module: keep percpu symbols in module's symtab
PATCH v2 1/1 module: keep percpu symbols in module's symtab
Currently, percpu symbols from .data..percpu ELF section of a module are not copied over and stored in final symtab array of struct module. Consequently such symbol cannot be returned via kallsyms API (for example kallsyms_lookup_name). This can be especially confusing when the percpu symbol is exported. Only its __ksymtab et al. are present in its symtab.
The culprit is in layout_and_allocate() function where SHF_ALLOC flag is dropped for .data..percpu section. There is in fact no need to copy the section to final struct module, because kernel module loader allocates extra percpu section by itself. Unfortunately only symbols from SHF_ALLOC sections are copied due to a check in is_core_symbol().
The patch changes is_core_symbol() function to copy over also percpu symbols (their st_shndx points to .data..percpu ELF section). We do it only if CONFIG_KALLSYMS_ALL is set to be consistent with the rest of the function (ELF section is SHF_ALLOC but !SHF_EXECINSTR). Finally elf_type() returns type 'a' for a percpu symbol because its address is absolute.
--- kernel/module.c | 15 +++++++++++---- 1 file changed, 11 insertions(+), 4 deletions(-)
kernel/module.c +11 -4
@@ -2401,7 +2401,7 @@ static char elf_type(const Elf_Sym *sym, const struct load_info *info) }if (sym->st_shndx ==SHN_UNDEF )return 'U' ; -if (sym->st_shndx ==SHN_ABS ) +if (sym->st_shndx ==SHN_ABS || sym->st_shndx == info->index.pcpu)return 'a' ;if (sym->st_shndx >=SHN_LORESERVE )return '?' ;
@@ -2430,7 +2430,7 @@ static char elf_type(const Elf_Sym *sym, const struct load_info *info) }static bool is_core_symbol (const Elf_Sym *src,const Elf_Shdr *sechdrs, -unsigned int shnum) +unsigned int shnum,unsigned int pcpundx) {const Elf_Shdr *sec;
@@ -2439,6 +2439,11 @@ static bool is_core_symbol(const Elf_Sym *src, const Elf_Shdr *sechdrs, || !src->st_name)return false; +#ifdef CONFIG_KALLSYMS_ALL +if (src->st_shndx == pcpundx) +return true; +#endif + sec = sechdrs + src->st_shndx; if (!(sec->sh_flags & SHF_ALLOC)#ifndef CONFIG_KALLSYMS_ALL
@@ -2476,7 +2481,8 @@ static void layout_symtab(struct module *mod, struct load_info *info)/* Compute total space required for the core symbols' strtab. */ for (ndst = i =0 ; i < nsrc; i++) {if (i ==0 || -is_core_symbol (src+i, info->sechdrs, info->hdr->e_shnum)) { +is_core_symbol (src+i, info->sechdrs, info->hdr->e_shnum, + info->index.pcpu)) { strtab_size +=strlen (&info->strtab[src[i].st_name])+1 ; ndst++; }
@@ -2518,7 +2524,8 @@ static void add_kallsyms(struct module *mod, const struct load_info *info) src = mod->symtab;for (ndst = i =0 ; i < mod->num_symtab; i++) {if (i ==0 || -is_core_symbol (src+i, info->sechdrs, info->hdr->e_shnum)) { +is_core_symbol (src+i, info->sechdrs, info->hdr->e_shnum, + info->index.pcpu)) { dst[ndst] = src[i]; dst[ndst++].st_name = s - mod->core_strtab; s += strlcpy(s, &mod->strtab[src[i].st_name],
Signature · 1 line
2.1.4
Rusty Russell
AppliedQuoted 13 lines · Miroslav Benes
Currently, percpu symbols from .data..percpu ELF section of a module are not copied over and stored in final symtab array of struct module. Consequently such symbol cannot be returned via kallsyms API (for example
kallsyms_lookup_name). This can be especially confusing when the percpu symbol is exported. Only its__ksymtabet al. are present in its symtab.The culprit is in
layout_and_allocate()function whereSHF_ALLOCflag is dropped for .data..percpu section. There is in fact no need to copy the section to final struct module, because kernel module loader allocates extra percpu section by itself. Unfortunately only symbols fromSHF_ALLOCsections are copied due to a check inis_core_symbol().
Quoted 8 lines
The patch changes
is_core_symbol()function to copy over also percpu symbols (theirst_shndxpoints to .data..percpu ELF section). We do it only ifCONFIG_KALLSYMS_ALLis set to be consistent with the rest of the function (ELF section isSHF_ALLOCbut !SHF_EXECINSTR). Finallyelf_type()returns type 'a' for a percpu symbol because its address is absolute.Signed-off-by Miroslav Benes <[email protected]>
Thanks, applied!
Cheers,
Rusty.