Current-Users archive

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

Re: Building PCC for "tools" is broken (missing symbol __USE)- PCC bug or NetBSD source tree error?



Then what's the best course of action? I'd just like to get a distribution with PCC instead of GCC for an old machine with limited disk space... it doesn't matter to me if PCC cannot compile all of userland. I was under the impression that PCC was able to compile a minimum working system (kernel and minimum userland) at some point in the past, but now needs to be updated so that it can compile a working distribution again.

-----Original Message----- From: Iain Hibbert
Sent: Wednesday, August 13, 2014 10:37 AM
To: William D. Jones
Cc: current-users%netbsd.org@localhost
Subject: Re: Building PCC for "tools" is broken (missing symbol __USE)- PCC bug or NetBSD source tree error?

On Mon, 11 Aug 2014, William D. Jones wrote:

2. When attempting to build userland with the following mk.conf, I receive an
error regarding a missing libgcc:
COPTS=-Os -Wno-error=uninitialized -Wno-error=maybe-uninitialized #This is
irrelevant, more than likely
MKPCC=yes
MKGCC=no
HAVE_GCC=4 #Define if MKGCC=no
#HAVE_PCC=1

Essentially, I want to use GCC to build a userland with PCC only for my 486 machine to save space and compiling time.I think this is a dependency error that the build system misses; it seems as if MKGCC isn't honored for certain Makefiles. Does anyone familiar with build.sh internals have any idea on how
to fix this?

I rather suspect this is unknown territory, with half-grown features
lying in wait to trap the unwary.

In the beginning, PCC support started to be added in the style of the GCC
support, but I think this is not fully tested (since PCC cannot yet build
everything). Latterly, Joerg has been working on LLVM support and he has
added the AVAILABLE_COMPILER functionality, but there are still remnants
of the previous setup, and I am not sure if there is yet a final design
for providing a NetBSD release with a different compiler, though as far as
I know, LLVM/Clang should be capable enough (there are still some
UNSUPPORTED_COMPILER.clang instances in the GCC code)

regards,
iain

--
William D. Jones
Rowan University | ECE | 2012
Member IEEE
Member Tau Beta Pi
thor0505%comcast.net@localhost
Message sent using 'Windows Live Mail' client.

Home | Main Index | Thread Index | Old Index