Skip to contentSkip to conversation

mptcp: allow priviledged operations from user namespaces

PATCH Superseded Thomas Haller 5 participants 13 messages Aug 5, 2022 – Aug 29, 2022

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 mptcp: allow priviledged operations from user namespaces

GENL_ADMIN_PERM checks that the user has CAP_NET_ADMIN in the initial namespace by calling netlink_capable(). Instead, use GENL_UNS_ADMIN_PERM which uses netlink_ns_capable(). This checks that the caller has CAP_NET_ADMIN in the current user namespace.

See also commit 4a92602aa1cd ('openvswitch: allow management from inside user namespaces') which introduced this mechanism. See also commit 5617c6cd6f84 ('nl80211: Allow privileged operations from user namespaces'), which introduced this for nl80211.

---
 net/mptcp/pm_netlink.c | 18 +++++++++---------
 1 file changed, 9 insertions(+), 9 deletions(-)
net/mptcp/pm_netlink.c +9 -9
index 291b5da42fdb..2c145cdc7bdc 100644
@@ -2218,17 +2218,17 @@ static const struct genl_small_ops mptcp_pm_ops[] = {
 	{
 		.cmd    = MPTCP_PM_CMD_ADD_ADDR,
 		.doit   = mptcp_nl_cmd_add_addr,
-		.flags  = GENL_ADMIN_PERM,
+		.flags  = GENL_UNS_ADMIN_PERM,
 	},
 	{
 		.cmd    = MPTCP_PM_CMD_DEL_ADDR,
 		.doit   = mptcp_nl_cmd_del_addr,
-		.flags  = GENL_ADMIN_PERM,
+		.flags  = GENL_UNS_ADMIN_PERM,
 	},
 	{
 		.cmd    = MPTCP_PM_CMD_FLUSH_ADDRS,
 		.doit   = mptcp_nl_cmd_flush_addrs,
-		.flags  = GENL_ADMIN_PERM,
+		.flags  = GENL_UNS_ADMIN_PERM,
 	},
 	{
 		.cmd    = MPTCP_PM_CMD_GET_ADDR,
@@ -2238,7 +2238,7 @@ static const struct genl_small_ops mptcp_pm_ops[] = {
 	{
 		.cmd    = MPTCP_PM_CMD_SET_LIMITS,
 		.doit   = mptcp_nl_cmd_set_limits,
-		.flags  = GENL_ADMIN_PERM,
+		.flags  = GENL_UNS_ADMIN_PERM,
 	},
 	{
 		.cmd    = MPTCP_PM_CMD_GET_LIMITS,
@@ -2247,27 +2247,27 @@ static const struct genl_small_ops mptcp_pm_ops[] = {
 	{
 		.cmd    = MPTCP_PM_CMD_SET_FLAGS,
 		.doit   = mptcp_nl_cmd_set_flags,
-		.flags  = GENL_ADMIN_PERM,
+		.flags  = GENL_UNS_ADMIN_PERM,
 	},
 	{
 		.cmd    = MPTCP_PM_CMD_ANNOUNCE,
 		.doit   = mptcp_nl_cmd_announce,
-		.flags  = GENL_ADMIN_PERM,
+		.flags  = GENL_UNS_ADMIN_PERM,
 	},
 	{
 		.cmd    = MPTCP_PM_CMD_REMOVE,
 		.doit   = mptcp_nl_cmd_remove,
-		.flags  = GENL_ADMIN_PERM,
+		.flags  = GENL_UNS_ADMIN_PERM,
 	},
 	{
 		.cmd    = MPTCP_PM_CMD_SUBFLOW_CREATE,
 		.doit   = mptcp_nl_cmd_sf_create,
-		.flags  = GENL_ADMIN_PERM,
+		.flags  = GENL_UNS_ADMIN_PERM,
 	},
 	{
 		.cmd    = MPTCP_PM_CMD_SUBFLOW_DESTROY,
 		.doit   = mptcp_nl_cmd_sf_destroy,
-		.flags  = GENL_ADMIN_PERM,
+		.flags  = GENL_UNS_ADMIN_PERM,
 	},
 };
 
Signature · 1 line
2.37.1
#1 lore raw

Re: mptcp: allow priviledged operations from user namespaces: Tests Results

Hi Thomas,

Thank you for your modifications, that's great!

