This will help prevent the CI loop from having build failures when
`checkpolicy` isn't available, when doing "randconfig" jobs.

Signed-off-by: Anthony PERARD <anthony.perard@citrix.com>
---
 Config.mk          | 6 ------
 xen/Makefile       | 2 ++
 xen/common/Kconfig | 7 +++++--
 3 files changed, 7 insertions(+), 8 deletions(-)

diff --git a/Config.mk b/Config.mk
index d08fa8d60dd7..97d3633706b3 100644
--- a/Config.mk
+++ b/Config.mk
@@ -137,12 +137,6 @@ export XEN_HAS_BUILD_ID=y
 build_id_linker := --build-id=sha1
 endif
 
-ifndef XEN_HAS_CHECKPOLICY
-    CHECKPOLICY ?= checkpolicy
-    XEN_HAS_CHECKPOLICY := $(shell $(CHECKPOLICY) -h 2>&1 | grep -q xen && echo y || echo n)
-    export XEN_HAS_CHECKPOLICY
-endif
-
 define buildmakevars2shellvars
     export PREFIX="$(prefix)";                                            \
     export XEN_SCRIPT_DIR="$(XEN_SCRIPT_DIR)";                            \
diff --git a/xen/Makefile b/xen/Makefile
index 8023680ffbf2..a60e49903d0c 100644
--- a/xen/Makefile
+++ b/xen/Makefile
@@ -17,6 +17,8 @@ export XEN_BUILD_HOST	?= $(shell hostname)
 PYTHON_INTERPRETER	:= $(word 1,$(shell which python3 python python2 2>/dev/null) python)
 export PYTHON		?= $(PYTHON_INTERPRETER)
 
+export CHECKPOLICY	?= checkpolicy
+
 export BASEDIR := $(CURDIR)
 export XEN_ROOT := $(BASEDIR)/..
 
diff --git a/xen/common/Kconfig b/xen/common/Kconfig
index 0ddd18e11af3..13537e460b8f 100644
--- a/xen/common/Kconfig
+++ b/xen/common/Kconfig
@@ -25,6 +25,9 @@ config GRANT_TABLE
 config HAS_ALTERNATIVE
 	bool
 
+config HAS_CHECKPOLICY
+	def_bool $(success,$(CHECKPOLICY) -h 2>&1 | grep -q xen)
+
 config HAS_COMPAT
 	bool
 
@@ -235,8 +238,8 @@ config XSM_FLASK_AVC_STATS
 
 config XSM_FLASK_POLICY
 	bool "Compile Xen with a built-in FLASK security policy"
-	default y if "$(XEN_HAS_CHECKPOLICY)" = "y"
-	depends on XSM_FLASK
+	default y
+	depends on XSM_FLASK && HAS_CHECKPOLICY
 	---help---
 	  This includes a default XSM policy in the hypervisor so that the
 	  bootloader does not need to load a policy to get sane behavior from an
-- 
Anthony PERARD

On Wed, Jul 14, 2021 at 12:23 PM Anthony PERARD
<anthony.perard@citrix.com> wrote:
>
> This will help prevent the CI loop from having build failures when
> `checkpolicy` isn't available, when doing "randconfig" jobs.
>
> Signed-off-by: Anthony PERARD <anthony.perard@citrix.com>

Reviewed-by: Jason Andryuk <jandryuk@gmail.com>

On 14/07/2021 17:17, Anthony PERARD wrote:
> This will help prevent the CI loop from having build failures when
> `checkpolicy` isn't available, when doing "randconfig" jobs.
>
> Signed-off-by: Anthony PERARD <anthony.perard@citrix.com>

Thankyou for doing this.

Acked-by: Andrew Cooper <andrew.cooper3@citrix.com>

On 14.07.2021 18:17, Anthony PERARD wrote:
> --- a/xen/common/Kconfig
> +++ b/xen/common/Kconfig
> @@ -25,6 +25,9 @@ config GRANT_TABLE
>  config HAS_ALTERNATIVE
>  	bool
>  
> +config HAS_CHECKPOLICY
> +	def_bool $(success,$(CHECKPOLICY) -h 2>&1 | grep -q xen)
> +

This is no different from other aspects of "Kconfig vs tool chain
capabilities" sent out last August to start a discussion about
whether we really want such. Besides Jürgen no-one cared to reply
iirc, which to me means no-one really cares one way or the other.
Which I didn't think was the case ... So here we are again, with
all the same questions still open.

I'm not going to nack the patch, because there's an immediate
purpose / need, but I also can't avoid commenting (and I won't
put my name on it in any positive way, i.e. also not as a
committer; if anything then to record my reservations).

Independent of this I'd like to raise the question of whether
the chosen placement is optimal. Other capability checks live
in xen/Kconfig.

Jan

On Thu, Jul 15, 2021 at 08:25:31AM +0200, Jan Beulich wrote:
> On 14.07.2021 18:17, Anthony PERARD wrote:
> > --- a/xen/common/Kconfig
> > +++ b/xen/common/Kconfig
> > @@ -25,6 +25,9 @@ config GRANT_TABLE
> >  config HAS_ALTERNATIVE
> >  	bool
> >  
> > +config HAS_CHECKPOLICY
> > +	def_bool $(success,$(CHECKPOLICY) -h 2>&1 | grep -q xen)
> > +
> 
> This is no different from other aspects of "Kconfig vs tool chain
> capabilities" sent out last August to start a discussion about
> whether we really want such. Besides Jürgen no-one cared to reply
> iirc, which to me means no-one really cares one way or the other.
> Which I didn't think was the case ... So here we are again, with
> all the same questions still open.

It's true, I don't really care either way. But with maybe a slight
preference for testing the environment every time `make` is run. But
there weren't really a precedent for testing in Makefile and using the
result in Kconfig (or I don't think there is).

> I'm not going to nack the patch, because there's an immediate
> purpose / need, but I also can't avoid commenting (and I won't
> put my name on it in any positive way, i.e. also not as a
> committer; if anything then to record my reservations).

I've prepared an update which test in Makefile, which I hope you'll like
better.

Thanks,

-- 
Anthony PERARD

On 15/07/2021 07:25, Jan Beulich wrote:
> On 14.07.2021 18:17, Anthony PERARD wrote:
>> --- a/xen/common/Kconfig
>> +++ b/xen/common/Kconfig
>> @@ -25,6 +25,9 @@ config GRANT_TABLE
>>  config HAS_ALTERNATIVE
>>  	bool
>>  
>> +config HAS_CHECKPOLICY
>> +	def_bool $(success,$(CHECKPOLICY) -h 2>&1 | grep -q xen)
>> +
> This is no different from other aspects of "Kconfig vs tool chain
> capabilities" sent out last August to start a discussion about
> whether we really want such. Besides Jürgen no-one cared to reply
> iirc, which to me means no-one really cares one way or the other.

You know full well that upgrading Kconfig was specifically to be able to
use this functionality, and you know full well that I firmly support
using this mechanism, because we've had both of these arguments several
times before.

The absence of replies doesn't mean people agree with you, or even that
they don't care.  It either means people didn't read the email, or
didn't have time to reply, or didn't feel like wasting time rehashing
the same arguments yet again with no hope for progress.


If you really insist on refusing to features specifically intended for
the improvement of our build processes, then call a vote so we can be
done with the argument for once and for all.

~Andrew

On 16.07.2021 15:15, Andrew Cooper wrote:
> On 15/07/2021 07:25, Jan Beulich wrote:
>> On 14.07.2021 18:17, Anthony PERARD wrote:
>>> --- a/xen/common/Kconfig
>>> +++ b/xen/common/Kconfig
>>> @@ -25,6 +25,9 @@ config GRANT_TABLE
>>>  config HAS_ALTERNATIVE
>>>  	bool
>>>  
>>> +config HAS_CHECKPOLICY
>>> +	def_bool $(success,$(CHECKPOLICY) -h 2>&1 | grep -q xen)
>>> +
>> This is no different from other aspects of "Kconfig vs tool chain
>> capabilities" sent out last August to start a discussion about
>> whether we really want such. Besides Jürgen no-one cared to reply
>> iirc, which to me means no-one really cares one way or the other.
> 
> You know full well that upgrading Kconfig was specifically to be able to
> use this functionality, and you know full well that I firmly support
> using this mechanism, because we've had both of these arguments several
> times before.
> 
> The absence of replies doesn't mean people agree with you, or even that
> they don't care.  It either means people didn't read the email, or
> didn't have time to reply, or didn't feel like wasting time rehashing
> the same arguments yet again with no hope for progress.
> 
> 
> If you really insist on refusing to features specifically intended for
> the improvement of our build processes, then call a vote so we can be
> done with the argument for once and for all.

I'm sorry Andrew, but this is not a way to discuss controversial items.
Back at the time you were pointing me at a discussion at a summit that
I didn't recall and hence presumably didn't attend for whatever reason.
Whatever may have been the result of such a discussion imo _has_ to be
under the precondition that no other contrary arguments arise. I do not
recall there having been spelled out up front this specific purpose of
upgrading kconfig, and even if it was spelled out, the ramifications
may not have become clear until the actual first uses were proposed.
It has to be possible to change views at such a point even for people
who did signal agreement earlier on. Not to speak of unaware ones.

Further iirc it was you who told me to start a mail thread about the
issue. Which I did. And now you say "... didn't feel like wasting time
rehashing the same arguments yet again with no hope for progress"? Can
you please point me to a place where those "same arguments" are put
down in writing, including reasons why they were either turned down or
considered of less relevance?

I can't help the feeling that when our opinions don't match you try to
silence me by whatever means you find suitable - ignoring my input,
claiming my earlier agreement, denying me any influence, etc. I'm
afraid I have to say that I don't think this is a way to run a
community project. The only two ways to get past my objections (or
really just reservations here) are to either convince me (which you
don't appear to be willing to) or to outvote me (which you haven't
tried at all so far if I'm not mistaken).

Jan

On Fri, Jul 16, 2021 at 02:15:28PM +0100, Andrew Cooper wrote:
> On 15/07/2021 07:25, Jan Beulich wrote:
> > On 14.07.2021 18:17, Anthony PERARD wrote:
> >> --- a/xen/common/Kconfig
> >> +++ b/xen/common/Kconfig
> >> @@ -25,6 +25,9 @@ config GRANT_TABLE
> >>  config HAS_ALTERNATIVE
> >>  	bool
> >>  
> >> +config HAS_CHECKPOLICY
> >> +	def_bool $(success,$(CHECKPOLICY) -h 2>&1 | grep -q xen)
> >> +
> > This is no different from other aspects of "Kconfig vs tool chain
> > capabilities" sent out last August to start a discussion about
> > whether we really want such. Besides Jürgen no-one cared to reply
> > iirc, which to me means no-one really cares one way or the other.
> 
> You know full well that upgrading Kconfig was specifically to be able to
> use this functionality, and you know full well that I firmly support

To be honest, I've upgraded Kconfig mostly because I needed to start
somewhere with upgrading our build system to look more like Kbuild. I
may have adding `CC --version` and some other CONFIG_* running shells,
but I don't think anymore that is a necessary things to do, it just look
cleaner.

> using this mechanism, because we've had both of these arguments several
> times before.

I'd like to read about your (or someone else's) arguments in favor of
doing more in Kconfig only, do you have links (or maybe subject,
keyword) to look at?

I think I understand Jan's arguments.

Cheers,

-- 
Anthony PERARD

> On Jul 14, 2021, at 5:17 PM, Anthony PERARD <anthony.perard@citrix.com> wrote:
> 
> This will help prevent the CI loop from having build failures when
> `checkpolicy` isn't available, when doing "randconfig" jobs.

Hang on, just to clarify what’s going on here.

‘randconfig’ is setting CONFIG_XSM_FLASK_POLICY in the .config file; and then when the build happens, we error out because one of the required components isn’t there.

What this patch does is to make it so that if someone explicitly sets CONFIG_XSM_FLASK_POLICY=y, but doesn’t have checkpolicy, the build system will silently disable the policy behind their backs without telling them?

Or does the randconfig test run kConfig one more time, at which point *then* the .config will be disabled?

The former I think is broken; the latter I think is fine.

 -George

On 16.07.21 17:26, George Dunlap wrote:
> 
> 
>> On Jul 14, 2021, at 5:17 PM, Anthony PERARD <anthony.perard@citrix.com> wrote:
>>
>> This will help prevent the CI loop from having build failures when
>> `checkpolicy` isn't available, when doing "randconfig" jobs.
> 
> Hang on, just to clarify what’s going on here.
> 
> ‘randconfig’ is setting CONFIG_XSM_FLASK_POLICY in the .config file; and then when the build happens, we error out because one of the required components isn’t there.
> 
> What this patch does is to make it so that if someone explicitly sets CONFIG_XSM_FLASK_POLICY=y, but doesn’t have checkpolicy, the build system will silently disable the policy behind their backs without telling them?
> 
> Or does the randconfig test run kConfig one more time, at which point *then* the .config will be disabled?
> 
> The former I think is broken; the latter I think is fine.

I still think that Kconfig should not disable explicit settings due to
tools not being available.

I'd be fine with trimodal settings for such parameters:

- Off
- On
- On if tools available

And the last one could even be the default.

But anyone wanting to test a specific option or to enable an option
should not be catched by surprise because some internal modification is
requesting another tool which happens not to be available.

BTW, the same applies to ./configure options for tools. I really don't
like that some components are silently not built because e.g. a header
file is not available. I'd rather have the possibility to tell
./configure that I want everything built and let the build fail if that
is not possible.


Juergen

On Fri, Jul 16, 2021 at 04:26:39PM +0100, George Dunlap wrote:
> 
> 
> > On Jul 14, 2021, at 5:17 PM, Anthony PERARD <anthony.perard@citrix.com> wrote:
> > 
> > This will help prevent the CI loop from having build failures when
> > `checkpolicy` isn't available, when doing "randconfig" jobs.
> 
> Hang on, just to clarify what’s going on here.
> 
> ‘randconfig’ is setting CONFIG_XSM_FLASK_POLICY in the .config file; and then when the build happens, we error out because one of the required components isn’t there.
> 
> What this patch does is to make it so that if someone explicitly sets CONFIG_XSM_FLASK_POLICY=y, but doesn’t have checkpolicy, the build system will silently disable the policy behind their backs without telling them?
> 
> Or does the randconfig test run kConfig one more time, at which point *then* the .config will be disabled?
> 
> The former I think is broken; the latter I think is fine.

That's right, Xen's build system is broken.

Kconfig doesn't say whether a given .config is correct or not, it
removes non existing CONFIG_* setting as well as those that are missing
dependencies. This isn't new, this is the default, this is how Linux is
using Kconfig.

But there's a way out of that.
There's an option to prevent Kconfig from updating .config, setting
KCONFIG_NOSILENTUPDATE in the environment (see docs/misc/kconfig.rst).
But that won't work as expected with the Xen build system because to
update the config via "syncconfig" doesn't prevent the build system from
building Xen (and probably fail later).
If it were working, build would fail, and user would have to run
"oldconfig" or other *config targets, and check the result (diff
.config.old .config).

Cheers,

-- 
Anthony PERARD

On 16/07/2021 16:26, George Dunlap wrote:
>
>> On Jul 14, 2021, at 5:17 PM, Anthony PERARD <anthony.perard@citrix.com> wrote:
>>
>> This will help prevent the CI loop from having build failures when
>> `checkpolicy` isn't available, when doing "randconfig" jobs.
> Hang on, just to clarify what’s going on here.
>
> ‘randconfig’ is setting CONFIG_XSM_FLASK_POLICY in the .config file; and then when the build happens, we error out because one of the required components isn’t there.
>
> What this patch does is to make it so that if someone explicitly sets CONFIG_XSM_FLASK_POLICY=y, but doesn’t have checkpolicy, the build system will silently disable the policy behind their backs without telling them?

Yes, but that's how ~everything in the Xen and Linux build works currently.

What this new version will do is produce a config/build combo, with the
config reporting that CONFIG_XSM_FLASK_POLICY was not active.

This is a damnsignt better than the "old" way of doing feature checks in
the makefiles, where there is no trace that the build system disabled a
feature because your compiler was too old.

~Andrew

On 16.07.2021 18:14, Andrew Cooper wrote:
> On 16/07/2021 16:26, George Dunlap wrote:
>>
>>> On Jul 14, 2021, at 5:17 PM, Anthony PERARD <anthony.perard@citrix.com> wrote:
>>>
>>> This will help prevent the CI loop from having build failures when
>>> `checkpolicy` isn't available, when doing "randconfig" jobs.
>> Hang on, just to clarify what’s going on here.
>>
>> ‘randconfig’ is setting CONFIG_XSM_FLASK_POLICY in the .config file; and then when the build happens, we error out because one of the required components isn’t there.
>>
>> What this patch does is to make it so that if someone explicitly sets CONFIG_XSM_FLASK_POLICY=y, but doesn’t have checkpolicy, the build system will silently disable the policy behind their backs without telling them?
> 
> Yes, but that's how ~everything in the Xen and Linux build works currently.
> 
> What this new version will do is produce a config/build combo, with the
> config reporting that CONFIG_XSM_FLASK_POLICY was not active.
> 
> This is a damnsignt better than the "old" way of doing feature checks in
> the makefiles, where there is no trace that the build system disabled a
> feature because your compiler was too old.

Disabling features at build time is of course no better than disabling
them at configure time. Instead, if a chose configuration cannot be built,
the build should fail - ideally with an error message clearly telling
people what the reason is and what they can do about it.

Jan

On Fri, Jul 16, 2021 at 04:26:39PM +0100, George Dunlap wrote:
> What this patch does is to make it so that if someone explicitly sets CONFIG_XSM_FLASK_POLICY=y, but doesn’t have checkpolicy, the build system will silently disable the policy behind their backs without telling them?

FYI, silenty disabling unavailable config options is actually
documented, in "INSTALL", section "Xen Hypervisor". ;-) Well kind of.

-- 
Anthony PERARD

This will help prevent the CI loop from having build failures when
`checkpolicy` isn't available, when doing "randconfig" jobs.

Also, move the check out of Config.mk and into xen/ build system.
Nothing in tools/ is using that information as it's done by
./configure.

Signed-off-by: Anthony PERARD <anthony.perard@citrix.com>
---
We might want to have a new Makefile for this kind of check that
Kconfig is going to use, just to keep the main Makefile a bit cleaner.
But maybe another time, if more are comming.

v2:
- move check to Makefile
---
 Config.mk          | 6 ------
 xen/Makefile       | 4 ++++
 xen/common/Kconfig | 4 ++--
 3 files changed, 6 insertions(+), 8 deletions(-)

diff --git a/Config.mk b/Config.mk
index d08fa8d60dd7..97d3633706b3 100644
--- a/Config.mk
+++ b/Config.mk
@@ -137,12 +137,6 @@ export XEN_HAS_BUILD_ID=y
 build_id_linker := --build-id=sha1
 endif
 
-ifndef XEN_HAS_CHECKPOLICY
-    CHECKPOLICY ?= checkpolicy
-    XEN_HAS_CHECKPOLICY := $(shell $(CHECKPOLICY) -h 2>&1 | grep -q xen && echo y || echo n)
-    export XEN_HAS_CHECKPOLICY
-endif
-
 define buildmakevars2shellvars
     export PREFIX="$(prefix)";                                            \
     export XEN_SCRIPT_DIR="$(XEN_SCRIPT_DIR)";                            \
diff --git a/xen/Makefile b/xen/Makefile
index 8023680ffbf2..045ddb18ad68 100644
--- a/xen/Makefile
+++ b/xen/Makefile
@@ -17,6 +17,8 @@ export XEN_BUILD_HOST	?= $(shell hostname)
 PYTHON_INTERPRETER	:= $(word 1,$(shell which python3 python python2 2>/dev/null) python)
 export PYTHON		?= $(PYTHON_INTERPRETER)
 
+export CHECKPOLICY	?= checkpolicy
+
 export BASEDIR := $(CURDIR)
 export XEN_ROOT := $(BASEDIR)/..
 
@@ -156,6 +158,8 @@ CFLAGS += $(CLANG_FLAGS)
 export CLANG_FLAGS
 endif
 
+export HAS_CHECKPOLICY := $(call success,$(CHECKPOLICY) -h 2>&1 | grep -q xen)
+
 export root-make-done := y
 endif # root-make-done
 
diff --git a/xen/common/Kconfig b/xen/common/Kconfig
index 0ddd18e11af3..a5ef3814f531 100644
--- a/xen/common/Kconfig
+++ b/xen/common/Kconfig
@@ -235,8 +235,8 @@ config XSM_FLASK_AVC_STATS
 
 config XSM_FLASK_POLICY
 	bool "Compile Xen with a built-in FLASK security policy"
-	default y if "$(XEN_HAS_CHECKPOLICY)" = "y"
-	depends on XSM_FLASK
+	default y
+	depends on XSM_FLASK && "$(HAS_CHECKPOLICY)"
 	---help---
 	  This includes a default XSM policy in the hypervisor so that the
 	  bootloader does not need to load a policy to get sane behavior from an
-- 
Anthony PERARD

On 16.07.2021 14:38, Anthony PERARD wrote:
> This will help prevent the CI loop from having build failures when
> `checkpolicy` isn't available, when doing "randconfig" jobs.
> 
> Also, move the check out of Config.mk and into xen/ build system.
> Nothing in tools/ is using that information as it's done by
> ./configure.
> 
> Signed-off-by: Anthony PERARD <anthony.perard@citrix.com>
> ---
> We might want to have a new Makefile for this kind of check that
> Kconfig is going to use, just to keep the main Makefile a bit cleaner.
> But maybe another time, if more are comming.
> 
> v2:
> - move check to Makefile

I'm afraid I don't understand:

> --- a/xen/Makefile
> +++ b/xen/Makefile
> @@ -17,6 +17,8 @@ export XEN_BUILD_HOST	?= $(shell hostname)
>  PYTHON_INTERPRETER	:= $(word 1,$(shell which python3 python python2 2>/dev/null) python)
>  export PYTHON		?= $(PYTHON_INTERPRETER)
>  
> +export CHECKPOLICY	?= checkpolicy
> +
>  export BASEDIR := $(CURDIR)
>  export XEN_ROOT := $(BASEDIR)/..
>  
> @@ -156,6 +158,8 @@ CFLAGS += $(CLANG_FLAGS)
>  export CLANG_FLAGS
>  endif
>  
> +export HAS_CHECKPOLICY := $(call success,$(CHECKPOLICY) -h 2>&1 | grep -q xen)

While the setting indeed gets obtained in a Makefile now, ...

> --- a/xen/common/Kconfig
> +++ b/xen/common/Kconfig
> @@ -235,8 +235,8 @@ config XSM_FLASK_AVC_STATS
>  
>  config XSM_FLASK_POLICY
>  	bool "Compile Xen with a built-in FLASK security policy"
> -	default y if "$(XEN_HAS_CHECKPOLICY)" = "y"
> -	depends on XSM_FLASK
> +	default y
> +	depends on XSM_FLASK && "$(HAS_CHECKPOLICY)"

... it's still used as a Kconfig dependency. This in particular
does not address George's concern about a setting silently getting
turned off behind the back of the person having enabled it (and
this could happen at any time, not just during the initial build,
where one might still remember to diff .config against
.config.old). The minimal thing imo is some kind of warning then.
Even better would be if the setting was left unchanged and the
build would fail; a solution for randconfig would then still be
needed of course. If we wanted to go this route, a tristate type
for the values may be unavoidable, along the lines of what Jürgen
has suggested. I wouldn't think of a global override though, but
a distinction of the origin of each option's setting. This might
be as simple as y vs Y for "positive" values and "# CONFIG_...
is not set" present (for visible options) or absent (for options
the user can't control) for "negative" ones. But yes, this would
likely be an intrusive change _and_ it would not be clear how to
transform existing .config-s, so is unlikely to be suitable for
the immediate issue at hand.

