Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > muc.lists.netbsd.tech.toolchain > #3631 > unrolled thread

re: gcc 16 bootstrap, stage 2 and 3 differ

Back to article view | Back to muc.lists.netbsd.tech.toolchain

This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by below is the oldest one visible, not the original post.

Contents

  re: gcc 16 bootstrap, stage 2 and 3 differ matthew green <mrg@eterna23.net> - 2026-08-16 08:43 +1000
    Re: gcc 16 bootstrap, stage 2 and 3 differ Thomas Klausner <wiz@netbsd.org> - 2026-08-16 19:51 +0200
      re: gcc 16 bootstrap, stage 2 and 3 differ matthew green <mrg@eterna23.net> - 2026-08-17 10:41 +1000
        re: gcc 16 bootstrap, stage 2 and 3 differ matthew green <mrg@eterna23.net> - 2026-08-17 13:22 +1000
          Re: gcc 16 bootstrap, stage 2 and 3 differ Thomas Klausner <wiz@netbsd.org> - 2026-08-17 09:37 +0200
            re: gcc 16 bootstrap, stage 2 and 3 differ matthew green <mrg@eterna23.net> - 2026-08-17 19:11 +1000

#3631 — re: gcc 16 bootstrap, stage 2 and 3 differ

> >   DWARF version:   5
>
> NetBSD's system compiler defaults to DWARF 4 because the CTF tools don't
> support DWARF 5--everyting else, binutils, gdb, etc. work fine with DWARF 5.

yeah - this was the only thing that i had any specific guess
about, i haven't tried gcc 16 yet.

would be really nice if someone updated the ctf tools -- the
upstream one supports dwarf5 i understand, but this should
not really matter for gcc 16 -- same as it shouldn't have
(and seems to not) mattered for gcc 15, etc.

as a random hack, try forcing "-gdwarf-5" into HOST_CFLAGS,
which would make the stage1 compiler use the same defaults..


.mrg.

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-admin@muc.de

[toc] | [next] | [standalone]

#3632

On Sun, Aug 16, 2026 at 08:43:25AM +0100, Matthew Green wrote:
> as a random hack, try forcing "-gdwarf-5" into HOST_CFLAGS,
> which would make the stage1 compiler use the same defaults..

I tried

Index: Makefile.common
===================================================================
RCS file: /cvsroot/pkgsrc/lang/gcc16/Makefile.common,v
retrieving revision 1.1
diff -u -r1.1 Makefile.common
--- Makefile.common	10 Aug 2026 21:41:34 -0000	1.1
+++ Makefile.common	16 Aug 2026 17:50:15 -0000
@@ -94,7 +94,8 @@
 # GCC 16.1.0's stage2/stage3 bootstrap comparison currently fails on
 # NetBSD/x86_64, so disable it. Revisit on updates.
 # https://gcc.gnu.org/pipermail/gcc-help/2026-August/145120.html
-CONFIGURE_ARGS+=	--disable-bootstrap
+#CONFIGURE_ARGS+=	--disable-bootstrap
+CFLAGS+=		-gdwarf-5
 .endif

 .if ${MACHINE_PLATFORM:MNetBSD-*-i386} || ${MACHINE_PLATFORM:MLinux-*-i386}

but that didn't change a thing.

I'm not sure what should have been the outcome, because the stage2
file was already in Dwarf 5 format, see diff:

--- /tmp/stage2.dwarfloc        2026-08-16 17:23:10.166040071 +0000
+++ /tmp/stage3.dwarfloc        2026-08-16 17:23:20.726532494 +0000
@@ -1,10 +1,10 @@

-/tmp/lang/gcc16/work/build/stage2-gcc/c/c-decl.o:     file format elf64-x86-64
+/tmp/lang/gcc16/work/build/stage3-gcc/c/c-decl.o:     file format elf64-x86-64

 Contents of the .debug_loclists section:

 Table at Offset 0
-  Length:          0x2665e
+  Length:          0x2667b
   DWARF version:   5
   Address size:    8
   Segment size:    0


Cheers,
 Thomas

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-admin@muc.de

[toc] | [prev] | [next] | [standalone]

#3633

hm, i can't reproduce this on my slightly older -current
(april), but recent pkgsrc.  that is a ryzen 3950x.

i'll try a newer -current next.


.mrg.

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-admin@muc.de

[toc] | [prev] | [next] | [standalone]

#3634

matthew green writes:
> hm, i can't reproduce this on my slightly older -current
> (april), but recent pkgsrc.  that is a ryzen 3950x.
>
> i'll try a newer -current next.

this one worked too.  fresh -current update on my test box,
building gcc16 worked.

i did *not* build a completely fresh pkgsrc, only lang/gcc16
itself updated, so i'll try a clean fully frseh pkgsrc too.


.mrg.

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-admin@muc.de

[toc] | [prev] | [next] | [standalone]

#3635

On Mon, Aug 17, 2026 at 01:22:09PM +0100, Matthew Green wrote:
> matthew green writes:
> > hm, i can't reproduce this on my slightly older -current
> > (april), but recent pkgsrc.  that is a ryzen 3950x.
> >
> > i'll try a newer -current next.
> 
> this one worked too.  fresh -current update on my test box,
> building gcc16 worked.
> 
> i did *not* build a completely fresh pkgsrc, only lang/gcc16
> itself updated, so i'll try a clean fully frseh pkgsrc too.

Well, the package works because of this section in Makefile.common:

.if !empty(PKGSRC_COMPILER:Mclang) || ${OPSYS} == "DragonFly" \
    || ${OPSYS} == "NetBSD"
# GCC 16.1.0's stage2/stage3 bootstrap comparison currently fails on
# NetBSD/x86_64, so disable it. Revisit on updates.
# https://gcc.gnu.org/pipermail/gcc-help/2026-August/145120.html
CONFIGURE_ARGS+=       --disable-bootstrap
.endif

Did you comment out that CONFIGURE_ARGS line when testing?
 Thomas

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-admin@muc.de

[toc] | [prev] | [next] | [standalone]

#3636

Thomas Klausner writes:
> Did you comment out that CONFIGURE_ARGS line when testing?

yes.  well, i did forget for the 2nd time, but when it finished
really fast, i realised and did it again.


.mrg.

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-admin@muc.de

[toc] | [prev] | [standalone]

Back to top | Article view | muc.lists.netbsd.tech.toolchain

csiph-web