Re: [PATCH v4] arm64: sysreg: Make mrs_s and msr_s macros work with Clang and LTO

From: Nick Desaulniers
Date: Tue Apr 23 2019 - 19:49:51 EST

Next message: Steven Rostedt: "Re: [RFC PATCH 1/1] x86/ftrace: make ftrace_int3_handler() not to skip fops invocation"
Previous message: Aubrey Li: "Re: [RFC PATCH v2 15/17] sched: Trivial forced-newidle balancer"
In reply to: Kees Cook: "[PATCH v4] arm64: sysreg: Make mrs_s and msr_s macros work with Clang and LTO"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

On Tue, Apr 23, 2019 at 4:26 PM Kees Cook <keescook@xxxxxxxxxxxx> wrote:
>
> Clang's integrated assembler does not allow assembly macros defined
> in one inline asm block using the .macro directive to be used across
> separate asm blocks. LLVM developers consider this a feature and not a
> bug, recommending code refactoring:
>
> https://bugs.llvm.org/show_bug.cgi?id=19749
>
> As binutils doesn't allow macros to be redefined, this change uses
> UNDEFINE_MRS_S and UNDEFINE_MSR_S to define corresponding macros
> in-place and workaround gcc and clang limitations on redefining macros
> across different assembler blocks.
>
> Specifically, the current state after preprocessing looks like this:
>
> asm volatile(".macro mXX_s ... .endm");
> void f()
> {
> asm volatile("mXX_s a, b");
> }
>
> With GCC, it gives macro redefinition error because sysreg.h is included
> in multiple source files, and assembler code for all of them is later
> combined for LTO (I've seen an intermediate file with hundreds of
> identical definitions).
>
> With clang, it gives macro undefined error because clang doesn't allow
> sharing macros between inline asm statements.
>
> I also seem to remember catching another sort of undefined error with
> GCC due to reordering of macro definition asm statement and generated
> asm code for function that uses the macro.
>
> The solution with defining and undefining for each use, while certainly
> not elegant, satisfies both GCC and clang, LTO and non-LTO.
>
> Co-developed-by: Alex Matveev <alxmtvv@xxxxxxxxx>
> Co-developed-by: Yury Norov <ynorov@xxxxxxxxxxxxxxxxxx>
> Co-developed-by: Sami Tolvanen <samitolvanen@xxxxxxxxxx>
> Signed-off-by: Kees Cook <keescook@xxxxxxxxxxxx>
> ---
> v4: move to using preprocessor entirely, split constraints for "rZ" case.
> v3: added more uses in irqflags, updated commit log, based on
> discussion in https://lore.kernel.org/patchwork/patch/851580/
> ---
> arch/arm64/include/asm/irqflags.h | 8 ++---
> arch/arm64/include/asm/kvm_hyp.h | 4 +--
> arch/arm64/include/asm/sysreg.h | 51 +++++++++++++++++++++++--------
> 3 files changed, 44 insertions(+), 19 deletions(-)
>
> diff --git a/arch/arm64/include/asm/irqflags.h b/arch/arm64/include/asm/irqflags.h
> index 43d8366c1e87..d359f28765cb 100644
> --- a/arch/arm64/include/asm/irqflags.h
> +++ b/arch/arm64/include/asm/irqflags.h
> @@ -43,7 +43,7 @@ static inline void arch_local_irq_enable(void)
> asm volatile(ALTERNATIVE(
> "msr daifclr, #2 // arch_local_irq_enable\n"
> "nop",
> - "msr_s " __stringify(SYS_ICC_PMR_EL1) ",%0\n"
> + __msr_s(SYS_ICC_PMR_EL1)
> "dsb sy",
> ARM64_HAS_IRQ_PRIO_MASKING)
> :
> @@ -55,7 +55,7 @@ static inline void arch_local_irq_disable(void)
> {
> asm volatile(ALTERNATIVE(
> "msr daifset, #2 // arch_local_irq_disable",
> - "msr_s " __stringify(SYS_ICC_PMR_EL1) ", %0",
> + __msr_s(SYS_ICC_PMR_EL1),
> ARM64_HAS_IRQ_PRIO_MASKING)
> :
> : "r" ((unsigned long) GIC_PRIO_IRQOFF)
> @@ -86,7 +86,7 @@ static inline unsigned long arch_local_save_flags(void)
> "mov %0, %1\n"
> "nop\n"
> "nop",
> - "mrs_s %0, " __stringify(SYS_ICC_PMR_EL1) "\n"
> + __mrs_s(SYS_ICC_PMR_EL1)
> "ands %1, %1, " __stringify(PSR_I_BIT) "\n"
> "csel %0, %0, %2, eq",
> ARM64_HAS_IRQ_PRIO_MASKING)
> @@ -116,7 +116,7 @@ static inline void arch_local_irq_restore(unsigned long flags)
> asm volatile(ALTERNATIVE(
> "msr daif, %0\n"
> "nop",
> - "msr_s " __stringify(SYS_ICC_PMR_EL1) ", %0\n"
> + __msr_s(SYS_ICC_PMR_EL1)
> "dsb sy",
> ARM64_HAS_IRQ_PRIO_MASKING)
> : "+r" (flags)
> diff --git a/arch/arm64/include/asm/kvm_hyp.h b/arch/arm64/include/asm/kvm_hyp.h
> index 4da765f2cca5..f2e681d657bb 100644
> --- a/arch/arm64/include/asm/kvm_hyp.h
> +++ b/arch/arm64/include/asm/kvm_hyp.h
> @@ -30,7 +30,7 @@
> ({ \
> u64 reg; \
> asm volatile(ALTERNATIVE("mrs %0, " __stringify(r##nvh),\
> - "mrs_s %0, " __stringify(r##vh),\
> + __mrs_s(r##vh), \
> ARM64_HAS_VIRT_HOST_EXTN) \
> : "=r" (reg)); \
> reg; \
> @@ -40,7 +40,7 @@
> do { \
> u64 __val = (u64)(v); \
> asm volatile(ALTERNATIVE("msr " __stringify(r##nvh) ", %x0",\
> - "msr_s " __stringify(r##vh) ", %x0",\
> + __msr_s_x0(r##vh), \
> ARM64_HAS_VIRT_HOST_EXTN) \
> : : "rZ" (__val)); \
> } while (0)
> diff --git a/arch/arm64/include/asm/sysreg.h b/arch/arm64/include/asm/sysreg.h
> index 5b267dec6194..21d8f6dd4d79 100644
> --- a/arch/arm64/include/asm/sysreg.h
> +++ b/arch/arm64/include/asm/sysreg.h
> @@ -746,20 +746,45 @@
> #include <linux/build_bug.h>
> #include <linux/types.h>
>
> -asm(
> -" .irp num,0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30\n"
> -" .equ .L__reg_num_x\\num, \\num\n"
> -" .endr\n"
> +#define __DEFINE_MRS_MSR_S_REGNUM \
> +" .irp num,0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30\n" \
> +" .equ .L__reg_num_x\\num, \\num\n" \
> +" .endr\n" \
> " .equ .L__reg_num_xzr, 31\n"
> -"\n"
> -" .macro mrs_s, rt, sreg\n"
> - __emit_inst(0xd5200000|(\\sreg)|(.L__reg_num_\\rt))
> +
> +#define DEFINE_MRS_S \
> + __DEFINE_MRS_MSR_S_REGNUM \
> +" .macro mrs_s, rt, sreg\n" \
> + __emit_inst(0xd5200000|(\\sreg)|(.L__reg_num_\\rt)) \
> " .endm\n"
> -"\n"
> -" .macro msr_s, sreg, rt\n"
> - __emit_inst(0xd5000000|(\\sreg)|(.L__reg_num_\\rt))
> +
> +#define DEFINE_MSR_S \
> + __DEFINE_MRS_MSR_S_REGNUM \
> +" .macro msr_s, sreg, rt\n" \
> + __emit_inst(0xd5000000|(\\sreg)|(.L__reg_num_\\rt)) \
> " .endm\n"
> -);
> +
> +#define UNDEFINE_MRS_S \
> +" .purgem mrs_s\n"
> +
> +#define UNDEFINE_MSR_S \
> +" .purgem msr_s\n"
> +
> +#define __mrs_s(r) \
> + DEFINE_MRS_S \
> +" mrs_s %0, " __stringify(r) "\n" \
> + UNDEFINE_MRS_S
> +
> +#define __msr_s(r) \
> + DEFINE_MSR_S \
> +" msr_s " __stringify(r) ", %0\n" \
> + UNDEFINE_MSR_S
> +
> +/* For use with "rZ" constraints. */
> +#define __msr_s_x0(r) \
> + DEFINE_MSR_S \
> +" msr_s " __stringify(r) ", %x0\n" \
> + UNDEFINE_MSR_S

Cool, thanks for sending this revision; it's much cleaner IMO. I
wonder (maybe Ard or Mark know) if we can drop the non-x0-variant of
msr and just use the x0-variant everywhere (but with the simpler name
of __msr_s)?

arch_local_irq_restore, arch_local_save_flags, arch_local_irq_disable,
and arch_local_irq_enable all seem to be dealing with unsigned longs
for %0, IIUC.

>
> /*
> * Unlike read_cpuid, calls to read_sysreg are never expected to be
> @@ -787,13 +812,13 @@ asm(
> */
> #define read_sysreg_s(r) ({ \
> u64 __val; \
> - asm volatile("mrs_s %0, " __stringify(r) : "=r" (__val)); \
> + asm volatile(__mrs_s(r) : "=r" (__val)); \
> __val; \
> })
>
> #define write_sysreg_s(v, r) do { \
> u64 __val = (u64)(v); \
> - asm volatile("msr_s " __stringify(r) ", %x0" : : "rZ" (__val)); \
> + asm volatile(__msr_s_x0(r) : : "rZ" (__val)); \
> } while (0)
>
> /*
> --
> 2.17.1
>
>
> --
> Kees Cook

--
Thanks,
~Nick Desaulniers

Next message: Steven Rostedt: "Re: [RFC PATCH 1/1] x86/ftrace: make ftrace_int3_handler() not to skip fops invocation"
Previous message: Aubrey Li: "Re: [RFC PATCH v2 15/17] sched: Trivial forced-newidle balancer"
In reply to: Kees Cook: "[PATCH v4] arm64: sysreg: Make mrs_s and msr_s macros work with Clang and LTO"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]