Index: projects/clang1000-import/lib/libusb/libusb_global_linux.h =================================================================== --- projects/clang1000-import/lib/libusb/libusb_global_linux.h (revision 358848) +++ projects/clang1000-import/lib/libusb/libusb_global_linux.h (revision 358849) @@ -1,80 +1,87 @@ /* $FreeBSD$ */ /*- * SPDX-License-Identifier: BSD-2-Clause-FreeBSD * * Copyright (c) 2013 Hans Petter Selasky. 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. */ #ifndef _LIBUSB_GLOBAL_LINUX_H_ #define _LIBUSB_GLOBAL_LINUX_H_ #define _XOPEN_SOURCE #define _BSD_SOURCE #ifdef __linux__ #define _POSIX_SOURCE #endif #define _POSIX_C_SOURCE 200809 #include #include #include #include #include #include #include #ifdef __linux__ #include #endif #include #include #include #include #include #include #include #include #include #include #include #include #define IOUSB(a) FBSD_L##a #ifndef __aligned #define __aligned(x) __attribute__((__aligned__(x))) #endif #ifndef __packed #define __packed __attribute__((__packed__)) #endif #ifndef strlcpy #define strlcpy(d,s,len) do { \ strncpy(d,s,len); \ ((char *)d)[(len) - 1] = 0; \ } while (0) #endif +#ifndef TAILQ_FOREACH_SAFE +#define TAILQ_FOREACH_SAFE(var, head, field, tvar) \ + for ((var) = TAILQ_FIRST((head)); \ + (var) && ((tvar) = TAILQ_NEXT((var), field), 1); \ + (var) = (tvar)) +#endif + #endif /* _LIBUSB_GLOBAL_LINUX_H_ */ Index: projects/clang1000-import/sys/amd64/vmm/intel/vmx.c =================================================================== --- projects/clang1000-import/sys/amd64/vmm/intel/vmx.c (revision 358848) +++ projects/clang1000-import/sys/amd64/vmm/intel/vmx.c (revision 358849) @@ -1,3839 +1,3896 @@ /*- * SPDX-License-Identifier: BSD-2-Clause-FreeBSD * * Copyright (c) 2011 NetApp, Inc. * All rights reserved. * Copyright (c) 2018 Joyent, Inc. * * 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 NETAPP, INC ``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 NETAPP, INC 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 __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "vmm_lapic.h" #include "vmm_host.h" #include "vmm_ioport.h" #include "vmm_ktr.h" #include "vmm_stat.h" #include "vatpic.h" #include "vlapic.h" #include "vlapic_priv.h" #include "ept.h" #include "vmx_cpufunc.h" #include "vmx.h" #include "vmx_msr.h" #include "x86.h" #include "vmx_controls.h" #define PINBASED_CTLS_ONE_SETTING \ (PINBASED_EXTINT_EXITING | \ PINBASED_NMI_EXITING | \ PINBASED_VIRTUAL_NMI) #define PINBASED_CTLS_ZERO_SETTING 0 #define PROCBASED_CTLS_WINDOW_SETTING \ (PROCBASED_INT_WINDOW_EXITING | \ PROCBASED_NMI_WINDOW_EXITING) #define PROCBASED_CTLS_ONE_SETTING \ (PROCBASED_SECONDARY_CONTROLS | \ PROCBASED_MWAIT_EXITING | \ PROCBASED_MONITOR_EXITING | \ PROCBASED_IO_EXITING | \ PROCBASED_MSR_BITMAPS | \ PROCBASED_CTLS_WINDOW_SETTING | \ PROCBASED_CR8_LOAD_EXITING | \ PROCBASED_CR8_STORE_EXITING) #define PROCBASED_CTLS_ZERO_SETTING \ (PROCBASED_CR3_LOAD_EXITING | \ PROCBASED_CR3_STORE_EXITING | \ PROCBASED_IO_BITMAPS) #define PROCBASED_CTLS2_ONE_SETTING PROCBASED2_ENABLE_EPT #define PROCBASED_CTLS2_ZERO_SETTING 0 #define VM_EXIT_CTLS_ONE_SETTING \ (VM_EXIT_SAVE_DEBUG_CONTROLS | \ VM_EXIT_HOST_LMA | \ VM_EXIT_SAVE_EFER | \ VM_EXIT_LOAD_EFER | \ VM_EXIT_ACKNOWLEDGE_INTERRUPT) #define VM_EXIT_CTLS_ZERO_SETTING 0 #define VM_ENTRY_CTLS_ONE_SETTING \ (VM_ENTRY_LOAD_DEBUG_CONTROLS | \ VM_ENTRY_LOAD_EFER) #define VM_ENTRY_CTLS_ZERO_SETTING \ (VM_ENTRY_INTO_SMM | \ VM_ENTRY_DEACTIVATE_DUAL_MONITOR) #define HANDLED 1 #define UNHANDLED 0 static MALLOC_DEFINE(M_VMX, "vmx", "vmx"); static MALLOC_DEFINE(M_VLAPIC, "vlapic", "vlapic"); SYSCTL_DECL(_hw_vmm); SYSCTL_NODE(_hw_vmm, OID_AUTO, vmx, CTLFLAG_RW | CTLFLAG_MPSAFE, NULL, NULL); int vmxon_enabled[MAXCPU]; static char vmxon_region[MAXCPU][PAGE_SIZE] __aligned(PAGE_SIZE); static uint32_t pinbased_ctls, procbased_ctls, procbased_ctls2; static uint32_t exit_ctls, entry_ctls; static uint64_t cr0_ones_mask, cr0_zeros_mask; SYSCTL_ULONG(_hw_vmm_vmx, OID_AUTO, cr0_ones_mask, CTLFLAG_RD, &cr0_ones_mask, 0, NULL); SYSCTL_ULONG(_hw_vmm_vmx, OID_AUTO, cr0_zeros_mask, CTLFLAG_RD, &cr0_zeros_mask, 0, NULL); static uint64_t cr4_ones_mask, cr4_zeros_mask; SYSCTL_ULONG(_hw_vmm_vmx, OID_AUTO, cr4_ones_mask, CTLFLAG_RD, &cr4_ones_mask, 0, NULL); SYSCTL_ULONG(_hw_vmm_vmx, OID_AUTO, cr4_zeros_mask, CTLFLAG_RD, &cr4_zeros_mask, 0, NULL); static int vmx_initialized; SYSCTL_INT(_hw_vmm_vmx, OID_AUTO, initialized, CTLFLAG_RD, &vmx_initialized, 0, "Intel VMX initialized"); /* * Optional capabilities */ static SYSCTL_NODE(_hw_vmm_vmx, OID_AUTO, cap, CTLFLAG_RW | CTLFLAG_MPSAFE, NULL, NULL); static int cap_halt_exit; SYSCTL_INT(_hw_vmm_vmx_cap, OID_AUTO, halt_exit, CTLFLAG_RD, &cap_halt_exit, 0, "HLT triggers a VM-exit"); static int cap_pause_exit; SYSCTL_INT(_hw_vmm_vmx_cap, OID_AUTO, pause_exit, CTLFLAG_RD, &cap_pause_exit, 0, "PAUSE triggers a VM-exit"); static int cap_unrestricted_guest; SYSCTL_INT(_hw_vmm_vmx_cap, OID_AUTO, unrestricted_guest, CTLFLAG_RD, &cap_unrestricted_guest, 0, "Unrestricted guests"); static int cap_monitor_trap; SYSCTL_INT(_hw_vmm_vmx_cap, OID_AUTO, monitor_trap, CTLFLAG_RD, &cap_monitor_trap, 0, "Monitor trap flag"); static int cap_invpcid; SYSCTL_INT(_hw_vmm_vmx_cap, OID_AUTO, invpcid, CTLFLAG_RD, &cap_invpcid, 0, "Guests are allowed to use INVPCID"); +static int tpr_shadowing; +SYSCTL_INT(_hw_vmm_vmx_cap, OID_AUTO, tpr_shadowing, CTLFLAG_RD, + &tpr_shadowing, 0, "TPR shadowing support"); + static int virtual_interrupt_delivery; SYSCTL_INT(_hw_vmm_vmx_cap, OID_AUTO, virtual_interrupt_delivery, CTLFLAG_RD, &virtual_interrupt_delivery, 0, "APICv virtual interrupt delivery support"); static int posted_interrupts; SYSCTL_INT(_hw_vmm_vmx_cap, OID_AUTO, posted_interrupts, CTLFLAG_RD, &posted_interrupts, 0, "APICv posted interrupt support"); static int pirvec = -1; SYSCTL_INT(_hw_vmm_vmx, OID_AUTO, posted_interrupt_vector, CTLFLAG_RD, &pirvec, 0, "APICv posted interrupt vector"); static struct unrhdr *vpid_unr; static u_int vpid_alloc_failed; SYSCTL_UINT(_hw_vmm_vmx, OID_AUTO, vpid_alloc_failed, CTLFLAG_RD, &vpid_alloc_failed, 0, NULL); int guest_l1d_flush; SYSCTL_INT(_hw_vmm_vmx, OID_AUTO, l1d_flush, CTLFLAG_RD, &guest_l1d_flush, 0, NULL); int guest_l1d_flush_sw; SYSCTL_INT(_hw_vmm_vmx, OID_AUTO, l1d_flush_sw, CTLFLAG_RD, &guest_l1d_flush_sw, 0, NULL); static struct msr_entry msr_load_list[1] __aligned(16); /* * The definitions of SDT probes for VMX. */ SDT_PROBE_DEFINE3(vmm, vmx, exit, entry, "struct vmx *", "int", "struct vm_exit *"); SDT_PROBE_DEFINE4(vmm, vmx, exit, taskswitch, "struct vmx *", "int", "struct vm_exit *", "struct vm_task_switch *"); SDT_PROBE_DEFINE4(vmm, vmx, exit, craccess, "struct vmx *", "int", "struct vm_exit *", "uint64_t"); SDT_PROBE_DEFINE4(vmm, vmx, exit, rdmsr, "struct vmx *", "int", "struct vm_exit *", "uint32_t"); SDT_PROBE_DEFINE5(vmm, vmx, exit, wrmsr, "struct vmx *", "int", "struct vm_exit *", "uint32_t", "uint64_t"); SDT_PROBE_DEFINE3(vmm, vmx, exit, halt, "struct vmx *", "int", "struct vm_exit *"); SDT_PROBE_DEFINE3(vmm, vmx, exit, mtrap, "struct vmx *", "int", "struct vm_exit *"); SDT_PROBE_DEFINE3(vmm, vmx, exit, pause, "struct vmx *", "int", "struct vm_exit *"); SDT_PROBE_DEFINE3(vmm, vmx, exit, intrwindow, "struct vmx *", "int", "struct vm_exit *"); SDT_PROBE_DEFINE4(vmm, vmx, exit, interrupt, "struct vmx *", "int", "struct vm_exit *", "uint32_t"); SDT_PROBE_DEFINE3(vmm, vmx, exit, nmiwindow, "struct vmx *", "int", "struct vm_exit *"); SDT_PROBE_DEFINE3(vmm, vmx, exit, inout, "struct vmx *", "int", "struct vm_exit *"); SDT_PROBE_DEFINE3(vmm, vmx, exit, cpuid, "struct vmx *", "int", "struct vm_exit *"); SDT_PROBE_DEFINE5(vmm, vmx, exit, exception, "struct vmx *", "int", "struct vm_exit *", "uint32_t", "int"); SDT_PROBE_DEFINE5(vmm, vmx, exit, nestedfault, "struct vmx *", "int", "struct vm_exit *", "uint64_t", "uint64_t"); SDT_PROBE_DEFINE4(vmm, vmx, exit, mmiofault, "struct vmx *", "int", "struct vm_exit *", "uint64_t"); SDT_PROBE_DEFINE3(vmm, vmx, exit, eoi, "struct vmx *", "int", "struct vm_exit *"); SDT_PROBE_DEFINE3(vmm, vmx, exit, apicaccess, "struct vmx *", "int", "struct vm_exit *"); SDT_PROBE_DEFINE4(vmm, vmx, exit, apicwrite, "struct vmx *", "int", "struct vm_exit *", "struct vlapic *"); SDT_PROBE_DEFINE3(vmm, vmx, exit, xsetbv, "struct vmx *", "int", "struct vm_exit *"); SDT_PROBE_DEFINE3(vmm, vmx, exit, monitor, "struct vmx *", "int", "struct vm_exit *"); SDT_PROBE_DEFINE3(vmm, vmx, exit, mwait, "struct vmx *", "int", "struct vm_exit *"); SDT_PROBE_DEFINE3(vmm, vmx, exit, vminsn, "struct vmx *", "int", "struct vm_exit *"); SDT_PROBE_DEFINE4(vmm, vmx, exit, unknown, "struct vmx *", "int", "struct vm_exit *", "uint32_t"); SDT_PROBE_DEFINE4(vmm, vmx, exit, return, "struct vmx *", "int", "struct vm_exit *", "int"); /* * Use the last page below 4GB as the APIC access address. This address is * occupied by the boot firmware so it is guaranteed that it will not conflict * with a page in system memory. */ #define APIC_ACCESS_ADDRESS 0xFFFFF000 static int vmx_getdesc(void *arg, int vcpu, int reg, struct seg_desc *desc); static int vmx_getreg(void *arg, int vcpu, int reg, uint64_t *retval); static int vmxctx_setreg(struct vmxctx *vmxctx, int reg, uint64_t val); static void vmx_inject_pir(struct vlapic *vlapic); #ifdef KTR static const char * exit_reason_to_str(int reason) { static char reasonbuf[32]; switch (reason) { case EXIT_REASON_EXCEPTION: return "exception"; case EXIT_REASON_EXT_INTR: return "extint"; case EXIT_REASON_TRIPLE_FAULT: return "triplefault"; case EXIT_REASON_INIT: return "init"; case EXIT_REASON_SIPI: return "sipi"; case EXIT_REASON_IO_SMI: return "iosmi"; case EXIT_REASON_SMI: return "smi"; case EXIT_REASON_INTR_WINDOW: return "intrwindow"; case EXIT_REASON_NMI_WINDOW: return "nmiwindow"; case EXIT_REASON_TASK_SWITCH: return "taskswitch"; case EXIT_REASON_CPUID: return "cpuid"; case EXIT_REASON_GETSEC: return "getsec"; case EXIT_REASON_HLT: return "hlt"; case EXIT_REASON_INVD: return "invd"; case EXIT_REASON_INVLPG: return "invlpg"; case EXIT_REASON_RDPMC: return "rdpmc"; case EXIT_REASON_RDTSC: return "rdtsc"; case EXIT_REASON_RSM: return "rsm"; case EXIT_REASON_VMCALL: return "vmcall"; case EXIT_REASON_VMCLEAR: return "vmclear"; case EXIT_REASON_VMLAUNCH: return "vmlaunch"; case EXIT_REASON_VMPTRLD: return "vmptrld"; case EXIT_REASON_VMPTRST: return "vmptrst"; case EXIT_REASON_VMREAD: return "vmread"; case EXIT_REASON_VMRESUME: return "vmresume"; case EXIT_REASON_VMWRITE: return "vmwrite"; case EXIT_REASON_VMXOFF: return "vmxoff"; case EXIT_REASON_VMXON: return "vmxon"; case EXIT_REASON_CR_ACCESS: return "craccess"; case EXIT_REASON_DR_ACCESS: return "draccess"; case EXIT_REASON_INOUT: return "inout"; case EXIT_REASON_RDMSR: return "rdmsr"; case EXIT_REASON_WRMSR: return "wrmsr"; case EXIT_REASON_INVAL_VMCS: return "invalvmcs"; case EXIT_REASON_INVAL_MSR: return "invalmsr"; case EXIT_REASON_MWAIT: return "mwait"; case EXIT_REASON_MTF: return "mtf"; case EXIT_REASON_MONITOR: return "monitor"; case EXIT_REASON_PAUSE: return "pause"; case EXIT_REASON_MCE_DURING_ENTRY: return "mce-during-entry"; case EXIT_REASON_TPR: return "tpr"; case EXIT_REASON_APIC_ACCESS: return "apic-access"; case EXIT_REASON_GDTR_IDTR: return "gdtridtr"; case EXIT_REASON_LDTR_TR: return "ldtrtr"; case EXIT_REASON_EPT_FAULT: return "eptfault"; case EXIT_REASON_EPT_MISCONFIG: return "eptmisconfig"; case EXIT_REASON_INVEPT: return "invept"; case EXIT_REASON_RDTSCP: return "rdtscp"; case EXIT_REASON_VMX_PREEMPT: return "vmxpreempt"; case EXIT_REASON_INVVPID: return "invvpid"; case EXIT_REASON_WBINVD: return "wbinvd"; case EXIT_REASON_XSETBV: return "xsetbv"; case EXIT_REASON_APIC_WRITE: return "apic-write"; default: snprintf(reasonbuf, sizeof(reasonbuf), "%d", reason); return (reasonbuf); } } #endif /* KTR */ static int vmx_allow_x2apic_msrs(struct vmx *vmx) { int i, error; error = 0; /* * Allow readonly access to the following x2APIC MSRs from the guest. */ error += guest_msr_ro(vmx, MSR_APIC_ID); error += guest_msr_ro(vmx, MSR_APIC_VERSION); error += guest_msr_ro(vmx, MSR_APIC_LDR); error += guest_msr_ro(vmx, MSR_APIC_SVR); for (i = 0; i < 8; i++) error += guest_msr_ro(vmx, MSR_APIC_ISR0 + i); for (i = 0; i < 8; i++) error += guest_msr_ro(vmx, MSR_APIC_TMR0 + i); for (i = 0; i < 8; i++) error += guest_msr_ro(vmx, MSR_APIC_IRR0 + i); error += guest_msr_ro(vmx, MSR_APIC_ESR); error += guest_msr_ro(vmx, MSR_APIC_LVT_TIMER); error += guest_msr_ro(vmx, MSR_APIC_LVT_THERMAL); error += guest_msr_ro(vmx, MSR_APIC_LVT_PCINT); error += guest_msr_ro(vmx, MSR_APIC_LVT_LINT0); error += guest_msr_ro(vmx, MSR_APIC_LVT_LINT1); error += guest_msr_ro(vmx, MSR_APIC_LVT_ERROR); error += guest_msr_ro(vmx, MSR_APIC_ICR_TIMER); error += guest_msr_ro(vmx, MSR_APIC_DCR_TIMER); error += guest_msr_ro(vmx, MSR_APIC_ICR); /* * Allow TPR, EOI and SELF_IPI MSRs to be read and written by the guest. * * These registers get special treatment described in the section * "Virtualizing MSR-Based APIC Accesses". */ error += guest_msr_rw(vmx, MSR_APIC_TPR); error += guest_msr_rw(vmx, MSR_APIC_EOI); error += guest_msr_rw(vmx, MSR_APIC_SELF_IPI); return (error); } u_long vmx_fix_cr0(u_long cr0) { return ((cr0 | cr0_ones_mask) & ~cr0_zeros_mask); } u_long vmx_fix_cr4(u_long cr4) { return ((cr4 | cr4_ones_mask) & ~cr4_zeros_mask); } static void vpid_free(int vpid) { if (vpid < 0 || vpid > 0xffff) panic("vpid_free: invalid vpid %d", vpid); /* * VPIDs [0,VM_MAXCPU] are special and are not allocated from * the unit number allocator. */ if (vpid > VM_MAXCPU) free_unr(vpid_unr, vpid); } static void vpid_alloc(uint16_t *vpid, int num) { int i, x; if (num <= 0 || num > VM_MAXCPU) panic("invalid number of vpids requested: %d", num); /* * If the "enable vpid" execution control is not enabled then the * VPID is required to be 0 for all vcpus. */ if ((procbased_ctls2 & PROCBASED2_ENABLE_VPID) == 0) { for (i = 0; i < num; i++) vpid[i] = 0; return; } /* * Allocate a unique VPID for each vcpu from the unit number allocator. */ for (i = 0; i < num; i++) { x = alloc_unr(vpid_unr); if (x == -1) break; else vpid[i] = x; } if (i < num) { atomic_add_int(&vpid_alloc_failed, 1); /* * If the unit number allocator does not have enough unique * VPIDs then we need to allocate from the [1,VM_MAXCPU] range. * * These VPIDs are not be unique across VMs but this does not * affect correctness because the combined mappings are also * tagged with the EP4TA which is unique for each VM. * * It is still sub-optimal because the invvpid will invalidate * combined mappings for a particular VPID across all EP4TAs. */ while (i-- > 0) vpid_free(vpid[i]); for (i = 0; i < num; i++) vpid[i] = i + 1; } } static void vpid_init(void) { /* * VPID 0 is required when the "enable VPID" execution control is * disabled. * * VPIDs [1,VM_MAXCPU] are used as the "overflow namespace" when the * unit number allocator does not have sufficient unique VPIDs to * satisfy the allocation. * * The remaining VPIDs are managed by the unit number allocator. */ vpid_unr = new_unrhdr(VM_MAXCPU + 1, 0xffff, NULL); } static void vmx_disable(void *arg __unused) { struct invvpid_desc invvpid_desc = { 0 }; struct invept_desc invept_desc = { 0 }; if (vmxon_enabled[curcpu]) { /* * See sections 25.3.3.3 and 25.3.3.4 in Intel Vol 3b. * * VMXON or VMXOFF are not required to invalidate any TLB * caching structures. This prevents potential retention of * cached information in the TLB between distinct VMX episodes. */ invvpid(INVVPID_TYPE_ALL_CONTEXTS, invvpid_desc); invept(INVEPT_TYPE_ALL_CONTEXTS, invept_desc); vmxoff(); } load_cr4(rcr4() & ~CR4_VMXE); } static int vmx_cleanup(void) { if (pirvec >= 0) lapic_ipi_free(pirvec); if (vpid_unr != NULL) { delete_unrhdr(vpid_unr); vpid_unr = NULL; } if (nmi_flush_l1d_sw == 1) nmi_flush_l1d_sw = 0; smp_rendezvous(NULL, vmx_disable, NULL, NULL); return (0); } static void vmx_enable(void *arg __unused) { int error; uint64_t feature_control; feature_control = rdmsr(MSR_IA32_FEATURE_CONTROL); if ((feature_control & IA32_FEATURE_CONTROL_LOCK) == 0 || (feature_control & IA32_FEATURE_CONTROL_VMX_EN) == 0) { wrmsr(MSR_IA32_FEATURE_CONTROL, feature_control | IA32_FEATURE_CONTROL_VMX_EN | IA32_FEATURE_CONTROL_LOCK); } load_cr4(rcr4() | CR4_VMXE); *(uint32_t *)vmxon_region[curcpu] = vmx_revision(); error = vmxon(vmxon_region[curcpu]); if (error == 0) vmxon_enabled[curcpu] = 1; } static void vmx_restore(void) { if (vmxon_enabled[curcpu]) vmxon(vmxon_region[curcpu]); } static int vmx_init(int ipinum) { - int error, use_tpr_shadow; + int error; uint64_t basic, fixed0, fixed1, feature_control; uint32_t tmp, procbased2_vid_bits; /* CPUID.1:ECX[bit 5] must be 1 for processor to support VMX */ if (!(cpu_feature2 & CPUID2_VMX)) { printf("vmx_init: processor does not support VMX operation\n"); return (ENXIO); } /* * Verify that MSR_IA32_FEATURE_CONTROL lock and VMXON enable bits * are set (bits 0 and 2 respectively). */ feature_control = rdmsr(MSR_IA32_FEATURE_CONTROL); if ((feature_control & IA32_FEATURE_CONTROL_LOCK) == 1 && (feature_control & IA32_FEATURE_CONTROL_VMX_EN) == 0) { printf("vmx_init: VMX operation disabled by BIOS\n"); return (ENXIO); } /* * Verify capabilities MSR_VMX_BASIC: * - bit 54 indicates support for INS/OUTS decoding */ basic = rdmsr(MSR_VMX_BASIC); if ((basic & (1UL << 54)) == 0) { printf("vmx_init: processor does not support desired basic " "capabilities\n"); return (EINVAL); } /* Check support for primary processor-based VM-execution controls */ error = vmx_set_ctlreg(MSR_VMX_PROCBASED_CTLS, MSR_VMX_TRUE_PROCBASED_CTLS, PROCBASED_CTLS_ONE_SETTING, PROCBASED_CTLS_ZERO_SETTING, &procbased_ctls); if (error) { printf("vmx_init: processor does not support desired primary " "processor-based controls\n"); return (error); } /* Clear the processor-based ctl bits that are set on demand */ procbased_ctls &= ~PROCBASED_CTLS_WINDOW_SETTING; /* Check support for secondary processor-based VM-execution controls */ error = vmx_set_ctlreg(MSR_VMX_PROCBASED_CTLS2, MSR_VMX_PROCBASED_CTLS2, PROCBASED_CTLS2_ONE_SETTING, PROCBASED_CTLS2_ZERO_SETTING, &procbased_ctls2); if (error) { printf("vmx_init: processor does not support desired secondary " "processor-based controls\n"); return (error); } /* Check support for VPID */ error = vmx_set_ctlreg(MSR_VMX_PROCBASED_CTLS2, MSR_VMX_PROCBASED_CTLS2, PROCBASED2_ENABLE_VPID, 0, &tmp); if (error == 0) procbased_ctls2 |= PROCBASED2_ENABLE_VPID; /* Check support for pin-based VM-execution controls */ error = vmx_set_ctlreg(MSR_VMX_PINBASED_CTLS, MSR_VMX_TRUE_PINBASED_CTLS, PINBASED_CTLS_ONE_SETTING, PINBASED_CTLS_ZERO_SETTING, &pinbased_ctls); if (error) { printf("vmx_init: processor does not support desired " "pin-based controls\n"); return (error); } /* Check support for VM-exit controls */ error = vmx_set_ctlreg(MSR_VMX_EXIT_CTLS, MSR_VMX_TRUE_EXIT_CTLS, VM_EXIT_CTLS_ONE_SETTING, VM_EXIT_CTLS_ZERO_SETTING, &exit_ctls); if (error) { printf("vmx_init: processor does not support desired " "exit controls\n"); return (error); } /* Check support for VM-entry controls */ error = vmx_set_ctlreg(MSR_VMX_ENTRY_CTLS, MSR_VMX_TRUE_ENTRY_CTLS, VM_ENTRY_CTLS_ONE_SETTING, VM_ENTRY_CTLS_ZERO_SETTING, &entry_ctls); if (error) { printf("vmx_init: processor does not support desired " "entry controls\n"); return (error); } /* * Check support for optional features by testing them * as individual bits */ cap_halt_exit = (vmx_set_ctlreg(MSR_VMX_PROCBASED_CTLS, MSR_VMX_TRUE_PROCBASED_CTLS, PROCBASED_HLT_EXITING, 0, &tmp) == 0); cap_monitor_trap = (vmx_set_ctlreg(MSR_VMX_PROCBASED_CTLS, MSR_VMX_PROCBASED_CTLS, PROCBASED_MTF, 0, &tmp) == 0); cap_pause_exit = (vmx_set_ctlreg(MSR_VMX_PROCBASED_CTLS, MSR_VMX_TRUE_PROCBASED_CTLS, PROCBASED_PAUSE_EXITING, 0, &tmp) == 0); cap_unrestricted_guest = (vmx_set_ctlreg(MSR_VMX_PROCBASED_CTLS2, MSR_VMX_PROCBASED_CTLS2, PROCBASED2_UNRESTRICTED_GUEST, 0, &tmp) == 0); cap_invpcid = (vmx_set_ctlreg(MSR_VMX_PROCBASED_CTLS2, MSR_VMX_PROCBASED_CTLS2, PROCBASED2_ENABLE_INVPCID, 0, &tmp) == 0); /* + * Check support for TPR shadow. + */ + error = vmx_set_ctlreg(MSR_VMX_PROCBASED_CTLS, + MSR_VMX_TRUE_PROCBASED_CTLS, PROCBASED_USE_TPR_SHADOW, 0, + &tmp); + if (error == 0) { + tpr_shadowing = 1; + TUNABLE_INT_FETCH("hw.vmm.vmx.use_tpr_shadowing", + &tpr_shadowing); + } + + if (tpr_shadowing) { + procbased_ctls |= PROCBASED_USE_TPR_SHADOW; + procbased_ctls &= ~PROCBASED_CR8_LOAD_EXITING; + procbased_ctls &= ~PROCBASED_CR8_STORE_EXITING; + } + + /* * Check support for virtual interrupt delivery. */ procbased2_vid_bits = (PROCBASED2_VIRTUALIZE_APIC_ACCESSES | PROCBASED2_VIRTUALIZE_X2APIC_MODE | PROCBASED2_APIC_REGISTER_VIRTUALIZATION | PROCBASED2_VIRTUAL_INTERRUPT_DELIVERY); - use_tpr_shadow = (vmx_set_ctlreg(MSR_VMX_PROCBASED_CTLS, - MSR_VMX_TRUE_PROCBASED_CTLS, PROCBASED_USE_TPR_SHADOW, 0, - &tmp) == 0); - error = vmx_set_ctlreg(MSR_VMX_PROCBASED_CTLS2, MSR_VMX_PROCBASED_CTLS2, procbased2_vid_bits, 0, &tmp); - if (error == 0 && use_tpr_shadow) { + if (error == 0 && tpr_shadowing) { virtual_interrupt_delivery = 1; TUNABLE_INT_FETCH("hw.vmm.vmx.use_apic_vid", &virtual_interrupt_delivery); } if (virtual_interrupt_delivery) { procbased_ctls |= PROCBASED_USE_TPR_SHADOW; procbased_ctls2 |= procbased2_vid_bits; procbased_ctls2 &= ~PROCBASED2_VIRTUALIZE_X2APIC_MODE; /* - * No need to emulate accesses to %CR8 if virtual - * interrupt delivery is enabled. - */ - procbased_ctls &= ~PROCBASED_CR8_LOAD_EXITING; - procbased_ctls &= ~PROCBASED_CR8_STORE_EXITING; - - /* * Check for Posted Interrupts only if Virtual Interrupt * Delivery is enabled. */ error = vmx_set_ctlreg(MSR_VMX_PINBASED_CTLS, MSR_VMX_TRUE_PINBASED_CTLS, PINBASED_POSTED_INTERRUPT, 0, &tmp); if (error == 0) { pirvec = lapic_ipi_alloc(pti ? &IDTVEC(justreturn1_pti) : &IDTVEC(justreturn)); if (pirvec < 0) { if (bootverbose) { printf("vmx_init: unable to allocate " "posted interrupt vector\n"); } } else { posted_interrupts = 1; TUNABLE_INT_FETCH("hw.vmm.vmx.use_apic_pir", &posted_interrupts); } } } if (posted_interrupts) pinbased_ctls |= PINBASED_POSTED_INTERRUPT; /* Initialize EPT */ error = ept_init(ipinum); if (error) { printf("vmx_init: ept initialization failed (%d)\n", error); return (error); } guest_l1d_flush = (cpu_ia32_arch_caps & IA32_ARCH_CAP_SKIP_L1DFL_VMENTRY) == 0; TUNABLE_INT_FETCH("hw.vmm.l1d_flush", &guest_l1d_flush); /* * L1D cache flush is enabled. Use IA32_FLUSH_CMD MSR when * available. Otherwise fall back to the software flush * method which loads enough data from the kernel text to * flush existing L1D content, both on VMX entry and on NMI * return. */ if (guest_l1d_flush) { if ((cpu_stdext_feature3 & CPUID_STDEXT3_L1D_FLUSH) == 0) { guest_l1d_flush_sw = 1; TUNABLE_INT_FETCH("hw.vmm.l1d_flush_sw", &guest_l1d_flush_sw); } if (guest_l1d_flush_sw) { if (nmi_flush_l1d_sw <= 1) nmi_flush_l1d_sw = 1; } else { msr_load_list[0].index = MSR_IA32_FLUSH_CMD; msr_load_list[0].val = IA32_FLUSH_CMD_L1D; } } /* * Stash the cr0 and cr4 bits that must be fixed to 0 or 1 */ fixed0 = rdmsr(MSR_VMX_CR0_FIXED0); fixed1 = rdmsr(MSR_VMX_CR0_FIXED1); cr0_ones_mask = fixed0 & fixed1; cr0_zeros_mask = ~fixed0 & ~fixed1; /* * CR0_PE and CR0_PG can be set to zero in VMX non-root operation * if unrestricted guest execution is allowed. */ if (cap_unrestricted_guest) cr0_ones_mask &= ~(CR0_PG | CR0_PE); /* * Do not allow the guest to set CR0_NW or CR0_CD. */ cr0_zeros_mask |= (CR0_NW | CR0_CD); fixed0 = rdmsr(MSR_VMX_CR4_FIXED0); fixed1 = rdmsr(MSR_VMX_CR4_FIXED1); cr4_ones_mask = fixed0 & fixed1; cr4_zeros_mask = ~fixed0 & ~fixed1; vpid_init(); vmx_msr_init(); /* enable VMX operation */ smp_rendezvous(NULL, vmx_enable, NULL, NULL); vmx_initialized = 1; return (0); } static void vmx_trigger_hostintr(int vector) { uintptr_t func; struct gate_descriptor *gd; gd = &idt[vector]; KASSERT(vector >= 32 && vector <= 255, ("vmx_trigger_hostintr: " "invalid vector %d", vector)); KASSERT(gd->gd_p == 1, ("gate descriptor for vector %d not present", vector)); KASSERT(gd->gd_type == SDT_SYSIGT, ("gate descriptor for vector %d " "has invalid type %d", vector, gd->gd_type)); KASSERT(gd->gd_dpl == SEL_KPL, ("gate descriptor for vector %d " "has invalid dpl %d", vector, gd->gd_dpl)); KASSERT(gd->gd_selector == GSEL(GCODE_SEL, SEL_KPL), ("gate descriptor " "for vector %d has invalid selector %d", vector, gd->gd_selector)); KASSERT(gd->gd_ist == 0, ("gate descriptor for vector %d has invalid " "IST %d", vector, gd->gd_ist)); func = ((long)gd->gd_hioffset << 16 | gd->gd_looffset); vmx_call_isr(func); } static int vmx_setup_cr_shadow(int which, struct vmcs *vmcs, uint32_t initial) { int error, mask_ident, shadow_ident; uint64_t mask_value; if (which != 0 && which != 4) panic("vmx_setup_cr_shadow: unknown cr%d", which); if (which == 0) { mask_ident = VMCS_CR0_MASK; mask_value = cr0_ones_mask | cr0_zeros_mask; shadow_ident = VMCS_CR0_SHADOW; } else { mask_ident = VMCS_CR4_MASK; mask_value = cr4_ones_mask | cr4_zeros_mask; shadow_ident = VMCS_CR4_SHADOW; } error = vmcs_setreg(vmcs, 0, VMCS_IDENT(mask_ident), mask_value); if (error) return (error); error = vmcs_setreg(vmcs, 0, VMCS_IDENT(shadow_ident), initial); if (error) return (error); return (0); } #define vmx_setup_cr0_shadow(vmcs,init) vmx_setup_cr_shadow(0, (vmcs), (init)) #define vmx_setup_cr4_shadow(vmcs,init) vmx_setup_cr_shadow(4, (vmcs), (init)) static void * vmx_vminit(struct vm *vm, pmap_t pmap) { uint16_t vpid[VM_MAXCPU]; int i, error; struct vmx *vmx; struct vmcs *vmcs; uint32_t exc_bitmap; uint16_t maxcpus; vmx = malloc(sizeof(struct vmx), M_VMX, M_WAITOK | M_ZERO); if ((uintptr_t)vmx & PAGE_MASK) { panic("malloc of struct vmx not aligned on %d byte boundary", PAGE_SIZE); } vmx->vm = vm; vmx->eptp = eptp(vtophys((vm_offset_t)pmap->pm_pml4)); /* * Clean up EPTP-tagged guest physical and combined mappings * * VMX transitions are not required to invalidate any guest physical * mappings. So, it may be possible for stale guest physical mappings * to be present in the processor TLBs. * * Combined mappings for this EP4TA are also invalidated for all VPIDs. */ ept_invalidate_mappings(vmx->eptp); msr_bitmap_initialize(vmx->msr_bitmap); /* * It is safe to allow direct access to MSR_GSBASE and MSR_FSBASE. * The guest FSBASE and GSBASE are saved and restored during * vm-exit and vm-entry respectively. The host FSBASE and GSBASE are * always restored from the vmcs host state area on vm-exit. * * The SYSENTER_CS/ESP/EIP MSRs are identical to FS/GSBASE in * how they are saved/restored so can be directly accessed by the * guest. * * MSR_EFER is saved and restored in the guest VMCS area on a * VM exit and entry respectively. It is also restored from the * host VMCS area on a VM exit. * * The TSC MSR is exposed read-only. Writes are disallowed as * that will impact the host TSC. If the guest does a write * the "use TSC offsetting" execution control is enabled and the * difference between the host TSC and the guest TSC is written * into the TSC offset in the VMCS. */ if (guest_msr_rw(vmx, MSR_GSBASE) || guest_msr_rw(vmx, MSR_FSBASE) || guest_msr_rw(vmx, MSR_SYSENTER_CS_MSR) || guest_msr_rw(vmx, MSR_SYSENTER_ESP_MSR) || guest_msr_rw(vmx, MSR_SYSENTER_EIP_MSR) || guest_msr_rw(vmx, MSR_EFER) || guest_msr_ro(vmx, MSR_TSC)) panic("vmx_vminit: error setting guest msr access"); vpid_alloc(vpid, VM_MAXCPU); if (virtual_interrupt_delivery) { error = vm_map_mmio(vm, DEFAULT_APIC_BASE, PAGE_SIZE, APIC_ACCESS_ADDRESS); /* XXX this should really return an error to the caller */ KASSERT(error == 0, ("vm_map_mmio(apicbase) error %d", error)); } maxcpus = vm_get_maxcpus(vm); for (i = 0; i < maxcpus; i++) { vmcs = &vmx->vmcs[i]; vmcs->identifier = vmx_revision(); error = vmclear(vmcs); if (error != 0) { panic("vmx_vminit: vmclear error %d on vcpu %d\n", error, i); } vmx_msr_guest_init(vmx, i); error = vmcs_init(vmcs); KASSERT(error == 0, ("vmcs_init error %d", error)); VMPTRLD(vmcs); error = 0; error += vmwrite(VMCS_HOST_RSP, (u_long)&vmx->ctx[i]); error += vmwrite(VMCS_EPTP, vmx->eptp); error += vmwrite(VMCS_PIN_BASED_CTLS, pinbased_ctls); error += vmwrite(VMCS_PRI_PROC_BASED_CTLS, procbased_ctls); error += vmwrite(VMCS_SEC_PROC_BASED_CTLS, procbased_ctls2); error += vmwrite(VMCS_EXIT_CTLS, exit_ctls); error += vmwrite(VMCS_ENTRY_CTLS, entry_ctls); error += vmwrite(VMCS_MSR_BITMAP, vtophys(vmx->msr_bitmap)); error += vmwrite(VMCS_VPID, vpid[i]); if (guest_l1d_flush && !guest_l1d_flush_sw) { vmcs_write(VMCS_ENTRY_MSR_LOAD, pmap_kextract( (vm_offset_t)&msr_load_list[0])); vmcs_write(VMCS_ENTRY_MSR_LOAD_COUNT, nitems(msr_load_list)); vmcs_write(VMCS_EXIT_MSR_STORE, 0); vmcs_write(VMCS_EXIT_MSR_STORE_COUNT, 0); } /* exception bitmap */ if (vcpu_trace_exceptions(vm, i)) exc_bitmap = 0xffffffff; else exc_bitmap = 1 << IDT_MC; error += vmwrite(VMCS_EXCEPTION_BITMAP, exc_bitmap); vmx->ctx[i].guest_dr6 = DBREG_DR6_RESERVED1; error += vmwrite(VMCS_GUEST_DR7, DBREG_DR7_RESERVED1); - if (virtual_interrupt_delivery) { - error += vmwrite(VMCS_APIC_ACCESS, APIC_ACCESS_ADDRESS); + if (tpr_shadowing) { error += vmwrite(VMCS_VIRTUAL_APIC, vtophys(&vmx->apic_page[i])); + } + + if (virtual_interrupt_delivery) { + error += vmwrite(VMCS_APIC_ACCESS, APIC_ACCESS_ADDRESS); error += vmwrite(VMCS_EOI_EXIT0, 0); error += vmwrite(VMCS_EOI_EXIT1, 0); error += vmwrite(VMCS_EOI_EXIT2, 0); error += vmwrite(VMCS_EOI_EXIT3, 0); } if (posted_interrupts) { error += vmwrite(VMCS_PIR_VECTOR, pirvec); error += vmwrite(VMCS_PIR_DESC, vtophys(&vmx->pir_desc[i])); } VMCLEAR(vmcs); KASSERT(error == 0, ("vmx_vminit: error customizing the vmcs")); vmx->cap[i].set = 0; vmx->cap[i].proc_ctls = procbased_ctls; vmx->cap[i].proc_ctls2 = procbased_ctls2; vmx->cap[i].exc_bitmap = exc_bitmap; vmx->state[i].nextrip = ~0; vmx->state[i].lastcpu = NOCPU; vmx->state[i].vpid = vpid[i]; /* * Set up the CR0/4 shadows, and init the read shadow * to the power-on register value from the Intel Sys Arch. * CR0 - 0x60000010 * CR4 - 0 */ error = vmx_setup_cr0_shadow(vmcs, 0x60000010); if (error != 0) panic("vmx_setup_cr0_shadow %d", error); error = vmx_setup_cr4_shadow(vmcs, 0); if (error != 0) panic("vmx_setup_cr4_shadow %d", error); vmx->ctx[i].pmap = pmap; } return (vmx); } static int vmx_handle_cpuid(struct vm *vm, int vcpu, struct vmxctx *vmxctx) { int handled, func; func = vmxctx->guest_rax; handled = x86_emulate_cpuid(vm, vcpu, (uint32_t*)(&vmxctx->guest_rax), (uint32_t*)(&vmxctx->guest_rbx), (uint32_t*)(&vmxctx->guest_rcx), (uint32_t*)(&vmxctx->guest_rdx)); return (handled); } static __inline void vmx_run_trace(struct vmx *vmx, int vcpu) { #ifdef KTR VCPU_CTR1(vmx->vm, vcpu, "Resume execution at %#lx", vmcs_guest_rip()); #endif } static __inline void vmx_exit_trace(struct vmx *vmx, int vcpu, uint64_t rip, uint32_t exit_reason, int handled) { #ifdef KTR VCPU_CTR3(vmx->vm, vcpu, "%s %s vmexit at 0x%0lx", handled ? "handled" : "unhandled", exit_reason_to_str(exit_reason), rip); #endif } static __inline void vmx_astpending_trace(struct vmx *vmx, int vcpu, uint64_t rip) { #ifdef KTR VCPU_CTR1(vmx->vm, vcpu, "astpending vmexit at 0x%0lx", rip); #endif } static VMM_STAT_INTEL(VCPU_INVVPID_SAVED, "Number of vpid invalidations saved"); static VMM_STAT_INTEL(VCPU_INVVPID_DONE, "Number of vpid invalidations done"); /* * Invalidate guest mappings identified by its vpid from the TLB. */ static __inline void vmx_invvpid(struct vmx *vmx, int vcpu, pmap_t pmap, int running) { struct vmxstate *vmxstate; struct invvpid_desc invvpid_desc; vmxstate = &vmx->state[vcpu]; if (vmxstate->vpid == 0) return; if (!running) { /* * Set the 'lastcpu' to an invalid host cpu. * * This will invalidate TLB entries tagged with the vcpu's * vpid the next time it runs via vmx_set_pcpu_defaults(). */ vmxstate->lastcpu = NOCPU; return; } KASSERT(curthread->td_critnest > 0, ("%s: vcpu %d running outside " "critical section", __func__, vcpu)); /* * Invalidate all mappings tagged with 'vpid' * * We do this because this vcpu was executing on a different host * cpu when it last ran. We do not track whether it invalidated * mappings associated with its 'vpid' during that run. So we must * assume that the mappings associated with 'vpid' on 'curcpu' are * stale and invalidate them. * * Note that we incur this penalty only when the scheduler chooses to * move the thread associated with this vcpu between host cpus. * * Note also that this will invalidate mappings tagged with 'vpid' * for "all" EP4TAs. */ if (pmap->pm_eptgen == vmx->eptgen[curcpu]) { invvpid_desc._res1 = 0; invvpid_desc._res2 = 0; invvpid_desc.vpid = vmxstate->vpid; invvpid_desc.linear_addr = 0; invvpid(INVVPID_TYPE_SINGLE_CONTEXT, invvpid_desc); vmm_stat_incr(vmx->vm, vcpu, VCPU_INVVPID_DONE, 1); } else { /* * The invvpid can be skipped if an invept is going to * be performed before entering the guest. The invept * will invalidate combined mappings tagged with * 'vmx->eptp' for all vpids. */ vmm_stat_incr(vmx->vm, vcpu, VCPU_INVVPID_SAVED, 1); } } static void vmx_set_pcpu_defaults(struct vmx *vmx, int vcpu, pmap_t pmap) { struct vmxstate *vmxstate; vmxstate = &vmx->state[vcpu]; if (vmxstate->lastcpu == curcpu) return; vmxstate->lastcpu = curcpu; vmm_stat_incr(vmx->vm, vcpu, VCPU_MIGRATIONS, 1); vmcs_write(VMCS_HOST_TR_BASE, vmm_get_host_trbase()); vmcs_write(VMCS_HOST_GDTR_BASE, vmm_get_host_gdtrbase()); vmcs_write(VMCS_HOST_GS_BASE, vmm_get_host_gsbase()); vmx_invvpid(vmx, vcpu, pmap, 1); } /* * We depend on 'procbased_ctls' to have the Interrupt Window Exiting bit set. */ CTASSERT((PROCBASED_CTLS_ONE_SETTING & PROCBASED_INT_WINDOW_EXITING) != 0); static void __inline vmx_set_int_window_exiting(struct vmx *vmx, int vcpu) { if ((vmx->cap[vcpu].proc_ctls & PROCBASED_INT_WINDOW_EXITING) == 0) { vmx->cap[vcpu].proc_ctls |= PROCBASED_INT_WINDOW_EXITING; vmcs_write(VMCS_PRI_PROC_BASED_CTLS, vmx->cap[vcpu].proc_ctls); VCPU_CTR0(vmx->vm, vcpu, "Enabling interrupt window exiting"); } } static void __inline vmx_clear_int_window_exiting(struct vmx *vmx, int vcpu) { KASSERT((vmx->cap[vcpu].proc_ctls & PROCBASED_INT_WINDOW_EXITING) != 0, ("intr_window_exiting not set: %#x", vmx->cap[vcpu].proc_ctls)); vmx->cap[vcpu].proc_ctls &= ~PROCBASED_INT_WINDOW_EXITING; vmcs_write(VMCS_PRI_PROC_BASED_CTLS, vmx->cap[vcpu].proc_ctls); VCPU_CTR0(vmx->vm, vcpu, "Disabling interrupt window exiting"); } static void __inline vmx_set_nmi_window_exiting(struct vmx *vmx, int vcpu) { if ((vmx->cap[vcpu].proc_ctls & PROCBASED_NMI_WINDOW_EXITING) == 0) { vmx->cap[vcpu].proc_ctls |= PROCBASED_NMI_WINDOW_EXITING; vmcs_write(VMCS_PRI_PROC_BASED_CTLS, vmx->cap[vcpu].proc_ctls); VCPU_CTR0(vmx->vm, vcpu, "Enabling NMI window exiting"); } } static void __inline vmx_clear_nmi_window_exiting(struct vmx *vmx, int vcpu) { KASSERT((vmx->cap[vcpu].proc_ctls & PROCBASED_NMI_WINDOW_EXITING) != 0, ("nmi_window_exiting not set %#x", vmx->cap[vcpu].proc_ctls)); vmx->cap[vcpu].proc_ctls &= ~PROCBASED_NMI_WINDOW_EXITING; vmcs_write(VMCS_PRI_PROC_BASED_CTLS, vmx->cap[vcpu].proc_ctls); VCPU_CTR0(vmx->vm, vcpu, "Disabling NMI window exiting"); } int vmx_set_tsc_offset(struct vmx *vmx, int vcpu, uint64_t offset) { int error; if ((vmx->cap[vcpu].proc_ctls & PROCBASED_TSC_OFFSET) == 0) { vmx->cap[vcpu].proc_ctls |= PROCBASED_TSC_OFFSET; vmcs_write(VMCS_PRI_PROC_BASED_CTLS, vmx->cap[vcpu].proc_ctls); VCPU_CTR0(vmx->vm, vcpu, "Enabling TSC offsetting"); } error = vmwrite(VMCS_TSC_OFFSET, offset); return (error); } #define NMI_BLOCKING (VMCS_INTERRUPTIBILITY_NMI_BLOCKING | \ VMCS_INTERRUPTIBILITY_MOVSS_BLOCKING) #define HWINTR_BLOCKING (VMCS_INTERRUPTIBILITY_STI_BLOCKING | \ VMCS_INTERRUPTIBILITY_MOVSS_BLOCKING) static void vmx_inject_nmi(struct vmx *vmx, int vcpu) { uint32_t gi, info; gi = vmcs_read(VMCS_GUEST_INTERRUPTIBILITY); KASSERT((gi & NMI_BLOCKING) == 0, ("vmx_inject_nmi: invalid guest " "interruptibility-state %#x", gi)); info = vmcs_read(VMCS_ENTRY_INTR_INFO); KASSERT((info & VMCS_INTR_VALID) == 0, ("vmx_inject_nmi: invalid " "VM-entry interruption information %#x", info)); /* * Inject the virtual NMI. The vector must be the NMI IDT entry * or the VMCS entry check will fail. */ info = IDT_NMI | VMCS_INTR_T_NMI | VMCS_INTR_VALID; vmcs_write(VMCS_ENTRY_INTR_INFO, info); VCPU_CTR0(vmx->vm, vcpu, "Injecting vNMI"); /* Clear the request */ vm_nmi_clear(vmx->vm, vcpu); } static void vmx_inject_interrupts(struct vmx *vmx, int vcpu, struct vlapic *vlapic, uint64_t guestrip) { int vector, need_nmi_exiting, extint_pending; uint64_t rflags, entryinfo; uint32_t gi, info; if (vmx->state[vcpu].nextrip != guestrip) { gi = vmcs_read(VMCS_GUEST_INTERRUPTIBILITY); if (gi & HWINTR_BLOCKING) { VCPU_CTR2(vmx->vm, vcpu, "Guest interrupt blocking " "cleared due to rip change: %#lx/%#lx", vmx->state[vcpu].nextrip, guestrip); gi &= ~HWINTR_BLOCKING; vmcs_write(VMCS_GUEST_INTERRUPTIBILITY, gi); } } if (vm_entry_intinfo(vmx->vm, vcpu, &entryinfo)) { KASSERT((entryinfo & VMCS_INTR_VALID) != 0, ("%s: entry " "intinfo is not valid: %#lx", __func__, entryinfo)); info = vmcs_read(VMCS_ENTRY_INTR_INFO); KASSERT((info & VMCS_INTR_VALID) == 0, ("%s: cannot inject " "pending exception: %#lx/%#x", __func__, entryinfo, info)); info = entryinfo; vector = info & 0xff; if (vector == IDT_BP || vector == IDT_OF) { /* * VT-x requires #BP and #OF to be injected as software * exceptions. */ info &= ~VMCS_INTR_T_MASK; info |= VMCS_INTR_T_SWEXCEPTION; } if (info & VMCS_INTR_DEL_ERRCODE) vmcs_write(VMCS_ENTRY_EXCEPTION_ERROR, entryinfo >> 32); vmcs_write(VMCS_ENTRY_INTR_INFO, info); } if (vm_nmi_pending(vmx->vm, vcpu)) { /* * If there are no conditions blocking NMI injection then * inject it directly here otherwise enable "NMI window * exiting" to inject it as soon as we can. * * We also check for STI_BLOCKING because some implementations * don't allow NMI injection in this case. If we are running * on a processor that doesn't have this restriction it will * immediately exit and the NMI will be injected in the * "NMI window exiting" handler. */ need_nmi_exiting = 1; gi = vmcs_read(VMCS_GUEST_INTERRUPTIBILITY); if ((gi & (HWINTR_BLOCKING | NMI_BLOCKING)) == 0) { info = vmcs_read(VMCS_ENTRY_INTR_INFO); if ((info & VMCS_INTR_VALID) == 0) { vmx_inject_nmi(vmx, vcpu); need_nmi_exiting = 0; } else { VCPU_CTR1(vmx->vm, vcpu, "Cannot inject NMI " "due to VM-entry intr info %#x", info); } } else { VCPU_CTR1(vmx->vm, vcpu, "Cannot inject NMI due to " "Guest Interruptibility-state %#x", gi); } if (need_nmi_exiting) vmx_set_nmi_window_exiting(vmx, vcpu); } extint_pending = vm_extint_pending(vmx->vm, vcpu); if (!extint_pending && virtual_interrupt_delivery) { vmx_inject_pir(vlapic); return; } /* * If interrupt-window exiting is already in effect then don't bother * checking for pending interrupts. This is just an optimization and * not needed for correctness. */ if ((vmx->cap[vcpu].proc_ctls & PROCBASED_INT_WINDOW_EXITING) != 0) { VCPU_CTR0(vmx->vm, vcpu, "Skip interrupt injection due to " "pending int_window_exiting"); return; } if (!extint_pending) { /* Ask the local apic for a vector to inject */ if (!vlapic_pending_intr(vlapic, &vector)) return; /* * From the Intel SDM, Volume 3, Section "Maskable * Hardware Interrupts": * - maskable interrupt vectors [16,255] can be delivered * through the local APIC. */ KASSERT(vector >= 16 && vector <= 255, ("invalid vector %d from local APIC", vector)); } else { /* Ask the legacy pic for a vector to inject */ vatpic_pending_intr(vmx->vm, &vector); /* * From the Intel SDM, Volume 3, Section "Maskable * Hardware Interrupts": * - maskable interrupt vectors [0,255] can be delivered * through the INTR pin. */ KASSERT(vector >= 0 && vector <= 255, ("invalid vector %d from INTR", vector)); } /* Check RFLAGS.IF and the interruptibility state of the guest */ rflags = vmcs_read(VMCS_GUEST_RFLAGS); if ((rflags & PSL_I) == 0) { VCPU_CTR2(vmx->vm, vcpu, "Cannot inject vector %d due to " "rflags %#lx", vector, rflags); goto cantinject; } gi = vmcs_read(VMCS_GUEST_INTERRUPTIBILITY); if (gi & HWINTR_BLOCKING) { VCPU_CTR2(vmx->vm, vcpu, "Cannot inject vector %d due to " "Guest Interruptibility-state %#x", vector, gi); goto cantinject; } info = vmcs_read(VMCS_ENTRY_INTR_INFO); if (info & VMCS_INTR_VALID) { /* * This is expected and could happen for multiple reasons: * - A vectoring VM-entry was aborted due to astpending * - A VM-exit happened during event injection. * - An exception was injected above. * - An NMI was injected above or after "NMI window exiting" */ VCPU_CTR2(vmx->vm, vcpu, "Cannot inject vector %d due to " "VM-entry intr info %#x", vector, info); goto cantinject; } /* Inject the interrupt */ info = VMCS_INTR_T_HWINTR | VMCS_INTR_VALID; info |= vector; vmcs_write(VMCS_ENTRY_INTR_INFO, info); if (!extint_pending) { /* Update the Local APIC ISR */ vlapic_intr_accepted(vlapic, vector); } else { vm_extint_clear(vmx->vm, vcpu); vatpic_intr_accepted(vmx->vm, vector); /* * After we accepted the current ExtINT the PIC may * have posted another one. If that is the case, set * the Interrupt Window Exiting execution control so * we can inject that one too. * * Also, interrupt window exiting allows us to inject any * pending APIC vector that was preempted by the ExtINT * as soon as possible. This applies both for the software * emulated vlapic and the hardware assisted virtual APIC. */ vmx_set_int_window_exiting(vmx, vcpu); } VCPU_CTR1(vmx->vm, vcpu, "Injecting hwintr at vector %d", vector); return; cantinject: /* * Set the Interrupt Window Exiting execution control so we can inject * the interrupt as soon as blocking condition goes away. */ vmx_set_int_window_exiting(vmx, vcpu); } /* * If the Virtual NMIs execution control is '1' then the logical processor * tracks virtual-NMI blocking in the Guest Interruptibility-state field of * the VMCS. An IRET instruction in VMX non-root operation will remove any * virtual-NMI blocking. * * This unblocking occurs even if the IRET causes a fault. In this case the * hypervisor needs to restore virtual-NMI blocking before resuming the guest. */ static void vmx_restore_nmi_blocking(struct vmx *vmx, int vcpuid) { uint32_t gi; VCPU_CTR0(vmx->vm, vcpuid, "Restore Virtual-NMI blocking"); gi = vmcs_read(VMCS_GUEST_INTERRUPTIBILITY); gi |= VMCS_INTERRUPTIBILITY_NMI_BLOCKING; vmcs_write(VMCS_GUEST_INTERRUPTIBILITY, gi); } static void vmx_clear_nmi_blocking(struct vmx *vmx, int vcpuid) { uint32_t gi; VCPU_CTR0(vmx->vm, vcpuid, "Clear Virtual-NMI blocking"); gi = vmcs_read(VMCS_GUEST_INTERRUPTIBILITY); gi &= ~VMCS_INTERRUPTIBILITY_NMI_BLOCKING; vmcs_write(VMCS_GUEST_INTERRUPTIBILITY, gi); } static void vmx_assert_nmi_blocking(struct vmx *vmx, int vcpuid) { uint32_t gi; gi = vmcs_read(VMCS_GUEST_INTERRUPTIBILITY); KASSERT(gi & VMCS_INTERRUPTIBILITY_NMI_BLOCKING, ("NMI blocking is not in effect %#x", gi)); } static int vmx_emulate_xsetbv(struct vmx *vmx, int vcpu, struct vm_exit *vmexit) { struct vmxctx *vmxctx; uint64_t xcrval; const struct xsave_limits *limits; vmxctx = &vmx->ctx[vcpu]; limits = vmm_get_xsave_limits(); /* * Note that the processor raises a GP# fault on its own if * xsetbv is executed for CPL != 0, so we do not have to * emulate that fault here. */ /* Only xcr0 is supported. */ if (vmxctx->guest_rcx != 0) { vm_inject_gp(vmx->vm, vcpu); return (HANDLED); } /* We only handle xcr0 if both the host and guest have XSAVE enabled. */ if (!limits->xsave_enabled || !(vmcs_read(VMCS_GUEST_CR4) & CR4_XSAVE)) { vm_inject_ud(vmx->vm, vcpu); return (HANDLED); } xcrval = vmxctx->guest_rdx << 32 | (vmxctx->guest_rax & 0xffffffff); if ((xcrval & ~limits->xcr0_allowed) != 0) { vm_inject_gp(vmx->vm, vcpu); return (HANDLED); } if (!(xcrval & XFEATURE_ENABLED_X87)) { vm_inject_gp(vmx->vm, vcpu); return (HANDLED); } /* AVX (YMM_Hi128) requires SSE. */ if (xcrval & XFEATURE_ENABLED_AVX && (xcrval & XFEATURE_AVX) != XFEATURE_AVX) { vm_inject_gp(vmx->vm, vcpu); return (HANDLED); } /* * AVX512 requires base AVX (YMM_Hi128) as well as OpMask, * ZMM_Hi256, and Hi16_ZMM. */ if (xcrval & XFEATURE_AVX512 && (xcrval & (XFEATURE_AVX512 | XFEATURE_AVX)) != (XFEATURE_AVX512 | XFEATURE_AVX)) { vm_inject_gp(vmx->vm, vcpu); return (HANDLED); } /* * Intel MPX requires both bound register state flags to be * set. */ if (((xcrval & XFEATURE_ENABLED_BNDREGS) != 0) != ((xcrval & XFEATURE_ENABLED_BNDCSR) != 0)) { vm_inject_gp(vmx->vm, vcpu); return (HANDLED); } /* * This runs "inside" vmrun() with the guest's FPU state, so * modifying xcr0 directly modifies the guest's xcr0, not the * host's. */ load_xcr(0, xcrval); return (HANDLED); } static uint64_t vmx_get_guest_reg(struct vmx *vmx, int vcpu, int ident) { const struct vmxctx *vmxctx; vmxctx = &vmx->ctx[vcpu]; switch (ident) { case 0: return (vmxctx->guest_rax); case 1: return (vmxctx->guest_rcx); case 2: return (vmxctx->guest_rdx); case 3: return (vmxctx->guest_rbx); case 4: return (vmcs_read(VMCS_GUEST_RSP)); case 5: return (vmxctx->guest_rbp); case 6: return (vmxctx->guest_rsi); case 7: return (vmxctx->guest_rdi); case 8: return (vmxctx->guest_r8); case 9: return (vmxctx->guest_r9); case 10: return (vmxctx->guest_r10); case 11: return (vmxctx->guest_r11); case 12: return (vmxctx->guest_r12); case 13: return (vmxctx->guest_r13); case 14: return (vmxctx->guest_r14); case 15: return (vmxctx->guest_r15); default: panic("invalid vmx register %d", ident); } } static void vmx_set_guest_reg(struct vmx *vmx, int vcpu, int ident, uint64_t regval) { struct vmxctx *vmxctx; vmxctx = &vmx->ctx[vcpu]; switch (ident) { case 0: vmxctx->guest_rax = regval; break; case 1: vmxctx->guest_rcx = regval; break; case 2: vmxctx->guest_rdx = regval; break; case 3: vmxctx->guest_rbx = regval; break; case 4: vmcs_write(VMCS_GUEST_RSP, regval); break; case 5: vmxctx->guest_rbp = regval; break; case 6: vmxctx->guest_rsi = regval; break; case 7: vmxctx->guest_rdi = regval; break; case 8: vmxctx->guest_r8 = regval; break; case 9: vmxctx->guest_r9 = regval; break; case 10: vmxctx->guest_r10 = regval; break; case 11: vmxctx->guest_r11 = regval; break; case 12: vmxctx->guest_r12 = regval; break; case 13: vmxctx->guest_r13 = regval; break; case 14: vmxctx->guest_r14 = regval; break; case 15: vmxctx->guest_r15 = regval; break; default: panic("invalid vmx register %d", ident); } } static int vmx_emulate_cr0_access(struct vmx *vmx, int vcpu, uint64_t exitqual) { uint64_t crval, regval; /* We only handle mov to %cr0 at this time */ if ((exitqual & 0xf0) != 0x00) return (UNHANDLED); regval = vmx_get_guest_reg(vmx, vcpu, (exitqual >> 8) & 0xf); vmcs_write(VMCS_CR0_SHADOW, regval); crval = regval | cr0_ones_mask; crval &= ~cr0_zeros_mask; vmcs_write(VMCS_GUEST_CR0, crval); if (regval & CR0_PG) { uint64_t efer, entry_ctls; /* * If CR0.PG is 1 and EFER.LME is 1 then EFER.LMA and * the "IA-32e mode guest" bit in VM-entry control must be * equal. */ efer = vmcs_read(VMCS_GUEST_IA32_EFER); if (efer & EFER_LME) { efer |= EFER_LMA; vmcs_write(VMCS_GUEST_IA32_EFER, efer); entry_ctls = vmcs_read(VMCS_ENTRY_CTLS); entry_ctls |= VM_ENTRY_GUEST_LMA; vmcs_write(VMCS_ENTRY_CTLS, entry_ctls); } } return (HANDLED); } static int vmx_emulate_cr4_access(struct vmx *vmx, int vcpu, uint64_t exitqual) { uint64_t crval, regval; /* We only handle mov to %cr4 at this time */ if ((exitqual & 0xf0) != 0x00) return (UNHANDLED); regval = vmx_get_guest_reg(vmx, vcpu, (exitqual >> 8) & 0xf); vmcs_write(VMCS_CR4_SHADOW, regval); crval = regval | cr4_ones_mask; crval &= ~cr4_zeros_mask; vmcs_write(VMCS_GUEST_CR4, crval); return (HANDLED); } static int vmx_emulate_cr8_access(struct vmx *vmx, int vcpu, uint64_t exitqual) { struct vlapic *vlapic; uint64_t cr8; int regnum; /* We only handle mov %cr8 to/from a register at this time. */ if ((exitqual & 0xe0) != 0x00) { return (UNHANDLED); } vlapic = vm_lapic(vmx->vm, vcpu); regnum = (exitqual >> 8) & 0xf; if (exitqual & 0x10) { cr8 = vlapic_get_cr8(vlapic); vmx_set_guest_reg(vmx, vcpu, regnum, cr8); } else { cr8 = vmx_get_guest_reg(vmx, vcpu, regnum); vlapic_set_cr8(vlapic, cr8); } return (HANDLED); } /* * From section "Guest Register State" in the Intel SDM: CPL = SS.DPL */ static int vmx_cpl(void) { uint32_t ssar; ssar = vmcs_read(VMCS_GUEST_SS_ACCESS_RIGHTS); return ((ssar >> 5) & 0x3); } static enum vm_cpu_mode vmx_cpu_mode(void) { uint32_t csar; if (vmcs_read(VMCS_GUEST_IA32_EFER) & EFER_LMA) { csar = vmcs_read(VMCS_GUEST_CS_ACCESS_RIGHTS); if (csar & 0x2000) return (CPU_MODE_64BIT); /* CS.L = 1 */ else return (CPU_MODE_COMPATIBILITY); } else if (vmcs_read(VMCS_GUEST_CR0) & CR0_PE) { return (CPU_MODE_PROTECTED); } else { return (CPU_MODE_REAL); } } static enum vm_paging_mode vmx_paging_mode(void) { if (!(vmcs_read(VMCS_GUEST_CR0) & CR0_PG)) return (PAGING_MODE_FLAT); if (!(vmcs_read(VMCS_GUEST_CR4) & CR4_PAE)) return (PAGING_MODE_32); if (vmcs_read(VMCS_GUEST_IA32_EFER) & EFER_LME) return (PAGING_MODE_64); else return (PAGING_MODE_PAE); } static uint64_t inout_str_index(struct vmx *vmx, int vcpuid, int in) { uint64_t val; int error; enum vm_reg_name reg; reg = in ? VM_REG_GUEST_RDI : VM_REG_GUEST_RSI; error = vmx_getreg(vmx, vcpuid, reg, &val); KASSERT(error == 0, ("%s: vmx_getreg error %d", __func__, error)); return (val); } static uint64_t inout_str_count(struct vmx *vmx, int vcpuid, int rep) { uint64_t val; int error; if (rep) { error = vmx_getreg(vmx, vcpuid, VM_REG_GUEST_RCX, &val); KASSERT(!error, ("%s: vmx_getreg error %d", __func__, error)); } else { val = 1; } return (val); } static int inout_str_addrsize(uint32_t inst_info) { uint32_t size; size = (inst_info >> 7) & 0x7; switch (size) { case 0: return (2); /* 16 bit */ case 1: return (4); /* 32 bit */ case 2: return (8); /* 64 bit */ default: panic("%s: invalid size encoding %d", __func__, size); } } static void inout_str_seginfo(struct vmx *vmx, int vcpuid, uint32_t inst_info, int in, struct vm_inout_str *vis) { int error, s; if (in) { vis->seg_name = VM_REG_GUEST_ES; } else { s = (inst_info >> 15) & 0x7; vis->seg_name = vm_segment_name(s); } error = vmx_getdesc(vmx, vcpuid, vis->seg_name, &vis->seg_desc); KASSERT(error == 0, ("%s: vmx_getdesc error %d", __func__, error)); } static void vmx_paging_info(struct vm_guest_paging *paging) { paging->cr3 = vmcs_guest_cr3(); paging->cpl = vmx_cpl(); paging->cpu_mode = vmx_cpu_mode(); paging->paging_mode = vmx_paging_mode(); } static void vmexit_inst_emul(struct vm_exit *vmexit, uint64_t gpa, uint64_t gla) { struct vm_guest_paging *paging; uint32_t csar; paging = &vmexit->u.inst_emul.paging; vmexit->exitcode = VM_EXITCODE_INST_EMUL; vmexit->inst_length = 0; vmexit->u.inst_emul.gpa = gpa; vmexit->u.inst_emul.gla = gla; vmx_paging_info(paging); switch (paging->cpu_mode) { case CPU_MODE_REAL: vmexit->u.inst_emul.cs_base = vmcs_read(VMCS_GUEST_CS_BASE); vmexit->u.inst_emul.cs_d = 0; break; case CPU_MODE_PROTECTED: case CPU_MODE_COMPATIBILITY: vmexit->u.inst_emul.cs_base = vmcs_read(VMCS_GUEST_CS_BASE); csar = vmcs_read(VMCS_GUEST_CS_ACCESS_RIGHTS); vmexit->u.inst_emul.cs_d = SEG_DESC_DEF32(csar); break; default: vmexit->u.inst_emul.cs_base = 0; vmexit->u.inst_emul.cs_d = 0; break; } vie_init(&vmexit->u.inst_emul.vie, NULL, 0); } static int ept_fault_type(uint64_t ept_qual) { int fault_type; if (ept_qual & EPT_VIOLATION_DATA_WRITE) fault_type = VM_PROT_WRITE; else if (ept_qual & EPT_VIOLATION_INST_FETCH) fault_type = VM_PROT_EXECUTE; else fault_type= VM_PROT_READ; return (fault_type); } static bool ept_emulation_fault(uint64_t ept_qual) { int read, write; /* EPT fault on an instruction fetch doesn't make sense here */ if (ept_qual & EPT_VIOLATION_INST_FETCH) return (false); /* EPT fault must be a read fault or a write fault */ read = ept_qual & EPT_VIOLATION_DATA_READ ? 1 : 0; write = ept_qual & EPT_VIOLATION_DATA_WRITE ? 1 : 0; if ((read | write) == 0) return (false); /* * The EPT violation must have been caused by accessing a * guest-physical address that is a translation of a guest-linear * address. */ if ((ept_qual & EPT_VIOLATION_GLA_VALID) == 0 || (ept_qual & EPT_VIOLATION_XLAT_VALID) == 0) { return (false); } return (true); } static __inline int apic_access_virtualization(struct vmx *vmx, int vcpuid) { uint32_t proc_ctls2; proc_ctls2 = vmx->cap[vcpuid].proc_ctls2; return ((proc_ctls2 & PROCBASED2_VIRTUALIZE_APIC_ACCESSES) ? 1 : 0); } static __inline int x2apic_virtualization(struct vmx *vmx, int vcpuid) { uint32_t proc_ctls2; proc_ctls2 = vmx->cap[vcpuid].proc_ctls2; return ((proc_ctls2 & PROCBASED2_VIRTUALIZE_X2APIC_MODE) ? 1 : 0); } static int vmx_handle_apic_write(struct vmx *vmx, int vcpuid, struct vlapic *vlapic, uint64_t qual) { int error, handled, offset; uint32_t *apic_regs, vector; bool retu; handled = HANDLED; offset = APIC_WRITE_OFFSET(qual); if (!apic_access_virtualization(vmx, vcpuid)) { /* * In general there should not be any APIC write VM-exits * unless APIC-access virtualization is enabled. * * However self-IPI virtualization can legitimately trigger * an APIC-write VM-exit so treat it specially. */ if (x2apic_virtualization(vmx, vcpuid) && offset == APIC_OFFSET_SELF_IPI) { apic_regs = (uint32_t *)(vlapic->apic_page); vector = apic_regs[APIC_OFFSET_SELF_IPI / 4]; vlapic_self_ipi_handler(vlapic, vector); return (HANDLED); } else return (UNHANDLED); } switch (offset) { case APIC_OFFSET_ID: vlapic_id_write_handler(vlapic); break; case APIC_OFFSET_LDR: vlapic_ldr_write_handler(vlapic); break; case APIC_OFFSET_DFR: vlapic_dfr_write_handler(vlapic); break; case APIC_OFFSET_SVR: vlapic_svr_write_handler(vlapic); break; case APIC_OFFSET_ESR: vlapic_esr_write_handler(vlapic); break; case APIC_OFFSET_ICR_LOW: retu = false; error = vlapic_icrlo_write_handler(vlapic, &retu); if (error != 0 || retu) handled = UNHANDLED; break; case APIC_OFFSET_CMCI_LVT: case APIC_OFFSET_TIMER_LVT ... APIC_OFFSET_ERROR_LVT: vlapic_lvt_write_handler(vlapic, offset); break; case APIC_OFFSET_TIMER_ICR: vlapic_icrtmr_write_handler(vlapic); break; case APIC_OFFSET_TIMER_DCR: vlapic_dcr_write_handler(vlapic); break; default: handled = UNHANDLED; break; } return (handled); } static bool apic_access_fault(struct vmx *vmx, int vcpuid, uint64_t gpa) { if (apic_access_virtualization(vmx, vcpuid) && (gpa >= DEFAULT_APIC_BASE && gpa < DEFAULT_APIC_BASE + PAGE_SIZE)) return (true); else return (false); } static int vmx_handle_apic_access(struct vmx *vmx, int vcpuid, struct vm_exit *vmexit) { uint64_t qual; int access_type, offset, allowed; if (!apic_access_virtualization(vmx, vcpuid)) return (UNHANDLED); qual = vmexit->u.vmx.exit_qualification; access_type = APIC_ACCESS_TYPE(qual); offset = APIC_ACCESS_OFFSET(qual); allowed = 0; if (access_type == 0) { /* * Read data access to the following registers is expected. */ switch (offset) { case APIC_OFFSET_APR: case APIC_OFFSET_PPR: case APIC_OFFSET_RRR: case APIC_OFFSET_CMCI_LVT: case APIC_OFFSET_TIMER_CCR: allowed = 1; break; default: break; } } else if (access_type == 1) { /* * Write data access to the following registers is expected. */ switch (offset) { case APIC_OFFSET_VER: case APIC_OFFSET_APR: case APIC_OFFSET_PPR: case APIC_OFFSET_RRR: case APIC_OFFSET_ISR0 ... APIC_OFFSET_ISR7: case APIC_OFFSET_TMR0 ... APIC_OFFSET_TMR7: case APIC_OFFSET_IRR0 ... APIC_OFFSET_IRR7: case APIC_OFFSET_CMCI_LVT: case APIC_OFFSET_TIMER_CCR: allowed = 1; break; default: break; } } if (allowed) { vmexit_inst_emul(vmexit, DEFAULT_APIC_BASE + offset, VIE_INVALID_GLA); } /* * Regardless of whether the APIC-access is allowed this handler * always returns UNHANDLED: * - if the access is allowed then it is handled by emulating the * instruction that caused the VM-exit (outside the critical section) * - if the access is not allowed then it will be converted to an * exitcode of VM_EXITCODE_VMX and will be dealt with in userland. */ return (UNHANDLED); } static enum task_switch_reason vmx_task_switch_reason(uint64_t qual) { int reason; reason = (qual >> 30) & 0x3; switch (reason) { case 0: return (TSR_CALL); case 1: return (TSR_IRET); case 2: return (TSR_JMP); case 3: return (TSR_IDT_GATE); default: panic("%s: invalid reason %d", __func__, reason); } } static int emulate_wrmsr(struct vmx *vmx, int vcpuid, u_int num, uint64_t val, bool *retu) { int error; if (lapic_msr(num)) error = lapic_wrmsr(vmx->vm, vcpuid, num, val, retu); else error = vmx_wrmsr(vmx, vcpuid, num, val, retu); return (error); } static int emulate_rdmsr(struct vmx *vmx, int vcpuid, u_int num, bool *retu) { struct vmxctx *vmxctx; uint64_t result; uint32_t eax, edx; int error; if (lapic_msr(num)) error = lapic_rdmsr(vmx->vm, vcpuid, num, &result, retu); else error = vmx_rdmsr(vmx, vcpuid, num, &result, retu); if (error == 0) { eax = result; vmxctx = &vmx->ctx[vcpuid]; error = vmxctx_setreg(vmxctx, VM_REG_GUEST_RAX, eax); KASSERT(error == 0, ("vmxctx_setreg(rax) error %d", error)); edx = result >> 32; error = vmxctx_setreg(vmxctx, VM_REG_GUEST_RDX, edx); KASSERT(error == 0, ("vmxctx_setreg(rdx) error %d", error)); } return (error); } static int vmx_exit_process(struct vmx *vmx, int vcpu, struct vm_exit *vmexit) { int error, errcode, errcode_valid, handled, in; struct vmxctx *vmxctx; struct vlapic *vlapic; struct vm_inout_str *vis; struct vm_task_switch *ts; uint32_t eax, ecx, edx, idtvec_info, idtvec_err, intr_info, inst_info; uint32_t intr_type, intr_vec, reason; uint64_t exitintinfo, qual, gpa; bool retu; CTASSERT((PINBASED_CTLS_ONE_SETTING & PINBASED_VIRTUAL_NMI) != 0); CTASSERT((PINBASED_CTLS_ONE_SETTING & PINBASED_NMI_EXITING) != 0); handled = UNHANDLED; vmxctx = &vmx->ctx[vcpu]; qual = vmexit->u.vmx.exit_qualification; reason = vmexit->u.vmx.exit_reason; vmexit->exitcode = VM_EXITCODE_BOGUS; vmm_stat_incr(vmx->vm, vcpu, VMEXIT_COUNT, 1); SDT_PROBE3(vmm, vmx, exit, entry, vmx, vcpu, vmexit); /* * VM-entry failures during or after loading guest state. * * These VM-exits are uncommon but must be handled specially * as most VM-exit fields are not populated as usual. */ if (__predict_false(reason == EXIT_REASON_MCE_DURING_ENTRY)) { VCPU_CTR0(vmx->vm, vcpu, "Handling MCE during VM-entry"); __asm __volatile("int $18"); return (1); } /* * VM exits that can be triggered during event delivery need to * be handled specially by re-injecting the event if the IDT * vectoring information field's valid bit is set. * * See "Information for VM Exits During Event Delivery" in Intel SDM * for details. */ idtvec_info = vmcs_idt_vectoring_info(); if (idtvec_info & VMCS_IDT_VEC_VALID) { idtvec_info &= ~(1 << 12); /* clear undefined bit */ exitintinfo = idtvec_info; if (idtvec_info & VMCS_IDT_VEC_ERRCODE_VALID) { idtvec_err = vmcs_idt_vectoring_err(); exitintinfo |= (uint64_t)idtvec_err << 32; } error = vm_exit_intinfo(vmx->vm, vcpu, exitintinfo); KASSERT(error == 0, ("%s: vm_set_intinfo error %d", __func__, error)); /* * If 'virtual NMIs' are being used and the VM-exit * happened while injecting an NMI during the previous * VM-entry, then clear "blocking by NMI" in the * Guest Interruptibility-State so the NMI can be * reinjected on the subsequent VM-entry. * * However, if the NMI was being delivered through a task * gate, then the new task must start execution with NMIs * blocked so don't clear NMI blocking in this case. */ intr_type = idtvec_info & VMCS_INTR_T_MASK; if (intr_type == VMCS_INTR_T_NMI) { if (reason != EXIT_REASON_TASK_SWITCH) vmx_clear_nmi_blocking(vmx, vcpu); else vmx_assert_nmi_blocking(vmx, vcpu); } /* * Update VM-entry instruction length if the event being * delivered was a software interrupt or software exception. */ if (intr_type == VMCS_INTR_T_SWINTR || intr_type == VMCS_INTR_T_PRIV_SWEXCEPTION || intr_type == VMCS_INTR_T_SWEXCEPTION) { vmcs_write(VMCS_ENTRY_INST_LENGTH, vmexit->inst_length); } } switch (reason) { case EXIT_REASON_TASK_SWITCH: ts = &vmexit->u.task_switch; ts->tsssel = qual & 0xffff; ts->reason = vmx_task_switch_reason(qual); ts->ext = 0; ts->errcode_valid = 0; vmx_paging_info(&ts->paging); /* * If the task switch was due to a CALL, JMP, IRET, software * interrupt (INT n) or software exception (INT3, INTO), * then the saved %rip references the instruction that caused * the task switch. The instruction length field in the VMCS * is valid in this case. * * In all other cases (e.g., NMI, hardware exception) the * saved %rip is one that would have been saved in the old TSS * had the task switch completed normally so the instruction * length field is not needed in this case and is explicitly * set to 0. */ if (ts->reason == TSR_IDT_GATE) { KASSERT(idtvec_info & VMCS_IDT_VEC_VALID, ("invalid idtvec_info %#x for IDT task switch", idtvec_info)); intr_type = idtvec_info & VMCS_INTR_T_MASK; if (intr_type != VMCS_INTR_T_SWINTR && intr_type != VMCS_INTR_T_SWEXCEPTION && intr_type != VMCS_INTR_T_PRIV_SWEXCEPTION) { /* Task switch triggered by external event */ ts->ext = 1; vmexit->inst_length = 0; if (idtvec_info & VMCS_IDT_VEC_ERRCODE_VALID) { ts->errcode_valid = 1; ts->errcode = vmcs_idt_vectoring_err(); } } } vmexit->exitcode = VM_EXITCODE_TASK_SWITCH; SDT_PROBE4(vmm, vmx, exit, taskswitch, vmx, vcpu, vmexit, ts); VCPU_CTR4(vmx->vm, vcpu, "task switch reason %d, tss 0x%04x, " "%s errcode 0x%016lx", ts->reason, ts->tsssel, ts->ext ? "external" : "internal", ((uint64_t)ts->errcode << 32) | ts->errcode_valid); break; case EXIT_REASON_CR_ACCESS: vmm_stat_incr(vmx->vm, vcpu, VMEXIT_CR_ACCESS, 1); SDT_PROBE4(vmm, vmx, exit, craccess, vmx, vcpu, vmexit, qual); switch (qual & 0xf) { case 0: handled = vmx_emulate_cr0_access(vmx, vcpu, qual); break; case 4: handled = vmx_emulate_cr4_access(vmx, vcpu, qual); break; case 8: handled = vmx_emulate_cr8_access(vmx, vcpu, qual); break; } break; case EXIT_REASON_RDMSR: vmm_stat_incr(vmx->vm, vcpu, VMEXIT_RDMSR, 1); retu = false; ecx = vmxctx->guest_rcx; VCPU_CTR1(vmx->vm, vcpu, "rdmsr 0x%08x", ecx); SDT_PROBE4(vmm, vmx, exit, rdmsr, vmx, vcpu, vmexit, ecx); error = emulate_rdmsr(vmx, vcpu, ecx, &retu); if (error) { vmexit->exitcode = VM_EXITCODE_RDMSR; vmexit->u.msr.code = ecx; } else if (!retu) { handled = HANDLED; } else { /* Return to userspace with a valid exitcode */ KASSERT(vmexit->exitcode != VM_EXITCODE_BOGUS, ("emulate_rdmsr retu with bogus exitcode")); } break; case EXIT_REASON_WRMSR: vmm_stat_incr(vmx->vm, vcpu, VMEXIT_WRMSR, 1); retu = false; eax = vmxctx->guest_rax; ecx = vmxctx->guest_rcx; edx = vmxctx->guest_rdx; VCPU_CTR2(vmx->vm, vcpu, "wrmsr 0x%08x value 0x%016lx", ecx, (uint64_t)edx << 32 | eax); SDT_PROBE5(vmm, vmx, exit, wrmsr, vmx, vmexit, vcpu, ecx, (uint64_t)edx << 32 | eax); error = emulate_wrmsr(vmx, vcpu, ecx, (uint64_t)edx << 32 | eax, &retu); if (error) { vmexit->exitcode = VM_EXITCODE_WRMSR; vmexit->u.msr.code = ecx; vmexit->u.msr.wval = (uint64_t)edx << 32 | eax; } else if (!retu) { handled = HANDLED; } else { /* Return to userspace with a valid exitcode */ KASSERT(vmexit->exitcode != VM_EXITCODE_BOGUS, ("emulate_wrmsr retu with bogus exitcode")); } break; case EXIT_REASON_HLT: vmm_stat_incr(vmx->vm, vcpu, VMEXIT_HLT, 1); SDT_PROBE3(vmm, vmx, exit, halt, vmx, vcpu, vmexit); vmexit->exitcode = VM_EXITCODE_HLT; vmexit->u.hlt.rflags = vmcs_read(VMCS_GUEST_RFLAGS); if (virtual_interrupt_delivery) vmexit->u.hlt.intr_status = vmcs_read(VMCS_GUEST_INTR_STATUS); else vmexit->u.hlt.intr_status = 0; break; case EXIT_REASON_MTF: vmm_stat_incr(vmx->vm, vcpu, VMEXIT_MTRAP, 1); SDT_PROBE3(vmm, vmx, exit, mtrap, vmx, vcpu, vmexit); vmexit->exitcode = VM_EXITCODE_MTRAP; vmexit->inst_length = 0; break; case EXIT_REASON_PAUSE: vmm_stat_incr(vmx->vm, vcpu, VMEXIT_PAUSE, 1); SDT_PROBE3(vmm, vmx, exit, pause, vmx, vcpu, vmexit); vmexit->exitcode = VM_EXITCODE_PAUSE; break; case EXIT_REASON_INTR_WINDOW: vmm_stat_incr(vmx->vm, vcpu, VMEXIT_INTR_WINDOW, 1); SDT_PROBE3(vmm, vmx, exit, intrwindow, vmx, vcpu, vmexit); vmx_clear_int_window_exiting(vmx, vcpu); return (1); case EXIT_REASON_EXT_INTR: /* * External interrupts serve only to cause VM exits and allow * the host interrupt handler to run. * * If this external interrupt triggers a virtual interrupt * to a VM, then that state will be recorded by the * host interrupt handler in the VM's softc. We will inject * this virtual interrupt during the subsequent VM enter. */ intr_info = vmcs_read(VMCS_EXIT_INTR_INFO); SDT_PROBE4(vmm, vmx, exit, interrupt, vmx, vcpu, vmexit, intr_info); /* * XXX: Ignore this exit if VMCS_INTR_VALID is not set. * This appears to be a bug in VMware Fusion? */ if (!(intr_info & VMCS_INTR_VALID)) return (1); KASSERT((intr_info & VMCS_INTR_VALID) != 0 && (intr_info & VMCS_INTR_T_MASK) == VMCS_INTR_T_HWINTR, ("VM exit interruption info invalid: %#x", intr_info)); vmx_trigger_hostintr(intr_info & 0xff); /* * This is special. We want to treat this as an 'handled' * VM-exit but not increment the instruction pointer. */ vmm_stat_incr(vmx->vm, vcpu, VMEXIT_EXTINT, 1); return (1); case EXIT_REASON_NMI_WINDOW: SDT_PROBE3(vmm, vmx, exit, nmiwindow, vmx, vcpu, vmexit); /* Exit to allow the pending virtual NMI to be injected */ if (vm_nmi_pending(vmx->vm, vcpu)) vmx_inject_nmi(vmx, vcpu); vmx_clear_nmi_window_exiting(vmx, vcpu); vmm_stat_incr(vmx->vm, vcpu, VMEXIT_NMI_WINDOW, 1); return (1); case EXIT_REASON_INOUT: vmm_stat_incr(vmx->vm, vcpu, VMEXIT_INOUT, 1); vmexit->exitcode = VM_EXITCODE_INOUT; vmexit->u.inout.bytes = (qual & 0x7) + 1; vmexit->u.inout.in = in = (qual & 0x8) ? 1 : 0; vmexit->u.inout.string = (qual & 0x10) ? 1 : 0; vmexit->u.inout.rep = (qual & 0x20) ? 1 : 0; vmexit->u.inout.port = (uint16_t)(qual >> 16); vmexit->u.inout.eax = (uint32_t)(vmxctx->guest_rax); if (vmexit->u.inout.string) { inst_info = vmcs_read(VMCS_EXIT_INSTRUCTION_INFO); vmexit->exitcode = VM_EXITCODE_INOUT_STR; vis = &vmexit->u.inout_str; vmx_paging_info(&vis->paging); vis->rflags = vmcs_read(VMCS_GUEST_RFLAGS); vis->cr0 = vmcs_read(VMCS_GUEST_CR0); vis->index = inout_str_index(vmx, vcpu, in); vis->count = inout_str_count(vmx, vcpu, vis->inout.rep); vis->addrsize = inout_str_addrsize(inst_info); inout_str_seginfo(vmx, vcpu, inst_info, in, vis); } SDT_PROBE3(vmm, vmx, exit, inout, vmx, vcpu, vmexit); break; case EXIT_REASON_CPUID: vmm_stat_incr(vmx->vm, vcpu, VMEXIT_CPUID, 1); SDT_PROBE3(vmm, vmx, exit, cpuid, vmx, vcpu, vmexit); handled = vmx_handle_cpuid(vmx->vm, vcpu, vmxctx); break; case EXIT_REASON_EXCEPTION: vmm_stat_incr(vmx->vm, vcpu, VMEXIT_EXCEPTION, 1); intr_info = vmcs_read(VMCS_EXIT_INTR_INFO); KASSERT((intr_info & VMCS_INTR_VALID) != 0, ("VM exit interruption info invalid: %#x", intr_info)); intr_vec = intr_info & 0xff; intr_type = intr_info & VMCS_INTR_T_MASK; /* * If Virtual NMIs control is 1 and the VM-exit is due to a * fault encountered during the execution of IRET then we must * restore the state of "virtual-NMI blocking" before resuming * the guest. * * See "Resuming Guest Software after Handling an Exception". * See "Information for VM Exits Due to Vectored Events". */ if ((idtvec_info & VMCS_IDT_VEC_VALID) == 0 && (intr_vec != IDT_DF) && (intr_info & EXIT_QUAL_NMIUDTI) != 0) vmx_restore_nmi_blocking(vmx, vcpu); /* * The NMI has already been handled in vmx_exit_handle_nmi(). */ if (intr_type == VMCS_INTR_T_NMI) return (1); /* * Call the machine check handler by hand. Also don't reflect * the machine check back into the guest. */ if (intr_vec == IDT_MC) { VCPU_CTR0(vmx->vm, vcpu, "Vectoring to MCE handler"); __asm __volatile("int $18"); return (1); } /* * If the hypervisor has requested user exits for * debug exceptions, bounce them out to userland. */ if (intr_type == VMCS_INTR_T_SWEXCEPTION && intr_vec == IDT_BP && (vmx->cap[vcpu].set & (1 << VM_CAP_BPT_EXIT))) { vmexit->exitcode = VM_EXITCODE_BPT; vmexit->u.bpt.inst_length = vmexit->inst_length; vmexit->inst_length = 0; break; } if (intr_vec == IDT_PF) { error = vmxctx_setreg(vmxctx, VM_REG_GUEST_CR2, qual); KASSERT(error == 0, ("%s: vmxctx_setreg(cr2) error %d", __func__, error)); } /* * Software exceptions exhibit trap-like behavior. This in * turn requires populating the VM-entry instruction length * so that the %rip in the trap frame is past the INT3/INTO * instruction. */ if (intr_type == VMCS_INTR_T_SWEXCEPTION) vmcs_write(VMCS_ENTRY_INST_LENGTH, vmexit->inst_length); /* Reflect all other exceptions back into the guest */ errcode_valid = errcode = 0; if (intr_info & VMCS_INTR_DEL_ERRCODE) { errcode_valid = 1; errcode = vmcs_read(VMCS_EXIT_INTR_ERRCODE); } VCPU_CTR2(vmx->vm, vcpu, "Reflecting exception %d/%#x into " "the guest", intr_vec, errcode); SDT_PROBE5(vmm, vmx, exit, exception, vmx, vcpu, vmexit, intr_vec, errcode); error = vm_inject_exception(vmx->vm, vcpu, intr_vec, errcode_valid, errcode, 0); KASSERT(error == 0, ("%s: vm_inject_exception error %d", __func__, error)); return (1); case EXIT_REASON_EPT_FAULT: /* * If 'gpa' lies within the address space allocated to * memory then this must be a nested page fault otherwise * this must be an instruction that accesses MMIO space. */ gpa = vmcs_gpa(); if (vm_mem_allocated(vmx->vm, vcpu, gpa) || apic_access_fault(vmx, vcpu, gpa)) { vmexit->exitcode = VM_EXITCODE_PAGING; vmexit->inst_length = 0; vmexit->u.paging.gpa = gpa; vmexit->u.paging.fault_type = ept_fault_type(qual); vmm_stat_incr(vmx->vm, vcpu, VMEXIT_NESTED_FAULT, 1); SDT_PROBE5(vmm, vmx, exit, nestedfault, vmx, vcpu, vmexit, gpa, qual); } else if (ept_emulation_fault(qual)) { vmexit_inst_emul(vmexit, gpa, vmcs_gla()); vmm_stat_incr(vmx->vm, vcpu, VMEXIT_INST_EMUL, 1); SDT_PROBE4(vmm, vmx, exit, mmiofault, vmx, vcpu, vmexit, gpa); } /* * If Virtual NMIs control is 1 and the VM-exit is due to an * EPT fault during the execution of IRET then we must restore * the state of "virtual-NMI blocking" before resuming. * * See description of "NMI unblocking due to IRET" in * "Exit Qualification for EPT Violations". */ if ((idtvec_info & VMCS_IDT_VEC_VALID) == 0 && (qual & EXIT_QUAL_NMIUDTI) != 0) vmx_restore_nmi_blocking(vmx, vcpu); break; case EXIT_REASON_VIRTUALIZED_EOI: vmexit->exitcode = VM_EXITCODE_IOAPIC_EOI; vmexit->u.ioapic_eoi.vector = qual & 0xFF; SDT_PROBE3(vmm, vmx, exit, eoi, vmx, vcpu, vmexit); vmexit->inst_length = 0; /* trap-like */ break; case EXIT_REASON_APIC_ACCESS: SDT_PROBE3(vmm, vmx, exit, apicaccess, vmx, vcpu, vmexit); handled = vmx_handle_apic_access(vmx, vcpu, vmexit); break; case EXIT_REASON_APIC_WRITE: /* * APIC-write VM exit is trap-like so the %rip is already * pointing to the next instruction. */ vmexit->inst_length = 0; vlapic = vm_lapic(vmx->vm, vcpu); SDT_PROBE4(vmm, vmx, exit, apicwrite, vmx, vcpu, vmexit, vlapic); handled = vmx_handle_apic_write(vmx, vcpu, vlapic, qual); break; case EXIT_REASON_XSETBV: SDT_PROBE3(vmm, vmx, exit, xsetbv, vmx, vcpu, vmexit); handled = vmx_emulate_xsetbv(vmx, vcpu, vmexit); break; case EXIT_REASON_MONITOR: SDT_PROBE3(vmm, vmx, exit, monitor, vmx, vcpu, vmexit); vmexit->exitcode = VM_EXITCODE_MONITOR; break; case EXIT_REASON_MWAIT: SDT_PROBE3(vmm, vmx, exit, mwait, vmx, vcpu, vmexit); vmexit->exitcode = VM_EXITCODE_MWAIT; break; + case EXIT_REASON_TPR: + vlapic = vm_lapic(vmx->vm, vcpu); + vlapic_sync_tpr(vlapic); + vmexit->inst_length = 0; + handled = HANDLED; + break; case EXIT_REASON_VMCALL: case EXIT_REASON_VMCLEAR: case EXIT_REASON_VMLAUNCH: case EXIT_REASON_VMPTRLD: case EXIT_REASON_VMPTRST: case EXIT_REASON_VMREAD: case EXIT_REASON_VMRESUME: case EXIT_REASON_VMWRITE: case EXIT_REASON_VMXOFF: case EXIT_REASON_VMXON: SDT_PROBE3(vmm, vmx, exit, vminsn, vmx, vcpu, vmexit); vmexit->exitcode = VM_EXITCODE_VMINSN; break; default: SDT_PROBE4(vmm, vmx, exit, unknown, vmx, vcpu, vmexit, reason); vmm_stat_incr(vmx->vm, vcpu, VMEXIT_UNKNOWN, 1); break; } if (handled) { /* * It is possible that control is returned to userland * even though we were able to handle the VM exit in the * kernel. * * In such a case we want to make sure that the userland * restarts guest execution at the instruction *after* * the one we just processed. Therefore we update the * guest rip in the VMCS and in 'vmexit'. */ vmexit->rip += vmexit->inst_length; vmexit->inst_length = 0; vmcs_write(VMCS_GUEST_RIP, vmexit->rip); } else { if (vmexit->exitcode == VM_EXITCODE_BOGUS) { /* * If this VM exit was not claimed by anybody then * treat it as a generic VMX exit. */ vmexit->exitcode = VM_EXITCODE_VMX; vmexit->u.vmx.status = VM_SUCCESS; vmexit->u.vmx.inst_type = 0; vmexit->u.vmx.inst_error = 0; } else { /* * The exitcode and collateral have been populated. * The VM exit will be processed further in userland. */ } } SDT_PROBE4(vmm, vmx, exit, return, vmx, vcpu, vmexit, handled); return (handled); } static __inline void vmx_exit_inst_error(struct vmxctx *vmxctx, int rc, struct vm_exit *vmexit) { KASSERT(vmxctx->inst_fail_status != VM_SUCCESS, ("vmx_exit_inst_error: invalid inst_fail_status %d", vmxctx->inst_fail_status)); vmexit->inst_length = 0; vmexit->exitcode = VM_EXITCODE_VMX; vmexit->u.vmx.status = vmxctx->inst_fail_status; vmexit->u.vmx.inst_error = vmcs_instruction_error(); vmexit->u.vmx.exit_reason = ~0; vmexit->u.vmx.exit_qualification = ~0; switch (rc) { case VMX_VMRESUME_ERROR: case VMX_VMLAUNCH_ERROR: case VMX_INVEPT_ERROR: vmexit->u.vmx.inst_type = rc; break; default: panic("vm_exit_inst_error: vmx_enter_guest returned %d", rc); } } /* * If the NMI-exiting VM execution control is set to '1' then an NMI in * non-root operation causes a VM-exit. NMI blocking is in effect so it is * sufficient to simply vector to the NMI handler via a software interrupt. * However, this must be done before maskable interrupts are enabled * otherwise the "iret" issued by an interrupt handler will incorrectly * clear NMI blocking. */ static __inline void vmx_exit_handle_nmi(struct vmx *vmx, int vcpuid, struct vm_exit *vmexit) { uint32_t intr_info; KASSERT((read_rflags() & PSL_I) == 0, ("interrupts enabled")); if (vmexit->u.vmx.exit_reason != EXIT_REASON_EXCEPTION) return; intr_info = vmcs_read(VMCS_EXIT_INTR_INFO); KASSERT((intr_info & VMCS_INTR_VALID) != 0, ("VM exit interruption info invalid: %#x", intr_info)); if ((intr_info & VMCS_INTR_T_MASK) == VMCS_INTR_T_NMI) { KASSERT((intr_info & 0xff) == IDT_NMI, ("VM exit due " "to NMI has invalid vector: %#x", intr_info)); VCPU_CTR0(vmx->vm, vcpuid, "Vectoring to NMI handler"); __asm __volatile("int $2"); } } static __inline void vmx_dr_enter_guest(struct vmxctx *vmxctx) { register_t rflags; /* Save host control debug registers. */ vmxctx->host_dr7 = rdr7(); vmxctx->host_debugctl = rdmsr(MSR_DEBUGCTLMSR); /* * Disable debugging in DR7 and DEBUGCTL to avoid triggering * exceptions in the host based on the guest DRx values. The * guest DR7 and DEBUGCTL are saved/restored in the VMCS. */ load_dr7(0); wrmsr(MSR_DEBUGCTLMSR, 0); /* * Disable single stepping the kernel to avoid corrupting the * guest DR6. A debugger might still be able to corrupt the * guest DR6 by setting a breakpoint after this point and then * single stepping. */ rflags = read_rflags(); vmxctx->host_tf = rflags & PSL_T; write_rflags(rflags & ~PSL_T); /* Save host debug registers. */ vmxctx->host_dr0 = rdr0(); vmxctx->host_dr1 = rdr1(); vmxctx->host_dr2 = rdr2(); vmxctx->host_dr3 = rdr3(); vmxctx->host_dr6 = rdr6(); /* Restore guest debug registers. */ load_dr0(vmxctx->guest_dr0); load_dr1(vmxctx->guest_dr1); load_dr2(vmxctx->guest_dr2); load_dr3(vmxctx->guest_dr3); load_dr6(vmxctx->guest_dr6); } static __inline void vmx_dr_leave_guest(struct vmxctx *vmxctx) { /* Save guest debug registers. */ vmxctx->guest_dr0 = rdr0(); vmxctx->guest_dr1 = rdr1(); vmxctx->guest_dr2 = rdr2(); vmxctx->guest_dr3 = rdr3(); vmxctx->guest_dr6 = rdr6(); /* * Restore host debug registers. Restore DR7, DEBUGCTL, and * PSL_T last. */ load_dr0(vmxctx->host_dr0); load_dr1(vmxctx->host_dr1); load_dr2(vmxctx->host_dr2); load_dr3(vmxctx->host_dr3); load_dr6(vmxctx->host_dr6); wrmsr(MSR_DEBUGCTLMSR, vmxctx->host_debugctl); load_dr7(vmxctx->host_dr7); write_rflags(read_rflags() | vmxctx->host_tf); } static int vmx_run(void *arg, int vcpu, register_t rip, pmap_t pmap, struct vm_eventinfo *evinfo) { int rc, handled, launched; struct vmx *vmx; struct vm *vm; struct vmxctx *vmxctx; struct vmcs *vmcs; struct vm_exit *vmexit; struct vlapic *vlapic; uint32_t exit_reason; struct region_descriptor gdtr, idtr; uint16_t ldt_sel; vmx = arg; vm = vmx->vm; vmcs = &vmx->vmcs[vcpu]; vmxctx = &vmx->ctx[vcpu]; vlapic = vm_lapic(vm, vcpu); vmexit = vm_exitinfo(vm, vcpu); launched = 0; KASSERT(vmxctx->pmap == pmap, ("pmap %p different than ctx pmap %p", pmap, vmxctx->pmap)); vmx_msr_guest_enter(vmx, vcpu); VMPTRLD(vmcs); /* * XXX * We do this every time because we may setup the virtual machine * from a different process than the one that actually runs it. * * If the life of a virtual machine was spent entirely in the context * of a single process we could do this once in vmx_vminit(). */ vmcs_write(VMCS_HOST_CR3, rcr3()); vmcs_write(VMCS_GUEST_RIP, rip); vmx_set_pcpu_defaults(vmx, vcpu, pmap); do { KASSERT(vmcs_guest_rip() == rip, ("%s: vmcs guest rip mismatch " "%#lx/%#lx", __func__, vmcs_guest_rip(), rip)); handled = UNHANDLED; /* * Interrupts are disabled from this point on until the * guest starts executing. This is done for the following * reasons: * * If an AST is asserted on this thread after the check below, * then the IPI_AST notification will not be lost, because it * will cause a VM exit due to external interrupt as soon as * the guest state is loaded. * * A posted interrupt after 'vmx_inject_interrupts()' will * not be "lost" because it will be held pending in the host * APIC because interrupts are disabled. The pending interrupt * will be recognized as soon as the guest state is loaded. * * The same reasoning applies to the IPI generated by * pmap_invalidate_ept(). */ disable_intr(); vmx_inject_interrupts(vmx, vcpu, vlapic, rip); /* * Check for vcpu suspension after injecting events because * vmx_inject_interrupts() can suspend the vcpu due to a * triple fault. */ if (vcpu_suspended(evinfo)) { enable_intr(); vm_exit_suspended(vmx->vm, vcpu, rip); break; } if (vcpu_rendezvous_pending(evinfo)) { enable_intr(); vm_exit_rendezvous(vmx->vm, vcpu, rip); break; } if (vcpu_reqidle(evinfo)) { enable_intr(); vm_exit_reqidle(vmx->vm, vcpu, rip); break; } if (vcpu_should_yield(vm, vcpu)) { enable_intr(); vm_exit_astpending(vmx->vm, vcpu, rip); vmx_astpending_trace(vmx, vcpu, rip); handled = HANDLED; break; } if (vcpu_debugged(vm, vcpu)) { enable_intr(); vm_exit_debug(vmx->vm, vcpu, rip); break; } /* + * If TPR Shadowing is enabled, the TPR Threshold + * must be updated right before entering the guest. + */ + if (tpr_shadowing && !virtual_interrupt_delivery) { + if ((vmx->cap[vcpu].proc_ctls & PROCBASED_USE_TPR_SHADOW) != 0) { + vmcs_write(VMCS_TPR_THRESHOLD, vlapic_get_cr8(vlapic)); + } + } + + /* * VM exits restore the base address but not the * limits of GDTR and IDTR. The VMCS only stores the * base address, so VM exits set the limits to 0xffff. * Save and restore the full GDTR and IDTR to restore * the limits. * * The VMCS does not save the LDTR at all, and VM * exits clear LDTR as if a NULL selector were loaded. * The userspace hypervisor probably doesn't use a * LDT, but save and restore it to be safe. */ sgdt(&gdtr); sidt(&idtr); ldt_sel = sldt(); vmx_run_trace(vmx, vcpu); vmx_dr_enter_guest(vmxctx); rc = vmx_enter_guest(vmxctx, vmx, launched); vmx_dr_leave_guest(vmxctx); bare_lgdt(&gdtr); lidt(&idtr); lldt(ldt_sel); /* Collect some information for VM exit processing */ vmexit->rip = rip = vmcs_guest_rip(); vmexit->inst_length = vmexit_instruction_length(); vmexit->u.vmx.exit_reason = exit_reason = vmcs_exit_reason(); vmexit->u.vmx.exit_qualification = vmcs_exit_qualification(); /* Update 'nextrip' */ vmx->state[vcpu].nextrip = rip; if (rc == VMX_GUEST_VMEXIT) { vmx_exit_handle_nmi(vmx, vcpu, vmexit); enable_intr(); handled = vmx_exit_process(vmx, vcpu, vmexit); } else { enable_intr(); vmx_exit_inst_error(vmxctx, rc, vmexit); } launched = 1; vmx_exit_trace(vmx, vcpu, rip, exit_reason, handled); rip = vmexit->rip; } while (handled); /* * If a VM exit has been handled then the exitcode must be BOGUS * If a VM exit is not handled then the exitcode must not be BOGUS */ if ((handled && vmexit->exitcode != VM_EXITCODE_BOGUS) || (!handled && vmexit->exitcode == VM_EXITCODE_BOGUS)) { panic("Mismatch between handled (%d) and exitcode (%d)", handled, vmexit->exitcode); } if (!handled) vmm_stat_incr(vm, vcpu, VMEXIT_USERSPACE, 1); VCPU_CTR1(vm, vcpu, "returning from vmx_run: exitcode %d", vmexit->exitcode); VMCLEAR(vmcs); vmx_msr_guest_exit(vmx, vcpu); return (0); } static void vmx_vmcleanup(void *arg) { int i; struct vmx *vmx = arg; uint16_t maxcpus; if (apic_access_virtualization(vmx, 0)) vm_unmap_mmio(vmx->vm, DEFAULT_APIC_BASE, PAGE_SIZE); maxcpus = vm_get_maxcpus(vmx->vm); for (i = 0; i < maxcpus; i++) vpid_free(vmx->state[i].vpid); free(vmx, M_VMX); return; } static register_t * vmxctx_regptr(struct vmxctx *vmxctx, int reg) { switch (reg) { case VM_REG_GUEST_RAX: return (&vmxctx->guest_rax); case VM_REG_GUEST_RBX: return (&vmxctx->guest_rbx); case VM_REG_GUEST_RCX: return (&vmxctx->guest_rcx); case VM_REG_GUEST_RDX: return (&vmxctx->guest_rdx); case VM_REG_GUEST_RSI: return (&vmxctx->guest_rsi); case VM_REG_GUEST_RDI: return (&vmxctx->guest_rdi); case VM_REG_GUEST_RBP: return (&vmxctx->guest_rbp); case VM_REG_GUEST_R8: return (&vmxctx->guest_r8); case VM_REG_GUEST_R9: return (&vmxctx->guest_r9); case VM_REG_GUEST_R10: return (&vmxctx->guest_r10); case VM_REG_GUEST_R11: return (&vmxctx->guest_r11); case VM_REG_GUEST_R12: return (&vmxctx->guest_r12); case VM_REG_GUEST_R13: return (&vmxctx->guest_r13); case VM_REG_GUEST_R14: return (&vmxctx->guest_r14); case VM_REG_GUEST_R15: return (&vmxctx->guest_r15); case VM_REG_GUEST_CR2: return (&vmxctx->guest_cr2); case VM_REG_GUEST_DR0: return (&vmxctx->guest_dr0); case VM_REG_GUEST_DR1: return (&vmxctx->guest_dr1); case VM_REG_GUEST_DR2: return (&vmxctx->guest_dr2); case VM_REG_GUEST_DR3: return (&vmxctx->guest_dr3); case VM_REG_GUEST_DR6: return (&vmxctx->guest_dr6); default: break; } return (NULL); } static int vmxctx_getreg(struct vmxctx *vmxctx, int reg, uint64_t *retval) { register_t *regp; if ((regp = vmxctx_regptr(vmxctx, reg)) != NULL) { *retval = *regp; return (0); } else return (EINVAL); } static int vmxctx_setreg(struct vmxctx *vmxctx, int reg, uint64_t val) { register_t *regp; if ((regp = vmxctx_regptr(vmxctx, reg)) != NULL) { *regp = val; return (0); } else return (EINVAL); } static int vmx_get_intr_shadow(struct vmx *vmx, int vcpu, int running, uint64_t *retval) { uint64_t gi; int error; error = vmcs_getreg(&vmx->vmcs[vcpu], running, VMCS_IDENT(VMCS_GUEST_INTERRUPTIBILITY), &gi); *retval = (gi & HWINTR_BLOCKING) ? 1 : 0; return (error); } static int vmx_modify_intr_shadow(struct vmx *vmx, int vcpu, int running, uint64_t val) { struct vmcs *vmcs; uint64_t gi; int error, ident; /* * Forcing the vcpu into an interrupt shadow is not supported. */ if (val) { error = EINVAL; goto done; } vmcs = &vmx->vmcs[vcpu]; ident = VMCS_IDENT(VMCS_GUEST_INTERRUPTIBILITY); error = vmcs_getreg(vmcs, running, ident, &gi); if (error == 0) { gi &= ~HWINTR_BLOCKING; error = vmcs_setreg(vmcs, running, ident, gi); } done: VCPU_CTR2(vmx->vm, vcpu, "Setting intr_shadow to %#lx %s", val, error ? "failed" : "succeeded"); return (error); } static int vmx_shadow_reg(int reg) { int shreg; shreg = -1; switch (reg) { case VM_REG_GUEST_CR0: shreg = VMCS_CR0_SHADOW; break; case VM_REG_GUEST_CR4: shreg = VMCS_CR4_SHADOW; break; default: break; } return (shreg); } static int vmx_getreg(void *arg, int vcpu, int reg, uint64_t *retval) { int running, hostcpu; struct vmx *vmx = arg; running = vcpu_is_running(vmx->vm, vcpu, &hostcpu); if (running && hostcpu != curcpu) panic("vmx_getreg: %s%d is running", vm_name(vmx->vm), vcpu); if (reg == VM_REG_GUEST_INTR_SHADOW) return (vmx_get_intr_shadow(vmx, vcpu, running, retval)); if (vmxctx_getreg(&vmx->ctx[vcpu], reg, retval) == 0) return (0); return (vmcs_getreg(&vmx->vmcs[vcpu], running, reg, retval)); } static int vmx_setreg(void *arg, int vcpu, int reg, uint64_t val) { int error, hostcpu, running, shadow; uint64_t ctls; pmap_t pmap; struct vmx *vmx = arg; running = vcpu_is_running(vmx->vm, vcpu, &hostcpu); if (running && hostcpu != curcpu) panic("vmx_setreg: %s%d is running", vm_name(vmx->vm), vcpu); if (reg == VM_REG_GUEST_INTR_SHADOW) return (vmx_modify_intr_shadow(vmx, vcpu, running, val)); if (vmxctx_setreg(&vmx->ctx[vcpu], reg, val) == 0) return (0); error = vmcs_setreg(&vmx->vmcs[vcpu], running, reg, val); if (error == 0) { /* * If the "load EFER" VM-entry control is 1 then the * value of EFER.LMA must be identical to "IA-32e mode guest" * bit in the VM-entry control. */ if ((entry_ctls & VM_ENTRY_LOAD_EFER) != 0 && (reg == VM_REG_GUEST_EFER)) { vmcs_getreg(&vmx->vmcs[vcpu], running, VMCS_IDENT(VMCS_ENTRY_CTLS), &ctls); if (val & EFER_LMA) ctls |= VM_ENTRY_GUEST_LMA; else ctls &= ~VM_ENTRY_GUEST_LMA; vmcs_setreg(&vmx->vmcs[vcpu], running, VMCS_IDENT(VMCS_ENTRY_CTLS), ctls); } shadow = vmx_shadow_reg(reg); if (shadow > 0) { /* * Store the unmodified value in the shadow */ error = vmcs_setreg(&vmx->vmcs[vcpu], running, VMCS_IDENT(shadow), val); } if (reg == VM_REG_GUEST_CR3) { /* * Invalidate the guest vcpu's TLB mappings to emulate * the behavior of updating %cr3. * * XXX the processor retains global mappings when %cr3 * is updated but vmx_invvpid() does not. */ pmap = vmx->ctx[vcpu].pmap; vmx_invvpid(vmx, vcpu, pmap, running); } } return (error); } static int vmx_getdesc(void *arg, int vcpu, int reg, struct seg_desc *desc) { int hostcpu, running; struct vmx *vmx = arg; running = vcpu_is_running(vmx->vm, vcpu, &hostcpu); if (running && hostcpu != curcpu) panic("vmx_getdesc: %s%d is running", vm_name(vmx->vm), vcpu); return (vmcs_getdesc(&vmx->vmcs[vcpu], running, reg, desc)); } static int vmx_setdesc(void *arg, int vcpu, int reg, struct seg_desc *desc) { int hostcpu, running; struct vmx *vmx = arg; running = vcpu_is_running(vmx->vm, vcpu, &hostcpu); if (running && hostcpu != curcpu) panic("vmx_setdesc: %s%d is running", vm_name(vmx->vm), vcpu); return (vmcs_setdesc(&vmx->vmcs[vcpu], running, reg, desc)); } static int vmx_getcap(void *arg, int vcpu, int type, int *retval) { struct vmx *vmx = arg; int vcap; int ret; ret = ENOENT; vcap = vmx->cap[vcpu].set; switch (type) { case VM_CAP_HALT_EXIT: if (cap_halt_exit) ret = 0; break; case VM_CAP_PAUSE_EXIT: if (cap_pause_exit) ret = 0; break; case VM_CAP_MTRAP_EXIT: if (cap_monitor_trap) ret = 0; break; case VM_CAP_UNRESTRICTED_GUEST: if (cap_unrestricted_guest) ret = 0; break; case VM_CAP_ENABLE_INVPCID: if (cap_invpcid) ret = 0; break; case VM_CAP_BPT_EXIT: ret = 0; break; default: break; } if (ret == 0) *retval = (vcap & (1 << type)) ? 1 : 0; return (ret); } static int vmx_setcap(void *arg, int vcpu, int type, int val) { struct vmx *vmx = arg; struct vmcs *vmcs = &vmx->vmcs[vcpu]; uint32_t baseval; uint32_t *pptr; int error; int flag; int reg; int retval; retval = ENOENT; pptr = NULL; switch (type) { case VM_CAP_HALT_EXIT: if (cap_halt_exit) { retval = 0; pptr = &vmx->cap[vcpu].proc_ctls; baseval = *pptr; flag = PROCBASED_HLT_EXITING; reg = VMCS_PRI_PROC_BASED_CTLS; } break; case VM_CAP_MTRAP_EXIT: if (cap_monitor_trap) { retval = 0; pptr = &vmx->cap[vcpu].proc_ctls; baseval = *pptr; flag = PROCBASED_MTF; reg = VMCS_PRI_PROC_BASED_CTLS; } break; case VM_CAP_PAUSE_EXIT: if (cap_pause_exit) { retval = 0; pptr = &vmx->cap[vcpu].proc_ctls; baseval = *pptr; flag = PROCBASED_PAUSE_EXITING; reg = VMCS_PRI_PROC_BASED_CTLS; } break; case VM_CAP_UNRESTRICTED_GUEST: if (cap_unrestricted_guest) { retval = 0; pptr = &vmx->cap[vcpu].proc_ctls2; baseval = *pptr; flag = PROCBASED2_UNRESTRICTED_GUEST; reg = VMCS_SEC_PROC_BASED_CTLS; } break; case VM_CAP_ENABLE_INVPCID: if (cap_invpcid) { retval = 0; pptr = &vmx->cap[vcpu].proc_ctls2; baseval = *pptr; flag = PROCBASED2_ENABLE_INVPCID; reg = VMCS_SEC_PROC_BASED_CTLS; } break; case VM_CAP_BPT_EXIT: retval = 0; /* Don't change the bitmap if we are tracing all exceptions. */ if (vmx->cap[vcpu].exc_bitmap != 0xffffffff) { pptr = &vmx->cap[vcpu].exc_bitmap; baseval = *pptr; flag = (1 << IDT_BP); reg = VMCS_EXCEPTION_BITMAP; } break; default: break; } if (retval) return (retval); if (pptr != NULL) { if (val) { baseval |= flag; } else { baseval &= ~flag; } VMPTRLD(vmcs); error = vmwrite(reg, baseval); VMCLEAR(vmcs); if (error) return (error); /* * Update optional stored flags, and record * setting */ *pptr = baseval; } if (val) { vmx->cap[vcpu].set |= (1 << type); } else { vmx->cap[vcpu].set &= ~(1 << type); } return (0); } struct vlapic_vtx { struct vlapic vlapic; struct pir_desc *pir_desc; struct vmx *vmx; u_int pending_prio; }; #define VPR_PRIO_BIT(vpr) (1 << ((vpr) >> 4)) #define VMX_CTR_PIR(vm, vcpuid, pir_desc, notify, vector, level, msg) \ do { \ VCPU_CTR2(vm, vcpuid, msg " assert %s-triggered vector %d", \ level ? "level" : "edge", vector); \ VCPU_CTR1(vm, vcpuid, msg " pir0 0x%016lx", pir_desc->pir[0]); \ VCPU_CTR1(vm, vcpuid, msg " pir1 0x%016lx", pir_desc->pir[1]); \ VCPU_CTR1(vm, vcpuid, msg " pir2 0x%016lx", pir_desc->pir[2]); \ VCPU_CTR1(vm, vcpuid, msg " pir3 0x%016lx", pir_desc->pir[3]); \ VCPU_CTR1(vm, vcpuid, msg " notify: %s", notify ? "yes" : "no");\ } while (0) /* * vlapic->ops handlers that utilize the APICv hardware assist described in * Chapter 29 of the Intel SDM. */ static int vmx_set_intr_ready(struct vlapic *vlapic, int vector, bool level) { struct vlapic_vtx *vlapic_vtx; struct pir_desc *pir_desc; uint64_t mask; int idx, notify = 0; vlapic_vtx = (struct vlapic_vtx *)vlapic; pir_desc = vlapic_vtx->pir_desc; /* * Keep track of interrupt requests in the PIR descriptor. This is * because the virtual APIC page pointed to by the VMCS cannot be * modified if the vcpu is running. */ idx = vector / 64; mask = 1UL << (vector % 64); atomic_set_long(&pir_desc->pir[idx], mask); /* * A notification is required whenever the 'pending' bit makes a * transition from 0->1. * * Even if the 'pending' bit is already asserted, notification about * the incoming interrupt may still be necessary. For example, if a * vCPU is HLTed with a high PPR, a low priority interrupt would cause * the 0->1 'pending' transition with a notification, but the vCPU * would ignore the interrupt for the time being. The same vCPU would * need to then be notified if a high-priority interrupt arrived which * satisfied the PPR. * * The priorities of interrupts injected while 'pending' is asserted * are tracked in a custom bitfield 'pending_prio'. Should the * to-be-injected interrupt exceed the priorities already present, the * notification is sent. The priorities recorded in 'pending_prio' are * cleared whenever the 'pending' bit makes another 0->1 transition. */ if (atomic_cmpset_long(&pir_desc->pending, 0, 1) != 0) { notify = 1; vlapic_vtx->pending_prio = 0; } else { const u_int old_prio = vlapic_vtx->pending_prio; const u_int prio_bit = VPR_PRIO_BIT(vector & APIC_TPR_INT); if ((old_prio & prio_bit) == 0 && prio_bit > old_prio) { atomic_set_int(&vlapic_vtx->pending_prio, prio_bit); notify = 1; } } VMX_CTR_PIR(vlapic->vm, vlapic->vcpuid, pir_desc, notify, vector, level, "vmx_set_intr_ready"); return (notify); } static int vmx_pending_intr(struct vlapic *vlapic, int *vecptr) { struct vlapic_vtx *vlapic_vtx; struct pir_desc *pir_desc; struct LAPIC *lapic; uint64_t pending, pirval; uint32_t ppr, vpr; int i; /* * This function is only expected to be called from the 'HLT' exit * handler which does not care about the vector that is pending. */ KASSERT(vecptr == NULL, ("vmx_pending_intr: vecptr must be NULL")); vlapic_vtx = (struct vlapic_vtx *)vlapic; pir_desc = vlapic_vtx->pir_desc; pending = atomic_load_acq_long(&pir_desc->pending); if (!pending) { /* * While a virtual interrupt may have already been * processed the actual delivery maybe pending the * interruptibility of the guest. Recognize a pending * interrupt by reevaluating virtual interrupts * following Section 29.2.1 in the Intel SDM Volume 3. */ struct vm_exit *vmexit; uint8_t rvi, ppr; vmexit = vm_exitinfo(vlapic->vm, vlapic->vcpuid); KASSERT(vmexit->exitcode == VM_EXITCODE_HLT, ("vmx_pending_intr: exitcode not 'HLT'")); rvi = vmexit->u.hlt.intr_status & APIC_TPR_INT; lapic = vlapic->apic_page; ppr = lapic->ppr & APIC_TPR_INT; if (rvi > ppr) { return (1); } return (0); } /* * If there is an interrupt pending then it will be recognized only * if its priority is greater than the processor priority. * * Special case: if the processor priority is zero then any pending * interrupt will be recognized. */ lapic = vlapic->apic_page; ppr = lapic->ppr & APIC_TPR_INT; if (ppr == 0) return (1); VCPU_CTR1(vlapic->vm, vlapic->vcpuid, "HLT with non-zero PPR %d", lapic->ppr); vpr = 0; for (i = 3; i >= 0; i--) { pirval = pir_desc->pir[i]; if (pirval != 0) { vpr = (i * 64 + flsl(pirval) - 1) & APIC_TPR_INT; break; } } /* * If the highest-priority pending interrupt falls short of the * processor priority of this vCPU, ensure that 'pending_prio' does not * have any stale bits which would preclude a higher-priority interrupt * from incurring a notification later. */ if (vpr <= ppr) { const u_int prio_bit = VPR_PRIO_BIT(vpr); const u_int old = vlapic_vtx->pending_prio; if (old > prio_bit && (old & prio_bit) == 0) { vlapic_vtx->pending_prio = prio_bit; } return (0); } return (1); } static void vmx_intr_accepted(struct vlapic *vlapic, int vector) { panic("vmx_intr_accepted: not expected to be called"); } static void vmx_set_tmr(struct vlapic *vlapic, int vector, bool level) { struct vlapic_vtx *vlapic_vtx; struct vmx *vmx; struct vmcs *vmcs; uint64_t mask, val; KASSERT(vector >= 0 && vector <= 255, ("invalid vector %d", vector)); KASSERT(!vcpu_is_running(vlapic->vm, vlapic->vcpuid, NULL), ("vmx_set_tmr: vcpu cannot be running")); vlapic_vtx = (struct vlapic_vtx *)vlapic; vmx = vlapic_vtx->vmx; vmcs = &vmx->vmcs[vlapic->vcpuid]; mask = 1UL << (vector % 64); VMPTRLD(vmcs); val = vmcs_read(VMCS_EOI_EXIT(vector)); if (level) val |= mask; else val &= ~mask; vmcs_write(VMCS_EOI_EXIT(vector), val); VMCLEAR(vmcs); } static void -vmx_enable_x2apic_mode(struct vlapic *vlapic) +vmx_enable_x2apic_mode_ts(struct vlapic *vlapic) { struct vmx *vmx; struct vmcs *vmcs; + uint32_t proc_ctls; + int vcpuid; + + vcpuid = vlapic->vcpuid; + vmx = ((struct vlapic_vtx *)vlapic)->vmx; + vmcs = &vmx->vmcs[vcpuid]; + + proc_ctls = vmx->cap[vcpuid].proc_ctls; + proc_ctls &= ~PROCBASED_USE_TPR_SHADOW; + proc_ctls |= PROCBASED_CR8_LOAD_EXITING; + proc_ctls |= PROCBASED_CR8_STORE_EXITING; + vmx->cap[vcpuid].proc_ctls = proc_ctls; + + VMPTRLD(vmcs); + vmcs_write(VMCS_PRI_PROC_BASED_CTLS, proc_ctls); + VMCLEAR(vmcs); +} + +static void +vmx_enable_x2apic_mode_vid(struct vlapic *vlapic) +{ + struct vmx *vmx; + struct vmcs *vmcs; uint32_t proc_ctls2; int vcpuid, error; vcpuid = vlapic->vcpuid; vmx = ((struct vlapic_vtx *)vlapic)->vmx; vmcs = &vmx->vmcs[vcpuid]; proc_ctls2 = vmx->cap[vcpuid].proc_ctls2; KASSERT((proc_ctls2 & PROCBASED2_VIRTUALIZE_APIC_ACCESSES) != 0, ("%s: invalid proc_ctls2 %#x", __func__, proc_ctls2)); proc_ctls2 &= ~PROCBASED2_VIRTUALIZE_APIC_ACCESSES; proc_ctls2 |= PROCBASED2_VIRTUALIZE_X2APIC_MODE; vmx->cap[vcpuid].proc_ctls2 = proc_ctls2; VMPTRLD(vmcs); vmcs_write(VMCS_SEC_PROC_BASED_CTLS, proc_ctls2); VMCLEAR(vmcs); if (vlapic->vcpuid == 0) { /* * The nested page table mappings are shared by all vcpus * so unmap the APIC access page just once. */ error = vm_unmap_mmio(vmx->vm, DEFAULT_APIC_BASE, PAGE_SIZE); KASSERT(error == 0, ("%s: vm_unmap_mmio error %d", __func__, error)); /* * The MSR bitmap is shared by all vcpus so modify it only * once in the context of vcpu 0. */ error = vmx_allow_x2apic_msrs(vmx); KASSERT(error == 0, ("%s: vmx_allow_x2apic_msrs error %d", __func__, error)); } } static void vmx_post_intr(struct vlapic *vlapic, int hostcpu) { ipi_cpu(hostcpu, pirvec); } /* * Transfer the pending interrupts in the PIR descriptor to the IRR * in the virtual APIC page. */ static void vmx_inject_pir(struct vlapic *vlapic) { struct vlapic_vtx *vlapic_vtx; struct pir_desc *pir_desc; struct LAPIC *lapic; uint64_t val, pirval; int rvi, pirbase = -1; uint16_t intr_status_old, intr_status_new; vlapic_vtx = (struct vlapic_vtx *)vlapic; pir_desc = vlapic_vtx->pir_desc; if (atomic_cmpset_long(&pir_desc->pending, 1, 0) == 0) { VCPU_CTR0(vlapic->vm, vlapic->vcpuid, "vmx_inject_pir: " "no posted interrupt pending"); return; } pirval = 0; pirbase = -1; lapic = vlapic->apic_page; val = atomic_readandclear_long(&pir_desc->pir[0]); if (val != 0) { lapic->irr0 |= val; lapic->irr1 |= val >> 32; pirbase = 0; pirval = val; } val = atomic_readandclear_long(&pir_desc->pir[1]); if (val != 0) { lapic->irr2 |= val; lapic->irr3 |= val >> 32; pirbase = 64; pirval = val; } val = atomic_readandclear_long(&pir_desc->pir[2]); if (val != 0) { lapic->irr4 |= val; lapic->irr5 |= val >> 32; pirbase = 128; pirval = val; } val = atomic_readandclear_long(&pir_desc->pir[3]); if (val != 0) { lapic->irr6 |= val; lapic->irr7 |= val >> 32; pirbase = 192; pirval = val; } VLAPIC_CTR_IRR(vlapic, "vmx_inject_pir"); /* * Update RVI so the processor can evaluate pending virtual * interrupts on VM-entry. * * It is possible for pirval to be 0 here, even though the * pending bit has been set. The scenario is: * CPU-Y is sending a posted interrupt to CPU-X, which * is running a guest and processing posted interrupts in h/w. * CPU-X will eventually exit and the state seen in s/w is * the pending bit set, but no PIR bits set. * * CPU-X CPU-Y * (vm running) (host running) * rx posted interrupt * CLEAR pending bit * SET PIR bit * READ/CLEAR PIR bits * SET pending bit * (vm exit) * pending bit set, PIR 0 */ if (pirval != 0) { rvi = pirbase + flsl(pirval) - 1; intr_status_old = vmcs_read(VMCS_GUEST_INTR_STATUS); intr_status_new = (intr_status_old & 0xFF00) | rvi; if (intr_status_new > intr_status_old) { vmcs_write(VMCS_GUEST_INTR_STATUS, intr_status_new); VCPU_CTR2(vlapic->vm, vlapic->vcpuid, "vmx_inject_pir: " "guest_intr_status changed from 0x%04x to 0x%04x", intr_status_old, intr_status_new); } } } static struct vlapic * vmx_vlapic_init(void *arg, int vcpuid) { struct vmx *vmx; struct vlapic *vlapic; struct vlapic_vtx *vlapic_vtx; vmx = arg; vlapic = malloc(sizeof(struct vlapic_vtx), M_VLAPIC, M_WAITOK | M_ZERO); vlapic->vm = vmx->vm; vlapic->vcpuid = vcpuid; vlapic->apic_page = (struct LAPIC *)&vmx->apic_page[vcpuid]; vlapic_vtx = (struct vlapic_vtx *)vlapic; vlapic_vtx->pir_desc = &vmx->pir_desc[vcpuid]; vlapic_vtx->vmx = vmx; + if (tpr_shadowing) { + vlapic->ops.enable_x2apic_mode = vmx_enable_x2apic_mode_ts; + } + if (virtual_interrupt_delivery) { vlapic->ops.set_intr_ready = vmx_set_intr_ready; vlapic->ops.pending_intr = vmx_pending_intr; vlapic->ops.intr_accepted = vmx_intr_accepted; vlapic->ops.set_tmr = vmx_set_tmr; - vlapic->ops.enable_x2apic_mode = vmx_enable_x2apic_mode; + vlapic->ops.enable_x2apic_mode = vmx_enable_x2apic_mode_vid; } if (posted_interrupts) vlapic->ops.post_intr = vmx_post_intr; vlapic_init(vlapic); return (vlapic); } static void vmx_vlapic_cleanup(void *arg, struct vlapic *vlapic) { vlapic_cleanup(vlapic); free(vlapic, M_VLAPIC); } struct vmm_ops vmm_ops_intel = { .init = vmx_init, .cleanup = vmx_cleanup, .resume = vmx_restore, .vminit = vmx_vminit, .vmrun = vmx_run, .vmcleanup = vmx_vmcleanup, .vmgetreg = vmx_getreg, .vmsetreg = vmx_setreg, .vmgetdesc = vmx_getdesc, .vmsetdesc = vmx_setdesc, .vmgetcap = vmx_getcap, .vmsetcap = vmx_setcap, .vmspace_alloc = ept_vmspace_alloc, .vmspace_free = ept_vmspace_free, .vlapic_init = vmx_vlapic_init, .vlapic_cleanup = vmx_vlapic_cleanup, }; Index: projects/clang1000-import/sys/amd64/vmm/io/vlapic.c =================================================================== --- projects/clang1000-import/sys/amd64/vmm/io/vlapic.c (revision 358848) +++ projects/clang1000-import/sys/amd64/vmm/io/vlapic.c (revision 358849) @@ -1,1645 +1,1652 @@ /*- * SPDX-License-Identifier: BSD-2-Clause-FreeBSD * * Copyright (c) 2011 NetApp, Inc. * All rights reserved. * Copyright (c) 2019 Joyent, Inc. * * 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 NETAPP, INC ``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 NETAPP, INC 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 __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include #include "vmm_lapic.h" #include "vmm_ktr.h" #include "vmm_stat.h" #include "vlapic.h" #include "vlapic_priv.h" #include "vioapic.h" #define PRIO(x) ((x) >> 4) #define VLAPIC_VERSION (16) #define x2apic(vlapic) (((vlapic)->msr_apicbase & APICBASE_X2APIC) ? 1 : 0) /* * The 'vlapic->timer_mtx' is used to provide mutual exclusion between the * vlapic_callout_handler() and vcpu accesses to: * - timer_freq_bt, timer_period_bt, timer_fire_bt * - timer LVT register */ #define VLAPIC_TIMER_LOCK(vlapic) mtx_lock_spin(&((vlapic)->timer_mtx)) #define VLAPIC_TIMER_UNLOCK(vlapic) mtx_unlock_spin(&((vlapic)->timer_mtx)) #define VLAPIC_TIMER_LOCKED(vlapic) mtx_owned(&((vlapic)->timer_mtx)) /* * APIC timer frequency: * - arbitrary but chosen to be in the ballpark of contemporary hardware. * - power-of-two to avoid loss of precision when converted to a bintime. */ #define VLAPIC_BUS_FREQ (128 * 1024 * 1024) static void vlapic_set_error(struct vlapic *, uint32_t, bool); static __inline uint32_t vlapic_get_id(struct vlapic *vlapic) { if (x2apic(vlapic)) return (vlapic->vcpuid); else return (vlapic->vcpuid << 24); } static uint32_t x2apic_ldr(struct vlapic *vlapic) { int apicid; uint32_t ldr; apicid = vlapic_get_id(vlapic); ldr = 1 << (apicid & 0xf); ldr |= (apicid & 0xffff0) << 12; return (ldr); } void vlapic_dfr_write_handler(struct vlapic *vlapic) { struct LAPIC *lapic; lapic = vlapic->apic_page; if (x2apic(vlapic)) { VM_CTR1(vlapic->vm, "ignoring write to DFR in x2apic mode: %#x", lapic->dfr); lapic->dfr = 0; return; } lapic->dfr &= APIC_DFR_MODEL_MASK; lapic->dfr |= APIC_DFR_RESERVED; if ((lapic->dfr & APIC_DFR_MODEL_MASK) == APIC_DFR_MODEL_FLAT) VLAPIC_CTR0(vlapic, "vlapic DFR in Flat Model"); else if ((lapic->dfr & APIC_DFR_MODEL_MASK) == APIC_DFR_MODEL_CLUSTER) VLAPIC_CTR0(vlapic, "vlapic DFR in Cluster Model"); else VLAPIC_CTR1(vlapic, "DFR in Unknown Model %#x", lapic->dfr); } void vlapic_ldr_write_handler(struct vlapic *vlapic) { struct LAPIC *lapic; lapic = vlapic->apic_page; /* LDR is read-only in x2apic mode */ if (x2apic(vlapic)) { VLAPIC_CTR1(vlapic, "ignoring write to LDR in x2apic mode: %#x", lapic->ldr); lapic->ldr = x2apic_ldr(vlapic); } else { lapic->ldr &= ~APIC_LDR_RESERVED; VLAPIC_CTR1(vlapic, "vlapic LDR set to %#x", lapic->ldr); } } void vlapic_id_write_handler(struct vlapic *vlapic) { struct LAPIC *lapic; /* * We don't allow the ID register to be modified so reset it back to * its default value. */ lapic = vlapic->apic_page; lapic->id = vlapic_get_id(vlapic); } static int vlapic_timer_divisor(uint32_t dcr) { switch (dcr & 0xB) { case APIC_TDCR_1: return (1); case APIC_TDCR_2: return (2); case APIC_TDCR_4: return (4); case APIC_TDCR_8: return (8); case APIC_TDCR_16: return (16); case APIC_TDCR_32: return (32); case APIC_TDCR_64: return (64); case APIC_TDCR_128: return (128); default: panic("vlapic_timer_divisor: invalid dcr 0x%08x", dcr); } } #if 0 static inline void vlapic_dump_lvt(uint32_t offset, uint32_t *lvt) { printf("Offset %x: lvt %08x (V:%02x DS:%x M:%x)\n", offset, *lvt, *lvt & APIC_LVTT_VECTOR, *lvt & APIC_LVTT_DS, *lvt & APIC_LVTT_M); } #endif static uint32_t vlapic_get_ccr(struct vlapic *vlapic) { struct bintime bt_now, bt_rem; struct LAPIC *lapic; uint32_t ccr; ccr = 0; lapic = vlapic->apic_page; VLAPIC_TIMER_LOCK(vlapic); if (callout_active(&vlapic->callout)) { /* * If the timer is scheduled to expire in the future then * compute the value of 'ccr' based on the remaining time. */ binuptime(&bt_now); if (bintime_cmp(&vlapic->timer_fire_bt, &bt_now, >)) { bt_rem = vlapic->timer_fire_bt; bintime_sub(&bt_rem, &bt_now); ccr += bt_rem.sec * BT2FREQ(&vlapic->timer_freq_bt); ccr += bt_rem.frac / vlapic->timer_freq_bt.frac; } } KASSERT(ccr <= lapic->icr_timer, ("vlapic_get_ccr: invalid ccr %#x, " "icr_timer is %#x", ccr, lapic->icr_timer)); VLAPIC_CTR2(vlapic, "vlapic ccr_timer = %#x, icr_timer = %#x", ccr, lapic->icr_timer); VLAPIC_TIMER_UNLOCK(vlapic); return (ccr); } void vlapic_dcr_write_handler(struct vlapic *vlapic) { struct LAPIC *lapic; int divisor; lapic = vlapic->apic_page; VLAPIC_TIMER_LOCK(vlapic); divisor = vlapic_timer_divisor(lapic->dcr_timer); VLAPIC_CTR2(vlapic, "vlapic dcr_timer=%#x, divisor=%d", lapic->dcr_timer, divisor); /* * Update the timer frequency and the timer period. * * XXX changes to the frequency divider will not take effect until * the timer is reloaded. */ FREQ2BT(VLAPIC_BUS_FREQ / divisor, &vlapic->timer_freq_bt); vlapic->timer_period_bt = vlapic->timer_freq_bt; bintime_mul(&vlapic->timer_period_bt, lapic->icr_timer); VLAPIC_TIMER_UNLOCK(vlapic); } void vlapic_esr_write_handler(struct vlapic *vlapic) { struct LAPIC *lapic; lapic = vlapic->apic_page; lapic->esr = vlapic->esr_pending; vlapic->esr_pending = 0; } int vlapic_set_intr_ready(struct vlapic *vlapic, int vector, bool level) { struct LAPIC *lapic; uint32_t *irrptr, *tmrptr, mask; int idx; KASSERT(vector >= 0 && vector < 256, ("invalid vector %d", vector)); lapic = vlapic->apic_page; if (!(lapic->svr & APIC_SVR_ENABLE)) { VLAPIC_CTR1(vlapic, "vlapic is software disabled, ignoring " "interrupt %d", vector); return (0); } if (vector < 16) { vlapic_set_error(vlapic, APIC_ESR_RECEIVE_ILLEGAL_VECTOR, false); VLAPIC_CTR1(vlapic, "vlapic ignoring interrupt to vector %d", vector); return (1); } if (vlapic->ops.set_intr_ready) return ((*vlapic->ops.set_intr_ready)(vlapic, vector, level)); idx = (vector / 32) * 4; mask = 1 << (vector % 32); irrptr = &lapic->irr0; atomic_set_int(&irrptr[idx], mask); /* * Verify that the trigger-mode of the interrupt matches with * the vlapic TMR registers. */ tmrptr = &lapic->tmr0; if ((tmrptr[idx] & mask) != (level ? mask : 0)) { VLAPIC_CTR3(vlapic, "vlapic TMR[%d] is 0x%08x but " "interrupt is %s-triggered", idx / 4, tmrptr[idx], level ? "level" : "edge"); } VLAPIC_CTR_IRR(vlapic, "vlapic_set_intr_ready"); return (1); } static __inline uint32_t * vlapic_get_lvtptr(struct vlapic *vlapic, uint32_t offset) { struct LAPIC *lapic = vlapic->apic_page; int i; switch (offset) { case APIC_OFFSET_CMCI_LVT: return (&lapic->lvt_cmci); case APIC_OFFSET_TIMER_LVT ... APIC_OFFSET_ERROR_LVT: i = (offset - APIC_OFFSET_TIMER_LVT) >> 2; return ((&lapic->lvt_timer) + i); default: panic("vlapic_get_lvt: invalid LVT\n"); } } static __inline int lvt_off_to_idx(uint32_t offset) { int index; switch (offset) { case APIC_OFFSET_CMCI_LVT: index = APIC_LVT_CMCI; break; case APIC_OFFSET_TIMER_LVT: index = APIC_LVT_TIMER; break; case APIC_OFFSET_THERM_LVT: index = APIC_LVT_THERMAL; break; case APIC_OFFSET_PERF_LVT: index = APIC_LVT_PMC; break; case APIC_OFFSET_LINT0_LVT: index = APIC_LVT_LINT0; break; case APIC_OFFSET_LINT1_LVT: index = APIC_LVT_LINT1; break; case APIC_OFFSET_ERROR_LVT: index = APIC_LVT_ERROR; break; default: index = -1; break; } KASSERT(index >= 0 && index <= VLAPIC_MAXLVT_INDEX, ("lvt_off_to_idx: " "invalid lvt index %d for offset %#x", index, offset)); return (index); } static __inline uint32_t vlapic_get_lvt(struct vlapic *vlapic, uint32_t offset) { int idx; uint32_t val; idx = lvt_off_to_idx(offset); val = atomic_load_acq_32(&vlapic->lvt_last[idx]); return (val); } void vlapic_lvt_write_handler(struct vlapic *vlapic, uint32_t offset) { uint32_t *lvtptr, mask, val; struct LAPIC *lapic; int idx; lapic = vlapic->apic_page; lvtptr = vlapic_get_lvtptr(vlapic, offset); val = *lvtptr; idx = lvt_off_to_idx(offset); if (!(lapic->svr & APIC_SVR_ENABLE)) val |= APIC_LVT_M; mask = APIC_LVT_M | APIC_LVT_DS | APIC_LVT_VECTOR; switch (offset) { case APIC_OFFSET_TIMER_LVT: mask |= APIC_LVTT_TM; break; case APIC_OFFSET_ERROR_LVT: break; case APIC_OFFSET_LINT0_LVT: case APIC_OFFSET_LINT1_LVT: mask |= APIC_LVT_TM | APIC_LVT_RIRR | APIC_LVT_IIPP; /* FALLTHROUGH */ default: mask |= APIC_LVT_DM; break; } val &= mask; *lvtptr = val; atomic_store_rel_32(&vlapic->lvt_last[idx], val); } static void vlapic_mask_lvts(struct vlapic *vlapic) { struct LAPIC *lapic = vlapic->apic_page; lapic->lvt_cmci |= APIC_LVT_M; vlapic_lvt_write_handler(vlapic, APIC_OFFSET_CMCI_LVT); lapic->lvt_timer |= APIC_LVT_M; vlapic_lvt_write_handler(vlapic, APIC_OFFSET_TIMER_LVT); lapic->lvt_thermal |= APIC_LVT_M; vlapic_lvt_write_handler(vlapic, APIC_OFFSET_THERM_LVT); lapic->lvt_pcint |= APIC_LVT_M; vlapic_lvt_write_handler(vlapic, APIC_OFFSET_PERF_LVT); lapic->lvt_lint0 |= APIC_LVT_M; vlapic_lvt_write_handler(vlapic, APIC_OFFSET_LINT0_LVT); lapic->lvt_lint1 |= APIC_LVT_M; vlapic_lvt_write_handler(vlapic, APIC_OFFSET_LINT1_LVT); lapic->lvt_error |= APIC_LVT_M; vlapic_lvt_write_handler(vlapic, APIC_OFFSET_ERROR_LVT); } static int vlapic_fire_lvt(struct vlapic *vlapic, u_int lvt) { uint32_t mode, reg, vec; reg = atomic_load_acq_32(&vlapic->lvt_last[lvt]); if (reg & APIC_LVT_M) return (0); vec = reg & APIC_LVT_VECTOR; mode = reg & APIC_LVT_DM; switch (mode) { case APIC_LVT_DM_FIXED: if (vec < 16) { vlapic_set_error(vlapic, APIC_ESR_SEND_ILLEGAL_VECTOR, lvt == APIC_LVT_ERROR); return (0); } if (vlapic_set_intr_ready(vlapic, vec, false)) vcpu_notify_event(vlapic->vm, vlapic->vcpuid, true); break; case APIC_LVT_DM_NMI: vm_inject_nmi(vlapic->vm, vlapic->vcpuid); break; case APIC_LVT_DM_EXTINT: vm_inject_extint(vlapic->vm, vlapic->vcpuid); break; default: // Other modes ignored return (0); } return (1); } #if 1 static void dump_isrvec_stk(struct vlapic *vlapic) { int i; uint32_t *isrptr; isrptr = &vlapic->apic_page->isr0; for (i = 0; i < 8; i++) printf("ISR%d 0x%08x\n", i, isrptr[i * 4]); for (i = 0; i <= vlapic->isrvec_stk_top; i++) printf("isrvec_stk[%d] = %d\n", i, vlapic->isrvec_stk[i]); } #endif /* * Algorithm adopted from section "Interrupt, Task and Processor Priority" * in Intel Architecture Manual Vol 3a. */ static void vlapic_update_ppr(struct vlapic *vlapic) { int isrvec, tpr, ppr; /* * Note that the value on the stack at index 0 is always 0. * * This is a placeholder for the value of ISRV when none of the * bits is set in the ISRx registers. */ isrvec = vlapic->isrvec_stk[vlapic->isrvec_stk_top]; tpr = vlapic->apic_page->tpr; #if 1 { int i, lastprio, curprio, vector, idx; uint32_t *isrptr; if (vlapic->isrvec_stk_top == 0 && isrvec != 0) panic("isrvec_stk is corrupted: %d", isrvec); /* * Make sure that the priority of the nested interrupts is * always increasing. */ lastprio = -1; for (i = 1; i <= vlapic->isrvec_stk_top; i++) { curprio = PRIO(vlapic->isrvec_stk[i]); if (curprio <= lastprio) { dump_isrvec_stk(vlapic); panic("isrvec_stk does not satisfy invariant"); } lastprio = curprio; } /* * Make sure that each bit set in the ISRx registers has a * corresponding entry on the isrvec stack. */ i = 1; isrptr = &vlapic->apic_page->isr0; for (vector = 0; vector < 256; vector++) { idx = (vector / 32) * 4; if (isrptr[idx] & (1 << (vector % 32))) { if (i > vlapic->isrvec_stk_top || vlapic->isrvec_stk[i] != vector) { dump_isrvec_stk(vlapic); panic("ISR and isrvec_stk out of sync"); } i++; } } } #endif if (PRIO(tpr) >= PRIO(isrvec)) ppr = tpr; else ppr = isrvec & 0xf0; vlapic->apic_page->ppr = ppr; VLAPIC_CTR1(vlapic, "vlapic_update_ppr 0x%02x", ppr); } +void +vlapic_sync_tpr(struct vlapic *vlapic) +{ + vlapic_update_ppr(vlapic); +} + static VMM_STAT(VLAPIC_GRATUITOUS_EOI, "EOI without any in-service interrupt"); static void vlapic_process_eoi(struct vlapic *vlapic) { struct LAPIC *lapic = vlapic->apic_page; uint32_t *isrptr, *tmrptr; int i, idx, bitpos, vector; isrptr = &lapic->isr0; tmrptr = &lapic->tmr0; for (i = 7; i >= 0; i--) { idx = i * 4; bitpos = fls(isrptr[idx]); if (bitpos-- != 0) { if (vlapic->isrvec_stk_top <= 0) { panic("invalid vlapic isrvec_stk_top %d", vlapic->isrvec_stk_top); } isrptr[idx] &= ~(1 << bitpos); vector = i * 32 + bitpos; VCPU_CTR1(vlapic->vm, vlapic->vcpuid, "EOI vector %d", vector); VLAPIC_CTR_ISR(vlapic, "vlapic_process_eoi"); vlapic->isrvec_stk_top--; vlapic_update_ppr(vlapic); if ((tmrptr[idx] & (1 << bitpos)) != 0) { vioapic_process_eoi(vlapic->vm, vlapic->vcpuid, vector); } return; } } VCPU_CTR0(vlapic->vm, vlapic->vcpuid, "Gratuitous EOI"); vmm_stat_incr(vlapic->vm, vlapic->vcpuid, VLAPIC_GRATUITOUS_EOI, 1); } static __inline int vlapic_get_lvt_field(uint32_t lvt, uint32_t mask) { return (lvt & mask); } static __inline int vlapic_periodic_timer(struct vlapic *vlapic) { uint32_t lvt; lvt = vlapic_get_lvt(vlapic, APIC_OFFSET_TIMER_LVT); return (vlapic_get_lvt_field(lvt, APIC_LVTT_TM_PERIODIC)); } static VMM_STAT(VLAPIC_INTR_ERROR, "error interrupts generated by vlapic"); static void vlapic_set_error(struct vlapic *vlapic, uint32_t mask, bool lvt_error) { vlapic->esr_pending |= mask; /* * Avoid infinite recursion if the error LVT itself is configured with * an illegal vector. */ if (lvt_error) return; if (vlapic_fire_lvt(vlapic, APIC_LVT_ERROR)) { vmm_stat_incr(vlapic->vm, vlapic->vcpuid, VLAPIC_INTR_ERROR, 1); } } static VMM_STAT(VLAPIC_INTR_TIMER, "timer interrupts generated by vlapic"); static void vlapic_fire_timer(struct vlapic *vlapic) { KASSERT(VLAPIC_TIMER_LOCKED(vlapic), ("vlapic_fire_timer not locked")); if (vlapic_fire_lvt(vlapic, APIC_LVT_TIMER)) { VLAPIC_CTR0(vlapic, "vlapic timer fired"); vmm_stat_incr(vlapic->vm, vlapic->vcpuid, VLAPIC_INTR_TIMER, 1); } } static VMM_STAT(VLAPIC_INTR_CMC, "corrected machine check interrupts generated by vlapic"); void vlapic_fire_cmci(struct vlapic *vlapic) { if (vlapic_fire_lvt(vlapic, APIC_LVT_CMCI)) { vmm_stat_incr(vlapic->vm, vlapic->vcpuid, VLAPIC_INTR_CMC, 1); } } static VMM_STAT_ARRAY(LVTS_TRIGGERRED, VLAPIC_MAXLVT_INDEX + 1, "lvts triggered"); int vlapic_trigger_lvt(struct vlapic *vlapic, int vector) { if (vlapic_enabled(vlapic) == false) { /* * When the local APIC is global/hardware disabled, * LINT[1:0] pins are configured as INTR and NMI pins, * respectively. */ switch (vector) { case APIC_LVT_LINT0: vm_inject_extint(vlapic->vm, vlapic->vcpuid); break; case APIC_LVT_LINT1: vm_inject_nmi(vlapic->vm, vlapic->vcpuid); break; default: break; } return (0); } switch (vector) { case APIC_LVT_LINT0: case APIC_LVT_LINT1: case APIC_LVT_TIMER: case APIC_LVT_ERROR: case APIC_LVT_PMC: case APIC_LVT_THERMAL: case APIC_LVT_CMCI: if (vlapic_fire_lvt(vlapic, vector)) { vmm_stat_array_incr(vlapic->vm, vlapic->vcpuid, LVTS_TRIGGERRED, vector, 1); } break; default: return (EINVAL); } return (0); } static void vlapic_callout_handler(void *arg) { struct vlapic *vlapic; struct bintime bt, btnow; sbintime_t rem_sbt; vlapic = arg; VLAPIC_TIMER_LOCK(vlapic); if (callout_pending(&vlapic->callout)) /* callout was reset */ goto done; if (!callout_active(&vlapic->callout)) /* callout was stopped */ goto done; callout_deactivate(&vlapic->callout); vlapic_fire_timer(vlapic); if (vlapic_periodic_timer(vlapic)) { binuptime(&btnow); KASSERT(bintime_cmp(&btnow, &vlapic->timer_fire_bt, >=), ("vlapic callout at %#lx.%#lx, expected at %#lx.#%lx", btnow.sec, btnow.frac, vlapic->timer_fire_bt.sec, vlapic->timer_fire_bt.frac)); /* * Compute the delta between when the timer was supposed to * fire and the present time. */ bt = btnow; bintime_sub(&bt, &vlapic->timer_fire_bt); rem_sbt = bttosbt(vlapic->timer_period_bt); if (bintime_cmp(&bt, &vlapic->timer_period_bt, <)) { /* * Adjust the time until the next countdown downward * to account for the lost time. */ rem_sbt -= bttosbt(bt); } else { /* * If the delta is greater than the timer period then * just reset our time base instead of trying to catch * up. */ vlapic->timer_fire_bt = btnow; VLAPIC_CTR2(vlapic, "vlapic timer lagging by %lu " "usecs, period is %lu usecs - resetting time base", bttosbt(bt) / SBT_1US, bttosbt(vlapic->timer_period_bt) / SBT_1US); } bintime_add(&vlapic->timer_fire_bt, &vlapic->timer_period_bt); callout_reset_sbt(&vlapic->callout, rem_sbt, 0, vlapic_callout_handler, vlapic, 0); } done: VLAPIC_TIMER_UNLOCK(vlapic); } void vlapic_icrtmr_write_handler(struct vlapic *vlapic) { struct LAPIC *lapic; sbintime_t sbt; uint32_t icr_timer; VLAPIC_TIMER_LOCK(vlapic); lapic = vlapic->apic_page; icr_timer = lapic->icr_timer; vlapic->timer_period_bt = vlapic->timer_freq_bt; bintime_mul(&vlapic->timer_period_bt, icr_timer); if (icr_timer != 0) { binuptime(&vlapic->timer_fire_bt); bintime_add(&vlapic->timer_fire_bt, &vlapic->timer_period_bt); sbt = bttosbt(vlapic->timer_period_bt); callout_reset_sbt(&vlapic->callout, sbt, 0, vlapic_callout_handler, vlapic, 0); } else callout_stop(&vlapic->callout); VLAPIC_TIMER_UNLOCK(vlapic); } /* * This function populates 'dmask' with the set of vcpus that match the * addressing specified by the (dest, phys, lowprio) tuple. * * 'x2apic_dest' specifies whether 'dest' is interpreted as x2APIC (32-bit) * or xAPIC (8-bit) destination field. */ static void vlapic_calcdest(struct vm *vm, cpuset_t *dmask, uint32_t dest, bool phys, bool lowprio, bool x2apic_dest) { struct vlapic *vlapic; uint32_t dfr, ldr, ldest, cluster; uint32_t mda_flat_ldest, mda_cluster_ldest, mda_ldest, mda_cluster_id; cpuset_t amask; int vcpuid; if ((x2apic_dest && dest == 0xffffffff) || (!x2apic_dest && dest == 0xff)) { /* * Broadcast in both logical and physical modes. */ *dmask = vm_active_cpus(vm); return; } if (phys) { /* * Physical mode: destination is APIC ID. */ CPU_ZERO(dmask); vcpuid = vm_apicid2vcpuid(vm, dest); amask = vm_active_cpus(vm); if (vcpuid < vm_get_maxcpus(vm) && CPU_ISSET(vcpuid, &amask)) CPU_SET(vcpuid, dmask); } else { /* * In the "Flat Model" the MDA is interpreted as an 8-bit wide * bitmask. This model is only available in the xAPIC mode. */ mda_flat_ldest = dest & 0xff; /* * In the "Cluster Model" the MDA is used to identify a * specific cluster and a set of APICs in that cluster. */ if (x2apic_dest) { mda_cluster_id = dest >> 16; mda_cluster_ldest = dest & 0xffff; } else { mda_cluster_id = (dest >> 4) & 0xf; mda_cluster_ldest = dest & 0xf; } /* * Logical mode: match each APIC that has a bit set * in its LDR that matches a bit in the ldest. */ CPU_ZERO(dmask); amask = vm_active_cpus(vm); while ((vcpuid = CPU_FFS(&amask)) != 0) { vcpuid--; CPU_CLR(vcpuid, &amask); vlapic = vm_lapic(vm, vcpuid); dfr = vlapic->apic_page->dfr; ldr = vlapic->apic_page->ldr; if ((dfr & APIC_DFR_MODEL_MASK) == APIC_DFR_MODEL_FLAT) { ldest = ldr >> 24; mda_ldest = mda_flat_ldest; } else if ((dfr & APIC_DFR_MODEL_MASK) == APIC_DFR_MODEL_CLUSTER) { if (x2apic(vlapic)) { cluster = ldr >> 16; ldest = ldr & 0xffff; } else { cluster = ldr >> 28; ldest = (ldr >> 24) & 0xf; } if (cluster != mda_cluster_id) continue; mda_ldest = mda_cluster_ldest; } else { /* * Guest has configured a bad logical * model for this vcpu - skip it. */ VLAPIC_CTR1(vlapic, "vlapic has bad logical " "model %x - cannot deliver interrupt", dfr); continue; } if ((mda_ldest & ldest) != 0) { CPU_SET(vcpuid, dmask); if (lowprio) break; } } } } static VMM_STAT_ARRAY(IPIS_SENT, VM_MAXCPU, "ipis sent to vcpu"); static void vlapic_set_tpr(struct vlapic *vlapic, uint8_t val) { struct LAPIC *lapic = vlapic->apic_page; if (lapic->tpr != val) { VCPU_CTR2(vlapic->vm, vlapic->vcpuid, "vlapic TPR changed " "from %#x to %#x", lapic->tpr, val); lapic->tpr = val; vlapic_update_ppr(vlapic); } } static uint8_t vlapic_get_tpr(struct vlapic *vlapic) { struct LAPIC *lapic = vlapic->apic_page; return (lapic->tpr); } void vlapic_set_cr8(struct vlapic *vlapic, uint64_t val) { uint8_t tpr; if (val & ~0xf) { vm_inject_gp(vlapic->vm, vlapic->vcpuid); return; } tpr = val << 4; vlapic_set_tpr(vlapic, tpr); } uint64_t vlapic_get_cr8(struct vlapic *vlapic) { uint8_t tpr; tpr = vlapic_get_tpr(vlapic); return (tpr >> 4); } int vlapic_icrlo_write_handler(struct vlapic *vlapic, bool *retu) { int i; bool phys; cpuset_t dmask; uint64_t icrval; uint32_t dest, vec, mode; struct vlapic *vlapic2; struct vm_exit *vmexit; struct LAPIC *lapic; uint16_t maxcpus; lapic = vlapic->apic_page; lapic->icr_lo &= ~APIC_DELSTAT_PEND; icrval = ((uint64_t)lapic->icr_hi << 32) | lapic->icr_lo; if (x2apic(vlapic)) dest = icrval >> 32; else dest = icrval >> (32 + 24); vec = icrval & APIC_VECTOR_MASK; mode = icrval & APIC_DELMODE_MASK; if (mode == APIC_DELMODE_FIXED && vec < 16) { vlapic_set_error(vlapic, APIC_ESR_SEND_ILLEGAL_VECTOR, false); VLAPIC_CTR1(vlapic, "Ignoring invalid IPI %d", vec); return (0); } VLAPIC_CTR2(vlapic, "icrlo 0x%016lx triggered ipi %d", icrval, vec); if (mode == APIC_DELMODE_FIXED || mode == APIC_DELMODE_NMI) { switch (icrval & APIC_DEST_MASK) { case APIC_DEST_DESTFLD: phys = ((icrval & APIC_DESTMODE_LOG) == 0); vlapic_calcdest(vlapic->vm, &dmask, dest, phys, false, x2apic(vlapic)); break; case APIC_DEST_SELF: CPU_SETOF(vlapic->vcpuid, &dmask); break; case APIC_DEST_ALLISELF: dmask = vm_active_cpus(vlapic->vm); break; case APIC_DEST_ALLESELF: dmask = vm_active_cpus(vlapic->vm); CPU_CLR(vlapic->vcpuid, &dmask); break; default: CPU_ZERO(&dmask); /* satisfy gcc */ break; } while ((i = CPU_FFS(&dmask)) != 0) { i--; CPU_CLR(i, &dmask); if (mode == APIC_DELMODE_FIXED) { lapic_intr_edge(vlapic->vm, i, vec); vmm_stat_array_incr(vlapic->vm, vlapic->vcpuid, IPIS_SENT, i, 1); VLAPIC_CTR2(vlapic, "vlapic sending ipi %d " "to vcpuid %d", vec, i); } else { vm_inject_nmi(vlapic->vm, i); VLAPIC_CTR1(vlapic, "vlapic sending ipi nmi " "to vcpuid %d", i); } } return (0); /* handled completely in the kernel */ } maxcpus = vm_get_maxcpus(vlapic->vm); if (mode == APIC_DELMODE_INIT) { if ((icrval & APIC_LEVEL_MASK) == APIC_LEVEL_DEASSERT) return (0); if (vlapic->vcpuid == 0 && dest != 0 && dest < maxcpus) { vlapic2 = vm_lapic(vlapic->vm, dest); /* move from INIT to waiting-for-SIPI state */ if (vlapic2->boot_state == BS_INIT) { vlapic2->boot_state = BS_SIPI; } return (0); } } if (mode == APIC_DELMODE_STARTUP) { if (vlapic->vcpuid == 0 && dest != 0 && dest < maxcpus) { vlapic2 = vm_lapic(vlapic->vm, dest); /* * Ignore SIPIs in any state other than wait-for-SIPI */ if (vlapic2->boot_state != BS_SIPI) return (0); vlapic2->boot_state = BS_RUNNING; *retu = true; vmexit = vm_exitinfo(vlapic->vm, vlapic->vcpuid); vmexit->exitcode = VM_EXITCODE_SPINUP_AP; vmexit->u.spinup_ap.vcpu = dest; vmexit->u.spinup_ap.rip = vec << PAGE_SHIFT; return (0); } } /* * This will cause a return to userland. */ return (1); } void vlapic_self_ipi_handler(struct vlapic *vlapic, uint64_t val) { int vec; KASSERT(x2apic(vlapic), ("SELF_IPI does not exist in xAPIC mode")); vec = val & 0xff; lapic_intr_edge(vlapic->vm, vlapic->vcpuid, vec); vmm_stat_array_incr(vlapic->vm, vlapic->vcpuid, IPIS_SENT, vlapic->vcpuid, 1); VLAPIC_CTR1(vlapic, "vlapic self-ipi %d", vec); } int vlapic_pending_intr(struct vlapic *vlapic, int *vecptr) { struct LAPIC *lapic = vlapic->apic_page; int idx, i, bitpos, vector; uint32_t *irrptr, val; + vlapic_update_ppr(vlapic); + if (vlapic->ops.pending_intr) return ((*vlapic->ops.pending_intr)(vlapic, vecptr)); irrptr = &lapic->irr0; for (i = 7; i >= 0; i--) { idx = i * 4; val = atomic_load_acq_int(&irrptr[idx]); bitpos = fls(val); if (bitpos != 0) { vector = i * 32 + (bitpos - 1); if (PRIO(vector) > PRIO(lapic->ppr)) { VLAPIC_CTR1(vlapic, "pending intr %d", vector); if (vecptr != NULL) *vecptr = vector; return (1); } else break; } } return (0); } void vlapic_intr_accepted(struct vlapic *vlapic, int vector) { struct LAPIC *lapic = vlapic->apic_page; uint32_t *irrptr, *isrptr; int idx, stk_top; if (vlapic->ops.intr_accepted) return ((*vlapic->ops.intr_accepted)(vlapic, vector)); /* * clear the ready bit for vector being accepted in irr * and set the vector as in service in isr. */ idx = (vector / 32) * 4; irrptr = &lapic->irr0; atomic_clear_int(&irrptr[idx], 1 << (vector % 32)); VLAPIC_CTR_IRR(vlapic, "vlapic_intr_accepted"); isrptr = &lapic->isr0; isrptr[idx] |= 1 << (vector % 32); VLAPIC_CTR_ISR(vlapic, "vlapic_intr_accepted"); /* * Update the PPR */ vlapic->isrvec_stk_top++; stk_top = vlapic->isrvec_stk_top; if (stk_top >= ISRVEC_STK_SIZE) panic("isrvec_stk_top overflow %d", stk_top); vlapic->isrvec_stk[stk_top] = vector; - vlapic_update_ppr(vlapic); } void vlapic_svr_write_handler(struct vlapic *vlapic) { struct LAPIC *lapic; uint32_t old, new, changed; lapic = vlapic->apic_page; new = lapic->svr; old = vlapic->svr_last; vlapic->svr_last = new; changed = old ^ new; if ((changed & APIC_SVR_ENABLE) != 0) { if ((new & APIC_SVR_ENABLE) == 0) { /* * The apic is now disabled so stop the apic timer * and mask all the LVT entries. */ VLAPIC_CTR0(vlapic, "vlapic is software-disabled"); VLAPIC_TIMER_LOCK(vlapic); callout_stop(&vlapic->callout); VLAPIC_TIMER_UNLOCK(vlapic); vlapic_mask_lvts(vlapic); } else { /* * The apic is now enabled so restart the apic timer * if it is configured in periodic mode. */ VLAPIC_CTR0(vlapic, "vlapic is software-enabled"); if (vlapic_periodic_timer(vlapic)) vlapic_icrtmr_write_handler(vlapic); } } } int vlapic_read(struct vlapic *vlapic, int mmio_access, uint64_t offset, uint64_t *data, bool *retu) { struct LAPIC *lapic = vlapic->apic_page; uint32_t *reg; int i; /* Ignore MMIO accesses in x2APIC mode */ if (x2apic(vlapic) && mmio_access) { VLAPIC_CTR1(vlapic, "MMIO read from offset %#lx in x2APIC mode", offset); *data = 0; goto done; } if (!x2apic(vlapic) && !mmio_access) { /* * XXX Generate GP fault for MSR accesses in xAPIC mode */ VLAPIC_CTR1(vlapic, "x2APIC MSR read from offset %#lx in " "xAPIC mode", offset); *data = 0; goto done; } if (offset > sizeof(*lapic)) { *data = 0; goto done; } offset &= ~3; switch(offset) { case APIC_OFFSET_ID: *data = lapic->id; break; case APIC_OFFSET_VER: *data = lapic->version; break; case APIC_OFFSET_TPR: *data = vlapic_get_tpr(vlapic); break; case APIC_OFFSET_APR: *data = lapic->apr; break; case APIC_OFFSET_PPR: *data = lapic->ppr; break; case APIC_OFFSET_EOI: *data = lapic->eoi; break; case APIC_OFFSET_LDR: *data = lapic->ldr; break; case APIC_OFFSET_DFR: *data = lapic->dfr; break; case APIC_OFFSET_SVR: *data = lapic->svr; break; case APIC_OFFSET_ISR0 ... APIC_OFFSET_ISR7: i = (offset - APIC_OFFSET_ISR0) >> 2; reg = &lapic->isr0; *data = *(reg + i); break; case APIC_OFFSET_TMR0 ... APIC_OFFSET_TMR7: i = (offset - APIC_OFFSET_TMR0) >> 2; reg = &lapic->tmr0; *data = *(reg + i); break; case APIC_OFFSET_IRR0 ... APIC_OFFSET_IRR7: i = (offset - APIC_OFFSET_IRR0) >> 2; reg = &lapic->irr0; *data = atomic_load_acq_int(reg + i); break; case APIC_OFFSET_ESR: *data = lapic->esr; break; case APIC_OFFSET_ICR_LOW: *data = lapic->icr_lo; if (x2apic(vlapic)) *data |= (uint64_t)lapic->icr_hi << 32; break; case APIC_OFFSET_ICR_HI: *data = lapic->icr_hi; break; case APIC_OFFSET_CMCI_LVT: case APIC_OFFSET_TIMER_LVT ... APIC_OFFSET_ERROR_LVT: *data = vlapic_get_lvt(vlapic, offset); #ifdef INVARIANTS reg = vlapic_get_lvtptr(vlapic, offset); KASSERT(*data == *reg, ("inconsistent lvt value at " "offset %#lx: %#lx/%#x", offset, *data, *reg)); #endif break; case APIC_OFFSET_TIMER_ICR: *data = lapic->icr_timer; break; case APIC_OFFSET_TIMER_CCR: *data = vlapic_get_ccr(vlapic); break; case APIC_OFFSET_TIMER_DCR: *data = lapic->dcr_timer; break; case APIC_OFFSET_SELF_IPI: /* * XXX generate a GP fault if vlapic is in x2apic mode */ *data = 0; break; case APIC_OFFSET_RRR: default: *data = 0; break; } done: VLAPIC_CTR2(vlapic, "vlapic read offset %#x, data %#lx", offset, *data); return 0; } int vlapic_write(struct vlapic *vlapic, int mmio_access, uint64_t offset, uint64_t data, bool *retu) { struct LAPIC *lapic = vlapic->apic_page; uint32_t *regptr; int retval; KASSERT((offset & 0xf) == 0 && offset < PAGE_SIZE, ("vlapic_write: invalid offset %#lx", offset)); VLAPIC_CTR2(vlapic, "vlapic write offset %#lx, data %#lx", offset, data); if (offset > sizeof(*lapic)) return (0); /* Ignore MMIO accesses in x2APIC mode */ if (x2apic(vlapic) && mmio_access) { VLAPIC_CTR2(vlapic, "MMIO write of %#lx to offset %#lx " "in x2APIC mode", data, offset); return (0); } /* * XXX Generate GP fault for MSR accesses in xAPIC mode */ if (!x2apic(vlapic) && !mmio_access) { VLAPIC_CTR2(vlapic, "x2APIC MSR write of %#lx to offset %#lx " "in xAPIC mode", data, offset); return (0); } retval = 0; switch(offset) { case APIC_OFFSET_ID: lapic->id = data; vlapic_id_write_handler(vlapic); break; case APIC_OFFSET_TPR: vlapic_set_tpr(vlapic, data & 0xff); break; case APIC_OFFSET_EOI: vlapic_process_eoi(vlapic); break; case APIC_OFFSET_LDR: lapic->ldr = data; vlapic_ldr_write_handler(vlapic); break; case APIC_OFFSET_DFR: lapic->dfr = data; vlapic_dfr_write_handler(vlapic); break; case APIC_OFFSET_SVR: lapic->svr = data; vlapic_svr_write_handler(vlapic); break; case APIC_OFFSET_ICR_LOW: lapic->icr_lo = data; if (x2apic(vlapic)) lapic->icr_hi = data >> 32; retval = vlapic_icrlo_write_handler(vlapic, retu); break; case APIC_OFFSET_ICR_HI: lapic->icr_hi = data; break; case APIC_OFFSET_CMCI_LVT: case APIC_OFFSET_TIMER_LVT ... APIC_OFFSET_ERROR_LVT: regptr = vlapic_get_lvtptr(vlapic, offset); *regptr = data; vlapic_lvt_write_handler(vlapic, offset); break; case APIC_OFFSET_TIMER_ICR: lapic->icr_timer = data; vlapic_icrtmr_write_handler(vlapic); break; case APIC_OFFSET_TIMER_DCR: lapic->dcr_timer = data; vlapic_dcr_write_handler(vlapic); break; case APIC_OFFSET_ESR: vlapic_esr_write_handler(vlapic); break; case APIC_OFFSET_SELF_IPI: if (x2apic(vlapic)) vlapic_self_ipi_handler(vlapic, data); break; case APIC_OFFSET_VER: case APIC_OFFSET_APR: case APIC_OFFSET_PPR: case APIC_OFFSET_RRR: case APIC_OFFSET_ISR0 ... APIC_OFFSET_ISR7: case APIC_OFFSET_TMR0 ... APIC_OFFSET_TMR7: case APIC_OFFSET_IRR0 ... APIC_OFFSET_IRR7: case APIC_OFFSET_TIMER_CCR: default: // Read only. break; } return (retval); } static void vlapic_reset(struct vlapic *vlapic) { struct LAPIC *lapic; lapic = vlapic->apic_page; bzero(lapic, sizeof(struct LAPIC)); lapic->id = vlapic_get_id(vlapic); lapic->version = VLAPIC_VERSION; lapic->version |= (VLAPIC_MAXLVT_INDEX << MAXLVTSHIFT); lapic->dfr = 0xffffffff; lapic->svr = APIC_SVR_VECTOR; vlapic_mask_lvts(vlapic); vlapic_reset_tmr(vlapic); lapic->dcr_timer = 0; vlapic_dcr_write_handler(vlapic); if (vlapic->vcpuid == 0) vlapic->boot_state = BS_RUNNING; /* BSP */ else vlapic->boot_state = BS_INIT; /* AP */ vlapic->svr_last = lapic->svr; } void vlapic_init(struct vlapic *vlapic) { KASSERT(vlapic->vm != NULL, ("vlapic_init: vm is not initialized")); KASSERT(vlapic->vcpuid >= 0 && vlapic->vcpuid < vm_get_maxcpus(vlapic->vm), ("vlapic_init: vcpuid is not initialized")); KASSERT(vlapic->apic_page != NULL, ("vlapic_init: apic_page is not " "initialized")); /* * If the vlapic is configured in x2apic mode then it will be * accessed in the critical section via the MSR emulation code. * * Therefore the timer mutex must be a spinlock because blockable * mutexes cannot be acquired in a critical section. */ mtx_init(&vlapic->timer_mtx, "vlapic timer mtx", NULL, MTX_SPIN); callout_init(&vlapic->callout, 1); vlapic->msr_apicbase = DEFAULT_APIC_BASE | APICBASE_ENABLED; if (vlapic->vcpuid == 0) vlapic->msr_apicbase |= APICBASE_BSP; vlapic_reset(vlapic); } void vlapic_cleanup(struct vlapic *vlapic) { callout_drain(&vlapic->callout); } uint64_t vlapic_get_apicbase(struct vlapic *vlapic) { return (vlapic->msr_apicbase); } int vlapic_set_apicbase(struct vlapic *vlapic, uint64_t new) { if (vlapic->msr_apicbase != new) { VLAPIC_CTR2(vlapic, "Changing APIC_BASE MSR from %#lx to %#lx " "not supported", vlapic->msr_apicbase, new); return (-1); } return (0); } void vlapic_set_x2apic_state(struct vm *vm, int vcpuid, enum x2apic_state state) { struct vlapic *vlapic; struct LAPIC *lapic; vlapic = vm_lapic(vm, vcpuid); if (state == X2APIC_DISABLED) vlapic->msr_apicbase &= ~APICBASE_X2APIC; else vlapic->msr_apicbase |= APICBASE_X2APIC; /* * Reset the local APIC registers whose values are mode-dependent. * * XXX this works because the APIC mode can be changed only at vcpu * initialization time. */ lapic = vlapic->apic_page; lapic->id = vlapic_get_id(vlapic); if (x2apic(vlapic)) { lapic->ldr = x2apic_ldr(vlapic); lapic->dfr = 0; } else { lapic->ldr = 0; lapic->dfr = 0xffffffff; } if (state == X2APIC_ENABLED) { if (vlapic->ops.enable_x2apic_mode) (*vlapic->ops.enable_x2apic_mode)(vlapic); } } void vlapic_deliver_intr(struct vm *vm, bool level, uint32_t dest, bool phys, int delmode, int vec) { bool lowprio; int vcpuid; cpuset_t dmask; if (delmode != IOART_DELFIXED && delmode != IOART_DELLOPRI && delmode != IOART_DELEXINT) { VM_CTR1(vm, "vlapic intr invalid delmode %#x", delmode); return; } lowprio = (delmode == IOART_DELLOPRI); /* * We don't provide any virtual interrupt redirection hardware so * all interrupts originating from the ioapic or MSI specify the * 'dest' in the legacy xAPIC format. */ vlapic_calcdest(vm, &dmask, dest, phys, lowprio, false); while ((vcpuid = CPU_FFS(&dmask)) != 0) { vcpuid--; CPU_CLR(vcpuid, &dmask); if (delmode == IOART_DELEXINT) { vm_inject_extint(vm, vcpuid); } else { lapic_set_intr(vm, vcpuid, vec, level); } } } void vlapic_post_intr(struct vlapic *vlapic, int hostcpu, int ipinum) { /* * Post an interrupt to the vcpu currently running on 'hostcpu'. * * This is done by leveraging features like Posted Interrupts (Intel) * Doorbell MSR (AMD AVIC) that avoid a VM exit. * * If neither of these features are available then fallback to * sending an IPI to 'hostcpu'. */ if (vlapic->ops.post_intr) (*vlapic->ops.post_intr)(vlapic, hostcpu); else ipi_cpu(hostcpu, ipinum); } bool vlapic_enabled(struct vlapic *vlapic) { struct LAPIC *lapic = vlapic->apic_page; if ((vlapic->msr_apicbase & APICBASE_ENABLED) != 0 && (lapic->svr & APIC_SVR_ENABLE) != 0) return (true); else return (false); } static void vlapic_set_tmr(struct vlapic *vlapic, int vector, bool level) { struct LAPIC *lapic; uint32_t *tmrptr, mask; int idx; lapic = vlapic->apic_page; tmrptr = &lapic->tmr0; idx = (vector / 32) * 4; mask = 1 << (vector % 32); if (level) tmrptr[idx] |= mask; else tmrptr[idx] &= ~mask; if (vlapic->ops.set_tmr != NULL) (*vlapic->ops.set_tmr)(vlapic, vector, level); } void vlapic_reset_tmr(struct vlapic *vlapic) { int vector; VLAPIC_CTR0(vlapic, "vlapic resetting all vectors to edge-triggered"); for (vector = 0; vector <= 255; vector++) vlapic_set_tmr(vlapic, vector, false); } void vlapic_set_tmr_level(struct vlapic *vlapic, uint32_t dest, bool phys, int delmode, int vector) { cpuset_t dmask; bool lowprio; KASSERT(vector >= 0 && vector <= 255, ("invalid vector %d", vector)); /* * A level trigger is valid only for fixed and lowprio delivery modes. */ if (delmode != APIC_DELMODE_FIXED && delmode != APIC_DELMODE_LOWPRIO) { VLAPIC_CTR1(vlapic, "Ignoring level trigger-mode for " "delivery-mode %d", delmode); return; } lowprio = (delmode == APIC_DELMODE_LOWPRIO); vlapic_calcdest(vlapic->vm, &dmask, dest, phys, lowprio, false); if (!CPU_ISSET(vlapic->vcpuid, &dmask)) return; VLAPIC_CTR1(vlapic, "vector %d set to level-triggered", vector); vlapic_set_tmr(vlapic, vector, true); } Index: projects/clang1000-import/sys/amd64/vmm/io/vlapic.h =================================================================== --- projects/clang1000-import/sys/amd64/vmm/io/vlapic.h (revision 358848) +++ projects/clang1000-import/sys/amd64/vmm/io/vlapic.h (revision 358849) @@ -1,110 +1,112 @@ /*- * SPDX-License-Identifier: BSD-2-Clause-FreeBSD * * Copyright (c) 2011 NetApp, Inc. * 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 NETAPP, INC ``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 NETAPP, INC 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. * * $FreeBSD$ */ #ifndef _VLAPIC_H_ #define _VLAPIC_H_ struct vm; enum x2apic_state; int vlapic_write(struct vlapic *vlapic, int mmio_access, uint64_t offset, uint64_t data, bool *retu); int vlapic_read(struct vlapic *vlapic, int mmio_access, uint64_t offset, uint64_t *data, bool *retu); /* * Returns 0 if there is no eligible vector that can be delivered to the * guest at this time and non-zero otherwise. * * If an eligible vector number is found and 'vecptr' is not NULL then it will * be stored in the location pointed to by 'vecptr'. * * Note that the vector does not automatically transition to the ISR as a * result of calling this function. */ int vlapic_pending_intr(struct vlapic *vlapic, int *vecptr); /* * Transition 'vector' from IRR to ISR. This function is called with the * vector returned by 'vlapic_pending_intr()' when the guest is able to * accept this interrupt (i.e. RFLAGS.IF = 1 and no conditions exist that * block interrupt delivery). */ void vlapic_intr_accepted(struct vlapic *vlapic, int vector); /* * Returns 1 if the vcpu needs to be notified of the interrupt and 0 otherwise. */ int vlapic_set_intr_ready(struct vlapic *vlapic, int vector, bool level); /* * Post an interrupt to the vcpu running on 'hostcpu'. This will use a * hardware assist if available (e.g. Posted Interrupt) or fall back to * sending an 'ipinum' to interrupt the 'hostcpu'. */ void vlapic_post_intr(struct vlapic *vlapic, int hostcpu, int ipinum); void vlapic_fire_cmci(struct vlapic *vlapic); int vlapic_trigger_lvt(struct vlapic *vlapic, int vector); +void vlapic_sync_tpr(struct vlapic *vlapic); + uint64_t vlapic_get_apicbase(struct vlapic *vlapic); int vlapic_set_apicbase(struct vlapic *vlapic, uint64_t val); void vlapic_set_x2apic_state(struct vm *vm, int vcpuid, enum x2apic_state s); bool vlapic_enabled(struct vlapic *vlapic); void vlapic_deliver_intr(struct vm *vm, bool level, uint32_t dest, bool phys, int delmode, int vec); /* Reset the trigger-mode bits for all vectors to be edge-triggered */ void vlapic_reset_tmr(struct vlapic *vlapic); /* * Set the trigger-mode bit associated with 'vector' to level-triggered if * the (dest,phys,delmode) tuple resolves to an interrupt being delivered to * this 'vlapic'. */ void vlapic_set_tmr_level(struct vlapic *vlapic, uint32_t dest, bool phys, int delmode, int vector); void vlapic_set_cr8(struct vlapic *vlapic, uint64_t val); uint64_t vlapic_get_cr8(struct vlapic *vlapic); /* APIC write handlers */ void vlapic_id_write_handler(struct vlapic *vlapic); void vlapic_ldr_write_handler(struct vlapic *vlapic); void vlapic_dfr_write_handler(struct vlapic *vlapic); void vlapic_svr_write_handler(struct vlapic *vlapic); void vlapic_esr_write_handler(struct vlapic *vlapic); int vlapic_icrlo_write_handler(struct vlapic *vlapic, bool *retu); void vlapic_icrtmr_write_handler(struct vlapic *vlapic); void vlapic_dcr_write_handler(struct vlapic *vlapic); void vlapic_lvt_write_handler(struct vlapic *vlapic, uint32_t offset); void vlapic_self_ipi_handler(struct vlapic *vlapic, uint64_t val); #endif /* _VLAPIC_H_ */ Index: projects/clang1000-import/sys/compat/linux/linux_ioctl.c =================================================================== --- projects/clang1000-import/sys/compat/linux/linux_ioctl.c (revision 358848) +++ projects/clang1000-import/sys/compat/linux/linux_ioctl.c (revision 358849) @@ -1,3752 +1,3755 @@ /*- * SPDX-License-Identifier: BSD-2-Clause-FreeBSD * * Copyright (c) 1994-1995 Søren Schmidt * 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 "opt_compat.h" #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef COMPAT_LINUX32 #include #include #else #include #include #endif #include #include #include #include #include #include #include #include #include #include #include CTASSERT(LINUX_IFNAMSIZ == IFNAMSIZ); static linux_ioctl_function_t linux_ioctl_cdrom; static linux_ioctl_function_t linux_ioctl_vfat; static linux_ioctl_function_t linux_ioctl_console; static linux_ioctl_function_t linux_ioctl_hdio; static linux_ioctl_function_t linux_ioctl_disk; static linux_ioctl_function_t linux_ioctl_socket; static linux_ioctl_function_t linux_ioctl_sound; static linux_ioctl_function_t linux_ioctl_termio; static linux_ioctl_function_t linux_ioctl_private; static linux_ioctl_function_t linux_ioctl_drm; static linux_ioctl_function_t linux_ioctl_sg; static linux_ioctl_function_t linux_ioctl_v4l; static linux_ioctl_function_t linux_ioctl_v4l2; static linux_ioctl_function_t linux_ioctl_special; static linux_ioctl_function_t linux_ioctl_fbsd_usb; static linux_ioctl_function_t linux_ioctl_evdev; static struct linux_ioctl_handler cdrom_handler = { linux_ioctl_cdrom, LINUX_IOCTL_CDROM_MIN, LINUX_IOCTL_CDROM_MAX }; static struct linux_ioctl_handler vfat_handler = { linux_ioctl_vfat, LINUX_IOCTL_VFAT_MIN, LINUX_IOCTL_VFAT_MAX }; static struct linux_ioctl_handler console_handler = { linux_ioctl_console, LINUX_IOCTL_CONSOLE_MIN, LINUX_IOCTL_CONSOLE_MAX }; static struct linux_ioctl_handler hdio_handler = { linux_ioctl_hdio, LINUX_IOCTL_HDIO_MIN, LINUX_IOCTL_HDIO_MAX }; static struct linux_ioctl_handler disk_handler = { linux_ioctl_disk, LINUX_IOCTL_DISK_MIN, LINUX_IOCTL_DISK_MAX }; static struct linux_ioctl_handler socket_handler = { linux_ioctl_socket, LINUX_IOCTL_SOCKET_MIN, LINUX_IOCTL_SOCKET_MAX }; static struct linux_ioctl_handler sound_handler = { linux_ioctl_sound, LINUX_IOCTL_SOUND_MIN, LINUX_IOCTL_SOUND_MAX }; static struct linux_ioctl_handler termio_handler = { linux_ioctl_termio, LINUX_IOCTL_TERMIO_MIN, LINUX_IOCTL_TERMIO_MAX }; static struct linux_ioctl_handler private_handler = { linux_ioctl_private, LINUX_IOCTL_PRIVATE_MIN, LINUX_IOCTL_PRIVATE_MAX }; static struct linux_ioctl_handler drm_handler = { linux_ioctl_drm, LINUX_IOCTL_DRM_MIN, LINUX_IOCTL_DRM_MAX }; static struct linux_ioctl_handler sg_handler = { linux_ioctl_sg, LINUX_IOCTL_SG_MIN, LINUX_IOCTL_SG_MAX }; static struct linux_ioctl_handler video_handler = { linux_ioctl_v4l, LINUX_IOCTL_VIDEO_MIN, LINUX_IOCTL_VIDEO_MAX }; static struct linux_ioctl_handler video2_handler = { linux_ioctl_v4l2, LINUX_IOCTL_VIDEO2_MIN, LINUX_IOCTL_VIDEO2_MAX }; static struct linux_ioctl_handler fbsd_usb = { linux_ioctl_fbsd_usb, FBSD_LUSB_MIN, FBSD_LUSB_MAX }; static struct linux_ioctl_handler evdev_handler = { linux_ioctl_evdev, LINUX_IOCTL_EVDEV_MIN, LINUX_IOCTL_EVDEV_MAX }; DATA_SET(linux_ioctl_handler_set, cdrom_handler); DATA_SET(linux_ioctl_handler_set, vfat_handler); DATA_SET(linux_ioctl_handler_set, console_handler); DATA_SET(linux_ioctl_handler_set, hdio_handler); DATA_SET(linux_ioctl_handler_set, disk_handler); DATA_SET(linux_ioctl_handler_set, socket_handler); DATA_SET(linux_ioctl_handler_set, sound_handler); DATA_SET(linux_ioctl_handler_set, termio_handler); DATA_SET(linux_ioctl_handler_set, private_handler); DATA_SET(linux_ioctl_handler_set, drm_handler); DATA_SET(linux_ioctl_handler_set, sg_handler); DATA_SET(linux_ioctl_handler_set, video_handler); DATA_SET(linux_ioctl_handler_set, video2_handler); DATA_SET(linux_ioctl_handler_set, fbsd_usb); DATA_SET(linux_ioctl_handler_set, evdev_handler); #ifdef __i386__ static TAILQ_HEAD(, linux_ioctl_handler_element) linux_ioctl_handlers = TAILQ_HEAD_INITIALIZER(linux_ioctl_handlers); static struct sx linux_ioctl_sx; SX_SYSINIT(linux_ioctl, &linux_ioctl_sx, "Linux ioctl handlers"); #else extern TAILQ_HEAD(, linux_ioctl_handler_element) linux_ioctl_handlers; extern struct sx linux_ioctl_sx; #endif #ifdef COMPAT_LINUX32 static TAILQ_HEAD(, linux_ioctl_handler_element) linux32_ioctl_handlers = TAILQ_HEAD_INITIALIZER(linux32_ioctl_handlers); #endif /* * hdio related ioctls for VMWare support */ struct linux_hd_geometry { u_int8_t heads; u_int8_t sectors; u_int16_t cylinders; u_int32_t start; }; struct linux_hd_big_geometry { u_int8_t heads; u_int8_t sectors; u_int32_t cylinders; u_int32_t start; }; static int linux_ioctl_hdio(struct thread *td, struct linux_ioctl_args *args) { struct file *fp; int error; u_int sectorsize, fwcylinders, fwheads, fwsectors; off_t mediasize, bytespercyl; error = fget(td, args->fd, &cap_ioctl_rights, &fp); if (error != 0) return (error); switch (args->cmd & 0xffff) { case LINUX_HDIO_GET_GEO: case LINUX_HDIO_GET_GEO_BIG: error = fo_ioctl(fp, DIOCGMEDIASIZE, (caddr_t)&mediasize, td->td_ucred, td); if (!error) error = fo_ioctl(fp, DIOCGSECTORSIZE, (caddr_t)§orsize, td->td_ucred, td); if (!error) error = fo_ioctl(fp, DIOCGFWHEADS, (caddr_t)&fwheads, td->td_ucred, td); if (!error) error = fo_ioctl(fp, DIOCGFWSECTORS, (caddr_t)&fwsectors, td->td_ucred, td); /* * XXX: DIOCGFIRSTOFFSET is not yet implemented, so * so pretend that GEOM always says 0. This is NOT VALID * for slices or partitions, only the per-disk raw devices. */ fdrop(fp, td); if (error) return (error); /* * 1. Calculate the number of bytes in a cylinder, * given the firmware's notion of heads and sectors * per cylinder. * 2. Calculate the number of cylinders, given the total * size of the media. * All internal calculations should have 64-bit precision. */ bytespercyl = (off_t) sectorsize * fwheads * fwsectors; fwcylinders = mediasize / bytespercyl; if ((args->cmd & 0xffff) == LINUX_HDIO_GET_GEO) { struct linux_hd_geometry hdg; hdg.cylinders = fwcylinders; hdg.heads = fwheads; hdg.sectors = fwsectors; hdg.start = 0; error = copyout(&hdg, (void *)args->arg, sizeof(hdg)); } else if ((args->cmd & 0xffff) == LINUX_HDIO_GET_GEO_BIG) { struct linux_hd_big_geometry hdbg; memset(&hdbg, 0, sizeof(hdbg)); hdbg.cylinders = fwcylinders; hdbg.heads = fwheads; hdbg.sectors = fwsectors; hdbg.start = 0; error = copyout(&hdbg, (void *)args->arg, sizeof(hdbg)); } return (error); break; default: /* XXX */ linux_msg(td, "ioctl fd=%d, cmd=0x%x ('%c',%d) is not implemented", args->fd, (int)(args->cmd & 0xffff), (int)(args->cmd & 0xff00) >> 8, (int)(args->cmd & 0xff)); break; } fdrop(fp, td); return (ENOIOCTL); } static int linux_ioctl_disk(struct thread *td, struct linux_ioctl_args *args) { struct file *fp; int error; u_int sectorsize; uint64_t blksize64; off_t mediasize; error = fget(td, args->fd, &cap_ioctl_rights, &fp); if (error != 0) return (error); switch (args->cmd & 0xffff) { case LINUX_BLKGETSIZE: error = fo_ioctl(fp, DIOCGSECTORSIZE, (caddr_t)§orsize, td->td_ucred, td); if (!error) error = fo_ioctl(fp, DIOCGMEDIASIZE, (caddr_t)&mediasize, td->td_ucred, td); fdrop(fp, td); if (error) return (error); sectorsize = mediasize / sectorsize; /* * XXX: How do we know we return the right size of integer ? */ return (copyout(§orsize, (void *)args->arg, sizeof(sectorsize))); break; case LINUX_BLKGETSIZE64: error = fo_ioctl(fp, DIOCGMEDIASIZE, (caddr_t)&mediasize, td->td_ucred, td); fdrop(fp, td); if (error) return (error); blksize64 = mediasize;; return (copyout(&blksize64, (void *)args->arg, sizeof(blksize64))); case LINUX_BLKSSZGET: error = fo_ioctl(fp, DIOCGSECTORSIZE, (caddr_t)§orsize, td->td_ucred, td); fdrop(fp, td); if (error) return (error); return (copyout(§orsize, (void *)args->arg, sizeof(sectorsize))); break; } fdrop(fp, td); return (ENOIOCTL); } /* * termio related ioctls */ struct linux_termio { unsigned short c_iflag; unsigned short c_oflag; unsigned short c_cflag; unsigned short c_lflag; unsigned char c_line; unsigned char c_cc[LINUX_NCC]; }; struct linux_termios { unsigned int c_iflag; unsigned int c_oflag; unsigned int c_cflag; unsigned int c_lflag; unsigned char c_line; unsigned char c_cc[LINUX_NCCS]; }; struct linux_winsize { unsigned short ws_row, ws_col; unsigned short ws_xpixel, ws_ypixel; }; struct speedtab { int sp_speed; /* Speed. */ int sp_code; /* Code. */ }; static struct speedtab sptab[] = { { B0, LINUX_B0 }, { B50, LINUX_B50 }, { B75, LINUX_B75 }, { B110, LINUX_B110 }, { B134, LINUX_B134 }, { B150, LINUX_B150 }, { B200, LINUX_B200 }, { B300, LINUX_B300 }, { B600, LINUX_B600 }, { B1200, LINUX_B1200 }, { B1800, LINUX_B1800 }, { B2400, LINUX_B2400 }, { B4800, LINUX_B4800 }, { B9600, LINUX_B9600 }, { B19200, LINUX_B19200 }, { B38400, LINUX_B38400 }, { B57600, LINUX_B57600 }, { B115200, LINUX_B115200 }, {-1, -1 } }; struct linux_serial_struct { int type; int line; int port; int irq; int flags; int xmit_fifo_size; int custom_divisor; int baud_base; unsigned short close_delay; char reserved_char[2]; int hub6; unsigned short closing_wait; unsigned short closing_wait2; int reserved[4]; }; static int linux_to_bsd_speed(int code, struct speedtab *table) { for ( ; table->sp_code != -1; table++) if (table->sp_code == code) return (table->sp_speed); return (-1); } static int bsd_to_linux_speed(int speed, struct speedtab *table) { for ( ; table->sp_speed != -1; table++) if (table->sp_speed == speed) return (table->sp_code); return (-1); } static void bsd_to_linux_termios(struct termios *bios, struct linux_termios *lios) { int i; lios->c_iflag = 0; if (bios->c_iflag & IGNBRK) lios->c_iflag |= LINUX_IGNBRK; if (bios->c_iflag & BRKINT) lios->c_iflag |= LINUX_BRKINT; if (bios->c_iflag & IGNPAR) lios->c_iflag |= LINUX_IGNPAR; if (bios->c_iflag & PARMRK) lios->c_iflag |= LINUX_PARMRK; if (bios->c_iflag & INPCK) lios->c_iflag |= LINUX_INPCK; if (bios->c_iflag & ISTRIP) lios->c_iflag |= LINUX_ISTRIP; if (bios->c_iflag & INLCR) lios->c_iflag |= LINUX_INLCR; if (bios->c_iflag & IGNCR) lios->c_iflag |= LINUX_IGNCR; if (bios->c_iflag & ICRNL) lios->c_iflag |= LINUX_ICRNL; if (bios->c_iflag & IXON) lios->c_iflag |= LINUX_IXON; if (bios->c_iflag & IXANY) lios->c_iflag |= LINUX_IXANY; if (bios->c_iflag & IXOFF) lios->c_iflag |= LINUX_IXOFF; if (bios->c_iflag & IMAXBEL) lios->c_iflag |= LINUX_IMAXBEL; lios->c_oflag = 0; if (bios->c_oflag & OPOST) lios->c_oflag |= LINUX_OPOST; if (bios->c_oflag & ONLCR) lios->c_oflag |= LINUX_ONLCR; if (bios->c_oflag & TAB3) lios->c_oflag |= LINUX_XTABS; lios->c_cflag = bsd_to_linux_speed(bios->c_ispeed, sptab); lios->c_cflag |= (bios->c_cflag & CSIZE) >> 4; if (bios->c_cflag & CSTOPB) lios->c_cflag |= LINUX_CSTOPB; if (bios->c_cflag & CREAD) lios->c_cflag |= LINUX_CREAD; if (bios->c_cflag & PARENB) lios->c_cflag |= LINUX_PARENB; if (bios->c_cflag & PARODD) lios->c_cflag |= LINUX_PARODD; if (bios->c_cflag & HUPCL) lios->c_cflag |= LINUX_HUPCL; if (bios->c_cflag & CLOCAL) lios->c_cflag |= LINUX_CLOCAL; if (bios->c_cflag & CRTSCTS) lios->c_cflag |= LINUX_CRTSCTS; lios->c_lflag = 0; if (bios->c_lflag & ISIG) lios->c_lflag |= LINUX_ISIG; if (bios->c_lflag & ICANON) lios->c_lflag |= LINUX_ICANON; if (bios->c_lflag & ECHO) lios->c_lflag |= LINUX_ECHO; if (bios->c_lflag & ECHOE) lios->c_lflag |= LINUX_ECHOE; if (bios->c_lflag & ECHOK) lios->c_lflag |= LINUX_ECHOK; if (bios->c_lflag & ECHONL) lios->c_lflag |= LINUX_ECHONL; if (bios->c_lflag & NOFLSH) lios->c_lflag |= LINUX_NOFLSH; if (bios->c_lflag & TOSTOP) lios->c_lflag |= LINUX_TOSTOP; if (bios->c_lflag & ECHOCTL) lios->c_lflag |= LINUX_ECHOCTL; if (bios->c_lflag & ECHOPRT) lios->c_lflag |= LINUX_ECHOPRT; if (bios->c_lflag & ECHOKE) lios->c_lflag |= LINUX_ECHOKE; if (bios->c_lflag & FLUSHO) lios->c_lflag |= LINUX_FLUSHO; if (bios->c_lflag & PENDIN) lios->c_lflag |= LINUX_PENDIN; if (bios->c_lflag & IEXTEN) lios->c_lflag |= LINUX_IEXTEN; for (i=0; ic_cc[i] = LINUX_POSIX_VDISABLE; lios->c_cc[LINUX_VINTR] = bios->c_cc[VINTR]; lios->c_cc[LINUX_VQUIT] = bios->c_cc[VQUIT]; lios->c_cc[LINUX_VERASE] = bios->c_cc[VERASE]; lios->c_cc[LINUX_VKILL] = bios->c_cc[VKILL]; lios->c_cc[LINUX_VEOF] = bios->c_cc[VEOF]; lios->c_cc[LINUX_VEOL] = bios->c_cc[VEOL]; lios->c_cc[LINUX_VMIN] = bios->c_cc[VMIN]; lios->c_cc[LINUX_VTIME] = bios->c_cc[VTIME]; lios->c_cc[LINUX_VEOL2] = bios->c_cc[VEOL2]; lios->c_cc[LINUX_VSUSP] = bios->c_cc[VSUSP]; lios->c_cc[LINUX_VSTART] = bios->c_cc[VSTART]; lios->c_cc[LINUX_VSTOP] = bios->c_cc[VSTOP]; lios->c_cc[LINUX_VREPRINT] = bios->c_cc[VREPRINT]; lios->c_cc[LINUX_VDISCARD] = bios->c_cc[VDISCARD]; lios->c_cc[LINUX_VWERASE] = bios->c_cc[VWERASE]; lios->c_cc[LINUX_VLNEXT] = bios->c_cc[VLNEXT]; if (linux_preserve_vstatus) lios->c_cc[LINUX_VSTATUS] = bios->c_cc[VSTATUS]; for (i=0; ic_cc[i] == _POSIX_VDISABLE) lios->c_cc[i] = LINUX_POSIX_VDISABLE; } lios->c_line = 0; } static void linux_to_bsd_termios(struct linux_termios *lios, struct termios *bios) { int i; bios->c_iflag = 0; if (lios->c_iflag & LINUX_IGNBRK) bios->c_iflag |= IGNBRK; if (lios->c_iflag & LINUX_BRKINT) bios->c_iflag |= BRKINT; if (lios->c_iflag & LINUX_IGNPAR) bios->c_iflag |= IGNPAR; if (lios->c_iflag & LINUX_PARMRK) bios->c_iflag |= PARMRK; if (lios->c_iflag & LINUX_INPCK) bios->c_iflag |= INPCK; if (lios->c_iflag & LINUX_ISTRIP) bios->c_iflag |= ISTRIP; if (lios->c_iflag & LINUX_INLCR) bios->c_iflag |= INLCR; if (lios->c_iflag & LINUX_IGNCR) bios->c_iflag |= IGNCR; if (lios->c_iflag & LINUX_ICRNL) bios->c_iflag |= ICRNL; if (lios->c_iflag & LINUX_IXON) bios->c_iflag |= IXON; if (lios->c_iflag & LINUX_IXANY) bios->c_iflag |= IXANY; if (lios->c_iflag & LINUX_IXOFF) bios->c_iflag |= IXOFF; if (lios->c_iflag & LINUX_IMAXBEL) bios->c_iflag |= IMAXBEL; bios->c_oflag = 0; if (lios->c_oflag & LINUX_OPOST) bios->c_oflag |= OPOST; if (lios->c_oflag & LINUX_ONLCR) bios->c_oflag |= ONLCR; if (lios->c_oflag & LINUX_XTABS) bios->c_oflag |= TAB3; bios->c_cflag = (lios->c_cflag & LINUX_CSIZE) << 4; if (lios->c_cflag & LINUX_CSTOPB) bios->c_cflag |= CSTOPB; if (lios->c_cflag & LINUX_CREAD) bios->c_cflag |= CREAD; if (lios->c_cflag & LINUX_PARENB) bios->c_cflag |= PARENB; if (lios->c_cflag & LINUX_PARODD) bios->c_cflag |= PARODD; if (lios->c_cflag & LINUX_HUPCL) bios->c_cflag |= HUPCL; if (lios->c_cflag & LINUX_CLOCAL) bios->c_cflag |= CLOCAL; if (lios->c_cflag & LINUX_CRTSCTS) bios->c_cflag |= CRTSCTS; bios->c_lflag = 0; if (lios->c_lflag & LINUX_ISIG) bios->c_lflag |= ISIG; if (lios->c_lflag & LINUX_ICANON) bios->c_lflag |= ICANON; if (lios->c_lflag & LINUX_ECHO) bios->c_lflag |= ECHO; if (lios->c_lflag & LINUX_ECHOE) bios->c_lflag |= ECHOE; if (lios->c_lflag & LINUX_ECHOK) bios->c_lflag |= ECHOK; if (lios->c_lflag & LINUX_ECHONL) bios->c_lflag |= ECHONL; if (lios->c_lflag & LINUX_NOFLSH) bios->c_lflag |= NOFLSH; if (lios->c_lflag & LINUX_TOSTOP) bios->c_lflag |= TOSTOP; if (lios->c_lflag & LINUX_ECHOCTL) bios->c_lflag |= ECHOCTL; if (lios->c_lflag & LINUX_ECHOPRT) bios->c_lflag |= ECHOPRT; if (lios->c_lflag & LINUX_ECHOKE) bios->c_lflag |= ECHOKE; if (lios->c_lflag & LINUX_FLUSHO) bios->c_lflag |= FLUSHO; if (lios->c_lflag & LINUX_PENDIN) bios->c_lflag |= PENDIN; if (lios->c_lflag & LINUX_IEXTEN) bios->c_lflag |= IEXTEN; for (i=0; ic_cc[i] = _POSIX_VDISABLE; bios->c_cc[VINTR] = lios->c_cc[LINUX_VINTR]; bios->c_cc[VQUIT] = lios->c_cc[LINUX_VQUIT]; bios->c_cc[VERASE] = lios->c_cc[LINUX_VERASE]; bios->c_cc[VKILL] = lios->c_cc[LINUX_VKILL]; bios->c_cc[VEOF] = lios->c_cc[LINUX_VEOF]; bios->c_cc[VEOL] = lios->c_cc[LINUX_VEOL]; bios->c_cc[VMIN] = lios->c_cc[LINUX_VMIN]; bios->c_cc[VTIME] = lios->c_cc[LINUX_VTIME]; bios->c_cc[VEOL2] = lios->c_cc[LINUX_VEOL2]; bios->c_cc[VSUSP] = lios->c_cc[LINUX_VSUSP]; bios->c_cc[VSTART] = lios->c_cc[LINUX_VSTART]; bios->c_cc[VSTOP] = lios->c_cc[LINUX_VSTOP]; bios->c_cc[VREPRINT] = lios->c_cc[LINUX_VREPRINT]; bios->c_cc[VDISCARD] = lios->c_cc[LINUX_VDISCARD]; bios->c_cc[VWERASE] = lios->c_cc[LINUX_VWERASE]; bios->c_cc[VLNEXT] = lios->c_cc[LINUX_VLNEXT]; if (linux_preserve_vstatus) bios->c_cc[VSTATUS] = lios->c_cc[LINUX_VSTATUS]; for (i=0; ic_cc[i] == LINUX_POSIX_VDISABLE) bios->c_cc[i] = _POSIX_VDISABLE; } bios->c_ispeed = bios->c_ospeed = linux_to_bsd_speed(lios->c_cflag & LINUX_CBAUD, sptab); } static void bsd_to_linux_termio(struct termios *bios, struct linux_termio *lio) { struct linux_termios lios; memset(lio, 0, sizeof(*lio)); bsd_to_linux_termios(bios, &lios); lio->c_iflag = lios.c_iflag; lio->c_oflag = lios.c_oflag; lio->c_cflag = lios.c_cflag; lio->c_lflag = lios.c_lflag; lio->c_line = lios.c_line; memcpy(lio->c_cc, lios.c_cc, LINUX_NCC); } static void linux_to_bsd_termio(struct linux_termio *lio, struct termios *bios) { struct linux_termios lios; int i; lios.c_iflag = lio->c_iflag; lios.c_oflag = lio->c_oflag; lios.c_cflag = lio->c_cflag; lios.c_lflag = lio->c_lflag; for (i=LINUX_NCC; ic_cc, LINUX_NCC); linux_to_bsd_termios(&lios, bios); } static int linux_ioctl_termio(struct thread *td, struct linux_ioctl_args *args) { struct termios bios; struct linux_termios lios; struct linux_termio lio; struct file *fp; int error; error = fget(td, args->fd, &cap_ioctl_rights, &fp); if (error != 0) return (error); switch (args->cmd & 0xffff) { case LINUX_TCGETS: error = fo_ioctl(fp, TIOCGETA, (caddr_t)&bios, td->td_ucred, td); if (error) break; bsd_to_linux_termios(&bios, &lios); error = copyout(&lios, (void *)args->arg, sizeof(lios)); break; case LINUX_TCSETS: error = copyin((void *)args->arg, &lios, sizeof(lios)); if (error) break; linux_to_bsd_termios(&lios, &bios); error = (fo_ioctl(fp, TIOCSETA, (caddr_t)&bios, td->td_ucred, td)); break; case LINUX_TCSETSW: error = copyin((void *)args->arg, &lios, sizeof(lios)); if (error) break; linux_to_bsd_termios(&lios, &bios); error = (fo_ioctl(fp, TIOCSETAW, (caddr_t)&bios, td->td_ucred, td)); break; case LINUX_TCSETSF: error = copyin((void *)args->arg, &lios, sizeof(lios)); if (error) break; linux_to_bsd_termios(&lios, &bios); error = (fo_ioctl(fp, TIOCSETAF, (caddr_t)&bios, td->td_ucred, td)); break; case LINUX_TCGETA: error = fo_ioctl(fp, TIOCGETA, (caddr_t)&bios, td->td_ucred, td); if (error) break; bsd_to_linux_termio(&bios, &lio); error = (copyout(&lio, (void *)args->arg, sizeof(lio))); break; case LINUX_TCSETA: error = copyin((void *)args->arg, &lio, sizeof(lio)); if (error) break; linux_to_bsd_termio(&lio, &bios); error = (fo_ioctl(fp, TIOCSETA, (caddr_t)&bios, td->td_ucred, td)); break; case LINUX_TCSETAW: error = copyin((void *)args->arg, &lio, sizeof(lio)); if (error) break; linux_to_bsd_termio(&lio, &bios); error = (fo_ioctl(fp, TIOCSETAW, (caddr_t)&bios, td->td_ucred, td)); break; case LINUX_TCSETAF: error = copyin((void *)args->arg, &lio, sizeof(lio)); if (error) break; linux_to_bsd_termio(&lio, &bios); error = (fo_ioctl(fp, TIOCSETAF, (caddr_t)&bios, td->td_ucred, td)); break; /* LINUX_TCSBRK */ case LINUX_TCXONC: { switch (args->arg) { case LINUX_TCOOFF: args->cmd = TIOCSTOP; break; case LINUX_TCOON: args->cmd = TIOCSTART; break; case LINUX_TCIOFF: case LINUX_TCION: { int c; struct write_args wr; error = fo_ioctl(fp, TIOCGETA, (caddr_t)&bios, td->td_ucred, td); if (error) break; fdrop(fp, td); c = (args->arg == LINUX_TCIOFF) ? VSTOP : VSTART; c = bios.c_cc[c]; if (c != _POSIX_VDISABLE) { wr.fd = args->fd; wr.buf = &c; wr.nbyte = sizeof(c); return (sys_write(td, &wr)); } else return (0); } default: fdrop(fp, td); return (EINVAL); } args->arg = 0; error = (sys_ioctl(td, (struct ioctl_args *)args)); break; } case LINUX_TCFLSH: { int val; switch (args->arg) { case LINUX_TCIFLUSH: val = FREAD; break; case LINUX_TCOFLUSH: val = FWRITE; break; case LINUX_TCIOFLUSH: val = FREAD | FWRITE; break; default: fdrop(fp, td); return (EINVAL); } error = (fo_ioctl(fp,TIOCFLUSH,(caddr_t)&val,td->td_ucred,td)); break; } case LINUX_TIOCEXCL: args->cmd = TIOCEXCL; error = (sys_ioctl(td, (struct ioctl_args *)args)); break; case LINUX_TIOCNXCL: args->cmd = TIOCNXCL; error = (sys_ioctl(td, (struct ioctl_args *)args)); break; case LINUX_TIOCSCTTY: args->cmd = TIOCSCTTY; error = (sys_ioctl(td, (struct ioctl_args *)args)); break; case LINUX_TIOCGPGRP: args->cmd = TIOCGPGRP; error = (sys_ioctl(td, (struct ioctl_args *)args)); break; case LINUX_TIOCSPGRP: args->cmd = TIOCSPGRP; error = (sys_ioctl(td, (struct ioctl_args *)args)); break; /* LINUX_TIOCOUTQ */ /* LINUX_TIOCSTI */ case LINUX_TIOCGWINSZ: args->cmd = TIOCGWINSZ; error = (sys_ioctl(td, (struct ioctl_args *)args)); break; case LINUX_TIOCSWINSZ: args->cmd = TIOCSWINSZ; error = (sys_ioctl(td, (struct ioctl_args *)args)); break; case LINUX_TIOCMGET: args->cmd = TIOCMGET; error = (sys_ioctl(td, (struct ioctl_args *)args)); break; case LINUX_TIOCMBIS: args->cmd = TIOCMBIS; error = (sys_ioctl(td, (struct ioctl_args *)args)); break; case LINUX_TIOCMBIC: args->cmd = TIOCMBIC; error = (sys_ioctl(td, (struct ioctl_args *)args)); break; case LINUX_TIOCMSET: args->cmd = TIOCMSET; error = (sys_ioctl(td, (struct ioctl_args *)args)); break; /* TIOCGSOFTCAR */ /* TIOCSSOFTCAR */ case LINUX_FIONREAD: /* LINUX_TIOCINQ */ args->cmd = FIONREAD; error = (sys_ioctl(td, (struct ioctl_args *)args)); break; /* LINUX_TIOCLINUX */ case LINUX_TIOCCONS: args->cmd = TIOCCONS; error = (sys_ioctl(td, (struct ioctl_args *)args)); break; case LINUX_TIOCGSERIAL: { struct linux_serial_struct lss; bzero(&lss, sizeof(lss)); lss.type = LINUX_PORT_16550A; lss.flags = 0; lss.close_delay = 0; error = copyout(&lss, (void *)args->arg, sizeof(lss)); break; } case LINUX_TIOCSSERIAL: { struct linux_serial_struct lss; error = copyin((void *)args->arg, &lss, sizeof(lss)); if (error) break; /* XXX - It really helps to have an implementation that * does nothing. NOT! */ error = 0; break; } case LINUX_TIOCPKT: args->cmd = TIOCPKT; error = (sys_ioctl(td, (struct ioctl_args *)args)); break; case LINUX_FIONBIO: args->cmd = FIONBIO; error = (sys_ioctl(td, (struct ioctl_args *)args)); break; case LINUX_TIOCNOTTY: args->cmd = TIOCNOTTY; error = (sys_ioctl(td, (struct ioctl_args *)args)); break; case LINUX_TIOCSETD: { int line; switch (args->arg) { case LINUX_N_TTY: line = TTYDISC; break; case LINUX_N_SLIP: line = SLIPDISC; break; case LINUX_N_PPP: line = PPPDISC; break; default: fdrop(fp, td); return (EINVAL); } error = (fo_ioctl(fp, TIOCSETD, (caddr_t)&line, td->td_ucred, td)); break; } case LINUX_TIOCGETD: { int linux_line; int bsd_line = TTYDISC; error = fo_ioctl(fp, TIOCGETD, (caddr_t)&bsd_line, td->td_ucred, td); if (error) break; switch (bsd_line) { case TTYDISC: linux_line = LINUX_N_TTY; break; case SLIPDISC: linux_line = LINUX_N_SLIP; break; case PPPDISC: linux_line = LINUX_N_PPP; break; default: fdrop(fp, td); return (EINVAL); } error = (copyout(&linux_line, (void *)args->arg, sizeof(int))); break; } /* LINUX_TCSBRKP */ /* LINUX_TIOCTTYGSTRUCT */ case LINUX_FIONCLEX: args->cmd = FIONCLEX; error = (sys_ioctl(td, (struct ioctl_args *)args)); break; case LINUX_FIOCLEX: args->cmd = FIOCLEX; error = (sys_ioctl(td, (struct ioctl_args *)args)); break; case LINUX_FIOASYNC: args->cmd = FIOASYNC; error = (sys_ioctl(td, (struct ioctl_args *)args)); break; /* LINUX_TIOCSERCONFIG */ /* LINUX_TIOCSERGWILD */ /* LINUX_TIOCSERSWILD */ /* LINUX_TIOCGLCKTRMIOS */ /* LINUX_TIOCSLCKTRMIOS */ case LINUX_TIOCSBRK: args->cmd = TIOCSBRK; error = (sys_ioctl(td, (struct ioctl_args *)args)); break; case LINUX_TIOCCBRK: args->cmd = TIOCCBRK; error = (sys_ioctl(td, (struct ioctl_args *)args)); break; case LINUX_TIOCGPTN: { int nb; error = fo_ioctl(fp, TIOCGPTN, (caddr_t)&nb, td->td_ucred, td); if (!error) error = copyout(&nb, (void *)args->arg, sizeof(int)); break; } case LINUX_TIOCSPTLCK: /* Our unlockpt() does nothing. */ error = 0; break; default: error = ENOIOCTL; break; } fdrop(fp, td); return (error); } /* * CDROM related ioctls */ struct linux_cdrom_msf { u_char cdmsf_min0; u_char cdmsf_sec0; u_char cdmsf_frame0; u_char cdmsf_min1; u_char cdmsf_sec1; u_char