/* ARGSUSED */
int
sys_ktrace(struct proc *curp, void *v, register_t *retval)
{
struct sys_ktrace_args /* {
syscallarg(const char *) fname;
syscallarg(int) ops;
syscallarg(int) facs;
syscallarg(pid_t) pid;
} */ *uap = v;
struct vnode *vp = NULL;
struct proc *p = NULL;
struct pgrp *pg;
int facs = SCARG(uap, facs) & ~((unsigned) KTRFAC_ROOT);
int ops = KTROP(SCARG(uap, ops));
int descend = SCARG(uap, ops) & KTRFLAG_DESCEND;
int ret = 0;
int error = 0;
struct nameidata nd;
curp->p_traceflag |= KTRFAC_ACTIVE;
if (ops != KTROP_CLEAR) {
/*
* an operation which requires a file argument.
*/
NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, fname),
curp);
if ((error = vn_open(&nd, FREAD|FWRITE|O_NOFOLLOW, 0)) != 0) {
curp->p_traceflag &= ~KTRFAC_ACTIVE;
return (error);
}
vp = nd.ni_vp;
VOP_UNLOCK(vp, 0, curp);
if (vp->v_type != VREG) {
(void) vn_close(vp, FREAD|FWRITE, curp->p_ucred, curp);
curp->p_traceflag &= ~KTRFAC_ACTIVE;
return (EACCES);
}
}
/*
* Clear all uses of the tracefile
*/
if (ops == KTROP_CLEARFILE) {
for (p = LIST_FIRST(&allproc); p; p = LIST_NEXT(p, p_list)) {
if (p->p_tracep == vp) {
if (ktrcanset(curp, p)) {
p->p_traceflag = 0;
ktrsettracevnode(p, NULL);
} else
error = EPERM;
}
}
goto done;
}
/*
* need something to (un)trace (XXX - why is this here?)
*/
if (!facs) {
error = EINVAL;
goto done;
}
/*
* do it
*/
if (SCARG(uap, pid) < 0) {
/*
* by process group
*/
pg = pgfind(-SCARG(uap, pid));
if (pg == NULL) {
error = ESRCH;
goto done;
}
LIST_FOREACH(p, &pg->pg_members, p_pglist)
if (descend)
ret |= ktrsetchildren(curp, p, ops, facs, vp);
else
ret |= ktrops(curp, p, ops, facs, vp);
} else {
/* ARGSUSED */
int
sys_execve(struct proc *p, void *v, register_t *retval)
{
struct sys_execve_args /* {
syscallarg(const char *) path;
syscallarg(char *const *) argp;
syscallarg(char *const *) envp;
} */ *uap = v;
int error;
struct exec_package pack;
struct nameidata nid;
struct vattr attr;
struct ucred *cred = p->p_ucred;
char *argp;
char * const *cpp, *dp, *sp;
struct process *pr = p->p_p;
long argc, envc;
size_t len, sgap;
#ifdef MACHINE_STACK_GROWS_UP
size_t slen;
#endif
char *stack;
struct ps_strings arginfo;
struct vmspace *vm = p->p_vmspace;
char **tmpfap;
extern struct emul emul_native;
#if NSYSTRACE > 0
int wassugid = ISSET(pr->ps_flags, PS_SUGID | PS_SUGIDEXEC);
size_t pathbuflen;
#endif
char *pathbuf = NULL;
/*
* Cheap solution to complicated problems.
* Mark this process as "leave me alone, I'm execing".
*/
atomic_setbits_int(&p->p_flag, P_INEXEC);
#if NSYSTRACE > 0
if (ISSET(p->p_flag, P_SYSTRACE)) {
systrace_execve0(p);
pathbuf = pool_get(&namei_pool, PR_WAITOK);
error = copyinstr(SCARG(uap, path), pathbuf, MAXPATHLEN,
&pathbuflen);
if (error != 0)
goto clrflag;
}
#endif
if (pathbuf != NULL) {
NDINIT(&nid, LOOKUP, NOFOLLOW, UIO_SYSSPACE, pathbuf, p);
} else {
NDINIT(&nid, LOOKUP, NOFOLLOW, UIO_USERSPACE,
SCARG(uap, path), p);
}
/*
* initialize the fields of the exec package.
