sputnik_ has quit [Read error: Connection reset by peer]
a5m has joined #linux-amlogic
lissyx has quit [Ping timeout: 245 seconds]
TobiasTheViking has quit [Ping timeout: 264 seconds]
tingoose has joined #linux-amlogic
wens has quit [Ping timeout: 265 seconds]
wens has joined #linux-amlogic
pomme has joined #linux-amlogic
<pomme>
Hello
<pomme>
I am trying to understand the documentation of the S905 and S905X given by Khadas and Odroid about the memory controller of those two SoC
<pomme>
According to that I understood so far and from the plan of the Khadas Vim and Odroid C2 the memory controller is onyl able of handling one channel (but two ranks per channel)
<pomme>
But in the register description of the DRAM the term channel appears so it bothers me
<pomme>
Does someone know if the memory controller is a one channel-two ranks or a two channels-two ranks ?
lissyx has joined #linux-amlogic
tingoose has quit [Ping timeout: 248 seconds]
lissyx has quit [Ping timeout: 248 seconds]
lissyx has joined #linux-amlogic
sputnik__ has quit [Remote host closed the connection]
<Joe-e19>
Does anyone know how to get self compiled mali.ko modules running? Or do the precompiled mali.ko module contain property code? Sovthat I can't get my custom kernel with custom mali.ko running?
<Joe-e19>
I got anything from the kernel GPLv2 open, except the mali.ko. Help or hints are welcomed. Thanks for ideas.
<pomme>
@Joe-e19 I am not expert but from what I read you need to have the same version for libMali.so and mali.ko
<pomme>
I do not know how ARM package their thing but I think that you need libMali.so r2p1_05rel0 [Revision 96995] and the source code of mali.ko for the same revision
<Joe-e19>
There are two version numbers: "In my libMali.so is this version: Mali online shader compiler r2p1_05rel0 [Revision 96995]
<Joe-e19>
And another version: 1.4 Linux-r3p1-01rel0 what release do I need now?" This is where I've ended with.
<Joe-e19>
pomme: I think the same but from where do I get the mali.ko sources?
<pomme>
Did not you already get them? You were talking about self compiled mali.ko at 14:48
<Joe-e19>
Yeah but the mali folder where the mali.ko is getting compiled out of it is not containing a .version file.
<Joe-e19>
Usually the sources contain a .version file. Mine does not. So I assume it's the wrong one.
<Joe-e19>
Why is the revision called: r2p1_05rel0 [Revision 96995]? Is it a custom revision from my manufacturer???
<Joe-e19>
I send a few emails to yifangdigital but they never respond.
<Joe-e19>
Yeah fuckk. I think yifangdigital may have made customizations to the mali.ko and just released without the customized code to me. Might be or?
<pomme>
From what I read on the arm website the modifications that can be done in mali.ko are mainly about power consumption not really about getting the thing works
<Joe-e19>
Without custom mali.ko I can't patch my kernel...
<Joe-e19>
Yeah for power management ;) but I get mali session ended with my own mali.ko module.
<pomme>
I think that your problem of ending session is that you do not have the same version for libmali.so and mali.ko
<pomme>
Maybe if you try to get a mali.ko and libmali.so from arm for the Mali 400 it will work?
<Joe-e19>
I think so but from where it must be for Amlogic 8726-MX SoC.
<Joe-e19>
I get with the libMali.so and my own mali.ko this error:
<Joe-e19>
[ 12.251389@0] Mali<2>: Session has ended
<Joe-e19>
[ 12.246390@0] Mali<2>: No handler for ioctl 0xC0048203 0x40D57B3C
<Joe-e19>
Does amlogic supply libMali.so precompiled mali drivers?
<pomme>
Maybe they provide the libMali.so, you can try to see on their website
<Joe-e19>
Okay.
afaerber has joined #linux-amlogic
<pomme>
But I found in the code in drivers/amlogic/mali/common/mali_kernel_core.c:182 a message which print the version of your driver under the form rXpY
<pomme>
So can you try to find all Mali related messages in dmesg? (dmesg | grep -i "mali")
<Joe-e19>
Yeah rXpY should be the revision but without the custom extansion which irritates me.
