[PATCH 00/16] cell patches for 2.6.20

@@ -805,7 +805,7 @@ static int __init create_spu(struct devi
 	if (ret)
 		goto out_unmap;
 	spin_lock_init(&spu->register_lock);
-	spu_mfc_sdr_set(spu, mfspr(SPRN_SDR1));
+	spu_mfc_sdr_setup(spu);
 	spu_mfc_sr1_set(spu, 0x33);
 	mutex_lock(&spu_mutex);
 

@@ -84,9 +84,9 @@ static void mfc_dsisr_set(struct spu *sp
 	out_be64(&spu->priv1->mfc_dsisr_RW, dsisr);
 }
 
-static void mfc_sdr_set(struct spu *spu, u64 sdr)
+static void mfc_sdr_setup(struct spu *spu)
 {
-	out_be64(&spu->priv1->mfc_sdr_RW, sdr);
+	out_be64(&spu->priv1->mfc_sdr_RW, mfspr(SPRN_SDR1));
 }
 
 static void mfc_sr1_set(struct spu *spu, u64 sr1)

@@ -146,7 +146,7 @@ const struct spu_priv1_ops spu_priv1_mmi
 	.mfc_dar_get = mfc_dar_get,
 	.mfc_dsisr_get = mfc_dsisr_get,
 	.mfc_dsisr_set = mfc_dsisr_set,
-	.mfc_sdr_set = mfc_sdr_set,
+	.mfc_sdr_setup = mfc_sdr_setup,
 	.mfc_sr1_set = mfc_sr1_set,
 	.mfc_sr1_get = mfc_sr1_get,
 	.mfc_tclass_id_set = mfc_tclass_id_set,

@@ -2165,9 +2165,6 @@ static void init_priv1(struct spu_state 
 	    MFC_STATE1_PROBLEM_STATE_MASK |
 	    MFC_STATE1_RELOCATE_MASK | MFC_STATE1_BUS_TLBIE_MASK;
 
-	/* Set storage description.  */
-	csa->priv1.mfc_sdr_RW = mfspr(SPRN_SDR1);
-
 	/* Enable OS-specific set of interrupts. */
 	csa->priv1.int_mask_class0_RW = CLASS0_ENABLE_DMA_ALIGNMENT_INTR |
 	    CLASS0_ENABLE_INVALID_DMA_COMMAND_INTR |

@@ -151,7 +151,6 @@ struct spu_priv1_collapsed {
 	u64 mfc_fir_chkstp_enable_RW;
 	u64 smf_sbi_signal_sel;
 	u64 smf_ato_signal_sel;
-	u64 mfc_sdr_RW;
 	u64 tlb_index_hint_RO;
 	u64 tlb_index_W;
 	u64 tlb_vpn_RW;

@@ -37,7 +37,7 @@ struct spu_priv1_ops
 	u64 (*mfc_dar_get) (struct spu *spu);
 	u64 (*mfc_dsisr_get) (struct spu *spu);
 	void (*mfc_dsisr_set) (struct spu *spu, u64 dsisr);
-	void (*mfc_sdr_set) (struct spu *spu, u64 sdr);
+	void (*mfc_sdr_setup) (struct spu *spu);
 	void (*mfc_sr1_set) (struct spu *spu, u64 sr1);
 	u64 (*mfc_sr1_get) (struct spu *spu);
 	void (*mfc_tclass_id_set) (struct spu *spu, u64 tclass_id);

@@ -112,9 +112,9 @@ spu_mfc_dsisr_set (struct spu *spu, u64 
 }
 
 static inline void
-spu_mfc_sdr_set (struct spu *spu, u64 sdr)
+spu_mfc_sdr_setup (struct spu *spu)
 {
-	spu_priv1_ops->mfc_sdr_set(spu, sdr);
+	spu_priv1_ops->mfc_sdr_setup(spu);
 }
 
 static inline void

--

@@ -25,4 +25,13 @@ config CBE_THERM
 	default m
 	depends on CBE_RAS
 
+config CBE_CPUFREQ
+	tristate "CBE frequency scaling"
+	depends on CBE_RAS && CPU_FREQ
+	default m
+	help
+	  This adds the cpufreq driver for Cell BE processors.
+	  For details, take a look at <file:Documentation/cpu-freq/>.
+	  If you don't have such processor, say N
+
 endmenu

