aboutsummaryrefslogtreecommitdiff
path: root/arch/x86/oprofile/op_model_amd.c
diff options
context:
space:
mode:
Diffstat (limited to 'arch/x86/oprofile/op_model_amd.c')
-rw-r--r--arch/x86/oprofile/op_model_amd.c285
1 files changed, 51 insertions, 234 deletions
diff --git a/arch/x86/oprofile/op_model_amd.c b/arch/x86/oprofile/op_model_amd.c
index a011bcc0f94..50d86c0e9ba 100644
--- a/arch/x86/oprofile/op_model_amd.c
+++ b/arch/x86/oprofile/op_model_amd.c
@@ -29,11 +29,10 @@
#include "op_x86_model.h"
#include "op_counter.h"
-#define NUM_COUNTERS 4
#ifdef CONFIG_OPROFILE_EVENT_MULTIPLEX
-#define NUM_VIRT_COUNTERS 32
+#define NUM_VIRT_COUNTERS 32
#else
-#define NUM_VIRT_COUNTERS NUM_COUNTERS
+#define NUM_VIRT_COUNTERS 0
#endif
#define OP_EVENT_MASK 0x0FFF
@@ -41,7 +40,8 @@
#define MSR_AMD_EVENTSEL_RESERVED ((0xFFFFFCF0ULL<<32)|(1ULL<<21))
-static unsigned long reset_value[NUM_VIRT_COUNTERS];
+static int num_counters;
+static unsigned long reset_value[OP_MAX_COUNTER];
#define IBS_FETCH_SIZE 6
#define IBS_OP_SIZE 12
@@ -68,62 +68,12 @@ static struct ibs_config ibs_config;
static struct ibs_state ibs_state;
/*
- * IBS cpuid feature detection
- */
-
-#define IBS_CPUID_FEATURES 0x8000001b
-
-/*
- * Same bit mask as for IBS cpuid feature flags (Fn8000_001B_EAX), but
- * bit 0 is used to indicate the existence of IBS.
- */
-#define IBS_CAPS_AVAIL (1U<<0)
-#define IBS_CAPS_FETCHSAM (1U<<1)
-#define IBS_CAPS_OPSAM (1U<<2)
-#define IBS_CAPS_RDWROPCNT (1U<<3)
-#define IBS_CAPS_OPCNT (1U<<4)
-#define IBS_CAPS_BRNTRGT (1U<<5)
-#define IBS_CAPS_OPCNTEXT (1U<<6)
-
-#define IBS_CAPS_DEFAULT (IBS_CAPS_AVAIL \
- | IBS_CAPS_FETCHSAM \
- | IBS_CAPS_OPSAM)
-
-/*
- * IBS APIC setup
- */
-#define IBSCTL 0x1cc
-#define IBSCTL_LVT_OFFSET_VALID (1ULL<<8)
-#define IBSCTL_LVT_OFFSET_MASK 0x0F
-
-/*
* IBS randomization macros
*/
#define IBS_RANDOM_BITS 12
#define IBS_RANDOM_MASK ((1ULL << IBS_RANDOM_BITS) - 1)
#define IBS_RANDOM_MAXCNT_OFFSET (1ULL << (IBS_RANDOM_BITS - 5))
-static u32 get_ibs_caps(void)
-{
- u32 ibs_caps;
- unsigned int max_level;
-
- if (!boot_cpu_has(X86_FEATURE_IBS))
- return 0;
-
- /* check IBS cpuid feature flags */
- max_level = cpuid_eax(0x80000000);
- if (max_level < IBS_CPUID_FEATURES)
- return IBS_CAPS_DEFAULT;
-
- ibs_caps = cpuid_eax(IBS_CPUID_FEATURES);
- if (!(ibs_caps & IBS_CAPS_AVAIL))
- /* cpuid flags not valid */
- return IBS_CAPS_DEFAULT;
-
- return ibs_caps;
-}
-
/*
* 16-bit Linear Feedback Shift Register (LFSR)
*
@@ -314,70 +264,6 @@ static void op_amd_stop_ibs(void)
wrmsrl(MSR_AMD64_IBSOPCTL, 0);
}
-static inline int eilvt_is_available(int offset)
-{
- /* check if we may assign a vector */
- return !setup_APIC_eilvt(offset, 0, APIC_EILVT_MSG_NMI, 1);
-}
-
-static inline int ibs_eilvt_valid(void)
-{
- int offset;
- u64 val;
-
- rdmsrl(MSR_AMD64_IBSCTL, val);
- offset = val & IBSCTL_LVT_OFFSET_MASK;
-
- if (!(val & IBSCTL_LVT_OFFSET_VALID)) {
- pr_err(FW_BUG "cpu %d, invalid IBS interrupt offset %d (MSR%08X=0x%016llx)\n",
- smp_processor_id(), offset, MSR_AMD64_IBSCTL, val);
