| Message ID | 20240504022301.35250-1-jhubbard@nvidia.com |
|---|---|
| State | New |
| Headers | show |
| Series | selftests/exec: build with -fPIE instead of -pie, to make clang happy | expand |
On Fri, May 03, 2024 at 07:23:01PM -0700, John Hubbard wrote: > clang doesn't deal well with "-pie -static": it warns that -pie is an > unused option here. Changing to "-fPIE -static" solves this problem for > clang, while keeping the gcc results identical. > > The problem is visible when building via: > > make LLVM=1 -C tools/testing/selftests > > Again: gcc 13 produces identical binaries for all of these programs, > both before and after this commit (using "-pie"), and after (using > "-fPIE"). > > Also, the runtime results are the same for both clang and gcc builds. Thanks for this! It got solved differently here: https://lkml.kernel.org/r/20240416152831.3199999-1-usama.anjum@collabora.com Does that work for you as well? -Kees
On 5/6/24 10:09 AM, Kees Cook wrote: > On Fri, May 03, 2024 at 07:23:01PM -0700, John Hubbard wrote: >> clang doesn't deal well with "-pie -static": it warns that -pie is an >> unused option here. Changing to "-fPIE -static" solves this problem for >> clang, while keeping the gcc results identical. >> >> The problem is visible when building via: >> >> make LLVM=1 -C tools/testing/selftests >> >> Again: gcc 13 produces identical binaries for all of these programs, >> both before and after this commit (using "-pie"), and after (using >> "-fPIE"). >> >> Also, the runtime results are the same for both clang and gcc builds. > > Thanks for this! It got solved differently here: > https://lkml.kernel.org/r/20240416152831.3199999-1-usama.anjum@collabora.com > > Does that work for you as well? > Yes it does. Please drop my patch and let's use yours instead, thanks! (As an aside, outside of the compiler sources themselves, it has proven quite difficult to find any in-depth documentation of linking options and behavior--now, and in the past.) thanks,
diff --git a/tools/testing/selftests/exec/Makefile b/tools/testing/selftests/exec/Makefile index fb4472ddffd8..b7b54d442378 100644 --- a/tools/testing/selftests/exec/Makefile +++ b/tools/testing/selftests/exec/Makefile @@ -29,8 +29,8 @@ $(OUTPUT)/execveat.denatured: $(OUTPUT)/execveat cp $< $@ chmod -x $@ $(OUTPUT)/load_address_4096: load_address.c - $(CC) $(CFLAGS) $(LDFLAGS) -Wl,-z,max-page-size=0x1000 -pie -static $< -o $@ + $(CC) $(CFLAGS) $(LDFLAGS) -Wl,-z,max-page-size=0x1000 -fPIE -static $< -o $@ $(OUTPUT)/load_address_2097152: load_address.c - $(CC) $(CFLAGS) $(LDFLAGS) -Wl,-z,max-page-size=0x200000 -pie -static $< -o $@ + $(CC) $(CFLAGS) $(LDFLAGS) -Wl,-z,max-page-size=0x200000 -fPIE -static $< -o $@ $(OUTPUT)/load_address_16777216: load_address.c - $(CC) $(CFLAGS) $(LDFLAGS) -Wl,-z,max-page-size=0x1000000 -pie -static $< -o $@ + $(CC) $(CFLAGS) $(LDFLAGS) -Wl,-z,max-page-size=0x1000000 -fPIE -static $< -o $@
clang doesn't deal well with "-pie -static": it warns that -pie is an unused option here. Changing to "-fPIE -static" solves this problem for clang, while keeping the gcc results identical. The problem is visible when building via: make LLVM=1 -C tools/testing/selftests Again: gcc 13 produces identical binaries for all of these programs, both before and after this commit (using "-pie"), and after (using "-fPIE"). Also, the runtime results are the same for both clang and gcc builds. Signed-off-by: John Hubbard <jhubbard@nvidia.com> --- tools/testing/selftests/exec/Makefile | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) base-commit: ddb4c3f25b7b95df3d6932db0b379d768a6ebdf7 prerequisite-patch-id: b901ece2a5b78503e2fb5480f20e304d36a0ea27