Dear Aurélien,

first of all, thanks a lot for your kind and proactive mail!

You actually made my morning this day. 😀️

On Sun, 2024-09-01 at 23:29 +0200, Aurélien COUDERC wrote:
> we have Qt6 failing to build on m68k.
> This is kind of good news because it means it is being built. 🙂
> It was not the case for Qt5 that had a build dep on firebird,
> missing on m68k and making in not buildable.

Yes, Firebird has always been a problem on m68k due to the very
same 16-bit alignment problem that Qt is stumbling over as well.

> So now we have it building and failing to build. The build failure
> message [1] made me think of previous discussion where IIRC you
> explained that the m68k architecture is the only one with a memory
> alignment of 2 bytes / 16 bit.

Yes, indeed. It is actually possible to tell GCC to use 32-bit alignment
on m68k with the help of the -malign-int flag [1] and we have been planning
to switch the m68k port to 32-bit alignment by default to fix these alignment
issues once and for all.

This will break the ABI which means we will have to rebuild the whole package
archive on m68k. But on the other hand, this will enable a lot of packages
on m68k which previously didn't work. So, in the long term, this particular
issue will be fixed once and for all.

> Like most teams we try to avoid carying too many downstream patches but it
> would be cool to have Qt built on all archs. So if a maintainable patch can
> make m68k build we (=Patrick) would be willing to have it in Debian.

This is a really kind offer, thanks a lot! I will create and test a patch and
send it with a bug report later this week. I am currently busy with my day job,
so I cannot look into immediately, but I will try to send it as soon as 
possible.

And, hopefully, these kind of kludges won't be necessary anymore in the 
foreseeable
future.

> Thanks for your help !

No, thank you for your kind words and offer!

Adrian

> [1] https://gcc.gnu.org/onlinedocs/gcc/M680x0-Options.html

-- 
 .''`.  John Paul Adrian Glaubitz
: :' :  Debian Developer
`. `'   Physicist
  `-    GPG: 62FF 8A75 84E0 2956 9546  0006 7426 3B37 F5B5 F913

Reply via email to