Hi Andrew,

On Fri, Dec 5, 2025 at 5:05 PM Andrew Cooper <[email protected]> wrote:
>
> On 05/12/2025 1:13 pm, Mykola Kvach wrote:
> > From: Mykola Kvach <[email protected]>
> >
> > Using PKG_CHECK_VAR() to obtain SYSTEMD_SLEEP_DIR reintroduced a build-time
> > dependency on systemd-devel. This breaks --enable-systemd builds that only
> > need the initscripts.
> >
> > Use the same defaulting scheme as other systemd paths: if SYSTEMD_SLEEP_DIR
> > isn't provided, set it to "$(prefix)/lib/systemd/system-sleep/".
> >
> > Fixes: e54a6cd6a1f3 ("systemd: Add hooks to stop/start xen-watchdog on 
> > suspend/resume")
> > Reported-by: Marek Marczykowski-Górecki <[email protected]>
> > Signed-off-by: Mykola Kvach <[email protected]>
> > ---
> > Build in which the regression was detected:
> > https://gitlab.com/xen-project/people/marmarek/xen/-/jobs/10959608099
> > ---
> >  m4/systemd.m4 | 2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/m4/systemd.m4 b/m4/systemd.m4
> > index c47a25ef93..ee684d3391 100644
> > --- a/m4/systemd.m4
> > +++ b/m4/systemd.m4
> > @@ -77,7 +77,7 @@ AC_DEFUN([AX_CHECK_SYSTEMD_LIBS], [
> >       ], [])
> >
> >       AS_IF([test "x$SYSTEMD_SLEEP_DIR" = x], [
> > -         PKG_CHECK_VAR([SYSTEMD_SLEEP_DIR], [systemd], [systemdsleepdir])
> > +         SYSTEMD_SLEEP_DIR="\$(prefix)/lib/systemd/system-sleep/"
> >       ], [])
> >
> >       AS_IF([test "x$SYSTEMD_SLEEP_DIR" = x], [
>
> Actually on second thoughts...  The build failure was hitting this
> second check, not the PKG_CHECK_VAR() itself.
>
> If systemd-devel is installed, we should use what it provides.  So,
> can't we just do this:
>
> diff --git a/m4/systemd.m4 b/m4/systemd.m4
> index c47a25ef9393..5fb5a7089ee3 100644
> --- a/m4/systemd.m4
> +++ b/m4/systemd.m4
> @@ -81,7 +81,7 @@ AC_DEFUN([AX_CHECK_SYSTEMD_LIBS], [
>         ], [])
>
>         AS_IF([test "x$SYSTEMD_SLEEP_DIR" = x], [
> -           AC_MSG_ERROR([SYSTEMD_SLEEP_DIR is unset])
> +           SYSTEMD_SLEEP_DIR="\$(prefix)/lib/systemd/system-sleep/"
>         ], [])
>  ])
>
>
> ?

Agreed. Using the systemd-provided value when available and falling
back to the default otherwise makes sense. I’ve sent v2 of this series
with your proposed change.

>
> ~Andrew


Best regards,
Mykola

Reply via email to