<narmstrong>
marc|gonzalez: yes it's expected with port/endpoint architecture
Daanct12 has quit [Quit: WeeChat 3.8]
<marc|gonzalez>
OK cool
<marc|gonzalez>
I've finished writing a script to generate a functional kernel image for the board
<marc|gonzalez>
narmstrong: do you want me to send something to fixup patch 3/3 of the pmu series?
<narmstrong>
marc|gonzalez: nop it was ok & I sent the fixes PR
<narmstrong>
HackerKkillinghi: maybe, I have a lot of difficulty understanding the difference between the 6, 8, & 8 revisions, usually we add a "0/3" cover letter where we can explain the goal and add a changelog
<marc|gonzalez>
narmstrong: even the typoed commit message in patch 3? :)
<narmstrong>
marc|gonzalez: I don't see any tupo
<narmstrong>
the commit title is good
<marc|gonzalez>
it's in the commit message :(
<marc|gonzalez>
Commit "perf/amlogic: resolve conflict between canvas & pmu" changed the base address. => wrong prefix because I had the wrong prefix for patch 2
<marc|gonzalez>
(not a deal breaker in the grand scheme of things)
<narmstrong>
marc|gonzalez: oh I see now, too late it's in the pipe
<marc|gonzalez>
I wasn't sure I had been clear yesterday. Turns out not ;)
<phh>
narmstrong: Hello, I tried migrating h264 (legacy vdec) out of esparser, I have, uh, results. I didn't dig very deep yet, but: one thing to work-around is that the vdec will read only 256B reads. vendor just do write size += 1024. I tried that and it doesn't work, except if I put more data in one read. So I'm thinking that it's just that the decoder requires to have SPS/PPS/... in one read. Does that make sense? How does v4l account for SPS/PPS/... NALs?
<phh>
Does it say something on how many NALs should be in one buffer?
naoki has quit [Quit: naoki]
<HackerKkillinghi>
narmstrong srroy for the difficulty
<HackerKkillinghi>
Yeah it is a bit messly
<HackerKkillinghi>
U can basiclly can ingnore
<HackerKkillinghi>
V6,7
<HackerKkillinghi>
V6 is as same as v5 bc i forgot to fix the issue when i sented the v6
<HackerKkillinghi>
V8 is basually as same as v7 . The reson i bump the version is bc i misstakly cut off the end of the patch when i resent the patch.The reson i resent yhe patch bc i send it with wrong email adress.
<HackerKkillinghi>
*The reson of why i resent the patch bc i send it with wrong email adress.
<narmstrong>
Ok, for next time try to use a cover letter and add those details in it
<HackerKkillinghi>
can you give me an example of cover letter
<HackerKkillinghi>
btw i saw this kind of lattle but i didnt know what it for
<HackerKkillinghi>
*kind of before
f_ has joined #linux-amlogic
f_[xmpp] has quit [Ping timeout: 265 seconds]
<HackerKkillinghi>
*this kind of lattle before
<f_>
?
<f_>
I see you're upstreaming your patches. Great!
<f_>
=)
<f_>
I should probably make the latest version of U-Boot work on the KII Pro so that then I can send a patch to U-Boot =)
<f_>
o/ chewitt
<HackerKkillinghi>
hey
<HackerKkillinghi>
F_ have you try to replace your stock bootloader of the kii pto
<HackerKkillinghi>
*pro
<leadf00t>
narmstrong: thank you! So there is hope. We picked up this development board that appears to have a MIPI connector: https://www.aliexpress.us/item/3256803453407727.html If the processor works well, we will be using it for our race car dashboard product.
<leadf00t>
narmstrong: thank you! So there is hope. We picked up this development board that appears to have a MIPI connector: https://www.aliexpress.us/item/3256803453407727.html If the processor works well, we will be using it for our race car dashboard product.
<f_>
I first need to know how I can recover it and where do I flash U-Boot.
<f_>
And I also need to properly port U-Boot.
<f_>
I'll try with the WeTek Hub's defconfig first.
<HackerKkillinghi>
No
<HackerKkillinghi>
Try p212
<f_>
Why?
<HackerKkillinghi>
, f_
<f_>
It uses a meson-gxl SoC, it's never going to work.
<HackerKkillinghi>
simple tv box from china are uselly are base on amlogic pXX
<f_>
The WeTek Hub uses the meson-gxbb SoC, which happens to be the same SoC used by the KII Pro.
<f_>
HackerKkillinghi: Yes I know.
<HackerKkillinghi>
Yeah
<HackerKkillinghi>
so WeTek Hub defconfig might also work
<f_>
I might also try with the WeTek Core 2 defconfig as well.
buzzmarshall has joined #linux-amlogic
marc|gonzalez has quit [Quit: Leaving.]
<f_>
s/Core/Play/
<HackerKkillinghi>
Check dm
<HackerKkillinghi>
f_
tomeu has quit [Quit: You have been kicked for being idle]
<f_>
Alright.
<f_>
I'll go back to work.
jacobk has quit [Quit: No Ping reply in 180 seconds.]
jacobk has joined #linux-amlogic
jacobk has quit [Ping timeout: 248 seconds]
<chewitt>
f_: the Hub/Play2 defconfigs are identical except for the board name and dtb
<chewitt>
but unless you have the original vendor u-boot sources to extract the FIP sources, this is a dead end
<chewitt>
or you need to have vendor Android ROM images that can be reflashed to emmc if (when) you break boot with the wrong FIP sources
<chewitt>
the FIP sources used for signing contain acs.bin which is normally tweaked by the box vendor to suit the particular RAM chips in that specific box
<chewitt>
if you don't have the original (correct) FIPs used with the box, you won't be able to sign u-boot correctly
<chewitt>
sign = build working image
<f_>
chewitt: I didn't touch the eMMC in any way so.
<chewitt>
if you don't have them (and you prob. don't) you can experiment with other gxbb FIP sources
<f_>
I should have the vendor U-Boot flashed somewhere.
<chewitt>
but if they are wrong for the box, it will not boot
<f_>
Along with FIPs.
<f_>
chewitt: Where are those supposed to be?
<chewitt>
and you will need to reflash Android to recover the box
<f_>
I have the vendor firmware on the eMMC. It stayed untouched while I was porting postmarketOS.
<f_>
So I still have the shady AOSP-based OS, and I still have the vendor U-Boot.
<f_>
And so I should be able to extract the FIPs.
<chewitt>
then you are fine, but you cannot update to upstream u-boot (which is what you said you'd do)
<chewitt>
as that requires you to build u-boot (which is easy) and then combine with with FIP sources in a signing recipe; to boot the box
<chewitt>
and if you don't have the FIP sources for the box (and you don't, you only have the compiled/signed output from them) you can't do that
<chewitt>
if the RAM chips in the box are the same specs as something we have FIP sources for; you may be able to borrow files from another board
<f_>
chewitt: No I meant..
<chewitt>
e.g. WeTek Core2 can boot using the Khadas VIM2 FIP sources
<chewitt>
but mostly.. this isn't likely to work, and when it doesn't work you need to reflash Android via Amlogic Burning Tool
<f_>
I meant replacing v2020.07 U-Boot with the latest version on the SD card....I was talking about a chainloaded U-Boot.
<chewitt>
ahh, okay
<f_>
But still. Good to know =)
<chewitt>
chainloading is not required and doesn't bring any benefits
<chewitt>
if anything, it probably makes things worse
<f_>
It does make ethernet work (for some weird reason?)
<chewitt>
sounds more like a broken/wrong device-tree file then
<chewitt>
the only reason LE supported chainload in the past, was working around a display issue and RAM size detection issue with some Beelink boxes
<chewitt>
once that was fixed in the kernel (and we used the right FIP sources) .. both issues were resolve and chainload was no longer required
<f_>
Sure.
<f_>
Honestly it would be ebtter to just flash a mainline U-Boot instead of the vendor one.
<f_>
*better
<f_>
If I can do that..
<HackerKkillinghi>
Chewitt
<chewitt>
f_: but you can't .. re-read my last few comments
<chewitt>
you are stuck with vendor u-boot *unless* you have the FIP sources for that specific box
<HackerKkillinghi>
You are worg The only benefits of chainloading is when you are dealing with locked bootloader.the bootm command is locked to boot verify boot image.
<HackerKkillinghi>
*locked to only boot verify boot image on xiaomi-once
<HackerKkillinghi>
we could use the fip from other box, right
<HackerKkillinghi>
,Chewitt
<phh>
if you have another box with the same signing...?
<HackerKkillinghi>
Tbh i dont think those cheap china box are signed in bl1
<f_>
< chewitt> if the RAM chips in the box are the same specs as something we have FIP sources for; you may be able to borrow files from another board
<f_>
I'm taking a break from Amlogic for now...Working on some projects :)
<f_>
Cya!
<HackerKkillinghi>
You can try that
<chewitt>
BL1 isn't the issue
f_ has quit [Quit: Lost terminal]
<HackerKkillinghi>
Isnt the bl1 verifty the bl2
<chewitt>
BL2 includes acs.bin which is modified by board vendors to suit the low-bin/cheap/reclaimed RAM chips used in boxes
<chewitt>
wrong timing data = no boot
<chewitt>
BL1 checks for a magic signing header at the right offset, and if present loads BL2
<chewitt>
It can also do checksumming of things (or similar) to lock boot, but that's rarely done
<HackerKkillinghi>
i think my xiaomi-once
<chewitt>
is one of the few
<HackerKkillinghi>
has did those thing
<chewitt>
the dreambox DVB boxes also have locked boot
<HackerKkillinghi>
maybe not
<chewitt>
one of the things I'd like to do for the FIP repo is collect data on the RAM chips used so people can see if they match
<HackerKkillinghi>
i think all google certified tv box did that
jacobk has joined #linux-amlogic
<HackerKkillinghi>
so xiaomi- once is not the few
<HackerKkillinghi>
maybe the china version of xiaomi- once dont have locked bootloader
<HackerKkillinghi>
let me tell you one thing about google certified tv box : it seem google is locking down those box chewitt
<HackerKkillinghi>
in older version of some google certified tv's stock firmw , the uboot can intercepted when it is booting which given a shell for running command.
<HackerKkillinghi>
but after some update the uboot cant intercepted when it is booting.
<HackerKkillinghi>
This seem to be happented on some google certified tv's box
<phh>
can you go into fastboot or "amlogic update" usb gadget mode?
<HackerKkillinghi>
yes
<HackerKkillinghi>
But manually run the command fastbot in uboot
<HackerKkillinghi>
(Android 8's uboot)
<phh>
then you can do `fastboot oem "any uboot command you like"` or `update bulkcmd "a uboot command"`
<HackerKkillinghi>
btw i tried on another device
<HackerKkillinghi>
the fastboot way didnt work
ldevulder has quit [Quit: Leaving]
<HackerKkillinghi>
Also one thing , xiaomi-once 's android 9 's uboot has to started down