codeworkx changed the topic of #teamhacksung to: HACKSUNG IS DEAD, HAIL HACKSUNG!
ankitkv has quit [Quit: Leaving]
koud has quit [Ping timeout: 260 seconds]
koud has joined #teamhacksung
krabador has quit [Quit: Sto andando via]
evcz is now known as Guest57239
meeeeeeee has joined #teamhacksung
meeeeeeee has quit [Changing host]
meeeeeeee has joined #teamhacksung
Guest57239 has quit [Read error: Operation timed out]
ramsrib has joined #teamhacksung
arcus_ has joined #teamhacksung
arcus has quit [Disconnected by services]
meeeeeeee is now known as evcz
evcz has quit [Read error: Connection reset by peer]
ramsrib has quit [Ping timeout: 240 seconds]
evcz has joined #teamhacksung
evcz has quit [Read error: Connection reset by peer]
evcz has joined #teamhacksung
DuperMan has joined #teamhacksung
DuperMan has left #teamhacksung [#teamhacksung]
Kaik541 has joined #teamhacksung
ramsrib has joined #teamhacksung
CoreISP has quit [Quit: Smoking]
lolwat has joined #teamhacksung
lolwat has quit [Read error: Connection reset by peer]
arcus_ has quit [Ping timeout: 245 seconds]
ricotz has joined #teamhacksung
Kaik541 has quit [Ping timeout: 265 seconds]
ThePot has joined #teamhacksung
Liquidcon has joined #teamhacksung
Liquidcon has left #teamhacksung [#teamhacksung]
Turl has quit [Ping timeout: 252 seconds]
Turl has joined #teamhacksung
Kaik541 has joined #teamhacksung
uberlaggydarwin has joined #teamhacksung
uberlaggydarwin has quit [Quit: Leaving.]
uberlaggydarwin has joined #teamhacksung
ThePot has quit [Read error: Connection reset by peer]
ThePot has joined #teamhacksung
<
geecko>
omg omg sw_sync_timeline_inc(fbdev->timeline, 1); is never called !
ramsrib has left #teamhacksung [#teamhacksung]
geecko has quit [Quit: Quitte]
Zhenech has joined #teamhacksung
krabador has joined #teamhacksung
The_Pot has joined #teamhacksung
ThePot has quit [Ping timeout: 272 seconds]
sbrissen_gone is now known as sbrissen
Turtuga has quit [Ping timeout: 246 seconds]
Turtuga has joined #teamhacksung
krabador has quit [Ping timeout: 260 seconds]
cuco has joined #teamhacksung
paulk-collins has joined #teamhacksung
ThePot has joined #teamhacksung
The_Pot has quit [Ping timeout: 264 seconds]
arcus has joined #teamhacksung
geecko has joined #teamhacksung
<
geecko>
xplodwild, got a problem
<
uberlaggydarwin>
oh yeah geecko
<
geecko>
my vps is fucking slow, it takes hours (really) to make something that took minutes before
<
geecko>
uberlaggydarwin, oh yeah? :D
<
uberlaggydarwin>
my friend loves you
<
uberlaggydarwin>
for the i9300 patches
<
geecko>
haha thanks !
<
geecko>
i'd like to love my galaxy note but it's not the case yet...
<
geecko>
although i think i found something
<
geecko>
but my build server is saying NO...
<
geecko>
uberlaggydarwin, btw codeworkx did 90% of the work on hwcomposer, but he didn't have proper kernel drivers at this time
<
geecko>
it was this summer
<
uberlaggydarwin>
ahh
<
uberlaggydarwin>
i'm not in the samsung "scene" anymore
<
xplodwild>
geecko: hmm?
<
uberlaggydarwin>
jumpd in the sony tree :D
<
geecko>
xplodwild, i've been running the 'make bootimage' command since one hour. "including ./cts/Android.mk ..."
<
geecko>
xplodwild, it usually takes 5 minutes to complete D:
<
geecko>
uberlaggydarwin, baaah, too easy :P
<
uberlaggydarwin>
try a -j4 or -j8
<
uberlaggydarwin>
yeah
<
uberlaggydarwin>
geecko: so much easier to get simple things working :)
<
geecko>
uberlaggydarwin, but there's no challenge ^^ i like to bitch about semi-functional things and fix them
<
uberlaggydarwin>
hehe
<
xplodwild>
geecko: give me your patch, i'll build a bootimage
<
geecko>
xplodwild, well it fixes nothing.
<
geecko>
xplodwild, it's to debug. i found something interesting
<
geecko>
this line is never reached although it should be !
<
xplodwild>
unless fbdev->support_fence isn't FENCE_SUPPORT
<
geecko>
xplodwild, weird, i already checked that, and it was good :P
<
xplodwild>
or it waits vsync forever
<
geecko>
there's something fishy here
<
geecko>
xplodwild, holy shit i forgot to breakfast n7000
<
geecko>
:facepalm:
krabador has joined #teamhacksung
<
uberlaggydarwin>
lol
<
geecko>
aaah ça build !!
<
bealtine>
it builds!!! ship it...now whats a memeroy fault?
<
bealtine>
you'd have to be in #c++ to understand that joke
<
geecko>
xplodwild, >__________________>
<
geecko>
xplodwild, in fact it's called.
<
geecko>
no clue again ;_;
<
xplodwild>
hmm... so let's sum up
<
xplodwild>
fences are registered and signaled
<
xplodwild>
they are active
<
geecko>
they are not signaled
<
xplodwild>
registered and active then
<
xplodwild>
the timeline is increased
<
xplodwild>
points are in the past
<
xplodwild>
but they're not signaled then?
<
geecko>
xplodwild, it's because some are signaled now, and i still have this black screen
<
geecko>
shit, i don't get it :(
<
geecko>
E/Fence ( 1888): GLConsumer::doGLFenceWaitLocked: fence 49 didn't signal in 3000 ms
ThePot has quit [Quit: Leaving]
<
xplodwild>
[d3ef3960] mali_fence: error
evcz has quit [Read error: Connection reset by peer]
evcz has joined #teamhacksung
<
geecko>
xplodwild,
<
geecko>
have a single timeline parent. They have 3 states: active, signaled, and error.
<
geecko>
A sync_pt is an abstract value which marks a point on a sync_timeline. Sync_pts
<
geecko>
They start in active state and transition, once, to either signaled (when the
<
geecko>
timeline counter advances beyond the sync_pt’s value) or error state.
krabador has quit [Quit: Sto andando via]
<
geecko>
sync_fence:
<
geecko>
Sync_fences are the primary primitives used by drivers to coordinate
<
geecko>
or may not have the same timeline parent. A sync_pt can only exist in one fence
<
geecko>
and the fence's list of sync_pts is immutable once created. Fences can be
<
geecko>
synchronization of their buffers. They are a collection of sync_pts which may
<
geecko>
waited on synchronously or asynchronously. Two fences can also be merged to
<
geecko>
create a third fence containing a copy of the two fences’ sync_pts. Fences are
<
geecko>
backed by file descriptors to allow userspace to coordinate the display pipeline
<
geecko>
dependencies.
<
geecko>
sync_timeline:
<
geecko>
A sync_timeline is an abstract monotonically increasing counter. In general,
<
geecko>
each driver/hardware block context will have one of these. They can be backed
<
geecko>
by the appropriate hardware or rely on the generic sw_sync implementation.
<
geecko>
Timelines are only ever created through their specific implementations
<
geecko>
(i.e. sw_sync.)
<
geecko>
so interesting
<
geecko>
$ grep -r sw_sync
<
geecko>
s3cfb_main.c:#include <linux/sw_sync.h>
<
geecko>
s3cfb.h: struct sw_sync_timeline *timeline;
<
geecko>
s3cfb_ops.c:#include <linux/sw_sync.h>
<
geecko>
s3cfb_main.c: fbdev[i]->timeline = sw_sync_timeline_create("s3c-fb");
<
geecko>
s3cfb_ops.c: sw_sync_timeline_inc(fbdev->timeline, 1);
<
geecko>
s3cfb_ops.c: pt = sw_sync_pt_create(fbdev->timeline, fbdev->timeline_max);
<
geecko>
s3cfb_ops.c: sw_sync_timeline_inc(fbdev->timeline, 1);
<
geecko>
s3cfb_ops.c: pt = sw_sync_pt_create(fbdev->timeline, fbdev->timeline_max);
xe4l has quit [Ping timeout: 240 seconds]
xe4l has joined #teamhacksung
<
xplodwild>
geecko: the timeline progresses
<
xplodwild>
but the fences never signal they're ready
<
geecko>
xplodwild, OH MY GOD
<
geecko>
xplodwild, I GOT IT WORKING
<
geecko>
/if (unlikely(fbdev->system_state == POWER_OFF)) {
<
xplodwild>
just that? ._.
<
geecko>
xplodwild, yeah! !!
<
geecko>
xplodwild, i have no background though
<
xplodwild>
in set_win_config?
<
geecko>
xplodwild, yep
evcz has quit [Read error: Connection reset by peer]
<
geecko>
... some things are still wrong
<
xplodwild>
such as?
<
geecko>
xplodwild, launching app, buffer doesn't get updated
<
geecko>
no background
<
xplodwild>
geecko: you commented that line you said right?
<
xplodwild>
so you make it 'think' the screen is turned off if I got that properly
<
geecko>
xplodwild, hehe yeah it doesn't makes sense :P
evcz has joined #teamhacksung
<
xplodwild>
geecko: thing is, sync-wise, both paths (if and else paths) do the same thing
<
xplodwild>
geecko: can you try inverting if ((fbdev->support_fence == FENCE_SUPPORT) && ret) { instead?
<
geecko>
xplodwild, yep, it makes more sense. just seen that
<
xplodwild>
or check ret actually
<
geecko>
xplodwild, ret. that's it
<
xplodwild>
dev_warn(fbdev->dev, "unrecognized window state %u",
<
xplodwild>
config->state);
<
xplodwild>
I think i might have seen this in our log
<
xplodwild>
have a dmsg around?
<
geecko>
xplodwild, dmesg | grep window => zero hits
<
xplodwild>
check the start date of your dmesg
<
xplodwild>
make sure you have the full log
Kaik541 has quit [Remote host closed the connection]
<
geecko>
oh yeah nevermind, it never takes this path
<
xplodwild>
oh yeah :P
<
xplodwild>
let me see in yesterday logs
evcz has quit [Read error: Connection reset by peer]
evcz has joined #teamhacksung
<
xplodwild>
we didn't paste any dmsg actually
<
geecko>
xplodwild, recompiling
evcz is now known as Guest68378
meeeeeeee has joined #teamhacksung
meeeeeeee has quit [Changing host]
meeeeeeee has joined #teamhacksung
<
xplodwild>
geecko: so? :P
Guest68378 has quit [Ping timeout: 246 seconds]
<
geecko>
xplodwild, errr wait i'm not so fast :P
<
xplodwild>
pfft :P
<
geecko>
xplodwild, i waited more than a year so why not a few more minutes :P
<
xplodwild>
because I need to leave in 5 minutes :P
<
geecko>
the right one
<
geecko>
xplodwild, so everything looks good... well, not on my screen :P
<
xplodwild>
brb in a few minutes
<
geecko>
xplodwild, only difference is this sw_sync_timeline_inc(fbdev->timeline, 1);
<
geecko>
trying to add it
pawitp has joined #teamhacksung
meeeeeeee has quit [Ping timeout: 245 seconds]
meeeeeeee has joined #teamhacksung
meeeeeeee has quit [Changing host]
meeeeeeee has joined #teamhacksung
meeeeeeee is now known as evcz
evcz is now known as Guest36480
meeeeeeee has joined #teamhacksung
meeeeeeee has quit [Changing host]
meeeeeeee has joined #teamhacksung
Guest36480 has quit [Ping timeout: 245 seconds]
<
geecko>
xplodwild, lol nice, when i comment out "ret = s3c_fb_set_win_buffer(fbdev, fb, config, regs);" i get a glowing screen
Kaik541 has joined #teamhacksung
krabador has joined #teamhacksung
plastikman_ has joined #teamhacksung
<
xplodwild>
geecko: so? :P
<
geecko>
xplodwild, not much :(
<
geecko>
xplodwild, i just hope it's not a hardware problem
<
geecko>
like, something not implemented and not documented
<
geecko>
xplodwild, i'm starting to think it's the layer 0 that derps
<
geecko>
xplodwild, because: no background, apps don't update, window switcher doesn't show any transparency
<
xplodwild>
or the blendmode
<
geecko>
see that? blend
plastikman_ has quit [Remote host closed the connection]
<
xplodwild>
what's the most recent smdk4210 kernel sauce we have?
krabador has quit [Read error: Connection reset by peer]
<
geecko>
xplodwild, i have n7000 xxlsz here
<
geecko>
xplodwild, it's jelly bean 4.1
<
geecko>
want the diff?
<
xplodwild>
geecko: just to test, what happens if you disable fence?
<
xplodwild>
if you invert FENCE_SUPPORT status
<
geecko>
xplodwild, where
<
xplodwild>
in kernel
<
xplodwild>
if you report unsupported
<
geecko>
xplodwild, i'm not sure it will do anything good :P
<
xplodwild>
true :P
white has joined #teamhacksung
ankitkv has joined #teamhacksung
<
geecko>
xplodwild, i should have coloured overlays now :D
<
geecko>
xplodwild, just to see how things are going on
CoreISP has joined #teamhacksung
plastikman has joined #teamhacksung
<
geecko>
xplodwild, look at chapter 16 ^^
RudyValencia has joined #teamhacksung
pawitp has quit [Quit: pawitp]
<
geecko>
xplodwild, all the registers for FIMD are described here :o
krabador has joined #teamhacksung
plastikman has quit [Ping timeout: 246 seconds]
<
xplodwild>
geecko: I can't see how that helps
<
geecko>
xplodwild, dunno :P
<
geecko>
xplodwild, i think there something wrong with the registers.
<
geecko>
xplodwild, epic stall
<
geecko>
o_o ......
<
geecko>
xplodwild, shit, mutex_unlock(&fbdev->update_regs_list_lock); is not called
<
geecko>
so, deadlock
<
geecko>
xplodwild, pff, recompiling the kernel takes so much time now
<
geecko>
i don't know why
<
geecko>
it's embarrassing
krabador has quit [Ping timeout: 260 seconds]
<
geecko>
xplodwild, i get the very first image of the bootanimation
<
geecko>
//list_add_tail(®s->list, &fbdev->update_regs_list);
<
geecko>
mutex_unlock(&fbdev->update_regs_list_lock);
<
geecko>
// &fbdev->update_regs_work);
<
geecko>
//queue_kthread_work(&fbdev->update_regs_worker,
krabador has joined #teamhacksung
<
geecko>
xplodwild, it's weird because sometimes i see the background
<
geecko>
very rare though
plastikman has joined #teamhacksung
krabador has quit [Quit: Sto andando via]
krabador has joined #teamhacksung
joethc has joined #teamhacksung
DAGr8 has quit [Remote host closed the connection]
white has quit [Remote host closed the connection]
krabador has quit [Quit: Sto andando via]
<
geecko>
xplodwild, i took a look a the diff between xxlsz from n7000 and the current one
krabador has joined #teamhacksung
<
geecko>
xplodwild, four more registers used. i disabled them, will see what happens :)
krabador_ has joined #teamhacksung
krabador_ has quit [Client Quit]
krabador has quit [Ping timeout: 260 seconds]
krabador has joined #teamhacksung
Kaik541 has quit [Ping timeout: 272 seconds]
krabador has quit [Quit: Sto andando via]
<
xplodwild>
geecko: no luck I guess? :P
<
geecko>
xplodwild, wiping my tears
<
geecko>
xplodwild, diff i9300 update12 -> n7000 xxlsz
<
geecko>
so it's reversed
<
geecko>
xplodwild, replaced s3cfb_ops.c with old version. should compile... let's see :P
paulk-collins has quit [Quit: Ex-Chat]
Billy__ has joined #teamhacksung
Billy__ has left #teamhacksung [#teamhacksung]
ankitkv has quit [Ping timeout: 248 seconds]
<
geecko>
xplodwild, forgot to remove the ifdefs :P
<
geecko>
i'm recompiling... again and again
ankitkv has joined #teamhacksung
<
geecko>
xplodwild, exact same behavior >_>
<
geecko>
as the original file without ifdefs
<
geecko>
xplodwild, oh, no.
<
geecko>
xplodwild, not exactly: "E/ExynosHWC( 1888): ioctl S3CFB_WIN_CONFIG failed: Bad address"
joethc has quit [Quit: Leaving]
<
geecko>
xplodwild, so it fails? interesting!!
uberlaggydarwin has quit [Quit: Leaving.]
sbrissen is now known as sbrissen_gone
CoreISP has quit [Quit: Smoking]
krabador has joined #teamhacksung
ricotz has quit [Quit: Ex-Chat]
rhcp has quit [Excess Flood]
rhcp has joined #teamhacksung
cuco has quit [Quit: Konversation terminated!]
plastikman has quit [Ping timeout: 264 seconds]
plastikman has joined #teamhacksung
krabador has left #teamhacksung ["Sto andando via"]
joethc has joined #teamhacksung
joethc has quit [Quit: Leaving]
plastikman has quit [Remote host closed the connection]
joethc has joined #teamhacksung
joethc has quit [Quit: Leaving]