big LITTLE is ARM's new Architecture focussing power/performance needs of modern
world. More information about big LITTLE can be found here:
http://www.arm.com/products/processors/technologies/biglittleprocessing.php
http://lwn.net/Articles/481055/
In order to keep cpufreq support for all big LITTLE platforms simple/generic,
this patch tries to add a generic cpufreq driver layer for all big LITTLE
platforms.
The driver is divided into two parts:
- Core driver: Generic and shared across all big LITTLE SoC's
- Glue drivers: Per platform drivers providing ops to the core driver
This patch adds in a generic glue driver which would extract information from
Device Tree.
Future SoC's can either reuse the DT glue or write their own depending on the
need.
Signed-off-by: Sudeep KarkadaNagesha <[email protected]>
Signed-off-by: Viresh Kumar <[email protected]>
---
This is pushed here:
http://git.linaro.org/gitweb?p=people/vireshk/linux.git;a=shortlog;h=refs/heads/cpufreq-biglittle
.../bindings/cpufreq/arm_big_little_dt.txt | 29 +++
MAINTAINERS | 11 +
drivers/cpufreq/Kconfig.arm | 13 +
drivers/cpufreq/Makefile | 4 +
drivers/cpufreq/arm_big_little.c | 278 +++++++++++++++++++++
drivers/cpufreq/arm_big_little.h | 40 +++
drivers/cpufreq/arm_big_little_dt.c | 125 +++++++++
7 files changed, 500 insertions(+)
create mode 100644 Documentation/devicetree/bindings/cpufreq/arm_big_little_dt.txt
create mode 100644 drivers/cpufreq/arm_big_little.c
create mode 100644 drivers/cpufreq/arm_big_little.h
create mode 100644 drivers/cpufreq/arm_big_little_dt.c
diff --git a/Documentation/devicetree/bindings/cpufreq/arm_big_little_dt.txt b/Documentation/devicetree/bindings/cpufreq/arm_big_little_dt.txt
new file mode 100644
index 0000000..6f534eb
--- /dev/null
+++ b/Documentation/devicetree/bindings/cpufreq/arm_big_little_dt.txt
@@ -0,0 +1,29 @@
+Generic ARM big LITTLE cpufreq driver's DT glue
+-----------------------------------------------
+
+It is DT specific glue layer for generic cpufreq driver for big LITTLE systems.
+
+Both required and optional properties listed below must be defined under node
+cluster*. * can be 0 or 1.
+
+Required properties:
+- freqs: List of all supported frequencies.
+
+Optional properties:
+- clock-latency: Specify the possible maximum transition latency for clock, in
+ unit of nanoseconds.
+
+Examples:
+
+cluster0: cluster@0 {
+ ..
+
+ freqs = <500000000 600000000 700000000 800000000 900000000 1000000000 1100000000 1200000000>;
+ clock-latency = <200000>;
+
+ ..
+
+ cores {
+ ..
+ };
+};
diff --git a/MAINTAINERS b/MAINTAINERS
index 554fd30..b14b749 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -2206,6 +2206,17 @@ S: Maintained
F: drivers/cpufreq/
F: include/linux/cpufreq.h
+CPU FREQUENCY DRIVERS - ARM BIG LITTLE
+M: Viresh Kumar <[email protected]>
+M: Sudeep KarkadaNagesha <[email protected]>
+L: [email protected]
+L: [email protected]
+W: http://www.arm.com/products/processors/technologies/biglittleprocessing.php
+S: Maintained
+F: drivers/cpufreq/arm_big_little.h
+F: drivers/cpufreq/arm_big_little.c
+F: drivers/cpufreq/arm_big_little_dt.c
+
CPUID/MSR DRIVER
M: "H. Peter Anvin" <[email protected]>
S: Maintained
diff --git a/drivers/cpufreq/Kconfig.arm b/drivers/cpufreq/Kconfig.arm
index 030ddf6..fdf54a9 100644
--- a/drivers/cpufreq/Kconfig.arm
+++ b/drivers/cpufreq/Kconfig.arm
@@ -2,6 +2,19 @@
# ARM CPU Frequency scaling drivers
#
+config ARM_BIG_LITTLE_CPUFREQ
+ tristate
+ depends on ARM_CPU_TOPOLOGY
+
+config ARM_DT_BL_CPUFREQ
+ tristate "Generic ARM big LITTLE CPUfreq driver probed via DT"
+ select ARM_BIG_LITTLE_CPUFREQ
+ depends on OF
+ default n
+ help
+ This enables the Generic CPUfreq driver for ARM big.LITTLE platform.
+ This gets frequency tables from DT.
