diff options
36 files changed, 919 insertions, 280 deletions
diff --git a/Makefile.inc1 b/Makefile.inc1 index 9dc4f2db4a6c..74c4598dd092 100644 --- a/Makefile.inc1 +++ b/Makefile.inc1 @@ -2240,7 +2240,7 @@ _default_flavor= -default _debug=-dbg . endif -create-dtb-package: +create-dtb-package: .PHONY @if [ -f ${KSTAGEDIR}/${DISTDIR}/dtb.plist ]; then \ ${SRCDIR}/release/packages/generate-ucl.lua \ PKGNAME "dtb" \ @@ -2265,9 +2265,12 @@ create-dtb-package: -o ${REPODIR}/${PKG_ABI}/${PKG_OUTPUT_DIR} ; \ fi -create-kernel-packages: .PHONY +create-kernel-packages: .PHONY create-kernel-flavored-packages create-dtb-package +create-kernel-flavored-packages: .PHONY +.ORDER: create-kernel-flavored-packages create-dtb-package + . for flavor in "" ${_debug} -create-kernel-packages: create-kernel-packages-flavor${flavor:C,^""$,${_default_flavor},} create-dtb-package +create-kernel-flavored-packages: create-kernel-packages-flavor${flavor:C,^""$,${_default_flavor},} create-kernel-packages-flavor${flavor:C,^""$,${_default_flavor},}: _pkgbootstrap .PHONY @cd ${KSTAGEDIR}/${DISTDIR} ; \ ${METALOG_SORT_CMD} ${KSTAGEDIR}/kernel.meta | \ diff --git a/etc/mtree/BSD.include.dist b/etc/mtree/BSD.include.dist index 28c4d91ac1c0..454657183c58 100644 --- a/etc/mtree/BSD.include.dist +++ b/etc/mtree/BSD.include.dist @@ -276,9 +276,7 @@ .. kadm5 .. - kdb5 - .. - krb5 + krb5 tags=package=kerberos-dev .. lib80211 .. diff --git a/etc/mtree/BSD.usr.dist b/etc/mtree/BSD.usr.dist index d7d839b94b96..6a8c155e5e73 100644 --- a/etc/mtree/BSD.usr.dist +++ b/etc/mtree/BSD.usr.dist @@ -85,15 +85,15 @@ .. i18n .. - krb5 - kdb + krb5 tags=package=kerberos-kdc + kdb tags=package=kerberos-kdc .. - plugins - kdb + plugins tags=package=kerberos-kdc + kdb tags=package=kerberos-kdc .. - preauth + preauth tags=package=kerberos-kdc .. - tls + tls tags=package=kerberos-kdc .. .. .. @@ -269,7 +269,7 @@ .. dtrace .. - et + et tags=package=kerberos-dev .. examples BSD_daemon diff --git a/etc/mtree/BSD.var.dist b/etc/mtree/BSD.var.dist index d8593e61c49e..b3372196f5f9 100644 --- a/etc/mtree/BSD.var.dist +++ b/etc/mtree/BSD.var.dist @@ -45,7 +45,7 @@ .. ipf mode=0700 tags=package=ipf .. - krb5kdc mode=0700 + krb5kdc mode=0700 tags=package=kerberos-kdc .. mtree .. diff --git a/lib/libbz2/Makefile b/lib/libbz2/Makefile index d773f202dd67..2aedbaed4328 100644 --- a/lib/libbz2/Makefile +++ b/lib/libbz2/Makefile @@ -13,4 +13,17 @@ CFLAGS+= -I${BZ2DIR} WARNS?= 3 +BZIP2_VERSION!= sed -n '/bzip2\/libbzip2 version /{s/.*version //;s/ of.*//p;q;}' ${BZ2DIR}/bzlib.h + +bzip2.pc: bzip2.pc.in + sed -e 's,@prefix@,/usr,g ; \ + s,@exec_prefix@,$${prefix},g ; \ + s,@libdir@,${LIBDIR},g ; \ + s,@sharedlibdir@,${SHLIBDIR},g ; \ + s,@includedir@,${INCLUDEDIR},g ; \ + s,@VERSION@,${BZIP2_VERSION},g ;' \ + ${.ALLSRC} > ${.TARGET} + +PCFILES= bzip2.pc + .include <bsd.lib.mk> diff --git a/lib/libbz2/bzip2.pc.in b/lib/libbz2/bzip2.pc.in new file mode 100644 index 000000000000..d91c9931a58a --- /dev/null +++ b/lib/libbz2/bzip2.pc.in @@ -0,0 +1,11 @@ +prefix=@prefix@ +exec_prefix=@exec_prefix@ +libdir=@libdir@ +sharedlibdir=@sharedlibdir@ +includedir=@includedir@ + +Name: bzip2 +Description: bzip2 compression library +Version: @VERSION@ +Libs: -L${libdir} -lbz2 +Cflags: -I${includedir} diff --git a/lib/libsys/getgroups.2 b/lib/libsys/getgroups.2 index 4881a65d532e..4e94b32d4e7b 100644 --- a/lib/libsys/getgroups.2 +++ b/lib/libsys/getgroups.2 @@ -33,7 +33,7 @@ .\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF .\" SUCH DAMAGE. .\" -.Dd September 17, 2025 +.Dd October 10, 2025 .Dt GETGROUPS 2 .Os .Sh NAME @@ -107,10 +107,8 @@ array. The .Fn getgroups system call conforms to -.St -p1003.1-2008 -with the additional properties that supplementary groups are reported in -strictly ascending order and the returned size coincides with the cardinal of -the set. +.St -p1003.1-2008 , +not reporting the effective group ID. .Sh HISTORY The .Fn getgroups @@ -121,8 +119,8 @@ Since .Fx 14.3 , the .Fn getgroups -system call has treated the supplementary groups as a set, reporting them in -strictly ascending order and returning the cardinal of the set. +system call has been reporting the supplementary groups in strictly ascending +order. .Pp Before .Fx 15.0 , @@ -138,15 +136,14 @@ system call gets the supplementary groups set in the array. In particular, as evoked in .Sx HISTORY , -it does not anymore retrieve the effective GID in the first slot of +it does not anymore retrieve the effective group ID in the first slot of .Fa gidset . -Programs should not make any assumption about which group is placed in the first -slot of -.Fa gidset -other than it being the supplementary group with smallest GID. +Programs that process this slot in a specific way must be modified to obtain the +effective group ID through other means, such as a call to +.Xr getegid 2 . .Pp -The effective GID is present in the supplementary groups set if and only if it -was explicitly set as a supplementary group. +The effective group ID is present in the supplementary groups set if and only if +it was explicitly set as a supplementary group. The function .Fn initgroups enforces that, while the diff --git a/libexec/rc/rc.d/virtual_oss b/libexec/rc/rc.d/virtual_oss index 4f5c34ce03f3..b9c830617385 100644 --- a/libexec/rc/rc.d/virtual_oss +++ b/libexec/rc/rc.d/virtual_oss @@ -1,8 +1,8 @@ #!/bin/sh # PROVIDE: virtual_oss -# REQUIRE: kld ldconfig -# BEFORE: LOGIN sndiod +# REQUIRE: NETWORKING kld ldconfig +# BEFORE: LOGIN # KEYWORD: shutdown . /etc/rc.subr diff --git a/release/tools/ec2-base.conf b/release/tools/ec2-base.conf index b6354db3d9d1..ffe2646240a5 100644 --- a/release/tools/ec2-base.conf +++ b/release/tools/ec2-base.conf @@ -36,5 +36,9 @@ vm_extra_pre_umount() { # Standard FreeBSD network configuration ec2_base_networking + # Add files from packages which weren't recorded in metalog + metalog_add_data ./usr/local/etc/dhclient.conf + metalog_add_data ./usr/local/etc/ssl/cert.pem + return 0 } diff --git a/release/tools/ec2-builder.conf b/release/tools/ec2-builder.conf index a55485fec0cd..bcea69331be5 100644 --- a/release/tools/ec2-builder.conf +++ b/release/tools/ec2-builder.conf @@ -66,5 +66,9 @@ vm_extra_pre_umount() { EOF metalog_add_data ./boot/loader.conf + # Add files from packages which weren't recorded in metalog + metalog_add_data ./usr/local/etc/dhclient.conf + metalog_add_data ./usr/local/etc/ssl/cert.pem + return 0 } diff --git a/release/tools/ec2-small.conf b/release/tools/ec2-small.conf index acaffbbc0c42..f12afec75a4f 100644 --- a/release/tools/ec2-small.conf +++ b/release/tools/ec2-small.conf @@ -49,5 +49,9 @@ vm_extra_pre_umount() { # Standard FreeBSD network configuration ec2_base_networking + # Add files from packages which weren't recorded in metalog + metalog_add_data ./usr/local/etc/dhclient.conf + metalog_add_data ./usr/local/etc/ssl/cert.pem + return 0 } diff --git a/release/tools/oci-image-notoolchain.conf b/release/tools/oci-image-notoolchain.conf index a769b53f9ff6..72a62657fa76 100644 --- a/release/tools/oci-image-notoolchain.conf +++ b/release/tools/oci-image-notoolchain.conf @@ -1,67 +1,25 @@ #! /bin/sh -# Build OCI container image with almost all packages suitable for jails, excluding compiler +# Build OCI container image with almost all packages suitable for jails, +# excluding toolchain. OCI_BASE_IMAGE=runtime oci_image_build() { set_cmd ${workdir} /bin/sh install_packages ${abi} ${workdir} \ - FreeBSD-acct \ - FreeBSD-at \ - FreeBSD-audit \ - FreeBSD-autofs \ - FreeBSD-blocklist \ - FreeBSD-bsnmp \ - FreeBSD-caroot \ - FreeBSD-certctl \ - FreeBSD-clibs \ - FreeBSD-console-tools \ - FreeBSD-cron \ - FreeBSD-csh \ + FreeBSD-bmake \ FreeBSD-dma \ - FreeBSD-ee \ - FreeBSD-fd \ - FreeBSD-fetch \ - FreeBSD-ftp \ FreeBSD-inetd \ FreeBSD-ipf \ FreeBSD-ipfw \ - FreeBSD-iscsi \ - FreeBSD-jail \ - FreeBSD-kerberos \ - FreeBSD-kerberos-lib \ - FreeBSD-libarchive \ - FreeBSD-libcompiler_rt-dev \ - FreeBSD-libexecinfo \ - FreeBSD-libucl \ - FreeBSD-locales \ - FreeBSD-lp \ - FreeBSD-mtree \ FreeBSD-natd \ FreeBSD-netmap \ - FreeBSD-newsyslog \ - FreeBSD-nfs \ - FreeBSD-nuageinit \ - FreeBSD-openssl \ - FreeBSD-periodic \ FreeBSD-pf \ - FreeBSD-pkg-bootstrap \ - FreeBSD-quotacheck \ - FreeBSD-rc \ - FreeBSD-rcmds \ - FreeBSD-rescue \ - FreeBSD-resolvconf \ - FreeBSD-runtime \ FreeBSD-ssh \ - FreeBSD-syslogd \ - FreeBSD-tcpd \ FreeBSD-telnet \ - FreeBSD-ufs \ - FreeBSD-unbound \ - FreeBSD-utilities \ - FreeBSD-vi \ - FreeBSD-yp \ + FreeBSD-xz \ FreeBSD-zfs \ - FreeBSD-zoneinfo + FreeBSD-set-minimal-jail } + diff --git a/release/tools/vmimage.subr b/release/tools/vmimage.subr index 067eb5ea48a8..15c4dd53e70b 100644 --- a/release/tools/vmimage.subr +++ b/release/tools/vmimage.subr @@ -213,6 +213,16 @@ vm_extra_install_packages() { install -y -r ${PKG_REPO_NAME} $pkg done metalog_add_data ./var/db/pkg/local.sqlite + + # Add some database files which are created by pkg triggers; + # at some point in the future the tools which create these + # files should probably learn how to record them in METALOG + # (which would simplify no-root installworld as well). + metalog_add_data ./etc/login.conf.db + metalog_add_data ./etc/passwd + metalog_add_data ./etc/pwd.db + metalog_add_data ./etc/spwd.db 600 + metalog_add_data ./var/db/services.db else if [ -n "${WITHOUT_QEMU}" ]; then return 0 diff --git a/sys/amd64/include/vmm.h b/sys/amd64/include/vmm.h index e35119af8572..66d8991d36e8 100644 --- a/sys/amd64/include/vmm.h +++ b/sys/amd64/include/vmm.h @@ -170,55 +170,63 @@ struct vm_eventinfo { int *iptr; /* reqidle cookie */ }; -typedef int (*vmm_init_func_t)(int ipinum); -typedef int (*vmm_cleanup_func_t)(void); -typedef void (*vmm_suspend_func_t)(void); -typedef void (*vmm_resume_func_t)(void); -typedef void * (*vmi_init_func_t)(struct vm *vm, struct pmap *pmap); -typedef int (*vmi_run_func_t)(void *vcpui, register_t rip, - struct pmap *pmap, struct vm_eventinfo *info); -typedef void (*vmi_cleanup_func_t)(void *vmi); -typedef void * (*vmi_vcpu_init_func_t)(void *vmi, struct vcpu *vcpu, - int vcpu_id); -typedef void (*vmi_vcpu_cleanup_func_t)(void *vcpui); -typedef int (*vmi_get_register_t)(void *vcpui, int num, uint64_t *retval); -typedef int (*vmi_set_register_t)(void *vcpui, int num, uint64_t val); -typedef int (*vmi_get_desc_t)(void *vcpui, int num, struct seg_desc *desc); -typedef int (*vmi_set_desc_t)(void *vcpui, int num, struct seg_desc *desc); -typedef int (*vmi_get_cap_t)(void *vcpui, int num, int *retval); -typedef int (*vmi_set_cap_t)(void *vcpui, int num, int val); -typedef struct vmspace * (*vmi_vmspace_alloc)(vm_offset_t min, vm_offset_t max); -typedef void (*vmi_vmspace_free)(struct vmspace *vmspace); -typedef struct vlapic * (*vmi_vlapic_init)(void *vcpui); -typedef void (*vmi_vlapic_cleanup)(struct vlapic *vlapic); -typedef int (*vmi_snapshot_vcpu_t)(void *vcpui, struct vm_snapshot_meta *meta); -typedef int (*vmi_restore_tsc_t)(void *vcpui, uint64_t now); +#define DECLARE_VMMOPS_FUNC(ret_type, opname, args) \ + typedef ret_type (*vmmops_##opname##_t) args; \ + ret_type vmmops_##opname args + +DECLARE_VMMOPS_FUNC(int, modinit, (int ipinum)); +DECLARE_VMMOPS_FUNC(int, modcleanup, (void)); +DECLARE_VMMOPS_FUNC(void, modresume, (void)); +DECLARE_VMMOPS_FUNC(void, modsuspend, (void)); +DECLARE_VMMOPS_FUNC(void *, init, (struct vm *vm, struct pmap *pmap)); +DECLARE_VMMOPS_FUNC(int, run, (void *vcpui, register_t pc, + struct pmap *pmap, struct vm_eventinfo *info)); +DECLARE_VMMOPS_FUNC(void, cleanup, (void *vmi)); +DECLARE_VMMOPS_FUNC(void *, vcpu_init, (void *vmi, struct vcpu *vcpu, + int vcpu_id)); +DECLARE_VMMOPS_FUNC(void, vcpu_cleanup, (void *vcpui)); +DECLARE_VMMOPS_FUNC(int, getreg, (void *vcpui, int num, uint64_t *retval)); +DECLARE_VMMOPS_FUNC(int, setreg, (void *vcpui, int num, uint64_t val)); +DECLARE_VMMOPS_FUNC(int, getdesc, (void *vcpui, int num, + struct seg_desc *desc)); +DECLARE_VMMOPS_FUNC(int, setdesc, (void *vcpui, int num, + struct seg_desc *desc)); +DECLARE_VMMOPS_FUNC(int, getcap, (void *vcpui, int num, int *retval)); +DECLARE_VMMOPS_FUNC(int, setcap, (void *vcpui, int num, int val)); +DECLARE_VMMOPS_FUNC(struct vmspace *, vmspace_alloc, + (vm_offset_t min, vm_offset_t max)); +DECLARE_VMMOPS_FUNC(void, vmspace_free, (struct vmspace *vmspace)); +DECLARE_VMMOPS_FUNC(struct vlapic *, vlapic_init, (void *vcpui)); +DECLARE_VMMOPS_FUNC(void, vlapic_cleanup, (struct vlapic *vlapic)); +DECLARE_VMMOPS_FUNC(int, vcpu_snapshot, (void *vcpui, + struct vm_snapshot_meta *meta)); +DECLARE_VMMOPS_FUNC(int, restore_tsc, (void *vcpui, uint64_t now)); struct vmm_ops { - vmm_init_func_t modinit; /* module wide initialization */ - vmm_cleanup_func_t modcleanup; - vmm_resume_func_t modsuspend; - vmm_resume_func_t modresume; - - vmi_init_func_t init; /* vm-specific initialization */ - vmi_run_func_t run; - vmi_cleanup_func_t cleanup; - vmi_vcpu_init_func_t vcpu_init; - vmi_vcpu_cleanup_func_t vcpu_cleanup; - vmi_get_register_t getreg; - vmi_set_register_t setreg; - vmi_get_desc_t getdesc; - vmi_set_desc_t setdesc; - vmi_get_cap_t getcap; - vmi_set_cap_t setcap; - vmi_vmspace_alloc vmspace_alloc; - vmi_vmspace_free vmspace_free; - vmi_vlapic_init vlapic_init; - vmi_vlapic_cleanup vlapic_cleanup; + vmmops_modinit_t modinit; /* module wide initialization */ + vmmops_modcleanup_t modcleanup; + vmmops_modresume_t modsuspend; + vmmops_modresume_t modresume; + + vmmops_init_t init; /* vm-specific initialization */ + vmmops_run_t run; + vmmops_cleanup_t cleanup; + vmmops_vcpu_init_t vcpu_init; + vmmops_vcpu_cleanup_t vcpu_cleanup; + vmmops_getreg_t getreg; + vmmops_setreg_t setreg; + vmmops_getdesc_t getdesc; + vmmops_setdesc_t setdesc; + vmmops_getcap_t getcap; + vmmops_setcap_t setcap; + vmmops_vmspace_alloc_t vmspace_alloc; + vmmops_vmspace_free_t vmspace_free; + vmmops_vlapic_init_t vlapic_init; + vmmops_vlapic_cleanup_t vlapic_cleanup; /* checkpoint operations */ - vmi_snapshot_vcpu_t vcpu_snapshot; - vmi_restore_tsc_t restore_tsc; + vmmops_vcpu_snapshot_t vcpu_snapshot; + vmmops_restore_tsc_t restore_tsc; }; extern const struct vmm_ops vmm_ops_intel; @@ -375,7 +383,6 @@ vcpu_should_yield(struct vcpu *vcpu) void *vcpu_stats(struct vcpu *vcpu); void vcpu_notify_event(struct vcpu *vcpu, bool lapic_intr); -struct vmspace *vm_vmspace(struct vm *vm); struct vm_mem *vm_mem(struct vm *vm); struct vatpic *vm_atpic(struct vm *vm); struct vatpit *vm_atpit(struct vm *vm); diff --git a/sys/amd64/vmm/vmm.c b/sys/amd64/vmm/vmm.c index c42da02d0bf6..2ac076551165 100644 --- a/sys/amd64/vmm/vmm.c +++ b/sys/amd64/vmm/vmm.c @@ -163,7 +163,6 @@ struct vm { void *rendezvous_arg; /* (x) [r] rendezvous func/arg */ vm_rendezvous_func_t rendezvous_func; struct mtx rendezvous_mtx; /* (o) rendezvous lock */ - struct vmspace *vmspace; /* (o) guest's address space */ struct vm_mem mem; /* (i) [m+v] guest memory */ char name[VM_MAX_NAMELEN+1]; /* (o) virtual machine name */ struct vcpu **vcpu; /* (o) guest vcpus */ @@ -201,7 +200,7 @@ vmmops_panic(void) } #define DEFINE_VMMOPS_IFUNC(ret_type, opname, args) \ - DEFINE_IFUNC(static, ret_type, vmmops_##opname, args) \ + DEFINE_IFUNC(, ret_type, vmmops_##opname, args) \ { \ if (vmm_is_intel()) \ return (vmm_ops_intel.opname); \ @@ -499,7 +498,7 @@ MODULE_VERSION(vmm, 1); static void vm_init(struct vm *vm, bool create) { - vm->cookie = vmmops_init(vm, vmspace_pmap(vm->vmspace)); + vm->cookie = vmmops_init(vm, vmspace_pmap(vm_vmspace(vm))); vm->iommu = NULL; vm->vioapic = vioapic_init(vm); vm->vhpet = vhpet_init(vm); @@ -584,7 +583,7 @@ int vm_create(const char *name, struct vm **retvm) { struct vm *vm; - struct vmspace *vmspace; + int error; /* * If vmm.ko could not be successfully initialized then don't attempt @@ -597,14 +596,13 @@ vm_create(const char *name, struct vm **retvm) VM_MAX_NAMELEN + 1) return (EINVAL); - vmspace = vmmops_vmspace_alloc(0, VM_MAXUSER_ADDRESS_LA48); - if (vmspace == NULL) - return (ENOMEM); - vm = malloc(sizeof(struct vm), M_VM, M_WAITOK | M_ZERO); + error = vm_mem_init(&vm->mem, 0, VM_MAXUSER_ADDRESS_LA48); + if (error != 0) { + free(vm, M_VM); + return (error); + } strcpy(vm->name, name); - vm->vmspace = vmspace; - vm_mem_init(&vm->mem); mtx_init(&vm->rendezvous_mtx, "vm rendezvous lock", 0, MTX_DEF); sx_init(&vm->vcpus_init_lock, "vm vcpus"); vm->vcpu = malloc(sizeof(*vm->vcpu) * vm_maxcpu, M_VM, M_WAITOK | @@ -685,9 +683,6 @@ vm_cleanup(struct vm *vm, bool destroy) if (destroy) { vm_mem_destroy(vm); - vmmops_vmspace_free(vm->vmspace); - vm->vmspace = NULL; - free(vm->vcpu, M_VM); sx_destroy(&vm->vcpus_init_lock); mtx_destroy(&vm->rendezvous_mtx); @@ -731,7 +726,7 @@ vm_map_mmio(struct vm *vm, vm_paddr_t gpa, size_t len, vm_paddr_t hpa) { vm_object_t obj; - if ((obj = vmm_mmio_alloc(vm->vmspace, gpa, len, hpa)) == NULL) + if ((obj = vmm_mmio_alloc(vm_vmspace(vm), gpa, len, hpa)) == NULL) return (ENOMEM); else return (0); @@ -741,19 +736,21 @@ int vm_unmap_mmio(struct vm *vm, vm_paddr_t gpa, size_t len) { - vmm_mmio_free(vm->vmspace, gpa, len); + vmm_mmio_free(vm_vmspace(vm), gpa, len); return (0); } static int vm_iommu_map(struct vm *vm) { + pmap_t pmap; vm_paddr_t gpa, hpa; struct vm_mem_map *mm; int error, i; sx_assert(&vm->mem.mem_segs_lock, SX_LOCKED); + pmap = vmspace_pmap(vm_vmspace(vm)); for (i = 0; i < VM_MAX_MEMMAPS; i++) { if (!vm_memseg_sysmem(vm, i)) continue; @@ -767,7 +764,7 @@ vm_iommu_map(struct vm *vm) mm->flags |= VM_MEMMAP_F_IOMMU; for (gpa = mm->gpa; gpa < mm->gpa + mm->len; gpa += PAGE_SIZE) { - hpa = pmap_extract(vmspace_pmap(vm->vmspace), gpa); + hpa = pmap_extract(pmap, gpa); /* * All mappings in the vmm vmspace must be @@ -816,7 +813,7 @@ vm_iommu_unmap(struct vm *vm) for (gpa = mm->gpa; gpa < mm->gpa + mm->len; gpa += PAGE_SIZE) { KASSERT(vm_page_wired(PHYS_TO_VM_PAGE(pmap_extract( - vmspace_pmap(vm->vmspace), gpa))), + vmspace_pmap(vm_vmspace(vm)), gpa))), ("vm_iommu_unmap: vm %p gpa %jx not wired", vm, (uintmax_t)gpa)); iommu_remove_mapping(vm->iommu, gpa, PAGE_SIZE); @@ -1249,7 +1246,7 @@ vm_handle_paging(struct vcpu *vcpu, bool *retu) ("vm_handle_paging: invalid fault_type %d", ftype)); if (ftype == VM_PROT_READ || ftype == VM_PROT_WRITE) { - rv = pmap_emulate_accessed_dirty(vmspace_pmap(vm->vmspace), + rv = pmap_emulate_accessed_dirty(vmspace_pmap(vm_vmspace(vm)), vme->u.paging.gpa, ftype); if (rv == 0) { VMM_CTR2(vcpu, "%s bit emulation for gpa %#lx", @@ -1259,7 +1256,7 @@ vm_handle_paging(struct vcpu *vcpu, bool *retu) } } - map = &vm->vmspace->vm_map; + map = &vm_vmspace(vm)->vm_map; rv = vm_fault(map, vme->u.paging.gpa, ftype, VM_FAULT_NORMAL, NULL); VMM_CTR3(vcpu, "vm_handle_paging rv = %d, gpa = %#lx, " @@ -1560,7 +1557,7 @@ vm_run(struct vcpu *vcpu) if (CPU_ISSET(vcpuid, &vm->suspended_cpus)) return (EINVAL); - pmap = vmspace_pmap(vm->vmspace); + pmap = vmspace_pmap(vm_vmspace(vm)); vme = &vcpu->exitinfo; evinfo.rptr = &vm->rendezvous_req_cpus; evinfo.sptr = &vm->suspend; @@ -2302,12 +2299,6 @@ vcpu_notify_event(struct vcpu *vcpu, bool lapic_intr) vcpu_unlock(vcpu); } -struct vmspace * -vm_vmspace(struct vm *vm) -{ - return (vm->vmspace); -} - struct vm_mem * vm_mem(struct vm *vm) { @@ -2519,7 +2510,7 @@ vm_get_rescnt(struct vcpu *vcpu, struct vmm_stat_type *stat) if (vcpu->vcpuid == 0) { vmm_stat_set(vcpu, VMM_MEM_RESIDENT, PAGE_SIZE * - vmspace_resident_count(vcpu->vm->vmspace)); + vmspace_resident_count(vm_vmspace(vcpu->vm))); } } @@ -2529,7 +2520,7 @@ vm_get_wiredcnt(struct vcpu *vcpu, struct vmm_stat_type *stat) if (vcpu->vcpuid == 0) { vmm_stat_set(vcpu, VMM_MEM_WIRED, PAGE_SIZE * - pmap_wired_count(vmspace_pmap(vcpu->vm->vmspace))); + pmap_wired_count(vmspace_pmap(vm_vmspace(vcpu->vm)))); } } diff --git a/sys/amd64/vmm/vmm_dev_machdep.c b/sys/amd64/vmm/vmm_dev_machdep.c index d8d2b460404c..dfebc9dcadbf 100644 --- a/sys/amd64/vmm/vmm_dev_machdep.c +++ b/sys/amd64/vmm/vmm_dev_machdep.c @@ -48,6 +48,7 @@ #include <x86/apicreg.h> #include <dev/vmm/vmm_dev.h> +#include <dev/vmm/vmm_mem.h> #include <dev/vmm/vmm_stat.h> #include "vmm_lapic.h" diff --git a/sys/arm64/include/vmm.h b/sys/arm64/include/vmm.h index e839b5dd92c9..84b286a60b38 100644 --- a/sys/arm64/include/vmm.h +++ b/sys/arm64/include/vmm.h @@ -143,6 +143,37 @@ struct vm_eventinfo { int *iptr; /* reqidle cookie */ }; +#define DECLARE_VMMOPS_FUNC(ret_type, opname, args) \ + ret_type vmmops_##opname args + +DECLARE_VMMOPS_FUNC(int, modinit, (int ipinum)); +DECLARE_VMMOPS_FUNC(int, modcleanup, (void)); +DECLARE_VMMOPS_FUNC(void *, init, (struct vm *vm, struct pmap *pmap)); +DECLARE_VMMOPS_FUNC(int, gla2gpa, (void *vcpui, struct vm_guest_paging *paging, + uint64_t gla, int prot, uint64_t *gpa, int *is_fault)); +DECLARE_VMMOPS_FUNC(int, run, (void *vcpui, register_t pc, struct pmap *pmap, + struct vm_eventinfo *info)); +DECLARE_VMMOPS_FUNC(void, cleanup, (void *vmi)); +DECLARE_VMMOPS_FUNC(void *, vcpu_init, (void *vmi, struct vcpu *vcpu, + int vcpu_id)); +DECLARE_VMMOPS_FUNC(void, vcpu_cleanup, (void *vcpui)); +DECLARE_VMMOPS_FUNC(int, exception, (void *vcpui, uint64_t esr, uint64_t far)); +DECLARE_VMMOPS_FUNC(int, getreg, (void *vcpui, int num, uint64_t *retval)); +DECLARE_VMMOPS_FUNC(int, setreg, (void *vcpui, int num, uint64_t val)); +DECLARE_VMMOPS_FUNC(int, getcap, (void *vcpui, int num, int *retval)); +DECLARE_VMMOPS_FUNC(int, setcap, (void *vcpui, int num, int val)); +DECLARE_VMMOPS_FUNC(struct vmspace *, vmspace_alloc, (vm_offset_t min, + vm_offset_t max)); +DECLARE_VMMOPS_FUNC(void, vmspace_free, (struct vmspace *vmspace)); +#ifdef notyet +#ifdef BHYVE_SNAPSHOT +DECLARE_VMMOPS_FUNC(int, snapshot, (void *vmi, struct vm_snapshot_meta *meta)); +DECLARE_VMMOPS_FUNC(int, vcpu_snapshot, (void *vcpui, + struct vm_snapshot_meta *meta)); +DECLARE_VMMOPS_FUNC(int, restore_tsc, (void *vcpui, uint64_t now)); +#endif +#endif + int vm_create(const char *name, struct vm **retvm); struct vcpu *vm_alloc_vcpu(struct vm *vm, int vcpuid); void vm_disable_vcpu_creation(struct vm *vm); @@ -232,7 +263,6 @@ vcpu_should_yield(struct vcpu *vcpu) void *vcpu_stats(struct vcpu *vcpu); void vcpu_notify_event(struct vcpu *vcpu); -struct vmspace *vm_vmspace(struct vm *vm); struct vm_mem *vm_mem(struct vm *vm); enum vm_reg_name vm_segment_name(int seg_encoding); diff --git a/sys/arm64/vmm/arm64.h b/sys/arm64/vmm/arm64.h index f9b74aef7188..f530dab05331 100644 --- a/sys/arm64/vmm/arm64.h +++ b/sys/arm64/vmm/arm64.h @@ -136,37 +136,6 @@ struct hyp { struct hypctx *ctx[]; }; -#define DEFINE_VMMOPS_IFUNC(ret_type, opname, args) \ - ret_type vmmops_##opname args; - -DEFINE_VMMOPS_IFUNC(int, modinit, (int ipinum)) -DEFINE_VMMOPS_IFUNC(int, modcleanup, (void)) -DEFINE_VMMOPS_IFUNC(void *, init, (struct vm *vm, struct pmap *pmap)) -DEFINE_VMMOPS_IFUNC(int, gla2gpa, (void *vcpui, struct vm_guest_paging *paging, - uint64_t gla, int prot, uint64_t *gpa, int *is_fault)) -DEFINE_VMMOPS_IFUNC(int, run, (void *vcpui, register_t pc, struct pmap *pmap, - struct vm_eventinfo *info)) -DEFINE_VMMOPS_IFUNC(void, cleanup, (void *vmi)) -DEFINE_VMMOPS_IFUNC(void *, vcpu_init, (void *vmi, struct vcpu *vcpu, - int vcpu_id)) -DEFINE_VMMOPS_IFUNC(void, vcpu_cleanup, (void *vcpui)) -DEFINE_VMMOPS_IFUNC(int, exception, (void *vcpui, uint64_t esr, uint64_t far)) -DEFINE_VMMOPS_IFUNC(int, getreg, (void *vcpui, int num, uint64_t *retval)) -DEFINE_VMMOPS_IFUNC(int, setreg, (void *vcpui, int num, uint64_t val)) -DEFINE_VMMOPS_IFUNC(int, getcap, (void *vcpui, int num, int *retval)) -DEFINE_VMMOPS_IFUNC(int, setcap, (void *vcpui, int num, int val)) -DEFINE_VMMOPS_IFUNC(struct vmspace *, vmspace_alloc, (vm_offset_t min, - vm_offset_t max)) -DEFINE_VMMOPS_IFUNC(void, vmspace_free, (struct vmspace *vmspace)) -#ifdef notyet -#ifdef BHYVE_SNAPSHOT -DEFINE_VMMOPS_IFUNC(int, snapshot, (void *vmi, struct vm_snapshot_meta *meta)) -DEFINE_VMMOPS_IFUNC(int, vcpu_snapshot, (void *vcpui, - struct vm_snapshot_meta *meta)) -DEFINE_VMMOPS_IFUNC(int, restore_tsc, (void *vcpui, uint64_t now)) -#endif -#endif - uint64_t vmm_call_hyp(uint64_t, ...); #if 0 diff --git a/sys/arm64/vmm/vmm.c b/sys/arm64/vmm/vmm.c index a551a2807183..aeda689f3b1a 100644 --- a/sys/arm64/vmm/vmm.c +++ b/sys/arm64/vmm/vmm.c @@ -88,7 +88,6 @@ struct vcpu { struct vfpstate *guestfpu; /* (a,i) guest fpu state */ }; -#define vcpu_lock_initialized(v) mtx_initialized(&((v)->mtx)) #define vcpu_lock_init(v) mtx_init(&((v)->mtx), "vcpu lock", 0, MTX_SPIN) #define vcpu_lock_destroy(v) mtx_destroy(&((v)->mtx)) #define vcpu_lock(v) mtx_lock_spin(&((v)->mtx)) @@ -126,7 +125,6 @@ struct vm { bool dying; /* (o) is dying */ volatile cpuset_t suspended_cpus; /* (i) suspended vcpus */ volatile cpuset_t halted_cpus; /* (x) cpus in a hard halt */ - struct vmspace *vmspace; /* (o) guest's address space */ struct vm_mem mem; /* (i) guest memory */ char name[VM_MAX_NAMELEN]; /* (o) virtual machine name */ struct vcpu **vcpu; /* (i) guest vcpus */ @@ -274,6 +272,7 @@ vcpu_cleanup(struct vcpu *vcpu, bool destroy) vmm_stat_free(vcpu->stats); fpu_save_area_free(vcpu->guestfpu); vcpu_lock_destroy(vcpu); + free(vcpu, M_VMM); } } @@ -407,7 +406,7 @@ vm_init(struct vm *vm, bool create) { int i; - vm->cookie = vmmops_init(vm, vmspace_pmap(vm->vmspace)); + vm->cookie = vmmops_init(vm, vmspace_pmap(vm_vmspace(vm))); MPASS(vm->cookie != NULL); CPU_ZERO(&vm->active_cpus); @@ -485,7 +484,7 @@ int vm_create(const char *name, struct vm **retvm) { struct vm *vm; - struct vmspace *vmspace; + int error; /* * If vmm.ko could not be successfully initialized then don't attempt @@ -497,14 +496,13 @@ vm_create(const char *name, struct vm **retvm) if (name == NULL || strlen(name) >= VM_MAX_NAMELEN) return (EINVAL); - vmspace = vmmops_vmspace_alloc(0, 1ul << 39); - if (vmspace == NULL) - return (ENOMEM); - vm = malloc(sizeof(struct vm), M_VMM, M_WAITOK | M_ZERO); + error = vm_mem_init(&vm->mem, 0, 1ul << 39); + if (error != 0) { + free(vm, M_VMM); + return (error); + } strcpy(vm->name, name); - vm->vmspace = vmspace; - vm_mem_init(&vm->mem); sx_init(&vm->vcpus_init_lock, "vm vcpus"); vm->sockets = 1; @@ -558,7 +556,7 @@ vm_cleanup(struct vm *vm, bool destroy) if (destroy) { vm_xlock_memsegs(vm); - pmap = vmspace_pmap(vm->vmspace); + pmap = vmspace_pmap(vm_vmspace(vm)); sched_pin(); PCPU_SET(curvmpmap, NULL); sched_unpin(); @@ -582,11 +580,6 @@ vm_cleanup(struct vm *vm, bool destroy) if (destroy) { vm_mem_destroy(vm); - vmmops_vmspace_free(vm->vmspace); - vm->vmspace = NULL; - - for (i = 0; i < vm->maxcpus; i++) - free(vm->vcpu[i], M_VMM); free(vm->vcpu, M_VMM); sx_destroy(&vm->vcpus_init_lock); } @@ -1090,12 +1083,6 @@ vcpu_notify_event(struct vcpu *vcpu) vcpu_unlock(vcpu); } -struct vmspace * -vm_vmspace(struct vm *vm) -{ - return (vm->vmspace); -} - struct vm_mem * vm_mem(struct vm *vm) { @@ -1416,7 +1403,7 @@ vm_handle_paging(struct vcpu *vcpu, bool *retu) vme = &vcpu->exitinfo; - pmap = vmspace_pmap(vcpu->vm->vmspace); + pmap = vmspace_pmap(vm_vmspace(vcpu->vm)); addr = vme->u.paging.gpa; esr = vme->u.paging.esr; @@ -1433,7 +1420,7 @@ vm_handle_paging(struct vcpu *vcpu, bool *retu) panic("%s: Invalid exception (esr = %lx)", __func__, esr); } - map = &vm->vmspace->vm_map; + map = &vm_vmspace(vm)->vm_map; rv = vm_fault(map, vme->u.paging.gpa, ftype, VM_FAULT_NORMAL, NULL); if (rv != KERN_SUCCESS) return (EFAULT); @@ -1507,7 +1494,7 @@ vm_run(struct vcpu *vcpu) if (CPU_ISSET(vcpuid, &vm->suspended_cpus)) return (EINVAL); - pmap = vmspace_pmap(vm->vmspace); + pmap = vmspace_pmap(vm_vmspace(vm)); vme = &vcpu->exitinfo; evinfo.rptr = NULL; evinfo.sptr = &vm->suspend; diff --git a/sys/dev/ice/ice_common.c b/sys/dev/ice/ice_common.c index ad4ea4c8e7a1..b895f661bc46 100644 --- a/sys/dev/ice/ice_common.c +++ b/sys/dev/ice/ice_common.c @@ -213,6 +213,15 @@ int ice_set_mac_type(struct ice_hw *hw) case ICE_DEV_ID_E830_L_QSFP: case ICE_DEV_ID_E830C_SFP: case ICE_DEV_ID_E830_L_SFP: + case ICE_DEV_ID_E835CC_BACKPLANE: + case ICE_DEV_ID_E835CC_QSFP56: + case ICE_DEV_ID_E835CC_SFP: + case ICE_DEV_ID_E835C_BACKPLANE: + case ICE_DEV_ID_E835C_QSFP: + case ICE_DEV_ID_E835C_SFP: + case ICE_DEV_ID_E835_L_BACKPLANE: + case ICE_DEV_ID_E835_L_QSFP: + case ICE_DEV_ID_E835_L_SFP: hw->mac_type = ICE_MAC_E830; break; default: diff --git a/sys/dev/ice/ice_devids.h b/sys/dev/ice/ice_devids.h index 3f91e9dfbcaf..74712c61ae8e 100644 --- a/sys/dev/ice/ice_devids.h +++ b/sys/dev/ice/ice_devids.h @@ -62,6 +62,24 @@ #define ICE_DEV_ID_E830C_SFP 0x12DA /* Intel(R) Ethernet Controller E830-L for SFP */ #define ICE_DEV_ID_E830_L_SFP 0x12DE +/* Intel(R) Ethernet Controller E835-CC for backplane */ +#define ICE_DEV_ID_E835CC_BACKPLANE 0x1248 +/* Intel(R) Ethernet Controller E835-CC for QSFP */ +#define ICE_DEV_ID_E835CC_QSFP56 0x1249 +/* Intel(R) Ethernet Controller E835-CC for SFP */ +#define ICE_DEV_ID_E835CC_SFP 0x124A +/* Intel(R) Ethernet Controller E835-C for backplane */ +#define ICE_DEV_ID_E835C_BACKPLANE 0x1261 +/* Intel(R) Ethernet Controller E835-C for QSFP */ +#define ICE_DEV_ID_E835C_QSFP 0x1262 +/* Intel(R) Ethernet Controller E835-C for SFP */ +#define ICE_DEV_ID_E835C_SFP 0x1263 +/* Intel(R) Ethernet Controller E835-L for backplane */ +#define ICE_DEV_ID_E835_L_BACKPLANE 0x1265 +/* Intel(R) Ethernet Controller E835-L for QSFP */ +#define ICE_DEV_ID_E835_L_QSFP 0x1266 +/* Intel(R) Ethernet Controller E835-L for SFP */ +#define ICE_DEV_ID_E835_L_SFP 0x1267 /* Intel(R) Ethernet Controller E810-C for backplane */ #define ICE_DEV_ID_E810C_BACKPLANE 0x1591 /* Intel(R) Ethernet Controller E810-C for QSFP */ diff --git a/sys/dev/ice/ice_drv_info.h b/sys/dev/ice/ice_drv_info.h index 2a51a7394424..46965f4124bc 100644 --- a/sys/dev/ice/ice_drv_info.h +++ b/sys/dev/ice/ice_drv_info.h @@ -218,6 +218,45 @@ static const pci_vendor_info_t ice_vendor_info_array[] = { "Intel(R) Ethernet Network Adapter E830-XXV-2"), PVIDV(ICE_INTEL_VENDOR_ID, ICE_DEV_ID_E830_L_SFP, "Intel(R) Ethernet Connection E830-L for SFP"), + PVIDV(ICE_INTEL_VENDOR_ID, ICE_DEV_ID_E835CC_BACKPLANE, + "Intel(R) Ethernet Connection E835-CC for backplane"), + PVIDV_OEM(ICE_INTEL_VENDOR_ID, ICE_DEV_ID_E835CC_QSFP56, + ICE_INTEL_VENDOR_ID, 0x0001, 0, + "Intel(R) Ethernet Network Adapter E835-C-Q2"), + PVIDV_OEM(ICE_INTEL_VENDOR_ID, ICE_DEV_ID_E835CC_QSFP56, + ICE_INTEL_VENDOR_ID, 0x0002, 0, + "Intel(R) Ethernet Network Adapter E835-C-Q2 for OCP 3.0"), + PVIDV_OEM(ICE_INTEL_VENDOR_ID, ICE_DEV_ID_E835CC_QSFP56, + ICE_INTEL_VENDOR_ID, 0x0003, 0, + "Intel(R) Ethernet Network Adapter E835-CC-Q1"), + PVIDV_OEM(ICE_INTEL_VENDOR_ID, ICE_DEV_ID_E835CC_QSFP56, + ICE_INTEL_VENDOR_ID, 0x0004, 0, + "Intel(R) Ethernet Network Adapter E835-CC-Q1 for OCP 3.0"), + PVIDV(ICE_INTEL_VENDOR_ID, ICE_DEV_ID_E835CC_QSFP56, + "Intel(R) Ethernet Connection E835-CC for QSFP56"), + PVIDV_OEM(ICE_INTEL_VENDOR_ID, ICE_DEV_ID_E835CC_SFP, + ICE_INTEL_VENDOR_ID, 0x0001, 0, + "Intel(R) Ethernet Network Adapter E835-XXV-2 for OCP 3.0"), + PVIDV_OEM(ICE_INTEL_VENDOR_ID, ICE_DEV_ID_E835CC_SFP, + ICE_INTEL_VENDOR_ID, 0x0003, 0, + "Intel(R) Ethernet Network Adapter E835-XXV-2"), + PVIDV_OEM(ICE_INTEL_VENDOR_ID, ICE_DEV_ID_E835CC_SFP, + ICE_INTEL_VENDOR_ID, 0x0004, 0, + "Intel(R) Ethernet Network Adapter E835-XXV-4 for OCP 3.0"), + PVIDV(ICE_INTEL_VENDOR_ID, ICE_DEV_ID_E835CC_SFP, + "Intel(R) Ethernet Connection E835-CC for SFP"), + PVIDV(ICE_INTEL_VENDOR_ID, ICE_DEV_ID_E835C_BACKPLANE, + "Intel(R) Ethernet Connection E835-C for backplane"), + PVIDV(ICE_INTEL_VENDOR_ID, ICE_DEV_ID_E835C_QSFP, + "Intel(R) Ethernet Connection E835-C for QSFP"), + PVIDV(ICE_INTEL_VENDOR_ID, ICE_DEV_ID_E835C_SFP, + "Intel(R) Ethernet Connection E835-C for SFP"), + PVIDV(ICE_INTEL_VENDOR_ID, ICE_DEV_ID_E835_L_BACKPLANE, + "Intel(R) Ethernet Connection E835-L for backplane"), + PVIDV(ICE_INTEL_VENDOR_ID, ICE_DEV_ID_E835_L_QSFP, + "Intel(R) Ethernet Connection E835-L for QSFP"), + PVIDV(ICE_INTEL_VENDOR_ID, ICE_DEV_ID_E835_L_SFP, + "Intel(R) Ethernet Connection E835-L for SFP"), PVIDV(ICE_INTEL_VENDOR_ID, ICE_DEV_ID_E825C_BACKPLANE, "Intel(R) Ethernet Connection E825-C for backplane"), PVIDV(ICE_INTEL_VENDOR_ID, ICE_DEV_ID_E825C_QSFP, diff --git a/sys/dev/ixgbe/if_ix.c b/sys/dev/ixgbe/if_ix.c index 6d08bd49bc04..1d36fd11f368 100644 --- a/sys/dev/ixgbe/if_ix.c +++ b/sys/dev/ixgbe/if_ix.c @@ -192,6 +192,8 @@ static int ixgbe_if_i2c_req(if_ctx_t, struct ifi2creq *); static bool ixgbe_if_needs_restart(if_ctx_t, enum iflib_restart_event); int ixgbe_intr(void *); +static int ixgbe_if_priv_ioctl(if_ctx_t ctx, u_long command, caddr_t data); + /************************************************************************ * Function prototypes ************************************************************************/ @@ -239,6 +241,13 @@ static void ixgbe_setup_vlan_hw_support(if_ctx_t); static void ixgbe_config_gpie(struct ixgbe_softc *); static void ixgbe_config_delay_values(struct ixgbe_softc *); +static void ixgbe_add_debug_sysctls(struct ixgbe_softc *sc); +static void ixgbe_add_debug_dump_sysctls(struct ixgbe_softc *sc); +static int ixgbe_debug_dump_ioctl(struct ixgbe_softc *sc, struct ifdrv *ifd); +static u8 ixgbe_debug_dump_print_cluster(struct ixgbe_softc *sc, + struct sbuf *sbuf, u8 cluster_id); +static int ixgbe_nvm_access_ioctl(struct ixgbe_softc *sc, struct ifdrv *ifd); + /* Sysctl handlers */ static int ixgbe_sysctl_flowcntl(SYSCTL_HANDLER_ARGS); static int ixgbe_sysctl_advertise(SYSCTL_HANDLER_ARGS); @@ -260,6 +269,9 @@ static int ixgbe_sysctl_wol_enable(SYSCTL_HANDLER_ARGS); static int ixgbe_sysctl_wufc(SYSCTL_HANDLER_ARGS); static int ixgbe_sysctl_tso_tcp_flags_mask(SYSCTL_HANDLER_ARGS); +static int ixgbe_sysctl_debug_dump_set_clusters(SYSCTL_HANDLER_ARGS); +static int ixgbe_sysctl_dump_debug_dump(SYSCTL_HANDLER_ARGS); + /* Deferred interrupt tasklets */ static void ixgbe_handle_msf(void *); static void ixgbe_handle_mod(void *); @@ -330,6 +342,7 @@ static device_method_t ixgbe_if_methods[] = { DEVMETHOD(ifdi_get_counter, ixgbe_if_get_counter), DEVMETHOD(ifdi_i2c_req, ixgbe_if_i2c_req), DEVMETHOD(ifdi_needs_restart, ixgbe_if_needs_restart), + DEVMETHOD(ifdi_priv_ioctl, ixgbe_if_priv_ioctl), #ifdef PCI_IOV DEVMETHOD(ifdi_iov_init, ixgbe_if_iov_init), DEVMETHOD(ifdi_iov_uninit, ixgbe_if_iov_uninit), @@ -1015,6 +1028,8 @@ ixgbe_if_attach_pre(if_ctx_t ctx) if (hw->mac.type == ixgbe_mac_E610) ixgbe_init_aci(hw); + sc->do_debug_dump = false; + if (hw->mac.ops.fw_recovery_mode && hw->mac.ops.fw_recovery_mode(hw)) { device_printf(dev, @@ -1395,6 +1410,248 @@ ixgbe_if_needs_restart(if_ctx_t ctx __unused, enum iflib_restart_event event) } /************************************************************************ + * ixgbe_if_priv_ioctl - Ioctl handler for driver + * + * Handler for custom driver specific ioctls + * + * return 0 on success, positive on failure + ************************************************************************/ +static int +ixgbe_if_priv_ioctl(if_ctx_t ctx, u_long command, caddr_t data) +{ + struct ixgbe_softc *sc = iflib_get_softc(ctx); + struct ifdrv *ifd; + device_t dev = sc->dev; + + /* Make sure the command type is valid */ + switch (command) { + case SIOCSDRVSPEC: + case SIOCGDRVSPEC: + /* Accepted commands */ + break; + case SIOCGPRIVATE_0: + /* + * Although we do not support this ioctl command, it's expected + * that iflib will forward it to the IFDI_PRIV_IOCTL handler. + * Do not print a message in this case. + */ + return (ENOTSUP); + default: + /* + * If we get a different command for this function, it's + * definitely unexpected, so log a message indicating what + * command we got for debugging purposes. + */ + device_printf(dev, + "%s: unexpected ioctl command %08lx\n", + __func__, command); + return (EINVAL); + } + + ifd = (struct ifdrv *)data; + + switch (ifd->ifd_cmd) { + case IXGBE_NVM_ACCESS: + IOCTL_DEBUGOUT("ioctl: NVM ACCESS"); + return (ixgbe_nvm_access_ioctl(sc, ifd)); + case IXGBE_DEBUG_DUMP: + IOCTL_DEBUGOUT("ioctl: DEBUG DUMP"); + return (ixgbe_debug_dump_ioctl(sc, ifd)); + default: + IOCTL_DEBUGOUT1( + "ioctl: UNKNOWN SIOC(S|G)DRVSPEC (0x%X) command\n", + (int)ifd->ifd_cmd); + return (EINVAL); + } + + return (0); +} + +/************************************************************************ + * ixgbe_nvm_access_ioctl + * + * Handles an NVM access ioctl request + ************************************************************************/ +static int +ixgbe_nvm_access_ioctl(struct ixgbe_softc *sc, struct ifdrv *ifd) +{ + struct ixgbe_nvm_access_data *data; + struct ixgbe_nvm_access_cmd *cmd; + struct ixgbe_hw *hw = &sc->hw; + size_t ifd_len = ifd->ifd_len; + size_t malloc_len; + device_t dev = sc->dev; + u8 *nvm_buffer; + s32 error = 0; + + /* + * ifioctl forwards SIOCxDRVSPEC to iflib without conducting + * a privilege check. Subsequently, iflib passes the ioctl to the driver + * without verifying privileges. To prevent non-privileged threads from + * accessing this interface, perform a privilege check at this point. + */ + error = priv_check(curthread, PRIV_DRIVER); + if (error) + return (error); + + if (ifd_len < sizeof(*cmd)) { + device_printf(dev, + "%s: ifdrv length is too small. Got %zu, " + "but expected %zu\n", + __func__, ifd_len, sizeof(*cmd)); + return (EINVAL); + } + + if (ifd->ifd_data == NULL) { + device_printf(dev, "%s: No ifd data buffer.\n", + __func__); + return (EINVAL); + } + + malloc_len = max(ifd_len, sizeof(*data) + sizeof(*cmd)); + + nvm_buffer = (u8 *)malloc(malloc_len, M_IXGBE, M_ZERO | M_NOWAIT); + if (!nvm_buffer) + return (ENOMEM); + + /* Copy the NVM access command and data in from user space */ + error = copyin(ifd->ifd_data, nvm_buffer, ifd_len); + if (error) { + device_printf(dev, "%s: Failed to copy data in, error: %d\n", + __func__, error); + goto cleanup_free_nvm_buffer; + } + + /* + * The NVM command structure is immediately followed by data which + * varies in size based on the command. + */ + cmd = (struct ixgbe_nvm_access_cmd *)nvm_buffer; + data = (struct ixgbe_nvm_access_data *) + (nvm_buffer + sizeof(struct ixgbe_nvm_access_cmd)); + + /* Handle the NVM access request */ + error = ixgbe_handle_nvm_access(hw, cmd, data); + if (error) { + device_printf(dev, "%s: NVM access request failed, error %d\n", + __func__, error); + } + + /* Copy the possibly modified contents of the handled request out */ + error = copyout(nvm_buffer, ifd->ifd_data, ifd_len); + if (error) { + device_printf(dev, "%s: Copying response back to " + "user space failed, error %d\n", + __func__, error); + goto cleanup_free_nvm_buffer; + } + +cleanup_free_nvm_buffer: + free(nvm_buffer, M_IXGBE); + return (error); +} + +/************************************************************************ + * ixgbe_debug_dump_ioctl + * + * Makes debug dump of internal FW/HW data. + ************************************************************************/ +static int +ixgbe_debug_dump_ioctl(struct ixgbe_softc *sc, struct ifdrv *ifd) +{ + struct ixgbe_debug_dump_cmd *dd_cmd; + struct ixgbe_hw *hw = &sc->hw; + size_t ifd_len = ifd->ifd_len; + device_t dev = sc->dev; + s32 error = 0; + + if (!(sc->feat_en & IXGBE_FEATURE_DBG_DUMP)) + return (ENODEV); + + /* Data returned from ACI command */ + u16 ret_buf_size = 0; + u16 ret_next_cluster = 0; + u16 ret_next_table = 0; + u32 ret_next_index = 0; + + /* + * ifioctl forwards SIOCxDRVSPEC to iflib without conducting + * a privilege check. Subsequently, iflib passes the ioctl to the driver + * without verifying privileges. To prevent non-privileged threads from + * accessing this interface, perform a privilege check at this point. + */ + error = priv_check(curthread, PRIV_DRIVER); + if (error) + return (error); + + if (ifd_len < sizeof(*dd_cmd)) { + device_printf(dev, + "%s: ifdrv length is too small. Got %zu, " + "but expected %zu\n", + __func__, ifd_len, sizeof(*dd_cmd)); + return (EINVAL); + } + + if (ifd->ifd_data == NULL) { + device_printf(dev, "%s: No ifd data buffer.\n", + __func__); + return (EINVAL); + } + + dd_cmd = (struct ixgbe_debug_dump_cmd *)malloc(ifd_len, M_IXGBE, + M_NOWAIT | M_ZERO); + if (!dd_cmd) { + error = -ENOMEM; + goto out; + } + /* copy data from userspace */ + error = copyin(ifd->ifd_data, dd_cmd, ifd_len); + if (error) { + device_printf(dev, "%s: Failed to copy data in, error: %d\n", + __func__, error); + goto out; + } + + /* ACI command requires buf_size arg to be grater than 0 */ + if (dd_cmd->data_size == 0) { + device_printf(dev, "%s: data_size must be greater than 0\n", + __func__); + error = EINVAL; + goto out; + } + + /* Zero the data buffer memory space */ + memset(dd_cmd->data, 0, ifd_len - sizeof(*dd_cmd)); + + error = ixgbe_aci_get_internal_data(hw, dd_cmd->cluster_id, + dd_cmd->table_id, dd_cmd->offset, dd_cmd->data, dd_cmd->data_size, + &ret_buf_size, &ret_next_cluster, &ret_next_table, &ret_next_index); + if (error) { + device_printf(dev, + "%s: Failed to get internal FW/HW data, error: %d\n", + __func__, error); + goto out; + } + + dd_cmd->cluster_id = ret_next_cluster; + dd_cmd->table_id = ret_next_table; + dd_cmd->offset = ret_next_index; + dd_cmd->data_size = ret_buf_size; + + error = copyout(dd_cmd, ifd->ifd_data, ifd->ifd_len); + if (error) { + device_printf(dev, + "%s: Failed to copy data out, error: %d\n", + __func__, error); + } + +out: + free(dd_cmd, M_IXGBE); + + return (error); +} + +/************************************************************************ * ixgbe_add_media_types ************************************************************************/ static void @@ -2883,6 +3140,264 @@ ixgbe_sysctl_interrupt_rate_handler(SYSCTL_HANDLER_ARGS) } /* ixgbe_sysctl_interrupt_rate_handler */ /************************************************************************ + * ixgbe_debug_dump_print_cluster + ************************************************************************/ +static u8 +ixgbe_debug_dump_print_cluster(struct ixgbe_softc *sc, struct sbuf *sbuf, + u8 cluster_id) +{ + u16 data_buf_size = IXGBE_ACI_MAX_BUFFER_SIZE; + device_t dev = sc->dev; + struct ixgbe_hw *hw = &sc->hw; + const u8 reserved_buf[8] = {}; + int max_aci_calls = 1000; + int error, counter = 0; + u8 *data_buf; + + /* Input parameters / loop variables */ + u16 table_id = 0; + u32 offset = 0; + + /* Data returned from ACI command */ + u16 ret_buf_size = 0; + u16 ret_next_cluster = 0; + u16 ret_next_table = 0; + u32 ret_next_index = 0; + + data_buf = (u8 *)malloc(data_buf_size, M_IXGBE, M_NOWAIT | M_ZERO); + if (!data_buf) + return (0); + + DEBUGOUT2("%s: dumping cluster id (relative) %d\n", + __func__, cluster_id); + + do { + DEBUGOUT3("table_id 0x%04x offset 0x%08x buf_size %d\n", + table_id, offset, data_buf_size); + + error = ixgbe_aci_get_internal_data(hw, cluster_id, table_id, + offset, data_buf, data_buf_size, &ret_buf_size, + &ret_next_cluster, &ret_next_table, &ret_next_index); + if (error) { + device_printf(dev, + "%s: Failed to get internal FW/HW data, error: %d, " + "last aci status: %d\n", + __func__, error, hw->aci.last_status); + break; + } + + DEBUGOUT3("ret_table_id 0x%04x ret_offset 0x%08x " + "ret_buf_size %d\n", + ret_next_table, ret_next_index, ret_buf_size); + + /* Print cluster id */ + u32 print_cluster_id = (u32)cluster_id; + sbuf_bcat(sbuf, &print_cluster_id, sizeof(print_cluster_id)); + /* Print table id */ + u32 print_table_id = (u32)table_id; + sbuf_bcat(sbuf, &print_table_id, sizeof(print_table_id)); + /* Print table length */ + u32 print_table_length = (u32)ret_buf_size; + sbuf_bcat(sbuf, &print_table_length, + sizeof(print_table_length)); + /* Print current offset */ + u32 print_curr_offset = offset; + sbuf_bcat(sbuf, &print_curr_offset, sizeof(print_curr_offset)); + /* Print reserved bytes */ + sbuf_bcat(sbuf, reserved_buf, sizeof(reserved_buf)); + /* Print data */ + sbuf_bcat(sbuf, data_buf, ret_buf_size); + + /* Prepare for the next loop spin */ + memset(data_buf, 0, data_buf_size); + + bool last_index = (ret_next_index == 0xffffffff); + bool last_table = ((ret_next_table == 0xff || + ret_next_table == 0xffff) && + last_index); + + if (last_table) { + /* End of the cluster */ + DEBUGOUT1("End of the cluster ID %d\n", cluster_id); + break; + } else if (last_index) { + /* End of the table */ + table_id = ret_next_table; + offset = 0; + } else { + /* More data left in the table */ + offset = ret_next_index; + } + } while (++counter < max_aci_calls); + + if (counter >= max_aci_calls) + device_printf(dev, "Exceeded nr of ACI calls for cluster %d\n", + cluster_id); + + free(data_buf, M_IXGBE); + + return (++cluster_id); +} /* ixgbe_print_debug_dump_cluster */ + +/************************************************************************ + * ixgbe_sysctl_debug_dump_set_clusters + * + * Sets the cluster to dump from FW when Debug Dump requested. + ************************************************************************/ +static int +ixgbe_sysctl_debug_dump_set_clusters(SYSCTL_HANDLER_ARGS) +{ + struct ixgbe_softc *sc = (struct ixgbe_softc *)arg1; + u32 clusters = sc->debug_dump_cluster_mask; + device_t dev = sc->dev; + int error; + + error = sysctl_handle_32(oidp, &clusters, 0, req); + if ((error) || !req->newptr) + return (error); + + if (clusters & ~(IXGBE_DBG_DUMP_VALID_CLUSTERS_MASK)) { + device_printf(dev, + "%s: Unrecognized parameter: %u\n", + __func__, clusters); + sc->debug_dump_cluster_mask = + IXGBE_ACI_DBG_DUMP_CLUSTER_ID_INVALID; + return (EINVAL); + } + + sc->debug_dump_cluster_mask = clusters; + + return (0); +} /* ixgbe_sysctl_debug_dump_set_clusters */ + +/************************************************************************ + * ixgbe_sysctl_dump_debug_dump + ************************************************************************/ +static int +ixgbe_sysctl_dump_debug_dump(SYSCTL_HANDLER_ARGS) +{ + struct ixgbe_softc *sc = (struct ixgbe_softc *)arg1; + device_t dev = sc->dev; + struct sbuf *sbuf; + int error = 0; + + UNREFERENCED_PARAMETER(arg2); + + if (!sc->do_debug_dump) { + if (req->oldptr == NULL && req->newptr == NULL) { + error = SYSCTL_OUT(req, 0, 0); + return (error); + } + + char input_buf[2] = ""; + error = sysctl_handle_string(oidp, input_buf, + sizeof(input_buf), req); + if ((error) || (req->newptr == NULL)) + return (error); + + if (input_buf[0] == '1') { + if (sc->debug_dump_cluster_mask == + IXGBE_ACI_DBG_DUMP_CLUSTER_ID_INVALID) { + device_printf(dev, + "Debug Dump failed because an invalid " + "cluster was specified.\n"); + return (EINVAL); + } + + sc->do_debug_dump = true; + return (0); + } + + return (EINVAL); + } + + /* Caller just wants the upper bound for size */ + if (req->oldptr == NULL && req->newptr == NULL) { + size_t est_output_len = IXGBE_DBG_DUMP_BASE_SIZE; + if (sc->debug_dump_cluster_mask & 0x2) + est_output_len += IXGBE_DBG_DUMP_BASE_SIZE; + error = SYSCTL_OUT(req, 0, est_output_len); + return (error); + } + + sbuf = sbuf_new_for_sysctl(NULL, NULL, 128, req); + sbuf_clear_flags(sbuf, SBUF_INCLUDENUL); + + DEBUGOUT("FW Debug Dump running...\n"); + + if (sc->debug_dump_cluster_mask) { + for (u8 id = 0; id <= IXGBE_ACI_DBG_DUMP_CLUSTER_ID_MAX; id++) { + if (sc->debug_dump_cluster_mask & BIT(id)) { + DEBUGOUT1("Dumping cluster ID %u...\n", id); + ixgbe_debug_dump_print_cluster(sc, sbuf, id); + } + } + } else { + u8 next_cluster_id = 0; + do { + DEBUGOUT1("Dumping cluster ID %u...\n", + next_cluster_id); + next_cluster_id = ixgbe_debug_dump_print_cluster(sc, + sbuf, next_cluster_id); + } while (next_cluster_id != 0 && + next_cluster_id <= IXGBE_ACI_DBG_DUMP_CLUSTER_ID_MAX); + } + + sbuf_finish(sbuf); + sbuf_delete(sbuf); + + sc->do_debug_dump = false; + + return (error); +} /* ixgbe_sysctl_dump_debug_dump */ + +/************************************************************************ + * ixgbe_add_debug_dump_sysctls + ************************************************************************/ +static void +ixgbe_add_debug_dump_sysctls(struct ixgbe_softc *sc) +{ + struct sysctl_oid_list *debug_list, *dump_list; + struct sysctl_oid *dump_node; + struct sysctl_ctx_list *ctx; + device_t dev = sc->dev; + + ctx = device_get_sysctl_ctx(dev); + debug_list = SYSCTL_CHILDREN(sc->debug_sysctls); + + dump_node = SYSCTL_ADD_NODE(ctx, debug_list, OID_AUTO, "dump", + CTLFLAG_RD, NULL, "Internal FW/HW Dump"); + dump_list = SYSCTL_CHILDREN(dump_node); + + SYSCTL_ADD_PROC(ctx, dump_list, OID_AUTO, "clusters", + CTLTYPE_U32 | CTLFLAG_RW, sc, 0, + ixgbe_sysctl_debug_dump_set_clusters, "SU", + IXGBE_SYSCTL_DESC_DEBUG_DUMP_SET_CLUSTER); + + SYSCTL_ADD_PROC(ctx, dump_list, OID_AUTO, "dump", + CTLTYPE_STRING | CTLFLAG_RW | CTLFLAG_MPSAFE, sc, 0, + ixgbe_sysctl_dump_debug_dump, "", + IXGBE_SYSCTL_DESC_DUMP_DEBUG_DUMP); +} /* ixgbe_add_debug_dump_sysctls */ + +static void +ixgbe_add_debug_sysctls(struct ixgbe_softc *sc) +{ + struct sysctl_oid_list *ctx_list; + struct sysctl_ctx_list *ctx; + device_t dev = sc->dev; + + ctx = device_get_sysctl_ctx(dev); + ctx_list = SYSCTL_CHILDREN(device_get_sysctl_tree(dev)); + + sc->debug_sysctls = SYSCTL_ADD_NODE(ctx, ctx_list, OID_AUTO, "debug", + CTLFLAG_RD, NULL, "Debug Sysctls"); + + if (sc->feat_en & IXGBE_FEATURE_DBG_DUMP) + ixgbe_add_debug_dump_sysctls(sc); +} /* ixgbe_add_debug_sysctls */ + +/************************************************************************ * ixgbe_add_device_sysctls ************************************************************************/ static void @@ -2992,6 +3507,8 @@ ixgbe_add_device_sysctls(if_ctx_t ctx) CTLTYPE_INT | CTLFLAG_RW, sc, 0, ixgbe_sysctl_eee_state, "I", "EEE Power Save State"); } + + ixgbe_add_debug_sysctls(sc); } /* ixgbe_add_device_sysctls */ /************************************************************************ @@ -5182,6 +5699,7 @@ ixgbe_init_device_features(struct ixgbe_softc *sc) break; case ixgbe_mac_E610: sc->feat_cap |= IXGBE_FEATURE_RECOVERY_MODE; + sc->feat_cap |= IXGBE_FEATURE_DBG_DUMP; break; default: break; @@ -5203,6 +5721,9 @@ ixgbe_init_device_features(struct ixgbe_softc *sc) /* Recovery mode */ if (sc->feat_cap & IXGBE_FEATURE_RECOVERY_MODE) sc->feat_en |= IXGBE_FEATURE_RECOVERY_MODE; + /* FW Debug Dump */ + if (sc->feat_cap & IXGBE_FEATURE_DBG_DUMP) + sc->feat_en |= IXGBE_FEATURE_DBG_DUMP; /* Enabled via global sysctl... */ /* Flow Director */ diff --git a/sys/dev/ixgbe/ixgbe.h b/sys/dev/ixgbe/ixgbe.h index 844064bf8543..624b71acabea 100644 --- a/sys/dev/ixgbe/ixgbe.h +++ b/sys/dev/ixgbe/ixgbe.h @@ -46,6 +46,7 @@ #include <sys/module.h> #include <sys/sockio.h> #include <sys/eventhandler.h> +#include <sys/priv.h> #include <net/if.h> #include <net/if_var.h> @@ -475,6 +476,20 @@ struct ixgbe_softc { u32 feat_cap; u32 feat_en; u16 lse_mask; + + struct sysctl_oid *debug_sysctls; + u32 debug_dump_cluster_mask; + bool do_debug_dump; +}; + +struct ixgbe_debug_dump_cmd { + u32 offset; /* offset to read/write from table, in bytes */ + u8 cluster_id; /* also used to get next cluster id */ + u16 table_id; + u16 data_size; /* size of data field, in bytes */ + u16 reserved1; + u32 reserved2; + u8 data[]; }; /* Precision Time Sync (IEEE 1588) defines */ @@ -499,6 +514,43 @@ struct ixgbe_softc { #define IXGBE_PHY_CURRENT_TEMP 0xC820 #define IXGBE_PHY_OVERTEMP_STATUS 0xC830 +/** + * The ioctl command number used by NVM update for accessing the driver for + * NVM access commands. + */ +#define IXGBE_NVM_ACCESS \ + (((((((('E' << 4) + '1') << 4) + 'K') << 4) + 'G') << 4) | 5) + +/* + * The ioctl command number used by a userspace tool for accessing the driver + * for getting debug dump data from the firmware. + */ +#define IXGBE_DEBUG_DUMP \ + (((((((('E' << 4) + '1') << 4) + 'K') << 4) + 'G') << 4) | 6) + +/* Debug Dump related definitions */ +#define IXGBE_ACI_DBG_DUMP_CLUSTER_ID_INVALID 0xFFFFFF +#define IXGBE_ACI_DBG_DUMP_CLUSTER_ID_BASE 50 +#define IXGBE_ACI_DBG_DUMP_CLUSTER_ID_MAX 1 + +#define IXGBE_DBG_DUMP_VALID_CLUSTERS_MASK 0x3 +#define IXGBE_DBG_DUMP_BASE_SIZE (2 * 1024 * 1024) + +#define IXGBE_SYSCTL_DESC_DEBUG_DUMP_SET_CLUSTER \ +"\nSelect clusters to dump with \"dump\" sysctl" \ +"\nFlags:" \ +"\n\t 0x1 - Link" \ +"\n\t 0x2 - Full CSR Space, excluding RCW registers" \ +"\n\t" \ +"\nUse \"sysctl -x\" to view flags properly." + +#define IXGBE_SYSCTL_DESC_DUMP_DEBUG_DUMP \ +"\nWrite 1 to output a FW debug dump containing the clusters " \ +"specified by the \"clusters\" sysctl" \ +"\nThe \"-b\" flag must be used in order to dump this data " \ +"as binary data because" \ +"\nthis data is opaque and not a string." + /* Sysctl help messages; displayed with sysctl -d */ #define IXGBE_SYSCTL_DESC_ADV_SPEED \ "\nControl advertised link speed using these flags:\n" \ diff --git a/sys/dev/ixgbe/ixgbe_features.h b/sys/dev/ixgbe/ixgbe_features.h index 0cef334a185f..bee9040319d8 100644 --- a/sys/dev/ixgbe/ixgbe_features.h +++ b/sys/dev/ixgbe/ixgbe_features.h @@ -57,6 +57,7 @@ #define IXGBE_FEATURE_LEGACY_IRQ (u32)(1 << 12) #define IXGBE_FEATURE_NEEDS_CTXD (u32)(1 << 13) #define IXGBE_FEATURE_RECOVERY_MODE (u32)(1 << 15) +#define IXGBE_FEATURE_DBG_DUMP (u32)(1 << 16) /* Check for OS support. Undefine features if not included in the OS */ #ifndef PCI_IOV diff --git a/sys/dev/nvme/nvme.h b/sys/dev/nvme/nvme.h index 17c5cdb4db87..57cb37907e65 100644 --- a/sys/dev/nvme/nvme.h +++ b/sys/dev/nvme/nvme.h @@ -1507,9 +1507,7 @@ struct nvme_namespace_data { uint8_t eui64[8]; /** lba format support */ - uint32_t lbaf[16]; - - uint8_t reserved7[192]; + uint32_t lbaf[64]; uint8_t vendor_specific[3712]; } __packed __aligned(4); @@ -2175,7 +2173,7 @@ void nvme_namespace_data_swapbytes(struct nvme_namespace_data *s __unused) s->anagrpid = le32toh(s->anagrpid); s->nvmsetid = le16toh(s->nvmsetid); s->endgid = le16toh(s->endgid); - for (i = 0; i < 16; i++) + for (i = 0; i < 64; i++) s->lbaf[i] = le32toh(s->lbaf[i]); #endif } diff --git a/sys/dev/vmm/vmm_mem.c b/sys/dev/vmm/vmm_mem.c index be59e37de33d..9df31c9ba133 100644 --- a/sys/dev/vmm/vmm_mem.c +++ b/sys/dev/vmm/vmm_mem.c @@ -26,10 +26,14 @@ static void vm_free_memmap(struct vm *vm, int ident); -void -vm_mem_init(struct vm_mem *mem) +int +vm_mem_init(struct vm_mem *mem, vm_offset_t lo, vm_offset_t hi) { + mem->mem_vmspace = vmmops_vmspace_alloc(lo, hi); + if (mem->mem_vmspace == NULL) + return (ENOMEM); sx_init(&mem->mem_segs_lock, "vm_mem_segs"); + return (0); } static bool @@ -93,10 +97,21 @@ vm_mem_destroy(struct vm *vm) for (int i = 0; i < VM_MAX_MEMSEGS; i++) vm_free_memseg(vm, i); + vmmops_vmspace_free(mem->mem_vmspace); + sx_xunlock(&mem->mem_segs_lock); sx_destroy(&mem->mem_segs_lock); } +struct vmspace * +vm_vmspace(struct vm *vm) +{ + struct vm_mem *mem; + + mem = vm_mem(vm); + return (mem->mem_vmspace); +} + void vm_slock_memsegs(struct vm *vm) { @@ -246,7 +261,7 @@ vm_mmap_memseg(struct vm *vm, vm_paddr_t gpa, int segid, vm_ooffset_t first, struct vm_mem *mem; struct vm_mem_seg *seg; struct vm_mem_map *m, *map; - struct vmspace *vmspace; + struct vm_map *vmmap; vm_ooffset_t last; int i, error; @@ -282,19 +297,19 @@ vm_mmap_memseg(struct vm *vm, vm_paddr_t gpa, int segid, vm_ooffset_t first, if (map == NULL) return (ENOSPC); - vmspace = vm_vmspace(vm); - error = vm_map_find(&vmspace->vm_map, seg->object, first, &gpa, - len, 0, VMFS_NO_SPACE, prot, prot, 0); + vmmap = &mem->mem_vmspace->vm_map; + error = vm_map_find(vmmap, seg->object, first, &gpa, len, 0, + VMFS_NO_SPACE, prot, prot, 0); if (error != KERN_SUCCESS) return (EFAULT); vm_object_reference(seg->object); if (flags & VM_MEMMAP_F_WIRED) { - error = vm_map_wire(&vmspace->vm_map, gpa, gpa + len, + error = vm_map_wire(vmmap, gpa, gpa + len, VM_MAP_WIRE_USER | VM_MAP_WIRE_NOHOLES); if (error != KERN_SUCCESS) { - vm_map_remove(&vmspace->vm_map, gpa, gpa + len); + vm_map_remove(vmmap, gpa, gpa + len); return (error == KERN_RESOURCE_SHORTAGE ? ENOMEM : EFAULT); } diff --git a/sys/dev/vmm/vmm_mem.h b/sys/dev/vmm/vmm_mem.h index 856470cf2590..f3d22058c7b8 100644 --- a/sys/dev/vmm/vmm_mem.h +++ b/sys/dev/vmm/vmm_mem.h @@ -36,6 +36,7 @@ enum { struct vm; struct vm_object; +struct vmspace; struct vm_mem_seg { size_t len; @@ -56,12 +57,15 @@ struct vm_mem { struct vm_mem_map mem_maps[VM_MAX_MEMMAPS]; struct vm_mem_seg mem_segs[VM_MAX_MEMSEGS]; struct sx mem_segs_lock; + struct vmspace *mem_vmspace; }; -void vm_mem_init(struct vm_mem *mem); +int vm_mem_init(struct vm_mem *mem, vm_offset_t lo, vm_offset_t hi); void vm_mem_cleanup(struct vm *vm); void vm_mem_destroy(struct vm *vm); +struct vmspace *vm_vmspace(struct vm *vm); + /* * APIs that modify the guest memory map require all vcpus to be frozen. */ diff --git a/sys/riscv/include/vmm.h b/sys/riscv/include/vmm.h index de7119dd534a..bc00474ed0fd 100644 --- a/sys/riscv/include/vmm.h +++ b/sys/riscv/include/vmm.h @@ -123,6 +123,29 @@ struct vm_eventinfo { int *iptr; /* reqidle cookie */ }; +#define DECLARE_VMMOPS_FUNC(ret_type, opname, args) \ + ret_type vmmops_##opname args + +DECLARE_VMMOPS_FUNC(int, modinit, (void)); +DECLARE_VMMOPS_FUNC(int, modcleanup, (void)); +DECLARE_VMMOPS_FUNC(void *, init, (struct vm *vm, struct pmap *pmap)); +DECLARE_VMMOPS_FUNC(int, gla2gpa, (void *vcpui, struct vm_guest_paging *paging, + uint64_t gla, int prot, uint64_t *gpa, int *is_fault)); +DECLARE_VMMOPS_FUNC(int, run, (void *vcpui, register_t pc, struct pmap *pmap, + struct vm_eventinfo *info)); +DECLARE_VMMOPS_FUNC(void, cleanup, (void *vmi)); +DECLARE_VMMOPS_FUNC(void *, vcpu_init, (void *vmi, struct vcpu *vcpu, + int vcpu_id)); +DECLARE_VMMOPS_FUNC(void, vcpu_cleanup, (void *vcpui)); +DECLARE_VMMOPS_FUNC(int, exception, (void *vcpui, uint64_t scause)); +DECLARE_VMMOPS_FUNC(int, getreg, (void *vcpui, int num, uint64_t *retval)); +DECLARE_VMMOPS_FUNC(int, setreg, (void *vcpui, int num, uint64_t val)); +DECLARE_VMMOPS_FUNC(int, getcap, (void *vcpui, int num, int *retval)); +DECLARE_VMMOPS_FUNC(int, setcap, (void *vcpui, int num, int val)); +DECLARE_VMMOPS_FUNC(struct vmspace *, vmspace_alloc, (vm_offset_t min, + vm_offset_t max)); +DECLARE_VMMOPS_FUNC(void, vmspace_free, (struct vmspace *vmspace)); + int vm_create(const char *name, struct vm **retvm); struct vcpu *vm_alloc_vcpu(struct vm *vm, int vcpuid); void vm_disable_vcpu_creation(struct vm *vm); @@ -212,7 +235,6 @@ vcpu_should_yield(struct vcpu *vcpu) void *vcpu_stats(struct vcpu *vcpu); void vcpu_notify_event(struct vcpu *vcpu); -struct vmspace *vm_vmspace(struct vm *vm); struct vm_mem *vm_mem(struct vm *vm); enum vm_reg_name vm_segment_name(int seg_encoding); diff --git a/sys/riscv/vmm/riscv.h b/sys/riscv/vmm/riscv.h index 870d0d6c5cd1..917a333520ed 100644 --- a/sys/riscv/vmm/riscv.h +++ b/sys/riscv/vmm/riscv.h @@ -122,29 +122,6 @@ struct hyptrap { uint64_t htinst; }; -#define DEFINE_VMMOPS_IFUNC(ret_type, opname, args) \ - ret_type vmmops_##opname args; - -DEFINE_VMMOPS_IFUNC(int, modinit, (void)) -DEFINE_VMMOPS_IFUNC(int, modcleanup, (void)) -DEFINE_VMMOPS_IFUNC(void *, init, (struct vm *vm, struct pmap *pmap)) -DEFINE_VMMOPS_IFUNC(int, gla2gpa, (void *vcpui, struct vm_guest_paging *paging, - uint64_t gla, int prot, uint64_t *gpa, int *is_fault)) -DEFINE_VMMOPS_IFUNC(int, run, (void *vcpui, register_t pc, struct pmap *pmap, - struct vm_eventinfo *info)) -DEFINE_VMMOPS_IFUNC(void, cleanup, (void *vmi)) -DEFINE_VMMOPS_IFUNC(void *, vcpu_init, (void *vmi, struct vcpu *vcpu, - int vcpu_id)) -DEFINE_VMMOPS_IFUNC(void, vcpu_cleanup, (void *vcpui)) -DEFINE_VMMOPS_IFUNC(int, exception, (void *vcpui, uint64_t scause)) -DEFINE_VMMOPS_IFUNC(int, getreg, (void *vcpui, int num, uint64_t *retval)) -DEFINE_VMMOPS_IFUNC(int, setreg, (void *vcpui, int num, uint64_t val)) -DEFINE_VMMOPS_IFUNC(int, getcap, (void *vcpui, int num, int *retval)) -DEFINE_VMMOPS_IFUNC(int, setcap, (void *vcpui, int num, int val)) -DEFINE_VMMOPS_IFUNC(struct vmspace *, vmspace_alloc, (vm_offset_t min, - vm_offset_t max)) -DEFINE_VMMOPS_IFUNC(void, vmspace_free, (struct vmspace *vmspace)) - #define dprintf(fmt, ...) struct hypctx *riscv_get_active_vcpu(void); diff --git a/sys/riscv/vmm/vmm.c b/sys/riscv/vmm/vmm.c index ec4514f70fa6..790dcc576507 100644 --- a/sys/riscv/vmm/vmm.c +++ b/sys/riscv/vmm/vmm.c @@ -92,7 +92,6 @@ struct vcpu { struct fpreg *guestfpu; /* (a,i) guest fpu state */ }; -#define vcpu_lock_initialized(v) mtx_initialized(&((v)->mtx)) #define vcpu_lock_init(v) mtx_init(&((v)->mtx), "vcpu lock", 0, MTX_SPIN) #define vcpu_lock_destroy(v) mtx_destroy(&((v)->mtx)) #define vcpu_lock(v) mtx_lock_spin(&((v)->mtx)) @@ -121,7 +120,6 @@ struct vm { bool dying; /* (o) is dying */ volatile cpuset_t suspended_cpus; /* (i) suspended vcpus */ volatile cpuset_t halted_cpus; /* (x) cpus in a hard halt */ - struct vmspace *vmspace; /* (o) guest's address space */ struct vm_mem mem; /* (i) [m+v] guest memory */ char name[VM_MAX_NAMELEN]; /* (o) virtual machine name */ struct vcpu **vcpu; /* (i) guest vcpus */ @@ -174,6 +172,7 @@ vcpu_cleanup(struct vcpu *vcpu, bool destroy) vmm_stat_free(vcpu->stats); fpu_save_area_free(vcpu->guestfpu); vcpu_lock_destroy(vcpu); + free(vcpu, M_VMM); } } @@ -285,7 +284,7 @@ vm_init(struct vm *vm, bool create) { int i; - vm->cookie = vmmops_init(vm, vmspace_pmap(vm->vmspace)); + vm->cookie = vmmops_init(vm, vmspace_pmap(vm_vmspace(vm))); MPASS(vm->cookie != NULL); CPU_ZERO(&vm->active_cpus); @@ -362,7 +361,7 @@ int vm_create(const char *name, struct vm **retvm) { struct vm *vm; - struct vmspace *vmspace; + int error; /* * If vmm.ko could not be successfully initialized then don't attempt @@ -374,14 +373,13 @@ vm_create(const char *name, struct vm **retvm) if (name == NULL || strlen(name) >= VM_MAX_NAMELEN) return (EINVAL); - vmspace = vmmops_vmspace_alloc(0, 1ul << 39); - if (vmspace == NULL) - return (ENOMEM); - vm = malloc(sizeof(struct vm), M_VMM, M_WAITOK | M_ZERO); + error = vm_mem_init(&vm->mem, 0, 1ul << 39); + if (error != 0) { + free(vm, M_VMM); + return (error); + } strcpy(vm->name, name); - vm->vmspace = vmspace; - vm_mem_init(&vm->mem); sx_init(&vm->vcpus_init_lock, "vm vcpus"); vm->sockets = 1; @@ -450,11 +448,6 @@ vm_cleanup(struct vm *vm, bool destroy) if (destroy) { vm_mem_destroy(vm); - vmmops_vmspace_free(vm->vmspace); - vm->vmspace = NULL; - - for (i = 0; i < vm->maxcpus; i++) - free(vm->vcpu[i], M_VMM); free(vm->vcpu, M_VMM); sx_destroy(&vm->vcpus_init_lock); } @@ -760,12 +753,6 @@ vcpu_notify_event(struct vcpu *vcpu) vcpu_unlock(vcpu); } -struct vmspace * -vm_vmspace(struct vm *vm) -{ - return (vm->vmspace); -} - struct vm_mem * vm_mem(struct vm *vm) { @@ -1084,7 +1071,7 @@ vm_handle_paging(struct vcpu *vcpu, bool *retu) vm = vcpu->vm; vme = &vcpu->exitinfo; - pmap = vmspace_pmap(vm->vmspace); + pmap = vmspace_pmap(vm_vmspace(vm)); addr = (vme->htval << 2) & ~(PAGE_SIZE - 1); dprintf("%s: %lx\n", __func__, addr); @@ -1107,7 +1094,7 @@ vm_handle_paging(struct vcpu *vcpu, bool *retu) if (pmap_fault(pmap, addr, ftype)) return (0); - map = &vm->vmspace->vm_map; + map = &vm_vmspace(vm)->vm_map; rv = vm_fault(map, addr, ftype, VM_FAULT_NORMAL, NULL); if (rv != KERN_SUCCESS) { printf("%s: vm_fault failed, addr %lx, ftype %d, err %d\n", @@ -1189,7 +1176,7 @@ vm_run(struct vcpu *vcpu) if (CPU_ISSET(vcpuid, &vm->suspended_cpus)) return (EINVAL); - pmap = vmspace_pmap(vm->vmspace); + pmap = vmspace_pmap(vm_vmspace(vm)); vme = &vcpu->exitinfo; evinfo.rptr = NULL; evinfo.sptr = &vm->suspend; diff --git a/sys/sys/user.h b/sys/sys/user.h index 3183f0792256..1704bc089d85 100644 --- a/sys/sys/user.h +++ b/sys/sys/user.h @@ -617,7 +617,8 @@ struct kinfo_vmobject { } kvo_type_spec; /* Type-specific union */ uint64_t kvo_me; /* Uniq handle for anon obj */ uint64_t kvo_laundry; /* Number of laundry pages. */ - uint64_t _kvo_qspare[5]; + uint64_t kvo_wired; /* Number of wired pages. */ + uint64_t _kvo_qspare[4]; uint32_t kvo_swapped; /* Number of swapped pages */ uint32_t kvo_flags; uint32_t _kvo_ispare[6]; diff --git a/sys/vm/vm_object.c b/sys/vm/vm_object.c index 6d9ea8bf9d93..5b4517d2bf0c 100644 --- a/sys/vm/vm_object.c +++ b/sys/vm/vm_object.c @@ -2522,15 +2522,13 @@ vm_object_list_handler(struct sysctl_req *req, bool swap_only) continue; } mtx_unlock(&vm_object_list_mtx); + + memset(kvo, 0, sizeof(*kvo)); kvo->kvo_size = ptoa(obj->size); kvo->kvo_resident = obj->resident_page_count; kvo->kvo_ref_count = obj->ref_count; kvo->kvo_shadow_count = atomic_load_int(&obj->shadow_count); kvo->kvo_memattr = obj->memattr; - kvo->kvo_active = 0; - kvo->kvo_inactive = 0; - kvo->kvo_laundry = 0; - kvo->kvo_flags = 0; if (!swap_only) { vm_page_iter_init(&pages, obj); VM_RADIX_FOREACH(m, &pages) { @@ -2549,12 +2547,12 @@ vm_object_list_handler(struct sysctl_req *req, bool swap_only) kvo->kvo_inactive++; else if (vm_page_in_laundry(m)) kvo->kvo_laundry++; + + if (vm_page_wired(m)) + kvo->kvo_wired++; } } - kvo->kvo_vn_fileid = 0; - kvo->kvo_vn_fsid = 0; - kvo->kvo_vn_fsid_freebsd11 = 0; freepath = NULL; fullpath = ""; vp = NULL; diff --git a/usr.bin/sockstat/tests/Makefile b/usr.bin/sockstat/tests/Makefile index 9971bca2d474..5412e9d842aa 100644 --- a/usr.bin/sockstat/tests/Makefile +++ b/usr.bin/sockstat/tests/Makefile @@ -1,5 +1,6 @@ ATF_TESTS_C+= sockstat_test -SRCS.sockstat_test= sockstat_test.c ../sockstat.c +SRCS.sockstat_test= sockstat_test.c sockstat.c +.PATH: ${.CURDIR:H} LIBADD= xo diff --git a/usr.bin/vmstat/vmstat.c b/usr.bin/vmstat/vmstat.c index 7a7c83fe1ac8..9b4d3a25ee07 100644 --- a/usr.bin/vmstat/vmstat.c +++ b/usr.bin/vmstat/vmstat.c @@ -1465,6 +1465,7 @@ display_object(struct kinfo_vmobject *kvo) xo_emit("{:active/%5ju} ", (uintmax_t)kvo->kvo_active); xo_emit("{:inactive/%5ju} ", (uintmax_t)kvo->kvo_inactive); xo_emit("{:laundry/%5ju} ", (uintmax_t)kvo->kvo_laundry); + xo_emit("{:wired/%5ju} ", (uintmax_t)kvo->kvo_wired); xo_emit("{:refcount/%3d} ", kvo->kvo_ref_count); xo_emit("{:shadowcount/%3d} ", kvo->kvo_shadow_count); @@ -1568,7 +1569,8 @@ doobjstat(void) return; } xo_emit("{T:RES/%5s} {T:ACT/%5s} {T:INACT/%5s} {T:LAUND/%5s} " - "{T:REF/%3s} {T:SHD/%3s} {T:CM/%2s} {T:TP/%3s} {T:PATH/%s}\n"); + "{T:WIRED/%5s} {T:REF/%3s} {T:SHD/%3s} {T:CM/%2s} {T:TP/%3s} " + "{T:PATH/%s}\n"); xo_open_list("object"); for (i = 0; i < cnt; i++) display_object(&kvo[i]); diff --git a/usr.sbin/rpc.tlsservd/rpc.tlsservd.c b/usr.sbin/rpc.tlsservd/rpc.tlsservd.c index f07385a2baa7..fb0501b2db4c 100644 --- a/usr.sbin/rpc.tlsservd/rpc.tlsservd.c +++ b/usr.sbin/rpc.tlsservd/rpc.tlsservd.c @@ -168,7 +168,12 @@ main(int argc, char **argv) rpctls_verbose = false; ncpu = (u_int)sysconf(_SC_NPROCESSORS_ONLN); +#ifdef notnow rpctls_maxthreads = ncpu > 1 ? ncpu / 2 : 1; +#else + /* XXX For now, until fixed properly!! */ + rpctls_maxthreads = 1; +#endif while ((ch = getopt_long(argc, argv, "2C:D:dhl:N:n:mp:r:uvWw", longopts, NULL)) != -1) { @@ -199,6 +204,8 @@ main(int argc, char **argv) if (rpctls_maxthreads < 1 || rpctls_maxthreads > ncpu) errx(1, "maximum threads must be between 1 and " "number of CPUs (%d)", ncpu); + /* XXX For now, until fixed properly!! */ + rpctls_maxthreads = 1; break; case 'n': hostname[0] = '@'; |