udp_sock_handler() takes a number of datagrams from sockets that depending
on their addresses could be forwarded either to the L2 interface ("tap")
or to another socket ("spliced"). In the latter case we can also only
send packets together if they have the same source port, and therefore
are sent via the same socket.
To reduce the total number of system calls we gather contiguous batches of
datagrams with the same destination interface and socket where applicable.
The determination of what the target is is made by udp_mmh_splice_port().
It returns the source port for splice packets and -1 for "tap" packets.
We find batches by looking ahead in our queue until we find a datagram
whose "splicefrom" port doesn't match the first in our current batch.
udp_mmh_splice_port() is moderately expensive, and unfortunately we
can call it twice on the same datagram: once as the (last + 1) entry
in one batch (to check it's not in that batch), then again as the
first entry in the next batch.
Avoid this by keeping track of the "splice port" in the metadata structure,
and filling it in one entry ahead of the one we're currently considering.
This is a bit subtle, but not that hard. It will also generalise better
when we have more complex possibilities based on the flow table.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
udp_mmh_splice_port() is used to determine if a UDP datagram can be
"spliced" (forwarded via a socket instead of tap). We only invoke it if
the origin socket has the 'splice' flag set.
Fold the checking of the flag into the helper itself, which makes the
caller simpler. It does mean we have a loop looking for a batch of
spliceable or non-spliceable packets even in the case where the flag is
clear. This shouldn't be that expensive though, since each call to
udp_mmh_splice_port() will return without accessing memory in that case.
In any case we're going to need a similar loop in more cases with upcoming
flow table work.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
sock_l4() creates, binds and otherwise prepares a new socket. It builds
the socket address to bind from separately provided address and port.
However, we have use cases coming up where it's more natural to construct
the socket address in the caller.
Prepare for this by adding sock_l4_sa() which takes a pre-constructed
socket address, and rewriting sock_l4() in terms of it.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
buf_size is set to sizeof(pkt_buf) by default. And it seems more correct
to provide the actual size of the buffer.
Later a buf_size of 0 will allow vhost-user mode to detect
guest memory buffers.
Signed-off-by: Laurent Vivier <lvivier@redhat.com>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
it was needed by a draft version of vhost-user, it is not needed
anymore.
Signed-off-by: Laurent Vivier <lvivier@redhat.com>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
As we are going to introduce the MODE_VU that will act like
the mode MODE_PASST, compare to MODE_PASTA rather than to add
a comparison to MODE_VU when we check for MODE_PASST.
Signed-off-by: Laurent Vivier <lvivier@redhat.com>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
We are going to introduce a variant of the function to use
vhost-user buffers rather than passt internal buffers.
Signed-off-by: Laurent Vivier <lvivier@redhat.com>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
This commit refactors the udp_update_hdr4() and udp_update_hdr6() functions
to improve code portability by replacing the udp_meta_t parameter with
more specific parameters for the IPv4 and IPv6 headers (iphdr/ipv6hdr)
and the source socket address (sockaddr_in/sockaddr_in6).
It also moves the tap_hdr_update() function call inside the udp_tap_send()
function not to have to pass the TAP header to udp_update_hdr4() and
udp_update_hdr6()
This refactor reduces complexity by making the functions more modular and
ensuring that each function operates on more narrowly scoped data structures.
This will facilitate future backend introduction like vhost-user.
Signed-off-by: Laurent Vivier <lvivier@redhat.com>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
Consolidate pool_tap4() and pool_tap6() into tap_flush_pools(),
and tap4_handler() and tap6_handler() into tap_handler().
Create a generic tap_add_packet() to consolidate packet
addition logic and reduce code duplication.
The purpose is to ease the export of these functions to use
them with the vhost-user backend.
Signed-off-by: Laurent Vivier <lvivier@redhat.com>
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
Move all the TCP parts using internal buffers to tcp_buf.c
and keep generic TCP management functions in tcp.c.
Add tcp_internal.h to export needed functions from tcp.c and
tcp_buf.h from tcp_buf.c
With this change we can use existing TCP functions with a
different kind of memory storage as for instance the shared
memory provided by the guest via vhost-user.
Signed-off-by: Laurent Vivier <lvivier@redhat.com>
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
This commit isolates the internal data structure management used for storing
data (e.g., tcp4_l2_flags_iov[], tcp6_l2_flags_iov[], tcp4_flags_ip[],
tcp4_flags[], ...) from the tcp_send_flag() function. The extracted
functionality is relocated to a new function named tcp_fill_flag_header().
tcp_fill_flag_header() is now a generic function that accepts parameters such
as struct tcphdr and a data pointer. tcp_send_flag() utilizes this parameter to
pass memory pointers from tcp4_l2_flags_iov[] and tcp6_l2_flags_iov[].
