llvm dev - Sep 2019 - An error of asm goto occured while compiling Linux kernel 5.3

Hi all,
I encountered an error while compiling Linux kernel 5.3 to IR.
My LLVM version is 9.0.0 release.
This error said "invalid operand for inline asm constraint
'i'" in
arch/x86/include/asm/jump_table.h.

The source code is
static __always_inline bool arch_static_branch(struct static_key *key,
bool branch)
{
asm_volatile_goto("1:"
".byte " __stringify(STATIC_KEY_INIT_NOP) "\n\t"
".pushsection __jump_table,  \"aw\" \n\t"
_ASM_ALIGN "\n\t"
".long 1b - ., %l[l_yes] - . \n\t"
_ASM_PTR "%c0 + %c1 - .\n\t"
".popsection \n\t"
: :  "i" (key), "i" (branch) : : l_yes);

return false;
l_yes:
return true;
}

The error pointed to asm_volatile_goto("1:"
I'm not sure whether this error is related to asm goto support.

Hi Will,

I'm adding cfe-dev since this potentially covers both areas.

On Thu, 26 Sep 2019 at 15:45, Will Lester via llvm-dev
<llvm-dev at lists.llvm.org> wrote:
> I encountered an error while compiling Linux kernel 5.3 to IR.
> My LLVM version is 9.0.0 release.
The error message you quote can only be emitted after LLVM has started
turning IR into machine code. What's the actual Clang command that
produces this error?
> This error said "invalid operand for inline asm constraint
'i'" in
> arch/x86/include/asm/jump_table.h.
This is a pretty fragile use of that constraint, since it's declaring
that something must be a constant when that can only happen via
optimizations. I'm not convinced that's what's going on here though.
Kernels aren't usually compiled at -O0 and I think I trust the Linux
developers to make sure users are passing an obvious constant.
> The source code is
> static __always_inline bool arch_static_branch(struct static_key *key,
> bool branch)
> {
> asm_volatile_goto("1:"
> ".byte " __stringify(STATIC_KEY_INIT_NOP) "\n\t"
> ".pushsection __jump_table,  \"aw\" \n\t"
> _ASM_ALIGN "\n\t"
> ".long 1b - ., %l[l_yes] - . \n\t"
> _ASM_PTR "%c0 + %c1 - .\n\t"
> ".popsection \n\t"
> : :  "i" (key), "i" (branch) : : l_yes);
>
> return false;
> l_yes:
> return true;
> }
>
> The error pointed to asm_volatile_goto("1:"
> I'm not sure whether this error is related to asm goto support.

On Fri, Sep 27, 2019 at 4:02 AM Tim Northover <t.p.northover at gmail.com>
wrote:
>
> Hi Will,
>
> I'm adding cfe-dev since this potentially covers both areas.
>
> On Thu, 26 Sep 2019 at 15:45, Will Lester via llvm-dev
> <llvm-dev at lists.llvm.org> wrote:
> > I encountered an error while compiling Linux kernel 5.3 to IR.
> > My LLVM version is 9.0.0 release.
>
> The error message you quote can only be emitted after LLVM has started
> turning IR into machine code. What's the actual Clang command that
> produces this error?
>
I have checked the building process and find the error actually happened when
building a .o file, which is  fs/nfs/flexfilelayout/flexfilelayout.o.

The Clang command is:

clang -Wp,-MD,fs/nfs/flexfilelayout/.flexfilelayout.o.d  -nostdinc
-isystem /usr/local/lib/clang/9.0.0/include -I./arch/x86/include
-I./arch/x86/include/generated  -I./include -I./arch/x86/include/uapi
-I./arch/x86/include/generated/uapi -I./include/uapi -I./include/generated/uapi
-include ./include/linux/kconfig.h -include ./include/linux/compiler_types.h
-D__KERNEL__ -Qunused-arguments -Wall -Wundef -Werror=strict-prototypes
-Wno-trigraphs -fno-strict-aliasing -fno-common -fshort-wchar -fno-PIE
-Werror=implicit-function-declaration -Werror=implicit-int -Wno-format-security
-std=gnu89 -no-integrated-as -Werror=unknown-warning-option -mno-sse
-mno-mmx -mno-sse2 -mno-3dnow -mno-avx -m64 -mno-80387 -mstack-alignment=8
-mtune=generic -mno-red-zone -mcmodel=kernel -DCONFIG_X86_X32_ABI
-DCONFIG_AS_CFI=1 -DCONFIG_AS_CFI_SIGNAL_FRAME=1 -DCONFIG_AS_CFI_SECTIONS=1
-DCONFIG_AS_SSSE3=1 -DCONFIG_AS_AVX=1 -DCONFIG_AS_AVX2=1
-DCONFIG_AS_AVX512=1 -DCONFIG_AS_SHA1_NI=1 -DCONFIG_AS_SHA256_NI=1
-Wno-sign-compare -fno-asynchronous-unwind-tables -mretpoline-external-thunk
-fno-delete-null-pointer-checks -Wno-address-of-packed-member -O0
-Wframe-larger-than=2048 -fstack-protector-strong -Wno-format-invalid-specifier
-Wno-gnu -Wno-tautological-compare -mno-global-merge -Wno-unused-const-variable
-fomit-frame-pointer -ftrivial-auto-var-init=pattern -g
-Wdeclaration-after-statement
-Wimplicit-fallthrough -Wvla -Wno-pointer-sign -fno-strict-overflow
-fno-merge-all-constants
-fno-stack-check -Werror=date-time -Werror=incompatible-pointer-types
-fcf-protection=none -Wno-initializer-overrides -Wno-format -Wno-sign-compare
-Wno-format-zero-length   -DMODULE 
-DKBUILD_BASENAME='"flexfilelayout"'
-DKBUILD_MODNAME='"nfs_layout_flexfiles"' -c -o
fs/nfs/flexfilelayout/flexfilelayout.o fs/nfs/flexfilelayout/flexfilelayout.c
> > This error said "invalid operand for inline asm constraint
'i'" in
> > arch/x86/include/asm/jump_table.h.
>
> This is a pretty fragile use of that constraint, since it's declaring
> that something must be a constant when that can only happen via
> optimizations. I'm not convinced that's what's going on here
though.
> Kernels aren't usually compiled at -O0 and I think I trust the Linux
> developers to make sure users are passing an obvious constant.
>
I am doing an analysis of kernel, therefore I use -O0 to keep details of
source code.
> > The source code is
> > static __always_inline bool arch_static_branch(struct static_key *key,
> > bool branch)
> > {
> > asm_volatile_goto("1:"
> > ".byte " __stringify(STATIC_KEY_INIT_NOP) "\n\t"
> > ".pushsection __jump_table,  \"aw\" \n\t"
> > _ASM_ALIGN "\n\t"
> > ".long 1b - ., %l[l_yes] - . \n\t"
> > _ASM_PTR "%c0 + %c1 - .\n\t"
> > ".popsection \n\t"
> > : :  "i" (key), "i" (branch) : : l_yes);
> >
> > return false;
> > l_yes:
> > return true;
> > }
> >
> > The error pointed to asm_volatile_goto("1:"
> > I'm not sure whether this error is related to asm goto support.