PANIC! 00000001

[ArsenShnurkov]

When I am trying to start "Run-HelloWorld.sh" (on debian),
the qemu launches,
SYSLINUX 3.72 says "Booting: MBI=0x00010090, entry=0x00400030",
then pink-over-blue text "PANIC! 00000001" appears at the top left corner of qemu window.

[ArsenShnurkov]

this sample code also gives panic message - http://pastehtml.com/view/ea137scgl.html

Mosa.Kernel.x86 VirtualPageAllocator.Reserve(67108864) have pages = 254 (and 16385 was requested)

TotalPages(1040640) < ReserveMemory(33554432)

i added command line argument -m 512M into the ./Run/Run-QEMU.sh

now the system reboots in line
https://github.com/mosa/MOSA-Project/blob/master/Source/Mosa.Kernel.x86/PageTable.cs#L42

[charsleysa]

Hi @ArsenShnurkov

I've tried to replicate your issues but cannot replicate them on Windows.

Can you please make sure that you have the latest copy of the repository.

Can you please make sure that your setup of QEMU matches the Windows setup.

[tgiphil]

I was able to re-create the problem on Windows by copying the Linux/Mono generated assemblies to Windows.

The test suite runs fine except for a few failed cases related to double/single floating point. But this would not explain the crashes.

[charsleysa]

@tgiphil could it possibly be something to do with Endianess?

If I recall correctly the field value emitter in the Metadata stage doesn't use an endian aware emitter since I kept that code mostly the same.

[ArsenShnurkov]

It is definitely related to Page Tables & Directory filling.
I rewrote Mosa.Kernel.x86 PageTable class
and it works for me now.

[tgiphil]

Great! Check it in and I will test on Windows.

[ArsenShnurkov]

it looks something like this now:

https://github.com/ArsenShnurkov/MOSA-Project/blob/fdc1b6e00a93a3f76a42cd717b6e8982705a44ea/Source/Mosa.Kernel.x86/PageTable.cs

[tgiphil]

This change exposed what I believe is a bug in the register allocator. It attempts to generate a move between two memory locations - which is not feasible on the x86.

I'll try to hunt it down and fix it tonight.

[tgiphil]

@ArsenShnurkov - I believe the bug has been fixed in ImproveRegisterAllocator branch in my repo. The register allocator was refactored. At the moment the very basic (no thrills) register allocator has been implemented and tested successfully on Windows. However, can you please test it on your Linux platform?

[ArsenShnurkov]

works for me (I just tested your branch as is, without any patches)