That works for SET, but didn't the OP also want TEST? The "Byte Alignment Feature" used to incur a performance hit; that may no longer be the case.
-- Shmuel (Seymour J.) Metz http://mason.gmu.edu/~smetz3 ________________________________________ From: IBM Mainframe Assembler List [[email protected]] on behalf of Charles Mills [[email protected]] Sent: Thursday, May 20, 2021 6:30 PM To: [email protected] Subject: Re: Macro to set a bit string No, I mean O into the target field; O into the appropriate byte of the bit array. O no longer requires alignment, right? So you just set the appropriate bit in positions 0-7 of the register and zero the remaining positions and O it against the target field. There is a three-byte "hangover." It won't hurt anything unless you happen to be at the end of a protection key or addressability boundary. That is what I meant by "if one could tolerate the storage reference 3 bytes beyond ..." It would be easy enough in many cases to assure three bytes of padding. No bits would ever be altered in those 3 bytes, but the hardware would see a storage reference. I suppose in the boundary case it could cause a cache miss. Charles -----Original Message----- From: IBM Mainframe Assembler List [mailto:[email protected]] On Behalf Of Seymour J Metz Sent: Thursday, May 20, 2021 2:42 PM To: [email protected] Subject: Re: Macro to set a bit string If you mean an O into the immediate field of a TM, that will wreak havoc with the pipeline, and will break when you call the routine a second time unless the bit offset is the same. "Things should be as simple as possible, but no simpler." -- Shmuel (Seymour J.) Metz http://mason.gmu.edu/~smetz3 ________________________________________ From: IBM Mainframe Assembler List [[email protected]] on behalf of Charles Mills [[email protected]] Sent: Thursday, May 20, 2021 4:32 PM To: [email protected] Subject: Re: Macro to set a bit string The BitPos table is clever. You could use that with an OC and get rid of the EX. No need to XR any of those registers. LAY and SRL 29 will clear R2 and R3, and high order bits are irrelevant to EX. It also occurred to me that if one could tolerate the storage reference (no alteration) 3 bytes beyond the target byte one could use O (Or register to storage) and avoid EX. Charles -----Original Message----- From: IBM Mainframe Assembler List [mailto:[email protected]] On Behalf Of Mike Hochee Sent: Thursday, May 20, 2021 12:58 PM To: [email protected] Subject: Re: Macro to set a bit string Hi João, The following (and it may have probs, as it was not tested), simply turns on the nth bit within an array. It could be changed to turn off the bit or perform some other bit boolean operation. You shouldn't need to iterate through array bytes or bits, as that would certainly slow things down. I tried to follow Charles' suggestions, but it is very rough and needs checks for out of bounds conditions, array size considerations, etc.. HTH, Mike * XR R2,R2 XR R3,R3 XR R4,R4 * LAY R2,12 <=== Input bit to turn on (rel to 0) SRDL R2,3 /8 to get bytes to bypass SRL R3,29 shift remainder into pos 61-63 IC R4,BitPos(R3) Prime R4 w/OI immediate value LA R5,BitArray(R2) A(target byte) in BitArray EX R4,OrBit Sets selected bit on * OrBit OI 0(R5),x'00' Set selected bit on in BitArray * BitArray DC XL256'00' MaxBits EQU L'BitArray*8 Max bits in array (rel to 0) * BitPos EQU * Ident bit in target byte Bit_0 DC BL1'10000000' Bit_1 DC BL1'01000000' Bit_2 DC BL1'00100000' Bit_3 DC BL1'00010000' Bit_4 DC BL1'00001000' Bit_5 DC BL1'00000100' Bit_6 DC BL1'00000010' Bit_7 DC BL1'00000001' *
