[PATCH] drm/amdgpu: fix S3 failure on specific platform

Change-Id: Ie932508ad6949f8bfc7c8db1f5874d3440d09fc6 Signed-off-by: Ken Wang --- drivers/gpu/drm/amd/amdgpu/amdgpu.h| 3 +++ drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 8 2 files changed, 11 insertions(+) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h

Re: [PATCH] drm/amdgpu: fix S3 failure on specific platform

sure, I will add the delay to io_wreg. From: Deucher, Alexander Sent: Tuesday, July 4, 2017 12:47:38 PM To: Wang, Ken; amd-gfx@lists.freedesktop.org Cc: Wang, Ken Subject: RE: [PATCH] drm/amdgpu: fix S3 failure on specific platform > -Original Message- >

Re: [PATCH 07/12] drm: export drm_gem_prime_dmabuf_ops

On 4 Jul. 2017 11:23, "Michel Dänzer" wrote: Adding the dri-devel list, since this is a core DRM patch. On 04/07/17 06:11 AM, Felix Kuehling wrote: > From: Christian König > > This allows drivers to check if a DMA-buf contains a GEM object or

[PATCH] drm/amdgpu: fix S3 failure on specific platform

From: Ken Wang Change-Id: Ie932508ad6949f8bfc7c8db1f5874d3440d09fc6 Signed-off-by: Ken Wang --- drivers/gpu/drm/amd/amdgpu/amdgpu.h| 3 +++ drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 8 2 files changed, 11 insertions(+) diff

RE: [PATCH] drm/amdgpu: fix S3 failure on specific platform

> -Original Message- > From: amd-gfx [mailto:amd-gfx-boun...@lists.freedesktop.org] On Behalf > Of Ken Wang > Sent: Tuesday, July 04, 2017 12:26 AM > To: amd-gfx@lists.freedesktop.org > Cc: Wang, Ken > Subject: [PATCH] drm/amdgpu: fix S3 failure on specific platform > > Change-Id:

[PATCH] drm/amdgpu: fix S3 failure on specific platform

Change-Id: Ie932508ad6949f8bfc7c8db1f5874d3440d09fc6 Signed-off-by: Ken Wang --- drivers/gpu/drm/amd/amdgpu/amdgpu.h| 3 +++ drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 8 2 files changed, 11 insertions(+) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h

RE: [PATCH 07/11] drm/amdgpu: rename GART to SYSVM

Distinguishing system vm and general vm is a good idea, but I'm not sure about renaming GTT to sysvm part, especially TTM TT stays there. Maybe we just need rename GART functions to SYSVM. Regards, David Zhou -Original Message- From: amd-gfx

RE: [PATCH 03/12] drm/amdgpu: Enable SDMA context switching for CIK

Reviewed-by: Chunming Zhou -Original Message- From: amd-gfx [mailto:amd-gfx-boun...@lists.freedesktop.org] On Behalf Of Felix Kuehling Sent: Tuesday, July 04, 2017 5:11 AM To: amd-gfx@lists.freedesktop.org Cc: Kuehling, Felix Subject: [PATCH

RE: [PATCH 04/12] drm/amdgpu: Make SDMA phase quantum configurable

Acked-by: Chunming Zhou -Original Message- From: amd-gfx [mailto:amd-gfx-boun...@lists.freedesktop.org] On Behalf Of Felix Kuehling Sent: Tuesday, July 04, 2017 5:11 AM To: amd-gfx@lists.freedesktop.org Cc: Kuehling, Felix Subject: [PATCH

Re: [PATCH 08/12] drm/amdgpu: disallow foreign BOs for UVD/VCE

On Mon, Jul 3, 2017 at 5:11 PM, Felix Kuehling wrote: > From: Christian König > > They don't support VM mode yet. > > Signed-off-by: Christian König > Reviewed-by: Felix Kuehling This could

Re: [PATCH 07/12] drm: export drm_gem_prime_dmabuf_ops

Adding the dri-devel list, since this is a core DRM patch. On 04/07/17 06:11 AM, Felix Kuehling wrote: > From: Christian König > > This allows drivers to check if a DMA-buf contains a GEM object or not. > > Signed-off-by: Christian König >

Re: [PATCH RFC 2/2] drm/amdgpu: Set/clear CPU_ACCESS flag on page fault and move to VRAM

On 03/07/17 08:47 PM, Christian König wrote: > Am 03.07.2017 um 11:49 schrieb Michel Dänzer: >>> Instead of messing with all this I suggest that we just add a jiffies >>> based timeout to the BO when we can clear the flag. For kernel BOs this >>> timeout is just infinity. >>> >>> Then we check in

Re: [PATCH 06/12] drm/amdgpu: Correctly establish the suspend/resume hook for amdkfd

On Tue, Jul 4, 2017 at 12:11 AM, Felix Kuehling wrote: > From: Yong Zhao > > Signed-off-by: Yong Zhao > Reviewed-by: Felix Kuehling > --- > drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 7 +++ >

[PATCH 08/12] drm/amdgpu: disallow foreign BOs for UVD/VCE

From: Christian König They don't support VM mode yet. Signed-off-by: Christian König Reviewed-by: Felix Kuehling --- drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)

[PATCH 05/12] drm/amdgpu: Send no-retry XNACK for all fault types

From: Jay Cornwall A subset of VM fault types currently send retry XNACK to the client. This causes a storm of interrupts from the VM to the host. Until the storm is throttled by other means send no-retry XNACK for all fault types instead. No change in behavior to the

[PATCH 00/12] Patches from amd-kfd-staging

Varios cleaned-up and some squashed patches from amd-kfg-staging that are not necessarily KFD-on-dGPU-specific. The intention is to minimize differences between amd-kfd-staging and upstream before reviewing KFD-specific changes for upstreaming. Patches 7-12 are a rebased (multiple times) patch

[PATCH 07/12] drm: export drm_gem_prime_dmabuf_ops

From: Christian König This allows drivers to check if a DMA-buf contains a GEM object or not. Signed-off-by: Christian König Reviewed-by: Felix Kuehling --- drivers/gpu/drm/drm_prime.c | 3 ++- include/drm/drmP.h

[PATCH 10/12] drm/amdgpu: separate BO from GEM object

From: Christian König This allows us to have multiple GEM objects for one BO. Signed-off-by: Christian König Reviewed-by: Felix Kuehling --- drivers/gpu/drm/amd/amdgpu/amdgpu.h| 12 +++--

[PATCH 01/12] drm/amdgpu: implement vm_operations_struct.access

Allows gdb to access contents of user mode mapped BOs. Signed-off-by: Felix Kuehling --- drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c | 130 +++- drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h | 2 + 2 files changed, 131 insertions(+), 1 deletion(-)

[PATCH 09/12] drm/amdgpu: disallow foreign BOs in the display path

From: Christian König Pinning them in other devices VRAM would obviously not work. Signed-off-by: Christian König Reviewed-by: Felix Kuehling --- drivers/gpu/drm/amd/amdgpu/amdgpu_display.c | 6 ++ 1 file

[PATCH 03/12] drm/amdgpu: Enable SDMA context switching for CIK

Enable SDMA context switching on CIK (copied from sdma_v3_0.c). Signed-off-by: Felix Kuehling --- drivers/gpu/drm/amd/amdgpu/cik_sdma.c | 30 ++ 1 file changed, 30 insertions(+) diff --git a/drivers/gpu/drm/amd/amdgpu/cik_sdma.c

[PATCH 06/12] drm/amdgpu: Correctly establish the suspend/resume hook for amdkfd

From: Yong Zhao Signed-off-by: Yong Zhao Reviewed-by: Felix Kuehling --- drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 7 +++ drivers/gpu/drm/amd/amdgpu/cik.c | 9 + 2 files changed, 8 insertions(+), 8

[PATCH 04/12] drm/amdgpu: Make SDMA phase quantum configurable

Set a configurable SDMA phase quantum when enabling SDMA context switching. The default value significantly reduces SDMA latency in page table updates when user-mode SDMA queues have concurrent activity, compared to the initial HW setting. Signed-off-by: Felix Kuehling

[PATCH 12/12] drm/amdgpu: enable foreign DMA-buf objects

From: Christian König We should be able to handle BOs from other instances as well. Signed-off-by: Christian König Reviewed-by: Felix Kuehling --- drivers/gpu/drm/amd/amdgpu/amdgpu.h | 2 ++

[PATCH 02/12] drm/amdgpu: Enable SDMA_CNTL.ATC_L1_ENABLE for SDMA on CZ

From: shaoyunl For GFX context, the ATC bit in SDMA*_GFX_VIRTUAL_ADDRESS can be cleared to perform in VM mode. For RLC context, to support ATC mode , ATC bit in SDMA*_RLC*_VIRTUAL_ADDRESS should be set. SDMA_CNTL.ATC_L1_ENABLE bit is global setting that enables the L1-L2

[PATCH 11/12] drm/amdgpu: handle foreign BOs in the VM mapping

From: Amber Lin Set the system bit for foreign BO mappings and use the remote VRAM BAR address as the VRAM base offset. Signed-off-by: Amber Lin Reviewed-by: Felix Kuehling --- drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 17

[PATCH] drm/amdgpu: check scratch registers to see if we need post (v2)

Rather than checking the CONGIG_MEMSIZE register as that may not be reliable on some APUs. v2: The scratch register is only used on CIK+ Reviewed-by: Christian König Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_device.c

Re: [PATCH v2 1/2] drm/ttm: Fix use-after-free in ttm_bo_clean_mm

I've gone ahead and pushed the two to our amd-staging-4.11 branch. Alex will certainly pick that up for his next fixes pull request. Thanks for the help, Christian. Am 03.07.2017 um 20:05 schrieb John Brooks: We unref the man->move fence in ttm_bo_clean_mm() and then call

[PATCH v2 2/2] drm/amdgpu: Don't call amd_powerplay_destroy() if we don't have powerplay

amd_powerplay_destroy() expects a handle pointing to a struct pp_instance. On chips without PowerPlay, pp_handle points to a struct amdgpu_device. The resulting attempt to kfree() fields of the wrong struct ends in fire: [ 91.560405] BUG: unable to handle kernel paging request at

[PATCH v2 1/2] drm/ttm: Fix use-after-free in ttm_bo_clean_mm

We unref the man->move fence in ttm_bo_clean_mm() and then call ttm_bo_force_list_clean() which waits on it, except the refcount is now zero so a warning is generated (or worse): [149492.279301] refcount_t: increment on 0; use-after-free. [149492.279309] [ cut here ]

RE: [PATCH] drm/amdgpu: trace VM flags as 64bits

> -Original Message- > From: amd-gfx [mailto:amd-gfx-boun...@lists.freedesktop.org] On Behalf > Of Christian König > Sent: Monday, July 03, 2017 9:25 AM > To: amd-gfx@lists.freedesktop.org > Subject: [PATCH] drm/amdgpu: trace VM flags as 64bits > > From: Christian König

RE: [PATCH] drm/amdgpu: remove stale TODO comment

> -Original Message- > From: amd-gfx [mailto:amd-gfx-boun...@lists.freedesktop.org] On Behalf > Of Christian König > Sent: Monday, July 03, 2017 9:22 AM > To: amd-gfx@lists.freedesktop.org > Subject: [PATCH] drm/amdgpu: remove stale TODO comment > > From: Christian König

RE: [PATCH 3/4] drm/amd/powerplay: move VI common AVFS code to smu7_smumgr.c

> -Original Message- > From: amd-gfx [mailto:amd-gfx-boun...@lists.freedesktop.org] On Behalf > Of Rex Zhu > Sent: Monday, July 03, 2017 6:14 AM > To: amd-gfx@lists.freedesktop.org > Cc: Zhu, Rex > Subject: [PATCH 3/4] drm/amd/powerplay: move VI common AVFS code to > smu7_smumgr.c > >

RE: [PATCH 1/4] drm/amd/powerplay: fix avfs state update error on polaris.

