Re: [PATCH] Linux: Add execveat system call wrapper

[Adhemerval Zanella <adhemerval.zanella@linaro.org>]

On 28/04/2020 14:50, Florian Weimer wrote:
> * Adhemerval Zanella via Libc-alpha:
> 
>> On 28/04/2020 09:20, Alexandra Hájková via Libc-alpha wrote:
>>
>>> +/* Execute the file FD refers to, overlaying the running program image.
>>> +   ARGV and ENVP are passed to the new program, as for `execve'.  */
>>> +int
>>> +execveat (int dirfd, const char *path, char *const argv[], char *const envp[],
>>> +          int flags)
>>> +{
>>> +  /* Avoid implicit array coercion in syscall macros.  */
>>> +  INLINE_SYSCALL_CALL (execveat, dirfd, path, &argv[0], &envp[0], flags);
>>> +#ifndef __ASSUME_EXECVEAT
>>> +  if (errno != ENOSYS)
>>> +    return -1;
>>> +
>>> +  int fd = openat (dirfd, path, flags | O_CLOEXEC);
>>> +  if (fd < 0)
>>> +    return -1;
>>> +
>>
>> Is this subject to a TOCTOU where a thread might close the fd by explicitly
>> reading /proc/self/fd/?
> 
> Sure, but that's as undefined as closing a file descriptor that the
> code has not opened.  This impacts any code in glibc which opens file
> descriptors.  I don't think it's a problem.
> 

Fair enough, I just want to avoid future issues with syscall fallback
machanisms.