Re: Addressing Scheme with 64 vs 63 bits
In [EMAIL PROTECTED], on 10/21/2008 at 05:30 PM, Scott Ford [EMAIL PROTECTED] said: Starting writing Assembler on a 360/20,, man 704 days... I won't try to justify the 704, but one (IBMAP) of the assemblers on, e.g., the 709, 7090, had a facility (QUAL) that HLA still doesn't have. And the convert instructions were cool. -- Shmuel (Seymour J.) Metz, SysProg and JOAT ISO position; see http://patriot.net/~shmuel/resume/brief.html We don't care. We don't have to care, we're Congress. (S877: The Shut up and Eat Your spam act of 2003) -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Addressing Scheme with 64 vs 63 bits
Robert A. Rosenberg wrote: There was also the secondary use of the high bit to signal AM24 vs AM31 in addresses used for branching to/from subroutines. This required replacing BALR with BASR and BR with BSM to do the AM Mode Switch. And, the low-order bit of the address now indicates switching to 64-bit mode using pointer-defined linkage. This has caused a number of software issues for code that tried to process those pointers using other instructions without first clearing bit 63. -- Edward E Jaffe Phoenix Software International, Inc 5200 W Century Blvd, Suite 800 Los Angeles, CA 90045 310-338-0400 x318 [EMAIL PROTECTED] http://www.phoenixsoftware.com/ -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Addressing Scheme with 64 vs 63 bits
The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main as well. [EMAIL PROTECTED] (Robert A. Rosenberg) writes: There was also the secondary use of the high bit to signal AM24 vs AM31 in addresses used for branching to/from subroutines. This required replacing BALR with BASR and BR with BSM to do the AM Mode Switch. 360/370 24bit psw had ILC/CC and program mask in adjacent byte; BAL/BALR not only saved the 24bit instruction address but also the next byte of the PSW (cc, and program mask). on return, not only could the calling/return address be restored, but SPM instruction would also be used to restore the program mask ... aka from principle of ops SPM programming notes: http://publibz.boulder.ibm.com/cgi-bin/bookmgr_OS390/BOOKS/dz9zr003/7.5.113?DT=20040504121320 1. Bits 34-39 of the general register may have been loaded from the PSW by execution of BRANCH AND LINK in the 24-bit addressing mode or by execution of INSERT PROGRAM MASK in either the 24-bit or 31-bit addressing mode. 2. SET PROGRAM MASK permits setting of the condition code and the mask bits in either the problem state or the supervisor state. 3. The program should take into consideration that the setting of the program mask can have a significant effect on subsequent execution of the program. Not only do the four mask bits control whether the corresponding interruptions occur, but the exponent-underflow and significance masks also determine the result which is obtained. ... snip .. BAS/BASR were introduced on 360/67 as part of supporting 32bit virtual addressing mode. retrenching to 370 ... not only was 360/67 32bit virtual addressing dropped ... but also the channel controller for multiprocessor support ... standard 360/67 multiprocessor not only allowed all processors to address all real storage but also all channels. standard 360 (and later 370) multiprocessor support only allowed two processors to address all of the (same) real storage ... but each processor was limited to only addressing their own, dedicated channels. some of the 360/67 control registers were also used to sense the switches on the channel controller (which governed the multiprocessor configuration settings ... not only for channels but also for real storage) ... these control register definitions were later taken over for access registers -- 40+yrs virtualization experience (since Jan68), online at home since Mar70 -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Addressing Scheme with 64 vs 63 bits
On Mon, 20 Oct 2008 22:00:00 -0500, Paul Gilmartin [EMAIL PROTECTED] wrote: ... I thought 31-bit was for compatibility with existing exploitation of the sign bit; ... Nope, it was the use of a X'80' in the high-order byte of a fullword to terminate a variable-length parameter list (of fullwords). Nope? Where do you perceive us to be in disagreement? ... There's a slight semantic difference. Sign vs. flag. Two different uses of the same high order bit, neither of which is compatable with a 32 bit address. Pat O'Keefe -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Addressing Scheme with 64 vs 63 bits
Right, but I think it's fairly clear that they were both talking about the same thing. Date: Tue, 21 Oct 2008 14:16:43 -0500 From: [EMAIL PROTECTED] Subject: Re: Addressing Scheme with 64 vs 63 bits To: IBM-MAIN@BAMA.UA.EDU On Mon, 20 Oct 2008 22:00:00 -0500, Paul Gilmartin [EMAIL PROTECTED] wrote: ... I thought 31-bit was for compatibility with existing exploitation of the sign bit; ... Nope, it was the use of a X'80' in the high-order byte of a fullword to terminate a variable-length parameter list (of fullwords). Nope? Where do you perceive us to be in disagreement? ... There's a slight semantic difference. Sign vs. flag. Two different uses of the same high order bit, neither of which is compatable with a 32 bit address. Pat O'Keefe _ Stay organized with simple drag and drop from Windows Live Hotmail. http://windowslive.com/Explore/hotmail?ocid=TXT_TAGLM_WL_hotmail_102008 -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Addressing Scheme with 64 vs 63 bits
On Tue, 21 Oct 2008 15:38:56 -0400, J R wrote: Right, but I think it's fairly clear that they were both talking about the same thing. ... and I chose the term simply because I can never remember which systems number bits left-to-right, and which right to left, or start with 0, or start with 1. And I think the IBM 704 didn't even number the sign bit -- it was simply S, and didn't (sometimes) participate in shift instructions. Date: Tue, 21 Oct 2008 14:16:43 -0500 From: [EMAIL PROTECTED] Nope? Where do you perceive us to be in disagreement? ... There's a slight semantic difference. Sign vs. flag. Two different uses of the same high order bit, neither of which is compatable with a 32 bit address. -- gil -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Addressing Scheme with 64 vs 63 bits
Gil -- same here...but I don't get hung up on semantics..I translate flag/sign when speaking in assembler-ese... Regards, Scott Ford Senior Systems Engineer [p] 678.266.3399 x304[m] 609-346-0399 identityforge.com This message is for the designated recipient only and may contain privileged, proprietary, or otherwise private information. If you have received it in error, please notify the sender immediately or let us know at [EMAIL PROTECTED] or [EMAIL PROTECTED], and then delete the original. Any other use of the email by you is prohibited. -Original Message- From: IBM Mainframe Discussion List [mailto:[EMAIL PROTECTED] On Behalf Of Paul Gilmartin Sent: Tuesday, October 21, 2008 4:25 PM To: IBM-MAIN@BAMA.UA.EDU Subject: Re: Addressing Scheme with 64 vs 63 bits On Tue, 21 Oct 2008 15:38:56 -0400, J R wrote: Right, but I think it's fairly clear that they were both talking about the same thing. ... and I chose the term simply because I can never remember which systems number bits left-to-right, and which right to left, or start with 0, or start with 1. And I think the IBM 704 didn't even number the sign bit -- it was simply S, and didn't (sometimes) participate in shift instructions. Date: Tue, 21 Oct 2008 14:16:43 -0500 From: [EMAIL PROTECTED] Nope? Where do you perceive us to be in disagreement? ... There's a slight semantic difference. Sign vs. flag. Two different uses of the same high order bit, neither of which is compatable with a 32 bit address. -- gil -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Addressing Scheme with 64 vs 63 bits
Paul Gilmartin wrote: ... and I chose the term simply because I can never remember which systems number bits left-to-right, and which right to left, or start with 0, or start with 1. And I think the IBM 704 didn't even number the sign bit -- it was simply S, and didn't (sometimes) participate in shift instructions. It's simple, the bits on the 360, et seq. are numbered in the same direction as on the 704, etc. And I really missed the instruction to shift only the sign bit from the AC to the MQ. Gerhard Postpischil Bradford, VT -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Addressing Scheme with 64 vs 63 bits
Yeah Gerhard and Gil, Starting writing Assembler on a 360/20,, man 704 days... Scott Ford Senior Systems Engineer [p] 678.266.3399 x304[m] 609-346-0399 identityforge.com This message is for the designated recipient only and may contain privileged, proprietary, or otherwise private information. If you have received it in error, please notify the sender immediately or let us know at [EMAIL PROTECTED] or [EMAIL PROTECTED], and then delete the original. Any other use of the email by you is prohibited. -Original Message- From: IBM Mainframe Discussion List [mailto:[EMAIL PROTECTED] On Behalf Of Gerhard Postpischil Sent: Tuesday, October 21, 2008 5:16 PM To: IBM-MAIN@BAMA.UA.EDU Subject: Re: Addressing Scheme with 64 vs 63 bits Paul Gilmartin wrote: ... and I chose the term simply because I can never remember which systems number bits left-to-right, and which right to left, or start with 0, or start with 1. And I think the IBM 704 didn't even number the sign bit -- it was simply S, and didn't (sometimes) participate in shift instructions. It's simple, the bits on the 360, et seq. are numbered in the same direction as on the 704, etc. And I really missed the instruction to shift only the sign bit from the AC to the MQ. Gerhard Postpischil Bradford, VT -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Addressing Scheme with 64 vs 63 bits
And why was it expected to be 63 bit ? Was there an expectation that one bit will be used to distinguish a 63 bit address from 31 / 24 bit addresses ? Mohammad On Sun, 19 Oct 2008 23:33:22 -0700, Edward Jaffe [EMAIL PROTECTED] wrote: It's analogous to why IBM implemented 64-bit addressing instead of the expected 63-bit. Both systems will address more data than all of the DASD in the world. But, the competitors would have convinced these idiots that 64-bit was superior to 63-bit. (Just a bit better. ;-) ) -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Addressing Scheme with 64 vs 63 bits
Mohammad Khan wrote: And why was it expected to be 63 bit ? Was there an expectation that one bit will be used to distinguish a 63 bit address from 31 / 24 bit addresses ? During the early discussions, many people expected ESAME to implement 63-bit addressing for reasons similar to why XA/370, ESA/370, and ESA/390 implemented 31-bit and not 32-bit addressing. -- Edward E Jaffe Phoenix Software International, Inc 5200 W Century Blvd, Suite 800 Los Angeles, CA 90045 310-338-0400 x318 [EMAIL PROTECTED] http://www.phoenixsoftware.com/ -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Addressing Scheme with 64 vs 63 bits
On Mon, 20 Oct 2008 09:57:12 -0700, Edward Jaffe wrote: Mohammad Khan wrote: And why was it expected to be 63 bit ? Was there an expectation that one bit will be used to distinguish a 63 bit address from 31 / 24 bit addresses ? No, because there was no prior convention of using a doubleword to contain a 31-bit address. During the early discussions, many people expected ESAME to implement 63-bit addressing for reasons similar to why XA/370, ESA/370, and ESA/390 implemented 31-bit and not 32-bit addressing. I thought 31-bit was for compatibility with existing exploitation of the sign bit; but there was no legacy convention of using the sign bit of shorter addresses stored in a doubleword to preserve compatibility with. -- gil -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Addressing Scheme with 64 vs 63 bits
On Mon, 20 Oct 2008 15:00:48 -0500, Paul Gilmartin [EMAIL PROTECTED] wrote: I thought 31-bit was for compatibility with existing exploitation of the sign bit; but there was no legacy convention of using the sign bit of shorter addresses stored in a doubleword to preserve compatibility with. Nope, it was the use of a X'80' in the high-order byte of a fullword to terminate a variable-length parameter list (of fullwords). And, FWIW, the S360/67 did implement (as an option) 32-bit addressing... Really caused problems when some customers had to get off of TSS and try to stuff things in 31-bit. W. Kevin Kelley IBM POK Lab -- z/OS Core Technical Development -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Addressing Scheme with 64 vs 63 bits
Nope, it was the use of a X'80' in the high-order byte of a fullword to terminate a variable-length parameter list (of fullwords). Isn't that the existing exploitation of the sign bit that Gil is alluding to? Date: Mon, 20 Oct 2008 20:31:08 -0500 From: [EMAIL PROTECTED] Subject: Re: Addressing Scheme with 64 vs 63 bits To: IBM-MAIN@BAMA.UA.EDU On Mon, 20 Oct 2008 15:00:48 -0500, Paul Gilmartin [EMAIL PROTECTED] wrote: I thought 31-bit was for compatibility with existing exploitation of the sign bit; but there was no legacy convention of using the sign bit of shorter addresses stored in a doubleword to preserve compatibility with. Nope, it was the use of a X'80' in the high-order byte of a fullword to terminate a variable-length parameter list (of fullwords). And, FWIW, the S360/67 did implement (as an option) 32-bit addressing... Really caused problems when some customers had to get off of TSS and try to stuff things in 31-bit. W. Kevin Kelley IBM POK Lab -- z/OS Core Technical Development _ Stay organized with simple drag and drop from Windows Live Hotmail. http://windowslive.com/Explore/hotmail?ocid=TXT_TAGLM_WL_hotmail_102008 -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Addressing Scheme with 64 vs 63 bits
On Mon, 20 Oct 2008 20:31:08 -0500, W. Kevin Kelley wrote: On Mon, 20 Oct 2008 15:00:48 -0500, Paul Gilmartin [EMAIL PROTECTED] wrote: I thought 31-bit was for compatibility with existing exploitation of the sign bit; ... Nope, it was the use of a X'80' in the high-order byte of a fullword to terminate a variable-length parameter list (of fullwords). Nope? Where do you perceive us to be in disagreement? Of course, with the transition from 370 to XA, the bit could no longer be set with MVI; OI was required. This almost certainly provoked an interminable thread on ASSEMBLER-LIST concerning the performance impact. I'd hardly be surprised, though, if OI on XA ran faster than MVI on 370. -- gil -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
Re: Addressing Scheme with 64 vs 63 bits
At 22:00 -0500 on 10/20/2008, Paul Gilmartin wrote about Re: Addressing Scheme with 64 vs 63 bits: On Mon, 20 Oct 2008 20:31:08 -0500, W. Kevin Kelley wrote: On Mon, 20 Oct 2008 15:00:48 -0500, Paul Gilmartin [EMAIL PROTECTED] wrote: I thought 31-bit was for compatibility with existing exploitation of the sign bit; ... Nope, it was the use of a X'80' in the high-order byte of a fullword to terminate a variable-length parameter list (of fullwords). Nope? Where do you perceive us to be in disagreement? Of course, with the transition from 370 to XA, the bit could no longer be set with MVI; OI was required. This almost certainly provoked an interminable thread on ASSEMBLER-LIST concerning the performance impact. I'd hardly be surprised, though, if OI on XA ran faster than MVI on 370. -- gil There was also the secondary use of the high bit to signal AM24 vs AM31 in addresses used for branching to/from subroutines. This required replacing BALR with BASR and BR with BSM to do the AM Mode Switch. -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html