This separation sets the stage for utilizing tcp_prepare_flags() to
set the memory provided by the guest via vhost-user in future developments.
Signed-off-by: Laurent Vivier <lvivier@redhat.com>
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
We have several functions which are used as callbacks for NS_CALL() which
only read their void * parameter, they don't write it. The
constParameterCallback warning in cppcheck 2.14.1 complains that this
parameter could be const void *, also pointing out that that would require
casting the function pointer when used as a callback.
Casting the function pointers seems substantially uglier than using a
non-const void * as the parameter, especially since in each case we cast
the void * to a const pointer of specific type immediately. So, suppress
these errors.
I think it would make logical sense to suppress this globally, but that
would cause unmatchedSuppression errors on earlier cppcheck versions. So,
instead individually suppress it, along with unmatchedSuppression in the
relevant places.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
Allow access to user_devpts.
$ pasta --version
pasta 0^20240510.g7288448-1.fc40.x86_64
...
$ awk '' < /dev/null
$ pasta --version
$
While this might be a awk bug it appears pasta should still have access
to devpts.
Signed-off-by: Derek Schrock <dereks@lifeofadishwasher.com>
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
Flow table entries need to be fully initialised before returning to the
main epoll loop. Commit 0060acd1 ("flow: Clarify and enforce flow state
transitions") now enforces that: once a flow is allocated we must either
cancel it, or activate it before returning to the main loop, or we will hit
an ASSERT().
Some error paths in tcp_conn_from_tap() weren't correctly updated for this
requirement - we can exit with a flow entry incompletely initialised.
Correct that by cancelling the flows in those situations.
I don't have enough information to be certain if this is the cause for
podman bug 22925, but it plausibly could be.
Fixes: 0060acd11b ("flow: Clarify and enforce flow state transitions")
Link: https://github.com/containers/podman/issues/22925
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
timespec_diff_ms() returns an int representing a duration in milliseconds.
This will overflow in about 25 days when an int is 32 bits. The way we
use this function, we're probably not going to get a result that long, but
it's not outrageously implausible. Use a long for safety.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
Functions and structures in lineread.c use plain int to record and report
the length of lines we receive. This means we truncate the result from
read(2) in some circumstances. Use ssize_t to avoid that.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
In conf() we parse a MAC address in two places, for the --ns-mac-addr and
the -M options. As well as duplicating code, the logic for this parsing
has several bugs:
* The most serious is that if the given string is shorter than a MAC
address should be, we'll access past the end of it.
* We don't check the endptr supplied by strtol() which means we could
ignore certain erroneous contents
* We never check the separator characters between each octet
* We ignore certain sorts of garbage that follow the MAC address
Correct all these bugs in a new parse_mac() helper.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
A negative bit index in a bitmap doesn't make sense. Avoid this by
construction by using unsigned indices. While we're there adjust
bitmap_isset() to return a bool instead of an int.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
We globally disabled this, with a justification lumped together with
several checks about braces. They don't really go together, the others
are essentially a stylistic choice which doesn't match our style. Omitting
brackets on macro parameters can lead to real and hard to track down bugs
if an expression is ever passed to the macro instead of a plain identifier.
We've only gotten away with the macros which trigger the warning, because
of other conventions its been unlikely to invoke them with anything other
than a simple identifier. Fix the macros, and enable the warning for the
future.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
The 'c' parameter is always passed exactly 'c'. The 'now' parameter is
always passed exactly 'now'.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
cppcheck 2.14.1 complains about the rport variable not being in as small
as scope as it could be. It's also only used once, so we might as well
just open code the calculation for it.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
The th pointer could be const, which causes a cppcheck warning on at least
some cppcheck versions (e.g. Cppcheck 2.13.0 in Fedora 40).
Fixes: e84a01e94c ("tcp: move seq_to_tap update to when frame is queued")
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
Now that we've simplified how usage() works, nothing ever sets the
log_to_stdout flag. Eliminate it.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
The message from usage() when given invalid options, or the -h / --help
option is currently printed by many calls to the info() function, also
used for runtime logging of informational messages.
That isn't useful: the usage message should always go to the terminal
(stdout or stderr), never syslog or a logfile. It should never be
filtered by priority. Really the only thing using the common logging
functions does is give more opportunities for something to go wrong.
Replace all the info() calls with direct fprintf() calls. This does mean
manually adding "\n" to each message. A little messy, but worth it for the
simplicity in other dimensions. While we're there make much heavier use
of single strings containing multiple lines of output text. That reduces
the number of fprintf calls, reducing visual clutter and making it easier
to see what the output will look like from the source.
Link: https://bugs.passt.top/show_bug.cgi?id=90
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
usage() does nothing but call print_usage() with EXIT_FAILURE as a
parameter. It's no more complex to just give that parameter at the single
call site. Eliminate it and rename print_usage() to just usage().
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
commit a469fc393f ("tcp, tap: Don't increase tap-side sequence counter for dropped frames")
delayed update of conn->seq_to_tap until the moment the corresponding
frame has been successfully pushed out. This has the advantage that we
immediately can make a new attempt to transmit a frame after a failed
trasnmit, rather than waiting for the peer to later discover a gap and
trigger the fast retransmit mechanism to solve the problem.
This approach has turned out to cause a problem with spurious sequence
number updates during peer-initiated retransmits, and we have realized
it may not be the best way to solve the above issue.
We now restore the previous method, by updating the said field at the
moment a frame is added to the outqueue. To retain the advantage of
having a quick re-attempt based on local failure detection, we now scan
through the part of the outqueue that had do be dropped, and restore the
sequence counter for each affected connection to the most appropriate
value.
Signed-off-by: Jon Maloy <jmaloy@redhat.com>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
Now:
- we don't open the PID file in main() anymore
- PID file and AF_UNIX socket are opened by pidfile_open() and
tap_sock_unix_open()
- write_pidfile() becomes pidfile_write()
Reported-by: Richard W.M. Jones <rjones@redhat.com>
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
Acked-by: Richard W.M. Jones <rjones@redhat.com>
We have pidfile_fd now, pid_file_fd would be quite ugly.
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
Reviewed-by: Richard W.M. Jones <rjones@redhat.com>
Otherwise, if the user runs us as root, and gives us paths that are
only accessible by root, we'll fail to open them, which might in turn
encourage users to change permissions or ownerships: definitely a bad
idea in terms of security.
Reported-by: Minxi Hou <mhou@redhat.com>
Reported-by: Richard W.M. Jones <rjones@redhat.com>
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
Acked-by: Richard W.M. Jones <rjones@redhat.com>
We won't call it from main() any longer: move it.
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
Reviewed-by: Richard W.M. Jones <rjones@redhat.com>
As I'm adding pidfile_open() in the next patch. The function comment
didn't match, by the way.
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
Reviewed-by: Richard W.M. Jones <rjones@redhat.com>
We'll need to open and bind the socket a while before listening to it,
so split that into two different functions. No functional changes
intended.
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
Reviewed-by: Richard W.M. Jones <rjones@redhat.com>
This is a remnant from the time we kept access to the original
filesystem and we could reinitialise the listening AF_UNIX socket.
Since commit 0515adceaa ("passt, pasta: Namespace-based sandboxing,
defer seccomp policy application"), however, we can't re-bind the
listening socket once we're up and running.
Drop the -1 initalisation and the corresponding check.
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
It has nothing to do with tap_sock_unix_init(). It used to be there as
that function could be called multiple times per passt instance, but
it's not the case anymore.
This also takes care of the fact that, with --fd, we wouldn't set the
initial MAC address, so we would need to wait for the guest to send us
an ARP packet before we could exchange data.
Fixes: 6b4e68383c ("passt, tap: Add --fd option")
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Acked-by: Richard W.M. Jones <rjones@redhat.com>
libguestfs tools have a good reason to run as root: if the guest image
is owned by root, it would be counterproductive to encourage users to
invoke them as non-root, as it would require changing permissions or
ownership of the image file.
And if they run as root, we'll start as root, too. Warn users we'll
switch to 'nobody', but don't tell them what to do.
Reported-by: Richard W.M. Jones <rjones@redhat.com>
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Reviewed-by: Richard W.M. Jones <rjones@redhat.com>
When we retrieve or copy host addresses we can include deprecated
addresses, which is not what we want. Adjust our logic to exclude them.
Similarly our tests can retrieve deprecated addresses, so exclude them
there too.
I hit this in practice because my router sometimes temporarily advertises
an fd00:: prefix before the real delegated IPv6 prefix. The deprecated
address can hang around for some time messing up my tests.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
We recently introduced this field to keep track of which side of a TCP flow
is the guest/tap facing one. Now that we generically record which pif each
side of each flow is connected to, we can easily derive that, and no longer
need to keep track of it explicitly.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
Currently we have no generic information flows apart from the type and
state, everything else is specific to the flow type. Start introducing
generic flow information by recording the pifs which the flow connects.
To keep track of what information is valid, introduce new flow states:
INI for when the initiating side information is complete, and TGT for
when both sides information is complete, but we haven't chosen the
flow type yet. For now, these states don't do an awful lot, but
they'll become more important as we add more generic information.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
Each flow in the flow table has two sides, 0 and 1, representing the
two interfaces between which passt/pasta will forward data for that flow.
