Re: [PATCH v2 1/3] util: validate whether O_DIRECT is supported after failure

[Markus Armbruster <armbru@redhat.com>]

Daniel P. Berrangé <berrange@redhat.com> writes:

> Currently we suggest that a filesystem may not support O_DIRECT after
> seeing an EINVAL. Other things can cause EINVAL though, so it is better
> to do an explicit check, and then report a definitive error message.
>
> Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
> ---
>  util/osdep.c | 15 +++++++++++----
>  1 file changed, 11 insertions(+), 4 deletions(-)
>
> diff --git a/util/osdep.c b/util/osdep.c
> index 4829c07ff6..e2b7507ee2 100644
> --- a/util/osdep.c
> +++ b/util/osdep.c
> @@ -332,9 +332,11 @@ int qemu_open(const char *name, int flags, ...)
>      }
>  
>  #ifdef O_CLOEXEC
> -    ret = open(name, flags | O_CLOEXEC, mode);
> -#else
> +    flags |= O_CLOEXEC;
> +#endif
>      ret = open(name, flags, mode);
> +
> +#ifndef O_CLOEXEC
>      if (ret >= 0) {
>          qemu_set_cloexec(ret);
>      }

I'd prefer something like

   #ifdef O_CLOEXEC
       flags |= O_CLOEXEC;
       ret = open(name, flags, mode);
   #else
       ret = open(name, flags, mode);
       if (ret >= 0) {
           qemu_set_cloexec(ret);
       }
   #endif

Continues to duplicate open(), but spares me the effort to fuse two
#ifdef sections in my head to understand what is being done in each
case.

> @@ -342,8 +344,13 @@ int qemu_open(const char *name, int flags, ...)
>  
>  #ifdef O_DIRECT
>      if (ret == -1 && errno == EINVAL && (flags & O_DIRECT)) {
> -        error_report("file system may not support O_DIRECT");
> -        errno = EINVAL; /* in case it was clobbered */
> +        int newflags = flags & ~O_DIRECT;
> +        ret = open(name, newflags, mode);

I'd prefer the more concise

           ret = open(name, flags & ~O_DIRECT, mode);

> +        if (ret != -1) {
> +            close(ret);
> +            error_report("file system does not support O_DIRECT");
> +            errno = EINVAL;
> +        }
>      }
>  #endif /* O_DIRECT */

The function now reports to stderr in just one of many failure modes.
That's wrong.  Looks like the next patch fixes this defect.  I'd swap
the two.