02:09
hpt has joined #systemtap
02:18
orivej has joined #systemtap
03:08
lijunlong has quit [Read error: Connection reset by peer]
03:08
lijunlong has joined #systemtap
04:06
orivej has quit [Ping timeout: 246 seconds]
04:13
orivej has joined #systemtap
04:24
khaled has joined #systemtap
05:56
orivej has quit [Ping timeout: 240 seconds]
05:58
orivej has joined #systemtap
06:48
fdalleau_away is now known as fdalleau
07:12
<
agentzh >
serhei: cycles are very helpful too.
07:31
orivej has quit [Ping timeout: 260 seconds]
08:54
lijunlong has quit [Ping timeout: 252 seconds]
08:56
lijunlong has joined #systemtap
09:11
lijunlong has quit [Read error: Connection reset by peer]
09:15
lijunlong has joined #systemtap
09:42
amerey_ has joined #systemtap
09:43
amerey has quit [Ping timeout: 240 seconds]
10:22
mjw has joined #systemtap
10:43
hpt has quit [Ping timeout: 240 seconds]
10:52
orivej has joined #systemtap
11:15
<
fche >
cycle counts are available via getmsr if nothing else, but also clock_gettime(CLOCK_MONOTONIC_RAW) might do
12:51
orivej has quit [Ping timeout: 265 seconds]
13:03
orivej has joined #systemtap
14:00
jistone_ has joined #systemtap
14:01
khaled_ has joined #systemtap
14:01
khaled has quit [Remote host closed the connection]
14:01
jistone_ is now known as jistone
14:01
jistone has joined #systemtap
14:01
jistone has quit [Changing host]
14:16
orivej has quit [Ping timeout: 240 seconds]
14:28
fdalleau is now known as fdalleau_away
14:53
fdalleau_away is now known as fdalleau
14:59
fdalleau is now known as fdalleau_away
14:59
fdalleau_away is now known as fdalleau
15:03
lijunlong has quit [Quit: Lost terminal]
15:12
fdalleau is now known as fdalleau_away
15:13
fdalleau_away is now known as fdalleau
15:39
tromey has joined #systemtap
17:35
mjw has quit [Quit: Leaving]
18:10
wcohen has joined #systemtap
18:49
<
agentzh >
fche: not sure if that's exposed to the ebpf world.
18:49
<
agentzh >
the ebpf helpers are very limited...
18:49
<
agentzh >
missing a lot of things.
18:50
<
fche >
ah yes. well there are some clocks
18:50
<
fche >
(had bpf/dyninst confused that moment)
18:50
<
agentzh >
yeah, there's time, not cycles.
18:50
<
agentzh >
like ns time.
18:51
<
agentzh >
dyninst's 100% CPU startup hit reminds me of the stock ebpf verifier :)
18:51
<
agentzh >
both eat a lot of CPU cycles upon startup.
18:51
<
agentzh >
for nontrivial things, of course.
18:53
<
agentzh >
seems like the bpf_probe_read_str() ebpf helper and etc enforce dst buffers on the bpf stack.
18:53
<
agentzh >
no joy...
18:54
<
agentzh >
one more reason to kill the ebpf verifier ;)
18:54
<
fche >
never going to give that up
18:54
<
agentzh >
considering how small the bpf stack limit is.
18:54
<
fche >
never going to let you down
18:54
<
fche >
never going to desert you
18:55
* agentzh
is talking to a bot named fche?
20:25
fdalleau is now known as fdalleau_away
20:35
tromey has quit [Quit: ERC (IRC client for Emacs 27.1)]
22:24
khaled_ has quit [Quit: Konversation terminated!]
22:26
orivej has joined #systemtap
22:31
orivej has quit [Ping timeout: 260 seconds]
22:31
orivej_ has joined #systemtap