Ferruh Yigit [Fri, 14 Jun 2019 17:19:39 +0000 (18:19 +0100)]
app/testpmd: fix crash
Testpmd tries to calculate mbuf size based on "max Rx packet size" and
"max MTU segment number".
When driver set a "nb_mtu_seg_max" to zero, it causes division by zero
segmentation fault in testpmd.
If the PMD set "nb_mtu_seg_max" to zero, testpmd shouldn't try to
calculate the mbuf size.
Fixes: 33f9630fc23d ("app/testpmd: create mbuf based on max supported segments") Signed-off-by: Ferruh Yigit <ferruh.yigit@intel.com>
Using access followed by open causes a static analysis warning
about Time of check versus Time of use. Also, access() and
open() have different UID permission checks.
This is not a serious problem; but easy to fix by using errno instead.
Coverity issue: 300870 Fixes: 4a928ef9f611 ("bus/pci: enable write combining during mapping") Cc: stable@dpdk.org Signed-off-by: Stephen Hemminger <stephen@networkplumber.org> Reviewed-by: David Marchand <david.marchand@redhat.com>
There is the patch [1] that uses master device Netlink socket
to retrieve master device link settings. This is not thread safe
because this resource may be in use by other call to the master
device itself. Using the same Netlink socket concurrently from
the multiple threads causes Netlink requests malfunction and
must be eliminated. The patch replaces master Netlink socket
with the socket from representor device.
[1] http://patches.dpdk.org/patch/53120/
Fixes: 0333b2f584d9 ("net/mlx5: inherit master link settings for representors") Cc: stable@dpdk.org Signed-off-by: Viacheslav Ovsiienko <viacheslavo@mellanox.com> Acked-by: Shahaf Shuler <shahafs@mellanox.com>
Matan Azrad [Thu, 30 May 2019 10:20:39 +0000 (10:20 +0000)]
net/mlx5: recover secondary process Tx errors
The SQ errors recovery mechanism in the PMD invokes a Verbs
functions to modify the RQ states in order to reset the SQ and to
reactivate it.
These Verbs functions are not allowed to be invoked from a secondary
process, hence the PMD skips the recovery when the error is captured
by secondary processes queues.
Using the DPDK IPC mechanism the secondary process can request Verbs
queues state modifications to be done synchronically by the primary
process.
Add support for secondary process Tx errors recovery.
Matan Azrad [Thu, 30 May 2019 10:20:38 +0000 (10:20 +0000)]
net/mlx5: recover secondary process Rx errors
The RQ errors recovery mechanism in the PMD invokes a Verbs functions to
modify the RQ states in order to reset the RQ and to reactivate it.
These Verbs functions are not allowed to be invoked from a secondary
process, hence the PMD skips the recovery when the error is captured by
secondary processes queues.
Using the DPDK IPC mechanism the secondary process can request Verbs
queues state modifications to be done synchronically by the primary
process.
Add support for secondary process Rx errors recovery.
Matan Azrad [Thu, 30 May 2019 10:20:37 +0000 (10:20 +0000)]
net/mlx5: handle Tx completion with error
When WQEs are posted to the HW to send packets, the PMD may get a
completion report with error from the HW, aka error CQE which is
associated to a bad WQE.
The error reason may be bad address, wrong lkey, bad sizes, etc.
that can wrongly be configured by the PMD or by the user.
Checking all the optional mistakes to prevent error CQEs doesn't make
sense due to performance impacts and huge complexity.
The error CQEs change the SQ state to error state what causes all the
next posted WQEs to be completed with CQE flush error forever.
Currently, the PMD doesn't handle Tx error CQEs and even may crashed
when one of them appears.
Extend the Tx data-path to detect these error CQEs, to report them by
the statistics error counters, to recover the SQ by moving the state
to ready again and adjusting the management variables appropriately.
Sometimes the error CQE root cause is very hard to debug and even may
be related to some corner cases which are not reproducible easily, hence
a dump file with debug information will be created for the first number
of error CQEs, this number can be configured by the PMD probe
parameters.
Matan Azrad [Thu, 30 May 2019 10:20:36 +0000 (10:20 +0000)]
net/mlx5: extend Rx completion with error handling
When WQEs are posted to the HW to receive packets, the PMD may receive
a completion report with error from the HW, aka error CQE which is
associated to a bad WQE.
The error reason may be bad address, wrong lkey, small buffer size,
etc. that can wrongly be configured by the PMD or by the user.
