6.18.14-rc1 review

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Sebastian Andrzej Siewior <[email protected]>

[ Upstream commit f6bc712877f24dc89bdfd7bdbf1a32f3b9960b34 ]

Passing IRQF_ONESHOT ensures that the interrupt source is masked until the secondary (threaded) handler is done. If only a primary handler is used then the flag makes no sense because the interrupt cannot fire (again) while its handler is running.

The flag also prevents force-threading of the primary handler and the irq-core will warn about this.

The flag was added to match the flag on the shared handler which uses a threaded handler and therefore IRQF_ONESHOT. This is no longer needed because devm_request_irq() now passes IRQF_COND_ONESHOT for this case.

Revert adding IRQF_ONESHOT to irqflags.

---
 drivers/platform/x86/intel/int0002_vgpio.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

index 6f5629dc3f8db..562e880256436 100644
@@ -206,8 +206,8 @@ static int int0002_probe(struct platform_device *pdev)
 	 * FIXME: augment this if we managed to pull handling of shared
 	 * IRQs into gpiolib.
 	 */
-	ret = devm_request_irq(dev, irq, int0002_irq,
-			       IRQF_ONESHOT | IRQF_SHARED, "INT0002", chip);
+	ret = devm_request_irq(dev, irq, int0002_irq, IRQF_SHARED, "INT0002",
+			       chip);
 	if (ret) {
 		dev_err(dev, "Error requesting IRQ %d: %d\n", irq, ret);
 		return ret;

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Sebastian Andrzej Siewior <[email protected]>

[ Upstream commit 5bfcdccb4d18d3909b7f87942be67fd6bdc00c1d ]

request_threaded_irq() is invoked with a primary and a secondary handler and no flags are passed. The primary handler is the same as irq_default_primary_handler() so there is no need to have an identical copy.

The lack of the IRQF_ONESHOT can be dangerous because the interrupt source is not masked while the threaded handler is active. This means, especially on LEVEL typed interrupt lines, the interrupt can fire again before the threaded handler had a chance to run.

Use the default primary interrupt handler by specifying NULL and set IRQF_ONESHOT so the interrupt source is masked until the secondary handler is done.

---
 drivers/iommu/amd/amd_iommu.h |  1 -
 drivers/iommu/amd/init.c      | 12 ++++--------
 drivers/iommu/amd/iommu.c     |  5 -----
 3 files changed, 4 insertions(+), 14 deletions(-)

index 9b4b589a54b57..bf77fdf5529f4 100644
@@ -15,7 +15,6 @@ irqreturn_t amd_iommu_int_thread(int irq, void *data);
 irqreturn_t amd_iommu_int_thread_evtlog(int irq, void *data);
 irqreturn_t amd_iommu_int_thread_pprlog(int irq, void *data);
 irqreturn_t amd_iommu_int_thread_galog(int irq, void *data);
-irqreturn_t amd_iommu_int_handler(int irq, void *data);
 void amd_iommu_restart_log(struct amd_iommu *iommu, const char *evt_type,
 			   u8 cntrl_intr, u8 cntrl_log,
 			   u32 status_run_mask, u32 status_overflow_mask);

index 034edce816d01..53afb1cb0a6fc 100644
@@ -2355,12 +2355,8 @@ static int iommu_setup_msi(struct amd_iommu *iommu)
 	if (r)
 		return r;
 
-	r = request_threaded_irq(iommu->dev->irq,
-				 amd_iommu_int_handler,
-				 amd_iommu_int_thread,
-				 0, "AMD-Vi",
-				 iommu);
-
+	r = request_threaded_irq(iommu->dev->irq, NULL, amd_iommu_int_thread,
+				 IRQF_ONESHOT, "AMD-Vi", iommu);
 	if (r) {
 		pci_disable_msi(iommu->dev);
 		return r;

@@ -2534,8 +2530,8 @@ static int __iommu_setup_intcapxt(struct amd_iommu *iommu, const char *devname,
 		return irq;
 	}
 
-	ret = request_threaded_irq(irq, amd_iommu_int_handler,
-				   thread_fn, 0, devname, iommu);
+	ret = request_threaded_irq(irq, NULL, thread_fn, IRQF_ONESHOT, devname,
+				   iommu);
 	if (ret) {
 		irq_domain_free_irqs(irq, 1);
 		irq_domain_remove(domain);

index 5914bef0c8c19..30dd482fe0953 100644
@@ -1146,11 +1146,6 @@ irqreturn_t amd_iommu_int_thread(int irq, void *data)
 	return IRQ_HANDLED;
 }
 
-irqreturn_t amd_iommu_int_handler(int irq, void *data)
-{
-	return IRQ_WAKE_THREAD;
-}
-
 /****************************************************************************
  *
  * IOMMU command queuing functions

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Sebastian Andrzej Siewior <[email protected]>

[ Upstream commit 28abed6569c87eab9071ab56c64433c2f0d9ce51 ]

There is no added value in btintel_pcie_msix_isr() compared to irq_default_primary_handler().

Using a threaded interrupt without a dedicated primary handler mandates the IRQF_ONESHOT flag to mask the interrupt source while the threaded handler is active. Otherwise the interrupt can fire again before the threaded handler had a chance to run.

Use the default primary interrupt handler by specifying NULL and set IRQF_ONESHOT so the interrupt source is masked until the secondary handler is done.

---
 drivers/bluetooth/btintel_pcie.c | 9 ++-------
 1 file changed, 2 insertions(+), 7 deletions(-)

index f280bcc61bbfb..c68a8de3025b0 100644
@@ -1430,11 +1430,6 @@ static void btintel_pcie_msix_rx_handle(struct btintel_pcie_data *data)
 	}
 }
 
-static irqreturn_t btintel_pcie_msix_isr(int irq, void *data)
-{
-	return IRQ_WAKE_THREAD;
-}
-
 static inline bool btintel_pcie_is_rxq_empty(struct btintel_pcie_data *data)
 {
 	return data->ia.cr_hia[BTINTEL_PCIE_RXQ_NUM] == data->ia.cr_tia[BTINTEL_PCIE_RXQ_NUM];

@@ -1536,9 +1531,9 @@ static int btintel_pcie_setup_irq(struct btintel_pcie_data *data)
 
 		err = devm_request_threaded_irq(&data->pdev->dev,
 						msix_entry->vector,
-						btintel_pcie_msix_isr,
+						NULL,
 						btintel_pcie_irq_msix_handler,
-						IRQF_SHARED,
+						IRQF_ONESHOT | IRQF_SHARED,
 						KBUILD_MODNAME,
 						msix_entry);
 		if (err) {

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Sebastian Andrzej Siewior <[email protected]>

[ Upstream commit bd81f07e9a27c341cd7e72be95eb0b7cf3910926 ]

There is no added value in efct_intr_msix() compared to irq_default_primary_handler().

Using a threaded interrupt without a dedicated primary handler mandates the IRQF_ONESHOT flag to mask the interrupt source while the threaded handler is active. Otherwise the interrupt can fire again before the threaded handler had a chance to run.

Use the default primary interrupt handler by specifying NULL and set IRQF_ONESHOT so the interrupt source is masked until the secondary handler is done.

---
 drivers/scsi/elx/efct/efct_driver.c | 8 +-------
 1 file changed, 1 insertion(+), 7 deletions(-)

index 1bd42f7db1773..528399f725d42 100644
@@ -415,12 +415,6 @@ efct_intr_thread(int irq, void *handle)
 	return IRQ_HANDLED;
 }
 
-static irqreturn_t
-efct_intr_msix(int irq, void *handle)
-{
-	return IRQ_WAKE_THREAD;
-}
-
 static int
 efct_setup_msix(struct efct *efct, u32 num_intrs)
 {

@@ -450,7 +444,7 @@ efct_setup_msix(struct efct *efct, u32 num_intrs)
 		intr_ctx->index = i;
 
 		rc = request_threaded_irq(pci_irq_vector(efct->pci, i),
-					  efct_intr_msix, efct_intr_thread, 0,
+					  NULL, efct_intr_thread, IRQF_ONESHOT,
 					  EFCT_DRIVER_NAME, intr_ctx);
 		if (rc) {
 			dev_err(&efct->pci->dev,

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Sebastian Andrzej Siewior <[email protected]>

[ Upstream commit 5c858d6c66304b4c7579582ec5235f02d43578ea ]

Passing IRQF_ONESHOT ensures that the interrupt source is masked until the secondary (threaded) handler is done. If only a primary handler is used then the flag makes no sense because the interrupt can not fire (again) while its handler is running.

The flag also prevents force-threading of the primary handler and the irq-core will warn about this.

Remove IRQF_ONESHOT from irqflags.

---
 drivers/edac/altera_edac.c | 11 ++++-------
 1 file changed, 4 insertions(+), 7 deletions(-)

index 0c5b94e64ea15..4edd2088c2db6 100644
@@ -1563,8 +1563,7 @@ static int altr_portb_setup(struct altr_edac_device_dev *device)
 		goto err_release_group_1;
 	}
 	rc = devm_request_irq(&altdev->ddev, altdev->sb_irq,
-			      prv->ecc_irq_handler,
-			      IRQF_ONESHOT | IRQF_TRIGGER_HIGH,
+			      prv->ecc_irq_handler, IRQF_TRIGGER_HIGH,
 			      ecc_name, altdev);
 	if (rc) {
 		edac_printk(KERN_ERR, EDAC_DEVICE, "PortB SBERR IRQ error\n");

@@ -1587,8 +1586,7 @@ static int altr_portb_setup(struct altr_edac_device_dev *device)
 		goto err_release_group_1;
 	}
 	rc = devm_request_irq(&altdev->ddev, altdev->db_irq,
-			      prv->ecc_irq_handler,
-			      IRQF_ONESHOT | IRQF_TRIGGER_HIGH,
+			      prv->ecc_irq_handler, IRQF_TRIGGER_HIGH,
 			      ecc_name, altdev);
 	if (rc) {
 		edac_printk(KERN_ERR, EDAC_DEVICE, "PortB DBERR IRQ error\n");

@@ -1970,8 +1968,7 @@ static int altr_edac_a10_device_add(struct altr_arria10_edac *edac,
 		goto err_release_group1;
 	}
 	rc = devm_request_irq(edac->dev, altdev->sb_irq, prv->ecc_irq_handler,
-			      IRQF_ONESHOT | IRQF_TRIGGER_HIGH,
-			      ecc_name, altdev);
+			      IRQF_TRIGGER_HIGH, ecc_name, altdev);
 	if (rc) {
 		edac_printk(KERN_ERR, EDAC_DEVICE, "No SBERR IRQ resource\n");
 		goto err_release_group1;

@@ -1993,7 +1990,7 @@ static int altr_edac_a10_device_add(struct altr_arria10_edac *edac,
 		goto err_release_group1;
 	}
 	rc = devm_request_irq(edac->dev, altdev->db_irq, prv->ecc_irq_handler,
-			      IRQF_ONESHOT | IRQF_TRIGGER_HIGH,
+			      IRQF_TRIGGER_HIGH,
 			      ecc_name, altdev);
 	if (rc) {
 		edac_printk(KERN_ERR, EDAC_DEVICE, "No DBERR IRQ resource\n");

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Sebastian Andrzej Siewior <[email protected]>

[ Upstream commit a7fb84ea70aae9a92a842932206e70ed1b3c7007 ]

Passing IRQF_ONESHOT ensures that the interrupt source is masked until the secondary (threaded) handler is done. If only a primary handler is used then the flag makes no sense because the interrupt can not fire (again) while its handler is running.

The flag also prevents force-threading of the primary handler and the irq-core will warn about this.

Remove IRQF_ONESHOT from irqflags.

---
 drivers/usb/typec/tcpm/fusb302.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

index 870a71f953f6c..19ff8217818e7 100644
@@ -1756,8 +1756,7 @@ static int fusb302_probe(struct i2c_client *client)
 	}
 
 	ret = request_irq(chip->gpio_int_n_irq, fusb302_irq_intn,
-			  IRQF_ONESHOT | IRQF_TRIGGER_LOW,
-			  "fsc_interrupt_int_n", chip);
+			  IRQF_TRIGGER_LOW, "fsc_interrupt_int_n", chip);
 	if (ret < 0) {
 		dev_err(dev, "cannot request IRQ for GPIO Int_N, ret=%d", ret);
 		goto tcpm_unregister_port;

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Sebastian Andrzej Siewior <[email protected]>

[ Upstream commit 18d28446231390e4ea3634fb16200865df2c6506 ]

Passing IRQF_ONESHOT ensures that the interrupt source is masked until the secondary (threaded) handler is done. If only a primary handler is used then the flag makes no sense because the interrupt can not fire (again) while its handler is running.

The flag also prevents force-threading of the primary handler and the irq-core will warn about this.

Remove IRQF_ONESHOT from irqflags.

---
 drivers/rtc/rtc-amlogic-a4.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

index a993d35e1d6b0..d766055d95848 100644
@@ -371,7 +371,7 @@ static int aml_rtc_probe(struct platform_device *pdev)
 	}
 
 	ret = devm_request_irq(dev, rtc->irq, aml_rtc_handler,
-			       IRQF_ONESHOT, "aml-rtc alarm", rtc);
+			       0, "aml-rtc alarm", rtc);
 	if (ret) {
 		dev_err_probe(dev, ret, "IRQ%d request failed, ret = %d\n",
 			      rtc->irq, ret);

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Sebastian Andrzej Siewior <[email protected]>

[ Upstream commit 553b4999cbe231b5011cb8db05a3092dec168aca ]

Using a threaded interrupt without a dedicated primary handler mandates the IRQF_ONESHOT flag to mask the interrupt source while the threaded handler is active. Otherwise the interrupt can fire again before the threaded handler had a chance to run.

Mark explained that this should not happen with this hardware since it is a slow irqchip which is behind an I2C/ SPI bus but the IRQ-core will refuse to accept such a handler.

Set IRQF_ONESHOT so the interrupt source is masked until the secondary handler is done.

---
 include/linux/mfd/wm8350/core.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

index 5f70d3b5d1b1a..097ef4dfcdac8 100644
@@ -667,7 +667,7 @@ static inline int wm8350_register_irq(struct wm8350 *wm8350, int irq,
 		return -ENODEV;
 
 	return request_threaded_irq(irq + wm8350->irq_base, NULL,
-				    handler, flags, name, data);
+				    handler, flags | IRQF_ONESHOT, name, data);
 }
 
 static inline void wm8350_free_irq(struct wm8350 *wm8350, int irq, void *data)

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Sebastian Andrzej Siewior <[email protected]>

[ Upstream commit ef92b98f5f6758a049898b53aa30476010db04fa ]

The interrupt handler iio_trigger_generic_data_rdy_poll() will invoke other interrupt handlers and this supposed to happen from hard interrupt context.

Use IRQF_NO_THREAD to forbid forced-threading.

---
 drivers/media/pci/mgb4/mgb4_trigger.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

index d7dddc5c8728e..10c23f0c833d5 100644
@@ -114,7 +114,7 @@ static int probe_trigger(struct iio_dev *indio_dev, int irq)
 	if (!st->trig)
 		return -ENOMEM;
 
-	ret = request_irq(irq, &iio_trigger_generic_data_rdy_poll, 0,
+	ret = request_irq(irq, &iio_trigger_generic_data_rdy_poll, IRQF_NO_THREAD,
 			  "mgb4-trigger", st->trig);
 	if (ret)
 		goto error_free_trig;

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Joel Fernandes <[email protected]>

[ Upstream commit 3cb3b27693bf30defb16aa096158a3b24583b8d2 ]

The defer params were not cleared in __dl_clear_params. Clear them.

Without this is some of my test cases are flaking and the DL timer is not starting correctly AFAICS.

---
 kernel/sched/deadline.c | 3 +++
 1 file changed, 3 insertions(+)

index c7a8717e837dd..72499cf2a1db5 100644
@@ -3591,6 +3591,9 @@ static void __dl_clear_params(struct sched_dl_entity *dl_se)
 	dl_se->dl_non_contending	= 0;
 	dl_se->dl_overrun		= 0;
 	dl_se->dl_server		= 0;
+	dl_se->dl_defer			= 0;
+	dl_se->dl_defer_running		= 0;
+	dl_se->dl_defer_armed		= 0;
 
 #ifdef CONFIG_RT_MUTEXES
 	dl_se->pi_se			= dl_se;

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Chen Jinghuang <[email protected]>

[ Upstream commit 94894c9c477e53bcea052e075c53f89df3d2a33e ]

CPU0 becomes overloaded when hosting a CPU-bound RT task, a non-CPU-bound RT task, and a CFS task stuck in kernel space. When other CPUs switch from RT to non-RT tasks, RT load balancing (LB) is triggered; with HAVE_RT_PUSH_IPI enabled, they send IPIs to CPU0 to drive the execution of rto_push_irq_work_func. During push_rt_task on CPU0,
if next_task->prio < rq->donor->prio, resched_curr() sets NEED_RESCHED and after the push operation completes, CPU0 calls rto_next_cpu(). Since only CPU0 is overloaded in this scenario, rto_next_cpu() should ideally return -1 (no further IPI needed).

However, multiple CPUs invoking tell_cpu_to_push() during LB increments rd->rto_loop_next. Even when rd->rto_cpu is set to -1, the mismatch between rd->rto_loop and rd->rto_loop_next forces rto_next_cpu() to restart its search from -1. With CPU0 remaining overloaded (satisfying rt_nr_migratory && rt_nr_total > 1), it gets reselected, causing CPU0 to queue irq_work to itself and send self-IPIs repeatedly. As long as CPU0 stays overloaded and other CPUs run pull_rt_tasks(), it falls into an infinite self-IPI loop, which triggers a CPU hardlockup due to continuous self-interrupts.

The trigging scenario is as follows:

cpu0 cpu1 cpu2
pull_rt_task
tell_cpu_to_push
<------------irq_work_queue_on
rto_push_irq_work_func

       push_rt_task
    resched_curr(rq)                                   pull_rt_task
    rto_next_cpu                                     tell_cpu_to_push
                      <-------------------------- atomic_inc(rto_loop_next)

rd->rto_loop != next
rto_next_cpu
irq_work_queue_on
rto_push_irq_work_func

Fix redundant self-IPI by filtering the initiating CPU in rto_next_cpu(). This solution has been verified to effectively eliminate spurious self-IPIs and prevent CPU hardlockup scenarios.

---
 kernel/sched/rt.c | 5 +++++
 1 file changed, 5 insertions(+)

index fb07dcfc60a24..d4d994fb8999a 100644
@@ -2100,6 +2100,7 @@ static void push_rt_tasks(struct rq *rq)
  */
 static int rto_next_cpu(struct root_domain *rd)
 {
+	int this_cpu = smp_processor_id();
 	int next;
 	int cpu;
 

@@ -2123,6 +2124,10 @@ static int rto_next_cpu(struct root_domain *rd)
 
 		rd->rto_cpu = cpu;
 
+		/* Do not send IPI to self */
+		if (cpu == this_cpu)
+			continue;
+
 		if (cpu < nr_cpu_ids)
 			return cpu;
 

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Titouan Ameline de Cadeville <[email protected]>

[ Upstream commit 46a03ea50b5f380bdb99178b8f90b39c6ba1f528 ]

Remove duplicate entries from the bprm_stack_limits KUnit test vector table. The duplicates do not add coverage and only increase test size.

---
 fs/tests/exec_kunit.c | 6 ------
 1 file changed, 6 deletions(-)

index 7c77d039680bb..f412d1a0f6bba 100644
@@ -87,9 +87,6 @@ static const struct bprm_stack_limits_result bprm_stack_limits_results[] = {
 	    .argc = 0, .envc = ARG_MAX / sizeof(void *) - 1 },
 	  .expected_argmin = ULONG_MAX - sizeof(void *) },
 	/* Raising rlim_stack / 4 to _STK_LIM / 4 * 3 will see more space. */
-	{ { .p = ULONG_MAX, .rlim_stack.rlim_cur = 4 * (_STK_LIM / 4 * 3),
-	    .argc = 0, .envc = 0 },
-	  .expected_argmin = ULONG_MAX - (_STK_LIM / 4 * 3) + sizeof(void *) },
 	{ { .p = ULONG_MAX, .rlim_stack.rlim_cur = 4 * (_STK_LIM / 4 * 3),
 	    .argc = 0, .envc = 0 },
 	  .expected_argmin = ULONG_MAX - (_STK_LIM / 4 * 3) + sizeof(void *) },

@@ -103,9 +100,6 @@ static const struct bprm_stack_limits_result bprm_stack_limits_results[] = {
 	{ { .p = ULONG_MAX, .rlim_stack.rlim_cur = 4 * _STK_LIM,
 	    .argc = 0, .envc = 0 },
 	  .expected_argmin = ULONG_MAX - (_STK_LIM / 4 * 3) + sizeof(void *) },
-	{ { .p = ULONG_MAX, .rlim_stack.rlim_cur = 4 * _STK_LIM,
-	    .argc = 0, .envc = 0 },
-	  .expected_argmin = ULONG_MAX - (_STK_LIM / 4 * 3) + sizeof(void *) },
 };
 
 static void exec_test_bprm_stack_limits(struct kunit *test)

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: KP Singh <[email protected]>

[ Upstream commit ea1535e28bb3773fc0b3cbd1f3842b808016990c ]

Practical BPF signatures are significantly smaller than KMALLOC_MAX_CACHE_SIZE

Allowing larger sizes opens the door for abuse by passing excessive size values and forcing the kernel into expensive allocation paths (via kmalloc_large or vmalloc).

---
 kernel/bpf/syscall.c | 7 +++++++
 1 file changed, 7 insertions(+)

index f39367765f0c4..2649e0472dfe0 100644
@@ -2825,6 +2825,13 @@ static int bpf_prog_verify_signature(struct bpf_prog *prog, union bpf_attr *attr
 	void *sig;
 	int err = 0;
 
+	/*
+	 * Don't attempt to use kmalloc_large or vmalloc for signatures.
+	 * Practical signature for BPF program should be below this limit.
+	 */
+	if (attr->signature_size > KMALLOC_MAX_CACHE_SIZE)
+		return -EINVAL;
+
 	if (system_keyring_id_check(attr->keyring_id) == 0)
 		key = bpf_lookup_system_key(attr->keyring_id);
 	else

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: KP Singh <[email protected]>

[ Upstream commit a2c86aa621c22f2a7e26c654f936d65cfff0aa91 ]

Currently, bpf_map_get_info_by_fd calculates and caches the hash of the map regardless of the map's frozen state.

This leads to a TOCTOU bug where userspace can call BPF_OBJ_GET_INFO_BY_FD to cache the hash and then modify the map contents before freezing.

Therefore, a trusted loader can be tricked into verifying the stale hash while loading the modified contents.

Fix this by returning -EPERM if the map is not frozen when the hash is requested. This ensures the hash is only generated for the final, immutable state of the map.

---
 kernel/bpf/syscall.c | 3 +++
 1 file changed, 3 insertions(+)

index 2649e0472dfe0..586ece78f783a 100644
@@ -5303,6 +5303,9 @@ static int bpf_map_get_info_by_fd(struct file *file,
 		if (info.hash_size != SHA256_DIGEST_SIZE)
 			return -EINVAL;
 
+		if (!READ_ONCE(map->frozen))
+			return -EPERM;
+
 		err = map->ops->map_get_hash(map, SHA256_DIGEST_SIZE, map->sha);
 		if (err != 0)
 			return err;

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Zilin Guan <[email protected]>

[ Upstream commit ccb679fdae2e62ed92fd9acb25ed809c0226fcc6 ]

The starfive_aes_aead_do_one_req() function allocates rctx->adata with kzalloc() but fails to free it if sg_copy_to_buffer() or starfive_aes_hw_init() fails, which lead to memory leaks.

Since rctx->adata is unconditionally freed after the write_adata operations, ensure consistent cleanup by freeing the allocation in these earlier error paths as well.

Compile tested only. Issue found using a prototype static analysis tool and code review.

---
 drivers/crypto/starfive/jh7110-aes.c | 9 +++++++--
 1 file changed, 7 insertions(+), 2 deletions(-)

index 426b24889af85..01195664cc7cd 100644
@@ -669,8 +669,10 @@ static int starfive_aes_aead_do_one_req(struct crypto_engine *engine, void *areq
 			return -ENOMEM;
 
 		if (sg_copy_to_buffer(req->src, sg_nents_for_len(req->src, cryp->assoclen),
-				      rctx->adata, cryp->assoclen) != cryp->assoclen)
+				      rctx->adata, cryp->assoclen) != cryp->assoclen) {
+			kfree(rctx->adata);
 			return -EINVAL;
+		}
 	}
 
 	if (cryp->total_in)

@@ -681,8 +683,11 @@ static int starfive_aes_aead_do_one_req(struct crypto_engine *engine, void *areq
 	ctx->rctx = rctx;
 
 	ret = starfive_aes_hw_init(ctx);
-	if (ret)
+	if (ret) {
+		if (cryp->assoclen)
+			kfree(rctx->adata);
 		return ret;
+	}
 
 	if (!cryp->assoclen)
 		goto write_text;

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Jonathan McDowell <[email protected]>

[ Upstream commit e74b96d77da9eb5ee1b603c937c2adab5134a04b ]

The HW RNG core allows for manual selection of which RNG device to use, but does not allow for no device to be enabled. It may be desirable to do this on systems with only a single suitable hardware RNG, where we need exclusive access to other functionality on this device. In particular when performing TPM firmware upgrades this lets us ensure the kernel does not try to access the device.

Before:

root@debian-qemu-efi:~# grep "" /sys/devices/virtual/misc/hw_random/rng_*
/sys/devices/virtual/misc/hw_random/rng_available:tpm-rng-0
/sys/devices/virtual/misc/hw_random/rng_current:tpm-rng-0
/sys/devices/virtual/misc/hw_random/rng_quality:1024
/sys/devices/virtual/misc/hw_random/rng_selected:0

After:

root@debian-qemu-efi:~# grep "" /sys/devices/virtual/misc/hw_random/rng_*
/sys/devices/virtual/misc/hw_random/rng_available:tpm-rng-0 none
/sys/devices/virtual/misc/hw_random/rng_current:tpm-rng-0
/sys/devices/virtual/misc/hw_random/rng_quality:1024
/sys/devices/virtual/misc/hw_random/rng_selected:0

root@debian-qemu-efi:~# echo none > /sys/devices/virtual/misc/hw_random/rng_current
root@debian-qemu-efi:~# grep "" /sys/devices/virtual/misc/hw_random/rng_*
/sys/devices/virtual/misc/hw_random/rng_available:tpm-rng-0 none
/sys/devices/virtual/misc/hw_random/rng_current:none
grep: /sys/devices/virtual/misc/hw_random/rng_quality: No such device
/sys/devices/virtual/misc/hw_random/rng_selected:1

(Observe using bpftrace no calls to TPM being made)

root@debian-qemu-efi:~# echo "" > /sys/devices/virtual/misc/hw_random/rng_current
root@debian-qemu-efi:~# grep "" /sys/devices/virtual/misc/hw_random/rng_*
/sys/devices/virtual/misc/hw_random/rng_available:tpm-rng-0 none
/sys/devices/virtual/misc/hw_random/rng_current:tpm-rng-0
/sys/devices/virtual/misc/hw_random/rng_quality:1024
/sys/devices/virtual/misc/hw_random/rng_selected:0

(Observe using bpftrace that calls to the TPM resume)

Stable-dep-of: cc2f39d6ac48 ("hwrng: core - use RCU and work_struct to fix race condition")

---
 drivers/char/hw_random/core.c | 9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)

index 018316f546215..56d888bebe0c0 100644
@@ -341,6 +341,9 @@ static ssize_t rng_current_store(struct device *dev,
 
 	if (sysfs_streq(buf, "")) {
 		err = enable_best_rng();
+	} else if (sysfs_streq(buf, "none")) {
+		cur_rng_set_by_user = 1;
+		drop_current_rng();
 	} else {
 		list_for_each_entry(rng, &rng_list, list) {
 			if (sysfs_streq(rng->name, buf)) {

@@ -392,7 +395,7 @@ static ssize_t rng_available_show(struct device *dev,
 		strlcat(buf, rng->name, PAGE_SIZE);
 		strlcat(buf, " ", PAGE_SIZE);
 	}
-	strlcat(buf, "\n", PAGE_SIZE);
+	strlcat(buf, "none\n", PAGE_SIZE);
 	mutex_unlock(&rng_mutex);
 
 	return strlen(buf);

@@ -544,8 +547,8 @@ int hwrng_register(struct hwrng *rng)
 	/* Adjust quality field to always have a proper value */
 	rng->quality = min_t(u16, min_t(u16, default_quality, 1024), rng->quality ?: 1024);
 
-	if (!current_rng ||
-	    (!cur_rng_set_by_user && rng->quality > current_rng->quality)) {
+	if (!cur_rng_set_by_user &&
+	    (!current_rng || rng->quality > current_rng->quality)) {
 		/*
 		 * Set new rng as current as the new rng source
 		 * provides better entropy quality and was not

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Lianjie Wang <[email protected]>

[ Upstream commit cc2f39d6ac48e6e3cb2d6240bc0d6df839dd0828 ]

Currently, hwrng_fill is not cleared until the hwrng_fillfn() thread exits. Since hwrng_unregister() reads hwrng_fill outside the rng_mutex lock, a concurrent hwrng_unregister() may call kthread_stop() again on the same task.

Additionally, if hwrng_unregister() is called immediately after hwrng_register(), the stopped thread may have never been executed. Thus, hwrng_fill remains dirty even after hwrng_unregister() returns. In this case, subsequent calls to hwrng_register() will fail to start new threads, and hwrng_unregister() will call kthread_stop() on the same freed task. In both cases, a use-after-free occurs:

refcount_t: addition on 0; use-after-free.
WARNING: ... at lib/refcount.c:25 refcount_warn_saturate+0xec/0x1c0

Call Trace:
 kthread_stop+0x181/0x360
 hwrng_unregister+0x288/0x380
 virtrng_remove+0xe3/0x200

This patch fixes the race by protecting the global hwrng_fill pointer inside the rng_mutex lock, so that hwrng_fillfn() thread is stopped only once, and calls to kthread_run() and kthread_stop() are serialized with the lock held.

To avoid deadlock in hwrng_fillfn() while being stopped with the lock held, we convert current_rng to RCU, so that get_current_rng() can read current_rng without holding the lock. To remove the lock from put_rng(), we also delay the actual cleanup into a work_struct.

Since get_current_rng() no longer returns ERR_PTR values, the IS_ERR() checks are removed from its callers.

With hwrng_fill protected by the rng_mutex lock, hwrng_fillfn() can no longer clear hwrng_fill itself. Therefore, if hwrng_fillfn() returns directly after current_rng is dropped, kthread_stop() would be called on a freed task_struct later. To fix this, hwrng_fillfn() calls schedule() now to keep the task alive until being stopped. The kthread_stop() call is also moved from hwrng_unregister() to drop_current_rng(), ensuring kthread_stop() is called on all possible paths where current_rng becomes NULL, so that the thread would not wait forever.

---
 drivers/char/hw_random/core.c | 168 +++++++++++++++++++++-------------
 include/linux/hw_random.h     |   2 +
 2 files changed, 107 insertions(+), 63 deletions(-)

index 56d888bebe0c0..036de7294bbda 100644
@@ -20,23 +20,25 @@
 #include <linux/miscdevice.h>
 #include <linux/module.h>
 #include <linux/random.h>
+#include <linux/rcupdate.h>
 #include <linux/sched.h>
 #include <linux/sched/signal.h>
 #include <linux/slab.h>
 #include <linux/string.h>
 #include <linux/uaccess.h>
+#include <linux/workqueue.h>
 
 #define RNG_MODULE_NAME		"hw_random"
 
 #define RNG_BUFFER_SIZE (SMP_CACHE_BYTES < 32 ? 32 : SMP_CACHE_BYTES)
 
-static struct hwrng *current_rng;
+static struct hwrng __rcu *current_rng;
 /* the current rng has been explicitly chosen by user via sysfs */
 static int cur_rng_set_by_user;
 static struct task_struct *hwrng_fill;
 /* list of registered rngs */
 static LIST_HEAD(rng_list);
-/* Protects rng_list and current_rng */
+/* Protects rng_list, hwrng_fill and updating on current_rng */
 static DEFINE_MUTEX(rng_mutex);
 /* Protects rng read functions, data_avail, rng_buffer and rng_fillbuf */
 static DEFINE_MUTEX(reading_mutex);

@@ -64,18 +66,39 @@ static size_t rng_buffer_size(void)
 	return RNG_BUFFER_SIZE;
 }
 
-static inline void cleanup_rng(struct kref *kref)
+static void cleanup_rng_work(struct work_struct *work)
 {
-	struct hwrng *rng = container_of(kref, struct hwrng, ref);
+	struct hwrng *rng = container_of(work, struct hwrng, cleanup_work);
+
+	/*
+	 * Hold rng_mutex here so we serialize in case they set_current_rng
+	 * on rng again immediately.
+	 */
+	mutex_lock(&rng_mutex);
+
+	/* Skip if rng has been reinitialized. */
+	if (kref_read(&rng->ref)) {
+		mutex_unlock(&rng_mutex);
+		return;
+	}
 
 	if (rng->cleanup)
 		rng->cleanup(rng);
 
 	complete(&rng->cleanup_done);
+	mutex_unlock(&rng_mutex);
+}
+
+static inline void cleanup_rng(struct kref *kref)
+{
+	struct hwrng *rng = container_of(kref, struct hwrng, ref);
+
+	schedule_work(&rng->cleanup_work);
 }
 
 static int set_current_rng(struct hwrng *rng)
 {
+	struct hwrng *old_rng;
 	int err;
 
 	BUG_ON(!mutex_is_locked(&rng_mutex));

@@ -84,8 +107,14 @@ static int set_current_rng(struct hwrng *rng)
 	if (err)
 		return err;
 
-	drop_current_rng();
-	current_rng = rng;
+	old_rng = rcu_dereference_protected(current_rng,
+					    lockdep_is_held(&rng_mutex));
+	rcu_assign_pointer(current_rng, rng);
+
+	if (old_rng) {
+		synchronize_rcu();
+		kref_put(&old_rng->ref, cleanup_rng);
+	}
 
 	/* if necessary, start hwrng thread */
 	if (!hwrng_fill) {

@@ -101,47 +130,56 @@ static int set_current_rng(struct hwrng *rng)
 
 static void drop_current_rng(void)
 {
-	BUG_ON(!mutex_is_locked(&rng_mutex));
-	if (!current_rng)
+	struct hwrng *rng;
+
+	rng = rcu_dereference_protected(current_rng,
+					lockdep_is_held(&rng_mutex));
+	if (!rng)
 		return;
 
+	RCU_INIT_POINTER(current_rng, NULL);
+	synchronize_rcu();
+
+	if (hwrng_fill) {
+		kthread_stop(hwrng_fill);
+		hwrng_fill = NULL;
+	}
+
 	/* decrease last reference for triggering the cleanup */
-	kref_put(&current_rng->ref, cleanup_rng);
-	current_rng = NULL;
+	kref_put(&rng->ref, cleanup_rng);
 }
 
-/* Returns ERR_PTR(), NULL or refcounted hwrng */
+/* Returns NULL or refcounted hwrng */
 static struct hwrng *get_current_rng_nolock(void)
 {
-	if (current_rng)
-		kref_get(&current_rng->ref);
+	struct hwrng *rng;
+
+	rng = rcu_dereference_protected(current_rng,
+					lockdep_is_held(&rng_mutex));
+	if (rng)
+		kref_get(&rng->ref);
 
-	return current_rng;
+	return rng;
 }
 
 static struct hwrng *get_current_rng(void)
 {
 	struct hwrng *rng;
 
-	if (mutex_lock_interruptible(&rng_mutex))
-		return ERR_PTR(-ERESTARTSYS);
+	rcu_read_lock();
+	rng = rcu_dereference(current_rng);
+	if (rng)
+		kref_get(&rng->ref);
 
-	rng = get_current_rng_nolock();
+	rcu_read_unlock();
 
-	mutex_unlock(&rng_mutex);
 	return rng;
 }
 
 static void put_rng(struct hwrng *rng)
 {
-	/*
-	 * Hold rng_mutex here so we serialize in case they set_current_rng
-	 * on rng again immediately.
-	 */
-	mutex_lock(&rng_mutex);
 	if (rng)
 		kref_put(&rng->ref, cleanup_rng);
-	mutex_unlock(&rng_mutex);
 }
 
 static int hwrng_init(struct hwrng *rng)

@@ -213,10 +251,6 @@ static ssize_t rng_dev_read(struct file *filp, char __user *buf,
 
 	while (size) {
 		rng = get_current_rng();
-		if (IS_ERR(rng)) {
-			err = PTR_ERR(rng);
-			goto out;
-		}
 		if (!rng) {
 			err = -ENODEV;
 			goto out;

@@ -303,7 +337,7 @@ static struct miscdevice rng_miscdev = {
 
 static int enable_best_rng(void)
 {
-	struct hwrng *rng, *new_rng = NULL;
+	struct hwrng *rng, *cur_rng, *new_rng = NULL;
 	int ret = -ENODEV;
 
 	BUG_ON(!mutex_is_locked(&rng_mutex));

@@ -321,7 +355,9 @@ static int enable_best_rng(void)
 			new_rng = rng;
 	}
 
-	ret = ((new_rng == current_rng) ? 0 : set_current_rng(new_rng));
+	cur_rng = rcu_dereference_protected(current_rng,
+					    lockdep_is_held(&rng_mutex));
+	ret = ((new_rng == cur_rng) ? 0 : set_current_rng(new_rng));
 	if (!ret)
 		cur_rng_set_by_user = 0;
 

@@ -371,8 +407,6 @@ static ssize_t rng_current_show(struct device *dev,
 	struct hwrng *rng;
 
 	rng = get_current_rng();
-	if (IS_ERR(rng))
-		return PTR_ERR(rng);
 
 	ret = sysfs_emit(buf, "%s\n", rng ? rng->name : "none");
 	put_rng(rng);

@@ -416,8 +450,6 @@ static ssize_t rng_quality_show(struct device *dev,
 	struct hwrng *rng;
 
 	rng = get_current_rng();
-	if (IS_ERR(rng))
-		return PTR_ERR(rng);
 
 	if (!rng) /* no need to put_rng */
 		return -ENODEV;

@@ -432,6 +464,7 @@ static ssize_t rng_quality_store(struct device *dev,
 				 struct device_attribute *attr,
 				 const char *buf, size_t len)
 {
+	struct hwrng *rng;
 	u16 quality;
 	int ret = -EINVAL;
 

@@ -448,12 +481,13 @@ static ssize_t rng_quality_store(struct device *dev,
 		goto out;
 	}
 
-	if (!current_rng) {
+	rng = rcu_dereference_protected(current_rng, lockdep_is_held(&rng_mutex));
+	if (!rng) {
 		ret = -ENODEV;
 		goto out;
 	}
 
-	current_rng->quality = quality;
+	rng->quality = quality;
 	current_quality = quality; /* obsolete */
 
 	/* the best available RNG may have changed */

@@ -489,8 +523,20 @@ static int hwrng_fillfn(void *unused)
 		struct hwrng *rng;
 
 		rng = get_current_rng();
-		if (IS_ERR(rng) || !rng)
+		if (!rng) {
+			/*
+			 * Keep the task_struct alive until kthread_stop()
+			 * is called to avoid UAF in drop_current_rng().
+			 */
+			while (!kthread_should_stop()) {
+				set_current_state(TASK_INTERRUPTIBLE);
+				if (!kthread_should_stop())
+					schedule();
+			}
+			set_current_state(TASK_RUNNING);
 			break;
+		}
+
 		mutex_lock(&reading_mutex);
 		rc = rng_get_data(rng, rng_fillbuf,
 				  rng_buffer_size(), 1);

@@ -518,14 +564,13 @@ static int hwrng_fillfn(void *unused)
 		add_hwgenerator_randomness((void *)rng_fillbuf, rc,
 					   entropy >> 10, true);
 	}
-	hwrng_fill = NULL;
 	return 0;
 }
 
 int hwrng_register(struct hwrng *rng)
 {
 	int err = -EINVAL;
-	struct hwrng *tmp;
+	struct hwrng *cur_rng, *tmp;
 
 	if (!rng->name || (!rng->data_read && !rng->read))
 		goto out;

@@ -540,6 +585,7 @@ int hwrng_register(struct hwrng *rng)
 	}
 	list_add_tail(&rng->list, &rng_list);
 
+	INIT_WORK(&rng->cleanup_work, cleanup_rng_work);
 	init_completion(&rng->cleanup_done);
 	complete(&rng->cleanup_done);
 	init_completion(&rng->dying);

@@ -547,16 +593,19 @@ int hwrng_register(struct hwrng *rng)
 	/* Adjust quality field to always have a proper value */
 	rng->quality = min_t(u16, min_t(u16, default_quality, 1024), rng->quality ?: 1024);
 
-	if (!cur_rng_set_by_user &&
-	    (!current_rng || rng->quality > current_rng->quality)) {
-		/*
-		 * Set new rng as current as the new rng source
-		 * provides better entropy quality and was not
-		 * chosen by userspace.
-		 */
-		err = set_current_rng(rng);
-		if (err)
-			goto out_unlock;
+	if (!cur_rng_set_by_user) {
+		cur_rng = rcu_dereference_protected(current_rng,
+						    lockdep_is_held(&rng_mutex));
+		if (!cur_rng || rng->quality > cur_rng->quality) {
+			/*
+			 * Set new rng as current as the new rng source
+			 * provides better entropy quality and was not
+			 * chosen by userspace.
+			 */
+			err = set_current_rng(rng);
+			if (err)
+				goto out_unlock;
+		}
 	}
 	mutex_unlock(&rng_mutex);
 	return 0;

@@ -569,14 +618,17 @@ EXPORT_SYMBOL_GPL(hwrng_register);
 
 void hwrng_unregister(struct hwrng *rng)
 {
-	struct hwrng *new_rng;
+	struct hwrng *cur_rng;
 	int err;
 
 	mutex_lock(&rng_mutex);
 
 	list_del(&rng->list);
 	complete_all(&rng->dying);
-	if (current_rng == rng) {
+
+	cur_rng = rcu_dereference_protected(current_rng,
+					    lockdep_is_held(&rng_mutex));
+	if (cur_rng == rng) {
 		err = enable_best_rng();
 		if (err) {
 			drop_current_rng();

@@ -584,17 +636,7 @@ void hwrng_unregister(struct hwrng *rng)
 		}
 	}
 
-	new_rng = get_current_rng_nolock();
-	if (list_empty(&rng_list)) {
-		mutex_unlock(&rng_mutex);
-		if (hwrng_fill)
-			kthread_stop(hwrng_fill);
-	} else
-		mutex_unlock(&rng_mutex);
-
-	if (new_rng)
-		put_rng(new_rng);
-
+	mutex_unlock(&rng_mutex);
 	wait_for_completion(&rng->cleanup_done);
 }
 EXPORT_SYMBOL_GPL(hwrng_unregister);

@@ -682,7 +724,7 @@ static int __init hwrng_modinit(void)
 static void __exit hwrng_modexit(void)
 {
 	mutex_lock(&rng_mutex);
-	BUG_ON(current_rng);
+	WARN_ON(rcu_access_pointer(current_rng));
 	kfree(rng_buffer);
 	kfree(rng_fillbuf);
 	mutex_unlock(&rng_mutex);

index b424555753b11..b77bc55a4cf35 100644
@@ -15,6 +15,7 @@
 #include <linux/completion.h>
 #include <linux/kref.h>
 #include <linux/types.h>
+#include <linux/workqueue_types.h>
 
 /**
  * struct hwrng - Hardware Random Number Generator driver

@@ -48,6 +49,7 @@ struct hwrng {
 	/* internal. */
 	struct list_head list;
 	struct kref ref;
+	struct work_struct cleanup_work;
 	struct completion cleanup_done;
 	struct completion dying;
 };

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Bastien Curutchet (eBPF Foundation) <[email protected]>

[ Upstream commit 3ab77f35a75eb236956c1e8ba8494ef18a75eae0 ]

AF_XDP features are tested by the test_xsk.sh script but not by the test_progs framework. The tests used by the script are defined in xksxceiver.c which can't be integrated in the test_progs framework as is.

Extract these test definitions from xskxceiver{.c/.h} to put them in new test_xsk{.c/.h} files.
Keep the main() function and its unshared dependencies in xksxceiver to avoid impacting the test_xsk.sh script which is often used to test real hardware.
Move ksft_test_result_*() calls to xskxceiver.c to keep the kselftest's report valid

Stable-dep-of: 42e41b2a0afa ("selftests/xsk: properly handle batch ending in the middle of a packet")

---
 tools/testing/selftests/bpf/Makefile     |    2 +-
 tools/testing/selftests/bpf/test_xsk.c   | 2420 ++++++++++++++++++++
 tools/testing/selftests/bpf/test_xsk.h   |  297 +++
 tools/testing/selftests/bpf/xskxceiver.c | 2545 +---------------------
 tools/testing/selftests/bpf/xskxceiver.h |  156 --
 5 files changed, 2762 insertions(+), 2658 deletions(-)
 create mode 100644 tools/testing/selftests/bpf/test_xsk.c
 create mode 100644 tools/testing/selftests/bpf/test_xsk.h

index e59b2bbf8d920..591e7e77f89ba 100644
@@ -805,7 +805,7 @@ $(OUTPUT)/test_verifier: test_verifier.c verifier/tests.h $(BPFOBJ) | $(OUTPUT)
 
 # Include find_bit.c to compile xskxceiver.
 EXTRA_SRC := $(TOOLSDIR)/lib/find_bit.c
-$(OUTPUT)/xskxceiver: $(EXTRA_SRC) xskxceiver.c xskxceiver.h $(OUTPUT)/network_helpers.o $(OUTPUT)/xsk.o $(OUTPUT)/xsk_xdp_progs.skel.h $(BPFOBJ) | $(OUTPUT)
+$(OUTPUT)/xskxceiver: $(EXTRA_SRC) test_xsk.c test_xsk.h xskxceiver.c xskxceiver.h $(OUTPUT)/network_helpers.o $(OUTPUT)/xsk.o $(OUTPUT)/xsk_xdp_progs.skel.h $(BPFOBJ) | $(OUTPUT)
 	$(call msg,BINARY,,$@)
 	$(Q)$(CC) $(CFLAGS) $(filter %.a %.o %.c,$^) $(LDLIBS) -o $@
 

2.51.0

Part of the diff is not shown here. View the raw message.

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Larysa Zaremba <[email protected]>

[ Upstream commit 42e41b2a0afa04ca49ee2725aadf90ccb058ed28 ]

Referenced commit reduced the scope of the variable pkt, so now it has to be reinitialized via pkt_stream_get_next_rx_pkt(), which also increments some counters. When the packet is interrupted by the batch ending, pkt stream therefore proceeds to the next packet, while xsk ring still contains the previous one, this results in a pkt_nb mismatch.

Decrement the affected counters when packet is interrupted.

---
 tools/testing/selftests/bpf/test_xsk.c | 2 ++
 1 file changed, 2 insertions(+)

index 02250f29f9946..b52f597ea9bd2 100644
@@ -1027,6 +1027,8 @@ static int __receive_pkts(struct test_spec *test, struct xsk_socket_info *xsk)
 			xsk_ring_prod__cancel(&umem->fq, nb_frags);
 		}
 		frags_processed -= nb_frags;
+		pkt_stream_cancel(pkt_stream);
+		pkts_sent--;
 	}
 
 	if (ifobj->use_fill_ring)

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Larysa Zaremba <[email protected]>

[ Upstream commit 88af9fefed412e4bea9a1a771cbe6fe347fa3507 ]

The issue occurs in TOO_MANY_FRAGS test case when xdp_zc_max_segs is set to an odd number.

