This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: [PATCH] mips: Fix argument passing for inlined syscalls on Linux [BZ #25523]
On 22/02/2020 19:40, Maciej W. Rozycki wrote:
> On Mon, 10 Feb 2020, WANG Xuerui wrote:
>
>> According to [gcc documentation][1], temporary variables must be used for
>> the desired content to not be call-clobbered.
>
> Why does it specifically matter here?
>
>> diff --git a/sysdeps/unix/sysv/linux/mips/mips32/sysdep.h b/sysdeps/unix/sysv/linux/mips/mips32/sysdep.h
>> index beefcf284b..c275d63f67 100644
>> --- a/sysdeps/unix/sysv/linux/mips/mips32/sysdep.h
>> +++ b/sysdeps/unix/sysv/linux/mips/mips32/sysdep.h
> [...]
>> @@ -202,11 +203,13 @@ union __mips_syscall_return
>> long _sys_result; \
>> \
>> { \
>> + long _arg1 = (long) (arg1); \
>> + long _arg2 = (long) (arg2); \
>> register long __s0 asm ("$16") __attribute__ ((unused)) \
>> = (number); \
>> register long __v0 asm ("$2"); \
>> - register long __a0 asm ("$4") = (long) (arg1); \
>> - register long __a1 asm ("$5") = (long) (arg2); \
>> + register long __a0 asm ("$4") = _arg1; \
>> + register long __a1 asm ("$5") = _arg2; \
>> register long __a3 asm ("$7"); \
>> __asm__ volatile ( \
>> ".set\tnoreorder\n\t" \
>
> Can e.g. `(long) (arg1)' end up as a library call? Do you have a case,
> even potential, where such clobbering actually happens?
On sysdeps/unix/sysv/linux/spawni.c:
188 if ((attr->__flags & POSIX_SPAWN_RESETIDS) != 0
189 && (local_seteuid (__getuid ()) != 0
190 || local_setegid (__getgid ()) != 0))
191 goto fail;
And local_seteuid/local_setegid are defined as:
sysdeps/unix/sysv/linux/local-setxid.h:
5 #ifdef __NR_setresuid32
6 # define local_seteuid(id) INLINE_SYSCALL (setresuid32, 3, -1, id, -1)
7 #else
8 # define local_seteuid(id) INLINE_SYSCALL (setresuid, 3, -1, id, -1)
9 #endif
10
11
12 #ifdef __NR_setresgid32
13 # define local_setegid(id) INLINE_SYSCALL (setresgid32, 3, -1, id, -1)
14 #else
15 # define local_setegid(id) INLINE_SYSCALL (setresgid, 3, -1, id, -1)
16 #endif
In any case, the previous usage of inline syscall is indeed fragile
and subject to such potential breakage.