Re: [PATCH v7 1/2] scripts: ftrace - move the sort-processing in ftrace_init

[Steven Rostedt <rostedt@goodmis.org>]

[-- Attachment #1: Type: text/plain, Size: 2755 bytes --]

On Tue,  7 Dec 2021 23:13:46 +0800
Yinan Liu <yinan@linux.alibaba.com> wrote:

> When the kernel starts, the initialization of ftrace takes
> up a portion of the time (approximately 6~8ms) to sort mcount
> addresses. We can save this time by moving mcount-sorting to
> compile time.
> 
> Signed-off-by: Yinan Liu <yinan@linux.alibaba.com>
> Reported-by: kernel test robot <lkp@intel.com>
> Reported-by: kernel test robot <oliver.sang@intel.com>
> ---
>  kernel/trace/ftrace.c   |  11 +++-
>  scripts/Makefile        |   6 +-
>  scripts/link-vmlinux.sh |   6 +-
>  scripts/sorttable.c     |   2 +
>  scripts/sorttable.h     | 120 +++++++++++++++++++++++++++++++++++++++-
>  5 files changed, 137 insertions(+), 8 deletions(-)
> 
> diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c
> index 30bc880c3849..9ca63df6553a 100644
> --- a/kernel/trace/ftrace.c
> +++ b/kernel/trace/ftrace.c
> @@ -6406,8 +6406,15 @@ static int ftrace_process_locs(struct module *mod,
>  	if (!count)
>  		return 0;
>  
> -	sort(start, count, sizeof(*start),
> -	     ftrace_cmp_ips, NULL);
> +	/*
> +	 * Sorting mcount in vmlinux at build time depend on
> +	 * CONFIG_BUILDTIME_TABLE_SORT, while mcount loc in
> +	 * modules can not be sorted at build time.
> +	 */
> +	if (!IS_ENABLED(CONFIG_BUILDTIME_TABLE_SORT) || mod) {
> +		sort(start, count, sizeof(*start),
> +		     ftrace_cmp_ips, NULL);
> +	}
>  
>  	start_pg = ftrace_allocate_pages(count);
>  	if (!start_pg)
> diff --git a/scripts/Makefile b/scripts/Makefile
> index 9adb6d247818..d27503469f4b 100644
> --- a/scripts/Makefile
> +++ b/scripts/Makefile
> @@ -17,6 +17,7 @@ hostprogs-always-$(CONFIG_SYSTEM_EXTRA_CERTIFICATE)	+= insert-sys-cert
>  hostprogs-always-$(CONFIG_SYSTEM_REVOCATION_LIST)	+= extract-cert
>  
>  HOSTCFLAGS_sorttable.o = -I$(srctree)/tools/include
> +HOSTLDLIBS_sorttable = -lpthread
>  HOSTCFLAGS_asn1_compiler.o = -I$(srctree)/include
>  HOSTCFLAGS_sign-file.o = $(CRYPTO_CFLAGS)
>  HOSTLDLIBS_sign-file = $(CRYPTO_LIBS)
> @@ -29,7 +30,10 @@ ARCH := x86
>  endif
>  HOSTCFLAGS_sorttable.o += -I$(srctree)/tools/arch/x86/include
>  HOSTCFLAGS_sorttable.o += -DUNWINDER_ORC_ENABLED
> -HOSTLDLIBS_sorttable = -lpthread
> +endif
> +
> +ifdef CONFIG_FUNCTION_TRACER
> +HOSTCFLAGS_sorttable.o += -DMCOUNT_SORT_ENABLED
>  endif
>

I dropped this patch again (note I pulled in patch 2, so you do not
need to resend it). The attached config fails to build. The reason is
that you need to test for CONFIG_DYNAMIC_FTRACE and not
CONFIG_FUNCTION_TRACER.  The function tracer can be statically enabled,
which means all the functions just call "mcount" (or whatever), and
they are never patched. This means that the mcount location is not
created.

Just resend a v8 of this patch. Thanks!

-- Steve

[-- Attachment #2: config-fail.gz --]
[-- Type: application/gzip, Size: 30299 bytes --]