Chas Williams [Tue, 27 Nov 2018 04:56:13 +0000 (23:56 -0500)]
ip_frag: fix IPv6 when MTU sizes not aligned to 8 bytes
The same issue was fixed on for the ipv4 version of this routine in
commit 8d4d3a4f7337 ("ip_frag: handle MTU sizes not aligned to 8 bytes").
Briefly, the size of an ipv6 header is always 40 bytes. With an MTU of
1500, this will never produce a multiple of 8 bytes for the frag_size
and this routine can never succeed. Since RTE_ASSERTS are disabled by
default, this failure is typically ignored.
To fix this, round down to the nearest 8 bytes and use this when
producing the fragments.
David Hunt [Fri, 14 Dec 2018 13:31:37 +0000 (13:31 +0000)]
examples/power: increase max cores to 256
Increase the number of addressable cores from 64 to 256. Also remove the
warning that incresing this number beyond 64 will cause problems (because
of the previous use of uint64_t masks). Now this number can be increased
significantly without causing problems.
Signed-off-by: David Hunt <david.hunt@intel.com> Reviewed-by: Anatoly Burakov <anatoly.burakov@intel.com>
David Hunt [Fri, 14 Dec 2018 13:31:35 +0000 (13:31 +0000)]
examples/power: remove mask functions
Since we're moving to allowing greater than 64 cores, the mask functions
that use uint64_t to perform functions on a masked set of cores are no
longer needed, so removing them.
Signed-off-by: David Hunt <david.hunt@intel.com> Reviewed-by: Anatoly Burakov <anatoly.burakov@intel.com>
David Hunt [Fri, 14 Dec 2018 13:31:34 +0000 (13:31 +0000)]
examples/power: change 64-bit masks to arrays
vm_power_manager currently makes use of uint64_t masks to keep track of
cores in use, limiting use of the app to only being able to manage the
first 64 cores in a multi-core system. Many modern systems have core
counts greater than 64, so this limitation needs to be removed.
This patch converts the relevant 64-bit masks to character arrays.
Signed-off-by: David Hunt <david.hunt@intel.com> Reviewed-by: Anatoly Burakov <anatoly.burakov@intel.com>
rte_timer_manage() adds expired timers to a "run list", and walks the
list, transitioning each timer from the PENDING to the RUNNING state.
If another lcore resets or stops the timer at precisely this
moment, the timer state would instead be set to CONFIG by that other
lcore, which would cause timer_manage() to skip over it. This is
expected behavior.
However, if a timer expires quickly enough, there exists the
following race condition that causes the timer_manage() routine to
misinterpret a timer in CONFIG state, resulting in lost timers:
- Thread A:
- starts a timer with rte_timer_reset()
- the timer is moved to CONFIG state
- the spinlock associated with the appropriate skiplist is acquired
- timer is inserted into the skiplist
- the spinlock is released
- Thread B:
- executes rte_timer_manage()
- find above timer as expired, add it to run list
- walk run list, see above timer still in CONFIG state, unlink it from
run list and continue on
- Thread A:
- move timer to PENDING state
- return from rte_timer_reset()
- timer is now in PENDING state, but not actually linked into a
pending list or a run list and will never get processed further
by rte_timer_manage()
This commit fixes this race condition by only releasing the spinlock
after the timer state has been transitioned from CONFIG to PENDING,
which prevents rte_timer_manage() from seeing an incorrect state.
Fixes: 9b15ba895b9f ("timer: use a skip list") Cc: stable@dpdk.org Signed-off-by: Erik Gabriel Carrillo <erik.g.carrillo@intel.com> Reviewed-by: Gavin Hu <gavin.hu@arm.com>
Tomasz Jozwiak [Fri, 14 Dec 2018 12:11:47 +0000 (13:11 +0100)]
compress/qat: fix dequeue error counter
This patch increments error counter (stats.dequeue_err_count)
in case of any error detection during qat_comp_process_response
function.
