Linux-ARM-Kernel Archive on lore.kernel.org
 help / color / mirror / Atom feed
* [PATCH 1/1 v7] ARM: imx: Added perf functionality to mmdc driver
From: Frank Li @ 2016-09-19 17:47 UTC (permalink / raw)
  To: linux-arm-kernel

From: Zhengyu Shen <zhengyu.shen@nxp.com>

MMDC is a multi-mode DDR controller that supports DDR3/DDR3L x16/x32/x64
and LPDDR2 two channel x16/x32 memory types. MMDC is configurable, high
performance, and optimized. MMDC is present on i.MX6 Quad and i.MX6
QuadPlus devices, but this driver only supports i.MX6 Quad at the moment.
MMDC provides registers for performance counters which read via this
driver to help debug memory throughput and similar issues.

$ perf stat -a -e mmdc/busy-cycles/,mmdc/read-accesses/,mmdc/read-bytes/,mmdc/total-cycles/,mmdc/write-accesses/,mmdc/write-bytes/ dd if=/dev/zero of=/dev/null bs=1M count=5000
Performance counter stats for 'dd if=/dev/zero of=/dev/null bs=1M count=5000':

         898021787      mmdc/busy-cycles/
          14819600      mmdc/read-accesses/
            471.30 MB   mmdc/read-bytes/
        2815419216      mmdc/total-cycles/
          13367354      mmdc/write-accesses/
            427.76 MB   mmdc/write-bytes/

       5.334757334 seconds time elapsed

Signed-off-by: Zhengyu Shen <zhengyu.shen@nxp.com>
Signed-off-by: Frank Li <frank.li@nxp.com>
---
Changes from v6 to v7
    use mmdc_pmu prefix
    remove unnecessary check
    improve group event check according to mark's feedback.
    check pmu_mmdc->mmdc_events[cfg] at event_add
    only check == 0 at event_del

Changes from v5 to v6
    Improve group event error handle

Changes from v4 to v5
    Remove mmdc_pmu:irq
    remove static variable cpuhp_mmdc_pmu
    remove spin_lock
    check is_sampling_event(event)
    remove unnecessary cast
    use hw_perf_event::prev_count

Changes from v3 to v4:
    Tested and fixed crash relating to removing events with perf fuzzer
    Adjusted formatting
    Moved all perf event code under CONFIG_PERF_EVENTS
        Switched cpuhp_setup_state to cpuhp_setup_state_nocalls

Changes from v2 to v3:
    Use WARN_ONCE instead of returning generic error values
    Replace CPU Notifiers with newer state machine hotplug
    Added additional checks on event_init for grouping and sampling
    Remove useless mmdc_enable_profiling function
    Added comments
    Moved start index of events from 0x01 to 0x00
    Added a counter to pmu_mmdc to only stop hrtimer after all events are finished
    Replace readl_relaxed and writel_relaxed with readl and writel
    Removed duplicate update function
    Used devm_kasprintf when naming mmdcs probed

Changes from v1 to v2:
    Added cpumask and migration handling support to driver
    Validated event during event_init
    Added code to properly stop counters
    Used perf_invalid_context instead of perf_sw_context
    Added hrtimer to poll for overflow
    Added better description
    Added support for multiple mmdcs

 arch/arm/mach-imx/mmdc.c | 459 ++++++++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 457 insertions(+), 2 deletions(-)

diff --git a/arch/arm/mach-imx/mmdc.c b/arch/arm/mach-imx/mmdc.c
index db9621c..1f70376 100644
--- a/arch/arm/mach-imx/mmdc.c
+++ b/arch/arm/mach-imx/mmdc.c
@@ -1,5 +1,5 @@
 /*
- * Copyright 2011 Freescale Semiconductor, Inc.
+ * Copyright 2011,2016 Freescale Semiconductor, Inc.
  * Copyright 2011 Linaro Ltd.
  *
  * The code contained herein is licensed under the GNU General Public
@@ -10,12 +10,16 @@
  * http://www.gnu.org/copyleft/gpl.html
  */
 
+#include <linux/hrtimer.h>
 #include <linux/init.h>
+#include <linux/interrupt.h>
 #include <linux/io.h>
 #include <linux/module.h>
 #include <linux/of.h>
 #include <linux/of_address.h>
 #include <linux/of_device.h>
+#include <linux/perf_event.h>
+#include <linux/slab.h>
 
 #include "common.h"
 
@@ -27,8 +31,458 @@
 #define BM_MMDC_MDMISC_DDR_TYPE	0x18
 #define BP_MMDC_MDMISC_DDR_TYPE	0x3
 
+#define TOTAL_CYCLES		0x0
+#define BUSY_CYCLES		0x1
+#define READ_ACCESSES		0x2
+#define WRITE_ACCESSES		0x3
+#define READ_BYTES		0x4
+#define WRITE_BYTES		0x5
+
+/* Enables, resets, freezes, overflow profiling*/
+#define DBG_DIS			0x0
+#define DBG_EN			0x1
+#define DBG_RST			0x2
+#define PRF_FRZ			0x4
+#define CYC_OVF			0x8
+
+#define MMDC_MADPCR0	0x410
+#define MMDC_MADPSR0	0x418
+#define MMDC_MADPSR1	0x41C
+#define MMDC_MADPSR2	0x420
+#define MMDC_MADPSR3	0x424
+#define MMDC_MADPSR4	0x428
+#define MMDC_MADPSR5	0x42C
+
+#define MMDC_NUM_COUNTERS	6
+
+#define to_mmdc_pmu(p) container_of(p, struct mmdc_pmu, pmu)
+
 static int ddr_type;
 
+#ifdef CONFIG_PERF_EVENTS
+
+static DEFINE_IDA(mmdc_ida);
+
+PMU_EVENT_ATTR_STRING(total-cycles, mmdc_pmu_total_cycles, "event=0x00")
+PMU_EVENT_ATTR_STRING(busy-cycles, mmdc_pmu_busy_cycles, "event=0x01")
+PMU_EVENT_ATTR_STRING(read-accesses, mmdc_pmu_read_accesses, "event=0x02")
+PMU_EVENT_ATTR_STRING(write-accesses, mmdc_pmu_write_accesses, "config=0x03")
+PMU_EVENT_ATTR_STRING(read-bytes, mmdc_pmu_read_bytes, "event=0x04")
+PMU_EVENT_ATTR_STRING(read-bytes.unit, mmdc_pmu_read_bytes_unit, "MB");
+PMU_EVENT_ATTR_STRING(read-bytes.scale, mmdc_pmu_read_bytes_scale, "0.000001");
+PMU_EVENT_ATTR_STRING(write-bytes, mmdc_pmu_write_bytes, "event=0x05")
+PMU_EVENT_ATTR_STRING(write-bytes.unit, mmdc_pmu_write_bytes_unit, "MB");
+PMU_EVENT_ATTR_STRING(write-bytes.scale, mmdc_pmu_write_bytes_scale, "0.000001");
+
+struct mmdc_pmu {
+	struct pmu pmu;
+	void __iomem *mmdc_base;
+	cpumask_t cpu;
+	struct hrtimer hrtimer;
+	unsigned int active_events;
+	struct device *dev;
+	struct perf_event *mmdc_events[MMDC_NUM_COUNTERS];
+	struct hlist_node node;
+};
+
+/*
+ * Polling period is set to one second, overflow of total-cycles (the fastest
+ * increasing counter) takes ten seconds so one second is safe
+ */
+static unsigned int mmdc_pmu_poll_period_us = 1000000;
+
+module_param_named(pmu_pmu_poll_period_us, mmdc_pmu_poll_period_us, uint,
+		S_IRUGO | S_IWUSR);
+
+static ktime_t mmdc_pmu_timer_period(void)
+{
+	return ns_to_ktime((u64)mmdc_pmu_poll_period_us * 1000);
+}
+
+static ssize_t mmdc_pmu_cpumask_show(struct device *dev,
+		struct device_attribute *attr, char *buf)
+{
+	struct mmdc_pmu *pmu_mmdc = dev_get_drvdata(dev);
+
+	return cpumap_print_to_pagebuf(true, buf, &pmu_mmdc->cpu);
+}
+
+static struct device_attribute mmdc_pmu_cpumask_attr =
+	__ATTR(cpumask, S_IRUGO, mmdc_pmu_cpumask_show, NULL);
+
+static struct attribute *mmdc_pmu_cpumask_attrs[] = {
+	&mmdc_pmu_cpumask_attr.attr,
+	NULL,
+};
+
+static struct attribute_group mmdc_pmu_cpumask_attr_group = {
+	.attrs = mmdc_pmu_cpumask_attrs,
+};
+
+static struct attribute *mmdc_pmu_events_attrs[] = {
+	&mmdc_pmu_total_cycles.attr.attr,
+	&mmdc_pmu_busy_cycles.attr.attr,
+	&mmdc_pmu_read_accesses.attr.attr,
+	&mmdc_pmu_write_accesses.attr.attr,
+	&mmdc_pmu_read_bytes.attr.attr,
+	&mmdc_pmu_read_bytes_unit.attr.attr,
+	&mmdc_pmu_read_bytes_scale.attr.attr,
+	&mmdc_pmu_write_bytes.attr.attr,
+	&mmdc_pmu_write_bytes_unit.attr.attr,
+	&mmdc_pmu_write_bytes_scale.attr.attr,
+	NULL,
+};
+
+static struct attribute_group mmdc_pmu_events_attr_group = {
+	.name = "events",
+	.attrs = mmdc_pmu_events_attrs,
+};
+
+PMU_FORMAT_ATTR(event, "config:0-63");
+static struct attribute *mmdc_pmu_format_attrs[] = {
+	&format_attr_event.attr,
+	NULL,
+};
+
+static struct attribute_group mmdc_pmu_format_attr_group = {
+	.name = "format",
+	.attrs = mmdc_pmu_format_attrs,
+};
+
+static const struct attribute_group *attr_groups[] = {
+	&mmdc_pmu_events_attr_group,
+	&mmdc_pmu_format_attr_group,
+	&mmdc_pmu_cpumask_attr_group,
+	NULL,
+};
+
+static u32 mmdc_pmu_read_counter(struct mmdc_pmu *pmu_mmdc, int cfg)
+{
+	void __iomem *mmdc_base, *reg;
+
+	mmdc_base = pmu_mmdc->mmdc_base;
+
+	switch (cfg) {
+	case TOTAL_CYCLES:
+		reg = mmdc_base + MMDC_MADPSR0;
+		break;
+	case BUSY_CYCLES:
+		reg = mmdc_base + MMDC_MADPSR1;
+		break;
+	case READ_ACCESSES:
+		reg = mmdc_base + MMDC_MADPSR2;
+		break;
+	case WRITE_ACCESSES:
+		reg = mmdc_base + MMDC_MADPSR3;
+		break;
+	case READ_BYTES:
+		reg = mmdc_base + MMDC_MADPSR4;
+		break;
+	case WRITE_BYTES:
+		reg = mmdc_base + MMDC_MADPSR5;
+		break;
+	default:
+		return WARN_ONCE(1,
+			"invalid configuration %d for mmdc counter", cfg);
+	}
+	return readl(reg);
+}
+
+static int mmdc_pmu_offline_cpu(unsigned int cpu, struct hlist_node *node)
+{
+	struct mmdc_pmu *pmu_mmdc = hlist_entry_safe(node, struct mmdc_pmu, node);
+	int target;
+
+	if (!cpumask_test_and_clear_cpu(cpu, &pmu_mmdc->cpu))
+		return 0;
+
+	target = cpumask_any_but(cpu_online_mask, cpu);
+	if (target >= nr_cpu_ids)
+		return 0;
+
+	perf_pmu_migrate_context(&pmu_mmdc->pmu, cpu, target);
+	cpumask_set_cpu(target, &pmu_mmdc->cpu);
+
+	return 0;
+}
+
+static bool mmdc_pmu_group_event_is_valid(struct perf_event *event,
+					  struct pmu *pmu,
+					  unsigned long *used_counters)
+{
+	int cfg = event->attr.config;
+
+	if (is_software_event(event))
+		return true;
+
+	if (event->pmu != pmu)
+		return false;
+
+	return !test_and_set_bit(cfg, used_counters);
+}
+
+/*
+ * Each event has a single fixed-purpose counter, so we can only have a
+ * single active event for each at any point in time. Here we just check
+ * for duplicates, and rely on mmdc_pmu_event_init to verify that the HW
+ * event numbers are valid.
+ */
+static bool mmdc_pmu_group_is_valid(struct perf_event *event)
+{
+	struct pmu *pmu = event->pmu;
+	struct perf_event *leader = event->group_leader;
+	struct perf_event *sibling;
+	unsigned long counter_mask = 0;
+
+	set_bit(leader->attr.config, &counter_mask);
+
+	if (event != leader) {
+		if (!mmdc_pmu_group_event_is_valid(event, pmu, &counter_mask))
+			return false;
+	}
+
+	list_for_each_entry(sibling, &leader->sibling_list, group_entry) {
+		if (!mmdc_pmu_group_event_is_valid(event, pmu, &counter_mask))
+			return false;
+	}
+
+	return true;
+}
+
+static int mmdc_pmu_event_init(struct perf_event *event)
+{
+	struct mmdc_pmu *pmu_mmdc = to_mmdc_pmu(event->pmu);
+	int cfg = event->attr.config;
+
+	if (event->attr.type != event->pmu->type)
+		return -ENOENT;
+
+	if (is_sampling_event(event) || event->attach_state & PERF_ATTACH_TASK)
+		return -EOPNOTSUPP;
+
+	if (event->cpu < 0) {
+		dev_warn(pmu_mmdc->dev, "Can't provide per-task data!\n");
+		return -EOPNOTSUPP;
+	}
+
+	if (event->attr.exclude_user		||
+			event->attr.exclude_kernel	||
+			event->attr.exclude_hv		||
+			event->attr.exclude_idle	||
+			event->attr.exclude_host	||
+			event->attr.exclude_guest	||
+			event->attr.sample_period)
+		return -EINVAL;
+
+	if (cfg < 0 || cfg >= MMDC_NUM_COUNTERS)
+		return -EINVAL;
+
+	if (!mmdc_pmu_group_is_valid(event))
+		return -EINVAL;
+
+	event->cpu = cpumask_first(&pmu_mmdc->cpu);
+	return 0;
+}
+
+static void mmdc_pmu_event_update(struct perf_event *event)
+{
+	struct mmdc_pmu *pmu_mmdc = to_mmdc_pmu(event->pmu);
+	struct hw_perf_event *hwc = &event->hw;
+	u64 delta, prev_raw_count, new_raw_count;
+
+	do {
+		prev_raw_count = local64_read(&hwc->prev_count);
+		new_raw_count = mmdc_pmu_read_counter(pmu_mmdc,
+						      event->attr.config);
+	} while (local64_cmpxchg(&hwc->prev_count, prev_raw_count,
+		new_raw_count) != prev_raw_count);
+
+	delta = (new_raw_count - prev_raw_count) & 0xFFFFFFFF;
+
+	local64_add(delta, &event->count);
+}
+
+static void mmdc_pmu_event_start(struct perf_event *event, int flags)
+{
+	struct mmdc_pmu *pmu_mmdc = to_mmdc_pmu(event->pmu);
+	struct hw_perf_event *hwc = &event->hw;
+	void __iomem *mmdc_base, *reg;
+
+	mmdc_base = pmu_mmdc->mmdc_base;
+	reg = mmdc_base + MMDC_MADPCR0;
+
+	/*
+	 * hrtimer is required because mmdc does not provide an interrupt so
+	 * polling is necessary
+	 */
+	hrtimer_start(&pmu_mmdc->hrtimer, mmdc_pmu_timer_period(),
+			HRTIMER_MODE_REL_PINNED);
+
+	local64_set(&hwc->prev_count, 0);
+
+	writel(DBG_RST, reg);
+	writel(DBG_EN, reg);
+}
+
+static int mmdc_pmu_event_add(struct perf_event *event, int flags)
+{
+	struct mmdc_pmu *pmu_mmdc = to_mmdc_pmu(event->pmu);
+	struct hw_perf_event *hwc = &event->hw;
+
+	int cfg = event->attr.config;
+
+	if (flags & PERF_EF_START)
+		mmdc_pmu_event_start(event, flags);
+
+	if (pmu_mmdc->mmdc_events[cfg] != NULL)
+		return -EAGAIN;
+
+	pmu_mmdc->mmdc_events[cfg] = event;
+	pmu_mmdc->active_events++;
+
+	local64_set(&hwc->prev_count, mmdc_pmu_read_counter(pmu_mmdc, cfg));
+
+	return 0;
+}
+
+static void mmdc_pmu_event_stop(struct perf_event *event, int flags)
+{
+	struct mmdc_pmu *pmu_mmdc = to_mmdc_pmu(event->pmu);
+	void __iomem *mmdc_base, *reg;
+
+	mmdc_base = pmu_mmdc->mmdc_base;
+	reg = mmdc_base + MMDC_MADPCR0;
+
+	writel(PRF_FRZ, reg);
+	mmdc_pmu_event_update(event);
+}
+
+static void mmdc_pmu_event_del(struct perf_event *event, int flags)
+{
+	struct mmdc_pmu *pmu_mmdc = to_mmdc_pmu(event->pmu);
+	int cfg = event->attr.config;
+
+	pmu_mmdc->mmdc_events[cfg] = NULL;
+	pmu_mmdc->active_events--;
+
+	if (pmu_mmdc->active_events == 0)
+		hrtimer_cancel(&pmu_mmdc->hrtimer);
+
+	mmdc_pmu_event_stop(event, PERF_EF_UPDATE);
+}
+
+static void mmdc_pmu_overflow_handler(struct mmdc_pmu *pmu_mmdc)
+{
+	int i;
+
+	for (i = 0; i < MMDC_NUM_COUNTERS; i++) {
+		struct perf_event *event = pmu_mmdc->mmdc_events[i];
+
+		if (event)
+			mmdc_pmu_event_update(event);
+	}
+}
+
+static enum hrtimer_restart mmdc_pmu_timer_handler(struct hrtimer *hrtimer)
+{
+	struct mmdc_pmu *pmu_mmdc = container_of(hrtimer, struct mmdc_pmu,
+			hrtimer);
+
+	mmdc_pmu_overflow_handler(pmu_mmdc);
+	hrtimer_forward_now(hrtimer, mmdc_pmu_timer_period());
+
+	return HRTIMER_RESTART;
+}
+
+static int mmdc_pmu_init(struct mmdc_pmu *pmu_mmdc,
+		void __iomem *mmdc_base, struct device *dev)
+{
+	int mmdc_num;
+
+	*pmu_mmdc = (struct mmdc_pmu) {
+		.pmu = (struct pmu) {
+			.task_ctx_nr    = perf_invalid_context,
+			.attr_groups    = attr_groups,
+			.event_init     = mmdc_pmu_event_init,
+			.add            = mmdc_pmu_event_add,
+			.del            = mmdc_pmu_event_del,
+			.start          = mmdc_pmu_event_start,
+			.stop           = mmdc_pmu_event_stop,
+			.read           = mmdc_pmu_event_update,
+		},
+		.mmdc_base = mmdc_base,
+		.dev = dev,
+		.active_events = 0,
+	};
+
+	mmdc_num = ida_simple_get(&mmdc_ida, 0, 0, GFP_KERNEL);
+
+	return mmdc_num;
+}
+
+static int imx_mmdc_remove(struct platform_device *pdev)
+{
+	struct mmdc_pmu *pmu_mmdc = platform_get_drvdata(pdev);
+
+	perf_pmu_unregister(&pmu_mmdc->pmu);
+	cpuhp_remove_state_nocalls(CPUHP_ONLINE);
+	kfree(pmu_mmdc);
+	return 0;
+}
+
+static int imx_mmdc_perf_init(struct platform_device *pdev, void __iomem *mmdc_base)
+{
+	struct mmdc_pmu *pmu_mmdc;
+	char *name;
+	int mmdc_num;
+	int ret;
+
+	pmu_mmdc = kzalloc(sizeof(*pmu_mmdc), GFP_KERNEL);
+	if (!pmu_mmdc) {
+		pr_err("failed to allocate PMU device!\n");
+		return -ENOMEM;
+	}
+
+	mmdc_num = mmdc_pmu_init(pmu_mmdc, mmdc_base, &pdev->dev);
+	if (mmdc_num == 0)
+		name = "mmdc";
+	else
+		name = devm_kasprintf(&pdev->dev,
+				GFP_KERNEL, "mmdc%d", mmdc_num);
+
+	hrtimer_init(&pmu_mmdc->hrtimer, CLOCK_MONOTONIC,
+			HRTIMER_MODE_REL);
+	pmu_mmdc->hrtimer.function = mmdc_pmu_timer_handler;
+
+	cpuhp_state_add_instance_nocalls(CPUHP_ONLINE,
+					 &pmu_mmdc->node);
+	cpumask_set_cpu(smp_processor_id(), &pmu_mmdc->cpu);
+	ret = cpuhp_setup_state_multi(CPUHP_AP_NOTIFY_ONLINE,
+				      "MMDC_ONLINE", NULL,
+				      mmdc_pmu_offline_cpu);
+	if (ret) {
+		pr_err("cpuhp_setup_state_multi failure\n");
+		goto pmu_register_err;
+	}
+
+	ret = perf_pmu_register(&(pmu_mmdc->pmu), name, -1);
+	platform_set_drvdata(pdev, pmu_mmdc);
+	if (ret)
+		goto pmu_register_err;
+	return 0;
+
+pmu_register_err:
+	pr_warn("MMDC Perf PMU failed (%d), disabled\n", ret);
+	hrtimer_cancel(&pmu_mmdc->hrtimer);
+	kfree(pmu_mmdc);
+	return ret;
+}
+
+#else
+#define imx_mmdc_remove NULL
+#define imx_mmdc_perf_init(pdev, mmdc_base) 0
+#endif
+
 static int imx_mmdc_probe(struct platform_device *pdev)
 {
 	struct device_node *np = pdev->dev.of_node;
@@ -62,7 +516,7 @@ static int imx_mmdc_probe(struct platform_device *pdev)
 		return -EBUSY;
 	}
 
-	return 0;
+	return imx_mmdc_perf_init(pdev, mmdc_base);
 }
 
 int imx_mmdc_get_ddr_type(void)
@@ -81,6 +535,7 @@ static struct platform_driver imx_mmdc_driver = {
 		.of_match_table = imx_mmdc_dt_ids,
 	},
 	.probe		= imx_mmdc_probe,
+	.remove		= imx_mmdc_remove,
 };
 
 static int __init imx_mmdc_init(void)
-- 
2.5.2

^ permalink raw reply related

* [PATCH] arm: dts: fix rk3066a based boards vdd_log voltage initialization
From: Boris Brezillon @ 2016-09-19 17:48 UTC (permalink / raw)
  To: linux-arm-kernel
In-Reply-To: <CAD=FV=Ufmvg1P8suOPQ53kpOD1pei4N2dOpKUCxZk001XKj_OA@mail.gmail.com>

On Mon, 19 Sep 2016 10:22:43 -0700
Doug Anderson <dianders@chromium.org> wrote:

> Hi,
> 
> On Mon, Sep 19, 2016 at 10:13 AM, Boris Brezillon
> <boris.brezillon@free-electrons.com> wrote:
> > Correct me if I'm wrong, but the main problem here is that, when we try
> > to detect the initial regulator state, we ran into a "missing entry in
> > the duty-cycle <-> voltage table" error, which then triggers an -EINVAL
> > error preventing the PWM regulator probe to succeed.
> >
> > Of course, adding an entry for the 0% dutycle case would solve the
> > issue, but I wonder if we should not allow "unknown value" at probe
> > time, and let the regulator user set the voltage output when it claims
> > it.
> >
> > Another option would be to fake a valid value in this case (choose the
> > closest entry in the voltage table?).  
> 
> If the goal is to avoid glitching the voltage at bootup, then the
> above doesn't really solve it.
> 
> For instance, you could have:
> 
> 0% - 1.5V
> 100% - 0.8V
> When configured as input: 1.1V
> Max useful value: 1.2V
> 
> So if firmware doesn't touch the PWM regulator then that means you're
> booting up at 1.1V.

Oh! I really didn't consider this case when developing the solution. I
was assuming that the firmware/bootloader would configure the PWM
correctly and leave it activating when booting the kernel.

> 
> If you read out the PWM itself, it will claim that it's running at 0%.
> While that's true, that doesn't mean that the voltage on the system is
> actually 1.5V because at boot the system had configured the pinmux as
> GPIO input.  ...so it's actually 1.1V because the PWM isn't actually
> controlling the pins.

The PWM chip has always claimed the pins and muxed them to the PWM IP.
So, this means it's broken from the beginning, and my patch is only
uncovering the problem (unless the pins stay configured as input until
the PWM is enabled, which I'm not sure is the case).

> 
> So if you want to avoid glitching the line then you want to make sure
> that you properly init to 1.1V, _not_ to 1.5V.
> 
> In the case above, it's possible that 1.5V isn't really so great for
> the hardware lifespan because 1.2V is the maximum useful value.  You
> could argue that the hardware is badly designed in this case, but I
> have certainly seen boards designed where the maximum achievable value
> is higher than the maximum "safe" value.

> 
> -Doug

^ permalink raw reply

* [PATCH] arm: dts: fix rk3066a based boards vdd_log voltage initialization
From: Doug Anderson @ 2016-09-19 17:52 UTC (permalink / raw)
  To: linux-arm-kernel
In-Reply-To: <20160919194844.0db7b307@bbrezillon>

Hi,

On Mon, Sep 19, 2016 at 10:48 AM, Boris Brezillon
<boris.brezillon@free-electrons.com> wrote:
> The PWM chip has always claimed the pins and muxed them to the PWM IP.
> So, this means it's broken from the beginning, and my patch is only
> uncovering the problem (unless the pins stay configured as input until
> the PWM is enabled, which I'm not sure is the case).

Such a solution is achievable with the pinctrl APIs pretty easily.
You might not be able to use the automatic "init" state but you can do
something similar and switch to an "active" state once the PWM is
actually turned on the first time.

-Doug

^ permalink raw reply

* [PATCH 1/1 v8] ARM: imx: Added perf functionality to mmdc driver
From: Frank Li @ 2016-09-19 17:57 UTC (permalink / raw)
  To: linux-arm-kernel

From: Zhengyu Shen <zhengyu.shen@nxp.com>

MMDC is a multi-mode DDR controller that supports DDR3/DDR3L x16/x32/x64
and LPDDR2 two channel x16/x32 memory types. MMDC is configurable, high
performance, and optimized. MMDC is present on i.MX6 Quad and i.MX6
QuadPlus devices, but this driver only supports i.MX6 Quad at the moment.
MMDC provides registers for performance counters which read via this
driver to help debug memory throughput and similar issues.

$ perf stat -a -e mmdc/busy-cycles/,mmdc/read-accesses/,mmdc/read-bytes/,mmdc/total-cycles/,mmdc/write-accesses/,mmdc/write-bytes/ dd if=/dev/zero of=/dev/null bs=1M count=5000
Performance counter stats for 'dd if=/dev/zero of=/dev/null bs=1M count=5000':

         898021787      mmdc/busy-cycles/
          14819600      mmdc/read-accesses/
            471.30 MB   mmdc/read-bytes/
        2815419216      mmdc/total-cycles/
          13367354      mmdc/write-accesses/
            427.76 MB   mmdc/write-bytes/

       5.334757334 seconds time elapsed

Signed-off-by: Zhengyu Shen <zhengyu.shen@nxp.com>
Signed-off-by: Frank Li <frank.li@nxp.com>
---
Changes from v7 to v8
    fix a group event check problem, should be slibling.

