Bug#877735: busybox: m68k build broken due to "-Os" in CFLAGS
On 02/05/2018 11:45 AM, Chris Boot wrote:
> Is this still a problem with current gcc on m68k? I see that gcc points
> at gcc-7 (>= 7.2.0-1~) in sid and, as far as I can tell, it also applies
> to m68k.
I haven't checked. I will verify this tonight. I am currently on the
train back home from FOSDEM and have limited mobile internet only.
> To me this is really a gcc bug, and the fix belongs there. busybox has a
> real interest in remaining small on several architectures, and I really
> don't believe a workaround like this belongs in the busybox package.
I don't disagree. It is a gcc bug, of course, when a package misbehaves
when it's miscompiled. However, fixing such bugs in gcc takes a bit more
time, but they will eventually get fixed.
This is why I asked for the workaround. There is nothing wrong with
carrying a small workaround in the debian/rules file if it helps relieving
issues porters have on some architectures.
There are many packages in Debian that have such workarounds, even for
architectures like armhf, armel, mips and so on, so it's not limited
to m68k.
Packages with such workarounds include: mesa, qtwebkit-opensource-src,
mpfr4 for example. So, yes, I absolutely agree with you that the underlying
issue is a bug in gcc which needs to be fixed. But until that has happened,
it would be very helpful for the m68k porters if the workaround can be
carried in the meantime in the debian/rules file.
My suggestion would be to add a comment with a reference to the gcc
upstream bug report to the workaround so that the busybox maintainers
can easily check whether the workaround is still necessary when they
upload a new version.
I will verify the issue with the latest gcc version tonight. If it still
exists, I will finally open the upstream bug report and add the URL
with the "forwarded" keyword to this bug report. It would then be highly
appreciated if you could add the workaround with the gcc bug report
referenced to the debian/rules file for the time being until the gcc
bug has been fixed.
Thanks,
Adrian
--
.''`. John Paul Adrian Glaubitz
: :' : Debian Developer - glaubitz@debian.org
`. `' Freie Universitaet Berlin - glaubitz@physik.fu-berlin.de
`- GPG: 62FF 8A75 84E0 2956 9546 0006 7426 3B37 F5B5 F913
Reply to: