Age | Commit message (Collapse) | Author | Files | Lines |
|
Type: fix
Fixes: 6bfd076
Change-Id: I8da83a8a755e5893da627c0555039c62975a5e84
Signed-off-by: Damjan Marion <damjan.marion@gmail.com>
|
|
Type: improvement
Change-Id: If20ae74658702399f626b1af7bd3519e05a5d422
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Type: improvement
Change-Id: I1911a9d6a0ca8e1d24ce7b95281886392081ffc6
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
clib_dt_read_from_sysfs is the only user of clib_dt_node_add_child, on
FreeBSD this causes a build failure due to an unused function. Place
clib_dt_node_add_child behind an include guard for linux only.
Type: fix
Change-Id: I9641ffd10d9d15e1e552f570cd386211d3b4ab39
Signed-off-by: Tom Jones <thj@freebsd.org>
|
|
Change-Id: Ief8e159b25d4fc4859c7116da6ff22c15bd3fff0
Type: feature
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
On FreeBSD the affinity returned by cpuset gives us a bitmask
documenting the available cpus. This can be modified if we are jailed or
are launched via the cpuset(1) command.
Initialise the clib bitmask otherwise it is always 0 and we cannot
configure any workers.
Type: improvement
Change-Id: I067e373ea440c6fbd03839d1cf103c8b97d8a0c6
Signed-off-by: Tom Jones <thj@freebsd.org>
|
|
Change-Id: Ibafa51f5fc98674d30e8758a0f9cc361c8b5c0fc
Type: refactor
Signed-off-by: Mohammed Hawari <mohammed@hawari.fr>
|
|
Cmake has a dedicated CMAKE_DL_LIBS variable to get libdl; use it
instead of trying to find it manually.
Type: improvement
Signed-off-by: Guillaume Solignac <gsoligna@cisco.com>
Change-Id: I9969c6be029f6a6e09cccaecd50e9eaf7d785b34
|
|
- make the format RFC9110 compliant
Type: fix
Change-Id: I4272562ca1082285a596ef866ab6c4f405c64bc5
Signed-off-by: Adrian Villin <avillin@cisco.com>
|
|
A string generated from format() may not be nul terminated.
Type: fix
Change-Id: I88452e446c3504d70758e9009c65be5466034d92
Signed-off-by: Ole Troan <otroan@employees.org>
|
|
Use get_mempolicy instead of move_pages to detect numa nodes. Avoids
issues whereby move_pages could stall if another rt thread spin waits
instead of yielding the cpu.
Type: improvement
Signed-off-by: Tatsushi Kanemoto <t-kanemoto@fujitsu.com>
Signed-off-by: Florin Coras <fcoras@cisco.com>
Change-Id: I7db2073ed82bb0b1e707130d9bd8f54bea333840
|
|
Add clib_mem_trace_dup which return copy of memory traces and
can be used to save memory trace in different formats.
Type: improvement
Change-Id: I731860cdc65e6c5255620a6778ce6c893a493b1d
Signed-off-by: Matus Fabian <matfabia@cisco.com>
|
|
After linux-v4.15, 'cpu MHz' in /proc/cpuinfo is CPU
real frequency on x86 if cpu support aperfmperf flags
more see linux commit:
https://github.com/torvalds/linux/commit/7d5905dc14a87805a59f3c5bf70173aac2bb18f8
Type: fix
Change-Id: Ib655a9a1c519104142120b343cd8ddbe0b6e50e6
Signed-off-by: Xiaoming Jiang <jiangxiaoming@outlook.com>
|
|
In Arm platform, targeted prefetching[1] is supported. Adding this in infra
to use it in future optimisation.
[1] https://developer.arm.com/documentation/101458/2404/Optimize/Prefetching-with---builtin-prefetch
Type: improvement
Change-Id: Id3e2a274628289707840a6b2ba4b83c5e4add60b
Signed-off-by: Niyaz Murshed <niyaz.murshed@arm.com>
|
|
Running VPP on a NUMA system with 5.19+ kernel outputs the following
error messages. 'show physmem' command confirms that VPP falls back to
using normal 4K pages instead of the preallocated 1G huge pages. The
root cause is that VPP uses move_pages()[1] to get the huge page node
information. However, this misbehaves on the 5.19+ kernel due to changes
introduced in its implementation[2].
Our proposed fix is retry obtaining NUMA node info with
get_mempolicy()[3] only if we see -ENOENT returned in status from
move_pages() and huge pages are used. Additionally, we use mincore()[4]
to check if pages are allocated and in memory to avoid the possibility
of get_mempolicy() falsely allocating a new page.
buffer [warn ]: numa[1] falling back to non-hugepage backed buffer pool ()
vpp# show physmem
used-pages 2 reserved-pages 16 default-page-size 1G lookup-page-size 4K
arena 'buffers-numa-0' pages 1 subpage-size 1G numa-node 0 shared fd 5
arena 'buffers-numa-1' pages 1 subpage-size 4K numa-node 1 shared fd 6
[1] https://man7.org/linux/man-pages/man2/move_pages.2.html
[2] https://lore.kernel.org/linux-mm/91da2c3b-96f1-bb03-8fff-4c38f31cb9be@huawei.com/
[3] https://man7.org/linux/man-pages/man2/get_mempolicy.2.html
[4] https://man7.org/linux/man-pages/man2/mincore.2.html
Type: fix
Signed-off-by: Jay Wang <jay.wang2@arm.com>
Change-Id: Ia423745423bb080404292333ef95455a4950ce0a
|
|
Use standard macros CPU_ISSET_S and CPU_ZERO_S and make code more
portable.
