Originally, the dcb utility only stopped receiving messages from a
socket when it found the attribute it was looking for. Cited commit
changed that, so that the utility will also stop when seeing an ACK
(NLMSG_ERROR message), by setting the NLM_F_ACK flag on requests.
This is problematic because it means a successful request will leave an
ACK in the socket receive buffer, causing the next request to bail
before reading its response.
Fix that by not stopping when finding the required attribute in a
response. Instead, stop on the subsequent ACK.
Fixes: 84c0369726 ("dcb: unblock mnl_socket_recvfrom if not message received")
Signed-off-by: Ido Schimmel <idosch@nvidia.com>
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Add new apptrust subcommand for the dcbnl apptrust extension object.
The apptrust command lets you specify a consecutive ordered list of
trusted selectors, which can be used by drivers to determine which
selectors are eligible (trusted) for packet prioritization, and in which
order.
Selectors are sent in a new nested attribute:
DCB_ATTR_IEEE_APP_TRUST_TABLE. The nest contains trusted selectors
encapsulated in either DCB_ATTR_IEEE_APP or DCB_ATTR_DCB_APP attributes,
for standard and non-standard selectors, respectively.
Example:
Trust selectors dscp and pcp, in that order
$ dcb apptrust set dev eth0 order dscp pcp
Signed-off-by: Daniel Machon <daniel.machon@microchip.com>
Reviewed-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: David Ahern <dsahern@kernel.org>
Add new pcp-prio parameter to the app subcommand, which can be used to
classify traffic based on PCP and DEI from the VLAN header. PCP and DEI
is specified in a combination of numerical and symbolic form, where 'de'
(drop-eligible) means DEI=1 and 'nd' (not-drop-eligible) means DEI=0.
Map PCP 1 and DEI 0 to priority 1
$ dcb app add dev eth0 pcp-prio 1nd:1
Map PCP 1 and DEI 1 to priority 1
$ dcb app add dev eth0 pcp-prio 1de:1
Internally, PCP and DEI is encoded in the protocol field of the dcb_app
struct. Each combination of PCP and DEI maps to a priority, thus needing
a range of 0-15. A well formed dcb_app entry for PCP/DEI
prioritization, could look like:
struct dcb_app pcp = {
.selector = DCB_APP_SEL_PCP,
.priority = 7,
.protocol = 15
}
For mapping PCP=7 and DEI=1 to Prio=7.
Also, three helper functions for translating between std and non-std APP
selectors, have been added to dcb_app.c and exposed through dcb.h.
Signed-off-by: Daniel Machon <daniel.machon@microchip.com>
Reviewed-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: David Ahern <dsahern@kernel.org>
Currently, the dcb command sinks to the kernel through the netlink
to obtain information. However, if the kernel fails to obtain infor-
mation or is not processed, the dcb command is suspended.
For example, if we don't implement dcbnl_ops->ieee_getpfc in the
kernel, the command "dcb pfc show dev eth1" will be stuck and subsequent
commands cannot be executed.
This patch adds the NLM_F_ACK flag to the netlink in mnlu_msg_prepare
to ensure that the kernel responds to user requests.
After the problem is solved, the execution result is as follows:
$ dcb pfc show dev eth1
Attribute not found: Success
Fixes: 67033d1c1c ("Add skeleton of a new tool, dcb")
Signed-off-by: Junxin Chen <chenjunxin1@huawei.com>
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
dcb, devlink, rdma, tipc and vdpa rely on libmnl to compile, so they
check for libmnl to be installed on their Makefiles.
This moves HAVE_MNL check from the tools to top-level Makefile, thus
avoiding to call their Makefiles if libmnl is not present.
Signed-off-by: Andrea Claudi <aclaudi@redhat.com>
Signed-off-by: David Ahern <dsahern@kernel.org>
Currently dcb decodes the response from "dcb app add" and "del" by
interpreting the returned attribute as u8. But the value stored there is
actually a negative errno value.
Additionally, "dcb app" currently shows two sets of messages, one in
dcb_set_attribute_attr_cb() where the issue is detected, and another as a
result of error return from that function.
The current state is as follows:
# dcb app add dev swp36 dscp-prio 20:2
Error when attempting to set attribute: Unknown error 239
Attribute write: No such file or directory
Fix the "unknown error" issue by correctly decoding the attribute as i8 and
negating it. Furthermore, set errno to that value, and let the top-level
"attribute write" error message show the correct message.
Initialize errno to 0 before the dcb_talk() dispatch, and make the error
print conditional on errno != 0. This way the few error messages that are
worth describing in the place where they are detected will not cause the
second error message to be printed.
The fixed reporting looks like this:
# dcb app add dev swp36 dscp-prio 20:2
Attribute write: File exists
Reported-by: Maksym Yaremchuk <maksymy@nvidia.com>
Signed-off-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
All the actual code exists, but we neglect to recognize "default-prio" as a
CLI key for selection of what to show.
Reported-by: Maksym Yaremchuk <maksymy@nvidia.com>
Signed-off-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Some installation of Clang are unhappy about the use of a hand-rolled
formatting strings, and emit warnings such as this one:
dcb.c:334:31: warning: format string is not a string literal
[-Wformat-nonliteral]
Rewrite the impacted code so that it always uses literal format strings.
Reported-by: Stephen Hemminger <stephen@networkplumber.org>
Signed-off-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
main() dinamically allocates dcb, but when dcb_help() is called it
returns without freeing it.
Fix this using a goto, as it is already done in the same function.
Fixes: 67033d1c1c ("Add skeleton of a new tool, dcb")
Signed-off-by: Andrea Claudi <aclaudi@redhat.com>
Reviewed-by: Petr Machata <me@pmachata.org>
Signed-off-by: David Ahern <dsahern@kernel.org>
dcb_cmd_app_show() is supposed to return EINVAL if an incorrect argument
is provided.
Fixes: 8e9bed1493 ("dcb: Add a subtool for the DCB APP object")
Signed-off-by: Andrea Claudi <aclaudi@redhat.com>
Reviewed-by: Petr Machata <me@pmachata.org>
Signed-off-by: David Ahern <dsahern@kernel.org>
In older distros we need bsd/stdlib.h but newer distro doesn't
need it. Also old distro will need libbsd-devel installed and newer
doesn't. To remove a possible dependency on libbsd-devel replace usage
of reallocarray to realloc.
dcb_app.c: In function ‘dcb_app_table_push’:
dcb_app.c:68:25: warning: implicit declaration of function ‘reallocarray’; did you mean ‘realloc’?
Fixes: 8e9bed1493 ("dcb: Add a subtool for the DCB APP object")
Signed-off-by: Roi Dayan <roid@nvidia.com>
Reviewed-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Since moving get_rate() and get_size() from tc to lib, on some
systems we fail to link because of missing math lib.
Move the functions that require math lib to their own c file
and add -lm to dcb that now use those functions.
../lib/libutil.a(utils.o): In function `get_rate':
utils.c:(.text+0x10dc): undefined reference to `floor'
../lib/libutil.a(utils.o): In function `get_size':
utils.c:(.text+0x1394): undefined reference to `floor'
../lib/libutil.a(json_print.o): In function `sprint_size':
json_print.c:(.text+0x14c0): undefined reference to `rint'
json_print.c:(.text+0x14f4): undefined reference to `rint'
json_print.c:(.text+0x157c): undefined reference to `rint'
Fixes: f3be0e6366 ("lib: Move get_rate(), get_rate64() from tc here")
Fixes: 44396bdfcc ("lib: Move get_size() from tc here")
Fixes: adbe5de966 ("lib: Move sprint_size() from tc here, add print_size()")
Signed-off-by: Roi Dayan <roid@nvidia.com>
Reviewed-by: Petr Machata <petrm@nvidia.com>
Tested-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
The Linux DCBX object is a 1-byte bitfield of flags that configure whether
the DCBX protocol is implemented in the device or in the host, and which
version of the protocol should be used. Add a tool to access the per-port
Linux DCBX object.
For example:
# dcb dcbx set dev eni1np1 host ieee
# dcb dcbx show dev eni1np1
host ieee
Signed-off-by: Petr Machata <me@pmachata.org>
Signed-off-by: David Ahern <dsahern@kernel.org>
DCB APP interfaces are standardized in 802.1q-2018, and allow configuration
of traffic prioritization rules based on several possible headers.
Add a dcb subtool for maintenance and display of the APP table. For
example:
# dcb app add dev eni1np1 dscp-prio 0:0 CS3:3 CS6:6
# dcb app show dev eni1np1
dscp-prio 0:0 CS3:3 CS6:6
# dcb app add dev eni1np1 dscp-prio CS3:4
# dcb app show dev eni1np1
dscp-prio 0:0 CS3:3 CS3:4 CS6:6
# dcb app replace dev eni1np1 dscp-prio CS3:5
# dcb app show dev eni1np1
dscp-prio 0:0 CS3:5 CS6:6
Signed-off-by: Petr Machata <me@pmachata.org>
Signed-off-by: David Ahern <dsahern@kernel.org>
Some DSCP values can be translated to symbolic names. That may not be
always desirable. Introduce a command-line option similar to other tools,
-N or --Numeric, to suppress this translation.
Signed-off-by: Petr Machata <me@pmachata.org>
Signed-off-by: David Ahern <dsahern@kernel.org>
The function dcb_get_attribute() assumes that the caller knows the exact
size of the looked-for payload. It also assumes that the response comes
wrapped in an DCB_ATTR_IEEE nest. The former assumption does not hold for
the IEEE APP table, which has variable size. The latter one does not hold
for DCBX, which is not IEEE-nested, and also for any CEE attributes, which
would come CEE-nested.
Factor out the payload extractor from the current dcb_get_attribute() code,
and put into a helper. Then rewrite dcb_get_attribute() compatibly in terms
of the new function. Introduce dcb_get_attribute_va() as a thin wrapper for
IEEE-nested access, and dcb_get_attribute_bare() for access to attributes
that are not nested.
Signed-off-by: Petr Machata <me@pmachata.org>
Signed-off-by: David Ahern <dsahern@kernel.org>
The function dcb_set_attribute() takes a fully-formed payload as an
argument. For callers that need to build a nested attribute, such as is the
case for DCB APP table, this is not great, because with libmnl, they would
need to construct a separate netlink message just to pluck out the payload
and hand it over to this function.
Currently, dcb_set_attribute() also always wraps the payload in an
DCB_ATTR_IEEE container, because that is what all the dcb subtools so far
needed. But that is not appropriate for DCBX in particular, and in fact a
handful other attributes, as well as any CEE payloads.
Instead, generalize this code by adding parameters for constructing a
custom payload and for fetching the response from a custom response
attribute. Then add dcb_set_attribute_va(), which takes a callback to
invoke in the right place for the nest to be built, and
dcb_set_attribute_bare(), which is similar to dcb_set_attribute(), but does
not encapsulate the payload in an IEEE container. Rewrite
dcb_set_attribute() compatibly in terms of the new functions.
Signed-off-by: Petr Machata <me@pmachata.org>
Signed-off-by: David Ahern <dsahern@kernel.org>
This to keep compatible with the major tools, ip and tc. Also
document the option in the man page, which was neglected.
Fixes: 67033d1c1c ("Add skeleton of a new tool, dcb")
Signed-off-by: Petr Machata <me@pmachata.org>
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
DCB socket buffer is allocated in dcb_init(), but never freed(). Free it
in dcb_fini().
Fixes: 67033d1c1c ("Add skeleton of a new tool, dcb")
Signed-off-by: Petr Machata <me@pmachata.org>
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
dcb currently sends all netlink messages with a type RTM_GETDCB, even the
set ones. Change to the appropriate type.
Fixes: 67033d1c1c ("Add skeleton of a new tool, dcb")
Signed-off-by: Petr Machata <me@pmachata.org>
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
DCBNL maxrate interfaces are an extension to the 802.1q DCB interfaces and
allow configuration of rate with which traffic in a given traffic class is
sent.
Add a dcb subtool to allow showing and tweaking of this per-TC maximum
rate. For example:
# dcb maxrate show dev eni1np1
tc-maxrate 0:25Gbit 1:25Gbit 2:25Gbit 3:25Gbit 4:25Gbit 5:25Gbit 6:100Gbit 7:25Gbit
Signed-off-by: Petr Machata <me@pmachata.org>
Signed-off-by: David Ahern <dsahern@gmail.com>
DCBNL buffer interfaces are an extension to the 802.1q DCB interfaces and
allow configuration of port headroom buffers.
Add a dcb subtool to allow showing and tweaking of buffer priority mapping
and buffer sizes. For example:
# dcb buf show dev eni1np1
prio-buffer 0:0 1:0 2:0 3:3 4:0 5:0 6:6 7:0
buffer-size 0:10000 1:0 2:0 3:70000 4:0 5:0 6:10000 7:0
total-size 221072
Signed-off-by: Petr Machata <me@pmachata.org>
Signed-off-by: David Ahern <dsahern@gmail.com>
Allow switching "dcb" into the ISO/IEC mode of units by passing -i.
Signed-off-by: Petr Machata <me@pmachata.org>
Signed-off-by: David Ahern <dsahern@gmail.com>
Allow selective display of statistical counters by passing -s.
Signed-off-by: Petr Machata <me@pmachata.org>
Signed-off-by: David Ahern <dsahern@gmail.com>
The DCB buffer object has a settable array of 32-bit quantities, and the
maxrate object of 64-bit ones. Adjust dcb_parse_mapping() and related
helpers to support 64-bit values in mappings, and add appropriate helpers.
Signed-off-by: Petr Machata <me@pmachata.org>
Signed-off-by: David Ahern <dsahern@gmail.com>
None, one, or many parameters can be given on the command line, but
the current synopsis allows only none or one. Fix it.
Signed-off-by: Petr Machata <me@pmachata.org>
Signed-off-by: David Ahern <dsahern@gmail.com>
"dcb ets show dev X help" currently shows full "ets" help instead of just
help for the show command. Fix it.
Signed-off-by: Petr Machata <me@pmachata.org>
Signed-off-by: David Ahern <dsahern@gmail.com>
getopt_long() currently includes "c" and "n" in the short option string.
These probably slipped in as a cut'n'paste, and are not actually accepted.
Remove them.
Signed-off-by: Petr Machata <me@pmachata.org>
Signed-off-by: David Ahern <dsahern@gmail.com>
The Linux DCB interface allows configuration of a broad range of
hardware-specific attributes, such as TC scheduling, flow control, per-port
buffer configuration, TC rate, etc. Add a new tool to show that
configuration and tweak it.
DCB allows configuration of several objects, and possibly could expand to
pre-standard CEE interfaces. Therefore the tool itself is a lean shell that
dispatches to subtools each dedicated to one of the objects.
Signed-off-by: Petr Machata <me@pmachata.org>
Signed-off-by: David Ahern <dsahern@gmail.com>
