17765f8de0
Provide an AVX2-based function using compiler intrinsics for TCP/IP-style checksums. The load/unpack/add idea and implementation is largely based on code from BESS (the Berkeley Extensible Software Switch) licensed as 3-Clause BSD, with a number of modifications to further decrease pipeline stalls and to minimise cache pollution. This speeds up considerably data paths from sockets to tap interfaces, decreasing overhead for checksum computation, with 16-64KiB packet buffers, from approximately 11% to 7%. The rest is just syscalls at this point. While at it, provide convenience targets in the Makefile for avx2, avx2_debug, and debug targets -- these simply add target-specific CFLAGS to the build. Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
35 lines
953 B
Makefile
35 lines
953 B
Makefile
CFLAGS += -Wall -Wextra -pedantic
|
|
CFLAGS += -DRLIMIT_STACK_VAL=$(shell ulimit -s)
|
|
|
|
all: passt pasta passt4netns qrap
|
|
|
|
avx2: CFLAGS += -Ofast -mavx2 -ftree-vectorize -funroll-loops
|
|
avx2: clean all
|
|
|
|
avx2_debug: CFLAGS += -Ofast -mavx2 -ftree-vectorize -funroll-loops -DDEBUG -g
|
|
avx2_debug: clean all
|
|
|
|
static: CFLAGS += -static
|
|
static: clean all
|
|
|
|
debug: CFLAGS += -static -DDEBUG -g
|
|
debug: clean all
|
|
|
|
passt: passt.c passt.h arp.c arp.h checksum.c checksum.h dhcp.c dhcp.h \
|
|
dhcpv6.c dhcpv6.h pcap.c pcap.h ndp.c ndp.h siphash.c siphash.h \
|
|
tap.c tap.h icmp.c icmp.h tcp.c tcp.h udp.c udp.h util.c util.h
|
|
$(CC) $(CFLAGS) passt.c arp.c checksum.c dhcp.c dhcpv6.c pcap.c ndp.c \
|
|
siphash.c tap.c icmp.c tcp.c udp.c util.c -o passt
|
|
|
|
pasta: passt
|
|
ln -s passt pasta
|
|
|
|
passt4netns: passt
|
|
ln -s passt passt4netns
|
|
|
|
qrap: qrap.c passt.h
|
|
$(CC) $(CFLAGS) -DARCH=\"$(shell uname -m)\" qrap.c -o qrap
|
|
|
|
.PHONY: clean
|
|
clean:
|
|
-${RM} passt *.o qrap pasta passt4netns
|