On 21/07/14 21:14, Jerome Glisse wrote:
On Mon, Jul 21, 2014 at 08:42:58PM +0300, Oded Gabbay wrote:
On 21/07/14 18:54, Jerome Glisse wrote:
On Mon, Jul 21, 2014 at 05:12:06PM +0300, Oded Gabbay wrote:
On 21/07/14 16:39, Christian König wrote:
Am 21.07.2014 14:36, schrieb Oded Gabbay:
On 20
On 21/07/14 21:22, Daniel Vetter wrote:
On Mon, Jul 21, 2014 at 7:28 PM, Oded Gabbay oded.gab...@amd.com wrote:
I'm not sure whether we can do the same trick with the hw scheduler. But
then unpinning hw contexts will drain the pipeline anyway, so I guess we
can just stop feeding the hw
On 21/07/14 21:59, Jerome Glisse wrote:
On Mon, Jul 21, 2014 at 09:36:44PM +0300, Oded Gabbay wrote:
On 21/07/14 21:14, Jerome Glisse wrote:
On Mon, Jul 21, 2014 at 08:42:58PM +0300, Oded Gabbay wrote:
On 21/07/14 18:54, Jerome Glisse wrote:
On Mon, Jul 21, 2014 at 05:12:06PM +0300, Oded
On 21/07/14 22:28, Jerome Glisse wrote:
On Mon, Jul 21, 2014 at 10:23:43PM +0300, Oded Gabbay wrote:
On 21/07/14 21:59, Jerome Glisse wrote:
On Mon, Jul 21, 2014 at 09:36:44PM +0300, Oded Gabbay wrote:
On 21/07/14 21:14, Jerome Glisse wrote:
On Mon, Jul 21, 2014 at 08:42:58PM +0300, Oded
On 22/07/14 02:05, Jerome Glisse wrote:
On Tue, Jul 22, 2014 at 12:56:13AM +0300, Oded Gabbay wrote:
On 21/07/14 22:28, Jerome Glisse wrote:
On Mon, Jul 21, 2014 at 10:23:43PM +0300, Oded Gabbay wrote:
On 21/07/14 21:59, Jerome Glisse wrote:
On Mon, Jul 21, 2014 at 09:36:44PM +0300, Oded
On 22/07/14 10:23, Daniel Vetter wrote:
On Mon, Jul 21, 2014 at 10:23:43PM +0300, Oded Gabbay wrote:
But Jerome, the core problem still remains in effect, even with your
suggestion. If an application, either via userspace queue or via ioctl,
submits a long-running kernel, than the CPU
On 22/07/14 10:28, Daniel Vetter wrote:
On Mon, Jul 21, 2014 at 03:03:07PM -0400, Jerome Glisse wrote:
On Mon, Jul 21, 2014 at 09:41:29PM +0300, Oded Gabbay wrote:
On 21/07/14 21:22, Daniel Vetter wrote:
On Mon, Jul 21, 2014 at 7:28 PM, Oded Gabbay oded.gab...@amd.com wrote:
I'm not sure
On 22/07/14 10:40, Daniel Vetter wrote:
On Tue, Jul 22, 2014 at 09:28:51AM +0200, Daniel Vetter wrote:
On Mon, Jul 21, 2014 at 03:03:07PM -0400, Jerome Glisse wrote:
On Mon, Jul 21, 2014 at 09:41:29PM +0300, Oded Gabbay wrote:
On 21/07/14 21:22, Daniel Vetter wrote:
On Mon, Jul 21, 2014 at 7
On 22/07/14 12:21, Daniel Vetter wrote:
On Tue, Jul 22, 2014 at 10:19 AM, Oded Gabbay oded.gab...@amd.com wrote:
Exactly, just prevent userspace from submitting more. And if you have
misbehaving userspace that submits too much, reset the gpu and tell it
that you're sorry but won't schedule any
On 22/07/14 14:15, Daniel Vetter wrote:
On Tue, Jul 22, 2014 at 12:52:43PM +0300, Oded Gabbay wrote:
On 22/07/14 12:21, Daniel Vetter wrote:
On Tue, Jul 22, 2014 at 10:19 AM, Oded Gabbay oded.gab...@amd.com wrote:
Exactly, just prevent userspace from submitting more. And if you have
On 23/07/14 10:05, Daniel Vetter wrote:
On Wed, Jul 23, 2014 at 8:50 AM, Oded Gabbay oded.gab...@amd.com wrote:
On 22/07/14 14:15, Daniel Vetter wrote:
On Tue, Jul 22, 2014 at 12:52:43PM +0300, Oded Gabbay wrote:
On 22/07/14 12:21, Daniel Vetter wrote:
On Tue, Jul 22, 2014 at 10:19 AM
From: Alexey Skidanov alexey.skida...@amd.com
Any kernel source registering the invalid PPR calback may include the header
file with PPR fault flags macros definitions.
