<chrs_>
WarheadsSE: what kernel version are you running?
<WarheadsSE>
at the moment, still from the chromium tree. Don't have it near me to double check, but 3.8.x i think.
<chrs_>
ok
<chrs_>
so not a mainline basically?
<chrs_>
have there been any major emmc improvements in mainline in say the past 6 months?
liquidAcid has joined #linux-exynos
<WarheadsSE>
Yeah, I am not running mainline on my XE503C32
<WarheadsSE>
It has even beena while since I tried it on my XE303C
<Wizzup>
Linux chrometwo 3.8.11 #24 SMP Thu Jun 19 19:55:15 CEST 2014 armv7l SAMSUNG EXYNOS5 (Flattened Device Tree) GNU/Linux
liquidAcid has quit [Quit: Leaving]
alexst has joined #linux-exynos
liquidAcid has joined #linux-exynos
alexst has quit [Ping timeout: 260 seconds]
liquidAcid has quit [Quit: Leaving]
liquidAcid has joined #linux-exynos
alexst has joined #linux-exynos
tfiga has quit [Remote host closed the connection]
alexst has quit [Ping timeout: 264 seconds]
alexst has joined #linux-exynos
tfiga has joined #linux-exynos
tfiga has quit [Remote host closed the connection]
tfiga has joined #linux-exynos
liquidAcid has quit [Ping timeout: 256 seconds]
liquidAcid has joined #linux-exynos
alexst has quit [Ping timeout: 240 seconds]
alexst has joined #linux-exynos
<Wizzup>
grr, the chrome config had CONFIG_SECURITY_YAMA=y
<chrs_>
YAMA?
alexst has quit [Ping timeout: 240 seconds]
<Wizzup>
chrs_: Some silly 'prevent ptrace' stuff
<chrs_>
why doesn't google just work in the mainline kernel?
<Wizzup>
'almost'
<chrs_>
i hate the fact that the mainline exynos support is spotty
<chrs_>
but chromeos works pretty smooth
<libv>
Wizzup: has that only recently been added to the config?
<libv>
as i used ptrace in september, and some hacks to the kernel, to grab the mali commandstream
<libv>
oh, googling tells me that it's been enabled for much longer already
<Wizzup>
libv: ptrace works if you're the parent
<Wizzup>
yeah I just always disable it
<libv>
what are you trying to grab btw?
<libv>
i was actually surprised how easy ptrace was to use
<Wizzup>
In this case I just wanted to strace a hanging irssi
<Wizzup>
Yeah, I've done quite a lot with ptrace
<Wizzup>
(and the API isn't that nice if you want more of less cross platform/arch stuff ;-)
<libv>
well, for my usecase, it seems good enough
<Wizzup>
:)
<libv>
and it has the possibility of no longer slowing things down by doing everything in process
<libv>
(on a multicore system, which is a bit of a given for mali t series)
<Wizzup>
Really?
<Wizzup>
How was that?
<libv>
but since i only captured and replayed the triangle, i really do not know
<libv>
for lima, i LD_PRELOAD, so everything is extra library call overhead in the same process
<libv>
for tamil, i couldn't get ld_preload to work reliably
<libv>
so i had to use ptrace
<libv>
and if my synapses hold up, you ptrace from a different process altogether
<Wizzup>
but ptrace makes stuff quite slow though?
<Wizzup>
As in, there's a _lot_ of context switches
<Wizzup>
(four, or eight for each syscall)
<libv>
true
<libv>
as said, i haven't tried
alexst has joined #linux-exynos
<Wizzup>
I actually noticed that for my ptrace syscall-tracing a single core was better than a dual core ;-) (all the context switches and switches between cpus made it slower)
<libv>
a single triangle frame is really not helpful
<Wizzup>
yeah, ok
<libv>
well, wait and see, a long while as i first need to start finishing off other stuff before i go back to that