03:23
hexdump0815 has quit [Ping timeout: 258 seconds]
03:25
hexdump0815 has joined #linux-amlogic
04:09
buzzmarshall has quit [Quit: Konversation terminated!]
04:33
camus has joined #linux-amlogic
04:38
camus has quit [Ping timeout: 260 seconds]
04:41
camus has joined #linux-amlogic
05:34
JohnnyonFlame has quit [Read error: Connection reset by peer]
07:19
ungeskriptet has joined #linux-amlogic
10:05
camus has quit [Ping timeout: 272 seconds]
10:36
f_[xmpp] has joined #linux-amlogic
10:40
f_[xmpp] has quit [Ping timeout: 240 seconds]
10:51
psydroid has joined #linux-amlogic
11:01
psydroid2 has joined #linux-amlogic
12:12
<
f_ >
Am comparing ghidra's decomp with my code.
12:12
<
f_ >
I already see some missing stuff..
12:12
<
f_ >
We were so close.
12:34
jacobk has joined #linux-amlogic
12:54
f_ has quit [Ping timeout: 255 seconds]
12:56
f_ has joined #linux-amlogic
12:59
<
f_ >
Tbh gxl DRAM init overall is a little similar to gxbb
13:03
f_ has left #linux-amlogic [#linux-amlogic]
13:03
f_ has joined #linux-amlogic
13:10
jacobk has quit [Ping timeout: 260 seconds]
13:14
<
f_ >
I see: ddrt_p->cfg_ddr_tccdl
13:14
<
f_ >
tccdl isn't present on gxbb IIRC
13:21
<
f_ >
ah seems to be only used for ddr4
13:42
jacobk has joined #linux-amlogic
14:15
<
f_ >
Seems like it still hangs at trying to read
14:15
<
f_ >
Let's just avoid reading from DRAM lol
14:16
<
f_ >
and call it done
14:16
* f_
back to comparing code
14:41
camus has joined #linux-amlogic
15:54
jacobk has quit [Ping timeout: 248 seconds]
16:00
<
f_ >
OK I first need to push ghidra files..
16:21
<
f_ >
Oh while at it let's also push some early decomp of aml_ddr.fw and SM1/G12B bl2.bin
16:25
<
f_ >
git branch: master, thelinuxmacbook:fip-unmod » diff bananapi-cm4io/aml_ddr.fw ../u-boot_librecomputer/fip/g12b/aml_ddr.fw
16:25
<
f_ >
Binary files bananapi-cm4io/aml_ddr.fw and ../u-boot_librecomputer/fip/g12b/aml_ddr.fw differ
16:26
<
f_ >
and since I don't have the cottonwood boards yet to test if they can both be used on solitude or alta..
16:26
<
f_ >
I'll upload librecomputer g12b aml_ddr.fw just to be safe.
16:27
<
f_ >
just need to document how to extract code from that file..thanks xdarklight BTW.
16:56
<
f_ >
As soon as I get the cottonwood boards I'll dump the bootrom for sure.
16:58
<
f_ >
And at least I won't have to deal with AArch32 BL2 nonsense.
17:03
<
f_ >
aml_ddr.fw seems like ddr firmware but it's just BL21 but for DRAM init
17:04
<
f_ >
(for those curious who didn't read logs ^^)
17:04
<
f_ >
I still think they have it because of size limits hardcoded in BL1.
17:04
<
f_ >
(they manually load aml_ddr.fw from BL2 it seems, too.)
17:05
<
f_ >
So reversing BL2 is a lot easier. Just figure out PLL init and copy-paste BL21 power init code, you're pretty much done.
17:06
<
f_ >
Which is why I probably won't be reversing BL2 on anything G12.
17:06
<
narmstrong >
Nice so bl2 is quite empty now
17:06
<
f_ >
narmstrong: Of course, BL2 has many more useless stuff like writing the board ID to saradc (IIRC)
17:07
<
f_ >
but you only need DRAM, PLL, and power init to boot.
17:07
<
f_ >
At least that's the case on gxbb and gxl
17:07
<
narmstrong >
Yep we can do the other stuff from u-boot
17:08
<
f_ >
ISTR having discovered BL2 PLL init code in U-Boot too (amlogic's fork).
17:08
<
f_ >
gxl BL2 PLL init code at least.
17:14
<
f_ >
the 2 lines with dxXlcdlr1
17:22
<
f_ >
narmstrong: I guess SM1 and G12B use the same BL2 binaries, correct?
