[NSWI004] Assignment 5: Bitmap shrinks unexpectedly

[Vojtech Horky]

Hello.

Dne 06. 01. 20 v 17:25 Péťa Emmerová napsal(a):
> Thank you very much for the reply.
> 
> We treated frame allocator similarly as kmalloc, so we tried to start 
> right next to the kernel end, and reserve our first frame for the bitmap 
> and its data. (The frame starts right after kernel_end.) Then we try to 
> split memory for heap and frame allocator.
> Do we have a bug in our approach?

Not in the approach but rather in the implementation ;-). Your 
frame_init() is still rather chaotic.

Regarding the reservation of the first frame - it depends on the amount 
of available memory and kernel size: sometimes it could fit even in the 
space after _kernel_end and before the first (free) frame.

And, please, bottom/in-line post on the ML, thanks!

- VH

> 
> po 6. 1. 2020 v 17:12 odesílatel Vojtech Horky <horky at d3s.mff.cuni.cz 
> <mailto:horky at d3s.mff.cuni.cz>> napsal:
> 
>     Hello.
> 
>     Dne 06. 01. 20 v 16:47 Péťa Emmerová napsal(a):
>      > Hello,
>      > we decided to use a bitmap for frame allocator. Firstly, we
>     initialize
>      > bitmap in frame_init, and then print out the length of it, which
>     seems
>      > to be fine. We then try to allocate the frames after the bitmap. But
>      > then, when we take a look at bitmap's length during a frame
>     allocation,
>      > it's length unexpectedly drops to 16. Moreover, although all
>     frames in
>      > bitmap should be set to zero, they aren't, they contain random
>     values.
>      > We would appreciate any help or hint how to work with the heap.
>      > Sincerely, Team Datel
> 
>     This sounds like a memory corruption. Have you shifted start of the
>     frames by the size of the storage for the bitmap?
> 
>     Looking at your code (f41f9d8a0), I think you do not account for the
>     fact that kernel occupies the first few frames of the physical memory.
> 
>     BTW, you have unreachable code in frame_free() and you do not restore
>     interrupts properly.
> 
>     Hope this helps,
>     - VH
>     _______________________________________________
>     NSWI004 mailing list
>     NSWI004 at d3s.mff.cuni.cz <mailto:NSWI004 at d3s.mff.cuni.cz>
>     https://d3s.mff.cuni.cz/mailman/listinfo/nswi004
> 
> 
> _______________________________________________
> NSWI004 mailing list
> NSWI004 at d3s.mff.cuni.cz
> https://d3s.mff.cuni.cz/mailman/listinfo/nswi004
>