<amstan>
bashintosh: you just have to reserve a bunch of memory
Sadneophyte has joined #linux-rockchip
<bashintosh>
amstan: did I tell you that you're the man? :) So that region of reserved memory will not be used?
lautriv has joined #linux-rockchip
<amstan>
bashintosh: i think so yes
<amstan>
we used that to workaround a bug where a small memory region couldn't be DMAd to, so we just called it invalid for everything to make it easy
<bashintosh>
amstan: Magic! I need to "simulate" a 1GB RAM environment (RK3288), where 2GB physical are available - will give a try to the patch. Thanks!!
<amstan>
bashintosh: so replace that region from that patch with 0x80000000 0xffffffff
<bashintosh>
amstan: the second value in the patch (0x1000000) is the lenght right? So should it be 0x80000000 as start region then 0x40000000 the lenght (~1GB)?
<amstan>
bashintosh: 0x40000000 for both actually
<bashintosh>
amstan: gotcha!
<amstan>
i tought you wanted to go from 4GB to 2GB, you need 2->1
<bashintosh>
Yes 2->1GB - let's see what happens ^_^
<amstan>
why are you doing this btw?
<bashintosh>
because I want to see how Lollipop performs with only 1GB of RAM, but I don't have a device RK3288 based with 1GB so I thought about "simulating" instead..
naobsd has joined #linux-rockchip
<bashintosh>
amstan: in theory Lollipop should handle 1GB pretty well compared to KitKat..
<amstan>
that display is nice, but if you run it with arm it'll be really slow
<amstan>
and i don't understand why something like that needs to be modular
<amstan>
is that rk3288? it says 1.5GHZ, 3288 runs at 1.8
<bashintosh>
amstan: why slow? RK3288 supports even higher resolutions, up to 4K
<amstan>
at 4k you barely have the ram latency to do anything else but move the pixels around
<amstan>
bandwidth*
<bashintosh>
amstan: some downclock it to 1.4GHz even. Heat is a concern, so for now 1.5GHz seems like a good tradeoff
<bashintosh>
well I have the 2k display right here and so far so good in terms of performance - with 2GB of RAM - that's why I want to try 1GB :)
<amstan>
you should do some memory speed tests when running at 2k vs 1080p
<amstan>
and see the difference
Sadneophyte has quit [Ping timeout: 246 seconds]
<bashintosh>
mmh, good tip - might not be so bad - been playing 4k videos with no trouble so far, but I got eDP running properly just last week so still in the works.
Sadneophyte has joined #linux-rockchip
Sadneophyte has quit [Ping timeout: 240 seconds]
levd1 has joined #linux-rockchip
levd has quit [Ping timeout: 256 seconds]
levd1 is now known as levd
Sadneophyte has joined #linux-rockchip
levd1 has joined #linux-rockchip
RayFlower has quit [Ping timeout: 246 seconds]
levd has quit [*.net *.split]
levd1 is now known as levd
RayFlower has joined #linux-rockchip
levd1 has joined #linux-rockchip
<bashintosh>
amstan: mmh, /proc/meminfo reports the same values with/without patch - should I see a difference there after I apply the patch?
levd has quit [Ping timeout: 260 seconds]
levd1 is now known as levd
<naobsd>
bashintosh: which loader and kernel config are you using?
<bashintosh>
naobsd: u-boot and default Firefly config with minor customisations
<naobsd>
any detail?
<bashintosh>
not using VGA/HDMI, using debugfs and USB-gadget and EHCI support disabled - I think that's it
Sadneophyte has quit [Ping timeout: 240 seconds]
levd has quit [Ping timeout: 256 seconds]
<naobsd>
well, detail about "u-boot" and "default Firefly config with minor customisation"
<naobsd>
any specific information?
levd has joined #linux-rockchip
<naobsd>
ah, lollipop image...
<naobsd>
bashintosh: please forgot previous questions. how did you update dtb?
levd has quit [Remote host closed the connection]
levd has joined #linux-rockchip
<bashintosh>
naobsd: I rebuilt the whole kernel :) then flashed resource/kernel
<naobsd>
bashintosh: are you sure boot partition doesn't have kernel and resource?
markm has joined #linux-rockchip
<bashintosh>
naobsd: I've used 'mkimage.sh withoutkernel' the first time I build the SDK - changes to kernel/DT do take effect when I flash kernel/resource (verified with different clocks init freqs for example)
<bashintosh>
s/build/built
<naobsd>
hm. then, sorry, no idea...
