H8300 - Addition of Pseudo registers has broken displaying of CCR and EXR registers

H8300 - Addition of Pseudo registers has broken displaying of CCR and EXR registers

[Dhananjay R. Deshpande]

Hi,

With the addition of Pseudo registers, "info reg" command doesn't show correct values of CCR and EXR registers. If you set these registers from GDB ( set $ccr = 0x55 )then it shows correct value but when the register is modified by instruction it always shows "0". This fails both with SIM and remote target ( stub )

Regards,
Dhananjay

Re: H8300 - Addition of Pseudo registers has broken displaying of CCR and EXR registers

[Andrew Cagney]

> Hi,
> 
> With the addition of Pseudo registers, "info reg" command doesn't show correct values of CCR and EXR registers. If you set these registers from GDB ( set $ccr = 0x55 )then it shows correct value but when the register is modified by instruction it always shows "0". This fails both with SIM and remote target ( stub )

Sounds like it is never fetched, or being fetched from the wrong place 
:-(  What does "maint print raw-registers" (which dumps the raw state of 
the register cache) display?   Are you seeing calls to the simulator for 
that register?

A testsuite/gdb.arch/ testcase is going to help here.  Especially to 
ensure that it doesn't break again.

enjoy,
Andrew

RE: H8300 - Addition of Pseudo registers has broken displaying of CCR and EXR registers

[Dhananjay R. Deshpande]

Hi Andrew,

> -----Original Message-----
> From: Andrew Cagney [mailto:cagney@gnu.org]

> Sounds like it is never fetched, or being fetched from the 
> wrong place 
> :-(  

I think this is because difference in raw and pseudo register size. 
The pseudo registers are of one byte and due to big endian we are getting 
wrong byte.

>What does "maint print raw-registers" (which dumps the 
> raw state of 
> the register cache) display?  

It display <cooked> for CCR and print cooked-registers displays zero.

> Are you seeing calls to the 
> simulator for 
> that register?

Yes.

> 
> A testsuite/gdb.arch/ testcase is going to help here.  Especially to 
> ensure that it doesn't break again.

I will try but its going to take some time to understand the test framework.
Here is the C code which can be used for testing - 

int main()
{
	asm("mov #0x55, r0l");
	asm("ldc r0l, ccr");
	asm("stc ccr, r1l");
}

Regards,
Dhananjay