diff options
Diffstat (limited to 'sys')
91 files changed, 1496 insertions, 643 deletions
diff --git a/sys/amd64/vmm/io/vioapic.c b/sys/amd64/vmm/io/vioapic.c index 8869dc1383e6..7df6193d6dc0 100644 --- a/sys/amd64/vmm/io/vioapic.c +++ b/sys/amd64/vmm/io/vioapic.c @@ -130,6 +130,15 @@ vioapic_send_intr(struct vioapic *vioapic, int pin) vector = low & IOART_INTVEC; dest = high >> APIC_ID_SHIFT; + /* + * Ideally we'd just call lapic_intr_msi() here with the + * constructed MSI instead of interpreting it for ourselves. + * But until/unless we support emulated IOMMUs with interrupt + * remapping, interpretation is simple. We just need to mask + * in the Extended Destination ID bits for the 15-bit + * enlightenment (http://david.woodhou.se/ExtDestId.pdf) + */ + dest |= ((high & APIC_EXT_ID_MASK) >> APIC_EXT_ID_SHIFT) << 8; vlapic_deliver_intr(vioapic->vm, level, dest, phys, delmode, vector); } diff --git a/sys/amd64/vmm/vmm_lapic.c b/sys/amd64/vmm/vmm_lapic.c index fd511733492e..0cae01f172ec 100644 --- a/sys/amd64/vmm/vmm_lapic.c +++ b/sys/amd64/vmm/vmm_lapic.c @@ -115,6 +115,11 @@ lapic_intr_msi(struct vm *vm, uint64_t addr, uint64_t msg) * physical otherwise. */ dest = (addr >> 12) & 0xff; + /* + * Extended Destination ID support uses bits 5-11 of the address: + * http://david.woodhou.se/ExtDestId.pdf + */ + dest |= ((addr >> 5) & 0x7f) << 8; phys = ((addr & (MSI_X86_ADDR_RH | MSI_X86_ADDR_LOG)) != (MSI_X86_ADDR_RH | MSI_X86_ADDR_LOG)); delmode = msg & APIC_DELMODE_MASK; diff --git a/sys/amd64/vmm/x86.c b/sys/amd64/vmm/x86.c index 366f1da9f850..2e2224595ab4 100644 --- a/sys/amd64/vmm/x86.c +++ b/sys/amd64/vmm/x86.c @@ -48,7 +48,12 @@ SYSCTL_DECL(_hw_vmm); static SYSCTL_NODE(_hw_vmm, OID_AUTO, topology, CTLFLAG_RD | CTLFLAG_MPSAFE, 0, NULL); -#define CPUID_VM_HIGH 0x40000000 +#define CPUID_VM_SIGNATURE 0x40000000 +#define CPUID_BHYVE_FEATURES 0x40000001 +#define CPUID_VM_HIGH CPUID_BHYVE_FEATURES + +/* Features advertised in CPUID_BHYVE_FEATURES %eax */ +#define CPUID_BHYVE_FEAT_EXT_DEST_ID (1UL << 0) /* MSI Extended Dest ID */ static const char bhyve_id[12] = "bhyve bhyve "; @@ -100,7 +105,7 @@ x86_emulate_cpuid(struct vcpu *vcpu, uint64_t *rax, uint64_t *rbx, if (cpu_exthigh != 0 && func >= 0x80000000) { if (func > cpu_exthigh) func = cpu_exthigh; - } else if (func >= 0x40000000) { + } else if (func >= CPUID_VM_SIGNATURE) { if (func > CPUID_VM_HIGH) func = CPUID_VM_HIGH; } else if (func > cpu_high) { @@ -601,13 +606,20 @@ x86_emulate_cpuid(struct vcpu *vcpu, uint64_t *rax, uint64_t *rbx, regs[3] = 0; break; - case 0x40000000: + case CPUID_VM_SIGNATURE: regs[0] = CPUID_VM_HIGH; bcopy(bhyve_id, ®s[1], 4); bcopy(bhyve_id + 4, ®s[2], 4); bcopy(bhyve_id + 8, ®s[3], 4); break; + case CPUID_BHYVE_FEATURES: + regs[0] = CPUID_BHYVE_FEAT_EXT_DEST_ID; + regs[1] = 0; + regs[2] = 0; + regs[3] = 0; + break; + default: default_leaf: /* diff --git a/sys/arm/allwinner/aw_gpio.c b/sys/arm/allwinner/aw_gpio.c index 2061e38a155f..f1b6f0bc9193 100644 --- a/sys/arm/allwinner/aw_gpio.c +++ b/sys/arm/allwinner/aw_gpio.c @@ -1162,11 +1162,12 @@ aw_gpio_attach(device_t dev) fdt_pinctrl_register(dev, "allwinner,pins"); fdt_pinctrl_configure_tree(dev); - sc->sc_busdev = gpiobus_attach_bus(dev); + sc->sc_busdev = gpiobus_add_bus(dev); if (sc->sc_busdev == NULL) goto fail; config_intrhook_oneshot(aw_gpio_enable_bank_supply, sc); + bus_attach_children(dev); return (0); diff --git a/sys/arm/allwinner/axp209.c b/sys/arm/allwinner/axp209.c index 239ead02d0e0..ff999a0f9b9b 100644 --- a/sys/arm/allwinner/axp209.c +++ b/sys/arm/allwinner/axp209.c @@ -1322,7 +1322,7 @@ axp2xx_attach(device_t dev) case AXP209: sc->pins = axp209_pins; sc->npins = nitems(axp209_pins); - sc->gpiodev = gpiobus_attach_bus(dev); + sc->gpiodev = gpiobus_add_bus(dev); sc->sensors = axp209_sensors; sc->nsensors = nitems(axp209_sensors); @@ -1333,7 +1333,7 @@ axp2xx_attach(device_t dev) case AXP221: sc->pins = axp221_pins; sc->npins = nitems(axp221_pins); - sc->gpiodev = gpiobus_attach_bus(dev); + sc->gpiodev = gpiobus_add_bus(dev); sc->sensors = axp221_sensors; sc->nsensors = nitems(axp221_sensors); @@ -1374,6 +1374,7 @@ axp2xx_attach(device_t dev) } } + bus_attach_children(dev); return (0); } diff --git a/sys/arm/allwinner/axp81x.c b/sys/arm/allwinner/axp81x.c index fc1a168595e5..71f0c8156a0d 100644 --- a/sys/arm/allwinner/axp81x.c +++ b/sys/arm/allwinner/axp81x.c @@ -1609,7 +1609,8 @@ axp8xx_attach(device_t dev) EVENTHANDLER_REGISTER(shutdown_final, axp8xx_shutdown, dev, SHUTDOWN_PRI_LAST); - sc->gpiodev = gpiobus_attach_bus(dev); + sc->gpiodev = gpiobus_add_bus(dev); + bus_attach_children(dev); return (0); } diff --git a/sys/arm/broadcom/bcm2835/bcm2835_gpio.c b/sys/arm/broadcom/bcm2835/bcm2835_gpio.c index 48d1d2af5abc..93ee5d7c8bd3 100644 --- a/sys/arm/broadcom/bcm2835/bcm2835_gpio.c +++ b/sys/arm/broadcom/bcm2835/bcm2835_gpio.c @@ -840,10 +840,11 @@ bcm_gpio_attach(device_t dev) fdt_pinctrl_register(dev, "brcm,pins"); fdt_pinctrl_configure_tree(dev); - sc->sc_busdev = gpiobus_attach_bus(dev); + sc->sc_busdev = gpiobus_add_bus(dev); if (sc->sc_busdev == NULL) goto fail; + bus_attach_children(dev); return (0); fail: diff --git a/sys/arm/broadcom/bcm2835/raspberrypi_gpio.c b/sys/arm/broadcom/bcm2835/raspberrypi_gpio.c index 5a0f5cf2b1b3..b286654c6f18 100644 --- a/sys/arm/broadcom/bcm2835/raspberrypi_gpio.c +++ b/sys/arm/broadcom/bcm2835/raspberrypi_gpio.c @@ -404,10 +404,11 @@ rpi_fw_gpio_attach(device_t dev) } } free(names, M_OFWPROP); - sc->sc_busdev = gpiobus_attach_bus(dev); + sc->sc_busdev = gpiobus_add_bus(dev); if (sc->sc_busdev == NULL) goto fail; + bus_attach_children(dev); return (0); fail: diff --git a/sys/arm/freescale/imx/imx_gpio.c b/sys/arm/freescale/imx/imx_gpio.c index 7610d28af90e..3b19ef1b5e67 100644 --- a/sys/arm/freescale/imx/imx_gpio.c +++ b/sys/arm/freescale/imx/imx_gpio.c @@ -861,13 +861,14 @@ imx51_gpio_attach(device_t dev) gpio_pic_register_isrcs(sc); intr_pic_register(dev, OF_xref_from_node(ofw_bus_get_node(dev))); #endif - sc->sc_busdev = gpiobus_attach_bus(dev); + sc->sc_busdev = gpiobus_add_bus(dev); if (sc->sc_busdev == NULL) { imx51_gpio_detach(dev); return (ENXIO); } + bus_attach_children(dev); return (0); } diff --git a/sys/arm/freescale/vybrid/vf_gpio.c b/sys/arm/freescale/vybrid/vf_gpio.c index c81524a8a27e..b4e1ba9af586 100644 --- a/sys/arm/freescale/vybrid/vf_gpio.c +++ b/sys/arm/freescale/vybrid/vf_gpio.c @@ -147,13 +147,14 @@ vf_gpio_attach(device_t dev) "vf_gpio%d.%d", device_get_unit(dev), i); } - sc->sc_busdev = gpiobus_attach_bus(dev); + sc->sc_busdev = gpiobus_add_bus(dev); if (sc->sc_busdev == NULL) { bus_release_resources(dev, vf_gpio_spec, sc->res); mtx_destroy(&sc->sc_mtx); return (ENXIO); } + bus_attach_children(dev); return (0); } diff --git a/sys/arm/mv/a37x0_gpio.c b/sys/arm/mv/a37x0_gpio.c index 86110ff87ab1..754663d2991e 100644 --- a/sys/arm/mv/a37x0_gpio.c +++ b/sys/arm/mv/a37x0_gpio.c @@ -291,10 +291,11 @@ a37x0_gpio_attach(device_t dev) if (sc->sc_npins > sc->sc_max_pins) return (ENXIO); - sc->sc_busdev = gpiobus_attach_bus(dev); + sc->sc_busdev = gpiobus_add_bus(dev); if (sc->sc_busdev == NULL) return (ENXIO); + bus_attach_children(dev); return (0); } diff --git a/sys/arm/mv/gpio.c b/sys/arm/mv/gpio.c index 934c00236153..b3c2314fb2d6 100644 --- a/sys/arm/mv/gpio.c +++ b/sys/arm/mv/gpio.c @@ -340,7 +340,7 @@ mv_gpio_attach(device_t dev) if (rv != 0) return (rv); - sc->sc_busdev = gpiobus_attach_bus(dev); + sc->sc_busdev = gpiobus_add_bus(dev); if (sc->sc_busdev == NULL) { mtx_destroy(&sc->mutex); bus_release_resource(dev, SYS_RES_IRQ, @@ -348,6 +348,7 @@ mv_gpio_attach(device_t dev) return (ENXIO); } + bus_attach_children(dev); return (0); } diff --git a/sys/arm/mv/mvebu_gpio.c b/sys/arm/mv/mvebu_gpio.c index 7acdfff539dc..4cc9b7030a65 100644 --- a/sys/arm/mv/mvebu_gpio.c +++ b/sys/arm/mv/mvebu_gpio.c @@ -804,12 +804,13 @@ mvebu_gpio_attach(device_t dev) } } - sc->busdev = gpiobus_attach_bus(dev); + sc->busdev = gpiobus_add_bus(dev); if (sc->busdev == NULL) { mvebu_gpio_detach(dev); return (ENXIO); } + bus_attach_children(dev); return (0); } diff --git a/sys/arm/nvidia/tegra_gpio.c b/sys/arm/nvidia/tegra_gpio.c index e37fd69a121e..aa34537352be 100644 --- a/sys/arm/nvidia/tegra_gpio.c +++ b/sys/arm/nvidia/tegra_gpio.c @@ -818,12 +818,13 @@ tegra_gpio_attach(device_t dev) return (ENXIO); } - sc->busdev = gpiobus_attach_bus(dev); + sc->busdev = gpiobus_add_bus(dev); if (sc->busdev == NULL) { tegra_gpio_detach(dev); return (ENXIO); } + bus_attach_children(dev); return (0); } diff --git a/sys/arm/ti/ti_gpio.c b/sys/arm/ti/ti_gpio.c index aceb3d63204e..01b9597a4418 100644 --- a/sys/arm/ti/ti_gpio.c +++ b/sys/arm/ti/ti_gpio.c @@ -674,12 +674,13 @@ ti_gpio_attach(device_t dev) } } - sc->sc_busdev = gpiobus_attach_bus(dev); + sc->sc_busdev = gpiobus_add_bus(dev); if (sc->sc_busdev == NULL) { ti_gpio_detach(dev); return (ENXIO); } + bus_attach_children(dev); return (0); } diff --git a/sys/arm/xilinx/zy7_gpio.c b/sys/arm/xilinx/zy7_gpio.c index 71b6fc3c0586..2434e43bf27c 100644 --- a/sys/arm/xilinx/zy7_gpio.c +++ b/sys/arm/xilinx/zy7_gpio.c @@ -441,12 +441,13 @@ zy7_gpio_attach(device_t dev) return (ENOMEM); } - sc->busdev = gpiobus_attach_bus(dev); + sc->busdev = gpiobus_add_bus(dev); if (sc->busdev == NULL) { zy7_gpio_detach(dev); return (ENOMEM); } + bus_attach_children(dev); return (0); } diff --git a/sys/arm64/apple/apple_pinctrl.c b/sys/arm64/apple/apple_pinctrl.c index ebaaccea1d99..c28b1c62d78c 100644 --- a/sys/arm64/apple/apple_pinctrl.c +++ b/sys/arm64/apple/apple_pinctrl.c @@ -171,12 +171,13 @@ apple_pinctrl_attach(device_t dev) OF_xref_from_node(ofw_bus_get_node(dev))); } - sc->sc_busdev = gpiobus_attach_bus(dev); + sc->sc_busdev = gpiobus_add_bus(dev); if (sc->sc_busdev == NULL) { device_printf(dev, "failed to attach gpiobus\n"); goto error; } + bus_attach_children(dev); return (0); error: mtx_destroy(&sc->sc_mtx); diff --git a/sys/arm64/rockchip/rk_gpio.c b/sys/arm64/rockchip/rk_gpio.c index 847bc7394dd0..61614f532634 100644 --- a/sys/arm64/rockchip/rk_gpio.c +++ b/sys/arm64/rockchip/rk_gpio.c @@ -371,12 +371,13 @@ rk_gpio_attach(device_t dev) sc->swporta_ddr = rk_gpio_read_4(sc, RK_GPIO_SWPORTA_DDR); RK_GPIO_UNLOCK(sc); - sc->sc_busdev = gpiobus_attach_bus(dev); + sc->sc_busdev = gpiobus_add_bus(dev); if (sc->sc_busdev == NULL) { rk_gpio_detach(dev); return (ENXIO); } + bus_attach_children(dev); return (0); } diff --git a/sys/arm64/rockchip/rk_grf_gpio.c b/sys/arm64/rockchip/rk_grf_gpio.c index 6818bd85bb95..6ac419889614 100644 --- a/sys/arm64/rockchip/rk_grf_gpio.c +++ b/sys/arm64/rockchip/rk_grf_gpio.c @@ -181,11 +181,12 @@ rk_grf_gpio_attach(device_t dev) return (ENXIO); } - sc->sc_busdev = gpiobus_attach_bus(dev); + sc->sc_busdev = gpiobus_add_bus(dev); if (sc->sc_busdev == NULL) { return (ENXIO); } + bus_attach_children(dev); return (0); } diff --git a/sys/cddl/boot/zfs/zfsimpl.h b/sys/cddl/boot/zfs/zfsimpl.h index 83d964360343..915aeeda3c9e 100644 --- a/sys/cddl/boot/zfs/zfsimpl.h +++ b/sys/cddl/boot/zfs/zfsimpl.h @@ -536,6 +536,12 @@ typedef struct zio_gbh { offsetof(vdev_label_t, vl_uberblock[(n) << VDEV_UBERBLOCK_SHIFT(vd)]) #define VDEV_UBERBLOCK_SIZE(vd) (1ULL << VDEV_UBERBLOCK_SHIFT(vd)) +#define ASHIFT_UBERBLOCK_SHIFT(ashift) \ + MIN(MAX(ashift, UBERBLOCK_SHIFT), \ + MAX_UBERBLOCK_SHIFT) +#define ASHIFT_UBERBLOCK_SIZE(ashift) \ + (1ULL << ASHIFT_UBERBLOCK_SHIFT(ashift)) + typedef struct vdev_phys { char vp_nvlist[VDEV_PHYS_SIZE - sizeof (zio_eck_t)]; zio_eck_t vp_zbt; diff --git a/sys/compat/freebsd32/freebsd32_syscall.h b/sys/compat/freebsd32/freebsd32_syscall.h index 971510ebb6b6..90cd21a80923 100644 --- a/sys/compat/freebsd32/freebsd32_syscall.h +++ b/sys/compat/freebsd32/freebsd32_syscall.h @@ -83,8 +83,8 @@ /* 76 is obsolete vhangup */ /* 77 is obsolete vlimit */ #define FREEBSD32_SYS_mincore 78 -#define FREEBSD32_SYS_getgroups 79 -#define FREEBSD32_SYS_setgroups 80 +#define FREEBSD32_SYS_freebsd14_getgroups 79 +#define FREEBSD32_SYS_freebsd14_setgroups 80 #define FREEBSD32_SYS_getpgrp 81 #define FREEBSD32_SYS_setpgid 82 #define FREEBSD32_SYS_freebsd32_setitimer 83 @@ -513,4 +513,6 @@ #define FREEBSD32_SYS_exterrctl 592 #define FREEBSD32_SYS_inotify_add_watch_at 593 #define FREEBSD32_SYS_inotify_rm_watch 594 -#define FREEBSD32_SYS_MAXSYSCALL 595 +#define FREEBSD32_SYS_getgroups 595 +#define FREEBSD32_SYS_setgroups 596 +#define FREEBSD32_SYS_MAXSYSCALL 597 diff --git a/sys/compat/freebsd32/freebsd32_syscalls.c b/sys/compat/freebsd32/freebsd32_syscalls.c index 79c414b675b1..f0f8d26554b5 100644 --- a/sys/compat/freebsd32/freebsd32_syscalls.c +++ b/sys/compat/freebsd32/freebsd32_syscalls.c @@ -84,8 +84,8 @@ const char *freebsd32_syscallnames[] = { "obs_vhangup", /* 76 = obsolete vhangup */ "obs_vlimit", /* 77 = obsolete vlimit */ "mincore", /* 78 = mincore */ - "getgroups", /* 79 = getgroups */ - "setgroups", /* 80 = setgroups */ + "compat14.getgroups", /* 79 = freebsd14 getgroups */ + "compat14.setgroups", /* 80 = freebsd14 setgroups */ "getpgrp", /* 81 = getpgrp */ "setpgid", /* 82 = setpgid */ "freebsd32_setitimer", /* 83 = freebsd32_setitimer */ @@ -600,4 +600,6 @@ const char *freebsd32_syscallnames[] = { "exterrctl", /* 592 = exterrctl */ "inotify_add_watch_at", /* 593 = inotify_add_watch_at */ "inotify_rm_watch", /* 594 = inotify_rm_watch */ + "getgroups", /* 595 = getgroups */ + "setgroups", /* 596 = setgroups */ }; diff --git a/sys/compat/freebsd32/freebsd32_sysent.c b/sys/compat/freebsd32/freebsd32_sysent.c index 1a4b0d87722c..12f1a346c3e9 100644 --- a/sys/compat/freebsd32/freebsd32_sysent.c +++ b/sys/compat/freebsd32/freebsd32_sysent.c @@ -146,8 +146,8 @@ struct sysent freebsd32_sysent[] = { { .sy_narg = 0, .sy_call = (sy_call_t *)nosys, .sy_auevent = AUE_NULL, .sy_flags = 0, .sy_thrcnt = SY_THR_ABSENT }, /* 76 = obsolete vhangup */ { .sy_narg = 0, .sy_call = (sy_call_t *)nosys, .sy_auevent = AUE_NULL, .sy_flags = 0, .sy_thrcnt = SY_THR_ABSENT }, /* 77 = obsolete vlimit */ { .sy_narg = AS(mincore_args), .sy_call = (sy_call_t *)sys_mincore, .sy_auevent = AUE_MINCORE, .sy_flags = SYF_CAPENABLED, .sy_thrcnt = SY_THR_STATIC }, /* 78 = mincore */ - { .sy_narg = AS(getgroups_args), .sy_call = (sy_call_t *)sys_getgroups, .sy_auevent = AUE_GETGROUPS, .sy_flags = SYF_CAPENABLED, .sy_thrcnt = SY_THR_STATIC }, /* 79 = getgroups */ - { .sy_narg = AS(setgroups_args), .sy_call = (sy_call_t *)sys_setgroups, .sy_auevent = AUE_SETGROUPS, .sy_flags = 0, .sy_thrcnt = SY_THR_STATIC }, /* 80 = setgroups */ + { compat14(AS(freebsd14_getgroups_args),getgroups), .sy_auevent = AUE_GETGROUPS, .sy_flags = SYF_CAPENABLED, .sy_thrcnt = SY_THR_STATIC }, /* 79 = freebsd14 getgroups */ + { compat14(AS(freebsd14_setgroups_args),setgroups), .sy_auevent = AUE_SETGROUPS, .sy_flags = 0, .sy_thrcnt = SY_THR_STATIC }, /* 80 = freebsd14 setgroups */ { .sy_narg = 0, .sy_call = (sy_call_t *)sys_getpgrp, .sy_auevent = AUE_GETPGRP, .sy_flags = SYF_CAPENABLED, .sy_thrcnt = SY_THR_STATIC }, /* 81 = getpgrp */ { .sy_narg = AS(setpgid_args), .sy_call = (sy_call_t *)sys_setpgid, .sy_auevent = AUE_SETPGRP, .sy_flags = 0, .sy_thrcnt = SY_THR_STATIC }, /* 82 = setpgid */ { .sy_narg = AS(freebsd32_setitimer_args), .sy_call = (sy_call_t *)freebsd32_setitimer, .sy_auevent = AUE_SETITIMER, .sy_flags = SYF_CAPENABLED, .sy_thrcnt = SY_THR_STATIC }, /* 83 = freebsd32_setitimer */ @@ -662,4 +662,6 @@ struct sysent freebsd32_sysent[] = { { .sy_narg = AS(exterrctl_args), .sy_call = (sy_call_t *)sys_exterrctl, .sy_auevent = AUE_NULL, .sy_flags = SYF_CAPENABLED, .sy_thrcnt = SY_THR_STATIC }, /* 592 = exterrctl */ { .sy_narg = AS(inotify_add_watch_at_args), .sy_call = (sy_call_t *)sys_inotify_add_watch_at, .sy_auevent = AUE_INOTIFY, .sy_flags = SYF_CAPENABLED, .sy_thrcnt = SY_THR_STATIC }, /* 593 = inotify_add_watch_at */ { .sy_narg = AS(inotify_rm_watch_args), .sy_call = (sy_call_t *)sys_inotify_rm_watch, .sy_auevent = AUE_INOTIFY, .sy_flags = SYF_CAPENABLED, .sy_thrcnt = SY_THR_STATIC }, /* 594 = inotify_rm_watch */ + { .sy_narg = AS(getgroups_args), .sy_call = (sy_call_t *)sys_getgroups, .sy_auevent = AUE_GETGROUPS, .sy_flags = SYF_CAPENABLED, .sy_thrcnt = SY_THR_STATIC }, /* 595 = getgroups */ + { .sy_narg = AS(setgroups_args), .sy_call = (sy_call_t *)sys_setgroups, .sy_auevent = AUE_SETGROUPS, .sy_flags = 0, .sy_thrcnt = SY_THR_STATIC }, /* 596 = setgroups */ }; diff --git a/sys/compat/freebsd32/freebsd32_systrace_args.c b/sys/compat/freebsd32/freebsd32_systrace_args.c index f9dc514bee7d..e471c5148021 100644 --- a/sys/compat/freebsd32/freebsd32_systrace_args.c +++ b/sys/compat/freebsd32/freebsd32_systrace_args.c @@ -457,22 +457,6 @@ systrace_args(int sysnum, void *params, uint64_t *uarg, int *n_args) *n_args = 3; break; } - /* getgroups */ - case 79: { - struct getgroups_args *p = params; - iarg[a++] = p->gidsetsize; /* int */ - uarg[a++] = (intptr_t)p->gidset; /* gid_t * */ - *n_args = 2; - break; - } - /* setgroups */ - case 80: { - struct setgroups_args *p = params; - iarg[a++] = p->gidsetsize; /* int */ - uarg[a++] = (intptr_t)p->gidset; /* const gid_t * */ - *n_args = 2; - break; - } /* getpgrp */ case 81: { *n_args = 0; @@ -3413,6 +3397,22 @@ systrace_args(int sysnum, void *params, uint64_t *uarg, int *n_args) *n_args = 2; break; } + /* getgroups */ + case 595: { + struct getgroups_args *p = params; + iarg[a++] = p->gidsetsize; /* int */ + uarg[a++] = (intptr_t)p->gidset; /* gid_t * */ + *n_args = 2; + break; + } + /* setgroups */ + case 596: { + struct setgroups_args *p = params; + iarg[a++] = p->gidsetsize; /* int */ + uarg[a++] = (intptr_t)p->gidset; /* const gid_t * */ + *n_args = 2; + break; + } default: *n_args = 0; break; @@ -4112,32 +4112,6 @@ systrace_entry_setargdesc(int sysnum, int ndx, char *desc, size_t descsz) break; }; break; - /* getgroups */ - case 79: - switch (ndx) { - case 0: - p = "int"; - break; - case 1: - p = "userland gid_t *"; - break; - default: - break; - }; - break; - /* setgroups */ - case 80: - switch (ndx) { - case 0: - p = "int"; - break; - case 1: - p = "userland const gid_t *"; - break; - default: - break; - }; - break; /* getpgrp */ case 81: break; @@ -9222,6 +9196,32 @@ systrace_entry_setargdesc(int sysnum, int ndx, char *desc, size_t descsz) break; }; break; + /* getgroups */ + case 595: + switch (ndx) { + case 0: + p = "int"; + break; + case 1: + p = "userland gid_t *"; + break; + default: + break; + }; + break; + /* setgroups */ + case 596: + switch (ndx) { + case 0: + p = "int"; + break; + case 1: + p = "userland const gid_t *"; + break; + default: + break; + }; + break; default: break; }; @@ -9488,16 +9488,6 @@ systrace_return_setargdesc(int sysnum, int ndx, char *desc, size_t descsz) if (ndx == 0 || ndx == 1) p = "int"; break; - /* getgroups */ - case 79: - if (ndx == 0 || ndx == 1) - p = "int"; - break; - /* setgroups */ - case 80: - if (ndx == 0 || ndx == 1) - p = "int"; - break; /* getpgrp */ case 81: /* setpgid */ @@ -11130,6 +11120,16 @@ systrace_return_setargdesc(int sysnum, int ndx, char *desc, size_t descsz) if (ndx == 0 || ndx == 1) p = "int"; break; + /* getgroups */ + case 595: + if (ndx == 0 || ndx == 1) + p = "int"; + break; + /* setgroups */ + case 596: + if (ndx == 0 || ndx == 1) + p = "int"; + break; default: break; }; diff --git a/sys/conf/files.arm64 b/sys/conf/files.arm64 index 641001efab5e..45966fea8041 100644 --- a/sys/conf/files.arm64 +++ b/sys/conf/files.arm64 @@ -178,6 +178,8 @@ crypto/des/des_enc.c optional netsmb crypto/openssl/ossl_aarch64.c optional ossl crypto/openssl/aarch64/chacha-armv8.S optional ossl \ compile-with "${CC} -c ${CFLAGS:N-mgeneral-regs-only} -I$S/crypto/openssl ${WERROR} ${.IMPSRC}" +crypto/openssl/aarch64/chacha-armv8-sve.S optional ossl \ + compile-with "${CC} -c ${CFLAGS:N-mgeneral-regs-only} -I$S/crypto/openssl ${WERROR} ${.IMPSRC}" crypto/openssl/aarch64/poly1305-armv8.S optional ossl \ compile-with "${CC} -c ${CFLAGS:N-mgeneral-regs-only} -I$S/crypto/openssl ${WERROR} ${.IMPSRC}" crypto/openssl/aarch64/sha1-armv8.S optional ossl \ diff --git a/sys/contrib/openzfs/module/zfs/dsl_deadlist.c b/sys/contrib/openzfs/module/zfs/dsl_deadlist.c index 475db3c89508..41ac72bf1c16 100644 --- a/sys/contrib/openzfs/module/zfs/dsl_deadlist.c +++ b/sys/contrib/openzfs/module/zfs/dsl_deadlist.c @@ -1049,7 +1049,8 @@ dsl_livelist_iterate(void *arg, const blkptr_t *bp, boolean_t bp_freed, ASSERT3U(BP_GET_PSIZE(bp), ==, BP_GET_PSIZE(&found->le_bp)); ASSERT3U(BP_GET_CHECKSUM(bp), ==, BP_GET_CHECKSUM(&found->le_bp)); - ASSERT3U(BP_GET_BIRTH(bp), ==, BP_GET_BIRTH(&found->le_bp)); + ASSERT3U(BP_GET_PHYSICAL_BIRTH(bp), ==, + BP_GET_PHYSICAL_BIRTH(&found->le_bp)); } if (bp_freed) { if (found == NULL) { diff --git a/sys/dev/amdgpio/amdgpio.c b/sys/dev/amdgpio/amdgpio.c index f39006d95805..2bd455c612b8 100644 --- a/sys/dev/amdgpio/amdgpio.c +++ b/sys/dev/amdgpio/amdgpio.c @@ -408,12 +408,13 @@ amdgpio_attach(device_t dev) GPIO_PIN_OUTPUT : GPIO_PIN_INPUT; } - sc->sc_busdev = gpiobus_attach_bus(dev); + sc->sc_busdev = gpiobus_add_bus(dev); if (sc->sc_busdev == NULL) { device_printf(dev, "could not attach gpiobus\n"); goto err_bus; } + bus_attach_children(dev); return (0); err_bus: diff --git a/sys/dev/bhnd/cores/chipc/chipc_gpio.c b/sys/dev/bhnd/cores/chipc/chipc_gpio.c index a110bdda5fa7..429de0fc1fd8 100644 --- a/sys/dev/bhnd/cores/chipc/chipc_gpio.c +++ b/sys/dev/bhnd/cores/chipc/chipc_gpio.c @@ -173,11 +173,13 @@ chipc_gpio_attach(device_t dev) if (CC_GPIO_QUIRK(sc, NO_GPIOC)) { sc->gpiobus = NULL; } else { - if ((sc->gpiobus = gpiobus_attach_bus(dev)) == NULL) { + if ((sc->gpiobus = gpiobus_add_bus(dev)) == NULL) { device_printf(dev, "failed to attach gpiobus\n"); error = ENXIO; goto failed; } + + bus_attach_children(dev); } /* Register as the bus GPIO provider */ diff --git a/sys/dev/ftgpio/ftgpio.c b/sys/dev/ftgpio/ftgpio.c index 7acfdd5b900e..68787b54bb16 100644 --- a/sys/dev/ftgpio/ftgpio.c +++ b/sys/dev/ftgpio/ftgpio.c @@ -398,12 +398,13 @@ ftgpio_attach(device_t dev) FTGPIO_VERBOSE_PRINTF(sc->dev, "groups GPIO1..GPIO6 enabled\n"); GPIO_UNLOCK(sc); - sc->busdev = gpiobus_attach_bus(dev); + sc->busdev = gpiobus_add_bus(dev); if (sc->busdev == NULL) { GPIO_LOCK_DESTROY(sc); return (ENXIO); } + bus_attach_children(dev); return (0); } diff --git a/sys/dev/gpio/acpi_gpiobus.c b/sys/dev/gpio/acpi_gpiobus.c index 94f4e5771266..170f23615416 100644 --- a/sys/dev/gpio/acpi_gpiobus.c +++ b/sys/dev/gpio/acpi_gpiobus.c @@ -383,7 +383,8 @@ acpi_gpiobus_detach(device_t dev) } static int -acpi_gpiobus_read_ivar(device_t dev, device_t child, int which, uintptr_t *result) +acpi_gpiobus_read_ivar(device_t dev, device_t child, int which, + uintptr_t *result) { struct acpi_gpiobus_ivar *devi = device_get_ivars(child); diff --git a/sys/dev/gpio/bytgpio.c b/sys/dev/gpio/bytgpio.c index f7b2a73ec6cb..5d685c155a03 100644 --- a/sys/dev/gpio/bytgpio.c +++ b/sys/dev/gpio/bytgpio.c @@ -608,7 +608,7 @@ bytgpio_attach(device_t dev) sc->sc_pad_funcs[pin] = val & BYTGPIO_PCONF0_FUNC_MASK; } - sc->sc_busdev = gpiobus_attach_bus(dev); + sc->sc_busdev = gpiobus_add_bus(dev); if (sc->sc_busdev == NULL) { BYTGPIO_LOCK_DESTROY(sc); bus_release_resource(dev, SYS_RES_MEMORY, @@ -616,6 +616,7 @@ bytgpio_attach(device_t dev) return (ENXIO); } + bus_attach_children(dev); return (0); error: diff --git a/sys/dev/gpio/chvgpio.c b/sys/dev/gpio/chvgpio.c index 199ad4d6f373..3273aad9242b 100644 --- a/sys/dev/gpio/chvgpio.c +++ b/sys/dev/gpio/chvgpio.c @@ -441,7 +441,7 @@ chvgpio_attach(device_t dev) bus_write_4(sc->sc_mem_res, CHVGPIO_INTERRUPT_MASK, 0); bus_write_4(sc->sc_mem_res, CHVGPIO_INTERRUPT_STATUS, 0xffff); - sc->sc_busdev = gpiobus_attach_bus(dev); + sc->sc_busdev = gpiobus_add_bus(dev); if (sc->sc_busdev == NULL) { CHVGPIO_LOCK_DESTROY(sc); bus_release_resource(dev, SYS_RES_MEMORY, @@ -451,6 +451,7 @@ chvgpio_attach(device_t dev) return (ENXIO); } + bus_attach_children(dev); return (0); } diff --git a/sys/dev/gpio/dwgpio/dwgpio.c b/sys/dev/gpio/dwgpio/dwgpio.c index 5acb99ca591e..3908113d5fd4 100644 --- a/sys/dev/gpio/dwgpio/dwgpio.c +++ b/sys/dev/gpio/dwgpio/dwgpio.c @@ -167,12 +167,13 @@ dwgpio_attach(device_t dev) snprintf(sc->gpio_pins[i].gp_name, GPIOMAXNAME, "dwgpio%d.%d", device_get_unit(dev), i); } - sc->busdev = gpiobus_attach_bus(dev); + sc->busdev = gpiobus_add_bus(dev); if (sc->busdev == NULL) { mtx_destroy(&sc->sc_mtx); return (ENXIO); } + bus_attach_children(dev); return (0); } diff --git a/sys/dev/gpio/gpiobus.c b/sys/dev/gpio/gpiobus.c index 764bcb7e6ee8..c25c41f43042 100644 --- a/sys/dev/gpio/gpiobus.c +++ b/sys/dev/gpio/gpiobus.c @@ -330,24 +330,6 @@ gpiobus_add_bus(device_t dev) return (busdev); } -/* - * Attach a gpiobus child. - * Note that the controller is expected - * to be fully initialized at this point. - */ -device_t -gpiobus_attach_bus(device_t dev) -{ - device_t busdev; - - busdev = gpiobus_add_bus(dev); - if (busdev == NULL) - return (NULL); - - bus_attach_children(dev); - return (busdev); -} - int gpiobus_detach_bus(device_t dev) { diff --git a/sys/dev/gpio/gpiobusvar.h b/sys/dev/gpio/gpiobusvar.h index 7f504236a774..0528efe45525 100644 --- a/sys/dev/gpio/gpiobusvar.h +++ b/sys/dev/gpio/gpiobusvar.h @@ -171,7 +171,6 @@ struct resource *gpio_alloc_intr_resource(device_t consumer_dev, int *rid, int gpio_check_flags(uint32_t, uint32_t); device_t gpiobus_add_bus(device_t); -device_t gpiobus_attach_bus(device_t); int gpiobus_detach_bus(device_t); #endif /* __GPIOBUS_H__ */ diff --git a/sys/dev/gpio/pl061.c b/sys/dev/gpio/pl061.c index 87d4310a6396..32109e5982bc 100644 --- a/sys/dev/gpio/pl061.c +++ b/sys/dev/gpio/pl061.c @@ -495,13 +495,14 @@ pl061_attach(device_t dev) goto free_isrc; } - sc->sc_busdev = gpiobus_attach_bus(dev); + sc->sc_busdev = gpiobus_add_bus(dev); if (sc->sc_busdev == NULL) { device_printf(dev, "couldn't attach gpio bus\n"); PL061_LOCK_DESTROY(sc); goto free_isrc; } + bus_attach_children(dev); return (0); free_isrc: diff --git a/sys/dev/gpio/qoriq_gpio.c b/sys/dev/gpio/qoriq_gpio.c index 8b44cd256c79..d11868a23751 100644 --- a/sys/dev/gpio/qoriq_gpio.c +++ b/sys/dev/gpio/qoriq_gpio.c @@ -379,12 +379,13 @@ qoriq_gpio_attach(device_t dev) OF_device_register_xref(OF_xref_from_node(ofw_bus_get_node(dev)), dev); - sc->busdev = gpiobus_attach_bus(dev); + sc->busdev = gpiobus_add_bus(dev); if (sc->busdev == NULL) { qoriq_gpio_detach(dev); return (ENOMEM); } + bus_attach_children(dev); return (0); } diff --git a/sys/dev/iicbus/gpio/pcf8574.c b/sys/dev/iicbus/gpio/pcf8574.c index ab6e2bc07d1f..bf60dec67557 100644 --- a/sys/dev/iicbus/gpio/pcf8574.c +++ b/sys/dev/iicbus/gpio/pcf8574.c @@ -142,12 +142,13 @@ pcf8574_attach(device_t dev) (void)pcf8574_write(sc, 0xff); sx_init(&sc->lock, "pcf8574"); - sc->busdev = gpiobus_attach_bus(dev); + sc->busdev = gpiobus_add_bus(dev); if (sc->busdev == NULL) { device_printf(dev, "Could not create busdev child\n"); sx_destroy(&sc->lock); return (ENXIO); } + bus_attach_children(dev); return (0); } @@ -158,9 +159,7 @@ pcf8574_detach(device_t dev) sc = device_get_softc(dev); - if (sc->busdev != NULL) - gpiobus_detach_bus(sc->busdev); - + gpiobus_detach_bus(dev); sx_destroy(&sc->lock); return (0); } diff --git a/sys/dev/iicbus/gpio/tca64xx.c b/sys/dev/iicbus/gpio/tca64xx.c index cd011ae9be75..ab8fedd3f8fd 100644 --- a/sys/dev/iicbus/gpio/tca64xx.c +++ b/sys/dev/iicbus/gpio/tca64xx.c @@ -262,7 +262,7 @@ tca64xx_attach(device_t dev) mtx_init(&sc->mtx, "tca64xx gpio", "gpio", MTX_DEF); OF_device_register_xref(OF_xref_from_node(ofw_bus_get_node(dev)), dev); - sc->busdev = gpiobus_attach_bus(dev); + sc->busdev = gpiobus_add_bus(dev); if (sc->busdev == NULL) { device_printf(dev, "Could not create busdev child\n"); return (ENXIO); @@ -281,6 +281,7 @@ tca64xx_attach(device_t dev) } #endif + bus_attach_children(dev); return (0); } @@ -291,9 +292,7 @@ tca64xx_detach(device_t dev) sc = device_get_softc(dev); - if (sc->busdev != NULL) - gpiobus_detach_bus(sc->busdev); - + gpiobus_detach_bus(dev); mtx_destroy(&sc->mtx); return (0); diff --git a/sys/dev/mlx5/mlx5_en/mlx5_en_flow_table.c b/sys/dev/mlx5/mlx5_en/mlx5_en_flow_table.c index 6e24395b5577..c45f02cdaf42 100644 --- a/sys/dev/mlx5/mlx5_en/mlx5_en_flow_table.c +++ b/sys/dev/mlx5/mlx5_en/mlx5_en_flow_table.c @@ -1783,8 +1783,8 @@ mlx5e_add_vxlan_rule(struct mlx5e_priv *priv, sa_family_t family, u_int port) el->refcount++; if (el->installed) return (0); - } - el = mlx5e_vxlan_alloc_db_el(priv, proto, port); + } else + el = mlx5e_vxlan_alloc_db_el(priv, proto, port); if ((if_getcapenable(priv->ifp) & IFCAP_VXLAN_HWCSUM) != 0) { err = mlx5e_add_vxlan_rule_from_db(priv, el); diff --git a/sys/dev/nctgpio/nctgpio.c b/sys/dev/nctgpio/nctgpio.c index 75ea1fbdba17..ddc2ceef7dfb 100644 --- a/sys/dev/nctgpio/nctgpio.c +++ b/sys/dev/nctgpio/nctgpio.c @@ -1258,13 +1258,14 @@ nct_attach(device_t dev) GPIO_UNLOCK(sc); - sc->busdev = gpiobus_attach_bus(dev); + sc->busdev = gpiobus_add_bus(dev); if (sc->busdev == NULL) { device_printf(dev, "failed to attach to gpiobus\n"); GPIO_LOCK_DESTROY(sc); return (ENXIO); } + bus_attach_children(dev); return (0); } diff --git a/sys/dev/p2sb/lewisburg_gpio.c b/sys/dev/p2sb/lewisburg_gpio.c index b45d7767602c..3be777ab9524 100644 --- a/sys/dev/p2sb/lewisburg_gpio.c +++ b/sys/dev/p2sb/lewisburg_gpio.c @@ -217,10 +217,11 @@ lbggpio_attach(device_t dev) } /* support gpio */ - sc->sc_busdev = gpiobus_attach_bus(dev); + sc->sc_busdev = gpiobus_add_bus(dev); if (sc->sc_busdev == NULL) return (ENXIO); + bus_attach_children(dev); return (0); } diff --git a/sys/dev/qcom_tlmm/qcom_tlmm_ipq4018.c b/sys/dev/qcom_tlmm/qcom_tlmm_ipq4018.c index 2d390cd449af..50f54b896748 100644 --- a/sys/dev/qcom_tlmm/qcom_tlmm_ipq4018.c +++ b/sys/dev/qcom_tlmm/qcom_tlmm_ipq4018.c @@ -346,13 +346,14 @@ qcom_tlmm_ipq4018_attach(device_t dev) fdt_pinctrl_register(dev, NULL); fdt_pinctrl_configure_by_name(dev, "default"); - sc->busdev = gpiobus_attach_bus(dev); + sc->busdev = gpiobus_add_bus(dev); if (sc->busdev == NULL) { device_printf(dev, "%s: failed to attach bus\n", __func__); qcom_tlmm_ipq4018_detach(dev); return (ENXIO); } + bus_attach_children(dev); return (0); } diff --git a/sys/dev/rccgpio/rccgpio.c b/sys/dev/rccgpio/rccgpio.c index b2b775b879ad..dafd0b511fa9 100644 --- a/sys/dev/rccgpio/rccgpio.c +++ b/sys/dev/rccgpio/rccgpio.c @@ -308,7 +308,7 @@ rcc_gpio_attach(device_t dev) RCC_WRITE(sc, RCC_GPIO_GP_LVL, sc->sc_output); /* Attach the gpiobus. */ - sc->sc_busdev = gpiobus_attach_bus(dev); + sc->sc_busdev = gpiobus_add_bus(dev); if (sc->sc_busdev == NULL) { bus_release_resource(dev, SYS_RES_IOPORT, sc->sc_io_rid, sc->sc_io_res); @@ -316,6 +316,7 @@ rcc_gpio_attach(device_t dev) return (ENXIO); } + bus_attach_children(dev); return (0); } diff --git a/sys/dev/uart/uart_cpu_acpi.c b/sys/dev/uart/uart_cpu_acpi.c index 7382c47a8db6..da77603f0093 100644 --- a/sys/dev/uart/uart_cpu_acpi.c +++ b/sys/dev/uart/uart_cpu_acpi.c @@ -44,23 +44,15 @@ #include <contrib/dev/acpica/include/accommon.h> #include <contrib/dev/acpica/include/actables.h> -static struct acpi_uart_compat_data * +static struct acpi_spcr_compat_data * uart_cpu_acpi_scan(uint8_t interface_type) { - struct acpi_uart_compat_data **cd, *curcd; + struct acpi_spcr_compat_data **cd, *curcd; int i; - SET_FOREACH(cd, uart_acpi_class_and_device_set) { + SET_FOREACH(cd, uart_acpi_spcr_class_set) { curcd = *cd; - for (i = 0; curcd[i].cd_hid != NULL; i++) { - if (curcd[i].cd_port_subtype == interface_type) - return (&curcd[i]); - } - } - - SET_FOREACH(cd, uart_acpi_class_set) { - curcd = *cd; - for (i = 0; curcd[i].cd_hid != NULL; i++) { + for (i = 0; curcd[i].cd_class != NULL; i++) { if (curcd[i].cd_port_subtype == interface_type) return (&curcd[i]); } @@ -143,7 +135,7 @@ uart_cpu_acpi_spcr(int devtype, struct uart_devinfo *di) { vm_paddr_t spcr_physaddr; ACPI_TABLE_SPCR *spcr; - struct acpi_uart_compat_data *cd; + struct acpi_spcr_compat_data *cd; struct uart_class *class; int error = ENXIO; @@ -237,7 +229,7 @@ uart_cpu_acpi_dbg2(struct uart_devinfo *di) ACPI_TABLE_DBG2 *dbg2; ACPI_DBG2_DEVICE *dbg2_dev; ACPI_GENERIC_ADDRESS *base_address; - struct acpi_uart_compat_data *cd; + struct acpi_spcr_compat_data *cd; struct uart_class *class; int error; bool found; diff --git a/sys/dev/uart/uart_cpu_acpi.h b/sys/dev/uart/uart_cpu_acpi.h index 94329e1f1349..218f643c7621 100644 --- a/sys/dev/uart/uart_cpu_acpi.h +++ b/sys/dev/uart/uart_cpu_acpi.h @@ -35,11 +35,18 @@ struct uart_class; +struct acpi_spcr_compat_data { + struct uart_class *cd_class; + uint16_t cd_port_subtype; +}; +SET_DECLARE(uart_acpi_spcr_class_set, struct acpi_spcr_compat_data); +#define UART_ACPI_SPCR_CLASS(data) \ + DATA_SET(uart_acpi_spcr_class_set, data) + struct acpi_uart_compat_data { const char *cd_hid; struct uart_class *cd_class; - uint16_t cd_port_subtype; int cd_regshft; int cd_regiowidth; int cd_rclk; @@ -56,14 +63,6 @@ SET_DECLARE(uart_acpi_class_and_device_set, struct acpi_uart_compat_data); #define UART_ACPI_CLASS_AND_DEVICE(data) \ DATA_SET(uart_acpi_class_and_device_set, data) -/* - * If your UART driver implements uart_class and custom device layer, - * then use UART_ACPI_CLASS for its declaration - */ -SET_DECLARE(uart_acpi_class_set, struct acpi_uart_compat_data); -#define UART_ACPI_CLASS(data) \ - DATA_SET(uart_acpi_class_set, data) - /* Try to initialize UART device from ACPI tables */ int uart_cpu_acpi_setup(int devtype, struct uart_devinfo *di); diff --git a/sys/dev/uart/uart_dev_ns8250.c b/sys/dev/uart/uart_dev_ns8250.c index 0f19ede6d9df..c38d50e54ad8 100644 --- a/sys/dev/uart/uart_dev_ns8250.c +++ b/sys/dev/uart/uart_dev_ns8250.c @@ -492,24 +492,32 @@ UART_CLASS(uart_ns8250_class); * XXX -- refactor out ACPI and FDT ifdefs */ #ifdef DEV_ACPI +static struct acpi_spcr_compat_data acpi_spcr_compat_data[] = { + { &uart_ns8250_class, ACPI_DBG2_16550_COMPATIBLE }, + { &uart_ns8250_class, ACPI_DBG2_16550_SUBSET }, + { &uart_ns8250_class, ACPI_DBG2_16550_WITH_GAS }, + { NULL, 0 }, +}; +UART_ACPI_SPCR_CLASS(acpi_spcr_compat_data); + static struct acpi_uart_compat_data acpi_compat_data[] = { - {"AMD0020", &uart_ns8250_class, 0, 2, 0, 48000000, UART_F_BUSY_DETECT, "AMD / Synopsys Designware UART"}, - {"AMDI0020", &uart_ns8250_class, 0, 2, 0, 48000000, UART_F_BUSY_DETECT, "AMD / Synopsys Designware UART"}, - {"APMC0D08", &uart_ns8250_class, ACPI_DBG2_16550_COMPATIBLE, 2, 4, 0, 0, "APM compatible UART"}, - {"MRVL0001", &uart_ns8250_class, ACPI_DBG2_16550_SUBSET, 2, 0, 200000000, UART_F_BUSY_DETECT, "Marvell / Synopsys Designware UART"}, - {"SCX0006", &uart_ns8250_class, 0, 2, 0, 62500000, UART_F_BUSY_DETECT, "SynQuacer / Synopsys Designware UART"}, - {"HISI0031", &uart_ns8250_class, 0, 2, 0, 200000000, UART_F_BUSY_DETECT, "HiSilicon / Synopsys Designware UART"}, - {"INTC1006", &uart_ns8250_class, 0, 2, 0, 25000000, 0, "Intel ARM64 UART"}, - {"NXP0018", &uart_ns8250_class, 0, 0, 0, 350000000, UART_F_BUSY_DETECT, "NXP / Synopsys Designware UART"}, - {"PNP0500", &uart_ns8250_class, 0, 0, 0, 0, 0, "Standard PC COM port"}, - {"PNP0501", &uart_ns8250_class, 0, 0, 0, 0, 0, "16550A-compatible COM port"}, - {"PNP0502", &uart_ns8250_class, 0, 0, 0, 0, 0, "Multiport serial device (non-intelligent 16550)"}, - {"PNP0510", &uart_ns8250_class, 0, 0, 0, 0, 0, "Generic IRDA-compatible device"}, - {"PNP0511", &uart_ns8250_class, 0, 0, 0, 0, 0, "Generic IRDA-compatible device"}, - {"WACF004", &uart_ns8250_class, 0, 0, 0, 0, 0, "Wacom Tablet PC Screen"}, - {"WACF00E", &uart_ns8250_class, 0, 0, 0, 0, 0, "Wacom Tablet PC Screen 00e"}, - {"FUJ02E5", &uart_ns8250_class, 0, 0, 0, 0, 0, "Wacom Tablet at FuS Lifebook T"}, - {NULL, NULL, 0, 0 , 0, 0, 0, NULL}, + {"AMD0020", &uart_ns8250_class, 2, 0, 48000000, UART_F_BUSY_DETECT, "AMD / Synopsys Designware UART"}, + {"AMDI0020", &uart_ns8250_class, 2, 0, 48000000, UART_F_BUSY_DETECT, "AMD / Synopsys Designware UART"}, + {"APMC0D08", &uart_ns8250_class, 2, 4, 0, 0, "APM compatible UART"}, + {"MRVL0001", &uart_ns8250_class, 2, 0, 200000000, UART_F_BUSY_DETECT, "Marvell / Synopsys Designware UART"}, + {"SCX0006", &uart_ns8250_class, 2, 0, 62500000, UART_F_BUSY_DETECT, "SynQuacer / Synopsys Designware UART"}, + {"HISI0031", &uart_ns8250_class, 2, 0, 200000000, UART_F_BUSY_DETECT, "HiSilicon / Synopsys Designware UART"}, + {"INTC1006", &uart_ns8250_class, 2, 0, 25000000, 0, "Intel ARM64 UART"}, + {"NXP0018", &uart_ns8250_class, 0, 0, 350000000, UART_F_BUSY_DETECT, "NXP / Synopsys Designware UART"}, + {"PNP0500", &uart_ns8250_class, 0, 0, 0, 0, "Standard PC COM port"}, + {"PNP0501", &uart_ns8250_class, 0, 0, 0, 0, "16550A-compatible COM port"}, + {"PNP0502", &uart_ns8250_class, 0, 0, 0, 0, "Multiport serial device (non-intelligent 16550)"}, + {"PNP0510", &uart_ns8250_class, 0, 0, 0, 0, "Generic IRDA-compatible device"}, + {"PNP0511", &uart_ns8250_class, 0, 0, 0, 0, "Generic IRDA-compatible device"}, + {"WACF004", &uart_ns8250_class, 0, 0, 0, 0, "Wacom Tablet PC Screen"}, + {"WACF00E", &uart_ns8250_class, 0, 0, 0, 0, "Wacom Tablet PC Screen 00e"}, + {"FUJ02E5", &uart_ns8250_class, 0, 0, 0, 0, "Wacom Tablet at FuS Lifebook T"}, + {NULL, NULL, 0 , 0, 0, 0, NULL}, }; UART_ACPI_CLASS_AND_DEVICE(acpi_compat_data); #endif diff --git a/sys/dev/uart/uart_dev_pl011.c b/sys/dev/uart/uart_dev_pl011.c index a0d5a5b1c7e2..6afc693cd347 100644 --- a/sys/dev/uart/uart_dev_pl011.c +++ b/sys/dev/uart/uart_dev_pl011.c @@ -391,11 +391,19 @@ UART_FDT_CLASS_AND_DEVICE(fdt_compat_data); #endif #ifdef DEV_ACPI +static struct acpi_spcr_compat_data acpi_spcr_compat_data[] = { + { &uart_pl011_class, ACPI_DBG2_ARM_PL011 }, + { &uart_pl011_class, ACPI_DBG2_ARM_SBSA_GENERIC }, + { &uart_pl011_class, ACPI_DBG2_ARM_SBSA_32BIT }, + { NULL, 0 }, +}; +UART_ACPI_SPCR_CLASS(acpi_spcr_compat_data); + static struct acpi_uart_compat_data acpi_compat_data[] = { - {"ARMH0011", &uart_pl011_class, ACPI_DBG2_ARM_PL011, 2, 0, 0, 0, "uart pl011"}, - {"ARMHB000", &uart_pl011_class, ACPI_DBG2_ARM_SBSA_GENERIC, 2, 0, 0, 0, "uart pl011"}, - {"ARMHB000", &uart_pl011_class, ACPI_DBG2_ARM_SBSA_32BIT, 2, 0, 0, 0, "uart pl011"}, - {NULL, NULL, 0, 0, 0, 0, 0, NULL}, + {"ARMH0011", &uart_pl011_class, 2, 0, 0, 0, "uart pl011"}, + {"ARMHB000", &uart_pl011_class, 2, 0, 0, 0, "uart pl011"}, + {"ARMHB000", &uart_pl011_class, 2, 0, 0, 0, "uart pl011"}, + {NULL, NULL, 0, 0, 0, 0, NULL}, }; UART_ACPI_CLASS_AND_DEVICE(acpi_compat_data); #endif diff --git a/sys/dev/usb/misc/cp2112.c b/sys/dev/usb/misc/cp2112.c index d4776ca342cb..201a3ec51ce4 100644 --- a/sys/dev/usb/misc/cp2112.c +++ b/sys/dev/usb/misc/cp2112.c @@ -708,11 +708,12 @@ cp2112gpio_attach(device_t dev) } } - sc->busdev = gpiobus_attach_bus(dev); + sc->busdev = gpiobus_add_bus(dev); if (sc->busdev == NULL) { - device_printf(dev, "gpiobus_attach_bus failed\n"); + device_printf(dev, "gpiobus_add_bus failed\n"); goto detach; } + bus_attach_children(dev); return (0); detach: diff --git a/sys/dev/usb/net/if_ipheth.c b/sys/dev/usb/net/if_ipheth.c index f70113c53eb4..cfa800707391 100644 --- a/sys/dev/usb/net/if_ipheth.c +++ b/sys/dev/usb/net/if_ipheth.c @@ -55,6 +55,7 @@ #include <net/if_var.h> #include <dev/usb/usb.h> +#include <dev/usb/usb_cdc.h> #include <dev/usb/usbdi.h> #include <dev/usb/usbdi_util.h> #include "usbdevs.h" @@ -81,6 +82,9 @@ static uether_fn_t ipheth_start; static uether_fn_t ipheth_setmulti; static uether_fn_t ipheth_setpromisc; +static ipheth_consumer_t ipheth_consume_read; +static ipheth_consumer_t ipheth_consume_read_ncm; + #ifdef USB_DEBUG static int ipheth_debug = 0; @@ -96,7 +100,31 @@ static const struct usb_config ipheth_config[IPHETH_N_TRANSFER] = { .direction = UE_DIR_RX, .frames = IPHETH_RX_FRAMES_MAX, .bufsize = (IPHETH_RX_FRAMES_MAX * MCLBYTES), - .flags = {.short_frames_ok = 1,.short_xfer_ok = 1,.ext_buffer = 1,}, + .flags = {.short_frames_ok = 1, .short_xfer_ok = 1, .ext_buffer = 1,}, + .callback = ipheth_bulk_read_callback, + .timeout = 0, /* no timeout */ + }, + + [IPHETH_BULK_TX] = { + .type = UE_BULK, + .endpoint = UE_ADDR_ANY, + .direction = UE_DIR_TX, + .frames = IPHETH_TX_FRAMES_MAX, + .bufsize = (IPHETH_TX_FRAMES_MAX * IPHETH_BUF_SIZE), + .flags = {.force_short_xfer = 1,}, + .callback = ipheth_bulk_write_callback, + .timeout = IPHETH_TX_TIMEOUT, + }, +}; + +static const struct usb_config ipheth_config_ncm[IPHETH_N_TRANSFER] = { + [IPHETH_BULK_RX] = { + .type = UE_BULK, + .endpoint = UE_ADDR_ANY, + .direction = UE_DIR_RX, + .frames = 1, + .bufsize = IPHETH_RX_NCM_BUF_SIZE, + .flags = {.short_frames_ok = 1, .short_xfer_ok = 1,}, .callback = ipheth_bulk_read_callback, .timeout = 0, /* no timeout */ }, @@ -204,6 +232,21 @@ ipheth_get_mac_addr(struct ipheth_softc *sc) return (0); } +static bool +ipheth_enable_ncm(struct ipheth_softc *sc) +{ + struct usb_device_request req; + + req.bmRequestType = UT_WRITE_VENDOR_INTERFACE; + req.bRequest = IPHETH_CMD_ENABLE_NCM; + USETW(req.wValue, 0); + req.wIndex[0] = sc->sc_iface_no; + req.wIndex[1] = 0; + USETW(req.wLength, 0); + + return (usbd_do_request(sc->sc_ue.ue_udev, NULL, &req, NULL) == 0); +} + static int ipheth_probe(device_t dev) { @@ -221,6 +264,7 @@ ipheth_attach(device_t dev) struct ipheth_softc *sc = device_get_softc(dev); struct usb_ether *ue = &sc->sc_ue; struct usb_attach_arg *uaa = device_get_ivars(dev); + const struct usb_config *config; int error; sc->sc_iface_no = uaa->info.bIfaceIndex; @@ -235,18 +279,29 @@ ipheth_attach(device_t dev) device_printf(dev, "Cannot set alternate setting\n"); goto detach; } - error = usbd_transfer_setup(uaa->device, &sc->sc_iface_no, - sc->sc_xfer, ipheth_config, IPHETH_N_TRANSFER, sc, &sc->sc_mtx); - if (error) { - device_printf(dev, "Cannot setup USB transfers\n"); - goto detach; - } + ue->ue_sc = sc; ue->ue_dev = dev; ue->ue_udev = uaa->device; ue->ue_mtx = &sc->sc_mtx; ue->ue_methods = &ipheth_ue_methods; + if (ipheth_enable_ncm(sc)) { + config = ipheth_config_ncm; + sc->is_ncm = true; + sc->consume = &ipheth_consume_read_ncm; + } else { + config = ipheth_config; + sc->consume = &ipheth_consume_read; + } + + error = usbd_transfer_setup(uaa->device, &sc->sc_iface_no, sc->sc_xfer, + config, IPHETH_N_TRANSFER, sc, &sc->sc_mtx); + if (error) { + device_printf(dev, "Cannot setup USB transfers\n"); + goto detach; + } + error = ipheth_get_mac_addr(sc); if (error) { device_printf(dev, "Cannot get MAC address\n"); @@ -389,12 +444,9 @@ ipheth_bulk_write_callback(struct usb_xfer *xfer, usb_error_t error) int actlen; int aframes; - usbd_xfer_status(xfer, &actlen, NULL, &aframes, NULL); - - DPRINTFN(1, "\n"); - switch (USB_GET_STATE(xfer)) { case USB_ST_TRANSFERRED: + usbd_xfer_status(xfer, &actlen, NULL, &aframes, NULL); DPRINTFN(11, "transfer complete: %u bytes in %u frames\n", actlen, aframes); @@ -471,53 +523,40 @@ ipheth_bulk_read_callback(struct usb_xfer *xfer, usb_error_t error) uint8_t x; int actlen; int aframes; - int len; - - usbd_xfer_status(xfer, &actlen, NULL, &aframes, NULL); switch (USB_GET_STATE(xfer)) { case USB_ST_TRANSFERRED: - + usbd_xfer_status(xfer, &actlen, NULL, &aframes, NULL); DPRINTF("received %u bytes in %u frames\n", actlen, aframes); - for (x = 0; x != aframes; x++) { - m = sc->sc_rx_buf[x]; - sc->sc_rx_buf[x] = NULL; - len = usbd_xfer_frame_len(xfer, x); - - if (len < (int)(sizeof(struct ether_header) + - IPHETH_RX_ADJ)) { - m_freem(m); - continue; - } - - m_adj(m, IPHETH_RX_ADJ); - - /* queue up mbuf */ - uether_rxmbuf(&sc->sc_ue, m, len - IPHETH_RX_ADJ); - } + for (x = 0; x != aframes; x++) + sc->consume(xfer, x); /* FALLTHROUGH */ case USB_ST_SETUP: - - for (x = 0; x != IPHETH_RX_FRAMES_MAX; x++) { - if (sc->sc_rx_buf[x] == NULL) { - m = uether_newbuf(); - if (m == NULL) - goto tr_stall; - - /* cancel alignment for ethernet */ - m_adj(m, ETHER_ALIGN); - - sc->sc_rx_buf[x] = m; - } else { - m = sc->sc_rx_buf[x]; + if (!sc->is_ncm) { + for (x = 0; x != IPHETH_RX_FRAMES_MAX; x++) { + if (sc->sc_rx_buf[x] == NULL) { + m = uether_newbuf(); + if (m == NULL) + goto tr_stall; + + /* cancel alignment for ethernet */ + m_adj(m, ETHER_ALIGN); + + sc->sc_rx_buf[x] = m; + } else { + m = sc->sc_rx_buf[x]; + } + usbd_xfer_set_frame_data(xfer, x, m->m_data, m->m_len); } - - usbd_xfer_set_frame_data(xfer, x, m->m_data, m->m_len); + usbd_xfer_set_frames(xfer, x); + } else { + usbd_xfer_set_frame_len(xfer, 0, + IPHETH_RX_NCM_BUF_SIZE); + usbd_xfer_set_frames(xfer, 1); } - /* set number of frames and start hardware */ - usbd_xfer_set_frames(xfer, x); + usbd_transfer_submit(xfer); /* flush any received frames */ uether_rxflush(&sc->sc_ue); @@ -539,3 +578,86 @@ ipheth_bulk_read_callback(struct usb_xfer *xfer, usb_error_t error) break; } } + +static void +ipheth_consume_read(struct usb_xfer *xfer, int x) +{ + struct ipheth_softc *sc = usbd_xfer_softc(xfer); + struct mbuf *m = sc->sc_rx_buf[x]; + int len; + + sc->sc_rx_buf[x] = NULL; + len = usbd_xfer_frame_len(xfer, x); + + if (len < (int)(sizeof(struct ether_header) + IPHETH_RX_ADJ)) { + m_freem(m); + return; + } + + m_adj(m, IPHETH_RX_ADJ); + + /* queue up mbuf */ + uether_rxmbuf(&sc->sc_ue, m, len - IPHETH_RX_ADJ); +} + +static void +ipheth_consume_read_ncm(struct usb_xfer *xfer, int x) +{ + struct ipheth_softc *sc = usbd_xfer_softc(xfer); + struct usb_page_cache *pc = usbd_xfer_get_frame(xfer, 0); + struct ncm_data_cache ncm; + if_t ifp = uether_getifp(&sc->sc_ue); + struct mbuf *new_buf; + int i, actlen; + uint16_t dp_offset, dp_len; + + usbd_xfer_status(xfer, &actlen, NULL, NULL, NULL); + + if (actlen < IPHETH_NCM_HEADER_SIZE) + return; + + usbd_copy_out(pc, 0, &ncm.hdr, sizeof(ncm.hdr)); + + if (UGETDW(ncm.hdr.dwSignature) != 0x484D434E) + return; + + /* Dpt follows the hdr on iOS */ + if (UGETW(ncm.hdr.wDptIndex) != (int)(sizeof(struct usb_ncm16_hdr))) + return; + + usbd_copy_out(pc, UGETW(ncm.hdr.wDptIndex), &ncm.dpt, sizeof(ncm.dpt)); + + if (UGETDW(ncm.dpt.dwSignature) != 0x304D434E) + return; + + usbd_copy_out(pc, UGETW(ncm.hdr.wDptIndex) + sizeof(ncm.dpt), &ncm.dp, + sizeof(ncm.dp)); + + for (i = 0; i < IPHETH_NCM_DPT_DP_NUM; ++i) { + dp_offset = UGETW(ncm.dp[i].wFrameIndex); + dp_len = UGETW(ncm.dp[i].wFrameLength); + + /* (3.3.1 USB CDC NCM spec v1.0) */ + if (dp_offset == 0 && dp_len == 0) + break; + + if (dp_offset < IPHETH_NCM_HEADER_SIZE || dp_offset >= actlen || + actlen < (dp_len + dp_offset) || + dp_len < sizeof(struct ether_header)) { + if_inc_counter(ifp, IFCOUNTER_IERRORS, 1); + continue; + } + if (dp_len > (MCLBYTES - ETHER_ALIGN)) { + if_inc_counter(ifp, IFCOUNTER_IQDROPS, 1); + continue; + } + + new_buf = uether_newbuf(); + if (new_buf == NULL) { + if_inc_counter(ifp, IFCOUNTER_IQDROPS, 1); + continue; + } + usbd_copy_out(pc, dp_offset, new_buf->m_data, dp_len); + uether_rxmbuf(&sc->sc_ue, new_buf, dp_len); + } +} diff --git a/sys/dev/usb/net/if_iphethvar.h b/sys/dev/usb/net/if_iphethvar.h index 203bb96b6f22..d637e8f67d01 100644 --- a/sys/dev/usb/net/if_iphethvar.h +++ b/sys/dev/usb/net/if_iphethvar.h @@ -41,6 +41,7 @@ #define IPHETH_BUF_SIZE 1514 #define IPHETH_TX_TIMEOUT 5000 /* ms */ +#define IPHETH_RX_NCM_BUF_SIZE 65536 #define IPHETH_RX_FRAMES_MAX 1 #define IPHETH_TX_FRAMES_MAX 8 @@ -55,10 +56,20 @@ #define IPHETH_CTRL_TIMEOUT 5000 /* ms */ #define IPHETH_CMD_GET_MACADDR 0x00 +#define IPHETH_CMD_ENABLE_NCM 0x04 #define IPHETH_CMD_CARRIER_CHECK 0x45 #define IPHETH_CARRIER_ON 0x04 +#define IPHETH_NCM_DPT_DP_NUM 22 +#define IPHETH_NCM_DPT_HEADER_SIZE \ + (sizeof(struct usb_ncm16_dpt) + \ + IPHETH_NCM_DPT_DP_NUM * sizeof(struct usb_ncm16_dp)) +#define IPHETH_NCM_HEADER_SIZE \ + (sizeof(struct usb_ncm16_hdr) + IPHETH_NCM_DPT_HEADER_SIZE) + +typedef void (ipheth_consumer_t)(struct usb_xfer *xfer, int idx); + enum { IPHETH_BULK_TX, IPHETH_BULK_RX, @@ -76,6 +87,16 @@ struct ipheth_softc { uint8_t sc_data[IPHETH_CTRL_BUF_SIZE]; uint8_t sc_iface_no; uint8_t sc_carrier_on; + + bool is_ncm; + + ipheth_consumer_t *consume; +}; + +struct ncm_data_cache { + struct usb_ncm16_hdr hdr; + struct usb_ncm16_dpt dpt; + struct usb_ncm16_dp dp[IPHETH_NCM_DPT_DP_NUM]; }; #define IPHETH_LOCK(_sc) mtx_lock(&(_sc)->sc_mtx) diff --git a/sys/dev/watchdog/watchdog.c b/sys/dev/watchdog/watchdog.c index e6b6dc1eac70..e1b2e08c3f10 100644 --- a/sys/dev/watchdog/watchdog.c +++ b/sys/dev/watchdog/watchdog.c @@ -50,11 +50,20 @@ #include <sys/syscallsubr.h> /* kern_clock_gettime() */ -static int wd_set_pretimeout(int newtimeout, int disableiftoolong); +#ifdef COMPAT_FREEBSD14 +#define WDIOCPATPAT_14 _IOW('W', 42, u_int) /* pat the watchdog */ +#define WDIOC_SETTIMEOUT_14 _IOW('W', 43, int) /* set/reset the timer */ +#define WDIOC_GETTIMEOUT_14 _IOR('W', 44, int) /* get total timeout */ +#define WDIOC_GETTIMELEFT_14 _IOR('W', 45, int) /* get time left */ +#define WDIOC_GETPRETIMEOUT_14 _IOR('W', 46, int) /* get the pre-timeout */ +#define WDIOC_SETPRETIMEOUT_14 _IOW('W', 47, int) /* set the pre-timeout */ +#endif + +static int wd_set_pretimeout(sbintime_t newtimeout, int disableiftoolong); static void wd_timeout_cb(void *arg); static struct callout wd_pretimeo_handle; -static int wd_pretimeout; +static sbintime_t wd_pretimeout; static int wd_pretimeout_act = WD_SOFT_LOG; static struct callout wd_softtimeo_handle; @@ -63,6 +72,8 @@ static int wd_softtimer; /* true = use softtimer instead of hardware static int wd_softtimeout_act = WD_SOFT_LOG; /* action for the software timeout */ static struct cdev *wd_dev; +static volatile sbintime_t wd_last_sbt; /* last timeout value (sbt) */ +static sbintime_t wd_last_sbt_sysctl; /* last timeout value (sbt) */ static volatile u_int wd_last_u; /* last timeout value set by kern_do_pat */ static u_int wd_last_u_sysctl; /* last timeout value set by kern_do_pat */ static u_int wd_last_u_sysctl_secs; /* wd_last_u in seconds */ @@ -73,6 +84,8 @@ SYSCTL_UINT(_hw_watchdog, OID_AUTO, wd_last_u, CTLFLAG_RD, &wd_last_u_sysctl, 0, "Watchdog last update time"); SYSCTL_UINT(_hw_watchdog, OID_AUTO, wd_last_u_secs, CTLFLAG_RD, &wd_last_u_sysctl_secs, 0, "Watchdog last update time"); +SYSCTL_SBINTIME_MSEC(_hw_watchdog, OID_AUTO, wd_last_msecs, CTLFLAG_RD, + &wd_last_sbt_sysctl, "Watchdog last update time (milliseconds)"); static int wd_lastpat_valid = 0; static time_t wd_lastpat = 0; /* when the watchdog was last patted */ @@ -80,105 +93,94 @@ static time_t wd_lastpat = 0; /* when the watchdog was last patted */ /* Hook for external software watchdog to register for use if needed */ void (*wdog_software_attach)(void); -static void -pow2ns_to_ts(int pow2ns, struct timespec *ts) +/* Legacy interface to watchdog. */ +int +wdog_kern_pat(u_int utim) { - uint64_t ns; + sbintime_t sbt; - ns = 1ULL << pow2ns; - ts->tv_sec = ns / 1000000000ULL; - ts->tv_nsec = ns % 1000000000ULL; -} + if ((utim & WD_LASTVAL) != 0 && (utim & WD_INTERVAL) > 0) + return (EINVAL); -static int -pow2ns_to_ticks(int pow2ns) -{ - struct timeval tv; - struct timespec ts; + if ((utim & WD_LASTVAL) != 0) { + return (wdog_control(WD_CTRL_RESET)); + } - pow2ns_to_ts(pow2ns, &ts); - TIMESPEC_TO_TIMEVAL(&tv, &ts); - return (tvtohz(&tv)); + utim &= WD_INTERVAL; + if (utim == WD_TO_NEVER) + sbt = 0; + else + sbt = nstosbt(1 << utim); + + return (wdog_kern_pat_sbt(sbt)); } -static int -seconds_to_pow2ns(int seconds) +int +wdog_control(int ctrl) { - uint64_t power; - uint64_t ns; - uint64_t shifted; - - ns = ((uint64_t)seconds) * 1000000000ULL; - power = flsll(ns); - shifted = 1ULL << power; - if (shifted <= ns) { - power++; + /* Disable takes precedence */ + if (ctrl == WD_CTRL_DISABLE) { + wdog_kern_pat(0); } - return (power); + + if ((ctrl & WD_CTRL_RESET) != 0) { + wdog_kern_pat_sbt(wd_last_sbt); + } else if ((ctrl & WD_CTRL_ENABLE) != 0) { + wdog_kern_pat_sbt(wd_last_sbt); + } + + return (0); } int -wdog_kern_pat(u_int utim) +wdog_kern_pat_sbt(sbintime_t sbt) { - int error; - static int first = 1; - - if ((utim & WD_LASTVAL) != 0 && (utim & WD_INTERVAL) > 0) - return (EINVAL); - - if ((utim & WD_LASTVAL) != 0) { - /* - * if WD_LASTVAL is set, fill in the bits for timeout - * from the saved value in wd_last_u. - */ - MPASS((wd_last_u & ~WD_INTERVAL) == 0); - utim &= ~WD_LASTVAL; - utim |= wd_last_u; - } else { - /* - * Otherwise save the new interval. - * This can be zero (to disable the watchdog) - */ - wd_last_u = (utim & WD_INTERVAL); + sbintime_t error_sbt = 0; + int pow2ns = 0; + int error = 0; + static bool first = true; + + /* legacy uses power-of-2-nanoseconds time. */ + if (sbt != 0) { + pow2ns = flsl(sbttons(sbt)); + } + if (wd_last_sbt != sbt) { + wd_last_u = pow2ns; wd_last_u_sysctl = wd_last_u; - wd_last_u_sysctl_secs = pow2ns_to_ticks(wd_last_u) / hz; + wd_last_u_sysctl_secs = sbt / SBT_1S; + + wd_last_sbt = sbt; } - if ((utim & WD_INTERVAL) == WD_TO_NEVER) { - utim = 0; - /* Assume all is well; watchdog signals failure. */ - error = 0; - } else { - /* Assume no watchdog available; watchdog flags success */ + if (sbt != 0) error = EOPNOTSUPP; - } + if (wd_softtimer) { - if (utim == 0) { + if (sbt == 0) { callout_stop(&wd_softtimeo_handle); } else { - (void) callout_reset(&wd_softtimeo_handle, - pow2ns_to_ticks(utim), wd_timeout_cb, "soft"); + (void) callout_reset_sbt(&wd_softtimeo_handle, + sbt, 0, wd_timeout_cb, "soft", 0); } error = 0; } else { - EVENTHANDLER_INVOKE(watchdog_list, utim, &error); + EVENTHANDLER_INVOKE(watchdog_sbt_list, sbt, &error_sbt, &error); + EVENTHANDLER_INVOKE(watchdog_list, pow2ns, &error); } /* - * If we no hardware watchdog responded, we have not tried to + * If no hardware watchdog responded, we have not tried to * attach an external software watchdog, and one is available, * attach it now and retry. */ - if (error == EOPNOTSUPP && first && *wdog_software_attach != NULL) { + if (error == EOPNOTSUPP && first && wdog_software_attach != NULL) { (*wdog_software_attach)(); - EVENTHANDLER_INVOKE(watchdog_list, utim, &error); + EVENTHANDLER_INVOKE(watchdog_sbt_list, sbt, &error_sbt, &error); + EVENTHANDLER_INVOKE(watchdog_list, pow2ns, &error); } - first = 0; + first = false; + /* TODO: Print a (rate limited?) warning if error_sbt is too far away */ wd_set_pretimeout(wd_pretimeout, true); - /* - * If we were able to arm/strobe the watchdog, then - * update the last time it was strobed for WDIOC_GETTIMELEFT - */ if (!error) { struct timespec ts; @@ -189,6 +191,7 @@ wdog_kern_pat(u_int utim) wd_lastpat_valid = 1; } } + return (error); } @@ -265,16 +268,14 @@ wd_timeout_cb(void *arg) * current actual watchdog timeout. */ static int -wd_set_pretimeout(int newtimeout, int disableiftoolong) +wd_set_pretimeout(sbintime_t newtimeout, int disableiftoolong) { - u_int utime; - struct timespec utime_ts; - int timeout_ticks; + sbintime_t utime; + sbintime_t timeout_left; - utime = wdog_kern_last_timeout(); - pow2ns_to_ts(utime, &utime_ts); + utime = wdog_kern_last_timeout_sbt(); /* do not permit a pre-timeout >= than the timeout. */ - if (newtimeout >= utime_ts.tv_sec) { + if (newtimeout >= utime) { /* * If 'disableiftoolong' then just fall through * so as to disable the pre-watchdog @@ -292,7 +293,7 @@ wd_set_pretimeout(int newtimeout, int disableiftoolong) return 0; } - timeout_ticks = pow2ns_to_ticks(utime) - (hz*newtimeout); + timeout_left = utime - newtimeout; #if 0 printf("wd_set_pretimeout: " "newtimeout: %d, " @@ -306,8 +307,8 @@ wd_set_pretimeout(int newtimeout, int disableiftoolong) #endif /* We determined the value is sane, so reset the callout */ - (void) callout_reset(&wd_pretimeo_handle, - timeout_ticks, wd_timeout_cb, "pre"); + (void) callout_reset_sbt(&wd_pretimeo_handle, + timeout_left, 0, wd_timeout_cb, "pre", 0); wd_pretimeout = newtimeout; return 0; } @@ -316,6 +317,7 @@ static int wd_ioctl(struct cdev *dev __unused, u_long cmd, caddr_t data, int flags __unused, struct thread *td) { + sbintime_t sb; u_int u; time_t timeleft; int error; @@ -351,29 +353,55 @@ wd_ioctl(struct cdev *dev __unused, u_long cmd, caddr_t data, error = EINVAL; } break; - case WDIOC_GETPRETIMEOUT: - *(int *)data = (int)wd_pretimeout; +#ifdef COMPAT_FREEBSD14 + case WDIOC_GETPRETIMEOUT_14: + *(int *)data = (int)(wd_pretimeout / SBT_1S); break; - case WDIOC_SETPRETIMEOUT: - error = wd_set_pretimeout(*(int *)data, false); + case WDIOC_SETPRETIMEOUT_14: + error = wd_set_pretimeout(*(int *)data * SBT_1S, false); break; - case WDIOC_GETTIMELEFT: + case WDIOC_GETTIMELEFT_14: error = wd_get_time_left(td, &timeleft); if (error) break; *(int *)data = (int)timeleft; break; - case WDIOC_SETTIMEOUT: + case WDIOC_SETTIMEOUT_14: u = *(u_int *)data; - error = wdog_kern_pat(seconds_to_pow2ns(u)); + error = wdog_kern_pat_sbt(mstosbt(u * 1000ULL)); break; - case WDIOC_GETTIMEOUT: + case WDIOC_GETTIMEOUT_14: u = wdog_kern_last_timeout(); *(u_int *)data = u; break; - case WDIOCPATPAT: + case WDIOCPATPAT_14: error = wd_ioctl_patpat(data); break; +#endif + + /* New API */ + case WDIOC_CONTROL: + wdog_control(*(int *)data); + break; + case WDIOC_SETTIMEOUT: + sb = *(sbintime_t *)data; + error = wdog_kern_pat_sbt(sb); + break; + case WDIOC_GETTIMEOUT: + *(sbintime_t *)data = wdog_kern_last_timeout_sbt(); + break; + case WDIOC_GETTIMELEFT: + error = wd_get_time_left(td, &timeleft); + if (error) + break; + *(sbintime_t *)data = (sbintime_t)timeleft * SBT_1S; + break; + case WDIOC_GETPRETIMEOUT: + *(sbintime_t *)data = wd_pretimeout; + break; + case WDIOC_SETPRETIMEOUT: + error = wd_set_pretimeout(*(sbintime_t *)data, false); + break; default: error = ENOIOCTL; break; @@ -392,6 +420,12 @@ wdog_kern_last_timeout(void) return (wd_last_u); } +sbintime_t +wdog_kern_last_timeout_sbt(void) +{ + return (wd_last_sbt); +} + static struct cdevsw wd_cdevsw = { .d_version = D_VERSION, .d_ioctl = wd_ioctl, diff --git a/sys/fs/nfs/nfs_commonkrpc.c b/sys/fs/nfs/nfs_commonkrpc.c index 0ae3b94bef89..1e4e8506790f 100644 --- a/sys/fs/nfs/nfs_commonkrpc.c +++ b/sys/fs/nfs/nfs_commonkrpc.c @@ -239,6 +239,7 @@ static bool nfscl_use_gss[NFSV42_NPROCS] = { true, true, true, + true, }; /* diff --git a/sys/fs/nfs/nfs_commonsubs.c b/sys/fs/nfs/nfs_commonsubs.c index 67e33193ecec..7f5b29ca2085 100644 --- a/sys/fs/nfs/nfs_commonsubs.c +++ b/sys/fs/nfs/nfs_commonsubs.c @@ -187,7 +187,7 @@ struct nfsv4_opflag nfsv4_opflag[NFSV42_NOPS] = { { 0, 0, 0, 0, LK_EXCLUSIVE, 1, 1 }, /* Read Plus */ { 0, 1, 0, 0, LK_SHARED, 1, 0 }, /* Seek */ { 0, 0, 0, 0, LK_EXCLUSIVE, 1, 1 }, /* Write Same */ - { 0, 0, 0, 0, LK_EXCLUSIVE, 1, 1 }, /* Clone */ + { 2, 1, 1, 0, LK_SHARED, 1, 0 }, /* Clone */ { 0, 1, 0, 0, LK_SHARED, 1, 1 }, /* Getxattr */ { 0, 1, 1, 1, LK_EXCLUSIVE, 1, 1 }, /* Setxattr */ { 0, 1, 0, 0, LK_SHARED, 1, 1 }, /* Listxattrs */ @@ -219,7 +219,7 @@ NFSD_VNET_DEFINE_STATIC(u_char *, nfsrv_dnsname) = NULL; static int nfs_bigreply[NFSV42_NPROCS] = { 0, 0, 0, 1, 0, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, - 1, 0, 0, 1, 0, 0, 0, 0, 0, 0 }; + 1, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0 }; /* local functions */ static int nfsrv_skipace(struct nfsrv_descript *nd, int *acesizep); @@ -310,6 +310,7 @@ static struct { { NFSV4OP_LAYOUTERROR, 1, "LayoutError", 11, }, { NFSV4OP_VERIFY, 3, "AppendWrite", 11, }, { NFSV4OP_OPENATTR, 3, "OpenAttr", 8, }, + { NFSV4OP_SAVEFH, 5, "Clone", 5, }, }; /* @@ -319,7 +320,7 @@ static int nfs_bigrequest[NFSV42_NPROCS] = { 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, - 0, 1, 0 + 0, 1, 0, 0 }; /* @@ -648,7 +649,7 @@ nfscl_fillsattr(struct nfsrv_descript *nd, struct vattr *vap, NFSSETBIT_ATTRBIT(&attrbits, NFSATTRBIT_TIMECREATE); (void) nfsv4_fillattr(nd, vp->v_mount, vp, NULL, vap, NULL, 0, &attrbits, NULL, NULL, 0, 0, 0, 0, (uint64_t)0, NULL, - false, false, false); + false, false, false, 0); break; } } @@ -1302,7 +1303,7 @@ nfsv4_loadattr(struct nfsrv_descript *nd, vnode_t vp, struct nfsv3_pathconf *pc, struct statfs *sbp, struct nfsstatfs *sfp, struct nfsfsinfo *fsp, NFSACL_T *aclp, int compare, int *retcmpp, u_int32_t *leasep, u_int32_t *rderrp, bool *has_namedattrp, - NFSPROC_T *p, struct ucred *cred) + uint32_t *clone_blksizep, NFSPROC_T *p, struct ucred *cred) { u_int32_t *tl; int i = 0, j, k, l = 0, m, bitpos, attrsum = 0; @@ -1437,6 +1438,13 @@ nfsv4_loadattr(struct nfsrv_descript *nd, vnode_t vp, NFSCLRBIT_ATTRBIT(&checkattrbits, NFSATTRBIT_SYSTEM); } + /* Some filesystems do not support block cloning */ + if (vp == NULL || VOP_PATHCONF(vp, + _PC_CLONE_BLKSIZE, &has_pathconf) != 0) + has_pathconf = 0; + if (has_pathconf == 0) + NFSCLRBIT_ATTRBIT(&checkattrbits, + NFSATTRBIT_CLONEBLKSIZE); if (!NFSEQUAL_ATTRBIT(&retattrbits, &checkattrbits) || retnotsup) *retcmpp = NFSERR_NOTSAME; @@ -2374,6 +2382,23 @@ nfsv4_loadattr(struct nfsrv_descript *nd, vnode_t vp, if (compare && !(*retcmpp) && i != nfs_srvmaxio) *retcmpp = NFSERR_NOTSAME; break; + case NFSATTRBIT_CLONEBLKSIZE: + NFSM_DISSECT(tl, uint32_t *, NFSX_UNSIGNED); + if (compare) { + if (!(*retcmpp)) { + if (vp == NULL || VOP_PATHCONF(vp, + _PC_CLONE_BLKSIZE, &has_pathconf) + != 0) + has_pathconf = 0; + if (has_pathconf != + fxdr_unsigned(uint32_t, *tl)) + *retcmpp = NFSERR_NOTSAME; + } + } else if (clone_blksizep != NULL) { + *clone_blksizep = fxdr_unsigned(uint32_t, *tl); + } + attrsum += NFSX_UNSIGNED; + break; case NFSATTRBIT_CHANGEATTRTYPE: NFSM_DISSECT(tl, uint32_t *, NFSX_UNSIGNED); if (compare) { @@ -2648,7 +2673,7 @@ nfsv4_fillattr(struct nfsrv_descript *nd, struct mount *mp, vnode_t vp, nfsattrbit_t *attrbitp, struct ucred *cred, NFSPROC_T *p, int isdgram, int reterr, int supports_nfsv4acls, int at_root, uint64_t mounted_on_fileno, struct statfs *pnfssf, bool xattrsupp, bool has_hiddensystem, - bool has_namedattr) + bool has_namedattr, uint32_t clone_blksize) { int bitpos, retnum = 0; u_int32_t *tl; @@ -2771,6 +2796,9 @@ nfsv4_fillattr(struct nfsrv_descript *nd, struct mount *mp, vnode_t vp, NFSCLRBIT_ATTRBIT(&attrbits, NFSATTRBIT_HIDDEN); NFSCLRBIT_ATTRBIT(&attrbits, NFSATTRBIT_SYSTEM); } + if (clone_blksize == 0) + NFSCLRBIT_ATTRBIT(&attrbits, + NFSATTRBIT_CLONEBLKSIZE); retnum += nfsrv_putattrbit(nd, &attrbits); break; case NFSATTRBIT_TYPE: @@ -3249,6 +3277,11 @@ nfsv4_fillattr(struct nfsrv_descript *nd, struct mount *mp, vnode_t vp, } retnum += NFSX_UNSIGNED; break; + case NFSATTRBIT_CLONEBLKSIZE: + NFSM_BUILD(tl, uint32_t *, NFSX_UNSIGNED); + *tl = txdr_unsigned(clone_blksize); + retnum += NFSX_UNSIGNED; + break; default: printf("EEK! Bad V4 attribute bitpos=%d\n", bitpos); } diff --git a/sys/fs/nfs/nfs_var.h b/sys/fs/nfs/nfs_var.h index 54f60a753c50..61083ecf2d66 100644 --- a/sys/fs/nfs/nfs_var.h +++ b/sys/fs/nfs/nfs_var.h @@ -286,6 +286,8 @@ int nfsrvd_deallocate(struct nfsrv_descript *, int, vnode_t, struct nfsexstuff *); int nfsrvd_copy_file_range(struct nfsrv_descript *, int, vnode_t, vnode_t, struct nfsexstuff *, struct nfsexstuff *); +int nfsrvd_clone(struct nfsrv_descript *, int, + vnode_t, vnode_t, struct nfsexstuff *, struct nfsexstuff *); int nfsrvd_seek(struct nfsrv_descript *, int, vnode_t, struct nfsexstuff *); int nfsrvd_getxattr(struct nfsrv_descript *, int, @@ -341,7 +343,8 @@ int nfsv4_loadattr(struct nfsrv_descript *, vnode_t, struct nfsvattr *, struct nfsfh **, fhandle_t *, int, struct nfsv3_pathconf *, struct statfs *, struct nfsstatfs *, struct nfsfsinfo *, NFSACL_T *, - int, int *, u_int32_t *, u_int32_t *, bool *, NFSPROC_T *, struct ucred *); + int, int *, u_int32_t *, u_int32_t *, bool *, uint32_t *, NFSPROC_T *, + struct ucred *); int nfsv4_lock(struct nfsv4lock *, int, int *, struct mtx *, struct mount *); void nfsv4_unlock(struct nfsv4lock *, int); void nfsv4_relref(struct nfsv4lock *); @@ -397,7 +400,7 @@ void nfsrv_wcc(struct nfsrv_descript *, int, struct nfsvattr *, int, int nfsv4_fillattr(struct nfsrv_descript *, struct mount *, vnode_t, NFSACL_T *, struct vattr *, fhandle_t *, int, nfsattrbit_t *, struct ucred *, NFSPROC_T *, int, int, int, int, uint64_t, struct statfs *, bool, bool, - bool); + bool, uint32_t); void nfsrv_fillattr(struct nfsrv_descript *, struct nfsvattr *); struct mbuf *nfsrv_adj(struct mbuf *, int, int); void nfsrv_postopattr(struct nfsrv_descript *, int, struct nfsvattr *); @@ -517,10 +520,10 @@ int nfsrpc_lock(struct nfsrv_descript *, struct nfsmount *, vnode_t, u_int8_t *, int, struct nfscllockowner *, int, int, u_int64_t, u_int64_t, short, struct ucred *, NFSPROC_T *, int); int nfsrpc_statfs(vnode_t, struct nfsstatfs *, struct nfsfsinfo *, uint32_t *, - struct ucred *, NFSPROC_T *, struct nfsvattr *, int *); + uint32_t *, struct ucred *, NFSPROC_T *, struct nfsvattr *, int *); int nfsrpc_fsinfo(vnode_t, struct nfsfsinfo *, struct ucred *, NFSPROC_T *, struct nfsvattr *, int *); -int nfsrpc_pathconf(vnode_t, struct nfsv3_pathconf *, bool *, +int nfsrpc_pathconf(vnode_t, struct nfsv3_pathconf *, bool *, uint32_t *, struct ucred *, NFSPROC_T *, struct nfsvattr *, int *); int nfsrpc_renew(struct nfsclclient *, struct nfsclds *, struct ucred *, NFSPROC_T *); @@ -562,6 +565,8 @@ int nfsrpc_deallocate(vnode_t, off_t, off_t, struct nfsvattr *, int *, int nfsrpc_copy_file_range(vnode_t, off_t *, vnode_t, off_t *, size_t *, unsigned int, int *, struct nfsvattr *, int *, struct nfsvattr *, struct ucred *, bool, bool *); +int nfsrpc_clone(vnode_t, off_t *, vnode_t, off_t *, size_t *, bool, + int *, struct nfsvattr *, int *, struct nfsvattr *, struct ucred *); int nfsrpc_seek(vnode_t, off_t *, bool *, int, struct ucred *, struct nfsvattr *, int *); int nfsrpc_getextattr(vnode_t, const char *, struct uio *, ssize_t *, @@ -668,7 +673,7 @@ int nfscl_nget(mount_t, vnode_t, struct nfsfh *, NFSPROC_T *nfscl_getparent(NFSPROC_T *); void nfscl_start_renewthread(struct nfsclclient *); void nfscl_loadsbinfo(struct nfsmount *, struct nfsstatfs *, void *); -void nfscl_loadfsinfo (struct nfsmount *, struct nfsfsinfo *); +void nfscl_loadfsinfo(struct nfsmount *, struct nfsfsinfo *, uint32_t); void nfscl_delegreturn(struct nfscldeleg *, int, struct nfsmount *, struct ucred *, NFSPROC_T *); void nfsrvd_cbinit(int); @@ -737,7 +742,7 @@ int nfsvno_updfilerev(vnode_t, struct nfsvattr *, struct nfsrv_descript *, int nfsvno_fillattr(struct nfsrv_descript *, struct mount *, vnode_t, struct nfsvattr *, fhandle_t *, int, nfsattrbit_t *, struct ucred *, NFSPROC_T *, int, int, int, int, uint64_t, bool, bool, - bool); + bool, uint32_t); int nfsrv_sattr(struct nfsrv_descript *, vnode_t, struct nfsvattr *, nfsattrbit_t *, NFSACL_T *, NFSPROC_T *); int nfsv4_sattr(struct nfsrv_descript *, vnode_t, struct nfsvattr *, nfsattrbit_t *, diff --git a/sys/fs/nfs/nfsport.h b/sys/fs/nfs/nfsport.h index c30b46261df0..bd6107187966 100644 --- a/sys/fs/nfs/nfsport.h +++ b/sys/fs/nfs/nfsport.h @@ -442,10 +442,13 @@ /* Do a NFSv4 Openattr. */ #define NFSPROC_OPENATTR 70 +/* Do a NFSv4.2 Clone. */ +#define NFSPROC_CLONE 71 + /* * Must be defined as one higher than the last NFSv4.2 Proc# above. */ -#define NFSV42_NPROCS 71 +#define NFSV42_NPROCS 72 /* Value of NFSV42_NPROCS for old nfsstats structure. (Always 69) */ #define NFSV42_OLDNPROCS 69 @@ -477,7 +480,7 @@ struct nfsstatsv1 { uint64_t readlink_bios; uint64_t biocache_readdirs; uint64_t readdir_bios; - uint64_t rpccnt[NFSV42_NPROCS + 9]; + uint64_t rpccnt[NFSV42_NPROCS + 8]; uint64_t rpcretries; uint64_t srvrpccnt[NFSV42_NOPS + NFSV4OP_FAKENOPS + 15]; uint64_t srvlayouts; diff --git a/sys/fs/nfs/nfsproto.h b/sys/fs/nfs/nfsproto.h index cb5a80e8df73..d628108bdc1a 100644 --- a/sys/fs/nfs/nfsproto.h +++ b/sys/fs/nfs/nfsproto.h @@ -411,10 +411,13 @@ /* Do a NFSv4 Openattr. */ #define NFSPROC_OPENATTR 70 +/* Do a NFSv4.2 Clone. */ +#define NFSPROC_CLONE 71 + /* * Must be defined as one higher than the last NFSv4.2 Proc# above. */ -#define NFSV42_NPROCS 71 +#define NFSV42_NPROCS 72 /* Value of NFSV42_NPROCS for old nfsstats structure. (Always 69) */ #define NFSV42_OLDNPROCS 69 @@ -1194,6 +1197,7 @@ struct nfsv3_sattr { NFSATTRBM_LAYOUTBLKSIZE | \ NFSATTRBM_LAYOUTALIGNMENT | \ NFSATTRBM_SUPPATTREXCLCREAT | \ + NFSATTRBM_CLONEBLKSIZE | \ NFSATTRBM_CHANGEATTRTYPE | \ NFSATTRBM_XATTRSUPPORT) @@ -1242,7 +1246,8 @@ struct nfsv3_sattr { * NFSATTRBIT_NFSV42 - Attributes only supported by NFSv4.2. */ #define NFSATTRBIT_NFSV42_2 \ - (NFSATTRBM_CHANGEATTRTYPE | \ + (NFSATTRBM_CLONEBLKSIZE | \ + NFSATTRBM_CHANGEATTRTYPE | \ NFSATTRBM_XATTRSUPPORT | \ NFSATTRBM_MODEUMASK) @@ -1415,7 +1420,7 @@ struct nfsv3_sattr { /* * NFSGETATTRBIT_STATFS2 - bits 64<->95 */ -#define NFSGETATTRBIT_STATFS2 0 +#define NFSGETATTRBIT_STATFS2 (NFSATTRBM_CLONEBLKSIZE) /* * Set of attributes for the equivalent of an nfsv3 pathconf rpc. @@ -1438,7 +1443,7 @@ struct nfsv3_sattr { /* * NFSGETATTRBIT_PATHCONF2 - bits 64<->95 */ -#define NFSGETATTRBIT_PATHCONF2 0 +#define NFSGETATTRBIT_PATHCONF2 (NFSATTRBM_CLONEBLKSIZE) /* * Sets of attributes required by readdir and readdirplus. diff --git a/sys/fs/nfsclient/nfs_clcomsubs.c b/sys/fs/nfsclient/nfs_clcomsubs.c index bca0bdcd0df1..05963074e53d 100644 --- a/sys/fs/nfsclient/nfs_clcomsubs.c +++ b/sys/fs/nfsclient/nfs_clcomsubs.c @@ -272,7 +272,7 @@ nfsm_loadattr(struct nfsrv_descript *nd, struct nfsvattr *nap) if (nd->nd_flag & ND_NFSV4) { error = nfsv4_loadattr(nd, NULL, nap, NULL, NULL, 0, NULL, NULL, NULL, NULL, NULL, 0, NULL, NULL, NULL, NULL, NULL, - NULL); + NULL, NULL); } else if (nd->nd_flag & ND_NFSV3) { NFSM_DISSECT(fp, struct nfs_fattr *, NFSX_V3FATTR); nap->na_type = nfsv34tov_type(fp->fa_type); diff --git a/sys/fs/nfsclient/nfs_clport.c b/sys/fs/nfsclient/nfs_clport.c index b25d967982a1..704aeeeabdf2 100644 --- a/sys/fs/nfsclient/nfs_clport.c +++ b/sys/fs/nfsclient/nfs_clport.c @@ -828,7 +828,7 @@ nfscl_wcc_data(struct nfsrv_descript *nd, struct vnode *vp, == (ND_NFSV4 | ND_V4WCCATTR)) { error = nfsv4_loadattr(nd, NULL, &nfsva, NULL, NULL, 0, NULL, NULL, NULL, NULL, NULL, 0, - NULL, NULL, NULL, NULL, NULL, NULL); + NULL, NULL, NULL, NULL, NULL, NULL, NULL); if (error) return (error); /* @@ -963,7 +963,8 @@ nfscl_loadsbinfo(struct nfsmount *nmp, struct nfsstatfs *sfp, void *statfs) * Use the fsinfo stuff to update the mount point. */ void -nfscl_loadfsinfo(struct nfsmount *nmp, struct nfsfsinfo *fsp) +nfscl_loadfsinfo(struct nfsmount *nmp, struct nfsfsinfo *fsp, + uint32_t clone_blksize) { if ((nmp->nm_wsize == 0 || fsp->fs_wtpref < nmp->nm_wsize) && @@ -1003,6 +1004,14 @@ nfscl_loadfsinfo(struct nfsmount *nmp, struct nfsfsinfo *fsp) fsp->fs_maxfilesize < nmp->nm_maxfilesize) nmp->nm_maxfilesize = fsp->fs_maxfilesize; nmp->nm_mountp->mnt_stat.f_iosize = newnfs_iosize(nmp); + + /* + * Although ZFS reports a clone_blksize of 16Mbytes, + * 128Kbytes usually works, so set it to that. + */ + if (clone_blksize > 128 * 1024) + clone_blksize = 128 * 1024; + nmp->nm_cloneblksize = clone_blksize; nmp->nm_state |= NFSSTA_GOTFSINFO; } diff --git a/sys/fs/nfsclient/nfs_clrpcops.c b/sys/fs/nfsclient/nfs_clrpcops.c index 920fcf7b8c61..4ec621de2eff 100644 --- a/sys/fs/nfsclient/nfs_clrpcops.c +++ b/sys/fs/nfsclient/nfs_clrpcops.c @@ -225,6 +225,9 @@ static int nfsrpc_layoutgetres(struct nfsmount *, vnode_t, uint8_t *, static int nfsrpc_copyrpc(vnode_t, off_t, vnode_t, off_t, size_t *, nfsv4stateid_t *, nfsv4stateid_t *, struct nfsvattr *, int *, struct nfsvattr *, int *, bool, int *, struct ucred *, NFSPROC_T *); +static int nfsrpc_clonerpc(vnode_t, off_t, vnode_t, off_t, size_t *, bool, + nfsv4stateid_t *, nfsv4stateid_t *, struct nfsvattr *, int *, + struct nfsvattr *, int *, struct ucred *, NFSPROC_T *); static int nfsrpc_seekrpc(vnode_t, off_t *, nfsv4stateid_t *, bool *, int, struct nfsvattr *, int *, struct ucred *); static struct mbuf *nfsm_split(struct mbuf *, uint64_t); @@ -696,7 +699,7 @@ nfsrpc_openrpc(struct nfsmount *nmp, vnode_t vp, u_int8_t *nfhp, int fhlen, ("nfsrpc_openrpc: Getattr repstat")); error = nfsv4_loadattr(nd, NULL, &nfsva, NULL, NULL, 0, NULL, NULL, NULL, NULL, NULL, 0, - NULL, NULL, NULL, NULL, p, cred); + NULL, NULL, NULL, NULL, NULL, p, cred); if (error) goto nfsmout; } @@ -1355,7 +1358,7 @@ nfsrpc_getattrnovp(struct nfsmount *nmp, u_int8_t *fhp, int fhlen, int syscred, if ((nd->nd_flag & ND_NFSV4) != 0) error = nfsv4_loadattr(nd, NULL, nap, NULL, NULL, 0, NULL, NULL, NULL, NULL, NULL, 0, NULL, leasep, NULL, - NULL, NULL, NULL); + NULL, NULL, NULL, NULL); else error = nfsm_loadattr(nd, nap); } else @@ -3597,7 +3600,7 @@ nfsrpc_readdir(vnode_t vp, struct uio *uiop, nfsuint64 *cookiep, nfsva.na_mntonfileno = UINT64_MAX; error = nfsv4_loadattr(nd, NULL, &nfsva, NULL, NULL, 0, NULL, NULL, NULL, NULL, NULL, 0, - NULL, NULL, NULL, NULL, p, cred); + NULL, NULL, NULL, NULL, NULL, p, cred); if (error) { dotdotfileid = dotfileid; } else if (gotmnton) { @@ -3847,7 +3850,7 @@ nfsrpc_readdir(vnode_t vp, struct uio *uiop, nfsuint64 *cookiep, nfsva.na_mntonfileno = UINT64_MAX; error = nfsv4_loadattr(nd, NULL, &nfsva, NULL, NULL, 0, NULL, NULL, NULL, NULL, NULL, 0, - NULL, NULL, &rderr, NULL, p, cred); + NULL, NULL, &rderr, NULL, NULL, p, cred); if (error) goto nfsmout; NFSM_DISSECT(tl, u_int32_t *, NFSX_UNSIGNED); @@ -4072,7 +4075,7 @@ nfsrpc_readdirplus(vnode_t vp, struct uio *uiop, nfsuint64 *cookiep, nfsva.na_mntonfileno = UINT64_MAX; error = nfsv4_loadattr(nd, NULL, &nfsva, NULL, NULL, 0, NULL, NULL, NULL, NULL, NULL, 0, - NULL, NULL, NULL, NULL, p, cred); + NULL, NULL, NULL, NULL, NULL, p, cred); if (error) { dotdotfileid = dotfileid; } else if (gotmnton) { @@ -4346,7 +4349,7 @@ nfsrpc_readdirplus(vnode_t vp, struct uio *uiop, nfsuint64 *cookiep, nfsva.na_mntonfileno = 0xffffffff; error = nfsv4_loadattr(nd, NULL, &nfsva, &nfhp, NULL, 0, NULL, NULL, NULL, NULL, NULL, 0, - NULL, NULL, &rderr, NULL, p, cred); + NULL, NULL, &rderr, NULL, NULL, p, cred); if (error) goto nfsmout; } @@ -4981,8 +4984,8 @@ nfsmout: */ int nfsrpc_statfs(vnode_t vp, struct nfsstatfs *sbp, struct nfsfsinfo *fsp, - uint32_t *leasep, struct ucred *cred, NFSPROC_T *p, struct nfsvattr *nap, - int *attrflagp) + uint32_t *leasep, uint32_t *cloneblksizep, struct ucred *cred, NFSPROC_T *p, + struct nfsvattr *nap, int *attrflagp) { u_int32_t *tl = NULL; struct nfsrv_descript nfsd, *nd = &nfsd; @@ -4991,6 +4994,8 @@ nfsrpc_statfs(vnode_t vp, struct nfsstatfs *sbp, struct nfsfsinfo *fsp, int error; *attrflagp = 0; + if (cloneblksizep != NULL) + *cloneblksizep = 0; nmp = VFSTONFS(vp->v_mount); if (NFSHASNFSV4(nmp)) { /* @@ -5009,7 +5014,7 @@ nfsrpc_statfs(vnode_t vp, struct nfsstatfs *sbp, struct nfsfsinfo *fsp, if (nd->nd_repstat == 0) { error = nfsv4_loadattr(nd, NULL, nap, NULL, NULL, 0, NULL, NULL, sbp, fsp, NULL, 0, NULL, leasep, NULL, - NULL, p, cred); + NULL, cloneblksizep, p, cred); if (!error) { nmp->nm_fsid[0] = nap->na_filesid[0]; nmp->nm_fsid[1] = nap->na_filesid[1]; @@ -5063,7 +5068,8 @@ nfsmout: */ int nfsrpc_pathconf(vnode_t vp, struct nfsv3_pathconf *pc, bool *has_namedattrp, - struct ucred *cred, NFSPROC_T *p, struct nfsvattr *nap, int *attrflagp) + uint32_t *clone_blksizep, struct ucred *cred, NFSPROC_T *p, + struct nfsvattr *nap, int *attrflagp) { struct nfsrv_descript nfsd, *nd = &nfsd; struct nfsmount *nmp; @@ -5074,6 +5080,7 @@ nfsrpc_pathconf(vnode_t vp, struct nfsv3_pathconf *pc, bool *has_namedattrp, *has_namedattrp = false; *attrflagp = 0; + *clone_blksizep = 0; nmp = VFSTONFS(vp->v_mount); if (NFSHASNFSV4(nmp)) { np = VTONFS(vp); @@ -5100,7 +5107,7 @@ nfsrpc_pathconf(vnode_t vp, struct nfsv3_pathconf *pc, bool *has_namedattrp, if (nd->nd_repstat == 0) { error = nfsv4_loadattr(nd, NULL, nap, NULL, NULL, 0, pc, NULL, NULL, NULL, NULL, 0, NULL, NULL, NULL, - has_namedattrp, p, cred); + has_namedattrp, clone_blksizep, p, cred); if (!error) *attrflagp = 1; } else { @@ -5395,7 +5402,8 @@ nfsrpc_getacl(vnode_t vp, struct ucred *cred, NFSPROC_T *p, struct acl *aclp) return (error); if (!nd->nd_repstat) error = nfsv4_loadattr(nd, vp, NULL, NULL, NULL, 0, NULL, - NULL, NULL, NULL, aclp, 0, NULL, NULL, NULL, NULL, p, cred); + NULL, NULL, NULL, aclp, 0, NULL, NULL, NULL, NULL, NULL, p, + cred); else error = nd->nd_repstat; m_freem(nd->nd_mrep); @@ -5437,7 +5445,7 @@ nfsrpc_setaclrpc(vnode_t vp, struct ucred *cred, NFSPROC_T *p, NFSSETBIT_ATTRBIT(&attrbits, NFSATTRBIT_ACL); (void) nfsv4_fillattr(nd, vp->v_mount, vp, aclp, NULL, NULL, 0, &attrbits, NULL, NULL, 0, 0, 0, 0, (uint64_t)0, NULL, false, false, - false); + false, 0); error = nfscl_request(nd, vp, p, cred); if (error) return (error); @@ -8496,7 +8504,7 @@ nfsrpc_openlayoutrpc(struct nfsmount *nmp, vnode_t vp, u_int8_t *nfhp, if (*++tl == 0) { error = nfsv4_loadattr(nd, NULL, &nfsva, NULL, NULL, 0, NULL, NULL, NULL, NULL, NULL, 0, - NULL, NULL, NULL, NULL, p, cred); + NULL, NULL, NULL, NULL, NULL, p, cred); if (error != 0) goto nfsmout; if (ndp != NULL) { @@ -9168,6 +9176,199 @@ nfsmout: } /* + * nfs clone operation. + */ +int +nfsrpc_clone(vnode_t invp, off_t *inoffp, vnode_t outvp, + off_t *outoffp, size_t *lenp, bool toeof, int *inattrflagp, + struct nfsvattr *innap, int *outattrflagp, struct nfsvattr *outnap, + struct ucred *cred) +{ + int error, expireret = 0, retrycnt; + uint32_t clidrev = 0; + struct nfsmount *nmp = VFSTONFS(invp->v_mount); + struct nfsfh *innfhp = NULL, *outnfhp = NULL; + nfsv4stateid_t instateid, outstateid; + void *inlckp, *outlckp; + + if (nmp->nm_clp != NULL) + clidrev = nmp->nm_clp->nfsc_clientidrev; + innfhp = VTONFS(invp)->n_fhp; + outnfhp = VTONFS(outvp)->n_fhp; + retrycnt = 0; + do { + /* Get both stateids. */ + inlckp = NULL; + nfscl_getstateid(invp, innfhp->nfh_fh, innfhp->nfh_len, + NFSV4OPEN_ACCESSREAD, 0, NULL, curthread, &instateid, + &inlckp); + outlckp = NULL; + nfscl_getstateid(outvp, outnfhp->nfh_fh, outnfhp->nfh_len, + NFSV4OPEN_ACCESSWRITE, 0, NULL, curthread, &outstateid, + &outlckp); + + error = nfsrpc_clonerpc(invp, *inoffp, outvp, *outoffp, lenp, + toeof, &instateid, &outstateid, innap, inattrflagp, outnap, + outattrflagp, cred, curthread); + if (error == 0) { + *inoffp += *lenp; + *outoffp += *lenp; + } else if (error == NFSERR_STALESTATEID) + nfscl_initiate_recovery(nmp->nm_clp); + if (inlckp != NULL) + nfscl_lockderef(inlckp); + if (outlckp != NULL) + nfscl_lockderef(outlckp); + if (error == NFSERR_GRACE || error == NFSERR_STALESTATEID || + error == NFSERR_STALEDONTRECOVER || error == NFSERR_DELAY || + error == NFSERR_OLDSTATEID || error == NFSERR_BADSESSION) { + (void) nfs_catnap(PZERO, error, "nfs_cfr"); + } else if ((error == NFSERR_EXPIRED || (!NFSHASINT(nmp) && + error == NFSERR_BADSTATEID)) && clidrev != 0) { + expireret = nfscl_hasexpired(nmp->nm_clp, clidrev, + curthread); + } else if (error == NFSERR_BADSTATEID && NFSHASINT(nmp)) { + error = EIO; + } + retrycnt++; + } while (error == NFSERR_GRACE || error == NFSERR_DELAY || + error == NFSERR_STALESTATEID || error == NFSERR_BADSESSION || + error == NFSERR_STALEDONTRECOVER || + (error == NFSERR_OLDSTATEID && retrycnt < 20) || + ((error == NFSERR_EXPIRED || error == NFSERR_BADSTATEID) && + expireret == 0 && clidrev != 0 && retrycnt < 4)); + if (error != 0 && (retrycnt >= 4 || + error == NFSERR_STALESTATEID || error == NFSERR_BADSESSION || + error == NFSERR_STALEDONTRECOVER)) + error = EIO; + return (error); +} + +/* + * The clone RPC. + */ +static int +nfsrpc_clonerpc(vnode_t invp, off_t inoff, vnode_t outvp, off_t outoff, + size_t *lenp, bool toeof, nfsv4stateid_t *instateidp, + nfsv4stateid_t *outstateidp, struct nfsvattr *innap, int *inattrflagp, + struct nfsvattr *outnap, int *outattrflagp, struct ucred *cred, + NFSPROC_T *p) +{ + uint32_t *tl, *opcntp; + int error; + struct nfsrv_descript nfsd; + struct nfsrv_descript *nd = &nfsd; + struct nfsmount *nmp; + nfsattrbit_t attrbits; + struct vattr va; + uint64_t len; + + nmp = VFSTONFS(invp->v_mount); + *inattrflagp = *outattrflagp = 0; + len = *lenp; + if (len == 0) + return (0); + if (toeof) + len = 0; + nfscl_reqstart(nd, NFSPROC_CLONE, nmp, VTONFS(invp)->n_fhp->nfh_fh, + VTONFS(invp)->n_fhp->nfh_len, &opcntp, NULL, 0, 0, cred); + /* + * First do a Setattr of atime to the server's clock + * time. The FreeBSD "collective" was of the opinion + * that setting atime was necessary for this syscall. + * Do the Setattr before the Clone, so that it can be + * handled well if the server replies NFSERR_DELAY to + * the Setattr operation. + */ + if ((nmp->nm_mountp->mnt_flag & MNT_NOATIME) == 0) { + NFSM_BUILD(tl, uint32_t *, NFSX_UNSIGNED); + *tl = txdr_unsigned(NFSV4OP_SETATTR); + nfsm_stateidtom(nd, instateidp, NFSSTATEID_PUTSTATEID); + VATTR_NULL(&va); + va.va_atime.tv_sec = va.va_atime.tv_nsec = 0; + va.va_vaflags = VA_UTIMES_NULL; + nfscl_fillsattr(nd, &va, invp, 0, 0); + /* Bump opcnt from 7 to 8. */ + *opcntp = txdr_unsigned(8); + } + + /* Now Getattr the invp attributes. */ + NFSM_BUILD(tl, uint32_t *, NFSX_UNSIGNED); + *tl = txdr_unsigned(NFSV4OP_GETATTR); + NFSGETATTR_ATTRBIT(&attrbits); + nfsrv_putattrbit(nd, &attrbits); + + /* Set outvp. */ + NFSM_BUILD(tl, uint32_t *, NFSX_UNSIGNED); + *tl = txdr_unsigned(NFSV4OP_PUTFH); + (void)nfsm_fhtom(nmp, nd, VTONFS(outvp)->n_fhp->nfh_fh, + VTONFS(outvp)->n_fhp->nfh_len, 0); + + /* Do the Clone. */ + NFSM_BUILD(tl, uint32_t *, NFSX_UNSIGNED); + *tl = txdr_unsigned(NFSV4OP_CLONE); + nfsm_stateidtom(nd, instateidp, NFSSTATEID_PUTSTATEID); + nfsm_stateidtom(nd, outstateidp, NFSSTATEID_PUTSTATEID); + NFSM_BUILD(tl, uint32_t *, 3 * NFSX_HYPER + NFSX_UNSIGNED); + txdr_hyper(inoff, tl); tl += 2; + txdr_hyper(outoff, tl); tl += 2; + txdr_hyper(len, tl); tl += 2; + + /* Get the outvp attributes. */ + *tl = txdr_unsigned(NFSV4OP_GETATTR); + NFSWRITEGETATTR_ATTRBIT(&attrbits); + nfsrv_putattrbit(nd, &attrbits); + + error = nfscl_request(nd, invp, p, cred); + if (error != 0) + return (error); + /* Skip over the Setattr reply. */ + if ((nd->nd_flag & ND_NOMOREDATA) == 0 && + (nmp->nm_mountp->mnt_flag & MNT_NOATIME) == 0) { + NFSM_DISSECT(tl, uint32_t *, 2 * NFSX_UNSIGNED); + if (*(tl + 1) == 0) { + error = nfsrv_getattrbits(nd, &attrbits, NULL, NULL); + if (error != 0) + goto nfsmout; + } else + nd->nd_flag |= ND_NOMOREDATA; + } + if ((nd->nd_flag & ND_NOMOREDATA) == 0) { + /* Get the input file's attributes. */ + NFSM_DISSECT(tl, uint32_t *, 2 * NFSX_UNSIGNED); + if (*(tl + 1) == 0) { + error = nfsm_loadattr(nd, innap); + if (error != 0) + goto nfsmout; + *inattrflagp = 1; + } else + nd->nd_flag |= ND_NOMOREDATA; + } + /* Skip over return stat for PutFH. */ + if ((nd->nd_flag & ND_NOMOREDATA) == 0) { + NFSM_DISSECT(tl, uint32_t *, 2 * NFSX_UNSIGNED); + if (*++tl != 0) + nd->nd_flag |= ND_NOMOREDATA; + } + /* Skip over return stat for Clone. */ + if ((nd->nd_flag & ND_NOMOREDATA) == 0) + NFSM_DISSECT(tl, uint32_t *, 2 * NFSX_UNSIGNED); + if (nd->nd_repstat == 0) { + NFSM_DISSECT(tl, uint32_t *, 2 * NFSX_UNSIGNED); + error = nfsm_loadattr(nd, outnap); + if (error == 0) + *outattrflagp = NFS_LATTR_NOSHRINK; + } else { + *lenp = 0; + } + if (error == 0) + error = nd->nd_repstat; +nfsmout: + m_freem(nd->nd_mrep); + return (error); +} + +/* * Seek operation. */ int @@ -9724,13 +9925,13 @@ nfscl_statfs(struct vnode *vp, struct ucred *cred, NFSPROC_T *td) struct nfsstatfs sb; struct mount *mp; struct nfsmount *nmp; - uint32_t lease; + uint32_t clone_blksize, lease; int attrflag, error; mp = vp->v_mount; nmp = VFSTONFS(mp); - error = nfsrpc_statfs(vp, &sb, &fs, &lease, cred, td, &nfsva, - &attrflag); + error = nfsrpc_statfs(vp, &sb, &fs, &lease, &clone_blksize, cred, td, + &nfsva, &attrflag); if (attrflag != 0) (void) nfscl_loadattrcache(&vp, &nfsva, NULL, 0, 1); if (error == 0) { @@ -9739,7 +9940,7 @@ nfscl_statfs(struct vnode *vp, struct ucred *cred, NFSPROC_T *td) nmp->nm_clp->nfsc_renew = NFSCL_RENEW(lease); NFSUNLOCKCLSTATE(); mtx_lock(&nmp->nm_mtx); - nfscl_loadfsinfo(nmp, &fs); + nfscl_loadfsinfo(nmp, &fs, clone_blksize); nfscl_loadsbinfo(nmp, &sb, &mp->mnt_stat); mp->mnt_stat.f_iosize = newnfs_iosize(nmp); mtx_unlock(&nmp->nm_mtx); diff --git a/sys/fs/nfsclient/nfs_clstate.c b/sys/fs/nfsclient/nfs_clstate.c index 99a781640c53..aa9d01fc4632 100644 --- a/sys/fs/nfsclient/nfs_clstate.c +++ b/sys/fs/nfsclient/nfs_clstate.c @@ -3701,7 +3701,7 @@ nfscl_docb(struct nfsrv_descript *nd, NFSPROC_T *p) if (!error) (void) nfsv4_fillattr(nd, NULL, NULL, NULL, &va, NULL, 0, &rattrbits, NULL, p, 0, 0, 0, 0, - (uint64_t)0, NULL, false, false, false); + (uint64_t)0, NULL, false, false, false, 0); break; case NFSV4OP_CBRECALL: NFSCL_DEBUG(4, "cbrecall\n"); diff --git a/sys/fs/nfsclient/nfs_clvfsops.c b/sys/fs/nfsclient/nfs_clvfsops.c index 0bd05c03885b..5ea7eab07632 100644 --- a/sys/fs/nfsclient/nfs_clvfsops.c +++ b/sys/fs/nfsclient/nfs_clvfsops.c @@ -292,8 +292,10 @@ nfs_statfs(struct mount *mp, struct statfs *sbp) int error = 0, attrflag, gotfsinfo = 0, ret; struct nfsnode *np; char *fakefh; + uint32_t clone_blksize; td = curthread; + clone_blksize = 0; error = vfs_busy(mp, MBF_NOWAIT); if (error) @@ -337,8 +339,8 @@ nfs_statfs(struct mount *mp, struct statfs *sbp) } else mtx_unlock(&nmp->nm_mtx); if (!error) - error = nfsrpc_statfs(vp, &sb, &fs, NULL, td->td_ucred, td, - &nfsva, &attrflag); + error = nfsrpc_statfs(vp, &sb, &fs, NULL, &clone_blksize, + td->td_ucred, td, &nfsva, &attrflag); if ((nmp->nm_privflag & NFSMNTP_FAKEROOTFH) != 0 && error == NFSERR_WRONGSEC) { /* Cannot get new stats, so return what is in mnt_stat. */ @@ -375,7 +377,7 @@ nfs_statfs(struct mount *mp, struct statfs *sbp) if (!error) { mtx_lock(&nmp->nm_mtx); if (gotfsinfo || (nmp->nm_flag & NFSMNT_NFSV4)) - nfscl_loadfsinfo(nmp, &fs); + nfscl_loadfsinfo(nmp, &fs, clone_blksize); nfscl_loadsbinfo(nmp, &sb, sbp); sbp->f_iosize = newnfs_iosize(nmp); mtx_unlock(&nmp->nm_mtx); @@ -408,7 +410,7 @@ ncl_fsinfo(struct nfsmount *nmp, struct vnode *vp, struct ucred *cred, if (attrflag) (void) nfscl_loadattrcache(&vp, &nfsva, NULL, 0, 1); mtx_lock(&nmp->nm_mtx); - nfscl_loadfsinfo(nmp, &fs); + nfscl_loadfsinfo(nmp, &fs, 0); mtx_unlock(&nmp->nm_mtx); } return (error); diff --git a/sys/fs/nfsclient/nfs_clvnops.c b/sys/fs/nfsclient/nfs_clvnops.c index fa451887e73e..6824ee6ef13d 100644 --- a/sys/fs/nfsclient/nfs_clvnops.c +++ b/sys/fs/nfsclient/nfs_clvnops.c @@ -4027,31 +4027,51 @@ nfs_copy_file_range(struct vop_copy_file_range_args *ap) struct vattr va, *vap; struct uio io; struct nfsmount *nmp; + struct nfsnode *np; size_t len, len2; ssize_t r; int error, inattrflag, outattrflag, ret, ret2, invp_lock; off_t inoff, outoff; - bool consecutive, must_commit, tryoutcred; + bool consecutive, must_commit, onevp, toeof, tryclone, tryoutcred; + bool mustclone; /* * NFSv4.2 Copy is not permitted for infile == outfile. + * The NFSv4.2 Clone operation does work on non-overlapping + * byte ranges in the same file, but only if offsets + * (and len if not to EOF) are aligned properly. * TODO: copy_file_range() between multiple NFS mountpoints + * --> This is not possible now, since each mount appears to + * the NFSv4.n server as a separate client. */ - if (invp == outvp || invp->v_mount != outvp->v_mount) { + if ((invp == outvp && (ap->a_flags & COPY_FILE_RANGE_CLONE) == 0) || + (invp != outvp && invp->v_mount != outvp->v_mount)) { generic_copy: return (ENOSYS); } - - invp_lock = LK_SHARED; + if (invp == outvp) { + onevp = true; + invp_lock = LK_EXCLUSIVE; + } else { + onevp = false; + invp_lock = LK_SHARED; + } + mustclone = false; + if (onevp || (ap->a_flags & COPY_FILE_RANGE_CLONE) != 0) + mustclone = true; relock: + inoff = *ap->a_inoffp; + outoff = *ap->a_outoffp; - /* Lock both vnodes, avoiding risk of deadlock. */ + /* Lock vnode(s), avoiding risk of deadlock. */ do { mp = NULL; error = vn_start_write(outvp, &mp, V_WAIT); if (error == 0) { error = vn_lock(outvp, LK_EXCLUSIVE); if (error == 0) { + if (onevp) + break; error = vn_lock(invp, invp_lock | LK_NOWAIT); if (error == 0) break; @@ -4071,16 +4091,24 @@ relock: return (error); /* - * More reasons to avoid nfs copy: not NFSv4.2, or explicitly - * disabled. + * More reasons to avoid nfs copy/clone: not NFSv4.2, explicitly + * disabled or requires cloning and unable to clone. + * Only clone if the clone_blksize attribute is supported + * and the clone_blksize is greater than 0. + * Alignment of offsets and length will be checked later. */ nmp = VFSTONFS(invp->v_mount); + np = VTONFS(invp); mtx_lock(&nmp->nm_mtx); + if ((nmp->nm_privflag & NFSMNTP_NOCOPY) != 0) + mustclone = true; if (!NFSHASNFSV4(nmp) || nmp->nm_minorvers < NFSV42_MINORVERSION || - (nmp->nm_privflag & NFSMNTP_NOCOPY) != 0) { + (mustclone && (!NFSISSET_ATTRBIT(&np->n_vattr.na_suppattr, + NFSATTRBIT_CLONEBLKSIZE) || nmp->nm_cloneblksize == 0))) { mtx_unlock(&nmp->nm_mtx); VOP_UNLOCK(invp); - VOP_UNLOCK(outvp); + if (!onevp) + VOP_UNLOCK(outvp); /* For onevp, same as invp. */ if (mp != NULL) vn_finished_write(mp); goto generic_copy; @@ -4111,6 +4139,8 @@ relock: invp_obj = invp->v_object; if (invp_obj != NULL && vm_object_mightbedirty(invp_obj)) { if (invp_lock != LK_EXCLUSIVE) { + KASSERT(!onevp, ("nfs_copy_file_range: " + "invp_lock LK_SHARED for onevp")); invp_lock = LK_EXCLUSIVE; VOP_UNLOCK(invp); VOP_UNLOCK(outvp); @@ -4134,10 +4164,10 @@ relock: else consecutive = false; mtx_unlock(&nmp->nm_mtx); - inoff = *ap->a_inoffp; - outoff = *ap->a_outoffp; tryoutcred = true; must_commit = false; + toeof = false; + if (error == 0) { vap = &VTONFS(invp)->n_vattr.na_vattr; error = VOP_GETATTR(invp, vap, ap->a_incred); @@ -4169,29 +4199,63 @@ relock: if (error == 0 && ret != 0) error = ret; } - } else if (inoff + len > vap->va_size) + } else if (inoff + len >= vap->va_size) { + toeof = true; *ap->a_lenp = len = vap->va_size - inoff; + } } else error = 0; } /* + * For cloning, the offsets must be clone blksize aligned and + * the len must be blksize aligned unless it goes to EOF on + * the input file. + */ + tryclone = false; + if (len > 0) { + if (error == 0 && NFSISSET_ATTRBIT(&np->n_vattr.na_suppattr, + NFSATTRBIT_CLONEBLKSIZE) && nmp->nm_cloneblksize != 0 && + (inoff % nmp->nm_cloneblksize) == 0 && + (outoff % nmp->nm_cloneblksize) == 0 && + (toeof || (len % nmp->nm_cloneblksize) == 0)) + tryclone = true; + else if (mustclone) + error = ENOSYS; + } + + /* * len will be set to 0 upon a successful Copy RPC. - * As such, this only loops when the Copy RPC needs to be retried. + * As such, this only loops when the Copy/Clone RPC needs to be retried. */ while (len > 0 && error == 0) { inattrflag = outattrflag = 0; len2 = len; - if (tryoutcred) - error = nfsrpc_copy_file_range(invp, ap->a_inoffp, - outvp, ap->a_outoffp, &len2, ap->a_flags, - &inattrflag, &innfsva, &outattrflag, &outnfsva, - ap->a_outcred, consecutive, &must_commit); - else - error = nfsrpc_copy_file_range(invp, ap->a_inoffp, - outvp, ap->a_outoffp, &len2, ap->a_flags, - &inattrflag, &innfsva, &outattrflag, &outnfsva, - ap->a_incred, consecutive, &must_commit); + if (tryclone) { + if (tryoutcred) + error = nfsrpc_clone(invp, ap->a_inoffp, outvp, + ap->a_outoffp, &len2, toeof, &inattrflag, + &innfsva, &outattrflag, &outnfsva, + ap->a_outcred); + else + error = nfsrpc_clone(invp, ap->a_inoffp, outvp, + ap->a_outoffp, &len2, toeof, &inattrflag, + &innfsva, &outattrflag, &outnfsva, + ap->a_incred); + } else { + if (tryoutcred) + error = nfsrpc_copy_file_range(invp, + ap->a_inoffp, outvp, ap->a_outoffp, &len2, + ap->a_flags, &inattrflag, &innfsva, + &outattrflag, &outnfsva, + ap->a_outcred, consecutive, &must_commit); + else + error = nfsrpc_copy_file_range(invp, + ap->a_inoffp, outvp, ap->a_outoffp, &len2, + ap->a_flags, &inattrflag, &innfsva, + &outattrflag, &outnfsva, + ap->a_incred, consecutive, &must_commit); + } if (inattrflag != 0) ret = nfscl_loadattrcache(&invp, &innfsva, NULL, 0, 1); if (outattrflag != 0) @@ -4230,6 +4294,13 @@ relock: /* Try again with incred. */ tryoutcred = false; error = 0; + } else if (tryclone && error != 0) { + if (mustclone) { + error = ENOSYS; + } else { + tryclone = false; + error = 0; + } } if (error == NFSERR_STALEWRITEVERF) { /* @@ -4243,11 +4314,12 @@ relock: } } VOP_UNLOCK(invp); - VOP_UNLOCK(outvp); + if (!onevp) + VOP_UNLOCK(outvp); /* For onevp, same as invp. */ if (mp != NULL) vn_finished_write(mp); if (error == NFSERR_NOTSUPP || error == NFSERR_OFFLOADNOREQS || - error == NFSERR_ACCES) { + error == NFSERR_ACCES || error == ENOSYS) { /* * Unlike the NFSv4.2 Copy, vn_generic_copy_file_range() can * use a_incred for the read and a_outcred for the write, so @@ -4255,7 +4327,7 @@ relock: * For NFSERR_NOTSUPP and NFSERR_OFFLOADNOREQS, the Copy can * never succeed, so disable it. */ - if (error != NFSERR_ACCES) { + if (error != NFSERR_ACCES && error != ENOSYS) { /* Can never do Copy on this mount. */ mtx_lock(&nmp->nm_mtx); nmp->nm_privflag |= NFSMNTP_NOCOPY; @@ -4596,6 +4668,7 @@ nfs_pathconf(struct vop_pathconf_args *ap) struct nfsmount *nmp; struct thread *td = curthread; off_t off; + uint32_t clone_blksize; bool eof, has_namedattr, named_enabled; int attrflag, error; struct nfsnode *np; @@ -4604,19 +4677,22 @@ nfs_pathconf(struct vop_pathconf_args *ap) np = VTONFS(vp); named_enabled = false; has_namedattr = false; + clone_blksize = 0; if ((NFS_ISV34(vp) && (ap->a_name == _PC_LINK_MAX || ap->a_name == _PC_NAME_MAX || ap->a_name == _PC_CHOWN_RESTRICTED || ap->a_name == _PC_NO_TRUNC)) || (NFS_ISV4(vp) && (ap->a_name == _PC_ACL_NFS4 || - ap->a_name == _PC_HAS_NAMEDATTR))) { + ap->a_name == _PC_HAS_NAMEDATTR || + ap->a_name == _PC_CLONE_BLKSIZE))) { /* * Since only the above 4 a_names are returned by the NFSv3 * Pathconf RPC, there is no point in doing it for others. * For NFSv4, the Pathconf RPC (actually a Getattr Op.) can - * be used for _PC_ACL_NFS4 and _PC_HAS_NAMEDATTR as well. + * be used for _PC_ACL_NFS4, _PC_HAS_NAMEDATTR and + * _PC_CLONE_BLKSIZE as well. */ - error = nfsrpc_pathconf(vp, &pc, &has_namedattr, td->td_ucred, - td, &nfsva, &attrflag); + error = nfsrpc_pathconf(vp, &pc, &has_namedattr, &clone_blksize, + td->td_ucred, td, &nfsva, &attrflag); if (attrflag != 0) (void) nfscl_loadattrcache(&vp, &nfsva, NULL, 0, 1); if (error != 0) @@ -4771,6 +4847,9 @@ nfs_pathconf(struct vop_pathconf_args *ap) else *ap->a_retval = 0; break; + case _PC_CLONE_BLKSIZE: + *ap->a_retval = clone_blksize; + break; default: error = vop_stdpathconf(ap); diff --git a/sys/fs/nfsclient/nfsmount.h b/sys/fs/nfsclient/nfsmount.h index 37b84a015dab..ef876dd30e59 100644 --- a/sys/fs/nfsclient/nfsmount.h +++ b/sys/fs/nfsclient/nfsmount.h @@ -87,6 +87,7 @@ struct nfsmount { /* unclipped, wraps to 0 */ struct __rpc_client *nm_aconn[NFS_MAXNCONN - 1]; /* Additional nconn */ /* Locked via nm_sockreq.nr_mtx */ + uint32_t nm_cloneblksize; /* Block cloning alignment */ u_int16_t nm_krbnamelen; /* Krb5 host principal, if any */ u_int16_t nm_dirpathlen; /* and mount dirpath, for V4 */ u_int16_t nm_srvkrbnamelen; /* and the server's target name */ diff --git a/sys/fs/nfsserver/nfs_nfsdport.c b/sys/fs/nfsserver/nfs_nfsdport.c index 8c427c66c156..b2966934f9b7 100644 --- a/sys/fs/nfsserver/nfs_nfsdport.c +++ b/sys/fs/nfsserver/nfs_nfsdport.c @@ -2113,7 +2113,8 @@ nfsvno_fillattr(struct nfsrv_descript *nd, struct mount *mp, struct vnode *vp, struct nfsvattr *nvap, fhandle_t *fhp, int rderror, nfsattrbit_t *attrbitp, struct ucred *cred, struct thread *p, int isdgram, int reterr, int supports_nfsv4acls, int at_root, uint64_t mounted_on_fileno, - bool xattrsupp, bool has_hiddensystem, bool has_namedattr) + bool xattrsupp, bool has_hiddensystem, bool has_namedattr, + uint32_t clone_blksize) { struct statfs *sf; int error; @@ -2130,9 +2131,11 @@ nfsvno_fillattr(struct nfsrv_descript *nd, struct mount *mp, struct vnode *vp, sf = NULL; } } + error = nfsv4_fillattr(nd, mp, vp, NULL, &nvap->na_vattr, fhp, rderror, attrbitp, cred, p, isdgram, reterr, supports_nfsv4acls, at_root, - mounted_on_fileno, sf, xattrsupp, has_hiddensystem, has_namedattr); + mounted_on_fileno, sf, xattrsupp, has_hiddensystem, has_namedattr, + clone_blksize); free(sf, M_TEMP); NFSEXITCODE2(0, nd); return (error); @@ -2441,7 +2444,7 @@ nfsrvd_readdirplus(struct nfsrv_descript *nd, int isdgram, struct vnode *vp, struct nfsexstuff *exp) { struct dirent *dp; - u_int32_t *tl; + uint32_t clone_blksize, *tl; int dirlen; char *cpos, *cend, *rbuf; struct vnode *nvp; @@ -2943,6 +2946,7 @@ again: xattrsupp = false; has_hiddensystem = false; has_namedattr = false; + clone_blksize = 0; if (nvp != NULL) { supports_nfsv4acls = nfs_supportsnfsv4acls(nvp); @@ -2966,6 +2970,11 @@ again: &pathval) != 0) pathval = 0; has_namedattr = pathval > 0; + pathval = 0; + if (VOP_PATHCONF(nvp, _PC_CLONE_BLKSIZE, + &pathval) != 0) + pathval = 0; + clone_blksize = pathval; NFSVOPUNLOCK(nvp); } else supports_nfsv4acls = 0; @@ -2986,14 +2995,16 @@ again: nd->nd_cred, p, isdgram, 0, supports_nfsv4acls, at_root, mounted_on_fileno, xattrsupp, - has_hiddensystem, has_namedattr); + has_hiddensystem, has_namedattr, + clone_blksize); } else { dirlen += nfsvno_fillattr(nd, new_mp, nvp, nvap, &nfh, r, &attrbits, nd->nd_cred, p, isdgram, 0, supports_nfsv4acls, at_root, mounted_on_fileno, xattrsupp, - has_hiddensystem, has_namedattr); + has_hiddensystem, has_namedattr, + clone_blksize); } if (nvp != NULL) vrele(nvp); @@ -3740,6 +3751,7 @@ nfsrv_v4rootexport(void *argp, struct ucred *cred, struct thread *p) NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, nfsexargp->fspec); if ((error = namei(&nd)) != 0) goto out; + NDFREE_PNBUF(&nd); error = nfsvno_getfh(nd.ni_vp, &fh, p); vrele(nd.ni_vp); if (!error) { @@ -5690,7 +5702,8 @@ nfsrv_writedsdorpc(struct nfsmount *nmp, fhandle_t *fhp, off_t off, int len, if ((nd->nd_flag & (ND_NOMOREDATA | ND_NFSV4 | ND_V4WCCATTR)) == (ND_NFSV4 | ND_V4WCCATTR)) { error = nfsv4_loadattr(nd, NULL, nap, NULL, NULL, 0, NULL, NULL, - NULL, NULL, NULL, 0, NULL, NULL, NULL, NULL, NULL, NULL); + NULL, NULL, NULL, 0, NULL, NULL, NULL, NULL, NULL, NULL, + NULL); NFSD_DEBUG(4, "nfsrv_writedsdorpc: wcc attr=%d\n", error); if (error != 0) goto nfsmout; @@ -5721,7 +5734,8 @@ nfsrv_writedsdorpc(struct nfsmount *nmp, fhandle_t *fhp, off_t off, int len, if (error == 0) { NFSM_DISSECT(tl, uint32_t *, 2 * NFSX_UNSIGNED); error = nfsv4_loadattr(nd, NULL, nap, NULL, NULL, 0, NULL, NULL, - NULL, NULL, NULL, 0, NULL, NULL, NULL, NULL, NULL, NULL); + NULL, NULL, NULL, 0, NULL, NULL, NULL, NULL, NULL, NULL, + NULL); } NFSD_DEBUG(4, "nfsrv_writedsdorpc: aft loadattr=%d\n", error); nfsmout: @@ -5887,7 +5901,8 @@ nfsrv_allocatedsdorpc(struct nfsmount *nmp, fhandle_t *fhp, off_t off, if (nd->nd_repstat == 0) { NFSM_DISSECT(tl, uint32_t *, 2 * NFSX_UNSIGNED); error = nfsv4_loadattr(nd, NULL, nap, NULL, NULL, 0, NULL, NULL, - NULL, NULL, NULL, 0, NULL, NULL, NULL, NULL, NULL, NULL); + NULL, NULL, NULL, 0, NULL, NULL, NULL, NULL, NULL, NULL, + NULL); } else error = nd->nd_repstat; NFSD_DEBUG(4, "nfsrv_allocatedsdorpc: aft loadattr=%d\n", error); @@ -6054,7 +6069,8 @@ nfsrv_deallocatedsdorpc(struct nfsmount *nmp, fhandle_t *fhp, off_t off, if ((nd->nd_flag & (ND_NOMOREDATA | ND_NFSV4 | ND_V4WCCATTR)) == (ND_NFSV4 | ND_V4WCCATTR)) { error = nfsv4_loadattr(nd, NULL, nap, NULL, NULL, 0, NULL, NULL, - NULL, NULL, NULL, 0, NULL, NULL, NULL, NULL, NULL, NULL); + NULL, NULL, NULL, 0, NULL, NULL, NULL, NULL, NULL, NULL, + NULL); NFSD_DEBUG(4, "nfsrv_deallocatedsdorpc: wcc attr=%d\n", error); if (error != 0) goto nfsmout; @@ -6068,7 +6084,8 @@ nfsrv_deallocatedsdorpc(struct nfsmount *nmp, fhandle_t *fhp, off_t off, if (nd->nd_repstat == 0) { NFSM_DISSECT(tl, uint32_t *, 2 * NFSX_UNSIGNED); error = nfsv4_loadattr(nd, NULL, nap, NULL, NULL, 0, NULL, NULL, - NULL, NULL, NULL, 0, NULL, NULL, NULL, NULL, NULL, NULL); + NULL, NULL, NULL, 0, NULL, NULL, NULL, NULL, NULL, NULL, + NULL); } else error = nd->nd_repstat; NFSD_DEBUG(4, "nfsrv_deallocatedsdorpc: aft loadattr=%d\n", error); @@ -6216,7 +6233,8 @@ nfsrv_setattrdsdorpc(fhandle_t *fhp, struct ucred *cred, NFSPROC_T *p, if ((nd->nd_flag & (ND_NOMOREDATA | ND_NFSV4 | ND_V4WCCATTR)) == (ND_NFSV4 | ND_V4WCCATTR)) { error = nfsv4_loadattr(nd, NULL, dsnap, NULL, NULL, 0, NULL, - NULL, NULL, NULL, NULL, 0, NULL, NULL, NULL, NULL, NULL, NULL); + NULL, NULL, NULL, NULL, 0, NULL, NULL, NULL, NULL, NULL, + NULL, NULL); NFSD_DEBUG(4, "nfsrv_setattrdsdorpc: wcc attr=%d\n", error); if (error != 0) goto nfsmout; @@ -6241,7 +6259,7 @@ nfsrv_setattrdsdorpc(fhandle_t *fhp, struct ucred *cred, NFSPROC_T *p, NFSM_DISSECT(tl, uint32_t *, 2 * NFSX_UNSIGNED); error = nfsv4_loadattr(nd, NULL, dsnap, NULL, NULL, 0, NULL, NULL, NULL, NULL, NULL, 0, NULL, NULL, NULL, NULL, NULL, - NULL); + NULL, NULL); } NFSD_DEBUG(4, "nfsrv_setattrdsdorpc: aft setattr loadattr=%d\n", error); nfsmout: @@ -6386,7 +6404,7 @@ nfsrv_setacldsdorpc(fhandle_t *fhp, struct ucred *cred, NFSPROC_T *p, * the same type (VREG). */ nfsv4_fillattr(nd, NULL, vp, aclp, NULL, NULL, 0, &attrbits, NULL, - NULL, 0, 0, 0, 0, 0, NULL, false, false, false); + NULL, 0, 0, 0, 0, 0, NULL, false, false, false, 0); error = newnfs_request(nd, nmp, NULL, &nmp->nm_sockreq, NULL, p, cred, NFS_PROG, NFS_VER4, NULL, 1, NULL, NULL); if (error != 0) { @@ -6530,7 +6548,7 @@ nfsrv_getattrdsrpc(fhandle_t *fhp, struct ucred *cred, NFSPROC_T *p, if (nd->nd_repstat == 0) { error = nfsv4_loadattr(nd, NULL, nap, NULL, NULL, 0, NULL, NULL, NULL, NULL, NULL, 0, NULL, NULL, NULL, - NULL, NULL, NULL); + NULL, NULL, NULL, NULL); /* * We can only save the updated values in the extended * attribute if the vp is exclusively locked. diff --git a/sys/fs/nfsserver/nfs_nfsdserv.c b/sys/fs/nfsserver/nfs_nfsdserv.c index 9eebcda548c6..a881968b03be 100644 --- a/sys/fs/nfsserver/nfs_nfsdserv.c +++ b/sys/fs/nfsserver/nfs_nfsdserv.c @@ -253,6 +253,7 @@ nfsrvd_getattr(struct nfsrv_descript *nd, int isdgram, size_t atsiz; long pathval; bool has_hiddensystem, has_namedattr, xattrsupp; + uint32_t clone_blksize; if (nd->nd_repstat) goto out; @@ -330,6 +331,11 @@ nfsrvd_getattr(struct nfsrv_descript *nd, int isdgram, &pathval) != 0) pathval = 0; has_namedattr = pathval > 0; + pathval = 0; + if (VOP_PATHCONF(vp, _PC_CLONE_BLKSIZE, + &pathval) != 0) + pathval = 0; + clone_blksize = pathval; mp = vp->v_mount; if (nfsrv_enable_crossmntpt != 0 && vp->v_type == VDIR && @@ -365,7 +371,7 @@ nfsrvd_getattr(struct nfsrv_descript *nd, int isdgram, isdgram, 1, supports_nfsv4acls, at_root, mounted_on_fileno, xattrsupp, has_hiddensystem, - has_namedattr); + has_namedattr, clone_blksize); vfs_unbusy(mp); } vrele(vp); @@ -4347,7 +4353,7 @@ nfsrvd_verify(struct nfsrv_descript *nd, int isdgram, if (!nd->nd_repstat) { nfsvno_getfs(&fs, isdgram); error = nfsv4_loadattr(nd, vp, &nva, NULL, &fh, fhsize, NULL, - sf, NULL, &fs, NULL, 1, &ret, NULL, NULL, NULL, p, + sf, NULL, &fs, NULL, 1, &ret, NULL, NULL, NULL, NULL, p, nd->nd_cred); if (!error) { if (nd->nd_procnum == NFSV4OP_NVERIFY) { @@ -6011,6 +6017,212 @@ nfsmout: } /* + * nfs clone service + */ +int +nfsrvd_clone(struct nfsrv_descript *nd, __unused int isdgram, + vnode_t vp, vnode_t tovp, struct nfsexstuff *exp, struct nfsexstuff *toexp) +{ + uint32_t *tl; + struct nfsvattr at; + int error = 0, ret; + off_t inoff, outoff; + uint64_t len; + size_t xfer; + struct nfsstate inst, outst, *instp = &inst, *outstp = &outst; + struct nfslock inlo, outlo, *inlop = &inlo, *outlop = &outlo; + nfsquad_t clientid; + nfsv4stateid_t stateid; + nfsattrbit_t attrbits; + void *rl_rcookie, *rl_wcookie; + long pathval; + + rl_rcookie = rl_wcookie = NULL; + pathval = 0; + if (nfsrv_maxcopyrange == 0 || nfsrv_devidcnt > 0 || + VOP_PATHCONF(vp, _PC_CLONE_BLKSIZE, &pathval) != 0 || + pathval == 0) { + /* + * For a pNFS server, reply NFSERR_NOTSUPP so that the client + * will not do the clone and will do I/O on the DS(s). + * If vfs.nfsd.maxcopyrange set to 0, disable Clone. + */ + nd->nd_repstat = NFSERR_NOTSUPP; + goto nfsmout; + } + NFSM_DISSECT(tl, uint32_t *, 2 * NFSX_STATEID + 3 * NFSX_HYPER); + instp->ls_flags = (NFSLCK_CHECK | NFSLCK_READACCESS); + inlop->lo_flags = NFSLCK_READ; + instp->ls_ownerlen = 0; + instp->ls_op = NULL; + instp->ls_uid = nd->nd_cred->cr_uid; + instp->ls_stateid.seqid = fxdr_unsigned(uint32_t, *tl++); + clientid.lval[0] = instp->ls_stateid.other[0] = *tl++; + clientid.lval[1] = instp->ls_stateid.other[1] = *tl++; + if ((nd->nd_flag & ND_IMPLIEDCLID) != 0) + clientid.qval = nd->nd_clientid.qval; + instp->ls_stateid.other[2] = *tl++; + outstp->ls_flags = (NFSLCK_CHECK | NFSLCK_WRITEACCESS); + outlop->lo_flags = NFSLCK_WRITE; + outstp->ls_ownerlen = 0; + outstp->ls_op = NULL; + outstp->ls_uid = nd->nd_cred->cr_uid; + outstp->ls_stateid.seqid = fxdr_unsigned(uint32_t, *tl++); + outstp->ls_stateid.other[0] = *tl++; + outstp->ls_stateid.other[1] = *tl++; + outstp->ls_stateid.other[2] = *tl++; + inoff = fxdr_hyper(tl); tl += 2; + inlop->lo_first = inoff; + outoff = fxdr_hyper(tl); tl += 2; + outlop->lo_first = outoff; + len = fxdr_hyper(tl); + if (len == 0) { + /* len == 0 means to EOF. */ + inlop->lo_end = OFF_MAX; + outlop->lo_end = OFF_MAX; + } else { + inlop->lo_end = inlop->lo_first + len; + outlop->lo_end = outlop->lo_first + len; + } + + if ((inoff > OFF_MAX || outoff > OFF_MAX || + inlop->lo_end > OFF_MAX || outlop->lo_end > OFF_MAX || + inlop->lo_end < inlop->lo_first || outlop->lo_end < + outlop->lo_first)) + nd->nd_repstat = NFSERR_INVAL; + + if (nd->nd_repstat == 0 && vp->v_type != VREG) + nd->nd_repstat = NFSERR_WRONGTYPE; + + /* Check permissions for the input file. */ + NFSZERO_ATTRBIT(&attrbits); + NFSSETBIT_ATTRBIT(&attrbits, NFSATTRBIT_OWNER); + ret = nfsvno_getattr(vp, &at, nd, curthread, 1, &attrbits); + if (nd->nd_repstat == 0) + nd->nd_repstat = ret; + if (nd->nd_repstat == 0 && (at.na_uid != nd->nd_cred->cr_uid || + NFSVNO_EXSTRICTACCESS(exp))) + nd->nd_repstat = nfsvno_accchk(vp, VREAD, nd->nd_cred, exp, + curthread, NFSACCCHK_ALLOWOWNER, NFSACCCHK_VPISLOCKED, + NULL); + if (nd->nd_repstat == 0) + nd->nd_repstat = nfsrv_lockctrl(vp, &instp, &inlop, NULL, + clientid, &stateid, exp, nd, curthread); + if (vp != tovp) { + NFSVOPUNLOCK(vp); + if (nd->nd_repstat != 0) + goto out; + + error = NFSVOPLOCK(tovp, LK_SHARED); + if (error != 0) + goto out; + pathval = 0; + if (VOP_PATHCONF(tovp, _PC_CLONE_BLKSIZE, &pathval) != 0 || + pathval == 0) + nd->nd_repstat = NFSERR_NOTSUPP; + else if (tovp->v_type != VREG) + nd->nd_repstat = NFSERR_WRONGTYPE; + } + + /* For the output file, we only need the Owner attribute. */ + ret = nfsvno_getattr(tovp, &at, nd, curthread, 1, &attrbits); + if (nd->nd_repstat == 0) + nd->nd_repstat = ret; + if (nd->nd_repstat == 0 && (at.na_uid != nd->nd_cred->cr_uid || + NFSVNO_EXSTRICTACCESS(exp))) + nd->nd_repstat = nfsvno_accchk(tovp, VWRITE, nd->nd_cred, toexp, + curthread, NFSACCCHK_ALLOWOWNER, NFSACCCHK_VPISLOCKED, + NULL); + if (nd->nd_repstat == 0) + nd->nd_repstat = nfsrv_lockctrl(tovp, &outstp, &outlop, NULL, + clientid, &stateid, toexp, nd, curthread); + NFSVOPUNLOCK(tovp); + + /* Range lock the byte ranges for both invp and outvp. */ + if (nd->nd_repstat == 0) { + for (;;) { + if (len == 0) + rl_wcookie = vn_rangelock_wlock(tovp, outoff, + OFF_MAX); + else + rl_wcookie = vn_rangelock_wlock(tovp, outoff, + outoff + len); + if (vp != tovp) { + if (len == 0) + rl_rcookie = vn_rangelock_tryrlock(vp, + inoff, OFF_MAX); + else + rl_rcookie = vn_rangelock_tryrlock(vp, + inoff, inoff + len); + if (rl_rcookie != NULL) + break; + } else { + rl_rcookie = NULL; + break; + } + vn_rangelock_unlock(tovp, rl_wcookie); + if (len == 0) + rl_rcookie = vn_rangelock_rlock(vp, inoff, + OFF_MAX); + else + rl_rcookie = vn_rangelock_rlock(vp, inoff, + inoff + len); + vn_rangelock_unlock(vp, rl_rcookie); + } + + error = NFSVOPLOCK(vp, LK_SHARED); + if (error == 0) { + ret = nfsvno_getattr(vp, &at, nd, curthread, 1, NULL); + if (ret == 0) { + /* + * Since invp is range locked, na_size should + * not change. + */ + if (len == 0 && at.na_size > inoff) + len = SSIZE_MAX; /* To EOF. */ + else if (inoff + len > at.na_size) + nd->nd_repstat = NFSERR_INVAL; + } + NFSVOPUNLOCK(vp); + if (ret != 0 && nd->nd_repstat == 0) + nd->nd_repstat = ret; + } else if (nd->nd_repstat == 0) + nd->nd_repstat = error; + } + + /* + * Do the actual copy to an upper limit of vfs.nfsd.maxcopyrange. + * This size limit can be set to limit the time a copy RPC will + * take. + */ + xfer = len; + if (nd->nd_repstat == 0) { + nd->nd_repstat = vn_copy_file_range(vp, &inoff, tovp, &outoff, + &xfer, COPY_FILE_RANGE_CLONE, nd->nd_cred, nd->nd_cred, + NULL); + if (nd->nd_repstat == ENOSYS) + nd->nd_repstat = NFSERR_INVAL; + } + + /* Unlock the ranges. */ + if (rl_rcookie != NULL) + vn_rangelock_unlock(vp, rl_rcookie); + if (rl_wcookie != NULL) + vn_rangelock_unlock(tovp, rl_wcookie); + +out: + vrele(vp); + vrele(tovp); + NFSEXITCODE2(error, nd); + return (error); +nfsmout: + vput(vp); + vrele(tovp); + NFSEXITCODE2(error, nd); + return (error); +} + +/* * nfs seek service */ int diff --git a/sys/fs/nfsserver/nfs_nfsdsocket.c b/sys/fs/nfsserver/nfs_nfsdsocket.c index d6832b4f74be..636c4735a131 100644 --- a/sys/fs/nfsserver/nfs_nfsdsocket.c +++ b/sys/fs/nfsserver/nfs_nfsdsocket.c @@ -371,7 +371,7 @@ int (*nfsrv4_ops2[NFSV42_NOPS])(struct nfsrv_descript *, (int (*)(struct nfsrv_descript *, int, vnode_t , vnode_t , struct nfsexstuff *, struct nfsexstuff *))0, (int (*)(struct nfsrv_descript *, int, vnode_t , vnode_t , struct nfsexstuff *, struct nfsexstuff *))0, (int (*)(struct nfsrv_descript *, int, vnode_t , vnode_t , struct nfsexstuff *, struct nfsexstuff *))0, - (int (*)(struct nfsrv_descript *, int, vnode_t , vnode_t , struct nfsexstuff *, struct nfsexstuff *))0, + nfsrvd_clone, (int (*)(struct nfsrv_descript *, int, vnode_t , vnode_t , struct nfsexstuff *, struct nfsexstuff *))0, (int (*)(struct nfsrv_descript *, int, vnode_t , vnode_t , struct nfsexstuff *, struct nfsexstuff *))0, (int (*)(struct nfsrv_descript *, int, vnode_t , vnode_t , struct nfsexstuff *, struct nfsexstuff *))0, diff --git a/sys/fs/nfsserver/nfs_nfsdstate.c b/sys/fs/nfsserver/nfs_nfsdstate.c index 2e27817389dd..111b0f26d0b5 100644 --- a/sys/fs/nfsserver/nfs_nfsdstate.c +++ b/sys/fs/nfsserver/nfs_nfsdstate.c @@ -4675,7 +4675,7 @@ errout: } else if (error == 0 && procnum == NFSV4OP_CBGETATTR) error = nfsv4_loadattr(nd, NULL, nap, NULL, NULL, 0, NULL, NULL, NULL, NULL, NULL, 0, NULL, NULL, NULL, - NULL, p, NULL); + NULL, NULL, p, NULL); m_freem(nd->nd_mrep); } NFSLOCKSTATE(); @@ -7731,6 +7731,7 @@ nfsrv_setdsserver(char *dspathp, char *mdspathp, NFSPROC_T *p, NFSD_DEBUG(4, "lookup=%d\n", error); if (error != 0) return (error); + NDFREE_PNBUF(&nd); if (nd.ni_vp->v_type != VDIR) { vput(nd.ni_vp); NFSD_DEBUG(4, "dspath not dir\n"); @@ -7767,6 +7768,7 @@ nfsrv_setdsserver(char *dspathp, char *mdspathp, NFSPROC_T *p, NFSD_DEBUG(4, "dsdirpath=%s lookup=%d\n", dsdirpath, error); if (error != 0) break; + NDFREE_PNBUF(&nd); if (nd.ni_vp->v_type != VDIR) { vput(nd.ni_vp); error = ENOTDIR; @@ -7795,6 +7797,7 @@ nfsrv_setdsserver(char *dspathp, char *mdspathp, NFSPROC_T *p, NFSD_DEBUG(4, "mds lookup=%d\n", error); if (error != 0) goto out; + NDFREE_PNBUF(&nd); if (nd.ni_vp->v_type != VDIR) { vput(nd.ni_vp); error = ENOTDIR; @@ -8654,6 +8657,7 @@ nfsrv_mdscopymr(char *mdspathp, char *dspathp, char *curdspathp, char *buf, NFSD_DEBUG(4, "lookup=%d\n", error); if (error != 0) return (error); + NDFREE_PNBUF(&nd); if (nd.ni_vp->v_type != VREG) { vput(nd.ni_vp); NFSD_DEBUG(4, "mdspath not reg\n"); @@ -8675,6 +8679,7 @@ nfsrv_mdscopymr(char *mdspathp, char *dspathp, char *curdspathp, char *buf, vput(vp); return (error); } + NDFREE_PNBUF(&nd); if (nd.ni_vp->v_type != VDIR) { vput(nd.ni_vp); vput(vp); @@ -8717,6 +8722,7 @@ nfsrv_mdscopymr(char *mdspathp, char *dspathp, char *curdspathp, char *buf, vput(curvp); return (error); } + NDFREE_PNBUF(&nd); if (nd.ni_vp->v_type != VDIR || nd.ni_vp == curvp) { vput(nd.ni_vp); vput(vp); diff --git a/sys/kern/init_sysent.c b/sys/kern/init_sysent.c index c0a5479c9634..fcd232cde21e 100644 --- a/sys/kern/init_sysent.c +++ b/sys/kern/init_sysent.c @@ -145,8 +145,8 @@ struct sysent sysent[] = { { .sy_narg = 0, .sy_call = (sy_call_t *)nosys, .sy_auevent = AUE_NULL, .sy_flags = 0, .sy_thrcnt = SY_THR_ABSENT }, /* 76 = obsolete vhangup */ { .sy_narg = 0, .sy_call = (sy_call_t *)nosys, .sy_auevent = AUE_NULL, .sy_flags = 0, .sy_thrcnt = SY_THR_ABSENT }, /* 77 = obsolete vlimit */ { .sy_narg = AS(mincore_args), .sy_call = (sy_call_t *)sys_mincore, .sy_auevent = AUE_MINCORE, .sy_flags = SYF_CAPENABLED, .sy_thrcnt = SY_THR_STATIC }, /* 78 = mincore */ - { .sy_narg = AS(getgroups_args), .sy_call = (sy_call_t *)sys_getgroups, .sy_auevent = AUE_GETGROUPS, .sy_flags = SYF_CAPENABLED, .sy_thrcnt = SY_THR_STATIC }, /* 79 = getgroups */ - { .sy_narg = AS(setgroups_args), .sy_call = (sy_call_t *)sys_setgroups, .sy_auevent = AUE_SETGROUPS, .sy_flags = 0, .sy_thrcnt = SY_THR_STATIC }, /* 80 = setgroups */ + { compat14(AS(freebsd14_getgroups_args),getgroups), .sy_auevent = AUE_GETGROUPS, .sy_flags = SYF_CAPENABLED, .sy_thrcnt = SY_THR_STATIC }, /* 79 = freebsd14 getgroups */ + { compat14(AS(freebsd14_setgroups_args),setgroups), .sy_auevent = AUE_SETGROUPS, .sy_flags = 0, .sy_thrcnt = SY_THR_STATIC }, /* 80 = freebsd14 setgroups */ { .sy_narg = 0, .sy_call = (sy_call_t *)sys_getpgrp, .sy_auevent = AUE_GETPGRP, .sy_flags = SYF_CAPENABLED, .sy_thrcnt = SY_THR_STATIC }, /* 81 = getpgrp */ { .sy_narg = AS(setpgid_args), .sy_call = (sy_call_t *)sys_setpgid, .sy_auevent = AUE_SETPGRP, .sy_flags = 0, .sy_thrcnt = SY_THR_STATIC }, /* 82 = setpgid */ { .sy_narg = AS(setitimer_args), .sy_call = (sy_call_t *)sys_setitimer, .sy_auevent = AUE_SETITIMER, .sy_flags = SYF_CAPENABLED, .sy_thrcnt = SY_THR_STATIC }, /* 83 = setitimer */ @@ -661,4 +661,6 @@ struct sysent sysent[] = { { .sy_narg = AS(exterrctl_args), .sy_call = (sy_call_t *)sys_exterrctl, .sy_auevent = AUE_NULL, .sy_flags = SYF_CAPENABLED, .sy_thrcnt = SY_THR_STATIC }, /* 592 = exterrctl */ { .sy_narg = AS(inotify_add_watch_at_args), .sy_call = (sy_call_t *)sys_inotify_add_watch_at, .sy_auevent = AUE_INOTIFY, .sy_flags = SYF_CAPENABLED, .sy_thrcnt = SY_THR_STATIC }, /* 593 = inotify_add_watch_at */ { .sy_narg = AS(inotify_rm_watch_args), .sy_call = (sy_call_t *)sys_inotify_rm_watch, .sy_auevent = AUE_INOTIFY, .sy_flags = SYF_CAPENABLED, .sy_thrcnt = SY_THR_STATIC }, /* 594 = inotify_rm_watch */ + { .sy_narg = AS(getgroups_args), .sy_call = (sy_call_t *)sys_getgroups, .sy_auevent = AUE_GETGROUPS, .sy_flags = SYF_CAPENABLED, .sy_thrcnt = SY_THR_STATIC }, /* 595 = getgroups */ + { .sy_narg = AS(setgroups_args), .sy_call = (sy_call_t *)sys_setgroups, .sy_auevent = AUE_SETGROUPS, .sy_flags = 0, .sy_thrcnt = SY_THR_STATIC }, /* 596 = setgroups */ }; diff --git a/sys/kern/kern_prot.c b/sys/kern/kern_prot.c index 2cd5b7069023..0ca42d640767 100644 --- a/sys/kern/kern_prot.c +++ b/sys/kern/kern_prot.c @@ -310,6 +310,39 @@ sys_getegid(struct thread *td, struct getegid_args *uap) return (0); } +#ifdef COMPAT_FREEBSD14 +int +freebsd14_getgroups(struct thread *td, struct freebsd14_getgroups_args *uap) +{ + struct ucred *cred; + int ngrp, error; + + cred = td->td_ucred; + + /* + * For FreeBSD < 15.0, we account for the egid being placed at the + * beginning of the group list prior to all supplementary groups. + */ + ngrp = cred->cr_ngroups + 1; + if (uap->gidsetsize == 0) { + error = 0; + goto out; + } else if (uap->gidsetsize < ngrp) { + return (EINVAL); + } + + error = copyout(&cred->cr_gid, uap->gidset, sizeof(gid_t)); + if (error == 0) + error = copyout(cred->cr_groups, uap->gidset + 1, + (ngrp - 1) * sizeof(gid_t)); + +out: + td->td_retval[0] = ngrp; + return (error); + +} +#endif /* COMPAT_FREEBSD14 */ + #ifndef _SYS_SYSPROTO_H_ struct getgroups_args { int gidsetsize; @@ -320,18 +353,11 @@ int sys_getgroups(struct thread *td, struct getgroups_args *uap) { struct ucred *cred; - gid_t *ugidset; int ngrp, error; cred = td->td_ucred; - /* - * cr_gid has been moved out of cr_groups, but we'll continue exporting - * the egid as groups[0] for the time being until we audit userland for - * any surprises. - */ - ngrp = cred->cr_ngroups + 1; - + ngrp = cred->cr_ngroups; if (uap->gidsetsize == 0) { error = 0; goto out; @@ -339,14 +365,7 @@ sys_getgroups(struct thread *td, struct getgroups_args *uap) if (uap->gidsetsize < ngrp) return (EINVAL); - ugidset = uap->gidset; - error = copyout(&cred->cr_gid, ugidset, sizeof(*ugidset)); - if (error != 0) - goto out; - - if (ngrp > 1) - error = copyout(cred->cr_groups, ugidset + 1, - (ngrp - 1) * sizeof(*ugidset)); + error = copyout(cred->cr_groups, uap->gidset, ngrp * sizeof(gid_t)); out: td->td_retval[0] = ngrp; return (error); @@ -1186,6 +1205,44 @@ fail: return (error); } +#ifdef COMPAT_FREEBSD14 +int +freebsd14_setgroups(struct thread *td, struct freebsd14_setgroups_args *uap) +{ + gid_t smallgroups[CRED_SMALLGROUPS_NB]; + gid_t *groups; + int gidsetsize, error; + + /* + * Before FreeBSD 15.0, we allow one more group to be supplied to + * account for the egid appearing before the supplementary groups. This + * may technically allow one more supplementary group for systems that + * did use the default NGROUPS_MAX if we round it back up to 1024. + */ + gidsetsize = uap->gidsetsize; + if (gidsetsize > ngroups_max + 1 || gidsetsize < 0) + return (EINVAL); + + if (gidsetsize > CRED_SMALLGROUPS_NB) + groups = malloc(gidsetsize * sizeof(gid_t), M_TEMP, M_WAITOK); + else + groups = smallgroups; + + error = copyin(uap->gidset, groups, gidsetsize * sizeof(gid_t)); + if (error == 0) { + int ngroups = gidsetsize > 0 ? gidsetsize - 1 /* egid */ : 0; + + error = kern_setgroups(td, &ngroups, groups + 1); + if (error == 0 && gidsetsize > 0) + td->td_proc->p_ucred->cr_gid = groups[0]; + } + + if (groups != smallgroups) + free(groups, M_TEMP); + return (error); +} +#endif /* COMPAT_FREEBSD14 */ + #ifndef _SYS_SYSPROTO_H_ struct setgroups_args { int gidsetsize; @@ -1210,8 +1267,7 @@ sys_setgroups(struct thread *td, struct setgroups_args *uap) * setgroups() differ. */ gidsetsize = uap->gidsetsize; - /* XXXKE Limit to ngroups_max when we change the userland interface. */ - if (gidsetsize > ngroups_max + 1 || gidsetsize < 0) + if (gidsetsize > ngroups_max || gidsetsize < 0) return (EINVAL); if (gidsetsize > CRED_SMALLGROUPS_NB) @@ -1238,35 +1294,17 @@ kern_setgroups(struct thread *td, int *ngrpp, gid_t *groups) struct proc *p = td->td_proc; struct ucred *newcred, *oldcred; int ngrp, error; - gid_t egid; ngrp = *ngrpp; /* Sanity check size. */ - /* XXXKE Limit to ngroups_max when we change the userland interface. */ - if (ngrp < 0 || ngrp > ngroups_max + 1) + if (ngrp < 0 || ngrp > ngroups_max) return (EINVAL); AUDIT_ARG_GROUPSET(groups, ngrp); - /* - * setgroups(0, NULL) is a legitimate way of clearing the groups vector - * on non-BSD systems (which generally do not have the egid in the - * groups[0]). We risk security holes when running non-BSD software if - * we do not do the same. So we allow and treat 0 for 'ngrp' specially - * below (twice). - */ - if (ngrp != 0) { - /* - * To maintain userland compat for now, we use the first group - * as our egid and we'll use the rest as our supplemental - * groups. - */ - egid = groups[0]; - ngrp--; - groups++; - groups_normalize(&ngrp, groups); - *ngrpp = ngrp; - } + groups_normalize(&ngrp, groups); + *ngrpp = ngrp; + newcred = crget(); crextend(newcred, ngrp); PROC_LOCK(p); @@ -1289,15 +1327,7 @@ kern_setgroups(struct thread *td, int *ngrpp, gid_t *groups) if (error) goto fail; - /* - * If some groups were passed, the first one is currently the desired - * egid. This code is to be removed (along with some commented block - * above) when setgroups() is changed to take only supplementary groups. - */ - if (ngrp != 0) - newcred->cr_gid = egid; crsetgroups_internal(newcred, ngrp, groups); - setsugid(p); proc_set_cred(p, newcred); PROC_UNLOCK(p); @@ -2891,7 +2921,8 @@ crextend(struct ucred *cr, int n) * Normalizes a set of groups to be applied to a 'struct ucred'. * * Normalization ensures that the supplementary groups are sorted in ascending - * order and do not contain duplicates. + * order and do not contain duplicates. This allows group_is_supplementary + * to do a binary search. */ static void groups_normalize(int *ngrp, gid_t *groups) diff --git a/sys/kern/syscalls.c b/sys/kern/syscalls.c index 09bf4d519927..4122f9261871 100644 --- a/sys/kern/syscalls.c +++ b/sys/kern/syscalls.c @@ -84,8 +84,8 @@ const char *syscallnames[] = { "obs_vhangup", /* 76 = obsolete vhangup */ "obs_vlimit", /* 77 = obsolete vlimit */ "mincore", /* 78 = mincore */ - "getgroups", /* 79 = getgroups */ - "setgroups", /* 80 = setgroups */ + "compat14.getgroups", /* 79 = freebsd14 getgroups */ + "compat14.setgroups", /* 80 = freebsd14 setgroups */ "getpgrp", /* 81 = getpgrp */ "setpgid", /* 82 = setpgid */ "setitimer", /* 83 = setitimer */ @@ -600,4 +600,6 @@ const char *syscallnames[] = { "exterrctl", /* 592 = exterrctl */ "inotify_add_watch_at", /* 593 = inotify_add_watch_at */ "inotify_rm_watch", /* 594 = inotify_rm_watch */ + "getgroups", /* 595 = getgroups */ + "setgroups", /* 596 = setgroups */ }; diff --git a/sys/kern/syscalls.master b/sys/kern/syscalls.master index 53b5d3cbbba9..fa64597d14a5 100644 --- a/sys/kern/syscalls.master +++ b/sys/kern/syscalls.master @@ -552,13 +552,13 @@ _Out_writes_bytes_(len/PAGE_SIZE) char *vec ); } -79 AUE_GETGROUPS STD|CAPENABLED { +79 AUE_GETGROUPS STD|CAPENABLED|COMPAT14 { int getgroups( int gidsetsize, _Out_writes_opt_(gidsetsize) gid_t *gidset ); } -80 AUE_SETGROUPS STD { +80 AUE_SETGROUPS STD|COMPAT14 { int setgroups( int gidsetsize, _In_reads_(gidsetsize) const gid_t *gidset @@ -3371,5 +3371,17 @@ int wd ); } +595 AUE_GETGROUPS STD|CAPENABLED { + int getgroups( + int gidsetsize, + _Out_writes_opt_(gidsetsize) gid_t *gidset + ); + } +596 AUE_SETGROUPS STD { + int setgroups( + int gidsetsize, + _In_reads_(gidsetsize) const gid_t *gidset + ); + } ; vim: syntax=off diff --git a/sys/kern/systrace_args.c b/sys/kern/systrace_args.c index 4dfc63924da9..2b1ea9eed8d4 100644 --- a/sys/kern/systrace_args.c +++ b/sys/kern/systrace_args.c @@ -454,22 +454,6 @@ systrace_args(int sysnum, void *params, uint64_t *uarg, int *n_args) *n_args = 3; break; } - /* getgroups */ - case 79: { - struct getgroups_args *p = params; - iarg[a++] = p->gidsetsize; /* int */ - uarg[a++] = (intptr_t)p->gidset; /* gid_t * */ - *n_args = 2; - break; - } - /* setgroups */ - case 80: { - struct setgroups_args *p = params; - iarg[a++] = p->gidsetsize; /* int */ - uarg[a++] = (intptr_t)p->gidset; /* const gid_t * */ - *n_args = 2; - break; - } /* getpgrp */ case 81: { *n_args = 0; @@ -3500,6 +3484,22 @@ systrace_args(int sysnum, void *params, uint64_t *uarg, int *n_args) *n_args = 2; break; } + /* getgroups */ + case 595: { + struct getgroups_args *p = params; + iarg[a++] = p->gidsetsize; /* int */ + uarg[a++] = (intptr_t)p->gidset; /* gid_t * */ + *n_args = 2; + break; + } + /* setgroups */ + case 596: { + struct setgroups_args *p = params; + iarg[a++] = p->gidsetsize; /* int */ + uarg[a++] = (intptr_t)p->gidset; /* const gid_t * */ + *n_args = 2; + break; + } default: *n_args = 0; break; @@ -4199,32 +4199,6 @@ systrace_entry_setargdesc(int sysnum, int ndx, char *desc, size_t descsz) break; }; break; - /* getgroups */ - case 79: - switch (ndx) { - case 0: - p = "int"; - break; - case 1: - p = "userland gid_t *"; - break; - default: - break; - }; - break; - /* setgroups */ - case 80: - switch (ndx) { - case 0: - p = "int"; - break; - case 1: - p = "userland const gid_t *"; - break; - default: - break; - }; - break; /* getpgrp */ case 81: break; @@ -9367,6 +9341,32 @@ systrace_entry_setargdesc(int sysnum, int ndx, char *desc, size_t descsz) break; }; break; + /* getgroups */ + case 595: + switch (ndx) { + case 0: + p = "int"; + break; + case 1: + p = "userland gid_t *"; + break; + default: + break; + }; + break; + /* setgroups */ + case 596: + switch (ndx) { + case 0: + p = "int"; + break; + case 1: + p = "userland const gid_t *"; + break; + default: + break; + }; + break; default: break; }; @@ -9633,16 +9633,6 @@ systrace_return_setargdesc(int sysnum, int ndx, char *desc, size_t descsz) if (ndx == 0 || ndx == 1) p = "int"; break; - /* getgroups */ - case 79: - if (ndx == 0 || ndx == 1) - p = "int"; - break; - /* setgroups */ - case 80: - if (ndx == 0 || ndx == 1) - p = "int"; - break; /* getpgrp */ case 81: /* setpgid */ @@ -11365,6 +11355,16 @@ systrace_return_setargdesc(int sysnum, int ndx, char *desc, size_t descsz) if (ndx == 0 || ndx == 1) p = "int"; break; + /* getgroups */ + case 595: + if (ndx == 0 || ndx == 1) + p = "int"; + break; + /* setgroups */ + case 596: + if (ndx == 0 || ndx == 1) + p = "int"; + break; default: break; }; diff --git a/sys/kern/vfs_inotify.c b/sys/kern/vfs_inotify.c index 746a5a39208e..b265a5ff3a62 100644 --- a/sys/kern/vfs_inotify.c +++ b/sys/kern/vfs_inotify.c @@ -801,6 +801,7 @@ vn_inotify_add_watch(struct vnode *vp, struct inotify_softc *sc, uint32_t mask, vn_lock(vp, LK_SHARED | LK_RETRY); if (error != 0) break; + NDFREE_PNBUF(&nd); vn_irflag_set_cond(nd.ni_vp, VIRF_INOTIFY_PARENT); vrele(nd.ni_vp); } diff --git a/sys/modules/ossl/Makefile b/sys/modules/ossl/Makefile index 7a92742d6b36..ac2c752e922e 100644 --- a/sys/modules/ossl/Makefile +++ b/sys/modules/ossl/Makefile @@ -29,6 +29,7 @@ SRCS.arm= \ SRCS.aarch64= \ chacha-armv8.S \ + chacha-armv8-sve.S \ poly1305-armv8.S \ sha1-armv8.S \ sha256-armv8.S \ diff --git a/sys/net/iflib.c b/sys/net/iflib.c index 6638c90882aa..2b43f6f19051 100644 --- a/sys/net/iflib.c +++ b/sys/net/iflib.c @@ -2910,7 +2910,9 @@ iflib_rxeof(iflib_rxq_t rxq, qidx_t budget) struct if_rxd_info ri; int err, budget_left, rx_bytes, rx_pkts; iflib_fl_t fl; +#if defined(INET6) || defined(INET) int lro_enabled; +#endif uint8_t retval = 0; /* @@ -2936,7 +2938,9 @@ iflib_rxeof(iflib_rxq_t rxq, qidx_t budget) return (retval); } +#if defined(INET6) || defined(INET) lro_enabled = (if_getcapenable(ifp) & IFCAP_LRO); +#endif /* pfil needs the vnet to be set */ CURVNET_SET_QUIET(if_getvnet(ifp)); diff --git a/sys/netinet/icmp6.h b/sys/netinet/icmp6.h index 7845b682f3e4..2ca5b3433e47 100644 --- a/sys/netinet/icmp6.h +++ b/sys/netinet/icmp6.h @@ -713,9 +713,6 @@ void icmp6_redirect_input(struct mbuf *, int); void icmp6_redirect_output(struct mbuf *, struct nhop_object *); int icmp6_ratelimit(const struct in6_addr *, const int, const int); -struct ip6ctlparam; -void icmp6_mtudisc_update(struct ip6ctlparam *, int); - /* XXX: is this the right place for these macros? */ #define icmp6_ifstat_inc(ifp, tag) \ do { \ diff --git a/sys/netinet/tcp_subr.c b/sys/netinet/tcp_subr.c index 1fce7c591639..2e039ebbfdd2 100644 --- a/sys/netinet/tcp_subr.c +++ b/sys/netinet/tcp_subr.c @@ -3205,7 +3205,7 @@ tcp6_next_pmtu(const struct icmp6_hdr *icmp6) * small, set to the min. */ if (mtu < IPV6_MMTU) - mtu = IPV6_MMTU - 8; /* XXXNP: what is the adjustment for? */ + mtu = IPV6_MMTU; return (mtu); } diff --git a/sys/netinet/tcp_timewait.c b/sys/netinet/tcp_timewait.c index c095fc8f7765..ce63fcf9ffc0 100644 --- a/sys/netinet/tcp_timewait.c +++ b/sys/netinet/tcp_timewait.c @@ -101,7 +101,7 @@ sysctl_net_inet_tcp_nolocaltimewait(SYSCTL_HANDLER_ARGS) if (error == 0 && req->newptr) { V_nolocaltimewait = new; gone_in(16, "net.inet.tcp.nolocaltimewait is obsolete." - " Use net.inet.tcp.local_msl instead.\n"); + " Use net.inet.tcp.msl_local instead.