Hi,
Re-sending this patch-set following the release of our user-space TPC
compiler and runtime library.
I would appreciate a review on this.
Thanks,
Oded
Oded Gabbay (1):
habanalabs: define uAPI to export FD for DMA-BUF
Tomer Tayar (1):
habanalabs: add support for dma-buf exporter
created to match that allocation.
Signed-off-by: Oded Gabbay
Reviewed-by: Tomer Tayar
Reviewed-by: Greg Kroah-Hartman
---
include/uapi/misc/habanalabs.h | 28 +++-
1 file changed, 27 insertions(+), 1 deletion(-)
diff --git a/include/uapi/misc/habanalabs.h b/include/uapi/misc
ance doesn't allow p2p.
Signed-off-by: Tomer Tayar
Reviewed-by: Oded Gabbay
Reviewed-by: Gal Pressman
Reviewed-by: Greg Kroah-Hartman
Signed-off-by: Oded Gabbay
---
drivers/misc/habanalabs/Kconfig | 1 +
drivers/misc/habanalabs/common/habanalabs.h | 22 +
drivers/m
On Tue, Sep 14, 2021 at 5:18 PM Daniel Vetter wrote:
>
> On Sun, Sep 12, 2021 at 07:53:07PM +0300, Oded Gabbay wrote:
> > Hi,
> > Re-sending this patch-set following the release of our user-space TPC
> > compiler and runtime library.
> >
> > I would apprecia
On Tue, Sep 14, 2021 at 7:12 PM Jason Gunthorpe wrote:
>
> On Tue, Sep 14, 2021 at 04:18:31PM +0200, Daniel Vetter wrote:
> > On Sun, Sep 12, 2021 at 07:53:07PM +0300, Oded Gabbay wrote:
> > > Hi,
> > > Re-sending this patch-set following the release of our use
On Thu, Sep 16, 2021 at 3:31 PM Daniel Vetter wrote:
>
> On Wed, Sep 15, 2021 at 10:45:36AM +0300, Oded Gabbay wrote:
> > On Tue, Sep 14, 2021 at 7:12 PM Jason Gunthorpe wrote:
> > >
> > > On Tue, Sep 14, 2021 at 04:18:31PM +0200, Daniel Vetter wrote:
> > &g
On Thu, Sep 16, 2021 at 3:44 PM Oded Gabbay wrote:
>
> On Thu, Sep 16, 2021 at 3:31 PM Daniel Vetter wrote:
> >
> > Maybe I got the device security model all wrong, but I thought Guadi is
> > single user, and the only thing it protects is the system against the
> &
On Fri, Sep 17, 2021 at 3:30 PM Daniel Vetter wrote:
>
> On Thu, Sep 16, 2021 at 10:10:14AM -0300, Jason Gunthorpe wrote:
> > On Thu, Sep 16, 2021 at 02:31:34PM +0200, Daniel Vetter wrote:
> > > On Wed, Sep 15, 2021 at 10:45:36AM +0300, Oded Gabbay wrote:
> > > >
On Sat, Sep 18, 2021 at 11:38 AM Oded Gabbay wrote:
>
> On Fri, Sep 17, 2021 at 3:30 PM Daniel Vetter wrote:
> >
> > On Thu, Sep 16, 2021 at 10:10:14AM -0300, Jason Gunthorpe wrote:
> > > On Thu, Sep 16, 2021 at 02:31:34PM +0200, Daniel Vetter wrote:
> > >
On Thu, Sep 23, 2021 at 12:22 PM Oded Gabbay wrote:
>
> On Sat, Sep 18, 2021 at 11:38 AM Oded Gabbay wrote:
> >
> > On Fri, Sep 17, 2021 at 3:30 PM Daniel Vetter wrote:
> > >
> > > On Thu, Sep 16, 2021 at 10:10:14AM -0300, Jason Gunthorpe wrote:
> > &g
On Tue, Sep 28, 2021 at 8:13 PM Jason Gunthorpe wrote:
>
> On Sun, Sep 12, 2021 at 07:53:08PM +0300, Oded Gabbay wrote:
> > /* HL_MEM_OP_* */
> > __u32 op;
> > - /* HL_MEM_* flags */
> > + /* HL_MEM_* flags.
> > + * For the HL_MEM_OP
On Tue, Sep 28, 2021 at 8:36 PM Jason Gunthorpe wrote:
>
> On Sun, Sep 12, 2021 at 07:53:09PM +0300, Oded Gabbay wrote:
> > From: Tomer Tayar
> >
> > Implement the calls to the dma-buf kernel api to create a dma-buf
> > object backed by FD.