TOO_MANY_FRAGS test case contains an invalid packet consisting of (xdp_zc_max_segs) frags. Every frag, even the last one has XDP_PKT_CONTD flag set. This packet is expected to be dropped. After that, there is a valid linear packet, which is expected to be received back.

Once (xdp_zc_max_segs) is an odd number, the last packet cannot be received, if packet forwarding between Rx and Tx interfaces relies on the ethernet header, e.g. checks for ETH_P_LOOPBACK. Packet is malformed, if all traffic is looped.

Turns out, sending function processes multiple invalid frags as if they were in 2-frag packets. So once the invalid mbuf packet contains an odd number of those, the valid packet after gets paired with the previous invalid descriptor, and hence does not get an ethernet header generated, so it is either dropped or malformed.

Make invalid packets in verbatim mode always have only a single frag. For such packets, number of frags is otherwise meaningless, as descriptor flags are pre-configured in verbatim mode and packet data is not generated for invalid descriptors.

---
 tools/testing/selftests/bpf/test_xsk.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

index b52f597ea9bd2..55d318c5c5e59 100644
@@ -431,7 +431,7 @@ static u32 pkt_nb_frags(u32 frame_size, struct pkt_stream *pkt_stream, struct pk
 	}
 
 	/* Search for the end of the packet in verbatim mode */
-	if (!pkt_continues(pkt->options))
+	if (!pkt_continues(pkt->options) || !pkt->valid)
 		return nb_frags;
 
 	next_frag = pkt_stream->current_pkt_nb;

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Sai Ritvik Tanksalkar <[email protected]>

[ Upstream commit 5669645c052f235726a85f443769b6fc02f66762 ]

persistent_ram_save_old() can be called multiple times for the same persistent_ram_zone (e.g., via ramoops_pstore_read -> ramoops_get_next_prz for PSTORE_TYPE_DMESG records).

Currently, the function only allocates prz->old_log when it is NULL, but it unconditionally updates prz->old_log_size to the current buffer size and then performs memcpy_fromio() using this new size. If the buffer size has grown since the first allocation (which can happen across different kernel boot cycles), this leads to:

1. A heap buffer overflow (OOB write) in the memcpy_fromio() calls
2. A subsequent OOB read when ramoops_pstore_read() accesses the buffer using the incorrect (larger) old_log_size

The KASAN splat would look similar to:

  BUG: KASAN: slab-out-of-bounds in ramoops_pstore_read+0x...
  Read of size N at addr ... by task ...

The conditions are likely extremely hard to hit:

0. Crash with a ramoops write of less-than-record-max-size bytes.
1. Reboot: ramoops registers, pstore_get_records(0) reads old crash, allocates old_log with size X
2. Crash handler registered, timer started (if pstore_update_ms >= 0)
3. Oops happens (non-fatal, system continues)
4. pstore_dump() writes oops via ramoops_pstore_write() size Y (>X)
5. pstore_new_entry = 1, pstore_timer_kick() called
6. System continues running (not a panic oops)
7. Timer fires after pstore_update_ms milliseconds
8. pstore_timefunc() → schedule_work() → pstore_dowork() → pstore_get_records(1)
9. ramoops_get_next_prz() → persistent_ram_save_old()
10. buffer_size() returns Y, but old_log is X bytes
11. Y > X: memcpy_fromio() overflows heap

Requirements:
- a prior crash record exists that did not fill the record size (almost impossible since the crash handler writes as much as it can possibly fit into the record, capped by max record size and the kmsg buffer almost always exceeds the max record size)
- pstore_update_ms >= 0 (disabled by default)
- Non-fatal oops (system survives)

Free and reallocate the buffer when the new size differs from the previously allocated size. This ensures old_log always has sufficient space for the data being copied.

---
 fs/pstore/ram_core.c | 11 +++++++++++
 1 file changed, 11 insertions(+)

index f1848cdd6d348..c9eaacdec37e4 100644
@@ -298,6 +298,17 @@ void persistent_ram_save_old(struct persistent_ram_zone *prz)
 	if (!size)
 		return;
 
+	/*
+	 * If the existing buffer is differently sized, free it so a new
+	 * one is allocated. This can happen when persistent_ram_save_old()
+	 * is called early in boot and later for a timer-triggered
+	 * survivable crash when the crash dumps don't match in size
+	 * (which would be extremely unlikely given kmsg buffers usually
+	 * exceed prz buffer sizes).
+	 */
+	if (prz->old_log && prz->old_log_size != size)
+		persistent_ram_free_old(prz);
+
 	if (!prz->old_log) {
 		persistent_ram_ecc_old(prz);
 		prz->old_log = kvzalloc(size, GFP_KERNEL);

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Jorge Ramirez-Ortiz <[email protected]>

[ Upstream commit 0fe01a7955f4fef97e7cc6d14bfc5931c660402b ]

Fail the driver probe if the region can't be mapped

---
 drivers/soc/qcom/smem.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

index f1d1b5aa5e4db..39177aa5793ad 100644
@@ -1215,7 +1215,9 @@ static int qcom_smem_probe(struct platform_device *pdev)
 		smem->item_count = qcom_smem_get_item_count(smem);
 		break;
 	case SMEM_GLOBAL_HEAP_VERSION:
-		qcom_smem_map_global(smem, size);
+		ret = qcom_smem_map_global(smem, size);
+		if (ret < 0)
+			return ret;
 		smem->item_count = SMEM_ITEM_COUNT;
 		break;
 	default:

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Dan Carpenter <[email protected]>

[ Upstream commit 7b5c7e83ac405ff9ecbdd92b37a477f4288f8814 ]

The snprintf() can't really overflow because we're writing a max of 42 bytes to a PAGE_SIZE buffer. But the limit calculation doesn't take the first 11 bytes that we wrote into consideration so the limit is not correct. Just fix it for correctness even though it doesn't affect runtime.

---
 drivers/edac/i5000_edac.c | 1 +
 1 file changed, 1 insertion(+)

index 4a1bebc1ff14d..471b8540d18b0 100644
@@ -1111,6 +1111,7 @@ static void calculate_dimm_size(struct i5000_pvt *pvt)
 
 	n = snprintf(p, space, "           ");
 	p += n;
+	space -= n;
 	for (branch = 0; branch < MAX_BRANCHES; branch++) {
 		n = snprintf(p, space, "       branch %d       | ", branch);
 		p += n;

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Dan Carpenter <[email protected]>

[ Upstream commit 72f12683611344853ab030fe7d19b23970ed2bd8 ]

The snprintf() can't really overflow because we're writing a max of 42 bytes to a PAGE_SIZE buffer. But my static checker complains because the limit calculation doesn't take the first 11 space characters that we wrote into the buffer into consideration. Fix this for the sake of correctness even though it doesn't affect runtime.

Also delete an earlier "space -= n;" which was not used.

---
 drivers/edac/i5400_edac.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

index b5cf25905b059..fb49a1d1df112 100644
@@ -1026,13 +1026,13 @@ static void calculate_dimm_size(struct i5400_pvt *pvt)
 		space -= n;
 	}
 
-	space -= n;
 	edac_dbg(2, "%s\n", mem_buffer);
 	p = mem_buffer;
 	space = PAGE_SIZE;
 
 	n = snprintf(p, space, "           ");
 	p += n;
+	space -= n;
 	for (branch = 0; branch < MAX_BRANCHES; branch++) {
 		n = snprintf(p, space, "       branch %d       | ", branch);
 		p += n;

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Sudeep Holla <[email protected]>

[ Upstream commit be4d4543f78074fbebd530ba5109d39a2a34e668 ]

The FF-A specification allows NOTIFICATION_INFO_GET to return either a 64-bit (FFA_FN64_SUCCESS) or a 32-bit (FFA_SUCCESS) response, depending on whether the firmware chooses the SMC64 or SMC32 calling convention.

The driver previously detected the response format by checking ret.a0, but still interpreted the returned ID lists (x3..x17 or w3..w7) as if they always followed the 64-bit SMC64 layout. In the SMC32 case, the upper 32 bits of each argument register are undefined by the calling convention, meaning the driver could read stale or garbage values when parsing notification IDs.

This resulted in incorrectly decoded partition/VCPU IDs whenever the FF-A firmware used an SMC32 return path.

Fix the issue by:

- Introducing logic to map list indices to the correct u16 offsets, depending on whether the response width matches the kernel word size or is a 32-bit response on a 64-bit kernel.
- Ensuring that the packed ID list is parsed using the proper layout, avoiding reads from undefined upper halves in the SMC32 case.

With this change, NOTIFICATION_INFO_GET now correctly interprets ID list entries regardless of the response width, aligning the driver with the FF-A specification.

Message-Id: <[email protected]>

---
 drivers/firmware/arm_ffa/driver.c | 33 +++++++++++++++++++++++++++----
 1 file changed, 29 insertions(+), 4 deletions(-)

index c72ee47565856..c501c3104b3a4 100644
@@ -981,10 +981,27 @@ static void __do_sched_recv_cb(u16 part_id, u16 vcpu, bool is_per_vcpu)
 	}
 }
 
+/*
+ * Map logical ID index to the u16 index within the packed ID list.
+ *
+ * For native responses (FF-A width == kernel word size), IDs are
+ * tightly packed: idx -> idx.
+ *
+ * For 32-bit responses on a 64-bit kernel, each 64-bit register
+ * contributes 4 x u16 values but only the lower 2 are defined; the
+ * upper 2 are garbage. This mapping skips those upper halves:
+ *   0,1,2,3,4,5,... -> 0,1,4,5,8,9,...
+ */
+static int list_idx_to_u16_idx(int idx, bool is_native_resp)
+{
+	return is_native_resp ? idx : idx + 2 * (idx >> 1);
+}
+
 static void ffa_notification_info_get(void)
 {
-	int idx, list, max_ids, lists_cnt, ids_processed, ids_count[MAX_IDS_64];
-	bool is_64b_resp;
+	int ids_processed, ids_count[MAX_IDS_64];
+	int idx, list, max_ids, lists_cnt;
+	bool is_64b_resp, is_native_resp;
 	ffa_value_t ret;
 	u64 id_list;
 

@@ -1001,6 +1018,7 @@ static void ffa_notification_info_get(void)
 		}
 
 		is_64b_resp = (ret.a0 == FFA_FN64_SUCCESS);
+		is_native_resp = (ret.a0 == FFA_FN_NATIVE(SUCCESS));
 
 		ids_processed = 0;
 		lists_cnt = FIELD_GET(NOTIFICATION_INFO_GET_ID_COUNT, ret.a2);

@@ -1017,12 +1035,16 @@ static void ffa_notification_info_get(void)
 
 		/* Process IDs */
 		for (list = 0; list < lists_cnt; list++) {
+			int u16_idx;
 			u16 vcpu_id, part_id, *packed_id_list = (u16 *)&ret.a3;
 
 			if (ids_processed >= max_ids - 1)
 				break;
 
-			part_id = packed_id_list[ids_processed++];
+			u16_idx = list_idx_to_u16_idx(ids_processed,
+						      is_native_resp);
+			part_id = packed_id_list[u16_idx];
+			ids_processed++;
 
 			if (ids_count[list] == 1) { /* Global Notification */
 				__do_sched_recv_cb(part_id, 0, false);

@@ -1034,7 +1056,10 @@ static void ffa_notification_info_get(void)
 				if (ids_processed >= max_ids - 1)
 					break;
 
-				vcpu_id = packed_id_list[ids_processed++];
+				u16_idx = list_idx_to_u16_idx(ids_processed,
+							      is_native_resp);
+				vcpu_id = packed_id_list[u16_idx];
+				ids_processed++;
 
 				__do_sched_recv_cb(part_id, vcpu_id, true);
 			}

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Alexander Stein <[email protected]>

[ Upstream commit 8401527abb5e3a00c867b6597b8e1b29c80c9824 ]

As per datasheet of the HDMI protection IC the CEC_IC pin has been configured as open-drain.

---
 arch/arm64/boot/dts/freescale/imx8mp-tqma8mpql-mba8mpxl.dts | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

index 4eedd00d83b9f..ac05c05193c5b 100644
@@ -860,7 +860,7 @@ pinctrl_hdmi: hdmigrp {
 		fsl,pins = <MX8MP_IOMUXC_HDMI_DDC_SCL__HDMIMIX_HDMI_SCL	0x400001c2>,
 			   <MX8MP_IOMUXC_HDMI_DDC_SDA__HDMIMIX_HDMI_SDA	0x400001c2>,
 			   <MX8MP_IOMUXC_HDMI_HPD__HDMIMIX_HDMI_HPD	0x40000010>,
-			   <MX8MP_IOMUXC_HDMI_CEC__HDMIMIX_HDMI_CEC	0x40000010>;
+			   <MX8MP_IOMUXC_HDMI_CEC__HDMIMIX_HDMI_CEC	0x40000030>;
 	};
 
 	pinctrl_hoggpio2: hoggpio2grp {

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Alexander Stein <[email protected]>

[ Upstream commit 53a5c1d98d1155ece4c9446c0fea55e17d08774a ]

As per datasheet of the HDMI protection IC the CEC_IC pin has been configured as open-drain.

---
 .../arm64/boot/dts/freescale/imx8mp-tqma8mpql-mba8mp-ras314.dts | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

index f7346b3d35fe5..a122f2ed5f531 100644
@@ -704,7 +704,7 @@ pinctrl_hdmi: hdmigrp {
 		fsl,pins = <MX8MP_IOMUXC_HDMI_DDC_SCL__HDMIMIX_HDMI_SCL	0x400001c2>,
 			   <MX8MP_IOMUXC_HDMI_DDC_SDA__HDMIMIX_HDMI_SDA	0x400001c2>,
 			   <MX8MP_IOMUXC_HDMI_HPD__HDMIMIX_HDMI_HPD	0x40000010>,
-			   <MX8MP_IOMUXC_HDMI_CEC__HDMIMIX_HDMI_CEC	0x40000154>;
+			   <MX8MP_IOMUXC_HDMI_CEC__HDMIMIX_HDMI_CEC	0x40000030>;
 	};
 
 	pinctrl_gpt1: gpt1grp {

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Haotian Zhang <[email protected]>

[ Upstream commit 1e07ebe744fb522983bd52a4a6148601675330c7 ]

When devm_platform_ioremap_resource() fails, it returns various error codes. Returning a hardcoded -ENOMEM masks the actual failure reason.

Use PTR_ERR() to propagate the actual error code returned by devm_platform_ioremap_resource() instead of -ENOMEM.

---
 drivers/clk/qcom/common.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

index 1215918867741..eec369d2173b5 100644
@@ -454,7 +454,7 @@ int qcom_cc_probe_by_index(struct platform_device *pdev, int index,
 
 	base = devm_platform_ioremap_resource(pdev, index);
 	if (IS_ERR(base))
-		return -ENOMEM;
+		return PTR_ERR(base);
 
 	regmap = devm_regmap_init_mmio(&pdev->dev, base, desc->config);
 	if (IS_ERR(regmap))

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Dmitry Baryshkov <[email protected]>

[ Upstream commit e814796dfcae8905682ac3ac2dd57f512a9f6726 ]

Historically sdm630.dtsi has used 1 byte length for the gpu_speed_bin cell, although it spans two bytes (offset 5, size 7 bits). It was being accepted by the kernel because before the commit 7a06ef751077 ("nvmem:
core: fix bit offsets of more than one byte") the kernel didn't have length check. After this commit nvmem core rejects QFPROM on sdm630 / sdm660, making GPU and USB unusable on those platforms.

Set the size of the gpu_speed_bin cell to 2 bytes, fixing the parsing error. While we are at it, update the length to 8 bits as pointed out by Alexey Minnekhanov.

---
 arch/arm64/boot/dts/qcom/sdm630.dtsi | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

index 8b1a45a4e56ed..b383e480a394d 100644
@@ -598,8 +598,8 @@ qusb2_hstx_trim: hstx-trim@240 {
 			};
 
 			gpu_speed_bin: gpu-speed-bin@41a0 {
-				reg = <0x41a2 0x1>;
-				bits = <5 7>;
+				reg = <0x41a2 0x2>;
+				bits = <5 8>;
 			};
 		};
 

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Casey Connolly <[email protected]>

[ Upstream commit c9b98b9dad9749bf2eb7336a6fca31a6af1039d7 ]

The touchscreen isn't enabled by bootloader and doesn't need to be enabled at boot, only when the driver probes, thus remove the regulator-boot-on property.

---
 arch/arm64/boot/dts/qcom/sdm845-oneplus-common.dtsi | 1 -
 1 file changed, 1 deletion(-)

index 51a9a276399ac..35d105ff689b9 100644
@@ -160,7 +160,6 @@ ts_1p8_supply: ts-1p8-regulator {
 
 		gpio = <&tlmm 88 0>;
 		enable-active-high;
-		regulator-boot-on;
 	};
 };
 

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Casey Connolly <[email protected]>

[ Upstream commit 45d1f42d3e84b5880cf9fab1eb24a7818320eeb7 ]

The panel regulator doesn't need to be always on, so remove this property.

---
 arch/arm64/boot/dts/qcom/sdm845-oneplus-common.dtsi | 1 -
 1 file changed, 1 deletion(-)

index 35d105ff689b9..7169da658dcd0 100644
@@ -256,7 +256,6 @@ vreg_l14a_1p88: ldo14 {
 			regulator-min-microvolt = <1800000>;
 			regulator-max-microvolt = <1800000>;
 			regulator-initial-mode = <RPMH_REGULATOR_MODE_HPM>;
-			regulator-always-on;
 		};
 
 		vreg_l17a_1p3: ldo17 {

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Casey Connolly <[email protected]>

[ Upstream commit ad33ee060be46794a03d033894c9db3a9d6c1a0f ]

This regulator is used only for the display, which is enabled by the bootloader and left on for continuous splash. Mark it as such.

---
 arch/arm64/boot/dts/qcom/sdm845-oneplus-common.dtsi | 1 +
 1 file changed, 1 insertion(+)

index 7169da658dcd0..1036305231b20 100644
@@ -256,6 +256,7 @@ vreg_l14a_1p88: ldo14 {
 			regulator-min-microvolt = <1800000>;
 			regulator-max-microvolt = <1800000>;
 			regulator-initial-mode = <RPMH_REGULATOR_MODE_HPM>;
+			regulator-boot-on;
 		};
 
 		vreg_l17a_1p3: ldo17 {

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Abel Vesa <[email protected]>

[ Upstream commit 3af51501e2b8c87564b5cda43b0e5c316cf54717 ]

It seems the USB combo SS1 and SS2 ref clocks have another gate, unlike the SS0. These gates are part of the TCSR clock controller.

At least on Dell XPS 13 (9345), if the ref clock provided by the TCSR clock controller for SS1 PHY is disabled on the clk_disable_unused late initcall, the PHY fails to initialize. It doesn't happen on the SS0 PHY and the SS2 is not used on this device.

This doesn't seem to be a problem on CRD though. It might be that the RPMh has a vote for it from some other consumer and does not actually disable it when ther kernel drops its vote.

Either way, these TCSR provided clocks seem to be the correct ones for the SS1 and SS2, so use them instead.

---
 arch/arm64/boot/dts/qcom/x1e80100.dtsi | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

index 662ad694cd914..3290fd8c2d6e6 100644
@@ -2910,7 +2910,7 @@ usb_1_ss1_qmpphy: phy@fda000 {
 			reg = <0 0x00fda000 0 0x4000>;
 
 			clocks = <&gcc GCC_USB3_SEC_PHY_AUX_CLK>,
-				 <&rpmhcc RPMH_CXO_CLK>,
+				 <&tcsr TCSR_USB4_1_CLKREF_EN>,
 				 <&gcc GCC_USB3_SEC_PHY_COM_AUX_CLK>,
 				 <&gcc GCC_USB3_SEC_PHY_PIPE_CLK>;
 			clock-names = "aux",

@@ -2981,7 +2981,7 @@ usb_1_ss2_qmpphy: phy@fdf000 {
 			reg = <0 0x00fdf000 0 0x4000>;
 
 			clocks = <&gcc GCC_USB3_TERT_PHY_AUX_CLK>,
-				 <&rpmhcc RPMH_CXO_CLK>,
+				 <&tcsr TCSR_USB4_2_CLKREF_EN>,
 				 <&gcc GCC_USB3_TERT_PHY_COM_AUX_CLK>,
 				 <&gcc GCC_USB3_TERT_PHY_PIPE_CLK>;
 			clock-names = "aux",

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Geert Uytterhoeven <[email protected]>

[ Upstream commit 44cfd102ce28e749a07bb0f1668cf932077b1175 ]

SW_LCD_EN is defined twice.

---
 arch/arm64/boot/dts/renesas/r9a09g047e57-smarc.dts | 1 -
 1 file changed, 1 deletion(-)

index 08e814c03fa85..ed6fcdc337a0b 100644
@@ -8,7 +8,6 @@
 /dts-v1/;
 
 /* Switch selection settings */
-#define SW_LCD_EN		0
 #define SW_GPIO8_CAN0_STB	0
 #define SW_GPIO9_CAN1_STB	0
 #define SW_LCD_EN		0

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Vladimir Zapolskiy <[email protected]>

[ Upstream commit 7db5fbe508deedec6c183d5056cf3c504c027f40 ]

Trivial change, Analog Devices vendor prefix is "adi", but there is a valid "ad" vendor prefix of another company, this may explain why the issue hasn't been discovered by the automatic tests.

A problem of not described compatible value is out of this change scope.

---
 arch/arm64/boot/dts/qcom/msm8994-msft-lumia-octagon.dtsi | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

index 4c983b10dd925..7ace3540ef0a0 100644
@@ -378,7 +378,7 @@ &blsp2_i2c1 {
 	status = "okay";
 
 	sideinteraction: touch@2c {
-		compatible = "ad,ad7147_captouch";
+		compatible = "adi,ad7147_captouch";
 		reg = <0x2c>;
 
 		pinctrl-names = "default", "sleep";

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Chen-Yu Tsai <[email protected]>

[ Upstream commit 0b2761eb1287bd9f62367cccf6626eb3107cef6f ]

The P66's device tree includes the reference design dtsi files, which defines a node and properties for the touchpanel in the common design. The P66 dts file then overrides all the properties to match its own design, but as the touchpanel model is different, a different schema is matched. This other schema uses a different name for the GPIO.

The original submission added the correct GPIO property, but did not delete the one inherited from the reference design, causing validation errors.

Explicitly delete the incorrect GPIO property.

---
 arch/arm/boot/dts/allwinner/sun5i-a13-utoo-p66.dts | 1 +
 1 file changed, 1 insertion(+)

index be486d28d04fa..428cab5a0e906 100644
@@ -102,6 +102,7 @@ &touchscreen {
 	/* The P66 uses a different EINT then the reference design */
 	interrupts = <6 9 IRQ_TYPE_EDGE_FALLING>; /* EINT9 (PG9) */
 	/* The icn8318 binding expects wake-gpios instead of power-gpios */
+	/delete-property/ power-gpios;
 	wake-gpios = <&pio 1 3 GPIO_ACTIVE_HIGH>; /* PB3 */
 	touchscreen-size-x = <800>;
 	touchscreen-size-y = <480>;

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Christophe Leroy <[email protected]>

[ Upstream commit 5fbc09eb0b4f4b1a4b33abebacbeee0d29f195e9 ]

Commit 74e19ef0ff80 ("uaccess: Add speculation barrier to copy_from_user()") added a redundant barrier_nospec() in copy_from_user(), because powerpc is already calling barrier_nospec() in allow_read_from_user() and allow_read_write_user(). But on other architectures that call to barrier_nospec() was missing. So change powerpc instead of reverting the above commit and having to fix other architectures one by one. This is now possible because barrier_nospec() has also been added in copy_from_user_iter().

Move barrier_nospec() out of allow_read_from_user() and allow_read_write_user(). This will also allow reuse of those functions when implementing masked user access which doesn't require barrier_nospec().

Don't add it back in raw_copy_from_user() as it is already called by copy_from_user() and copy_from_user_iter().

---
 arch/powerpc/include/asm/kup.h     | 2 --
 arch/powerpc/include/asm/uaccess.h | 4 ++++
 2 files changed, 4 insertions(+), 2 deletions(-)

index dab63b82a8d4f..f2009d7c8cfa7 100644
@@ -134,7 +134,6 @@ static __always_inline void kuap_assert_locked(void)
 
 static __always_inline void allow_read_from_user(const void __user *from, unsigned long size)
 {
-	barrier_nospec();
 	allow_user_access(NULL, from, size, KUAP_READ);
 }
 

@@ -146,7 +145,6 @@ static __always_inline void allow_write_to_user(void __user *to, unsigned long s
 static __always_inline void allow_read_write_user(void __user *to, const void __user *from,
 						  unsigned long size)
 {
-	barrier_nospec();
 	allow_user_access(to, from, size, KUAP_READ_WRITE);
 }
 

index 4f5a46a77fa2b..3987a5c33558b 100644
@@ -301,6 +301,7 @@ do {								\
 	__typeof__(sizeof(*(ptr))) __gu_size = sizeof(*(ptr));	\
 								\
 	might_fault();					\
+	barrier_nospec();					\
 	allow_read_from_user(__gu_addr, __gu_size);		\
 	__get_user_size_allowed(__gu_val, __gu_addr, __gu_size, __gu_err);	\
 	prevent_read_from_user(__gu_addr, __gu_size);		\

@@ -329,6 +330,7 @@ raw_copy_in_user(void __user *to, const void __user *from, unsigned long n)
 {
 	unsigned long ret;
 
+	barrier_nospec();
 	allow_read_write_user(to, from, n);
 	ret = __copy_tofrom_user(to, from, n);
 	prevent_read_write_user(to, from, n);

@@ -415,6 +417,7 @@ static __must_check __always_inline bool user_access_begin(const void __user *pt
 
 	might_fault();
 
+	barrier_nospec();
 	allow_read_write_user((void __user *)ptr, ptr, len);
 	return true;
 }

@@ -431,6 +434,7 @@ user_read_access_begin(const void __user *ptr, size_t len)
 
 	might_fault();
 
+	barrier_nospec();
 	allow_read_from_user(ptr, len);
 	return true;
 }

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Haotian Zhang <[email protected]>

[ Upstream commit 0da7824734d8d83e6a844dd0207f071cb0c50cf4 ]

If cmd_db_magic_matches() fails after memremap() succeeds, the function returns -EINVAL without unmapping the memory region, causing a potential resource leak.

Switch to devm_memremap to automatically manage the map resource.

---
 drivers/soc/qcom/cmd-db.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

index ae66c2623d250..84a75d8c4b702 100644
@@ -349,15 +349,16 @@ static int cmd_db_dev_probe(struct platform_device *pdev)
 		return -EINVAL;
 	}
 
-	cmd_db_header = memremap(rmem->base, rmem->size, MEMREMAP_WC);
-	if (!cmd_db_header) {
-		ret = -ENOMEM;
+	cmd_db_header = devm_memremap(&pdev->dev, rmem->base, rmem->size, MEMREMAP_WC);
+	if (IS_ERR(cmd_db_header)) {
+		ret = PTR_ERR(cmd_db_header);
 		cmd_db_header = NULL;
 		return ret;
 	}
 
 	if (!cmd_db_magic_matches(cmd_db_header)) {
 		dev_err(&pdev->dev, "Invalid Command DB Magic\n");
+		cmd_db_header = NULL;
 		return -EINVAL;
 	}
 

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Zilin Guan <[email protected]>

[ Upstream commit 6259094ee806fb813ca95894c65fb80e2ec98bf1 ]

In svs_enable_debug_write(), the buf allocated by memdup_user_nul() is leaked if kstrtoint() fails.

Fix this by using __free(kfree) to automatically free buf, eliminating the need for explicit kfree() calls and preventing leaks.

---
 drivers/soc/mediatek/mtk-svs.c | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)

index f45537546553e..99edecb204f25 100644
@@ -9,6 +9,7 @@
 #include <linux/bits.h>
 #include <linux/clk.h>
 #include <linux/completion.h>
+#include <linux/cleanup.h>
 #include <linux/cpu.h>
 #include <linux/cpuidle.h>
 #include <linux/debugfs.h>

@@ -789,7 +790,7 @@ static ssize_t svs_enable_debug_write(struct file *filp,
 	struct svs_bank *svsb = file_inode(filp)->i_private;
 	struct svs_platform *svsp = dev_get_drvdata(svsb->dev);
 	int enabled, ret;
-	char *buf = NULL;
+	char *buf __free(kfree) = NULL;
 
 	if (count >= PAGE_SIZE)
 		return -EINVAL;

@@ -807,8 +808,6 @@ static ssize_t svs_enable_debug_write(struct file *filp,
 		svsb->mode_support = SVSB_MODE_ALL_DISABLE;
 	}
 
-	kfree(buf);
-
 	return count;
 }
 

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Narayana Murty N <[email protected]>

[ Upstream commit 815a8d2feb5615ae7f0b5befd206af0b0160614c ]

The recent commit 1010b4c012b0 ("powerpc/eeh: Make EEH driver device hotplug safe") restructured the EEH driver to improve synchronization with the PCI hotplug layer.

However, it inadvertently moved pci_lock_rescan_remove() outside its intended scope in eeh_handle_normal_event(), leading to broken PCI error reporting and improper EEH event triggering. Specifically, eeh_handle_normal_event() acquired pci_lock_rescan_remove() before calling eeh_pe_bus_get(), but eeh_pe_bus_get() itself attempts to acquire the same lock internally, causing nested locking and disrupting normal EEH event handling paths.

This patch adds a boolean parameter do_lock to _eeh_pe_bus_get(), with two public wrappers:
eeh_pe_bus_get() with locking enabled. eeh_pe_bus_get_nolock() that skips locking.

Callers that already hold pci_lock_rescan_remove() now use eeh_pe_bus_get_nolock() to avoid recursive lock acquisition.

Additionally, pci_lock_rescan_remove() calls are restored to the correct position—after eeh_pe_bus_get() and immediately before iterating affected PEs and devices. This ensures EEH-triggered PCI removes occur under proper bus rescan locking without recursive lock contention.

The eeh_pe_loc_get() function has been split into two functions:
eeh_pe_loc_get(struct eeh_pe *pe) which retrieves the loc for given PE. eeh_pe_loc_get_bus(struct pci_bus *bus) which retrieves the location code for given bus.

This resolves lockdep warnings such as:
<snip>

[   84.964298] [    T928] ============================================
[   84.964304] [    T928] WARNING: possible recursive locking detected
[   84.964311] [    T928] 6.18.0-rc3 #51 Not tainted
[   84.964315] [    T928] --------------------------------------------
[   84.964320] [    T928] eehd/928 is trying to acquire lock:
[   84.964324] [    T928] c000000003b29d58 (pci_rescan_remove_lock){+.+.}-{3:3}, at: pci_lock_rescan_remove+0x28/0x40
[   84.964342] [    T928]
                       but task is already holding lock:
[   84.964347] [    T928] c000000003b29d58 (pci_rescan_remove_lock){+.+.}-{3:3}, at: pci_lock_rescan_remove+0x28/0x40
[   84.964357] [    T928]
                       other info that might help us debug this:
[   84.964363] [    T928]  Possible unsafe locking scenario:

[   84.964367] [    T928]        CPU0
[   84.964370] [    T928]        ----
[   84.964373] [    T928]   lock(pci_rescan_remove_lock);
[   84.964378] [    T928]   lock(pci_rescan_remove_lock);
[   84.964383] [    T928]
                       *** DEADLOCK ***

[   84.964388] [    T928]  May be due to missing lock nesting notation

[   84.964393] [    T928] 1 lock held by eehd/928:
[   84.964397] [    T928]  #0: c000000003b29d58 (pci_rescan_remove_lock){+.+.}-{3:3}, at: pci_lock_rescan_remove+0x28/0x40
[   84.964408] [    T928]
                       stack backtrace:
[   84.964414] [    T928] CPU: 2 UID: 0 PID: 928 Comm: eehd Not tainted 6.18.0-rc3 #51 VOLUNTARY
[   84.964417] [    T928] Hardware name: IBM,9080-HEX POWER10 (architected) 0x800200 0xf000006 of:IBM,FW1060.00 (NH1060_022) hv:phyp pSeries
[   84.964419] [    T928] Call Trace:
[   84.964420] [    T928] [c0000011a7157990] [c000000001705de4] dump_stack_lvl+0xc8/0x130 (unreliable)
[   84.964424] [    T928] [c0000011a71579d0] [c0000000002f66e0] print_deadlock_bug+0x430/0x440
[   84.964428] [    T928] [c0000011a7157a70] [c0000000002fd0c0] __lock_acquire+0x1530/0x2d80
[   84.964431] [    T928] [c0000011a7157ba0] [c0000000002fea54] lock_acquire+0x144/0x410
[   84.964433] [    T928] [c0000011a7157cb0] [c0000011a7157cb0] __mutex_lock+0xf4/0x1050
[   84.964436] [    T928] [c0000011a7157e00] [c000000000de21d8] pci_lock_rescan_remove+0x28/0x40
[   84.964439] [    T928] [c0000011a7157e20] [c00000000004ed98] eeh_pe_bus_get+0x48/0xc0
[   84.964442] [    T928] [c0000011a7157e50] [c000000000050434] eeh_handle_normal_event+0x64/0xa60
[   84.964446] [    T928] [c0000011a7157f30] [c000000000051de8] eeh_event_handler+0xf8/0x190
[   84.964450] [    T928] [c0000011a7157f90] [c0000000002747ac] kthread+0x16c/0x180
[   84.964453] [    T928] [c0000011a7157fe0] [c00000000000ded8] start_kernel_thread+0x14/0x18

</snip>

---
 arch/powerpc/include/asm/eeh.h   |  2 +
 arch/powerpc/kernel/eeh_driver.c | 11 ++---
 arch/powerpc/kernel/eeh_pe.c     | 74 ++++++++++++++++++++++++++++++--
 3 files changed, 78 insertions(+), 9 deletions(-)

index 5e34611de9ef4..b7ebb4ac2c710 100644
@@ -289,6 +289,8 @@ void eeh_pe_dev_traverse(struct eeh_pe *root,
 void eeh_pe_restore_bars(struct eeh_pe *pe);
 const char *eeh_pe_loc_get(struct eeh_pe *pe);
 struct pci_bus *eeh_pe_bus_get(struct eeh_pe *pe);
+const char *eeh_pe_loc_get_bus(struct pci_bus *bus);
+struct pci_bus *eeh_pe_bus_get_nolock(struct eeh_pe *pe);
 
 void eeh_show_enabled(void);
 int __init eeh_init(struct eeh_ops *ops);

index ef78ff77cf8f2..028f691585323 100644
@@ -846,7 +846,7 @@ void eeh_handle_normal_event(struct eeh_pe *pe)
 
 	pci_lock_rescan_remove();
 
-	bus = eeh_pe_bus_get(pe);
+	bus = eeh_pe_bus_get_nolock(pe);
 	if (!bus) {
 		pr_err("%s: Cannot find PCI bus for PHB#%x-PE#%x\n",
 			__func__, pe->phb->global_number, pe->addr);

@@ -886,14 +886,15 @@ void eeh_handle_normal_event(struct eeh_pe *pe)
 	/* Log the event */
 	if (pe->type & EEH_PE_PHB) {
 		pr_err("EEH: Recovering PHB#%x, location: %s\n",
-			pe->phb->global_number, eeh_pe_loc_get(pe));
+			pe->phb->global_number, eeh_pe_loc_get_bus(bus));
 	} else {
 		struct eeh_pe *phb_pe = eeh_phb_pe_get(pe->phb);
 
 		pr_err("EEH: Recovering PHB#%x-PE#%x\n",
 		       pe->phb->global_number, pe->addr);
 		pr_err("EEH: PE location: %s, PHB location: %s\n",
-		       eeh_pe_loc_get(pe), eeh_pe_loc_get(phb_pe));
+		       eeh_pe_loc_get_bus(bus),
+		       eeh_pe_loc_get_bus(eeh_pe_bus_get_nolock(phb_pe)));
 	}
 
 #ifdef CONFIG_STACKTRACE

@@ -1098,7 +1099,7 @@ void eeh_handle_normal_event(struct eeh_pe *pe)
 		eeh_pe_state_clear(pe, EEH_PE_PRI_BUS, true);
 		eeh_pe_dev_mode_mark(pe, EEH_DEV_REMOVED);
 
-		bus = eeh_pe_bus_get(pe);
+		bus = eeh_pe_bus_get_nolock(pe);
 		if (bus)
 			pci_hp_remove_devices(bus);
 		else

@@ -1222,7 +1223,7 @@ void eeh_handle_special_event(void)
 				    (phb_pe->state & EEH_PE_RECOVERING))
 					continue;
 
-				bus = eeh_pe_bus_get(phb_pe);
+				bus = eeh_pe_bus_get_nolock(phb_pe);
 				if (!bus) {
 					pr_err("%s: Cannot find PCI bus for "
 					       "PHB#%x-PE#%x\n",

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Lad Prabhakar <[email protected]>

[ Upstream commit a1b1ee0348f889ec262482e16e9ff670617db7b0 ]

Switch SD0 write-protect detection to a GPIO on the RZ/T2H and RZ/N2H EVKs. Both boards use a full-size SD card slot on the SD0 channel with a dedicated WP pin.

The RZ/T2H and RZ/N2H SoCs use of_data_rcar_gen3, which sets MMC_CAP2_NO_WRITE_PROTECT and causes the core to ignore the WP signal unless a wp-gpios property is provided. Describe the WP pin as a GPIO to allow the MMC core to evaluate the write-protect status correctly.

---
 arch/arm64/boot/dts/renesas/rzt2h-n2h-evk-common.dtsi | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

index 5c91002c99c48..5384a43837c1d 100644
@@ -154,8 +154,7 @@ data-pins {
 		ctrl-pins {
 			pinmux = <RZT2H_PORT_PINMUX(12, 0, 0x29)>, /* SD0_CLK */
 				 <RZT2H_PORT_PINMUX(12, 1, 0x29)>, /* SD0_CMD */
-				 <RZT2H_PORT_PINMUX(22, 5, 0x29)>, /* SD0_CD */
-				 <RZT2H_PORT_PINMUX(22, 6, 0x29)>; /* SD0_WP */
+				 <RZT2H_PORT_PINMUX(22, 5, 0x29)>; /* SD0_CD */
 		};
 	};
 

@@ -212,6 +211,7 @@ &sdhi0 {
 	pinctrl-names = "default", "state_uhs";
 	vmmc-supply = <&reg_3p3v>;
 	vqmmc-supply = <&vqmmc_sdhi0>;
+	wp-gpios = <&pinctrl RZT2H_GPIO(22, 6) GPIO_ACTIVE_HIGH>;
 	bus-width = <4>;
 	sd-uhs-sdr50;
 	sd-uhs-sdr104;

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Vladimir Zapolskiy <[email protected]>

[ Upstream commit 71630e581a0e34c03757f5c1706f57c853b92555 ]

Motor Control PWM depends on its own supply clock, the clock gate control is present in TIMCLK_CTRL1 register.

---
 arch/arm/boot/dts/nxp/lpc/lpc32xx.dtsi | 1 +
 1 file changed, 1 insertion(+)

index 2236901a00313..8e9ed93da129e 100644
@@ -302,6 +302,7 @@ i2c2: i2c@400a8000 {
 			mpwm: pwm@400e8000 {
 				compatible = "nxp,lpc3220-motor-pwm";
 				reg = <0x400e8000 0x78>;
+				clocks = <&clk LPC32XX_CLK_MCPWM>;
 				#pwm-cells = <3>;
 				status = "disabled";
 			};

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: AngeloGioacchino Del Regno <[email protected]>

[ Upstream commit b1fc81a986c9b8089db31e21a372cc8b6514e900 ]

Rename "piins-bt-wakeup" to "pins-bt-wakeup" to fix a dtbs_check warning happening due to this typo.

---
 arch/arm64/boot/dts/mediatek/mt8183-kukui-jacuzzi-pico6.dts | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

index cce326aec1aa5..40af5656d6f15 100644
@@ -91,7 +91,7 @@ bluetooth@2 {
 
 &pio {
 	bt_pins_wakeup: bt-pins-wakeup {
-		piins-bt-wakeup {
+		pins-bt-wakeup {
 			pinmux = <PINMUX_GPIO42__FUNC_GPIO42>;
 			input-enable;
 		};

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Jerome Brunet <[email protected]>

[ Upstream commit 86124a8becb43eed3103f2459399daee8af2c99d ]

The amlogic MMC driver operate with the assumption that MMC clock is configured to provide 24MHz. It uses this path for low rates such as 400kHz.

This assumption did hold true until but it now, but it is apparently not the case with s4. The clock has been reported to provide 1GHz instead. This is most likely due to how the bootloader is using the MMC clock on this platform.

Regardless of why the MMC clock rate is 1GHz, if the MMC driver expects 24MHz, the clock should be properly assigned, so assign it.

---
 arch/arm64/boot/dts/amlogic/meson-s4.dtsi | 3 +++
 1 file changed, 3 insertions(+)

index 9d99ed2994dfa..f314f07062abe 100644
@@ -838,6 +838,9 @@ sd: mmc@fe08a000 {
 			clock-names = "core", "clkin0", "clkin1";
 			resets = <&reset RESET_SD_EMMC_B>;
 			status = "disabled";
+
+			assigned-clocks = <&clkc_periphs CLKID_SD_EMMC_B>;
+			assigned-clock-rates = <24000000>;
 		};
 
 		emmc: mmc@fe08c000 {

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Jerome Brunet <[email protected]>

[ Upstream commit 3a115d42922cffc91b303992eadf220111d66c31 ]

MMC A and C are mis-represented as having their "clkin0" input connected to xtal while it is actually connected to the MMC clock, probably in an attempt to provide 24MHz to the device on this input.

Fix this and assign the clock to 24MHz to actually provide the required rate.

---
 arch/arm64/boot/dts/amlogic/meson-s4.dtsi | 10 ++++++++--
 1 file changed, 8 insertions(+), 2 deletions(-)

index f314f07062abe..dfc0a30a6e61b 100644
@@ -819,13 +819,16 @@ sdio: mmc@fe088000 {
 			reg = <0x0 0xfe088000 0x0 0x800>;
 			interrupts = <GIC_SPI 176 IRQ_TYPE_LEVEL_HIGH>;
 			clocks = <&clkc_periphs CLKID_SDEMMC_A>,
-				 <&xtal>,
+				 <&clkc_periphs CLKID_SD_EMMC_A>,
 				 <&clkc_pll CLKID_FCLK_DIV2>;
 			clock-names = "core", "clkin0", "clkin1";
 			resets = <&reset RESET_SD_EMMC_A>;
 			cap-sdio-irq;
 			keep-power-in-suspend;
 			status = "disabled";
+
+			assigned-clocks = <&clkc_periphs CLKID_SD_EMMC_A>;
+			assigned-clock-rates = <24000000>;
 		};
 
 		sd: mmc@fe08a000 {

@@ -848,13 +851,16 @@ emmc: mmc@fe08c000 {
 			reg = <0x0 0xfe08c000 0x0 0x800>;
 			interrupts = <GIC_SPI 178 IRQ_TYPE_EDGE_RISING>;
 			clocks = <&clkc_periphs CLKID_NAND>,
-				 <&xtal>,
+				 <&clkc_periphs CLKID_SD_EMMC_C>,
 				 <&clkc_pll CLKID_FCLK_DIV2>;
 			clock-names = "core", "clkin0", "clkin1";
 			resets = <&reset RESET_NAND_EMMC>;
 			no-sdio;
 			no-sd;
 			status = "disabled";
+
+			assigned-clocks = <&clkc_periphs CLKID_SD_EMMC_C>;
+			assigned-clock-rates = <24000000>;
 		};
 	};
 };

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Abhash Kumar Jha <[email protected]>

[ Upstream commit 24c9d5fb8bbf5e8c9e6fc2beffeb80ac2da83de4 ]

The device tree nodes should be ordered by unit addresses in ascending order.

Correct the order by moving the c71_3 DSP node at the end as it has a higher unit address.

Stable-dep-of: 61acc4428a7f ("arm64: dts: ti: k3-j784s4-j742s2-main-common.dtsi: Refactor watchdog instances for j784s4")

---
 arch/arm64/boot/dts/ti/k3-j784s4-main.dtsi | 26 +++++++++++-----------
 1 file changed, 13 insertions(+), 13 deletions(-)

index 0160fe0da9838..5b7830a3c0975 100644
@@ -6,19 +6,6 @@
  */
 
 &cbass_main {
-	c71_3: dsp@67800000 {
-		compatible = "ti,j721s2-c71-dsp";
-		reg = <0x00 0x67800000 0x00 0x00080000>,
-		      <0x00 0x67e00000 0x00 0x0000c000>;
-		reg-names = "l2sram", "l1dram";
-		resets = <&k3_reset 40 1>;
-		firmware-name = "j784s4-c71_3-fw";
-		ti,sci = <&sms>;
-		ti,sci-dev-id = <40>;
-		ti,sci-proc-ids = <0x33 0xff>;
-		status = "disabled";
-	};
-
 	pcie2_rc: pcie@2920000 {
 		compatible = "ti,j784s4-pcie-host";
 		reg = <0x00 0x02920000 0x00 0x1000>,

@@ -113,6 +100,19 @@ serdes2: serdes@5020000 {
 			status = "disabled";
 		};
 	};
+
+	c71_3: dsp@67800000 {
+		compatible = "ti,j721s2-c71-dsp";
+		reg = <0x00 0x67800000 0x00 0x00080000>,
+		      <0x00 0x67e00000 0x00 0x0000c000>;
+		reg-names = "l2sram", "l1dram";
+		resets = <&k3_reset 40 1>;
+		firmware-name = "j784s4-c71_3-fw";
+		ti,sci = <&sms>;
+		ti,sci-dev-id = <40>;
+		ti,sci-proc-ids = <0x33 0xff>;
+		status = "disabled";
+	};
 };
 
 &scm_conf {

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Abhash Kumar Jha <[email protected]>

[ Upstream commit 61acc4428a7f52e0a13e226ba76f2ce2ca66c065 ]

Each A72 core has one watchdog instance associated with it. Since j742s2 has 4 A72 cores, the common file should not define 8 watchdog instances.

Refactor the last 4 extra watchdogs from the common file to j784s4 specific file, as j784s4 has 8 A72 cores and thus hardware description requires 8 watchdog instances.

---
 .../dts/ti/k3-j784s4-j742s2-main-common.dtsi  | 36 -------------------
 arch/arm64/boot/dts/ti/k3-j784s4-main.dtsi    | 36 +++++++++++++++++++
 2 files changed, 36 insertions(+), 36 deletions(-)

index 9cc0901d58fbf..c2636e624f18b 100644
@@ -2378,42 +2378,6 @@ watchdog3: watchdog@2230000 {
 		assigned-clock-parents = <&k3_clks 351 4>;
 	};
 
-	watchdog4: watchdog@2240000 {
-		compatible = "ti,j7-rti-wdt";
-		reg = <0x00 0x2240000 0x00 0x100>;
-		clocks = <&k3_clks 352 0>;
-		power-domains = <&k3_pds 352 TI_SCI_PD_EXCLUSIVE>;
-		assigned-clocks = <&k3_clks 352 0>;
-		assigned-clock-parents = <&k3_clks 352 4>;
-	};
-
-	watchdog5: watchdog@2250000 {
-		compatible = "ti,j7-rti-wdt";
-		reg = <0x00 0x2250000 0x00 0x100>;
-		clocks = <&k3_clks 353 0>;
-		power-domains = <&k3_pds 353 TI_SCI_PD_EXCLUSIVE>;
-		assigned-clocks = <&k3_clks 353 0>;
-		assigned-clock-parents = <&k3_clks 353 4>;
-	};
-
-	watchdog6: watchdog@2260000 {
-		compatible = "ti,j7-rti-wdt";
-		reg = <0x00 0x2260000 0x00 0x100>;
-		clocks = <&k3_clks 354 0>;
-		power-domains = <&k3_pds 354 TI_SCI_PD_EXCLUSIVE>;
-		assigned-clocks = <&k3_clks 354 0>;
-		assigned-clock-parents = <&k3_clks 354 4>;
-	};
-
-	watchdog7: watchdog@2270000 {
-		compatible = "ti,j7-rti-wdt";
-		reg = <0x00 0x2270000 0x00 0x100>;
-		clocks = <&k3_clks 355 0>;
-		power-domains = <&k3_pds 355 TI_SCI_PD_EXCLUSIVE>;
-		assigned-clocks = <&k3_clks 355 0>;
-		assigned-clock-parents = <&k3_clks 355 4>;
-	};
-
 	/*
 	 * The following RTI instances are coupled with MCU R5Fs, c7x and
 	 * GPU so keeping them reserved as these will be used by their

index 5b7830a3c0975..78fcd0c40abcf 100644
@@ -6,6 +6,42 @@
  */
 
 &cbass_main {
+	watchdog4: watchdog@2240000 {
+		compatible = "ti,j7-rti-wdt";
+		reg = <0x00 0x2240000 0x00 0x100>;
+		clocks = <&k3_clks 352 0>;
+		power-domains = <&k3_pds 352 TI_SCI_PD_EXCLUSIVE>;
+		assigned-clocks = <&k3_clks 352 0>;
+		assigned-clock-parents = <&k3_clks 352 4>;
+	};
+
+	watchdog5: watchdog@2250000 {
+		compatible = "ti,j7-rti-wdt";
+		reg = <0x00 0x2250000 0x00 0x100>;
+		clocks = <&k3_clks 353 0>;
+		power-domains = <&k3_pds 353 TI_SCI_PD_EXCLUSIVE>;
+		assigned-clocks = <&k3_clks 353 0>;
+		assigned-clock-parents = <&k3_clks 353 4>;
+	};
+
+	watchdog6: watchdog@2260000 {
+		compatible = "ti,j7-rti-wdt";
+		reg = <0x00 0x2260000 0x00 0x100>;
+		clocks = <&k3_clks 354 0>;
+		power-domains = <&k3_pds 354 TI_SCI_PD_EXCLUSIVE>;
+		assigned-clocks = <&k3_clks 354 0>;
+		assigned-clock-parents = <&k3_clks 354 4>;
+	};
+
+	watchdog7: watchdog@2270000 {
+		compatible = "ti,j7-rti-wdt";
+		reg = <0x00 0x2270000 0x00 0x100>;
+		clocks = <&k3_clks 355 0>;
+		power-domains = <&k3_pds 355 TI_SCI_PD_EXCLUSIVE>;
+		assigned-clocks = <&k3_clks 355 0>;
+		assigned-clock-parents = <&k3_clks 355 4>;
+	};
+
 	pcie2_rc: pcie@2920000 {
 		compatible = "ti,j784s4-pcie-host";
 		reg = <0x00 0x02920000 0x00 0x1000>,

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Dmitry Baryshkov <[email protected]>

[ Upstream commit ccef4b2703ff5b0de0b1bda30a0de3026d52eb19 ]

The header has a function which calls pr_err(). Don't require users of the header to include <linux/printk.h> and include it here.

---
 include/linux/soc/qcom/ubwc.h | 1 +
 1 file changed, 1 insertion(+)

index 1ed8b1b16bc90..d9dfc9edc1b2f 100644
@@ -8,6 +8,7 @@
 #define __QCOM_UBWC_H__
 
 #include <linux/bits.h>
+#include <linux/printk.h>
 #include <linux/types.h>
 
 struct qcom_ubwc_cfg_data {

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Haotian Zhang <[email protected]>

[ Upstream commit 3bd4edd67b034f8e1f61c86e0eb098de6179e3f2 ]

Although unlikely, devm_pm_runtime_enable() can fail due to memory allocations. Without proper error handling, the subsequent pm_runtime_resume_and_get() call may operate on incorrectly initialized runtime PM state.

Add error handling to check the return value of
devm_pm_runtime_enable() and return on failure.

---
 drivers/hwspinlock/omap_hwspinlock.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

index 27b47b8623c09..2d8de835bc242 100644
@@ -88,7 +88,9 @@ static int omap_hwspinlock_probe(struct platform_device *pdev)
 	 * make sure the module is enabled and clocked before reading
 	 * the module SYSSTATUS register
 	 */
-	devm_pm_runtime_enable(&pdev->dev);
+	ret = devm_pm_runtime_enable(&pdev->dev);
+	if (ret)
+		return ret;
 	ret = pm_runtime_resume_and_get(&pdev->dev);
 	if (ret < 0)
 		return ret;

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Jerome Brunet <[email protected]>

[ Upstream commit 69330fd2368371c4eb47d60ace6bca09763d24a0 ]

The amlogic MMC driver operate with the assumption that MMC clock is configured to provide 24MHz. It uses this path for low rates such as 400kHz.

Assign the clocks to make sure they are properly configured

---
 arch/arm64/boot/dts/amlogic/amlogic-c3.dtsi | 7 +++++++
 1 file changed, 7 insertions(+)

index 07aaaf71ea9ae..f226df3ce153b 100644
@@ -969,6 +969,10 @@ sdio: mmc@88000 {
 				no-sd;
 				resets = <&reset RESET_SD_EMMC_A>;
 				status = "disabled";
+
+				assigned-clocks = <&clkc_periphs CLKID_SD_EMMC_A>;
+				assigned-clock-rates = <24000000>;
+
 			};
 
 			sd: mmc@8a000 {

@@ -984,6 +988,9 @@ sd: mmc@8a000 {
 				no-sdio;
 				resets = <&reset RESET_SD_EMMC_B>;
 				status = "disabled";
+
+				assigned-clocks = <&clkc_periphs CLKID_SD_EMMC_B>;
+				assigned-clock-rates = <24000000>;
 			};
 
 			nand: nand-controller@8d000 {

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Jerome Brunet <[email protected]>

[ Upstream commit 13d3fe2318ef6e46d6fcfe13bc373827fdf2aeac ]

The amlogic MMC driver operate with the assumption that MMC clock is configured to provide 24MHz. It uses this path for low rates such as 400kHz.

Assign the clocks to make sure they are properly configured

---
 arch/arm64/boot/dts/amlogic/meson-axg.dtsi | 6 ++++++
 1 file changed, 6 insertions(+)

index 04fb130ac7c6a..bbf94a1f92a10 100644
@@ -1960,6 +1960,9 @@ sd_emmc_b: mmc@5000 {
 					<&clkc CLKID_FCLK_DIV2>;
 				clock-names = "core", "clkin0", "clkin1";
 				resets = <&reset RESET_SD_EMMC_B>;
+
+				assigned-clocks = <&clkc CLKID_SD_EMMC_B_CLK0>;
+				assigned-clock-rates = <24000000>;
 			};
 
 			sd_emmc_c: mmc@7000 {

@@ -1972,6 +1975,9 @@ sd_emmc_c: mmc@7000 {
 					<&clkc CLKID_FCLK_DIV2>;
 				clock-names = "core", "clkin0", "clkin1";
 				resets = <&reset RESET_SD_EMMC_C>;
+
+				assigned-clocks = <&clkc CLKID_SD_EMMC_C_CLK0>;
+				assigned-clock-rates = <24000000>;
 			};
 
 			nfc: nand-controller@7800 {

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Jerome Brunet <[email protected]>

[ Upstream commit 406706559046eebc09a31e8ae5e78620bfd746fe ]

The amlogic MMC driver operate with the assumption that MMC clock is configured to provide 24MHz. It uses this path for low rates such as 400kHz.

Assign the clocks to make sure they are properly configured

---
 arch/arm64/boot/dts/amlogic/meson-gxbb.dtsi | 9 +++++++++
 arch/arm64/boot/dts/amlogic/meson-gxl.dtsi  | 9 +++++++++
 2 files changed, 18 insertions(+)

index f69923da07feb..a9c830a570cc6 100644
@@ -824,6 +824,9 @@ &sd_emmc_a {
 		 <&clkc CLKID_FCLK_DIV2>;
 	clock-names = "core", "clkin0", "clkin1";
 	resets = <&reset RESET_SD_EMMC_A>;
+
+	assigned-clocks = <&clkc CLKID_SD_EMMC_A_CLK0>;
+	assigned-clock-rates = <24000000>;
 };
 
 &sd_emmc_b {

@@ -832,6 +835,9 @@ &sd_emmc_b {
 		 <&clkc CLKID_FCLK_DIV2>;
 	clock-names = "core", "clkin0", "clkin1";
 	resets = <&reset RESET_SD_EMMC_B>;
+
+	assigned-clocks = <&clkc CLKID_SD_EMMC_B_CLK0>;
+	assigned-clock-rates = <24000000>;
 };
 
 &sd_emmc_c {

@@ -840,6 +846,9 @@ &sd_emmc_c {
 		 <&clkc CLKID_FCLK_DIV2>;
 	clock-names = "core", "clkin0", "clkin1";
 	resets = <&reset RESET_SD_EMMC_C>;
+
+	assigned-clocks = <&clkc CLKID_SD_EMMC_C_CLK0>;
+	assigned-clock-rates = <24000000>;
 };
 
 &simplefb_hdmi {

index ba535010a3c91..e202d84f06720 100644
@@ -894,6 +894,9 @@ &sd_emmc_a {
 		 <&clkc CLKID_FCLK_DIV2>;
 	clock-names = "core", "clkin0", "clkin1";
 	resets = <&reset RESET_SD_EMMC_A>;
+
+	assigned-clocks = <&clkc CLKID_SD_EMMC_A_CLK0>;
+	assigned-clock-rates = <24000000>;
 };
 
 &sd_emmc_b {

@@ -902,6 +905,9 @@ &sd_emmc_b {
 		 <&clkc CLKID_FCLK_DIV2>;
 	clock-names = "core", "clkin0", "clkin1";
 	resets = <&reset RESET_SD_EMMC_B>;
+
+	assigned-clocks = <&clkc CLKID_SD_EMMC_B_CLK0>;
+	assigned-clock-rates = <24000000>;
 };
 
 &sd_emmc_c {

@@ -910,6 +916,9 @@ &sd_emmc_c {
 		 <&clkc CLKID_FCLK_DIV2>;
 	clock-names = "core", "clkin0", "clkin1";
 	resets = <&reset RESET_SD_EMMC_C>;
+
+	assigned-clocks = <&clkc CLKID_SD_EMMC_C_CLK0>;
+	assigned-clock-rates = <24000000>;
 };
 
 &simplefb_hdmi {

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Jerome Brunet <[email protected]>

[ Upstream commit be2ff5fdb0e83e32d4ec4e68a69875cec0d14621 ]

The amlogic MMC driver operate with the assumption that MMC clock is configured to provide 24MHz. It uses this path for low rates such as 400kHz.

Assign the clocks to make sure they are properly configured

---
 arch/arm64/boot/dts/amlogic/meson-g12-common.dtsi | 6 ++++++
 1 file changed, 6 insertions(+)

index dcc927a9da802..6724405eaa6f5 100644
@@ -2443,6 +2443,9 @@ sd_emmc_b: mmc@ffe05000 {
 				 <&clkc CLKID_FCLK_DIV2>;
 			clock-names = "core", "clkin0", "clkin1";
 			resets = <&reset RESET_SD_EMMC_B>;
+
+			assigned-clocks = <&clkc CLKID_SD_EMMC_B_CLK0>;
+			assigned-clock-rates = <24000000>;
 		};
 
 		sd_emmc_c: mmc@ffe07000 {

@@ -2455,6 +2458,9 @@ sd_emmc_c: mmc@ffe07000 {
 				 <&clkc CLKID_FCLK_DIV2>;
 			clock-names = "core", "clkin0", "clkin1";
 			resets = <&reset RESET_SD_EMMC_C>;
+
+			assigned-clocks = <&clkc CLKID_SD_EMMC_C_CLK0>;
+			assigned-clock-rates = <24000000>;
 		};
 
 		usb: usb@ffe09000 {

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Jerome Brunet <[email protected]>

[ Upstream commit 3c941feaa363f1573a501452391ddf513394c84b ]

The amlogic MMC driver operate with the assumption that MMC clock is configured to provide 24MHz. It uses this path for low rates such as 400kHz.

Assign the clock to make sure it is properly configured

---
 arch/arm64/boot/dts/amlogic/meson-g12-common.dtsi | 3 +++
 1 file changed, 3 insertions(+)

index 6724405eaa6f5..8d8ab775404d9 100644
@@ -2431,6 +2431,9 @@ sd_emmc_a: mmc@ffe03000 {
 				 <&clkc CLKID_FCLK_DIV2>;
 			clock-names = "core", "clkin0", "clkin1";
 			resets = <&reset RESET_SD_EMMC_A>;
+
+			assigned-clocks = <&clkc CLKID_SD_EMMC_A_CLK0>;
+			assigned-clock-rates = <24000000>;
 		};
 
 		sd_emmc_b: mmc@ffe05000 {

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Dmitry Baryshkov <[email protected]>

[ Upstream commit c5dc4812f6bf397b82290c540085e9ec98b47b30 ]

Follow commit 9c92d36b0b1e ("arm64: dts: qcom: qrb2210-rb1: Fix UART3 wakeup IRQ storm") and apply the similar fix to the RB2 platform.

Having RX / TX pins as pull up and wakup interrupt as high-level triggered generates an interrupt storm when trying to suspend the device. Avoid the storm by using the falling edge trigger (as all other platforms do).

---
 arch/arm64/boot/dts/qcom/qrb4210-rb2.dts | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

index bdf2d66e40c62..44ca3e61c33d4 100644
@@ -694,7 +694,7 @@ sdc2_card_det_n: sd-card-det-n-state {
 
 &uart3 {
 	interrupts-extended = <&intc GIC_SPI 330 IRQ_TYPE_LEVEL_HIGH>,
-			      <&tlmm 11 IRQ_TYPE_LEVEL_HIGH>;
+			      <&tlmm 11 IRQ_TYPE_EDGE_FALLING>;
 	pinctrl-0 = <&uart3_default>;
 	pinctrl-1 = <&uart3_sleep>;
 	pinctrl-names = "default", "sleep";

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Dmitry Baryshkov <[email protected]>

[ Upstream commit 8bfb696ccdc5bcfad7a45b84c2c8a36757070e19 ]

On SDM845 SPI uses hardware-provided chip select, while specifying cs-gpio makes the driver request GPIO pin, which on DB845c conflicts with the normal host controllers pinctrl entry.

Drop the cs-gpios property to restore SPI functionality.

---
 arch/arm64/boot/dts/qcom/sdm845-db845c.dts | 1 -
 1 file changed, 1 deletion(-)

index 8abf3e909502f..384be2f8b1411 100644
@@ -850,7 +850,6 @@ &spi0 {
 	status = "okay";
 	pinctrl-names = "default";
 	pinctrl-0 = <&qup_spi0_default>;
-	cs-gpios = <&tlmm 3 GPIO_ACTIVE_LOW>;
 
 	can@0 {
 		compatible = "microchip,mcp2517fd";

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Dmitry Baryshkov <[email protected]>

[ Upstream commit c303e89f7f17c29981d09f8beaaf60937ae8b1f2 ]

Specify power supply for the second chain / antenna output of the onboard WiFi chip.

---
 arch/arm64/boot/dts/qcom/sdm845-db845c.dts | 7 +++++++
 1 file changed, 7 insertions(+)

index 384be2f8b1411..5147d6d3cc26b 100644
@@ -379,6 +379,12 @@ vreg_l21a_2p95: ldo21 {
 			regulator-initial-mode = <RPMH_REGULATOR_MODE_HPM>;
 		};
 
+		vreg_l23a_3p3: ldo23 {
+			regulator-min-microvolt = <3300000>;
+			regulator-max-microvolt = <3312000>;
+			regulator-initial-mode = <RPMH_REGULATOR_MODE_HPM>;
+		};
+
 		vreg_l24a_3p075: ldo24 {
 			regulator-min-microvolt = <3088000>;
 			regulator-max-microvolt = <3088000>;

@@ -1155,6 +1161,7 @@ &wifi {
 	vdd-1.8-xo-supply = <&vreg_l7a_1p8>;
 	vdd-1.3-rfa-supply = <&vreg_l17a_1p3>;
 	vdd-3.3-ch0-supply = <&vreg_l25a_3p3>;
+	vdd-3.3-ch1-supply = <&vreg_l23a_3p3>;
 
 	qcom,snoc-host-cap-8bit-quirk;
 	qcom,calibration-variant = "Thundercomm_DB845C";

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Jonathan Marek <[email protected]>

[ Upstream commit b38dd256e11a4c8bd5a893e11fc42d493939c907 ]

Unlike the phone SoCs this was copied from, x1e has a 40-bit physical bus. The upper address space is used to support more than 32GB of memory.

This fixes issues when DMA buffers are allocated outside the 36-bit range.

---
 arch/arm64/boot/dts/qcom/x1e80100.dtsi | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

index 3290fd8c2d6e6..512a75da4f13f 100644
@@ -791,8 +791,8 @@ soc: soc@0 {
 
 		#address-cells = <2>;
 		#size-cells = <2>;
-		dma-ranges = <0 0 0 0 0x10 0>;
-		ranges = <0 0 0 0 0x10 0>;
+		dma-ranges = <0 0 0 0 0x100 0>;
+		ranges = <0 0 0 0 0x100 0>;
 
 		gcc: clock-controller@100000 {
 			compatible = "qcom,x1e80100-gcc";

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Viken Dadhaniya <[email protected]>

[ Upstream commit dda4bdd325326dd67ae4401f4f3d35b9cf781e3f ]

QUP devices are currently marked with opp-shared in their OPP table, causing the kernel to treat them as part of a shared OPP domain. This leads to the qcom_geni_serial driver failing to probe with error -EBUSY (-16).

Remove the opp-shared property to ensure the OPP framework treats the QUP OPP table as device-specific, allowing the serial driver to probe successfully

---
 arch/arm64/boot/dts/qcom/sm6150.dtsi | 1 -
 1 file changed, 1 deletion(-)

index 64e7c9dbafc70..363b9f436cd0a 100644
@@ -398,7 +398,6 @@ cdsp_smp2p_in: slave-kernel {
 
 	qup_opp_table: opp-table-qup {
 		compatible = "operating-points-v2";
-		opp-shared;
 
 		opp-75000000 {
 			opp-hz = /bits/ 64 <75000000>;

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Konrad Dybcio <[email protected]>

[ Upstream commit 0fdcc948929a6d673bd0f90631dd6e42090c3dbd ]

Describe the GPU register regions, with the former existing but not being used much if at all on this silicon, and the latter containing various debugging levers generally related to dumping the state of the IP upon a crash.

---
 arch/arm64/boot/dts/qcom/qcm2290.dtsi | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

index 3b0ba590ee825..e0e400fdd2497 100644
@@ -1503,8 +1503,12 @@ usb_dwc3_ss: endpoint {
 
 		gpu: gpu@5900000 {
 			compatible = "qcom,adreno-07000200", "qcom,adreno";
-			reg = <0x0 0x05900000 0x0 0x40000>;
-			reg-names = "kgsl_3d0_reg_memory";
+			reg = <0x0 0x05900000 0x0 0x40000>,
+			      <0x0 0x0599e000 0x0 0x1000>,
+			      <0x0 0x05961000 0x0 0x800>;
+			reg-names = "kgsl_3d0_reg_memory",
+				    "cx_mem",
+				    "cx_dbgc";
 
 			interrupts = <GIC_SPI 177 IRQ_TYPE_LEVEL_HIGH>;
 

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Konrad Dybcio <[email protected]>

[ Upstream commit 78c13dac18cf0e6f6cbc6ea85d4f967e6cca9562 ]

Describe the GPU register regions, with the former existing but not being used much if at all on this silicon, and the latter containing various debugging levers generally related to dumping the state of the IP upon a crash.

---
 arch/arm64/boot/dts/qcom/sm6115.dtsi | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

index 91fc36b59abf9..8b8395f6a2dfc 100644
@@ -1715,8 +1715,12 @@ usb_dwc3_ss: endpoint {
 
 		gpu: gpu@5900000 {
 			compatible = "qcom,adreno-610.0", "qcom,adreno";
-			reg = <0x0 0x05900000 0x0 0x40000>;
-			reg-names = "kgsl_3d0_reg_memory";
+			reg = <0x0 0x05900000 0x0 0x40000>,
+			      <0x0 0x0599e000 0x0 0x1000>,
+			      <0x0 0x05961000 0x0 0x800>;
+			reg-names = "kgsl_3d0_reg_memory",
+				    "cx_mem",
+				    "cx_dbgc";
 
 			/* There's no (real) GMU, so we have to handle quite a bunch of clocks! */
 			clocks = <&gpucc GPU_CC_GX_GFX3D_CLK>,

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Junhui Liu <[email protected]>

[ Upstream commit c7a5e01e229d21e0560d78bd645b4f7398667ce4 ]

The driver doesn't use any symbols depending on CONFIG_OF, so drop the dependency. Also, enable it by default when ARCH_CANAAN is selected.

---
 drivers/reset/Kconfig | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

index 78b7078478d46..b3b9e0f9d8c48 100644
@@ -150,7 +150,7 @@ config RESET_K210
 config RESET_K230
 	tristate "Reset controller driver for Canaan Kendryte K230 SoC"
 	depends on ARCH_CANAAN || COMPILE_TEST
-	depends on OF
+	default ARCH_CANAAN
 	help
 	  Support for the Canaan Kendryte K230 RISC-V SoC reset controller.
 	  Say Y if you want to control reset signals provided by this

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Boris Brezillon <[email protected]>

[ Upstream commit 3c0a60195b37af83bbbaf223cd3a78945bace49e ]

We have seen a few cases where the whole memory subsystem is blocked and flush operations never complete. When that happens, we want to:

- schedule a reset, so we can recover from this situation
- in the reset path, we need to reset the pending_reqs so we can send new commands after the reset
- if more panthor_gpu_flush_caches() operations are queued after the timeout, we skip them and return -EIO directly to avoid needless waits (the memory block won't miraculously work again)

Note that we drop the WARN_ON()s because these hangs can be triggered with buggy GPU jobs created by the UMD, and there's no way we can prevent it. We do keep the error messages though.

v2:
- New patch

v3:
- Collect R-b
- Explicitly mention the fact we dropped the WARN_ON()s in the commit message

v4:
- No changes

---
 drivers/gpu/drm/panthor/panthor_gpu.c | 19 ++++++++++++-------
 1 file changed, 12 insertions(+), 7 deletions(-)

index db69449a5be09..b92c9e738dbfa 100644
@@ -259,38 +259,42 @@ int panthor_gpu_l2_power_on(struct panthor_device *ptdev)
 int panthor_gpu_flush_caches(struct panthor_device *ptdev,
 			     u32 l2, u32 lsc, u32 other)
 {
-	bool timedout = false;
 	unsigned long flags;
+	int ret = 0;
 
 	/* Serialize cache flush operations. */
 	guard(mutex)(&ptdev->gpu->cache_flush_lock);
 
 	spin_lock_irqsave(&ptdev->gpu->reqs_lock, flags);
-	if (!drm_WARN_ON(&ptdev->base,
-			 ptdev->gpu->pending_reqs & GPU_IRQ_CLEAN_CACHES_COMPLETED)) {
+	if (!(ptdev->gpu->pending_reqs & GPU_IRQ_CLEAN_CACHES_COMPLETED)) {
 		ptdev->gpu->pending_reqs |= GPU_IRQ_CLEAN_CACHES_COMPLETED;
 		gpu_write(ptdev, GPU_CMD, GPU_FLUSH_CACHES(l2, lsc, other));
+	} else {
+		ret = -EIO;
 	}
 	spin_unlock_irqrestore(&ptdev->gpu->reqs_lock, flags);
 
+	if (ret)
+		return ret;
+
 	if (!wait_event_timeout(ptdev->gpu->reqs_acked,
 				!(ptdev->gpu->pending_reqs & GPU_IRQ_CLEAN_CACHES_COMPLETED),
 				msecs_to_jiffies(100))) {
 		spin_lock_irqsave(&ptdev->gpu->reqs_lock, flags);
 		if ((ptdev->gpu->pending_reqs & GPU_IRQ_CLEAN_CACHES_COMPLETED) != 0 &&
 		    !(gpu_read(ptdev, GPU_INT_RAWSTAT) & GPU_IRQ_CLEAN_CACHES_COMPLETED))
-			timedout = true;
+			ret = -ETIMEDOUT;
 		else
 			ptdev->gpu->pending_reqs &= ~GPU_IRQ_CLEAN_CACHES_COMPLETED;
 		spin_unlock_irqrestore(&ptdev->gpu->reqs_lock, flags);
 	}
 
-	if (timedout) {
+	if (ret) {
+		panthor_device_schedule_reset(ptdev);
 		drm_err(&ptdev->base, "Flush caches timeout");
-		return -ETIMEDOUT;
 	}
 
-	return 0;
+	return ret;
 }
 
 /**

@@ -330,6 +334,7 @@ int panthor_gpu_soft_reset(struct panthor_device *ptdev)
 		return -ETIMEDOUT;
 	}
 
+	ptdev->gpu->pending_reqs = 0;
 	return 0;
 }
 

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Boris Brezillon <[email protected]>

[ Upstream commit a3c2d0b40b108bd45d44f6c1dfa33c39d577adcd ]

We have a full tick when the remaining time to the next tick is zero, not the other way around. Declare a full_tick variable so we don't get that test wrong in other places.

v2:
- Add R-b

v3:
- Collect R-b

---
 drivers/gpu/drm/panthor/panthor_sched.c | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

index 881a07ffbabc8..6e4667cbe1b82 100644
@@ -2387,6 +2387,7 @@ static void tick_work(struct work_struct *work)
 	u64 remaining_jiffies = 0, resched_delay;
 	u64 now = get_jiffies_64();
 	int prio, ret, cookie;
+	bool full_tick;
 
 	if (!drm_dev_enter(&ptdev->base, &cookie))
 		return;

@@ -2398,15 +2399,17 @@ static void tick_work(struct work_struct *work)
 	if (time_before64(now, sched->resched_target))
 		remaining_jiffies = sched->resched_target - now;
 
+	full_tick = remaining_jiffies == 0;
+
 	mutex_lock(&sched->lock);
 	if (panthor_device_reset_is_pending(sched->ptdev))
 		goto out_unlock;
 
-	tick_ctx_init(sched, &ctx, remaining_jiffies != 0);
+	tick_ctx_init(sched, &ctx, full_tick);
 	if (ctx.csg_upd_failed_mask)
 		goto out_cleanup_ctx;
 
-	if (remaining_jiffies) {
+	if (!full_tick) {
 		/* Scheduling forced in the middle of a tick. Only RT groups
 		 * can preempt non-RT ones. Currently running RT groups can't be
 		 * preempted.

2.51.0

6.18-stable review patch. If anyone has any objections, please let me know.

------------------

From: Boris Brezillon <[email protected]>

[ Upstream commit 55429c51d5db3db24c2ad561944c6a0ca922d476 ]

When rotating group priorities, we want the group with the highest priority to go back to the end of the queue, and all other active groups to get their priority bumped, otherwise some groups will never get a chance to run with the highest priority. This implies moving the rotation itself to tick_work(), and only dealing with old group ordering in tick_ctx_insert_old_group().

v2:
- Add R-b
- Fix the commit message

v3:
- Drop the full_tick argument in tick_ctx_init()
- Collect R-b

---
 drivers/gpu/drm/panthor/panthor_sched.c | 52 +++++++++++++++----------
 1 file changed, 31 insertions(+), 21 deletions(-)

index 6e4667cbe1b82..03062169b8d33 100644
@@ -1989,31 +1989,22 @@ tick_ctx_pick_groups_from_list(const struct panthor_scheduler *sched,
 static void
 tick_ctx_insert_old_group(struct panthor_scheduler *sched,
 			  struct panthor_sched_tick_ctx *ctx,
-			  struct panthor_group *group,
-			  bool full_tick)
+			  struct panthor_group *group)
 {
 	struct panthor_csg_slot *csg_slot = &sched->csg_slots[group->csg_id];
 	struct panthor_group *other_group;
 
-	if (!full_tick) {
-		list_add_tail(&group->run_node, &ctx->old_groups[group->priority]);
-		return;
-	}
-
-	/* Rotate to make sure groups with lower CSG slot
-	 * priorities have a chance to get a higher CSG slot
-	 * priority next time they get picked. This priority
-	 * has an impact on resource request ordering, so it's
-	 * important to make sure we don't let one group starve
-	 * all other groups with the same group priority.
-	 */
+	/* Class groups in descending priority order so we can easily rotate. */
 	list_for_each_entry(other_group,
 			    &ctx->old_groups[csg_slot->group->priority],
 			    run_node) {
 		struct panthor_csg_slot *other_csg_slot = &sched->csg_slots[other_group->csg_id];
 
-		if (other_csg_slot->priority > csg_slot->priority) {
-			list_add_tail(&csg_slot->group->run_node, &other_group->run_node);
+		/* Our group has a higher prio than the one we're testing against,
+		 * place it just before.
+		 */
+		if (csg_slot->priority > other_csg_slot->priority) {
+			list_add_tail(&group->run_node, &other_group->run_node);
 			return;
 		}
 	}

@@ -2023,8 +2014,7 @@ tick_ctx_insert_old_group(struct panthor_scheduler *sched,
 
 static void
 tick_ctx_init(struct panthor_scheduler *sched,
-	      struct panthor_sched_tick_ctx *ctx,
-	      bool full_tick)
+	      struct panthor_sched_tick_ctx *ctx)
 {
 	struct panthor_device *ptdev = sched->ptdev;
 	struct panthor_csg_slots_upd_ctx upd_ctx;

@@ -2062,7 +2052,7 @@ tick_ctx_init(struct panthor_scheduler *sched,
 				group->fatal_queues |= GENMASK(group->queue_count - 1, 0);
 		}
 
-		tick_ctx_insert_old_group(sched, ctx, group, full_tick);
+		tick_ctx_insert_old_group(sched, ctx, group);
 		csgs_upd_ctx_queue_reqs(ptdev, &upd_ctx, i,
 					csg_iface->output->ack ^ CSG_STATUS_UPDATE,
 					CSG_STATUS_UPDATE);

@@ -2405,7 +2395,7 @@ static void tick_work(struct work_struct *work)
 	if (panthor_device_reset_is_pending(sched->ptdev))
 		goto out_unlock;
 
-	tick_ctx_init(sched, &ctx, full_tick);
+	tick_ctx_init(sched, &ctx);
 	if (ctx.csg_upd_failed_mask)
 		goto out_cleanup_ctx;
 

@@ -2431,9 +2421,29 @@ static void tick_work(struct work_struct *work)
 	for (prio = PANTHOR_CSG_PRIORITY_COUNT - 1;
 	     prio >= 0 && !tick_ctx_is_full(sched, &ctx);
 	     prio--) {
+		struct panthor_group *old_highest_prio_group =
+			list_first_entry_or_null(&ctx.old_groups[prio],
+						 struct panthor_group, run_node);
+
+		/* Pull out the group with the highest prio for rotation. */
+		if (old_highest_prio_group)
+			list_del(&old_highest_prio_group->run_node);
+
+		/* Re-insert old active groups so they get a chance to run with higher prio. */
+		tick_ctx_pick_groups_from_list(sched, &ctx, &ctx.old_groups[prio], true, true);
+
+		/* Fill the remaining slots with runnable groups. */
 		tick_ctx_pick_groups_from_list(sched, &ctx, &sched->groups.runnable[prio],
 					       true, false);
-		tick_ctx_pick_groups_from_list(sched, &ctx, &ctx.old_groups[prio], true, true);
+
+		/* Re-insert the old group with the highest prio, and give it a chance to be
+		 * scheduled again (but with a lower prio) if there's room left.
+		 */
+		if (old_highest_prio_group) {
+			list_add_tail(&old_highest_prio_group->run_node, &ctx.old_groups[prio]);
+			tick_ctx_pick_groups_from_list(sched, &ctx, &ctx.old_groups[prio],
+						       true, true);
+		}
 	}
 
 	/* If we have free CSG slots left, pick idle groups */