*/
if (pathbuf != NULL)
pack.ep_name = pathbuf;
else
pack.ep_name = (char *)SCARG(uap, path);
pack.ep_hdr = malloc(exec_maxhdrsz, M_EXEC, M_WAITOK);
pack.ep_hdrlen = exec_maxhdrsz;
pack.ep_hdrvalid = 0;
pack.ep_ndp = &nid;
pack.ep_interp = NULL;
pack.ep_emul_arg = NULL;
VMCMDSET_INIT(&pack.ep_vmcmds);
pack.ep_vap = &attr;
pack.ep_emul = &emul_native;
pack.ep_flags = 0;
/* see if we can run it. */
if ((error = check_exec(p, &pack)) != 0) {
goto freehdr;
}
/* XXX -- THE FOLLOWING SECTION NEEDS MAJOR CLEANUP */
/* allocate an argument buffer */
argp = (char *) uvm_km_valloc_wait(exec_map, NCARGS);
#ifdef DIAGNOSTIC
if (argp == NULL)
panic("execve: argp == NULL");
#endif
dp = argp;
argc = 0;
/* copy the fake args list, if there's one, freeing it as we go */
if (pack.ep_flags & EXEC_HASARGL) {
tmpfap = pack.ep_fa;
while (*tmpfap != NULL) {
char *cp;
cp = *tmpfap;
while (*cp)
*dp++ = *cp++;
*dp++ = '\0';
free(*tmpfap, M_EXEC);
tmpfap++; argc++;
}
free(pack.ep_fa, M_EXEC);
pack.ep_flags &= ~EXEC_HASARGL;
}
/* Now get argv & environment */
if (!(cpp = SCARG(uap, argp))) {
error = EFAULT;
goto bad;
}
if (pack.ep_flags & EXEC_SKIPARG)
cpp++;
while (1) {
len = argp + ARG_MAX - dp;
if ((error = copyin(cpp, &sp, sizeof(sp))) != 0)
goto bad;
if (!sp)
break;
if ((error = copyinstr(sp, dp, len, &len)) != 0) {
if (error == ENAMETOOLONG)
error = E2BIG;
goto bad;
}
dp += len;
cpp++;
argc++;
}
envc = 0;
/* environment does not need to be there */
if ((cpp = SCARG(uap, envp)) != NULL ) {
while (1) {
len = argp + ARG_MAX - dp;
if ((error = copyin(cpp, &sp, sizeof(sp))) != 0)
goto bad;
if (!sp)
break;
if ((error = copyinstr(sp, dp, len, &len)) != 0) {
if (error == ENAMETOOLONG)
error = E2BIG;
goto bad;
}
dp += len;
cpp++;
envc++;
}
}
dp = (char *)ALIGN(dp);
sgap = STACKGAPLEN;
if (stackgap_random != 0)
sgap += (arc4random() * ALIGNBYTES) & (stackgap_random - 1);
#ifdef MACHINE_STACK_GROWS_UP
sgap = ALIGN(sgap);
#endif
/* Now check if args & environ fit into new stack */
len = ((argc + envc + 2 + pack.ep_emul->e_arglen) * sizeof(char *) +
sizeof(long) + dp + sgap + sizeof(struct ps_strings)) - argp;
len = ALIGN(len); /* make the stack "safely" aligned */
if (len > pack.ep_ssize) { /* in effect, compare to initial limit */
error = ENOMEM;
goto bad;
}
/* adjust "active stack depth" for process VSZ */
pack.ep_ssize = len; /* maybe should go elsewhere, but... */
/*
* Prepare vmspace for remapping. Note that uvmspace_exec can replace
* p_vmspace!
