Your biggest OSdev /facepalm moment

[pcmattman]

nekros: I was about to say "Just find + replace" but then I remembered ASM comments are usually semicolons

[Solar]

Hmmm...

One for the regex masters: search all semicolons that

• don't end the line and
• don't follow a "for("

...and replace them with //. If that doesn't do the trick, your C++ style sucks.

[nekros]

It was already fixed when I wrote this.

[neon]

The several times that I posted looking for help only to find and solve the problem 5 minutes later after posting.

[nekros]

Funny when you ask something out loud you end up answering it the moment you finish your sentence.

[nekros]

I think I just experienced the biggest one on my record yet. Bootloader, was being screwy because I set sp to the start of the stack...

[Ferrarius]

spending over an hour trying to locate a mistake in the IDT or any other code, only to find out I didn't correctly remap the second PIC.

[pcmattman]

The several times that I posted looking for help only to find and solve the problem 5 minutes later after posting.

I'm in that boat. It's like posting the question is a catalyst for the answer to come...

[PHPnerd]

Some time ago, with my previous PMM, I got always no pages. Have been searching for 2 weeks. Problem was in find free frame function

Code: Select all

uint32_t *bitmap; // is initialized, dont worry :P
uint32_t bitmapSize; // in bits
uint32_t i = 0;
for(i = 0; i <(bitmapSize >> 3); i++)
{
      if(bitmap[i] != 0xffffffff) // have != here, so when some bit is empty, it skips it!!!, and when all full, next statement checks where empty bit is, but its nowhere.
            continue; // fast break if full
      if() // do rest stuff, search through 32 bit finding empty one
}

really stupid

// PHPnerd

[earlz]

using code insertion on my IRQ handlers and injecting 'mov al,0x...' after the 'popa'

[NickJohnson]

I setup a special linking script and code that would collect all of the init code in my kernel and free it after everything was set up. Unfortunately, in my overzealous marking for deletion, I made it get rid of the main loop, so when the freeing function returned... kpow! Somehow this took me three hours to figure out.

Or, equally dumb: I had my GDT being loaded before I switched on paging, and it was still in lower memory. I could unmap the kernel image that was in lower memory, but whenever I did an interrupt, it would reload the segments, not find an actual GDT, and triple fault. I kept looking through my memory manager to try and figure out where the heck it was faulting, when the actual problem was in the first ten instructions in the kernel.

[alethiophile]

I initialized some space for a disk address packet to 0xdeadbeef so I could find it in a hex editor, then wondered why my read_sectors kept saying 'LBA out of range'.

[Troy Martin]

Ye gods, this could be a five-page post

One of them would be accidentally adding a * before a variable name. Unexpected output was everywhere. In the form of divide errors. Unfortunately I don't have a screenshot...

[pcmattman]

The other day:

Code: Select all

void assert(bool b)
{
    ERROR("Assertion failed.");
    Processor::breakpoint();
}

Took me a while to figure out why all my assertions were failing