diff options
author | Evan Quan <evan.quan@amd.com> | 2023-08-11 19:40:31 +0300 |
---|---|---|
committer | Greg Kroah-Hartman <gregkh@linuxfoundation.org> | 2023-08-16 19:27:23 +0300 |
commit | 0f19195d639764d68f6f316dda363ba29821e5bc (patch) | |
tree | 6ac517383f3cc0025a65a6970624cb931314113e /drivers/gpu/drm/amd/pm/powerplay/amd_powerplay.c | |
parent | b064f9ccf11cb4008a1f0e3c96099041b121af80 (diff) | |
download | linux-0f19195d639764d68f6f316dda363ba29821e5bc.tar.xz |
drm/amd/pm: avoid unintentional shutdown due to temperature momentary fluctuation
commit b75efe88b20c2be28b67e2821a794cc183e32374 upstream
An intentional delay is added on soft ctf triggered. Then there will
be a double check for the GPU temperature before taking further
action. This can avoid unintended shutdown due to temperature
momentary fluctuation.
Signed-off-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Lijo Lazar <lijo.lazar@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
[ Hand-modified because:
* XCP support added to amdgpu.h in kernel 6.5
and is not necessary for this fix.
* SMU microcode initialization moved in
32806038aa76 ("drm/amd: Load SMU microcode during early_init") ]
Link: https://gitlab.freedesktop.org/drm/amd/-/issues/1267
Link: https://gitlab.freedesktop.org/drm/amd/-/issues/2779
Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Diffstat (limited to 'drivers/gpu/drm/amd/pm/powerplay/amd_powerplay.c')
-rw-r--r-- | drivers/gpu/drm/amd/pm/powerplay/amd_powerplay.c | 48 |
1 files changed, 48 insertions, 0 deletions
diff --git a/drivers/gpu/drm/amd/pm/powerplay/amd_powerplay.c b/drivers/gpu/drm/amd/pm/powerplay/amd_powerplay.c index 3f4a476d7802..179e1c593a53 100644 --- a/drivers/gpu/drm/amd/pm/powerplay/amd_powerplay.c +++ b/drivers/gpu/drm/amd/pm/powerplay/amd_powerplay.c @@ -26,6 +26,7 @@ #include <linux/gfp.h> #include <linux/slab.h> #include <linux/firmware.h> +#include <linux/reboot.h> #include "amd_shared.h" #include "amd_powerplay.h" #include "power_state.h" @@ -91,6 +92,45 @@ static int pp_early_init(void *handle) return 0; } +static void pp_swctf_delayed_work_handler(struct work_struct *work) +{ + struct pp_hwmgr *hwmgr = + container_of(work, struct pp_hwmgr, swctf_delayed_work.work); + struct amdgpu_device *adev = hwmgr->adev; + struct amdgpu_dpm_thermal *range = + &adev->pm.dpm.thermal; + uint32_t gpu_temperature, size; + int ret; + + /* + * If the hotspot/edge temperature is confirmed as below SW CTF setting point + * after the delay enforced, nothing will be done. + * Otherwise, a graceful shutdown will be performed to prevent further damage. + */ + if (range->sw_ctf_threshold && + hwmgr->hwmgr_func->read_sensor) { + ret = hwmgr->hwmgr_func->read_sensor(hwmgr, + AMDGPU_PP_SENSOR_HOTSPOT_TEMP, + &gpu_temperature, + &size); + /* + * For some legacy ASICs, hotspot temperature retrieving might be not + * supported. Check the edge temperature instead then. + */ + if (ret == -EOPNOTSUPP) + ret = hwmgr->hwmgr_func->read_sensor(hwmgr, + AMDGPU_PP_SENSOR_EDGE_TEMP, + &gpu_temperature, + &size); + if (!ret && gpu_temperature / 1000 < range->sw_ctf_threshold) + return; + } + + dev_emerg(adev->dev, "ERROR: GPU over temperature range(SW CTF) detected!\n"); + dev_emerg(adev->dev, "ERROR: System is going to shutdown due to GPU SW CTF!\n"); + orderly_poweroff(true); +} + static int pp_sw_init(void *handle) { struct amdgpu_device *adev = handle; @@ -101,6 +141,10 @@ static int pp_sw_init(void *handle) pr_debug("powerplay sw init %s\n", ret ? "failed" : "successfully"); + if (!ret) + INIT_DELAYED_WORK(&hwmgr->swctf_delayed_work, + pp_swctf_delayed_work_handler); + return ret; } @@ -136,6 +180,8 @@ static int pp_hw_fini(void *handle) struct amdgpu_device *adev = handle; struct pp_hwmgr *hwmgr = adev->powerplay.pp_handle; + cancel_delayed_work_sync(&hwmgr->swctf_delayed_work); + hwmgr_hw_fini(hwmgr); return 0; @@ -222,6 +268,8 @@ static int pp_suspend(void *handle) struct amdgpu_device *adev = handle; struct pp_hwmgr *hwmgr = adev->powerplay.pp_handle; + cancel_delayed_work_sync(&hwmgr->swctf_delayed_work); + return hwmgr_suspend(hwmgr); } |