<Joe-e19>
I use my soldered UART to see mali messages fly
<pomme>
You manage to get the rXpY of your mali driver???
<pomme>
I will try to see what is MALI_DEBUG_PRINT maybe you need to compile the kernel with a debug option
<Joe-e19>
Aha this looks good. It's pretty nice that you help me. You might solve this issue.
dlan has quit [Ping timeout: 245 seconds]
<pomme>
Apparently MALI_DEBUG_PRINT is a macro which check if the level given as first arg is inferior or equal to mali_debug_level
<pomme>
In your case the level given as arg is 2
<Joe-e19>
Yes init.rc says the following: insmod /boot/ump.ko ump_debug_level=2
<Joe-e19>
insmod /boot/mali.ko mali_debug_level=2
<pomme>
But MALI_DEBUG_PRINT internally call MALI_PRINTF which directly print "Mali<level>"
<pomme>
Then MALI_PRINTF is another macro which called _mali_osk_dbgmsg which called vprintk
<Joe-e19>
Hm... I just have to insmod the module without the 2?
<pomme>
So since you showed me messages wich printed "Mali<2>" I think that either you miss some Mali message in dmesg or the session exit before printing the rXpY message
<Joe-e19>
Will a working mali.ko the precompiled one print out the rxpy version?
<pomme>
I think it should, but I will try to read the calls
dlan has joined #linux-amlogic
<Joe-e19>
Okay. Would it be of any help to upload my dmesg for you?
<pomme>
Yes it would be cool if you can upload the output of dmesg | grep -i "mali"
<Joe-e19>
Okay I can't do dmesg I can only get it via uart ttl converter. Brb i'm booting my notebook and restart my tablet with precompiled mali.ko
<pomme>
Ok
<pomme>
I search a bit into the code and as the dmesg message indicate it I suspect that the libmali.so does a ioctl(FLAG, ...) with a FLAG which is unknown for mali.ko
<Joe-e19>
So the tablet booted and show me Mali<2> output ima uploading the "dmesg" to my dropbox.
<pomme>
Cool !
<Joe-e19>
Need to copy the link from the notebook takes a bit
<Joe-e19>
Does the "dmesg" contain something interesting for mali.ko module?
<pomme>
got it 462:[ 6.897923@1] Mali<2>: Found Mali GPU Mali-400 MP r1p0
<pomme>
So you need to find on the arm website (or another source but arm is prefferable) the libmali.so r1p0
<Joe-e19>
Okay r1p0 is the version... damn old like 2011 or 2012...
<pomme>
Yes it is a pretty old version
<pomme>
But the bug is "normal" if you use a newer libmali.so with and old mali.ko
<Joe-e19>
The precompiled mali.ko showed now r1po means that the libMali.so has too r1p0.
<Joe-e19>
My Android firmware already contains the library libMali.so which is working with the precompiled mali.ko.
<Joe-e19>
But why do I need to find a libMali.so?? I already have one...
Barada has quit [Quit: Barada]
<Joe-e19>
The thing is that the mali.ko compiled by me doesn't work with the precompiled libMali.so
<Joe-e19>
I think I got it now: the customer provided me an too old mali.ko source inside the kernel source and i have a too new libMali.so? Should be now correct?
<Joe-e19>
libMali.so was also provided by the vendor with the mali.ko source.
<pomme>
Yes your libmali is too new
<pomme>
I think
<pomme>
But yifangdigital also give you a libmali.so???
<pomme>
Hum
<pomme>
I think I did not understand everything
<pomme>
What is the precompiled mali.ko you are talking about?
<Joe-e19>
Yifangdigital is the manufacturer of the tablet who manufactured the mainboard screen and touchscreen and the chassis. Except the AXP202 chip and the Amlogic 8726-MX SoC and the hynix 4x 256 MB RAM chips.
<Joe-e19>
The precompiled mali.ko is the mali.ko which was chipped with the device as stock Android.