Changes from v6 to v7
    use mmdc_pmu prefix
    remove unnecessary check
    improve group event check according to mark's feedback.
    check pmu_mmdc->mmdc_events[cfg] at event_add
    only check == 0 at event_del

Changes from v5 to v6
    Improve group event error handle

Changes from v4 to v5
    Remove mmdc_pmu:irq
    remove static variable cpuhp_mmdc_pmu
    remove spin_lock
    check is_sampling_event(event)
    remove unnecessary cast
    use hw_perf_event::prev_count

Changes from v3 to v4:
    Tested and fixed crash relating to removing events with perf fuzzer
    Adjusted formatting
    Moved all perf event code under CONFIG_PERF_EVENTS
        Switched cpuhp_setup_state to cpuhp_setup_state_nocalls

Changes from v2 to v3:
    Use WARN_ONCE instead of returning generic error values
    Replace CPU Notifiers with newer state machine hotplug
    Added additional checks on event_init for grouping and sampling
    Remove useless mmdc_enable_profiling function
    Added comments
    Moved start index of events from 0x01 to 0x00
    Added a counter to pmu_mmdc to only stop hrtimer after all events are finished
    Replace readl_relaxed and writel_relaxed with readl and writel
    Removed duplicate update function
    Used devm_kasprintf when naming mmdcs probed

Changes from v1 to v2:
    Added cpumask and migration handling support to driver
    Validated event during event_init
    Added code to properly stop counters
    Used perf_invalid_context instead of perf_sw_context
    Added hrtimer to poll for overflow
    Added better description
    Added support for multiple mmdcs

 arch/arm/mach-imx/mmdc.c | 459 ++++++++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 457 insertions(+), 2 deletions(-)

diff --git a/arch/arm/mach-imx/mmdc.c b/arch/arm/mach-imx/mmdc.c
index db9621c..d82d14c 100644
--- a/arch/arm/mach-imx/mmdc.c
+++ b/arch/arm/mach-imx/mmdc.c
@@ -1,5 +1,5 @@
 /*
- * Copyright 2011 Freescale Semiconductor, Inc.
+ * Copyright 2011,2016 Freescale Semiconductor, Inc.
  * Copyright 2011 Linaro Ltd.
  *
  * The code contained herein is licensed under the GNU General Public
@@ -10,12 +10,16 @@
  * http://www.gnu.org/copyleft/gpl.html
  */
 
+#include <linux/hrtimer.h>
 #include <linux/init.h>
+#include <linux/interrupt.h>
 #include <linux/io.h>
 #include <linux/module.h>
 #include <linux/of.h>
 #include <linux/of_address.h>
 #include <linux/of_device.h>
+#include <linux/perf_event.h>
+#include <linux/slab.h>
 
 #include "common.h"
 
@@ -27,8 +31,458 @@
 #define BM_MMDC_MDMISC_DDR_TYPE	0x18
 #define BP_MMDC_MDMISC_DDR_TYPE	0x3
 
+#define TOTAL_CYCLES		0x0
+#define BUSY_CYCLES		0x1
+#define READ_ACCESSES		0x2
+#define WRITE_ACCESSES		0x3
+#define READ_BYTES		0x4
+#define WRITE_BYTES		0x5
+
+/* Enables, resets, freezes, overflow profiling*/
+#define DBG_DIS			0x0
+#define DBG_EN			0x1
+#define DBG_RST			0x2
+#define PRF_FRZ			0x4
+#define CYC_OVF			0x8
+
+#define MMDC_MADPCR0	0x410
+#define MMDC_MADPSR0	0x418
+#define MMDC_MADPSR1	0x41C
+#define MMDC_MADPSR2	0x420
+#define MMDC_MADPSR3	0x424
+#define MMDC_MADPSR4	0x428
+#define MMDC_MADPSR5	0x42C
+
+#define MMDC_NUM_COUNTERS	6
+
+#define to_mmdc_pmu(p) container_of(p, struct mmdc_pmu, pmu)
+
 static int ddr_type;
 
+#ifdef CONFIG_PERF_EVENTS
+
+static DEFINE_IDA(mmdc_ida);
+
+PMU_EVENT_ATTR_STRING(total-cycles, mmdc_pmu_total_cycles, "event=0x00")
+PMU_EVENT_ATTR_STRING(busy-cycles, mmdc_pmu_busy_cycles, "event=0x01")
+PMU_EVENT_ATTR_STRING(read-accesses, mmdc_pmu_read_accesses, "event=0x02")
+PMU_EVENT_ATTR_STRING(write-accesses, mmdc_pmu_write_accesses, "config=0x03")
+PMU_EVENT_ATTR_STRING(read-bytes, mmdc_pmu_read_bytes, "event=0x04")
+PMU_EVENT_ATTR_STRING(read-bytes.unit, mmdc_pmu_read_bytes_unit, "MB");
+PMU_EVENT_ATTR_STRING(read-bytes.scale, mmdc_pmu_read_bytes_scale, "0.000001");
+PMU_EVENT_ATTR_STRING(write-bytes, mmdc_pmu_write_bytes, "event=0x05")
+PMU_EVENT_ATTR_STRING(write-bytes.unit, mmdc_pmu_write_bytes_unit, "MB");
+PMU_EVENT_ATTR_STRING(write-bytes.scale, mmdc_pmu_write_bytes_scale, "0.000001");
+
+struct mmdc_pmu {
+	struct pmu pmu;
+	void __iomem *mmdc_base;
+	cpumask_t cpu;
+	struct hrtimer hrtimer;
+	unsigned int active_events;
+	struct device *dev;
+	struct perf_event *mmdc_events[MMDC_NUM_COUNTERS];
+	struct hlist_node node;
+};
+
+/*
+ * Polling period is set to one second, overflow of total-cycles (the fastest
+ * increasing counter) takes ten seconds so one second is safe
+ */
+static unsigned int mmdc_pmu_poll_period_us = 1000000;
+
+module_param_named(pmu_pmu_poll_period_us, mmdc_pmu_poll_period_us, uint,
+		S_IRUGO | S_IWUSR);
+
+static ktime_t mmdc_pmu_timer_period(void)
+{
+	return ns_to_ktime((u64)mmdc_pmu_poll_period_us * 1000);
+}
+
+static ssize_t mmdc_pmu_cpumask_show(struct device *dev,
+		struct device_attribute *attr, char *buf)
+{
+	struct mmdc_pmu *pmu_mmdc = dev_get_drvdata(dev);
+
+	return cpumap_print_to_pagebuf(true, buf, &pmu_mmdc->cpu);
+}
+
+static struct device_attribute mmdc_pmu_cpumask_attr =
+	__ATTR(cpumask, S_IRUGO, mmdc_pmu_cpumask_show, NULL);
+
+static struct attribute *mmdc_pmu_cpumask_attrs[] = {
+	&mmdc_pmu_cpumask_attr.attr,
+	NULL,
+};
+
+static struct attribute_group mmdc_pmu_cpumask_attr_group = {
+	.attrs = mmdc_pmu_cpumask_attrs,
+};
+
+static struct attribute *mmdc_pmu_events_attrs[] = {
+	&mmdc_pmu_total_cycles.attr.attr,
+	&mmdc_pmu_busy_cycles.attr.attr,
+	&mmdc_pmu_read_accesses.attr.attr,
+	&mmdc_pmu_write_accesses.attr.attr,
+	&mmdc_pmu_read_bytes.attr.attr,
+	&mmdc_pmu_read_bytes_unit.attr.attr,
+	&mmdc_pmu_read_bytes_scale.attr.attr,
+	&mmdc_pmu_write_bytes.attr.attr,
+	&mmdc_pmu_write_bytes_unit.attr.attr,
+	&mmdc_pmu_write_bytes_scale.attr.attr,
+	NULL,
+};
+
+static struct attribute_group mmdc_pmu_events_attr_group = {
+	.name = "events",
+	.attrs = mmdc_pmu_events_attrs,
+};
+
+PMU_FORMAT_ATTR(event, "config:0-63");
+static struct attribute *mmdc_pmu_format_attrs[] = {
+	&format_attr_event.attr,
+	NULL,
+};
+
+static struct attribute_group mmdc_pmu_format_attr_group = {
+	.name = "format",
+	.attrs = mmdc_pmu_format_attrs,
+};
+
+static const struct attribute_group *attr_groups[] = {
+	&mmdc_pmu_events_attr_group,
+	&mmdc_pmu_format_attr_group,
+	&mmdc_pmu_cpumask_attr_group,
+	NULL,
+};
+
+static u32 mmdc_pmu_read_counter(struct mmdc_pmu *pmu_mmdc, int cfg)
+{
+	void __iomem *mmdc_base, *reg;
+
+	mmdc_base = pmu_mmdc->mmdc_base;
+
+	switch (cfg) {
+	case TOTAL_CYCLES:
+		reg = mmdc_base + MMDC_MADPSR0;
+		break;
+	case BUSY_CYCLES:
+		reg = mmdc_base + MMDC_MADPSR1;
+		break;
+	case READ_ACCESSES:
+		reg = mmdc_base + MMDC_MADPSR2;
+		break;
+	case WRITE_ACCESSES:
+		reg = mmdc_base + MMDC_MADPSR3;
+		break;
+	case READ_BYTES:
+		reg = mmdc_base + MMDC_MADPSR4;
+		break;
+	case WRITE_BYTES:
+		reg = mmdc_base + MMDC_MADPSR5;
+		break;
+	default:
+		return WARN_ONCE(1,
+			"invalid configuration %d for mmdc counter", cfg);
+	}
+	return readl(reg);
+}
+
+static int mmdc_pmu_offline_cpu(unsigned int cpu, struct hlist_node *node)
+{
+	struct mmdc_pmu *pmu_mmdc = hlist_entry_safe(node, struct mmdc_pmu, node);
+	int target;
+
+	if (!cpumask_test_and_clear_cpu(cpu, &pmu_mmdc->cpu))
+		return 0;
+
+	target = cpumask_any_but(cpu_online_mask, cpu);
+	if (target >= nr_cpu_ids)
+		return 0;
+
+	perf_pmu_migrate_context(&pmu_mmdc->pmu, cpu, target);
+	cpumask_set_cpu(target, &pmu_mmdc->cpu);
+
+	return 0;
+}
+
+static bool mmdc_pmu_group_event_is_valid(struct perf_event *event,
+					  struct pmu *pmu,
+					  unsigned long *used_counters)
+{
+	int cfg = event->attr.config;
+
+	if (is_software_event(event))
+		return true;
+
+	if (event->pmu != pmu)
+		return false;
+
+	return !test_and_set_bit(cfg, used_counters);
+}
+
+/*
+ * Each event has a single fixed-purpose counter, so we can only have a
+ * single active event for each at any point in time. Here we just check
+ * for duplicates, and rely on mmdc_pmu_event_init to verify that the HW
+ * event numbers are valid.
+ */
+static bool mmdc_pmu_group_is_valid(struct perf_event *event)
+{
+	struct pmu *pmu = event->pmu;
+	struct perf_event *leader = event->group_leader;
+	struct perf_event *sibling;
+	unsigned long counter_mask = 0;
+
+	set_bit(leader->attr.config, &counter_mask);
+
+	if (event != leader) {
+		if (!mmdc_pmu_group_event_is_valid(event, pmu, &counter_mask))
+			return false;
+	}
+
+	list_for_each_entry(sibling, &leader->sibling_list, group_entry) {
+		if (!mmdc_pmu_group_event_is_valid(sibling, pmu, &counter_mask))
+			return false;
+	}
+
+	return true;
+}
+
+static int mmdc_pmu_event_init(struct perf_event *event)
+{
+	struct mmdc_pmu *pmu_mmdc = to_mmdc_pmu(event->pmu);
+	int cfg = event->attr.config;
+
+	if (event->attr.type != event->pmu->type)
+		return -ENOENT;
+
+	if (is_sampling_event(event) || event->attach_state & PERF_ATTACH_TASK)
+		return -EOPNOTSUPP;
+
+	if (event->cpu < 0) {
+		dev_warn(pmu_mmdc->dev, "Can't provide per-task data!\n");
+		return -EOPNOTSUPP;
+	}
+
+	if (event->attr.exclude_user		||
+			event->attr.exclude_kernel	||
+			event->attr.exclude_hv		||
+			event->attr.exclude_idle	||
+			event->attr.exclude_host	||
+			event->attr.exclude_guest	||
+			event->attr.sample_period)
+		return -EINVAL;
+
+	if (cfg < 0 || cfg >= MMDC_NUM_COUNTERS)
+		return -EINVAL;
+
+	if (!mmdc_pmu_group_is_valid(event))
+		return -EINVAL;
+
+	event->cpu = cpumask_first(&pmu_mmdc->cpu);
+	return 0;
+}
+
+static void mmdc_pmu_event_update(struct perf_event *event)
+{
+	struct mmdc_pmu *pmu_mmdc = to_mmdc_pmu(event->pmu);
+	struct hw_perf_event *hwc = &event->hw;
+	u64 delta, prev_raw_count, new_raw_count;
+
+	do {
+		prev_raw_count = local64_read(&hwc->prev_count);
+		new_raw_count = mmdc_pmu_read_counter(pmu_mmdc,
+						      event->attr.config);
+	} while (local64_cmpxchg(&hwc->prev_count, prev_raw_count,
+		new_raw_count) != prev_raw_count);
+
+	delta = (new_raw_count - prev_raw_count) & 0xFFFFFFFF;
+
+	local64_add(delta, &event->count);
+}
+
+static void mmdc_pmu_event_start(struct perf_event *event, int flags)
+{
+	struct mmdc_pmu *pmu_mmdc = to_mmdc_pmu(event->pmu);
+	struct hw_perf_event *hwc = &event->hw;
+	void __iomem *mmdc_base, *reg;
+
+	mmdc_base = pmu_mmdc->mmdc_base;
+	reg = mmdc_base + MMDC_MADPCR0;
+
+	/*
+	 * hrtimer is required because mmdc does not provide an interrupt so
+	 * polling is necessary
+	 */
+	hrtimer_start(&pmu_mmdc->hrtimer, mmdc_pmu_timer_period(),
+			HRTIMER_MODE_REL_PINNED);
+
+	local64_set(&hwc->prev_count, 0);
+
+	writel(DBG_RST, reg);
+	writel(DBG_EN, reg);
+}
+
+static int mmdc_pmu_event_add(struct perf_event *event, int flags)
+{
+	struct mmdc_pmu *pmu_mmdc = to_mmdc_pmu(event->pmu);
+	struct hw_perf_event *hwc = &event->hw;
+
+	int cfg = event->attr.config;
+
+	if (flags & PERF_EF_START)
+		mmdc_pmu_event_start(event, flags);
+
+	if (pmu_mmdc->mmdc_events[cfg] != NULL)
+		return -EAGAIN;
+
+	pmu_mmdc->mmdc_events[cfg] = event;
+	pmu_mmdc->active_events++;
+
+	local64_set(&hwc->prev_count, mmdc_pmu_read_counter(pmu_mmdc, cfg));
+
+	return 0;
+}
+
+static void mmdc_pmu_event_stop(struct perf_event *event, int flags)
+{
+	struct mmdc_pmu *pmu_mmdc = to_mmdc_pmu(event->pmu);
+	void __iomem *mmdc_base, *reg;
+
+	mmdc_base = pmu_mmdc->mmdc_base;
+	reg = mmdc_base + MMDC_MADPCR0;
+
+	writel(PRF_FRZ, reg);
+	mmdc_pmu_event_update(event);
+}
+
+static void mmdc_pmu_event_del(struct perf_event *event, int flags)
+{
+	struct mmdc_pmu *pmu_mmdc = to_mmdc_pmu(event->pmu);
+	int cfg = event->attr.config;
+
+	pmu_mmdc->mmdc_events[cfg] = NULL;
+	pmu_mmdc->active_events--;
+
+	if (pmu_mmdc->active_events == 0)
+		hrtimer_cancel(&pmu_mmdc->hrtimer);
+
+	mmdc_pmu_event_stop(event, PERF_EF_UPDATE);
+}
+
+static void mmdc_pmu_overflow_handler(struct mmdc_pmu *pmu_mmdc)
+{
+	int i;
+
+	for (i = 0; i < MMDC_NUM_COUNTERS; i++) {
+		struct perf_event *event = pmu_mmdc->mmdc_events[i];
+
+		if (event)
+			mmdc_pmu_event_update(event);
+	}
+}
+
+static enum hrtimer_restart mmdc_pmu_timer_handler(struct hrtimer *hrtimer)
+{
+	struct mmdc_pmu *pmu_mmdc = container_of(hrtimer, struct mmdc_pmu,
+			hrtimer);
+
+	mmdc_pmu_overflow_handler(pmu_mmdc);
+	hrtimer_forward_now(hrtimer, mmdc_pmu_timer_period());
+
+	return HRTIMER_RESTART;
+}
+
+static int mmdc_pmu_init(struct mmdc_pmu *pmu_mmdc,
+		void __iomem *mmdc_base, struct device *dev)
+{
+	int mmdc_num;
+
+	*pmu_mmdc = (struct mmdc_pmu) {
+		.pmu = (struct pmu) {
+			.task_ctx_nr    = perf_invalid_context,
+			.attr_groups    = attr_groups,
+			.event_init     = mmdc_pmu_event_init,
+			.add            = mmdc_pmu_event_add,
+			.del            = mmdc_pmu_event_del,
+			.start          = mmdc_pmu_event_start,
+			.stop           = mmdc_pmu_event_stop,
+			.read           = mmdc_pmu_event_update,
+		},
+		.mmdc_base = mmdc_base,
+		.dev = dev,
+		.active_events = 0,
+	};
+
+	mmdc_num = ida_simple_get(&mmdc_ida, 0, 0, GFP_KERNEL);
+
+	return mmdc_num;
+}
+
+static int imx_mmdc_remove(struct platform_device *pdev)
+{
+	struct mmdc_pmu *pmu_mmdc = platform_get_drvdata(pdev);
+
+	perf_pmu_unregister(&pmu_mmdc->pmu);
+	cpuhp_remove_state_nocalls(CPUHP_ONLINE);
+	kfree(pmu_mmdc);
+	return 0;
+}
+
+static int imx_mmdc_perf_init(struct platform_device *pdev, void __iomem *mmdc_base)
+{
+	struct mmdc_pmu *pmu_mmdc;
+	char *name;
+	int mmdc_num;
+	int ret;
+
+	pmu_mmdc = kzalloc(sizeof(*pmu_mmdc), GFP_KERNEL);
+	if (!pmu_mmdc) {
+		pr_err("failed to allocate PMU device!\n");
+		return -ENOMEM;
+	}
+
+	mmdc_num = mmdc_pmu_init(pmu_mmdc, mmdc_base, &pdev->dev);
+	if (mmdc_num == 0)
+		name = "mmdc";
+	else
+		name = devm_kasprintf(&pdev->dev,
+				GFP_KERNEL, "mmdc%d", mmdc_num);
+
+	hrtimer_init(&pmu_mmdc->hrtimer, CLOCK_MONOTONIC,
+			HRTIMER_MODE_REL);
+	pmu_mmdc->hrtimer.function = mmdc_pmu_timer_handler;
+
+	cpuhp_state_add_instance_nocalls(CPUHP_ONLINE,
+					 &pmu_mmdc->node);
+	cpumask_set_cpu(smp_processor_id(), &pmu_mmdc->cpu);
+	ret = cpuhp_setup_state_multi(CPUHP_AP_NOTIFY_ONLINE,
+				      "MMDC_ONLINE", NULL,
+				      mmdc_pmu_offline_cpu);
+	if (ret) {
+		pr_err("cpuhp_setup_state_multi failure\n");
+		goto pmu_register_err;
+	}
+
+	ret = perf_pmu_register(&(pmu_mmdc->pmu), name, -1);
+	platform_set_drvdata(pdev, pmu_mmdc);
+	if (ret)
+		goto pmu_register_err;
+	return 0;
+
+pmu_register_err:
+	pr_warn("MMDC Perf PMU failed (%d), disabled\n", ret);
+	hrtimer_cancel(&pmu_mmdc->hrtimer);
+	kfree(pmu_mmdc);
+	return ret;
+}
+
+#else
+#define imx_mmdc_remove NULL
+#define imx_mmdc_perf_init(pdev, mmdc_base) 0
+#endif
+
 static int imx_mmdc_probe(struct platform_device *pdev)
 {
 	struct device_node *np = pdev->dev.of_node;
@@ -62,7 +516,7 @@ static int imx_mmdc_probe(struct platform_device *pdev)
 		return -EBUSY;
 	}
 
-	return 0;
+	return imx_mmdc_perf_init(pdev, mmdc_base);
 }
 
 int imx_mmdc_get_ddr_type(void)
@@ -81,6 +535,7 @@ static struct platform_driver imx_mmdc_driver = {
 		.of_match_table = imx_mmdc_dt_ids,
 	},
 	.probe		= imx_mmdc_probe,
+	.remove		= imx_mmdc_remove,
 };
 
 static int __init imx_mmdc_init(void)
-- 
2.5.2

^ permalink raw reply related

* [RFC] arm64: Ensure proper addressing for ldnp/stnp
From: Robin Murphy @ 2016-09-19 18:01 UTC (permalink / raw)
  To: linux-arm-kernel
In-Reply-To: <1474306586-25118-1-git-send-email-bdegraaf@codeaurora.org>

On 19/09/16 18:36, Brent DeGraaf wrote:
> According to section 6.3.8 of the ARM Programmer's Guide, non-temporal
> loads and stores do not verify that address dependency is met between a
> load of an address to a register and a subsequent non-temporal load or
> store using that address on the executing PE. Therefore, context switch
> code and subroutine calls that use non-temporally accessed addresses as
> parameters that might depend on a load of an address into an argument
> register must ensure that ordering requirements are met by introducing
> a barrier prior to the successive non-temporal access.  Add appropriate
> barriers whereever this specific situation comes into play.
> 
> Signed-off-by: Brent DeGraaf <bdegraaf@codeaurora.org>
> ---
>  arch/arm64/kernel/entry.S  | 1 +
>  arch/arm64/lib/copy_page.S | 2 ++
>  2 files changed, 3 insertions(+)
> 
> diff --git a/arch/arm64/kernel/entry.S b/arch/arm64/kernel/entry.S
> index 441420c..982c4d3 100644
> --- a/arch/arm64/kernel/entry.S
> +++ b/arch/arm64/kernel/entry.S
> @@ -679,6 +679,7 @@ ENTRY(cpu_switch_to)
>  	ldp	x27, x28, [x8], #16
>  	ldp	x29, x9, [x8], #16
>  	ldr	lr, [x8]
> +	dmb	nshld	// Existence of instructions with loose load-use dependencies (e.g. ldnp/stnp) make this barrier necessary
>  	mov	sp, x9
>  	and	x9, x9, #~(THREAD_SIZE - 1)
>  	msr	sp_el0, x9
> diff --git a/arch/arm64/lib/copy_page.S b/arch/arm64/lib/copy_page.S
> index 4c1e700..21c6892 100644
> --- a/arch/arm64/lib/copy_page.S
> +++ b/arch/arm64/lib/copy_page.S
> @@ -47,6 +47,8 @@ alternative_endif
>  	ldp	x14, x15, [x1, #96]
>  	ldp	x16, x17, [x1, #112]
>  
> +	dmb	nshld // In case x0 (for stnp) is dependent on a load

The ARMv8 ARM (B2.7.2 in issue j) says that when an address dependency
exists between a load and a subsequent LDNP, *other* observers may
observe those accesses in any order. How's that related to an STNP on
the same CPU?

Robin.

> +
>  	mov	x18, #(PAGE_SIZE - 128)
>  	add	x1, x1, #128
>  1:
> 

^ permalink raw reply

* [PATCH] arm: dts: fix rk3066a based boards vdd_log voltage initialization
From: Boris Brezillon @ 2016-09-19 18:06 UTC (permalink / raw)
  To: linux-arm-kernel
In-Reply-To: <CAD=FV=UyW5cXGGmq5xUzZ1aV5NzbBYGOz1gO9V1itrFU-PigAw@mail.gmail.com>

On Mon, 19 Sep 2016 10:52:51 -0700
Doug Anderson <dianders@chromium.org> wrote:

> Hi,
> 
> On Mon, Sep 19, 2016 at 10:48 AM, Boris Brezillon
> <boris.brezillon@free-electrons.com> wrote:
> > The PWM chip has always claimed the pins and muxed them to the PWM IP.
> > So, this means it's broken from the beginning, and my patch is only
> > uncovering the problem (unless the pins stay configured as input until
> > the PWM is enabled, which I'm not sure is the case).  
> 
> Such a solution is achievable with the pinctrl APIs pretty easily.
> You might not be able to use the automatic "init" state but you can do
> something similar and switch to an "active" state once the PWM is
> actually turned on the first time.

But is it really the case here (I don't see any code requesting a
specific pinmux depending on the PWM state)?

Anyway, we really need to handle this case, we should define the
typical voltage when the PWM is disabled. Same as what you suggested
with voltage-when-input, but with a different naming (since the concept
of pinmux is PWM hardware/driver specific).

	voltage-when-pwm-disabled = <...>;

^ permalink raw reply

* find_child_device() on ACPI node works only once
From: Timur Tabi @ 2016-09-19 18:08 UTC (permalink / raw)
  To: linux-arm-kernel

I need help getting my ACPI table parsing code to work correctly.  My 
driver fails to find the child device in a node when the driver is 
unloaded and reloaded.  I suspect that I'm not freeing a resource correctly.

Here is my ACPI table (trimmed):

Device (MAC0) {
	Name (_HID, "QCOM8070")
	Name (_UID, 0)
	Method (_CRS, 0x0, Serialized) {
		Name (RBUF, ResourceTemplate() {
			...
			Return (RBUF)
		}
	}
	Device (IPHY) {
		Name (_HID, "QCOM8071")

		Method (_CRS, 0x0, Serialized) {
			Name (RBUF, ResourceTemplate () {
				...
			Return (RBUF)
		}
	}


So QCOM8071 is a child device for QCOM8070.  The driver probes on 
QCOM8070 and then manually instantiates QCOM8071.

static const struct acpi_device_id emac_acpi_match[] = {
	{
		.id = "QCOM8070",
	},
	{}
};
MODULE_DEVICE_TABLE(acpi, emac_acpi_match);

static struct platform_driver emac_platform_driver = {
	.probe	= emac_probe,
	.remove	= emac_remove,
	.driver = {
		.name		= "qcom-emac",
		.acpi_match_table = ACPI_PTR(emac_acpi_match),
	},
};

In emac_probe, I do this:

struct device *dev;

dev = device_find_child(&pdev->dev, NULL, emac_sgmii_acpi_match);
adpt->sgmii_pdev = to_platform_device(dev);

And here's emac_sgmii_acpi_match:

> static int emac_sgmii_acpi_match(struct device *dev, void *data)
> {
> 	static const struct acpi_device_id match_table[] = {
> 		{
> 			.id = "QCOM8071",
> 		},
> 		{}
> 	};
> 	const struct acpi_device_id *id = acpi_match_device(match_table, dev);
>
> 	return !!id;
> }

When my driver exits, I call

	platform_device_unregister(adpt->sgmii_pdev);

So here's the problem: the second time my driver is loaded, 
emac_sgmii_acpi_match() always returns false.  The QCOM8071 child node 
no longer exists.  It's apparently hidden in some way.

I've been trying for several days to debug this, and I'm at my wits' 
end.  I'm assuming that I'm not freeing the resource properly, but other 
than calling platform_device_unregister(), what else should I do?

-- 
Qualcomm Datacenter Technologies, Inc. as an affiliate of Qualcomm
Technologies, Inc.  Qualcomm Technologies, Inc. is a member of the
Code Aurora Forum, a Linux Foundation Collaborative Project.

^ permalink raw reply

* [PATCH V6 3/5] PCI: thunder-pem: Allow to probe PEM-specific register range for ACPI case
From: Bjorn Helgaas @ 2016-09-19 18:09 UTC (permalink / raw)
  To: linux-arm-kernel
In-Reply-To: <1473449047-10499-4-git-send-email-tn@semihalf.com>

On Fri, Sep 09, 2016 at 09:24:05PM +0200, Tomasz Nowicki wrote:
> thunder-pem driver stands for being ACPI based PCI host controller.
> However, there is no standard way to describe its PEM-specific register
> ranges in ACPI tables. Thus we add thunder_pem_init() ACPI extension
> to obtain hardcoded addresses from static resource array.
> Although it is not pretty, it prevents from creating standard mechanism to
> handle similar cases in future.
> 
> Signed-off-by: Tomasz Nowicki <tn@semihalf.com>
> ---
>  drivers/pci/host/pci-thunder-pem.c | 61 ++++++++++++++++++++++++++++++--------
>  1 file changed, 48 insertions(+), 13 deletions(-)
> 
> diff --git a/drivers/pci/host/pci-thunder-pem.c b/drivers/pci/host/pci-thunder-pem.c
> index 6abaf80..b048761 100644
> --- a/drivers/pci/host/pci-thunder-pem.c
> +++ b/drivers/pci/host/pci-thunder-pem.c
> @@ -18,6 +18,7 @@
>  #include <linux/init.h>
>  #include <linux/of_address.h>
>  #include <linux/of_pci.h>
> +#include <linux/pci-acpi.h>
>  #include <linux/pci-ecam.h>
>  #include <linux/platform_device.h>
>  
> @@ -284,6 +285,40 @@ static int thunder_pem_config_write(struct pci_bus *bus, unsigned int devfn,
>  	return pci_generic_config_write(bus, devfn, where, size, val);
>  }
>  
> +#ifdef CONFIG_ACPI
> +static struct resource thunder_pem_reg_res[] = {
> +	[4] = DEFINE_RES_MEM(0x87e0c0000000UL, SZ_16M),
> +	[5] = DEFINE_RES_MEM(0x87e0c1000000UL, SZ_16M),
> +	[6] = DEFINE_RES_MEM(0x87e0c2000000UL, SZ_16M),
> +	[7] = DEFINE_RES_MEM(0x87e0c3000000UL, SZ_16M),
> +	[8] = DEFINE_RES_MEM(0x87e0c4000000UL, SZ_16M),
> +	[9] = DEFINE_RES_MEM(0x87e0c5000000UL, SZ_16M),
> +	[14] = DEFINE_RES_MEM(0x97e0c0000000UL, SZ_16M),
> +	[15] = DEFINE_RES_MEM(0x97e0c1000000UL, SZ_16M),
> +	[16] = DEFINE_RES_MEM(0x97e0c2000000UL, SZ_16M),
> +	[17] = DEFINE_RES_MEM(0x97e0c3000000UL, SZ_16M),
> +	[18] = DEFINE_RES_MEM(0x97e0c4000000UL, SZ_16M),
> +	[19] = DEFINE_RES_MEM(0x97e0c5000000UL, SZ_16M),

1) The "correct" way to discover the resources consumed by an ACPI
   device is to use the _CRS method.  I know there are some issues
   there for bridges (not the fault of ThunderX!) because there's not
   a good way to distinguish windows from resources consumed directly
   by the bridge.

   But we should either do this correctly, or include a comment about
   why we're doing it wrong, so we don't give the impression that this
   is the right way to do it.

   I seem to recall some discussion about why we're doing it this way,
   but I don't remember the details.  It'd be nice to include a
   summary here.

2) This is a little weird because here we define the resource size as
   16MB, in the OF case we get the resource size from OF, in either
   case we ioremap 64K of it, and then as far as I can tell, we only
   ever access PEM_CFG_WR and PEM_CFG_RD, at offsets 0x28 and 0x30
   into the space.

   If the hardware actually decodes the entire 16MB, we should ioremap
   the whole 16MB.  (Strictly speaking, drivers only need to ioremap
   the parts they're using, but in this case nobody claims the entire
   resource because of deficiencies in the ACPI and OF cores, so the
   driver should ioremap the entire thing to help prevent conflicts
   with other devices.)

   It'd be nice if we didn't have the 64KB magic number.  I think
   using devm_ioremap_resource() would be nice.

> +};
> +
> +static struct resource *thunder_pem_acpi_res(struct pci_config_window *cfg)
> +{
> +	struct acpi_device *adev = to_acpi_device(cfg->parent);
> +	struct acpi_pci_root *root = acpi_driver_data(adev);
> +
> +	if ((root->segment >= 4 && root->segment <= 9) ||
> +	    (root->segment >= 14 && root->segment <= 19))
> +		return &thunder_pem_reg_res[root->segment];
> +
> +	return NULL;
> +}
> +#else
> +static struct resource *thunder_pem_acpi_res(struct pci_config_window *cfg)
> +{
> +	return NULL;
> +}
> +#endif
> +
>  static int thunder_pem_init(struct pci_config_window *cfg)
>  {
>  	struct device *dev = cfg->parent;
> @@ -292,24 +327,24 @@ static int thunder_pem_init(struct pci_config_window *cfg)
>  	struct thunder_pem_pci *pem_pci;
>  	struct platform_device *pdev;
>  
> -	/* Only OF support for now */
> -	if (!dev->of_node)
> -		return -EINVAL;
> -
>  	pem_pci = devm_kzalloc(dev, sizeof(*pem_pci), GFP_KERNEL);
>  	if (!pem_pci)
>  		return -ENOMEM;
>  
> -	pdev = to_platform_device(dev);
> -
> -	/*
> -	 * The second register range is the PEM bridge to the PCIe
> -	 * bus.  It has a different config access method than those
> -	 * devices behind the bridge.
> -	 */
> -	res_pem = platform_get_resource(pdev, IORESOURCE_MEM, 1);
> +	if (acpi_disabled) {
> +		pdev = to_platform_device(dev);
> +
> +		/*
> +		 * The second register range is the PEM bridge to the PCIe
> +		 * bus.  It has a different config access method than those
> +		 * devices behind the bridge.
> +		 */
> +		res_pem = platform_get_resource(pdev, IORESOURCE_MEM, 1);
> +	} else {
> +		res_pem = thunder_pem_acpi_res(cfg);
> +	}
>  	if (!res_pem) {
> -		dev_err(dev, "missing \"reg[1]\"property\n");
> +		dev_err(dev, "missing configuration region\n");
>  		return -EINVAL;
>  	}
>  
> -- 
> 1.9.1
> 

^ permalink raw reply

* [PATCH] arm: dts: fix rk3066a based boards vdd_log voltage initialization
From: Doug Anderson @ 2016-09-19 18:12 UTC (permalink / raw)
  To: linux-arm-kernel
In-Reply-To: <20160919200627.3ca1aae4@bbrezillon>

Hi,

On Mon, Sep 19, 2016 at 11:06 AM, Boris Brezillon
<boris.brezillon@free-electrons.com> wrote:
> On Mon, 19 Sep 2016 10:52:51 -0700
> Doug Anderson <dianders@chromium.org> wrote:
>
>> Hi,
>>
>> On Mon, Sep 19, 2016 at 10:48 AM, Boris Brezillon
>> <boris.brezillon@free-electrons.com> wrote:
>> > The PWM chip has always claimed the pins and muxed them to the PWM IP.
>> > So, this means it's broken from the beginning, and my patch is only
>> > uncovering the problem (unless the pins stay configured as input until
>> > the PWM is enabled, which I'm not sure is the case).
>>
>> Such a solution is achievable with the pinctrl APIs pretty easily.
>> You might not be able to use the automatic "init" state but you can do
>> something similar and switch to an "active" state once the PWM is
>> actually turned on the first time.
>
> But is it really the case here (I don't see any code requesting a
> specific pinmux depending on the PWM state)?

It is not happening right now as far as I know.  ...but that's a bug.

> Anyway, we really need to handle this case, we should define the
> typical voltage when the PWM is disabled. Same as what you suggested
> with voltage-when-input, but with a different naming (since the concept
> of pinmux is PWM hardware/driver specific).
>
>         voltage-when-pwm-disabled = <...>;

Voltage when disabled and voltage when input are two different states.
A disabled PWM will typically either drive high or low (depending on
where it was when you turned it off).  Not all "disabled" states will
mean that the pin is configured as an input.

^ permalink raw reply

* [PATCH v2 0/3] Armada 7k/8k CP110 system controller fixes
From: Stephen Boyd @ 2016-09-19 18:15 UTC (permalink / raw)
  To: linux-arm-kernel
In-Reply-To: <CAPv3WKdoas6nJngFh2bLU0wvz2OZMga6G40UMPkfz9aunRawVw@mail.gmail.com>

On 09/19, Marcin Wojtas wrote:
> Hi Stephen,
> 
> Did you have any chance to take a look at v2? Do you have any remarks?
> 

Hmm I don't have any of these patches in my queue. Probably
because there was some bug in the second patch so I assumed it
would go for another round. Please fix that bug and resend.

-- 
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
a Linux Foundation Collaborative Project

^ permalink raw reply

* [PATCH v2 0/3] Armada 7k/8k CP110 system controller fixes
From: Marcin Wojtas @ 2016-09-19 18:18 UTC (permalink / raw)
  To: linux-arm-kernel
In-Reply-To: <20160919181549.GE7243@codeaurora.org>

Sure, thanks. I'll send it right away.

Best regards,
Marcin

2016-09-19 20:15 GMT+02:00 Stephen Boyd <sboyd@codeaurora.org>:
> On 09/19, Marcin Wojtas wrote:
>> Hi Stephen,
>>
>> Did you have any chance to take a look at v2? Do you have any remarks?
>>
>
> Hmm I don't have any of these patches in my queue. Probably
> because there was some bug in the second patch so I assumed it
> would go for another round. Please fix that bug and resend.
>
> --
> Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
> a Linux Foundation Collaborative Project

^ permalink raw reply

* [RFC] arm64: Ensure proper addressing for ldnp/stnp
From: bdegraaf at codeaurora.org @ 2016-09-19 18:25 UTC (permalink / raw)
  To: linux-arm-kernel
In-Reply-To: <be5f5bfd-fd41-d202-cec9-0ca3fcad28db@arm.com>

On 2016-09-19 14:01, Robin Murphy wrote:
> On 19/09/16 18:36, Brent DeGraaf wrote:
>> According to section 6.3.8 of the ARM Programmer's Guide, non-temporal
>> loads and stores do not verify that address dependency is met between 
>> a
>> load of an address to a register and a subsequent non-temporal load or
>> store using that address on the executing PE. Therefore, context 
>> switch
>> code and subroutine calls that use non-temporally accessed addresses 
>> as
>> parameters that might depend on a load of an address into an argument
>> register must ensure that ordering requirements are met by introducing
>> a barrier prior to the successive non-temporal access.  Add 
>> appropriate
>> barriers whereever this specific situation comes into play.
>> 
>> Signed-off-by: Brent DeGraaf <bdegraaf@codeaurora.org>
>> ---
>>  arch/arm64/kernel/entry.S  | 1 +
>>  arch/arm64/lib/copy_page.S | 2 ++
>>  2 files changed, 3 insertions(+)
>> 
>> diff --git a/arch/arm64/kernel/entry.S b/arch/arm64/kernel/entry.S
>> index 441420c..982c4d3 100644
>> --- a/arch/arm64/kernel/entry.S
>> +++ b/arch/arm64/kernel/entry.S
>> @@ -679,6 +679,7 @@ ENTRY(cpu_switch_to)
>>  	ldp	x27, x28, [x8], #16
>>  	ldp	x29, x9, [x8], #16
>>  	ldr	lr, [x8]
>> +	dmb	nshld	// Existence of instructions with loose load-use 
>> dependencies (e.g. ldnp/stnp) make this barrier necessary
>>  	mov	sp, x9
>>  	and	x9, x9, #~(THREAD_SIZE - 1)
>>  	msr	sp_el0, x9
>> diff --git a/arch/arm64/lib/copy_page.S b/arch/arm64/lib/copy_page.S
>> index 4c1e700..21c6892 100644
>> --- a/arch/arm64/lib/copy_page.S
>> +++ b/arch/arm64/lib/copy_page.S
>> @@ -47,6 +47,8 @@ alternative_endif
>>  	ldp	x14, x15, [x1, #96]
>>  	ldp	x16, x17, [x1, #112]
>> 
>> +	dmb	nshld // In case x0 (for stnp) is dependent on a load
> 
> The ARMv8 ARM (B2.7.2 in issue j) says that when an address dependency
> exists between a load and a subsequent LDNP, *other* observers may
> observe those accesses in any order. How's that related to an STNP on
> the same CPU?
> 
> Robin.
> 
>> +
>>  	mov	x18, #(PAGE_SIZE - 128)
>>  	add	x1, x1, #128
>>  1:
>> 

Yes, I have seen the section in the ARM ARM about this. But the 
Programmer's Guide goes further, even providing a concrete example:

"Non-temporal loads and stores relax the memory ordering 
requirements...the LDNP instruction might
be observed before the preceding LDR instruction, which can result in 
reading from an unpredictable
address in X0.

For example:
LDR X0, [X3]
LDNP X2, X1, [X0]
To correct the above, you need an explicit load barrier:
LDR X0, [X3]
DMB NSHLD
LDNP X2, X1, [X0]"

Did the ARM ARM leave this out?  Or is the Programmer's Guide section 
incorrect?

Thanks for your comments,
Brent

^ permalink raw reply

* [RFC] arm64: Ensure proper addressing for ldnp/stnp
From: Laura Abbott @ 2016-09-19 18:28 UTC (permalink / raw)
  To: linux-arm-kernel
In-Reply-To: <1474306586-25118-1-git-send-email-bdegraaf@codeaurora.org>

On 09/19/2016 10:36 AM, Brent DeGraaf wrote:
> According to section 6.3.8 of the ARM Programmer's Guide, non-temporal
> loads and stores do not verify that address dependency is met between a
> load of an address to a register and a subsequent non-temporal load or
> store using that address on the executing PE. Therefore, context switch
> code and subroutine calls that use non-temporally accessed addresses as
> parameters that might depend on a load of an address into an argument
> register must ensure that ordering requirements are met by introducing
> a barrier prior to the successive non-temporal access.  Add appropriate
> barriers whereever this specific situation comes into play.
>

Was this found by code inspection or is there a (public) exciting test
case to observe this behavior?

Thanks,
Laura

> Signed-off-by: Brent DeGraaf <bdegraaf@codeaurora.org>
> ---
>  arch/arm64/kernel/entry.S  | 1 +
>  arch/arm64/lib/copy_page.S | 2 ++
>  2 files changed, 3 insertions(+)
>
> diff --git a/arch/arm64/kernel/entry.S b/arch/arm64/kernel/entry.S
> index 441420c..982c4d3 100644
> --- a/arch/arm64/kernel/entry.S
> +++ b/arch/arm64/kernel/entry.S
> @@ -679,6 +679,7 @@ ENTRY(cpu_switch_to)
>  	ldp	x27, x28, [x8], #16
>  	ldp	x29, x9, [x8], #16
>  	ldr	lr, [x8]
> +	dmb	nshld	// Existence of instructions with loose load-use dependencies (e.g. ldnp/stnp) make this barrier necessary
>  	mov	sp, x9
>  	and	x9, x9, #~(THREAD_SIZE - 1)
>  	msr	sp_el0, x9
> diff --git a/arch/arm64/lib/copy_page.S b/arch/arm64/lib/copy_page.S
> index 4c1e700..21c6892 100644
> --- a/arch/arm64/lib/copy_page.S
> +++ b/arch/arm64/lib/copy_page.S
> @@ -47,6 +47,8 @@ alternative_endif
>  	ldp	x14, x15, [x1, #96]
>  	ldp	x16, x17, [x1, #112]
>
> +	dmb	nshld // In case x0 (for stnp) is dependent on a load
> +
>  	mov	x18, #(PAGE_SIZE - 128)
>  	add	x1, x1, #128
>  1:
>

^ permalink raw reply

* [PATCH] arm: dts: fix rk3066a based boards vdd_log voltage initialization
From: Boris Brezillon @ 2016-09-19 18:31 UTC (permalink / raw)
  To: linux-arm-kernel
In-Reply-To: <CAD=FV=U_BQez2eCqm7mX9kj0dcEsnBKiFWcLAhXXJD+yTZhACw@mail.gmail.com>

On Mon, 19 Sep 2016 11:12:12 -0700
Doug Anderson <dianders@chromium.org> wrote:

> Hi,
> 
> On Mon, Sep 19, 2016 at 11:06 AM, Boris Brezillon
> <boris.brezillon@free-electrons.com> wrote:
> > On Mon, 19 Sep 2016 10:52:51 -0700
> > Doug Anderson <dianders@chromium.org> wrote:
> >  
> >> Hi,
> >>
> >> On Mon, Sep 19, 2016 at 10:48 AM, Boris Brezillon
> >> <boris.brezillon@free-electrons.com> wrote:  
> >> > The PWM chip has always claimed the pins and muxed them to the PWM IP.
> >> > So, this means it's broken from the beginning, and my patch is only
> >> > uncovering the problem (unless the pins stay configured as input until
> >> > the PWM is enabled, which I'm not sure is the case).  
> >>
> >> Such a solution is achievable with the pinctrl APIs pretty easily.
> >> You might not be able to use the automatic "init" state but you can do
> >> something similar and switch to an "active" state once the PWM is
> >> actually turned on the first time.  
> >
> > But is it really the case here (I don't see any code requesting a
> > specific pinmux depending on the PWM state)?  
> 
> It is not happening right now as far as I know.  ...but that's a bug.

Okay.

> 
> > Anyway, we really need to handle this case, we should define the
> > typical voltage when the PWM is disabled. Same as what you suggested
> > with voltage-when-input, but with a different naming (since the concept
> > of pinmux is PWM hardware/driver specific).
> >
> >         voltage-when-pwm-disabled = <...>;  
> 
> Voltage when disabled and voltage when input are two different states.
> A disabled PWM will typically either drive high or low (depending on
> where it was when you turned it off).  Not all "disabled" states will
> mean that the pin is configured as an input.

Well, my point was that, from the regulator PoV, a PWM is either
enabled or disabled. The pinmux config when the PWM is disabled
depends on the PWM driver/hardware, and I don't think it's a good idea
to expose this information at the regulator level, hence the name
"voltage-when-pwm-disabled". Anyway, let's stop bikeshedding, no
matter the name, I think we both agree that a new DT property is needed
to handle this case :-).

^ permalink raw reply

* [RFC] arm64: Ensure proper addressing for ldnp/stnp
From: bdegraaf at codeaurora.org @ 2016-09-19 18:41 UTC (permalink / raw)
  To: linux-arm-kernel
In-Reply-To: <50fcc549-d785-6332-83bd-292a4f44e8d4@redhat.com>

On 2016-09-19 14:28, Laura Abbott wrote:
> On 09/19/2016 10:36 AM, Brent DeGraaf wrote:
>> According to section 6.3.8 of the ARM Programmer's Guide, non-temporal
>> loads and stores do not verify that address dependency is met between 
>> a
>> load of an address to a register and a subsequent non-temporal load or
>> store using that address on the executing PE. Therefore, context 
>> switch
>> code and subroutine calls that use non-temporally accessed addresses 
>> as
>> parameters that might depend on a load of an address into an argument
>> register must ensure that ordering requirements are met by introducing
>> a barrier prior to the successive non-temporal access.  Add 
>> appropriate
>> barriers whereever this specific situation comes into play.
>> 
> 
> Was this found by code inspection or is there a (public) exciting test
> case to observe this behavior?
> 
> Thanks,
> Laura
> 

Code inspection only.

Brent

^ permalink raw reply

* [PATCH] arm64, numa: Add cpu_to_node() implementation.
From: David Daney @ 2016-09-19 18:49 UTC (permalink / raw)
  To: linux-arm-kernel

From: David Daney <david.daney@cavium.com>

The wq_numa_init() function makes a private CPU to node map by calling
cpu_to_node() early in the boot process, before the non-boot CPUs are
brought online.  Since the default implementation of cpu_to_node()
returns zero for CPUs that have never been brought online, the
workqueue system's view is that *all* CPUs are on node zero.

When the unbound workqueue for a non-zero node is created, the
tsk_cpus_allowed() for the worker threads is the empty set because
there are, in the view of the workqueue system, no CPUs on non-zero
nodes.  The code in try_to_wake_up() using this empty cpumask ends up
using the cpumask empty set value of NR_CPUS as an index into the
per-CPU area pointer array, and gets garbage as it is one past the end
of the array.  This results in:

[    0.881970] Unable to handle kernel paging request at virtual address fffffb1008b926a4
[    1.970095] pgd = fffffc00094b0000
[    1.973530] [fffffb1008b926a4] *pgd=0000000000000000, *pud=0000000000000000, *pmd=0000000000000000
[    1.982610] Internal error: Oops: 96000004 [#1] SMP
[    1.987541] Modules linked in:
[    1.990631] CPU: 48 PID: 295 Comm: cpuhp/48 Tainted: G        W       4.8.0-rc6-preempt-vol+ #9
[    1.999435] Hardware name: Cavium ThunderX CN88XX board (DT)
[    2.005159] task: fffffe0fe89cc300 task.stack: fffffe0fe8b8c000
[    2.011158] PC is at try_to_wake_up+0x194/0x34c
[    2.015737] LR is at try_to_wake_up+0x150/0x34c
[    2.020318] pc : [<fffffc00080e7468>] lr : [<fffffc00080e7424>] pstate: 600000c5
[    2.027803] sp : fffffe0fe8b8fb10
[    2.031149] x29: fffffe0fe8b8fb10 x28: 0000000000000000
[    2.036522] x27: fffffc0008c63bc8 x26: 0000000000001000
[    2.041896] x25: fffffc0008c63c80 x24: fffffc0008bfb200
[    2.047270] x23: 00000000000000c0 x22: 0000000000000004
[    2.052642] x21: fffffe0fe89d25bc x20: 0000000000001000
[    2.058014] x19: fffffe0fe89d1d00 x18: 0000000000000000
[    2.063386] x17: 0000000000000000 x16: 0000000000000000
[    2.068760] x15: 0000000000000018 x14: 0000000000000000
[    2.074133] x13: 0000000000000000 x12: 0000000000000000
[    2.079505] x11: 0000000000000000 x10: 0000000000000000
[    2.084879] x9 : 0000000000000000 x8 : 0000000000000000
[    2.090251] x7 : 0000000000000040 x6 : 0000000000000000
[    2.095621] x5 : ffffffffffffffff x4 : 0000000000000000
[    2.100991] x3 : 0000000000000000 x2 : 0000000000000000
[    2.106364] x1 : fffffc0008be4c24 x0 : ffffff0ffffada80
[    2.111737]
[    2.113236] Process cpuhp/48 (pid: 295, stack limit = 0xfffffe0fe8b8c020)
[    2.120102] Stack: (0xfffffe0fe8b8fb10 to 0xfffffe0fe8b90000)
[    2.125914] fb00:                                   fffffe0fe8b8fb80 fffffc00080e7648
.
.
.
[    2.442859] Call trace:
[    2.445327] Exception stack(0xfffffe0fe8b8f940 to 0xfffffe0fe8b8fa70)
[    2.451843] f940: fffffe0fe89d1d00 0000040000000000 fffffe0fe8b8fb10 fffffc00080e7468
[    2.459767] f960: fffffe0fe8b8f980 fffffc00080e4958 ffffff0ff91ab200 fffffc00080e4b64
[    2.467690] f980: fffffe0fe8b8f9d0 fffffc00080e515c fffffe0fe8b8fa80 0000000000000000
[    2.475614] f9a0: fffffe0fe8b8f9d0 fffffc00080e58e4 fffffe0fe8b8fa80 0000000000000000
[    2.483540] f9c0: fffffe0fe8d10000 0000000000000040 fffffe0fe8b8fa50 fffffc00080e5ac4
[    2.491465] f9e0: ffffff0ffffada80 fffffc0008be4c24 0000000000000000 0000000000000000
[    2.499387] fa00: 0000000000000000 ffffffffffffffff 0000000000000000 0000000000000040
[    2.507309] fa20: 0000000000000000 0000000000000000 0000000000000000 0000000000000000
[    2.515233] fa40: 0000000000000000 0000000000000000 0000000000000000 0000000000000018
[    2.523156] fa60: 0000000000000000 0000000000000000
[    2.528089] [<fffffc00080e7468>] try_to_wake_up+0x194/0x34c
[    2.533723] [<fffffc00080e7648>] wake_up_process+0x28/0x34
[    2.539275] [<fffffc00080d3764>] create_worker+0x110/0x19c
[    2.544824] [<fffffc00080d69dc>] alloc_unbound_pwq+0x3cc/0x4b0
[    2.550724] [<fffffc00080d6bcc>] wq_update_unbound_numa+0x10c/0x1e4
[    2.557066] [<fffffc00080d7d78>] workqueue_online_cpu+0x220/0x28c
[    2.563234] [<fffffc00080bd288>] cpuhp_invoke_callback+0x6c/0x168
[    2.569398] [<fffffc00080bdf74>] cpuhp_up_callbacks+0x44/0xe4
[    2.575210] [<fffffc00080be194>] cpuhp_thread_fun+0x13c/0x148
[    2.581027] [<fffffc00080dfbac>] smpboot_thread_fn+0x19c/0x1a8
[    2.586929] [<fffffc00080dbd64>] kthread+0xdc/0xf0
[    2.591776] [<fffffc0008083380>] ret_from_fork+0x10/0x50
[    2.597147] Code: b00057e1 91304021 91005021 b8626822 (b8606821)
[    2.603464] ---[ end trace 58c0cd36b88802bc ]---
[    2.608138] Kernel panic - not syncing: Fatal exception

Fix by supplying a cpu_to_node() implementation that returns correct
node mappings.

Cc: <stable@vger.kernel.org> # 4.7.x-
Signed-off-by: David Daney <david.daney@cavium.com>

---
 arch/arm64/include/asm/topology.h |  3 +++
 arch/arm64/mm/numa.c              | 18 ++++++++++++++++++
 2 files changed, 21 insertions(+)

diff --git a/arch/arm64/include/asm/topology.h b/arch/arm64/include/asm/topology.h
index 8b57339..8d935447 100644
--- a/arch/arm64/include/asm/topology.h
+++ b/arch/arm64/include/asm/topology.h
@@ -30,6 +30,9 @@ int pcibus_to_node(struct pci_bus *bus);
 				 cpu_all_mask :				\
 				 cpumask_of_node(pcibus_to_node(bus)))
 
+int cpu_to_node(int cpu);
+#define cpu_to_node cpu_to_node
+
 #endif /* CONFIG_NUMA */
 
 #include <asm-generic/topology.h>
diff --git a/arch/arm64/mm/numa.c b/arch/arm64/mm/numa.c
index 5bb15ea..e76281b 100644
--- a/arch/arm64/mm/numa.c
+++ b/arch/arm64/mm/numa.c
@@ -130,6 +130,24 @@ void __init early_map_cpu_to_node(unsigned int cpu, int nid)
 	cpu_to_node_map[cpu] = nid;
 }
 
+int cpu_to_node(int cpu)
+{
+	int nid;
+
+	/*
+	 * Return 0 for unknown mapping so that we report something
+	 * sensible if firmware doesn't supply a proper mapping.
+	 */
+	if (cpu < 0 || cpu >= NR_CPUS)
+		return 0;
+
+	nid = cpu_to_node_map[cpu];
+	if (nid == NUMA_NO_NODE)
+		nid = 0;
+	return nid;
+}
+EXPORT_SYMBOL(cpu_to_node);
+
 /**
  * numa_add_memblk - Set node id to memblk
  * @nid: NUMA node ID of the new memblk
-- 
1.8.3.1

^ permalink raw reply related

* [PATCH v3] ASoC: exynos: organize the asoc audio into a menu
From: Randy Li @ 2016-09-19 18:57 UTC (permalink / raw)
  To: linux-arm-kernel

It is simple sound card time, we could assign different codec
to a interface without making a specific driver for it. The SPDIF
and I2S interface for Samsung would be possible used by
simple-sound-card, but not sure about the PCM.

Those S3C time entries are left alone as I don't think any new board
would need them.

Signed-off-by: Randy Li <ayaka@soulik.info>
---
 sound/soc/samsung/Kconfig | 57 ++++++++++++++++++++++++-----------------------
 1 file changed, 29 insertions(+), 28 deletions(-)

diff --git a/sound/soc/samsung/Kconfig b/sound/soc/samsung/Kconfig
index 7b722b0..f6023b4 100644
--- a/sound/soc/samsung/Kconfig
+++ b/sound/soc/samsung/Kconfig
@@ -1,12 +1,14 @@
-config SND_SOC_SAMSUNG
+menuconfig SND_SOC_SAMSUNG
 	tristate "ASoC support for Samsung"
 	depends on (PLAT_SAMSUNG || ARCH_EXYNOS)
 	select SND_SOC_GENERIC_DMAENGINE_PCM
-	help
+	---help---
 	  Say Y or M if you want to add support for codecs attached to
 	  the Samsung SoCs' Audio interfaces. You will also need to
 	  select the audio interfaces to support below.
 
+if SND_SOC_SAMSUNG
+
 config SND_S3C24XX_I2S
 	tristate
 
@@ -18,22 +20,22 @@ config SND_S3C2412_SOC_I2S
 	select SND_S3C_I2SV2_SOC
 
 config SND_SAMSUNG_PCM
-	tristate
+	tristate "Samsung PCM interface support"
 
 config SND_SAMSUNG_AC97
 	tristate
 	select SND_SOC_AC97_BUS
 
 config SND_SAMSUNG_SPDIF
-	tristate
+	tristate "Samsung SPDIF transmitter support"
 	select SND_SOC_SPDIF
 
 config SND_SAMSUNG_I2S
-	tristate
+	tristate "Samsung I2S interface support"
 
 config SND_SOC_SAMSUNG_NEO1973_WM8753
 	tristate "Audio support for Openmoko Neo1973 Smartphones (GTA02)"
-	depends on SND_SOC_SAMSUNG && MACH_NEO1973_GTA02
+	depends on MACH_NEO1973_GTA02
 	select SND_S3C24XX_I2S
 	select SND_SOC_WM8753
 	select SND_SOC_BT_SCO
@@ -43,7 +45,7 @@ config SND_SOC_SAMSUNG_NEO1973_WM8753
 
 config SND_SOC_SAMSUNG_JIVE_WM8750
 	tristate "SoC I2S Audio support for Jive"
-	depends on SND_SOC_SAMSUNG && MACH_JIVE && I2C
+	depends on MACH_JIVE && I2C
 	select SND_SOC_WM8750
 	select SND_S3C2412_SOC_I2S
 	help
@@ -51,7 +53,7 @@ config SND_SOC_SAMSUNG_JIVE_WM8750
 
 config SND_SOC_SAMSUNG_SMDK_WM8580
 	tristate "SoC I2S Audio support for WM8580 on SMDK"
-	depends on SND_SOC_SAMSUNG && (MACH_SMDK6410 || MACH_SMDKC100 || MACH_SMDKV210 || MACH_SMDKC110)
+	depends on MACH_SMDK6410 || MACH_SMDKC100 || MACH_SMDKV210 || MACH_SMDKC110
 	depends on I2C
 	select SND_SOC_WM8580
 	select SND_SAMSUNG_I2S
@@ -60,7 +62,6 @@ config SND_SOC_SAMSUNG_SMDK_WM8580
 
 config SND_SOC_SAMSUNG_SMDK_WM8994
 	tristate "SoC I2S Audio support for WM8994 on SMDK"
-	depends on SND_SOC_SAMSUNG
 	depends on I2C=y
 	select MFD_WM8994
 	select SND_SOC_WM8994
@@ -70,7 +71,7 @@ config SND_SOC_SAMSUNG_SMDK_WM8994
 
 config SND_SOC_SAMSUNG_SMDK2443_WM9710
 	tristate "SoC AC97 Audio support for SMDK2443 - WM9710"
-	depends on SND_SOC_SAMSUNG && MACH_SMDK2443
+	depends on MACH_SMDK2443
 	select AC97_BUS
 	select SND_SOC_AC97_CODEC
 	select SND_SAMSUNG_AC97
@@ -80,7 +81,7 @@ config SND_SOC_SAMSUNG_SMDK2443_WM9710
 
 config SND_SOC_SAMSUNG_LN2440SBC_ALC650
 	tristate "SoC AC97 Audio support for LN2440SBC - ALC650"
-	depends on SND_SOC_SAMSUNG && ARCH_S3C24XX
+	depends on ARCH_S3C24XX
 	select AC97_BUS
 	select SND_SOC_AC97_CODEC
 	select SND_SAMSUNG_AC97
@@ -90,7 +91,7 @@ config SND_SOC_SAMSUNG_LN2440SBC_ALC650
 
 config SND_SOC_SAMSUNG_S3C24XX_UDA134X
 	tristate "SoC I2S Audio support UDA134X wired to a S3C24XX"
-	depends on SND_SOC_SAMSUNG && ARCH_S3C24XX
+	depends on ARCH_S3C24XX
 	select SND_S3C24XX_I2S
 	select SND_SOC_L3
 	select SND_SOC_UDA134X
@@ -102,21 +103,21 @@ config SND_SOC_SAMSUNG_SIMTEC
 
 config SND_SOC_SAMSUNG_SIMTEC_TLV320AIC23
 	tristate "SoC I2S Audio support for TLV320AIC23 on Simtec boards"
-	depends on SND_SOC_SAMSUNG && ARCH_S3C24XX && I2C
+	depends on ARCH_S3C24XX && I2C
 	select SND_S3C24XX_I2S
 	select SND_SOC_TLV320AIC23_I2C
 	select SND_SOC_SAMSUNG_SIMTEC
 
 config SND_SOC_SAMSUNG_SIMTEC_HERMES
 	tristate "SoC I2S Audio support for Simtec Hermes board"
-	depends on SND_SOC_SAMSUNG && ARCH_S3C24XX && I2C
+	depends on ARCH_S3C24XX && I2C
 	select SND_S3C24XX_I2S
 	select SND_SOC_TLV320AIC3X
 	select SND_SOC_SAMSUNG_SIMTEC
 
 config SND_SOC_SAMSUNG_H1940_UDA1380
 	tristate "Audio support for the HP iPAQ H1940"
-	depends on SND_SOC_SAMSUNG && ARCH_H1940 && I2C
+	depends on ARCH_H1940 && I2C
 	select SND_S3C24XX_I2S
 	select SND_SOC_UDA1380
 	help
@@ -124,7 +125,7 @@ config SND_SOC_SAMSUNG_H1940_UDA1380
 
 config SND_SOC_SAMSUNG_RX1950_UDA1380
 	tristate "Audio support for the HP iPAQ RX1950"
-	depends on SND_SOC_SAMSUNG && MACH_RX1950 && I2C
+	depends on MACH_RX1950 && I2C
 	select SND_S3C24XX_I2S
 	select SND_SOC_UDA1380
 	help
@@ -132,7 +133,7 @@ config SND_SOC_SAMSUNG_RX1950_UDA1380
 
 config SND_SOC_SAMSUNG_SMDK_WM9713
 	tristate "SoC AC97 Audio support for SMDK with WM9713"
-	depends on SND_SOC_SAMSUNG && (MACH_SMDK6410 || MACH_SMDKC100 || MACH_SMDKV210 || MACH_SMDKC110)
+	depends on MACH_SMDK6410 || MACH_SMDKC100 || MACH_SMDKV210 || MACH_SMDKC110
 	select SND_SOC_WM9713
 	select SND_SAMSUNG_AC97
 	help
@@ -140,20 +141,19 @@ config SND_SOC_SAMSUNG_SMDK_WM9713
 
 config SND_SOC_SMARTQ
 	tristate "SoC I2S Audio support for SmartQ board"
-	depends on SND_SOC_SAMSUNG && MACH_SMARTQ && I2C
+	depends on MACH_SMARTQ && I2C
 	select SND_SAMSUNG_I2S
 	select SND_SOC_WM8750
 
 config SND_SOC_SAMSUNG_SMDK_SPDIF
 	tristate "SoC S/PDIF Audio support for SMDK"
-	depends on SND_SOC_SAMSUNG
 	select SND_SAMSUNG_SPDIF
 	help
 	  Say Y if you want to add support for SoC S/PDIF audio on the SMDK.
 
 config SND_SOC_SMDK_WM8580_PCM
 	tristate "SoC PCM Audio support for WM8580 on SMDK"
-	depends on SND_SOC_SAMSUNG && (MACH_SMDKV210 || MACH_SMDKC110)
+	depends on MACH_SMDKV210 || MACH_SMDKC110
 	depends on I2C
 	select SND_SOC_WM8580
 	select SND_SAMSUNG_PCM
@@ -162,7 +162,6 @@ config SND_SOC_SMDK_WM8580_PCM
 
 config SND_SOC_SMDK_WM8994_PCM
 	tristate "SoC PCM Audio support for WM8994 on SMDK"
-	depends on SND_SOC_SAMSUNG
 	depends on I2C=y
 	select MFD_WM8994
 	select SND_SOC_WM8994
@@ -172,7 +171,7 @@ config SND_SOC_SMDK_WM8994_PCM
 
 config SND_SOC_SPEYSIDE
 	tristate "Audio support for Wolfson Speyside"
-	depends on SND_SOC_SAMSUNG && I2C && SPI_MASTER
+	depends on I2C && SPI_MASTER
 	depends on MACH_WLF_CRAGG_6410 || COMPILE_TEST
 	select SND_SAMSUNG_I2S
 	select SND_SOC_WM8996
@@ -182,14 +181,14 @@ config SND_SOC_SPEYSIDE
 
 config SND_SOC_TOBERMORY
 	tristate "Audio support for Wolfson Tobermory"
-	depends on SND_SOC_SAMSUNG && INPUT && I2C
+	depends on INPUT && I2C
 	depends on MACH_WLF_CRAGG_6410 || COMPILE_TEST
 	select SND_SAMSUNG_I2S
 	select SND_SOC_WM8962
 
 config SND_SOC_BELLS
 	tristate "Audio support for Wolfson Bells"
-	depends on SND_SOC_SAMSUNG && MFD_ARIZONA && I2C && SPI_MASTER
+	depends on MFD_ARIZONA && I2C && SPI_MASTER
 	depends on MACH_WLF_CRAGG_6410 || COMPILE_TEST
 	select SND_SAMSUNG_I2S
 	select SND_SOC_WM5102
@@ -200,7 +199,7 @@ config SND_SOC_BELLS
 
 config SND_SOC_LOWLAND
 	tristate "Audio support for Wolfson Lowland"
-	depends on SND_SOC_SAMSUNG && I2C
+	depends on I2C
 	depends on MACH_WLF_CRAGG_6410 || COMPILE_TEST
 	select SND_SAMSUNG_I2S
 	select SND_SOC_WM5100
@@ -208,7 +207,7 @@ config SND_SOC_LOWLAND
 
 config SND_SOC_LITTLEMILL
 	tristate "Audio support for Wolfson Littlemill"
-	depends on SND_SOC_SAMSUNG && I2C
+	depends on I2C
 	depends on MACH_WLF_CRAGG_6410 || COMPILE_TEST
 	select SND_SAMSUNG_I2S
 	select MFD_WM8994
@@ -216,7 +215,7 @@ config SND_SOC_LITTLEMILL
 
 config SND_SOC_SNOW
 	tristate "Audio support for Google Snow boards"
-	depends on SND_SOC_SAMSUNG && I2C
+	depends on I2C
 	select SND_SOC_MAX98090
 	select SND_SOC_MAX98095
 	select SND_SAMSUNG_I2S
@@ -226,6 +225,8 @@ config SND_SOC_SNOW
 
 config SND_SOC_ARNDALE_RT5631_ALC5631
         tristate "Audio support for RT5631(ALC5631) on Arndale Board"
-        depends on SND_SOC_SAMSUNG && I2C
+        depends on I2C
         select SND_SAMSUNG_I2S
         select SND_SOC_RT5631
+
+endif #SND_SOC_SAMSUNG
-- 
2.7.4

^ permalink raw reply related

* [RFC PATCH v3 0/2] adding panel claa070wp03xg support for exynos
From: Randy Li @ 2016-09-19 19:02 UTC (permalink / raw)
  To: linux-arm-kernel

The timings issue is still here, this version is just some modifications
in dts file.

Randy Li (2):
  ARM: dts: samsung: add rga-lvds panel in itop elite
  drm/panel: Add support for Chunghwa CLAA070WP03XG panel

 .../display/panel/chunghwa,claa070wp03xg.txt       |  7 +++
 arch/arm/boot/dts/exynos4412-itop-elite.dts        | 54 +++++++++++++++++++++-
 drivers/gpu/drm/panel/panel-simple.c               | 27 +++++++++++
 3 files changed, 86 insertions(+), 2 deletions(-)
 create mode 100644 Documentation/devicetree/bindings/display/panel/chunghwa,claa070wp03xg.txt

-- 
2.7.4

^ permalink raw reply

* [RFC PATCH v3 1/2] ARM: dts: samsung: add rga-lvds panel in itop elite
From: Randy Li @ 2016-09-19 19:02 UTC (permalink / raw)
  To: linux-arm-kernel
In-Reply-To: <1474311771-23321-1-git-send-email-ayaka@soulik.info>

It is actually a lvds panel connected through a rga-lvds bridge.
The touchscreen is communicated with i2c bus but the driver is not
support now.

Signed-off-by: Randy Li <ayaka@soulik.info>
---
 arch/arm/boot/dts/exynos4412-itop-elite.dts | 54 +++++++++++++++++++++++++++--
 1 file changed, 52 insertions(+), 2 deletions(-)

diff --git a/arch/arm/boot/dts/exynos4412-itop-elite.dts b/arch/arm/boot/dts/exynos4412-itop-elite.dts
index b08705e..5153522 100644
--- a/arch/arm/boot/dts/exynos4412-itop-elite.dts
+++ b/arch/arm/boot/dts/exynos4412-itop-elite.dts
@@ -138,6 +138,36 @@
 		assigned-clocks = <&clock CLK_MOUT_CAM0>;
 		assigned-clock-parents = <&clock CLK_XUSBXTI>;
 	};
+
+	vcc_sys_lcd: sys-lcd {
+		compatible = "regulator-fixed";
+		regulator-name = "vcc_5v";
+		regulator-min-microvolt = <5000000>;
+		regulator-max-microvolt = <5000000>;
+		gpio = <&gpl0 4 GPIO_ACTIVE_HIGH>;
+	};
+
+	panel: panel at 0 {
+		compatible = "chunghwa,claa070wp03xg";
+
+		power-supply = <&vcc_sys_lcd>;
+		enable-gpios = <&gpl0 2 GPIO_ACTIVE_HIGH>;
+		backlight = <&bl>;
+
+		port {
+			lcd_ep: endpoint {
+				remote-endpoint = <&rga_lvds>;
+			};
+		};
+	};
+
+	bl: backlight {
+		compatible = "pwm-backlight";
+		pwms = <&pwm 1 5000000 PWM_POLARITY_INVERTED>;
+		brightness-levels = <0 5 12 16 32 64 128 255>;
+		default-brightness-level = <5>;
+		power-supply = <&vcc_sys_lcd>;
+	};
 };
 
 &adc {
@@ -171,11 +201,31 @@
 	assigned-clock-rates = <0>, <176000000>;
 };
 
+&fimd {
+	pinctrl-0 = <&lcd_clk &lcd_data24>;
+	pinctrl-names = "default";
+	status = "okay";
+	ports {
+		#address-cells = <1>;
+		#size-cells = <0>;
+		port at 3 {
+			reg = <3>;
+			rga_lvds: endpoint {
+				remote-endpoint = <&lcd_ep>;
+			};
+		};
+	};
+};
+
 &hsotg {
 	dr_mode = "peripheral";
 	status = "okay";
 };
 
+&i2c_3 {
+	status = "okay";
+};
+
 &i2c_4 {
 	samsung,i2c-sda-delay = <100>;
 	samsung,i2c-slave-addr = <0x10>;
@@ -215,9 +265,9 @@
 
 &pwm {
 	status = "okay";
-	pinctrl-0 = <&pwm0_out>;
+	pinctrl-0 = <&pwm0_out &pwm1_out>;
 	pinctrl-names = "default";
-	samsung,pwm-outputs = <0>;
+	samsung,pwm-outputs = <0>, <1>;
 };
 
 &sdhci_2 {
-- 
2.7.4

^ permalink raw reply related

* [RFC PATCH v3 2/2] drm/panel: Add support for Chunghwa CLAA070WP03XG panel
From: Randy Li @ 2016-09-19 19:02 UTC (permalink / raw)
  To: linux-arm-kernel
In-Reply-To: <1474311771-23321-1-git-send-email-ayaka@soulik.info>

The Chunghwa CLAA070WP03XG is a 7" 1280x800 panel, which can be
supported by the simple panel driver.

Signed-off-by: Randy Li <ayaka@soulik.info>
---
 .../display/panel/chunghwa,claa070wp03xg.txt       |  7 ++++++
 drivers/gpu/drm/panel/panel-simple.c               | 27 ++++++++++++++++++++++
 2 files changed, 34 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/display/panel/chunghwa,claa070wp03xg.txt

diff --git a/Documentation/devicetree/bindings/display/panel/chunghwa,claa070wp03xg.txt b/Documentation/devicetree/bindings/display/panel/chunghwa,claa070wp03xg.txt
new file mode 100644
index 0000000..dd22685
--- /dev/null
+++ b/Documentation/devicetree/bindings/display/panel/chunghwa,claa070wp03xg.txt
@@ -0,0 +1,7 @@
+Chunghwa Picture Tubes Ltd. 7" WXGA TFT LCD panel
+
+Required properties:
+- compatible: should be "chunghwa,claa070wp03xg"
+
+This binding is compatible with the simple-panel binding, which is specified
+in simple-panel.txt in this directory.
diff --git a/drivers/gpu/drm/panel/panel-simple.c b/drivers/gpu/drm/panel/panel-simple.c
index f178998..3204e6b 100644
--- a/drivers/gpu/drm/panel/panel-simple.c
+++ b/drivers/gpu/drm/panel/panel-simple.c
@@ -583,6 +583,30 @@ static const struct panel_desc avic_tm070ddh03 = {
 	},
 };
 
+static const struct drm_display_mode chunghwa_claa070wp03xg_mode = {
+	.clock = 66770,
+	.hdisplay = 800,
+	.hsync_start = 800 + 49,
+	.hsync_end = 800 + 49 + 33,
+	.htotal = 800 + 49 + 33 + 17,
+	.vdisplay = 1280,
+	.vsync_start = 1280 + 1,
+	.vsync_end = 1280 + 1 + 7,
+	.vtotal = 1280 + 1 + 7 + 15,
+	.vrefresh = 60,
+	.flags = DRM_MODE_FLAG_NVSYNC | DRM_MODE_FLAG_NHSYNC,
+};
+
+static const struct panel_desc chunghwa_claa070wp03xg = {
+	.modes = &chunghwa_claa070wp03xg_mode,
+	.num_modes = 1,
+	.bpc = 6,
+	.size = {
+		.width = 94,
+		.height = 150,
+	},
+};
+
 static const struct drm_display_mode chunghwa_claa101wa01a_mode = {
 	.clock = 72070,
 	.hdisplay = 1366,
@@ -1544,6 +1568,9 @@ static const struct of_device_id platform_of_match[] = {
 		.compatible = "avic,tm070ddh03",
 		.data = &avic_tm070ddh03,
 	}, {
+		.compatible = "chunghwa,claa070wp03xg",
+		.data = &chunghwa_claa070wp03xg,
+	}, {
 		.compatible = "chunghwa,claa101wa01a",
 		.data = &chunghwa_claa101wa01a
 	}, {
-- 
2.7.4

^ permalink raw reply related

* [PATCH 0/4] kill get_clock_tick_rate()
From: Robert Jarzmik @ 2016-09-19 19:12 UTC (permalink / raw)
  To: linux-arm-kernel

Hi,

This serie aims at killing get_clock_tick_rate() from pxa and sa1100 and replace
it by clock API calls.