@@ -4,6 +4,7 @@ obj-$(CONFIG_PPC_CELL_NATIVE)		+= interr
 obj-$(CONFIG_CBE_RAS)			+= ras.o
 
 obj-$(CONFIG_CBE_THERM)			+= cbe_thermal.o
+obj-$(CONFIG_CBE_CPUFREQ)		+= cbe_cpufreq.o
 
 ifeq ($(CONFIG_SMP),y)
 obj-$(CONFIG_PPC_CELL_NATIVE)		+= smp.o

@@ -0,0 +1,248 @@
+/*
+ * cpufreq driver for the cell processor
+ *
+ * (C) Copyright IBM Deutschland Entwicklung GmbH 2005
+ *
+ * Author: Christian Krafft <krafft@de.ibm.com>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2, or (at your option)
+ * any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+
+#include <linux/cpufreq.h>
+#include <linux/timer.h>
+
+#include <asm/hw_irq.h>
+#include <asm/io.h>
+#include <asm/processor.h>
+#include <asm/prom.h>
+#include <asm/time.h>
+
+#include "cbe_regs.h"
+
+static DEFINE_MUTEX(cbe_switch_mutex);
+
+
+/* the CBE supports an 8 step frequency scaling */
+static struct cpufreq_frequency_table cbe_freqs[] = {
+	{1,	0},
+	{2,	0},
+	{3,	0},
+	{4,	0},
+	{5,	0},
+	{6,	0},
+	{8,	0},
+	{10,	0},
+	{0,	CPUFREQ_TABLE_END},
+};
+
+/* to write to MIC register */
+static u64 MIC_Slow_Fast_Timer_table[] = {
+	[0 ... 7] = 0x007fc00000000000ull,
+};
+
+/* more values for the MIC */
+static u64 MIC_Slow_Next_Timer_table[] = {
+	0x0000240000000000ull,
+	0x0000268000000000ull,
+	0x000029C000000000ull,
+	0x00002D0000000000ull,
+	0x0000300000000000ull,
+	0x0000334000000000ull,
+	0x000039C000000000ull,
+	0x00003FC000000000ull,
+};
+
+/*
+ * hardware specific functions
+ */
+
+static int get_pmode(int cpu)
+{
+	int ret;
+	struct cbe_pmd_regs __iomem *pmd_regs;
+
+	pmd_regs = cbe_get_cpu_pmd_regs(cpu);
+	ret = in_be64(&pmd_regs->pmsr) & 0x07;
+
+	return ret;
+}
+
+static int set_pmode(int cpu, unsigned int pmode)
+{
+	struct cbe_pmd_regs __iomem *pmd_regs;
+	struct cbe_mic_tm_regs __iomem *mic_tm_regs;
+	u64 flags;
+	u64 value;
+
+	local_irq_save(flags);
+
+	mic_tm_regs = cbe_get_cpu_mic_tm_regs(cpu);
+	pmd_regs = cbe_get_cpu_pmd_regs(cpu);
+
+	pr_debug("pm register is mapped at %p\n", &pmd_regs->pmcr);
+	pr_debug("mic register is mapped at %p\n", &mic_tm_regs->slow_fast_timer_0);
+
+	out_be64(&mic_tm_regs->slow_fast_timer_0, MIC_Slow_Fast_Timer_table[pmode]);
+	out_be64(&mic_tm_regs->slow_fast_timer_1, MIC_Slow_Fast_Timer_table[pmode]);
+
+	out_be64(&mic_tm_regs->slow_next_timer_0, MIC_Slow_Next_Timer_table[pmode]);
+	out_be64(&mic_tm_regs->slow_next_timer_1, MIC_Slow_Next_Timer_table[pmode]);
+
+	value = in_be64(&pmd_regs->pmcr);
+	/* set bits to zero */
+	value &= 0xFFFFFFFFFFFFFFF8ull;
+	/* set bits to next pmode */
+	value |= pmode;
+
+	out_be64(&pmd_regs->pmcr, value);
+
+	/* wait until new pmode appears in status register */
+	value = in_be64(&pmd_regs->pmsr) & 0x07;
+	while(value != pmode) {
+		cpu_relax();
+		value = in_be64(&pmd_regs->pmsr) & 0x07;
+	}
+
+	local_irq_restore(flags);
+
+	return 0;
+}
+
+/*
+ * cpufreq functions
+ */
+
+static int cbe_cpufreq_cpu_init (struct cpufreq_policy *policy)
+{
+	u32 *max_freq;
+	int i, cur_pmode;
+	struct device_node *cpu;
+
+	cpu = of_get_cpu_node(policy->cpu, NULL);
+
+	if(!cpu)
+		return -ENODEV;
+
+	pr_debug("init cpufreq on CPU %d\n", policy->cpu);
+
+	max_freq = (u32*) get_property(cpu, "clock-frequency", NULL);
+
+	if(!max_freq)
+		return -EINVAL;
+
+	// we need the freq in kHz
+	*max_freq /= 1000;
+
+	pr_debug("max clock-frequency is at %u kHz\n", *max_freq);
+	pr_debug("initializing frequency table\n");
+
+	// initialize frequency table
+	for (i=0; cbe_freqs[i].frequency!=CPUFREQ_TABLE_END; i++) {
+		cbe_freqs[i].frequency = *max_freq / cbe_freqs[i].index;
+		pr_debug("%d: %d\n", i, cbe_freqs[i].frequency);
+	}
+
+	policy->governor = CPUFREQ_DEFAULT_GOVERNOR;
+	/* if DEBUG is enabled set_pmode() measures the correct latency of a transition */
+	policy->cpuinfo.transition_latency = 25000;
+
+	cur_pmode = get_pmode(policy->cpu);
+	pr_debug("current pmode is at %d\n",cur_pmode);
+
+	policy->cur = cbe_freqs[cur_pmode].frequency;
+
+#ifdef CONFIG_SMP
+	policy->cpus = cpu_sibling_map[policy->cpu];
+#endif
+
+	cpufreq_frequency_table_get_attr (cbe_freqs, policy->cpu);
+
+	/* this ensures that policy->cpuinfo_min and policy->cpuinfo_max are set correctly */
+	return cpufreq_frequency_table_cpuinfo (policy, cbe_freqs);
+}
+
+static int cbe_cpufreq_cpu_exit(struct cpufreq_policy *policy)
+{
+	cpufreq_frequency_table_put_attr(policy->cpu);
+	return 0;
+}
+
+static int cbe_cpufreq_verify(struct cpufreq_policy *policy)
+{
+	return cpufreq_frequency_table_verify(policy, cbe_freqs);
+}
+
+
+static int cbe_cpufreq_target(struct cpufreq_policy *policy, unsigned int target_freq,
+			    unsigned int relation)
+{
+	int rc;
+	struct cpufreq_freqs freqs;
+	int cbe_pmode_new;
+
+	cpufreq_frequency_table_target(policy,
+				       cbe_freqs,
+				       target_freq,
+				       relation,
+				       &cbe_pmode_new);
+
+	freqs.old = policy->cur;
+	freqs.new = cbe_freqs[cbe_pmode_new].frequency;
+	freqs.cpu = policy->cpu;
+
+	mutex_lock (&cbe_switch_mutex);
+	cpufreq_notify_transition(&freqs, CPUFREQ_PRECHANGE);
+
+	pr_debug("setting frequency for cpu %d to %d kHz, 1/%d of max frequency\n",
+		 policy->cpu,
+		 cbe_freqs[cbe_pmode_new].frequency,
+		 cbe_freqs[cbe_pmode_new].index);
+
+	rc = set_pmode(policy->cpu, cbe_pmode_new);
+	cpufreq_notify_transition(&freqs, CPUFREQ_POSTCHANGE);
+	mutex_unlock(&cbe_switch_mutex);
+
+	return rc;
+}
+
+static struct cpufreq_driver cbe_cpufreq_driver = {
+	.verify		= cbe_cpufreq_verify,
+	.target		= cbe_cpufreq_target,
+	.init		= cbe_cpufreq_cpu_init,
+	.exit		= cbe_cpufreq_cpu_exit,
+	.name		= "cbe-cpufreq",
+	.owner		= THIS_MODULE,
+	.flags		= CPUFREQ_CONST_LOOPS,
+};
+
+/*
+ * module init and destoy
+ */
+
+static int __init cbe_cpufreq_init(void)
+{
+	return cpufreq_register_driver(&cbe_cpufreq_driver);
+}
+
+static void __exit cbe_cpufreq_exit(void)
+{
+	cpufreq_unregister_driver(&cbe_cpufreq_driver);
+}
+
+module_init(cbe_cpufreq_init);
+module_exit(cbe_cpufreq_exit);
+
+MODULE_LICENSE("GPL");
+MODULE_AUTHOR("Christian Krafft <krafft@de.ibm.com>");

