Has anyone else noticed their umask being set to 022 after an exec() in
recent kernels?  I see this with the i386 port, and am wondering if it's
a problem local to me, or if it's simply that nobody's noticed it yet.

I first started noticing this on 8 June thanks to a "make trip" failure
in the plan9/rc package, and didn't really chase it down hoping it would
go away after all the recent memory locking fixes etc were in place.

I'm willing to accept that it may be a compilation weirdness, since I
build with -march=i686 -mfancy-math-387 -O3 and have had USE_EGCS_HAIFA=1
(and, unrelatedly, I387_LIBM?=yes) in my /etc/mk.conf for ages.   On another
box I use -march=pentium and the other stuff above; it sees the same symptoms.

Anyway, the symptoms are shown off below.

Is this PR-worthy?

	Sean.
- --
$ ls
$ umask 
0022
$ touch a
$ ls -l
total 0
-rw-r--r--  1 smd  wheel  0 Jun  8 17:35 a
$ umask 027
$ touch b
$ ls -l
total 0
-rw-r--r--  1 smd  wheel  0 Jun  8 17:35 a
-rw-r--r--  1 smd  wheel  0 Jun  8 17:35 b
$ echo > c
$ ls -l
total 1
-rw-r--r--  1 smd  wheel  0 Jun  8 17:35 a
-rw-r--r--  1 smd  wheel  0 Jun  8 17:35 b
-rw-r-----  1 smd  wheel  1 Jun  8 17:35 c
$ dd if=/dev/null of=e
0+0 records in
0+0 records out
0 bytes transferred in 1 secs (0 bytes/sec)
$ ls -l
total 1
-rw-r--r--  1 smd  wheel  0 Jun  8 17:35 a
-rw-r--r--  1 smd  wheel  0 Jun  8 17:35 b
-rw-r-----  1 smd  wheel  1 Jun  8 17:35 c
-rw-r--r--  1 smd  wheel  0 Jun  8 17:35 e
$ umask
0027
$ csh
smd% umask
22
smd% umask 027
smd% umask
27
smd% csh
smd% umask
22