[PATCH] fix insidious bug with init section identification in the kernel module loader

[PATCH] fix insidious bug with init section identification in the kernel module loader

[James Bottomley]

[-- Attachment #1: Type: text/plain, Size: 847 bytes --]

This problem manifests itself nastily on parisc, where the linker seems
to generate large number of elf sections, often one for each non static
function, with names like

.text.<function name>

The problem is that the kernel/module.c uses the following test to
identify init sections:

                           || strstr(secstrings + s->sh_name, ".init"))

Which will pass if ".init" is matched anywhere in the elf section name. 
Obviously, any parisc sections for functions that begin with "init"
match, and are spuriously dumped into the init code.

The fix that works for me on parisc is just to check the beginning of
the string for ".init".  However, I can't find any document that
guarantees this to be correct.  On the other hand, no non-init section
will begin with ".init", so the patch should be safe, just may be non
optimal.

James


[-- Attachment #2: tmp.diff --]
[-- Type: text/plain, Size: 1208 bytes --]

===== kernel/module.c 1.86 vs edited =====
--- 1.86/kernel/module.c	Wed Jun 11 00:55:09 2003
+++ edited/kernel/module.c	Thu Jun 12 15:46:13 2003
@@ -1194,7 +1194,8 @@
 			if ((s->sh_flags & masks[m][0]) != masks[m][0]
 			    || (s->sh_flags & masks[m][1])
 			    || s->sh_entsize != ~0UL
-			    || strstr(secstrings + s->sh_name, ".init"))
+			    || strncmp(secstrings + s->sh_name,
+				       ".init", 5) == 0)
 				continue;
 			s->sh_entsize = get_offset(&mod->core_size, s);
 			DEBUGP("\t%s\n", secstrings + s->sh_name);
@@ -1209,7 +1210,8 @@
 			if ((s->sh_flags & masks[m][0]) != masks[m][0]
 			    || (s->sh_flags & masks[m][1])
 			    || s->sh_entsize != ~0UL
-			    || !strstr(secstrings + s->sh_name, ".init"))
+			    || strncmp(secstrings + s->sh_name,
+				       ".init", 5) != 0)
 				continue;
 			s->sh_entsize = (get_offset(&mod->init_size, s)
 					 | INIT_OFFSET_MASK);
@@ -1413,7 +1415,7 @@
 		}
 #ifndef CONFIG_MODULE_UNLOAD
 		/* Don't load .exit sections */
-		if (strstr(secstrings+sechdrs[i].sh_name, ".exit"))
+		if (strncmp(secstrings+sechdrs[i].sh_name, ".exit", 5) == 0)
 			sechdrs[i].sh_flags &= ~(unsigned long)SHF_ALLOC;
 #endif
 	}

Re: [PATCH] fix insidious bug with init section identification in the kernel module loader

[Rusty Russell]

In message <1055452158.2117.52.camel@mulgrave> you write:
> This problem manifests itself nastily on parisc, where the linker seems
> to generate large number of elf sections, often one for each non static
> function, with names like
> 
> .text.<function name>

-ffunction-sections, perhaps?

> The problem is that the kernel/module.c uses the following test to
> identify init sections:
> 
>                            || strstr(secstrings + s->sh_name, ".init"))
> 
> Which will pass if ".init" is matched anywhere in the elf section name. 
> Obviously, any parisc sections for functions that begin with "init"
> match, and are spuriously dumped into the init code.

Yep.  I agree with your fix: there are some archs which haven't
changed over from ".data.init" to ".init.data", and ".data.exit" to
".exit.data", but as you point out, the worst that happens is they
don't discard stuff they could.

Linus, please apply.
Rusty.
--
  Anyone who quotes me in their sig is an idiot. -- Rusty Russell.

===== kernel/module.c 1.86 vs edited =====
--- 1.86/kernel/module.c	Wed Jun 11 00:55:09 2003
+++ edited/kernel/module.c	Thu Jun 12 15:46:13 2003
@@ -1194,7 +1194,8 @@
 			if ((s->sh_flags & masks[m][0]) != masks[m][0]
 			    || (s->sh_flags & masks[m][1])
 			    || s->sh_entsize != ~0UL
-			    || strstr(secstrings + s->sh_name, ".init"))
+			    || strncmp(secstrings + s->sh_name,
+				       ".init", 5) == 0)
 				continue;
 			s->sh_entsize = get_offset(&mod->core_size, s);
 			DEBUGP("\t%s\n", secstrings + s->sh_name);
@@ -1209,7 +1210,8 @@
 			if ((s->sh_flags & masks[m][0]) != masks[m][0]
 			    || (s->sh_flags & masks[m][1])
 			    || s->sh_entsize != ~0UL
-			    || !strstr(secstrings + s->sh_name, ".init"))
+			    || strncmp(secstrings + s->sh_name,
+				       ".init", 5) != 0)
 				continue;
 			s->sh_entsize = (get_offset(&mod->init_size, s)
 					 | INIT_OFFSET_MASK);
@@ -1413,7 +1415,7 @@
 		}
 #ifndef CONFIG_MODULE_UNLOAD
 		/* Don't load .exit sections */
-		if (strstr(secstrings+sechdrs[i].sh_name, ".exit"))
+		if (strncmp(secstrings+sechdrs[i].sh_name, ".exit", 5) == 0)
 			sechdrs[i].sh_flags &= ~(unsigned long)SHF_ALLOC;
 #endif
 	}

Re: [PATCH] fix insidious bug with init section identification in the kernel module loader

[James Bottomley]

On Thu, 2003-06-12 at 20:18, Rusty Russell wrote:
> In message <1055452158.2117.52.camel@mulgrave> you write:
> > This problem manifests itself nastily on parisc, where the linker seems
> > to generate large number of elf sections, often one for each non static
> > function, with names like
> > 
> > .text.<function name>
> 
> -ffunction-sections, perhaps?

Yes, we only have nineteen bit branch relocations on the parisc1.1, so
we usually have to interleave trampoline stubs in between the function
sections to get from one end of the kernel to the other.

I suppose as long as we never get a module bigger than 256k, we don't
actually need to split modules up this way...

James