Skip to contentSkip to conversation

perf, x86: Enable PEBS mode automatically for mem-{loads,stores} v2

PATCH Stephane Eranian 1 participant 1 message May 31, 2013 2 messages not in the archive

lore
Share & tools
lore

A message that is not in the archive (sent off-list or not yet received): [email protected]

A message that is not in the archive (sent off-list or not yet received): [email protected]

#2 lore raw
Quoted 10 lines · Andi Kleen

From: Andi Kleen <[email protected]>

With the earlier patches to automatically try cpu// and add a precise sys attribute, we can now enable PEBS for the mem-loads, mem-stores events everywhere.

This allows to use

perf record -e mem-loads ...

Quoted 8 lines

instead of

perf record -e cpu/mem-loads/p ...

Nehalem and Sandy Bridge only use precise mode == 1 as the off-by-one IP correction is quite expensive here (user can still override) On Haswell as it's basically free always use precise=2

I think overhead should not guide the choice here. You want to expose a uniform behavior across CPUs as much as possible. I suggest you use precise=2 whenever it is available. In other words, try to return the load/store instruction address each time you can.

Quoted 42 lines

Cc: [email protected]
v2: Different white space

---
 arch/x86/kernel/cpu/perf_event_intel.c | 14 +++++++++-----
 1 file changed, 9 insertions(+), 5 deletions(-)
arch/x86/kernel/cpu/perf_event_intel.c +9 -6
index b1d4966..914243c 100644
@@ -176,9 +176,12 @@ static struct extra_reg intel_snbep_extra_regs[] __read_mostly = {
        EVENT_EXTRA_END
 };
 
-EVENT_ATTR_STR(mem-loads, mem_ld_nhm, "event=0x0b,umask=0x10,ldlat=3");
-EVENT_ATTR_STR(mem-loads, mem_ld_snb, "event=0xcd,umask=0x1,ldlat=3");
-EVENT_ATTR_STR(mem-stores, mem_st_snb, "event=0xcd,umask=0x2");
+EVENT_ATTR_STR(mem-loads, mem_ld_nhm,
+               "event=0x0b,umask=0x10,ldlat=3,precise=1");
+EVENT_ATTR_STR(mem-loads, mem_ld_snb,
+               "event=0xcd,umask=0x1,ldlat=3,precise=1");
+EVENT_ATTR_STR(mem-stores, mem_st_snb,
+               "event=0xcd,umask=0x2,precise=1");
 
 struct attribute *nhm_events_attrs[] = {
        EVENT_PTR(mem_ld_nhm),
@@ -2033,8 +2036,9 @@ static __init void intel_nehalem_quirk(void)
        }
 }
 
-EVENT_ATTR_STR(mem-loads,      mem_ld_hsw,     "event=0xcd,umask=0x1,ldlat=3");
-EVENT_ATTR_STR(mem-stores,     mem_st_hsw,     "event=0xd0,umask=0x82")
+EVENT_ATTR_STR(mem-loads,      mem_ld_hsw,
+               "event=0xcd,umask=0x1,ldlat=3,precise=2");
+EVENT_ATTR_STR(mem-stores,     mem_st_hsw,  "event=0xd0,umask=0x82,precise=2")
 
 static struct attribute *hsw_events_attrs[] = {
        EVENT_PTR(mem_ld_hsw),
--

1.8.1.4