<Joe-e19>
I bougt the Intenso Tab 814 here in Germany with stock Android and there is a mali.ko binary on the devide which I called precompiled mali.ko ;)
<pomme>
Ah ok !
<pomme>
So yes the precomplied mali.ko is r1p0 and if you used the precomplied version you need to have a libMali.so r1p0
<pomme>
You can download the sources for mali.ko and download the corresponding libMali.so
<pomme>
Yes but be sure that the libMali.so is also for Mali 400
<Joe-e19>
Luckily this tablet (Intenso Tab 814) exists as another version called Prestigio PMP5580C.
<Joe-e19>
You can google the Prestigio and the Intenso tablet and you see how similar they are.
<Joe-e19>
For the Prestigio PMP5580C do exist a custom rom without custom kernel (the guys have just modded the initramfs) but the customrom has so many bugs.
<pomme>
Oh that's cool! Maybe you can try to get the custom kernel
<pomme>
And you can also check Lineageos (the heir of CyanogenMod), maybe they support your tablet
<Joe-e19>
I have the firmware from the prestigio in 3 different versions maybe one libMali.so will work.
<pomme>
I hope
<Joe-e19>
No they don't support my tablet, because I'm the guy who is going to make this tablet supported. ^^
<pomme>
Ah ok! Courage!
<Joe-e19>
Yes courage and patience is my 2 th name. Lol.
<pomme>
Yes patience is really important in computer science. And particullarly in kernel hacking and all the things which gravitates around lol
<Joe-e19>
Nice. I'm a good employer than. I'm 19 and go to school here in Germany and learn about computers.
<Joe-e19>
In the school we have never programmed in C/C++ which is the i
dsd_ has joined #linux-amlogic
<Joe-e19>
Best programming language or not even assembly. We just use Windows at home I use like 99 % Debian Linux.
<pomme>
It is a cool thing that you try to hack a bit but do not burn steps, you will not learn the mystery of C or assembly in 1st year of undergraduate in computer science
<Joe-e19>
I might have a custom mali.ko kernel image! I just kept them nice I'll try them.
<pomme>
795:[ 7.144266@0] Mali<2>: Found Mali GPU Mali-400 MP r1p0
<pomme>
So yifangdigital made the thing well and they gave the corresponding sources for their precomplied mali.ko
<Joe-e19>
Yes.
<pomme>
Hum
<pomme>
But that's strange because you also have message like Mali<2>: No handler for ioctl 0xC0048203 0x40E3FB3C for the custom mali.ko...
<Joe-e19>
Yeah. For me it looks like the mali sources is missing code... This can't be.
<pomme>
Yes I think it too...
<pomme>
I think you will need to try the prestigio firmware :s
<Joe-e19>
The Prestigio firmware comes without any sources because Russia.........
<pomme>
><"""
<pomme>
It is for sure a GPL violation
<Joe-e19>
Yeah it's a pain with companies and put the GPL sources out.
<Joe-e19>
If I tell you now something it will break the wall:
<Joe-e19>
My Intenso Tab 814 uses a U-Boot bootloader by stock shipped and they give me no sources.
<Joe-e19>
They just provide the kernel sources as we learned now mali.ko custom code is missing and the GPLv2 code for the used bootloader is missing.
<pomme>
I think your only solution is to get a mali.ko libmali.so couple from ARM for the Mali 400
<pomme>
...They screwed
<pomme>
But sometimes it is not intentionnaly, some small companies are just not aware of the GPL
<pomme>
But all this consolidate me in one thinking: being mailine is very important!
<Joe-e19>
mailine?
<pomme>
mainline* sorry
<Joe-e19>
You mean that the tablet software should be mainline to be good?
<Joe-e19>
The tablets bootloader is from 2011 this is not mainline. I know now what Intenso and Yifangdigital did...
<pomme>
I mean that companies must put their changes into the mainline kernel instead of keeping them for them
<Joe-e19>
...they just sold these devices and just wanted to generate money without comply to GPLv2
<pomme>
I do not think so because they gave you the sources of their kernel?
<Joe-e19>
Yeah they gave me it, but the mali.ko isn't working as we learned now.
