Re: [Qemu-devel] [PATCH v3] qga: add mountpoint usage to GuestFilesystemInfo

[Eric Blake]

On 05/30/2018 09:19 PM, Chen Hanxiao wrote:
> From: Chen Hanxiao <address@hidden>
>
> This patch adds support for getting the usage of mounted
> filesystem.
> It's very useful when we try to monitor guest's filesystem.
>
> Cc: Michael Roth <address@hidden>
> Cc: Eric Blake <address@hidden>
>
> Signed-off-by: Chen Hanxiao <address@hidden>
> ---

> @@ -1079,7 +1083,19 @@ static GuestFilesystemInfo *build_guest_fsinfo(struct 
> FsMount *mount,
>       fs->type = g_strdup(mount->devtype);
>       build_guest_fsinfo_for_device(devpath, fs, errp);
>   
> +    if (statvfs(fs->mountpoint, &buf)) {
> +        error_setg_errno(errp, errno, "Failed to get statvfs");
> +        return NULL;
> +    }

Failing where we used to succeed is not nice.  Perhaps it would be 
better to make usage a best-effort query; and ignore statvfs() failure,

> +
> +    used = buf.f_blocks - buf.f_bfree;
> +    nonroot_total = used + buf.f_bavail;
> +    usage = (double) used / nonroot_total;
> +
> +    fs->usage = usage;
> +
>       g_free(devpath);
> +
>       return fs;
>   }
>   
> diff --git a/qga/qapi-schema.json b/qga/qapi-schema.json
> index 17884c7c70..68b9f60824 100644
> --- a/qga/qapi-schema.json
> +++ b/qga/qapi-schema.json
> @@ -846,13 +846,14 @@
>   # @name: disk name
>   # @mountpoint: mount point path
>   # @type: file system type string
> +# @usage: file system usage, fraction between 0 and 1 (since 3.0)
>   # @disk: an array of disk hardware information that the volume lies on,
>   #        which may be empty if the disk type is not supported
>   #
>   # Since: 2.2
>   ##
>   { 'struct': 'GuestFilesystemInfo',
> -  'data': {'name': 'str', 'mountpoint': 'str', 'type': 'str',
> +  'data': {'name': 'str', 'mountpoint': 'str', 'type': 'str', 'usage': 
> 'number',

...by instead marking this optional, as in '*usage'; and only setting 
has_usage=true when statvfs() succeeds.

Also, hitting exactly 80 columns already feels long; and if you go with 
my suggestion, you'd be lengthening it even more, so wrap the new member 
to the next line to keep length under 80.

--
Eric Blake, Principal Software Engineer
Red Hat, Inc.           +1-919-301-3266
Virtualization:  qemu.org | libvirt.org