Add IAR RISC-V 32 Embedded Extension Support (#588)

Adds RV32E support to the IAR port. This is done by
reducing our register usage to the first 16 registers
only.

Influenced by changes in https://github.com/FreeRTOS/FreeRTOS-Kernel/pull/543

Signed-off-by: Gaurav Aggarwal <aggarg@amazon.com>
Co-authored-by: Gaurav Aggarwal <aggarg@amazon.com>

portable/GCC/RISC-V/portASM.S

@@ -208,12 +208,12 @@ pxPortInitialiseStack:
     store_x x0, 0(a0)                   /* Critical nesting count starts at 0 for every task. */
 
 #ifdef __riscv_32e
-    addi a0, a0, -(6 * portWORD_SIZE)   /* Space for registers x11-x15. */
+    addi a0, a0, -(6 * portWORD_SIZE)   /* Space for registers x10-x15. */
 #else
-    addi a0, a0, -(22 * portWORD_SIZE)  /* Space for registers x11-x31. */
+    addi a0, a0, -(22 * portWORD_SIZE)  /* Space for registers x10-x31. */
 #endif
     store_x a2, 0(a0)                   /* Task parameters (pvParameters parameter) goes into register X10/a0 on the stack. */
-    addi a0, a0, -(6 * portWORD_SIZE)   /* Space for registers x5-x9. */
+    addi a0, a0, -(6 * portWORD_SIZE)   /* Space for registers x5-x9 + taskReturnAddress. */
     load_x t0, xTaskReturnAddress
     store_x t0, 0(a0)                   /* Return address onto the stack. */
     addi t0, x0, portasmADDITIONAL_CONTEXT_SIZE /* The number of chip specific additional registers. */

portable/IAR/RISC-V/portASM.s

@@ -107,35 +107,35 @@ definitions. */
 	CODE
 
 portUPDATE_MTIMER_COMPARE_REGISTER MACRO
-    load_x t0, pullMachineTimerCompareRegister  /* Load address of compare register into t0. */
+    load_x a0, pullMachineTimerCompareRegister  /* Load address of compare register into a0. */
-    load_x t1, pullNextTime                     /* Load the address of ullNextTime into t1. */
+    load_x a1, pullNextTime                     /* Load the address of ullNextTime into a1. */
 
     #if( __riscv_xlen == 32 )
 
         /* Update the 64-bit mtimer compare match value in two 32-bit writes. */
-        li t4, -1
+        li a4, -1
-        lw t2, 0(t1)                /* Load the low word of ullNextTime into t2. */
+        lw a2, 0(a1)                /* Load the low word of ullNextTime into a2. */
-        lw t3, 4(t1)                /* Load the high word of ullNextTime into t3. */
+        lw a3, 4(a1)                /* Load the high word of ullNextTime into a3. */
-        sw t4, 0(t0)                /* Low word no smaller than old value to start with - will be overwritten below. */
+        sw a4, 0(a0)                /* Low word no smaller than old value to start with - will be overwritten below. */
-        sw t3, 4(t0)                /* Store high word of ullNextTime into compare register.  No smaller than new value. */
+        sw a3, 4(a0)                /* Store high word of ullNextTime into compare register.  No smaller than new value. */
-        sw t2, 0(t0)                /* Store low word of ullNextTime into compare register. */
+        sw a2, 0(a0)                /* Store low word of ullNextTime into compare register. */
         lw t0, uxTimerIncrementsForOneTick  /* Load the value of ullTimerIncrementForOneTick into t0 (could this be optimized by storing in an array next to pullNextTime?). */
-        add t4, t0, t2              /* Add the low word of ullNextTime to the timer increments for one tick (assumes timer increment for one tick fits in 32-bits). */
+        add a4, t0, a2              /* Add the low word of ullNextTime to the timer increments for one tick (assumes timer increment for one tick fits in 32-bits). */
-        sltu t5, t4, t2             /* See if the sum of low words overflowed (what about the zero case?). */
+        sltu t1, a4, a2             /* See if the sum of low words overflowed (what about the zero case?). */
-        add t6, t3, t5              /* Add overflow to high word of ullNextTime. */
+        add t2, a3, t1              /* Add overflow to high word of ullNextTime. */
-        sw t4, 0(t1)                /* Store new low word of ullNextTime. */
+        sw a4, 0(a1)                /* Store new low word of ullNextTime. */
-        sw t6, 4(t1)                /* Store new high word of ullNextTime. */
+        sw t2, 4(a1)                /* Store new high word of ullNextTime. */
 
     #endif /* __riscv_xlen == 32 */
 
     #if( __riscv_xlen == 64 )
 
