Index: head/sys/arm64/arm64/copyinout.S =================================================================== --- head/sys/arm64/arm64/copyinout.S (revision 297614) +++ head/sys/arm64/arm64/copyinout.S (revision 297615) @@ -1,218 +1,220 @@ /*- * Copyright (c) 2015 The FreeBSD Foundation * All rights reserved. * * This software was developed by Andrew Turner under * sponsorship from the FreeBSD Foundation. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * */ #include __FBSDID("$FreeBSD$"); #include +#include + #include "assym.s" /* * Fault handler for the copy{in,out} functions below. */ ENTRY(copyio_fault) SET_FAULT_HANDLER(xzr, x1) /* Clear the handler */ copyio_fault_nopcb: mov x0, #EFAULT ret END(copyio_fault) /* * Copies from a kernel to user address * * int copyout(const void *kaddr, void *udaddr, size_t len) */ ENTRY(copyout) cbz x2, 1f adds x3, x1, x2 b.cs copyio_fault_nopcb ldr x4, =VM_MAXUSER_ADDRESS cmp x3, x4 b.hi copyio_fault_nopcb b copycommon 1: mov x0, xzr /* return 0 */ ret END(copyout) /* * Copies from a user to kernel address * * int copyin(const void *uaddr, void *kdaddr, size_t len) */ ENTRY(copyin) cbz x2, 1f adds x3, x0, x2 b.cs copyio_fault_nopcb ldr x4, =VM_MAXUSER_ADDRESS cmp x3, x4 b.hi copyio_fault_nopcb b copycommon 1: mov x0, xzr /* return 0 */ ret END(copyin) /* * Copies a string from a user to kernel address * * int copyinstr(const void *udaddr, void *kaddr, size_t len, size_t *done) */ ENTRY(copyinstr) mov x5, xzr /* count = 0 */ mov w4, #1 /* If zero return faulure */ cbz x2, 3f /* If len == 0 then skip loop */ adr x6, copyio_fault /* Get the handler address */ SET_FAULT_HANDLER(x6, x7) /* Set the handler */ ldr x7, =VM_MAXUSER_ADDRESS 1: cmp x0, x7 b.cs copyio_fault ldrb w4, [x0], #1 /* Load from uaddr */ strb w4, [x1], #1 /* Store in kaddr */ add x5, x5, #1 /* count++ */ cbz w4, 2f /* Break when NUL-terminated */ sub x2, x2, #1 /* len-- */ cbnz x2, 1b 2: SET_FAULT_HANDLER(xzr, x7) /* Clear the handler */ 3: cbz x3, 4f /* Check if done != NULL */ str x5, [x3] /* done = count */ 4: mov w1, #ENAMETOOLONG /* Load ENAMETOOLONG to return if failed */ cmp w4, #0 /* Check if we saved the NUL-terminator */ csel w0, wzr, w1, eq /* If so return success, else failure */ ret END(copyinstr) /* * Local helper * * x0 - src pointer * x1 - dst pointer * x2 - size * lr - the return address, so jump here instead of calling * * This function is optimized to minimize concurrent memory accesses. In * present form it is suited for cores with a single memory prefetching * unit. * ARM64TODO: * Consider using separate functions for each ARM64 core. Adding memory * access interleaving might increase a total throughput on A57 or A72. */ .text .align 4 .local copycommon .type copycommon,@function copycommon: adr x6, copyio_fault /* Get the handler address */ SET_FAULT_HANDLER(x6, x7) /* Set the handler */ /* Check alignment */ orr x3, x0, x1 ands x3, x3, 0x07 b.eq aligned /* Unaligned is byte by byte copy */ byte_by_byte: ldrb w3, [x0], #0x01 strb w3, [x1], #0x01 subs x2, x2, #0x01 b.ne byte_by_byte b ending aligned: cmp x2, #0x10 b.lt lead_out cmp x2, #0x40 b.lt by_dwords_start /* Block copy */ lsr x15, x2, #0x06 by_blocks: ldp x3, x4, [x0], #0x10 ldp x5, x6, [x0], #0x10 ldp x7, x8, [x0], #0x10 ldp x9, x10, [x0], #0x10 stp x3, x4, [x1], #0x10 stp x5, x6, [x1], #0x10 stp x7, x8, [x1], #0x10 stp x9, x10, [x1], #0x10 subs x15, x15, #0x01 b.ne by_blocks and x2, x2, #0x3f by_dwords_start: lsr x15, x2, #0x04 cbz x15, lead_out by_dwords: ldp x3, x4, [x0], #0x10 stp x3, x4, [x1], #0x10 subs x15, x15, #0x01 b.ne by_dwords /* Less than 16 bytes to copy */ lead_out: tbz x2, #0x03, last_word ldr x3, [x0], #0x08 str x3, [x1], #0x08 last_word: tbz x2, #0x02, last_hword ldr w3, [x0], #0x04 str w3, [x1], #0x04 last_hword: tbz x2, #0x01, last_byte ldrh w3, [x0], #0x02 strh w3, [x1], #0x02 last_byte: tbz x2, #0x00, ending ldrb w3, [x0] strb w3, [x1] ending: SET_FAULT_HANDLER(xzr, x7) /* Clear the handler */ mov x0, xzr /* return 0 */ ret .size copycommon, . - copycommon Index: head/sys/arm64/arm64/genassym.c =================================================================== --- head/sys/arm64/arm64/genassym.c (revision 297614) +++ head/sys/arm64/arm64/genassym.c (revision 297615) @@ -1,68 +1,65 @@ /*- * Copyright (c) 2004 Olivier Houchard * Copyright (c) 2014 Andrew Turner * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include -ASSYM(KERNBASE, KERNBASE); -ASSYM(VM_MAXUSER_ADDRESS, VM_MAXUSER_ADDRESS); - ASSYM(TDF_ASTPENDING, TDF_ASTPENDING); ASSYM(TDF_NEEDRESCHED, TDF_NEEDRESCHED); ASSYM(PCPU_SIZE, sizeof(struct pcpu)); ASSYM(PC_CURPCB, offsetof(struct pcpu, pc_curpcb)); ASSYM(PC_CURTHREAD, offsetof(struct pcpu, pc_curthread)); /* Size of pcb, rounded to keep stack alignment */ ASSYM(PCB_SIZE, roundup2(sizeof(struct pcb), STACKALIGNBYTES + 1)); ASSYM(PCB_SINGLE_STEP_SHIFT, PCB_SINGLE_STEP_SHIFT); ASSYM(PCB_REGS, offsetof(struct pcb, pcb_x)); ASSYM(PCB_SP, offsetof(struct pcb, pcb_sp)); ASSYM(PCB_L0ADDR, offsetof(struct pcb, pcb_l0addr)); ASSYM(PCB_ONFAULT, offsetof(struct pcb, pcb_onfault)); ASSYM(PCB_FLAGS, offsetof(struct pcb, pcb_flags)); ASSYM(SF_UC, offsetof(struct sigframe, sf_uc)); ASSYM(TD_PCB, offsetof(struct thread, td_pcb)); ASSYM(TD_FLAGS, offsetof(struct thread, td_flags)); ASSYM(TD_LOCK, offsetof(struct thread, td_lock)); ASSYM(TF_SIZE, sizeof(struct trapframe)); ASSYM(TF_SP, offsetof(struct trapframe, tf_sp)); ASSYM(TF_ELR, offsetof(struct trapframe, tf_elr)); ASSYM(TF_X, offsetof(struct trapframe, tf_x)); Index: head/sys/arm64/arm64/locore.S =================================================================== --- head/sys/arm64/arm64/locore.S (revision 297614) +++ head/sys/arm64/arm64/locore.S (revision 297615) @@ -1,681 +1,682 @@ /*- * Copyright (c) 2012-2014 Andrew Turner * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * $FreeBSD$ */ #include "assym.s" #include "opt_kstack_pages.h" #include #include #include #include #include #include +#include #define VIRT_BITS 48 .globl kernbase .set kernbase, KERNBASE #define DEVICE_MEM 0 #define NORMAL_UNCACHED 1 #define NORMAL_MEM 2 /* * We assume: * MMU on with an identity map, or off * D-Cache: off * I-Cache: on or off * We are loaded at a 2MiB aligned address */ .text .globl _start _start: /* Drop to EL1 */ bl drop_to_el1 /* * Disable the MMU. We may have entered the kernel with it on and * will need to update the tables later. If this has been set up * with anything other than a VA == PA map then this will fail, * but in this case the code to find where we are running from * would have also failed. */ dsb sy mrs x2, sctlr_el1 bic x2, x2, SCTLR_M msr sctlr_el1, x2 isb /* Set the context id */ msr contextidr_el1, xzr /* Get the virt -> phys offset */ bl get_virt_delta /* * At this point: * x29 = PA - VA * x28 = Our physical load address */ /* Create the page tables */ bl create_pagetables /* * At this point: * x27 = TTBR0 table * x26 = Kernel L1 table * x24 = TTBR1 table */ /* Enable the mmu */ bl start_mmu /* Jump to the virtual address space */ ldr x15, .Lvirtdone br x15 virtdone: /* Set up the stack */ adr x25, initstack_end mov sp, x25 sub sp, sp, #PCB_SIZE /* Zero the BSS */ ldr x15, .Lbss ldr x14, .Lend 1: str xzr, [x15], #8 cmp x15, x14 b.lo 1b /* Backup the module pointer */ mov x1, x0 /* Make the page table base a virtual address */ sub x26, x26, x29 sub x24, x24, x29 sub sp, sp, #(64 * 4) mov x0, sp /* Degate the delda so it is VA -> PA */ neg x29, x29 str x1, [x0] /* modulep */ str x26, [x0, 8] /* kern_l1pt */ str x29, [x0, 16] /* kern_delta */ str x25, [x0, 24] /* kern_stack */ str x24, [x0, 32] /* kern_l0pt */ /* trace back starts here */ mov fp, #0 /* Branch to C code */ bl initarm bl mi_startup /* We should not get here */ brk 0 .align 3 .Lvirtdone: .quad virtdone .Lbss: .quad __bss_start .Lend: .quad _end #ifdef SMP /* * mpentry(unsigned long) * * Called