narmstrong changed the topic of #linux-amlogic to: Amlogic mainline kernel development discussion - our wiki http://linux-meson.com/ - ml linux-amlogic@lists.infradead.org - official channel moved from Freenode - publicly logged on https://libera.irclog.whitequark.org/linux-amlogic
vagrantc has quit [Quit: leaving]
jacobk has quit [Ping timeout: 248 seconds]
chewitt_ has joined #linux-amlogic
jacobk has joined #linux-amlogic
chewitt has quit [Ping timeout: 255 seconds]
jacobk has quit [Ping timeout: 260 seconds]
chewitt_ is now known as chewitt
jacobk has joined #linux-amlogic
hexdump0815 has quit [Ping timeout: 245 seconds]
hexdump0815 has joined #linux-amlogic
jacobk has quit [Ping timeout: 272 seconds]
JohnnyonFlame has joined #linux-amlogic
naoki has joined #linux-amlogic
jacobk has joined #linux-amlogic
camus has quit [Ping timeout: 255 seconds]
camus has joined #linux-amlogic
jacobk has quit [Ping timeout: 272 seconds]
jacobk has joined #linux-amlogic
naoki has quit [Quit: naoki]
ldevulder has joined #linux-amlogic
JohnnyonFlame has quit [Read error: Connection reset by peer]
mripard has joined #linux-amlogic
mripard has quit [Quit: mripard]
mripard has joined #linux-amlogic
ldevulder has quit [Quit: Leaving]
<f_> hmm
<f_> cottonwood uses DDR4 IIRC while bpi-cm4 uses lpddr4
<f_> minute: will the io board include an SD card slot or USB to communicate with the bootrom?
<f_> ah wait
<f_> alta also uses lpddr4
<f_> yeah I guess 2 different A311D boards is better than one :)
<minute> f_: both are available
<f_> Is it a custom io board or bpi?
<minute> f_: bpi
<f_> ok
<minute> we have our own but for your case i think it's too much added complexity, and you can't access usb as a gadget
<f_> hmm looks like it's around the same size as my gxbb board
<f_> minute: Not like accessing USB as a gadget is very useful
<f_> I usually just use it to write to eMMC, but on my set-top box I never used it.
<f_> and just wrote to an SD card, forced it to boot to it by erasing the eMMC
<f_> But yeah honestly USB gadget is just convienient..not something necessary
<f_> Unless, of course, I'd try getting SPL DFU to work
<f_> anyway, either io board is fine, as long as I can at least boot from SD.
repk_ has quit [Ping timeout: 240 seconds]
<f_> minute: so yeah sure
<f_> Shall I send my address via email?
<f_> Am tempted to extract the KII Pro bl2.bin from my eMMC backup
<f_> I'm convinced that bl2.bin is very out-of-date and very old.
<f_> well..not really out-of-date
repk_ has joined #linux-amlogic
<f_> yeah very old..Built : 13:55:05, Jan 28 2016
<f_> Very early acs too.
<f_> ddrs, ddrt, and plls struct 'version' is 1
<f_> that's how old and early it is.
<f_> I bet it's as old as ODROID-C2 BL2
<f_> and just by looking at the hex dump BL30 and BL301 are separate it seems.
<f_> would be nice to <del>sue videostrong</del> ask them kindly to share U-Boot sources
<f_> but I doubt they'll be useful for anything..
psydroid has joined #linux-amlogic
<minute> f_: yes please, lukas@mntre.com
<chewitt> f_ most of the odm shops seem to have deleted their s905 bsp sources
<chewitt> ultimately some dev has them locally on a server/workstation box and each "dump" of sources is around 80-90GB in size
<chewitt> each device they work on reinvents the wheel
<narmstrong> I think amlogic tries to erase those old chips from internet
<narmstrong> but it's not how it works :-p some random hackers will keep them alive forever
<chewitt> i.e. download fresh bsp dump from Amlogic > hack it until the board works > ship the project > keep sources until support contract period expires > delete to free up disk
<f_> narmstrong: some crazy hackers will keep them alive
<f_> chewitt: repeat
<chewitt> most of the time they don't bother deleting until the disk is full, then delete the oldest thing
<f_> I doubt manufacturers try to reinvent the hardware either. I'd expect they base their box on a reference design, add dvb and casing, ship it
<chewitt> despite the availability of really big disks these days most s905 things already departed this world :)
<f_> perhaps also add some more stuff
<f_> chewitt: This is where IA would be handy.
<chewitt> Amlogic themselves designed these things to be copy/paste-able
<f_> Someone must've archived the code in IA, just need to find it..
<phh> that's largely why they are succesful
<f_> chewitt: amlogic themselves copy-paste stuff after each soc release
<chewitt> as long as you don't deviate too far from the original design the bsp "works" and you don't expend much effort on software
<f_> ^ every single vendor bsp
<chewitt> and that's why we so do love bsp code :)
<chewitt> the same mistakes repeated ad infinitum
<f_> :)
<f_> minute: will do that, thanks.