Our CI did some validations and here is its report:

- KVM Validation: normal:
- Unstable: 2 failed test(s): kunit selftest_mptcp_join 🔴:
- Task: https://cirrus-ci.com/task/4847158800154624
- Summary: https://api.cirrus-ci.com/v1/artifact/task/4847158800154624/summary/summary.txt

- KVM Validation: debug:
- Unstable: 2 failed test(s): kunit packetdrill_add_addr 🔴:
- Task: https://cirrus-ci.com/task/5973058706997248
- Summary: https://api.cirrus-ci.com/v1/artifact/task/5973058706997248/summary/summary.txt

Initiator: Patchew Applier
Commits: https://github.com/multipath-tcp/mptcp_net-next/commits/c384b8e857ac

If there are some issues, you can reproduce them using the same environment as the one used by the CI thanks to a docker image, e.g.:

    $ cd [kernel source code]
    $ docker run -v "${PWD}:${PWD}:rw" -w "${PWD}" --privileged --rm -it \
        --pull always mptcp/mptcp-upstream-virtme-docker:latest \
        auto-debug

For more details:

https://github.com/multipath-tcp/mptcp-upstream-virtme-docker

Please note that despite all the efforts that have been already done to have a stable tests suite when executed on a public CI like here, it is possible some reported issues are not due to your modifications. Still, do not hesitate to help us improve that ;-)

Cheers,
MPTCP GH Action bot
Bot operated by Matthieu Baerts (Tessares)

#2 lore raw
In reply to Thomas Haller
Quoted 11 lines · Thomas Haller

GENL_ADMIN_PERM checks that the user has CAP_NET_ADMIN in the initial namespace by calling netlink_capable(). Instead, use GENL_UNS_ADMIN_PERM which uses netlink_ns_capable(). This checks that the caller has CAP_NET_ADMIN in the current user namespace.

See also commit 4a92602aa1cd ('openvswitch: allow management from inside user namespaces') which introduced this mechanism. See also commit 5617c6cd6f84 ('nl80211: Allow privileged operations from user namespaces'), which introduced this for nl80211.

Hi Thomas -

Thanks for the patch! This does seem like a good idea to me (and it seems to work ok with our MPTCP tests), but I'd like to get some more community input before merging (Paolo/Florian?).

We also need to figure out which branch this belongs in (we usually tag the patch subject with [PATCH mptcp-net] or [PATCH mptcp-next] to indicate this). If we upstream it to the net branch it will get in to 6.0-rcX and be eligible for backporting to the stable branches. Looks like it would need manual backporting to pre-5.19 kernels due to the incremental changes to mptcp_pm_ops over time.

To upstream this to net it also needs a Fixes tag:

Fixes 702c2f646d42 ("mptcp: netlink: allow userspace-driven subflow establishment")

Thomas, do you have any thoughts/preferences on upstreaming to net or net-next?