> >
> > We b
On Wed, Sep 29, 2021 at 12:17 AM Oded Gabbay wrote:
>
> On Tue, Sep 28, 2021 at 8:36 PM Jason Gunthorpe wrote:
> >
> > On Sun, Sep 12, 2021 at 07:53:09PM +0300, Oded Gabbay wrote:
> > > From: Tomer Tayar
> > >
> > > Implement the calls to the dma-bu
Hi,
I'm sending v7 after the latest review from Jason.
All the changes are detailed in the commit messages.
Dave, I'll appreciate if you can also a-b this patchset.
Thanks,
Oded
Oded Gabbay (1):
habanalabs: define uAPI to export FD for DMA-BUF
Tomer Tayar (1):
habanalabs: add s
created to match that allocation.
Signed-off-by: Oded Gabbay
Reviewed-by: Tomer Tayar
Reviewed-by: Greg Kroah-Hartman
Acked-by: Daniel Vetter
---
Changes in v7:
- Change the type of the fd variable returned from IOCTL to be __s32
include/uapi/misc/habanalabs.h | 28 +++-
1
ance doesn't allow p2p.
Signed-off-by: Tomer Tayar
Reviewed-by: Oded Gabbay
Reviewed-by: Gal Pressman
Reviewed-by: Greg Kroah-Hartman
Acked-by: Daniel Vetter
Signed-off-by: Oded Gabbay
---
Changes in v7:
- rename structure hl_dmabuf_wrapper to hl_dmabuf_priv
- remove change that wasn
On Sun, Oct 3, 2021 at 12:08 PM Oded Gabbay wrote:
>
> Hi,
> I'm sending v7 after the latest review from Jason.
> All the changes are detailed in the commit messages.
>
> Dave, I'll appreciate if you can also a-b this patchset.
>
> Thanks,
> Oded
Hi,
I w
On Mon, Jun 21, 2021 at 5:12 PM Jason Gunthorpe wrote:
>
> On Mon, Jun 21, 2021 at 03:02:10PM +0200, Greg KH wrote:
> > On Mon, Jun 21, 2021 at 02:28:48PM +0200, Daniel Vetter wrote:
>
> > > Also I'm wondering which is the other driver that we share buffers
> > > with. The gaudi stuff doesn't have
On Mon, Jun 21, 2021 at 9:27 PM Daniel Vetter wrote:
>
> On Mon, Jun 21, 2021 at 7:55 PM Jason Gunthorpe wrote:
> > On Mon, Jun 21, 2021 at 07:26:14PM +0300, Oded Gabbay wrote:
> > > On Mon, Jun 21, 2021 at 5:12 PM Jason Gunthorpe wrote:
> > > >
> > >
On Tue, Jun 22, 2021 at 9:37 AM Christian König
wrote:
>
> Am 22.06.21 um 01:29 schrieb Jason Gunthorpe:
> > On Mon, Jun 21, 2021 at 10:24:16PM +0300, Oded Gabbay wrote:
> >
> >> Another thing I want to emphasize is that we are doing p2p only
> >> through the
On Tue, Jun 22, 2021 at 3:01 PM Jason Gunthorpe wrote:
>
> On Tue, Jun 22, 2021 at 11:42:27AM +0300, Oded Gabbay wrote:
> > On Tue, Jun 22, 2021 at 9:37 AM Christian König
> > wrote:
> > >
> > > Am 22.06.21 um 01:29 schrieb Jason Gunthorpe:
> > > >
On Tue, Jun 22, 2021 at 3:15 PM Jason Gunthorpe wrote:
>
> On Tue, Jun 22, 2021 at 03:04:30PM +0300, Oded Gabbay wrote:
> > On Tue, Jun 22, 2021 at 3:01 PM Jason Gunthorpe wrote:
> > >
> > > On Tue, Jun 22, 2021 at 11:42:27AM +0300, Oded Gabbay wrote:
> >
On Tue, Jun 22, 2021 at 6:11 PM Jason Gunthorpe wrote:
>
> On Tue, Jun 22, 2021 at 04:12:26PM +0300, Oded Gabbay wrote:
>
> > > 1) Setting sg_page to NULL
> > > 2) 'mapping' pages for P2P DMA without going through the iommu
> > > 3) Allowing P2P DMA
On Tue, Jun 22, 2021 at 6:28 PM Jason Gunthorpe wrote:
>
> On Tue, Jun 22, 2021 at 05:24:08PM +0200, Christian König wrote:
>
> > > > I will take two GAUDI devices and use one as an exporter and one as an
> > > > importer. I want to see that the solution works end-to-end, with real
> > > > device
On Tue, Jun 22, 2021 at 6:31 PM Christian König
wrote:
>
>
>
> Am 22.06.21 um 17:28 schrieb Jason Gunthorpe:
> > On Tue, Jun 22, 2021 at 05:24:08PM +0200, Christian König wrote:
> >
> I will take two GAUDI devices and use one as an exporter and one as an
> importer. I want to see that th
On Wed, Jun 23, 2021 at 11:57 AM Christian König
wrote:
>
> Am 22.06.21 um 18:05 schrieb Jason Gunthorpe:
> > On Tue, Jun 22, 2021 at 05:48:10PM +0200, Christian König wrote:
> >> Am 22.06.21 um 17:40 schrieb Jason Gunthorpe:
> >>> On Tue, Jun 22, 2021 at 05:29:01PM +0200, Christian König wrote:
>
On Wed, Jun 23, 2021 at 9:24 PM Jason Gunthorpe wrote:
>
> On Wed, Jun 23, 2021 at 10:57:35AM +0200, Christian König wrote:
>
> > > > No it isn't. It makes devices depend on allocating struct pages for
> > > > their
> > > > BARs which is not necessary nor desired.
> > > Which dramatically reduces
On Wed, Jun 23, 2021 at 9:50 PM Jason Gunthorpe wrote:
>
> On Wed, Jun 23, 2021 at 09:43:04PM +0300, Oded Gabbay wrote:
>
> > Can you please explain why it is so important to (allow) access them
> > through the CPU ?
>
> It is not so much important, as it reflects
On Wed, Jun 23, 2021 at 10:34 PM Jason Gunthorpe wrote:
>
> On Wed, Jun 23, 2021 at 10:00:29PM +0300, Oded Gabbay wrote:
> > On Wed, Jun 23, 2021 at 9:50 PM Jason Gunthorpe wrote:
> > >
> > > On Wed, Jun 23, 2021 at 09:43:04PM +0300, Oded Gabbay wrote:
> > >
on a bare-metal environment with IOMMU enabled, on a sky-lake CPU
with a white-listed PCIe bridge (to make the pci_p2pdma_distance_many happy).
Greg, I hope this will be good enough for you to merge this code.
Thanks,
Oded
Oded Gabbay (1):
habanalabs: define uAPI to export FD for DMA-BUF
Tome
created to match that allocation.
Signed-off-by: Oded Gabbay
Reviewed-by: Tomer Tayar
---
include/uapi/misc/habanalabs.h | 28 +++-
1 file changed, 27 insertions(+), 1 deletion(-)
diff --git a/include/uapi/misc/habanalabs.h b/include/uapi/misc/habanalabs.h
index 18765eb75b65
low p2p.
Signed-off-by: Tomer Tayar
Reviewed-by: Oded Gabbay
Reviewed-by: Gal Pressman
Signed-off-by: Oded Gabbay
---
Changes in v4:
- Using dma_map_resources() to map the physical address of the PCI BAR
to a DMA'able address that the other device can use
- Check the p2p distance u
On Mon, Jul 5, 2021 at 7:52 PM Jason Gunthorpe wrote:
>
> On Mon, Jul 05, 2021 at 04:03:14PM +0300, Oded Gabbay wrote:
>
> > + rc = sg_alloc_table(*sgt, nents, GFP_KERNEL | __GFP_ZERO);
> > + if (rc)
> > + goto error_free;
>
> If you are not g
On Tue, Jul 6, 2021 at 11:40 AM Daniel Vetter wrote:
>
> On Mon, Jul 05, 2021 at 04:03:12PM +0300, Oded Gabbay wrote:
> > Hi,
> > I'm sending v4 of this patch-set following the long email thread.
> > I want to thank Jason for reviewing v3 and pointing out the errors
On Tue, Jul 6, 2021 at 3:23 PM Daniel Vetter wrote:
>
> On Tue, Jul 06, 2021 at 02:21:10PM +0200, Christoph Hellwig wrote:
> > On Tue, Jul 06, 2021 at 10:40:37AM +0200, Daniel Vetter wrote:
> > > > Greg, I hope this will be good enough for you to merge this code.
> > >
> > > So we're officially go
On Tue, Jul 6, 2021 at 4:17 PM Daniel Vetter wrote:
>
> On Tue, Jul 6, 2021 at 2:46 PM Oded Gabbay wrote:
> >
> > On Tue, Jul 6, 2021 at 3:23 PM Daniel Vetter wrote:
> > >
> > > On Tue, Jul 06, 2021 at 02:21:10PM +0200, Christoph Hellwig wrote:
> > &g
On Tue, Jul 6, 2021, 16:54 Jason Gunthorpe wrote:
> On Tue, Jul 06, 2021 at 12:44:49PM +0300, Oded Gabbay wrote:
>
> > > > + /* In case we got a large memory area to export, we need to
> divide it
> > > > + * to smaller areas because each e
fixes are in the changelog in the commit message of
the second patch.