cdmsf_frame1; }; struct linux_cdrom_tochdr { u_char cdth_trk0; u_char cdth_trk1; }; union linux_cdrom_addr { struct { u_char minute; u_char second; u_char frame; } msf; int lba; }; struct linux_cdrom_tocentry { u_char cdte_track; u_char cdte_adr:4; u_char cdte_ctrl:4; u_char cdte_format; union linux_cdrom_addr cdte_addr; u_char cdte_datamode; }; struct linux_cdrom_subchnl { u_char cdsc_format; u_char cdsc_audiostatus; u_char cdsc_adr:4; u_char cdsc_ctrl:4; u_char cdsc_trk; u_char cdsc_ind; union linux_cdrom_addr cdsc_absaddr; union linux_cdrom_addr cdsc_reladdr; }; struct l_cdrom_read_audio { union linux_cdrom_addr addr; u_char addr_format; l_int nframes; u_char *buf; }; struct l_dvd_layer { u_char book_version:4; u_char book_type:4; u_char min_rate:4; u_char disc_size:4; u_char layer_type:4; u_char track_path:1; u_char nlayers:2; u_char track_density:4; u_char linear_density:4; u_char bca:1; u_int32_t start_sector; u_int32_t end_sector; u_int32_t end_sector_l0; }; struct l_dvd_physical { u_char type; u_char layer_num; struct l_dvd_layer layer[4]; }; struct l_dvd_copyright { u_char type; u_char layer_num; u_char cpst; u_char rmi; }; struct l_dvd_disckey { u_char type; l_uint agid:2; u_char value[2048]; }; struct l_dvd_bca { u_char type; l_int len; u_char value[188]; }; struct l_dvd_manufact { u_char type; u_char layer_num; l_int len; u_char value[2048]; }; typedef union { u_char type; struct l_dvd_physical physical; struct l_dvd_copyright copyright; struct l_dvd_disckey disckey; struct l_dvd_bca bca; struct l_dvd_manufact manufact; } l_dvd_struct; typedef u_char l_dvd_key[5]; typedef u_char l_dvd_challenge[10]; struct l_dvd_lu_send_agid { u_char type; l_uint agid:2; }; struct l_dvd_host_send_challenge { u_char type; l_uint agid:2; l_dvd_challenge chal; }; struct l_dvd_send_key { u_char type; l_uint agid:2; l_dvd_key key; }; struct l_dvd_lu_send_challenge { u_char type; l_uint agid:2; l_dvd_challenge chal; }; struct l_dvd_lu_send_title_key { u_char type; l_uint agid:2; l_dvd_key title_key; l_int lba; l_uint cpm:1; l_uint cp_sec:1; l_uint cgms:2; }; struct l_dvd_lu_send_asf { u_char type; l_uint agid:2; l_uint asf:1; }; struct l_dvd_host_send_rpcstate { u_char type; u_char pdrc; }; struct l_dvd_lu_send_rpcstate { u_char type:2; u_char vra:3; u_char ucca:3; u_char region_mask; u_char rpc_scheme; }; typedef union { u_char type; struct l_dvd_lu_send_agid lsa; struct l_dvd_host_send_challenge hsc; struct l_dvd_send_key lsk; struct l_dvd_lu_send_challenge lsc; struct l_dvd_send_key hsk; struct l_dvd_lu_send_title_key lstk; struct l_dvd_lu_send_asf lsasf; struct l_dvd_host_send_rpcstate hrpcs; struct l_dvd_lu_send_rpcstate lrpcs; } l_dvd_authinfo; static void bsd_to_linux_msf_lba(u_char af, union msf_lba *bp, union linux_cdrom_addr *lp) { if (af == CD_LBA_FORMAT) lp->lba = bp->lba; else { lp->msf.minute = bp->msf.minute; lp->msf.second = bp->msf.second; lp->msf.frame = bp->msf.frame; } } static void set_linux_cdrom_addr(union linux_cdrom_addr *addr, int format, int lba) { if (format == LINUX_CDROM_MSF) { addr->msf.frame = lba % 75; lba /= 75; lba += 2; addr->msf.second = lba % 60; addr->msf.minute = lba / 60; } else addr->lba = lba; } static int linux_to_bsd_dvd_struct(l_dvd_struct *lp, struct dvd_struct *bp) { bp->format = lp->type; switch (bp->format) { case DVD_STRUCT_PHYSICAL: if (bp->layer_num >= 4) return (EINVAL); bp->layer_num = lp->physical.layer_num; break; case DVD_STRUCT_COPYRIGHT: bp->layer_num = lp->copyright.layer_num; break; case DVD_STRUCT_DISCKEY: bp->agid = lp->disckey.agid; break; case DVD_STRUCT_BCA: case DVD_STRUCT_MANUFACT: break; default: return (EINVAL); } return (0); } static int bsd_to_linux_dvd_struct(struct dvd_struct *bp, l_dvd_struct *lp) { switch (bp->format) { case DVD_STRUCT_PHYSICAL: { struct dvd_layer *blp = (struct dvd_layer *)bp->data; struct l_dvd_layer *llp = &lp->physical.layer[bp->layer_num]; memset(llp, 0, sizeof(*llp)); llp->book_version = blp->book_version; llp->book_type = blp->book_type; llp->min_rate = blp->max_rate; llp->disc_size = blp->disc_size; llp->layer_type = blp->layer_type; llp->track_path = blp->track_path; llp->nlayers = blp->nlayers; llp->track_density = blp->track_density; llp->linear_density = blp->linear_density; llp->bca = blp->bca; llp->start_sector = blp->start_sector; llp->end_sector = blp->end_sector; llp->end_sector_l0 = blp->end_sector_l0; break; } case DVD_STRUCT_COPYRIGHT: lp->copyright.cpst = bp->cpst; lp->copyright.rmi = bp->rmi; break; case DVD_STRUCT_DISCKEY: memcpy(lp->disckey.value, bp->data, sizeof(lp->disckey.value)); break; case DVD_STRUCT_BCA: lp->bca.len = bp->length; memcpy(lp->bca.value, bp->data, sizeof(lp->bca.value)); break; case DVD_STRUCT_MANUFACT: lp->manufact.len = bp->length; memcpy(lp->manufact.value, bp->data, sizeof(lp->manufact.value)); /* lp->manufact.layer_num is unused in Linux (redhat 7.0). */ break; default: return (EINVAL); } return (0); } static int linux_to_bsd_dvd_authinfo(l_dvd_authinfo *lp, int *bcode, struct dvd_authinfo *bp) { switch (lp->type) { case LINUX_DVD_LU_SEND_AGID: *bcode = DVDIOCREPORTKEY; bp->format = DVD_REPORT_AGID; bp->agid = lp->lsa.agid; break; case LINUX_DVD_HOST_SEND_CHALLENGE: *bcode = DVDIOCSENDKEY; bp->format = DVD_SEND_CHALLENGE; bp->agid = lp->hsc.agid; memcpy(bp->keychal, lp->hsc.chal, 10); break; case LINUX_DVD_LU_SEND_KEY1: *bcode = DVDIOCREPORTKEY; bp->format = DVD_REPORT_KEY1; bp->agid = lp->lsk.agid; break; case LINUX_DVD_LU_SEND_CHALLENGE: *bcode = DVDIOCREPORTKEY; bp->format = DVD_REPORT_CHALLENGE; bp->agid = lp->lsc.agid; break; case LINUX_DVD_HOST_SEND_KEY2: *bcode = DVDIOCSENDKEY; bp->format = DVD_SEND_KEY2; bp->agid = lp->hsk.agid; memcpy(bp->keychal, lp->hsk.key, 5); break; case LINUX_DVD_LU_SEND_TITLE_KEY: *bcode = DVDIOCREPORTKEY; bp->format = DVD_REPORT_TITLE_KEY; bp->agid = lp->lstk.agid; bp->lba = lp->lstk.lba; break; case LINUX_DVD_LU_SEND_ASF: *bcode = DVDIOCREPORTKEY; bp->format = DVD_REPORT_ASF; bp->agid = lp->lsasf.agid; break; case LINUX_DVD_INVALIDATE_AGID: *bcode = DVDIOCREPORTKEY; bp->format = DVD_INVALIDATE_AGID; bp->agid = lp->lsa.agid; break; case LINUX_DVD_LU_SEND_RPC_STATE: *bcode = DVDIOCREPORTKEY; bp->format = DVD_REPORT_RPC; break; case LINUX_DVD_HOST_SEND_RPC_STATE: *bcode = DVDIOCSENDKEY; bp->format = DVD_SEND_RPC; bp->region = lp->hrpcs.pdrc; break; default: return (EINVAL); } return (0); } static int bsd_to_linux_dvd_authinfo(struct dvd_authinfo *bp, l_dvd_authinfo *lp) { switch (lp->type) { case LINUX_DVD_LU_SEND_AGID: lp->lsa.agid = bp->agid; break; case LINUX_DVD_HOST_SEND_CHALLENGE: lp->type = LINUX_DVD_LU_SEND_KEY1; break; case LINUX_DVD_LU_SEND_KEY1: memcpy(lp->lsk.key, bp->keychal, sizeof(lp->lsk.key)); break; case LINUX_DVD_LU_SEND_CHALLENGE: memcpy(lp->lsc.chal, bp->keychal, sizeof(lp->lsc.chal)); break; case LINUX_DVD_HOST_SEND_KEY2: lp->type = LINUX_DVD_AUTH_ESTABLISHED; break; case LINUX_DVD_LU_SEND_TITLE_KEY: memcpy(lp->lstk.title_key, bp->keychal, sizeof(lp->lstk.title_key)); lp->lstk.cpm = bp->cpm; lp->lstk.cp_sec = bp->cp_sec; lp->lstk.cgms = bp->cgms; break; case LINUX_DVD_LU_SEND_ASF: lp->lsasf.asf = bp->asf; break; case LINUX_DVD_INVALIDATE_AGID: break; case LINUX_DVD_LU_SEND_RPC_STATE: lp->lrpcs.type = bp->reg_type; lp->lrpcs.vra = bp->vend_rsts; lp->lrpcs.ucca = bp->user_rsts; lp->lrpcs.region_mask = bp->region; lp->lrpcs.rpc_scheme = bp->rpc_scheme; break; case LINUX_DVD_HOST_SEND_RPC_STATE: break; default: return (EINVAL); } return (0); } static int linux_ioctl_cdrom(struct thread *td, struct linux_ioctl_args *args) { struct file *fp; int error; error = fget(td, args->fd, &cap_ioctl_rights, &fp); if (error != 0) return (error); switch (args->cmd & 0xffff) { case LINUX_CDROMPAUSE: args->cmd = CDIOCPAUSE; error = (sys_ioctl(td, (struct ioctl_args *)args)); break; case LINUX_CDROMRESUME: args->cmd = CDIOCRESUME; error = (sys_ioctl(td, (struct ioctl_args *)args)); break; case LINUX_CDROMPLAYMSF: args->cmd = CDIOCPLAYMSF; error = (sys_ioctl(td, (struct ioctl_args *)args)); break; case LINUX_CDROMPLAYTRKIND: args->cmd = CDIOCPLAYTRACKS; error = (sys_ioctl(td, (struct ioctl_args *)args)); break; case LINUX_CDROMREADTOCHDR: { struct ioc_toc_header th; struct linux_cdrom_tochdr lth; error = fo_ioctl(fp, CDIOREADTOCHEADER, (caddr_t)&th, td->td_ucred, td); if (!error) { lth.cdth_trk0 = th.starting_track; lth.cdth_trk1 = th.ending_track; copyout(<h, (void *)args->arg, sizeof(lth)); } break; } case LINUX_CDROMREADTOCENTRY: { struct linux_cdrom_tocentry lte; struct ioc_read_toc_single_entry irtse; error = copyin((void *)args->arg, <e, sizeof(lte)); if (error) break; irtse.address_format = lte.cdte_format; irtse.track = lte.cdte_track; error = fo_ioctl(fp, CDIOREADTOCENTRY, (caddr_t)&irtse, td->td_ucred, td); if (!error) { lte.cdte_ctrl = irtse.entry.control; lte.cdte_adr = irtse.entry.addr_type; bsd_to_linux_msf_lba(irtse.address_format, &irtse.entry.addr, <e.cdte_addr); error = copyout(<e, (void *)args->arg, sizeof(lte)); } break; } case LINUX_CDROMSTOP: args->cmd = CDIOCSTOP; error = (sys_ioctl(td, (struct ioctl_args *)args)); break; case LINUX_CDROMSTART: args->cmd = CDIOCSTART; error = (sys_ioctl(td, (struct ioctl_args *)args)); break; case LINUX_CDROMEJECT: args->cmd = CDIOCEJECT; error = (sys_ioctl(td, (struct ioctl_args *)args)); break; /* LINUX_CDROMVOLCTRL */ case LINUX_CDROMSUBCHNL: { struct linux_cdrom_subchnl sc; struct ioc_read_subchannel bsdsc; struct cd_sub_channel_info bsdinfo; error = copyin((void *)args->arg, &sc, sizeof(sc)); if (error) break; /* * Invoke the native ioctl and bounce the returned data through * the userspace buffer. This works because the Linux structure * is the same size as our structures for the subchannel header * and position data. */ bsdsc.address_format = CD_LBA_FORMAT; bsdsc.data_format = CD_CURRENT_POSITION; bsdsc.track = 0; bsdsc.data_len = sizeof(sc); bsdsc.data = (void *)args->arg; error = fo_ioctl(fp, CDIOCREADSUBCHANNEL, (caddr_t)&bsdsc, td->td_ucred, td); if (error) break; error = copyin((void *)args->arg, &bsdinfo, sizeof(bsdinfo)); if (error) break; sc.cdsc_audiostatus = bsdinfo.header.audio_status; sc.cdsc_adr = bsdinfo.what.position.addr_type; sc.cdsc_ctrl = bsdinfo.what.position.control; sc.cdsc_trk = bsdinfo.what.position.track_number; sc.cdsc_ind = bsdinfo.what.position.index_number; set_linux_cdrom_addr(&sc.cdsc_absaddr, sc.cdsc_format, bsdinfo.what.position.absaddr.lba); set_linux_cdrom_addr(&sc.cdsc_reladdr, sc.cdsc_format, bsdinfo.what.position.reladdr.lba); error = copyout(&sc, (void *)args->arg, sizeof(sc)); break; } /* LINUX_CDROMREADMODE2 */ /* LINUX_CDROMREADMODE1 */ /* LINUX_CDROMREADAUDIO */ /* LINUX_CDROMEJECT_SW */ /* LINUX_CDROMMULTISESSION */ /* LINUX_CDROM_GET_UPC */ case LINUX_CDROMRESET: args->cmd = CDIOCRESET; error = (sys_ioctl(td, (struct ioctl_args *)args)); break; /* LINUX_CDROMVOLREAD */ /* LINUX_CDROMREADRAW */ /* LINUX_CDROMREADCOOKED */ /* LINUX_CDROMSEEK */ /* LINUX_CDROMPLAYBLK */ /* LINUX_CDROMREADALL */ /* LINUX_CDROMCLOSETRAY */ /* LINUX_CDROMLOADFROMSLOT */ /* LINUX_CDROMGETSPINDOWN */ /* LINUX_CDROMSETSPINDOWN */ /* LINUX_CDROM_SET_OPTIONS */ /* LINUX_CDROM_CLEAR_OPTIONS */ /* LINUX_CDROM_SELECT_SPEED */ /* LINUX_CDROM_SELECT_DISC */ /* LINUX_CDROM_MEDIA_CHANGED */ /* LINUX_CDROM_DRIVE_STATUS */ /* LINUX_CDROM_DISC_STATUS */ /* LINUX_CDROM_CHANGER_NSLOTS */ /* LINUX_CDROM_LOCKDOOR */ /* LINUX_CDROM_DEBUG */ /* LINUX_CDROM_GET_CAPABILITY */ /* LINUX_CDROMAUDIOBUFSIZ */ case LINUX_DVD_READ_STRUCT: { l_dvd_struct *lds; struct dvd_struct *bds; lds = malloc(sizeof(*lds), M_LINUX, M_WAITOK); bds = malloc(sizeof(*bds), M_LINUX, M_WAITOK); error = copyin((void *)args->arg, lds, sizeof(*lds)); if (error) goto out; error = linux_to_bsd_dvd_struct(lds, bds); if (error) goto out; error = fo_ioctl(fp, DVDIOCREADSTRUCTURE, (caddr_t)bds, td->td_ucred, td); if (error) goto out; error = bsd_to_linux_dvd_struct(bds, lds); if (error) goto out; error = copyout(lds, (void *)args->arg, sizeof(*lds)); out: free(bds, M_LINUX); free(lds, M_LINUX); break; } /* LINUX_DVD_WRITE_STRUCT */ case LINUX_DVD_AUTH: { l_dvd_authinfo lda; struct dvd_authinfo bda; int bcode; error = copyin((void *)args->arg, &lda, sizeof(lda)); if (error) break; error = linux_to_bsd_dvd_authinfo(&lda, &bcode, &bda); if (error) break; error = fo_ioctl(fp, bcode, (caddr_t)&bda, td->td_ucred, td); if (error) { if (lda.type == LINUX_DVD_HOST_SEND_KEY2) { lda.type = LINUX_DVD_AUTH_FAILURE; copyout(&lda, (void *)args->arg, sizeof(lda)); } break; } error = bsd_to_linux_dvd_authinfo(&bda, &lda); if (error) break; error = copyout(&lda, (void *)args->arg, sizeof(lda)); break; } case LINUX_SCSI_GET_BUS_NUMBER: { struct sg_scsi_id id; error = fo_ioctl(fp, SG_GET_SCSI_ID, (caddr_t)&id, td->td_ucred, td); if (error) break; error = copyout(&id.channel, (void *)args->arg, sizeof(int)); break; } case LINUX_SCSI_GET_IDLUN: { struct sg_scsi_id id; struct scsi_idlun idl; error = fo_ioctl(fp, SG_GET_SCSI_ID, (caddr_t)&id, td->td_ucred, td); if (error) break; idl.dev_id = (id.scsi_id & 0xff) + ((id.lun & 0xff) << 8) + ((id.channel & 0xff) << 16) + ((id.host_no & 0xff) << 24); idl.host_unique_id = id.host_no; error = copyout(&idl, (void *)args->arg, sizeof(idl)); break; } /* LINUX_CDROM_SEND_PACKET */ /* LINUX_CDROM_NEXT_WRITABLE */ /* LINUX_CDROM_LAST_WRITTEN */ default: error = ENOIOCTL; break; } fdrop(fp, td); return (error); } static int linux_ioctl_vfat(struct thread *td, struct linux_ioctl_args *args) { return (ENOTTY); } /* * Sound related ioctls */ struct linux_old_mixer_info { char id[16]; char name[32]; }; static u_int32_t dirbits[4] = { IOC_VOID, IOC_IN, IOC_OUT, IOC_INOUT }; #define SETDIR(c) (((c) & ~IOC_DIRMASK) | dirbits[args->cmd >> 30]) static int linux_ioctl_sound(struct thread *td, struct linux_ioctl_args *args) { switch (args->cmd & 0xffff) { case LINUX_SOUND_MIXER_WRITE_VOLUME: args->cmd = SETDIR(SOUND_MIXER_WRITE_VOLUME); return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SOUND_MIXER_WRITE_BASS: args->cmd = SETDIR(SOUND_MIXER_WRITE_BASS); return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SOUND_MIXER_WRITE_TREBLE: args->cmd = SETDIR(SOUND_MIXER_WRITE_TREBLE); return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SOUND_MIXER_WRITE_SYNTH: args->cmd = SETDIR(SOUND_MIXER_WRITE_SYNTH); return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SOUND_MIXER_WRITE_PCM: args->cmd = SETDIR(SOUND_MIXER_WRITE_PCM); return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SOUND_MIXER_WRITE_SPEAKER: args->cmd = SETDIR(SOUND_MIXER_WRITE_SPEAKER); return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SOUND_MIXER_WRITE_LINE: args->cmd = SETDIR(SOUND_MIXER_WRITE_LINE); return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SOUND_MIXER_WRITE_MIC: args->cmd = SETDIR(SOUND_MIXER_WRITE_MIC); return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SOUND_MIXER_WRITE_CD: args->cmd = SETDIR(SOUND_MIXER_WRITE_CD); return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SOUND_MIXER_WRITE_IMIX: args->cmd = SETDIR(SOUND_MIXER_WRITE_IMIX); return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SOUND_MIXER_WRITE_ALTPCM: args->cmd = SETDIR(SOUND_MIXER_WRITE_ALTPCM); return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SOUND_MIXER_WRITE_RECLEV: args->cmd = SETDIR(SOUND_MIXER_WRITE_RECLEV); return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SOUND_MIXER_WRITE_IGAIN: args->cmd = SETDIR(SOUND_MIXER_WRITE_IGAIN); return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SOUND_MIXER_WRITE_OGAIN: args->cmd = SETDIR(SOUND_MIXER_WRITE_OGAIN); return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SOUND_MIXER_WRITE_LINE1: args->cmd = SETDIR(SOUND_MIXER_WRITE_LINE1); return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SOUND_MIXER_WRITE_LINE2: args->cmd = SETDIR(SOUND_MIXER_WRITE_LINE2); return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SOUND_MIXER_WRITE_LINE3: args->cmd = SETDIR(SOUND_MIXER_WRITE_LINE3); return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SOUND_MIXER_INFO: { /* Key on encoded length */ switch ((args->cmd >> 16) & 0x1fff) { case 0x005c: { /* SOUND_MIXER_INFO */ args->cmd = SOUND_MIXER_INFO; return (sys_ioctl(td, (struct ioctl_args *)args)); } case 0x0030: { /* SOUND_OLD_MIXER_INFO */ struct linux_old_mixer_info info; bzero(&info, sizeof(info)); strncpy(info.id, "OSS", sizeof(info.id) - 1); strncpy(info.name, "FreeBSD OSS Mixer", sizeof(info.name) - 1); copyout(&info, (void *)args->arg, sizeof(info)); return (0); } default: return (ENOIOCTL); } break; } case LINUX_OSS_GETVERSION: { int version = linux_get_oss_version(td); return (copyout(&version, (void *)args->arg, sizeof(int))); } case LINUX_SOUND_MIXER_READ_STEREODEVS: args->cmd = SOUND_MIXER_READ_STEREODEVS; return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SOUND_MIXER_READ_CAPS: args->cmd = SOUND_MIXER_READ_CAPS; return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SOUND_MIXER_READ_RECMASK: args->cmd = SOUND_MIXER_READ_RECMASK; return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SOUND_MIXER_READ_DEVMASK: args->cmd = SOUND_MIXER_READ_DEVMASK; return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SOUND_MIXER_WRITE_RECSRC: args->cmd = SETDIR(SOUND_MIXER_WRITE_RECSRC); return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SNDCTL_DSP_RESET: args->cmd = SNDCTL_DSP_RESET; return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SNDCTL_DSP_SYNC: args->cmd = SNDCTL_DSP_SYNC; return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SNDCTL_DSP_SPEED: args->cmd = SNDCTL_DSP_SPEED; return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SNDCTL_DSP_STEREO: args->cmd = SNDCTL_DSP_STEREO; return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SNDCTL_DSP_GETBLKSIZE: /* LINUX_SNDCTL_DSP_SETBLKSIZE */ args->cmd = SNDCTL_DSP_GETBLKSIZE; return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SNDCTL_DSP_SETFMT: args->cmd = SNDCTL_DSP_SETFMT; return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SOUND_PCM_WRITE_CHANNELS: args->cmd = SOUND_PCM_WRITE_CHANNELS; return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SOUND_PCM_WRITE_FILTER: args->cmd = SOUND_PCM_WRITE_FILTER; return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SNDCTL_DSP_POST: args->cmd = SNDCTL_DSP_POST; return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SNDCTL_DSP_SUBDIVIDE: args->cmd = SNDCTL_DSP_SUBDIVIDE; return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SNDCTL_DSP_SETFRAGMENT: args->cmd = SNDCTL_DSP_SETFRAGMENT; return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SNDCTL_DSP_GETFMTS: args->cmd = SNDCTL_DSP_GETFMTS; return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SNDCTL_DSP_GETOSPACE: args->cmd = SNDCTL_DSP_GETOSPACE; return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SNDCTL_DSP_GETISPACE: args->cmd = SNDCTL_DSP_GETISPACE; return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SNDCTL_DSP_NONBLOCK: args->cmd = SNDCTL_DSP_NONBLOCK; return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SNDCTL_DSP_GETCAPS: args->cmd = SNDCTL_DSP_GETCAPS; return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SNDCTL_DSP_SETTRIGGER: /* LINUX_SNDCTL_GETTRIGGER */ args->cmd = SNDCTL_DSP_SETTRIGGER; return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SNDCTL_DSP_GETIPTR: args->cmd = SNDCTL_DSP_GETIPTR; return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SNDCTL_DSP_GETOPTR: args->cmd = SNDCTL_DSP_GETOPTR; return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SNDCTL_DSP_SETDUPLEX: args->cmd = SNDCTL_DSP_SETDUPLEX; return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SNDCTL_DSP_GETODELAY: args->cmd = SNDCTL_DSP_GETODELAY; return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SNDCTL_SEQ_RESET: args->cmd = SNDCTL_SEQ_RESET; return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SNDCTL_SEQ_SYNC: args->cmd = SNDCTL_SEQ_SYNC; return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SNDCTL_SYNTH_INFO: args->cmd = SNDCTL_SYNTH_INFO; return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SNDCTL_SEQ_CTRLRATE: args->cmd = SNDCTL_SEQ_CTRLRATE; return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SNDCTL_SEQ_GETOUTCOUNT: args->cmd = SNDCTL_SEQ_GETOUTCOUNT; return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SNDCTL_SEQ_GETINCOUNT: args->cmd = SNDCTL_SEQ_GETINCOUNT; return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SNDCTL_SEQ_PERCMODE: args->cmd = SNDCTL_SEQ_PERCMODE; return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SNDCTL_FM_LOAD_INSTR: args->cmd = SNDCTL_FM_LOAD_INSTR; return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SNDCTL_SEQ_TESTMIDI: args->cmd = SNDCTL_SEQ_TESTMIDI; return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SNDCTL_SEQ_RESETSAMPLES: args->cmd = SNDCTL_SEQ_RESETSAMPLES; return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SNDCTL_SEQ_NRSYNTHS: args->cmd = SNDCTL_SEQ_NRSYNTHS; return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SNDCTL_SEQ_NRMIDIS: args->cmd = SNDCTL_SEQ_NRMIDIS; return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SNDCTL_MIDI_INFO: args->cmd = SNDCTL_MIDI_INFO; return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SNDCTL_SEQ_TRESHOLD: args->cmd = SNDCTL_SEQ_TRESHOLD; return (sys_ioctl(td, (struct ioctl_args *)args)); case LINUX_SNDCTL_SYNTH_MEMAVL: args->cmd = SNDCTL_SYNTH_MEMAVL; return (sys_ioctl(td, (struct ioctl_args *)args)); } return (ENOIOCTL); } /* * Console related ioctls */ static int linux_ioctl_console(struct thread *td, struct linux_ioctl_args *args) { struct file *fp; int error; error = fget(td, args->fd, &cap_ioctl_rights, &fp); if (error != 0) return (error); switch (args->cmd & 0xffff) { case LINUX_KIOCSOUND: args->cmd = KIOCSOUND; error = (sys_ioctl(td, (struct ioctl_args *)args)); break; case LINUX_KDMKTONE: args->cmd = KDMKTONE; error = (sys_ioctl(td, (struct ioctl_args *)args)); break; case LINUX_KDGETLED: args->cmd = KDGETLED; error = (sys_ioctl(td, (struct ioctl_args *)args)); break; case LINUX_KDSETLED: args->cmd = KDSETLED; error = (sys_ioctl(td, (struct ioctl_args *)args)); break; case LINUX_KDSETMODE: args->cmd = KDSETMODE; error = (sys_ioctl(td, (struct ioctl_args *)args)); break; case LINUX_KDGETMODE: args->cmd = KDGETMODE; error = (sys_ioctl(td, (struct ioctl_args *)args)); break; case LINUX_KDGKBMODE: args->cmd = KDGKBMODE; error = (sys_ioctl(td, (struct ioctl_args *)args)); break; case LINUX_KDSKBMODE: { int kbdmode; switch (args->arg) { case LINUX_KBD_RAW: kbdmode = K_RAW; break; case LINUX_KBD_XLATE: kbdmode = K_XLATE; break; case LINUX_KBD_MEDIUMRAW: kbdmode = K_RAW; break; default: fdrop(fp, td); return (EINVAL); } error = (fo_ioctl(fp, KDSKBMODE, (caddr_t)&kbdmode, td->td_ucred, td)); break; } case LINUX_VT_OPENQRY: args->cmd = VT_OPENQRY; error = (sys_ioctl(td, (struct ioctl_args *)args)); break; case LINUX_VT_GETMODE: args->cmd = VT_GETMODE; error = (sys_ioctl(td, (struct ioctl_args *)args)); break; case LINUX_VT_SETMODE: { struct vt_mode mode; if ((error = copyin((void *)args->arg, &mode, sizeof(mode)))) break; if (LINUX_SIG_VALID(mode.relsig)) mode.relsig = linux_to_bsd_signal(mode.relsig); else mode.relsig = 0; if (LINUX_SIG_VALID(mode.acqsig)) mode.acqsig = linux_to_bsd_signal(mode.acqsig); else mode.acqsig = 0; /* XXX. Linux ignores frsig and set it to 0. */ mode.frsig = 0; if ((error = copyout(&mode, (void *)args->arg, sizeof(mode)))) break; args->cmd = VT_SETMODE; error = (sys_ioctl(td, (struct ioctl_args *)args)); break; } case LINUX_VT_GETSTATE: args->cmd = VT_GETACTIVE; error = (sys_ioctl(td, (struct ioctl_args *)args)); break; case LINUX_VT_RELDISP: args->cmd = VT_RELDISP; error = (sys_ioctl(td, (struct ioctl_args *)args)); break; case LINUX_VT_ACTIVATE: args->cmd = VT_ACTIVATE; error = (sys_ioctl(td, (struct ioctl_args *)args)); break; case LINUX_VT_WAITACTIVE: args->cmd = VT_WAITACTIVE; error = (sys_ioctl(td, (struct ioctl_args *)args)); break; default: error = ENOIOCTL; break; } fdrop(fp, td); return (error); } /* * Implement the SIOCGIFNAME ioctl */ static int linux_ioctl_ifname(struct thread *td, struct l_ifreq *uifr) { struct l_ifreq ifr; struct ifnet *ifp; int error, ethno, index; error = copyin(uifr, &ifr, sizeof(ifr)); if (error != 0) return (error); CURVNET_SET(TD_TO_VNET(curthread)); IFNET_RLOCK(); index = 1; /* ifr.ifr_ifindex starts from 1 */ ethno = 0; error = ENODEV; CK_STAILQ_FOREACH(ifp, &V_ifnet, if_link) { if (ifr.ifr_ifindex == index) { if (IFP_IS_ETH(ifp)) snprintf(ifr.ifr_name, LINUX_IFNAMSIZ, "eth%d", ethno); else strlcpy(ifr.ifr_name, ifp->if_xname, LINUX_IFNAMSIZ); error = 0; break; } if (IFP_IS_ETH(ifp)) ethno++; index++; } IFNET_RUNLOCK(); if (error == 0) error = copyout(&ifr, uifr, sizeof(ifr)); CURVNET_RESTORE(); return (error); } /* * Implement the SIOCGIFCONF ioctl */ static int linux_ifconf(struct thread *td, struct ifconf *uifc) { #ifdef COMPAT_LINUX32 struct l_ifconf ifc; #else struct ifconf ifc; #endif struct l_ifreq ifr; struct ifnet *ifp; struct ifaddr *ifa; struct sbuf *sb; int error, ethno, full = 0, valid_len, max_len; error = copyin(uifc, &ifc, sizeof(ifc)); if (error != 0) return (error); max_len = MAXPHYS - 1; CURVNET_SET(TD_TO_VNET(td)); /* handle the 'request buffer size' case */ if ((l_uintptr_t)ifc.ifc_buf == PTROUT(NULL)) { ifc.ifc_len = 0; IFNET_RLOCK(); CK_STAILQ_FOREACH(ifp, &V_ifnet, if_link) { CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) { struct sockaddr *sa = ifa->ifa_addr; if (sa->sa_family == AF_INET) ifc.ifc_len += sizeof(ifr); } } IFNET_RUNLOCK(); error = copyout(&ifc, uifc, sizeof(ifc)); CURVNET_RESTORE(); return (error); } if (ifc.ifc_len <= 0) { CURVNET_RESTORE(); return (EINVAL); } again: /* Keep track of eth interfaces */ ethno = 0; if (ifc.ifc_len <= max_len) { max_len = ifc.ifc_len; full = 1; } sb = sbuf_new(NULL, NULL, max_len + 1, SBUF_FIXEDLEN); max_len = 0; valid_len = 0; /* Return all AF_INET addresses of all interfaces */ IFNET_RLOCK(); CK_STAILQ_FOREACH(ifp, &V_ifnet, if_link) { int addrs = 0; bzero(&ifr, sizeof(ifr)); if (IFP_IS_ETH(ifp)) snprintf(ifr.ifr_name, LINUX_IFNAMSIZ, "eth%d", ethno++); else strlcpy(ifr.ifr_name, ifp->if_xname, LINUX_IFNAMSIZ); /* Walk the address list */ CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) { struct sockaddr *sa = ifa->ifa_addr; if (sa->sa_family == AF_INET) { ifr.ifr_addr.sa_family = LINUX_AF_INET; memcpy(ifr.ifr_addr.sa_data, sa->sa_data, sizeof(ifr.ifr_addr.sa_data)); sbuf_bcat(sb, &ifr, sizeof(ifr)); max_len += sizeof(ifr); addrs++; } if (sbuf_error(sb) == 0) valid_len = sbuf_len(sb); } if (addrs == 0) { bzero((caddr_t)&ifr.ifr_addr, sizeof(ifr.ifr_addr)); sbuf_bcat(sb, &ifr, sizeof(ifr)); max_len += sizeof(ifr); if (sbuf_error(sb) == 0) valid_len = sbuf_len(sb); } } IFNET_RUNLOCK(); if (valid_len != max_len && !full) { sbuf_delete(sb); goto again; } ifc.ifc_len = valid_len; sbuf_finish(sb); error = copyout(sbuf_data(sb), PTRIN(ifc.ifc_buf), ifc.ifc_len); if (error == 0) error = copyout(&ifc, uifc, sizeof(ifc)); sbuf_delete(sb); CURVNET_RESTORE(); return (error); } static int linux_gifflags(struct thread *td, struct ifnet *ifp, struct l_ifreq *ifr) { l_short flags; linux_ifflags(ifp, &flags); return (copyout(&flags, &ifr->ifr_flags, sizeof(flags))); } static int linux_gifhwaddr(struct ifnet *ifp, struct l_ifreq *ifr) { struct l_sockaddr lsa; if (linux_ifhwaddr(ifp, &lsa) != 0) return (ENOENT); return (copyout(&lsa, &ifr->ifr_hwaddr, sizeof(lsa))); } /* * If we fault in bsd_to_linux_ifreq() then we will fault when we call * the native ioctl(). Thus, we don't really need to check the return * value of this function. */ static int bsd_to_linux_ifreq(struct ifreq *arg) { struct ifreq ifr; size_t ifr_len = sizeof(struct ifreq); int error; if ((error = copyin(arg, &ifr, ifr_len))) return (error); *(u_short *)&ifr.ifr_addr = ifr.ifr_addr.sa_family; error = copyout(&ifr, arg, ifr_len); return (error); } /* * Socket related ioctls */ static int linux_ioctl_socket(struct thread *td, struct linux_ioctl_args *args) { char lifname[LINUX_IFNAMSIZ], ifname[IFNAMSIZ]; struct ifnet *ifp; struct file *fp; int error, type; ifp = NULL; error = 0; error = fget(td, args->fd, &cap_ioctl_rights, &fp); if (error != 0) return (error); type = fp->f_type; fdrop(fp, td); if (type != DTYPE_SOCKET) { /* not a socket - probably a tap / vmnet device */ switch (args->cmd) { case LINUX_SIOCGIFADDR: case LINUX_SIOCSIFADDR: case LINUX_SIOCGIFFLAGS: return (linux_ioctl_special(td, args)); default: return (ENOIOCTL); } } switch (args->cmd & 0xffff) { case LINUX_FIOGETOWN: case LINUX_FIOSETOWN: case LINUX_SIOCADDMULTI: case LINUX_SIOCATMARK: case LINUX_SIOCDELMULTI: case LINUX_SIOCGIFNAME: case LINUX_SIOCGIFCONF: case LINUX_SIOCGPGRP: case LINUX_SIOCSPGRP: case LINUX_SIOCGIFCOUNT: /* these ioctls don't take an interface name */ break; case LINUX_SIOCGIFFLAGS: case LINUX_SIOCGIFADDR: case LINUX_SIOCSIFADDR: case LINUX_SIOCGIFDSTADDR: case LINUX_SIOCGIFBRDADDR: case LINUX_SIOCGIFNETMASK: case LINUX_SIOCSIFNETMASK: case LINUX_SIOCGIFMTU: case LINUX_SIOCSIFMTU: case LINUX_SIOCSIFNAME: case LINUX_SIOCGIFHWADDR: case LINUX_SIOCSIFHWADDR: case LINUX_SIOCDEVPRIVATE: case LINUX_SIOCDEVPRIVATE+1: case LINUX_SIOCGIFINDEX: /* copy in the interface name and translate it. */ error = copyin((void *)args->arg, lifname, LINUX_IFNAMSIZ); if (error != 0) return (error); memset(ifname, 0, sizeof(ifname)); ifp = ifname_linux_to_bsd(td, lifname, ifname); if (ifp == NULL) return (EINVAL); /* * We need to copy it back out in case we pass the * request on to our native ioctl(), which will expect * the ifreq to be in user space and have the correct * interface name. */ error = copyout(ifname, (void *)args->arg, IFNAMSIZ); if (error != 0) return (error); break; default: return (ENOIOCTL); } switch (args->cmd & 0xffff) { case LINUX_FIOSETOWN: args->cmd = FIOSETOWN; error = sys_ioctl(td, (struct ioctl_args *)args); break; case LINUX_SIOCSPGRP: args->cmd = SIOCSPGRP; error = sys_ioctl(td, (struct ioctl_args *)args); break; case LINUX_FIOGETOWN: args->cmd = FIOGETOWN; error = sys_ioctl(td, (struct ioctl_args *)args); break; case LINUX_SIOCGPGRP: args->cmd = SIOCGPGRP; error = sys_ioctl(td, (struct ioctl_args *)args); break; case LINUX_SIOCATMARK: args->cmd = SIOCATMARK; error = sys_ioctl(td, (struct ioctl_args *)args); break; /* LINUX_SIOCGSTAMP */ case LINUX_SIOCGIFNAME: error = linux_ioctl_ifname(td, (struct l_ifreq *)args->arg); break; case LINUX_SIOCGIFCONF: error = linux_ifconf(td, (struct ifconf *)args->arg); break; case LINUX_SIOCGIFFLAGS: args->cmd = SIOCGIFFLAGS; error = linux_gifflags(td, ifp, (struct l_ifreq *)args->arg); break; case LINUX_SIOCGIFADDR: args->cmd = SIOCGIFADDR; error = sys_ioctl(td, (struct ioctl_args *)args); bsd_to_linux_ifreq((struct ifreq *)args->arg); break; case LINUX_SIOCSIFADDR: /* XXX probably doesn't work, included for completeness */ args->cmd = SIOCSIFADDR; error = sys_ioctl(td, (struct ioctl_args *)args); break; case LINUX_SIOCGIFDSTADDR: args->cmd = SIOCGIFDSTADDR; error = sys_ioctl(td, (struct ioctl_args *)args); bsd_to_linux_ifreq((struct ifreq *)args->arg); break; case LINUX_SIOCGIFBRDADDR: args->cmd = SIOCGIFBRDADDR; error = sys_ioctl(td, (struct ioctl_args *)args); bsd_to_linux_ifreq((struct ifreq *)args->arg); break; case LINUX_SIOCGIFNETMASK: args->cmd = SIOCGIFNETMASK; error = sys_ioctl(td, (struct ioctl_args *)args); bsd_to_linux_ifreq((struct ifreq *)args->arg); break; case LINUX_SIOCSIFNETMASK: error = ENOIOCTL; break; case LINUX_SIOCGIFMTU: args->cmd = SIOCGIFMTU; error = sys_ioctl(td, (struct ioctl_args *)args); break; case LINUX_SIOCSIFMTU: args->cmd = SIOCSIFMTU; error = sys_ioctl(td, (struct ioctl_args *)args); break; case LINUX_SIOCSIFNAME: error = ENOIOCTL; break; case LINUX_SIOCGIFHWADDR: error = linux_gifhwaddr(ifp, (struct l_ifreq *)args->arg); break; case LINUX_SIOCSIFHWADDR: error = ENOIOCTL; break; case LINUX_SIOCADDMULTI: args->cmd = SIOCADDMULTI; error = sys_ioctl(td, (struct ioctl_args *)args); break; case LINUX_SIOCDELMULTI: args->cmd = SIOCDELMULTI; error = sys_ioctl(td, (struct ioctl_args *)args); break; case LINUX_SIOCGIFINDEX: args->cmd = SIOCGIFINDEX; error = sys_ioctl(td, (struct ioctl_args *)args); break; case LINUX_SIOCGIFCOUNT: error = 0; break; /* * XXX This is slightly bogus, but these ioctls are currently * XXX only used by the aironet (if_an) network driver. */ case LINUX_SIOCDEVPRIVATE: args->cmd = SIOCGPRIVATE_0; error = sys_ioctl(td, (struct ioctl_args *)args); break; case LINUX_SIOCDEVPRIVATE+1: args->cmd = SIOCGPRIVATE_1; error = sys_ioctl(td, (struct ioctl_args *)args); break; } if (ifp != NULL) /* restore the original interface name */ copyout(lifname, (void *)args->arg, LINUX_IFNAMSIZ); return (error); } /* * Device private ioctl handler */ static int linux_ioctl_private(struct thread *td, struct linux_ioctl_args *args) { struct file *fp; int error, type; error = fget(td, args->fd, &cap_ioctl_rights, &fp); if (error != 0) return (error); type = fp->f_type; fdrop(fp, td); if (type == DTYPE_SOCKET) return (linux_ioctl_socket(td, args)); return (ENOIOCTL); } /* * DRM ioctl handler (sys/dev/drm) */ static int linux_ioctl_drm(struct thread *td, struct linux_ioctl_args *args) { args->cmd = SETDIR(args->cmd); return (sys_ioctl(td, (struct ioctl_args *)args)); } #ifdef COMPAT_LINUX32 #define CP(src,dst,fld) do { (dst).fld = (src).fld; } while (0) #define PTRIN_CP(src,dst,fld) \ do { (dst).fld = PTRIN((src).fld); } while (0) #define PTROUT_CP(src,dst,fld) \ do { (dst).fld = PTROUT((src).fld); } while (0) static int linux_ioctl_sg_io(struct thread *td, struct linux_ioctl_args *args) { struct sg_io_hdr io; struct sg_io_hdr32 io32; struct file *fp; int error; error = fget(td, args->fd, &cap_ioctl_rights, &fp); if (error != 0) { printf("sg_linux_ioctl: fget returned %d\n", error); return (error); } if ((error = copyin((void *)args->arg, &io32, sizeof(io32))) != 0) goto out; CP(io32, io, interface_id); CP(io32, io, dxfer_direction); CP(io32, io, cmd_len); CP(io32, io, mx_sb_len); CP(io32, io, iovec_count); CP(io32, io, dxfer_len); PTRIN_CP(io32, io, dxferp); PTRIN_CP(io32, io, cmdp); PTRIN_CP(io32, io, sbp); CP(io32, io, timeout); CP(io32, io, flags); CP(io32, io, pack_id); PTRIN_CP(io32, io, usr_ptr); CP(io32, io, status); CP(io32, io, masked_status); CP(io32, io, msg_status); CP(io32, io, sb_len_wr); CP(io32, io, host_status); CP(io32, io, driver_status); CP(io32, io, resid); CP(io32, io, duration); CP(io32, io, info); if ((error = fo_ioctl(fp, SG_IO, (caddr_t)&io, td->td_ucred, td)) != 0) goto out; CP(io, io32, interface_id); CP(io, io32, dxfer_direction); CP(io, io32, cmd_len); CP(io, io32, mx_sb_len); CP(io, io32, iovec_count); CP(io, io32, dxfer_len); PTROUT_CP(io, io32, dxferp); PTROUT_CP(io, io32, cmdp); PTROUT_CP(io, io32, sbp); CP(io, io32, timeout); CP(io, io32, flags); CP(io, io32, pack_id); PTROUT_CP(io, io32, usr_ptr); CP(io, io32, status); CP(io, io32, masked_status); CP(io, io32, msg_status); CP(io, io32, sb_len_wr); CP(io, io32, host_status); CP(io, io32, driver_status); CP(io, io32, resid); CP(io, io32, duration); CP(io, io32, info); error = copyout(&io32, (void *)args->arg, sizeof(io32)); out: fdrop(fp, td); return (error); } #endif static int linux_ioctl_sg(struct thread *td, struct linux_ioctl_args *args) { switch (args->cmd) { case LINUX_SG_GET_VERSION_NUM: args->cmd = SG_GET_VERSION_NUM; break; case LINUX_SG_SET_TIMEOUT: args->cmd = SG_SET_TIMEOUT; break; case LINUX_SG_GET_TIMEOUT: args->cmd = SG_GET_TIMEOUT; break; case LINUX_SG_IO: args->cmd = SG_IO; #ifdef COMPAT_LINUX32 return (linux_ioctl_sg_io(td, args)); #endif break; case LINUX_SG_GET_RESERVED_SIZE: args->cmd = SG_GET_RESERVED_SIZE; break; case LINUX_SG_GET_SCSI_ID: args->cmd = SG_GET_SCSI_ID; break; case LINUX_SG_GET_SG_TABLESIZE: args->cmd = SG_GET_SG_TABLESIZE; break; default: return (ENODEV); } return (sys_ioctl(td, (struct ioctl_args *)args)); } /* * Video4Linux (V4L) ioctl handler */ static int linux_to_bsd_v4l_tuner(struct l_video_tuner *lvt, struct video_tuner *vt) { vt->tuner = lvt->tuner; strlcpy(vt->name, lvt->name, LINUX_VIDEO_TUNER_NAME_SIZE); vt->rangelow = lvt->rangelow; /* possible long size conversion */ vt->rangehigh = lvt->rangehigh; /* possible long size conversion */ vt->flags = lvt->flags; vt->mode = lvt->mode; vt->signal = lvt->signal; return (0); } static int bsd_to_linux_v4l_tuner(struct video_tuner *vt, struct l_video_tuner *lvt) { lvt->tuner = vt->tuner; strlcpy(lvt->name, vt->name, LINUX_VIDEO_TUNER_NAME_SIZE); lvt->rangelow = vt->rangelow; /* possible long size conversion */ lvt->rangehigh = vt->rangehigh; /* possible long size conversion */ lvt->flags = vt->flags; lvt->mode = vt->mode; lvt->signal = vt->signal; return (0); } #ifdef COMPAT_LINUX_V4L_CLIPLIST static int linux_to_bsd_v4l_clip(struct l_video_clip *lvc, struct video_clip *vc) { vc->x = lvc->x; vc->y = lvc->y; vc->width = lvc->width; vc->height = lvc->height; vc->next = PTRIN(lvc->next); /* possible pointer size conversion */ return (0); } #endif static int linux_to_bsd_v4l_window(struct l_video_window *lvw, struct video_window *vw) { vw->x = lvw->x; vw->y = lvw->y; vw->width = lvw->width; vw->height = lvw->height; vw->chromakey = lvw->chromakey; vw->flags = lvw->flags; vw->clips = PTRIN(lvw->clips); /* possible pointer size conversion */ vw->clipcount = lvw->clipcount; return (0); } static int bsd_to_linux_v4l_window(struct video_window *vw, struct l_video_window *lvw) { memset(lvw, 0, sizeof(*lvw)); lvw->x = vw->x; lvw->y = vw->y; lvw->width = vw->width; lvw->height = vw->height; lvw->chromakey = vw->chromakey; lvw->flags = vw->flags; lvw->clips = PTROUT(vw->clips); /* possible pointer size conversion */ lvw->clipcount = vw->clipcount; return (0); } static int linux_to_bsd_v4l_buffer(struct l_video_buffer *lvb, struct video_buffer *vb) { vb->base = PTRIN(lvb->base); /* possible pointer size conversion */ vb->height = lvb->height; vb->width = lvb->width; vb->depth = lvb->depth; vb->bytesperline = lvb->bytesperline; return (0); } static int bsd_to_linux_v4l_buffer(struct video_buffer *vb, struct l_video_buffer *lvb) { lvb->base = PTROUT(vb->base); /* possible pointer size conversion */ lvb->height = vb->height; lvb->width = vb->width; lvb->depth = vb->depth; lvb->bytesperline = vb->bytesperline; return (0); } static int linux_to_bsd_v4l_code(struct l_video_code *lvc, struct video_code *vc) { strlcpy(vc->loadwhat, lvc->loadwhat, LINUX_VIDEO_CODE_LOADWHAT_SIZE); vc->datasize = lvc->datasize; vc->data = PTRIN(lvc->data); /* possible pointer size conversion */ return (0); } #ifdef COMPAT_LINUX_V4L_CLIPLIST static int linux_v4l_clip_copy(void *lvc, struct video_clip **ppvc) { int error; struct video_clip vclip; struct l_video_clip l_vclip; error = copyin(lvc, &l_vclip, sizeof(l_vclip)); if (error) return (error); linux_to_bsd_v4l_clip(&l_vclip, &vclip); /* XXX: If there can be no concurrency: s/M_NOWAIT/M_WAITOK/ */ if ((*ppvc = malloc(sizeof(**ppvc), M_LINUX, M_NOWAIT)) == NULL) return (ENOMEM); /* XXX: Linux has no ENOMEM here. */ memcpy(*ppvc, &vclip, sizeof(vclip)); (*ppvc)->next = NULL; return (0); } static int linux_v4l_cliplist_free(struct video_window *vw) { struct video_clip **ppvc; struct video_clip **ppvc_next; for (ppvc = &(vw->clips); *ppvc != NULL; ppvc = ppvc_next) { ppvc_next = &((*ppvc)->next); free(*ppvc, M_LINUX); } vw->clips = NULL; return (0); } static int linux_v4l_cliplist_copy(struct l_video_window *lvw, struct video_window *vw) { int error; int clipcount; void *plvc; struct video_clip **ppvc; /* * XXX: The cliplist is used to pass in a list of clipping * rectangles or, if clipcount == VIDEO_CLIP_BITMAP, a * clipping bitmap. Some Linux apps, however, appear to * leave cliplist and clips uninitialized. In any case, * the cliplist is not used by pwc(4), at the time of * writing, FreeBSD's only V4L driver. When a driver * that uses the cliplist is developed, this code may * need re-examiniation. */ error = 0; clipcount = vw->clipcount; if (clipcount == VIDEO_CLIP_BITMAP) { /* * In this case, the pointer (clips) is overloaded * to be a "void *" to a bitmap, therefore there * is no struct video_clip to copy now. */ } else if (clipcount > 0 && clipcount <= 16384) { /* * Clips points to list of clip rectangles, so * copy the list. * * XXX: Upper limit of 16384 was used here to try to * avoid cases when clipcount and clips pointer * are uninitialized and therefore have high random * values, as is the case in the Linux Skype * application. The value 16384 was chosen as that * is what is used in the Linux stradis(4) MPEG * decoder driver, the only place we found an * example of cliplist use. */ plvc = PTRIN(lvw->clips); vw->clips = NULL; ppvc = &(vw->clips); while (clipcount-- > 0) { if (plvc == NULL) { error = EFAULT; break; } else { error = linux_v4l_clip_copy(plvc, ppvc); if (error) { linux_v4l_cliplist_free(vw); break; } } ppvc = &((*ppvc)->next); plvc = PTRIN(((struct l_video_clip *) plvc)->next); } } else { /* * clipcount == 0 or negative (but not VIDEO_CLIP_BITMAP) * Force cliplist to null. */ vw->clipcount = 0; vw->clips = NULL; } return (error); } #endif static int linux_ioctl_v4l(struct thread *td, struct linux_ioctl_args *args) { struct file *fp; int error; struct video_tuner vtun; struct video_window vwin; struct video_buffer vbuf; struct video_code vcode; struct l_video_tuner l_vtun; struct l_video_window l_vwin; struct l_video_buffer l_vbuf; struct l_video_code l_vcode; switch (args->cmd & 0xffff) { case LINUX_VIDIOCGCAP: args->cmd = VIDIOCGCAP; break; case LINUX_VIDIOCGCHAN: args->cmd = VIDIOCGCHAN; break; case LINUX_VIDIOCSCHAN: args->cmd = VIDIOCSCHAN; break; case LINUX_VIDIOCGTUNER: error = fget(td, args->fd, &cap_ioctl_rights, &fp); if (error != 0) return (error); error = copyin((void *) args->arg, &l_vtun, sizeof(l_vtun)); if (error) { fdrop(fp, td); return (error); } linux_to_bsd_v4l_tuner(&l_vtun, &vtun); error = fo_ioctl(fp, VIDIOCGTUNER, &vtun, td->td_ucred, td); if (!error) { bsd_to_linux_v4l_tuner(&vtun, &l_vtun); error = copyout(&l_vtun, (void *) args->arg, sizeof(l_vtun)); } fdrop(fp, td); return (error); case LINUX_VIDIOCSTUNER: error = fget(td, args->fd, &cap_ioctl_rights, &fp); if (error != 0) return (error); error = copyin((void *) args->arg, &l_vtun, sizeof(l_vtun)); if (error) { fdrop(fp, td); return (error); } linux_to_bsd_v4l_tuner(&l_vtun, &vtun); error = fo_ioctl(fp, VIDIOCSTUNER, &vtun, td->td_ucred, td); fdrop(fp, td); return (error); case LINUX_VIDIOCGPICT: args->cmd = VIDIOCGPICT; break; case LINUX_VIDIOCSPICT: args->cmd = VIDIOCSPICT; break; case LINUX_VIDIOCCAPTURE: args->cmd = VIDIOCCAPTURE; break; case LINUX_VIDIOCGWIN: error = fget(td, args->fd, &cap_ioctl_rights, &fp); if (error != 0) return (error); error = fo_ioctl(fp, VIDIOCGWIN, &vwin, td->td_ucred, td); if (!error) { bsd_to_linux_v4l_window(&vwin, &l_vwin); error = copyout(&l_vwin, (void *) args->arg, sizeof(l_vwin)); } fdrop(fp, td); return (error); case LINUX_VIDIOCSWIN: error = fget(td, args->fd, &cap_ioctl_rights, &fp); if (error != 0) return (error); error = copyin((void *) args->arg, &l_vwin, sizeof(l_vwin)); if (error) { fdrop(fp, td); return (error); } linux_to_bsd_v4l_window(&l_vwin, &vwin); #ifdef COMPAT_LINUX_V4L_CLIPLIST error = linux_v4l_cliplist_copy(&l_vwin, &vwin); if (error) { fdrop(fp, td); return (error); } #endif error = fo_ioctl(fp, VIDIOCSWIN, &vwin, td->td_ucred, td); fdrop(fp, td); #ifdef COMPAT_LINUX_V4L_CLIPLIST linux_v4l_cliplist_free(&vwin); #endif return (error); case LINUX_VIDIOCGFBUF: error = fget(td, args->fd, &cap_ioctl_rights, &fp); if (error != 0) return (error); error = fo_ioctl(fp, VIDIOCGFBUF, &vbuf, td->td_ucred, td); if (!error) { bsd_to_linux_v4l_buffer(&vbuf, &l_vbuf); error = copyout(&l_vbuf, (void *) args->arg, sizeof(l_vbuf)); } fdrop(fp, td); return (error); case LINUX_VIDIOCSFBUF: error = fget(td, args->fd, &cap_ioctl_rights, &fp); if (error != 0) return (error); error = copyin((void *) args->arg, &l_vbuf, sizeof(l_vbuf)); if (error) { fdrop(fp, td); return (error); } linux_to_bsd_v4l_buffer(&l_vbuf, &vbuf); error = fo_ioctl(fp, VIDIOCSFBUF, &vbuf, td->td_ucred, td); fdrop(fp, td); return (error); case LINUX_VIDIOCKEY: args->cmd = VIDIOCKEY; break; case LINUX_VIDIOCGFREQ: args->cmd = VIDIOCGFREQ; break; case LINUX_VIDIOCSFREQ: args->cmd = VIDIOCSFREQ; break; case LINUX_VIDIOCGAUDIO: args->cmd = VIDIOCGAUDIO; break; case LINUX_VIDIOCSAUDIO: args->cmd = VIDIOCSAUDIO; break; case LINUX_VIDIOCSYNC: args->cmd = VIDIOCSYNC; break; case LINUX_VIDIOCMCAPTURE: args->cmd = VIDIOCMCAPTURE; break; case LINUX_VIDIOCGMBUF: args->cmd = VIDIOCGMBUF; break; case LINUX_VIDIOCGUNIT: args->cmd = VIDIOCGUNIT; break; case LINUX_VIDIOCGCAPTURE: args->cmd = VIDIOCGCAPTURE; break; case LINUX_VIDIOCSCAPTURE: args->cmd = VIDIOCSCAPTURE; break; case LINUX_VIDIOCSPLAYMODE: args->cmd = VIDIOCSPLAYMODE; break; case LINUX_VIDIOCSWRITEMODE: args->cmd = VIDIOCSWRITEMODE; break; case LINUX_VIDIOCGPLAYINFO: args->cmd = VIDIOCGPLAYINFO; break; case LINUX_VIDIOCSMICROCODE: error = fget(td, args->fd, &cap_ioctl_rights, &fp); if (error != 0) return (error); error = copyin((void *) args->arg, &l_vcode, sizeof(l_vcode)); if (error) { fdrop(fp, td); return (error); } linux_to_bsd_v4l_code(&l_vcode, &vcode); error = fo_ioctl(fp, VIDIOCSMICROCODE, &vcode, td->td_ucred, td); fdrop(fp, td); return (error); case LINUX_VIDIOCGVBIFMT: args->cmd = VIDIOCGVBIFMT; break; case LINUX_VIDIOCSVBIFMT: args->cmd = VIDIOCSVBIFMT; break; default: return (ENOIOCTL); } error = sys_ioctl(td, (struct ioctl_args *)args); return (error); } /* * Special ioctl handler */ static int linux_ioctl_special(struct thread *td, struct linux_ioctl_args *args) { int error; switch (args->cmd) { case LINUX_SIOCGIFADDR: args->cmd = SIOCGIFADDR; error = sys_ioctl(td, (struct ioctl_args *)args); break; case LINUX_SIOCSIFADDR: args->cmd = SIOCSIFADDR; error = sys_ioctl(td, (struct ioctl_args *)args); break; case LINUX_SIOCGIFFLAGS: args->cmd = SIOCGIFFLAGS; error = sys_ioctl(td, (struct ioctl_args *)args); break; default: error = ENOIOCTL; } return (error); } static int linux_to_bsd_v4l2_standard(struct l_v4l2_standard *lvstd, struct v4l2_standard *vstd) { vstd->index = lvstd->index; vstd->id = lvstd->id; CTASSERT(sizeof(vstd->name) == sizeof(lvstd->name)); memcpy(vstd->name, lvstd->name, sizeof(vstd->name)); vstd->frameperiod = lvstd->frameperiod; vstd->framelines = lvstd->framelines; CTASSERT(sizeof(vstd->reserved) == sizeof(lvstd->reserved)); memcpy(vstd->reserved, lvstd->reserved, sizeof(vstd->reserved)); return (0); } static int bsd_to_linux_v4l2_standard(struct v4l2_standard *vstd, struct l_v4l2_standard *lvstd) { lvstd->index = vstd->index; lvstd->id = vstd->id; CTASSERT(sizeof(vstd->name) == sizeof(lvstd->name)); memcpy(lvstd->name, vstd->name, sizeof(lvstd->name)); lvstd->frameperiod = vstd->frameperiod; lvstd->framelines = vstd->framelines; CTASSERT(sizeof(vstd->reserved) == sizeof(lvstd->reserved)); memcpy(lvstd->reserved, vstd->reserved, sizeof(lvstd->reserved)); return (0); } static int linux_to_bsd_v4l2_buffer(struct l_v4l2_buffer *lvb, struct v4l2_buffer *vb) { vb->index = lvb->index; vb->type = lvb->type; vb->bytesused = lvb->bytesused; vb->flags = lvb->flags; vb->field = lvb->field; vb->timestamp.tv_sec = lvb->timestamp.tv_sec; vb->timestamp.tv_usec = lvb->timestamp.tv_usec; memcpy(&vb->timecode, &lvb->timecode, sizeof (lvb->timecode)); vb->sequence = lvb->sequence; vb->memory = lvb->memory; if (lvb->memory == V4L2_MEMORY_USERPTR) /* possible pointer size conversion */ vb->m.userptr = (unsigned long)PTRIN(lvb->m.userptr); else vb->m.offset = lvb->m.offset; vb->length = lvb->length; vb->input = lvb->input; vb->reserved = lvb->reserved; return (0); } static int bsd_to_linux_v4l2_buffer(struct v4l2_buffer *vb, struct l_v4l2_buffer *lvb) { lvb->index = vb->index; lvb->type = vb->type; lvb->bytesused = vb->bytesused; lvb->flags = vb->flags; lvb->field = vb->field; lvb->timestamp.tv_sec = vb->timestamp.tv_sec; lvb->timestamp.tv_usec = vb->timestamp.tv_usec; memcpy(&lvb->timecode, &vb->timecode, sizeof (vb->timecode)); lvb->sequence = vb->sequence; lvb->memory = vb->memory; if (vb->memory == V4L2_MEMORY_USERPTR) /* possible pointer size conversion */ lvb->m.userptr = PTROUT(vb->m.userptr); else lvb->m.offset = vb->m.offset; lvb->length = vb->length; lvb->input = vb->input; lvb->reserved = vb->reserved; return (0); } static int linux_to_bsd_v4l2_format(struct l_v4l2_format *lvf, struct v4l2_format *vf) { vf->type = lvf->type; if (lvf->type == V4L2_BUF_TYPE_VIDEO_OVERLAY #ifdef V4L2_BUF_TYPE_VIDEO_OUTPUT_OVERLAY || lvf->type == V4L2_BUF_TYPE_VIDEO_OUTPUT_OVERLAY #endif ) /* * XXX TODO - needs 32 -> 64 bit conversion: * (unused by webcams?) */ return (EINVAL); memcpy(&vf->fmt, &lvf->fmt, sizeof(vf->fmt)); return (0); } static int bsd_to_linux_v4l2_format(struct v4l2_format *vf, struct l_v4l2_format *lvf) { lvf->type = vf->type; if (vf->type == V4L2_BUF_TYPE_VIDEO_OVERLAY #ifdef V4L2_BUF_TYPE_VIDEO_OUTPUT_OVERLAY || vf->type == V4L2_BUF_TYPE_VIDEO_OUTPUT_OVERLAY #endif ) /* * XXX TODO - needs 32 -> 64 bit conversion: * (unused by webcams?) */ return (EINVAL); memcpy(&lvf->fmt, &vf->fmt, sizeof(vf->fmt)); return (0); } static int linux_ioctl_v4l2(struct thread *td, struct linux_ioctl_args *args) { struct file *fp; int error; struct v4l2_format vformat; struct l_v4l2_format l_vformat; struct v4l2_standard vstd; struct l_v4l2_standard l_vstd; struct l_v4l2_buffer l_vbuf; struct v4l2_buffer vbuf; struct v4l2_input vinp; switch (args->cmd & 0xffff) { case LINUX_VIDIOC_RESERVED: case LINUX_VIDIOC_LOG_STATUS: if ((args->cmd & IOC_DIRMASK) != LINUX_IOC_VOID) return (ENOIOCTL); args->cmd = (args->cmd & 0xffff) | IOC_VOID; break; case LINUX_VIDIOC_OVERLAY: case LINUX_VIDIOC_STREAMON: case LINUX_VIDIOC_STREAMOFF: case LINUX_VIDIOC_S_STD: case LINUX_VIDIOC_S_TUNER: case LINUX_VIDIOC_S_AUDIO: case LINUX_VIDIOC_S_AUDOUT: case LINUX_VIDIOC_S_MODULATOR: case LINUX_VIDIOC_S_FREQUENCY: case LINUX_VIDIOC_S_CROP: case LINUX_VIDIOC_S_JPEGCOMP: case LINUX_VIDIOC_S_PRIORITY: case LINUX_VIDIOC_DBG_S_REGISTER: case LINUX_VIDIOC_S_HW_FREQ_SEEK: case LINUX_VIDIOC_SUBSCRIBE_EVENT: case LINUX_VIDIOC_UNSUBSCRIBE_EVENT: args->cmd = (args->cmd & ~IOC_DIRMASK) | IOC_IN; break; case LINUX_VIDIOC_QUERYCAP: case LINUX_VIDIOC_G_STD: case LINUX_VIDIOC_G_AUDIO: case LINUX_VIDIOC_G_INPUT: case LINUX_VIDIOC_G_OUTPUT: case LINUX_VIDIOC_G_AUDOUT: case LINUX_VIDIOC_G_JPEGCOMP: case LINUX_VIDIOC_QUERYSTD: case LINUX_VIDIOC_G_PRIORITY: case LINUX_VIDIOC_QUERY_DV_PRESET: args->cmd = (args->cmd & ~IOC_DIRMASK) | IOC_OUT; break; case LINUX_VIDIOC_ENUM_FMT: case LINUX_VIDIOC_REQBUFS: case LINUX_VIDIOC_G_PARM: case LINUX_VIDIOC_S_PARM: case LINUX_VIDIOC_G_CTRL: case LINUX_VIDIOC_S_CTRL: case LINUX_VIDIOC_G_TUNER: case LINUX_VIDIOC_QUERYCTRL: case LINUX_VIDIOC_QUERYMENU: case LINUX_VIDIOC_S_INPUT: case LINUX_VIDIOC_S_OUTPUT: case LINUX_VIDIOC_ENUMOUTPUT: case LINUX_VIDIOC_G_MODULATOR: case LINUX_VIDIOC_G_FREQUENCY: case LINUX_VIDIOC_CROPCAP: case LINUX_VIDIOC_G_CROP: case LINUX_VIDIOC_ENUMAUDIO: case LINUX_VIDIOC_ENUMAUDOUT: case LINUX_VIDIOC_G_SLICED_VBI_CAP: #ifdef VIDIOC_ENUM_FRAMESIZES case LINUX_VIDIOC_ENUM_FRAMESIZES: case LINUX_VIDIOC_ENUM_FRAMEINTERVALS: case LINUX_VIDIOC_ENCODER_CMD: case LINUX_VIDIOC_TRY_ENCODER_CMD: #endif case LINUX_VIDIOC_DBG_G_REGISTER: case LINUX_VIDIOC_DBG_G_CHIP_IDENT: case LINUX_VIDIOC_ENUM_DV_PRESETS: case LINUX_VIDIOC_S_DV_PRESET: case LINUX_VIDIOC_G_DV_PRESET: case LINUX_VIDIOC_S_DV_TIMINGS: case LINUX_VIDIOC_G_DV_TIMINGS: args->cmd = (args->cmd & ~IOC_DIRMASK) | IOC_INOUT; break; case LINUX_VIDIOC_G_FMT: case LINUX_VIDIOC_S_FMT: case LINUX_VIDIOC_TRY_FMT: error = copyin((void *)args->arg, &l_vformat, sizeof(l_vformat)); if (error) return (error); error = fget(td, args->fd, &cap_ioctl_rights, &fp); if (error) return (error); if (linux_to_bsd_v4l2_format(&l_vformat, &vformat) != 0) error = EINVAL; else if ((args->cmd & 0xffff) == LINUX_VIDIOC_G_FMT) error = fo_ioctl(fp, VIDIOC_G_FMT, &vformat, td->td_ucred, td); else if ((args->cmd & 0xffff) == LINUX_VIDIOC_S_FMT) error = fo_ioctl(fp, VIDIOC_S_FMT, &vformat, td->td_ucred, td); else error = fo_ioctl(fp, VIDIOC_TRY_FMT, &vformat, td->td_ucred, td); bsd_to_linux_v4l2_format(&vformat, &l_vformat); copyout(&l_vformat, (void *)args->arg, sizeof(l_vformat)); fdrop(fp, td); return (error); case LINUX_VIDIOC_ENUMSTD: error = copyin((void *)args->arg, &l_vstd, sizeof(l_vstd)); if (error) return (error); linux_to_bsd_v4l2_standard(&l_vstd, &vstd); error = fget(td, args->fd, &cap_ioctl_rights, &fp); if (error) return (error); error = fo_ioctl(fp, VIDIOC_ENUMSTD, (caddr_t)&vstd, td->td_ucred, td); if (error) { fdrop(fp, td); return (error); } bsd_to_linux_v4l2_standard(&vstd, &l_vstd); error = copyout(&l_vstd, (void *)args->arg, sizeof(l_vstd)); fdrop(fp, td); return (error); case LINUX_VIDIOC_ENUMINPUT: /* * The Linux struct l_v4l2_input differs only in size, * it has no padding at the end. */ error = copyin((void *)args->arg, &vinp, sizeof(struct l_v4l2_input)); if (error != 0) return (error); error = fget(td, args->fd, &cap_ioctl_rights, &fp); if (error != 0) return (error); error = fo_ioctl(fp, VIDIOC_ENUMINPUT, (caddr_t)&vinp, td->td_ucred, td); if (error) { fdrop(fp, td); return (error); } error = copyout(&vinp, (void *)args->arg, sizeof(struct l_v4l2_input)); fdrop(fp, td); return (error); case LINUX_VIDIOC_QUERYBUF: case LINUX_VIDIOC_QBUF: case LINUX_VIDIOC_DQBUF: error = copyin((void *)args->arg, &l_vbuf, sizeof(l_vbuf)); if (error) return (error); error = fget(td, args->fd, &cap_ioctl_rights, &fp); if (error) return (error); linux_to_bsd_v4l2_buffer(&l_vbuf, &vbuf); if ((args->cmd & 0xffff) == LINUX_VIDIOC_QUERYBUF) error = fo_ioctl(fp, VIDIOC_QUERYBUF, &vbuf, td->td_ucred, td); else if ((args->cmd & 0xffff) == LINUX_VIDIOC_QBUF) error = fo_ioctl(fp, VIDIOC_QBUF, &vbuf, td->td_ucred, td); else error = fo_ioctl(fp, VIDIOC_DQBUF, &vbuf, td->td_ucred, td); bsd_to_linux_v4l2_buffer(&vbuf, &l_vbuf); copyout(&l_vbuf, (void *)args->arg, sizeof(l_vbuf)); fdrop(fp, td); return (error); /* * XXX TODO - these need 32 -> 64 bit conversion: * (are any of them needed for webcams?) */ case LINUX_VIDIOC_G_FBUF: case LINUX_VIDIOC_S_FBUF: case LINUX_VIDIOC_G_EXT_CTRLS: case LINUX_VIDIOC_S_EXT_CTRLS: case LINUX_VIDIOC_TRY_EXT_CTRLS: case LINUX_VIDIOC_DQEVENT: default: return (ENOIOCTL); } error = sys_ioctl(td, (struct ioctl_args *)args); return (error); } /* * Support for emulators/linux-libusb. This port uses FBSD_LUSB* macros * instead of USB* ones. This lets us to provide correct values for cmd. * 0xffffffe0 -- 0xffffffff range seemed to be the least collision-prone. */ static int linux_ioctl_fbsd_usb(struct thread *td, struct linux_ioctl_args *args) { int error; error = 0; switch (args->cmd) { case FBSD_LUSB_DEVICEENUMERATE: args->cmd = USB_DEVICEENUMERATE; break; case FBSD_LUSB_DEV_QUIRK_ADD: args->cmd = USB_DEV_QUIRK_ADD; break; case FBSD_LUSB_DEV_QUIRK_GET: args->cmd = USB_DEV_QUIRK_GET; break; case FBSD_LUSB_DEV_QUIRK_REMOVE: args->cmd = USB_DEV_QUIRK_REMOVE; break; case FBSD_LUSB_DO_REQUEST: args->cmd = USB_DO_REQUEST; break; case FBSD_LUSB_FS_CLEAR_STALL_SYNC: args->cmd = USB_FS_CLEAR_STALL_SYNC; break; case FBSD_LUSB_FS_CLOSE: args->cmd = USB_FS_CLOSE; break; case FBSD_LUSB_FS_COMPLETE: args->cmd = USB_FS_COMPLETE; break; case FBSD_LUSB_FS_INIT: args->cmd = USB_FS_INIT; break; case FBSD_LUSB_FS_OPEN: args->cmd = USB_FS_OPEN; break; case FBSD_LUSB_FS_START: args->cmd = USB_FS_START; break; case FBSD_LUSB_FS_STOP: args->cmd = USB_FS_STOP; break; case FBSD_LUSB_FS_UNINIT: args->cmd = USB_FS_UNINIT; break; case FBSD_LUSB_GET_CONFIG: args->cmd = USB_GET_CONFIG; break; case FBSD_LUSB_GET_DEVICEINFO: args->cmd = USB_GET_DEVICEINFO; break; case FBSD_LUSB_GET_DEVICE_DESC: args->cmd = USB_GET_DEVICE_DESC; break; case FBSD_LUSB_GET_FULL_DESC: args->cmd = USB_GET_FULL_DESC; break; case FBSD_LUSB_GET_IFACE_DRIVER: args->cmd = USB_GET_IFACE_DRIVER; break; case FBSD_LUSB_GET_PLUGTIME: args->cmd = USB_GET_PLUGTIME; break; case FBSD_LUSB_GET_POWER_MODE: args->cmd = USB_GET_POWER_MODE; break; case FBSD_LUSB_GET_REPORT_DESC: args->cmd = USB_GET_REPORT_DESC; break; case FBSD_LUSB_GET_REPORT_ID: args->cmd = USB_GET_REPORT_ID; break; case FBSD_LUSB_GET_TEMPLATE: args->cmd = USB_GET_TEMPLATE; break; case FBSD_LUSB_IFACE_DRIVER_ACTIVE: args->cmd = USB_IFACE_DRIVER_ACTIVE; break; case FBSD_LUSB_IFACE_DRIVER_DETACH: args->cmd = USB_IFACE_DRIVER_DETACH; break; case FBSD_LUSB_QUIRK_NAME_GET: args->cmd = USB_QUIRK_NAME_GET; break; case FBSD_LUSB_READ_DIR: args->cmd = USB_READ_DIR; break; case FBSD_LUSB_SET_ALTINTERFACE: args->cmd = USB_SET_ALTINTERFACE; break; case FBSD_LUSB_SET_CONFIG: args->cmd = USB_SET_CONFIG; break; case FBSD_LUSB_SET_IMMED: args->cmd = USB_SET_IMMED; break; case FBSD_LUSB_SET_POWER_MODE: args->cmd = USB_SET_POWER_MODE; break; case FBSD_LUSB_SET_TEMPLATE: args->cmd = USB_SET_TEMPLATE; break; case FBSD_LUSB_FS_OPEN_STREAM: args->cmd = USB_FS_OPEN_STREAM; break; case FBSD_LUSB_GET_DEV_PORT_PATH: args->cmd = USB_GET_DEV_PORT_PATH; break; case FBSD_LUSB_GET_POWER_USAGE: args->cmd = USB_GET_POWER_USAGE; break; + case FBSD_LUSB_DEVICESTATS: + args->cmd = USB_DEVICESTATS; + break; default: error = ENOIOCTL; } if (error != ENOIOCTL) error = sys_ioctl(td, (struct ioctl_args *)args); return (error); } /* * Some evdev ioctls must be translated. * - EVIOCGMTSLOTS is a IOC_READ ioctl on Linux although it has input data * (must be IOC_INOUT on FreeBSD). * - On Linux, EVIOCGRAB, EVIOCREVOKE and EVIOCRMFF are defined as _IOW with * an int argument. You don't pass an int pointer to the ioctl(), however, * but just the int directly. On FreeBSD, they are defined as _IOWINT for * this to work. */ static int linux_ioctl_evdev(struct thread *td, struct linux_ioctl_args *args) { struct file *fp; clockid_t clock; int error; args->cmd = SETDIR(args->cmd); switch (args->cmd) { case (EVIOCGRAB & ~IOC_DIRMASK) | IOC_IN: args->cmd = EVIOCGRAB; break; case (EVIOCREVOKE & ~IOC_DIRMASK) | IOC_IN: args->cmd = EVIOCREVOKE; break; case (EVIOCRMFF & ~IOC_DIRMASK) | IOC_IN: args->cmd = EVIOCRMFF; break; case EVIOCSCLOCKID: { error = copyin(PTRIN(args->arg), &clock, sizeof(clock)); if (error != 0) return (error); if (clock & ~(LINUX_IOCTL_EVDEV_CLK)) return (EINVAL); error = linux_to_native_clockid(&clock, clock); if (error != 0) return (error); error = fget(td, args->fd, &cap_ioctl_rights, &fp); if (error != 0) return (error); error = fo_ioctl(fp, EVIOCSCLOCKID, &clock, td->td_ucred, td); fdrop(fp, td); return (error); } default: break; } if (IOCBASECMD(args->cmd) == ((EVIOCGMTSLOTS(0) & ~IOC_DIRMASK) | IOC_OUT)) args->cmd = (args->cmd & ~IOC_DIRMASK) | IOC_INOUT; return (sys_ioctl(td, (struct ioctl_args *)args)); } /* * main ioctl syscall function */ int linux_ioctl(struct thread *td, struct linux_ioctl_args *args) { struct file *fp; struct linux_ioctl_handler_element *he; int error, cmd; error = fget(td, args->fd, &cap_ioctl_rights, &fp); if (error != 0) return (error); if ((fp->f_flag & (FREAD|FWRITE)) == 0) { fdrop(fp, td); return (EBADF); } /* Iterate over the ioctl handlers */ cmd = args->cmd & 0xffff; sx_slock(&linux_ioctl_sx); mtx_lock(&Giant); #ifdef COMPAT_LINUX32 TAILQ_FOREACH(he, &linux32_ioctl_handlers, list) { if (cmd >= he->low && cmd <= he->high) { error = (*he->func)(td, args); if (error != ENOIOCTL) { mtx_unlock(&Giant); sx_sunlock(&linux_ioctl_sx); fdrop(fp, td); return (error); } } } #endif TAILQ_FOREACH(he, &linux_ioctl_handlers, list) { if (cmd >= he->low && cmd <= he->high) { error = (*he->func)(td, args); if (error != ENOIOCTL) { mtx_unlock(&Giant); sx_sunlock(&linux_ioctl_sx); fdrop(fp, td); return (error); } } } mtx_unlock(&Giant); sx_sunlock(&linux_ioctl_sx); fdrop(fp, td); switch (args->cmd & 0xffff) { case LINUX_BTRFS_IOC_CLONE: case LINUX_FS_IOC_FIEMAP: return (ENOTSUP); default: linux_msg(td, "ioctl fd=%d, cmd=0x%x ('%c',%d) is not implemented", args->fd, (int)(args->cmd & 0xffff), (int)(args->cmd & 0xff00) >> 8, (int)(args->cmd & 0xff)); break; } return (EINVAL); } int linux_ioctl_register_handler(struct linux_ioctl_handler *h) { struct linux_ioctl_handler_element *he, *cur; if (h == NULL || h->func == NULL) return (EINVAL); /* * Reuse the element if the handler is already on the list, otherwise * create a new element. */ sx_xlock(&linux_ioctl_sx); TAILQ_FOREACH(he, &linux_ioctl_handlers, list) { if (he->func == h->func) break; } if (he == NULL) { he = malloc(sizeof(*he), M_LINUX, M_WAITOK); he->func = h->func; } else TAILQ_REMOVE(&linux_ioctl_handlers, he, list); /* Initialize range information. */ he->low = h->low; he->high = h->high; he->span = h->high - h->low + 1; /* Add the element to the list, sorted on span. */ TAILQ_FOREACH(cur, &linux_ioctl_handlers, list) { if (cur->span > he->span) { TAILQ_INSERT_BEFORE(cur, he, list); sx_xunlock(&linux_ioctl_sx); return (0); } } TAILQ_INSERT_TAIL(&linux_ioctl_handlers, he, list); sx_xunlock(&linux_ioctl_sx); return (0); } int linux_ioctl_unregister_handler(struct linux_ioctl_handler *h) { struct linux_ioctl_handler_element *he; if (h == NULL || h->func == NULL) return (EINVAL); sx_xlock(&linux_ioctl_sx); TAILQ_FOREACH(he, &linux_ioctl_handlers, list) { if (he->func == h->func) { TAILQ_REMOVE(&linux_ioctl_handlers, he, list); sx_xunlock(&linux_ioctl_sx); free(he, M_LINUX); return (0); } } sx_xunlock(&linux_ioctl_sx); return (EINVAL); } #ifdef COMPAT_LINUX32 int linux32_ioctl_register_handler(struct linux_ioctl_handler *h) { struct linux_ioctl_handler_element *he, *cur; if (h == NULL || h->func == NULL) return (EINVAL); /* * Reuse the element if the handler is already on the list, otherwise * create a new element. */ sx_xlock(&linux_ioctl_sx); TAILQ_FOREACH(he, &linux32_ioctl_handlers, list) { if (he->func == h->func) break; } if (he == NULL) { he = malloc(sizeof(*he), M_LINUX, M_WAITOK); he->func = h->func; } else TAILQ_REMOVE(&linux32_ioctl_handlers, he, list); /* Initialize range information. */ he->low = h->low; he->high = h->high; he->span = h->high - h->low + 1; /* Add the element to the list, sorted on span. */ TAILQ_FOREACH(cur, &linux32_ioctl_handlers, list) { if (cur->span > he->span) { TAILQ_INSERT_BEFORE(cur, he, list); sx_xunlock(&linux_ioctl_sx); return (0); } } TAILQ_INSERT_TAIL(&linux32_ioctl_handlers, he, list); sx_xunlock(&linux_ioctl_sx); return (0); } int linux32_ioctl_unregister_handler(struct linux_ioctl_handler *h) { struct linux_ioctl_handler_element *he; if (h == NULL || h->func == NULL) return (EINVAL); sx_xlock(&linux_ioctl_sx); TAILQ_FOREACH(he, &linux32_ioctl_handlers, list) { if (he->func == h->func) { TAILQ_REMOVE(&linux32_ioctl_handlers, he, list); sx_xunlock(&linux_ioctl_sx); free(he, M_LINUX); return (0); } } sx_xunlock(&linux_ioctl_sx); return (EINVAL); } #endif Index: projects/clang1000-import/sys/compat/linux/linux_ioctl.h =================================================================== --- projects/clang1000-import/sys/compat/linux/linux_ioctl.h (revision 358848) +++ projects/clang1000-import/sys/compat/linux/linux_ioctl.h (revision 358849) @@ -1,790 +1,791 @@ /*- * SPDX-License-Identifier: BSD-2-Clause-FreeBSD * * Copyright (c) 1999 Marcel Moolenaar * 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. * * $FreeBSD$ */ #ifndef _LINUX_IOCTL_H_ #define _LINUX_IOCTL_H_ /* * ioctl * * XXX comments in Linux' indicate these * could be arch-dependant... */ #define LINUX_IOC_VOID 0 #define LINUX_IOC_IN 0x40000000 #define LINUX_IOC_OUT 0x80000000 #define LINUX_IOC_INOUT (LINUX_IOC_IN|LINUX_IOC_OUT) /* * disk */ #define LINUX_BLKROSET 0x125d #define LINUX_BLKROGET 0x125e #define LINUX_BLKRRPART 0x125f #define LINUX_BLKGETSIZE 0x1260 #define LINUX_BLKFLSBUF 0x1261 #define LINUX_BLKRASET 0x1262 #define LINUX_BLKRAGET 0x1263 #define LINUX_BLKFRASET 0x1264 #define LINUX_BLKFRAGET 0x1265 #define LINUX_BLKSECTSET 0x1266 #define LINUX_BLKSECTGET 0x1267 #define LINUX_BLKSSZGET 0x1268 #define LINUX_BLKGETSIZE64 0x1272 #define LINUX_IOCTL_DISK_MIN LINUX_BLKROSET #define LINUX_IOCTL_DISK_MAX LINUX_BLKGETSIZE64 /* * hdio */ #define LINUX_HDIO_GET_GEO 0x0301 #define LINUX_HDIO_GET_IDENTITY 0x030D /* not yet implemented */ #define LINUX_HDIO_GET_GEO_BIG 0x0330 #define LINUX_IOCTL_HDIO_MIN LINUX_HDIO_GET_GEO #define LINUX_IOCTL_HDIO_MAX LINUX_HDIO_GET_GEO_BIG /* * cdrom */ #define LINUX_CDROMPAUSE 0x5301 #define LINUX_CDROMRESUME 0x5302 #define LINUX_CDROMPLAYMSF 0x5303 #define LINUX_CDROMPLAYTRKIND 0x5304 #define LINUX_CDROMREADTOCHDR 0x5305 #define LINUX_CDROMREADTOCENTRY 0x5306 #define LINUX_CDROMSTOP 0x5307 #define LINUX_CDROMSTART 0x5308 #define LINUX_CDROMEJECT 0x5309 #define LINUX_CDROMVOLCTRL 0x530a #define LINUX_CDROMSUBCHNL 0x530b #define LINUX_CDROMREADMODE2 0x530c #define LINUX_CDROMREADMODE1 0x530d #define LINUX_CDROMREADAUDIO 0x530e #define LINUX_CDROMEJECT_SW 0x530f #define LINUX_CDROMMULTISESSION 0x5310 #define LINUX_CDROM_GET_UPC 