| 32,7 → 32,11 |
|---|
| #include <arch/mm/page.h> |
| #include <align.h> |
| |
| #define STACK_ITEMS 19 |
| |
| #define FRS_TO_SAVE 30 |
| #define STACK_ITEMS (19 + FRS_TO_SAVE*2) |
| //#define STACK_ITEMS 19 |
| /* 30*2 for FPU registers */ |
| #define STACK_FRAME_SIZE ALIGN_UP((STACK_ITEMS*STACK_ITEM_SIZE) + STACK_SCRATCH_AREA_SIZE, STACK_ALIGNMENT) |
| |
| #if (STACK_ITEMS % 2 == 0) |
| 78,6 → 82,19 |
|---|
| /* |
| * Note that r24-r31 from bank 0 can be used only as long as PSR.ic = 0. |
| */ |
| |
| /*Set up FPU as in interrupred*/ |
| mov r24=psr |
| mov r25=cr.ipsr |
| mov r26=(PSR_DFH_MASK) |
| mov r27=(~(PSR_DFH_MASK));; |
| and r26=r25,r26 |
| and r24=r24,r27;; |
| or r24=r24,r26;; |
| mov psr.l=r24;; |
| srlz.i |
| srlz.d;; |
| |
| mov r24 = cr.iip |
| mov r25 = cr.ipsr |
| mov r26 = cr.iipa |
| 195,7 → 212,7 |
|---|
| /* 16. RSE switch to interrupted context */ |
| cover /* allocate zerro size frame (step 1 (from Intel Docs)) */ |
| |
| add r31 = STACK_SCRATCH_AREA_SIZE, r12 ;; |
| add r31 = (STACK_SCRATCH_AREA_SIZE+(FRS_TO_SAVE*2*8)), r12 ;; |
| |
| ld8 r30 = [r31], +8 ;; /* load ar.bsp */ |
| ld8 r29 = [r31], +8 ;; /* load ar.bspstore */ |
| 230,11 → 247,19 |
|---|
| ld8 r25 = [r31], +8 ;; /* load cr.ipsr */ |
| ld8 r24 = [r31], +8 ;; /* load cr.iip */ |
| |
| mov cr.iip = r24 |
| mov cr.ipsr = r25 |
| mov cr.iip = r24;; |
| mov cr.iipa = r26 |
| mov cr.isr = r27 |
| mov cr.ifa = r28 |
| /*Set up FPU as in exception*/ |
| mov r24=psr |
| mov r26=(PSR_DFH_MASK) |
| mov r27=(~(PSR_DFH_MASK));; |
| and r25=r25,r27 |
| and r24=r24,r26;; |
| or r25=r25,r24;; |
| mov cr.ipsr = r25 |
| |
| |
| /* 18. restore predicate registers from memory stack */ |
| ld8 r29 = [r31], +8 ;; /* load predicate registers */ |
| 315,6 → 340,57 |
|---|
| mov loc45 = r30 |
| mov loc46 = r31 |
| |
| mov r24=96 + STACK_SCRATCH_AREA_SIZE |
| mov r25=112 + STACK_SCRATCH_AREA_SIZE |
| mov r26=0 + STACK_SCRATCH_AREA_SIZE |
| mov r27=16 + STACK_SCRATCH_AREA_SIZE |
| mov r28=32 + STACK_SCRATCH_AREA_SIZE |
| mov r29=48 + STACK_SCRATCH_AREA_SIZE |
| mov r30=64 + STACK_SCRATCH_AREA_SIZE |
| mov r31=80 + STACK_SCRATCH_AREA_SIZE;; |
| add r24=r12,r24 |
| add r25=r12,r25 |
| add r26=r12,r26 |
| add r27=r12,r27 |
| add r28=r12,r28 |
| add r29=r12,r29 |
| add r30=r12,r30 |
| add r31=r12,r31;; |
| |
| stf.spill [r26]=f2,0x80 |
| stf.spill [r27]=f3,0x80 |
| stf.spill [r28]=f4,0x80 |
| stf.spill [r29]=f5,0x80 |
| stf.spill [r30]=f6,0x80 |
| stf.spill [r31]=f7,0x80;; |
| |
| stf.spill [r24]=f8,0x80 |
| stf.spill [r25]=f9,0x80 |
| stf.spill [r26]=f10,0x80 |
| stf.spill [r27]=f11,0x80 |
| stf.spill [r28]=f12,0x80 |
| stf.spill [r29]=f13,0x80 |
| stf.spill [r30]=f14,0x80 |
| stf.spill [r31]=f15,0x80;; |
| |
| stf.spill [r24]=f16,0x80 |
| stf.spill [r25]=f17,0x80 |
| stf.spill [r26]=f18,0x80 |
| stf.spill [r27]=f19,0x80 |
| stf.spill [r28]=f20,0x80 |
| stf.spill [r29]=f21,0x80 |
| stf.spill [r30]=f22,0x80 |
| stf.spill [r31]=f23,0x80;; |
| |
| stf.spill [r24]=f24,0x80 |
| stf.spill [r25]=f25,0x80 |
| stf.spill [r26]=f26,0x80 |
| stf.spill [r27]=f27,0x80 |
| stf.spill [r28]=f28,0x80 |
| stf.spill [r29]=f29,0x80 |
| stf.spill [r30]=f30,0x80 |
| stf.spill [r31]=f31,0x80;; |
| |
| /* preserve Floating point status register */ |
| mov loc47 = ar.fpsr |
| |
| 343,6 → 419,59 |
|---|
| |
| /* 13. restore general and floating-point registers */ |
| /* TODO: restore floating-point context */ |
| mov r24=96 + STACK_SCRATCH_AREA_SIZE |
| mov r25=112 + STACK_SCRATCH_AREA_SIZE |
| mov r26=0 + STACK_SCRATCH_AREA_SIZE |
| mov r27=16 + STACK_SCRATCH_AREA_SIZE |
| mov r28=32 + STACK_SCRATCH_AREA_SIZE |
| mov r29=48 + STACK_SCRATCH_AREA_SIZE |
| mov r30=64 + STACK_SCRATCH_AREA_SIZE |
| mov r31=80 + STACK_SCRATCH_AREA_SIZE;; |
| add r24=r12,r24 |
| add r25=r12,r25 |
| add r26=r12,r26 |
| add r27=r12,r27 |
| add r28=r12,r28 |
| add r29=r12,r29 |
| add r30=r12,r30 |
| add r31=r12,r31;; |
| |
| |
| ldf.fill f2=[r26],0x80 |
| ldf.fill f3=[r27],0x80 |
| ldf.fill f4=[r28],0x80 |
| ldf.fill f5=[r29],0x80 |
| ldf.fill f6=[r30],0x80 |
| ldf.fill f7=[r31],0x80;; |
| |
| ldf.fill f8=[r24],0x80 |
| ldf.fill f9=[r25],0x80 |
| ldf.fill f10=[r26],0x80 |
| ldf.fill f11=[r27],0x80 |
| ldf.fill f12=[r28],0x80 |
| ldf.fill f13=[r29],0x80 |
| ldf.fill f14=[r30],0x80 |
| ldf.fill f15=[r31],0x80;; |
| |
| ldf.fill f16=[r24],0x80 |
| ldf.fill f17=[r25],0x80 |
| ldf.fill f18=[r26],0x80 |
| ldf.fill f19=[r27],0x80 |
| ldf.fill f20=[r28],0x80 |
| ldf.fill f21=[r29],0x80 |
| ldf.fill f22=[r30],0x80 |
| ldf.fill f23=[r31],0x80;; |
| |
| ldf.fill f24=[r24],0x80 |
| ldf.fill f25=[r25],0x80 |
| ldf.fill f26=[r26],0x80 |
| ldf.fill f27=[r27],0x80 |
| ldf.fill f28=[r28],0x80 |
| ldf.fill f29=[r29],0x80 |
| ldf.fill f30=[r30],0x80 |
| ldf.fill f31=[r31],0x80;; |
| |
| |
| mov r1 = loc17 |
| mov r2 = loc18 |
| mov r3 = loc19 |
| 478,3 → 607,7 |
|---|
| HEAVYWEIGHT_HANDLER 0x7d00 |
| HEAVYWEIGHT_HANDLER 0x7e00 |
| HEAVYWEIGHT_HANDLER 0x7f00 |
| |
| |
| |
| |