Cannot build guix as part of guix system reconfigure after commit 224d437fb4 on aarch64

Recipients:(address . bug-guix@gnu.org)

Message-ID:CAPWcbYHjBoqd0vsLOknF7Yk86wUiC5P_ppTguv8G=fziJ=FJeQ@mail.gmail.com

Hi,

I'm new to guix so I'm really struggling to debug this strange issue.
On the latest master I can run 'guix pull', and the latest guix seems
to build just fine, however when I go to do 'guix system reconfigure',
building guix fails the check phase. I've attached the log file. After
some help on IRC, I understand that during 'guix system reconfigure'
ends up building a previous version of itself via the "boot-stripping"
process in gnu/packages/package-management.scm. I followed this linked
list of commits backwards and found that commits prior to commit
224d437fb4 which updates this guix package definition to version 14
which is commit 2a621f1 build just fine, however all the versions of
that package afterwards fail at the same place during 'guix system
reconfigure', in the check phase, with the same tests failing. At this
point, I think I lack sufficient knowledge of guix to debug this issue
further and it looks like cuirass is so backlogged on aarch64 that is
still hasn't gotten to building these guix versions so I can't compare
with it's build attempts. Any help would be greatly appreciated as I'd
really like to be running the latest guix from master.

Thanks!

Attachment: 7w4qq23xj1jhk0972x7q8s5q918pl3-guix-1.3.0-14.2a621f1.drv

Leo Famulari wrote on 2 Jan 03:13 +0100

Recipients:(name . Aiko Kyle)(address . aikokyle@gmail.com)(address . 52943@debbugs.gnu.org)

Message-ID:YdEKYg2eDo2b7b6h@jasmine.lan

On Sat, Jan 01, 2022 at 04:54:51PM -0700, Aiko Kyle wrote:

Toggle quote (18 lines)> I'm new to guix so I'm really struggling to debug this strange issue.
> On the latest master I can run 'guix pull', and the latest guix seems
> to build just fine, however when I go to do 'guix system reconfigure',
> building guix fails the check phase. I've attached the log file. After
> some help on IRC, I understand that during 'guix system reconfigure'
> ends up building a previous version of itself via the "boot-stripping"
> process in gnu/packages/package-management.scm. I followed this linked
> list of commits backwards and found that commits prior to commit
> 224d437fb4 which updates this guix package definition to version 14
> which is commit 2a621f1 build just fine, however all the versions of
> that package afterwards fail at the same place during 'guix system
> reconfigure', in the check phase, with the same tests failing. At this
> point, I think I lack sufficient knowledge of guix to debug this issue
> further and it looks like cuirass is so backlogged on aarch64 that is
> still hasn't gotten to building these guix versions so I can't compare
> with it's build attempts. Any help would be greatly appreciated as I'd
> really like to be running the latest guix from master.

Thanks, this has already been reported as

https://issues.guix.gnu.org/52908

Leo Famulari wrote on 2 Jan 03:14 +0100

(no subject)

Recipients:(name . GNU bug tracker automated control server)(address . control@debbugs.gnu.org)

Message-ID:YdEKgAUl76AdwYB5@jasmine.lan

merge 52908 52943

Aiko Kyle wrote on 2 Jan 05:11 +0100

Re: bug#52943: Cannot build guix as part of guix system reconfigure after commit 224d437fb4 on aarch64

Recipients:(name . Leo Famulari)(address . leo@famulari.name)(address . 52943@debbugs.gnu.org)

Message-ID:CAPWcbYEu4hcP6KQq8pumbk5+9uALiKdibR_kJ8KER-kCf9mDug@mail.gmail.com

Attached is the output of guix system shepherd-graph

gnu/system/examples/desktop.tmpl on commit 2a49ddb513

Attachment: graph.dot

Leo Famulari wrote on 2 Jan 05:51 +0100

Recipients:(name . Aiko Kyle)(address . aikokyle@gmail.com)(address . 52943@debbugs.gnu.org)

Message-ID:YdEvVvT7k01JrYA2@jasmine.lan

On Sat, Jan 01, 2022 at 09:11:49PM -0700, Aiko Kyle wrote:

Toggle quote (3 lines)

> Attached is the output of guix system shepherd-graph

> gnu/system/examples/desktop.tmpl on commit 2a49ddb513

Thanks, and I've attached a graph of the same operating system
declaration from the same revision of Guix but for x86_64, along with
rendered images of both graphs.

The graphs clearly show that xorg-server is provided in two places on
aarch64. The question is why?

digraph "Guix shepherd-service" {
  "user-file-systems" [label = "user-file-systems", shape = box, fontname = sans];
  "file-systems" -> "user-file-systems" [color = darkviolet];
  "file-systems" [label = "file-systems", shape = box, fontname = sans];
  "user-processes" -> "file-systems" [color = dimgrey];
  "user-homes" -> "file-systems" [color = darkgoldenrod];
  "urandom-seed" -> "file-systems" [color = magenta];
  "user-processes" [label = "user-processes", shape = box, fontname = sans];
  "nscd" -> "user-processes" [color = darkseagreen];
  "guix-daemon" -> "user-processes" [color = dimgrey];
  "syslogd" -> "user-processes" [color = darkseagreen];
  "term-tty6" -> "user-processes" [color = peachpuff4];
  "term-tty5" -> "user-processes" [color = red];
  "term-tty4" -> "user-processes" [color = darkseagreen];
  "term-tty3" -> "user-processes" [color = darkgoldenrod];
  "term-tty2" -> "user-processes" [color = dimgrey];
  "term-tty1" -> "user-processes" [color = magenta];
  "term-auto" -> "user-processes" [color = red];
  "ntpd" -> "user-processes" [color = red];
  "dbus-system" -> "user-processes" [color = peachpuff4];
  "avahi-daemon" -> "user-processes" [color = cyan3];
  "wpa-supplicant" -> "user-processes" [color = dimgrey];
  "networking" -> "user-processes" [color = darkviolet];
  "xorg-server" -> "user-processes" [color = red];
  "mcron" -> "user-processes" [color = darkgoldenrod];
  "nscd" [label = "nscd", shape = box, fontname = sans];
  "guix-daemon" [label = "guix-daemon", shape = box, fontname = sans];
  "syslogd" [label = "syslogd", shape = box, fontname = sans];
  "dbus-system" -> "syslogd" [color = peachpuff4];
  "wpa-supplicant" -> "syslogd" [color = dimgrey];
  "dbus-system" [label = "dbus-system", shape = box, fontname = sans];
  "elogind" -> "dbus-system" [color = darkviolet];
  "upower-daemon" -> "dbus-system" [color = dimgrey];
  "avahi-daemon" -> "dbus-system" [color = cyan3];
  "wpa-supplicant" -> "dbus-system" [color = dimgrey];
  "networking" -> "dbus-system" [color = darkviolet];
  "xorg-server" -> "dbus-system" [color = red];
  "elogind" [label = "elogind", shape = box, fontname = sans];
  "upower-daemon" [label = "upower-daemon", shape = box, fontname = sans];
  "avahi-daemon" [label = "avahi-daemon", shape = box, fontname = sans];
  "wpa-supplicant" [label = "wpa-supplicant", shape = box, fontname = sans];
  "networking" -> "wpa-supplicant" [color = darkviolet];
  "networking" [label = "networking", shape = box, fontname = sans];
  "ntpd" -> "networking" [color = red];
  "avahi-daemon" -> "networking" [color = cyan3];
  "ntpd" [label = "ntpd", shape = box, fontname = sans];
  "xorg-server" [label = "xorg-server", shape = box, fontname = sans];
  "term-tty6" [label = "term-tty6", shape = box, fontname = sans];
  "console-font-tty6" -> "term-tty6" [color = darkviolet];
  "console-font-tty6" [label = "console-font-tty6", shape = box, fontname = sans];
  "term-tty5" [label = "term-tty5", shape = box, fontname = sans];
  "console-font-tty5" -> "term-tty5" [color = peachpuff4];
  "console-font-tty5" [label = "console-font-tty5", shape = box, fontname = sans];
  "term-tty4" [label = "term-tty4", shape = box, fontname = sans];
  "console-font-tty4" -> "term-tty4" [color = red];
  "console-font-tty4" [label = "console-font-tty4", shape = box, fontname = sans];
  "term-tty3" [label = "term-tty3", shape = box, fontname = sans];
  "console-font-tty3" -> "term-tty3" [color = cyan3];
  "console-font-tty3" [label = "console-font-tty3", shape = box, fontname = sans];
  "term-tty2" [label = "term-tty2", shape = box, fontname = sans];
  "console-font-tty2" -> "term-tty2" [color = blue];
  "console-font-tty2" [label = "console-font-tty2", shape = box, fontname = sans];
  "term-tty1" [label = "term-tty1", shape = box, fontname = sans];
  "console-font-tty1" -> "term-tty1" [color = darkseagreen];
  "console-font-tty1" [label = "console-font-tty1", shape = box, fontname = sans];
  "term-auto" [label = "term-auto", shape = box, fontname = sans];
  "mcron" [label = "mcron", shape = box, fontname = sans];
  "user-homes" [label = "user-homes", shape = box, fontname = sans];
  "user-processes" -> "user-homes" [color = dimgrey];
  "urandom-seed" [label = "urandom-seed", shape = box, fontname = sans];
  "user-processes" -> "urandom-seed" [color = dimgrey];
  "root-file-system" [label = "root-file-system", shape = box, fontname = sans];
  "file-systems" -> "root-file-system" [color = darkviolet];
  "file-system-/boot/efi" -> "root-file-system" [color = cyan3];
  "file-system-/dev/pts" -> "root-file-system" [color = red];
  "file-system-/sys/kernel/debug" -> "root-file-system" [color = cyan3];
  "file-system-/dev/shm" -> "root-file-system" [color = peachpuff4];
  "file-system-/sys/firmware/efi/efivars" -> "root-file-system" [color = red];
  "file-system-/gnu/store" -> "root-file-system" [color = magenta];
  "file-system-/run/systemd" -> "root-file-system" [color = darkseagreen];
  "file-system-/run/user" -> "root-file-system" [color = magenta];
  "file-system-/sys/fs/cgroup/elogind" -> "root-file-system" [color = dimgrey];
  "file-system-/sys/fs/cgroup" -> "root-file-system" [color = dimgrey];
  "file-system-/sys/fs/cgroup/cpuset" -> "root-file-system" [color = blue];
  "file-system-/sys/fs/cgroup/cpu" -> "root-file-system" [color = cyan3];
  "file-system-/sys/fs/cgroup/cpuacct" -> "root-file-system" [color = magenta];
  "file-system-/sys/fs/cgroup/memory" -> "root-file-system" [color = darkgoldenrod];
  "file-system-/sys/fs/cgroup/devices" -> "root-file-system" [color = magenta];
  "file-system-/sys/fs/cgroup/freezer" -> "root-file-system" [color = red];
  "file-system-/sys/fs/cgroup/blkio" -> "root-file-system" [color = peachpuff4];
  "file-system-/sys/fs/cgroup/perf_event" -> "root-file-system" [color = red];
  "file-system-/sys/fs/cgroup/pids" -> "root-file-system" [color = darkviolet];
  "file-system-/var/cache/fontconfig" -> "root-file-system" [color = magenta];
  "udev" -> "root-file-system" [color = cyan3];
  "virtual-terminal" -> "root-file-system" [color = peachpuff4];
  "file-system-/boot/efi" [label = "file-system-/boot/efi", shape = box, fontname = sans];
  "file-systems" -> "file-system-/boot/efi" [color = darkviolet];
  "file-system-/dev/pts" [label = "file-system-/dev/pts", shape = box, fontname = sans];
  "file-systems" -> "file-system-/dev/pts" [color = darkviolet];
  "file-system-/sys/kernel/debug" [label = "file-system-/sys/kernel/debug", shape = box, fontname = sans];
  "file-systems" -> "file-system-/sys/kernel/debug" [color = darkviolet];
  "file-system-/dev/shm" [label = "file-system-/dev/shm", shape = box, fontname = sans];
  "file-systems" -> "file-system-/dev/shm" [color = darkviolet];
  "file-system-/sys/firmware/efi/efivars" [label = "file-system-/sys/firmware/efi/efivars", shape = box, fontname = sans];
  "file-systems" -> "file-system-/sys/firmware/efi/efivars" [color = darkviolet];
  "file-system-/gnu/store" [label = "file-system-/gnu/store", shape = box, fontname = sans];
  "file-systems" -> "file-system-/gnu/store" [color = darkviolet];
  "file-system-/run/systemd" [label = "file-system-/run/systemd", shape = box, fontname = sans];
  "file-systems" -> "file-system-/run/systemd" [color = darkviolet];
  "file-system-/run/user" [label = "file-system-/run/user", shape = box, fontname = sans];
  "file-systems" -> "file-system-/run/user" [color = darkviolet];
  "file-system-/sys/fs/cgroup/elogind" [label = "file-system-/sys/fs/cgroup/elogind", shape = box, fontname = sans];
  "file-systems" -> "file-system-/sys/fs/cgroup/elogind" [color = darkviolet];
  "file-system-/sys/fs/cgroup" [label = "file-system-/sys/fs/cgroup", shape = box, fontname = sans];
  "file-systems" -> "file-system-/sys/fs/cgroup" [color = darkviolet];
  "file-system-/sys/fs/cgroup/elogind" -> "file-system-/sys/fs/cgroup" [color = dimgrey];
  "file-system-/sys/fs/cgroup/cpuset" -> "file-system-/sys/fs/cgroup" [color = blue];
  "file-system-/sys/fs/cgroup/cpu" -> "file-system-/sys/fs/cgroup" [color = cyan3];
  "file-system-/sys/fs/cgroup/cpuacct" -> "file-system-/sys/fs/cgroup" [color = magenta];
  "file-system-/sys/fs/cgroup/memory" -> "file-system-/sys/fs/cgroup" [color = darkgoldenrod];
  "file-system-/sys/fs/cgroup/devices" -> "file-system-/sys/fs/cgroup" [color = magenta];
  "file-system-/sys/fs/cgroup/freezer" -> "file-system-/sys/fs/cgroup" [color = red];
  "file-system-/sys/fs/cgroup/blkio" -> "file-system-/sys/fs/cgroup" [color = peachpuff4];
  "file-system-/sys/fs/cgroup/perf_event" -> "file-system-/sys/fs/cgroup" [color = red];
  "file-system-/sys/fs/cgroup/pids" -> "file-system-/sys/fs/cgroup" [color = darkviolet];
  "file-system-/sys/fs/cgroup/cpuset" [label = "file-system-/sys/fs/cgroup/cpuset", shape = box, fontname = sans];
  "file-systems" -> "file-system-/sys/fs/cgroup/cpuset" [color = darkviolet];
  "file-system-/sys/fs/cgroup/cpu" [label = "file-system-/sys/fs/cgroup/cpu", shape = box, fontname = sans];
  "file-systems" -> "file-system-/sys/fs/cgroup/cpu" [color = darkviolet];
  "file-system-/sys/fs/cgroup/cpuacct" [label = "file-system-/sys/fs/cgroup/cpuacct", shape = box, fontname = sans];
  "file-systems" -> "file-system-/sys/fs/cgroup/cpuacct" [color = darkviolet];
  "file-system-/sys/fs/cgroup/memory" [label = "file-system-/sys/fs/cgroup/memory", shape = box, fontname = sans];
  "file-systems" -> "file-system-/sys/fs/cgroup/memory" [color = darkviolet];
  "file-system-/sys/fs/cgroup/devices" [label = "file-system-/sys/fs/cgroup/devices", shape = box, fontname = sans];
  "file-systems" -> "file-system-/sys/fs/cgroup/devices" [color = darkviolet];
  "file-system-/sys/fs/cgroup/freezer" [label = "file-system-/sys/fs/cgroup/freezer", shape = box, fontname = sans];
  "file-systems" -> "file-system-/sys/fs/cgroup/freezer" [color = darkviolet];
  "file-system-/sys/fs/cgroup/blkio" [label = "file-system-/sys/fs/cgroup/blkio", shape = box, fontname = sans];
  "file-systems" -> "file-system-/sys/fs/cgroup/blkio" [color = darkviolet];
  "file-system-/sys/fs/cgroup/perf_event" [label = "file-system-/sys/fs/cgroup/perf_event", shape = box, fontname = sans];
  "file-systems" -> "file-system-/sys/fs/cgroup/perf_event" [color = darkviolet];
  "file-system-/sys/fs/cgroup/pids" [label = "file-system-/sys/fs/cgroup/pids", shape = box, fontname = sans];
  "file-systems" -> "file-system-/sys/fs/cgroup/pids" [color = darkviolet];
  "file-system-/var/cache/fontconfig" [label = "file-system-/var/cache/fontconfig", shape = box, fontname = sans];
  "file-systems" -> "file-system-/var/cache/fontconfig" [color = darkviolet];
  "udev" [label = "udev", shape = box, fontname = sans];
  "swap-/swapfile" -> "udev" [color = darkseagreen];
  "file-system-/boot/efi" -> "udev" [color = cyan3];
  "file-system-/dev/pts" -> "udev" [color = red];
  "file-system-/sys/kernel/debug" -> "udev" [color = cyan3];
  "file-system-/dev/shm" -> "udev" [color = peachpuff4];
  "file-system-/sys/firmware/efi/efivars" -> "udev" [color = red];
  "file-system-/gnu/store" -> "udev" [color = magenta];
  "file-system-/run/systemd" -> "udev" [color = darkseagreen];
  "file-system-/run/user" -> "udev" [color = magenta];
  "file-system-/sys/fs/cgroup/elogind" -> "udev" [color = dimgrey];
  "file-system-/sys/fs/cgroup" -> "udev" [color = dimgrey];
  "file-system-/sys/fs/cgroup/cpuset" -> "udev" [color = blue];
  "file-system-/sys/fs/cgroup/cpu" -> "udev" [color = cyan3];
  "file-system-/sys/fs/cgroup/cpuacct" -> "udev" [color = magenta];
  "file-system-/sys/fs/cgroup/memory" -> "udev" [color = darkgoldenrod];
  "file-system-/sys/fs/cgroup/devices" -> "udev" [color = magenta];
  "file-system-/sys/fs/cgroup/freezer" -> "udev" [color = red];
  "file-system-/sys/fs/cgroup/blkio" -> "udev" [color = peachpuff4];
  "file-system-/sys/fs/cgroup/perf_event" -> "udev" [color = red];
  "file-system-/sys/fs/cgroup/pids" -> "udev" [color = darkviolet];
  "file-system-/var/cache/fontconfig" -> "udev" [color = magenta];
  "urandom-seed" -> "udev" [color = magenta];
  "term-tty6" -> "udev" [color = peachpuff4];
  "term-tty5" -> "udev" [color = red];
  "term-tty4" -> "udev" [color = darkseagreen];
  "term-tty3" -> "udev" [color = darkgoldenrod];
  "term-tty2" -> "udev" [color = dimgrey];
  "term-tty1" -> "udev" [color = magenta];
  "term-auto" -> "udev" [color = red];
  "upower-daemon" -> "udev" [color = dimgrey];
  "xorg-server" -> "udev" [color = red];
  "swap-/swapfile" [label = "swap-/swapfile", shape = box, fontname = sans];
  "virtual-terminal" [label = "virtual-terminal", shape = box, fontname = sans];
  "term-tty6" -> "virtual-terminal" [color = peachpuff4];
  "term-tty5" -> "virtual-terminal" [color = red];
  "term-tty4" -> "virtual-terminal" [color = darkseagreen];
  "term-tty3" -> "virtual-terminal" [color = darkgoldenrod];
  "term-tty2" -> "virtual-terminal" [color = dimgrey];
  "term-tty1" -> "virtual-terminal" [color = magenta];
  "host-name" [label = "host-name", shape = box, fontname = sans];
  "term-tty6" -> "host-name" [color = peachpuff4];
  "term-tty5" -> "host-name" [color = red];
  "term-tty4" -> "host-name" [color = darkseagreen];
  "term-tty3" -> "host-name" [color = darkgoldenrod];
  "term-tty2" -> "host-name" [color = dimgrey];
  "term-tty1" -> "host-name" [color = magenta];
  "term-auto" -> "host-name" [color = red];
  "xorg-server" -> "host-name" [color = red];
  "sysctl" [label = "sysctl", shape = box, fontname = sans];
  "loopback" [label = "loopback", shape = box, fontname = sans];
  "wpa-supplicant" -> "loopback" [color = dimgrey];
  "networking" -> "loopback" [color = darkviolet];

