| Age | Commit message (Collapse) | Author |
|---|
|
Use the same RX buffer size which iproute2 uses as default, in order to
avoid (or rather, delay as much as possible) ENOBUFS conditions when
receiving a large burst of netlink notifications.
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
|
|
Once we're notified about pending data on the netlink event socket,
try to receive as much messages as possible in order to empty the socket
buffer. In contrast to the previous implementation which received one
message per epoll read notification, this avoids some premature ENOBUFS
situations when a huge burst of messages arrives, e.g. due to kernel
side GC runs on the neighbor table.
Since libnl's nl_recvmsgs*() functions do not expose the underlying
-1 error returned by the recvmsg() calls, we need to indirectly check
for errors through errno variable, so clear it out both before the
recvmsgs() call and after successfully completing a message reception
callback invocation.
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
|
|
Only store the callback reference in the registry once the listener
context was successfully initialized, to avoid keeping a lingering
entry when invalid parameters are passed.
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
|
|
Signed-off-by: Packet Please <pktpls@systemli.org>
|
|
Signed-off-by: Packet Please <pktpls@systemli.org>
|
|
Get rid of most __APPLE__ guards by introducing a central platform.c unit
providing drop-in replacements for missing APIs.
Also move system signal definitions into the new platform file to be able
to share them with the upcoming debug library.
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
|
|
Fixes maintaining the listener array
Signed-off-by: Felix Fietkau <nbd@nbd.name>
|
|
Arguments stored in the registry need an extra reference
Signed-off-by: Felix Fietkau <nbd@nbd.name>
|
|
Similar to nl80211 uloop listener
Signed-off-by: Felix Fietkau <nbd@nbd.name>
|
|
Some pecularities in the encoding of the IFLA_AF_SPEC attribute make it
unsuitable for table driven parsing/generation.
To solve this issue, introduce specific datatype handling for IFLA_AF_SPEC
and parse/generate the RTA depending on the address family of the containing
netlink message.
Also add some missing constants while we're at it.
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
|
|
Decode IFLA_STATS64 attribute and make contained counters available to ucode.
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
|
|
For certain operations, such as bringing up interfaces, it is required to
initialize the ifi_change mask in the ifinfomsg struct.
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
|
|
So far the NETLINK_GET_STRICT_CHK socket flag was only set on the implicit
socket creation performed during the first request and ignored for
subsequent ones which made it impossible to perform only some requests
with enabled strict checking.
Modify the logic to check the flag state for every request and change
it if needed. This allows performing both strict and non-strict requests
over the same connection.
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
|
|
rtnl: fix segmentation fault on parsing linkinfo RTA without data
|
|
rtnl: zero request message headers
|
|
rtnl: fix premature netlink reply receive abort
|
|
Some link types, such as veth, yield an IFLA_LINKINFO nla without an
embedded IFLA_INFO_DATA / INFLA_INFO_SLAVE_DATA nla which causes the
nla converter to dereference a NULL nla pointer.
Properly deal with such cases and check for the existence of the
child nla before attempting to parse it.
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
|
|
For route netlink request messages having a header struct, uc_nl_request()
invokes nlmsg_reserve() to reserve room for the struct data but the
nlmsg_reserve() function only zeroes additional alignment bytes, not the
actual reserved buffer space.
Extend the existing logic to explicitly zero out the reserved header space
in order to avoid sending uninitialized struct member values to the kernel.
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
|
|
The nl_recvmsgs() logic in uc_nl_request() incorrectly stopped reading
the socket before the netlink ACK message was handled for non-multipart
replies.
This caused subsequent requests to incorrectly receive the ACK of the
previous request, leading to a failure to receive the actual reply.
Fix this issue by continue reading the socket until either the finish
callback for multipart (dump) messages or the ack callback for non-
multipart messages was received.
This fix is basically the same as the one applied to the nl80211 module
in 54ef6c0 ("nl80211: fix premature netlink reply receive abort").
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
|
|
Some nested RTAs such as IFLA_INET_CONF do not contain actual sub-RTAs but
just an array of integers. Avoid calling a no-op `nla_parse()` for such
attributes to suppress the non-harmful leftover bytes warning emitted by
libnl.
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
|
|
This avoid triggering -Werror=unused-result compilation failures with
certain toolchains.
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
|
|
- Parse integer literals as unsigned numeric values in order to be able
to represent the entire unsigned 64bit value range
- Stop parsing minus-prefixed integer literals as negative numbers but
treat them as separate minus operator followed by a positive integer
instead
- Only store unsigned numeric constants in bytecode
- Rework numeric comparison logic to be able to handle full 64bit
unsigned integers
- If possible, yield unsigned 64 bit results for additions
- Simplify numeric value conversion API
- Compile code with -fwrapv for defined signed overflow semantics
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
|
|
# ucode -mrtnl -Rs 'printf("%.J", rtnl.request(rtnl.const.RTM_GETROUTE, 0, { dst: "8.8.8.8" }))'
{
"family": 2,
"tos": 0,
"protocol": 0,
"scope": 0,
"type": 1,
"flags": 512,
"dst": "8.8.8.8/32",
"oif": "onboard",
"gateway": "10.11.12.13",
"prefsrc": "10.11.12.7",
"cacheinfo": {
"clntref": 2,
"lastuse": 0,
"expires": 0,
"error": 0,
"used": 0,
"id": 0,
"ts": 0,
"tsage": 0
},
"table": 254,
"uid": 0
}
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
|
|
# ucode -mrtnl -Rs 'printf("%.J", rtnl.request(rtnl.const.RTM_GETLINK, 0, { dev: "lo" }))'
{
"family": 0,
"type": 772,
"dev": "lo",
"flags": 65609,
"address": "00:00:00:00:00:00",
"broadcast": "00:00:00:00:00:00",
"txqlen": 1000,
"mtu": 65536,
"carrier": true,
"linkmode": 0,
"operstate": 0,
"num_tx_queues": 1,
"num_rx_queues": 1,
"af_spec": {
"inet": {
"conf": {
"forwarding": 1,
"mc_forwarding": 0,
"proxy_arp": 0,
"accept_redirects": 1,
"secure_redirects": 1,
"send_redirects": 1,
"shared_media": 1,
"rp_filter": 0,
"accept_source_route": 1,
"bootp_relay": 0,
"log_martians": 0,
"tag": 0,
"arpfilter": 0,
"medium_id": 0,
"noxfrm": 1,
"nopolicy": 1,
"force_igmp_version": 0,
"arp_announce": 0,
"arp_ignore": 0,
"promote_secondaries": 0,
"arp_accept": 0,
"arp_notify": 0,
"accept_local": 0,
"src_vmark": 0,
"proxy_arp_pvlan": 0,
"route_localnet": 0,
"igmpv2_unsolicited_report_interval": 10000,
"igmpv3_unsolicited_report_interval": 1000,
"ignore_routes_with_linkdown": 0,
"drop_unicast_in_l2_multicast": 0,
"drop_gratuitous_arp": 0,
"bc_forwarding": 0
}
},
"inet6": {
"mode": 0,
"flags": 2147483648,
"conf": {
"forwarding": 0,
"hoplimit": 64,
"mtu6": 65536,
"accept_ra": 1,
"accept_redirects": 1,
"autoconf": 1,
"dad_transmits": 1,
"rtr_solicits": -1,
"rtr_solicit_interval": 4000,
"rtr_solicit_delay": 1000,
"use_tempaddr": -1,
"temp_valid_lft": 604800,
"temp_prefered_lft": 86400,
"regen_max_retry": 3,
"max_desync_factor": 600,
"max_addresses": 16,
"force_mld_version": 0,
"accept_ra_defrtr": 1,
"accept_ra_pinfo": 1,
"accept_ra_rtr_pref": 1,
"rtr_probe_interval": 60000,
"accept_ra_rt_info_max_plen": 0,
"proxy_ndp": 0,
"optimistic_dad": 0,
"accept_source_route": 0,
"mc_forwarding": 0,
"disable_ipv6": 0,
"accept_dad": -1,
"force_tllao": 0,
"ndisc_notify": 0,
"mldv1_unsolicited_report_interval": 10000,
"mldv2_unsolicited_report_interval": 1000,
"suppress_frag_ndisc": 1,
"accept_ra_from_local": 0,
"use_optimistic": 0,
"accept_ra_mtu": 1,
"stable_secret": 0,
"use_oif_addrs_only": 0,
"accept_ra_min_hop_limit": 1,
"ignore_routes_with_linkdown": 0,
"drop_unicast_in_l2_multicast": 0,
"drop_unsolicited_na": 0,
"keep_addr_on_down": 0,
"rtr_solicit_max_interval": 3600000,
"seg6_enabled": 0,
"seg6_require_hmac": 0,
"enhanced_dad": 1,
"addr_gen_mode": 0,
"disable_policy": 0,
"accept_ra_rt_info_min_plen": 0,
"ndisc_tclass": 0,
"rpl_seg_enabled": 0
}
}
},
"proto_down": false,
"group": 0,
"ifname": "lo"
}
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
|
|
Some netlink replies contain attributes with embeded payloads whose
length depends on the kernel version, e.g. IFLA_INET_CONF and
IFLA_INET6_CONF.
Deal with such cases by allowing the payloads to be shorter than
the expected size and by skipping struct members which would lead
to out-of-bounds accesses.
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
|
|
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
|
