Extent options for read-only /var handling
cover Extent options for read-only /var handling
This series is based on the work of Yann to allow variants for handling /var in respect to a read-only rootfs.
It contains a new version to use an overlayfs,
a version using a systemd-generator,
and a few more patches to improve the situation under a read-only root.
Further it tries to establish some buildroot runtime directories, namely /run/.br and /tmp/.br.
The /run/.br should used for any runtime directories/files that are specific tu buildroot. Since /run is expected to only contain a modest amount of content, /tmp/.br is also available for "fatter" stuff.
Those patches are tested to solve the /var issue while behaving nicely within systemd and not requiring special processing or modification of the rootfs.
The patches further allow systemd and its services to boot without errors, even if /var is plainly read-only.
Norbert Lange (6):
system: add overlayfs option for ro /var package/skeleton-init-systemd: force permissions for buildroot runtime
dirs
system: add systemd generator for /var package/skeleton-init-systemd: fix var.mount options package/systemd: handle creation of /var/log/journal package/systemd: create the /boot directory
Yann E. MORIN (1):
system: introduce a choice for /var management
.../skeleton-init-systemd/00-buildroot.conf | 7 +++ .../generator/br-mount-generator | 19 ++++++++ .../overlayfs/[email protected] | 26 ++++++++++ .../overlayfs/[email protected] | 26 ++++++++++ .../overlayfs/overlay.mount.in | 16 +++++++ .../skeleton-init-systemd.mk | 26 ++++++++++ package/skeleton-init-systemd/var.mount | 2 +- package/systemd/systemd.mk | 8 ++++ system/Config.in | 48 +++++++++++++++++-- 9 files changed, 174 insertions(+), 4 deletions(-) create mode 100644 package/skeleton-init-systemd/00-buildroot.conf create mode 100755 package/skeleton-init-systemd/generator/br-mount-generator create mode 100644 package/skeleton-init-systemd/overlayfs/[email protected] create mode 100644 package/skeleton-init-systemd/overlayfs/[email protected] create mode 100644 package/skeleton-init-systemd/overlayfs/overlay.mount.in
Signature · 6 lines
2.39.0 _______________________________________________ buildroot mailing list [email protected] https://lists.buildroot.org/mailman/listinfo/buildroot
PATCH 1/7 system: introduce a choice for /var management
From: "Yann E. MORIN" <[email protected]>
Currently, we have a single solution to handle the /var content on a read-only root filesystem, and users can opt out of using it.
We're going to introduce another solution (based on an overlayfs), which is incompatible with using the factory, so we'll have a choice of three mutually exclusive options.
Introduce that choice now with just the existing /var factory, and a new option to opt out of it.
--- system/Config.in | 22 +++++++++++++++++++--- 1 file changed, 19 insertions(+), 3 deletions(-)
system/Config.in +19 -3
@@ -156,10 +156,16 @@ endchoice if BR2_INIT_SYSTEMD -config BR2_INIT_SYSTEMD_VAR_FACTORY - bool "build a factory to populate a tmpfs on /var" - default y # legacy +choice + bool "/var management" + default BR2_INIT_SYSTEMD_VAR_FACTORY # legacy depends on !BR2_TARGET_GENERIC_REMOUNT_ROOTFS_RW + help + Select how Buildroot provides a read-write /var when the + rootfs is not remounted read-write. + +config BR2_INIT_SYSTEMD_VAR_FACTORY + bool "build a factory to populate a tmpfs" help Build a factory of the content of /var as installed by packages, mount a tmpfs on /var at runtime, so that
@@ -181,6 +187,16 @@ config BR2_INIT_SYSTEMD_VAR_FACTORY var.mount unit, that overrides the What and Type, and possibly the Options and After, fields. +config BR2_INIT_SYSTEMD_VAR_NONE + bool "do nothing" + help + Choose this if you have custom dispositions (like one or more + of a post-build script, a fakeroot script, systemd units, an + initramfs, or something else) that prepare /var to be writable + on a read-only rootfs. + +endchoice + config BR2_INIT_SYSTEMD_POPULATE_TMPFILES bool "trigger systemd-tmpfiles during build" default y # legacy
Signature · 6 lines
2.39.0 _______________________________________________ buildroot mailing list [email protected] https://lists.buildroot.org/mailman/listinfo/buildroot
PATCH 2/7 system: add overlayfs option for ro /var
In reply to Norbert LangeThis commit adds an alternative that has the following characteristics:
- Dont depend on anything being available, except the
API File Systems [1].
As /var is meant to be available before normal and even some early services are running.
- Be a clean drop-in, that can be trivially added / removed.
- Depend on overlayfs being available in the kernel.
- Units are supposed to be reusable for custom solutions.
Further, a few subdirectories are introduced, which hopefully can serve as template for similar uses.
Buildroot owns /run/.br, bind mounts end up in /run/.br/bnd, overlay directories in /run/.br/ovl.
The sequence is rather important, and the systemd mount units serve as reliable "sequence point" (cant be replicated otherwise).
What we need to do is mount /var in its usable state, and block units depending on this before we are finished.
This includes:
- Other mounts targeting a subdirectory in /var.
- Units using 'RequiresMountsFor' targeting /var or subdirectories.
Most notable basic.target which most units depend upon by default.
- Units that are (transitively) odered after one of the above.
As comparison, the /var factory introduces issues here, as systemd-tmpfiles-setup.service will require all mounts to be finished before, as well als journald running.
(That means /var will already be used while and before initialization).
The necessary steps are:
1. Bindmount /var to /run/.br/bnd/var
This is done by using the template [email protected].
2. Mount an tmpfs and prepare the overlay at /run/.br/ovl/var
This is done by using the template [email protected], the mounting command is separated so that users can further customize the behaviour.
The choice of using a service seems more flexible, as the mount units lack several options of the mount command. As a gut feeling the fileystems mounted under /run should not be accessible or propagate their mounts, hence --make-private.
(Or you might not use a mount at all).
3. Finally mount the overlay with a normal mount unit
Dependencies on the template-instances of 1. and 2. will ensure correct odering before.
By nature off being a systemd mount unit the remaining units will be automatically order after this unit if necessary.