}

Attachment: graph-x86_64.png

Attachment: graph-aarch64.png

Aiko Kyle wrote on 2 Jan 07:36 +0100

Recipients:(name . Leo Famulari)(address . leo@famulari.name)(address . 52943@debbugs.gnu.org)

Message-ID:CAPWcbYEjhSWsXR41ivHUZfthUje_i0rc1P8p_8KYqw8ibtjJEg@mail.gmail.com

On Sat, Jan 1, 2022 at 9:51 PM Leo Famulari <leo@famulari.name> wrote:

Toggle quote (8 lines)

> Thanks, and I've attached a graph of the same operating system

> declaration from the same revision of Guix but for x86_64, along with

> rendered images of both graphs.

> The graphs clearly show that xorg-server is provided in two places on

> aarch64. The question is why?

Thanks Leo for the help on IRC!

What we seem to have figured out is that commit 49599fab56 is almost
certainly the culprit. So the issue is that somehow using
sddm-service-type in %desktop-services on aarch64 instead of
gdm-service-type which is used an x86_64 after that commit results in
xorg-server-service-type being included twice, likely though
gdm-service-type still being included somewhere.
set-xorg-configuration is a potential culprit, although so far I
haven't' been able to make the test pass by modifying the call to
set-xorg-configuration in desktop.tmpl to explicitly pass
sddm-service-type.

Akira Kyle wrote on 3 Jan 22:38 +0100

Recipients:(name . Leo Famulari)(address . leo@famulari.name)(address . 52943@debbugs.gnu.org)

Message-ID:CAPWcbYEUitbYDjfhfwwNJvysUkPmt-RLVjuNvQ=U0hRbBT=jLQ@mail.gmail.com

The attached patch fixes the failing guix-system.sh test on aarch64.

However there are now other tests that are failing.

Attachment: 0001-tests-guix-system-Fix-on-non-x86_64-systems.patch

Akira Kyle wrote on 4 Jan 02:13 +0100

Recipients:(name . Leo Famulari)(address . leo@famulari.name)(address . 52943@debbugs.gnu.org)

Message-ID:CAPWcbYHDeB_ebMK6iRd4r3PDZ+v64MQjBvkF0GTbqrQMHiSHdw@mail.gmail.com

Attached is the current log file for the test suite on commit 92faad0

with the patch applied.

Attachment: test-suite.log

Akira Kyle wrote on 4 Jan 02:16 +0100

Recipients:(name . Leo Famulari)(address . leo@famulari.name)(address . 52943@debbugs.gnu.org)

Message-ID:CAPWcbYEK0SrJxEa7f6zram8O5zGM_H5eGdHRi=Tf1OnTQMud8w@mail.gmail.com

Some of these failing tests have been reported on guix-devel already:

https://lists.gnu.org/archive/html/guix-devel/2021-12/msg00158.html

On Mon, Jan 3, 2022 at 6:13 PM Akira Kyle <akira@akirakyle.com> wrote:

Toggle quote (3 lines)

> Attached is the current log file for the test suite on commit 92faad0

> with the patch applied.

Chris Marusich wrote on 9 Jan 03:40 +0100

Recipients:(address . control@debbugs.gnu.org)

Message-ID:87wnj9d435.fsf@gmail.com

# Although at first it did seem like 52908 was the sole cause of this

# bug, Aiko reported in 52908 that there are still more problems needing

# resolution. So, this should be regarded as a separate bug, after all.

unmerge 52943

reopen 52943

Chris

PGP: https://savannah.gnu.org/people/viewgpg.php?user_id=106836

Chris Marusich wrote on 9 Jan 05:39 +0100

Recipients:(name . Aiko Kyle)(address . aikokyle@gmail.com)(address . 52943@debbugs.gnu.org)

Message-ID:87o84lcyk1.fsf@gmail.com

Aiko Kyle <aikokyle@gmail.com> writes:

Toggle quote (4 lines)

> On the latest master I can run 'guix pull', and the latest guix seems

> to build just fine, however when I go to do 'guix system reconfigure',

> building guix fails the check phase.

Aiko has reported that the fix for 52908 did indeed fix the

"tests/guix-system.sh" problem on aarch64-linux. This means the

"service 'xorg-server' provided more than once" error has been fixed.

However, fixing that issue has revealed another. Aiko said:

Toggle quote (3 lines)

> guix system reconfigure is still failing for me on aarch64 due to the

> test 'file-needed/recursive' in tests/gremlin.scm failing.

So the current status is that Aiko still can't do "guix reconfigure" on

master after running "guix pull". For this reason, I have re-opened

this bug and unmerged it from 52908, since it's not exactly the same

issue any more.

At the moment, the file-needed/recursive bug that is preventing Aiko

from running "guix reconfigure" appears to be specific to aarch64-linux.

The problem was reported on guix-devel here:

https://lists.gnu.org/archive/html/guix-devel/2022-01/msg00019.html

Pierre Langlois did some investigation and found this information,

posted in the email thread above:

Toggle quote (145 lines)> I've also been trying to fix this test but without much luck.  It
> does look similar to this issue for ppc64 [0], where the `ldd/ld.so'
> beaviour isn't the same as what the gremlin guile module does. However
> the patch proposed there isn't fixing the issue for me either on
> aarch64.
> 
> [0]: https://issues.guix.gnu.org/52940,
> 
> Maybe we can compare notes and a solution will come up :-).  So the test
> fails because 'truth', the result from `ldd', has ld-linux-aarch64.so
> listed while 'needed', from the gremlin guile module doesn't:
> 
> --8<---------------cut here---------------start------------->8---
> (truth ;; result from `ldd libguile.so'
>  ("/gnu/store/...-gcc-10.3.0-lib/lib/libgcc_s.so.1"
>   "/gnu/store/...-glibc-2.33/lib/ld-linux-aarch64.so.1"   ;; This isn't
>                                                           ;; in gremlins
>   "/gnu/store/...-glibc-2.33/lib/libc.so.6"
>   "/gnu/store/...-glibc-2.33/lib/libcrypt.so.1"
>   "/gnu/store/...-glibc-2.33/lib/libdl.so.2"
>   "/gnu/store/...-glibc-2.33/lib/libm.so.6"
>   "/gnu/store/...-glibc-2.33/lib/libpthread.so.0"
>   "/gnu/store/...-guile-3.0.7/lib/libguile-3.0.so.1"
>   "/gnu/store/...-libffi-3.3/lib/libffi.so.7"
>   "/gnu/store/...-libgc-8.0.4/lib/libgc.so.1"
>   "/gnu/store/...-libunistring-0.9.10/lib/libunistring.so.2"))
> 
> (needed  ;; result from gremlin
>  ("/gnu/store/...-gcc-10.3.0-lib/lib/libgcc_s.so.1"
>   "/gnu/store/...-glibc-2.33/lib/libc.so.6"
>   "/gnu/store/...-glibc-2.33/lib/libcrypt.so.1"
>   "/gnu/store/...-glibc-2.33/lib/libdl.so.2"
>   "/gnu/store/...-glibc-2.33/lib/libm.so.6"
>   "/gnu/store/...-glibc-2.33/lib/libpthread.so.0"
>   "/gnu/store/...-guile-3.0.7/lib/libguile-3.0.so.1"
>   "/gnu/store/...-libffi-3.3/lib/libffi.so.7"
>   "/gnu/store/...-libgc-8.0.4/lib/libgc.so.1"
>   "/gnu/store/...-libunistring-0.9.10/lib/libunistring.so.2"))
> --8<---------------cut here---------------end--------------->8---
> 
> Digging a bit more I started comparing x86_64 and aarch64 binaries and
> noticed that libguile.so didn't have the same dynamic section:
> 
> --8<---------------cut here---------------start------------->8---
> # On aarch64:
> $ objdump -x 
> /gnu/store/pqw0c33k2h8n2snpchnyvx7w617kk31s-guile-3.0.7/lib/libguile-3.0.so.1.4.0
>                         
>                                                                                 
>                                                         
> /gnu/store/pqw0c33k2h8n2snpchnyvx7w617kk31s-guile-3.0.7/lib/libguile-3.0.so.1.4.0:
>      file format elf64-littleaarch64    
> ...
> Dynamic Section:                                                                
>                                                         
>   NEEDED               libgc.so.1                                               
>                                                         
>   NEEDED               libpthread.so.0                                          
>                                                         
>   NEEDED               libffi.so.7                                              
>                                                         
>   NEEDED               libunistring.so.2                                        
>                                                         
>   NEEDED               libcrypt.so.1                                            
>                                                         
>   NEEDED               libdl.so.2                                               
>                                                         
>   NEEDED               libm.so.6                                                
>                                                         
>   NEEDED               libgcc_s.so.1                                            
>                                                         
>   NEEDED               libc.so.6                                                
>                                                         
>   SONAME               libguile-3.0.so.1    
> ...
> 
> # On x86_64:
> $ objdump -x 
> /gnu/store/3h3jn0745ngd87zp83k5smwhykxvdfgf-guile-3.0.7/lib/libguile-3.0.so.1.4.0
> 
> /gnu/store/3h3jn0745ngd87zp83k5smwhykxvdfgf-guile-3.0.7/lib/libguile-3.0.so.1.4.0:
>      file format elf64-x86-64
> ...
> Dynamic Section:
>   NEEDED               libgc.so.1
>   NEEDED               libpthread.so.0
>   NEEDED               libffi.so.7
>   NEEDED               libunistring.so.2
>   NEEDED               libcrypt.so.1
>   NEEDED               libdl.so.2
>   NEEDED               libm.so.6
>   NEEDED               libgcc_s.so.1
>   NEEDED               libc.so.6
>   NEEDED               ld-linux-x86-64.so.2  # ld-linux-<arch>.so is here
>   SONAME               libguile-3.0.so.1
> ...
> --8<---------------cut here---------------end--------------->8---
> 
> On aarch64, ld-linux-<arch> is missing.  I'm not sure if this is an
> issue with our aarch64 port or if that's OK. It's interesting though
> that if you run `ldd' on libguile on aarch64, the dynamic linker is
> added to the list, even though it's not in the dynamic section:
> 
> --8<---------------cut here---------------start------------->8---
> # On aarch64
> $ ldd 
> /gnu/store/pqw0c33k2h8n2snpchnyvx7w617kk31s-guile-3.0.7/lib/libguile-3.0.so.1.4.0
>         linux-vdso.so.1 (0x0000ffff96fab000)
>         libgc.so.1 => 
> /gnu/store/1gkpbfxjx2sbchjhf19yjm4a8vkir0nm-libgc-8.0.4/lib/libgc.so.1 
> (0x0000ffff96d88000)
>         libpthread.so.0 => 
> /gnu/store/skfhr2f8b7jnnpiarrrcjn6qx0xzfw00-glibc-2.33/lib/libpthread.so.0 
> (0x0000ffff96d59000)
>         libffi.so.7 => 
> /gnu/store/hjirgm7pwmc2biqz6d0fc1ajr3ha4v14-libffi-3.3/lib/libffi.so.7 
> (0x0000ffff96d40000)
>         libunistring.so.2 => 
> /gnu/store/4k552fq1p6q73mr9ww7g5y3m77p7cfbm-libunistring-0.9.10/lib/libunistring.so.2
>  (0x0000ffff96bb4000)
>         libcrypt.so.1 => 
> /gnu/store/skfhr2f8b7jnnpiarrrcjn6qx0xzfw00-glibc-2.33/lib/libcrypt.so.1 
> (0x0000ffff96b6d000)
>         libdl.so.2 => 
> /gnu/store/skfhr2f8b7jnnpiarrrcjn6qx0xzfw00-glibc-2.33/lib/libdl.so.2 
> (0x0000ffff96b59000)
>         libm.so.6 => 
> /gnu/store/skfhr2f8b7jnnpiarrrcjn6qx0xzfw00-glibc-2.33/lib/libm.so.6 
> (0x0000ffff96ab0000)
>         libgcc_s.so.1 => 
> /gnu/store/47snyrq6pq6896m9dysp2s5vx53m6x08-gcc-10.3.0-lib/lib/libgcc_s.so.1 
> (0x0000ffff96a8b000)
>         libc.so.6 => 
> /gnu/store/skfhr2f8b7jnnpiarrrcjn6qx0xzfw00-glibc-2.33/lib/libc.so.6 
> (0x0000ffff96917000)
>         
> /gnu/store/skfhr2f8b7jnnpiarrrcjn6qx0xzfw00-glibc-2.33/lib/ld-linux-aarch64.so.1
>  (0x0000ffff96f79000)
>         # ^
> --8<---------------cut here---------------end--------------->8---
> 
> We could adapt the test to add the dynamic linker, emulating `ldd', but
> I'm curious if anybody deeply familiar with ELF and dynamic linking
> might have an idea what's going on.

To summarize, this seems to be the problem: On aarch64-linux, when an
ELF file's dynamic section does not contain a NEEDED entry for
ld-linux-aarch64.so.1, file-needed/recursive dutifully omits that entry,
but ldd prints it (even though it is actually missing from the file's
dynamic section).  This causes the file-needed/recursive test in
tests/gremlin.scm to fail because the set of entries returned by
file-needed/recursive differs from the set of entries returned by ldd.

Although this may sound similar to 52940, it is different.  Bug 52940
was an issue where, when an ELF file's dynamic section contains two
entries in RUNPATH that are different strings but refer to the same
directory (which does happen on powerpc64le-linux), it causes
file-needed/recursive to include in its result two entries that are
different strings but refer to the same file.  We decided that such
redundant entries are probably benign, so we resolved 52940 by changing
the test to compare entries by file equality, rather than string
equality.  However, this is not the same issue as described by Aiko and
Pierre above, so it makes sense that the fix for 52940 did not fix the
test for aarch64-linux.

So, the separate aarch64-linux problem with file-needed/recursive
persists.  For comparison, on powerpc64le-linux, when an ELF file's
dynamic section does not contain a NEEDED entry for "ld64.so.2" (I
believe this is the powerpc64le-linux equivalent of
ld-linux-aarch64.so.1), both ldd and file-needed/recursive include
ld64.so.2, despite the fact that there is no actual NEEDED entry for it
in the ELF file.  Here is what I see on a powerpc64le-linux system:

Toggle snippet (85 lines)marusich@suzaku ~/guix-master [env]$ type -P guile
/gnu/store/qr79b2m6cfdj8ar7g0psqg4hglm6djfm-profile/bin/guile
marusich@suzaku ~/guix-master [env]$ type -P ldd
/gnu/store/qr79b2m6cfdj8ar7g0psqg4hglm6djfm-profile/bin/ldd
marusich@suzaku ~/guix-master [env]$ ldd /gnu/store/qr79b2m6cfdj8ar7g0psqg4hglm6djfm-profile/bin/guile
        linux-vdso64.so.1 (0x00007fff89220000)
        libguile-3.0.so.1 => /gnu/store/47s31zvkhk2ixqn0z7gq6hpz7j7cn9dl-guile-3.0.7/lib/libguile-3.0.so.1 (0x00007fff89050000)
        libgc.so.1 => /gnu/store/csqz3w2axyci2qm79xsj11cpfxqh7zb1-libgc-8.0.4/lib/libgc.so.1 (0x00007fff88fb0000)
        libpthread.so.0 => /gnu/store/3jr84ajdz821y480c454pqrswxbhgzlq-glibc-2.33/lib/libpthread.so.0 (0x00007fff88f60000)
        libffi.so.7 => /gnu/store/kmgva3hbpk1bv0gvx5s4w01n0fdvd2l9-libffi-3.3/lib/../lib/libffi.so.7 (0x00007fff88f30000)
        libunistring.so.2 => /gnu/store/zf87w2xccli6yvxpplfwd82gcgm6jyrd-libunistring-0.9.10/lib/libunistring.so.2 (0x00007fff88d80000)
        libcrypt.so.1 => /gnu/store/3jr84ajdz821y480c454pqrswxbhgzlq-glibc-2.33/lib/libcrypt.so.1 (0x00007fff88d30000)
        libdl.so.2 => /gnu/store/3jr84ajdz821y480c454pqrswxbhgzlq-glibc-2.33/lib/libdl.so.2 (0x00007fff88d00000)
        libm.so.6 => /gnu/store/3jr84ajdz821y480c454pqrswxbhgzlq-glibc-2.33/lib/libm.so.6 (0x00007fff88bb0000)
        libgcc_s.so.1 => /gnu/store/l0hnwrv8ma03shjg84a03s05wmj7a0sk-gcc-10.3.0-lib/lib/libgcc_s.so.1 (0x00007fff88b70000)
        libc.so.6 => /gnu/store/3jr84ajdz821y480c454pqrswxbhgzlq-glibc-2.33/lib/libc.so.6 (0x00007fff88950000)
        /gnu/store/3jr84ajdz821y480c454pqrswxbhgzlq-glibc-2.33/lib/ld64.so.2 (0x00007fff89240000)
marusich@suzaku ~/guix-master [env]$ readelf -d /gnu/store/qr79b2m6cfdj8ar7g0psqg4hglm6djfm-profile/bin/guile

Dynamic section at offset 0xfc60 contains 37 entries:
  Tag        Type                         Name/Value
 0x0000000000000001 (NEEDED)             Shared library: [libguile-3.0.so.1]
 0x0000000000000001 (NEEDED)             Shared library: [libgc.so.1]
 0x0000000000000001 (NEEDED)             Shared library: [libpthread.so.0]
 0x0000000000000001 (NEEDED)             Shared library: [libffi.so.7]
 0x0000000000000001 (NEEDED)             Shared library: [libunistring.so.2]
 0x0000000000000001 (NEEDED)             Shared library: [libcrypt.so.1]
 0x0000000000000001 (NEEDED)             Shared library: [libdl.so.2]
 0x0000000000000001 (NEEDED)             Shared library: [libm.so.6]
 0x0000000000000001 (NEEDED)             Shared library: [libgcc_s.so.1]
 0x0000000000000001 (NEEDED)             Shared library: [libc.so.6]
 0x000000000000001d (RUNPATH)            Library runpath: [/gnu/store/47s31zvkhk2ixqn0z7gq6hpz7j7cn9dl-guile-3.0.7/lib:/gnu/store/csqz3w2axyci2qm79xsj11cpfxqh7zb1-libgc-8.0.4/lib:/gnu/store/kmgva3hbpk1bv0gvx5s4w01n0fdvd2l9-libffi-3.3/lib/../lib:/gnu/store/zf87w2xccli6yvxpplfwd82gcgm6jyrd-libunistring-0.9.10/lib:/gnu/store/3jr84ajdz821y480c454pqrswxbhgzlq-glibc-2.33/lib:/gnu/store/l0hnwrv8ma03shjg84a03s05wmj7a0sk-gcc-10.3.0-lib/lib:/gnu/store/3jr84ajdz821y480c454pqrswxbhgzlq-glibc-2.33/lib/../lib:/gnu/store/l0hnwrv8ma03shjg84a03s05wmj7a0sk-gcc-10.3.0-lib/lib/gcc/powerpc64le-unknown-linux-gnu/10.3.0/../../../../lib]
 0x000000000000000c (INIT)               0x100009e0
 0x000000000000000d (FINI)               0x10000f30
 0x0000000000000019 (INIT_ARRAY)         0x1001fc50
 0x000000000000001b (INIT_ARRAYSZ)       8 (bytes)
 0x000000000000001a (FINI_ARRAY)         0x1001fc58
 0x000000000000001c (FINI_ARRAYSZ)       8 (bytes)
 0x0000000000000004 (HASH)               0x100002a0
 0x000000006ffffef5 (GNU_HASH)           0x100002f8
 0x0000000000000005 (STRTAB)             0x100004a8
 0x0000000000000006 (SYMTAB)             0x10000328
 0x000000000000000a (STRSZ)              894 (bytes)
 0x000000000000000b (SYMENT)             24 (bytes)
 0x0000000000000015 (DEBUG)              0x0
 0x0000000000000003 (PLTGOT)             0x10020000
 0x0000000000000002 (PLTRELSZ)           216 (bytes)
 0x0000000000000014 (PLTREL)             RELA
 0x0000000000000017 (JMPREL)             0x100008e8
 0x0000000070000000 (PPC64_GLINK)        0x10000eec
 0x0000000070000003 (PPC64_OPT)          0x0
 0x0000000000000007 (RELA)               0x10000888
 0x0000000000000008 (RELASZ)             96 (bytes)
 0x0000000000000009 (RELAENT)            24 (bytes)
 0x000000006ffffffe (VERNEED)            0x10000848
 0x000000006fffffff (VERNEEDNUM)         2
 0x000000006ffffff0 (VERSYM)             0x10000826
 0x0000000000000000 (NULL)               0x0
marusich@suzaku ~/guix-master [env]$ ./pre-inst-env guix repl
GNU Guile 3.0.7
Copyright (C) 1995-2021 Free Software Foundation, Inc.

Guile comes with ABSOLUTELY NO WARRANTY; for details type `,show w'.
This program is free software, and you are welcome to redistribute it
under certain conditions; type `,show c' for details.

