Conor Walsh [Thu, 19 Nov 2020 10:16:59 +0000 (10:16 +0000)]
build: fix gcc warning requiring Wformat
On some CentOS/RHEL systems using gcc 8.3.1 to compile dpdk, gcc shows a
warning on every build step saying that -Wformat-nonliteral and
-Wformat-security warnings will be ignored unless -Wformat is
also specified as a compiler flag. When the build is run with -werror
the build will fail due to these warnings.
Exact warning returned:
cc1: error: -Wformat-nonliteral ignored without -Wformat
[-Werror=format-nonliteral]
cc1: error: -Wformat-security ignored without -Wformat
[-Werror=format-security]
cc1: all warnings being treated as errors
This patch adds the -Wformat flag to config/meson.build. The warning id
181 has also been suppressed in icc as icc was showing false positives
with -Wformat enabled.
Fixes: 524a0d5d66b9 ("build: enable extra warnings with meson") Cc: stable@dpdk.org Signed-off-by: Conor Walsh <conor.walsh@intel.com> Tested-by: Lingli Chen <linglix.chen@intel.com>
Nick Connolly [Sun, 22 Nov 2020 08:45:15 +0000 (08:45 +0000)]
eal/windows: fix linkage with MinGW
Linking with the 'pci' driver when building with MinGW on
Windows fails with undefined symbol 'GUID_DEVCLASS_NET'.
This occurs because devguid.h is included in rte_windows.h
before INITGUID is defined.
Move the include of devguid.h after the definition of INITGUID.
Fixes: b762221ac24f ("bus/pci: support Windows with bifurcated drivers") Cc: stable@dpdk.org Signed-off-by: Nick Connolly <nick.connolly@mayadata.io> Reviewed-by: Tal Shnaiderman <talshn@nvidia.com>
Yunjian Wang [Thu, 12 Nov 2020 09:31:15 +0000 (17:31 +0800)]
malloc: fix style in free list index computation
Cleanup code style issue reported by kernel checkpatch. As follows:
* ERROR:CODE_INDENT: code indent should use tabs where possible
* ERROR:SPACING: spaces required around that '?' (ctx:VxE)
* WARNING:INDENTED_LABEL: labels should not be indented
Fixes: b0489e7bca2f ("malloc: fix linear complexity") Cc: stable@dpdk.org Signed-off-by: Yunjian Wang <wangyunjian@huawei.com> Acked-by: Anatoly Burakov <anatoly.burakov@intel.com>
Gregory Etelson [Tue, 17 Nov 2020 18:49:40 +0000 (20:49 +0200)]
net/mlx5: fix representor interrupts handler
Representor is a port in DPDK that is connected to a VF in such a way
that assuming there are no offload flows, each packet that is sent
from the VF will be received by the corresponding representor. While
each packet that is sent to a representor will be received by the VF.
This is very useful in case of SRIOV mode, where the first packet that
is sent by the VF will be received by the DPDK application which will
decide if this flow should be offloaded to the E-Switch.
Representor shares interrupts handler with host PF over the PCI
address. Therefore, after PF completes its interrupts handler
initialization, no additional actions required for representor.
Gregory Etelson [Tue, 17 Nov 2020 18:49:39 +0000 (20:49 +0200)]
examples/l3fwd-power: check packet types after start
l3fwd-power uses `--parse-ptype' parameter to query egress packets
type. Before that feature is enabled l3fwd-power verifies PMD ability
to advertise supported packet types with
rte_eth_dev_get_supported_ptypes().
The above ethdev function API contains this note:
"
Better to invoke this API after the device is already started or rx burst
function is decided, to obtain correct supported ptypes.
"
The patch updates code for PMDs that register supported types after
dev_start.
Andrey Vesnovaty [Sun, 22 Nov 2020 14:49:51 +0000 (16:49 +0200)]
net/mlx5: fix shared RSS action update
The shared RSS action update was not operational due to lack
of kernel driver support of TIR object modification.
This commit introduces the workaround to support shared RSS
action modify using an indirect queue table update instead of
touching TIR object directly.
Limitations: the only supported RSS property to update is queues, the
rest of the properties ignored.
Fixes: d2046c09aa64 ("net/mlx5: support shared action for RSS") Signed-off-by: Andrey Vesnovaty <andreyv@nvidia.com> Acked-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
Changing the allocation scheme to improve mbufs locality caused mbufs
overrun in some cases. Revert the previous replenish logic back.
Calculate a number of unused mbufs and replenish max this number of mbufs.
Mark the last 4 mbufs as fake mbufs to prevent overflowing into consumed
mbufs in the future. Keep the consumed index and the produced index 4 mbufs
apart for this purpose.
Replenish some mbufs only in case the consumed index is within the
replenish threshold of the produced index in order to retain the cache
locality for the vectorized MPRQ routine.
Benoît Ganne [Sun, 22 Nov 2020 10:04:04 +0000 (12:04 +0200)]
net/mlx5: allow unknown link speed
mlx5 PMD refuses to update link state if link speed is defined but
status is down or if link speed is undefined but status is up, even if
the ioctl() succeeded.
This prevents application to detect link up/down event, especially when
the link speed is not correctly detected.
Commit [1] allowed returning unknown link speed, so now PMD allows
the return of unknown link speed in the above case.
Due to some old kernel driver bug, link speed wasn't detected properly.
Xiaoyu Min [Mon, 16 Nov 2020 07:55:15 +0000 (15:55 +0800)]
app/flow-perf: fix raw encapsulation size
The rte_flow_item_eth and rte_flow_item_vlan items are refined.
The structs do not exactly represent the packet bits captured on the
wire anymore so add_*_header functions should use real header instead of
the using rte_flow_item_* struct.
Replace the rte_flow_item_* with the existing corresponding rte_*_hdr.
Fixes: 09315fc83861 ("ethdev: add VLAN attributes to ethernet and VLAN items") Signed-off-by: Xiaoyu Min <jackmin@nvidia.com>
Dekel Peled [Mon, 16 Nov 2020 07:55:18 +0000 (15:55 +0800)]
net/softnic: fix header size calculation
The rte_flow_item_eth and rte_flow_item_vlan items were updated in [1].
The rte_flow_item_ipv6 item was updated in [2].
The structs now contain additional metadata following the header data.
The size to use for match should be the header data size only, and
not the size of the whole struct.
This patch replaces the rte_flow_item_* with the corresponding rte_*_hdr.
Fixes: 09315fc83861 ("ethdev: add VLAN attributes to ethernet and VLAN items") Fixes: ad976bd40d28 ("ethdev: add extensions attributes to IPv6 item") Signed-off-by: Dekel Peled <dekelp@nvidia.com> Acked-by: Jasvinder Singh <jasvinder.singh@intel.com>
Xiaoyu Min [Mon, 16 Nov 2020 07:55:14 +0000 (15:55 +0800)]
net/mlx5: fix encap/decap limit for hairpin flow split
The rte_flow_item_eth and rte_flow_item_vlan items are refined.
The structs do not exactly represent the packet bits captured on the
wire anymore.
Should use real header instead of the whole struct.
Replace the rte_flow_item_* with the existing corresponding rte_*_hdr.
Fixes: 09315fc83861 ("ethdev: add VLAN attributes to ethernet and VLAN items") Fixes: f9210259cac7 ("net/mlx5: fix raw encap/decap limit") Signed-off-by: Xiaoyu Min <jackmin@nvidia.com> Acked-by: Matan Azrad <matan@nvidia.com>
The rte_flow_item_eth and rte_flow_item_vlan items are refined.
The structs do not exactly represent the packet bits captured on the
wire anymore so should only copy real header instead of the whole struct.
Replace the rte_flow_item_* with the existing corresponding rte_*_hdr.
Fixes: 09315fc83861 ("ethdev: add VLAN attributes to ethernet and VLAN items") Signed-off-by: Xiaoyu Min <jackmin@nvidia.com> Acked-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
Michael Baum [Wed, 18 Nov 2020 17:00:11 +0000 (17:00 +0000)]
regex/mlx5: fix leak on queue setup failure
In regex QP setup, the PMD creates some SQ objects.
When SQ object creation is failed, the previous SQ objects memory were
not freed what caused a memory leak.
Free them.
Fixes: 54fa1f6a67d7 ("regex/mlx5: add teardown for fastpath buffers") Signed-off-by: Michael Baum <michaelba@nvidia.com> Acked-by: Ori Kam <orika@nvidia.com>
Michael Baum [Wed, 18 Nov 2020 17:00:10 +0000 (17:00 +0000)]
regex/mlx5: fix type of setup constants
The constant representing the size of the metadata is defined as a
regular number (32-bit signed), even though all of its uses request an
unsigned int variable.
Similarly the constant representing the maximal output is also defined
as a regular number, even though all of its uses request an unsigned int
variable.
Change the type of the above constants to unsigned.
Fixes: 5f41b66d12cd ("regex/mlx5: setup fast path") Cc: stable@dpdk.org Signed-off-by: Michael Baum <michaelba@nvidia.com> Acked-by: Ori Kam <orika@nvidia.com>
Michael Baum [Wed, 18 Nov 2020 17:00:09 +0000 (17:00 +0000)]
regex/mlx5: improve error messages in RXP rules flush
During the rules flush, the rxp_poll_csr_for_value function is called
twice. The rxp_poll_csr_for_value function can fail for two reasons:
1. It could not read the value from register, in which case the
function returns -1.
2. It read a value, but not the value it expected to receive. In this
case it returns -EBUSY.
When the function fails it prints an error message that is relevant only
for a second type of failure. Moreover, for failure of the first type it
prints a value of an uninitialized variable.
In case of success, the function prints a debug message about the number
of cycles it took. This line was probably copied by mistake, since the
variable it reads from, is always equal to 0 and is not an indicator of
the number of cycles.
Remove the incorrect line about the cycles, and reduce the error print
only for the relevant error.
Fixes: b34d816363b5 ("regex/mlx5: support rules import") Cc: stable@dpdk.org Signed-off-by: Michael Baum <michaelba@nvidia.com> Acked-by: Ori Kam <orika@nvidia.com>
Michael Baum [Wed, 18 Nov 2020 17:00:08 +0000 (17:00 +0000)]
regex/mlx5: check DevX register write
The rxp_flush_rules function tries to read and write to the register
several times using DevX API, and when it fails the function returns an
error.
Similarly the rxp_init_eng function also tries to write to the register
several times, and if writing is failed, it returns an error too.
Both functions have one write that the function does not check if it
succeeded, overriding the return value from the write function without
using it.
