<jekstrand>
airlied: That sounds entirely possible
LexSfX has quit [Ping timeout: 480 seconds]
LexSfX has joined #dri-devel
<jekstrand>
PRIMITIVE_ID
<airlied>
oh AMD_DEBUG=ib is the dump
<jekstrand>
found it, I think
<airlied>
hmmm flat shade differs
<jekstrand>
or not
camus has joined #dri-devel
Company has quit [Read error: Connection reset by peer]
camus1 has quit [Ping timeout: 480 seconds]
<airlied>
si_shader_nir.c:151 is the spot it breaks from
<airlied>
jekstrand: might push prim id out over 32
<airlied>
yeah that's why it breaks, PRIMITIVE_ID goes 31->32
<airlied>
oh not maybe not that
<airlied>
that code seems horribly fragile, not really sure what is going wrong :-P
* jekstrand
looks for things that are 5 bits
nabaiste^ has joined #dri-devel
<jekstrand>
I thought maybe it was nir_io_semantics::location
<airlied>
nah it's a bug in the code
<airlied>
just not sure what exactly
<airlied>
ah location is coming from there alright
<airlied>
the thing is the correct path just happens to not hit the PRIMID check
<airlied>
and the new path hits it because you moved the value
<jekstrand>
WAT?
<airlied>
in the working path semantic is 32, and prim is 31
<airlied>
so the forced flat path never hits
<airlied>
you move prim id, the broken path has semantic 32, prim 32, boom force flat
<airlied>
it seems to rely on VARYINGs being 32 onwards
<airlied>
and system value fs inputs being below that
<airlied>
I wonder should it be checking VARYING_SLOT_PRIMITIVE_ID
<airlied>
but I think we'd need a mareko for that diagnosis
mclasen has quit [Ping timeout: 480 seconds]
ybogdano has quit [Ping timeout: 480 seconds]
<jekstrand>
Yeah, I got nothin'
<jekstrand>
I tagged mareko on the bug and told him to check the backlog. Maybe he'll show up and fix it tomorrow.
<airlied>
it's definitely a system value coming in as a input problem
<airlied>
jekstrand: the one test I can find seems to suggest VARYING_SLOT_PRIMITIVE_ID
<airlied>
is what we should be seeing in there
<jekstrand>
airlied: I can believe it
<jekstrand>
airlied: It appears in some ways to try and handle both
<jekstrand>
airlied: Yeah, I think what's happening is that SYSTEM_VALUE_PRIMITIVE_ID is now lining up with VARYING_SLOT_VAR0
<jekstrand>
I suspect that should just be VARYING_SLOT_PRIMITIVE_ID
<jekstrand>
I"m going to run that through CI just for grins
<jekstrand>
It's got to be. Looking at surrounding code, SYSTEM_VALUE_* makes no sense in that context.
i-garrison has quit []
<jekstrand>
Ok, that's enough for tonight. I think I've convinced myself. I don't know that I believe that code in general but I believe my change as much as any of it. We'll see if mareko agrees in the morning.
<jekstrand>
airlied: CI says my s/SYSTEM_VALUE/VARYING_SLOT/ patch works. Hopefully mareko will think so too.
<jekstrand>
airlied: Thanks for the debug help
jewins has quit [Ping timeout: 480 seconds]
tzimmermann has joined #dri-devel
mbrost has quit [Ping timeout: 480 seconds]
Duke`` has quit [Ping timeout: 480 seconds]
mattrope has quit [Ping timeout: 480 seconds]
danvet has joined #dri-devel
alanc has quit [Remote host closed the connection]
alanc has joined #dri-devel
camus1 has joined #dri-devel
mszyprow has joined #dri-devel
camus has quit [Ping timeout: 480 seconds]
mlankhorst has joined #dri-devel
<tzimmermann>
javierm, if you don't mind, i'll update some of the odd pci drivers with module_driver()
<javierm>
tzimmermann: sure, of course I don't
<tzimmermann>
i leave the regular module_pci_driver() to you. there are some drivers, such as i915, which probably cannot be updated in any meaningful way
<javierm>
tzimmermann: Ok. I also noticed that many drivers have a custom module init/exit just because they call platform_register_drivers() instead of platform_register_driver()
<javierm>
tzimmermann: I plan to propose a module_platform_drivers() macro to include/linux/platform_device.h
thellstrom has joined #dri-devel
camus has joined #dri-devel
camus1 has quit [Ping timeout: 480 seconds]
rasterman has joined #dri-devel
jkrzyszt_ has joined #dri-devel
<tzimmermann>
makes sense
tursulin has joined #dri-devel
pcercuei has joined #dri-devel
pnowack has joined #dri-devel
gawin has joined #dri-devel
camus1 has joined #dri-devel
camus has quit [Read error: Connection reset by peer]
lemonzest has quit [Quit: WeeChat 3.3]
tarceri has quit [Remote host closed the connection]
lemonzest has joined #dri-devel
Wanderer has quit [Remote host closed the connection]
mareko has quit [Remote host closed the connection]
marcheu has quit [Remote host closed the connection]
mareko has joined #dri-devel
dri-logger has quit [Remote host closed the connection]
dri-logger has joined #dri-devel
glisse has quit [Remote host closed the connection]
glisse has joined #dri-devel
mslusarz has quit [Remote host closed the connection]
mslusarz has joined #dri-devel
marcheu has joined #dri-devel
x512[m] has joined #dri-devel
x512 has joined #dri-devel
x512 has quit []
mclasen has joined #dri-devel
ngcortes has quit [Remote host closed the connection]
thellstrom has quit [Remote host closed the connection]
thellstrom has joined #dri-devel
camus has joined #dri-devel
camus1 has quit [Remote host closed the connection]
YuGiOhJCJ has joined #dri-devel
i-garrison has quit [Ping timeout: 480 seconds]
<tzimmermann>
there are 4 vmwgfx patches in drm-misc-next. can i cherry-pick them into drm-misc-next-fixes?
<tzimmermann>
zackr ^
iive has joined #dri-devel
Company has joined #dri-devel
camus has quit []
i-garrison has joined #dri-devel
i-garrison has quit []
i-garrison has joined #dri-devel
i-garrison has quit []
i-garrison has joined #dri-devel
YuGiOhJCJ has quit [Quit: YuGiOhJCJ]
nabaiste^ has quit [Remote host closed the connection]
jewins has joined #dri-devel
<tzimmermann>
javierm, i did some work on other PCI drivers. now that i've seen your patchset, i'm not sure if i should send it out
Bennett has joined #dri-devel
mlankhorst has quit [Ping timeout: 480 seconds]
rasterman has quit [Quit: Gettin' stinky!]
rasterman has joined #dri-devel
<javierm>
tzimmermann: are there conflicts with those? I only included one PCI driver (drm/hisilicon/hibmc)
<tzimmermann>
javierm, what i have is basically the same as yours. i'll send out an RFC patchset. if you can, please cherry-pick whatever you find useful und merge the macros into yours
<tzimmermann>
and if it really doesn't fit in, ignore it :)
nsneck has joined #dri-devel
<javierm>
tzimmermann: Ok. And agree with your feedback that we could have separate header files for each macro
<tzimmermann>
javierm, for each macro would be optimal, but if you add too many header files, people will complain :)
camus has joined #dri-devel
<tzimmermann>
i worked on drm_drv.h first and then realized that half of drm includes this file. so i moved these macros elsewhere
<javierm>
tzimmermann: yeah, let's go with a <drm/drm_module.h> then
<javierm>
tzimmermann: and sorry, I thought you saw the series this morning and that's why you mentioned moving PCI drivers to use module_driver() as a follow-up
<javierm>
since I didn't modify the existing DRM drivers that are already handling nomodeset
<tzimmermann>
javierm, oh the mail says 1:37. that is 1:37 am?
<tzimmermann>
no need to apologize
glisse is now known as Guest8702
glisse has joined #dri-devel
<javierm>
tzimmermann: yes, I couldn't sleep yesterday :)
mlankhorst has joined #dri-devel
dri-logg1r has joined #dri-devel
Guest8702 has quit [Ping timeout: 480 seconds]
dri-logger has quit [Ping timeout: 480 seconds]
<javierm>
tzimmermann: btw, it bothers me that pci_register_driver() and platform_driver_register() are named inconsistently
marcheu has quit [Remote host closed the connection]
marcheu has joined #dri-devel
mareko has quit [Remote host closed the connection]
mareko has joined #dri-devel
mslusarz has quit [Remote host closed the connection]
mslusarz has joined #dri-devel
<tzimmermann>
javierm, i know what you mean. but that's probably the result of how the kernel development works. some things are not consistent when they should be. sometimes i feel an urge to address some of that. and then i see that i'd have to fix >100 places. (and than i work on something else)
<javierm>
tzimmermann: yeah, that ship has already sailed. Hence the bothering
<javierm>
tzimmermann: anyway, what do you think about posting a v3 with your patch #1, a patch on top to add my drm_module_platform_driver(), your patch #2 and then a patch to convert a single platform driver ?
<javierm>
to avoid patch bombing the list and once we have those in place, we can post all the other patches for the remaining drivers
mslusarz has quit [Read error: Connection reset by peer]
mslusarz has joined #dri-devel
<tzimmermann>
javierm, that would work as well. i'd pick 3 or 4 conversion patches for each bus to give more examples. but maybe wait until next week to give people time to comment on the current patches.
glisse has quit [Remote host closed the connection]
glisse has joined #dri-devel
dri-logg1r has quit [Remote host closed the connection]
dri-logger has joined #dri-devel
<javierm>
tzimmermann: Ok. Let's do that. I just want to avoid posthing ~40 patches for eacy iteration since already flooded the list too much this week
<javierm>
*each
<tzimmermann>
:D
<tzimmermann>
danvet, if you have a moment, i'd like to ask for your comment on https://lore.kernel.org/dri-devel/20211215104318.18866-1-tzimmermann@suse.de/ this moves displayport helpers into a separate module. im ok with or without subdirectories. i just cannot bring myself to use names like <drm/drm_dp/drm_dp_*.h>
adjtm has quit [Remote host closed the connection]
adjtm has joined #dri-devel
gawin has quit [Ping timeout: 480 seconds]
<zackr>
tzimmermann: that would be much appreciated!
<tzimmermann>
zackr, ok
mattrope has joined #dri-devel
MrCooper has quit [Remote host closed the connection]
MrCooper has joined #dri-devel
<tzimmermann>
zackr, done
<zackr>
tzimmermann: that's great. thank you!
tursulin has quit [Quit: Konversation terminated!]
mbrost has joined #dri-devel
oneforall2 has quit [Remote host closed the connection]
alatiera6 has joined #dri-devel
alatiera has quit [Ping timeout: 480 seconds]
alatiera6 has quit []
alatiera has joined #dri-devel
oneforall2 has joined #dri-devel
Duke`` has joined #dri-devel
tzimmermann has quit [Quit: Leaving]
aravind has quit [Ping timeout: 480 seconds]
Hi-Angel has quit [Remote host closed the connection]
gouchi has joined #dri-devel
pendingchaos_ has joined #dri-devel
pendingchaos has quit [Read error: Connection reset by peer]
jkrzyszt_ has quit [Ping timeout: 480 seconds]
adjtm has quit [Quit: Leaving]
pendingchaos_ has quit [Read error: No route to host]
pendingchaos has joined #dri-devel
camus has quit [Remote host closed the connection]
<eric_engestrom>
hey everyone! sorry about the delay with the mesa release, the covid booster shot was kicking my ass for the last 2.5 days :(
<eric_engestrom>
I'll do the release now :)
<dj-death>
eric_engestrom: oh... nasty
<dj-death>
glad you feel better
ngcortes has joined #dri-devel
Bennett has quit [Remote host closed the connection]
ybogdano has joined #dri-devel
ngcortes has quit []
ngcortes has joined #dri-devel
<eric_engestrom>
yeah, I've definitely fallen into the worst category for the symptoms (each covid shot, although the 2nd dose was the worst, I had a really hard time breathing for a day or so), but at the same time it makes me so thankful that I got to experience this from the vaccine and not from the actual virus :]
<eric_engestrom>
(hoping I never actually catch it)
<alyssa>
eric_engestrom: virutalized virus
<alyssa>
docker covid-19 containers, they're all range
<alyssa>
manage your vaccines with k8s
<alyssa>
:-p
<eric_engestrom>
alyssa: haha
mlankhorst has quit [Ping timeout: 480 seconds]
mlankhorst has joined #dri-devel
adjtm has joined #dri-devel
Kayden has quit [Quit: Leaving]
ella-0 has joined #dri-devel
ella-0_ has quit [Read error: Connection reset by peer]
gawin has quit [Ping timeout: 480 seconds]
bluebugs has quit [Quit: Leaving]
mszyprow has joined #dri-devel
pcercuei has quit [Quit: dodo]
Duke`` has quit [Ping timeout: 480 seconds]
oneforall2 has quit [Quit: Leaving]
pnowack has quit [Quit: pnowack]
oneforall2 has joined #dri-devel
gouchi has quit [Remote host closed the connection]
rasterman has quit [Quit: Gettin' stinky!]
iive has quit [Ping timeout: 480 seconds]
mlankhorst has quit [Ping timeout: 480 seconds]
iive has joined #dri-devel
* jekstrand
turned in his Intel badge today
<jenatali>
:O
<anholt>
jekstrand: where to next?
<jekstrand>
Not saying quite next but, never fear, I'm not going far.
<jekstrand>
*quite yet
<anholt>
hopefully the immediate plan is "somewhere tropical, with few laptops"
<jekstrand>
No. Christmas will be spent somewhere very non-tropical (-:
danvet has quit [Ping timeout: 480 seconds]
<jekstrand>
But I've got about 5 weeks before I start the new gig so I've got some time to kick back.
<alyssa>
not far and not tropical... hmm....
<alyssa>
Houston?
<alyssa>
Austin?
<FLHerne>
Still Mesa?
<alyssa>
...Phoenix?
<jekstrand>
FLHerne: Yeah, still Mesa
<alyssa>
jekstrand: I was close, guessing Phoenix!
<jekstrand>
alyssa: :P
mszyprow has quit [Ping timeout: 480 seconds]
<FLHerne>
jekstrand: You're joining the secret Nvidia open driver project that's been implausibly rumoured to be happening for the last decade?
<jekstrand>
FLHerne: Anything is possible!
<jekstrand>
FLHerne: But not everything is probable. :-P
<Sachiel>
he'll be joining a team that exclusively works on WSI
<jekstrand>
I can say that people at my new employer has promised me fairly definitively that WSI won't be my problem. :-D
<FLHerne>
a project to revive the old Intel vec4 architecture, but scaled up many times with extremely strange memory segmentation
<jekstrand>
Yup, but with the fp64 design from IVB. HSW was way too competent. :P
<alyssa>
jekstrand: that's exactly like someone who's about to do WSI for NVIDIA would say
<emersion>
looks like your new employer knows how to trick new employees :P
<eric_engestrom>
jekstrand: are you going to be managing the people whose problem WSI will be?
<jekstrand>
I'm sure I will do some WSI but only if it's my idea.
<jekstrand>
eric_engestrom: I won't be managing people
<jekstrand>
One of the nice things about giving all my Intel HW back is that I no longer have an Iron Lake so I can't fix strange crocus bugs anyore. (-:
<alyssa>
That's exactly what a manag-- ok, I'll stop 😝
<bnieuwenhuizen>
jekstrand: getting reintroduced to GL?
<jekstrand>
bnieuwenhuizen: I think I'll be able to keep avoiding GL for the most part, but we'll see. :)
<bnieuwenhuizen>
just pitch Zink :)
<alyssa>
anholt: What is AOS?
<jekstrand>
alyssa: array-of-struct
<anholt>
alyssa: aka vector isas.
<alyssa>
got it
<jekstrand>
anholt: Is that the one that transposes matrices?
<alyssa>
We have lots of non-intel vec4 hw in-tree ... at least for vec4 mali I don't remember this being a real issue?
<jekstrand>
I'd have to run blame on things to be sure but I think it did gain some real perf in something back in the day
<alyssa>
anholt: Even if flip_matrices stays, GLSL vectorizer can go I think
<anholt>
yeah, you get to avoid a chain of dependencies
<alyssa>
since the algorithm there seems strictly worse than what nir has
<anholt>
yeah, so I definitely remove glsl vectorizer in the other MR
<alyssa>
right yes, too many MR's :)
<eric_engestrom>
jekstrand: I just noticed only the first commit of !13341 was backported to 21.3; no objection if I backport the other commit as well? 21.3 is seeing that failure as well