Make the data of the *context registers 32 bits wide. #981
Conversation
xml/hwbp_registers.xml
Outdated
| In this section XLEN refers to the effective XLEN in the current execution | ||
| mode. This is DXLEN when in Debug Mode. | ||
| On systems where those values of XLEN can differ, this is handled as | ||
| follows. |
There was a problem hiding this comment.
I don't think that "those" makes sense anymore. If we're going to remove the MXLEN reference, shouldn't we also remove DXLEN since elsewhere it says that XLEN is DXLEN in debug mode?
Maybe: "In this section XLEN refers to the effective XLEN in the current execution mode. On systems where XLEN values can differ between modes, this is handled as follows."
There was a problem hiding this comment.
I kept DXLEN in there to just make it extra clear than when the hart is halted DXLEN applies. But you're right it's not necessary, and I guess it is odd to only mention that one and not any other modes. I'll simplify as you suggest.
| @@ -300,7 +302,8 @@ same project unless stated otherwise. | |||
| hypervisor directly. | |||
| ==== | |||
|
|
|||
| <field name="hcontext" bits="XLEN-1:0" access="WARL" reset="0"> | |||
| <field name="0" bits="XLEN-1:14" access="R" reset="0" /> | |||
| <field name="hcontext" bits="13:0" access="WARL" reset="0"> | |||
There was a problem hiding this comment.
Is this really 13:0 or 31:0? It's "recommended" to implement no more than 14 bits but that isn't a requirement. The required max seems like it used to be XLEN and is now 32.
There was a problem hiding this comment.
I wanted to change it to 14 bits because there is no value in implementing additional bits. None of our triggers look at bits higher than bit 13 in this register. Or did I miss something?
There was a problem hiding this comment.
You're right. I forgot that the width of mhvalue basically makes it useless to have anything wider than 14 bits. I think that we should change the final sentence of the mcontext register's hcontext field to be:
If the H extension is implemented, it’s recommended to implement no more than 7 bits on RV32.
Because it would now be impossible to implement more than 14 on RV64.
xml/hwbp_registers.xml
Outdated
| 0. It's recommended to implement no more than 16 bits on RV32, and | ||
| 34 on RV64. | ||
| 32 on RV64. |
There was a problem hiding this comment.
Now I notice that this recommendation is confusing because it's not possible to implement more than 32 bits. We can just drop everything after the comma.
There was a problem hiding this comment.
I had read this as:
- It's recommended to implement no more than 16 bits on RV32.
- It's recommended to implement 32 bits on RV64.
But I agree it's ambiguous.
Maybe we should scrap the "no more" altogether, and just recommend 16 and 32? It's already clear that fewer are allowed.
E.g.
It's recommended to implement 16 bits on RV32, and 32 on RV64.
There was a problem hiding this comment.
I'm OK with your proposal but I'd eliminate the comma and then maybe say "bits" to be more symmetric: "It's recommended to implement 16 bits on RV32 and 32 bits on RV64."
xml/hwbp_registers.xml
Outdated
| 0. It's recommended to implement 16 bits on RV32, and 32 bits on | ||
| RV64. |
There was a problem hiding this comment.
| 0. It's recommended to implement 16 bits on RV32, and 32 bits on | |
| RV64. | |
| 0. It's recommended to implement 16 bits on RV32 and 32 bits on | |
| RV64. |
Clarify bit recommendations on scontext and mcontext.
Not just M-Mode and D-Mode.
The commit provdes the change between debug spec 1.0.0-rc1 and 1.0.0-rc2 Reference: riscv/riscv-debug-spec#981
The commit provdes the change between debug spec 1.0.0-rc1 and 1.0.0-rc2 Reference: riscv/riscv-debug-spec#981
Still todo, is figure out what to call XLEN there. In the introduction we mention what XLEN means for Debug and M-Mode, but some of these registers are accessible in other modes also.