linux-kernel - Re: [PATCH v4 03/10] objtool: Handle different entry size of rodata

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <4993a8d0-d08b-d9c1-f13f-934af75c3103@loongson.cn>
Date: Tue, 26 Nov 2024 21:22:22 +0800
From: Tiezhu Yang <yangtiezhu@...ngson.cn>
To: Josh Poimboeuf <jpoimboe@...nel.org>
Cc: Huacai Chen <chenhuacai@...nel.org>, Peter Zijlstra
 <peterz@...radead.org>, loongarch@...ts.linux.dev,
 linux-kernel@...r.kernel.org
Subject: Re: [PATCH v4 03/10] objtool: Handle different entry size of rodata

On 11/26/2024 06:59 PM, Tiezhu Yang wrote:
> On 11/26/2024 03:02 PM, Josh Poimboeuf wrote:
>> On Fri, Nov 22, 2024 at 12:49:58PM +0800, Tiezhu Yang wrote:
>>> @@ -2107,8 +2112,15 @@ static int add_jump_table(struct objtool_file
>>> *file, struct instruction *insn,
>>>          if (reloc != table && reloc == next_table)
>>>              break;
>>>
>>> +        /* Handle the special cases compiled with Clang on LoongArch */
>>
>> This comment is not helpful at all.  A comment is only needed if the
>> code is not already obvious.  In that case it should describe what is
>> being done and why.
>
> Will remove it.
>
>>> +        if (file->elf->ehdr.e_machine == EM_LOONGARCH &&
>>> +            reloc_type(reloc) == R_LARCH_32_PCREL)
>>> +            rodata_entry_size = 4;
>>> +        else
>>> +            rodata_entry_size = 8;
>>
>> Is this really loongarch-specific or is it only related to the size of
>> the reloc?
>
> This is related with the reloc type, but it may be wrong to only
> check the reloc type to assign the value of rodata_entry_size,
> becasue the value of reloc type for different archs may be same,
> so it needs to check ehdr.e_machine and relocation type.
>
>> Can this be abstracted out to a reloc_size() function like
>> so?
>>
>>
>> https://git.kernel.org/pub/scm/linux/kernel/git/jpoimboe/linux.git/tree/tools/objtool/klp-diff.c?h=objtool-diff#n834
>>
>>
>> maybe it could live in elf.h.
>
> OK, if I understand your comment correctly, this should be an
> arch-specific function defined in
> tools/objtool/arch/*/include/arch/elf.h, otherwise it also needs to
> check ehdr.e_machine
> in tools/objtool/include/objtool/elf.h.

There are only macro definitions in
tools/objtool/arch/*/include/arch/elf.h,
so I think it is better to add reloc_size() in
tools/objtool/include/objtool/elf.h,
like this:

static inline unsigned int reloc_size(struct elf *elf, struct reloc *reloc)
{
         if (elf->ehdr.e_machine == EM_X86_64) {
                 switch (reloc_type(reloc)) {
                 case R_X86_64_32:
                 case R_X86_64_PC32:
                         return 4;
                 case R_X86_64_64:
                 case R_X86_64_PC64:
                         return 8;
                 default:
                         ERROR("unknown X86_64 reloc type");
                 }
         } else if (elf->ehdr.e_machine == EM_LOONGARCH) {
                 switch (reloc_type(reloc)) {
                 case R_LARCH_32:
                 case R_LARCH_32_PCREL:
                         return 4;
                 case R_LARCH_64:
                 case R_LARCH_64_PCREL:
                         return 8;
                 default:
                         ERROR("unknown LoongArch reloc type");
                 }
         } else {
                 return 8;
         }
}

and call it in check.c, like this:

rodata_entry_size = reloc_size(file->elf, reloc);

What do you think?

Thanks,
Tiezhu