Couple of questions/bugs?

[Bob S]

Z390,

I've run into a couple of problems that I'd like to bring up.

1.  The TRT instruction does not clear the high half of R1 if the CPU is in AMODE 64.

      LGHI   R1,-1

     TRT     FIELD,TRT_TABLE

     IF (NZ)

          R1 high half is still high-values/not a 64 bit address.

    ENDIF

Workaround is to do a LGHI R1,0 before the TRT.

2.  When using extended precision binary floating point numbers, I get unexpected IEEE underflow exceptions when the exponent isn't close to zero.  Are extended precision numbers converted internally to DBs and processed that way?  Just wondering if there are some artificial limits in place for extended precision operations.

Workaround is to just use DBs.  Side issue.  It would be nice if the FPC was displayed somewhere on program check dumps.  I get it by a STFPC in an ESTAE routine and see the DXC code of x'10.'

Regards,

Bob Schulien

[John Ganci]

I see Abe opened three issues for your two concerns: TRT amode64 GR1 issue,, unexpected IEEE underflow, no FPC in dump. Comments.

1. Agreed, TRT (and TRTR, EDMK) do not properly set GR1 when addressing mode is 64. Looking at the code this should be an easy fix.

2. FPC does appear in a dump What version of z390 are you using? Sample output from v1.8.2:

$ z390/work/z390/bash/exec z390/tests/TABCODE noinit dump
15:21:48 TABCODE   EZ390 START USING z390 v1.8.2-jg ON J2SE 17.0.12 01/20/25
EZ390I Trace Table Entry      800FFEA0 90ECD00C     STM  
EZ390I Trace Table Entry      800FFEA4 18CF         LR  
EZ390I Trace Table Entry      800FFEA6 58201000     L    
EZ390I Trace Table Entry      800FFEAA 48302000     LH  
EZ390I Trace Table Entry      800FFEAE 1233         LTR  
EZ390I Trace Table Entry      800FFEB0 4780C032     BE  
EZ390I Trace Table Entry      800FFED2 41400064     LA  
EZ390I Trace Table Entry      800FFED6 1814         LR  
EZ390E error  11 ABEND PSW=07850600 800FFED8 0A0D         SVC   ABEND U0100
 R0-R3 00000000800FFEA0 0000000000000064 0000000000002304 0000000000000000
 R4-R7 0000000000000064 0000000000000000 0000000000000000 0000000000000000
 R8-RB 0000000000000000 0000000000000000 0000000000000000 0000000000000000
 RC-RF 00000000800FFEA0 0000000000002100 0000000000002018 00000000800FFEA0

  FPC  70000000
 F0-F3 0000000000000000 0000000000000000 0000000000000000 0000000000000000
 F4-F7 0000000000000000 0000000000000000 0000000000000000 0000000000000000
 F8-FB 0000000000000000 0000000000000000 0000000000000000 0000000000000000
 FC-FF 0000000000000000 0000000000000000 0000000000000000 0000000000000000

 TIOT NO DCB ENTRIES FOUND

...

3. Please supply a simple program that illustrates the unexpected IEEE underflow exception so that we are able duplicate the error you are seeing.

Regards, John Ganci

--
You received this message because you are subscribed to the Google Groups "z390" group.
To unsubscribe from this group and stop receiving emails from it, send an email to z390+uns...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/z390/b178fe7b-ac05-4f99-920b-6ba3f06abf5cn%40googlegroups.com.

[John Ganci]

PS.

I looked at the issues on github. You did supply sample code for the IEEE problem The output indicates that you are using z390 v1.8.1. If you upgrade to v1.8.2 you should see the FPC in a dump.

Regards, John Ganci

[Abe Kornelis]

Bob,

to download pre-compiled version of z390, latest release, please visit
https://github.com/z390development/z390/releases/tag/v1.8.2

At the bottom of the page, select the zip file to download.

If you want to install from source, please select the appropriate
source code archive.

If you need additional instructions for upgrading/installing,
please let me know.

Kind regards,
Abe
===

Op 20/01/2025 om 22:42 schreef John Ganci:

[Abe Kornelis]

Bob,

I have closed the request for showing FPC on dumps.

Kind regards,
Abe
===

Op 20/01/2025 om 22:35 schreef John Ganci: