gm2
[Top][All Lists]
Advanced

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

Re: Building gm2 devel branch on gcc-12


From: Gaius Mulley
Subject: Re: Building gm2 devel branch on gcc-12
Date: Wed, 16 Feb 2022 12:58:31 +0000

[oops forgot to reply to the mailing list]

there have been a number of bugfixes to Makefile.in in the gcc-12 branch of gm2 (addressing errors reported via Solaris platform testing).

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101389

Hopefully the parallel build would now work?



On Wed, Feb 16, 2022 at 1:47 AM john o goyo <jog37@riddermarkfarm.ca> wrote:
Greetings, Gaius.

On 2022-02-15 17:51, Gaius Mulley wrote:
> oh wow - I had no idea it was taking that long to compile.   On the
> raspberry pi 4 (8GB) (filesystem on ssd over usb3) using make -j 4 a
> build and regression test for --enable-languages=m2,c,c++,lto takes
> over 24 hours and I thought that was slow.  Removing lto reduces build
> times - but stresses gcc less.  Maybe make -j 4 is non optimal as well

It was only compiling gcc/gimple-match.cc that took so long.

Another oddity that I have yet to investigate is that make -j<anything>
failed on the gm2 portion on Solaris boxes but sequential make succeeded.

Sincerely,
jog


>
> regards,
> Gaius
>
>
> On Tue, Feb 15, 2022 at 10:09 PM john o goyo <jog37@riddermarkfarm.ca
> <mailto:jog37@riddermarkfarm.ca>> wrote:
>
>     On 2022-02-15 14:21, Gaius Mulley wrote:
>     > john o goyo <jog37@riddermarkfarm.ca
>     <mailto:jog37@riddermarkfarm.ca>> writes:
>     >
>     >> Greetings, Gaius.
>     >>
>     >> On 2022-02-15 12:25, Gaius Mulley wrote:
>     >>> john o goyo <jog37@riddermarkfarm.ca
>     <mailto:jog37@riddermarkfarm.ca>> writes:
>     >>>
>     >>>> Greetings, all.
>     >>>>
>     >>>> Now the build has stopped as follows.
>     >>>>
>     >>>> gmake[3]: Entering directory '/home/build/opt/gcc/bld/gm2/gcc'
>     >>>> unset CC ; m2/boot-bin/mklink -s --langc++ --exit --name
>     mainmcinit.c
>     >>>> ../../../src/gcc-git/gcc/m2/init/mcinit
>     >>>> m2/boot-bin/mklink: Shared object "libstdc++.so.6" not found
>     >>>>
>     >>>> Note:
>     >>>>
>     >>>> duplex$ lf /usr/lib/libstdc++.so*
>     >>>> /usr/lib/libstdc++.so@ /usr/lib/libstdc++.so.9@
>     >>>> /usr/lib/libstdc++.so.9.0
>     >>>>
>     >>>>
>     >>>> I do not understand this.  Should mklink not accept libstdc++.so
>     >>>> rather than the specific libstdc++.so.6?
>     >>>>
>     >>>> Sincerely,
>     >>>> john
>     >>> Hi John,
>     >>>
>     >>> that is odd - mklink is now built by the C++ compiler (a
>     recent change
>     >>> from around a month ago).  Which C++ compiler built mklink on your
>     >>> system?
>     >> duplex$ g++ -v
>     >> Using built-in specs.
>     >> [...]
>     >> gcc version 7.5.0 (nb4 20200810)
>     >>
>     >>
>     >>>     Any chance you can post the build log?
>     >> Of course but where is it?  #6-)
>     > ah yes :-).  Is it possible to restart the build and redirect
>     stdout and
>     > error to a file - it might be huge but could be useful to
>     eyeball the
>     > build commands for mklink and the above (possibly!),
>
>     Will do.  (Incidentally, after two days, gimple-match.cc was still
>     compiling.  I compiled by hand, removing the optimization, and went
>     well.  This must be a worst-case scenario. #6-)
>
>     Sincerely,
>     john
>
>     >
>     > regards,
>     > Gaius
>
>


reply via email to

[Prev in Thread] Current Thread [Next in Thread]