Re: [PATCH] drm/amdkfd: Fix printing pointer cast

2017-12-10 Thread Oded Gabbay
Applied to -next Thanks, Oded On Mon, Dec 4, 2017 at 4:05 PM, Felix Kuehling wrote: > Thanks. This change is Reviewed-by: Felix Kuehling > > > On 2017-12-04 06:50 AM, Kent Russell wrote: >> Just print a pointer instead of casting >> >> v2: Remove the 0x prefix, since %p prints that automatically

Re: [PATCH 04/14] drm/amdkfd: Fix oversubscription accounting

2017-12-10 Thread Oded Gabbay
On Tue, Dec 5, 2017 at 9:27 PM, Felix Kuehling wrote: > On 2017-12-05 03:10 AM, Oded Gabbay wrote: >> On Tue, Nov 28, 2017 at 1:29 AM, Felix Kuehling >> wrote: >>> Don't count SDMA queues towards compute HQD oversubscription when >>> deciding whether to create a chained runlist. >>> >>> Signed-o

Re: [PATCH 08/14] drm/amdkfd: Get reference to lead_thread task struct

2017-12-10 Thread Oded Gabbay
On Tue, Dec 5, 2017 at 9:14 PM, Felix Kuehling wrote: > On 2017-12-05 03:43 AM, Oded Gabbay wrote: >> On Tue, Nov 28, 2017 at 1:29 AM, Felix Kuehling >> wrote: >>> Increment the kfd_process.lead_thread's reference counter to make >>> it safe to dereference. This is needed for getting a safe refe

Re: [PATCH 10/14] drm/amdkfd: Use ref count to prevent kfd_process destruction

2017-12-10 Thread Oded Gabbay
On Fri, Dec 1, 2017 at 11:17 PM, Felix Kuehling wrote: > > On 2017-11-28 04:52 AM, Christian König wrote: >> Am 28.11.2017 um 00:29 schrieb Felix Kuehling: >>> Use a reference counter instead of a lock to prevent process >>> destruction while functions running out of process context are using >>>

Re: [PATCH 09/14] drm/amdkfd: Make kfd_process reference counted

2017-12-10 Thread Oded Gabbay
On Tue, Nov 28, 2017 at 1:29 AM, Felix Kuehling wrote: > This will be used to elliminate the use of the process lock for > preventing concurrent process destruction. This will simplify lock > dependencies between KFD and KGD. > > This also simplifies the process destruction in a few ways: > * Don'

Re: [PATCH 11/14] drm/amdkfd: Return NULL if kfd_lookup_process_by_pasid fails

2017-12-10 Thread Oded Gabbay
On Tue, Nov 28, 2017 at 1:29 AM, Felix Kuehling wrote: > From: Yong Zhao > > If no matching process is found, return NULL instead of a pointer > to the last process in the kfd_processes_table. > > Signed-off-by: Yong Zhao > Signed-off-by: Felix Kuehling > --- > drivers/gpu/drm/amd/amdkfd/kfd_p

Re: [PATCH 12/14] drm/amdkfd: Reduce nesting in kfd_create_process_device_data

2017-12-10 Thread Oded Gabbay
On Tue, Nov 28, 2017 at 1:29 AM, Felix Kuehling wrote: > Signed-off-by: Felix Kuehling > --- > drivers/gpu/drm/amd/amdkfd/kfd_process.c | 23 --- > 1 file changed, 12 insertions(+), 11 deletions(-) > > diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_process.c > b/drivers/gpu/drm

Re: [PATCH 13/14] drm/amdkfd: Factor PDD destruction out of kfd_process_wq_release

2017-12-10 Thread Oded Gabbay
On Tue, Nov 28, 2017 at 1:29 AM, Felix Kuehling wrote: > Signed-off-by: Felix Kuehling > --- > drivers/gpu/drm/amd/amdkfd/kfd_process.c | 40 > +++- > 1 file changed, 24 insertions(+), 16 deletions(-) > > diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_process.c > b/dri

Re: [PATCH 14/14] drm/amdkfd: Simplify locking during process creation

2017-12-10 Thread Oded Gabbay
On Tue, Nov 28, 2017 at 1:29 AM, Felix Kuehling wrote: > From: Yong Zhao > > Also fixes error handling if kfd_process_init_cwsr fails. > > Signed-off-by: Yong Zhao > Signed-off-by: Felix Kuehling > --- > drivers/gpu/drm/amd/amdkfd/kfd_process.c | 46 > +++- > 1 fil

Re: [PATCH 00/37] KFD dGPU topology and initialization

2017-12-10 Thread Oded Gabbay
On Sat, Dec 9, 2017 at 6:08 AM, Felix Kuehling wrote: > This patch series adds support for dGPU topology to KFD and implements > everything needed to initialize KFD on dGPUs. I'm excited! > > This is still missing dGPU memory management APIs, so it's not going to > be able to run any user mode t

Re: [Ask for help] ATI HD 7470 card on arm64 platform

2017-12-10 Thread Xinliang Liu
On 7 December 2017 at 19:11, Michel Dänzer wrote: > On 2017-12-07 10:35 AM, Xinliang Liu wrote: >> Hi, >> >> We are trying ATI HD 7470 card on arm64 platform D05 with kernel 4.1.40. >> But it failed with kernel log: >> radeon 0005:01:00: ring 0 stalled for more than 1msec >> radeon 0005:01:00:

RE: [PATCH] drm/amd/amdgpu: set gtt size according to system memory size only

2017-12-10 Thread He, Roger
-Original Message- From: Michel Dänzer [mailto:mic...@daenzer.net] Sent: Saturday, December 09, 2017 1:26 AM To: Koenig, Christian ; He, Roger Cc: amd-gfx@lists.freedesktop.org Subject: Re: [PATCH] drm/amd/amdgpu: set gtt size according to system memory size only On 2017-12-07 07:07 PM,

Re: [PATCH 1/2] drm/amdgpu: reverse PDBs order

2017-12-10 Thread Chunming Zhou
On 2017年12月08日 22:58, Alex Deucher wrote: On Fri, Dec 8, 2017 at 5:56 AM, Chunming Zhou wrote: The hiberachy of page table is as below, which aligns hw names. PDB2->PDB1->PDB0->PTB, accordingly: level3 --- PDB2 level2 --- PDB1 level1 --- PDB0 level0 --- PTB What's the advantage of this chang

Re: [PATCH 1/2] drm/amdgpu: reverse PDBs order

2017-12-10 Thread Alex Deucher
On Sun, Dec 10, 2017 at 10:01 PM, Chunming Zhou wrote: > > > On 2017年12月08日 22:58, Alex Deucher wrote: >> >> On Fri, Dec 8, 2017 at 5:56 AM, Chunming Zhou wrote: >>> >>> The hiberachy of page table is as below, which aligns hw names. >>> PDB2->PDB1->PDB0->PTB, accordingly: >>> level3 --- PDB2 >>>

Re: [PATCH 1/8] drm/amdgpu: stop joining PDEs

2017-12-10 Thread Chunming Zhou
On 2017年12月09日 00:41, Christian König wrote: That doesn't hit any more most of the time anyway. agree, there is little chance that two PDBs are continue, Reviewed-by: Chunming Zhou Signed-off-by: Christian König --- drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 41 ++

Re: [PATCH 2/8] drm/amdgpu: update one PDE at a time

2017-12-10 Thread Chunming Zhou
On 2017年12月09日 00:41, Christian König wrote: Horrible inefficient, but avoids problems when the root PD size becomes to big. Signed-off-by: Christian König --- drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 81 +++--- 1 file changed, 36 insertions(+), 45 deletions(-)

Re: [PATCH 3/8] drm/amdgpu: avoid the modulo in amdgpu_vm_get_entry

2017-12-10 Thread Chunming Zhou
Reviewed-by: Chunming Zhou On 2017年12月09日 00:41, Christian König wrote: We can do this with a simple mask as well. Signed-off-by: Christian König --- drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/drivers/gpu/drm/amd/amdgpu

Re: [PATCH 4/8] drm/amdgpu: remove last_entry_used from the VM code

2017-12-10 Thread Chunming Zhou
On 2017年12月09日 00:41, Christian König wrote: Not needed any more. Signed-off-by: Christian König --- drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 52 +++--- drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h | 1 - 2 files changed, 29 insertions(+), 24 deletions(-) diff --gi

Re: [PATCH 5/8] drm/amdgpu: remove keeping the addr of the VM PDs

2017-12-10 Thread Chunming Zhou
Reviewed-by: Chunming Zhou On 2017年12月09日 00:41, Christian König wrote: No more double house keeping. Signed-off-by: Christian König --- drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 13 - drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h | 2 +- 2 files changed, 5 insertions(+), 10 deletio

Re: [PATCH 6/8] drm/amdgpu: batch PDE updates again

2017-12-10 Thread Chunming Zhou
Reviewed-by: Chunming Zhou On 2017年12月09日 00:41, Christian König wrote: Now instead of one submission for each PDE batch them together over all PDs who need an update. Signed-off-by: Christian König --- drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 206 +++-- 1 file

[PATCH] drm/amdgpu: set default temperature range

2017-12-10 Thread Satyajit Sahu
Setting default values for min and maxi threshold temperatures. Hwmgr can override these values if it has necessary tables. Reviewed-by: Alex Deucher Signed-off-by: Satyajit Sahu --- drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c | 4 drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.h | 3 +++ 2

Re: [PATCH 8/8] drm/amdgpu: implement 2+1 PD support for Raven

2017-12-10 Thread Chunming Zhou
It is a coincidence, I also am trying to implement this, still under debug. For yours, seems not ready yet, right? We should handle 64KB native page to 16 * 4KB sub-PTB for TF case, which is the only verified option by HW. For TF case, the number entries and shift of PTB is a bit different from

Re: [PATCH 2/2] drm/amdgpu: fix pte index calculation

2017-12-10 Thread Chunming Zhou
On 2017年12月08日 22:00, Christian König wrote: What is wrong with the old approach? When the num entries are less than one page(512), the old approach will result in error. e.g. when we enable translate further feature, the ptb will only have 32 entries with 64KB---16 * 4KB selection. I wou

Re: [PATCH 1/2] drm/amdgpu: reverse PDBs order

2017-12-10 Thread Chunming Zhou
On 2017年12月08日 22:06, Christian König wrote: Am 08.12.2017 um 11:56 schrieb Chunming Zhou: The hiberachy of page table is as below, which aligns hw names. PDB2->PDB1->PDB0->PTB, accordingly: level3 --- PDB2 level2 --- PDB1 level1 --- PDB0 level0 --- PTB Change-Id: I2d748e5e96cffe18294c104c4b1