<y0g1>
Does A20 from Cubietruck has got support for hardware IOMMU?
<MoeIcenowy>
I think the answer is no...
<y0g1>
I would like to know if I can setup in linux kernel config options: "ARM Ltd. System MMU (SMMU) Support" and "ARMv7/v8 Long Descriptor Format" and "ARMv7/v8 Short Descriptor Format"
<y0g1>
MoeIcenowy: thx, but if I will mark the options? it will work or not?
<MoeIcenowy>
I think it will be safe not to select them
<y0g1>
ok
y0g1 has quit [Quit: thx for help]
bugzc has quit [Ping timeout: 250 seconds]
<mripard>
MoeIcenowy: there's no IOMMU on the Allwinner SoCs
<mripard>
so we basically don't care about those options
<mripard>
we can enable or disable it, it doesn't matter.
cnxsoft has quit [Remote host closed the connection]
cnxsoft has joined #linux-sunxi
<montjoie>
mripard: you do not have made any comment on my dt sid patch
<mripard>
my comment would have been the same one than Jean-Francois
<mripard>
and you didn't answer
apritzel has joined #linux-sunxi
<montjoie>
oups I didnt see any comment
massi has joined #linux-sunxi
<montjoie>
gmail spam...
yann has quit [Ping timeout: 250 seconds]
apritzel has quit [Ping timeout: 265 seconds]
<MoeIcenowy>
mripard: yes...
<MoeIcenowy>
mripard: is anyone working on A10/13/20 CCU?
fvogt_vps is now known as DimStar
DimStar is now known as fvogt_vps
<mripard>
MoeIcenowy: yes, it's done, I still need to send the patches
<mripard>
if you want to help, converting the various PRCM clocks would be great :)
benoitm974 has joined #linux-sunxi
<wens>
i wonder how we're supposed to reuse the current bindings
<wens>
seems doable, but involves some ugly code
jailbox has quit [Ping timeout: 244 seconds]
<mripard>
what current bindings are you talking about ?
<Wizzup>
igraltist: awesome!
repka_ has joined #linux-sunxi
arete74 has quit [Ping timeout: 250 seconds]
<MoeIcenowy>
mripard: PRCM clocks?
<MoeIcenowy>
make a sun8i-a23-r-ccu ?
<MoeIcenowy>
like sun8i-a23-r-pinctrl :-)
repka has quit [Ping timeout: 256 seconds]
jailbox has joined #linux-sunxi
<MoeIcenowy>
(but I maybe more interested in sun8i-h3-r-ccu, as my h3 device have debug UART, my A33 devices have only UART muxed with MMC
repka_ has quit [Quit: Leaving]
benoitm974 has quit [Remote host closed the connection]
repka has joined #linux-sunxi
<mripard>
MoeIcenowy: yes, that's what I meant :)
repka_ has joined #linux-sunxi
<MoeIcenowy>
but the r clocks are mysterious... should I refer to current dt?
<MoeIcenowy>
or is there any accurate documents?
<wens>
mripard: i was more inclined to keep the name 'sunXi-aXX-prcm'
<MoeIcenowy>
is it possible to make r-ccu a subnode of prcm?
benoitm974 has joined #linux-sunxi
<mripard>
wens: the PRCM is the whole block, not just the clocks, right?
<wens>
MoeIcenowy: that would not be a proper description
<wens>
mripard: yes, which also contains a power domain controls and what not
<MoeIcenowy>
to be honest I know nothing about prcm clock...
<wens>
MoeIcenowy: there's no real documentation for the whole block, just bits and pieces
<MoeIcenowy>
but what in prcm needs clock?
<wens>
the best documented one is probably the a80
repka has quit [Ping timeout: 260 seconds]
<wens>
MoeIcenowy: RSB uses it
<MoeIcenowy>
I know rsb and ir
<wens>
ah crap, the NMI IRQ is also in that block
repka_ has quit [Client Quit]
repka has joined #linux-sunxi
yann has joined #linux-sunxi
<MoeIcenowy>
according to sun8i-a23-a33.dtsi, there's clock gate for r-pio, r-timer, rsb, r-uart, r-i2c
florianH has joined #linux-sunxi
<wens>
MoeIcenowy: yes, there's also a clock tree in there (cpu-s -> ahb-s -> apb-s -> gates), and mod clocks for some things
<igraltist>
Wizzup: its was not so difficult to setup
<MoeIcenowy>
cpu-s is just ar100_clk?
