This series address self-tests failures for udp gro-related tests.
The first patch addresses the main problem I observe locally - the XDP program required by such tests, xdp_dummy, is currently build in the ebpf self-tests directory, not available if/when the user targets net only. Arguably is more a refactor than a fix, but still targeting net to hopefully
The second patch fixes the integration of such tests with the build system.
Patch 3/3 fixes sporadic failures due to races.
Tested with:
make -C tools/testing/selftests/ TARGETS=net install ./tools/testing/selftests/kselftest_install/run_kselftest.sh \ -t "net:udpgro_bench.sh net:udpgro.sh net:udpgro_fwd.sh \ net:udpgro_frglist.sh net:veth.sh"
no failures.
Paolo Abeni (3): selftests: net: remove dependency on ebpf tests selftests: net: included needed helper in the install targets selftests: net: explicitly wait for listener ready
tools/testing/selftests/net/Makefile | 6 ++++-- tools/testing/selftests/net/udpgro.sh | 4 ++-- tools/testing/selftests/net/udpgro_bench.sh | 4 ++-- tools/testing/selftests/net/udpgro_frglist.sh | 6 +++--- tools/testing/selftests/net/udpgro_fwd.sh | 8 +++++--- tools/testing/selftests/net/veth.sh | 4 ++-- tools/testing/selftests/net/xdp_dummy.c | 13 +++++++++++++ 7 files changed, 31 insertions(+), 14 deletions(-) create mode 100644 tools/testing/selftests/net/xdp_dummy.c
Several net tests requires an XDP program build under the ebpf directory, and error out if such program is not available.
That makes running successful net test hard, let's duplicate into the net dir the [very small] program, re-using the existing rules to build it, and finally dropping the bogus dependency.
Signed-off-by: Paolo Abeni pabeni@redhat.com --- tools/testing/selftests/net/Makefile | 5 +++-- tools/testing/selftests/net/udpgro.sh | 4 ++-- tools/testing/selftests/net/udpgro_bench.sh | 4 ++-- tools/testing/selftests/net/udpgro_frglist.sh | 6 +++--- tools/testing/selftests/net/udpgro_fwd.sh | 2 +- tools/testing/selftests/net/veth.sh | 4 ++-- tools/testing/selftests/net/xdp_dummy.c | 13 +++++++++++++ 7 files changed, 26 insertions(+), 12 deletions(-) create mode 100644 tools/testing/selftests/net/xdp_dummy.c
diff --git a/tools/testing/selftests/net/Makefile b/tools/testing/selftests/net/Makefile index 50818075e566..304d8b852ef0 100644 --- a/tools/testing/selftests/net/Makefile +++ b/tools/testing/selftests/net/Makefile @@ -84,6 +84,7 @@ TEST_PROGS += sctp_vrf.sh TEST_GEN_FILES += sctp_hello TEST_GEN_FILES += csum TEST_GEN_FILES += nat6to4.o +TEST_GEN_FILES += xdp_dummy.o TEST_GEN_FILES += ip_local_port_range TEST_GEN_FILES += bind_wildcard TEST_PROGS += test_vxlan_mdb.sh @@ -104,7 +105,7 @@ $(OUTPUT)/tcp_inq: LDLIBS += -lpthread $(OUTPUT)/bind_bhash: LDLIBS += -lpthread $(OUTPUT)/io_uring_zerocopy_tx: CFLAGS += -I../../../include/
-# Rules to generate bpf obj nat6to4.o +# Rules to generate bpf objs CLANG ?= clang SCRATCH_DIR := $(OUTPUT)/tools BUILD_DIR := $(SCRATCH_DIR)/build @@ -139,7 +140,7 @@ endif
CLANG_SYS_INCLUDES = $(call get_sys_includes,$(CLANG),$(CLANG_TARGET_ARCH))
-$(OUTPUT)/nat6to4.o: nat6to4.c $(BPFOBJ) | $(MAKE_DIRS) +$(OUTPUT)/nat6to4.o $(OUTPUT)/xdp_dummy.o: $(OUTPUT)/%.o : %.c $(BPFOBJ) | $(MAKE_DIRS) $(CLANG) -O2 --target=bpf -c $< $(CCINCLUDE) $(CLANG_SYS_INCLUDES) -o $@
$(BPFOBJ): $(wildcard $(BPFDIR)/*.[ch] $(BPFDIR)/Makefile) \ diff --git a/tools/testing/selftests/net/udpgro.sh b/tools/testing/selftests/net/udpgro.sh index af5dc57c8ce9..8802604148dd 100755 --- a/tools/testing/selftests/net/udpgro.sh +++ b/tools/testing/selftests/net/udpgro.sh @@ -7,7 +7,7 @@ source net_helper.sh
readonly PEER_NS="ns-peer-$(mktemp -u XXXXXX)"
-BPF_FILE="../bpf/xdp_dummy.bpf.o" +BPF_FILE="xdp_dummy.o"
# set global exit status, but never reset nonzero one. check_err() @@ -197,7 +197,7 @@ run_all() { }
if [ ! -f ${BPF_FILE} ]; then - echo "Missing ${BPF_FILE}. Build bpf selftest first" + echo "Missing ${BPF_FILE}. Run 'make' first" exit -1 fi
diff --git a/tools/testing/selftests/net/udpgro_bench.sh b/tools/testing/selftests/net/udpgro_bench.sh index cb664679b434..7080eae5312b 100755 --- a/tools/testing/selftests/net/udpgro_bench.sh +++ b/tools/testing/selftests/net/udpgro_bench.sh @@ -7,7 +7,7 @@ source net_helper.sh
readonly PEER_NS="ns-peer-$(mktemp -u XXXXXX)"
-BPF_FILE="../bpf/xdp_dummy.bpf.o" +BPF_FILE="xdp_dummy.o"
cleanup() { local -r jobs="$(jobs -p)" @@ -84,7 +84,7 @@ run_all() { }
if [ ! -f ${BPF_FILE} ]; then - echo "Missing ${BPF_FILE}. Build bpf selftest first" + echo "Missing ${BPF_FILE}. Run 'make' first" exit -1 fi
diff --git a/tools/testing/selftests/net/udpgro_frglist.sh b/tools/testing/selftests/net/udpgro_frglist.sh index dd47fa96f6b3..e1ff645bd3d1 100755 --- a/tools/testing/selftests/net/udpgro_frglist.sh +++ b/tools/testing/selftests/net/udpgro_frglist.sh @@ -7,7 +7,7 @@ source net_helper.sh
readonly PEER_NS="ns-peer-$(mktemp -u XXXXXX)"
-BPF_FILE="../bpf/xdp_dummy.bpf.o" +BPF_FILE="xdp_dummy.o"
cleanup() { local -r jobs="$(jobs -p)" @@ -85,12 +85,12 @@ run_all() { }
if [ ! -f ${BPF_FILE} ]; then - echo "Missing ${BPF_FILE}. Build bpf selftest first" + echo "Missing ${BPF_FILE}. Run 'make' first" exit -1 fi
if [ ! -f nat6to4.o ]; then - echo "Missing nat6to4 helper. Build bpf nat6to4.o selftest first" + echo "Missing nat6to4 helper. Run 'make' first" exit -1 fi
diff --git a/tools/testing/selftests/net/udpgro_fwd.sh b/tools/testing/selftests/net/udpgro_fwd.sh index c079565add39..5fa8659ab13d 100755 --- a/tools/testing/selftests/net/udpgro_fwd.sh +++ b/tools/testing/selftests/net/udpgro_fwd.sh @@ -1,7 +1,7 @@ #!/bin/bash # SPDX-License-Identifier: GPL-2.0
-BPF_FILE="../bpf/xdp_dummy.bpf.o" +BPF_FILE="xdp_dummy.o" readonly BASE="ns-$(mktemp -u XXXXXX)" readonly SRC=2 readonly DST=1 diff --git a/tools/testing/selftests/net/veth.sh b/tools/testing/selftests/net/veth.sh index 2d073595c620..27574bbf2d63 100755 --- a/tools/testing/selftests/net/veth.sh +++ b/tools/testing/selftests/net/veth.sh @@ -1,7 +1,7 @@ #!/bin/sh # SPDX-License-Identifier: GPL-2.0
-BPF_FILE="../bpf/xdp_dummy.bpf.o" +BPF_FILE="xdp_dummy.o" readonly STATS="$(mktemp -p /tmp ns-XXXXXX)" readonly BASE=`basename $STATS` readonly SRC=2 @@ -218,7 +218,7 @@ while getopts "hs:" option; do done
if [ ! -f ${BPF_FILE} ]; then - echo "Missing ${BPF_FILE}. Build bpf selftest first" + echo "Missing ${BPF_FILE}. Run 'make' first" exit 1 fi
diff --git a/tools/testing/selftests/net/xdp_dummy.c b/tools/testing/selftests/net/xdp_dummy.c new file mode 100644 index 000000000000..d988b2e0cee8 --- /dev/null +++ b/tools/testing/selftests/net/xdp_dummy.c @@ -0,0 +1,13 @@ +// SPDX-License-Identifier: GPL-2.0 + +#define KBUILD_MODNAME "xdp_dummy" +#include <linux/bpf.h> +#include <bpf/bpf_helpers.h> + +SEC("xdp") +int xdp_dummy_prog(struct xdp_md *ctx) +{ + return XDP_PASS; +} + +char _license[] SEC("license") = "GPL";
Paolo Abeni wrote:
Several net tests requires an XDP program build under the ebpf directory, and error out if such program is not available.
