- if no region is big enough to accommodate need_n_bits, we can scan
past the end of the bitmask
- rewrite the search loop for clarity
Type: fix
Change-Id: I82637bc91eb7e7ae922785bc626fa224638cb4e4
Signed-off-by: Benoît Ganne <bganne@cisco.com>
Adding two elements to the full svm queue passes exact
bounds check, therefore tail gets overwritten w/o any
waiting. Fix it with requiring at lease two free slots.
Type: fix
Signed-off-by: Vladislav Grishenko <themiron@yandex-team.ru>
Change-Id: I704ee606c47b523952cb965f848339ae1988cb60
Linux doesn't support the Linux idiom of using lseek and a write to set
the size of a file, instead use ftruncate to accomplish the same effect.
This change is taken from the Nanoteq VPP port commit:
04a1b19b37
Type: improvement
Change-Id: Ie0b83e751b8b8f20b6814e5c9f760035747dfad9
Signed-off-by: Tom Jones <thj@freebsd.org>
FreeBSD's siginfo_t doesn't have the ptr value used on linux to encode a
token. Fow now recover the code from siginfo_t and leave the opaque
value as 0.
Type: improvement
Change-Id: Ic6ef96f018b326b99d68534a4488cfc49fdf16e1
Signed-off-by: Tom Jones <thj@freebsd.org>
On FreeBSD we need to include stdint.h to get __WORDSIZE.
Type: improvement
Change-Id: I784aa0856fd8298f33016b7ccedecfa3672d5429
Signed-off-by: Tom Jones <thj@freebsd.org>
Move from synchronous flushing of io and ctrl events from transports to
applications to an async model via a new session_input input node that
runs in interrupt mode. Events are coalesced per application worker.
On the one hand, this helps by minimizing message queue locking churn.
And on the other, it opens the possibility for further optimizations of
event message generation, obviates need for rx rescheduling rpcs and is
a first step towards a fully async data/io rx path.
Type: improvement
Signed-off-by: Florin Coras <fcoras@cisco.com>
Change-Id: Id6bebcb65fc9feef8aa02ddf1af6d9ba6f6745ce
When checking for CLIB_SANITIZE_ADDR to enable specific behavior for
AddressSanitizer, we must have vppinfra/clib.h included as it is defined
there.
Type: fix
Change-Id: I9060c3c29c1289d28596c215a1d1709b2ea7c84e
Signed-off-by: Benoît Ganne <bganne@cisco.com>
add hugepage for vcl configure and svm
Type: feature
Signed-off-by: Junfeng Wang <drenfong.wang@intel.com>
Change-Id: I6a8905e3fec23d840e629114b1e5a403d0a258ef
svm_msg_q_size_to_alloc must return a valid base address, if it fails
pass up the error for handling
Type: fix
Change-Id: I408492f65f646862122acb9a187819b3bbf4f91c
Signed-off-by: Ofer Heifetz <oferh@marvell.com>
Make sure producer drops lock when it waits for empty ring slot.
Type: fix
Signed-off-by: Florin Coras <fcoras@cisco.com>
Change-Id: Id77d54ee8c01bed20c9eaf5ad372ed4b1e9fa712
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>
- support of in-place growth of vectors (if there is available space next to
existing alloc)
- drops the need for alloc_aligned_at_offset from memory allocator,
which allows easier swap to different memory allocator and reduces
malloc overhead
- rework of pool and vec macros to inline functions to improve debuggability
- fix alignment - in many cases macros were not using native alignment
of the particular datatype. Explicitly setting alignment with XXX_aligned()
versions of the macro is not needed anymore in > 99% of cases
- fix ASAN usage
- avoid use of vector of voids, this was root cause of several bugs
found in vec_* and pool_* function where sizeof() was used on voids
instead of real vector data type
- introduce minimal alignment which is currently 8 bytes, vectors will
be always aligned at least to that value (underlay allocator actually always
provide 16-byte aligned allocs)
Type: improvement
Change-Id: I20f4b081bb13bbf7bc0ace85cc4e301787f12fdf
Signed-off-by: Damjan Marion <damarion@cisco.com>
In addition to returning the number of bytes also update the number of
segments to reflect the number used.
Type: improvement
Signed-off-by: Florin Coras <fcoras@cisco.com>
Change-Id: Ia87dc2aa62cea38b18dfa83df94dc2abe29d5121
Add the ability to set a threshold on the fifo that will
be used to notify caller when de-queue from the fifo reached
a value that is less than or equal to the threshold.
Type: feature
Change-Id: I70ba1a05f783ce5247409e9beebe4e336b7c1eb5
Signed-off-by: Michal Kalderon <mkalderon@marvell.com>
Signed-off-by: Yuval Caduri <cyuval@marvell.com>
Simplifies allocation of fifos as fifo segment and segment manager
indices can be set at alloc time.
Type: improvement
Signed-off-by: Florin Coras <fcoras@cisco.com>
Change-Id: Ibd357b3ff0279d8deefcdcb17010b4068007ccb7
This patch ensures that fifo segment has at least the size that was
requested during allocation.
Type: improvement
Signed-off-by: Filip Tehlar <ftehlar@cisco.com>
Change-Id: Iea8a885ac290183e25e5c8f9163bba226c5efa15
Allocate per app pair segments with space for more than one fifo.
Type: feature
Signed-off-by: Florin Coras <fcoras@cisco.com>
Change-Id: Ib96fe12b899cb14ff20c0be607814011e2c3fc6a
This avoids chunk allocation/collection deadlocks if either of the sides
crashes.
Type: improvement
Signed-off-by: Florin Coras <fcoras@cisco.com>
Change-Id: I98619e6e035fa8688889ca34db2143c8898732df
Add option to use per app private segments for app to vpp message
queues, as opposed to exposing internal message queues segment.
When so configured, internal message queues are still polled by the
session queue node but external app message queues are handled by a new
input node (appsl-rx-mqs-input) that runs in interrupt state. Signaling
of the node, when mqs receive new messages, is done through eventfds
epolled by worker epoll input nodes.
Type: feature
Signed-off-by: Florin Coras <fcoras@cisco.com>
Change-Id: Iffe8ce5a9944a56a14e6d0f492a850cb9e392d16
1. When producer invokes svm_msg_q_wait() in svm_msg_q_lock_and_alloc_msg_w_ring(), queue mutex is held by itself.
2. Sometimes, svm msg queue is not full and ring is full, svm_msg_q_wait() do nothing with mutex held, consumer will blocking at svm_msg_q_send_signal().
Type: fix
Signed-off-by: nandfan <fanyufei521@outlook.com>
Signed-off-by: Florin Coras <fcoras@cisco.com>
Change-Id: Ib90b87ab76534cd42e9a4c3e11703e80d93ca678
Avoid changing the header on attach as it may be in use. Instead, as for
chunks, allocate header to be collected on detach.
Type: fix
Signed-off-by: Florin Coras <fcoras@cisco.com>
Change-Id: Ib316ecb5d61ae161032869b6f6a1863f1105a1d9