I'd like to have this reviewed and acked afterwards at least by :
 - Russell for the mach-sa1100 and sa1100_wdt.c
 - Stephen or Michael for the clk-pxa25x.c
 - Thomas or Daniel for the clocksource part
 - Wim for sa1100_wdt.c

If the review converges, I'd also like to take it through the pxa/for-next tree,
or alternatively Russell's tree if there is a dependency I'm not seeing, for
v4.10 cycle.

Cheers.

Robert Jarzmik (3):
  ARM: sa11x0/pxa: acquire timer rate from the clock rate
  watchdog: sa11x0/pxa: get rid of get_clock_tick_rate
  ARM: sa11x0/pxa: get rid of get_clock_tick_rate

Russell King - ARM Linux (1):
  clk: pxa25x: OSTIMER0 clocks from the main oscillator

 arch/arm/mach-pxa/generic.c                  | 18 +-----------------
 arch/arm/mach-pxa/include/mach/hardware.h    |  2 --
 arch/arm/mach-sa1100/generic.c               |  2 +-
 arch/arm/mach-sa1100/include/mach/hardware.h |  4 ----
 drivers/clk/pxa/clk-pxa25x.c                 |  2 +-
 drivers/clocksource/pxa_timer.c              | 11 +++++------
 drivers/watchdog/sa1100_wdt.c                | 24 +++++++++++++++++++++++-
 include/clocksource/pxa.h                    |  3 +--
 8 files changed, 32 insertions(+), 34 deletions(-)

-- 
2.1.4

^ permalink raw reply

* [PATCH 1/4] clk: pxa25x: OSTIMER0 clocks from the main oscillator
From: Robert Jarzmik @ 2016-09-19 19:12 UTC (permalink / raw)
  To: linux-arm-kernel
In-Reply-To: <1474312335-20997-1-git-send-email-robert.jarzmik@free.fr>

From: Russell King - ARM Linux <linux@armlinux.org.uk>

The OSTIMER0 clock ticks at the main oscillator rate, not the 32kHz
oscillator rate.  Ensure that it is parented to the main oscillator.

Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>
Acked-by: Robert Jarzmik <robert.jarzmik@free.fr>
Tested-by: Robert Jarzmik <robert.jarzmik@free.fr>
---
 drivers/clk/pxa/clk-pxa25x.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/clk/pxa/clk-pxa25x.c b/drivers/clk/pxa/clk-pxa25x.c
index a98b98e2a9e4..56b0a6027e38 100644
--- a/drivers/clk/pxa/clk-pxa25x.c
+++ b/drivers/clk/pxa/clk-pxa25x.c
@@ -230,7 +230,7 @@ static struct dummy_clk dummy_clks[] __initdata = {
 	DUMMY_CLK("GPIO11_CLK", NULL, "osc_3_6864mhz"),
 	DUMMY_CLK("GPIO12_CLK", NULL, "osc_32_768khz"),
 	DUMMY_CLK(NULL, "sa1100-rtc", "osc_32_768khz"),
-	DUMMY_CLK("OSTIMER0", NULL, "osc_32_768khz"),
+	DUMMY_CLK("OSTIMER0", NULL, "osc_3_6864mhz"),
 	DUMMY_CLK("UARTCLK", "pxa2xx-ir", "STUART"),
 };
 
-- 
2.1.4

^ permalink raw reply related

* [PATCH 2/4] ARM: sa11x0/pxa: acquire timer rate from the clock rate
From: Robert Jarzmik @ 2016-09-19 19:12 UTC (permalink / raw)
  To: linux-arm-kernel
In-Reply-To: <1474312335-20997-1-git-send-email-robert.jarzmik@free.fr>

As both pxa and sa1100 provide a clock to the timer, the rate can be
inferred from the clock rather than hard encoded in a functional call.

This patch changes the pxa timer to have a mandatory clock which is used
as the timer rate.

Signed-off-by: Robert Jarzmik <robert.jarzmik@free.fr>
---
 arch/arm/mach-pxa/generic.c     |  3 +--
 arch/arm/mach-sa1100/generic.c  |  2 +-
 drivers/clocksource/pxa_timer.c | 11 +++++------
 include/clocksource/pxa.h       |  3 +--
 4 files changed, 8 insertions(+), 11 deletions(-)

diff --git a/arch/arm/mach-pxa/generic.c b/arch/arm/mach-pxa/generic.c
index ec510ecf8370..2a9297991d74 100644
--- a/arch/arm/mach-pxa/generic.c
+++ b/arch/arm/mach-pxa/generic.c
@@ -69,8 +69,7 @@ void __init pxa_timer_init(void)
 		pxa27x_clocks_init();
 	if (cpu_is_pxa3xx())
 		pxa3xx_clocks_init();
-	pxa_timer_nodt_init(IRQ_OST0, io_p2v(0x40a00000),
-			    get_clock_tick_rate());
+	pxa_timer_nodt_init(IRQ_OST0, io_p2v(0x40a00000));
 }
 
 /*
diff --git a/arch/arm/mach-sa1100/generic.c b/arch/arm/mach-sa1100/generic.c
index 3e09beddb6e8..2eb00691b07d 100644
--- a/arch/arm/mach-sa1100/generic.c
+++ b/arch/arm/mach-sa1100/generic.c
@@ -378,7 +378,7 @@ void __init sa1100_map_io(void)
 
 void __init sa1100_timer_init(void)
 {
-	pxa_timer_nodt_init(IRQ_OST0, io_p2v(0x90000000), 3686400);
+	pxa_timer_nodt_init(IRQ_OST0, io_p2v(0x90000000));
 }
 
 static struct resource irq_resource =
diff --git a/drivers/clocksource/pxa_timer.c b/drivers/clocksource/pxa_timer.c
index 3e1cb512f3ce..9cae38eebec2 100644
--- a/drivers/clocksource/pxa_timer.c
+++ b/drivers/clocksource/pxa_timer.c
@@ -220,17 +220,16 @@ CLOCKSOURCE_OF_DECLARE(pxa_timer, "marvell,pxa-timer", pxa_timer_dt_init);
 /*
  * Legacy timer init for non device-tree boards.
  */
-void __init pxa_timer_nodt_init(int irq, void __iomem *base,
-	unsigned long clock_tick_rate)
+void __init pxa_timer_nodt_init(int irq, void __iomem *base)
 {
 	struct clk *clk;
 
 	timer_base = base;
 	clk = clk_get(NULL, "OSTIMER0");
-	if (clk && !IS_ERR(clk))
+	if (clk && !IS_ERR(clk)) {
 		clk_prepare_enable(clk);
-	else
+		pxa_timer_common_init(irq, clk_get_rate(clk));
+	} else {
 		pr_crit("%s: unable to get clk\n", __func__);
-
-	pxa_timer_common_init(irq, clock_tick_rate);
+	}
 }
diff --git a/include/clocksource/pxa.h b/include/clocksource/pxa.h
index 1efbe5a66958..a9a0f03024a4 100644
--- a/include/clocksource/pxa.h
+++ b/include/clocksource/pxa.h
@@ -12,7 +12,6 @@
 #ifndef _CLOCKSOURCE_PXA_H
 #define _CLOCKSOURCE_PXA_H
 
-extern void pxa_timer_nodt_init(int irq, void __iomem *base,
-			   unsigned long clock_tick_rate);
+extern void pxa_timer_nodt_init(int irq, void __iomem *base);
 
 #endif
-- 
2.1.4

^ permalink raw reply related

* [PATCH 3/4] watchdog: sa11x0/pxa: get rid of get_clock_tick_rate
From: Robert Jarzmik @ 2016-09-19 19:12 UTC (permalink / raw)
  To: linux-arm-kernel
In-Reply-To: <1474312335-20997-1-git-send-email-robert.jarzmik@free.fr>

The OS timer rate used for the watchdog can now be fetched from the
standard clock API. This will remove the last user of
get_clock_tick_rate() in both pxa and sa11x0 architectures.

Signed-off-by: Robert Jarzmik <robert.jarzmik@free.fr>
---
 drivers/watchdog/sa1100_wdt.c | 24 +++++++++++++++++++++++-
 1 file changed, 23 insertions(+), 1 deletion(-)

diff --git a/drivers/watchdog/sa1100_wdt.c b/drivers/watchdog/sa1100_wdt.c
index e1d39a1e9628..8965e3f536c3 100644
--- a/drivers/watchdog/sa1100_wdt.c
+++ b/drivers/watchdog/sa1100_wdt.c
@@ -22,6 +22,7 @@
 
 #include <linux/module.h>
 #include <linux/moduleparam.h>
+#include <linux/clk.h>
 #include <linux/types.h>
 #include <linux/kernel.h>
 #include <linux/fs.h>
@@ -155,12 +156,27 @@ static struct miscdevice sa1100dog_miscdev = {
 };
 
 static int margin __initdata = 60;		/* (secs) Default is 1 minute */
+static struct clk *clk;
 
 static int __init sa1100dog_init(void)
 {
 	int ret;
 
-	oscr_freq = get_clock_tick_rate();
+	clk = clk_get(NULL, "OSTIMER0");
+	if (IS_ERR(clk)) {
+		pr_err("SA1100/PXA2xx Watchdog Timer: clock not found: %d\n",
+		       (int) PTR_ERR(clk));
+		return PTR_ERR(clk);
+	}
+
+	ret = clk_prepare_enable(clk);
+	if (ret) {
+		pr_err("SA1100/PXA2xx Watchdog Timer: clock failed to prepare+enable: %d\n",
+		       ret);
+		goto err;
+	}
+
+	oscr_freq = clk_get_rate(clk);
 
 	/*
 	 * Read the reset status, and save it for later.  If
@@ -176,11 +192,17 @@ static int __init sa1100dog_init(void)
 		pr_info("SA1100/PXA2xx Watchdog Timer: timer margin %d sec\n",
 			margin);
 	return ret;
+err:
+	clk_disable_unprepare(clk);
+	clk_put(clk);
+	return ret;
 }
 
 static void __exit sa1100dog_exit(void)
 {
 	misc_deregister(&sa1100dog_miscdev);
+	clk_disable_unprepare(clk);
+	clk_put(clk);
 }
 
 module_init(sa1100dog_init);
-- 
2.1.4

^ permalink raw reply related

* [PATCH 4/4] ARM: sa11x0/pxa: get rid of get_clock_tick_rate
From: Robert Jarzmik @ 2016-09-19 19:12 UTC (permalink / raw)
  To: linux-arm-kernel
In-Reply-To: <1474312335-20997-1-git-send-email-robert.jarzmik@free.fr>

The last user of this function is gone, so remove it. The clock API
should now be used to get clock rates.

Signed-off-by: Robert Jarzmik <robert.jarzmik@free.fr>
---
 arch/arm/mach-pxa/generic.c                  | 15 ---------------
 arch/arm/mach-pxa/include/mach/hardware.h    |  2 --
 arch/arm/mach-sa1100/include/mach/hardware.h |  4 ----
 3 files changed, 21 deletions(-)

diff --git a/arch/arm/mach-pxa/generic.c b/arch/arm/mach-pxa/generic.c
index 2a9297991d74..cb73a9723d0e 100644
--- a/arch/arm/mach-pxa/generic.c
+++ b/arch/arm/mach-pxa/generic.c
@@ -43,21 +43,6 @@ void clear_reset_status(unsigned int mask)
 	}
 }
 
-unsigned long get_clock_tick_rate(void)
-{
-	unsigned long clock_tick_rate;
-
-	if (cpu_is_pxa25x())
-		clock_tick_rate = 3686400;
-	else if (machine_is_mainstone())
-		clock_tick_rate = 3249600;
-	else
-		clock_tick_rate = 3250000;
-
-	return clock_tick_rate;
-}
-EXPORT_SYMBOL(get_clock_tick_rate);
-
 /*
  * For non device-tree builds, keep legacy timer init
  */
diff --git a/arch/arm/mach-pxa/include/mach/hardware.h b/arch/arm/mach-pxa/include/mach/hardware.h
index 8d63c211b22f..55064124ca4e 100644
--- a/arch/arm/mach-pxa/include/mach/hardware.h
+++ b/arch/arm/mach-pxa/include/mach/hardware.h
@@ -303,8 +303,6 @@
  */
 extern unsigned int get_memclk_frequency_10khz(void);
 
-/* return the clock tick rate of the OS timer */
-extern unsigned long get_clock_tick_rate(void);
 #endif
 
 #endif  /* _ASM_ARCH_HARDWARE_H */
diff --git a/arch/arm/mach-sa1100/include/mach/hardware.h b/arch/arm/mach-sa1100/include/mach/hardware.h
index cbedd75a9d65..474844a93e29 100644
--- a/arch/arm/mach-sa1100/include/mach/hardware.h
+++ b/arch/arm/mach-sa1100/include/mach/hardware.h
@@ -61,10 +61,6 @@
 # define __REG(x)	(*((volatile unsigned long __iomem *)io_p2v(x)))
 # define __PREG(x)	(io_v2p((unsigned long)&(x)))
 
-static inline unsigned long get_clock_tick_rate(void)
-{
-	return 3686400;
-}
 #else
 
 # define __REG(x)	io_p2v(x)
-- 
2.1.4

^ permalink raw reply related


This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox