret;
}
int amdgpu_ci_write_smc_sram_dword(struct amdgpu_device *adev,
u32 smc_address, u32 value, u32 limit)
{
unsigned long flags;
int ret;
spin_lock_irqsave(&adev->smc_idx_lock, flags);
ret = ci_set_smc_sram_address(adev, smc_address, limit);
if (ret == 0)
WREG32(mmSMC_IND_DATA_0, value);
spin_unlock_irqrestore(&adev->smc_idx_lock, flags);
return ret;
}

75
drivers/gpu/drm/amd/amdgpu/cik.c
View File

@ -1741,6 +1741,69 @@ static bool cik_need_full_reset(struct amdgpu_device *adev)
return true;
}
static void cik_get_pcie_usage(struct amdgpu_device *adev, uint64_t *count0,
uint64_t *count1)
{
uint32_t perfctr = 0;
uint64_t cnt0_of, cnt1_of;
int tmp;
/* This reports 0 on APUs, so return to avoid writing/reading registers
* that may or may not be different from their GPU counterparts
*/
if (adev->flags & AMD_IS_APU)
return;
/* Set the 2 events that we wish to watch, defined above */
/* Reg 40 is # received msgs, Reg 104 is # of posted requests sent */
perfctr = REG_SET_FIELD(perfctr, PCIE_PERF_CNTL_TXCLK, EVENT0_SEL, 40);
perfctr = REG_SET_FIELD(perfctr, PCIE_PERF_CNTL_TXCLK, EVENT1_SEL, 104);
/* Write to enable desired perf counters */
WREG32_PCIE(ixPCIE_PERF_CNTL_TXCLK, perfctr);
/* Zero out and enable the perf counters
* Write 0x5:
* Bit 0 = Start all counters(1)
* Bit 2 = Global counter reset enable(1)
*/
WREG32_PCIE(ixPCIE_PERF_COUNT_CNTL, 0x00000005);
msleep(1000);
/* Load the shadow and disable the perf counters
* Write 0x2:
* Bit 0 = Stop counters(0)
* Bit 1 = Load the shadow counters(1)
*/
WREG32_PCIE(ixPCIE_PERF_COUNT_CNTL, 0x00000002);
/* Read register values to get any >32bit overflow */
tmp = RREG32_PCIE(ixPCIE_PERF_CNTL_TXCLK);
cnt0_of = REG_GET_FIELD(tmp, PCIE_PERF_CNTL_TXCLK, COUNTER0_UPPER);
cnt1_of = REG_GET_FIELD(tmp, PCIE_PERF_CNTL_TXCLK, COUNTER1_UPPER);
/* Get the values and add the overflow */
*count0 = RREG32_PCIE(ixPCIE_PERF_COUNT0_TXCLK) | (cnt0_of << 32);
*count1 = RREG32_PCIE(ixPCIE_PERF_COUNT1_TXCLK) | (cnt1_of << 32);
}
static bool cik_need_reset_on_init(struct amdgpu_device *adev)
{
u32 clock_cntl, pc;
if (adev->flags & AMD_IS_APU)
return false;
/* check if the SMC is already running */
clock_cntl = RREG32_SMC(ixSMC_SYSCON_CLOCK_CNTL_0);
pc = RREG32_SMC(ixSMC_PC_C);
if ((0 == REG_GET_FIELD(clock_cntl, SMC_SYSCON_CLOCK_CNTL_0, ck_disable)) &&
(0x20100 <= pc))
return true;
return false;
}
static const struct amdgpu_asic_funcs cik_asic_funcs =
{
.read_disabled_bios = &cik_read_disabled_bios,
@ -1756,6 +1819,8 @@ static const struct amdgpu_asic_funcs cik_asic_funcs =
.invalidate_hdp = &cik_invalidate_hdp,
.need_full_reset = &cik_need_full_reset,
.init_doorbell_index = &legacy_doorbell_index_init,
.get_pcie_usage = &cik_get_pcie_usage,
.need_reset_on_init = &cik_need_reset_on_init,
};
static int cik_common_early_init(void *handle)
@ -2005,10 +2070,7 @@ int cik_set_ip_blocks(struct amdgpu_device *adev)
amdgpu_device_ip_block_add(adev, &cik_ih_ip_block);
amdgpu_device_ip_block_add(adev, &gfx_v7_2_ip_block);
amdgpu_device_ip_block_add(adev, &cik_sdma_ip_block);
if (amdgpu_dpm == -1)
amdgpu_device_ip_block_add(adev, &pp_smu_ip_block);
else
amdgpu_device_ip_block_add(adev, &ci_smu_ip_block);
amdgpu_device_ip_block_add(adev, &pp_smu_ip_block);
if (adev->enable_virtual_display)
amdgpu_device_ip_block_add(adev, &dce_virtual_ip_block);
#if defined(CONFIG_DRM_AMD_DC)
@ -2026,10 +2088,7 @@ int cik_set_ip_blocks(struct amdgpu_device *adev)
amdgpu_device_ip_block_add(adev, &cik_ih_ip_block);
amdgpu_device_ip_block_add(adev, &gfx_v7_3_ip_block);
amdgpu_device_ip_block_add(adev, &cik_sdma_ip_block);
if (amdgpu_dpm == -1)
amdgpu_device_ip_block_add(adev, &pp_smu_ip_block);
else
amdgpu_device_ip_block_add(adev, &ci_smu_ip_block);
amdgpu_device_ip_block_add(adev, &pp_smu_ip_block);
if (adev->enable_virtual_display)
amdgpu_device_ip_block_add(adev, &dce_virtual_ip_block);
#if defined(CONFIG_DRM_AMD_DC)

1
drivers/gpu/drm/amd/amdgpu/cik_dpm.h
View File

@ -24,7 +24,6 @@
#ifndef __CIK_DPM_H__
#define __CIK_DPM_H__
extern const struct amdgpu_ip_block_version ci_smu_ip_block;
extern const struct amdgpu_ip_block_version kv_smu_ip_block;
#endif

36
drivers/gpu/drm/amd/amdgpu/cik_ih.c
View File

@ -103,9 +103,9 @@ static void cik_ih_disable_interrupts(struct amdgpu_device *adev)
*/
static int cik_ih_irq_init(struct amdgpu_device *adev)
{
struct amdgpu_ih_ring *ih = &adev->irq.ih;
int rb_bufsz;
u32 interrupt_cntl, ih_cntl, ih_rb_cntl;
u64 wptr_off;
/* disable irqs */
cik_ih_disable_interrupts(adev);
@ -131,9 +131,8 @@ static int cik_ih_irq_init(struct amdgpu_device *adev)
ih_rb_cntl |= IH_RB_CNTL__WPTR_WRITEBACK_ENABLE_MASK;
/* set the writeback address whether it's enabled or not */
wptr_off = adev->wb.gpu_addr + (adev->irq.ih.wptr_offs * 4);
WREG32(mmIH_RB_WPTR_ADDR_LO, lower_32_bits(wptr_off));
WREG32(mmIH_RB_WPTR_ADDR_HI, upper_32_bits(wptr_off) & 0xFF);
WREG32(mmIH_RB_WPTR_ADDR_LO, lower_32_bits(ih->wptr_addr));
WREG32(mmIH_RB_WPTR_ADDR_HI, upper_32_bits(ih->wptr_addr) & 0xFF);
WREG32(mmIH_RB_CNTL, ih_rb_cntl);
@ -183,11 +182,12 @@ static void cik_ih_irq_disable(struct amdgpu_device *adev)
* Used by cik_irq_process().
* Returns the value of the wptr.
*/
static u32 cik_ih_get_wptr(struct amdgpu_device *adev)
static u32 cik_ih_get_wptr(struct amdgpu_device *adev,
struct amdgpu_ih_ring *ih)
{
u32 wptr, tmp;
wptr = le32_to_cpu(adev->wb.wb[adev->irq.ih.wptr_offs]);
wptr = le32_to_cpu(*ih->wptr_cpu);
if (wptr & IH_RB_WPTR__RB_OVERFLOW_MASK) {
wptr &= ~IH_RB_WPTR__RB_OVERFLOW_MASK;
@ -196,13 +196,13 @@ static u32 cik_ih_get_wptr(struct amdgpu_device *adev)
* this should allow us to catchup.
*/
dev_warn(adev->dev, "IH ring buffer overflow (0x%08X, 0x%08X, 0x%08X)\n",
wptr, adev->irq.ih.rptr, (wptr + 16) & adev->irq.ih.ptr_mask);
adev->irq.ih.rptr = (wptr + 16) & adev->irq.ih.ptr_mask;
wptr, ih->rptr, (wptr + 16) & ih->ptr_mask);
ih->rptr = (wptr + 16) & ih->ptr_mask;
tmp = RREG32(mmIH_RB_CNTL);
tmp |= IH_RB_CNTL__WPTR_OVERFLOW_CLEAR_MASK;
WREG32(mmIH_RB_CNTL, tmp);
}
return (wptr & adev->irq.ih.ptr_mask);
return (wptr & ih->ptr_mask);
}
/* CIK IV Ring
@ -237,16 +237,17 @@ static u32 cik_ih_get_wptr(struct amdgpu_device *adev)
* position and also advance the position.
*/
static void cik_ih_decode_iv(struct amdgpu_device *adev,
struct amdgpu_ih_ring *ih,
struct amdgpu_iv_entry *entry)
{
/* wptr/rptr are in bytes! */
u32 ring_index = adev->irq.ih.rptr >> 2;
u32 ring_index = ih->rptr >> 2;
uint32_t dw[4];
dw[0] = le32_to_cpu(adev->irq.ih.ring[ring_index + 0]);
dw[1] = le32_to_cpu(adev->irq.ih.ring[ring_index + 1]);
dw[2] = le32_to_cpu(adev->irq.ih.ring[ring_index + 2]);
dw[3] = le32_to_cpu(adev->irq.ih.ring[ring_index + 3]);
dw[0] = le32_to_cpu(ih->ring[ring_index + 0]);
dw[1] = le32_to_cpu(ih->ring[ring_index + 1]);
dw[2] = le32_to_cpu(ih->ring[ring_index + 2]);
dw[3] = le32_to_cpu(ih->ring[ring_index + 3]);
entry->client_id = AMDGPU_IRQ_CLIENTID_LEGACY;
entry->src_id = dw[0] & 0xff;
@ -256,7 +257,7 @@ static void cik_ih_decode_iv(struct amdgpu_device *adev,
entry->pasid = (dw[2] >> 16) & 0xffff;
/* wptr/rptr are in bytes! */
adev->irq.ih.rptr += 16;
ih->rptr += 16;
}
/**
@ -266,9 +267,10 @@ static void cik_ih_decode_iv(struct amdgpu_device *adev,
*
* Set the IH ring buffer rptr.
*/
static void cik_ih_set_rptr(struct amdgpu_device *adev)
static void cik_ih_set_rptr(struct amdgpu_device *adev,
struct amdgpu_ih_ring *ih)
{
WREG32(mmIH_RB_RPTR, adev->irq.ih.rptr);
WREG32(mmIH_RB_RPTR, ih->rptr);
}
static int cik_ih_early_init(void *handle)

2
drivers/gpu/drm/amd/amdgpu/cik_sdma.c
View File

@ -220,7 +220,7 @@ static void cik_sdma_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count)
static void cik_sdma_ring_emit_ib(struct amdgpu_ring *ring,
struct amdgpu_job *job,
struct amdgpu_ib *ib,
bool ctx_switch)
uint32_t flags)
{
unsigned vmid = AMDGPU_JOB_GET_VMID(job);
u32 extra_bits = vmid & 0xf;

40
drivers/gpu/drm/amd/amdgpu/cz_ih.c
View File

@ -103,9 +103,9 @@ static void cz_ih_disable_interrupts(struct amdgpu_device *adev)
*/
static int cz_ih_irq_init(struct amdgpu_device *adev)
{
int rb_bufsz;
struct amdgpu_ih_ring *ih = &adev->irq.ih;
u32 interrupt_cntl, ih_cntl, ih_rb_cntl;
u64 wptr_off;
int rb_bufsz;
/* disable irqs */
cz_ih_disable_interrupts(adev);
@ -133,9 +133,8 @@ static int cz_ih_irq_init(struct amdgpu_device *adev)
ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, WPTR_WRITEBACK_ENABLE, 1);
/* set the writeback address whether it's enabled or not */
wptr_off = adev->wb.gpu_addr + (adev->irq.ih.wptr_offs * 4);
WREG32(mmIH_RB_WPTR_ADDR_LO, lower_32_bits(wptr_off));
WREG32(mmIH_RB_WPTR_ADDR_HI, upper_32_bits(wptr_off) & 0xFF);
WREG32(mmIH_RB_WPTR_ADDR_LO, lower_32_bits(ih->wptr_addr));
WREG32(mmIH_RB_WPTR_ADDR_HI, upper_32_bits(ih->wptr_addr) & 0xFF);
WREG32(mmIH_RB_CNTL, ih_rb_cntl);
@ -185,11 +184,12 @@ static void cz_ih_irq_disable(struct amdgpu_device *adev)
* Used by cz_irq_process(VI).
* Returns the value of the wptr.
*/
static u32 cz_ih_get_wptr(struct amdgpu_device *adev)
static u32 cz_ih_get_wptr(struct amdgpu_device *adev,
struct amdgpu_ih_ring *ih)
{
u32 wptr, tmp;
wptr = le32_to_cpu(adev->wb.wb[adev->irq.ih.wptr_offs]);
wptr = le32_to_cpu(*ih->wptr_cpu);
if (REG_GET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW)) {
wptr = REG_SET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW, 0);
@ -198,13 +198,13 @@ static u32 cz_ih_get_wptr(struct amdgpu_device *adev)
* this should allow us to catchup.
*/
dev_warn(adev->dev, "IH ring buffer overflow (0x%08X, 0x%08X, 0x%08X)\n",
wptr, adev->irq.ih.rptr, (wptr + 16) & adev->irq.ih.ptr_mask);
adev->irq.ih.rptr = (wptr + 16) & adev->irq.ih.ptr_mask;
wptr, ih->rptr, (wptr + 16) & ih->ptr_mask);
ih->rptr = (wptr + 16) & ih->ptr_mask;
tmp = RREG32(mmIH_RB_CNTL);
tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, WPTR_OVERFLOW_CLEAR, 1);
WREG32(mmIH_RB_CNTL, tmp);
}
return (wptr & adev->irq.ih.ptr_mask);
return (wptr & ih->ptr_mask);
}
/**
@ -216,16 +216,17 @@ static u32 cz_ih_get_wptr(struct amdgpu_device *adev)
* position and also advance the position.
*/
static void cz_ih_decode_iv(struct amdgpu_device *adev,
struct amdgpu_iv_entry *entry)
struct amdgpu_ih_ring *ih,
struct amdgpu_iv_entry *entry)
{
/* wptr/rptr are in bytes! */
u32 ring_index = adev->irq.ih.rptr >> 2;
u32 ring_index = ih->rptr >> 2;
uint32_t dw[4];
dw[0] = le32_to_cpu(adev->irq.ih.ring[ring_index + 0]);
dw[1] = le32_to_cpu(adev->irq.ih.ring[ring_index + 1]);
dw[2] = le32_to_cpu(adev->irq.ih.ring[ring_index + 2]);
dw[3] = le32_to_cpu(adev->irq.ih.ring[ring_index + 3]);
dw[0] = le32_to_cpu(ih->ring[ring_index + 0]);
dw[1] = le32_to_cpu(ih->ring[ring_index + 1]);
dw[2] = le32_to_cpu(ih->ring[ring_index + 2]);
dw[3] = le32_to_cpu(ih->ring[ring_index + 3]);
entry->client_id = AMDGPU_IRQ_CLIENTID_LEGACY;
entry->src_id = dw[0] & 0xff;
@ -235,7 +236,7 @@ static void cz_ih_decode_iv(struct amdgpu_device *adev,
entry->pasid = (dw[2] >> 16) & 0xffff;
/* wptr/rptr are in bytes! */
adev->irq.ih.rptr += 16;
ih->rptr += 16;
}
/**
@ -245,9 +246,10 @@ static void cz_ih_decode_iv(struct amdgpu_device *adev,
*
* Set the IH ring buffer rptr.
*/
static void cz_ih_set_rptr(struct amdgpu_device *adev)
static void cz_ih_set_rptr(struct amdgpu_device *adev,
struct amdgpu_ih_ring *ih)
{
WREG32(mmIH_RB_RPTR, adev->irq.ih.rptr);
WREG32(mmIH_RB_RPTR, ih->rptr);
}
static int cz_ih_early_init(void *handle)

17
drivers/gpu/drm/amd/amdgpu/dce_virtual.c
View File

@ -167,19 +167,6 @@ static void dce_virtual_crtc_disable(struct drm_crtc *crtc)
struct amdgpu_crtc *amdgpu_crtc = to_amdgpu_crtc(crtc);
dce_virtual_crtc_dpms(crtc, DRM_MODE_DPMS_OFF);
if (crtc->primary->fb) {
int r;
struct amdgpu_bo *abo;
abo = gem_to_amdgpu_bo(crtc->primary->fb->obj[0]);
r = amdgpu_bo_reserve(abo, true);
if (unlikely(r))
DRM_ERROR("failed to reserve abo before unpin\n");
else {
amdgpu_bo_unpin(abo);
amdgpu_bo_unreserve(abo);
}
}
amdgpu_crtc->pll_id = ATOM_PPLL_INVALID;
amdgpu_crtc->encoder = NULL;
@ -692,7 +679,9 @@ static int dce_virtual_pageflip(struct amdgpu_device *adev,
spin_unlock_irqrestore(&adev->ddev->event_lock, flags);
drm_crtc_vblank_put(&amdgpu_crtc->base);
schedule_work(&works->unpin_work);
amdgpu_bo_unref(&works->old_abo);
kfree(works->shared);
kfree(works);
return 0;
}

4
drivers/gpu/drm/amd/amdgpu/gfx_v6_0.c
View File

@ -1842,13 +1842,13 @@ static void gfx_v6_0_ring_emit_fence(struct amdgpu_ring *ring, u64 addr,
static void gfx_v6_0_ring_emit_ib(struct amdgpu_ring *ring,
struct amdgpu_job *job,
struct amdgpu_ib *ib,
bool ctx_switch)
uint32_t flags)
{
unsigned vmid = AMDGPU_JOB_GET_VMID(job);
u32 header, control = 0;
/* insert SWITCH_BUFFER packet before first IB in the ring frame */
if (ctx_switch) {
if (flags & AMDGPU_HAVE_CTX_SWITCH) {
amdgpu_ring_write(ring, PACKET3(PACKET3_SWITCH_BUFFER, 0));
amdgpu_ring_write(ring, 0);
}

25
drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c
View File

@ -2228,13 +2228,13 @@ static void gfx_v7_0_ring_emit_fence_compute(struct amdgpu_ring *ring,
static void gfx_v7_0_ring_emit_ib_gfx(struct amdgpu_ring *ring,
struct amdgpu_job *job,
struct amdgpu_ib *ib,
bool ctx_switch)
uint32_t flags)
{
unsigned vmid = AMDGPU_JOB_GET_VMID(job);
u32 header, control = 0;
/* insert SWITCH_BUFFER packet before first IB in the ring frame */
if (ctx_switch) {
if (flags & AMDGPU_HAVE_CTX_SWITCH) {
amdgpu_ring_write(ring, PACKET3(PACKET3_SWITCH_BUFFER, 0));
amdgpu_ring_write(ring, 0);
}
@ -2259,11 +2259,27 @@ static void gfx_v7_0_ring_emit_ib_gfx(struct amdgpu_ring *ring,
static void gfx_v7_0_ring_emit_ib_compute(struct amdgpu_ring *ring,
struct amdgpu_job *job,
struct amdgpu_ib *ib,
bool ctx_switch)
uint32_t flags)
{
unsigned vmid = AMDGPU_JOB_GET_VMID(job);
u32 control = INDIRECT_BUFFER_VALID | ib->length_dw | (vmid << 24);
/* Currently, there is a high possibility to get wave ID mismatch
* between ME and GDS, leading to a hw deadlock, because ME generates
* different wave IDs than the GDS expects. This situation happens
* randomly when at least 5 compute pipes use GDS ordered append.
* The wave IDs generated by ME are also wrong after suspend/resume.
* Those are probably bugs somewhere else in the kernel driver.
*
* Writing GDS_COMPUTE_MAX_WAVE_ID resets wave ID counters in ME and
* GDS to 0 for this ring (me/pipe).
*/
if (ib->flags & AMDGPU_IB_FLAG_RESET_GDS_MAX_WAVE_ID) {
amdgpu_ring_write(ring, PACKET3(PACKET3_SET_CONFIG_REG, 1));
amdgpu_ring_write(ring, mmGDS_COMPUTE_MAX_WAVE_ID - PACKET3_SET_CONFIG_REG_START);
amdgpu_ring_write(ring, ring->adev->gds.gds_compute_max_wave_id);
}
amdgpu_ring_write(ring, PACKET3(PACKET3_INDIRECT_BUFFER, 2));
amdgpu_ring_write(ring,
#ifdef __BIG_ENDIAN
@ -5000,7 +5016,7 @@ static const struct amdgpu_ring_funcs gfx_v7_0_ring_funcs_compute = {
7 + /* gfx_v7_0_ring_emit_pipeline_sync */
CIK_FLUSH_GPU_TLB_NUM_WREG * 5 + 7 + /* gfx_v7_0_ring_emit_vm_flush */
7 + 7 + 7, /* gfx_v7_0_ring_emit_fence_compute x3 for user fence, vm fence */
.emit_ib_size = 4, /* gfx_v7_0_ring_emit_ib_compute */
.emit_ib_size = 7, /* gfx_v7_0_ring_emit_ib_compute */
.emit_ib = gfx_v7_0_ring_emit_ib_compute,
.emit_fence = gfx_v7_0_ring_emit_fence_compute,
.emit_pipeline_sync = gfx_v7_0_ring_emit_pipeline_sync,
@ -5057,6 +5073,7 @@ static void gfx_v7_0_set_gds_init(struct amdgpu_device *adev)
adev->gds.mem.total_size = RREG32(mmGDS_VMID0_SIZE);
adev->gds.gws.total_size = 64;
adev->gds.oa.total_size = 16;
adev->gds.gds_compute_max_wave_id = RREG32(mmGDS_COMPUTE_MAX_WAVE_ID);
if (adev->gds.mem.total_size == 64 * 1024) {
adev->gds.mem.gfx_partition_size = 4096;

75
drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
View File

@ -4233,7 +4233,6 @@ static int gfx_v8_0_cp_gfx_resume(struct amdgpu_device *adev)
u32 tmp;
u32 rb_bufsz;
u64 rb_addr, rptr_addr, wptr_gpu_addr;
int r;
/* Set the write pointer delay */
WREG32(mmCP_RB_WPTR_DELAY, 0);
@ -4278,9 +4277,8 @@ static int gfx_v8_0_cp_gfx_resume(struct amdgpu_device *adev)
amdgpu_ring_clear_ring(ring);
gfx_v8_0_cp_gfx_start(adev);
ring->sched.ready = true;
r = amdgpu_ring_test_helper(ring);
return r;
return 0;
}
static void gfx_v8_0_cp_compute_enable(struct amdgpu_device *adev, bool enable)
@ -4369,10 +4367,9 @@ static int gfx_v8_0_kiq_kcq_enable(struct amdgpu_device *adev)
amdgpu_ring_write(kiq_ring, upper_32_bits(wptr_addr));
}
r = amdgpu_ring_test_helper(kiq_ring);
if (r)
DRM_ERROR("KCQ enable failed\n");
return r;
amdgpu_ring_commit(kiq_ring);
return 0;
}
static int gfx_v8_0_deactivate_hqd(struct amdgpu_device *adev, u32 req)
@ -4709,18 +4706,34 @@ static int gfx_v8_0_kcq_resume(struct amdgpu_device *adev)
if (r)
goto done;
/* Test KCQs - reversing the order of rings seems to fix ring test failure
* after GPU reset
*/
for (i = adev->gfx.num_compute_rings - 1; i >= 0; i--) {
ring = &adev->gfx.compute_ring[i];
r = amdgpu_ring_test_helper(ring);
}
done:
return r;
}
static int gfx_v8_0_cp_test_all_rings(struct amdgpu_device *adev)
{
int r, i;
struct amdgpu_ring *ring;
/* collect all the ring_tests here, gfx, kiq, compute */
ring = &adev->gfx.gfx_ring[0];
r = amdgpu_ring_test_helper(ring);
if (r)
return r;
ring = &adev->gfx.kiq.ring;
r = amdgpu_ring_test_helper(ring);
if (r)
return r;
for (i = 0; i < adev->gfx.num_compute_rings; i++) {
ring = &adev->gfx.compute_ring[i];
amdgpu_ring_test_helper(ring);
}
return 0;
}
static int gfx_v8_0_cp_resume(struct amdgpu_device *adev)
{
int r;
@ -4739,6 +4752,11 @@ static int gfx_v8_0_cp_resume(struct amdgpu_device *adev)
r = gfx_v8_0_kcq_resume(adev);
if (r)
return r;
r = gfx_v8_0_cp_test_all_rings(adev);
if (r)
return r;
gfx_v8_0_enable_gui_idle_interrupt(adev, true);
return 0;
@ -5086,6 +5104,8 @@ static int gfx_v8_0_post_soft_reset(void *handle)
REG_GET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET, SOFT_RESET_GFX))
gfx_v8_0_cp_gfx_resume(adev);
gfx_v8_0_cp_test_all_rings(adev);
adev->gfx.rlc.funcs->start(adev);
return 0;
@ -6027,7 +6047,7 @@ static void gfx_v8_0_ring_emit_vgt_flush(struct amdgpu_ring *ring)
static void gfx_v8_0_ring_emit_ib_gfx(struct amdgpu_ring *ring,
struct amdgpu_job *job,
struct amdgpu_ib *ib,
bool ctx_switch)
uint32_t flags)
{
unsigned vmid = AMDGPU_JOB_GET_VMID(job);
u32 header, control = 0;
@ -6059,11 +6079,27 @@ static void gfx_v8_0_ring_emit_ib_gfx(struct amdgpu_ring *ring,
static void gfx_v8_0_ring_emit_ib_compute(struct amdgpu_ring *ring,
struct amdgpu_job *job,
struct amdgpu_ib *ib,
bool ctx_switch)
uint32_t flags)
{
unsigned vmid = AMDGPU_JOB_GET_VMID(job);
u32 control = INDIRECT_BUFFER_VALID | ib->length_dw | (vmid << 24);
/* Currently, there is a high possibility to get wave ID mismatch
* between ME and GDS, leading to a hw deadlock, because ME generates
* different wave IDs than the GDS expects. This situation happens
* randomly when at least 5 compute pipes use GDS ordered append.
* The wave IDs generated by ME are also wrong after suspend/resume.
* Those are probably bugs somewhere else in the kernel driver.
*
* Writing GDS_COMPUTE_MAX_WAVE_ID resets wave ID counters in ME and
* GDS to 0 for this ring (me/pipe).
*/
if (ib->flags & AMDGPU_IB_FLAG_RESET_GDS_MAX_WAVE_ID) {
amdgpu_ring_write(ring, PACKET3(PACKET3_SET_CONFIG_REG, 1));
amdgpu_ring_write(ring, mmGDS_COMPUTE_MAX_WAVE_ID - PACKET3_SET_CONFIG_REG_START);
amdgpu_ring_write(ring, ring->adev->gds.gds_compute_max_wave_id);
}
amdgpu_ring_write(ring, PACKET3(PACKET3_INDIRECT_BUFFER, 2));
amdgpu_ring_write(ring,
#ifdef __BIG_ENDIAN
@ -6870,7 +6906,7 @@ static const struct amdgpu_ring_funcs gfx_v8_0_ring_funcs_compute = {
7 + /* gfx_v8_0_ring_emit_pipeline_sync */
VI_FLUSH_GPU_TLB_NUM_WREG * 5 + 7 + /* gfx_v8_0_ring_emit_vm_flush */
7 + 7 + 7, /* gfx_v8_0_ring_emit_fence_compute x3 for user fence, vm fence */
.emit_ib_size = 4, /* gfx_v8_0_ring_emit_ib_compute */
.emit_ib_size = 7, /* gfx_v8_0_ring_emit_ib_compute */
.emit_ib = gfx_v8_0_ring_emit_ib_compute,
.emit_fence = gfx_v8_0_ring_emit_fence_compute,
.emit_pipeline_sync = gfx_v8_0_ring_emit_pipeline_sync,
@ -6900,7 +6936,7 @@ static const struct amdgpu_ring_funcs gfx_v8_0_ring_funcs_kiq = {
7 + /* gfx_v8_0_ring_emit_pipeline_sync */
17 + /* gfx_v8_0_ring_emit_vm_flush */
7 + 7 + 7, /* gfx_v8_0_ring_emit_fence_kiq x3 for user fence, vm fence */
.emit_ib_size = 4, /* gfx_v8_0_ring_emit_ib_compute */
.emit_ib_size = 7, /* gfx_v8_0_ring_emit_ib_compute */
.emit_fence = gfx_v8_0_ring_emit_fence_kiq,
.test_ring = gfx_v8_0_ring_test_ring,
.insert_nop = amdgpu_ring_insert_nop,
@ -6976,6 +7012,7 @@ static void gfx_v8_0_set_gds_init(struct amdgpu_device *adev)
adev->gds.mem.total_size = RREG32(mmGDS_VMID0_SIZE);
adev->gds.gws.total_size = 64;
adev->gds.oa.total_size = 16;
adev->gds.gds_compute_max_wave_id = RREG32(mmGDS_COMPUTE_MAX_WAVE_ID);
if (adev->gds.mem.total_size == 64 * 1024) {
adev->gds.mem.gfx_partition_size = 4096;

58
drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
View File

@ -113,7 +113,10 @@ static const struct soc15_reg_golden golden_settings_gc_9_0[] =
SOC15_REG_GOLDEN_VALUE(GC, 0, mmTCP_CHAN_STEER_HI, 0xffffffff, 0x4a2c0e68),
SOC15_REG_GOLDEN_VALUE(GC, 0, mmTCP_CHAN_STEER_LO, 0xffffffff, 0xb5d3f197),
SOC15_REG_GOLDEN_VALUE(GC, 0, mmVGT_CACHE_INVALIDATION, 0x3fff3af3, 0x19200000),
SOC15_REG_GOLDEN_VALUE(GC, 0, mmVGT_GS_MAX_WAVE_ID, 0x00000fff, 0x000003ff)
SOC15_REG_GOLDEN_VALUE(GC, 0, mmVGT_GS_MAX_WAVE_ID, 0x00000fff, 0x000003ff),
SOC15_REG_GOLDEN_VALUE(GC, 0, mmCP_MEC1_F32_INT_DIS, 0x00000000, 0x00000800),
SOC15_REG_GOLDEN_VALUE(GC, 0, mmCP_MEC2_F32_INT_DIS, 0x00000000, 0x00000800),
SOC15_REG_GOLDEN_VALUE(GC, 0, mmCP_DEBUG, 0x00000000, 0x00008000)
};
static const struct soc15_reg_golden golden_settings_gc_9_0_vg10[] =
@ -135,10 +138,7 @@ static const struct soc15_reg_golden golden_settings_gc_9_0_vg10[] =
SOC15_REG_GOLDEN_VALUE(GC, 0, mmRMI_UTCL1_CNTL2, 0x00030000, 0x00020000),
SOC15_REG_GOLDEN_VALUE(GC, 0, mmSPI_CONFIG_CNTL_1, 0x0000000f, 0x01000107),
SOC15_REG_GOLDEN_VALUE(GC, 0, mmTD_CNTL, 0x00001800, 0x00000800),
SOC15_REG_GOLDEN_VALUE(GC, 0, mmWD_UTCL1_CNTL, 0x08000000, 0x08000080),
SOC15_REG_GOLDEN_VALUE(GC, 0, mmCP_MEC1_F32_INT_DIS, 0x00000000, 0x00000800),
SOC15_REG_GOLDEN_VALUE(GC, 0, mmCP_MEC2_F32_INT_DIS, 0x00000000, 0x00000800),
SOC15_REG_GOLDEN_VALUE(GC, 0, mmCP_DEBUG, 0x00000000, 0x00008000)
SOC15_REG_GOLDEN_VALUE(GC, 0, mmWD_UTCL1_CNTL, 0x08000000, 0x08000080)
};
static const struct soc15_reg_golden golden_settings_gc_9_0_vg20[] =
@ -3587,6 +3587,8 @@ static void gfx_v9_0_update_medium_grain_clock_gating(struct amdgpu_device *adev
{
uint32_t data, def;
amdgpu_gfx_rlc_enter_safe_mode(adev);
/* It is disabled by HW by default */
if (enable && (adev->cg_flags & AMD_CG_SUPPORT_GFX_MGCG)) {
/* 1 - RLC_CGTT_MGCG_OVERRIDE */
@ -3651,6 +3653,8 @@ static void gfx_v9_0_update_medium_grain_clock_gating(struct amdgpu_device *adev
WREG32_SOC15(GC, 0, mmCP_MEM_SLP_CNTL, data);
}
}
amdgpu_gfx_rlc_exit_safe_mode(adev);
}
static void gfx_v9_0_update_3d_clock_gating(struct amdgpu_device *adev,
@ -3968,7 +3972,7 @@ static void gfx_v9_0_ring_emit_hdp_flush(struct amdgpu_ring *ring)
static void gfx_v9_0_ring_emit_ib_gfx(struct amdgpu_ring *ring,
struct amdgpu_job *job,
struct amdgpu_ib *ib,
bool ctx_switch)
uint32_t flags)
{
unsigned vmid = AMDGPU_JOB_GET_VMID(job);
u32 header, control = 0;
@ -4001,11 +4005,27 @@ static void gfx_v9_0_ring_emit_ib_gfx(struct amdgpu_ring *ring,
static void gfx_v9_0_ring_emit_ib_compute(struct amdgpu_ring *ring,
struct amdgpu_job *job,
struct amdgpu_ib *ib,
bool ctx_switch)
uint32_t flags)
{
unsigned vmid = AMDGPU_JOB_GET_VMID(job);
u32 control = INDIRECT_BUFFER_VALID | ib->length_dw | (vmid << 24);
/* Currently, there is a high possibility to get wave ID mismatch
* between ME and GDS, leading to a hw deadlock, because ME generates
* different wave IDs than the GDS expects. This situation happens
* randomly when at least 5 compute pipes use GDS ordered append.
* The wave IDs generated by ME are also wrong after suspend/resume.
* Those are probably bugs somewhere else in the kernel driver.
*
* Writing GDS_COMPUTE_MAX_WAVE_ID resets wave ID counters in ME and
* GDS to 0 for this ring (me/pipe).
*/
if (ib->flags & AMDGPU_IB_FLAG_RESET_GDS_MAX_WAVE_ID) {
amdgpu_ring_write(ring, PACKET3(PACKET3_SET_CONFIG_REG, 1));
amdgpu_ring_write(ring, mmGDS_COMPUTE_MAX_WAVE_ID);
amdgpu_ring_write(ring, ring->adev->gds.gds_compute_max_wave_id);
}
amdgpu_ring_write(ring, PACKET3(PACKET3_INDIRECT_BUFFER, 2));
BUG_ON(ib->gpu_addr & 0x3); /* Dword align */
amdgpu_ring_write(ring,
@ -4725,7 +4745,7 @@ static const struct amdgpu_ring_funcs gfx_v9_0_ring_funcs_compute = {
SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 7 +
2 + /* gfx_v9_0_ring_emit_vm_flush */
8 + 8 + 8, /* gfx_v9_0_ring_emit_fence x3 for user fence, vm fence */
.emit_ib_size = 4, /* gfx_v9_0_ring_emit_ib_compute */
.emit_ib_size = 7, /* gfx_v9_0_ring_emit_ib_compute */
.emit_ib = gfx_v9_0_ring_emit_ib_compute,
.emit_fence = gfx_v9_0_ring_emit_fence,
.emit_pipeline_sync = gfx_v9_0_ring_emit_pipeline_sync,
@ -4760,7 +4780,7 @@ static const struct amdgpu_ring_funcs gfx_v9_0_ring_funcs_kiq = {
SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 7 +
2 + /* gfx_v9_0_ring_emit_vm_flush */
8 + 8 + 8, /* gfx_v9_0_ring_emit_fence_kiq x3 for user fence, vm fence */
.emit_ib_size = 4, /* gfx_v9_0_ring_emit_ib_compute */
.emit_ib_size = 7, /* gfx_v9_0_ring_emit_ib_compute */
.emit_fence = gfx_v9_0_ring_emit_fence_kiq,
.test_ring = gfx_v9_0_ring_test_ring,
.insert_nop = amdgpu_ring_insert_nop,
@ -4842,6 +4862,26 @@ static void gfx_v9_0_set_gds_init(struct amdgpu_device *adev)
break;
}
switch (adev->asic_type) {
case CHIP_VEGA10:
case CHIP_VEGA20:
adev->gds.gds_compute_max_wave_id = 0x7ff;
break;
case CHIP_VEGA12:
adev->gds.gds_compute_max_wave_id = 0x27f;
break;
case CHIP_RAVEN:
if (adev->rev_id >= 0x8)
adev->gds.gds_compute_max_wave_id = 0x77; /* raven2 */
else
adev->gds.gds_compute_max_wave_id = 0x15f; /* raven1 */
break;
default:
/* this really depends on the chip */
adev->gds.gds_compute_max_wave_id = 0x7ff;
break;
}
adev->gds.gws.total_size = 64;
adev->gds.oa.total_size = 16;

3
drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c
View File

@ -1471,8 +1471,9 @@ static int gmc_v8_0_process_interrupt(struct amdgpu_device *adev,
gmc_v8_0_set_fault_enable_default(adev, false);
if (printk_ratelimit()) {
struct amdgpu_task_info task_info = { 0 };
struct amdgpu_task_info task_info;
memset(&task_info, 0, sizeof(struct amdgpu_task_info));
amdgpu_vm_get_task_info(adev, entry->pasid, &task_info);
dev_err(adev->dev, "GPU fault detected: %d 0x%08x for process %s pid %d thread %s pid %d\n",

14
drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
View File

@ -305,6 +305,7 @@ static int gmc_v9_0_process_interrupt(struct amdgpu_device *adev,
struct amdgpu_iv_entry *entry)
{
struct amdgpu_vmhub *hub = &adev->vmhub[entry->vmid_src];
bool retry_fault = !!(entry->src_data[1] & 0x80);
uint32_t status = 0;
u64 addr;
@ -320,13 +321,16 @@ static int gmc_v9_0_process_interrupt(struct amdgpu_device *adev,
}
if (printk_ratelimit()) {
struct amdgpu_task_info task_info = { 0 };
struct amdgpu_task_info task_info;
memset(&task_info, 0, sizeof(struct amdgpu_task_info));
amdgpu_vm_get_task_info(adev, entry->pasid, &task_info);
dev_err(adev->dev,
"[%s] VMC page fault (src_id:%u ring:%u vmid:%u pasid:%u, for process %s pid %d thread %s pid %d)\n",
"[%s] %s page fault (src_id:%u ring:%u vmid:%u "
"pasid:%u, for process %s pid %d thread %s pid %d)\n",
entry->vmid_src ? "mmhub" : "gfxhub",
retry_fault ? "retry" : "no-retry",
entry->src_id, entry->ring_id, entry->vmid,
entry->pasid, task_info.process_name, task_info.tgid,
task_info.task_name, task_info.pid);
@ -961,7 +965,11 @@ static int gmc_v9_0_sw_init(void *handle)
* vm size is 256TB (48bit), maximum size of Vega10,
* block size 512 (9bit)
*/
amdgpu_vm_adjust_size(adev, 256 * 1024, 9, 3, 48);
/* sriov restrict max_pfn below AMDGPU_GMC_HOLE */
if (amdgpu_sriov_vf(adev))
amdgpu_vm_adjust_size(adev, 256 * 1024, 9, 3, 47);
else
amdgpu_vm_adjust_size(adev, 256 * 1024, 9, 3, 48);
break;
default:
break;

