cURL / Mailing Lists / curl-library / Single Mail


Re: TLS certificate verification

From: ( \\/\\/|||\ <>
Date: Sun, 21 Jun 2015 19:07:38 -0300

You seriously don't know how to get the full data response off the
socket??? And you claim libcurl is "the most powerful on the planet" ???

It's not just on SEARCH ALL, it's ANY search parameter that returns a large
response. No search command = *no imap support* dude... just like no
multiple recipients = *no smtp support* ... stop claiming shit that isn't

No time to detail every bug I find in libcurl, If you want the binary, let
me know. Anyway, I'm done w/libcurl, goodbye.

On Sun, Jun 21, 2015 at 3:41 AM, Ray Satiro via curl-library <> wrote:

> On 6/20/2015 7:28 PM, (( \/\/|||"'""/'")) ((\"""" )) (( ))\\\"\\"\ wrote:
>> 1. I have found conditions on custom IMAP command to get thousands of
>> envelopes in one FETCH where callback function tried to allocate huge
>> amounts of memory, more than response size, making Callback return -1, and
>> curl_easy_perform return something like "cannot write file" or similar.
>> Can't remember exact error...
>> 2. When issuing "select inbox" + "search unseen", on 1 mail inbox the
>> data returned in body/header is ALWAYS wrong. Happens consistently on 1
>> inbox has probably 9 or 10k unseen mails. It's a MAJOR error here and makes
>> libcurl custom command responses 100% unreliable. I attached some pictures
>> to better describe and show libcurl incorrect response & OpenSSL correct
>> response. The response should be many thousands of UID's and libcurl cuts
>> it down to several hundred and concats a UID.
>> 3. I found a way to make libcurl do an unhandled exception crash on
>> setting a custom command with a 300 byte string, although inconsistent
>> happens when sending/recieving many commands back to back. I spent a few
>> hours debugging it to try to do an exploit but I couldn't get control of
>> RIP so I don't think it's exploitable, but at best it's a denial of
>> service. Crash happens on curl_easy_setopt(CUSTOM_COMMAND, "\x41 * 300") -
>> dies when calling curl_easy_perform(). I have my clients code in there and
>> it's a lot so I can't send source but when I have more free time, if
>> needed, I could send you a binary to show the crash. It's definitely in
>> libcurl though.
> Point 2 is definitely a serious bug [1] and it is listed in known bugs
> [2]. I don't know of a good (or even a not-so-good) way to fix it. Points 1
> & 3 are either bugs or user error there isn't enough information to tell.
> You'd need to provide a self contained source example with the minimum
> amount of code needed to reproduce, no binary.
> [1]:
> [2]:
> -------------------------------------------------------------------
> List admin:
> Etiquette:

List admin:
Received on 2015-06-22