Type: improvement
Change-Id: I8e6a0fd79b90de8c6c33cad8882680eae2ca50ef
Signed-off-by: Renato Botelho do Couto <renato@netgate.com>
|
|
os_exit() and os_puts() are not exported, preventing linkage with
libvppinfra if those are not defined.
Type: fix
Change-Id: I7e3b583147e1348f078afeac3f793fe150405009
Signed-off-by: Benoît Ganne <bganne@cisco.com>
|
|
- use libunwrap which seems to be industry standard
- display traceback on console if running interactive or with syslog
disabled (color output unless nocolor specified)
- print hexdump of offending code
- print library filename for each stack frame
Type: improvement
Change-Id: I61d3056251b87076be0578ccda300aa311c222ef
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
On non-glibc systems, execinfo is the only option available, but the lib
is old and can crash when unwinding. We now can use libunwind to unroll
it instead of using execinfo.h.
Type: improvement
Signed-off-by: Guillaume Solignac <gsoligna@cisco.com>
Change-Id: I8b55f7aca97261a2efb5dca998889d5e0645939a
|
|
Type: improvement
Change-Id: I5253f2b1b51493f2d4d0f451ad4c2208fd8f6bf2
Signed-off-by: Tom Jones <thj@freebsd.org>
|
|
Fix out of buffer access after mhash_unset
Add format mhash pair functions
Add related mhash tests, similar to hash ones
Type: fix
Signed-off-by: Vladislav Grishenko <themiron@yandex-team.ru>
Change-Id: Idbefd7d32c4cd16b55d84ad4006c38251b4e2c33
|
|
Upgrade cJSON library to patchlevel 17.
Replace internal realloc added earlier 36217e3ca.
Type: fix
Fixes: 36217e3ca
Change-Id: I7d8a80dc4241e9f952895d24adca8fa2d873e746
Signed-off-by: Ole Troan <otroan@employees.org>
|
|
This reverts commit 71c32a898941e32b5d4f865b50fbe775560c582d.
Type: fix
Reason for revert: vnet pinning is not considered in this patch.
This causes keywords 'workers' and 'skip-cores' to be broken,
as well as keyword 'main-core auto' introduced in this patch.
If this patch is ever reconsidered, it must account for vnet
pinning fix in commit https://gerrit.fd.io/r/c/vpp/+/40711.
Change-Id: I1f3154a6c7e830b100f824375aa00e95b192f7f3
Signed-off-by: hsandid <halsandi@cisco.com>
|
|
Type: improvement
Change-Id: I5fa90dd26d6cac2cefee009c97dd5ff8dac3ee1e
Signed-off-by: Tom Jones <thj@freebsd.org>
|
|
Type: improvement
Change-Id: Ie3d6cba886e0779949e8d574f82cff62abc7ebed
Signed-off-by: Tom Jones <thj@freebsd.org>
|
|
FreeBSD has a different interface to request memory mapping information.
Add a FreeBSD specific method for reading physical addresses and make it
available at build time.
Type: improvement
Change-Id: I3588dde8e0a6f6d53333040245341ed09cebef9d
Signed-off-by: Tom Jones <thj@freebsd.org>
|
|
Add a unix method for getting the current executable name. This is
implemented to match the readlink api for existing calls.
Type: improvement
Change-Id: Id06a55892d09d0b305a56b55a424f53ffb685a72
Signed-off-by: Tom Jones <thj@freebsd.org>
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Change-Id: Ic1fa3bd164e80c2ca1146be001870da0238a5f2e
Type: improvement
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Use mask_load_zero to avoid out-of-buffer reads in vectorized
function versions.
Type: fix
Signed-off-by: Dmitry Valter <d-valter@yandex-team.com>
Change-Id: I12bcb817ccf2db210c1c99fdfa444dc3f540035b
|
|
vl_api_prefix_t.len is 1 byte only, but unformat %d writes 4 bytes
add helper functions unformat_u(8|16) which don't write more than
appropriate amount of bytes
fix other similar errors in vpp/api/types.c
Type: fix
Change-Id: I74a61a377147c373f8c25ed083052b2287763c39
Signed-off-by: Klement Sekera <klement.sekera@gmail.com>
|
|
Properly handle small buffers in the AVX256 version of array_mask_u32.
