From: Yumei Huang <yuhuang@redhat.com>
To: Stefano Brivio <sbrivio@redhat.com>
Cc: passt-dev@passt.top, david@gibson.dropbear.id.au
Subject: Re: [PATCH] conf, pasta: Add --no-tap option
Date: Tue, 13 Jan 2026 19:20:47 +0800 [thread overview]
Message-ID: <CANsz47kuSByrnTQaRj46jZ=EgBK3DCC2OiyA+Tdi=bOMGS9=+Q@mail.gmail.com> (raw)
In-Reply-To: <20260110191226.570a3f0b@elisabeth>
On Sun, Jan 11, 2026 at 2:12 AM Stefano Brivio <sbrivio@redhat.com> wrote:
>
> On Mon, 29 Dec 2025 17:55:58 +0800
> Yumei Huang <yuhuang@redhat.com> wrote:
>
> > This patch introduces a mode where we only forward loopback connections
> > and traffic between two namespaces (via the loopback interface, 'lo'),
> > without a tap device.
> >
> > With this, podman can support forwarding ::1 in custom networks when using
> > rootlesskit for forwarding ports.
> >
> > In --no-tap mode, --host-lo-to-ns-lo, --no-icmp and --no-ra is automatically
> > enabled. Options requiring a tap device (--ns-ifname, --ns-mac-addr,
> > --config-net, --outbound-if4/6) are rejected.
> >
> > Link: https://bugs.passt.top/show_bug.cgi?id=149
> > Signed-off-by: Yumei Huang <yuhuang@redhat.com>
> > ---
> > conf.c | 56 +++++++++++++++++++++++++++++++++++++++++---------------
> > fwd.c | 3 +++
> > passt.1 | 5 +++++
> > passt.h | 2 ++
> > pasta.c | 3 +++
> > tap.c | 11 +++++++----
> > 6 files changed, 61 insertions(+), 19 deletions(-)
> >
> > diff --git a/conf.c b/conf.c
> > index 84ae12b..353d0a5 100644
> > --- a/conf.c
> > +++ b/conf.c
> > @@ -1049,7 +1049,8 @@ pasta_opts:
> > " --no-copy-addrs DEPRECATED:\n"
> > " Don't copy all addresses to namespace\n"
> > " --ns-mac-addr ADDR Set MAC address on tap interface\n"
> > - " --no-splice Disable inbound socket splicing\n");
> > + " --no-splice Disable inbound socket splicing\n"
> > + " --no-tap Don't create tap device\n");
> >
> > passt_exit(status);
> > }
> > @@ -1451,6 +1452,7 @@ void conf(struct ctx *c, int argc, char **argv)
> > {"no-ndp", no_argument, &c->no_ndp, 1 },
> > {"no-ra", no_argument, &c->no_ra, 1 },
> > {"no-splice", no_argument, &c->no_splice, 1 },
> > + {"no-tap", no_argument, &c->no_tap, 1 },
> > {"freebind", no_argument, &c->freebind, 1 },
> > {"no-map-gw", no_argument, &no_map_gw, 1 },
> > {"ipv4-only", no_argument, NULL, '4' },
> > @@ -1947,8 +1949,11 @@ void conf(struct ctx *c, int argc, char **argv)
> > }
> > } while (name != -1);
> >
> > - if (c->mode != MODE_PASTA)
> > + if (c->mode != MODE_PASTA) {
> > c->no_splice = 1;
> > + if (c->no_tap)
> > + die("--no-tap is for pasta mode only");
> > + }
> >
> > if (c->mode == MODE_PASTA && !c->pasta_conf_ns) {
> > if (copy_routes_opt)
> > @@ -1957,6 +1962,25 @@ void conf(struct ctx *c, int argc, char **argv)
> > die("--no-copy-addrs needs --config-net");
> > }
> >
> > + if (c->mode == MODE_PASTA && c->no_tap) {
> > + if (c->no_splice)
> > + die("--no-tap is incompatible with --no-splice");
>
> I'm not sure if you need this for other reasons, but as long as it's
> called --no-tap, it's not really incompatible with --no-splice.
I will update it to --splice-only
>
> Maybe users just want to get a disconnected namespace for whatever
> reason ('pasta' is shorter to type than 'unshare -rUn').
>
> > + if (*c->ip4.ifname_out || *c->ip6.ifname_out)
> > + die("--no-tap is incompatible with --outbound-if4/6");
> > + if (*c->pasta_ifn)
> > + die("--no-tap is incompatible with --ns-ifname");
> > + if (*c->guest_mac)
> > + die("--no-tap is incompatible with --ns-mac-addr");
> > + if (c->pasta_conf_ns)
> > + die("--no-tap is incompatible with --config-net");
>
> I guess all these checks are to save some checks later, which looks like
> a good reason to have them here.
>
> If not, though, I don't think we *really* need to tell the user that
> --ns-ifname will be ignored with --no-tap.
