* [PATCH v2 0/8] accel/ivpu: Fixes for 6.3
@ 2023-03-23 12:54 Stanislaw Gruszka
2023-03-23 12:54 ` [PATCH v2 1/8] accel/ivpu: Do not access HW registers after unbind Stanislaw Gruszka
` (7 more replies)
0 siblings, 8 replies; 13+ messages in thread
From: Stanislaw Gruszka @ 2023-03-23 12:54 UTC (permalink / raw)
To: dri-devel; +Cc: Stanislaw Gruszka, Oded Gabbay, Jeffrey Hugo, Jacek Lawrynowicz
Various fixes intended for linux 6.3 relase.
Patch 6 is dependency for patch 7.
v2:
- add Fixes tags
- fix printing error on unplug
- clarify that SSID=65 is valid in the commit message
- change commit messages for patch 4 and 7
- add patch 8 with new fix
Andrzej Kacprowski (1):
accel/ivpu: Fix IPC buffer header status field value
Stanislaw Gruszka (7):
accel/ivpu: Do not access HW registers after unbind
accel/ivpu: Cancel recovery work
accel/ivpu: Do not use SSID 1
accel/ivpu: Fix power down sequence
accel/ivpu: Disable buttress on device removal
accel/ivpu: Remove support for 1 tile SKUs
accel/ivpu: Fix VPU clock calculation
drivers/accel/ivpu/ivpu_drv.c | 18 +++--
drivers/accel/ivpu/ivpu_drv.h | 7 +-
drivers/accel/ivpu/ivpu_hw_mtl.c | 113 ++++++++++---------------------
drivers/accel/ivpu/ivpu_ipc.h | 2 +-
drivers/accel/ivpu/ivpu_job.c | 11 ++-
drivers/accel/ivpu/ivpu_pm.c | 17 ++++-
drivers/accel/ivpu/ivpu_pm.h | 1 +
7 files changed, 80 insertions(+), 89 deletions(-)
--
2.25.1
^ permalink raw reply [flat|nested] 13+ messages in thread
* [PATCH v2 1/8] accel/ivpu: Do not access HW registers after unbind
2023-03-23 12:54 [PATCH v2 0/8] accel/ivpu: Fixes for 6.3 Stanislaw Gruszka
@ 2023-03-23 12:54 ` Stanislaw Gruszka
2023-03-23 12:54 ` [PATCH v2 2/8] accel/ivpu: Cancel recovery work Stanislaw Gruszka
` (6 subsequent siblings)
7 siblings, 0 replies; 13+ messages in thread
From: Stanislaw Gruszka @ 2023-03-23 12:54 UTC (permalink / raw)
To: dri-devel; +Cc: Stanislaw Gruszka, Oded Gabbay, Jeffrey Hugo, Jacek Lawrynowicz
We should not access hardware after we unbind from the bus.
Use drm_dev_enter() / drm_dev_exit() to mark code sections where
hardware is accessed (and not already protected by other locks)
and drm_dev_unplug() to mark device is gone.
Fixes: 35b137630f08 ("accel/ivpu: Introduce a new DRM driver for Intel VPU")
Signed-off-by: Stanislaw Gruszka <stanislaw.gruszka@linux.intel.com>
Reviewed-by: Jeffrey Hugo <quic_jhugo@quicinc.com>
---
drivers/accel/ivpu/ivpu_drv.c | 8 ++++++--
drivers/accel/ivpu/ivpu_drv.h | 1 +
drivers/accel/ivpu/ivpu_job.c | 11 +++++++++--
3 files changed, 16 insertions(+), 4 deletions(-)
diff --git a/drivers/accel/ivpu/ivpu_drv.c b/drivers/accel/ivpu/ivpu_drv.c
index 231f29bb5025..ac06bbfca920 100644
--- a/drivers/accel/ivpu/ivpu_drv.c
+++ b/drivers/accel/ivpu/ivpu_drv.c
@@ -8,7 +8,6 @@
#include <linux/pci.h>
#include <drm/drm_accel.h>
-#include <drm/drm_drv.h>
#include <drm/drm_file.h>
#include <drm/drm_gem.h>
#include <drm/drm_ioctl.h>
@@ -118,6 +117,10 @@ static int ivpu_get_param_ioctl(struct drm_device *dev, void *data, struct drm_f
struct pci_dev *pdev = to_pci_dev(vdev->drm.dev);
struct drm_ivpu_param *args = data;
int ret = 0;
+ int idx;
+
+ if (!drm_dev_enter(dev, &idx))
+ return -ENODEV;
switch (args->param) {
case DRM_IVPU_PARAM_DEVICE_ID:
@@ -171,6 +174,7 @@ static int ivpu_get_param_ioctl(struct drm_device *dev, void *data, struct drm_f
break;
}
+ drm_dev_exit(idx);
return ret;
}
@@ -622,7 +626,7 @@ static void ivpu_remove(struct pci_dev *pdev)
{
struct ivpu_device *vdev = pci_get_drvdata(pdev);
- drm_dev_unregister(&vdev->drm);
+ drm_dev_unplug(&vdev->drm);
ivpu_dev_fini(vdev);
}
diff --git a/drivers/accel/ivpu/ivpu_drv.h b/drivers/accel/ivpu/ivpu_drv.h
index f47b4965db2e..1b2aa05840ad 100644
--- a/drivers/accel/ivpu/ivpu_drv.h
+++ b/drivers/accel/ivpu/ivpu_drv.h
@@ -7,6 +7,7 @@
#define __IVPU_DRV_H__
#include <drm/drm_device.h>
+#include <drm/drm_drv.h>
#include <drm/drm_managed.h>
#include <drm/drm_mm.h>
#include <drm/drm_print.h>
diff --git a/drivers/accel/ivpu/ivpu_job.c b/drivers/accel/ivpu/ivpu_job.c
