[RFC 3/4] Add test for --decode-fds=eventfd

[Sahil]

Hi,

Thank you for the feedback.

On Monday, April 29, 2024 4:41:06 PM IST Dmitry V. Levin wrote:
> On Mon, Apr 29, 2024 at 09:01:46AM +0530, Sahil Siddiq wrote:
> [...]
> > +    struct pollfd fds[1];
> > +    fds[0].fd = efd2;
> > +    fds[0].events = POLLIN;
> > +
> > +    int ret = poll(fds, 1, -1);
> > +    if (ret == -1)
> > +        perror_msg_and_fail("eventfd");
> > +
> > +    printf("poll([{fd=%d<eventfd-count:%"PRIu64">, events=POLLIN}], 1, -1) = %d ([{fd=%d, revents=POLLIN}])\n",
> > +               efd2, efd_counter, ret, fd2);
> 
> The use of poll is problematic as this system call is not implemented on
> recently introduced architectures.

Oh, I wasn't aware of this. Could you give me an example of a recently
introduced architecture? Is openRISC an example of such an architecture?
How do I get the list of architectures in which a syscall hasn't been
implemented?

Thanks,
Sahil