Re: toolchain/48627: CTFCONVERT is set depending on properties of the build host

To: toolchain-manager%netbsd.org@localhost,gnats-admin%netbsd.org@localhost,netbsd-bugs%netbsd.org@localhost,martin%NetBSD.org@localhost
Subject: Re: toolchain/48627: CTFCONVERT is set depending on properties of the build host
From: David Holland <dholland-bugs%netbsd.org@localhost>
Date: Sat, 26 Apr 2014 05:30:02 +0000 (UTC)

The following reply was made to PR toolchain/48627; it has been noted by GNATS.

From: David Holland <dholland-bugs%netbsd.org@localhost>
To: gnats-bugs%netbsd.org@localhost
Cc: 
Subject: Re: toolchain/48627: CTFCONVERT is set depending on properties of
 the build host
Date: Sat, 26 Apr 2014 05:29:51 +0000

 Whole pile of messages not sent to gnats.
 
    ------
 
 From: Ryota Ozaki <ozaki-r%NetBSD.org@localhost>
 To: Christos Zoulas <christos%zoulas.com@localhost>, 
martin%netbsd.org@localhost
 Cc: toolchain-manager%netbsd.org@localhost, netbsd-bugs%netbsd.org@localhost,
        gnats-admin%netbsd.org@localhost
 Subject: Re: toolchain/48627: CTFCONVERT is set depending on properties of the
        build host
 Date: Wed, 5 Mar 2014 10:33:33 +0900
 
 Could I commit the patch? Christos?
 
 The fix also helps my work :)
 
 Regards,
   ozaki-r
 
 From: Christos Zoulas <christos%zoulas.com@localhost>
 To: Ryota Ozaki <ozaki-r%NetBSD.org@localhost>, martin%netbsd.org@localhost
 Cc: toolchain-manager%netbsd.org@localhost, netbsd-bugs%netbsd.org@localhost,
        gnats-admin%netbsd.org@localhost
 Subject: Re: toolchain/48627: CTFCONVERT is set depending on properties of the
        build host
 Date: Wed, 5 Mar 2014 10:33:08 -0500
 
 On Mar 5, 10:33am, ozaki-r%NetBSD.org@localhost (Ryota Ozaki) wrote:
 -- Subject: Re: toolchain/48627: CTFCONVERT is set depending on properties of
 
 | Could I commit the patch? Christos?
 | 
 | The fix also helps my work :)
 
 The bsd.own.mk fix is probably fine (but does not really make a difference),
 but the sys.mk cannot possibly work because sys.mk is the first file to be
 loaded by make:
 
 $ cd /usr/src/usr.bin/make
 $ ktrace make
 $ kdump | grep NAMI
   3968      1 ktrace   NAMI  "./make"
   3968      1 ktrace   NAMI  "/usr/bin/make"
   3968      1 ktrace   NAMI  "/usr/libexec/ld.elf_so"
   3968      1 make     NAMI  "/etc/ld.so.conf"
   3968      1 make     NAMI  "/usr/lib/libutil.so.7"
   3968      1 make     NAMI  "/usr/lib/libgcc_s.so.1"
   3968      1 make     NAMI  "/usr/lib/libc.so.12"
   3968      1 make     NAMI  "/etc/malloc.conf"
   3968      1 make     NAMI  "/usr/src/usr.bin/make"
   3968      1 make     NAMI  "/usr/src/usr.bin/make"
   3968      1 make     NAMI  "/usr/src/usr.bin/make"
   3968      1 make     NAMI  "/usr/src/usr.bin/make"
   3968      1 make     NAMI  "/usr/src/usr.bin/make"
   3968      1 make     NAMI  "."
   3968      1 make     NAMI  "/usr/src/usr.bin/make/obj.amd64"
   3968      1 make     NAMI  "/usr/src/usr.bin/make/obj.amd64"
   3968      1 make     NAMI  "."
   3968      1 make     NAMI  "/usr/share/mk"
   3968      1 make     NAMI  "/usr/share/mk/sys.mk"
 
 So, I don't think that the patch is correct. How have you tested that it works?
 
 christos
 
 From: Ryota Ozaki <ozaki-r%NetBSD.org@localhost>
 To: Christos Zoulas <christos%zoulas.com@localhost>
 Cc: martin%netbsd.org@localhost, toolchain-manager%netbsd.org@localhost, 
netbsd-bugs%netbsd.org@localhost,
        gnats-admin%netbsd.org@localhost
 Subject: Re: toolchain/48627: CTFCONVERT is set depending on properties of the
        build host
 Date: Thu, 6 Mar 2014 10:14:23 +0900
 
 On Thu, Mar 6, 2014 at 12:33 AM, Christos Zoulas 
<christos%zoulas.com@localhost> wrote:
 > On Mar 5, 10:33am, ozaki-r%NetBSD.org@localhost (Ryota Ozaki) wrote:
 > -- Subject: Re: toolchain/48627: CTFCONVERT is set depending on properties of
 >
 > | Could I commit the patch? Christos?
 > |
 > | The fix also helps my work :)
 >
 > The bsd.own.mk fix is probably fine (but does not really make a difference),
 
 This fix makes sense when we build tools with MKDTRACE but build kernel w/o
 MKDTRACE (i.e., build.sh -V MKDTRACE=yes tools; build.sh kernel=GENERIC);
 without the fix, nbctfconvert unexpectedly runs w/o MKDTRACE.
 
