David Marchand [Tue, 16 Feb 2016 20:37:03 +0000 (21:37 +0100)]
eal: introduce PCI ioport API
Most of the code is inspired on virtio driver.
rte_pci_ioport structure is filled at map time with anything needed for later
read / write calls.
At the moment, base field is used to store a x86 ioport (uint16_t) and will
be reused for other arches.
Signed-off-by: David Marchand <david.marchand@6wind.com> Tested-by: Santosh Shukla <sshukla@mvista.com> Reviewed-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
David Marchand [Tue, 16 Feb 2016 20:37:02 +0000 (21:37 +0100)]
virtio: fix check when mapping PCI resources
According to the api, rte_eal_pci_map_device is only successful when
returning 0.
Fixes: 6ba1f63b5ab0 ("virtio: support specification 1.0") Signed-off-by: David Marchand <david.marchand@6wind.com> Acked-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
David Marchand [Tue, 16 Feb 2016 20:37:01 +0000 (21:37 +0100)]
virtio: fix FreeBSD build
Fixes: c52afa68d763 ("virtio: move left PCI stuff in the right file") Signed-off-by: David Marchand <david.marchand@6wind.com> Acked-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
Thomas Monjalon [Tue, 2 Feb 2016 23:10:26 +0000 (00:10 +0100)]
eal: remove compiler optimization workaround
The compiler optimization was disabled a long time ago
without describing what was the exact issue.
Maybe it does not apply anymore.
As it looks unneeded, let's remove this strange pragma.
Signed-off-by: Thomas Monjalon <thomas.monjalon@6wind.com>
Thomas Monjalon [Tue, 2 Feb 2016 23:10:24 +0000 (00:10 +0100)]
eal/arm: adapt CPU flags check to the arch
The structure feature_entry does not need leaf/subleaf
which were copied from x86 CPUID implementation.
On x86, a valid flag is detected with the non-zero leaf value.
This check is replaced by a check with a dummy "none" register.
Signed-off-by: Thomas Monjalon <thomas.monjalon@6wind.com> Acked-by: Jerin Jacob <jerin.jacob@caviumnetworks.com> Tested-by: Jerin Jacob <jerin.jacob@caviumnetworks.com>
Thomas Monjalon [Tue, 2 Feb 2016 23:10:23 +0000 (00:10 +0100)]
eal: move CPU flag functions out of headers
The patch c344eab3ee has moved the hardware definition of CPU flags.
Now the functions checking these hardware flags are also moved.
The function rte_cpu_get_flag_enabled() is no more inline.
The benefits are:
- remove rte_cpu_feature_table from the ABI (recently added)
- hide hardware details from the API
- allow to adapt structures per arch (done in next patch)
Signed-off-by: Thomas Monjalon <thomas.monjalon@6wind.com> Acked-by: Jerin Jacob <jerin.jacob@caviumnetworks.com>
Thomas Monjalon [Tue, 2 Feb 2016 22:59:49 +0000 (23:59 +0100)]
eal: get CPU flag name
The new function rte_cpu_get_flag_name() is added to the EAL API.
It is implemented (duplicated) in each arch because the next patch
will remove the public exposure of the feature tables.
Signed-off-by: Thomas Monjalon <thomas.monjalon@6wind.com>
Thomas Monjalon [Fri, 5 Feb 2016 14:43:56 +0000 (15:43 +0100)]
examples: fix build dependencies
When building for ARM some examples were failing to compile because
of some dependencies disabled.
Declaring these dependencies prevent from trying to compile some
not supported examples.
Signed-off-by: Thomas Monjalon <thomas.monjalon@6wind.com>
Jerin Jacob [Fri, 12 Feb 2016 11:13:51 +0000 (16:43 +0530)]
examples/distributor: fix build for non-x86 arch
_mm_prefetch is defined only in x86 compilers.