Thanks,
Mat

Quoted 76 lines

---
net/mptcp/pm_netlink.c | 18 +++++++++---------

1 file changed, 9 insertions(+), 9 deletions(-)
net/mptcp/pm_netlink.c +9 -9
index 291b5da42fdb..2c145cdc7bdc 100644
@@ -2218,17 +2218,17 @@ static const struct genl_small_ops mptcp_pm_ops[] = {
 	{
 		.cmd    = MPTCP_PM_CMD_ADD_ADDR,
 		.doit   = mptcp_nl_cmd_add_addr,
-		.flags  = GENL_ADMIN_PERM,
+		.flags  = GENL_UNS_ADMIN_PERM,
 	},
 	{
 		.cmd    = MPTCP_PM_CMD_DEL_ADDR,
 		.doit   = mptcp_nl_cmd_del_addr,
-		.flags  = GENL_ADMIN_PERM,
+		.flags  = GENL_UNS_ADMIN_PERM,
 	},
 	{
 		.cmd    = MPTCP_PM_CMD_FLUSH_ADDRS,
 		.doit   = mptcp_nl_cmd_flush_addrs,
-		.flags  = GENL_ADMIN_PERM,
+		.flags  = GENL_UNS_ADMIN_PERM,
 	},
 	{
 		.cmd    = MPTCP_PM_CMD_GET_ADDR,
@@ -2238,7 +2238,7 @@ static const struct genl_small_ops mptcp_pm_ops[] = {
 	{
 		.cmd    = MPTCP_PM_CMD_SET_LIMITS,
 		.doit   = mptcp_nl_cmd_set_limits,
-		.flags  = GENL_ADMIN_PERM,
+		.flags  = GENL_UNS_ADMIN_PERM,
 	},
 	{
 		.cmd    = MPTCP_PM_CMD_GET_LIMITS,
@@ -2247,27 +2247,27 @@ static const struct genl_small_ops mptcp_pm_ops[] = {
 	{
 		.cmd    = MPTCP_PM_CMD_SET_FLAGS,
 		.doit   = mptcp_nl_cmd_set_flags,
-		.flags  = GENL_ADMIN_PERM,
+		.flags  = GENL_UNS_ADMIN_PERM,
 	},
 	{
 		.cmd    = MPTCP_PM_CMD_ANNOUNCE,
 		.doit   = mptcp_nl_cmd_announce,
-		.flags  = GENL_ADMIN_PERM,
+		.flags  = GENL_UNS_ADMIN_PERM,
 	},
 	{
 		.cmd    = MPTCP_PM_CMD_REMOVE,
 		.doit   = mptcp_nl_cmd_remove,
-		.flags  = GENL_ADMIN_PERM,
+		.flags  = GENL_UNS_ADMIN_PERM,
 	},
 	{
 		.cmd    = MPTCP_PM_CMD_SUBFLOW_CREATE,
 		.doit   = mptcp_nl_cmd_sf_create,
-		.flags  = GENL_ADMIN_PERM,
+		.flags  = GENL_UNS_ADMIN_PERM,
 	},
 	{
 		.cmd    = MPTCP_PM_CMD_SUBFLOW_DESTROY,
 		.doit   = mptcp_nl_cmd_sf_destroy,
-		.flags  = GENL_ADMIN_PERM,
+		.flags  = GENL_UNS_ADMIN_PERM,
 	},
 };
 
Signature · 1 line
2.37.1

--
Mat Martineau
Intel

#3 lore raw

Hi Mat,

Quoted 17 lines · Mat Martineau
Quoted 7 lines · Thomas Haller

GENL_ADMIN_PERM checks that the user has CAP_NET_ADMIN in the initial
namespace by calling netlink_capable(). Instead, use GENL_UNS_ADMIN_PERM
which uses netlink_ns_capable(). This checks that the caller has CAP_NET_ADMIN in the current user namespace.

Quoted 7 lines

See also commit 4a92602aa1cd ('openvswitch: allow management from inside
user namespaces') which introduced this mechanism. See also commit 5617c6cd6f84 ('nl80211: Allow privileged operations from user namespaces'), which introduced this for nl80211.

Quoted 7 lines

Hi Thomas -

Thanks for the patch! This does seem like a good idea to me (and it seems
to work ok with our MPTCP tests), but I'd like to get some more community
input before merging (Paolo/Florian?).

thanks for your feedback.

The background for this is that we run unit tests for NetworkManager in a separate network namespace, started as non-root user. Configuring MPTCP endpoints does currently not work in that environment.

Quoted 12 lines

We also need to figure out which branch this belongs in (we usually tag
the patch subject with [PATCH mptcp-net] or [PATCH mptcp-next] to indicate
this). If we upstream it to the net branch it will get in to 6.0-rcX and
be eligible for backporting to the stable branches. Looks like it would
need manual backporting to pre-5.19 kernels due to the incremental changes
to mptcp_pm_ops over time.

Quoted 8 lines

To upstream this to net it also needs a Fixes tag:

Fixes 702c2f646d42 ("mptcp: netlink: allow userspace-driven subflow

establishment")

Thomas, do you have any thoughts/preferences on upstreaming to net or net-next?

I don't have a strong opinion or need. Thanks for explaining the process.

Ffrom what you said, it sounds as this would make sense as [PATCH mptcp-net]. I will send a v2 with the "Fixes:" line.

