| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-ID: <CAK7-dKYDqV97K+hbw651zbu3-UZ1WSOf2a07uqWWtiRSfJV_zQ@mail.gmail.com> Date: Mon, 17 Nov 2025 15:37:50 +0900 From: Hoyeon Lee <hoyeon.lee@...e.com> To: Yonghong Song <yonghong.song@...ux.dev> Cc: bpf@...r.kernel.org, Andrii Nakryiko <andrii@...nel.org>, Eduard Zingerman <eddyz87@...il.com>, Alexei Starovoitov <ast@...nel.org>, Daniel Borkmann <daniel@...earbox.net>, Martin KaFai Lau <martin.lau@...ux.dev>, Song Liu <song@...nel.org>, John Fastabend <john.fastabend@...il.com>, KP Singh <kpsingh@...nel.org>, Stanislav Fomichev <sdf@...ichev.me>, Hao Luo <haoluo@...gle.com>, Jiri Olsa <jolsa@...nel.org>, Shuah Khan <shuah@...nel.org>, Nathan Chancellor <nathan@...nel.org>, Nick Desaulniers <nick.desaulniers+lkml@...il.com>, Bill Wendling <morbo@...gle.com>, Justin Stitt <justinstitt@...gle.com>, linux-kselftest@...r.kernel.org, linux-kernel@...r.kernel.org, llvm@...ts.linux.dev Subject: Re: [bpf-next v1 5/5] selftests/bpf: propagate LLVM toolchain to runqslower build On Mon, Nov 17, 2025 at 3:04 PM Yonghong Song <yonghong.song@...ux.dev> wrote: > > > > On 11/15/25 2:55 PM, Hoyeon Lee wrote: > > The selftests/bpf invokes a nested make when building runqslower, but > > LLVM toolchain version (clang/llvm-strip) is not propagated. As a > > result, runqslower is built with system default clang, not respecting > > specified LLVM version. > > > > # LLVM=-21 make -C tools/testing/selftests/bpf > > ... > > make feature_display=0 -C /bpf/tools/bpf/runqslower \ > > OUTPUT=/bpf/tools/testing/selftests/bpf/tools/build/runqslower/ \ > > BPFOBJ_OUTPUT=/bpf/tools/testing/selftests/bpf/tools/build/libbpf/ \ > > BPFOBJ=/bpf/tools/testing/selftests/bpf/tools/build/libbpf/libbpf.a \ > > BPF_INCLUDE=/bpf/tools/testing/selftests/bpf/tools/include \ > > BPFTOOL_OUTPUT=/bpf/tools/testing/selftests/bpf/tools/build/bpftool/ \ > > VMLINUX_BTF=/sys/kernel/btf/vmlinux BPF_TARGET_ENDIAN=--target=bpfel \ > > EXTRA_CFLAGS='-g -O0 ' EXTRA_LDFLAGS=' ' && \ > > cp /bpf/tools/testing/selftests/bpf/tools/build/runqslower/runqslower \ > > /bpf/tools/testing/selftests/bpf/runqslower > > clang -g -O2 --target=bpfel -I/bpf/tools/testing/selftests/bpf/tools/build/runqslower/ \ > > -I/bpf/tools/testing/selftests/bpf/tools/include -I/bpf/tools/include/uapi \ > > -c runqslower.bpf.c -o /bpf/tools/testing/selftests/bpf/tools/build/runqslower/runqslower.bpf.o && \ > > llvm-strip -g /bpf/tools/testing/selftests/bpf/tools/build/runqslower//runqslower.bpf.o > > /bin/sh: 1: clang: not found > > I tried with LLVM=-20 make -C tools/testing/selftests/bpf in my system and > there is no build error. > > Also could you try with command line > make -C tools/testing/selftests/bpf LLVM=1 > for clang build kernel or selftests, LLVM=1 is recommended as it > encodes a bunch of clang command lines: > CC = $(LLVM_PREFIX)clang$(LLVM_SUFFIX) > LD = $(LLVM_PREFIX)ld.lld$(LLVM_SUFFIX) > AR = $(LLVM_PREFIX)llvm-ar$(LLVM_SUFFIX) > NM = $(LLVM_PREFIX)llvm-nm$(LLVM_SUFFIX) > OBJCOPY = $(LLVM_PREFIX)llvm-objcopy$(LLVM_SUFFIX) > OBJDUMP = $(LLVM_PREFIX)llvm-objdump$(LLVM_SUFFIX) > READELF = $(LLVM_PREFIX)llvm-readelf$(LLVM_SUFFIX) > STRIP = $(LLVM_PREFIX)llvm-strip$(LLVM_SUFFIX) > > Thanks for the review. Just to clarify, the issue is not the build failure itself. The error "clang: not found" only appeared because I intentionally did not set update-alternatives to avoid falling back to the system default compiler. The real issue is that the runqslower sub-make invokes "clang" instead of "clang-21" when LLVM=-21 is specified. This shows that the selected LLVM toolchain version is not being propagated into the nested build. LLVM=1 works well for general builds, but in this case the intention is to verify that a specific LLVM version is consistently applied across all sub-makes. That propagation does not currently happen, and the patch addresses exactly that. Thanks again for taking a look. > > > > > > Explicitly propagate CLANG and LLVM_STRIP to the runqslower sub-make so > > that the LLVM toolchain selection from lib.mk is preserved. > > > > Signed-off-by: Hoyeon Lee <hoyeon.lee@...e.com> > > --- > > tools/testing/selftests/bpf/Makefile | 1 + > > tools/testing/selftests/lib.mk | 1 + > > 2 files changed, 2 insertions(+) > > > > diff --git a/tools/testing/selftests/bpf/Makefile b/tools/testing/selftests/bpf/Makefile > > index 34ea23c63bd5..79ab69920dca 100644 > > --- a/tools/testing/selftests/bpf/Makefile > > +++ b/tools/testing/selftests/bpf/Makefile > > @@ -306,6 +306,7 @@ endif > > > > $(OUTPUT)/runqslower: $(BPFOBJ) | $(DEFAULT_BPFTOOL) $(RUNQSLOWER_OUTPUT) > > $(Q)$(MAKE) $(submake_extras) -C $(TOOLSDIR)/bpf/runqslower \ > > + CLANG=$(CLANG) LLVM_STRIP=$(LLVM_STRIP) \ > > OUTPUT=$(RUNQSLOWER_OUTPUT) VMLINUX_BTF=$(VMLINUX_BTF) \ > > BPFTOOL_OUTPUT=$(HOST_BUILD_DIR)/bpftool/ \ > > BPFOBJ_OUTPUT=$(BUILD_DIR)/libbpf/ \ > > diff --git a/tools/testing/selftests/lib.mk b/tools/testing/selftests/lib.mk > > index a448fae57831..f14255b2afbd 100644 > > --- a/tools/testing/selftests/lib.mk > > +++ b/tools/testing/selftests/lib.mk > > @@ -8,6 +8,7 @@ LLVM_SUFFIX := $(LLVM) > > endif > > > > CLANG := $(LLVM_PREFIX)clang$(LLVM_SUFFIX) > > +LLVM_STRIP := $(LLVM_PREFIX)llvm-strip$(LLVM_SUFFIX) > > > > CLANG_TARGET_FLAGS_arm := arm-linux-gnueabi > > CLANG_TARGET_FLAGS_arm64 := aarch64-linux-gnu >
Powered by blists - more mailing lists