/* * CDDL HEADER START * * The contents of this file are subject to the terms of the * Common Development and Distribution License (the "License"). * You may not use this file except in compliance with the License. * * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE * or http://www.opensolaris.org/os/licensing. * See the License for the specific language governing permissions * and limitations under the License. * * When distributing Covered Code, include this CDDL HEADER in each * file and include the License file at usr/src/OPENSOLARIS.LICENSE. * If applicable, add the following below this CDDL HEADER, with the * fields enclosed by brackets "[]" replaced with your own identifying * information: Portions Copyright [yyyy] [name of copyright owner] * * CDDL HEADER END */ /* * Copyright 2007 Sun Microsystems, Inc. All rights reserved. * Use is subject to license terms. * * Assembly code support for the Olympus-C module */ #pragma ident "%Z%%M% %I% %E% SMI" #if !defined(lint) #include "assym.h" #endif /* lint */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef TRAPTRACE #include #endif /* TRAPTRACE */ /* * Macro that flushes the entire Ecache. * * arg1 = ecache size * arg2 = ecache linesize * arg3 = ecache flush address - Not used for olympus-C */ #define ECACHE_FLUSHALL(arg1, arg2, arg3, tmp1) \ mov ASI_L2_CTRL_U2_FLUSH, arg1; \ mov ASI_L2_CTRL_RW_ADDR, arg2; \ stxa arg1, [arg2]ASI_L2_CTRL /* * SPARC64-VI MMU and Cache operations. */ #if defined(lint) /* ARGSUSED */ void vtag_flushpage(caddr_t vaddr, uint64_t sfmmup) {} #else /* lint */ ENTRY_NP(vtag_flushpage) /* * flush page from the tlb * * %o0 = vaddr * %o1 = sfmmup */ rdpr %pstate, %o5 #ifdef DEBUG PANIC_IF_INTR_DISABLED_PSTR(%o5, opl_di_l3, %g1) #endif /* DEBUG */ /* * disable ints */ andn %o5, PSTATE_IE, %o4 wrpr %o4, 0, %pstate /* * Then, blow out the tlb * Interrupts are disabled to prevent the primary ctx register * from changing underneath us. */ sethi %hi(ksfmmup), %o3 ldx [%o3 + %lo(ksfmmup)], %o3 cmp %o3, %o1 bne,pt %xcc, 1f ! if not kernel as, go to 1 sethi %hi(FLUSH_ADDR), %o3 /* * For Kernel demaps use primary. type = page implicitly */ stxa %g0, [%o0]ASI_DTLB_DEMAP /* dmmu flush for KCONTEXT */ stxa %g0, [%o0]ASI_ITLB_DEMAP /* immu flush for KCONTEXT */ flush %o3 retl wrpr %g0, %o5, %pstate /* enable interrupts */ 1: /* * User demap. We need to set the primary context properly. * Secondary context cannot be used for SPARC64-VI IMMU. * %o0 = vaddr * %o1 = sfmmup * %o3 = FLUSH_ADDR */ SFMMU_CPU_CNUM(%o1, %g1, %g2) ! %g1 = sfmmu cnum on this CPU ldub [%o1 + SFMMU_CEXT], %o4 ! %o4 = sfmmup->sfmmu_cext sll %o4, CTXREG_EXT_SHIFT, %o4 or %g1, %o4, %g1 ! %g1 = primary pgsz | cnum wrpr %g0, 1, %tl set MMU_PCONTEXT, %o4 or DEMAP_PRIMARY | DEMAP_PAGE_TYPE, %o0, %o0 ldxa [%o4]ASI_DMMU, %o2 ! %o2 = save old ctxnum srlx %o2, CTXREG_NEXT_SHIFT, %o1 ! need to preserve nucleus pgsz sllx %o1, CTXREG_NEXT_SHIFT, %o1 ! %o1 = nucleus pgsz or %g1, %o1, %g1 ! %g1 = nucleus pgsz | primary pgsz | cnum stxa %g1, [%o4]ASI_DMMU ! wr new ctxum stxa %g0, [%o0]ASI_DTLB_DEMAP stxa %g0, [%o0]ASI_ITLB_DEMAP stxa %o2, [%o4]ASI_DMMU /* restore old ctxnum */ flush %o3 wrpr %g0, 0, %tl retl wrpr %g0, %o5, %pstate /* enable interrupts */ SET_SIZE(vtag_flushpage) #endif /* lint */ #if defined(lint) void vtag_flushall(void) {} #else /* lint */ ENTRY_NP2(vtag_flushall, demap_all) /* * flush the tlb */ sethi %hi(FLUSH_ADDR), %o3 set DEMAP_ALL_TYPE, %g1 stxa %g0, [%g1]ASI_DTLB_DEMAP stxa %g0, [%g1]ASI_ITLB_DEMAP flush %o3 retl nop SET_SIZE(demap_all) SET_SIZE(vtag_flushall) #endif /* lint */ #if defined(lint) /* ARGSUSED */ void vtag_flushpage_tl1(uint64_t vaddr, uint64_t sfmmup) {} #else /* lint */ ENTRY_NP(vtag_flushpage_tl1) /* * x-trap to flush page from tlb and tsb * * %g1 = vaddr, zero-extended on 32-bit kernel * %g2 = sfmmup * * assumes TSBE_TAG = 0 */ srln %g1, MMU_PAGESHIFT, %g1 sethi %hi(ksfmmup), %g3 ldx [%g3 + %lo(ksfmmup)], %g3 cmp %g3, %g2 bne,pt %xcc, 1f ! if not kernel as, go to 1 slln %g1, MMU_PAGESHIFT, %g1 /* g1 = vaddr */ /* We need to demap in the kernel context */ or DEMAP_NUCLEUS | DEMAP_PAGE_TYPE, %g1, %g1 stxa %g0, [%g1]ASI_DTLB_DEMAP stxa %g0, [%g1]ASI_ITLB_DEMAP retry 1: /* We need to demap in a user context */ or DEMAP_PRIMARY | DEMAP_PAGE_TYPE, %g1, %g1 SFMMU_CPU_CNUM(%g2, %g6, %g3) ! %g6 = sfmmu cnum on this CPU ldub [%g2 + SFMMU_CEXT], %g4 ! %g4 = sfmmup->cext sll %g4, CTXREG_EXT_SHIFT, %g4 or %g6, %g4, %g6 ! %g6 = primary pgsz | cnum set MMU_PCONTEXT, %g4 ldxa [%g4]ASI_DMMU, %g5 ! %g5 = save old ctxnum srlx %g5, CTXREG_NEXT_SHIFT, %g2 ! %g2 = nucleus pgsz sllx %g2, CTXREG_NEXT_SHIFT, %g2 ! preserve nucleus pgsz or %g6, %g2, %g6 ! %g6 = nucleus pgsz | primary pgsz | cnum stxa %g6, [%g4]ASI_DMMU ! wr new ctxum stxa %g0, [%g1]ASI_DTLB_DEMAP stxa %g0, [%g1]ASI_ITLB_DEMAP stxa %g5, [%g4]ASI_DMMU ! restore old ctxnum retry SET_SIZE(vtag_flushpage_tl1) #endif /* lint */ #if defined(lint) /* ARGSUSED */ void vtag_flush_pgcnt_tl1(uint64_t vaddr, uint64_t sfmmup_pgcnt) {} #else /* lint */ ENTRY_NP(vtag_flush_pgcnt_tl1) /* * x-trap to flush pgcnt MMU_PAGESIZE pages from tlb * * %g1 = vaddr, zero-extended on 32-bit kernel * %g2 = * * NOTE: this handler relies on the fact that no * interrupts or traps can occur during the loop * issuing the TLB_DEMAP operations. It is assumed * that interrupts are disabled and this code is * fetching from the kernel locked text address. * * assumes TSBE_TAG = 0 */ set SFMMU_PGCNT_MASK, %g4 and %g4, %g2, %g3 /* g3 = pgcnt - 1 */ add %g3, 1, %g3 /* g3 = pgcnt */ andn %g2, SFMMU_PGCNT_MASK, %g2 /* g2 = sfmmup */ srln %g1, MMU_PAGESHIFT, %g1 sethi %hi(ksfmmup), %g4 ldx [%g4 + %lo(ksfmmup)], %g4 cmp %g4, %g2 bne,pn %xcc, 1f /* if not kernel as, go to 1 */ slln %g1, MMU_PAGESHIFT, %g1 /* g1 = vaddr */ /* We need to demap in the kernel context */ or DEMAP_NUCLEUS | DEMAP_PAGE_TYPE, %g1, %g1 set MMU_PAGESIZE, %g2 /* g2 = pgsize */ sethi %hi(FLUSH_ADDR), %g5 4: stxa %g0, [%g1]ASI_DTLB_DEMAP stxa %g0, [%g1]ASI_ITLB_DEMAP flush %g5 ! flush required by immu deccc %g3 /* decr pgcnt */ bnz,pt %icc,4b add %g1, %g2, %g1 /* next page */ retry 1: /* * We need to demap in a user context * * g2 = sfmmup * g3 = pgcnt */ SFMMU_CPU_CNUM(%g2, %g5, %g6) ! %g5 = sfmmu cnum on this CPU or DEMAP_PRIMARY | DEMAP_PAGE_TYPE, %g1, %g1 ldub [%g2 + SFMMU_CEXT], %g4 ! %g4 = sfmmup->cext sll %g4, CTXREG_EXT_SHIFT, %g4 or %g5, %g4, %g5 set MMU_PCONTEXT, %g4 ldxa [%g4]ASI_DMMU, %g6 /* rd old ctxnum */ srlx %g6, CTXREG_NEXT_SHIFT, %g2 /* %g2 = nucleus pgsz */ sllx %g2, CTXREG_NEXT_SHIFT, %g2 /* preserve nucleus pgsz */ or %g5, %g2, %g5 /* %g5 = nucleus pgsz | primary pgsz | cnum */ stxa %g5, [%g4]ASI_DMMU /* wr new ctxum */ set MMU_PAGESIZE, %g2 /* g2 = pgsize */ sethi %hi(FLUSH_ADDR), %g5 3: stxa %g0, [%g1]ASI_DTLB_DEMAP stxa %g0, [%g1]ASI_ITLB_DEMAP flush %g5 ! flush required by immu deccc %g3 /* decr pgcnt */ bnz,pt %icc,3b add %g1, %g2, %g1 /* next page */ stxa %g6, [%g4]ASI_DMMU /* restore old ctxnum */ retry SET_SIZE(vtag_flush_pgcnt_tl1) #endif /* lint */ #if defined(lint) /*ARGSUSED*/ void vtag_flushall_tl1(uint64_t dummy1, uint64_t dummy2) {} #else /* lint */ ENTRY_NP(vtag_flushall_tl1) /* * x-trap to flush tlb */ set DEMAP_ALL_TYPE, %g4 stxa %g0, [%g4]ASI_DTLB_DEMAP stxa %g0, [%g4]ASI_ITLB_DEMAP retry SET_SIZE(vtag_flushall_tl1) #endif /* lint */ /* * VAC (virtual address conflict) does not apply to OPL. * VAC resolution is managed by the Olympus processor hardware. * As a result, all OPL VAC flushing routines are no-ops. */ #if defined(lint) /* ARGSUSED */ void vac_flushpage(pfn_t pfnum, int vcolor) {} #else /* lint */ ENTRY(vac_flushpage) retl nop SET_SIZE(vac_flushpage) #endif /* lint */ #if defined(lint) /* ARGSUSED */ void vac_flushpage_tl1(uint64_t pfnum, uint64_t vcolor) {} #else /* lint */ ENTRY_NP(vac_flushpage_tl1) retry SET_SIZE(vac_flushpage_tl1) #endif /* lint */ #if defined(lint) /* ARGSUSED */ void vac_flushcolor(int vcolor, pfn_t pfnum) {} #else /* lint */ ENTRY(vac_flushcolor) retl nop SET_SIZE(vac_flushcolor) #endif /* lint */ #if defined(lint) /* ARGSUSED */ void vac_flushcolor_tl1(uint64_t vcolor, uint64_t pfnum) {} #else /* lint */ ENTRY(vac_flushcolor_tl1) retry SET_SIZE(vac_flushcolor_tl1) #endif /* lint */ #if defined(lint) int idsr_busy(void) { return (0); } #else /* lint */ /* * Determine whether or not the IDSR is busy. * Entry: no arguments * Returns: 1 if busy, 0 otherwise */ ENTRY(idsr_busy) ldxa [%g0]ASI_INTR_DISPATCH_STATUS, %g1 clr %o0 btst IDSR_BUSY, %g1 bz,a,pt %xcc, 1f mov 1, %o0 1: retl nop SET_SIZE(idsr_busy) #endif /* lint */ #if defined(lint) /* ARGSUSED */ void init_mondo(xcfunc_t *func, uint64_t arg1, uint64_t arg2) {} /* ARGSUSED */ void init_mondo_nocheck(xcfunc_t *func, uint64_t arg1, uint64_t arg2) {} #else /* lint */ .global _dispatch_status_busy _dispatch_status_busy: .asciz "ASI_INTR_DISPATCH_STATUS error: busy" .align 4 /* * Setup interrupt dispatch data registers * Entry: * %o0 - function or inumber to call * %o1, %o2 - arguments (2 uint64_t's) */ .seg "text" ENTRY(init_mondo) #ifdef DEBUG ! ! IDSR should not be busy at the moment ! ldxa [%g0]ASI_INTR_DISPATCH_STATUS, %g1 btst IDSR_BUSY, %g1 bz,pt %xcc, 1f nop sethi %hi(_dispatch_status_busy), %o0 call panic or %o0, %lo(_dispatch_status_busy), %o0 #endif /* DEBUG */ ALTENTRY(init_mondo_nocheck) ! ! interrupt vector dispatch data reg 0 ! 1: mov IDDR_0, %g1 mov IDDR_1, %g2 mov IDDR_2, %g3 stxa %o0, [%g1]ASI_INTR_DISPATCH ! ! interrupt vector dispatch data reg 1 ! stxa %o1, [%g2]ASI_INTR_DISPATCH ! ! interrupt vector dispatch data reg 2 ! stxa %o2, [%g3]ASI_INTR_DISPATCH membar #Sync retl nop SET_SIZE(init_mondo_nocheck) SET_SIZE(init_mondo) #endif /* lint */ #if defined(lint) /* ARGSUSED */ void shipit(int upaid, int bn) { return; } #else /* lint */ /* * Ship mondo to aid using busy/nack pair bn */ ENTRY_NP(shipit) sll %o0, IDCR_PID_SHIFT, %g1 ! IDCR<23:14> = agent id sll %o1, IDCR_BN_SHIFT, %g2 ! IDCR<28:24> = b/n pair or %g1, IDCR_OFFSET, %g1 ! IDCR<13:0> = 0x70 or %g1, %g2, %g1 stxa %g0, [%g1]ASI_INTR_DISPATCH ! interrupt vector dispatch membar #Sync retl nop SET_SIZE(shipit) #endif /* lint */ #if defined(lint) /* ARGSUSED */ void flush_instr_mem(caddr_t vaddr, size_t len) {} #else /* lint */ /* * flush_instr_mem: * Flush 1 page of the I-$ starting at vaddr * %o0 vaddr * %o1 bytes to be flushed * * SPARC64-VI maintains consistency of the on-chip Instruction Cache with * the stores from all processors so that a FLUSH instruction is only needed * to ensure pipeline is consistent. This means a single flush is sufficient at * the end of a sequence of stores that updates the instruction stream to * ensure correct operation. */ ENTRY(flush_instr_mem) flush %o0 ! address irrelevant retl nop SET_SIZE(flush_instr_mem) #endif /* lint */ /* * flush_ecache: * %o0 - 64 bit physical address * %o1 - ecache size * %o2 - ecache linesize */ #if defined(lint) /*ARGSUSED*/ void flush_ecache(uint64_t physaddr, size_t ecache_size, size_t ecache_linesize) {} #else /* !lint */ ENTRY(flush_ecache) /* * Flush the entire Ecache. */ ECACHE_FLUSHALL(%o1, %o2, %o0, %o4) retl nop SET_SIZE(flush_ecache) #endif /* lint */ #if defined(lint) /*ARGSUSED*/ void kdi_flush_idcache(int dcache_size, int dcache_lsize, int icache_size, int icache_lsize) { } #else /* lint */ /* * I/D cache flushing is not needed for OPL processors */ ENTRY(kdi_flush_idcache) retl nop SET_SIZE(kdi_flush_idcache) #endif /* lint */ #ifdef TRAPTRACE /* * Simplified trap trace macro for OPL. Adapted from us3. */ #define OPL_TRAPTRACE(ptr, scr1, scr2, label) \ CPU_INDEX(scr1, ptr); \ sll scr1, TRAPTR_SIZE_SHIFT, scr1; \ set trap_trace_ctl, ptr; \ add ptr, scr1, scr1; \ ld [scr1 + TRAPTR_LIMIT], ptr; \ tst ptr; \ be,pn %icc, label/**/1; \ ldx [scr1 + TRAPTR_PBASE], ptr; \ ld [scr1 + TRAPTR_OFFSET], scr1; \ add ptr, scr1, ptr; \ rd %asi, scr2; \ wr %g0, TRAPTR_ASI, %asi; \ rd STICK, scr1; \ stxa scr1, [ptr + TRAP_ENT_TICK]%asi; \ rdpr %tl, scr1; \ stha scr1, [ptr + TRAP_ENT_TL]%asi; \ rdpr %tt, scr1; \ stha scr1, [ptr + TRAP_ENT_TT]%asi; \ rdpr %tpc, scr1; \ stna scr1, [ptr + TRAP_ENT_TPC]%asi; \ rdpr %tstate, scr1; \ stxa scr1, [ptr + TRAP_ENT_TSTATE]%asi; \ stna %sp, [ptr + TRAP_ENT_SP]%asi; \ stna %g0, [ptr + TRAP_ENT_TR]%asi; \ stna %g0, [ptr + TRAP_ENT_F1]%asi; \ stna %g0, [ptr + TRAP_ENT_F2]%asi; \ stna %g0, [ptr + TRAP_ENT_F3]%asi; \ stna %g0, [ptr + TRAP_ENT_F4]%asi; \ wr %g0, scr2, %asi; \ CPU_INDEX(ptr, scr1); \ sll ptr, TRAPTR_SIZE_SHIFT, ptr; \ set trap_trace_ctl, scr1; \ add scr1, ptr, ptr; \ ld [ptr + TRAPTR_OFFSET], scr1; \ ld [ptr + TRAPTR_LIMIT], scr2; \ st scr1, [ptr + TRAPTR_LAST_OFFSET]; \ add scr1, TRAP_ENT_SIZE, scr1; \ sub scr2, TRAP_ENT_SIZE, scr2; \ cmp scr1, scr2; \ movge %icc, 0, scr1; \ st scr1, [ptr + TRAPTR_OFFSET]; \ label/**/1: #endif /* TRAPTRACE */ /* * Macros facilitating error handling. */ /* * Save alternative global registers reg1, reg2, reg3 * to scratchpad registers 1, 2, 3 respectively. */ #define OPL_SAVE_GLOBAL(reg1, reg2, reg3) \ stxa reg1, [%g0]ASI_SCRATCHPAD ;\ mov OPL_SCRATCHPAD_SAVE_AG2, reg1 ;\ stxa reg2, [reg1]ASI_SCRATCHPAD ;\ mov OPL_SCRATCHPAD_SAVE_AG3, reg1 ;\ stxa reg3, [reg1]ASI_SCRATCHPAD /* * Restore alternative global registers reg1, reg2, reg3 * from scratchpad registers 1, 2, 3 respectively. */ #define OPL_RESTORE_GLOBAL(reg1, reg2, reg3) \ mov OPL_SCRATCHPAD_SAVE_AG3, reg1 ;\ ldxa [reg1]ASI_SCRATCHPAD, reg3 ;\ mov OPL_SCRATCHPAD_SAVE_AG2, reg1 ;\ ldxa [reg1]ASI_SCRATCHPAD, reg2 ;\ ldxa [%g0]ASI_SCRATCHPAD, reg1 /* * Logs value `val' into the member `offset' of a structure * at physical address `pa' */ #define LOG_REG(pa, offset, val) \ add pa, offset, pa ;\ stxa val, [pa]ASI_MEM #define FLUSH_ALL_TLB(tmp1) \ set DEMAP_ALL_TYPE, tmp1 ;\ stxa %g0, [tmp1]ASI_ITLB_DEMAP ;\ stxa %g0, [tmp1]ASI_DTLB_DEMAP ;\ sethi %hi(FLUSH_ADDR), tmp1 ;\ flush tmp1 /* * Extracts the Physaddr to Logging Buffer field of the OPL_SCRATCHPAD_ERRLOG * scratch register by zeroing all other fields. Result is in pa. */ #define LOG_ADDR(pa) \ mov OPL_SCRATCHPAD_ERRLOG, pa ;\ ldxa [pa]ASI_SCRATCHPAD, pa ;\ sllx pa, 64-ERRLOG_REG_EIDR_SHIFT, pa ;\ srlx pa, 64-ERRLOG_REG_EIDR_SHIFT+ERRLOG_REG_ERR_SHIFT, pa ;\ sllx pa, ERRLOG_REG_ERR_SHIFT, pa /* * Advance the per-cpu error log buffer pointer to the next * ERRLOG_SZ entry, making sure that it will modulo (wraparound) * ERRLOG_BUFSIZ boundary. The args logpa, bufmask, tmp are * unused input registers for this macro. * * Algorithm: * 1. logpa = contents of errorlog scratchpad register * 2. bufmask = ERRLOG_BUFSIZ - 1 * 3. tmp = logpa & ~(bufmask) (tmp is now logbase) * 4. logpa += ERRLOG_SZ * 5. logpa = logpa & bufmask (get new offset to logbase) * 4. logpa = tmp | logpa * 7. write logpa back into errorlog scratchpad register * * new logpa = (logpa & ~bufmask) | ((logpa + ERRLOG_SZ) & bufmask) * */ #define UPDATE_LOGADD(logpa, bufmask, tmp) \ set OPL_SCRATCHPAD_ERRLOG, tmp ;\ ldxa [tmp]ASI_SCRATCHPAD, logpa ;\ set (ERRLOG_BUFSZ-1), bufmask ;\ andn logpa, bufmask, tmp ;\ add logpa, ERRLOG_SZ, logpa ;\ and logpa, bufmask, logpa ;\ or tmp, logpa, logpa ;\ set OPL_SCRATCHPAD_ERRLOG, tmp ;\ stxa logpa, [tmp]ASI_SCRATCHPAD /* Log error status registers into the log buffer */ #define LOG_SYNC_REG(sfsr, sfar, tmp) \ LOG_ADDR(tmp) ;\ LOG_REG(tmp, LOG_SFSR_OFF, sfsr) ;\ LOG_ADDR(tmp) ;\ mov tmp, sfsr ;\ LOG_REG(tmp, LOG_SFAR_OFF, sfar) ;\ rd STICK, sfar ;\ mov sfsr, tmp ;\ LOG_REG(tmp, LOG_STICK_OFF, sfar) ;\ rdpr %tl, tmp ;\ sllx tmp, 32, sfar ;\ rdpr %tt, tmp ;\ or sfar, tmp, sfar ;\ mov sfsr, tmp ;\ LOG_REG(tmp, LOG_TL_OFF, sfar) ;\ set OPL_SCRATCHPAD_ERRLOG, tmp ;\ ldxa [tmp]ASI_SCRATCHPAD, sfar ;\ mov sfsr, tmp ;\ LOG_REG(tmp, LOG_ASI3_OFF, sfar) ;\ rdpr %tpc, sfar ;\ mov sfsr, tmp ;\ LOG_REG(tmp, LOG_TPC_OFF, sfar) ;\ UPDATE_LOGADD(sfsr, sfar, tmp) #define LOG_UGER_REG(uger, tmp, tmp2) \ LOG_ADDR(tmp) ;\ mov tmp, tmp2 ;\ LOG_REG(tmp2, LOG_UGER_OFF, uger) ;\ mov tmp, uger ;\ rd STICK, tmp2 ;\ LOG_REG(tmp, LOG_STICK_OFF, tmp2) ;\ rdpr %tl, tmp ;\ sllx tmp, 32, tmp2 ;\ rdpr %tt, tmp ;\ or tmp2, tmp, tmp2 ;\ mov uger, tmp ;\ LOG_REG(tmp, LOG_TL_OFF, tmp2) ;\ set OPL_SCRATCHPAD_ERRLOG, tmp2 ;\ ldxa [tmp2]ASI_SCRATCHPAD, tmp2 ;\ mov uger, tmp ;\ LOG_REG(tmp, LOG_ASI3_OFF, tmp2) ;\ rdpr %tstate, tmp2 ;\ mov uger, tmp ;\ LOG_REG(tmp, LOG_TSTATE_OFF, tmp2) ;\ rdpr %tpc, tmp2 ;\ mov uger, tmp ;\ LOG_REG(tmp, LOG_TPC_OFF, tmp2) ;\ UPDATE_LOGADD(uger, tmp, tmp2) /* * Scrub the STICK_COMPARE register to clear error by updating * it to a reasonable value for interrupt generation. * Ensure that we observe the CPU_ENABLE flag so that we * don't accidentally enable TICK interrupt in STICK_COMPARE * i.e. no clock interrupt will be generated if CPU_ENABLE flag * is off. */ #define UPDATE_STICK_COMPARE(tmp1, tmp2) \ CPU_ADDR(tmp1, tmp2) ;\ lduh [tmp1 + CPU_FLAGS], tmp2 ;\ andcc tmp2, CPU_ENABLE, %g0 ;\ set OPL_UGER_STICK_DIFF, tmp2 ;\ rd STICK, tmp1 ;\ add tmp1, tmp2, tmp1 ;\ mov 1, tmp2 ;\ sllx tmp2, TICKINT_DIS_SHFT, tmp2 ;\ or tmp1, tmp2, tmp2 ;\ movnz %xcc, tmp1, tmp2 ;\ wr tmp2, %g0, STICK_COMPARE /* * Reset registers that may be corrupted by IAUG_CRE error. * To update interrupt handling related registers force the * clock interrupt. */ #define IAG_CRE(tmp1, tmp2) \ set OPL_SCRATCHPAD_ERRLOG, tmp1 ;\ ldxa [tmp1]ASI_SCRATCHPAD, tmp1 ;\ srlx tmp1, ERRLOG_REG_EIDR_SHIFT, tmp1 ;\ set ERRLOG_REG_EIDR_MASK, tmp2 ;\ and tmp1, tmp2, tmp1 ;\ stxa tmp1, [%g0]ASI_EIDR ;\ wr %g0, 0, SOFTINT ;\ sethi %hi(hres_last_tick), tmp1 ;\ ldx [tmp1 + %lo(hres_last_tick)], tmp1 ;\ set OPL_UGER_STICK_DIFF, tmp2 ;\ add tmp1, tmp2, tmp1 ;\ wr tmp1, %g0, STICK ;\ UPDATE_STICK_COMPARE(tmp1, tmp2) #define CLEAR_FPREGS(tmp) \ wr %g0, FPRS_FEF, %fprs ;\ wr %g0, %g0, %gsr ;\ sethi %hi(opl_clr_freg), tmp ;\ or tmp, %lo(opl_clr_freg), tmp ;\ ldx [tmp], %fsr ;\ fzero %d0 ;\ fzero %d2 ;\ fzero %d4 ;\ fzero %d6 ;\ fzero %d8 ;\ fzero %d10 ;\ fzero %d12 ;\ fzero %d14 ;\ fzero %d16 ;\ fzero %d18 ;\ fzero %d20 ;\ fzero %d22 ;\ fzero %d24 ;\ fzero %d26 ;\ fzero %d28 ;\ fzero %d30 ;\ fzero %d32 ;\ fzero %d34 ;\ fzero %d36 ;\ fzero %d38 ;\ fzero %d40 ;\ fzero %d42 ;\ fzero %d44 ;\ fzero %d46 ;\ fzero %d48 ;\ fzero %d50 ;\ fzero %d52 ;\ fzero %d54 ;\ fzero %d56 ;\ fzero %d58 ;\ fzero %d60 ;\ fzero %d62 ;\ wr %g0, %g0, %fprs #define CLEAR_GLOBALS() \ mov %g0, %g1 ;\ mov %g0, %g2 ;\ mov %g0, %g3 ;\ mov %g0, %g4 ;\ mov %g0, %g5 ;\ mov %g0, %g6 ;\ mov %g0, %g7 /* * We do not clear the alternative globals here because they * are scratch registers, i.e. there is no code that reads from * them without write to them firstly. In other words every * read always follows write that makes extra write to the * alternative globals unnecessary. */ #define CLEAR_GEN_REGS(tmp1, label) \ set TSTATE_KERN, tmp1 ;\ wrpr %g0, tmp1, %tstate ;\ mov %g0, %y ;\ mov %g0, %asi ;\ mov %g0, %ccr ;\ mov %g0, %l0 ;\ mov %g0, %l1 ;\ mov %g0, %l2 ;\ mov %g0, %l3 ;\ mov %g0, %l4 ;\ mov %g0, %l5 ;\ mov %g0, %l6 ;\ mov %g0, %l7 ;\ mov %g0, %i0 ;\ mov %g0, %i1 ;\ mov %g0, %i2 ;\ mov %g0, %i3 ;\ mov %g0, %i4 ;\ mov %g0, %i5 ;\ mov %g0, %i6 ;\ mov %g0, %i7 ;\ mov %g0, %o1 ;\ mov %g0, %o2 ;\ mov %g0, %o3 ;\ mov %g0, %o4 ;\ mov %g0, %o5 ;\ mov %g0, %o6 ;\ mov %g0, %o7 ;\ mov %g0, %o0 ;\ mov %g0, %g4 ;\ mov %g0, %g5 ;\ mov %g0, %g6 ;\ mov %g0, %g7 ;\ rdpr %tl, tmp1 ;\ cmp tmp1, 1 ;\ be,pt %xcc, label/**/1 ;\ rdpr %pstate, tmp1 ;\ wrpr tmp1, PSTATE_AG|PSTATE_IG, %pstate ;\ CLEAR_GLOBALS() ;\ rdpr %pstate, tmp1 ;\ wrpr tmp1, PSTATE_IG|PSTATE_MG, %pstate ;\ CLEAR_GLOBALS() ;\ rdpr %pstate, tmp1 ;\ wrpr tmp1, PSTATE_MG|PSTATE_AG, %pstate ;\ ba,pt %xcc, label/**/2 ;\ nop ;\ label/**/1: ;\ wrpr tmp1, PSTATE_AG, %pstate ;\ CLEAR_GLOBALS() ;\ rdpr %pstate, tmp1 ;\ wrpr tmp1, PSTATE_AG, %pstate ;\ label/**/2: /* * Reset all window related registers */ #define RESET_WINREG(tmp) \ sethi %hi(nwin_minus_one), tmp ;\ ld [tmp + %lo(nwin_minus_one)], tmp ;\ wrpr %g0, tmp, %cwp ;\ wrpr %g0, tmp, %cleanwin ;\ sub tmp, 1, tmp ;\ wrpr %g0, tmp, %cansave ;\ wrpr %g0, %g0, %canrestore ;\ wrpr %g0, %g0, %otherwin ;\ wrpr %g0, PIL_MAX, %pil ;\ wrpr %g0, WSTATE_KERN, %wstate #define RESET_PREV_TSTATE(tmp1, tmp2, label) \ rdpr %tl, tmp1 ;\ subcc tmp1, 1, tmp1 ;\ bz,pt %xcc, label/**/1 ;\ nop ;\ wrpr tmp1, %g0, %tl ;\ set TSTATE_KERN, tmp2 ;\ wrpr tmp2, %g0, %tstate ;\ wrpr %g0, %g0, %tpc ;\ wrpr %g0, %g0, %tnpc ;\ add tmp1, 1, tmp1 ;\ wrpr tmp1, %g0, %tl ;\ label/**/1: /* * %pstate, %pc, %npc are propagated to %tstate, %tpc, %tnpc, * and we reset these regiseter here. */ #define RESET_CUR_TSTATE(tmp) \ set TSTATE_KERN, tmp ;\ wrpr %g0, tmp, %tstate ;\ wrpr %g0, 0, %tpc ;\ wrpr %g0, 0, %tnpc ;\ RESET_WINREG(tmp) /* * In case of urgent errors some MMU registers may be * corrupted, so we set here some reasonable values for * them. Note that resetting MMU registers also reset the context * info, we will need to reset the window registers to prevent * spill/fill that depends on context info for correct behaviour. * Note that the TLBs must be flushed before programming the context * registers. */ #if !defined(lint) #define RESET_MMU_REGS(tmp1, tmp2, tmp3) \ FLUSH_ALL_TLB(tmp1) ;\ set MMU_PCONTEXT, tmp1 ;\ sethi %hi(kcontextreg), tmp2 ;\ ldx [tmp2 + %lo(kcontextreg)], tmp2 ;\ stxa tmp2, [tmp1]ASI_DMMU ;\ set MMU_SCONTEXT, tmp1 ;\ stxa tmp2, [tmp1]ASI_DMMU ;\ sethi %hi(ktsb_base), tmp1 ;\ ldx [tmp1 + %lo(ktsb_base)], tmp2 ;\ mov MMU_TSB, tmp3 ;\ stxa tmp2, [tmp3]ASI_IMMU ;\ stxa tmp2, [tmp3]ASI_DMMU ;\ membar #Sync ;\ RESET_WINREG(tmp1) #define RESET_TSB_TAGPTR(tmp) \ set MMU_TAG_ACCESS, tmp ;\ stxa %g0, [tmp]ASI_IMMU ;\ stxa %g0, [tmp]ASI_DMMU ;\ membar #Sync #endif /* lint */ /* * In case of errors in the MMU_TSB_PREFETCH registers we have to * reset them. We can use "0" as the reset value, this way we set * the "V" bit of the registers to 0, which will disable the prefetch * so the values of the other fields are irrelevant. */ #if !defined(lint) #define RESET_TSB_PREFETCH(tmp) \ set VA_UTSBPREF_8K, tmp ;\ stxa %g0, [tmp]ASI_ITSB_PREFETCH ;\ set VA_UTSBPREF_4M, tmp ;\ stxa %g0, [tmp]ASI_ITSB_PREFETCH ;\ set VA_KTSBPREF_8K, tmp ;\ stxa %g0, [tmp]ASI_ITSB_PREFETCH ;\ set VA_KTSBPREF_4M, tmp ;\ stxa %g0, [tmp]ASI_ITSB_PREFETCH ;\ set VA_UTSBPREF_8K, tmp ;\ stxa %g0, [tmp]ASI_DTSB_PREFETCH ;\ set VA_UTSBPREF_4M, tmp ;\ stxa %g0, [tmp]ASI_DTSB_PREFETCH ;\ set VA_KTSBPREF_8K, tmp ;\ stxa %g0, [tmp]ASI_DTSB_PREFETCH ;\ set VA_KTSBPREF_4M, tmp ;\ stxa %g0, [tmp]ASI_DTSB_PREFETCH #endif /* lint */ /* * In case of errors in the MMU_SHARED_CONTEXT register we have to * reset its value. We can use "0" as the reset value, it will put * 0 in the IV field disabling the shared context support, and * making values of all the other fields of the register irrelevant. */ #if !defined(lint) #define RESET_SHARED_CTXT(tmp) \ set MMU_SHARED_CONTEXT, tmp ;\ stxa %g0, [tmp]ASI_DMMU #endif /* lint */ /* * RESET_TO_PRIV() * * In many cases, we need to force the thread into privilege mode because * privilege mode is only thing in which the system continue to work * due to undeterminable user mode information that come from register * corruption. * * - opl_uger_ctxt * If the error is secondary TSB related register parity, we have no idea * what value is supposed to be for it. * * The below three cases %tstate is not accessible until it is overwritten * with some value, so we have no clue if the thread was running on user mode * or not * - opl_uger_pstate * If the error is %pstate parity, it propagates to %tstate. * - opl_uger_tstate * No need to say the reason * - opl_uger_r * If the error is %ccr or %asi parity, it propagates to %tstate * * For the above four cases, user mode info may not be available for * sys_trap() and user_trap() to work consistently. So we have to force * the thread into privilege mode. * * Forcing the thread to privilege mode requires forcing * regular %g7 to be CPU_THREAD. Because if it was running on user mode, * %g7 will be set in user_trap(). Also since the %sp may be in * an inconsistent state, we need to do a stack reset and switch to * something we know i.e. current thread's kernel stack. * We also reset the window registers and MMU registers just to * make sure. * * To set regular %g7, we need to clear PSTATE_AG bit and need to * use one local