Fixes: 3cc14fc48e35 ("compress/qat: check that correct firmware is in use") Fixes: 32842f2a6d7d ("compress/qat: create FW request and process response") Cc: stable@dpdk.org Signed-off-by: Tomasz Jozwiak <tomaszx.jozwiak@intel.com> Acked-by: Fiona Trahe <fiona.trahe@intel.com>
Tomasz Jozwiak [Fri, 14 Dec 2018 11:19:31 +0000 (12:19 +0100)]
compress/qat: fix return on building request error
This patch fixes error status which should be set inside
qat_comp_build_request function in case any errors are detected.
In these cases op.status is set to
RTE_COMP_OP_STATUS_INVALID_ARGS to help application debug.
Fixes: 1947bd18580b ("compress/qat: support scatter-gather buffers") Cc: stable@dpdk.org Signed-off-by: Tomasz Jozwiak <tomaszx.jozwiak@intel.com> Acked-by: Fiona Trahe <fiona.trahe@intel.com>
Tomasz Jozwiak [Fri, 14 Dec 2018 11:19:30 +0000 (12:19 +0100)]
common/qat: remove check of valid firmware response
This reverts commit d09973f6c4df ("common/qat: fix for invalid
response from firmware") due to incorrectly reporting failures
on some older firmware versions.
Fixes: d09973f6c4df ("common/qat: fix for invalid response from firmware") Cc: stable@dpdk.org Signed-off-by: Tomasz Jozwiak <tomaszx.jozwiak@intel.com> Acked-by: Fiona Trahe <fiona.trahe@intel.com>
Kamil Chalupnik [Fri, 7 Dec 2018 15:15:34 +0000 (16:15 +0100)]
app/bbdev: enhance interrupt test
Improvements added to interrupt test:
- test is run in loop (number of iterations is specified by
TEST_REPETITIONS define) which ensures more accurate results
- mapping cores to thread parameteres was put in order.
Master core is always set at first index. It fixes problem with
running test for only one core
Signed-off-by: Kamil Chalupnik <kamilx.chalupnik@intel.com> Acked-by: Amr Mokhtar <amr.mokhtar@intel.com>
Kamil Chalupnik [Fri, 7 Dec 2018 15:15:32 +0000 (16:15 +0100)]
bbdev: enhance throughput test
Improvements added to throughput test:
- test is run in loop (number of iterations is specified by
TEST_REPETITIONS define) which ensures more accurate results
- length of input data is calculated based on amount of CBs in TB
- maximum number of decoding iterations is gathered from results
- added new functions responsible for printing results
- small fixes for memory management
Signed-off-by: Kamil Chalupnik <kamilx.chalupnik@intel.com> Acked-by: Amr Mokhtar <amr.mokhtar@intel.com>
Nikhil Rao [Mon, 17 Dec 2018 04:39:41 +0000 (10:09 +0530)]
eventdev: fix eth Tx adapter queue count checks
rte_event_eth_tx_adapter_queue_add() - add a check
that returns an error if the ethdev has zero Tx queues
configured.
rte_event_eth_tx_adapter_queue_del() - remove the
checks for ethdev queue count, instead check for
queues added to the adapter which maybe different
from the current ethdev queue count.
app/eventdev: detect deadlock for timer event producer
If timer events get dropped for some reason, the thread that launched
producer and worker cores will never exit, because the deadlock check
doesn't currently apply to the event timer adapter case. This commit
fixes this.
Fixes: d008f20bce23 ("app/eventdev: add event timer adapter as a producer") Cc: stable@dpdk.org Signed-off-by: Erik Gabriel Carrillo <erik.g.carrillo@intel.com> Acked-by: Jerin Jacob <jerin.jacob@caviumnetworks.com>
Jiayu Hu [Mon, 17 Dec 2018 06:05:39 +0000 (14:05 +0800)]
gro: fix overflow of payload length calculation
When the packet length is smaller than the header length,
the calculated payload length will be overflowed and result
in incorrect reassembly behaviors.