Checking all the optional mistakes to prevent error CQEs doesn't make
sense due to performance impacts, moreover, some error CQEs can be
triggered because of the packets coming from the wire when the DPDK
application has no any control.
Most of the error CQE types change the RQ state to error state what
causes all the next received packets to be dropped by the HW and to be
completed with CQE flush error forever.
The current solution detects these error CQEs and even reports the
errors to the user by the statistics error counters but without
recovery, so if the RQ inserted to the error state it never moves to
ready state again and all the next packets ever will be dropped.
Extend the error CQEs handling for recovery by moving the state to
ready again, and rearranging all the RQ WQEs and the management
variables appropriately.
Sometimes the error CQE root cause is very hard to debug and even may
be related to some corner cases which are not reproducible easily,
hence a dump file with debug information will be created for the first
number of error CQEs, this number can be configured by the PMD probe
parameters.
Matan Azrad [Thu, 30 May 2019 10:20:34 +0000 (10:20 +0000)]
net/mlx5: mitigate Rx doorbell memory barrier
The RQ WQEs must be written in the memory before the HW gets the RQ
doorbell, hence a memory barrier should be triggered after the WQEs
writing and before the doorbell writing.
The current code used rte_wmb barrier which ensures that all the memory
stores were done while it is enough to use rte_cio_wmb barrier for the
local memory stores because the WQEs are in local memory.
Matan Azrad [Thu, 30 May 2019 10:20:33 +0000 (10:20 +0000)]
net/mlx5: fix device arguments error detection
When bad device arguments are added to the DPDK command line, the PMD
ignores all the command line arguments specified by the user and uses
the default values instead.
This behavior doesn't make sense because the user intention is to force
some device parameters and expects to get an error in case of
problematic issues with the arguments.
Stop probing and report an error in case of problematic command line
arguments.
Fixes: e72dd09b614e ("net/mlx5: add support for configuration through kvargs") Cc: stable@dpdk.org Signed-off-by: Matan Azrad <matan@mellanox.com> Acked-by: Shahaf Shuler <shahafs@mellanox.com>
Matan Azrad [Thu, 30 May 2019 10:20:31 +0000 (10:20 +0000)]
net/mlx5: remove Rx queues indexes correlation
There is a full correlation between the CQE indexes to the WQE indexes
in the vectorized Rx queues management.
When the RQ is inserted to the reset state, the correlation may break
because the HW starts the RQ polling from index 0 while the CQ polling
continues regularly.
As an arrangement to CQE errors handling, when the RQ can be reset,
the correlation dependence should be removed from all the Rx queues
index managements.
Remove the aforementioned dependence from the vectorized Rx burst
functions.
Haiyue Wang [Fri, 7 Jun 2019 15:56:21 +0000 (23:56 +0800)]
net/ixgbevf: add full link status check option
To get the VF's link status by calling 'rte_eth_link_get_nowait()', the
VF not only check PF's physical link status, but also check the mailbox
running status. And mailbox checking will generate mailbox interrupt in
PF, it will be worse if many VFs are running in the system, the PF will
have to handle many interrrupts.
Normally, checking the PF's physical link status is enough for nowait.
For different scenarios, adding an 'pflink_fullchk' option to control
whether to check the link fully or not.
Fixes: 91546fb62e67 ("net/ixgbevf: fix link state") Cc: stable@dpdk.org Signed-off-by: Haiyue Wang <haiyue.wang@intel.com> Acked-by: Scott Daniels <daniels@research.att.com> Acked-by: Qi Zhang <qi.z.zhang@intel.com>
Kalesh AP [Sat, 8 Jun 2019 19:22:06 +0000 (23:22 +0400)]
net/bnxt: fix xstats
If the HWRM_PORT_QSTATS_EXT fails to initialize
fw_rx_port_stats_ext_size or fw_tx_port_stats_ext_size,
the driver can end up passing junk statistics to the application.
Instead of relying on the application to initialize the xstats
buffer before calling the xstats_get dev_op, memset xstats
with zeros to avoid returning or displaying incorrect statistics.
Also fixed the buffer starting offset.
Fixes: f55e12f33416 ("net/bnxt: support extended port counters") Cc: stable@dpdk.org Signed-off-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com> Signed-off-by: Ajit Khaparde <ajit.khaparde@broadcom.com> Reviewed-by: Lance Richardson <lance.richardson@broadcom.com>
Somnath Kotur [Sat, 8 Jun 2019 19:22:05 +0000 (23:22 +0400)]
net/bnxt: use configured MTU during load
The MTU value of a port can be (re)configured out-of-band.
