[PATCH 2/3] drm/amd/pm/smu11: poll BACO status after RPM BACO exits
Alex Deucher
alexdeucher at gmail.com
Tue Nov 29 17:52:33 UTC 2022
Once these patches land, can we revert these changes?
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=0c85c067c9d9d7a1b2cc2e01a236d5d0d4a872b5
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=192039f12233c9063d040266e7c98188c7c89dec
Alex
On Tue, Nov 22, 2022 at 8:44 PM Guchun Chen <guchun.chen at amd.com> wrote:
>
> After executing BACO exit, driver needs to poll the status
> to ensure FW has completed BACO exit sequence to prevent
> timing issue.
>
> v2: use usleep_range to replace msleep to fix checkpatch.pl warnings
>
> Signed-off-by: Guchun Chen <guchun.chen at amd.com>
> ---
> .../gpu/drm/amd/pm/swsmu/smu11/smu_v11_0.c | 24 ++++++++++++++++++-
> 1 file changed, 23 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu11/smu_v11_0.c b/drivers/gpu/drm/amd/pm/swsmu/smu11/smu_v11_0.c
> index ad5f6a15a1d7..ad66d57aa102 100644
> --- a/drivers/gpu/drm/amd/pm/swsmu/smu11/smu_v11_0.c
> +++ b/drivers/gpu/drm/amd/pm/swsmu/smu11/smu_v11_0.c
> @@ -79,6 +79,17 @@ MODULE_FIRMWARE("amdgpu/beige_goby_smc.bin");
> #define mmTHM_BACO_CNTL_ARCT 0xA7
> #define mmTHM_BACO_CNTL_ARCT_BASE_IDX 0
>
> +static void smu_v11_0_poll_baco_exit(struct smu_context *smu)
> +{
> + struct amdgpu_device *adev = smu->adev;
> + uint32_t data, loop = 0;
> +
> + do {
> + usleep_range(1000, 1100);
> + data = RREG32_SOC15(THM, 0, mmTHM_BACO_CNTL);
> + } while ((data & 0x100) && (++loop < 100));
> +}
> +
> int smu_v11_0_init_microcode(struct smu_context *smu)
> {
> struct amdgpu_device *adev = smu->adev;
> @@ -1689,7 +1700,18 @@ int smu_v11_0_baco_enter(struct smu_context *smu)
>
> int smu_v11_0_baco_exit(struct smu_context *smu)
> {
> - return smu_v11_0_baco_set_state(smu, SMU_BACO_STATE_EXIT);
> + int ret;
> +
> + ret = smu_v11_0_baco_set_state(smu, SMU_BACO_STATE_EXIT);
> + if (!ret) {
> + /*
> + * Poll BACO exit status to ensure FW has completed
> + * BACO exit process to avoid timing issues.
> + */
> + smu_v11_0_poll_baco_exit(smu);
> + }
> +
> + return ret;
> }
>
> int smu_v11_0_mode1_reset(struct smu_context *smu)
> --
> 2.25.1
>
More information about the amd-gfx
mailing list