You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@roller.apache.org by Allen Gilliland <al...@sun.com> on 2007/04/19 22:22:30 UTC

removing cookieLogin check from struts2 profile form

I just wanted to send this out to make sure I am not wrong about this. 
In the new struts2 profile update form I have removed a small amount of 
code which tried to prevent users who were logged in via a 'cookieLogin' 
from changing their passwords.  I don't really know the history behind 
this code and don't even know if it was working anymore, but I am 
removing it because ...

1. It *looks* like it doesn't even take effect anymore because I can't 
find anywhere in the code which sets the "cookieLogin" attribute on a 
session to trigger the behavior.

2. The old code wasn't actually preventing this scenario, it was only 
discouraging it.  In the old action & jsp it would simply not display 
the password form fields when the user had logged in via cookie, but 
technically if you submitted the form with the right fields the action 
would accept them and modify the password.

3. If this is still a valid security constraint that we want to imply 
then I think we can do it in a better way, i.e. to ask for the users 
existing password on the profile update form at all times.  Personally I 
think this would be overkill, but it's not necessarily unreasonable.

So, if anyone who knows more about this old "cookieLogin" check thinks 
it still needs to be in the code, please say so.

-- Allen

Re: removing cookieLogin check from struts2 profile form

Posted by Allen Gilliland <al...@sun.com>.

Matt Raible wrote:
> On 4/19/07, Allen Gilliland <al...@sun.com> wrote:
>>
>>
>> Matt Raible wrote:
>> > The old code probably did some checking in the edit() method to see if
>> > the user logged in with Remember Me. If so, it set that value, which
>> > the JSP checked and hid the fields for.
>>
>> I looked at that, but all that I can see is a conditional in the edit()
>> method which adds an action message if it's a cookie login, and in the
>> jsp it doesn't show the password fields if it's a cookie login.  I
>> couldn't find anywhere in the code where that cookie login state is
>> actually set, the only thing i can think of is if it happens inside of
>> the Acegi code.
>>
> 
> That's weird - here's what I use in AppFuse in the UserAction.edit() 
> method:
> 
>    // if user logged in with remember me, display a warning that they
> can't change passwords
>    log.debug("checking for remember me login...");
> 
>    AuthenticationTrustResolver resolver = new
> AuthenticationTrustResolverImpl();
>    SecurityContext ctx = SecurityContextHolder.getContext();
> 
>    if (ctx != null) {
>        Authentication auth = ctx.getAuthentication();
> 
>        if (resolver.isRememberMe(auth)) {
>            getSession().setAttribute("cookieLogin", "true");
>            saveMessage(getText("userProfile.cookieLogin"));
>        }
>    }

the sruts1 yourProfile action is still untouched and there is no code 
like that in there.  there is none of that Acegi stuff about checking if 
the auth came from the remember me service, but it does look for the 
attribute "cookieLogin" in the session and set an action message if it 
exists.

in any case, i don't think that matters anyways.  the real question is, 
do we think there is a need to enforce greater security on the profile 
edit page to prevent unwanted folks from modifying a profile?  if we do 
then the most appropriate way to do that is to always ask for the users 
existing password before allowing a profile change.

my personal opinion is that this would be wasted effort.  if someone is 
somehow granted a session via the remember me service who is not the 
actual owner of that account then the security has already been 
breached, whether they can change the account password or not doesn't 
really matter.  currently the code would let that person ... change the 
account email address, post entries on the weblog, delete anything from 
the weblog, delete the entire weblog, invite other accounts to the 
weblog, and on and on.  the fact is that our security model is such that 
if someone is given a session we have to expect that session is properly 
authenticated.  we have no valid use cases where someone should be given 
a partially authenticated session.

