[POWERPC] cell: PPU Oprofile cleanup patch

This is a clean up patch that includes the following changes:

 -Some comments were added to clarify the code based on feedback
  from the community.
 -The write_pm_cntrl() and set_count_mode() were passed a
  structure element from a global variable.  The argument was
  removed so the functions now just operate on the global directly.
 -The set_pm_event() function call in the cell_virtual_cntr()
  routine was moved to a for-loop before the for_each_cpu loop

Signed-off-by: Carl Love <carll@us.ibm.com>
Signed-off-by: Maynard Johnson <mpjohn@us.ibm.com>
Signed-off-by: Arnd Bergmann <arnd.bergmann@de.ibm.com>

diff --git a/arch/powerpc/oprofile/op_model_cell.c b/arch/powerpc/oprofile/op_model_cell.c
index 2eb15f3..8d4a958 100644 (file)
--- a/arch/powerpc/oprofile/op_model_cell.c
+++ b/arch/powerpc/oprofile/op_model_cell.c
@@ -41,8 +41,12 @@
 #define PPU_CYCLES_EVENT_NUM 1 /*  event number for CYCLES */
 #define CBE_COUNT_ALL_CYCLES 0x42800000        /* PPU cycle event specifier */
 
-#define NUM_THREADS 2
-#define VIRT_CNTR_SW_TIME_NS 100000000 // 0.5 seconds
+#define NUM_THREADS 2         /* number of physical threads in
+                              * physical processor
+                              */
+#define NUM_TRACE_BUS_WORDS 4
+#define NUM_INPUT_BUS_WORDS 2
+
 
 struct pmc_cntrl_data {
        unsigned long vcntr;
@@ -93,7 +97,6 @@ static struct {
        u32 pm07_cntrl[NR_PHYS_CTRS];
 } pm_regs;
 
-
 #define GET_SUB_UNIT(x) ((x & 0x0000f000) >> 12)
 #define GET_BUS_WORD(x) ((x & 0x000000f0) >> 4)
 #define GET_BUS_TYPE(x) ((x & 0x00000300) >> 8)
@@ -101,7 +104,6 @@ static struct {
 #define GET_COUNT_CYCLES(x) (x & 0x00000001)
 #define GET_INPUT_CONTROL(x) ((x & 0x00000004) >> 2)
 
-
 static DEFINE_PER_CPU(unsigned long[NR_PHYS_CTRS], pmc_values);
 
 static struct pmc_cntrl_data pmc_cntrl[NUM_THREADS][NR_PHYS_CTRS];
@@ -129,8 +131,8 @@ static spinlock_t virt_cntr_lock = SPIN_LOCK_UNLOCKED;
 
 static u32 ctr_enabled;
 
-static unsigned char trace_bus[4];
-static unsigned char input_bus[2];
+static unsigned char trace_bus[NUM_TRACE_BUS_WORDS];
+static unsigned char input_bus[NUM_INPUT_BUS_WORDS];
 
 /*
  * Firmware interface functions
@@ -183,7 +185,8 @@ static void pm_rtas_activate_signals(u32 node, u32 count)
        for (j = 0; j < count; j++) {
                /* fw expects physical cpu # */
                pm_signal_local[j].cpu = node;
-               pm_signal_local[j].signal_group = pm_signal[j].signal_group;
+               pm_signal_local[j].signal_group
+                       = pm_signal[j].signal_group;
                pm_signal_local[j].bus_word = pm_signal[j].bus_word;
                pm_signal_local[j].sub_unit = pm_signal[j].sub_unit;
                pm_signal_local[j].bit = pm_signal[j].bit;
@@ -232,13 +235,21 @@ static void set_pm_event(u32 ctr, int event, u32 unit_mask)
 
        p->signal_group = event / 100;
        p->bus_word = bus_word;
-       p->sub_unit = unit_mask & 0x0000f000;
+       p->sub_unit = (unit_mask & 0x0000f000) >> 12;
 
        pm_regs.pm07_cntrl[ctr] = 0;
        pm_regs.pm07_cntrl[ctr] |= PM07_CTR_COUNT_CYCLES(count_cycles);
        pm_regs.pm07_cntrl[ctr] |= PM07_CTR_POLARITY(polarity);
        pm_regs.pm07_cntrl[ctr] |= PM07_CTR_INPUT_CONTROL(input_control);
 
