Neither one of those instructions would help. I have to take the IP
address and NC it with a mask (say, x'FFFF0000') to get a more generic
ip address.
Tony Thigpen
-----Original Message -----
From: Zierdt Richard A
Sent: 05/29/2012 01:19 PM
A relatively new machine instruction, FIND LEFTMOST ONE (FLOGR R1,R2) may be
helpful here (if the extended-immediate facility is available).
FLOGR returns in R1 the bit position (0-63) of the first one-bit encountered in
a 64-bit register (R2), left-to-right.
If the register is zero, then 64 (40 hex) is returned. There's a little more to
this, described in the POP manual.
Another instruction, perhaps less likely to help in this case, is POPULATION
COUNT (POPCNT R1,R2), which counts the number of one-bits in each of the eight
bytes of a 64-bit register. Essentially, it returns eight sums.
-----Original Message-----
From: IBM Mainframe Assembler List [mailto:[email protected]] On
Behalf Of robin
Sent: Tuesday, May 29, 2012 8:00 AM
To: [email protected]
Subject: Re: Generating a bit mask
From: Tony Thigpen
Sent: Tuesday, 29 May 2012 12:51 PM
I am working with some IPv6 addresses that are 128 bits long. The input
to the program is an ip name or address and the number of high bits
that I need to use for a compare against another ip address.
So, I want to create run time routine to build the a 16 byte 'and' bit
field based on the input number of bits. (I will then 'and' both
addresses and then use a simple CLC.)
A good way is to determine the bit start position.
Put 8 bits in a GPR. If shift is 1 thru 7, shift right the required no. of
bits.
Then store that.
For the end byte, do the same, only shift left.
For the bytes between the initial byte and end byte, store zero or more
characters (each of 8 one-bits).
Although a table built at compile time might be a reasonable solution,
How so? You say that the number of bits is input -- i.e. not a compile-time
value.
I thought this would be a good time to learn a new trick if someone
has one they can share.
