From de626e68a9a1613fdb184b157953b83428108c32 Mon Sep 17 00:00:00 2001 From: Eugene Grosbein Date: Sun, 16 Aug 2020 11:37:28 +0000 Subject: [PATCH 01/10] Unbreak nvd(8) build as module outside of kernel build environment. --- sys/modules/nvd/Makefile | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/sys/modules/nvd/Makefile b/sys/modules/nvd/Makefile index d47ac918475..950309498bd 100644 --- a/sys/modules/nvd/Makefile +++ b/sys/modules/nvd/Makefile @@ -3,6 +3,6 @@ .PATH: ${SRCTOP}/sys/dev/nvd KMOD= nvd -SRCS= nvd.c opt_geom.h device_if.h bus_if.h +SRCS= nvd.c opt_geom.h device_if.h bus_if.h pci_if.h .include From f5d30f7f76067fa53901af94d92f27781d911521 Mon Sep 17 00:00:00 2001 From: Michael Tuexen Date: Sun, 16 Aug 2020 11:50:37 +0000 Subject: [PATCH 02/10] Improve the handling of concurrent send() calls for SCTP sockets, especially when having the explicit EOR mode enabled. Reported by: Megan2013678@protonmail.com Reported by: syzbot+bc02585076c3cc977f9b@syzkaller.appspotmail.com MFC after: 3 days --- sys/netinet/sctp_input.c | 3 +-- sys/netinet/sctp_output.c | 18 +++++++++++++++--- sys/netinet/sctp_pcb.c | 9 +++++++-- sys/netinet/sctp_structs.h | 1 + sys/netinet/sctp_usrreq.c | 1 + sys/netinet/sctputil.c | 17 +++++------------ sys/netinet/sctputil.h | 2 +- 7 files changed, 31 insertions(+), 20 deletions(-) diff --git a/sys/netinet/sctp_input.c b/sys/netinet/sctp_input.c index 9cb1edba154..6b5ef69439c 100644 --- a/sys/netinet/sctp_input.c +++ b/sys/netinet/sctp_input.c @@ -829,7 +829,6 @@ sctp_handle_abort(struct sctp_abort_chunk *abort, #ifdef SCTP_ASOCLOG_OF_TSNS sctp_print_out_track_log(stcb); #endif - SCTP_ADD_SUBSTATE(stcb, SCTP_STATE_WAS_ABORTED); (void)sctp_free_assoc(stcb->sctp_ep, stcb, SCTP_NORMAL_PROC, SCTP_FROM_SCTP_INPUT + SCTP_LOC_8); SCTPDBG(SCTP_DEBUG_INPUT2, "sctp_handle_abort: finished\n"); @@ -1866,7 +1865,7 @@ sctp_process_cookie_existing(struct mbuf *m, int iphlen, int offset, /* send up all the data */ SCTP_TCB_SEND_LOCK(stcb); - sctp_report_all_outbound(stcb, 0, 1, SCTP_SO_LOCKED); + sctp_report_all_outbound(stcb, 0, SCTP_SO_LOCKED); for (i = 0; i < stcb->asoc.streamoutcnt; i++) { stcb->asoc.strmout[i].chunks_on_queues = 0; #if defined(SCTP_DETAILED_STR_STATS) diff --git a/sys/netinet/sctp_output.c b/sys/netinet/sctp_output.c index 91ee9ee8181..7c0b7b253f4 100644 --- a/sys/netinet/sctp_output.c +++ b/sys/netinet/sctp_output.c @@ -13148,12 +13148,21 @@ skip_preblock: if (sinfo_flags & SCTP_UNORDERED) { SCTP_STAT_INCR(sctps_sends_with_unord); } + sp->processing = 1; TAILQ_INSERT_TAIL(&strm->outqueue, sp, next); stcb->asoc.ss_functions.sctp_ss_add_to_stream(stcb, asoc, strm, sp, 1); SCTP_TCB_SEND_UNLOCK(stcb); } else { SCTP_TCB_SEND_LOCK(stcb); sp = TAILQ_LAST(&strm->outqueue, sctp_streamhead); + if (sp->processing) { + SCTP_TCB_SEND_UNLOCK(stcb); + SCTP_LTRACE_ERR_RET(inp, stcb, net, SCTP_FROM_SCTP_OUTPUT, EINVAL); + error = EINVAL; + goto out; + } else { + sp->processing = 1; + } SCTP_TCB_SEND_UNLOCK(stcb); if (sp == NULL) { /* ???? Huh ??? last msg is gone */ @@ -13195,13 +13204,14 @@ skip_preblock: } /* Update the mbuf and count */ SCTP_TCB_SEND_LOCK(stcb); - if (stcb->asoc.state & SCTP_STATE_ABOUT_TO_BE_FREED) { + if ((stcb->asoc.state & SCTP_STATE_ABOUT_TO_BE_FREED) || + (stcb->asoc.state & SCTP_STATE_WAS_ABORTED)) { /* * we need to get out. Peer probably * aborted. */ sctp_m_freem(mm); - if (stcb->asoc.state & SCTP_PCB_FLAGS_WAS_ABORTED) { + if (stcb->asoc.state & SCTP_STATE_WAS_ABORTED) { SCTP_LTRACE_ERR_RET(NULL, stcb, NULL, SCTP_FROM_SCTP_OUTPUT, ECONNRESET); error = ECONNRESET; } @@ -13405,7 +13415,8 @@ skip_preblock: } } SCTP_TCB_SEND_LOCK(stcb); - if (stcb->asoc.state & SCTP_STATE_ABOUT_TO_BE_FREED) { + if ((stcb->asoc.state & SCTP_STATE_ABOUT_TO_BE_FREED) || + (stcb->asoc.state & SCTP_STATE_WAS_ABORTED)) { SCTP_TCB_SEND_UNLOCK(stcb); goto out_unlocked; } @@ -13421,6 +13432,7 @@ skip_preblock: strm->last_msg_incomplete = 0; asoc->stream_locked = 0; } + sp->processing = 0; } else { SCTP_PRINTF("Huh no sp TSNH?\n"); strm->last_msg_incomplete = 0; diff --git a/sys/netinet/sctp_pcb.c b/sys/netinet/sctp_pcb.c index 73926f2b4cb..aa70abbe7de 100644 --- a/sys/netinet/sctp_pcb.c +++ b/sys/netinet/sctp_pcb.c @@ -4733,6 +4733,7 @@ sctp_free_assoc(struct sctp_inpcb *inp, struct sctp_tcb *stcb, int from_inpcbfre /* there is no asoc, really TSNH :-0 */ return (1); } + SCTP_TCB_SEND_LOCK(stcb); if (stcb->asoc.alternate) { sctp_free_remote_addr(stcb->asoc.alternate); stcb->asoc.alternate = NULL; @@ -4767,6 +4768,7 @@ sctp_free_assoc(struct sctp_inpcb *inp, struct sctp_tcb *stcb, int from_inpcbfre /* nope, reader or writer in the way */ sctp_timer_start(SCTP_TIMER_TYPE_ASOCKILL, inp, stcb, NULL); /* no asoc destroyed */ + SCTP_TCB_SEND_UNLOCK(stcb); SCTP_TCB_UNLOCK(stcb); #ifdef SCTP_LOG_CLOSING sctp_log_closing(inp, stcb, 8); @@ -4835,6 +4837,7 @@ sctp_free_assoc(struct sctp_inpcb *inp, struct sctp_tcb *stcb, int from_inpcbfre SCTP_CLEAR_SUBSTATE(stcb, SCTP_STATE_IN_ACCEPT_QUEUE); sctp_timer_start(SCTP_TIMER_TYPE_ASOCKILL, inp, stcb, NULL); } + SCTP_TCB_SEND_UNLOCK(stcb); SCTP_TCB_UNLOCK(stcb); if ((inp->sctp_flags & SCTP_PCB_FLAGS_SOCKET_ALLGONE) || (inp->sctp_flags & SCTP_PCB_FLAGS_SOCKET_GONE)) @@ -4868,10 +4871,12 @@ sctp_free_assoc(struct sctp_inpcb *inp, struct sctp_tcb *stcb, int from_inpcbfre if (from_inpcbfree == SCTP_NORMAL_PROC) { atomic_add_int(&stcb->asoc.refcnt, 1); + SCTP_TCB_SEND_UNLOCK(stcb); SCTP_TCB_UNLOCK(stcb); SCTP_INP_INFO_WLOCK(); SCTP_INP_WLOCK(inp); SCTP_TCB_LOCK(stcb); + SCTP_TCB_SEND_LOCK(stcb); } /* Double check the GONE flag */ if ((inp->sctp_flags & SCTP_PCB_FLAGS_SOCKET_ALLGONE) || @@ -4919,6 +4924,7 @@ sctp_free_assoc(struct sctp_inpcb *inp, struct sctp_tcb *stcb, int from_inpcbfre SCTP_INP_INFO_WUNLOCK(); SCTP_INP_WUNLOCK(inp); } + SCTP_TCB_SEND_UNLOCK(stcb); SCTP_TCB_UNLOCK(stcb); return (0); } @@ -4950,7 +4956,6 @@ sctp_free_assoc(struct sctp_inpcb *inp, struct sctp_tcb *stcb, int from_inpcbfre * in case. */ /* anything on the wheel needs to be removed */ - SCTP_TCB_SEND_LOCK(stcb); for (i = 0; i < asoc->streamoutcnt; i++) { struct sctp_stream_out *outs; @@ -4981,7 +4986,6 @@ sctp_free_assoc(struct sctp_inpcb *inp, struct sctp_tcb *stcb, int from_inpcbfre sctp_free_a_strmoq(stcb, sp, SCTP_SO_LOCKED); } } - SCTP_TCB_SEND_UNLOCK(stcb); /* sa_ignore FREED_MEMORY */ TAILQ_FOREACH_SAFE(strrst, &asoc->resetHead, next_resp, nstrrst) { TAILQ_REMOVE(&asoc->resetHead, strrst, next_resp); @@ -5183,6 +5187,7 @@ sctp_free_assoc(struct sctp_inpcb *inp, struct sctp_tcb *stcb, int from_inpcbfre /* Insert new items here :> */ /* Get rid of LOCK */ + SCTP_TCB_SEND_UNLOCK(stcb); SCTP_TCB_UNLOCK(stcb); SCTP_TCB_LOCK_DESTROY(stcb); SCTP_TCB_SEND_LOCK_DESTROY(stcb); diff --git a/sys/netinet/sctp_structs.h b/sys/netinet/sctp_structs.h index 82429ec4278..d35388d31a4 100644 --- a/sys/netinet/sctp_structs.h +++ b/sys/netinet/sctp_structs.h @@ -534,6 +534,7 @@ struct sctp_stream_queue_pending { uint8_t sender_all_done; uint8_t put_last_out; uint8_t discard_rest; + uint8_t processing; }; /* diff --git a/sys/netinet/sctp_usrreq.c b/sys/netinet/sctp_usrreq.c index 9d9320608fb..5437f426883 100644 --- a/sys/netinet/sctp_usrreq.c +++ b/sys/netinet/sctp_usrreq.c @@ -190,6 +190,7 @@ sctp_notify(struct sctp_inpcb *inp, } else if ((icmp_code == ICMP_UNREACH_PROTOCOL) || (icmp_code == ICMP_UNREACH_PORT)) { /* Treat it like an ABORT. */ + SCTP_ADD_SUBSTATE(stcb, SCTP_STATE_WAS_ABORTED); sctp_abort_notification(stcb, 1, 0, NULL, SCTP_SO_NOT_LOCKED); (void)sctp_free_assoc(inp, stcb, SCTP_NORMAL_PROC, SCTP_FROM_SCTP_USRREQ + SCTP_LOC_2); diff --git a/sys/netinet/sctputil.c b/sys/netinet/sctputil.c index 74e64566009..a4072561e81 100644 --- a/sys/netinet/sctputil.c +++ b/sys/netinet/sctputil.c @@ -4239,7 +4239,7 @@ sctp_ulp_notify(uint32_t notification, struct sctp_tcb *stcb, } void -sctp_report_all_outbound(struct sctp_tcb *stcb, uint16_t error, int holds_lock, int so_locked) +sctp_report_all_outbound(struct sctp_tcb *stcb, uint16_t error, int so_locked) { struct sctp_association *asoc; struct sctp_stream_out *outs; @@ -4261,9 +4261,6 @@ sctp_report_all_outbound(struct sctp_tcb *stcb, uint16_t error, int holds_lock, return; } /* now through all the gunk freeing chunks */ - if (holds_lock == 0) { - SCTP_TCB_SEND_LOCK(stcb); - } /* sent queue SHOULD be empty */ TAILQ_FOREACH_SAFE(chk, &asoc->sent_queue, sctp_next, nchk) { TAILQ_REMOVE(&asoc->sent_queue, chk, sctp_next); @@ -4340,10 +4337,6 @@ sctp_report_all_outbound(struct sctp_tcb *stcb, uint16_t error, int holds_lock, /* sa_ignore FREED_MEMORY */ } } - - if (holds_lock == 0) { - SCTP_TCB_SEND_UNLOCK(stcb); - } } void @@ -4363,8 +4356,11 @@ sctp_abort_notification(struct sctp_tcb *stcb, uint8_t from_peer, uint16_t error (stcb->asoc.state & SCTP_STATE_CLOSED_SOCKET)) { return; } + SCTP_TCB_SEND_LOCK(stcb); + SCTP_ADD_SUBSTATE(stcb, SCTP_STATE_WAS_ABORTED); /* Tell them we lost the asoc */ - sctp_report_all_outbound(stcb, error, 0, so_locked); + sctp_report_all_outbound(stcb, error, so_locked); + SCTP_TCB_SEND_UNLOCK(stcb); if (from_peer) { sctp_ulp_notify(SCTP_NOTIFY_ASSOC_REM_ABORTED, stcb, error, abort, so_locked); } else { @@ -4393,7 +4389,6 @@ sctp_abort_association(struct sctp_inpcb *inp, struct sctp_tcb *stcb, if (stcb != NULL) { /* We have a TCB to abort, send notification too */ sctp_abort_notification(stcb, 0, 0, NULL, SCTP_SO_NOT_LOCKED); - SCTP_ADD_SUBSTATE(stcb, SCTP_STATE_WAS_ABORTED); /* Ok, now lets free it */ SCTP_STAT_INCR_COUNTER32(sctps_aborted); if ((SCTP_GET_STATE(stcb) == SCTP_STATE_OPEN) || @@ -4482,8 +4477,6 @@ sctp_abort_an_association(struct sctp_inpcb *inp, struct sctp_tcb *stcb, } } return; - } else { - SCTP_ADD_SUBSTATE(stcb, SCTP_STATE_WAS_ABORTED); } /* notify the peer */ sctp_send_abort_tcb(stcb, op_err, so_locked); diff --git a/sys/netinet/sctputil.h b/sys/netinet/sctputil.h index b0b1e622f8b..7a6d3b8e5b9 100644 --- a/sys/netinet/sctputil.h +++ b/sys/netinet/sctputil.h @@ -164,7 +164,7 @@ void sctp_stop_timers_for_shutdown(struct sctp_tcb *); /* Stop all timers for association and remote addresses. */ void sctp_stop_association_timers(struct sctp_tcb *, bool); -void sctp_report_all_outbound(struct sctp_tcb *, uint16_t, int, int); +void sctp_report_all_outbound(struct sctp_tcb *, uint16_t, int); int sctp_expand_mapping_array(struct sctp_association *, uint32_t); From 1aca6504534c2339789cc8cce84378ab4947524d Mon Sep 17 00:00:00 2001 From: Eugene Grosbein Date: Sun, 16 Aug 2020 12:59:43 +0000 Subject: [PATCH 03/10] Unbreak building cp2112(8) as a module outside of kernel build environment. --- sys/modules/usb/cp2112/Makefile | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/sys/modules/usb/cp2112/Makefile b/sys/modules/usb/cp2112/Makefile index dd0213842ec..eb9652f8797 100644 --- a/sys/modules/usb/cp2112/Makefile +++ b/sys/modules/usb/cp2112/Makefile @@ -31,7 +31,7 @@ S= ${SRCTOP}/sys KMOD= cp2112 SRCS= cp2112.c -SRCS+= opt_bus.h opt_usb.h +SRCS+= opt_bus.h opt_platform.h opt_usb.h SRCS+= device_if.h bus_if.h gpio_if.h iicbus_if.h usb_if.h usbdevs.h .include From d59f3890c36601f4f62bb4a0b7db977e42339588 Mon Sep 17 00:00:00 2001 From: Michael Tuexen Date: Sun, 16 Aug 2020 13:31:14 +0000 Subject: [PATCH 04/10] Remove a line which is needed and was added in https://svnweb.freebsd.org/changeset/base/364268 MFC after: 3 days --- sys/netinet/sctp_usrreq.c | 1 - 1 file changed, 1 deletion(-) diff --git a/sys/netinet/sctp_usrreq.c b/sys/netinet/sctp_usrreq.c index 5437f426883..9d9320608fb 100644 --- a/sys/netinet/sctp_usrreq.c +++ b/sys/netinet/sctp_usrreq.c @@ -190,7 +190,6 @@ sctp_notify(struct sctp_inpcb *inp, } else if ((icmp_code == ICMP_UNREACH_PROTOCOL) || (icmp_code == ICMP_UNREACH_PORT)) { /* Treat it like an ABORT. */ - SCTP_ADD_SUBSTATE(stcb, SCTP_STATE_WAS_ABORTED); sctp_abort_notification(stcb, 1, 0, NULL, SCTP_SO_NOT_LOCKED); (void)sctp_free_assoc(inp, stcb, SCTP_NORMAL_PROC, SCTP_FROM_SCTP_USRREQ + SCTP_LOC_2); From a92a971bbb94ad5b44e2a5bbdc669ad3ae762c8d Mon Sep 17 00:00:00 2001 From: Mateusz Guzik Date: Sun, 16 Aug 2020 17:18:54 +0000 Subject: [PATCH 05/10] vfs: remove the thread argument from vget It was already asserted to be curthread. Semantic patch: @@ expression arg1, arg2, arg3; @@ - vget(arg1, arg2, arg3) + vget(arg1, arg2) --- sys/compat/linuxkpi/common/include/linux/fs.h | 2 +- sys/fs/autofs/autofs_vnops.c | 2 +- sys/fs/ext2fs/ext2_vfsops.c | 4 ++-- sys/fs/fdescfs/fdesc_vfsops.c | 2 +- sys/fs/fdescfs/fdesc_vnops.c | 4 ++-- sys/fs/fuse/fuse_vfsops.c | 2 +- sys/fs/msdosfs/msdosfs_vfsops.c | 2 +- sys/fs/nfsclient/nfs_clvfsops.c | 2 +- sys/fs/nullfs/null_vfsops.c | 2 +- sys/fs/pseudofs/pseudofs_vncache.c | 4 ++-- sys/fs/smbfs/smbfs_node.c | 2 +- sys/fs/smbfs/smbfs_vfsops.c | 2 +- sys/fs/tmpfs/tmpfs_subr.c | 2 +- sys/fs/tmpfs/tmpfs_vfsops.c | 3 +-- sys/kern/uipc_mqueue.c | 2 +- sys/kern/vfs_default.c | 2 +- sys/kern/vfs_subr.c | 6 ++---- sys/sys/vnode.h | 2 +- sys/ufs/ffs/ffs_alloc.c | 2 +- sys/ufs/ffs/ffs_softdep.c | 3 +-- sys/ufs/ffs/ffs_vfsops.c | 7 +++---- sys/ufs/ufs/ufs_quota.c | 7 +++---- sys/vm/vm_fault.c | 4 ++-- sys/vm/vm_mmap.c | 4 ++-- sys/vm/vm_pageout.c | 2 +- 25 files changed, 35 insertions(+), 41 deletions(-) diff --git a/sys/compat/linuxkpi/common/include/linux/fs.h b/sys/compat/linuxkpi/common/include/linux/fs.h index 7f5993bd775..f400fc7cbfd 100644 --- a/sys/compat/linuxkpi/common/include/linux/fs.h +++ b/sys/compat/linuxkpi/common/include/linux/fs.h @@ -260,7 +260,7 @@ igrab(struct inode *inode) { int error; - error = vget(inode, 0, curthread); + error = vget(inode, 0); if (error) return (NULL); diff --git a/sys/fs/autofs/autofs_vnops.c b/sys/fs/autofs/autofs_vnops.c index e9e058933e5..a9bfcc56b65 100644 --- a/sys/fs/autofs/autofs_vnops.c +++ b/sys/fs/autofs/autofs_vnops.c @@ -656,7 +656,7 @@ autofs_node_vn(struct autofs_node *anp, struct mount *mp, int flags, vp = anp->an_vnode; if (vp != NULL) { - error = vget(vp, flags | LK_RETRY, curthread); + error = vget(vp, flags | LK_RETRY); if (error != 0) { AUTOFS_WARN("vget failed with error %d", error); sx_xunlock(&anp->an_vnode_lock); diff --git a/sys/fs/ext2fs/ext2_vfsops.c b/sys/fs/ext2fs/ext2_vfsops.c index f40fb7202e4..5aff9bdb2ff 100644 --- a/sys/fs/ext2fs/ext2_vfsops.c +++ b/sys/fs/ext2fs/ext2_vfsops.c @@ -799,7 +799,7 @@ loop: /* * Step 4: invalidate all cached file data. */ - if (vget(vp, LK_EXCLUSIVE | LK_INTERLOCK, td)) { + if (vget(vp, LK_EXCLUSIVE | LK_INTERLOCK)) { MNT_VNODE_FOREACH_ALL_ABORT(mp, mvp); goto loop; } @@ -1158,7 +1158,7 @@ loop: VI_UNLOCK(vp); continue; } - error = vget(vp, LK_EXCLUSIVE | LK_NOWAIT | LK_INTERLOCK, td); + error = vget(vp, LK_EXCLUSIVE | LK_NOWAIT | LK_INTERLOCK); if (error) { if (error == ENOENT) { MNT_VNODE_FOREACH_ALL_ABORT(mp, mvp); diff --git a/sys/fs/fdescfs/fdesc_vfsops.c b/sys/fs/fdescfs/fdesc_vfsops.c index e6aa9b9a234..e7878f115ac 100644 --- a/sys/fs/fdescfs/fdesc_vfsops.c +++ b/sys/fs/fdescfs/fdesc_vfsops.c @@ -163,7 +163,7 @@ fdesc_root(struct mount *mp, int flags, struct vnode **vpp) * Return locked reference to root. */ vp = VFSTOFDESC(mp)->f_root; - vget(vp, LK_EXCLUSIVE | LK_RETRY, curthread); + vget(vp, LK_EXCLUSIVE | LK_RETRY); *vpp = vp; return (0); } diff --git a/sys/fs/fdescfs/fdesc_vnops.c b/sys/fs/fdescfs/fdesc_vnops.c index 732d679f6de..87fcb7160bb 100644 --- a/sys/fs/fdescfs/fdesc_vnops.c +++ b/sys/fs/fdescfs/fdesc_vnops.c @@ -182,7 +182,7 @@ loop: vp = fd->fd_vnode; VI_LOCK(vp); mtx_unlock(&fdesc_hashmtx); - if (vget(vp, LK_EXCLUSIVE | LK_INTERLOCK, td)) + if (vget(vp, LK_EXCLUSIVE | LK_INTERLOCK)) goto loop; *vpp = vp; return (0); @@ -232,7 +232,7 @@ loop: vp2 = fd2->fd_vnode; VI_LOCK(vp2); mtx_unlock(&fdesc_hashmtx); - error = vget(vp2, LK_EXCLUSIVE | LK_INTERLOCK, td); + error = vget(vp2, LK_EXCLUSIVE | LK_INTERLOCK); /* Someone beat us, dec use count and wait for reclaim */ vgone(vp); vput(vp); diff --git a/sys/fs/fuse/fuse_vfsops.c b/sys/fs/fuse/fuse_vfsops.c index 93e2b922b76..ce7ffeda249 100644 --- a/sys/fs/fuse/fuse_vfsops.c +++ b/sys/fs/fuse/fuse_vfsops.c @@ -610,7 +610,7 @@ fuse_vfsop_root(struct mount *mp, int lkflags, struct vnode **vpp) int err = 0; if (data->vroot != NULL) { - err = vget(data->vroot, lkflags, curthread); + err = vget(data->vroot, lkflags); if (err == 0) *vpp = data->vroot; } else { diff --git a/sys/fs/msdosfs/msdosfs_vfsops.c b/sys/fs/msdosfs/msdosfs_vfsops.c index 2d0c10d519e..8d068646906 100644 --- a/sys/fs/msdosfs/msdosfs_vfsops.c +++ b/sys/fs/msdosfs/msdosfs_vfsops.c @@ -921,7 +921,7 @@ loop: VI_UNLOCK(vp); continue; } - error = vget(vp, LK_EXCLUSIVE | LK_NOWAIT | LK_INTERLOCK, td); + error = vget(vp, LK_EXCLUSIVE | LK_NOWAIT | LK_INTERLOCK); if (error) { if (error == ENOENT) { MNT_VNODE_FOREACH_ALL_ABORT(mp, nvp); diff --git a/sys/fs/nfsclient/nfs_clvfsops.c b/sys/fs/nfsclient/nfs_clvfsops.c index e97d42f4a38..3a12af68bfe 100644 --- a/sys/fs/nfsclient/nfs_clvfsops.c +++ b/sys/fs/nfsclient/nfs_clvfsops.c @@ -1828,7 +1828,7 @@ loop: VI_UNLOCK(vp); continue; } - if (vget(vp, LK_EXCLUSIVE | LK_INTERLOCK, td)) { + if (vget(vp, LK_EXCLUSIVE | LK_INTERLOCK)) { MNT_VNODE_FOREACH_ALL_ABORT(mp, mvp); goto loop; } diff --git a/sys/fs/nullfs/null_vfsops.c b/sys/fs/nullfs/null_vfsops.c index cdce2872e23..a3d23d35018 100644 --- a/sys/fs/nullfs/null_vfsops.c +++ b/sys/fs/nullfs/null_vfsops.c @@ -281,7 +281,7 @@ nullfs_root(mp, flags, vpp) NULLFSDEBUG("nullfs_root(mp = %p, vp = %p)\n", mp, mntdata->nullm_lowerrootvp); - error = vget(mntdata->nullm_lowerrootvp, flags, curthread); + error = vget(mntdata->nullm_lowerrootvp, flags); if (error == 0) { error = null_nodeget(mp, mntdata->nullm_lowerrootvp, &vp); if (error == 0) { diff --git a/sys/fs/pseudofs/pseudofs_vncache.c b/sys/fs/pseudofs/pseudofs_vncache.c index 6fd73babaf3..8d6f56f6fe2 100644 --- a/sys/fs/pseudofs/pseudofs_vncache.c +++ b/sys/fs/pseudofs/pseudofs_vncache.c @@ -136,7 +136,7 @@ retry: vp = pvd->pvd_vnode; VI_LOCK(vp); mtx_unlock(&pfs_vncache_mutex); - if (vget(vp, LK_EXCLUSIVE | LK_INTERLOCK, curthread) == 0) { + if (vget(vp, LK_EXCLUSIVE | LK_INTERLOCK) == 0) { ++pfs_vncache_hits; *vpp = vp; /* @@ -218,7 +218,7 @@ retry2: vp = pvd2->pvd_vnode; VI_LOCK(vp); mtx_unlock(&pfs_vncache_mutex); - if (vget(vp, LK_EXCLUSIVE | LK_INTERLOCK, curthread) == 0) { + if (vget(vp, LK_EXCLUSIVE | LK_INTERLOCK) == 0) { ++pfs_vncache_hits; vgone(*vpp); vput(*vpp); diff --git a/sys/fs/smbfs/smbfs_node.c b/sys/fs/smbfs/smbfs_node.c index 16b5de93846..f544d1019e8 100644 --- a/sys/fs/smbfs/smbfs_node.c +++ b/sys/fs/smbfs/smbfs_node.c @@ -124,7 +124,7 @@ smbfs_node_alloc(struct mount *mp, struct vnode *dvp, const char *dirnm, if (dvp == NULL) return EINVAL; vp = VTOSMB(VTOSMB(dvp)->n_parent)->n_vnode; - error = vget(vp, LK_EXCLUSIVE, td); + error = vget(vp, LK_EXCLUSIVE); if (error == 0) *vpp = vp; return error; diff --git a/sys/fs/smbfs/smbfs_vfsops.c b/sys/fs/smbfs/smbfs_vfsops.c index a85f0672052..12a5525990f 100644 --- a/sys/fs/smbfs/smbfs_vfsops.c +++ b/sys/fs/smbfs/smbfs_vfsops.c @@ -329,7 +329,7 @@ smbfs_root(struct mount *mp, int flags, struct vnode **vpp) if (smp->sm_root) { *vpp = SMBTOV(smp->sm_root); - return vget(*vpp, LK_EXCLUSIVE | LK_RETRY, td); + return vget(*vpp, LK_EXCLUSIVE | LK_RETRY); } scred = smbfs_malloc_scred(); smb_makescred(scred, td, cred); diff --git a/sys/fs/tmpfs/tmpfs_subr.c b/sys/fs/tmpfs/tmpfs_subr.c index 55a6390f683..2bf1cb9cc69 100644 --- a/sys/fs/tmpfs/tmpfs_subr.c +++ b/sys/fs/tmpfs/tmpfs_subr.c @@ -620,7 +620,7 @@ loop: goto loop; } TMPFS_NODE_UNLOCK(node); - error = vget(vp, lkflag | LK_INTERLOCK, curthread); + error = vget(vp, lkflag | LK_INTERLOCK); if (error == ENOENT) { TMPFS_NODE_LOCK(node); goto loop; diff --git a/sys/fs/tmpfs/tmpfs_vfsops.c b/sys/fs/tmpfs/tmpfs_vfsops.c index bb2ae154af9..c98b6126eec 100644 --- a/sys/fs/tmpfs/tmpfs_vfsops.c +++ b/sys/fs/tmpfs/tmpfs_vfsops.c @@ -130,8 +130,7 @@ tmpfs_update_mtime(struct mount *mp, bool lazy) * metadata changes now. */ if (!lazy || obj->generation != obj->cleangeneration) { - if (vget(vp, LK_EXCLUSIVE | LK_INTERLOCK, - curthread) != 0) + if (vget(vp, LK_EXCLUSIVE | LK_INTERLOCK) != 0) continue; tmpfs_check_mtime(vp); if (!lazy) diff --git a/sys/kern/uipc_mqueue.c b/sys/kern/uipc_mqueue.c index 018630ca7da..28fc1d84329 100644 --- a/sys/kern/uipc_mqueue.c +++ b/sys/kern/uipc_mqueue.c @@ -754,7 +754,7 @@ mqfs_allocv(struct mount *mp, struct vnode **vpp, struct mqfs_node *pn) found: *vpp = vd->mv_vnode; sx_xunlock(&mqfs->mi_lock); - error = vget(*vpp, LK_RETRY | LK_EXCLUSIVE, curthread); + error = vget(*vpp, LK_RETRY | LK_EXCLUSIVE); vdrop(*vpp); return (error); } diff --git a/sys/kern/vfs_default.c b/sys/kern/vfs_default.c index 57465506d7d..c8e2947b613 100644 --- a/sys/kern/vfs_default.c +++ b/sys/kern/vfs_default.c @@ -1348,7 +1348,7 @@ loop: VI_UNLOCK(vp); continue; } - if ((error = vget(vp, lockreq, td)) != 0) { + if ((error = vget(vp, lockreq)) != 0) { if (error == ENOENT) { MNT_VNODE_FOREACH_ALL_ABORT(mp, mvp); goto loop; diff --git a/sys/kern/vfs_subr.c b/sys/kern/vfs_subr.c index 03bfa129b16..a8a26e28a0a 100644 --- a/sys/kern/vfs_subr.c +++ b/sys/kern/vfs_subr.c @@ -2856,12 +2856,10 @@ vget_abort(struct vnode *vp, enum vgetstate vs) } int -vget(struct vnode *vp, int flags, struct thread *td) +vget(struct vnode *vp, int flags) { enum vgetstate vs; - MPASS(td == curthread); - vs = vget_prep(vp); return (vget_finish(vp, flags, vs)); } @@ -4684,7 +4682,7 @@ vfs_periodic_msync_inactive(struct mount *mp, int flags) VI_UNLOCK(vp); continue; } - if (vget(vp, lkflags, td) == 0) { + if (vget(vp, lkflags) == 0) { obj = vp->v_object; if (obj != NULL && (vp->v_vflag & VV_NOSYNC) == 0) { VM_OBJECT_WLOCK(obj); diff --git a/sys/sys/vnode.h b/sys/sys/vnode.h index 9d736483d00..7d87f70da53 100644 --- a/sys/sys/vnode.h +++ b/sys/sys/vnode.h @@ -680,7 +680,7 @@ void vlazy(struct vnode *); void vdrop(struct vnode *); void vdropl(struct vnode *); int vflush(struct mount *mp, int rootrefs, int flags, struct thread *td); -int vget(struct vnode *vp, int flags, struct thread *td); +int vget(struct vnode *vp, int flags); enum vgetstate vget_prep_smr(struct vnode *vp); enum vgetstate vget_prep(struct vnode *vp); int vget_finish(struct vnode *vp, int flags, enum vgetstate vs); diff --git a/sys/ufs/ffs/ffs_alloc.c b/sys/ufs/ffs/ffs_alloc.c index d8ccdec718a..221f3bf157f 100644 --- a/sys/ufs/ffs/ffs_alloc.c +++ b/sys/ufs/ffs/ffs_alloc.c @@ -3463,7 +3463,7 @@ sysctl_ffs_fsck(SYSCTL_HANDLER_ARGS) */ pwd = pwd_hold(td); dvp = pwd->pwd_cdir; - if ((error = vget(dvp, LK_EXCLUSIVE, td)) != 0) { + if ((error = vget(dvp, LK_EXCLUSIVE)) != 0) { vput(fdvp); pwd_drop(pwd); break; diff --git a/sys/ufs/ffs/ffs_softdep.c b/sys/ufs/ffs/ffs_softdep.c index f97b08160db..577ddfa7e8d 100644 --- a/sys/ufs/ffs/ffs_softdep.c +++ b/sys/ufs/ffs/ffs_softdep.c @@ -13624,8 +13624,7 @@ softdep_request_cleanup_flush(mp, ump) VI_UNLOCK(lvp); continue; } - if (vget(lvp, LK_EXCLUSIVE | LK_INTERLOCK | LK_NOWAIT, - td) != 0) { + if (vget(lvp, LK_EXCLUSIVE | LK_INTERLOCK | LK_NOWAIT) != 0) { failed_vnode = 1; continue; } diff --git a/sys/ufs/ffs/ffs_vfsops.c b/sys/ufs/ffs/ffs_vfsops.c index 834930b5e99..e02af7499c1 100644 --- a/sys/ufs/ffs/ffs_vfsops.c +++ b/sys/ufs/ffs/ffs_vfsops.c @@ -974,7 +974,7 @@ loop: /* * Step 4: invalidate all cached file data. */ - if (vget(vp, LK_EXCLUSIVE | LK_INTERLOCK, td)) { + if (vget(vp, LK_EXCLUSIVE | LK_INTERLOCK)) { MNT_VNODE_FOREACH_ALL_ABORT(mp, mvp); goto loop; } @@ -1758,8 +1758,7 @@ ffs_sync_lazy(mp) VI_UNLOCK(vp); continue; } - if ((error = vget(vp, LK_EXCLUSIVE | LK_NOWAIT | LK_INTERLOCK, - td)) != 0) + if ((error = vget(vp, LK_EXCLUSIVE | LK_NOWAIT | LK_INTERLOCK)) != 0) continue; #ifdef QUOTA qsyncvp(vp); @@ -1856,7 +1855,7 @@ loop: VI_UNLOCK(vp); continue; } - if ((error = vget(vp, lockreq, td)) != 0) { + if ((error = vget(vp, lockreq)) != 0) { if (error == ENOENT || error == ENOLCK) { MNT_VNODE_FOREACH_ALL_ABORT(mp, mvp); goto loop; diff --git a/sys/ufs/ufs/ufs_quota.c b/sys/ufs/ufs/ufs_quota.c index f59ff2a1c18..47c8133adc1 100644 --- a/sys/ufs/ufs/ufs_quota.c +++ b/sys/ufs/ufs/ufs_quota.c @@ -613,7 +613,7 @@ quotaon(struct thread *td, struct mount *mp, int type, void *fname) */ again: MNT_VNODE_FOREACH_ALL(vp, mp, mvp) { - if (vget(vp, LK_EXCLUSIVE | LK_INTERLOCK, td)) { + if (vget(vp, LK_EXCLUSIVE | LK_INTERLOCK)) { MNT_VNODE_FOREACH_ALL_ABORT(mp, mvp); goto again; } @@ -680,7 +680,7 @@ again: VI_UNLOCK(vp); continue; } - if (vget(vp, LK_EXCLUSIVE | LK_INTERLOCK, td)) { + if (vget(vp, LK_EXCLUSIVE | LK_INTERLOCK)) { MNT_VNODE_FOREACH_ALL_ABORT(mp, mvp); goto again; } @@ -1064,7 +1064,6 @@ int qsync(struct mount *mp) { struct ufsmount *ump = VFSTOUFS(mp); - struct thread *td = curthread; /* XXX */ struct vnode *vp, *mvp; struct dquot *dq; int i, error; @@ -1088,7 +1087,7 @@ again: VI_UNLOCK(vp); continue; } - error = vget(vp, LK_EXCLUSIVE | LK_INTERLOCK, td); + error = vget(vp, LK_EXCLUSIVE | LK_INTERLOCK); if (error) { if (error == ENOENT) { MNT_VNODE_FOREACH_ALL_ABORT(mp, mvp); diff --git a/sys/vm/vm_fault.c b/sys/vm/vm_fault.c index d28ec3a366e..cc67e131f2d 100644 --- a/sys/vm/vm_fault.c +++ b/sys/vm/vm_fault.c @@ -673,7 +673,7 @@ vm_fault_lock_vnode(struct faultstate *fs, bool objlocked) * paging-in-progress count incremented. Otherwise, we could * deadlock. */ - error = vget(vp, locked | LK_CANRECURSE | LK_NOWAIT, curthread); + error = vget(vp, locked | LK_CANRECURSE | LK_NOWAIT); if (error == 0) { fs->vp = vp; return (KERN_SUCCESS); @@ -684,7 +684,7 @@ vm_fault_lock_vnode(struct faultstate *fs, bool objlocked) unlock_and_deallocate(fs); else fault_deallocate(fs); - error = vget(vp, locked | LK_RETRY | LK_CANRECURSE, curthread); + error = vget(vp, locked | LK_RETRY | LK_CANRECURSE); vdrop(vp); fs->vp = vp; KASSERT(error == 0, ("vm_fault: vget failed %d", error)); diff --git a/sys/vm/vm_mmap.c b/sys/vm/vm_mmap.c index 7add69a58f0..24f542edfa3 100644 --- a/sys/vm/vm_mmap.c +++ b/sys/vm/vm_mmap.c @@ -1284,7 +1284,7 @@ vm_mmap_vnode(struct thread *td, vm_size_t objsize, cred = td->td_ucred; writex = (*maxprotp & VM_PROT_WRITE) != 0 && (*flagsp & MAP_SHARED) != 0; - if ((error = vget(vp, LK_SHARED, td)) != 0) + if ((error = vget(vp, LK_SHARED)) != 0) return (error); AUDIT_ARG_VNODE1(vp); foff = *foffp; @@ -1305,7 +1305,7 @@ vm_mmap_vnode(struct thread *td, vm_size_t objsize, * Bypass filesystems obey the mpsafety of the * underlying fs. Tmpfs never bypasses. */ - error = vget(vp, LK_SHARED, td); + error = vget(vp, LK_SHARED); if (error != 0) return (error); } diff --git a/sys/vm/vm_pageout.c b/sys/vm/vm_pageout.c index a882a17f3f3..34a3a4f9048 100644 --- a/sys/vm/vm_pageout.c +++ b/sys/vm/vm_pageout.c @@ -643,7 +643,7 @@ vm_pageout_clean(vm_page_t m, int *numpagedout) VM_OBJECT_WUNLOCK(object); lockmode = MNT_SHARED_WRITES(vp->v_mount) ? LK_SHARED : LK_EXCLUSIVE; - if (vget(vp, lockmode | LK_TIMELOCK, curthread)) { + if (vget(vp, lockmode | LK_TIMELOCK)) { vp = NULL; error = EDEADLK; goto unlock_mp; From 1abe36567f289e660ac5c55b97598272b8a1334c Mon Sep 17 00:00:00 2001 From: Mateusz Guzik Date: Sun, 16 Aug 2020 17:19:23 +0000 Subject: [PATCH 06/10] tmpfs: use vget_prep/vget_finish instead of vget + vnode --- sys/fs/tmpfs/tmpfs_subr.c | 7 +++---- 1 file changed, 3 insertions(+), 4 deletions(-) diff --git a/sys/fs/tmpfs/tmpfs_subr.c b/sys/fs/tmpfs/tmpfs_subr.c index 2bf1cb9cc69..f691fc12822 100644 --- a/sys/fs/tmpfs/tmpfs_subr.c +++ b/sys/fs/tmpfs/tmpfs_subr.c @@ -588,6 +588,7 @@ tmpfs_alloc_vp(struct mount *mp, struct tmpfs_node *node, int lkflag, struct vnode **vpp) { struct vnode *vp; + enum vgetstate vs; struct tmpfs_mount *tm; vm_object_t object; int error; @@ -600,18 +601,15 @@ loop: TMPFS_NODE_ASSERT_LOCKED(node); if ((vp = node->tn_vnode) != NULL) { MPASS((node->tn_vpstate & TMPFS_VNODE_DOOMED) == 0); - VI_LOCK(vp); if ((node->tn_type == VDIR && node->tn_dir.tn_parent == NULL) || (VN_IS_DOOMED(vp) && (lkflag & LK_NOWAIT) != 0)) { - VI_UNLOCK(vp); TMPFS_NODE_UNLOCK(node); error = ENOENT; vp = NULL; goto out; } if (VN_IS_DOOMED(vp)) { - VI_UNLOCK(vp); node->tn_vpstate |= TMPFS_VNODE_WRECLAIM; while ((node->tn_vpstate & TMPFS_VNODE_WRECLAIM) != 0) { msleep(&node->tn_vnode, TMPFS_NODE_MTX(node), @@ -619,8 +617,9 @@ loop: } goto loop; } + vs = vget_prep(vp); TMPFS_NODE_UNLOCK(node); - error = vget(vp, lkflag | LK_INTERLOCK); + error = vget_finish(vp, lkflag, vs); if (error == ENOENT) { TMPFS_NODE_LOCK(node); goto loop; From 494c0f2a83f8bcf397435b2a664998bf7df140f9 Mon Sep 17 00:00:00 2001 From: Mateusz Guzik Date: Sun, 16 Aug 2020 17:55:20 +0000 Subject: [PATCH 07/10] vfs: mark HASBUF as an internal flag There is no setter for cn_pnbuf. --- sys/kern/vfs_lookup.c | 3 +-- sys/sys/namei.h | 29 +++++++++++++++-------------- 2 files changed, 16 insertions(+), 16 deletions(-) diff --git a/sys/kern/vfs_lookup.c b/sys/kern/vfs_lookup.c index 43b36dccb02..83a7664a2bc 100644 --- a/sys/kern/vfs_lookup.c +++ b/sys/kern/vfs_lookup.c @@ -504,8 +504,7 @@ namei(struct nameidata *ndp) * Get a buffer for the name to be translated, and copy the * name into the buffer. */ - if ((cnp->cn_flags & HASBUF) == 0) - cnp->cn_pnbuf = uma_zalloc(namei_zone, M_WAITOK); + cnp->cn_pnbuf = uma_zalloc(namei_zone, M_WAITOK); if (ndp->ni_segflg == UIO_SYSSPACE) error = copystr(ndp->ni_dirp, cnp->cn_pnbuf, MAXPATHLEN, &ndp->ni_pathlen); diff --git a/sys/sys/namei.h b/sys/sys/namei.h index 42d39e7330a..9e32853146f 100644 --- a/sys/sys/namei.h +++ b/sys/sys/namei.h @@ -149,21 +149,21 @@ int cache_fplookup(struct nameidata *ndp, enum cache_fpl_status *status, * buffer and for vrele'ing ni_startdir. */ #define RDONLY 0x00000200 /* lookup with read-only semantics */ -#define HASBUF 0x00000400 /* has allocated pathname buffer */ -#define SAVENAME 0x00000800 /* save pathname buffer */ -#define SAVESTART 0x00001000 /* save starting directory */ -#define ISWHITEOUT 0x00002000 /* found whiteout */ -#define DOWHITEOUT 0x00004000 /* do whiteouts */ -#define WILLBEDIR 0x00008000 /* new files will be dirs; allow trailing / */ -#define ISOPEN 0x00010000 /* caller is opening; return a real vnode. */ -#define NOCROSSMOUNT 0x00020000 /* do not cross mount points */ -#define NOMACCHECK 0x00040000 /* do not perform MAC checks */ -#define AUDITVNODE1 0x00080000 /* audit the looked up vnode information */ -#define AUDITVNODE2 0x00100000 /* audit the looked up vnode information */ -#define NOCAPCHECK 0x00200000 /* do not perform capability checks */ +#define SAVENAME 0x00000400 /* save pathname buffer */ +#define SAVESTART 0x00000800 /* save starting directory */ +#define ISWHITEOUT 0x00001000 /* found whiteout */ +#define DOWHITEOUT 0x00002000 /* do whiteouts */ +#define WILLBEDIR 0x00004000 /* new files will be dirs; allow trailing / */ +#define ISOPEN 0x00008000 /* caller is opening; return a real vnode. */ +#define NOCROSSMOUNT 0x00010000 /* do not cross mount points */ +#define NOMACCHECK 0x00020000 /* do not perform MAC checks */ +#define AUDITVNODE1 0x00040000 /* audit the looked up vnode information */ +#define AUDITVNODE2 0x00080000 /* audit the looked up vnode information */ +#define NOCAPCHECK 0x00100000 /* do not perform capability checks */ /* UNUSED 0x00400000 */ +/* UNUSED 0x00200000 */ /* UNUSED 0x00800000 */ -/* UNUSED 0x01000000 */ +#define HASBUF 0x01000000 /* has allocated pathname buffer */ #define NOEXECCHECK 0x02000000 /* do not perform exec check on dir */ #define MAKEENTRY 0x04000000 /* entry is to be added to name cache */ #define ISSYMLINK 0x08000000 /* symlink needs interpretation */ @@ -176,7 +176,8 @@ int cache_fplookup(struct nameidata *ndp, enum cache_fpl_status *status, * Flags which must not be passed in by callers. */ #define NAMEI_INTERNAL_FLAGS \ - (NOEXECCHECK | MAKEENTRY | ISSYMLINK | ISLASTCN | ISDOTDOT | TRAILINGSLASH) + (HASBUF | NOEXECCHECK | MAKEENTRY | ISSYMLINK | ISLASTCN | ISDOTDOT | \ + TRAILINGSLASH) /* * Namei results flags From f1780be99c5aefa59d9076defd226c13e6bb13ae Mon Sep 17 00:00:00 2001 From: Mateusz Guzik Date: Sun, 16 Aug 2020 17:56:30 +0000 Subject: [PATCH 08/10] Bump __FreeBSD_version after vget() and HASBUF change --- sys/sys/param.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/sys/sys/param.h b/sys/sys/param.h index 17130031999..76393482710 100644 --- a/sys/sys/param.h +++ b/sys/sys/param.h @@ -60,7 +60,7 @@ * in the range 5 to 9. */ #undef __FreeBSD_version -#define __FreeBSD_version 1300108 /* Master, propagated to newvers */ +#define __FreeBSD_version 1300109 /* Master, propagated to newvers */ /* * __FreeBSD_kernel__ indicates that this system uses the kernel of FreeBSD, From 832dc76b6368fc1fa57b67f65d83c15cc646d2e4 Mon Sep 17 00:00:00 2001 From: Mariusz Zaborski Date: Sun, 16 Aug 2020 18:12:21 +0000 Subject: [PATCH 09/10] libcasper: Introduce cap_net a network service for Casper. Reviewed by: emaste, markj (previous version), bcr (man page) Differential Revision: https://reviews.freebsd.org/D24688 --- lib/libcasper/services/Makefile | 1 + lib/libcasper/services/cap_dns/Makefile | 5 - lib/libcasper/services/cap_dns/cap_dns.3 | 5 +- lib/libcasper/services/cap_net/Makefile | 48 + lib/libcasper/services/cap_net/cap_net.3 | 287 ++++ lib/libcasper/services/cap_net/cap_net.c | 1385 +++++++++++++++++ lib/libcasper/services/cap_net/cap_net.h | 165 ++ lib/libcasper/services/cap_net/tests/Makefile | 16 + .../services/cap_net/tests/net_test.c | 1160 ++++++++++++++ share/mk/src.libnames.mk | 2 + 10 files changed, 3068 insertions(+), 6 deletions(-) create mode 100644 lib/libcasper/services/cap_net/Makefile create mode 100644 lib/libcasper/services/cap_net/cap_net.3 create mode 100644 lib/libcasper/services/cap_net/cap_net.c create mode 100644 lib/libcasper/services/cap_net/cap_net.h create mode 100644 lib/libcasper/services/cap_net/tests/Makefile create mode 100644 lib/libcasper/services/cap_net/tests/net_test.c diff --git a/lib/libcasper/services/Makefile b/lib/libcasper/services/Makefile index fb4b8c34f31..8fcb2e1e5c6 100644 --- a/lib/libcasper/services/Makefile +++ b/lib/libcasper/services/Makefile @@ -5,6 +5,7 @@ SUBDIR= cap_dns SUBDIR+= cap_fileargs SUBDIR+= cap_grp +SUBDIR+= cap_net SUBDIR+= cap_pwd SUBDIR+= cap_sysctl SUBDIR+= cap_syslog diff --git a/lib/libcasper/services/cap_dns/Makefile b/lib/libcasper/services/cap_dns/Makefile index d97e42404ac..b59b329d065 100644 --- a/lib/libcasper/services/cap_dns/Makefile +++ b/lib/libcasper/services/cap_dns/Makefile @@ -27,11 +27,6 @@ SUBDIR.${MK_TESTS}+= tests MAN+= cap_dns.3 MLINKS+=cap_dns.3 libcap_dns.3 -MLINKS+=cap_dns.3 cap_gethostbyname.3 -MLINKS+=cap_dns.3 cap_gethostbyname2.3 -MLINKS+=cap_dns.3 cap_gethostbyaddr.3 -MLINKS+=cap_dns.3 cap_getaddrinfo.3 -MLINKS+=cap_dns.3 cap_getnameinfo.3 MLINKS+=cap_dns.3 cap_dns_type_limit.3 MLINKS+=cap_dns.3 cap_dns_family_limit.3 diff --git a/lib/libcasper/services/cap_dns/cap_dns.3 b/lib/libcasper/services/cap_dns/cap_dns.3 index e36a7a90950..104e37f53d8 100644 --- a/lib/libcasper/services/cap_dns/cap_dns.3 +++ b/lib/libcasper/services/cap_dns/cap_dns.3 @@ -24,7 +24,7 @@ .\" .\" $FreeBSD$ .\" -.Dd May 5, 2020 +.Dd August 15, 2020 .Dt CAP_DNS 3 .Os .Sh NAME @@ -58,6 +58,9 @@ .Fn cap_dns_family_limit "const cap_channel_t *chan" "const int *families" "size_t nfamilies" .Sh DESCRIPTION .Bf -symbolic +This service is obsolete and +.Xr cap_net 3 +should be used instead. The .Fn cap_getaddrinfo , and diff --git a/lib/libcasper/services/cap_net/Makefile b/lib/libcasper/services/cap_net/Makefile new file mode 100644 index 00000000000..70acac1cc9a --- /dev/null +++ b/lib/libcasper/services/cap_net/Makefile @@ -0,0 +1,48 @@ +# $FreeBSD$ + +SHLIBDIR?= /lib/casper + +.include + +PACKAGE=libcasper + +SHLIB_MAJOR= 1 +INCSDIR?= ${INCLUDEDIR}/casper + +.if ${MK_CASPER} != "no" +SHLIB= cap_net + +SRCS= cap_net.c +.endif + +INCS= cap_net.h + +LIBADD= nv + +CFLAGS+=-I${.CURDIR} +CFLAGS+=-DWITH_CASPER + +HAS_TESTS= +SUBDIR.${MK_TESTS}+= tests + +MAN+= cap_net.3 + +MLINKS+=cap_net.3 libcap_net.3 +MLINKS+=cap_net.3 cap_bind.3 +MLINKS+=cap_net.3 cap_connect.3 +MLINKS+=cap_net.3 cap_net_free.3 +MLINKS+=cap_net.3 cap_net_limit.3 +MLINKS+=cap_net.3 cap_net_limit_addr2name.3 +MLINKS+=cap_net.3 cap_net_limit_addr2name_family.3 +MLINKS+=cap_net.3 cap_net_limit_bind.3 +MLINKS+=cap_net.3 cap_net_limit_connect.3 +MLINKS+=cap_net.3 cap_net_limit_init.3 +MLINKS+=cap_net.3 cap_net_limit_name2addr.3 +MLINKS+=cap_net.3 cap_net_limit_name2addr_family.3 +MLINKS+=cap_net.3 cap_getaddrinfo.3 +MLINKS+=cap_net.3 cap_gethostbyaddr.3 +MLINKS+=cap_net.3 cap_gethostbyname.3 +MLINKS+=cap_net.3 cap_gethostbyname2.3 +MLINKS+=cap_net.3 cap_getnameinfo.3 + +.include diff --git a/lib/libcasper/services/cap_net/cap_net.3 b/lib/libcasper/services/cap_net/cap_net.3 new file mode 100644 index 00000000000..cd0b4450fda --- /dev/null +++ b/lib/libcasper/services/cap_net/cap_net.3 @@ -0,0 +1,287 @@ +.\" Copyright (c) 2020 Mariusz Zaborski +.\" +.\" Redistribution and use in source and binary forms, with or without +.\" modification, are permitted provided that the following conditions +.\" are met: +.\" 1. Redistributions of source code must retain the above copyright +.\" notice, this list of conditions and the following disclaimer. +.\" 2. Redistributions in binary form must reproduce the above copyright +.\" notice, this list of conditions and the following disclaimer in the +.\" documentation and/or other materials provided with the distribution. +.\" +.\" THIS SOFTWARE IS PROVIDED BY THE AUTHORS AND CONTRIBUTORS ``AS IS'' AND +.\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE +.\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE +.\" ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHORS OR CONTRIBUTORS BE LIABLE +.\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +.\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS +.\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) +.\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT +.\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY +.\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF +.\" SUCH DAMAGE. +.\" +.\" $FreeBSD$ +.\" +.Dd August 15, 2020 +.Dt CAP_NET 3 +.Os +.Sh NAME +.Nm cap_bind , +.Nm cap_connect , +.Nm cap_getaddrinfo , +.Nm cap_gethostbyaddr , +.Nm cap_gethostbyname , +.Nm cap_gethostbyname2 , +.Nm cap_getnameinfo , +.Nm cap_net_free , +.Nm cap_net_limit , +.Nm cap_net_limit_addr2name , +.Nm cap_net_limit_addr2name_family , +.Nm cap_net_limit_bind , +.Nm cap_net_limit_connect , +.Nm cap_net_limit_init , +.Nm cap_net_limit_name2addr , +.Nm cap_net_limit_name2addr_family , +.Nd "library for networking in capability mode" +.Sh LIBRARY +.Lb libcap_net +.Sh SYNOPSIS +.In sys/nv.h +.In libcasper.h +.In casper/cap_net.h +.Ft int +.Fn cap_bind "cap_channel_t *chan" "int s" "const struct sockaddr *addr" "socklen_t addrlen" +.Ft int +.Fn cap_connect "cap_channel_t *chan" "int s" "const struct sockaddr *name" "socklen_t namelen" +.Ft int +.Fn cap_getaddrinfo "cap_channel_t *chan" "const char *hostname" "const char *servname" "const struct addrinfo *hints" "struct addrinfo **res" +.Ft int +.Fn cap_getnameinfo "cap_channel_t *chan" "const struct sockaddr *sa" "socklen_t salen" "char *host" "size_t hostlen" "char *serv" "size_t servlen" "int flags" +.Ft "struct hostent *" +.Fn cap_gethostbyname "const cap_channel_t *chan" "const char *name" +.Ft "struct hostent *" +.Fn cap_gethostbyname2 "const cap_channel_t *chan" "const char *name" "int af" +.Ft "struct hostent *" +.Fn cap_gethostbyaddr "const cap_channel_t *chan" "const void *addr" "socklen_t len" "int af" +.Ft "cap_net_limit_t *" +.Fn cap_net_limit_init "cap_channel_t *chan" "uint64_t mode" +.Ft int +.Fn cap_net_limit "cap_net_limit_t *limit" +.Ft void +.Fn cap_net_free "cap_net_limit_t *limit" +.Ft "cap_net_limit_t *" +.Fn cap_net_limit_addr2name_family "cap_net_limit_t *limit" "int *family" "size_t size" +.Ft "cap_net_limit_t *" +.Fn cap_net_limit_addr2name "cap_net_limit_t *limit" "const struct sockaddr *sa" "socklen_t salen" +.Ft "cap_net_limit_t *" +.Fn cap_net_limit_name2addr_family "cap_net_limit_t *limit" "int *family" "size_t size" +.Ft "cap_net_limit_t *" +.Fn cap_net_limit_name2addr "cap_net_limit_t *limit" "const char *name" "const char *serv" +.Ft "cap_net_limit_t *" +.Fn cap_net_limit_connect "cap_net_limit_t *limit" "const struct sockaddr *sa" "socklen_t salen" +.Ft "cap_net_limit_t *" +.Fn cap_net_limit_bind "cap_net_limit_t *limit" "const struct sockaddr *sa" "socklen_t salen" +.Sh DESCRIPTION +.Pp +The functions +.Fn cap_bind, +.Fn cap_connect, +.Fn cap_gethostbyname , +.Fn cap_gethostbyname2 , +.Fn cap_gethostbyaddr +and +.Fn cap_getnameinfo +are respectively equivalent to +.Xr bind 2 , +.Xr connect 2 , +.Xr gethostbyname 3 , +.Xr gethostbyname2 3 , +.Xr gethostbyaddr 3 +and +.Xr getnameinfo 3 +except that the connection to the +.Nm system.net +service needs to be provided. +.Sh LIMITS +By default, the cap_net capability provides unrestricted access to the network +namespace. +Applications typically only require access to a small portion of the network +namespace: +.Fn cap_net_limit +interface can be used to restrict access to the network. +.Fn cap_net_limit_init +returns an opaque limit handle used to store a list of capabilities. +The +.Fv mode +restricts the functionality of the service. +Modes are encoded using the following flags: +.Pp +.Bd -literal -offset indent -compact +CAPNET_ADDR2NAME reverse DNS lookups are allowed with + cap_getnameinfo +CAPNET_NAME2ADDR name resolution is allowed with + cap_getaddrinfo +CAPNET_DEPRECATED_ADDR2NAME reverse DNS lookups are allowed with + cap_gethostbyaddr +CAPNET_DEPRECATED_NAME2ADDR name resolution is allowed with + cap_gethostbyname and cap_gethostbyname2 +CAPNET_BIND bind syscall is allowed +CAPNET_CONNECT connect syscall is allowed +CAPNET_CONNECTDNS connect syscall is allowed to the values + returned from privies call to + the cap_getaddrinfo or cap_gethostbyname +.Ed +.Pp +.Fn cap_net_limit_addr2name_family +limits the +.Fn cap_getnameinfo +and +.Fn cap_gethostbyaddr +to do reverse DNS lookups to specific family (AF_INET, AF_INET6, etc.) +.Pp +.Fn cap_net_limit_addr2name +limits the +.Fn cap_getnameinfo +and +.Fn cap_gethostbyaddr +to do reverse DNS lookups only on those specific structures. +.Pp +.Fn cap_net_limit_name2addr_family +limits the +.Fn cap_getaddrinfo , +.Fn cap_gethostbyname +and +.Fn cap_gethostbyname2 +to do the name resolution on specific family (AF_INET, AF_INET6, etc.) +.Pp +.Fn cap_net_limit_addr2name +restricts +.Fn cap_getaddrinfo , +.Fn cap_gethostbyname +and +.Fn cap_gethostbyname2 +to a set of domains. +.Pp +.Fn cap_net_limit_bind +limits +.Fn cap_bind +to bind only on those specific structures. +.Pp +.Fn cap_net_limit_connect +limits +.Fn cap_connect +to connect only on those specific structures. +If the CAPNET_CONNECTDNS is set the limits are extended to the values returned +by +.Fn cap_getaddrinfo , +.Fn cap_gethostbyname +and +.Fn cap_gethostbyname2 +In case of the +.Fn cap_getaddrinfo +the restriction is strict. +In case of the +.Fn cap_gethostbyname +and +.Fn cap_gethostbyname2 +any port will be accepted in the +.Fn cap_connect +function. +.Pp +.Fn cap_net_limit +applies a set of sysctl limits to the capability, denying access to sysctl +variables not belonging to the set. +.Pp +Once a set of limits is applied, subsequent calls to +.Fn cap_net_limit +will fail unless the new set is a subset of the current set. +.Pp +The +.Fn cap_net_limit +will consume the limits. +If the +.Fn cap_net_limit +was not called the rights may be freed using +.Fn cap_net_free . +Multiple calls to +.Fn cap_net_limit_addr2name_family , +.Fn cap_net_limit_addr2name , +.Fn cap_net_limit_name2addr_family , +.Fn cap_net_limit_name2addr , +.Fn cap_net_limit_connect , +and +.Fn cap_net_limit_bind +is supported, each call is extending preview capabilities. +.Sh EXAMPLES +The following example first opens a capability to casper and then uses this +capability to create the +.Nm system.net +casper service and uses it to resolve a host and connect to it. +.Bd -literal +cap_channel_t *capcas, *capnet; +cap_net_limit_t *limit; +int familylimit, error, s; +const char *host = "example.com"; +struct addrinfo hints, *res; + +/* Open capability to Casper. */ +capcas = cap_init(); +if (capcas == NULL) + err(1, "Unable to contact Casper"); + +/* Cache NLA for gai_strerror. */ +caph_cache_catpages(); + +/* Enter capability mode sandbox. */ +if (caph_enter_casper() < 0) + err(1, "Unable to enter capability mode"); + +/* Use Casper capability to create capability to the system.net service. */ +capnet = cap_service_open(capcas, "system.net"); +if (capnet == NULL) + err(1, "Unable to open system.net service"); + +/* Close Casper capability. */ +cap_close(capcas); + +/* Limit system.net to reserve IPv4 addresses, to host example.com . */ +limit = cap_net_limit_init(capnet, CAPNET_NAME2ADDR | CAPNET_CONNECTDNS); +if (limit == NULL) + err(1, "Unable to create limits."); +cap_net_limit_name2addr(limit, host, "80"); +familylimit = AF_INET; +cap_net_limit_name2addr_family(limit, &familylimit, 1); +if (cap_net_limit(limit) < 0) + err(1, "Unable to apply limits."); + +/* Find IP addresses for the given host. */ +memset(&hints, 0, sizeof(hints)); +hints.ai_family = AF_INET; +hints.ai_socktype = SOCK_STREAM; + +error = cap_getaddrinfo(capnet, host, "80", &hints, &res); +if (error != 0) + errx(1, "cap_getaddrinfo(): %s: %s", host, gai_strerror(error)); + +s = socket(res->ai_family, res->ai_socktype, res->ai_protocol); +if (s < 0) + err(1, "Unable to create socket"); + +if (cap_connect(capnet, s, res->ai_addr, res->ai_addrlen) < 0) + err(1, "Unable