0x5311 #define LINUX_CDROMRESET 0x5312 #define LINUX_CDROMVOLREAD 0x5313 #define LINUX_CDROMREADRAW 0x5314 #define LINUX_CDROMREADCOOKED 0x5315 #define LINUX_CDROMSEEK 0x5316 #define LINUX_CDROMPLAYBLK 0x5317 #define LINUX_CDROMREADALL 0x5318 #define LINUX_CDROMCLOSETRAY 0x5319 #define LINUX_CDROMLOADFROMSLOT 0x531a #define LINUX_CDROMGETSPINDOWN 0x531d #define LINUX_CDROMSETSPINDOWN 0x531e #define LINUX_CDROM_SET_OPTIONS 0x5320 #define LINUX_CDROM_CLEAR_OPTIONS 0x5321 #define LINUX_CDROM_SELECT_SPEED 0x5322 #define LINUX_CDROM_SELECT_DISC 0x5323 #define LINUX_CDROM_MEDIA_CHANGED 0x5325 #define LINUX_CDROM_DRIVE_STATUS 0x5326 #define LINUX_CDROM_DISC_STATUS 0x5327 #define LINUX_CDROM_CHANGER_NSLOTS 0x5328 #define LINUX_CDROM_LOCKDOOR 0x5329 #define LINUX_CDROM_DEBUG 0x5330 #define LINUX_CDROM_GET_CAPABILITY 0x5331 #define LINUX_CDROMAUDIOBUFSIZ 0x5382 #define LINUX_SCSI_GET_IDLUN 0x5382 #define LINUX_SCSI_GET_BUS_NUMBER 0x5386 #define LINUX_DVD_READ_STRUCT 0x5390 #define LINUX_DVD_WRITE_STRUCT 0x5391 #define LINUX_DVD_AUTH 0x5392 #define LINUX_CDROM_SEND_PACKET 0x5393 #define LINUX_CDROM_NEXT_WRITABLE 0x5394 #define LINUX_CDROM_LAST_WRITTEN 0x5395 #define LINUX_IOCTL_CDROM_MIN LINUX_CDROMPAUSE #define LINUX_IOCTL_CDROM_MAX LINUX_CDROM_LAST_WRITTEN #define LINUX_CDROM_LBA 0x01 #define LINUX_CDROM_MSF 0x02 #define LINUX_DVD_LU_SEND_AGID 0 #define LINUX_DVD_HOST_SEND_CHALLENGE 1 #define LINUX_DVD_LU_SEND_KEY1 2 #define LINUX_DVD_LU_SEND_CHALLENGE 3 #define LINUX_DVD_HOST_SEND_KEY2 4 #define LINUX_DVD_AUTH_ESTABLISHED 5 #define LINUX_DVD_AUTH_FAILURE 6 #define LINUX_DVD_LU_SEND_TITLE_KEY 7 #define LINUX_DVD_LU_SEND_ASF 8 #define LINUX_DVD_INVALIDATE_AGID 9 #define LINUX_DVD_LU_SEND_RPC_STATE 10 #define LINUX_DVD_HOST_SEND_RPC_STATE 11 /* * SG */ #define LINUX_SG_SET_TIMEOUT 0x2201 #define LINUX_SG_GET_TIMEOUT 0x2202 #define LINUX_SG_EMULATED_HOST 0x2203 #define LINUX_SG_SET_TRANSFORM 0x2204 #define LINUX_SG_GET_TRANSFORM 0x2205 #define LINUX_SG_GET_COMMAND_Q 0x2270 #define LINUX_SG_SET_COMMAND_Q 0x2271 #define LINUX_SG_SET_RESERVED_SIZE 0x2275 #define LINUX_SG_GET_RESERVED_SIZE 0x2272 #define LINUX_SG_GET_SCSI_ID 0x2276 #define LINUX_SG_SET_FORCE_LOW_DMA 0x2279 #define LINUX_SG_GET_LOW_DMA 0x227a #define LINUX_SG_SET_FORCE_PACK_ID 0x227b #define LINUX_SG_GET_PACK_ID 0x227c #define LINUX_SG_GET_NUM_WAITING 0x227d #define LINUX_SG_SET_DEBUG 0x227e #define LINUX_SG_GET_SG_TABLESIZE 0x227f #define LINUX_SG_GET_VERSION_NUM 0x2282 #define LINUX_SG_NEXT_CMD_LEN 0x2283 #define LINUX_SG_SCSI_RESET 0x2284 #define LINUX_SG_IO 0x2285 #define LINUX_SG_GET_REQUEST_TABLE 0x2286 #define LINUX_SG_SET_KEEP_ORPHAN 0x2287 #define LINUX_SG_GET_KEEP_ORPHAN 0x2288 #define LINUX_SG_GET_ACCESS_COUNT 0x2289 #define LINUX_IOCTL_SG_MIN 0x2200 #define LINUX_IOCTL_SG_MAX 0x22ff /* * VFAT */ #define LINUX_VFAT_READDIR_BOTH 0x7201 #define LINUX_IOCTL_VFAT_MIN LINUX_VFAT_READDIR_BOTH #define LINUX_IOCTL_VFAT_MAX LINUX_VFAT_READDIR_BOTH /* * console */ #define LINUX_KIOCSOUND 0x4B2F #define LINUX_KDMKTONE 0x4B30 #define LINUX_KDGETLED 0x4B31 #define LINUX_KDSETLED 0x4B32 #define LINUX_KDSETMODE 0x4B3A #define LINUX_KDGETMODE 0x4B3B #define LINUX_KDGKBMODE 0x4B44 #define LINUX_KDSKBMODE 0x4B45 #define LINUX_VT_OPENQRY 0x5600 #define LINUX_VT_GETMODE 0x5601 #define LINUX_VT_SETMODE 0x5602 #define LINUX_VT_GETSTATE 0x5603 #define LINUX_VT_RELDISP 0x5605 #define LINUX_VT_ACTIVATE 0x5606 #define LINUX_VT_WAITACTIVE 0x5607 #define LINUX_IOCTL_CONSOLE_MIN LINUX_KIOCSOUND #define LINUX_IOCTL_CONSOLE_MAX LINUX_VT_WAITACTIVE #define LINUX_LED_SCR 0x01 #define LINUX_LED_NUM 0x02 #define LINUX_LED_CAP 0x04 #define LINUX_KD_TEXT 0x0 #define LINUX_KD_GRAPHICS 0x1 #define LINUX_KD_TEXT0 0x2 #define LINUX_KD_TEXT1 0x3 #define LINUX_KBD_RAW 0 #define LINUX_KBD_XLATE 1 #define LINUX_KBD_MEDIUMRAW 2 /* * socket */ #define LINUX_FIOSETOWN 0x8901 #define LINUX_SIOCSPGRP 0x8902 #define LINUX_FIOGETOWN 0x8903 #define LINUX_SIOCGPGRP 0x8904 #define LINUX_SIOCATMARK 0x8905 #define LINUX_SIOCGSTAMP 0x8906 #define LINUX_SIOCGIFNAME 0x8910 #define LINUX_SIOCGIFCONF 0x8912 #define LINUX_SIOCGIFFLAGS 0x8913 #define LINUX_SIOCGIFADDR 0x8915 #define LINUX_SIOCSIFADDR 0x8916 #define LINUX_SIOCGIFDSTADDR 0x8917 #define LINUX_SIOCGIFBRDADDR 0x8919 #define LINUX_SIOCGIFNETMASK 0x891b #define LINUX_SIOCSIFNETMASK 0x891c #define LINUX_SIOCGIFMTU 0x8921 #define LINUX_SIOCSIFMTU 0x8922 #define LINUX_SIOCSIFNAME 0x8923 #define LINUX_SIOCSIFHWADDR 0x8924 #define LINUX_SIOCGIFHWADDR 0x8927 #define LINUX_SIOCADDMULTI 0x8931 #define LINUX_SIOCDELMULTI 0x8932 #define LINUX_SIOCGIFINDEX 0x8933 #define LINUX_SIOGIFINDEX LINUX_SIOCGIFINDEX #define LINUX_SIOCGIFCOUNT 0x8938 #define LINUX_IOCTL_SOCKET_MIN LINUX_FIOSETOWN #define LINUX_IOCTL_SOCKET_MAX LINUX_SIOCGIFCOUNT /* * Device private ioctl calls */ #define LINUX_SIOCDEVPRIVATE 0x89F0 /* to 89FF */ #define LINUX_IOCTL_PRIVATE_MIN LINUX_SIOCDEVPRIVATE #define LINUX_IOCTL_PRIVATE_MAX LINUX_SIOCDEVPRIVATE+0xf /* * sound */ #define LINUX_SOUND_MIXER_WRITE_VOLUME 0x4d00 #define LINUX_SOUND_MIXER_WRITE_BASS 0x4d01 #define LINUX_SOUND_MIXER_WRITE_TREBLE 0x4d02 #define LINUX_SOUND_MIXER_WRITE_SYNTH 0x4d03 #define LINUX_SOUND_MIXER_WRITE_PCM 0x4d04 #define LINUX_SOUND_MIXER_WRITE_SPEAKER 0x4d05 #define LINUX_SOUND_MIXER_WRITE_LINE 0x4d06 #define LINUX_SOUND_MIXER_WRITE_MIC 0x4d07 #define LINUX_SOUND_MIXER_WRITE_CD 0x4d08 #define LINUX_SOUND_MIXER_WRITE_IMIX 0x4d09 #define LINUX_SOUND_MIXER_WRITE_ALTPCM 0x4d0A #define LINUX_SOUND_MIXER_WRITE_RECLEV 0x4d0B #define LINUX_SOUND_MIXER_WRITE_IGAIN 0x4d0C #define LINUX_SOUND_MIXER_WRITE_OGAIN 0x4d0D #define LINUX_SOUND_MIXER_WRITE_LINE1 0x4d0E #define LINUX_SOUND_MIXER_WRITE_LINE2 0x4d0F #define LINUX_SOUND_MIXER_WRITE_LINE3 0x4d10 #define LINUX_SOUND_MIXER_INFO 0x4d65 #define LINUX_OSS_GETVERSION 0x4d76 #define LINUX_SOUND_MIXER_READ_STEREODEVS 0x4dfb #define LINUX_SOUND_MIXER_READ_CAPS 0x4dfc #define LINUX_SOUND_MIXER_READ_RECMASK 0x4dfd #define LINUX_SOUND_MIXER_READ_DEVMASK 0x4dfe #define LINUX_SOUND_MIXER_WRITE_RECSRC 0x4dff #define LINUX_SNDCTL_DSP_RESET 0x5000 #define LINUX_SNDCTL_DSP_SYNC 0x5001 #define LINUX_SNDCTL_DSP_SPEED 0x5002 #define LINUX_SNDCTL_DSP_STEREO 0x5003 #define LINUX_SNDCTL_DSP_GETBLKSIZE 0x5004 #define LINUX_SNDCTL_DSP_SETBLKSIZE LINUX_SNDCTL_DSP_GETBLKSIZE #define LINUX_SNDCTL_DSP_SETFMT 0x5005 #define LINUX_SOUND_PCM_WRITE_CHANNELS 0x5006 #define LINUX_SOUND_PCM_WRITE_FILTER 0x5007 #define LINUX_SNDCTL_DSP_POST 0x5008 #define LINUX_SNDCTL_DSP_SUBDIVIDE 0x5009 #define LINUX_SNDCTL_DSP_SETFRAGMENT 0x500A #define LINUX_SNDCTL_DSP_GETFMTS 0x500B #define LINUX_SNDCTL_DSP_GETOSPACE 0x500C #define LINUX_SNDCTL_DSP_GETISPACE 0x500D #define LINUX_SNDCTL_DSP_NONBLOCK 0x500E #define LINUX_SNDCTL_DSP_GETCAPS 0x500F #define LINUX_SNDCTL_DSP_GETTRIGGER 0x5010 #define LINUX_SNDCTL_DSP_SETTRIGGER LINUX_SNDCTL_DSP_GETTRIGGER #define LINUX_SNDCTL_DSP_GETIPTR 0x5011 #define LINUX_SNDCTL_DSP_GETOPTR 0x5012 #define LINUX_SNDCTL_DSP_SETDUPLEX 0x5016 #define LINUX_SNDCTL_DSP_GETODELAY 0x5017 #define LINUX_SNDCTL_SEQ_RESET 0x5100 #define LINUX_SNDCTL_SEQ_SYNC 0x5101 #define LINUX_SNDCTL_SYNTH_INFO 0x5102 #define LINUX_SNDCTL_SEQ_CTRLRATE 0x5103 #define LINUX_SNDCTL_SEQ_GETOUTCOUNT 0x5104 #define LINUX_SNDCTL_SEQ_GETINCOUNT 0x5105 #define LINUX_SNDCTL_SEQ_PERCMODE 0x5106 #define LINUX_SNDCTL_FM_LOAD_INSTR 0x5107 #define LINUX_SNDCTL_SEQ_TESTMIDI 0x5108 #define LINUX_SNDCTL_SEQ_RESETSAMPLES 0x5109 #define LINUX_SNDCTL_SEQ_NRSYNTHS 0x510A #define LINUX_SNDCTL_SEQ_NRMIDIS 0x510B #define LINUX_SNDCTL_MIDI_INFO 0x510C #define LINUX_SNDCTL_SEQ_TRESHOLD 0x510D #define LINUX_SNDCTL_SYNTH_MEMAVL 0x510E #define LINUX_IOCTL_SOUND_MIN LINUX_SOUND_MIXER_WRITE_VOLUME #define LINUX_IOCTL_SOUND_MAX LINUX_SNDCTL_SYNTH_MEMAVL /* * termio */ #define LINUX_TCGETS 0x5401 #define LINUX_TCSETS 0x5402 #define LINUX_TCSETSW 0x5403 #define LINUX_TCSETSF 0x5404 #define LINUX_TCGETA 0x5405 #define LINUX_TCSETA 0x5406 #define LINUX_TCSETAW 0x5407 #define LINUX_TCSETAF 0x5408 #define LINUX_TCSBRK 0x5409 #define LINUX_TCXONC 0x540A #define LINUX_TCFLSH 0x540B #define LINUX_TIOCEXCL 0x540C #define LINUX_TIOCNXCL 0x540D #define LINUX_TIOCSCTTY 0x540E #define LINUX_TIOCGPGRP 0x540F #define LINUX_TIOCSPGRP 0x5410 #define LINUX_TIOCOUTQ 0x5411 #define LINUX_TIOCSTI 0x5412 #define LINUX_TIOCGWINSZ 0x5413 #define LINUX_TIOCSWINSZ 0x5414 #define LINUX_TIOCMGET 0x5415 #define LINUX_TIOCMBIS 0x5416 #define LINUX_TIOCMBIC 0x5417 #define LINUX_TIOCMSET 0x5418 #define LINUX_TIOCGSOFTCAR 0x5419 #define LINUX_TIOCSSOFTCAR 0x541A #define LINUX_FIONREAD 0x541B #define LINUX_TIOCINQ FIONREAD #define LINUX_TIOCLINUX 0x541C #define LINUX_TIOCCONS 0x541D #define LINUX_TIOCGSERIAL 0x541E #define LINUX_TIOCSSERIAL 0x541F #define LINUX_TIOCPKT 0x5420 #define LINUX_FIONBIO 0x5421 #define LINUX_TIOCNOTTY 0x5422 #define LINUX_TIOCSETD 0x5423 #define LINUX_TIOCGETD 0x5424 #define LINUX_TCSBRKP 0x5425 #define LINUX_TIOCTTYGSTRUCT 0x5426 #define LINUX_TIOCSBRK 0x5427 #define LINUX_TIOCCBRK 0x5428 #define LINUX_TIOCGPTN 0x5430 #define LINUX_TIOCSPTLCK 0x5431 #define LINUX_FIONCLEX 0x5450 #define LINUX_FIOCLEX 0x5451 #define LINUX_FIOASYNC 0x5452 #define LINUX_TIOCSERCONFIG 0x5453 #define LINUX_TIOCSERGWILD 0x5454 #define LINUX_TIOCSERSWILD 0x5455 #define LINUX_TIOCGLCKTRMIOS 0x5456 #define LINUX_TIOCSLCKTRMIOS 0x5457 #define LINUX_IOCTL_TERMIO_MIN LINUX_TCGETS #define LINUX_IOCTL_TERMIO_MAX LINUX_TIOCSLCKTRMIOS /* arguments for tcflow() and LINUX_TCXONC */ #define LINUX_TCOOFF 0 #define LINUX_TCOON 1 #define LINUX_TCIOFF 2 #define LINUX_TCION 3 /* arguments for tcflush() and LINUX_TCFLSH */ #define LINUX_TCIFLUSH 0 #define LINUX_TCOFLUSH 1 #define LINUX_TCIOFLUSH 2 /* line disciplines */ #define LINUX_N_TTY 0 #define LINUX_N_SLIP 1 #define LINUX_N_MOUSE 2 #define LINUX_N_PPP 3 /* Linux termio c_cc values */ #define LINUX_VINTR 0 #define LINUX_VQUIT 1 #define LINUX_VERASE 2 #define LINUX_VKILL 3 #define LINUX_VEOF 4 #define LINUX_VTIME 5 #define LINUX_VMIN 6 #define LINUX_VSWTC 7 #define LINUX_NCC 8 /* Linux termios c_cc values */ /* In addition to the termio values */ #define LINUX_VSTART 8 #define LINUX_VSTOP 9 #define LINUX_VSUSP 10 #define LINUX_VEOL 11 #define LINUX_VREPRINT 12 #define LINUX_VDISCARD 13 #define LINUX_VWERASE 14 #define LINUX_VLNEXT 15 #define LINUX_VEOL2 16 #define LINUX_VSTATUS 18 #define LINUX_NCCS 19 #define LINUX_POSIX_VDISABLE '\0' /* Linux c_iflag masks */ #define LINUX_IGNBRK 0x0000001 #define LINUX_BRKINT 0x0000002 #define LINUX_IGNPAR 0x0000004 #define LINUX_PARMRK 0x0000008 #define LINUX_INPCK 0x0000010 #define LINUX_ISTRIP 0x0000020 #define LINUX_INLCR 0x0000040 #define LINUX_IGNCR 0x0000080 #define LINUX_ICRNL 0x0000100 #define LINUX_IUCLC 0x0000200 #define LINUX_IXON 0x0000400 #define LINUX_IXANY 0x0000800 #define LINUX_IXOFF 0x0001000 #define LINUX_IMAXBEL 0x0002000 /* Linux c_oflag masks */ #define LINUX_OPOST 0x0000001 #define LINUX_OLCUC 0x0000002 #define LINUX_ONLCR 0x0000004 #define LINUX_OCRNL 0x0000008 #define LINUX_ONOCR 0x0000010 #define LINUX_ONLRET 0x0000020 #define LINUX_OFILL 0x0000040 #define LINUX_OFDEL 0x0000080 #define LINUX_NLDLY 0x0000100 #define LINUX_NL0 0x0000000 #define LINUX_NL1 0x0000100 #define LINUX_CRDLY 0x0000600 #define LINUX_CR0 0x0000000 #define LINUX_CR1 0x0000200 #define LINUX_CR2 0x0000400 #define LINUX_CR3 0x0000600 #define LINUX_TABDLY 0x0001800 #define LINUX_TAB0 0x0000000 #define LINUX_TAB1 0x0000800 #define LINUX_TAB2 0x0001000 #define LINUX_TAB3 0x0001800 #define LINUX_XTABS 0x0001800 #define LINUX_BSDLY 0x0002000 #define LINUX_BS0 0x0000000 #define LINUX_BS1 0x0002000 #define LINUX_VTDLY 0x0004000 #define LINUX_VT0 0x0000000 #define LINUX_VT1 0x0004000 #define LINUX_FFDLY 0x0008000 #define LINUX_FF0 0x0000000 #define LINUX_FF1 0x0008000 #define LINUX_CBAUD 0x0000100f #define LINUX_B0 0x00000000 #define LINUX_B50 0x00000001 #define LINUX_B75 0x00000002 #define LINUX_B110 0x00000003 #define LINUX_B134 0x00000004 #define LINUX_B150 0x00000005 #define LINUX_B200 0x00000006 #define LINUX_B300 0x00000007 #define LINUX_B600 0x00000008 #define LINUX_B1200 0x00000009 #define LINUX_B1800 0x0000000a #define LINUX_B2400 0x0000000b #define LINUX_B4800 0x0000000c #define LINUX_B9600 0x0000000d #define LINUX_B19200 0x0000000e #define LINUX_B38400 0x0000000f #define LINUX_EXTA LINUX_B19200 #define LINUX_EXTB LINUX_B38400 #define LINUX_CBAUDEX 0x00001000 #define LINUX_B57600 0x00001001 #define LINUX_B115200 0x00001002 #define LINUX_CSIZE 0x00000030 #define LINUX_CS5 0x00000000 #define LINUX_CS6 0x00000010 #define LINUX_CS7 0x00000020 #define LINUX_CS8 0x00000030 #define LINUX_CSTOPB 0x00000040 #define LINUX_CREAD 0x00000080 #define LINUX_PARENB 0x00000100 #define LINUX_PARODD 0x00000200 #define LINUX_HUPCL 0x00000400 #define LINUX_CLOCAL 0x00000800 #define LINUX_CRTSCTS 0x80000000 /* Linux c_lflag masks */ #define LINUX_ISIG 0x00000001 #define LINUX_ICANON 0x00000002 #define LINUX_XCASE 0x00000004 #define LINUX_ECHO 0x00000008 #define LINUX_ECHOE 0x00000010 #define LINUX_ECHOK 0x00000020 #define LINUX_ECHONL 0x00000040 #define LINUX_NOFLSH 0x00000080 #define LINUX_TOSTOP 0x00000100 #define LINUX_ECHOCTL 0x00000200 #define LINUX_ECHOPRT 0x00000400 #define LINUX_ECHOKE 0x00000800 #define LINUX_FLUSHO 0x00001000 #define LINUX_PENDIN 0x00002000 #define LINUX_IEXTEN 0x00008000 /* serial_struct values for TIOC[GS]SERIAL ioctls */ #define LINUX_ASYNC_CLOSING_WAIT_INF 0 #define LINUX_ASYNC_CLOSING_WAIT_NONE 65535 #define LINUX_PORT_UNKNOWN 0 #define LINUX_PORT_8250 1 #define LINUX_PORT_16450 2 #define LINUX_PORT_16550 3 #define LINUX_PORT_16550A 4 #define LINUX_PORT_CIRRUS 5 #define LINUX_PORT_16650 6 #define LINUX_PORT_MAX 6 #define LINUX_ASYNC_HUP_NOTIFY 0x0001 #define LINUX_ASYNC_FOURPORT 0x0002 #define LINUX_ASYNC_SAK 0x0004 #define LINUX_ASYNC_SPLIT_TERMIOS 0x0008 #define LINUX_ASYNC_SPD_MASK 0x0030 #define LINUX_ASYNC_SPD_HI 0x0010 #define LINUX_ASYNC_SPD_VHI 0x0020 #define LINUX_ASYNC_SPD_CUST 0x0030 #define LINUX_ASYNC_SKIP_TEST 0x0040 #define LINUX_ASYNC_AUTO_IRQ 0x0080 #define LINUX_ASYNC_SESSION_LOCKOUT 0x0100 #define LINUX_ASYNC_PGRP_LOCKOUT 0x0200 #define LINUX_ASYNC_CALLOUT_NOHUP 0x0400 #define LINUX_ASYNC_FLAGS 0x0FFF #define LINUX_IOCTL_DRM_MIN 0x6400 #define LINUX_IOCTL_DRM_MAX 0x64ff /* * video */ #define LINUX_VIDIOCGCAP 0x7601 #define LINUX_VIDIOCGCHAN 0x7602 #define LINUX_VIDIOCSCHAN 0x7603 #define LINUX_VIDIOCGTUNER 0x7604 #define LINUX_VIDIOCSTUNER 0x7605 #define LINUX_VIDIOCGPICT 0x7606 #define LINUX_VIDIOCSPICT 0x7607 #define LINUX_VIDIOCCAPTURE 0x7608 #define LINUX_VIDIOCGWIN 0x7609 #define LINUX_VIDIOCSWIN 0x760a #define LINUX_VIDIOCGFBUF 0x760b #define LINUX_VIDIOCSFBUF 0x760c #define LINUX_VIDIOCKEY 0x760d #define LINUX_VIDIOCGFREQ 0x760e #define LINUX_VIDIOCSFREQ 0x760f #define LINUX_VIDIOCGAUDIO 0x7610 #define LINUX_VIDIOCSAUDIO 0x7611 #define LINUX_VIDIOCSYNC 0x7623 #define LINUX_VIDIOCMCAPTURE 0x7613 #define LINUX_VIDIOCGMBUF 0x7614 #define LINUX_VIDIOCGUNIT 0x7615 #define LINUX_VIDIOCGCAPTURE 0x7616 #define LINUX_VIDIOCSCAPTURE 0x7617 #define LINUX_VIDIOCSPLAYMODE 0x7618 #define LINUX_VIDIOCSWRITEMODE 0x7619 #define LINUX_VIDIOCGPLAYINFO 0x761a #define LINUX_VIDIOCSMICROCODE 0x761b #define LINUX_VIDIOCGVBIFMT 0x761c #define LINUX_VIDIOCSVBIFMT 0x761d #define LINUX_IOCTL_VIDEO_MIN LINUX_VIDIOCGCAP #define LINUX_IOCTL_VIDEO_MAX LINUX_VIDIOCSVBIFMT /* videodev2 aka V4L2 */ #define LINUX_VIDIOC_QUERYCAP 0x5600 /* 0x80685600 */ #define LINUX_VIDIOC_RESERVED 0x5601 /* 0x00005601 */ #define LINUX_VIDIOC_ENUM_FMT 0x5602 /* 0xc0405602 */ #define LINUX_VIDIOC_G_FMT 0x5604 /* 0xc0cc5604 */ #define LINUX_VIDIOC_S_FMT 0x5605 /* 0xc0cc5605 */ #define LINUX_VIDIOC_REQBUFS 0x5608 /* 0xc0145608 */ #define LINUX_VIDIOC_QUERYBUF 0x5609 /* 0xc0445609 */ #define LINUX_VIDIOC_G_FBUF 0x560a /* 0x802c560a */ #define LINUX_VIDIOC_S_FBUF 0x560b /* 0x402c560b */ #define LINUX_VIDIOC_OVERLAY 0x560e /* 0x4004560e */ #define LINUX_VIDIOC_QBUF 0x560f /* 0xc044560f */ #define LINUX_VIDIOC_DQBUF 0x5611 /* 0xc0445611 */ #define LINUX_VIDIOC_STREAMON 0x5612 /* 0x40045612 */ #define LINUX_VIDIOC_STREAMOFF 0x5613 /* 0x40045613 */ #define LINUX_VIDIOC_G_PARM 0x5615 /* 0xc0cc5615 */ #define LINUX_VIDIOC_S_PARM 0x5616 /* 0xc0cc5616 */ #define LINUX_VIDIOC_G_STD 0x5617 /* 0x80085617 */ #define LINUX_VIDIOC_S_STD 0x5618 /* 0x40085618 */ #define LINUX_VIDIOC_ENUMSTD 0x5619 /* 0xc0405619 */ #define LINUX_VIDIOC_ENUMINPUT 0x561a /* 0xc04c561a */ #define LINUX_VIDIOC_G_CTRL 0x561b /* 0xc008561b */ #define LINUX_VIDIOC_S_CTRL 0x561c /* 0xc008561c */ #define LINUX_VIDIOC_G_TUNER 0x561d /* 0xc054561d */ #define LINUX_VIDIOC_S_TUNER 0x561e /* 0x4054561e */ #define LINUX_VIDIOC_G_AUDIO 0x5621 /* 0x80345621 */ #define LINUX_VIDIOC_S_AUDIO 0x5622 /* 0x40345622 */ #define LINUX_VIDIOC_QUERYCTRL 0x5624 /* 0xc0445624 */ #define LINUX_VIDIOC_QUERYMENU 0x5625 /* 0xc02c5625 */ #define LINUX_VIDIOC_G_INPUT 0x5626 /* 0x80045626 */ #define LINUX_VIDIOC_S_INPUT 0x5627 /* 0xc0045627 */ #define LINUX_VIDIOC_G_OUTPUT 0x562e /* 0x8004562e */ #define LINUX_VIDIOC_S_OUTPUT 0x562f /* 0xc004562f */ #define LINUX_VIDIOC_ENUMOUTPUT 0x5630 /* 0xc0485630 */ #define LINUX_VIDIOC_G_AUDOUT 0x5631 /* 0x80345631 */ #define LINUX_VIDIOC_S_AUDOUT 0x5632 /* 0x40345632 */ #define LINUX_VIDIOC_G_MODULATOR 0x5636 /* 0xc0445636 */ #define LINUX_VIDIOC_S_MODULATOR 0x5637 /* 0x40445637 */ #define LINUX_VIDIOC_G_FREQUENCY 0x5638 /* 0xc02c5638 */ #define LINUX_VIDIOC_S_FREQUENCY 0x5639 /* 0x402c5639 */ #define LINUX_VIDIOC_CROPCAP 0x563a /* 0xc02c563a */ #define LINUX_VIDIOC_G_CROP 0x563b /* 0xc014563b */ #define LINUX_VIDIOC_S_CROP 0x563c /* 0x4014563c */ #define LINUX_VIDIOC_G_JPEGCOMP 0x563d /* 0x808c563d */ #define LINUX_VIDIOC_S_JPEGCOMP 0x563e /* 0x408c563e */ #define LINUX_VIDIOC_QUERYSTD 0x563f /* 0x8008563f */ #define LINUX_VIDIOC_TRY_FMT 0x5640 /* 0xc0cc5640 */ #define LINUX_VIDIOC_ENUMAUDIO 0x5641 /* 0xc0345641 */ #define LINUX_VIDIOC_ENUMAUDOUT 0x5642 /* 0xc0345642 */ #define LINUX_VIDIOC_G_PRIORITY 0x5643 /* 0x80045643 */ #define LINUX_VIDIOC_S_PRIORITY 0x5644 /* 0x40045644 */ #define LINUX_VIDIOC_G_SLICED_VBI_CAP 0x5645 /* 0xc0745645 */ #define LINUX_VIDIOC_LOG_STATUS 0x5646 /* 0x00005646 */ #define LINUX_VIDIOC_G_EXT_CTRLS 0x5647 /* 0xc0185647 */ #define LINUX_VIDIOC_S_EXT_CTRLS 0x5648 /* 0xc0185648 */ #define LINUX_VIDIOC_TRY_EXT_CTRLS 0x5649 /* 0xc0185649 */ #define LINUX_VIDIOC_ENUM_FRAMESIZES 0x564a /* 0xc02c564a */ #define LINUX_VIDIOC_ENUM_FRAMEINTERVALS 0x564b /* 0xc034564b */ #define LINUX_VIDIOC_G_ENC_INDEX 0x564c /* 0x8818564c */ #define LINUX_VIDIOC_ENCODER_CMD 0x564d /* 0xc028564d */ #define LINUX_VIDIOC_TRY_ENCODER_CMD 0x564e /* 0xc028564e */ #define LINUX_VIDIOC_DBG_S_REGISTER 0x564f /* 0x4038564f */ #define LINUX_VIDIOC_DBG_G_REGISTER 0x5650 /* 0xc0385650 */ #define LINUX_VIDIOC_DBG_G_CHIP_IDENT 0x5651 /* 0xc02c5651 */ #define LINUX_VIDIOC_S_HW_FREQ_SEEK 0x5652 /* 0x40305652 */ #define LINUX_VIDIOC_ENUM_DV_PRESETS 0x5653 /* 0xc0405653 */ #define LINUX_VIDIOC_S_DV_PRESET 0x5654 /* 0xc0145654 */ #define LINUX_VIDIOC_G_DV_PRESET 0x5655 /* 0xc0145655 */ #define LINUX_VIDIOC_QUERY_DV_PRESET 0x5656 /* 0x80145656 */ #define LINUX_VIDIOC_S_DV_TIMINGS 0x5657 /* 0xc0845657 */ #define LINUX_VIDIOC_G_DV_TIMINGS 0x5658 /* 0xc0845658 */ #define LINUX_VIDIOC_DQEVENT 0x5659 /* 0x80785659 */ #define LINUX_VIDIOC_SUBSCRIBE_EVENT 0x565a /* 0x4020565a */ #define LINUX_VIDIOC_UNSUBSCRIBE_EVENT 0x565b /* 0x4020565b */ #define LINUX_VIDIOC_OVERLAY_OLD 0x560e /* 0xc004560e */ #define LINUX_VIDIOC_S_PARM_OLD 0x5616 /* 0x40cc5616 */ #define LINUX_VIDIOC_S_CTRL_OLD 0x561c /* 0x4008561c */ #define LINUX_VIDIOC_G_AUDIO_OLD 0x5621 /* 0xc0345621 */ #define LINUX_VIDIOC_G_AUDOUT_OLD 0x5631 /* 0xc0345631 */ #define LINUX_VIDIOC_CROPCAP_OLD 0x563a /* 0x802c563a */ #define LINUX_IOCTL_VIDEO2_MIN LINUX_VIDIOC_QUERYCAP #define LINUX_IOCTL_VIDEO2_MAX LINUX_VIDIOC_UNSUBSCRIBE_EVENT /* * Our libusb(8) calls emulated within linux(4). */ #define FBSD_LUSB_DEVICEENUMERATE 0xffff #define FBSD_LUSB_DEV_QUIRK_ADD 0xfffe #define FBSD_LUSB_DEV_QUIRK_GET 0xfffd #define FBSD_LUSB_DEV_QUIRK_REMOVE 0xfffc #define FBSD_LUSB_DO_REQUEST 0xfffb #define FBSD_LUSB_FS_CLEAR_STALL_SYNC 0xfffa #define FBSD_LUSB_FS_CLOSE 0xfff9 #define FBSD_LUSB_FS_COMPLETE 0xfff8 #define FBSD_LUSB_FS_INIT 0xfff7 #define FBSD_LUSB_FS_OPEN 0xfff6 #define FBSD_LUSB_FS_START 0xfff5 #define FBSD_LUSB_FS_STOP 0xfff4 #define FBSD_LUSB_FS_UNINIT 0xfff3 #define FBSD_LUSB_GET_CONFIG 0xfff2 #define FBSD_LUSB_GET_DEVICEINFO 0xfff1 #define FBSD_LUSB_GET_DEVICE_DESC 0xfff0 #define FBSD_LUSB_GET_FULL_DESC 0xffef #define FBSD_LUSB_GET_IFACE_DRIVER 0xffee #define FBSD_LUSB_GET_PLUGTIME 0xffed #define FBSD_LUSB_GET_POWER_MODE 0xffec #define FBSD_LUSB_GET_REPORT_DESC 0xffeb #define FBSD_LUSB_GET_REPORT_ID 0xffea #define FBSD_LUSB_GET_TEMPLATE 0xffe9 #define FBSD_LUSB_IFACE_DRIVER_ACTIVE 0xffe8 #define FBSD_LUSB_IFACE_DRIVER_DETACH 0xffe7 #define FBSD_LUSB_QUIRK_NAME_GET 0xffe6 #define FBSD_LUSB_READ_DIR 0xffe5 #define FBSD_LUSB_SET_ALTINTERFACE 0xffe4 #define FBSD_LUSB_SET_CONFIG 0xffe3 #define FBSD_LUSB_SET_IMMED 0xffe2 #define FBSD_LUSB_SET_POWER_MODE 0xffe1 #define FBSD_LUSB_SET_TEMPLATE 0xffe0 #define FBSD_LUSB_FS_OPEN_STREAM 0xffdf #define FBSD_LUSB_GET_DEV_PORT_PATH 0xffde #define FBSD_LUSB_GET_POWER_USAGE 0xffdd +#define FBSD_LUSB_DEVICESTATS 0xffdc #define FBSD_LUSB_MAX 0xffff -#define FBSD_LUSB_MIN 0xffdd +#define FBSD_LUSB_MIN 0xffdc /* * Linux btrfs clone operation */ #define LINUX_BTRFS_IOC_CLONE 0x9409 /* 0x40049409 */ #define LINUX_FS_IOC_FIEMAP 0x660b /* * Linux evdev ioctl min and max */ #define LINUX_IOCTL_EVDEV_MIN 0x4500 #define LINUX_IOCTL_EVDEV_MAX 0x45ff #define LINUX_IOCTL_EVDEV_CLK LINUX_CLOCK_REALTIME | \ LINUX_CLOCK_MONOTONIC |LINUX_CLOCK_BOOTTIME /* * Pluggable ioctl handlers */ struct linux_ioctl_args; struct thread; typedef int linux_ioctl_function_t(struct thread *, struct linux_ioctl_args *); struct linux_ioctl_handler { linux_ioctl_function_t *func; int low, high; }; struct linux_ioctl_handler_element { TAILQ_ENTRY(linux_ioctl_handler_element) list; int (*func)(struct thread *, struct linux_ioctl_args *); int low, high, span; }; int linux_ioctl_register_handler(struct linux_ioctl_handler *h); int linux_ioctl_unregister_handler(struct linux_ioctl_handler *h); #ifdef COMPAT_LINUX32 int linux32_ioctl_register_handler(struct linux_ioctl_handler *h); int linux32_ioctl_unregister_handler(struct linux_ioctl_handler *h); #endif #endif /* !_LINUX_IOCTL_H_ */ Index: projects/clang1000-import/sys/sys/param.h =================================================================== --- projects/clang1000-import/sys/sys/param.h (revision 358848) +++ projects/clang1000-import/sys/sys/param.h (revision 358849) @@ -1,368 +1,368 @@ /*- * SPDX-License-Identifier: BSD-3-Clause * * Copyright (c) 1982, 1986, 1989, 1993 * The Regents of the University of California. All rights reserved. * (c) UNIX System Laboratories, Inc. * All or some portions of this file are derived from material licensed * to the University of California by American Telephone and Telegraph * Co. or Unix System Laboratories, Inc. and are reproduced herein with * the permission of UNIX System Laboratories, Inc. * * 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. * * @(#)param.h 8.3 (Berkeley) 4/4/95 * $FreeBSD$ */ #ifndef _SYS_PARAM_H_ #define _SYS_PARAM_H_ #include #define BSD 199506 /* System version (year & month). */ #define BSD4_3 1 #define BSD4_4 1 /* * __FreeBSD_version numbers are documented in the Porter's Handbook. * If you bump the version for any reason, you should update the documentation * there. * Currently this lives here in the doc/ repository: * * head/en_US.ISO8859-1/books/porters-handbook/versions/chapter.xml * * scheme is: Rxx * 'R' is in the range 0 to 4 if this is a release branch or * X.0-CURRENT before releng/X.0 is created, otherwise 'R' is * in the range 5 to 9. */ #undef __FreeBSD_version -#define __FreeBSD_version 1300082 /* Master, propagated to newvers */ +#define __FreeBSD_version 1300083 /* Master, propagated to newvers */ /* * __FreeBSD_kernel__ indicates that this system uses the kernel of FreeBSD, * which by definition is always true on FreeBSD. This macro is also defined * on other systems that use the kernel of FreeBSD, such as GNU/kFreeBSD. * * It is tempting to use this macro in userland code when we want to enable * kernel-specific routines, and in fact it's fine to do this in code that * is part of FreeBSD itself. However, be aware that as presence of this * macro is still not widespread (e.g. older FreeBSD versions, 3rd party * compilers, etc), it is STRONGLY DISCOURAGED to check for this macro in * external applications without also checking for __FreeBSD__ as an * alternative. */ #undef __FreeBSD_kernel__ #define __FreeBSD_kernel__ #if defined(_KERNEL) || defined(IN_RTLD) #define P_OSREL_SIGWAIT 700000 #define P_OSREL_SIGSEGV 700004 #define P_OSREL_MAP_ANON 800104 #define P_OSREL_MAP_FSTRICT 1100036 #define P_OSREL_SHUTDOWN_ENOTCONN 1100077 #define P_OSREL_MAP_GUARD 1200035 #define P_OSREL_WRFSBASE 1200041 #define P_OSREL_CK_CYLGRP 1200046 #define P_OSREL_VMTOTAL64 1200054 #define P_OSREL_CK_SUPERBLOCK 1300000 #define P_OSREL_CK_INODE 1300005 #define P_OSREL_POWERPC_NEW_AUX_ARGS 1300070 #define P_OSREL_MAJOR(x) ((x) / 100000) #endif #ifndef LOCORE #include #endif /* * Machine-independent constants (some used in following include files). * Redefined constants are from POSIX 1003.1 limits file. * * MAXCOMLEN should be >= sizeof(ac_comm) (see ) */ #include #define MAXCOMLEN 19 /* max command name remembered */ #define MAXINTERP PATH_MAX /* max interpreter file name length */ #define MAXLOGNAME 33 /* max login name length (incl. NUL) */ #define MAXUPRC CHILD_MAX /* max simultaneous processes */ #define NCARGS ARG_MAX /* max bytes for an exec function */ #define NGROUPS (NGROUPS_MAX+1) /* max number groups */ #define NOFILE OPEN_MAX /* max open files per process */ #define NOGROUP 65535 /* marker for empty group set member */ #define MAXHOSTNAMELEN 256 /* max hostname size */ #define SPECNAMELEN 255 /* max length of devicename */ /* More types and definitions used throughout the kernel. */ #ifdef _KERNEL #include #include #ifndef LOCORE #include #include #endif #ifndef FALSE #define FALSE 0 #endif #ifndef TRUE #define TRUE 1 #endif #endif #ifndef _KERNEL /* Signals. */ #include #endif /* Machine type dependent parameters. */ #include #ifndef _KERNEL #include #endif #ifndef DEV_BSHIFT #define DEV_BSHIFT 9 /* log2(DEV_BSIZE) */ #endif #define DEV_BSIZE (1<>PAGE_SHIFT) #endif /* * btodb() is messy and perhaps slow because `bytes' may be an off_t. We * want to shift an unsigned type to avoid sign extension and we don't * want to widen `bytes' unnecessarily. Assume that the result fits in * a daddr_t. */ #ifndef btodb #define btodb(bytes) /* calculates (bytes / DEV_BSIZE) */ \ (sizeof (bytes) > sizeof(long) \ ? (daddr_t)((unsigned long long)(bytes) >> DEV_BSHIFT) \ : (daddr_t)((unsigned long)(bytes) >> DEV_BSHIFT)) #endif #ifndef dbtob #define dbtob(db) /* calculates (db * DEV_BSIZE) */ \ ((off_t)(db) << DEV_BSHIFT) #endif #define PRIMASK 0x0ff #define PCATCH 0x100 /* OR'd with pri for tsleep to check signals */ #define PDROP 0x200 /* OR'd with pri to stop re-entry of interlock mutex */ #define NZERO 0 /* default "nice" */ #define NBBY 8 /* number of bits in a byte */ #define NBPW sizeof(int) /* number of bytes per word (integer) */ #define CMASK 022 /* default file mask: S_IWGRP|S_IWOTH */ #define NODEV (dev_t)(-1) /* non-existent device */ /* * File system parameters and macros. * * MAXBSIZE - Filesystems are made out of blocks of at most MAXBSIZE bytes * per block. MAXBSIZE may be made larger without effecting * any existing filesystems as long as it does not exceed MAXPHYS, * and may be made smaller at the risk of not being able to use * filesystems which require a block size exceeding MAXBSIZE. * * MAXBCACHEBUF - Maximum size of a buffer in the buffer cache. This must * be >= MAXBSIZE and can be set differently for different * architectures by defining it in . * Making this larger allows NFS to do larger reads/writes. * * BKVASIZE - Nominal buffer space per buffer, in bytes. BKVASIZE is the * minimum KVM memory reservation the kernel is willing to make. * Filesystems can of course request smaller chunks. Actual * backing memory uses a chunk size of a page (PAGE_SIZE). * The default value here can be overridden on a per-architecture * basis by defining it in . * * If you make BKVASIZE too small you risk seriously fragmenting * the buffer KVM map which may slow things down a bit. If you * make it too big the kernel will not be able to optimally use * the KVM memory reserved for the buffer cache and will wind * up with too-few buffers. * * The default is 16384, roughly 2x the block size used by a * normal UFS filesystem. */ #define MAXBSIZE 65536 /* must be power of 2 */ #ifndef MAXBCACHEBUF #define MAXBCACHEBUF MAXBSIZE /* must be a power of 2 >= MAXBSIZE */ #endif #ifndef BKVASIZE #define BKVASIZE 16384 /* must be power of 2 */ #endif #define BKVAMASK (BKVASIZE-1) /* * MAXPATHLEN defines the longest permissible path length after expanding * symbolic links. It is used to allocate a temporary buffer from the buffer * pool in which to do the name expansion, hence should be a power of two, * and must be less than or equal to MAXBSIZE. MAXSYMLINKS defines the * maximum number of symbolic links that may be expanded in a path name. * It should be set high enough to allow all legitimate uses, but halt * infinite loops reasonably quickly. */ #define MAXPATHLEN PATH_MAX #define MAXSYMLINKS 32 /* Bit map related macros. */ #define setbit(a,i) (((unsigned char *)(a))[(i)/NBBY] |= 1<<((i)%NBBY)) #define clrbit(a,i) (((unsigned char *)(a))[(i)/NBBY] &= ~(1<<((i)%NBBY))) #define isset(a,i) \ (((const unsigned char *)(a))[(i)/NBBY] & (1<<((i)%NBBY))) #define isclr(a,i) \ ((((const unsigned char *)(a))[(i)/NBBY] & (1<<((i)%NBBY))) == 0) /* Macros for counting and rounding. */ #ifndef howmany #define howmany(x, y) (((x)+((y)-1))/(y)) #endif #define nitems(x) (sizeof((x)) / sizeof((x)[0])) #define rounddown(x, y) (((x)/(y))*(y)) #define rounddown2(x, y) ((x)&(~((y)-1))) /* if y is power of two */ #define roundup(x, y) ((((x)+((y)-1))/(y))*(y)) /* to any y */ #define roundup2(x, y) (((x)+((y)-1))&(~((y)-1))) /* if y is powers of two */ #define powerof2(x) ((((x)-1)&(x))==0) /* Macros for min/max. */ #define MIN(a,b) (((a)<(b))?(a):(b)) #define MAX(a,b) (((a)>(b))?(a):(b)) #ifdef _KERNEL /* * Basic byte order function prototypes for non-inline functions. */ #ifndef LOCORE #ifndef _BYTEORDER_PROTOTYPED #define _BYTEORDER_PROTOTYPED __BEGIN_DECLS __uint32_t htonl(__uint32_t); __uint16_t htons(__uint16_t); __uint32_t ntohl(__uint32_t); __uint16_t ntohs(__uint16_t); __END_DECLS #endif #endif #ifndef _BYTEORDER_FUNC_DEFINED #define _BYTEORDER_FUNC_DEFINED #define htonl(x) __htonl(x) #define htons(x) __htons(x) #define ntohl(x) __ntohl(x) #define ntohs(x) __ntohs(x) #endif /* !_BYTEORDER_FUNC_DEFINED */ #endif /* _KERNEL */ /* * Scale factor for scaled integers used to count %cpu time and load avgs. * * The number of CPU `tick's that map to a unique `%age' can be expressed * by the formula (1 / (2 ^ (FSHIFT - 11))). The maximum load average that * can be calculated (assuming 32 bits) can be closely approximated using * the formula (2 ^ (2 * (16 - FSHIFT))) for (FSHIFT < 15). * * For the scheduler to maintain a 1:1 mapping of CPU `tick' to `%age', * FSHIFT must be at least 11; this gives us a maximum load avg of ~1024. */ #define FSHIFT 11 /* bits to right of fixed binary point */ #define FSCALE (1<> (PAGE_SHIFT - DEV_BSHIFT)) #define ctodb(db) /* calculates pages to devblks */ \ ((db) << (PAGE_SHIFT - DEV_BSHIFT)) /* * Old spelling of __containerof(). */ #define member2struct(s, m, x) \ ((struct s *)(void *)((char *)(x) - offsetof(struct s, m))) /* * Access a variable length array that has been declared as a fixed * length array. */ #define __PAST_END(array, offset) (((__typeof__(*(array)) *)(array))[offset]) #endif /* _SYS_PARAM_H_ */ Index: projects/clang1000-import =================================================================== --- projects/clang1000-import (revision 358848) +++ projects/clang1000-import (revision 358849) Property changes on: projects/clang1000-import ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head:r358832-358848