Enter `,help' for help.
scheme@(guix-user)> ,use (guix build gremlin)
scheme@(guix-user)> ,pp (file-needed/recursive "/gnu/store/qr79b2m6cfdj8ar7g0psqg4hglm6djfm-profile/bin/guile")
$1 = ("/gnu/store/3jr84ajdz821y480c454pqrswxbhgzlq-glibc-2.33/lib/libc.so.6"
 "/gnu/store/l0hnwrv8ma03shjg84a03s05wmj7a0sk-gcc-10.3.0-lib/lib/libgcc_s.so.1"
 "/gnu/store/3jr84ajdz821y480c454pqrswxbhgzlq-glibc-2.33/lib/libm.so.6"
 "/gnu/store/3jr84ajdz821y480c454pqrswxbhgzlq-glibc-2.33/lib/libdl.so.2"
 "/gnu/store/3jr84ajdz821y480c454pqrswxbhgzlq-glibc-2.33/lib/libcrypt.so.1"
 "/gnu/store/zf87w2xccli6yvxpplfwd82gcgm6jyrd-libunistring-0.9.10/lib/libunistring.so.2"
 "/gnu/store/kmgva3hbpk1bv0gvx5s4w01n0fdvd2l9-libffi-3.3/lib/../lib/libffi.so.7"
 "/gnu/store/3jr84ajdz821y480c454pqrswxbhgzlq-glibc-2.33/lib/libpthread.so.0"
 "/gnu/store/csqz3w2axyci2qm79xsj11cpfxqh7zb1-libgc-8.0.4/lib/libgc.so.1"
 "/gnu/store/47s31zvkhk2ixqn0z7gq6hpz7j7cn9dl-guile-3.0.7/lib/libguile-3.0.so.1"
 "/gnu/store/3jr84ajdz821y480c454pqrswxbhgzlq-glibc-2.33/lib/ld64.so.2"
 "/gnu/store/3jr84ajdz821y480c454pqrswxbhgzlq-glibc-2.33/lib/../lib/libc.so.6")
$2 = ("ld64.so.2")
scheme@(guix-user)>

I don't know if it's related, but I just noticed this: it's a little
strange that in the above output (for powerpc64le-linux), ld64.so.2 is
included in the second value ($2).  This is supposedly the list of .so
file names that could not be found.  It's strange that ld64.so.2 shows
up in $2 because it seems to contradict the fact that it is included in
$1, which is the list of files that were found successfully.

Since Pierre shared information about the libguile shared object
specifically, I'll do the same here.  On powerpc64le-linux, the
"ld64.so.2" entry is present in the dynamic section of the
/gnu/store/47s31zvkhk2ixqn0z7gq6hpz7j7cn9dl-guile-3.0.7/lib/libguile-3.0.so.1
ELF file:

Toggle snippet (16 lines)marusich@suzaku ~/guix-master [env]$ readelf -d /gnu/store/47s31zvkhk2ixqn0z7gq6hpz7j7cn9dl-guile-3.0.7/lib/libguile-3.0.so.1| grep -e NEEDED -e RUNPATH
 0x0000000000000001 (NEEDED)             Shared library: [libgc.so.1]
 0x0000000000000001 (NEEDED)             Shared library: [libpthread.so.0]
 0x0000000000000001 (NEEDED)             Shared library: [libffi.so.7]
 0x0000000000000001 (NEEDED)             Shared library: [libunistring.so.2]
 0x0000000000000001 (NEEDED)             Shared library: [libcrypt.so.1]
 0x0000000000000001 (NEEDED)             Shared library: [libdl.so.2]
 0x0000000000000001 (NEEDED)             Shared library: [libm.so.6]
 0x0000000000000001 (NEEDED)             Shared library: [libgcc_s.so.1]
 0x0000000000000001 (NEEDED)             Shared library: [libc.so.6]
 0x0000000000000001 (NEEDED)             Shared library: [ld64.so.2]
 0x000000000000001d (RUNPATH)            Library runpath: [/gnu/store/csqz3w2axyci2qm79xsj11cpfxqh7zb1-libgc-8.0.4/lib:/gnu/store/kmgva3hbpk1bv0gvx5s4w01n0fdvd2l9-libffi-3.3/lib/../lib:/gnu/store/zf87w2xccli6yvxpplfwd82gcgm6jyrd-libunistr
ing-0.9.10/lib:/gnu/store/3jr84ajdz821y480c454pqrswxbhgzlq-glibc-2.33/lib:/gnu/store/l0hnwrv8ma03shjg84a03s05wmj7a0sk-gcc-10.3.0-lib/lib:/gnu/store/3jr84ajdz821y480c454pqrswxbhgzlq-glibc-2.33/lib/../lib:/gnu/store/l0hnwrv8ma03shjg84a03s05
wmj7a0sk-gcc-10.3.0-lib/lib/gcc/powerpc64le-unknown-linux-gnu/10.3.0/../../../../lib]

Maybe this information can help somehow.

It seems fishy that on aarch64-linux, there is no NEEDED entry for

ld-linux-aarch64.so.1 in the ELF files under consideration. As shown

above, a similar entry DOES exist on both x86_64-linux and

powerpc64le-linux. For this reason, it seems plausible that maybe the

missing NEEDED entry is bad. However, I don't really know enough to say

whether it's indicative of a problem with our aarch64-linux port. Is

there an aarch64-linux or ELF expert in the room who can help? :-)

It also seems fishy that, on powerpc64le-linux, file-needed/recursive

apparently resolves ld64.so.2 successfully, even though it

simultaneously includes it in the "failed to resolve" list. Confusing

as that is to me, I don't know if that's really related to the

aarch64-linux issue.

More investigation is needed...

Chris

PGP: https://savannah.gnu.org/people/viewgpg.php?user_id=106836

Pierre Langlois wrote on 15 Jan 14:27 +0100

Recipients:(name . Chris Marusich)(address . cmmarusich@gmail.com)

Message-ID:874k65rukb.fsf@gmx.com

Hi Chris,

Chris Marusich <cmmarusich@gmail.com> writes:

(snip)

Toggle quote (16 lines)> It seems fishy that on aarch64-linux, there is no NEEDED entry for
> ld-linux-aarch64.so.1 in the ELF files under consideration.  As shown
> above, a similar entry DOES exist on both x86_64-linux and
> powerpc64le-linux.  For this reason, it seems plausible that maybe the
> missing NEEDED entry is bad.  However, I don't really know enough to say
> whether it's indicative of a problem with our aarch64-linux port.  Is
> there an aarch64-linux or ELF expert in the room who can help?  :-)
>
> It also seems fishy that, on powerpc64le-linux, file-needed/recursive
> apparently resolves ld64.so.2 successfully, even though it
> simultaneously includes it in the "failed to resolve" list.  Confusing
> as that is to me, I don't know if that's really related to the
> aarch64-linux issue.
>
> More investigation is needed...

I'm afraid I don't have any new insight if this is an issue or just
working as intended. Given we have a limited bandwidth to address this
thoroughly, would it make sense to apply a temporary work-around in the
mean time? I'd be good for at least guix to build on aarch64 in the 1.4
release.

I have the attached patch in my tree for instance (along with an update
of the guix package), and I've not noticed any issues on a rockpro64,
running cgit, DHCP and dnsmasq. That's just anecdotal :-), but I'm also
thinking if we unblock the guix package then the farm might catch other
issues that could help getting to the bottom of this.

WDYT?

Thanks,
Pierre

From cdd2c995b7c03400a9702a6b354655b0584b727c Mon Sep 17 00:00:00 2001

---

tests/gremlin.scm | 11 +++++++++--

1 file changed, 9 insertions(+), 2 deletions(-)

Toggle diff (31 lines)diff --git a/tests/gremlin.scm b/tests/gremlin.scm
index 9e0017337a..c9b05ae21a 100644
--- a/tests/gremlin.scm
+++ b/tests/gremlin.scm
@@ -20,9 +20,11 @@
 (define-module (test-gremlin)
   #:use-module (guix elf)
   #:use-module (guix tests)
-  #:use-module ((guix utils) #:select (call-with-temporary-directory))
+  #:use-module ((guix utils) #:select (call-with-temporary-directory
+                                       target-aarch64?))
   #:use-module (guix build utils)
   #:use-module (guix build gremlin)
+  #:use-module (gnu packages bootstrap)
   #:use-module (srfi srfi-1)
   #:use-module (srfi srfi-26)
   #:use-module (srfi srfi-34)
@@ -99,7 +101,12 @@ (define ground-truth
                 (or (string-prefix? "linux-vdso.so" entry)
                     (string-prefix? "linux-vdso32.so" entry) ;32-bit powerpc
                     (string-prefix? "linux-vdso64.so" entry) ;64-bit powerpc
-                    (string-prefix? "linux-gate.so" entry))) ;i386
+                    (string-prefix? "linux-gate.so" entry)   ;i386
+                    ;; FIXME: Binaries on aarch64 do not always include the
+                    ;; dynamic linker in their NEEDED section, it's unclear if that's OK, see
+                    ;; ttps://bugs.gnu.org/52943.
+                    (and (target-aarch64?)
+                         (string-contains entry (glibc-dynamic-linker)))))
               (read-ldd-output pipe)))

     (and (zero? (close-pipe pipe))

2.34.0

Chris Marusich wrote on 16 Jan 02:00 +0100

Recipients:(name . Pierre Langlois)(address . pierre.langlois@gmx.com)

Message-ID:87czksa40e.fsf@gmail.com

Hi Pierre,

FYI, it looks like you included bug-guix@gnu.org in the CC list of your

last email. If you do that, I believe it will create a new bug report,

which may not be what you intended to do.

Pierre Langlois <pierre.langlois@gmx.com> writes:

Toggle quote (14 lines)> I'm afraid I don't have any new insight if this is an issue or just
> working as intended. Given we have a limited bandwidth to address this
> thoroughly, would it make sense to apply a temporary work-around in the
> mean time? I'd be good for at least guix to build on aarch64 in the 1.4
> release.
>
> I have the attached patch in my tree for instance (along with an update
> of the guix package), and I've not noticed any issues on a rockpro64,
> running cgit, DHCP and dnsmasq. That's just anecdotal :-), but I'm also
> thinking if we unblock the guix package then the farm might catch other
> issues that could help getting to the bottom of this.
>
> WDYT?

I agree with you. That said, I don't personally run any aarch64-linux

machines, and I don't plan to investigate the aarch64-linux issue

further right now.

Although I will defer to anyone who actually runs aarch64-linux and

knows better, at the moment to me it seems reasonable to commit a patch

like the one you've proposed in order to ensure that the "guix" package

builds successfully on that platform for the 1.4.0 release. The fact

that you have successfully built and used various pieces of software on

aarch64-linux with this patch suggests that maybe the current behavior

is OK, after all.

I've made some rather trivial modifications to your patch, mainly to

make it conform to our required ChangeLog format. If nobody has further

comments, I will commit the attached version to master in about 24

hours.

Chris

PGP: https://savannah.gnu.org/people/viewgpg.php?user_id=106836

From f502bab1e09276982acfd3ac0c151b241f7aa07d Mon Sep 17 00:00:00 2001

Fixes: https://issues.guix.gnu.org/52943.

* tests/gremlin.scm (file-needed/recursive)[ground-truth]: On aarch64-linux,

remove the dynamic linker from this list.

---

tests/gremlin.scm | 12 ++++++++++--

1 file changed, 10 insertions(+), 2 deletions(-)

Toggle diff (43 lines)diff --git a/tests/gremlin.scm b/tests/gremlin.scm
index 9e0017337a..3dbb8d3643 100644
--- a/tests/gremlin.scm
+++ b/tests/gremlin.scm
@@ -1,6 +1,7 @@
 ;;; GNU Guix --- Functional package management for GNU
 ;;; Copyright © 2015, 2018, 2020, 2022 Ludovic Courtès <ludo@gnu.org>
 ;;; Copyright © 2022 Chris Marusich <cmmarusich@gmail.com>
+;;; Copyright © 2022 Pierre Langlois <pierre.langlois@gmx.com>
 ;;;
 ;;; This file is part of GNU Guix.
 ;;;
@@ -20,9 +21,11 @@
 (define-module (test-gremlin)
   #:use-module (guix elf)
   #:use-module (guix tests)
-  #:use-module ((guix utils) #:select (call-with-temporary-directory))
+  #:use-module ((guix utils) #:select (call-with-temporary-directory
+                                       target-aarch64?))
   #:use-module (guix build utils)
   #:use-module (guix build gremlin)
+  #:use-module (gnu packages bootstrap)
   #:use-module (srfi srfi-1)
   #:use-module (srfi srfi-26)
   #:use-module (srfi srfi-34)
@@ -99,7 +102,12 @@ (define ground-truth
                 (or (string-prefix? "linux-vdso.so" entry)
                     (string-prefix? "linux-vdso32.so" entry) ;32-bit powerpc
                     (string-prefix? "linux-vdso64.so" entry) ;64-bit powerpc
-                    (string-prefix? "linux-gate.so" entry))) ;i386
+                    (string-prefix? "linux-gate.so" entry)   ;i386
+                    ;; FIXME: ELF files on aarch64 do not always include a
+                    ;; NEEDED entry for the dynamic linker, and it is unclear
+                    ;; if that is OK.  See: https://issues.guix.gnu.org/52943
+                    (and (target-aarch64?)
+                         (string-contains entry (glibc-dynamic-linker)))))
               (read-ldd-output pipe)))
 
     (and (zero? (close-pipe pipe))

base-commit: 172bd0b5cde2609389fd16d18862b5b612c4b000
-- 
2.33.1

Chris Marusich wrote on 16 Jan 02:02 +0100

(no subject)

Recipients:(address . control@debbugs.gnu.org)

Message-ID:878rvga3wt.fsf@gmail.com

# "guix" package fails to build on aarch64 on master, thus "guix pull"

# will also fail.

block 53214 by 52943

Chris

PGP: https://savannah.gnu.org/people/viewgpg.php?user_id=106836

Chris Marusich wrote on 17 Jan 05:46 +0100

Re: bug#52943: Cannot build guix as part of guix system reconfigure after commit 224d437fb4 on aarch64

Recipients:(name . Pierre Langlois)(address . pierre.langlois@gmx.com)

Message-ID:87y23f7yw1.fsf@gmail.com

Hi,

Chris Marusich <cmmarusich@gmail.com> writes:

Toggle quote (3 lines)

> If nobody has further comments, I will commit the attached version to

> master in about 24 hours.

I've committed this in 24c3485bb3ffc05e687ef6513ac287b8d3048bab. We

still need to update the "guix" package, so even if you run "guix pull"

right now, you won't be able to successfully build "guix" after pulling

on aarch64-linux yet.

I will coordinate with the other people who are helping to make the

1.4.0 release to ensure that this fix makes it into the 1.4.0 release.

See this guix-devel email thread for details:

https://lists.gnu.org/archive/html/guix-devel/2022-01/msg00296.html

Chris

PGP: https://savannah.gnu.org/people/viewgpg.php?user_id=106836

Pierre Langlois wrote on 17 Jan 19:51 +0100

Recipients:(name . Chris Marusich)(address . cmmarusich@gmail.com)

Message-ID:878rvefb54.fsf@gmx.com

Hi,

Chris Marusich <cmmarusich@gmail.com> writes:

Toggle quote (19 lines)> [[PGP Signed Part:Undecided]]
> Hi,
>
> Chris Marusich <cmmarusich@gmail.com> writes:
>
>> If nobody has further comments, I will commit the attached version to
>> master in about 24 hours.
>
> I've committed this in 24c3485bb3ffc05e687ef6513ac287b8d3048bab.  We
> still need to update the "guix" package, so even if you run "guix pull"
> right now, you won't be able to successfully build "guix" after pulling
> on aarch64-linux yet.
>
> I will coordinate with the other people who are helping to make the
> 1.4.0 release to ensure that this fix makes it into the 1.4.0 release.
> See this guix-devel email thread for details:
>
>   https://lists.gnu.org/archive/html/guix-devel/2022-01/msg00296.html

That's awesome, thank you!

Pierre

Vagrant Cascadian wrote on 18 Jan 06:33 +0100

Recipients:

Message-ID:87bl09d2wi.fsf@yucca

On 2022-01-17, Pierre Langlois wrote:

Toggle quote (19 lines)> Chris Marusich <cmmarusich@gmail.com> writes:
>> Chris Marusich <cmmarusich@gmail.com> writes:

>>> If nobody has further comments, I will commit the attached version to
>>> master in about 24 hours.
>>
>> I've committed this in 24c3485bb3ffc05e687ef6513ac287b8d3048bab.  We
>> still need to update the "guix" package, so even if you run "guix pull"
>> right now, you won't be able to successfully build "guix" after pulling
>> on aarch64-linux yet.
>>
>> I will coordinate with the other people who are helping to make the
>> 1.4.0 release to ensure that this fix makes it into the 1.4.0 release.
>> See this guix-devel email thread for details:
>>
>>   https://lists.gnu.org/archive/html/guix-devel/2022-01/msg00296.html
>
> That's awesome, thank you!