As example, consider using /dev/sdc1 as upper layer for var, this can be archived by swapping out just the line executing the mount itself:
[Service]
ExecStart=
ExecStart=/bin/mount --make-private -n /dev/sdc1 ${OVERLAY_DIR}
[1] - https://www.freedesktop.org/wiki/Software/systemd/APIFileSystems/
--- .../overlayfs/[email protected] | 26 +++++++++++++++++++ .../overlayfs/[email protected] | 26 +++++++++++++++++++ .../overlayfs/overlay.mount.in | 16 ++++++++++++ .../skeleton-init-systemd.mk | 17 ++++++++++++ system/Config.in | 13 ++++++++++ 5 files changed, 98 insertions(+) create mode 100644 package/skeleton-init-systemd/overlayfs/[email protected] create mode 100644 package/skeleton-init-systemd/overlayfs/[email protected] create mode 100644 package/skeleton-init-systemd/overlayfs/overlay.mount.in
package/skeleton-init-systemd/overlayfs/[email protected] new file +26 -0
@@ -0,0 +1,26 @@ +[Unit] +Description=Bind-mount rootfs directory (/%I) to /run +Documentation=man:file-hierarchy(7) +ConditionPathIsSymbolicLink=!/%I +DefaultDependencies=no +Conflicts=umount.target +Before=umount.target + +# Needs to run after rootfs is properly mounted +# and before regular mounts might interfere. +After=systemd-remount-fs.service +Before=local-fs-pre.target + +[Service] +Type=oneshot +RemainAfterExit=yes +# dont fail if common dirs already exist +ExecStartPre=/bin/mkdir -m755 -p /run/.br +ExecStartPre=/bin/mkdir -m700 -p /run/.br/bnd + +ExecStartPre=/bin/mkdir -m700 -p /run/.br/bnd/%I +ExecStart=/bin/mount --make-private -n --bind -o ro /%I /run/.br/bnd/%I + +# lazy unmount, dont block shutdown under any circumstances +ExecStop=/bin/umount -n -l /run/.br/bnd/%I +ExecStopPost=-/bin/rmdir /run/.br/bnd/%I
package/skeleton-init-systemd/overlayfs/[email protected] new file +26 -0
@@ -0,0 +1,26 @@ +[Unit] +Description=Mount and prepare tmpfs for overlay (/%I) +Documentation=man:file-hierarchy(7) +ConditionPathIsSymbolicLink=!/%I +DefaultDependencies=no +Conflicts=umount.target +Before=local-fs.target umount.target +# prepare for systemd mount units aswell +RequiresMountsFor=/run/.br/ovl/%I + +[Service] +Environment="OVERLAY_DIR=/run/.br/ovl/%I" +Type=oneshot +RemainAfterExit=yes +# dont fail if common dirs already exist +ExecStartPre=/bin/mkdir -m755 -p /run/.br +ExecStartPre=/bin/mkdir -m700 -p /run/.br/ovl + +ExecStartPre=/bin/mkdir -m700 -p ${OVERLAY_DIR} +# Create an override and edit this line for customization +ExecStart=/bin/mount --make-private -n -t tmpfs tmpfs_br_ovl ${OVERLAY_DIR} +ExecStartPost=/bin/mkdir -p ${OVERLAY_DIR}/up ${OVERLAY_DIR}/wd + +# lazy unmount, dont block shutdown under any circumstances +ExecStop=/bin/umount -n -l ${OVERLAY_DIR} +ExecStopPost=/bin/rmdir ${OVERLAY_DIR}
package/skeleton-init-systemd/overlayfs/overlay.mount.in new file +16 -0
@@ -0,0 +1,16 @@ +[Unit] +Description=Variable storage (/@PATH@) +Documentation=man:file-hierarchy(7) +ConditionPathIsSymbolicLink=!/@PATH@ + +After=br-bindmount-run@@[email protected] +Requires=br-bindmount-run@@[email protected] + +After=br-overlay-prepare@@[email protected] +BindsTo=br-overlay-prepare@@[email protected] + +[Mount] +Type=overlay +What=br_ovl_@PATH@ +Where=/@PATH@ +Options=redirect_dir=on,index=on,xino=on,lowerdir=/run/.br/bnd/@PATH@,upperdir=/run/.br/ovl/@PATH@/up,workdir=/run/.br/ovl/@PATH@/wd
package/skeleton-init-systemd/skeleton-init-systemd.mk +17 -0
@@ -57,6 +57,23 @@ define SKELETON_INIT_SYSTEMD_PRE_ROOTFS_VAR endef SKELETON_INIT_SYSTEMD_ROOTFS_PRE_CMD_HOOKS += SKELETON_INIT_SYSTEMD_PRE_ROOTFS_VAR endif # BR2_INIT_SYSTEMD_VAR_FACTORY + +ifeq ($(BR2_INIT_SYSTEMD_VAR_OVERLAYFS),y) +define SKELETON_INIT_SYSTEMD_LINUX_CONFIG_FIXUPS + $(call KCONFIG_ENABLE_OPT,CONFIG_OVERLAY_FS) +endef +define SKELETON_INIT_SYSTEMD_POST_INSTALL_VAR_OVERLAYFS + sed 's,@PATH@,var,g' $(SKELETON_INIT_SYSTEMD_PKGDIR)/overlayfs/overlay.mount.in >$(@D)/var.mount + $(INSTALL) -D -m 0644 $(@D)/var.mount $(TARGET_DIR)/usr/lib/systemd/system/var.mount + $(INSTALL) -D -m 0644 $(SKELETON_INIT_SYSTEMD_PKGDIR)/overlayfs/[email protected] \ + $(TARGET_DIR)/usr/lib/systemd/system/[email protected] + $(INSTALL) -D -m 0644 $(SKELETON_INIT_SYSTEMD_PKGDIR)/overlayfs/[email protected] \ + $(TARGET_DIR)/usr/lib/systemd/system/[email protected] + # /var mount gets pulled in automatically by basic.target +endef +SKELETON_INIT_SYSTEMD_POST_INSTALL_TARGET_HOOKS += SKELETON_INIT_SYSTEMD_POST_INSTALL_VAR_OVERLAYFS +endif # BR2_INIT_SYSTEMD_VAR_OVERLAYFS + endif # BR2_TARGET_GENERIC_REMOUNT_ROOTFS_RW ifeq ($(BR2_INIT_SYSTEMD_POPULATE_TMPFILES),y)
system/Config.in +13 -0
@@ -187,6 +187,19 @@ config BR2_INIT_SYSTEMD_VAR_FACTORY var.mount unit, that overrides the What and Type, and possibly the Options and After, fields. +config BR2_INIT_SYSTEMD_VAR_OVERLAYFS + bool "mount an overlayfs backed by a tmpfs" + help + Mount an overlayfs on /var, with the upper as a tmpfs. + + You can customize the backing storage with an override, for + ecample create this file and add the content following: + /lib/systemd/system/[email protected]/mount.conf + + [Service] + ExecStart= + ExecStart=/bin/mount --make-private -n /dev/sdc1 ${OVERLAY_DIR} + config BR2_INIT_SYSTEMD_VAR_NONE bool "do nothing" help
Signature · 6 lines
2.39.0 _______________________________________________ buildroot mailing list [email protected] https://lists.buildroot.org/mailman/listinfo/buildroot
Re: [Buildroot] [PATCH 2/7] system: add overlayfs option for ro /var
Hi Norbert,
Quoted 13 lines · Norbert Lange
This commit adds an alternative that has the following characteristics:
(quoted diff) +0
-4- Dont depend on anything being available, except the API File Systems [1]. As /var is meant to be available before normal and even some early services are running. - Be a clean drop-in, that can be trivially added / removed. - Depend on overlayfs being available in the kernel. - Units are supposed to be reusable for custom solutions.
[snip]
Quoted 5 lines
Cc Yann E. MORIN <[email protected]>Cc Romain Naour <[email protected]>Cc Jérémy Rosen <[email protected]>Signed-off-by Norbert Lange <[email protected]>
In the end, I ended up applying Yann's v4 instead, with a few more
modifications, as reported in that thread. I did take inspiration from this patch - mostly from the commit message.
For your information, I'll give some feedback on this patch as well.
Quoted 25 lines
--- .../overlayfs/[email protected] | 26 +++++++++++++++++++ .../overlayfs/[email protected] | 26 +++++++++++++++++++ .../overlayfs/overlay.mount.in | 16 ++++++++++++ .../skeleton-init-systemd.mk | 17 ++++++++++++ system/Config.in | 13 ++++++++++ 5 files changed, 98 insertions(+)create mode 100644 package/skeleton-init-systemd/overlayfs/[email protected] create mode 100644 package/skeleton-init-systemd/overlayfs/[email protected] create mode 100644 package/skeleton-init-systemd/overlayfs/overlay.mount.in
package/skeleton-init-systemd/overlayfs/[email protected] new file +8
-0@@ -0,0 +1,26 @@ +[Unit] +Description=Bind-mount rootfs directory (/%I) to /run +Documentation=man:file-hierarchy(7) +ConditionPathIsSymbolicLink=!/%I +DefaultDependencies=no +Conflicts=umount.target +Before=umount.target +
package/skeleton-init-systemd/overlayfs/[email protected] line 18
Quoted 10 lines
package/skeleton-init-systemd/overlayfs/[email protected] new file +10
-0+# Needs to run after rootfs is properly mounted +# and before regular mounts might interfere. +After=systemd-remount-fs.service +Before=local-fs-pre.target + +[Service] +Type=oneshot +RemainAfterExit=yes +# dont fail if common dirs already exist +ExecStartPre=/bin/mkdir -m755 -p /run/.br
There's no reason to make it a hidden directory. Also, there is no reason to
abbreviate the directory name so much.
package/skeleton-init-systemd/overlayfs/[email protected] line 19
Quoted 1 line
package/skeleton-init-systemd/overlayfs/[email protected] new file +1
-0+ExecStartPre=/bin/mkdir -m700 -p /run/.br/bnd
I don't see why it's useful to have a separate base directory for bind mounts.
When you do a bind mount like this, it's always for a very specific purpose - not always for an overlay, but in any case, it makes more sense to mount on a mount point that is more related to the sepecific purpose.
For the overlay case, it's very convenient to have everything that the overlay
builds on next to each other, i.e. lower, upper, work.