Thus we move them to include/linux/amd-iommu.h
Signed-off-by: Alexey Skidanov alexey.skida...@amd.com
Signed-off-by: Oded
From: Alexey Skidanov alexey.skida...@amd.com
The pasid wasn't properly initialized before caling to invalid PPR calback
Signed-off-by: Alexey Skidanov alexey.skida...@amd.com
Signed-off-by: Oded Gabbay oded.gab...@amd.com
---
drivers/iommu/amd_iommu_v2.c | 1 +
1 file changed, 1 insertion
On 21/07/14 02:09, Jerome Glisse wrote:
On Thu, Jul 17, 2014 at 04:29:28PM +0300, Oded Gabbay wrote:
From: Ben Goz ben@amd.com
Signed-off-by: Ben Goz ben@amd.com
Signed-off-by: Oded Gabbay oded.gab...@amd.com
---
drivers/gpu/drm/radeon/amdkfd/kfd_chardev.c | 133
On 21/07/14 05:45, Jerome Glisse wrote:
On Thu, Jul 17, 2014 at 04:29:23PM +0300, Oded Gabbay wrote:
From: Ben Goz ben@amd.com
This patch adds a new parameter to the amdkfd driver. This parameter enables
the user to select the scheduling policy of the CP. The choices are:
* CP Scheduling
On 21/07/14 05:42, Jerome Glisse wrote:
On Thu, Jul 17, 2014 at 04:29:22PM +0300, Oded Gabbay wrote:
From: Ben Goz ben@amd.com
The kernel queue module enables the amdkfd to establish kernel queues, not
exposed to user space.
The kernel queues are used for HIQ (HSA Interface Queue
On 21/07/14 02:06, Jerome Glisse wrote:
On Thu, Jul 17, 2014 at 04:29:20PM +0300, Oded Gabbay wrote:
From: Ben Goz ben@amd.com
The queue module enables allocating and initializing queues uniformly.
Signed-off-by: Ben Goz ben@amd.com
Signed-off-by: Oded Gabbay oded.gab...@amd.com
On 21/07/14 02:04, Jerome Glisse wrote:
On Thu, Jul 17, 2014 at 04:29:19PM +0300, Oded Gabbay wrote:
This patch adds the functions to bind and unbind pasid from a device through
the amd_iommu driver.
The unbind function is called when the mm_struct of the process is released.
The bind
On 21/07/14 01:37, Jerome Glisse wrote:
On Thu, Jul 17, 2014 at 04:29:17PM +0300, Oded Gabbay wrote:
From: Evgeny Pinchuk evgeny.pinc...@amd.com
This patch adds the topology module to the driver. The topology is exposed to
userspace through the sysfs.
The calls to add and remove a device
On 21/07/14 01:37, Jerome Glisse wrote:
On Thu, Jul 17, 2014 at 04:29:17PM +0300, Oded Gabbay wrote:
From: Evgeny Pinchuk evgeny.pinc...@amd.com
This patch adds the topology module to the driver. The topology is exposed to
userspace through the sysfs.
The calls to add and remove a device
On 27/07/14 14:15, Oded Gabbay wrote:
On 21/07/14 01:37, Jerome Glisse wrote:
On Thu, Jul 17, 2014 at 04:29:17PM +0300, Oded Gabbay wrote:
From: Evgeny Pinchuk evgeny.pinc...@amd.com
This patch adds the topology module to the driver. The topology is exposed to
userspace through the sysfs
On 14/07/14 10:58, Christian König wrote:
Am 14.07.2014 09:38, schrieb Michel Dänzer:
On 11.07.2014 06:50, Oded Gabbay wrote:
@@ -5876,8 +5871,13 @@ int cik_ib_parse(struct radeon_device *rdev, struct
radeon_ib *ib)
*/
int cik_vm_init(struct radeon_device *rdev)
{
-/* number of VMs
On 11/07/14 22:22, Jerome Glisse wrote:
On Fri, Jul 11, 2014 at 06:56:12PM +, Bridgman, John wrote:
From: Jerome Glisse [mailto:j.gli...@gmail.com]
Sent: Friday, July 11, 2014 2:52 PM
To: Bridgman, John
Cc: Oded Gabbay; David Airlie; Deucher, Alexander; linux-
ker...@vger.kernel.org; dri-de
On 11/07/14 20:28, Joe Perches wrote:
On Fri, 2014-07-11 at 13:04 -0400, Jerome Glisse wrote:
On Fri, Jul 11, 2014 at 12:50:09AM +0300, Oded Gabbay wrote:
[]
+static long kfd_ioctl(struct file *, unsigned int, unsigned long);
Nitpick, avoid unsigned int just use unsigned.