FWIW, I can also confirm that updating the "guix" package to a version

that contains 24c3485bb3ffc05e687ef6513ac287b8d3048bab allows the "guix"

package (which contains the fairly important "guix-daemon"), to build

again. Thanks for getting it that far!

Haven't been able to upgrade my aarch64 systems without local patches

for over a month now... glad to see a fix is *almost* ready!

Is there anything in particular holding back upating the guix package

(other than the huge amount of big merges in progress, release

preparation, etc. and etc.)?

Thanks everyone!

live well,

vagrant

-----BEGIN PGP SIGNATURE-----

iHUEARYKAB0WIQRlgHNhO/zFx+LkXUXcUY/If5cWqgUCYeZRDgAKCRDcUY/If5cW

qkRZAQCUce0XWZ6VxG+eGmNA7h6bj5MGJ0VjfTmVdY9hgoWFQwEApzoFpZecnBaM

54+wP5hdHXUuOUhFQ7fwachLllWajAQ=

=iOg/

-----END PGP SIGNATURE-----

Vagrant Cascadian wrote on 19 Jan 20:28 +0100

Recipients:

Message-ID:87k0eva5k3.fsf@yucca

On 2022-01-17, Vagrant Cascadian wrote:

Toggle quote (32 lines)> On 2022-01-17, Pierre Langlois wrote:
>> Chris Marusich <cmmarusich@gmail.com> writes:
>>> Chris Marusich <cmmarusich@gmail.com> writes:
>
>>>> If nobody has further comments, I will commit the attached version to
>>>> master in about 24 hours.
>>>
>>> I've committed this in 24c3485bb3ffc05e687ef6513ac287b8d3048bab.  We
>>> still need to update the "guix" package, so even if you run "guix pull"
>>> right now, you won't be able to successfully build "guix" after pulling
>>> on aarch64-linux yet.
>>>
>>> I will coordinate with the other people who are helping to make the
>>> 1.4.0 release to ensure that this fix makes it into the 1.4.0 release.
>>> See this guix-devel email thread for details:
>>>
>>>   https://lists.gnu.org/archive/html/guix-devel/2022-01/msg00296.html
>>
>> That's awesome, thank you!
>
> FWIW, I can also confirm that updating the "guix" package to a version
> that contains 24c3485bb3ffc05e687ef6513ac287b8d3048bab allows the "guix"
> package (which contains the fairly important "guix-daemon"), to build
> again. Thanks for getting it that far!
>
> Haven't been able to upgrade my aarch64 systems without local patches
> for over a month now... glad to see a fix is *almost* ready!
>
> Is there anything in particular holding back upating the guix package
> (other than the huge amount of big merges in progress, release
> preparation, etc. and etc.)?

I went ahead and pushed to master:

f758ede583ef9662963873750206540f58ff3c45 gnu: guix: update to 1.3.0-22.24c3485.

I tried building a newer version, but there were new test suite failures

on both aarch64 and x86_64 :/

Happy aarch64ing!

live well,

vagrant

-----BEGIN PGP SIGNATURE-----

iHUEARYKAB0WIQRlgHNhO/zFx+LkXUXcUY/If5cWqgUCYehmXAAKCRDcUY/If5cW

qjknAQC5mqWy3lTzwBCAJgtRXThKl2if9gwiSffnhdCW+V1AlQEAq8KG+P671uaq

t6ueyOG3j/cNVblfBMs+PnircgQPFQs=

=ybvY

-----END PGP SIGNATURE-----

Closed

Leo Famulari wrote on 29 Jan 19:33 +0100

Recipients:

Message-ID:YfWId5SVaChgck4O@jasmine.lan

On Wed, Jan 19, 2022 at 11:28:28AM -0800, Vagrant Cascadian wrote:

Toggle quote (3 lines)

> I tried building a newer version, but there were new test suite failures

> on both aarch64 and x86_64 :/

Since the 'guix' package still does not build on aarch64, I'm reopening

this bug.

Leo Famulari wrote on 29 Jan 19:36 +0100

(no subject)

Recipients:(address . control@debbugs.gnu.org)

Message-ID:YfWJQ+BQBuvTlSQg@jasmine.lan

merge 53616 52943

Pierre Langlois wrote on 29 Jan 21:54 +0100

Re: bug#52943: Cannot build guix as part of guix system reconfigure after commit 224d437fb4 on aarch64

Recipients:(name . Leo Famulari)(address . leo@famulari.name)

Message-ID:87r18qb6l2.fsf@gmx.com

Hi Leo,

Leo Famulari <leo@famulari.name> writes:

Toggle quote (7 lines)

> On Wed, Jan 19, 2022 at 11:28:28AM -0800, Vagrant Cascadian wrote:

>> I tried building a newer version, but there were new test suite failures

>> on both aarch64 and x86_64 :/

> Since the 'guix' package still does not build on aarch64, I'm reopening

> this bug.

Oh it doesn't? What hash are you on?

I'm currently on b14a1cdef571c0901f716f4e752a97e8028ccbbd on a RockPro64

and it's working OK, maybe something changed since then although this

hash isn't very old. I see the farm does not have any substitutes yet.

Thanks,

Pierre

Leo Famulari wrote on 29 Jan 22:06 +0100

Recipients:(name . Pierre Langlois)(address . pierre.langlois@gmx.com)

Message-ID:YfWsUoWtOhMDlLm9@jasmine.lan

On Sat, Jan 29, 2022 at 08:54:24PM +0000, Pierre Langlois wrote:

Toggle quote (5 lines)

> > Since the 'guix' package still does not build on aarch64, I'm reopening

> > this bug.

> Oh it doesn't? What hash are you on?

I'm not using aarch64, so I can't give a commit hash. I am observing the

build farm's support for aarch64.

Toggle quote (4 lines)

> I'm currently on b14a1cdef571c0901f716f4e752a97e8028ccbbd on a RockPro64

> and it's working OK, maybe something changed since then although this

> hash isn't very old. I see the farm does not have any substitutes yet.

As you point out, the build farm cannot build it:

https://ci.guix.gnu.org/search?query=spec%3Amaster+system%3Aaarch64-linux+guix-1.3.0

Maybe it's a problem with in the 'guix' package, maybe in the some other

package. But the effect for users is the same, and they report things

like this:

https://issues.guix.gnu.org/53616

Pierre Langlois wrote on 29 Jan 22:35 +0100

Recipients:(name . Leo Famulari)(address . leo@famulari.name)

Message-ID:87zgne9q2g.fsf@gmx.com

Leo Famulari <leo@famulari.name> writes:

Toggle quote (23 lines)> On Sat, Jan 29, 2022 at 08:54:24PM +0000, Pierre Langlois wrote:
>> > Since the 'guix' package still does not build on aarch64, I'm reopening
>> > this bug.
>> 
>> Oh it doesn't? What hash are you on?
>
> I'm not using aarch64, so I can't give a commit hash. I am observing the
> build farm's support for aarch64.
>
>> I'm currently on b14a1cdef571c0901f716f4e752a97e8028ccbbd on a RockPro64
>> and it's working OK, maybe something changed since then although this
>> hash isn't very old. I see the farm does not have any substitutes yet.
>
> As you point out, the build farm cannot build it:
>
> https://ci.guix.gnu.org/search?query=spec%3Amaster+system%3Aaarch64-linux+guix-1.3.0
>
> Maybe it's a problem with in the 'guix' package, maybe in the some other
> package. But the effect for users is the same, and they report things
> like this:
>
> https://issues.guix.gnu.org/53616

Oh I see, I'm not familiar with how the infra works, but it looks like

aarch64 jobs aren't currently scheduled at all: https://ci.guix.gnu.org/build/337714/details

Maybe we could restart it and see if we can reproduce the problem? I

wonder if there's a good reason for jobs not being scheduled, I saw

recent commits regarding new aarch64 HW being set up.

I'll be happy to help debug build issues when we know more of course!

Thanks,

Pierre

Ludovic Courtès wrote on 1 Feb 09:21 +0100

control message for bug #52943

Recipients:(address . control@debbugs.gnu.org)

Message-ID:875ypz80b0.fsf@gnu.org

severity 52943 important

quit

is:open	open issues
is:done	closed issues
submitter:<who>	search issue submitter
author:<who>	search by message author
date:yesterday..now	search by issue date
mdate:3m..2d	search by message date