Use rte_prefetch_non_temporal() abstraction instead of _mm_prefetch(x, 0)
to in-order to build distributor application for non x86 platforms
Signed-off-by: Jerin Jacob <jerin.jacob@caviumnetworks.com>
Jerin Jacob [Fri, 12 Feb 2016 11:13:50 +0000 (16:43 +0530)]
eal: introduce non-temporal prefetch
non-temporal/transient/stream version of rte_prefetch0()
The non-temporal prefetch is intended as a prefetch hint that processor
will use the prefetched data only once or short period,
unlike the rte_prefetch0() function which imply that
prefetched data to use repeatedly.
Signed-off-by: Jerin Jacob <jerin.jacob@caviumnetworks.com> Acked-by: Jan Viktorin <viktorin@rehivetech.com>
Jerin Jacob [Fri, 29 Jan 2016 07:45:52 +0000 (13:15 +0530)]
eal: introduce new cache line macros
- RTE_CACHE_LINE_MIN_SIZE(Supported minimum cache line size)
- __rte_cache_min_aligned(Force minimum cache line alignment)
- RTE_CACHE_LINE_SIZE_LOG2(Express cache line size in terms of log2)
Suggested-by: Konstantin Ananyev <konstantin.ananyev@intel.com> Signed-off-by: Jerin Jacob <jerin.jacob@caviumnetworks.com>
Jerin Jacob [Mon, 7 Dec 2015 14:22:50 +0000 (19:52 +0530)]
config: clean cache line size selection scheme
by default, all the targets will be configured with the 64-byte cache line
size, targets which have different cache line size can be overridden
through target specific config file.
Selected ThunderX and power8 as CONFIG_RTE_CACHE_LINE_SIZE=128 targets
based on existing configuration.
Signed-off-by: Jerin Jacob <jerin.jacob@caviumnetworks.com> Acked-by: Thomas Monjalon <thomas.monjalon@6wind.com>
Thomas Monjalon [Fri, 5 Feb 2016 21:20:08 +0000 (22:20 +0100)]
config: remove obsolete machine descriptions
More and more machines and architectures are added without keeping
the lists up-to-date.
Replace the lists with a pointer to the reference directory.
The same kind of pointer is used for the supported compilers and environments.
Signed-off-by: Thomas Monjalon <thomas.monjalon@6wind.com> Acked-by: Konstantin Ananyev <konstantin.ananyev@intel.com>
Bruce Richardson [Wed, 10 Feb 2016 17:02:12 +0000 (17:02 +0000)]
doc: rename release notes 2.3 to 16.04
Updated release documentation to reflect new numbering scheme.
Signed-off-by: Bruce Richardson <bruce.richardson@intel.com> Signed-off-by: John McNamara <john.mcnamara@intel.com> Acked-by: Thomas Monjalon <thomas.monjalon@6wind.com>
Also, added zero padding to the month so that it appear as 16.04 and
not 16.4 in "make showversion" and rte_version().
Signed-off-by: Bruce Richardson <bruce.richardson@intel.com> Signed-off-by: John McNamara <john.mcnamara@intel.com> Acked-by: Thomas Monjalon <thomas.monjalon@6wind.com>
Thomas Monjalon [Mon, 8 Feb 2016 10:30:07 +0000 (11:30 +0100)]
doc: drop old naming of the project
It was requested by Intel, more than one year ago, to replace the name
"Intel DPDK" by "DPDK".
Some references to the old name were still in some docs and code comments,
leading to confusion.
Fixes: ac8ada004c12 ("doc: remove Intel references from release notes") Signed-off-by: Thomas Monjalon <thomas.monjalon@6wind.com> Acked-by: Bruce Richardson <bruce.richardson@intel.com>
Kamil Rytarowski [Thu, 28 Jan 2016 13:13:54 +0000 (14:13 +0100)]
eal/linux: support built-in kernel modules
Currently rte_eal_check_module() detects Linux kernel modules via reading
/proc/modules. Built-in ones aren't listed there and therefore they are not
being found.
Add support for checking built-in modules with parsing the sysfs files
This commit obsoletes the /proc/modules parsing approach.
