Patchwork FT32: support for FT32B processor - part 1

login
register
mail settings
Submitter James Bowman
Date Oct. 4, 2017, 4:51 p.m.
Message ID <2BB0A51F073B384698CACFD1D5A30FCC0DED1F49@glaexch1>
Download mbox | patch
Permalink /patch/23330/
State New
Headers show

Comments

James Bowman - Oct. 4, 2017, 4:51 p.m.
FT32B is a new FT32 family member. It has a code
compression scheme, which requires the use of linker
relaxations. The change is quite large, so submission
is in several parts.

Part 1 adds a 15-bit instruction field, and CPU-specific functions for
the code compression that are used in binutils and GDB.

This patch contains the gdb changes. The corresponding binutils patch is
https://sourceware.org/ml/binutils/2017-10/msg00010.html

OK to commit?

James.

gdb/ChangeLog:

2017-10-04  James Bowman  <james.bowman@ftdichip.com>

	* sim/ft32/interp.c (step_once): Replace FT32_FLD_K8 with K15.

---
Kevin Buettner - Oct. 7, 2017, 5:01 p.m.
On Wed, 4 Oct 2017 16:51:02 +0000
James Bowman <james.bowman@ftdichip.com> wrote:

> FT32B is a new FT32 family member. It has a code
> compression scheme, which requires the use of linker
> relaxations. The change is quite large, so submission
> is in several parts.
> 
> Part 1 adds a 15-bit instruction field, and CPU-specific functions for
> the code compression that are used in binutils and GDB.
> 
> This patch contains the gdb changes. The corresponding binutils patch is
> https://sourceware.org/ml/binutils/2017-10/msg00010.html
> 
> OK to commit?

Hi James,

I've looked over your patch. While I have not researched the specifics
of the FT32B, your patch looks reasonable to me.

I would approve this patch if I could, but after looking at
sim/MAINTAINERS it seems that only Mike Frysinger may approve this
patch.

Kevin
Simon Marchi - Oct. 9, 2017, 2:38 p.m.
On 2017-10-07 01:01 PM, Kevin Buettner wrote:
> On Wed, 4 Oct 2017 16:51:02 +0000
> James Bowman <james.bowman@ftdichip.com> wrote:
> 
>> FT32B is a new FT32 family member. It has a code
>> compression scheme, which requires the use of linker
>> relaxations. The change is quite large, so submission
>> is in several parts.
>>
>> Part 1 adds a 15-bit instruction field, and CPU-specific functions for
>> the code compression that are used in binutils and GDB.
>>
>> This patch contains the gdb changes. The corresponding binutils patch is
>> https://sourceware.org/ml/binutils/2017-10/msg00010.html
>>
>> OK to commit?
> 
> Hi James,
> 
> I've looked over your patch. While I have not researched the specifics
> of the FT32B, your patch looks reasonable to me.
> 
> I would approve this patch if I could, but after looking at
> sim/MAINTAINERS it seems that only Mike Frysinger may approve this
> patch.
> 
> Kevin
> 

There is no maintainer listed for ft32 architecture, both in gdb/ and sim/.
Maybe it would be useful to have some, so arch-specific changes can go through
them (just throwing the idea out there).

Simon

Patch

diff --git a/sim/ft32/interp.c b/sim/ft32/interp.c
index b6fedc1..3bc08ee 100644
--- a/sim/ft32/interp.c
+++ b/sim/ft32/interp.c
@@ -332,7 +332,7 @@  step_once (SIM_DESC sd)
   uint32_t pa;
   uint32_t aa;
   uint32_t k16;
-  uint32_t k8;
+  uint32_t k15;
   uint32_t al;
   uint32_t r_1v;
   uint32_t rimmv;
@@ -372,7 +372,11 @@  step_once (SIM_DESC sd)
   pa   =              (inst >> FT32_FLD_PA_BIT) & LSBS (FT32_FLD_PA_SIZ);
   aa   =              (inst >> FT32_FLD_AA_BIT) & LSBS (FT32_FLD_AA_SIZ);
   k16  =              (inst >> FT32_FLD_K16_BIT) & LSBS (FT32_FLD_K16_SIZ);
-  k8   = nsigned (8,  (inst >> FT32_FLD_K8_BIT) & LSBS (FT32_FLD_K8_SIZ));
+  k15  =              (inst >> FT32_FLD_K15_BIT) & LSBS (FT32_FLD_K15_SIZ);
+  if (k15 & 0x80)
+    k15 ^= 0x7f00;
+  if (k15 & 0x4000)
+    k15 -= 0x8000;
   al   =              (inst >> FT32_FLD_AL_BIT) & LSBS (FT32_FLD_AL_SIZ);
 
   r_1v = cpu->state.regs[r_1];
@@ -499,7 +503,7 @@  step_once (SIM_DESC sd)
       break;
 
     case FT32_PAT_LPMI:
-      cpu->state.regs[r_d] = ft32_read_item (sd, dw, cpu->state.regs[r_1] + k8);
+      cpu->state.regs[r_d] = ft32_read_item (sd, dw, cpu->state.regs[r_1] + k15);
       cpu->state.cycles += 1;
       break;
 
@@ -508,7 +512,7 @@  step_once (SIM_DESC sd)
       break;
 
     case FT32_PAT_STI:
-      cpu_mem_write (sd, dw, cpu->state.regs[r_d] + k8, cpu->state.regs[r_1]);
+      cpu_mem_write (sd, dw, cpu->state.regs[r_d] + k15, cpu->state.regs[r_1]);
       break;
 
     case FT32_PAT_LDA:
@@ -517,7 +521,7 @@  step_once (SIM_DESC sd)
       break;
 
     case FT32_PAT_LDI:
-      cpu->state.regs[r_d] = cpu_mem_read (sd, dw, cpu->state.regs[r_1] + k8);
+      cpu->state.regs[r_d] = cpu_mem_read (sd, dw, cpu->state.regs[r_1] + k15);
       cpu->state.cycles += 1;
       break;
 
@@ -534,8 +538,8 @@  step_once (SIM_DESC sd)
     case FT32_PAT_EXI:
       {
 	uint32_t tmp;
-	tmp = cpu_mem_read (sd, dw, cpu->state.regs[r_1] + k8);
-	cpu_mem_write (sd, dw, cpu->state.regs[r_1] + k8, cpu->state.regs[r_d]);
+	tmp = cpu_mem_read (sd, dw, cpu->state.regs[r_1] + k15);
+	cpu_mem_write (sd, dw, cpu->state.regs[r_1] + k15, cpu->state.regs[r_d]);
 	cpu->state.regs[r_d] = tmp;
 	cpu->state.cycles += 1;
       }