Bug#419035: rott: crashes on start

Fabian Greffrath · External Since: Aug 03, 2007 Posts: 3

Dear ROTTers,

last weekend I finally found the time to implement what Darren Salt (and
several 64bit-porting documentations the web) suggested and replaced all
unsinged and signed long variables in the ROTT source code with the
corresponfing (u)int32_t variables. Please find the dpatch at:

http://www.geocities.com/fab666_2000/10-32bitisms.dpatch.zip

You need to copy it into the rott-1.0/debian/patches directory and
append '10-32bitisms' to the debian/patches/00list file.

Please rebuild rott on AMD64 and try if it works for you. Please check
especially if sound and writing/loading of savegames work.
If it happens to be still buggy I highly appreciate further help!!!

Thanks for testing!

Cheers,
Fabian

--
To UNSUBSCRIBE, email to debian-bugs-rc-REQUEST.RemoveThis@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster.RemoveThis@lists.debian.org

Fabian Greffrath · External Since: Aug 03, 2007 Posts: 3

Robert Isaac schrieb:
> The full version of the game was playable on Sarge without issues
> (32-bit compiled from upstream CVS). The breakage seemed to happen
> when compiled on an Etch system (if that's helpful) this is both the
> Debian package and Upstream's CVS.
Sorry, but I don't know what you want to tell me...

The "full version" is playable on Sarge when taken from upstream CVS.
How about the Shareware version (or what do you mean by "full version")?
Have you tested the Debian package on amd64-Sarge, too?

What does "32-bit compiled" mean?
Of course ROTT runs on 32-bit systems, we are talking about problems on
64-bit platforms.

However, are you sure you are running an "amd64" flavour kernel in your
Sarge install?!

Cheers,
Fabian

--
To UNSUBSCRIBE, email to debian-bugs-rc-REQUEST.DeleteThis@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster.DeleteThis@lists.debian.org

Fabian Greffrath · External Since: Aug 03, 2007 Posts: 3

Robert Isaac schrieb:
> There is no difference between the full retail version and the
> shareware version, it is just a compile switch that allows the loading
> of more wad files.
>
I know, that's why I was wondering why it should work with the full
version but not with the shareware version which is shipped in the
Debian package.

> I had compiled the game in a 32 bit chroot on my sarge amd64 box.
>
It is *supposed* to work in a 32bit chroot even on a 64bit system.
> The last time I tried the Debian package (and upstream's CVS) they
> both crashed on startup in a 32 bit chroot AND in the 64 bit
> environment this was on Etch.
>
This is however strange...

--
To UNSUBSCRIBE, email to debian-bugs-rc-REQUEST DeleteThis @lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster DeleteThis @lists.debian.org

Dmitry V. Rutsky · External Since: Mar 26, 2007 Posts: 6

On Monday 06 August 2007 08:59:59 am Fabian Greffrath wrote:
> thank you very much for debugging!
>
> I have merged both of our patches and made them available under:
> http://www.geocities.com/fab666_2000/10-32bitisms_v2.dpatch.zip

Good, although I won't have time to check that out for another couple of
weeks.

> This should be a good starting point for further debugging. Wink

>
> > I've got other things to do for the time being, so I don't plan to see
> > into these issues any time soon. Hope somebody else will do that Smile

> > By the way, GCC just screams about these issues in it's warnings, like
> > ===============================================
> > rt_actor.c: In function â€˜FindAddressesâ€™:
> > rt_actor.c:1004: warning: cast from pointer to integer of different size
> > rt_actor.c:1012: warning: cast from pointer to integer of different size
> > ===============================================
> > Probably fixing all these warnings in the first place will solve the
> > issues.
>
> However, do you consider the game playable?

Perhaps, but I'm not playing until music and game loading is fixed.

Technically this bug could be closed as the game no longer crashes on start
with these patches, had not Darren retitled it as having to do with
32bitisms. I don't think there's an awful lot to fix, it probably just needs
fixing all those screenfuls of GCC warnings with proper type changes, as
Darren pointed out. GDB just makes it easy to do incrementally in a manner
that is less likely to break code, and more fun as the exact moment of
corruption can be traced.

May the source be with you,
--
-- Dmitry Rutsky