From: Roman Bogorodskiy <bogorodskiy@gmail.com>
To: Kostiantyn Kostiuk <kkostiuk@redhat.com>
Cc: qemu-devel@nongnu.org, Brad Smith <brad@comstyle.com>,
Michael Roth <michael.roth@amd.com>
Subject: Re: [PATCH v2] qga: do not run hwclock(8) on non-Linux POSIX OSes
Date: Wed, 28 Oct 2026 11:56:33 +0100 [thread overview]
Message-ID: <auHU4RuV_XTSV0Vn@tulp> (raw)
In-Reply-To: <CAPMcbCpMtRgkWkEPj5xns7xMVCzDOjBcwp8m944q+tTtEPj1Tg@mail.gmail.com>
Kostiantyn Kostiuk wrote:
> On Thu, Oct 15, 2026 at 5:24 PM Roman Bogorodskiy <bogorodskiy@gmail.com>
> wrote:
>
> > Currently, qmp_guest_set_time() sets system time using
> > the settimeofday(2) system call and then runs the hwclock(8)
> > util to set hardware time.
> >
> > hwclock(8) is Linux only, so current implementation
> > always fails while trying to run it on non-Linux POSIX OSes,
> > such as BSDs (FreeBSD, OpenBSD, NetBSD) and illumos.
> >
> > Update the code not to run it on non-Linux. The settimeofday(2)
> > call sets both system and hardware time on these systems.
> > For example, on FreeBSD settimeofday() calls settime()
> > which calls resettodr() [1] which updates RTC clock [2].
> > It is similar on OpenBSD and NetBSD, though implementations
> > differ a bit.
> >
> > Additional detail: on FreeBSD RTC clock setting can be
> > disabled with the `machdep.disable_rtc_set` sysctl.
> > As we cannot do anything about it, do not handle that
> > case separately.
> >
> > Non-Linux POSIX OSes do not provide a common utility
> > equivalent to hwclock -s, so report an unsupported operation
> > when no time is supplied.
> >
> > 1: https://cgit.freebsd.org/src/tree/sys/kern/kern_time.c
> > 2: https://cgit.freebsd.org/src/tree/sys/kern/subr_rtc.c
> >
> > Signed-off-by: Roman Bogorodskiy <bogorodskiy@gmail.com>
> > ---
> > Changes since v1:
> >
> > Replace "#if defined(CONFIG_BSD)" condition with
> > "#if !defined(CONFIG_LINUX)" as hwclock(8) is Linux only.
> >
> > qga/commands-posix.c | 14 ++++++++++++++
> > 1 file changed, 14 insertions(+)
> >
> > diff --git a/qga/commands-posix.c b/qga/commands-posix.c
> > index 837e285b6d..ffd1744b74 100644
> > --- a/qga/commands-posix.c
> > +++ b/qga/commands-posix.c
> > @@ -322,6 +322,20 @@ void qmp_guest_set_time(bool has_time, int64_t
> > time_ns, Error **errp)
> > }
> > }
> >
> > +#if !defined(CONFIG_LINUX)
> > + /*
> > + * BSDs (FreeBSD, OpenBSD, NetBSD) and illumos do not have the
> > + * hwclock(8) tool. These systems update hardware clock when
> > + * settimeofday() sets system time, so no additional actions
> > + * are needed. Setting system time from RTC is not supported.
> > + */
> > + if (!has_time) {
> > + error_setg(errp, "Setting system time from RTC is not supported");
> > + }
> > +
> > + return;
> > +#endif
> > +
> >
>
> As only Linux (not POSIX) and Windows support this function
> lets move qmp_guest_set_time to commands-linux and update QAPI with
> 'if': { 'any': ['CONFIG_LINUX', 'CONFIG_WIN32'] }
Hi Kostiantyn,
Sorry if my commit message does not explain the situation clearly enough.
BSDs and illumos do support "guest-set-time" with "time" specified.
On these systems, calling settimeofday() is enough for that, it sets
both system time and hardware clock, so lack of the hwclock(8) does not
prevent it from working as intended.
Only "guest-set-time" without the "time" argument (i.e. setting system time
from the hardware clock) is not supported indeed, which we indicate with
an error message.
Therefore I think keeping qmp_guest_set_time() in commands-posix.c and
keeping the proposed conditional behavior is appropriate.
>
> Best Regards,
> Kostiantyn Kostiuk.
next prev parent reply other threads:[~2026-10-28 10:57 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-10-15 14:23 [PATCH v2] qga: do not run hwclock(8) on non-Linux POSIX OSes Roman Bogorodskiy
2026-10-28 9:03 ` Kostiantyn Kostiuk
2026-10-28 10:56 ` Roman Bogorodskiy [this message]
2026-10-28 11:51 ` Kostiantyn Kostiuk
2026-10-28 11:51 ` Kostiantyn Kostiuk
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=auHU4RuV_XTSV0Vn@tulp \
--to=bogorodskiy@gmail.com \
--cc=brad@comstyle.com \
--cc=kkostiuk@redhat.com \
--cc=michael.roth@amd.com \
--cc=qemu-devel@nongnu.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
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.