| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Date: Tue, 02 Nov 2010 21:58:23 +0800
From: Lin Ming <ming.m.lin@...el.com>
To: Peter Zijlstra <a.p.zijlstra@...llo.nl>
Cc: Ingo Molnar <mingo@...hat.com>,
Frederic Weisbecker <fweisbec@...il.com>,
Arjan van de Ven <arjan@...radead.org>,
Stephane Eranian <eranian@...gle.com>,
"robert.richter@....com" <robert.richter@....com>,
Cyrill Gorcunov <gorcunov@...il.com>,
"paulus@...ba.org" <paulus@...ba.org>,
Thomas Gleixner <tglx@...utronix.de>,
"H. Peter Anvin" <hpa@...or.com>,
CoreyAshford <cjashfor@...ux.vnet.ibm.com>,
lkml <linux-kernel@...r.kernel.org>
Subject: Re: [DRAFT PATCH 0/3] perf: Add Intel Nehalem uncore pmu support
On Tue, 2010-11-02 at 20:29 +0800, Peter Zijlstra wrote:
> On Tue, 2010-11-02 at 15:27 +0800, Lin Ming wrote:
> > Any comment is very appreciated.
>
> Right, so I was hoping to use the sysfs bits to expose things, I'll try
> and get around to looking at your latest effort in that area soonish.
> I'll try and sit down with gregkh one of these days to talk it over.
>
> I'm not too sure about 1/3's change to x86_perf_event_update(), but its
> not too aweful, the change to x86_perf_event_set_period() however does
> look quite gruesome.
>
> It might make sense to simple duplicate that code in the uncore bits,.
> dunno.
How about below?
diff --git a/arch/x86/kernel/cpu/perf_event_intel_uncore.c b/arch/x86/kernel/cpu/perf_event_intel_uncore.c
index fafa0f9..b22aa95 100644
--- a/arch/x86/kernel/cpu/perf_event_intel_uncore.c
+++ b/arch/x86/kernel/cpu/perf_event_intel_uncore.c
@@ -80,10 +80,52 @@ static int uncore_pmu_event_init(struct perf_event *event)
return 0;
}
+static int
+uncore_perf_event_set_period(struct perf_event *event)
+{
+ struct hw_perf_event *hwc = &event->hw;
+ s64 left = local64_read(&hwc->period_left);
+ s64 period = hwc->sample_period;
+ u64 max_period = (1ULL << UNCORE_NUM_COUNTERS) - 1;
+ int ret = 0, idx = hwc->idx;
+
+ /*
+ * If we are way outside a reasonable range then just skip forward:
+ */
+ if (unlikely(left <= -period)) {
+ left = period;
+ local64_set(&hwc->period_left, left);
+ hwc->last_period = period;
+ ret = 1;
+ }
+
+ if (unlikely(left <= 0)) {
+ left += period;
+ local64_set(&hwc->period_left, left);
+ hwc->last_period = period;
+ ret = 1;
+ }
+
+ if (left > max_period)
+ left = max_period;
+
+ /*
+ * The hw event starts counting from this event offset,
+ * mark it to be able to extra future deltas:
+ */
+ local64_set(&hwc->prev_count, (u64)-left);
+
+ wrmsrl(hwc->event_base + idx, (u64)(-left) & max_period);
+
+ perf_event_update_userpage(event);
+
+ return ret;
+}
+
static void uncore_pmu_start(struct perf_event *event, int flags)
{
if (flags & PERF_EF_RELOAD)
- x86_perf_event_set_period(event);
+ uncore_perf_event_set_period(event);
uncore_pmu_enable_event(event);
@@ -200,7 +242,7 @@ static inline void uncore_pmu_ack_status(u64 ack)
static int uncore_pmu_save_and_restart(struct perf_event *event)
{
x86_perf_event_update(event, UNCORE_CNTVAL_BITS);
- return x86_perf_event_set_period(event);
+ return uncore_perf_event_set_period(event);
}
int uncore_pmu_handle_irq(struct pt_regs *regs)
>
> 2/3 looks ok, but I think it would be nice if it would be more self
> contained, that is, not be part of the include mess and possibly have
> its own NMI_DIE notifier entry.
How about below to make uncore code more self contained?
I'll look at the NMI DIE notifier thing later.
diff --git a/arch/x86/include/asm/perf_event.h b/arch/x86/include/asm/perf_event.h
index 550e26b..8df4e13 100644
--- a/arch/x86/include/asm/perf_event.h
+++ b/arch/x86/include/asm/perf_event.h
@@ -127,6 +127,7 @@ union cpuid10_edx {
#ifdef CONFIG_PERF_EVENTS
extern void init_hw_perf_events(void);
extern void perf_events_lapic_init(void);
+extern void init_uncore_pmu(void);
#define PERF_EVENT_INDEX_OFFSET 0
@@ -138,6 +139,7 @@ extern void perf_events_lapic_init(void);
#define PERF_EFLAGS_EXACT (1UL << 3)
struct pt_regs;
+extern int uncore_pmu_handle_irq(struct pt_regs *regs);
extern unsigned long perf_instruction_pointer(struct pt_regs *regs);
extern unsigned long perf_misc_flags(struct pt_regs *regs);
#define perf_misc_flags(regs) perf_misc_flags(regs)
diff --git a/arch/x86/kernel/cpu/Makefile b/arch/x86/kernel/cpu/Makefile
index 3f0ebe4..db4bf99 100644
--- a/arch/x86/kernel/cpu/Makefile
+++ b/arch/x86/kernel/cpu/Makefile
@@ -27,6 +27,7 @@ obj-$(CONFIG_CPU_SUP_TRANSMETA_32) += transmeta.o
obj-$(CONFIG_CPU_SUP_UMC_32) += umc.o
obj-$(CONFIG_PERF_EVENTS) += perf_event.o
+obj-$(CONFIG_PERF_EVENTS) += perf_event_intel_uncore.o
obj-$(CONFIG_X86_MCE) += mcheck/
obj-$(CONFIG_MTRR) += mtrr/
diff --git a/arch/x86/kernel/cpu/perf_event.c b/arch/x86/kernel/cpu/perf_event.c
index cca07b4..330e4f4 100644
--- a/arch/x86/kernel/cpu/perf_event.c
+++ b/arch/x86/kernel/cpu/perf_event.c
@@ -1215,8 +1215,6 @@ struct pmu_nmi_state {
static DEFINE_PER_CPU(struct pmu_nmi_state, pmu_nmi);
-static int uncore_pmu_handle_irq(struct pt_regs *regs);
-
static int __kprobes
perf_event_nmi_handler(struct notifier_block *self,
unsigned long cmd, void *__args)
@@ -1308,7 +1306,6 @@ x86_get_event_constraints(struct cpu_hw_events *cpuc, struct perf_event *event)
#include "perf_event_intel_lbr.c"
#include "perf_event_intel_ds.c"
#include "perf_event_intel.c"
-#include "perf_event_intel_uncore.c"
static int __cpuinit
x86_pmu_notifier(struct notifier_block *self, unsigned long action, void *hcpu)
diff --git a/arch/x86/kernel/cpu/perf_event_intel_uncore.c b/arch/x86/kernel/cpu/perf_event_intel_uncore.c
index b22aa95..b9f15f2 100644
--- a/arch/x86/kernel/cpu/perf_event_intel_uncore.c
+++ b/arch/x86/kernel/cpu/perf_event_intel_uncore.c
@@ -3,6 +3,7 @@
static struct node_hw_events uncore_events[MAX_NUMNODES];
static DEFINE_PER_CPU(struct uncore_cpu_hw_events, uncore_cpu_hw_events);
static bool uncore_pmu_initialized;
+static atomic_t active_events;
static void uncore_pmu_enable_event(struct perf_event *event)
{
diff --git a/arch/x86/kernel/cpu/perf_event_intel_uncore.h b/arch/x86/kernel/cpu/perf_event_intel_uncore.h
index 33b9b5e..0a5e6d4 100644
--- a/arch/x86/kernel/cpu/perf_event_intel_uncore.h
+++ b/arch/x86/kernel/cpu/perf_event_intel_uncore.h
@@ -78,3 +78,4 @@ struct uncore_cpu_hw_events {
unsigned long active_mask[BITS_TO_LONGS(UNCORE_NUM_COUNTERS)];
};
+extern u64 x86_perf_event_update(struct perf_event *event, int cntval_bits);
>
> All in all, Thanks for doing this, its a good start!
Thanks for the quick response.
Lin Ming
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists