Skip to contentSkip to conversation

selftests/bpf: add bpf-gcc support

PATCH In review Ilya Leoshkevich 3 participants 3 messages Sep 12, 2019 – Sep 16, 2019

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

PATCH v2 1/1 selftests/bpf: add bpf-gcc support

Now that binutils and gcc support for BPF is upstream, make use of it in BPF selftests using alu32-like approach. Share as much as possible of CFLAGS calculation with clang.

Fixes only obvious issues, leaving more complex ones for later:
- Use gcc-provided bpf-helpers.h instead of manually defining the helpers, change bpf_helpers.h include guard to avoid conflict.
- Include <linux/stddef.h> for __always_inline.
- Add $(OUTPUT)/../usr/include to include path in order to use local kernel headers instead of system kernel headers when building with O=.

In order to activate the bpf-gcc support, one needs to configure binutils and gcc with --target=bpf and make them available in $PATH. In particular, gcc must be installed as bpf-gcc, which is the default.

Right now with binutils 25a2915e8dba and gcc r275589 only a handful of tests work:

	# ./test_progs_bpf_gcc
	# Summary: 7/39 PASSED, 1 SKIPPED, 98 FAILED

The reason for those failures are as follows:

- Build errors:
- error: too many function arguments for eBPF for __always_inline functions read_str_var and read_map_var - must be inlining issue, and for process_l3_headers_v6, which relies on optimizing away function arguments.
- error: indirect call in function, which are not supported by eBPF where there are no obvious indirect calls in the source calls, e.g. in __encap_ipip_none.
- error: field 'lock' has incomplete type for fields of struct bpf_spin_lock type - bpf_spin_lock is re#defined by bpf-helpers.h, so its usage is sensitive to order of #includes.
- error: eBPF stack limit exceeded in sysctl_tcp_mem.
- Load errors:
- Missing object files due to above build errors.
- libbpf: failed to create map (name: 'test_ver.bss').
- libbpf: object file doesn't contain bpf program.
- `libbpf: Program '.text' contains unrecognized relo data pointing to

    section 0`.

- libbpf: BTF is required, but is missing or corrupted - no BTF support in gcc yet.

---
v1->v2: Use bpf-helpers.h, fix a few obvious compatibility problems.

 tools/testing/selftests/bpf/Makefile          | 65 ++++++++++++++-----
 tools/testing/selftests/bpf/bpf_helpers.h     | 24 ++++---
 .../testing/selftests/bpf/progs/test_tc_edt.c |  1 +
 3 files changed, 67 insertions(+), 23 deletions(-)
tools/testing/selftests/bpf/Makefile +50 -15
index 7f3196af1ae4..6889c19a628c 100644
@@ -17,6 +17,7 @@ LLC		?= llc
 LLVM_OBJCOPY	?= llvm-objcopy
 LLVM_READELF	?= llvm-readelf
 BTF_PAHOLE	?= pahole
+BPF_GCC		?= $(shell command -v bpf-gcc;)
 CFLAGS += -g -Wall -O2 -I$(APIDIR) -I$(LIBDIR) -I$(BPFDIR) -I$(GENDIR) $(GENFLAGS) -I../../../include \
 	  -Dbpf_prog_load=bpf_prog_test_load \
 	  -Dbpf_load_program=bpf_test_load_program
@@ -46,6 +47,10 @@ ifneq ($(SUBREG_CODEGEN),)
 TEST_GEN_FILES += $(patsubst %.o,alu32/%.o, $(BPF_OBJ_FILES))
 endif
 
+ifneq ($(BPF_GCC),)
+TEST_GEN_FILES += $(patsubst %.o,bpf_gcc/%.o, $(BPF_OBJ_FILES))
+endif
+
 # Order correspond to 'make run_tests' order
 TEST_PROGS := test_kmod.sh \
 	test_libbpf.sh \
@@ -137,16 +142,19 @@ endif
 #
 # Use '-idirafter': Don't interfere with include mechanics except where the
 # build would have failed anyways.
-CLANG_SYS_INCLUDES := $(shell $(CLANG) -v -E - </dev/null 2>&1 \
+define get_sys_includes
+$(shell $(1) -v -E - </dev/null 2>&1 \
 	| sed -n '/<...> search starts here:/,/End of search list./{ s| \(/.*\)|-idirafter \1|p }')
+endef
+CLANG_SYS_INCLUDES = $(call get_sys_includes,$(CLANG))
+BPF_CFLAGS = -I. -I./include/uapi -I../../../include/uapi \
+	     -I$(OUTPUT)/../usr/include -D__TARGET_ARCH_$(SRCARCH)
 
-CLANG_FLAGS = -I. -I./include/uapi -I../../../include/uapi \
-	      $(CLANG_SYS_INCLUDES) \
-	      -Wno-compare-distinct-pointer-types \
-	      -D__TARGET_ARCH_$(SRCARCH)
+CLANG_CFLAGS = $(CLANG_SYS_INCLUDES) \
+	       -Wno-compare-distinct-pointer-types
 
-$(OUTPUT)/test_l4lb_noinline.o: CLANG_FLAGS += -fno-inline
-$(OUTPUT)/test_xdp_noinline.o: CLANG_FLAGS += -fno-inline
+$(OUTPUT)/test_l4lb_noinline.o: BPF_CFLAGS += -fno-inline
+$(OUTPUT)/test_xdp_noinline.o: BPF_CFLAGS += -fno-inline
 
 $(OUTPUT)/test_queue_map.o: test_queue_stack_map.h
 $(OUTPUT)/test_stack_map.o: test_queue_stack_map.h
@@ -163,12 +171,12 @@ BTF_LLVM_PROBE := $(shell echo "int main() { return 0; }" | \
 			  /bin/rm -f ./llvm_btf_verify.o)
 
 ifneq ($(BTF_LLVM_PROBE),)
-	CLANG_FLAGS += -g
+	BPF_CFLAGS += -g
 else
 ifneq ($(BTF_LLC_PROBE),)
 ifneq ($(BTF_PAHOLE_PROBE),)
 ifneq ($(BTF_OBJCOPY_PROBE),)
-	CLANG_FLAGS += -g
+	BPF_CFLAGS += -g
 	LLC_FLAGS += -mattr=dwarfris
 	DWARF2BTF = y
 endif
@@ -202,8 +210,8 @@ $(ALU32_BUILD_DIR)/test_progs_32: prog_tests/*.c
 
 $(ALU32_BUILD_DIR)/%.o: progs/%.c $(ALU32_BUILD_DIR)/test_progs_32 \
 					| $(ALU32_BUILD_DIR)
-	($(CLANG) $(CLANG_FLAGS) -O2 -target bpf -emit-llvm -c $< -o - || \
-		echo "clang failed") | \
+	($(CLANG) $(BPF_CFLAGS) $(CLANG_CFLAGS) -O2 -target bpf -emit-llvm \
+		-c $< -o - || echo "clang failed") | \
 	$(LLC) -march=bpf -mattr=+alu32 -mcpu=$(CPU) $(LLC_FLAGS) \
 		-filetype=obj -o $@
 ifeq ($(DWARF2BTF),y)
@@ -211,10 +219,37 @@ ifeq ($(DWARF2BTF),y)
 endif
 endif
 
