X1000 CFE Patches: Testers Welcome (Fear not: real-time, no reflashing) v.02 NEW!

@Hypex
I remember how you love to mess things up in all ways so no surprise :) now lets step by step: if you rawwrite your iso on usbstick, thn you should do "boot -fs=iso usbdisk0:amigaboot.of". If you format usb with ffs then you should do "boot -fs=amigafs usbdisk0:amigaboot.of". If you do it wrong you will have "couldnt init filesystem" and co. If you not run usb patch then you will have no bootable usb device found. If you create stick by rawwrite way from iso, then -fs=iso, if ffs then be sure it have bootable flag and -fs=amigafs is used.

Just make it all clear and step by step. Be sure you have prepare stick correctly before trying to boot from.

Be sure you reattach your stick too after usb patch applied too.

@kas1e
I tested usb_fix.elf with USB DVD and install .iso:



it works, amigaboot.of is loaded, but in amibaboot.of bootloader there is not visible USB CD - only HDDs.
I can load kickstart from DH0 or DH0x, and than boot from USB CD.
But still I must have kickstart on disk.



test of USB flashdisk - wit FFS filesystem:


the same behavior like before - USB Flashdisk not visible in amigaboot.of menu.

It is probably limitation of amigaboot.of
Or did I some mistake and booting fully from USB without disk is possible?

@sailor
Something wrong, boot fully from USB and from USB CD-ROm with patch fully possible now, of course, that the whole point of it.

See how it looks like when i "boot -fs=iso usbdisk0:amigaboot.of" from USB CD-rom and x1000 installation ISO :



So first one is my CD.

The same if i use stick after, which formatted as FFS and _marked to be bootable in mediatool box of course(!)_ :



So this definitely should works as you can see, but something still not.

My bet that on your 2gfx setup again some addresses shifts, and node didn't creates correctly.

Can you plz check that:

1). reboot clean.
2). run no patches, just single "usb_fix.elf"
3). insert USB CD and Stick.

At this point on serial you should see [USB] unit 0 attached and [USB] unit 1 attached amount others (those are prints from my attach() hooks).

Now, plz do "show devtree", and upload it somewhere (posting there will be too huge and bork the forum). Or if you will post it there, post then inside of quote or code blocks as always.

Thanks !

@Sailor
In your fist output i find something strange when you use USB CD-Rom (i.e. the first output):

1). you run sata_fix, then run usb_fix : that ok.
2). Then i see that:



It says that you connected usb mass storage. Not a CD-Drive ! Then you do : dir -fs=iso usbdisk1: , but that strange: if you connect just a USB stick, it then "usbdisk0:" , not "usbdisk1:".

Maybe you simple mess the outputs ? And the first one was for usb-mass storage instead, while second one was for USB CD-Rom ?


In any case, try to do plz just that:

1). power off/ power on.
2). run sata_fix / usb_fix
3). attach FFS formatted usb stick with created boot system like written there : https://g-a-p.me/inside/x5kbootstick.html. "boot" flag is must in mediatoolbox.
4). You should see:

USBMASS: Unit 0 connected
[USB] unit 0 attached


That mean usbdisk0.

5). do "boot -fs=amigafs usbdisk0:amigaboot.of".

At this point you should see usb partition there.

Then, once this work, you can try USB CD-Rom, same steps, but you connect USB CD-Rom and boot from it, and see if it have listed.

Also, for second test you can instert firstly usbstick, and then cdrom, and only then cdrom will be usbdisk1: If you will attach only usb cdrom, it will be usbstick0:.

Let me know how it goes.

@kas1e

I don't love to mess things up. I prefer to run a clean system but if some OS update breaks my system I'm forced to compensate. I'm still not over Update 2, where other people also needed to mix and match components as well before the hot fix. I had hoped Update 3 would give me a clean break, but I too found that Update 3 broke my system as well, and that was testing from a fresh install incrementally updated. I can't accept a system that will likely freeze on boot and where USB2 transfers don't work at all while inserting a stick locks up the system for half a minute. Right now Update 1 is solid for me, so until they find the rare bugs introduced in the last two updates, or I find new replacement hardware compatible, this situation won't change for me. It is not a mess I love.

However, you are certainly messing with CFE here. Do you love to mess with CFE? Why can't you leave the poor CFE alone?

In any case I tried again without any patches. CFE still could not load amigaboot.of from my isofs formatted USB. Need to go back to my notes, or forums with info, as it looks like isofs is not supported by CFE on USB media.

I used Linux to image the USB stick with OS4 install ISO. Because RawWrite sucks and now I see why OS4 guides don't even mention it sometimes.

I expect FFS will work. But it's a pain to setup. And that MS commodity can be buggy. So far as I know it cannot be scripted. Unless we have support programs that can format a USB drive as FFS without needing user intervention.

Also I found something last night. I was doing a repair on a FAT USB stick just in case it was breaking CFE. I ran GParted and it found an RDSK on the USB stick! It kept saying the checksum was wrong. This is very odd as that stick has been mounted numerous times on Linux and it knows it is FAT. GParted refused to identify it as FAT. But Disks has no issue. I must have used it as FFS in the past. So where ever I reformatted as FAT did not properly erase it. It still had a boot record. Make me wonder if CFE looks and this is part of why CFE cannot read FAT reliably.

@kas1eQuote:

I am glad that at least with some configuration it is works as expected.
According to point 2) I tested both patches: with usb_fix only and with usb_fix + sata_fix. In both cases it has the same result ( no USB Kickstart directory visible. )

I have in my standard HW config allways USB flashdisk with ext2 partition and linux kernels. It is connected with hub + keyboard to first USB port.
So in every of my attempts above there was two USB devices:
usbdisk0: .. ext2 with linux kernels
usbdisk1: .. AmigaOS boot media, or Flash or CD.
Maybe amigaboot.of search for usbdisk0 only and if found no kickstart it skips usbdisk1.

So I check it without ext2 usb and without second gfx
I will test again in the evening.

@Hypex
Quote:

So it mean you doing it wrong then.. I do have created ffs formatted usb stick, set it as bootable, and surely can do "boot -fs=amigafs usbdisk0:amigaboot.of", just without usb_patch it will not find as bootable.

@Sailor
Quote:

I tried with second gfx already: that works fine too. So that not the case.

Quote:

No no, both are ok , tested myself, it works.

But, what i see in your log:



See, it says Unit1 connected, but unit 0 attached (by me). For me it looks strange. If you have no USB attached and then simple attach one usb stick, you should have unit0 connected / unit0 attached. But there we have unit1 connected, but unit0 attached. Like, i create node in this case for wrong unit.

I tried to reproduce it like this :

I keep in one port some stick, then run patches, then insert second stick and have:



But in your log it definitely messes up, i only need to find a way how to reproduce this "unit 1 connected" but "unit 0 attached"... maybe hub cause such differences ?

@Sailor
Good new !! I am able to reproduce ! Once i attach the HUB to first port and put there stick + usb keyabord, i have exactly the same :



And of course node creates for wrong unit -> not visibly as boot media. So no need for additional test, it's bug on my own, will fix it today.

@kas1eQuote:

Thank you very much, you are great!

@Sailor (and others too of course)
And what interesting, it seems that CFE of x1000 "auto enable" only first USBs port on motherboards back?

I mean, we for long time have this issue : if the USB stick stays in the usb port other than port 1 we need to replug them, but port1 seems autodetect all fine. So i may try to find how it done, and do autodetection of some other ones (like ones connected by the case's wires to motherboard)

@kas1e
I suppose that only devices in usb port 1 is automatically detected.

In Nemo TRM i foun no point, but in X1000 quickstart guide is:


and only in X1000 Linux installation guide is some clue:

@kas1eQuote:
Cross fingers - it will be a very nice feature

@All
First page updated with new archive. All the patches updated, readme.txt included (with all the info about). In brief what was added since last in archive:

-- all the fixes we do there (Tuvok, i fixed this "auto execute" too, all the things TearsOfMe and Sailor report)
-- new commands "editenv" and "pager" (see readme.txt for details)
-- sfs2 patch - read/load/boot from sfs2 medias
-- usb patch - that one you know already (sailor plz test if all fine now)

The only one which a bit "out" now from archive is Radeon based patch, i need to think more about how to integrate it all better, so for next version probably.

Have fun and let me know how it goes, thanks ! The more tests you do, the better it will be in end :)

@kas1e

Quote:

No, it's just that I like to test a few combinations. A CD image is easier to work with and can be transferred on any OS. Plus on OS4 it's the fastest way to write all data as it avoids any filesystem slowing it down. An A1/XE also happens to support a CD image being booted from CD or USB. Even the the PC world took time to catch up with a hybrid ISO/USB image.

Anyway I went back to FFS but could not get it to work fully. Amigaboot menu loaded fine and it was able to load modules off USB. But it crashed immediately after. Someone else also had some trouble, but I recall some interrupt message. Mine doesn't get that far.

A patch log:

A boot log:

Will try latest patches I just noticed were released. BTW I'm testing USB with a USB1.1 hub as my Apple keyboard plugged into my X1000 only has USB1.1.

I also had a go at bboot. But CFE for some reason is having trouble reading FFS from USB as well. This used to work fine. I copied that and created a Kickstart.zip to test. Allowed time for copy to settle. Then reboot. The files didn't show up in CFE! No sign in dir command. I'm used to this happening in FAT because of the buggy fatfs driver, causing people everywhere to fat shame CFE, but not FFS. It's gone whack! No idea what the issue is, all the other files on there can be seen. But copying new files can't be seen by CFE. I thought DOS7 was supported?

It got more bizarre. Despite not seeing the Kickstart.zip it was able to load it in with the ramdisk command. But refused to work with load command. Even then ramdisk wasn't correct as it kept returning address as 0x24000000 despite me specifying address. But it butt refused to load bboot. No sign anywhere! Even added an extension in case CFE only accepts extended files. Gave up and just loaded if off my HDD. Which just like my USB had the file copied to it right before a reboot. It loaded it but crashed instantly.

Quote:

I never used to see it. Because I had an Apple keyboard with USB ports I just used to plug my sticks in that. So I had always been able to easily use USB sticks in CFE without realising or had forgotten about the limitations. But USB was detected in one front case port. However, it never worked. It indicates it mounted it but it never does. Even after the patch it doesn't work. I had a usbdisk0 inserted so plugged in a usbdisk1 but it couldn't see any files on it. Well, in the case CFE could read it at all, since FAT32 is so unreliable.

@Hypex
Sure you use CORRECT version of CFE as noted in readme and in first post in this topic ? The one which "2012_june_08". That memory patches, if you use different version all addresses will be wrong.

Also are you copy to FFS stick correct system files from which you actually can boot ? If it pure x1000 iso copy, you may have no luck because of HD/RX issues, etc. You should copy your working system partition from which you 100% can boot till working workbench if you ever have one in last years.

And:



Wtf ! :) you have 15 boot partitions ?:) looks like memory corruption ?

Quote:
If it works before, and didn't now, then again something with your hardware or the things you do. I can normally read and boot from FFS USB stick for sure, just without patch, i at least can read from FFS Stick with no problems too.

Your issues with "no files" on FFS can be simple maybe you just after copy do reboot too fast, and USB (especially old ones) need some seconds to cope with copy.

Quote:

While load command suck (at least ith combination of amigafs) , but it was noted before : you need to specify in -max a size of the binary 1:1 , then it will work. So -max should be called -size, like that:



(And of course bboot should be the one Balaton post in the other thread, not the 0.8 one..)

I mean, you need one time to note every quirks of CFE to some text file and used to it :) It's all the same bugs and quirks which you need to remember one time and used to them :)

Because for now, its more looks like you in total mess even without patches and everything working/not-working and you don't know why and when. At least that sounds and looks like that. Adding on top of this mess a hacking-firmware patches is call to more mess, for sure :)

@kas1e
Test #1:
usbdisk0:ext2, usbdisk1:ffs
ALL OK:

@kas1e
Test #2:
usbdisk0:ext2, usbdisk1:DVD
ALL OK:

@kas1e
Test #3:
usbdisk0:ext2 - in first port +hub +kbd
usbdisk1:DVD usbdisk2: FFS flash - both in front USBs:


- all devices was inserted before usb.elf run, but devices was not detected
- then I reinsert DVD, USB DVD is visible and redable
- then I reinsert FFS flash, not seen in show devices, but visible on show usb
If I dir it, there is error:

But - overall this patch is very nice improvement, I think that to fix every possible combination is not necessary. Great thing is, that we can boot from USB!

I will test the devices connected to read ( motherboard slots )