[PATCH] module: use MODULE_SYMBOL_PREFIX with module_layout

[PATCH] module: use MODULE_SYMBOL_PREFIX with module_layout

[Rusty Russell]

From: Mike Frysinger <vapier@gentoo.org>

The check_modstruct_version() needs to look up the symbol "module_layout"
in the kernel, but it does so literally and not by a C identifier.  The
trouble is that it does not include a symbol prefix for those ports that
need it (like the Blackfin and H8300 port).  So make sure we tack on the
MODULE_SYMBOL_PREFIX define to the front of it.

Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
---
 kernel/module.c |    3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/kernel/module.c b/kernel/module.c
--- a/kernel/module.c
+++ b/kernel/module.c
@@ -1068,7 +1068,8 @@ static inline int check_modstruct_versio
 {
 	const unsigned long *crc;
 
-	if (!find_symbol("module_layout", NULL, &crc, true, false))
+	if (!find_symbol(MODULE_SYMBOL_PREFIX "module_layout", NULL,
+			 &crc, true, false))
 		BUG();
 	return check_version(sechdrs, versindex, "module_layout", mod, crc);
 }

Re: module version magic and arches with symbol prefixes

[Rusty Russell]

On Fri, 19 Jun 2009 12:54:44 am Mike Frysinger wrote:
> the current check_modstruct_version() does this:
> {
>     const unsigned long *crc;
>
>     if (!find_symbol("module_layout", NULL, &crc, true, false))
>         BUG();
>     return check_version(sechdrs, versindex, "module_layout", mod, crc);
> }
> the trouble here is that it looks for a literal "module_layout" symbol
> and for ports that have symbol prefixes (a quick check shows Blackfin
> & h8300), this aint going to work.

MODULE_SYMBOL_PREFIX is the fix for this, ie:

	if (!find_symbol(MODULE_SYMBOL_PREFIX "module_layout), ...

> also, using BUG() here seems pretty damn harsh.  wouldnt it make more
> sense to do something like:
>     if (WARN_ON(!find_symbol("module_layout", NULL, &crc, true, false)))
>         return 0;
> this way the module is simply not loaded rather than killing the kernel

No, it means the kernel didn't build properly.  Better to fail spectacularly: 
ideally we'd do this find in an init routine and guarantee a boot crash.

Thanks,
Rusty.

[PATCH] module: use MODULE_SYMBOL_PREFIX with module_layout

[Mike Frysinger]

The check_modstruct_version() needs to look up the symbol "module_layout"
in the kernel, but it does so literally and not by a C identifier.  The
trouble is that it does not include a symbol prefix for those ports that
need it (like the Blackfin and H8300 port).  So make sure we tack on the
MODULE_SYMBOL_PREFIX define to the front of it.

Signed-off-by: Mike Frysinger <vapier@gentoo.org>
---
 kernel/module.c |    3 ++-
 1 files changed, 2 insertions(+), 1 deletions(-)

diff --git a/kernel/module.c b/kernel/module.c
index 80036bc..7850bef 100644
--- a/kernel/module.c
+++ b/kernel/module.c
@@ -1053,7 +1053,8 @@ static inline int check_modstruct_version(Elf_Shdr *sechdrs,
 {
 	const unsigned long *crc;
 
-	if (!find_symbol("module_layout", NULL, &crc, true, false))
+	if (!find_symbol(MODULE_SYMBOL_PREFIX "module_layout", NULL,
+	                 &crc, true, false))
 		BUG();
 	return check_version(sechdrs, versindex, "module_layout", mod, crc);
 }
-- 
1.6.3.1

Re: [PATCH] module: use MODULE_SYMBOL_PREFIX with module_layout

[Rusty Russell]

On Fri, 19 Jun 2009 05:19:22 pm Mike Frysinger wrote:
> The check_modstruct_version() needs to look up the symbol "module_layout"
> in the kernel, but it does so literally and not by a C identifier.  The
> trouble is that it does not include a symbol prefix for those ports that
> need it (like the Blackfin and H8300 port).  So make sure we tack on the
> MODULE_SYMBOL_PREFIX define to the front of it.

Thanks, applied.  I'm surprised this took so long to find.

Rusty.

Re: [PATCH] module: use MODULE_SYMBOL_PREFIX with module_layout

[Mike Frysinger]

On Fri, Jun 19, 2009 at 10:02, Rusty Russell wrote:
> On Fri, 19 Jun 2009 05:19:22 pm Mike Frysinger wrote:
>> The check_modstruct_version() needs to look up the symbol "module_layout"
>> in the kernel, but it does so literally and not by a C identifier.  The
>> trouble is that it does not include a symbol prefix for those ports that
>> need it (like the Blackfin and H8300 port).  So make sure we tack on the
>> MODULE_SYMBOL_PREFIX define to the front of it.
>
> Thanks, applied.  I'm surprised this took so long to find.

well, technically i first noticed on 2007-10-24, but the fix that went
in internally sucked and people forgot about it ;)
-mike