Re: [PATCH] linux-user: Add AT_EXECFN and AT_EXECFD auxval

[Laurent Vivier]

Le 02/03/2020 à 19:18, Lirong Yuan a écrit :
> On Mon, Mar 2, 2020 at 6:39 AM Laurent Vivier <address@hidden> wrote:
>>
>> Le 21/02/2020 à 21:28, Lirong Yuan a écrit :
>>> This change adds the support for AT_EXECFN and AT_EXECFD auxval.
>>
>> Why do we need AT_EXECFD?
>>
>> AT_EXECFD is normally only used with binfmt_misc so I don't see any use
>> cases for it with QEMU.
>>
>> For AT_EXECFN, according to kernel commit
>>
>>   651910874633 execve filename: document and export via auxiliary vector
>>
>> It sould be like readlink("/proc/self/exe",), and thus I think we should
>> use realpath() like we have in syscall.c for TARGET_NR_readlink:
>>
>>  8843     case TARGET_NR_readlink:
>> ...
>>  8854                 char real[PATH_MAX], *temp;
>>  8855                 temp = realpath(exec_path, real);
>> ...
>>
>> Thanks,
>> Laurent
>>
>>>
>>> Signed-off-by: Lirong Yuan <address@hidden>
>>> ---
>>>  linux-user/elfload.c | 13 +++++++++----
>>>  1 file changed, 9 insertions(+), 4 deletions(-)
>>>
>>> diff --git a/linux-user/elfload.c b/linux-user/elfload.c
>>> index f3080a1635..7e0f3042f1 100644
>>> --- a/linux-user/elfload.c
>>> +++ b/linux-user/elfload.c
>>> @@ -1568,7 +1568,7 @@ struct exec
>>>                                   ~(abi_ulong)(TARGET_ELF_EXEC_PAGESIZE-1))
>>>  #define TARGET_ELF_PAGEOFFSET(_v) ((_v) & (TARGET_ELF_EXEC_PAGESIZE-1))
>>>
>>> -#define DLINFO_ITEMS 15
>>> +#define DLINFO_ITEMS 17
>>>
>>>  static inline void memcpy_fromfs(void * to, const void * from, unsigned 
>>> long n)
>>>  {
>>> @@ -1888,11 +1888,14 @@ static abi_ulong loader_build_fdpic_loadmap(struct 
>>> image_info *info, abi_ulong s
>>>      return sp;
>>>  }
>>>
>>> -static abi_ulong create_elf_tables(abi_ulong p, int argc, int envc,
>>> +static abi_ulong create_elf_tables(struct linux_binprm *bprm,
>>>                                     struct elfhdr *exec,
>>>                                     struct image_info *info,
>>>                                     struct image_info *interp_info)
>>>  {
>>> +    abi_ulong p = bprm->p;
>>> +    int argc = bprm->argc;
>>> +    int envc = bprm->envc;
>>>      abi_ulong sp;
>>>      abi_ulong u_argc, u_argv, u_envp, u_auxv;
>>>      int size;
>>> @@ -2032,6 +2035,8 @@ static abi_ulong create_elf_tables(abi_ulong p, int 
>>> argc, int envc,
>>>      NEW_AUX_ENT(AT_CLKTCK, (abi_ulong) sysconf(_SC_CLK_TCK));
>>>      NEW_AUX_ENT(AT_RANDOM, (abi_ulong) u_rand_bytes);
>>>      NEW_AUX_ENT(AT_SECURE, (abi_ulong) qemu_getauxval(AT_SECURE));
>>> +    NEW_AUX_ENT(AT_EXECFN, info->file_string);
>>> +    NEW_AUX_ENT(AT_EXECFD, bprm->fd);
>>>
>>>  #ifdef ELF_HWCAP2
>>>      NEW_AUX_ENT(AT_HWCAP2, (abi_ulong) ELF_HWCAP2);
>>> @@ -2870,8 +2875,8 @@ int load_elf_binary(struct linux_binprm *bprm, struct 
>>> image_info *info)
>>>  #endif
>>>      }
>>>
>>> -    bprm->p = create_elf_tables(bprm->p, bprm->argc, bprm->envc, &elf_ex,
>>> -                                info, (elf_interpreter ? &interp_info : 
>>> NULL));
>>> +    bprm->p = create_elf_tables(bprm, &elf_ex, info,
>>> +                                (elf_interpreter ? &interp_info : NULL));
>>>      info->start_stack = bprm->p;
>>>
>>>      /* If we have an interpreter, set that as the program's entry point.
>>>
>>
> 
> Hi Laurent,

Hi Lirong,

> I added support for AT_EXECFD because I thought it might be useful to
> implement all types that getauxval could take as an argument.
> Would you prefer that it be removed?

I think providing the AT_EXECFD to the target binary could make it think
it has been run directly by binfmt_misc (as an interpreter itself), and
that is not the case (qemu is run by binfmt_misc and is the interpreter).

So I would prefer you remove it.

> For AT_EXECFN, there are two questions that we considered:
> 1) What should it return?
> Since QEMU is emulating running the guest program, the function should
> return the file name of the guest program (info->file_string), rather
> than the QEMU program itself, which we get from
> qemu_getauxval(AT_EXECFN).
> 
> 2) Should it return the full path or as is?
> We tested the behavior of getauxval with a simple test program on
> Linux, and it turned out that it returned file path as is. For
> example,
> $ ./test
> getauxval(AT_EXECFN): ./test
> $ /usr/local/home/tmp/test
> getauxval(AT_EXECFN): /usr/local/home/tmp/test

If you have compared the result with the real one it's fine for me.

Resend tou patch without the AT_EXECFD part and it will be good.

Thanks,
Laurent