On Fri, Nov 23, 2012 at 07:35:36PM +0900, satoru takeuchi wrote:

Hi,

2012/11/22 Greg Kroah-Hartman [off-list ref]:

quoted

3.6-stable review patch.  If anyone has any objections, please let me know.

------------------

From: Rusty Russell <redacted>

commit 59ef28b1f14899b10d6b2682c7057ca00a9a3f47 upstream.

Masaki found and patched a kallsyms issue: the last symbol in a
module's symtab wasn't transferred.  This is because we manually copy
the zero'th entry (which is always empty) then copy the rest in a loop
starting at 1, though from src[0].  His fix was minimal, I prefer to
rewrite the loops in more standard form.

There are two loops: one to get the size, and one to copy.  Make these
identical: always count entry 0 and any defined symbol in an allocated
non-init section.

This bug exists since the following commit was introduced.
   module: reduce symbol table for loaded modules (v2)
   commit: 4a4962263f07d14660849ec134ee42b63e95ea9a

LKML: http://lkml.org/lkml/2012/10/24/27
Reported-by: Masaki Kimura <redacted>
Signed-off-by: Rusty Russell <redacted>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

---
 kernel/module.c |   27 ++++++++++++++++-----------
 1 file changed, 16 insertions(+), 11 deletions(-)

--- a/kernel/module.c
+++ b/kernel/module.c

@@ -2273,12 +2273,17 @@ static void layout_symtab(struct module
        src = (void *)info->hdr + symsect->sh_offset;
        nsrc = symsect->sh_size / sizeof(*src);

+       /* strtab always starts with a nul, so offset 0 is the empty string. */
+       strtab_size = 1;
+

I suspect above code sniped is not needed since the size of src[0](always 1) is
counted at the following code("strtab_size += strlen(..) + 1;").

quoted

        /* Compute total space required for the core symbols' strtab. */
-       for (ndst = i = strtab_size = 1; i < nsrc; ++i, ++src)
-               if (is_core_symbol(src, info->sechdrs, info->hdr->e_shnum)) {
-                       strtab_size += strlen(&info->strtab[src->st_name]) + 1;
+       for (ndst = i = 0; i < nsrc; i++) {
+               if (i == 0 ||
+                   is_core_symbol(src+i, info->sechdrs, info->hdr->e_shnum)) {

src[i] rather than src+1 is better as kallsyms();

quoted

+                       strtab_size += strlen(&info->strtab[src[i].st_name])+1;
                        ndst++;
                }
+       }

        /* Append room for core symbols at end of core part. */
        info->symoffs = ALIGN(mod->core_size, symsect->sh_addralign ?: 1);

@@ -2312,15 +2317,15 @@ static void add_kallsyms(struct module *
        mod->core_symtab = dst = mod->module_core + info->symoffs;
        mod->core_strtab = s = mod->module_core + info->stroffs;
        src = mod->symtab;
-       *dst = *src;
        *s++ = 0;

This "*s++"line is also not needed since the content of src[0] is copied in the
following loop.

If my understanding is correct, the allsym table becomes as follows by
applying this patch.

==================
\0\0<first core symbol>\0<second core symbol>\0... # "\0" means null
character in this case
==================

It should be as follows.

====================
\0<first core symbol>\0<second core symbol>\0...
====================

This patch would work, but need extra one byte per module. Olease let
us know if I misunderstand something.
I didn't confirm with real machine.

Please confirm with a real machine, and if the problem is there, please
submit a patch to get it changed in Linus's tree, which is where this
patch came from.

thanks,

greg k-h