User Agent: Mozilla/5.0 (Windows NT 6.1) AppleWebKit/535.8 (KHTML, like Gecko) Chrome/17.0.942.0 Safari/535.8 Steps to reproduce: Well, i'm reporting only because, this can do spam. But, risk to security don't have any. Guys, if you want fix, there are the demo: https://firstname.lastname@example.org
Yeah, this is pretty minor. But we should still hash-token it the same way we did for createaccount.
Created attachment 598315 [details] [diff] [review] patch - v1
Comment on attachment 598315 [details] [diff] [review] patch - v1 If someone accesses token.cgi directly, it will now get the "Suspicious Action" confirmation page (which I can live with, though I still don't think this bug must be fixed). But if the user clicks the confirmation button, it fails complaining that the page doesn't exist.
Please leave this bug alone. Just because a bug is not fixed immediately doesn't mean it will never be fixed. And if we decide it's minor enough to not be fixed, we will close it as wontfix, not invalid.
Created attachment 649151 [details] [diff] [review] patch, v2 The reason of the error I mentioned in comment 3 is because both global/confirm-action.html.tmpl and account/auth/login-small.html.tmpl use a variable named "script_name", and so login-small.html.tmpl was overwritting the value passed to confirm-action.html.tmpl. The fix is to simply rename the variable in login-small.html.tmpl to something else.
Comment on attachment 649151 [details] [diff] [review] patch, v2 r=reed works as expected, and direct access to token.cgi with email@example.com&a=reqpw shows the suspicious action page, and I am able to continue passed it fine.
I will have to backport the patch to 4.2 as token.cgi is very different between 4.2 and 4.4.
Created attachment 649311 [details] [diff] [review] patch for 4.2, v1 Backport for 4.2. No code change, only fixed a conflict in token.cgi.
Comment on attachment 649311 [details] [diff] [review] patch for 4.2, v1 Technically, you're doing the check in token.cgi in a different spot than trunk, but not sure it really matters. Though, I suspect it makes more sense to get the super-basic checks done (like valid e-mail address format) before doing token check (which can take more perf), but whatever...
Committing to: bzr+ssh://firstname.lastname@example.org/bugzilla/trunk/ modified token.cgi modified template/en/default/account/auth/login-small.html.tmpl modified template/en/default/account/auth/login.html.tmpl Committed revision 8330. Committing to: bzr+ssh://email@example.com/bugzilla/4.2/ modified token.cgi modified template/en/default/account/auth/login-small.html.tmpl modified template/en/default/account/auth/login.html.tmpl Committed revision 8114.