by a core when it is being brought online. * The data in x0 is passed straight to init_secondary. */ ENTRY(mpentry) /* Disable interrupts */ msr daifset, #2 /* Drop to EL1 */ bl drop_to_el1 /* Set the context id */ msr contextidr_el1, x1 /* Load the kernel page table */ adr x24, pagetable_l0_ttbr1 /* Load the identity page table */ adr x27, pagetable_l0_ttbr0 /* Enable the mmu */ bl start_mmu /* Jump to the virtual address space */ ldr x15, =mp_virtdone br x15 mp_virtdone: ldr x4, =secondary_stacks mov x5, #(PAGE_SIZE * KSTACK_PAGES) mul x5, x0, x5 add sp, x4, x5 b init_secondary END(mpentry) #endif /* * If we are started in EL2, configure the required hypervisor * registers and drop to EL1. */ drop_to_el1: mrs x1, CurrentEL lsr x1, x1, #2 cmp x1, #0x2 b.eq 1f ret 1: /* Configure the Hypervisor */ mov x2, #(HCR_RW) msr hcr_el2, x2 /* Load the Virtualization Process ID Register */ mrs x2, midr_el1 msr vpidr_el2, x2 /* Load the Virtualization Multiprocess ID Register */ mrs x2, mpidr_el1 msr vmpidr_el2, x2 /* Set the bits that need to be 1 in sctlr_el1 */ ldr x2, .Lsctlr_res1 msr sctlr_el1, x2 /* Don't trap to EL2 for exceptions */ mov x2, #CPTR_RES1 msr cptr_el2, x2 /* Don't trap to EL2 for CP15 traps */ msr hstr_el2, xzr /* Enable access to the physical timers at EL1 */ mrs x2, cnthctl_el2 orr x2, x2, #(CNTHCTL_EL1PCTEN | CNTHCTL_EL1PCEN) msr cnthctl_el2, x2 /* Set the counter offset to a known value */ msr cntvoff_el2, xzr /* Hypervisor trap functions */ adr x2, hyp_vectors msr vbar_el2, x2 mov x2, #(PSR_F | PSR_I | PSR_A | PSR_D | PSR_M_EL1h) msr spsr_el2, x2 /* Configure GICv3 CPU interface */ mrs x2, id_aa64pfr0_el1 /* Extract GIC bits from the register */ ubfx x2, x2, #ID_AA64PFR0_GIC_SHIFT, #ID_AA64PFR0_GIC_BITS /* GIC[3:0] == 0001 - GIC CPU interface via special regs. supported */ cmp x2, #(ID_AA64PFR0_GIC_CPUIF_EN >> ID_AA64PFR0_GIC_SHIFT) b.ne 2f mrs x2, icc_sre_el2 orr x2, x2, #ICC_SRE_EL2_EN /* Enable access from insecure EL1 */ orr x2, x2, #ICC_SRE_EL2_SRE /* Enable system registers */ msr icc_sre_el2, x2 2: /* Set the address to return to our return address */ msr elr_el2, x30 isb eret .align 3 .Lsctlr_res1: .quad SCTLR_RES1 #define VECT_EMPTY \ .align 7; \ 1: b 1b .align 11 hyp_vectors: VECT_EMPTY /* Synchronous EL2t */ VECT_EMPTY /* IRQ EL2t */ VECT_EMPTY /* FIQ EL2t */ VECT_EMPTY /* Error EL2t */ VECT_EMPTY /* Synchronous EL2h */ VECT_EMPTY /* IRQ EL2h */ VECT_EMPTY /* FIQ EL2h */ VECT_EMPTY /* Error EL2h */ VECT_EMPTY /* Synchronous 64-bit EL1 */ VECT_EMPTY /* IRQ 64-bit EL1 */ VECT_EMPTY /* FIQ 64-bit EL1 */ VECT_EMPTY /* Error 64-bit EL1 */ VECT_EMPTY /* Synchronous 32-bit EL1 */ VECT_EMPTY /* IRQ 32-bit EL1 */ VECT_EMPTY /* FIQ 32-bit EL1 */ VECT_EMPTY /* Error 32-bit EL1 */ /* * Get the delta between the physical address we were loaded to and the * virtual address we expect to run from. This is used when building the * initial page table. */ get_virt_delta: /* Load the physical address of virt_map */ adr x29, virt_map /* Load the virtual address of virt_map stored in virt_map */ ldr x28, [x29] /* Find PA - VA as PA' = VA' - VA + PA = VA' + (PA - VA) = VA' + x29 */ sub x29, x29, x28 /* Find the load address for the kernel */ mov x28, #(KERNBASE) add x28, x28, x29 ret .align 3 virt_map: .quad virt_map /* * This builds the page tables containing the identity map, and the kernel * virtual map. * * It relys on: * We were loaded to an address that is on a 2MiB boundary * All the memory must not cross a 1GiB boundaty * x28 contains the physical address we were loaded from * * TODO: This is out of date. * There are at least 5 pages before that address for the page tables * The pages used are: * - The identity (PA = VA) table (TTBR0) * - The Kernel L1 table (TTBR1)(not yet) * - The PA != VA L2 table to jump into (not yet) * - The FDT L2 table (not yet) */ create_pagetables: /* Save the Link register */ mov x5, x30 /* Clean the page table */ adr