diff mbox series

[2/3] objtool: move libelf check out of top Makefile

Message ID 1539681053-24388-3-git-send-email-yamada.masahiro@socionext.com
State New
Headers show
Series kbuild: add scripts/Makefile.toolcheck to check necessary host tools | expand

Commit Message

Masahiro Yamada Oct. 16, 2018, 9:10 a.m. UTC
Bernd Edlinger reports:

  The next make after an oldconfig reads in the outdated
  include/config/auto.conf which can kill the make before
  it is able to call the syncconfig target.

  $ make defconfig
  *** Default configuration is based on 'x86_64_defconfig'
  $ make
  scripts/kconfig/conf  --syncconfig Kconfig
  Makefile:936: *** "Cannot generate ORC metadata for CONFIG_UNWINDER_ORC=y,
  please install libelf-dev, libelf-devel or elfutils-libelf-devel".  Stop.

  $ sed -i s/CONFIG_UNWINDER_ORC=y// .configs
  $ make oldconfig
  Choose kernel unwinder
  > 1. ORC unwinder (UNWINDER_ORC) (NEW)

    2. Frame pointer unwinder (UNWINDER_FRAME_POINTER)
  choice[1-2?]: 2
  $ make
  Makefile:936: *** "Cannot generate ORC metadata for CONFIG_UNWINDER_ORC=y,
  please install libelf-dev, libelf-devel or elfutils-libelf-devel".  Stop.

I tried to fix it by moving the libelf evaluation to the Kconfig
stage [1], but it was rejected.

Kconfig is now able to run shell commands to evaluate any prerequisites
needed to enable CONFIG options.

For example, as shown in commit 2a61f4747eea ("stack-protector: test
compiler capability in Kconfig and drop AUTO mode"), the new Kconfig
syntax is useful to evaluate target compiler capabilities.

However, disabling the CONFIG option silently just because of missing
libelf would not be a preferred behavior; in this case, installing an
appropriate package will solve the problem. Hence, this check will be
kept in a makefile, but somewhere else than the top Makefile.

Move the check to scripts/Makefile.toolcheck so that it is run after
syncconfig.

Another behavioral change is, missing libelf for CONFIG_STACK_VALIDATION
was previously a warning, but now a error.

[1] https://patchwork.kernel.org/patch/10516049/

Reported-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>

---

 Makefile                   | 21 +++------------------
 scripts/Makefile.build     |  2 --
 scripts/Makefile.toolcheck |  5 +++++
 3 files changed, 8 insertions(+), 20 deletions(-)

-- 
2.7.4

Comments

Josh Poimboeuf Oct. 16, 2018, 2:25 p.m. UTC | #1
On Tue, Oct 16, 2018 at 06:10:52PM +0900, Masahiro Yamada wrote:
> Another behavioral change is, missing libelf for CONFIG_STACK_VALIDATION

> was previously a warning, but now a error.


This behavioral change should be fine.  It was already an error for
UNWINDER_ORC, so this would only upgrade a warning to an error for
people using STACK_VALIDATION without ORC, which should be a small
number of people by this point.

> diff --git a/scripts/Makefile.toolcheck b/scripts/Makefile.toolcheck

> index f3c165d..bc26fc0 100644

> --- a/scripts/Makefile.toolcheck

> +++ b/scripts/Makefile.toolcheck

> @@ -12,6 +12,11 @@ include include/config/auto.conf

>  __toolcheck:

>  	@:

>  

> +chk_stack_validation = echo "int main() {}" | $(HOSTCC) -xc -o /dev/null -lelf -

> +msg_stack_validation = "libelf is necessary for building the objtool." \

> +		       "Please install libelf-dev, libelf-devel or elfutils-libelf-devel."

> +toolcheck-$(CONFIG_STACK_VALIDATION) += stack_validation

> +

>  PHONY += $(toolcheck-y)

>  __toolcheck: $(toolcheck-y)


This is a nice improvement.

It would probably be a good idea to clarify to the user which config
option(s) are the cause for the error, by putting
"CONFIG_STACK_VALIDATION" in the error string, for example.

Though, for this particular case, it would be clearer to have a
different error, based on which option is enabled, like we had before.

Like:


ifdef CONFIG_UNWINDER_ORC

chk_unwinder_orc = echo "int main() {}" | $(HOSTCC) -xc -o /dev/null -lelf -
msg_unwinder_orc = "Cannot build objtool to generate ORC metadata for CONFIG_UNWINDER_ORC=y.  " \
		       "Please install libelf-dev, libelf-devel or elfutils-libelf-devel."
toolcheck-$(CONFIG_UNWINDER_ORC) += unwinder_orc

else

chk_stack_validation = echo "int main() {}" | $(HOSTCC) -xc -o /dev/null -lelf -
msg_stack_validation = "Cannot build objtool for CONFIG_STACK_VALIDATION=y.  " \
		       "Please install libelf-dev, libelf-devel or elfutils-libelf-devel."
toolcheck-$(CONFIG_STACK_VALIDATION) += stack_validation

endif


What do you think?

-- 
Josh
Masahiro Yamada Oct. 16, 2018, 3:51 p.m. UTC | #2
On Tue, Oct 16, 2018 at 11:25 PM Josh Poimboeuf <jpoimboe@redhat.com> wrote:
>

> On Tue, Oct 16, 2018 at 06:10:52PM +0900, Masahiro Yamada wrote:

> > Another behavioral change is, missing libelf for CONFIG_STACK_VALIDATION

> > was previously a warning, but now a error.

>

> This behavioral change should be fine.  It was already an error for

> UNWINDER_ORC, so this would only upgrade a warning to an error for

> people using STACK_VALIDATION without ORC, which should be a small

> number of people by this point.

>

> > diff --git a/scripts/Makefile.toolcheck b/scripts/Makefile.toolcheck

> > index f3c165d..bc26fc0 100644

> > --- a/scripts/Makefile.toolcheck

> > +++ b/scripts/Makefile.toolcheck

> > @@ -12,6 +12,11 @@ include include/config/auto.conf

> >  __toolcheck:

> >       @:

> >

> > +chk_stack_validation = echo "int main() {}" | $(HOSTCC) -xc -o /dev/null -lelf -

> > +msg_stack_validation = "libelf is necessary for building the objtool." \

> > +                    "Please install libelf-dev, libelf-devel or elfutils-libelf-devel."

> > +toolcheck-$(CONFIG_STACK_VALIDATION) += stack_validation

> > +

> >  PHONY += $(toolcheck-y)

> >  __toolcheck: $(toolcheck-y)

>

> This is a nice improvement.

>

> It would probably be a good idea to clarify to the user which config

> option(s) are the cause for the error, by putting

> "CONFIG_STACK_VALIDATION" in the error string, for example.

>

> Though, for this particular case, it would be clearer to have a

> different error, based on which option is enabled, like we had before.

>

> Like:

>

>

> ifdef CONFIG_UNWINDER_ORC

>

> chk_unwinder_orc = echo "int main() {}" | $(HOSTCC) -xc -o /dev/null -lelf -

> msg_unwinder_orc = "Cannot build objtool to generate ORC metadata for CONFIG_UNWINDER_ORC=y.  " \

>                        "Please install libelf-dev, libelf-devel or elfutils-libelf-devel."

> toolcheck-$(CONFIG_UNWINDER_ORC) += unwinder_orc

>

> else

>

> chk_stack_validation = echo "int main() {}" | $(HOSTCC) -xc -o /dev/null -lelf -

> msg_stack_validation = "Cannot build objtool for CONFIG_STACK_VALIDATION=y.  " \

>                        "Please install libelf-dev, libelf-devel or elfutils-libelf-devel."

> toolcheck-$(CONFIG_STACK_VALIDATION) += stack_validation

>

> endif

>

>

> What do you think?



It is ugly.

Do you need such detailed information like ORC metadata stuff here?

This Makefile aims to error out, showing why the build failed.
That's it.



-- 
Best Regards
Masahiro Yamada
Josh Poimboeuf Oct. 16, 2018, 4:15 p.m. UTC | #3
On Wed, Oct 17, 2018 at 12:51:40AM +0900, Masahiro Yamada wrote:
> > ifdef CONFIG_UNWINDER_ORC

> >

> > chk_unwinder_orc = echo "int main() {}" | $(HOSTCC) -xc -o /dev/null -lelf -

> > msg_unwinder_orc = "Cannot build objtool to generate ORC metadata for CONFIG_UNWINDER_ORC=y.  " \

> >                        "Please install libelf-dev, libelf-devel or elfutils-libelf-devel."

> > toolcheck-$(CONFIG_UNWINDER_ORC) += unwinder_orc

> >

> > else

> >

> > chk_stack_validation = echo "int main() {}" | $(HOSTCC) -xc -o /dev/null -lelf -

> > msg_stack_validation = "Cannot build objtool for CONFIG_STACK_VALIDATION=y.  " \

> >                        "Please install libelf-dev, libelf-devel or elfutils-libelf-devel."

> > toolcheck-$(CONFIG_STACK_VALIDATION) += stack_validation

> >

> > endif

> >

> >

> > What do you think?

> 

> 

> It is ugly.

> 

> Do you need such detailed information like ORC metadata stuff here?

> 

> This Makefile aims to error out, showing why the build failed.

> That's it.


Yeah, it is kind of ugly.  But the "showing why the build failed" part
is important.  I was trying to give the user a clear error message,
similar to what we have today.

Without context, the user won't know what objtool is, or why it needs to
be built.

If we have just a single error message for all cases, it should at least
mention the config option.  Like

   "Cannot build objtool for CONFIG_STACK_VALIDATION."

But then, most users will only have that enabled because of ORC.  So an
ORC-specific message would be more appropriate in most cases.

So maybe it can just be something more vague:

msg_stack_validation = "Cannot build objtool for CONFIG_UNWINDER_ORC and/or CONFIG_STACK_VALIDATION.  " \
		       "Please install libelf-dev, libelf-devel or elfutils-libelf-devel."

That would probably be good enough.  Then we could drop the ugly ifdef.

-- 
Josh
Masahiro Yamada Oct. 19, 2018, 6:04 a.m. UTC | #4
Hi Josh,


On Wed, Oct 17, 2018 at 1:16 AM Josh Poimboeuf <jpoimboe@redhat.com> wrote:
>

> On Wed, Oct 17, 2018 at 12:51:40AM +0900, Masahiro Yamada wrote:

> > > ifdef CONFIG_UNWINDER_ORC

> > >

> > > chk_unwinder_orc = echo "int main() {}" | $(HOSTCC) -xc -o /dev/null -lelf -

> > > msg_unwinder_orc = "Cannot build objtool to generate ORC metadata for CONFIG_UNWINDER_ORC=y.  " \

> > >                        "Please install libelf-dev, libelf-devel or elfutils-libelf-devel."

> > > toolcheck-$(CONFIG_UNWINDER_ORC) += unwinder_orc

> > >

> > > else

> > >

> > > chk_stack_validation = echo "int main() {}" | $(HOSTCC) -xc -o /dev/null -lelf -

> > > msg_stack_validation = "Cannot build objtool for CONFIG_STACK_VALIDATION=y.  " \

> > >                        "Please install libelf-dev, libelf-devel or elfutils-libelf-devel."

> > > toolcheck-$(CONFIG_STACK_VALIDATION) += stack_validation

> > >

> > > endif

> > >

> > >

> > > What do you think?

> >

> >

> > It is ugly.

> >

> > Do you need such detailed information like ORC metadata stuff here?

> >

> > This Makefile aims to error out, showing why the build failed.

> > That's it.

>

> Yeah, it is kind of ugly.  But the "showing why the build failed" part

> is important.  I was trying to give the user a clear error message,

> similar to what we have today.

>

> Without context, the user won't know what objtool is, or why it needs to

> be built.

>

> If we have just a single error message for all cases, it should at least

> mention the config option.  Like

>

>    "Cannot build objtool for CONFIG_STACK_VALIDATION."

>

> But then, most users will only have that enabled because of ORC.  So an

> ORC-specific message would be more appropriate in most cases.

>

> So maybe it can just be something more vague:

>

> msg_stack_validation = "Cannot build objtool for CONFIG_UNWINDER_ORC and/or CONFIG_STACK_VALIDATION.  " \

>                        "Please install libelf-dev, libelf-devel or elfutils-libelf-devel."

>

> That would probably be good enough.  Then we could drop the ugly ifdef.



Fair point, but I am confused by the current
STACK_VALIDATION / UNWINDER_ORC logic.

In my understanding, objtool is
an all-in-one object check/manipulation tool.

STACK_VALIDATION and UNWINDER_ORC
is a selection of a sub-command, 'check' or 'orc generate'.

(Correct me if am wrong.)


However, STACK_VALIDATION is still used to
decide whether or not to compile the objtool.


Adding a new symbol OBJTOOL would clarify the logic.



config OBJTOOL
        bool

config STACK_VALIDATION
        bool "Compile-time stack metadata validation"
        depends on HAVE_STACK_VALIDATION
        select OBJTOOL
        ...


config UNWINDER_ORC
        bool "ORC unwinder"
        depends on X86_64
        select OBJTOOL
        ...




-- 
Best Regards
Masahiro Yamada
Josh Poimboeuf Oct. 19, 2018, 7:28 p.m. UTC | #5
On Fri, Oct 19, 2018 at 03:04:22PM +0900, Masahiro Yamada wrote:
> Hi Josh,

> 

> 

> On Wed, Oct 17, 2018 at 1:16 AM Josh Poimboeuf <jpoimboe@redhat.com> wrote:

> >

> > On Wed, Oct 17, 2018 at 12:51:40AM +0900, Masahiro Yamada wrote:

> > > > ifdef CONFIG_UNWINDER_ORC

> > > >

> > > > chk_unwinder_orc = echo "int main() {}" | $(HOSTCC) -xc -o /dev/null -lelf -

> > > > msg_unwinder_orc = "Cannot build objtool to generate ORC metadata for CONFIG_UNWINDER_ORC=y.  " \

> > > >                        "Please install libelf-dev, libelf-devel or elfutils-libelf-devel."

> > > > toolcheck-$(CONFIG_UNWINDER_ORC) += unwinder_orc

> > > >

> > > > else

> > > >

> > > > chk_stack_validation = echo "int main() {}" | $(HOSTCC) -xc -o /dev/null -lelf -

> > > > msg_stack_validation = "Cannot build objtool for CONFIG_STACK_VALIDATION=y.  " \

> > > >                        "Please install libelf-dev, libelf-devel or elfutils-libelf-devel."

> > > > toolcheck-$(CONFIG_STACK_VALIDATION) += stack_validation

> > > >

> > > > endif

> > > >

> > > >

> > > > What do you think?

> > >

> > >

> > > It is ugly.

> > >

> > > Do you need such detailed information like ORC metadata stuff here?

> > >

> > > This Makefile aims to error out, showing why the build failed.

> > > That's it.

> >

> > Yeah, it is kind of ugly.  But the "showing why the build failed" part

> > is important.  I was trying to give the user a clear error message,

> > similar to what we have today.

> >

> > Without context, the user won't know what objtool is, or why it needs to

> > be built.

> >

> > If we have just a single error message for all cases, it should at least

> > mention the config option.  Like

> >

> >    "Cannot build objtool for CONFIG_STACK_VALIDATION."

> >

> > But then, most users will only have that enabled because of ORC.  So an

> > ORC-specific message would be more appropriate in most cases.

> >

> > So maybe it can just be something more vague:

> >

> > msg_stack_validation = "Cannot build objtool for CONFIG_UNWINDER_ORC and/or CONFIG_STACK_VALIDATION.  " \

> >                        "Please install libelf-dev, libelf-devel or elfutils-libelf-devel."

> >

> > That would probably be good enough.  Then we could drop the ugly ifdef.

> 

> 

> Fair point, but I am confused by the current

> STACK_VALIDATION / UNWINDER_ORC logic.

> 

> In my understanding, objtool is

> an all-in-one object check/manipulation tool.

> 

> STACK_VALIDATION and UNWINDER_ORC

> is a selection of a sub-command, 'check' or 'orc generate'.

> 

> (Correct me if am wrong.)

> 

> 

> However, STACK_VALIDATION is still used to

> decide whether or not to compile the objtool.

> 

> 

> Adding a new symbol OBJTOOL would clarify the logic.