There was one issue with your proposal to set the orig_nents to 0. I did
that, but I also had to restore it to nents before calling sg_free_table
because that function uses orig_nents to iterate.
Thanks,
Oded
Oded Gabb
created to match that allocation.
Signed-off-by: Oded Gabbay
Reviewed-by: Tomer Tayar
---
include/uapi/misc/habanalabs.h | 28 +++-
1 file changed, 27 insertions(+), 1 deletion(-)
diff --git a/include/uapi/misc/habanalabs.h b/include/uapi/misc/habanalabs.h
index 18765eb75b65
ance doesn't allow p2p.
Signed-off-by: Tomer Tayar
Reviewed-by: Oded Gabbay
Reviewed-by: Gal Pressman
Signed-off-by: Oded Gabbay
---
Changes in v5:
- Use dma_get_max_seg_size() to get the max segment size of the importer.
Use that value when we convert the pages array into an SGT, to
On Fri, Dec 23, 2022 at 9:37 PM Shashank Sharma wrote:
>
> This patch adds skeleton code for usermode queue creation. It
> typically contains:
> - A new structure to keep all the user queue data in one place.
> - An IOCTL function to create/free a usermode queue.
> - A function to generate unique
On Fri, Jun 17, 2022 at 8:20 PM Sierra Guiza, Alejandro (Alex)
wrote:
>
>
> On 6/17/2022 4:40 AM, David Hildenbrand wrote:
> > On 31.05.22 22:00, Alex Sierra wrote:
> >> Device memory that is cache coherent from device and CPU point of view.
> >> This is used on platforms that have an advanced sys
On Mon, Jun 20, 2022 at 3:33 AM Alistair Popple wrote:
>
>
> Oded Gabbay writes:
>
> > On Fri, Jun 17, 2022 at 8:20 PM Sierra Guiza, Alejandro (Alex)
> > wrote:
> >>
> >>
> >> On 6/17/2022 4:40 AM, David Hildenbrand wrote:
> >> > O
On Mon, Jun 20, 2022 at 11:50 AM Alistair Popple wrote:
>
>
> Oded Gabbay writes:
>
> > On Mon, Jun 20, 2022 at 3:33 AM Alistair Popple wrote:
> >>
> >>
> >> Oded Gabbay writes:
> >>
> >> > On Fri, Jun 17, 2022 at 8:20 PM Sierra
q->properties.is_active = true;
> retval = mqd_mgr->load_mqd(mqd_mgr, q->mqd, q->pipe,
> - q->queue, &q->properties,
> - q->process->mm);
> +
On Wed, Jul 11, 2018 at 3:41 PM Arnd Bergmann wrote:
>
> getrawmonotonic64() and get_monotonic_boottime64() are deprecated
> because of the nonstandard naming.
>
> The replacement functions ktime_get_raw_ns() and ktime_get_boot_ns()
> also simplify the callers.
>
> Reviewed-by: Felix Kuehling .
>
Hi Felix,
Thanks for the patch-set. Applied to -next.
Oded
On Thu, Jul 12, 2018 at 10:52 AM Christian König
wrote:
>
> Patches which don't already have my rb are Acked-by: Christian König
> .
>
> Regards,
> Christian.
>
> Am 12.07.2018 um 04:32 schrieb Felix Kuehling:
> > This series fixes some
Hi Felix,
Thanks for the patch-set. Applied to -next.
Oded
On Fri, Jul 13, 2018 at 11:24 PM Alex Deucher wrote:
>
> On Fri, Jul 13, 2018 at 4:17 PM, Felix Kuehling
> wrote:
> > Raven refers to Ryzen APUs with integrated GFXv9 GPU.
> > This patch series completes Raven support for KFD:
> >
> > *
_CU_MASK \
> + AMDKFD_IOW(0x1A, struct kfd_ioctl_set_cu_mask_args)
> +
> #define AMDKFD_COMMAND_START 0x01
> -#define AMDKFD_COMMAND_END 0x1A
> +#define AMDKFD_COMMAND_END 0x1B
>
> #endif
> --
> 2.7.4
>
++ b/drivers/gpu/drm/amd/amdkfd/kfd_process_queue_manager.c
> @@ -325,6 +325,8 @@ int pqm_destroy_queue(struct process_queue_manager *pqm,
> unsigned int qid)
> if (retval != -ETIME)
> goto err_destroy_queue;
>
Hi Felix,
Both patches applied to -next.
Oded
On Tue, Jul 17, 2018 at 7:14 PM Alex Deucher wrote:
>
> On Mon, Jul 16, 2018 at 7:10 PM, Felix Kuehling
> wrote:
> > User mode queue submissions don't go through KFD. Therefore we don't
> > know exactly when compute is idle or not idle. We use the ex
Hi Dave,
This is amdkfd pull for 4.19. The major changes are:
- Add Raven support. Raven refers to Ryzen APUs with integrated GFXv9 GPU.
- Integrate GPU reset support
In addition, there are a couple of small fixes and improvements, such as:
- Better handling and reporting to user of VM faults
-
Do you have IOMMUv2 enabled on this chipset ?
If not, amdkfd won't include raven support.
Oded
On Mon, Jul 30, 2018 at 4:33 PM Paul Menzel
wrote:
>
> Dear Felix, dear Yong,
>
>
> On 07/13/18 22:17, Felix Kühling wrote:
> > From: Yong Zhao
> >
> > Add DID and kfd_device_info for Raven.
> >
> > Si
On Wed, Aug 22, 2018 at 7:44 PM Linus Torvalds
wrote:
> One of the complex ones is the amdgpu driver. It does a
> "use_mm(mmptr)" deep deep in the guts of a macro that ends up being
> used in fa few places, and it's very hard to tell if it's right.
>
> It looks almost certainly buggy (there is no
On Wed, Aug 22, 2018 at 10:58 PM Linus Torvalds
wrote:
>
> On Wed, Aug 22, 2018 at 12:37 PM Oded Gabbay wrote:
> >
> > Having said that, I think we *are* protected by the mmu_notifier
> > release because if the process suddenly dies, we will gracefully clean
> > th
On Thu, Aug 23, 2018 at 4:34 AM David Airlie wrote:
>
> On Thu, Aug 23, 2018 at 8:25 AM, Felix Kuehling
> wrote:
> > Hi all,
> >
> > Oded has offered to make me co-maintainer of KFD, as he's super busy at
> > work and less responsive than he used to be.
> >
> > At the same time we're about to se
On Tue, Aug 6, 2019 at 10:31 AM Christoph Hellwig wrote:
>
> Btw, who maintains amkfd these days? MAINTAINERS still lists
> Oded, but he seems to have moved on to Habanalabs and maintains that
> drivers now while not having any action on amdkfd for over a year.
I've sent a patch to update the MA
On Wed, Feb 8, 2017 at 6:23 PM, Andres Rodriguez wrote:
> Hey Felix,
>
> Thanks for the pointer to the ROCm mqd commit. I like that the workarounds
> are easy to spot. I'll add that to a new patch series I'm working on for
> some bug-fixes for perf being lower on pipes other than pipe 0.
>
> I hav
we should merge this patch-set.
Oded
On Wed, Feb 8, 2017 at 9:47 PM, Andres Rodriguez wrote:
> Thank you Oded.
>
> - Andres
>
>
> On 2017-02-08 02:32 PM, Oded Gabbay wrote:
>>
>> On Wed, Feb 8, 2017 at 6:23 PM, Andres Rodriguez
>> wrote:
>>>
>&g
lid queue enabled by amdgpu: %d\n", i);
> break;
> }
>
> John/Felix,
>
> Any chance I could borrow a carrizo/kaveri for a few days? Or maybe you
> could help me run some final tests on this patch series?
>
> - Andres
>
t; Where can I find a repo with kfdtest?
>
> I tried looking here bit couldn't find it:
>
> https://cgit.freedesktop.org/~gabbayo/
>
> -Andres
>
>
>
> On 2017-02-10 05:35 AM, Oded Gabbay wrote:
>>
>> So the warning in dmesg is gone of course, but the test
Hi Andres,
A little suggestion, if I may.
Could you please send future versions of this patch-set with [PATCH
vX], where X is the version number ?
That would help (at least me) to track the evolution of this patch-set quicker.
And another thing, if you haven't done it so far (because I didn't
chec
On Fri, Apr 14, 2017 at 12:35 AM, Andres Rodriguez wrote:
> Make amdgpu the owner of all per-pipe state of the HQDs.
>
> This change will allow us to split the queues between kfd and amdgpu
> with a queue granularity instead of pipe granularity.
>
> This patch fixes kfd allocating an HDP_EOP regio
take some time.
Oded
On Sun, Apr 16, 2017 at 3:12 AM, Andres Rodriguez wrote:
> Hey Oded,
>
> I've had problems with 4.12-wip on its own. Just wanted to confirm if
> you picked up this fix for a bug in drm-next:
> "drm: Fix get_property logic fumble"
>
> Regards,
ted patch with 2 bugs fixed:
>
> 1. the offset into the HPD bo was incorrect
> 2. the compute_pipe_init function had an mec/me mismatch, so registers were
> being set for the wrong ME
>
> Regards,
> Andres
>
> On 2017-04-16 01:25 AM, Oded Gabbay wrote:
>>
>>
also gone over your code and it seems fine.