+
config ARM_OMAP2PLUS_CPUFREQ
bool "TI OMAP2+"
depends on ARCH_OMAP2PLUS
diff --git a/drivers/cpufreq/Makefile b/drivers/cpufreq/Makefile
index 863fd18..d1b0832 100644
--- a/drivers/cpufreq/Makefile
+++ b/drivers/cpufreq/Makefile
@@ -44,6 +44,10 @@ obj-$(CONFIG_X86_INTEL_PSTATE) += intel_pstate.o
##################################################################################
# ARM SoC drivers
+obj-$(CONFIG_ARM_BIG_LITTLE_CPUFREQ) += arm_big_little.o
+# big LITTLE per platform glues. Keep DT_BL_CPUFREQ as the last entry in all big
+# LITTLE drivers, so that it is probed last.
+obj-$(CONFIG_ARM_DT_BL_CPUFREQ) += arm_big_little_dt.o
obj-$(CONFIG_UX500_SOC_DB8500) += dbx500-cpufreq.o
obj-$(CONFIG_ARM_S3C2416_CPUFREQ) += s3c2416-cpufreq.o
obj-$(CONFIG_ARM_S3C64XX_CPUFREQ) += s3c64xx-cpufreq.o
diff --git a/drivers/cpufreq/arm_big_little.c b/drivers/cpufreq/arm_big_little.c
new file mode 100644
index 0000000..0d6de0e
--- /dev/null
+++ b/drivers/cpufreq/arm_big_little.c
@@ -0,0 +1,278 @@
+/*
+ * ARM big.LITTLE Platforms CPUFreq support
+ *
+ * Copyright (C) 2013 ARM Ltd.
+ * Sudeep KarkadaNagesha <[email protected]>
+ *
+ * Copyright (C) 2013 Linaro.
+ * Viresh Kumar <[email protected]>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed "as is" WITHOUT ANY WARRANTY of any
+ * kind, whether express or implied; without even the implied warranty
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ */
+
+#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
+
+#include <linux/clk.h>
+#include <linux/cpufreq.h>
+#include <linux/cpumask.h>
+#include <linux/export.h>
+#include <linux/of_platform.h>
+#include <linux/slab.h>
+#include <linux/topology.h>
+#include <linux/types.h>
+
+#include "arm_big_little.h"
+
+/* Currently we support only two clusters */
+#define MAX_CLUSTERS 2
+
+static struct cpufreq_arm_bL_ops *arm_bL_ops;
+static struct clk *clk[MAX_CLUSTERS];
+static struct cpufreq_frequency_table *freq_table[MAX_CLUSTERS];
+static atomic_t cluster_usage[MAX_CLUSTERS] = {ATOMIC_INIT(0), ATOMIC_INIT(0)};
+
+static int cpu_to_cluster(int cpu)
+{
+ return topology_physical_package_id(cpu);
+}
+
+static unsigned int bL_cpufreq_get(unsigned int cpu)
+{
+ u32 cur_cluster = cpu_to_cluster(cpu);
+
+ return clk_get_rate(clk[cur_cluster]) / 1000;
+}
+
+/* Validate policy frequency range */
+static int bL_cpufreq_verify_policy(struct cpufreq_policy *policy)
+{
+ u32 cur_cluster = cpu_to_cluster(policy->cpu);
+
+ return cpufreq_frequency_table_verify(policy, freq_table[cur_cluster]);
+}
+
+/* Set clock frequency */
+static int bL_cpufreq_set_target(struct cpufreq_policy *policy,
+ unsigned int target_freq, unsigned int relation)
+{
+ struct cpufreq_freqs freqs;
+ u32 cpu = policy->cpu, freq_tab_idx, cur_cluster;
+ int ret = 0;
+
+ cur_cluster = cpu_to_cluster(policy->cpu);
+
+ freqs.old = bL_cpufreq_get(policy->cpu);
+
+ /* Determine valid target frequency using freq_table */
+ cpufreq_frequency_table_target(policy, freq_table[cur_cluster],
+ target_freq, relation, &freq_tab_idx);
+ freqs.new = freq_table[cur_cluster][freq_tab_idx].frequency;
+
+ freqs.cpu = policy->cpu;
+
+ pr_debug("%s: cpu: %d, cluster: %d, oldfreq: %d, target freq: %d, new freq: %d\n",
+ __func__, cpu, cur_cluster, freqs.old, target_freq,
+ freqs.new);
+
+ if (freqs.old == freqs.new)
+ return 0;
+
+ for_each_cpu(freqs.cpu, policy->cpus)
+ cpufreq_notify_transition(&freqs, CPUFREQ_PRECHANGE);
+
+ ret = clk_set_rate(clk[cur_cluster], freqs.new * 1000);
+ if (ret) {
+ pr_err("clk_set_rate failed: %d\n", ret);
+ return ret;
+ }
+
+ policy->cur = freqs.new;
+
+ for_each_cpu(freqs.cpu, policy->cpus)
+ cpufreq_notify_transition(&freqs, CPUFREQ_POSTCHANGE);
+
+ return ret;
+}
+
