[PATCH 0/6] accel/ivpu: Changes for 6.15

[PATCH 0/6] accel/ivpu: Changes for 6.15

[Jacek Lawrynowicz]

Most notable is the addition of hardware fault injection support which allows
to test error handling paths in the driver.

Jacek Lawrynowicz (2):
  accel/ivpu: Add support for hardware fault injection
  accel/ivpu: Update last_busy in IRQ handler

Karol Wachowski (3):
  accel/ivpu: Fix missing MMU events if file_priv is unbound
  accel/ivpu: Turn on HWS by default on all platforms
  accel/ivpu: Move recovery work to system_unbound_wq

Tomasz Rusinowicz (1):
  accel/ivpu: Enable recovery and adjust timeouts for fpga

 drivers/accel/ivpu/ivpu_debugfs.c   |  5 +++
 drivers/accel/ivpu/ivpu_fw.c        | 10 +----
 drivers/accel/ivpu/ivpu_hw.c        | 33 +++++++++++----
 drivers/accel/ivpu/ivpu_hw_ip.c     |  4 +-
 drivers/accel/ivpu/ivpu_hw_reg_io.h | 64 +++++++++++++++++------------
 drivers/accel/ivpu/ivpu_job.c       |  8 +++-
 drivers/accel/ivpu/ivpu_pm.c        |  7 +---
 7 files changed, 78 insertions(+), 53 deletions(-)

--
2.45.1

[PATCH 1/6] accel/ivpu: Add support for hardware fault injection

[Jacek Lawrynowicz]

This commit introduces the capability to simulate hardware faults for
testing purposes. The new `fail_hw` fault can be injected in
`ivpu_hw_reg_poll_fld()`, which is used in various parts of the driver
to wait for the hardware to reach a specific state. This allows to test
failures during NPU boot and shutdown, IPC message handling and more.

Fault injection can be enabled using debugfs or a module parameter.

Reviewed-by: Maciej Falkowski <maciej.falkowski@linux.intel.com>
Signed-off-by: Jacek Lawrynowicz <jacek.lawrynowicz@linux.intel.com>
---
 drivers/accel/ivpu/ivpu_debugfs.c   |  5 +++
 drivers/accel/ivpu/ivpu_hw.c        | 14 +++++++
 drivers/accel/ivpu/ivpu_hw_ip.c     |  4 +-
 drivers/accel/ivpu/ivpu_hw_reg_io.h | 64 +++++++++++++++++------------
 4 files changed, 58 insertions(+), 29 deletions(-)

diff --git a/drivers/accel/ivpu/ivpu_debugfs.c b/drivers/accel/ivpu/ivpu_debugfs.c
index 8180b95ed69dc..e79715c53f2a0 100644
--- a/drivers/accel/ivpu/ivpu_debugfs.c
+++ b/drivers/accel/ivpu/ivpu_debugfs.c
@@ -4,6 +4,7 @@
  */
 
 #include <linux/debugfs.h>
+#include <linux/fault-inject.h>
 
 #include <drm/drm_debugfs.h>
 #include <drm/drm_file.h>
@@ -430,4 +431,8 @@ void ivpu_debugfs_init(struct ivpu_device *vdev)
 				    debugfs_root, vdev, &fw_profiling_freq_fops);
 		debugfs_create_file("dct", 0644, debugfs_root, vdev, &ivpu_dct_fops);
 	}
+
+#ifdef CONFIG_FAULT_INJECTION
+	fault_create_debugfs_attr("fail_hw", debugfs_root, &ivpu_hw_failure);
+#endif
 }
diff --git a/drivers/accel/ivpu/ivpu_hw.c b/drivers/accel/ivpu/ivpu_hw.c
index e332f19ab51de..8099ab047bfe6 100644
--- a/drivers/accel/ivpu/ivpu_hw.c
+++ b/drivers/accel/ivpu/ivpu_hw.c
@@ -9,6 +9,15 @@
 #include "ivpu_hw_ip.h"
 
 #include <linux/dmi.h>
+#include <linux/fault-inject.h>
+
+#ifdef CONFIG_FAULT_INJECTION
+DECLARE_FAULT_ATTR(ivpu_hw_failure);
+
+static char *ivpu_fail_hw;
+module_param_named_unsafe(fail_hw, ivpu_fail_hw, charp, 0444);
+MODULE_PARM_DESC(fail_hw, "<interval>,<probability>,<space>,<times>");
+#endif
 
 static char *platform_to_str(u32 platform)
 {
@@ -247,6 +256,11 @@ int ivpu_hw_init(struct ivpu_device *vdev)
 	timeouts_init(vdev);
 	atomic_set(&vdev->hw->firewall_irq_counter, 0);
 
+#ifdef CONFIG_FAULT_INJECTION
+	if (ivpu_fail_hw)
+		setup_fault_attr(&ivpu_hw_failure, ivpu_fail_hw);
+#endif
+
 	return 0;
 }
 
diff --git a/drivers/accel/ivpu/ivpu_hw_ip.c b/drivers/accel/ivpu/ivpu_hw_ip.c
index 029dd065614b2..823f6a57dc546 100644
--- a/drivers/accel/ivpu/ivpu_hw_ip.c
+++ b/drivers/accel/ivpu/ivpu_hw_ip.c
@@ -968,14 +968,14 @@ void ivpu_hw_ip_wdt_disable(struct ivpu_device *vdev)
 
 static u32 ipc_rx_count_get_37xx(struct ivpu_device *vdev)
 {
-	u32 count = REGV_RD32_SILENT(VPU_37XX_HOST_SS_TIM_IPC_FIFO_STAT);
+	u32 count = readl(vdev->regv + VPU_37XX_HOST_SS_TIM_IPC_FIFO_STAT);
 
 	return REG_GET_FLD(VPU_37XX_HOST_SS_TIM_IPC_FIFO_STAT, FILL_LEVEL, count);
 }
 
 static u32 ipc_rx_count_get_40xx(struct ivpu_device *vdev)
 {
-	u32 count = REGV_RD32_SILENT(VPU_40XX_HOST_SS_TIM_IPC_FIFO_STAT);
+	u32 count = readl(vdev->regv + VPU_40XX_HOST_SS_TIM_IPC_FIFO_STAT);
 
 	return REG_GET_FLD(VPU_40XX_HOST_SS_TIM_IPC_FIFO_STAT, FILL_LEVEL, count);
 }
diff --git a/drivers/accel/ivpu/ivpu_hw_reg_io.h b/drivers/accel/ivpu/ivpu_hw_reg_io.h
index 79b3f441eac4d..66259b0ead026 100644
--- a/drivers/accel/ivpu/ivpu_hw_reg_io.h
+++ b/drivers/accel/ivpu/ivpu_hw_reg_io.h
@@ -7,6 +7,7 @@
 #define __IVPU_HW_REG_IO_H__
 
 #include <linux/bitfield.h>
+#include <linux/fault-inject.h>
 #include <linux/io.h>
 #include <linux/iopoll.h>
 
@@ -16,13 +17,11 @@
 #define REG_IO_ERROR      0xffffffff
 
 #define REGB_RD32(reg)          ivpu_hw_reg_rd32(vdev, vdev->regb, (reg), #reg, __func__)
-#define REGB_RD32_SILENT(reg)   readl(vdev->regb + (reg))
 #define REGB_RD64(reg)          ivpu_hw_reg_rd64(vdev, vdev->regb, (reg), #reg, __func__)
 #define REGB_WR32(reg, val)     ivpu_hw_reg_wr32(vdev, vdev->regb, (reg), (val), #reg, __func__)
 #define REGB_WR64(reg, val)     ivpu_hw_reg_wr64(vdev, vdev->regb, (reg), (val), #reg, __func__)
 
 #define REGV_RD32(reg)          ivpu_hw_reg_rd32(vdev, vdev->regv, (reg), #reg, __func__)
-#define REGV_RD32_SILENT(reg)   readl(vdev->regv + (reg))
 #define REGV_RD64(reg)          ivpu_hw_reg_rd64(vdev, vdev->regv, (reg), #reg, __func__)
 #define REGV_WR32(reg, val)     ivpu_hw_reg_wr32(vdev, vdev->regv, (reg), (val), #reg, __func__)
 #define REGV_WR64(reg, val)     ivpu_hw_reg_wr64(vdev, vdev->regv, (reg), (val), #reg, __func__)
@@ -47,31 +46,42 @@
 #define REG_TEST_FLD_NUM(REG, FLD, num, val) \
 	((num) == FIELD_GET(REG##_##FLD##_MASK, val))
 
-#define REGB_POLL_FLD(reg, fld, val, timeout_us) \
-({ \
-	u32 var; \
-	int r; \
-	ivpu_dbg(vdev, REG, "%s : %s (0x%08x) Polling field %s started (expected 0x%x)\n", \
-		 __func__, #reg, reg, #fld, val); \
-	r = read_poll_timeout(REGB_RD32_SILENT, var, (FIELD_GET(reg##_##fld##_MASK, var) == (val)),\
-			      REG_POLL_SLEEP_US, timeout_us, false, (reg)); \
-	ivpu_dbg(vdev, REG, "%s : %s (0x%08x) Polling field %s %s (reg val 0x%08x)\n", \
-		 __func__, #reg, reg, #fld, r ? "ETIMEDOUT" : "OK", var); \
-	r; \
-})
-
-#define REGV_POLL_FLD(reg, fld, val, timeout_us) \
-({ \
-	u32 var; \
-	int r; \
-	ivpu_dbg(vdev, REG, "%s : %s (0x%08x) Polling field %s started (expected 0x%x)\n", \
-		 __func__, #reg, reg, #fld, val); \
-	r = read_poll_timeout(REGV_RD32_SILENT, var, (FIELD_GET(reg##_##fld##_MASK, var) == (val)),\
-			      REG_POLL_SLEEP_US, timeout_us, false, (reg)); \
-	ivpu_dbg(vdev, REG, "%s : %s (0x%08x) Polling field %s %s (reg val 0x%08x)\n", \
-		 __func__, #reg, reg, #fld, r ? "ETIMEDOUT" : "OK", var); \
-	r; \
-})
+#define REGB_POLL_FLD(reg, fld, exp_fld_val, timeout_us) \
+	ivpu_hw_reg_poll_fld(vdev, vdev->regb, reg, reg##_##fld##_MASK, \
+			     FIELD_PREP(reg##_##fld##_MASK, exp_fld_val), timeout_us, \
+			     __func__, #reg, #fld)
+
+#define REGV_POLL_FLD(reg, fld, exp_fld_val, timeout_us) \
+	ivpu_hw_reg_poll_fld(vdev, vdev->regv, reg, reg##_##fld##_MASK, \
+			     FIELD_PREP(reg##_##fld##_MASK, exp_fld_val), timeout_us, \
+			     __func__, #reg, #fld)
+
+extern struct fault_attr ivpu_hw_failure;
+
+static inline int __must_check
+ivpu_hw_reg_poll_fld(struct ivpu_device *vdev, void __iomem *base,
+		     u32 reg_offset, u32 reg_mask, u32 exp_masked_val, u32 timeout_us,
+		     const char *func_name, const char *reg_name, const char *fld_name)
+{
+	u32 reg_val;
+	int ret;
+
+	ivpu_dbg(vdev, REG, "%s : %s (0x%08x) POLL %s started (exp_val 0x%x)\n",
+		 func_name, reg_name, reg_offset, fld_name, exp_masked_val);
+
+	ret = read_poll_timeout(readl, reg_val, (reg_val & reg_mask) == exp_masked_val,
+				REG_POLL_SLEEP_US, timeout_us, false, base + reg_offset);
+
+#ifdef CONFIG_FAULT_INJECTION
+	if (should_fail(&ivpu_hw_failure, 1))
+		ret = -ETIMEDOUT;
+#endif
+
+	ivpu_dbg(vdev, REG, "%s : %s (0x%08x) POLL %s %s (reg_val 0x%08x)\n",
+		 func_name, reg_name, reg_offset, fld_name, ret ? "ETIMEDOUT" : "OK", reg_val);
+
+	return ret;
+}
 
 static inline u32
 ivpu_hw_reg_rd32(struct ivpu_device *vdev, void __iomem *base, u32 reg,
-- 
2.45.1

[PATCH 2/6] accel/ivpu: Update last_busy in IRQ handler

[Jacek Lawrynowicz]

Call pm_runtime_mark_last_busy() in top half of IRQ handler to prevent
device from being runtime suspended before bottom half is executed on
a workqueue.

