Re: [MiNT] 68000 kernel compile

[Paul Wratt <paul.wratt@gmail.com>]

2009/12/7 Vincent Rivière <vincent.riviere@freesbee.fr>:
> Paul Wratt wrote:
>>
>> while doing a "make all" for XaAES I noticed a lot of warnings during
>> the first part of the build process, which is the 68000 kernel build,
>> and wondered if this would not lead to an explanation as to why MiNT
>> currently does not run on an ST (according to other posts I have read,
>> and one bug report)
>
> Personally I have very few knowledge of the FreeMiNT kernel.
>
> I tried once to get it work on Steem, it crashed with bombs. But maybe I did
> something wrong. Or something is incompatible with Steem hard disk
> emulation. This certainly requires some investigation.
>
> About the warnings, every occurrence is specific. It may be a harmless
> indication, or a big issue. Every warning line has to be examined
> separately.
>
> A very evil warning is "dereferencing type-punned pointer will break
> strict-aliasing rules" in GCC 4.x. Is is usually harmless, but not always.
> The strict-aliasing rules lead to better optimization, but can break some
> old code. If unsure, this optimization can be disabled with
> -fno-strict-aliasing.
>
> Without any clue, we can't establish a direct link between the compilation
> warnings and the ST crash.
>
> --
> Vincent Rivière
>

I will get to kernel "stuff" using debuggers etc, but that will be
after some solid HighWire development and bug fixing, which will
require use of debugger. I am looking forward to using debugger in
STEem, it will be a fist for me, using debugger on ST as opposed to
using one on ARAnyM under MiNT. The last debugger I looked at was
built into PacifiST, and before that Devpac on an Amstard..

I see tho, that Alan is "on the job", so this is a good thing. BTW
those warnings, my interpretation of them prompted this thread, not
that they were serious, but that they could be, and I did not see
similar ones on other CPU builds.

Paul