From: Jean-Louis Dupond <jean-louis@dupond.be>
To: Konstantin Kostiuk <kkostiuk@redhat.com>
Cc: qemu-devel@nongnu.org, michael.roth@amd.com
Subject: Re: [PATCH] qga: skip bind mounts in fs list
Date: Mon, 28 Oct 2024 13:15:18 +0100 [thread overview]
Message-ID: <25723cba-8c60-401e-8b9a-731430f0970d@dupond.be> (raw)
In-Reply-To: <CAPMcbCrkzq3dGSJBkYZ_bU8qYTvXDokgJGkh563gg8OHGqB1wQ@mail.gmail.com>
[-- Attachment #1: Type: text/plain, Size: 5002 bytes --]
Hi Konstantin,
Thanks for your response.
What I observed was when running CloudLinux is that with a /tmp on a
loop device, is that the underlying fs was first freezed, and then the
/tmp was getting a freeze call.
But this was hanging, because it couldn't freeze as the underlying fs
was already freezed.
So the whole system became unresponsive expect if you send a unfreeze
via sysrq.
Build a qemu-ga with this patch included, and then it worked fine.
Because it skips the bind mounts and therefor made sure that the loop
device was first freezed and afterwards the underlying fs.
Which works fine :)
I did not see real kernel crashes, so that was not debugged/tested.
Thanks
Jean-Louis
On 28/10/2024 11:57, Konstantin Kostiuk wrote:
> Hi Jean-Louis,
>
> Thanks for your patch. I hope next week, I will test and review this
> patch.
>
> Just a question, did you have a chance to test that this patch fix
> kernel crash?
>
> Best Regards,
> Konstantin Kostiuk.
>
>
> On Fri, Oct 25, 2024 at 1:06 PM Jean-Louis Dupond
> <jean-louis@dupond.be> wrote:
>
> On 9/10/2024 10:34, Jean-Louis Dupond wrote:
> > On 2/10/2024 12:06, Jean-Louis Dupond wrote:
> >> The filesystem list in build_fs_mount_list should skip bind mounts.
> >> This because we end up in locking situations when doing
> fsFreeze. Like
> >> mentioned in [1] and [2].
> >>
> >> Next to that, the build_fs_mount_list call did a fallback via
> >> build_fs_mount_list_from_mtab if mountinfo did not exist.
> >> There it skipped bind mounts, but this is broken for newer OS.
> >> This as mounts does not return the path of the bind mount but the
> >> underlying dev/partition, so S_ISDIR will never return true in
> >> dev_major_minor call.
> >>
> >> This patch simply checks the existing devmajor:devminor tuple
> in the
> >> mounts, and if it already exists, this means we have the same
> devices
> >> mounted again, a bind mount. So skip this.
> >>
> >> Same approach is used in open-vm-tools [3].
> >>
> >> [1]: https://gitlab.com/qemu-project/qemu/-/issues/592
> >> [2]: https://gitlab.com/qemu-project/qemu/-/issues/520
> >> [3]:
> >>
> https://github.com/vmware/open-vm-tools/commit/d58847b497e212737007958c945af1df22a8ab58
> >>
> >> Signed-off-by: Jean-Louis Dupond <jean-louis@dupond.be>
> >> ---
> >> qga/commands-linux.c | 25 +++++++++++++++++++++++++
> >> 1 file changed, 25 insertions(+)
> >>
> >> diff --git a/qga/commands-linux.c b/qga/commands-linux.c
> >> index 51d5e3d927..426b040ab8 100644
> >> --- a/qga/commands-linux.c
> >> +++ b/qga/commands-linux.c
> >> @@ -59,6 +59,22 @@ static int dev_major_minor(const char *devpath,
> >> return -1;
> >> }
> >> +/*
> >> + * Check if we already have the devmajor:devminor in the mounts
> >> + * If thats the case return true.
> >> + */
> >> +static bool dev_exists(FsMountList *mounts, unsigned int
> devmajor,
> >> unsigned int devminor)
> >> +{
> >> + FsMount *mount;
> >> +
> >> + QTAILQ_FOREACH(mount, mounts, next) {
> >> + if (mount->devmajor == devmajor && mount->devminor ==
> >> devminor) {
> >> + return true;
> >> + }
> >> + }
> >> + return false;
> >> +}
> >> +
> >> static bool build_fs_mount_list_from_mtab(FsMountList *mounts,
> >> Error **errp)
> >> {
> >> struct mntent *ment;
> >> @@ -89,6 +105,10 @@ static bool
> >> build_fs_mount_list_from_mtab(FsMountList *mounts, Error **errp)
> >> /* Skip bind mounts */
> >> continue;
> >> }
> >> + if (dev_exists(mounts, devmajor, devminor)) {
> >> + /* Skip already existing devices (bind mounts) */
> >> + continue;
> >> + }
> >> mount = g_new0(FsMount, 1);
> >> mount->dirname = g_strdup(ment->mnt_dir);
> >> @@ -172,6 +192,11 @@ bool build_fs_mount_list(FsMountList *mounts,
> >> Error **errp)
> >> }
> >> }
> >> + if (dev_exists(mounts, devmajor, devminor)) {
> >> + /* Skip already existing devices (bind mounts) */
> >> + continue;
> >> + }
> >> +
> >> mount = g_new0(FsMount, 1);
> >> mount->dirname = g_strdup(line + dir_s);
> >> mount->devtype = g_strdup(dash + type_s);
> >
> >
> > Ping + add kkostiuk@redhat.com as I missed him in the initial mail.
> >
>
> Any chance on a review or getting it merged?
> Think it's a good (of course ;)) improvement.
>
[-- Attachment #2: Type: text/html, Size: 8818 bytes --]
next prev parent reply other threads:[~2024-10-28 12:16 UTC|newest]
Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top
2024-10-02 10:06 [PATCH] qga: skip bind mounts in fs list Jean-Louis Dupond
2024-10-09 8:34 ` Jean-Louis Dupond
2024-10-25 10:06 ` Jean-Louis Dupond
2024-10-28 10:57 ` Konstantin Kostiuk
2024-10-28 12:15 ` Jean-Louis Dupond [this message]
2024-11-26 10:40 ` Konstantin 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=25723cba-8c60-401e-8b9a-731430f0970d@dupond.be \
--to=jean-louis@dupond.be \
--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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).