Add a check for this writing, and return an error in case of failure.
Fixes: b34d816363b5 ("regex/mlx5: support rules import") Fixes: e3dbbf718ebc ("regex/mlx5: support configuration") Cc: stable@dpdk.org Signed-off-by: Michael Baum <michaelba@nvidia.com> Acked-by: Ori Kam <orika@nvidia.com>
Michael Baum [Wed, 18 Nov 2020 17:00:07 +0000 (17:00 +0000)]
regex/mlx5: remove unnecessary initializations
The rxp_poll_csr_for_value function defines a variable named ret. It is
the return value of the function, and it is updated to 0 by default
later in the function.
Similarly the rxp_init_rtru function also defines a variable named ret.
The function assigns into it return values from functions during the
function.
In both functions they initialize the ret variable when defining it.
however, in both cases they do not use any ret variable before assigning
into them different values, so the initializations are unnecessary.
Clean the aforementioned unnecessary initializations.
Fixes: e3dbbf718ebc ("regex/mlx5: support configuration") Fixes: b34d816363b5 ("regex/mlx5: support rules import") Cc: stable@dpdk.org Signed-off-by: Michael Baum <michaelba@nvidia.com> Acked-by: Ori Kam <orika@nvidia.com>
Michael Baum [Wed, 18 Nov 2020 17:00:06 +0000 (17:00 +0000)]
regex/mlx5: fix RXP engines iterator size
The mlx5_regex_rules_db_import function goes over all engines in the
loop and program rxp rules.
The iterator of the loop is called id and the variable representing the
number of engines is called priv->nb_engines.
The id variable is of uint8_t type while the priv->nb_engines variable
is of uint32_t type. The size of the priv->nb_engines variable is much
larger than the number of iterations allowed by the id type.
Theoretically there may be a situation where the value of the
priv->nb_engines will be greater than can be represented by 8 bits and
the loop will never end.
Change the type of id to uint32_t.
Fixes: b34d816363b5 ("regex/mlx5: support rules import") Cc: stable@dpdk.org Signed-off-by: Michael Baum <michaelba@nvidia.com> Acked-by: Ori Kam <orika@nvidia.com>
Michael Baum [Wed, 18 Nov 2020 17:00:05 +0000 (17:00 +0000)]
regex/mlx5: fix crash on probe failure
The mlx5_regex_pci_probe function allocates a mlx5_regex_priv structure
using rte_zmalloc.
If the allocation fails, the function jumps to the dev_error label in
order to release previously allocated resources in the function.
However, in the dev_error label it attempts to refer to the internal
fields of the priv structure and if its allocation fails (as in this
case) it is actually dereferencing to NULL.
Replace the jump with an error label.
Fixes: 1db6ebd4ef58 ("regex/mlx5: fix crash on initialization failure") Cc: stable@dpdk.org Signed-off-by: Michael Baum <michaelba@nvidia.com> Acked-by: Ori Kam <orika@nvidia.com>
Shijith Thotton [Thu, 19 Nov 2020 08:46:11 +0000 (14:16 +0530)]
event/octeontx2: unlink queues during port release
Unlinking queues from port should be done during port release. Doing it
during device re-configuration could result in segfault as ports array
is re-allocated based on new number of ports.
Fixes: f7ac8b66b23c ("event/octeontx2: support linking queues to ports") Cc: stable@dpdk.org Signed-off-by: Shijith Thotton <sthotton@marvell.com> Signed-off-by: Pavan Nikhilesh <pbhagavatula@marvell.com>
Pavan Nikhilesh [Fri, 20 Nov 2020 10:41:46 +0000 (16:11 +0530)]
event/octeontx2: fix unconditional Tx flush
Fix unconditional Tx flush, in case of Tx only we need to check if
work slot is non-empty before issuing flush.
Also, in packet retransmit cases added check for the reference
count and flush the work slot only for the last packet.
Chengchang Tang [Fri, 20 Nov 2020 11:27:34 +0000 (19:27 +0800)]
net/hns3: fix HW exception for unbalanced Rx/Tx queues
For kupeng 930, there are 3 registers to control the enable
status of a TQP(i.e. task queue pair, include a txq and a rxq).
One of them controls whether the TQP is enabled, and the other
two controls whether the rxq and txq are enabled. The registers
used to control the enabled status of the rxq and txq are enabled
by default. Therefore, after the TQP is enabled, the rxq and txq
are enabled by default.
Currently, when the number of rxq is not equal to the number of txq, the
unused rxqs or txqs are not disabled by driver, so these unused queues
will be enabled in this situation. And the related HW rings have not
been initialized which could lead to a hardware exception.
This patch fix it by disable these unused queues during enable the TQPs.
Fixes: fa29fe45a7b4 ("net/hns3: support queue start and stop") Cc: stable@dpdk.org Signed-off-by: Chengchang Tang <tangchengchang@huawei.com> Signed-off-by: Lijun Ou <oulijun@huawei.com>
Huisong Li [Fri, 20 Nov 2020 11:27:33 +0000 (19:27 +0800)]
net/hns3: fix crash with multi-TC
The HW and SW DCB configurations need to be updated only
after the DCB configuration information changed. But the
change of tx/rx queue number is ignored. If user decreases
the number of tx queue after configuring multi-TC, the queue
mapping information in hns3_tc_queue_info can not be updated.
And then accessing the released queue resource in
"hns3_init_tx_ring_tc" will trigger a segment fault.
Fixes: 62e3ccc2b94c ("net/hns3: support flow control") Cc: stable@dpdk.org Signed-off-by: Huisong Li <lihuisong@huawei.com> Signed-off-by: Lijun Ou <oulijun@huawei.com>
Thomas Monjalon [Fri, 20 Nov 2020 09:54:38 +0000 (10:54 +0100)]
ethdev: rename a flow shared action error code
In the experimental function rte_flow_shared_action_destroy()
introduced in DPDK 20.11, the errno ETOOMANYREFS was used.
This errno is not always available on Windows,
so it is preferred using EBUSY instead.
Signed-off-by: Thomas Monjalon <thomas@monjalon.net> Reviewed-by: Tal Shnaiderman <talshn@nvidia.com> Tested-by: Tal Shnaiderman <talshn@nvidia.com> Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>
Guoyang Zhou [Fri, 20 Nov 2020 09:11:29 +0000 (17:11 +0800)]
net/hinic: fix hugepage memory leaks
The 'nic_io' memory is set to NULL before it is freed, this is wrong,
and the freeing of some private port resources should be moved from the
".remove(device)" function to the ".dev_close(port)" function.
Fixes: f30e69b41f94 ("ethdev: add device flag to bypass auto-filled queue xstats") Cc: stable@dpdk.org Signed-off-by: Guoyang Zhou <zhouguoyang@huawei.com>
Dekel Peled [Wed, 18 Nov 2020 08:59:56 +0000 (10:59 +0200)]
net/mlx5: fix input register for ASO object
Existing code uses the hard-coded value REG_C_5 as input for function
mlx5dv_dr_action_create_flow_hit().
This patch updates function mlx5_flow_get_reg_id() to return the
selected REG_C value for ASO Flow Hit operation.
The returned value is used, after reducing offset REG_C_0, as input
for function mlx5dv_dr_action_create_flow_hit().
Fixes: f935ed4b645a ("net/mlx5: support flow hit action for aging") Signed-off-by: Dekel Peled <dekelp@nvidia.com> Acked-by: Matan Azrad <matan@nvidia.com>
Dekel Peled [Wed, 18 Nov 2020 08:59:55 +0000 (10:59 +0200)]
net/mlx5: fix memory leak on ASO age close
Recent patch introduced the use of ASO flow hit action for age action.
The relevant management struct uses dynamically allocated memory.
This memory was not freed on closing.
This patch adds memory freeing as needed.
Fixes: f935ed4b645a ("net/mlx5: support flow hit action for aging") Signed-off-by: Dekel Peled <dekelp@nvidia.com> Acked-by: Matan Azrad <matan@nvidia.com>
Gregory Etelson [Thu, 19 Nov 2020 11:10:24 +0000 (13:10 +0200)]
app/testpmd: fix flow tunnel commands
testpmd provides commands to test tunnel offload rte_flow
capabilities. Testpmd tunnel commands allow to configure new ofloaded
tunnel types, list existing offloaded tunnels and destroy existing
offloaded tunnels.
Tunnel offload commands allowed parameters repetition. For example,
the following commands were accepted:
testpmd> flow tunnel create 0 create 1 type vxlan
or
testpmd> flow tunnel list 0 list 1
Current patch fixed that fault. Correct tunnel commands syntax is:
testpmd> flow tunnel create <port> type <tunnel type>
testpmd> flow tunnel list <port>
testpmd> flow tunnel destroy <port> id <tunnel id>
Fixes: 1b9f274623b8 ("app/testpmd: add commands for tunnel offload") Signed-off-by: Gregory Etelson <getelson@nvidia.com> Acked-by: Ori Kam <orika@nvidia.com> Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>
Gregory Etelson [Wed, 18 Nov 2020 18:23:59 +0000 (20:23 +0200)]
net/mlx5: fix restore info in non-tunnel traffic
Tunnel offload API provides applications with ability to restore
packet outer headers after partial offload. Exact feature execution
depends on hardware abilities and PMD implementation. Hardware that is
supported by MLX5 PMD places a mark on a packet after partial offload.
PMD decodes that mark and provides application with required
information.
Application can call the restore API for packets that are part of
offloaded tunnel and not. It's up to a PMD to provide correct
information.
Current MLX5 tunnel offload implementation does not allow applications
to use flow MARK actions. It is restricted to tunnel offload use only.
This fault was triggered by application that did not activate tunnel
offload and called the restore API with a marked packet. The PMD tried
to decode the mark value and crashed. The patch decodes mark value
only if tunnel offload is active.
Bing Zhao [Thu, 19 Nov 2020 02:41:32 +0000 (10:41 +0800)]
net/mlx5: fix eCPRI item value with mask
When creating a flow with eCPRI item, the mask and the value are both
needed in order to build the matching criteria.
In the current implementation, the unused value bits clear operation
was missed when filling the mask and value fields. For the value, the
bits not required were not masked with the mask provided. Indeed,
this action is not mandatory. But when creating a flow in the root
table, the kernel driver got involved and a check would prevent this
flow from being created. The same flow could be created successfully
with the userspace rdma-core on the non-root tables.
An AND operation needs to be added to clear the unused bits in the
value when building the matching criteria. Then the same flow can be
created successfully no matter with kernel driver or with rdma-core.
Suanming Mou [Thu, 19 Nov 2020 01:19:47 +0000 (09:19 +0800)]
net/mlx5: fix sample and mirror flow action deletion
The sample and mirror action objects are maintained on the list
shared between the ports belonging to the same multiport Infiniband
device(between representors).
The actions in the NIC steering domains might contain the references
to the sub-flow action objects created over the given port. The action
deletion might happen in the context of the different port and on the
deletion of referenced objects the incorrect port might be specified.
To avoid this we should save the port on what the sub-flow actions
were created and then use this saved port for sub-flow action release.
This commit saves the create device in the sample and mirror actions
struct to avoid using the incorrect port device in releasing.
Fixes: 19784141692e ("net/mlx5: make sample and mirror action thread safe") Signed-off-by: Suanming Mou <suanmingm@nvidia.com> Reviewed-by: Jiawei Wang <jiaweiw@nvidia.com> Acked-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
Suanming Mou [Wed, 18 Nov 2020 02:20:24 +0000 (10:20 +0800)]
net/mlx5: fix header reformat action hash key
Currently, header reformat action uses the hash list 32-bit key
generated in header reformat register function directly. The key will
not be recalculated in the hash list function.
As the 64-bit key is composed of the 32-bit attributes and 32-bit
reformat buffer csum, the hash list function only gets 32-bit key
directly will take the attribute part only, csum part will be ignored.
For different header reformat actions, the attributes can be the same,
while the buffer will be different. Only take the attribute part causes
lots of the conflicts.
This commits adds the attribute part and the significant different csum
part for the key.
The MLX5_ENCAPSULATION_DECISION_SIZE constant is used
to check the raw encap/decap actions for the raw header
size. The header is constructed of the rte_xxx_hdr
structures instead of rte items. Hence, constant
must be defined with rte_xxx_hdr structure sizes.
Fixes: 50f576d657d7 ("net/mlx5: fix VLAN actions in meter") Cc: stable@dpdk.org Signed-off-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com> Acked-by: Ori Kam <orika@nvidia.com> Acked-by: Suanming Mou <suanmingm@nvidia.com>
Xueming Li [Tue, 17 Nov 2020 11:01:38 +0000 (11:01 +0000)]
net/mlx5: set representor to first PF in bonding mode
When the representor device was set to PF1 in bonding mode, iterating
device iterator that looking for representors by bonding device failed
to match PF0 pci address with PF1 address. So detaching PF bonding
device only detached all representors on PF0.
This patch registers all representors of PF1 with PF0 as PCI device.
Signed-off-by: Xueming Li <xuemingl@nvidia.com> Reviewed-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
net/mlx5: fix WQE counter assert in free completion queue
The following assertion fails in case RTE_ENABLE_ASSERT is enabled:
PANIC in mlx5_tx_handle_completion():
assert "(txq->fcqs[txq->cq_ci & txq->cqe_m] >> 16)
== cqe->wqe_counter" failed
The free completion queue only contains an expected WQE counter if
RTE_LIBRTE_MLX5_DEBUG is enabled as well. Thus enabling
RTE_ENABLE_ASSERT alone causes the assert to fail.
Compile the assert conditionally only if RTE_ENABLE_ASSERT is enabled.
The kernel can use two approaches to distinguish the E-Switch
source vport in the packet metadata - either with dedicated
source_port field or register C0. To eliminate the extra source
vport matching in the hardware the source_port field can be
set to specific values (0xFFFF) for the wire source port.
This match can be applied to recognize wire port only in FDB
domain. Missing the register C0 match in the NIC Rx domain causes
incorrect representor steering within shared IB device ports
and must be always specified (if kernel uses this approach).
Harman Kalra [Mon, 16 Nov 2020 14:41:55 +0000 (20:11 +0530)]
common/octeontx2: fix mbox response by syncing data types
Syncing the data types of msix offset mbox response structure
fields wrt rvu_pfvf structure as in msix offset mbox handler,
respective block LFs response field is initialised with rvu_pfvf
and mismatch in the data types of respective fields may lead
to issues.
This mismatch in data types resulted in a bug where no sso msix
offsets are returned. For 256 SSO hwgrps, 256 msix offsets
are required, but since num of sso lf field (rsp->sso) in
response structure is u8 and is initialised with same field
in pfvf structure viz u16, value 256 corresponds to 0 hence
no sso_msixoff.
Similar mbox changes are required in kernel side also which are
available in marvell SDK version SDK-11.0.
Fixes: 2b71657c8660 ("common/octeontx2: add mbox request and response definition") Cc: stable@dpdk.org Signed-off-by: Harman Kalra <hkalra@marvell.com> Acked-by: Jerin Jacob <jerinj@marvell.com>
Didier Pallard [Tue, 17 Nov 2020 11:26:47 +0000 (12:26 +0100)]
net/mlx5: fix Rx descriptor status
Three bugs in rx_queue_count function:
- One entry may contain several segments, so 'used' must be multiplied
by number of segments per entry to properly reflect the queue usage.
- The number of cqes is equals to (1U << rxq->elts_n) - 1 in SPRQ mode.
The range returned by rx_queue_count should be the number of entries
used in queue, so it ranges from 0 to max number of entries
in queue, not this number minus one.
- For MPRQ mode, we need to take into account of the number of strd.
Maxime Leroy [Tue, 17 Nov 2020 11:26:46 +0000 (12:26 +0100)]
net/mlx5: fix Rx queue count calculation
The commit d2d57605522d ("net/mlx5: fix Rx queue count calculation") is
incorrect because the count calculation is wrong for the next cqe:
Example:
Compressed Set of packets 1 | Compressed Set of packets 2
C | a | e0 | e1 | e2 | e3 | e4 | e5 | C | a | e0
There are 2 compressed set of packets in the first queue. For the first
set, n is computed correctly.
But for the second, n is not computed properly. Because the zip context
is for the first set. The second set is not yet decompressed, so
there are no context.
To fix the issue, we should only use the zip context for the first CQEs
series.
Xiaoyu Min [Mon, 16 Nov 2020 02:07:59 +0000 (10:07 +0800)]
net/mlx5: fix RSS queue type validation
When the RSS queues' types are not uniformed, i.e, mixed with normal Rx
queue and hairpin queue, PMD accept this flow after commit[1] instead of
rejecting it.
This because commit[1] creates Rx queue object as DevX type via DevX API
instead of IBV type via Verbs, in which the latter will check the queues'
type when creating Verbs ind table but the former doesn't check when
creating DevX ind table.
However, in any case, logically PMD should check whether the input
configuration of RSS action is reasonable or not, which should
include queues' type check as well as the others.
So add the check of RSS queues' type in validation function to fix issue.
[1]:
commit 6deb19e1b2d2 ("net/mlx5: separate Rx queue object creations")
Fixes: 63bd16292c3a ("net/mlx5: support RSS on hairpin") Cc: stable@dpdk.org Signed-off-by: Xiaoyu Min <jackmin@nvidia.com> Acked-by: Matan Azrad <matan@nvidia.com>
Received packets can be aligned to the size of the cache line on
PCI transactions. This could improve performance by avoiding
partial cache line writes in exchange for increased PCI bandwidth.
This feature is supposed to be controlled by the rxq_pkt_pad_en
devarg and it is true for an RxQ created via the Verbs API.
But in the DevX API case, it is erroneously controlled by the
rxq_cqe_pad_en devarg instead, which is in charge of the CQE
padding instead and should not control the RxQ creation.
Fix DevX RxQ creation by using the proper configuration flag for
Rx packet padding that is being set by the rxq_pkt_pad_en devarg.
Fixes: dc9ceff73c99 ("net/mlx5: create advanced RxQ via DevX") Cc: stable@dpdk.org Signed-off-by: Alexander Kozyrev <akozyrev@nvidia.com> Acked-by: Matan Azrad <matan@nvidia.com>
Gregory Etelson [Mon, 16 Nov 2020 14:02:23 +0000 (16:02 +0200)]
net/mlx5: fix crash in tunnel offload setup
The new flow table resource management API triggered a PMD crash in
tunnel offload mode, when tunnel match flow rule was inserted before
tunnel set rule.
Reason for the crash was double flow table registration. The table was
registered by the tunnel offload code for the first time and once
more by PMD code, as part of general table processing. The table
counter was decremented only once during the rule destruction and
caused a resource leak that triggered the crash.
The patch updates PMD registration with tunnel offload parameters and
removes table registration in tunnel related code.
The original patch was removing active tunnel offload objects from a
tunnels db list without checking its reference counter value.
That action was leading to a PMD crash.
Current patch isolates tunnels db list into a separate API. That API
manages MT protection of the tunnel offload db.
Fixes: 5b38d8cd4663 ("net/mlx5: make tunnel hub list thread safe") Signed-off-by: Gregory Etelson <getelson@nvidia.com> Acked-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
Gregory Etelson [Mon, 16 Nov 2020 14:02:21 +0000 (16:02 +0200)]
net/mlx5: fix tunnel offload object allocation
The original patch allocated tunnel offload objects with invalid
indexes. As the result, PMD tunnel object allocation failed.
In this patch indexed pool provides both an index and memory for a new
tunnel offload object.
Also tunnel offload ipool moved to dv enabled code only.
Fixes: 4ae8825c5085 ("net/mlx5: use indexed pool as id generator") Signed-off-by: Gregory Etelson <getelson@nvidia.com> Acked-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
Gregory Etelson [Mon, 16 Nov 2020 14:02:20 +0000 (16:02 +0200)]
net/mlx5: fix structure passing method in function call
Tunnel offload implementation introduced 64 bit-field flow_grp_info
structure. Since the structure size is 64 bits, the code passed that
type by value in function calls.
The patch changes that structure passing method to reference.
Gregory Etelson [Mon, 16 Nov 2020 14:02:19 +0000 (16:02 +0200)]
net/mlx5: fix build with Direct Verbs disabled
Tunnel offload API is implemented for Direct Verbs environment only.
Current patch re-arranges tunnel related functions for compilation in
non Direct Verbs setups to prevent compilation failures. The patch
does not introduce new functions.
Ivan Malov [Mon, 16 Nov 2020 19:40:29 +0000 (22:40 +0300)]
net/sfc: rely on native thread safety in flow mechanism
The changeset that introduced common flow API thread safety
in fact introduced double locking to this particular PMD as
RTE flow API implementation in the PMD has been thread-safe
since the day zero. State this by setting the corresponding
device flag to skip locking imposed by generic RTE flow API.
Fixes: 80d1a9aff7f6 ("ethdev: make flow API thread safe") Signed-off-by: Ivan Malov <ivan.malov@oktetlabs.ru> Acked-by: Andrew Rybchenko <andrew.rybchenko@oktetlabs.ru>
Ivan Malov [Mon, 16 Nov 2020 19:23:03 +0000 (22:23 +0300)]
net/sfc: fix double free on encap transfer flow rule path
Don't free the outer match spec by its pointer in the parsing context
if it has already been tracked by an entry in the outer rule registry.
Fixes: dadff137931c ("net/sfc: support encap flow items in transfer rules") Signed-off-by: Ivan Malov <ivan.malov@oktetlabs.ru> Acked-by: Andrew Rybchenko <andrew.rybchenko@oktetlabs.ru>
Bruce Richardson [Mon, 16 Nov 2020 09:33:37 +0000 (09:33 +0000)]
examples/multi_process: add SPDX license tag to makefiles
The makefiles for some of the multi_process example files were missing
SPDX license headers, so add them. These Makefiles completely replaced
the older makefiles and were written from scratch, not based on previous
versions, so add a new copyright year on them.
Fixes: 13abe17c3cd4 ("examples/multi_process: convert to pkg-config-based build") Suggested-by: Stephen Hemminger <stephen@networkplumber.org> Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
Thomas Monjalon [Tue, 17 Nov 2020 10:38:26 +0000 (11:38 +0100)]
devtools: rename build test verbosity variables
The verbosity was meant to be set with options -v and -vv,
or possibly with the environment variables TEST_MESON_BUILD_VERBOSE
and TEST_MESON_BUILD_VERY_VERBOSE.
It is decided to keep only the options -v and -vv,
so the variables are renamed with lower case, marking them as privates.
The handling of the verbosity level is also moved upper in the script,
closer to other initializations.
Signed-off-by: Thomas Monjalon <thomas@monjalon.net> Reviewed-by: David Marchand <david.marchand@redhat.com>
Anatoly Burakov [Thu, 19 Nov 2020 10:41:56 +0000 (10:41 +0000)]
doc: allow external build
Currently, when building sphinx documentation, the build will only
succeed if being run from the build system, because the conf.py script
expects DPDK_VERSION environment variable to be set, and crashes if it
is not.
However, there are certain external tools (such as sphinx documentation
preview extensions for certain IDE's) that use live preview and thus
rely on running their own sphinx commands. In these cases, it is useful
to permit building sphinx documentation without specifying the
DPDK_VERSION environment variable. The version string is the only thing
preventing manual sphinx build commands from working.
Fix the conf.py to use "None" as a version string in cases when
DPDK_VERSION environment variable is not set.
Signed-off-by: Anatoly Burakov <anatoly.burakov@intel.com> Acked-by: Bruce Richardson <bruce.richardson@intel.com>
Gaetan Rivet [Mon, 16 Nov 2020 10:12:12 +0000 (11:12 +0100)]
bus/pci: fix comment explaining device naming
The original triple negative was hard to read and the attempt
to improve the formulation was commendable, unfortunately the new
comment is the inverse of correct.
Fixes: a65a34a85ebf ("eal: replace usage of blacklist/whitelist in enums") Reported-by: David Marchand <david.marchand@redhat.com> Signed-off-by: Gaetan Rivet <grive@u256.net> Acked-by: Luca Boccassi <bluca@debian.org> Reviewed-by: David Marchand <david.marchand@redhat.com>
David Marchand [Mon, 16 Nov 2020 10:06:18 +0000 (11:06 +0100)]
bus/vdev: fix comment
RTE_DEV_WHITELISTED is now replaced with RTE_DEV_ALLOWED.
Fixes: a65a34a85ebf ("eal: replace usage of blacklist/whitelist in enums") Signed-off-by: David Marchand <david.marchand@redhat.com> Acked-by: Thomas Monjalon <thomas@monjalon.net>
eal: replace usage of blacklist/whitelist in enums
Rename the enum values in the EAL include files.
As a backward compatible temporary migration tool, define
a replacement mapping for old values.
The old names relating to blacklist and whitelist are replaced
by block list and allow list, but applications may be using the
older compatibility macros, marked as deprecated.
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org> Acked-by: Luca Boccassi <bluca@debian.org> Acked-by: Gaetan Rivet <grive@u256.net> Acked-by: Hemant Agrawal <hemant.agrawal@nxp.com> Signed-off-by: Thomas Monjalon <thomas@monjalon.net>
Fix the detection of instruction pattern with multiple emits followed
by TX. Once detected, this is one of the instruction patterns that is
internally replaced with a single optimized instruction, as long as
none of the instructions to be replaced is referenced by a jump
instruction. The fix enforces this check for the TX instruction of
the pattern.
Conor Walsh [Tue, 10 Nov 2020 11:03:13 +0000 (11:03 +0000)]
test/ring: fix build for O1 optimization
When DPDK is compiled with gcc < 9 with the optimization level set to 1
gcc sees zcd in test_ring.h as possibly being uninitialised. To correct
this error if statements from _st_ring_dequeue_bulk and
_st_ring_enqueue_bulk were corrected within test_ring_mt_peek_stress_zc.c
Fixes: f72299fd157d ("test/ring: add stress tests for zero copy API") Signed-off-by: Conor Walsh <conor.walsh@intel.com> Reviewed-by: Honnappa Nagarahalli <honnappa.nagarahalli@arm.com>
David Marchand [Sat, 14 Nov 2020 09:05:33 +0000 (10:05 +0100)]
examples: restore trace point
Before make removal, those examples were built with experimental flag
for tracepoints to be compiled in but the pkg-config part of those
makefiles were missed.
Fixes: 78d44153de8f ("ethdev: add tracepoints") Cc: stable@dpdk.org Signed-off-by: David Marchand <david.marchand@redhat.com> Acked-by: Bruce Richardson <bruce.richardson@intel.com>
David Marchand [Sat, 14 Nov 2020 09:05:30 +0000 (10:05 +0100)]
examples/performance-thread: fix build with pkg-config
main.c: In function ‘lthread_tx’:
main.c:2091:25: error: implicit declaration of function ‘sched_getcpu’;
did you mean ‘sched_getparam’? [-Werror=implicit-function-declaration]
2091 | tx_conf->conf.cpu_id = sched_getcpu();
| ^~~~~~~~~~~~
| sched_getparam
cc1: all warnings being treated as errors
Explicitly pass _GNU_SOURCE and include missing header (rather than
rely on automagic inclusion from other system headers).
David Marchand [Sat, 14 Nov 2020 09:05:29 +0000 (10:05 +0100)]
examples/ntb: fix clean target
When introducing this example, the cleanup from commit 7e9562a107f1
("examples: fix make clean when using pkg-config") was missed.
Fixes: c5eebf85badc ("examples/ntb: add example for NTB") Cc: stable@dpdk.org Signed-off-by: David Marchand <david.marchand@redhat.com> Acked-by: Bruce Richardson <bruce.richardson@intel.com> Acked-by: Xiaoyun Li <xiaoyun.li@intel.com>
David Marchand [Sat, 14 Nov 2020 09:05:28 +0000 (10:05 +0100)]
examples/l3fwd-graph: fix pkg-config usage
This example missed the fixes from commit 69b1bb49ed82
("examples: hide error for missing pkg-config path flag") and
commit 12a652a02b08 ("examples: fix build with old pkg-config").
David Marchand [Sat, 14 Nov 2020 09:05:26 +0000 (10:05 +0100)]
examples/l2fwd-crypto: fix build with pkg-config
Two issues fixed here.
First add the experimental flag.
Then fix a link issue with the crypto scheduler driver:
/usr/bin/ld: /tmp/cchr7aHA.o: in function `main':
main.c:(.text.startup+0x1673): undefined reference to
`rte_cryptodev_scheduler_workers_get'
collect2: error: ld returned 1 exit status
Fixes: e3bcb99a5e13 ("examples/l2fwd-crypto: limit number of sessions") Fixes: 261bbff75e34 ("examples: use separate crypto session mempools") Cc: stable@dpdk.org Signed-off-by: David Marchand <david.marchand@redhat.com>
David Marchand [Sat, 14 Nov 2020 09:05:25 +0000 (10:05 +0100)]
examples/kni: fix build with pkg-config
rm -f build/kni build/kni-static build/kni-shared
test -d build && rmdir -p build || true
[...]
/usr/bin/ld: /tmp/cc72ssnK.o: undefined reference to symbol
'pthread_join@@GLIBC_2.2.5'
This example explicitly call pthread API and should be linked against
the pthread library.
Fixes: 724beb913b44 ("examples/kni: monitor and update link state continually") Cc: stable@dpdk.org Signed-off-by: David Marchand <david.marchand@redhat.com> Acked-by: Bruce Richardson <bruce.richardson@intel.com>
David Marchand [Sat, 14 Nov 2020 09:05:24 +0000 (10:05 +0100)]
examples/ipsec-gw: fix build with pkg-config
flow.c: In function ‘parse_flow_tokens’:
flow.c:153:23: error: taking address of packed member of ‘struct
rte_ipv4_hdr’ may result in an unaligned pointer value
[-Werror=address-of-packed-member]
153 | if (ipv4_addr_cpy(&rule->ipv4.spec.hdr.src_addr,
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~
flow.c:154:9: error: taking address of packed member of ‘struct
rte_ipv4_hdr’ may result in an unaligned pointer value
[-Werror=address-of-packed-member]
154 | &rule->ipv4.mask.hdr.src_addr,
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~
flow.c:170:23: error: taking address of packed member of ‘struct
rte_ipv4_hdr’ may result in an unaligned pointer value
[-Werror=address-of-packed-member]
170 | if (ipv4_addr_cpy(&rule->ipv4.spec.hdr.dst_addr,
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~
flow.c:171:9: error: taking address of packed member of ‘struct
rte_ipv4_hdr’ may result in an unaligned pointer value
[-Werror=address-of-packed-member]
171 | &rule->ipv4.mask.hdr.dst_addr,
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~
cc1: all warnings being treated as errors
Meson build is fine since we waive those warnings.
Replicate it for make.
Fixes: 8e693616fcb2 ("examples/ipsec-secgw: enable flow based distribution") Cc: stable@dpdk.org Signed-off-by: David Marchand <david.marchand@redhat.com> Acked-by: Bruce Richardson <bruce.richardson@intel.com>
git.droids-corp.org / dpdk.git / log