Thread (10 messages) 10 messages, 4 authors, 2020-10-13

Re: [Linux-kernel-mentees] [PATCH v2 2/2] kbuild: use interpreters to invoke scripts

From: Masahiro Yamada <masahiroy@kernel.org>
Date: 2020-10-13 16:03:17
Also in: linux-arm-kernel, linux-kbuild, lkml

On Tue, Oct 13, 2020 at 4:03 AM Bernd Petrovitsch
[off-list ref] wrote:
Hi all!

On 12/10/2020 18:42, Ujjwal Kumar wrote:
quoted
On 12/10/20 11:50 pm, Lukas Bulwahn wrote:
quoted

On Mon, 12 Oct 2020, Ujjwal Kumar wrote:
quoted
We cannot rely on execute bits to be set on files in the repository.
The build script should use the explicit interpreter when invoking any
script from the repository.

Link: https://lore.kernel.org/lkml/20200830174409.c24c3f67addcce0cea9a9d4c@linux-foundation.org/ (local)
Link: https://lore.kernel.org/lkml/202008271102.FEB906C88@keescook/ (local)

Suggested-by: Andrew Morton <akpm@linux-foundation.org>
Suggested-by: Kees Cook <redacted>
Suggested-by: Lukas Bulwahn <lukas.bulwahn@gmail.com>
Signed-off-by: Ujjwal Kumar <redacted>
---
 Makefile                          | 4 ++--
 arch/arm64/kernel/vdso/Makefile   | 2 +-
 arch/arm64/kernel/vdso32/Makefile | 2 +-
 arch/ia64/Makefile                | 4 ++--
 arch/nds32/kernel/vdso/Makefile   | 2 +-
 scripts/Makefile.build            | 2 +-
 scripts/Makefile.package          | 4 ++--
 7 files changed, 10 insertions(+), 10 deletions(-)
diff --git a/Makefile b/Makefile
index 0af7945caa61..df20e71dd7c8 100644
--- a/Makefile
+++ b/Makefile
@@ -1256,7 +1256,7 @@ include/generated/utsrelease.h: include/config/kernel.release FORCE
 PHONY += headerdep
 headerdep:
    $(Q)find $(srctree)/include/ -name '*.h' | xargs --max-args 1 \
-   $(srctree)/scripts/headerdep.pl -I$(srctree)/include
+   $(PERL) $(srctree)/scripts/headerdep.pl -I$(srctree)/include

 # ---------------------------------------------------------------------------
 # Kernel headers
@@ -1312,7 +1312,7 @@ PHONY += kselftest-merge
 kselftest-merge:
    $(if $(wildcard $(objtree)/.config),, $(error No .config exists, config your kernel first!))
    $(Q)find $(srctree)/tools/testing/selftests -name config | \
-           xargs $(srctree)/scripts/kconfig/merge_config.sh -m $(objtree)/.config
+           xargs $(CONFIG_SHELL) $(srctree)/scripts/kconfig/merge_config.sh -m $(objtree)/.config
    $(Q)$(MAKE) -f $(srctree)/Makefile olddefconfig

 # ---------------------------------------------------------------------------
diff --git a/arch/arm64/kernel/vdso/Makefile b/arch/arm64/kernel/vdso/Makefile
index edccdb77c53e..fb07804b7fc1 100644
--- a/arch/arm64/kernel/vdso/Makefile
+++ b/arch/arm64/kernel/vdso/Makefile
@@ -65,7 +65,7 @@ $(obj)/%.so: $(obj)/%.so.dbg FORCE
 # Generate VDSO offsets using helper script
 gen-vdsosym := $(srctree)/$(src)/gen_vdso_offsets.sh
 quiet_cmd_vdsosym = VDSOSYM $@
-      cmd_vdsosym = $(NM) $< | $(gen-vdsosym) | LC_ALL=C sort > $@
+      cmd_vdsosym = $(NM) $< | $(CONFIG_SHELL) $(gen-vdsosym) | LC_ALL=C sort > $@

 include/generated/vdso-offsets.h: $(obj)/vdso.so.dbg FORCE
    $(call if_changed,vdsosym)
diff --git a/arch/arm64/kernel/vdso32/Makefile b/arch/arm64/kernel/vdso32/Makefile
index 7f96a1a9f68c..617c9ac58156 100644
--- a/arch/arm64/kernel/vdso32/Makefile
+++ b/arch/arm64/kernel/vdso32/Makefile
@@ -205,7 +205,7 @@ quiet_cmd_vdsomunge = MUNGE   $@
 gen-vdsosym := $(srctree)/$(src)/../vdso/gen_vdso_offsets.sh
 quiet_cmd_vdsosym = VDSOSYM $@
 # The AArch64 nm should be able to read an AArch32 binary
-      cmd_vdsosym = $(NM) $< | $(gen-vdsosym) | LC_ALL=C sort > $@
+      cmd_vdsosym = $(NM) $< | $(CONFIG_SHELL) $(gen-vdsosym) | LC_ALL=C sort > $@

 # Install commands for the unstripped file
 quiet_cmd_vdso_install = INSTALL32 $@
diff --git a/arch/ia64/Makefile b/arch/ia64/Makefile
index 703b1c4f6d12..86d42a2d09cb 100644
--- a/arch/ia64/Makefile
+++ b/arch/ia64/Makefile
@@ -27,8 +27,8 @@ cflags-y  := -pipe $(EXTRA) -ffixed-r13 -mfixed-range=f12-f15,f32-f127 \
               -falign-functions=32 -frename-registers -fno-optimize-sibling-calls
 KBUILD_CFLAGS_KERNEL := -mconstant-gp

-GAS_STATUS = $(shell $(srctree)/arch/ia64/scripts/check-gas "$(CC)" "$(OBJDUMP)")
-KBUILD_CPPFLAGS += $(shell $(srctree)/arch/ia64/scripts/toolchain-flags "$(CC)" "$(OBJDUMP)" "$(READELF)")
+GAS_STATUS = $(shell $(CONFIG_SHELL) $(srctree)/arch/ia64/scripts/check-gas "$(CC)" "$(OBJDUMP)")
+KBUILD_CPPFLAGS += $(shell $(CONFIG_SHELL) $(srctree)/arch/ia64/scripts/toolchain-flags "$(CC)" "$(OBJDUMP)" "$(READELF)")
Here is an instance of what Masahiro-san pointed out being wrong.

Ujjwal, will you send a v3?
Following is the quoted text from the reply mail from Masahiro
quoted
quoted
-GAS_STATUS     = $(shell $(srctree)/arch/ia64/scripts/check-gas "$(CC)" "$(OBJDUMP)")
-KBUILD_CPPFLAGS += $(shell $(srctree)/arch/ia64/scripts/toolchain-flags "$(CC)" "$(OBJDUMP)" "$(READELF)")
+GAS_STATUS     = $($(CONFIG_SHELL) $(srctree)/arch/ia64/scripts/check-gas "$(CC)" "$(OBJDUMP)")
+KBUILD_CPPFLAGS += $($(CONFIG_SHELL) $(srctree)/arch/ia64/scripts/toolchain-flags "$(CC)" "$(OBJDUMP)" "$(READELF)")


These changes look wrong to me.

$($(CONFIG_SHELL)    ->  $(shell $(CONFIG_SHELL)
From the above text, I understand as follows:
Did you actually *test* that (expecially) these lines work
afterwards as good as before?
quoted
That my proposed change:
$(shell $(src...)    ->  $($(CONFIG_SHELL) $(src...)

is WRONG
Yup, as it's in a Makefile and that's a Makefile construct.
quoted
and in the next line he suggested the required correction.
That being:
$($(CONFIG_SHELL)    ->  $(shell $(CONFIG_SHELL)
Such stuff should generally not be needed as the to-be-used
shell can be set in Makefiles via a "SHELL = " assignment
(defaulting to /bin/sh - what else;-).
Flags for the shell can BTW set with ".SHELLFLAGS = ".

You are talking about a different thing.



Take the current code as an example:

$(shell $(srctree)/arch/ia64/scripts/check-gas "$(CC)" "$(OBJDUMP)")


Here are two shell invocations.


[1]
The command
$(srctree)/arch/ia64/scripts/check-gas "$(CC)" "$(OBJDUMP)"
is run in /bin/sh because the default value of SHELL is /bin/sh.


[2]
The script, arch/ia64/scripts/check-gas, is run in /bin/sh
because the hash-bang (the first line of check-gas)
specifies #!/bin/sh




Bernd is talking about [1].

In contrast, this patch is addressing [2] because
Andrew Morton suggested to run scripts without relying
on the executable bit.
(and, after this patch, we run scripts without relying
on the hash-bang because we now specify the interpreter.)


Of course, [1] and [2] can be different.


I always want to use /bin/sh for [1],
so please do not use bash-extension inside $(shell  ...)


You have more choices for [2].

If arch/ia64/scripts/check-gas had been written with bash-extension,
the code would have been changed into:

$(shell $(BASH) $(srctree)/arch/ia64/scripts/check-gas "$(CC)" "$(OBJDUMP)")


I hope this will be clearer.



So please
-) learn basic "Makefile" + "make" before brainlessly patching
   a Makefile.
-) actually testy your changes to make sure the patch didn't
   broke anything
-) and - last but not least - check if there isn't a shell
   already set (and which).

MfG,
        Bernd
--
There is no cloud, just other people computers.
-- https://static.fsf.org/nosvn/stickers/thereisnocloud.svg


--
Best Regards
Masahiro Yamada
_______________________________________________
Linux-kernel-mentees mailing list
Linux-kernel-mentees@lists.linuxfoundation.org
https://lists.linuxfoundation.org/mailman/listinfo/linux-kernel-mentees
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help