diff options
author | Linus Torvalds <torvalds@ppc970.osdl.org> | 2005-04-16 15:20:36 -0700 |
---|---|---|
committer | Linus Torvalds <torvalds@ppc970.osdl.org> | 2005-04-16 15:20:36 -0700 |
commit | 1da177e4c3f41524e886b7f1b8a0c1fc7321cac2 (patch) | |
tree | 0bba044c4ce775e45a88a51686b5d9f90697ea9d /arch/parisc/kernel/entry.S |
Linux-2.6.12-rc2
Initial git repository build. I'm not bothering with the full history,
even though we have it. We can create a separate "historical" git
archive of that later if we want to, and in the meantime it's about
3.2GB when imported into git - space that would just make the early
git days unnecessarily complicated, when we don't have a lot of good
infrastructure for it.
Let it rip!
Diffstat (limited to 'arch/parisc/kernel/entry.S')
-rw-r--r-- | arch/parisc/kernel/entry.S | 2426 |
1 files changed, 2426 insertions, 0 deletions
diff --git a/arch/parisc/kernel/entry.S b/arch/parisc/kernel/entry.S new file mode 100644 index 000000000000..ee58d37dbb27 --- /dev/null +++ b/arch/parisc/kernel/entry.S @@ -0,0 +1,2426 @@ +/* + * Linux/PA-RISC Project (http://www.parisc-linux.org/) + * + * kernel entry points (interruptions, system call wrappers) + * Copyright (C) 1999,2000 Philipp Rumpf + * Copyright (C) 1999 SuSE GmbH Nuernberg + * Copyright (C) 2000 Hewlett-Packard (John Marvin) + * Copyright (C) 1999 Hewlett-Packard (Frank Rowand) + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2, or (at your option) + * any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. + */ + +#include <linux/config.h> +#include <asm/offsets.h> + +/* we have the following possibilities to act on an interruption: + * - handle in assembly and use shadowed registers only + * - save registers to kernel stack and handle in assembly or C */ + + +#include <asm/assembly.h> /* for LDREG/STREG defines */ +#include <asm/pgtable.h> +#include <asm/psw.h> +#include <asm/signal.h> +#include <asm/unistd.h> +#include <asm/thread_info.h> + +#ifdef __LP64__ +#define CMPIB cmpib,* +#define CMPB cmpb,* +#define COND(x) *x + + .level 2.0w +#else +#define CMPIB cmpib, +#define CMPB cmpb, +#define COND(x) x + + .level 2.0 +#endif + + .import pa_dbit_lock,data + + /* space_to_prot macro creates a prot id from a space id */ + +#if (SPACEID_SHIFT) == 0 + .macro space_to_prot spc prot + depd,z \spc,62,31,\prot + .endm +#else + .macro space_to_prot spc prot + extrd,u \spc,(64 - (SPACEID_SHIFT)),32,\prot + .endm +#endif + + /* Switch to virtual mapping, trashing only %r1 */ + .macro virt_map + rsm PSW_SM_Q,%r0 + tovirt_r1 %r29 + mfsp %sr7, %r1 + or,= %r0,%r1,%r0 /* Only save sr7 in sr3 if sr7 != 0 */ + mtsp %r1, %sr3 + mtsp %r0, %sr4 + mtsp %r0, %sr5 + mtsp %r0, %sr6 + mtsp %r0, %sr7 + load32 KERNEL_PSW, %r1 + mtctl %r1, %cr22 + mtctl %r0, %cr17 /* Clear IIASQ tail */ + mtctl %r0, %cr17 /* Clear IIASQ head */ + load32 4f, %r1 + mtctl %r1, %cr18 /* Set IIAOQ tail */ + ldo 4(%r1), %r1 + mtctl %r1, %cr18 /* Set IIAOQ head */ + rfir + nop +4: + .endm + + /* + * The "get_stack" macros are responsible for determining the + * kernel stack value. + * + * For Faults: + * If sr7 == 0 + * Already using a kernel stack, so call the + * get_stack_use_r30 macro to push a pt_regs structure + * on the stack, and store registers there. + * else + * Need to set up a kernel stack, so call the + * get_stack_use_cr30 macro to set up a pointer + * to the pt_regs structure contained within the + * task pointer pointed to by cr30. Set the stack + * pointer to point to the end of the task structure. + * + * For Interrupts: + * If sr7 == 0 + * Already using a kernel stack, check to see if r30 + * is already pointing to the per processor interrupt + * stack. If it is, call the get_stack_use_r30 macro + * to push a pt_regs structure on the stack, and store + * registers there. Otherwise, call get_stack_use_cr31 + * to get a pointer to the base of the interrupt stack + * and push a pt_regs structure on that stack. + * else + * Need to set up a kernel stack, so call the + * get_stack_use_cr30 macro to set up a pointer + * to the pt_regs structure contained within the + * task pointer pointed to by cr30. Set the stack + * pointer to point to the end of the task structure. + * N.B: We don't use the interrupt stack for the + * first interrupt from userland, because signals/ + * resched's are processed when returning to userland, + * and we can sleep in those cases. + * + * Note that we use shadowed registers for temps until + * we can save %r26 and %r29. %r26 is used to preserve + * %r8 (a shadowed register) which temporarily contained + * either the fault type ("code") or the eirr. We need + * to use a non-shadowed register to carry the value over + * the rfir in virt_map. We use %r26 since this value winds + * up being passed as the argument to either do_cpu_irq_mask + * or handle_interruption. %r29 is used to hold a pointer + * the register save area, and once again, it needs to + * be a non-shadowed register so that it survives the rfir. + * + * N.B. TASK_SZ_ALGN and PT_SZ_ALGN include space for a stack frame. + */ + + .macro get_stack_use_cr30 + + /* we save the registers in the task struct */ + + mfctl %cr30, %r1 + tophys %r1,%r9 + LDREG TI_TASK(%r9), %r1 /* thread_info -> task_struct */ + tophys %r1,%r9 + ldo TASK_REGS(%r9),%r9 + STREG %r30, PT_GR30(%r9) + STREG %r29,PT_GR29(%r9) + STREG %r26,PT_GR26(%r9) + copy %r9,%r29 + mfctl %cr30, %r1 + ldo THREAD_SZ_ALGN(%r1), %r30 + .endm + + .macro get_stack_use_r30 + + /* we put a struct pt_regs on the stack and save the registers there */ + + tophys %r30,%r9 + STREG %r30,PT_GR30(%r9) + ldo PT_SZ_ALGN(%r30),%r30 + STREG %r29,PT_GR29(%r9) + STREG %r26,PT_GR26(%r9) + copy %r9,%r29 + .endm + + .macro rest_stack + LDREG PT_GR1(%r29), %r1 + LDREG PT_GR30(%r29),%r30 + LDREG PT_GR29(%r29),%r29 + .endm + + /* default interruption handler + * (calls traps.c:handle_interruption) */ + .macro def code + b intr_save + ldi \code, %r8 + .align 32 + .endm + + /* Interrupt interruption handler + * (calls irq.c:do_cpu_irq_mask) */ + .macro extint code + b intr_extint + mfsp %sr7,%r16 + .align 32 + .endm + + .import os_hpmc, code + + /* HPMC handler */ + .macro hpmc code + nop /* must be a NOP, will be patched later */ + load32 PA(os_hpmc), %r3 + bv,n 0(%r3) + nop + .word 0 /* checksum (will be patched) */ + .word PA(os_hpmc) /* address of handler */ + .word 0 /* length of handler */ + .endm + + /* + * Performance Note: Instructions will be moved up into + * this part of the code later on, once we are sure + * that the tlb miss handlers are close to final form. + */ + + /* Register definitions for tlb miss handler macros */ + + va = r8 /* virtual address for which the trap occured */ + spc = r24 /* space for which the trap occured */ + +#ifndef __LP64__ + + /* + * itlb miss interruption handler (parisc 1.1 - 32 bit) + */ + + .macro itlb_11 code + + mfctl %pcsq, spc + b itlb_miss_11 + mfctl %pcoq, va + + .align 32 + .endm +#endif + + /* + * itlb miss interruption handler (parisc 2.0) + */ + + .macro itlb_20 code + mfctl %pcsq, spc +#ifdef __LP64__ + b itlb_miss_20w +#else + b itlb_miss_20 +#endif + mfctl %pcoq, va + + .align 32 + .endm + +#ifndef __LP64__ + /* + * naitlb miss interruption handler (parisc 1.1 - 32 bit) + * + * Note: naitlb misses will be treated + * as an ordinary itlb miss for now. + * However, note that naitlb misses + * have the faulting address in the + * IOR/ISR. + */ + + .macro naitlb_11 code + + mfctl %isr,spc + b itlb_miss_11 + mfctl %ior,va + /* FIXME: If user causes a naitlb miss, the priv level may not be in + * lower bits of va, where the itlb miss handler is expecting them + */ + + .align 32 + .endm +#endif + + /* + * naitlb miss interruption handler (parisc 2.0) + * + * Note: naitlb misses will be treated + * as an ordinary itlb miss for now. + * However, note that naitlb misses + * have the faulting address in the + * IOR/ISR. + */ + + .macro naitlb_20 code + + mfctl %isr,spc +#ifdef __LP64__ + b itlb_miss_20w +#else + b itlb_miss_20 +#endif + mfctl %ior,va + /* FIXME: If user causes a naitlb miss, the priv level may not be in + * lower bits of va, where the itlb miss handler is expecting them + */ + + .align 32 + .endm + +#ifndef __LP64__ + /* + * dtlb miss interruption handler (parisc 1.1 - 32 bit) + */ + + .macro dtlb_11 code + + mfctl %isr, spc + b dtlb_miss_11 + mfctl %ior, va + + .align 32 + .endm +#endif + + /* + * dtlb miss interruption handler (parisc 2.0) + */ + + .macro dtlb_20 code + + mfctl %isr, spc +#ifdef __LP64__ + b dtlb_miss_20w +#else + b dtlb_miss_20 +#endif + mfctl %ior, va + + .align 32 + .endm + +#ifndef __LP64__ + /* nadtlb miss interruption handler (parisc 1.1 - 32 bit) */ + + .macro nadtlb_11 code + + mfctl %isr,spc + b nadtlb_miss_11 + mfctl %ior,va + + .align 32 + .endm +#endif + + /* nadtlb miss interruption handler (parisc 2.0) */ + + .macro nadtlb_20 code + + mfctl %isr,spc +#ifdef __LP64__ + b nadtlb_miss_20w +#else + b nadtlb_miss_20 +#endif + mfctl %ior,va + + .align 32 + .endm + +#ifndef __LP64__ + /* + * dirty bit trap interruption handler (parisc 1.1 - 32 bit) + */ + + .macro dbit_11 code + + mfctl %isr,spc + b dbit_trap_11 + mfctl %ior,va + + .align 32 + .endm +#endif + + /* + * dirty bit trap interruption handler (parisc 2.0) + */ + + .macro dbit_20 code + + mfctl %isr,spc +#ifdef __LP64__ + b dbit_trap_20w +#else + b dbit_trap_20 +#endif + mfctl %ior,va + + .align 32 + .endm + + /* The following are simple 32 vs 64 bit instruction + * abstractions for the macros */ + .macro EXTR reg1,start,length,reg2 +#ifdef __LP64__ + extrd,u \reg1,32+\start,\length,\reg2 +#else + extrw,u \reg1,\start,\length,\reg2 +#endif + .endm + + .macro DEP reg1,start,length,reg2 +#ifdef __LP64__ + depd \reg1,32+\start,\length,\reg2 +#else + depw \reg1,\start,\length,\reg2 +#endif + .endm + + .macro DEPI val,start,length,reg +#ifdef __LP64__ + depdi \val,32+\start,\length,\reg +#else + depwi \val,\start,\length,\reg +#endif + .endm + + /* In LP64, the space contains part of the upper 32 bits of the + * fault. We have to extract this and place it in the va, + * zeroing the corresponding bits in the space register */ + .macro space_adjust spc,va,tmp +#ifdef __LP64__ + extrd,u \spc,63,SPACEID_SHIFT,\tmp + depd %r0,63,SPACEID_SHIFT,\spc + depd \tmp,31,SPACEID_SHIFT,\va +#endif + .endm + + .import swapper_pg_dir,code + + /* Get the pgd. For faults on space zero (kernel space), this + * is simply swapper_pg_dir. For user space faults, the + * pgd is stored in %cr25 */ + .macro get_pgd spc,reg + ldil L%PA(swapper_pg_dir),\reg + ldo R%PA(swapper_pg_dir)(\reg),\reg + or,COND(=) %r0,\spc,%r0 + mfctl %cr25,\reg + .endm + + /* + space_check(spc,tmp,fault) + + spc - The space we saw the fault with. + tmp - The place to store the current space. + fault - Function to call on failure. + + Only allow faults on different spaces from the + currently active one if we're the kernel + + */ + .macro space_check spc,tmp,fault + mfsp %sr7,\tmp + or,COND(<>) %r0,\spc,%r0 /* user may execute gateway page + * as kernel, so defeat the space + * check if it is */ + copy \spc,\tmp + or,COND(=) %r0,\tmp,%r0 /* nullify if executing as kernel */ + cmpb,COND(<>),n \tmp,\spc,\fault + .endm + + /* Look up a PTE in a 2-Level scheme (faulting at each + * level if the entry isn't present + * + * NOTE: we use ldw even for LP64, since the short pointers + * can address up to 1TB + */ + .macro L2_ptep pmd,pte,index,va,fault +#if PT_NLEVELS == 3 + EXTR \va,31-ASM_PMD_SHIFT,ASM_BITS_PER_PMD,\index +#else + EXTR \va,31-ASM_PGDIR_SHIFT,ASM_BITS_PER_PGD,\index +#endif + DEP %r0,31,PAGE_SHIFT,\pmd /* clear offset */ + copy %r0,\pte + ldw,s \index(\pmd),\pmd + bb,>=,n \pmd,_PxD_PRESENT_BIT,\fault + DEP %r0,31,PxD_FLAG_SHIFT,\pmd /* clear flags */ + copy \pmd,%r9 +#ifdef __LP64__ + shld %r9,PxD_VALUE_SHIFT,\pmd +#else + shlw %r9,PxD_VALUE_SHIFT,\pmd +#endif + EXTR \va,31-PAGE_SHIFT,ASM_BITS_PER_PTE,\index + DEP %r0,31,PAGE_SHIFT,\pmd /* clear offset */ + shladd \index,BITS_PER_PTE_ENTRY,\pmd,\pmd + LDREG %r0(\pmd),\pte /* pmd is now pte */ + bb,>=,n \pte,_PAGE_PRESENT_BIT,\fault + .endm + + /* Look up PTE in a 3-Level scheme. + * + * Here we implement a Hybrid L2/L3 scheme: we allocate the + * first pmd adjacent to the pgd. This means that we can + * subtract a constant offset to get to it. The pmd and pgd + * sizes are arranged so that a single pmd covers 4GB (giving + * a full LP64 process access to 8TB) so our lookups are + * effectively L2 for the first 4GB of the kernel (i.e. for + * all ILP32 processes and all the kernel for machines with + * under 4GB of memory) */ + .macro L3_ptep pgd,pte,index,va,fault + extrd,u \va,63-ASM_PGDIR_SHIFT,ASM_BITS_PER_PGD,\index + copy %r0,\pte + extrd,u,*= \va,31,32,%r0 + ldw,s \index(\pgd),\pgd + extrd,u,*= \va,31,32,%r0 + bb,>=,n \pgd,_PxD_PRESENT_BIT,\fault + extrd,u,*= \va,31,32,%r0 + shld \pgd,PxD_VALUE_SHIFT,\index + extrd,u,*= \va,31,32,%r0 + copy \index,\pgd + extrd,u,*<> \va,31,32,%r0 + ldo ASM_PGD_PMD_OFFSET(\pgd),\pgd + L2_ptep \pgd,\pte,\index,\va,\fault + .endm + + /* Set the _PAGE_ACCESSED bit of the PTE. Be clever and + * don't needlessly dirty the cache line if it was already set */ + .macro update_ptep ptep,pte,tmp,tmp1 + ldi _PAGE_ACCESSED,\tmp1 + or \tmp1,\pte,\tmp + and,COND(<>) \tmp1,\pte,%r0 + STREG \tmp,0(\ptep) + .endm + + /* Set the dirty bit (and accessed bit). No need to be + * clever, this is only used from the dirty fault */ + .macro update_dirty ptep,pte,tmp + ldi _PAGE_ACCESSED|_PAGE_DIRTY,\tmp + or \tmp,\pte,\pte + STREG \pte,0(\ptep) + .endm + + /* Convert the pte and prot to tlb insertion values. How + * this happens is quite subtle, read below */ + .macro make_insert_tlb spc,pte,prot + space_to_prot \spc \prot /* create prot id from space */ + /* The following is the real subtlety. This is depositing + * T <-> _PAGE_REFTRAP + * D <-> _PAGE_DIRTY + * B <-> _PAGE_DMB (memory break) + * + * Then incredible subtlety: The access rights are + * _PAGE_GATEWAY _PAGE_EXEC _PAGE_READ + * See 3-14 of the parisc 2.0 manual + * + * Finally, _PAGE_READ goes in the top bit of PL1 (so we + * trigger an access rights trap in user space if the user + * tries to read an unreadable page */ + depd \pte,8,7,\prot + + /* PAGE_USER indicates the page can be read with user privileges, + * so deposit X1|11 to PL1|PL2 (remember the upper bit of PL1 + * contains _PAGE_READ */ + extrd,u,*= \pte,_PAGE_USER_BIT+32,1,%r0 + depdi 7,11,3,\prot + /* If we're a gateway page, drop PL2 back to zero for promotion + * to kernel privilege (so we can execute the page as kernel). + * Any privilege promotion page always denys read and write */ + extrd,u,*= \pte,_PAGE_GATEWAY_BIT+32,1,%r0 + depd %r0,11,2,\prot /* If Gateway, Set PL2 to 0 */ + + /* Get rid of prot bits and convert to page addr for iitlbt */ + + depd %r0,63,PAGE_SHIFT,\pte + extrd,u \pte,56,32,\pte + .endm + + /* Identical macro to make_insert_tlb above, except it + * makes the tlb entry for the differently formatted pa11 + * insertion instructions */ + .macro make_insert_tlb_11 spc,pte,prot + zdep \spc,30,15,\prot + dep \pte,8,7,\prot + extru,= \pte,_PAGE_NO_CACHE_BIT,1,%r0 + depi 1,12,1,\prot + extru,= \pte,_PAGE_USER_BIT,1,%r0 + depi 7,11,3,\prot /* Set for user space (1 rsvd for read) */ + extru,= \pte,_PAGE_GATEWAY_BIT,1,%r0 + depi 0,11,2,\prot /* If Gateway, Set PL2 to 0 */ + + /* Get rid of prot bits and convert to page addr for iitlba */ + + depi 0,31,12,\pte + extru \pte,24,25,\pte + + .endm + + /* This is for ILP32 PA2.0 only. The TLB insertion needs + * to extend into I/O space if the address is 0xfXXXXXXX + * so we extend the f's into the top word of the pte in + * this case */ + .macro f_extend pte,tmp + extrd,s \pte,42,4,\tmp + addi,<> 1,\tmp,%r0 + extrd,s \pte,63,25,\pte + .endm + + /* The alias region is an 8MB aligned 16MB to do clear and + * copy user pages at addresses congruent with the user + * virtual address. + * + * To use the alias page, you set %r26 up with the to TLB + * entry (identifying the physical page) and %r23 up with + * the from tlb entry (or nothing if only a to entry---for + * clear_user_page_asm) */ + .macro do_alias spc,tmp,tmp1,va,pte,prot,fault + cmpib,COND(<>),n 0,\spc,\fault + ldil L%(TMPALIAS_MAP_START),\tmp +#if defined(__LP64__) && (TMPALIAS_MAP_START >= 0x80000000) + /* on LP64, ldi will sign extend into the upper 32 bits, + * which is behaviour we don't want */ + depdi 0,31,32,\tmp +#endif + copy \va,\tmp1 + DEPI 0,31,23,\tmp1 + cmpb,COND(<>),n \tmp,\tmp1,\fault + ldi (_PAGE_DIRTY|_PAGE_WRITE|_PAGE_READ),\prot + depd,z \prot,8,7,\prot + /* + * OK, it is in the temp alias region, check whether "from" or "to". + * Check "subtle" note in pacache.S re: r23/r26. + */ +#ifdef __LP64__ + extrd,u,*= \va,41,1,%r0 +#else + extrw,u,= \va,9,1,%r0 +#endif + or,COND(tr) %r23,%r0,\pte + or %r26,%r0,\pte + .endm + + + /* + * Align fault_vector_20 on 4K boundary so that both + * fault_vector_11 and fault_vector_20 are on the + * same page. This is only necessary as long as we + * write protect the kernel text, which we may stop + * doing once we use large page translations to cover + * the static part of the kernel address space. + */ + + .export fault_vector_20 + + .text + + .align 4096 + +fault_vector_20: + /* First vector is invalid (0) */ + .ascii "cows can fly" + .byte 0 + .align 32 + + hpmc 1 + def 2 + def 3 + extint 4 + def 5 + itlb_20 6 + def 7 + def 8 + def 9 + def 10 + def 11 + def 12 + def 13 + def 14 + dtlb_20 15 +#if 0 + naitlb_20 16 +#else + def 16 +#endif + nadtlb_20 17 + def 18 + def 19 + dbit_20 20 + def 21 + def 22 + def 23 + def 24 + def 25 + def 26 + def 27 + def 28 + def 29 + def 30 + def 31 + +#ifndef __LP64__ + + .export fault_vector_11 + + .align 2048 + +fault_vector_11: + /* First vector is invalid (0) */ + .ascii "cows can fly" + .byte 0 + .align 32 + + hpmc 1 + def 2 + def 3 + extint 4 + def 5 + itlb_11 6 + def 7 + def 8 + def 9 + def 10 + def 11 + def 12 + def 13 + def 14 + dtlb_11 15 +#if 0 + naitlb_11 16 +#else + def 16 +#endif + nadtlb_11 17 + def 18 + def 19 + dbit_11 20 + def 21 + def 22 + def 23 + def 24 + def 25 + def 26 + def 27 + def 28 + def 29 + def 30 + def 31 + +#endif + + .import handle_interruption,code + .import do_cpu_irq_mask,code + + /* + * r26 = function to be called + * r25 = argument to pass in + * r24 = flags for do_fork() + * + * Kernel threads don't ever return, so they don't need + * a true register context. We just save away the arguments + * for copy_thread/ret_ to properly set up the child. + */ + +#define CLONE_VM 0x100 /* Must agree with <linux/sched.h> */ +#define CLONE_UNTRACED 0x00800000 + + .export __kernel_thread, code + .import do_fork +__kernel_thread: + STREG %r2, -RP_OFFSET(%r30) + + copy %r30, %r1 + ldo PT_SZ_ALGN(%r30),%r30 +#ifdef __LP64__ + /* Yo, function pointers in wide mode are little structs... -PB */ + ldd 24(%r26), %r2 + STREG %r2, PT_GR27(%r1) /* Store childs %dp */ + ldd 16(%r26), %r26 + + STREG %r22, PT_GR22(%r1) /* save r22 (arg5) */ + copy %r0, %r22 /* user_tid */ +#endif + STREG %r26, PT_GR26(%r1) /* Store function & argument for child */ + STREG %r25, PT_GR25(%r1) + ldil L%CLONE_UNTRACED, %r26 + ldo CLONE_VM(%r26), %r26 /* Force CLONE_VM since only init_mm */ + or %r26, %r24, %r26 /* will have kernel mappings. */ + ldi 1, %r25 /* stack_start, signals kernel thread */ + stw %r0, -52(%r30) /* user_tid */ +#ifdef __LP64__ + ldo -16(%r30),%r29 /* Reference param save area */ +#endif + BL do_fork, %r2 + copy %r1, %r24 /* pt_regs */ + + /* Parent Returns here */ + + LDREG -PT_SZ_ALGN-RP_OFFSET(%r30), %r2 + ldo -PT_SZ_ALGN(%r30), %r30 + bv %r0(%r2) + nop + + /* + * Child Returns here + * + * copy_thread moved args from temp save area set up above + * into task save area. + */ + + .export ret_from_kernel_thread +ret_from_kernel_thread: + + /* Call schedule_tail first though */ + BL schedule_tail, %r2 + nop + + LDREG TI_TASK-THREAD_SZ_ALGN(%r30), %r1 + LDREG TASK_PT_GR25(%r1), %r26 +#ifdef __LP64__ + LDREG TASK_PT_GR27(%r1), %r27 + LDREG TASK_PT_GR22(%r1), %r22 +#endif + LDREG TASK_PT_GR26(%r1), %r1 + ble 0(%sr7, %r1) + copy %r31, %r2 + +#ifdef __LP64__ + ldo -16(%r30),%r29 /* Reference param save area */ + loadgp /* Thread could have been in a module */ +#endif + b sys_exit + ldi 0, %r26 + + .import sys_execve, code + .export __execve, code +__execve: + copy %r2, %r15 + copy %r30, %r16 + ldo PT_SZ_ALGN(%r30), %r30 + STREG %r26, PT_GR26(%r16) + STREG %r25, PT_GR25(%r16) + STREG %r24, PT_GR24(%r16) +#ifdef __LP64__ + ldo -16(%r30),%r29 /* Reference param save area */ +#endif + BL sys_execve, %r2 + copy %r16, %r26 + + cmpib,=,n 0,%r28,intr_return /* forward */ + + /* yes, this will trap and die. */ + copy %r15, %r2 + copy %r16, %r30 + bv %r0(%r2) + nop + + .align 4 + + /* + * struct task_struct *_switch_to(struct task_struct *prev, + * struct task_struct *next) + * + * switch kernel stacks and return prev */ + .export _switch_to, code +_switch_to: + STREG %r2, -RP_OFFSET(%r30) + + callee_save + + load32 _switch_to_ret, %r2 + + STREG %r2, TASK_PT_KPC(%r26) + LDREG TASK_PT_KPC(%r25), %r2 + + STREG %r30, TASK_PT_KSP(%r26) + LDREG TASK_PT_KSP(%r25), %r30 + LDREG TASK_THREAD_INFO(%r25), %r25 + bv %r0(%r2) + mtctl %r25,%cr30 + +_switch_to_ret: + mtctl %r0, %cr0 /* Needed for single stepping */ + callee_rest + + LDREG -RP_OFFSET(%r30), %r2 + bv %r0(%r2) + copy %r26, %r28 + + /* + * Common rfi return path for interruptions, kernel execve, and + * sys_rt_sigreturn (sometimes). The sys_rt_sigreturn syscall will + * return via this path if the signal was received when the process + * was running; if the process was blocked on a syscall then the + * normal syscall_exit path is used. All syscalls for traced + * proceses exit via intr_restore. + * + * XXX If any syscalls that change a processes space id ever exit + * this way, then we will need to copy %sr3 in to PT_SR[3..7], and + * adjust IASQ[0..1]. + * + * Note that the following code uses a "relied upon translation". + * See the parisc ACD for details. The ssm is necessary due to a + * PCXT bug. + */ + + .align 4096 + + .export syscall_exit_rfi +syscall_exit_rfi: + mfctl %cr30,%r16 + LDREG TI_TASK(%r16), %r16 /* thread_info -> task_struct */ + ldo TASK_REGS(%r16),%r16 + /* Force iaoq to userspace, as the user has had access to our current + * context via sigcontext. Also Filter the PSW for the same reason. + */ + LDREG PT_IAOQ0(%r16),%r19 + depi 3,31,2,%r19 + STREG %r19,PT_IAOQ0(%r16) + LDREG PT_IAOQ1(%r16),%r19 + depi 