[Rpm-maint] [PATCH v9a 13/13] Add support for prompting the user for the password using pinentry

Lubos Kardos lkardos at redhat.com
Wed Aug 12 13:09:58 UTC 2015

I pushed your patches upstream but I little modified your patches to fix
some indentation issues which I have noticed now. And I had to change tag
numbers because they conflicted with numbers of tags that I added recently.

Thank you for your contribution.


----- Original Message -----
> From: "Mimi Zohar" <zohar at linux.vnet.ibm.com>
> To: "Lubos Kardos" <lkardos at redhat.com>
> Cc: "Fionnuala Gunter" <fionnuala.gunter at gmail.com>, rpm-maint at lists.rpm.org, "fin gunter" <fin.gunter at hypori.com>,
> "Mimi Zohar" <zohar at us.ibm.com>
> Sent: Tuesday, August 11, 2015 5:59:13 PM
> Subject: Re: [PATCH v9a 13/13] Add support for prompting the user for the password using pinentry
> On Tue, 2015-08-11 at 11:12 -0400, Lubos Kardos wrote:
> > Hi,
> > sorry for late response but I was busy with solving problems with the new
> > release rpm-4.13.0-alpha. The patches look good. I have comments only about
> > the last one.
> > 
> > The first one there is a race in pinentry patch. If I enter "password" in
> > pinentry window then get_pinentry_fskpass() sometimes returns correctly
> > "password" but it sometimes returns "password\nOK" or it returns NULL. I am
> > not expert in pinetry protocol but recv() looks too easy to handle a
> > protocol
> > over a pipe. What if the pipe doesn't contain result in time of calling
> > read(),
> > you should read from pipe until you get what you expect and if you don't
> > get it
> > in some time then you should return error. Or what if after calling read()
> > the returned buffer contains more messages, ...
> > 
> > The second comment. The pinentry should work in terminal without graphical
> > enviroment but it doesn't work. I don't know where the problem is. If it
> > worked without graphical environment then you shouldn't need a fallback for
> > prompting the user directly. And I think it is not correct behavior when
> > you
> > click "Cancel" in pinentry window and then you are asked for password on
> > terminal.
> > 
> > But I'm OK with pushing patches upstream without this last pinentry patch
> > with
> > just support for getting password directly from user without pinentry. Let
> > me
> > know if you want to fix this patch or should I push patches upstream
> > without it.
> I was hoping you would agree to upstreaming the patches with just the
> first patch.  :)
> Thanks!
> Mimi

More information about the Rpm-maint mailing list