2015-12-18

<montjoie> emac
<montjoie> for opipc it seems that I receive frame with my driver but without getting interupt for it, but it is a progress anyway

2015-12-14

2015-12-01

<wens> montjoie: congrats

2015-11-30

<camh> montjoie++ for MII. Happy to help test. I have orange pi plus and pc to test with
<montjoie> yeah the h3 MII is working
<deskwizard> montjoie, 16gb wouldnt be enough wouldnt it ?
<montjoie> deskwizard: and lots of ram for compiling on ram disk

2015-11-29

<montjoie> it seems that they forget that a64 is 64bit
<montjoie> and the crypto engine does less functionnality for RSA
<montjoie_> interesting the a64 datasheet, the emac is the same as h3 but with heavy copy/paste, since dma regsiter take 32bit address

2015-11-27

<ssvb> montjoie: the 3.4 kernel works just fine on H3 with the U-Boot bootloader, but U-Boot needs to define the machine id and also CONFIG_SYS_BOOTM_LEN needs to be increased
<montjoie> ssvb: yes I try to made the MII working and I think I gave clk problem, need to compare with a 3.4
<ssvb> montjoie: trying to boot the 3.4 kernel on H3?
<montjoie> thanks for the info
<NiteHawk> montjoie: sunxi-fel should manage that just fine. but: you need a suitable combination of u-boot and kernel. recent mainline u-boot might not boot older 3.4 kernels without some tweaks: https://linux-sunxi.org/Mainline_U-Boot#Legacy_kernel_won.27t_start
<montjoie> hello does sunxi-fel could boot an old 3.4 woth a script.bin ? or I need usb-boot
<jelle> montjoie: yeah I know hans
<montjoie> jelle: Got also an orange pi pc, with the two repo you could boot it (without network)
<libv> montjoie: :)
<montjoie> and now back to H3 ethernet driver:)
<libv> montjoie: thanks
<montjoie> I will add a mainline section in /h3
<montjoie> nothing in one place
<libv> montjoie: explain/rephrase please?
<GeorgeF> montjoie: thank you, I will have a look
<montjoie> libv: too dispatched
<montjoie> branch sunxi-wip
<montjoie> GeorgeF: see hans de goede repo

2015-11-24

<montjoie> but no network
<montjoie> ssvb: it works http://pastebin.com/1a8cUMDu thanks
<montjoie> hans
<montjoie> ssvb: I use hand u-boot (branch sunxi-wip)
<ssvb> codekipper: if montjoie used that u-boot branch, then this perfectly explains his kernel oopses
<ssvb> montjoie: or some other fork?
<ssvb> montjoie: is your U-Boot from http://git.denx.de/?p=u-boot.git;a=summary ?
<ssvb> montjoie: OK, but the oops is still an indication that something might be not very reliable, and the DRAM clock speed is the primary suspect
<montjoie> ssvb: I will try the downgrade also
<montjoie> I miss boot.scr
<codekipper> montjoie: looks like you're missing boot.scr/uEnv.txt on the card
<ssvb> montjoie: for a rather conservative choice, just change it in U-Boot defconfig to 480 MHz
<ssvb> montjoie: you can try to downclock DRAM from 672 MHz to something lower
<montjoie> but the kernel crash after http://pastebin.com/mTKXWDkF
<montjoie> ssvb: I can start uImage at hand so there something in mmc
<ssvb> montjoie: there is still no USB or Ethernet support for H3 in U-Boot
<ssvb> montjoie: do you have anything on the sd card?
<montjoie> does someone see something wrong with http://pastebin.com/j1nsBMGa, my u-boot wont autostart anything

2015-11-23

<montjoie> does someone see something wrong with http://pastebin.com/j1nsBMGa, my u-boot wont autostart anything
<montjoie> plaes: thanks for the test

2015-11-22

<plaes> montjoie: could you add to your sunxi page that rngtest comes from rng-tools package on Debian
<plaes> montjoie: o/

2015-11-21

<montjoie> could someone with an orangepipc share me their u-boot-sunxi-with-spl.bin ? I cannot made one that works