That makes running successful net test hard, let's duplicate into the net dir the [very small] program, re-using the existing rules to build it, and finally dropping the bogus dependency.
Signed-off-by: Paolo Abeni pabeni@redhat.com
tools/testing/selftests/net/Makefile | 5 +++-- tools/testing/selftests/net/udpgro.sh | 4 ++-- tools/testing/selftests/net/udpgro_bench.sh | 4 ++-- tools/testing/selftests/net/udpgro_frglist.sh | 6 +++--- tools/testing/selftests/net/udpgro_fwd.sh | 2 +- tools/testing/selftests/net/veth.sh | 4 ++-- tools/testing/selftests/net/xdp_dummy.c | 13 +++++++++++++ 7 files changed, 26 insertions(+), 12 deletions(-) create mode 100644 tools/testing/selftests/net/xdp_dummy.c
diff --git a/tools/testing/selftests/net/Makefile b/tools/testing/selftests/net/Makefile index 50818075e566..304d8b852ef0 100644 --- a/tools/testing/selftests/net/Makefile +++ b/tools/testing/selftests/net/Makefile @@ -84,6 +84,7 @@ TEST_PROGS += sctp_vrf.sh TEST_GEN_FILES += sctp_hello TEST_GEN_FILES += csum TEST_GEN_FILES += nat6to4.o +TEST_GEN_FILES += xdp_dummy.o TEST_GEN_FILES += ip_local_port_range TEST_GEN_FILES += bind_wildcard TEST_PROGS += test_vxlan_mdb.sh @@ -104,7 +105,7 @@ $(OUTPUT)/tcp_inq: LDLIBS += -lpthread $(OUTPUT)/bind_bhash: LDLIBS += -lpthread $(OUTPUT)/io_uring_zerocopy_tx: CFLAGS += -I../../../include/ -# Rules to generate bpf obj nat6to4.o +# Rules to generate bpf objs CLANG ?= clang SCRATCH_DIR := $(OUTPUT)/tools BUILD_DIR := $(SCRATCH_DIR)/build @@ -139,7 +140,7 @@ endif CLANG_SYS_INCLUDES = $(call get_sys_includes,$(CLANG),$(CLANG_TARGET_ARCH)) -$(OUTPUT)/nat6to4.o: nat6to4.c $(BPFOBJ) | $(MAKE_DIRS) +$(OUTPUT)/nat6to4.o $(OUTPUT)/xdp_dummy.o: $(OUTPUT)/%.o : %.c $(BPFOBJ) | $(MAKE_DIRS) $(CLANG) -O2 --target=bpf -c $< $(CCINCLUDE) $(CLANG_SYS_INCLUDES) -o $@
is the "$(OUTPUT)/%.o :" intentional or a leftover from editing?
On Wed, 2024-01-24 at 20:10 -0500, Willem de Bruijn wrote:
Paolo Abeni wrote:
Several net tests requires an XDP program build under the ebpf directory, and error out if such program is not available.
That makes running successful net test hard, let's duplicate into the net dir the [very small] program, re-using the existing rules to build it, and finally dropping the bogus dependency.
Signed-off-by: Paolo Abeni pabeni@redhat.com
tools/testing/selftests/net/Makefile | 5 +++-- tools/testing/selftests/net/udpgro.sh | 4 ++-- tools/testing/selftests/net/udpgro_bench.sh | 4 ++-- tools/testing/selftests/net/udpgro_frglist.sh | 6 +++--- tools/testing/selftests/net/udpgro_fwd.sh | 2 +- tools/testing/selftests/net/veth.sh | 4 ++-- tools/testing/selftests/net/xdp_dummy.c | 13 +++++++++++++ 7 files changed, 26 insertions(+), 12 deletions(-) create mode 100644 tools/testing/selftests/net/xdp_dummy.c
diff --git a/tools/testing/selftests/net/Makefile b/tools/testing/selftests/net/Makefile index 50818075e566..304d8b852ef0 100644 --- a/tools/testing/selftests/net/Makefile +++ b/tools/testing/selftests/net/Makefile @@ -84,6 +84,7 @@ TEST_PROGS += sctp_vrf.sh TEST_GEN_FILES += sctp_hello TEST_GEN_FILES += csum TEST_GEN_FILES += nat6to4.o +TEST_GEN_FILES += xdp_dummy.o TEST_GEN_FILES += ip_local_port_range TEST_GEN_FILES += bind_wildcard TEST_PROGS += test_vxlan_mdb.sh @@ -104,7 +105,7 @@ $(OUTPUT)/tcp_inq: LDLIBS += -lpthread $(OUTPUT)/bind_bhash: LDLIBS += -lpthread $(OUTPUT)/io_uring_zerocopy_tx: CFLAGS += -I../../../include/ -# Rules to generate bpf obj nat6to4.o +# Rules to generate bpf objs CLANG ?= clang SCRATCH_DIR := $(OUTPUT)/tools BUILD_DIR := $(SCRATCH_DIR)/build @@ -139,7 +140,7 @@ endif CLANG_SYS_INCLUDES = $(call get_sys_includes,$(CLANG),$(CLANG_TARGET_ARCH)) -$(OUTPUT)/nat6to4.o: nat6to4.c $(BPFOBJ) | $(MAKE_DIRS) +$(OUTPUT)/nat6to4.o $(OUTPUT)/xdp_dummy.o: $(OUTPUT)/%.o : %.c $(BPFOBJ) | $(MAKE_DIRS) $(CLANG) -O2 --target=bpf -c $< $(CCINCLUDE) $(CLANG_SYS_INCLUDES) -o $@
is the "$(OUTPUT)/%.o :" intentional or a leftover from editing?
Is intentional and AFAICS required to let this rule being selected when the output directory is not an empty string (the target and the pre-req will be in different directories).
Cheers,
Paolo
Paolo Abeni wrote:
On Wed, 2024-01-24 at 20:10 -0500, Willem de Bruijn wrote:
Paolo Abeni wrote:
Several net tests requires an XDP program build under the ebpf directory, and error out if such program is not available.
That makes running successful net test hard, let's duplicate into the net dir the [very small] program, re-using the existing rules to build it, and finally dropping the bogus dependency.
Signed-off-by: Paolo Abeni pabeni@redhat.com
tools/testing/selftests/net/Makefile | 5 +++-- tools/testing/selftests/net/udpgro.sh | 4 ++-- tools/testing/selftests/net/udpgro_bench.sh | 4 ++-- tools/testing/selftests/net/udpgro_frglist.sh | 6 +++--- tools/testing/selftests/net/udpgro_fwd.sh | 2 +- tools/testing/selftests/net/veth.sh | 4 ++-- tools/testing/selftests/net/xdp_dummy.c | 13 +++++++++++++ 7 files changed, 26 insertions(+), 12 deletions(-) create mode 100644 tools/testing/selftests/net/xdp_dummy.c
diff --git a/tools/testing/selftests/net/Makefile b/tools/testing/selftests/net/Makefile index 50818075e566..304d8b852ef0 100644 --- a/tools/testing/selftests/net/Makefile +++ b/tools/testing/selftests/net/Makefile @@ -84,6 +84,7 @@ TEST_PROGS += sctp_vrf.sh TEST_GEN_FILES += sctp_hello TEST_GEN_FILES += csum TEST_GEN_FILES += nat6to4.o +TEST_GEN_FILES += xdp_dummy.o TEST_GEN_FILES += ip_local_port_range TEST_GEN_FILES += bind_wildcard TEST_PROGS += test_vxlan_mdb.sh @@ -104,7 +105,7 @@ $(OUTPUT)/tcp_inq: LDLIBS += -lpthread $(OUTPUT)/bind_bhash: LDLIBS += -lpthread $(OUTPUT)/io_uring_zerocopy_tx: CFLAGS += -I../../../include/ -# Rules to generate bpf obj nat6to4.o +# Rules to generate bpf objs CLANG ?= clang SCRATCH_DIR := $(OUTPUT)/tools BUILD_DIR := $(SCRATCH_DIR)/build @@ -139,7 +140,7 @@ endif CLANG_SYS_INCLUDES = $(call get_sys_includes,$(CLANG),$(CLANG_TARGET_ARCH)) -$(OUTPUT)/nat6to4.o: nat6to4.c $(BPFOBJ) | $(MAKE_DIRS) +$(OUTPUT)/nat6to4.o $(OUTPUT)/xdp_dummy.o: $(OUTPUT)/%.o : %.c $(BPFOBJ) | $(MAKE_DIRS) $(CLANG) -O2 --target=bpf -c $< $(CCINCLUDE) $(CLANG_SYS_INCLUDES) -o $@
is the "$(OUTPUT)/%.o :" intentional or a leftover from editing?
Is intentional and AFAICS required to let this rule being selected when the output directory is not an empty string (the target and the pre-req will be in different directories).
Thanks. I don't understand why. Sorry to harp on this small point, but you've verified that the build fails without? Is it perhaps due to that "$(MAKE_DIRS)" order-only-prerequisite? But nat6to4 on its own did not need this.
Substition references could add a second colon in a rule, but otherwise I cannot find a reference to this repeated colon syntax.
Don't waste time on my behalf if you're sure this is correct. I just can't add a reviewed tag if I don't understand it -- but that tag is hardly essential.
On Thu, 2024-01-25 at 09:27 -0500, Willem de Bruijn wrote:
Paolo Abeni wrote:
On Wed, 2024-01-24 at 20:10 -0500, Willem de Bruijn wrote:
Paolo Abeni wrote:
Several net tests requires an XDP program build under the ebpf directory, and error out if such program is not available.
That makes running successful net test hard, let's duplicate into the net dir the [very small] program, re-using the existing rules to build it, and finally dropping the bogus dependency.
Signed-off-by: Paolo Abeni pabeni@redhat.com
tools/testing/selftests/net/Makefile | 5 +++-- tools/testing/selftests/net/udpgro.sh | 4 ++-- tools/testing/selftests/net/udpgro_bench.sh | 4 ++-- tools/testing/selftests/net/udpgro_frglist.sh | 6 +++--- tools/testing/selftests/net/udpgro_fwd.sh | 2 +- tools/testing/selftests/net/veth.sh | 4 ++-- tools/testing/selftests/net/xdp_dummy.c | 13 +++++++++++++ 7 files changed, 26 insertions(+), 12 deletions(-) create mode 100644 tools/testing/selftests/net/xdp_dummy.c
diff --git a/tools/testing/selftests/net/Makefile b/tools/testing/selftests/net/Makefile index 50818075e566..304d8b852ef0 100644 --- a/tools/testing/selftests/net/Makefile +++ b/tools/testing/selftests/net/Makefile @@ -84,6 +84,7 @@ TEST_PROGS += sctp_vrf.sh TEST_GEN_FILES += sctp_hello TEST_GEN_FILES += csum TEST_GEN_FILES += nat6to4.o +TEST_GEN_FILES += xdp_dummy.o TEST_GEN_FILES += ip_local_port_range TEST_GEN_FILES += bind_wildcard TEST_PROGS += test_vxlan_mdb.sh @@ -104,7 +105,7 @@ $(OUTPUT)/tcp_inq: LDLIBS += -lpthread $(OUTPUT)/bind_bhash: LDLIBS += -lpthread $(OUTPUT)/io_uring_zerocopy_tx: CFLAGS += -I../../../include/ -# Rules to generate bpf obj nat6to4.o +# Rules to generate bpf objs CLANG ?= clang SCRATCH_DIR := $(OUTPUT)/tools BUILD_DIR := $(SCRATCH_DIR)/build @@ -139,7 +140,7 @@ endif CLANG_SYS_INCLUDES = $(call get_sys_includes,$(CLANG),$(CLANG_TARGET_ARCH)) -$(OUTPUT)/nat6to4.o: nat6to4.c $(BPFOBJ) | $(MAKE_DIRS) +$(OUTPUT)/nat6to4.o $(OUTPUT)/xdp_dummy.o: $(OUTPUT)/%.o : %.c $(BPFOBJ) | $(MAKE_DIRS) $(CLANG) -O2 --target=bpf -c $< $(CCINCLUDE) $(CLANG_SYS_INCLUDES) -o $@
is the "$(OUTPUT)/%.o :" intentional or a leftover from editing?