36
drivers/gpu/drm/amd/amdgpu/iceland_ih.c
View File

@ -103,9 +103,9 @@ static void iceland_ih_disable_interrupts(struct amdgpu_device *adev)
*/
static int iceland_ih_irq_init(struct amdgpu_device *adev)
{
struct amdgpu_ih_ring *ih = &adev->irq.ih;
int rb_bufsz;
u32 interrupt_cntl, ih_cntl, ih_rb_cntl;
u64 wptr_off;
/* disable irqs */
iceland_ih_disable_interrupts(adev);
@ -133,9 +133,8 @@ static int iceland_ih_irq_init(struct amdgpu_device *adev)
ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, WPTR_WRITEBACK_ENABLE, 1);
/* set the writeback address whether it's enabled or not */
wptr_off = adev->wb.gpu_addr + (adev->irq.ih.wptr_offs * 4);
WREG32(mmIH_RB_WPTR_ADDR_LO, lower_32_bits(wptr_off));
WREG32(mmIH_RB_WPTR_ADDR_HI, upper_32_bits(wptr_off) & 0xFF);
WREG32(mmIH_RB_WPTR_ADDR_LO, lower_32_bits(ih->wptr_addr));
WREG32(mmIH_RB_WPTR_ADDR_HI, upper_32_bits(ih->wptr_addr) & 0xFF);
WREG32(mmIH_RB_CNTL, ih_rb_cntl);
@ -185,11 +184,12 @@ static void iceland_ih_irq_disable(struct amdgpu_device *adev)
* Used by cz_irq_process(VI).
* Returns the value of the wptr.
*/
static u32 iceland_ih_get_wptr(struct amdgpu_device *adev)
static u32 iceland_ih_get_wptr(struct amdgpu_device *adev,
struct amdgpu_ih_ring *ih)
{
u32 wptr, tmp;
wptr = le32_to_cpu(adev->wb.wb[adev->irq.ih.wptr_offs]);
wptr = le32_to_cpu(*ih->wptr_cpu);
if (REG_GET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW)) {
wptr = REG_SET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW, 0);
@ -198,13 +198,13 @@ static u32 iceland_ih_get_wptr(struct amdgpu_device *adev)
* this should allow us to catchup.
*/
dev_warn(adev->dev, "IH ring buffer overflow (0x%08X, 0x%08X, 0x%08X)\n",
wptr, adev->irq.ih.rptr, (wptr + 16) & adev->irq.ih.ptr_mask);
adev->irq.ih.rptr = (wptr + 16) & adev->irq.ih.ptr_mask;
wptr, ih->rptr, (wptr + 16) & ih->ptr_mask);
ih->rptr = (wptr + 16) & ih->ptr_mask;
tmp = RREG32(mmIH_RB_CNTL);
tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, WPTR_OVERFLOW_CLEAR, 1);
WREG32(mmIH_RB_CNTL, tmp);
}
return (wptr & adev->irq.ih.ptr_mask);
return (wptr & ih->ptr_mask);
}
/**
@ -216,16 +216,17 @@ static u32 iceland_ih_get_wptr(struct amdgpu_device *adev)
* position and also advance the position.
*/
static void iceland_ih_decode_iv(struct amdgpu_device *adev,
struct amdgpu_ih_ring *ih,
struct amdgpu_iv_entry *entry)
{
/* wptr/rptr are in bytes! */
u32 ring_index = adev->irq.ih.rptr >> 2;
u32 ring_index = ih->rptr >> 2;
uint32_t dw[4];
dw[0] = le32_to_cpu(adev->irq.ih.ring[ring_index + 0]);
dw[1] = le32_to_cpu(adev->irq.ih.ring[ring_index + 1]);
dw[2] = le32_to_cpu(adev->irq.ih.ring[ring_index + 2]);
dw[3] = le32_to_cpu(adev->irq.ih.ring[ring_index + 3]);
dw[0] = le32_to_cpu(ih->ring[ring_index + 0]);
dw[1] = le32_to_cpu(ih->ring[ring_index + 1]);
dw[2] = le32_to_cpu(ih->ring[ring_index + 2]);
dw[3] = le32_to_cpu(ih->ring[ring_index + 3]);
entry->client_id = AMDGPU_IRQ_CLIENTID_LEGACY;
entry->src_id = dw[0] & 0xff;
@ -235,7 +236,7 @@ static void iceland_ih_decode_iv(struct amdgpu_device *adev,
entry->pasid = (dw[2] >> 16) & 0xffff;
/* wptr/rptr are in bytes! */
adev->irq.ih.rptr += 16;
ih->rptr += 16;
}
/**
@ -245,9 +246,10 @@ static void iceland_ih_decode_iv(struct amdgpu_device *adev,
*
* Set the IH ring buffer rptr.
*/
static void iceland_ih_set_rptr(struct amdgpu_device *adev)
static void iceland_ih_set_rptr(struct amdgpu_device *adev,
struct amdgpu_ih_ring *ih)
{
WREG32(mmIH_RB_RPTR, adev->irq.ih.rptr);
WREG32(mmIH_RB_RPTR, ih->rptr);
}
static int iceland_ih_early_init(void *handle)

2
drivers/gpu/drm/amd/amdgpu/mxgpu_ai.c
View File

@ -174,7 +174,7 @@ static int xgpu_ai_send_access_requests(struct amdgpu_device *adev,
return r;
}
/* Retrieve checksum from mailbox2 */
if (req == IDH_REQ_GPU_INIT_ACCESS) {
if (req == IDH_REQ_GPU_INIT_ACCESS || req == IDH_REQ_GPU_RESET_ACCESS) {
adev->virt.fw_reserve.checksum_key =
RREG32_NO_KIQ(SOC15_REG_OFFSET(NBIO, 0,
mmBIF_BX_PF0_MAILBOX_MSGBUF_RCV_DW2));

10
drivers/gpu/drm/amd/amdgpu/nbio_v6_1.c
View File

@ -27,13 +27,9 @@
#include "nbio/nbio_6_1_default.h"
#include "nbio/nbio_6_1_offset.h"
#include "nbio/nbio_6_1_sh_mask.h"
#include "nbio/nbio_6_1_smn.h"
#include "vega10_enum.h"
#define smnCPM_CONTROL 0x11180460
#define smnPCIE_CNTL2 0x11180070
#define smnPCIE_CONFIG_CNTL 0x11180044
#define smnPCIE_CI_CNTL 0x11180080
static u32 nbio_v6_1_get_rev_id(struct amdgpu_device *adev)
{
u32 tmp = RREG32_SOC15(NBIO, 0, mmRCC_DEV0_EPF0_STRAP0);
@ -72,7 +68,7 @@ static u32 nbio_v6_1_get_memsize(struct amdgpu_device *adev)
}
static void nbio_v6_1_sdma_doorbell_range(struct amdgpu_device *adev, int instance,
bool use_doorbell, int doorbell_index)
bool use_doorbell, int doorbell_index, int doorbell_size)
{
u32 reg = instance == 0 ? SOC15_REG_OFFSET(NBIO, 0, mmBIF_SDMA0_DOORBELL_RANGE) :
SOC15_REG_OFFSET(NBIO, 0, mmBIF_SDMA1_DOORBELL_RANGE);
@ -81,7 +77,7 @@ static void nbio_v6_1_sdma_doorbell_range(struct amdgpu_device *adev, int instan
if (use_doorbell) {
doorbell_range = REG_SET_FIELD(doorbell_range, BIF_SDMA0_DOORBELL_RANGE, OFFSET, doorbell_index);
doorbell_range = REG_SET_FIELD(doorbell_range, BIF_SDMA0_DOORBELL_RANGE, SIZE, 2);
doorbell_range = REG_SET_FIELD(doorbell_range, BIF_SDMA0_DOORBELL_RANGE, SIZE, doorbell_size);
} else
doorbell_range = REG_SET_FIELD(doorbell_range, BIF_SDMA0_DOORBELL_RANGE, SIZE, 0);

8
drivers/gpu/drm/amd/amdgpu/nbio_v7_0.c
View File

@ -27,13 +27,11 @@
#include "nbio/nbio_7_0_default.h"
#include "nbio/nbio_7_0_offset.h"
#include "nbio/nbio_7_0_sh_mask.h"
#include "nbio/nbio_7_0_smn.h"
#include "vega10_enum.h"
#define smnNBIF_MGCG_CTRL_LCLK 0x1013a05c
#define smnCPM_CONTROL 0x11180460
#define smnPCIE_CNTL2 0x11180070
static u32 nbio_v7_0_get_rev_id(struct amdgpu_device *adev)
{
u32 tmp = RREG32_SOC15(NBIO, 0, mmRCC_DEV0_EPF0_STRAP0);
@ -69,7 +67,7 @@ static u32 nbio_v7_0_get_memsize(struct amdgpu_device *adev)
}
static void nbio_v7_0_sdma_doorbell_range(struct amdgpu_device *adev, int instance,
bool use_doorbell, int doorbell_index)
bool use_doorbell, int doorbell_index, int doorbell_size)
{
u32 reg = instance == 0 ? SOC15_REG_OFFSET(NBIO, 0, mmBIF_SDMA0_DOORBELL_RANGE) :
SOC15_REG_OFFSET(NBIO, 0, mmBIF_SDMA1_DOORBELL_RANGE);
@ -78,7 +76,7 @@ static void nbio_v7_0_sdma_doorbell_range(struct amdgpu_device *adev, int instan
if (use_doorbell) {
doorbell_range = REG_SET_FIELD(doorbell_range, BIF_SDMA0_DOORBELL_RANGE, OFFSET, doorbell_index);
doorbell_range = REG_SET_FIELD(doorbell_range, BIF_SDMA0_DOORBELL_RANGE, SIZE, 2);
doorbell_range = REG_SET_FIELD(doorbell_range, BIF_SDMA0_DOORBELL_RANGE, SIZE, doorbell_size);
} else
doorbell_range = REG_SET_FIELD(doorbell_range, BIF_SDMA0_DOORBELL_RANGE, SIZE, 0);

9
drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c
View File

@ -26,13 +26,10 @@
#include "nbio/nbio_7_4_offset.h"
#include "nbio/nbio_7_4_sh_mask.h"
#include "nbio/nbio_7_4_0_smn.h"
#define smnNBIF_MGCG_CTRL_LCLK 0x1013a21c
#define smnCPM_CONTROL 0x11180460
#define smnPCIE_CNTL2 0x11180070
#define smnPCIE_CI_CNTL 0x11180080
static u32 nbio_v7_4_get_rev_id(struct amdgpu_device *adev)
{
u32 tmp = RREG32_SOC15(NBIO, 0, mmRCC_DEV0_EPF0_STRAP0);
@ -68,7 +65,7 @@ static u32 nbio_v7_4_get_memsize(struct amdgpu_device *adev)
}
static void nbio_v7_4_sdma_doorbell_range(struct amdgpu_device *adev, int instance,
bool use_doorbell, int doorbell_index)
bool use_doorbell, int doorbell_index, int doorbell_size)
{
u32 reg = instance == 0 ? SOC15_REG_OFFSET(NBIO, 0, mmBIF_SDMA0_DOORBELL_RANGE) :
SOC15_REG_OFFSET(NBIO, 0, mmBIF_SDMA1_DOORBELL_RANGE);
@ -77,7 +74,7 @@ static void nbio_v7_4_sdma_doorbell_range(struct amdgpu_device *adev, int instan
if (use_doorbell) {
doorbell_range = REG_SET_FIELD(doorbell_range, BIF_SDMA0_DOORBELL_RANGE, OFFSET, doorbell_index);
doorbell_range = REG_SET_FIELD(doorbell_range, BIF_SDMA0_DOORBELL_RANGE, SIZE, 2);
doorbell_range = REG_SET_FIELD(doorbell_range, BIF_SDMA0_DOORBELL_RANGE, SIZE, doorbell_size);
} else
doorbell_range = REG_SET_FIELD(doorbell_range, BIF_SDMA0_DOORBELL_RANGE, SIZE, 0);

2
drivers/gpu/drm/amd/amdgpu/psp_gfx_if.h
View File

@ -191,7 +191,7 @@ enum psp_gfx_fw_type
GFX_FW_TYPE_MMSCH = 19,
GFX_FW_TYPE_RLC_RESTORE_LIST_GPM_MEM = 20,
GFX_FW_TYPE_RLC_RESTORE_LIST_SRM_MEM = 21,
GFX_FW_TYPE_RLC_RESTORE_LIST_CNTL = 22,
GFX_FW_TYPE_RLC_RESTORE_LIST_SRM_CNTL = 22,
GFX_FW_TYPE_UVD1 = 23,
GFX_FW_TYPE_MAX = 24
};

90
drivers/gpu/drm/amd/amdgpu/psp_v10_0.c
View File

@ -38,75 +38,6 @@ MODULE_FIRMWARE("amdgpu/raven_asd.bin");
MODULE_FIRMWARE("amdgpu/picasso_asd.bin");
MODULE_FIRMWARE("amdgpu/raven2_asd.bin");
static int
psp_v10_0_get_fw_type(struct amdgpu_firmware_info *ucode, enum psp_gfx_fw_type *type)
{
switch(ucode->ucode_id) {
case AMDGPU_UCODE_ID_SDMA0:
*type = GFX_FW_TYPE_SDMA0;
break;
case AMDGPU_UCODE_ID_SDMA1:
*type = GFX_FW_TYPE_SDMA1;
break;
case AMDGPU_UCODE_ID_CP_CE:
*type = GFX_FW_TYPE_CP_CE;
break;
case AMDGPU_UCODE_ID_CP_PFP:
*type = GFX_FW_TYPE_CP_PFP;
break;
case AMDGPU_UCODE_ID_CP_ME:
*type = GFX_FW_TYPE_CP_ME;
break;
case AMDGPU_UCODE_ID_CP_MEC1:
*type = GFX_FW_TYPE_CP_MEC;
break;
case AMDGPU_UCODE_ID_CP_MEC1_JT:
*type = GFX_FW_TYPE_CP_MEC_ME1;
break;
case AMDGPU_UCODE_ID_CP_MEC2:
*type = GFX_FW_TYPE_CP_MEC;
break;
case AMDGPU_UCODE_ID_CP_MEC2_JT:
*type = GFX_FW_TYPE_CP_MEC_ME2;
break;
case AMDGPU_UCODE_ID_RLC_G:
*type = GFX_FW_TYPE_RLC_G;
break;
case AMDGPU_UCODE_ID_RLC_RESTORE_LIST_CNTL:
*type = GFX_FW_TYPE_RLC_RESTORE_LIST_CNTL;
break;
case AMDGPU_UCODE_ID_RLC_RESTORE_LIST_GPM_MEM:
*type = GFX_FW_TYPE_RLC_RESTORE_LIST_GPM_MEM;
break;
case AMDGPU_UCODE_ID_RLC_RESTORE_LIST_SRM_MEM:
*type = GFX_FW_TYPE_RLC_RESTORE_LIST_SRM_MEM;
break;
case AMDGPU_UCODE_ID_SMC:
*type = GFX_FW_TYPE_SMU;
break;
case AMDGPU_UCODE_ID_UVD:
*type = GFX_FW_TYPE_UVD;
break;
case AMDGPU_UCODE_ID_VCE:
*type = GFX_FW_TYPE_VCE;
break;
case AMDGPU_UCODE_ID_VCN:
*type = GFX_FW_TYPE_VCN;
break;
case AMDGPU_UCODE_ID_DMCU_ERAM:
*type = GFX_FW_TYPE_DMCU_ERAM;
break;
case AMDGPU_UCODE_ID_DMCU_INTV:
*type = GFX_FW_TYPE_DMCU_ISR;
break;
case AMDGPU_UCODE_ID_MAXIMUM:
default:
return -EINVAL;
}
return 0;
}
static int psp_v10_0_init_microcode(struct psp_context *psp)
{
struct amdgpu_device *adev = psp->adev;
@ -158,26 +89,6 @@ static int psp_v10_0_init_microcode(struct psp_context *psp)
return err;
}
static int psp_v10_0_prep_cmd_buf(struct amdgpu_firmware_info *ucode,
struct psp_gfx_cmd_resp *cmd)
{
int ret;
uint64_t fw_mem_mc_addr = ucode->mc_addr;
memset(cmd, 0, sizeof(struct psp_gfx_cmd_resp));
cmd->cmd_id = GFX_CMD_ID_LOAD_IP_FW;
cmd->cmd.cmd_load_ip_fw.fw_phy_addr_lo = lower_32_bits(fw_mem_mc_addr);
cmd->cmd.cmd_load_ip_fw.fw_phy_addr_hi = upper_32_bits(fw_mem_mc_addr);
cmd->cmd.cmd_load_ip_fw.fw_size = ucode->ucode_size;
ret = psp_v10_0_get_fw_type(ucode, &cmd->cmd.cmd_load_ip_fw.fw_type);
if (ret)
DRM_ERROR("Unknown firmware type\n");
return ret;
}
static int psp_v10_0_ring_init(struct psp_context *psp,
enum psp_ring_type ring_type)
{
@ -454,7 +365,6 @@ static int psp_v10_0_mode1_reset(struct psp_context *psp)
static const struct psp_funcs psp_v10_0_funcs = {
.init_microcode = psp_v10_0_init_microcode,
.prep_cmd_buf = psp_v10_0_prep_cmd_buf,
.ring_init = psp_v10_0_ring_init,
.ring_create = psp_v10_0_ring_create,
.ring_stop = psp_v10_0_ring_stop,

75
drivers/gpu/drm/amd/amdgpu/psp_v11_0.c
View File

@ -40,60 +40,6 @@ MODULE_FIRMWARE("amdgpu/vega20_ta.bin");
/* address block */
#define smnMP1_FIRMWARE_FLAGS 0x3010024
static int
psp_v11_0_get_fw_type(struct amdgpu_firmware_info *ucode, enum psp_gfx_fw_type *type)
{
switch (ucode->ucode_id) {
case AMDGPU_UCODE_ID_SDMA0:
*type = GFX_FW_TYPE_SDMA0;
break;
case AMDGPU_UCODE_ID_SDMA1:
*type = GFX_FW_TYPE_SDMA1;
break;
case AMDGPU_UCODE_ID_CP_CE:
*type = GFX_FW_TYPE_CP_CE;
break;
case AMDGPU_UCODE_ID_CP_PFP:
*type = GFX_FW_TYPE_CP_PFP;
break;
case AMDGPU_UCODE_ID_CP_ME:
*type = GFX_FW_TYPE_CP_ME;
break;
case AMDGPU_UCODE_ID_CP_MEC1:
*type = GFX_FW_TYPE_CP_MEC;
break;
case AMDGPU_UCODE_ID_CP_MEC1_JT:
*type = GFX_FW_TYPE_CP_MEC_ME1;
break;
case AMDGPU_UCODE_ID_CP_MEC2:
*type = GFX_FW_TYPE_CP_MEC;
break;
case AMDGPU_UCODE_ID_CP_MEC2_JT:
*type = GFX_FW_TYPE_CP_MEC_ME2;
break;
case AMDGPU_UCODE_ID_RLC_G:
*type = GFX_FW_TYPE_RLC_G;
break;
case AMDGPU_UCODE_ID_SMC:
*type = GFX_FW_TYPE_SMU;
break;
case AMDGPU_UCODE_ID_UVD:
*type = GFX_FW_TYPE_UVD;
break;
case AMDGPU_UCODE_ID_VCE:
*type = GFX_FW_TYPE_VCE;
break;
case AMDGPU_UCODE_ID_UVD1:
*type = GFX_FW_TYPE_UVD1;
break;
case AMDGPU_UCODE_ID_MAXIMUM:
default:
return -EINVAL;
}
return 0;
}
static int psp_v11_0_init_microcode(struct psp_context *psp)
{
struct amdgpu_device *adev = psp->adev;
@ -267,26 +213,6 @@ static int psp_v11_0_bootloader_load_sos(struct psp_context *psp)
return ret;
}
static int psp_v11_0_prep_cmd_buf(struct amdgpu_firmware_info *ucode,
struct psp_gfx_cmd_resp *cmd)
{
int ret;
uint64_t fw_mem_mc_addr = ucode->mc_addr;
memset(cmd, 0, sizeof(struct psp_gfx_cmd_resp));
cmd->cmd_id = GFX_CMD_ID_LOAD_IP_FW;
cmd->cmd.cmd_load_ip_fw.fw_phy_addr_lo = lower_32_bits(fw_mem_mc_addr);
cmd->cmd.cmd_load_ip_fw.fw_phy_addr_hi = upper_32_bits(fw_mem_mc_addr);
cmd->cmd.cmd_load_ip_fw.fw_size = ucode->ucode_size;
ret = psp_v11_0_get_fw_type(ucode, &cmd->cmd.cmd_load_ip_fw.fw_type);
if (ret)
DRM_ERROR("Unknown firmware type\n");
return ret;
}
static int psp_v11_0_ring_init(struct psp_context *psp,
enum psp_ring_type ring_type)
{
@ -753,7 +679,6 @@ static const struct psp_funcs psp_v11_0_funcs = {
.init_microcode = psp_v11_0_init_microcode,
.bootloader_load_sysdrv = psp_v11_0_bootloader_load_sysdrv,
.bootloader_load_sos = psp_v11_0_bootloader_load_sos,
.prep_cmd_buf = psp_v11_0_prep_cmd_buf,
.ring_init = psp_v11_0_ring_init,
.ring_create = psp_v11_0_ring_create,
.ring_stop = psp_v11_0_ring_stop,

