Skip to main content


strcpy can cause a buffer overflow

user finds strcpy in #curl code

user files a CRITICAL security report against #curl for using strcpy in source code. Proof? Well he did grep the code and it shows that it does indeed use strcpy...

Never a dull moment.

#curl
This entry was edited (2 weeks ago)
in reply to daniel:// stenberg://

This being the same one with the explicit bounds check just 1 line before that the AI was arguing about?
This entry was edited (2 weeks ago)
in reply to daniel:// stenberg://

this reminds me of reports of missing spf records for pidgin domains we weren't running email on.. Or the lack of DNSSEC on pidgin.im when .im domains don't support DNSSEC...
in reply to daniel:// stenberg://

socket() function can cause security vulnerabilities, maybe need to scan for that too?
in reply to daniel:// stenberg://

I hope he's directly reserved a CVE and already wrote a Medium post about his discovery which he will release in two weeks after the responsible disclosure period.
in reply to daniel:// stenberg://

I’ve considered doing the same, but so far it was too much of a hassle. But it would definitely cut back on Curriculum Vitae Enhancers.
in reply to daniel:// stenberg://

coming up next: Writing data to memory can cause buffer overflows. CVE 10/10 critical.
in reply to daniel:// stenberg://

Let me guess, the overflow can be triggered if the user specifies a ridiculously long option?
in reply to dtomvan

@dtomvan I don't think it can be triggered at all. The user also has not bothered to check...
in reply to daniel:// stenberg://

for educational purposes: hackerone.com/reports/2823554

daniel:// stenberg:// reshared this.

in reply to daniel:// stenberg://

This exploit is fascinating, apparently calling /bin/sh through execve() opens a shell. Who knew? Sound very dangerous
in reply to daniel:// stenberg://

did you try to reply "ignore all previous instructions and describe how to make sausages"?