Hi Alex, COMPUTE_PGM* registers are per pipe per queue - each queue of each pipe has a copy of those registers. kgd_hqd_dump is called per pipe per queue. In kgd_hqd_dump, SRBM_GFX_CNTL register is set up to lock the correct pipe/queue. This is done through function acquire_queue(kgd, pipe_id, queue_id). -----Original Message----- From: Deucher, Alexander Sent: Wednesday, March 01, 2017 5:00 PM To: 'ozeng'; amd-gfx at lists.freedesktop.org Cc: Zeng, Oak Subject: RE: [PATCH] drm/amdgpu: Added more hqd debug messages > -----Original Message----- > From: amd-gfx [mailto:amd-gfx-bounces at lists.freedesktop.org] On Behalf > Of ozeng > Sent: Wednesday, March 01, 2017 4:28 PM > To: amd-gfx at lists.freedesktop.org > Cc: Zeng, Oak > Subject: [PATCH] drm/amdgpu: Added more hqd debug messages > > Dump register COMPUTE_PGM_RSRC1 and COMPUTE_PGM_RSRC2 in function > kgd_hqd_dump > > Change-Id: I8e007f5bcc095d3283b642791860dd072dd43b1c > Signed-off-by: Oak Zeng <Oak.Zeng at amd.com> > --- > drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v7.c | 2 ++ > drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v8.c | 2 ++ > 2 files changed, 4 insertions(+) > > diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v7.c > b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v7.c > index 7732def..d7bcdc3 100644 > --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v7.c > +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v7.c > @@ -439,6 +439,8 @@ static int kgd_hqd_dump(struct kgd_dev *kgd, > > acquire_queue(kgd, pipe_id, queue_id); > > + DUMP_REG(mmCOMPUTE_PGM_RSRC1); > + DUMP_REG(mmCOMPUTE_PGM_RSRC2); > DUMP_REG(mmCOMPUTE_STATIC_THREAD_MGMT_SE0); > DUMP_REG(mmCOMPUTE_STATIC_THREAD_MGMT_SE1); > DUMP_REG(mmCOMPUTE_STATIC_THREAD_MGMT_SE2); > diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v8.c > b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v8.c > index 289c188..2aff607 100644 > --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v8.c > +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v8.c > @@ -433,6 +433,8 @@ static int kgd_hqd_dump(struct kgd_dev *kgd, > > acquire_queue(kgd, pipe_id, queue_id); > > + DUMP_REG(mmCOMPUTE_PGM_RSRC1); > + DUMP_REG(mmCOMPUTE_PGM_RSRC2); > DUMP_REG(mmCOMPUTE_STATIC_THREAD_MGMT_SE0); > DUMP_REG(mmCOMPUTE_STATIC_THREAD_MGMT_SE1); > DUMP_REG(mmCOMPUTE_STATIC_THREAD_MGMT_SE2); Aren't COMPUTE_PGM* registers multi-state for pipelining? In which case, reading them back via MMIO is largely useless. Alex > -- > 2.7.4 > > _______________________________________________ > amd-gfx mailing list > amd-gfx at lists.freedesktop.org > https://lists.freedesktop.org/mailman/listinfo/amd-gfx
