[v2] modpost: Don't let "driver"s reference .exit.*

Commit Message

Uwe Kleine-König Sept. 30, 2023, 4:52 p.m. UTC

Drivers must not reference functions marked with __exit as these likely
are not available when the code is built-in.

There are few creative offenders uncovered for example in ARCH=amd64
allmodconfig builds. So only trigger the section mismatch warning for
W=1 builds.

The dual rule that drivers must not reference .init.* is implemented
since commit 0db252452378 ("modpost: don't allow *driver to reference
.init.*") which however missed that .exit.* should be handled in the
same way.

Thanks to Masahiro Yamada and Arnd Bergmann who gave valuable hints to
find this improvement.

Signed-off-by: Uwe Kleine-König <[e-mail]>
---
Hello,

changes since (implicit) v1, sent with Message-Id:
[e-mail]:

 - enable the warning about .data -> .exit.* only in W=1 builds to keep
   normal builds without warnings. *sigh*
 - improved commit log and mention the above item.
 - updated the code comment to match the code

Thanks
Uwe

 scripts/mod/modpost.c | 15 +++++++++++++--
 1 file changed, 13 insertions(+), 2 deletions(-)


base-commit: 6465e260f48790807eef06b583b38ca9789b6072

Comments

Masahiro Yamada Sept. 30, 2023, 8:33 p.m. UTC | #1

On Sun, Oct 1, 2023 at 1:52 AM Uwe Kleine-König
<[e-mail]> wrote:
>
> Drivers must not reference functions marked with __exit as these likely
> are not available when the code is built-in.
>
> There are few creative offenders uncovered for example in ARCH=amd64
> allmodconfig builds. So only trigger the section mismatch warning for
> W=1 builds.
>
> The dual rule that drivers must not reference .init.* is implemented
> since commit 0db252452378 ("modpost: don't allow *driver to reference
> .init.*") which however missed that .exit.* should be handled in the
> same way.
>
> Thanks to Masahiro Yamada and Arnd Bergmann who gave valuable hints to
> find this improvement.
>
> Signed-off-by: Uwe Kleine-König <[e-mail]>


Thanks.

If there is no objection from anybody, I will apply this
to my fixes branch and send a pull request.


Then, I hope you (or somebody) will volunteer to fix broken drivers.





--
Best Regards
Masahiro Yamada

Uwe Kleine-König Sept. 30, 2023, 8:48 p.m. UTC | #2

Hello Masahiro,

On Sun, Oct 01, 2023 at 05:33:54AM +0900, Masahiro Yamada wrote:
> Then, I hope you (or somebody) will volunteer to fix broken drivers.

I already added that to my todo list. However it's otherwise non-empty,
so I won't promise to tackle that quickly.

Best regards
Uwe

Patch

diff --git a/scripts/mod/modpost.c b/scripts/mod/modpost.c
index de499dce5265..b3dee80497cb 100644
--- a/scripts/mod/modpost.c
+++ b/scripts/mod/modpost.c
@@ -1015,9 +1015,20 @@  static int secref_whitelist(const char *fromsec, const char *fromsym,
 				    "*_console")))
 		return 0;
 
-	/* symbols in data sections that may refer to meminit/exit sections */
+	/* symbols in data sections that may refer to meminit sections */
 	if (match(fromsec, PATTERNS(DATA_SECTIONS)) &&
-	    match(tosec, PATTERNS(ALL_XXXINIT_SECTIONS, ALL_EXIT_SECTIONS)) &&
+	    match(tosec, PATTERNS(ALL_XXXINIT_SECTIONS, ALL_XXXEXIT_SECTIONS)) &&
+	    match(fromsym, PATTERNS("*driver")))
+		return 0;
+
+	/*
+	 * symbols in data sections must not refer to .exit.*, but there are
+	 * quite a few offenders, so hide these unless for W=1 builds until
+	 * these are fixed.
+	 */
+	if (!extra_warn &&
+	    match(fromsec, PATTERNS(DATA_SECTIONS)) &&
+	    match(tosec, PATTERNS(EXIT_SECTIONS)) &&
 	    match(fromsym, PATTERNS("*driver")))
 		return 0;