Is intentional and AFAICS required to let this rule being selected when the output directory is not an empty string (the target and the pre-req will be in different directories).
Thanks. I don't understand why. Sorry to harp on this small point, but you've verified that the build fails without? Is it perhaps due to that "$(MAKE_DIRS)" order-only-prerequisite? But nat6to4 on its own did not need this.
I tried quite a bit of permutation (all others failing) before selecting this one (shame on me, with a stackoverflow hint [!!!]).
But I finally found the relevant documentation reference:
https://www.gnu.org/software/make/manual/make.html#Static-Pattern
A simpler wildcard rule would not be enough, as the already existing wildcard used to build plain c files will take precedence.
nat6to4 did not need this fancy syntax, as it was a simple, single target single pre-req rule - that takes precedence on the mentioned wildcard.
Please let me know if the above clarifies a bit the scenario.
Cheers,
Paolo
Paolo Abeni wrote:
On Thu, 2024-01-25 at 09:27 -0500, Willem de Bruijn wrote:
Paolo Abeni wrote:
On Wed, 2024-01-24 at 20:10 -0500, Willem de Bruijn wrote:
Paolo Abeni wrote:
Several net tests requires an XDP program build under the ebpf directory, and error out if such program is not available.
That makes running successful net test hard, let's duplicate into the net dir the [very small] program, re-using the existing rules to build it, and finally dropping the bogus dependency.
Signed-off-by: Paolo Abeni pabeni@redhat.com
tools/testing/selftests/net/Makefile | 5 +++-- tools/testing/selftests/net/udpgro.sh | 4 ++-- tools/testing/selftests/net/udpgro_bench.sh | 4 ++-- tools/testing/selftests/net/udpgro_frglist.sh | 6 +++--- tools/testing/selftests/net/udpgro_fwd.sh | 2 +- tools/testing/selftests/net/veth.sh | 4 ++-- tools/testing/selftests/net/xdp_dummy.c | 13 +++++++++++++ 7 files changed, 26 insertions(+), 12 deletions(-) create mode 100644 tools/testing/selftests/net/xdp_dummy.c
diff --git a/tools/testing/selftests/net/Makefile b/tools/testing/selftests/net/Makefile index 50818075e566..304d8b852ef0 100644 --- a/tools/testing/selftests/net/Makefile +++ b/tools/testing/selftests/net/Makefile @@ -84,6 +84,7 @@ TEST_PROGS += sctp_vrf.sh TEST_GEN_FILES += sctp_hello TEST_GEN_FILES += csum TEST_GEN_FILES += nat6to4.o +TEST_GEN_FILES += xdp_dummy.o TEST_GEN_FILES += ip_local_port_range TEST_GEN_FILES += bind_wildcard TEST_PROGS += test_vxlan_mdb.sh @@ -104,7 +105,7 @@ $(OUTPUT)/tcp_inq: LDLIBS += -lpthread $(OUTPUT)/bind_bhash: LDLIBS += -lpthread $(OUTPUT)/io_uring_zerocopy_tx: CFLAGS += -I../../../include/ -# Rules to generate bpf obj nat6to4.o +# Rules to generate bpf objs CLANG ?= clang SCRATCH_DIR := $(OUTPUT)/tools BUILD_DIR := $(SCRATCH_DIR)/build @@ -139,7 +140,7 @@ endif CLANG_SYS_INCLUDES = $(call get_sys_includes,$(CLANG),$(CLANG_TARGET_ARCH)) -$(OUTPUT)/nat6to4.o: nat6to4.c $(BPFOBJ) | $(MAKE_DIRS) +$(OUTPUT)/nat6to4.o $(OUTPUT)/xdp_dummy.o: $(OUTPUT)/%.o : %.c $(BPFOBJ) | $(MAKE_DIRS) $(CLANG) -O2 --target=bpf -c $< $(CCINCLUDE) $(CLANG_SYS_INCLUDES) -o $@
is the "$(OUTPUT)/%.o :" intentional or a leftover from editing?