index 94068aedf97c..910301fae435 100644
--- a/drivers/accel/ivpu/ivpu_job.c
+++ b/drivers/accel/ivpu/ivpu_job.c
@@ -489,12 +489,12 @@ ivpu_job_prepare_bos_for_submit(struct drm_file *file, struct ivpu_job *job, u32
int ivpu_submit_ioctl(struct drm_device *dev, void *data, struct drm_file *file)
{
- int ret = 0;
struct ivpu_file_priv *file_priv = file->driver_priv;
struct ivpu_device *vdev = file_priv->vdev;
struct drm_ivpu_submit *params = data;
struct ivpu_job *job;
u32 *buf_handles;
+ int idx, ret;
if (params->engine > DRM_IVPU_ENGINE_COPY)
return -EINVAL;
@@ -523,6 +523,11 @@ int ivpu_submit_ioctl(struct drm_device *dev, void *data, struct drm_file *file)
goto free_handles;
}
+ if (!drm_dev_enter(&vdev->drm, &idx)) {
+ ret = -ENODEV;
+ goto free_handles;
+ }
+
ivpu_dbg(vdev, JOB, "Submit ioctl: ctx %u buf_count %u\n",
file_priv->ctx.id, params->buffer_count);
@@ -530,7 +535,7 @@ int ivpu_submit_ioctl(struct drm_device *dev, void *data, struct drm_file *file)
if (!job) {
ivpu_err(vdev, "Failed to create job\n");
ret = -ENOMEM;
- goto free_handles;
+ goto dev_exit;
}
ret = ivpu_job_prepare_bos_for_submit(file, job, buf_handles, params->buffer_count,
@@ -548,6 +553,8 @@ int ivpu_submit_ioctl(struct drm_device *dev, void *data, struct drm_file *file)
job_put:
job_put(job);
+dev_exit:
+ drm_dev_exit(idx);
free_handles:
kfree(buf_handles);
--
2.25.1
^ permalink raw reply related [flat|nested] 13+ messages in thread
* [PATCH v2 2/8] accel/ivpu: Cancel recovery work
2023-03-23 12:54 [PATCH v2 0/8] accel/ivpu: Fixes for 6.3 Stanislaw Gruszka
2023-03-23 12:54 ` [PATCH v2 1/8] accel/ivpu: Do not access HW registers after unbind Stanislaw Gruszka
@ 2023-03-23 12:54 ` Stanislaw Gruszka
2023-03-23 14:59 ` Jeffrey Hugo
2023-03-23 12:54 ` [PATCH v2 3/8] accel/ivpu: Do not use SSID 1 Stanislaw Gruszka
` (5 subsequent siblings)
7 siblings, 1 reply; 13+ messages in thread
From: Stanislaw Gruszka @ 2023-03-23 12:54 UTC (permalink / raw)
To: dri-devel; +Cc: Stanislaw Gruszka, Oded Gabbay, Jeffrey Hugo, Jacek Lawrynowicz
Prevent running recovery_work after device is removed.
Fixes: 852be13f3bd3 ("accel/ivpu: Add PM support")
Signed-off-by: Stanislaw Gruszka <stanislaw.gruszka@linux.intel.com>
---
drivers/accel/ivpu/ivpu_drv.c | 2 ++
drivers/accel/ivpu/ivpu_pm.c | 17 ++++++++++++++---
drivers/accel/ivpu/ivpu_pm.h | 1 +
3 files changed, 17 insertions(+), 3 deletions(-)
diff --git a/drivers/accel/ivpu/ivpu_drv.c b/drivers/accel/ivpu/ivpu_drv.c
index ac06bbfca920..d9e311b40348 100644
--- a/drivers/accel/ivpu/ivpu_drv.c
+++ b/drivers/accel/ivpu/ivpu_drv.c
@@ -580,6 +580,8 @@ static void ivpu_dev_fini(struct ivpu_device *vdev)
ivpu_pm_disable(vdev);
ivpu_shutdown(vdev);
ivpu_job_done_thread_fini(vdev);
+ ivpu_pm_cancel_recovery(vdev);
+
ivpu_ipc_fini(vdev);
ivpu_fw_fini(vdev);
ivpu_mmu_global_context_fini(vdev);
diff --git a/drivers/accel/ivpu/ivpu_pm.c b/drivers/accel/ivpu/ivpu_pm.c
index a880f1dd857e..da0bbc46a024 100644
--- a/drivers/accel/ivpu/ivpu_pm.c
+++ b/drivers/accel/ivpu/ivpu_pm.c
@@ -98,12 +98,18 @@ static int ivpu_resume(struct ivpu_device *vdev)
static void ivpu_pm_recovery_work(struct work_struct *work)
{
struct ivpu_pm_info *pm = container_of(work, struct ivpu_pm_info, recovery_work);
- struct ivpu_device *vdev = pm->vdev;
+ struct ivpu_device *vdev = pm->vdev;
char *evt[2] = {"IVPU_PM_EVENT=IVPU_RECOVER", NULL};
int ret;
- ret = pci_reset_function(to_pci_dev(vdev->drm.dev));
- if (ret)
+retry:
+ ret = pci_try_reset_function(to_pci_dev(vdev->drm.dev));
+ if (ret == -EAGAIN && !drm_dev_is_unplugged(&vdev->drm)) {
+ cond_resched();
+ goto retry;
+ }
+
+ if (ret && ret != -EAGAIN)
ivpu_err(vdev, "Failed to reset VPU: %d\n", ret);
kobject_uevent_env(&vdev->drm.dev->kobj, KOBJ_CHANGE, evt);
@@ -302,6 +308,11 @@ int ivpu_pm_init(struct ivpu_device *vdev)
return 0;
}
+void ivpu_pm_cancel_recovery(struct ivpu_device *vdev)
+{
+ cancel_work_sync(&vdev->pm->recovery_work);
+}
+
void ivpu_pm_enable(struct ivpu_device *vdev)
{
struct device *dev = vdev->drm.dev;