Quoted 20 lines
package/skeleton-init-systemd/overlayfs/[email protected] new file +7
-0+ +ExecStartPre=/bin/mkdir -m700 -p /run/.br/bnd/%I +ExecStart=/bin/mount --make-private -n --bind -o ro /%I /run/.br/bnd/%I + +# lazy unmount, dont block shutdown under any circumstances +ExecStop=/bin/umount -n -l /run/.br/bnd/%I +ExecStopPost=-/bin/rmdir /run/.br/bnd/%Ipackage/skeleton-init-systemd/overlayfs/[email protected] new file +7
-0@@ -0,0 +1,26 @@ +[Unit] +Description=Mount and prepare tmpfs for overlay (/%I) +Documentation=man:file-hierarchy(7) +ConditionPathIsSymbolicLink=!/%I +DefaultDependencies=no +Conflicts=umount.target +Before=local-fs.target umount.target
package/skeleton-init-systemd/overlayfs/[email protected] line 17
Quoted 10 lines
package/skeleton-init-systemd/overlayfs/[email protected] new file +10
-0+# prepare for systemd mount units aswell +RequiresMountsFor=/run/.br/ovl/%I + +[Service] +Environment="OVERLAY_DIR=/run/.br/ovl/%I" +Type=oneshot +RemainAfterExit=yes +# dont fail if common dirs already exist +ExecStartPre=/bin/mkdir -m755 -p /run/.br +ExecStartPre=/bin/mkdir -m700 -p /run/.br/ovl
I'm not sure if there's a good reason to make this mode 700 instead of the
default 755. If there is a good reason, I'd accept a patch that modifies the prepare-var-overlay.service that I ended up applying.
package/skeleton-init-systemd/overlayfs/[email protected] line 21
Quoted 3 lines
package/skeleton-init-systemd/overlayfs/[email protected] new file +3
-0+ExecStartPre=/bin/mkdir -m700 -p ${OVERLAY_DIR} +# Create an override and edit this line for customization +ExecStart=/bin/mount --make-private -n -t tmpfs tmpfs_br_ovl ${OVERLAY_DIR}
There is in fact no reason to mount anything here - /run is already a tmpfs
with all the required properties. We can just reuse it.
Quoted 11 lines
package/skeleton-init-systemd/overlayfs/[email protected] new file +5
-0+ExecStartPost=/bin/mkdir -p ${OVERLAY_DIR}/up ${OVERLAY_DIR}/wd + +# lazy unmount, dont block shutdown under any circumstances +ExecStop=/bin/umount -n -l ${OVERLAY_DIR} +ExecStopPost=/bin/rmdir ${OVERLAY_DIR}package/skeleton-init-systemd/overlayfs/overlay.mount.in new file +0
-0
package/skeleton-init-systemd/overlayfs/overlay.mount.in line 4
Quoted 4 lines
package/skeleton-init-systemd/overlayfs/overlay.mount.in new file +4
-0@@ -0,0 +1,16 @@ +[Unit] +Description=Variable storage (/@PATH@) +Documentation=man:file-hierarchy(7) +ConditionPathIsSymbolicLink=!/@PATH@
There's too much @PATH@ here. I actually think we can completely avoid it by
using %p or %P as appropriate everywhere.
I think that that way, we can indeed reuse both the overlay.mount.in and
[email protected] for other overlays. I unfortunately didn't have time to implement and test that, so instead I committed it without the %p/%P and without a template service.
If you end up implementing it as a template, please make sure to also include
a runtime test for it, e.g. with /etc as the overlay. Or even better something deeper - for the test it could be something silly, like /usr/share.
Quoted 22 lines
package/skeleton-init-systemd/overlayfs/overlay.mount.in new file +12
-0+ +After=br-bindmount-run@@[email protected] +Requires=br-bindmount-run@@[email protected] + +After=br-overlay-prepare@@[email protected] +BindsTo=br-overlay-prepare@@[email protected] + +[Mount] +Type=overlay +What=br_ovl_@PATH@ +Where=/@PATH@ +Options=redirect_dir=on,index=on,xino=on,lowerdir=/run/.br/bnd/@PATH@,upperdir=/run/.br/ovl/@PATH@/up,workdir=/run/.br/ovl/@PATH@/wdpackage/skeleton-init-systemd/skeleton-init-systemd.mk +2
-0@@ -57,6 +57,23 @@ define SKELETON_INIT_SYSTEMD_PRE_ROOTFS_VAR endef SKELETON_INIT_SYSTEMD_ROOTFS_PRE_CMD_HOOKS += SKELETON_INIT_SYSTEMD_PRE_ROOTFS_VAR endif # BR2_INIT_SYSTEMD_VAR_FACTORY + +ifeq ($(BR2_INIT_SYSTEMD_VAR_OVERLAYFS),y)
package/skeleton-init-systemd/skeleton-init-systemd.mk define SKELETON_INIT_SYSTEMD_PRE_ROOTFS_VAR · line 71
Quoted 10 lines
package/skeleton-init-systemd/skeleton-init-systemd.mk +10
-0+define SKELETON_INIT_SYSTEMD_LINUX_CONFIG_FIXUPS + $(call KCONFIG_ENABLE_OPT,CONFIG_OVERLAY_FS) +endef +define SKELETON_INIT_SYSTEMD_POST_INSTALL_VAR_OVERLAYFS + sed 's,@PATH@,var,g' $(SKELETON_INIT_SYSTEMD_PKGDIR)/overlayfs/overlay.mount.in >$(@D)/var.mount + $(INSTALL) -D -m 0644 $(@D)/var.mount $(TARGET_DIR)/usr/lib/systemd/system/var.mount + $(INSTALL) -D -m 0644 $(SKELETON_INIT_SYSTEMD_PKGDIR)/overlayfs/[email protected] \ + $(TARGET_DIR)/usr/lib/systemd/system/[email protected] + $(INSTALL) -D -m 0644 $(SKELETON_INIT_SYSTEMD_PKGDIR)/overlayfs/[email protected] \ + $(TARGET_DIR)/usr/lib/systemd/system/[email protected]
If it's templates, they can actually be installed independently of theBR2_INIT_SYSTEMD_VAR_OVERLAYFS option - maybe you want to use a factory for /var but an overlay for /etc...
I've marked this patch as Superseded (because I took Yann's v4). The rest of
the series I'll look at later.
Regards,
Arnout
Quoted 22 lines
package/skeleton-init-systemd/skeleton-init-systemd.mk +5
-0+ # /var mount gets pulled in automatically by basic.target +endef +SKELETON_INIT_SYSTEMD_POST_INSTALL_TARGET_HOOKS += SKELETON_INIT_SYSTEMD_POST_INSTALL_VAR_OVERLAYFS +endif # BR2_INIT_SYSTEMD_VAR_OVERLAYFS + endif # BR2_TARGET_GENERIC_REMOUNT_ROOTFS_RW ifeq ($(BR2_INIT_SYSTEMD_POPULATE_TMPFILES),y)system/Config.in +6
-0@@ -187,6 +187,19 @@ config BR2_INIT_SYSTEMD_VAR_FACTORY var.mount unit, that overrides the What and Type, and possibly the Options and After, fields. +config BR2_INIT_SYSTEMD_VAR_OVERLAYFS + bool "mount an overlayfs backed by a tmpfs" + help + Mount an overlayfs on /var, with the upper as a tmpfs. + + You can customize the backing storage with an override, for
system/Config.in config BR2_INIT_SYSTEMD_VAR_FACTORY · line 205
Quoted 10 lines
system/Config.in +7
-0+ ecample create this file and add the content following: + /lib/systemd/system/[email protected]/mount.conf + + [Service] + ExecStart= + ExecStart=/bin/mount --make-private -n /dev/sdc1 ${OVERLAY_DIR} + config BR2_INIT_SYSTEMD_VAR_NONE bool "do nothing" help
_______________________________________________
buildroot mailing list
[email protected]
https://lists.buildroot.org/mailman/listinfo/buildroot
Re: [Buildroot] [PATCH 2/7] system: add overlayfs option for ro /var
Am So., 8. Okt. 2023 um 20:37 Uhr schrieb Arnout Vandecappelle <[email protected]>:
Quoted 25 lines
Hi Norbert,
Quoted 13 lines · Norbert Lange
This commit adds an alternative that has the following characteristics:
(quoted diff) +0
-4- Dont depend on anything being available, except the API File Systems [1]. As /var is meant to be available before normal and even some early services are running. - Be a clean drop-in, that can be trivially added / removed. - Depend on overlayfs being available in the kernel. - Units are supposed to be reusable for custom solutions.[snip]
Quoted 5 lines
Cc Yann E. MORIN <[email protected]>Cc Romain Naour <[email protected]>Cc Jérémy Rosen <[email protected]>Signed-off-by Norbert Lange <[email protected]>
Quoted 5 lines
In the end, I ended up applying Yann's v4 instead, with a few more
modifications, as reported in that thread. I did take inspiration from this patch - mostly from the commit message.For your information, I'll give some feedback on this patch as well.
