Index: head/lib/libc/sys/ktrace.2 =================================================================== --- head/lib/libc/sys/ktrace.2 (revision 226268) +++ head/lib/libc/sys/ktrace.2 (revision 226269) @@ -1,201 +1,202 @@ .\" Copyright (c) 1993 .\" The Regents of the University of California. All rights reserved. .\" .\" Redistribution and use in source and binary forms, with or without .\" modification, are permitted provided that the following conditions .\" are met: .\" 1. Redistributions of source code must retain the above copyright .\" notice, this list of conditions and the following disclaimer. .\" 2. Redistributions in binary form must reproduce the above copyright .\" notice, this list of conditions and the following disclaimer in the .\" documentation and/or other materials provided with the distribution. .\" 4. Neither the name of the University nor the names of its contributors .\" may be used to endorse or promote products derived from this software .\" without specific prior written permission. .\" .\" THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND .\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE .\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE .\" ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE .\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL .\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS .\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) .\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT .\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY .\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF .\" SUCH DAMAGE. .\" .\" @(#)ktrace.2 8.1 (Berkeley) 6/4/93 .\" $FreeBSD$ .\" -.Dd October 9, 2011 +.Dd October 10, 2011 .Dt KTRACE 2 .Os .Sh NAME .Nm ktrace .Nd process tracing .Sh LIBRARY .Lb libc .Sh SYNOPSIS .In sys/param.h .In sys/time.h .In sys/uio.h .In sys/ktrace.h .Ft int .Fn ktrace "const char *tracefile" "int ops" "int trpoints" "int pid" .Sh DESCRIPTION The .Fn ktrace system call enables or disables tracing of one or more processes. Users may only trace their own processes. Only the super-user can trace setuid or setgid programs. .Pp The .Fa tracefile argument gives the pathname of the file to be used for tracing. The file must exist and be a regular file writable by the calling process. All trace records are always appended to the file, so the file must be truncated to zero length to discard previous trace data. If tracing points are being disabled (see KTROP_CLEAR below), .Fa tracefile may be NULL. .Pp The .Fa ops argument specifies the requested ktrace operation. The defined operations are: .Bl -column KTRFLAG_DESCENDXXX -offset indent -.It "KTROP_SET Enable trace points specified in" +.It "KTROP_SET Enable trace points specified in" .Fa trpoints . -.It "KTROP_CLEAR Disable trace points specified in +.It "KTROP_CLEAR Disable trace points specified in" .Fa trpoints . .It "KTROP_CLEARFILE Stop all tracing." .It "KTRFLAG_DESCEND The tracing change should apply to the" specified process and all its current children. .El .Pp The .Fa trpoints argument specifies the trace points of interest. The defined trace points are: .Bl -column KTRFAC_PROCCTORXXX -offset indent .It "KTRFAC_SYSCALL Trace system calls." .It "KTRFAC_SYSRET Trace return values from system calls." .It "KTRFAC_NAMEI Trace name lookup operations." .It "KTRFAC_GENIO Trace all I/O (note that this option can" generate much output). .It "KTRFAC_PSIG Trace posted signals." .It "KTRFAC_CSW Trace context switch points." .It "KTRFAC_USER Trace application-specific events." .It "KTRFAC_STRUCT Trace certain data structures." .It "KTRFAC_SYSCTL Trace sysctls." .It "KTRFAC_PROCCTOR Trace process construction." .It "KTRFAC_PROCDTOR Trace process destruction." +.It "KTRFAC_CAPFAIL Trace capability failures." .It "KTRFAC_INHERIT Inherit tracing to future children." .El .Pp Each tracing event outputs a record composed of a generic header followed by a trace point specific structure. The generic header is: .Bd -literal struct ktr_header { int ktr_len; /* length of buf */ short ktr_type; /* trace record type */ pid_t ktr_pid; /* process id */ char ktr_comm[MAXCOMLEN+1]; /* command name */ struct timeval ktr_time; /* timestamp */ intptr_t ktr_tid; /* was ktr_buffer */ }; .Ed .Pp The .Va ktr_len field specifies the length of the .Va ktr_type data that follows this header. The .Va ktr_pid and .Va ktr_comm fields specify the process and command generating the record. The .Va ktr_time field gives the time (with microsecond resolution) that the record was generated. The .Va ktr_tid field holds a threadid. .Pp The generic header is followed by .Va ktr_len bytes of a .Va ktr_type record. The type specific records are defined in the .In sys/ktrace.h include file. .Sh SYSCTL TUNABLES The following .Xr sysctl 8 tunables influence the behaviour of .Fn ktrace : .Bl -tag -width indent .It Va kern.ktrace.geniosize bounds the amount of data a traced I/O request will log to the trace file. .It Va kern.ktrace.request_pool bounds the number of trace events being logged at a time. .El .Pp Sysctl tunables that control process debuggability (as determined by .Xr p_candebug 9 ) also affect the operation of .Fn ktrace . .Sh RETURN VALUES .Rv -std ktrace .Sh ERRORS The .Fn ktrace system call will fail if: .Bl -tag -width Er .It Bq Er ENOTDIR A component of the path prefix is not a directory. .It Bq Er ENAMETOOLONG A component of a pathname exceeded 255 characters, or an entire path name exceeded 1023 characters. .It Bq Er ENOENT The named tracefile does not exist. .It Bq Er EACCES Search permission is denied for a component of the path prefix. .It Bq Er ELOOP Too many symbolic links were encountered in translating the pathname. .It Bq Er EIO An I/O error occurred while reading from or writing to the file system. .It Bq Er ENOSYS The kernel was not compiled with .Nm support. .El .Pp A thread may be unable to log one or more tracing events due to a temporary shortage of resources. This condition is remembered by the kernel, and the next tracing request that succeeds will have the flag .Li KTR_DROP set in its .Va ktr_type field. .Sh SEE ALSO .Xr kdump 1 , .Xr ktrace 1 , .Xr utrace 2 , .Xr sysctl 8 , .Xr p_candebug 9 .Sh HISTORY The .Fn ktrace system call first appeared in .Bx 4.4 . Index: head/sys/kern/kern_ktrace.c =================================================================== --- head/sys/kern/kern_ktrace.c (revision 226268) +++ head/sys/kern/kern_ktrace.c (revision 226269) @@ -1,1224 +1,1245 @@ /*- * Copyright (c) 1989, 1993 * The Regents of the University of California. * Copyright (c) 2005 Robert N. M. Watson * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * @(#)kern_ktrace.c 8.2 (Berkeley) 9/23/93 */ #include __FBSDID("$FreeBSD$"); #include "opt_ktrace.h" #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include /* * The ktrace facility allows the tracing of certain key events in user space * processes, such as system calls, signal delivery, context switches, and * user generated events using utrace(2). It works by streaming event * records and data to a vnode associated with the process using the * ktrace(2) system call. In general, records can be written directly from * the context that generates the event. One important exception to this is * during a context switch, where sleeping is not permitted. To handle this * case, trace events are generated using in-kernel ktr_request records, and * then delivered to disk at a convenient moment -- either immediately, the * next traceable event, at system call return, or at process exit. * * When dealing with multiple threads or processes writing to the same event * log, ordering guarantees are weak: specifically, if an event has multiple * records (i.e., system call enter and return), they may be interlaced with * records from another event. Process and thread ID information is provided * in the record, and user applications can de-interlace events if required. */ static MALLOC_DEFINE(M_KTRACE, "KTRACE", "KTRACE"); #ifdef KTRACE FEATURE(ktrace, "Kernel support for system-call tracing"); #ifndef KTRACE_REQUEST_POOL #define KTRACE_REQUEST_POOL 100 #endif struct ktr_request { struct ktr_header ktr_header; void *ktr_buffer; union { struct ktr_proc_ctor ktr_proc_ctor; + struct ktr_cap_fail ktr_cap_fail; struct ktr_syscall ktr_syscall; struct ktr_sysret ktr_sysret; struct ktr_genio ktr_genio; struct ktr_psig ktr_psig; struct ktr_csw ktr_csw; } ktr_data; STAILQ_ENTRY(ktr_request) ktr_list; }; static int data_lengths[] = { 0, /* none */ offsetof(struct ktr_syscall, ktr_args), /* KTR_SYSCALL */ sizeof(struct ktr_sysret), /* KTR_SYSRET */ 0, /* KTR_NAMEI */ sizeof(struct ktr_genio), /* KTR_GENIO */ sizeof(struct ktr_psig), /* KTR_PSIG */ sizeof(struct ktr_csw), /* KTR_CSW */ 0, /* KTR_USER */ 0, /* KTR_STRUCT */ 0, /* KTR_SYSCTL */ sizeof(struct ktr_proc_ctor), /* KTR_PROCCTOR */ 0, /* KTR_PROCDTOR */ + sizeof(struct ktr_cap_fail), /* KTR_CAPFAIL */ }; static STAILQ_HEAD(, ktr_request) ktr_free; static SYSCTL_NODE(_kern, OID_AUTO, ktrace, CTLFLAG_RD, 0, "KTRACE options"); static u_int ktr_requestpool = KTRACE_REQUEST_POOL; TUNABLE_INT("kern.ktrace.request_pool", &ktr_requestpool); static u_int ktr_geniosize = PAGE_SIZE; TUNABLE_INT("kern.ktrace.genio_size", &ktr_geniosize); SYSCTL_UINT(_kern_ktrace, OID_AUTO, genio_size, CTLFLAG_RW, &ktr_geniosize, 0, "Maximum size of genio event payload"); static int print_message = 1; static struct mtx ktrace_mtx; static struct sx ktrace_sx; static void ktrace_init(void *dummy); static int sysctl_kern_ktrace_request_pool(SYSCTL_HANDLER_ARGS); static u_int ktrace_resize_pool(u_int oldsize, u_int newsize); static struct ktr_request *ktr_getrequest_entered(struct thread *td, int type); static struct ktr_request *ktr_getrequest(int type); static void ktr_submitrequest(struct thread *td, struct ktr_request *req); static void ktr_freeproc(struct proc *p, struct ucred **uc, struct vnode **vp); static void ktr_freerequest(struct ktr_request *req); static void ktr_freerequest_locked(struct ktr_request *req); static void ktr_writerequest(struct thread *td, struct ktr_request *req); static int ktrcanset(struct thread *,struct proc *); static int ktrsetchildren(struct thread *,struct proc *,int,int,struct vnode *); static int ktrops(struct thread *,struct proc *,int,int,struct vnode *); static void ktrprocctor_entered(struct thread *, struct proc *); /* * ktrace itself generates events, such as context switches, which we do not * wish to trace. Maintain a flag, TDP_INKTRACE, on each thread to determine * whether or not it is in a region where tracing of events should be * suppressed. */ static void ktrace_enter(struct thread *td) { KASSERT(!