* [Buildroot] [PATCH 1/1] package/systemd: bump to version 251.3 @ 2022-07-19 13:04 Norbert Lange 2022-07-19 16:21 ` Yann E. MORIN 0 siblings, 1 reply; 6+ messages in thread From: Norbert Lange @ 2022-07-19 13:04 UTC (permalink / raw) To: buildroot; +Cc: Norbert Lange, Yann E. MORIN Add config option for the new sysupdate feature. Remove upstreamed patch. Update options, manually setdefault options, remove option not available anymore. Set default-user-shell based on wether a couple packages are enabled. Set default-locale to C, this seems the only safe default, users can override this in `/etc/locale.conf`. Signed-off-by: Norbert Lange <nolange79@gmail.com> --- ...define-MOVE_MOUNT_T_EMPTY_PATH-if-mi.patch | 33 ------------------- package/systemd/Config.in | 12 +++++++ package/systemd/systemd.hash | 4 +-- package/systemd/systemd.mk | 20 +++++++++-- 4 files changed, 32 insertions(+), 37 deletions(-) delete mode 100644 package/systemd/0001-missing-syscall-define-MOVE_MOUNT_T_EMPTY_PATH-if-mi.patch diff --git a/package/systemd/0001-missing-syscall-define-MOVE_MOUNT_T_EMPTY_PATH-if-mi.patch b/package/systemd/0001-missing-syscall-define-MOVE_MOUNT_T_EMPTY_PATH-if-mi.patch deleted file mode 100644 index 0934886acb..0000000000 --- a/package/systemd/0001-missing-syscall-define-MOVE_MOUNT_T_EMPTY_PATH-if-mi.patch +++ /dev/null @@ -1,33 +0,0 @@ -From 16c132a6a0389e0eaea70c4ad95dbfd1637ec5ba Mon Sep 17 00:00:00 2001 -From: Romain Naour <romain.naour@smile.fr> -Date: Fri, 7 Jan 2022 22:25:23 +0100 -Subject: [PATCH] missing-syscall: define MOVE_MOUNT_T_EMPTY_PATH if missing - -MOVE_MOUNT_T_EMPTY_PATH has been added to systemd 250 by [1] -but it's defined in kernel headers since version 5.2. - -[1] c7bf079bbc19e3b409acc0c7acc3e14749211fe2 - -Signed-off-by: Romain Naour <romain.naour@smile.fr> ---- - src/basic/missing_syscall.h | 4 ++++ - 1 file changed, 4 insertions(+) - -diff --git a/src/basic/missing_syscall.h b/src/basic/missing_syscall.h -index 8267b1a90c..793d111c55 100644 ---- a/src/basic/missing_syscall.h -+++ b/src/basic/missing_syscall.h -@@ -569,6 +569,10 @@ static inline int missing_open_tree( - #define MOVE_MOUNT_F_EMPTY_PATH 0x00000004 /* Empty from path permitted */ - #endif - -+#ifndef MOVE_MOUNT_T_EMPTY_PATH -+#define MOVE_MOUNT_T_EMPTY_PATH 0x00000040 /* Empty to path permitted */ -+#endif -+ - static inline int missing_move_mount( - int from_dfd, - const char *from_pathname, --- -2.31.1 - diff --git a/package/systemd/Config.in b/package/systemd/Config.in index d5d88404de..9f1a778e59 100644 --- a/package/systemd/Config.in +++ b/package/systemd/Config.in @@ -475,6 +475,18 @@ config BR2_PACKAGE_SYSTEMD_SYSEXT https://www.freedesktop.org/software/systemd/man/systemd-sysext.html +config BR2_PACKAGE_SYSTEMD_SYSUPDATE + bool "enable sysupdate support" + select BR2_PACKAGE_OPENSSL + select BR2_PACKAGE_OPENSSL_FORCE_LIBOPENSSL + select BR2_PACKAGE_UTIL_LINUX_LIBFDISK + help + systemd-sysupdate atomically updates the host OS, container + images, portable service images or other sources, based on + the transfer configuration files described in sysupdate.d. + + https://www.freedesktop.org/software/systemd/man/systemd-sysupdate.html + config BR2_PACKAGE_SYSTEMD_SYSUSERS bool "enable sysusers support" help diff --git a/package/systemd/systemd.hash b/package/systemd/systemd.hash index 3572b25965..838ed13fd5 100644 --- a/package/systemd/systemd.hash +++ b/package/systemd/systemd.hash @@ -1,5 +1,5 @@ # sha256 locally computed -sha256 d2bda9d225da11dc9ff48b48e59fc36798d3e66902ed400a9f78fa370c596864 systemd-250.4.tar.gz +sha256 11c9c3fcc13d689615dd2ea79161d1b992087cf1e2a23856d7d7ade0efbddc54 systemd-251.3.tar.gz sha256 ab15fd526bd8dd18a9e77ebc139656bf4d33e97fc7238cd11bf60e2b9b8666c6 LICENSE.GPL2 sha256 dc626520dcd53a22f727af3ee42c770e56c97a64fe3adb063799d8ab032fe551 LICENSE.LGPL2.1 sha256 e5a8645ad94aab24e312dd0c6be2aa54236eb9374480b1b14ea5c61598874fd5 LICENSES/BSD-2-Clause.txt @@ -11,4 +11,4 @@ sha256 790ac93fb2859097bdda4cf08b5a4feb5e479d0cb2c74f403248241bc3e7c216 LICENS sha256 b85dcd3e453d05982552c52b5fc9e0bdd6d23c6f8e844b984a88af32570b0cc0 LICENSES/MIT.txt sha256 2a98749f6bec00dfaed86fa9c1edea871aaae4f11ee4d100b671aaf4ce353a73 LICENSES/murmurhash2-public-domain.txt sha256 1a7adaa2c86cedfd6c7f5c0c7c72fd6d3e02cd0c9593f21fdb53c89bb2b130ec LICENSES/OFL-1.1.txt -sha256 11801e931f252252a16eac8299465510d0a82c36bfd9ac8aea9b202b76d2f82b LICENSES/README.md +sha256 177ff657698bc940f9888b7f69e5d734b0933c2859ea264fde8de1700cec40dd LICENSES/README.md diff --git a/package/systemd/systemd.mk b/package/systemd/systemd.mk index 46a4e8de2c..b6fffc553f 100644 --- a/package/systemd/systemd.mk +++ b/package/systemd/systemd.mk @@ -19,7 +19,7 @@ # - Diff sysusers.d with the previous version # - Diff factory/etc/nsswitch.conf with the previous version # (details are often sprinkled around in README and manpages) -SYSTEMD_VERSION = 250.4 +SYSTEMD_VERSION = 251.3 SYSTEMD_SITE = $(call github,systemd,systemd-stable,v$(SYSTEMD_VERSION)) SYSTEMD_LICENSE = \ LGPL-2.1+, \ @@ -62,6 +62,8 @@ SYSTEMD_PROVIDES = udev SYSTEMD_CONF_OPTS += \ -Ddefault-hierarchy=unified \ + -Ddefault-compression=auto \ + -Ddefault-locale='C' \ -Didn=true \ -Dima=false \ -Dkexec-path=/usr/sbin/kexec \ @@ -472,6 +474,20 @@ else SYSTEMD_CONF_OPTS += -Dsysext=false endif +ifeq ($(BR2_PACKAGE_SYSTEMD_SYSUPDATE),y) +SYSTEMD_CONF_OPTS += -Dsysupdate=true +else +SYSTEMD_CONF_OPTS += -Dsysupdate=false +endif + +ifeq ($(BR2_PACKAGE_BASH),y) +SYSTEMD_CONF_OPTS += -Ddefault-user-shell=/bin/bash +else ifeq ($(BR2_PACKAGE_ZSH),y) +SYSTEMD_CONF_OPTS += -Ddefault-user-shell=/bin/zsh +else +SYSTEMD_CONF_OPTS += -Ddefault-user-shell=/bin/sh +endif + ifeq ($(BR2_PACKAGE_SYSTEMD_NETWORKD),y) SYSTEMD_CONF_OPTS += -Dnetworkd=true SYSTEMD_NETWORKD_USER = systemd-network -1 systemd-network -1 * - - - systemd Network Management @@ -546,7 +562,6 @@ SYSTEMD_DEPENDENCIES += gnu-efi SYSTEMD_CONF_OPTS += \ -Defi=true \ -Dgnu-efi=true \ - -Defi-cc=$(TARGET_CC) \ -Defi-ld=bfd \ -Defi-libdir=$(STAGING_DIR)/usr/lib \ -Defi-includedir=$(STAGING_DIR)/usr/include/efi @@ -797,6 +812,7 @@ HOST_SYSTEMD_CONF_OPTS = \ -Dmachined=false \ -Dportabled=false \ -Dsysext=false \ + -Dsysupdate=false \ -Duserdb=false \ -Dhomed=false \ -Dnetworkd=false \ -- 2.35.1 _______________________________________________ buildroot mailing list buildroot@buildroot.org https://lists.buildroot.org/mailman/listinfo/buildroot ^ permalink raw reply related [flat|nested] 6+ messages in thread
* Re: [Buildroot] [PATCH 1/1] package/systemd: bump to version 251.3 2022-07-19 13:04 [Buildroot] [PATCH 1/1] package/systemd: bump to version 251.3 Norbert Lange @ 2022-07-19 16:21 ` Yann E. MORIN 2022-07-19 21:56 ` James Hilliard 0 siblings, 1 reply; 6+ messages in thread From: Yann E. MORIN @ 2022-07-19 16:21 UTC (permalink / raw) To: Norbert Lange; +Cc: buildroot Norbert, All, On 2022-07-19 15:04 +0200, Norbert Lange spake thusly: > Add config option for the new sysupdate feature. Honestly, with a package so complex and critical like systemd, I would prefer that new features not be added during the bump, but as followup patches. The bump should add a force-disable of new features (when known), or just leave them as auto-detected or upstream-decided defaults. > Remove upstreamed patch. > > Update options, manually setdefault options, remove option Sorry, I don't immediately see what "manually setdefault" means... Is there a missing space, and this should be "manually set default" ? Or is "setdefault" and actual systemd concept? (genuinely asking) > not available anymore. > > Set default-user-shell based on wether a couple packages > are enabled. > > Set default-locale to C, this seems the only safe default, How does that play with BR2_ENABLE_LOCALE_WHITELIST? I think that we would want to use the first word of BR2_ENABLE_LOCALE_WHITELIST is it is set, falling back to just 'C' otherwise. Furthermore, that is not a novelty with v251; v250 already had that option, so that should be a separate, preliminary patch. > users can override this in `/etc/locale.conf`. > Signed-off-by: Norbert Lange <nolange79@gmail.com> > --- [--SNIP--] > diff --git a/package/systemd/systemd.hash b/package/systemd/systemd.hash > index 3572b25965..838ed13fd5 100644 > --- a/package/systemd/systemd.hash > +++ b/package/systemd/systemd.hash > @@ -1,5 +1,5 @@ > # sha256 locally computed > -sha256 d2bda9d225da11dc9ff48b48e59fc36798d3e66902ed400a9f78fa370c596864 systemd-250.4.tar.gz > +sha256 11c9c3fcc13d689615dd2ea79161d1b992087cf1e2a23856d7d7ade0efbddc54 systemd-251.3.tar.gz > sha256 ab15fd526bd8dd18a9e77ebc139656bf4d33e97fc7238cd11bf60e2b9b8666c6 LICENSE.GPL2 > sha256 dc626520dcd53a22f727af3ee42c770e56c97a64fe3adb063799d8ab032fe551 LICENSE.LGPL2.1 > sha256 e5a8645ad94aab24e312dd0c6be2aa54236eb9374480b1b14ea5c61598874fd5 LICENSES/BSD-2-Clause.txt > @@ -11,4 +11,4 @@ sha256 790ac93fb2859097bdda4cf08b5a4feb5e479d0cb2c74f403248241bc3e7c216 LICENS > sha256 b85dcd3e453d05982552c52b5fc9e0bdd6d23c6f8e844b984a88af32570b0cc0 LICENSES/MIT.txt > sha256 2a98749f6bec00dfaed86fa9c1edea871aaae4f11ee4d100b671aaf4ce353a73 LICENSES/murmurhash2-public-domain.txt > sha256 1a7adaa2c86cedfd6c7f5c0c7c72fd6d3e02cd0c9593f21fdb53c89bb2b130ec LICENSES/OFL-1.1.txt > -sha256 11801e931f252252a16eac8299465510d0a82c36bfd9ac8aea9b202b76d2f82b LICENSES/README.md > +sha256 177ff657698bc940f9888b7f69e5d734b0933c2859ea264fde8de1700cec40dd LICENSES/README.md What has changed in that license file? How does it impact (or not) the licensing info we carry? We have hashes for license files, so that we know they licensing terms change when a package is updated. Silently changing the hash kind of defeats the purpose for those hashes... > diff --git a/package/systemd/systemd.mk b/package/systemd/systemd.mk > index 46a4e8de2c..b6fffc553f 100644 > --- a/package/systemd/systemd.mk > +++ b/package/systemd/systemd.mk [--SNIP--] > @@ -472,6 +474,20 @@ else > SYSTEMD_CONF_OPTS += -Dsysext=false > endif > > +ifeq ($(BR2_PACKAGE_SYSTEMD_SYSUPDATE),y) > +SYSTEMD_CONF_OPTS += -Dsysupdate=true > +else > +SYSTEMD_CONF_OPTS += -Dsysupdate=false > +endif > + > +ifeq ($(BR2_PACKAGE_BASH),y) > +SYSTEMD_CONF_OPTS += -Ddefault-user-shell=/bin/bash > +else ifeq ($(BR2_PACKAGE_ZSH),y) > +SYSTEMD_CONF_OPTS += -Ddefault-user-shell=/bin/zsh > +else > +SYSTEMD_CONF_OPTS += -Ddefault-user-shell=/bin/sh > +endif We already have BR2_SYSTEM_BIN_SH which defines what /bin/sh is, and so I think we should use exactly /bin/sh as the default shell: ifneq ($(BR2_SYSTEM_BIN_SH_NONE),y) SYSTEMD_CONF_OPTS += -Ddefault-user-shell=/bin/sh endif If there is not default shell for the system, then there is none for users as well. Systemd-defined users that want a specific shell will have to explicitly request it: if they have a requirement for a specific shell, they can't rely on whatever the default is, so usng /bin/sh is exactly what we want. If BR2_SYSTEM_BIN_SH is empty, it is perfectly possible that there is no shell available at all (no standalone shell, no shell in busybox), so we don't want to set any default either. Regards, Yann E. MORIN. -- .-----------------.--------------------.------------------.