17:23
<
narmstrong >
Hmm Nop but they should be close
17:28
<
f_ >
$ diff aml-a311d-cc/bl2_new.bin aml-s905d3-cc/bl2_new.bin
17:28
<
f_ >
Binary files aml-a311d-cc/bl2_new.bin and aml-s905d3-cc/bl2_new.bin differ
17:32
<
f_ >
G12A then I guess?
17:40
<
lvrp16 >
f_: I will check in a bit. I forgot.
17:41
<
lvrp16 >
It is using a very old bl chain.
17:41
<
lvrp16 >
I have to update it.
17:47
hexdump0815 has quit [Quit: WeeChat 1.9.1]
17:47
hexdump0815 has joined #linux-amlogic
17:51
buzzmarshall has joined #linux-amlogic
17:55
vagrantc has joined #linux-amlogic
17:58
<
f_ >
BTW seeing that you're already working on V1.0 now..will I get V1.0 instead of V0.2?
18:03
CyReVolt has joined #linux-amlogic
18:05
<
CyReVolt >
I've just commit initial notes on the newer `adnl` protocol and tool, which also has commands to write to and read from memory.
18:05
<
CyReVolt >
committed*
18:05
<
CyReVolt >
That should let us do the same on newer SoCs, though I haven't tried yet. Capturing some USB traffic is on the agenda.
18:06
<
f_ >
adnl is based on fastboot
18:06
<
CyReVolt >
It's really just a hacked up fastboot with some specific extra commands and such. Their adnl CLI tool is based on the older C++ implementation.
18:06
<
f_ >
The output of that command did look familiar to me...
18:08
<
CyReVolt >
the weirdest part is the Lua library and the extensions they built into their libraries - I suspect it's for their testing
18:09
<
f_ >
CyReVolt: have you got gxbb hw?
18:10
<
f_ >
Nice to see adnl reverse-engineering work
18:10
<
CyReVolt >
I have no gxbb boards, no. Can you recommend one? :)
18:11
<
f_ >
I guess ODROID-C2, Videostrong KII Pro or WeTek Play2/Hub could be good candidates
18:12
<
f_ >
but there are many set-top boxes claiming to be a KII Pro...
18:12
<
CyReVolt >
ok gotcha
18:13
<
f_ >
so either get a C2 or WeTek Play2/Hub
18:13
<
f_ >
>Sign in to GitHub
18:13
<
f_ >
to continue to Gist
18:13
<
CyReVolt >
I had no idea how to configure the kernel. It's the Android TV aarch64 kernel with like 300 places of `#if CONFIG_AMLOGIC`
18:14
<
CyReVolt >
ah hang on
18:14
<
CyReVolt >
gotta omit the `edit` in the URL
18:18
<
f_ >
> Hello, we are in BL2X world !
18:19
<
f_ >
Never heard of all this lol
18:21
<
CyReVolt >
that is on S905X4 :-)
18:27
psydroid2 has quit [Remote host closed the connection]
18:33
<
f_ >
Yeah I guess it's a new family
18:34
<
f_ >
these logs are ridiculous though lol
18:44
<
lvrp16 >
f_: there's v01 which is v0.1. v02 is v0.2. v1 will be v1 and the only difference is the addition of a switch and some changes to the SD card setup that does not affect the software.
18:44
<
lvrp16 >
v02 and v1 is understood to be the same
18:45
<
lvrp16 >
only v01 were prototypes
18:45
<
lvrp16 >
CyReVolt: they got S905X5 soon :D
18:45
<
lvrp16 >
too many models in rapid succession
18:52
<
f_ >
Hello, we are in BL2X world !
18:56
<
CyReVolt >
not very surprising, probably also somewhere in the Metaverse market I suppose?
19:01
<
f_ >
implies there are multiple BL2*
19:01
<
f_ >
Honestly I expect junk with only 1% being needed init
19:02
<
f_ >
But I don't have any SC2 boards so
*shrug*
19:03
<
f_ >
+ there doesn't seem to be any fip recipe for it avilable publicly...
20:25
<
CyReVolt >
that includes a signed BL2X
20:49
CyReVolt has quit [Quit: Leaving]
21:24
<
phh >
narmstrong: X4's AO CPU is RISC-V if that's what's surprising you?
21:47
<
f_ >
Surprising indeed. It used to be a Cortex-M
22:14
vagrantc has quit [Quit: leaving]
22:36
vagrantc has joined #linux-amlogic
23:14
jacobk has joined #linux-amlogic
23:31
jacobk has quit [Ping timeout: 240 seconds]
23:55
psydroid has quit [Remote host closed the connection]