> -Original Message- > From: amd-gfx [mailto:amd-gfx-boun...@lists.freedesktop.org] On Behalf > Of Rex Zhu > Sent: Monday, July 03, 2017 6:14 AM > To: amd-gfx@lists.freedesktop.org > Cc: Zhu, Rex > Subject: [PATCH 1/4] drm/amd/powerplay: fix avfs state update error on > polaris. > >

RE: [PATCH 4/4] drm/amd/powerplay: add avfs check for old asics on Vi.

> -Original Message- > From: amd-gfx [mailto:amd-gfx-boun...@lists.freedesktop.org] On Behalf > Of Rex Zhu > Sent: Monday, July 03, 2017 6:14 AM > To: amd-gfx@lists.freedesktop.org > Cc: Zhu, Rex > Subject: [PATCH 4/4] drm/amd/powerplay: add avfs check for old asics on Vi. > >

RE: [PATCH] drm/amdgpu: fix vulkan test performance drop and hang on VI

> -Original Message- > From: amd-gfx [mailto:amd-gfx-boun...@lists.freedesktop.org] On Behalf > Of Rex Zhu > Sent: Monday, July 03, 2017 6:13 AM > To: amd-gfx@lists.freedesktop.org > Cc: Zhu, Rex > Subject: [PATCH] drm/amdgpu: fix vulkan test performance drop and hang > on VI > > caused

Re: [PATCH] drm/amdgpu: trace VM flags as 64bits

Reviewed-by: Alex Xie On 2017-07-03 09:25 AM, Christian König wrote: From: Christian König Otherwise the upper bits are lost. Signed-off-by: Christian König --- drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h | 18

[PATCH] drm/amdgpu: trace VM flags as 64bits

From: Christian König Otherwise the upper bits are lost. Signed-off-by: Christian König --- drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h | 18 +- 1 file changed, 9 insertions(+), 9 deletions(-) diff --git

[PATCH] drm/amdgpu: remove stale TODO comment

From: Christian König That is already fixed. Signed-off-by: Christian König --- drivers/gpu/drm/amd/amdgpu/amdgpu_object.c | 1 - 1 file changed, 1 deletion(-) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c

Re: Deprecation of AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED

On Mon, Jul 3, 2017 at 12:08 PM, Michel Dänzer wrote: > On 30/06/17 08:43 PM, Marek Olšák wrote: >> >> I don't know what is being talked about here anymore, but I wouldn't >> like to use CPU_ACCESS_REQUIRED or CPU_ACCESS_REALLY_REQUIRED in >> userspace. The reason is that

Re: [PATCH] drm/amdgpu: Make KIQ read/write register routine be atomic

Hi Shaoyun looks you want to make KIQ reg access be atomic & UN-interruptible, I think most user of KIQ reg access is not in atomic context, so your patch only benefit for the place using KIQ from IRQ. why not implement another KIQ reg access function ? e.g. :

Re: [PATCH RFC 2/2] drm/amdgpu: Set/clear CPU_ACCESS flag on page fault and move to VRAM

Am 03.07.2017 um 11:49 schrieb Michel Dänzer: Instead of messing with all this I suggest that we just add a jiffies based timeout to the BO when we can clear the flag. For kernel BOs this timeout is just infinity. Then we check in amdgpu_cs_bo_validate() before generating the placements if we

[PATCH 3/4] drm/amd/powerplay: move VI common AVFS code to smu7_smumgr.c

Change-Id: I2bee3e700281a57ad77132794187ef45d2d79dcd Signed-off-by: Rex Zhu --- drivers/gpu/drm/amd/powerplay/inc/smumgr.h | 3 + drivers/gpu/drm/amd/powerplay/smumgr/fiji_smc.c| 6 +- drivers/gpu/drm/amd/powerplay/smumgr/fiji_smumgr.c | 73 ++

[PATCH 2/4] drm/amd/powerplay: refine avfs enable code on fiji.

1. simplify avfs state switch. 2. delete save/restore VFT table functions as not support by fiji. 3. implement thermal_avfs_enable funciton. Change-Id: I5e28672591d77b1ec9e3406d0fc7d42566831a08 Signed-off-by: Rex Zhu --- drivers/gpu/drm/amd/powerplay/smumgr/fiji_smc.c|

[PATCH 4/4] drm/amd/powerplay: add avfs check for old asics on Vi.

Change-Id: I1737ba27ae2a8f5c579ccb541ced9cb979ffd1ff Signed-off-by: Rex Zhu --- drivers/gpu/drm/amd/powerplay/hwmgr/smu7_hwmgr.c | 9 + 1 file changed, 9 insertions(+) diff --git a/drivers/gpu/drm/amd/powerplay/hwmgr/smu7_hwmgr.c

[PATCH] drm/amdgpu: fix vulkan test performance drop and hang on VI

caused by not program dynamic_cu_mask_addr in the KIQ MQD. v2: create struct vi_mqd_allocation in FB which will contain 1. PM4 MQD structure. 2. Write Pointer Poll Memory. 3. Read Pointer Report Memory 4. Dynamic CU Mask. 5. Dynamic RB Mask. Change-Id: I22c840f1bf8d365f7df33a27d6b11e1aea8f2958

Re: Deprecation of AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED

On 30/06/17 08:43 PM, Marek Olšák wrote: > > I don't know what is being talked about here anymore, but I wouldn't > like to use CPU_ACCESS_REQUIRED or CPU_ACCESS_REALLY_REQUIRED in > userspace. The reason is that userspace doesn't and can't know whether > CPU access will be required, and the

Re: [PATCH RFC v2] drm/amdgpu: Set/clear CPU_ACCESS flag on page fault and move to VRAM

On 01/07/17 12:31 AM, John Brooks wrote: > When a BO is moved to VRAM, clear AMDGPU_BO_FLAG_CPU_ACCESS. This allows it > to potentially later move to invisible VRAM if the CPU does not access it > again. > > Setting the CPU_ACCESS flag in amdgpu_bo_fault_reserve_notify() also means > that we can

Re: [PATCH RFC 2/2] drm/amdgpu: Set/clear CPU_ACCESS flag on page fault and move to VRAM

On 03/07/17 04:06 PM, Christian König wrote: > Am 03.07.2017 um 03:34 schrieb Michel Dänzer: > >> [...] I suggested clearing the flag here to John on IRC. The >> idea is briefly described in the commit log, let me elaborate a bit on >> that: >> >> When a BO is moved to VRAM which has the

[PATCH 11/11] drm/amdgpu: add sysvm_size

From: Christian König Limit the size of the SYSVM. This saves us a bunch of visible VRAM, but also limitates the maximum BO size we can swap out. v2: rebased and cleaned up after GART to SYSVM rename. Signed-off-by: Christian König ---

[PATCH 10/11] drm/amdgpu: setup GTT size directly from module parameter

From: Christian König Instead of relying on the sysvm_size to be the same as the module parameter. Signed-off-by: Christian König --- drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c | 12 +--- 1 file changed, 9 insertions(+), 3 deletions(-)

[PATCH 08/11] drm/amdgpu: move SYSVM struct and function into amdgpu_sysvm.h

From: Christian König No functional change. Signed-off-by: Christian König --- drivers/gpu/drm/amd/amdgpu/amdgpu.h | 48 +-- drivers/gpu/drm/amd/amdgpu/amdgpu_sysvm.h | 77 +++

[PATCH 03/11] drm/amdgpu: use the GTT windows for BO moves v2

From: Christian König This way we don't need to map the full BO at a time any more. v2: use fixed windows for src/dst Signed-off-by: Christian König --- drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c | 125 +++-

[PATCH 09/11] drm/amdgpu: move amdgpu_sysvm_location into amdgpu_sysvm.c as well

From: Christian König No intended functional change. Signed-off-by: Christian König --- drivers/gpu/drm/amd/amdgpu/amdgpu.h| 1 - drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 36

[PATCH 07/11] drm/amdgpu: rename GART to SYSVM

From: Christian König Just mass rename all names related to the hardware GART/GTT functions to SYSVM. The name of symbols related to the TTM TT domain stay the same. This should improve the distinction between the two. Signed-off-by: Christian König

[PATCH 04/11] drm/amdgpu: stop mapping BOs to GTT

From: Christian König No need to map BOs to GTT on eviction and intermediate transfers any more. Signed-off-by: Christian König --- drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c | 19 ++- 1 file changed, 2 insertions(+), 17

[PATCH 01/11] drm/amdgpu: reserve the first 2x512 of GART

From: Christian König We want to use them as remap address space. Signed-off-by: Christian König --- drivers/gpu/drm/amd/amdgpu/amdgpu_gtt_mgr.c | 5 - drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h | 3 +++ 2 files changed, 7

[PATCH 02/11] drm/amdgpu: add amdgpu_gart_map function v2

From: Christian König This allows us to write the mapped PTEs into an IB instead of the table directly. v2: fix build with debugfs enabled, remove unused assignment Signed-off-by: Christian König --- drivers/gpu/drm/amd/amdgpu/amdgpu.h

[PATCH 05/11] drm/amdgpu: remove maximum BO size limitation v2

From: Christian König We can finally remove this now. v2: remove now unused max_size variable as well. Signed-off-by: Christian König --- drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c | 12 1 file changed, 12 deletions(-) diff --git

[PATCH 06/11] drm/amdgpu: use TTM values instead of MC values for the info queries

From: Christian König Use the TTM values instead of the hardware config here. Signed-off-by: Christian König --- drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c | 10 ++ 1 file changed, 6 insertions(+), 4 deletions(-) diff --git

Re: [PATCH] drm: amd: amdgpu: constify ttm_place structures.

Am 02.07.2017 um 11:13 schrieb Arvind Yadav: ttm_place are not supposed to change at runtime. All functions working with ttm_place provided by work with const ttm_place. So mark the non-const structs as const. Signed-off-by: Arvind Yadav Reviewed-by: Christian

Re: [PATCH] drm: radeon: radeon_ttm: constify ttm_place structures.

Am 02.07.2017 um 11:06 schrieb Arvind Yadav: ttm_place are not supposed to change at runtime. All functions working with ttm_place provided by work with const ttm_place. So mark the non-const structs as const. File size before: text data bss dec hex filename 9235

Re: [PATCH] drm: radeon: constify drm_prop_enum_list structures.

Am 01.07.2017 um 11:47 schrieb Arvind Yadav: drm_prop_enum_lists are not supposed to change at runtime. All functions working with drm_prop_enum_list provided by work with const drm_prop_enum_list. So mark the non-const structs as const. File size before: text data bss

Re: [PATCH RFC 2/2] drm/amdgpu: Set/clear CPU_ACCESS flag on page fault and move to VRAM

Am 03.07.2017 um 03:34 schrieb Michel Dänzer: On 02/07/17 09:52 PM, Christian König wrote: Am 30.06.2017 um 17:18 schrieb John Brooks: When a BO is moved to VRAM, clear AMDGPU_BO_FLAG_CPU_ACCESS. This allows it to potentially later move to invisible VRAM if the CPU does not access it again.