- return 0;
- }
-
- if (!eilvt_is_available(offset)) {
- pr_err(FW_BUG "cpu %d, IBS interrupt offset %d not available (MSR%08X=0x%016llx)\n",
- smp_processor_id(), offset, MSR_AMD64_IBSCTL, val);
- return 0;
- }
-
- return 1;
-}
-
-static inline int get_ibs_offset(void)
-{
- u64 val;
-
- rdmsrl(MSR_AMD64_IBSCTL, val);
- if (!(val & IBSCTL_LVT_OFFSET_VALID))
- return -EINVAL;
-
- return val & IBSCTL_LVT_OFFSET_MASK;
-}
-
-static void setup_APIC_ibs(void)
-{
- int offset;
-
- offset = get_ibs_offset();
- if (offset < 0)
- goto failed;
-
- if (!setup_APIC_eilvt(offset, 0, APIC_EILVT_MSG_NMI, 0))
- return;
-failed:
- pr_warn("oprofile: IBS APIC setup failed on cpu #%d\n",
- smp_processor_id());
-}
-
-static void clear_APIC_ibs(void)
-{
- int offset;
-
- offset = get_ibs_offset();
- if (offset >= 0)
- setup_APIC_eilvt(offset, 0, APIC_EILVT_MSG_FIX, 1);
-}
-
#ifdef CONFIG_OPROFILE_EVENT_MULTIPLEX
static void op_mux_switch_ctrl(struct op_x86_model_spec const *model,
@@ -387,7 +273,7 @@ static void op_mux_switch_ctrl(struct op_x86_model_spec const *model,
int i;
/* enable active counters */
- for (i = 0; i < NUM_COUNTERS; ++i) {
+ for (i = 0; i < num_counters; ++i) {
int virt = op_x86_phys_to_virt(i);
if (!reset_value[virt])
continue;
@@ -406,7 +292,7 @@ static void op_amd_shutdown(struct op_msrs const * const msrs)
{
int i;
- for (i = 0; i < NUM_COUNTERS; ++i) {
+ for (i = 0; i < num_counters; ++i) {
if (!msrs->counters[i].addr)
continue;
release_perfctr_nmi(MSR_K7_PERFCTR0 + i);
@@ -418,7 +304,7 @@ static int op_amd_fill_in_addresses(struct op_msrs * const msrs)
{
int i;
- for (i = 0; i < NUM_COUNTERS; i++) {
+ for (i = 0; i < num_counters; i++) {
if (!reserve_perfctr_nmi(MSR_K7_PERFCTR0 + i))
goto fail;
if (!reserve_evntsel_nmi(MSR_K7_EVNTSEL0 + i)) {
@@ -426,8 +312,13 @@ static int op_amd_fill_in_addresses(struct op_msrs * const msrs)
goto fail;
}
/* both registers must be reserved */
- msrs->counters[i].addr = MSR_K7_PERFCTR0 + i;
- msrs->controls[i].addr = MSR_K7_EVNTSEL0 + i;
+ if (num_counters == AMD64_NUM_COUNTERS_CORE) {
+ msrs->counters[i].addr = MSR_F15H_PERF_CTR + (i << 1);
+ msrs->controls[i].addr = MSR_F15H_PERF_CTL + (i << 1);
+ } else {
+ msrs->controls[i].addr = MSR_K7_EVNTSEL0 + i;
+ msrs->counters[i].addr = MSR_K7_PERFCTR0 + i;
+ }
continue;
fail:
if (!counter_config[i].enabled)
@@ -447,7 +338,7 @@ static void op_amd_setup_ctrs(struct op_x86_model_spec const *model,
int i;
/* setup reset_value */
- for (i = 0; i < NUM_VIRT_COUNTERS; ++i) {
+ for (i = 0; i < OP_MAX_COUNTER; ++i) {
if (counter_config[i].enabled
&& msrs->counters[op_x86_virt_to_phys(i)].addr)
reset_value[i] = counter_config[i].count;
@@ -456,7 +347,7 @@ static void op_amd_setup_ctrs(struct op_x86_model_spec const *model,
}
/* clear all counters */
- for (i = 0; i < NUM_COUNTERS; ++i) {
+ for (i = 0; i < num_counters; ++i) {
if (!msrs->controls[i].addr)
continue;
rdmsrl(msrs->controls[i].addr, val);
@@ -472,7 +363,7 @@ static void op_amd_setup_ctrs(struct op_x86_model_spec const *model,
}
/* enable active counters */
- for (i = 0; i < NUM_COUNTERS; ++i) {
+ for (i = 0; i < num_counters; ++i) {
int virt = op_x86_phys_to_virt(i);
if (!reset_value[virt])
continue;
@@ -486,15 +377,6 @@ static void op_amd_setup_ctrs(struct op_x86_model_spec const *model,
val |= op_x86_get_ctrl(model, &counter_config[virt]);
wrmsrl(msrs->controls[i].addr, val);
}
-
- if (ibs_caps)
- setup_APIC_ibs();
-}
-
-static void op_amd_cpu_shutdown(void)
-{
- if (ibs_caps)
- clear_APIC_ibs();
}
static int op_amd_check_ctrs(struct pt_regs * const regs,
@@ -503,7 +385,7 @@ static int op_amd_check_ctrs(struct pt_regs * const regs,
u64 val;
int i;
- for (i = 0; i < NUM_COUNTERS; ++i) {
+ for (i = 0; i < num_counters; ++i) {
int virt = op_x86_phys_to_virt(i);
if (!reset_value[virt])
continue;
@@ -526,7 +408,7 @@ static void op_amd_start(struct op_msrs const * const msrs)
u64 val;
int i;
- for (i = 0; i < NUM_COUNTERS; ++i) {
+ for (i = 0; i < num_counters; ++i) {
if (!reset_value[op_x86_phys_to_virt(i)])
continue;
rdmsrl(msrs->controls[i].addr, val);
@@ -546,7 +428,7 @@ static void op_amd_stop(struct op_msrs const * const msrs)
* Subtle: stop on all counters to avoid race with setting our
* pm callback
*/
- for (i = 0; i < NUM_COUNTERS; ++i) {
+ for (i = 0; i < num_counters; ++i) {
if (!reset_value[op_x86_phys_to_virt(i)])
continue;
rdmsrl(msrs->controls[i].addr, val);
@@ -557,80 +439,11 @@ static void op_amd_stop(struct op_msrs const * const msrs)
op_amd_stop_ibs();
}
-static int setup_ibs_ctl(int ibs_eilvt_off)
-{
- struct pci_dev *cpu_cfg;
- int nodes;
- u32 value = 0;
-
- nodes = 0;
- cpu_cfg = NULL;
- do {
- cpu_cfg = pci_get_device(PCI_VENDOR_ID_AMD,
- PCI_DEVICE_ID_AMD_10H_NB_MISC,
- cpu_cfg);
- if (!cpu_cfg)
- break;
- ++nodes;
- pci_write_config_dword(cpu_cfg, IBSCTL, ibs_eilvt_off
- | IBSCTL_LVT_OFFSET_VALID);
- pci_read_config_dword(cpu_cfg, IBSCTL, &value);
- if (value != (ibs_eilvt_off | IBSCTL_LVT_OFFSET_VALID)) {
- pci_dev_put(cpu_cfg);
- printk(KERN_DEBUG "Failed to setup IBS LVT offset, "
- "IBSCTL = 0x%08x\n", value);
- return -EINVAL;
- }
- } while (1);
-
- if (!nodes) {
- printk(KERN_DEBUG "No CPU node configured for IBS\n");
- return -ENODEV;
- }
-
- return 0;
-}
-
-static int force_ibs_eilvt_setup(void)
-{
- int i;
- int ret;
-
- /* find the next free available EILVT entry */
- for (i = 1; i < 4; i++) {
- if (!eilvt_is_available(i))
- continue;
- ret = setup_ibs_ctl(i);
- if (ret)
- return ret;
- return 0;
- }
-
- printk(KERN_DEBUG "No EILVT entry available\n");
-
- return -EBUSY;
-}
-
-static int __init_ibs_nmi(void)
-{
- int ret;
-
- if (ibs_eilvt_valid())
- return 0;
-
- ret = force_ibs_eilvt_setup();
- if (ret)
- return ret;
-
- if (!ibs_eilvt_valid())
- return -EFAULT;
-
- pr_err(FW_BUG "workaround enabled for IBS LVT offset\n");
-
- return 0;
-}
+/*
+ * check and reserve APIC extended interrupt LVT offset for IBS if
+ * available
+ */
-/* initialize the APIC for the IBS interrupts if available */
static void init_ibs(void)
{
ibs_caps = get_ibs_caps();
@@ -638,25 +451,19 @@ static void init_ibs(void)
if (!ibs_caps)
return;
- if (__init_ibs_nmi()) {
- ibs_caps = 0;
- return;
- }
-
- printk(KERN_INFO "oprofile: AMD IBS detected (0x%08x)\n",
- (unsigned)ibs_caps);
+ printk(KERN_INFO "oprofile: AMD IBS detected (0x%08x)\n", ibs_caps);
}
-static int (*create_arch_files)(struct super_block *sb, struct dentry *root);
+static int (*create_arch_files)(struct dentry *root);
-static int setup_ibs_files(struct super_block *sb, struct dentry *root)
+static int setup_ibs_files(struct dentry *root)
{
struct dentry *dir;
int ret = 0;
/* architecture specific files */
if (create_arch_files)
- ret = create_arch_files(sb, root);
+ ret = create_arch_files(root);
if (ret)
return ret;
@@ -672,50 +479,60 @@ static int setup_ibs_files(struct super_block *sb, struct dentry *root)
ibs_config.max_cnt_op = 250000;
if (ibs_caps & IBS_CAPS_FETCHSAM) {
- dir = oprofilefs_mkdir(sb, root, "ibs_fetch");
- oprofilefs_create_ulong(sb, dir, "enable",
+ dir = oprofilefs_mkdir(root, "ibs_fetch");
+ oprofilefs_create_ulong(dir, "enable",
&ibs_config.fetch_enabled);
- oprofilefs_create_ulong(sb, dir, "max_count",
+ oprofilefs_create_ulong(dir, "max_count",
&ibs_config.max_cnt_fetch);
- oprofilefs_create_ulong(sb, dir, "rand_enable",
+ oprofilefs_create_ulong(dir, "rand_enable",
&ibs_config.rand_en);
}
if (ibs_caps & IBS_CAPS_OPSAM) {
- dir = oprofilefs_mkdir(sb, root, "ibs_op");
- oprofilefs_create_ulong(sb, dir, "enable",
+ dir = oprofilefs_mkdir(root, "ibs_op");
+ oprofilefs_create_ulong(dir, "enable",
&ibs_config.op_enabled);
- oprofilefs_create_ulong(sb, dir, "max_count",
+ oprofilefs_create_ulong(dir, "max_count",
&ibs_config.max_cnt_op);
if (ibs_caps & IBS_CAPS_OPCNT)
- oprofilefs_create_ulong(sb, dir, "dispatched_ops",
+ oprofilefs_create_ulong(dir, "dispatched_ops",
&ibs_config.dispatched_ops);
if (ibs_caps & IBS_CAPS_BRNTRGT)
- oprofilefs_create_ulong(sb, dir, "branch_target",
+ oprofilefs_create_ulong(dir, "branch_target",
&ibs_config.branch_target);
}
return 0;
}
+struct op_x86_model_spec op_amd_spec;
+
static int op_amd_init(struct oprofile_operations *ops)
{
init_ibs();
create_arch_files = ops->create_files;
ops->create_files = setup_ibs_files;
+
+ if (boot_cpu_data.x86 == 0x15) {
+ num_counters = AMD64_NUM_COUNTERS_CORE;
+ } else {
+ num_counters = AMD64_NUM_COUNTERS;
+ }
+
+ op_amd_spec.num_counters = num_counters;
+ op_amd_spec.num_controls = num_counters;
+ op_amd_spec.num_virt_counters = max(num_counters, NUM_VIRT_COUNTERS);
+
return 0;
}
struct op_x86_model_spec op_amd_spec = {
- .num_counters = NUM_COUNTERS,
- .num_controls = NUM_COUNTERS,
- .num_virt_counters = NUM_VIRT_COUNTERS,
+ /* num_counters/num_controls filled in at runtime */
.reserved = MSR_AMD_EVENTSEL_RESERVED,
.event_mask = OP_EVENT_MASK,
.init = op_amd_init,
.fill_in_addresses = &op_amd_fill_in_addresses,
.setup_ctrs = &op_amd_setup_ctrs,
- .cpu_down = &op_amd_cpu_shutdown,
.check_ctrs = &op_amd_check_ctrs,
.start = &op_amd_start,
.stop = &op_amd_stop,
generated by cgit v1.2.3-70-g09d2 (git 2.46.0) at 2025-12-06 12:51:27 +0000