<chewitt> narmstrong and myself both have the bpi cm4 and its io board if helps with testing in the future
buzzmarshall has joined #linux-amlogic
jacobk has quit [Ping timeout: 272 seconds]
jacobk has joined #linux-amlogic
<f_> minute: sent
<f_> about sdks, IA has an archive of the openlinux.amlogic.org website
<f_> the old one
<chewitt> that's probably too old
<f_> yeah circa 2013
<chewitt> and IIRC the stuff available is only the kernel, not the full sdk
<f_> Yeah only the kernel
<f_> Haven't used it much tbh
<f_> so I don't quite know what they supply.
<f_> but moving to a wiki and a protected git repo was a mistake.
<chewitt> ^ thats the first s905 release, and it's been moved behind a paywall
<chewitt> git ssh access requires a key that you've been issued
<f_> Indeed
<chewitt> from conversations I've had with senior people .. lots of their customerbase wants to create drm encumbered product
<chewitt> and obscurity via closed sources is seen as a big benefit
<f_> Of course
<chewitt> (as part of the product security plan)
<f_> Set-top boxes are full of DRM junk.
<chewitt> the Amlogic folks I've spoken to are generally okay with or at worst indifferent to sharing code .. but their customers are firmly in the 'no' camp
<chewitt> and customers is what pays salaries
<f_> Except for a handful of SBC manufacturers.
<chewitt> and end-users are not customers, so our wishes aren't really factored into the conversation :)
<f_> lol
jacobk has quit [Ping timeout: 260 seconds]
<f_> chewitt: I guess marketing also says no
<chewitt> marketing is responsible for advertising what customers want to hear
<chewitt> sales is responsible for saying it and agreeing to it
<chewitt> engineering largely does what its told
jacobk has joined #linux-amlogic
<f_> yup.
jacobk has quit [Ping timeout: 260 seconds]
vagrantc has joined #linux-amlogic
Danct12 has quit [Quit: A-lined: This user has been AViVA-lined!]
chewitt has quit [Quit: Zzz..]
Danct12 has joined #linux-amlogic
Danct12 has quit [Client Quit]
Danct12 has joined #linux-amlogic
Danct12 has quit [Read error: Connection reset by peer]
Danct12 has joined #linux-amlogic
vagrantc has quit [Quit: leaving]
hexdump0815 has quit [Quit: WeeChat 1.9.1]
hexdump0815 has joined #linux-amlogic
jacobk has joined #linux-amlogic
jacobk has quit [Ping timeout: 248 seconds]
<vkoskiv> Hey all, I'm trying to play around with hw h264 decoding on the A311D & meson_vdec. ffplay seems to detect and talks to /dev/video0 to set up playback, but the current version on debian segfaults in v4l2_dequeue_v4l2buf. I compiled from the latest HEAD, and that one instead repeats lines like this a bunch:
<vkoskiv> [h264_v4l2m2m @ 0xffff74074ef0] output POLLERR6KB sq= 0B f=0/0
vagrantc has joined #linux-amlogic
<vkoskiv> It's entirely possible, and even likely that I'm just misunderstanding something or using it wrong somehow.
<vkoskiv> (And I'm aware meson_vdec is in staging, I'm not expecting perfection here yet)
<vkoskiv> Invocation currently is: ./ffplay_g -vcodec h264_v4l2m2m ~/bbb_sunflower_1080p_60fps_normal.mp4
<vkoskiv> I was using mpv before, that seemed to behave a bit more logically, I saw in the strace that it set the hardware decoder up in a reasonable manner, but then after about a dozen VIDIOC_QUERYBUF ioctls, it quits.
<vkoskiv> I think that too crashes in libavcodec, same function.
<vkoskiv> My linux video knowledge isn't very broad, so all I know is that the decoder is there, responds to ioctls in a reasonable way, but then something goes boom later for some reason.
<vkoskiv> Oh, and I've monitored the 'vdec' row in /proc/interrupts to try to see if it's actually using the hardware decoder, and none of the counts have increased at any point. Not sure if that's relevant here, though.
<vkoskiv> And I'm running linux 6.5.0-2-reform2-arm64 here.
<f_> Is that MNT Reform?
<f_> A311D Reform?
<vkoskiv> Yep!
<vkoskiv> With the fairly new RCM4 adapter board
<f_> :D one more A311D Reform user
<f_> tbh I never tried using vdec on any amlogic soc..
<narmstrong> The h264 decoder is broken on a311d, you may have more chances decoding a vp9 stream
hexdump0815 has quit [Ping timeout: 240 seconds]
<vkoskiv> Ah, I'll give that a go
hexdump0815 has joined #linux-amlogic
<f_> narmstrong: how well does vdec run on gxbb/gxl?
<vkoskiv> narmstrong: I guess I can learn about the state of h264 on the mailing list, right?
vagrantc has quit [Quit: leaving]
psydroid has quit [Quit: KVIrc 5.0.0 Aria http://www.kvirc.net/]
<vkoskiv> narmstrong: I get the same output with VP9 as with the h.264 video. At least in strace it looks mostly similar, and I get the same segfault
<vkoskiv> partial strace from mpv: https://pastebin.com/bvSkpRdW
kbingham has quit [Remote host closed the connection]
kbingham has joined #linux-amlogic
jacobk has joined #linux-amlogic
vagrantc has joined #linux-amlogic
kbingham has quit [Quit: ZNC - http://znc.in]
kbingham has joined #linux-amlogic