Fixes: 1e4cf4d6d4fb ("gro: cleanup") Fixes: 9e0b9d2ec0f4 ("gro: support VxLAN GRO") Cc: stable@dpdk.org Signed-off-by: Jiayu Hu <jiayu.hu@intel.com>
Anatoly Burakov [Tue, 13 Nov 2018 15:54:44 +0000 (15:54 +0000)]
eal: clean up unused files on initialization
When creating process data structures, EAL will create many files
in EAL runtime directory. Because we allow multiple secondary
processes to run, each secondary process gets their own unique
file. With many secondary processes running and exiting on the
system, runtime directory will, over time, create enormous amounts
of sockets, fbarray files and other stuff that just sits there
unused because the process that allocated it has died a long time
ago. This may lead to exhaustion of disk (or RAM) space in the
runtime directory.
Fix this by removing every unlocked file at initialization that
matches either socket or fbarray naming convention. We cannot be
sure of any other files, so we'll leave them alone. Also, remove
similar code from mp socket code.
We do it at the end of init, rather than at the beginning, because
secondary process will use primary process' data structures even
if the primary itself has died, and we don't want to remove those
before we lock them.
Luca Boccassi [Wed, 5 Dec 2018 14:25:45 +0000 (14:25 +0000)]
doc: fix garbage text in generated HTML guides
If the user has MAKEFLAGS set in the environment when building the
documentation, the doc/guides/conf.py script which calls "make"
exclusively to get the project version might pick up garbage from
stdout, like:
<title>FAQ — Data Plane Development Kit make[2]:
Entering directory '/build/1st/dpdk-18.11/doc/guides'
18.11.0
make[2]: Leaving directory '/build/1st/dpdk-18.11'
documentation</title>
Override MAKEFLAGS in the Python subprocess call to avoid this issue.
The DPAA bus support code put out messages like:
dpaax: read_memory_node(): Unable to glob device-tree memory node:
(/proc/device-tree/memory[@0-9]*/reg)(3)
dpaax: PA->VA translation not available;
dpaax: Expect performance impact.
These are unnecessary and likely to confuse the end user.
Fix this by doing nothing if bus is empty.
NDIS multi-queue support is only in WS2012 or later. Check the NDIS
version to limit to single queue on older versions. Similar code
exists in Linux driver.
When using multiple queues, there was a race with the kernel
in setting up the second channel. This regression is due to a kernel change
which does not allow accessing sysfs files for Hyper-V channels that are not opened.
The fix is simple, just move the logic to detect not ready
sub channels earlier in the existing loop.
Fixes: 831dba47bd36 ("bus/vmbus: add Hyper-V virtual bus support") Cc: stable@dpdk.org Reported-by: Mohammed Gamal <mgamal@redhat.com> Signed-off-by: Stephen Hemminger <sthemmin@microsoft.com>
David Marchand [Tue, 4 Dec 2018 08:28:31 +0000 (09:28 +0100)]
log: add missing experimental tag
When rte_log_register_type_and_pick_level() has been introduced, it has
been correctly added to the EXPERIMENTAL section of the eal map and the
symbol itself has been marked at its definition.
However, the declaration of this symbol in rte_log.h is missing the
__rte_experimental tag.
Because of this, a user can try to call this symbol without being aware
this is an experimental api (neither compilation nor link warning).
Fixes: b22e77c02620 ("eal: register log type and pick level from args") Cc: stable@dpdk.org Signed-off-by: David Marchand <david.marchand@redhat.com> Acked-by: Neil Horman <nhorman@tuxdriver.com>
Tiwei Bie [Mon, 3 Dec 2018 02:58:57 +0000 (10:58 +0800)]
pdump: remove deprecated APIs
We already changed to use generic IPC in pdump since below commit:
commit 660098d61f57 ("pdump: use generic multi-process channel")
The `rte_pdump_set_socket_dir()`, the `path` parameter of
`rte_pdump_init()` and the `enum rte_pdump_socktype` have been
deprecated since then. This commit removes these deprecated
APIs and also bumps the pdump ABI.
Timothy Redaelli [Tue, 27 Nov 2018 19:35:32 +0000 (20:35 +0100)]
mk: do not install meson.build in usertools
In commit 7dd34c71de2a ("usertools: install with meson") meson.build was
added in usertools directory and so it's copied to $(datadir)/usertools
with "make install".