<Wizzup>
I'm going to build a kernel for it now
<wens>
MoeIcenowy: yes
<MoeIcenowy>
what have mod clks?
<MoeIcenowy>
I know ir have one
<wens>
1-wire
<wens>
and i2s (on a80)
<MoeIcenowy>
oh it's still a simpler driver than main ccu :-)
<MoeIcenowy>
maybe we should start from sun8i-a23-r-ccu
<MoeIcenowy>
it's most simple
<wens>
i prefer we think it through first though
<MoeIcenowy>
oh...
<MoeIcenowy>
but how should the dt be organized?
<MoeIcenowy>
prcm@xxx { ccu@xxx { };}; ?
<wens>
ideally it's just one block, with no subnodes
<MoeIcenowy>
and then probe all subblocks via mfd?
<wens>
what sub-blocks are left?
Worf has joined #linux-sunxi
<wens>
for the a23 the kernel only uses clocks, resets, and the codec controls
<MoeIcenowy>
yes... now clocks and resets are still subnodes
<mripard>
wens: how can that work without subnodes ?
<wens>
depends on the kernel API i guess, but most device tree related calls let you specify which node you want to tie a resource to
<wens>
such as clocks, resets
<wens>
the prcm node would have a bunch of #X-cells for each kind of resource
<mripard>
and then a huge driver that does everything, from clocks to i2c?
<wens>
what i2c? the i2c is a separate block in the RTC domain
<wens>
you could still use mfd for the driver portion, just not the bindings
<wens>
i.e. have sub-devices in the kernel, but no sub-nodes in the DT
<mripard>
I'm against that
<mripard>
this is not something that scales
<mripard>
you can have a look at qschulz ADC patches if you want any proof of that
<mripard>
we had to fix stuff in basically all the drivers that we used, including some generic ones
<mripard>
just because we had to do it that way
<mripard>
so unless if it is for pure masochism (which might turn out to be a good argument :)) there's no reason to do that
<wens>
ok
<mripard>
(plus, not having to deal with MFD at all is also good)
<wens>
how about just a simple-bus / simple-mfd parent node
<wens>
and sub-nodes for 1) clocks, 2) resets, 3) analog codec controls (a23+)
<wens>
and on the a80 there's also the nmi interrupt controller in there
<wens>
at least the registers are grouped together
<mripard>
that would work for me
<wens>
clocks are mostly [0x0, 0x60)-ish, resets are [0xb0, 0xb4)
<wens>
on the a80/a83 the osc24M control is an outlier, at the very end of the block @0x1f4
apritzel has joined #linux-sunxi
<pcduino>
is it worth buying an A80 yet?
<wens>
pcduino: in terms of?
IgorPec has quit [Read error: Connection reset by peer]
<pcduino>
are most features working and ready to use?
<mripard>
pcduino: it's also kind of a chicken and egg issue
<mripard>
the support is also not very advanced because there's not a lot of people that have one
<pcduino>
ah so about 40%
<mripard>
and there's not a lot of people that have one because the support is not very advanced :)
<pcduino>
I will get one when finished with the A20
<jelle>
wonder how well the big.little works in the sdk version
<mripard>
jelle: probably not that great, it's probably using the IKS, and not the EAS
mhlavink has joined #linux-sunxi
matthias_bgg has joined #linux-sunxi
<Wizzup>
igraltist: I'm presuming that it wouldn't work on A10? I'm unable to select KVM in my kernel config, but I use my A10 to compile kernels for my A10 and A20 devices
<igraltist>
Wizzup: i dont know. i just have a cubietruck its with a20
<Wizzup>
ok, and you could just check the KVM option under Virtualisation?
<igraltist>
yes if you mean to enable theam
<Wizzup>
yes, I'm currently in kernel menuconfig
<Wizzup>
Ah. I had to enable LPAE
<igraltist>
yes this too
<Wizzup>
Now I can see it
<Wizzup>
Let's see how well it works with 4.7.8 grsec :D
<pcduino>
igraltist: do you have lvds working on the cubietruck?
<igraltist>
no is a headless machine
benoitm974 has quit []
<MoeIcenowy>
Wizzup: cubietruck with grsec?
<MoeIcenowy>
interwsting
<igraltist>
i have rsbac on it
<Wizzup>
MoeIcenowy: LIME2
<MoeIcenowy>
also interesting.
<Wizzup>
I've been running grsec for over 1-2 years now on my LIMEs
<Wizzup>
works fine
<MoeIcenowy>
with 3.4 or mainline?