Signed-off-by: Kamil Rytarowski <kamil.rytarowski@caviumnetworks.com> Acked-by: David Marchand <david.marchand@6wind.com> Acked-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
Kamil Rytarowski [Thu, 28 Jan 2016 13:13:53 +0000 (14:13 +0100)]
tools: support binding to built-in kernel modules
Currently dpdk_nic_bind.py detects Linux kernel modules via reading
/proc/modules. Built-in ones aren't listed there and therefore they are
not being found by the script.
Add support for checking built-in modules with parsing the sysfs files.
This commit obsoletes the /proc/modules parsing approach.
Signed-off-by: Kamil Rytarowski <kamil.rytarowski@caviumnetworks.com> Acked-by: Yuanhan Liu <yuanhan.liu@linux.intel.com>
Zhihong Wang [Thu, 4 Feb 2016 02:12:34 +0000 (21:12 -0500)]
eal/x86: fix build with clang for old AVX
When configuring RTE_MACHINE to "default", rte_memcpy implementation
is the default one (old AVX).
In this code, clang raises a warning thanks to -Wsometimes-uninitialized:
rte_memcpy.h:838:6: error:
variable 'srcofs' is used uninitialized whenever 'if' condition is false
if (dstofss > 0) {
^~~~~~~~~~~
rte_memcpy.h:849:6: note: uninitialized use occurs here
if (srcofs == 0) {
^~~~~~
It is fixed by moving srcofs initialization out of the condition.
Also dstofss calculation is corrected.
Fixes: 1ae817f9f887 ("eal/x86: tune memcpy for platforms without AVX512") Reported-by: Pablo de Lara <pablo.de.lara.guarch@intel.com> Signed-off-by: Zhihong Wang <zhihong.wang@intel.com>
Yuanhan Liu [Tue, 2 Feb 2016 13:48:19 +0000 (21:48 +0800)]
virtio: support specification 1.0
Modern (v1.0) virtio pci device defines several pci capabilities.
Each cap has a configure structure corresponding to it, and the
cap.bar and cap.offset fields tell us where to find it.
Firstly, we map the pci resources by rte_eal_pci_map_device().
We then could easily locate a cfg structure by:
Therefore, the entrance of enabling modern (v1.0) pci device support
is to iterate the pci capability lists, and to locate some configs
we care; and they are:
- common cfg
For generic virtio and virtqueue configuration, such as setting/getting
features, enabling a specific queue, and so on.
- nofity cfg
Combining with `queue_notify_off' from common cfg, we could use it to
notify a specific virt queue.
- device cfg
Where virtio_net_config structure is located.
- isr cfg
Where to read isr (interrupt status).
If any of above cap is not found, we fallback to the legacy virtio
handling.
If succeed, hw->vtpci_ops is assigned to modern_ops, where all
operations are implemented by reading/writing a (or few) specific
configuration space from above 4 cfg structures. And that's basically
how this patch works.
Besides those changes, virtio 1.0 introduces a new status field:
FEATURES_OK, which is set after features negotiation is done.
Yuanhan Liu [Tue, 2 Feb 2016 13:48:18 +0000 (21:48 +0800)]
pci: export device mapping functions
Normally we could set RTE_PCI_DRV_NEED_MAPPING flag so that eal will
invoke pci_map_device internally for us. From that point view, there
is no need to export pci_map_device.
However, for virtio pmd driver, which is designed to work without
binding UIO (or something similar first), pci_map_device() will fail,
which ends up with virtio pmd driver being skipped. Therefore, we can
not set RTE_PCI_DRV_NEED_MAPPING blindly at virtio pmd driver.
Therefore, this patch exports pci_map_device, and let virtio pmd call
it when necessary.
Yuanhan Liu [Tue, 2 Feb 2016 13:48:17 +0000 (21:48 +0800)]
virtio: retrieve header size from device setting
The mergeable virtio net hdr format has been the standard and the
only virtio net hdr format since virtio 1.0. Therefore, we can
not hardcode hdr_size to "sizeof(struct virtio_net_hdr)" any more
at virtio_recv_pkts(), otherwise, there would be a mismatch of
hdr size from rte_vhost_enqueue_burst() and virtio_recv_pkts(),
leading a packet corruption.
Instead, we should retrieve it from hw->vtnet_hdr_size; we will
do proper settings at eth_virtio_dev_init() in later patches.
Yuanhan Liu [Tue, 2 Feb 2016 13:48:14 +0000 (21:48 +0800)]
virtio: introduce PCI implementation structure
Introduce struct virtio_pci_ops, to let legacy virtio (v0.95) and
modern virtio (1.0) have different implementation regarding to a
specific pci action, such as read host status.
With that, this patch reimplements all exported pci functions, in
a way like:
So that we need pay attention to those pci related functions only
while adding virtio 1.0 support.
This patch introduced a new vtpci function, vtpci_init(), to do
proper virtio pci settings. It's pretty simple so far: just sets
hw->vtpci_ops to legacy_ops as we don't support 1.0 yet.
Ferruh Yigit [Thu, 28 Jan 2016 12:20:23 +0000 (12:20 +0000)]
eal: move cpu flags out of headers
Move cpu_feature_table array from arch specific rte_cpuflags.h files to
new arch specific rte_cpuflags.c files.
Main motivation is to escape from static variable declarations in
header files. cpu_feature_table has many copies in final binary, even
exist in some object files that does not use this variable at all.
And this can be a sample to create architecture specific source files
and move some functions which are not performance sensitive from
architecture header files to source files.
Anatoly Burakov [Thu, 28 Jan 2016 11:57:54 +0000 (11:57 +0000)]
vfio: support no-IOMMU mode
This commit is adding a generic mechanism to support multiple IOMMU
types. For now, it's only type 1 (x86 IOMMU) and no-IOMMU (a special
VFIO mode that doesn't use IOMMU at all), but it's easily extended
by adding necessary definitions to eal_vfio.h, and DMA mapping
functions to eal_pci_vfio.c.
Since type 1 IOMMU module is no longer necessary to have VFIO,
we fix the module check to check for vfio-pci instead. It's not
ideal and triggers VFIO checks more often (and thus produces more
error output, which was the reason behind the module check in the
first place), so we compensate for that by providing more verbose
logging, indicating whether VFIO initialization has succeeded or
failed.
Zhihong Wang [Mon, 18 Jan 2016 03:05:12 +0000 (22:05 -0500)]
eal/x86: optimize memcpy for AVX512 platforms
Implement AVX512 memcpy and choose the right implementation based on
predefined macros, to make full utilization of hardware resources and
deliver high performance.
In current DPDK, memcpy holds a large proportion of execution time in
libs like Vhost, especially for large packets, and this patch can bring
considerable benefits for AVX512 platforms.
The implementation is based on the current DPDK memcpy framework, some
background introduction can be found in these threads:
http://dpdk.org/ml/archives/dev/2014-November/008158.html
http://dpdk.org/ml/archives/dev/2015-January/011800.html
Signed-off-by: Zhihong Wang <zhihong.wang@intel.com>
Zhihong Wang [Wed, 30 Dec 2015 21:59:51 +0000 (16:59 -0500)]
examples/l3fwd: handle SIGINT and SIGTERM
Handle SIGINT and SIGTERM in l3fwd.
Signed-off-by: Zhihong Wang <zhihong.wang@intel.com> Acked-by: Michael Qiu <michael.qiu@intel.com> Acked-by: Konstantin Ananyev <konstantin.ananyev@intel.com>
Zhihong Wang [Wed, 30 Dec 2015 21:59:50 +0000 (16:59 -0500)]
examples/l2fwd: handle SIGINT and SIGTERM
Handle SIGINT and SIGTERM in l2fwd.
Signed-off-by: Zhihong Wang <zhihong.wang@intel.com> Acked-by: Michael Qiu <michael.qiu@intel.com> Acked-by: Stephen Hemminger <stephen@networkplumber.org>
Depending on non-doc targets being built before and the setting of DESTDIR
the copy of the examples dir being part of install-doc could in some cases
fail with a non existent "$(DESTDIR)$(datadir)" target directory.
Add the conditional rte_mkdir for that to avoid the issue.
Signed-off-by: Christian Ehrhardt <christian.ehrhardt@canonical.com> Acked-by: Thomas Monjalon <thomas.monjalon@6wind.com>
Wenzhuo Lu [Fri, 20 Nov 2015 07:17:52 +0000 (15:17 +0800)]
ixgbe/base: move X550 MDIO clock speed init
The x550 MDIO clock speed must be configured prior to first MDIO read or
write. The default MDIO clock speed is not valid, therefore the driver
is configuring a valid speed prior to reading the copper PHY device id.
Wenzhuo Lu [Fri, 20 Nov 2015 07:17:56 +0000 (15:17 +0800)]
ixgbe/base: prevent X550 KR PHY reset in init
This patch removes KR PHY reset from ixgbe_init_phy_ops_X550em. Since
this function is meant to initialize function pointers for detected PHY
type. Internal PHY reset was moved to ixgbe_setup_internal_phy_t_x550em
which will now detect which mode does internal PHY work in, and setup it
as required.
Wenzhuo Lu [Fri, 20 Nov 2015 07:17:43 +0000 (15:17 +0800)]
ixgbe/base: avoid needless copper PHY access
Avoid a needless PHY access on copper phys to save the 10ms wait
time for each PHY access. A helper function is introduced to
actually do the register access and process the contents.
Wenzhuo Lu [Fri, 20 Nov 2015 07:17:50 +0000 (15:17 +0800)]
ixgbe/base: add flow director drop queue
This patch adds ixgbe_set_fdir_drop_queue_82599 for enabling and
setting flow director drop queue, and adds sets drop no match in
ixgbe_init_fdir_perfect_82599 for x550.
Wenzhuo Lu [Fri, 20 Nov 2015 07:17:44 +0000 (15:17 +0800)]
ixgbe/base: remove wait and check flow director signature addition
Waiting for FDIRCMD completion is an expensive thing to do in the
transmit hot path. This wait was added to catch problems with perfect
filter rules, and, at least in the Linux driver, there is no error
check anyway, so there is no point to adding the delay. So do not wait
for completion. Change the return of the function to void, since it has
no meaningful return value.
Wenzhuo Lu [Fri, 20 Nov 2015 07:17:48 +0000 (15:17 +0800)]
ixgbe/base: add flow control ethertype for filtering
This patch adds the flow control ethertype to the defines for the
ETQF filter list. This only adds the define. Each driver
can add this ethertype to the filter. This is needed to prevent
denial of service by malicious VFs sending out flow control
packets.
Wenzhuo Lu [Fri, 20 Nov 2015 07:17:53 +0000 (15:17 +0800)]
ixgbe/base: fix Tx hang in CEE mode
Currently credit_refill and credit_max could be zero for a TC and that
is causing Tx hang for CEE mode configuration, so to fix that have at
min credit assigned to a TC and that is as what IEEE mode already does.
Fixes: 29c673401c4d ("doc: improve Linux guide layout") Signed-off-by: John McNamara <john.mcnamara@intel.com> Acked-by: Bernard Iremonger <bernard.iremonger@intel.com>
Thomas Monjalon [Thu, 17 Dec 2015 11:11:38 +0000 (12:11 +0100)]
doc: remove DPDK from guide titles
In HTML and PDF guides, it is clear in the header that the doc
is related to the DPDK.
So "DPDK" is redundant and can be removed from FAQ and release notes
titles to improve consistency.
Signed-off-by: Thomas Monjalon <thomas.monjalon@6wind.com> Acked-by: John McNamara <john.mcnamara@intel.com>
Jingjing Wu [Tue, 15 Dec 2015 16:23:07 +0000 (00:23 +0800)]
i40e: fix flow director index sign
Coverity issue reported like
CID 119268 (#1 of 1): Unintended sign extension
(SIGN_EXTENSION)sign_extension: Suspicious implicit sign extension:
vsi_id with type unsigned short (16 bits, unsigned) is promoted in
vsi_id << 23 to type int (32 bits, signed), then sign-extended to type
unsigned long (64 bits, unsigned). If vsi_id << 23 is greater than
0x7FFFFFFF, the upper bits of the result will all be 1.
Fixes: 88ebc2b7f976 ("i40e: extend flow director to support VF") Signed-off-by: Jingjing Wu <jingjing.wu@intel.com> Acked-by: John McNamara <john.mcnamara@intel.com>
Jingjing Wu [Tue, 15 Dec 2015 14:52:48 +0000 (22:52 +0800)]
i40e: fix max frame size
In FreeBsd driver, the max frame size is changed to MTU, but not
keep the default value defined in DataSheet. When DPDK runs on that
NIC, the configured value is not expected.
This patch sets the max frame size to default when initialization.
Robin Jarry [Tue, 15 Dec 2015 16:05:01 +0000 (17:05 +0100)]
ixgbe: restore imissed stat counter
This counter was left unmodified. Restore it in ixgbe_dev_stats_get.
The ierrors counter still includes imissed for ixgbe. This behaviour is
not consistent amongst all drivers. Another patch may be needed to unify
the meaning of the ierrors counter.
Fixes: 5e50ad1c1b63 ("ixgbe: add specific stats") Signed-off-by: Robin Jarry <robin.jarry@6wind.com> Acked-by: Harry van Haaren <harry.van.haaren@intel.com>
Panu Matilainen [Tue, 15 Dec 2015 13:55:15 +0000 (15:55 +0200)]
scripts: fix ABI validator when revision is a tag
Commit 9cbae2aa64eb managed to break the only previously supported
case where a tag is used as a revision, due to git show output
differing between tags and other objects. The hash is on the last
line of the output in both cases though so just grab that.
Fixes: 9cbae2aa64eb ("scripts: support any git revisions as ABI validation range") Signed-off-by: Panu Matilainen <pmatilai@redhat.com> Acked-by: Neil Horman <nhorman@tuxdriver.com>
John McNamara [Tue, 15 Dec 2015 11:12:01 +0000 (11:12 +0000)]
doc: remove unused references from faq
The faq refers to Linux*, with an asterisk, without any equivalent
note or footnote. This is a legacy from older versions of the docs.
This update removes it.
Signed-off-by: John McNamara <john.mcnamara@intel.com>
John McNamara [Tue, 15 Dec 2015 10:10:49 +0000 (10:10 +0000)]
doc: clean up index files
Remove **Contents** and |Today| from the rst doc index files since
these are already added automatically to PDF files and are of
little value to the Html files where the Contents is shown in a
sidebar.
Signed-off-by: John McNamara <john.mcnamara@intel.com> Acked-by: Harry van Haaren <harry.van.haaren@intel.com>
Harry van Haaren [Tue, 15 Dec 2015 12:10:30 +0000 (12:10 +0000)]
doc: add statistics fixes in release notes
This patch updates the release notes to include the changes
made (by me), which were not appropriately documented at the time.
Hence, this patch fixes a number of missing docs,
Fixes: e81a315e5dc5 ("ixgbe: add MAC short packet discard count to Rx errors") Fixes: 48dd1a82a615 ("ixgbe: remove mac fault counts from Rx errors") Fixes: 256ff05a9cae ("ixgbe: fix Rx errors statistics for UDP checksum")
Also, the CRC byte removal was not added to release notes, so Fixes: 156c5a8cf913 ("e1000: remove CRC size from byte counters") Fixes: c03fcee9abbd ("ixgbe: remove CRC size from byte counters") Fixes: 0834d1524dee ("i40e: remove CRC size from byte counters") Signed-off-by: Harry van Haaren <harry.van.haaren@intel.com> Acked-by: John McNamara <john.mcnamara@intel.com>
git.droids-corp.org / dpdk.git / log