LSI 33k ?

LSI 33k ?

[Kevin Buettner]

Does anyone know anything about the LSI 33k?  Okay, more specifically, did
support for this core ever actually work in GDB?

In mips-tdep.c, I see the following:

    /* Names of LSI 33k registers.  */

    char *mips_lsi33k_reg_names[] = {
            "epc",      "hi",   "lo",   "sr",   "cause","badvaddr",
            "dcic", "bpc",  "bda",  "",     "",     "",     "",      "",
            "",     "",     "",     "",     "",     "",     "",      "",
            ...

With the exception of IRIX, the first line for all of the other regname
sets look like this:

	    "sr",	"lo",	"hi",	"bad",	"cause","pc",

I'm not concerned about IRIX, because BADVADDR_REGNUM, CAUSE_REGNUM,
PC_REGNUM, etc. are redefined in the appropriate tm-*.h file.

The LSI 33k rearranges these registers, apparently even the PC_REGNUM
value, but nowhere do I see a PC_REGNUM definition that's specific to
the LSI 33k.  This is bad because if the LSI 33k registers are really
laid out as indicated by lsi33_reg_names[], then there's no way that
GDB is actually obtaining the correct PC value whenever it uses PC_REGNUM.
(It'll be using the "baddvaddr" value...)

I thought for a while that perhaps the LSI 33k was simply bit-rotted,
but after doing some digging in the old Cygnus CVS repository, I don't
see how it could have ever worked.

Does anyone know?

Kevin

Re: LSI 33k ?

[Stan Shebs]

It might never have worked, or the correct code might not be in CVS 
anywhere.
I have a vague memory that it was for Array Technology ca 1992-3, and in 
those
earlier days of Cygnus, some shipments were made "by hand", and not all 
of the
customer's bits exactly corresponded with a CVS tag.  remote-array.c 
might also
have an answer in the form of jiggered-around register numbering.

Stan

Kevin Buettner wrote:

>Does anyone know anything about the LSI 33k?  Okay, more specifically, did
>support for this core ever actually work in GDB?
>
>In mips-tdep.c, I see the following:
>
>    /* Names of LSI 33k registers.  */
>
>    char *mips_lsi33k_reg_names[] = {
>            "epc",      "hi",   "lo",   "sr",   "cause","badvaddr",
>            "dcic", "bpc",  "bda",  "",     "",     "",     "",      "",
>            "",     "",     "",     "",     "",     "",     "",      "",
>            ...
>
>With the exception of IRIX, the first line for all of the other regname
>sets look like this:
>
>	    "sr",	"lo",	"hi",	"bad",	"cause","pc",
>
>I'm not concerned about IRIX, because BADVADDR_REGNUM, CAUSE_REGNUM,
>PC_REGNUM, etc. are redefined in the appropriate tm-*.h file.
>
>The LSI 33k rearranges these registers, apparently even the PC_REGNUM
>value, but nowhere do I see a PC_REGNUM definition that's specific to
>the LSI 33k.  This is bad because if the LSI 33k registers are really
>laid out as indicated by lsi33_reg_names[], then there's no way that
>GDB is actually obtaining the correct PC value whenever it uses PC_REGNUM.
>(It'll be using the "baddvaddr" value...)
>
>I thought for a while that perhaps the LSI 33k was simply bit-rotted,
>but after doing some digging in the old Cygnus CVS repository, I don't
>see how it could have ever worked.
>
>Does anyone know?
>
>Kevin
>
>  
>