So, this patch-set is:
Acked-by: Oded Gabbay
> On 2017-04-21 04:02 PM, Andres Rodriguez wrote:
>>
>> V4 updates:
>> * Rebased onto latest 4.12-wip (ba16d6c), since I got test reports that
>> the
>>patch series f
ts(local_mem_size);
> + buf[6] = upper_32_bits(local_mem_size);
>
> for (i = 0, hashout = 0; i < 7; i++)
> hashout ^= hash_32(buf[i], KFD_GPU_ID_HASH_WIDTH);
> --
> 2.7.4
>
> ___
> amd-gfx maili
> amd-gfx mailing list
> amd-gfx@lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/amd-gfx
With the above comment, this patch is:
Reviewed-by: Oded Gabbay
___
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
On Sat, Sep 3, 2016 at 5:49 AM, Edward O'Callaghan
wrote:
> Signed-off-by: Edward O'Callaghan
> ---
> drivers/gpu/drm/amd/amdkfd/kfd_dbgmgr.c | 1 +
> 1 file changed, 1 insertion(+)
>
> diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_dbgmgr.c
> b/drivers/gpu/drm/amd/amdkfd/kfd_dbgmgr.c
> index 56d6
On Sat, Sep 10, 2016 at 4:31 AM, Edward O'Callaghan
wrote:
> Dereference the one time and unpack the lower and upper 32bit
> portions with the proper kernel helper macros.
>
> Signed-off-by: Edward O'Callaghan
> Reviewed-by: Oded Gabbay
> ---
> drivers/gpu/dr
On Sat, Sep 10, 2016 at 4:31 AM, Edward O'Callaghan
wrote:
> Signed-off-by: Edward O'Callaghan
> Reviewed-by: Oded Gabbay
> ---
> drivers/gpu/drm/amd/amdkfd/kfd_kernel_queue.c | 3 +++
> 1 file changed, 3 insertions(+)
>
> diff --git a/drivers/gpu/drm/amd/amdkfd/
On Sat, Sep 10, 2016 at 4:31 AM, Edward O'Callaghan
wrote:
> Ensure we return a NULL on the fail branch so that the call
> site may BUG_ON() it.
>
> Signed-off-by: Edward O'Callaghan
> ---
> drivers/gpu/drm/amd/amdkfd/kfd_process.c | 5 +++--
> 1 file changed, 3 insertions(+), 2 deletions(-)
>
>
On Sat, Sep 10, 2016 at 4:31 AM, Edward O'Callaghan
wrote:
> pqm_uninit() will be called in kfd_process_notifier_release(), which
> is when the process exits. Calling it in kfd_unbind_process_from_device()
> is duplicate and in fact incorrect, because pqm should not be
> uninitalized as long as th
* We don't call amd_iommu_unbind_pasid() here
> +* because the IOMMU called us.
> +*/
> + pdd->bound = false;
>
> - srcu_read_unlock(&kfd_processes_srcu, idx);
> + mutex_unlock(&p->mutex);
> }
>
> struct kfd_proce
utimeout;
>
> m = get_sdma_mqd(mqd);
> sdma_base_addr = get_sdma_base_addr(m);
> @@ -396,7 +397,7 @@ static int kgd_hqd_sdma_destroy(struct kgd_dev *kgd, void
> *mqd,
> temp = RREG32(sdma_base_addr + mmSDMA0_
On Sat, Sep 10, 2016 at 4:31 AM, Edward O'Callaghan
wrote:
> Signed-off-by: Edward O'Callaghan
Please give a little explanation why this new function is needed.
In addition, I don't think "radeonkfd/amdkfd: .." is good as we never
used this kind of title before.
Let's just use "drm/amdkfd: ..."
opertices' by reference
> ___
> amd-gfx mailing list
> amd-gfx@lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/amd-gfx
All patches in the series are
Reviewed-by: Oded Gabbay
I added them to my -next tree
Christian is correct. That is not portable to other architectures
Oded
On Tue, Sep 27, 2016 at 3:52 PM, Christian König
wrote:
> NAK, don't use bitfields to decode hardware values. They aren't portable.
>
> Regards,
> Christian.