72
drivers/gpu/drm/amd/amdgpu/psp_v3_1.c
View File

@ -47,57 +47,6 @@ MODULE_FIRMWARE("amdgpu/vega12_asd.bin");
static uint32_t sos_old_versions[] = {1517616, 1510592, 1448594, 1446554};
static int
psp_v3_1_get_fw_type(struct amdgpu_firmware_info *ucode, enum psp_gfx_fw_type *type)
{
switch(ucode->ucode_id) {
case AMDGPU_UCODE_ID_SDMA0:
*type = GFX_FW_TYPE_SDMA0;
break;
case AMDGPU_UCODE_ID_SDMA1:
*type = GFX_FW_TYPE_SDMA1;
break;
case AMDGPU_UCODE_ID_CP_CE:
*type = GFX_FW_TYPE_CP_CE;
break;
case AMDGPU_UCODE_ID_CP_PFP:
*type = GFX_FW_TYPE_CP_PFP;
break;
case AMDGPU_UCODE_ID_CP_ME:
*type = GFX_FW_TYPE_CP_ME;
break;
case AMDGPU_UCODE_ID_CP_MEC1:
*type = GFX_FW_TYPE_CP_MEC;
break;
case AMDGPU_UCODE_ID_CP_MEC1_JT:
*type = GFX_FW_TYPE_CP_MEC_ME1;
break;
case AMDGPU_UCODE_ID_CP_MEC2:
*type = GFX_FW_TYPE_CP_MEC;
break;
case AMDGPU_UCODE_ID_CP_MEC2_JT:
*type = GFX_FW_TYPE_CP_MEC_ME2;
break;
case AMDGPU_UCODE_ID_RLC_G:
*type = GFX_FW_TYPE_RLC_G;
break;
case AMDGPU_UCODE_ID_SMC:
*type = GFX_FW_TYPE_SMU;
break;
case AMDGPU_UCODE_ID_UVD:
*type = GFX_FW_TYPE_UVD;
break;
case AMDGPU_UCODE_ID_VCE:
*type = GFX_FW_TYPE_VCE;
break;
case AMDGPU_UCODE_ID_MAXIMUM:
default:
return -EINVAL;
}
return 0;
}
static int psp_v3_1_init_microcode(struct psp_context *psp)
{
struct amdgpu_device *adev = psp->adev;
@ -277,26 +226,6 @@ static int psp_v3_1_bootloader_load_sos(struct psp_context *psp)
return ret;
}
static int psp_v3_1_prep_cmd_buf(struct amdgpu_firmware_info *ucode,
struct psp_gfx_cmd_resp *cmd)
{
int ret;
uint64_t fw_mem_mc_addr = ucode->mc_addr;
memset(cmd, 0, sizeof(struct psp_gfx_cmd_resp));
cmd->cmd_id = GFX_CMD_ID_LOAD_IP_FW;
cmd->cmd.cmd_load_ip_fw.fw_phy_addr_lo = lower_32_bits(fw_mem_mc_addr);
cmd->cmd.cmd_load_ip_fw.fw_phy_addr_hi = upper_32_bits(fw_mem_mc_addr);
cmd->cmd.cmd_load_ip_fw.fw_size = ucode->ucode_size;
ret = psp_v3_1_get_fw_type(ucode, &cmd->cmd.cmd_load_ip_fw.fw_type);
if (ret)
DRM_ERROR("Unknown firmware type\n");
return ret;
}
static int psp_v3_1_ring_init(struct psp_context *psp,
enum psp_ring_type ring_type)
{
@ -615,7 +544,6 @@ static const struct psp_funcs psp_v3_1_funcs = {
.init_microcode = psp_v3_1_init_microcode,
.bootloader_load_sysdrv = psp_v3_1_bootloader_load_sysdrv,
.bootloader_load_sos = psp_v3_1_bootloader_load_sos,
.prep_cmd_buf = psp_v3_1_prep_cmd_buf,
.ring_init = psp_v3_1_ring_init,
.ring_create = psp_v3_1_ring_create,
.ring_stop = psp_v3_1_ring_stop,

2
drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c
View File

@ -247,7 +247,7 @@ static void sdma_v2_4_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count)
static void sdma_v2_4_ring_emit_ib(struct amdgpu_ring *ring,
struct amdgpu_job *job,
struct amdgpu_ib *ib,
bool ctx_switch)
uint32_t flags)
{
unsigned vmid = AMDGPU_JOB_GET_VMID(job);

5
drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c
View File

@ -421,7 +421,7 @@ static void sdma_v3_0_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count)
static void sdma_v3_0_ring_emit_ib(struct amdgpu_ring *ring,
struct amdgpu_job *job,
struct amdgpu_ib *ib,
bool ctx_switch)
uint32_t flags)
{
unsigned vmid = AMDGPU_JOB_GET_VMID(job);
@ -1145,8 +1145,7 @@ static int sdma_v3_0_sw_init(void *handle)
ring->ring_obj = NULL;
if (!amdgpu_sriov_vf(adev)) {
ring->use_doorbell = true;
ring->doorbell_index = (i == 0) ?
adev->doorbell_index.sdma_engine0 : adev->doorbell_index.sdma_engine1;
ring->doorbell_index = adev->doorbell_index.sdma_engine[i];
} else {
ring->use_pollmem = true;
}

15
drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c
View File

@ -78,7 +78,6 @@ static const struct soc15_reg_golden golden_settings_sdma_4[] = {
SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_RLC1_RB_WPTR_POLL_CNTL, 0x0000fff0, 0x00403000),
SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_UTCL1_PAGE, 0x000003ff, 0x000003c0),
SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_UTCL1_WATERMK, 0xfc000000, 0x00000000),
SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_CHICKEN_BITS, 0xfe931f07, 0x02831f07),
SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_CLK_CTRL, 0xffffffff, 0x3f000100),
SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_GFX_IB_CNTL, 0x800f0100, 0x00000100),
SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_GFX_RB_WPTR_POLL_CNTL, 0x0000fff0, 0x00403000),
@ -96,6 +95,7 @@ static const struct soc15_reg_golden golden_settings_sdma_4[] = {
static const struct soc15_reg_golden golden_settings_sdma_vg10[] = {
SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_GB_ADDR_CONFIG, 0x0018773f, 0x00104002),
SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_GB_ADDR_CONFIG_READ, 0x0018773f, 0x00104002),
SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_CHICKEN_BITS, 0xfe931f07, 0x02831d07),
SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_GB_ADDR_CONFIG, 0x0018773f, 0x00104002),
SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_GB_ADDR_CONFIG_READ, 0x0018773f, 0x00104002)
};
@ -103,6 +103,7 @@ static const struct soc15_reg_golden golden_settings_sdma_vg10[] = {
static const struct soc15_reg_golden golden_settings_sdma_vg12[] = {
SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_GB_ADDR_CONFIG, 0x0018773f, 0x00104001),
SOC15_REG_GOLDEN_VALUE(SDMA0, 0, mmSDMA0_GB_ADDR_CONFIG_READ, 0x0018773f, 0x00104001),
SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_CHICKEN_BITS, 0xfe931f07, 0x02831d07),
SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_GB_ADDR_CONFIG, 0x0018773f, 0x00104001),
SOC15_REG_GOLDEN_VALUE(SDMA1, 0, mmSDMA1_GB_ADDR_CONFIG_READ, 0x0018773f, 0x00104001)
};
@ -499,7 +500,7 @@ static void sdma_v4_0_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count)
static void sdma_v4_0_ring_emit_ib(struct amdgpu_ring *ring,
struct amdgpu_job *job,
struct amdgpu_ib *ib,
bool ctx_switch)
uint32_t flags)
{
unsigned vmid = AMDGPU_JOB_GET_VMID(job);
@ -833,8 +834,6 @@ static void sdma_v4_0_gfx_resume(struct amdgpu_device *adev, unsigned int i)
OFFSET, ring->doorbell_index);
WREG32_SDMA(i, mmSDMA0_GFX_DOORBELL, doorbell);
WREG32_SDMA(i, mmSDMA0_GFX_DOORBELL_OFFSET, doorbell_offset);
adev->nbio_funcs->sdma_doorbell_range(adev, i, ring->use_doorbell,
ring->doorbell_index);
sdma_v4_0_ring_set_wptr(ring);
@ -1521,9 +1520,7 @@ static int sdma_v4_0_sw_init(void *handle)
ring->use_doorbell?"true":"false");
/* doorbell size is 2 dwords, get DWORD offset */
ring->doorbell_index = (i == 0) ?
(adev->doorbell_index.sdma_engine0 << 1)
: (adev->doorbell_index.sdma_engine1 << 1);
ring->doorbell_index = adev->doorbell_index.sdma_engine[i] << 1;
sprintf(ring->name, "sdma%d", i);
r = amdgpu_ring_init(adev, ring, 1024,
@ -1542,9 +1539,7 @@ static int sdma_v4_0_sw_init(void *handle)
/* paging queue use same doorbell index/routing as gfx queue
* with 0x400 (4096 dwords) offset on second doorbell page
*/
ring->doorbell_index = (i == 0) ?
(adev->doorbell_index.sdma_engine0 << 1)
: (adev->doorbell_index.sdma_engine1 << 1);
ring->doorbell_index = adev->doorbell_index.sdma_engine[i] << 1;
ring->doorbell_index += 0x400;
sprintf(ring->name, "page%d", i);

54
drivers/gpu/drm/amd/amdgpu/si.c
View File

@ -47,6 +47,7 @@
#include "dce/dce_6_0_d.h"
#include "uvd/uvd_4_0_d.h"
#include "bif/bif_3_0_d.h"
#include "bif/bif_3_0_sh_mask.h"
static const u32 tahiti_golden_registers[] =
{
@ -1258,6 +1259,11 @@ static bool si_need_full_reset(struct amdgpu_device *adev)
return true;
}
static bool si_need_reset_on_init(struct amdgpu_device *adev)
{
return false;
}
static int si_get_pcie_lanes(struct amdgpu_device *adev)
{
u32 link_width_cntl;
@ -1323,6 +1329,52 @@ static void si_set_pcie_lanes(struct amdgpu_device *adev, int lanes)
WREG32_PCIE_PORT(PCIE_LC_LINK_WIDTH_CNTL, link_width_cntl);
}
static void si_get_pcie_usage(struct amdgpu_device *adev, uint64_t *count0,
uint64_t *count1)
{
uint32_t perfctr = 0;
uint64_t cnt0_of, cnt1_of;
int tmp;
/* This reports 0 on APUs, so return to avoid writing/reading registers
* that may or may not be different from their GPU counterparts
*/
if (adev->flags & AMD_IS_APU)
return;
/* Set the 2 events that we wish to watch, defined above */
/* Reg 40 is # received msgs, Reg 104 is # of posted requests sent */
perfctr = REG_SET_FIELD(perfctr, PCIE_PERF_CNTL_TXCLK, EVENT0_SEL, 40);
perfctr = REG_SET_FIELD(perfctr, PCIE_PERF_CNTL_TXCLK, EVENT1_SEL, 104);
/* Write to enable desired perf counters */
WREG32_PCIE(ixPCIE_PERF_CNTL_TXCLK, perfctr);
/* Zero out and enable the perf counters
* Write 0x5:
* Bit 0 = Start all counters(1)
* Bit 2 = Global counter reset enable(1)
*/
WREG32_PCIE(ixPCIE_PERF_COUNT_CNTL, 0x00000005);
msleep(1000);
/* Load the shadow and disable the perf counters
* Write 0x2:
* Bit 0 = Stop counters(0)
* Bit 1 = Load the shadow counters(1)
*/
WREG32_PCIE(ixPCIE_PERF_COUNT_CNTL, 0x00000002);
/* Read register values to get any >32bit overflow */
tmp = RREG32_PCIE(ixPCIE_PERF_CNTL_TXCLK);
cnt0_of = REG_GET_FIELD(tmp, PCIE_PERF_CNTL_TXCLK, COUNTER0_UPPER);
cnt1_of = REG_GET_FIELD(tmp, PCIE_PERF_CNTL_TXCLK, COUNTER1_UPPER);
/* Get the values and add the overflow */
*count0 = RREG32_PCIE(ixPCIE_PERF_COUNT0_TXCLK) | (cnt0_of << 32);
*count1 = RREG32_PCIE(ixPCIE_PERF_COUNT1_TXCLK) | (cnt1_of << 32);
}
static const struct amdgpu_asic_funcs si_asic_funcs =
{
.read_disabled_bios = &si_read_disabled_bios,
@ -1339,6 +1391,8 @@ static const struct amdgpu_asic_funcs si_asic_funcs =
.flush_hdp = &si_flush_hdp,
.invalidate_hdp = &si_invalidate_hdp,
.need_full_reset = &si_need_full_reset,
.get_pcie_usage = &si_get_pcie_usage,
.need_reset_on_init = &si_need_reset_on_init,
};
static uint32_t si_get_rev_id(struct amdgpu_device *adev)

2
drivers/gpu/drm/amd/amdgpu/si_dma.c
View File

@ -63,7 +63,7 @@ static void si_dma_ring_set_wptr(struct amdgpu_ring *ring)
static void si_dma_ring_emit_ib(struct amdgpu_ring *ring,
struct amdgpu_job *job,
struct amdgpu_ib *ib,
bool ctx_switch)
uint32_t flags)
{
unsigned vmid = AMDGPU_JOB_GET_VMID(job);
/* The indirect buffer packet must end on an 8 DW boundary in the DMA ring.

38
drivers/gpu/drm/amd/amdgpu/si_ih.c
View File

@ -57,9 +57,9 @@ static void si_ih_disable_interrupts(struct amdgpu_device *adev)
static int si_ih_irq_init(struct amdgpu_device *adev)
{
struct amdgpu_ih_ring *ih = &adev->irq.ih;
int rb_bufsz;
u32 interrupt_cntl, ih_cntl, ih_rb_cntl;
u64 wptr_off;
si_ih_disable_interrupts(adev);
WREG32(INTERRUPT_CNTL2, adev->irq.ih.gpu_addr >> 8);
@ -76,9 +76,8 @@ static int si_ih_irq_init(struct amdgpu_device *adev)
(rb_bufsz << 1) |
IH_WPTR_WRITEBACK_ENABLE;
wptr_off = adev->wb.gpu_addr + (adev->irq.ih.wptr_offs * 4);
WREG32(IH_RB_WPTR_ADDR_LO, lower_32_bits(wptr_off));
WREG32(IH_RB_WPTR_ADDR_HI, upper_32_bits(wptr_off) & 0xFF);
WREG32(IH_RB_WPTR_ADDR_LO, lower_32_bits(ih->wptr_addr));
WREG32(IH_RB_WPTR_ADDR_HI, upper_32_bits(ih->wptr_addr) & 0xFF);
WREG32(IH_RB_CNTL, ih_rb_cntl);
WREG32(IH_RB_RPTR, 0);
WREG32(IH_RB_WPTR, 0);
@ -100,34 +99,36 @@ static void si_ih_irq_disable(struct amdgpu_device *adev)
mdelay(1);
}
static u32 si_ih_get_wptr(struct amdgpu_device *adev)
static u32 si_ih_get_wptr(struct amdgpu_device *adev,
struct amdgpu_ih_ring *ih)
{
u32 wptr, tmp;
wptr = le32_to_cpu(adev->wb.wb[adev->irq.ih.wptr_offs]);
wptr = le32_to_cpu(*ih->wptr_cpu);
if (wptr & IH_RB_WPTR__RB_OVERFLOW_MASK) {
wptr &= ~IH_RB_WPTR__RB_OVERFLOW_MASK;
dev_warn(adev->dev, "IH ring buffer overflow (0x%08X, 0x%08X, 0x%08X)\n",
wptr, adev->irq.ih.rptr, (wptr + 16) & adev->irq.ih.ptr_mask);
adev->irq.ih.rptr = (wptr + 16) & adev->irq.ih.ptr_mask;
wptr, ih->rptr, (wptr + 16) & ih->ptr_mask);
ih->rptr = (wptr + 16) & ih->ptr_mask;
tmp = RREG32(IH_RB_CNTL);
tmp |= IH_RB_CNTL__WPTR_OVERFLOW_CLEAR_MASK;
WREG32(IH_RB_CNTL, tmp);
}
return (wptr & adev->irq.ih.ptr_mask);
return (wptr & ih->ptr_mask);
}
static void si_ih_decode_iv(struct amdgpu_device *adev,
struct amdgpu_iv_entry *entry)
struct amdgpu_ih_ring *ih,
struct amdgpu_iv_entry *entry)
{
u32 ring_index = adev->irq.ih.rptr >> 2;
u32 ring_index = ih->rptr >> 2;
uint32_t dw[4];
dw[0] = le32_to_cpu(adev->irq.ih.ring[ring_index + 0]);
dw[1] = le32_to_cpu(adev->irq.ih.ring[ring_index + 1]);
dw[2] = le32_to_cpu(adev->irq.ih.ring[ring_index + 2]);
dw[3] = le32_to_cpu(adev->irq.ih.ring[ring_index + 3]);
dw[0] = le32_to_cpu(ih->ring[ring_index + 0]);
dw[1] = le32_to_cpu(ih->ring[ring_index + 1]);
dw[2] = le32_to_cpu(ih->ring[ring_index + 2]);
dw[3] = le32_to_cpu(ih->ring[ring_index + 3]);
entry->client_id = AMDGPU_IRQ_CLIENTID_LEGACY;
entry->src_id = dw[0] & 0xff;
@ -135,12 +136,13 @@ static void si_ih_decode_iv(struct amdgpu_device *adev,
entry->ring_id = dw[2] & 0xff;
entry->vmid = (dw[2] >> 8) & 0xff;
adev->irq.ih.rptr += 16;
ih->rptr += 16;
}
static void si_ih_set_rptr(struct amdgpu_device *adev)
static void si_ih_set_rptr(struct amdgpu_device *adev,
struct amdgpu_ih_ring *ih)
{
WREG32(IH_RB_RPTR, adev->irq.ih.rptr);
WREG32(IH_RB_RPTR, ih->rptr);
}
static int si_ih_early_init(void *handle)

167
drivers/gpu/drm/amd/amdgpu/soc15.c
View File

@ -43,6 +43,10 @@
#include "hdp/hdp_4_0_sh_mask.h"
#include "smuio/smuio_9_0_offset.h"
#include "smuio/smuio_9_0_sh_mask.h"
#include "nbio/nbio_7_0_default.h"
#include "nbio/nbio_7_0_sh_mask.h"
#include "nbio/nbio_7_0_smn.h"
#include "mp/mp_9_0_offset.h"
#include "soc15.h"
#include "soc15_common.h"
@ -385,14 +389,13 @@ void soc15_program_register_sequence(struct amdgpu_device *adev,
}
static int soc15_asic_reset(struct amdgpu_device *adev)
static int soc15_asic_mode1_reset(struct amdgpu_device *adev)
{
u32 i;
amdgpu_atombios_scratch_regs_engine_hung(adev, true);
dev_info(adev->dev, "GPU reset\n");
dev_info(adev->dev, "GPU mode1 reset\n");
/* disable BM */
pci_clear_master(adev->pdev);
@ -417,6 +420,63 @@ static int soc15_asic_reset(struct amdgpu_device *adev)
return 0;
}
static int soc15_asic_get_baco_capability(struct amdgpu_device *adev, bool *cap)
{
void *pp_handle = adev->powerplay.pp_handle;
const struct amd_pm_funcs *pp_funcs = adev->powerplay.pp_funcs;
if (!pp_funcs || !pp_funcs->get_asic_baco_capability) {
*cap = false;
return -ENOENT;
}
return pp_funcs->get_asic_baco_capability(pp_handle, cap);
}
static int soc15_asic_baco_reset(struct amdgpu_device *adev)
{
void *pp_handle = adev->powerplay.pp_handle;
const struct amd_pm_funcs *pp_funcs = adev->powerplay.pp_funcs;
if (!pp_funcs ||!pp_funcs->get_asic_baco_state ||!pp_funcs->set_asic_baco_state)
return -ENOENT;
/* enter BACO state */
if (pp_funcs->set_asic_baco_state(pp_handle, 1))
return -EIO;
/* exit BACO state */
if (pp_funcs->set_asic_baco_state(pp_handle, 0))
return -EIO;
dev_info(adev->dev, "GPU BACO reset\n");
return 0;
}
static int soc15_asic_reset(struct amdgpu_device *adev)
{
int ret;
bool baco_reset;
switch (adev->asic_type) {
case CHIP_VEGA10:
case CHIP_VEGA20:
soc15_asic_get_baco_capability(adev, &baco_reset);
break;
default:
baco_reset = false;
break;
}
if (baco_reset)
ret = soc15_asic_baco_reset(adev);
else
ret = soc15_asic_mode1_reset(adev);
return ret;
}
/*static int soc15_set_uvd_clock(struct amdgpu_device *adev, u32 clock,
u32 cntl_reg, u32 status_reg)
{
@ -535,10 +595,12 @@ int soc15_set_ip_blocks(struct amdgpu_device *adev)
amdgpu_device_ip_block_add(adev, &vega10_common_ip_block);
amdgpu_device_ip_block_add(adev, &gmc_v9_0_ip_block);
amdgpu_device_ip_block_add(adev, &vega10_ih_ip_block);
if (adev->asic_type == CHIP_VEGA20)
amdgpu_device_ip_block_add(adev, &psp_v11_0_ip_block);
else
amdgpu_device_ip_block_add(adev, &psp_v3_1_ip_block);
if (likely(adev->firmware.load_type == AMDGPU_FW_LOAD_PSP)) {
if (adev->asic_type == CHIP_VEGA20)
amdgpu_device_ip_block_add(adev, &psp_v11_0_ip_block);
else
amdgpu_device_ip_block_add(adev, &psp_v3_1_ip_block);
}
amdgpu_device_ip_block_add(adev, &gfx_v9_0_ip_block);
amdgpu_device_ip_block_add(adev, &sdma_v4_0_ip_block);
if (!amdgpu_sriov_vf(adev))
@ -560,7 +622,8 @@ int soc15_set_ip_blocks(struct amdgpu_device *adev)
amdgpu_device_ip_block_add(adev, &vega10_common_ip_block);
amdgpu_device_ip_block_add(adev, &gmc_v9_0_ip_block);
amdgpu_device_ip_block_add(adev, &vega10_ih_ip_block);
amdgpu_device_ip_block_add(adev, &psp_v10_0_ip_block);
if (likely(adev->firmware.load_type == AMDGPU_FW_LOAD_PSP))
amdgpu_device_ip_block_add(adev, &psp_v10_0_ip_block);
amdgpu_device_ip_block_add(adev, &gfx_v9_0_ip_block);
amdgpu_device_ip_block_add(adev, &sdma_v4_0_ip_block);
amdgpu_device_ip_block_add(adev, &pp_smu_ip_block);
@ -601,6 +664,68 @@ static bool soc15_need_full_reset(struct amdgpu_device *adev)
/* change this when we implement soft reset */
return true;
}
static void soc15_get_pcie_usage(struct amdgpu_device *adev, uint64_t *count0,
uint64_t *count1)
{
uint32_t perfctr = 0;
uint64_t cnt0_of, cnt1_of;
int tmp;
/* This reports 0 on APUs, so return to avoid writing/reading registers
* that may or may not be different from their GPU counterparts
*/
if (adev->flags & AMD_IS_APU)
return;
/* Set the 2 events that we wish to watch, defined above */
/* Reg 40 is # received msgs, Reg 104 is # of posted requests sent */
perfctr = REG_SET_FIELD(perfctr, PCIE_PERF_CNTL_TXCLK, EVENT0_SEL, 40);
perfctr = REG_SET_FIELD(perfctr, PCIE_PERF_CNTL_TXCLK, EVENT1_SEL, 104);
/* Write to enable desired perf counters */
WREG32_PCIE(smnPCIE_PERF_CNTL_TXCLK, perfctr);
/* Zero out and enable the perf counters
* Write 0x5:
* Bit 0 = Start all counters(1)
* Bit 2 = Global counter reset enable(1)
*/
WREG32_PCIE(smnPCIE_PERF_COUNT_CNTL, 0x00000005);
msleep(1000);
/* Load the shadow and disable the perf counters
* Write 0x2:
* Bit 0 = Stop counters(0)
* Bit 1 = Load the shadow counters(1)
*/
WREG32_PCIE(smnPCIE_PERF_COUNT_CNTL, 0x00000002);
/* Read register values to get any >32bit overflow */
tmp = RREG32_PCIE(smnPCIE_PERF_CNTL_TXCLK);
cnt0_of = REG_GET_FIELD(tmp, PCIE_PERF_CNTL_TXCLK, COUNTER0_UPPER);
cnt1_of = REG_GET_FIELD(tmp, PCIE_PERF_CNTL_TXCLK, COUNTER1_UPPER);
/* Get the values and add the overflow */
*count0 = RREG32_PCIE(smnPCIE_PERF_COUNT0_TXCLK) | (cnt0_of << 32);
*count1 = RREG32_PCIE(smnPCIE_PERF_COUNT1_TXCLK) | (cnt1_of << 32);
}
static bool soc15_need_reset_on_init(struct amdgpu_device *adev)
{
u32 sol_reg;
if (adev->flags & AMD_IS_APU)
return false;
/* Check sOS sign of life register to confirm sys driver and sOS
* are already been loaded.
*/
sol_reg = RREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_81);
if (sol_reg)
return true;
return false;
}
static const struct amdgpu_asic_funcs soc15_asic_funcs =
{
@ -617,6 +742,8 @@ static const struct amdgpu_asic_funcs soc15_asic_funcs =
.invalidate_hdp = &soc15_invalidate_hdp,
.need_full_reset = &soc15_need_full_reset,
.init_doorbell_index = &vega10_doorbell_index_init,
.get_pcie_usage = &soc15_get_pcie_usage,
.need_reset_on_init = &soc15_need_reset_on_init,
};
static const struct amdgpu_asic_funcs vega20_asic_funcs =
@ -634,6 +761,8 @@ static const struct amdgpu_asic_funcs vega20_asic_funcs =
.invalidate_hdp = &soc15_invalidate_hdp,
.need_full_reset = &soc15_need_full_reset,
.init_doorbell_index = &vega20_doorbell_index_init,
.get_pcie_usage = &soc15_get_pcie_usage,
.need_reset_on_init = &soc15_need_reset_on_init,
};
static int soc15_common_early_init(void *handle)
@ -840,6 +969,22 @@ static int soc15_common_sw_fini(void *handle)
return 0;
}
static void soc15_doorbell_range_init(struct amdgpu_device *adev)
{
int i;
struct amdgpu_ring *ring;
for (i = 0; i < adev->sdma.num_instances; i++) {
ring = &adev->sdma.instance[i].ring;
adev->nbio_funcs->sdma_doorbell_range(adev, i,
ring->use_doorbell, ring->doorbell_index,
adev->doorbell_index.sdma_doorbell_range);
}
adev->nbio_funcs->ih_doorbell_range(adev, adev->irq.ih.use_doorbell,
adev->irq.ih.doorbell_index);
}
static int soc15_common_hw_init(void *handle)
{
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
@ -852,6 +997,12 @@ static int soc15_common_hw_init(void *handle)
adev->nbio_funcs->init_registers(adev);
/* enable the doorbell aperture */
soc15_enable_doorbell_aperture(adev, true);
/* HW doorbell routing policy: doorbell writing not
* in SDMA/IH/MM/ACV range will be routed to CP. So
* we need to init SDMA/IH/MM/ACV doorbell range prior
* to CP ip block init and ring test.
*/
soc15_doorbell_range_init(adev);
return 0;
}

60
drivers/gpu/drm/amd/amdgpu/tonga_ih.c
View File

@ -99,9 +99,9 @@ static void tonga_ih_disable_interrupts(struct amdgpu_device *adev)
*/
static int tonga_ih_irq_init(struct amdgpu_device *adev)
{
int rb_bufsz;
u32 interrupt_cntl, ih_rb_cntl, ih_doorbell_rtpr;
u64 wptr_off;
struct amdgpu_ih_ring *ih = &adev->irq.ih;
int rb_bufsz;
/* disable irqs */
tonga_ih_disable_interrupts(adev);
@ -118,10 +118,7 @@ static int tonga_ih_irq_init(struct amdgpu_device *adev)
WREG32(mmINTERRUPT_CNTL, interrupt_cntl);
/* Ring Buffer base. [39:8] of 40-bit address of the beginning of the ring buffer*/
if (adev->irq.ih.use_bus_addr)
WREG32(mmIH_RB_BASE, adev->irq.ih.rb_dma_addr >> 8);
else
WREG32(mmIH_RB_BASE, adev->irq.ih.gpu_addr >> 8);
WREG32(mmIH_RB_BASE, ih->gpu_addr >> 8);
rb_bufsz = order_base_2(adev->irq.ih.ring_size / 4);
ih_rb_cntl = REG_SET_FIELD(0, IH_RB_CNTL, WPTR_OVERFLOW_CLEAR, 1);
@ -136,12 +133,8 @@ static int tonga_ih_irq_init(struct amdgpu_device *adev)
WREG32(mmIH_RB_CNTL, ih_rb_cntl);
/* set the writeback address whether it's enabled or not */
if (adev->irq.ih.use_bus_addr)
wptr_off = adev->irq.ih.rb_dma_addr + (adev->irq.ih.wptr_offs * 4);
else
wptr_off = adev->wb.gpu_addr + (adev->irq.ih.wptr_offs * 4);
WREG32(mmIH_RB_WPTR_ADDR_LO, lower_32_bits(wptr_off));
WREG32(mmIH_RB_WPTR_ADDR_HI, upper_32_bits(wptr_off) & 0xFF);
WREG32(mmIH_RB_WPTR_ADDR_LO, lower_32_bits(ih->wptr_addr));
WREG32(mmIH_RB_WPTR_ADDR_HI, upper_32_bits(ih->wptr_addr) & 0xFF);
/* set rptr, wptr to 0 */
WREG32(mmIH_RB_RPTR, 0);
@ -193,14 +186,12 @@ static void tonga_ih_irq_disable(struct amdgpu_device *adev)
* Used by cz_irq_process(VI).
* Returns the value of the wptr.
*/
static u32 tonga_ih_get_wptr(struct amdgpu_device *adev)
static u32 tonga_ih_get_wptr(struct amdgpu_device *adev,
struct amdgpu_ih_ring *ih)
{
u32 wptr, tmp;
if (adev->irq.ih.use_bus_addr)
wptr = le32_to_cpu(adev->irq.ih.ring[adev->irq.ih.wptr_offs]);
else
wptr = le32_to_cpu(adev->wb.wb[adev->irq.ih.wptr_offs]);
wptr = le32_to_cpu(*ih->wptr_cpu);
if (REG_GET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW)) {
wptr = REG_SET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW, 0);
@ -209,13 +200,13 @@ static u32 tonga_ih_get_wptr(struct amdgpu_device *adev)
* this should allow us to catchup.
*/
dev_warn(adev->dev, "IH ring buffer overflow (0x%08X, 0x%08X, 0x%08X)\n",
wptr, adev->irq.ih.rptr, (wptr + 16) & adev->irq.ih.ptr_mask);
adev->irq.ih.rptr = (wptr + 16) & adev->irq.ih.ptr_mask;
wptr, ih->rptr, (wptr + 16) & ih->ptr_mask);
ih->rptr = (wptr + 16) & ih->ptr_mask;
tmp = RREG32(mmIH_RB_CNTL);
tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, WPTR_OVERFLOW_CLEAR, 1);
WREG32(mmIH_RB_CNTL, tmp);
}
return (wptr & adev->irq.ih.ptr_mask);
return (wptr & ih->ptr_mask);
}
/**
@ -227,16 +218,17 @@ static u32 tonga_ih_get_wptr(struct amdgpu_device *adev)
* position and also advance the position.
*/
static void tonga_ih_decode_iv(struct amdgpu_device *adev,
struct amdgpu_iv_entry *entry)
struct amdgpu_ih_ring *ih,
struct amdgpu_iv_entry *entry)
{
/* wptr/rptr are in bytes! */
u32 ring_index = adev->irq.ih.rptr >> 2;
u32 ring_index = ih->rptr >> 2;
uint32_t dw[4];
dw[0] = le32_to_cpu(adev->irq.ih.ring[ring_index + 0]);
dw[1] = le32_to_cpu(adev->irq.ih.ring[ring_index + 1]);
dw[2] = le32_to_cpu(adev->irq.ih.ring[ring_index + 2]);
dw[3] = le32_to_cpu(adev->irq.ih.ring[ring_index + 3]);
dw[0] = le32_to_cpu(ih->ring[ring_index + 0]);
dw[1] = le32_to_cpu(ih->ring[ring_index + 1]);
dw[2] = le32_to_cpu(ih->ring[ring_index + 2]);
dw[3] = le32_to_cpu(ih->ring[ring_index + 3]);
entry->client_id = AMDGPU_IRQ_CLIENTID_LEGACY;
entry->src_id = dw[0] & 0xff;
@ -246,7 +238,7 @@ static void tonga_ih_decode_iv(struct amdgpu_device *adev,
entry->pasid = (dw[2] >> 16) & 0xffff;
/* wptr/rptr are in bytes! */
adev->irq.ih.rptr += 16;
ih->rptr += 16;
}
/**
@ -256,17 +248,15 @@ static void tonga_ih_decode_iv(struct amdgpu_device *adev,
*
* Set the IH ring buffer rptr.
*/
static void tonga_ih_set_rptr(struct amdgpu_device *adev)
static void tonga_ih_set_rptr(struct amdgpu_device *adev,
struct amdgpu_ih_ring *ih)
{
if (adev->irq.ih.use_doorbell) {
if (ih->use_doorbell) {
/* XXX check if swapping is necessary on BE */
if (adev->irq.ih.use_bus_addr)
adev->irq.ih.ring[adev->irq.ih.rptr_offs] = adev->irq.ih.rptr;
else
adev->wb.wb[adev->irq.ih.rptr_offs] = adev->irq.ih.rptr;
WDOORBELL32(adev->irq.ih.doorbell_index, adev->irq.ih.rptr);
*ih->rptr_cpu = ih->rptr;
WDOORBELL32(ih->doorbell_index, ih->rptr);
} else {
WREG32(mmIH_RB_RPTR, adev->irq.ih.rptr);
WREG32(mmIH_RB_RPTR, ih->rptr);
}
}

2
drivers/gpu/drm/amd/amdgpu/uvd_v4_2.c
View File

@ -511,7 +511,7 @@ static int uvd_v4_2_ring_test_ring(struct amdgpu_ring *ring)
static void uvd_v4_2_ring_emit_ib(struct amdgpu_ring *ring,
struct amdgpu_job *job,
struct amdgpu_ib *ib,
bool ctx_switch)
uint32_t flags)
{
amdgpu_ring_write(ring, PACKET0(mmUVD_RBC_IB_BASE, 0));
amdgpu_ring_write(ring, ib->gpu_addr);

2
drivers/gpu/drm/amd/amdgpu/uvd_v5_0.c
View File

@ -526,7 +526,7 @@ static int uvd_v5_0_ring_test_ring(struct amdgpu_ring *ring)
static void uvd_v5_0_ring_emit_ib(struct amdgpu_ring *ring,
struct amdgpu_job *job,
struct amdgpu_ib *ib,
bool ctx_switch)
uint32_t flags)
{
amdgpu_ring_write(ring, PACKET0(mmUVD_LMI_RBC_IB_64BIT_BAR_LOW, 0));
amdgpu_ring_write(ring, lower_32_bits(ib->gpu_addr));

4
drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c
View File

@ -977,7 +977,7 @@ static int uvd_v6_0_ring_test_ring(struct amdgpu_ring *ring)
static void uvd_v6_0_ring_emit_ib(struct amdgpu_ring *ring,
struct amdgpu_job *job,
struct amdgpu_ib *ib,
bool ctx_switch)
uint32_t flags)
{
unsigned vmid = AMDGPU_JOB_GET_VMID(job);
@ -1003,7 +1003,7 @@ static void uvd_v6_0_ring_emit_ib(struct amdgpu_ring *ring,
static void uvd_v6_0_enc_ring_emit_ib(struct amdgpu_ring *ring,
struct amdgpu_job *job,
struct amdgpu_ib *ib,
bool ctx_switch)
uint32_t flags)
{
unsigned vmid = AMDGPU_JOB_GET_VMID(job);

4
drivers/gpu/drm/amd/amdgpu/uvd_v7_0.c
View File

@ -1272,7 +1272,7 @@ static int uvd_v7_0_ring_patch_cs_in_place(struct amdgpu_cs_parser *p,
static void uvd_v7_0_ring_emit_ib(struct amdgpu_ring *ring,
struct amdgpu_job *job,
struct amdgpu_ib *ib,
bool ctx_switch)
uint32_t flags)
{
struct amdgpu_device *adev = ring->adev;
unsigned vmid = AMDGPU_JOB_GET_VMID(job);
@ -1303,7 +1303,7 @@ static void uvd_v7_0_ring_emit_ib(struct amdgpu_ring *ring,
static void uvd_v7_0_enc_ring_emit_ib(struct amdgpu_ring *ring,
struct amdgpu_job *job,
struct amdgpu_ib *ib,
bool ctx_switch)
uint32_t flags)
{
unsigned vmid = AMDGPU_JOB_GET_VMID(job);

2
drivers/gpu/drm/amd/amdgpu/vce_v3_0.c
View File

@ -834,7 +834,7 @@ static void vce_v3_0_get_clockgating_state(void *handle, u32 *flags)
static void vce_v3_0_ring_emit_ib(struct amdgpu_ring *ring,
struct amdgpu_job *job,
struct amdgpu_ib *ib,
bool ctx_switch)
uint32_t flags)
{
unsigned vmid = AMDGPU_JOB_GET_VMID(job);