diff --git a/drivers/accel/ivpu/ivpu_pm.h b/drivers/accel/ivpu/ivpu_pm.h
index dc1b3758e13f..baca98187255 100644
--- a/drivers/accel/ivpu/ivpu_pm.h
+++ b/drivers/accel/ivpu/ivpu_pm.h
@@ -21,6 +21,7 @@ struct ivpu_pm_info {
int ivpu_pm_init(struct ivpu_device *vdev);
void ivpu_pm_enable(struct ivpu_device *vdev);
void ivpu_pm_disable(struct ivpu_device *vdev);
+void ivpu_pm_cancel_recovery(struct ivpu_device *vdev);
int ivpu_pm_suspend_cb(struct device *dev);
int ivpu_pm_resume_cb(struct device *dev);
--
2.25.1
^ permalink raw reply related [flat|nested] 13+ messages in thread
* [PATCH v2 3/8] accel/ivpu: Do not use SSID 1
2023-03-23 12:54 [PATCH v2 0/8] accel/ivpu: Fixes for 6.3 Stanislaw Gruszka
2023-03-23 12:54 ` [PATCH v2 1/8] accel/ivpu: Do not access HW registers after unbind Stanislaw Gruszka
2023-03-23 12:54 ` [PATCH v2 2/8] accel/ivpu: Cancel recovery work Stanislaw Gruszka
@ 2023-03-23 12:54 ` Stanislaw Gruszka
2023-03-23 15:00 ` Jeffrey Hugo
2023-03-23 12:55 ` [PATCH v2 4/8] accel/ivpu: Fix power down sequence Stanislaw Gruszka
` (4 subsequent siblings)
7 siblings, 1 reply; 13+ messages in thread
From: Stanislaw Gruszka @ 2023-03-23 12:54 UTC (permalink / raw)
To: dri-devel
Cc: Stanislaw Gruszka, Oded Gabbay, Jeffrey Hugo, Jacek Lawrynowicz,
Andrzej Kacprowski
The SSID=1 is used by the firmware as default value in
case SSID mapping is not initialized. This allows
detecting use of miss-configured memory contexts.
The future FW versions may not allow using SSID=1.
SSID=65 is valid value, number of contexts are limited
by number of available command queues, but SSID can be
any u16 value.
Fixes: 35b137630f08 ("accel/ivpu: Introduce a new DRM driver for Intel VPU")
Co-developed-by: Andrzej Kacprowski <andrzej.kacprowski@linux.intel.com>
Signed-off-by: Andrzej Kacprowski <andrzej.kacprowski@linux.intel.com>
Signed-off-by: Stanislaw Gruszka <stanislaw.gruszka@linux.intel.com>
---
drivers/accel/ivpu/ivpu_drv.c | 4 ++--
drivers/accel/ivpu/ivpu_drv.h | 5 ++++-
2 files changed, 6 insertions(+), 3 deletions(-)
diff --git a/drivers/accel/ivpu/ivpu_drv.c b/drivers/accel/ivpu/ivpu_drv.c
index d9e311b40348..70245cf84593 100644
--- a/drivers/accel/ivpu/ivpu_drv.c
+++ b/drivers/accel/ivpu/ivpu_drv.c
@@ -474,8 +474,8 @@ static int ivpu_dev_init(struct ivpu_device *vdev)
vdev->hw->ops = &ivpu_hw_mtl_ops;
vdev->platform = IVPU_PLATFORM_INVALID;
- vdev->context_xa_limit.min = IVPU_GLOBAL_CONTEXT_MMU_SSID + 1;
- vdev->context_xa_limit.max = IVPU_CONTEXT_LIMIT;
+ vdev->context_xa_limit.min = IVPU_USER_CONTEXT_MIN_SSID;
+ vdev->context_xa_limit.max = IVPU_USER_CONTEXT_MAX_SSID;
atomic64_set(&vdev->unique_id_counter, 0);
xa_init_flags(&vdev->context_xa, XA_FLAGS_ALLOC);
xa_init_flags(&vdev->submitted_jobs_xa, XA_FLAGS_ALLOC1);
diff --git a/drivers/accel/ivpu/ivpu_drv.h b/drivers/accel/ivpu/ivpu_drv.h
index 1b2aa05840ad..ef12a38e06e1 100644
--- a/drivers/accel/ivpu/ivpu_drv.h
+++ b/drivers/accel/ivpu/ivpu_drv.h
@@ -25,7 +25,10 @@
#define PCI_DEVICE_ID_MTL 0x7d1d
#define IVPU_GLOBAL_CONTEXT_MMU_SSID 0
-#define IVPU_CONTEXT_LIMIT 64
+/* SSID 1 is used by the VPU to represent invalid context */
+#define IVPU_USER_CONTEXT_MIN_SSID 2
+#define IVPU_USER_CONTEXT_MAX_SSID (IVPU_USER_CONTEXT_MIN_SSID + 63)
+
#define IVPU_NUM_ENGINES 2
#define IVPU_PLATFORM_SILICON 0
--
2.25.1
^ permalink raw reply related [flat|nested] 13+ messages in thread
* [PATCH v2 4/8] accel/ivpu: Fix power down sequence
2023-03-23 12:54 [PATCH v2 0/8] accel/ivpu: Fixes for 6.3 Stanislaw Gruszka
` (2 preceding siblings ...)
2023-03-23 12:54 ` [PATCH v2 3/8] accel/ivpu: Do not use SSID 1 Stanislaw Gruszka
@ 2023-03-23 12:55 ` Stanislaw Gruszka
2023-03-23 12:55 ` [PATCH v2 5/8] accel/ivpu: Disable buttress on device removal Stanislaw Gruszka
` (3 subsequent siblings)
7 siblings, 0 replies; 13+ messages in thread
From: Stanislaw Gruszka @ 2023-03-23 12:55 UTC (permalink / raw)
To: dri-devel; +Cc: Stanislaw Gruszka, Oded Gabbay, Jeffrey Hugo, Jacek Lawrynowicz
Remove FPGA workaround on power_down to skip checking for noc
quiescent state.
