| Message ID | 20240807190236.136388-1-ayaka@soulik.info (mailing list archive) |
|---|---|
| State | Changes Requested |
| Delegated to: | Netdev Maintainers |
| Headers | show |
| Series | [net-next,v3] net: tuntap: add ioctl() TUNGETQUEUEINDEX to fetch queue index | expand |
On Wed, Aug 7, 2024 at 3:02 PM Randy Li <ayaka@soulik.info> wrote: > > We need the queue index in qdisc mapping rule and the result > value for the steering eBPF. > There was no way to know that. This commit message is very short. There's an ongoing conversation in v2 on whether this feature is needed, given that tc and tun already have a variety of ways to both drop and steer traffic. So a bit too soon for a v3. The big question that the commit message should answer is why this feature is needed given the alternatives. Essentially, in v4 please include the short conclusion of the discussion. Also add a Link: to the full discussion on lore. > Changelog: > v3: > fixes two style issues in the previous commit > v2: > Fixes the flow when the queue is disabled in the tap type device. > Put this ioctl() under the lock protection for the tun device. > > > Signed-off-by: Randy Li <ayaka@soulik.info> > --- > drivers/net/tap.c | 10 ++++++++++ > drivers/net/tun.c | 13 +++++++++++++ > include/uapi/linux/if_tun.h | 1 + > 3 files changed, 24 insertions(+) > > diff --git a/drivers/net/tap.c b/drivers/net/tap.c > index 77574f7a3bd4..bbd717cf78a5 100644 > --- a/drivers/net/tap.c > +++ b/drivers/net/tap.c > @@ -1120,6 +1120,16 @@ static long tap_ioctl(struct file *file, unsigned int cmd, > rtnl_unlock(); > return ret; > > + case TUNGETQUEUEINDEX: > + rtnl_lock(); > + if (!q->enabled) > + ret = -EINVAL; > + else > + ret = put_user(q->queue_index, up); > + > + rtnl_unlock(); > + return ret; > + > case SIOCGIFHWADDR: > rtnl_lock(); > tap = tap_get_tap_dev(q); > diff --git a/drivers/net/tun.c b/drivers/net/tun.c > index 1d06c560c5e6..0c527ccdeab3 100644 > --- a/drivers/net/tun.c > +++ b/drivers/net/tun.c > @@ -3151,6 +3151,19 @@ static long __tun_chr_ioctl(struct file *file, unsigned int cmd, > tfile->ifindex = ifindex; > goto unlock; > } > + if (cmd == TUNGETQUEUEINDEX) { > + ret = -EINVAL; > + if (tfile->detached) > + goto unlock; > + > + ret = -EFAULT; > + if (put_user(tfile->queue_index, (unsigned int __user *)argp)) > + goto unlock; > + > + ret = 0; > + goto unlock; > + } > + > > ret = -EBADFD; > if (!tun) > diff --git a/include/uapi/linux/if_tun.h b/include/uapi/linux/if_tun.h > index 287cdc81c939..2668ca3b06a5 100644 > --- a/include/uapi/linux/if_tun.h > +++ b/include/uapi/linux/if_tun.h > @@ -61,6 +61,7 @@ > #define TUNSETFILTEREBPF _IOR('T', 225, int) > #define TUNSETCARRIER _IOW('T', 226, int) > #define TUNGETDEVNETNS _IO('T', 227) > +#define TUNGETQUEUEINDEX _IOR('T', 228, unsigned int) > > /* TUNSETIFF ifr flags */ > #define IFF_TUN 0x0001 > -- > 2.45.2 >
diff --git a/drivers/net/tap.c b/drivers/net/tap.c index 77574f7a3bd4..bbd717cf78a5 100644 --- a/drivers/net/tap.c +++ b/drivers/net/tap.c @@ -1120,6 +1120,16 @@ static long tap_ioctl(struct file *file, unsigned int cmd, rtnl_unlock(); return ret; + case TUNGETQUEUEINDEX: + rtnl_lock(); + if (!q->enabled) + ret = -EINVAL; + else + ret = put_user(q->queue_index, up); + + rtnl_unlock(); + return ret; + case SIOCGIFHWADDR: rtnl_lock(); tap = tap_get_tap_dev(q); diff --git a/drivers/net/tun.c b/drivers/net/tun.c index 1d06c560c5e6..0c527ccdeab3 100644 --- a/drivers/net/tun.c +++ b/drivers/net/tun.c @@ -3151,6 +3151,19 @@ static long __tun_chr_ioctl(struct file *file, unsigned int cmd, tfile->ifindex = ifindex; goto unlock; } + if (cmd == TUNGETQUEUEINDEX) { + ret = -EINVAL; + if (tfile->detached) + goto unlock; + + ret = -EFAULT; + if (put_user(tfile->queue_index, (unsigned int __user *)argp)) + goto unlock; + + ret = 0; + goto unlock; + } + ret = -EBADFD; if (!tun) diff --git a/include/uapi/linux/if_tun.h b/include/uapi/linux/if_tun.h index 287cdc81c939..2668ca3b06a5 100644 --- a/include/uapi/linux/if_tun.h +++ b/include/uapi/linux/if_tun.h @@ -61,6 +61,7 @@ #define TUNSETFILTEREBPF _IOR('T', 225, int) #define TUNSETCARRIER _IOW('T', 226, int) #define TUNGETDEVNETNS _IO('T', 227) +#define TUNGETQUEUEINDEX _IOR('T', 228, unsigned int) /* TUNSETIFF ifr flags */ #define IFF_TUN 0x0001
We need the queue index in qdisc mapping rule and the result value for the steering eBPF. There was no way to know that. Changelog: v3: fixes two style issues in the previous commit v2: Fixes the flow when the queue is disabled in the tap type device. Put this ioctl() under the lock protection for the tun device. Signed-off-by: Randy Li <ayaka@soulik.info> --- drivers/net/tap.c | 10 ++++++++++ drivers/net/tun.c | 13 +++++++++++++ include/uapi/linux/if_tun.h | 1 + 3 files changed, 24 insertions(+)