+       /* Some of the islands signal selection is based on 64 bit words.
+        * The debug bus words are 32 bits, the input words to the performance
+        * counters are defined as 32 bits.  Need to convert the 64 bit island
+        * specification to the appropriate 32 input bit and bus word for the
+        * performance counter event selection.  See the CELL Performance
+        * monitoring signals manual and the Perf cntr hardware descriptions
+        * for the details.
+        */
        if (input_control == 0) {
                if (signal_bit > 31) {
                        signal_bit -= 32;
@@ -259,12 +270,12 @@ static void set_pm_event(u32 ctr, int event, u32 unit_mask)
                p->bit = signal_bit;
        }
 
-       for (i = 0; i < 4; i++) {
+       for (i = 0; i < NUM_TRACE_BUS_WORDS; i++) {
                if (bus_word & (1 << i)) {
                        pm_regs.debug_bus_control |=
                            (bus_type << (31 - (2 * i) + 1));
 
-                       for (j = 0; j < 2; j++) {
+                       for (j = 0; j < NUM_INPUT_BUS_WORDS; j++) {
                                if (input_bus[j] == 0xff) {
                                        input_bus[j] = i;
                                        pm_regs.group_control |=
@@ -278,52 +289,58 @@ out:
        ;
 }
 
-static void write_pm_cntrl(int cpu, struct pm_cntrl *pm_cntrl)
+static void write_pm_cntrl(int cpu)
 {
-       /* Oprofile will use 32 bit counters, set bits 7:10 to 0 */
+       /* Oprofile will use 32 bit counters, set bits 7:10 to 0
+        * pmregs.pm_cntrl is a global
+        */
+
        u32 val = 0;
-       if (pm_cntrl->enable == 1)
+       if (pm_regs.pm_cntrl.enable == 1)
                val |= CBE_PM_ENABLE_PERF_MON;
 
-       if (pm_cntrl->stop_at_max == 1)
+       if (pm_regs.pm_cntrl.stop_at_max == 1)
                val |= CBE_PM_STOP_AT_MAX;
 
-       if (pm_cntrl->trace_mode == 1)
-               val |= CBE_PM_TRACE_MODE_SET(pm_cntrl->trace_mode);
+       if (pm_regs.pm_cntrl.trace_mode == 1)
+               val |= CBE_PM_TRACE_MODE_SET(pm_regs.pm_cntrl.trace_mode);
 
-       if (pm_cntrl->freeze == 1)
+       if (pm_regs.pm_cntrl.freeze == 1)
                val |= CBE_PM_FREEZE_ALL_CTRS;
 
        /* Routine set_count_mode must be called previously to set
         * the count mode based on the user selection of user and kernel.
         */
-       val |= CBE_PM_COUNT_MODE_SET(pm_cntrl->count_mode);
+       val |= CBE_PM_COUNT_MODE_SET(pm_regs.pm_cntrl.count_mode);
        cbe_write_pm(cpu, pm_control, val);
 }
 
 static inline void
-set_count_mode(u32 kernel, u32 user, struct pm_cntrl *pm_cntrl)
+set_count_mode(u32 kernel, u32 user)
 {
        /* The user must specify user and kernel if they want them. If
-        *  neither is specified, OProfile will count in hypervisor mode
+        *  neither is specified, OProfile will count in hypervisor mode.
+        *  pm_regs.pm_cntrl is a global
         */
        if (kernel) {
                if (user)
-                       pm_cntrl->count_mode = CBE_COUNT_ALL_MODES;
+                       pm_regs.pm_cntrl.count_mode = CBE_COUNT_ALL_MODES;
                else
-                       pm_cntrl->count_mode = CBE_COUNT_SUPERVISOR_MODE;
+                       pm_regs.pm_cntrl.count_mode =
+                               CBE_COUNT_SUPERVISOR_MODE;
        } else {
                if (user)
-                       pm_cntrl->count_mode = CBE_COUNT_PROBLEM_MODE;
+                       pm_regs.pm_cntrl.count_mode = CBE_COUNT_PROBLEM_MODE;
                else
-                       pm_cntrl->count_mode = CBE_COUNT_HYPERVISOR_MODE;
+                       pm_regs.pm_cntrl.count_mode =
+                               CBE_COUNT_HYPERVISOR_MODE;
        }
 }
 
 static inline void enable_ctr(u32 cpu, u32 ctr, u32 * pm07_cntrl)
 {
 
-       pm07_cntrl[ctr] |= PM07_CTR_ENABLE(1);
+       pm07_cntrl[ctr] |= CBE_PM_CTR_ENABLE;
        cbe_write_pm07_control(cpu, ctr, pm07_cntrl[ctr]);
 }
 