--------------------. | Yann E. MORIN | Real-Time Embedded | /"\ ASCII RIBBON | Erics' conspiracy: | | +33 662 376 056 | Software Designer | \ / CAMPAIGN | ___ | | +33 561 099 427 `------------.-------: X AGAINST | \e/ There is no | | http://ymorin.is-a-geek.org/ | _/*\_ | / \ HTML MAIL | v conspiracy. | '------------------------------^-------^------------------^--------------------' _______________________________________________ buildroot mailing list buildroot@buildroot.org https://lists.buildroot.org/mailman/listinfo/buildroot ^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [Buildroot] [PATCH 1/1] package/systemd: bump to version 251.3 2022-07-19 16:21 ` Yann E. MORIN @ 2022-07-19 21:56 ` James Hilliard 2022-07-20 14:51 ` Norbert Lange 0 siblings, 1 reply; 6+ messages in thread From: James Hilliard @ 2022-07-19 21:56 UTC (permalink / raw) To: Yann E. MORIN; +Cc: Norbert Lange, buildroot FYI I had sent a similar bump earlier: https://patchwork.ozlabs.org/project/buildroot/patch/20220522011729.117324-1-james.hilliard1@gmail.com/ On Tue, Jul 19, 2022 at 10:22 AM Yann E. MORIN <yann.morin.1998@free.fr> wrote: > > Norbert, All, > > On 2022-07-19 15:04 +0200, Norbert Lange spake thusly: > > Add config option for the new sysupdate feature. > > Honestly, with a package so complex and critical like systemd, I would > prefer that new features not be added during the bump, but as followup > patches. The bump should add a force-disable of new features (when > known), or just leave them as auto-detected or upstream-decided > defaults. > > > Remove upstreamed patch. > > > > Update options, manually setdefault options, remove option > > Sorry, I don't immediately see what "manually setdefault" means... > Is there a missing space, and this should be "manually set default" ? > Or is "setdefault" and actual systemd concept? > > (genuinely asking) > > > not available anymore. > > > > Set default-user-shell based on wether a couple packages > > are enabled. > > > > Set default-locale to C, this seems the only safe default, > > How does that play with BR2_ENABLE_LOCALE_WHITELIST? I think that we > would want to use the first word of BR2_ENABLE_LOCALE_WHITELIST is it is > set, falling back to just 'C' otherwise. Yeah, we should use 'C.UTF-8' since it's now supported by localedef, I had sent a patch earlier changing the default but there were uclibc compatibility issues, since systemd is incompatible with uclibc there's probably a better approach: https://patchwork.ozlabs.org/project/buildroot/patch/20220517192439.1034654-1-james.hilliard1@gmail.com/ > > Furthermore, that is not a novelty with v251; v250 already had that > option, so that should be a separate, preliminary patch. > > > users can override this in `/etc/locale.conf`. > > > Signed-off-by: Norbert Lange <nolange79@gmail.com> > > --- > [--SNIP--] > > diff --git a/package/systemd/systemd.hash b/package/systemd/systemd.hash > > index 3572b25965..838ed13fd5 100644 > > --- a/package/systemd/systemd.hash > > +++ b/package/systemd/systemd.hash > > @@ -1,5 +1,5 @@ > > # sha256 locally computed > > -sha256 d2bda9d225da11dc9ff48b48e59fc36798d3e66902ed400a9f78fa370c596864 systemd-250.4.tar.gz > > +sha256 11c9c3fcc13d689615dd2ea79161d1b992087cf1e2a23856d7d7ade0efbddc54 systemd-251.3.tar.gz > > sha256 ab15fd526bd8dd18a9e77ebc139656bf4d33e97fc7238cd11bf60e2b9b8666c6 LICENSE.GPL2 > > sha256 dc626520dcd53a22f727af3ee42c770e56c97a64fe3adb063799d8ab032fe551 LICENSE.LGPL2.1 > > sha256 e5a8645ad94aab24e312dd0c6be2aa54236eb9374480b1b14ea5c61598874fd5 LICENSES/BSD-2-Clause.txt > > @@ -11,4 +11,4 @@ sha256 790ac93fb2859097bdda4cf08b5a4feb5e479d0cb2c74f403248241bc3e7c216 LICENS > > sha256 b85dcd3e453d05982552c52b5fc9e0bdd6d23c6f8e844b984a88af32570b0cc0 LICENSES/MIT.txt > > sha256 2a98749f6bec00dfaed86fa9c1edea871aaae4f11ee4d100b671aaf4ce353a73 LICENSES/murmurhash2-public-domain.txt > > sha256 1a7adaa2c86cedfd6c7f5c0c7c72fd6d3e02cd0c9593f21fdb53c89bb2b130ec LICENSES/OFL-1.1.txt > > -sha256 11801e931f252252a16eac8299465510d0a82c36bfd9ac8aea9b202b76d2f82b LICENSES/README.md > > +sha256 177ff657698bc940f9888b7f69e5d734b0933c2859ea264fde8de1700cec40dd LICENSES/README.md > > What has changed in that license file? How does it impact (or not) the > licensing info we carry? > > We have hashes for license files, so that we know they licensing terms > change when a package is updated. Silently changing the hash kind of > defeats the purpose for those hashes... > > > diff --git a/package/systemd/systemd.mk b/package/systemd/systemd.mk > > index 46a4e8de2c..b6fffc553f 100644 > > --- a/package/systemd/systemd.mk > > +++ b/package/systemd/systemd.mk > [--SNIP--] > > @@ -472,6 +474,20 @@ else > > SYSTEMD_CONF_OPTS += -Dsysext=false > > endif > > > > +ifeq ($(BR2_PACKAGE_SYSTEMD_SYSUPDATE),y) > > +SYSTEMD_CONF_OPTS += -Dsysupdate=true > > +else > > +SYSTEMD_CONF_OPTS += -Dsysupdate=false > > +endif > > + > > +ifeq ($(BR2_PACKAGE_BASH),y) > > +SYSTEMD_CONF_OPTS += -Ddefault-user-shell=/bin/bash > > +else ifeq ($(BR2_PACKAGE_ZSH),y) > > +SYSTEMD_CONF_OPTS += -Ddefault-user-shell=/bin/zsh > > +else > > +SYSTEMD_CONF_OPTS += -Ddefault-user-shell=/bin/sh > > +endif > > We already have BR2_SYSTEM_BIN_SH which defines what /bin/sh is, and so > I think we should use exactly /bin/sh as the default shell: > > ifneq ($(BR2_SYSTEM_BIN_SH_NONE),y) > SYSTEMD_CONF_OPTS += -Ddefault-user-shell=/bin/sh > endif > > If there is not default shell for the system, then there is none for > users as well. Systemd-defined users that want a specific shell will > have to explicitly request it: if they have a requirement for a specific > shell, they can't rely on whatever the default is, so usng /bin/sh is > exactly what we want. > > If BR2_SYSTEM_BIN_SH is empty, it is perfectly possible that there is no > shell available at all (no standalone shell, no shell in busybox), so we > don't want to set any default either. > > Regards, > Yann E. MORIN. > > -- > .-----------------.--------------------.------------------.--------------------. > | Yann E. MORIN | Real-Time Embedded | /"\ ASCII RIBBON | Erics' conspiracy: | > | +33 662 376 056 | Software Designer | \ / CAMPAIGN | ___ | > | +33 561 099 427 `------------.-------: X AGAINST | \e/ There is no | > | http://ymorin.is-a-geek.org/ | _/*\_ | / \ HTML MAIL | v conspiracy. | > '------------------------------^-------^------------------^--------------------' > _______________________________________________ > buildroot mailing list > buildroot@buildroot.org > https://lists.buildroot.org/mailman/listinfo/buildroot _______________________________________________ buildroot mailing list buildroot@buildroot.org https://lists.buildroot.org/mailman/listinfo/buildroot ^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [Buildroot] [PATCH 1/1] package/systemd: bump to version 251.3 2022-07-19 21:56 ` James Hilliard @ 2022-07-20 14:51 ` Norbert Lange 2022-07-20 16:52 ` James Hilliard 0 siblings, 1 reply; 6+ messages in thread From: Norbert Lange @ 2022-07-20 14:51 UTC (permalink / raw) To: James Hilliard; +Cc: Yann E. MORIN, buildroot Am Di., 19. Juli 2022 um 23:56 Uhr schrieb James Hilliard <james.hilliard1@gmail.com>: > > FYI I had sent a similar bump earlier: > https://patchwork.ozlabs.org/project/buildroot/patch/20220522011729.117324-1-james.hilliard1@gmail.com/ > > On Tue, Jul 19, 2022 at 10:22 AM Yann E. MORIN <yann.morin.1998@free.fr> wrote: > > > > Norbert, All, > > > > On 2022-07-19 15:04 +0200, Norbert Lange spake thusly: > > > Add config option for the new sysupdate feature. > > > > Honestly, with a package so complex and critical like systemd, I would > > prefer that new features not be added during the bump, but as followup > > patches. The bump should add a force-disable of new features (when > > known), or just leave them as auto-detected or upstream-decided > > defaults. Honestly, I got scolded every time I just left it force-disabled ;) Someone even got to the great lengths of typing the steps into systemd.mk: # - If there are new services: # - create new options for them (if they really are optional); The features are disabled by default though, so it should not cause havoc. > > > > > Remove upstreamed patch. > > > > > > Update options, manually setdefault options, remove option > > > > Sorry, I don't immediately see what "manually setdefault" means... > > Is there a missing space, and this should be "manually set default" ? > > Or is "setdefault" and actual systemd concept? > > > > (genuinely asking) Missing space > > > > > not available anymore. > > > > > > Set default-user-shell based on wether a couple packages > > > are enabled. > > > > > > Set default-locale to C, this seems the only safe default, > > > > How does that play with BR2_ENABLE_LOCALE_WHITELIST? I think that we > > would want to use the first word of BR2_ENABLE_LOCALE_WHITELIST is it is > > set, falling back to just 'C' otherwise. > > Yeah, we should use 'C.UTF-8' since it's now supported by localedef, I had > sent a patch earlier changing the default but there were uclibc compatibility > issues, since systemd is incompatible with uclibc there's probably a better > approach: > https://patchwork.ozlabs.org/project/buildroot/patch/20220517192439.1034654-1-james.hilliard1@gmail.com/ > I on not really familiar how this should be detected, especially if there is an external toolchain used and it should definitely only be one of 'C' or 'C.UTF-8'. The /usr/lib/C.UTF-8 directory has around 2MB, and for my usecase I don't need or want it - unless you need to correctly display/sort strings, apps just copy a binary blob and don't need to know any details. if there is *any* doubt it should be C, and currently there is alot doubt. > > > > Furthermore, that is not a novelty with v251; v250 already had that > > option, so that should be a separate, preliminary patch. > > > > > users can override this in `/etc/locale.conf`. > > > > > Signed-off-by: Norbert Lange <nolange79@gmail.com> > > > --- > > [--SNIP--] > > > diff --git a/package/systemd/systemd.hash b/package/systemd/systemd.hash > > > index 3572b25965..838ed13fd5 100644 > > > --- a/package/systemd/systemd.hash > > > +++ b/package/systemd/systemd.hash > > > @@ -1,5 +1,5 @@ > > > # sha256 locally computed > > > -sha256 d2bda9d225da11dc9ff48b48e59fc36798d3e66902ed400a9f78fa370c596864 systemd-250.4.tar.gz > > > +sha256 11c9c3fcc13d689615dd2ea79161d1b992087cf1e2a23856d7d7ade0efbddc54 systemd-251.3.tar.gz > > > sha256 ab15fd526bd8dd18a9e77ebc139656bf4d33e97fc7238cd11bf60e2b9b8666c6 LICENSE.GPL2 > > > sha256 dc626520dcd53a22f727af3ee42c770e56c97a64fe3adb063799d8ab032fe551 LICENSE.LGPL2.1 > > > sha256 e5a8645ad94aab24e312dd0c6be2aa54236eb9374480b1b14ea5c61598874fd5 LICENSES/BSD-2-Clause.txt > > > @@ -11,4 +11,4 @@ sha256 790ac93fb2859097bdda4cf08b5a4feb5e479d0cb2c74f403248241bc3e7c216 LICENS > > > sha256 b85dcd3e453d05982552c52b5fc9e0bdd6d23c6f8e844b984a88af32570b0cc0 LICENSES/MIT.txt > > > sha256 2a98749f6bec00dfaed86fa9c1edea871aaae4f11ee4d100b671aaf4ce353a73 LICENSES/murmurhash2-public-domain.txt > > > sha256 1a7adaa2c86cedfd6c7f5c0c7c72fd6d3e02cd0c9593f21fdb53c89bb2b130ec LICENSES/OFL-1.1.txt > > > -sha256 