2015-08-21 17:44:07

by Max Filippov

[permalink] [raw]
Subject: [PATCH v3] xtensa: improve vmlinux.lds.S sed post-processing

Current sed script makes assumptions about the structure of rules that
group .text sections in the vmlinux linker script. These assumptions
get broken occasionally, e.g.: 779c88c94c34 "ARM: 8321/1: asm-generic:
introduce.text.fixup input section", or 9bebe9e5b0f3 "kbuild: Fix
.text.unlikely placement".

Rewrite sed rules so that they don't depend on number/arrangement of text
sections in *(...) blocks.

Signed-off-by: Max Filippov <[email protected]>
---
This fixes breakage in the linux-next. Chris, could you please apply it to
your for_next branch?

Changes v2 -> v3:
- rebase to the current master

Changes v1 -> v2:
- don't change arbitrary sections containing .text, only .text, .ref.text
and .[a-z]*it.text

arch/xtensa/kernel/Makefile | 9 +++++----
1 file changed, 5 insertions(+), 4 deletions(-)

diff --git a/arch/xtensa/kernel/Makefile b/arch/xtensa/kernel/Makefile
index d3a0f0f..ad93217 100644
--- a/arch/xtensa/kernel/Makefile
+++ b/arch/xtensa/kernel/Makefile
@@ -27,10 +27,11 @@ AFLAGS_head.o += -mtext-section-literals
#
# Replicate rules in scripts/Makefile.build

-sed-y = -e 's/\*(\(\.[a-z]*it\|\.ref\|\)\.text)/*(\1.literal \1.text)/g' \
- -e 's/\.text\.unlikely/.literal.unlikely .text.unlikely/g' \
- -e 's/\*(\(\.text .*\))/*(.literal \1)/g' \
- -e 's/\*(\(\.text\.[a-z]*\))/*(\1.literal \1)/g'
+sed-y = -e ':a; s/\*(\([^)]*\)\.text\.unlikely/*(\1.literal.unlikely .{text}.unlikely/; ta; ' \
+ -e ':b; s/\*(\([^)]*\)\.text\(\.[a-z]*\)/*(\1.{text}\2.literal .{text}\2/; tb; ' \
+ -e ':c; s/\*(\([^)]*\)\(\.[a-z]*it\|\.ref\)\.text/*(\1\2.literal \2.{text}/; tc; ' \
+ -e ':d; s/\*(\([^)]\+ \|\)\.text/*(\1.literal .{text}/; td; ' \
+ -e 's/\.{text}/.text/g'

quiet_cmd__cpp_lds_S = LDS $@
cmd__cpp_lds_S = $(CPP) $(cpp_flags) -P -C -Uxtensa -D__ASSEMBLY__ $< \
--
1.8.1.4


2015-08-21 18:23:01

by Chris Zankel

[permalink] [raw]
Subject: Re: [PATCH v3] xtensa: improve vmlinux.lds.S sed post-processing

Thanks Max. Running a test build and will add to xtensa-next.

On Fri, Aug 21, 2015 at 10:43 AM, Max Filippov <[email protected]> wrote:
> Current sed script makes assumptions about the structure of rules that
> group .text sections in the vmlinux linker script. These assumptions
> get broken occasionally, e.g.: 779c88c94c34 "ARM: 8321/1: asm-generic:
> introduce.text.fixup input section", or 9bebe9e5b0f3 "kbuild: Fix
> .text.unlikely placement".
>
> Rewrite sed rules so that they don't depend on number/arrangement of text
> sections in *(...) blocks.
>
> Signed-off-by: Max Filippov <[email protected]>
> ---
> This fixes breakage in the linux-next. Chris, could you please apply it to
> your for_next branch?
>
> Changes v2 -> v3:
> - rebase to the current master
>
> Changes v1 -> v2:
> - don't change arbitrary sections containing .text, only .text, .ref.text
> and .[a-z]*it.text
>
> arch/xtensa/kernel/Makefile | 9 +++++----
> 1 file changed, 5 insertions(+), 4 deletions(-)
>
> diff --git a/arch/xtensa/kernel/Makefile b/arch/xtensa/kernel/Makefile
> index d3a0f0f..ad93217 100644
> --- a/arch/xtensa/kernel/Makefile
> +++ b/arch/xtensa/kernel/Makefile
> @@ -27,10 +27,11 @@ AFLAGS_head.o += -mtext-section-literals
> #
> # Replicate rules in scripts/Makefile.build
>
> -sed-y = -e 's/\*(\(\.[a-z]*it\|\.ref\|\)\.text)/*(\1.literal \1.text)/g' \
> - -e 's/\.text\.unlikely/.literal.unlikely .text.unlikely/g' \
> - -e 's/\*(\(\.text .*\))/*(.literal \1)/g' \
> - -e 's/\*(\(\.text\.[a-z]*\))/*(\1.literal \1)/g'
> +sed-y = -e ':a; s/\*(\([^)]*\)\.text\.unlikely/*(\1.literal.unlikely .{text}.unlikely/; ta; ' \
> + -e ':b; s/\*(\([^)]*\)\.text\(\.[a-z]*\)/*(\1.{text}\2.literal .{text}\2/; tb; ' \
> + -e ':c; s/\*(\([^)]*\)\(\.[a-z]*it\|\.ref\)\.text/*(\1\2.literal \2.{text}/; tc; ' \
> + -e ':d; s/\*(\([^)]\+ \|\)\.text/*(\1.literal .{text}/; td; ' \
> + -e 's/\.{text}/.text/g'
>
> quiet_cmd__cpp_lds_S = LDS $@
> cmd__cpp_lds_S = $(CPP) $(cpp_flags) -P -C -Uxtensa -D__ASSEMBLY__ $< \
> --
> 1.8.1.4
>