Thomas

Quoted 84 lines
Quoted 79 lines

---
net/mptcp/pm_netlink.c | 18 +++++++++---------

1 file changed, 9 insertions(+), 9 deletions(-)
net/mptcp/pm_netlink.c +3 -3
index 291b5da42fdb..2c145cdc7bdc 100644
@@ -2218,17 +2218,17 @@ static const struct genl_small_ops
 mptcp_pm_ops[] = {
         {
                 .cmd    = MPTCP_PM_CMD_ADD_ADDR,
                 .doit   = mptcp_nl_cmd_add_addr,
-               .flags  = GENL_ADMIN_PERM,
+               .flags  = GENL_UNS_ADMIN_PERM,
         },
         {
                 .cmd    = MPTCP_PM_CMD_DEL_ADDR,
                 .doit   = mptcp_nl_cmd_del_addr,
-               .flags  = GENL_ADMIN_PERM,
+               .flags  = GENL_UNS_ADMIN_PERM,
         },
         {
                 .cmd    = MPTCP_PM_CMD_FLUSH_ADDRS,
                 .doit   = mptcp_nl_cmd_flush_addrs,
-               .flags  = GENL_ADMIN_PERM,
+               .flags  = GENL_UNS_ADMIN_PERM,
         },
         {

                .cmd    = MPTCP_PM_CMD_GET_ADDR,

net/mptcp/pm_netlink.c +1 -1
@@ -2238,7 +2238,7 @@ static const struct genl_small_ops
 mptcp_pm_ops[] = {
         {
                 .cmd    = MPTCP_PM_CMD_SET_LIMITS,
                 .doit   = mptcp_nl_cmd_set_limits,
-               .flags  = GENL_ADMIN_PERM,
+               .flags  = GENL_UNS_ADMIN_PERM,
         },
         {

                .cmd    = MPTCP_PM_CMD_GET_LIMITS,

net/mptcp/pm_netlink.c +5 -5
@@ -2247,27 +2247,27 @@ static const struct genl_small_ops
 mptcp_pm_ops[] = {
         {
                 .cmd    = MPTCP_PM_CMD_SET_FLAGS,
                 .doit   = mptcp_nl_cmd_set_flags,
-               .flags  = GENL_ADMIN_PERM,
+               .flags  = GENL_UNS_ADMIN_PERM,
         },
         {
                 .cmd    = MPTCP_PM_CMD_ANNOUNCE,
                 .doit   = mptcp_nl_cmd_announce,
-               .flags  = GENL_ADMIN_PERM,
+               .flags  = GENL_UNS_ADMIN_PERM,
         },
         {
                 .cmd    = MPTCP_PM_CMD_REMOVE,
                 .doit   = mptcp_nl_cmd_remove,
-               .flags  = GENL_ADMIN_PERM,
+               .flags  = GENL_UNS_ADMIN_PERM,
         },
         {
                 .cmd    = MPTCP_PM_CMD_SUBFLOW_CREATE,
                 .doit   = mptcp_nl_cmd_sf_create,
-               .flags  = GENL_ADMIN_PERM,
+               .flags  = GENL_UNS_ADMIN_PERM,
         },
         {
                 .cmd    = MPTCP_PM_CMD_SUBFLOW_DESTROY,
                 .doit   = mptcp_nl_cmd_sf_destroy,
-               .flags  = GENL_ADMIN_PERM,
+               .flags  = GENL_UNS_ADMIN_PERM,
         },
 };
Signature · 2 lines
-- 
2.37.1

--
Mat Martineau
Intel

#4 lore raw
In reply to Mat Martineau
Quoted 12 lines · Mat Martineau
Quoted 10 lines · Thomas Haller

GENL_ADMIN_PERM checks that the user has CAP_NET_ADMIN in the initial namespace by calling netlink_capable(). Instead, use GENL_UNS_ADMIN_PERM which uses netlink_ns_capable(). This checks that the caller has CAP_NET_ADMIN in the current user namespace.

See also commit 4a92602aa1cd ('openvswitch: allow management from inside user namespaces') which introduced this mechanism. See also commit 5617c6cd6f84 ('nl80211: Allow privileged operations from user namespaces'), which introduced this for nl80211.

Quoted 7 lines
Quoted 1 line

Hi Thomas -

Thanks for the patch! This does seem like a good idea to me (and it seems to work ok with our MPTCP tests), but I'd like to get some more community input before merging (Paolo/Florian?).

Looks good to me, I don't see any places where we'd accept e.g. a token from a different netns.

mptcp_nl_cmd_add_addr should probably be changed to use GFP_KERNEL_ACCOUNT so that the added entries can be limited via memcg.

mptcp_userspace_pm_append_new_local_addr() helper needs a bit of rework for that, but I don't think this has do be done right away.

Personally I'd go for -next though.