+/* translate the integer array into cpufreq_frequency_table entries */
+struct cpufreq_frequency_table *
+arm_bL_copy_table_from_array(unsigned int *table, int count)
+{
+ int i;
+
+ struct cpufreq_frequency_table *freq_table;
+
+ pr_debug("%s: table: %p, count: %d\n", __func__, table, count);
+
+ freq_table = kmalloc(sizeof(*freq_table) * (count + 1), GFP_KERNEL);
+ if (!freq_table)
+ return NULL;
+
+ for (i = 0; i < count; i++) {
+ pr_debug("%s: index: %d, freq: %d\n", __func__, i, table[i]);
+ freq_table[i].index = i;
+ freq_table[i].frequency = table[i]; /* in kHZ */
+ }
+
+ freq_table[i].index = count;
+ freq_table[i].frequency = CPUFREQ_TABLE_END;
+
+ return freq_table;
+}
+EXPORT_SYMBOL_GPL(arm_bL_copy_table_from_array);
+
+void arm_bL_free_freq_table(u32 cluster)
+{
+ pr_debug("%s: free freq table\n", __func__);
+
+ kfree(freq_table[cluster]);
+}
+EXPORT_SYMBOL_GPL(arm_bL_free_freq_table);
+
+static void put_cluster_clk_and_freq_table(u32 cluster)
+{
+ if (!atomic_dec_return(&cluster_usage[cluster])) {
+ clk_put(clk[cluster]);
+ clk[cluster] = NULL;
+ arm_bL_ops->put_freq_tbl(cluster);
+ freq_table[cluster] = NULL;
+ pr_debug("%s: cluster: %d\n", __func__, cluster);
+ }
+}
+
+static int get_cluster_clk_and_freq_table(u32 cluster)
+{
+ char name[9] = "cluster";
+ int count;
+
+ if (atomic_inc_return(&cluster_usage[cluster]) != 1)
+ return 0;
+
+ freq_table[cluster] = arm_bL_ops->get_freq_tbl(cluster, &count);
+ if (!freq_table[cluster])
+ goto atomic_dec;
+
+ name[7] = cluster + '0';
+ clk[cluster] = clk_get(NULL, name);
+ if (!IS_ERR_OR_NULL(clk[cluster])) {
+ pr_debug("%s: clk: %p & freq table: %p, cluster: %d\n",
+ __func__, clk[cluster], freq_table[cluster],
+ cluster);
+ return 0;
+ }
+
+ arm_bL_ops->put_freq_tbl(cluster);
+
+atomic_dec:
+ atomic_dec(&cluster_usage[cluster]);
+ pr_err("%s: Failed to get data for cluster: %d\n", __func__, cluster);
+ return -ENODATA;
+}
+
+/* Per-CPU initialization */
+static int bL_cpufreq_init(struct cpufreq_policy *policy)
+{
+ u32 cur_cluster = cpu_to_cluster(policy->cpu);
+ int result;
+
+ result = get_cluster_clk_and_freq_table(cur_cluster);
+ if (result)
+ return result;
+
+ result = cpufreq_frequency_table_cpuinfo(policy,
+ freq_table[cur_cluster]);
+ if (result) {
+ pr_err("CPU %d, cluster: %d invalid freq table\n", policy->cpu,
+ cur_cluster);
+ put_cluster_clk_and_freq_table(cur_cluster);
+ return result;
+ }
+
+ cpufreq_frequency_table_get_attr(freq_table[cur_cluster], policy->cpu);
+
+ policy->cpuinfo.transition_latency =
+ arm_bL_ops->get_transition_latency(cur_cluster);
+ policy->cur = bL_cpufreq_get(policy->cpu);
+
+ cpumask_copy(policy->cpus, topology_core_cpumask(policy->cpu));
+
+ pr_info("CPU %d initialized\n", policy->cpu);
+ return 0;
+}
+
+static int bL_cpufreq_exit(struct cpufreq_policy *policy)
+{
+ put_cluster_clk_and_freq_table(cpu_to_cluster(policy->cpu));
+ pr_debug("%s: Exited, cpu: %d\n", __func__, policy->cpu);
+
+ return 0;
+}
+
+/* Export freq_table to sysfs */
+static struct freq_attr *bL_cpufreq_attr[] = {
+ &cpufreq_freq_attr_scaling_available_freqs,
+ NULL,
+};
+
+static struct cpufreq_driver bL_cpufreq_driver = {
+ .name = "arm-big-little",
+ .flags = CPUFREQ_STICKY,
+ .verify = bL_cpufreq_verify_policy,
+ .target = bL_cpufreq_set_target,
+ .get = bL_cpufreq_get,
+ .init = bL_cpufreq_init,
+ .exit = bL_cpufreq_exit,
+ .attr = bL_cpufreq_attr,
+};
+
+int bL_cpufreq_register(struct cpufreq_arm_bL_ops *ops)
+{
+ int ret;
+
+ if (arm_bL_ops) {
+ pr_debug("%s: Already registered: %s, exiting\n", __func__,
+ arm_bL_ops->name);
+ return -EBUSY;
+ }
+
+ if (!ops || !strlen(ops->name) || !ops->get_freq_tbl) {
+ pr_err("%s: Invalid arm_bL_ops, exiting\n", __func__);
+ return -ENODEV;
+ }
+
+ arm_bL_ops = ops;
+
+ ret = cpufreq_register_driver(&bL_cpufreq_driver);
+ if (ret) {
+ pr_info("%s: Failed registering platform driver: %s, err: %d\n",
+ __func__, ops->name, ret);
+ arm_bL_ops = NULL;
+ } else {
+ pr_info("%s: Registered platform driver: %s\n", __func__,
+ ops->name);
+ }
+
+ return ret;
+}
+EXPORT_SYMBOL_GPL(bL_cpufreq_register);
+
+void bL_cpufreq_unregister(struct cpufreq_arm_bL_ops *ops)
+{
+ if (arm_bL_ops != ops) {
+ pr_info("%s: Registered with: %s, can't unregister, exiting\n",
+ __func__, arm_bL_ops->name);
+ }
+
+ cpufreq_unregister_driver(&bL_cpufreq_driver);
+ pr_info("%s: Un-registered platform driver: %s\n", __func__,
+ arm_bL_ops->name);
+ arm_bL_ops = NULL;
+}
+EXPORT_SYMBOL_GPL(bL_cpufreq_unregister);
diff --git a/drivers/cpufreq/arm_big_little.h b/drivers/cpufreq/arm_big_little.h
new file mode 100644
index 0000000..049e15d
--- /dev/null
+++ b/drivers/cpufreq/arm_big_little.h
@@ -0,0 +1,40 @@
+/*
+ * ARM big.LITTLE platform's CPUFreq header file
+ *
+ * Copyright (C) 2013 ARM Ltd.
+ * Sudeep KarkadaNagesha <[email protected]>
+ *
+ * Copyright (C) 2013 Linaro.
+ * Viresh Kumar <[email protected]>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed "as is" WITHOUT ANY WARRANTY of any
+ * kind, whether express or implied; without even the implied warranty
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ */
+#ifndef CPUFREQ_ARM_BIG_LITTLE_H
+#define CPUFREQ_ARM_BIG_LITTLE_H
+
+#include <linux/cpufreq.h>
+#include <linux/types.h>
+
+struct cpufreq_arm_bL_ops {
+ char name[CPUFREQ_NAME_LEN];
+ u32 (*get_transition_latency)(u32 cluster);
+ struct cpufreq_frequency_table *(*get_freq_tbl)(u32 cluster,
+ int *count);
+ void (*put_freq_tbl)(u32 cluster);
+};
+
+struct cpufreq_frequency_table *
+arm_bL_copy_table_from_array(unsigned int *table, int count);
+void arm_bL_free_freq_table(u32 cluster);
+
+int bL_cpufreq_register(struct cpufreq_arm_bL_ops *ops);
+void bL_cpufreq_unregister(struct cpufreq_arm_bL_ops *ops);
+
+#endif /* CPUFREQ_ARM_BIG_LITTLE_H */
diff --git a/drivers/cpufreq/arm_big_little_dt.c b/drivers/cpufreq/arm_big_little_dt.c
new file mode 100644
index 0000000..c1eee50
--- /dev/null
+++ b/drivers/cpufreq/arm_big_little_dt.c
@@ -0,0 +1,125 @@
+/*
+ * Generic big.LITTLE CPUFreq Interface driver
+ *
+ * It provides necessary ops to arm_big_little cpufreq driver and gets
+ * Frequency information from Device Tree. Freq table in DT must be in KHz.
+ *
+ * Copyright (C) 2013 Linaro.
+ * Viresh Kumar <[email protected]>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed "as is" WITHOUT ANY WARRANTY of any
+ * kind, whether express or implied; without even the implied warranty
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ */
+
+#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
+
+#include <linux/cpufreq.h>
+#include <linux/export.h>
+#include <linux/module.h>
+#include <linux/of.h>
+#include <linux/slab.h>
+#include <linux/types.h>
+#include "arm_big_little.h"
+
+static struct cpufreq_frequency_table *generic_get_freq_tbl(u32 cluster,
+ int *count)
+{
+ struct device_node *np = NULL;
+ const struct property *pp;
+ unsigned int *table = NULL;
+ int cluster_id;
+ struct cpufreq_frequency_table *cpufreq_table;
+
+ while ((np = of_find_node_by_name(np, "cluster"))) {
+ if (of_property_read_u32(np, "reg", &cluster_id))
+ continue;
+
+ if (cluster_id != cluster)
+ continue;
+
+ pp = of_find_property(np, "freqs", NULL);
+ if (!pp)
+ continue;
+
+ *count = pp->length / sizeof(u32);
+ if (!*count)
+ continue;
+
+ table = kmalloc(sizeof(*table) * (*count), GFP_KERNEL);
+ if (!table) {
+ pr_err("%s: Failed to allocate memory for table\n",
+ __func__);
+ return NULL;
+ }
+
+ of_property_read_u32_array(np, "freqs", table, *count);
+ break;
+ }
+
+ if (!table) {
+ pr_err("%s: Unable to retrieve Freq table from Device Tree",
+ __func__);
+ return NULL;
+ }
+
+ cpufreq_table = arm_bL_copy_table_from_array(table, *count);
+ kfree(table);
+
+ return cpufreq_table;
+}
+
+static void generic_put_freq_tbl(u32 cluster)
+{
+ arm_bL_free_freq_table(cluster);
+}
+
+static u32 get_transition_latency(u32 cluster)
+{
+ struct device_node *np = NULL;
+ u32 transition_latency = CPUFREQ_ETERNAL;
+ int cluster_id;
+
+ while ((np = of_find_node_by_name(np, "cluster"))) {
+ if (of_property_read_u32(np, "reg", &cluster_id))
+ continue;
+
+ if (cluster_id != cluster)
+ continue;
+
+ of_property_read_u32(np, "clock-latency", &transition_latency);
+ return transition_latency;
+ }
+
+ pr_err("%s: Unable to get clock-latency from DT, use CPUFREQ_ETERNAL",
+ __func__);
+
+ return transition_latency;
+}
+
+static struct cpufreq_arm_bL_ops generic_bL_ops = {
+ .name = "generic-bl",
+ .get_transition_latency = get_transition_latency,
+ .get_freq_tbl = generic_get_freq_tbl,
+ .put_freq_tbl = generic_put_freq_tbl,
+};
+
+static int generic_bL_init(void)
+{
+ return bL_cpufreq_register(&generic_bL_ops);
+}
+module_init(generic_bL_init);
+
+static void generic_bL_exit(void)
+{
+ return bL_cpufreq_unregister(&generic_bL_ops);
+}
+module_exit(generic_bL_exit);
+
+MODULE_DESCRIPTION("Generic ARM big LITTLE cpufreq driver");
+MODULE_LICENSE("GPL");
--
1.7.12.rc2.18.g61b472e
On Tue, Mar 05, 2013 at 12:52:41PM +0800, Viresh Kumar wrote:
> +static void put_cluster_clk_and_freq_table(u32 cluster)
> +{
> + if (!atomic_dec_return(&cluster_usage[cluster])) {
> + clk_put(clk[cluster]);
> + clk[cluster] = NULL;
What's the point in setting the clk to NULL?
> +static int get_cluster_clk_and_freq_table(u32 cluster)
> +{
> + char name[9] = "cluster";
> + int count;
> +
> + if (atomic_inc_return(&cluster_usage[cluster]) != 1)
> + return 0;
> +
> + freq_table[cluster] = arm_bL_ops->get_freq_tbl(cluster, &count);
> + if (!freq_table[cluster])
> + goto atomic_dec;
> +
> + name[7] = cluster + '0';
> + clk[cluster] = clk_get(NULL, name);
> + if (!IS_ERR_OR_NULL(clk[cluster])) {
NAK. Two reasons.
1. IS_ERR_OR_NULL. You know about this, it's been on the list several
times.
2. Maybe clk_get_sys() rather than clk_get() and use a sensible device
name ("cpu-cluster.%u" maybe) rather than a connection name with a
NULL device?
On 5 March 2013 18:52, Russell King - ARM Linux <[email protected]> wrote:
> On Tue, Mar 05, 2013 at 12:52:41PM +0800, Viresh Kumar wrote:
>> +static void put_cluster_clk_and_freq_table(u32 cluster)
>> +{
>> + if (!atomic_dec_return(&cluster_usage[cluster])) {
>> + clk_put(clk[cluster]);
>> + clk[cluster] = NULL;
>
> What's the point in setting the clk to NULL?
I couldn't find one and the same is true for freq_table[] too.
>> +static int get_cluster_clk_and_freq_table(u32 cluster)
>> +{
>> + char name[9] = "cluster";
>> + int count;
>> +
>> + if (atomic_inc_return(&cluster_usage[cluster]) != 1)
>> + return 0;
>> +
>> + freq_table[cluster] = arm_bL_ops->get_freq_tbl(cluster, &count);
>> + if (!freq_table[cluster])
>> + goto atomic_dec;
>> +
>> + name[7] = cluster + '0';
>> + clk[cluster] = clk_get(NULL, name);
>> + if (!IS_ERR_OR_NULL(clk[cluster])) {
>
> NAK. Two reasons.
>
> 1. IS_ERR_OR_NULL. You know about this, it's been on the list several
> times.
AAHHHHHH .. How can i mess up with this concept.. I am really feeling bad now.
> 2. Maybe clk_get_sys() rather than clk_get() and use a sensible device
> name ("cpu-cluster.%u" maybe) rather than a connection name with a
> NULL device?
That's a good comment (rather than pointing at some stupid mistake), I will
probably keep the same name for the device as well.
So how does below fix look to you?
----------x-----------------x-----------------
diff --git a/drivers/cpufreq/arm_big_little.c b/drivers/cpufreq/arm_big_little.c
index 0d6de0e..2486b9a 100644
--- a/drivers/cpufreq/arm_big_little.c
+++ b/drivers/cpufreq/arm_big_little.c
@@ -140,9 +140,7 @@ static void put_cluster_clk_and_freq_table(u32 cluster)
{
if (!atomic_dec_return(&cluster_usage[cluster])) {
clk_put(clk[cluster]);
- clk[cluster] = NULL;
arm_bL_ops->put_freq_tbl(cluster);
- freq_table[cluster] = NULL;
pr_debug("%s: cluster: %d\n", __func__, cluster);
}
}
@@ -160,8 +158,8 @@ static int get_cluster_clk_and_freq_table(u32 cluster)
goto atomic_dec;
name[7] = cluster + '0';
- clk[cluster] = clk_get(NULL, name);
- if (!IS_ERR_OR_NULL(clk[cluster])) {
+ clk[cluster] = clk_get_sys(name, NULL);
+ if (!IS_ERR(clk[cluster])) {
pr_debug("%s: clk: %p & freq table: %p, cluster: %d\n",
__func__, clk[cluster], freq_table[cluster],
cluster);
On Wed, Mar 06, 2013 at 12:38:36AM +0800, Viresh Kumar wrote:
> So how does below fix look to you?
Much better, but I think you want a better device name than just "clusterN".
Also, please get rid of that "default n" - n is the default default default
default default default.... there's no need to specify it. :)
On 7 March 2013 01:25, Russell King - ARM Linux <[email protected]> wrote:
> On Wed, Mar 06, 2013 at 12:38:36AM +0800, Viresh Kumar wrote:
>> So how does below fix look to you?
>
> Much better, but I think you want a better device name than just "clusterN".
I will try to find some other name, maybe cpu-cluster (as what you suggested
initially).
> Also, please get rid of that "default n" - n is the default default default
> default default default.... there's no need to specify it. :)
Ok :)
On 5 March 2013 18:52, Russell King - ARM Linux <[email protected]> wrote:
> On Tue, Mar 05, 2013 at 12:52:41PM +0800, Viresh Kumar wrote:
>> + clk[cluster] = clk_get(NULL, name);
>> + if (!IS_ERR_OR_NULL(clk[cluster])) {
>
> NAK. Two reasons.
>
> 1. IS_ERR_OR_NULL. You know about this, it's been on the list several
> times.
Hey, i had a second thought about this one and i have some other opinion
here. This is a cpufreq driver and we need clock support for sure here, we
can't work without it. And so here is the latest fixup:
diff --git a/drivers/cpufreq/Kconfig.arm b/drivers/cpufreq/Kconfig.arm
index fdf54a9..87b7e48 100644
--- a/drivers/cpufreq/Kconfig.arm
+++ b/drivers/cpufreq/Kconfig.arm
@@ -9,8 +9,7 @@ config ARM_BIG_LITTLE_CPUFREQ
config ARM_DT_BL_CPUFREQ
tristate "Generic ARM big LITTLE CPUfreq driver probed via DT"
select ARM_BIG_LITTLE_CPUFREQ
- depends on OF
- default n
+ depends on OF && HAVE_CLK
help
This enables the Generic CPUfreq driver for ARM big.LITTLE platform.
This gets frequency tables from DT.
diff --git a/drivers/cpufreq/arm_big_little.c b/drivers/cpufreq/arm_big_little.c
index 2486b9a..d1fdc65 100644
--- a/drivers/cpufreq/arm_big_little.c
+++ b/drivers/cpufreq/arm_big_little.c
@@ -147,7 +147,7 @@ static void put_cluster_clk_and_freq_table(u32 cluster)
static int get_cluster_clk_and_freq_table(u32 cluster)
{
- char name[9] = "cluster";
+ char name[9] = "cpu-cluster";
int count;
if (atomic_inc_return(&cluster_usage[cluster]) != 1)
@@ -159,7 +159,7 @@ static int get_cluster_clk_and_freq_table(u32 cluster)
name[7] = cluster + '0';
clk[cluster] = clk_get_sys(name, NULL);
- if (!IS_ERR(clk[cluster])) {
+ if (!IS_ERR_OR_NULL(clk[cluster])) {
pr_debug("%s: clk: %p & freq table: %p, cluster: %d\n",
__func__, clk[cluster], freq_table[cluster],
cluster);
On Wed, Mar 6, 2013 at 4:32 PM, Viresh Kumar <[email protected]> wrote:
> clk[cluster] = clk_get_sys(name, NULL);
> - if (!IS_ERR(clk[cluster])) {
> + if (!IS_ERR_OR_NULL(clk[cluster])) {
> pr_debug("%s: clk: %p & freq table: %p, cluster: %d\n",
> __func__, clk[cluster], freq_table[cluster],
> cluster);
You seem pretty attached to IS_ERR_OR_NULL here.
Harvey
On Thu, Mar 07, 2013 at 08:32:37AM +0800, Viresh Kumar wrote:
> On 5 March 2013 18:52, Russell King - ARM Linux <[email protected]> wrote:
> > On Tue, Mar 05, 2013 at 12:52:41PM +0800, Viresh Kumar wrote:
> >> + clk[cluster] = clk_get(NULL, name);
> >> + if (!IS_ERR_OR_NULL(clk[cluster])) {
> >
> > NAK. Two reasons.
> >
> > 1. IS_ERR_OR_NULL. You know about this, it's been on the list several
> > times.
>
> Hey, i had a second thought about this one and i have some other opinion
> here. This is a cpufreq driver and we need clock support for sure here, we
> can't work without it. And so here is the latest fixup:
NAK. You just don't understand.
On 7 March 2013 08:49, Harvey Harrison <[email protected]> wrote:
> On Wed, Mar 6, 2013 at 4:32 PM, Viresh Kumar <[email protected]> wrote:
>
>> clk[cluster] = clk_get_sys(name, NULL);
>> - if (!IS_ERR(clk[cluster])) {
>> + if (!IS_ERR_OR_NULL(clk[cluster])) {
>> pr_debug("%s: clk: %p & freq table: %p, cluster: %d\n",
>> __func__, clk[cluster], freq_table[cluster],
>> cluster);
>
>
> You seem pretty attached to IS_ERR_OR_NULL here.
Haha. Not really. I just wanted to get more logical conclusion out. Please check
the other mail (where i would reply to Russell), which might have more
discussion
around this.
On Tue, Mar 05, 2013 at 12:52:41PM +0800, Viresh Kumar wrote:
> +++ b/Documentation/devicetree/bindings/cpufreq/arm_big_little_dt.txt
> @@ -0,0 +1,29 @@
> +Generic ARM big LITTLE cpufreq driver's DT glue
> +-----------------------------------------------
> +
> +It is DT specific glue layer for generic cpufreq driver for big LITTLE systems.
> +
> +Both required and optional properties listed below must be defined under node
> +cluster*. * can be 0 or 1.
> +
> +Required properties:
> +- freqs: List of all supported frequencies.
> +
> +Optional properties:
> +- clock-latency: Specify the possible maximum transition latency for clock, in
> + unit of nanoseconds.
> +
> +Examples:
> +
> +cluster0: cluster@0 {
> + ..
> +
> + freqs = <500000000 600000000 700000000 800000000 900000000 1000000000 1100000000 1200000000>;
> + clock-latency = <200000>;
> +
> + ..
> +
> + cores {
> + ..
> + };
> +};
This binding makes no sense to me. It needs to be substantially better
documented, not just a couple of sentences that people that understand
bit.LITTLE thoroughly can make sense of.
It also duplicates the cpu binding. I suspect this should instead be done
through additions of the cpu bindings instead of duplication. So this needs to
be substantially reworked.
-Olof
On Wed, Mar 6, 2013 at 5:46 PM, Olof Johansson <[email protected]> wrote:
> This binding makes no sense to me. It needs to be substantially better
> documented, not just a couple of sentences that people that understand
> bit.LITTLE thoroughly can make sense of.
>
> It also duplicates the cpu binding. I suspect this should instead be done
> through additions of the cpu bindings instead of duplication. So this needs to
> be substantially reworked.
Also, _ALWAYS_ cc devicetree-discuss on new bindings. get_maintainer
tells you to do so.
-Olof
On 7 March 2013 08:51, Russell King - ARM Linux <[email protected]> wrote:
> On Thu, Mar 07, 2013 at 08:32:37AM +0800, Viresh Kumar wrote:
>> On 5 March 2013 18:52, Russell King - ARM Linux <[email protected]> wrote:
>> > On Tue, Mar 05, 2013 at 12:52:41PM +0800, Viresh Kumar wrote:
>> >> + clk[cluster] = clk_get(NULL, name);
>> >> + if (!IS_ERR_OR_NULL(clk[cluster])) {
>> >
>> > NAK. Two reasons.
>> >
>> > 1. IS_ERR_OR_NULL. You know about this, it's been on the list several
>> > times.
>>
>> Hey, i had a second thought about this one and i have some other opinion
>> here. This is a cpufreq driver and we need clock support for sure here, we
>> can't work without it. And so here is the latest fixup:
>
> NAK. You just don't understand.
Poor me!!
I still remember the huge discussions we had during "clk: Add non
CONFIG_HAVE_CLK routines" patchset.
For others: https://lkml.org/lkml/2012/4/24/389
Back to the discussion, I understand that clk_get() just returns a cookie and
NULL is not an error and so it shouldn't be treated specially. And that's what
we do with most of our drivers as all other clk routines (clk_get[set]_rate())
have safe guards against the NULL clk, and they wouldn't complain.
The special case we have in a cpufreq driver is, we can't work with
zero returned
from clk_get_rate()... That will make cpufreq driver work badly.
And that's why i got two additions
- depends on HAVE_CLK
- and using IS_ERR_OR_NULL
as NULL returned from clk_get is not a error but is not acceptable to
the cpufreq
driver.
Because i now have the HAVE_CLK dependency i am not sure if we can get NULL
returned from clk_get() at all. Can you get around something here?
--
viresh
On 7 March 2013 09:50, Olof Johansson <[email protected]> wrote:
> On Wed, Mar 6, 2013 at 5:46 PM, Olof Johansson <[email protected]> wrote:
>
>> This binding makes no sense to me. It needs to be substantially better
>> documented, not just a couple of sentences that people that understand
>> bit.LITTLE thoroughly can make sense of.
>>
>> It also duplicates the cpu binding. I suspect this should instead be done
>> through additions of the cpu bindings instead of duplication. So this needs to
>> be substantially reworked.
Actually, i wasn't getting in new bindings for cpu/cluster for big
LITTLE but was
just trying to get bindings for getting freq-table for cpufreq driver.
Lorenzo (cc'd) actually has a patch for getting the bindings in and he
is looking
to send them soon. I know these can change after some review and the plan was
i will fix cpufreq bindings again once that patch is in. We don't have
any mainlined
hardware for it for now and so it wouldn't break anything.
> Also, _ALWAYS_ cc devicetree-discuss on new bindings. get_maintainer
> tells you to do so.
I knew it and i messed up with it. Will surely take care of it next time.
On Thu, Mar 07, 2013 at 10:03:28AM +0800, Viresh Kumar wrote:
> On 7 March 2013 08:51, Russell King - ARM Linux <[email protected]> wrote:
> > On Thu, Mar 07, 2013 at 08:32:37AM +0800, Viresh Kumar wrote:
> >> On 5 March 2013 18:52, Russell King - ARM Linux <[email protected]> wrote:
> >> > On Tue, Mar 05, 2013 at 12:52:41PM +0800, Viresh Kumar wrote:
> >> >> + clk[cluster] = clk_get(NULL, name);
> >> >> + if (!IS_ERR_OR_NULL(clk[cluster])) {
> >> >
> >> > NAK. Two reasons.
> >> >
> >> > 1. IS_ERR_OR_NULL. You know about this, it's been on the list several
> >> > times.
> >>
> >> Hey, i had a second thought about this one and i have some other opinion
> >> here. This is a cpufreq driver and we need clock support for sure here, we
> >> can't work without it. And so here is the latest fixup:
> >
> > NAK. You just don't understand.
>
> Poor me!!
>
> I still remember the huge discussions we had during "clk: Add non
> CONFIG_HAVE_CLK routines" patchset.
>
> For others: https://lkml.org/lkml/2012/4/24/389
>
> Back to the discussion, I understand that clk_get() just returns a cookie and
> NULL is not an error and so it shouldn't be treated specially. And that's what
> we do with most of our drivers as all other clk routines (clk_get[set]_rate())
> have safe guards against the NULL clk, and they wouldn't complain.
>
> The special case we have in a cpufreq driver is, we can't work with
> zero returned
> from clk_get_rate()... That will make cpufreq driver work badly.
So how is this different from any other clock which may also return zero
from its clk_get_rate() ?
If that's the condition you want to check for, call clk_get_rate() after
a successful clk_get*() and check for the condition. Don't go treating
the cookie somehow specially. You're *assuming* a behaviour that is
inappropriate for the side of the interface you're working with.
On 7 March 2013 19:49, Russell King - ARM Linux <[email protected]> wrote:
> So how is this different from any other clock which may also return zero
> from its clk_get_rate() ?
>
> If that's the condition you want to check for, call clk_get_rate() after
> a successful clk_get*() and check for the condition. Don't go treating
> the cookie somehow specially. You're *assuming* a behaviour that is
> inappropriate for the side of the interface you're working with.
Okay. I will replace the earlier fixup with following:
diff --git a/drivers/cpufreq/Kconfig.arm b/drivers/cpufreq/Kconfig.arm
index fdf54a9..87b7e48 100644
--- a/drivers/cpufreq/Kconfig.arm
+++ b/drivers/cpufreq/Kconfig.arm
@@ -9,8 +9,7 @@ config ARM_BIG_LITTLE_CPUFREQ
config ARM_DT_BL_CPUFREQ
tristate "Generic ARM big LITTLE CPUfreq driver probed via DT"
select ARM_BIG_LITTLE_CPUFREQ
- depends on OF
- default n
+ depends on OF && HAVE_CLK
help
This enables the Generic CPUfreq driver for ARM big.LITTLE platform.
This gets frequency tables from DT.
diff --git a/drivers/cpufreq/arm_big_little.c b/drivers/cpufreq/arm_big_little.c
index 2486b9a..a41fd89 100644
--- a/drivers/cpufreq/arm_big_little.c
+++ b/drivers/cpufreq/arm_big_little.c
@@ -147,7 +147,7 @@ static void put_cluster_clk_and_freq_table(u32 cluster)
static int get_cluster_clk_and_freq_table(u32 cluster)
{
- char name[9] = "cluster";
+ char name[9] = "cpu-cluster";
int count;
if (atomic_inc_return(&cluster_usage[cluster]) != 1)
For more clarity i will resend this patch now will all updates.