Put VPU in reset before powering it down and skip manipulating
registers that are reset by the VPU reset.
This fixes power down errors where VPU is powered down just after VPU
is booted.
Fixes: 35b137630f08 ("accel/ivpu: Introduce a new DRM driver for Intel VPU")
Co-developed-by: Jacek Lawrynowicz <jacek.lawrynowicz@linux.intel.com>
Signed-off-by: Jacek Lawrynowicz <jacek.lawrynowicz@linux.intel.com>
Signed-off-by: Stanislaw Gruszka <stanislaw.gruszka@linux.intel.com>
Reviewed-by: Jeffrey Hugo <quic_jhugo@quicinc.com>
---
drivers/accel/ivpu/ivpu_hw_mtl.c | 37 ++------------------------------
1 file changed, 2 insertions(+), 35 deletions(-)
diff --git a/drivers/accel/ivpu/ivpu_hw_mtl.c b/drivers/accel/ivpu/ivpu_hw_mtl.c
index 62bfaa9081c4..70ca6de78060 100644
--- a/drivers/accel/ivpu/ivpu_hw_mtl.c
+++ b/drivers/accel/ivpu/ivpu_hw_mtl.c
@@ -403,11 +403,6 @@ static int ivpu_boot_host_ss_axi_enable(struct ivpu_device *vdev)
return ivpu_boot_host_ss_axi_drive(vdev, true);
}
-static int ivpu_boot_host_ss_axi_disable(struct ivpu_device *vdev)
-{
- return ivpu_boot_host_ss_axi_drive(vdev, false);
-}
-
static int ivpu_boot_host_ss_top_noc_drive(struct ivpu_device *vdev, bool enable)
{
int ret;
@@ -441,11 +436,6 @@ static int ivpu_boot_host_ss_top_noc_enable(struct ivpu_device *vdev)
return ivpu_boot_host_ss_top_noc_drive(vdev, true);
}
-static int ivpu_boot_host_ss_top_noc_disable(struct ivpu_device *vdev)
-{
- return ivpu_boot_host_ss_top_noc_drive(vdev, false);
-}
-
static void ivpu_boot_pwr_island_trickle_drive(struct ivpu_device *vdev, bool enable)
{
u32 val = REGV_RD32(MTL_VPU_HOST_SS_AON_PWR_ISLAND_TRICKLE_EN0);
@@ -504,16 +494,6 @@ static void ivpu_boot_dpu_active_drive(struct ivpu_device *vdev, bool enable)
REGV_WR32(MTL_VPU_HOST_SS_AON_DPU_ACTIVE, val);
}
-static int ivpu_boot_pwr_domain_disable(struct ivpu_device *vdev)
-{
- ivpu_boot_dpu_active_drive(vdev, false);
- ivpu_boot_pwr_island_isolation_drive(vdev, true);
- ivpu_boot_pwr_island_trickle_drive(vdev, false);
- ivpu_boot_pwr_island_drive(vdev, false);
-
- return ivpu_boot_wait_for_pwr_island_status(vdev, 0x0);
-}
-
static int ivpu_boot_pwr_domain_enable(struct ivpu_device *vdev)
{
int ret;
@@ -797,21 +777,8 @@ static int ivpu_hw_mtl_power_down(struct ivpu_device *vdev)
{
int ret = 0;
- /* FPGA requires manual clearing of IP_Reset bit by enabling quiescent state */
- if (ivpu_is_fpga(vdev)) {
- if (ivpu_boot_host_ss_top_noc_disable(vdev)) {
- ivpu_err(vdev, "Failed to disable TOP NOC\n");
- ret = -EIO;
- }
-
- if (ivpu_boot_host_ss_axi_disable(vdev)) {
- ivpu_err(vdev, "Failed to disable AXI\n");
- ret = -EIO;
- }
- }
-
- if (ivpu_boot_pwr_domain_disable(vdev)) {
- ivpu_err(vdev, "Failed to disable power domain\n");
+ if (ivpu_hw_mtl_reset(vdev)) {
+ ivpu_err(vdev, "Failed to reset the VPU\n");
ret = -EIO;
}
--
2.25.1
^ permalink raw reply related [flat|nested] 13+ messages in thread
* [PATCH v2 5/8] accel/ivpu: Disable buttress on device removal
2023-03-23 12:54 [PATCH v2 0/8] accel/ivpu: Fixes for 6.3 Stanislaw Gruszka
` (3 preceding siblings ...)
2023-03-23 12:55 ` [PATCH v2 4/8] accel/ivpu: Fix power down sequence Stanislaw Gruszka
@ 2023-03-23 12:55 ` Stanislaw Gruszka
2023-03-23 12:55 ` [PATCH v2 6/8] accel/ivpu: Remove support for 1 tile SKUs Stanislaw Gruszka
` (2 subsequent siblings)
7 siblings, 0 replies; 13+ messages in thread
From: Stanislaw Gruszka @ 2023-03-23 12:55 UTC (permalink / raw)
To: dri-devel; +Cc: Stanislaw Gruszka, Oded Gabbay, Jeffrey Hugo, Jacek Lawrynowicz
Use pci_set_power_state() to disable buttress when device is removed.
This is workaround of hardware bug that hangs the system.
Additionally not disabling buttress prevents CPU enter deeper Pkg-C
states when the driver is unloaded or fail to probe.
Fixes: 35b137630f08 ("accel/ivpu: Introduce a new DRM driver for Intel VPU")
Signed-off-by: Stanislaw Gruszka <stanislaw.gruszka@linux.intel.com>
Reviewed-by: Jeffrey Hugo <quic_jhugo@quicinc.com>
---
drivers/accel/ivpu/ivpu_drv.c | 4 ++++
drivers/accel/ivpu/ivpu_drv.h | 1 +
drivers/accel/ivpu/ivpu_hw_mtl.c | 1 +
3 files changed, 6 insertions(+)
diff --git a/drivers/accel/ivpu/ivpu_drv.c b/drivers/accel/ivpu/ivpu_drv.c
index 70245cf84593..6a320a73e3cc 100644
--- a/drivers/accel/ivpu/ivpu_drv.c
+++ b/drivers/accel/ivpu/ivpu_drv.c
@@ -569,6 +569,8 @@ static int ivpu_dev_init(struct ivpu_device *vdev)
ivpu_mmu_global_context_fini(vdev);
err_power_down:
ivpu_hw_power_down(vdev);
+ if (IVPU_WA(d3hot_after_power_off))
+ pci_set_power_state(to_pci_dev(vdev->drm.dev), PCI_D3hot);
err_xa_destroy:
xa_destroy(&vdev->submitted_jobs_xa);
xa_destroy(&vdev->context_xa);
@@ -579,6 +581,8 @@ static void ivpu_dev_fini(struct ivpu_device *vdev)
{
ivpu_pm_disable(vdev);
ivpu_shutdown(vdev);
+ if (IVPU_WA(d3hot_after_power_off))
+ pci_set_power_state(to_pci_dev(vdev->drm.dev), PCI_D3hot);
ivpu_job_done_thread_fini(vdev);
ivpu_pm_cancel_recovery(vdev);
diff --git a/drivers/accel/ivpu/ivpu_drv.h b/drivers/accel/ivpu/ivpu_drv.h
index ef12a38e06e1..d3013fbd13b3 100644
--- a/drivers/accel/ivpu/ivpu_drv.h
+++ b/drivers/accel/ivpu/ivpu_drv.h
@@ -74,6 +74,7 @@
struct ivpu_wa_table {
bool punit_disabled;
bool clear_runtime_mem;
+ bool d3hot_after_power_off;
};
struct ivpu_hw_info;
diff --git a/drivers/accel/ivpu/ivpu_hw_mtl.c b/drivers/accel/ivpu/ivpu_hw_mtl.c
index 70ca6de78060..133ba33d2866 100644
--- a/drivers/accel/ivpu/ivpu_hw_mtl.c
+++ b/drivers/accel/ivpu/ivpu_hw_mtl.c
@@ -101,6 +101,7 @@ static void ivpu_hw_wa_init(struct ivpu_device *vdev)
{
vdev->wa.punit_disabled = ivpu_is_fpga(vdev);
vdev->wa.clear_runtime_mem = false;
+ vdev->wa.d3hot_after_power_off = true;
}
static void ivpu_hw_timeouts_init(struct ivpu_device *vdev)
--
2.25.1
^ permalink raw reply related [flat|nested] 13+ messages in thread
* [PATCH v2 6/8] accel/ivpu: Remove support for 1 tile SKUs
2023-03-23 12:54 [PATCH v2 0/8] accel/ivpu: Fixes for 6.3 Stanislaw Gruszka
` (4 preceding siblings ...)
2023-03-23 12:55 ` [PATCH v2 5/8] accel/ivpu: Disable buttress on device removal Stanislaw Gruszka
@ 2023-03-23 12:55 ` Stanislaw Gruszka
2023-03-23 12:55 ` [PATCH v2 7/8] accel/ivpu: Fix VPU clock calculation Stanislaw Gruszka
2023-03-23 12:55 ` [PATCH v2 8/8] accel/ivpu: Fix IPC buffer header status field value Stanislaw Gruszka
7 siblings, 0 replies; 13+ messages in thread
From: Stanislaw Gruszka @ 2023-03-23 12:55 UTC (permalink / raw)
To: dri-devel
Cc: Stanislaw Gruszka, Oded Gabbay, Jeffrey Hugo, Jacek Lawrynowicz,
Andrzej Kacprowski
The support for single tile SKUs was dropped from MTL.
Note that we can still boot the VPU with 1-tile work point
config - this is independent from number of tiles present
in the VPU.
Co-developed-by: Andrzej Kacprowski <andrzej.kacprowski@linux.intel.com>
Signed-off-by: Andrzej Kacprowski <andrzej.kacprowski@linux.intel.com>
Signed-off-by: Stanislaw Gruszka <stanislaw.gruszka@linux.intel.com>
Reviewed-by: Jeffrey Hugo <quic_jhugo@quicinc.com>
---
drivers/accel/ivpu/ivpu_hw_mtl.c | 59 ++++++++++----------------------
1 file changed, 18 insertions(+), 41 deletions(-)
diff --git a/drivers/accel/ivpu/ivpu_hw_mtl.c b/drivers/accel/ivpu/ivpu_hw_mtl.c
index 133ba33d2866..98c8a4aa25f0 100644
--- a/drivers/accel/ivpu/ivpu_hw_mtl.c
+++ b/drivers/accel/ivpu/ivpu_hw_mtl.c
@@ -12,20 +12,20 @@
#include "ivpu_mmu.h"
#include "ivpu_pm.h"
-#define TILE_FUSE_ENABLE_BOTH 0x0
-#define TILE_FUSE_ENABLE_UPPER 0x1