 BTW, nbctfconvert to the kernel unfortunately fails for now, so running it w/o
 MKDTRACE annoyed me.
 
 > but the sys.mk cannot possibly work because sys.mk is the first file to be
 > loaded by make:
 >
 > $ cd /usr/src/usr.bin/make
 > $ ktrace make
 > $ kdump | grep NAMI
 >   3968      1 ktrace   NAMI  "./make"
 >   3968      1 ktrace   NAMI  "/usr/bin/make"
 >   3968      1 ktrace   NAMI  "/usr/libexec/ld.elf_so"
 >   3968      1 make     NAMI  "/etc/ld.so.conf"
 >   3968      1 make     NAMI  "/usr/lib/libutil.so.7"
 >   3968      1 make     NAMI  "/usr/lib/libgcc_s.so.1"
 >   3968      1 make     NAMI  "/usr/lib/libc.so.12"
 >   3968      1 make     NAMI  "/etc/malloc.conf"
 >   3968      1 make     NAMI  "/usr/src/usr.bin/make"
 >   3968      1 make     NAMI  "/usr/src/usr.bin/make"
 >   3968      1 make     NAMI  "/usr/src/usr.bin/make"
 >   3968      1 make     NAMI  "/usr/src/usr.bin/make"
 >   3968      1 make     NAMI  "/usr/src/usr.bin/make"
 >   3968      1 make     NAMI  "."
 >   3968      1 make     NAMI  "/usr/src/usr.bin/make/obj.amd64"
 >   3968      1 make     NAMI  "/usr/src/usr.bin/make/obj.amd64"
 >   3968      1 make     NAMI  "."
 >   3968      1 make     NAMI  "/usr/share/mk"
 >   3968      1 make     NAMI  "/usr/share/mk/sys.mk"
 >
 > So, I don't think that the patch is correct. How have you tested that it 
 > works?
 
 I tested only for my case and I thought the fix could be applied to the other.
 I'm sorry, I jumped to conclusions.
 
   ozaki-r
 
 >
 > christos
 
 From: Masao Uebayashi <uebayasi%gmail.com@localhost>
 To: Ryota Ozaki <ozaki-r%netbsd.org@localhost>
 Cc: Christos Zoulas <christos%zoulas.com@localhost>, Martin Husemann 
<martin%netbsd.org@localhost>,
        toolchain-manager%netbsd.org@localhost, 
netbsd-bugs%netbsd.org@localhost,
        gnats-admin%netbsd.org@localhost
 Subject: Re: toolchain/48627: CTFCONVERT is set depending on properties of the
        build host
 Date: Thu, 6 Mar 2014 10:44:35 +0900
 
 Ideally CTFCONVERT'ed objects have dedicated rule (.go -> .ctfo) and
 things become much cleaner.
 
 (Why do installed executables have no suffixe on Unix?)
 
 From: Christos Zoulas <christos%zoulas.com@localhost>
 To: Ryota Ozaki <ozaki-r%NetBSD.org@localhost>
 Cc: martin%netbsd.org@localhost, toolchain-manager%netbsd.org@localhost, 
netbsd-bugs%netbsd.org@localhost,
        gnats-admin%netbsd.org@localhost
 Subject: Re: toolchain/48627: CTFCONVERT is set depending on properties of the
        build host
 Date: Wed, 5 Mar 2014 21:17:25 -0500
 
 On Mar 6, 10:14am, ozaki-r%NetBSD.org@localhost (Ryota Ozaki) wrote:
 -- Subject: Re: toolchain/48627: CTFCONVERT is set depending on properties of
 
 | This fix makes sense when we build tools with MKDTRACE but build kernel w/o
 | MKDTRACE (i.e., build.sh -V MKDTRACE=yes tools; build.sh kernel=GENERIC);
 | without the fix, nbctfconvert unexpectedly runs w/o MKDTRACE.
 | 
 | BTW, nbctfconvert to the kernel unfortunately fails for now, so running it 
w/o
 | MKDTRACE annoyed me.
 | 
 
 [stuff deleted]
 
 | I tested only for my case and I thought the fix could be applied to the 
other.
 | I'm sorry, I jumped to conclusions.
 
 That's the problem with sys.mk; We can't put variables there because people
 will think that they can set those variables in /etc/mk.conf and have it work
 properly. I still don't know the right answer.
 
 christos

Follow-Ups:
- Re: toolchain/48627: CTFCONVERT is set depending on properties of the build host
  - From: Tetsuya Isaki

Prev by Date: Re: install/48724: Set lists not updated for files made obsolete by GCC bump to 4.8.3
Next by Date: Re: port-cobalt/46170 (NetBSD/cobalt 6.0_BETA does not boot)
Previous by Thread: Re: toolchain/48627: CTFCONVERT is set depending on properties of the build host
Next by Thread: Re: toolchain/48627: CTFCONVERT is set depending on properties of the build host
Indexes:

Home | Main Index | Thread Index | Old Index