Re: toolchain/57378: Issue with set/checkflist

[Sean Cole <scole%NetBSD.org@localhost>]

The following reply was made to PR toolchain/57378; it has been noted by GNATS.

From: Sean Cole <scole%NetBSD.org@localhost>
To: Taylor R Campbell <riastradh%NetBSD.org@localhost>
Cc: bch <brad.harder%gmail.com@localhost>, gnats-bugs%netbsd.org@localhost, gnats-admin%netbsd.org@localhost,
        netbsd-bugs%netbsd.org@localhost, toolchain-manager%netbsd.org@localhost
Subject: Re: toolchain/57378: Issue with set/checkflist
Date: Thu, 25 May 2023 08:34:28 +0700

 Taylor R Campbell <riastradh%NetBSD.org@localhost> writes:
 
 >> Date: Wed, 24 May 2023 12:37:36 -0700
 >> From: bch <brad.harder%gmail.com@localhost>
 >> 
 >> On Tue, Apr 25, 2023 at 04:40 Taylor R Campbell <riastradh%netbsd.org@localhost>
 >> wrote:
 >> 
 >> > > Date: Tue, 25 Apr 2023 05:35:00 +0000 (UTC)
 >> > > From: Bch <beceeh%shaw.ca@localhost>
 >> > >
 >> > > =======  1 extra files in DESTDIR  =========
 >> > > Files in DESTDIR but missing from flist.
 >> > > File is obsolete or flist is out of date ?
 >> > > ------------------------------------------
 >> > > ./usr/libexec/ld.elf_so
 >> >
 >> >  I haven't seen this and it hasn't turned up in the autobuilds.  Is
 >> > this from a clean objdir?  If not, can you try that?
 >> 
 >> Hey riastradh - I think I've isolated it to MKDYNAMICROOT = no.
 >> 
 >> Happy to take instruction to test this further - I'm interested in
 >> understanding our flist facilities more, this might be a good opportunity.
 >
 > It looks like there is a spurious `dynamicroot' condition on the
 > /usr/libexec/ld.elf_so entry, which should instead be unconditional.
 >
 > The condition may have been added by mistake by scole@ in 2016 for
 > this PR to fix various ia64 build issues:
 >
 > https://gnats.netbsd.org/51261
 >
 > The PR has a comment about it suggesting it was intentional:
 >
 >  M distrib/sets/lists/base/shl.mi # change dynamicroot for ./lib/npf*,
 > usr/libexec/ld.elf_so
 >
 > But I don't think that makes sense, because /usr/libexec/ld.elf_so is
 > always there (it's a regular file with MKDYNAMICROOT=no, a symlink
 > with MKDYNAMICROOT=yes), whereas /libexec/ld.elf_so is correctly
 > conditioned on `dynamicroot'.
 
 Feel free to change to whatever makes sense.
 
 I am not sure of the exact details for this particular change anymore,
 but have a vague recollection of ld.elf not yet being implemented for
 ia64 and trying to work around that.
 
 Also, it might be helpful to understand why the issue is showing up
 now.  I think I committed this about 7 years ago.
 
 Thanks