Omitting inputs and outputs mbuf creation for BaseBand Null Device
as inputs and outputs data do not exist for Null Device
Fixes: b2a4654f082b ("mempool: check for zero size creation")
Cc: pablo.de.lara.gua...@intel.com
Signed-off-by: Kamil Chalupnik
---
app/test-bbdev/test_bbdev_perf.c | 14 ++
Tested-by: Peng, Yuan
> NIC Fortville 4*10G
> ethtool -i ens785f1
> driver: i40e
> version: 2.4.10
> firmware-version: 6.01 0x80003205 1.1691.0 DPDK version: 18.11-rc0
> commit 76b9d9de5c7d747c381027156aac07735cb1bc0c
> FlexRAN-1.6.0
> Os: 4.4.0-62-generic
> ICC:parallel_studio_xe_2018_update1_p
> -Original Message-
> From: Peng, Yuan
> Sent: Friday 17 August 2018 10:39
> To: Chalupnik, KamilX ; Mokhtar, Amr
> ; De Lara Guarch, Pablo
>
> Cc: dev@dpdk.org; sta...@dpdk.org; Chalupnik, KamilX
> ; Peng, Yuan
> Subject: RE: [dpdk-dev] [PATCH] app/testbbdev: fix inputs mbuf creation
Hotplug is an important feature for use-cases like the datacenter device's
fail-safe and for SRIOV Live Migration in SDN/NFV. It could bring higher
flexibility and continuality to networking services in multiple use-cases
in the industry. So let's see how DPDK can help users implement hotplug
solut
A memory failure and system crash can be caused if a device is hotplugged
out but the application can still access the device by MMIO.
This patch introduces bus ops to handle memory failures of illegal access,
especially for hotplug. Each bus can implement its own case-dependent
logic to handle th
This patch aims to add a helper to iterate through all buses to find the
relevant bus to handle the sigbus error.
Signed-off-by: Jeff Guo
Acked-by: Shaopeng He
---
v10->v9:
refine commit log
---
lib/librte_eal/common/eal_common_bus.c | 43 ++
lib/librte_eal/commo
This patch implements the ops to handle memory failures on the PCI bus. It
avoids MMIO read/write errors by creating a new dummy memory to remap the
memory where the failure is.
Signed-off-by: Jeff Guo
Acked-by: Shaopeng He
---
v10->v9:
change pci bus ops name
---
drivers/bus/pci/pci_common.c
When a device is hotplugged out, the PCI resource is released in the
kernel, the UIO file descriptor will disappear and the irq will be
released. After this, a kernel crash will be caused if the igb uio driver
tries to access or release these resources.
And more, uio_remove will be called unexpect
The mechanism can initially register the sigbus handler after the device
event monitor is enabled. When a sigbus error is captured, the mechanism
will check the failure address and accordingly remap the invalid memory
for the corresponding device. It could prevent the application from
crashing when
This patch implements the ops for the PCI bus sigbus handler. It finds the
PCI device that is being hotplugged out and calls the relevant ops of the
memory failure handler to handle the failure of the device.
Signed-off-by: Jeff Guo
Acked-by: Shaopeng He
---
v10->v9:
refine doc.
---
drivers/bus
This patch aims to add a helper to iterate through all buses to find the
relevant bus to handle the sigbus error.
Signed-off-by: Jeff Guo
Acked-by: Shaopeng He
---
v10->v9:
refine commit log
---
lib/librte_eal/common/eal_common_bus.c | 43 ++
lib/librte_eal/commo
When a device is hotplugged out, a sigbus error will occur of the datapath
can still read/write to the device. A handler is required here to capture
the sigbus signal and handle it appropriately.
This patch introduces bus ops to handle sigbus errors. Each bus can
implement its own case-dependent l
This patch use testpmd for example to show how to use failure handle
mechanism for hotplug in app.
Signed-off-by: Jeff Guo
---
v10->v9:
use new APIs to manage hotplug handling.
---
app/test-pmd/testpmd.c | 27 ++-
1 file changed, 22 insertions(+), 5 deletions(-)
diff --g
We currently have eal event and ethdev event for ethdev hotplug. Some
ethdev's need to use an eal event to detect hotplug behaviors. Previously,
we needed to register an eal event callback in the application, but this
potentially causes a race condition between the eal event process to the
ethdev e
Implement a couple of eal device event handler install/unintall APIs in
Ethdev as below:
- rte_eth_dev_event_handler_install
- rte_eth_dev_event_handler_uninstall
Each ethdev could call these APIs to install eal event handler in the PMDs,
with each callback corresponding to one port, and process
This patch aims to enable hotplug detection in i40e PMD.
First, it sets the “RTE_PCI_DRV_INTR_RMV” flag in drv_flags to
announce the ability of hotplug processing. Then, it calls the
“rte_eth_dev_event_handler_install” API to install the device
event handler for ethdev. When the eal device event b
This patch aims to enable hotplug detection in ixgbe PMD.
First, it sets the “RTE_PCI_DRV_INTR_RMV” flag in drv_flags to
announce the ability of hotplug processing. Then, it calls the
“rte_eth_dev_event_handler_install” API to install the device
event handler for ethdev. When the eal device event
Since we can now use the PMDs to manage the eal event for hotplug, we no
longer need to register the device event callback in the application
anymore. This patch removes the redundant code.
Signed-off-by: Jeff Guo
Acked-by: Wenzhuo Lu
---
v6->v5:
refine commit log
---
app/test-pmd/testpmd.c | 7
As we may know that the process of hotplug is different between igb_uio
and vfio. For igb_uio, it could use uevent notification and memory
failure handle mechanism for hotplug. But for vfio, when device is be
hotplug-out, the uevent can not be detected immediately, because of the
vfio kernel module
Add a new req event in eal device event for vfio hotplug. When the req
request send from the vfio kernel module be detected, vfio userpace
driver could use this event to notify the app to handler it.
Signed-off-by: Jeff Guo
---
lib/librte_eal/common/include/rte_dev.h | 1 +
1 file changed, 1 ins
Add a new req notifier in eal interrupt for enable vfio hotplug.
Signed-off-by: Jeff Guo
---
lib/librte_eal/common/include/rte_eal_interrupts.h | 1 +
lib/librte_eal/linuxapp/eal/eal_interrupts.c | 71 ++
2 files changed, 72 insertions(+)
diff --git a/lib/librte_eal/c
This patch modify the device event callback process function name to be
more explicit, and exposure the API from private to public. The drivers
and apps would directly use this API to process device event callback.
Signed-off-by: Jeff Guo
---
lib/librte_eal/common/eal_common_dev.c | 5 +++--
l
There are some extended interrupt types in vfio pci device except from the
existing interrupts, such as err and req notifier, it could be useful for
device error monitoring. And these corresponding interrupt handler is
different from the other interrupt handler that register in PMDs, so a new
inter
This patch will add req notifier processing to enable hotplug for vfio.
When device is be hotplug-out, the vfio kernel module will sent req
notifier to request user space to release the allocated resources. It
could use the bus failure mechanism to trigger the currently hotplug
procedure of user sp
Allocation failures of shadow used ring and batched copy array
are not recoverable and leads to the segmentation faults like
this on the receiving/transmission path:
Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 0x7f913fecf0 (LWP 43625)]
in copy_desc_to_mbuf () at
Add following testpmd run-time commands to support test of TM packet
marking:
set port tm mark ip_ecn
set port tm mark ip_dscp
set port tm mark vlan_dei
Signed-off-by: Krzysztof Kanas
---
app/test-pmd/cmdline.c | 3 +
app/test-pmd/cmdline_tm.c
Hi Nikhil,
Few comments inline.
Nit: Please use --in-reply-to while sending next versions.
(https://core.dpdk.org/contribute/)
Thanks,
Pavan
On Fri, Aug 17, 2018 at 09:50:52AM +0530, Nikhil Rao wrote:
>
> This patch adds tests for the eth Tx adapter APIs. It also
> tests the data path for
DPDK 17.11 introduced the IOMMU feature which caused additional
DPDK memory requirement per vhostuser device as part of
iotlb_init(). Today this is done unconditionally (from DPDK
17.11 onwards) i.e. irrespective of IOMMU feature being
enabled on the vhostuser device, iotlb is initialized. This
bre
> I guess the try-locks still does not solve the overhead of multiple readers
> contending the counter. It just provides a non-blocking version of the same
> algorithm.
DPDK project does not use any rwlock for solving any overhead problem
of multiple reader (at least I did not find it). For non
On 17.08.2018 23:19, Nitin Katiyar wrote:
> DPDK 17.11 introduced the IOMMU feature which caused additional
> DPDK memory requirement per vhostuser device as part of
> iotlb_init(). Today this is done unconditionally (from DPDK
> 17.11 onwards) i.e. irrespective of IOMMU feature being
> enabled on
LMTST does not guarantee packet content get synced with L2C when
HW access the packet for transmitting.
Adding coherent IO write barrier will make sure HW sees the correct
packet if its modified.
Fixes: 9e747589bd4c ("net/octeontx: add packet transmit burst function")
Cc: sta...@dpdk.org
Reporte
On Friday 17 August 2018 08:00 PM, Jerin Jacob wrote:
> LMTST does not guarantee packet content get synced with L2C when
> HW access the packet for transmitting.
>
> Adding coherent IO write barrier will make sure HW sees the correct
> packet if its modified.
>
> Fixes: 9e747589bd4c ("net/octeont
-Original Message-
From: Michel Machado
Sent: Friday, August 17, 2018 8:35 AM
To: Honnappa Nagarahalli ; Fu, Qiaobin
; Richardson, Bruce ; De Lara
Guarch, Pablo
Cc: dev@dpdk.org; Doucette, Cody, Joseph ; Wang, Yipeng1
; Wiles, Keith ; Gobriel, Sameh
; Tai, Charlie ; Stephen
Hemmin
https://bugs.dpdk.org/show_bug.cgi?id=84
Bug ID: 84
Summary: Primary process broadcasts vdevs during each
secondary's vdev scan
Product: DPDK
Version: unspecified
Hardware: All
OS: All
Status: CO
When bond slave devices cannot transmit all packets in bufs array,
tx_burst callback shall merge the un-transmitted packets back to
bufs array. Recent merge logic introduced a bug which causes
invalid mbuf addresses being written to bufs array.
When caller frees the un-transmitted packets, due to i
35 matches
Mail list logo