11801e931f252252a16eac8299465510d0a82c36bfd9ac8aea9b202b76d2f82b LICENSES/README.md > > > +sha256 177ff657698bc940f9888b7f69e5d734b0933c2859ea264fde8de1700cec40dd LICENSES/README.md > > > > What has changed in that license file? How does it impact (or not) the > > licensing info we carry? > > > > We have hashes for license files, so that we know they licensing terms > > change when a package is updated. Silently changing the hash kind of > > defeats the purpose for those hashes... One line got added, no changes in the license text itself : * the following sources are licensed under the **CC0-1.0** license: - src/systemctl/systemd-sysv-install.SKELETON - tools/check-includes.pl - all examples under man/ + - config files and examples under /network The line is still valid IMHO: "CC0-1.0 (few source files, see LICENSES/README.md)" > > > > > diff --git a/package/systemd/systemd.mk b/package/systemd/systemd.mk > > > index 46a4e8de2c..b6fffc553f 100644 > > > --- a/package/systemd/systemd.mk > > > +++ b/package/systemd/systemd.mk > > [--SNIP--] > > > @@ -472,6 +474,20 @@ else > > > SYSTEMD_CONF_OPTS += -Dsysext=false > > > endif > > > > > > +ifeq ($(BR2_PACKAGE_SYSTEMD_SYSUPDATE),y) > > > +SYSTEMD_CONF_OPTS += -Dsysupdate=true > > > +else > > > +SYSTEMD_CONF_OPTS += -Dsysupdate=false > > > +endif > > > + > > > +ifeq ($(BR2_PACKAGE_BASH),y) > > > +SYSTEMD_CONF_OPTS += -Ddefault-user-shell=/bin/bash > > > +else ifeq ($(BR2_PACKAGE_ZSH),y) > > > +SYSTEMD_CONF_OPTS += -Ddefault-user-shell=/bin/zsh > > > +else > > > +SYSTEMD_CONF_OPTS += -Ddefault-user-shell=/bin/sh > > > +endif > > > > We already have BR2_SYSTEM_BIN_SH which defines what /bin/sh is, and so > > I think we should use exactly /bin/sh as the default shell: > > > > ifneq ($(BR2_SYSTEM_BIN_SH_NONE),y) > > SYSTEMD_CONF_OPTS += -Ddefault-user-shell=/bin/sh > > endif > > > > If there is not default shell for the system, then there is none for > > users as well. Systemd-defined users that want a specific shell will > > have to explicitly request it: if they have a requirement for a specific > > shell, they can't rely on whatever the default is, so usng /bin/sh is > > exactly what we want. > > > > If BR2_SYSTEM_BIN_SH is empty, it is perfectly possible that there is no > > shell available at all (no standalone shell, no shell in busybox), so we > > don't want to set any default either. If you don't set anything, then '/bin/sh' will be used, so I guess ill just set that unconditionally. > > > > Regards, > > Yann E. MORIN. > > > > -- > > .-----------------.--------------------.------------------.--------------------. > > | Yann E. MORIN | Real-Time Embedded | /"\ ASCII RIBBON | Erics' conspiracy: | > > | +33 662 376 056 | Software Designer | \ / CAMPAIGN | ___ | > > | +33 561 099 427 `------------.-------: X AGAINST | \e/ There is no | > > | http://ymorin.is-a-geek.org/ | _/*\_ | / \ HTML MAIL | v conspiracy. | > > '------------------------------^-------^------------------^--------------------' > > _______________________________________________ > > buildroot mailing list > > buildroot@buildroot.org > > https://lists.buildroot.org/mailman/listinfo/buildroot Regards, Norbert _______________________________________________ buildroot mailing list buildroot@buildroot.org https://lists.buildroot.org/mailman/listinfo/buildroot ^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [Buildroot] [PATCH 1/1] package/systemd: bump to version 251.3 2022-07-20 14:51 ` Norbert Lange @ 2022-07-20 16:52 ` James Hilliard 2022-07-21 14:40 ` Norbert Lange 0 siblings, 1 reply; 6+ messages in thread From: James Hilliard @ 2022-07-20 16:52 UTC (permalink / raw) To: Norbert Lange; +Cc: Yann E. MORIN, buildroot On Wed, Jul 20, 2022 at 8:51 AM Norbert Lange <nolange79@gmail.com> wrote: > > Am Di., 19. Juli 2022 um 23:56 Uhr schrieb James Hilliard > <james.hilliard1@gmail.com>: > > > > FYI I had sent a similar bump earlier: > > https://patchwork.ozlabs.org/project/buildroot/patch/20220522011729.117324-1-james.hilliard1@gmail.com/ > > > > On Tue, Jul 19, 2022 at 10:22 AM Yann E. MORIN <yann.morin.1998@free.fr> wrote: > > > > > > Norbert, All, > > > > > > On 2022-07-19 15:04 +0200, Norbert Lange spake thusly: > > > > Add config option for the new sysupdate feature. > > > > > > Honestly, with a package so complex and critical like systemd, I would > > > prefer that new features not be added during the bump, but as followup > > > patches. The bump should add a force-disable of new features (when > > > known), or just leave them as auto-detected or upstream-decided > > > defaults. > > Honestly, I got scolded every time I just left it force-disabled ;) > Someone even got to the great lengths of typing the steps into systemd.mk: > > # - If there are new services: > # - create new options for them (if they really are optional); > > The features are disabled by default though, so it should not > cause havoc. > > > > > > > > Remove upstreamed patch. > > > > > > > > Update options, manually setdefault options, remove option > > > > > > Sorry, I don't immediately see what "manually setdefault" means... > > > Is there a missing space, and this should be "manually set default" ? > > > Or is "setdefault" and actual systemd concept? > > > > > > (genuinely asking) > > Missing space > > > > > > > > not available anymore. > > > > > > > > Set default-user-shell based on wether a couple packages > > > > are enabled. > > > > > > > > Set default-locale to C, this seems the only safe default, > > > > > > How does that play with BR2_ENABLE_LOCALE_WHITELIST? I think that we > > > would want to