@@ -139,7 +139,19 @@ CONFIG_RTAS_FLASH=y
 CONFIG_MMIO_NVRAM=y
 # CONFIG_PPC_MPC106 is not set
 # CONFIG_PPC_970_NAP is not set
-# CONFIG_CPU_FREQ is not set
+CONFIG_CPU_FREQ=y
+CONFIG_CPU_FREQ_TABLE=y
+CONFIG_CPU_FREQ_DEBUG=y
+CONFIG_CPU_FREQ_STAT=y
+# CONFIG_CPU_FREQ_STAT_DETAILS is not set
+CONFIG_CPU_FREQ_DEFAULT_GOV_PERFORMANCE=y
+# CONFIG_CPU_FREQ_DEFAULT_GOV_USERSPACE is not set
+CONFIG_CPU_FREQ_GOV_PERFORMANCE=y
+CONFIG_CPU_FREQ_GOV_POWERSAVE=y
+CONFIG_CPU_FREQ_GOV_USERSPACE=y
+CONFIG_CPU_FREQ_GOV_ONDEMAND=y
+CONFIG_CPU_FREQ_GOV_CONSERVATIVE=y
+# CONFIG_CPU_FREQ_PMAC64 is not set
 # CONFIG_WANT_EARLY_SERIAL is not set
 # CONFIG_MPIC is not set
 

@@ -150,6 +162,7 @@ CONFIG_SPU_FS=m
 CONFIG_SPU_BASE=y
 CONFIG_CBE_RAS=y
 CONFIG_CBE_THERM=m
+CONFIG_CBE_CPUFREQ=m
 
 #
 # Kernel options

@@ -1672,6 +1672,7 @@ struct device_node *of_get_cpu_node(int 
 	}
 	return NULL;
 }
+EXPORT_SYMBOL(of_get_cpu_node);
 
