DPDK patches and discussions
 help / color / mirror / Atom feed
From: Anatoly Burakov <anatoly.burakov@intel.com>
To: dev@dpdk.org, Robin Jarry <rjarry@redhat.com>
Subject: [PATCH v1 1/1] usertools/devbind: allow changing UID/GID for VFIO
Date: Mon,  2 Sep 2024 14:57:32 +0100	[thread overview]
Message-ID: <4cd0282dabfa59e715028ecf255468529655b487.1725285449.git.anatoly.burakov@intel.com> (raw)

Currently, when binding a device to VFIO, the UID/GID for the device will
always stay as system default (`root`). Yet, when running DPDK as non-root
user, one has to change the UID/GID of the device to match the user's
UID/GID to use the device.

This patch adds an option to `dpdk-devbind.py` to change the UID/GID of
the device when binding it to VFIO.

Signed-off-by: Anatoly Burakov <anatoly.burakov@intel.com>
---
 usertools/dpdk-devbind.py | 41 +++++++++++++++++++++++++++++++++++++--
 1 file changed, 39 insertions(+), 2 deletions(-)

diff --git a/usertools/dpdk-devbind.py b/usertools/dpdk-devbind.py
index 078e8c387b..37e2b9972d 100755
--- a/usertools/dpdk-devbind.py
+++ b/usertools/dpdk-devbind.py
@@ -8,6 +8,8 @@
 import subprocess
 import argparse
 import platform
+import grp
+import pwd
 
 from glob import glob
 from os.path import exists, basename
@@ -107,6 +109,8 @@
 b_flag = None
 status_flag = False
 force_flag = False
+vfio_uid = ""
+vfio_gid = ""
 args = []
 
 
@@ -462,6 +466,22 @@ def bind_one(dev_id, driver, force):
                      % (dev_id, filename, err))
 
 
+def own_one(dev_id, uid, gid):
+    """Set the IOMMU group ownership for a device"""
+    # find IOMMU group for a particular device
+    iommu_grp_base_path = os.path.join("/sys/bus/pci/devices", dev_id, "iommu_group")
+    try:
+        iommu_grp = os.path.basename(os.readlink(iommu_grp_base_path))
+        # we found IOMMU group, now find the device
+        dev_path = os.path.join("/dev/vfio", iommu_grp)
+        # set the ownership
+        _uid = pwd.getpwnam(uid).pw_uid if uid else -1
+        _gid = grp.getgrnam(gid).gr_gid if gid else -1
+        os.chown(dev_path, _uid, _gid)
+    except OSError as err:
+        sys.exit(f"Error: failed to read IOMMU group for {dev_id}: {err}")
+
+
 def unbind_all(dev_list, force=False):
     """Unbind method, takes a list of device locations"""
 
@@ -482,7 +502,7 @@ def unbind_all(dev_list, force=False):
         unbind_one(d, force)
 
 
-def bind_all(dev_list, driver, force=False):
+def bind_all(dev_list, driver, uid, gid, force=False):
     """Bind method, takes a list of device locations"""
     global devices
 
@@ -510,6 +530,9 @@ def bind_all(dev_list, driver, force=False):
 
     for d in dev_list:
         bind_one(d, driver, force)
+        # if we're binding to vfio-pci, set the IOMMU user/group ownership if one was specified
+        if driver == "vfio-pci" and (uid or gid):
+            own_one(d, uid, gid)
 
     # For kernels < 3.15 when binding devices to a generic driver
     # (i.e. one that doesn't have a PCI ID table) using new_id, some devices
@@ -662,6 +685,8 @@ def parse_args():
     global status_dev
     global force_flag
     global args
+    global vfio_uid
+    global vfio_gid
 
     parser = argparse.ArgumentParser(
         description='Utility to bind and unbind devices from Linux kernel',
@@ -707,6 +732,12 @@ def parse_args():
         '--unbind',
         action='store_true',
         help="Unbind a device (equivalent to \"-b none\")")
+    parser.add_argument(
+        "-U", "--uid", help="For VFIO, specify the UID to set IOMMU group ownership"
+    )
+    parser.add_argument(
+        "-G", "--gid", help="For VFIO, specify the GID to set IOMMU group ownership"
+    )
     parser.add_argument(
         '--force',
         action='store_true',
@@ -737,6 +768,10 @@ def parse_args():
         b_flag = opt.bind
     elif opt.unbind:
         b_flag = "none"
+    if opt.uid:
+        vfio_uid = opt.uid
+    if opt.gid:
+        vfio_gid = opt.gid
     args = opt.devices
 
     if not b_flag and not status_flag:
@@ -764,11 +799,13 @@ def do_arg_actions():
     global status_flag
     global force_flag
     global args
+    global vfio_uid
+    global vfio_gid
 
     if b_flag in ["none", "None"]:
         unbind_all(args, force_flag)
     elif b_flag is not None:
-        bind_all(args, b_flag, force_flag)
+        bind_all(args, b_flag, vfio_uid, vfio_gid, force_flag)
     if status_flag:
         if b_flag is not None:
             clear_data()
-- 
2.43.5


             reply	other threads:[~2024-09-02 13:57 UTC|newest]

Thread overview: 2+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-09-02 13:57 Anatoly Burakov [this message]
2024-09-03  9:11 ` Burakov, Anatoly

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=4cd0282dabfa59e715028ecf255468529655b487.1725285449.git.anatoly.burakov@intel.com \
    --to=anatoly.burakov@intel.com \
    --cc=dev@dpdk.org \
    --cc=rjarry@redhat.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;
as well as URLs for NNTP newsgroup(s).