         /* Update the 64-bit mtimer compare match value. */
-        ld t2, 0(t1)                /* Load ullNextTime into t2. */
+        ld t2, 0(a1)                /* Load ullNextTime into t2. */
-        sd t2, 0(t0)                /* Store ullNextTime into compare register. */
+        sd t2, 0(a0)                /* Store ullNextTime into compare register. */
         ld t0, uxTimerIncrementsForOneTick  /* Load the value of ullTimerIncrementForOneTick into t0 (could this be optimized by storing in an array next to pullNextTime?). */
         add t4, t0, t2              /* Add ullNextTime to the timer increments for one tick. */
-        sd t4, 0(t1)                /* Store ullNextTime. */
+        sd t4, 0(a1)                /* Store ullNextTime. */
 
     #endif /* __riscv_xlen == 64 */
 	ENDM
@@ -205,7 +205,7 @@ portUPDATE_MTIMER_COMPARE_REGISTER MACRO
  * pxCode
  */
 pxPortInitialiseStack:
-    csrr t0, CSR_MSTATUS					/* Obtain current mstatus value. */
+    csrr t0, CSR_MSTATUS                /* Obtain current mstatus value. */
     andi t0, t0, ~0x8                   /* Ensure interrupts are disabled when the stack is restored within an ISR.  Required when a task is created after the schedulre has been started, otherwise interrupts would be disabled anyway. */
     addi t1, x0, 0x188                  /* Generate the value 0x1880, which are the MPIE and MPP bits to set in mstatus. */
     slli t1, t1, 4
@@ -215,9 +215,13 @@ pxPortInitialiseStack:
     store_x t0, 0(a0)                   /* mstatus onto the stack. */
     addi a0, a0, -portWORD_SIZE         /* Space for critical nesting count. */
     store_x x0, 0(a0)                   /* Critical nesting count starts at 0 for every task. */
-    addi a0, a0, -(22 * portWORD_SIZE)  /* Space for registers x11-x31. */
+#ifdef __riscv_32e
+    addi a0, a0, -(6 * portWORD_SIZE)   /* Space for registers x10-15. */
+#else
+    addi a0, a0, -(22 * portWORD_SIZE)  /* Space for registers x10-x31. */
+#endif
     store_x a2, 0(a0)                   /* Task parameters (pvParameters parameter) goes into register X10/a0 on the stack. */
-    addi a0, a0, -(6 * portWORD_SIZE)   /* Space for registers x5-x9. */
+    addi a0, a0, -(6 * portWORD_SIZE)   /* Space for registers x5-x9 + taskReturnAddress. */
     load_x t0, xTaskReturnAddress
     store_x t0, 0(a0)                   /* Return address onto the stack. */
     addi t0, x0, portasmADDITIONAL_CONTEXT_SIZE /* The number of chip specific additional registers. */
@@ -250,6 +254,7 @@ xPortStartFirstTask:
     load_x  x13, 10 * portWORD_SIZE( sp )   /* a3 */
     load_x  x14, 11 * portWORD_SIZE( sp )   /* a4 */
     load_x  x15, 12 * portWORD_SIZE( sp )   /* a5 */
+#ifndef __riscv_32e
     load_x  x16, 13 * portWORD_SIZE( sp )   /* a6 */
     load_x  x17, 14 * portWORD_SIZE( sp )   /* a7 */
     load_x  x18, 15 * portWORD_SIZE( sp )   /* s2 */
@@ -266,12 +271,13 @@ xPortStartFirstTask:
     load_x  x29, 26 * portWORD_SIZE( sp )   /* t4 */
     load_x  x30, 27 * portWORD_SIZE( sp )   /* t5 */
     load_x  x31, 28 * portWORD_SIZE( sp )   /* t6 */
+#endif
 
-    load_x  x5, 29 * portWORD_SIZE( sp )    /* Obtain xCriticalNesting value for this task from task's stack. */
+    load_x  x5, portCRITICAL_NESTING_OFFSET * portWORD_SIZE( sp )    /* Obtain xCriticalNesting value for this task from task's stack. */
     load_x  x6, pxCriticalNesting           /* Load the address of xCriticalNesting into x6. */
     store_x x5, 0( x6 )                     /* Restore the critical nesting value for this task. */
 
-    load_x  x5, 30 * portWORD_SIZE( sp )    /* Initial mstatus into x5 (t0). */
+    load_x  x5, portMSTATUS_OFFSET * portWORD_SIZE( sp )    /* Initial mstatus into x5 (t0). */
     addi    x5, x5, 0x08                    /* Set MIE bit so the first task starts with interrupts enabled - required as returns with ret not eret. */
     csrrw   x0, CSR_MSTATUS, x5             /* Interrupts enabled from here! */
 

portable/IAR/RISC-V/portContext.h

@@ -48,7 +48,15 @@
  * portasmRESTORE_ADDITIONAL_REGISTERS macros - which can be defined in a chip
  * specific version of freertos_risc_v_chip_specific_extensions.h.  See the
  * notes at the top of portASM.S file. */
-#define portCONTEXT_SIZE ( 31 * portWORD_SIZE )
+#ifdef __riscv_32e
+    #define portCONTEXT_SIZE ( 15 * portWORD_SIZE )
+    #define portCRITICAL_NESTING_OFFSET 13
+    #define portMSTATUS_OFFSET  14
+#else
+    #define portCONTEXT_SIZE ( 31 * portWORD_SIZE )
+    #define portCRITICAL_NESTING_OFFSET 29
+    #define portMSTATUS_OFFSET  30
+#endif
 
   EXTERN pxCurrentTCB
   EXTERN xISRStackTop
@@ -71,6 +79,7 @@ portcontextSAVE_CONTEXT_INTERNAL MACRO
     store_x x13, 10 * portWORD_SIZE( sp )
     store_x x14, 11 * portWORD_SIZE( sp )
     store_x x15, 12 * portWORD_SIZE( sp )
+#ifndef __riscv_32e
     store_x x16, 13 * portWORD_SIZE( sp )
     store_x x17, 14 * portWORD_SIZE( sp )
     store_x x18, 15 * portWORD_SIZE( sp )
@@ -87,12 +96,13 @@ portcontextSAVE_CONTEXT_INTERNAL MACRO
     store_x x29, 26 * portWORD_SIZE( sp )
     store_x x30, 27 * portWORD_SIZE( sp )
     store_x x31, 28 * portWORD_SIZE( sp )
+#endif
 
     load_x  t0, xCriticalNesting         /* Load the value of xCriticalNesting into t0. */
-    store_x t0, 29 * portWORD_SIZE( sp ) /* Store the critical nesting value to the stack. */
+    store_x t0, portCRITICAL_NESTING_OFFSET * portWORD_SIZE( sp ) /* Store the critical nesting value to the stack. */
 
     csrr t0, mstatus                     /* Required for MPIE bit. */
-    store_x t0, 30 * portWORD_SIZE( sp )
+    store_x t0, portMSTATUS_OFFSET * portWORD_SIZE( sp )
 
     portasmSAVE_ADDITIONAL_REGISTERS     /* Defined in freertos_risc_v_chip_specific_extensions.h to save any registers unique to the RISC-V implementation. */
 
@@ -123,7 +133,7 @@ portcontextSAVE_INTERRUPT_CONTEXT MACRO
 
 portcontextRESTORE_CONTEXT MACRO
     load_x  t1, pxCurrentTCB                /* Load pxCurrentTCB. */
-        load_x  sp, 0( t1 )                 /* Read sp from first TCB member. */
+    load_x  sp, 0( t1 )                     /* Read sp from first TCB member. */
 
     /* Load mepc with the address of the instruction in the task to run next. */
     load_x t0, 0( sp )
@@ -133,10 +143,10 @@ portcontextRESTORE_CONTEXT MACRO
     portasmRESTORE_ADDITIONAL_REGISTERS
 
     /* Load mstatus with the interrupt enable bits used by the task. */
-    load_x  t0, 30 * portWORD_SIZE( sp )
+    load_x  t0, portMSTATUS_OFFSET * portWORD_SIZE( sp )
     csrw mstatus, t0                        /* Required for MPIE bit. */
 
-    load_x  t0, 29 * portWORD_SIZE( sp )    /* Obtain xCriticalNesting value for this task from task's stack. */
+    load_x  t0, portCRITICAL_NESTING_OFFSET * portWORD_SIZE( sp )    /* Obtain xCriticalNesting value for this task from task's stack. */
     load_x  t1, pxCriticalNesting           /* Load the address of xCriticalNesting into t1. */
     store_x t0, 0( t1 )                     /* Restore the critical nesting value for this task. */
 
@@ -152,6 +162,7 @@ portcontextRESTORE_CONTEXT MACRO
     load_x  x13, 10 * portWORD_SIZE( sp )
     load_x  x14, 11 * portWORD_SIZE( sp )
     load_x  x15, 12 * portWORD_SIZE( sp )
+#ifndef __riscv_32e
     load_x  x16, 13 * portWORD_SIZE( sp )
     load_x  x17, 14 * portWORD_SIZE( sp )
     load_x  x18, 15 * portWORD_SIZE( sp )
@@ -168,6 +179,7 @@ portcontextRESTORE_CONTEXT MACRO
     load_x  x29, 26 * portWORD_SIZE( sp )
     load_x  x30, 27 * portWORD_SIZE( sp )
     load_x  x31, 28 * portWORD_SIZE( sp )
+#endif
     addi sp, sp, portCONTEXT_SIZE
 
     mret

portable/IAR/RISC-V/portmacro.h

@@ -82,7 +82,11 @@ typedef portUBASE_TYPE TickType_t;
 /* Architecture specifics. */
 #define portSTACK_GROWTH            ( -1 )
 #define portTICK_PERIOD_MS          ( ( TickType_t ) 1000 / configTICK_RATE_HZ )
-#define portBYTE_ALIGNMENT          16
+#ifdef __riscv_32e
+    #define portBYTE_ALIGNMENT          8   /* RV32E uses RISC-V EABI with reduced stack alignment requirements. */
+#else
+    #define portBYTE_ALIGNMENT          16
+#endif
 /*-----------------------------------------------------------*/
 
 /* Scheduler utilities. */