Re: [PATCH] LoongArch: vDSO: Emit GNU_EH_FRAME correctly

From: Xi Ruoyao

Date: Wed Feb 25 2026 - 21:36:37 EST

On Thu, 2026-02-26 at 09:30 +0800, Jinyang He wrote:
> On 2026-02-25 18:45, Xi Ruoyao wrote:
>
> > With -fno-asynchronous-unwind-tables and --no-eh-frame-hdr (the
> > default
> > of the linker), the GNU_EH_FRAME segment (specified by vdso.lds.S)
> > is
> > empty. This is not valid, as the current DWARF specification
> > mandates
> > the first byte of the EH frame to be the version number 1. It
> > causes
> > some unwinders to complain, for example the ClickHouse query
> > profiler
> > spams the log with messages:
> >
> >      clickhouse-server[365854]: libunwind: unsupported .eh_frame_hdr
> >      version: 127 at 7ffffffb0000
> >
> > Here "127" is just the byte located at the p_vaddr (0, i.e. the
> > beginning of the vDSO) of the empty GNU_EH_FRAME segment.
> > Cross-checking with /proc/365854/maps has also proven 7ffffffb0000
> > is
> > the start of vDSO in the process VM image.
> >
> > In LoongArch the -fno-asynchronous-unwind-tables option seems just a
> > MIPS legacy, and MIPS only uses this option to satisfy the MIPS-
> > specific
> > "genvdso" program, per the commit cfd75c2db17e ("MIPS: VDSO:
> > Explicitly
> > use -fno-asynchronous-unwind-tables"). IIUC it indicates some
> > inherent
> > limitation of the MIPS ELF ABI and has nothing to do with
> > LoongArch. So
> > we can simply flip it over to -fasynchronous-unwind-tables and pass
> > --eh-frame-hdr for linking the vDSO, allowing the profilers to
> > unwind the
> > stack for statistics even if the sample point is taken when the PC
> > is in
> > the vDSO.
> >
> > However simply adjusting the options above would exploit an issue:
> > when
> > the libgcc unwinder saw the invalid GNU_EH_FRAME segment, it
> > silently
> > falled back to a machine-specific routine to match the code pattern
> > of
> > rt_sigreturn and extract the registers saved in the sigframe if the
> > code
> > pattern is matched. As unwinding from signal handlers is vital for
> > libgcc to support pthread cancellation etc., the fall-back routine
> > had
> > been silently keeping the LoongArch Linux systems functioning since
> > Linux 5.19. But when we start to emit GNU_EH_FRAME with the correct
> > format, fall-back routine will no longer be used and libgcc will
> > fail
> > to unwind the sigframe, and unwinding from signal handlers will no
> > longer work, causing dozens of glibc test failures. To make it
> > possible
> > to unwind from signal handlers again, it's necessary to code the
> > unwind
> > info in __vdso_rt_sigreturn via .cfi_* directives.
> >
> > The offsets in the .cfi_* directives depend on the layout of struct
> > sigframe, notably the offset of sigcontect in the sigframe. To use
> > the
> > offset in the assembly file, factor out struct sigframe into a
> > header to
> > allow asm-offsets.c to output the offset for assembly.
> >
> > To work around a long-term issue in the libgcc unwinder (the pc is
> > unconditionally substracted by 1: doing so is technically incorrect
> > for
> > a signal frame), a nop instruction is included with the two real
> > instructions in __vdso_rt_sigreturn in the same FDE PC range. The
> > same
> > hack has been used on x86 for a long time.
> >
> > Fixes: c6b99bed6b8f ("LoongArch: Add VDSO and VSYSCALL support")
> > Cc: stable@xxxxxxxxxxxxxxx
> > Signed-off-by: Xi Ruoyao <xry111@xxxxxxxxxxx>
> > ---
> > arch/loongarch/include/asm/sigframe.h | 22 ++++++++++++++
> > arch/loongarch/kernel/asm-offsets.c   | 2 ++
> > arch/loongarch/kernel/signal.c        | 6 +---
> > arch/loongarch/vdso/Makefile          | 4 +--
> > arch/loongarch/vdso/sigreturn.S       | 44
> > ++++++++++++++++++++++++---
> > 5 files changed, 67 insertions(+), 11 deletions(-)
> > create mode 100644 arch/loongarch/include/asm/sigframe.h
> >
> > diff --git a/arch/loongarch/include/asm/sigframe.h
> > b/arch/loongarch/include/asm/sigframe.h
> > new file mode 100644
> > index 000000000000..6889bcf5dc88
> > --- /dev/null
> > +++ b/arch/loongarch/include/asm/sigframe.h
> > @@ -0,0 +1,22 @@
> > +/* SPDX-License-Identifier: GPL-2.0+ */
> > +/*
> > + * Separated from arch/loongarch/kernel/signal.c:
> > + *
> > + * Author: Hanlu Li <lihanlu@xxxxxxxxxxx>
> > + *         Huacai Chen <chenhuacai@xxxxxxxxxxx>
> > + * Copyright (C) 2020-2022 Loongson Technology Corporation Limited
> > + *
> > + * Derived from MIPS:
> > + * Copyright (C) 1991, 1992 Linus Torvalds
> > + * Copyright (C) 1994 - 2000 Ralf Baechle
> > + * Copyright (C) 1999, 2000 Silicon Graphics, Inc.
> > + * Copyright (C) 2014, Imagination Technologies Ltd.
> > + */
> > +
> > +#include <uapi/asm/ucontext.h>
> > +#include <asm/siginfo.h>
> > +
> > +struct rt_sigframe {
> > + struct siginfo rs_info;
> > + struct ucontext rs_uctx;
> > +};
> > diff --git a/arch/loongarch/kernel/asm-offsets.c
> > b/arch/loongarch/kernel/asm-offsets.c
> > index 3017c7157600..2cc953f113ac 100644
> > --- a/arch/loongarch/kernel/asm-offsets.c
> > +++ b/arch/loongarch/kernel/asm-offsets.c
> > @@ -16,6 +16,7 @@
> > #include <asm/ptrace.h>
> > #include <asm/processor.h>
> > #include <asm/ftrace.h>
> > +#include <asm/sigframe.h>
> > #include <vdso/datapage.h>
> >
> > static void __used output_ptreg_defines(void)
> > @@ -220,6 +221,7 @@ static void __used output_sc_defines(void)
> >    COMMENT("Linux sigcontext offsets.");
> >    OFFSET(SC_REGS, sigcontext, sc_regs);
> >    OFFSET(SC_PC, sigcontext, sc_pc);
> > + OFFSET(RT_SIGFRAME_SC, rt_sigframe, rs_uctx.uc_mcontext);
> >    BLANK();
> > }
> >
> > diff --git a/arch/loongarch/kernel/signal.c
> > b/arch/loongarch/kernel/signal.c
> > index c9f7ca778364..e297d54ea638 100644
> > --- a/arch/loongarch/kernel/signal.c
> > +++ b/arch/loongarch/kernel/signal.c
> > @@ -37,6 +37,7 @@
> > #include <asm/lbt.h>
> > #include <asm/ucontext.h>
> > #include <asm/vdso.h>
> > +#include <asm/sigframe.h>
> >
> > #ifdef DEBUG_SIG
> > # define DEBUGP(fmt, args...) printk("%s: " fmt, __func__,
> > ##args)
> > @@ -51,11 +52,6 @@
> > #define lock_lbt_owner() ({ preempt_disable();
> > pagefault_disable(); })
> > #define unlock_lbt_owner() ({ pagefault_enable();
> > preempt_enable(); })
> >
> > -struct rt_sigframe {
> > - struct siginfo rs_info;
> > - struct ucontext rs_uctx;
> > -};
> > -
> > struct _ctx_layout {
> >    struct sctx_info *addr;
> >    unsigned int size;
> > diff --git a/arch/loongarch/vdso/Makefile
> > b/arch/loongarch/vdso/Makefile
> > index 520f1513f07d..294c16b9517f 100644
> > --- a/arch/loongarch/vdso/Makefile
> > +++ b/arch/loongarch/vdso/Makefile
> > @@ -26,7 +26,7 @@ cflags-vdso := $(ccflags-vdso) \
> >    $(filter -W%,$(filter-out -Wa$(comma)%,$(KBUILD_CFLAGS))) \
> >    -std=gnu11 -fms-extensions -O2 -g -fno-strict-aliasing -
> > fno-common -fno-builtin \
> >    -fno-stack-protector -fno-jump-tables -
> > DDISABLE_BRANCH_PROFILING \
> > - $(call cc-option, -fno-asynchronous-unwind-tables) \
> > + $(call cc-option, -fasynchronous-unwind-tables) \
> >    $(call cc-option, -fno-stack-protector)
> > aflags-vdso := $(ccflags-vdso) \
> >    -D__ASSEMBLY__ -Wa,-gdwarf-2
> > @@ -41,7 +41,7 @@ endif
> >
> > # VDSO linker flags.
> > ldflags-y := -Bsymbolic --no-undefined -soname=linux-vdso.so.1 \
> > - $(filter -E%,$(KBUILD_CFLAGS)) -shared --build-id -T
> > + $(filter -E%,$(KBUILD_CFLAGS)) -shared --build-id --eh-
> > frame-hdr -T
> >
> > #
> > # Shared build commands.
> > diff --git a/arch/loongarch/vdso/sigreturn.S
> > b/arch/loongarch/vdso/sigreturn.S
> > index 9cb3c58fad03..e46c1deacb9e 100644
> > --- a/arch/loongarch/vdso/sigreturn.S
> > +++ b/arch/loongarch/vdso/sigreturn.S
> > @@ -12,13 +12,49 @@
> >
> > #include <asm/regdef.h>
> > #include <asm/asm.h>
> > +#include <asm/asm-offsets.h>
> >
> >    .section .text
> > - .cfi_sections .debug_frame
> >
> > -SYM_FUNC_START(__vdso_rt_sigreturn)
> > + .cfi_startproc
> > + .cfi_signal_frame
> >
> > + /*
> > + * There is a struct rt_sigframe at $sp, set CFA to the
> > address of
> > + * the struct sigcontext in the rt_sigframe to simplify the
> > + * offsets below.
> > + */
> > + .cfi_def_cfa 3, RT_SIGFRAME_SC
> > +
> > + /*
> > + * 72 is DWARF 2 CFA column for the return address from a
> > signal
> > + * handler context on LoongArch, i.e. the PC stored in the
> > + * sigcontext.
> > + */
> Could we use `.cfi_return_column 0` here? Since I recommended
> llvm::libunwind that restore register to zero not destroy it
> and we can use it as a hack, [1]. But I don't know whether the
> libgcc works ok or not in this case.
> Otherwise I think we should update [2] first.
>
> Others look good to me.
>
> [1] https://reviews.llvm.org/D137010#3907282
> [2] https://github.com/loongson/la-abi-specs/blob/release/ladwarf.adoc

I'll take a look but I think [2] needs an update no matter if we use 0
or 72. I.e. if we use 0 we should document the slot for the signal
frame RA instead of $r0.

--
Xi Ruoyao <xry111@xxxxxxxxxxx>