<Wizzup>
mainline
<pcduino>
plaes: should I check changes into this branch?
benoitm974 has joined #linux-sunxi
benoitm974 has quit [Client Quit]
<MoeIcenowy>
pcduino: I have an A33 tablet LVDS work
<MoeIcenowy>
but I used simplefb
<pcduino>
igraltist: are you using NAND?
<pcduino>
I have no fb0
<igraltist>
pcduino: not working at moment
<mripard>
Wizzup: yeah, KVM won't work on the A10
<MoeIcenowy>
A10 do not have virtualization.
<MoeIcenowy>
(Because the core is Cortex-A8
<pcduino>
MoeIcenowy: A33 is sun8i
<Wizzup>
mripard: makes sense. thanks.
<MoeIcenowy>
pcduino: but I think A10 should work in this way
<mripard>
Wizzup: you need something with a cortex-a7, so anything newer than the A20/A31
<MoeIcenowy>
altering U-Boot config
<pcduino>
MoeIcenowy: plaes is working on the A10 LVDS and I am looking into the A20
<MoeIcenowy>
pcduino: it's only the DRM works
<MoeIcenowy>
making it just able to display something do not need DRM
<MoeIcenowy>
simplefb is enough
<pcduino>
I also need to work out how to add the additional gpio for the pcduino3b
<pcduino>
this is my first time using device tree :)
<MoeIcenowy>
I think chuwi-v7-cw0825 tablet is an example of A10 with LVDS
<MoeIcenowy>
oh
<Wizzup>
mripard: newer than A20? I thought the A20 should do kvm?
<MoeIcenowy>
it's not a good simple
<MoeIcenowy>
Wizzup: yes A20 do
<MoeIcenowy>
pcduino: a good example
<MoeIcenowy>
Hyundai A7HD
<MoeIcenowy>
it uses a LVDS screen
<MoeIcenowy>
and it's A10
<pcduino>
nothing in dts folder for *hy*
<mripard>
Wizzup: sorry, A20 and A31 included
<igraltist>
has the a10 good aes support?
<pcduino>
my bad
* pcduino
falm palms
<MoeIcenowy>
pcduino: please refer to U-Boot configs
<MoeIcenowy>
notice that U-Boot configs mix upper case and lower case
<MoeIcenowy>
(It's the U-Boot config magic... I've got two LVDS devices work, one A31s and another A33.
<pcduino>
sun4i-a10-hyundai-a7hd has no framebuffer@ chosen. Oh I was looking in mainine arch/arm/boot/dts. I made my own config for the pcduino 3b copiled it then copied it into the u-boot is this the wrong approach?
<ElBarto>
montjoie: but usually I just read the sources and learn from that when I don't know stuff
<ElBarto>
also man pages
<montjoie>
thanks, but I searched some more like ldd3 or kernelnewbies
<montjoie>
explaining API by API
<ElBarto>
I don't think we have such doc
<ElBarto>
the book "The design and implementaton of FreeBSD" might be what you're looking for
Leepty has quit [Ping timeout: 260 seconds]
<montjoie>
ok
Putti has joined #linux-sunxi
Leepty has joined #linux-sunxi
leviathanch has joined #linux-sunxi
<igraltist>
i read that the orange pi got to hot can someone confirm this?
jemk has quit [Ping timeout: 258 seconds]
<KotCzarny>
igraltist: there are many oranges
jemk has joined #linux-sunxi
IgorPec has quit [Ping timeout: 256 seconds]
<igraltist>
the smallest
<igraltist>
orange pi one
<KotCzarny>
in short: bigger board with more copper == better heat dissipation, higher voltage == higher temp, bad vendor kernel settings + small board == too hot
<KotCzarny>
if you use armbian it should be already optimized
<KotCzarny>
and it affects any board, not only oranges
<igraltist>
the question was more in general
<igraltist>
like the old p4 was a nice heater
<Net147>
especially the P4 prescott (tres hot)
<KotCzarny>
p4 has nothing on athlons
<KotCzarny>
;)
<KotCzarny>
but tbh, h3 has maximum temperature of 125C, he he
<Net147>
fry an egg on it =P
<KotCzarny>
well, you can use it for soldering :P
matthias_bgg has quit [Quit: Leaving]
<Net147>
I think there was a VIA Eden processor commercial a long time ago showing prehistoric humans evolving from using campfire as heater to using exhaust heat from old computers...
<KotCzarny>
i do that still
<KotCzarny>
when gaming my room gets noticably hotter ;)