Reviewed-by: Karol Wachowski <karol.wachowski@intel.com>
Signed-off-by: Jacek Lawrynowicz <jacek.lawrynowicz@linux.intel.com>
---
 drivers/accel/ivpu/ivpu_hw.c | 9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)

diff --git a/drivers/accel/ivpu/ivpu_hw.c b/drivers/accel/ivpu/ivpu_hw.c
index 8099ab047bfe6..4199f0bbb435b 100644
--- a/drivers/accel/ivpu/ivpu_hw.c
+++ b/drivers/accel/ivpu/ivpu_hw.c
@@ -10,6 +10,7 @@
 
 #include <linux/dmi.h>
 #include <linux/fault-inject.h>
+#include <linux/pm_runtime.h>
 
 #ifdef CONFIG_FAULT_INJECTION
 DECLARE_FAULT_ATTR(ivpu_hw_failure);
@@ -331,7 +332,9 @@ irqreturn_t ivpu_hw_irq_handler(int irq, void *ptr)
 	/* Re-enable global interrupts to re-trigger MSI for pending interrupts */
 	ivpu_hw_btrs_global_int_enable(vdev);
 
-	if (ip_handled || btrs_handled)
-		return IRQ_HANDLED;
-	return IRQ_NONE;
+	if (!ip_handled && !btrs_handled)
+		return IRQ_NONE;
+
+	pm_runtime_mark_last_busy(vdev->drm.dev);
+	return IRQ_HANDLED;
 }
-- 
2.45.1

[PATCH 3/6] accel/ivpu: Fix missing MMU events if file_priv is unbound

[Jacek Lawrynowicz]

From: Karol Wachowski <karol.wachowski@intel.com>

Move the ivpu_mmu_discard_events() function to the common portion of
the abort work function. This ensures it is called only once, even if
there are no faulty contexts in context_xa, to guarantee that MMU events
are discarded and new events are not missed.

Reviewed-by: Jacek Lawrynowicz <jacek.lawrynowicz@linux.intel.com>
Signed-off-by: Karol Wachowski <karol.wachowski@intel.com>
Signed-off-by: Jacek Lawrynowicz <jacek.lawrynowicz@linux.intel.com>
---
 drivers/accel/ivpu/ivpu_job.c | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)

diff --git a/drivers/accel/ivpu/ivpu_job.c b/drivers/accel/ivpu/ivpu_job.c
index 3c162ac41a1dc..c1013f511efa6 100644
--- a/drivers/accel/ivpu/ivpu_job.c
+++ b/drivers/accel/ivpu/ivpu_job.c
@@ -375,7 +375,6 @@ void ivpu_context_abort_locked(struct ivpu_file_priv *file_priv)
 		ivpu_jsm_context_release(vdev, file_priv->ctx.id);
 
 	ivpu_mmu_disable_ssid_events(vdev, file_priv->ctx.id);
-	ivpu_mmu_discard_events(vdev);
 
 	file_priv->aborted = true;
 }
@@ -980,6 +979,13 @@ void ivpu_context_abort_work_fn(struct work_struct *work)
 	}
 	mutex_unlock(&vdev->context_list_lock);
 
+	/*
+	 * We will not receive new MMU event interrupts until existing events are discarded
+	 * however, we want to discard these events only after aborting the faulty context
+	 * to avoid generating new faults from that context
+	 */
+	ivpu_mmu_discard_events(vdev);
+
 	if (vdev->fw->sched_mode != VPU_SCHEDULING_MODE_HW)
 		return;
 
-- 
2.45.1

[PATCH 4/6] accel/ivpu: Turn on HWS by default on all platforms

[Jacek Lawrynowicz]

From: Karol Wachowski <karol.wachowski@intel.com>

Hardware scheduling (HWS) is supposed to be supported on all existing
platform with recent FW including pre-silicon ones. Turn on HWS by
default.

Reviewed-by: Jacek Lawrynowicz <jacek.lawrynowicz@linux.intel.com>
Signed-off-by: Karol Wachowski <karol.wachowski@intel.com>
Signed-off-by: Jacek Lawrynowicz <jacek.lawrynowicz@linux.intel.com>
---
 drivers/accel/ivpu/ivpu_fw.c | 10 +---------
 1 file changed, 1 insertion(+), 9 deletions(-)

diff --git a/drivers/accel/ivpu/ivpu_fw.c b/drivers/accel/ivpu/ivpu_fw.c
index 6cf1fb826d1ba..cfe8f79ec5a6f 100644
--- a/drivers/accel/ivpu/ivpu_fw.c
+++ b/drivers/accel/ivpu/ivpu_fw.c
@@ -148,15 +148,7 @@ ivpu_fw_sched_mode_select(struct ivpu_device *vdev, const struct vpu_firmware_he
 	if (IVPU_FW_CHECK_API_VER_LT(vdev, fw_hdr, JSM, 3, 24))
 		return VPU_SCHEDULING_MODE_OS;
 
