The server must check client reachability (we chose to do that by checking a
cookie) before destroying the existing association (RFC 6347 section 4.2.8).
Let's make sure we do, by having a proxy-in-the-middle inject a ClientHello -
the server should notice, but not destroy the connection.
Signed-off-by: Manuel Pégourié-Gonnard <manuel.pegourie-gonnard@arm.com>
To prevent dropping the same message over and over again, the UDP proxy
test application programs/test/udp_proxy _logically_ maintains a mapping
from records to the number of times the record has already been dropped,
and stops dropping once a configurable threshold (currently 2) is passed.
However, the actual implementation deviates from this logical view
in two crucial respects:
- To keep the implementation simple and independent of
implementations of suitable map interfaces, it only counts how
many times a record of a given _size_ has been dropped, and
stops dropping further records of that size once the configurable
threshold is passed. Of course, this is not fail-proof, but a
good enough approximation for the proxy, and it allows to use
an inefficient but simple array for the required map.
- The implementation mixes datagram lengths and record lengths:
When deciding whether it is allowed to drop a datagram, it
uses the total datagram size as a lookup index into the map
counting the number of times a package has been dropped. However,
when updating this map, the UDP proxy traverses the datagram
record by record, and updates the mapping at the level of record
lengths.
Apart from this inconsistency, the introduction of the Connection ID
feature leads to yet another problem: The CID length is not part of
the record header but dynamically negotiated during (potentially
encrypted!) handshakes, and it is hence impossible for a passive traffic
analyzer (in this case our UDP proxy) to reliably parse record headers;
especially, it isn't possible to reliably infer the length of a record,
nor to dissect a datagram into records.
The previous implementation of the UDP proxy was not CID-aware and
assumed that the record length would always reside at offsets 11, 12
in the DTLS record header, which would allow it to iterate through
the datagram record by record. As mentioned, this is no longer possible
for CID-based records, and the current implementation can run into
a buffer overflow in this case (because it doesn't validate that
the record length is not larger than what remains in the datagram).
This commit removes the inconsistency in datagram vs. record length
and resolves the buffer overflow issue by not attempting any dissection
of datagrams into records, and instead only counting how often _datagrams_
of a particular size have been dropped.
There is only one practical situation where this makes a difference:
If datagram packing is used by default but disabled on retransmission
(which OpenSSL has been seen to do), it can happen that we drop a
datagram in its initial transmission, then also drop some of its records
when they retransmitted one-by-one afterwards, yet still keeping the
drop-counter at 1 instead of 2. However, even in this situation, we'll
correctly count the number of droppings from that point on and eventually
stop dropping, because the peer will not fall back to using packing
and hence use stable record lengths.
This commit adds the command line option 'bad_cid' to the UDP proxy
`./programs/test/udp_proxy`. It takes a non-negative integral value N,
which if not 0 has the effect of duplicating every 1:N CID records
and modifying the CID in the first copy sent.
This is to exercise the stacks documented behaviour on receipt
of unexpected CIDs.
It is important to send the record with the unexpected CID first,
because otherwise the packet would be dropped already during
replay protection (the same holds for the implementation of the
existing 'bad_ad' option).
ApplicationData records are not protected against loss by DTLS
and our test applications ssl_client2 and ssl_server2 don't
implement any retransmission scheme to deal with loss of the
data they exchange. Therefore, the UDP proxy programs/test/udp_proxy
does not drop ApplicationData records.
With the introduction of the Connection ID, encrypted ApplicationData
records cannot be recognized as such by inspecting the record content
type, as the latter is always set to the CID specific content type for
protected records using CIDs, while the actual content type is hidden
in the plaintext.
To keep tests working, this commit adds CID records to the list of
content types which are protected against dropping by the UDP proxy.
Previously, the UDP proxy could only remember one delayed message
for future transmission; if two messages were delayed in succession,
without another one being normally forwarded in between,
the message that got delayed first would be dropped.
This commit enhances the UDP proxy to allow to delay an arbitrary
(compile-time fixed) number of messages in succession.
MSVC rightfully complained that there was some conversion from `size_t`
to `unsigned int` that could come with a loss of data. This commit
re-types the corresponding struct field `ctx_buffer::len` to `size_t`.
Also, the function `ctx_buffer_append` has an integer return value
which is supposed to be the (positive) length of the appended data
on success, and a check is inserted that the data to be appended does
not exceed MAX_INT in length.
The UDP proxy corrupts application data at the end of the datagram. If
there are multiple DTLS records within the same datagram, this leads
to the wrong message being corrupted. This commit always corrupts the
beginning of the message to prevent this.
Overall, the UDP proxy needs reworking if it is supposed to reliably
support multiple records within a single datagram, because it
determines its actions from the type of the first record in the
current datagram only.
This commit provides the new option pack=TIME for the udp proxy
./programs/test/udp_proxy. If used, udp packets with the same
destination will be queued and concatenated for up to TIME
milliseconds before being delivered.
This is useful to test how mbed TLS's deals with multiple DTLS records
within a single datagram.
The library/net.c and its corresponding include/mbedtls/net.h file are
renamed to library/net_sockets.c and include/mbedtls/net_sockets.h
respectively. This is to avoid naming collisions in projects which also
have files with the common name 'net'.
* development: (100 commits)
Update Changelog for the mem-measure branch
Fix issues introduced when rebasing
Fix compile error in memory_buffer_alloc_selftest
Code cosmetics
Add curve25519 to ecc-heap.sh
Add curve25519 to the benchmark program
Fix compile issue when buffer_alloc not available
New script ecc-heap.sh
Fix unused variable issue in some configs
Rm usunused member in private struct
Add heap usage for PK in benchmark
Use memory_buffer_alloc() in benchmark if available
Only define mode_func if mode is enabled (CBC etc)
PKCS8 encrypted key depend on PKCS5 or PKCS12
Disable SRV_C for client measurement
Output stack+heap usage with massif
Enable NIST_OPTIM by default for config-suite-b
Refactor memory.sh
Adapt memory.sh to config-suite-b
Adapt mini-client for config-suite-b.h
...
Conflicts:
ChangeLog
include/polarssl/net.h
library/Makefile
library/error.c
library/ssl_tls.c
programs/Makefile
programs/ssl/ssl_client2.c
programs/ssl/ssl_server2.c
tests/Makefile
There seemed to be some race conditions with server closing its fd right after
sending HelloVerifyRequest causing the proxy to exit after a failed read.
