/// @file\r
-/// Contains an implementation of setjmp and longjmp for the\r
-/// Itanium-based architecture.\r
+/// Contains an implementation of longjmp for the Itanium-based architecture.\r
///\r
-/// Copyright (c) 2006, Intel Corporation \r
-/// All rights reserved. This program and the accompanying materials \r
-/// are licensed and made available under the terms and conditions of the BSD License \r
-/// which accompanies this distribution. The full text of the license may be found at \r
-/// http://opensource.org/licenses/bsd-license.php \r
-/// \r
-/// THE PROGRAM IS DISTRIBUTED UNDER THE BSD LICENSE ON AN "AS IS" BASIS, \r
-/// WITHOUT WARRANTIES OR REPRESENTATIONS OF ANY KIND, EITHER EXPRESS OR IMPLIED. \r
-/// \r
-/// Module Name: setjmp.s\r
+/// Copyright (c) 2006, Intel Corporation\r
+/// All rights reserved. This program and the accompanying materials\r
+/// are licensed and made available under the terms and conditions of the BSD License\r
+/// which accompanies this distribution. The full text of the license may be found at\r
+/// http://opensource.org/licenses/bsd-license.php\r
+///\r
+/// THE PROGRAM IS DISTRIBUTED UNDER THE BSD LICENSE ON AN "AS IS" BASIS,\r
+/// WITHOUT WARRANTIES OR REPRESENTATIONS OF ANY KIND, EITHER EXPRESS OR IMPLIED.\r
+///\r
+/// Module Name: longjmp.s\r
///\r
///\r
\r
- .file "setjmp.s"\r
+.auto\r
+.text\r
\r
-#include "asm.h"\r
-#include "ia_64gen.h"\r
+.global InternalAssertJumpBuffer\r
+.type InternalAssertJumpBuffer, @function\r
\r
-/// int SetJump(struct jmp_buffer save)\r
-///\r
-/// Setup a non-local goto.\r
-///\r
-/// Description:\r
-///\r
-/// SetJump stores the current register set in the area pointed to\r
-/// by "save". It returns zero. Subsequent calls to "LongJump" will\r
-/// restore the registers and return non-zero to the same location.\r
-///\r
-/// On entry, r32 contains the pointer to the jmp_buffer\r
-///\r
+.proc SetJump\r
+.type SetJump, @function\r
+SetJump::\r
+ alloc loc0 = ar.pfs, 1, 2, 1, 0\r
+ mov loc1 = b0\r
+ mov out0 = in0\r
+\r
+ brl.call.sptk.many b0 = InternalAssertJumpBuffer\r
+\r
+ mov r14 = ar.unat\r
+ mov r15 = ar.bsp\r
+ add r10 = 0x10*20, in0\r
+\r
+ stf.spill.nta [in0] = f2, 0x10\r
+ st8.spill.nta [r10] = r4, 8\r
+ mov r21 = b1\r
+\r
+ stf.spill.nta [in0] = f3, 0x10\r
+ st8.spill.nta [r10] = r5, 8\r
+ mov r22 = b2\r
\r
-PROCEDURE_ENTRY(SetJump)\r
- //\r
- // Make sure buffer is aligned at 16byte boundary\r
- //\r
- mov r32 = r33 \r
+ stf.spill.nta [in0] = f4, 0x10\r
+ st8.spill.nta [r10] = r6, 8\r
+ mov r23 = b3\r
\r
- add r10 = -0x10,r0 ;; // mask the lower 4 bits\r
- and r32 = r32, r10;; \r
- add r32 = 0x10, r32;; // move to next 16 byte boundary\r
+ stf.spill.nta [in0] = f5, 0x10\r
+ st8.spill.nta [r10] = r7, 8\r
+ mov r24 = b4\r
\r
- add r10 = J_PREDS, r32 // skip Unats & pfs save area\r
- add r11 = J_BSP, r32\r
- //\r
- // save immediate context\r
- //\r
- mov r2 = ar.bsp // save backing store pointer\r
- mov r3 = pr // save predicates\r
- ;;\r
- //\r
- // save user Unat register\r
- //\r
- mov r16 = ar.lc // save loop count register\r
- mov r14 = ar.unat // save user Unat register\r
+ stf.spill.nta [in0] = f16, 0x10\r
+ st8.spill.nta [r10] = sp, 8\r
+ mov r25 = b5\r
\r
- st8 [r10] = r3, J_LC-J_PREDS\r
- st8 [r11] = r2, J_R4-J_BSP\r
- ;;\r
- st8 [r10] = r16, J_R5-J_LC\r
- st8 [r32] = r14, J_NATS // Note: Unat at the \r
- // beginning of the save area\r
- mov r15 = ar.pfs\r
- ;;\r
- //\r
- // save preserved general registers & NaT's\r
- //\r
- st8.spill [r11] = r4, J_R6-J_R4\r
- ;;\r
- st8.spill [r10] = r5, J_R7-J_R5 \r
- ;;\r
- st8.spill [r11] = r6, J_SP-J_R6\r
- ;;\r
- st8.spill [r10] = r7, J_F3-J_R7 \r
- ;;\r
- st8.spill [r11] = sp, J_F2-J_SP\r
- ;;\r
- //\r
- // save spilled Unat and pfs registers\r
- //\r
- mov r2 = ar.unat // save Unat register after spill\r
- ;;\r
- st8 [r32] = r2, J_PFS-J_NATS // save unat for spilled regs\r
- ;;\r
- st8 [r32] = r15 // save pfs\r
- //\r
- // save floating registers \r
- //\r
- stf.spill [r11] = f2, J_F4-J_F2\r
- stf.spill [r10] = f3, J_F5-J_F3 \r
- ;;\r
- stf.spill [r11] = f4, J_F16-J_F4\r
- stf.spill [r10] = f5, J_F17-J_F5 \r
- ;;\r
- stf.spill [r11] = f16, J_F18-J_F16\r
- stf.spill [r10] = f17, J_F19-J_F17 \r
- ;;\r
- stf.spill [r11] = f18, J_F20-J_F18\r
- stf.spill [r10] = f19, J_F21-J_F19 \r
- ;;\r
- stf.spill [r11] = f20, J_F22-J_F20\r
- stf.spill [r10] = f21, J_F23-J_F21 \r
- ;;\r
- stf.spill [r11] = f22, J_F24-J_F22\r
- stf.spill [r10] = f23, J_F25-J_F23 \r
- ;;\r
- stf.spill [r11] = f24, J_F26-J_F24\r
- stf.spill [r10] = f25, J_F27-J_F25 \r
- ;;\r
- stf.spill [r11] = f26, J_F28-J_F26\r
- stf.spill [r10] = f27, J_F29-J_F27 \r
- ;;\r
- stf.spill [r11] = f28, J_F30-J_F28\r
- stf.spill [r10] = f29, J_F31-J_F29 \r
- ;;\r
- stf.spill [r11] = f30, J_FPSR-J_F30\r
- stf.spill [r10] = f31, J_B0-J_F31 // size of f31 + fpsr\r
- //\r
- // save FPSR register & branch registers\r
- //\r
- mov r2 = ar.fpsr // save fpsr register\r
- mov r3 = b0 \r
- ;;\r
- st8 [r11] = r2, J_B1-J_FPSR\r
- st8 [r10] = r3, J_B2-J_B0\r
- mov r2 = b1\r
- mov r3 = b2 \r
- ;;\r
- st8 [r11] = r2, J_B3-J_B1\r
- st8 [r10] = r3, J_B4-J_B2\r
- mov r2 = b3\r
- mov r3 = b4 \r
- ;;\r
- st8 [r11] = r2, J_B5-J_B3\r
- st8 [r10] = r3\r
- mov r2 = b5 \r
- ;;\r
- st8 [r11] = r2\r
- ;;\r
- //\r
- // return\r
- //\r
- mov r8 = r0 // return 0 from setjmp\r
- mov ar.unat = r14 // restore unat\r
- br.ret.sptk b0\r
