Re: [RFC PATCH] x86_64: Add "-m elf_i386" when linking i386 object files.
Added Rui, an LLD code owner. This patch contains the approach we discussed earlier during LLD development when faced this issue first time (and as a result of the discussion, the same fix was suggested: https://bugzilla.kernel.org/show_bug.cgi?id=194091#c0), so I think it is fine. Best regards, George | Developer | Access Softek, Inc От: Tri Vo Отправлено: 12 декабря 2018 г. 22:48 Кому: Nick Desaulniers Копия: b...@alien8.de; x...@kernel.org; t...@linutronix.de; mi...@redhat.com; George Rimar; Dmitry Golovin; Bill Wendling; h...@zytor.com; linux-kernel@vger.kernel.org Тема: Re: [RFC PATCH] x86_64: Add "-m elf_i386" when linking i386 object files. CAUTION: This email originated from outside of the organization. Do not click links or open attachments unless you recognize the sender and know the content is safe. Adding appropriate people, lists. On Tue, Dec 11, 2018 at 1:07 PM Nick Desaulniers wrote: > > On Mon, Dec 10, 2018 at 2:50 PM Nick Desaulniers > wrote: > > > > On Mon, Dec 10, 2018 at 2:26 PM Tri Vo wrote: > > > > > > From: George Rimar > > > > > > Linux kernel uses OUTPUT_FORMAT in it's linker scripts. Most of the time > > > -m option is passed to the linker with correct architecture, but > > > sometimes (at least for x86_64) the -m option contradicts OUTPUT_FORMAT > > > directive. Specifically, arch/x86/boot and arch/x86/realmode/rm modules > > > have i386 object files, but are linked with -m elf_x86_64 linker flag > > > when building for x86_64. > > > > > > BFD and Gold linkers are OK with this, but lld fails: > > > ld.lld: error: arch/x86/realmode/rm/header.o is incompatible with > > > elf_x86_64 > > > > > > Suggested fix: just add correct -m after incorrect one (it overrides > > > it), so the linker invocation looks like this: ld -m elf_x86_64 -z > > > max-page-size=0x20 -m elf_i386 --emit-relocs -T realmode.lds > > > header.o trampoline_64.o stack.o reboot.o -o realmode.elf (it will also > > > work with GNU ld, because it supports OUTPUT_FORMAT and just ignores -m > > > options if this directive is in the linker script). > > > > > > Tested by building x86_64 kernel with GNU gcc/ld toolchain and booting > > > it in QEMU. > > > > > > Suggested-by: Dmitry Golovin > > > Signed-off-by: Tri Vo > > > Tested-by: Tri Vo > > > > This fixes the following linkage error I observe when linking an x86 > > kernel with LLD: > > > > ``` > > LD arch/x86/realmode/rm/realmode.elf > > ld.lld: error: arch/x86/realmode/rm/header.o is incompatible with elf_x86_64 > > ld.lld: error: arch/x86/realmode/rm/trampoline_64.o is incompatible > > with elf_x86_64 > > ld.lld: error: arch/x86/realmode/rm/stack.o is incompatible with elf_x86_64 > > ld.lld: error: arch/x86/realmode/rm/reboot.o is incompatible with elf_x86_64 > > ld.lld: error: arch/x86/realmode/rm/wakeup_asm.o is incompatible with > > elf_x86_64 > > ld.lld: error: arch/x86/realmode/rm/wakemain.o is incompatible with > > elf_x86_64 > > ld.lld: error: arch/x86/realmode/rm/video-mode.o is incompatible with > > elf_x86_64 > > ld.lld: error: arch/x86/realmode/rm/copy.o is incompatible with elf_x86_64 > > ld.lld: error: arch/x86/realmode/rm/bioscall.o is incompatible with > > elf_x86_64 > > ld.lld: error: arch/x86/realmode/rm/regs.o is incompatible with elf_x86_64 > > ld.lld: error: arch/x86/realmode/rm/video-vga.o is incompatible with > > elf_x86_64 > > ld.lld: error: arch/x86/realmode/rm/video-vesa.o is incompatible with > > elf_x86_64 > > ld.lld: error: arch/x86/realmode/rm/video-bios.o is incompatible with > > elf_x86_64 > > arch/x86/realmode/rm/Makefile:55: recipe for target > > 'arch/x86/realmode/rm/realmode.elf' failed > > ``` > > Tested-by: Nick Desaulniers > > > > Looks like we still have a few other (unrelated) issues to track down > > with LLD, but this gets us one step closer. > > > > Thanks for sending this, Tri! > > Just some additional thoughts on this: > > The kernel is adding -m elf_x86_64 to the linker command line > parameters, but then relying on binutils tie-breaking-behavior by > specifying a different architecture via linker script. So it's > ambiguous which arch the kernel is trying to link for, and the kernel > gets lucky/relies on the way ld.bfd happens to resolve this ambiguity > without warning. An alternative fix may be to just filter out/remove > the -m elf_x86_64 for this target's linker flags. > > > > > > ---
Re: [PATCH v3] x86_64: Add "-m elf_i386" when linking i386 object files.
Signed-off-by: George Rimar Best regards, George | Developer | Access Softek, Inc От: Tri Vo Отправлено: 11 января 2019 г. 23:10 Кому: x...@kernel.org; t...@linutronix.de; mi...@redhat.com; b...@alien8.de; h...@zytor.com Копия: George Rimar; d...@golovin.in; mo...@google.com; ndesaulni...@google.com; m...@suse.de; r...@google.com; linux-kernel@vger.kernel.org; Tri Vo Тема: [PATCH v3] x86_64: Add "-m elf_i386" when linking i386 object files. CAUTION: This email originated from outside of the organization. Do not click links or open attachments unless you recognize the sender and know the content is safe. If you suspect potential phishing or spam email, report it to reports...@accesssoftek.com From: George Rimar Linux kernel uses OUTPUT_FORMAT in it's linker scripts. Most of the time -m option is passed to the linker with correct architecture, but sometimes (at least for x86_64) the -m option contradicts OUTPUT_FORMAT directive. Specifically, arch/x86/boot and arch/x86/realmode/rm modules have i386 object files, but are linked with -m elf_x86_64 linker flag when building for x86_64. "man ld" doesn't explicitly state any tie-breakers between -m and OUTPUT_FORMAT. BFD and Gold linkers override -m value with OUTPUT_FORMAT. But LLVM lld has a different behavior. When supplied with contradicting -m and OUTPUT_FORMAT values it fails with the following error message: ld.lld: error: arch/x86/realmode/rm/header.o is incompatible with elf_x86_64 Suggested fix: just add correct -m after incorrect one (it overrides it), so the linker invocation looks like this: ld -m elf_x86_64 -z max-page-size=0x20 -m elf_i386 --emit-relocs -T realmode.lds header.o trampoline_64.o stack.o reboot.o -o realmode.elf This is not a functional change for GNU ld, because (although not explicitly documented) it already overrides -m EMULATION with OUTPUT_FORMAT. Tested by building x86_64 kernel with GNU gcc/ld toolchain and booting it in QEMU. Suggested-by: Dmitry Golovin Signed-off-by: George Rimar Signed-off-by: Tri Vo Tested-by: Tri Vo Tested-by: Nick Desaulniers --- v2: updated commit message to clarify that ld documentation is ambiguous w.r.t -m vs OUTPUT_FORMAT behavior. v3: fixed/added SOB and "Tested-by" fields. arch/x86/boot/Makefile| 2 +- arch/x86/realmode/rm/Makefile | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/arch/x86/boot/Makefile b/arch/x86/boot/Makefile index 9b5adae9cc40..e2839b5c246c 100644 --- a/arch/x86/boot/Makefile +++ b/arch/x86/boot/Makefile @@ -100,7 +100,7 @@ $(obj)/zoffset.h: $(obj)/compressed/vmlinux FORCE AFLAGS_header.o += -I$(objtree)/$(obj) $(obj)/header.o: $(obj)/zoffset.h -LDFLAGS_setup.elf := -T +LDFLAGS_setup.elf := -m elf_i386 -T $(obj)/setup.elf: $(src)/setup.ld $(SETUP_OBJS) FORCE $(call if_changed,ld) diff --git a/arch/x86/realmode/rm/Makefile b/arch/x86/realmode/rm/Makefile index 4463fa72db94..96cb20de08af 100644 --- a/arch/x86/realmode/rm/Makefile +++ b/arch/x86/realmode/rm/Makefile @@ -47,7 +47,7 @@ $(obj)/pasyms.h: $(REALMODE_OBJS) FORCE targets += realmode.lds $(obj)/realmode.lds: $(obj)/pasyms.h -LDFLAGS_realmode.elf := --emit-relocs -T +LDFLAGS_realmode.elf := -m elf_i386 --emit-relocs -T CPPFLAGS_realmode.lds += -P -C -I$(objtree)/$(obj) targets += realmode.elf -- 2.20.1.97.g81188d93c3-goog