2015-11-20

<montjoie> it is necessary to be sure it is working
<montjoie> fl_0: cat /dev/hwrng | rngtest
<fl_0> montjoie: how can i test ne newly built module?
<fl_0> montjoie: sun4i-ss module compiled without problems here
<fl_0> montjoie: thank you for your email
<montjoie> plaes: thanks
<plaes> montjoie: typo: otionnal
<montjoie> Hello I have updated the patch http://sunxi.montjoie.ovh/patchs_current/0001-crypto-sun4i-ss-support-the-Security-System-PRNG.patch if someone want to test it (wens ?:))

2015-11-19

<igraltist> montjoie: nothing
<montjoie> igraltist: what said dmesg ?
<igraltist> montjoie http://paste.debian.net/333974/\

2015-11-17

<montjoie> my major fear is the sha1 error
<wens> montjoie: and the a31 already has ahb_ss at 150, prng failing
<wens> montjoie: that requires some uboot modifications :|
<montjoie> could you test to put ahb_ss to 150 on any failing board ?
<montjoie> thanks
<montjoie> wens I dont remember how to see those numbers, could you give the patch to the debugfs ?
<montjoie> or perhaps underclocked ?:)
<montjoie> but perhaps 84 is too much
<montjoie> oups > 24
<montjoie> the datasheet require < 24
<montjoie> wens: and same for A31
<fl_0> montjoie: I can confirm that the driver loads perfectly now
<montjoie> I was thinking to a sort of clock probleme, but the requirement are the same between A31 and A20
<montjoie> logicay the prng is unrelated with those error
<montjoie> wens: does you have always those sha1 error on probe ?
<montjoie> but last time all was totally the same
<montjoie> I will recheck datasheet for hwrng
<montjoie> same error than A10, but strange you got also an error on sha1
<wens> montjoie: A33 doesn't look good ^
<fl_0> montjoie: good morning.
<montjoie> I will resentence the line on the site
<montjoie> in fact A10 is supported, but I got errors on it
<wens> montjoie: a33 maybe? your webpage says a10 is not supported?
<montjoie> igraltist: could you paste the fail ?
<montjoie> wens: thanks for testing, could you test it on non-A20 board ?
<montjoie> fl_0: the link given by Turl is good for solving this issue
<wens> montjoie: ^
<wens> montjoie: running rngtest on my cb2

2015-11-16

<fl_0> montjoie: ping?
<igraltist> montjoie: now the /dev/hwrng exit but test fails
<montjoie> thanks for testing
<montjoie> and then test it with cat /dev/hwrng | rngtest
<igraltist> montjoie: i have already this patched
<montjoie> the security system is in mainline only since 4.3, before that you need the four patch for adding it
<igraltist> montjoie: ok
<montjoie> igraltist: the patch is for 4.3+
<igraltist> montjoie: i use 4.1
<igraltist> montjoie: againts whisch kernel is this patch
<montjoie> cajg: yes It is the on-chip crypto
<cajg> Nice, montjoie, I hadn't seen you project.
<montjoie> hello I seek tester for the Security System PRNG (A10/A20/A31 etc...) http://sunxi.montjoie.ovh/patchs_current/0001-crypto-sun4i-ss-support-the-Security-System-PRNG.patch

2015-11-15

<montjoie> oh my god it is the first time I open the SDK, they have zipped allk .o files in kernel tree:)

2015-11-13

<mripard> montjoie: you should *really* put the version number of your patches.
<wens> montjoie: gmail warns that the sender may not be who it claims to be
<montjoie> wens: mripard do you have receive my sun4i-ss statesize patch today ? nothing show up in lkml but gmail said everythings ok

2015-11-03

<montjoie> thanks vishnup, no hurry, I just feared to miss them
<vishnup> montjoie: I promise, I'll post it within 24 hours
<vishnup> montjoie: not yet,

2015-11-02

<montjoie> hello does someone know if vishnup released his uboot patch for A83T ?

2015-10-25

<jelle> montjoie: obviously not
<montjoie> jelly: with a mainline kernel ?
<montjoie> and I seek people with H3/mainline
<montjoie> hello, anybody here with the devboard a83T ? it automaticly shutdown when I power up like hdmi/ethernet is necessary to continue booting

2015-09-25

<kenny> montjoie: either way, as long as the rich os can't get them. Your patches for SS made me remember it again, but it doesn't seem like much info exists on getting stuff running in there.
<montjoie> or perhaps you speak about secure ram
<montjoie> kenny: I want that also, but the storage is some efuse and I do not know the number of write possible on it

2015-09-22

<montjoie> but all my tries fail with DMA never ending, or worst, bad transfer
<montjoie> but I convinced that playing with DMA engine DDMA_PARA_REG register could boost something
<montjoie> probably like a "minimal len for dma"
<montjoie> with Kconfig or module_param I dont know
<montjoie> I agree, in final I will create an option for letting user to choose
<mripard> montjoie: yeah, but the CPU is busy doing other things during that time
<montjoie> for disk encryption, using DMA give speed/4
<montjoie> lossing 10 or 20% is ok for me, but 70%...
<montjoie> I think we can do better, but the documentation is lacking on some register
<montjoie> yes
<montjoie> WORSE than everything
<montjoie> <horror> I have updated my bench for AES 128 on http://sunxi.montjoie.ovh/ with DMA results for the security system</horror>

2015-09-16

<ricardocrudo> montjoie: write data and read back
<montjoie> for the moment no, I need to finish DMA support
<plaes> montjoie: btw, I have my A10 device now quite usable, if you still need A10 tests
<montjoie> what kind of use do you speek ?

2015-09-01

<mripard> montjoie_: DMA has never been a guarantee of performances
<montjoie_> very very bad
<montjoie_> bpi-user the DMA need to be used by driver, for example I am writing the support for it in crypto, but for the moment very bad performance

2015-08-27

<montjoie> plaes: the test end on first fail,so it do not check aes 192 and aes 256
<plaes> montjoie: and test with 1000: http://fpaste.org/260191/06930881/
<montjoie> perhaps you need to set the list of cipher for the engine in openssl.cnf
<montjoie> I see but I never tested the AF_ALG engine for openssl
<montjoie> plaes: ok thanks, it is confirming the bug
<plaes> montjoie: something like this: Decrypt error at request 0 pos=270448 of len=524288 key=16
<montjoie> plaes: updated
<montjoie> plaes: oups I believed to disable the build of af_alg tool, I will update the git
<plaes> montjoie: is it missing af_alg library?
<montjoie> and run cryptodev_test test aes 1000
<montjoie> a corruption bug that I try to solve, could you try https://github.com/montjoie/cryptotest
<plaes> montjoie: which bug?
<montjoie> hello I seek people who use the Security System driver, for testingif they hit the same bug than me.

2015-08-23

<montjoie> yes
<plaes> montjoie: should 'openssl speed -engine af_alg' work with sunxi-ss?
<plaes> montjoie: o/
<montjoie> but I note for later to do a scp bench
<montjoie> and the best performance will be available thougth cryptodev via the openssl cryptodev engine
<montjoie> lauri: on sunxi.montjoie.ovh you will have all details, but the hardware does not support arcfour

2015-08-07

<montjoie> just add the device
<montjoie> but nothing on my patch add reset control
<montjoie> sunxi
<montjoie> yes on suxi.montjoie.ovh (perhaps some are missing or old)
<wens> montjoie: do you have a31/a23/a33 patches (adding reset control to ss driver and dtsi bits) for crypto?

2015-08-04

<ganbold> montjoie: nice

2015-08-03

<montjoie> arg random failure with my test tool
<fullstop> montjoie: I completely understand. No worries!
<montjoie> Turl: I need to stabilize and then I bench, this week probably
<montjoie> fullstop: thanks but I prefer to not read allwinner source
<Turl> montjoie: nice :) how is performance?
<montjoie> yes
<montjoie> Good news The Security System DMA seems to work now

2015-07-30

<montjoie> wens: yes but I need to redone them with latest neon optimization
<wens> montjoie: did you do benchmarks comparing ss with arm-asm/neon implementations for aes/sha-1?

2015-07-29

<montjoie> ganbold: the status could be seen at http://sunxi.montjoie.ovh/
<montjoie> yes

2015-07-26

<montjoie> yes LOADADDR=0x40008000
<dodo_> montjoie: hmn... there is nothing about that in linux-sunxi's wiki (http://linux-sunxi.org/Mainline_Kernel_Howto)
<montjoie> dodo_: you need to provide it
<montjoie> Specify LOADADDR on the commandline to build an uImage

2015-07-25

<montjoie> you will need to wait for A8x driver
<montjoie> same for dsa
<montjoie> and no rsa bench since A20 SS does not have it
<ganbold> montjoie: yeah I saw some of those, I just wanted to see rsa2048, dsa2048 case
<montjoie> ganbold: all my bench are on http://sunxi.montjoie.ovh/ but I need to redone them. (they are 6 month old)
<ganbold> montjoie: ping, do you have some benchmark numbers related to openssl (with kernel with and without A20 security system patches)? I'm interested in "openssl speed rsa2048 dsa2048" output

2015-07-20

<montjoie> hello I still search people with A10s, A31, A13 and A33 for testing my crypto driver on them.

2015-06-07

<montjoie> wens: certainly the driver will be common with A80
<montjoie> yes wens
<wens> montjoie: you're working on SS for A83?

2015-06-01

<montjoie> hello, does someone work on A83T ?

2015-04-22

<Montjoie> or use af_alg
<Montjoie> so you have few choice, patch openssh or disable seccomp
<sehraf> Montjoie: ah, ok :)
<Montjoie> it is related to secomp
<Montjoie> bug referenced on my site:)
<sehraf> Montjoie: did you try the ss module + cryptodev + sshd (+ openssl for cryptodev) ? whenever i've loaded cryptodev and the ss module i can't establish a new ssh connection to my cubieboard
<Montjoie> I still seek people with A10 A10s A33 for testing the Security System driver on it? i have updated sunxi.montjoie.ovh with all necessary dt patch for thoses SoC

2015-04-21

<Montjoie> does someone use the A20 watchdog ? how do you have configured it ?

2015-04-17

<Montjoie> you need the Security System patch also
<Montjoie> it is linked in the page
<kurain> but I don't know how to contact you. ;( Montjoie
<Montjoie> kurain, hi
<kurain> Montjoie: hi
<kurain> montjoie[home]: really glad to see yor patch
<Montjoie> I send it a few time, but I doesnt want to spam it
<plaes> Montjoie: this is the first time I saw your link
<Montjoie> I seek people with A10 A10s A33 also for test
<Montjoie> kurain, is you want, I can send you a DT patch for adding support to A31
<Montjoie> All details could be found here http://sunxi.montjoie.ovh/
<Montjoie> and according to datasheet it is the same
<Montjoie> kurain, if you want I have already done PRNG for A20

2015-03-27

<Montjoie> Security System now support ECB, working on other cipher mode...

2015-03-18

<montjoie[home]> but for the moment my driver pass the rngtest tool
<montjoie[home]> for comparing methods to configure it
<montjoie[home]> MY123: if you can share your driver for the prng
<MY123> Montjoie, but RDRAND is worse :-)
<MY123> Montjoie, bad quality
<Montjoie> A20 PRNG is killing me

2015-03-17

<Montjoie> it seems to be the better solution
<atsampson> Montjoie: the alternative is to provide a way of interrupting the power input, but some electronics will still be required
<Montjoie> the problem with the link, is the need to weld
<Montjoie> I found http://linux-sunxi.org/Cubieboard/Reset but I expecting other way
<Montjoie> does someone know a way to hard reboot cubieboard via gpio or other (for implementing a hard reboot at distance)

2015-03-05

<Montjoie> with gentoo, no systemd:)

2015-02-19

<driv3l_> Montjoie - Thanks for the info. Any plans to support any other mode besides CBC? CTR?
<Montjoie> i need to finish the proper changelog
<Montjoie> driv3l, the major fact is to fix a possible crash with ipsec
<driv3l> Montjoie - You released an update to the SS patch (v6). What has changed?

2015-02-16

<T0mW> montjoie[home]: seems all the info I come up with is not related to the sunxi, or refers to the older samsung way.
<T0mW> montjoie[home]: googling for two days now...
<montjoie[home]> do like me, get the datasheet and search for an already done driver
<montjoie[home]> they certainly comes from datasheet

2015-02-12

<Turl> montjoie[home]: if you do setenforce 0 it's not
<montjoie[home]> selinux is never annoying!!!

2015-02-10

<montjoie[home]> thanks
<lioka> montjoie[home]: i'm using -machive virt with qemu, if that counts as something special
<lioka> montjoie[home]: sata
<montjoie[home]> i would says, which real storage, sata or nbd ?
<lioka> montjoie[home]: virtio-blk for storage
<lioka> montjoie[home]: good enough with kvm
<montjoie[home]> and performance ? what storage do you use ?
<lioka> montjoie[home]: with qemu ? nope
<montjoie[home]> lioka: do you have done something spetial to compil/use it ?
<lioka> Montjoie: i'm
<atsampson> Montjoie: there are certainly people who use Xen on the A20: http://wiki.xen.org/wiki/Xen_ARM_with_Virtualization_Extensions/Allwinner
<Montjoie> A20 theoricaly have extention for virt
<Montjoie> does some people use qemu on their cubieboard ?

2015-02-02

<premoboss> Montjoie, olny if the entropy pool is full
<Montjoie> I do not know the best way for getting a good seed, perhaps /dev/random is good
<premoboss> Montjoie, ok. maybe i can read ADC to have a random seed
<Montjoie> a PRNG need a seed to generate random number, so the random is not so random
<premoboss> Montjoie, can you give me more details? i dont know wht PRNG is
<Montjoie> premoboss, the A20 have a PRNG but no real random generator

2015-01-28

<interrobang__> but Montjoie has no patch given :/
<ssvb> interrobang__: ask montjoie[home] about the current status of the hardware cryptographic accelerator in the mainline kernel
<interrobang__> Montjoie, montjoie[home], hey!;-) -- You can find the latest patch here (TODO) <- when?
<interrobang__> montjoie[home], sorry, how can i get the patch? dont find any repo or download link https://lkml.org/lkml/2014/10/19/61
<interrobang__> montjoie[home], which kernel do you use on cubieboard?
<montjoie[home]> the patch for 3.4 is old and have bug
<montjoie[home]> sunxi-ss is not supported anymore on 3.4 anyway
<montjoie[home]> interrobang__: try mainline kernel
<interrobang__> Montjoie, sorry
<Turl> Montjoie: yeah :)
<Montjoie> what about hw monitoring ?

2014-12-31

<montjoie[home]> yes with sslciphersuite
<montjoie[home]> the cipher is for each client/server session
<montjoie[home]> you dont need to regen a cetificate
<montjoie[home]> this is two separate problem
<montjoie[home]> you need to use a cipher based on aes cbc
<montjoie[home]> hardware
<mossroy> montjoie[home] : thanks for your answer
<montjoie[home]> I will update my site with more example for user space use
<montjoie[home]> mossroy: your cipher is aes gcm so not one supported
<mossroy> I'm trying to use the Security System (hardware crypto acceleration) on a Olinuxino A20 Micro, with the patch from montjoie

2014-12-04

<Montjoie> yes 3.17 for example
<speakman> Montjoie: latest like in latest official kernel?
<Montjoie> speakman, premoboard latest linux kernel are pretty stable also

2014-11-24

<montjoie[home]> and no sata:(

2014-10-27

<Montjoie> with writesl I got some more performance improvment, I am happy
<Montjoie> Yes, I need to conver french decimal
<wens> Montjoie: comma in the numbers should be a decimal point?
<Montjoie> For people wanting to have numbers for Security System bench, http://sunxi.montjoie.ovh

2014-10-24

<montjoie[home]> yeah up to 10% speed improvment with use of writesl() in the crypto accelerator

2014-10-21

<Montjoie> drivers/net/ethernet/allwinner/sun4i-emac.c
<Montjoie> gmac/emac

2014-10-19

<montjoie[home]> wens: thanks I doesnt know that it exists that
<montjoie[home]> I still seek people with A10 A80 A13 A10s A33 A31 A23 for testing the crypto accelerator
<montjoie[home]> yeah I took finaly the time to send the patch

2014-10-17

<Montjoie> you can test right now, pick the patch and add it to your dtb
<Montjoie> according to the manual
<Montjoie> I dev on A20, A10 normally have the same
<linkmauve1> Montjoie, on which platform is that? Is that crypto hardware present on A10?
<Montjoie> for sha1 up to 50%, md5 is up to 15%
<Montjoie> yes that is the other reason:)
<Montjoie> I will work on dma when the basic driver will be accepted
<Montjoie> no dma
<Montjoie> and I have just sent the last task of the eudyptula challenge so more freetime tofinish the patch