Re: [RFC PATCH glibc 1/3] glibc: Perform rseq(2) registration at C startup and thread creation (v18)

[Mathieu Desnoyers <mathieu.desnoyers@efficios.com>]

----- On Apr 30, 2020, at 4:34 PM, Florian Weimer fweimer@redhat.com wrote:

> * Mathieu Desnoyers:
> 
>> I'm actually wondering about "EPERM". Is it sometimes used by seccomp
>> filters to refuse a system call ?
> 
> Yes, it's the default for systemd-nspawn.  I have argued against it, but
> unsuccessfully.  It breaks all kinds of stuff inside glibc, too.

OK, so how about this errno handling ?

static inline void
rseq_register_current_thread (void)
{
  int ret;

  if (__rseq_abi.cpu_id != RSEQ_CPU_ID_UNINITIALIZED)
    __libc_fatal ("glibc fatal error: "
                  "rseq already initialized for this thread\n");
  ret = INTERNAL_SYSCALL_CALL (rseq, &__rseq_abi, sizeof (struct rseq),
                              0, RSEQ_SIG);
  if (INTERNAL_SYSCALL_ERROR_P (ret))
    {
      const char *msg = NULL;

      switch (INTERNAL_SYSCALL_ERRNO (ret))
        {
        case ENOSYS:    /* rseq system call not implemented.  */
        case EPERM:     /* rseq system call filtered by seccomp.  */
        case EACCES:    /* rseq system call filtered by seccomp.  */
          __rseq_abi.cpu_id = RSEQ_CPU_ID_REGISTRATION_FAILED;
          break;
        case EINVAL:
          msg = "glibc fatal error: rseq already registered for this thread\n";
          break;
        case EBUSY:
          msg = "glibc fatal error: rseq parameters are invalid\n";
          break;
        case EFAULT:
          msg = "glibc fatal error: rseq is an invalid address\n";
          break;
        default:
          msg = "glibc fatal error: unexpected rseq errno\n";
          break;
        }
      if (msg)
        __libc_fatal (msg);
    }
}

Thanks,

Mathieu



-- 
Mathieu Desnoyers
EfficiOS Inc.
http://www.efficios.com