You have done everything right. The installer probably has a bug. It was the same for me and for "sailorMH" from the link you provided.
"Booting image at 00500000 ... Bad Magic Number Failed loading kernel (or kickstart) image file(s)"
Means you can't load the kernel file/wrong path. Most likely you also have a non-generated image/kernel like mine.
The linux boot partition is supposed to be the first of the linux partitions, but not the first partition on the system. I will write you in detail when I am at the computer - I will put it in this message.
Ok thanks my friend for the information, so it is a problem of the Linux install Jessie installer.
Also I just used the Install A1 iso to install without additional boot/kernel images which according to the readme should not be necessary.
MacStudio ARM M1 Max Qemu//Pegasos2 AmigaOs4.1 FE / AmigaOne x5000/40 AmigaOs4.1 FE
Volker was able to solve the problem, the sound output with es1370 (sb128) now works perfectly and also the network problems were solved!
It was actually interrupt problems.
@Balaton
This does not only affect the AmigaOneXe emulation, but also Pegasos2 where the PCI mapping might be wrong. Volker described it as follows:
Quote Volker: The disturbed audio playback was an interrupt problem with the AmigaOne XE. As I wrote before, the es1370 device uses interrupts to tell the Amiga driver which part of the audio buffer was played back. This interrupt was masked in the interrupt controller. So the interrupt simply did not arrive.
To fix the problem PCI cards have to be 'plugged' into the correct PCI slot. I found a picture of the three currently possible PCI slots of the AmigaOne XE here at https://www.amigaportal.cz/node/159407.
For the sound to work with es1370 the card has to be plugged into one of the slots 00:08.0 - 00.0a.0. So e.g.
With this I have audio without playback problems and the network works as well.QuoteEnd
For correct audio output, please use this command....
-device es1370,addr=0x08
and for the network this...
-device rtl8139,addr=0x09
YouTube:
It will also finally fix the network problem under the Pegasos 2 emulation, but for that we also need to find out the PCI bus addresses if they are not the same.
There is a lot of debug output to see, don't be surprised it will behave the same as under real hardware, but for now I am still testing.
MacStudio ARM M1 Max Qemu//Pegasos2 AmigaOs4.1 FE / AmigaOne x5000/40 AmigaOs4.1 FE
I changed to PIO still has "disco mode" in the console I found information that on the real Amiga A1 people also had problems with this. I will try to find information on how to solve this. changinglm "libata.force" I added : noncq norst transfer mode and "libata.dma"
@Maijestro It can be done, but without a second installation of linux under qemu on ppc it will be a pain. You can also check at the end of the installation what is going on. -but if you are not familiar with Linux it is better not to do this. You will only get frustrated.
edit:
The problem appears to be the same as #182 - IRQ This would explain starting from the installer cd and running all modules in the target system causes a problem with the disk controller. From what I've noticed the problem starts when the USB modules are loaded.
edit:
> Even better you can pass an actual BIOS ROM for a Rage128Pro and then you may be able to get picture even in firmware and don't need an additional VGA device.
Yes it works, on romfile - thanks On pegasos it didn't work.
Edited by smarkusg on 2023/10/3 7:40:22 Edited by smarkusg on 2023/10/3 13:57:56
smarkusg wrote:@balaton @Maijestro It can be done, but without a second installation of linux under qemu on ppc it will be a pain. You can also check at the end of the installation what is going on. -but if you are not familiar with Linux it is better not to do this. You will only get frustrated.
For me the project Linux is finished
It services only to improve the Pegasos2 and AmigaOneXe emulation. We have now a complete 2 emulation with AmigaOs4.1 the only thing missing are the sources to U-Boot for AmigaOneXe to make it more user friendly.
SDL1 problems no longer exist under the AmigaOneXe emulation and it is as fast as the Pegasos2 emulation.
The only problem that remains is that under the Pegasos2 emulation SDL1 games/programs run worse. But don't worry we will solve that problem too.
MacStudio ARM M1 Max Qemu//Pegasos2 AmigaOs4.1 FE / AmigaOne x5000/40 AmigaOs4.1 FE
@smarkusg[quote]smarkusg wrote:Linux Qemu AmigaONE XE using ati-vga and BIOS ROM Rage128Pro Not yet set up everything - now testing
This looks really good....good job. Thanks for the videos.
