Performance tests of the c_can driver led to the patch that gives the series its name. I also added two patches not really related to the topic of the series. Run test succesfully on a custom board having two CAN ports. I connected the CAN1 port to the CAN2 port with a cable. Then I opened two terminals. On one I issued a dump command and on the other the transmit command used in the tests described in https://marc.info/?l=linux-can&m=139746476821294&w=2. Terminal-1: $ ip link set can1 type can bitrate <bitrate> $ ip link set up can1 $ candump can1 >/tmp/can-test-<bitrate> Terminal-2 $ ip link set can0 type can bitrate <bitrate> $ ip link set up can0 $ time cangen can0 -g0 -p1 -I5A5 -L0 -x -n 1000000 Then I applied the following commands to the file generated by the dump command: $ wc -l </tmp/can-test-<bitrate> # ca $ egrep -v ' can1 5A5 \[0\]' /tmp/can-test-<bitrate> | wc -l # cb I repeated the tests for 1000000, 500000, 250000 and 125000 bitrates, before and after applying the series. Here are the results: Before applying the series: bitrate time ca cb 125000 6m 42.71s 1000000 0 250000 3m 23.28s 1000000 0 500000 1m 44.04s 1000000 0 1000000 1m 8.44s 1000000 0 After applying ring-FIFO series: bitrate time ca cb 125000 6m 40.48s 1000000 0 250000 3m 20.80s 1000000 0 500000 1m 42.56s 1000000 0 1000000 1m 7.89s 1000000 0 Changes in v3: - Remove the transmission spin_lock. - Use IF_RX in c_can_do_tx(). Changes in v2: - Move c_can_get_tx_free() from c_can_main.c to c_can.h. Dario Binacchi (4): can: c_can: remove struct c_can_priv::priv field can: c_can: exit c_can_do_tx() early if no frames have been sent can: c_can: support tx ring algorithm can: c_can: cache frames to operate as a true FIFO drivers/net/can/c_can/c_can.h | 25 ++++++- drivers/net/can/c_can/c_can_main.c | 95 +++++++++++++++++++------- drivers/net/can/c_can/c_can_platform.c | 1 - 3 files changed, 94 insertions(+), 27 deletions(-) -- 2.17.1
