Re: missing privsep in ckqueue function

2016-02-29 Thread Chris Bennett
On Mon, Feb 29, 2016 at 10:05:03AM -0700, Todd C. Miller wrote: > On Mon, 29 Feb 2016 09:55:45 -0700, "Todd C. Miller" wrote: > > > Most of the PRIV_START / PRIV_END should be removed. There are a > > few instances where we need to drop setgid when opening files, > > however. Removing those call

Re: missing privsep in ckqueue function

2016-02-29 Thread Todd C. Miller
On Mon, 29 Feb 2016 09:55:45 -0700, "Todd C. Miller" wrote: > Most of the PRIV_START / PRIV_END should be removed. There are a > few instances where we need to drop setgid when opening files, > however. Removing those calls needs to be done very carefully. It is also worth rethinking whether lp

Re: missing privsep in ckqueue function

2016-02-29 Thread Todd C. Miller
On Mon, 29 Feb 2016 09:48:32 -0700, Theo de Raadt wrote: > PRIV_START / PRIV_END is not privsep by any means. It is the > old cron-style "drop id, do action, regain id" model. Most of the PRIV_START / PRIV_END should be removed. There are a few instances where we need to drop setgid when openin

Re: missing privsep in ckqueue function

2016-02-29 Thread Theo de Raadt
PRIV_START / PRIV_END is not privsep by any means. It is the old cron-style "drop id, do action, regain id" model. I don't think you understand what is being done here. > I have a diff out there right now on these files but I noticed the > following: > > > /* > * Scan the current directory an

missing privsep in ckqueue function

2016-02-29 Thread Chris Bennett
I have a diff out there right now on these files but I noticed the following: /* * Scan the current directory and make a list of daemon files sorted by * creation time. * Return the number of entries and a pointer to the list. */ int getq(struct queue ***namelist) { struct dirent *d;