[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [libmicrohttpd] NTLM request loops when CURLOPT_FORBID_REUSE is set

From: Christian Grothoff
Subject: Re: [libmicrohttpd] NTLM request loops when CURLOPT_FORBID_REUSE is set
Date: Tue, 17 Jun 2014 12:04:07 +0200
User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:24.0) Gecko/20100101 Icedove/24.5.0

Dear Frank,

This is not the cURL mailinglist ;-).  Also, cURL (as opposed to supports about a dozen crypto backends,
so your problems may depend on the specific backend you had the
misfortune of linking against.

I have in the past experienced that some SSL libraries do not
play nice with certain other SSL libraries under certain
cipher configurations.  So when you report such issues, you
should always be very specific as to which crypto libraries
with which cipher settings were involved on both sides. Often
the HTTP libraries (client/server) are not to blame, and
diagnosing such issues without detailed knowledge of the full
stack can be rather difficult.

Happy hacking!


On 06/17/2014 11:58 AM, Frank Meier wrote:
> Hi
> In our Application we normally do requests without HTTP keepalive
> Now when we use NTLM this does not work anymore with this settings. I
> expected that libcurl would use the same connection for the NTLM
> authentication (type1) request and the following "real" (type3) request
> and then would drop the connection.
> The problem then is, that libcurl is starting to loop indefinitely
> unless CURLOPT_TIMEOUT is configured:
> 1) Libcurl sends request with NTLM type1 header
> 2) Server sends 401 with NTLM type2 header
> 3) Libcurl drops connection
> 4) Libcurl sends again a request with NTLM type1 header (and so on...)
> When CURLOPT_FORBID_REUSE is set to 0 everything works fine.
> In my opinion the connection should not be dropped after the first
> request but one can argue, if CURLOPT_FORBID_REUSE is set, this is just
> what happens. But If so why does it still work when
> In any case I think libcurl should not start looping and should abort
> after the second request.
> BTW I tested with version 7.37.0 and actual checkout from git
> cheers Frank

Attachment: 0x48426C7E.asc
Description: application/pgp-keys

Attachment: signature.asc
Description: OpenPGP digital signature

reply via email to

[Prev in Thread] Current Thread [Next in Thread]