 #ifdef DEBUG
 static struct debugfs_blob_wrapper flat_dt_blob;

@@ -65,6 +65,7 @@ cpumask_t cpu_sibling_map[NR_CPUS] = { [
 
 EXPORT_SYMBOL(cpu_online_map);
 EXPORT_SYMBOL(cpu_possible_map);
+EXPORT_SYMBOL(cpu_sibling_map);
 
 /* SMP operations for this machine */
 struct smp_ops_t *smp_ops;

--

@@ -39,6 +39,13 @@
 #define cpu_should_die()	0
 #endif
 
+static int __init powersave_off(char *arg)
+{
+	ppc_md.power_save = NULL;
+	return 0;
+}
+__setup("powersave=off", powersave_off);
+
 /*
  * The body of the idle task.
  */

@@ -38,32 +38,16 @@
 #include "pervasive.h"
 #include "cbe_regs.h"
 
-static DEFINE_SPINLOCK(cbe_pervasive_lock);
-
-static void __init cbe_enable_pause_zero(void)
+static void cbe_power_save(void)
 {
-	unsigned long thread_switch_control;
-	unsigned long temp_register;
-	struct cbe_pmd_regs __iomem *pregs;
-
-	spin_lock_irq(&cbe_pervasive_lock);
-	pregs = cbe_get_cpu_pmd_regs(smp_processor_id());
-	if (pregs == NULL)
-		goto out;
-
-	pr_debug("Power Management: CPU %d\n", smp_processor_id());
-
-	 /* Enable Pause(0) control bit */
-	temp_register = in_be64(&pregs->pmcr);
-
-	out_be64(&pregs->pmcr,
-		 temp_register | CBE_PMD_PAUSE_ZERO_CONTROL);
+	unsigned long ctrl, thread_switch_control;
+	ctrl = mfspr(SPRN_CTRLF);
 
 	/* Enable DEC and EE interrupt request */
 	thread_switch_control  = mfspr(SPRN_TSC_CELL);
 	thread_switch_control |= TSC_CELL_EE_ENABLE | TSC_CELL_EE_BOOST;
 
-	switch ((mfspr(SPRN_CTRLF) & CTRL_CT)) {
+	switch (ctrl & CTRL_CT) {
 	case CTRL_CT0:
 		thread_switch_control |= TSC_CELL_DEC_ENABLE_0;
 		break;

@@ -75,58 +59,21 @@ static void __init cbe_enable_pause_zero
 			__FUNCTION__);
 		break;
 	}
-
 	mtspr(SPRN_TSC_CELL, thread_switch_control);
 
-out:
-	spin_unlock_irq(&cbe_pervasive_lock);
-}
-
-static void cbe_idle(void)
-{
-	unsigned long ctrl;
-
-	/* Why do we do that on every idle ? Couldn't that be done once for
-	 * all or do we lose the state some way ? Also, the pmcr
-	 * register setting, that can't be set once at boot ? We really want
-	 * to move that away in order to implement a simple powersave
+	/*
+	 * go into low thread priority, medium priority will be
+	 * restored for us after wake-up.
 	 */
-	cbe_enable_pause_zero();
+	HMT_low();
 
-	while (1) {
-		if (!need_resched()) {
-			local_irq_disable();
-			while (!need_resched()) {
-				/* go into low thread priority */
-				HMT_low();
-
-				/*
-				 * atomically disable thread execution
-				 * and runlatch.
-				 * External and Decrementer exceptions
-				 * are still handled when the thread
-				 * is disabled but now enter in
-				 * cbe_system_reset_exception()
-				 */
-				ctrl = mfspr(SPRN_CTRLF);
-				ctrl &= ~(CTRL_RUNLATCH | CTRL_TE);
-				mtspr(SPRN_CTRLT, ctrl);
-			}
-			/* restore thread prio */
-			HMT_medium();
-			local_irq_enable();
-		}
-
-		/*
-		 * turn runlatch on again before scheduling the
-		 * process we just woke up
-		 */
-		ppc64_runlatch_on();
-
-		preempt_enable_no_resched();
-		schedule();
-		preempt_disable();
-	}
+	/*
+	 * atomically disable thread execution and runlatch.
+	 * External and Decrementer exceptions are still handled when the
+	 * thread is disabled but now enter in cbe_system_reset_exception()
+	 */
+	ctrl &= ~(CTRL_RUNLATCH | CTRL_TE);
+	mtspr(SPRN_CTRLT, ctrl);
 }
 
 static int cbe_system_reset_exception(struct pt_regs *regs)

@@ -158,9 +105,20 @@ static int cbe_system_reset_exception(st
 
 void __init cbe_pervasive_init(void)
 {
+	int cpu;
 	if (!cpu_has_feature(CPU_FTR_PAUSE_ZERO))
 		return;
 