-- Allen


> 
>> >
>> > A better solution for the JSP is to remove the fields completely, then
>> > use a Preparable interface on UserAction (or whatever its called) to
>> > refetch the user on save. That way, all the values from the database
>> > would be retained, and any ones specified in request parameters (from
>> > the form) would be overwritten.
>>
>> Well, we do lookup the user when a profile save happens and we only set
>> the properties which are allowed to be changed.  I don't think it's good
>> usability to remove the password fields though.  If someone is logged in
>> via a remember me cookie then they should be in equal standing with any
>> other logged in user.
> 
> I'm fine with that, but when I first started digging into the Remember
> Me stuff (years ago) a lot of folks recommended that it not be given
> the same status as a user who logs in with a keyboard.
> 
> Matt
> 
>>
>> -- Allen
>>
>>
>> >
>> > Matt
>> >
>> > On 4/19/07, Allen Gilliland <al...@sun.com> wrote:
>> >> I just wanted to send this out to make sure I am not wrong about this.
>> >> In the new struts2 profile update form I have removed a small 
>> amount of
>> >> code which tried to prevent users who were logged in via a 
>> 'cookieLogin'
>> >> from changing their passwords.  I don't really know the history behind
>> >> this code and don't even know if it was working anymore, but I am
>> >> removing it because ...
>> >>
>> >> 1. It *looks* like it doesn't even take effect anymore because I can't
>> >> find anywhere in the code which sets the "cookieLogin" attribute on a
>> >> session to trigger the behavior.
>> >>
>> >> 2. The old code wasn't actually preventing this scenario, it was only
>> >> discouraging it.  In the old action & jsp it would simply not display
>> >> the password form fields when the user had logged in via cookie, but
>> >> technically if you submitted the form with the right fields the action
>> >> would accept them and modify the password.
>> >>
>> >> 3. If this is still a valid security constraint that we want to imply
>> >> then I think we can do it in a better way, i.e. to ask for the users
>> >> existing password on the profile update form at all times.  
>> Personally I
>> >> think this would be overkill, but it's not necessarily unreasonable.
>> >>
>> >> So, if anyone who knows more about this old "cookieLogin" check thinks
>> >> it still needs to be in the code, please say so.
>> >>
>> >> -- Allen
>> >>
>> >
>> >
>>
> 
> 

Re: removing cookieLogin check from struts2 profile form

Posted by Matt Raible <ma...@raibledesigns.com>.
On 4/19/07, Allen Gilliland <al...@sun.com> wrote:
>
>
> Matt Raible wrote:
> > The old code probably did some checking in the edit() method to see if
> > the user logged in with Remember Me. If so, it set that value, which
> > the JSP checked and hid the fields for.
>
> I looked at that, but all that I can see is a conditional in the edit()
> method which adds an action message if it's a cookie login, and in the
> jsp it doesn't show the password fields if it's a cookie login.  I
> couldn't find anywhere in the code where that cookie login state is
> actually set, the only thing i can think of is if it happens inside of
> the Acegi code.
>

That's weird - here's what I use in AppFuse in the UserAction.edit() method:

    // if user logged in with remember me, display a warning that they
can't change passwords
    log.debug("checking for remember me login...");

    AuthenticationTrustResolver resolver = new
AuthenticationTrustResolverImpl();
    SecurityContext ctx = SecurityContextHolder.getContext();

    if (ctx != null) {
        Authentication auth = ctx.getAuthentication();

        if (resolver.isRememberMe(auth)) {
            getSession().setAttribute("cookieLogin", "true");
            saveMessage(getText("userProfile.cookieLogin"));
        }
    }

> >
> > A better solution for the JSP is to remove the fields completely, then
> > use a Preparable interface on UserAction (or whatever its called) to
> > refetch the user on save. That way, all the values from the database
> > would be retained, and any ones specified in request parameters (from
> > the form) would be overwritten.
>
> Well, we do lookup the user when a profile save happens and we only set
> the properties which are allowed to be changed.  I don't think it's good
> usability to remove the password fields though.  If someone is logged in
> via a remember me cookie then they should be in equal standing with any
> other logged in user.

I'm fine with that, but when I first started digging into the Remember
Me stuff (years ago) a lot of folks recommended that it not be given
the same status as a user who logs in with a keyboard.

Matt

>
> -- Allen
>
>
> >
> > Matt
> >
> > On 4/19/07, Allen Gilliland <al...@sun.com> wrote:
> >> I just wanted to send this out to make sure I am not wrong about this.
> >> In the new struts2 profile update form I have removed a small amount of
> >> code which tried to prevent users who were logged in via a 'cookieLogin'
> >> from changing their passwords.  I don't really know the history behind
> >> this code and don't even know if it was working anymore, but I am
> >> removing it because ...
> >>
> >> 1. It *looks* like it doesn't even take effect anymore because I can't
> >> find anywhere in the code which sets the "cookieLogin" attribute on a
> >> session to trigger the behavior.
> >>
> >> 2. The old code wasn't actually preventing this scenario, it was only
> >> discouraging it.  In the old action & jsp it would simply not display
> >> the password form fields when the user had logged in via cookie, but
> >> technically if you submitted the form with the right fields the action
> >> would accept them and modify the password.
> >>
> >> 3. If this is still a valid security constraint that we want to imply
> >> then I think we can do it in a better way, i.e. to ask for the users
> >> existing password on the profile update form at all times.  Personally I
> >> think this would be overkill, but it's not necessarily unreasonable.
> >>
> >> So, if anyone who knows more about this old "cookieLogin" check thinks
> >> it still needs to be in the code, please say so.
> >>
> >> -- Allen
> >>
> >
> >
>


-- 
http://raibledesigns.com

Re: removing cookieLogin check from struts2 profile form

Posted by Allen Gilliland <al...@sun.com>.

Matt Raible wrote:
> The old code probably did some checking in the edit() method to see if
> the user logged in with Remember Me. If so, it set that value, which
> the JSP checked and hid the fields for.

I looked at that, but all that I can see is a conditional in the edit() 
method which adds an action message if it's a cookie login, and in the 
jsp it doesn't show the password fields if it's a cookie login.  I 
couldn't find anywhere in the code where that cookie login state is 
actually set, the only thing i can think of is if it happens inside of 
the Acegi code.


> 
> A better solution for the JSP is to remove the fields completely, then
> use a Preparable interface on UserAction (or whatever its called) to
> refetch the user on save. That way, all the values from the database
> would be retained, and any ones specified in request parameters (from
> the form) would be overwritten.

Well, we do lookup the user when a profile save happens and we only set 
the properties which are allowed to be changed.  I don't think it's good 
usability to remove the password fields though.  If someone is logged in 
via a remember me cookie then they should be in equal standing with any 
other logged in user.

-- Allen


> 
> Matt
> 
> On 4/19/07, Allen Gilliland <al...@sun.com> wrote:
>> I just wanted to send this out to make sure I am not wrong about this.
>> In the new struts2 profile update form I have removed a small amount of
>> code which tried to prevent users who were logged in via a 'cookieLogin'
>> from changing their passwords.  I don't really know the history behind
>> this code and don't even know if it was working anymore, but I am
>> removing it because ...
>>
>> 1. It *looks* like it doesn't even take effect anymore because I can't
>> find anywhere in the code which sets the "cookieLogin" attribute on a
>> session to trigger the behavior.
>>
>> 2. The old code wasn't actually preventing this scenario, it was only
>> discouraging it.  In the old action & jsp it would simply not display
>> the password form fields when the user had logged in via cookie, but
>> technically if you submitted the form with the right fields the action
>> would accept them and modify the password.
>>
>> 3. If this is still a valid security constraint that we want to imply
>> then I think we can do it in a better way, i.e. to ask for the users
>> existing password on the profile update form at all times.  Personally I
>> think this would be overkill, but it's not necessarily unreasonable.
>>
>> So, if anyone who knows more about this old "cookieLogin" check thinks
>> it still needs to be in the code, please say so.
>>
>> -- Allen
>>
> 
> 

Re: removing cookieLogin check from struts2 profile form

Posted by Matt Raible <ma...@raibledesigns.com>.
The old code probably did some checking in the edit() method to see if
the user logged in with Remember Me. If so, it set that value, which
the JSP checked and hid the fields for.

A better solution for the JSP is to remove the fields completely, then
use a Preparable interface on UserAction (or whatever its called) to
refetch the user on save. That way, all the values from the database
would be retained, and any ones specified in request parameters (from
the form) would be overwritten.

Matt

On 4/19/07, Allen Gilliland <al...@sun.com> wrote:
> I just wanted to send this out to make sure I am not wrong about this.
> In the new struts2 profile update form I have removed a small amount of
> code which tried to prevent users who were logged in via a 'cookieLogin'
> from changing their passwords.  I don't really know the history behind
> this code and don't even know if it was working anymore, but I am
> removing it because ...
>
> 1. It *looks* like it doesn't even take effect anymore because I can't
> find anywhere in the code which sets the "cookieLogin" attribute on a
> session to trigger the behavior.
>
> 2. The old code wasn't actually preventing this scenario, it was only
> discouraging it.  In the old action & jsp it would simply not display
> the password form fields when the user had logged in via cookie, but
> technically if you submitted the form with the right fields the action
> would accept them and modify the password.
>
> 3. If this is still a valid security constraint that we want to imply
> then I think we can do it in a better way, i.e. to ask for the users
> existing password on the profile update form at all times.  Personally I
> think this would be overkill, but it's not necessarily unreasonable.
>
> So, if anyone who knows more about this old "cookieLogin" check thinks
> it still needs to be in the code, please say so.
>
> -- Allen
>


-- 
http://raibledesigns.com