ZeroPain 0.08

With ZeroPain 0.08 running on O5.25 OS_Memory,20,-1 is reporting “Compatibility page is disabled”. Is that as it should be?

SYS "XOS_Module",18,"ZeroPain" TO r0%,r1%
IF r0%=18 THEN
  PRINT "ZeroPain Module loaded"
ELSE
  PRINT "ZeroPain Module not loaded"
ENDIF
SYS "OS_Memory",20,-1 TO ,r1%
CASE r1% OF
WHEN 0  : PRINT "Compatibility page disabled"
WHEN 1  : PRINT "Compatibility page enabled"
WHEN -1 : PRINT "Compatibility not supported"
ENDCASE

I see some items removed from ZeroPain.

I took read this to mean that ZeroPain only logs.

It is not that clear cut in that the painfull SecureSockets 1.04 does require ZeroPain 0.08 to start.
Fortunately there is a fixed SecureSockets so this example is somewhat academic.

I should add that nothing appears to be actually broken.

Is that as it should be?

Yes, otherwise it wouldn’t be able to trap read accesses to zero page and log them.

I think that is how ZeroPain works – if the compatibility page is not there an abort happens, and if ZP is running it traps the abort, logs it then emulates the read. If the compatibility page is there, it simply satisfies the read with dummy information and there is no abort. I believe the ZP emulation returns similar data to that in the compatibility page.

[Jeffrey beat me to it typing his more concise message!]

Thanks for the explanations. That was how I understood ZeroPain 0.07 to work, I managed to think something had changed with 0.08.

With ref to VRPC-DL RO 5.25 (17-Apr-2018)

!Zap (12 Jul 2015 tnk-11)

Can’t seem to modify ARM code – ie Module.

The ‘Enter’ key does’t seem to respond – anyone else see this.

Thanks

Running ‘Protector’ (Zero PAGE Schutzprogramm) by Andreas Feldner.
Seems there is a option of:-

*Protect 2

switches writing and read protection for the Zeropage (have
I never tried)

I wonder if this could be used to log Zero Pain errors – for ones using ‘Low Vector’ builds – ie like the emulators?

We tend to use Adrian Lees tools from http://www.sendiri.co.uk/ when doing zero page tests on RO4.x and low vector builds. It’s a bit of a pain to have to turn them on/off constantly to cope with various legacy bits and bobs which cause problems, but it gives a few extra options, esp if you’re using virtual systems.

I wonder if this could be used to log Zero Pain errors – for ones using ‘Low Vector’ builds – ie like the emulators?

Strikes me as a good idea to provide something of a testbed on standard off the shelf portable machines (laptop PeeCee)

Would be a useful idea perhaps, but I suspect that when using RISC OS versions older than 5.25 the number of zero page accesses would increase with age, which could easily swamp the ones anyone was interested in – particularly when older applications were also used. So any logging would probably need some means to restrict the logs in some way … and istr some discussion in the early days indicating that was not simple.

For versions 5.25 or later, will a low vector build still be required?