Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Date:   Tue, 18 Feb 2020 17:05:54 +0100
From:   Jessica Yu <jeyu@kernel.org>
To:     Matthias Maennich <maennich@google.com>
Cc:     Masahiro Yamada <yamada.masahiro@socionext.com>,
        linux-kernel@vger.kernel.org, Martijn Coenen <maco@android.com>
Subject: Re: [PATCH] modpost: return error if module is missing ns imports
 and MODULE_ALLOW_MISSING_NAMESPACE_IMPORTS=n
Message-ID: <20200218160553.GA18056@linux-8ccs>
References: <20200214143709.6490-1-jeyu@kernel.org>
 <20200217145644.GA221719@google.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Disposition: inline
In-Reply-To: <20200217145644.GA221719@google.com>
User-Agent: Mutt/1.10.1 (2018-07-13)
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

+++ Matthias Maennich [17/02/20 14:56 +0000]:
>Hi Jessica!
>
>On Fri, Feb 14, 2020 at 03:37:09PM +0100, Jessica Yu wrote:
>>Currently when CONFIG_MODULE_ALLOW_MISSING_NAMESPACE_IMPORTS=n, modpost only warns
>>when a module is missing namespace imports. Under this configuration, such a module
>>cannot be loaded into the kernel anyway, as the module loader would reject it.
>>We might as well return a build error when a module is missing namespace imports
>>under CONFIG_MODULE_ALLOW_MISSING_NAMESPACE_IMPORTS=n, so that the build
>>warning does not go ignored/unnoticed.
>
>I generally agree with the idea of the patch. Thanks for working on
>this! I also can't remember any reason why I did not write it like this
>initially. Probably just because I introduced this configuration option
>quite late in the development process of the initial patches.
>
>>
>>Signed-off-by: Jessica Yu <jeyu@kernel.org>
>>---
>>scripts/Makefile.modpost |  1 +
>>scripts/mod/modpost.c    | 19 +++++++++++++++----
>>2 files changed, 16 insertions(+), 4 deletions(-)
>>
>>diff --git a/scripts/Makefile.modpost b/scripts/Makefile.modpost
>>index b4d3f2d122ac..a53660f910a9 100644
>>--- a/scripts/Makefile.modpost
>>+++ b/scripts/Makefile.modpost
>>@@ -53,6 +53,7 @@ MODPOST = scripts/mod/modpost						\
>>	$(if $(KBUILD_EXTMOD),$(addprefix -e ,$(KBUILD_EXTRA_SYMBOLS)))	\
>>	$(if $(KBUILD_EXTMOD),-o $(modulesymfile))			\
>>	$(if $(CONFIG_SECTION_MISMATCH_WARN_ONLY),,-E)			\
>>+	$(if $(CONFIG_MODULE_ALLOW_MISSING_NAMESPACE_IMPORTS),,-N) 	\
>>	$(if $(KBUILD_MODPOST_WARN),-w)
>>
>>ifdef MODPOST_VMLINUX
>>diff --git a/scripts/mod/modpost.c b/scripts/mod/modpost.c
>>index 7edfdb2f4497..53e966f7d557 100644
>>--- a/scripts/mod/modpost.c
>>+++ b/scripts/mod/modpost.c
>>@@ -39,6 +39,8 @@ static int sec_mismatch_count = 0;
>>static int sec_mismatch_fatal = 0;
>>/* ignore missing files */
>>static int ignore_missing_files;
>>+/* Return an error when there are missing namespace imports */
>>+static int missing_ns_import_error = 0;
>
>A more suitable name is maybe missing_ns_import_is_error or follow the
>naming of the config option: allow_missing_ns_imports (with default = 1).
>
>>
>>enum export {
>>	export_plain,      export_unused,     export_gpl,
>>@@ -2216,9 +2218,15 @@ static int check_exports(struct module *mod)
>>
>>		if (exp->namespace &&
>>		    !module_imports_namespace(mod, exp->namespace)) {
>>-			warn("module %s uses symbol %s from namespace %s, but does not import it.\n",
>>-			     basename, exp->name, exp->namespace);
>>-			add_namespace(&mod->missing_namespaces, exp->namespace);
>>+			if (missing_ns_import_error) {
>>+				merror("module %s uses symbol %s from namespace %s, but does not import it.\n",
>>+					basename, exp->name, exp->namespace);
>
>I would like to avoid the code duplication here. The string literal is
>identical for both cases.

Hm, but one is a call to merror() and the other to warn(). The
previous if (warn_unresolved) block does the same thing. I am not sure
how to simplify it to one call without introducing macro magic or
overcomplicating things. Or were you thinking of something else?

>>+				err = 1;
>
>Also, if we fail here, we might as well help the user to fix it by
>suggesting to run `make nsdeps` (once per failed modpost run). Speaking
>of which, `make nsdeps` is currently broken by this patch as it relies
>on a successful (yet warning-full) build of the modules. So, in case of
>`make nsdeps`, we probably have to omit the -N flag again when invoking
>modpost.

Good catch! Since KBUILD_NSDEPS is set when running `make nsdeps`,
maybe we can do something like:

diff --git a/scripts/Makefile.modpost b/scripts/Makefile.modpost
index a53660f910a9..145703ef8d3a 100644
--- a/scripts/Makefile.modpost
+++ b/scripts/Makefile.modpost
@@ -53,7 +53,7 @@ MODPOST = scripts/mod/modpost                                         \
        $(if $(KBUILD_EXTMOD),$(addprefix -e ,$(KBUILD_EXTRA_SYMBOLS))) \
        $(if $(KBUILD_EXTMOD),-o $(modulesymfile))                      \
        $(if $(CONFIG_SECTION_MISMATCH_WARN_ONLY),,-E)                  \
-       $(if $(CONFIG_MODULE_ALLOW_MISSING_NAMESPACE_IMPORTS),,-N)      \
+       $(if $(CONFIG_MODULE_ALLOW_MISSING_NAMESPACE_IMPORTS),,$(if $(KBUILD_NSDEPS),,-N))      \
        $(if $(KBUILD_MODPOST_WARN),-w)

 ifdef MODPOST_VMLINUX

Thanks for the review!