@@ -365,6 +382,14 @@ static void cell_virtual_cntr(unsigned long data)
        hdw_thread = 1 ^ hdw_thread;
        next_hdw_thread = hdw_thread;
 
+       for (i = 0; i < num_counters; i++)
+       /* There are some per thread events.  Must do the
+        * set event, for the thread that is being started
+        */
+               set_pm_event(i,
+                       pmc_cntrl[next_hdw_thread][i].evnts,
+                       pmc_cntrl[next_hdw_thread][i].masks);
+
        /* The following is done only once per each node, but
         * we need cpu #, not node #, to pass to the cbe_xxx functions.
         */
@@ -385,12 +410,13 @@ static void cell_virtual_cntr(unsigned long data)
                            == 0xFFFFFFFF)
                                /* If the cntr value is 0xffffffff, we must
                                 * reset that to 0xfffffff0 when the current
-                                * thread is restarted.  This will generate a new
-                                * interrupt and make sure that we never restore
-                                * the counters to the max value.  If the counters
-                                * were restored to the max value, they do not
-                                * increment and no interrupts are generated.  Hence
-                                * no more samples will be collected on that cpu.
+                                * thread is restarted.  This will generate a
+                                * new interrupt and make sure that we never
+                                * restore the counters to the max value.  If
+                                * the counters were restored to the max value,
+                                * they do not increment and no interrupts are
+                                * generated.  Hence no more samples will be
+                                * collected on that cpu.
                                 */
                                cbe_write_ctr(cpu, i, 0xFFFFFFF0);
                        else
@@ -410,9 +436,6 @@ static void cell_virtual_cntr(unsigned long data)
                                 * Must do the set event, enable_cntr
                                 * for each cpu.
                                 */
-                               set_pm_event(i,
-                                    pmc_cntrl[next_hdw_thread][i].evnts,
-                                    pmc_cntrl[next_hdw_thread][i].masks);
                                enable_ctr(cpu, i,
                                           pm_regs.pm07_cntrl);
                        } else {
@@ -465,8 +488,7 @@ cell_reg_setup(struct op_counter_config *ctr,
        pm_regs.pm_cntrl.trace_mode = 0;
        pm_regs.pm_cntrl.freeze = 1;
 
-       set_count_mode(sys->enable_kernel, sys->enable_user,
-                      &pm_regs.pm_cntrl);
+       set_count_mode(sys->enable_kernel, sys->enable_user);
 
        /* Setup the thread 0 events */
        for (i = 0; i < num_ctrs; ++i) {
@@ -498,10 +520,10 @@ cell_reg_setup(struct op_counter_config *ctr,
                pmc_cntrl[1][i].vcntr = i;
        }
 
-       for (i = 0; i < 4; i++)
+       for (i = 0; i < NUM_TRACE_BUS_WORDS; i++)
                trace_bus[i] = 0xff;
 
-       for (i = 0; i < 2; i++)
+       for (i = 0; i < NUM_INPUT_BUS_WORDS; i++)
                input_bus[i] = 0xff;
 
        /* Our counters count up, and "count" refers to
@@ -560,7 +582,7 @@ static void cell_cpu_setup(struct op_counter_config *cntr)
        cbe_write_pm(cpu, pm_start_stop, 0);
        cbe_write_pm(cpu, group_control, pm_regs.group_control);
        cbe_write_pm(cpu, debug_bus_control, pm_regs.debug_bus_control);
-       write_pm_cntrl(cpu, &pm_regs.pm_cntrl);
+       write_pm_cntrl(cpu);
 
        for (i = 0; i < num_counters; ++i) {
                if (ctr_enabled & (1 << i)) {
@@ -602,7 +624,7 @@ static void cell_global_start(struct op_counter_config *ctr)
                        }
                }
 
-               cbe_clear_pm_interrupts(cpu);
+               cbe_get_and_clear_pm_interrupts(cpu);
                cbe_enable_pm_interrupts(cpu, hdw_thread, interrupt_mask);
                cbe_enable_pm(cpu);
        }
@@ -672,7 +694,7 @@ cell_handle_interrupt(struct pt_regs *regs, struct op_counter_config *ctr)
 
        cbe_disable_pm(cpu);
 
-       interrupt_mask = cbe_clear_pm_interrupts(cpu);
+       interrupt_mask = cbe_get_and_clear_pm_interrupts(cpu);
 