to connect to host"); +.Ed +.Sh SEE ALSO +.Xr bind 2 , +.Xr cap_enter 2 , +.Xr connect 2 , +.Xr caph_enter 3 , +.Xr err 3 , +.Xr gethostbyaddr 3 , +.Xr gethostbyname 3 , +.Xr gethostbyname2 3 , +.Xr getnameinfo 3 , +.Xr capsicum 4 , +.Xr nv 9 +.Sh AUTHORS +.An Mariusz Zaborski Aq Mt oshogbo@FreeBSD.org diff --git a/lib/libcasper/services/cap_net/cap_net.c b/lib/libcasper/services/cap_net/cap_net.c new file mode 100644 index 00000000000..a2078b3335d --- /dev/null +++ b/lib/libcasper/services/cap_net/cap_net.c @@ -0,0 +1,1385 @@ +/*- + * SPDX-License-Identifier: BSD-2-Clause-FreeBSD + * + * Copyright (c) 2020 Mariusz Zaborski + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHORS AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHORS OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + */ + +#include +__FBSDID("$FreeBSD$"); + +#include +#include +#include +#include +#include + +#include +#include +#include +#include +#include +#include + +#include +#include + +#include "cap_net.h" + +#define CAPNET_MASK (CAPNET_ADDR2NAME | CAPNET_NAME2ADDR \ + CAPNET_DEPRECATED_ADDR2NAME | CAPNET_DEPRECATED_NAME2ADDR | \ + CAPNET_CONNECT | CAPNET_BIND | CAPNET_CONNECTDNS) + +/* + * Defines for the names of the limits. + * XXX: we should convert all string constats to this to avoid typos. + */ +#define LIMIT_NV_BIND "bind" +#define LIMIT_NV_CONNECT "connect" +#define LIMIT_NV_ADDR2NAME "addr2name" +#define LIMIT_NV_NAME2ADDR "name2addr" + +struct cap_net_limit { + cap_channel_t *cnl_chan; + uint64_t cnl_mode; + nvlist_t *cnl_addr2name; + nvlist_t *cnl_name2addr; + nvlist_t *cnl_connect; + nvlist_t *cnl_bind; +}; + +static struct hostent hent; + +static void +hostent_free(struct hostent *hp) +{ + unsigned int ii; + + free(hp->h_name); + hp->h_name = NULL; + if (hp->h_aliases != NULL) { + for (ii = 0; hp->h_aliases[ii] != NULL; ii++) + free(hp->h_aliases[ii]); + free(hp->h_aliases); + hp->h_aliases = NULL; + } + if (hp->h_addr_list != NULL) { + for (ii = 0; hp->h_addr_list[ii] != NULL; ii++) + free(hp->h_addr_list[ii]); + free(hp->h_addr_list); + hp->h_addr_list = NULL; + } +} + +static struct hostent * +hostent_unpack(const nvlist_t *nvl, struct hostent *hp) +{ + unsigned int ii, nitems; + char nvlname[64]; + int n; + + hostent_free(hp); + + hp->h_name = strdup(nvlist_get_string(nvl, "name")); + if (hp->h_name == NULL) + goto fail; + hp->h_addrtype = (int)nvlist_get_number(nvl, "addrtype"); + hp->h_length = (int)nvlist_get_number(nvl, "length"); + + nitems = (unsigned int)nvlist_get_number(nvl, "naliases"); + hp->h_aliases = calloc(sizeof(hp->h_aliases[0]), nitems + 1); + if (hp->h_aliases == NULL) + goto fail; + for (ii = 0; ii < nitems; ii++) { + n = snprintf(nvlname, sizeof(nvlname), "alias%u", ii); + assert(n > 0 && n < (int)sizeof(nvlname)); + hp->h_aliases[ii] = + strdup(nvlist_get_string(nvl, nvlname)); + if (hp->h_aliases[ii] == NULL) + goto fail; + } + hp->h_aliases[ii] = NULL; + + nitems = (unsigned int)nvlist_get_number(nvl, "naddrs"); + hp->h_addr_list = calloc(sizeof(hp->h_addr_list[0]), nitems + 1); + if (hp->h_addr_list == NULL) + goto fail; + for (ii = 0; ii < nitems; ii++) { + hp->h_addr_list[ii] = malloc(hp->h_length); + if (hp->h_addr_list[ii] == NULL) + goto fail; + n = snprintf(nvlname, sizeof(nvlname), "addr%u", ii); + assert(n > 0 && n < (int)sizeof(nvlname)); + bcopy(nvlist_get_binary(nvl, nvlname, NULL), + hp->h_addr_list[ii], hp->h_length); + } + hp->h_addr_list[ii] = NULL; + + return (hp); +fail: + hostent_free(hp); + h_errno = NO_RECOVERY; + return (NULL); +} + +static int +request_cb(cap_channel_t *chan, const char *name, int s, + const struct sockaddr *saddr, socklen_t len) +{ + nvlist_t *nvl; + int serrno; + + nvl = nvlist_create(0); + nvlist_add_string(nvl, "cmd", name); + nvlist_add_descriptor(nvl, "s", s); + nvlist_add_binary(nvl, "saddr", saddr, len); + + nvl = cap_xfer_nvlist(chan, nvl); + if (nvl == NULL) + return (-1); + + if (nvlist_get_number(nvl, "error") != 0) { + serrno = (int)nvlist_get_number(nvl, "error"); + nvlist_destroy(nvl); + errno = serrno; + return (-1); + } + + s = dup2(s, nvlist_get_descriptor(nvl, "s")); + nvlist_destroy(nvl); + + return (s == -1 ? -1 : 0); +} + +int +cap_bind(cap_channel_t *chan, int s, const struct sockaddr *addr, + socklen_t addrlen) +{ + + return (request_cb(chan, LIMIT_NV_BIND, s, addr, addrlen)); +} + +int +cap_connect(cap_channel_t *chan, int s, const struct sockaddr *name, + socklen_t namelen) +{ + + return (request_cb(chan, LIMIT_NV_CONNECT, s, name, namelen)); +} + + +struct hostent * +cap_gethostbyname(cap_channel_t *chan, const char *name) +{ + + return (cap_gethostbyname2(chan, name, AF_INET)); +} + +struct hostent * +cap_gethostbyname2(cap_channel_t *chan, const char *name, int af) +{ + struct hostent *hp; + nvlist_t *nvl; + + nvl = nvlist_create(0); + nvlist_add_string(nvl, "cmd", "gethostbyname"); + nvlist_add_number(nvl, "family", (uint64_t)af); + nvlist_add_string(nvl, "name", name); + nvl = cap_xfer_nvlist(chan, nvl); + if (nvl == NULL) { + h_errno = NO_RECOVERY; + return (NULL); + } + if (nvlist_get_number(nvl, "error") != 0) { + h_errno = (int)nvlist_get_number(nvl, "error"); + nvlist_destroy(nvl); + return (NULL); + } + + hp = hostent_unpack(nvl, &hent); + nvlist_destroy(nvl); + return (hp); +} + +struct hostent * +cap_gethostbyaddr(cap_channel_t *chan, const void *addr, socklen_t len, + int af) +{ + struct hostent *hp; + nvlist_t *nvl; + + nvl = nvlist_create(0); + nvlist_add_string(nvl, "cmd", "gethostbyaddr"); + nvlist_add_binary(nvl, "addr", addr, (size_t)len); + nvlist_add_number(nvl, "family", (uint64_t)af); + nvl = cap_xfer_nvlist(chan, nvl); + if (nvl == NULL) { + h_errno = NO_RECOVERY; + return (NULL); + } + if (nvlist_get_number(nvl, "error") != 0) { + h_errno = (int)nvlist_get_number(nvl, "error"); + nvlist_destroy(nvl); + return (NULL); + } + hp = hostent_unpack(nvl, &hent); + nvlist_destroy(nvl); + return (hp); +} + +static struct addrinfo * +addrinfo_unpack(const nvlist_t *nvl) +{ + struct addrinfo *ai; + const void *addr; + size_t addrlen; + const char *canonname; + + addr = nvlist_get_binary(nvl, "ai_addr", &addrlen); + ai = malloc(sizeof(*ai) + addrlen); + if (ai == NULL) + return (NULL); + ai->ai_flags = (int)nvlist_get_number(nvl, "ai_flags"); + ai->ai_family = (int)nvlist_get_number(nvl, "ai_family"); + ai->ai_socktype = (int)nvlist_get_number(nvl, "ai_socktype"); + ai->ai_protocol = (int)nvlist_get_number(nvl, "ai_protocol"); + ai->ai_addrlen = (socklen_t)addrlen; + canonname = dnvlist_get_string(nvl, "ai_canonname", NULL); + if (canonname != NULL) { + ai->ai_canonname = strdup(canonname); + if (ai->ai_canonname == NULL) { + free(ai); + return (NULL); + } + } else { + ai->ai_canonname = NULL; + } + ai->ai_addr = (void *)(ai + 1); + bcopy(addr, ai->ai_addr, addrlen); + ai->ai_next = NULL; + + return (ai); +} + +int +cap_getaddrinfo(cap_channel_t *chan, const char *hostname, const char *servname, + const struct addrinfo *hints, struct addrinfo **res) +{ + struct addrinfo *firstai, *prevai, *curai; + unsigned int ii; + const nvlist_t *nvlai; + char nvlname[64]; + nvlist_t *nvl; + int error, n; + + nvl = nvlist_create(0); + nvlist_add_string(nvl, "cmd", "getaddrinfo"); + if (hostname != NULL) + nvlist_add_string(nvl, "hostname", hostname); + if (servname != NULL) + nvlist_add_string(nvl, "servname", servname); + if (hints != NULL) { + nvlist_add_number(nvl, "hints.ai_flags", + (uint64_t)hints->ai_flags); + nvlist_add_number(nvl, "hints.ai_family", + (uint64_t)hints->ai_family); + nvlist_add_number(nvl, "hints.ai_socktype", + (uint64_t)hints->ai_socktype); + nvlist_add_number(nvl, "hints.ai_protocol", + (uint64_t)hints->ai_protocol); + } + nvl = cap_xfer_nvlist(chan, nvl); + if (nvl == NULL) + return (EAI_MEMORY); + if (nvlist_get_number(nvl, "error") != 0) { + error = (int)nvlist_get_number(nvl, "error"); + nvlist_destroy(nvl); + return (error); + } + + nvlai = NULL; + firstai = prevai = curai = NULL; + for (ii = 0; ; ii++) { + n = snprintf(nvlname, sizeof(nvlname), "res%u", ii); + assert(n > 0 && n < (int)sizeof(nvlname)); + if (!nvlist_exists_nvlist(nvl, nvlname)) + break; + nvlai = nvlist_get_nvlist(nvl, nvlname); + curai = addrinfo_unpack(nvlai); + if (curai == NULL) + return (EAI_MEMORY); + if (prevai != NULL) + prevai->ai_next = curai; + else + firstai = curai; + prevai = curai; + } + nvlist_destroy(nvl); + if (curai == NULL && nvlai != NULL) { + if (firstai == NULL) + freeaddrinfo(firstai); + return (EAI_MEMORY); + } + + *res = firstai; + return (0); +} + +int +cap_getnameinfo(cap_channel_t *chan, const struct sockaddr *sa, socklen_t salen, + char *host, size_t hostlen, char *serv, size_t servlen, int flags) +{ + nvlist_t *nvl; + int error; + + nvl = nvlist_create(0); + nvlist_add_string(nvl, "cmd", "getnameinfo"); + nvlist_add_number(nvl, "hostlen", (uint64_t)hostlen); + nvlist_add_number(nvl, "servlen", (uint64_t)servlen); + nvlist_add_binary(nvl, "sa", sa, (size_t)salen); + nvlist_add_number(nvl, "flags", (uint64_t)flags); + nvl = cap_xfer_nvlist(chan, nvl); + if (nvl == NULL) + return (EAI_MEMORY); + if (nvlist_get_number(nvl, "error") != 0) { + error = (int)nvlist_get_number(nvl, "error"); + nvlist_destroy(nvl); + return (error); + } + + if (host != NULL && nvlist_exists_string(nvl, "host")) + strlcpy(host, nvlist_get_string(nvl, "host"), hostlen + 1); + if (serv != NULL && nvlist_exists_string(nvl, "serv")) + strlcpy(serv, nvlist_get_string(nvl, "serv"), servlen + 1); + nvlist_destroy(nvl); + return (0); +} + +cap_net_limit_t * +cap_net_limit_init(cap_channel_t *chan, uint64_t mode) +{ + cap_net_limit_t *limit; + + limit = calloc(1, sizeof(*limit)); + if (limit != NULL) { + limit->cnl_mode = mode; + limit->cnl_chan = chan; + limit->cnl_addr2name = nvlist_create(0); + limit->cnl_name2addr = nvlist_create(0); + limit->cnl_connect = nvlist_create(0); + limit->cnl_bind = nvlist_create(0); + } + + return (limit); +} + +static void +pack_limit(nvlist_t *lnvl, const char *name, nvlist_t *limit) +{ + + if (!nvlist_empty(limit)) { + nvlist_move_nvlist(lnvl, name, limit); + } else { + nvlist_destroy(limit); + } +} + +int +cap_net_limit(cap_net_limit_t *limit) +{ + nvlist_t *lnvl; + cap_channel_t *chan; + + lnvl = nvlist_create(0); + nvlist_add_number(lnvl, "mode", limit->cnl_mode); + + pack_limit(lnvl, LIMIT_NV_ADDR2NAME, limit->cnl_addr2name); + pack_limit(lnvl, LIMIT_NV_NAME2ADDR, limit->cnl_name2addr); + pack_limit(lnvl, LIMIT_NV_CONNECT, limit->cnl_connect); + pack_limit(lnvl, LIMIT_NV_BIND, limit->cnl_bind); + + chan = limit->cnl_chan; + free(limit); + + return (cap_limit_set(chan, lnvl)); +} + +void +cap_net_free(cap_net_limit_t *limit) +{ + + if (limit == NULL) + return; + + nvlist_destroy(limit->cnl_addr2name); + nvlist_destroy(limit->cnl_name2addr); + nvlist_destroy(limit->cnl_connect); + nvlist_destroy(limit->cnl_bind); + + free(limit); +} + +static void +pack_family(nvlist_t *nvl, int *family, size_t size) +{ + size_t i; + + i = 0; + if (!nvlist_exists_number_array(nvl, "family")) { + uint64_t val; + + val = family[0]; + nvlist_add_number_array(nvl, "family", &val, 1); + i += 1; + } + + for (; i < size; i++) { + nvlist_append_number_array(nvl, "family", family[i]); + } +} + +static void +pack_sockaddr(nvlist_t *res, const struct sockaddr *sa, socklen_t salen) +{ + nvlist_t *nvl; + + if (!nvlist_exists_nvlist(res, "sockaddr")) { + nvl = nvlist_create(NV_FLAG_NO_UNIQUE); + } else { + nvl = nvlist_take_nvlist(res, "sockaddr"); + } + + nvlist_add_binary(nvl, "", sa, salen); + nvlist_move_nvlist(res, "sockaddr", nvl); +} + +cap_net_limit_t * +cap_net_limit_addr2name_family(cap_net_limit_t *limit, int *family, size_t size) +{ + + pack_family(limit->cnl_addr2name, family, size); + return (limit); +} + +cap_net_limit_t * +cap_net_limit_name2addr_family(cap_net_limit_t *limit, int *family, size_t size) +{ + + pack_family(limit->cnl_name2addr, family, size); + return (limit); +} + +cap_net_limit_t * +cap_net_limit_name2addr(cap_net_limit_t *limit, const char *host, + const char *serv) +{ + nvlist_t *nvl; + + if (!nvlist_exists_nvlist(limit->cnl_name2addr, "hosts")) { + nvl = nvlist_create(NV_FLAG_NO_UNIQUE); + } else { + nvl = nvlist_take_nvlist(limit->cnl_name2addr, "hosts"); + } + + nvlist_add_string(nvl, + host != NULL ? host : "", + serv != NULL ? serv : ""); + + nvlist_move_nvlist(limit->cnl_name2addr, "hosts", nvl); + return (limit); +} + +cap_net_limit_t * +cap_net_limit_addr2name(cap_net_limit_t *limit, const struct sockaddr *sa, + socklen_t salen) +{ + + pack_sockaddr(limit->cnl_addr2name, sa, salen); + return (limit); +} + + +cap_net_limit_t * +cap_net_limit_connect(cap_net_limit_t *limit, const struct sockaddr *sa, + socklen_t salen) +{ + + pack_sockaddr(limit->cnl_connect, sa, salen); + return (limit); +} + +cap_net_limit_t * +cap_net_limit_bind(cap_net_limit_t *limit, const struct sockaddr *sa, + socklen_t salen) +{ + + pack_sockaddr(limit->cnl_bind, sa, salen); + return (limit); +} + +/* + * Service functions. + */ + +static nvlist_t *capdnscache; + +static void +net_add_sockaddr_to_cache(struct sockaddr *sa, socklen_t salen, bool deprecated) +{ + void *cookie; + + if (capdnscache == NULL) { + capdnscache = nvlist_create(NV_FLAG_NO_UNIQUE); + } else { + /* Lets keep it clean. Look for dups. */ + cookie = NULL; + while (nvlist_next(capdnscache, NULL, &cookie) != NULL) { + const void *data; + size_t size; + + assert(cnvlist_type(cookie) == NV_TYPE_BINARY); + + data = cnvlist_get_binary(cookie, &size); + if (salen != size) + continue; + if (memcmp(data, sa, size) == 0) + return; + } + } + + nvlist_add_binary(capdnscache, deprecated ? "d" : "", sa, salen); +} + +static void +net_add_hostent_to_cache(const char *address, size_t asize, int family) +{ + + if (family != AF_INET && family != AF_INET6) + return; + + if (family == AF_INET6) { + struct sockaddr_in6 connaddr; + + memset(&connaddr, 0, sizeof(connaddr)); + connaddr.sin6_family = AF_INET6; + memcpy((char *)&connaddr.sin6_addr, address, asize); + connaddr.sin6_port = 0; + + net_add_sockaddr_to_cache((struct sockaddr *)&connaddr, + sizeof(connaddr), true); + } else { + struct sockaddr_in connaddr; + + memset(&connaddr, 0, sizeof(connaddr)); + connaddr.sin_family = AF_INET; + memcpy((char *)&connaddr.sin_addr.s_addr, address, asize); + connaddr.sin_port = 0; + + net_add_sockaddr_to_cache((struct sockaddr *)&connaddr, + sizeof(connaddr), true); + } +} + +static bool +net_allowed_mode(const nvlist_t *limits, uint64_t mode) +{ + + if (limits == NULL) + return (true); + + return ((nvlist_get_number(limits, "mode") & mode) == mode); +} + +static bool +net_allowed_family(const nvlist_t *limits, int family) +{ + const uint64_t *allowedfamily; + size_t i, allsize; + + if (limits == NULL) + return (true); + + /* If there are no familes at all, allow any mode. */ + if (!nvlist_exists_number_array(limits, "family")) + return (true); + + allowedfamily = nvlist_get_number_array(limits, "family", &allsize); + for (i = 0; i < allsize; i++) { + /* XXX: what with AF_UNSPEC? */ + if (allowedfamily[i] == (uint64_t)family) { + return (true); + } + } + + return (false); +} + +static bool +net_allowed_bsaddr_impl(const nvlist_t *salimits, const void *saddr, + size_t saddrsize) +{ + void *cookie; + const void *limit; + size_t limitsize; + + cookie = NULL; + while (nvlist_next(salimits, NULL, &cookie) != NULL) { + limit = cnvlist_get_binary(cookie, &limitsize); + + if (limitsize != saddrsize) { + continue; + } + if (memcmp(limit, saddr, limitsize) == 0) { + return (true); + } + + /* + * In case of deprecated version (gethostbyname) we have to + * ignore port, because there is no such info in the hostent. + * Suporting only AF_INET and AF_INET6. + */ + if (strcmp(cnvlist_name(cookie), "d") != 0 || + (saddrsize != sizeof(struct sockaddr_in) && + saddrsize != sizeof(struct sockaddr_in6))) { + continue; + } + if (saddrsize == sizeof(struct sockaddr_in)) { + const struct sockaddr_in *saddrptr; + struct sockaddr_in sockaddr; + + saddrptr = (const struct sockaddr_in *)saddr; + memcpy(&sockaddr, limit, sizeof(sockaddr)); + sockaddr.sin_port = saddrptr->sin_port; + + if (memcmp(&sockaddr, saddr, saddrsize) == 0) { + return (true); + } + } else if (saddrsize == sizeof(struct sockaddr_in6)) { + const struct sockaddr_in6 *saddrptr; + struct sockaddr_in6 sockaddr; + + saddrptr = (const struct sockaddr_in6 *)saddr; + memcpy(&sockaddr, limit, sizeof(sockaddr)); + sockaddr.sin6_port = saddrptr->sin6_port; + + if (memcmp(&sockaddr, saddr, saddrsize) == 0) { + return (true); + } + } + } + + return (false); +} + +static bool +net_allowed_bsaddr(const nvlist_t *limits, const void *saddr, size_t saddrsize) +{ + + if (limits == NULL) + return (true); + + if (!nvlist_exists_nvlist(limits, "sockaddr")) + return (true); + + return (net_allowed_bsaddr_impl(nvlist_get_nvlist(limits, "sockaddr"), + saddr, saddrsize)); +} + +static bool +net_allowed_hosts(const nvlist_t *limits, const char *name, const char *srvname) +{ + void *cookie; + const nvlist_t *hlimits; + const char *testname, *testsrvname; + + if (limits == NULL) { + return (true); + } + + /* If there are no hosts at all, allow any. */ + if (!nvlist_exists_nvlist(limits, "hosts")) { + return (true); + } + + cookie = NULL; + testname = (name == NULL ? "" : name); + testsrvname = (srvname == NULL ? "" : srvname); + hlimits = nvlist_get_nvlist(limits, "hosts"); + while (nvlist_next(hlimits, NULL, &cookie) != NULL) { + if (strcmp(cnvlist_name(cookie), "") != 0 && + strcmp(cnvlist_name(cookie), testname) != 0) { + continue; + } + + if (strcmp(cnvlist_get_string(cookie), "") != 0 && + strcmp(cnvlist_get_string(cookie), testsrvname) != 0) { + continue; + } + + return (true); + } + + return (false); +} + +static void +hostent_pack(const struct hostent *hp, nvlist_t *nvl, bool addtocache) +{ + unsigned int ii; + char nvlname[64]; + int n; + + nvlist_add_string(nvl, "name", hp->h_name); + nvlist_add_number(nvl, "addrtype", (uint64_t)hp->h_addrtype); + nvlist_add_number(nvl, "length", (uint64_t)hp->h_length); + + if (hp->h_aliases == NULL) { + nvlist_add_number(nvl, "naliases", 0); + } else { + for (ii = 0; hp->h_aliases[ii] != NULL; ii++) { + n = snprintf(nvlname, sizeof(nvlname), "alias%u", ii); + assert(n > 0 && n < (int)sizeof(nvlname)); + nvlist_add_string(nvl, nvlname, hp->h_aliases[ii]); + } + nvlist_add_number(nvl, "naliases", (uint64_t)ii); + } + + if (hp->h_addr_list == NULL) { + nvlist_add_number(nvl, "naddrs", 0); + } else { + for (ii = 0; hp->h_addr_list[ii] != NULL; ii++) { + n = snprintf(nvlname, sizeof(nvlname), "addr%u", ii); + assert(n > 0 && n < (int)sizeof(nvlname)); + nvlist_add_binary(nvl, nvlname, hp->h_addr_list[ii], + (size_t)hp->h_length); + if (addtocache) { + net_add_hostent_to_cache(hp->h_addr_list[ii], + hp->h_length, hp->h_addrtype); + } + } + nvlist_add_number(nvl, "naddrs", (uint64_t)ii); + } +} + +static int +net_gethostbyname(const nvlist_t *limits, const nvlist_t *nvlin, + nvlist_t *nvlout) +{ + struct hostent *hp; + int family; + const nvlist_t *funclimit; + const char *name; + bool dnscache; + + if (!net_allowed_mode(limits, CAPNET_DEPRECATED_NAME2ADDR)) + return (ENOTCAPABLE); + + dnscache = net_allowed_mode(limits, CAPNET_CONNECTDNS); + funclimit = dnvlist_get_nvlist(limits, LIMIT_NV_NAME2ADDR, NULL); + + family = (int)nvlist_get_number(nvlin, "family"); + if (!net_allowed_family(funclimit, family)) + return (ENOTCAPABLE); + + name = nvlist_get_string(nvlin, "name"); + if (!net_allowed_hosts(funclimit, name, "")) + return (ENOTCAPABLE); + + hp = gethostbyname2(name, family); + if (hp == NULL) + return (h_errno); + hostent_pack(hp, nvlout, dnscache); + return (0); +} + +static int +net_gethostbyaddr(const nvlist_t *limits, const nvlist_t *nvlin, + nvlist_t *nvlout) +{ + struct hostent *hp; + const void *addr; + size_t addrsize; + int family; + const nvlist_t *funclimit; + + if (!net_allowed_mode(limits, CAPNET_DEPRECATED_ADDR2NAME)) + return (ENOTCAPABLE); + + funclimit = dnvlist_get_nvlist(limits, LIMIT_NV_ADDR2NAME, NULL); + + family = (int)nvlist_get_number(nvlin, "family"); + if (!net_allowed_family(funclimit, family)) + return (ENOTCAPABLE); + + addr = nvlist_get_binary(nvlin, "addr", &addrsize); + if (!net_allowed_bsaddr(funclimit, addr, addrsize)) + return (ENOTCAPABLE); + + hp = gethostbyaddr(addr, (socklen_t)addrsize, family); + if (hp == NULL) + return (h_errno); + hostent_pack(hp, nvlout, false); + return (0); +} + +static int +net_getnameinfo(const nvlist_t *limits, const nvlist_t *nvlin, nvlist_t *nvlout) +{ + struct sockaddr_storage sast; + const void *sabin; + char *host, *serv; + size_t sabinsize, hostlen, servlen; + socklen_t salen; + int error, flags; + const nvlist_t *funclimit; + + if (!net_allowed_mode(limits, CAPNET_ADDR2NAME)) + return (ENOTCAPABLE); + funclimit = dnvlist_get_nvlist(limits, LIMIT_NV_ADDR2NAME, NULL); + + error = 0; + host = serv = NULL; + memset(&sast, 0, sizeof(sast)); + + hostlen = (size_t)nvlist_get_number(nvlin, "hostlen"); + servlen = (size_t)nvlist_get_number(nvlin, "servlen"); + + if (hostlen > 0) { + host = calloc(1, hostlen + 1); + if (host == NULL) { + error = EAI_MEMORY; + goto out; + } + } + if (servlen > 0) { + serv = calloc(1, servlen + 1); + if (serv == NULL) { + error = EAI_MEMORY; + goto out; + } + } + + sabin = nvlist_get_binary(nvlin, "sa", &sabinsize); + if (sabinsize > sizeof(sast)) { + error = EAI_FAIL; + goto out; + } + if (!net_allowed_bsaddr(funclimit, sabin, sabinsize)) + return (ENOTCAPABLE); + + memcpy(&sast, sabin, sabinsize); + salen = (socklen_t)sabinsize; + + if ((sast.ss_family != AF_INET || + salen != sizeof(struct sockaddr_in)) && + (sast.ss_family != AF_INET6 || + salen != sizeof(struct sockaddr_in6))) { + error = EAI_FAIL; + goto out; + } + + if (!net_allowed_family(funclimit, (int)sast.ss_family)) { + error = ENOTCAPABLE; + goto out; + } + + flags = (int)nvlist_get_number(nvlin, "flags"); + + error = getnameinfo((struct sockaddr *)&sast, salen, host, hostlen, + serv, servlen, flags); + if (error != 0) + goto out; + + if (host != NULL) + nvlist_move_string(nvlout, "host", host); + if (serv != NULL) + nvlist_move_string(nvlout, "serv", serv); +out: + if (error != 0) { + free(host); + free(serv); + } + return (error); +} + +static nvlist_t * +addrinfo_pack(const struct addrinfo *ai) +{ + nvlist_t *nvl; + + nvl = nvlist_create(0); + nvlist_add_number(nvl, "ai_flags", (uint64_t)ai->ai_flags); + nvlist_add_number(nvl, "ai_family", (uint64_t)ai->ai_family); + nvlist_add_number(nvl, "ai_socktype", (uint64_t)ai->ai_socktype); + nvlist_add_number(nvl, "ai_protocol", (uint64_t)ai->ai_protocol); + nvlist_add_binary(nvl, "ai_addr", ai->ai_addr, (size_t)ai->ai_addrlen); + if (ai->ai_canonname != NULL) + nvlist_add_string(nvl, "ai_canonname", ai->ai_canonname); + + return (nvl); +} + +static int +net_getaddrinfo(const nvlist_t *limits, const nvlist_t *nvlin, nvlist_t *nvlout) +{ + struct addrinfo hints, *hintsp, *res, *cur; + const char *hostname, *servname; + char nvlname[64]; + nvlist_t *elem; + unsigned int ii; + int error, family, n; + const nvlist_t *funclimit; + bool dnscache; + + if (!net_allowed_mode(limits, CAPNET_NAME2ADDR)) + return (ENOTCAPABLE); + dnscache = net_allowed_mode(limits, CAPNET_CONNECTDNS); + funclimit = dnvlist_get_nvlist(limits, LIMIT_NV_NAME2ADDR, NULL); + + hostname = dnvlist_get_string(nvlin, "hostname", NULL); + servname = dnvlist_get_string(nvlin, "servname", NULL); + if (nvlist_exists_number(nvlin, "hints.ai_flags")) { + hints.ai_flags = (int)nvlist_get_number(nvlin, + "hints.ai_flags"); + hints.ai_family = (int)nvlist_get_number(nvlin, + "hints.ai_family"); + hints.ai_socktype = (int)nvlist_get_number(nvlin, + "hints.ai_socktype"); + hints.ai_protocol = (int)nvlist_get_number(nvlin, + "hints.ai_protocol"); + hints.ai_addrlen = 0; + hints.ai_addr = NULL; + hints.ai_canonname = NULL; + hints.ai_next = NULL; + hintsp = &hints; + family = hints.ai_family; + } else { + hintsp = NULL; + family = AF_UNSPEC; + } + + if (!net_allowed_family(funclimit, family)) + return (ENOTCAPABLE); + if (!net_allowed_hosts(funclimit, hostname, servname)) + return (ENOTCAPABLE); + error = getaddrinfo(hostname, servname, hintsp, &res); + if (error != 0) { + goto out; + } + + for (cur = res, ii = 0; cur != NULL; cur = cur->ai_next, ii++) { + elem = addrinfo_pack(cur); + n = snprintf(nvlname, sizeof(nvlname), "res%u", ii); + assert(n > 0 && n < (int)sizeof(nvlname)); + nvlist_move_nvlist(nvlout, nvlname, elem); + if (dnscache) { + net_add_sockaddr_to_cache(cur->ai_addr, + cur->ai_addrlen, false); + } + } + + freeaddrinfo(res); + error = 0; +out: + return (error); +} + +static int +net_bind(const nvlist_t *limits, nvlist_t *nvlin, nvlist_t *nvlout) +{ + int socket, serrno; + const void *saddr; + size_t len; + const nvlist_t *funclimit; + + if (!net_allowed_mode(limits, CAPNET_BIND)) + return (ENOTCAPABLE); + funclimit = dnvlist_get_nvlist(limits, LIMIT_NV_BIND, NULL); + + saddr = nvlist_get_binary(nvlin, "saddr", &len); + + if (!net_allowed_bsaddr(funclimit, saddr, len)) + return (ENOTCAPABLE); + + socket = nvlist_take_descriptor(nvlin, "s"); + if (bind(socket, saddr, len) < 0) { + serrno = errno; + close(socket); + return (serrno); + } + + nvlist_move_descriptor(nvlout, "s", socket); + + return (0); +} + +static int +net_connect(const nvlist_t *limits, nvlist_t *nvlin, nvlist_t *nvlout) +{ + int socket, serrno; + const void *saddr; + const nvlist_t *funclimit; + size_t len; + bool conn, conndns; + + conn = net_allowed_mode(limits, CAPNET_CONNECT); + conndns = net_allowed_mode(limits, CAPNET_CONNECTDNS); + + if (!conn && !conndns) + return (ENOTCAPABLE); + + funclimit = dnvlist_get_nvlist(limits, LIMIT_NV_CONNECT, NULL); + + saddr = nvlist_get_binary(nvlin, "saddr", &len); + if (conn && !net_allowed_bsaddr(funclimit, saddr, len)) { + return (ENOTCAPABLE); + } else if (conndns && (capdnscache == NULL || + !net_allowed_bsaddr_impl(capdnscache, saddr, len))) { + return (ENOTCAPABLE); + } + socket = dup(nvlist_get_descriptor(nvlin, "s")); + if (connect(socket, saddr, len) < 0) { + serrno = errno; + close(socket); + return (serrno); + } + + nvlist_move_descriptor(nvlout, "s", socket); + + return (0); +} + +static bool +verify_only_sa_newlimts(const nvlist_t *oldfunclimits, + const nvlist_t *newfunclimit) +{ + void *cookie; + + cookie = NULL; + while (nvlist_next(newfunclimit, NULL, &cookie) != NULL) { + void *sacookie; + + if (strcmp(cnvlist_name(cookie), "sockaddr") != 0) + return (false); + + if (cnvlist_type(cookie) != NV_TYPE_NVLIST) + return (false); + + sacookie = NULL; + while (nvlist_next(cnvlist_get_nvlist(cookie), NULL, + &sacookie) != NULL) { + const void *sa; + size_t sasize; + + if (cnvlist_type(sacookie) != NV_TYPE_BINARY) + return (false); + + sa = cnvlist_get_binary(sacookie, &sasize); + if (!net_allowed_bsaddr(oldfunclimits, sa, sasize)) + return (false); + } + } + + return (true); +} + +static bool +verify_bind_newlimts(const nvlist_t *oldlimits, + const nvlist_t *newfunclimit) +{ + const nvlist_t *oldfunclimits; + + oldfunclimits = NULL; + if (oldlimits != NULL) { + oldfunclimits = dnvlist_get_nvlist(oldlimits, LIMIT_NV_BIND, + NULL); + } + + return (verify_only_sa_newlimts(oldfunclimits, newfunclimit)); +} + + +static bool +verify_connect_newlimits(const nvlist_t *oldlimits, + const nvlist_t *newfunclimit) +{ + const nvlist_t *oldfunclimits; + + oldfunclimits = NULL; + if (oldlimits != NULL) { + oldfunclimits = dnvlist_get_nvlist(oldlimits, LIMIT_NV_CONNECT, + NULL); + } + + return (verify_only_sa_newlimts(oldfunclimits, newfunclimit)); +} + +static bool +verify_addr2name_newlimits(const nvlist_t *oldlimits, + const nvlist_t *newfunclimit) +{ + void *cookie; + const nvlist_t *oldfunclimits; + + oldfunclimits = NULL; + if (oldlimits != NULL) { + oldfunclimits = dnvlist_get_nvlist(oldlimits, + LIMIT_NV_ADDR2NAME, NULL); + } + + cookie = NULL; + while (nvlist_next(newfunclimit, NULL, &cookie) != NULL) { + if (strcmp(cnvlist_name(cookie), "sockaddr") == 0) { + void *sacookie; + + if (cnvlist_type(cookie) != NV_TYPE_NVLIST) + return (false); + + sacookie = NULL; + while (nvlist_next(cnvlist_get_nvlist(cookie), NULL, + &sacookie) != NULL) { + const void *sa; + size_t sasize; + + if (cnvlist_type(sacookie) != NV_TYPE_BINARY) + return (false); + + sa = cnvlist_get_binary(sacookie, &sasize); + if (!net_allowed_bsaddr(oldfunclimits, sa, + sasize)) { + return (false); + } + } + } else if (strcmp(cnvlist_name(cookie), "family") == 0) { + size_t i, sfamilies; + const uint64_t *families; + + if (cnvlist_type(cookie) != NV_TYPE_NUMBER_ARRAY) + return (false); + + families = cnvlist_get_number_array(cookie, &sfamilies); + for (i = 0; i < sfamilies; i++) { + if (!net_allowed_family(oldfunclimits, + families[i])) { + return (false); + } + } + } else { + return (false); + } + } + + return (true); +} + +static bool +verify_name2addr_newlimits(const nvlist_t *oldlimits, + const nvlist_t *newfunclimit) +{ + void *cookie; + const nvlist_t *oldfunclimits; + + oldfunclimits = NULL; + if (oldlimits != NULL) { + oldfunclimits = dnvlist_get_nvlist(oldlimits, + LIMIT_NV_ADDR2NAME, NULL); + } + + cookie = NULL; + while (nvlist_next(newfunclimit, NULL, &cookie) != NULL) { + if (strcmp(cnvlist_name(cookie), "hosts") == 0) { + void *hostcookie; + + if (cnvlist_type(cookie) != NV_TYPE_NVLIST) + return (false); + + hostcookie = NULL; + while (nvlist_next(cnvlist_get_nvlist(cookie), NULL, + &hostcookie) != NULL) { + if (cnvlist_type(hostcookie) != NV_TYPE_STRING) + return (false); + + if (!net_allowed_hosts(oldfunclimits, + cnvlist_name(hostcookie), + cnvlist_get_string(hostcookie))) { + return (false); + } + } + } else if (strcmp(cnvlist_name(cookie), "family") == 0) { + size_t i, sfamilies; + const uint64_t *families; + + if (cnvlist_type(cookie) != NV_TYPE_NUMBER_ARRAY) + return (false); + + families = cnvlist_get_number_array(cookie, &sfamilies); + for (i = 0; i < sfamilies; i++) { + if (!net_allowed_family(oldfunclimits, + families[i])) { + return (false); + } + } + } else { + return (false); + } + } + + return (true); +} + +static int +net_limit(const nvlist_t *oldlimits, const nvlist_t *newlimits) +{ + const char *name; + void *cookie; + bool hasmode, hasconnect, hasbind, hasaddr2name, hasname2addr; + + /* + * Modes: + * ADDR2NAME: + * getnameinfo + * DEPRECATED_ADDR2NAME: + * gethostbyaddr + * + * NAME2ADDR: + * getaddrinfo + * DEPRECATED_NAME2ADDR: + * gethostbyname + * + * Limit scheme: + * mode : NV_TYPE_NUMBER + * connect : NV_TYPE_NVLIST + * sockaddr : NV_TYPE_NVLIST + * "" : NV_TYPE_BINARY + * ... : NV_TYPE_BINARY + * bind : NV_TYPE_NVLIST + * sockaddr : NV_TYPE_NVLIST + * "" : NV_TYPE_BINARY + * ... : NV_TYPE_BINARY + * addr2name : NV_TYPE_NVLIST + * family : NV_TYPE_NUMBER_ARRAY + * sockaddr : NV_TYPE_NVLIST + * "" : NV_TYPE_BINARY + * ... : NV_TYPE_BINARY + * name2addr : NV_TYPE_NVLIST + * family : NV_TYPE_NUMBER + * hosts : NV_TYPE_NVLIST + * host : servname : NV_TYPE_STRING + */ + + hasmode = false; + hasconnect = false; + hasbind = false; + hasaddr2name = false; + hasname2addr = false; + + cookie = NULL; + while ((name = nvlist_next(newlimits, NULL, &cookie)) != NULL) { + if (strcmp(name, "mode") == 0) { + if (cnvlist_type(cookie) != NV_TYPE_NUMBER) { + return (NO_RECOVERY); + } + if (!net_allowed_mode(oldlimits, + cnvlist_get_number(cookie))) { + return (ENOTCAPABLE); + } + hasmode = true; + continue; + } + + if (cnvlist_type(cookie) != NV_TYPE_NVLIST) { + return (NO_RECOVERY); + } + + if (strcmp(name, LIMIT_NV_BIND) == 0) { + hasbind = true; + if (!verify_bind_newlimts(oldlimits, + cnvlist_get_nvlist(cookie))) { + return (ENOTCAPABLE); + } + } else if (strcmp(name, LIMIT_NV_CONNECT) == 0) { + hasconnect = true; + if (!verify_connect_newlimits(oldlimits, + cnvlist_get_nvlist(cookie))) { + return (ENOTCAPABLE); + } + } else if (strcmp(name, LIMIT_NV_ADDR2NAME) == 0) { + hasaddr2name = true; + if (!verify_addr2name_newlimits(oldlimits, + cnvlist_get_nvlist(cookie))) { + return (ENOTCAPABLE); + } + } else if (strcmp(name, LIMIT_NV_NAME2ADDR) == 0) { + hasname2addr = true; + if (!verify_name2addr_newlimits(oldlimits, + cnvlist_get_nvlist(cookie))) { + return (ENOTCAPABLE); + } + } + } + + /* Mode is required. */ + if (!hasmode) + return (ENOTCAPABLE); + + /* + * If the new limit doesn't mention mode or family we have to + * check if the current limit does have those. Missing mode or + * family in the limit means that all modes or families are + * allowed. + */ + if (oldlimits == NULL) + return (0); + if (!hasconnect && nvlist_exists(oldlimits, LIMIT_NV_BIND)) + return (ENOTCAPABLE); + if (!hasconnect && nvlist_exists(oldlimits, LIMIT_NV_CONNECT)) + return (ENOTCAPABLE); + if (!hasaddr2name && nvlist_exists(oldlimits, LIMIT_NV_ADDR2NAME)) + return (ENOTCAPABLE); + if (!hasname2addr && nvlist_exists(oldlimits, LIMIT_NV_NAME2ADDR)) + return (ENOTCAPABLE); + return (0); +} + +static int +net_command(const char *cmd, const nvlist_t *limits, nvlist_t *nvlin, + nvlist_t *nvlout) +{ + + if (strcmp(cmd, "bind") == 0) + return (net_bind(limits, nvlin, nvlout)); + else if (strcmp(cmd, "connect") == 0) + return (net_connect(limits, nvlin, nvlout)); + else if (strcmp(cmd, "gethostbyname") == 0) + return (net_gethostbyname(limits, nvlin, nvlout)); + else if (strcmp(cmd, "gethostbyaddr") == 0) + return (net_gethostbyaddr(limits, nvlin, nvlout)); + else if (strcmp(cmd, "getnameinfo") == 0) + return (net_getnameinfo(limits, nvlin, nvlout)); + else if (strcmp(cmd, "getaddrinfo") == 0) + return (net_getaddrinfo(limits, nvlin, nvlout)); + + return (EINVAL); +} + +CREATE_SERVICE("system.net", net_limit, net_command, 0); diff --git a/lib/libcasper/services/cap_net/cap_net.h b/lib/libcasper/services/cap_net/cap_net.h new file mode 100644 index 00000000000..59943c70072 --- /dev/null +++ b/lib/libcasper/services/cap_net/cap_net.h @@ -0,0 +1,165 @@ +/*- + * SPDX-License-Identifier: BSD-2-Clause-FreeBSD + * + * Copyright (c) 2020 Mariusz Zaborski + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHORS AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHORS OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + * + * $FreeBSD$ + */ + +#ifndef _CAP_NETWORK_H_ +#define _CAP_NETWORK_H_ + +#ifdef HAVE_CASPER +#define WITH_CASPER +#endif + +#include +#include + +#include + +struct addrinfo; +struct hostent; + +struct cap_net_limit; +typedef struct cap_net_limit cap_net_limit_t; + +#define CAPNET_ADDR2NAME (0x01) +#define CAPNET_NAME2ADDR (0x02) +#define CAPNET_DEPRECATED_ADDR2NAME (0x04) +#define CAPNET_DEPRECATED_NAME2ADDR (0x08) +#define CAPNET_CONNECT (0x10) +#define CAPNET_BIND (0x20) +#define CAPNET_CONNECTDNS (0x40) + +#ifdef WITH_CASPER +/* Capability functions. */ +int cap_bind(cap_channel_t *chan, int s, const struct sockaddr *addr, + socklen_t addrlen); +int cap_connect(cap_channel_t *chan, int s, const struct sockaddr *name, + socklen_t namelen); + +int cap_getaddrinfo(cap_channel_t *chan, const char *hostname, + const char *servname, const struct addrinfo *hints, struct addrinfo **res); +int cap_getnameinfo(cap_channel_t *chan, const struct sockaddr *sa, + socklen_t salen, char *host, size_t hostlen, char *serv, size_t servlen, + int flags); + +/* Limit functions. */ +cap_net_limit_t *cap_net_limit_init(cap_channel_t *chan, uint64_t mode); +int cap_net_limit(cap_net_limit_t *limit); +void cap_net_free(cap_net_limit_t *limit); + +cap_net_limit_t *cap_net_limit_addr2name_family(cap_net_limit_t *limit, + int *family, size_t size); +cap_net_limit_t *cap_net_limit_addr2name(cap_net_limit_t *limit, + const struct sockaddr *sa, socklen_t salen); + +cap_net_limit_t *cap_net_limit_name2addr_family(cap_net_limit_t *limit, + int *family, size_t size); +cap_net_limit_t *cap_net_limit_name2addr(cap_net_limit_t *limit, + const char *name, const char *serv); + +cap_net_limit_t *cap_net_limit_connect(cap_net_limit_t *limit, + const struct sockaddr *sa, socklen_t salen); + +cap_net_limit_t *cap_net_limit_bind(cap_net_limit_t *limit, + const struct sockaddr *sa, socklen_t salen); + +/* Deprecated functions. */ +struct hostent *cap_gethostbyname(cap_channel_t *chan, const char *name); +struct hostent *cap_gethostbyname2(cap_channel_t *chan, const char *name, + int af); +struct hostent *cap_gethostbyaddr(cap_channel_t *chan, const void *addr, + socklen_t len, int af); +#else +/* Capability functions. */ +#define cap_bind(chan, s, addr, addrlen) \ + bind(s, addr, addrlen) +#define cap_connect(chan, s, name, namelen) \ + connect(s, name, namelen) +#define cap_getaddrinfo(chan, hostname, servname, hints, res) \ + getaddrinfo(hostname, servname, hints, res) +#define cap_getnameinfo(chan, sa, salen, host, hostlen, serv, servlen, flags) \ + getnameinfo(sa, salen, host, hostlen, serv, servlen, flags) + +/* Limit functions. */ +#define cap_net_limit_init(chan, mode) ((cap_net_limit_t *)malloc(8)) +#define cap_net_free(limit) free(limit) +static inline int +cap_net_limit(cap_net_limit_t *limit) +{ + free(limit); + return (0); +} + +static inline cap_net_limit_t * +cap_net_limit_addr2name_family(cap_net_limit_t *limit, + int *family __unused, size_t size __unused) +{ + return (limit); +} + +static inline cap_net_limit_t * +cap_net_limit_addr2name(cap_net_limit_t *limit, + const struct sockaddr *sa __unused, socklen_t salen __unused) +{ + return (limit); +} + +static inline cap_net_limit_t * +cap_net_limit_name2addr_family(cap_net_limit_t *limit, + int *family __unused, size_t size __unused) +{ + return (limit); +} + +static inline cap_net_limit_t * +cap_net_limit_name2addr(cap_net_limit_t *limit, + const char *name __unused, const char *serv __unused) +{ + return (limit); +} + +static inline cap_net_limit_t * +cap_net_limit_connect(cap_net_limit_t *limit, + const struct sockaddr *sa __unused, socklen_t salen __unused) +{ + return (limit); +} + +static inline cap_net_limit_t * +cap_net_limit_bind(cap_net_limit_t *limit, + const struct sockaddr *sa __unused, socklen_t salen __unused) +{ + return (limit); +} + +/* Deprecated functions. */ +#define cap_gethostbyname(chan, name) gethostbyname(name) +#define cap_gethostbyname2(chan, name, type) gethostbyname2(name, type) +#define cap_gethostbyaddr(chan, addr, len, type) gethostbyaddr(addr, len, type) +#endif + +#endif /* !_CAP_NETWORK_H_ */ diff --git a/lib/libcasper/services/cap_net/tests/Makefile b/lib/libcasper/services/cap_net/tests/Makefile new file mode 100644 index 00000000000..915a2dd3546 --- /dev/null +++ b/lib/libcasper/services/cap_net/tests/Makefile @@ -0,0 +1,16 @@ +# $FreeBSD$ + +.include + +ATF_TESTS_C= net_test + +.if ${MK_CASPER} != "no" +LIBADD+= casper +LIBADD+= cap_net +CFLAGS+=-DWITH_CASPER +.endif +LIBADD+= nv + +WARNS?= 3 + +.include diff --git a/lib/libcasper/services/cap_net/tests/net_test.c b/lib/libcasper/services/cap_net/tests/net_test.c new file mode 100644 index 00000000000..d73270095e1 --- /dev/null +++ b/lib/libcasper/services/cap_net/tests/net_test.c @@ -0,0 +1,1160 @@ +/*- + * Copyright (c) 2020 Mariusz Zaborski + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS + * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED + * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR + * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS + * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR + * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF + * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS + * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN + * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) + * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE + * POSSIBILITY OF SUCH DAMAGE. + */ + +#include +__FBSDID("$FreeBSD$"); + +#include +#include +#include +#include + +#include +#include + +#include + +#include +#include + +#define TEST_DOMAIN_0 "example.com" +#define TEST_DOMAIN_1 "freebsd.org" +#define TEST_IPV4 "1.1.1.1" +#define TEST_IPV6 "2001:4860:4860::8888" +#define TEST_BIND_IPV4 "127.0.0.1" + +static cap_channel_t * +create_network_service(void) +{ + cap_channel_t *capcas, *capnet; + + capcas = cap_init(); + ATF_REQUIRE(capcas != NULL); + + capnet = cap_service_open(capcas, "system.net"); + ATF_REQUIRE(capnet != NULL); + + cap_close(capcas); + return (capnet); +} + +static int +test_getnameinfo_v4(cap_channel_t *chan, int family, const char *ip) +{ + struct sockaddr_in ipaddr; + char capfn[MAXHOSTNAMELEN]; + char origfn[MAXHOSTNAMELEN]; + int ret; + + memset(&ipaddr, 0, sizeof(ipaddr)); + ipaddr.sin_family = family; + inet_pton(family, ip, &ipaddr.sin_addr); + + ret = cap_getnameinfo(chan, (struct sockaddr *)&ipaddr, sizeof(ipaddr), + capfn, sizeof(capfn), NULL, 0, NI_NAMEREQD); + if (ret != 0) { + return (ret); + } + + ret = getnameinfo((struct sockaddr *)&ipaddr, sizeof(ipaddr), origfn, + sizeof(origfn), NULL, 0, NI_NAMEREQD); + ATF_REQUIRE(ret == 0); + ATF_REQUIRE(strcmp(origfn, capfn) == 0); + + return (0); +} + +static int +test_getnameinfo_v6(cap_channel_t *chan, const char *ip) +{ + struct sockaddr_in6 ipaddr; + char capfn[MAXHOSTNAMELEN]; + char origfn[MAXHOSTNAMELEN]; + int ret; + + memset(&ipaddr, 0, sizeof(ipaddr)); + ipaddr.sin6_family = AF_INET6; + inet_pton(AF_INET6, ip, &ipaddr.sin6_addr); + + ret = cap_getnameinfo(chan, (struct sockaddr *)&ipaddr, sizeof(ipaddr), + capfn, sizeof(capfn), NULL, 0, NI_NAMEREQD); + if (ret != 0) { + return (ret); + } + + ret = getnameinfo((struct sockaddr *)&ipaddr, sizeof(ipaddr), origfn, + sizeof(origfn), NULL, 0, NI_NAMEREQD); + ATF_REQUIRE(ret == 0); + ATF_REQUIRE(strcmp(origfn, capfn) == 0); + + return (0); +} + +static int +test_getnameinfo(cap_channel_t *chan, int family, const char *ip) +{ + + if (family == AF_INET6) { + return (test_getnameinfo_v6(chan, ip)); + } + + return (test_getnameinfo_v4(chan, family, ip)); +} + +static int +test_gethostbyaddr_v4(cap_channel_t *chan, int family, const char *ip) +{ + struct in_addr ipaddr; + struct hostent *caphp, *orighp; + + memset(&ipaddr, 0, sizeof(ipaddr)); + inet_pton(AF_INET, ip, &ipaddr); + + caphp = cap_gethostbyaddr(chan, &ipaddr, sizeof(ipaddr), family); + if (caphp == NULL) { + return (h_errno); + } + + orighp = gethostbyaddr(&ipaddr, sizeof(ipaddr), family); + ATF_REQUIRE(orighp != NULL); + ATF_REQUIRE(strcmp(caphp->h_name, caphp->h_name) == 0); + + return (0); +} + +static int +test_gethostbyaddr_v6(cap_channel_t *chan, const char *ip) +{ + struct in6_addr ipaddr; + struct hostent *caphp, *orighp; + + memset(&ipaddr, 0, sizeof(ipaddr)); + inet_pton(AF_INET6, ip, &ipaddr); + + caphp = cap_gethostbyaddr(chan, &ipaddr, sizeof(ipaddr), AF_INET6); + if (caphp == NULL) + return (h_errno); + + orighp = gethostbyaddr(&ipaddr, sizeof(ipaddr), AF_INET6); + ATF_REQUIRE(orighp != NULL); + ATF_REQUIRE(strcmp(caphp->h_name, caphp->h_name) == 0); + + return (0); +} + +static int +test_gethostbyaddr(cap_channel_t *chan, int family, const char *ip) +{ + + if (family == AF_INET6) { + return (test_gethostbyaddr_v6(chan, ip)); + } else { + return (test_gethostbyaddr_v4(chan, family, ip)); + } +} + +static int +test_getaddrinfo(cap_channel_t *chan, int family, const char *domain, + const char *servname) +{ + struct addrinfo hints, *capres, *origres, *res0, *res1; + bool found; + int ret; + + memset(&hints, 0, sizeof(hints)); + hints.ai_family = family; + hints.ai_socktype = SOCK_STREAM; + + ret = cap_getaddrinfo(chan, domain, servname, &hints, &capres); + if (ret != 0) { + return (ret); + } + + ret = getaddrinfo(domain, servname, &hints, &origres); + ATF_REQUIRE(ret == 0); + + for (res0 = capres; res0 != NULL; res0 = res0->ai_next) { + found = false; + for (res1 = origres; res1 != NULL; res1 = res1->ai_next) { + if (res1->ai_addrlen == res0->ai_addrlen && + memcmp(res1->ai_addr, res0->ai_addr, + res0->ai_addrlen) == 0) { + found = true; + break; + } + } + ATF_REQUIRE(found); + } + + freeaddrinfo(capres); + freeaddrinfo(origres); + return (0); +} + +static int +test_gethostbyname(cap_channel_t *chan, int family, const char *domain) +{ + struct hostent *caphp, *orighp; + + caphp = cap_gethostbyname2(chan, domain, family); + if (caphp == NULL) { + return (h_errno); + } + + orighp = gethostbyname2(domain, family); + ATF_REQUIRE(orighp != NULL); + ATF_REQUIRE(strcmp(caphp->h_name, orighp->h_name) == 0); + + return (0); +} + +static int +test_bind(cap_channel_t *chan, const char *ip) +{ + struct sockaddr_in ipv4; + int capfd, ret, serrno; + + capfd = socket(AF_INET, SOCK_STREAM, 0); + ATF_REQUIRE(capfd > 0); + + memset(&ipv4, 0, sizeof(ipv4)); + ipv4.sin_family = AF_INET; + inet_pton(AF_INET, ip, &ipv4.sin_addr); + + ret = cap_bind(chan, capfd, (struct sockaddr *)&ipv4, sizeof(ipv4)); + serrno = errno; + close(capfd); + + return (ret < 0 ? serrno : 0); +} + +static int +test_connect(cap_channel_t *chan, const char *ip, unsigned short port) +{ + struct sockaddr_in ipv4; + int capfd, ret, serrno; + + capfd = socket(AF_INET, SOCK_STREAM, 0); + ATF_REQUIRE(capfd > 0); + + memset(&ipv4, 0, sizeof(ipv4)); + ipv4.sin_family = AF_INET; + ipv4.sin_port = htons(port); + inet_pton(AF_INET, ip, &ipv4.sin_addr); + + ret = cap_connect(chan, capfd, (struct sockaddr *)&ipv4, sizeof(ipv4)); + serrno = errno; + close(capfd); + + return (ret < 0 ? serrno : 0); +} + +static void +test_extend_mode(cap_channel_t *capnet, int current) +{ + cap_net_limit_t *limit; + const int rights[] = { + CAPNET_ADDR2NAME, + CAPNET_NAME2ADDR, + CAPNET_DEPRECATED_ADDR2NAME, + CAPNET_DEPRECATED_NAME2ADDR, + CAPNET_CONNECT, + CAPNET_BIND, + CAPNET_CONNECTDNS + }; + size_t i; + + for (i = 0; i < nitems(rights); i++) { + if (current == rights[i]) + continue; + + limit = cap_net_limit_init(capnet, current | rights[i]); + ATF_REQUIRE(limit != NULL); + ATF_REQUIRE(cap_net_limit(limit) != 0); + } +} + +ATF_TC_WITHOUT_HEAD(capnet__limits_addr2name_mode); +ATF_TC_BODY(capnet__limits_addr2name_mode, tc) +{ + cap_channel_t *capnet; + cap_net_limit_t *limit; + + capnet = create_network_service(); + + /* LIMIT */ + limit = cap_net_limit_init(capnet, CAPNET_ADDR2NAME); + ATF_REQUIRE(limit != NULL); + ATF_REQUIRE(cap_net_limit(limit) == 0); + + /* ALLOWED */ + ATF_REQUIRE(test_getnameinfo(capnet, AF_INET, TEST_IPV4) == 0); + + /* DISALLOWED */ + ATF_REQUIRE(test_gethostbyname(capnet, AF_INET, TEST_DOMAIN_0) == + ENOTCAPABLE); + ATF_REQUIRE(test_gethostbyaddr(capnet, AF_INET, TEST_IPV4) == + ENOTCAPABLE); + ATF_REQUIRE(test_getaddrinfo(capnet, AF_INET, TEST_DOMAIN_0, NULL) == + ENOTCAPABLE); + ATF_REQUIRE(test_bind(capnet, TEST_BIND_IPV4) == ENOTCAPABLE); + ATF_REQUIRE(test_connect(capnet, TEST_IPV4, 80) == ENOTCAPABLE); + + test_extend_mode(capnet, CAPNET_ADDR2NAME); + + cap_close(capnet); +} + +ATF_TC_WITHOUT_HEAD(capnet__limits_addr2name_family); +ATF_TC_BODY(capnet__limits_addr2name_family, tc) +{ + cap_channel_t *capnet; + cap_net_limit_t *limit; + int family[] = { AF_INET6, AF_INET }; + + capnet = create_network_service(); + + /* Limit to AF_INET6 and AF_INET. */ + limit = cap_net_limit_init(capnet, CAPNET_ADDR2NAME); + ATF_REQUIRE(limit != NULL); + cap_net_limit_addr2name_family(limit, family, nitems(family)); + ATF_REQUIRE(cap_net_limit(limit) == 0); + + ATF_REQUIRE(test_getnameinfo(capnet, AF_INET, TEST_IPV4) == 0); + ATF_REQUIRE(test_getnameinfo(capnet, AF_INET6, TEST_IPV6) == 0); + + /* Limit to AF_INET6 and AF_INET. */ + limit = cap_net_limit_init(capnet, CAPNET_ADDR2NAME); + ATF_REQUIRE(limit != NULL); + cap_net_limit_addr2name_family(limit, &family[0], 1); + cap_net_limit_addr2name_family(limit, &family[1], 1); + ATF_REQUIRE(cap_net_limit(limit) == 0); + + ATF_REQUIRE(test_getnameinfo(capnet, AF_INET, TEST_IPV4) == 0); + ATF_REQUIRE(test_getnameinfo(capnet, AF_INET6, TEST_IPV6) == 0); + + /* Limit to AF_INET6. */ + limit = cap_net_limit_init(capnet, CAPNET_ADDR2NAME); + ATF_REQUIRE(limit != NULL); + cap_net_limit_addr2name_family(limit, family, 1); + ATF_REQUIRE(cap_net_limit(limit) == 0); + + ATF_REQUIRE(test_getnameinfo(capnet, AF_INET, TEST_IPV4) == + ENOTCAPABLE); + ATF_REQUIRE(test_getnameinfo(capnet, AF_INET6, TEST_IPV6) == 0); + + /* Unable to set empty limits. Empty limits means full access. */ + limit = cap_net_limit_init(capnet, CAPNET_ADDR2NAME); + ATF_REQUIRE(cap_net_limit(limit) != 0); + + cap_close(capnet); +} + +ATF_TC_WITHOUT_HEAD(capnet__limits_addr2name); +ATF_TC_BODY(capnet__limits_addr2name, tc) +{ + cap_channel_t *capnet; + cap_net_limit_t *limit; + struct sockaddr_in ipaddrv4; + struct sockaddr_in6 ipaddrv6; + + capnet = create_network_service(); + + /* Limit to TEST_IPV4 and TEST_IPV6. */ + memset(&ipaddrv4, 0, sizeof(ipaddrv4)); + memset(&ipaddrv6, 0, sizeof(ipaddrv6)); + + ipaddrv4.sin_family = AF_INET; + inet_pton(AF_INET, TEST_IPV4, &ipaddrv4.sin_addr); + + ipaddrv6.sin6_family = AF_INET6; + inet_pton(AF_INET6, TEST_IPV6, &ipaddrv6.sin6_addr); + + limit = cap_net_limit_init(capnet, CAPNET_ADDR2NAME); + ATF_REQUIRE(limit != NULL); + + cap_net_limit_addr2name(limit, (struct sockaddr *)&ipaddrv4, + sizeof(ipaddrv4)); + cap_net_limit_addr2name(limit, (struct sockaddr *)&ipaddrv6, + sizeof(ipaddrv6)); + ATF_REQUIRE(cap_net_limit(limit) == 0); + + ATF_REQUIRE(test_getnameinfo(capnet, AF_INET, TEST_IPV4) == 0); + ATF_REQUIRE(test_getnameinfo(capnet, AF_INET6, TEST_IPV6) == 0); + ATF_REQUIRE(test_getnameinfo(capnet, AF_INET, "127.0.0.1") == + ENOTCAPABLE); + + /* Limit to AF_INET. */ + limit = cap_net_limit_init(capnet, CAPNET_ADDR2NAME); + ATF_REQUIRE(limit != NULL); + cap_net_limit_addr2name(limit, (struct sockaddr *)&ipaddrv4, + sizeof(ipaddrv4)); + ATF_REQUIRE(cap_net_limit(limit) == 0); + + ATF_REQUIRE(test_getnameinfo(capnet, AF_INET, TEST_IPV4) == 0); + ATF_REQUIRE(test_getnameinfo(capnet, AF_INET6, TEST_IPV6) == + ENOTCAPABLE); + ATF_REQUIRE(test_getnameinfo(capnet, AF_INET, "127.0.0.1") == + ENOTCAPABLE); + + /* Unable to set empty limits. Empty limits means full access. */ + limit = cap_net_limit_init(capnet, CAPNET_ADDR2NAME); + ATF_REQUIRE(cap_net_limit(limit) != 0); + + cap_close(capnet); +} + +ATF_TC_WITHOUT_HEAD(capnet__limits_deprecated_addr2name_mode); +ATF_TC_BODY(capnet__limits_deprecated_addr2name_mode, tc) +{ + cap_channel_t *capnet; + cap_net_limit_t *limit; + + capnet = create_network_service(); + + /* LIMIT */ + limit = cap_net_limit_init(capnet, CAPNET_DEPRECATED_ADDR2NAME); + ATF_REQUIRE(limit != NULL); + ATF_REQUIRE(cap_net_limit(limit) == 0); + + /* ALLOWED */ + ATF_REQUIRE(test_gethostbyaddr(capnet, AF_INET, TEST_IPV4) == 0); + + /* DISALLOWED */ + ATF_REQUIRE(test_gethostbyname(capnet, AF_INET, TEST_DOMAIN_0) == + ENOTCAPABLE); + ATF_REQUIRE(test_getnameinfo(capnet, AF_INET, TEST_IPV4) == + ENOTCAPABLE); + ATF_REQUIRE(test_getaddrinfo(capnet, AF_INET, TEST_DOMAIN_0, NULL) == + ENOTCAPABLE); + ATF_REQUIRE(test_bind(capnet, TEST_BIND_IPV4) == ENOTCAPABLE); + ATF_REQUIRE(test_connect(capnet, TEST_IPV4, 80) == ENOTCAPABLE); + + cap_close(capnet); +} + +ATF_TC_WITHOUT_HEAD(capnet__limits_deprecated_addr2name_family); +ATF_TC_BODY(capnet__limits_deprecated_addr2name_family, tc) +{ + cap_channel_t *capnet; + cap_net_limit_t *limit; + int family[] = { AF_INET6, AF_INET }; + + capnet = create_network_service(); + + /* Limit to AF_INET6 and AF_INET. */ + limit = cap_net_limit_init(capnet, CAPNET_DEPRECATED_ADDR2NAME); + ATF_REQUIRE(limit != NULL); + cap_net_limit_addr2name_family(limit, family, nitems(family)); + ATF_REQUIRE(cap_net_limit(limit) == 0); + + ATF_REQUIRE(test_gethostbyaddr(capnet, AF_INET, TEST_IPV4) == 0); + ATF_REQUIRE(test_gethostbyaddr(capnet, AF_INET6, TEST_IPV6) == 0); + ATF_REQUIRE(test_gethostbyaddr(capnet, PF_LINK, TEST_IPV4) == + ENOTCAPABLE); + + /* Limit to AF_INET6 and AF_INET. */ + limit = cap_net_limit_init(capnet, CAPNET_DEPRECATED_ADDR2NAME); + ATF_REQUIRE(limit != NULL); + cap_net_limit_addr2name_family(limit, &family[0], 1); + cap_net_limit_addr2name_family(limit, &family[1], 1); + ATF_REQUIRE(cap_net_limit(limit) == 0); + + ATF_REQUIRE(test_gethostbyaddr(capnet, AF_INET, TEST_IPV4) == 0); + ATF_REQUIRE(test_gethostbyaddr(capnet, AF_INET6, TEST_IPV6) == 0); + ATF_REQUIRE(test_gethostbyaddr(capnet, PF_LINK, TEST_IPV4) == + ENOTCAPABLE); + + /* Limit to AF_INET6. */ + limit = cap_net_limit_init(capnet, CAPNET_DEPRECATED_ADDR2NAME); + ATF_REQUIRE(limit != NULL); + cap_net_limit_addr2name_family(limit, family, 1); + ATF_REQUIRE(cap_net_limit(limit) == 0); + + ATF_REQUIRE(test_gethostbyaddr(capnet, AF_INET, TEST_IPV4) == + ENOTCAPABLE); + ATF_REQUIRE(test_gethostbyaddr(capnet, AF_INET6, TEST_IPV6) == 0); + ATF_REQUIRE(test_gethostbyaddr(capnet, PF_LINK, TEST_IPV4) == + ENOTCAPABLE); + + /* Unable to set empty limits. Empty limits means full access. */ + limit = cap_net_limit_init(capnet, CAPNET_DEPRECATED_ADDR2NAME); + ATF_REQUIRE(cap_net_limit(limit) != 0); + + cap_close(capnet); +} + +ATF_TC_WITHOUT_HEAD(capnet__limits_deprecated_addr2name); +ATF_TC_BODY(capnet__limits_deprecated_addr2name, tc) +{ + cap_channel_t *capnet; + cap_net_limit_t *limit; + struct in_addr ipaddrv4; + struct in6_addr ipaddrv6; + + capnet = create_network_service(); + + /* Limit to TEST_IPV4 and TEST_IPV6. */ + memset(&ipaddrv4, 0, sizeof(ipaddrv4)); + memset(&ipaddrv6, 0, sizeof(ipaddrv6)); + + inet_pton(AF_INET, TEST_IPV4, &ipaddrv4); + inet_pton(AF_INET6, TEST_IPV6, &ipaddrv6); + + limit = cap_net_limit_init(capnet, CAPNET_DEPRECATED_ADDR2NAME); + ATF_REQUIRE(limit != NULL); + + cap_net_limit_addr2name(limit, (struct sockaddr *)&ipaddrv4, + sizeof(ipaddrv4)); + cap_net_limit_addr2name(limit, (struct sockaddr *)&ipaddrv6, + sizeof(ipaddrv6)); + ATF_REQUIRE(cap_net_limit(limit) == 0); + + ATF_REQUIRE(test_gethostbyaddr(capnet, AF_INET, TEST_IPV4) == 0); + ATF_REQUIRE(test_gethostbyaddr(capnet, AF_INET6, TEST_IPV6) == 0); + ATF_REQUIRE(test_gethostbyaddr(capnet, AF_INET, "127.0.0.1") == + ENOTCAPABLE); + + /* Limit to AF_INET. */ + limit = cap_net_limit_init(capnet, CAPNET_DEPRECATED_ADDR2NAME); + ATF_REQUIRE(limit != NULL); + cap_net_limit_addr2name(limit, (struct sockaddr *)&ipaddrv4, + sizeof(ipaddrv4)); + ATF_REQUIRE(cap_net_limit(limit) == 0); + + ATF_REQUIRE(test_gethostbyaddr(capnet, AF_INET, TEST_IPV4) == 0); + ATF_REQUIRE(test_gethostbyaddr(capnet, AF_INET6, TEST_IPV6) == + ENOTCAPABLE); + ATF_REQUIRE(test_gethostbyaddr(capnet, AF_INET, "127.0.0.1") == + ENOTCAPABLE); + + /* Unable to set empty limits. Empty limits means full access. */ + limit = cap_net_limit_init(capnet, CAPNET_DEPRECATED_ADDR2NAME); + ATF_REQUIRE(cap_net_limit(limit) != 0); + + cap_close(capnet); +} + + +ATF_TC_WITHOUT_HEAD(capnet__limits_name2addr_mode); +ATF_TC_BODY(capnet__limits_name2addr_mode, tc) +{ + cap_channel_t *capnet; + cap_net_limit_t *limit; + + capnet = create_network_service(); + + /* LIMIT */ + limit = cap_net_limit_init(capnet, CAPNET_NAME2ADDR); + ATF_REQUIRE(limit != NULL); + ATF_REQUIRE(cap_net_limit(limit) == 0); + + /* ALLOWED */ + ATF_REQUIRE(test_getaddrinfo(capnet, AF_INET, TEST_DOMAIN_0, NULL) == + 0); + + /* DISALLOWED */ + ATF_REQUIRE( + test_gethostbyname(capnet, AF_INET, TEST_DOMAIN_0) == ENOTCAPABLE); + ATF_REQUIRE(test_getnameinfo(capnet, AF_INET, TEST_IPV4) == + ENOTCAPABLE); + ATF_REQUIRE(test_gethostbyaddr(capnet, AF_INET, TEST_IPV4) == + ENOTCAPABLE); + ATF_REQUIRE(test_bind(capnet, TEST_BIND_IPV4) == ENOTCAPABLE); + ATF_REQUIRE(test_connect(capnet, TEST_IPV4, 80) == ENOTCAPABLE); + + test_extend_mode(capnet, CAPNET_ADDR2NAME); + + cap_close(capnet); +} + +ATF_TC_WITHOUT_HEAD(capnet__limits_name2addr_hosts); +ATF_TC_BODY(capnet__limits_name2addr_hosts, tc) +{ + cap_channel_t *capnet; + cap_net_limit_t *limit; + + capnet = create_network_service(); + + /* Limit to TEST_DOMAIN_0 and localhost only. */ + limit = cap_net_limit_init(capnet, CAPNET_NAME2ADDR); + ATF_REQUIRE(limit != NULL); + cap_net_limit_name2addr(limit, TEST_DOMAIN_0, NULL); + cap_net_limit_name2addr(limit, "localhost", NULL); + ATF_REQUIRE(cap_net_limit(limit) == 0); + + ATF_REQUIRE(test_getaddrinfo(capnet, AF_INET, TEST_DOMAIN_0, NULL) == + 0); + ATF_REQUIRE(test_getaddrinfo(capnet, AF_INET, "localhost", NULL) == 0); + ATF_REQUIRE(test_getaddrinfo(capnet, AF_INET, TEST_DOMAIN_1, NULL) == + ENOTCAPABLE); + + /* Limit to TEST_DOMAIN_0 only. */ + limit = cap_net_limit_init(capnet, CAPNET_NAME2ADDR); + ATF_REQUIRE(limit != NULL); + cap_net_limit_name2addr(limit, TEST_DOMAIN_0, NULL); + ATF_REQUIRE(cap_net_limit(limit) == 0); + + ATF_REQUIRE(test_getaddrinfo(capnet, AF_INET, "localhost", NULL) == + ENOTCAPABLE); + ATF_REQUIRE(test_getaddrinfo(capnet, AF_INET, TEST_DOMAIN_1, NULL) == + ENOTCAPABLE); + ATF_REQUIRE(test_getaddrinfo(capnet, AF_INET, TEST_DOMAIN_0, NULL) == + 0); + + /* Unable to set empty limits. Empty limits means full access. */ + limit = cap_net_limit_init(capnet, CAPNET_NAME2ADDR); + ATF_REQUIRE(cap_net_limit(limit) != 0); + + cap_close(capnet); +} + +ATF_TC_WITHOUT_HEAD(capnet__limits_name2addr_hosts_servnames_strict); +ATF_TC_BODY(capnet__limits_name2addr_hosts_servnames_strict, tc) +{ + cap_channel_t *capnet; + cap_net_limit_t *limit; + + capnet = create_network_service(); + + /* + * Limit to TEST_DOMAIN_0 and HTTP service. + */ + limit = cap_net_limit_init(capnet, CAPNET_NAME2ADDR); + ATF_REQUIRE(limit != NULL); + cap_net_limit_name2addr(limit, TEST_DOMAIN_0, "http"); + ATF_REQUIRE(cap_net_limit(limit) == 0); + + ATF_REQUIRE(test_getaddrinfo(capnet, AF_INET, TEST_DOMAIN_0, "http") == + 0); + ATF_REQUIRE(test_getaddrinfo(capnet, AF_INET, TEST_DOMAIN_0, NULL) == + ENOTCAPABLE); + ATF_REQUIRE(test_getaddrinfo(capnet, AF_INET, TEST_DOMAIN_0, "snmp") == + ENOTCAPABLE); + ATF_REQUIRE(test_getaddrinfo(capnet, AF_INET, TEST_DOMAIN_1, "http") == + ENOTCAPABLE); + + /* Unable to set empty limits. Empty limits means full access. */ + limit = cap_net_limit_init(capnet, CAPNET_NAME2ADDR); + ATF_REQUIRE(cap_net_limit(limit) != 0); + + cap_close(capnet); +} + +ATF_TC_WITHOUT_HEAD(capnet__limits_name2addr_hosts_servnames_mix); +ATF_TC_BODY(capnet__limits_name2addr_hosts_servnames_mix, tc) +{ + cap_channel_t *capnet; + cap_net_limit_t *limit; + + capnet = create_network_service(); + + /* + * Limit to TEST_DOMAIN_0 and any servnamex, and any domain with + * servname HTTP. + */ + limit = cap_net_limit_init(capnet, CAPNET_NAME2ADDR); + ATF_REQUIRE(limit != NULL); + cap_net_limit_name2addr(limit, TEST_DOMAIN_0, NULL); + cap_net_limit_name2addr(limit, NULL, "http"); + ATF_REQUIRE(cap_net_limit(limit) == 0); + + ATF_REQUIRE(test_getaddrinfo(capnet, AF_INET, TEST_DOMAIN_0, "http") == + 0); + ATF_REQUIRE(test_getaddrinfo(capnet, AF_INET, TEST_DOMAIN_0, NULL) == + 0); + ATF_REQUIRE(test_getaddrinfo(capnet, AF_INET, TEST_DOMAIN_1, "http") == + 0); + ATF_REQUIRE(test_getaddrinfo(capnet, AF_INET, TEST_DOMAIN_0, NULL) == + 0); + ATF_REQUIRE(test_getaddrinfo(capnet, AF_INET, TEST_DOMAIN_1, "snmp") == + ENOTCAPABLE); + + /* Limit to HTTTP servname only. */ + limit = cap_net_limit_init(capnet, CAPNET_NAME2ADDR); + ATF_REQUIRE(limit != NULL); + cap_net_limit_name2addr(limit, NULL, "http"); + ATF_REQUIRE(cap_net_limit(limit) == 0); + + ATF_REQUIRE(test_getaddrinfo(capnet, AF_INET, TEST_DOMAIN_0, "http") == + 0); + ATF_REQUIRE(test_getaddrinfo(capnet, AF_INET, TEST_DOMAIN_0, NULL) == + ENOTCAPABLE); + ATF_REQUIRE(test_getaddrinfo(capnet, AF_INET, TEST_DOMAIN_1, "http") == + 0); + ATF_REQUIRE(test_getaddrinfo(capnet, AF_INET, TEST_DOMAIN_0, NULL) == + ENOTCAPABLE); + ATF_REQUIRE(test_getaddrinfo(capnet, AF_INET, TEST_DOMAIN_1, "snmp") == + ENOTCAPABLE); + + /* Unable to set empty limits. Empty limits means full access. */ + limit = cap_net_limit_init(capnet, CAPNET_NAME2ADDR); + ATF_REQUIRE(cap_net_limit(limit) != 0); + + cap_close(capnet); +} + +ATF_TC_WITHOUT_HEAD(capnet__limits_name2addr_family); +ATF_TC_BODY(capnet__limits_name2addr_family, tc) +{ + cap_channel_t *capnet; + cap_net_limit_t *limit; + int family[] = { AF_INET6, AF_INET }; + + capnet = create_network_service(); + + /* Limit to AF_INET and AF_INET6. */ + limit = cap_net_limit_init(capnet, CAPNET_NAME2ADDR); + ATF_REQUIRE(limit != NULL); + cap_net_limit_name2addr(limit, TEST_DOMAIN_0, NULL); + cap_net_limit_name2addr_family(limit, family, nitems(family)); + ATF_REQUIRE(cap_net_limit(limit) == 0); + + ATF_REQUIRE(test_getaddrinfo(capnet, AF_INET, TEST_DOMAIN_0, NULL) == + 0); + ATF_REQUIRE(test_getaddrinfo(capnet, AF_INET6, TEST_DOMAIN_0, NULL) == + 0); + ATF_REQUIRE(test_getaddrinfo(capnet, PF_LINK, TEST_DOMAIN_0, NULL) == + ENOTCAPABLE); + + /* Limit to AF_INET and AF_INET6. */ + limit = cap_net_limit_init(capnet, CAPNET_NAME2ADDR); + ATF_REQUIRE(limit != NULL); + cap_net_limit_name2addr(limit, TEST_DOMAIN_0, NULL); + cap_net_limit_name2addr_family(limit, &family[0], 1); + cap_net_limit_name2addr_family(limit, &family[1], 1); + ATF_REQUIRE(cap_net_limit(limit) == 0); + + ATF_REQUIRE(test_getaddrinfo(capnet, AF_INET, TEST_DOMAIN_0, NULL) == + 0); + ATF_REQUIRE(test_getaddrinfo(capnet, AF_INET6, TEST_DOMAIN_0, NULL) == + 0); + ATF_REQUIRE(test_getaddrinfo(capnet, PF_LINK, TEST_DOMAIN_0, NULL) == + ENOTCAPABLE); + + /* Limit to AF_INET6 only. */ + limit = cap_net_limit_init(capnet, CAPNET_NAME2ADDR); + ATF_REQUIRE(limit != NULL); + cap_net_limit_name2addr(limit, TEST_DOMAIN_0, NULL); + cap_net_limit_name2addr_family(limit, family, 1); + ATF_REQUIRE(cap_net_limit(limit) == 0); + + ATF_REQUIRE(test_getaddrinfo(capnet, AF_INET, TEST_DOMAIN_0, NULL) == + ENOTCAPABLE); + ATF_REQUIRE(test_getaddrinfo(capnet, AF_INET6, TEST_DOMAIN_0, NULL) == + 0); + ATF_REQUIRE(test_getaddrinfo(capnet, PF_LINK, TEST_DOMAIN_0, NULL) == + ENOTCAPABLE); + + /* Unable to set empty limits. Empty limits means full access. */ + limit = cap_net_limit_init(capnet, CAPNET_NAME2ADDR); + ATF_REQUIRE(cap_net_limit(limit) != 0); + + cap_close(capnet); +} + +ATF_TC_WITHOUT_HEAD(capnet__limits_deprecated_name2addr_mode); +ATF_TC_BODY(capnet__limits_deprecated_name2addr_mode, tc) +{ + cap_channel_t *capnet; + cap_net_limit_t *limit; + + capnet = create_network_service(); + + /* LIMIT */ + limit = cap_net_limit_init(capnet, CAPNET_DEPRECATED_NAME2ADDR); + ATF_REQUIRE(limit != NULL); + ATF_REQUIRE(cap_net_limit(limit) == 0); + + /* ALLOWED */ + ATF_REQUIRE(test_gethostbyname(capnet, AF_INET, TEST_DOMAIN_0) == 0); + + /* DISALLOWED */ + ATF_REQUIRE(test_getnameinfo(capnet, AF_INET, TEST_IPV4) == + ENOTCAPABLE); + ATF_REQUIRE(test_gethostbyaddr(capnet, AF_INET, TEST_IPV4) == + ENOTCAPABLE); + ATF_REQUIRE(test_getaddrinfo(capnet, AF_INET, TEST_DOMAIN_0, NULL) == + ENOTCAPABLE); + ATF_REQUIRE(test_bind(capnet, TEST_BIND_IPV4) == ENOTCAPABLE); + ATF_REQUIRE(test_connect(capnet, TEST_IPV4, 80) == ENOTCAPABLE); + + test_extend_mode(capnet, CAPNET_ADDR2NAME); + + cap_close(capnet); +} + +ATF_TC_WITHOUT_HEAD(capnet__limits_deprecated_name2addr_hosts); +ATF_TC_BODY(capnet__limits_deprecated_name2addr_hosts, tc) +{ + cap_channel_t *capnet; + cap_net_limit_t *limit; + + capnet = create_network_service(); + + /* Limit to TEST_DOMAIN_0 and localhost only. */ + limit = cap_net_limit_init(capnet, CAPNET_DEPRECATED_NAME2ADDR); + ATF_REQUIRE(limit != NULL); + cap_net_limit_name2addr(limit, TEST_DOMAIN_0, NULL); + cap_net_limit_name2addr(limit, "localhost", NULL); + ATF_REQUIRE(cap_net_limit(limit) == 0); + + ATF_REQUIRE(test_gethostbyname(capnet, AF_INET, TEST_DOMAIN_0) == 0); + ATF_REQUIRE(test_gethostbyname(capnet, AF_INET, "localhost") == 0); + ATF_REQUIRE( + test_gethostbyname(capnet, AF_INET, TEST_DOMAIN_1) == ENOTCAPABLE); + + /* Limit to TEST_DOMAIN_0 only. */ + limit = cap_net_limit_init(capnet, CAPNET_DEPRECATED_NAME2ADDR); + ATF_REQUIRE(limit != NULL); + cap_net_limit_name2addr(limit, TEST_DOMAIN_0, NULL); + ATF_REQUIRE(cap_net_limit(limit) == 0); + + ATF_REQUIRE( + test_gethostbyname(capnet, AF_INET, "localhost") == ENOTCAPABLE); + ATF_REQUIRE( + test_gethostbyname(capnet, AF_INET, TEST_DOMAIN_1) == ENOTCAPABLE); + ATF_REQUIRE(test_gethostbyname(capnet, AF_INET, TEST_DOMAIN_0) == 0); + + /* Unable to set empty limits. Empty limits means full access. */ + limit = cap_net_limit_init(capnet, CAPNET_DEPRECATED_NAME2ADDR); + ATF_REQUIRE(cap_net_limit(limit) != 0); + + cap_close(capnet); +} + +ATF_TC_WITHOUT_HEAD(capnet__limits_deprecated_name2addr_family); +ATF_TC_BODY(capnet__limits_deprecated_name2addr_family, tc) +{ + cap_channel_t *capnet; + cap_net_limit_t *limit; + int family[] = { AF_INET6, AF_INET }; + + capnet = create_network_service(); + + /* Limit to AF_INET and AF_INET6. */ + limit = cap_net_limit_init(capnet, CAPNET_DEPRECATED_NAME2ADDR); + ATF_REQUIRE(limit != NULL); + cap_net_limit_name2addr(limit, TEST_DOMAIN_0, NULL); + cap_net_limit_name2addr_family(limit, family, nitems(family)); + ATF_REQUIRE(cap_net_limit(limit) == 0); + + ATF_REQUIRE(test_gethostbyname(capnet, AF_INET, TEST_DOMAIN_0) == 0); + ATF_REQUIRE(test_gethostbyname(capnet, AF_INET6, TEST_DOMAIN_0) == 0); + ATF_REQUIRE( + test_gethostbyname(capnet, PF_LINK, TEST_DOMAIN_0) == ENOTCAPABLE); + + /* Limit to AF_INET and AF_INET6. */ + limit = cap_net_limit_init(capnet, CAPNET_DEPRECATED_NAME2ADDR); + ATF_REQUIRE(limit != NULL); + cap_net_limit_name2addr(limit, TEST_DOMAIN_0, NULL); + cap_net_limit_name2addr_family(limit, &family[0], 1); + cap_net_limit_name2addr_family(limit, &family[1], 1); + ATF_REQUIRE(cap_net_limit(limit) == 0); + + ATF_REQUIRE(test_gethostbyname(capnet, AF_INET, TEST_DOMAIN_0) == 0); + ATF_REQUIRE(test_gethostbyname(capnet, AF_INET6, TEST_DOMAIN_0) == 0); + ATF_REQUIRE( + test_gethostbyname(capnet, PF_LINK, TEST_DOMAIN_0) == ENOTCAPABLE); + + /* Limit to AF_INET6 only. */ + limit = cap_net_limit_init(capnet, CAPNET_DEPRECATED_NAME2ADDR); + ATF_REQUIRE(limit != NULL); + cap_net_limit_name2addr(limit, TEST_DOMAIN_0, NULL); + cap_net_limit_name2addr_family(limit, family, 1); + ATF_REQUIRE(cap_net_limit(limit) == 0); + + ATF_REQUIRE( + test_gethostbyname(capnet, AF_INET, TEST_DOMAIN_0) == ENOTCAPABLE); + ATF_REQUIRE(test_gethostbyname(capnet, AF_INET6, TEST_DOMAIN_0) == 0); + ATF_REQUIRE( + test_gethostbyname(capnet, PF_LINK, TEST_DOMAIN_0) == ENOTCAPABLE); + + /* Unable to set empty limits. Empty limits means full access. */ + limit = cap_net_limit_init(capnet, CAPNET_DEPRECATED_NAME2ADDR); + ATF_REQUIRE(cap_net_limit(limit) != 0); + + cap_close(capnet); +} + +ATF_TC_WITHOUT_HEAD(capnet__limits_bind_mode); +ATF_TC_BODY(capnet__limits_bind_mode, tc) +{ + cap_channel_t *capnet; + cap_net_limit_t *limit; + + capnet = create_network_service(); + + /* LIMIT */ + limit = cap_net_limit_init(capnet, CAPNET_BIND); + ATF_REQUIRE(limit != NULL); + ATF_REQUIRE(cap_net_limit(limit) == 0); + + /* ALLOWED */ + ATF_REQUIRE(test_bind(capnet, TEST_BIND_IPV4) == 0); + + /* DISALLOWED */ + ATF_REQUIRE( + test_gethostbyname(capnet, AF_INET, TEST_DOMAIN_0) == ENOTCAPABLE); + ATF_REQUIRE(test_getnameinfo(capnet, AF_INET, TEST_IPV4) == + ENOTCAPABLE); + ATF_REQUIRE(test_gethostbyaddr(capnet, AF_INET, TEST_IPV4) == + ENOTCAPABLE); + ATF_REQUIRE(test_getaddrinfo(capnet, AF_INET, TEST_DOMAIN_0, NULL) == + ENOTCAPABLE); + ATF_REQUIRE(test_connect(capnet, TEST_IPV4, 80) == ENOTCAPABLE); + + test_extend_mode(capnet, CAPNET_ADDR2NAME); + + cap_close(capnet); +} + +ATF_TC_WITHOUT_HEAD(capnet__limits_bind); +ATF_TC_BODY(capnet__limits_bind, tc) +{ + cap_channel_t *capnet; + cap_net_limit_t *limit; + struct sockaddr_in ipv4; + + capnet = create_network_service(); + + limit = cap_net_limit_init(capnet, CAPNET_BIND); + ATF_REQUIRE(limit != NULL); + + memset(&ipv4, 0, sizeof(ipv4)); + ipv4.sin_family = AF_INET; + inet_pton(AF_INET, TEST_BIND_IPV4, &ipv4.sin_addr); + + cap_net_limit_bind(limit, (struct sockaddr *)&ipv4, sizeof(ipv4)); + ATF_REQUIRE(cap_net_limit(limit) == 0); + + ATF_REQUIRE(test_bind(capnet, TEST_BIND_IPV4) == 0); + ATF_REQUIRE(test_bind(capnet, "127.0.0.2") == ENOTCAPABLE); + + cap_close(capnet); +} + +ATF_TC_WITHOUT_HEAD(capnet__limits_connect_mode); +ATF_TC_BODY(capnet__limits_connect_mode, tc) +{ + cap_channel_t *capnet; + cap_net_limit_t *limit; + + capnet = create_network_service(); + + /* LIMIT */ + limit = cap_net_limit_init(capnet, CAPNET_CONNECT); + ATF_REQUIRE(limit != NULL); + ATF_REQUIRE(cap_net_limit(limit) == 0); + + /* ALLOWED */ + ATF_REQUIRE(test_connect(capnet, TEST_IPV4, 80) == 0); + + /* DISALLOWED */ + ATF_REQUIRE( + test_gethostbyname(capnet, AF_INET, TEST_DOMAIN_0) == ENOTCAPABLE); + ATF_REQUIRE(test_getnameinfo(capnet, AF_INET, TEST_IPV4) == + ENOTCAPABLE); + ATF_REQUIRE(test_gethostbyaddr(capnet, AF_INET, TEST_IPV4) == + ENOTCAPABLE); + ATF_REQUIRE(test_getaddrinfo(capnet, AF_INET, TEST_DOMAIN_0, NULL) == + ENOTCAPABLE); + ATF_REQUIRE(test_bind(capnet, TEST_BIND_IPV4) == ENOTCAPABLE); + + test_extend_mode(capnet, CAPNET_ADDR2NAME); + + cap_close(capnet); +} + +ATF_TC_WITHOUT_HEAD(capnet__limits_connect); +ATF_TC_BODY(capnet__limits_connect, tc) +{ + cap_channel_t *capnet; + cap_net_limit_t *limit; + struct sockaddr_in ipv4; + + capnet = create_network_service(); + + /* Limit only to TEST_IPV4 on port 80 and 443. */ + limit = cap_net_limit_init(capnet, CAPNET_CONNECT); + ATF_REQUIRE(limit != NULL); + memset(&ipv4, 0, sizeof(ipv4)); + ipv4.sin_family = AF_INET; + ipv4.sin_port = htons(80); + inet_pton(AF_INET, TEST_IPV4, &ipv4.sin_addr); + cap_net_limit_connect(limit, (struct sockaddr *)&ipv4, sizeof(ipv4)); + + ipv4.sin_port = htons(443); + cap_net_limit_connect(limit, (struct sockaddr *)&ipv4, sizeof(ipv4)); + ATF_REQUIRE(cap_net_limit(limit) == 0); + + ATF_REQUIRE(test_connect(capnet, TEST_IPV4, 80) == 0); + ATF_REQUIRE(test_connect(capnet, "8.8.8.8", 80) == ENOTCAPABLE); + ATF_REQUIRE(test_connect(capnet, "8.8.8.8", 433) == ENOTCAPABLE); + ATF_REQUIRE(test_connect(capnet, TEST_IPV4, 443) == 0); + + /* Limit only to TEST_IPV4 on port 443. */ + limit = cap_net_limit_init(capnet, CAPNET_CONNECT); + cap_net_limit_connect(limit, (struct sockaddr *)&ipv4, sizeof(ipv4)); + ATF_REQUIRE(cap_net_limit(limit) == 0); + + ATF_REQUIRE(test_connect(capnet, TEST_IPV4, 433) == ENOTCAPABLE); + ATF_REQUIRE(test_connect(capnet, "8.8.8.8", 80) == ENOTCAPABLE); + ATF_REQUIRE(test_connect(capnet, "8.8.8.8", 433) == ENOTCAPABLE); + ATF_REQUIRE(test_connect(capnet, TEST_IPV4, 443) == 0); + + /* Unable to set empty limits. Empty limits means full access. */ + limit = cap_net_limit_init(capnet, CAPNET_CONNECT); + ATF_REQUIRE(cap_net_limit(limit) != 0); + + cap_close(capnet); +} + +ATF_TC_WITHOUT_HEAD(capnet__limits_connecttodns); +ATF_TC_BODY(capnet__limits_connecttodns, tc) +{ + cap_channel_t *capnet; + cap_net_limit_t *limit; + struct addrinfo hints, *capres, *res; + int family[] = { AF_INET }; + + capnet = create_network_service(); + + limit = cap_net_limit_init(capnet, CAPNET_CONNECTDNS | + CAPNET_NAME2ADDR); + ATF_REQUIRE(limit != NULL); + cap_net_limit_name2addr(limit, TEST_IPV4, "80"); + cap_net_limit_name2addr_family(limit, family, 1); + ATF_REQUIRE(cap_net_limit(limit) == 0); + + memset(&hints, 0, sizeof(hints)); + hints.ai_family = AF_INET; + hints.ai_socktype = SOCK_STREAM; + + ATF_REQUIRE(test_connect(capnet, "8.8.8.8", 433) == ENOTCAPABLE); + ATF_REQUIRE(cap_getaddrinfo(capnet, TEST_IPV4, "80", &hints, &capres) == + 0); + ATF_REQUIRE(test_connect(capnet, "8.8.8.8", 433) == ENOTCAPABLE); + + for (res = capres; res != NULL; res = res->ai_next) { + int s; + + ATF_REQUIRE(res->ai_family == AF_INET); + ATF_REQUIRE(res->ai_socktype == SOCK_STREAM); + + s = socket(res->ai_family, res->ai_socktype, res->ai_protocol); + ATF_REQUIRE(s >= 0); + + ATF_REQUIRE(cap_connect(capnet, s, res->ai_addr, + res->ai_addrlen) == 0); + close(s); + } + + freeaddrinfo(capres); + cap_close(capnet); +} + + +ATF_TC_WITHOUT_HEAD(capnet__limits_deprecated_connecttodns); +ATF_TC_BODY(capnet__limits_deprecated_connecttodns, tc) +{ + cap_channel_t *capnet; + cap_net_limit_t *limit; + struct hostent *caphp; + struct in_addr ipaddr; + struct sockaddr_in connaddr; + int family[] = { AF_INET }; + int i; + + capnet = create_network_service(); + + limit = cap_net_limit_init(capnet, CAPNET_CONNECTDNS | + CAPNET_DEPRECATED_NAME2ADDR); + ATF_REQUIRE(limit != NULL); + cap_net_limit_name2addr(limit, TEST_IPV4, NULL); + cap_net_limit_name2addr_family(limit, family, 1); + ATF_REQUIRE(cap_net_limit(limit) == 0); + + memset(&ipaddr, 0, sizeof(ipaddr)); + inet_pton(AF_INET, TEST_IPV4, &ipaddr); + + ATF_REQUIRE(test_connect(capnet, "8.8.8.8", 433) == ENOTCAPABLE); + caphp = cap_gethostbyname2(capnet, TEST_IPV4, AF_INET); + ATF_REQUIRE(caphp != NULL); + ATF_REQUIRE(caphp->h_addrtype == AF_INET); + ATF_REQUIRE(test_connect(capnet, "8.8.8.8", 433) == ENOTCAPABLE); + + for (i = 0; caphp->h_addr_list[i] != NULL; i++) { + int s; + + s = socket(AF_INET, SOCK_STREAM, 0); + ATF_REQUIRE(s >= 0); + + memset(&connaddr, 0, sizeof(connaddr)); + connaddr.sin_family = AF_INET; + memcpy((char *)&connaddr.sin_addr.s_addr, + (char *)caphp->h_addr_list[i], caphp->h_length); + connaddr.sin_port = htons(80); + + ATF_REQUIRE(cap_connect(capnet, s, (struct sockaddr *)&connaddr, + sizeof(connaddr)) == 0); + close(s); + } + + cap_close(capnet); +} + +ATF_TP_ADD_TCS(tp) +{ + + ATF_TP_ADD_TC(tp, capnet__limits_addr2name_mode); + ATF_TP_ADD_TC(tp, capnet__limits_addr2name_family); + ATF_TP_ADD_TC(tp, capnet__limits_addr2name); + + ATF_TP_ADD_TC(tp, capnet__limits_deprecated_addr2name_mode); + ATF_TP_ADD_TC(tp, capnet__limits_deprecated_addr2name_family); + ATF_TP_ADD_TC(tp, capnet__limits_deprecated_addr2name); + + ATF_TP_ADD_TC(tp, capnet__limits_name2addr_mode); + ATF_TP_ADD_TC(tp, capnet__limits_name2addr_hosts); + ATF_TP_ADD_TC(tp, capnet__limits_name2addr_hosts_servnames_strict); + ATF_TP_ADD_TC(tp, capnet__limits_name2addr_hosts_servnames_mix); + ATF_TP_ADD_TC(tp, capnet__limits_name2addr_family); + + ATF_TP_ADD_TC(tp, capnet__limits_deprecated_name2addr_mode); + ATF_TP_ADD_TC(tp, capnet__limits_deprecated_name2addr_hosts); + ATF_TP_ADD_TC(tp, capnet__limits_deprecated_name2addr_family); + + ATF_TP_ADD_TC(tp, capnet__limits_bind_mode); + ATF_TP_ADD_TC(tp, capnet__limits_bind); + + ATF_TP_ADD_TC(tp, capnet__limits_connect_mode); + ATF_TP_ADD_TC(tp, capnet__limits_connect); + + ATF_TP_ADD_TC(tp, capnet__limits_connecttodns); + ATF_TP_ADD_TC(tp, capnet__limits_deprecated_connecttodns); + + return (atf_no_error()); +} diff --git a/share/mk/src.libnames.mk b/share/mk/src.libnames.mk index c2c3d6c4110..2488527a5a6 100644 --- a/share/mk/src.libnames.mk +++ b/share/mk/src.libnames.mk @@ -89,6 +89,7 @@ _LIBRARIES= \ cap_dns \ cap_fileargs \ cap_grp \ + cap_net \ cap_pwd \ cap_sysctl \ cap_syslog \ @@ -637,6 +638,7 @@ LIBBSNMPDIR= ${OBJTOP}/lib/libbsnmp/libbsnmp LIBCASPERDIR= ${OBJTOP}/lib/libcasper/libcasper LIBCAP_DNSDIR= ${OBJTOP}/lib/libcasper/services/cap_dns LIBCAP_GRPDIR= ${OBJTOP}/lib/libcasper/services/cap_grp +LIBCAP_NETDIR= ${OBJTOP}/lib/libcasper/services/cap_net LIBCAP_PWDDIR= ${OBJTOP}/lib/libcasper/services/cap_pwd LIBCAP_SYSCTLDIR= ${OBJTOP}/lib/libcasper/services/cap_sysctl LIBCAP_SYSLOGDIR= ${OBJTOP}/lib/libcasper/services/cap_syslog From 8ab4becab09f0053f70a29a7bfb0f639e9519ec4 Mon Sep 17 00:00:00 2001 From: Mateusz Guzik Date: Sun, 16 Aug 2020 18:21:21 +0000 Subject: [PATCH 10/10] vfs: use namei_zone for getcwd allocations instead of malloc. Note that this should probably be wrapped with a dedicated API and other vn_getcwd callers did not get converted. --- sys/kern/vfs_cache.c | 6 ++---- 1 file changed, 2 insertions(+), 4 deletions(-) diff --git a/sys/kern/vfs_cache.c b/sys/kern/vfs_cache.c index bc2fc0c5812..91cafa2699f 100644 --- a/sys/kern/vfs_cache.c +++ b/sys/kern/vfs_cache.c @@ -2427,11 +2427,11 @@ sys___getcwd(struct thread *td, struct __getcwd_args *uap) if (buflen > MAXPATHLEN) buflen = MAXPATHLEN; - buf = malloc(buflen, M_TEMP, M_WAITOK); + buf = uma_zalloc(namei_zone, M_WAITOK); error = vn_getcwd(td, buf, &retbuf, &buflen); if (error == 0) error = copyout(retbuf, uap->buf, buflen); - free(buf, M_TEMP); + uma_zfree(namei_zone, buf); return (error); } @@ -2983,8 +2983,6 @@ DB_SHOW_COMMAND(vpath, db_show_vpath) #endif -extern uma_zone_t namei_zone; - static bool __read_frequently cache_fast_lookup = true; SYSCTL_BOOL(_vfs, OID_AUTO, cache_fast_lookup, CTLFLAG_RW, &cache_fast_lookup, 0, "");