I suggest
On 11/07/14 19:24, Jerome Glisse wrote:
On Thu, Jul 10, 2014 at 03:38:33PM -0700, Joe Perches wrote:
On Fri, 2014-07-11 at 00:50 +0300, Oded Gabbay wrote:
This patch adds the interface between the radeon driver and the kfd
driver. The interface implementation is contained in
radeon_kfd.c
On 11/07/14 19:36, Jerome Glisse wrote:
On Fri, Jul 11, 2014 at 12:50:08AM +0300, Oded Gabbay wrote:
The KFD driver should be loaded when the radeon driver is loaded and
should be finalized when the radeon driver is removed.
This patch adds a function call to initialize kfd from radeon_init
On 11/07/14 21:25, Jerome Glisse wrote:
On Fri, Jul 11, 2014 at 12:50:11AM +0300, Oded Gabbay wrote:
This patch adds the code base of the scheduler, which handles queue
creation, deletion and scheduling on the CP of the GPU.
Signed-off-by: Oded Gabbay oded.gab...@amd.com
I would rather see
On 17/07/14 15:29, Christian König wrote:
Am 17.07.2014 13:57, schrieb Oded Gabbay:
On 11/07/14 19:36, Jerome Glisse wrote:
On Fri, Jul 11, 2014 at 12:50:08AM +0300, Oded Gabbay wrote:
The KFD driver should be loaded when the radeon driver is loaded and
should be finalized when the radeon
in mm_struct and remove the use of the new notification chain in mmput.
Instead, I will try to use the mmu release notifier.
Signed-off-by: Oded Gabbay oded.gab...@amd.com
---
include/linux/mm_types.h | 14 ++
1 file changed, 14 insertions(+)
diff --git a/include/linux/mm_types.h b/include
radeon and amdkfd share the doorbell aperture.
radeon sets it up, takes the doorbells required for its own rings
and reports the setup to amdkfd.
radeon reserved doorbells are at the start of the doorbell aperture.
Signed-off-by: Oded Gabbay oded.gab...@amd.com
---
drivers/gpu/drm/radeon
To support HSA on KV, we need to limit the number of vmids and pipes
that are available for radeon's use with KV.
This patch reserves VMIDs 8-15 for amdkfd (so radeon can only use VMIDs
0-7) and also makes radeon thinks that KV has only a single MEC with a single
pipe in it
Signed-off-by: Oded
amdkfd should set interrupts for pipes 1-7.
Signed-off-by: Oded Gabbay oded.gab...@amd.com
---
drivers/gpu/drm/radeon/cik.c | 71 +---
1 file changed, 1 insertion(+), 70 deletions(-)
diff --git a/drivers/gpu/drm/radeon/cik.c b/drivers/gpu/drm/radeon/cik.c
From: Ben Goz ben@amd.com
The kernel queue module enables the amdkfd to establish kernel queues, not
exposed to user space.
The kernel queues are used for HIQ (HSA Interface Queue) and DIQ (Debug
Interface Queue) operations
Signed-off-by: Ben Goz ben@amd.com
Signed-off-by: Oded Gabbay
From: Alexey Skidanov alexey.skida...@amd.com
Signed-off-by: Alexey Skidanov alexey.skida...@amd.com
Signed-off-by: Oded Gabbay oded.gab...@amd.com
---
drivers/gpu/drm/radeon/amdkfd/kfd_chardev.c | 40 -
drivers/gpu/drm/radeon/amdkfd/kfd_priv.h| 5
2 files
From: Evgeny Pinchuk evgeny.pinc...@amd.com
Signed-off-by: Ben Goz ben@amd.com
Signed-off-by: Oded Gabbay oded.gab...@amd.com
---
drivers/gpu/drm/radeon/amdkfd/kfd_chardev.c | 46 +++--
drivers/gpu/drm/radeon/amdkfd/kfd_device.c | 2 ++
drivers/gpu/drm/radeon/amdkfd
From: Ben Goz ben@amd.com
The queue module enables allocating and initializing queues uniformly.
Signed-off-by: Ben Goz ben@amd.com
Signed-off-by: Oded Gabbay oded.gab...@amd.com
---
drivers/gpu/drm/radeon/amdkfd/Makefile| 2 +-
drivers/gpu/drm/radeon/amdkfd/kfd_priv.h | 48
This patch adds the amdkfd skeleton driver. The driver does nothing except
define a /dev/kfd device.
It returns -ENODEV on all amdkfd IOCTLs.
Signed-off-by: Oded Gabbay oded.gab...@amd.com
---
drivers/gpu/drm/radeon/Kconfig | 2 +
drivers/gpu/drm/radeon/Makefile | 2
performance counters
Signed-off-by: Oded Gabbay oded.gab...@amd.com
---
include/uapi/linux/kfd_ioctl.h | 133 +
1 file changed, 133 insertions(+)
create mode 100644 include/uapi/linux/kfd_ioctl.h
diff --git a/include/uapi/linux/kfd_ioctl.h b/include/uapi/linux
The kfd_process object is created when a process opens /dev/kfd and is closed
when the mm_struct of that process is teared-down.
Signed-off-by: Andrew Lewycky andrew.lewy...@amd.com
Signed-off-by: Oded Gabbay oded.gab...@amd.com
---
drivers/gpu/drm/radeon/amdkfd/Makefile | 4 +-
drivers/gpu
at this stage of the patchset.
Signed-off-by: Oded Gabbay oded.gab...@amd.com
---
drivers/gpu/drm/radeon/amdkfd/kfd_device.c | 80 -
drivers/gpu/drm/radeon/amdkfd/kfd_priv.h| 1 +
drivers/gpu/drm/radeon/amdkfd/kfd_process.c | 12 +
3 files changed, 92 insertions(+), 1
From: Evgeny Pinchuk evgeny.pinc...@amd.com
Signed-off-by: Evgeny Pinchuk evgeny.pinc...@amd.com
Signed-off-by: Oded Gabbay oded.gab...@amd.com
---
drivers/gpu/drm/radeon/amdkfd/kfd_chardev.c | 29 -
1 file changed, 28 insertions(+), 1 deletion(-)
diff --git
Scheduling
Signed-off-by: Ben Goz ben@amd.com
Signed-off-by: Oded Gabbay oded.gab...@amd.com
---
drivers/gpu/drm/radeon/amdkfd/kfd_module.c | 4
drivers/gpu/drm/radeon/amdkfd/kfd_priv.h | 9 +
2 files changed, 13 insertions(+)
diff --git a/drivers/gpu/drm/radeon/amdkfd
possible for us to lose
interrupts because we have no back-pressure to the hardware.
Signed-off-by: Andrew Lewycky andrew.lewy...@amd.com
Signed-off-by: Oded Gabbay oded.gab...@amd.com
---
drivers/gpu/drm/radeon/amdkfd/Makefile| 3 +-
drivers/gpu/drm/radeon/amdkfd/kfd_device.c| 16
From: Andrew Lewycky andrew.lewy...@amd.com
Signed-off-by: Andrew Lewycky andrew.lewy...@amd.com
Signed-off-by: Oded Gabbay oded.gab...@amd.com
---
drivers/gpu/drm/radeon/amdkfd/kfd_chardev.c | 51 -
1 file changed, 50 insertions(+), 1 deletion(-)
diff --git
Signed-off-by: Oded Gabbay oded.gab...@amd.com
---
CREDITS | 7 +++
MAINTAINERS | 10 ++
2 files changed, 17 insertions(+)
diff --git a/CREDITS b/CREDITS
index 03343bf..ae47364 100644
--- a/CREDITS
+++ b/CREDITS
@@ -1197,6 +1197,13 @@ S: R. Tocantins, 89 - Cristo Rei
S: 80050
From: Ben Goz ben@amd.com
The packet manager module builds PM4 packets for the sole use of the CP
scheduler. Those packets are used by the HIQ to submit runlists to the CP.
Signed-off-by: Ben Goz ben@amd.com
Signed-off-by: Oded Gabbay oded.gab...@amd.com
---
drivers/gpu/drm/radeon
From: Ben Goz ben@amd.com
The mqd_manager module handles MQD data structures. MQD stands for Memory Queue
Descriptor, which is used by the H/W to keep the usermode queue state in memory.
Signed-off-by: Ben Goz ben@amd.com
Signed-off-by: Oded Gabbay oded.gab...@amd.com
---
drivers/gpu
.
Signed-off-by: Oded Gabbay oded.gab...@amd.com
---
drivers/gpu/drm/radeon/Makefile | 1 +
drivers/gpu/drm/radeon/cik.c| 9 +
drivers/gpu/drm/radeon/cik_reg.h| 65 +
drivers/gpu/drm/radeon/cikd.h | 51 +++-
drivers/gpu/drm/radeon/radeon.h | 3 +
drivers/gpu
Implementing a lock for selecting and accessing shader engines and arrays.
This lock will make sure that radeon and amdkfd are not colliding when
accessing shader engines and arrays with GRBM_GFX_INDEX register.
Signed-off-by: Oded Gabbay oded.gab...@amd.com
---
drivers/gpu/drm/radeon/cik.c
-off-by: Oded Gabbay oded.gab...@amd.com
---
drivers/gpu/drm/radeon/amdkfd/Makefile |2 +-
drivers/gpu/drm/radeon/amdkfd/kfd_crat.h | 294 +++
drivers/gpu/drm/radeon/amdkfd/kfd_device.c |7 +
drivers/gpu/drm/radeon/amdkfd/kfd_module.c |7 +
drivers/gpu/drm/radeon
supports suspend/resume operation.
Signed-off-by: Ben Goz ben@amd.com
Signed-off-by: Oded Gabbay oded.gab...@amd.com
---
drivers/gpu/drm/radeon/amdkfd/Makefile | 2 +-
drivers/gpu/drm/radeon/amdkfd/kfd_device.c | 26 +-
.../drm/radeon/amdkfd/kfd_device_queue_manager.c | 985
From: Ben Goz ben@amd.com
Signed-off-by: Ben Goz ben@amd.com
Signed-off-by: Oded Gabbay oded.gab...@amd.com
---
drivers/gpu/drm/radeon/amdkfd/kfd_chardev.c | 133 +++-
drivers/gpu/drm/radeon/amdkfd/kfd_priv.h| 8 ++
2 files changed, 138 insertions(+), 3
@amd.com
Signed-off-by: Oded Gabbay oded.gab...@amd.com
---
drivers/gpu/drm/radeon/amdkfd/Makefile | 3 +-
drivers/gpu/drm/radeon/amdkfd/kfd_priv.h | 17 +
drivers/gpu/drm/radeon/amdkfd/kfd_process.c| 13 +
.../drm/radeon/amdkfd/kfd_process_queue_manager.c | 343
the create/destroy/update queue IOCTLs
Evgeny Pinchuk (3):
amdkfd: Add topology module to amdkfd
amdkfd: Implement the Get Clock Counters IOCTL
amdkfd: Implement the PMC Acquire/Release IOCTLs
Oded Gabbay (10):
mm: Add kfd_process pointer to mm_struct
drm/radeon: reduce number of free
if I can drop the kfd_process
in mm_struct and remove the use of the new notification chain in mmput.
Instead, I will try to use the mmu release notifier.
Signed-off-by: Oded Gabbay oded.gab...@amd.com
---
include/linux/mm_types.h | 14 ++
1 file changed, 14 insertions(+)
diff --git
On 17/07/14 17:12, Jerome Glisse wrote:
On Thu, Jul 17, 2014 at 04:57:58PM +0300, Oded Gabbay wrote:
Forgot to add mm mailing list. Sorry.
This patch enables the amdkfd driver to retrieve the kfd_process
object from the process's mm_struct. This is needed because kfd_process
lifespan is bound
with mmu_notifier.
Signed-off-by: Peter Zijlstra pet...@infradead.org
Signed-off-by: Jérôme Glisse jgli...@redhat.com
Signed-off-by: Oded Gabbay oded.gab...@amd.com
---
include/linux/mmu_notifier.h | 6 ++
mm/mmu_notifier.c| 40 +++-
2 files changed, 45
On 21/07/14 05:33, Jerome Glisse wrote:
On Thu, Jul 17, 2014 at 04:29:21PM +0300, Oded Gabbay wrote:
From: Ben Goz ben@amd.com
The mqd_manager module handles MQD data structures. MQD stands for Memory
Queue Descriptor, which is used by the H/W to keep the usermode queue state
On 21/07/14 02:02, Jerome Glisse wrote:
On Thu, Jul 17, 2014 at 04:29:18PM +0300, Oded Gabbay wrote:
From: Andrew Lewycky andrew.lewy...@amd.com
This patch adds the process module and 4 helper modules:
- kfd_process, which handles process which open /dev/kfd
- kfd_doorbell, which provides
On 20/07/14 20:09, Jerome Glisse wrote:
On Thu, Jul 17, 2014 at 04:29:16PM +0300, Oded Gabbay wrote:
This patch adds the amdkfd skeleton driver. The driver does nothing except
define a /dev/kfd device.
It returns -ENODEV on all amdkfd IOCTLs.
Signed-off-by: Oded Gabbay oded.gab...@amd.com
On 20/07/14 19:54, Jerome Glisse wrote:
On Thu, Jul 17, 2014 at 04:29:15PM +0300, Oded Gabbay wrote:
- KFD_IOC_GET_VERSION:
Retrieves the interface version of amdkfd
- KFD_IOC_CREATE_QUEUE:
Creates a usermode queue that runs on a specific GPU device
- KFD_IOC_DESTROY_QUEUE
On 20/07/14 20:35, Jerome Glisse wrote:
On Thu, Jul 17, 2014 at 04:29:13PM +0300, Oded Gabbay wrote:
This patch adds the interface between the radeon driver and the amdkfd
driver.
The interface implementation is contained in radeon_kfd.c and radeon_kfd.h.
The interface itself
, schrieb Oded Gabbay:
On 20/07/14 20:46, Jerome Glisse wrote:
[snip!!]
My BlackBerry thumb thanks you ;)
The main questions here are if it's avoid able to pin down
the memory and if the memory is pinned down at driver load,
by request from userspace or by anything else.
As far as I can see
On 24/07/14 21:47, Jerome Glisse wrote:
On Thu, Jul 24, 2014 at 01:35:53PM -0400, Alex Deucher wrote:
On Thu, Jul 24, 2014 at 11:44 AM, Jerome Glisse j.gli...@gmail.com wrote:
On Thu, Jul 24, 2014 at 01:01:41AM +0300, Oded Gabbay wrote:
On 24/07/14 00:46, Bridgman, John wrote:
-Original
functions
v3: add documentation
Signed-off-by: Ben Goz ben@amd.com
Signed-off-by: Oded Gabbay oded.gab...@amd.com
---
drivers/gpu/drm/radeon/amdkfd/Makefile | 2 +-
drivers/gpu/drm/radeon/amdkfd/kfd_packet_manager.c | 495 +
drivers/gpu/drm/radeon/amdkfd
supports suspend/resume operation.
v3: change device_init
v3: use new gart allocation functions
v3: Add documentation
Signed-off-by: Ben Goz ben@amd.com
Signed-off-by: Oded Gabbay oded.gab...@amd.com
---
drivers/gpu/drm/radeon/amdkfd/Makefile | 2 +-
drivers/gpu/drm/radeon/amdkfd
From: Alexey Skidanov alexey.skida...@amd.com
v3: fix debug msg
Signed-off-by: Alexey Skidanov alexey.skida...@amd.com
Signed-off-by: Oded Gabbay oded.gab...@amd.com
---
drivers/gpu/drm/radeon/amdkfd/kfd_chardev.c | 47 -
drivers/gpu/drm/radeon/amdkfd/kfd_priv.h
code to support mmap of doorbell pages from userspace
v3: Add documentation for apertures
Signed-off-by: Andrew Lewycky andrew.lewy...@amd.com
Signed-off-by: Oded Gabbay oded.gab...@amd.com
---
drivers/gpu/drm/radeon/amdkfd/Makefile | 4 +-
drivers/gpu/drm/radeon/amdkfd/kfd_aperture.c | 350
From: Andrew Lewycky andrew.lewy...@amd.com
Signed-off-by: Andrew Lewycky andrew.lewy...@amd.com
Signed-off-by: Oded Gabbay oded.gab...@amd.com
---
drivers/gpu/drm/radeon/amdkfd/kfd_chardev.c | 51 -
1 file changed, 50 insertions(+), 1 deletion(-)
diff --git
From: Ben Goz ben@amd.com
v3: remove use of internal typedefs
v3: fix debug prints
v3: add checks for parameters
v3: use doorbell address from user
Signed-off-by: Ben Goz ben@amd.com
Signed-off-by: Oded Gabbay oded.gab...@amd.com
---
drivers/gpu/drm/radeon/amdkfd/kfd_chardev.c
functions
Signed-off-by: Ben Goz ben@amd.com
Signed-off-by: Oded Gabbay oded.gab...@amd.com
---
drivers/gpu/drm/radeon/amdkfd/Makefile | 3 +-
.../drm/radeon/amdkfd/kfd_device_queue_manager.h | 101 +++
drivers/gpu/drm/radeon/amdkfd/kfd_kernel_queue.c | 330 ++
drivers
queues per process instead of define
v3: use doorbell address from user
Signed-off-by: Ben Goz ben@amd.com
Signed-off-by: Oded Gabbay oded.gab...@amd.com
---
drivers/gpu/drm/radeon/amdkfd/Makefile | 3 +-
drivers/gpu/drm/radeon/amdkfd/kfd_priv.h | 17 +
drivers/gpu/drm
-by: Ben Goz ben@amd.com
Signed-off-by: Oded Gabbay oded.gab...@amd.com
---
drivers/gpu/drm/radeon/amdkfd/kfd_module.c | 12
drivers/gpu/drm/radeon/amdkfd/kfd_priv.h | 29 +
2 files changed, 41 insertions(+)
diff --git a/drivers/gpu/drm/radeon/amdkfd
to use linux macros
v3: use new gart allocation functions
v3: Add documentation
Signed-off-by: Ben Goz ben@amd.com
Signed-off-by: Oded Gabbay oded.gab...@amd.com
---
drivers/gpu/drm/radeon/amdkfd/Makefile | 2 +-
drivers/gpu/drm/radeon/amdkfd/cik_regs.h| 220
which are compliant with the HSA
spec).