+ stf.spill.nta [in0] = f17, 0x10\r
+ st8.nta [r10] = loc1, 8\r
+ mov r16 = pr\r
\r
-PROCEDURE_EXIT(SetJump)\r
+ stf.spill.nta [in0] = f18, 0x10\r
+ st8.nta [r10] = r21, 8\r
+ mov r17 = ar.lc\r
\r
+ stf.spill.nta [in0] = f19, 0x10\r
+ st8.nta [r10] = r22, 8\r
\r
-//\r
-// void _LongJump(struct jmp_buffer *)\r
-//\r
-// Perform a non-local goto.\r
-//\r
-// Description:\r
-//\r
-// LongJump initializes the register set to the values saved by a\r
-// previous 'SetJump' and jumps to the return location saved by that\r
-// 'SetJump'. This has the effect of unwinding the stack and returning\r
-// for a second time to the 'SetJump'.\r
-//\r
+ stf.spill.nta [in0] = f20, 0x10\r
+ st8.nta [r10] = r23, 8\r
\r
-PROCEDURE_ENTRY(_LongJump)\r
- //\r
- // Make sure buffer is aligned at 16byte boundary\r
- //\r
- mov r32 = r33 \r
+ stf.spill.nta [in0] = f21, 0x10\r
+ st8.nta [r10] = r24, 8\r
\r
- add r10 = -0x10,r0 ;; // mask the lower 4 bits\r
- and r32 = r32, r10;; \r
- add r32 = 0x10, r32;; // move to next 16 byte boundary\r
+ stf.spill.nta [in0] = f22, 0x10\r
+ st8.nta [r10] = r25, 8\r
\r
- //\r
- // caching the return value as we do invala in the end\r
- //\r
-/// mov r8 = r33 // return value\r
- mov r8 = 1 // For now return hard coded 1\r
+ stf.spill.nta [in0] = f23, 0x10\r
+ mov r18 = ar.unat\r
\r
- //\r
- // get immediate context\r
- //\r
- mov r14 = ar.rsc // get user RSC conf \r
- add r10 = J_PFS, r32 // get address of pfs\r
- add r11 = J_NATS, r32\r
- ;;\r
- ld8 r15 = [r10], J_BSP-J_PFS // get pfs\r
- ld8 r2 = [r11], J_LC-J_NATS // get unat for spilled regs\r
- ;;\r
- mov ar.unat = r2\r
- ;;\r
- ld8 r16 = [r10], J_PREDS-J_BSP // get backing store pointer\r
- mov ar.rsc = r0 // put RSE in enforced lazy \r
- mov ar.pfs = r15\r
- ;;\r
- \r
- //\r
- // while returning from longjmp the BSPSTORE and BSP needs to be\r
- // same and discard all the registers allocated after we did\r
- // setjmp. Also, we need to generate the RNAT register since we\r
- // did not flushed the RSE on setjmp.\r
- //\r
- mov r17 = ar.bspstore // get current BSPSTORE\r
- ;;\r
- cmp.ltu p6,p7 = r17, r16 // is it less than BSP of \r
-(p6) br.spnt.few .flush_rse\r
- mov r19 = ar.rnat // get current RNAT\r
- ;;\r
- loadrs // invalidate dirty regs\r
- br.sptk.many .restore_rnat // restore RNAT\r
+ stf.spill.nta [in0] = f24, 0x10\r
+ st8.nta [r10] = r14, 8 // UNAT\r
\r
-.flush_rse:\r
- flushrs\r
- ;;\r
- mov r19 = ar.rnat // get current RNAT\r
- mov r17 = r16 // current BSPSTORE\r
- ;;\r
-.restore_rnat:\r
- //\r
- // check if RNAT is saved between saved BSP and curr BSPSTORE\r
- //\r
- dep r18 = 1,r16,3,6 // get RNAT address\r
- ;;\r
- cmp.ltu p8,p9 = r18, r17 // RNAT saved on RSE\r
- ;;\r
-(p8) ld8 r19 = [r18] // get RNAT from RSE\r
- ;;\r
- mov ar.bspstore = r16 // set new BSPSTORE \r
- ;;\r
- mov ar.rnat = r19 // restore RNAT\r
- mov ar.rsc = r14 // restore RSC conf\r
+ stf.spill.nta [in0] = f25, 0x10\r
+ st8.nta [r10] = r18, 8 // UNAT after spill\r
\r
+ stf.spill.nta [in0] = f26, 0x10\r
+ st8.nta [r10] = loc0, 8 // PFS\r
\r
- ld8 r3 = [r11], J_R4-J_LC // get lc register\r
- ld8 r2 = [r10], J_R5-J_PREDS // get predicates\r
- ;;\r
- mov pr = r2, -1\r
- mov ar.lc = r3\r
- //\r
- // restore preserved general registers & NaT's\r
- //\r
- ld8.fill r4 = [r11], J_R6-J_R4\r
- ;;\r
- ld8.fill r5 = [r10], J_R7-J_R5 \r
- ld8.fill r6 = [r11], J_SP-J_R6\r
- ;;\r
- ld8.fill r7 = [r10], J_F2-J_R7\r
- ld8.fill sp = [r11], J_F3-J_SP\r
- ;;\r
- //\r
- // restore floating registers \r
- //\r
- ldf.fill f2 = [r10], J_F4-J_F2\r
- ldf.fill f3 = [r11], J_F5-J_F3 \r
- ;;\r
- ldf.fill f4 = [r10], J_F16-J_F4\r
- ldf.fill f5 = [r11], J_F17-J_F5 \r
- ;;\r
- ldf.fill f16 = [r10], J_F18-J_F16\r
- ldf.fill f17 = [r11], J_F19-J_F17\r
- ;;\r
- ldf.fill f18 = [r10], J_F20-J_F18\r
- ldf.fill f19 = [r11], J_F21-J_F19\r
- ;;\r
- ldf.fill f20 = [r10], J_F22-J_F20\r
- ldf.fill f21 = [r11], J_F23-J_F21\r
- ;;\r
- ldf.fill f22 = [r10], J_F24-J_F22\r
- ldf.fill f23 = [r11], J_F25-J_F23 \r
- ;;\r
- ldf.fill f24 = [r10], J_F26-J_F24\r
- ldf.fill f25 = [r11], J_F27-J_F25\r
- ;;\r
- ldf.fill f26 = [r10], J_F28-J_F26\r
- ldf.fill f27 = [r11], J_F29-J_F27\r
- ;;\r
- ldf.fill f28 = [r10], J_F30-J_F28\r
- ldf.fill f29 = [r11], J_F31-J_F29 \r
- ;;\r
- ldf.fill f30 = [r10], J_FPSR-J_F30\r
- ldf.fill f31 = [r11], J_B0-J_F31 ;;\r
+ stf.spill.nta [in0] = f27, 0x10\r
+ st8.nta [r10] = r15, 8 // BSP\r
+ mov r8 = 0\r
\r
- //\r
- // restore branch registers and fpsr\r
- //\r
- ld8 r16 = [r10], J_B1-J_FPSR // get fpsr\r
- ld8 r17 = [r11], J_B2-J_B0 // get return pointer\r
- ;;\r
- mov ar.fpsr = r16\r
- mov b0 = r17\r
- ld8 r2 = [r10], J_B3-J_B1\r
- ld8 r3 = [r11], J_B4-J_B2\r
- ;;\r
- mov b1 = r2\r
- mov b2 = r3\r
- ld8 r2 = [r10], J_B5-J_B3\r
- ld8 r3 = [r11]\r
- ;;\r
- mov b3 = r2\r
- mov b4 = r3 \r
- ld8 r2 = [r10]\r
- ld8 r21 = [r32] // get user unat\r
- ;;\r
- mov b5 = r2\r
- mov ar.unat = r21\r
+ stf.spill.nta [in0] = f28, 0x10\r
+ mov r19 = ar.fpsr\r
\r
- //\r
- // invalidate ALAT\r
- //\r
- invala ;;\r
+ stf.spill.nta [in0] = f29, 0x10\r
+ st8.nta [r10] = r16, 8 // PR\r
+ mov ar.pfs = loc0\r
\r
- br.ret.sptk b0\r
-PROCEDURE_EXIT(_LongJump)\r
+ stf.spill.nta [in0] = f30, 0x10\r
+ st8.nta [r10] = r17, 8 // LC\r
+ mov b0 = loc1\r
\r
+ stf.spill.nta [in0] = f31, 0x10\r
+ st8.nta [r10] = r19 // FPSR\r
\r
+ mov ar.unat = r14\r
+ br.ret.sptk b0\r
+.endp SetJump\r
projects / mirror_edk2.git / blobdiff