[tip:x86/build] x86/build: Specify elf_i386 linker emulation explicitly for i386 objects
Commit-ID: 927185c124d62a9a4d35878d7f6d432a166b74e3 Gitweb: https://git.kernel.org/tip/927185c124d62a9a4d35878d7f6d432a166b74e3 Author: George Rimar AuthorDate: Fri, 11 Jan 2019 12:10:12 -0800 Committer: Borislav Petkov CommitDate: Sat, 12 Jan 2019 00:11:39 +0100 x86/build: Specify elf_i386 linker emulation explicitly for i386 objects The kernel uses the OUTPUT_FORMAT linker script command in it's linker scripts. Most of the time, the -m option is passed to the linker with correct architecture, but sometimes (at least for x86_64) the -m option contradicts the OUTPUT_FORMAT directive. Specifically, arch/x86/boot and arch/x86/realmode/rm produce i386 object files, but are linked with the -m elf_x86_64 linker flag when building for x86_64. The GNU linker manpage doesn't explicitly state any tie-breakers between -m and OUTPUT_FORMAT. But with BFD and Gold linkers, OUTPUT_FORMAT overrides the emulation value specified with the -m option. LLVM lld has a different behavior, however. When supplied with contradicting -m and OUTPUT_FORMAT values it fails with the following error message: ld.lld: error: arch/x86/realmode/rm/header.o is incompatible with elf_x86_64 Therefore, just add the correct -m after the incorrect one (it overrides it), so the linker invocation looks like this: ld -m elf_x86_64 -z max-page-size=0x20 -m elf_i386 --emit-relocs -T \ realmode.lds header.o trampoline_64.o stack.o reboot.o -o realmode.elf This is not a functional change for GNU ld, because (although not explicitly documented) OUTPUT_FORMAT overrides -m EMULATION. Tested by building x86_64 kernel with GNU gcc/ld toolchain and booting it in QEMU. [ bp: massage and clarify text. ] Suggested-by: Dmitry Golovin Signed-off-by: George Rimar Signed-off-by: Tri Vo Signed-off-by: Borislav Petkov Tested-by: Tri Vo Tested-by: Nick Desaulniers Cc: "H. Peter Anvin" Cc: Ingo Molnar Cc: Michael Matz Cc: Thomas Gleixner Cc: mo...@google.com Cc: ndesaulni...@google.com Cc: r...@google.com Cc: x86-ml Link: https://lkml.kernel.org/r/20190111201012.71210-1-tr...@android.com --- arch/x86/boot/Makefile| 2 +- arch/x86/realmode/rm/Makefile | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/arch/x86/boot/Makefile b/arch/x86/boot/Makefile index 9b5adae9cc40..e2839b5c246c 100644 --- a/arch/x86/boot/Makefile +++ b/arch/x86/boot/Makefile @@ -100,7 +100,7 @@ $(obj)/zoffset.h: $(obj)/compressed/vmlinux FORCE AFLAGS_header.o += -I$(objtree)/$(obj) $(obj)/header.o: $(obj)/zoffset.h -LDFLAGS_setup.elf := -T +LDFLAGS_setup.elf := -m elf_i386 -T $(obj)/setup.elf: $(src)/setup.ld $(SETUP_OBJS) FORCE $(call if_changed,ld) diff --git a/arch/x86/realmode/rm/Makefile b/arch/x86/realmode/rm/Makefile index 4463fa72db94..96cb20de08af 100644 --- a/arch/x86/realmode/rm/Makefile +++ b/arch/x86/realmode/rm/Makefile @@ -47,7 +47,7 @@ $(obj)/pasyms.h: $(REALMODE_OBJS) FORCE targets += realmode.lds $(obj)/realmode.lds: $(obj)/pasyms.h -LDFLAGS_realmode.elf := --emit-relocs -T +LDFLAGS_realmode.elf := -m elf_i386 --emit-relocs -T CPPFLAGS_realmode.lds += -P -C -I$(objtree)/$(obj) targets += realmode.elf