To recap, amdkfd exposes a different kind of topology than the one exposed by
/sys/devices/system/cpu/cpu even though it may contain similar information.
Signed-off-by: Evgeny Pinchuk evgeny.pinc...@amd.com
Signed-off-by: Oded Gabbay oded.gab...@amd.com
of the amdkfd device
- KFD_IOC_UPDATE_QUEUE:
Updates configuration of an existing usermode queue
(v3) remove pragma pack
(v3) remove pmc ioctls
(v3) add parameter for doorbell offset
(v3) add comment on counters
Signed-off-by: Oded Gabbay oded.gab...@amd.com
---
include/uapi/linux
Implementing a lock for selecting and accessing shader engines and arrays.
This lock will make sure that radeon and amdkfd are not colliding when
accessing shader engines and arrays with GRBM_GFX_INDEX register.
Signed-off-by: Oded Gabbay oded.gab...@amd.com
---
drivers/gpu/drm/radeon/cik.c
Signed-off-by: Oded Gabbay oded.gab...@amd.com
---
CREDITS | 7 +++
MAINTAINERS | 10 ++
2 files changed, 17 insertions(+)
diff --git a/CREDITS b/CREDITS
index 28ee151..e9628d5 100644
--- a/CREDITS
+++ b/CREDITS
@@ -1197,6 +1197,13 @@ S: R. Tocantins, 89 - Cristo Rei
S: 80050
for us to lose
interrupts because we have no back-pressure to the hardware.
v3: change device init
v3: make sure spin lock is taken only if init is complete
v3: move bool to end of struct
Signed-off-by: Andrew Lewycky andrew.lewy...@amd.com
Signed-off-by: Oded Gabbay oded.gab...@amd.com
This patch adds the amdkfd skeleton driver. The driver does nothing except
define a /dev/kfd device.
It returns -ENODEV on all amdkfd IOCTLs.
(v3) move bool to end of struct
(v3) remove pmc ioctls
(v3) add meaningful error message for ioctl error
Signed-off-by: Oded Gabbay oded.gab...@amd.com
radeon and amdkfd share the doorbell aperture.
radeon sets it up, takes the doorbells required for its own rings
and reports the setup to amdkfd.
radeon reserved doorbells are at the start of the doorbell aperture.
Signed-off-by: Oded Gabbay oded.gab...@amd.com
---
drivers/gpu/drm/radeon
to use linux macros
(v3) Add documentation for the interface
Signed-off-by: Oded Gabbay oded.gab...@amd.com
---
drivers/gpu/drm/radeon/Makefile | 1 +
drivers/gpu/drm/radeon/cik.c| 9 +
drivers/gpu/drm/radeon/cik_reg.h| 65 +
drivers/gpu/drm/radeon/cikd.h | 51
of the patchset.
Signed-off-by: Oded Gabbay oded.gab...@amd.com
---
drivers/gpu/drm/radeon/amdkfd/kfd_device.c | 86 -
drivers/gpu/drm/radeon/amdkfd/kfd_priv.h| 1 +
drivers/gpu/drm/radeon/amdkfd/kfd_process.c | 12
3 files changed, 98 insertions(+), 1
From: Evgeny Pinchuk evgeny.pinc...@amd.com
Signed-off-by: Evgeny Pinchuk evgeny.pinc...@amd.com
Signed-off-by: Oded Gabbay oded.gab...@amd.com
---
drivers/gpu/drm/radeon/amdkfd/kfd_chardev.c | 29 -
1 file changed, 28 insertions(+), 1 deletion(-)
diff --git
amdkfd should set interrupts for pipes 1-7.
Signed-off-by: Oded Gabbay oded.gab...@amd.com
---
drivers/gpu/drm/radeon/cik.c | 71 +---
1 file changed, 1 insertion(+), 70 deletions(-)
diff --git a/drivers/gpu/drm/radeon/cik.c b/drivers/gpu/drm/radeon/cik.c
packet manager module
amdkfd: Add process queue manager module
amdkfd: Add device queue manager module
amdkfd: Implement the create/destroy/update queue IOCTLs
Evgeny Pinchuk (2):
amdkfd: Add topology module to amdkfd
amdkfd: Implement the Get Clock Counters IOCTL
Oded Gabbay (9):
drm
From: Ben Goz ben@amd.com
The queue module enables allocating and initializing queues uniformly.
