sites.google.com Open in urlscan Pro
2607:f8b0:400d:c0d::65  Public Scan

Form analysis
0 forms found in the DOM

Text Content

Search this site





Skip to main content

Skip to navigation

nemo20000
 * Home
 * !RMA
 * AutoCLIspc
 * CerilicaAlphabets
 * ColourFiler
 * Data Transfer Protocol
 * DeepKeys
 * Desktop
 * Filer
 * FixUpDown
 * HaikuError
 * IconMover
 * !InAWindow
 * MenuBadges
 * MenuState
 * MetaSprite
 * ModCredits
 * nemoBasic
 * Patches
 * Toys
 * UTF-8
 * VDU
 * WriteReg
 * ThriftyLZW
 * Links

nemo20000





WRITEREG

© nemo 2002-2020


WRITEREG IS A DEBUGGING MODULE THAT DISPLAYS THE CONTENTS OF ANY REGISTER, OR
THE PROCESSOR FLAGS.

The current version is 1.11 (22 Aug 2022) [32bit]

This version is not compatible with RISC OS 6.

This module provides a number of SWIs for displaying the contents of any
register without side-effects. As well as the usual parameter registers it can
display R10-R15, plus the flags, CPSR and SPSR. It will even display the SWI’s X
Flag... because it can.

Important: As of 1.09 it now preserves ALL flags on exit, including V. This
means you need to call the X form of the SWIs if there’s any possibility of the
V flag being set on entry, or the kernel’s SWI Handler will call the Error
Handler on exit. (and StrongARM or later is now required)



RO5.24 BUG

RISC OS 5 has a history of either misunderstanding APIs or unilaterally deciding
to retrospectively change them in incompatible ways, and 5.24 is no different.
Previous versions misinterpreted error numbers (and hilariously generated an
error claiming that the error was erroneous). But 5.24 has gone further, 
deciding to break the API contract specified on RPM page 1-26 which states:

As the [SWI] exits, RISC OS checks the V flag. If it is set[...] then RISC OS
looks at bit 17 (the X bit) of the SWI number: If it is set then control returns
to your program, and you should deal with the error yourself.

5.24 changes this. If the V flag is set on exit and the X bit was set in the
SWI, it may decide to corrupt R0 entirely. It does this because some RO5
developer decided that they knew better than the author of the module and the
author of the code calling it, and that thirty years of compatibility can be
thrown away because of how very clever that one person is.

I’m sick to the back teeth of dealing with the consequences of people messing
with things they don’t understand, so I now no longer regard RO5 as a RISC OS
compatible operating system. It is somewhat compatible, but is evidently not
guaranteed to stay so.

I will henceforth refer to it as DKOS.


SWIS



WRITE_R0 - WRITE_R15

Displays the contents of the caller’s register. In the case of R14 and R15 in
26bit modes, this will include flags. The value is displayed as an 8 digit
hexadecimal number followed by a space.



WRITE_PC

Displays the caller’s PC in hex. This never includes flags.



WRITE_ALL

Displays all of the caller’s registers in a 4×4 grid.



WRITE_MANY

Displays registers selected by R8, in up to a 4×4 grid. R8b0 set displays R0, b1
displays R1, and so on.



WRITE_FLAGS

Displays the flags and mode in hex.



WRITE_NZCV

Displays the caller’s flags as four characters followed by a space.



WRITE_CPSR, WRITE_SPSR

Displays the CPSR and SPSR in hex on those processors that support them.



WRITE_X

Tells you in hex whether you called Write_X or XWrite_X. This is more clever
than useful.


RISC OS 6

RO6 contains a significant change to stack handling which severely complicates
low-level things like WriteReg and VectorExtend. I have chosen to not support it
because of this. However, I am open to persuasion if there’s anyone out there
running Six. I suspect that the vast majority of users have RO5 on new hardware
or RO4 on an emulator.


nemo is on Twitter

Google Sites
Report abuse