3,31,2,%r19 + STREG %r19,PT_IAOQ1(%r16) + LDREG PT_PSW(%r16),%r19 + load32 USER_PSW_MASK,%r1 +#ifdef __LP64__ + load32 USER_PSW_HI_MASK,%r20 + depd %r20,31,32,%r1 +#endif + and %r19,%r1,%r19 /* Mask out bits that user shouldn't play with */ + load32 USER_PSW,%r1 + or %r19,%r1,%r19 /* Make sure default USER_PSW bits are set */ + STREG %r19,PT_PSW(%r16) + + /* + * If we aren't being traced, we never saved space registers + * (we don't store them in the sigcontext), so set them + * to "proper" values now (otherwise we'll wind up restoring + * whatever was last stored in the task structure, which might + * be inconsistent if an interrupt occured while on the gateway + * page) Note that we may be "trashing" values the user put in + * them, but we don't support the the user changing them. + */ + + STREG %r0,PT_SR2(%r16) + mfsp %sr3,%r19 + STREG %r19,PT_SR0(%r16) + STREG %r19,PT_SR1(%r16) + STREG %r19,PT_SR3(%r16) + STREG %r19,PT_SR4(%r16) + STREG %r19,PT_SR5(%r16) + STREG %r19,PT_SR6(%r16) + STREG %r19,PT_SR7(%r16) + +intr_return: + /* NOTE: Need to enable interrupts incase we schedule. */ + ssm PSW_SM_I, %r0 + + /* Check for software interrupts */ + + .import irq_stat,data + + load32 irq_stat,%r19 +#ifdef CONFIG_SMP + mfctl %cr30,%r1 + ldw TI_CPU(%r1),%r1 /* get cpu # - int */ + /* shift left ____cacheline_aligned (aka L1_CACHE_BYTES) amount + ** irq_stat[] is defined using ____cacheline_aligned. + */ +#ifdef __LP64__ + shld %r1, 6, %r20 +#else + shlw %r1, 5, %r20 +#endif + add %r19,%r20,%r19 /* now have &irq_stat[smp_processor_id()] */ +#endif /* CONFIG_SMP */ + + LDREG IRQSTAT_SIRQ_PEND(%r19),%r20 /* hardirq.h: unsigned long */ + cmpib,<>,n 0,%r20,intr_do_softirq /* forward */ + +intr_check_resched: + + /* check for reschedule */ + mfctl %cr30,%r1 + LDREG TI_FLAGS(%r1),%r19 /* sched.h: TIF_NEED_RESCHED */ + bb,<,n %r19,31-TIF_NEED_RESCHED,intr_do_resched /* forward */ + +intr_check_sig: + /* As above */ + mfctl %cr30,%r1 + LDREG TI_FLAGS(%r1),%r19 /* sched.h: TIF_SIGPENDING */ + bb,<,n %r19, 31-TIF_SIGPENDING, intr_do_signal /* forward */ + +intr_restore: + copy %r16,%r29 + ldo PT_FR31(%r29),%r1 + rest_fp %r1 + rest_general %r29 + + /* Create a "relied upon translation" PA 2.0 Arch. F-5 */ + ssm 0,%r0 + nop + nop + nop + nop + nop + nop + nop + tophys_r1 %r29 + rsm (PSW_SM_Q|PSW_SM_P|PSW_SM_D|PSW_SM_I),%r0 + + /* Restore space id's and special cr's from PT_REGS + * structure pointed to by r29 */ + rest_specials %r29 + + /* Important: Note that rest_stack restores r29 + * last (we are using it)! It also restores r1 and r30. */ + rest_stack + + rfi + nop + nop + nop + nop + nop + nop + nop + nop + + .import do_softirq,code +intr_do_softirq: + bl do_softirq,%r2 +#ifdef __LP64__ + ldo -16(%r30),%r29 /* Reference param save area */ +#else + nop +#endif + b intr_check_resched + nop + + .import schedule,code +intr_do_resched: + /* Only do reschedule if we are returning to user space */ + LDREG PT_IASQ0(%r16), %r20 + CMPIB= 0,%r20,intr_restore /* backward */ + nop + LDREG PT_IASQ1(%r16), %r20 + CMPIB= 0,%r20,intr_restore /* backward */ + nop + +#ifdef __LP64__ + ldo -16(%r30),%r29 /* Reference param save area */ +#endif + + ldil L%intr_check_sig, %r2 + b schedule + ldo R%intr_check_sig(%r2), %r2 + + + .import do_signal,code +intr_do_signal: + /* + This check is critical to having LWS + working. The IASQ is zero on the gateway + page and we cannot deliver any signals until + we get off the gateway page. + + Only do signals if we are returning to user space + */ + LDREG PT_IASQ0(%r16), %r20 + CMPIB= 0,%r20,intr_restore /* backward */ + nop + LDREG PT_IASQ1(%r16), %r20 + CMPIB= 0,%r20,intr_restore /* backward */ + nop + + copy %r0, %r24 /* unsigned long in_syscall */ + copy %r16, %r25 /* struct pt_regs *regs */ +#ifdef __LP64__ + ldo -16(%r30),%r29 /* Reference param save area */ +#endif + + BL do_signal,%r2 + copy %r0, %r26 /* sigset_t *oldset = NULL */ + + b intr_check_sig + nop + + /* + * External interrupts. + */ + +intr_extint: + CMPIB=,n 0,%r16,1f + get_stack_use_cr30 + b,n 3f + +1: +#if 0 /* Interrupt Stack support not working yet! */ + mfctl %cr31,%r1 + copy %r30,%r17 + /* FIXME! depi below has hardcoded idea of interrupt stack size (32k)*/ +#ifdef __LP64__ + depdi 0,63,15,%r17 +#else + depi 0,31,15,%r17 +#endif + CMPB=,n %r1,%r17,2f + get_stack_use_cr31 + b,n 3f +#endif +2: + get_stack_use_r30 + +3: + save_specials %r29 + virt_map + save_general %r29 + + ldo PT_FR0(%r29), %r24 + save_fp %r24 + + loadgp + + copy %r29, %r26 /* arg0 is pt_regs */ + copy %r29, %r16 /* save pt_regs */ + + ldil L%intr_return, %r2 + +#ifdef __LP64__ + ldo -16(%r30),%r29 /* Reference param save area */ +#endif + + b do_cpu_irq_mask + ldo R%intr_return(%r2), %r2 /* return to intr_return, not here */ + + + /* Generic interruptions (illegal insn, unaligned, page fault, etc) */ + + .export intr_save, code /* for os_hpmc */ + +intr_save: + mfsp %sr7,%r16 + CMPIB=,n 0,%r16,1f + get_stack_use_cr30 + b 2f + copy %r8,%r26 + +1: + get_stack_use_r30 + copy %r8,%r26 + +2: + save_specials %r29 + + /* If this trap is a itlb miss, skip saving/adjusting isr/ior */ + + /* + * FIXME: 1) Use a #define for the hardwired "6" below (and in + * traps.c. + * 2) Once we start executing code above 4 Gb, we need + * to adjust iasq/iaoq here in the same way we + * adjust isr/ior below. + */ + + CMPIB=,n 6,%r26,skip_save_ior + + /* save_specials left ipsw value in r8 for us to test */ + + mfctl %cr20, %r16 /* isr */ + mfctl %cr21, %r17 /* ior */ + +#ifdef __LP64__ + /* + * If the interrupted code was running with W bit off (32 bit), + * clear the b bits (bits 0 & 1) in the ior. + */ + extrd,u,*<> %r8,PSW_W_BIT,1,%r0 + depdi 0,1,2,%r17 + + /* + * FIXME: This code has hardwired assumptions about the split + * between space bits and offset bits. This will change + * when we allow alternate page sizes. + */ + + /* adjust isr/ior. */ + + extrd,u %r16,63,7,%r1 /* get high bits from isr for ior */ + depd %r1,31,7,%r17 /* deposit them into ior */ + depdi 0,63,7,%r16 /* clear them from isr */ +#endif + STREG %r16, PT_ISR(%r29) + STREG %r17, PT_IOR(%r29) + + +skip_save_ior: + virt_map + save_general %r29 + + ldo PT_FR0(%r29), %r25 + save_fp %r25 + + loadgp + + copy %r29, %r25 /* arg1 is pt_regs */ +#ifdef __LP64__ + ldo -16(%r30),%r29 /* Reference param save area */ +#endif + + ldil L%intr_check_sig, %r2 + copy %r25, %r16 /* save pt_regs */ + + b handle_interruption + ldo R%intr_check_sig(%r2), %r2 + + + /* + * Note for all tlb miss handlers: + * + * cr24 contains a pointer to the kernel address space + * page directory. + * + * cr25 contains a pointer to the current user address + * space page directory. + * + * sr3 will contain the space id of the user address space + * of the current running thread while that thread is + * running in the kernel. + */ + + /* + * register number allocations. Note that these are all + * in the shadowed registers + */ + + t0 = r1 /* temporary register 0 */ + va = r8 /* virtual address for which the trap occured */ + t1 = r9 /* temporary register 1 */ + pte = r16 /* pte/phys page # */ + prot = r17 /* prot bits */ + spc = r24 /* space for which the trap occured */ + ptp = r25 /* page directory/page table pointer */ + +#ifdef __LP64__ + +dtlb_miss_20w: + space_adjust spc,va,t0 + get_pgd spc,ptp + space_check spc,t0,dtlb_fault + + L3_ptep ptp,pte,t0,va,dtlb_check_alias_20w + + update_ptep ptp,pte,t0,t1 + + make_insert_tlb spc,pte,prot + + idtlbt pte,prot + + rfir + nop + +dtlb_check_ali |