lanefu changed the topic of #armbian-rockchip to: Armbian - Linux for ARM development boards | Rockchip SoC | www.armbian.com | This channel is relayed to the equivalent Discord channel | this channel is logged
DC-IRC has quit [Remote host closed the connection]
DC-IRC has joined #armbian-rockchip
<DC-IRC> <terrorpup> no console, keeps reverting to Chinese
<DC-IRC> <terrorpup> I for the last 20 years I have done Linux/Unix as a job, expect now I am more management ( WHICH I HATE )
<DC-IRC> <terrorpup> I love working on the boards because they are keeping me fresh
<DC-IRC> <terrorpup> Day in, Day Out, I work RH ( Expect in my Home, which is openSUSE, Armbian, Debian, FreeBSD, and one or two Ubuntu boxes, which drive me up the wall, if you are going to use debian, just use debian, that my saying. )
<DC-IRC> <terrorpup> Anyway guys @nicod_sbc / @tenkawa42 thank you, my son is proofing my blog for the BPi-P2 Pro, tomorrow I will do a setup video. I am a lot like @nicod_sbc I am not sure what to run it on. It's cute, but, file system and locale for me are issue. I got Wifi to work.
<DC-IRC> <terrorpup> Nite Guys
<montjoie> Hello I seek people with an rk3568 willing to help testing the crypto driver on it
<DC-IRC> <rpardini> montjoie: nice, I've a 3568 and a 3566 (dunno if it applies as well) somewhat available for testing. can we get rng? 😄 also in u-boot? (wanna kaslr)
<montjoie> rpardini no rng yet:( (but this is planned)
<DC-IRC> <rpardini> okie, and I suppose `CRYPTO_DEV_ROCKCHIP2=y` and `CRYPTO_DEV_ROCKCHIP2_DEBUG=y`
<montjoie> yes
<DC-IRC> <rpardini> ok will try and report back
<DC-IRC> <rpardini> montjoie: your changes have cherry-picked clean on 6.5.5 -- do you need 6.6-rc4 and/or linux-next as base?
<montjoie> rpardini: I need only recent changes on crypto_engine, so probably <=6.4 wont work
<DC-IRC> <rpardini> I'll give it a try on v6.5.5 first, since that's what we have setup for rockchip64-edge
<montjoie> thanks
<DC-IRC> <rpardini> no game on 6.5.y, need 6.6-rc1 at least for Mr. Xu's crypto_engine skcipher_engine_alg
<DC-IRC> <rpardini> I'll take a small detour for rebaseing
<DC-IRC> <tenkawa42> @terrorpup figure out anything on the language settings?
<DC-IRC> <tenkawa42> the locales are what I told you but your console output is handled somewhere else..
<DC-IRC> <tenkawa42> everything in locales is handled by those variables
<DC-IRC> <terrorpup> No, I will send you a DM, I spent the rest of the night writing a blog
<DC-IRC> <terrorpup> I do want to share with other, but I will wait and give if asked.
<DC-IRC> <rpardini> montjoie: rk2 crypto probes OK on 3568. I installed cryptodev and ran their hashcrypt_speed test. rk2-sha1 gives around 59mb/s with 64k blocks.
<DC-IRC> <rpardini> /sys/kernel/debug/rk3588_crypto/stats says 'rk3588-crypto fe380000.crypto requests: 668319' but is otherwise all zeros.
<DC-IRC> <rpardini> similar results on rk3566 too.
<DC-IRC> <rpardini> let me know if theres something else I should test for...
<DC-IRC> <rpardini> I'll be pushing 6.6-rc4 to some of Armbian's `-edge` kernels soon and will include your patches so more folks can test too.
<montjoie> rpardini could you pastebin the whole /sys/kernel/debug/rk3588_crypto/stats ?
<DC-IRC> <rpardini> montjoie: here, 3568 https://paste.next.armbian.com/ahigisebup and for 3566 https://paste.next.armbian.com/timederehu
<DC-IRC> <rpardini> also a dmesg + crypto stats on 3568 for completeness https://paste.next.armbian.com/uwatabebaq
<montjoie> rpardini having 0 in flalback is good, it means the HW really handle it
<montjoie> do you want I add an tested-by on my patch serie (or I can CC you and you will send it on mailling list)
<montjoie> probably I need to change the DT patch since I can move crypto node on rk356x
<DC-IRC> <rpardini> sure. thanks for your work!!
