Re: [PATCH v2 9/9] docs: secret: Unify and sanitize examples on how to set secret value

On Mon, Jan 27, 2020 at 09:59:01 +0100, Peter Krempa wrote: > On Fri, Jan 24, 2020 at 10:41:00 -0600, Eric Blake wrote: > > On 1/24/20 10:08 AM, Peter Krempa wrote: > > > Discourage passing secrets as commandline arguments. > > > > > > Signed-off-by: Peter Krempa <pkrempa(a)redhat.com&gt; > > > --- > > > docs/formatsecret.html.in | 88 +++++++++++++++++++++++++-------------- > > > 1 file changed, 57 insertions(+), 31 deletions(-) > > > > > > > > > > > + <h2><a id="settingSecrets">Setting secret values in virsh</a></h2> > > > + > > > > > + </pre> > > > + > > > + <p> > > > + The secret can also be set via an argument, but note that other users > > > + may see it in the process listing output. The secret must be base64 > > > + encoded. > > > > Is this last sentence still accurate, given that you can pass --plain to > > avoid base64 encoding? > > I didn't allow to use --plain together with passing it on the command > line. --plain works only in conjunction with --file. I didn't see a > point of adding new features to an insecure way of doing things. > > > Should the note use <b> or other formatting to call attention to the > > security risk of doing it this way? > > Yeah, I can add it. > I propose the following change to emphasize the fact that it's insecure: diff --git a/docs/formatsecret.html.in b/docs/formatsecret.html.in index 6c2d5e02a6..695370e150 100644 --- a/docs/formatsecret.html.in +++ b/docs/formatsecret.html.in @@ -398,9 +398,10 @@ Secret value set </pre> <p> - The secret can also be set via an argument, but note that other users - may see it in the process listing output. The secret must be base64 - encoded. + <b>WARNING</b>The following approach is <b>insecure</b> and deprecated. + The secret can also be set via an argument. Note that other users may see + the actual secret in the process listing! + The secret must be base64 encoded. </p>