From: "Jakob Klocker" <j.klocker@proxmox.com>
To: "David Riley" <d.riley@proxmox.com>, <pve-devel@lists.proxmox.com>
Subject: Re: [PATCH pve-storage 2/2] fix #6050: cifs: normalize share and subdir paths for mount and status
Date: Tue, 18 Aug 2026 12:23:33 +0200 [thread overview]
Message-ID: <DKRZV3N7Q1RV.15YJQKI4EALYG@proxmox.com> (raw)
In-Reply-To: <20260807093657.23649-3-d.riley@proxmox.com>
Comments inline. I'd also suggest mentioning why you didn't reuse the
`canonpath` function you used in the previous patch here.
On Fri Aug 7, 2026 at 11:36 AM CEST, David Riley wrote:
> Manual edits to the storage configuration bypass strict API
> validation. If a user adds extra or trailing slashes to the share or
> subdirectory paths, it causes two distinct issues for CIFS storages:
>
> * The Linux VFS CIFS module expects strict UNC [0] formatting and
> will reject the mount attempt with a 'Malformed UNC in devname'
> error.
> * The internal path used by the status daemon fails to match the
> normalized path reported in /proc/mounts, causing working storages
> to continuously report as 'inactive'.
>
> Sanitize the share and subdirectory strings by stripping leading and
> trailing slashes before executing mounts, verifying mount status, and
> testing the SMB connection. This ensures working storages mount
> successfully and are correctly recognized as active.
>
> [0] https://docs.kernel.org/admin-guide/cifs/usage.html#cifs-vfs-mount-options
>
> Link: https://bugzilla.proxmox.com/show_bug.cgi?id=6050
> Signed-off-by: David Riley <d.riley@proxmox.com>
> ---
> src/PVE/Storage/CIFSPlugin.pm | 32 ++++++++++++++++++++++++++++----
> 1 file changed, 28 insertions(+), 4 deletions(-)
>
> diff --git a/src/PVE/Storage/CIFSPlugin.pm b/src/PVE/Storage/CIFSPlugin.pm
> index 54f0f4e..78ed88c 100644
> --- a/src/PVE/Storage/CIFSPlugin.pm
> +++ b/src/PVE/Storage/CIFSPlugin.pm
> @@ -2,15 +2,20 @@ package PVE::Storage::CIFSPlugin;
>
> use strict;
> use warnings;
> +
> +use File::Path;
> use Net::IP;
> +
> use PVE::Tools qw(run_command);
> use PVE::ProcFSTools;
> -use File::Path;
> use PVE::Storage::Plugin;
> use PVE::JSONSchema qw(get_standard_option);
>
> use base qw(PVE::Storage::Plugin);
>
> +my $RM_LEADING_SLASHES = qr{^/+};
> +my $RM_TRAILING_SLASHES = qr{/+$};
Since you're removing the slashes at three different locations in this
file, I'd prefere having a function which does that. Would make the code
easier to read. You could also think about moving that function into the
common module, since I can imagine this is something we could use from
time to time. What do you think?
> +
> # CIFS helper functions
>
> sub cifs_is_mounted : prototype($$) {
> @@ -19,8 +24,16 @@ sub cifs_is_mounted : prototype($$) {
> my ($mountpoint, $server, $share) = $scfg->@{ 'path', 'server', 'share' };
> my $subdir = $scfg->{subdir} // '';
>
> + for my $path ($share, $subdir) {
Is stripping share actually needed? The documented workflow
(pvesm scan cifs <address>) never returns slashes, and since share
is a single name rather than a path, slashes there look more like
invalid input than a formatting variant.
> + $path =~ s/$RM_LEADING_SLASHES// if $path;
> + $path =~ s/$RM_TRAILING_SLASHES// if $path;
> + }
> +
> $server = "[$server]" if Net::IP::ip_is_ipv6($server);
> - my $source = "//${server}/$share$subdir";
> + my $source = "//${server}";
> + $source .= "/$share";
> + $source .= "/$subdir" if $subdir;
> +
> $mountdata = PVE::ProcFSTools::parse_proc_mounts() if !$mountdata;
>
> return $mountpoint if grep {
> @@ -83,8 +96,15 @@ sub cifs_mount : prototype($$$$$) {
> my ($mountpoint, $server, $share, $options) = $scfg->@{ 'path', 'server', 'share', 'options' };
> my $subdir = $scfg->{subdir} // '';
>
> + for my $path ($share, $subdir) {
> + $path =~ s/$RM_LEADING_SLASHES// if $path;
> + $path =~ s/$RM_TRAILING_SLASHES// if $path;
> + }
> +
> $server = "[$server]" if Net::IP::ip_is_ipv6($server);
> - my $source = "//${server}/$share$subdir";
> + my $source = "//${server}";
> + $source .= "/$share";
> + $source .= "/$subdir" if $subdir;
>
> my $cmd = ['/bin/mount', '-t', 'cifs', $source, $mountpoint, '-o', 'soft'];
>
> @@ -285,7 +305,11 @@ sub deactivate_storage {
> sub check_connection {
> my ($class, $storeid, $scfg) = @_;
>
> - my $servicename = '//' . $scfg->{server} . '/' . $scfg->{share};
> + my $share = $scfg->{share};
> + $share =~ s/$RM_LEADING_SLASHES// if $share;
> + $share =~ s/$RM_TRAILING_SLASHES// if $share;
> +
> + my $servicename = '//' . $scfg->{server} . '/' . $share;
>
> my $cmd = ['/usr/bin/smbclient', $servicename, '-d', '0'];
>
next prev parent reply other threads:[~2026-08-18 10:23 UTC|newest]
Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-08-07 9:36 [PATCH storage 0/2] fix #6050: normalize paths in mount status checks David Riley
2026-08-07 9:36 ` [PATCH pve-storage 1/2] fix #6050: nfs: strip trailing slashes from mount status check David Riley
2026-08-18 9:35 ` Jakob Klocker
2026-08-07 9:36 ` [PATCH pve-storage 2/2] fix #6050: cifs: normalize share and subdir paths for mount and status David Riley
2026-08-18 10:23 ` Jakob Klocker [this message]
2026-08-18 12:33 ` David Riley
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=DKRZV3N7Q1RV.15YJQKI4EALYG@proxmox.com \
--to=j.klocker@proxmox.com \
--cc=d.riley@proxmox.com \
--cc=pve-devel@lists.proxmox.com \
/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