-	ppc_md.idle_loop = cbe_idle;
+	for_each_possible_cpu(cpu) {
+		struct cbe_pmd_regs __iomem *regs = cbe_get_cpu_pmd_regs(cpu);
+		if (!regs)
+			continue;
+
+		 /* Enable Pause(0) control bit */
+		out_be64(&regs->pmcr, in_be64(&regs->pmcr) |
+					    CBE_PMD_PAUSE_ZERO_CONTROL);
+	}
+
+	ppc_md.power_save = cbe_power_save;
 	ppc_md.system_reset_exception = cbe_system_reset_exception;
 }

--

@@ -20,4 +20,9 @@ config CBE_RAS
 	bool "RAS features for bare metal Cell BE"
 	default y
 
+config CBE_THERM
+	tristate "CBE thermal support"
+	default m
+	depends on CBE_RAS
+
 endmenu

@@ -3,6 +3,8 @@ obj-$(CONFIG_PPC_CELL_NATIVE)		+= interr
 					   pmu.o
 obj-$(CONFIG_CBE_RAS)			+= ras.o
 
+obj-$(CONFIG_CBE_THERM)			+= cbe_thermal.o
+
 ifeq ($(CONFIG_SMP),y)
 obj-$(CONFIG_PPC_CELL_NATIVE)		+= smp.o
 endif

@@ -149,6 +149,7 @@ CONFIG_MMIO_NVRAM=y
 CONFIG_SPU_FS=m
 CONFIG_SPU_BASE=y
 CONFIG_CBE_RAS=y
+CONFIG_CBE_THERM=m
 
 #
 # Kernel options

@@ -0,0 +1,225 @@
+/*
+ * thermal support for the cell processor
+ *
+ * (C) Copyright IBM Deutschland Entwicklung GmbH 2005
+ *
+ * Author: Christian Krafft <krafft@de.ibm.com>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2, or (at your option)
+ * any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+
+#include <linux/module.h>
+#include <linux/sysdev.h>
+#include <linux/kernel.h>
+#include <linux/cpu.h>
+#include <asm/spu.h>
+#include <asm/io.h>
+#include <asm/prom.h>
+
+#include "cbe_regs.h"
+
+static struct cbe_pmd_regs __iomem *get_pmd_regs(struct sys_device *sysdev)
+{
+	struct spu *spu;
+
+	spu = container_of(sysdev, struct spu, sysdev);
+
+	return cbe_get_pmd_regs(spu->devnode);
+}
+
+/* returns the value for a given spu in a given register */
+static u8 spu_read_register_value(struct sys_device *sysdev, union spe_reg __iomem *reg)
+{
+	unsigned int *id;
+	union spe_reg value;
+	struct spu *spu;
+
+	/* getting the id from the reg attribute will not work on future device-tree layouts
+	 * in future we should store the id to the spu struct and use it here */
+	spu = container_of(sysdev, struct spu, sysdev);
+	id = (unsigned int *)get_property(spu->devnode, "reg", NULL);
+	value.val = in_be64(&reg->val);
+
+	return value.spe[*id];
+}
+
+static ssize_t spu_show_temp(struct sys_device *sysdev, char *buf)
+{
+	int value;
+	struct cbe_pmd_regs __iomem *pmd_regs;
+
+	pmd_regs = get_pmd_regs(sysdev);
+
+	value = spu_read_register_value(sysdev, &pmd_regs->ts_ctsr1);
+	/* clear all other bits */
+	value &= 0x3F;
+	/* temp is stored in steps of 2 degrees */
+	value *= 2;
+	/* base temp is 65 degrees */
+	value += 65;
+
+	return sprintf(buf, "%d\n", (int) value);
+}
+
+static ssize_t ppe_show_temp(struct sys_device *sysdev, char *buf, int pos)
+{
+	struct cbe_pmd_regs __iomem *pmd_regs;
+	u64 value;
+
+	pmd_regs = cbe_get_cpu_pmd_regs(sysdev->id);
+	value = in_be64(&pmd_regs->ts_ctsr2);
+
+	/* access the corresponding byte */
+	value >>= pos;
+	/* clear all other bits */
+	value &= 0x3F;
+	/* temp is stored in steps of 2 degrees */
+	value *= 2;
+	/* base temp is 65 degrees */
+	value += 65;
+
+	return sprintf(buf, "%d\n", (int) value);
+}
+
+
+/* shows the temperature of the DTS on the PPE,
+ * located near the linear thermal sensor */
+static ssize_t ppe_show_temp0(struct sys_device *sysdev, char *buf)
+{
+	return ppe_show_temp(sysdev, buf, 32);
+}
+
+/* shows the temperature of the second DTS on the PPE */
+static ssize_t ppe_show_temp1(struct sys_device *sysdev, char *buf)
+{
+	return ppe_show_temp(sysdev, buf, 0);
+}
+
+static struct sysdev_attribute attr_spu_temperature = {
+	.attr = {.name = "temperature", .mode = 0400 },
+	.show = spu_show_temp,
+};
+
+static struct attribute *spu_attributes[] = {
+	&attr_spu_temperature.attr,
+};
+
+static struct attribute_group spu_attribute_group = {
+	.name	= "thermal",
+	.attrs	= spu_attributes,
+};
+
+static struct sysdev_attribute attr_ppe_temperature0 = {
+	.attr = {.name = "temperature0", .mode = 0400 },
+	.show = ppe_show_temp0,
+};
+
+static struct sysdev_attribute attr_ppe_temperature1 = {
+	.attr = {.name = "temperature1", .mode = 0400 },
+	.show = ppe_show_temp1,
+};
+
+static struct attribute *ppe_attributes[] = {
+	&attr_ppe_temperature0.attr,
+	&attr_ppe_temperature1.attr,
+};
+
+static struct attribute_group ppe_attribute_group = {
+	.name	= "thermal",
+	.attrs	= ppe_attributes,
+};
+
+/*
+ * initialize throttling with default values
+ */
+static void __init init_default_values(void)
+{
+	int cpu;
+	struct cbe_pmd_regs __iomem *pmd_regs;
+	struct sys_device *sysdev;
+	union ppe_spe_reg tpr;
+	union spe_reg str1;
+	u64 str2;
+	union spe_reg cr1;
+	u64 cr2;
+
+	/* TPR defaults */
+	/* ppe
+	 *	1F - no full stop
+	 *	08 - dynamic throttling starts if over 80 degrees
+	 *	03 - dynamic throttling ceases if below 70 degrees */
+	tpr.ppe = 0x1F0803;
+	/* spe
+	 *	10 - full stopped when over 96 degrees
+	 *	08 - dynamic throttling starts if over 80 degrees
+	 *	03 - dynamic throttling ceases if below 70 degrees
+	 */
+	tpr.spe = 0x100803;
+
+	/* STR defaults */
+	/* str1
+	 *	10 - stop 16 of 32 cycles
+	 */
+	str1.val = 0x1010101010101010ull;
+	/* str2
+	 *	10 - stop 16 of 32 cycles
+	 */
+	str2 = 0x10;
+
+	/* CR defaults */
+	/* cr1
+	 *	4 - normal operation
+	 */
+	cr1.val = 0x0404040404040404ull;
+	/* cr2
+	 *	4 - normal operation
+	 */
+	cr2 = 0x04;
+
+	for_each_possible_cpu (cpu) {
+		pr_debug("processing cpu %d\n", cpu);
+		sysdev = get_cpu_sysdev(cpu);
+		pmd_regs = cbe_get_cpu_pmd_regs(sysdev->id);
+
+		out_be64(&pmd_regs->tm_str2, str2);
+		out_be64(&pmd_regs->tm_str1.val, str1.val);
+		out_be64(&pmd_regs->tm_tpr.val, tpr.val);
+		out_be64(&pmd_regs->tm_cr1.val, cr1.val);
+		out_be64(&pmd_regs->tm_cr2, cr2);
+	}
+}
+
+
+static int __init thermal_init(void)
+{
+	init_default_values();
+
+	spu_add_sysdev_attr_group(&spu_attribute_group);
+	cpu_add_sysdev_attr_group(&ppe_attribute_group);
+
+	return 0;
+}
+module_init(thermal_init);
+
+static void __exit thermal_exit(void)
+{
+	spu_remove_sysdev_attr_group(&spu_attribute_group);
+	cpu_remove_sysdev_attr_group(&ppe_attribute_group);
+}
+module_exit(thermal_exit);
+
+MODULE_LICENSE("GPL");
+MODULE_AUTHOR("Christian Krafft <krafft@de.ibm.com>");
+

--

@@ -102,7 +102,7 @@ static inline int check_spu_isolate(stru
 	 *     saved at this time.
 	 */
 	isolate_state = SPU_STATUS_ISOLATED_STATE |
-	    SPU_STATUS_ISOLATED_LOAD_STAUTUS | SPU_STATUS_ISOLATED_EXIT_STAUTUS;
+	    SPU_STATUS_ISOLATED_LOAD_STATUS | SPU_STATUS_ISOLATED_EXIT_STATUS;
 	return (in_be32(&prob->spu_status_R) & isolate_state) ? 1 : 0;
 }
 

@@ -1046,12 +1046,12 @@ static inline int suspend_spe(struct spu
 	 */
 	if (in_be32(&prob->spu_status_R) & SPU_STATUS_RUNNING) {
 		if (in_be32(&prob->spu_status_R) &
-		    SPU_STATUS_ISOLATED_EXIT_STAUTUS) {
+		    SPU_STATUS_ISOLATED_EXIT_STATUS) {
 			POLL_WHILE_TRUE(in_be32(&prob->spu_status_R) &
 					SPU_STATUS_RUNNING);
 		}
 		if ((in_be32(&prob->spu_status_R) &
-		     SPU_STATUS_ISOLATED_LOAD_STAUTUS)
+		     SPU_STATUS_ISOLATED_LOAD_STATUS)
 		    || (in_be32(&prob->spu_status_R) &
 			SPU_STATUS_ISOLATED_STATE)) {
 			out_be32(&prob->spu_runcntl_RW, SPU_RUNCNTL_STOP);