Jan

On Mon, Jul 19, 2021 at 09:37:06AM +0200, Jan Beulich wrote:
> On 16.07.2021 14:38, Anthony PERARD wrote:
> > +export HAS_CHECKPOLICY := $(call success,$(CHECKPOLICY) -h 2>&1 | grep -q xen)
> 
> While the setting indeed gets obtained in a Makefile now, ...
> 
> > --- a/xen/common/Kconfig
> > +++ b/xen/common/Kconfig
> > @@ -235,8 +235,8 @@ config XSM_FLASK_AVC_STATS
> >  
> >  config XSM_FLASK_POLICY
> >  	bool "Compile Xen with a built-in FLASK security policy"
> > -	default y if "$(XEN_HAS_CHECKPOLICY)" = "y"
> > -	depends on XSM_FLASK
> > +	default y
> > +	depends on XSM_FLASK && "$(HAS_CHECKPOLICY)"
> 
> ... it's still used as a Kconfig dependency. This in particular
> does not address George's concern about a setting silently getting
> turned off behind the back of the person having enabled it (and

This patch v2 wasn't meant to address George's concern which didn't
exist at the time this v2 was sent... I was trying to address yours.

But it seems that "George's concern" is part of your issues with
Kconfig too, which I missed when trying to right this v2.

Anyway, those two patches are the only way I'm going to try to fix the
random build failure in the GitLab CI, I'm not going to try to fix
issues with the use of Kconfig for now. In the mean time either v1 or v2
is committed, or will just keep getting random build failure in the
GitLab CI.

-- 
Anthony PERARD

On 19.07.2021 12:47, Anthony PERARD wrote:
> On Mon, Jul 19, 2021 at 09:37:06AM +0200, Jan Beulich wrote:
>> On 16.07.2021 14:38, Anthony PERARD wrote:
>>> +export HAS_CHECKPOLICY := $(call success,$(CHECKPOLICY) -h 2>&1 | grep -q xen)
>>
>> While the setting indeed gets obtained in a Makefile now, ...
>>
>>> --- a/xen/common/Kconfig
>>> +++ b/xen/common/Kconfig
>>> @@ -235,8 +235,8 @@ config XSM_FLASK_AVC_STATS
>>>  
>>>  config XSM_FLASK_POLICY
>>>  	bool "Compile Xen with a built-in FLASK security policy"
>>> -	default y if "$(XEN_HAS_CHECKPOLICY)" = "y"
>>> -	depends on XSM_FLASK
>>> +	default y
>>> +	depends on XSM_FLASK && "$(HAS_CHECKPOLICY)"
>>
>> ... it's still used as a Kconfig dependency. This in particular
>> does not address George's concern about a setting silently getting
>> turned off behind the back of the person having enabled it (and
> 
> This patch v2 wasn't meant to address George's concern which didn't
> exist at the time this v2 was sent... I was trying to address yours.
> 
> But it seems that "George's concern" is part of your issues with
> Kconfig too, which I missed when trying to right this v2.
> 
> Anyway, those two patches are the only way I'm going to try to fix the
> random build failure in the GitLab CI, I'm not going to try to fix
> issues with the use of Kconfig for now. In the mean time either v1 or v2
> is committed, or will just keep getting random build failure in the
> GitLab CI.

Fair enough. I actually think that randconfig shouldn't act quite as
randomly as it does. But what's sensible as behavior there really
depends heavily on the future intentions with .config. If we follow
Linux'es model (which Andrew advocates for), its randomness would be
limited by options which could get randomly set getting further
altered by environmental conditions. Hence that would limit what can
actually be tested, but it would avoid failures resulting from the
environment not matching the chose settings.

Otoh with our current model (largely, leaving aside the few
environment checks we've already got) what is being asked for is
what is going to get built. But failure from environmental
constraints shouldn't be treated the same as failure from bad
interaction of options; it's (aiui) the latter which randconfig is
supposed to point out.

Jan

On 19 Jul 2021, at 12:04, Jan Beulich <jbeulich@suse.com<mailto:jbeulich@suse.com>> wrote:

On 19.07.2021 12:47, Anthony PERARD wrote:
On Mon, Jul 19, 2021 at 09:37:06AM +0200, Jan Beulich wrote:
On 16.07.2021 14:38, Anthony PERARD wrote:
+export HAS_CHECKPOLICY := $(call success,$(CHECKPOLICY) -h 2>&1 | grep -q xen)

While the setting indeed gets obtained in a Makefile now, ...

--- a/xen/common/Kconfig
+++ b/xen/common/Kconfig
@@ -235,8 +235,8 @@ config XSM_FLASK_AVC_STATS

config XSM_FLASK_POLICY
bool "Compile Xen with a built-in FLASK security policy"
- default y if "$(XEN_HAS_CHECKPOLICY)" = "y"
- depends on XSM_FLASK
+ default y
+ depends on XSM_FLASK && "$(HAS_CHECKPOLICY)"

... it's still used as a Kconfig dependency. This in particular
does not address George's concern about a setting silently getting
turned off behind the back of the person having enabled it (and

This patch v2 wasn't meant to address George's concern which didn't
exist at the time this v2 was sent... I was trying to address yours.

But it seems that "George's concern" is part of your issues with
Kconfig too, which I missed when trying to right this v2.

Anyway, those two patches are the only way I'm going to try to fix the
random build failure in the GitLab CI, I'm not going to try to fix
issues with the use of Kconfig for now. In the mean time either v1 or v2
is committed, or will just keep getting random build failure in the
GitLab CI.

Fair enough. I actually think that randconfig shouldn't act quite as
randomly as it does.

It sounds like randconfig currently works appropriately for how KConfig is meant to work.  That is, for good or for ill, KConfig seems designed to silently discard options with missing dependencies (either internal or external); so generating a config with missing dependencies to hand to it is “sensible”.


But what's sensible as behavior there really
depends heavily on the future intentions with .config. If we follow
Linux'es model (which Andrew advocates for), its randomness would be
limited by options which could get randomly set getting further
altered by environmental conditions. Hence that would limit what can
actually be tested, but it would avoid failures resulting from the
environment not matching the chose settings.

Otoh with our current model (largely, leaving aside the few
environment checks we've already got) what is being asked for is
what is going to get built. But failure from environmental
constraints shouldn't be treated the same as failure from bad
interaction of options; it's (aiui) the latter which randconfig is
supposed to point out.

Right; so another solution to the ‘randconfig CI loop’ issue would be to have the pipeline pass (or some non-pass non-fail state if possible) when then build failed because there were missing dependencies.  Maybe one way of doing that would be to have a “make check-dependencies” target that would run first.

That said, since it’s basically known that “silently disabling things it can’t build" is a quirk of Kconfig, I’m less inclined to object to v1 of the patch.

 -George

On 16/07/2021 13:38, Anthony PERARD wrote:
> This will help prevent the CI loop from having build failures when
> `checkpolicy` isn't available, when doing "randconfig" jobs.
>
> Also, move the check out of Config.mk and into xen/ build system.
> Nothing in tools/ is using that information as it's done by
> ./configure.
>
> Signed-off-by: Anthony PERARD <anthony.perard@citrix.com>
> ---
> We might want to have a new Makefile for this kind of check that
> Kconfig is going to use, just to keep the main Makefile a bit cleaner.
> But maybe another time, if more are comming.

I won't nack this patch, but I'm a very firm -2 against it.

v1 is the correct way to go.

~Andrew