2019-09-26 00:51:05

by Johannes Berg

Subject: static EXPORT_SYMBOL checker causes false positives on ARCH=um

Hi,

With the new commit 15bfc2348d54 ("modpost: check for static
EXPORT_SYMBOL* functions") we get a lot of warnings on ARCH=um builds:

WARNING: "rename" [vmlinux] is a static EXPORT_SYMBOL
WARNING: "lseek" [vmlinux] is a static EXPORT_SYMBOL
WARNING: "ftruncate64" [vmlinux] is a static EXPORT_SYMBOL
[...]
see https://p.sipsolutions.net/7232995f34907b9d.txt


This hack fixes *most* of them:

diff --git a/scripts/mod/modpost.c b/scripts/mod/modpost.c
index 820eed87fb43..3e443563ebea 100644
--- a/scripts/mod/modpost.c
+++ b/scripts/mod/modpost.c
@@ -1931,12 +1931,18 @@ static void check_sec_ref(struct module *mod, const char *modname,
static char *remove_dot(char *s)
{
size_t n = strcspn(s, ".");
+ char *at;

if (n && s[n]) {
size_t m = strspn(s + n + 1, "0123456789");
if (m && (s[n + m] == '.' || s[n + m] == 0))
s[n] = 0;
}
+
+ at = strchr(s, '@');
+ if (at)
+ *at = 0;
+
return s;
}


(but obviously just serves to give you an idea of what's going on).


With that, only two remain for me:

WARNING: "__guard" [vmlinux] is a static EXPORT_SYMBOL
WARNING: "__stack_smash_handler" [vmlinux] is a static EXPORT_SYMBOL

and I think that's because they don't even exist at all, so arguably the
code shouldn't export them, but I didn't find a way to detect at build
time if -fstack-protector was enabled or not?

Any thoughts?

Thanks,
johannes

2019-09-26 08:02:17

by Masahiro Yamada

Subject: Re: static EXPORT_SYMBOL checker causes false positives on ARCH=um

Hi Johannes,

On Tue, Sep 24, 2019 at 5:17 PM Johannes Berg <[email protected]> wrote:
>
> Hi,
>
> With the new commit 15bfc2348d54 ("modpost: check for static
> EXPORT_SYMBOL* functions") we get a lot of warnings on ARCH=um builds:
>
> WARNING: "rename" [vmlinux] is a static EXPORT_SYMBOL
> WARNING: "lseek" [vmlinux] is a static EXPORT_SYMBOL
> WARNING: "ftruncate64" [vmlinux] is a static EXPORT_SYMBOL
> [...]
> see https://p.sipsolutions.net/7232995f34907b9d.txt

Thanks for the report.

I think this should work:
https://lore.kernel.org/patchwork/patch/1131017/

It fixes "__guard" and "__stack_smash_handler" too.


Thanks.



>
>
> This hack fixes *most* of them:
>
> diff --git a/scripts/mod/modpost.c b/scripts/mod/modpost.c
> index 820eed87fb43..3e443563ebea 100644
> --- a/scripts/mod/modpost.c
> +++ b/scripts/mod/modpost.c
> @@ -1931,12 +1931,18 @@ static void check_sec_ref(struct module *mod, const char *modname,
> static char *remove_dot(char *s)
> {
> size_t n = strcspn(s, ".");
> + char *at;
>
> if (n && s[n]) {
> size_t m = strspn(s + n + 1, "0123456789");
> if (m && (s[n + m] == '.' || s[n + m] == 0))
> s[n] = 0;
> }
> +
> + at = strchr(s, '@');
> + if (at)
> + *at = 0;
> +
> return s;
> }
>
>
> (but obviously just serves to give you an idea of what's going on).
>
>
> With that, only two remain for me:
>
> WARNING: "__guard" [vmlinux] is a static EXPORT_SYMBOL
> WARNING: "__stack_smash_handler" [vmlinux] is a static EXPORT_SYMBOL
>
> and I think that's because they don't even exist at all, so arguably the
> code shouldn't export them, but I didn't find a way to detect at build
> time if -fstack-protector was enabled or not?
>
> Any thoughts?
>
> Thanks,
> johannes
>


--
Best Regards
Masahiro Yamada