2017-11-15

<willmore> montjoie, I'd like to use 12V to distribute power to the boards. Each board would have a polyfuse of the appropriate rating followed by a DC/DC converter which feeds the board.
<montjoie> willmore: what you would recommand for limiting power, just a simple fuse ?
<willmore> montjoie, same here. I guess we live on the edge. I wouldn't recommend it to people whos electrical skills I'm not confident of.
<montjoie> I use one without power limiting....
<montjoie> I tried as suggested by icenowy[m] to drop DRAM_CLK to 572
<montjoie> on my pine64 I got python[24274]: unhandled level 2 translation fault (11) at 0x00000000, esr 0x92000006, this is reproductibbe any idea ?
<montjoie> wow a 3G RAM variant!
<montjoie> icenowy[m]: so should work out of the box ?
<icenowy[m]> montjoie: another dw part for pcie :-)
<montjoie> the PCIE seems really low documented
<montjoie> icenowy[m]: yeah seems dedicated:(
<montjoie> and a second Crypto engine dedicated to AES...
<montjoie> I was speaking about EPHY ethernet
<montjoie> and now I understand better CE, the doc said "only different algo run in parallel"...
<montjoie> XTS in crypto engine!
<montjoie> wow an EPHY register list
<montjoie> I see some little add in EMAC doc
<montjoie> yeah

2017-11-14

<smaeul> montjoie: http://sholland.org/paste/a64_timer_failures_1.log is from a few hours of all 4 cores spinning reading the timer

2017-11-13

<montjoie> does it was sent for upstream ?
<montjoie> and why I get it only now, and not before ?
<smaeul> montjoie: this isn't an ARM errata, even though it's with the ARM architectural timer, since the timer is implemented outside the IP block provided by ARM
<montjoie> smaeul: I have all A53 ERRATA enabled
<smaeul> montjoie: icenowy[m]: do you have timer bug workaround applied on A64?
<montjoie> but strange it worked before, perhaps recent uboot made DRAM higher ?
<montjoie> yes, got it only on pine64
<montjoie> I hit also some strange internal compiler error
<montjoie> yeah I hit some rcu/hungtask/other on a64 since some days
<montjoie> wens: will fix

2017-11-12

<montjoie> no

2017-11-11

<montjoie> but now I see it
<montjoie> my mail reading lag
<montjoie> now PHY is working
<montjoie> icenowy[m]: you re right
<icenowy[m]> montjoie: pine64 don't use reg_gmac_3v3 at gpio
<montjoie> I believed that it can be found un schematics
<montjoie> perhaps but I need tofind it
<montjoie> perhaps i didnt search the right term
<montjoie> how to fill reg_gmac_3v3 ? (find which GPIO is used). I read pine64 schematics but found nothing

2017-11-10

<montjoie> wens: with your config on next it works, so a strange unrelated config is necessary
<wens> montjoie: nothing comes to mind :/
<montjoie> so unrelated
<montjoie> I miss only REGULATOR_PWM
<montjoie> mmc probe is deffered and seems never reprobed
<montjoie> wens: I have same MMC_xxx and AXP_xxx options than you, what can I miss for making MMC working ?
<wens> montjoie: probably
<montjoie> bad dtb update ?
<wens> montjoie: yes it works now
<montjoie> wens: "ok, this time it worked" <= so it dwmac work ?
<montjoie> wens: .config need to be sorted for a correct diff sorry...
<montjoie> icenowy[m]: I have AXP compiled in
<wens> the duplicate reg_dc1sw is likely because montjoie and I both posted patches adding it
<wens> montjoie: if I didn't have it selected, it wouldn't even probe, and you wouldn't see the errors I posted
<icenowy[m]> maybe montjoie didn't compile in AXP support?
<wens> montjoie: CONFIG_DWMAC_SUN8I=y
<montjoie> wens: you do nto have select DWMAC-sUN8I
<montjoie> wens: my bpim64 does not have the same problem than you (timeout TX) but mdio is found
<montjoie> wens: sorry didnt compare well
<montjoie> wens: could my problem be "missing AXP patches" ?
<montjoie> wens: yes please
<wens> montjoie: for arm64?
<montjoie> wens: could you share your working config ?
<wens> montjoie: no idea... I'm running custom config. arm64 defconfig is too fat

2017-11-09

<montjoie> so I miss something in https://paste.pound-python.org/show/aO9WqPQ3IJAUs8iofkkm/ but I have checked sunxi_defconfig and saw nothing missing in GPIO/MMC
<montjoie> wens: on my bpim64 I got the same wait for root device with sunxi-next
<montjoie> I do not see why it do that
<montjoie> I need to try sunxi-next
<montjoie> all my arm64 wont boot...
<montjoie> ouch its big

2017-11-08

<montjoie> anyway my issue seems not uboot/ATF related
<montjoie> but uboot is from git master:)
<ElBarto> montjoie: got it
<montjoie> ElBarto: I was saying beelding edge VS old image gievn apritzel 1 year ago (the first one)
<ElBarto> montjoie: ok, I was wondering if I missed a merge in atf from arm :) thanks
<montjoie> as documented in uboot
<montjoie> ElBarto: ATF from apritzel
<montjoie> KotCzarny: I will probably bisect anyway, for finding what breaks
<ElBarto> montjoie: bleeding edge ATF for pine64 ? or apritzel fork ?
<KotCzarny> montjoie: so, want the files for testing?
<montjoie> KotCzarny: on bpim64 its old apritzel uboot, on pine64 its beeding edge uboot/ATF. Both with linux-next
<KotCzarny> montjoie: my opiwin (a64) booted
<montjoie> my H5 (neo plus2) didnt work from the beginning, all my arm64 leaves me
<montjoie> yes
<montjoie> and my .config worked some month ago
<montjoie> yes it is pine64 and bpim64, but their kernel will be too old
<KotCzarny> montjoie: is your device supported by armbian?
<montjoie> yes
<montjoie> so something prevent the load
<montjoie> and so not now
<montjoie> but on last success boot, I see 2 "sunxi-mmc xxxx" lines
<montjoie> I have spam of hub 1-0:1.0: activate --> -22 also
<montjoie> but never see it after
<montjoie> sunxi-mmc 1c0f000.mmc: could not find pctldev for node /soc/pinctrl@1c20800/mmc0-pins, deferring probe
<montjoie> waiting for root
<montjoie> it seems that MMC probe is defered... forever
<montjoie> KotCzarny: setting LOW does not fix the problem
<montjoie> KotCzarny: will try it, but why it worked before...

2017-11-07

<KotCzarny> montjoie: i've used armbian's kernel dtb and it booted from sdcard
<montjoie> pine64/bpim4 was last tested on 4.12.0-rc4-next-20170606+
<montjoie> no I said that lots of problem hit arm64...
<montjoie> yes lkml.kernel.org/r/20171107083337.89952-1-kirill.shutemov@linux.intel.com
<montjoie> arm64 is broken by some other commits also
<montjoie> it worked before
<montjoie> do not know
<montjoie> got that problem on a64 also

2017-11-06

<montjoie> DWMAC_SUN8I (in menuconfig drivers net ethernet stmmacà
<codekipper_> montjoie: what configs do I need set for ethernet?.
<codekipper_> montjoie: on 20171018 it wouldn't start the rootfs, on the current tags it barfs straight after starting the kernel
<montjoie> codekipper: what is your crash problem on your pine64 ?
<montjoie> codekipper: I will try to bisect and praying that my mail got answered
<codekipper> montjoie: linux-next hasn't worked for me on my pine64 for a few weeks...https://kernelci.org/boot/all/job/next/branch/master/kernel/next-20171012/ seems to be the last booting tag

2017-11-05

<montjoie> anybody with a sun50i booting recent linux-next ?
<montjoie> wens: boot fail also without ubsan..
<montjoie> anybody with recent boot on bpim64 with ubsan enabled ?

2017-11-02

<montjoie> since I do not use mali and co, it should be doable
<montjoie> (I note, test BE kernel on future lab)

2017-11-01

<montjoie> ah ah distcc to the rescue! time to use all my boards
<montjoie> wens: sine linux-next are not published how to know the commit id for "fixes:" tag
<montjoie> strange the kbuild robot does not detect it
<montjoie> I need to add a one by one add patch and compile in my check... that's my fault
<montjoie> send the remaining ?
<montjoie> wens: mripard how can I fix that ?
<montjoie> the reg 3v3 is in another patch
<montjoie> arg
<montjoie> wens: ?
<wens> montjoie: looks like your dts patches don't quite compile

2017-10-31

<montjoie> should work, but you probably need to replace uboot and dtb also
<montjoie> they are ready and tested
<montjoie> mripard: do you want also patchs for A83T ? (and bpm3)

2017-10-30

<mripard> montjoie: awesome, thanks
<montjoie> will send next iteration in some hours
<montjoie> mripard: sorry I just came back from we
<mripard> montjoie: can you send a new iteration of your DT patches today?

2017-10-27

<montjoie> dwmac-sun8i merged in net!

2017-10-23

<montjoie> hoping the dwmac V8 will be the last one

2017-10-22

<montjoie> uart work, mmc work forever
<MoeIcenowy> montjoie: I think yes
<montjoie> hello MoeIcenowy does linux-next support R40 MMC ?

2017-10-20

<montjoie> do not miss "lava in a box" tuesday:)

2017-10-19

<montjoie> perhaps even IPV6 CRC is done
<montjoie> this part is common, so dwmac-sun8i do what stmmac do
<montjoie> wens: why ?
<montjoie> but only for IP
<montjoie> for me it work both in sun8i-emac and dwmac-sun8i
<montjoie> wens: auto CRC ? IP CRC ?
<wens> montjoie: IIRC auto CRC doesn't work with sun8i emac, right?
<montjoie> martinayotte: thanks for the test

2017-10-18

<martinayotte> montjoie : I've got my rescue SDcard, then I've try "reg = <7>;" and got "eth0: stmmac_open: Cannot attach to PHY (error: -19)" ... Now I'll try with "reg = <1>;" ...
<martinayotte> montjoie : I've just put 7, and my board doesn't boot any more ... coincidence ? since it was from eMMC, I need to prepare an emergency SDCard to boot.
<montjoie> probably 7 is the right
<montjoie> martinayotte: warning: loss of ethernet if it is not the right value
<montjoie> martinayotte: try to change in DT the reg = <0> to 1 or 7
<martinayotte> montjoie : I've a bpim2+ with Armbian Next 4.13.0, what/how should look to help you ?
<montjoie> I seek someone with a bpim2+ which could confirme the address of the etherney PHY (1 or 7)

2017-10-17

<montjoie> crypto ECC
<MoeIcenowy> montjoie: H6 also mentions ECC
<montjoie> old usermanual of H3 speaked about ECC also
<montjoie> I didnt bench RSA for now
<MoeIcenowy> montjoie: P.S. CE on H6 claims RSA
<montjoie> for speed no
<montjoie> mmmmmmhhh joker
<ElBarto> montjoie: does the crypto accel is useful on A64 which have the crypto instruction ?
<montjoie> I believed it was totaly done
<MoeIcenowy> montjoie: done? on which variants?
<montjoie> ElBarto: crypto since dwmac was already done
<ElBarto> montjoie: which one ? crypto ?
<montjoie> I wanted to port my driver to BSD but some ENOTIME

2017-10-08

<montjoie> needed swap on nfs
<montjoie> bulding gcc on H3 is a disaster:)

2017-10-06

<montjoie> you need the dwmac-sun8i driver now:)
<montjoie> v6 is for sun8i-emac, its old and unmaintained
<antony> montjoie v6 is the latest? I tested v5. It works great on Nanopi Neo Plus2. I am wondering, would it help if I send a Tested-by?
<paintenzero> montjoie: wens: Thank you!
<montjoie> I need to send a better mail for my latest problem, since nobody answer it
<montjoie> near latest:)
<wens> paintenzero: fetch montjoie's latest patches
<wens> montjoie: if the discussion about EMAC mdio-mux continues, I wonder if we can bring back support for a83t/a64 first?

2017-10-05

<paintenzero> montjoie: Do you have all cpu cores online? The problem is I have only 1 core available on any 4.x kernel I compiled :( How do you make your kernel run all 4 cores?
<montjoie> the filename is dwmac-sunxi.c but yeah the modulename is sun7i-dwmac
<montjoie> as I read the mail, its clearly a pinctrl problem (out of my knowledge)
<montjoie> A TRAP!
<montjoie> its dwmac-sunxi not dwmac-sun8i:)
<montjoie> no
<montjoie> I do not know well gmac on A20
<KotCzarny> i think montjoie might be able to answer that
<montjoie> (many orangepi) (many 4.x)
<montjoie> but your question is unprecise
<montjoie> yes

2017-09-28

<ElBarto> montjoie: thanks
<ElBarto> montjoie: what is conmux ?
<wens> montjoie: it's co-hosted with OSS (that's LinuxCon, ContainerCon, CloudCon rolled into one) and the kernel summit, so it's really huge
<montjoie> wow ELCE is full of conference
<montjoie> but the "lab in a box" demo will have my software stuff
<montjoie> not moving
<montjoie> yes
<wens> montjoie: you are with baylibre now?
<montjoie> it's planned
<montjoie> I have limited material, my child have it as a target for regain stuff
<KotCzarny> montjoie: yu can stabilize the construction by adding one column on the left
<montjoie> only one lab use allwinner, I think I know who to contact for fixing that
<MoeIcenowy> montjoie: for kernelci there's a problem now
<montjoie> I create the new baylibre lab:)
<wens> montjoie: are you adding a new lab to kernelci?
<montjoie> currently you will have to document a yaml file with your setup and it will generate all config files
<montjoie> when my stuff is ready, I will upload it
<montjoie> with conmux
<montjoie> (old picture, it is higher now)
<montjoie> lego tower to the rescue http://sunxi.montjoie.ovh/cluster.jpg
<montjoie> If you need help to setup a custom lava/kernelci does not hesitate to ask me:)
<plaes> montjoie: cool, I actually received bunch of extra hw (relay controller + power supplies) last night for setting up CI ;)
<wens> montjoie: huh?
<montjoie> since I work now on kernelci, I goal to add more allwinner devices on it:)

2017-09-25

<montjoie> wens: so could you merge the current serie ? or do you prefer I resend with the additionnal patch ?
<montjoie> wens: the changes you ask on DT are unrelated to the serie (fix warning) so I will made then in a separate patch (furthermore it is big)

2017-09-24

<montjoie> 48h without mail
<montjoie> gmail burn!!!!

2017-09-23

<lurchi_> montjoie: git send-email ?
<montjoie> pfff cannot send patch, gmail hating mutt...

2017-09-21

<montjoie> many
<montjoie> MoeIcenowy: it seems that something add may mac address
<MoeIcenowy> montjoie: today I'm testing Ethernet on my OPi PC using 4.13 with the removal reverted, and the MAC says "dwmac-sun8i 1c30000.ethernet eth0: Too many address, switching to promiscuous"

2017-09-19

<montjoie> argulp: with my branch no need to hack DT:)
<argulp> Merci montjoie, I'm checking Documentation/devicetree/bindings/net
<montjoie> KotCzarny: a little more, the internal/external is now represented with a mdio-mux
<KotCzarny> montjoie: is it just dt fix or something more?
<montjoie> I wait for rob answer on mainling list
<montjoie> and updating my github branch...
<montjoie> like updating http://linux-sunxi.org/Sun8i_emac
<KotCzarny> montjoie: maybe you should create simple howto on wiki?
<montjoie> sun8i-emac is obsolete
<montjoie> it is te only one
<montjoie> ok thanks
<montjoie> ok, I see bits of infos on mail, I will try to centralize them
<MoeIcenowy> montjoie: interesting... but I'm not doing on it
<montjoie> GIC broken, TZPC etc...
<montjoie> MoeIcenowy: could you make an entry in the wiki on secure boot breakage ? like a secure status matrix:)

2017-09-17

<montjoie> so the mac could control only one
<montjoie> no the register for selecting it is boolean
<montjoie> Putti: vagrantc yes another series is necesseray, but I wait on rob answer to andrew comment about mdio child status

2017-09-08

<montjoie> forgeted to do it:(
<montjoie> Moe_Icenowy: I sent the series this morning
<montjoie> I will send the serie today
<montjoie> I will try
<Moe_Icenowy> montjoie: is it still possible to get dwmac-sun8i work in 4.14?

2017-09-05

<montjoie> vagrantc: yes just need to send patches but net-next is closed
<montjoie> vagrantc: how to represent the mdio-mux of h3
<montjoie> and I need the mdio-mux patch to be accepted (reviewedby ffaineli yesterday)
<montjoie> MoeIcenowy: the cover letter is ready BUT net-next is closed
<MoeIcenowy> montjoie says that he lacks a proper cover letter for the ready patchset

2017-09-04

<MoeIcenowy> montjoie: I now doubt the "Die bond ID" in SS/CE is used by Allwinner to check the chip variant

2017-09-02

<montjoie> series is ready, just need to finish cover letter
<montjoie> MoeIcenowy: bringing back in 4.14 is my goal
<MoeIcenowy> montjoie: is it possible to get A64/A83T EMAC support back in 4.14?

2017-09-01

<montjoie> too late
<montjoie> and perhaps after review some patch will squash "doc for example"
<montjoie> my plan was to send in the series "4 patch revert² then patch for mux"
<montjoie> plaes: I mean revert the revert
<MoeIcenowy> montjoie: we are all M-started ;-)
<plaes> montjoie: IIRC that stuff was already reverted
<montjoie> perhaps for the first wave I keep patch splitted and in final they will be squashed
<montjoie> Do I fusion my patchs or I do in step "revert the revert then patch it"
<montjoie> sorry I mean mripard:
<montjoie> MoeIcenowy: wens I will send my serie for dwmac-sun8i mux. BUT I need to restore DT.

2017-08-31

<montjoie> I patch it for testing sunxi hardware
<montjoie> vu libkcpi know only software algs
<montjoie> yes
<MoeIcenowy> montjoie: P.S. can libkcapi also use other algs from sun4i-ss?
<montjoie> CONFIG_CRYPTO_TEST
<MoeIcenowy> montjoie: which kernel option is for tcrypt?
<montjoie> MoeIcenowy: see http://sunxi.montjoie.ovh/#prng but you need libkcapi
<montjoie> let me check if instructions for doing it are good
<montjoie> you could also test rng
<montjoie> modprobe tcrypt
<montjoie> if no error is printed, all is good
<montjoie> you should see sun4i-ss algos
<montjoie> see cat /proc/crypto
<MoeIcenowy> montjoie: it says "[ 0.755981] sun4i-ss 1c15000.crypto-engine: Die ID 7"
<montjoie> yes
<montjoie> V3s will have a mux... of one MDIO
<montjoie> MoeIcenowy: V3s will need mux
<montjoie> for most algos
<montjoie> when each algo is registred, tests are done
<montjoie> yes
<montjoie> so the driver is used at load
<montjoie> MoeIcenowy: at load, crypto api autotest some algos
<montjoie> modprobe it, you should rapidly see if it works:)
<MoeIcenowy> montjoie: how to test whether sun4i-ss is working on a SoC?

2017-08-30

<montjoie> pfff my emac mux series goes big, finding too many bugs

2017-08-29

<montjoie> miasma: latest uboot git
<miasma> montjoie: you're using u-boot v2007.something?
<montjoie> miasma: do not hit this problem
<miasma> montjoie: what about "sunxi SPL version mismatch: found 0x02 > maximum supported 0x01. You need a more recent version of this (sunxi-tools) fel utility."
<miasma> montjoie: why don't the devs fix their scripts? e.g. dtoc has some simple issues like using 'print variable' instead of 'print(variable)'
<montjoie> miasma: solved it by created a "python" shell script which do python2 and export PATH=.:$PATH

2017-08-26

<montjoie> yeah I will try install armbian