\n"); } return (error); } diff --git a/sys/netinet6/icmp6.c b/sys/netinet6/icmp6.c index d89515d7eda5..f98381499b2d 100644 --- a/sys/netinet6/icmp6.c +++ b/sys/netinet6/icmp6.c @@ -159,6 +159,7 @@ static int ni6_addrs(struct icmp6_nodeinfo *, struct mbuf *, static int ni6_store_addrs(struct icmp6_nodeinfo *, struct icmp6_nodeinfo *, struct ifnet *, int); static int icmp6_notify_error(struct mbuf **, int, int); +static void icmp6_mtudisc_update(struct ip6ctlparam *ip6cp); /* * Kernel module interface for updating icmp6stat. The argument is an index @@ -1115,7 +1116,7 @@ icmp6_notify_error(struct mbuf **mp, int off, int icmp6len) if (icmp6type == ICMP6_PACKET_TOO_BIG) { notifymtu = ntohl(icmp6->icmp6_mtu); ip6cp.ip6c_cmdarg = (void *)¬ifymtu; - icmp6_mtudisc_update(&ip6cp, 1); /*XXX*/ + icmp6_mtudisc_update(&ip6cp); } if (ip6_ctlprotox[nxt] != NULL) @@ -1130,47 +1131,18 @@ icmp6_notify_error(struct mbuf **mp, int off, int icmp6len) return (-1); } -void -icmp6_mtudisc_update(struct ip6ctlparam *ip6cp, int validated) +static void +icmp6_mtudisc_update(struct ip6ctlparam *ip6cp) { struct in6_addr *dst = &ip6cp->ip6c_finaldst->sin6_addr; struct icmp6_hdr *icmp6 = ip6cp->ip6c_icmp6; - struct mbuf *m = ip6cp->ip6c_m; /* will be necessary for scope issue */ + struct mbuf *m = ip6cp->ip6c_m; u_int mtu = ntohl(icmp6->icmp6_mtu); struct in_conninfo inc; uint32_t max_mtu; -#if 0 - /* - * RFC2460 section 5, last paragraph. - * even though minimum link MTU for IPv6 is IPV6_MMTU, - * we may see ICMPv6 too big with mtu < IPV6_MMTU - * due to packet translator in the middle. - * see ip6_output() and ip6_getpmtu() "alwaysfrag" case for - * special handling. - */ if (mtu < IPV6_MMTU) return; -#endif - - /* - * we reject ICMPv6 too big with abnormally small value. - * XXX what is the good definition of "abnormally small"? - */ - if (mtu < sizeof(struct ip6_hdr) + sizeof(struct ip6_frag) + 8) - return; - - if (!validated) - return; - - /* - * In case the suggested mtu is less than IPV6_MMTU, we - * only need to remember that it was for above mentioned - * "alwaysfrag" case. - * Try to be as close to the spec as possible. - */ - if (mtu < IPV6_MMTU) - mtu = IPV6_MMTU - 8; bzero(&inc, sizeof(inc)); inc.inc_fibnum = M_GETFIB(m); diff --git a/sys/netinet6/ip6_output.c b/sys/netinet6/ip6_output.c index ed71c58fffbe..6299ce6e146b 100644 --- a/sys/netinet6/ip6_output.c +++ b/sys/netinet6/ip6_output.c @@ -145,11 +145,10 @@ static int ip6_insertfraghdr(struct mbuf *, struct mbuf *, int, struct ip6_frag **); static int ip6_insert_jumboopt(struct ip6_exthdrs *, u_int32_t); static int ip6_splithdr(struct mbuf *, struct ip6_exthdrs *); -static int ip6_getpmtu(struct route_in6 *, int, - struct ifnet *, const struct in6_addr *, u_long *, int *, u_int, - u_int); -static int ip6_calcmtu(struct ifnet *, const struct in6_addr *, u_long, - u_long *, int *, u_int); +static void ip6_getpmtu(struct route_in6 *, int, + struct ifnet *, const struct in6_addr *, u_long *, u_int, u_int); +static void ip6_calcmtu(struct ifnet *, const struct in6_addr *, u_long, + u_long *, u_int); static int ip6_getpmtu_ctl(u_int, const struct in6_addr *, u_long *); static int copypktopts(struct ip6_pktopts *, struct ip6_pktopts *, int); @@ -418,7 +417,7 @@ ip6_output(struct mbuf *m0, struct ip6_pktopts *opt, int vlan_pcp = -1; struct in6_ifaddr *ia = NULL; u_long mtu; - int alwaysfrag, dontfrag; + int dontfrag; u_int32_t optlen, plen = 0, unfragpartlen; struct ip6_exthdrs exthdrs; struct in6_addr src0, dst0; @@ -939,12 +938,10 @@ nonh6lookup: *ifpp = ifp; /* Determine path MTU. */ - if ((error = ip6_getpmtu(ro_pmtu, ro != ro_pmtu, ifp, &ip6->ip6_dst, - &mtu, &alwaysfrag, fibnum, *nexthdrp)) != 0) - goto bad; - KASSERT(mtu > 0, ("%s:%d: mtu %ld, ro_pmtu %p ro %p ifp %p " - "alwaysfrag %d fibnum %u\n", __func__, __LINE__, mtu, ro_pmtu, ro, - ifp, alwaysfrag, fibnum)); + ip6_getpmtu(ro_pmtu, ro != ro_pmtu, ifp, &ip6->ip6_dst, &mtu, fibnum, + *nexthdrp); + KASSERT(mtu > 0, ("%s:%d: mtu %ld, ro_pmtu %p ro %p ifp %p fibnum %u", + __func__, __LINE__, mtu, ro_pmtu, ro, ifp, fibnum)); /* * The caller of this function may specify to use the minimum MTU @@ -1121,20 +1118,13 @@ passout: * Send the packet to the outgoing interface. * If necessary, do IPv6 fragmentation before sending. * - * The logic here is rather complex: - * 1: normal case (dontfrag == 0, alwaysfrag == 0) + * 1: normal case (dontfrag == 0) * 1-a: send as is if tlen <= path mtu * 1-b: fragment if tlen > path mtu * * 2: if user asks us not to fragment (dontfrag == 1) * 2-a: send as is if tlen <= interface mtu * 2-b: error if tlen > interface mtu - * - * 3: if we always need to attach fragment header (alwaysfrag == 1) - * always fragment - * - * 4: if dontfrag == 1 && alwaysfrag == 1 - * error, as we cannot handle this conflicting request. */ sw_csum = m->m_pkthdr.csum_flags; if (!hdrsplit) { @@ -1157,14 +1147,9 @@ passout: dontfrag = 1; else dontfrag = 0; - if (dontfrag && alwaysfrag) { /* Case 4. */ - /* Conflicting request - can't transmit. */ - error = EMSGSIZE; - goto bad; - } if (dontfrag && tlen > IN6_LINKMTU(ifp) && !tso) { /* Case 2-b. */ /* - * Even if the DONTFRAG option is specified, we cannot send the + * If the DONTFRAG option is specified, we cannot send the * packet when the data length is larger than the MTU of the * outgoing interface. * Notify the error by sending IPV6_PATHMTU ancillary data if @@ -1178,7 +1163,7 @@ passout: } /* Transmit packet without fragmentation. */ - if (dontfrag || (!alwaysfrag && tlen <= mtu)) { /* Cases 1-a and 2-a. */ + if (dontfrag || tlen <= mtu) { /* Cases 1-a and 2-a. */ struct in6_ifaddr *ia6; ip6 = mtod(m, struct ip6_hdr *); @@ -1194,7 +1179,7 @@ passout: goto done; } - /* Try to fragment the packet. Cases 1-b and 3. */ + /* Try to fragment the packet. Case 1-b. */ if (mtu < IPV6_MMTU) { /* Path MTU cannot be less than IPV6_MMTU. */ error = EMSGSIZE; @@ -1478,9 +1463,10 @@ ip6_getpmtu_ctl(u_int fibnum, const struct in6_addr *dst, u_long *mtup) NET_EPOCH_ENTER(et); nh = fib6_lookup(fibnum, &kdst, scopeid, NHR_NONE, 0); - if (nh != NULL) - error = ip6_calcmtu(nh->nh_ifp, dst, nh->nh_mtu, mtup, NULL, 0); - else + if (nh != NULL) { + ip6_calcmtu(nh->nh_ifp, dst, nh->nh_mtu, mtup, 0); + error = 0; + } else error = EHOSTUNREACH; NET_EPOCH_EXIT(et); @@ -1494,13 +1480,12 @@ ip6_getpmtu_ctl(u_int fibnum, const struct in6_addr *dst, u_long *mtup) * inside @ro_pmtu to avoid subsequent route lookups after packet * filter processing. * - * Stores mtu and always-frag value into @mtup and @alwaysfragp. - * Returns 0 on success. + * Stores mtu into @mtup. */ -static int +static void ip6_getpmtu(struct route_in6 *ro_pmtu, int do_lookup, struct ifnet *ifp, const struct in6_addr *dst, u_long *mtup, - int *alwaysfragp, u_int fibnum, u_int proto) + u_int fibnum, u_int proto) { struct nhop_object *nh; struct in6_addr kdst; @@ -1544,65 +1529,41 @@ ip6_getpmtu(struct route_in6 *ro_pmtu, int do_lookup, if (ro_pmtu != NULL && ro_pmtu->ro_nh != NULL) mtu = ro_pmtu->ro_nh->nh_mtu; - return (ip6_calcmtu(ifp, dst, mtu, mtup, alwaysfragp, proto)); + ip6_calcmtu(ifp, dst, mtu, mtup, proto); } /* * Calculate MTU based on transmit @ifp, route mtu @rt_mtu and * hostcache data for @dst. - * Stores mtu and always-frag value into @mtup and @alwaysfragp. - * - * Returns 0 on success. + * Stores mtu into @mtup. */ -static int +static void ip6_calcmtu(struct ifnet *ifp, const struct in6_addr *dst, u_long rt_mtu, - u_long *mtup, int *alwaysfragp, u_int proto) + u_long *mtup, u_int proto) { u_long mtu = 0; - int alwaysfrag = 0; - int error = 0; if (rt_mtu > 0) { - u_int32_t ifmtu; - struct in_conninfo inc; - - bzero(&inc, sizeof(inc)); - inc.inc_flags |= INC_ISIPV6; - inc.inc6_faddr = *dst; - - ifmtu = IN6_LINKMTU(ifp); + /* Skip the hostcache if the protocol handles PMTU changes. */ + if (proto != IPPROTO_TCP && proto != IPPROTO_SCTP) { + struct in_conninfo inc = { + .inc_flags = INC_ISIPV6, + .inc6_faddr = *dst, + }; - /* TCP is known to react to pmtu changes so skip hc */ - if (proto != IPPROTO_TCP) mtu = tcp_hc_getmtu(&inc); + } if (mtu) mtu = min(mtu, rt_mtu); else mtu = rt_mtu; - if (mtu == 0) - mtu = ifmtu; - else if (mtu < IPV6_MMTU) { - /* - * RFC2460 section 5, last paragraph: - * if we record ICMPv6 too big message with - * mtu < IPV6_MMTU, transmit packets sized IPV6_MMTU - * or smaller, with framgent header attached. - * (fragment header is needed regardless from the - * packet size, for translators to identify packets) - */ - alwaysfrag = 1; - mtu = IPV6_MMTU; - } - } else if (ifp) { + } + + if (mtu == 0) mtu = IN6_LINKMTU(ifp); - } else - error = EHOSTUNREACH; /* XXX */ *mtup = mtu; - if (alwaysfragp) - *alwaysfragp = alwaysfrag; - return (error); } /* diff --git a/sys/powerpc/mpc85xx/mpc85xx_gpio.c b/sys/powerpc/mpc85xx/mpc85xx_gpio.c index cb96d768adef..7353ed7bac7b 100644 --- a/sys/powerpc/mpc85xx/mpc85xx_gpio.c +++ b/sys/powerpc/mpc85xx/mpc85xx_gpio.c @@ -228,12 +228,13 @@ mpc85xx_gpio_attach(device_t dev) OF_device_register_xref(OF_xref_from_node(ofw_bus_get_node(dev)), dev); - sc->busdev = gpiobus_attach_bus(dev); + sc->busdev = gpiobus_add_bus(dev); if (sc->busdev == NULL) { mpc85xx_gpio_detach(dev); return (ENOMEM); } + bus_attach_children(dev); return (0); } diff --git a/sys/riscv/sifive/sifive_gpio.c b/sys/riscv/sifive/sifive_gpio.c index ef68d2b39da3..98bff2f72082 100644 --- a/sys/riscv/sifive/sifive_gpio.c +++ b/sys/riscv/sifive/sifive_gpio.c @@ -157,13 +157,14 @@ sfgpio_attach(device_t dev) sc->gpio_pins[i].gp_name[GPIOMAXNAME - 1] = '\0'; } - sc->busdev = gpiobus_attach_bus(dev); + sc->busdev = gpiobus_add_bus(dev); if (sc->busdev == NULL) { device_printf(dev, "Cannot attach gpiobus\n"); error = ENXIO; goto fail; } + bus_attach_children(dev); return (0); fail: diff --git a/sys/riscv/starfive/jh7110_gpio.c b/sys/riscv/starfive/jh7110_gpio.c index 452a3306b4a1..1ed7d9f42259 100644 --- a/sys/riscv/starfive/jh7110_gpio.c +++ b/sys/riscv/starfive/jh7110_gpio.c @@ -321,13 +321,14 @@ jh7110_gpio_attach(device_t dev) JH7110_GPIO_WRITE(sc, GPIOE_1, 0); JH7110_GPIO_WRITE(sc, GPIOEN, 1); - sc->busdev = gpiobus_attach_bus(dev); + sc->busdev = gpiobus_add_bus(dev); if (sc->busdev == NULL) { device_printf(dev, "Cannot attach gpiobus\n"); jh7110_gpio_detach(dev); return (ENXIO); } + bus_attach_children(dev); return (0); } diff --git a/sys/rpc/auth.h b/sys/rpc/auth.h index 3d58fb19536b..33c33ffd594d 100644 --- a/sys/rpc/auth.h +++ b/sys/rpc/auth.h @@ -246,6 +246,19 @@ extern AUTH *authunix_create_default(void); /* takes no parameters */ extern AUTH *authnone_create(void); /* takes no parameters */ extern AUTH *authtls_create(void); /* takes no parameters */ __END_DECLS +/* + * DES style authentication + * AUTH *authsecdes_create(servername, window, timehost, ckey) + * char *servername; - network name of server + * u_int window; - time to live + * const char *timehost; - optional hostname to sync with + * des_block *ckey; - optional conversation key to use + */ +__BEGIN_DECLS +extern AUTH *authdes_create (char *, u_int, struct sockaddr *, des_block *); +extern AUTH *authdes_seccreate (const char *, const u_int, const char *, + const des_block *); +__END_DECLS __BEGIN_DECLS extern bool_t xdr_opaque_auth (XDR *, struct opaque_auth *); @@ -267,6 +280,19 @@ extern void passwd2des ( char *, char * ); __END_DECLS /* + * + * These routines interface to the keyserv daemon + * + */ +__BEGIN_DECLS +extern int key_decryptsession(const char *, des_block *); +extern int key_encryptsession(const char *, des_block *); +extern int key_gendes(des_block *); +extern int key_setsecret(const char *); +extern int key_secretkey_is_set(void); +__END_DECLS + +/* * Publickey routines. */ __BEGIN_DECLS diff --git a/sys/sys/param.h b/sys/sys/param.h index b9942c7bc2fd..c410a6ee666f 100644 --- a/sys/sys/param.h +++ b/sys/sys/param.h @@ -74,7 +74,7 @@ * cannot include sys/param.h and should only be updated here. */ #undef __FreeBSD_version -#define __FreeBSD_version 1500059 +#define __FreeBSD_version 1500060 /* * __FreeBSD_kernel__ indicates that this system uses the kernel of FreeBSD, diff --git a/sys/sys/syscall.h b/sys/sys/syscall.h index d703a11fda01..2d6903967e15 100644 --- a/sys/sys/syscall.h +++ b/sys/sys/syscall.h @@ -85,8 +85,8 @@ /* 76 is obsolete vhangup */ /* 77 is obsolete vlimit */ #define SYS_mincore 78 -#define SYS_getgroups 79 -#define SYS_setgroups 80 +#define SYS_freebsd14_getgroups 79 +#define SYS_freebsd14_setgroups 80 #define SYS_getpgrp 81 #define SYS_setpgid 82 #define SYS_setitimer 83 @@ -533,4 +533,6 @@ #define SYS_exterrctl 592 #define SYS_inotify_add_watch_at 593 #define SYS_inotify_rm_watch 594 -#define SYS_MAXSYSCALL 595 +#define SYS_getgroups 595 +#define SYS_setgroups 596 +#define SYS_MAXSYSCALL 597 diff --git a/sys/sys/syscall.mk b/sys/sys/syscall.mk index b7ded62cacb4..d1172c2dc7bf 100644 --- a/sys/sys/syscall.mk +++ b/sys/sys/syscall.mk @@ -65,8 +65,8 @@ MIASM = \ mprotect.o \ madvise.o \ mincore.o \ - getgroups.o \ - setgroups.o \ + freebsd14_getgroups.o \ + freebsd14_setgroups.o \ getpgrp.o \ setpgid.o \ setitimer.o \ @@ -436,4 +436,6 @@ MIASM = \ setcred.o \ exterrctl.o \ inotify_add_watch_at.o \ - inotify_rm_watch.o + inotify_rm_watch.o \ + getgroups.o \ + setgroups.o diff --git a/sys/sys/sysproto.h b/sys/sys/sysproto.h index 8d666f9c8ee9..98311a6dbf94 100644 --- a/sys/sys/sysproto.h +++ b/sys/sys/sysproto.h @@ -273,14 +273,6 @@ struct mincore_args { char len_l_[PADL_(size_t)]; size_t len; char len_r_[PADR_(size_t)]; char vec_l_[PADL_(char *)]; char * vec; char vec_r_[PADR_(char *)]; }; -struct getgroups_args { - char gidsetsize_l_[PADL_(int)]; int gidsetsize; char gidsetsize_r_[PADR_(int)]; - char gidset_l_[PADL_(gid_t *)]; gid_t * gidset; char gidset_r_[PADR_(gid_t *)]; -}; -struct setgroups_args { - char gidsetsize_l_[PADL_(int)]; int gidsetsize; char gidsetsize_r_[PADR_(int)]; - char gidset_l_[PADL_(const gid_t *)]; const gid_t * gidset; char gidset_r_[PADR_(const gid_t *)]; -}; struct getpgrp_args { syscallarg_t dummy; }; @@ -1901,6 +1893,14 @@ struct inotify_rm_watch_args { char fd_l_[PADL_(int)]; int fd; char fd_r_[PADR_(int)]; char wd_l_[PADL_(int)]; int wd; char wd_r_[PADR_(int)]; }; +struct getgroups_args { + char gidsetsize_l_[PADL_(int)]; int gidsetsize; char gidsetsize_r_[PADR_(int)]; + char gidset_l_[PADL_(gid_t *)]; gid_t * gidset; char gidset_r_[PADR_(gid_t *)]; +}; +struct setgroups_args { + char gidsetsize_l_[PADL_(int)]; int gidsetsize; char gidsetsize_r_[PADR_(int)]; + char gidset_l_[PADL_(const gid_t *)]; const gid_t * gidset; char gidset_r_[PADR_(const gid_t *)]; +}; int sys__exit(struct thread *, struct _exit_args *); int sys_fork(struct thread *, struct fork_args *); int sys_read(struct thread *, struct read_args *); @@ -1957,8 +1957,6 @@ int sys_munmap(struct thread *, struct munmap_args *); int sys_mprotect(struct thread *, struct mprotect_args *); int sys_madvise(struct thread *, struct madvise_args *); int sys_mincore(struct thread *, struct mincore_args *); -int sys_getgroups(struct thread *, struct getgroups_args *); -int sys_setgroups(struct thread *, struct setgroups_args *); int sys_getpgrp(struct thread *, struct getpgrp_args *); int sys_setpgid(struct thread *, struct setpgid_args *); int sys_setitimer(struct thread *, struct setitimer_args *); @@ -2305,6 +2303,8 @@ int sys_setcred(struct thread *, struct setcred_args *); int sys_exterrctl(struct thread *, struct exterrctl_args *); int sys_inotify_add_watch_at(struct thread *, struct inotify_add_watch_at_args *); int sys_inotify_rm_watch(struct thread *, struct inotify_rm_watch_args *); +int sys_getgroups(struct thread *, struct getgroups_args *); +int sys_setgroups(struct thread *, struct setgroups_args *); #ifdef COMPAT_43 @@ -2799,6 +2799,16 @@ int freebsd13_swapoff(struct thread *, struct freebsd13_swapoff_args *); #ifdef COMPAT_FREEBSD14 +struct freebsd14_getgroups_args { + char gidsetsize_l_[PADL_(int)]; int gidsetsize; char gidsetsize_r_[PADR_(int)]; + char gidset_l_[PADL_(gid_t *)]; gid_t * gidset; char gidset_r_[PADR_(gid_t *)]; +}; +struct freebsd14_setgroups_args { + char gidsetsize_l_[PADL_(int)]; int gidsetsize; char gidsetsize_r_[PADR_(int)]; + char gidset_l_[PADL_(const gid_t *)]; const gid_t * gidset; char gidset_r_[PADR_(const gid_t *)]; +}; +int freebsd14_getgroups(struct thread *, struct freebsd14_getgroups_args *); +int freebsd14_setgroups(struct thread *, struct freebsd14_setgroups_args *); #endif /* COMPAT_FREEBSD14 */ @@ -2873,8 +2883,8 @@ int freebsd13_swapoff(struct thread *, struct freebsd13_swapoff_args *); #define SYS_AUE_mprotect AUE_MPROTECT #define SYS_AUE_madvise AUE_MADVISE #define SYS_AUE_mincore AUE_MINCORE -#define SYS_AUE_getgroups AUE_GETGROUPS -#define SYS_AUE_setgroups AUE_SETGROUPS +#define SYS_AUE_freebsd14_getgroups AUE_GETGROUPS +#define SYS_AUE_freebsd14_setgroups AUE_SETGROUPS #define SYS_AUE_getpgrp AUE_GETPGRP #define SYS_AUE_setpgid AUE_SETPGRP #define SYS_AUE_setitimer AUE_SETITIMER @@ -3289,6 +3299,8 @@ int freebsd13_swapoff(struct thread *, struct freebsd13_swapoff_args *); #define SYS_AUE_exterrctl AUE_NULL #define SYS_AUE_inotify_add_watch_at AUE_INOTIFY #define SYS_AUE_inotify_rm_watch AUE_INOTIFY +#define SYS_AUE_getgroups AUE_GETGROUPS +#define SYS_AUE_setgroups AUE_SETGROUPS #undef PAD_ #undef PADL_ diff --git a/sys/sys/watchdog.h b/sys/sys/watchdog.h index 4a16b18509f5..8401d343a6b7 100644 --- a/sys/sys/watchdog.h +++ b/sys/sys/watchdog.h @@ -32,15 +32,16 @@ #define _SYS_WATCHDOG_H #include <sys/ioccom.h> +#include <sys/_types.h> #define _PATH_WATCHDOG "fido" -#define WDIOCPATPAT _IOW('W', 42, u_int) /* pat the watchdog */ -#define WDIOC_SETTIMEOUT _IOW('W', 43, int) /* set/reset the timer */ -#define WDIOC_GETTIMEOUT _IOR('W', 44, int) /* get total timeout */ -#define WDIOC_GETTIMELEFT _IOR('W', 45, int) /* get time left */ -#define WDIOC_GETPRETIMEOUT _IOR('W', 46, int) /* get the pre-timeout */ -#define WDIOC_SETPRETIMEOUT _IOW('W', 47, int) /* set the pre-timeout */ +#define WDIOC_PATPAT _IOW('W', 52, sbintime_t) /* pat the watchdog */ +#define WDIOC_SETTIMEOUT _IOW('W', 53, sbintime_t) /* set/reset the timer */ +#define WDIOC_GETTIMEOUT _IOR('W', 54, sbintime_t) /* get total timeout */ +#define WDIOC_GETTIMELEFT _IOR('W', 55, sbintime_t) /* get time left */ +#define WDIOC_GETPRETIMEOUT _IOR('W', 56, sbintime_t) /* get the pre-timeout */ +#define WDIOC_SETPRETIMEOUT _IOW('W', 57, sbintime_t) /* set the pre-timeout */ /* set the action when a pre-timeout occurs see: WD_SOFT_* */ #define WDIOC_SETPRETIMEOUTACT _IOW('W', 48, int) @@ -48,6 +49,8 @@ #define WDIOC_SETSOFT _IOW('W', 49, int) #define WDIOC_SETSOFTTIMEOUTACT _IOW('W', 50, int) +#define WDIOC_CONTROL _IOW('W', 51, int) /* configure watchdog */ + #define WD_ACTIVE 0x8000000 /* * Watchdog reset, timeout set to value in WD_INTERVAL field. @@ -93,6 +96,11 @@ #define WD_TO_64SEC 36 #define WD_TO_128SEC 37 +/* Control options for WDIOC_CONTROL */ +#define WD_CTRL_DISABLE 0x00000000 +#define WD_CTRL_ENABLE 0x00000001 +#define WD_CTRL_RESET 0x00000002 + /* action on pre-timeout trigger */ #define WD_SOFT_PANIC 0x01 /* panic */ #define WD_SOFT_DDB 0x02 /* enter debugger */ @@ -105,11 +113,16 @@ #include <sys/_eventhandler.h> typedef void (*watchdog_fn)(void *, u_int, int *); +typedef void (*watchdog_sbt_fn)(void *, sbintime_t, sbintime_t *, int *); EVENTHANDLER_DECLARE(watchdog_list, watchdog_fn); +EVENTHANDLER_DECLARE(watchdog_sbt_list, watchdog_sbt_fn); u_int wdog_kern_last_timeout(void); int wdog_kern_pat(u_int utim); +sbintime_t wdog_kern_last_timeout_sbt(void); +int wdog_kern_pat_sbt(sbintime_t utim); +int wdog_control(int ctrl); /* * The following function pointer is used to attach a software watchdog diff --git a/sys/x86/include/apicreg.h b/sys/x86/include/apicreg.h index d610d7f11a1c..1252647fbab3 100644 --- a/sys/x86/include/apicreg.h +++ b/sys/x86/include/apicreg.h @@ -296,6 +296,8 @@ typedef struct IOAPIC ioapic_t; /* constants relating to APIC ID registers */ #define APIC_ID_MASK 0xff000000 #define APIC_ID_SHIFT 24 +#define APIC_EXT_ID_MASK 0x00fe0000 +#define APIC_EXT_ID_SHIFT 17 #define APIC_ID_CLUSTER 0xf0 #define APIC_ID_CLUSTER_ID 0x0f #define APIC_MAX_CLUSTER 0xe |