AmigaOS 3.2 kickstart ROM crashing with NetBSD boot blocks

[a4k-oerx]

Originally Posted by Chain-Q

As I wrote at the start of post #5, the Alert was generated by bootstrap in the ROM ..

Thanks for clarification. So the initial statement "alert is not generated by any of the kickstart components" seems to be wrong.

[Chain-Q]

Originally Posted by a4k-oerx

So the initial statement "alert is not generated by any of the kickstart components" seems to be wrong.

As far as I can tell, yes. As I wrote already, the bootblock's code contains no explicit calls to exec.library's Alert function anywhere (it doesn't even have Alert() declared it its own headers/stubs it uses to interact with AmigaOS calls), this means any Alerts must come from code surrounding the bootblock, which this early during boot can't really be anything else but the Kickstart, or some other ROM that might be present in the system.

[abs]

The change is now in NetBSD HEAD, and should get a pullup into the upcoming netbsd-10

(Link to change on NetBSD git mirror)
https://github.com/NetBSD/src/commit...e4fc3e9adbbe80

Full release builds of NetBSD with the change (plus Roc Vallès's new device-streams binaries) are at:

• https://sync.absd.org/amiga/current/
• https://sync.absd.org/amiga/netbsd-10/

Official builds with the changes will start to show up at http://nyftp.netbsd.org/

Thanks again to Karoly (and for the others tracking down the fix)

@a4k-oerx and anyone else interested in having a play with NetBSD/amiga - I think its safe to say we're actively working on making it easier to get started with NetBSD, both on real and emulated hardware - please drop a message in here or port-amiga@netbsd.org with any issues you hit (including "there are just too many damn steps!") as it gives useful insight into the best places to look to make improvements form the perspective of someone who is not already familiar with the setup

[abs]

Clarification - sync.absd.org has the complete distribution less X11 sets (to speed up building)

[Thomas Richter]

Originally Posted by a4k-oerx

Thanks for clarification. So the initial statement "alert is not generated by any of the kickstart components" seems to be wrong.

No, it is right. The alert is (was) generated by the NetBSD code not being able to get access to the console.

[Chain-Q]

Originally Posted by Thomas Richter

No, it is right. The alert is (was) generated by the NetBSD code not being able to get access to the console.

No, it is wrong. The OS generated the Alert, because the bootblock returned with non-zero exit code. You literally wrote the Alert is "not even formatted or allocated correctly" but the bootblock has no code at all to generate any sort of alert, it just returns to the caller code (i.e. the OS) with non-zero, if it can't open console.device (or any other similar error).

But I'm hoping we're not arguing on schematics, i.e. "the bootblock generated something" vs. "the bootblock caused the OS to generate something"...

[a4k-oerx]

Originally Posted by Thomas Richter

No, it is right. The alert is (was) generated by the NetBSD code not being able to get access to the console.

Can you show the _LVOAlert() call in NetBSD code which generates this "not correctly allocated or formatted" alert? That would be helpful, thanks.

[karloch]

Thanks everyone for the work done fixing the NetBSD bootblock. I have just recorded a video showing it working well with Kickstart 47.111.

[ Show youtube player ]