When your OS goes crazy - Screenshots

[Vijfhoek]

Something went wrong with my printf implementation...

[BrightLight]

It's because I forgot one line of code:

Code: Select all

mov byte[is_program_running], 0

[jnc100]

Significant interleaving of serial port log messages whilst I was stress-testing the scheduler:

Code: Select all

vfmoLodfs./
tyggs.vf
es:r. Logmoengetr: dferenisoterng ins_messsfag .rmectagssaore ge y:loloop
op
voofstfs: ent ent: ererining messag Moge lunootp
m: esmosauntingg te loop
ysos.rootfs to /

I should probably do something about that.

Regards,
John.

[hgoel]

Not really sure why this is happening, something to do with my double buffering implementation


EDIT: I eventually figured it out, main issue was the size of the kernel after including that wallpaper data as a header exported from GIMP, it'd go from 17kb to 8MB
EDIT2: Actually seems like the problem might be something deeper than that, everything is fine if I copy the translated GIMP output into an array first and then just copy that to the framebuffer.
EDIT3: Finally found the problem, buffer overflow because GIMPS macro assumes that data will only be read once.

[Quaker763]

I have this strange suspicion it didn't actually allocate a block....

[hgoel]

Well, setup multithreading, AHC and EXT2 parsing since my last post so I was trying my hand at loading an image from disk, this is the spectacular close but not there yet outcome:

[tuxie]

When I finally managed to compile my kernel from a year ago again:

[sortie]

Editing console rendering code always goes wrong. Today the redraw logic (used for scrolling and such) only updated a fourth of the screen, while the normal single-character rendering code continued to work.

[Vercas]

This photo is a bit old, but it shows quite accurately what happens when you don't check the framebuffer type (pixels or chars) that GRUB gives you:


On the other hand, it means that I used the rest of the info correctly.

[abhoriel]

I was adjusting my panic() routine to send NMIs to the other cores causing them to also dump stack traces.
as they all got the NMI at the same time they all dumped their stack traces at the same time, they got interleaved

[ostylk]

Hmm.. Looks fine.
Let's test out my new scroll() function

... fail

[Vercas]

A couple of days old, it's a test of my object allocator with SMP.
Unsynchronized serial output FTW.

[Awe2K]

When something really went wrong:
Printing text when SUDDENLY interrupt arrives

[Techel]

ThiHellos iHellos HelloHelloindeeHellod verHelloy fuHellonny

[nicknytko]

bug in my printDec( ) caused an infinite loop, and also a cool gradient