utils_disk: check the mount point by findmnt cmd

[leidwang]

Currently, if we run the test in image mode, the mount path check will not match because the directory is redirected,
so use findmnt to solve this problem.

ID: 3444

[leidwang]

Hi @hellohellenmao Could you please check if this MR can fix the error that you encountered, thanks.

[hellohellenmao]

Hi @hellohellenmao Could you please check if this MR can fix the error that you encountered, thanks.

I do not think so, for my case is invoking the umount https://github.com/avocado-framework/avocado-vt/blob/9f599feb7a75c756727e5f874f67ecb0e38bcc00/virttest/utils_disk.py#L158C36-L158C39 here to do the umounting, so could you please update the code here either?
Thanks

[leidwang]

Hi @hellohellenmao Could you please check if this MR can fix the error that you encountered, thanks.

I do not think so, for my case is invoking the umount https://github.com/avocado-framework/avocado-vt/blob/9f599feb7a75c756727e5f874f67ecb0e38bcc00/virttest/utils_disk.py#L158C36-L158C39 here to do the umounting, so could you please update the code here either? Thanks

Updated the code, could you please add your logs to the issue, thanks.

[hellohellenmao]

Still hit the issue with the patch , does it work from your side @leidwang

(1/1) Host_RHEL.m9.u6.ovmf.qcow2.virtio_scsi.up.virtio_net.bootc.Guest.RHEL.9.6.0.x86_64.io-github-autotest-qemu.virtio_fs_sandbox.none.q35: ERROR: Failed to remove directory rm -rf /mnt/myfs from remote machine: Shell command failed: 'rm -rf /mnt/myfs' (status: 1, output: "rm: cannot remove '/mnt/myfs': Device or resource busy\n")

[leidwang]

Hi @PaulYuuu Could you please review this MR? Thanks.

[leidwang]

Hi @PaulYuuu @hellohellenmao Would you please help review this MR, thanks.

[YvanY0]

Almost LGTM, but you should rebase the branch to fix the CI issue.

[leidwang]

Hi @PaulYuuu Could you please help involve others to review this PR?Thanks.

[YvanY0]

Hi @PaulYuuu Could you please help involve others to review this PR?Thanks.

As this is related to block feature, I think @qingwangrh @YongxueHong should be the right persons to review. And, not only us use this, we should also include libvirt people to review if needed.

[qingwangrh]

hi, @leidwang , if we want fewer changes, how about changing "cat /proc/mount" to "mount"
or " mount|grep %s" % mount_str to get precise result. that is one line changed. Could you please help to try it?

[leidwang]

hi, @leidwang , if we want fewer changes, how about changing "cat /proc/mount" to "mount" or " mount|grep %s" % mount_str to get precise result. that is one line changed. Could you please help to try it?

Hi @qingwangrh findmnt provides a more structured output that can display mount information in a tree format, table format, or JSON format(-J). Its output is easier to read and parse. I would more prefer to use findmnt to check mount info and update currently code accordingly, feel free to add any comments about that? @PaulYuuu @YongxueHong

[YvanY0]

hi, @leidwang , if we want fewer changes, how about changing "cat /proc/mount" to "mount" or " mount|grep %s" % mount_str to get precise result. that is one line changed. Could you please help to try it?

Hi @qingwangrh findmnt provides a more structured output that can display mount information in a tree format, table format, or JSON format(-J). Its output is easier to read and parse. I would more prefer to use findmnt to check mount info and update currently code accordingly, feel free to add any comments about that? @PaulYuuu @YongxueHong

Agree. And mount | grep can only check the target path, and cannot grep the soft link.

[YongxueHong]

Hi @leidwang
Where does it redirect to? Could you give more? Thanks.

CC @nickzhq
Please help check if the vt_utils.filesystem module needs to be updated.
Thanks.

[qingwangrh]

It makes sense to use findmnt if we want to support softlink.

mount | grep $(readlink -f xxx) also may support softlink

The findmnt may get better readability output,
Both methods LGTM

[leidwang]

Hi @leidwang Where does it redirect to? Could you give more? Thanks.

Sure, for example, if we mount nfs to /home/mount_test, in bootc host, it will shown as /var/home/mount_test instead of /home/mount_test.

