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!
yann has quit [Ping timeout: 256 seconds]
cwabbott has quit [Quit: cwabbott]
cwabbott has joined #lima
jernej has quit [Quit: Free ZNC ~ Powered by LunarBNC: https://LunarBNC.net]
jernej has joined #lima
dddddd has quit [Ping timeout: 260 seconds]
buzzmarshall has quit [Remote host closed the connection]
Barada has joined #lima
yann has joined #lima
monstr has joined #lima
psydread has left #lima [#lima]
psydread has joined #lima
Elpaulo has quit [Quit: Elpaulo]
dddddd has joined #lima
Barada has quit [Quit: Barada]
deesix has joined #lima
dddddd_ has joined #lima
dddddd has quit [Ping timeout: 246 seconds]
dddddd__ has joined #lima
deesix_ has quit [Ping timeout: 246 seconds]
dddddd_ has quit [Ping timeout: 264 seconds]
deesix has quit [Ping timeout: 264 seconds]
dddddd__ has quit [Ping timeout: 240 seconds]
deesix has joined #lima
dddddd__ has joined #lima
dddddd__ is now known as dddddd
monstr has quit [Ping timeout: 240 seconds]
<lool-> Marex: Hi!
<lool-> Marex: I think tjaalton was in touch about trying lima on a zcu106 (Mali 400MP)
<lool-> Marex: I'm the one at the receiving end of the instructions - thanks!!
<lool-> Marex: I tried using the device tree fragment that tjaalton passed on to me, but I couldn't get it to work; there's an existing device tree fragment in linux-xlnx master (5.4), I just enabled the GPU node
<lool-> but I'm building linux-xlnx with both the xlnx drm and the lima drm
<lool-> and that's what I see: https://paste.ubuntu.com/p/2gJn6zvtZK/
kaspter has quit [Remote host closed the connection]
kaspter has joined #lima
<rellla> lool-: do you have a link to your kernel source?
<lool-> master has a 5.4 rebase of the xlnx BSP
<Marex> lool-: hi
<Marex> lool-: I use linux-stable, without any of the xilinx crud
<Marex> oh
<Marex> lool-: are you using zynqmp-clk-ccf.inc or zynqmp-clk.inc in your DT ?
<Marex> lool-: linux 5.4 didn't have the xilinx firmware-based clock control (aka. awfulness), so I went with fixed clock tree (zynqmp-clk.dtsi instead of zynqmp-clk-ccf.dtsi)
<Marex> lool-: my guess would be that what happens is that lima tries to get clock rate from one of those clock (or claim the clock) and fails , because something which you need to investigate
<Marex> lool-: look at the clocks and clock-names properties of the GPU node in DT and look where the bus clock phandle points to
<lool-> Ok, great hints, I'll try to look into this
<lool-> that tree is using zynqmp-clk-ccf.dtsi
<Marex> lool-: do you even have clock-names = ... "bus" in the DT ? you can even check it via /proc/device-tree or /sys/(whatever the name was)
<lool-> sorry, was taking lunch
<lool-> Marex: there are in zynqmp.dtsi for eg USB attached peripherals
<Marex> lool-: and for the GPU ?
<lool-> no
<Marex> well, there's your issue :)
<Marex> I think I shared a patch yesterday for the GPU node
<lool-> ah right, I tried this snippet, but it didn't work; let me try mixing the clock-names in there
<lool-> (sorry, will be super slow today as I have many back to back calls and I'm trying to progress this in the background :)
monstr has joined #lima
chewitt has joined #lima
<lool-> hi monstr!
<monstr> hi lool-
<monstr> lool- let's talk about it tmr in better time. Just ping me tmr
<lool-> ack
monstr has quit [Remote host closed the connection]
ornxka has quit [Quit: No Ping reply in 180 seconds.]
ornxka has joined #lima
gcl has quit [Quit: leaving]
<Marex> lool-: so did you get it going now ?
<lool-> Marex: so when I use the dt fragment you've pasted earlier, I get an error about the IRQ pmu from the lima driver:
<lool-> (and a kernel panic while parsing the dt)
<Marex> lool-: isn't that optional IRQ ?
<Marex> lool-: I only ever tried it on mainline 5.7-rc and 5.4.40 with Lima patches backported, I have no idea whether there might be a problem in one of those 1200+ vendorkernel patches on top of outdated Linux 5.4.0, sorry
<lool-> yup
<lool-> so I'm looking at Documentation/devicetree/bindings/gpu/arm,mali-utgard.yaml
<lool-> and it seems to list 4-5 interrupt names
<lool-> but indeed only first 2 required
<Marex> look at drivers/gpu/drm/lima/lima_device.c
<lool-> there's indeed some delta with mainline, but didn't seem to be too much about adding/removing dt dependencies, more internal rework
<Marex> you very likely need all the patches between 5.4 and at least 5.7rc, there are a lot of fixes there
<Marex> or just switch to 5.7rc or next
cwabbott has quit [Ping timeout: 272 seconds]
cwabbott has joined #lima
<lool-> I took a quick look at rc5, but it doesn't seem to have nearly enough drivers to get the board to boot or display anything
<lool-> I guess I could try to get lima working headless for now and then try to backport it into the xlnx tree, sounds like quite the journey :)
psydread has quit [Remote host closed the connection]