FW will be returning this configured MTU as part of func_qcfg cmd.
Driver to use this value during load time.
Somnath Kotur [Sat, 8 Jun 2019 19:22:04 +0000 (23:22 +0400)]
net/bnxt: check for null completion ring doorbell
It is observed that sometimes during init, the bnxt_int_handler() gets
invoked while the cpr->cp_db.doorbell is not yet initialized. Check for
the same and return.
John Daley [Thu, 6 Jun 2019 15:26:58 +0000 (08:26 -0700)]
net/enic: remove flow locks
There is no requirement for thread safety in the flow PMD code and no
need for the locks.
Fixes: 6ced137607d0 ("net/enic: flow API for NICs with advanced filters enabled") Cc: stable@dpdk.org Signed-off-by: John Daley <johndale@cisco.com> Reviewed-by: Hyong Youb Kim <hyonkim@cisco.com>
John Daley [Thu, 6 Jun 2019 15:26:57 +0000 (08:26 -0700)]
net/enic: remove flow count action support
The firmware in 1400 series VIC adapters which would support COUNT
flow action was postponed and reworked. The capability will be
re-added in a future release when the firmware is available.
This reverts the following commits.
commit 86df6c4e2fce ("net/enic: support flow counter action")
commit 1b4ce87dc5e6 ("net/enic: fix counter action")
Cc: stable@dpdk.org Signed-off-by: John Daley <johndale@cisco.com> Reviewed-by: Hyong Youb Kim <hyonkim@cisco.com>
When Rx interrupts are disabled, we simply disable rearm when
the interrupt fires the next time. So, the next packet will
trigger interrupt (if it is not happened yet after previous Rx
burst processing).
Signed-off-by: Georgiy Levashov <georgiy.levashov@oktetlabs.ru> Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
Shahed Shaikh [Tue, 4 Jun 2019 18:53:52 +0000 (11:53 -0700)]
net/bnx2x: fix supported max Rx/Tx descriptor count
Driver does not provide limit on number Rx and Tx descriptors per queue,
this may result in application configuring 64k descriptors (default set
by rte_eth_dev_info_get()) and further result in issues in PMD and HW
flows due to unsupported number.
Shahed Shaikh [Tue, 4 Jun 2019 18:53:51 +0000 (11:53 -0700)]
net/bnx2x: fix link state
Don't call bnx2x_link_status_update() from bnx2x_link_update().
Actual use case of bnx2x_link_status_update() is to update the link
status in shared memory between driver and MFW, and not to get the
link status from HW.
So ideally, bnx2x_link_status_update() should be called when there
is an actual link event or change in link status.
Calling bnx2x_link_status_update() from bnx2x_link_update() may
corrupt the data of link status in shared memory and result
in inconsistent state of link.
Shahed Shaikh [Tue, 4 Jun 2019 18:53:49 +0000 (11:53 -0700)]
net/bnx2x: fix interrupt flood
PMD sets up and clears the slow path interrupt status block in dev_start
and dev_stop flow and slow path interrupt status block DMA memory for
device is allocated in dev_configure flow.
This situation creates a state where, after dev_stop is called, and if
there is a slow path interrupt from device, PMD sees the old value of
status block consumer in dev_start flow, since DMA memory for status
block belongs to old configuration and dev_start will result in
new slow path interrupt status block configuration.
And since PMD fails to ack new slow path interrupt with correct status
block consumer value, device continues to trigger interrupt causing an
interrupt flood.
Fix is to create and destroy status block DMA memory in dev_start and
dev_stop flow instead of dev_configure and dev_close flow.
Shahed Shaikh [Tue, 4 Jun 2019 18:53:48 +0000 (11:53 -0700)]
net/bnx2x: fix packet drop
Patch "8bd31421c593 ("net/bnx2x: fix ramrod timeout")"
introduced a regression where sc->scan_fp flags is
set for unexpectedly long time. So the slow path completion
handler flow is run unnecessarily which walks over receive
descriptor ring of fast path and drops the data packets while looking
for slow path completion descriptor out of fast path ring.
This issue is seen under heavy traffic with link events happening
in background.
Sunil Kumar Kori [Mon, 27 May 2019 08:35:29 +0000 (14:05 +0530)]
app/testpmd: create mbuf based on max supported segments
Configuring buffer size based following parameters:
- max-pkt-len
- max supported segments per MTU
Buffer size are configured as given below:
- If platform supports infinite segments per packet then default
buffer size is used.
- If platform supports nb_mtu_seg_max segments then buffer size
is configured as (max-pkt-len / nb_mtu_seg_max) + headroom
Signed-off-by: Sunil Kumar Kori <skori@marvell.com> Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>
Xiaolong Ye [Thu, 30 May 2019 09:07:05 +0000 (17:07 +0800)]
net/af_xdp: enable zero copy by external mbuf
Implement zero copy of af_xdp pmd through mbuf's external memory
mechanism to achieve high performance.
This patch also provides a new parameter "pmd_zero_copy" for user, so they
can choose to enable zero copy of af_xdp pmd or not.
To be clear, "zero copy" here is different from the "zero copy mode" of
AF_XDP, it is about zero copy between af_xdp umem and mbuf used in dpdk
application.
The failsafe driver device info had several issues in the
info it reported in dev_info_get:
- it cleared dev_info->device set in rte_eth_dev_info_get
- many fields (for example max_rx_queue) should be the minimum
of all sub devices
- it reported tx capa for the active transmit device, but
the device may change.
There was enough messed up that ended up reworking the info_get
handler. There is no need to save current values or have a
template for defaults.
Fixes: 4e31ee26ed51 ("net/failsafe: report actual device capabilities") Cc: stable@dpdk.org Signed-off-by: Stephen Hemminger <sthemmin@microsoft.com> Acked-by: Gaetan Rivet <gaetan.rivet@6wind.com>
Mesut Ali Ergin [Thu, 6 Jun 2019 16:30:42 +0000 (09:30 -0700)]
net/i40e: fix uninitialized value
Initializes mark_spec pointer to NULL.
Coverity issue: 341075 Fixes: 0bbcfc706a2b ("net/i40e: support MARK and RSS flow action") Signed-off-by: Mesut Ali Ergin <mesut.a.ergin@intel.com> Acked-by: Kevin Traynor <ktraynor@redhat.com>
net/sfc: advertise offload capabilities by Tx datapaths
Tx datapath feature bits were useful on migration from the old offload API
to the new one. However, right now it just adds indirection which
complicates code reading and understanding. Also addition of a new
offloads requires addition of a new feature bits and makes patches longer
and harder to understand. So, remove feature bits which correspond to Tx
offloads and simply advertise device and per-queue offloads directly.
Generic code could still mask some offloads if running HW or FW does not
support it.
Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com> Reviewed-by: Igor Romanov <igor.romanov@oktetlabs.ru> Reviewed-by: Ivan Malov <ivan.malov@oktetlabs.ru>
net/sfc: advertise offload capabilities by Rx datapaths
Rx datapath feature bits were useful on migration from the old offload API
to the new one. However, right now it just adds indirection which
complicates code reading and understanding. Also addition of a new
offloads requires addition of a new feature bits and makes patches longer
and harder to understand. So, remove feature bits which correspond to Rx
offloads and simply advertise device and per-queue offloads directly.
Generic code could still mask some offloads if running HW or FW does not
support it.
Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com> Reviewed-by: Igor Romanov <igor.romanov@oktetlabs.ru> Reviewed-by: Ivan Malov <ivan.malov@oktetlabs.ru>
Jakub Grajciar [Thu, 6 Jun 2019 11:38:50 +0000 (13:38 +0200)]
net/memif: introduce memory interface PMD
Shared memory packet interface (memif) PMD allows for DPDK and any other
client using memif (DPDK, VPP, libmemif) to communicate using shared
memory. The created device transmits packets in a raw format. It can be
used with Ethernet mode, IP mode, or Punt/Inject. At this moment, only
Ethernet mode is supported in DPDK memif implementation. Memif is Linux
only.
Signed-off-by: Jakub Grajciar <jgrajcia@cisco.com> Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>
Maxime Coquelin [Wed, 5 Jun 2019 10:00:38 +0000 (12:00 +0200)]
net/virtio: fix segment length in mergeable packed Rx
Head segment data_len field is wrongly summed with the length
of all the segments of the chain, whereas it should be the
length of the first segment only.
Maxime Coquelin [Wed, 5 Jun 2019 10:00:37 +0000 (12:00 +0200)]
net/virtio: fix mergeable Rx with segmented packet
After having dequeued a burst of descriptors, there may be a
need to dequeue a few more if the last packet was segmented
and not complete. When it happens, the extra segments were
not properly attached to the mbuf chain, and so were lost.
Also, head segment data_len field is wrongly summed with
the length of all the segments of the chain.
This patch fixes both the mbuf chaining and head segment's
data_len field
Fixes: bcac5aa207f8 ("net/virtio: improve batching in mergeable path") Cc: stable@dpdk.org Reported-by: Yaroslav Brustinov <ybrustin@cisco.com> Reviewed-by: David Marchand <david.marchand@redhat.com> Reviewed-by: Tiwei Bie <tiwei.bie@intel.com> Signed-off-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Maxime Coquelin [Wed, 5 Jun 2019 10:00:36 +0000 (12:00 +0200)]
net/virtio: fix in-order Rx with segmented packet
After having dequeued a burst of descriptors, there may be a
need to dequeue a few more if the last packet was segmented
and not complete. When it happens, the extra segments were
not properly attached to the mbuf chain, and so were lost.
Also, head segment data_len field is wrongly summed with
the length of all the segments of the chain.
This patch fixes both the mbuf chaining and head segment's
data_len field.
Fixes: e5f456a98d3c ("net/virtio: support in-order Rx and Tx") Cc: stable@dpdk.org Reported-by: Yaroslav Brustinov <ybrustin@cisco.com> Reviewed-by: David Marchand <david.marchand@redhat.com> Reviewed-by: Tiwei Bie <tiwei.bie@intel.com> Signed-off-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Maxime Coquelin [Wed, 29 May 2019 13:04:20 +0000 (15:04 +0200)]
eal/x86: force inlining of all memcpy and mov helpers
Some helpers in the header file are forced inlined other are
only inlined, this patch forces inline for all.
It will avoid it to be embedded as functions when called multiple
times in the same object file. For example, when we added packed
ring support in vhost-user library, rte_memcpy_generic got no
more inlined.
Signed-off-by: Maxime Coquelin <maxime.coquelin@redhat.com> Acked-by: Bruce Richardson <bruce.richardson@intel.com>
Maxime Coquelin [Wed, 29 May 2019 13:04:18 +0000 (15:04 +0200)]
vhost: do not inline unlikely fragmented buffers code
Handling of fragmented virtio-net header and indirect descriptors
tables was implemented to fix CVE-2018-1059. It should never
happen with healthy guests and so is already considered as
unlikely code path.
This patch moves these bits into non-inline dedicated functions
to reduce the I-cache pressure.
Signed-off-by: Maxime Coquelin <maxime.coquelin@redhat.com> Reviewed-by: Tiwei Bie <tiwei.bie@intel.com>
David Harton [Thu, 16 May 2019 18:28:03 +0000 (14:28 -0400)]
net/i40e: eliminate weak symbols in data path
Use of weak symbols can hide makefile errors especially when
custom makefiles are used. Removing the use of weak symbols
to avoid a stub function being linked in production code.
Signed-off-by: David Harton <dharton@cisco.com> Acked-by: Bruce Richardson <bruce.richardson@intel.com>
William Tu [Fri, 31 May 2019 16:52:42 +0000 (09:52 -0700)]
net/af_xdp: fix remove path
When users call rte_eth_dev_close() and rte_dev_remove(), the af_xdp
pmd return -1 (EPERM) due to eth_dev == NULL.
Since the af_xdp pmd driver advertises RTE_ETH_DEV_CLOSE_REMOVE, all
the resources are freed on rte_eth_dev_close(). rte_dev_remove() tries
to detach device and subsequently calls rte_pmd_af_xdp_remove() that
tries to free already freed resources and fails.
Fix it by return success.
Fixes: f1debd77efaf ("net/af_xdp: introduce AF_XDP PMD") Cc: stable@dpdk.org
Reported-at: https://patchwork.ozlabs.org/patch/1106528/ Suggested-by: Ilya Maximets <i.maximets@samsung.com> Signed-off-by: William Tu <u9012063@gmail.com> Acked-by: Xiaolong Ye <xiaolong.ye@intel.com>
Ajit Khaparde [Wed, 29 May 2019 21:02:24 +0000 (17:02 -0400)]
net/bnxt: fix RSS RETA indirection table ops
We are trying to update the indirection table for all the VNICs.
We should update the table only for the default vnic0.
Fix the reta update function to only update table entries that are
selected by the update mask. Translate queue number to firmware
group ID when updating an entry.
Fix reta query op to only return table entries as identfied by the
provided mask. Translate firmware group IDs to queue numbers.
Removed extraneous code from bnxt_reta_query_op().
Ajit Khaparde [Wed, 29 May 2019 21:02:20 +0000 (17:02 -0400)]
net/bnxt: update release notes
Update release doc briefly describing updates to bnxt PMD done during
19.05 release, including transmit optimization changes in the commits
identified by the "Fixes:" tags below.
git.droids-corp.org / dpdk.git / log