I also had a lot of time today and tested the AmigaOne XE emulation with Qemu, my conclusion is that it is quite usable and works well.
There are no SDL1 problems, but the Pegasos2 emulation is a bit faster.
Since this emulation is relatively new, it is probably not yet as optimized as the Pegasos2 emulation, but the AmigaOneXe emulation is also very fun and should already be faster than Sam460 and x1000 raw CPU performance.
It is good that they have set up a Linux system under AmigaOne, it certainly helps to make some things even better.
MacStudio ARM M1 Max Qemu//Pegasos2 AmigaOs4.1 FE / AmigaOne x5000/40 AmigaOs4.1 FE
The problem appears to be the same as #182 - IRQ This would explain starting from the installer cd and running all modules in the target system causes a problem with the disk controller. From what I've noticed the problem starts when the USB modules are loaded.
So you say all you did was setting addr for ES1370 and rtl8139 and nothing else to get it booting? All the devices in the VIA chip are hard coded to use the address as on real hardware because VIA chip is created with correct address so all IDE, USB, AC97 should not be affected by PCI cards. This does not explain the error about /boot/a1boot to me.
I can change the code to start assigning PCI devices starting from address 8, I already did that but could not test it other than in firmware they now show up at correct address. But since there are only 3 slots then you probably will need to specify address for -device VGA and -device sm501 otherwise they will take 8 and 9 so thre will only be one slot left, so maybe it's not much better that way. (By the way I never understood why people write single digits in hex when they are the same as decimal so if you want to type less you could just use addr=9 or addr=10 which is still one key less than addr=0xa.)
Quote:
Yes it works, on romfile - thanks On pegasos it didn't work.
I think pegasos2 firmware has a radeon driver but knows nothing about rage128pro or plain VGA so probably does not work. I think it worked with cirrus-vga with the correct Bochs vgabios-cirrus but still gave a lot of BIOS emulator errors on serial. Maybe the problem is just the BIOS emulator in pegasos2 rom as even with QEMU vgabios it gives only errors but you can still type commands in blank QEMU window and boot anyway just don't see output on VGA.
@Maijestro I can't understand how SDL has no problems on amigaone but has on pegasos2. There's nothing different in graphics and CPU emulation between these machines, all the difference are the firmware and north bridge which should not have anthing to do with graphics so only plausible explanation to me is that this SDL version has some code that detects it's running on amigaone and does something differenetly which is missing or does not work on pegasos2. Are the sources of the SDL port that has the problem available? Maybe somebody could check what that's doing differently on amigaone.
Similarly speed difference between pegasos2 and amigaone cannot be optimised in QEMU as these use the same code in QEMU so if there are differences then that likely comes from AmigaOS doing something differently. The amigaone emulation is just adding a minimal ArticiaS emulation that does nothing than creating the PCI bus and forwading access to it and the board code that only creates the devices that are mostly the same as in pegasos2 so there isn't anything in there that could be optimised.
It's great that Volker debugged the sound issue and found a solution. But that does not explain the broken sound with via-ac97 and the issue on pegasos2 is probably not the same as that generally works and only fails occasionally so it can't be interrupts masked. I also don't quite yet understand why using different slot helps as all PCI interrupts are routed to the same ISA IRQ so unless the firmware or AmigaOS only checks if there's a card in the slot to enable the interrupts then it would still get interrupts. If it just needs a card in one slot then only specifying addr=8 for any one card would also fix the problem. Or maybe when it gets an interrupt it only checks the cards at specific addresses so it misses cards using other addresses? Maybe you could forward Volker's original explanation to me for better understanding.(According to the picture Volker referred to network could also use addr=6 then.)
1. "/boot/a1boot". Under "/boot/a1boot" is mounted the "boot" partition where the boot image and boot settings are located. By default it should be "/boot", but this partition should not be larger than the appropriate size. The kernel itself takes up a lot of space after unpacking. That is why there is a "boot" directory on the root partition. The boot image is generated from this and placed on the "boot" partition, i.e. "/boot/a1boot". The "boot" partition is marked as bootable and the boot software looks there for "a1boot.conf" where it finds information on where and what to boot. There is nothing special there as I wrote after loading the system.
Here is a screen shot of what it looks like after loading the cd.
It's placed where it shouldn't be according to the "sailor" list from the link @Maijestro gave, it should be 8,9,10
The system loads if you assign the sound and network card under 8.9
There are two packages on the installation CD in addition to the debian packages: firmware-linux-nonfree_0.43_all.deb linux-image-3.16.0-4-book3s-amigaone-smp+_3.16.7-ckt11-1+deb8u5_powerpc.deb
this one is with "nonfree" maybe that's why it is loaded separately during installation firmware-linux-nonfree_0.43_all.deb
2. SDL SDL problem was checked by @kas1e on real hardware in 16bit mode in this topic:
"Tested old XRick from os4depot on real pegasos2 on both 32bit modes and 16 bit modes on my AGP Radeon9250 card : no issues. All shown correctly. For sake of tests i check with each test via Ranger that i do have 16bit and 32bit modes on the XRick screen, and yeah, old Xrick works fine on real peg2 with radeon9250 both 32 and 16 bits output."
One thing I forgot to ask if @kas1e has a pegasos with G4. On Qemu under pegasos G3 there is no problem with SDL1. In addition SDL2 works on G3 and G4 correctly.
One last test to make sure all is checked is to run the Eemu pegasos g4 using passthrough The only person who can do this is "MartinW" because he has the right przet przet and knowledge. Currently his PC has become a parts donor for his Amiga x5000 .... when he finds the time he will check it
3. currently no sound card works on Qemu amigaos under Linux. I have to check exactly. And yes I will move the Debian partitions under pegasos to have two identical systems to compare running only with different kernels.
balaton wrote:@Maijestro I can't understand how SDL has no problems on amigaone but has on pegasos2. There's nothing different in graphics and CPU emulation between these machines, all the difference are the firmware and north bridge which should not have anthing to do with graphics so only plausible explanation to me is that this SDL version has some code that detects it's running on amigaone and does something differenetly which is missing or does not work on pegasos2. Are the sources of the SDL port that has the problem available? Maybe somebody could check what that's doing differently on amigaone.
You wrote something about Northbridge, maybe it is exactly this problem. @LiveForIt had already pointed it out at the time, but maybe you missed it, it was in another thread.
Quote:
Was there some kind of byte swap support in northbridge? Perhaps it does something wrong.
looks like you 4 bytes swaped with next 4 bytes. (8bit GFX mode only way you get correct colors)
Perhaps its swapping 8bytes instead of swapping typical 4 bytes. Pixel 1,2,3,4,5,6,7,8 resvered as 8,7,6,5,4,3,2,1
whats expected is 1,2,3,4,1,2,3,4 revered as 4,3,2,1,4,3,2,1
If writes directly into bitmap buffer instead of using WritePixelArray, strange things can happen.
Odd things can happen with JIT, for example two 32bit instruction can be combined to from a 64bit instruction, maybe assumption is not always right. A reserved 64bit operation, vs 2 x revered 32bit operations won’t actually produce the same result.
On the fast I have found only a small SDL test demo with Source http://os4depot.net/?function=showfil ... e=demo/misc/parallax4.lha on AmigaOne runs the demo without problems Pegasos2 with distorted picture. It also affects only the G4 CPU with G3 CPU works SDL under Pegasos2.
With SD2 there are no problems on both machines, but especially many older tools/programs/games made on SDL1 are not compatible anymore. It also does not affect everything that was ported with SDL1.
Quote:
Similarly speed difference between pegasos2 and amigaone cannot be optimised in QEMU as these use the same code in QEMU so if there are differences then that likely comes from AmigaOS doing something differently. The amigaone emulation is just adding a minimal ArticiaS emulation that does nothing than creating the PCI bus and forwading access to it and the board code that only creates the devices that are mostly the same as in pegasos2 so there isn't anything in there that could be optimised.
Ok, I understand, both machines use the same software (AmigaOs4.1 FE Update 2/Enhancer2.2), the differences are also minimal, noticeable when doing graphics intensive things like ScummVM/video/streaming YouTube. But since we don't have 3D acceleration anyway, it doesn't matter, it's still impressively fast.
It would be nice if we could get NVRAM support for the AmigaOne emulation so that we could permanently store the U-Boot settings, for the Pegasos2 emulation this would not be mandatory.
Quote:
Maybe you could forward Volker's original explanation to me for better understanding.(According to the picture Volker referred to network could also use addr=6 then.)
All the information I have, you can see above where he describes how to solve the problem. I have only translated it into English.Post #182
rtl8139,addr=6 does not work, the driver is recognized by AmigaOs4.1, but with it no internet connection can be established, with addr=8 however it works.
Pegasos2 and AmigaOne use rtl8139.device (53.4) with newer version the internet connection breaks on both machines.
I also noticed that the transfer rates vary a lot, I downloaded a 300MB file from Os4depot, from 960kb-9 MBit/s, but well it's an emulation.
Edited by Maijestro on 2023/10/4 9:31:15 Edited by Maijestro on 2023/10/4 9:32:57 Edited by Maijestro on 2023/10/4 10:08:09
MacStudio ARM M1 Max Qemu//Pegasos2 AmigaOs4.1 FE / AmigaOne x5000/40 AmigaOs4.1 FE
rtl8139,addr=6 does not work, the driver is recognized by AmigaOs4.1, but with it no internet connection can be established, with addr=8 however it works.
0.6.0 is the onboard 3Com ethernet chip of the AmigaOne, make sure you don't have the driver for it installed when trying to use 0.6.0 for something else.
@smarkusg The only reason a small boot partition is used is that the old U-Boot, or more likely SLB_v2, only supports ext2fs/ext3fs with 128 bytes inodes. Small partitions may still use 128 bytes inodes by default, but large ones use 256 bytes by default and you have to use "mke2fs -I128 ..." for formatting it or U-Boot/SLB can't access it. If that's OK for you could use a single 128 bytes inode ext2fs partition with everything (AmigaOne /boot/a1boot config and kernels, Pegasos2 kernels and whatever else it may need, and the / root partition) for testing.
@Maijestro Byte swap support in north bridge is not likely to cause it because:
- No such thing is emulated on amigaone and it works so it does not seem to use this feature
- It's not emulated for mv64361 (north bridge of pegasos2) but I've added warnings that you should see with -d unimp if the guest writes the bits that enables byte swap and they are not shown so it's not used by AmigaOS on pegasos2 either
- The byte swap support in mv64361 is per memory window not per device so it would byte swap all PCI devices so if it were used then either all of them would work or none of them but not most of them are working and only one screen mode is failing.
So I think this must be something within SDL 1 source itself and we would need the source of that library not a program using it to check what it does.
With same code I did not mean AmigaOS 4.1 FE but that amigaone and pegasos2 both use mostly the same code within QEMU to emulate these machines so all devices and PCI and sound and graphics are the same code in QEMU, only difference is that pegasos2 has mv64361 instead of articias where articias in QEMU is really nothing just a place holder for PCI devices, no other features of this chip are emulated as nothing seems to be used other than PCI access and IRQ acknowledge and mapping memory addresses. The vt82c686b and vt8231 south bridges emulation also share most code, the only difference is that some registers are at different address but the functionality is the same. So all that's different between these machines must come from firmware and guest code (i.e. AmigaOS versions might have differenceies) plus software might detect on what machine they are running and do some things differently. There's nothing in QEMU that would explain these differences to me.
OK that setting addr fixes the problem but I still don't fully understand what's happening. Maybe I should ask Volker then. This does not explain broken sound with via-ac97 for example so there's probably something more there. And you say that newer network driver versions are unstable also on amigaone the same way as on pegasos2 so maybe we have another IRQ problem affecting both machines or there's some problem in these drivers?
@smarkusg What's not clear is what you did to avoid the error about mounting /boot/a1boot that you got first and prevented booting. This is what others who want to install Linux might also want to know. I don't see how does changing address of PCI devices could avoid that mount error.
I think @Hans said Radeon had big endian 16bit mode so it would work with that but sm501 and newer Radeons have little endian which SDL might not detect correctly on pegasos2. But then why does it work on amigaone? Maybe @kas1e can also test pass through sometimes although he seems to be busy with a lot of things. Probably the easiest is to try to get the source for the failing SDL and see what it does.
About sound, in your video there were some errors about codec 1 missing or something. Only the output part of via-ac97 is emulated, I did not bother to implement input so in case the driver needs that then maybe it fails becuase of missing input channels or it expects the codec to be different? I think hardware can support two codecs but only one is present in these machines but I'd expect Linux to be less picky about this and find what's there and use it. In any case ES1370 should probably work as that's a PCI card or some other PCI sound card QEMU can emulate.
@joerg Do drivers bind to slot addresses? I thought they check PCI device IDs and don't even load when no such card is present so how can that be a problem?
@balaton Accessing onboard hardware like the AmigaOne 3Com ethernet chip can have some small differences to general PCI access and the 3Com driver may have a special case for the AmigaOne onboard ethernet. It doesn't even have to be the AmigaOS driver, U-Boot supports network booting and has to configure the onboard ethernet chip and may not check what's in 0.6.0, on real hardware it's not possible that there is something else, write some config or I/O data to the emulated RTL8139 making it fail later.
balaton wrote:@Maijestro OK that setting addr fixes the problem but I still don't fully understand what's happening. Maybe I should ask Volker then. This does not explain broken sound with via-ac97 for example so there's probably something more there. And you say that newer network driver versions are unstable also on amigaone the same way as on pegasos2 so maybe we have another IRQ problem affecting both machines or there's some problem in these drivers?
ac97 does not work incorrectly, but not at all under AmigaOneXe emulation, only sb128 had an incorrect sound output, which is now fixed with the correct address assignment.
Volker wrote me that he can't explain why ac97 doesn't work under AmigaOne emulation, but he is investigating the problem further.
The rtl8139 network card is the most used PCI card on AmigaNG, so I think the driver is ok. This was also written by @joerg. Unless PCI devices on real AmigaOne and Pegasos2 hardware behave differently and the driver was specially adapted for this....which is not expected under emulation. But I honestly have no idea about these things.
MacStudio ARM M1 Max Qemu//Pegasos2 AmigaOs4.1 FE / AmigaOne x5000/40 AmigaOs4.1 FE
A problem with the Pegasos2 may still be that AFAIK all PCI slots use the same IRQ. On the AmigaOne each of the 4 PCI slots (1xPCI66, 3xPCI33) uses a different IRQ, and it can even be configured in U-Boot, check for example the picture in 8. in https://www.amigaportal.cz/node/159407 but that's only the case for the PCI slots, not for onboard hardware like the VIA parts (PATA, USB, etc.) or 3Com ethernet. Not all AmigaOS PCI drivers may support shared IRQs. Of course you'd get the same problem on a real Pegasos2, but if there is such a problem the estimated 20-50 users using a real Pegasos2 with AmigaOS 4.x may never have noticed it, especially since they probably all use the onboard ethernet and not a RTL8139 PCI card.
joerg wrote:A problem with the Pegasos2 may still be that AFAIK all PCI slots use Of course you'd get the same problem on a real Pegasos2, but if there is such a problem the estimated 20-50 users using a real Pegasos2 with AmigaOS 4.x may never have noticed it, especially since they probably all use the onboard ethernet and not a RTL8139 PCI card.
hmmm...ok and I thought the Rtl8139 network card would have been a very common card on AmigaNG. Then it could well be a driver problem after all.....damn I'm confused and could explain why newer versions of this driver no longer work properly. But there is no one who can confirm it, so again it remains a theory.
So we are back to the very beginning of this problem.......
MacStudio ARM M1 Max Qemu//Pegasos2 AmigaOs4.1 FE / AmigaOne x5000/40 AmigaOs4.1 FE
@Maijestro Using a RTL8139 PCI card on AmigaOS 4.1 is common, but using a real Pegasos2 for AmigaOS 4.1 isn't. It can only be a driver problem if it's working correctly with the AmigaOne emulation but doesn't work with the Pegasos2 emulation. For the RTL8139 driver that's not the case, but it could be for other PCI drivers. Not relevant for the QEmu emulation, except for using a real PCI(e) card with pass-through, but for example Hans wrote that his Radeon HD/RX drivers don't support shared IRQs, and that's very likely the case for some other PCI drivers as well.
News feed
OS4 Feedback forum
OS4Depot Feedback forum
Amigabounty forum
OpenAmiga forum
AmiCygnix forum
ABC shell forum
AmiKit forum
Cinnamon Writer forum
CodeBench forum
Digital Universe forum
Dopus 5 forum
E-UAE forum
Gnash forum
Ibrowse forum
JAmiga forum
Odyssey forum
OWB forum
Qt forum
SmartFileSystem forum
Timberwolf forum
TouchDevice forum
TuneNet forum
Unsatisfactory Software forum
Donate