Happy that there finally is some movement, still its weird that after a long back-and-forth
a modified version is committed before feedback.
Quoted 31 lines
Quoted 31 lines
--- .../overlayfs/[email protected] | 26 +++++++++++++++++++ .../overlayfs/[email protected] | 26 +++++++++++++++++++ .../overlayfs/overlay.mount.in | 16 ++++++++++++ .../skeleton-init-systemd.mk | 17 ++++++++++++ system/Config.in | 13 ++++++++++ 5 files changed, 98 insertions(+)create mode 100644 package/skeleton-init-systemd/overlayfs/[email protected] create mode 100644 package/skeleton-init-systemd/overlayfs/[email protected] create mode 100644 package/skeleton-init-systemd/overlayfs/overlay.mount.in
package/skeleton-init-systemd/overlayfs/[email protected] new file +13
-0@@ -0,0 +1,26 @@ +[Unit] +Description=Bind-mount rootfs directory (/%I) to /run +Documentation=man:file-hierarchy(7) +ConditionPathIsSymbolicLink=!/%I +DefaultDependencies=no +Conflicts=umount.target +Before=umount.target + +# Needs to run after rootfs is properly mounted +# and before regular mounts might interfere. +After=systemd-remount-fs.service +Before=local-fs-pre.target +
Quoted 8 lines
package/skeleton-init-systemd/overlayfs/[email protected] line 18
Quoted 5 lines
package/skeleton-init-systemd/overlayfs/[email protected] new file +5
-0+[Service] +Type=oneshot +RemainAfterExit=yes +# dont fail if common dirs already exist +ExecStartPre=/bin/mkdir -m755 -p /run/.brThere's no reason to make it a hidden directory. Also, there is no reason to
abbreviate the directory name so much.
should be short imho, to avoid running into path-length problems. its a technical detail, hidden from users
Quoted 7 lines
package/skeleton-init-systemd/overlayfs/[email protected] line 19
Quoted 2 lines
package/skeleton-init-systemd/overlayfs/[email protected] new file +1
-0+ExecStartPre=/bin/mkdir -m700 -p /run/.br/bndI don't see why it's useful to have a separate base directory for bind mounts.
When you do a bind mount like this, it's always for a very specific purpose - not always for an overlay, but in any case, it makes more sense to mount on a mount point that is more related to the sepecific purpose.
Because /run/.br could be for anything else later, and /run/.br/bnd should mirror
the filesystem root.
Quoted 3 lines
For the overlay case, it's very convenient to have everything that the overlay
builds on next to each other, i.e. lower, upper, work.
Yeah, there was a reason for not doing that, had to do with being able to re-use that system of generic mount templates, ie. /run/.br/bnd would always mirror a part of the fs root, without changed hierarchies.
ie. you could either use a single mount / to /run/.br/bnd, or say /etc to /run/.br/bnd/etc and
/var to /run/.br/bnd/var. It would not effect the overlay units.
Quoted 8 lines
package/skeleton-init-systemd/overlayfs/[email protected] line 26
Quoted 8 lines
package/skeleton-init-systemd/overlayfs/[email protected] new file +7
-0+ +ExecStartPre=/bin/mkdir -m700 -p /run/.br/bnd/%I +ExecStart=/bin/mount --make-private -n --bind -o ro /%I /run/.br/bnd/%I + +# lazy unmount, dont block shutdown under any circumstances +ExecStop=/bin/umount -n -l /run/.br/bnd/%I +ExecStopPost=-/bin/rmdir /run/.br/bnd/%I
This got left out aswell:
ExecStopPost=-/bin/rmdir /run/.br/bnd/%I
Removing it, would allow new accesses to the filesystem thats in progress of being lazily unmounted.
Quoted 19 lines
Quoted 19 lines
package/skeleton-init-systemd/overlayfs/[email protected] new file +13
-0@@ -0,0 +1,26 @@ +[Unit] +Description=Mount and prepare tmpfs for overlay (/%I) +Documentation=man:file-hierarchy(7) +ConditionPathIsSymbolicLink=!/%I +DefaultDependencies=no +Conflicts=umount.target +Before=local-fs.target umount.target +# prepare for systemd mount units aswell +RequiresMountsFor=/run/.br/ovl/%I + +[Service] +Environment="OVERLAY_DIR=/run/.br/ovl/%I" +Type=oneshot
Quoted 8 lines
package/skeleton-init-systemd/overlayfs/[email protected] line 17
Quoted 4 lines
package/skeleton-init-systemd/overlayfs/[email protected] new file +4
-0+RemainAfterExit=yes +# dont fail if common dirs already exist +ExecStartPre=/bin/mkdir -m755 -p /run/.br +ExecStartPre=/bin/mkdir -m700 -p /run/.br/ovlI'm not sure if there's a good reason to make this mode 700 instead of the
default 755. If there is a good reason, I'd accept a patch that modifies the prepare-var-overlay.service that I ended up applying.
No one should access this directories directly, its being defensive
Quoted 7 lines
package/skeleton-init-systemd/overlayfs/[email protected] line 21
Quoted 4 lines
package/skeleton-init-systemd/overlayfs/[email protected] new file +3
-0+ExecStartPre=/bin/mkdir -m700 -p ${OVERLAY_DIR} +# Create an override and edit this line for customization +ExecStart=/bin/mount --make-private -n -t tmpfs tmpfs_br_ovl ${OVERLAY_DIR}There is in fact no reason to mount anything here - /run is already a tmpfs
with all the required properties. We can just reuse it.
Yes, problem is that the /run mount is limited in size, not configurable (parameters are in systemd's sources) and
filling /var with stuff should never exhaust one of the most critical filesystems.
Quoted 20 lines
Quoted 12 lines
package/skeleton-init-systemd/overlayfs/[email protected] new file +5
-0+ExecStartPost=/bin/mkdir -p ${OVERLAY_DIR}/up ${OVERLAY_DIR}/wd + +# lazy unmount, dont block shutdown under any circumstances +ExecStop=/bin/umount -n -l ${OVERLAY_DIR} +ExecStopPost=/bin/rmdir ${OVERLAY_DIR}package/skeleton-init-systemd/overlayfs/overlay.mount.in new file +0
-0package/skeleton-init-systemd/overlayfs/overlay.mount.in line 4
Quoted 4 lines
package/skeleton-init-systemd/overlayfs/overlay.mount.in new file +4
-0@@ -0,0 +1,16 @@ +[Unit] +Description=Variable storage (/@PATH@) +Documentation=man:file-hierarchy(7) +ConditionPathIsSymbolicLink=!/@PATH@There's too much @PATH@ here. I actually think we can completely avoid it by
using %p or %P as appropriate everywhere.
Quoted 8 lines
I think that that way, we can indeed reuse both the overlay.mount.in and
[email protected] for other overlays. I unfortunately didn't have time to implement and test that, so instead I committed it without the %p/%P and without a template service.If you end up implementing it as a template, please make sure to also include
a runtime test for it, e.g. with /etc as the overlay. Or even better something deeper - for the test it could be something silly, like/usr/share.
Yeah, gonna rebase my stuff and pickup where I stopped long ago, gonna take a while.