use the first word of BR2_ENABLE_LOCALE_WHITELIST is it is > > > set, falling back to just 'C' otherwise. > > > > Yeah, we should use 'C.UTF-8' since it's now supported by localedef, I had > > sent a patch earlier changing the default but there were uclibc compatibility > > issues, since systemd is incompatible with uclibc there's probably a better > > approach: > > https://patchwork.ozlabs.org/project/buildroot/patch/20220517192439.1034654-1-james.hilliard1@gmail.com/ > > > > I on not really familiar how this should be detected, especially if > there is an external > toolchain used and it should definitely only be one of 'C' or 'C.UTF-8'. > > The /usr/lib/C.UTF-8 directory has around 2MB, and for my usecase > I don't need or want it - unless you need to correctly display/sort strings, > apps just copy a binary blob and don't need to know any details. Main issue is that the regular 'C' locale tends to cause lots of issues for anything that needs UTF-8 support, which can include a lot of rather unexpected things such as simply extracting tar archives with libarchive. I think python may also not work correctly without a UTF-8 locale: https://peps.python.org/pep-0538/ I think the only reliably sane locale is C.UTF-8 which is what most distros have been moving towards. However there may be cases where just a C locale is desired so we may want to have an option for that still, it should not however be the default IMO. > > if there is *any* doubt it should be C, and currently there is alot doubt. Upstream considers C.UTF-8 the preferred default locale: https://github.com/systemd/systemd/commit/c9bd01ae902d66acd8647a1cc0167bc0d246cbc8 > > > > > > > Furthermore, that is not a novelty with v251; v250 already had that > > > option, so that should be a separate, preliminary patch. > > > > > > > users can override this in `/etc/locale.conf`. > > > > > > > Signed-off-by: Norbert Lange <nolange79@gmail.com> > > > > --- > > > [--SNIP--] > > > > diff --git a/package/systemd/systemd.hash b/package/systemd/systemd.hash > > > > index 3572b25965..838ed13fd5 100644 > > > > --- a/package/systemd/systemd.hash > > > > +++ b/package/systemd/systemd.hash > > > > @@ -1,5 +1,5 @@ > > > > # sha256 locally computed > > > > -sha256 d2bda9d225da11dc9ff48b48e59fc36798d3e66902ed400a9f78fa370c596864 systemd-250.4.tar.gz > > > > +sha256 11c9c3fcc13d689615dd2ea79161d1b992087cf1e2a23856d7d7ade0efbddc54 systemd-251.3.tar.gz > > > > sha256 ab15fd526bd8dd18a9e77ebc139656bf4d33e97fc7238cd11bf60e2b9b8666c6 LICENSE.GPL2 > > > > sha256 dc626520dcd53a22f727af3ee42c770e56c97a64fe3adb063799d8ab032fe551 LICENSE.LGPL2.1 > > > > sha256 e5a8645ad94aab24e312dd0c6be2aa54236eb9374480b1b14ea5c61598874fd5 LICENSES/BSD-2-Clause.txt > > > > @@ -11,4 +11,4 @@ sha256 790ac93fb2859097bdda4cf08b5a4feb5e479d0cb2c74f403248241bc3e7c216 LICENS > > > > sha256 b85dcd3e453d05982552c52b5fc9e0bdd6d23c6f8e844b984a88af32570b0cc0 LICENSES/MIT.txt > > > > sha256 2a98749f6bec00dfaed86fa9c1edea871aaae4f11ee4d100b671aaf4ce353a73 LICENSES/murmurhash2-public-domain.txt > > > > sha256 1a7adaa2c86cedfd6c7f5c0c7c72fd6d3e02cd0c9593f21fdb53c89bb2b130ec LICENSES/OFL-1.1.txt > > > > -sha256 11801e931f252252a16eac8299465510d0a82c36bfd9ac8aea9b202b76d2f82b LICENSES/README.md > > > > +sha256 177ff657698bc940f9888b7f69e5d734b0933c2859ea264fde8de1700cec40dd LICENSES/README.md > > > > > > What has changed in that license file? How does it impact (or not) the > > > licensing info we carry? > > > > > > We have hashes for license files, so that we know they licensing terms > > > change when a package is updated. Silently changing the hash kind of > > > defeats the purpose for those hashes... > > One line got added, no changes in the license text itself : > > * the following sources are licensed under the **CC0-1.0** license: > - src/systemctl/systemd-sysv-install.SKELETON > - tools/check-includes.pl > - all examples under man/ > + - config files and examples under /network > > The line is still valid IMHO: > "CC0-1.0 (few source files, see LICENSES/README.md)" > > > > > > > > diff --git a/package/systemd/systemd.mk b/package/systemd/systemd.mk > > > > index 46a4e8de2c..b6fffc553f 100644 > > > > --- a/package/systemd/systemd.mk > > > > +++ b/package/systemd/systemd.mk > > > [--SNIP--] > > > > @@ -472,6 +474,20 @@ else > > > > SYSTEMD_CONF_OPTS += -Dsysext=false > > > > endif > > > > > > > > +ifeq ($(BR2_PACKAGE_SYSTEMD_SYSUPDATE),y) > > > > +SYSTEMD_CONF_OPTS += -Dsysupdate=true > > > > +else > > > > +SYSTEMD_CONF_OPTS += -Dsysupdate=false > > > > +endif > > > > + > > > > +ifeq ($(BR2_PACKAGE_BASH),y) > > > > +SYSTEMD_CONF_OPTS += -Ddefault-user-shell=/bin/bash > > > > +else ifeq ($(BR2_PACKAGE_ZSH),y) > > > > +SYSTEMD_CONF_OPTS += -Ddefault-user-shell=/bin/zsh > > > > +else > > > > +SYSTEMD_CONF_OPTS += -Ddefault-user-shell=/bin/sh > > > > +endif > > > > > > We already have BR2_SYSTEM_BIN_SH which defines what /bin/sh is, and so > > > I think we should use exactly /bin/sh as the default shell: > > > > > > ifneq ($(BR2_SYSTEM_BIN_SH_NONE),y) > > > SYSTEMD_CONF_OPTS += -Ddefault-user-shell=/bin/sh > > > endif > > > > > > If there is not default shell for the system, then there is none for > > > users as well. Systemd-defined users that want a specific shell will > > > have to explicitly request it: if they have a requirement for a specific > > > shell, they can't rely on whatever the default is, so usng /bin/sh is > > > exactly what we want. > > > > > > If BR2_SYSTEM_BIN_SH is empty, it is perfectly possible that there is no > > > shell available at all (no standalone shell, no shell in busybox), so we > > > don't want to set any default either. > > If you don't set anything, then '/bin/sh' will be used, so I guess ill > just set that > unconditionally. Well /bin/bash is the default if nothing is specified so we should override that: https://github.com/systemd/systemd-stable/blob/v251.3/meson_options.txt#L223-L224 > > > > > > > Regards, > > > Yann E. MORIN. > > > > > > -- > > > .