Age | Commit message (Collapse) | Author | Files | Lines |
|
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
|
|
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
|
|
Avoid direct sysfs reads when possible...
Type: improvement
Change-Id: I2b84cd18f3da47925d068951f24b79b5b6e20bb1
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
more generic version of clib_sysfs_link_to_name with support for
format strings...
Type: improvement
Change-Id: I0cb263748970378c661415196eb7e08450370677
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Type: fix
Signed-off-by: Dave Barach <dave@barachs.net>
Change-Id: Idbdfdf2d3fdbb64366f50d5a7458c4073a4f2746
|
|
Account for the potential of sysconf() returning -1 if it can not
get the page size and make it a fatal error.
Coverity: 277313
Type: fix
Signed-off-by: Andrew Yourtchenko <ayourtch@gmail.com>
Change-Id: I8cae6a35ec2f745c37f1fe6557e5fa66720b4628
|
|
fixes coverity 282527
Type: fix
Fixes: fecb2524ab
Signed-off-by: Andrew Yourtchenko <ayourtch@gmail.com>
Change-Id: I9ac72c3802e66369a8f24c92451e33f22c058f24
|
|
Type: fix
fixes: 561ae5d
Change-Id: I0d98f5b43bc9ab5d31463b285177a11a10b864d2
Signed-off-by: Damjan Marion <dmarion@me.com>
|
|
Type: refactor
Change-Id: I5ca142ec1557d5b5c3806b43553ad9d3b5ea1112
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Use of _vec_len() to set vector length breaks address sanitizer.
Users should use vec_set_len(), vec_inc_len(), vec_dec_len () instead.
Type: improvement
Change-Id: I441ae948771eb21c23a61f3ff9163bdad74a2cb8
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
The release of the map_lock is a bit too early, it needs to wrap
mprotect(), otherwise the access rights of hdr may be deprived of
other worker threads.
Type: fix
Signed-off-by: liuyacan <liuyacan@corp.netease.com>
Change-Id: I9cb89d49a097ec6d5ae9fd2af2bddf967ec946e9
|
|
i.e.
memory {
default-hugepage-size 1G
}
Type: improvement
Change-Id: I822afb51712ae92f4e4992b8ffa33dcb15ccaef1
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
* Add clib_socket_init support for abstract sockets
if name starts with an '@'
* Add clib_socket_init_netns to open socket in netns
* Add clib_netns_open
Type: feature
Change-Id: I89637ad657c702ec38ddecb5c03a1673d0dfb104
Signed-off-by: Nathan Skrzypczak <nathan.skrzypczak@gmail.com>
|
|
while one mprotect PROT_NONE on hdr->next or hdr->prev,
the other one with the PROT_NONE is unmap at the same time,
cause SIGSEGV.
Type: fix
Signed-off-by: arikachen <eaglesora@gmail.com>
Change-Id: I21c0497da140c9654b566e47f767a90346715ed8
|
|
Type: refactor
Change-Id: Ia5a7ebd5c8492326f5bf2f1264f1c5ef38ccb4dc
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
For portabiliy reasons it is better to have all wrapped in clib code.
I.e. instead of using getcpu() we have clib_get_current_numa_node () and
clib_get_current_cpu_id().
Type: refactor
Change-Id: I29b52d7f29bc7f93873402c4070561f564b71c63
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Type: fix
Signed-off-by: Florin Coras <fcoras@cisco.com>
Change-Id: I23d1dda86c781ac077dbee7cb0e1ddeaa328c660
|
|
Type: fix
Apparently snprintf is not thread safe!?
Thread 6 "vpp_wk_3" received signal SIGSEGV, Segmentation fault.
[Switching to Thread 0x7ffeed1a3700 (LWP 819375)]
0x00007ffff58fc2b4 in __GI__IO_default_xsputn (n=<optimized out>,
data=<optimized out>, f=<optimized out>) at genops.c:394
394 genops.c: No such file or directory.
(gdb) bt
0 0x00007ffff58fc2b4 in __GI__IO_default_xsputn (n=<optimized out>,
data=<optimized out>, f=<optimized out>) at genops.c:394
1 __GI__IO_default_xsputn (f=0x7ffeec35e050, data=<optimized out>,
n=9) at genops.c:370
2 0x00007ffff58e127c in __vfprintf_internal (s=s@entry=0x7ffeec35e050,
format=format@entry=0x7ffff5f833ce "%s",
ap=ap@entry=0x7ffeec35e1d0, mode_flags=mode_flags@entry=2) at
../libio/libioP.h:948
3 0x00007ffff58f611a in __vsnprintf_internal (string=0x7ffed57ed018
"", maxlen=<optimized out>, format=0x7ffff5f833ce "%s",
args=args@entry=0x7ffeec35e1d0, mode_flags=2) at vsnprintf.c:114
4 0x00007ffff5997fd1 in ___snprintf_chk (s=<optimized out>,
maxlen=<optimized out>, flag=<optimized out>, slen=<optimized out>,
format=<optimized out>) at snprintf_chk.c:38
5 0x00007ffff5f7fd74 in clib_mem_vm_map_internal (base=<optimized
out>, log2_page_sz=<optimized out>, size=<optimized out>,
fd=<optimized out>, offset=<optimized out>, name=0x0) at
/scratch/fcoras/vpp/src/vppinfra/linux/mem.c:502
Signed-off-by: Florin Coras <fcoras@cisco.com>
Change-Id: I8ff7cfed422273dbdc55068772e11f5e550a9b60
|
|
Type: improvement
Change-Id: I1ab1b100000b4d7212c58e10312e16e7527bd333
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Type: fix
Change-Id: I90b537eba7f56dbb1928f0ab5f157218affb43f5
Signed-off-by: Benoît Ganne <bganne@cisco.com>
|
|
Type: improvement
Change-Id: I57a9f85f7df1fc48656b72592349f4c544302f77
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Type: fix
Signed-off-by: Dave Barach <dave@barachs.net>
Change-Id: I654747d618cc4fe99b7774827303769fe43392ed
|
|
This patch adds smal header in front of dlmalloc space, and it stores
some additional information about the heap.
Immediate benefit of this patch is that we know the underlying page size
si we can display heap page statistics / real memory usage.
Type: improvement
Change-Id: Ibd6989cc2f2f64630ab08734c9552e15029c5f3f
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Type: improvement
Change-Id: Ie04302c576869bc7bfaa9f13ed2ea8a403a393d4
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Type: fix
Change-Id: I941b92ebec66ccd9c9e9a12d485844313c9f117b
Signed-off-by: Benoît Ganne <bganne@cisco.com>
Signed-off-by: Mohsin Kazmi <sykazmi@cisco.com>
|
|
Type: improvement
Signed-off-by: Florin Coras <fcoras@cisco.com>
Change-Id: I33c6a5d1686cc32a6cde149083256d6cf0770fc5
|
|
Type: improvement
Change-Id: I81a7fb71c8ce0c0d22e326a4ddd01bc1c1aea5f7
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Type: improvement
Change-Id: I9973bce20a0a2a8a7e227cf96518de5b79374425
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Type: improvement
Change-Id: I04752c011e4ca58f56aa53f6ae27bae93a5c4590
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Type: improvement
Change-Id: I381fc3dec8580208d0e24637d791af69011aa83b
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Type: improvement
Change-Id: I298aadfdf17d98dfb1ada1ec4f87e0821e6aeb7f
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Type: improvement
Change-Id: Ie44dbf9396cfed19dba153810b7bd76ce5377cd4
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Type: fix
Ticket: VPP-1837
Signed-off-by: Dave Barach <dave@barachs.net>
Change-Id: I9ec87d2293d8f92c3e488a0f61083cf815ac496c
|
|
Template instances can allocate BIHASH_KVP_PER_PAGE data records
tangent to the bucket, to remove a dependent read / prefetch.
Template instances can ask for immediate memory allocation, to avoid
several branches in the lookup path.
Clean up l2 fib, gpb plugin codes: use clib_bihash_get_bucket(...)
Use hugepages for bihash allocation arenas
Type: improvement
Signed-off-by: Dave Barach <dave@barachs.net>
Signed-off-by: Damjan Marion <damarion@cisco.com>
Change-Id: I92fc11bc58e48d84e2d61f44580916dd1c56361c
|
|
Type: feature
Signed-off-by: Florin Coras <fcoras@cisco.com>
Change-Id: I999836a7893a89aac5243b111eac35fddd03e2a6
|
|
Type: fix
Change-Id: Ifd61c0683c85fe7340965c225ed23e46ec88e01a
Signed-off-by: Benoît Ganne <bganne@cisco.com>
|
|
Change-Id: Ibfdcec60567ec357205fa137257f2d8cba44b01c
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
With this simple tweak, vpp is OK for "make test" or non-dpdk driver
operation over a kernel which doesn't support hugepages at all.
Tested on a Raspberry-pi 3B+ running the Ubuntu 18.04 64-bit server
beta image. It's amazing that such a minimal system will not only
self-compile vpp, but also run the unit tests. It's slow, but what do
you expect for that amount of money?
Change-Id: I7f080f871c828cc4b44be13809495278fadbd717
Signed-off-by: Dave Barach <dave@barachs.net>
|
|
Change-Id: Iecceffe06a92660976ebb58cd3cbec4be8931db0
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Change-Id: I5ff713ad0b254c74c5622e3b9425cca365b5ee97
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Change-Id: Ic4c46bc733afae8bf0d8146623ed15633928de30
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Change-Id: Ibec32c6df32f4cd9889d378e244f170c93ad295b
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Change-Id: Iffba7ebe5af8fadc0251f3a10022739d45f394ce
Signed-off-by: Juraj Sloboda <jsloboda@cisco.com>
|
|
Starting with kernel 4.14 hugepage fd can be retrieved with
memfd_systemcall
Change-Id: I0f3bd6d0a7757ffe7b98e83763502013ac763ecb
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Change-Id: I2779626d745badb63386efcf729da7a094a4f297
Signed-off-by: Haiyang Tan <haiyangtan@tencent.com>
|
|
Change-Id: I8691a10493d159a97574550c111f07722960a7cd
Signed-off-by: Haiyang Tan <haiyangtan@tencent.com>
|
|
Move the binary api segment above 4gb
Change-Id: I40e8aa7a97722a32397f5a538b5ff8344c50d408
Signed-off-by: Dave Barach <dave@barachs.net>
|
|
Change-Id: I6c1c855cf5fc2ee06f1c7ddd6576ca16cd556fdd
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Change-Id: Icfb99a09726c01e96ff14967afbafa4116e02eff
Signed-off-by: Dave Barach <dbarach@cisco.com>
|
|
Use (u64) cast to ensure proper address calculations.
Change-Id: I6bad50010b140189f1b0af177e55da0045bd7a93
Signed-off-by: Lee Roberts <lee.roberts@hpe.com>
|