@@ -1085,7 +1085,7 @@ static inline void clear_spu_status(stru
 	 */
 	if (!(in_be32(&prob->spu_status_R) & SPU_STATUS_RUNNING)) {
 		if (in_be32(&prob->spu_status_R) &
-		    SPU_STATUS_ISOLATED_EXIT_STAUTUS) {
+		    SPU_STATUS_ISOLATED_EXIT_STATUS) {
 			spu_mfc_sr1_set(spu,
 					MFC_STATE1_MASTER_RUN_CONTROL_MASK);
 			eieio();

@@ -1095,7 +1095,7 @@ static inline void clear_spu_status(stru
 					SPU_STATUS_RUNNING);
 		}
 		if ((in_be32(&prob->spu_status_R) &
-		     SPU_STATUS_ISOLATED_LOAD_STAUTUS)
+		     SPU_STATUS_ISOLATED_LOAD_STATUS)
 		    || (in_be32(&prob->spu_status_R) &
 			SPU_STATUS_ISOLATED_STATE)) {
 			spu_mfc_sr1_set(spu,

@@ -291,8 +291,8 @@ struct spu_problem {
 #define SPU_STATUS_INVALID_INSTR        0x20
 #define SPU_STATUS_INVALID_CH           0x40
 #define SPU_STATUS_ISOLATED_STATE       0x80
-#define SPU_STATUS_ISOLATED_LOAD_STAUTUS 0x200
-#define SPU_STATUS_ISOLATED_EXIT_STAUTUS 0x400
+#define SPU_STATUS_ISOLATED_LOAD_STATUS 0x200
+#define SPU_STATUS_ISOLATED_EXIT_STATUS 0x400
 	u8  pad_0x4028_0x402c[0x4];				/* 0x4028 */
 	u32 spu_spe_R;						/* 0x402c */
 	u8  pad_0x4030_0x4034[0x4];				/* 0x4030 */

--

@@ -1531,3 +1531,25 @@ struct tree_descr spufs_dir_contents[] =
 	{ "object-id", &spufs_object_id_ops, 0666, },
 	{},
 };
+
+struct tree_descr spufs_dir_nosched_contents[] = {
+	{ "mem",  &spufs_mem_fops,  0666, },
+	{ "mbox", &spufs_mbox_fops, 0444, },
+	{ "ibox", &spufs_ibox_fops, 0444, },
+	{ "wbox", &spufs_wbox_fops, 0222, },
+	{ "mbox_stat", &spufs_mbox_stat_fops, 0444, },
+	{ "ibox_stat", &spufs_ibox_stat_fops, 0444, },
+	{ "wbox_stat", &spufs_wbox_stat_fops, 0444, },
+	{ "signal1", &spufs_signal1_fops, 0666, },
+	{ "signal2", &spufs_signal2_fops, 0666, },
+	{ "signal1_type", &spufs_signal1_type, 0666, },
+	{ "signal2_type", &spufs_signal2_type, 0666, },
+	{ "mss", &spufs_mss_fops, 0666, },
+	{ "mfc", &spufs_mfc_fops, 0666, },
+	{ "cntl", &spufs_cntl_fops,  0666, },
+	{ "npc", &spufs_npc_ops, 0666, },
+	{ "psmap", &spufs_psmap_fops, 0666, },
+	{ "phys-id", &spufs_id_ops, 0666, },
+	{ "object-id", &spufs_object_id_ops, 0666, },
+	{},
+};

@@ -219,8 +219,11 @@ static char *spu_hw_get_ls(struct spu_co
 
 static void spu_hw_runcntl_write(struct spu_context *ctx, u32 val)
 {
-	eieio();
+	spin_lock_irq(&ctx->spu->register_lock);
+	if (val & SPU_RUNCNTL_ISOLATE)
+		out_be64(&ctx->spu->priv2->spu_privcntl_RW, 4LL);
 	out_be32(&ctx->spu->problem->spu_runcntl_RW, val);
+	spin_unlock_irq(&ctx->spu->register_lock);
 }
 
 static void spu_hw_runcntl_stop(struct spu_context *ctx)

@@ -258,7 +258,12 @@ spufs_mkdir(struct inode *dir, struct de
 
 	inode->i_op = &spufs_dir_inode_operations;
 	inode->i_fop = &simple_dir_operations;
-	ret = spufs_fill_dir(dentry, spufs_dir_contents, mode, ctx);
+	if (flags & SPU_CREATE_NOSCHED)
+		ret = spufs_fill_dir(dentry, spufs_dir_nosched_contents,
+					 mode, ctx);
+	else
+		ret = spufs_fill_dir(dentry, spufs_dir_contents, mode, ctx);
+
 	if (ret)
 		goto out_free_ctx;
 

@@ -307,6 +312,16 @@ static int spufs_create_context(struct i
 {
 	int ret;
 
+	ret = -EPERM;
+	if ((flags & SPU_CREATE_NOSCHED) &&
+	    !capable(CAP_SYS_NICE))
+		goto out_unlock;
+
+	ret = -EINVAL;
+	if ((flags & (SPU_CREATE_NOSCHED | SPU_CREATE_ISOLATE))
+	    == SPU_CREATE_ISOLATE)
+		goto out_unlock;
+
 	ret = spufs_mkdir(inode, dentry, flags, mode & S_IRWXUGO);
 	if (ret)
 		goto out_unlock;

@@ -51,11 +51,17 @@ static inline int spu_stopped(struct spu
 static inline int spu_run_init(struct spu_context *ctx, u32 * npc)
 {
 	int ret;
+	unsigned long runcntl = SPU_RUNCNTL_RUNNABLE;
 
 	if ((ret = spu_acquire_runnable(ctx)) != 0)
 		return ret;
-	ctx->ops->npc_write(ctx, *npc);
-	ctx->ops->runcntl_write(ctx, SPU_RUNCNTL_RUNNABLE);
+
+	if (ctx->flags & SPU_CREATE_ISOLATE)
+		runcntl |= SPU_RUNCNTL_ISOLATE;
+	else
+		ctx->ops->npc_write(ctx, *npc);
+
+	ctx->ops->runcntl_write(ctx, runcntl);
 	return 0;
 }
 

@@ -135,6 +135,7 @@ struct spufs_inode_info {
 	container_of(inode, struct spufs_inode_info, vfs_inode)
 
 extern struct tree_descr spufs_dir_contents[];
+extern struct tree_descr spufs_dir_nosched_contents[];
 
 /* system call implementation */
 long spufs_run_spu(struct file *file,

@@ -1916,6 +1916,51 @@ static void save_lscsa(struct spu_state 
 	wait_spu_stopped(prev, spu);	/* Step 57. */
 }
 
+static void force_spu_isolate_exit(struct spu *spu)
+{
+	struct spu_problem __iomem *prob = spu->problem;
+	struct spu_priv2 __iomem *priv2 = spu->priv2;
+
+	/* Stop SPE execution and wait for completion. */
+	out_be32(&prob->spu_runcntl_RW, SPU_RUNCNTL_STOP);
+	iobarrier_rw();
+	POLL_WHILE_TRUE(in_be32(&prob->spu_status_R) & SPU_STATUS_RUNNING);
+
+	/* Restart SPE master runcntl. */
+	spu_mfc_sr1_set(spu, MFC_STATE1_MASTER_RUN_CONTROL_MASK);
+	iobarrier_w();
+
+	/* Initiate isolate exit request and wait for completion. */
+	out_be64(&priv2->spu_privcntl_RW, 4LL);
+	iobarrier_w();
+	out_be32(&prob->spu_runcntl_RW, 2);
+	iobarrier_rw();
+	POLL_WHILE_FALSE((in_be32(&prob->spu_status_R)
+				& SPU_STATUS_STOPPED_BY_STOP));
+
+	/* Reset load request to normal. */
+	out_be64(&priv2->spu_privcntl_RW, SPU_PRIVCNT_LOAD_REQUEST_NORMAL);
+	iobarrier_w();
+}
+
+/**
+ * stop_spu_isolate
+ *	Check SPU run-control state and force isolated
+ *	exit function as necessary.
+ */
+static void stop_spu_isolate(struct spu *spu)
+{
+	struct spu_problem __iomem *prob = spu->problem;
+
+	if (in_be32(&prob->spu_status_R) & SPU_STATUS_ISOLATED_STATE) {
+		/* The SPU is in isolated state; the only way
+		 * to get it out is to perform an isolated
+		 * exit (clean) operation.
+		 */
+		force_spu_isolate_exit(spu);
+	}
+}
+
 static void harvest(struct spu_state *prev, struct spu *spu)
 {
 	/*

@@ -1928,6 +1973,7 @@ static void harvest(struct spu_state *pr
 	inhibit_user_access(prev, spu);	        /* Step 3.  */
 	terminate_spu_app(prev, spu);	        /* Step 4.  */
 	set_switch_pending(prev, spu);	        /* Step 5.  */
+	stop_spu_isolate(spu);			/* NEW.     */
 	remove_other_spu_access(prev, spu);	/* Step 6.  */
 	suspend_mfc(prev, spu);	                /* Step 7.  */
 	wait_suspend_mfc_complete(prev, spu);	/* Step 8.  */

@@ -2096,11 +2142,11 @@ int spu_save(struct spu_state *prev, str
 	acquire_spu_lock(spu);	        /* Step 1.     */
 	rc = __do_spu_save(prev, spu);	/* Steps 2-53. */
 	release_spu_lock(spu);
-	if (rc) {
+	if (rc != 0 && rc != 2 && rc != 6) {
 		panic("%s failed on SPU[%d], rc=%d.\n",
 		      __func__, spu->number, rc);
 	}
-	return rc;
+	return 0;
 }
 EXPORT_SYMBOL_GPL(spu_save);