-	switch (ivpu_device_id(vdev)) {
-	case PCI_DEVICE_ID_MTL:
-	case PCI_DEVICE_ID_ARL:
-	case PCI_DEVICE_ID_LNL:
-	case PCI_DEVICE_ID_PTL_P:
-		return VPU_SCHEDULING_MODE_HW;
-	default:
-		return VPU_SCHEDULING_MODE_OS;
-	}
+	return VPU_SCHEDULING_MODE_HW;
 }
 
 static int ivpu_fw_parse(struct ivpu_device *vdev)
-- 
2.45.1

[PATCH 5/6] accel/ivpu: Enable recovery and adjust timeouts for fpga

[Jacek Lawrynowicz]

From: Tomasz Rusinowicz <tomasz.rusinowicz@intel.com>

Recovery now works on fpga. JSM state dump timeout needs to
be really long for the new fpga model releases.

Enable punit on fpga.

Reviewed-by: Jacek Lawrynowicz <jacek.lawrynowicz@linux.intel.com>
Signed-off-by: Tomasz Rusinowicz <tomasz.rusinowicz@intel.com>
Signed-off-by: Jacek Lawrynowicz <jacek.lawrynowicz@linux.intel.com>
---
 drivers/accel/ivpu/ivpu_hw.c | 10 +++++-----
 drivers/accel/ivpu/ivpu_pm.c |  5 -----
 2 files changed, 5 insertions(+), 10 deletions(-)

diff --git a/drivers/accel/ivpu/ivpu_hw.c b/drivers/accel/ivpu/ivpu_hw.c
index 4199f0bbb435b..2057907c5aa6a 100644
--- a/drivers/accel/ivpu/ivpu_hw.c
+++ b/drivers/accel/ivpu/ivpu_hw.c
@@ -58,7 +58,7 @@ static void platform_init(struct ivpu_device *vdev)
 
 static void wa_init(struct ivpu_device *vdev)
 {
-	vdev->wa.punit_disabled = ivpu_is_fpga(vdev);
+	vdev->wa.punit_disabled = false;
 	vdev->wa.clear_runtime_mem = false;
 
 	if (ivpu_hw_btrs_gen(vdev) == IVPU_HW_BTRS_MTL)
@@ -87,12 +87,12 @@ static void timeouts_init(struct ivpu_device *vdev)
 		vdev->timeout.autosuspend = -1;
 		vdev->timeout.d0i3_entry_msg = -1;
 	} else if (ivpu_is_fpga(vdev)) {
-		vdev->timeout.boot = 100000;
-		vdev->timeout.jsm = 50000;
-		vdev->timeout.tdr = 2000000;
+		vdev->timeout.boot = 50;
+		vdev->timeout.jsm = 15000;
+		vdev->timeout.tdr = 30000;
 		vdev->timeout.autosuspend = -1;
 		vdev->timeout.d0i3_entry_msg = 500;
-		vdev->timeout.state_dump_msg = 10;
+		vdev->timeout.state_dump_msg = 10000;
 	} else if (ivpu_is_simics(vdev)) {
 		vdev->timeout.boot = 50;
 		vdev->timeout.jsm = 500;
diff --git a/drivers/accel/ivpu/ivpu_pm.c b/drivers/accel/ivpu/ivpu_pm.c
index f41b3bfe40af4..8f6222d157204 100644
--- a/drivers/accel/ivpu/ivpu_pm.c
+++ b/drivers/accel/ivpu/ivpu_pm.c
@@ -161,11 +161,6 @@ void ivpu_pm_trigger_recovery(struct ivpu_device *vdev, const char *reason)
 		return;
 	}
 
-	if (ivpu_is_fpga(vdev)) {
-		ivpu_err(vdev, "Recovery not available on FPGA\n");
-		return;
-	}
-
 	/* Trigger recovery if it's not in progress */
 	if (atomic_cmpxchg(&vdev->pm->reset_pending, 0, 1) == 0) {
 		ivpu_hw_diagnose_failure(vdev);
-- 
2.45.1

[PATCH 6/6] accel/ivpu: Move recovery work to system_unbound_wq

[Jacek Lawrynowicz]

From: Karol Wachowski <karol.wachowski@intel.com>

Recovery work doesn't need to be bound to any specific CPU, so move it
to unbound workqueue to improve execution time and system latency.