<Joe-e19>
So the sources they have given me is not complete.
<pomme>
Yes, they made a mistake while giving the sources...
<Joe-e19>
Really?
<pomme>
Apparently it is the case
lissyx has joined #linux-amlogic
<Joe-e19>
I think that Intenso don't want that someone like me is pushing updates out for this tablet. So that they further earn money from other and newer tablets.
<Joe-e19>
I emailed the company via support@intenso.de and asked where the full sources and the bootloader source is. They have pointed me just to the link and told me that all is there.
<pomme>
Remail them with the proof that the sources does not match but do not be aggresive
<Joe-e19>
Of course there is "just" the kernel without working mali.ko sources... there is no bootloader source inside the kernel archive.
<Joe-e19>
Okay I showed no agressive behaviour so far. ;) So I should ask them for working mali.ko sources?
<pomme>
Yes, mail them and bring the proof that you think that their sources does not correspond
<pomme>
do not*
<Joe-e19>
How does the proof look that the sources don't match? The revisions are both the same...
<pomme>
The behavior
<Joe-e19>
Okay. What if the company isn't going to believe me?
<pomme>
And begin together the negociation with the company
<Joe-e19>
Okay.
<Joe-e19>
I'll eamil Intenso this evening which is like in 2 hours.
<pomme>
I hope they will cooperate
<Joe-e19>
I hope too.
<Joe-e19>
What should I put in the subject if my email?
<Joe-e19>
Your kernel sources are incomplete??
<pomme>
Kernel sources do not match precompiled kernel
<pomme>
I think
<Joe-e19>
Okay. The only thing that doesn't work too with the custom kernel is the camera from the tablet.
<pomme>
Tell them too
<Joe-e19>
I think the kernel config just uses another camera driver...
<pomme>
But what is the problem when you use the source mali.ko
<pomme>
You do not have anything printed on the screen
<Joe-e19>
Yeah the backlight from the screen is on and the only thing i see is a black screen which is powered on but nothing showing.
<Joe-e19>
But the backlight is on... no Android bootanimation follows as it should...
<Joe-e19>
When I use the source mali.ko I get no handler for .... via UART and the black screen which is powered on (the backlight). That's all.
<pomme>
But since the screen prints nothing how can you know that the front camera does not work too?
<Joe-e19>
Oops I have a custom kernel with the precompiled mali.ko from Intenso, okay? ;)
<pomme>
Oh ok !
<pomme>
Sorry I am bit loosed between custom kernel, source kernel and precompiled one
<pomme>
lost*
<Joe-e19>
No problem.
<pomme>
So with the precompiled one the front camera is not working... *sigh*
<Joe-e19>
Would it be possible to just use the precompiled mali.ko and a newer Android version then Jelly Bean like one like KitKat or LolliPop?
<pomme>
Normally you can
<pomme>
You just need to hope that all the functions that are used in the precompiled mali.ko still exists in a newer kernel (Android 4.4 for example)
<Joe-e19>
Yeah that's why the customrom KitKat maybe exists. But the guy had no access to the kernel sources because he is from Russia and the tablet is from Russia.
<Joe-e19>
So I should be able to get KitKat running with precompiled mali.ko and with selinux enabled in the kernel config???
<pomme>
You can hope to get it work
<pomme>
But I will not claim that it will work directly
<Joe-e19>
Okay I'll start to hack around this while awaiting Intenso's reply to the incomplete kernel sources...
<pomme>
It is a good idea :)
<Joe-e19>
The only thing I want to do with this tablet is just to get KitKat with a custom kernel running and change the mtd partitions via the kernel config file.
<Joe-e19>
On my side I see no problem why the precompiled mali.ko module should stop me from that plan.
<pomme>
if you have the .config of the precompiled kernel you should use it to compile KitKat
<Joe-e19>
As long as my kernel doesn't need to increase the 3.0.8-localversion number.
<pomme>
Why do you care about the kernel version?
<Joe-e19>
I have the config file. ;) The only thing is the camera that doesn't work so far.
<Joe-e19>
Because I've heared that the kernel version needs to be like 4.x.x to be secure against security valnuerabilities.
<pomme>
Oh yes! But I did not understand the "need to increase"
<Joe-e19>
Ah okay.
<Joe-e19>
It won't be necessary to have a kernel like 3.x.x something newer than 3.0.8 for KitKat?
<pomme>
KitKat comes with its own kernel version no?
<Joe-e19>
True. ^^ I should be able to get even Nougat running... lol.
<Joe-e19>
Or?
<pomme>
So you do not need to care about 3.0.8
<Joe-e19>
Yeah. But I won't be able to get LineageOS officially supported... :((
<pomme>
You just try to compile KitKat with your .config and add the precomplied mali.ko and the libmali.so too
tingoose has joined #linux-amlogic
<pomme>
LineageOS need that kernel 3.0.8 is supported?
a5m has quit [Remote host closed the connection]
<Joe-e19>
Yes. And then I enable swap in my kernel config, good idea? Cause swap is deactivated in the original .config.
<Joe-e19>
No LineageOS requires a kernel device tree which fully complies with the GPLv2 or GPLv3... :'(
<pomme>
As you wish, I am not a great fan of swap in the 21th century
<pomme>
Oh yes for now if Intenso does not reply you will not comply with GPL
<Joe-e19>
What is an alternative to swap in the 21th century?
<pomme>
For me: nothing
<Joe-e19>
Zram??
<Joe-e19>
More RAM is your alternative? :D
ldevulder has quit [Quit: Leaving]
<pomme>
We already have enough RAM
<pomme>
PC have on average 6-8GB and smartphone 2GB
<Joe-e19>
Okay my tablet has 1 GB. LOL.
<pomme>
If you do not launch firefox with 100000 tabs it should be ok
<pomme>
How many NAND do you have ?
<Joe-e19>
8 GB lol.
<Joe-e19>
I can increase it with a 64 GB MicroSDCard.
<pomme>
Oh in this case you can give 1GB of swap if you want
<pomme>
I think that the need of swap really depends of your use case
<pomme>
If you want to play Fortnite or PUBG yes you will (but I think that the SoC will not be able to display them at a good framerate)
<Joe-e19>
I'm planning to get LolliPop running and make my microsdcard to ext4 2th oartition and intigrate it into /data as extansion to the 8 GB
<pomme>
If you want to YouTube then you will not need swap
<Joe-e19>
I'm definitively not planning to play that. I'm not a gamer.
<Joe-e19>
I'm looking YouTube videos :)
<pomme>
Swap was used in a time where DRAM was very little (some MB) and expensive
<Joe-e19>
Okay. Lol.
<pomme>
Now it is bigger and less expensive (ok it was cheaper in 2013 compared to now)
<Joe-e19>
Today RAM goes like 64 GB
<pomme>
I do not swap is shitty, it is just for me less useful now
<pomme>
say*
<Joe-e19>
My computer has 12 GB 32 GB would be possible
<Joe-e19>
Yeah my computer has no swap too just for hibernate lol.
<Joe-e19>
My /data partition from my tablet is 877 MB little that's why I'm going to my custom kernel effort.
<pomme>
Wow that's really small!
<Joe-e19>
I have a partition on the 8 GB Nand which is 6 GB and holds a microsdcard which I want to replace with the /data
<Joe-e19>
My plan is to make a 7 GB /data by merging the both partitions and then I'm satisfied.
<pomme>
I am not really good in partitionning so I will not be able to say it is feasible
<pomme>
if it is*
<Joe-e19>
I'm right now happy that I'm not the one who isn't able to compile the kernel with working mali.ko. I'm let's say happy about that I made nothing broken on my side.
<Joe-e19>
Luckily Intenso did.
<Joe-e19>
The tablets nand is partitioned not physicalls via a MBR it's just partitioned with MTD containers.
<pomme>
I do not know that a MTD containers is
<Joe-e19>
On the Nand are badblock filesystems which hold upoin them fat32 or ext4 if needed.
<Joe-e19>
Yeah mtd frames are having a label and a starting and an ending point with the hardware numbers like 0x000000 to 0x00003000 holds the bootloader etc.
<Joe-e19>
In the kernel config is an option that allow to specify the mtd partition endings and beginning to my needs.
<Joe-e19>
Of course I won't be able to change the location (beginning and ending) of the bootloader, kernel and recovery partitions. But I just need to get partitions like /system and the others bigger.
<pomme>
I do not get the point of using MTD since NAND in a smartphone is used as a block device
<Joe-e19>
Hahah I don't have a nand with a badblock management chip mine is just raw.
<Joe-e19>
So I must use a filesystem with badblock management like yaffs2 offer.
<Joe-e19>
This tablet is damn old. Released in 2013 and I'm hacking it in 2018 lol.
<pomme>
I am not sure that I understood well. You are telling me that there is no wear leveling or erase managing for your NAND?
<Joe-e19>
This tablet is so cheap that I'm not even scared to break it with my experiments.
<Joe-e19>
Yeah there is no managing it's managed by software and not by a hardware chip which is used today.
<pomme>
Ok, I understand better now
<Joe-e19>
That's why my nand partitions are: /dev/block/mtdblock0 and not /dev/block/emmcblk0 etc.
<pomme>
You bought it just to hack it?
<Joe-e19>
This emmc just shows that the Nand is like a sd card. Even SDCards have a small controller infront of the nand which is wear levelling and badblcok management.
<Joe-e19>
Yeah I just bought it to use it for one or two years and than I'm hacking it. :D
lissyx has quit [Ping timeout: 245 seconds]
<Joe-e19>
I did play games with this tablet but somehow it turned out that the tablet has a too old Android version without ART runtime and no swap which would blow the 1 GB RAM issue away.
<Joe-e19>
And the ~800 MB /data which required me to app2sd use which slowed the tablet down. Lol.
<Joe-e19>
I like hacking.
<pomme>
Continue in this way!
<Joe-e19>
Yeah I'm the smartest in my class. Only me and one other person are using Linux as operating system.
<pomme>
Using Linux does not make you smarter than others. I know people that use Linux that are just fucking git lol
<Joe-e19>
I'm using Linux since I'm 12 years old...
<Joe-e19>
Hahaha that's true Linux is a big tool which enables you to do much.
<Joe-e19>
One another example: rooting your Android device means not that you are smart but it enables you to be or get smart. Correct?
<pomme>
You should follow computer science studies and particullarly operating systems one
<Joe-e19>
Rooting doesn't come with a newer and more powerful Android OS.
<pomme>
Maybe, I never rooted my phone because I did not need it
<Joe-e19>
Will Linux kick MS Windows oneday away?
<pomme>
I hope but the more the time pass the less I am optimist on this question
<Joe-e19>
I'm also not rooted since three months I had the prestigio firmwares on my intenso tablet and the customroms with alot of bugs.
<Joe-e19>
Yeah I hope too
<Joe-e19>
As I said already no customrom had a custom kernel the only thing which was custom are the ramdisk and the operating system. So I might have luck cause of the 99 % kernel source.
<Joe-e19>
6:46 PM here right now. What's your time?
<pomme>
Same and I think I will not delay to go back home lol
<pomme>
So have a nice day and have fun hacking your tablet :D
<Joe-e19>
Yeah. Thanks for your time! :)
vagrantc has joined #linux-amlogic
<Joe-e19>
You enlightened me alot.
<Joe-e19>
See ya. IRC is so nice.
<pomme>
You are welcome!
pomme has quit [Quit: Page closed]
lissyx has joined #linux-amlogic
return0e_ has joined #linux-amlogic
<webczat>
general question to arm developers. :D is there any nice board that can be used for me (as a hobbyist) to do os programming? as in making os from scratch or something like that. like to do that I would need full hardware specs
<webczat>
i am not sure if I would buy such a board, currently just researching options
<webczat>
also another related question: all such boards do have *some* software? as in, the SOC ROM stuff? like I can never boot directly into a reset vector and start from there
tingoose has quit [Ping timeout: 245 seconds]
<narmstrong>
modern ones have ROMs
<narmstrong>
you can look at microcontrollers like the STM32's, the CPU will directly boot on a flash address and you have full HW specification and documentation
<narmstrong>
but it's a Cortex-M based SoC, without MMU
<webczat>
narmstrong: so no fun :D
<narmstrong>
but it's cool enough to experiment OS Programming
<narmstrong>
yes it is, try to make your own RTOS is real fun
<narmstrong>
when you understand all the logic in the context switching and write it, it's real fun
<webczat>
I am specifically into OS programming, or os like. not into making electronic stuff. that means I would need mmu and probably even some onboard storage.
<narmstrong>
do you mean drivers writing ?
<webczat>
narmstrong: no
<webczat>
narmstrong: I mean what I said, something like os kernel dev. but probably not without mmu. after doing a bit of x86 stuff I would consider no paging = no fun :D
<webczat>
oh well, or *less* fun.
<webczat>
probably same as having code in rom that runs before mine, but that is less of a problem than no mmu.
<webczat>
of course that means x86 was far less interesting as you had the whole firmware before yourself
<webczat>
and (as I am blind) also less accessible. I probably did not have anything like serial port/null modem cable at that time, and also bios does not talk over it so
<webczat>
arm sbc's and so are better in that regard because even on chip roms talk over uart and you know that the thing just restarted and refuses to load anything from emmc for any reason
<narmstrong>
yeah sure
<webczat>
well do you need to initializer ram? :D i mean if I would ever write the code that would be in rom, do you need to initialize ram or do something like that?
<webczat>
i mean how much less fun do I have without writing code for the rom (as in what the rom does that I wouldn't need to do later usually)
<narmstrong>
it depends the complexity of the SoC
<webczat>
not counting program loading
<narmstrong>
but the RAM should be already initialized
<webczat>
like after rom or on power up
<narmstrong>
and if not, you can run some code that does not need RAM to execute itself, in assembly before jumping into C
<webczat>
narmstrong: yeah, unless I would not execute from rom because then :) ram must be active to load anything
<narmstrong>
yep
<webczat>
still, roms are rather not reprogrammable by me and already populated?
<webczat>
well and btw how big are they? for example my c2 shows the rom boot messages and they are rather wague... cryptic even, like it had no place to put larger strings
<webczat>
oh well another thing to play with in armv8 is el3 software :D
<webczat>
yes I already know what EL3 is even though I had actually never learned ARM stuff... because I have no use for it right now
<narmstrong>
it will hard to find an armv8 SoC with a reprogrammable ROM stage
<narmstrong>
qemu is your friend...
<webczat>
well... are there any at all? isn't it like it's hard to find *any* armv8's with open hardware specs that are available?
<narmstrong>
hardly
<webczat>
narmstrong: well, except that qemu is an emulator so some nice things I would normally have to do on a real board are probably skipped because it is an emulator
<narmstrong>
you could maybe use the RPi
<narmstrong>
some guys rewrote the GPU firmware loading the code on the ARM core
<narmstrong>
with this, you can start the ARM core from something loaded from the VPU without any code running before
<narmstrong>
but the SoC itself really sucks
<webczat>
well interesting how... when you boot the rpi, arm core is probably started isn't it?
<narmstrong>
no, the VPU starts first
<webczat>
well weird
<webczat>
weird decision
<narmstrong>
yes, the whole SoC is weirds
afaerber has quit [Quit: Leaving]
<narmstrong>
they don't use the ARM GIC, no physical ethernet interface, ...
<webczat>
boo
<webczat>
narmstrong: well I believe this rom thing is not a hard requirement for me, although I would love if I could program from the very beginning. I believe my current hard requirement is armv8 board with available specs, enough to do everything there even after I get loaded by rom into the ram
<webczat>
possibly another hard requirement is jtag (if I get what jtag is, never used it)
<webczat>
and uarts. although I would rather have problems finding a board without at least one
<webczat>
of course not even sure if I would even buy such a thing, but if I would find anything that suits my needs then at some point i could
<narmstrong>
good luck !
<webczat>
narmstrong: searching around beagle boards, they probably suit my needs but they are not armv8