<bashintosh>
:) thanks anyway
<bashintosh>
I wonder what reads the "reserved-memory" node from the DT - maybe it's currently ignored :(
<naobsd>
I cannot find any memory node in firefly dts
<naobsd>
I guess it's detected & modified by u-boot
<bashintosh>
right - I noticed that too - seems to be only in the DTS that amstan pointed me to - trying to find the full source file
<naobsd>
anyway, I'm not sure, it needs confirmation
<bashintosh>
naobsd: I guess you're right - I run "grep -RnisI --color=tty reserved-memory *" in the kernel root directory, which only returned the patch I just added :(
<bashintosh>
naobsd: amstan adding "mem=1G" to the kernel parameters did the magic ^_^
GriefNorth has joined #linux-rockchip
phoenix has joined #linux-rockchip
phoenix is now known as Guest30758
<Guest30758>
does anyone here have access to a recent version of the rockchip linux NAND driver patch?
<Guest30758>
(apparently labelled "[RK292X&RK30&RK31&RK32_ANDROID-SDK] linux-nand-driver_Patch_V2.8" or similar according to previous logs of this channel)
levd1 has joined #linux-rockchip
levd has quit [Ping timeout: 246 seconds]
levd1 is now known as levd
levd1 has joined #linux-rockchip
GriefNorth has quit [Remote host closed the connection]
levd has quit [Ping timeout: 250 seconds]
levd1 is now known as levd
GriefNorth has joined #linux-rockchip
GriefNorth has quit [Quit: WeeChat 1.3]
GriefNorth has joined #linux-rockchip
chrs_ has quit [K-Lined]
tizbac has quit [K-Lined]
bashintosh has quit [Ping timeout: 256 seconds]
chrs_ has joined #linux-rockchip
levd has quit [Ping timeout: 264 seconds]
levd has joined #linux-rockchip
bashintosh has joined #linux-rockchip
Luke-Jr has quit [Read error: Connection reset by peer]
Luke-Jr has joined #linux-rockchip
levd has quit [Ping timeout: 252 seconds]
levd has joined #linux-rockchip
lautriv_ has joined #linux-rockchip
lautriv has quit [Ping timeout: 244 seconds]
markm has quit [Ping timeout: 240 seconds]
rory096 has quit [Ping timeout: 272 seconds]
mcan has quit [Read error: Connection reset by peer]
mcan_ has joined #linux-rockchip
sanjay has joined #linux-rockchip
<sanjay>
hello
mcan_ has quit [Read error: Connection reset by peer]
mcan_ has joined #linux-rockchip
sanjay is now known as sanjay85
premoboss has joined #linux-rockchip
GriefNorth has quit [Ping timeout: 246 seconds]
nighty-_ has joined #linux-rockchip
tizbac has joined #linux-rockchip
GriefNorth has joined #linux-rockchip
markm has joined #linux-rockchip
naobsd has quit [Quit: naobsd]
levd1 has joined #linux-rockchip
levd has quit [Ping timeout: 268 seconds]
levd1 is now known as levd
levd1 has joined #linux-rockchip
levd has quit [Ping timeout: 268 seconds]
levd1 is now known as levd
xcasex has quit [Read error: Connection reset by peer]
xcasex has joined #linux-rockchip
levd1 has joined #linux-rockchip
levd has quit [Ping timeout: 256 seconds]
levd1 is now known as levd
nighty-_ has quit [Ping timeout: 272 seconds]
sanjay85 has quit [Ping timeout: 246 seconds]
nighty-_ has joined #linux-rockchip
mcan_ has quit [Read error: Connection reset by peer]
levd has quit [Ping timeout: 252 seconds]
Guest30758 has quit [Quit: Leaving]
Helle has quit [Ping timeout: 260 seconds]
Helle has joined #linux-rockchip
Helle has quit [Ping timeout: 256 seconds]
Helle has joined #linux-rockchip
premoboss has quit [Remote host closed the connection]
naobsd has joined #linux-rockchip
ssvb has quit [Ping timeout: 252 seconds]
<lautriv_>
just had a look on the wiki how rockchips kernels differ and supposed to be build, the "rm -rf $INSTALL_MOD_PATH" is redundant since make modules_install cares about actual content.