x6, pagetable mov x26, x6 adr x27, pagetable_end 1: stp xzr, xzr, [x6], #16 stp xzr, xzr, [x6], #16 stp xzr, xzr, [x6], #16 stp xzr, xzr, [x6], #16 cmp x6, x27 b.lo 1b /* * Build the TTBR1 maps. */ /* Find the size of the kernel */ mov x6, #(KERNBASE) ldr x7, .Lend /* Find the end - begin */ sub x8, x7, x6 /* Get the number of l2 pages to allocate, rounded down */ lsr x10, x8, #(L2_SHIFT) /* Add 8 MiB for any rounding above and the module data */ add x10, x10, #4 /* Create the kernel space L2 table */ mov x6, x26 mov x7, #NORMAL_MEM mov x8, #(KERNBASE & L2_BLOCK_MASK) mov x9, x28 bl build_l2_block_pagetable /* Move to the l1 table */ add x26, x26, #PAGE_SIZE /* Link the l1 -> l2 table */ mov x9, x6 mov x6, x26 bl link_l1_pagetable /* Move to the l0 table */ add x24, x26, #PAGE_SIZE /* Link the l0 -> l1 table */ mov x9, x6 mov x6, x24 bl link_l0_pagetable /* * Build the TTBR0 maps. */ add x27, x24, #PAGE_SIZE mov x6, x27 /* The initial page table */ #if defined(SOCDEV_PA) && defined(SOCDEV_VA) /* Create a table for the UART */ mov x7, #DEVICE_MEM mov x8, #(SOCDEV_VA) /* VA start */ mov x9, #(SOCDEV_PA) /* PA start */ mov x10, #1 bl build_l1_block_pagetable #endif /* Create the VA = PA map */ mov x7, #NORMAL_UNCACHED /* Uncached as it's only needed early on */ mov x9, x27 mov x8, x9 /* VA start (== PA start) */ mov x10, #1 bl build_l1_block_pagetable /* Move to the l0 table */ add x27, x27, #PAGE_SIZE /* Link the l0 -> l1 table */ mov x9, x6 mov x6, x27 bl link_l0_pagetable /* Restore the Link register */ mov x30, x5 ret /* * Builds an L0 -> L1 table descriptor * * This is a link for a 512GiB block of memory with up to 1GiB regions mapped * within it by build_l1_block_pagetable. * * x6 = L0 table * x8 = Virtual Address * x9 = L1 PA (trashed) * x11, x12 and x13 are trashed */ link_l0_pagetable: /* * Link an L0 -> L1 table entry. */ /* Find the table index */ lsr x11, x8, #L0_SHIFT and x11, x11, #L0_ADDR_MASK /* Build the L0 block entry */ mov x12, #L0_TABLE /* Only use the output address bits */ lsr x9, x9, #12 orr x12, x12, x9, lsl #12 /* Store the entry */ str x12, [x6, x11, lsl #3] ret /* * Builds an L1 -> L2 table descriptor * * This is a link for a 1GiB block of memory with up to 2MiB regions mapped * within it by build_l2_block_pagetable. * * x6 = L1 table * x8 = Virtual Address * x9 = L2 PA (trashed) * x11, x12 and x13 are trashed */ link_l1_pagetable: /* * Link an L1 -> L2 table entry. */ /* Find the table index */ lsr x11, x8, #L1_SHIFT and x11, x11, #Ln_ADDR_MASK /* Build the L1 block entry */ mov x12, #L1_TABLE /* Only use the output address bits */ lsr x9, x9, #12 orr x12, x12, x9, lsl #12 /* Store the entry */ str x12, [x6, x11, lsl #3] ret /* * Builds count 1 GiB page table entry * x6 = L1 table * x7 = Type (0 = Device, 1 = Normal) * x8 = VA start * x9 = PA start (trashed) * x10 = Entry count (TODO) * x11, x12 and x13 are trashed */ build_l1_block_pagetable: /* * Build the L1 table entry. */ /* Find the table index */ lsr x11, x8, #L1_SHIFT and x11, x11, #Ln_ADDR_MASK /* Build the L1 block entry */ lsl x12, x7, #2 orr x12, x12, #L1_BLOCK orr x12, x12, #(ATTR_AF) #ifdef SMP orr x12, x12, ATTR_SH(ATTR_SH_IS) #endif /* Only use the output address bits */ lsr x9, x9, #L1_SHIFT /* Set the physical address for this virtual address */ 1: orr x12, x12, x9, lsl #L1_SHIFT /* Store the entry */ str x12, [x6, x11, lsl #3] /* Clear the address bits */ and x12, x12, #ATTR_MASK_L sub x10, x10, #1 add x11, x11, #1 add x9, x9, #1 cbnz x10, 1b 2: ret /* * Builds count 2 MiB page table entry * x6 = L2 table * x7 = Type (0 = Device, 1 = Normal) * x8 = VA start * x9 = PA start (trashed) * x10 = Entry count (TODO) * x11, x12 and x13 are trashed */ build_l2_block_pagetable: /* * Build the L2 table entry. */ /* Find the table index */ lsr x11, x8, #L2_SHIFT and x11, x11, #Ln_ADDR_MASK /* Build the L2 block entry */ lsl x12, x7, #2 orr x12, x12, #L2_BLOCK orr x12, x12, #(ATTR_AF) #ifdef