riscv now uses -fpatchable-function-entry with dynamic ftrace after
the `afc76b8b8` commit, which means recordmcount shouldn't be called
to create the __mcount_loc section before the vmlinux linking.

We should select FTRACE_MCOUNT_USE_PATCHABLE_FUNCTION_ENTRY to tell
Makefile not to run recordmcount.

Signed-off-by: Song Shuai <suagrfillet@gmail.com>
---
 arch/riscv/Kconfig | 1 +
 1 file changed, 1 insertion(+)

diff --git a/arch/riscv/Kconfig b/arch/riscv/Kconfig
index c22f58155948..25db8cea876a 100644
--- a/arch/riscv/Kconfig
+++ b/arch/riscv/Kconfig
@@ -246,6 +246,7 @@ config ARCH_RV64I
 	select HAVE_DYNAMIC_FTRACE if !XIP_KERNEL && MMU && $(cc-option,-fpatchable-function-entry=8)
 	select HAVE_DYNAMIC_FTRACE_WITH_REGS if HAVE_DYNAMIC_FTRACE
 	select HAVE_FTRACE_MCOUNT_RECORD if !XIP_KERNEL
+	select FTRACE_MCOUNT_USE_PATCHABLE_FUNCTION_ENTRY if DYNAMIC_FTRACE
 	select HAVE_FUNCTION_GRAPH_TRACER
 	select HAVE_FUNCTION_TRACER if !XIP_KERNEL
 	select SWIOTLB if MMU
-- 
2.20.1

Song Shuai <suagrfillet@gmail.com> 于2022年9月23日周五 10:03写道：
>
> riscv now uses -fpatchable-function-entry with dynamic ftrace after
> the `afc76b8b8` commit, which means recordmcount shouldn't be called
> to create the __mcount_loc section before the vmlinux linking.
>
> We should select FTRACE_MCOUNT_USE_PATCHABLE_FUNCTION_ENTRY to tell
> Makefile not to run recordmcount.
>
Hi,Guo:

Can this patch squeeze into your v3 series ?

Thanks,
Song
> Signed-off-by: Song Shuai <suagrfillet@gmail.com>
> ---
>  arch/riscv/Kconfig | 1 +
>  1 file changed, 1 insertion(+)
>
> diff --git a/arch/riscv/Kconfig b/arch/riscv/Kconfig
> index c22f58155948..25db8cea876a 100644
> --- a/arch/riscv/Kconfig
> +++ b/arch/riscv/Kconfig
> @@ -246,6 +246,7 @@ config ARCH_RV64I
>         select HAVE_DYNAMIC_FTRACE if !XIP_KERNEL && MMU && $(cc-option,-fpatchable-function-entry=8)
>         select HAVE_DYNAMIC_FTRACE_WITH_REGS if HAVE_DYNAMIC_FTRACE
>         select HAVE_FTRACE_MCOUNT_RECORD if !XIP_KERNEL
> +       select FTRACE_MCOUNT_USE_PATCHABLE_FUNCTION_ENTRY if DYNAMIC_FTRACE
>         select HAVE_FUNCTION_GRAPH_TRACER
>         select HAVE_FUNCTION_TRACER if !XIP_KERNEL
>         select SWIOTLB if MMU
> --
> 2.20.1
>

On Tue, Nov 22, 2022 at 6:11 PM Song Shuai <suagrfillet@gmail.com> wrote:
>
> Song Shuai <suagrfillet@gmail.com> 于2022年9月23日周五 10:03写道：
> >
> > riscv now uses -fpatchable-function-entry with dynamic ftrace after
> > the `afc76b8b8` commit, which means recordmcount shouldn't be called
The ref commit format needs to be corrected. Others:
Tested-by: Guo Ren <guoren@kernel.org>

> > to create the __mcount_loc section before the vmlinux linking.
> >
> > We should select FTRACE_MCOUNT_USE_PATCHABLE_FUNCTION_ENTRY to tell
> > Makefile not to run recordmcount.
> >
> Hi,Guo:
>
> Can this patch squeeze into your v3 series ?
They don't have a dependency, and it's unnecessary. A single patch is
easy for the maintainer to pick up.

>
> Thanks,
> Song
> > Signed-off-by: Song Shuai <suagrfillet@gmail.com>
> > ---
> >  arch/riscv/Kconfig | 1 +
> >  1 file changed, 1 insertion(+)
> >
> > diff --git a/arch/riscv/Kconfig b/arch/riscv/Kconfig
> > index c22f58155948..25db8cea876a 100644
> > --- a/arch/riscv/Kconfig
> > +++ b/arch/riscv/Kconfig
> > @@ -246,6 +246,7 @@ config ARCH_RV64I
> >         select HAVE_DYNAMIC_FTRACE if !XIP_KERNEL && MMU && $(cc-option,-fpatchable-function-entry=8)
> >         select HAVE_DYNAMIC_FTRACE_WITH_REGS if HAVE_DYNAMIC_FTRACE
> >         select HAVE_FTRACE_MCOUNT_RECORD if !XIP_KERNEL
> > +       select FTRACE_MCOUNT_USE_PATCHABLE_FUNCTION_ENTRY if DYNAMIC_FTRACE
> >         select HAVE_FUNCTION_GRAPH_TRACER
> >         select HAVE_FUNCTION_TRACER if !XIP_KERNEL
> >         select SWIOTLB if MMU
> > --
> > 2.20.1
> >



-- 
Best Regards
 Guo Ren

Guo Ren <guoren@kernel.org> 于2022年11月28日周一 21:47写道：
>
> On Tue, Nov 22, 2022 at 6:11 PM Song Shuai <suagrfillet@gmail.com> wrote:
> >
> > Song Shuai <suagrfillet@gmail.com> 于2022年9月23日周五 10:03写道：
> > >
> > > riscv now uses -fpatchable-function-entry with dynamic ftrace after
> > > the `afc76b8b8` commit, which means recordmcount shouldn't be called
> The ref commit format needs to be corrected. Others:
> Tested-by: Guo Ren <guoren@kernel.org>
>
Thanks for your test.
I'll correct this ref in the next version.
> > > to create the __mcount_loc section before the vmlinux linking.
> > >
> > > We should select FTRACE_MCOUNT_USE_PATCHABLE_FUNCTION_ENTRY to tell
> > > Makefile not to run recordmcount.
> > >
> > Hi,Guo:
> >
> > Can this patch squeeze into your v3 series ?
> They don't have a dependency, and it's unnecessary. A single patch is
> easy for the maintainer to pick up.
>
Got it.
> >
> > Thanks,
> > Song
> > > Signed-off-by: Song Shuai <suagrfillet@gmail.com>
> > > ---
> > >  arch/riscv/Kconfig | 1 +
> > >  1 file changed, 1 insertion(+)
> > >
> > > diff --git a/arch/riscv/Kconfig b/arch/riscv/Kconfig
> > > index c22f58155948..25db8cea876a 100644
> > > --- a/arch/riscv/Kconfig
> > > +++ b/arch/riscv/Kconfig
> > > @@ -246,6 +246,7 @@ config ARCH_RV64I
> > >         select HAVE_DYNAMIC_FTRACE if !XIP_KERNEL && MMU && $(cc-option,-fpatchable-function-entry=8)
> > >         select HAVE_DYNAMIC_FTRACE_WITH_REGS if HAVE_DYNAMIC_FTRACE
> > >         select HAVE_FTRACE_MCOUNT_RECORD if !XIP_KERNEL
> > > +       select FTRACE_MCOUNT_USE_PATCHABLE_FUNCTION_ENTRY if DYNAMIC_FTRACE
> > >         select HAVE_FUNCTION_GRAPH_TRACER
> > >         select HAVE_FUNCTION_TRACER if !XIP_KERNEL
> > >         select SWIOTLB if MMU
> > > --
> > > 2.20.1
> > >
>
>
>
> --
> Best Regards
>  Guo Ren



-- 
Thanks,
Song