Reviewed-by: Jacek Lawrynowicz <jacek.lawrynowicz@linux.intel.com>
Signed-off-by: Karol Wachowski <karol.wachowski@intel.com>
Signed-off-by: Jacek Lawrynowicz <jacek.lawrynowicz@linux.intel.com>
---
 drivers/accel/ivpu/ivpu_pm.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/accel/ivpu/ivpu_pm.c b/drivers/accel/ivpu/ivpu_pm.c
index 8f6222d157204..d3db944ad8643 100644
--- a/drivers/accel/ivpu/ivpu_pm.c
+++ b/drivers/accel/ivpu/ivpu_pm.c
@@ -165,7 +165,7 @@ void ivpu_pm_trigger_recovery(struct ivpu_device *vdev, const char *reason)
 	if (atomic_cmpxchg(&vdev->pm->reset_pending, 0, 1) == 0) {
 		ivpu_hw_diagnose_failure(vdev);
 		ivpu_hw_irq_disable(vdev); /* Disable IRQ early to protect from IRQ storm */
-		queue_work(system_long_wq, &vdev->pm->recovery_work);
+		queue_work(system_unbound_wq, &vdev->pm->recovery_work);
 	}
 }
 
-- 
2.45.1

Re: [PATCH 1/6] accel/ivpu: Add support for hardware fault injection

[Jeffrey Hugo]

On 1/29/2025 5:56 AM, Jacek Lawrynowicz wrote:
> This commit introduces the capability to simulate hardware faults for

Nit - "This commit" is redundant.

> testing purposes. The new `fail_hw` fault can be injected in
> `ivpu_hw_reg_poll_fld()`, which is used in various parts of the driver
> to wait for the hardware to reach a specific state. This allows to test
> failures during NPU boot and shutdown, IPC message handling and more.
> 
> Fault injection can be enabled using debugfs or a module parameter.
> 
> Reviewed-by: Maciej Falkowski <maciej.falkowski@linux.intel.com>
> Signed-off-by: Jacek Lawrynowicz <jacek.lawrynowicz@linux.intel.com>

Reviewed-by: Jeffrey Hugo <quic_jhugo@quicinc.com>

Re: [PATCH 2/6] accel/ivpu: Update last_busy in IRQ handler

[Jeffrey Hugo]

On 1/29/2025 5:56 AM, Jacek Lawrynowicz wrote:
> Call pm_runtime_mark_last_busy() in top half of IRQ handler to prevent
> device from being runtime suspended before bottom half is executed on
> a workqueue.
> 
> Reviewed-by: Karol Wachowski <karol.wachowski@intel.com>
> Signed-off-by: Jacek Lawrynowicz <jacek.lawrynowicz@linux.intel.com>

Reviewed-by: Jeffrey Hugo <quic_jhugo@quicinc.com>

Re: [PATCH 3/6] accel/ivpu: Fix missing MMU events if file_priv is unbound

[Jeffrey Hugo]

On 1/29/2025 5:56 AM, Jacek Lawrynowicz wrote:
> From: Karol Wachowski <karol.wachowski@intel.com>
> 
> Move the ivpu_mmu_discard_events() function to the common portion of
> the abort work function. This ensures it is called only once, even if
> there are no faulty contexts in context_xa, to guarantee that MMU events
> are discarded and new events are not missed.
> 
> Reviewed-by: Jacek Lawrynowicz <jacek.lawrynowicz@linux.intel.com>
> Signed-off-by: Karol Wachowski <karol.wachowski@intel.com>
> Signed-off-by: Jacek Lawrynowicz <jacek.lawrynowicz@linux.intel.com>

Reviewed-by: Jeffrey Hugo <quic_jhugo@quicinc.com>

Re: [PATCH 4/6] accel/ivpu: Turn on HWS by default on all platforms

[Jeffrey Hugo]

On 1/29/2025 5:56 AM, Jacek Lawrynowicz wrote:
> From: Karol Wachowski <karol.wachowski@intel.com>
> 
> Hardware scheduling (HWS) is supposed to be supported on all existing
> platform with recent FW including pre-silicon ones. Turn on HWS by
> default.

Is there released firmware which does not have this enabled/supported? 
Should this be a "on by default, if FW VER > X"?

-Jeff

Re: [PATCH 5/6] accel/ivpu: Enable recovery and adjust timeouts for fpga

[Jeffrey Hugo]

On 1/29/2025 5:56 AM, Jacek Lawrynowicz wrote:
> From: Tomasz Rusinowicz <tomasz.rusinowicz@intel.com>
> 
> Recovery now works on fpga. JSM state dump timeout needs to
> be really long for the new fpga model releases.
> 
> Enable punit on fpga.
> 
> Reviewed-by: Jacek Lawrynowicz <jacek.lawrynowicz@linux.intel.com>
> Signed-off-by: Tomasz Rusinowicz <tomasz.rusinowicz@intel.com>
> Signed-off-by: Jacek Lawrynowicz <jacek.lawrynowicz@linux.intel.com>

Reviewed-by: Jeffrey Hugo <quic_jhugo@quicinc.com>

Re: [PATCH 6/6] accel/ivpu: Move recovery work to system_unbound_wq

[Jeffrey Hugo]

On 1/29/2025 5:56 AM, Jacek Lawrynowicz wrote:
> From: Karol Wachowski <karol.wachowski@intel.com>
> 
> Recovery work doesn't need to be bound to any specific CPU, so move it
> to unbound workqueue to improve execution time and system latency.
> 
> Reviewed-by: Jacek Lawrynowicz <jacek.lawrynowicz@linux.intel.com>
> Signed-off-by: Karol Wachowski <karol.wachowski@intel.com>
> Signed-off-by: Jacek Lawrynowicz <jacek.lawrynowicz@linux.intel.com>

Reviewed-by: Jeffrey Hugo <quic_jhugo@quicinc.com>

Re: [PATCH 4/6] accel/ivpu: Turn on HWS by default on all platforms

[Jacek Lawrynowicz]

Yes and there is a check for FW version in ivpu_fw_sched_mode_select() that verifies this.
It is just above the changed lines from this patch.

On 1/31/2025 7:47 PM, Jeffrey Hugo wrote:
> On 1/29/2025 5:56 AM, Jacek Lawrynowicz wrote:
>> From: Karol Wachowski <karol.wachowski@intel.com>
>>
>> Hardware scheduling (HWS) is supposed to be supported on all existing
>> platform with recent FW including pre-silicon ones. Turn on HWS by
>> default.
> 
> Is there released firmware which does not have this enabled/supported? Should this be a "on by default, if FW VER > X"?
> 
> -Jeff

Re: [PATCH 1/6] accel/ivpu: Add support for hardware fault injection

[Jacek Lawrynowicz]

On 1/31/2025 7:41 PM, Jeffrey Hugo wrote:
> On 1/29/2025 5:56 AM, Jacek Lawrynowicz wrote:
>> This commit introduces the capability to simulate hardware faults for
> 
> Nit - "This commit" is redundant.

Sure, removed.

Re: [PATCH 0/6] accel/ivpu: Changes for 6.15

[Jacek Lawrynowicz]

Applied to drm-misc-next

On 1/29/2025 1:56 PM, Jacek Lawrynowicz wrote:
> Most notable is the addition of hardware fault injection support which allows
> to test error handling paths in the driver.
> 
> Jacek Lawrynowicz (2):
>   accel/ivpu: Add support for hardware fault injection
>   accel/ivpu: Update last_busy in IRQ handler
> 
> Karol Wachowski (3):
>   accel/ivpu: Fix missing MMU events if file_priv is unbound
>   accel/ivpu: Turn on HWS by default on all platforms
>   accel/ivpu: Move recovery work to system_unbound_wq
> 
> Tomasz Rusinowicz (1):
>   accel/ivpu: Enable recovery and adjust timeouts for fpga
> 
>  drivers/accel/ivpu/ivpu_debugfs.c   |  5 +++
>  drivers/accel/ivpu/ivpu_fw.c        | 10 +----
>  drivers/accel/ivpu/ivpu_hw.c        | 33 +++++++++++----
>  drivers/accel/ivpu/ivpu_hw_ip.c     |  4 +-
>  drivers/accel/ivpu/ivpu_hw_reg_io.h | 64 +++++++++++++++++------------
>  drivers/accel/ivpu/ivpu_job.c       |  8 +++-
>  drivers/accel/ivpu/ivpu_pm.c        |  7 +---
>  7 files changed, 78 insertions(+), 53 deletions(-)
> 
> --
> 2.45.1