SMP orr x12, x12, ATTR_SH(ATTR_SH_IS) #endif /* Only use the output address bits */ lsr x9, x9, #L2_SHIFT /* Set the physical address for this virtual address */ 1: orr x12, x12, x9, lsl #L2_SHIFT /* Store the entry */ str x12, [x6, x11, lsl #3] /* Clear the address bits */ and x12, x12, #ATTR_MASK_L sub x10, x10, #1 add x11, x11, #1 add x9, x9, #1 cbnz x10, 1b 2: ret start_mmu: dsb sy /* Load the exception vectors */ ldr x2, =exception_vectors msr vbar_el1, x2 /* Load ttbr0 and ttbr1 */ msr ttbr0_el1, x27 msr ttbr1_el1, x24 isb /* Clear the Monitor Debug System control register */ msr mdscr_el1, xzr /* Invalidate the TLB */ tlbi vmalle1is ldr x2, mair msr mair_el1, x2 /* * Setup TCR according to PARange bits from ID_AA64MMFR0_EL1. */ ldr x2, tcr mrs x3, id_aa64mmfr0_el1 bfi x2, x3, #32, #3 msr tcr_el1, x2 /* Setup SCTLR */ ldr x2, sctlr_set ldr x3, sctlr_clear mrs x1, sctlr_el1 bic x1, x1, x3 /* Clear the required bits */ orr x1, x1, x2 /* Set the required bits */ msr sctlr_el1, x1 isb ret .align 3 mair: /* Device Normal, no cache Normal, write-back */ .quad MAIR_ATTR(0x00, 0) | MAIR_ATTR(0x44, 1) | MAIR_ATTR(0xff, 2) tcr: .quad (TCR_TxSZ(64 - VIRT_BITS) | TCR_ASID_16 | TCR_TG1_4K | \ TCR_CACHE_ATTRS | TCR_SMP_ATTRS) sctlr_set: /* Bits to set */ .quad (SCTLR_UCI | SCTLR_nTWE | SCTLR_nTWI | SCTLR_UCT | SCTLR_DZE | \ SCTLR_I | SCTLR_SED | SCTLR_SA0 | SCTLR_SA | SCTLR_C | SCTLR_M) sctlr_clear: /* Bits to clear */ .quad (SCTLR_EE | SCTLR_EOE | SCTLR_WXN | SCTLR_UMA | SCTLR_ITD | \ SCTLR_THEE | SCTLR_CP15BEN | SCTLR_A) .globl abort abort: b abort //.section .init_pagetable .align 12 /* 4KiB aligned */ /* * 3 initial tables (in the following order): * L2 for kernel (High addresses) * L1 for kernel * L1 for user (Low addresses) */ pagetable: .space PAGE_SIZE pagetable_l1_ttbr1: .space PAGE_SIZE pagetable_l0_ttbr1: .space PAGE_SIZE pagetable_l1_ttbr0: .space PAGE_SIZE pagetable_l0_ttbr0: .space PAGE_SIZE pagetable_end: el2_pagetable: .space PAGE_SIZE .globl init_pt_va init_pt_va: .quad pagetable /* XXX: Keep page tables VA */ .align 4 initstack: .space (PAGE_SIZE * KSTACK_PAGES) initstack_end: ENTRY(sigcode) mov x0, sp add x0, x0, #SF_UC 1: mov x8, #SYS_sigreturn svc 0 /* sigreturn failed, exit */ mov x8, #SYS_exit svc 0 b 1b END(sigcode) /* This may be copied to the stack, keep it 16-byte aligned */ .align 3 esigcode: .data .align 3 .global szsigcode szsigcode: .quad esigcode - sigcode Index: head/sys/arm64/arm64/support.S =================================================================== --- head/sys/arm64/arm64/support.S (revision 297614) +++ head/sys/arm64/arm64/support.S (revision 297615) @@ -1,328 +1,329 @@ /*- * Copyright (c) 2014 Andrew Turner * Copyright (c) 2014-2015 The FreeBSD Foundation * All rights reserved. * * Portions of this software were developed by Andrew Turner * under sponsorship from the FreeBSD Foundation * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * */ #include __FBSDID("$FreeBSD$"); #include #include +#include #include "assym.s" /* * One of the fu* or su* functions failed, return -1. */ ENTRY(fsu_fault) SET_FAULT_HANDLER(xzr, x1) /* Reset the handler function */ fsu_fault_nopcb: mov x0, #-1 ret END(fsu_fault) /* * int casueword32(volatile uint32_t *, uint32_t, uint32_t *, uint32_t) */ ENTRY(casueword32) ldr x4, =(VM_MAXUSER_ADDRESS-3) cmp x0, x4 b.cs fsu_fault_nopcb adr x6, fsu_fault /* Load the fault handler */ SET_FAULT_HANDLER(x6, x4) /* And set it */ 1: ldxr w4, [x0] /* Load-exclusive the data */ cmp w4, w1 /* Compare */ b.ne 2f /* Not equal, exit */ stxr w5, w3, [x0] /* Store the new data */ cbnz w5, 1b /* Retry on failure */ 2: SET_FAULT_HANDLER(xzr, x5) /* Reset the fault handler */ str w4, [x2] /* Store the read data */ mov x0, #0 /* Success */ ret /* Return */ END(casueword32) /* * int casueword(volatile u_long *, u_long, u_long *, u_long) */ ENTRY(casueword) ldr x4, =(VM_MAXUSER_ADDRESS-7) cmp x0, x4 b.cs fsu_fault_nopcb