> 

> 

> 

> config OBJTOOL

>         bool

> 

> config STACK_VALIDATION

>         bool "Compile-time stack metadata validation"

>         depends on HAVE_STACK_VALIDATION

>         select OBJTOOL

>         ...

> 

> 

> config UNWINDER_ORC

>         bool "ORC unwinder"

>         depends on X86_64

>         select OBJTOOL

>         ...


While 'orc generate' and 'check' are indeed separate subcommands of
objtool, the functionality of 'orc generate' is actually a superset of
the functionality of 'check'.  In other words, ORC generation relies on
the stack validation feature, which is consistent with the current
config logic.

-- 
Josh
diff mbox series

Patch

diff --git a/Makefile b/Makefile
index 23a204a..71940b7 100644
--- a/Makefile
+++ b/Makefile
@@ -949,23 +949,6 @@  mod_sign_cmd = true
 endif
 export mod_sign_cmd
 
-ifdef CONFIG_STACK_VALIDATION
-  has_libelf := $(call try-run,\
-		echo "int main() {}" | $(HOSTCC) -xc -o /dev/null -lelf -,1,0)
-  ifeq ($(has_libelf),1)
-    objtool_target := tools/objtool FORCE
-  else
-    ifdef CONFIG_UNWINDER_ORC
-      $(error "Cannot generate ORC metadata for CONFIG_UNWINDER_ORC=y, please install libelf-dev, libelf-devel or elfutils-libelf-devel")
-    else
-      $(warning "Cannot use CONFIG_STACK_VALIDATION=y, please install libelf-dev, libelf-devel or elfutils-libelf-devel")
-    endif
-    SKIP_STACK_VALIDATION := 1
-    export SKIP_STACK_VALIDATION
-  endif
-endif
-
-
 ifeq ($(KBUILD_EXTMOD),)
 core-y		+= kernel/ certs/ mm/ fs/ ipc/ security/ crypto/ block/
 
@@ -1115,7 +1098,9 @@  uapi-asm-generic:
 	            src=uapi/asm obj=arch/$(SRCARCH)/include/generated/uapi/asm
 
 PHONY += prepare-objtool
-prepare-objtool: $(objtool_target)
+ifdef CONFIG_STACK_VALIDATION
+prepare-objtool: tools/objtool
+endif
 
 # Generate some files
 # ---------------------------------------------------------------------------
diff --git a/scripts/Makefile.build b/scripts/Makefile.build
index 54da4b0..e9dabe4 100644
--- a/scripts/Makefile.build
+++ b/scripts/Makefile.build
@@ -233,7 +233,6 @@  endif # CC_USING_RECORD_MCOUNT
 endif # CONFIG_FTRACE_MCOUNT_RECORD
 
 ifdef CONFIG_STACK_VALIDATION
-ifneq ($(SKIP_STACK_VALIDATION),1)
 
 __objtool_obj := $(objtree)/tools/objtool/objtool
 
@@ -270,7 +269,6 @@  objtool_obj = $(if $(patsubst y%,, \
 	$(OBJECT_FILES_NON_STANDARD_$(basetarget).o)$(OBJECT_FILES_NON_STANDARD)n), \
 	$(__objtool_obj))
 
-endif # SKIP_STACK_VALIDATION
 endif # CONFIG_STACK_VALIDATION
 
 # Rebuild all objects when objtool changes, or is enabled/disabled.
diff --git a/scripts/Makefile.toolcheck b/scripts/Makefile.toolcheck
index f3c165d..bc26fc0 100644
--- a/scripts/Makefile.toolcheck
+++ b/scripts/Makefile.toolcheck
@@ -12,6 +12,11 @@  include include/config/auto.conf
 __toolcheck:
 	@:
 
+chk_stack_validation = echo "int main() {}" | $(HOSTCC) -xc -o /dev/null -lelf -
+msg_stack_validation = "libelf is necessary for building the objtool." \
+		       "Please install libelf-dev, libelf-devel or elfutils-libelf-devel."
+toolcheck-$(CONFIG_STACK_VALIDATION) += stack_validation
+
 PHONY += $(toolcheck-y)
 __toolcheck: $(toolcheck-y)