>
> One thing that might confuse users, though, is this:
>
> $ ./pasta --no-tap --mtu 1500 -- ip l
> 1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN mode DEFAULT group default qlen 1000
> link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
>
> or even this:
>
> $ ./pasta --no-tap -a 192.0.2.1 -- ip a
> 1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group default qlen 1000
> link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
> inet 127.0.0.1/8 scope host lo
> valid_lft forever preferred_lft forever
> inet6 ::1/128 scope host proto kernel_lo
> valid_lft forever preferred_lft forever
>
> but I would rather *not* add conditions and checks for those even if
> there's a *slight* potential for confusion, otherwise this becomes
> really long. And it's really not worth it, I think.
Then I guess we only need the c->no_splice check, right?
>
> > +
> > + c->host_lo_to_ns_lo = 1;
> > + c->no_icmp = 1;
> > + c->no_ra = 1;
> > + c->no_dns = 1;
> > + c->no_dns_search = 1;
> > + }
> > +
> > if (!ifi4 && *c->ip4.ifname_out)
> > ifi4 = if_nametoindex(c->ip4.ifname_out);
> >
> > @@ -1980,9 +2004,9 @@ void conf(struct ctx *c, int argc, char **argv)
> > log_conf_parsed = true; /* Stop printing everything */
> >
> > nl_sock_init(c, false);
> > - if (!v6_only)
> > + if (!v6_only && !c->no_tap)
> > c->ifi4 = conf_ip4(ifi4, &c->ip4);
> > - if (!v4_only)
> > + if (!v4_only && !c->no_tap)
> > c->ifi6 = conf_ip6(ifi6, &c->ip6);
> >
> > if (c->ifi4 && c->mtu < IPV4_MIN_MTU) {
> > @@ -1998,30 +2022,32 @@ void conf(struct ctx *c, int argc, char **argv)
> > (*c->ip6.ifname_out && !c->ifi6))
> > die("External interface not usable");
> >
> > - if (!c->ifi4 && !c->ifi6 && !*c->pasta_ifn) {
> > + if (!c->ifi4 && !c->ifi6 && !*c->pasta_ifn && !c->no_tap) {
>
> You already checked that !*c->pasta_ifn above.
I guess the check above (aka. if (*c->pasta_ifn && c->no_tap)) doesn't
affect this one? If c->pasta_ifn is assigned, we won't come to the
check !c->no_tap here. Otherwise, we do need to check !c->no_tap.
>
> > strncpy(c->pasta_ifn, pasta_default_ifn,
> > sizeof(c->pasta_ifn) - 1);
> > }
> >
> > if (!c->ifi4 && !v6_only) {
> > - info("IPv4: no external interface as template, use local mode");
> > -
> > - conf_ip4_local(&c->ip4);
> > + if (!c->no_tap) {
> > + info("IPv4: no external interface as template, use local mode");
> > + conf_ip4_local(&c->ip4);
> > + }
> > c->ifi4 = -1;
> > }
> >
> > if (!c->ifi6 && !v4_only) {
> > - info("IPv6: no external interface as template, use local mode");
> > -
> > - conf_ip6_local(&c->ip6);
> > + if (!c->no_tap) {
> > + info("IPv6: no external interface as template, use local mode");
> > + conf_ip6_local(&c->ip6);
> > + }
> > c->ifi6 = -1;
> > }
> >
> > - if (c->ifi4 && !no_map_gw &&
> > + if (c->ifi4 > 0 && !no_map_gw &&
> > IN4_IS_ADDR_UNSPECIFIED(&c->ip4.map_host_loopback))
> > c->ip4.map_host_loopback = c->ip4.guest_gw;
> >
> > - if (c->ifi6 && !no_map_gw &&
> > + if (c->ifi6 > 0 && !no_map_gw &&
> > IN6_IS_ADDR_UNSPECIFIED(&c->ip6.map_host_loopback))
> > c->ip6.map_host_loopback = c->ip6.guest_gw;
> >
> > @@ -2116,10 +2142,10 @@ void conf(struct ctx *c, int argc, char **argv)
> > conf_ports(c, name, optarg, &c->udp.fwd_out);
> > } while (name != -1);
> >
> > - if (!c->ifi4)
> > + if (c->ifi4 <= 0)
> > c->no_dhcp = 1;
> >
> > - if (!c->ifi6) {
> > + if (c->ifi6 <= 0) {
> > c->no_ndp = 1;
> > c->no_dhcpv6 = 1;
> > } else if (IN6_IS_ADDR_UNSPECIFIED(&c->ip6.addr)) {
> > diff --git a/fwd.c b/fwd.c
> > index 44a0e10..2f4a89a 100644
> > --- a/fwd.c
> > +++ b/fwd.c
> > @@ -780,6 +780,9 @@ uint8_t fwd_nat_from_host(const struct ctx *c, uint8_t proto,
> > return PIF_SPLICE;
> > }
> >
> > + if (c->no_tap)
> > + return PIF_NONE;
> > +
> > if (!nat_inbound(c, &ini->eaddr, &tgt->oaddr)) {
> > if (inany_v4(&ini->eaddr)) {
> > if (IN4_IS_ADDR_UNSPECIFIED(&c->ip4.our_tap_addr))
> > diff --git a/passt.1 b/passt.1
> > index db0d662..2d643f7 100644
> > --- a/passt.1
> > +++ b/passt.1
> > @@ -755,6 +755,11 @@ Default is to let the tap driver build a pseudorandom hardware address.
> > Disable the bypass path for inbound, local traffic. See the section \fBHandling
> > of local traffic in pasta\fR in the \fBNOTES\fR for more details.
> >
> > +.TP
> > +.BR \-\-no-tap
> > +Do not create a tap device in the namespace. In this mode, only local loopback
> > +traffic between namespaces is forwarded using splice.
>
> "Using splice" isn't really clear, and it's not entirely correct in the
> case of UDP: there's no splice() system call there, even if we call
> some UDP flows "spliced" for analogy with TCP.
>
> Maybe just omit it, say:
>
> [...] In this mode, \fIpasta\fR only
> forwards loopback traffic between namespaces.
Do you think we still need "Do not create a tap device in the
namespace" after updating it to --splice-only?
>
> ?
>
> > +
> > .SH EXAMPLES
> >
> > .SS \fBpasta
> > diff --git a/passt.h b/passt.h
> > index 79d01dd..0c1ec4c 100644
> > --- a/passt.h
> > +++ b/passt.h
> > @@ -200,6 +200,7 @@ struct ip6_ctx {
> > * @no_ndp: Disable NDP handler altogether
> > * @no_ra: Disable router advertisements
> > * @no_splice: Disable socket splicing for inbound traffic
> > + * @no_tap: Do not create tap device
> > * @host_lo_to_ns_lo: Map host loopback addresses to ns loopback addresses
> > * @freebind: Allow binding of non-local addresses for forwarding
> > * @low_wmem: Low probed net.core.wmem_max
> > @@ -277,6 +278,7 @@ struct ctx {
> > int no_ndp;
> > int no_ra;
> > int no_splice;
> > + int no_tap;
> > int host_lo_to_ns_lo;
> > int freebind;
> >
> > diff --git a/pasta.c b/pasta.c
> > index 0ddd6b0..3510ec5 100644
> > --- a/pasta.c
> > +++ b/pasta.c
> > @@ -316,6 +316,9 @@ void pasta_ns_conf(struct ctx *c)
> > die("Couldn't bring up loopback interface in namespace: %s",
> > strerror_(-rc));
> >
> > + if (c->no_tap)
> > + return;
> > +
> > /* Get or set MAC in target namespace */
> > if (MAC_IS_ZERO(c->guest_mac))
> > nl_link_get_mac(nl_sock_ns, c->pasta_ifi, c->guest_mac);
> > diff --git a/tap.c b/tap.c
> > index 9d1344b..9b4eedc 100644
> > --- a/tap.c
> > +++ b/tap.c
> > @@ -1491,13 +1491,16 @@ static int tap_ns_tun(void *arg)
> > */
> > static void tap_sock_tun_init(struct ctx *c)
> > {
> > - NS_CALL(tap_ns_tun, c);
> > - if (c->fd_tap == -1)
> > - die("Failed to set up tap device in namespace");
> > + if (!c->no_tap) {
> > + NS_CALL(tap_ns_tun, c);
> > + if (c->fd_tap == -1)
> > + die("Failed to set up tap device in namespace");
> > + }
> >
> > pasta_ns_conf(c);
> >
> > - tap_start_connection(c);
> > + if (!c->no_tap)
> > + tap_start_connection(c);
> > }
> >
> > /**
>
> Other than that, minus pending comments, it all looks good to me.
Thank for the review and comments!
>
> --
> Stefano
>
--
Thanks,
Yumei Huang
prev parent reply other threads:[~2026-01-13 11:21 UTC|newest]
Thread overview: 16+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-12-29 9:55 Yumei Huang
2025-12-31 15:07 ` Stefano Brivio
2026-01-05 4:18 ` David Gibson
2026-01-05 8:53 ` Yumei Huang
2026-01-10 18:12 ` Stefano Brivio
2026-01-12 4:26 ` David Gibson
2026-01-13 0:12 ` Stefano Brivio
2026-01-13 2:39 ` David Gibson
2026-01-13 9:57 ` Yumei Huang
2026-01-05 13:48 ` Paul Holzinger
2026-01-05 21:10 ` Stefano Brivio
2026-01-07 15:20 ` Paul Holzinger
2026-01-10 18:12 ` Stefano Brivio
2026-01-12 8:20 ` Yumei Huang
2026-01-10 18:12 ` Stefano Brivio
2026-01-13 11:20 ` Yumei Huang [this message]
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='CANsz47kuSByrnTQaRj46jZ=EgBK3DCC2OiyA+Tdi=bOMGS9=+Q@mail.gmail.com' \
--to=yuhuang@redhat.com \
--cc=david@gibson.dropbear.id.au \
--cc=passt-dev@passt.top \
--cc=sbrivio@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.
Code repositories for project(s) associated with this public inbox
https://passt.top/passt
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 IMAP folder(s).