        /* If the interrupt mask has been cleared, then the virt cntr
         * has cleared the interrupt.  When the thread that generated

diff --git a/arch/powerpc/platforms/cell/pmu.c b/arch/powerpc/platforms/cell/pmu.c
index d04ae16..66ca4b5 100644 (file)
--- a/arch/powerpc/platforms/cell/pmu.c
+++ b/arch/powerpc/platforms/cell/pmu.c
@@ -345,18 +345,12 @@ EXPORT_SYMBOL_GPL(cbe_read_trace_buffer);
  * Enabling/disabling interrupts for the entire performance monitoring unit.
  */
 
-u32 cbe_query_pm_interrupts(u32 cpu)
-{
-       return cbe_read_pm(cpu, pm_status);
-}
-EXPORT_SYMBOL_GPL(cbe_query_pm_interrupts);
-
-u32 cbe_clear_pm_interrupts(u32 cpu)
+u32 cbe_get_and_clear_pm_interrupts(u32 cpu)
 {
        /* Reading pm_status clears the interrupt bits. */
-       return cbe_query_pm_interrupts(cpu);
+       return cbe_read_pm(cpu, pm_status);
 }
-EXPORT_SYMBOL_GPL(cbe_clear_pm_interrupts);
+EXPORT_SYMBOL_GPL(cbe_get_and_clear_pm_interrupts);
 
 void cbe_enable_pm_interrupts(u32 cpu, u32 thread, u32 mask)
 {
@@ -371,7 +365,7 @@ EXPORT_SYMBOL_GPL(cbe_enable_pm_interrupts);
 
 void cbe_disable_pm_interrupts(u32 cpu)
 {
-       cbe_clear_pm_interrupts(cpu);
+       cbe_get_and_clear_pm_interrupts(cpu);
        cbe_write_pm(cpu, pm_status, 0);
 }
 EXPORT_SYMBOL_GPL(cbe_disable_pm_interrupts);

diff --git a/include/asm-powerpc/cell-pmu.h b/include/asm-powerpc/cell-pmu.h
index e8c2ebd..35b9577 100644 (file)
--- a/include/asm-powerpc/cell-pmu.h
+++ b/include/asm-powerpc/cell-pmu.h
@@ -53,6 +53,11 @@
 #define CBE_PM_CTR_POLARITY                0x01000000
 #define CBE_PM_CTR_COUNT_CYCLES            0x00800000
 #define CBE_PM_CTR_ENABLE                  0x00400000
+#define PM07_CTR_INPUT_MUX(x)              (((x) & 0x3F) << 26)
+#define PM07_CTR_INPUT_CONTROL(x)          (((x) & 1) << 25)
+#define PM07_CTR_POLARITY(x)               (((x) & 1) << 24)
+#define PM07_CTR_COUNT_CYCLES(x)           (((x) & 1) << 23)
+#define PM07_CTR_ENABLE(x)                 (((x) & 1) << 22)
 
 /* Macros for the pm_status register. */
 #define CBE_PM_CTR_OVERFLOW_INTR(ctr)      (1 << (31 - ((ctr) & 7)))
@@ -89,8 +94,7 @@ extern void cbe_read_trace_buffer(u32 cpu, u64 *buf);
 
 extern void cbe_enable_pm_interrupts(u32 cpu, u32 thread, u32 mask);
 extern void cbe_disable_pm_interrupts(u32 cpu);
-extern u32  cbe_query_pm_interrupts(u32 cpu);
-extern u32  cbe_clear_pm_interrupts(u32 cpu);
+extern u32  cbe_get_and_clear_pm_interrupts(u32 cpu);
 extern void cbe_sync_irq(int node);
 
 /* Utility functions, macros */
@@ -103,11 +107,4 @@ extern u32 cbe_get_hw_thread_id(int cpu);
 #define CBE_COUNT_PROBLEM_MODE          2
 #define CBE_COUNT_ALL_MODES             3
 
-/* Macros for the pm07_control registers. */
-#define PM07_CTR_INPUT_MUX(x)                    (((x) & 0x3F) << 26)
-#define PM07_CTR_INPUT_CONTROL(x)                (((x) & 1) << 25)
-#define PM07_CTR_POLARITY(x)                     (((x) & 1) << 24)
-#define PM07_CTR_COUNT_CYCLES(x)                 (((x) & 1) << 23)
-#define PM07_CTR_ENABLE(x)                       (((x) & 1) << 22)
-
 #endif /* __ASM_CELL_PMU_H__ */