(td->td_pflags & TDP_INKTRACE), ("ktrace_enter: flag set")); td->td_pflags |= TDP_INKTRACE; } static void ktrace_exit(struct thread *td) { KASSERT(td->td_pflags & TDP_INKTRACE, ("ktrace_exit: flag not set")); td->td_pflags &= ~TDP_INKTRACE; } static void ktrace_assert(struct thread *td) { KASSERT(td->td_pflags & TDP_INKTRACE, ("ktrace_assert: flag not set")); } static void ktrace_init(void *dummy) { struct ktr_request *req; int i; mtx_init(&ktrace_mtx, "ktrace", NULL, MTX_DEF | MTX_QUIET); sx_init(&ktrace_sx, "ktrace_sx"); STAILQ_INIT(&ktr_free); for (i = 0; i < ktr_requestpool; i++) { req = malloc(sizeof(struct ktr_request), M_KTRACE, M_WAITOK); STAILQ_INSERT_HEAD(&ktr_free, req, ktr_list); } } SYSINIT(ktrace_init, SI_SUB_KTRACE, SI_ORDER_ANY, ktrace_init, NULL); static int sysctl_kern_ktrace_request_pool(SYSCTL_HANDLER_ARGS) { struct thread *td; u_int newsize, oldsize, wantsize; int error; /* Handle easy read-only case first to avoid warnings from GCC. */ if (!req->newptr) { oldsize = ktr_requestpool; return (SYSCTL_OUT(req, &oldsize, sizeof(u_int))); } error = SYSCTL_IN(req, &wantsize, sizeof(u_int)); if (error) return (error); td = curthread; ktrace_enter(td); oldsize = ktr_requestpool; newsize = ktrace_resize_pool(oldsize, wantsize); ktrace_exit(td); error = SYSCTL_OUT(req, &oldsize, sizeof(u_int)); if (error) return (error); if (wantsize > oldsize && newsize < wantsize) return (ENOSPC); return (0); } SYSCTL_PROC(_kern_ktrace, OID_AUTO, request_pool, CTLTYPE_UINT|CTLFLAG_RW, &ktr_requestpool, 0, sysctl_kern_ktrace_request_pool, "IU", "Pool buffer size for ktrace(1)"); static u_int ktrace_resize_pool(u_int oldsize, u_int newsize) { STAILQ_HEAD(, ktr_request) ktr_new; struct ktr_request *req; int bound; print_message = 1; bound = newsize - oldsize; if (bound == 0) return (ktr_requestpool); if (bound < 0) { mtx_lock(&ktrace_mtx); /* Shrink pool down to newsize if possible. */ while (bound++ < 0) { req = STAILQ_FIRST(&ktr_free); if (req == NULL) break; STAILQ_REMOVE_HEAD(&ktr_free, ktr_list); ktr_requestpool--; free(req, M_KTRACE); } } else { /* Grow pool up to newsize. */ STAILQ_INIT(&ktr_new); while (bound-- > 0) { req = malloc(sizeof(struct ktr_request), M_KTRACE, M_WAITOK); STAILQ_INSERT_HEAD(&ktr_new, req, ktr_list); } mtx_lock(&ktrace_mtx); STAILQ_CONCAT(&ktr_free, &ktr_new); ktr_requestpool += (newsize - oldsize); } mtx_unlock(&ktrace_mtx); return (ktr_requestpool); } /* ktr_getrequest() assumes that ktr_comm[] is the same size as td_name[]. */ CTASSERT(sizeof(((struct ktr_header *)NULL)->ktr_comm) == (sizeof((struct thread *)NULL)->td_name)); static struct ktr_request * ktr_getrequest_entered(struct thread *td, int type) { struct ktr_request *req; struct proc *p = td->td_proc; int pm; mtx_lock(&ktrace_mtx); if (!KTRCHECK(td, type)) { mtx_unlock(&ktrace_mtx); return (NULL); } req = STAILQ_FIRST(&ktr_free); if (req != NULL) { STAILQ_REMOVE_HEAD(&ktr_free, ktr_list); req->ktr_header.ktr_type = type; if (p->p_traceflag & KTRFAC_DROP) { req->ktr_header.ktr_type |= KTR_DROP; p->p_traceflag &= ~KTRFAC_DROP; } mtx_unlock(&ktrace_mtx); microtime(&req->ktr_header.ktr_time); req->ktr_header.ktr_pid = p->p_pid; req->ktr_header.ktr_tid = td->td_tid; bcopy(td->td_name, req->ktr_header.ktr_comm, sizeof(req->ktr_header.ktr_comm)); req->ktr_buffer = NULL; req->ktr_header.ktr_len = 0; } else { p->p_traceflag |= KTRFAC_DROP; pm = print_message; print_message = 0; mtx_unlock(&ktrace_mtx); if (pm) printf("Out of ktrace request objects.\n"); } return (req); } static struct ktr_request * ktr_getrequest(int type) { struct thread *td = curthread; struct ktr_request *req; ktrace_enter(td); req = ktr_getrequest_entered(td, type); if (req == NULL) ktrace_exit(td); return (req); } /* * Some trace generation environments don't permit direct access to VFS, * such as during a context switch where sleeping is not allowed. Under these * circumstances, queue a request to the thread to be written asynchronously * later. */ static void ktr_enqueuerequest(struct thread *td, struct ktr_request *req) { mtx_lock(&ktrace_mtx); STAILQ_INSERT_TAIL(&td->td_proc->p_ktr, req, ktr_list); mtx_unlock(&ktrace_mtx); } /* * Drain any pending ktrace records from the per-thread queue to disk. This * is used both internally before committing other records, and also on * system call return. We drain all the ones we can find at the time when * drain is requested, but don't keep draining after that as those events * may be approximately "after" the current event. */ static void ktr_drain(struct thread *td) { struct ktr_request *queued_req; STAILQ_HEAD(, ktr_request) local_queue; ktrace_assert(td); sx_assert(&ktrace_sx, SX_XLOCKED); STAILQ_INIT(&local_queue); if (!STAILQ_EMPTY(&td->td_proc->p_ktr)) { mtx_lock(&ktrace_mtx); STAILQ_CONCAT(&local_queue, &td->td_proc->p_ktr); mtx_unlock(&ktrace_mtx); while ((queued_req = STAILQ_FIRST(&local_queue))) { STAILQ_REMOVE_HEAD(&local_queue, ktr_list); ktr_writerequest(td, queued_req); ktr_freerequest(queued_req); } } } /* * Submit a trace record for immediate commit to disk -- to be used only * where entering VFS is OK. First drain any pending records that may have * been cached in the thread. */ static void ktr_submitrequest(struct thread *td, struct ktr_request *req) { ktrace_assert(td); sx_xlock(&ktrace_sx); ktr_drain(td); ktr_writerequest(td, req); ktr_freerequest(req); sx_xunlock(&ktrace_sx); ktrace_exit(td); } static void ktr_freerequest(struct ktr_request *req) { mtx_lock(&ktrace_mtx); ktr_freerequest_locked(req); mtx_unlock(&ktrace_mtx); } static void ktr_freerequest_locked(struct ktr_request *req) { mtx_assert(&ktrace_mtx, MA_OWNED); if (req->ktr_buffer != NULL) free(req->ktr_buffer, M_KTRACE); STAILQ_INSERT_HEAD(&ktr_free, req, ktr_list); } /* * Disable tracing for a process and release all associated resources. * The caller is responsible for releasing a reference on the returned * vnode and credentials. */ static void ktr_freeproc(struct proc *p, struct ucred **uc, struct vnode **vp) { struct ktr_request *req; PROC_LOCK_ASSERT(p, MA_OWNED); mtx_assert(&ktrace_mtx, MA_OWNED); *uc = p->p_tracecred; p->p_tracecred = NULL; if (vp != NULL) *vp = p->p_tracevp; p->p_tracevp = NULL; p->p_traceflag = 0; while ((req = STAILQ_FIRST(&p->p_ktr)) != NULL) { STAILQ_REMOVE_HEAD(&p->p_ktr, ktr_list); ktr_freerequest_locked(req); } } void ktrsyscall(code, narg, args) int code, narg; register_t args[]; { struct ktr_request *req; struct ktr_syscall *ktp; size_t buflen; char *buf = NULL; buflen = sizeof(register_t) * narg; if (buflen > 0) { buf = malloc(buflen, M_KTRACE, M_WAITOK); bcopy(args, buf, buflen); } req = ktr_getrequest(KTR_SYSCALL); if (req == NULL) { if (buf != NULL) free(buf, M_KTRACE); return; } ktp = &req->ktr_data.ktr_syscall; ktp->ktr_code = code; ktp->ktr_narg = narg; if (buflen > 0) { req->ktr_header.ktr_len = buflen; req->ktr_buffer = buf; } ktr_submitrequest(curthread, req); } void ktrsysret(code, error, retval) int code, error; register_t retval; { struct ktr_request *req; struct ktr_sysret *ktp; req = ktr_getrequest(KTR_SYSRET); if (req == NULL) return; ktp = &req->ktr_data.ktr_sysret; ktp->ktr_code = code; ktp->ktr_error = error; ktp->ktr_retval = retval; /* what about val2 ? */ ktr_submitrequest(curthread, req); } /* * When a setuid process execs, disable tracing. * * XXX: We toss any pending asynchronous records. */ void ktrprocexec(struct proc *p, struct ucred **uc, struct vnode **vp) { PROC_LOCK_ASSERT(p, MA_OWNED); mtx_lock(&ktrace_mtx); ktr_freeproc(p, uc, vp); mtx_unlock(&ktrace_mtx); } /* * When a process exits, drain per-process asynchronous trace records * and disable tracing. */ void ktrprocexit(struct thread *td) { struct ktr_request *req; struct proc *p; struct ucred *cred; struct vnode *vp; int vfslocked; p = td->td_proc; if (p->p_traceflag == 0) return; ktrace_enter(td); req = ktr_getrequest_entered(td, KTR_PROCDTOR); if (req != NULL) ktr_enqueuerequest(td, req); sx_xlock(&ktrace_sx); ktr_drain(td); sx_xunlock(&ktrace_sx); PROC_LOCK(p); mtx_lock(&ktrace_mtx); ktr_freeproc(p, &cred, &vp); mtx_unlock(&ktrace_mtx); PROC_UNLOCK(p); if (vp != NULL) { vfslocked = VFS_LOCK_GIANT(vp->v_mount); vrele(vp); VFS_UNLOCK_GIANT(vfslocked); } if (cred != NULL) crfree(cred); ktrace_exit(td); } static void ktrprocctor_entered(struct thread *td, struct proc *p) { struct ktr_proc_ctor *ktp; struct ktr_request *req; struct thread *td2; ktrace_assert(td); td2 = FIRST_THREAD_IN_PROC(p); req = ktr_getrequest_entered(td2, KTR_PROCCTOR); if (req == NULL) return; ktp = &req->ktr_data.ktr_proc_ctor; ktp->sv_flags = p->p_sysent->sv_flags; ktr_enqueuerequest(td2, req); } void ktrprocctor(struct proc *p) { struct thread *td = curthread; if ((p->p_traceflag & KTRFAC_MASK) == 0) return; ktrace_enter(td); ktrprocctor_entered(td, p); ktrace_exit(td); } /* * When a process forks, enable tracing in the new process if needed. */ void ktrprocfork(struct proc *p1, struct proc *p2) { PROC_LOCK(p1); mtx_lock(&ktrace_mtx); KASSERT(p2->p_tracevp == NULL, ("new process has a ktrace vnode")); if (p1->p_traceflag & KTRFAC_INHERIT) { p2->p_traceflag = p1->p_traceflag; if ((p2->p_tracevp = p1->p_tracevp) != NULL) { VREF(p2->p_tracevp); KASSERT(p1->p_tracecred != NULL, ("ktrace vnode with no cred")); p2->p_tracecred = crhold(p1->p_tracecred); } } mtx_unlock(&ktrace_mtx); PROC_UNLOCK(p1); ktrprocctor(p2); } /* * When a thread returns, drain any asynchronous records generated by the * system call. */ void ktruserret(struct thread *td) { ktrace_enter(td); sx_xlock(&ktrace_sx); ktr_drain(td); sx_xunlock(&ktrace_sx); ktrace_exit(td); } void ktrnamei(path) char *path; { struct ktr_request *req; int namelen; char *buf = NULL; namelen = strlen(path); if (namelen > 0) { buf = malloc(namelen, M_KTRACE, M_WAITOK); bcopy(path, buf, namelen); } req = ktr_getrequest(KTR_NAMEI); if (req == NULL) { if (buf != NULL) free(buf, M_KTRACE); return; } if (namelen > 0) { req->ktr_header.ktr_len = namelen; req->ktr_buffer = buf; } ktr_submitrequest(curthread, req); } void ktrsysctl(name, namelen) int *name; u_int namelen; { struct ktr_request *req; u_int mib[CTL_MAXNAME + 2]; char *mibname; size_t mibnamelen; int error; /* Lookup name of mib. */ KASSERT(namelen <= CTL_MAXNAME, ("sysctl MIB too long")); mib[0] = 0; mib[1] = 1; bcopy(name, mib + 2, namelen * sizeof(*name)); mibnamelen = 128; mibname = malloc(mibnamelen, M_KTRACE, M_WAITOK); error = kernel_sysctl(curthread, mib, namelen + 2, mibname, &mibnamelen, NULL, 0, &mibnamelen, 0); if (error) { free(mibname, M_KTRACE); return; } req = ktr_getrequest(KTR_SYSCTL); if (req == NULL) { free(mibname, M_KTRACE); return; } req->ktr_header.ktr_len = mibnamelen; req->ktr_buffer = mibname; ktr_submitrequest(curthread, req); } void ktrgenio(fd, rw, uio, error) int fd; enum uio_rw rw; struct uio *uio; int error; { struct ktr_request *req; struct ktr_genio *ktg; int datalen; char *buf; if (error) { free(uio, M_IOV); return; } uio->uio_offset = 0; uio->uio_rw = UIO_WRITE; datalen = imin(uio->uio_resid, ktr_geniosize); buf = malloc(datalen, M_KTRACE, M_WAITOK); error = uiomove(buf, datalen, uio); free(uio, M_IOV); if (error) { free(buf, M_KTRACE); return; } req = ktr_getrequest(KTR_GENIO); if (req == NULL) { free(buf, M_KTRACE); return; } ktg = &req->ktr_data.ktr_genio; ktg->ktr_fd = fd; ktg->ktr_rw = rw; req->ktr_header.ktr_len = datalen; req->ktr_buffer = buf; ktr_submitrequest(curthread, req); } void ktrpsig(sig, action, mask, code) int sig; sig_t action; sigset_t *mask; int code; { struct thread *td = curthread; struct ktr_request *req; struct ktr_psig *kp; req = ktr_getrequest(KTR_PSIG); if (req == NULL) return; kp = &req->ktr_data.ktr_psig; kp->signo = (char)sig; kp->action = action; kp->mask = *mask; kp->code = code; ktr_enqueuerequest(td, req); ktrace_exit(td); } void ktrcsw(out, user) int out, user; { struct thread *td = curthread; struct ktr_request *req; struct ktr_csw *kc; req = ktr_getrequest(KTR_CSW); if (req == NULL) return; kc = &req->ktr_data.ktr_csw; kc->out = out; kc->user = user; ktr_enqueuerequest(td, req); ktrace_exit(td); } void ktrstruct(name, data, datalen) const char *name; void *data; size_t datalen; { struct ktr_request *req; char *buf = NULL; size_t buflen; if (!data) datalen = 0; buflen = strlen(name) + 1 + datalen; buf = malloc(buflen, M_KTRACE, M_WAITOK); strcpy(buf, name); bcopy(data, buf + strlen(name) + 1, datalen); if ((req = ktr_getrequest(KTR_STRUCT)) == NULL) { free(buf, M_KTRACE); return; } req->ktr_buffer = buf; req->ktr_header.ktr_len = buflen; ktr_submitrequest(curthread, req); +} + +void +ktrcapfail(needed, held) + cap_rights_t needed; + cap_rights_t held; +{ + struct thread *td = curthread; + struct ktr_request *req; + struct ktr_cap_fail *kcf; + + req = ktr_getrequest(KTR_CAPFAIL); + if (req == NULL) + return; + kcf = &req->ktr_data.ktr_cap_fail; + kcf->cap_needed = needed; + kcf->cap_held = held; + ktr_enqueuerequest(td, req); + ktrace_exit(td); } #endif /* KTRACE */ /* Interface and common routines */ #ifndef _SYS_SYSPROTO_H_ struct ktrace_args { char *fname; int ops; int facs; int pid; }; #endif /* ARGSUSED */ int sys_ktrace(td, uap) struct thread *td; register struct ktrace_args *uap; { #ifdef KTRACE register struct vnode *vp = NULL; register struct proc *p; struct pgrp *pg; int facs = uap->facs & ~KTRFAC_ROOT; int ops = KTROP(uap->ops); int descend = uap->ops & KTRFLAG_DESCEND; int nfound, ret = 0; int flags, error = 0, vfslocked; struct nameidata nd; struct ucred *cred; /* * Need something to (un)trace. */ if (ops != KTROP_CLEARFILE && facs == 0) return (EINVAL); ktrace_enter(td); if (ops != KTROP_CLEAR) { /* * an operation which requires a file argument. */ NDINIT(&nd, LOOKUP, NOFOLLOW | MPSAFE, UIO_USERSPACE, uap->fname, td); flags = FREAD | FWRITE | O_NOFOLLOW; error = vn_open(&nd, &flags, 0, NULL); if (error) { ktrace_exit(td); return (error); } vfslocked = NDHASGIANT(&nd); NDFREE(&nd, NDF_ONLY_PNBUF); vp = nd.ni_vp; VOP_UNLOCK(vp, 0); if (vp->v_type != VREG) { (void) vn_close(vp, FREAD|FWRITE, td->td_ucred, td); VFS_UNLOCK_GIANT(vfslocked); ktrace_exit(td); return (EACCES); } VFS_UNLOCK_GIANT(vfslocked); } /* * Clear all uses of the tracefile. */ if (ops == KTROP_CLEARFILE) { int vrele_count; vrele_count = 0; sx_slock(&allproc_lock); FOREACH_PROC_IN_SYSTEM(p) { PROC_LOCK(p); if (p->p_tracevp == vp) { if (ktrcanset(td, p)) { mtx_lock(&ktrace_mtx); ktr_freeproc(p, &cred, NULL); mtx_unlock(&ktrace_mtx); vrele_count++; crfree(cred); } else error = EPERM; } PROC_UNLOCK(p); } sx_sunlock(&allproc_lock); if (vrele_count > 0) { vfslocked = VFS_LOCK_GIANT(vp->v_mount); while (vrele_count-- > 0) vrele(vp); VFS_UNLOCK_GIANT(vfslocked); } goto done; } /* * do it */ sx_slock(&proctree_lock); if (uap->pid < 0) { /* * by process group */ pg = pgfind(-uap->pid); if (pg == NULL) { sx_sunlock(&proctree_lock); error = ESRCH; goto done; } /* * ktrops() may call vrele(). Lock pg_members * by the proctree_lock rather than pg_mtx. */ PGRP_UNLOCK(pg); nfound = 0; LIST_FOREACH(p, &pg->pg_members, p_pglist) { PROC_LOCK(p); if (p->p_state == PRS_NEW || p_cansee(td, p) != 0) { PROC_UNLOCK(p); continue; } nfound++; if (descend) ret |= ktrsetchildren(td, p, ops, facs, vp); else ret |= ktrops(td, p, ops, facs, vp); } if (nfound == 0) { sx_sunlock(&proctree_lock); error = ESRCH; goto done; } } else { /* * by pid */ p = pfind(uap->pid); if (p == NULL) error = ESRCH; else error = p_cansee(td, p); if (error) { if (p != NULL) PROC_UNLOCK(p); sx_sunlock(&proctree_lock); goto done; } if (descend) ret |= ktrsetchildren(td, p, ops, facs, vp); else ret |= ktrops(td, p, ops, facs, vp); } sx_sunlock(&proctree_lock); if (!ret) error = EPERM; done: if (vp != NULL) { vfslocked = VFS_LOCK_GIANT(vp->v_mount); (void) vn_close(vp, FWRITE, td->td_ucred, td); VFS_UNLOCK_GIANT(vfslocked); } ktrace_exit(td); return (error); #else /* !KTRACE */ return (ENOSYS); #endif /* KTRACE */ } /* ARGSUSED */ int sys_utrace(td, uap) struct thread *td; register struct utrace_args *uap; { #ifdef KTRACE struct ktr_request *req; void *cp; int error; if (!KTRPOINT(td, KTR_USER)) return (0); if (uap->len > KTR_USER_MAXLEN) return (EINVAL); cp = malloc(uap->len, M_KTRACE, M_WAITOK); error = copyin(uap->addr, cp, uap->len); if (error) { free(cp, M_KTRACE); return (error); } req = ktr_getrequest(KTR_USER); if (req == NULL) { free(cp, M_KTRACE); return (ENOMEM); } req->ktr_buffer = cp; req->ktr_header.ktr_len = uap->len; ktr_submitrequest(td, req); return (0); #else /* !KTRACE */ return (ENOSYS); #endif /* KTRACE */ } #ifdef KTRACE static int ktrops(td, p, ops, facs, vp) struct thread *td; struct proc *p; int ops, facs; struct vnode *vp; { struct vnode *tracevp = NULL; struct ucred *tracecred = NULL; PROC_LOCK_ASSERT(p, MA_OWNED); if (!ktrcanset(td, p)) { PROC_UNLOCK(p); return (0); } if (p->p_flag & P_WEXIT) { /* If the process is exiting, just ignore it. */ PROC_UNLOCK(p); return (1); } mtx_lock(&ktrace_mtx); if (ops == KTROP_SET) { if (p->p_tracevp != vp) { /* * if trace file already in use, relinquish below */ tracevp = p->p_tracevp; VREF(vp); p->p_tracevp = vp; } if (p->p_tracecred != td->td_ucred) { tracecred = p->p_tracecred; p->p_tracecred = crhold(td->td_ucred); } p->p_traceflag |= facs; if (priv_check(td, PRIV_KTRACE) == 0) p->p_traceflag |= KTRFAC_ROOT; } else { /* KTROP_CLEAR */ if (((p->p_traceflag &= ~facs) & KTRFAC_MASK) == 0) /* no more tracing */ ktr_freeproc(p, &tracecred, &tracevp); } mtx_unlock(&ktrace_mtx); if ((p->p_traceflag & KTRFAC_MASK) != 0) ktrprocctor_entered(td, p); PROC_UNLOCK(p); if (tracevp != NULL) { int vfslocked; vfslocked = VFS_LOCK_GIANT(tracevp->v_mount); vrele(tracevp); VFS_UNLOCK_GIANT(vfslocked); } if (tracecred != NULL) crfree(tracecred); return (1); } static int ktrsetchildren(td, top, ops, facs, vp) struct thread *td; struct proc *top; int ops, facs; struct vnode *vp; { register struct proc *p; register int ret = 0; p = top; PROC_LOCK_ASSERT(p, MA_OWNED); sx_assert(&proctree_lock, SX_LOCKED); for (;;) { ret |= ktrops(td, p, ops, facs, vp); /* * If this process has children, descend to them next, * otherwise do any siblings, and if done with this level, * follow back up the tree (but not past top). */ if (!LIST_EMPTY(&p->p_children)) p = LIST_FIRST(&p->p_children); else for (;;) { if (p == top) return (ret); if (LIST_NEXT(p, p_sibling)) { p = LIST_NEXT(p, p_sibling); break; } p = p->p_pptr; } PROC_LOCK(p); } /*NOTREACHED*/ } static void ktr_writerequest(struct thread *td, struct ktr_request *req) { struct ktr_header *kth; struct vnode *vp; struct proc *p; struct ucred *cred; struct uio auio; struct iovec aiov[3]; struct mount *mp; int datalen, buflen, vrele_count; int error, vfslocked; /* * We hold the vnode and credential for use in I/O in case ktrace is * disabled on the process as we write out the request. * * XXXRW: This is not ideal: we could end up performing a write after * the vnode has been closed. */ mtx_lock(&ktrace_mtx); vp = td->td_proc->p_tracevp; cred = td->td_proc->p_tracecred; /* * If vp is NULL, the vp has been cleared out from under this * request, so just drop it. Make sure the credential and vnode are * in sync: we should have both or neither. */ if (vp == NULL) { KASSERT(cred == NULL, ("ktr_writerequest: cred != NULL")); mtx_unlock(&ktrace_mtx); return; } VREF(vp); KASSERT(cred != NULL, ("ktr_writerequest: cred == NULL")); crhold(cred); mtx_unlock(&ktrace_mtx); kth = &req->ktr_header; KASSERT(((u_short)kth->ktr_type & ~KTR_DROP) < sizeof(data_lengths) / sizeof(data_lengths[0]), ("data_lengths array overflow")); datalen = data_lengths[(u_short)kth->ktr_type & ~KTR_DROP]; buflen = kth->ktr_len; auio.uio_iov = &aiov[0]; auio.uio_offset = 0; auio.uio_segflg = UIO_SYSSPACE; auio.uio_rw = UIO_WRITE; aiov[0].iov_base = (caddr_t)kth; aiov[0].iov_len = sizeof(struct ktr_header); auio.uio_resid = sizeof(struct ktr_header); auio.uio_iovcnt = 1; auio.uio_td = td; if (datalen != 0) { aiov[1].iov_base = (caddr_t)&req->ktr_data; aiov[1].iov_len = datalen; auio.uio_resid += datalen; auio.uio_iovcnt++; kth->ktr_len += datalen; } if (buflen != 0) { KASSERT(req->ktr_buffer != NULL, ("ktrace: nothing to write")); aiov[auio.uio_iovcnt].iov_base = req->ktr_buffer; aiov[auio.uio_iovcnt].iov_len = buflen; auio.uio_resid += buflen; auio.uio_iovcnt++; } vfslocked = VFS_LOCK_GIANT(vp->v_mount); vn_start_write(vp, &mp, V_WAIT); vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); #ifdef MAC error = mac_vnode_check_write(cred, NOCRED, vp); if (error == 0) #endif error = VOP_WRITE(vp, &auio, IO_UNIT | IO_APPEND, cred); VOP_UNLOCK(vp, 0); vn_finished_write(mp); crfree(cred); if (!error) { vrele(vp); VFS_UNLOCK_GIANT(vfslocked); return; } VFS_UNLOCK_GIANT(vfslocked); /* * If error encountered, give up tracing on this vnode. We defer * all the vrele()'s on the vnode until after we are finished walking * the various lists to avoid needlessly holding locks. * NB: at this point we still hold the vnode reference that must * not go away as we need the valid vnode to compare with. Thus let * vrele_count start at 1 and the reference will be freed * by the loop at the end after our last use of vp. */ log(LOG_NOTICE, "ktrace write failed, errno %d, tracing stopped\n", error); vrele_count = 1; /* * First, clear this vnode from being used by any processes in the * system. * XXX - If one process gets an EPERM writing to the vnode, should * we really do this? Other processes might have suitable * credentials for the operation. */ cred = NULL; sx_slock(&allproc_lock); FOREACH_PROC_IN_SYSTEM(p) { PROC_LOCK(p); if (p->p_tracevp == vp) { mtx_lock(&ktrace_mtx); ktr_freeproc(p, &cred, NULL); mtx_unlock(&ktrace_mtx); vrele_count++; } PROC_UNLOCK(p); if (cred != NULL) { crfree(cred); cred = NULL; } } sx_sunlock(&allproc_lock); vfslocked = VFS_LOCK_GIANT(vp->v_mount); while (vrele_count-- > 0) vrele(vp); VFS_UNLOCK_GIANT(vfslocked); } /* * Return true if caller has permission to set the ktracing state * of target. Essentially, the target can't possess any * more permissions than the caller. KTRFAC_ROOT signifies that * root previously set the tracing status on the target process, and * so, only root may further change it. */ static int ktrcanset(td, targetp) struct thread *td; struct proc *targetp; { PROC_LOCK_ASSERT(targetp, MA_OWNED); if (targetp->p_traceflag & KTRFAC_ROOT && priv_check(td, PRIV_KTRACE)) return (0); if (p_candebug(td, targetp) != 0) return (0); return (1); } #endif /* KTRACE */ Index: head/sys/kern/sys_capability.c =================================================================== --- head/sys/kern/sys_capability.c (revision 226268) +++ head/sys/kern/sys_capability.c (revision 226269) @@ -1,553 +1,561 @@ /*- * Copyright (c) 2008-2011 Robert N. M. Watson * Copyright (c) 2010-2011 Jonathan Anderson * All rights reserved. * * This software was developed at the University of Cambridge Computer * Laboratory with support from a grant from Google, Inc. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ /* * FreeBSD kernel capability facility. * * Two kernel features are implemented here: capability mode, a sandboxed mode * of execution for processes, and capabilities, a refinement on file * descriptors that allows fine-grained control over operations on the file * descriptor. Collectively, these allow processes to run in the style of a * historic "capability system" in which they can use only resources * explicitly delegated to them. This model is enforced by restricting access * to global namespaces in capability mode. * * Capabilities wrap other file descriptor types, binding them to a constant * rights mask set when the capability is created. New capabilities may be * derived from existing capabilities, but only if they have the same or a * strict subset of the rights on the original capability. * * System calls permitted in capability mode are defined in capabilities.conf; * calls must be carefully audited for safety to ensure that they don't allow * escape from a sandbox. Some calls permit only a subset of operations in * capability mode -- for example, shm_open(2) is limited to creating * anonymous, rather than named, POSIX shared memory objects. */ #include "opt_capsicum.h" +#include "opt_ktrace.h" #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include +#include +#include #include #include #include #ifdef CAPABILITY_MODE FEATURE(security_capability_mode, "Capsicum Capability Mode"); /* * System call to enter capability mode for the process. */ int sys_cap_enter(struct thread *td, struct cap_enter_args *uap) { struct ucred *newcred, *oldcred; struct proc *p; if (IN_CAPABILITY_MODE(td)) return (0); newcred = crget(); p = td->td_proc; PROC_LOCK(p); oldcred = p->p_ucred; crcopy(newcred, oldcred); newcred->cr_flags |= CRED_FLAG_CAPMODE; p->p_ucred = newcred; PROC_UNLOCK(p); crfree(oldcred); return (0); } /* * System call to query whether the process is in capability mode. */ int sys_cap_getmode(struct thread *td, struct cap_getmode_args *uap) { u_int i; i = (IN_CAPABILITY_MODE(td)) ? 1 : 0; return (copyout(&i, uap->modep, sizeof(i))); } #else /* !CAPABILITY_MODE */ int sys_cap_enter(struct thread *td, struct cap_enter_args *uap) { return (ENOSYS); } int sys_cap_getmode(struct thread *td, struct cap_getmode_args *uap) { return (ENOSYS); } #endif /* CAPABILITY_MODE */ #ifdef CAPABILITIES FEATURE(security_capabilities, "Capsicum Capabilities"); /* * struct capability describes a capability, and is hung off of its struct * file f_data field. cap_file and cap_rightss are static once hooked up, as * neither the object it references nor the rights it encapsulates are * permitted to change. */ struct capability { struct file *cap_object; /* Underlying object's file. */ struct file *cap_file; /* Back-pointer to cap's file. */ cap_rights_t cap_rights; /* Mask of rights on object. */ }; /* * Capabilities have a fileops vector, but in practice none should ever be * called except for fo_close, as the capability will normally not be * returned during a file descriptor lookup in the system call code. */ static fo_rdwr_t capability_read; static fo_rdwr_t capability_write; static fo_truncate_t capability_truncate; static fo_ioctl_t capability_ioctl; static fo_poll_t capability_poll; static fo_kqfilter_t capability_kqfilter; static fo_stat_t capability_stat; static fo_close_t capability_close; static fo_chmod_t capability_chmod; static fo_chown_t capability_chown; static struct fileops capability_ops = { .fo_read = capability_read, .fo_write = capability_write, .fo_truncate = capability_truncate, .fo_ioctl = capability_ioctl, .fo_poll = capability_poll, .fo_kqfilter = capability_kqfilter, .fo_stat = capability_stat, .fo_close = capability_close, .fo_chmod = capability_chmod, .fo_chown = capability_chown, .fo_flags = DFLAG_PASSABLE, }; static struct fileops capability_ops_unpassable = { .fo_read = capability_read, .fo_write = capability_write, .fo_truncate = capability_truncate, .fo_ioctl = capability_ioctl, .fo_poll = capability_poll, .fo_kqfilter = capability_kqfilter, .fo_stat = capability_stat, .fo_close = capability_close, .fo_chmod = capability_chmod, .fo_chown = capability_chown, .fo_flags = 0, }; static uma_zone_t capability_zone; static void capability_init(void *dummy __unused) { capability_zone = uma_zcreate("capability", sizeof(struct capability), NULL, NULL, NULL, NULL, UMA_ALIGN_PTR, 0); if (capability_zone == NULL) panic("capability_init: capability_zone not initialized"); } SYSINIT(vfs, SI_SUB_VFS, SI_ORDER_ANY, capability_init, NULL); /* * Test whether a capability grants the requested rights. */ static int cap_check(struct capability *c, cap_rights_t rights) { - if ((c->cap_rights | rights) != c->cap_rights) + if ((c->cap_rights | rights) != c->cap_rights) { +#ifdef KTRACE + if (KTRPOINT(curthread, KTR_CAPFAIL)) + ktrcapfail(rights, c->cap_rights); +#endif return (ENOTCAPABLE); + } return (0); } /* * Extract rights from a capability for monitoring purposes -- not for use in * any other way, as we want to keep all capability permission evaluation in * this one file. */ cap_rights_t cap_rights(struct file *fp_cap) { struct capability *c; KASSERT(fp_cap->f_type == DTYPE_CAPABILITY, ("cap_rights: !capability")); c = fp_cap->f_data; return (c->cap_rights); } /* * System call to create a new capability reference to either an existing * file object or an an existing capability. */ int sys_cap_new(struct thread *td, struct cap_new_args *uap) { int error, capfd; int fd = uap->fd; struct file *fp; cap_rights_t rights = uap->rights; AUDIT_ARG_FD(fd); AUDIT_ARG_RIGHTS(rights); error = fget(td, fd, rights, &fp); if (error) return (error); AUDIT_ARG_FILE(td->td_proc, fp); error = kern_capwrap(td, fp, rights, &capfd); if (error) return (error); /* * Release our reference to the file (kern_capwrap has held a reference * for the filedesc array). */ fdrop(fp, td); td->td_retval[0] = capfd; return (0); } /* * System call to query the rights mask associated with a capability. */ int sys_cap_getrights(struct thread *td, struct cap_getrights_args *uap) { struct capability *cp; struct file *fp; int error; AUDIT_ARG_FD(uap->fd); error = fgetcap(td, uap->fd, &fp); if (error) return (error); cp = fp->f_data; error = copyout(&cp->cap_rights, uap->rightsp, sizeof(*uap->rightsp)); fdrop(fp, td); return (error); } /* * Create a capability to wrap around an existing file. */ int kern_capwrap(struct thread *td, struct file *fp, cap_rights_t rights, int *capfdp) { struct capability *cp, *cp_old; struct file *fp_object, *fcapp; int error; if ((rights | CAP_MASK_VALID) != CAP_MASK_VALID) return (EINVAL); /* * If a new capability is being derived from an existing capability, * then the new capability rights must be a subset of the existing * rights. */ if (fp->f_type == DTYPE_CAPABILITY) { cp_old = fp->f_data; if ((cp_old->cap_rights | rights) != cp_old->cap_rights) return (ENOTCAPABLE); } /* * Allocate a new file descriptor to hang the capability off of. */ error = falloc(td, &fcapp, capfdp, fp->f_flag); if (error) return (error); /* * Rather than nesting capabilities, directly reference the object an * existing capability references. There's nothing else interesting * to preserve for future use, as we've incorporated the previous * rights mask into the new one. This prevents us from having to * deal with capability chains. */ if (fp->f_type == DTYPE_CAPABILITY) fp_object = ((struct capability *)fp->f_data)->cap_object; else fp_object = fp; fhold(fp_object); cp = uma_zalloc(capability_zone, M_WAITOK | M_ZERO); cp->cap_rights = rights; cp->cap_object = fp_object; cp->cap_file = fcapp; if (fp->f_flag & DFLAG_PASSABLE) finit(fcapp, fp->f_flag, DTYPE_CAPABILITY, cp, &capability_ops); else finit(fcapp, fp->f_flag, DTYPE_CAPABILITY, cp, &capability_ops_unpassable); /* * Release our private reference (the proc filedesc still has one). */ fdrop(fcapp, td); return (0); } /* * Given a file descriptor, test it against a capability rights mask and then * return the file descriptor on which to actually perform the requested * operation. As long as the reference to fp_cap remains valid, the returned * pointer in *fp will remain valid, so no extra reference management is * required, and the caller should fdrop() fp_cap as normal when done with * both. */ int cap_funwrap(struct file *fp_cap, cap_rights_t rights, struct file **fpp) { struct capability *c; int error; if (fp_cap->f_type != DTYPE_CAPABILITY) { *fpp = fp_cap; return (0); } c = fp_cap->f_data; error = cap_check(c, rights); if (error) return (error); *fpp = c->cap_object; return (0); } /* * Slightly different routine for memory mapping file descriptors: unwrap the * capability and check CAP_MMAP, but also return a bitmask representing the * maximum mapping rights the capability allows on the object. */ int cap_funwrap_mmap(struct file *fp_cap, cap_rights_t rights, u_char *maxprotp, struct file **fpp) { struct capability *c; u_char maxprot; int error; if (fp_cap->f_type != DTYPE_CAPABILITY) { *fpp = fp_cap; *maxprotp = VM_PROT_ALL; return (0); } c = fp_cap->f_data; error = cap_check(c, rights | CAP_MMAP); if (error) return (error); *fpp = c->cap_object; maxprot = 0; if (c->cap_rights & CAP_READ) maxprot |= VM_PROT_READ; if (c->cap_rights & CAP_WRITE) maxprot |= VM_PROT_WRITE; if (c->cap_rights & CAP_MAPEXEC) maxprot |= VM_PROT_EXECUTE; *maxprotp = maxprot; return (0); } /* * When a capability is closed, simply drop the reference on the underlying * object and free the capability. fdrop() will handle the case where the * underlying object also needs to close, and the caller will have already * performed any object-specific lock or mqueue handling. */ static int capability_close(struct file *fp, struct thread *td) { struct capability *c; struct file *fp_object; KASSERT(fp->f_type == DTYPE_CAPABILITY, ("capability_close: !capability")); c = fp->f_data; fp->f_ops = &badfileops; fp->f_data = NULL; fp_object = c->cap_object; uma_zfree(capability_zone, c); return (fdrop(fp_object, td)); } /* * In general, file descriptor operations should never make it to the * capability, only the underlying file descriptor operation vector, so panic * if we do turn up here. */ static int capability_read(struct file *fp, struct uio *uio, struct ucred *active_cred, int flags, struct thread *td) { panic("capability_read"); } static int capability_write(struct file *fp, struct uio *uio, struct ucred *active_cred, int flags, struct thread *td) { panic("capability_write"); } static int capability_truncate(struct file *fp, off_t length, struct ucred *active_cred, struct thread *td) { panic("capability_truncate"); } static int capability_ioctl(struct file *fp, u_long com, void *data, struct ucred *active_cred, struct thread *td) { panic("capability_ioctl"); } static int capability_poll(struct file *fp, int events, struct ucred *active_cred, struct thread *td) { panic("capability_poll"); } static int capability_kqfilter(struct file *fp, struct knote *kn) { panic("capability_kqfilter"); } static int capability_stat(struct file *fp, struct stat *sb, struct ucred *active_cred, struct thread *td) { panic("capability_stat"); } int capability_chmod(struct file *fp, mode_t mode, struct ucred *active_cred, struct thread *td) { panic("capability_chmod"); } int capability_chown(struct file *fp, uid_t uid, gid_t gid, struct ucred *active_cred, struct thread *td) { panic("capability_chown"); } #else /* !CAPABILITIES */ /* * Stub Capability functions for when options CAPABILITIES isn't compiled * into the kernel. */ int sys_cap_new(struct thread *td, struct cap_new_args *uap) { return (ENOSYS); } int sys_cap_getrights(struct thread *td, struct cap_getrights_args *uap) { return (ENOSYS); } int cap_funwrap(struct file *fp_cap, cap_rights_t rights, struct file **fpp) { KASSERT(fp_cap->f_type != DTYPE_CAPABILITY, ("cap_funwrap: saw capability")); *fpp = fp_cap; return (0); } int cap_funwrap_mmap(struct file *fp_cap, cap_rights_t rights, u_char *maxprotp, struct file **fpp) { KASSERT(fp_cap->f_type != DTYPE_CAPABILITY, ("cap_funwrap_mmap: saw capability")); *fpp = fp_cap; *maxprotp = VM_PROT_ALL; return (0); } #endif /* CAPABILITIES */ Index: head/sys/sys/ktrace.h =================================================================== --- head/sys/sys/ktrace.h (revision 226268) +++ head/sys/sys/ktrace.h (revision 226269) @@ -1,239 +1,250 @@ /*- * Copyright (c) 1988, 1993 * The Regents of the University of California. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * @(#)ktrace.h 8.1 (Berkeley) 6/2/93 * $FreeBSD$ */ #ifndef _SYS_KTRACE_H_ #define _SYS_KTRACE_H_ /* * operations to ktrace system call (KTROP(op)) */ #define KTROP_SET 0 /* set trace points */ #define KTROP_CLEAR 1 /* clear trace points */ #define KTROP_CLEARFILE 2 /* stop all tracing to file */ #define KTROP(o) ((o)&3) /* macro to extract operation */ /* * flags (ORed in with operation) */ #define KTRFLAG_DESCEND 4 /* perform op on all children too */ /* * ktrace record header */ struct ktr_header { int ktr_len; /* length of buf */ short ktr_type; /* trace record type */ pid_t ktr_pid; /* process id */ char ktr_comm[MAXCOMLEN + 1];/* command name */ struct timeval ktr_time; /* timestamp */ intptr_t ktr_tid; /* was ktr_buffer */ }; /* * Test for kernel trace point (MP SAFE). * * KTRCHECK() just checks that the type is enabled and is only for * internal use in the ktrace subsystem. KTRPOINT() checks against * ktrace recursion as well as checking that the type is enabled and * is the public interface. */ #define KTRCHECK(td, type) ((td)->td_proc->p_traceflag & (1 << type)) #define KTRPOINT(td, type) \ (KTRCHECK((td), (type)) && !((td)->td_pflags & TDP_INKTRACE)) #define KTRCHECKDRAIN(td) (!(STAILQ_EMPTY(&(td)->td_proc->p_ktr))) #define KTRUSERRET(td) do { \ if (KTRCHECKDRAIN(td)) \ ktruserret(td); \ } while (0) /* * ktrace record types */ /* * KTR_SYSCALL - system call record */ #define KTR_SYSCALL 1 struct ktr_syscall { short ktr_code; /* syscall number */ short ktr_narg; /* number of arguments */ /* * followed by ktr_narg register_t */ register_t ktr_args[1]; }; /* * KTR_SYSRET - return from system call record */ #define KTR_SYSRET 2 struct ktr_sysret { short ktr_code; short ktr_eosys; int ktr_error; register_t ktr_retval; }; /* * KTR_NAMEI - namei record */ #define KTR_NAMEI 3 /* record contains pathname */ /* * KTR_GENIO - trace generic process i/o */ #define KTR_GENIO 4 struct ktr_genio { int ktr_fd; enum uio_rw ktr_rw; /* * followed by data successfully read/written */ }; /* * KTR_PSIG - trace processed signal */ #define KTR_PSIG 5 struct ktr_psig { int signo; sig_t action; int code; sigset_t mask; }; /* * KTR_CSW - trace context switches */ #define KTR_CSW 6 struct ktr_csw { int out; /* 1 if switch out, 0 if switch in */ int user; /* 1 if usermode (ivcsw), 0 if kernel (vcsw) */ }; /* * KTR_USER - data coming from userland */ #define KTR_USER_MAXLEN 2048 /* maximum length of passed data */ #define KTR_USER 7 /* * KTR_STRUCT - misc. structs */ #define KTR_STRUCT 8 /* * record contains null-terminated struct name followed by * struct contents */ struct sockaddr; struct stat; struct sysentvec; /* * KTR_SYSCTL - name of a sysctl MIB */ #define KTR_SYSCTL 9 /* record contains null-terminated MIB name */ /* * KTR_PROCCTOR - trace process creation (multiple ABI support) */ #define KTR_PROCCTOR 10 struct ktr_proc_ctor { u_int sv_flags; /* struct sysentvec sv_flags copy */ }; /* * KTR_PROCDTOR - trace process destruction (multiple ABI support) */ #define KTR_PROCDTOR 11 /* + * KTR_CAPFAIL - trace capability check failures + */ +#define KTR_CAPFAIL 12 +struct ktr_cap_fail { + cap_rights_t cap_needed; + cap_rights_t cap_held; +}; + +/* * KTR_DROP - If this bit is set in ktr_type, then at least one event * between the previous record and this record was dropped. */ #define KTR_DROP 0x8000 /* * kernel trace points (in p_traceflag) */ #define KTRFAC_MASK 0x00ffffff #define KTRFAC_SYSCALL (1<sa_len) #define ktrstat(s) \ ktrstruct("stat", (s), sizeof(struct stat)) #else #include __BEGIN_DECLS int ktrace(const char *, int, int, pid_t); int utrace(const void *, size_t); __END_DECLS #endif #endif Index: head/usr.bin/kdump/kdump.c =================================================================== --- head/usr.bin/kdump/kdump.c (revision 226268) +++ head/usr.bin/kdump/kdump.c (revision 226269) @@ -1,1624 +1,1661 @@ /*- * Copyright (c) 1988, 1993 * The Regents of the University of California. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #ifndef lint static const char copyright[] = "@(#) Copyright (c) 1988, 1993\n\ The Regents of the University of California. All rights reserved.\n"; #endif /* not lint */ #ifndef lint #if 0 static char sccsid[] = "@(#)kdump.c 8.1 (Berkeley) 6/6/93"; #endif #endif /* not lint */ #include __FBSDID("$FreeBSD$"); #define _KERNEL extern int errno; #include #undef _KERNEL #include #include #define _KERNEL #include #undef _KERNEL #include #include #include #include #include #include #include #include #ifdef IPX #include #include #endif #ifdef NETATALK #include #endif #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "ktrace.h" #include "kdump_subr.h" u_int abidump(struct ktr_header *); int fetchprocinfo(struct ktr_header *, u_int *); int fread_tail(void *, int, int); void dumpheader(struct ktr_header *); void ktrsyscall(struct ktr_syscall *, u_int); void ktrsysret(struct ktr_sysret *, u_int); void ktrnamei(char *, int); void hexdump(char *, int, int); void visdump(char *, int, int); void ktrgenio(struct ktr_genio *, int); void ktrpsig(struct ktr_psig *); void ktrcsw(struct ktr_csw *); void ktruser(int, unsigned char *); void ktrsockaddr(struct sockaddr *); void ktrstat(struct stat *); void ktrstruct(char *, size_t); +void ktrcapfail(struct ktr_cap_fail *); void usage(void); void ioctlname(unsigned long, int); int timestamp, decimal, fancy = 1, suppressdata, tail, threads, maxdata, resolv = 0, abiflag = 0; const char *tracefile = DEF_TRACEFILE; struct ktr_header ktr_header; #define TIME_FORMAT "%b %e %T %Y" #define eqs(s1, s2) (strcmp((s1), (s2)) == 0) #define print_number(i,n,c) do { \ if (decimal) \ printf("%c%jd", c, (intmax_t)*i); \ else \ printf("%c%#jx", c, (uintmax_t)(u_register_t)*i); \ i++; \ n--; \ c = ','; \ } while (0) #if defined(__amd64__) || defined(__i386__) void linux_ktrsyscall(struct ktr_syscall *); void linux_ktrsysret(struct ktr_sysret *); extern char *linux_syscallnames[]; extern int nlinux_syscalls; /* * from linux.h * Linux syscalls return negative errno's, we do positive and map them */ static int bsd_to_linux_errno[ELAST + 1] = { -0, -1, -2, -3, -4, -5, -6, -7, -8, -9, -10, -35, -12, -13, -14, -15, -16, -17, -18, -19, -20, -21, -22, -23, -24, -25, -26, -27, -28, -29, -30, -31, -32, -33, -34, -11,-115,-114, -88, -89, -90, -91, -92, -93, -94, -95, -96, -97, -98, -99, -100,-101,-102,-103,-104,-105,-106,-107,-108,-109, -110,-111, -40, -36,-112,-113, -39, -11, -87,-122, -116, -66, -6, -6, -6, -6, -6, -37, -38, -9, -6, -6, -43, -42, -75,-125, -84, -95, -16, -74, -72, -67, -71 }; #endif struct proc_info { TAILQ_ENTRY(proc_info) info; u_int sv_flags; pid_t pid; }; TAILQ_HEAD(trace_procs, proc_info) trace_procs; int main(int argc, char *argv[]) { int ch, ktrlen, size; void *m; int trpoints = ALL_POINTS; int drop_logged; pid_t pid = 0; u_int sv_flags; setlocale(LC_CTYPE, ""); while ((ch = getopt(argc,argv,"f:dElm:np:AHRrsTt:")) != -1) switch (ch) { case 'A': abiflag = 1; break; case 'f': tracefile = optarg; break; case 'd': decimal = 1; break; case 'l': tail = 1; break; case 'm': maxdata = atoi(optarg); break; case 'n': fancy = 0; break; case 'p': pid = atoi(optarg); break; case 'r': resolv = 1; break; case 's': suppressdata = 1; break; case 'E': timestamp = 3; /* elapsed timestamp */ break; case 'H': threads = 1; break; case 'R': timestamp = 2; /* relative timestamp */ break; case 'T': timestamp = 1; break; case 't': trpoints = getpoints(optarg); if (trpoints < 0) errx(1, "unknown trace point in %s", optarg); break; default: usage(); } if (argc > optind) usage(); m = malloc(size = 1025); if (m == NULL) errx(1, "%s", strerror(ENOMEM)); if (!freopen(tracefile, "r", stdin)) err(1, "%s", tracefile); TAILQ_INIT(&trace_procs); drop_logged = 0; while (fread_tail(&ktr_header, sizeof(struct ktr_header), 1)) { if (ktr_header.ktr_type & KTR_DROP) { ktr_header.ktr_type &= ~KTR_DROP; if (!drop_logged && threads) { printf( "%6jd %6jd %-8.*s Events dropped.\n", (intmax_t)ktr_header.ktr_pid, ktr_header.ktr_tid > 0 ? (intmax_t)ktr_header.ktr_tid : 0, MAXCOMLEN, ktr_header.ktr_comm); drop_logged = 1; } else if (!drop_logged) { printf("%6jd %-8.*s Events dropped.\n", (intmax_t)ktr_header.ktr_pid, MAXCOMLEN, ktr_header.ktr_comm); drop_logged = 1; } } if (trpoints & (1< size) { m = realloc(m, ktrlen+1); if (m == NULL) errx(1, "%s", strerror(ENOMEM)); size = ktrlen; } if (ktrlen && fread_tail(m, ktrlen, 1) == 0) errx(1, "data too short"); if (fetchprocinfo(&ktr_header, (u_int *)m) != 0) continue; sv_flags = abidump(&ktr_header); if (pid && ktr_header.ktr_pid != pid) continue; if ((trpoints & (1<ktr_type) { case KTR_PROCCTOR: TAILQ_FOREACH(pi, &trace_procs, info) { if (pi->pid == kth->ktr_pid) { TAILQ_REMOVE(&trace_procs, pi, info); break; } } pi = malloc(sizeof(struct proc_info)); if (pi == NULL) errx(1, "%s", strerror(ENOMEM)); pi->sv_flags = *flags; pi->pid = kth->ktr_pid; TAILQ_INSERT_TAIL(&trace_procs, pi, info); return (1); case KTR_PROCDTOR: TAILQ_FOREACH(pi, &trace_procs, info) { if (pi->pid == kth->ktr_pid) { TAILQ_REMOVE(&trace_procs, pi, info); free(pi); break; } } return (1); } return (0); } u_int abidump(struct ktr_header *kth) { struct proc_info *pi; const char *abi; const char *arch; u_int flags = 0; TAILQ_FOREACH(pi, &trace_procs, info) { if (pi->pid == kth->ktr_pid) { flags = pi->sv_flags; break; } } if (abiflag == 0) return (flags); switch (flags & SV_ABI_MASK) { case SV_ABI_LINUX: abi = "L"; break; case SV_ABI_FREEBSD: abi = "F"; break; default: abi = "U"; break; } if (flags != 0) { if (flags & SV_LP64) arch = "64"; else arch = "32"; } else arch = "00"; printf("%s%s ", abi, arch); return (flags); } void dumpheader(struct ktr_header *kth) { static char unknown[64]; static struct timeval prevtime, temp; const char *type; switch (kth->ktr_type) { case KTR_SYSCALL: type = "CALL"; break; case KTR_SYSRET: type = "RET "; break; case KTR_NAMEI: type = "NAMI"; break; case KTR_GENIO: type = "GIO "; break; case KTR_PSIG: type = "PSIG"; break; case KTR_CSW: type = "CSW "; break; case KTR_USER: type = "USER"; break; case KTR_STRUCT: type = "STRU"; break; case KTR_SYSCTL: type = "SCTL"; break; case KTR_PROCCTOR: /* FALLTHROUGH */ case KTR_PROCDTOR: return; + case KTR_CAPFAIL: + type = "CAP "; + break; default: sprintf(unknown, "UNKNOWN(%d)", kth->ktr_type); type = unknown; } /* * The ktr_tid field was previously the ktr_buffer field, which held * the kernel pointer value for the buffer associated with data * following the record header. It now holds a threadid, but only * for trace files after the change. Older trace files still contain * kernel pointers. Detect this and suppress the results by printing * negative tid's as 0. */ if (threads) printf("%6jd %6jd %-8.*s ", (intmax_t)kth->ktr_pid, kth->ktr_tid > 0 ? (intmax_t)kth->ktr_tid : 0, MAXCOMLEN, kth->ktr_comm); else printf("%6jd %-8.*s ", (intmax_t)kth->ktr_pid, MAXCOMLEN, kth->ktr_comm); if (timestamp) { if (timestamp == 3) { if (prevtime.tv_sec == 0) prevtime = kth->ktr_time; timevalsub(&kth->ktr_time, &prevtime); } if (timestamp == 2) { temp = kth->ktr_time; timevalsub(&kth->ktr_time, &prevtime); prevtime = temp; } printf("%jd.%06ld ", (intmax_t)kth->ktr_time.tv_sec, kth->ktr_time.tv_usec); } printf("%s ", type); } #include #define KTRACE #include #undef KTRACE int nsyscalls = sizeof (syscallnames) / sizeof (syscallnames[0]); void ktrsyscall(struct ktr_syscall *ktr, u_int flags) { int narg = ktr->ktr_narg; register_t *ip; + intmax_t arg; if ((flags != 0 && ((flags & SV_ABI_MASK) != SV_ABI_FREEBSD)) || (ktr->ktr_code >= nsyscalls || ktr->ktr_code < 0)) printf("[%d]", ktr->ktr_code); else printf("%s", syscallnames[ktr->ktr_code]); ip = &ktr->ktr_args[0]; if (narg) { char c = '('; if (fancy && (flags == 0 || (flags & SV_ABI_MASK) == SV_ABI_FREEBSD)) { switch (ktr->ktr_code) { case SYS_ioctl: { const char *cp; print_number(ip, narg, c); putchar(c); ioctlname(*ip, decimal); c = ','; ip++; narg--; break; } case SYS_ptrace: putchar('('); ptraceopname(*ip); c = ','; ip++; narg--; break; case SYS_access: case SYS_eaccess: print_number(ip, narg, c); putchar(','); accessmodename(*ip); ip++; narg--; break; case SYS_open: print_number(ip, narg, c); putchar(','); flagsandmodename(ip[0], ip[1], decimal); ip += 2; narg -= 2; break; case SYS_wait4: print_number(ip, narg, c); print_number(ip, narg, c); putchar(','); wait4optname(*ip); ip++; narg--; break; case SYS_chmod: case SYS_fchmod: case SYS_lchmod: print_number(ip, narg, c); putchar(','); modename(*ip); ip++; narg--; break; case SYS_mknod: print_number(ip, narg, c); putchar(','); modename(*ip); ip++; narg--; break; case SYS_getfsstat: print_number(ip, narg, c); print_number(ip, narg, c); putchar(','); getfsstatflagsname(*ip); ip++; narg--; break; case SYS_mount: print_number(ip, narg, c); print_number(ip, narg, c); putchar(','); mountflagsname(*ip); ip++; narg--; break; case SYS_unmount: print_number(ip, narg, c); putchar(','); mountflagsname(*ip); ip++; narg--; break; case SYS_recvmsg: case SYS_sendmsg: print_number(ip, narg, c); print_number(ip, narg, c); putchar(','); sendrecvflagsname(*ip); ip++; narg--; break; case SYS_recvfrom: case SYS_sendto: print_number(ip, narg, c); print_number(ip, narg, c); print_number(ip, narg, c); putchar(','); sendrecvflagsname(*ip); ip++; narg--; break; case SYS_chflags: case SYS_fchflags: case SYS_lchflags: print_number(ip, narg, c); putchar(','); modename(*ip); ip++; narg--; break; case SYS_kill: print_number(ip, narg, c); putchar(','); signame(*ip); ip++; narg--; break; case SYS_reboot: putchar('('); rebootoptname(*ip); ip++; narg--; break; case SYS_umask: putchar('('); modename(*ip); ip++; narg--; break; case SYS_msync: print_number(ip, narg, c); print_number(ip, narg, c); putchar(','); msyncflagsname(*ip); ip++; narg--; break; #ifdef SYS_freebsd6_mmap case SYS_freebsd6_mmap: print_number(ip, narg, c); print_number(ip, narg, c); putchar(','); mmapprotname(*ip); putchar(','); ip++; narg--; mmapflagsname(*ip); ip++; narg--; break; #endif case SYS_mmap: print_number(ip, narg, c); print_number(ip, narg, c); putchar(','); mmapprotname(*ip); putchar(','); ip++; narg--; mmapflagsname(*ip); ip++; narg--; break; case SYS_mprotect: print_number(ip, narg, c); print_number(ip, narg, c); putchar(','); mmapprotname(*ip); ip++; narg--; break; case SYS_madvise: print_number(ip, narg, c); print_number(ip, narg, c); putchar(','); madvisebehavname(*ip); ip++; narg--; break; case SYS_setpriority: print_number(ip, narg, c); print_number(ip, narg, c); putchar(','); prioname(*ip); ip++; narg--; break; case SYS_fcntl: print_number(ip, narg, c); putchar(','); fcntlcmdname(ip[0], ip[1], decimal); ip += 2; narg -= 2; break; case SYS_socket: { int sockdomain; putchar('('); sockdomain = *ip; sockdomainname(sockdomain); ip++; narg--; putchar(','); socktypename(*ip); ip++; narg--; if (sockdomain == PF_INET || sockdomain == PF_INET6) { putchar(','); sockipprotoname(*ip); ip++; narg--; } c = ','; break; } case SYS_setsockopt: case SYS_getsockopt: print_number(ip, narg, c); putchar(','); sockoptlevelname(*ip, decimal); if (*ip == SOL_SOCKET) { ip++; narg--; putchar(','); sockoptname(*ip); } ip++; narg--; break; #ifdef SYS_freebsd6_lseek case SYS_freebsd6_lseek: print_number(ip, narg, c); /* Hidden 'pad' argument, not in lseek(2) */ print_number(ip, narg, c); print_number(ip, narg, c); putchar(','); whencename(*ip); ip++; narg--; break; #endif case SYS_lseek: print_number(ip, narg, c); /* Hidden 'pad' argument, not in lseek(2) */ print_number(ip, narg, c); putchar(','); whencename(*ip); ip++; narg--; break; case SYS_flock: print_number(ip, narg, c); putchar(','); flockname(*ip); ip++; narg--; break; case SYS_mkfifo: case SYS_mkdir: print_number(ip, narg, c); putchar(','); modename(*ip); ip++; narg--; break; case SYS_shutdown: print_number(ip, narg, c); putchar(','); shutdownhowname(*ip); ip++; narg--; break; case SYS_socketpair: putchar('('); sockdomainname(*ip); ip++; narg--; putchar(','); socktypename(*ip); ip++; narg--; c = ','; break; case SYS_getrlimit: case SYS_setrlimit: putchar('('); rlimitname(*ip); ip++; narg--; c = ','; break; case SYS_quotactl: print_number(ip, narg, c); putchar(','); quotactlname(*ip); ip++; narg--; c = ','; break; case SYS_nfssvc: putchar('('); nfssvcname(*ip); ip++; narg--; c = ','; break; case SYS_rtprio: putchar('('); rtprioname(*ip); ip++; narg--; c = ','; break; case SYS___semctl: print_number(ip, narg, c); print_number(ip, narg, c); putchar(','); semctlname(*ip); ip++; narg--; break; case SYS_semget: print_number(ip, narg, c); print_number(ip, narg, c); putchar(','); semgetname(*ip); ip++; narg--; break; case SYS_msgctl: print_number(ip, narg, c); putchar(','); shmctlname(*ip); ip++; narg--; break; case SYS_shmat: print_number(ip, narg, c); print_number(ip, narg, c); putchar(','); shmatname(*ip); ip++; narg--; break; case SYS_shmctl: print_number(ip, narg, c); putchar(','); shmctlname(*ip); ip++; narg--; break; case SYS_minherit: print_number(ip, narg, c); print_number(ip, narg, c); putchar(','); minheritname(*ip); ip++; narg--; break; case SYS_rfork: putchar('('); rforkname(*ip); ip++; narg--; c = ','; break; case SYS_lio_listio: putchar('('); lio_listioname(*ip); ip++; narg--; c = ','; break; case SYS_mlockall: putchar('('); mlockallname(*ip); ip++; narg--; break; case SYS_sched_setscheduler: print_number(ip, narg, c); putchar(','); schedpolicyname(*ip); ip++; narg--; break; case SYS_sched_get_priority_max: case SYS_sched_get_priority_min: putchar('('); schedpolicyname(*ip); ip++; narg--; break; case SYS_sendfile: print_number(ip, narg, c); print_number(ip, narg, c); print_number(ip, narg, c); print_number(ip, narg, c); print_number(ip, narg, c); print_number(ip, narg, c); putchar(','); sendfileflagsname(*ip); ip++; narg--; break; case SYS_kldsym: print_number(ip, narg, c); putchar(','); kldsymcmdname(*ip); ip++; narg--; break; case SYS_sigprocmask: putchar('('); sigprocmaskhowname(*ip); ip++; narg--; c = ','; break; case SYS___acl_get_file: case SYS___acl_set_file: case SYS___acl_get_fd: case SYS___acl_set_fd: case SYS___acl_delete_file: case SYS___acl_delete_fd: case SYS___acl_aclcheck_file: case SYS___acl_aclcheck_fd: case SYS___acl_get_link: case SYS___acl_set_link: case SYS___acl_delete_link: case SYS___acl_aclcheck_link: print_number(ip, narg, c); putchar(','); acltypename(*ip); ip++; narg--; break; case SYS_sigaction: putchar('('); signame(*ip); ip++; narg--; c = ','; break; case SYS_extattrctl: print_number(ip, narg, c); putchar(','); extattrctlname(*ip); ip++; narg--; break; case SYS_nmount: print_number(ip, narg, c); print_number(ip, narg, c); putchar(','); mountflagsname(*ip); ip++; narg--; break; case SYS_thr_create: print_number(ip, narg, c); print_number(ip, narg, c); putchar(','); thrcreateflagsname(*ip); ip++; narg--; break; case SYS_thr_kill: print_number(ip, narg, c); putchar(','); signame(*ip); ip++; narg--; break; case SYS_kldunloadf: print_number(ip, narg, c); putchar(','); kldunloadfflagsname(*ip); ip++; narg--; break; - case SYS_cap_new: - print_number(ip, narg, c); - putchar(','); - capname(*ip); - ip++; - narg--; + case SYS_cap_new: + print_number(ip, narg, c); + putchar(','); + arg = *ip; + ip++; + narg--; + /* + * Hack: the second argument is a + * cap_rights_t, which 64 bits wide, so on + * 32-bit systems, it is split between two + * registers. + * + * Since sizeof() is not evaluated by the + * preprocessor, we can't use an #ifdef, + * but the compiler will probably optimize + * the code out anyway. + */ + if (sizeof(cap_rights_t) > sizeof(register_t)) { +#if _BYTE_ORDER == _LITTLE_ENDIAN + arg = ((intmax_t)*ip << 32) + arg; +#else + arg = (arg << 32) + *ip; +#endif + ip++; + narg--; + } + capname(arg); break; } } while (narg > 0) { print_number(ip, narg, c); } putchar(')'); } putchar('\n'); } void ktrsysret(struct ktr_sysret *ktr, u_int flags) { register_t ret = ktr->ktr_retval; int error = ktr->ktr_error; int code = ktr->ktr_code; if ((flags != 0 && ((flags & SV_ABI_MASK) != SV_ABI_FREEBSD)) || (code >= nsyscalls || code < 0)) printf("[%d] ", code); else printf("%s ", syscallnames[code]); if (error == 0) { if (fancy) { printf("%ld", (long)ret); if (ret < 0 || ret > 9) printf("/%#lx", (unsigned long)ret); } else { if (decimal) printf("%ld", (long)ret); else printf("%#lx", (unsigned long)ret); } } else if (error == ERESTART) printf("RESTART"); else if (error == EJUSTRETURN) printf("JUSTRETURN"); else { printf("-1 errno %d", ktr->ktr_error); if (fancy) printf(" %s", strerror(ktr->ktr_error)); } putchar('\n'); } void ktrnamei(char *cp, int len) { printf("\"%.