Machine Type

It would be nice to have a SYS call from RISC OS that would identify the machine platform and revision as HALDeviceComms_GPIO does. I have tried SYS “HALDeviceComms_GPIO” but it doesn’t work.

SYS “OS_Hardware” ,&403,0,0,0,0,0,0,0,4 TO r0%,r1%,r2%
PRINT"Processor “;~(?(r2%+2))
PRINT”Type &“;~!(r2%+64)
PRINT”Revision &";~!(r2%+76)

This thread seems to cover the same topic, even by the same author. It looks like testing for the feature is the done thing rather than holding a big table of every machine ever made (and those that are yet to be)

SYS “OS_Hardw …

For those pasting, there seems to be a problem with SmartQuotes. It don’t work here!

And a ‘SWI name not known’ error means Virtual Risc PC? Or Select?

The Wiki page is in the HAL category, which IIRC was introduced in RISC OS 5. It’s therefore probably not available in Select.

Many thanks. I have added a few lines to RISCOSmark so that I now get both the machine and the filing system being tested:

  120real%=TRUE
  130ON ERROR real%=FALSE
  140IF real% THEN SYS "OS_Hardware",&403,0,0,0,0,0,0,0,4 TO r0%,r1%,r2%
  150ON ERROR OFF
  160do%=FALSE
  170IF real%=0 THEN
  180  aa$="Virtual RPC etc."
  190  *help UtilityModule { > Pipe:$.sysin }
  200  sys%=OPENIN("Pipe:$.sysin")
  210  REPEAT
  220    A$=GET$#sys%
  230    B$="Module is: "
  240    IF LEFT$(A$,LEN(B$))=B$ THEN
  250      aa$+=" RISC OS ":do%=2
  260    ELSE
  270      IF do% THEN aa$+=A$:do%-=1
  280    ENDIF
  300  UNTIL EOF#sys%
  310  CLOSE#sys%
  320ELSE
  330  SYS "XOS_GenerateError",!(r2%+12) TO aa$
  331  b%=INSTR(aa$," GPIO interface")
  332  IF b%>0 THEN aa$=MID$(aa$,1,b%-1)
  340  aa$+=" processor &"+STR$~(?(r2%+2))+" type &"+STR$~!(r2%+64)+" revision &"+STR$~!(r2%+76)
  350ENDIF
  370logfile$="Pipe:$.mlog"+RIGHT$("0000"+STR$(t% MOD 100000),5)
  380log%=OPENOUT(logfile$)
  390ON ERROR ON ERROR OFF:PRINT"Error (see logfile)":BPUT#log%,REPORT$+" in line "+STR$ERL:CLOSE#log%:OSCLI("SetType "+logfile$+" FFF"):OSCLI("Filer_Run "+logfile$):END
  441IF real% THEN
  450  *help modules { > Pipe:$.sysin }
  460  sys%=OPENIN("Pipe:$.sysin")
  470  REPEAT
  480    A$=GET$#sys%
  490    B$="MOS Utilities"
  500    IF LEFT$(A$,LEN(B$))=B$ THEN BPUT#log%,A$
  510    B$="FileSwitch"
  520    IF LEFT$(A$,LEN(B$))=B$ THEN BPUT#log%,A$
  530    B$="C Library"
  540    IF LEFT$(A$,LEN(B$))=B$ THEN BPUT#log%,A$
  550    B$="USBDriver"
  560    IF LEFT$(A$,LEN(B$))=B$ THEN BPUT#log%,A$
  570    B$="FileCore"
  580    IF LEFT$(A$,LEN(B$))=B$ THEN BPUT#log%,A$
  590    B$="SCSIFS"
  600    IF LEFT$(A$,LEN(B$))=B$ THEN BPUT#log%,A$
  610  UNTIL EOF#sys%
  620  CLOSE#sys%
  621ENDIF
  640BPUT#log%,aa$+" tested at "+TIME$
  650SYS "OS_ReadVarVal","FileSwitch$CurrentFilingSystem"+CHR$0,block%,256,0,3 TO ,,pp%
  660CFS$=""
  670FOR i%=0 TO pp%-1
  680  CFS$+=CHR$(block%?i%)
  690NEXT
  700SYS "OS_ReadVarVal","FileSwitch$"+CFS$+"$CSD"+CHR$0,block%,256,0,3 TO ,,pp%
  710CSD$=""
  720FOR i%=0 TO pp%-1
  730  CSD$+=CHR$(block%?i%)
  740NEXT
  750BPUT#log%,"Filing system: "+CFS$+CSD$
...

Which gives the following output:

RISCOSmark 1.01 (14 May 2003)
Comparison with RiscPC SA 202MHz running RISC OS 4.02 800x600,256
(HD benchmarks are in kilobytes/sec)

MOS Utilities   5.23 (10 Aug 2015)
FileSwitch      2.83 (11 Nov 2014)
C Library       5.86 (30 Jun 2015)
USBDriver       0.74 (04 Jul 2015)
FileCore        3.66 (28 Mar 2015)
SCSIFS          1.28 (06 Oct 2013)
iMX6 processor &3 type &0 revision &16000000 tested at Wed,23 Sep 2015.08:33:23
Filing system: LanMan98:NAS.$.SpeedTests
Graphics Resoloution: 1680x1050, 16M colours

Test                                                   Benchmark
Processor - Looped instructions (cache)                  1883774    1059%
Memory - Multiple register transfer                         7200    4444%
Rectangle Copy - Graphics acceleration test                  296     122%
Icon Plotting - 16 colour sprite with mask                 11776     588%
Draw Path - Stroke narrow line                              5479     351%
Draw Fill - Plot filled shape                               3851     263%
HD Read - Block load 8MB file                               1832      61%
HD Write - Block save 8MB file                             13768     452%
FS Read - Byte stream file in                               1370     661%
FS Write - Byte stream file out                              551     286%

Hi Chris can you post a copy of the altered program i can’t find it any where.

I have added a few lines to RISCOSmark

Done that. Very nice addition, thanks.

I did find one tiny little snag, due to user incompetence, having forgotten how romark places its test file. It comes unglued on this Raspberry Pi if the CSD has not been explicitly set, as it isn’t after a boot.

System variable 'FileSwitch$SCSI$CSD' not found in line 76

*show FileSwitch*
FileSwitch$CurrentFilingSystem : SCSI
FileSwitch$SpecialField :
FileSwitch$TemporaryFilingSystem : SCSI
*

N.B. My line numbers that is, after the update, this is the line :-

SYS "OS_ReadVarVal","FileSwitch$"+CFS$+"$CSD"+CHR$0,block%,256,0,3 TO ,,pp%

I added this to catch the unset state :-

SYS "XOS_ReadVarVal","FileSwitch$"+CFS$+"$CSD"+CHR$0,block%,-1,0,3 TO ,,pp%
IF pp%=0 THEN
  CSD$=" - CSD not set"
ELSE
  SYS "OS_ReadVarVal","FileSwitch$"+CFS$+"$CSD"+CHR$0,block%,256,0,3 TO ,,pp%
  CSD$=""
  FOR i%=0 TO pp%-1
    CSD$+=CHR$(block%?i%)
  NEXT
ENDIF

Yes it comes ‘unglued’ as you say (on any platform) if the CSD is not set. I always run !MultiTask [available from !Store or here] which sets the CSD to the last opened directory or last double clicked file so I don’t have to bother with a full pathname in my BASIC programmes. Then RISCOSMark tests the directory (and filesystem) from which it is run.

Uploaded the tweaked RISCOSMark here with the suggested correction for when the CSD is not set. It uses 8Mbytes for the direct load/save test and counts read speed after the file to be read has been created. I chose 8Mbytes so that it was larger than any memory cache.

!MultiTask

I haven’t come by that one before, I’ll give it a look, sounds interesting.

Then RISCOSMark tests the directory (and filesystem) from which it is run.

That would be more intuitive.

Thanks for that.

IF pp%=0 THEN
CSD$=" - CSD not set"

Wouldn’t it make more sense to assume $ instead of complaining the CSD isn’t set?

Wouldn’t it make more sense to assume $ instead of complaining the CSD isn’t set?

Probably. However the CSD may be set to a different filing system than the current filing system and I think RISCOSMark tests the CSD (@) rather than the current filing system so complaining is perhaps prudent.

However the CSD may be set to a different filing system than the current filing system

I believe that’s the reason for using a little

Set MyProg$Dir <Obey$Dir> 

Yes, exactly.

Well, it seems that my code shown above has a wrong value in…

PRINT”Revision &";~!(r2%+68) is correct.

The reason I got the +76 value is in the GPIO module the detection code uses the variable set-up in castle/sources/kernel/hdr.GPIODevice, and when the code is assembled it seems the wrong value is used.
This has caused the GPIO module to fail with my new test builds and has caused lots of “what am I doing wrong” moments.
Why this value is incorrect is still a unknown to me but using a number instead of the variable works….