>
>
> Am 27.09.2016 um 14:47 schrieb Edward O'Callaghan:
>>
>> Sign
heck if there is over subscription */
> if ((sched_policy ==
> KFD_SCHED_POLICY_HWS_NO_OVERSUBSCRIPTION) &&
> - ((dev->dqm->processes_count >= VMID_PER_DEVICE) ||
> + ((dev->dqm->processes_count >= dev->vm_info.vmid_num_kfd) ||
> (dev->dqm->queue_count >= PIPE_PER_ME_CP_SCHEDULING *
> QUEUES_PER_PIPE))) {
> pr_err("kfd: over-subscription is not allowed in
> radeon_kfd.sched_policy == 1\n");
> retval = -EPERM;
> --
> 2.7.4
>
> ___
> amd-gfx mailing list
> amd-gfx@lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/amd-gfx
With the above comment fixed, this patch is:
Reviewed-by: Oded Gabbay
___
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
On Mon, Sep 11, 2017 at 10:29 PM, Deucher, Alexander
wrote:
>> -Original Message-
>> From: Kuehling, Felix
>> Sent: Wednesday, September 06, 2017 5:54 PM
>> To: amd-gfx@lists.freedesktop.org; Deucher, Alexander; Oded Gabbay;
>> Koenig, Christian
>>
ck_in_mhz: Retrieves maximum GPU clock in MHz
> *
> + * @alloc_pasid: Allocate a PASID
> + * @free_pasid: Free a PASID
> + *
> * @program_sh_mem_settings: A function that should initiate the memory
> * properties such as main aperture memory type (cache / non cached) and
> * secondary aperture base address, size and memory type.
> @@ -160,6 +163,9 @@ struct kfd2kgd_calls {
>
> uint32_t (*get_max_engine_clock_in_mhz)(struct kgd_dev *kgd);
>
> + int (*alloc_pasid)(unsigned int bits);
> + void (*free_pasid)(unsigned int pasid);
> +
> /* Register access functions */
> void (*program_sh_mem_settings)(struct kgd_dev *kgd, uint32_t vmid,
> uint32_t sh_mem_config, uint32_t sh_mem_ape1_base,
> --
> 2.7.4
>
> ___
> amd-gfx mailing list
> amd-gfx@lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/amd-gfx
This patch is:
Reviewed-by: Oded Gabbay
___
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
return (struct radeon_device *)kgd;
> --
> 2.7.4
>
> ___
> amd-gfx mailing list
> amd-gfx@lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/amd-gfx
This patch is:
Reviewed-by: Oded Gabbay
___
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
i++;
> + }
> +
> + if (!kfd2kgd)
> + return false;
> + }
>
> - mutex_unlock(&pasid_mutex);
> + r = kfd2kgd->alloc_pasid(pasid_bits);
>
> - return found;
> + return r > 0 ? r : 0;
> }
>
> void kfd_pasid_free
gt;mutex);
>
> @@ -288,6 +289,9 @@ static struct kfd_process *create_process(const struct
> task_struct *thread)
> if (process->pasid == 0)
> goto err_alloc_pasid;
>
> + if (kfd_alloc_process_doorbells(process) < 0)
> +
On Sat, Sep 16, 2017 at 2:43 AM, Felix Kuehling wrote:
> From: Andres Rodriguez
>
> Set the default permissions of /dev/kfd to be more than just root
> accessible 600.
>
I don't think that's acceptable.
You need to use udev rules file for that.
Oded
> Signed-off-by: Andres Rodriguez
> Reviewed
On Sat, Sep 16, 2017 at 2:42 AM, Felix Kuehling wrote:
> From: Yong Zhao
>
> There are already CHIP_* definitions under amd_shared.h file on amdgpu
> side, so KFD should reuse them rather than defining new ones.
>
> Using enum for asic type requires default cases on switch statements
> to prevent
iommu_invalid_ppr_cb);
> +
> + err = kfd->dqm->ops.start(kfd->dqm);
> + if (err) {
> + dev_err(kfd_device,
> + "Error starting queue manager for device %x:%x\n",
> + kfd->pdev->vendor, kfd->pdev->device);
> + goto dqm_start_error;
> }
>
> - return 0;
> + return err;
> +
> +dqm_start_error:
> + amd_iommu_free_device(kfd->pdev);
> +
> + return err;
> }
>
> /* This is called directly from KGD at ISR. */
> --
> 2.7.4
>
> ___
> amd-gfx mailing list
> amd-gfx@lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/amd-gfx
This patch is:
Reviewed-by: Oded Gabbay
___
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
the paths in the code in order to understand it.
> + mutex_unlock(&p->mutex);
> +
> + if (temp_bound == PDD_BOUND)
> + amd_iommu_unbind_pasid(dev->pdev, temp_pasid);
> + }
> +
> + srcu_read_unlock(&kfd_processes_srcu, idx);
> +}
> +
> +void kfd_process_iommu_unbind_callback(struct kfd_dev *dev, unsigned int
> pasid)
> {
> struct kfd_process *p;
> struct kfd_process_device *pdd;
> @@ -432,15 +500,6 @@ void kfd_unbind_process_from_device(struct kfd_dev *dev,
> unsigned int pasid)
> pdd->reset_wavefronts = false;
> }
>
> - /*
> -* Just mark pdd as unbound, because we still need it
> -* to call amd_iommu_unbind_pasid() in when the
> -* process exits.
> -* We don't call amd_iommu_unbind_pasid() here
> -* because the IOMMU called us.
> -*/
> - pdd->bound = false;
> -
> mutex_unlock(&p->mutex);
> }
>
> --
> 2.7.4
>
> ___
> amd-gfx mailing list
> amd-gfx@lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/amd-gfx
With the above minor comments fixed, this patch is:
Reviewed-by: Oded Gabbay
___
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
rm/amd/amdkfd/kfd_priv.h
> +++ b/drivers/gpu/drm/amd/amdkfd/kfd_priv.h
> @@ -673,11 +673,8 @@ int amdkfd_fence_wait_timeout(unsigned int *fence_addr,
>
> /* Packet Manager */
>
> -#define KFD_HIQ_TIMEOUT (500)
> -
> #define KFD_FENCE_COMP
__
> amd-gfx mailing list
> amd-gfx@lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/amd-gfx
This patch is:
Reviewed-by: Oded Gabbay
___
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
dqm->ops.uninitialize = uninitialize_nocpsch;
> + dqm->ops.uninitialize = uninitialize;
> dqm->ops.set_cache_memory_policy = set_cache_memory_policy;
> break;
> default:
> --
> 2.7.4
>
> ___
> amd-gfx mailing list
> amd-gfx@lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/amd-gfx
This patch is:
Reviewed-by: Oded Gabbay
___
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
rocess_queue_manager *pqm,
> case KFD_QUEUE_TYPE_COMPUTE:
> /* check if there is over subscription */
> if ((sched_policy ==
> KFD_SCHED_POLICY_HWS_NO_OVERSUBSCRIPTION) &&
> - ((dev->dqm->processes_count >= VMID
On Sat, Sep 16, 2017 at 2:43 AM, Felix Kuehling wrote:
> From: Yong Zhao
>
> Avoid intermediate negative numbers when doing calculations with a mix
> of signed and unsigned variables where implicit conversions can lead
> to unexpected results.
>
> When kernel queue buffer wraps around to 0, we ne
> struct list_head signal_event_pages;
> u32 next_nonsignal_event_id;
> size_t signal_event_count;
> + bool signal_event_limit_reached;
> };
>
> /**
> --
> 2.7.4
>
> _______
> amd-gfx mailing list
> am
On Fri, Sep 8, 2017 at 5:13 PM, Colin King wrote:
> From: Colin Ian King
>
> The call to kfd_device_by_id can potentially return null, so check that
> dev is null and return with -EINVAL to avoid a null pointer dereference.
>
> Detected by CoverityScan CID#1454629 ("Dereference null return value"
bugger. 4096 is good enough and
> the new debug trap handler will use events from that pool.
>
> Thanks
> Hari
>
>
>
> -Original Message-
> From: Kuehling, Felix
> Sent: Thursday, September 21, 2017 9:36 PM
> To: amd-gfx@lists.freedesktop.org; Oded Gabbay
t_timeout(unsigned int *fence_addr,
> unsigned int fence_value,
> - unsigned long timeout);
> + unsigned int timeout_ms);
>
> /* Packet Manager */
>
> --
> 2.7.4
>
This patch is:
Reviewed-by: Oded Gabbay
___
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
@ -112,11 +114,6 @@ enum cache_policy {
> cache_policy_noncoherent
> };
>
> -enum asic_family_type {
> - CHIP_KAVERI = 0,
> - CHIP_CARRIZO
> -};
> -
> struct kfd_event_interrupt_class {
> bool (*interrupt_isr)(struct kfd_dev *dev,
>
= rptr) {
> + *buffer_ptr = NULL;
> + return -ENOMEM;
> + }
> + /* fill nops, roll back and start at position 0 */
> while (wptr > 0) {
> queue_address[wptr] = kq->nop_packet;
>
On Thu, Sep 21, 2017 at 1:10 AM, Felix Kuehling wrote:
> Small self-contained KFD fixes that don't introduce any new
> functionality or features. These may be suitable to include in
> drm-fixes for 4.14.
>
Thanks!
I added patches 5, 9 and 10 to my -fixes tree and I will send them
shortly to Dave
1 - 100 of 326 matches
Mail list logo