+ifneq ($(BPF_GCC),)
+GCC_SYS_INCLUDES = $(call get_sys_includes,gcc)
+IS_LITTLE_ENDIAN = $(shell $(CC) -dM -E - </dev/null | \
+			grep 'define __BYTE_ORDER__ __ORDER_LITTLE_ENDIAN__')
+ifeq ($(IS_LITTLE_ENDIAN),)
+MENDIAN=-mbig-endian
+else
+MENDIAN=-mlittle-endian
+endif
+BPF_GCC_CFLAGS = $(GCC_SYS_INCLUDES) $(MENDIAN)
+BPF_GCC_BUILD_DIR = $(OUTPUT)/bpf_gcc
+TEST_CUSTOM_PROGS += $(BPF_GCC_BUILD_DIR)/test_progs_bpf_gcc
+$(BPF_GCC_BUILD_DIR):
+	mkdir -p $@
+
+$(BPF_GCC_BUILD_DIR)/urandom_read: $(OUTPUT)/urandom_read | $(BPF_GCC_BUILD_DIR)
+	cp $< $@
+
+$(BPF_GCC_BUILD_DIR)/test_progs_bpf_gcc: $(OUTPUT)/test_progs \
+					 | $(BPF_GCC_BUILD_DIR)
+	cp $< $@
+
+$(BPF_GCC_BUILD_DIR)/%.o: progs/%.c $(BPF_GCC_BUILD_DIR)/test_progs_bpf_gcc \
+			  | $(BPF_GCC_BUILD_DIR)
+	$(BPF_GCC) $(BPF_CFLAGS) $(BPF_GCC_CFLAGS) -O2 -c $< -o $@
+endif
+
 # Have one program compiled without "-target bpf" to test whether libbpf loads
 # it successfully
 $(OUTPUT)/test_xdp.o: progs/test_xdp.c
-	($(CLANG) $(CLANG_FLAGS) -O2 -emit-llvm -c $< -o - || \
+	($(CLANG) $(BPF_CFLAGS) $(CLANG_CFLAGS) -O2 -emit-llvm -c $< -o - || \
 		echo "clang failed") | \
 	$(LLC) -march=bpf -mcpu=$(CPU) $(LLC_FLAGS) -filetype=obj -o $@
 ifeq ($(DWARF2BTF),y)
@@ -222,8 +257,8 @@ ifeq ($(DWARF2BTF),y)
 endif
 
 $(OUTPUT)/%.o: progs/%.c
-	($(CLANG) $(CLANG_FLAGS) -O2 -target bpf -emit-llvm -c $< -o - || \
-		echo "clang failed") | \
+	($(CLANG) $(BPF_CFLAGS) $(CLANG_CFLAGS) -O2 -target bpf -emit-llvm \
+		-c $< -o - || echo "clang failed") | \
 	$(LLC) -march=bpf -mcpu=$(CPU) $(LLC_FLAGS) -filetype=obj -o $@
 ifeq ($(DWARF2BTF),y)
 	$(BTF_PAHOLE) -J $@
@@ -282,6 +317,6 @@ $(VERIFIER_TESTS_H): $(VERIFIER_TEST_FILES) | $(VERIFIER_TESTS_DIR)
 		  echo '#endif' \
 		 ) > $(VERIFIER_TESTS_H))
 
-EXTRA_CLEAN := $(TEST_CUSTOM_PROGS) $(ALU32_BUILD_DIR) \
+EXTRA_CLEAN := $(TEST_CUSTOM_PROGS) $(ALU32_BUILD_DIR) $(BPF_GCC_BUILD_DIR) \
 	$(VERIFIER_TESTS_H) $(PROG_TESTS_H) $(MAP_TESTS_H) \
 	feature
tools/testing/selftests/bpf/bpf_helpers.h +16 -8
index 6c4930bc6e2e..54a50699bbfd 100644
@@ -1,12 +1,6 @@
 /* SPDX-License-Identifier: (LGPL-2.1 OR BSD-2-Clause) */
-#ifndef __BPF_HELPERS_H
-#define __BPF_HELPERS_H
-
-/* helper macro to place programs, maps, license in
- * different sections in elf_bpf file. Section names
- * are interpreted by elf_bpf loader
- */
-#define SEC(NAME) __attribute__((section(NAME), used))
+#ifndef __BPF_HELPERS__
+#define __BPF_HELPERS__
 
 #define __uint(name, val) int (*name)[val]
 #define __type(name, val) val *name
@@ -19,6 +13,14 @@
 			 ##__VA_ARGS__);		\
 })
 
+#ifdef __clang__
+
+/* helper macro to place programs, maps, license in
+ * different sections in elf_bpf file. Section names
+ * are interpreted by elf_bpf loader
+ */
+#define SEC(NAME) __attribute__((section(NAME), used))
+
 /* helper functions called from eBPF programs written in C */
 static void *(*bpf_map_lookup_elem)(void *map, const void *key) =
 	(void *) BPF_FUNC_map_lookup_elem;
@@ -256,6 +258,12 @@ struct bpf_map_def {
 	unsigned int numa_node;
 };
 
+#else
+
+#include <bpf-helpers.h>
+
+#endif
+
 #define BPF_ANNOTATE_KV_PAIR(name, type_key, type_val)		\
 	struct ____btf_map_##name {				\
 		type_key key;					\
tools/testing/selftests/bpf/progs/test_tc_edt.c +1 -0
index 3af64c470d64..0961415ba477 100644
@@ -2,6 +2,7 @@
 #include <stdint.h>
 #include <linux/bpf.h>
 #include <linux/if_ether.h>
+#include <linux/stddef.h>
 #include <linux/in.h>
 #include <linux/ip.h>
 #include <linux/pkt_cls.h>
Signature · 1 line
2.21.0
#1 lore raw

Right now with binutils 25a2915e8dba and gcc r275589 only a handful of tests work:

# ./test_progs_bpf_gcc
# Summary: 7/39 PASSED, 1 SKIPPED, 98 FAILED

The reason for those failures are as follows:

- Build errors:
- error: too many function arguments for eBPF for __always_inline functions read_str_var and read_map_var - must be inlining issue, and for process_l3_headers_v6, which relies on optimizing away function arguments.
- error: indirect call in function, which are not supported by eBPF where there are no obvious indirect calls in the source calls, e.g. in __encap_ipip_none.
- error: field 'lock' has incomplete type for fields of struct bpf_spin_lock type - bpf_spin_lock is re#defined by bpf-helpers.h, so its usage is sensitive to order of #includes.
- error: eBPF stack limit exceeded in sysctl_tcp_mem.
- Load errors:
- Missing object files due to above build errors.
- libbpf: failed to create map (name: 'test_ver.bss').
- libbpf: object file doesn't contain bpf program.
- libbpf: Program '.text' contains unrecognized relo data pointing to section 0.
- libbpf: BTF is required, but is missing or corrupted - no BTF support in gcc yet.

Thanks for the nice report.
That will keep me busy for a little while :)

maintainer
#2 lore raw
In reply to Ilya Leoshkevich
Quoted 40 lines · Ilya Leoshkevich

Now that binutils and gcc support for BPF is upstream, make use of it in BPF selftests using alu32-like approach. Share as much as possible of CFLAGS calculation with clang.

Fixes only obvious issues, leaving more complex ones for later:
- Use gcc-provided bpf-helpers.h instead of manually defining the helpers, change bpf_helpers.h include guard to avoid conflict.
- Include <linux/stddef.h> for __always_inline.
- Add $(OUTPUT)/../usr/include to include path in order to use local kernel headers instead of system kernel headers when building with O=.

In order to activate the bpf-gcc support, one needs to configure binutils and gcc with --target=bpf and make them available in $PATH. In particular, gcc must be installed as bpf-gcc, which is the default.

Right now with binutils 25a2915e8dba and gcc r275589 only a handful of tests work:

	# ./test_progs_bpf_gcc
	# Summary: 7/39 PASSED, 1 SKIPPED, 98 FAILED

The reason for those failures are as follows:

- Build errors:
- error: too many function arguments for eBPF for __always_inline functions read_str_var and read_map_var - must be inlining issue, and for process_l3_headers_v6, which relies on optimizing away function arguments.
- error: indirect call in function, which are not supported by eBPF where there are no obvious indirect calls in the source calls, e.g. in __encap_ipip_none.
- error: field 'lock' has incomplete type for fields of struct bpf_spin_lock type - bpf_spin_lock is re#defined by bpf-helpers.h, so its usage is sensitive to order of #includes.
- error: eBPF stack limit exceeded in sysctl_tcp_mem.
- Load errors:
- Missing object files due to above build errors.
- libbpf: failed to create map (name: 'test_ver.bss').
- libbpf: object file doesn't contain bpf program.
- `libbpf: Program '.text' contains unrecognized relo data pointing to

Quoted 7 lines
     section 0`.

- libbpf: BTF is required, but is missing or corrupted - no BTF support in gcc yet.

---
v1->v2: Use bpf-helpers.h, fix a few obvious compatibility problems.

I think it's a good starting point unblocking Jose and others to run the BPF test suite against bpf-gcc. Longer term we might want to do further refactoring to really cleanly split llvm vs gcc dependency, and compare test results back to back to make sure it's consistent behavior. Anyway, applied, thanks!