-#define TILE_FUSE_ENABLE_LOWER 0x2
-
-#define TILE_SKU_BOTH_MTL 0x3630
-#define TILE_SKU_LOWER_MTL 0x3631
-#define TILE_SKU_UPPER_MTL 0x3632
+#define TILE_FUSE_ENABLE_BOTH 0x0
+#define TILE_SKU_BOTH_MTL 0x3630
/* Work point configuration values */
-#define WP_CONFIG_1_TILE_5_3_RATIO 0x0101
-#define WP_CONFIG_1_TILE_4_3_RATIO 0x0102
-#define WP_CONFIG_2_TILE_5_3_RATIO 0x0201
-#define WP_CONFIG_2_TILE_4_3_RATIO 0x0202
-#define WP_CONFIG_0_TILE_PLL_OFF 0x0000
+#define CONFIG_1_TILE 0x01
+#define CONFIG_2_TILE 0x02
+#define PLL_RATIO_5_3 0x01
+#define PLL_RATIO_4_3 0x02
+#define WP_CONFIG(tile, ratio) (((tile) << 8) | (ratio))
+#define WP_CONFIG_1_TILE_5_3_RATIO WP_CONFIG(CONFIG_1_TILE, PLL_RATIO_5_3)
+#define WP_CONFIG_1_TILE_4_3_RATIO WP_CONFIG(CONFIG_1_TILE, PLL_RATIO_4_3)
+#define WP_CONFIG_2_TILE_5_3_RATIO WP_CONFIG(CONFIG_2_TILE, PLL_RATIO_5_3)
+#define WP_CONFIG_2_TILE_4_3_RATIO WP_CONFIG(CONFIG_2_TILE, PLL_RATIO_4_3)
+#define WP_CONFIG_0_TILE_PLL_OFF WP_CONFIG(0, 0)
#define PLL_REF_CLK_FREQ (50 * 1000000)
#define PLL_SIMULATION_FREQ (10 * 1000000)
@@ -219,7 +219,8 @@ static int ivpu_pll_drive(struct ivpu_device *vdev, bool enable)
config = 0;
}
- ivpu_dbg(vdev, PM, "PLL workpoint request: %d Hz\n", PLL_RATIO_TO_FREQ(target_ratio));
+ ivpu_dbg(vdev, PM, "PLL workpoint request: config 0x%04x pll ratio 0x%x\n",
+ config, target_ratio);
ret = ivpu_pll_cmd_send(vdev, hw->pll.min_ratio, hw->pll.max_ratio, target_ratio, config);
if (ret) {
@@ -610,34 +611,10 @@ static int ivpu_boot_d0i3_drive(struct ivpu_device *vdev, bool enable)
static int ivpu_hw_mtl_info_init(struct ivpu_device *vdev)
{
struct ivpu_hw_info *hw = vdev->hw;
- u32 tile_fuse;
-
- tile_fuse = REGB_RD32(MTL_BUTTRESS_TILE_FUSE);
- if (!REG_TEST_FLD(MTL_BUTTRESS_TILE_FUSE, VALID, tile_fuse))
- ivpu_warn(vdev, "Tile Fuse: Invalid (0x%x)\n", tile_fuse);
-
- hw->tile_fuse = REG_GET_FLD(MTL_BUTTRESS_TILE_FUSE, SKU, tile_fuse);
- switch (hw->tile_fuse) {
- case TILE_FUSE_ENABLE_LOWER:
- hw->sku = TILE_SKU_LOWER_MTL;
- hw->config = WP_CONFIG_1_TILE_5_3_RATIO;
- ivpu_dbg(vdev, MISC, "Tile Fuse: Enable Lower\n");
- break;
- case TILE_FUSE_ENABLE_UPPER:
- hw->sku = TILE_SKU_UPPER_MTL;
- hw->config = WP_CONFIG_1_TILE_4_3_RATIO;
- ivpu_dbg(vdev, MISC, "Tile Fuse: Enable Upper\n");
- break;
- case TILE_FUSE_ENABLE_BOTH:
- hw->sku = TILE_SKU_BOTH_MTL;
- hw->config = WP_CONFIG_2_TILE_5_3_RATIO;
- ivpu_dbg(vdev, MISC, "Tile Fuse: Enable Both\n");
- break;
- default:
- hw->config = WP_CONFIG_0_TILE_PLL_OFF;
- ivpu_dbg(vdev, MISC, "Tile Fuse: Disable\n");
- break;
- }
+
+ hw->tile_fuse = TILE_FUSE_ENABLE_BOTH;
+ hw->sku = TILE_SKU_BOTH_MTL;
+ hw->config = WP_CONFIG_2_TILE_4_3_RATIO;
ivpu_pll_init_frequency_ratios(vdev);
--
2.25.1
^ permalink raw reply related [flat|nested] 13+ messages in thread
* [PATCH v2 7/8] accel/ivpu: Fix VPU clock calculation
2023-03-23 12:54 [PATCH v2 0/8] accel/ivpu: Fixes for 6.3 Stanislaw Gruszka
` (5 preceding siblings ...)
2023-03-23 12:55 ` [PATCH v2 6/8] accel/ivpu: Remove support for 1 tile SKUs Stanislaw Gruszka
@ 2023-03-23 12:55 ` Stanislaw Gruszka
2023-03-23 15:01 ` Jeffrey Hugo
2023-03-23 12:55 ` [PATCH v2 8/8] accel/ivpu: Fix IPC buffer header status field value Stanislaw Gruszka
7 siblings, 1 reply; 13+ messages in thread
From: Stanislaw Gruszka @ 2023-03-23 12:55 UTC (permalink / raw)
To: dri-devel
Cc: Stanislaw Gruszka, Oded Gabbay, Jeffrey Hugo, Jacek Lawrynowicz,
Andrzej Kacprowski
The driver calculates the wrong frequency because it ignores
the workpoint config and this cause undesired power/performance
characteristics. Fix this by using the workpoint config in
the freq calculations.