This patch avoids to copy meson.build when installing usertools with
"make install".
Gavin Hu [Tue, 11 Dec 2018 03:55:50 +0000 (11:55 +0800)]
config: enable C11 memory model for armv8 with meson
This patch makes the configuration based on makefile and the
configuration based on meson to be the same.
Fixes: c6e536e38437 ("build: add more implementers IDs and PNs for ARM") Cc: stable@dpdk.org Signed-off-by: Gavin Hu <gavin.hu@arm.com> Reviewed-by: Ruifeng Wang <ruifeng.wang@arm.com> Reviewed-by: Honnappa Nagarahalli <honnappa.nagarahalli@arm.com> Acked-by: Bruce Richardson <bruce.richardson@intel.com>
Jerin Jacob [Sun, 16 Dec 2018 15:43:25 +0000 (15:43 +0000)]
maintainers: update Cavium email ids
Following Marvell's acquisition of Cavium, we need to update all the
Cavium maintainer's entries to point to our new e-mail addresses.
Update maintainers as they are no longer working for Cavium.
Thanks to Harish Patil for his support and development of our various
dpdk drivers.
Thomas Monjalon [Mon, 26 Nov 2018 20:26:59 +0000 (21:26 +0100)]
pkg: remove RPM recipe
This RPM recipe is not well maintained and probably not used a lot.
Each distribution has its own constraints and recipes.
It may not be a good idea to try maintaining packaging recipes in
the project itself.
Signed-off-by: Thomas Monjalon <thomas@monjalon.net> Acked-by: Bruce Richardson <bruce.richardson@intel.com>
Zhirun Yan [Thu, 13 Dec 2018 12:49:28 +0000 (12:49 +0000)]
net/i40e: remove redundant reset of queue number
Before this patch, there are two functions will call
i40e_dev_free_queues to free queues. For rte_eth_dev_close(), its
redundant because of duplication. For rte_eth_dev_reset() its
redundant because of not necessary, since following dev_configure
is required after dev_reset and it will be updated correctly.
This patch removes redundant code in i40e_dev_free_queues().
Fixes: 6b4537128394 ("i40e: free queue memory when closing") Cc: stable@dpdk.org Signed-off-by: Zhirun Yan <zhirun.yan@intel.com> Signed-off-by: Haiyue Wang <haiyue.wang@intel.com> Acked-by: Qi Zhang <qi.z.zhang@intel.com>
Didier Pallard [Tue, 11 Dec 2018 17:25:01 +0000 (18:25 +0100)]
net/i40e: revert fix offload not supported mask
This reverts
commit 09a62d75691f ("net/i40e: fix offload not supported mask")
Contrary to what is said in above patch commit log,
I40E_TX_OFFLOAD_NOTSUP_MASK is the mask of Tx offload bits that are part
of PKT_TX_OFFLOAD_MASK but not included in I40E_TX_OFFLOAD_MASK.
Above patch erroneously includes all PKT_RX_OFFLOAD_ bits in the
I40E_TX_OFFLOAD_NOTSUP_MASK, this is not what is expected.
Restore the initial xor that gives the expected result.
Maxime Coquelin [Tue, 27 Nov 2018 10:54:27 +0000 (11:54 +0100)]
net/virtio: allocate vrings on device NUMA node
When a guest is spanned on multiple NUMA nodes and
multiple Virtio devices are spanned onto these nodes,
we expect that their ring memory is allocated in the
right memory node.
Otherwise, vCPUs from node A may be polling Virtio rings
allocated on node B, which would increase QPI bandwidth
and impact performance.
Signed-off-by: Maxime Coquelin <maxime.coquelin@redhat.com> Reviewed-by: David Marchand <david.marchand@redhat.com>
Didier Pallard [Tue, 11 Dec 2018 13:00:28 +0000 (14:00 +0100)]
drivers/net: fix several Tx prepare functions
Since below commit, several tx_prep functions are broken, they fail to
pass supported Tx offload features check:
PKT_TX_IPVx must be set when any PKT_TX_L4 checksum is requested,
but these values are not present in the mask of supported Tx offloads
of several drivers that advertise PKT_TX_L4_MASK.