Quoted 29 lines
Quoted 29 lines
package/skeleton-init-systemd/overlayfs/overlay.mount.in new file +12
-0+ +After=br-bindmount-run@@[email protected] +Requires=br-bindmount-run@@[email protected] + +After=br-overlay-prepare@@[email protected] +BindsTo=br-overlay-prepare@@[email protected] + +[Mount] +Type=overlay +What=br_ovl_@PATH@ +Where=/@PATH@ +Options=redirect_dir=on,index=on,xino=on,lowerdir=/run/.br/bnd/@PATH@,upperdir=/run/.br/ovl/@PATH@/up,workdir=/run/.br/ovl/@PATH@/wdpackage/skeleton-init-systemd/skeleton-init-systemd.mk +8
-0@@ -57,6 +57,23 @@ define SKELETON_INIT_SYSTEMD_PRE_ROOTFS_VAR endef SKELETON_INIT_SYSTEMD_ROOTFS_PRE_CMD_HOOKS += SKELETON_INIT_SYSTEMD_PRE_ROOTFS_VAR endif # BR2_INIT_SYSTEMD_VAR_FACTORY + +ifeq ($(BR2_INIT_SYSTEMD_VAR_OVERLAYFS),y) +define SKELETON_INIT_SYSTEMD_LINUX_CONFIG_FIXUPS + $(call KCONFIG_ENABLE_OPT,CONFIG_OVERLAY_FS) +endef +define SKELETON_INIT_SYSTEMD_POST_INSTALL_VAR_OVERLAYFS + sed 's,@PATH@,var,g' $(SKELETON_INIT_SYSTEMD_PKGDIR)/overlayfs/overlay.mount.in >$(@D)/var.mount + $(INSTALL) -D -m 0644 $(@D)/var.mount $(TARGET_DIR)/usr/lib/systemd/system/var.mount
Quoted 8 lines
package/skeleton-init-systemd/skeleton-init-systemd.mk define SKELETON_INIT_SYSTEMD_PRE_ROOTFS_VAR · line 71
Quoted 4 lines
package/skeleton-init-systemd/skeleton-init-systemd.mk +4
-0+ $(INSTALL) -D -m 0644 $(SKELETON_INIT_SYSTEMD_PKGDIR)/overlayfs/[email protected] \ + $(TARGET_DIR)/usr/lib/systemd/system/[email protected] + $(INSTALL) -D -m 0644 $(SKELETON_INIT_SYSTEMD_PKGDIR)/overlayfs/[email protected] \ + $(TARGET_DIR)/usr/lib/systemd/system/[email protected]If it's templates, they can actually be installed independently of the
BR2_INIT_SYSTEMD_VAR_OVERLAYFSoption - maybe you want to use a factory for /var but an overlay for /etc...
Sure, but not sure if it should be installed unconditionally. Easy to just copy those files into a
rootfs-overlay.
Quoted 4 lines
I've marked this patch as Superseded (because I took Yann's v4). The rest of
the series I'll look at later.
Ok
Quoted 4 lines
Regards,
Arnout
Regards,
Norbert
_______________________________________________
buildroot mailing list
[email protected]
https://lists.buildroot.org/mailman/listinfo/buildroot
PATCH 3/7 package/skeleton-init-systemd: force permissions for buildroot runtime dirs
In reply to Norbert LangeA previous commit introduced /run/.br and /tmp/.br and a few subdirectories.
Add a tmpfile to ensure the existance, ownership and permissions.
--- package/skeleton-init-systemd/00-buildroot.conf | 7 +++++++ package/skeleton-init-systemd/skeleton-init-systemd.mk | 1 + 2 files changed, 8 insertions(+) create mode 100644 package/skeleton-init-systemd/00-buildroot.conf
package/skeleton-init-systemd/00-buildroot.conf new file +7 -0
@@ -0,0 +1,7 @@ +# Create buildroot API directories and fix permissions + +d /run/.br 0755 root root - +d /tmp/.br 0755 root root - + +z /run/.br/bnd 0700 root root - +z /run/.br/ovl 0700 root root -
package/skeleton-init-systemd/skeleton-init-systemd.mk +1 -0
@@ -91,6 +91,7 @@ define SKELETON_INIT_SYSTEMD_INSTALL_TARGET_CMDS ln -s ../run $(TARGET_DIR)/var/run # prevent install scripts to create var/lock as directory ln -s ../run/lock $(TARGET_DIR)/var/lock + install -D -m644 $(SKELETON_INIT_SYSTEMD_PKGDIR)/00-buildroot.conf $(TARGET_DIR)/usr/lib/tmpfiles.d/00-buildroot.conf install -D -m644 $(SKELETON_INIT_SYSTEMD_PKGDIR)/legacy.conf $(TARGET_DIR)/usr/lib/tmpfiles.d/legacy.conf $(SKELETON_INIT_SYSTEMD_ROOT_RO_OR_RW) endef
Signature · 6 lines
2.39.0 _______________________________________________ buildroot mailing list [email protected] https://lists.buildroot.org/mailman/listinfo/buildroot
Re: [Buildroot] [PATCH 3/7] package/skeleton-init-systemd: force permissions for buildroot runtime dirs
Hi Norbert,
Obviously, this doesn't apply any more because I committed a different tree.
However...
Quoted 17 lines · Norbert Lange
A previous commit introduced
/run/.brand/tmp/.brand a few subdirectories.
Add a tmpfile to ensure the existance, ownership and permissions.Signed-off-by Norbert Lange <[email protected]>--- package/skeleton-init-systemd/00-buildroot.conf | 7 +++++++ package/skeleton-init-systemd/skeleton-init-systemd.mk | 1 + 2 files changed, 8 insertions(+)create mode 100644
package/skeleton-init-systemd/00-buildroot.confpackage/skeleton-init-systemd/00-buildroot.conf new file +0
-0
package/skeleton-init-systemd/00-buildroot.conf line 4
Quoted 4 lines
package/skeleton-init-systemd/00-buildroot.conf new file +4
-0@@ -0,0 +1,7 @@ +# Create buildroot API directories and fix permissions + +d /run/.br 0755 root root - +d /tmp/.br 0755 root root -
I don't think this tmp directory is used anywhere?
package/skeleton-init-systemd/00-buildroot.conf line 7
Quoted 2 lines
package/skeleton-init-systemd/00-buildroot.conf new file +2
-0+z /run/.br/bnd 0700 root root - +z /run/.br/ovl 0700 root root -
What's the point of doing this? These directories are created by the overlay
service file. systemd-tmpfiles is started after local-fs.target and the /var overlay mount comes before, so it's the overlay service file that has to create them. You could say "hey maybe there's some other service that makes use of those directories" - but if so, that's another service that is going to come before local-fs.target, so it _also_ has to create those directories itself...
There's a bit of a problem with systemd-tmpfiles, that it doesn't allow to
specify dependencies/ordering. That's in fact one of the reasons the factory approach doesn't work so well. systemd-tmpfiles only works for services after sysinit.target, really - and therefore doesn't work for anything related to local filesystems.
IMHO it would have been better to put tmpfile blocks in service definitions.
But that's not the world we live in :-)
So, I've marked as Rejected, but feel free to argue my points.
Regards,
Arnout
package/skeleton-init-systemd/skeleton-init-systemd.mk define SKELETON_INIT_SYSTEMD_INSTALL_TARGET_CMDS · line 97
Quoted 12 lines
package/skeleton-init-systemd/skeleton-init-systemd.mk +1
-0@@ -91,6 +91,7 @@ define SKELETON_INIT_SYSTEMD_INSTALL_TARGET_CMDS ln -s ../run $(TARGET_DIR)/var/run # prevent install scripts to create var/lock as directory ln -s ../run/lock $(TARGET_DIR)/var/lock + install -D -m644 $(SKELETON_INIT_SYSTEMD_PKGDIR)/00-buildroot.conf $(TARGET_DIR)/usr/lib/tmpfiles.d/00-buildroot.conf install -D -m644 $(SKELETON_INIT_SYSTEMD_PKGDIR)/legacy.conf $(TARGET_DIR)/usr/lib/tmpfiles.d/legacy.conf $(SKELETON_INIT_SYSTEMD_ROOT_RO_OR_RW) endef
_______________________________________________
buildroot mailing list
[email protected]
https://lists.buildroot.org/mailman/listinfo/buildroot
Re: [Buildroot] [PATCH 3/7] package/skeleton-init-systemd: force permissions for buildroot runtime dirs
Am Di., 10. Okt. 2023 um 21:32 Uhr schrieb Arnout Vandecappelle <[email protected]>:
Quoted 18 lines
Hi Norbert,
Obviously, this doesn't apply any more because I committed a different tree.
However...Quoted 11 lines · Norbert Lange
A previous commit introduced
/run/.brand/tmp/.brand a few subdirectories.
Add a tmpfile to ensure the existance, ownership and permissions.Signed-off-by Norbert Lange <[email protected]>--- package/skeleton-init-systemd/00-buildroot.conf | 7 +++++++ package/skeleton-init-systemd/skeleton-init-systemd.mk | 1 + 2 files changed, 8 insertions(+)create mode 100644
package/skeleton-init-systemd/00-buildroot.conf
Quoted 12 lines
package/skeleton-init-systemd/00-buildroot.conf line 4
Quoted 10 lines
package/skeleton-init-systemd/00-buildroot.conf new file +4
-0@@ -0,0 +1,7 @@ +# Create buildroot API directories and fix permissions + +d /run/.br 0755 root root - +d /tmp/.br 0755 root root -I don't think this tmp directory is used anywhere?
Not yet, these would reserve the paths. One usage would
be to use /tmp/.br as base for overlay "upper" dirs, reusing the /tmp filesystem instead of mounting an extra tmpfs. Using /run for anything else but small files is a bad idea, as I mentioned in response to the commit "package/skeleton-init-systemd: add option to use overlayfs on /var".
Quoted 10 lines
package/skeleton-init-systemd/00-buildroot.conf line 7
Quoted 3 lines
package/skeleton-init-systemd/00-buildroot.conf new file +2
-0+z /run/.br/bnd 0700 root root - +z /run/.br/ovl 0700 root root -What's the point of doing this? These directories are created by the overlay
service file. systemd-tmpfiles is started after local-fs.target and the /var overlay mount comes before, so it's the overlay service file that has to create them. You could say "hey maybe there's some other service that makes use of those directories" - but if so, that's another service that is going to come before local-fs.target, so it _also_ has to create those directories itself...
Not all overlays/services have to be enabled, you generally should use tmpfiles.d
to ensure owner/mode of some sort of protected files. Prevents hijacking.
Think of mounting an overlay on-demand, running an system update, testing said update then "commiting" the (merged) overlay to a blockdevice. if anything breaks, you reboot and you are running the unmodified working system.
Quoted 9 lines
There's a bit of a problem with systemd-tmpfiles, that it doesn't allow to
specify dependencies/ordering. That's in fact one of the reasons the factory approach doesn't work so well. systemd-tmpfiles only works for services after sysinit.target, really - and therefore doesn't work for anything related to local filesystems.IMHO it would have been better to put tmpfile blocks in service definitions.
But that's not the world we live in :-)
That's not what they are for. tmpfiles.d should set up the filesystem whether services are enabled or not. Erases a whole set of problems.
Quoted 2 lines
So, I've marked as Rejected, but feel free to argue my points.
The argument would be to create & protect paths buildroot services could be using.
Quoted 4 lines
Regards,
Arnout
Regards, Norbert
_______________________________________________
buildroot mailing list
[email protected]
https://lists.buildroot.org/mailman/listinfo/buildroot
PATCH 4/7 system: add systemd generator for /var
In reply to Norbert LangeThis commit adds an alternative that has the following characteristics:
- Dont depend on anything being available, except the
API File Systems [1].
- Be a clean drop-in, that can be trivially added / removed.
- Runs before systemd loads its configuration.
Could be extended to handle more tricky paths like /etc.
For more explaination, see the commit introducing the overlay method.
The implementation doesnt focus on being expandable but simple, in the future there could be an attempt to source shell scripts for configuration.
So that multiple paths can be covered (/etc),
or the mounts and handling specified
(do-nothing, copy-over or overlay original contents).
The intent is to have a direct replacement for the var factory method.
[1] - https://www.freedesktop.org/wiki/Software/systemd/APIFileSystems/
--- .../generator/br-mount-generator | 19 +++++++++++++++++++ .../skeleton-init-systemd.mk | 10 +++++++++- system/Config.in | 13 +++++++++++++ 3 files changed, 41 insertions(+), 1 deletion(-) create mode 100755 package/skeleton-init-systemd/generator/br-mount-generator
package/skeleton-init-systemd/generator/br-mount-generator new file +19 -0
@@ -0,0 +1,19 @@ +#!/bin/sh +set -e +# SPDX-License-Identifier: MIT +# Note: doing heavy stuff here is not recommended, but this way +# the ordering is guaranteed. Be quick about everything. +# Know that the generator can be run again later. + +[ "${SYSTEMD_IN_INITRD-0}" = 1 ] && exit +grep '^tmpfs_br_var /var tmpfs' /proc/self/mounts >/dev/null && exit + +TMPMOUNT=/run/.br/.tmpvar +trap "umount -l $TMPMOUNT || :; rmdir $TMPMOUNT || :" 0 + +mkdir -m755 -p $TMPMOUNT +mount -n -t tmpfs -o nosuid,nodev,size=50% tmpfs_br_var $TMPMOUNT + +cp -r -p /var/. $TMPMOUNT +# this is a shared mount so --move wont work +mount -n --bind $TMPMOUNT /var
package/skeleton-init-systemd/skeleton-init-systemd.mk +9 -1
@@ -69,11 +69,19 @@ define SKELETON_INIT_SYSTEMD_POST_INSTALL_VAR_OVERLAYFS $(TARGET_DIR)/usr/lib/systemd/system/[email protected] $(INSTALL) -D -m 0644 $(SKELETON_INIT_SYSTEMD_PKGDIR)/overlayfs/[email protected] \ $(TARGET_DIR)/usr/lib/systemd/system/[email protected] - # /var mount gets pulled in automatically by basic.target + # var.mount gets pulled in automatically by basic.target endef SKELETON_INIT_SYSTEMD_POST_INSTALL_TARGET_HOOKS += SKELETON_INIT_SYSTEMD_POST_INSTALL_VAR_OVERLAYFS endif # BR2_INIT_SYSTEMD_VAR_OVERLAYFS +ifeq ($(BR2_INIT_SYSTEMD_VAR_GENERATOR),y) +define SKELETON_INIT_SYSTEMD_POST_INSTALL_VAR_GENERATOR + $(INSTALL) -D -m 0755 $(SKELETON_INIT_SYSTEMD_PKGDIR)/generator/br-mount-generator \ + $(TARGET_DIR)/usr/lib/systemd/system-generators/br-var-mount-generator +endef +SKELETON_INIT_SYSTEMD_POST_INSTALL_TARGET_HOOKS += SKELETON_INIT_SYSTEMD_POST_INSTALL_VAR_GENERATOR +endif # BR2_INIT_SYSTEMD_VAR_GENERATOR + endif # BR2_TARGET_GENERIC_REMOUNT_ROOTFS_RW ifeq ($(BR2_INIT_SYSTEMD_POPULATE_TMPFILES),y)
system/Config.in +13 -0
@@ -200,6 +200,19 @@ config BR2_INIT_SYSTEMD_VAR_OVERLAYFS ExecStart= ExecStart=/bin/mount --make-private -n /dev/sdc1 ${OVERLAY_DIR} +config BR2_INIT_SYSTEMD_VAR_GENERATOR + bool "use a systemd generator to mount and populate a tmpfs" + help + Mount an tmpfs on /var, with the the original contents copied. + + The logic is contained in a systemd generator, to ensure this + is done early and ordered before anything can access /var. + + Unlike the var factory, nothing needs to be prepared and + the populating happens before any unit accesses /var. + + Unlike the var factory, it does not need an overlayfs. + config BR2_INIT_SYSTEMD_VAR_NONE bool "do nothing" help
Signature · 6 lines
2.39.0 _______________________________________________ buildroot mailing list [email protected] https://lists.buildroot.org/mailman/listinfo/buildroot
Re: [Buildroot] [PATCH 4/7] system: add systemd generator for /var
Hi Norbert,
I'm not entirely sure about this one, but not completely against it either.
Since I do have a few comments below, I've marked as Changes Requested.
Quoted 1 line · Norbert Lange
This commit adds an alternative that has the following characteristics:
This commit adds an alternative /var management method that has the following
characteristics:
Review
Quoted 9 lines
(quoted diff) +0
-3- Dont depend on anything being available, except the API File Systems [1]. - Be a clean drop-in, that can be trivially added / removed. - Runs before systemd loads its configuration. Could be extended to handle more tricky paths like /etc.
Well, for /etc there is no real reason for it to be writeable (but not
persistent), is there? There are some debugging situations where it could be useful, but then I think you could just as well use the overlayfs approach.
What is missing here still is why you'd prefer this over the overlay approach.
That's explained in the help text though. Not that I really believe there's a good reason though - see below.
Quoted 1 line
For more explaination, see the commit introducing the overlay method.
That's commit 10c637ab06d935bbce4c833c4ab9695cdb32b6c8 now. I put most of your
explanation in the commit message so this reference should still apply.
explaination -> explanation
Quoted 2 lines
The implementation doesnt focus on being expandable but simple,
doesn't
Quoted 5 lines
in the future there could be an attempt to source shell scripts for configuration.
So that multiple paths can be covered (/etc),
or the mounts and handling specified
(do-nothing, copy-over or overlay original contents).
I don't think it makes sense to refer to an unproven theoretical extension
like trying for some kind of half-baked writeable overlay without using overlayfs. Well, I shouldn't be talking, my current project is in fact doing exactly that - but let me tell you, it's half baked :-)
Quoted 21 lines
The intent is to have a direct replacement for the var factory method.
[1] - https://www.freedesktop.org/wiki/Software/systemd/APIFileSystems/
Cc Yann E. MORIN <[email protected]>Cc Romain Naour <[email protected]>Cc Jérémy Rosen <[email protected]>Signed-off-by Norbert Lange <[email protected]>--- .../generator/br-mount-generator | 19 +++++++++++++++++++ .../skeleton-init-systemd.mk | 10 +++++++++- system/Config.in | 13 +++++++++++++ 3 files changed, 41 insertions(+), 1 deletion(-)create mode 100755 package/skeleton-init-systemd/generator/br-mount-generator
package/skeleton-init-systemd/generator/br-mount-generator new file +0
-0
package/skeleton-init-systemd/generator/br-mount-generator line 3
Quoted 3 lines
package/skeleton-init-systemd/generator/br-mount-generator new file +3
-0@@ -0,0 +1,19 @@ +#!/bin/sh +set -e +# SPDX-License-Identifier: MIT
The MIT license suggest that you copied this from somewhere else. If so, you
must refer to the original - at least to the original author, as required by the MIT license.
package/skeleton-init-systemd/generator/br-mount-generator line 8
Quoted 5 lines
package/skeleton-init-systemd/generator/br-mount-generator new file +5
-0+# Note: doing heavy stuff here is not recommended, but this way +# the ordering is guaranteed. Be quick about everything. +# Know that the generator can be run again later. + +[ "${SYSTEMD_IN_INITRD-0}" = 1 ] && exit
I don't like this redundant default of 0 much. But that's just minor.
package/skeleton-init-systemd/generator/br-mount-generator line 9
Quoted 1 line
package/skeleton-init-systemd/generator/br-mount-generator new file +1
-0+grep '^tmpfs_br_var /var tmpfs' /proc/self/mounts >/dev/null && exit
Why >/dev/null instead of grep -q?
package/skeleton-init-systemd/generator/br-mount-generator line 18
Quoted 8 lines
package/skeleton-init-systemd/generator/br-mount-generator new file +8
-0+ +TMPMOUNT=/run/.br/.tmpvar +trap "umount -l $TMPMOUNT || :; rmdir $TMPMOUNT || :" 0 + +mkdir -m755 -p $TMPMOUNT +mount -n -t tmpfs -o nosuid,nodev,size=50% tmpfs_br_var $TMPMOUNT > + +cp -r -p /var/. $TMPMOUNT +# this is a shared mount so --move wont work
"this" seems to refer to /run/.br/.tmpvar, which is _not_ a shared mount (and
it's anyway not relevant if it is). /run is a shared mount and that is the reason you can't --move it.
And perhaps add "so instead, bind-mount and unmount $TMPMOUNT afterwards (in
the trap handler)."
package/skeleton-init-systemd/skeleton-init-systemd.mk define SKELETON_INIT_SYSTEMD_POST_INSTALL_VAR_OVERLAYFS · line 72
Quoted 11 lines
package/skeleton-init-systemd/generator/br-mount-generator new file +1
-0+mount -n --bind $TMPMOUNT /varpackage/skeleton-init-systemd/skeleton-init-systemd.mk +1
-1@@ -69,11 +69,19 @@ define SKELETON_INIT_SYSTEMD_POST_INSTALL_VAR_OVERLAYFS $(TARGET_DIR)/usr/lib/systemd/system/[email protected] $(INSTALL) -D -m 0644 $(SKELETON_INIT_SYSTEMD_PKGDIR)/overlayfs/[email protected] \ $(TARGET_DIR)/usr/lib/systemd/system/[email protected] - # /var mount gets pulled in automatically by basic.target + # var.mount gets pulled in automatically by basic.target
If you're anyway going to correct this comment, correct it correctly :-) It's
pulled in by local-fs.target, not basic.target.
package/skeleton-init-systemd/skeleton-init-systemd.mk define SKELETON_INIT_SYSTEMD_POST_INSTALL_VAR_OVERLAYFS · line 80
Quoted 8 lines
package/skeleton-init-systemd/skeleton-init-systemd.mk +4
-0endef SKELETON_INIT_SYSTEMD_POST_INSTALL_TARGET_HOOKS += SKELETON_INIT_SYSTEMD_POST_INSTALL_VAR_OVERLAYFS endif # BR2_INIT_SYSTEMD_VAR_OVERLAYFS +ifeq ($(BR2_INIT_SYSTEMD_VAR_GENERATOR),y) +define SKELETON_INIT_SYSTEMD_POST_INSTALL_VAR_GENERATOR + $(INSTALL) -D -m 0755 $(SKELETON_INIT_SYSTEMD_PKGDIR)/generator/br-mount-generator \ + $(TARGET_DIR)/usr/lib/systemd/system-generators/br-var-mount-generator
Please keep the same name in PKGDIR. I know the intention is to keep it
general so it can be used for other things (e.g. /etc), but it _isn't_ general so don't try pretending it is.
Quoted 15 lines
package/skeleton-init-systemd/skeleton-init-systemd.mk +4
-0+endef +SKELETON_INIT_SYSTEMD_POST_INSTALL_TARGET_HOOKS += SKELETON_INIT_SYSTEMD_POST_INSTALL_VAR_GENERATOR +endif # BR2_INIT_SYSTEMD_VAR_GENERATOR + endif # BR2_TARGET_GENERIC_REMOUNT_ROOTFS_RW ifeq ($(BR2_INIT_SYSTEMD_POPULATE_TMPFILES),y)system/Config.in +0
-0@@ -200,6 +200,19 @@ config BR2_INIT_SYSTEMD_VAR_OVERLAYFS ExecStart= ExecStart=/bin/mount --make-private -n /dev/sdc1 ${OVERLAY_DIR}
system/Config.in config BR2_INIT_SYSTEMD_VAR_OVERLAYFS · line 212
Quoted 10 lines
system/Config.in +10
-0+config BR2_INIT_SYSTEMD_VAR_GENERATOR + bool "use a systemd generator to mount and populate a tmpfs" + help + Mount an tmpfs on /var, with the the original contents copied. + + The logic is contained in a systemd generator, to ensure this + is done early and ordered before anything can access /var. + + Unlike the var factory, nothing needs to be prepared and + the populating happens before any unit accesses /var.
Well, nothing is _allowed_ to access /var before local-fs-pre.target, exactly
because /var may be mounted (and mounting something depends on local-fs.target). If you do need to access /var, you need to do special magic like journald does and wait with using it until it gets mounted.
Do you have an actual proven use case of something that requires /var before
local-fs-pre.target (and that isn't an unacceptable hack)?
system/Config.in config BR2_INIT_SYSTEMD_VAR_OVERLAYFS · line 214
Quoted 1 line
system/Config.in +1
-0+ Unlike the var factory, it does not need an overlayfs.
If this is the only reason, then I don't think it's worth the bother to
replace the factory.
Regards,
Arnout
system/Config.in config BR2_INIT_SYSTEMD_VAR_OVERLAYFS · line 218
Quoted 4 lines
system/Config.in +1
-0+ config BR2_INIT_SYSTEMD_VAR_NONE bool "do nothing" help
_______________________________________________
buildroot mailing list
[email protected]
https://lists.buildroot.org/mailman/listinfo/buildroot
PATCH 5/7 package/skeleton-init-systemd: fix var.mount options
In reply to Norbert LangeThe mode should reflect the mount directory, and 755 is the default - so drop the mount option.
--- package/skeleton-init-systemd/var.mount | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
package/skeleton-init-systemd/var.mount +1 -1
@@ -12,4 +12,4 @@ After=swap.target What=tmpfs Where=/var Type=tmpfs -Options=mode=1777,strictatime,nosuid,nodev,size=50%%,nr_inodes=1m +Options=nosuid,nodev,size=50%%,nr_inodes=1m
Signature · 6 lines
2.39.0 _______________________________________________ buildroot mailing list [email protected] https://lists.buildroot.org/mailman/listinfo/buildroot
Re: [Buildroot] [PATCH 5/7] package/skeleton-init-systemd: fix var.mount options
Quoted 5 lines · Norbert Lange
The mode should reflect the mount directory, and 755 is the default - so drop the mount option.
Cc Yann E. MORIN <[email protected]>Signed-off-by Norbert Lange <[email protected]>
Applied to master, thanks.
Regards,
Arnout
Quoted 9 lines
--- package/skeleton-init-systemd/var.mount | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)package/skeleton-init-systemd/var.mount +0
-0
package/skeleton-init-systemd/var.mount After=swap.target · line 15
Quoted 5 lines
package/skeleton-init-systemd/var.mount +1
-1@@ -12,4 +12,4 @@ After=swap.target What=tmpfs Where=/var Type=tmpfs -Options=mode=1777,strictatime,nosuid,nodev,size=50%%,nr_inodes=1m +Options=nosuid,nodev,size=50%%,nr_inodes=1m
_______________________________________________
buildroot mailing list
[email protected]
https://lists.buildroot.org/mailman/listinfo/buildroot
PATCH 6/7 package/systemd: handle creation of /var/log/journal
In reply to Norbert Langejournald will always have a runtime log in /run/systemd/journal, and a persistent one in /var/log/journal under certain conditions. By default it will check for the existence of that directory.
When /var is not backed by mass-storage you typically dont want your log duplicated and journald should be tuned to account for having only RAM available.
Buildroot will now create this directory conditionally together with it's handling of /var under a read-only filesystem.
It is rather easy for users to opt-in by either creating that directory or editing /etc/systemd/journald.conf.
--- package/systemd/systemd.mk | 7 +++++++ 1 file changed, 7 insertions(+)
package/systemd/systemd.mk +7 -0
@@ -61,6 +61,7 @@ SYSTEMD_SELINUX_MODULES = systemd udev xdg SYSTEMD_PROVIDES = udev SYSTEMD_CONF_OPTS += \ + -Dcreate-log-dirs=false \ -Ddefault-hierarchy=unified \ -Didn=true \ -Dima=false \
@@ -100,6 +101,10 @@ ifeq ($(BR2_nios2),y) SYSTEMD_LDFLAGS = $(TARGET_LDFLAGS) -Wl,--no-fatal-warnings endif +ifeq ($(BR2_TARGET_GENERIC_REMOUNT_ROOTFS_RW),y) +SYSTEMD_JOURNALD_PERMISSIONS = /var/log/journal d 2755 root systemd-journal - - - - - +endif + ifeq ($(BR2_PACKAGE_ACL),y) SYSTEMD_DEPENDENCIES += acl SYSTEMD_CONF_OPTS += -Dacl=true
@@ -603,6 +608,7 @@ define SYSTEMD_PERMISSIONS /var/lib/private d 700 0 0 - - - - - /var/log/private d 700 0 0 - - - - - /var/cache/private d 700 0 0 - - - - - + $(SYSTEMD_JOURNALD_PERMISSIONS) $(SYSTEMD_LOGIND_PERMISSIONS) $(SYSTEMD_MACHINED_PERMISSIONS) $(SYSTEMD_HOMED_PERMISSIONS)
@@ -782,6 +788,7 @@ HOST_SYSTEMD_CONF_OPTS = \ --libdir=lib \ --sysconfdir=/etc \ --localstatedir=/var \ + -Dcreate-log-dirs=false \ -Dmode=release \ -Dutmp=false \ -Dhibernate=false \
Signature · 6 lines
2.39.0 _______________________________________________ buildroot mailing list [email protected] https://lists.buildroot.org/mailman/listinfo/buildroot
Re: [Buildroot] [PATCH 6/7] package/systemd: handle creation of /var/log/journal
Quoted 10 lines · Norbert Lange
journald will always have a runtime log in
/run/systemd/journal, and a persistent one in/var/log/journalunder certain conditions. By default it will check for the existence of that directory.When /var is not backed by mass-storage you typically dont want your log duplicated and journald should be tuned to account for having only RAM available.
Buildroot will now create this directory conditionally together with it's handling of /var under a read-only filesystem.
The patch now only creates the directory if rootfs-rw is enabled.
This is actually not great IMHO - we have too many things hinging on that
config setting that give meaning to it which doesn't really exist. Still, there's no better way to do it, really.
Quoted 2 lines
It is rather easy for users to opt-in by either creating that directory or editing
/etc/systemd/journald.conf.
We really should have documentation of how to do this, and a runtime test that
proves that it works.
But OK, I anyway already committed this to master, thanks.
Regards,
Arnout
Quoted 38 lines
Signed-off-by Norbert Lange <[email protected]>--- package/systemd/systemd.mk | 7 +++++++ 1 file changed, 7 insertions(+)package/systemd/systemd.mk +6
-0@@ -61,6 +61,7 @@ SYSTEMD_SELINUX_MODULES = systemd udev xdg SYSTEMD_PROVIDES = udev SYSTEMD_CONF_OPTS += \ + -Dcreate-log-dirs=false \ -Ddefault-hierarchy=unified \ -Didn=true \ -Dima=false \@@ -100,6 +101,10 @@ ifeq ($(BR2_nios2),y) SYSTEMD_LDFLAGS = $(TARGET_LDFLAGS) -Wl,--no-fatal-warnings endif +ifeq ($(BR2_TARGET_GENERIC_REMOUNT_ROOTFS_RW),y) +SYSTEMD_JOURNALD_PERMISSIONS = /var/log/journal d 2755 root systemd-journal - - - - - +endif + ifeq ($(BR2_PACKAGE_ACL),y) SYSTEMD_DEPENDENCIES += acl SYSTEMD_CONF_OPTS += -Dacl=true@@ -603,6 +608,7 @@ define SYSTEMD_PERMISSIONS /var/lib/private d 700 0 0 - - - - - /var/log/private d 700 0 0 - - - - - /var/cache/private d 700 0 0 - - - - - + $(SYSTEMD_JOURNALD_PERMISSIONS) $(SYSTEMD_LOGIND_PERMISSIONS) $(SYSTEMD_MACHINED_PERMISSIONS) $(SYSTEMD_HOMED_PERMISSIONS)
package/systemd/systemd.mk HOST_SYSTEMD_CONF_OPTS = \ · line 794
Quoted 7 lines
package/systemd/systemd.mk +1
-0@@ -782,6 +788,7 @@ HOST_SYSTEMD_CONF_OPTS = \ --libdir=lib \ --sysconfdir=/etc \ --localstatedir=/var \ + -Dcreate-log-dirs=false \ -Dmode=release \ -Dutmp=false \ -Dhibernate=false \
_______________________________________________
buildroot mailing list
[email protected]
https://lists.buildroot.org/mailman/listinfo/buildroot
PATCH 7/7 package/systemd: create the /boot directory
In reply to Norbert LangeThe gpt-auto-generator can generate the boot.mount unit, and report an error if this directory cant be created (read-only fs).
--- package/systemd/systemd.mk | 1 + 1 file changed, 1 insertion(+)
package/systemd/systemd.mk +1 -0
@@ -603,6 +603,7 @@ define SYSTEMD_INSTALL_IMAGES_CMDS endef define SYSTEMD_PERMISSIONS + /boot d 700 0 0 - - - - - /var/spool d 755 0 0 - - - - - /var/lib d 755 0 0 - - - - - /var/lib/private d 700 0 0 - - - - -
Signature · 6 lines
2.39.0 _______________________________________________ buildroot mailing list [email protected] https://lists.buildroot.org/mailman/listinfo/buildroot
Re: [Buildroot] [PATCH 7/7] package/systemd: create the /boot directory
Quoted 4 lines · Norbert Lange
The gpt-auto-generator can generate the boot.mount unit, and report an error if this directory cant be created (read-only fs).
Signed-off-by Norbert Lange <[email protected]>
Applied to master, thanks.
Regards,
Arnout
Quoted 9 lines
--- package/systemd/systemd.mk | 1 + 1 file changed, 1 insertion(+)package/systemd/systemd.mk +0
-0
package/systemd/systemd.mk define SYSTEMD_INSTALL_IMAGES_CMDS · line 609
Quoted 7 lines
package/systemd/systemd.mk +1
-0@@ -603,6 +603,7 @@ define SYSTEMD_INSTALL_IMAGES_CMDS endef define SYSTEMD_PERMISSIONS + /boot d 700 0 0 - - - - - /var/spool d 755 0 0 - - - - - /var/lib d 755 0 0 - - - - - /var/lib/private d 700 0 0 - - - - -
_______________________________________________
buildroot mailing list
[email protected]
https://lists.buildroot.org/mailman/listinfo/buildroot