Fixes: 35b137630f08 ("accel/ivpu: Introduce a new DRM driver for Intel VPU")
Co-developed-by: Andrzej Kacprowski <andrzej.kacprowski@linux.intel.com>
Signed-off-by: Andrzej Kacprowski <andrzej.kacprowski@linux.intel.com>
Signed-off-by: Stanislaw Gruszka <stanislaw.gruszka@linux.intel.com>
---
drivers/accel/ivpu/ivpu_hw_mtl.c | 16 ++++++++++++++--
1 file changed, 14 insertions(+), 2 deletions(-)
diff --git a/drivers/accel/ivpu/ivpu_hw_mtl.c b/drivers/accel/ivpu/ivpu_hw_mtl.c
index 98c8a4aa25f0..382ec127be8e 100644
--- a/drivers/accel/ivpu/ivpu_hw_mtl.c
+++ b/drivers/accel/ivpu/ivpu_hw_mtl.c
@@ -29,7 +29,6 @@
#define PLL_REF_CLK_FREQ (50 * 1000000)
#define PLL_SIMULATION_FREQ (10 * 1000000)
-#define PLL_RATIO_TO_FREQ(x) ((x) * PLL_REF_CLK_FREQ)
#define PLL_DEFAULT_EPP_VALUE 0x80
#define TIM_SAFE_ENABLE 0xf1d0dead
@@ -789,6 +788,19 @@ static void ivpu_hw_mtl_wdt_disable(struct ivpu_device *vdev)
REGV_WR32(MTL_VPU_CPU_SS_TIM_GEN_CONFIG, val);
}
+static u32 ivpu_hw_mtl_pll_to_freq(u32 ratio, u32 config)
+{
+ u32 pll_clock = PLL_REF_CLK_FREQ * ratio;
+ u32 cpu_clock;
+
+ if ((config & 0xff) == PLL_RATIO_4_3)
+ cpu_clock = pll_clock * 2 / 4;
+ else
+ cpu_clock = pll_clock * 2 / 5;
+
+ return cpu_clock;
+}
+
/* Register indirect accesses */
static u32 ivpu_hw_mtl_reg_pll_freq_get(struct ivpu_device *vdev)
{
@@ -800,7 +812,7 @@ static u32 ivpu_hw_mtl_reg_pll_freq_get(struct ivpu_device *vdev)
if (!ivpu_is_silicon(vdev))
return PLL_SIMULATION_FREQ;
- return PLL_RATIO_TO_FREQ(pll_curr_ratio);
+ return ivpu_hw_mtl_pll_to_freq(pll_curr_ratio, vdev->hw->config);
}
static u32 ivpu_hw_mtl_reg_telemetry_offset_get(struct ivpu_device *vdev)
--
2.25.1
^ permalink raw reply related [flat|nested] 13+ messages in thread
* [PATCH v2 8/8] accel/ivpu: Fix IPC buffer header status field value
2023-03-23 12:54 [PATCH v2 0/8] accel/ivpu: Fixes for 6.3 Stanislaw Gruszka
` (6 preceding siblings ...)
2023-03-23 12:55 ` [PATCH v2 7/8] accel/ivpu: Fix VPU clock calculation Stanislaw Gruszka
@ 2023-03-23 12:55 ` Stanislaw Gruszka
2023-03-23 15:02 ` Jeffrey Hugo
7 siblings, 1 reply; 13+ messages in thread
From: Stanislaw Gruszka @ 2023-03-23 12:55 UTC (permalink / raw)
To: dri-devel
Cc: Andrzej Kacprowski, Oded Gabbay, Jeffrey Hugo, Jacek Lawrynowicz,
Stanislaw Gruszka
From: Andrzej Kacprowski <andrzej.kacprowski@linux.intel.com>
IPC messages transmitted to the device must be marked as
allocated - status field must be set to 1. The VPU driver
has IVPU_IPC_HDR_ALLOCATED incorrectly defined. Future VPU
firmware versions will reject all IPC messages with invalid
status and will not work with a VPU driver that is missing
this fix.
Fixes: 5d7422cfb498 ("accel/ivpu: Add IPC driver and JSM messages")
Signed-off-by: Andrzej Kacprowski <andrzej.kacprowski@linux.intel.com>
Signed-off-by: Stanislaw Gruszka <stanislaw.gruszka@linux.intel.com>
---
drivers/accel/ivpu/ivpu_ipc.h | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/accel/ivpu/ivpu_ipc.h b/drivers/accel/ivpu/ivpu_ipc.h
index 9838202ecfad..68f5b6668e00 100644
--- a/drivers/accel/ivpu/ivpu_ipc.h
+++ b/drivers/accel/ivpu/ivpu_ipc.h
@@ -21,7 +21,7 @@ struct ivpu_bo;
#define IVPU_IPC_ALIGNMENT 64
#define IVPU_IPC_HDR_FREE 0
-#define IVPU_IPC_HDR_ALLOCATED 0
+#define IVPU_IPC_HDR_ALLOCATED 1
/**
* struct ivpu_ipc_hdr - The IPC message header structure, exchanged
--
2.25.1
^ permalink raw reply related [flat|nested] 13+ messages in thread
* Re: [PATCH v2 2/8] accel/ivpu: Cancel recovery work
2023-03-23 12:54 ` [PATCH v2 2/8] accel/ivpu: Cancel recovery work Stanislaw Gruszka
@ 2023-03-23 14:59 ` Jeffrey Hugo
0 siblings, 0 replies; 13+ messages in thread
From: Jeffrey Hugo @ 2023-03-23 14:59 UTC (permalink / raw)
To: Stanislaw Gruszka, dri-devel; +Cc: Oded Gabbay, Jacek Lawrynowicz
On 3/23/2023 6:54 AM, Stanislaw Gruszka wrote:
> Prevent running recovery_work after device is removed.
