[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [MiNT] zview



Jo Even Skarstein wrote:

> On Mon, 17 Oct 2011 10:26:51 +0200, Helmut Karlowski
> <helmut.karlowski@ish.de> wrote:
>
> > True, but sometimes you find old bugs or poor code by implementing new
> > things,
> > but the main priority is of course to make it rock-solid, and that's why
> > I think it's not very good that XaAES uses kmalloc/kfree for nearly every
> > little chunk of memory. I only may help to debug the hard way ;)
>
> A big problem with MiNT and XaAES is that it's a lot of work to find out
> how things works. There is no document that explains the purpose of each
> source-file, how the build works, the design of the object-renderer, how
> messages are implemented, the internal proc/app-structs... I would love to
> contribute to XaAES and the kernel, but it's incredibly hard to understand
> how everything connects. If you understand this it's much easier to track
> down bugs or add features.

That's true of course. Without aranym and writing tons of debug-output it would be
nearly impossible (for me) to get into it. I still don't always know
what is done by what thread in XaAES and so on.

> > Another bug may lie in XaAES executing USERDEFS in user-mode, which is
> > also complicated.
>
> I like that. XaAES is pretty good at recovering from bugs in USERDEFs. As
> a programmer I find this very handy.

Sure it's generally better. But there's at least one app (papryus5-demo)
that seems to rely on USERDEFS being executed in super-mode and then
later the kernel mixes up super- and user-stack when checking
stack-boudaries, and kills it. But this affects only this app, so it's
not a big issue.

-Helmut