Hi All. We've had a similar issue to the one ment...
# cfml-general
n
Hi All. We've had a similar issue to the one mentioned above - ajax calls to cfcs breaking in a few cases. We're testing but my question is whether the jvm argument fix (Dcoldfusion.runtime.remotemethod.matchArguments=false) can be handled one a site by site basis (similar to another recent upgrade where the flag could be set in Application.cfc). So, we'd ideally like to use the latest update as is, but then use a flag on a site by site basis to role back this one arguments matching item on specific sites. Anyone try that?
s
it being a jvm argument I think limits you to applying it to all sites under the server
I haven’t seen any docs around an application var for it like they did for the implicit scope search
though reading the notes it seems it will only trigger if you overload the function
is that something you have implemented?
@carehart is the person to ask for the minutiae
or @foundeo as well
(apologies @carehart for tagging you directly)
n
@salted Thanks for the response. I think we got it sorted out.
s
oh great, can you share what fixed it?
I’m about to apply updates to several servers (2023) so if I encounter it would be good to know the fix/workaround
c
@salted, no worries. @nickg no, the workaround can't be applied at the app level. There are only 2 solutions (for now). First, yes the change in the update is what's breaking ajax or other remote method calls into cfc remote methods...but in ways beyond what Adobe indicates in the technote. Besides it affecting args you EXPLICITLY pass in, the problem is that if you have any url vars defined in the code (including application.cfc/cfm), those are (and it turns out always have been) passed in IMPLICITLY on any such remote method calls. As such, the problem can (for now) be fixed only by that jvm arg or by defining any such url vars that are being implicitly passed in. I discuss this also in my blog post on the update.
s
@carehart thanks for your input I immediately thought that you’d have already addressed it so glad I tagged you
is it only the URL scope that impacts this change?
(I’ll go and read your blog post in case I’ve asked a moot q)
c
That's all I've found so far (as it's all anyone else had identified). I'd not yet considered testing other scopes. Good thinking.
s
@carehart if it’s not too much to impose can you confirm if this is a critical update for all systems or is the critical fix something that applies to specific use cases i.e. if you use this then it’s critical otherwise you can apply in good time
I only ask because the previous couple updates have only been critical if you have X implemented
c
Adobe never makes that distinction, and there's not enough info to know from the apsb. That said, it currently indicates that there's no known exploits on the wild. Their words.
s
hmm so usual story of not enough detail 😕
I always go to the CVEs when they quote them to get the proper idea of severity
but you can’t rely on it not being seen in the wild either as that’s the nature of things like zero days
c
I'd not found them offering much, either. Things can change over time, of course
s
I am as always supportive of your efforts for clarity, lol
this latest update is a quandary for me too as I have very few clients that use methods with remote so I have to balance my obligation to update promply with the time needed to fix the issue if I don’t use the jvm arg
not to mention your caveat of cfajaxproxy
c
Thanks. I'll add also that I'd looked to the blog of Brian Reilly (hoyahaxa.com), who is one of the acknowledged identifiers of the vulns addressed. He also has nothing on these yet.
s
Oh that’s a name I’ve not heard for some time I didn’t know he was still active in that area, thanks for that
As of today I only really follow yourself and Pete when new updates are released
c
Btw, it's ANY calls in, not JUST those generated by cfajaxproxy. I'd updated my post earlier today, if you saw it previously
s
I get a red warning with any update only to downgrade it when I dig in to the notes and find out it applies to situations when you use x feature
It’s 2:30am here now so I’ve bookmarked your blog post for tomorrow - I’ve got as far as the mysql connector warning. That being said, the adobe team haven’t fixed an issue I raised previously of the updates adding in files to the jetty folder even if you don’t have some features installed which then breaks your install so
c
To be clear, this update did address some issue related to jetty, so keep hope alive
s
Oh well wonders will never cease!
I can’t find my previous message about it due to the slack subscription limit so I might have to boot up a sandbox to replicate it as it’s one of those issues that you will encounter, fix and forget to tell anyone that you got it and how you fixed it
c
👍
(I've confirmed that it is indeed only url vars that are being passed implicitly into remote method calls, not form or cookie. And as I discuss in my post, the fact cf implicitly passes in url vars to such remote methods is NOT new. It's just new that this change trips over that, if they're not defined as method arguments.)