-----------------.--------------------.------------------.--------------------. > > > | Yann E. MORIN | Real-Time Embedded | /"\ ASCII RIBBON | Erics' conspiracy: | > > > | +33 662 376 056 | Software Designer | \ / CAMPAIGN | ___ | > > > | +33 561 099 427 `------------.-------: X AGAINST | \e/ There is no | > > > | http://ymorin.is-a-geek.org/ | _/*\_ | / \ HTML MAIL | v conspiracy. | > > > '------------------------------^-------^------------------^--------------------' > > > _______________________________________________ > > > buildroot mailing list > > > buildroot@buildroot.org > > > https://lists.buildroot.org/mailman/listinfo/buildroot > > Regards, Norbert _______________________________________________ buildroot mailing list buildroot@buildroot.org https://lists.buildroot.org/mailman/listinfo/buildroot ^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [Buildroot] [PATCH 1/1] package/systemd: bump to version 251.3 2022-07-20 16:52 ` James Hilliard @ 2022-07-21 14:40 ` Norbert Lange 0 siblings, 0 replies; 6+ messages in thread From: Norbert Lange @ 2022-07-21 14:40 UTC (permalink / raw) To: James Hilliard; +Cc: Yann E. MORIN, buildroot Am Mi., 20. Juli 2022 um 18:52 Uhr schrieb James Hilliard <james.hilliard1@gmail.com>: > > On Wed, Jul 20, 2022 at 8:51 AM Norbert Lange <nolange79@gmail.com> wrote: > > > > Am Di., 19. Juli 2022 um 23:56 Uhr schrieb James Hilliard > > <james.hilliard1@gmail.com>: > > > > > > FYI I had sent a similar bump earlier: > > > https://patchwork.ozlabs.org/project/buildroot/patch/20220522011729.117324-1-james.hilliard1@gmail.com/ > > > > > > On Tue, Jul 19, 2022 at 10:22 AM Yann E. MORIN <yann.morin.1998@free.fr> wrote: > > > > > > > > Norbert, All, > > > > > > > > On 2022-07-19 15:04 +0200, Norbert Lange spake thusly: > > > > > Add config option for the new sysupdate feature. > > > > > > > > Honestly, with a package so complex and critical like systemd, I would > > > > prefer that new features not be added during the bump, but as followup > > > > patches. The bump should add a force-disable of new features (when > > > > known), or just leave them as auto-detected or upstream-decided > > > > defaults. > > > > Honestly, I got scolded every time I just left it force-disabled ;) > > Someone even got to the great lengths of typing the steps into systemd.mk: > > > > # - If there are new services: > > # - create new options for them (if they really are optional); > > > > The features are disabled by default though, so it should not > > cause havoc. > > > > > > > > > > > Remove upstreamed patch. > > > > > > > > > > Update options, manually setdefault options, remove option > > > > > > > > Sorry, I don't immediately see what "manually setdefault" means... > > > > Is there a missing space, and this should be "manually set default" ? > > > > Or is "setdefault" and actual systemd concept? > > > > > > > > (genuinely asking) > > > > Missing space > > > > > > > > > > > not available anymore. > > > > > > > > > > Set default-user-shell based on wether a couple packages > > > > > are enabled. > > > > > > > > > > Set default-locale to C, this seems the only safe default, > > > > > > > > How does that play with BR2_ENABLE_LOCALE_WHITELIST? I think that we > > > > would want to use the first word of BR2_ENABLE_LOCALE_WHITELIST is it is > > > > set, falling back to just 'C' otherwise. > > > > > > Yeah, we should use 'C.UTF-8' since it's now supported by localedef, I had > > > sent a patch earlier changing the default but there were uclibc compatibility > > > issues, since systemd is incompatible with uclibc there's probably a better > > > approach: > > > https://patchwork.ozlabs.org/project/buildroot/patch/20220517192439.1034654-1-james.hilliard1@gmail.com/ > > > > > > > I on not really familiar how this should be detected, especially if > > there is an external > > toolchain used and it should definitely only be one of 'C' or 'C.UTF-8'. > > > > The /usr/lib/C.UTF-8 directory has around 2MB, and for my usecase > > I don't need or want it - unless you need to correctly display/sort strings, > > apps just copy a binary blob and don't need to know any details. > > Main issue is that the regular 'C' locale tends to cause lots of > issues for anything > that needs UTF-8 support, which can include a lot of rather unexpected things > such as simply extracting tar archives with libarchive. > > I think python may also not work correctly without a UTF-8 locale: > https://peps.python.org/pep-0538/ Yeah, that's a rather "fat" installation tho > I think the only reliably sane locale is C.UTF-8 which is what most distros > have been moving towards. However there may be cases where just a C > locale is desired so we may want to have an option for that still, it should not > however be the default IMO. As said, that's 2MB for stuff you probably don't use in alot applications where you would use buildroot instead of a full-blown distro. > > > > > if there is *any* doubt it should be C, and currently there is alot doubt. > > Upstream considers C.UTF-8 the preferred default locale: > https://github.com/systemd/systemd/commit/c9bd01ae902d66acd8647a1cc0167bc0d246cbc8 Thats ok for their primary target (redhat and other full distros), they also assume bash to be a reasonable default. I would not know how to message to buuldroot wether my external toolchain with an older glibc is "fit" for C.UTF-8. I will leave it at that for now, just because it doesnt make matters worse and I dont want to block the version bump. > > > > > > > > > > > Furthermore, that is not a novelty with v251; v250 already had that > > > > option, so that should be a separate, preliminary patch. > > > > > > > > > users can override this in `/etc/locale.conf`. > > > > > > > > > Signed-off-by: Norbert Lange <nolange79@gmail.com> > > > > > --- > > > > [--SNIP--] > > > > > diff --git a/package/systemd/systemd.hash b/package/systemd/systemd.hash > > > > > index 3572b25965..838ed13fd5 100644 > > > > > --- a/package/systemd/systemd.hash > > > > > +++ b/package/systemd/systemd.hash > > > > > @@ -1,5 +1,5 @@ > > > > > # sha256 locally computed > > > > > -sha256 d2bda9d225da11dc9ff48b48e59fc36798d3e66902ed400a9f78fa370c596864 systemd-250.4.tar.gz > > > > > +sha256 11c9c3fcc13d689615dd2ea79161d1b992087cf1e2a23856d7d7ade0efbddc54 systemd-251.3.tar.gz > > > > > sha256 ab15fd526bd8dd18a9e77ebc139656bf4d33e97fc7238cd11bf60e2b9b8666c6 LICENSE.GPL2 > > > > > sha256 dc626520dcd53a22f727af3ee42c770e56c97a64fe3adb063799d8ab032fe551 LICENSE.LGPL2.1 > > > > > sha256 e5a8645ad94aab24e312dd0c6be2aa54236eb9374480b1b14ea5c61598874fd5 LICENSES/BSD-2-Clause.txt > > > > > @@ -11,4 +11,4 @@ sha256 790ac93fb2859097bdda4cf08b5a4feb5e479d0cb2c74f403248241bc3e7c216 LICENS > > > > > sha256 b85dcd3e453d05982552c52b5fc9e0bdd6d23c6f8e844b984a88af32570b0cc0 LICENSES/MIT.txt > > > > > sha256 2a98749f6bec00dfaed86fa9c1edea871aaae4f11ee4d100b671aaf4ce353a73 LICENSES/murmurhash2-public-domain.txt > > > > > sha256 1a7adaa2c86cedfd6c7f5c0c7c72fd6d3e02cd0c9593f21fdb53c89bb2b130ec LICENSES/OFL-1.1.txt > > > > > -sha256 11801e931f252252a16eac8299465510d0a82c36bfd9ac8aea9b202b76d2f82b LICENSES/README.md > > > > > +sha256 177ff657698bc940f9888b7f69e5d734b0933c2859ea264fde8de1700cec40dd LICENSES/README.md > > > > > > > > What has changed in that license file? How does it impact (or not) the > > > > licensing info we carry? > > > > > > > > We have hashes for license files, so that we know they licensing terms > > > > change when a package is updated. Silently changing the hash kind of > > > > defeats the purpose for those hashes... > > > > One line got added, no changes in the license text itself : > > > > * the following sources are licensed under the **CC0-1.0** license: > > - src/systemctl/systemd-sysv-install.SKELETON > > - tools/check-includes.pl > > - all examples under man/ > > + - config files and examples under /network > > > > The line is still valid IMHO: > > "CC0-1.0 (few source files, see LICENSES/README.md)" > > > > > > > > > > > diff --git a/package/systemd/systemd.mk b/package/systemd/systemd.mk > > > > > index 46a4e8de2c..b6fffc553f 100644 > > > > > --- a/package/systemd/systemd.mk > > > > > +++ b/package/systemd/systemd.mk > > > > [--SNIP--] > > > > > @@ -472,6 +474,20 @@ else > > > > > SYSTEMD_CONF_OPTS += -Dsysext=false > > > > > endif > > > > > > > > > > +ifeq ($(BR2_PACKAGE_SYSTEMD_SYSUPDATE),y) > > > > > +SYSTEMD_CONF_OPTS += -Dsysupdate=true > > > > > +else > > > > > +SYSTEMD_CONF_OPTS += -Dsysupdate=false > > > > > +endif > > > > > + > > > > > +ifeq ($(BR2_PACKAGE_BASH),y) > > > > > +SYSTEMD_CONF_OPTS += -Ddefault-user-shell=/bin/bash > > > > > +else ifeq ($(BR2_PACKAGE_ZSH),y) > > > > > +SYSTEMD_CONF_OPTS += -Ddefault-user-shell=/bin/zsh > > > > > +else > > > > > +SYSTEMD_CONF_OPTS += -Ddefault-user-shell=/bin/sh > > > > > +endif > > > > > > > > We already have BR2_SYSTEM_BIN_SH which defines what /bin/sh is, and so > > > > I think we should use exactly /bin/sh as the default shell: > > > > > > > > ifneq ($(BR2_SYSTEM_BIN_SH_NONE),y) > > > > SYSTEMD_CONF_OPTS += -Ddefault-user-shell=/bin/sh > > > > endif > > > > > > > > If there is not default shell for the system, then there is none for > > > > users as well. Systemd-defined users that want a specific shell will > > > > have to explicitly request it: if they have a requirement for a specific > > > > shell, they can't rely on whatever the default is, so usng /bin/sh is > > > > exactly what we want. > > > > > > > > If BR2_SYSTEM_BIN_SH is empty, it is perfectly possible that there is no > > > > shell available at all (no standalone shell, no shell in busybox), so we > > > > don't want to set any default either. > > > > If you don't set anything, then '/bin/sh' will be used, so I guess ill > > just set that > > unconditionally. > > Well /bin/bash is the default if nothing is specified so we should > override that: > https://github.com/systemd/systemd-stable/blob/v251.3/meson_options.txt#L223-L224 OK Regards, Norbert _______________________________________________ buildroot mailing list buildroot@buildroot.org https://lists.buildroot.org/mailman/listinfo/buildroot ^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2022-07-21 14:41 UTC | newest] Thread overview: 6+ messages (download: mbox.gz follow: Atom feed -- links below jump to the message on this page -- 2022-07-19 13:04 [Buildroot] [PATCH 1/1] package/systemd: bump to version 251.3 Norbert Lange 2022-07-19 16:21 ` Yann E. MORIN 2022-07-19 21:56 ` James Hilliard 2022-07-20 14:51 ` Norbert Lange 2022-07-20 16:52 ` James Hilliard 2022-07-21 14:40 ` Norbert Lange
This is an external index of several public inboxes, see mirroring instructions on how to clone and mirror all data and code used by this external index.