nighty-_ has quit [Quit: Disappears in a puff of smoke]
<lautriv_>
need some suggestions on the kernels config. my box is about 80% radxa but differs on some details, especially larger Flash/RAM and probably some GPIO, thingie is a tv-box may i go rather for the rk3188_Tronsmart_MK908_galland_defconfig or maybe rk3188_dongle_defconfig ?
<rperier>
if you want to lose weight, I have better than a diet... just read a clktree driver for a soc inside the kernel... :D (I am reading the clktree driver for zynq , for my needs at work... sweet baby jesus...)
<rperier>
:D
<rperier>
(it's a metaphor of course)
nighty-_ has joined #linux-rockchip
nighty-_ has quit [Max SendQ exceeded]
nighty-_ has joined #linux-rockchip
ssvb has joined #linux-rockchip
RayFlower has quit [Ping timeout: 246 seconds]
RayFlower has joined #linux-rockchip
<mmind00>
rperier: yeah, clock drivers are fun ;-)
<rperier>
especially when you refer all your clocks by "<&clkc ID>" from the devicetree, it's very helpful :D
<rperier>
(well, clock-names is supposed to help yes... )
<rperier>
something like the rockchip kernel is nice, because you use macro-like constants
<naobsd>
Orion R68 rk3368 box was arrived
<naobsd>
there are kind UART2 TX/RX mark on board
<naobsd>
mmind00: is somewhat-stable good for rk3368 too?
<mmind00>
naobsd: of course, but there aren't any differences between mainline and my dev-branch for the rk3368 currently
<naobsd>
I see, thanks
<naobsd>
I have to prepare somewhat working dts at first ;)
<mmind00>
naobsd: I'd guess they'll have followed the reference layout, so it should be very similar to the r89 (except that your board is nearly square)
GriefNorth has quit [Ping timeout: 265 seconds]
<naobsd>
I guess so
<mmind00>
naobsd: mainly look if a real pmic magically appeared ;-) ... on the r89 it's just a syr82x for the cpu and some gpio-controlled switches
nighty-_ has quit [Ping timeout: 246 seconds]
<naobsd>
hmm, on stock kernel,
<naobsd>
only regulator-dummy, vdd_arm, and vdd_logic are available...?
nighty-_ has joined #linux-rockchip
<naobsd>
oops. Orion R68 SDK must be very w.i.p. state...
<naobsd>
anyway
<naobsd>
diff -u rk3368-box-r88.dts rk3368-box.dts shows very little difference
rory096 has joined #linux-rockchip
<naobsd>
power key, dwc power, led
<naobsd>
mmm, r88.dts in Android sdk says dwc power gpio is active low...
<naobsd>
mmind00: is dwc power gpio really active high on r88?
<mrutland>
naobsd: I was able to get v4.3-rc1 happily running on an Orion r68 meta using the r88 dts and mmind00's clock patch.
<naobsd>
mrutland: oh
<mmind00>
naobsd: as I'm booting my rootfs from an usb stick currently, I guess so :-)
<naobsd>
yes, it should work
<naobsd>
dwc host might be always on on r68
<mrutland>
The UART2 pins are standard 115200n8 TTL
<naobsd>
mmind00: android sdk dts says "gpio0 4" is host, "gpio0 5" is otg, both active low
<naobsd>
mmind00: but mainline r88.dts says both are "gpio0 4"
<naobsd>
I guess it's different board(rev?)
<naobsd>
anyway, minor difference for r68
<mmind00>
naobsd: according to my schematics, gpio0_4 is host_vbus_drv ... gpio0_5 just connects to TP703
<mmind00>
naobsd: and the vcc_host_5v coming form there connects to both usb ports the r88 board has
<naobsd>
mmind00: probably dts in android sdk is for different board or rev
<mrutland>
naobsd: It's also possible that it's wrong in the android SDK -- connecting OTG to my board with the stock kernel killed the board each time...
