Re: [PATCH RFC] improve the error message in "xl list --long"

[Andrew Cooper <andrew.cooper3@citrix.com>]

On 17/10/13 17:21, Bamvor Jian Zhang wrote:
> with this patch, xl will raise a proper error when execute "xl list --long"
> for existed domain but no domain configuration found. it would encounter
> if other toolstack(like libvirt) create the vm, xl could saw the domain
> name but could not get the domain configuration.
>
> there is a commit a76377f1 check the return value of
> libxl_read_file_contents in libxl_userdata_retrieve. but skip the ENOENT.
> it is almost three years before. i do not know the senario about that
> commit. it seems that it is not safe to simplely remove the ENOENT.
>
> Signed-off-by: Bamvor Jian Zhang <bjzhang@suse.com>
> ---
>  tools/libxl/libxl_dom.c | 6 ++++++
>  1 file changed, 6 insertions(+)
>
> diff --git a/tools/libxl/libxl_dom.c b/tools/libxl/libxl_dom.c
> index 356f920..8daf51a 100644
> --- a/tools/libxl/libxl_dom.c
> +++ b/tools/libxl/libxl_dom.c
> @@ -1655,6 +1655,12 @@ int libxl_userdata_retrieve(libxl_ctx *ctx, uint32_t domid,
>          goto out;
>      }
>  
> +    if (access(filename, F_OK) != 0) {
> +        LIBXL__LOG(ctx, LIBXL__LOG_ERROR, "Unknown domain configuration. Is domain id '%u' owned by another libxl toolstack?", domid);

This line is too long,  perhaps putting the start of the string on a new
line, and splitting again at the %u so the grepable parts of the string
stay contiguous.

Finally, "another libxl toolstack" is too specific (XAPI domains being a
prime example which would fall over here).  I would suggest just
"another toolstack".

~Andrew

> +        rc = ERROR_INVAL;
> +        goto out;
> +    }
> +
>      e = libxl_read_file_contents(ctx, filename, data_r ? &data : 0, &datalen);
>      if (e && errno != ENOENT) {
>          rc = ERROR_FAIL;