ChanServ changed the topic of #lima to: Development channel for open source lima driver for ARM Mali4** GPUs - Kernel has landed in mainline, userspace driver is part of mesa - Logs at https://people.freedesktop.org/~cbrill/dri-log/index.php?channel=lima and https://freenode.irclog.whitequark.org/lima - Contact ARM for binary driver support!
narmstrong has quit [Read error: Connection reset by peer]
narmstrong has joined #lima
ezequielg has joined #lima
lvrp16 has quit [Ping timeout: 246 seconds]
ezequielg has quit [Read error: Connection reset by peer]
austriancoder has quit [Ping timeout: 250 seconds]
robher has quit [Ping timeout: 245 seconds]
narmstrong has quit [Ping timeout: 245 seconds]
ric96 has quit [Ping timeout: 248 seconds]
megi has quit [Ping timeout: 240 seconds]
ezequielg has joined #lima
ezequielg has quit [Ping timeout: 272 seconds]
ezequielg has joined #lima
ezequielg has quit [Read error: Connection reset by peer]
ezequielg has joined #lima
ezequielg has quit [Ping timeout: 272 seconds]
dddddd has quit [Ping timeout: 265 seconds]
lvrp16 has joined #lima
lvrp16 has quit [Excess Flood]
anarsoul has quit [Ping timeout: 268 seconds]
anarsoul has joined #lima
Barada has joined #lima
Barada1 has joined #lima
Barada has quit [Ping timeout: 272 seconds]
Barada1 is now known as Barada
<anarsoul> rellla: it looks like you may need to split lima_blend_factor() into several functions
<anarsoul> these extra 'ifs' don't look good
<anarsoul> rellla: btw, values returned by lima_blend_factor() look like bitmap to me
<anarsoul> e.g.
<anarsoul> PIPE_BLENDFACTOR_SRC_COLOR = 0, PIPE_BLENDFACTOR_SRC_ALPHA = 16
<anarsoul> PIPE_BLENDFACTOR_DST_COLOR = 1, PIPE_BLENDFACTOR_DST_ALPHA = 17
<anarsoul> so
<anarsoul> (1 << 4) means alpha
<anarsoul> then
<anarsoul> PIPE_BLENDFACTOR_INV_SRC_COLOR = 8, PIPE_BLENDFACTOR_INV_SRC_ALPHA = 24
<anarsoul> to (1 << 3) means INV
<anarsoul> I guess last 3 bits are SRC = 0, DST = 1, CONST = 2, ZERO = 3, SATURATE = 4, ??? = 5, ??? = 6, ??? = 7
<anarsoul> let's take a look at panfrost sources :)
austriancoder has joined #lima
<anarsoul> nah, it's different there
ric96 has joined #lima
narmstrong has joined #lima
lvrp16 has joined #lima
ezequielg has joined #lima
robher has joined #lima
anarsoul|c has joined #lima
yann has quit [Ping timeout: 268 seconds]
yuq825 has joined #lima
yann|work has joined #lima
marex-cloud has joined #lima
<rellla> anarsoul: i used more helper functions now. we have more code in the end but i think it better explains now what mali does.
monstr has joined #lima
megi has joined #lima
anarsoul|c has quit [Quit: Connection closed for inactivity]
warpme_ has joined #lima
kaspter has quit [Quit: kaspter]
dddddd has joined #lima
megi has quit [Ping timeout: 268 seconds]
yuq825 has quit [Remote host closed the connection]
<rellla> anarsoul: i have 40/40 passes with dEQP-GLES2.functional.default_vertex_attrib.* on a cubieboard1(A10) with current master but (old) 5.0.0 kernel
<rellla> https://gitlab.freedesktop.org/mesa/mesa/commit/6f394343b1f704f8b98a24add7f4106e72e2db7b broke dEQP-GLES2.functional.shaders.algorithm.hsl_to_rgb_vertex btw
<rellla> iirc you asked about the breaking tests: http://imkreisrum.de/deqp/mesa_d8e05edb/deqp-lima-skips.txt
<rellla> crashes simply crash without kernel error. i did not investigate more.
<rellla> kernel crashes show the gp task error
megi has joined #lima
<rellla> and the others simply run out of memory (maybe due to some endless loop?) all other tests finish cleanly. either with a pass or a fail
<enunes> rellla: some time ago I wrote this to avoid crashing the entire test run due to bugs like that: https://paste.centos.org/view/raw/6ae0af4c
<rellla> enunes: thanks
<rellla> dEQP-GLES2.functional.shaders.random.texture.fragment.106 is also failing nowadays btw. but this one i didn't bisect. maybe it's the opt_algebraic patch as well
<enunes> this was already in the skip list right?
Barada has quit [Quit: Barada]
<rellla> which one?
<enunes> dEQP-GLES2.functional.shaders.random.*
<rellla> not in mine. this is, what i use: http://imkreisrum.de/deqp/mesa_d8e05edb/deqp-lima-skips.txt
<enunes> I copied the one I have from you a couple of weeks ago probably
<enunes> maybe I should update it now that you've been fixing all the tests
<rellla> there have been some .* in it
<rellla> now we have just single tests
<rellla> the skip and fails lists here http://imkreisrum.de/deqp/ are produced an A10 but should be the same for Mali-450
<rellla> they are done with current master and my blend commit.
<rellla> erm no :/
<enunes> ok I'll update it and generate a new caselist to use from now on
<rellla> it's just current master without the blend commit
<rellla> dEQP-GLES2.functional.fragment_ops.blend.* should also go away once merged
<rellla> enunes: it's your liveness commit that breaks dEQP-GLES2.functional.shaders.random.texture.fragment.106 btw :p
<enunes> rellla: damn, and I spent a massive amount of effort testing that
<enunes> but that one was still on my skip list
<enunes> anything else breaks or just that?
<rellla> just that one with yours and the other one i mentioned with the opt_algebraic patch
<enunes> what is the opt_algenraic patch?
yuq825 has joined #lima
<enunes> I don't see something immediately obvious, it could be that that test only worked before due to luck
yuq825 has quit [Quit: Leaving.]
<enunes> ah ok
anarsoul|c has joined #lima
yann|work has quit [Ping timeout: 272 seconds]
monstr has quit [Remote host closed the connection]
yann|work has joined #lima
<anarsoul> rellla: enunes: so attribute tests randomly fail on H3 :\
<anarsoul> sometimes they pass, sometimes they don't