<mrutland>
Though I haven't tried that with mainline, so it could be wrong there too :/
<naobsd>
mmm I have to attach TV and mouse to use "su" on stock android... ;)
<lautriv_>
naobsd, no UART ?
<mrutland>
naobsd: If you power it on while holding the reset button, you'll get a root shell on the UART
cosm has joined #linux-rockchip
<naobsd>
well
<naobsd>
I know well about how android works
<naobsd>
UART works of course
<naobsd>
if you're using user build (released ver.) firmware, shell on console is not root
<naobsd>
and when type "su", superuser app asks grant it or not. this is the reason I have to attach TV/mouse
<naobsd>
and if holding button for loader mode while power on, and only if OTG is not connected, it goes recovery mode which gives root shell on console
<naobsd>
hmm, interesting, SD card is working if I removed UART2 RX/TX wires w/o reboot etc
<naobsd>
ok, my box seems fine
<naobsd>
let's start hacking :)
<lautriv_>
naobsd, had this happen when i started on this box, appears to be sensible to ground-loops.
<naobsd>
lautriv_: sorry, I'm not sure what's "this"
sanjay85 has joined #linux-rockchip
<sanjay85>
hi
<sanjay85>
naobsd> are u there ? :)
<lautriv_>
naobsd, the wire from USB via ftdi to your pins has a specific lenght where the GND is put on the host, the PSU has another wire and is not conducting GND to the host, one more wire via USB/OTG may produce a hovering signal because the different resistance of those lines.
<naobsd>
sanjay85: hi
<naobsd>
lautriv_: I don't have any problem
<lautriv_>
naobsd, except the SD when UART plugged
<sanjay85>
how are you ?
<sanjay85>
i just have a question
<sanjay85>
i'm very sorry for disturb
<sanjay85>
is there a way, to have a "volatile" partition ?
<sanjay85>
because mmcblk0 is volatile
<sanjay85>
i can write a file
<sanjay85>
execute
<sanjay85>
without any error
<sanjay85>
but when umount / mount
<sanjay85>
nothing
<sanjay85>
i can format, it's working, then when mount, file is coming back
<sanjay85>
i mount /system, copy debugfs, into /system
<sanjay85>
i can execute debugfs
<sanjay85>
when umount, mount => debugfs disapair
<sanjay85>
same as e2fsck, repair all error, then when re-execute e2fsck, exactly same error inode
<sanjay85>
it's like volatile
<naobsd>
lautriv_: I'm talking about rk3368 now, no problem
<naobsd>
sanjay85: technically possible, but I guess it's just defected
<sanjay85>
so mmc is dead
<sanjay85>
hardward
<sanjay85>
hardware
karlp has quit [Ping timeout: 250 seconds]
<naobsd>
just my guessing, not sure at all
<naobsd>
oops
<sanjay85>
ok
<sanjay85>
thank you
<naobsd>
somewhat-stable is somewhat booted...but I have to modify cmdline ;)
karlp has joined #linux-rockchip
<sanjay85>
ok
<sanjay85>
can you explain me quickly somethink
<sanjay85>
when i wrote
<sanjay85>
on sdcard
<sanjay85>
d if=idb_sector_0.enc of=${DEV} conv=sync,fsync seek=64
<sanjay85>
etc
<sanjay85>
i can boot with sdcard
<sanjay85>
but only recovery and rockchip loader
<sanjay85>
is there a way to see /system into sdcard ?
<sanjay85>
i really like kernel and system is not in sdcard, if i want to modify kernel and fstab for /system into sdcard
<sanjay85>
i can mount anything from sdcard
<naobsd>
sanjay85: hard to explain with few words...a lot of things should be done
<sanjay85>
just wan't to know the start ?
<sanjay85>
from where i must start ?
<naobsd>
sanjay85: I think there is no quick start guide for it
<sanjay85>
ok
<sanjay85>
:(
<sanjay85>
thank you
<naobsd>
mrutland: which CMDLINE are you using for mainline on R68?
<naobsd>
mmm earlycon=uart8250,mmio32,0xff690000 should be ok...