# mkdir mount_test
# mount ***:/kvmqeauto/iso /home/mount_test/
# cat /proc/mounts
......
***:/kvmqeauto/iso /var/home/mount_test nfs ro,relatime,vers=3,rsize=65536,wsize=65536,namlen=255,hard,proto=tcp,timeo=600,retrans=2,sec=sys,mountaddr=***,mountvers=3,mountport=635,mountproto=udp,local_lock=none,addr=*** 0 0

CC @nickzhq Please help check if the vt_utils.filesystem module needs to be updated. Thanks.

[YongxueHong]

Hi @leidwang Where does it redirect to? Could you give more? Thanks.

Sure, for example, if we mount nfs to /home/mount_test, in bootc host, it will shown as /var/home/mount_test instead of /home/mount_test.

# mkdir mount_test
# mount ***:/kvmqeauto/iso /home/mount_test/
# cat /proc/mounts
......
***:/kvmqeauto/iso /var/home/mount_test nfs ro,relatime,vers=3,rsize=65536,wsize=65536,namlen=255,hard,proto=tcp,timeo=600,retrans=2,sec=sys,mountaddr=***,mountvers=3,mountport=635,mountproto=udp,local_lock=none,addr=*** 0 0

Hi @leidwang
Thanks for illustrating. I am just curious why the mount point will be redirected to another?
Did it just happen with the image mode?
Would you know more?
Thanks.

CC @nickzhq Please help check if the vt_utils.filesystem module needs to be updated. Thanks.

[leidwang]

Hi @leidwang Where does it redirect to? Could you give more? Thanks.

Sure, for example, if we mount nfs to /home/mount_test, in bootc host, it will shown as /var/home/mount_test instead of /home/mount_test.

# mkdir mount_test
# mount ***:/kvmqeauto/iso /home/mount_test/
# cat /proc/mounts
......
***:/kvmqeauto/iso /var/home/mount_test nfs ro,relatime,vers=3,rsize=65536,wsize=65536,namlen=255,hard,proto=tcp,timeo=600,retrans=2,sec=sys,mountaddr=***,mountvers=3,mountport=635,mountproto=udp,local_lock=none,addr=*** 0 0

Hi @leidwang Thanks for illustrating. I am just curious why the mount point will be redirected to another? Did it just happen with the image mode? Would you know more? Thanks.

Yes, it only happens in image mode OS, it should be related to image mode os file system is read-only, only /var and /etc are writable.

CC @nickzhq Please help check if the vt_utils.filesystem module needs to be updated. Thanks.

[YvanY0]

I don't think it's only related to image mode, no one can promise users will not create a softlink.

For instance, avocado will point the iso path to /root/avocado/data/avocado-vt/isos/, our internal usage will link it /home/kvm_autotest_root/iso/

[leidwang]

Hi @PaulYuuu @YongxueHong I have updated the code according to our offline discussion, would you please check it again?Thanks.

[YongxueHong]

Hi @leidwang
The code is good for me.
Could you help run some test cases in non-image mode to verify if it will introduce a regression issue?
Thanks in advance.

@hellohellenmao @qingwangrh
Could you help review it? Since there are some updates with it.
Thanks.

[YongxueHong]

Looks like mount_result = None is useless.

[qingwangrh]

change the commit message "check the moint point by findmnt cmd" -> check the mount point by findmnt cmd
Other LGTM

[leidwang]

Hi @leidwang The code is good for me. Could you help run some test cases in non-image mode to verify if it will introduce a regression issue? Thanks in advance.

@hellohellenmao @qingwangrh Could you help review it? Since there are some updates with it. Thanks.

Run the installation test case as it will call this function.

JOB LOG : /root/avocado/job-results/job-2025-07-03T10.37-7bff2d1/job.log
(1/2) Host_RHEL.m10.u.ovmf.qcow2.virtio_scsi.up.virtio_net.Guest.Win2022.x86_64.io-github-autotest-qemu.unattended_install.cdrom.extra_cdrom_ks.default_install.aio_threads.q35: STARTED
(1/2) Host_RHEL.m10.u.ovmf.qcow2.virtio_scsi.up.virtio_net.Guest.Win2022.x86_64.io-github-autotest-qemu.unattended_install.cdrom.extra_cdrom_ks.default_install.aio_threads.q35: PASS (631.99 s)

[leidwang]

Tested this PR on image mode and package mode, all works well.

[YongxueHong]

LGTM.

[leidwang]

Hi @YongxueHong @PaulYuuu If there is no more comments for this MR, would you please help merge it?Thanks.