Is intentional and AFAICS required to let this rule being selected when the output directory is not an empty string (the target and the pre-req will be in different directories).
Thanks. I don't understand why. Sorry to harp on this small point, but you've verified that the build fails without? Is it perhaps due to that "$(MAKE_DIRS)" order-only-prerequisite? But nat6to4 on its own did not need this.
I tried quite a bit of permutation (all others failing) before selecting this one (shame on me, with a stackoverflow hint [!!!]).
But I finally found the relevant documentation reference:
https://www.gnu.org/software/make/manual/make.html#Static-Pattern
A simpler wildcard rule would not be enough, as the already existing wildcard used to build plain c files will take precedence.
nat6to4 did not need this fancy syntax, as it was a simple, single target single pre-req rule - that takes precedence on the mentioned wildcard.
Please let me know if the above clarifies a bit the scenario.
Reviewed-by: Willem de Bruijn willemb@google.com
Thanks for looking that up!
So the wildcard is needed for the %.c in the new rule. Makes sense.
The blamed commit below introduce a dependency in some net self-tests towards a newly introduce helper script.
Such script is currently not included into the TEST_PROGS_EXTENDED list and thus is not installed, causing failure for the relevant tests when executed from the install dir.
Fix the issue updating the install targets.
Fixes: 3bdd9fd29cb0 ("selftests/net: synchronize udpgro tests' tx and rx connection") Signed-off-by: Paolo Abeni pabeni@redhat.com --- tools/testing/selftests/net/Makefile | 1 + 1 file changed, 1 insertion(+)
diff --git a/tools/testing/selftests/net/Makefile b/tools/testing/selftests/net/Makefile index 304d8b852ef0..48c6f93b8149 100644 --- a/tools/testing/selftests/net/Makefile +++ b/tools/testing/selftests/net/Makefile @@ -55,6 +55,7 @@ TEST_PROGS += rps_default_mask.sh TEST_PROGS += big_tcp.sh TEST_PROGS_EXTENDED := in_netns.sh setup_loopback.sh setup_veth.sh TEST_PROGS_EXTENDED += toeplitz_client.sh toeplitz.sh lib.sh +TEST_PROGS_EXTENDED += net_helper.sh TEST_GEN_FILES = socket nettest TEST_GEN_FILES += psock_fanout psock_tpacket msg_zerocopy reuseport_addr_any TEST_GEN_FILES += tcp_mmap tcp_inq psock_snd txring_overwrite
Paolo Abeni wrote:
The blamed commit below introduce a dependency in some net self-tests towards a newly introduce helper script.
Such script is currently not included into the TEST_PROGS_EXTENDED list and thus is not installed, causing failure for the relevant tests when executed from the install dir.
Fix the issue updating the install targets.
Fixes: 3bdd9fd29cb0 ("selftests/net: synchronize udpgro tests' tx and rx connection") Signed-off-by: Paolo Abeni pabeni@redhat.com
Reviewed-by: Willem de Bruijn willemb@google.com
The UDP GRO forwarding test still hard-code an arbitrary pause to wait for the UDP listener becoming ready in background.
That causes sporadic failures depending on the host load.
Replace the sleep with the existing helper waiting for the desired port being exposed.
Fixes: a062260a9d5f ("selftests: net: add UDP GRO forwarding self-tests") Signed-off-by: Paolo Abeni pabeni@redhat.com --- tools/testing/selftests/net/udpgro_fwd.sh | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-)
diff --git a/tools/testing/selftests/net/udpgro_fwd.sh b/tools/testing/selftests/net/udpgro_fwd.sh index 5fa8659ab13d..d6b9c759043c 100755 --- a/tools/testing/selftests/net/udpgro_fwd.sh +++ b/tools/testing/selftests/net/udpgro_fwd.sh @@ -1,6 +1,8 @@ #!/bin/bash # SPDX-License-Identifier: GPL-2.0
+source net_helper.sh + BPF_FILE="xdp_dummy.o" readonly BASE="ns-$(mktemp -u XXXXXX)" readonly SRC=2 @@ -119,7 +121,7 @@ run_test() { ip netns exec $NS_DST $ipt -A INPUT -p udp --dport 8000 ip netns exec $NS_DST ./udpgso_bench_rx -C 1000 -R 10 -n 10 -l 1300 $rx_args & local spid=$! - sleep 0.1 + wait_local_port_listen "$NS_DST" 8000 udp ip netns exec $NS_SRC ./udpgso_bench_tx $family -M 1 -s 13000 -S 1300 -D $dst local retc=$? wait $spid @@ -168,7 +170,7 @@ run_bench() { ip netns exec $NS_DST bash -c "echo 2 > /sys/class/net/veth$DST/queues/rx-0/rps_cpus" ip netns exec $NS_DST taskset 0x2 ./udpgso_bench_rx -C 1000 -R 10 & local spid=$! - sleep 0.1 + wait_local_port_listen "$NS_DST" 8000 udp ip netns exec $NS_SRC taskset 0x1 ./udpgso_bench_tx $family -l 3 -S 1300 -D $dst local retc=$? wait $spid
Paolo Abeni wrote:
The UDP GRO forwarding test still hard-code an arbitrary pause to wait for the UDP listener becoming ready in background.
That causes sporadic failures depending on the host load.
Replace the sleep with the existing helper waiting for the desired port being exposed.
Fixes: a062260a9d5f ("selftests: net: add UDP GRO forwarding self-tests") Signed-off-by: Paolo Abeni pabeni@redhat.com
Reviewed-by: Willem de Bruijn willemb@google.com
On Wed, 24 Jan 2024 22:33:19 +0100 Paolo Abeni wrote:
This series address self-tests failures for udp gro-related tests.
The first patch addresses the main problem I observe locally - the XDP program required by such tests, xdp_dummy, is currently build in the ebpf self-tests directory, not available if/when the user targets net only. Arguably is more a refactor than a fix, but still targeting net to hopefully
The second patch fixes the integration of such tests with the build system.
Patch 3/3 fixes sporadic failures due to races.
Tested with:
make -C tools/testing/selftests/ TARGETS=net install ./tools/testing/selftests/kselftest_install/run_kselftest.sh \ -t "net:udpgro_bench.sh net:udpgro.sh net:udpgro_fwd.sh \ net:udpgro_frglist.sh net:veth.sh"
no failures.
This series got into net-next-2024-01-25--00-00, looking at the outputs it turns out that the iproute2 was crashing somewhere in libbpf :( I rebuilt latest iproute2 without libbpf support. It loads xdp_dummy.o just fine, so hopefully that's good enough for now, we'll see for sure in about 3 hours :)
Hello:
This series was applied to netdev/net.git (main) by Jakub Kicinski kuba@kernel.org:
On Wed, 24 Jan 2024 22:33:19 +0100 you wrote:
This series address self-tests failures for udp gro-related tests.
The first patch addresses the main problem I observe locally - the XDP program required by such tests, xdp_dummy, is currently build in the ebpf self-tests directory, not available if/when the user targets net only. Arguably is more a refactor than a fix, but still targeting net to hopefully
[...]
Here is the summary with links: - [net,1/3] selftests: net: remove dependency on ebpf tests https://git.kernel.org/netdev/net/c/98cb12eb52a7 - [net,2/3] selftests: net: included needed helper in the install targets https://git.kernel.org/netdev/net/c/f5173fe3e13b - [net,3/3] selftests: net: explicitly wait for listener ready https://git.kernel.org/netdev/net/c/4acffb66630a
You are awesome, thank you!
linux-kselftest-mirror@lists.linaro.org
-
Jakub Kicinski -
Paolo Abeni -
patchwork-bot+netdevbpf＠kernel.org -
Willem de Bruijn