Re: net: integer overflow in ip_idents_reserve

On 12/16/2014 06:09 PM, Hannes Frederic Sowa wrote:

On Tue, Dec 16, 2014, at 22:47, Eric Dumazet wrote:

quoted

quoted

On Tue, 2014-12-16 at 16:19 -0500, Sasha Levin wrote:

quoted

quoted

Hi Eric,

While fuzzing with trinity on a -next kernel with the undefined behaviour
sanitizer path, I've observed the following warning in code which was
introduced in 04ca6973f7 ("ip: make IP identifiers less predictable"):

This is a false positive.

Also we compile the whole kernel with -fno-strict-overflow, so every
report of signed overflow leading to undefined behavior is probably a
false positive. I don't know if it is worth to try to get rid of them, I
doubt it.

I reported this one because there's usually some code to handle overflow
in code that expects that and here there was none (I could see).

For example, the ntp code had a few cases where a user could generate
overflows and mess up quite a few things (he got what he asked for -
problems).


Thanks,
Sasha