#5 lore raw
Quoted 21 lines · Florian Westphal
Quoted 12 lines · Mat Martineau
Quoted 10 lines · Thomas Haller

GENL_ADMIN_PERM checks that the user has CAP_NET_ADMIN in the initial namespace by calling netlink_capable(). Instead, use GENL_UNS_ADMIN_PERM which uses netlink_ns_capable(). This checks that the caller has CAP_NET_ADMIN in the current user namespace.

See also commit 4a92602aa1cd ('openvswitch: allow management from inside user namespaces') which introduced this mechanism. See also commit 5617c6cd6f84 ('nl80211: Allow privileged operations from user namespaces'), which introduced this for nl80211.

Quoted 7 lines
Quoted 1 line

Hi Thomas -

Thanks for the patch! This does seem like a good idea to me (and it seems to work ok with our MPTCP tests), but I'd like to get some more community input before merging (Paolo/Florian?).

Quoted 8 lines

Looks good to me, I don't see any places where we'd accept e.g. a token from a different netns.

mptcp_nl_cmd_add_addr should probably be changed to use GFP_KERNEL_ACCOUNT so that the added entries can be limited via memcg.

mptcp_userspace_pm_append_new_local_addr() helper needs a bit of rework for that, but I don't think this has do be done right away.

Actually its fine; the allocation is accounted vs the socket.

mptcp_nl_cmd_add_addr has the only "permanent" kmalloc that I can see.

#6 lore raw

PATCH v2 1/2 mptcp: allow priviledged operations from user namespaces

In reply to Mat Martineau

GENL_ADMIN_PERM checks that the user has CAP_NET_ADMIN in the initial namespace by calling netlink_capable(). Instead, use GENL_UNS_ADMIN_PERM which uses netlink_ns_capable(). This checks that the caller has CAP_NET_ADMIN in the current user namespace.

See also commit 4a92602aa1cd ('openvswitch: allow management from inside user namespaces') which introduced this mechanism. See also commit 5617c6cd6f84 ('nl80211: Allow privileged operations from user namespaces'), which introduced this for nl80211.

---
 net/mptcp/pm_netlink.c | 18 +++++++++---------
 1 file changed, 9 insertions(+), 9 deletions(-)
net/mptcp/pm_netlink.c +9 -9
index 291b5da42fdb..2c145cdc7bdc 100644
@@ -2218,17 +2218,17 @@ static const struct genl_small_ops mptcp_pm_ops[] = {
 	{
 		.cmd    = MPTCP_PM_CMD_ADD_ADDR,
 		.doit   = mptcp_nl_cmd_add_addr,
-		.flags  = GENL_ADMIN_PERM,
+		.flags  = GENL_UNS_ADMIN_PERM,
 	},
 	{
 		.cmd    = MPTCP_PM_CMD_DEL_ADDR,
 		.doit   = mptcp_nl_cmd_del_addr,
-		.flags  = GENL_ADMIN_PERM,
+		.flags  = GENL_UNS_ADMIN_PERM,
 	},
 	{
 		.cmd    = MPTCP_PM_CMD_FLUSH_ADDRS,
 		.doit   = mptcp_nl_cmd_flush_addrs,
-		.flags  = GENL_ADMIN_PERM,
+		.flags  = GENL_UNS_ADMIN_PERM,
 	},
 	{
 		.cmd    = MPTCP_PM_CMD_GET_ADDR,
@@ -2238,7 +2238,7 @@ static const struct genl_small_ops mptcp_pm_ops[] = {
 	{
 		.cmd    = MPTCP_PM_CMD_SET_LIMITS,
 		.doit   = mptcp_nl_cmd_set_limits,
-		.flags  = GENL_ADMIN_PERM,
+		.flags  = GENL_UNS_ADMIN_PERM,
 	},
 	{
 		.cmd    = MPTCP_PM_CMD_GET_LIMITS,
@@ -2247,27 +2247,27 @@ static const struct genl_small_ops mptcp_pm_ops[] = {
 	{
 		.cmd    = MPTCP_PM_CMD_SET_FLAGS,
 		.doit   = mptcp_nl_cmd_set_flags,
-		.flags  = GENL_ADMIN_PERM,
+		.flags  = GENL_UNS_ADMIN_PERM,
 	},
 	{
 		.cmd    = MPTCP_PM_CMD_ANNOUNCE,
 		.doit   = mptcp_nl_cmd_announce,
-		.flags  = GENL_ADMIN_PERM,
+		.flags  = GENL_UNS_ADMIN_PERM,
 	},
 	{
 		.cmd    = MPTCP_PM_CMD_REMOVE,
 		.doit   = mptcp_nl_cmd_remove,
-		.flags  = GENL_ADMIN_PERM,
+		.flags  = GENL_UNS_ADMIN_PERM,
 	},
 	{
 		.cmd    = MPTCP_PM_CMD_SUBFLOW_CREATE,
 		.doit   = mptcp_nl_cmd_sf_create,
-		.flags  = GENL_ADMIN_PERM,
+		.flags  = GENL_UNS_ADMIN_PERM,
 	},
 	{
 		.cmd    = MPTCP_PM_CMD_SUBFLOW_DESTROY,
 		.doit   = mptcp_nl_cmd_sf_destroy,
-		.flags  = GENL_ADMIN_PERM,
+		.flags  = GENL_UNS_ADMIN_PERM,
 	},
 };
 
Signature · 1 line
2.37.1
#7 lore raw

PATCH v2 2/2 mptcp: account memory allocation in mptcp_nl_cmd_add_addr() to user

Now that non-root users can configure MPTCP endpoints, account the memory allocation to the user.

---
 net/mptcp/pm_netlink.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
net/mptcp/pm_netlink.c +1 -1
index 2c145cdc7bdc..79469178a2a8 100644
@@ -1327,7 +1327,7 @@ static int mptcp_nl_cmd_add_addr(struct sk_buff *skb, struct genl_info *info)
 		return -EINVAL;
 	}
 
-	entry = kmalloc(sizeof(*entry), GFP_KERNEL);
+	entry = kmalloc(sizeof(*entry), GFP_KERNEL_ACCOUNT);
 	if (!entry) {
 		GENL_SET_ERR_MSG(info, "can't allocate addr");
 		return -ENOMEM;
Signature · 1 line
2.37.1
Reviewed-by on v2 2/2
#8 lore raw

Re: [PATCH mptcp-next v2 2/2] mptcp: account memory allocation in mptcp_nl_cmd_add_addr() to user

Quoted 4 lines · Thomas Haller

Now that non-root users can configure MPTCP endpoints, account the memory allocation to the user.

Looks good:

Quoted 21 lines

---
net/mptcp/pm_netlink.c | 2 +-

1 file changed, 1 insertion(+), 1 deletion(-)
net/mptcp/pm_netlink.c +1 -1
index 2c145cdc7bdc..79469178a2a8 100644
@@ -1327,7 +1327,7 @@ static int mptcp_nl_cmd_add_addr(struct sk_buff *skb, struct genl_info *info)
 		return -EINVAL;
 	}
 
-	entry = kmalloc(sizeof(*entry), GFP_KERNEL);
+	entry = kmalloc(sizeof(*entry), GFP_KERNEL_ACCOUNT);
 	if (!entry) {
 		GENL_SET_ERR_MSG(info, "can't allocate addr");
 		return -ENOMEM;
Signature · 1 line
2.37.1

--
Mat Martineau
Intel

Reviewed-by on v2 1/2
#9 lore raw
In reply to Thomas Haller
Quoted 11 lines · Thomas Haller

GENL_ADMIN_PERM checks that the user has CAP_NET_ADMIN in the initial namespace by calling netlink_capable(). Instead, use GENL_UNS_ADMIN_PERM which uses netlink_ns_capable(). This checks that the caller has CAP_NET_ADMIN in the current user namespace.

See also commit 4a92602aa1cd ('openvswitch: allow management from inside user namespaces') which introduced this mechanism. See also commit 5617c6cd6f84 ('nl80211: Allow privileged operations from user namespaces'), which introduced this for nl80211.