Type: fix
Signed-off-by: Dmitry Valter <d-valter@yandex-team.com>
Fixes: f62ed3f9c1ec3e8db36f63d6a54f46b7bea43723
Change-Id: Idf1cb43ccf37bd7c439d11e4f68fe30064c6d09a
|
|
Type: feature
Auto core pinning now fetches vpp cpu affinity list
using pthread api. This enables us to do core-pinning in
environments where the host cpu list does not necessarily align
with cpus available to vpp
Change-Id: Ife8c2a2351c08c5c6c4fdf7c729eeff2697bc39a
Signed-off-by: hsandid <halsandi@cisco.com>
|
|
Type: improvement
Change-Id: I303a8a7333d44f6eab902d4d61023cfa96760dd2
Signed-off-by: Tom Jones <thj@freebsd.org>
|
|
Avoid direct sysfs reads when possible...
Type: improvement
Change-Id: I2b84cd18f3da47925d068951f24b79b5b6e20bb1
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Type: refactor
Change-Id: I5235bf3e9aff58af6ba2c14e8c6529c4fc9ec86c
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Change-Id: I0fcda3e7afaab037bd12d0079d6639c6cbe8647e
Type: improvement
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Type: improvement
Change-Id: I6f99f09c7724ce656a4f41a1d5f9c88d74c00faf
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
FreeBSD has its own set of syscalls for getting current CPU and NUMA
domain information. Stub out these calls and return CPU 0 and NUMA domain
0 as placeholders until we bring in FreeBSD specific calls.
Type: improvement
Change-Id: Id61df0273b0bcc6acf4844ee626e4f246f9f217b
Signed-off-by: Tom Jones <thj@freebsd.org>
|
|
Type: improvement
Change-Id: I3a48ec4fd4a4063b6e3792b87def7bb796fc947c
Signed-off-by: Tom Jones <thj@freebsd.org>
|
|
Working from the implementation in linux/mem.c add FreeBSD specific
functionality. This duplicates parts of the Linux implementation and a
depuplication job could be run in the future.
Stub out some parts of the API for now, they are either use unavailable
features on FreeBSD or require further implementation than this initial
implementation.
Type: improvement
Change-Id: I1e443e32304d19776a9a4d5e34adfa16ec919427
Signed-off-by: Tom Jones <thj@freebsd.org>
|
|
As a placeholder stub out clib_perf functions on FreeBSD, the interface
for performance counters on FreeBSD is different and will require its
own implementation.
Type: improvement
Change-Id: I61511ebe5ee1fd6fe93d3179c7843c811f47a846
Signed-off-by: Tom Jones <thj@freebsd.org>
|
|
FreeBSD has support netlink since 13.2, but keeps its header in a
slightly different place. Correctly include netlink/netlink.h for
FreeBSD.
Type: improvement
Change-Id: Iec78ca72db74ea7f9c874e7039c7d17b6321f9dc
Signed-off-by: Tom Jones <thj@freebsd.org>
|
|
Place Linux specific features behind CLIB_LINUX. The previous #ifdef
calls would still pass even if CLIB_LINUX is set to 0 in clib.h, instead
use #if.
Type: improvement
Change-Id: Ibe4424eee4bc0065f54764b017b9a56a3e854380
Signed-off-by: Tom Jones <thj@freebsd.org>
|
|
Finding the program counter in the ucontext struct is platform and
architecture specific, place the Linux checks inside an #ifdef and add a
look up for the FreeBSD amd64 specific naming.
Type: improvement
Change-Id: I42fcef5f20227c23d84acee336e37c4870146bb4
Signed-off-by: Tom Jones <thj@freebsd.org>
|
|
Type: improvement
Change-Id: If136e0854dbaecfa62fc009ca2b0b94e8d9a56f2
Signed-off-by: Tom Jones <thj@freebsd.org>
|
|
Don't use the MAP_HUGETLB mmap flag on FreeBSD as it isn't available on
the platform.
Type: improvement
Change-Id: Ie2046601c693f9ef12cf8113d6dbd87c322f1cf0
Signed-off-by: Tom Jones <thj@freebsd.org>
|
|
FreeBSD doesn't support sysfs or preallocating hugepages directly. In
lieu of refactoring clib_sysfs calls place this one behind an #ifdef
linux.
Type: improvement
Change-Id: I4079ca70175e4cec44d913db2b0407a3ca6306c9
Signed-off-by: Tom Jones <thj@freebsd.org>
|
|
FreeBSD has platform specific APIs for interfacing with performance
counters. Until we bring in FreeBSD support, don't build perfmon on
platforms which aren't linux.
Type: improvement
Change-Id: Iabc418922c0d2e851c6098508761135f53433ae8
Signed-off-by: Tom Jones <thj@freebsd.org>
|
|
The SIGPWR signal is specific to Linux, place it behind a Linux flag to
help the build on FreeBSD.
Type: improvement
Change-Id: Ie36c2ceb201afff98e92bbe64595c3953b11c457
Signed-off-by: Tom Jones <thj@freebsd.org>
|
|
Handle non-even n_elts for the larger array instead of reading past
the source buffer.
Type: fix
Fixes: f62ed3f9c1ec3e8db36f63d6a54f46b7bea43723
Signed-off-by: Dmitry Valter <d-valter@yandex-team.com>
Change-Id: Ic1708a3f33fe71ca752345b5c77b6ae7a2d42bcd
|