*s\"\n", len, cp); } void hexdump(char *p, int len, int screenwidth) { int n, i; int width; width = 0; do { width += 2; i = 13; /* base offset */ i += (width / 2) + 1; /* spaces every second byte */ i += (width * 2); /* width of bytes */ i += 3; /* " |" */ i += width; /* each byte */ i += 1; /* "|" */ } while (i < screenwidth); width -= 2; for (n = 0; n < len; n += width) { for (i = n; i < n + width; i++) { if ((i % width) == 0) { /* beginning of line */ printf(" 0x%04x", i); } if ((i % 2) == 0) { printf(" "); } if (i < len) printf("%02x", p[i] & 0xff); else printf(" "); } printf(" |"); for (i = n; i < n + width; i++) { if (i >= len) break; if (p[i] >= ' ' && p[i] <= '~') printf("%c", p[i]); else printf("."); } printf("|\n"); } if ((i % width) != 0) printf("\n"); } void visdump(char *dp, int datalen, int screenwidth) { int col = 0; char *cp; int width; char visbuf[5]; printf(" \""); col = 8; for (;datalen > 0; datalen--, dp++) { vis(visbuf, *dp, VIS_CSTYLE, *(dp+1)); cp = visbuf; /* * Keep track of printables and * space chars (like fold(1)). */ if (col == 0) { putchar('\t'); col = 8; } switch(*cp) { case '\n': col = 0; putchar('\n'); continue; case '\t': width = 8 - (col&07); break; default: width = strlen(cp); } if (col + width > (screenwidth-2)) { printf("\\\n\t"); col = 8; } col += width; do { putchar(*cp++); } while (*cp); } if (col == 0) printf(" "); printf("\"\n"); } void ktrgenio(struct ktr_genio *ktr, int len) { int datalen = len - sizeof (struct ktr_genio); char *dp = (char *)ktr + sizeof (struct ktr_genio); static int screenwidth = 0; int i, binary; if (screenwidth == 0) { struct winsize ws; if (fancy && ioctl(fileno(stderr), TIOCGWINSZ, &ws) != -1 && ws.ws_col > 8) screenwidth = ws.ws_col; else screenwidth = 80; } printf("fd %d %s %d byte%s\n", ktr->ktr_fd, ktr->ktr_rw == UIO_READ ? "read" : "wrote", datalen, datalen == 1 ? "" : "s"); if (suppressdata) return; if (maxdata && datalen > maxdata) datalen = maxdata; for (i = 0, binary = 0; i < datalen && binary == 0; i++) { if (dp[i] >= 32 && dp[i] < 127) continue; if (dp[i] == 10 || dp[i] == 13 || dp[i] == 0 || dp[i] == 9) continue; binary = 1; } if (binary) hexdump(dp, datalen, screenwidth); else visdump(dp, datalen, screenwidth); } const char *signames[] = { "NULL", "HUP", "INT", "QUIT", "ILL", "TRAP", "IOT", /* 1 - 6 */ "EMT", "FPE", "KILL", "BUS", "SEGV", "SYS", /* 7 - 12 */ "PIPE", "ALRM", "TERM", "URG", "STOP", "TSTP", /* 13 - 18 */ "CONT", "CHLD", "TTIN", "TTOU", "IO", "XCPU", /* 19 - 24 */ "XFSZ", "VTALRM", "PROF", "WINCH", "29", "USR1", /* 25 - 30 */ "USR2", NULL, /* 31 - 32 */ }; void ktrpsig(struct ktr_psig *psig) { if (psig->signo > 0 && psig->signo < NSIG) printf("SIG%s ", signames[psig->signo]); else printf("SIG %d ", psig->signo); if (psig->action == SIG_DFL) printf("SIG_DFL code=0x%x\n", psig->code); else { printf("caught handler=0x%lx mask=0x%x code=0x%x\n", (u_long)psig->action, psig->mask.__bits[0], psig->code); } } void ktrcsw(struct ktr_csw *cs) { printf("%s %s\n", cs->out ? "stop" : "resume", cs->user ? "user" : "kernel"); } #define UTRACE_DLOPEN_START 1 #define UTRACE_DLOPEN_STOP 2 #define UTRACE_DLCLOSE_START 3 #define UTRACE_DLCLOSE_STOP 4 #define UTRACE_LOAD_OBJECT 5 #define UTRACE_UNLOAD_OBJECT 6 #define UTRACE_ADD_RUNDEP 7 #define UTRACE_PRELOAD_FINISHED 8 #define UTRACE_INIT_CALL 9 #define UTRACE_FINI_CALL 10 struct utrace_rtld { char sig[4]; /* 'RTLD' */ int event; void *handle; void *mapbase; size_t mapsize; int refcnt; char name[MAXPATHLEN]; }; void ktruser_rtld(int len, unsigned char *p) { struct utrace_rtld *ut = (struct utrace_rtld *)p; void *parent; int mode; switch (ut->event) { case UTRACE_DLOPEN_START: mode = ut->refcnt; printf("dlopen(%s, ", ut->name); switch (mode & RTLD_MODEMASK) { case RTLD_NOW: printf("RTLD_NOW"); break; case RTLD_LAZY: printf("RTLD_LAZY"); break; default: printf("%#x", mode & RTLD_MODEMASK); } if (mode & RTLD_GLOBAL) printf(" | RTLD_GLOBAL"); if (mode & RTLD_TRACE) printf(" | RTLD_TRACE"); if (mode & ~(RTLD_MODEMASK | RTLD_GLOBAL | RTLD_TRACE)) printf(" | %#x", mode & ~(RTLD_MODEMASK | RTLD_GLOBAL | RTLD_TRACE)); printf(")\n"); break; case UTRACE_DLOPEN_STOP: printf("%p = dlopen(%s) ref %d\n", ut->handle, ut->name, ut->refcnt); break; case UTRACE_DLCLOSE_START: printf("dlclose(%p) (%s, %d)\n", ut->handle, ut->name, ut->refcnt); break; case UTRACE_DLCLOSE_STOP: printf("dlclose(%p) finished\n", ut->handle); break; case UTRACE_LOAD_OBJECT: printf("RTLD: loaded %p @ %p - %p (%s)\n", ut->handle, ut->mapbase, (char *)ut->mapbase + ut->mapsize - 1, ut->name); break; case UTRACE_UNLOAD_OBJECT: printf("RTLD: unloaded %p @ %p - %p (%s)\n", ut->handle, ut->mapbase, (char *)ut->mapbase + ut->mapsize - 1, ut->name); break; case UTRACE_ADD_RUNDEP: parent = ut->mapbase; printf("RTLD: %p now depends on %p (%s, %d)\n", parent, ut->handle, ut->name, ut->refcnt); break; case UTRACE_PRELOAD_FINISHED: printf("RTLD: LD_PRELOAD finished\n"); break; case UTRACE_INIT_CALL: printf("RTLD: init %p for %p (%s)\n", ut->mapbase, ut->handle, ut->name); break; case UTRACE_FINI_CALL: printf("RTLD: fini %p for %p (%s)\n", ut->mapbase, ut->handle, ut->name); break; default: p += 4; len -= 4; printf("RTLD: %d ", len); while (len--) if (decimal) printf(" %d", *p++); else printf(" %02x", *p++); printf("\n"); } } struct utrace_malloc { void *p; size_t s; void *r; }; void ktruser_malloc(int len, unsigned char *p) { struct utrace_malloc *ut = (struct utrace_malloc *)p; if (ut->p == (void *)(intptr_t)(-1)) printf("malloc_init()\n"); else if (ut->s == 0) printf("free(%p)\n", ut->p); else if (ut->p == NULL) printf("%p = malloc(%zu)\n", ut->r, ut->s); else printf("%p = realloc(%p, %zu)\n", ut->r, ut->p, ut->s); } void ktruser(int len, unsigned char *p) { if (len >= 8 && bcmp(p, "RTLD", 4) == 0) { ktruser_rtld(len, p); return; } if (len == sizeof(struct utrace_malloc)) { ktruser_malloc(len, p); return; } printf("%d ", len); while (len--) if (decimal) printf(" %d", *p++); else printf(" %02x", *p++); printf("\n"); } void ktrsockaddr(struct sockaddr *sa) { /* TODO: Support additional address families #include struct sockaddr_natm *natm; #include struct sockaddr_nb *nb; */ char addr[64]; /* * note: ktrstruct() has already verified that sa points to a * buffer at least sizeof(struct sockaddr) bytes long and exactly * sa->sa_len bytes long. */ printf("struct sockaddr { "); sockfamilyname(sa->sa_family); printf(", "); #define check_sockaddr_len(n) \ if (sa_##n->s##n##_len < sizeof(struct sockaddr_##n)) { \ printf("invalid"); \ break; \ } switch(sa->sa_family) { case AF_INET: { struct sockaddr_in *sa_in; sa_in = (struct sockaddr_in *)sa; check_sockaddr_len(in); inet_ntop(AF_INET, &sa_in->sin_addr, addr, sizeof addr); printf("%s:%u", addr, ntohs(sa_in->sin_port)); break; } #ifdef NETATALK case AF_APPLETALK: { struct sockaddr_at *sa_at; struct netrange *nr; sa_at = (struct sockaddr_at *)sa; check_sockaddr_len(at); nr = &sa_at->sat_range.r_netrange; printf("%d.%d, %d-%d, %d", ntohs(sa_at->sat_addr.s_net), sa_at->sat_addr.s_node, ntohs(nr->nr_firstnet), ntohs(nr->nr_lastnet), nr->nr_phase); break; } #endif case AF_INET6: { struct sockaddr_in6 *sa_in6; sa_in6 = (struct sockaddr_in6 *)sa; check_sockaddr_len(in6); inet_ntop(AF_INET6, &sa_in6->sin6_addr, addr, sizeof addr); printf("[%s]:%u", addr, htons(sa_in6->sin6_port)); break; } #ifdef IPX case AF_IPX: { struct sockaddr_ipx *sa_ipx; sa_ipx = (struct sockaddr_ipx *)sa; check_sockaddr_len(ipx); /* XXX wish we had ipx_ntop */ printf("%s", ipx_ntoa(sa_ipx->sipx_addr)); break; } #endif case AF_UNIX: { struct sockaddr_un *sa_un; sa_un = (struct sockaddr_un *)sa; check_sockaddr_len(un); printf("%.*s", (int)sizeof(sa_un->sun_path), sa_un->sun_path); break; } default: printf("unknown address family"); } printf(" }\n"); } void ktrstat(struct stat *statp) { char mode[12], timestr[PATH_MAX + 4]; struct passwd *pwd; struct group *grp; struct tm *tm; /* * note: ktrstruct() has already verified that statp points to a * buffer exactly sizeof(struct stat) bytes long. */ printf("struct stat {"); strmode(statp->st_mode, mode); printf("dev=%ju, ino=%ju, mode=%s, nlink=%ju, ", (uintmax_t)statp->st_dev, (uintmax_t)statp->st_ino, mode, (uintmax_t)statp->st_nlink); if (resolv == 0 || (pwd = getpwuid(statp->st_uid)) == NULL) printf("uid=%ju, ", (uintmax_t)statp->st_uid); else printf("uid=\"%s\", ", pwd->pw_name); if (resolv == 0 || (grp = getgrgid(statp->st_gid)) == NULL) printf("gid=%ju, ", (uintmax_t)statp->st_gid); else printf("gid=\"%s\", ", grp->gr_name); printf("rdev=%ju, ", (uintmax_t)statp->st_rdev); printf("atime="); if (resolv == 0) printf("%jd", (intmax_t)statp->st_atim.tv_sec); else { tm = localtime(&statp->st_atim.tv_sec); strftime(timestr, sizeof(timestr), TIME_FORMAT, tm); printf("\"%s\"", timestr); } if (statp->st_atim.tv_nsec != 0) printf(".%09ld, ", statp->st_atim.tv_nsec); else printf(", "); printf("stime="); if (resolv == 0) printf("%jd", (intmax_t)statp->st_mtim.tv_sec); else { tm = localtime(&statp->st_mtim.tv_sec); strftime(timestr, sizeof(timestr), TIME_FORMAT, tm); printf("\"%s\"", timestr); } if (statp->st_mtim.tv_nsec != 0) printf(".%09ld, ", statp->st_mtim.tv_nsec); else printf(", "); printf("ctime="); if (resolv == 0) printf("%jd", (intmax_t)statp->st_ctim.tv_sec); else { tm = localtime(&statp->st_ctim.tv_sec); strftime(timestr, sizeof(timestr), TIME_FORMAT, tm); printf("\"%s\"", timestr); } if (statp->st_ctim.tv_nsec != 0) printf(".%09ld, ", statp->st_ctim.tv_nsec); else printf(", "); printf("birthtime="); if (resolv == 0) printf("%jd", (intmax_t)statp->st_birthtim.tv_sec); else { tm = localtime(&statp->st_birthtim.tv_sec); strftime(timestr, sizeof(timestr), TIME_FORMAT, tm); printf("\"%s\"", timestr); } if (statp->st_birthtim.tv_nsec != 0) printf(".%09ld, ", statp->st_birthtim.tv_nsec); else printf(", "); printf("size=%jd, blksize=%ju, blocks=%jd, flags=0x%x", (uintmax_t)statp->st_size, (uintmax_t)statp->st_blksize, (intmax_t)statp->st_blocks, statp->st_flags); printf(" }\n"); } void ktrstruct(char *buf, size_t buflen) { char *name, *data; size_t namelen, datalen; int i; struct stat sb; struct sockaddr_storage ss; for (name = buf, namelen = 0; namelen < buflen && name[namelen] != '\0'; ++namelen) /* nothing */; if (namelen == buflen) goto invalid; if (name[namelen] != '\0') goto invalid; data = buf + namelen + 1; datalen = buflen - namelen - 1; if (datalen == 0) goto invalid; /* sanity check */ for (i = 0; i < namelen; ++i) if (!isalpha((unsigned char)name[i])) goto invalid; if (strcmp(name, "stat") == 0) { if (datalen != sizeof(struct stat)) goto invalid; memcpy(&sb, data, datalen); ktrstat(&sb); } else if (strcmp(name, "sockaddr") == 0) { if (datalen > sizeof(ss)) goto invalid; memcpy(&ss, data, datalen); if (datalen < sizeof(struct sockaddr) || datalen != ss.ss_len) goto invalid; ktrsockaddr((struct sockaddr *)&ss); } else { printf("unknown structure\n"); } return; invalid: printf("invalid record\n"); +} + +void +ktrcapfail(struct ktr_cap_fail *ktr) +{ + printf("needed "); + capname((intmax_t)ktr->cap_needed); + printf(" held "); + capname((intmax_t)ktr->cap_held); } #if defined(__amd64__) || defined(__i386__) void linux_ktrsyscall(struct ktr_syscall *ktr) { int narg = ktr->ktr_narg; register_t *ip; if (ktr->ktr_code >= nlinux_syscalls || ktr->ktr_code < 0) printf("[%d]", ktr->ktr_code); else printf("%s", linux_syscallnames[ktr->ktr_code]); ip = &ktr->ktr_args[0]; if (narg) { char c = '('; while (narg > 0) print_number(ip, narg, c); putchar(')'); } putchar('\n'); } void linux_ktrsysret(struct ktr_sysret *ktr) { register_t ret = ktr->ktr_retval; int error = ktr->ktr_error; int code = ktr->ktr_code; if (code >= nlinux_syscalls || code < 0) printf("[%d] ", code); else printf("%s ", linux_syscallnames[code]); if (error == 0) { if (fancy) { printf("%ld", (long)ret); if (ret < 0 || ret > 9) printf("/%#lx", (unsigned long)ret); } else { if (decimal) printf("%ld", (long)ret); else printf("%#lx", (unsigned long)ret); } } else if (error == ERESTART) printf("RESTART"); else if (error == EJUSTRETURN) printf("JUSTRETURN"); else { if (ktr->ktr_error <= ELAST + 1) error = abs(bsd_to_linux_errno[ktr->ktr_error]); else error = 999; printf("-1 errno %d", error); if (fancy) printf(" %s", strerror(ktr->ktr_error)); } putchar('\n'); } #endif void usage(void) { fprintf(stderr, "usage: kdump [-dEnlHRrsTA] [-f trfile] " "[-m maxdata] [-p pid] [-t trstr]\n"); exit(1); } Index: head/usr.bin/ktrace/ktrace.1 =================================================================== --- head/usr.bin/ktrace/ktrace.1 (revision 226268) +++ head/usr.bin/ktrace/ktrace.1 (revision 226269) @@ -1,181 +1,183 @@ .\" Copyright (c) 1990, 1993 .\" The Regents of the University of California. All rights reserved. .\" .\" Redistribution and use in source and binary forms, with or without .\" modification, are permitted provided that the following conditions .\" are met: .\" 1. Redistributions of source code must retain the above copyright .\" notice, this list of conditions and the following disclaimer. .\" 2. Redistributions in binary form must reproduce the above copyright .\" notice, this list of conditions and the following disclaimer in the .\" documentation and/or other materials provided with the distribution. .\" 4. Neither the name of the University nor the names of its contributors .\" may be used to endorse or promote products derived from this software .\" without specific prior written permission. .\" .\" THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND .\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE .\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE .\" ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE .\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL .\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS .\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) .\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT .\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY .\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF .\" SUCH DAMAGE. .\" .\" @(#)ktrace.1 8.1 (Berkeley) 6/6/93 .\" $FreeBSD$ .\" -.Dd February 23, 2008 +.Dd October 10, 2011 .Dt KTRACE 1 .Os .Sh NAME .Nm ktrace .Nd enable kernel process tracing .Sh SYNOPSIS .Nm .Op Fl aCcdi .Op Fl f Ar trfile .Op Fl g Ar pgrp | Fl p Ar pid .Op Fl t Ar trstr .Nm .Op Fl adi .Op Fl f Ar trfile .Op Fl t Ar trstr .Ar command .Sh DESCRIPTION The .Nm utility enables kernel trace logging for the specified processes. Kernel trace data is logged to the file .Pa ktrace.out . The kernel operations that are traced include system calls, namei translations, signal processing, and .Tn I/O . .Pp Once tracing is enabled on a process, trace data will be logged until either the process exits or the trace point is cleared. A traced process can generate enormous amounts of log data quickly; It is strongly suggested that users memorize how to disable tracing before attempting to trace a process. The following command is sufficient to disable tracing on all user-owned processes, and, if executed by root, all processes: .Pp .Dl \&$ ktrace -C .Pp The trace file is not human readable; use .Xr kdump 1 to decode it. .Pp The utility may be used only with a kernel that has been built with the .Dq KTRACE option in the kernel configuration file. .Pp The options are: .Bl -tag -width indent .It Fl a Append to the trace file instead of recreating it. .It Fl C Disable tracing on all user-owned processes, and, if executed by root, all processes in the system. .It Fl c Clear the trace points associated with the specified file or processes. .It Fl d Descendants; perform the operation for all current children of the designated processes. .It Fl f Ar trfile Log trace records to .Ar trfile instead of .Pa ktrace.out . .It Fl g Ar pgid Enable (disable) tracing on all processes in the process group (only one .Fl g flag is permitted). .It Fl i Inherit; pass the trace flags to all future children of the designated processes. .It Fl p Ar pid Enable (disable) tracing on the indicated process id (only one .Fl p flag is permitted). .It Fl t Ar trstr The string argument represents the kernel trace points, one per letter. The following table equates the letters with the tracepoints: .Pp .Bl -tag -width flag -compact .It Cm c trace system calls .It Cm i trace .Tn I/O .It Cm n trace namei translations +.It Cm p +trace capability check failures .It Cm s trace signal processing .It Cm t trace various structures .It Cm u userland traces .It Cm w context switches .It Cm y trace .Xr sysctl 3 requests .It Cm + trace the default set of trace points - -.Cm c , i , n , s , t , u , y +.Cm c , i , n , p , s , t , u , y .El .It Ar command Execute .Ar command with the specified trace flags. .El .Pp The .Fl p , .Fl g , and .Ar command options are mutually exclusive. .Sh EXAMPLES # trace all kernel operations of process id 34 .Dl $ ktrace -p 34 .Pp # trace all kernel operations of processes in process group 15 and # pass the trace flags to all current and future children .Dl $ ktrace -idg 15 .Pp # disable all tracing of process 65 .Dl $ ktrace -cp 65 .Pp # disable tracing signals on process 70 and all current children .Dl $ ktrace -t s -cdp 70 .Pp # enable tracing of .Tn I/O on process 67 .Dl $ ktrace -ti -p 67 .Pp # run the command "w", tracing only system calls .Dl $ ktrace -tc w .Pp # disable all tracing to the file "tracedata" .Dl $ ktrace -c -f tracedata .Pp # disable tracing of all user-owned processes .Dl $ ktrace -C .Sh SEE ALSO .Xr kdump 1 .Sh HISTORY The .Nm command appeared in .Bx 4.4 . .Sh BUGS Only works if .Ar trfile is a regular file. Index: head/usr.bin/ktrace/ktrace.h =================================================================== --- head/usr.bin/ktrace/ktrace.h (revision 226268) +++ head/usr.bin/ktrace/ktrace.h (revision 226269) @@ -1,43 +1,43 @@ /*- * Copyright (c) 1988, 1993 * The Regents of the University of California. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * * @(#)ktrace.h 8.1 (Berkeley) 6/6/93 * $FreeBSD$ */ #define DEF_POINTS (KTRFAC_SYSCALL | KTRFAC_SYSRET | KTRFAC_NAMEI | \ KTRFAC_GENIO | KTRFAC_PSIG | KTRFAC_USER | \ - KTRFAC_STRUCT | KTRFAC_SYSCTL) + KTRFAC_STRUCT | KTRFAC_SYSCTL | KTRFAC_CAPFAIL) #define PROC_ABI_POINTS (KTRFAC_PROCCTOR | KTRFAC_PROCDTOR) #define ALL_POINTS (DEF_POINTS | KTRFAC_CSW | PROC_ABI_POINTS) #define DEF_TRACEFILE "ktrace.out" int getpoints(char *); Index: head/usr.bin/ktrace/subr.c =================================================================== --- head/usr.bin/ktrace/subr.c (revision 226268) +++ head/usr.bin/ktrace/subr.c (revision 226269) @@ -1,123 +1,126 @@ /*- * Copyright (c) 1988, 1993 * The Regents of the University of California. All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #if 0 #ifndef lint static char sccsid[] = "@(#)subr.c 8.1 (Berkeley) 6/6/93"; #endif /* not lint */ #endif #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include "ktrace.h" void timevaladd(struct timeval *, struct timeval *); void timevalsub(struct timeval *, struct timeval *); void timevalfix(struct timeval *); int getpoints(char *s) { int facs = 0; while (*s) { switch(*s) { case 'c': facs |= KTRFAC_SYSCALL | KTRFAC_SYSRET; break; + case 'i': + facs |= KTRFAC_GENIO; + break; case 'n': facs |= KTRFAC_NAMEI; break; - case 'i': - facs |= KTRFAC_GENIO; + case 'p': + facs |= KTRFAC_CAPFAIL; break; case 's': facs |= KTRFAC_PSIG; break; case 't': facs |= KTRFAC_STRUCT; break; case 'u': facs |= KTRFAC_USER; break; case 'w': facs |= KTRFAC_CSW; break; case 'y': facs |= KTRFAC_SYSCTL; break; case '+': facs |= DEF_POINTS; break; default: return (-1); } s++; } return (facs); } void timevaladd(struct timeval *t1, struct timeval *t2) { t1->tv_sec += t2->tv_sec; t1->tv_usec += t2->tv_usec; timevalfix(t1); } void timevalsub(struct timeval *t1, struct timeval *t2) { t1->tv_sec -= t2->tv_sec; t1->tv_usec -= t2->tv_usec; timevalfix(t1); } void timevalfix(struct timeval *t1) { if (t1->tv_usec < 0) { t1->tv_sec--; t1->tv_usec += 1000000; } if (t1->tv_usec >= 1000000) { t1->tv_sec++; t1->tv_usec -= 1000000; } }