DragonFly commits List (threaded) for 2005-08
Re: cvs commit: src/sys/netinet6 raw_ip6.c
:Content-Type: text/plain; charset=US-ASCII; format=flowed
:On 11.08.2005, at 19:31, Matthew Dillon wrote:
:> Same with this one. It is unclear that the malloc should be
:> allowed to fail at all, M_INTWAIT should probably be used instead
:> of M_NOWAIT.
:Checking all code actually using inp->in6p_icmp6filt showed me that
:other places actually accept it to be NULL. Thus the
:ICMP6_FILTER_SETPASSALL was the only one unconditionally using the
Ok, but you also have to look at it from an operational point
e.g. when a user loads a module, they don't expect the load to
fail because an M_NOWAIT malloc failed.
In this case it appears to be handling a filter command
(initiated by a user ?). Not sure, but it sure sounds like
something a user would not expect to fail.