adr x6, fsu_fault /* Load the fault handler */ SET_FAULT_HANDLER(x6, x4) /* And set it */ 1: ldxr x4, [x0] /* Load-exclusive the data */ cmp x4, x1 /* Compare */ b.ne 2f /* Not equal, exit */ stxr w5, x3, [x0] /* Store the new data */ cbnz w5, 1b /* Retry on failure */ 2: SET_FAULT_HANDLER(xzr, x5) /* Reset the fault handler */ str x4, [x2] /* Store the read data */ mov x0, #0 /* Success */ ret /* Return */ END(casueword) /* * int fubyte(volatile const void *) */ ENTRY(fubyte) ldr x1, =VM_MAXUSER_ADDRESS cmp x0, x1 b.cs fsu_fault_nopcb adr x6, fsu_fault /* Load the fault handler */ SET_FAULT_HANDLER(x6, x1) /* And set it */ ldrb w0, [x0] /* Try loading the data */ SET_FAULT_HANDLER(xzr, x1) /* Reset the fault handler */ ret /* Return */ END(fubyte) /* * int fuword(volatile const void *) */ ENTRY(fuword16) ldr x1, =(VM_MAXUSER_ADDRESS-1) cmp x0, x1 b.cs fsu_fault_nopcb adr x6, fsu_fault /* Load the fault handler */ SET_FAULT_HANDLER(x6, x1) /* And set it */ ldrh w0, [x0] /* Try loading the data */ SET_FAULT_HANDLER(xzr, x1) /* Reset the fault handler */ ret /* Return */ END(fuword16) /* * int32_t fueword32(volatile const void *, int32_t *) */ ENTRY(fueword32) ldr x2, =(VM_MAXUSER_ADDRESS-3) cmp x0, x2 b.cs fsu_fault_nopcb adr x6, fsu_fault /* Load the fault handler */ SET_FAULT_HANDLER(x6, x2) /* And set it */ ldr w0, [x0] /* Try loading the data */ SET_FAULT_HANDLER(xzr, x2) /* Reset the fault handler */ str w0, [x1] /* Save the data in kernel space */ mov w0, #0 /* Success */ ret /* Return */ END(fueword32) /* * long fueword(volatile const void *, int64_t *) * int64_t fueword64(volatile const void *, int64_t *) */ ENTRY(fueword) EENTRY(fueword64) ldr x2, =(VM_MAXUSER_ADDRESS-7) cmp x0, x2 b.cs fsu_fault_nopcb adr x6, fsu_fault /* Load the fault handler */ SET_FAULT_HANDLER(x6, x2) /* And set it */ ldr x0, [x0] /* Try loading the data */ SET_FAULT_HANDLER(xzr, x2) /* Reset the fault handler */ str x0, [x1] /* Save the data in kernel space */ mov x0, #0 /* Success */ ret /* Return */ EEND(fueword64) END(fueword) /* * int subyte(volatile void *, int) */ ENTRY(subyte) ldr x2, =VM_MAXUSER_ADDRESS cmp x0, x2 b.cs fsu_fault_nopcb adr x6, fsu_fault /* Load the fault handler */ SET_FAULT_HANDLER(x6, x2) /* And set it */ strb w1, [x0] /* Try storing the data */ SET_FAULT_HANDLER(xzr, x2) /* Reset the fault handler */ mov x0, #0 /* Success */ ret /* Return */ END(subyte) /* * int suword16(volatile void *, int) */ ENTRY(suword16) ldr x2, =(VM_MAXUSER_ADDRESS-1) cmp x0, x2 b.cs fsu_fault_nopcb adr x6, fsu_fault /* Load the fault handler */ SET_FAULT_HANDLER(x6, x2) /* And set it */ strh w1, [x0] /* Try storing the data */ SET_FAULT_HANDLER(xzr, x2) /* Reset the fault handler */ mov x0, #0 /* Success */ ret /* Return */ END(suword16) /* * int suword32(volatile void *, int) */ ENTRY(suword32) ldr x2, =(VM_MAXUSER_ADDRESS-3) cmp x0, x2 b.cs fsu_fault_nopcb adr x6, fsu_fault /* Load the fault handler */ SET_FAULT_HANDLER(x6, x2) /* And set it */ str w1, [x0] /* Try storing the data */ SET_FAULT_HANDLER(xzr, x2) /* Reset the fault handler */ mov x0, #0 /* Success */ ret /* Return */ END(suword32) /* * int suword(volatile void *, long) */ ENTRY(suword) EENTRY(suword64) ldr x2, =(VM_MAXUSER_ADDRESS-7) cmp x0, x2 b.cs fsu_fault_nopcb adr x6, fsu_fault /* Load the fault handler */ SET_FAULT_HANDLER(x6, x2) /* And set it */ str x1, [x0] /* Try storing the data */ SET_FAULT_HANDLER(xzr, x2) /* Reset the fault handler */ mov x0, #0 /* Success */ ret /* Return */ EEND(suword64) END(suword) /* * fuswintr and suswintr are just like fusword and susword except that if * the page is not in memory or would cause a trap, then we return an error. * The important thing is to prevent sleep() and switch(). */ /* * Special handler so the trap code knows not to sleep. */ ENTRY(fsu_intr_fault) SET_FAULT_HANDLER(xzr, x1) /* Reset the handler function */ mov x0, #-1 ret END(fsu_fault) /* * int fuswintr(void *) */ ENTRY(fuswintr) ldr x1, =(VM_MAXUSER_ADDRESS-3) cmp x0, x1 b.cs fsu_fault_nopcb adr x6, fsu_intr_fault /* Load the fault handler */ SET_FAULT_HANDLER(x6, x1) /* And set it */ ldr w0, [x0] /* Try loading the data */ SET_FAULT_HANDLER(xzr, x1) /* Reset the fault handler */ ret /* Return */ END(fuswintr) /* * int suswintr(void *base, int word) */ ENTRY(suswintr) ldr x2, =(VM_MAXUSER_ADDRESS-3) cmp x0, x2 b.cs fsu_fault_nopcb adr x6, fsu_intr_fault /* Load the fault handler */ SET_FAULT_HANDLER(x6, x2) /* And set it */ str w1, [x0] /* Try storing the data */ SET_FAULT_HANDLER(xzr, x2) /* Reset the fault handler */ mov x0, #0 /* Success */ ret /* Return */ END(suswintr) ENTRY(setjmp) /* Store the stack pointer */ mov x8, sp str x8, [x0], #8 /* Store the general purpose registers and lr */ stp x19, x20, [x0], #16 stp x21, x22, [x0], #16 stp x23, x24, [x0], #16 stp x25, x26, [x0], #16 stp x27, x28, [x0], #16 stp x29, lr, [x0], #16 /* Return value */ mov x0, #0 ret END(setjmp) ENTRY(longjmp) /* Restore the stack pointer */ ldr x8, [x0], #8 mov sp, x8 /* Restore the general purpose registers and lr */ ldp x19, x20, [x0], #16 ldp x21, x22, [x0], #16 ldp x23, x24, [x0], #16 ldp x25, x26, [x0], #16 ldp x27, x28, [x0], #16 ldp x29, lr, [x0], #16 /* Load the return value */ mov x0, x1 ret END(longjmp) /* * pagezero, simple implementation */ ENTRY(pagezero_simple) add x1, x0, #PAGE_SIZE 1: stp xzr, xzr, [x0], #0x10 stp xzr, xzr, [x0], #0x10 stp xzr, xzr, [x0], #0x10 stp xzr, xzr, [x0], #0x10 cmp x0, x1 b.ne 1b ret END(pagezero_simple) /* * pagezero, cache assisted */ ENTRY(pagezero_cache) add x1, x0, #PAGE_SIZE ldr x2, =dczva_line_size ldr x2, [x2] 1: dc zva, x0 add x0, x0, x2 cmp x0, x1 b.ne 1b ret END(pagezero_cache) Index: head/sys/arm64/include/vmparam.h =================================================================== --- head/sys/arm64/include/vmparam.h (revision 297614) +++ head/sys/arm64/include/vmparam.h (revision 297615) @@ -1,238 +1,242 @@ /*- * Copyright (c) 1990 The Regents of the University of California. * All rights reserved. * Copyright (c) 1994 John S. Dyson * All rights reserved. * * This code is derived from software contributed to Berkeley by * William Jolitz. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * from: @(#)vmparam.h 5.9 (Berkeley) 5/12/91 * from: FreeBSD: src/sys/i386/include/vmparam.h,v 1.33 2000/03/30 * $FreeBSD$ */ #ifndef _MACHINE_VMPARAM_H_ #define _MACHINE_VMPARAM_H_ /* * Virtual memory related constants, all in bytes */ #ifndef MAXTSIZ #define MAXTSIZ (1*1024*1024*1024) /* max text size */ #endif #ifndef DFLDSIZ #define DFLDSIZ (128*1024*1024) /* initial data size limit */ #endif #ifndef MAXDSIZ #define MAXDSIZ (1*1024*1024*1024) /* max data size */ #endif #ifndef DFLSSIZ #define DFLSSIZ (128*1024*1024) /* initial stack size limit */ #endif #ifndef MAXSSIZ #define MAXSSIZ (1*1024*1024*1024) /* max stack size */ #endif #ifndef SGROWSIZ #define SGROWSIZ (128*1024) /* amount to grow stack */ #endif /* * The physical address space is sparsely populated. */ #define VM_PHYSSEG_SPARSE /* * The number of PHYSSEG entries must be one greater than the number * of phys_avail entries because the phys_avail entry that spans the * largest physical address that is accessible by ISA DMA is split * into two PHYSSEG entries. */ #define VM_PHYSSEG_MAX 64 /* * Create two free page pools: VM_FREEPOOL_DEFAULT is the default pool * from which physical pages are allocated and VM_FREEPOOL_DIRECT is * the pool from which physical pages for small UMA objects are * allocated. */ #define VM_NFREEPOOL 2 #define VM_FREEPOOL_DEFAULT 0 #define VM_FREEPOOL_DIRECT 1 /* * Create two free page lists: VM_FREELIST_DEFAULT is for physical * pages that are above the largest physical address that is * accessible by ISA DMA and VM_FREELIST_ISADMA is for physical pages * that are below that address. */ #define VM_NFREELIST 2 #define VM_FREELIST_DEFAULT 0 #define VM_FREELIST_ISADMA 1 /* * An allocation size of 16MB is supported in order to optimize the * use of the direct map by UMA. Specifically, a cache line contains * at most four TTEs, collectively mapping 16MB of physical memory. * By reducing the number of distinct 16MB "pages" that are used by UMA, * the physical memory allocator reduces the likelihood of both 4MB * page TLB misses and cache misses caused by 4MB page TLB misses. */ #define VM_NFREEORDER 12 /* * Enable superpage reservations: 1 level. */ #ifndef VM_NRESERVLEVEL #define VM_NRESERVLEVEL 1 #endif /* * Level 0 reservations consist of 512 pages. */ #ifndef VM_LEVEL_0_ORDER #define VM_LEVEL_0_ORDER 9 #endif /** * Address space layout. * * ARMv8 implements up to a 48 bit virtual address space. The address space is * split into 2 regions at each end of the 64 bit address space, with an * out of range "hole" in the middle. * * We limit the size of the two spaces to 39 bits each. * * Upper region: 0xffffffffffffffff * 0xffffff8000000000 * * Hole: 0xffffff7fffffffff * 0x0000008000000000 * * Lower region: 0x0000007fffffffff * 0x0000000000000000 * * We use the upper region for the kernel, and the lower region for userland. * * We define some interesting address constants: * * VM_MIN_ADDRESS and VM_MAX_ADDRESS define the start and end of the entire * 64 bit address space, mostly just for convenience. * * VM_MIN_KERNEL_ADDRESS and VM_MAX_KERNEL_ADDRESS define the start and end of * mappable kernel virtual address space. * * VM_MIN_USER_ADDRESS and VM_MAX_USER_ADDRESS define the start and end of the * user address space. */ #define VM_MIN_ADDRESS (0x0000000000000000UL) #define VM_MAX_ADDRESS (0xffffffffffffffffUL) /* 32 GiB of kernel addresses */ #define VM_MIN_KERNEL_ADDRESS (0xffffff8000000000UL) #define VM_MAX_KERNEL_ADDRESS (0xffffff8800000000UL) /* Direct Map for 128 GiB of PA: 0x0 - 0x1fffffffff */ #define DMAP_MIN_ADDRESS (0xffffffc000000000UL) #define DMAP_MAX_ADDRESS (0xffffffdfffffffffUL) -extern vm_paddr_t dmap_phys_base; #define DMAP_MIN_PHYSADDR (dmap_phys_base) #define DMAP_MAX_PHYSADDR (dmap_phys_base + (DMAP_MAX_ADDRESS - DMAP_MIN_ADDRESS)) /* True if pa is in the dmap range */ #define PHYS_IN_DMAP(pa) ((pa) >= DMAP_MIN_PHYSADDR && \ (pa) <= DMAP_MAX_PHYSADDR) /* True if va is in the dmap range */ #define VIRT_IN_DMAP(va) ((va) >= DMAP_MIN_ADDRESS && \ (va) <= DMAP_MAX_ADDRESS) #define PHYS_TO_DMAP(pa) \ ({ \ KASSERT(PHYS_IN_DMAP(pa), \ ("%s: PA out of range, PA: 0x%lx", __func__, \ (vm_paddr_t)(pa))); \ ((pa) - dmap_phys_base) | DMAP_MIN_ADDRESS; \ }) #define DMAP_TO_PHYS(va) \ ({ \ KASSERT(VIRT_IN_DMAP(va), \ ("%s: VA out of range, VA: 0x%lx", __func__, \ (vm_offset_t)(va))); \ ((va) & ~DMAP_MIN_ADDRESS) + dmap_phys_base; \ }) #define VM_MIN_USER_ADDRESS (0x0000000000000000UL) #define VM_MAX_USER_ADDRESS (0x0001000000000000UL) #define VM_MINUSER_ADDRESS (VM_MIN_USER_ADDRESS) #define VM_MAXUSER_ADDRESS (VM_MAX_USER_ADDRESS) #define KERNBASE (VM_MIN_KERNEL_ADDRESS) #define SHAREDPAGE (VM_MAXUSER_ADDRESS - PAGE_SIZE) #define USRSTACK SHAREDPAGE /* * How many physical pages per kmem arena virtual page. */ #ifndef VM_KMEM_SIZE_SCALE #define VM_KMEM_SIZE_SCALE (3) #endif /* * Optional floor (in bytes) on the size of the kmem arena. */ #ifndef VM_KMEM_SIZE_MIN #define VM_KMEM_SIZE_MIN (16 * 1024 * 1024) #endif /* * Optional ceiling (in bytes) on the size of the kmem arena: 60% of the * kernel map. */ #ifndef VM_KMEM_SIZE_MAX #define VM_KMEM_SIZE_MAX ((VM_MAX_KERNEL_ADDRESS - \ VM_MIN_KERNEL_ADDRESS + 1) * 3 / 5) #endif /* * Initial pagein size of beginning of executable file. */ #ifndef VM_INITIAL_PAGEIN #define VM_INITIAL_PAGEIN 16 #endif #define UMA_MD_SMALL_ALLOC +#ifndef LOCORE + +extern vm_paddr_t dmap_phys_base; extern u_int tsb_kernel_ldd_phys; extern vm_offset_t vm_max_kernel_address; extern vm_offset_t init_pt_va; + +#endif #define ZERO_REGION_SIZE (64 * 1024) /* 64KB */ #endif /* !_MACHINE_VMPARAM_H_ */