So any packet sent to those drivers with a L4 checksum request and
one of PKT_TX_IPVx bit set is rejected by the tx prepare function.
Hyong Youb Kim [Mon, 10 Dec 2018 18:28:55 +0000 (10:28 -0800)]
net/enic: support multicast filtering
The VIC hardware has 64 MAC filters per vNIC, which can be either
unicast or multicast. Use one half for unicast and the other half for
multicast, as the VIC kernel drivers for Linux and Windows do.
Signed-off-by: Hyong Youb Kim <hyonkim@cisco.com> Reviewed-by: John Daley <johndale@cisco.com>
net/cxgbe: fix control queue mbuf pool naming convention
When multiple Chelsio adapters are present, the current naming
convention of using underlying physical port number of the adapter
creates conflicts when allocating control queue mbuf pool for
multiple adapters and hence results in allocation failure. Fix by
using port_id from rte_eth_dev_data, instead of physical port number.
Fixes: 3a3aaabc7c77 ("net/cxgbe: add control queue to communicate filter requests") Cc: stable@dpdk.org Signed-off-by: Rahul Lakkireddy <rahul.lakkireddy@chelsio.com>
Dekel Peled [Tue, 4 Dec 2018 13:52:02 +0000 (15:52 +0200)]
app/testpmd: fix MPLSoGRE encapsulation
In function cmd_set_mplsogre_encap_parsed(), MPLS label value was
set in mplsogre_encap_conf struct without the required offset.
As a result the value was copied incorrectly into
rte_flow_item_mpls struct.
This patch sets MPLS label value in appropriate location at
mplsogre_encap_conf struct, so it is correctly copied to
rte_flow_item_mpls struct.
Fixes: 3e77031be855 ("app/testpmd: add MPLSoGRE encapsulation") Cc: stable@dpdk.org Signed-off-by: Dekel Peled <dekelp@mellanox.com> Acked-by: Ori Kam <orika@mellanox.com>
Dekel Peled [Tue, 4 Dec 2018 13:51:33 +0000 (15:51 +0200)]
app/testpmd: fix MPLSoUDP encapsulation
In function cmd_set_mplsoudp_encap_parsed(), MPLS label value was
set in mplsoudp_encap_conf struct without the required offset.
As a result the value was copied incorrectly into
rte_flow_item_mpls struct.
This patch sets MPLS label value in appropriate location at
mplsoudp_encap_conf struct, so it is correctly copied to
rte_flow_item_mpls struct.
Fixes: a1191d39cb57 ("app/testpmd: add MPLSoUDP encapsulation") Cc: stable@dpdk.org Signed-off-by: Dekel Peled <dekelp@mellanox.com> Acked-by: Ori Kam <orika@mellanox.com>
Dekel Peled [Wed, 28 Nov 2018 09:16:17 +0000 (11:16 +0200)]
net/mlx5: fix validation of Rx queue number
Function mlx5_ctrl_flow_vlan() is used to set the rss rule in
MLX5 PMD, using priv->reta_idx_n as number of Rx queues.
This number is passed to mlx5_flow_validate_action_rss(), which
attempts to access the Rx queues at priv->rxqs.
In case priv->rxqs_n is 0, priv->rxqs is empty, and
mlx5_flow_validate_action_rss() will crash with segmentation fault.
priv->reta_idx_n can never be 0, even if priv->rxqs_n is set to 0.
But when priv->rxqs_n is set to 0, setting the rss rule is invalid.
This patch updates mlx5_ctrl_flow_vlan(), if priv->rxqs_n is 0 the
function will fail with EINVAL errno.
Zhirun Yan [Mon, 10 Dec 2018 14:08:58 +0000 (14:08 +0000)]
net/i40e: fix port close
Port reset will call i40evf_uninit_vf() to release resource. It wants
to call i40evf_dev_close() to release resources. Before this patch,
hw->adapter_stopped was used to mark the status about start/stop and
close. So it will never call i40evf_dev_close() after stopping the port.
This patch added hw->adapter_closed flag in i40evf_dev_close() and
i40e_dev_close() to control the status of close.
Fixes: 4861cde46116 ("i40e: new poll mode driver") Cc: stable@dpdk.org Signed-off-by: Zhirun Yan <zhirun.yan@intel.com> Signed-off-by: Haiyue Wang <haiyue.wang@intel.com> Acked-by: Qi Zhang <qi.z.zhang@intel.com>
Xiaoyun Li [Tue, 4 Dec 2018 04:37:57 +0000 (12:37 +0800)]
net/ixgbe: fix overwriting RSS RETA
When starting the device, the RSS table is initialized. So the RSS
update before device_start would be overwritten. This patch allows users
to update the RSS reta table before device_start.
Fixes: db5b65301dde ("ethdev: allow to set RSS hash computation flags and/or key") Cc: stable@dpdk.org Signed-off-by: Xiaoyun Li <xiaoyun.li@intel.com> Acked-by: Qi Zhang <qi.z.zhang@intel.com>
Xiaoyun Li [Mon, 3 Dec 2018 08:26:32 +0000 (16:26 +0800)]
net/i40e: fix overwriting RSS RETA
When starting the device, the RSS table is initialized. So the RSS
update before dev_start would be overwritten. This patch allows users
to update the RSS reta table before dev_start and adjusts the order
to set entries sequentially.
Fixes: 4861cde46116 ("i40e: new poll mode driver") Cc: stable@dpdk.org Signed-off-by: Xiaoyun Li <xiaoyun.li@intel.com> Acked-by: Qi Zhang <qi.z.zhang@intel.com>
Qiming Yang [Fri, 30 Nov 2018 09:05:58 +0000 (17:05 +0800)]
net/i40e: fix getting RSS configuration
RSS hash configure get API (i40e_dev_rss_hash_conf_get()) didn't check
the return value of i40e_get_rss_key().
i40e_dev_rss_hash_conf_get() will return success even getting RSS hash
key failed. This patch fixes the issue.
Fixes: d0a349409bd7 ("i40e: support AQ based RSS config") Cc: stable@dpdk.org Signed-off-by: Qiming Yang <qiming.yang@intel.com> Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>
Radu Nicolau [Wed, 14 Nov 2018 12:19:09 +0000 (12:19 +0000)]
examples/bond: fix crash when there is no active slave
If bond_ethdev_rx_burst() called more times with no active slaves
the active slave index will point out of bounds, resulting in a
segfault.
The configured slaves needs to be checked, and if none became active
there is no point going further.
Do not start the packet processing threads until all configured
slaves become active.
Fixes: cc7e8ae84faa ("examples/bond: add example application for link bonding mode 6") Cc: stable@dpdk.org Signed-off-by: Radu Nicolau <radu.nicolau@intel.com> Acked-by: Chas Williams <chas3@att.com>
Saleh Alsouqi [Wed, 21 Nov 2018 14:47:24 +0000 (14:47 +0000)]
app/testpmd: fix MPLS BoS bit default value
The Bottom-of-Stack (bos) bit of MPLS indicates
whether its the last MPLS layer (1) or not (0).
Indicating that the encapsulating MPLS is the
last MPLS layer in the packet as the default
behavior is more appropriate since multiple
encapsulation actions is not supported.
Rasesh Mody [Sat, 17 Nov 2018 09:37:40 +0000 (09:37 +0000)]
net/bnx2x: cleanup info logs
Reduced number of INFO logs in BNX2X PMD by converting some INFO
logs to DEBUG and few NOTICE logs to INFO, removing extra new lines,
printing banner bar once for the adapter and device specific info.
Fixes: ba7eeb035a5f ("net/bnx2x: fix logging to include device name") Fixes: 540a211084a7 ("bnx2x: driver core") Cc: stable@dpdk.org Signed-off-by: Rasesh Mody <rasesh.mody@cavium.com>
Tom Barbette [Fri, 23 Nov 2018 08:03:37 +0000 (09:03 +0100)]
net/mlx5: report imissed statistics
The imissed counters (number of packets dropped because the queues were
full) were actually reported through xstats as "rx_out_of_buffer"
but was not reported through stats.
Following a recent discussion on the ML, as there is no way to tell the
user if a counter is implemented or not, this should be considered a
bug. For example, user looking at imissed will think the packets are
lost before reaching the device.
Signed-off-by: Tom Barbette <barbette@kth.se> Acked-by: Shahaf Shuler <shahafs@mellanox.com>
Take the 'other interrupt' into account when setting up
MSI-X interrupts and use the proper mask when enabling it.
Also, rearm the MSI-X vector after the LSC interrupt fires.
This change allows both LSC and RXQ interrupts to work at
the same time when using MSI-X interrupts.
Cc: stable@dpdk.org Signed-off-by: Timmons C. Player <timmons.player@spirent.com> Acked-by: Wei Zhao <wei.zhao1@intel.com>
David Marchand [Fri, 30 Nov 2018 12:32:02 +0000 (13:32 +0100)]
devtools: fix symbol check when adding experimental section
The incriminated commit did relax the condition to catch all sections
but dropped the + removal which can trigger false detection of the
special EXPERIMENTAL section when adding symbols and the section in the
same patch.
Fixes: 7281cf520f89 ("devtools: relax rule for identifying symbol section") Cc: stable@dpdk.org Signed-off-by: David Marchand <david.marchand@redhat.com> Acked-by: Neil Horman <nhorman@tuxdriver.com>
Anatoly Burakov [Mon, 26 Nov 2018 10:57:03 +0000 (10:57 +0000)]
mem: fix division by zero in no-NUMA mode
When RTE_EAL_NUMA_AWARE_HUGEPAGES is set to "n", not all memtypes
will be valid, because we skip some due to not supporting other
NUMA nodes, leading to a division by zero error down the line
because the necessary memtype fields weren't populated.
Fix it by limiting number of memtypes to number of memtypes we
have actually created.
Fixes: 1dd342d0fdc4 ("mem: improve segment list preallocation") Cc: stable@dpdk.org Signed-off-by: Anatoly Burakov <anatoly.burakov@intel.com> Tested-by: David Hunt <david.hunt@intel.com>
Darek Stojaczyk [Fri, 23 Nov 2018 15:43:28 +0000 (16:43 +0100)]
eal: fix devargs reference after probing failure
Even if a device failed to plug, it's still a device
object that references the devargs. Those devargs will
be freed automatically together with the device, but
freeing them any earlier - like it's done in the hotplug
error handling path right now - will give us a dangling
pointer and a segfault scenario.
Consider the following case:
* secondary process receives the hotplug request IPC message
* devargs are either created or updated
* the bus is scanned
* a new device object is created with the latest devargs
* the device can't be plugged for whatever reason,
bus->plug returns error
* the devargs are freed, even though they're still referenced
by the device object on the bus
For PCI devices, the generic device name comes from
a buffer within the devargs. Freeing those will make
EAL segfault whenever the device name is checked.
This patch just prevents the hotplug error handling
path from removing the devargs when there's a device
that references them. This is done by simply exiting
early from the hotplug function. As mentioned in the
beginning, those devargs will be freed later, together
with the device itself.
Fixes: 7e8b26650146 ("eal: fix hotplug add / remove") Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com> Acked-by: Maxime Coquelin <maxime.coquelin@redhat.com> Acked-by: Thomas Monjalon <thomas@monjalon.net>
Darek Stojaczyk [Fri, 23 Nov 2018 14:11:07 +0000 (15:11 +0100)]
eal: fix devargs leak on multi-process detach request
Device detach triggered through IPC leaked some memory.
It allocated a devargs objects just to use it for
parsing the devargs string in order to retrieve the
device name. Those devargs weren't passed anywhere
and were never freed.
First of all, let's put those devargs on the stack,
so they doesn't need to be freed. Then free the
additional arguments string as soon as it's allocated,
because we won't need it.
Fixes: ac9e4a17370f ("eal: support attach/detach shared device from secondary") Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com> Acked-by: Qi Zhang <qi.z.zhang@intel.com>
git.droids-corp.org / dpdk.git / log