[Pdbg] [PATCH 2/2] libsbefifo: Implement long read timeout for chip-ops

[Joel Stanley]

On Tue, 22 Feb 2022 at 02:51, Joel Stanley <joel at jms.id.au> wrote:

> +int sbefifo_set_long_timeout(struct sbefifo_context *sctx)
> +{
> +       unsigned int long_timeout = 30;
> +       int rc;
> +
> +       LOG("long_timeout: %u sec\n", long_timeout);
> +       rc = ioctl(sctx->fd, FSI_SBEFIFO_READ_TIMEOUT, &long_timeout);
> +       if (rc == -1 && errno == EINVAL) {
> +               /* Do not fail if kernel does not implement ioctl */
> +               rc = 0;
> +       }
> +
> +       return rc;
> +}
> +
> +int sbefifo_reset_timeout(struct sbefifo_context *sctx)
> +{
> +       unsigned int timeout = 0;
> +       int rc;
> +
> +       LOG("reset_timeout\n");
> +       rc = ioctl(sctx->fd, FSI_SBEFIFO_READ_TIMEOUT, &timeout);
> +       if (rc == -1 && errno == EINVAL) {

The kernel will set rc = -1 and errno to ENOTTY if the ioctl is not supported.

We should test for that instead of EINVAL, or even not test if the
ioctl failed at all?

> +               /* Do not fail if kernel does not implement ioctl */
> +               rc = 0;
> +       }
> +
> +       return rc;
> +}
> +
>  void sbefifo_debug(const char *fmt, ...)
>  {
>         va_list ap;
> --
> 2.34.1
>