Correct way to hard reset in the bootblock on 1.3-3.1?

[a/b]

Originally Posted by admiral

New vectors can be set to be "followers" that just do an indirect call to the matching vectors at 0. Thus delaying everything vectored just a few cycles.
This would break some very, very tight software on plain 68010 running at the standard A500 7MHz clock... but nothing else.

Don't think we are on the same page. I'm talking about 010+ (VBR is not 0 at boot time), and no MMU or MMU not actively used by virus. If you do something like (fairly common for older, but not only, code):

  move.l #sv_code,$80.w
  trap #0
svr_code:

Since VBR is not 0 you won't end up in SV mode and will crash when you use privileged stuff or when SV stack is trashed (because your new a7 is now USP, and not SSP). Plus all your interrupts doing nothing and you basically re-enabled system interrupts while trashing the system.

All things considered, hopefully :P, trying to get to SV mode *manually* seems to be a chicked and egg situation (unless 68000): you need VBR to try (hello MMU) to patch a vector, and you need SV to get VBR, so as Ross said Supervisor() seems to be the most feasible way because you'd have to break something to close that door. And if only ROM code is allowed through, what good is that? As soon as you run e.g. SetPatch you probably get a nice guru, or an unstable system if all the exceptions are consumed by virus to let them slide.

[Photon]

Originally Posted by Thomas Richter

No. Run MuForce, and you'll get a hit, but AbsExecBase will remain unchainged.

Does this mean that

move.l 4.w,a6
st.w 7
jsr ColdReboot(a6)

gives a high risk of the ColdReboot function not completing correctly until Execbase is rebuilt by ROM? That's all I care about.

[ross]

Originally Posted by Photon

Does this mean that

move.l 4.w,a6
st.w 7
jsr ColdReboot(a6)

gives a high risk of the ColdReboot function not completing correctly until Execbase is rebuilt by ROM? That's all I care about.

This *should* work because ColdReboot do not seems to read execbase but trust the passed A6, anyway it is probably unsupported.
(EDIT: it internally calls system functions for the caches, supervisor mode, etc)

But I think Thomas was referring to the MMU trapping (and preventing) you from writing to execbase.

st 7.w