Thanks Thomas, looks good:

Florian had mentioned preferring net-next in the discussion of v1 (which Thomas has agreed with by labeling this for mptcp-next), and I agree. I don't think it quite meets the bar for -net or stable backporting and it would be easier to explain that "6.1 and later support user namespaces for MPTCP generic netlink commands". (Matthieu, if someone makes a convincing case for -net, it's up to you :) )

- Mat

Quoted 75 lines

---
net/mptcp/pm_netlink.c | 18 +++++++++---------

1 file changed, 9 insertions(+), 9 deletions(-)
net/mptcp/pm_netlink.c +9 -9
index 291b5da42fdb..2c145cdc7bdc 100644
@@ -2218,17 +2218,17 @@ static const struct genl_small_ops mptcp_pm_ops[] = {
 	{
 		.cmd    = MPTCP_PM_CMD_ADD_ADDR,
 		.doit   = mptcp_nl_cmd_add_addr,
-		.flags  = GENL_ADMIN_PERM,
+		.flags  = GENL_UNS_ADMIN_PERM,
 	},
 	{
 		.cmd    = MPTCP_PM_CMD_DEL_ADDR,
 		.doit   = mptcp_nl_cmd_del_addr,
-		.flags  = GENL_ADMIN_PERM,
+		.flags  = GENL_UNS_ADMIN_PERM,
 	},
 	{
 		.cmd    = MPTCP_PM_CMD_FLUSH_ADDRS,
 		.doit   = mptcp_nl_cmd_flush_addrs,
-		.flags  = GENL_ADMIN_PERM,
+		.flags  = GENL_UNS_ADMIN_PERM,
 	},
 	{
 		.cmd    = MPTCP_PM_CMD_GET_ADDR,
@@ -2238,7 +2238,7 @@ static const struct genl_small_ops mptcp_pm_ops[] = {
 	{
 		.cmd    = MPTCP_PM_CMD_SET_LIMITS,
 		.doit   = mptcp_nl_cmd_set_limits,
-		.flags  = GENL_ADMIN_PERM,
+		.flags  = GENL_UNS_ADMIN_PERM,
 	},
 	{
 		.cmd    = MPTCP_PM_CMD_GET_LIMITS,
@@ -2247,27 +2247,27 @@ static const struct genl_small_ops mptcp_pm_ops[] = {
 	{
 		.cmd    = MPTCP_PM_CMD_SET_FLAGS,
 		.doit   = mptcp_nl_cmd_set_flags,
-		.flags  = GENL_ADMIN_PERM,
+		.flags  = GENL_UNS_ADMIN_PERM,
 	},
 	{
 		.cmd    = MPTCP_PM_CMD_ANNOUNCE,
 		.doit   = mptcp_nl_cmd_announce,
-		.flags  = GENL_ADMIN_PERM,
+		.flags  = GENL_UNS_ADMIN_PERM,
 	},
 	{
 		.cmd    = MPTCP_PM_CMD_REMOVE,
 		.doit   = mptcp_nl_cmd_remove,
-		.flags  = GENL_ADMIN_PERM,
+		.flags  = GENL_UNS_ADMIN_PERM,
 	},
 	{
 		.cmd    = MPTCP_PM_CMD_SUBFLOW_CREATE,
 		.doit   = mptcp_nl_cmd_sf_create,
-		.flags  = GENL_ADMIN_PERM,
+		.flags  = GENL_UNS_ADMIN_PERM,
 	},
 	{
 		.cmd    = MPTCP_PM_CMD_SUBFLOW_DESTROY,
 		.doit   = mptcp_nl_cmd_sf_destroy,
-		.flags  = GENL_ADMIN_PERM,
+		.flags  = GENL_UNS_ADMIN_PERM,
 	},
 };
 
Signature · 1 line
2.37.1

--
Mat Martineau
Intel

#10 lore raw

Hi Thomas, Mat, Florian,

Quoted 12 lines · Mat Martineau
Quoted 10 lines · Thomas Haller

GENL_ADMIN_PERM checks that the user has CAP_NET_ADMIN in the initial namespace by calling netlink_capable(). Instead, use GENL_UNS_ADMIN_PERM which uses netlink_ns_capable(). This checks that the caller has CAP_NET_ADMIN in the current user namespace.

See also commit 4a92602aa1cd ('openvswitch: allow management from inside user namespaces') which introduced this mechanism. See also commit 5617c6cd6f84 ('nl80211: Allow privileged operations from user namespaces'), which introduced this for nl80211.

Quoted 5 lines
Quoted 1 line

Thanks Thomas, looks good:

Thank you for the patches and the reviews!

Quoted 6 lines

Florian had mentioned preferring net-next in the discussion of v1 (which Thomas has agreed with by labeling this for mptcp-next), and I agree. I don't think it quite meets the bar for -net or stable backporting and it would be easier to explain that "6.1 and later support user namespaces for MPTCP generic netlink commands". (Matthieu, if someone makes a convincing case for -net, it's up to you :) )

I initially thought it was more a bug-fix but I also agree with Florian. Also I see that the two mentioned commits above don't have a Fixes tag.

The only exception I found and related to this flag was for l2tp: the commit 2abe05234f2e ("l2tp: Allow management of tunnels and session in user namespace") has been selected by davem and backported to v5.4.

https://lore.kernel.org/stable/[email protected]/
(see patch 03/19)

Anyway, your 2 patches are now in our tree (feat. for net-next) with Mat's RvB tag and without a typo (s/priviledged/privileged/) + a small fix for checkpatch related to how the commit are mentioned, nothing important:

New patches for t/upstream:
- 3d542a6c45ea: mptcp: allow privileged operations from user namespaces
- 11bdb1959854: mptcp: account memory allocation in mptcp_nl_cmd_add_addr() to user
- Results: c4a0ae952875..004104cc8a77 (export)

Tests are now in progress:

https://cirrus-ci.com/github/multipath-tcp/mptcp_net-next/export/20220811T101116

@Thomas: BTW, thank you for maintaining libnl!
Funny that you sent these patches to MPTCP while earlier this week I sent the latest libnl version to Debian :)

https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1016867
https://tracker.debian.org/pkg/libnl3

Cheers,
Matt

Signature · 2 lines
Tessares | Belgium | Hybrid Access Solutions
www.tessares.net
#11 lore raw

Hi Matthieu,

Quoted 29 lines · Matthieu Baerts

Hi Thomas, Mat, Florian,

On 11/08/2022 01:08, Mat Martineau wrote:
...

Quoted 4 lines

Thanks Thomas, looks good:

Thank you for the patches and the reviews!

Quoted 1 line

[...]

I initially thought it was more a bug-fix but I also agree with Florian.
Also I see that the two mentioned commits above don't have a Fixes tag.

The only exception I found and related to this flag was for l2tp: the commit 2abe05234f2e ("l2tp: Allow management of tunnels and session in
user namespace") has been selected by davem and backported to v5.4.

https://lore.kernel.org/stable/[email protected]/
(see patch 03/19)

Quoted 5 lines

Anyway, your 2 patches are now in our tree (feat. for net-next) with Mat's RvB tag and without a typo (s/priviledged/privileged/) + a small
fix for checkpatch related to how the commit are mentioned, nothing important:

Thank you for the adjustments, the feedback and applying!!

Quoted 13 lines

New patches for t/upstream:
- 3d542a6c45ea: mptcp: allow privileged operations from user namespaces
- 11bdb1959854: mptcp: account memory allocation in mptcp_nl_cmd_add_addr() to user
- Results: c4a0ae952875..004104cc8a77 (export)

Tests are now in progress:

https://cirrus-ci.com/github/multipath-tcp/mptcp_net-next/export/20220811T101116

Quoted 6 lines

@Thomas: BTW, thank you for maintaining libnl!
Funny that you sent these patches to MPTCP while earlier this week I sent the latest libnl version to Debian :)

https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1016867
https://tracker.debian.org/pkg/libnl3

Oh, that's nice. After I rather neglected libnl for quite some time, I want do to a better job in the future. :)

On an unrelated note: upcoming NetworkManager 1.40.0 release will have basic MPTCP support and configure endpoints. Similar to what mptcpd does. This was the backstory for the patches.

best,
Thomas

#12 lore raw

Hi Thomas,

Quoted 1 line · Thomas Haller

On Thu, 2022-08-11 at 12:34 +0200, Matthieu Baerts wrote:

(...)

Quoted 9 lines
Quoted 6 lines

@Thomas: BTW, thank you for maintaining libnl!
Funny that you sent these patches to MPTCP while earlier this week I sent the latest libnl version to Debian :)

https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1016867
https://tracker.debian.org/pkg/libnl3

Oh, that's nice. After I rather neglected libnl for quite some time, I want do to a better job in the future. :)

:-)

Quoted 3 lines

On an unrelated note: upcoming NetworkManager 1.40.0 release will have basic MPTCP support and configure endpoints. Similar to what mptcpd does. This was the backstory for the patches.

That's a really good news! Thank you for having worked on that, preparing a setup to be used with MPTCP is going to be easier now!

Cheers,
Matt

Signature · 2 lines
Tessares | Belgium | Hybrid Access Solutions
www.tessares.net