2
drivers/gpu/drm/amd/amdgpu/vce_v4_0.c
View File

@ -947,7 +947,7 @@ static int vce_v4_0_set_powergating_state(void *handle,
#endif
static void vce_v4_0_ring_emit_ib(struct amdgpu_ring *ring, struct amdgpu_job *job,
struct amdgpu_ib *ib, bool ctx_switch)
struct amdgpu_ib *ib, uint32_t flags)
{
unsigned vmid = AMDGPU_JOB_GET_VMID(job);

6
drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c
View File

@ -1371,7 +1371,7 @@ static void vcn_v1_0_dec_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64
static void vcn_v1_0_dec_ring_emit_ib(struct amdgpu_ring *ring,
struct amdgpu_job *job,
struct amdgpu_ib *ib,
bool ctx_switch)
uint32_t flags)
{
struct amdgpu_device *adev = ring->adev;
unsigned vmid = AMDGPU_JOB_GET_VMID(job);
@ -1531,7 +1531,7 @@ static void vcn_v1_0_enc_ring_insert_end(struct amdgpu_ring *ring)
static void vcn_v1_0_enc_ring_emit_ib(struct amdgpu_ring *ring,
struct amdgpu_job *job,
struct amdgpu_ib *ib,
bool ctx_switch)
uint32_t flags)
{
unsigned vmid = AMDGPU_JOB_GET_VMID(job);
@ -1736,7 +1736,7 @@ static void vcn_v1_0_jpeg_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u6
static void vcn_v1_0_jpeg_ring_emit_ib(struct amdgpu_ring *ring,
struct amdgpu_job *job,
struct amdgpu_ib *ib,
bool ctx_switch)
uint32_t flags)
{
struct amdgpu_device *adev = ring->adev;
unsigned vmid = AMDGPU_JOB_GET_VMID(job);

318
drivers/gpu/drm/amd/amdgpu/vega10_ih.c
View File

@ -50,6 +50,22 @@ static void vega10_ih_enable_interrupts(struct amdgpu_device *adev)
ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, ENABLE_INTR, 1);
WREG32_SOC15(OSSSYS, 0, mmIH_RB_CNTL, ih_rb_cntl);
adev->irq.ih.enabled = true;
if (adev->irq.ih1.ring_size) {
ih_rb_cntl = RREG32_SOC15(OSSSYS, 0, mmIH_RB_CNTL_RING1);
ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL_RING1,
RB_ENABLE, 1);
WREG32_SOC15(OSSSYS, 0, mmIH_RB_CNTL_RING1, ih_rb_cntl);
adev->irq.ih1.enabled = true;
}
if (adev->irq.ih2.ring_size) {
ih_rb_cntl = RREG32_SOC15(OSSSYS, 0, mmIH_RB_CNTL_RING2);
ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL_RING2,
RB_ENABLE, 1);
WREG32_SOC15(OSSSYS, 0, mmIH_RB_CNTL_RING2, ih_rb_cntl);
adev->irq.ih2.enabled = true;
}
}
/**
@ -71,6 +87,53 @@ static void vega10_ih_disable_interrupts(struct amdgpu_device *adev)
WREG32_SOC15(OSSSYS, 0, mmIH_RB_WPTR, 0);
adev->irq.ih.enabled = false;
adev->irq.ih.rptr = 0;
if (adev->irq.ih1.ring_size) {
ih_rb_cntl = RREG32_SOC15(OSSSYS, 0, mmIH_RB_CNTL_RING1);
ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL_RING1,
RB_ENABLE, 0);
WREG32_SOC15(OSSSYS, 0, mmIH_RB_CNTL_RING1, ih_rb_cntl);
/* set rptr, wptr to 0 */
WREG32_SOC15(OSSSYS, 0, mmIH_RB_RPTR_RING1, 0);
WREG32_SOC15(OSSSYS, 0, mmIH_RB_WPTR_RING1, 0);
adev->irq.ih1.enabled = false;
adev->irq.ih1.rptr = 0;
}
if (adev->irq.ih2.ring_size) {
ih_rb_cntl = RREG32_SOC15(OSSSYS, 0, mmIH_RB_CNTL_RING2);
ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL_RING2,
RB_ENABLE, 0);
WREG32_SOC15(OSSSYS, 0, mmIH_RB_CNTL_RING2, ih_rb_cntl);
/* set rptr, wptr to 0 */
WREG32_SOC15(OSSSYS, 0, mmIH_RB_RPTR_RING2, 0);
WREG32_SOC15(OSSSYS, 0, mmIH_RB_WPTR_RING2, 0);
adev->irq.ih2.enabled = false;
adev->irq.ih2.rptr = 0;
}
}
static uint32_t vega10_ih_rb_cntl(struct amdgpu_ih_ring *ih, uint32_t ih_rb_cntl)
{
int rb_bufsz = order_base_2(ih->ring_size / 4);
ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL,
MC_SPACE, ih->use_bus_addr ? 1 : 4);
ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL,
WPTR_OVERFLOW_CLEAR, 1);
ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL,
WPTR_OVERFLOW_ENABLE, 1);
ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, RB_SIZE, rb_bufsz);
/* Ring Buffer write pointer writeback. If enabled, IH_RB_WPTR register
* value is written to memory
*/
ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL,
WPTR_WRITEBACK_ENABLE, 1);
ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, MC_SNOOP, 1);
ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, MC_RO, 0);
ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, MC_VMID, 0);
return ih_rb_cntl;
}
/**
@ -86,50 +149,32 @@ static void vega10_ih_disable_interrupts(struct amdgpu_device *adev)
*/
static int vega10_ih_irq_init(struct amdgpu_device *adev)
{
struct amdgpu_ih_ring *ih;
int ret = 0;
int rb_bufsz;
u32 ih_rb_cntl, ih_doorbell_rtpr;
u32 tmp;
u64 wptr_off;
/* disable irqs */
vega10_ih_disable_interrupts(adev);
adev->nbio_funcs->ih_control(adev);
ih_rb_cntl = RREG32_SOC15(OSSSYS, 0, mmIH_RB_CNTL);
ih = &adev->irq.ih;
/* Ring Buffer base. [39:8] of 40-bit address of the beginning of the ring buffer*/
if (adev->irq.ih.use_bus_addr) {
WREG32_SOC15(OSSSYS, 0, mmIH_RB_BASE, adev->irq.ih.rb_dma_addr >> 8);
WREG32_SOC15(OSSSYS, 0, mmIH_RB_BASE_HI, ((u64)adev->irq.ih.rb_dma_addr >> 40) & 0xff);
ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, MC_SPACE, 1);
} else {
WREG32_SOC15(OSSSYS, 0, mmIH_RB_BASE, adev->irq.ih.gpu_addr >> 8);
WREG32_SOC15(OSSSYS, 0, mmIH_RB_BASE_HI, (adev->irq.ih.gpu_addr >> 40) & 0xff);
ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, MC_SPACE, 4);
}
rb_bufsz = order_base_2(adev->irq.ih.ring_size / 4);
ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, WPTR_OVERFLOW_CLEAR, 1);
ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, WPTR_OVERFLOW_ENABLE, 1);
ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, RB_SIZE, rb_bufsz);
/* Ring Buffer write pointer writeback. If enabled, IH_RB_WPTR register value is written to memory */
ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, WPTR_WRITEBACK_ENABLE, 1);
ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, MC_SNOOP, 1);
ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, MC_RO, 0);
ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, MC_VMID, 0);
if (adev->irq.msi_enabled)
ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, RPTR_REARM, 1);
WREG32_SOC15(OSSSYS, 0, mmIH_RB_BASE, ih->gpu_addr >> 8);
WREG32_SOC15(OSSSYS, 0, mmIH_RB_BASE_HI, (ih->gpu_addr >> 40) & 0xff);
ih_rb_cntl = RREG32_SOC15(OSSSYS, 0, mmIH_RB_CNTL);
ih_rb_cntl = vega10_ih_rb_cntl(ih, ih_rb_cntl);
ih_rb_cntl = REG_SET_FIELD(ih_rb_cntl, IH_RB_CNTL, RPTR_REARM,
!!adev->irq.msi_enabled);
WREG32_SOC15(OSSSYS, 0, mmIH_RB_CNTL, ih_rb_cntl);
/* set the writeback address whether it's enabled or not */
if (adev->irq.ih.use_bus_addr)
wptr_off = adev->irq.ih.rb_dma_addr + (adev->irq.ih.wptr_offs * 4);
else
wptr_off = adev->wb.gpu_addr + (adev->irq.ih.wptr_offs * 4);
WREG32_SOC15(OSSSYS, 0, mmIH_RB_WPTR_ADDR_LO, lower_32_bits(wptr_off));
WREG32_SOC15(OSSSYS, 0, mmIH_RB_WPTR_ADDR_HI, upper_32_bits(wptr_off) & 0xFFFF);
WREG32_SOC15(OSSSYS, 0, mmIH_RB_WPTR_ADDR_LO,
lower_32_bits(ih->wptr_addr));
WREG32_SOC15(OSSSYS, 0, mmIH_RB_WPTR_ADDR_HI,
upper_32_bits(ih->wptr_addr) & 0xFFFF);
/* set rptr, wptr to 0 */
WREG32_SOC15(OSSSYS, 0, mmIH_RB_RPTR, 0);
@ -137,17 +182,48 @@ static int vega10_ih_irq_init(struct amdgpu_device *adev)
ih_doorbell_rtpr = RREG32_SOC15(OSSSYS, 0, mmIH_DOORBELL_RPTR);
if (adev->irq.ih.use_doorbell) {
ih_doorbell_rtpr = REG_SET_FIELD(ih_doorbell_rtpr, IH_DOORBELL_RPTR,
OFFSET, adev->irq.ih.doorbell_index);
ih_doorbell_rtpr = REG_SET_FIELD(ih_doorbell_rtpr, IH_DOORBELL_RPTR,
ih_doorbell_rtpr = REG_SET_FIELD(ih_doorbell_rtpr,
IH_DOORBELL_RPTR, OFFSET,
adev->irq.ih.doorbell_index);
ih_doorbell_rtpr = REG_SET_FIELD(ih_doorbell_rtpr,
IH_DOORBELL_RPTR,
ENABLE, 1);
} else {
ih_doorbell_rtpr = REG_SET_FIELD(ih_doorbell_rtpr, IH_DOORBELL_RPTR,
ih_doorbell_rtpr = REG_SET_FIELD(ih_doorbell_rtpr,
IH_DOORBELL_RPTR,
ENABLE, 0);
}
WREG32_SOC15(OSSSYS, 0, mmIH_DOORBELL_RPTR, ih_doorbell_rtpr);
adev->nbio_funcs->ih_doorbell_range(adev, adev->irq.ih.use_doorbell,
adev->irq.ih.doorbell_index);
ih = &adev->irq.ih1;
if (ih->ring_size) {
WREG32_SOC15(OSSSYS, 0, mmIH_RB_BASE_RING1, ih->gpu_addr >> 8);
WREG32_SOC15(OSSSYS, 0, mmIH_RB_BASE_HI_RING1,
(ih->gpu_addr >> 40) & 0xff);
ih_rb_cntl = RREG32_SOC15(OSSSYS, 0, mmIH_RB_CNTL_RING1);
ih_rb_cntl = vega10_ih_rb_cntl(ih, ih_rb_cntl);
WREG32_SOC15(OSSSYS, 0, mmIH_RB_CNTL_RING1, ih_rb_cntl);
/* set rptr, wptr to 0 */
WREG32_SOC15(OSSSYS, 0, mmIH_RB_RPTR_RING1, 0);
WREG32_SOC15(OSSSYS, 0, mmIH_RB_WPTR_RING1, 0);
}
ih = &adev->irq.ih2;
if (ih->ring_size) {
WREG32_SOC15(OSSSYS, 0, mmIH_RB_BASE_RING2, ih->gpu_addr >> 8);
WREG32_SOC15(OSSSYS, 0, mmIH_RB_BASE_HI_RING2,
(ih->gpu_addr >> 40) & 0xff);
ih_rb_cntl = RREG32_SOC15(OSSSYS, 0, mmIH_RB_CNTL_RING1);
ih_rb_cntl = vega10_ih_rb_cntl(ih, ih_rb_cntl);
WREG32_SOC15(OSSSYS, 0, mmIH_RB_CNTL_RING2, ih_rb_cntl);
/* set rptr, wptr to 0 */
WREG32_SOC15(OSSSYS, 0, mmIH_RB_RPTR_RING2, 0);
WREG32_SOC15(OSSSYS, 0, mmIH_RB_WPTR_RING2, 0);
}
tmp = RREG32_SOC15(OSSSYS, 0, mmIH_STORM_CLIENT_LIST_CNTL);
tmp = REG_SET_FIELD(tmp, IH_STORM_CLIENT_LIST_CNTL,
@ -191,32 +267,58 @@ static void vega10_ih_irq_disable(struct amdgpu_device *adev)
* ring buffer overflow and deal with it.
* Returns the value of the wptr.
*/
static u32 vega10_ih_get_wptr(struct amdgpu_device *adev)
static u32 vega10_ih_get_wptr(struct amdgpu_device *adev,
struct amdgpu_ih_ring *ih)
{
u32 wptr, tmp;
u32 wptr, reg, tmp;
if (adev->irq.ih.use_bus_addr)
wptr = le32_to_cpu(adev->irq.ih.ring[adev->irq.ih.wptr_offs]);
wptr = le32_to_cpu(*ih->wptr_cpu);
if (!REG_GET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW))
goto out;
/* Double check that the overflow wasn't already cleared. */
if (ih == &adev->irq.ih)
reg = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_WPTR);
else if (ih == &adev->irq.ih1)
reg = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_WPTR_RING1);
else if (ih == &adev->irq.ih2)
reg = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_WPTR_RING2);
else
wptr = le32_to_cpu(adev->wb.wb[adev->irq.ih.wptr_offs]);
BUG();
if (REG_GET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW)) {
wptr = REG_SET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW, 0);
wptr = RREG32_NO_KIQ(reg);
if (!REG_GET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW))
goto out;
/* When a ring buffer overflow happen start parsing interrupt
* from the last not overwritten vector (wptr + 32). Hopefully
* this should allow us to catchup.
*/
tmp = (wptr + 32) & adev->irq.ih.ptr_mask;
dev_warn(adev->dev, "IH ring buffer overflow (0x%08X, 0x%08X, 0x%08X)\n",
wptr, adev->irq.ih.rptr, tmp);
adev->irq.ih.rptr = tmp;
wptr = REG_SET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW, 0);
tmp = RREG32_NO_KIQ(SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_CNTL));
tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, WPTR_OVERFLOW_CLEAR, 1);
WREG32_NO_KIQ(SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_CNTL), tmp);
}
return (wptr & adev->irq.ih.ptr_mask);
/* When a ring buffer overflow happen start parsing interrupt
* from the last not overwritten vector (wptr + 32). Hopefully
* this should allow us to catchup.
*/
tmp = (wptr + 32) & ih->ptr_mask;
dev_warn(adev->dev, "IH ring buffer overflow "
"(0x%08X, 0x%08X, 0x%08X)\n",
wptr, ih->rptr, tmp);
ih->rptr = tmp;
if (ih == &adev->irq.ih)
reg = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_CNTL);
else if (ih == &adev->irq.ih1)
reg = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_CNTL_RING1);
else if (ih == &adev->irq.ih2)
reg = SOC15_REG_OFFSET(OSSSYS, 0, mmIH_RB_CNTL_RING2);
else
BUG();
tmp = RREG32_NO_KIQ(reg);
tmp = REG_SET_FIELD(tmp, IH_RB_CNTL, WPTR_OVERFLOW_CLEAR, 1);
WREG32_NO_KIQ(reg, tmp);
out:
return (wptr & ih->ptr_mask);
}
/**
@ -228,20 +330,21 @@ static u32 vega10_ih_get_wptr(struct amdgpu_device *adev)
* position and also advance the position.
*/
static void vega10_ih_decode_iv(struct amdgpu_device *adev,
struct amdgpu_iv_entry *entry)
struct amdgpu_ih_ring *ih,
struct amdgpu_iv_entry *entry)
{
/* wptr/rptr are in bytes! */
u32 ring_index = adev->irq.ih.rptr >> 2;
u32 ring_index = ih->rptr >> 2;
uint32_t dw[8];
dw[0] = le32_to_cpu(adev->irq.ih.ring[ring_index + 0]);
dw[1] = le32_to_cpu(adev->irq.ih.ring[ring_index + 1]);
dw[2] = le32_to_cpu(adev->irq.ih.ring[ring_index + 2]);
dw[3] = le32_to_cpu(adev->irq.ih.ring[ring_index + 3]);
dw[4] = le32_to_cpu(adev->irq.ih.ring[ring_index + 4]);
dw[5] = le32_to_cpu(adev->irq.ih.ring[ring_index + 5]);
dw[6] = le32_to_cpu(adev->irq.ih.ring[ring_index + 6]);
dw[7] = le32_to_cpu(adev->irq.ih.ring[ring_index + 7]);
dw[0] = le32_to_cpu(ih->ring[ring_index + 0]);
dw[1] = le32_to_cpu(ih->ring[ring_index + 1]);
dw[2] = le32_to_cpu(ih->ring[ring_index + 2]);
dw[3] = le32_to_cpu(ih->ring[ring_index + 3]);
dw[4] = le32_to_cpu(ih->ring[ring_index + 4]);
dw[5] = le32_to_cpu(ih->ring[ring_index + 5]);
dw[6] = le32_to_cpu(ih->ring[ring_index + 6]);
dw[7] = le32_to_cpu(ih->ring[ring_index + 7]);
entry->client_id = dw[0] & 0xff;
entry->src_id = (dw[0] >> 8) & 0xff;
@ -257,9 +360,8 @@ static void vega10_ih_decode_iv(struct amdgpu_device *adev,
entry->src_data[2] = dw[6];
entry->src_data[3] = dw[7];
/* wptr/rptr are in bytes! */
adev->irq.ih.rptr += 32;
ih->rptr += 32;
}
/**
@ -269,37 +371,95 @@ static void vega10_ih_decode_iv(struct amdgpu_device *adev,
*
* Set the IH ring buffer rptr.
*/
static void vega10_ih_set_rptr(struct amdgpu_device *adev)
static void vega10_ih_set_rptr(struct amdgpu_device *adev,
struct amdgpu_ih_ring *ih)
{
if (adev->irq.ih.use_doorbell) {
if (ih->use_doorbell) {
/* XXX check if swapping is necessary on BE */
if (adev->irq.ih.use_bus_addr)
adev->irq.ih.ring[adev->irq.ih.rptr_offs] = adev->irq.ih.rptr;
else
adev->wb.wb[adev->irq.ih.rptr_offs] = adev->irq.ih.rptr;
WDOORBELL32(adev->irq.ih.doorbell_index, adev->irq.ih.rptr);
} else {
WREG32_SOC15(OSSSYS, 0, mmIH_RB_RPTR, adev->irq.ih.rptr);
*ih->rptr_cpu = ih->rptr;
WDOORBELL32(ih->doorbell_index, ih->rptr);
} else if (ih == &adev->irq.ih) {
WREG32_SOC15(OSSSYS, 0, mmIH_RB_RPTR, ih->rptr);
} else if (ih == &adev->irq.ih1) {
WREG32_SOC15(OSSSYS, 0, mmIH_RB_RPTR_RING1, ih->rptr);
} else if (ih == &adev->irq.ih2) {
WREG32_SOC15(OSSSYS, 0, mmIH_RB_RPTR_RING2, ih->rptr);
}
}
/**
* vega10_ih_self_irq - dispatch work for ring 1 and 2
*
* @adev: amdgpu_device pointer
* @source: irq source
* @entry: IV with WPTR update
*
* Update the WPTR from the IV and schedule work to handle the entries.
*/
static int vega10_ih_self_irq(struct amdgpu_device *adev,
struct amdgpu_irq_src *source,
struct amdgpu_iv_entry *entry)
{
uint32_t wptr = cpu_to_le32(entry->src_data[0]);
switch (entry->ring_id) {
case 1:
*adev->irq.ih1.wptr_cpu = wptr;
schedule_work(&adev->irq.ih1_work);
break;
case 2:
*adev->irq.ih2.wptr_cpu = wptr;
schedule_work(&adev->irq.ih2_work);
break;
default: break;
}
return 0;
}
static const struct amdgpu_irq_src_funcs vega10_ih_self_irq_funcs = {
.process = vega10_ih_self_irq,
};
static void vega10_ih_set_self_irq_funcs(struct amdgpu_device *adev)
{
adev->irq.self_irq.num_types = 0;
adev->irq.self_irq.funcs = &vega10_ih_self_irq_funcs;
}
static int vega10_ih_early_init(void *handle)
{
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
vega10_ih_set_interrupt_funcs(adev);
vega10_ih_set_self_irq_funcs(adev);
return 0;
}
static int vega10_ih_sw_init(void *handle)
{
int r;
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
int r;
r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_IH, 0,
&adev->irq.self_irq);
if (r)
return r;
r = amdgpu_ih_ring_init(adev, &adev->irq.ih, 256 * 1024, true);
if (r)
return r;
if (adev->asic_type == CHIP_VEGA10) {
r = amdgpu_ih_ring_init(adev, &adev->irq.ih1, PAGE_SIZE, true);
if (r)
return r;
r = amdgpu_ih_ring_init(adev, &adev->irq.ih2, PAGE_SIZE, true);
if (r)
return r;
}
/* TODO add doorbell for IH1 & IH2 as well */
adev->irq.ih.use_doorbell = true;
adev->irq.ih.doorbell_index = adev->doorbell_index.ih << 1;
@ -313,6 +473,8 @@ static int vega10_ih_sw_fini(void *handle)
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
amdgpu_irq_fini(adev);
amdgpu_ih_ring_fini(adev, &adev->irq.ih2);
amdgpu_ih_ring_fini(adev, &adev->irq.ih1);
amdgpu_ih_ring_fini(adev, &adev->irq.ih);
return 0;

