[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <20241204064754.GL1245331@unreal>
Date: Wed, 4 Dec 2024 08:47:54 +0200
From: Leon Romanovsky <leon@...nel.org>
To: Bjorn Helgaas <helgaas@...nel.org>
Cc: Stephen Hemminger <stephen@...workplumber.org>,
Krzysztof Wilczyński <kw@...ux.com>,
linux-pci@...r.kernel.org, Ariel Almog <ariela@...dia.com>,
Aditya Prabhune <aprabhune@...dia.com>,
Hannes Reinecke <hare@...e.de>,
Heiner Kallweit <hkallweit1@...il.com>,
Arun Easi <aeasi@...vell.com>, Jonathan Chocron <jonnyc@...zon.com>,
Bert Kenward <bkenward@...arflare.com>,
Matt Carlson <mcarlson@...adcom.com>,
Kai-Heng Feng <kai.heng.feng@...onical.com>,
Jean Delvare <jdelvare@...e.de>,
Alex Williamson <alex.williamson@...hat.com>,
linux-kernel@...r.kernel.org, netdev@...r.kernel.org,
Jakub Kicinski <kuba@...nel.org>,
Thomas Weißschuh <linux@...ssschuh.net>,
Kees Cook <kees@...nel.org>,
"Gustavo A. R. Silva" <gustavoars@...nel.org>,
linux-hardening@...r.kernel.org
Subject: Re: [PATCH v3] PCI/sysfs: Change read permissions for VPD attributes
On Tue, Dec 03, 2024 at 02:36:25PM -0600, Bjorn Helgaas wrote:
> [+cc Linux hardening folks for any security/reliability concerns]
>
> On Tue, Dec 03, 2024 at 07:40:27PM +0200, Leon Romanovsky wrote:
> > On Tue, Dec 03, 2024 at 09:24:56AM -0800, Stephen Hemminger wrote:
> > > On Tue, 3 Dec 2024 14:15:28 +0200
> > > Leon Romanovsky <leon@...nel.org> wrote:
> > >
> > > > The Vital Product Data (VPD) attribute is not readable by regular
> > > > user without root permissions. Such restriction is not needed at
> > > > all for Mellanox devices, as data presented in that VPD is not
> > > > sensitive and access to the HW is safe and well tested.
> > > >
> > > > This change changes the permissions of the VPD attribute to be accessible
> > > > for read by all users for Mellanox devices, while write continue to be
> > > > restricted to root only.
> > > >
> > > > The main use case is to remove need to have root/setuid permissions
> > > > while using monitoring library [1].
> > > >
> > > > [leonro@vm ~]$ lspci |grep nox
> > > > 00:09.0 Ethernet controller: Mellanox Technologies MT2910 Family [ConnectX-7]
> > > >
> > > > Before:
> > > > [leonro@vm ~]$ ls -al /sys/bus/pci/devices/0000:00:09.0/vpd
> > > > -rw------- 1 root root 0 Nov 13 12:30 /sys/bus/pci/devices/0000:00:09.0/vpd
> > > > After:
> > > > [leonro@vm ~]$ ls -al /sys/bus/pci/devices/0000:00:09.0/vpd
> > > > -rw-r--r-- 1 root root 0 Nov 13 12:30 /sys/bus/pci/devices/0000:00:09.0/vpd
> > > >
> > > > [1] https://developer.nvidia.com/management-library-nvml
> > > > Signed-off-by: Leon Romanovsky <leonro@...dia.com>
> > > > ---
> > > > Changelog:
> > > > v3:
> > > > * Used | to change file attributes
> > > > * Remove WARN_ON
> > > > v2: https://lore.kernel.org/all/61a0fa74461c15edfae76222522fa445c28bec34.1731502431.git.leon@kernel.org
> > > > * Another implementation to make sure that user is presented with
> > > > correct permissions without need for driver intervention.
> > > > v1: https://lore.kernel.org/all/cover.1731005223.git.leonro@nvidia.com
> > > > * Changed implementation from open-read-to-everyone to be opt-in
> > > > * Removed stable and Fixes tags, as it seems like feature now.
> > > > v0:
> > > > https://lore.kernel.org/all/65791906154e3e5ea12ea49127cf7c707325ca56.1730102428.git.leonro@nvidia.com/
> > > > ---
> > > > drivers/pci/vpd.c | 7 +++++++
> > > > 1 file changed, 7 insertions(+)
> > > >
> > > > diff --git a/drivers/pci/vpd.c b/drivers/pci/vpd.c
> > > > index a469bcbc0da7..a7aa54203321 100644
> > > > --- a/drivers/pci/vpd.c
> > > > +++ b/drivers/pci/vpd.c
> > > > @@ -332,6 +332,13 @@ static umode_t vpd_attr_is_visible(struct kobject *kobj,
> > > > if (!pdev->vpd.cap)
> > > > return 0;
> > > >
> > > > + /*
> > > > + * Mellanox devices have implementation that allows VPD read by
> > > > + * unprivileged users, so just add needed bits to allow read.
> > > > + */
> > > > + if (unlikely(pdev->vendor == PCI_VENDOR_ID_MELLANOX))
> > > > + return a->attr.mode | 0044;
> > > > +
> > > > return a->attr.mode;
> > > > }
> > >
> > > Could this be with other vendor specific quirks instead?
> >
> > In previous versions, I asked Bjorn about using quirks and the answer
> > was that quirks are mainly to fix HW defects fixes and this change doesn't
> > belong to that category.
> >
> > https://lore.kernel.org/linux-pci/20241111214804.GA1820183@bhelgaas/
>
> That previous proposal was driver-based, so VPD would only be readable
> by unprivileged users after mlx5 was loaded. VPD would be readable at
> any time with either a quirk or the current patch. The quirk would
> require a new bit in pci_dev but has the advantage of getting the
> Mellanox grunge out of the generic code.
>
> My biggest concerns are that this exposes VPD data of unknown
> sensitivity and exercises the sometimes-problematic device VPD
> protocol for very little user benefit. IIUC, the monitoring library
> only wants this to identify the specific device variant in the user
> interface; it doesn't need it to actually *use* the device.
>
> We think these concerns are minimal for these devices (and I guess for
> *all* present and future Mellanox devices), but I don't think it's a
> great precedent.
Yes, and we can always move this "if ..." to quirks once second device
will appear.
Thanks
>
> Bjorn
Powered by blists - more mailing lists