<DC-IRC> <rpardini> ```
<DC-IRC> <rpardini> Tested-by: Ricardo Pardini <ricardo@pardini.net>
<DC-IRC> <rpardini> ```
<montjoie> instead of both on rk3566 AND rk3568
<DC-IRC> <rpardini> oh -- yeah it looks like the same.
<DC-IRC> <tenkawa42> montjoie: will this work on all 3568's or only certain ones?
<montjoie> I wanted test to be sure
<montjoie> yes on all 3568
<DC-IRC> <tenkawa42> Excellent
<DC-IRC> <rpardini> I suppose mainline-enabled 3588 folks will wanna test it as well
<montjoie> this is on SoC, so not board related
<DC-IRC> <tenkawa42> montjoie: One thing I've learned with Rockchip/crypto in general... never assume anything
<DC-IRC> <tenkawa42> I have RK3588s patches that are not SoC independent
<DC-IRC> <tenkawa42> they won't work on Khadas
<DC-IRC> <tenkawa42> (They apply just fine but the logic itself is incorrect)
<DC-IRC> <tenkawa42> @rpardini speaking of Khadas... who/is anyone maintaining the Edge2 for Armbian?
<montjoie> tenkawa42, I know, like TRNG produces random number but vendor likes to break this ah ah ah ah
<DC-IRC> <tenkawa42> omg yes
<DC-IRC> <rpardini> @tenkawa42 kedge2 is only on vendor kernel AFAIK
<DC-IRC> <tenkawa42> @rpardini yeah was afraid you were going to say that
<DC-IRC> <tenkawa42> that thing is a mess
<DC-IRC> <tenkawa42> For such a great potential in hardware.... terrible kernel...
<DC-IRC> <rpardini> mainline makes no sense right now for kedge2, board has no eth, no pci; and usb3 is not working. so you got an uart-enabled paperweight on mainline I think. vendor should work reasonably well.
<DC-IRC> <tenkawa42> LOL
<DC-IRC> <tenkawa42> Not sure its big&heavy enough to be a paperweight
<DC-IRC> <rpardini> I'm still waiting for a kedge2 and that uart-board-thingy so I can fix the u-boot. For now the oowow images are counting on oowow's (broken) u-boot in SPI from factory. which is bad.
<DC-IRC> <tenkawa42> I think I have heavier coasters
<DC-IRC> <tenkawa42> Yeah for some reason the uart speed/flow seems wrong
<DC-IRC> <tenkawa42> it outputs.. but its all garbage at all speeds/settings
<DC-IRC> <rpardini> have you set your ttl adapter to 3.3v level? someone else had the same symptom when set to 5v level. (not talking vcc here)
<DC-IRC> <tenkawa42> Mine only has 3.3v
<DC-IRC> <tenkawa42> its a adafruit one made for the rpi
<DC-IRC> <tenkawa42> I want to say it worked at one time but a firmware/update of some sort scrambled it
<DC-IRC> <tenkawa42> (on the Edge2)
<DC-IRC> <rpardini> yeah I dunno. Igor swears it works for him. You and Nico swear it only spits out garbage.
<DC-IRC> <rpardini> I'll decide if/when I get one
<DC-IRC> <tenkawa42> I do see some notes in Khadas docs though that make me wonder about the dts too
<DC-IRC> <tenkawa42> It talks about the uart and gpio having to be "chosen"
<DC-IRC> <tenkawa42> only one or other
<DC-IRC> <tenkawa42> which seems odd
<DC-IRC> <tenkawa42> Ah this is if you want to remove it completely
<DC-IRC> <rpardini> well we've been through this. you should get blXX messages in 1.5mbaud no matter what bootloader or kernel or dts you have.
<DC-IRC> <tenkawa42> Yeah I get those
<DC-IRC> <tenkawa42> It goes crazy once the kernel loads
<DC-IRC> <tenkawa42> That sounds like cmdline doesn't it?
<montjoie> rpardini I will update my branch with proper dt patch probably tomorow, if you can delay pushing to armbian
<DC-IRC> <rpardini> montjoie: no worries. it's in a batch with my 6.6-rc4 bumps, so if needs be I'll update it by the weekend.
<montjoie> since rk3566 is tested, I could merge some rk356x patch
nekomancer has joined #armbian-rockchip
nekomancer has quit [Client Quit]