>
> Fixes: 852be13f3bd3 ("accel/ivpu: Add PM support")
> Signed-off-by: Stanislaw Gruszka <stanislaw.gruszka@linux.intel.com>
Reviewed-by: Jeffrey Hugo <quic_jhugo@quicinc.com>
^ permalink raw reply [flat|nested] 13+ messages in thread
* Re: [PATCH v2 3/8] accel/ivpu: Do not use SSID 1
2023-03-23 12:54 ` [PATCH v2 3/8] accel/ivpu: Do not use SSID 1 Stanislaw Gruszka
@ 2023-03-23 15:00 ` Jeffrey Hugo
0 siblings, 0 replies; 13+ messages in thread
From: Jeffrey Hugo @ 2023-03-23 15:00 UTC (permalink / raw)
To: Stanislaw Gruszka, dri-devel
Cc: Andrzej Kacprowski, Oded Gabbay, Jacek Lawrynowicz
On 3/23/2023 6:54 AM, Stanislaw Gruszka wrote:
> The SSID=1 is used by the firmware as default value in
> case SSID mapping is not initialized. This allows
> detecting use of miss-configured memory contexts.
> The future FW versions may not allow using SSID=1.
>
> SSID=65 is valid value, number of contexts are limited
> by number of available command queues, but SSID can be
> any u16 value.
>
> Fixes: 35b137630f08 ("accel/ivpu: Introduce a new DRM driver for Intel VPU")
> Co-developed-by: Andrzej Kacprowski <andrzej.kacprowski@linux.intel.com>
> Signed-off-by: Andrzej Kacprowski <andrzej.kacprowski@linux.intel.com>
> Signed-off-by: Stanislaw Gruszka <stanislaw.gruszka@linux.intel.com>
Reviewed-by: Jeffrey Hugo <quic_jhugo@quicinc.com>
^ permalink raw reply [flat|nested] 13+ messages in thread
* Re: [PATCH v2 7/8] accel/ivpu: Fix VPU clock calculation
2023-03-23 12:55 ` [PATCH v2 7/8] accel/ivpu: Fix VPU clock calculation Stanislaw Gruszka
@ 2023-03-23 15:01 ` Jeffrey Hugo
0 siblings, 0 replies; 13+ messages in thread
From: Jeffrey Hugo @ 2023-03-23 15:01 UTC (permalink / raw)
To: Stanislaw Gruszka, dri-devel
Cc: Andrzej Kacprowski, Oded Gabbay, Jacek Lawrynowicz
On 3/23/2023 6:55 AM, Stanislaw Gruszka wrote:
> The driver calculates the wrong frequency because it ignores
> the workpoint config and this cause undesired power/performance
> characteristics. Fix this by using the workpoint config in
> the freq calculations.
>
> Fixes: 35b137630f08 ("accel/ivpu: Introduce a new DRM driver for Intel VPU")
> Co-developed-by: Andrzej Kacprowski <andrzej.kacprowski@linux.intel.com>
> Signed-off-by: Andrzej Kacprowski <andrzej.kacprowski@linux.intel.com>
> Signed-off-by: Stanislaw Gruszka <stanislaw.gruszka@linux.intel.com>
Reviewed-by: Jeffrey Hugo <quic_jhugo@quicinc.com>
^ permalink raw reply [flat|nested] 13+ messages in thread
* Re: [PATCH v2 8/8] accel/ivpu: Fix IPC buffer header status field value
2023-03-23 12:55 ` [PATCH v2 8/8] accel/ivpu: Fix IPC buffer header status field value Stanislaw Gruszka
@ 2023-03-23 15:02 ` Jeffrey Hugo
0 siblings, 0 replies; 13+ messages in thread
From: Jeffrey Hugo @ 2023-03-23 15:02 UTC (permalink / raw)
To: Stanislaw Gruszka, dri-devel
Cc: Andrzej Kacprowski, Oded Gabbay, Jacek Lawrynowicz
On 3/23/2023 6:55 AM, Stanislaw Gruszka wrote:
> From: Andrzej Kacprowski <andrzej.kacprowski@linux.intel.com>
>
> IPC messages transmitted to the device must be marked as
> allocated - status field must be set to 1. The VPU driver
> has IVPU_IPC_HDR_ALLOCATED incorrectly defined. Future VPU
> firmware versions will reject all IPC messages with invalid
> status and will not work with a VPU driver that is missing
> this fix.
>
> Fixes: 5d7422cfb498 ("accel/ivpu: Add IPC driver and JSM messages")
> Signed-off-by: Andrzej Kacprowski <andrzej.kacprowski@linux.intel.com>
> Signed-off-by: Stanislaw Gruszka <stanislaw.gruszka@linux.intel.com>
Reviewed-by: Jeffrey Hugo <quic_jhugo@quicinc.com>
^ permalink raw reply [flat|nested] 13+ messages in thread
end of thread, other threads:[~2023-03-23 15:02 UTC | newest]
Thread overview: 13+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2023-03-23 12:54 [PATCH v2 0/8] accel/ivpu: Fixes for 6.3 Stanislaw Gruszka
2023-03-23 12:54 ` [PATCH v2 1/8] accel/ivpu: Do not access HW registers after unbind Stanislaw Gruszka
2023-03-23 12:54 ` [PATCH v2 2/8] accel/ivpu: Cancel recovery work Stanislaw Gruszka
2023-03-23 14:59 ` Jeffrey Hugo
2023-03-23 12:54 ` [PATCH v2 3/8] accel/ivpu: Do not use SSID 1 Stanislaw Gruszka
2023-03-23 15:00 ` Jeffrey Hugo
2023-03-23 12:55 ` [PATCH v2 4/8] accel/ivpu: Fix power down sequence Stanislaw Gruszka
2023-03-23 12:55 ` [PATCH v2 5/8] accel/ivpu: Disable buttress on device removal Stanislaw Gruszka
2023-03-23 12:55 ` [PATCH v2 6/8] accel/ivpu: Remove support for 1 tile SKUs Stanislaw Gruszka
2023-03-23 12:55 ` [PATCH v2 7/8] accel/ivpu: Fix VPU clock calculation Stanislaw Gruszka
2023-03-23 15:01 ` Jeffrey Hugo
2023-03-23 12:55 ` [PATCH v2 8/8] accel/ivpu: Fix IPC buffer header status field value Stanislaw Gruszka
2023-03-23 15:02 ` Jeffrey Hugo
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox