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
camus has joined #linux-amlogic
adamg_ has quit [Quit: Connection closed for inactivity]
buzzmarshall has quit [Quit: Konversation terminated!]
jacobk has quit [Ping timeout: 245 seconds]
jacobk has joined #linux-amlogic
vagrantc has quit [Quit: leaving]
hexdump0815 has quit [Ping timeout: 264 seconds]
hexdump0815 has joined #linux-amlogic
vagrantc has joined #linux-amlogic
jacobk has quit [Ping timeout: 260 seconds]
jacobk has joined #linux-amlogic
camus has quit [Ping timeout: 256 seconds]
vagrantc has quit [Quit: leaving]
jkl has quit [Quit: Gone.]
jkl has joined #linux-amlogic
camus has joined #linux-amlogic
camus has quit [Remote host closed the connection]
camus has joined #linux-amlogic
JohnnyonFlame has quit [Read error: Connection reset by peer]
<narmstrong> f_: what would be the point of tpl? It would still need to have mmc code to load SPL, would it enable to have access to more sram ? If yes, then yes since on g12 there’s much more ddr setup to handle and we will be heavily limited by the rom load size
<f_> narmstrong: True, and on g12 the ROM can load more than 48K IIRC
<f_> I should dump the bootROM on my cm4 and cottonwood boards, however
<f_> narmstrong: and my guess with aml_ddr.fw is that they wanted to make it look like actual dram firmware
<f_> Speaking of dram firmware, just for fun I think I should test I.MX8MQ firmware. It'll probably not boot however
<narmstrong> f_: didn't they move all the ddr code out of bl2 into those .fw files ?
<f_> They moved it only to aml_ddr.fw, I think the rest were supplied by Synopsys
<narmstrong> ack, so they did a sort of tpl, but instead of adding a new stage they added loadable objects in a very complex way
<f_> but oh well, my guess with bl2 using the same codebase seems to be partially false
<f_> Seems like they rebased on tf-a v1.3 at some point
<f_> (hence some bl2.v3.bin binaries in libretech-amlogic-boot)
<f_> narmstrong: I think Synopsys is responsible for this. But oh well, that does mean I can skip all reverse engineering and just use aml_ddr.fw in SPL!
<f_> (which I won't do)
<f_> (it's a blob nonetheless, so reversing it would be better, at least in my opinion)
<narmstrong> yup
<lvrp16> f_: heroing it hard 🤣 more pressure on me to sell more boards to make use of all the work.
<lvrp16> I am in the middle of solving a critical user issue, the necessity to flash MicroSD cards. Should be done in a week.
<f_> lvrp16: haha, pretty ironic that S905X boots fine while S905 b0rks lol
<f_> I wonder if there was some potato magic involved...
<lvrp16> I pick winners.
<lvrp16> 🤣
<f_> I still want to fixup S905 boot before upstreaming. Besides, I've been preparing for upstreaming yesterday, cleaning up my code.
<lvrp16> I keep tabs on semiconductor cycles, eg. flip flop. S905 was a significant change from S805. There were bound to be issues from a new generation.
<lvrp16> Never use the first chip from a new process or architecture.
<f_> yeah I also feel like S905 was more of an experiment
<lvrp16> Almost universal, always buy the 2nd to last model year of a car platform.
<lvrp16> New platform, new undiscovered issues. Take a while to round out.
<f_> S905X DRAM did change a little compared to S905. It didn't change too much, fortunately. DMC seems to not have been modified much
<f_> lvrp16: and I didn't respect that rule by using the first 64-bit MacBook lol
<lvrp16> Haha yeah, the shiny new thing is a trap.
<f_> (it was identical to the previous, 32-bit MacBook but instead had a new CPU)
<f_> lvrp16: tbh it was pretty much identical to an X60 with a C2D mod, mostly.
<lvrp16> If you look at how much better rounded M2 is vs M1.
<f_> lvrp16: except it wasn't shiny new when I used it :P, I usually use old hardware and that MacBook was no exception: it was from 2006
<lvrp16> The only things I like new are monitors.
<f_> ^^
<f_> (and keyboards)
<f_> Honestly my 2011 hp is more than enough for me
<narmstrong> i have the first intel 32bit only macbook, now it's a brick :-p
<f_> narmstrong: tried installing coreboot?
<narmstrong> f_: no, but I have too much old stuff to make boot again to care^^
<f_> I heard these 32-bit MacBooks had read-only flash while the 64-bit version had read/write flash
<f_> narmstrong: lol
<f_> Might try installing more ram if ghidra and chromium continue to suck all my RAM though
<f_> (making it unable to enter suspend "Out of memory", I have to close ghidra and chromium, especially if I have gitlab open)
<narmstrong> i have even better, an HP Jornada 720
<f_> I'm not using it to its limits, even. I have 8 GB of ram installed but it can actually handle up to 32 GB
<narmstrong> this thing has the OS on a read-only flash
<narmstrong> *rom
<f_> narmstrong: HP Web..thing?
<f_> oh haha
<narmstrong> it should run mainline linux, i never tried...
<narmstrong> all the blog posts about it are very old, it's hard to know^^
<f_> I've an elitebook
<f_> I mostly use it as a desktop, with a monitor connected
<f_> hmm S905 cannot boot linux anymore?
<f_> >sha fail.
<f_> I like this more.
<f_> sha fail...hmm.......
<f_> This seems to come from the SCP.
<f_> This is getting weird.
<f_> Added some udelays and it also boots, but linux doesn't boot unless I plug the box into the wall
<f_> hmm looks like the oopses are page map-related?
<f_> and I think I removed some DTPR-related writel's....this could be the issue
<f_> No, it doesn't boot unless I plug in HDMI, so weird...
camus has quit [Ping timeout: 264 seconds]
psydroid has joined #linux-amlogic
<f_> lvrp16: S905 fun!
<f_> narmstrong: Probably not an SCP issue
<f_> I cloned my repo, reset to d6c460b39f548ee6e55531df06394316ad694d78, added SCP fixes to boot, and no kernel oops at all.
<f_> So this is definitely a regression.
<f_> I'll probably try to bisect.
ungeskriptet has quit [Quit: The Lounge - https://thelounge.chat]
ungeskriptet has joined #linux-amlogic
jacobk has quit [Ping timeout: 245 seconds]
camus has joined #linux-amlogic
<f_> Kwiboo: Want to test yourself?
jacobk has joined #linux-amlogic
jacobk has quit [Ping timeout: 255 seconds]
f_[xmpp] has joined #linux-amlogic
ungeskriptet9 has joined #linux-amlogic
ungeskriptet has quit [Ping timeout: 264 seconds]
ungeskriptet9 is now known as ungeskriptet
psydroid has quit [Quit: KVIrc 5.0.0 Aria http://www.kvirc.net/]
<f_[xmpp]> Looks like a gcc issue?
kenny has quit [Quit: WeeChat 4.1.1]
adamg_ has joined #linux-amlogic
jacobk has joined #linux-amlogic
jacobk has quit [Ping timeout: 268 seconds]
<f_> I confirmed those oopses were indeed a regression!
<f_> I've been testing on a lot of commits...
<f_> lvrp16: I wonder why lepotato can't power from USB-A correctly while my box (which requires a 12V power supply) can power from USB-A with HDMI just fine?
<f_> Is S905 using less power than S905X?
<f_> Seems like that regression is an old one.
<f_> I think I found it!
<narmstrong> my lepotato works fine from usb-a only
<f_> with HDMI?
<f_> Then it must be my eSATAp port not providing enough power :S
<f_> Anyway, commit 108f895f7e57e5a151bbb6091ce8fc8a1f918bea
<f_> onwards, I can reproduce the issue.
<f_> anything before works fine (with the SCP firmware issue fix)
<f_> 7 minutes and it still didn't panic.
<f_> That tells me that the issue is DRAM init-related
<f_> 8 minutes and 40 seconds.
<f_> 20 minutes. I guess that was the issue!
<f_> Aha. The rank mode I was using uses *2* channels...apparently?
<f_> and I removed support for 2 channels, so it might explain why it didn't work exactly as expected!
<f_> Issue is clearly unreliable DRAM init on my set-top box specifically.
<f_> I couldn't reproduce on lepotato because I got the 1 GB version :S
<f_> (I$ asked for the 1 GB version)
<f_> s/$//
<lvrp16> If you need the 2GB version, let me know.
<f_> so that I have 2 lepotatoes?
<f_> *so I'd then...
<lvrp16> Mashed Potatoes
* f_ gives lvrp16 some mashed potatoes.
<f_> Actually it looks like the 1 GB version is just the 2 GB version but with only 2 RAM chips, and I can still see traces.
jacobk has joined #linux-amlogic
_whitelogger has joined #linux-amlogic