5
drivers/gpu/drm/amd/amdgpu/vega10_reg_init.c
View File

@ -70,8 +70,8 @@ void vega10_doorbell_index_init(struct amdgpu_device *adev)
adev->doorbell_index.userqueue_start = AMDGPU_DOORBELL64_USERQUEUE_START;
adev->doorbell_index.userqueue_end = AMDGPU_DOORBELL64_USERQUEUE_END;
adev->doorbell_index.gfx_ring0 = AMDGPU_DOORBELL64_GFX_RING0;
adev->doorbell_index.sdma_engine0 = AMDGPU_DOORBELL64_sDMA_ENGINE0;
adev->doorbell_index.sdma_engine1 = AMDGPU_DOORBELL64_sDMA_ENGINE1;
adev->doorbell_index.sdma_engine[0] = AMDGPU_DOORBELL64_sDMA_ENGINE0;
adev->doorbell_index.sdma_engine[1] = AMDGPU_DOORBELL64_sDMA_ENGINE1;
adev->doorbell_index.ih = AMDGPU_DOORBELL64_IH;
adev->doorbell_index.uvd_vce.uvd_ring0_1 = AMDGPU_DOORBELL64_UVD_RING0_1;
adev->doorbell_index.uvd_vce.uvd_ring2_3 = AMDGPU_DOORBELL64_UVD_RING2_3;
@ -83,5 +83,6 @@ void vega10_doorbell_index_init(struct amdgpu_device *adev)
adev->doorbell_index.uvd_vce.vce_ring6_7 = AMDGPU_DOORBELL64_VCE_RING6_7;
/* In unit of dword doorbell */
adev->doorbell_index.max_assignment = AMDGPU_DOORBELL64_MAX_ASSIGNMENT << 1;
adev->doorbell_index.sdma_doorbell_range = 4;
}

17
drivers/gpu/drm/amd/amdgpu/vega20_reg_init.c
View File

@ -68,14 +68,14 @@ void vega20_doorbell_index_init(struct amdgpu_device *adev)
adev->doorbell_index.userqueue_start = AMDGPU_VEGA20_DOORBELL_USERQUEUE_START;
adev->doorbell_index.userqueue_end = AMDGPU_VEGA20_DOORBELL_USERQUEUE_END;
adev->doorbell_index.gfx_ring0 = AMDGPU_VEGA20_DOORBELL_GFX_RING0;
adev->doorbell_index.sdma_engine0 = AMDGPU_VEGA20_DOORBELL_sDMA_ENGINE0;
adev->doorbell_index.sdma_engine1 = AMDGPU_VEGA20_DOORBELL_sDMA_ENGINE1;
adev->doorbell_index.sdma_engine2 = AMDGPU_VEGA20_DOORBELL_sDMA_ENGINE2;
adev->doorbell_index.sdma_engine3 = AMDGPU_VEGA20_DOORBELL_sDMA_ENGINE3;
adev->doorbell_index.sdma_engine4 = AMDGPU_VEGA20_DOORBELL_sDMA_ENGINE4;
adev->doorbell_index.sdma_engine5 = AMDGPU_VEGA20_DOORBELL_sDMA_ENGINE5;
adev->doorbell_index.sdma_engine6 = AMDGPU_VEGA20_DOORBELL_sDMA_ENGINE6;
adev->doorbell_index.sdma_engine7 = AMDGPU_VEGA20_DOORBELL_sDMA_ENGINE7;
adev->doorbell_index.sdma_engine[0] = AMDGPU_VEGA20_DOORBELL_sDMA_ENGINE0;
adev->doorbell_index.sdma_engine[1] = AMDGPU_VEGA20_DOORBELL_sDMA_ENGINE1;
adev->doorbell_index.sdma_engine[2] = AMDGPU_VEGA20_DOORBELL_sDMA_ENGINE2;
adev->doorbell_index.sdma_engine[3] = AMDGPU_VEGA20_DOORBELL_sDMA_ENGINE3;
adev->doorbell_index.sdma_engine[4] = AMDGPU_VEGA20_DOORBELL_sDMA_ENGINE4;
adev->doorbell_index.sdma_engine[5] = AMDGPU_VEGA20_DOORBELL_sDMA_ENGINE5;
adev->doorbell_index.sdma_engine[6] = AMDGPU_VEGA20_DOORBELL_sDMA_ENGINE6;
adev->doorbell_index.sdma_engine[7] = AMDGPU_VEGA20_DOORBELL_sDMA_ENGINE7;
adev->doorbell_index.ih = AMDGPU_VEGA20_DOORBELL_IH;
adev->doorbell_index.uvd_vce.uvd_ring0_1 = AMDGPU_VEGA20_DOORBELL64_UVD_RING0_1;
adev->doorbell_index.uvd_vce.uvd_ring2_3 = AMDGPU_VEGA20_DOORBELL64_UVD_RING2_3;
@ -86,5 +86,6 @@ void vega20_doorbell_index_init(struct amdgpu_device *adev)
adev->doorbell_index.uvd_vce.vce_ring4_5 = AMDGPU_VEGA20_DOORBELL64_VCE_RING4_5;
adev->doorbell_index.uvd_vce.vce_ring6_7 = AMDGPU_VEGA20_DOORBELL64_VCE_RING6_7;
adev->doorbell_index.max_assignment = AMDGPU_VEGA20_DOORBELL_MAX_ASSIGNMENT << 1;
adev->doorbell_index.sdma_doorbell_range = 20;
}

69
drivers/gpu/drm/amd/amdgpu/vi.c
View File

@ -941,6 +941,69 @@ static bool vi_need_full_reset(struct amdgpu_device *adev)
}
}
static void vi_get_pcie_usage(struct amdgpu_device *adev, uint64_t *count0,
uint64_t *count1)
{
uint32_t perfctr = 0;
uint64_t cnt0_of, cnt1_of;
int tmp;
/* This reports 0 on APUs, so return to avoid writing/reading registers
* that may or may not be different from their GPU counterparts
*/
if (adev->flags & AMD_IS_APU)
return;
/* Set the 2 events that we wish to watch, defined above */
/* Reg 40 is # received msgs, Reg 104 is # of posted requests sent */
perfctr = REG_SET_FIELD(perfctr, PCIE_PERF_CNTL_TXCLK, EVENT0_SEL, 40);
perfctr = REG_SET_FIELD(perfctr, PCIE_PERF_CNTL_TXCLK, EVENT1_SEL, 104);
/* Write to enable desired perf counters */
WREG32_PCIE(ixPCIE_PERF_CNTL_TXCLK, perfctr);
/* Zero out and enable the perf counters
* Write 0x5:
* Bit 0 = Start all counters(1)
* Bit 2 = Global counter reset enable(1)
*/
WREG32_PCIE(ixPCIE_PERF_COUNT_CNTL, 0x00000005);
msleep(1000);
/* Load the shadow and disable the perf counters
* Write 0x2:
* Bit 0 = Stop counters(0)
* Bit 1 = Load the shadow counters(1)
*/
WREG32_PCIE(ixPCIE_PERF_COUNT_CNTL, 0x00000002);
/* Read register values to get any >32bit overflow */
tmp = RREG32_PCIE(ixPCIE_PERF_CNTL_TXCLK);
cnt0_of = REG_GET_FIELD(tmp, PCIE_PERF_CNTL_TXCLK, COUNTER0_UPPER);
cnt1_of = REG_GET_FIELD(tmp, PCIE_PERF_CNTL_TXCLK, COUNTER1_UPPER);
/* Get the values and add the overflow */
*count0 = RREG32_PCIE(ixPCIE_PERF_COUNT0_TXCLK) | (cnt0_of << 32);
*count1 = RREG32_PCIE(ixPCIE_PERF_COUNT1_TXCLK) | (cnt1_of << 32);
}
static bool vi_need_reset_on_init(struct amdgpu_device *adev)
{
u32 clock_cntl, pc;
if (adev->flags & AMD_IS_APU)
return false;
/* check if the SMC is already running */
clock_cntl = RREG32_SMC(ixSMC_SYSCON_CLOCK_CNTL_0);
pc = RREG32_SMC(ixSMC_PC_C);
if ((0 == REG_GET_FIELD(clock_cntl, SMC_SYSCON_CLOCK_CNTL_0, ck_disable)) &&
(0x20100 <= pc))
return true;
return false;
}
static const struct amdgpu_asic_funcs vi_asic_funcs =
{
.read_disabled_bios = &vi_read_disabled_bios,
@ -956,6 +1019,8 @@ static const struct amdgpu_asic_funcs vi_asic_funcs =
.invalidate_hdp = &vi_invalidate_hdp,
.need_full_reset = &vi_need_full_reset,
.init_doorbell_index = &legacy_doorbell_index_init,
.get_pcie_usage = &vi_get_pcie_usage,
.need_reset_on_init = &vi_need_reset_on_init,
};
#define CZ_REV_BRISTOL(rev) \
@ -1726,8 +1791,8 @@ void legacy_doorbell_index_init(struct amdgpu_device *adev)
adev->doorbell_index.mec_ring6 = AMDGPU_DOORBELL_MEC_RING6;
adev->doorbell_index.mec_ring7 = AMDGPU_DOORBELL_MEC_RING7;
adev->doorbell_index.gfx_ring0 = AMDGPU_DOORBELL_GFX_RING0;
adev->doorbell_index.sdma_engine0 = AMDGPU_DOORBELL_sDMA_ENGINE0;
adev->doorbell_index.sdma_engine1 = AMDGPU_DOORBELL_sDMA_ENGINE1;
adev->doorbell_index.sdma_engine[0] = AMDGPU_DOORBELL_sDMA_ENGINE0;
adev->doorbell_index.sdma_engine[1] = AMDGPU_DOORBELL_sDMA_ENGINE1;
adev->doorbell_index.ih = AMDGPU_DOORBELL_IH;
adev->doorbell_index.max_assignment = AMDGPU_DOORBELL_MAX_ASSIGNMENT;
}

4
drivers/gpu/drm/amd/amdkfd/Kconfig
View File

@ -4,8 +4,8 @@
config HSA_AMD
bool "HSA kernel driver for AMD GPU devices"
depends on DRM_AMDGPU && X86_64
imply AMD_IOMMU_V2
depends on DRM_AMDGPU && (X86_64 || ARM64)
imply AMD_IOMMU_V2 if X86_64
select MMU_NOTIFIER
help
Enable this if you want to use HSA features on AMD GPU devices.

8
drivers/gpu/drm/amd/amdkfd/kfd_crat.c
View File

@ -863,6 +863,7 @@ static int kfd_fill_mem_info_for_cpu(int numa_node_id, int *avail_size,
return 0;
}
#if CONFIG_X86_64
static int kfd_fill_iolink_info_for_cpu(int numa_node_id, int *avail_size,
uint32_t *num_entries,
struct crat_subtype_iolink *sub_type_hdr)
@ -905,6 +906,7 @@ static int kfd_fill_iolink_info_for_cpu(int numa_node_id, int *avail_size,
return 0;
}
#endif
/* kfd_create_vcrat_image_cpu - Create Virtual CRAT for CPU
*
@ -920,7 +922,9 @@ static int kfd_create_vcrat_image_cpu(void *pcrat_image, size_t *size)
struct crat_subtype_generic *sub_type_hdr;
int avail_size = *size;
int numa_node_id;
#ifdef CONFIG_X86_64
uint32_t entries = 0;
#endif
int ret = 0;
if (!pcrat_image || avail_size < VCRAT_SIZE_FOR_CPU)
@ -982,6 +986,7 @@ static int kfd_create_vcrat_image_cpu(void *pcrat_image, size_t *size)
sub_type_hdr->length);
/* Fill in Subtype: IO Link */
#ifdef CONFIG_X86_64
ret = kfd_fill_iolink_info_for_cpu(numa_node_id, &avail_size,
&entries,
(struct crat_subtype_iolink *)sub_type_hdr);
@ -992,6 +997,9 @@ static int kfd_create_vcrat_image_cpu(void *pcrat_image, size_t *size)
sub_type_hdr = (typeof(sub_type_hdr))((char *)sub_type_hdr +
sub_type_hdr->length * entries);
#else
pr_info("IO link not available for non x86 platforms\n");
#endif
crat_table->num_domains++;
}

31
drivers/gpu/drm/amd/amdkfd/kfd_module.c
View File

@ -23,22 +23,7 @@
#include <linux/sched.h>
#include <linux/device.h>
#include "kfd_priv.h"
static const struct kgd2kfd_calls kgd2kfd = {
.exit = kgd2kfd_exit,
.probe = kgd2kfd_probe,
.device_init = kgd2kfd_device_init,
.device_exit = kgd2kfd_device_exit,
.interrupt = kgd2kfd_interrupt,
.suspend = kgd2kfd_suspend,
.resume = kgd2kfd_resume,
.quiesce_mm = kgd2kfd_quiesce_mm,
.resume_mm = kgd2kfd_resume_mm,
.schedule_evict_and_restore_process =
kgd2kfd_schedule_evict_and_restore_process,
.pre_reset = kgd2kfd_pre_reset,
.post_reset = kgd2kfd_post_reset,
};
#include "amdgpu_amdkfd.h"
static int kfd_init(void)
{
@ -91,20 +76,10 @@ static void kfd_exit(void)
kfd_chardev_exit();
}
int kgd2kfd_init(unsigned int interface_version,
const struct kgd2kfd_calls **g2f)
int kgd2kfd_init()
{
int err;
err = kfd_init();
if (err)
return err;
*g2f = &kgd2kfd;
return 0;
return kfd_init();
}
EXPORT_SYMBOL(kgd2kfd_init);
void kgd2kfd_exit(void)
{

22
drivers/gpu/drm/amd/amdkfd/kfd_priv.h
View File

@ -266,14 +266,6 @@ struct kfd_dev {
bool pci_atomic_requested;
};
/* KGD2KFD callbacks */
void kgd2kfd_exit(void);
struct kfd_dev *kgd2kfd_probe(struct kgd_dev *kgd,
struct pci_dev *pdev, const struct kfd2kgd_calls *f2g);
bool kgd2kfd_device_init(struct kfd_dev *kfd,
const struct kgd2kfd_shared_resources *gpu_resources);
void kgd2kfd_device_exit(struct kfd_dev *kfd);
enum kfd_mempool {
KFD_MEMPOOL_SYSTEM_CACHEABLE = 1,
KFD_MEMPOOL_SYSTEM_WRITECOMBINE = 2,
@ -541,11 +533,6 @@ struct qcm_process_device {
/* Approx. time before evicting the process again */
#define PROCESS_ACTIVE_TIME_MS 10
int kgd2kfd_quiesce_mm(struct mm_struct *mm);
int kgd2kfd_resume_mm(struct mm_struct *mm);
int kgd2kfd_schedule_evict_and_restore_process(struct mm_struct *mm,
struct dma_fence *fence);
/* 8 byte handle containing GPU ID in the most significant 4 bytes and
* idr_handle in the least significant 4 bytes
*/
@ -800,20 +787,11 @@ int kfd_numa_node_to_apic_id(int numa_node_id);
/* Interrupts */
int kfd_interrupt_init(struct kfd_dev *dev);
void kfd_interrupt_exit(struct kfd_dev *dev);
void kgd2kfd_interrupt(struct kfd_dev *kfd, const void *ih_ring_entry);
bool enqueue_ih_ring_entry(struct kfd_dev *kfd, const void *ih_ring_entry);
bool interrupt_is_wanted(struct kfd_dev *dev,
const uint32_t *ih_ring_entry,
uint32_t *patched_ihre, bool *flag);
/* Power Management */
void kgd2kfd_suspend(struct kfd_dev *kfd);
int kgd2kfd_resume(struct kfd_dev *kfd);
/* GPU reset */
int kgd2kfd_pre_reset(struct kfd_dev *kfd);
int kgd2kfd_post_reset(struct kfd_dev *kfd);
/* amdkfd Apertures */
int kfd_init_apertures(struct kfd_process *process);

21
drivers/gpu/drm/amd/amdkfd/kfd_topology.c
View File

@ -1093,8 +1093,6 @@ static uint32_t kfd_generate_gpu_id(struct kfd_dev *gpu)
* the GPU device is not already present in the topology device
* list then return NULL. This means a new topology device has to
* be created for this GPU.
* TODO: Rather than assiging @gpu to first topology device withtout
* gpu attached, it will better to have more stringent check.
*/
static struct kfd_topology_device *kfd_assign_gpu(struct kfd_dev *gpu)
{
@ -1102,12 +1100,20 @@ static struct kfd_topology_device *kfd_assign_gpu(struct kfd_dev *gpu)
struct kfd_topology_device *out_dev = NULL;
down_write(&topology_lock);
list_for_each_entry(dev, &topology_device_list, list)
list_for_each_entry(dev, &topology_device_list, list) {
/* Discrete GPUs need their own topology device list
* entries. Don't assign them to CPU/APU nodes.
*/
if (!gpu->device_info->needs_iommu_device &&
dev->node_props.cpu_cores_count)
continue;
if (!dev->gpu && (dev->node_props.simd_count > 0)) {
dev->gpu = gpu;
out_dev = dev;
break;
}
}
up_write(&topology_lock);
return out_dev;
}
@ -1392,7 +1398,6 @@ int kfd_topology_enum_kfd_devices(uint8_t idx, struct kfd_dev **kdev)
static int kfd_cpumask_to_apic_id(const struct cpumask *cpumask)
{
const struct cpuinfo_x86 *cpuinfo;
int first_cpu_of_numa_node;
if (!cpumask || cpumask == cpu_none_mask)
@ -1400,9 +1405,11 @@ static int kfd_cpumask_to_apic_id(const struct cpumask *cpumask)
first_cpu_of_numa_node = cpumask_first(cpumask);
if (first_cpu_of_numa_node >= nr_cpu_ids)
return -1;
cpuinfo = &cpu_data(first_cpu_of_numa_node);
return cpuinfo->apicid;
#ifdef CONFIG_X86_64
return cpu_data(first_cpu_of_numa_node).apicid;
#else
return first_cpu_of_numa_node;
#endif
}
/* kfd_numa_node_to_apic_id - Returns the APIC ID of the first logical processor