register. Note that we are panicking and will never * unwind back so it is ok to clobber a local. * * If the thread was running in user mode, the %tpc value itself might be * within the range of OBP addresses. %tpc must be forced to be zero to prevent * sys_trap() from going to prom_trap() * */ #define RESET_TO_PRIV(tmp, tmp1, tmp2, local) \ RESET_MMU_REGS(tmp, tmp1, tmp2) ;\ CPU_ADDR(tmp, tmp1) ;\ ldx [tmp + CPU_THREAD], local ;\ ldx [local + T_STACK], tmp ;\ sub tmp, STACK_BIAS, %sp ;\ rdpr %pstate, tmp ;\ wrpr tmp, PSTATE_AG, %pstate ;\ mov local, %g7 ;\ rdpr %pstate, local ;\ wrpr local, PSTATE_AG, %pstate ;\ wrpr %g0, 1, %tl ;\ set TSTATE_KERN, tmp ;\ rdpr %cwp, tmp1 ;\ or tmp, tmp1, tmp ;\ wrpr tmp, %g0, %tstate ;\ wrpr %g0, %tpc #if defined(lint) void ce_err(void) {} #else /* lint */ /* * We normally don't expect CE traps since we disable the * 0x63 trap reporting at the start of day. There is a * small window before we disable them, so let check for * it. Otherwise, panic. */ .align 128 ENTRY_NP(ce_err) mov AFSR_ECR, %g1 ldxa [%g1]ASI_ECR, %g1 andcc %g1, ASI_ECR_RTE_UE | ASI_ECR_RTE_CEDG, %g0 bz,pn %xcc, 1f nop retry 1: /* * We did disabled the 0x63 trap reporting. * This shouldn't happen - panic. */ set trap, %g1 rdpr %tt, %g3 sethi %hi(sys_trap), %g5 jmp %g5 + %lo(sys_trap) sub %g0, 1, %g4 SET_SIZE(ce_err) #endif /* lint */ #if defined(lint) void ce_err_tl1(void) {} #else /* lint */ /* * We don't use trap for CE detection. */ ENTRY_NP(ce_err_tl1) set trap, %g1 rdpr %tt, %g3 sethi %hi(sys_trap), %g5 jmp %g5 + %lo(sys_trap) sub %g0, 1, %g4 SET_SIZE(ce_err_tl1) #endif /* lint */ #if defined(lint) void async_err(void) {} #else /* lint */ /* * async_err is the default handler for IAE/DAE traps. * For OPL, we patch in the right handler at start of day. * But if a IAE/DAE trap get generated before the handler * is patched, panic. */ ENTRY_NP(async_err) set trap, %g1 rdpr %tt, %g3 sethi %hi(sys_trap), %g5 jmp %g5 + %lo(sys_trap) sub %g0, 1, %g4 SET_SIZE(async_err) #endif /* lint */ #if defined(lint) void opl_sync_trap(void) {} #else /* lint */ .seg ".data" .global opl_clr_freg .global opl_cpu0_err_log .align 16 opl_clr_freg: .word 0 .align 16 .align MMU_PAGESIZE opl_cpu0_err_log: .skip MMU_PAGESIZE /* * Common synchronous error trap handler (tt=0xA, 0x32) * All TL=0 and TL>0 0xA and 0x32 traps vector to this handler. * The error handling can be best summarized as follows: * 0. Do TRAPTRACE if enabled. * 1. Save globals %g1, %g2 & %g3 onto the scratchpad regs. * 2. The SFSR register is read and verified as valid by checking * SFSR.FV bit being set. If the SFSR.FV is not set, the * error cases cannot be decoded/determined and the SFPAR * register that contain the physical faultaddr is also * not valid. Also the SPFAR is only valid for UE/TO/BERR error * cases. Assuming the SFSR.FV is valid: * - BERR(bus error)/TO(timeout)/UE case * If any of these error cases are detected, read the SFPAR * to get the faultaddress. Generate ereport. * - TLB Parity case (only recoverable case) * For DAE, read SFAR for the faultaddress. For IAE, * use %tpc for faultaddress (SFAR is not valid in IAE) * Flush all the tlbs. * Subtract one from the recoverable error count stored in * the error log scratch register. If the threshold limit * is reached (zero) - generate ereport. Else * restore globals and retry (no ereport is generated). * - TLB Multiple hits * For DAE, read SFAR for the faultaddress. For IAE, * use %tpc for faultaddress (SFAR is not valid in IAE). * Flush all tlbs and generate ereport. * 3. TL=0 and TL>0 considerations * - Since both TL=0 & TL>1 traps are made to vector into * the same handler, the underlying assumption/design here is * that any nested error condition (if happens) occurs only * in the handler and the system is assumed to eventually * Red-mode. With this philosophy in mind, the recoverable * TLB Parity error case never check the TL level before it * retry. Note that this is ok for the TL>1 case (assuming we * don't have a nested error) since we always save the globals * %g1, %g2 & %g3 whenever we enter this trap handler. * - Additional TL=0 vs TL>1 handling includes: * - For UE error occuring under TL>1, special handling * is added to prevent the unlikely chance of a cpu-lockup * when a UE was originally detected in user stack and * the spill trap handler taken from sys_trap() so happened * to reference the same UE location. Under the above * condition (TL>1 and UE error), paranoid code is added * to reset window regs so that spill traps can't happen * during the unwind back to TL=0 handling. * Note that we can do that because we are not returning * back. * 4. Ereport generation. * - Ereport generation is performed when we unwind to the TL=0 * handling code via sys_trap(). on_trap()/lofault protection * will apply there. * */ ENTRY_NP(opl_sync_trap) #ifdef TRAPTRACE OPL_TRAPTRACE(%g1, %g2, %g3, opl_sync_trap_lb) rdpr %tt, %g1 #endif /* TRAPTRACE */ cmp %g1, T_INSTR_ERROR bne,pt %xcc, 0f mov MMU_SFSR, %g3 ldxa [%g3]ASI_IMMU, %g1 ! IAE trap case tt = 0xa andcc %g1, SFSR_FV, %g0 bz,a,pn %xcc, 2f ! Branch if SFSR is invalid and rdpr %tpc, %g2 ! use %tpc for faultaddr instead sethi %hi(SFSR_UE|SFSR_BERR|SFSR_TO), %g3 andcc %g1, %g3, %g0 ! Check for UE/BERR/TO errors bz,a,pt %xcc, 1f ! Branch if not UE/BERR/TO and rdpr %tpc, %g2 ! use %tpc as faultaddr set OPL_MMU_SFPAR, %g3 ! In the UE/BERR/TO cases, use ba,pt %xcc, 2f ! SFPAR as faultaddr ldxa [%g3]ASI_IMMU, %g2 0: ldxa [%g3]ASI_DMMU, %g1 ! DAE trap case tt = 0x32 andcc %g1, SFSR_FV, %g0 bnz,pt %xcc, 7f ! branch if SFSR.FV is valid mov MMU_SFAR, %g2 ! set %g2 to use SFAR ba,pt %xcc, 2f ! SFSR.FV is not valid, read SFAR ldxa [%g2]ASI_DMMU, %g2 ! for faultaddr 7: sethi %hi(SFSR_UE|SFSR_BERR|SFSR_TO), %g3 andcc %g1, %g3, %g0 ! Check UE/BERR/TO for valid SFPAR movnz %xcc, OPL_MMU_SFPAR, %g2 ! Use SFPAR instead of SFAR for ldxa [%g2]ASI_DMMU, %g2 ! faultaddr 1: sethi %hi(SFSR_TLB_PRT), %g3 andcc %g1, %g3, %g0 bz,pt %xcc, 8f ! branch for TLB multi-hit check nop /* * This is the TLB parity error case and it is the * only retryable error case. * Only %g1, %g2 and %g3 are allowed */ FLUSH_ALL_TLB(%g3) set OPL_SCRATCHPAD_ERRLOG, %g3 ldxa [%g3]ASI_SCRATCHPAD, %g3 ! Read errlog scratchreg and %g3, ERRLOG_REG_NUMERR_MASK, %g3! Extract the error count subcc %g3, 1, %g0 ! Subtract one from the count bz,pn %xcc, 2f ! too many TLB parity errs in a certain nop ! period, branch to generate ereport LOG_SYNC_REG(%g1, %g2, %g3) ! Record into the error log set OPL_SCRATCHPAD_ERRLOG, %g3 ldxa [%g3]ASI_SCRATCHPAD, %g2 sub %g2, 1, %g2 ! decrement error counter by 1 stxa %g2, [%g3]ASI_SCRATCHPAD ! update the errlog scratchreg OPL_RESTORE_GLOBAL(%g1, %g2, %g3) retry 8: sethi %hi(SFSR_TLB_MUL), %g3 andcc %g1, %g3, %g0 bz,pt %xcc, 2f ! check for the TLB multi-hit errors nop FLUSH_ALL_TLB(%g3) 2: /* * non-retryable error handling * now we can use other registers since * we will not be returning back */ mov %g1, %g5 ! %g5 = SFSR mov %g2, %g6 ! %g6 = SFPAR or SFAR/tpc LOG_SYNC_REG(%g1, %g2, %g3) ! Record into the error log /* * Special case for UE on user stack. * There is a possibility that the same error may come back here * by touching the same UE in spill trap handler taken from * sys_trap(). It ends up with an infinite loop causing a cpu lockup. * Conditions for this handling this case are: * - SFSR_FV is valid and SFSR_UE is set * - we are at TL > 1 * If the above conditions are true, we force %cansave to be a * big number to prevent spill trap in sys_trap(). Note that * we will not be returning back. */ rdpr %tt, %g4 ! %g4 == ttype rdpr %tl, %g1 ! %g1 == tl cmp %g1, 1 ! Check if TL == 1 be,pt %xcc, 3f ! branch if we came from TL=0 nop andcc %g5, SFSR_FV, %g0 ! see if SFSR.FV is valid bz,pn %xcc, 4f ! branch, checking UE is meaningless sethi %hi(SFSR_UE), %g2 andcc %g5, %g2, %g0 ! check for UE bz,pt %xcc, 4f ! branch if not UE nop RESET_WINREG(%g1) ! reset windows to prevent spills 4: RESET_USER_RTT_REGS(%g2, %g3, 5f) 5: mov %g5, %g3 ! pass SFSR to the 3rd arg mov %g6, %g2 ! pass SFAR to the 2nd arg set opl_cpu_isync_tl1_error, %g1 set opl_cpu_dsync_tl1_error, %g6 cmp %g4, T_INSTR_ERROR movne %icc, %g6, %g1 ba,pt %icc, 6f nop 3: mov %g5, %g3 ! pass SFSR to the 3rd arg mov %g6, %g2 ! pass SFAR to the 2nd arg set opl_cpu_isync_tl0_error, %g1 set opl_cpu_dsync_tl0_error, %g6 cmp %g4, T_INSTR_ERROR movne %icc, %g6, %g1 6: sethi %hi(sys_trap), %g5 jmp %g5 + %lo(sys_trap) mov PIL_15, %g4 SET_SIZE(opl_sync_trap) #endif /* lint */ #if defined(lint) void opl_uger_trap(void) {} #else /* lint */ /* * Common Urgent error trap handler (tt=0x40) * All TL=0 and TL>0 0x40 traps vector to this handler. * The error handling can be best summarized as follows: * 1. Read the Urgent error status register (UGERSR) * Faultaddress is N/A here and it is not collected. * 2. Check to see if we have a multiple errors case * If so, we enable WEAK_ED (weak error detection) bit * to prevent any potential error storms and branch directly * to generate ereport. (we don't decode/handle individual * error cases when we get a multiple error situation) * 3. Now look for the recoverable error cases which include * IUG_DTLB, IUG_ITLB or COREERR errors. If any of the * recoverable errors are detected, do the following: * - Flush all tlbs. * - Verify that we came from TL=0, if not, generate * ereport. Note that the reason we don't recover * at TL>0 is because the AGs might be corrupted or * inconsistent. We can't save/restore them into * the scratchpad regs like we did for opl_sync_trap(). * - Check the INSTEND[5:4] bits in the UGERSR. If the * value is 0x3 (11b), this error is not recoverable. * Generate ereport. * - Subtract one from the recoverable error count stored in * the error log scratch register. If the threshold limit * is reached (zero) - generate ereport. * - If the count is within the limit, update the count * in the error log register (subtract one). Log the error * info in the log buffer. Capture traptrace if enabled. * Retry (no ereport generated) * 4. The rest of the error cases are unrecoverable and will * be handled according (flushing regs, etc as required). * For details on these error cases (UGER_CRE, UGER_CTXT, etc..) * consult the OPL cpu/mem philosophy doc. * Ereport will be generated for these errors. * 5. Ereport generation. * - Ereport generation for urgent error trap always * result in a panic when we unwind to the TL=0 handling * code via sys_trap(). on_trap()/lofault protection do * not apply there. */ ENTRY_NP(opl_uger_trap) set ASI_UGERSR, %g2 ldxa [%g2]ASI_AFSR, %g1 ! Read the UGERSR reg set UGESR_MULTI, %g2 andcc %g1, %g2, %g0 ! Check for Multi-errs bz,pt %xcc, opl_uger_is_recover ! branch if not Multi-errs nop set AFSR_ECR, %g2 ldxa [%g2]ASI_AFSR, %g3 ! Enable Weak error or %g3, ASI_ECR_WEAK_ED, %g3 ! detect mode to prevent stxa %g3, [%g2]ASI_AFSR ! potential error storms ba %xcc, opl_uger_panic1 nop opl_uger_is_recover: set UGESR_CAN_RECOVER, %g2 ! Check for recoverable andcc %g1, %g2, %g0 ! errors i.e.IUG_DTLB, bz,pt %xcc, opl_uger_cre ! IUG_ITLB or COREERR nop /* * Fall thru to handle recoverable case * Need to do the following additional checks to determine * if this is indeed recoverable. * 1. Error trap came from TL=0 and * 2. INSTEND[5:4] bits in UGERSR is not 0x3 * 3. Recoverable error count limit not reached * */ FLUSH_ALL_TLB(%g3) rdpr %tl, %g3 ! Read TL cmp %g3, 1 ! Check if we came from TL=0 bne,pt %xcc, opl_uger_panic ! branch if came from TL>0 nop srlx %g1, 4, %g2 ! shift INSTEND[5:4] -> [1:0] and %g2, 3, %g2 ! extract the shifted [1:0] bits cmp %g2, 3 ! check if INSTEND is recoverable be,pt %xcc, opl_uger_panic ! panic if ([1:0] = 11b) nop set OPL_SCRATCHPAD_ERRLOG, %g3 ldxa [%g3]ASI_SCRATCHPAD, %g2 ! Read errlog scratch reg and %g2, ERRLOG_REG_NUMERR_MASK, %g3! Extract error count and subcc %g3, 1, %g3 ! subtract one from it bz,pt %xcc, opl_uger_panic ! If count reached zero, too many nop ! errors, branch to generate ereport sub %g2, 1, %g2 ! Subtract one from the count set OPL_SCRATCHPAD_ERRLOG, %g3 ! and write back the updated stxa %g2, [%g3]ASI_SCRATCHPAD ! count into the errlog reg LOG_UGER_REG(%g1, %g2, %g3) ! Log the error info #ifdef TRAPTRACE OPL_TRAPTRACE(%g1, %g2, %g3, opl_uger_trap_lb) #endif /* TRAPTRACE */ retry ! retry - no ereport /* * Process the rest of the unrecoverable error cases * All error cases below ultimately branch to either * opl_uger_panic or opl_uger_panic1. * opl_uger_panic1 is the same as opl_uger_panic except * for the additional execution of the RESET_TO_PRIV() * macro that does a heavy handed reset. Read the * comments for RESET_TO_PRIV() macro for more info. */ opl_uger_cre: set UGESR_IAUG_CRE, %g2 andcc %g1, %g2, %g0 bz,pt %xcc, opl_uger_ctxt nop IAG_CRE(%g2, %g3) set AFSR_ECR, %g2 ldxa [%g2]ASI_AFSR, %g3 or %g3, ASI_ECR_WEAK_ED, %g3 stxa %g3, [%g2]ASI_AFSR ba %xcc, opl_uger_panic nop opl_uger_ctxt: set UGESR_IAUG_TSBCTXT, %g2 andcc %g1, %g2, %g0 bz,pt %xcc, opl_uger_tsbp nop GET_CPU_IMPL(%g2) cmp %g2, JUPITER_IMPL bne %xcc, 1f nop RESET_SHARED_CTXT(%g2) 1: RESET_MMU_REGS(%g2, %g3, %g4) ba %xcc, opl_uger_panic nop opl_uger_tsbp: set UGESR_IUG_TSBP, %g2 andcc %g1, %g2, %g0 bz,pt %xcc, opl_uger_pstate nop GET_CPU_IMPL(%g2) cmp %g2, JUPITER_IMPL bne %xcc, 1f nop RESET_TSB_PREFETCH(%g2) 1: RESET_TSB_TAGPTR(%g2) /* * IUG_TSBP error may corrupt MMU registers * Reset them here. */ RESET_MMU_REGS(%g2, %g3, %g4) ba %xcc, opl_uger_panic nop opl_uger_pstate: set UGESR_IUG_PSTATE, %g2 andcc %g1, %g2, %g0 bz,pt %xcc, opl_uger_tstate nop RESET_CUR_TSTATE(%g2) ba %xcc, opl_uger_panic1 nop opl_uger_tstate: set UGESR_IUG_TSTATE, %g2 andcc %g1, %g2, %g0 bz,pt %xcc, opl_uger_f nop RESET_PREV_TSTATE(%g2, %g3, opl_uger_tstate_1) ba %xcc, opl_uger_panic1 nop opl_uger_f: set UGESR_IUG_F, %g2 andcc %g1, %g2, %g0 bz,pt %xcc, opl_uger_r nop CLEAR_FPREGS(%g2) ba %xcc, opl_uger_panic nop opl_uger_r: set UGESR_IUG_R, %g2 andcc %g1, %g2, %g0 bz,pt %xcc, opl_uger_panic1 nop CLEAR_GEN_REGS(%g2, opl_uger_r_1) ba %xcc, opl_uger_panic1 nop opl_uger_panic: mov %g1, %g2 ! %g2 = arg #1 LOG_UGER_REG(%g1, %g3, %g4) ba %xcc, opl_uger_panic_cmn nop opl_uger_panic1: mov %g1, %g2 ! %g2 = arg #1 LOG_UGER_REG(%g1, %g3, %g4) RESET_TO_PRIV(%g1, %g3, %g4, %l0) /* * Set up the argument for sys_trap. * %g2 = arg #1 already set above */ opl_uger_panic_cmn: RESET_USER_RTT_REGS(%g4, %g5, 1f) 1: rdpr %tl, %g3 ! arg #2 set opl_cpu_urgent_error, %g1 ! pc sethi %hi(sys_trap), %g5 jmp %g5 + %lo(sys_trap) mov PIL_15, %g4 SET_SIZE(opl_uger_trap) #endif /* lint */ #if defined(lint) void opl_ta3_trap(void) {} void opl_cleanw_subr(void) {} #else /* lint */ /* * OPL ta3 support (note please, that win_reg * area size for each cpu is 2^7 bytes) */ #define RESTORE_WREGS(tmp1, tmp2) \ CPU_INDEX(tmp1, tmp2) ;\ sethi %hi(opl_ta3_save), tmp2 ;\ ldx [tmp2 +%lo(opl_ta3_save)], tmp2 ;\ sllx tmp1, 7, tmp1 ;\ add tmp2, tmp1, tmp2 ;\ ldx [tmp2 + 0], %l0 ;\ ldx [tmp2 + 8], %l1 ;\ ldx [tmp2 + 16], %l2 ;\ ldx [tmp2 + 24], %l3 ;\ ldx [tmp2 + 32], %l4 ;\ ldx [tmp2 + 40], %l5 ;\ ldx [tmp2 + 48], %l6 ;\ ldx [tmp2 + 56], %l7 ;\ ldx [tmp2 + 64], %i0 ;\ ldx [tmp2 + 72], %i1 ;\ ldx [tmp2 + 80], %i2 ;\ ldx [tmp2 + 88], %i3 ;\ ldx [tmp2 + 96], %i4 ;\ ldx [tmp2 + 104], %i5 ;\ ldx [tmp2 + 112], %i6 ;\ ldx [tmp2 + 120], %i7 #define SAVE_WREGS(tmp1, tmp2) \ CPU_INDEX(tmp1, tmp2) ;\ sethi %hi(opl_ta3_save), tmp2 ;\ ldx [tmp2 +%lo(opl_ta3_save)], tmp2 ;\ sllx tmp1, 7, tmp1 ;\ add tmp2, tmp1, tmp2 ;\ stx %l0, [tmp2 + 0] ;\ stx %l1, [tmp2 + 8] ;\ stx %l2, [tmp2 + 16] ;\ stx %l3, [tmp2 + 24] ;\ stx %l4, [tmp2 + 32] ;\ stx %l5, [tmp2 + 40] ;\ stx %l6, [tmp2 + 48] ;\ stx %l7, [tmp2 + 56] ;\ stx %i0, [tmp2 + 64] ;\ stx %i1, [tmp2 + 72] ;\ stx %i2, [tmp2 + 80] ;\ stx %i3, [tmp2 + 88] ;\ stx %i4, [tmp2 + 96] ;\ stx %i5, [tmp2 + 104] ;\ stx %i6, [tmp2 + 112] ;\ stx %i7, [tmp2 + 120] /* * The purpose of this function is to make sure that the restore * instruction after the flushw does not cause a fill trap. The sun4u * fill trap handler can not handle a tlb fault of an unmapped stack * except at the restore instruction at user_rtt. On OPL systems the * stack can get unmapped between the flushw and restore instructions * since multiple strands share the tlb. */ ENTRY_NP(opl_ta3_trap) set trap, %g1 mov T_FLUSHW, %g3 sub %g0, 1, %g4 rdpr %cwp, %g5 SAVE_WREGS(%g2, %g6) save flushw rdpr %cwp, %g6 wrpr %g5, %cwp RESTORE_WREGS(%g2, %g5) wrpr %g6, %cwp restored restore ba,a fast_trap_done SET_SIZE(opl_ta3_trap) ENTRY_NP(opl_cleanw_subr) set trap, %g1 mov T_FLUSHW, %g3 sub %g0, 1, %g4 rdpr %cwp, %g5 SAVE_WREGS(%g2, %g6) save flushw rdpr %cwp, %g6 wrpr %g5, %cwp RESTORE_WREGS(%g2, %g5) wrpr %g6, %cwp restored restore jmp %g7 nop SET_SIZE(opl_cleanw_subr) #endif /* lint */ #if defined(lint) void opl_serr_instr(void) {} #else /* lint */ /* * The actual trap handler for tt=0x0a, and tt=0x32 */ ENTRY_NP(opl_serr_instr) OPL_SAVE_GLOBAL(%g1,%g2,%g3) sethi %hi(opl_sync_trap), %g3 jmp %g3 + %lo(opl_sync_trap) rdpr %tt, %g1 .align 32 SET_SIZE(opl_serr_instr) #endif /* lint */ #if defined(lint) void opl_ugerr_instr(void) {} #else /* lint */ /* * The actual trap handler for tt=0x40 */ ENTRY_NP(opl_ugerr_instr) sethi %hi(opl_uger_trap), %g3 jmp %g3 + %lo(opl_uger_trap) nop .align 32 SET_SIZE(opl_ugerr_instr) #endif /* lint */ #if defined(lint) void opl_ta3_instr(void) {} #else /* lint */ /* * The actual trap handler for tt=0x103 (flushw) */ ENTRY_NP(opl_ta3_instr) sethi %hi(opl_ta3_trap), %g3 jmp %g3 + %lo(opl_ta3_trap) nop .align 32 SET_SIZE(opl_ta3_instr) #endif /* lint */ #if defined(lint) void opl_ta4_instr(void) {} #else /* lint */ /* * The patch for the .clean_windows code */ ENTRY_NP(opl_ta4_instr) sethi %hi(opl_cleanw_subr), %g3 add %g3, %lo(opl_cleanw_subr), %g3 jmpl %g3, %g7 add %g7, 8, %g7 nop nop nop SET_SIZE(opl_ta4_instr) #endif /* lint */ #if defined(lint) /* * Get timestamp (stick). */ /* ARGSUSED */ void stick_timestamp(int64_t *ts) { } #else /* lint */ ENTRY_NP(stick_timestamp) rd STICK, %g1 ! read stick reg sllx %g1, 1, %g1 srlx %g1, 1, %g1 ! clear npt bit retl stx %g1, [%o0] ! store the timestamp SET_SIZE(stick_timestamp) #endif /* lint */ #if defined(lint) /* * Set STICK adjusted by skew. */ /* ARGSUSED */ void stick_adj(int64_t skew) { } #else /* lint */ ENTRY_NP(stick_adj) rdpr %pstate, %g1 ! save processor state andn %g1, PSTATE_IE, %g3 ba 1f ! cache align stick adj wrpr %g0, %g3, %pstate ! turn off interrupts .align 16 1: nop rd STICK, %g4 ! read stick reg add %g4, %o0, %o1 ! adjust stick with skew wr %o1, %g0, STICK ! write stick reg retl wrpr %g1, %pstate ! restore processor state SET_SIZE(stick_adj) #endif /* lint */ #if defined(lint) /* * Debugger-specific stick retrieval */ /*ARGSUSED*/ int kdi_get_stick(uint64_t *stickp) { return (0); } #else /* lint */ ENTRY_NP(kdi_get_stick) rd STICK, %g1 stx %g1, [%o0] retl mov %g0, %o0 SET_SIZE(kdi_get_stick) #endif /* lint */ #if defined(lint) /*ARGSUSED*/ int dtrace_blksuword32(uintptr_t addr, uint32_t *data, int tryagain) { return (0); } #else ENTRY(dtrace_blksuword32) save %sp, -SA(MINFRAME + 4), %sp rdpr %pstate, %l1 andn %l1, PSTATE_IE, %l2 ! disable interrupts to wrpr %g0, %l2, %pstate ! protect our FPU diddling rd %fprs, %l0 andcc %l0, FPRS_FEF, %g0 bz,a,pt %xcc, 1f ! if the fpu is disabled wr %g0, FPRS_FEF, %fprs ! ... enable the fpu st %f0, [%fp + STACK_BIAS - 4] ! save %f0 to the stack 1: set 0f, %l5 /* * We're about to write a block full or either total garbage * (not kernel data, don't worry) or user floating-point data * (so it only _looks_ like garbage). */ ld [%i1], %f0 ! modify the block membar #Sync stn %l5, [THREAD_REG + T_LOFAULT] ! set up the lofault handler stda %d0, [%i0]ASI_BLK_COMMIT_S ! store the modified block membar #Sync flush %i0 ! flush instruction pipeline stn %g0, [THREAD_REG + T_LOFAULT] ! remove the lofault handler bz,a,pt %xcc, 1f wr %g0, %l0, %fprs ! restore %fprs ld [%fp + STACK_BIAS - 4], %f0 ! restore %f0 1: wrpr %g0, %l1, %pstate ! restore interrupts ret restore %g0, %g0, %o0 0: membar #Sync stn %g0, [THREAD_REG + T_LOFAULT] ! remove the lofault handler bz,a,pt %xcc, 1f wr %g0, %l0, %fprs ! restore %fprs ld [%fp + STACK_BIAS - 4], %f0 ! restore %f0 1: wrpr %g0, %l1, %pstate ! restore interrupts /* * If tryagain is set (%i2) we tail-call dtrace_blksuword32_err() * which deals with watchpoints. Otherwise, just return -1. */ brnz,pt %i2, 1f nop ret restore %g0, -1, %o0 1: call dtrace_blksuword32_err restore SET_SIZE(dtrace_blksuword32) #endif /* lint */ #if defined(lint) /*ARGSUSED*/ void ras_cntr_reset(void *arg) { } #else ENTRY_NP(ras_cntr_reset) set OPL_SCRATCHPAD_ERRLOG, %o1 ldxa [%o1]ASI_SCRATCHPAD, %o0 or %o0, ERRLOG_REG_NUMERR_MASK, %o0 retl stxa %o0, [%o1]ASI_SCRATCHPAD SET_SIZE(ras_cntr_reset) #endif /* lint */ #if defined(lint) /* ARGSUSED */ void opl_error_setup(uint64_t cpu_err_log_pa) { } #else /* lint */ ENTRY_NP(opl_error_setup) /* * Initialize the error log scratchpad register */ ldxa [%g0]ASI_EIDR, %o2 sethi %hi(ERRLOG_REG_EIDR_MASK), %o1 or %o1, %lo(ERRLOG_REG_EIDR_MASK), %o1 and %o2, %o1, %o3 sllx %o3, ERRLOG_REG_EIDR_SHIFT, %o2 or %o2, %o0, %o3 or %o3, ERRLOG_REG_NUMERR_MASK, %o0 set OPL_SCRATCHPAD_ERRLOG, %o1 stxa %o0, [%o1]ASI_SCRATCHPAD /* * Disable all restrainable error traps */ mov AFSR_ECR, %o1 ldxa [%o1]ASI_AFSR, %o0 andn %o0, ASI_ECR_RTE_UE|ASI_ECR_RTE_CEDG, %o0 retl stxa %o0, [%o1]ASI_AFSR SET_SIZE(opl_error_setup) #endif /* lint */ #if defined(lint) /* ARGSUSED */ void opl_mpg_enable(void) { } #else /* lint */ ENTRY_NP(opl_mpg_enable) /* * Enable MMU translating multiple page sizes for * sITLB and sDTLB. */ mov LSU_MCNTL, %o0 ldxa [%o0] ASI_MCNTL, %o1 or %o1, MCNTL_MPG_SITLB | MCNTL_MPG_SDTLB, %o1 retl stxa %o1, [%o0] ASI_MCNTL SET_SIZE(opl_mpg_enable) #endif /* lint */ #if defined(lint) /* * This function is called for each (enabled) CPU. We use it to * initialize error handling related registers. */ /*ARGSUSED*/ void cpu_feature_init(void) {} #else /* lint */ ENTRY(cpu_feature_init) ! ! get the device_id and store the device_id ! in the appropriate cpunodes structure ! given the cpus index ! CPU_INDEX(%o0, %o1) mulx %o0, CPU_NODE_SIZE, %o0 set cpunodes + DEVICE_ID, %o1 ldxa [%g0] ASI_DEVICE_SERIAL_ID, %o2 stx %o2, [%o0 + %o1] ! ! initialize CPU registers ! ba opl_cpu_reg_init nop SET_SIZE(cpu_feature_init) #endif /* lint */ #if defined(lint) void cpu_cleartickpnt(void) {} #else /* lint */ /* * Clear the NPT (non-privileged trap) bit in the %tick/%stick * registers. In an effort to make the change in the * tick/stick counter as consistent as possible, we disable * all interrupts while we're changing the registers. We also * ensure that the read and write instructions are in the same * line in the instruction cache. */ ENTRY_NP(cpu_clearticknpt) rdpr %pstate, %g1 /* save processor state */ andn %g1, PSTATE_IE, %g3 /* turn off */ wrpr %g0, %g3, %pstate /* interrupts */ rdpr %tick, %g2 /* get tick register */ brgez,pn %g2, 1f /* if NPT bit off, we're done */ mov 1, %g3 /* create mask */ sllx %g3, 63, %g3 /* for NPT bit */ ba,a,pt %xcc, 2f .align 8 /* Ensure rd/wr in same i$ line */ 2: rdpr %tick, %g2 /* get tick register */ wrpr %g3, %g2, %tick /* write tick register, */ /* clearing NPT bit */ 1: rd STICK, %g2 /* get stick register */ brgez,pn %g2, 3f /* if NPT bit off, we're done */ mov 1, %g3 /* create mask */ sllx %g3, 63, %g3 /* for NPT bit */ ba,a,pt %xcc, 4f .align 8 /* Ensure rd/wr in same i$ line */ 4: rd STICK, %g2 /* get stick register */ wr %g3, %g2, STICK /* write stick register, */ /* clearing NPT bit */ 3: jmp %g4 + 4 wrpr %g0, %g1, %pstate /* restore processor state */ SET_SIZE(cpu_clearticknpt) #endif /* lint */ #if defined(lint) void cpu_halt_cpu(void) {} void cpu_smt_pause(void) {} #else /* lint */ /* * Halt the current strand with the suspend instruction. * The compiler/asm currently does not support this suspend * instruction mnemonic, use byte code for now. */ ENTRY_NP(cpu_halt_cpu) .word 0x81b01040 retl nop SET_SIZE(cpu_halt_cpu) /* * Pause the current strand with the sleep instruction. * The compiler/asm currently does not support this sleep * instruction mnemonic, use byte code for now. */ ENTRY_NP(cpu_smt_pause) .word 0x81b01060 retl nop SET_SIZE(cpu_smt_pause) #endif /* lint */