The vtpci_ops assignment needs the 'hw->port_id' as an input parameter.
That said, we should set 'hw->port_id' firstly, then do the vtpci_ops
assignment, while the code does reversely. That would result to a crash
when more than one virtio devices are used, because we keep assigning
proper vtpci_ops to virtio_hw_internal[0]->vtpci_ops, leaving the pointer
for other ports being NULL.
Reverse the order fixes this issue.
Fixes: 9470427c88e1 ("net/virtio: do not store PCI device pointer at shared memory")
Cc: stable@dpdk.org
Reported-by: Lei Yao <lei.a.yao@intel.com>
Signed-off-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
When CONFIG_RTE_VIRTIO_USER is disabled (default on FreeBSD),
the virtio driver cannot be compiled:
librte_pmd_virtio.a(virtio_ethdev.o): In function `eth_virtio_dev_init':
(.text+0x1eba): undefined reference to `virtio_user_ops'
Reported-by: Andrew Rybchenko <arybchenko@solarflare.com>
Signed-off-by: Thomas Monjalon <thomas.monjalon@6wind.com>
When the virtio PMD is used on top of a vhost that does not support
offloads, Rx offload capabilities are still advertised by
virtio_dev_info_get(). But if an application tries to start the PMD with
Rx offloads enabled (rxmode.hw_ip_checksum = 1), the initialization of
the device will fail with -ENOTSUP and the following log:
rx ip checksum not available on this host
This patch fixes the Rx offload capabilities returned by
virtio_dev_info_get() to be consistent with features advertised by the
host.
Fixes: 96cb6711939e ("net/virtio: support Rx checksum offload")
Fixes: 86d59b21468a ("net/virtio: support LRO")
Cc: stable@dpdk.org
Signed-off-by: Olivier Matz <olivier.matz@6wind.com>
Acked-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
When closing virtio devices, close eventfds, free the struct to
store queue/irq mapping.
Signed-off-by: Jianfeng Tan <jianfeng.tan@intel.com>
Tested-by: Lei Yao <lei.a.yao@intel.com>
Acked-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
When virtio devices get stopped, tell the kernel to unbind the
mapping between interrupts and eventfds.
Note: it behaves differently from other NICs which close eventfds,
free struct. In virtio, we do those things when close device in
following patch.
Signed-off-by: Jianfeng Tan <jianfeng.tan@intel.com>
Tested-by: Lei Yao <lei.a.yao@intel.com>
Acked-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
This patch mainly allocates structure to store queue/irq mapping,
and configure queue/irq mapping down through PCI ops. It also creates
eventfds for each Rx queue and tell the kernel about the eventfd/intr
binding.
Note: So far, we hard-code 1:1 queue/irq mapping (each rx queue has
one exclusive interrupt), like this:
vec 0 -> config irq
vec 1 -> rxq0
vec 2 -> rxq1
...
which means, the "vectors" option of QEMU should be configured with
a value >= N+1 (N is the number of the queue pairs).
Signed-off-by: Jianfeng Tan <jianfeng.tan@intel.com>
Tested-by: Lei Yao <lei.a.yao@intel.com>
Acked-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
This patch implements interrupt enable/disable functions for each
Rx queue. And we rely on flags of avail queue as the hint for virtio
device to interrupt virtio driver or not.
Signed-off-by: Jianfeng Tan <jianfeng.tan@intel.com>
Tested-by: Lei Yao <lei.a.yao@intel.com>
Acked-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
Under interrupt mode, rx_descriptor_done is used as an indicator
for applications to check if some number of packets are ready to
be received.
This patch enables this by checking used ring's local consumed idx
with shared (with backend) idx.
Signed-off-by: Jianfeng Tan <jianfeng.tan@intel.com>
Tested-by: Lei Yao <lei.a.yao@intel.com>
Acked-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
We need to define a prototype for such wrapper, which makes thing
too complicated. Remove wrapper and call set_config_irq directly.
Suggested-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
Signed-off-by: Jianfeng Tan <jianfeng.tan@intel.com>
Tested-by: Lei Yao <lei.a.yao@intel.com>
Acked-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
The LSC flag is decided according to if VIRTIO_NET_F_STATUS feature
is negotiated. Copy the PCI info after the judgement will rewrite
the correct result.
Fixes: 198ab33677c9 ("net/virtio: move device initialization in a function")
CC: stable@dpdk.org
Signed-off-by: Jianfeng Tan <jianfeng.tan@intel.com>
Tested-by: Lei Yao <lei.a.yao@intel.com>
Acked-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
hw->dev, a pointer to pci_dev, was actually not used, until the
refactor of decouping from PCI device. This would somehow break
the multiple process again, since "hw" is stored at shared memory,
while "pci_dev" is not: the primary and secondary process could
have different address for it, while just one value is allowed.
Thus we should not store it to "hw", instead, we could retrieve
it from the "eth_dev->device" field.
Fixes: ae34410a8a8a ("ethdev: move info filling of PCI into drivers")
Fixes: eac901ce29be ("ethdev: decouple from PCI device")
Signed-off-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
Since commit 0e1b45a284b4 ("ethdev: decouple interrupt handling from
PCI device"), intr_handle is stored at eth_dev struct, that we could
use it directly. Thus there is no need to get it from hw.
Signed-off-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
The introduce of virtio 1.0 support brings yet another set of ops, badly,
it's not handled correctly, that it breaks the multiple process support.
The issue is the data/function pointer may vary from different processes,
and the old used to do one time set (for primary process only). That
said, the function pointer the secondary process saw is actually from the
primary process space. Accessing it could likely result to a crash.
Kudos to the last patches, we now be able to maintain those info that may
vary among different process locally, meaning every process could have its
own copy for each of them, with the correct value set. And this is what
this patch does:
- remap the PCI (IO port for legacy device and memory map for modern
device)
- set vtpci_ops correctly
After that, multiple process would work like a charm. (At least, it
passed my fuzzy test)
Fixes: b8f04520ad71 ("virtio: use PCI ioport API")
Fixes: d5bbeefca826 ("virtio: introduce PCI implementation structure")
Fixes: 6ba1f63b5ab0 ("virtio: support specification 1.0")
Cc: stable@dpdk.org
Reported-by: Juho Snellman <jsnell@iki.fi>
Reported-by: Yaron Illouz <yaroni@radcom.com>
Signed-off-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
We used to store the vtpci_ops at virtio_hw structure. The struct,
however, is stored in shared memory. That means only one value is
allowed. For the multiple process model, however, the address of
vtpci_ops should be different among different processes.
Take virtio PMD as example, the vtpci_ops is set by the primary
process, based on its own process space. If we access that address
from the secondary process, that would be an illegal memory access,
A crash then might happen.
To make the multiple process model work, we need store the vtpci_ops
in local memory but not in a shared memory. This is what the patch
does: a local virtio_hw_internal array of size RTE_MAX_ETHPORTS is
allocated. This new structure is used to store all these kind of
info in a non-shared memory. Current, we have:
- vtpci_ops
- rte_pci_ioport
- virtio pci mapped memory, such as common_cfg.
The later two will be done in coming patches. Later patches would also
set them correctly for secondary process, so that the multiple process
model could work.
Signed-off-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
If the primary enables the vector Rx/Tx path, the current code would
let the secondary always choose the non vector Rx/Tx path. This results
to a Rx/Tx method mismatch between primary and secondary process. Werid
errors then may happen, something like:
PMD: virtio_xmit_pkts() tx: virtqueue_enqueue error: -14
Fix it by choosing the correct Rx/Tx callbacks for the secondary process.
That is, use vector path if it's given.
Fixes: 8d8393fb1861 ("virtio: pick simple Rx/Tx")
Cc: stable@dpdk.org
Signed-off-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
Attaching and detaching ethernet ports from an application
is not the same thing as physically removing a PCI device,
so clarify the flags indicating support. All PCI devices
are assumed to be physically removable, so no flag is
necessary in the PCI layer.
Signed-off-by: Ben Walker <benjamin.walker@intel.com>
The function rte_eth_xstats_get() return an array of tuples (id,
value). The value is the statistic counter, while the id references a
name in the array returned by rte_eth_xstats_get_name().
Today, each 'id' returned by rte_eth_xstats_get() is equal to the index
in the returned array, making this value useless. It also prevents a
driver from having different indexes for names and value, like in the
example below:
rte_eth_xstats_get_name() returns:
0: "rx0_stat"
1: "rx1_stat"
2: ...
7: "rx7_stat"
8: "tx0_stat"
9: "tx1_stat"
...
15: "tx7_stat"
rte_eth_xstats_get() returns:
0: id=0, val=<stat> ("rx0_stat")
1: id=1, val=<stat> ("rx1_stat")
2: id=8, val=<stat> ("tx0_stat")
3: id=9, val=<stat> ("tx1_stat")
This patch fixes the drivers to set the 'id' in their ethdev->xstats_get()
(except e1000 which was already doing it), and fixes ethdev by not setting
the 'id' field to the index of the table for pmd-specific stats: instead,
they should just be shifted by the max number of generic statistics.
Fixes: bd6aa172cf35 ("ethdev: fetch extended statistics with integer ids")
Signed-off-by: Olivier Matz <olivier.matz@6wind.com>
Acked-by: Remy Horton <remy.horton@intel.com>
This makes struct rte_eth_dev independent of struct rte_pci_device by
replacing it with a pointer to the generic struct rte_device.
Signed-off-by: Jan Blunck <jblunck@infradead.org>
Acked-by: Shreyansh Jain <shreyansh.jain@nxp.com>
Only the drivers itself can decide if it could fill PCI information fields
of dev_info.
Signed-off-by: Jan Blunck <jblunck@infradead.org>
Acked-by: Shreyansh Jain <shreyansh.jain@nxp.com>
We don't need to depend on rte_eth_dev->pci_dev to differentiate between
the virtio_user and the virtio_pci case. Instead we can use the private
virtio_hw struct to get that information.
Signed-off-by: Jan Blunck <jblunck@infradead.org>
Acked-by: Shreyansh Jain <shreyansh.jain@nxp.com>
This adds a helper to get the rte_intr_handle from the virtio_hw. This is
safe to do since the usage of the helper is guarded by RTE_ETH_DEV_INTR_LSC
which is only set if we found a PCI device during initialization.
Signed-off-by: Jan Blunck <jblunck@infradead.org>
Acked-by: Shreyansh Jain <shreyansh.jain@nxp.com>
This is overwritten in rte_eth_dev_info_get().
Signed-off-by: Jan Blunck <jblunck@infradead.org>
Reviewed-by: David Marchand <david.marchand@6wind.com>
Add a new macro RTE_PMD_REGISTER_KMOD_DEP() that allows a driver to
declare the list of kernel modules required to run properly.
Today, most PCI drivers require uio/vfio.
Signed-off-by: Olivier Matz <olivier.matz@6wind.com>
Acked-by: Fiona Trahe <fiona.trahe@intel.com>
Acked-by: Adrien Mazarguil <adrien.mazarguil@6wind.com>
When queue number shrinks to 1 from X, the following code stops us
sending the multiple queue ctrl message:
if (nb_queues > 1) {
if (virtio_set_multiple_queues(dev, nb_queues) != 0)
return -EINVAL;
}
This ends up with still X queues being enabled, which is obviously
wrong. Fix it by replacing the check with a multiple queue enabled
or not check.
Fixes: 823ad647950a ("virtio: support multiple queues")
Signed-off-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
From the virtio spec of view, multiple-queue is always enabled/disabled
in queue pairs. DPDK somehow allows the case when Tx and Rx queue number
are different.
Currently, virtio PMD get the queue pair number from the nb_rx_queues
field, which could be an issue when Tx queue number > Rx queue number.
Say, 2 Tx queues and 1 Rx queues. This would end up with 1 quues being
enabled. Which is wrong.
The fix is straightforward. Just pick a bigger number and enable that many
of queues.
Fixes: 823ad647950a ("virtio: support multiple queues")
Signed-off-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
The "hw->started" field was introduced to stop touching queues
on restart. We never touches queues on restart any more, thus
it's safe to remove this flag.
Signed-off-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Invoking vtpci_reinit_complete() at port start stage doesn't make any
sense, instead, it should be done at the end of dev init stage.
So move it here.
Signed-off-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
The only piece of code of virtio_dev_rxtx_start() is actually doing
queue configure/setup work. So, move it to corresponding queue_setup
callback.
Once that is done, virtio_dev_rxtx_start() becomes an empty function,
thus it's being removed.
Signed-off-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
virtio_dev_vring_start() is actually doing the vring initiation job.
And the vring initiation job should be done at the dev init stage, as
stated with great details in former commit.
So move it there, and rename it to virtio_init_vring().
Signed-off-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Queue allocation should be done once, since the queue related info (such
as vring addreess) will only be informed to the vhost-user backend once
without virtio device reset.
That means, if you allocate queues again after the vhost-user negotiation,
the vhost-user backend will not be informed any more. Leading to a state
that the vring info mismatches between virtio PMD driver and vhost-backend:
the driver switches to the new address has just been allocated, while the
vhost-backend still sticks to the old address has been assigned in the init
stage.
Unfortunately, that is exactly how the virtio driver is coded so far: queue
allocation is done at queue_setup stage (when rte_eth_tx/rx_queue_setup is
invoked). This is wrong, because queue_setup can be invoked several times.
For example,
$ start_testpmd.sh ... --txq=1 --rxq=1 ...
> port stop 0
> port config all txq 1 # just trigger the queue_setup callback again
> port config all rxq 1
> port start 0
The right way to do is allocate the queues in the init stage, so that the
vring info could be persistent with the vhost-user backend.
Besides that, we should allocate max_queue pairs the device supports, but
not nr queue pairs firstly configured, to make following case work.
$ start_testpmd.sh ... --txq=1 --rxq=1 ...
> port stop 0
> port config all txq 2
> port config all rxq 2
> port start 0
Since the allocation is switched to init stage, the free should also
moved from the rx/tx_queue_release to dev close stage. That leading we
could do nothing an empty rx/tx_queue_release() implementation.
Signed-off-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
Let rxq/txq/cq be the union field of the virtqueue struct. This would
simplifies the vq allocation a bit: we don't need calculate the vq_size
any more based on the queue type.
Signed-off-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Instead of setting up a queue memzone name like "port0_rxq0", "port0_txq0",
it could be simplified a bit to something like "port0_vq0", "port0_vq1" ...
Meanwhile, the code is also simplified a bit.
Signed-off-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
This reverts commit 9a0615af7746 ("virtio: fix restart"); conflict is
manually addressed.
Kyle reported an issue with above commit
qemu-kvm: Guest moved used index from 5 to 1
with following steps,
1) Start my virtio interfaces
2) Send some traffic into/out of the interfaces
3) Stop the interfaces
4) Start the interfaces
5) Send some more traffic
And here are some quotes from Kyle's analysis,
Prior to the patch, if an interface were stopped then started, without
restarting the application, the queues would be left as-is, because
hw->started would be set to 1. Now, calling stop sets hw->started to 0,
which means the next call to start will "touch the queues". This is the
unintended side-effect that causes the problem.
We should not touch the queues once the init is done, otherwise, the vring
state of virtio PMD driver and vhost-user would be inconsistent, leading
some issue like above.
Thus this patch is reverted.
Fixes: 9a0615af7746 ("virtio: fix restart")
Reported-by: Kyle Larose <klarose@sandvine.com>
Signed-off-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
add cb_arg parameter to the _rte_eth_dev_callback_process function.
Adding a parameter to this function allows passing information
to the application when an eth device event occurs such as
a VF to PF message.
This allows the application to decide if a particular function
is permitted.
Signed-off-by: Bernard Iremonger <bernard.iremonger@intel.com>
Signed-off-by: Alex Zelezniak <alexz@att.com>
All macros related to driver registeration renamed from DRIVER_*
to RTE_PMD_*
This includes:
DRIVER_REGISTER_PCI -> RTE_PMD_REGISTER_PCI
DRIVER_REGISTER_PCI_TABLE -> RTE_PMD_REGISTER_PCI_TABLE
DRIVER_REGISTER_VDEV -> RTE_PMD_REGISTER_VDEV
DRIVER_REGISTER_PARAM_STRING -> RTE_PMD_REGISTER_PARAM_STRING
DRIVER_EXPORT_* -> RTE_PMD_EXPORT_*
Fix PMDINFOGEN tool to look for matches of RTE_PMD_REGISTER_*.
Signed-off-by: Shreyansh Jain <shreyansh.jain@nxp.com>
Acked-by: Thomas Monjalon <thomas.monjalon@6wind.com>
Add the ability to reset the virtio device in the configure callback
if the features flag changed since previous reset. This will be possible
with the introduction of offload support in next commits.
Signed-off-by: Olivier Matz <olivier.matz@6wind.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Acked-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
Move the configuration of control queue in the configure callback.
This is needed by next commit, which introduces the reinitialization
of the device in the configure callback to change the feature flags.
Therefore, the control queue will have to be restarted at the same
place.
As virtio_dev_cq_queue_setup() is called from a place where
config->max_virtqueue_pairs is not available, we need to store this in
the private structure. It replaces max_rx_queues and max_tx_queues which
have the same value. The log showing the value of max_rx_queues and
max_tx_queues is also removed since config->max_virtqueue_pairs is
already displayed above.
Signed-off-by: Olivier Matz <olivier.matz@6wind.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Acked-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
Move all code related to device initialization in a new function
virtio_init_device().
This commit brings no functional change, it prepares the next commits
that will add the offload support. For that, it will be needed to
reinitialize the device from ethdev->configure(), using this new
function.
Signed-off-by: Olivier Matz <olivier.matz@6wind.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Acked-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
Add modern device id and rename VIRTIO_PCI_DEVICEID_MIN to
VIRTIO_PCI_LEGACY_DEVICEID_NET. While at it, remove unused macros too.
Signed-off-by: Jason Wang <jasowang@redhat.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Acked-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
The driver name has been lost with the eal rework.
Restore it.
Fixes: c830cb295411 ("drivers: use PCI registration macro")
Signed-off-by: David Marchand <david.marchand@6wind.com>
Acked-by: Shreyansh Jain <shreyansh.jain@nxp.com>
Acked-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
Virtio interfaces do not currently allow the user to specify a particular
Maximum Transmission Unit (MTU). Consequently, the MTU of Virtio interfaces
is typically set to the Ethernet default value of 1500.
This is problematic in the case of cloud deployments, in which a specific
(and potentially non-standard) MTU needs to be set by a DHCP server, which
needs to be honored by all interfaces across the traffic path.To acheive
this Virtio interfaces should support setting of MTU.
In case when GRE/VXLAN tunneling is used for internal communication, there
will be an overhead added by the infrastructure in the packet over and
above the ETHER MTU of 1518. So to take care of this overhead in these
cases the DHCP server corrects the L3 MTU to 1454. But since virtio
interfaces was not having the MTU set functionality that MTU sent by the
DHCP server was ignored and the instance will still send packets with 1500
MTU which after encapsulation will become more than 1518 and eventually
gets dropped in the infrastructure.
By adding an additional 'set_mtu' function to the Virtio driver, we can
honor the MTU sent by the DHCP server. The dhcp server/controller can
then leverage this 'set_mtu' functionality to resolve the above
mentioned issue of packets getting dropped due to incorrect size.
Signed-off-by: Souvik Dey <sodey@sonusnet.com>
Reviewed-by: Mark Kavanagh <mark.b.kavanagh@intel.com>
Acked-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
We have a stats named "size_1024_1517_packets", while the code
actually counts the range "[1024, 1518]", which is obviously wrong.
The code is as follows in the function virtio_update_packet_stats.
else if (s < 1519)
stats->size_bins[6]++;
We could either fix it by correcting the "if" check in the code,
or fix it by just renaming the stats to conform to the code. The
latter solution is taken because that's what the RFC2819 suggests.
Fixes: 76d4c652e07d ("virtio: add extended stats")
Signed-off-by: Zhiyong Yang <zhiyong.yang@intel.com>
Acked-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
Remove the 'name' member from rte_pci_driver and move to generic
rte_driver.
Most of the PMD drivers were initially using DRIVER_REGISTER_PCI(<name>..)
as well as assigning a name to eth_driver.pci_drv.name member.
In this patch, only the original DRIVER_REGISTER_PCI(<name>..) name has
been populated into the rte_driver.name member - assignments through
eth_driver has been removed.
Signed-off-by: Jan Viktorin <viktorin@rehivetech.com>
[Shreyansh: Rebase and expand changes to newly added files]
Signed-off-by: Shreyansh Jain <shreyansh.jain@nxp.com>
Acked-by: David Marchand <david.marchand@6wind.com>
Simplify crypto and ethdev pci drivers init by using newly introduced
init macros and helpers.
Those drivers then don't need to register as "rte_driver"s anymore.
Exceptions:
- virtio and mlx* use RTE_INIT directly as they have custom initialization
steps.
- VDEV devices are not modified - they continue to use PMD_REGISTER_DRIVER.
Update documentation for replacing an example referring to
PMD_REGISTER_DRIVER.
Signed-off-by: David Marchand <david.marchand@6wind.com>
Signed-off-by: Shreyansh Jain <shreyansh.jain@nxp.com>
As discussed in the past release, driver names are modified
to be more consistent, and the future driver should follow
this new convention.
Driver names consist of:
"driver category"_"driver folder name"_"optional extra name".
For example:
- Crypto null driver -> "crypto_null"
- Network IXGBE VF driver -> "net_ixgbe_vf"
Signed-off-by: Pablo de Lara <pablo.de.lara.guarch@intel.com>