1483
drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
View File

File diff suppressed because it is too large Load Diff

48
drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crc.c
View File

@ -64,8 +64,10 @@ amdgpu_dm_crtc_verify_crc_source(struct drm_crtc *crtc, const char *src_name,
int amdgpu_dm_crtc_set_crc_source(struct drm_crtc *crtc, const char *src_name)
{
struct amdgpu_device *adev = crtc->dev->dev_private;
struct dm_crtc_state *crtc_state = to_dm_crtc_state(crtc->state);
struct dc_stream_state *stream_state = crtc_state->stream;
bool enable;
enum amdgpu_dm_pipe_crc_source source = dm_parse_crc_source(src_name);
@ -80,29 +82,33 @@ int amdgpu_dm_crtc_set_crc_source(struct drm_crtc *crtc, const char *src_name)
return -EINVAL;
}
/* When enabling CRC, we should also disable dithering. */
if (source == AMDGPU_DM_PIPE_CRC_SOURCE_AUTO) {
if (dc_stream_configure_crc(stream_state->ctx->dc,
stream_state,
true, true)) {
crtc_state->crc_enabled = true;
dc_stream_set_dither_option(stream_state,
DITHER_OPTION_TRUN8);
}
else
return -EINVAL;
} else {
if (dc_stream_configure_crc(stream_state->ctx->dc,
stream_state,
false, false)) {
crtc_state->crc_enabled = false;
dc_stream_set_dither_option(stream_state,
DITHER_OPTION_DEFAULT);
}
else
return -EINVAL;
enable = (source == AMDGPU_DM_PIPE_CRC_SOURCE_AUTO);
mutex_lock(&adev->dm.dc_lock);
if (!dc_stream_configure_crc(stream_state->ctx->dc, stream_state,
enable, enable)) {
mutex_unlock(&adev->dm.dc_lock);
return -EINVAL;
}
/* When enabling CRC, we should also disable dithering. */
dc_stream_set_dither_option(stream_state,
enable ? DITHER_OPTION_TRUN8
: DITHER_OPTION_DEFAULT);
mutex_unlock(&adev->dm.dc_lock);
/*
* Reading the CRC requires the vblank interrupt handler to be
* enabled. Keep a reference until CRC capture stops.
*/
if (!crtc_state->crc_enabled && enable)
drm_crtc_vblank_get(crtc);
else if (crtc_state->crc_enabled && !enable)
drm_crtc_vblank_put(crtc);
crtc_state->crc_enabled = enable;
/* Reset crc_skipped on dm state */
crtc_state->crc_skip_count = 0;
return 0;

49
drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_debugfs.c
View File

@ -783,6 +783,45 @@ static ssize_t dtn_log_write(
return size;
}
/*
* Backlight at this moment. Read only.
* As written to display, taking ABM and backlight lut into account.
* Ranges from 0x0 to 0x10000 (= 100% PWM)
*/
static int current_backlight_read(struct seq_file *m, void *data)
{
struct drm_info_node *node = (struct drm_info_node *)m->private;
struct drm_device *dev = node->minor->dev;
struct amdgpu_device *adev = dev->dev_private;
struct dc *dc = adev->dm.dc;
unsigned int backlight = dc_get_current_backlight_pwm(dc);
seq_printf(m, "0x%x\n", backlight);
return 0;
}
/*
* Backlight value that is being approached. Read only.
* As written to display, taking ABM and backlight lut into account.
* Ranges from 0x0 to 0x10000 (= 100% PWM)
*/
static int target_backlight_read(struct seq_file *m, void *data)
{
struct drm_info_node *node = (struct drm_info_node *)m->private;
struct drm_device *dev = node->minor->dev;
struct amdgpu_device *adev = dev->dev_private;
struct dc *dc = adev->dm.dc;
unsigned int backlight = dc_get_target_backlight_pwm(dc);
seq_printf(m, "0x%x\n", backlight);
return 0;
}
static const struct drm_info_list amdgpu_dm_debugfs_list[] = {
{"amdgpu_current_backlight_pwm", &current_backlight_read},
{"amdgpu_target_backlight_pwm", &target_backlight_read},
};
int dtn_debugfs_init(struct amdgpu_device *adev)
{
static const struct file_operations dtn_log_fops = {
@ -793,9 +832,15 @@ int dtn_debugfs_init(struct amdgpu_device *adev)
};
struct drm_minor *minor = adev->ddev->primary;
struct dentry *root = minor->debugfs_root;
struct dentry *ent, *root = minor->debugfs_root;
int ret;
struct dentry *ent = debugfs_create_file(
ret = amdgpu_debugfs_add_files(adev, amdgpu_dm_debugfs_list,
ARRAY_SIZE(amdgpu_dm_debugfs_list));
if (ret)
return ret;
ent = debugfs_create_file(
"amdgpu_dm_dtn_log",
0644,
root,

15
drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_helpers.c
View File

@ -192,7 +192,7 @@ bool dm_helpers_dp_mst_write_payload_allocation_table(
int bpp = 0;
int pbn = 0;
aconnector = stream->sink->priv;
aconnector = (struct amdgpu_dm_connector *)stream->dm_stream_context;
if (!aconnector || !aconnector->mst_port)
return false;
@ -205,7 +205,7 @@ bool dm_helpers_dp_mst_write_payload_allocation_table(
mst_port = aconnector->port;
if (enable) {
clock = stream->timing.pix_clk_khz;
clock = stream->timing.pix_clk_100hz / 10;
switch (stream->timing.display_color_depth) {
@ -263,6 +263,13 @@ bool dm_helpers_dp_mst_write_payload_allocation_table(
return true;
}
/*
* poll pending down reply before clear payload allocation table
*/
void dm_helpers_dp_mst_poll_pending_down_reply(
struct dc_context *ctx,
const struct dc_link *link)
{}
/*
* Clear payload allocation table before enable MST DP link.
@ -284,7 +291,7 @@ bool dm_helpers_dp_mst_poll_for_allocation_change_trigger(
struct drm_dp_mst_topology_mgr *mst_mgr;
int ret;
aconnector = stream->sink->priv;
aconnector = (struct amdgpu_dm_connector *)stream->dm_stream_context;
if (!aconnector || !aconnector->mst_port)
return false;
@ -312,7 +319,7 @@ bool dm_helpers_dp_mst_send_payload_allocation(
struct drm_dp_mst_port *mst_port;
int ret;
aconnector = stream->sink->priv;
aconnector = (struct amdgpu_dm_connector *)stream->dm_stream_context;
if (!aconnector || !aconnector->mst_port)
return false;

85
drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.c
View File

@ -35,6 +35,8 @@
#include "dc_link_ddc.h"
#include "i2caux_interface.h"
/* #define TRACE_DPCD */
#ifdef TRACE_DPCD
@ -81,80 +83,24 @@ static ssize_t dm_dp_aux_transfer(struct drm_dp_aux *aux,
struct drm_dp_aux_msg *msg)
{
ssize_t result = 0;
enum i2caux_transaction_action action;
enum aux_transaction_type type;
struct aux_payload payload;
if (WARN_ON(msg->size > 16))
return -E2BIG;
switch (msg->request & ~DP_AUX_I2C_MOT) {
case DP_AUX_NATIVE_READ:
type = AUX_TRANSACTION_TYPE_DP;
action = I2CAUX_TRANSACTION_ACTION_DP_READ;
payload.address = msg->address;
payload.data = msg->buffer;
payload.length = msg->size;
payload.reply = &msg->reply;
payload.i2c_over_aux = (msg->request & DP_AUX_NATIVE_WRITE) == 0;
payload.write = (msg->request & DP_AUX_I2C_READ) == 0;
payload.mot = (msg->request & DP_AUX_I2C_MOT) != 0;
payload.defer_delay = 0;
result = dc_link_aux_transfer(TO_DM_AUX(aux)->ddc_service,
msg->address,
&msg->reply,
msg->buffer,
msg->size,
type,
action);
break;
case DP_AUX_NATIVE_WRITE:
type = AUX_TRANSACTION_TYPE_DP;
action = I2CAUX_TRANSACTION_ACTION_DP_WRITE;
result = dc_link_aux_transfer(TO_DM_AUX(aux)->ddc_service, &payload);
dc_link_aux_transfer(TO_DM_AUX(aux)->ddc_service,
msg->address,
&msg->reply,
msg->buffer,
msg->size,
type,
action);
if (payload.write)
result = msg->size;
break;
case DP_AUX_I2C_READ:
type = AUX_TRANSACTION_TYPE_I2C;
if (msg->request & DP_AUX_I2C_MOT)
action = I2CAUX_TRANSACTION_ACTION_I2C_READ_MOT;
else
action = I2CAUX_TRANSACTION_ACTION_I2C_READ;
result = dc_link_aux_transfer(TO_DM_AUX(aux)->ddc_service,
msg->address,
&msg->reply,
msg->buffer,
msg->size,
type,
action);
break;
case DP_AUX_I2C_WRITE:
type = AUX_TRANSACTION_TYPE_I2C;
if (msg->request & DP_AUX_I2C_MOT)
action = I2CAUX_TRANSACTION_ACTION_I2C_WRITE_MOT;
else
action = I2CAUX_TRANSACTION_ACTION_I2C_WRITE;
dc_link_aux_transfer(TO_DM_AUX(aux)->ddc_service,
msg->address,
&msg->reply,
msg->buffer,
msg->size,
type,
action);
result = msg->size;
break;
default:
return -EINVAL;
}
#ifdef TRACE_DPCD
log_dpcd(msg->request,
msg->address,
msg->buffer,
msg->size,
r == DDC_RESULT_SUCESSFULL);
#endif
if (result < 0) /* DC doesn't know about kernel error codes */
result = -EIO;
@ -228,6 +174,11 @@ static int dm_dp_mst_get_modes(struct drm_connector *connector)
aconnector->edid = edid;
}
if (aconnector->dc_sink && aconnector->dc_sink->sink_signal == SIGNAL_TYPE_VIRTUAL) {
dc_sink_release(aconnector->dc_sink);
aconnector->dc_sink = NULL;
}
if (!aconnector->dc_sink) {
struct dc_sink *dc_sink;
struct dc_sink_init_data init_params = {

57
drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_pp_smu.c
View File

@ -559,6 +559,58 @@ void pp_rv_set_pme_wa_enable(struct pp_smu *pp)
pp_funcs->notify_smu_enable_pwe(pp_handle);
}
void pp_rv_set_active_display_count(struct pp_smu *pp, int count)
{
const struct dc_context *ctx = pp->dm;
struct amdgpu_device *adev = ctx->driver_context;
void *pp_handle = adev->powerplay.pp_handle;
const struct amd_pm_funcs *pp_funcs = adev->powerplay.pp_funcs;
if (!pp_funcs || !pp_funcs->set_active_display_count)
return;
pp_funcs->set_active_display_count(pp_handle, count);
}
void pp_rv_set_min_deep_sleep_dcfclk(struct pp_smu *pp, int clock)
{
const struct dc_context *ctx = pp->dm;
struct amdgpu_device *adev = ctx->driver_context;
void *pp_handle = adev->powerplay.pp_handle;
const struct amd_pm_funcs *pp_funcs = adev->powerplay.pp_funcs;
if (!pp_funcs || !pp_funcs->set_min_deep_sleep_dcefclk)
return;
pp_funcs->set_min_deep_sleep_dcefclk(pp_handle, clock);
}
void pp_rv_set_hard_min_dcefclk_by_freq(struct pp_smu *pp, int clock)
{
const struct dc_context *ctx = pp->dm;
struct amdgpu_device *adev = ctx->driver_context;
void *pp_handle = adev->powerplay.pp_handle;
const struct amd_pm_funcs *pp_funcs = adev->powerplay.pp_funcs;
if (!pp_funcs || !pp_funcs->set_hard_min_dcefclk_by_freq)
return;
pp_funcs->set_hard_min_dcefclk_by_freq(pp_handle, clock);
}
void pp_rv_set_hard_min_fclk_by_freq(struct pp_smu *pp, int mhz)
{
const struct dc_context *ctx = pp->dm;
struct amdgpu_device *adev = ctx->driver_context;
void *pp_handle = adev->powerplay.pp_handle;
const struct amd_pm_funcs *pp_funcs = adev->powerplay.pp_funcs;
if (!pp_funcs || !pp_funcs->set_hard_min_fclk_by_freq)
return;
pp_funcs->set_hard_min_fclk_by_freq(pp_handle, mhz);
}
void dm_pp_get_funcs_rv(
struct dc_context *ctx,
struct pp_smu_funcs_rv *funcs)
@ -567,4 +619,9 @@ void dm_pp_get_funcs_rv(
funcs->set_display_requirement = pp_rv_set_display_requirement;
funcs->set_wm_ranges = pp_rv_set_wm_ranges;
funcs->set_pme_wa_enable = pp_rv_set_pme_wa_enable;
funcs->set_display_count = pp_rv_set_active_display_count;
funcs->set_min_deep_sleep_dcfclk = pp_rv_set_min_deep_sleep_dcfclk;
funcs->set_hard_min_dcfclk_by_freq = pp_rv_set_hard_min_dcefclk_by_freq;
funcs->set_hard_min_fclk_by_freq = pp_rv_set_hard_min_fclk_by_freq;
}

5
drivers/gpu/drm/amd/display/dc/Makefile
View File

@ -23,7 +23,7 @@
# Makefile for Display Core (dc) component.
#
DC_LIBS = basics bios calcs dce gpio i2caux irq virtual
DC_LIBS = basics bios calcs dce gpio irq virtual
ifdef CONFIG_DRM_AMD_DC_DCN1_0
DC_LIBS += dcn10 dml
@ -41,7 +41,8 @@ AMD_DC = $(addsuffix /Makefile, $(addprefix $(FULL_AMD_DISPLAY_PATH)/dc/,$(DC_LI
include $(AMD_DC)
DISPLAY_CORE = dc.o dc_link.o dc_resource.o dc_hw_sequencer.o dc_sink.o \
dc_surface.o dc_link_hwss.o dc_link_dp.o dc_link_ddc.o dc_debug.o dc_stream.o
dc_surface.o dc_link_hwss.o dc_link_dp.o dc_link_ddc.o dc_debug.o dc_stream.o \
dc_vm_helper.o
AMD_DISPLAY_CORE = $(addprefix $(AMDDALPATH)/dc/core/,$(DISPLAY_CORE))

14
drivers/gpu/drm/amd/display/dc/bios/bios_parser.c
View File

@ -835,18 +835,6 @@ static enum bp_result bios_parser_enable_crtc(
return bp->cmd_tbl.enable_crtc(bp, id, enable);
}
static enum bp_result bios_parser_crtc_source_select(
struct dc_bios *dcb,
struct bp_crtc_source_select *bp_params)
{
struct bios_parser *bp = BP_FROM_DCB(dcb);
if (!bp->cmd_tbl.select_crtc_source)
return BP_RESULT_FAILURE;
return bp->cmd_tbl.select_crtc_source(bp, bp_params);
}
static enum bp_result bios_parser_enable_disp_power_gating(
struct dc_bios *dcb,
enum controller_id controller_id,
@ -2842,8 +2830,6 @@ static const struct dc_vbios_funcs vbios_funcs = {
.program_crtc_timing = bios_parser_program_crtc_timing, /* still use. should probably retire and program directly */
.crtc_source_select = bios_parser_crtc_source_select, /* still use. should probably retire and program directly */
.program_display_engine_pll = bios_parser_program_display_engine_pll,
.enable_disp_power_gating = bios_parser_enable_disp_power_gating,

16
drivers/gpu/drm/amd/display/dc/bios/bios_parser2.c
View File

@ -1083,18 +1083,6 @@ static enum bp_result bios_parser_enable_crtc(
return bp->cmd_tbl.enable_crtc(bp, id, enable);
}
static enum bp_result bios_parser_crtc_source_select(
struct dc_bios *dcb,
struct bp_crtc_source_select *bp_params)
{
struct bios_parser *bp = BP_FROM_DCB(dcb);
if (!bp->cmd_tbl.select_crtc_source)
return BP_RESULT_FAILURE;
return bp->cmd_tbl.select_crtc_source(bp, bp_params);
}
static enum bp_result bios_parser_enable_disp_power_gating(
struct dc_bios *dcb,
enum controller_id controller_id,
@ -1899,8 +1887,6 @@ static const struct dc_vbios_funcs vbios_funcs = {
.is_accelerated_mode = bios_parser_is_accelerated_mode,
.is_active_display = bios_is_active_display,
.set_scratch_critical_state = bios_parser_set_scratch_critical_state,
@ -1917,8 +1903,6 @@ static const struct dc_vbios_funcs vbios_funcs = {
.program_crtc_timing = bios_parser_program_crtc_timing,
.crtc_source_select = bios_parser_crtc_source_select,
.enable_disp_power_gating = bios_parser_enable_disp_power_gating,
.bios_parser_destroy = firmware_parser_destroy,

96
drivers/gpu/drm/amd/display/dc/bios/bios_parser_helper.c
View File

@ -83,101 +83,7 @@ uint32_t bios_get_vga_enabled_displays(
{
uint32_t active_disp = 1;
if (bios->regs->BIOS_SCRATCH_3) /*follow up with other asic, todo*/
active_disp = REG_READ(BIOS_SCRATCH_3) & 0XFFFF;
active_disp = REG_READ(BIOS_SCRATCH_3) & 0XFFFF;
return active_disp;
}
bool bios_is_active_display(
struct dc_bios *bios,
enum signal_type signal,
const struct connector_device_tag_info *device_tag)
{
uint32_t active = 0;
uint32_t connected = 0;
uint32_t bios_scratch_0 = 0;
uint32_t bios_scratch_3 = 0;
switch (signal) {
case SIGNAL_TYPE_DVI_SINGLE_LINK:
case SIGNAL_TYPE_DVI_DUAL_LINK:
case SIGNAL_TYPE_HDMI_TYPE_A:
case SIGNAL_TYPE_DISPLAY_PORT:
case SIGNAL_TYPE_DISPLAY_PORT_MST:
{
if (device_tag->dev_id.device_type == DEVICE_TYPE_DFP) {
switch (device_tag->dev_id.enum_id) {
case 1:
{
active = ATOM_S3_DFP1_ACTIVE;
connected = 0x0008; //ATOM_DISPLAY_DFP1_CONNECT
}
break;
case 2:
{
active = ATOM_S3_DFP2_ACTIVE;
connected = 0x0080; //ATOM_DISPLAY_DFP2_CONNECT
}
break;
case 3:
{
active = ATOM_S3_DFP3_ACTIVE;
connected = 0x0200; //ATOM_DISPLAY_DFP3_CONNECT
}
break;
case 4:
{
active = ATOM_S3_DFP4_ACTIVE;
connected = 0x0400; //ATOM_DISPLAY_DFP4_CONNECT
}
break;
case 5:
{
active = ATOM_S3_DFP5_ACTIVE;
connected = 0x0800; //ATOM_DISPLAY_DFP5_CONNECT
}
break;
case 6:
{
active = ATOM_S3_DFP6_ACTIVE;
connected = 0x0040; //ATOM_DISPLAY_DFP6_CONNECT
}
break;
default:
break;
}
}
}
break;
case SIGNAL_TYPE_LVDS:
case SIGNAL_TYPE_EDP:
{
active = ATOM_S3_LCD1_ACTIVE;
connected = 0x0002; //ATOM_DISPLAY_LCD1_CONNECT
}
break;
default:
break;
}
if (bios->regs->BIOS_SCRATCH_0) /*follow up with other asic, todo*/
bios_scratch_0 = REG_READ(BIOS_SCRATCH_0);
if (bios->regs->BIOS_SCRATCH_3) /*follow up with other asic, todo*/
bios_scratch_3 = REG_READ(BIOS_SCRATCH_3);
bios_scratch_3 &= ATOM_S3_DEVICE_ACTIVE_MASK;
if ((active & bios_scratch_3) && (connected & bios_scratch_0))
return true;
return false;
}

Some files were not shown because too many files have changed in this diff Show More