Which side is which is currently up to the protocol specific code: TCP
uses side 0 for the host/"sock" side and 1 for the guest/"tap" side, except
for spliced connections where it uses 0 for the initiating side and 1 for
the target side. ICMP also uses 0 for the host/"sock" side and 1 for the
guest/"tap" side, but in its case the latter is always also the initiating
side.
Make this generically consistent by always using side 0 for the initiating
side and 1 for the target side. This doesn't simplify a lot for now, and
arguably makes TCP slightly more complex, since we add an extra field to
the connection structure to record which is the guest facing side. This is
an interim change, which we'll be able to remove later.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>q
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
Flows move over several different states in their lifetime. The rules for
these are documented in comments, but they're pretty complex and a number
of the transitions are implicit, which makes this pretty fragile and
error prone.
Change the code to explicitly track the states in a field. Make all
transitions explicit and logged. To the extent that it's practical in C,
enforce what can and can't be done in various states with ASSERT()s.
While we're at it, tweak the docs to clarify the restrictions on each state
a bit.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
This adds some extra inany helpers for comparing an inany address to
addresses of a specific family (including special addresses), and building
an inany from an IPv4 address (either statically or at runtime).
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
The flow dispatches deferred and timer handling for flows centrally, but
needs to call into protocol specific code for the handling of individual
flows. Currently this passes a general union flow *. It makes more sense
to pass the specific relevant flow type structure. That brings the check
on the flow type adjacent to casting to the union variant which it tags.
Arguably, this is a slight abstraction violation since it involves the
generic flow code using protocol specific types. It's already calling into
protocol specific functions, so I don't think this really makes any
difference.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
When reporting errors, we sometimes want to show a relevant socket address.
Doing so by extracting the various relevant fields can be pretty awkward,
so introduce a sockaddr_ntop() helper to make it simpler. For now we just
have one user in tcp.c, but I have further upcoming patches which can make
use of it.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
Commit b686afa2 introduced the invalid apparmor rule
`mount options=(rw, runbindable) /,` since runbindable mount rules
cannot have a source.
Therefore running aa-logprof/aa-genprof will trigger errors (see
https://bugs.launchpad.net/ubuntu/+source/apparmor/+bug/2065685)
$ sudo aa-logprof
ERROR: Operation {'runbindable'} cannot have a source. Source = AARE('/')
This patch fixes it to the intended behavior.
Link: https://bugs.launchpad.net/ubuntu/+source/apparmor/+bug/2065685
Fixes: b686afa23e ("apparmor: Explicitly pass options we use while remounting root filesystem")
Signed-off-by: Maxime Bélair <maxime.belair@canonical.com>
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
For some unknown reason "owner" makes it impossible to open bind mounted
netns references as apparmor denies it. In the kernel denied log entry
we see ouid=0 but it is not clear why that is as the actual file is
owned by the real (rootless) user id.
In abstractions/pasta there is already `@{run}/user/@{uid}/**` without
owner set for the same reason as this path contains the netns path by
default when running under Podman.
Fixes: 72884484b0 ("apparmor: allow read access on /tmp for pasta")
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
clang-tidy 18.1.1 in Fedora 40 complains about every #define of an integral
value, suggesting it be converted to an enum. Although that's certainly
possible, it's of dubious value and results in some awkward arrangements on
out codebase. Suppress it globally.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
In conf() we temporarily set the forwarding mode variables to 0 - an
invalid value, so that we can check later if they've been set by the
intervening logic. clang-tidy 18.1.1 in Fedora 40 now complains about
this. Satisfy it by giving an name in the enum to the 0 value.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
We take care of this in nl_addr_dup(): if the interface index
associated to an address doesn't match the selected host interface
(ifa->ifa_index != ifi_src), we don't copy that address.
But for routes, we just unconditionally update the interface index to
match the index in the target namespace, even if the source interface
didn't match.
This might happen in two cases: with a pre-4.20 kernel without support
for NETLINK_GET_STRICT_CHK, which won't filter routes based on the
interface we pass in the request, as reported by runsisi, and any
kernel with support for multipath routes where any of the nexthops
refers to an unrelated host interface.
In both cases, check the index of the source interface, and avoid
copying unrelated routes.
Reported-by: runsisi <runsisi@hust.edu.cn>
Link: https://bugs.passt.top/show_bug.cgi?id=86
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
Tested-by: runsisi <runsisi@hust.edu.cn>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
The podman CI on debian runs tests based on /tmp but pasta is failing
there because it is unable to open the netns path as the open for read
access is denied.
Link: https://github.com/containers/podman/issues/22625
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