v3: remove typedef
v3: break pr_debug to one line
v3: remove memset
v3: add documentation
Signed-off-by: Ben Goz ben@amd.com
Signed-off-by: Oded Gabbay oded.gab...@amd.com
---
drivers/gpu
for static vmid allocation in radeon
Signed-off-by: Oded Gabbay oded.gab...@amd.com
---
drivers/gpu/drm/radeon/cik.c | 48 +--
drivers/gpu/drm/radeon/cikd.h | 2 ++
2 files changed, 26 insertions(+), 24 deletions(-)
diff --git a/drivers/gpu/drm/radeon/cik.c b
On 05/08/14 20:11, David Herrmann wrote:
Hi
On Tue, Aug 5, 2014 at 5:30 PM, Oded Gabbay oded.gab...@amd.com wrote:
Hi,
Here is the v3 patch set of amdkfd.
This version contains changes and fixes to code, as agreed on during the
review
of the v2 patch set.
The major changes
On 29/07/14 19:18, Joerg Roedel wrote:
From: Joerg Roedel jroe...@suse.de
Add calls to the new mmu_notifier_invalidate_range()
function to all places if the VMM that need it.
Signed-off-by: Joerg Roedel jroe...@suse.de
---
include/linux/mmu_notifier.h | 28
Hi,
Could anyone please review this patch ?
Thanks,
Oded
On 18/10/14 00:43, Oded Gabbay wrote:
This patch makes use of the new delayed mmu release notifier feature in
mm code. This is necessary because on the one hand amd_iommu_unbind_pasid
must be called explicitly during the tear
notifiers.
4. (new code in this patch) amd_iommu_unbind_pasid() sets a delayed notifier,
using the delayed mmu release notifier feature (new in 3.17),
which does the actual release later, after the iteration over the list of
mmu notifiers is over.
Signed-off-by: Oded Gabbay oded.gab...@amd.com
Sure, no problem
Oded
On 10/27/2014 05:13 PM, Joerg Roedel wrote:
Hi Oded,
can you please test these patches with the KFD driver and make sure
nothing breaks for you? I really like this improvement and it would be
great to send it upstream for v3.19.
Thanks,
Joerg
On Fri, Oct
Buy Kaveri ;)
I'm sure Intel will be happy to contribute some $$$ to AMD ;)
Oded
On 10/27/2014 05:35 PM, Jesse Barnes wrote:
Thanks, I have no way of testing this, but I'm hopeful. :)
Jesse
On Mon, 27 Oct 2014 17:15:45 +0200
Oded Gabbay oded.gab...@amd.com wrote:
Sure, no problem
Hi Joerg and Jesse,
I tested our amdkfd driver with your patches applied (kernel 3.17.1).
I run OpenCL tests, Aparapi/Sumatra (Java) and OpenMP
All tests passed and I didn't see any kernel error messages.
So:
Tested-by: Oded Gabbay oded.gab...@amd.com
Oded
On 10/27/2014 05:35 PM, Jesse
On 11/21/2014 12:24 PM, Paul Bolle wrote:
On Sat, 2014-11-08 at 20:37 +0200, Oded Gabbay wrote:
This patch adds the amdkfd skeleton driver. The driver does nothing except
define a /dev/kfd device.
It returns -ENODEV on all amdkfd IOCTLs.
v3: Move bool field to the end of structure
On 11/21/2014 07:04 PM, Jim Davis wrote:
Building with the attached random configuration file,
drivers/gpu/drm/amd/amdkfd/kfd_doorbell.c: In function ‘kfd_doorbell_init’:
drivers/gpu/drm/amd/amdkfd/kfd_doorbell.c:97:2: error: implicit
declaration of function ‘ioremap’
This patch fixes a compilation error when using certain configuration by
including the file io.h in kfd_doorbell.c
Signed-off-by: Oded Gabbay oded.gab...@amd.com
---
drivers/gpu/drm/amd/amdkfd/kfd_doorbell.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/drivers/gpu/drm/amd/amdkfd
Signed-off-by: Oded Gabbay oded.gab...@amd.com
---
drivers/gpu/drm/amd/amdkfd/Kconfig | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/gpu/drm/amd/amdkfd/Kconfig
b/drivers/gpu/drm/amd/amdkfd/Kconfig
index e13c67c..8dfac37 100644
--- a/drivers/gpu/drm/amd/amdkfd/Kconfig
This patch removes the dependency of amdkfd upon DRM_AMDGPU symbol in amdkfd's
Kconfig file.
This is done because amdgpu driver is not yet upstreamed and therefore,
DRM_AMDGPU symbol is not present in any Kconfig file.
Reviewed-by: Alex Deucher alexander.deuc...@amd.com
Signed-off-by: Oded
101 - 200 of 1658 matches
Mail list logo