2 ms·
CVE's aren't free. The aftermath of issuing a CVE for this bug would have massive consequences all over the industry, fixing the bug and leaving the past - whe
by jacquesm 3y ago
CVE's aren't free.
The aftermath of issuing a CVE for this bug would have massive consequences all over the industry, fixing the bug and leaving the past - where you probably would have become aware of this by now if it affected you and you probably should have a look at your code regardless to see if the situation can occur - means that on the next scheduled update of glibc everybody will have the fix. If you have a way of getting this to escalate to RCE or some other nastiness, especially if it can be done remotely in some application that is networked and that uses qsort (which should be most of them so if it is that easy then it should be quick to prove).
Note that even something as mundane as the UTF-8 encoding scheme led to an exploit so it may well be possible. But I think classifying this as a security bug rather than just as a bug at this point is premature.
Consider being on the receiving side of a mandatory (for instance: regulated industry) upgrade requirement of all of your systems because a CVE was issued would carry substantial costs even if the risk of those systems being exploited was very small. So the glibc team likely took such costs into account and absent an easy or obvious way to exploit this that seems for the moment to be the right call. But it could change.
- n_plus_1_acc 3y agoRequiring no CVE is obviously a proxy for no vulnerabilities, and it's as close as you can get. I think if you are in an industry that values this, the correct thing to do is check if you use qsort and check your comparision functions. Now. To encourage this, it would be good thing for glibc to issue a CVE.