*/
uvmspace_exec(p, VM_MIN_ADDRESS, VM_MAXUSER_ADDRESS);
vm = p->p_vmspace;
/* Now map address space */
vm->vm_taddr = (char *)pack.ep_taddr;
vm->vm_tsize = atop(round_page(pack.ep_tsize));
vm->vm_daddr = (char *)pack.ep_daddr;
vm->vm_dsize = atop(round_page(pack.ep_dsize));
vm->vm_dused = 0;
vm->vm_ssize = atop(round_page(pack.ep_ssize));
vm->vm_maxsaddr = (char *)pack.ep_maxsaddr;
vm->vm_minsaddr = (char *)pack.ep_minsaddr;
/* create the new process's VM space by running the vmcmds */
#ifdef DIAGNOSTIC
if (pack.ep_vmcmds.evs_used == 0)
panic("execve: no vmcmds");
#endif
error = exec_process_vmcmds(p, &pack);
/* if an error happened, deallocate and punt */
if (error)
goto exec_abort;
/* remember information about the process */
arginfo.ps_nargvstr = argc;
arginfo.ps_nenvstr = envc;
#ifdef MACHINE_STACK_GROWS_UP
stack = (char *)USRSTACK + sizeof(arginfo) + sgap;
slen = len - sizeof(arginfo) - sgap;
#else
stack = (char *)(USRSTACK - len);
#endif
/* Now copy argc, args & environ to new stack */
if (!(*pack.ep_emul->e_copyargs)(&pack, &arginfo, stack, argp))
goto exec_abort;
/* copy out the process's ps_strings structure */
if (copyout(&arginfo, (char *)PS_STRINGS, sizeof(arginfo)))
goto exec_abort;
stopprofclock(p); /* stop profiling */
fdcloseexec(p); /* handle close on exec */
execsigs(p); /* reset caught signals */
TCB_SET(p, NULL); /* reset the TCB address */
/* set command name & other accounting info */
len = min(nid.ni_cnd.cn_namelen, MAXCOMLEN);
bcopy(nid.ni_cnd.cn_nameptr, p->p_comm, len);
p->p_comm[len] = 0;
p->p_acflag &= ~AFORK;
/* record proc's vnode, for use by procfs and others */
if (p->p_textvp)
vrele(p->p_textvp);
vref(pack.ep_vp);
p->p_textvp = pack.ep_vp;
atomic_setbits_int(&pr->ps_flags, PS_EXEC);
if (pr->ps_flags & PS_PPWAIT) {
atomic_clearbits_int(&pr->ps_flags, PS_PPWAIT);
atomic_clearbits_int(&pr->ps_pptr->ps_flags, PS_ISPWAIT);
wakeup(pr->ps_pptr);
}
/*
* If process does execve() while it has a mismatched real,
* effective, or saved uid/gid, we set PS_SUGIDEXEC.
*/
if (p->p_ucred->cr_uid != p->p_cred->p_ruid ||
p->p_ucred->cr_uid != p->p_cred->p_svuid ||
p->p_ucred->cr_gid != p->p_cred->p_rgid ||
p->p_ucred->cr_gid != p->p_cred->p_svgid)
atomic_setbits_int(&pr->ps_flags, PS_SUGIDEXEC);
else
atomic_clearbits_int(&pr->ps_flags, PS_SUGIDEXEC);
/*
* deal with set[ug]id.
* MNT_NOEXEC has already been used to disable s[ug]id.
*/
if ((attr.va_mode & (VSUID | VSGID)) && proc_cansugid(p)) {
int i;
atomic_setbits_int(&pr->ps_flags, PS_SUGID|PS_SUGIDEXEC);
#ifdef KTRACE
/*
* If process is being ktraced, turn off - unless
* root set it.
*/
if (p->p_tracep && !(p->p_traceflag & KTRFAC_ROOT)) {
p->p_traceflag = 0;
ktrsettracevnode(p, NULL);
}
#endif
p->p_ucred = crcopy(cred);
if (attr.va_mode & VSUID)
p->p_ucred->cr_uid = attr.va_uid;
if (attr.va_mode & VSGID)
p->p_ucred->cr_gid = attr.va_gid;
/*
* For set[ug]id processes, a few caveats apply to
* stdin, stdout, and stderr.
*/
for (i = 0; i < 3; i++) {
struct file *fp = NULL;
/*
* NOTE - This will never return NULL because of
* immature fds. The file descriptor table is not
* shared because we're suid.
*/
fp = fd_getfile(p->p_fd, i);
#ifdef PROCFS
/*
* Close descriptors that are writing to procfs.
*/
if (fp && fp->f_type == DTYPE_VNODE &&
((struct vnode *)(fp->f_data))->v_tag == VT_PROCFS &&
(fp->f_flag & FWRITE)) {
fdrelease(p, i);
fp = NULL;
}
#endif
/*
* Ensure that stdin, stdout, and stderr are already
* allocated. We do not want userland to accidentally
* allocate descriptors in this range which has implied
* meaning to libc.
*/
if (fp == NULL) {
short flags = FREAD | (i == 0 ? 0 : FWRITE);
struct vnode *vp;
int indx;
if ((error = falloc(p, &fp, &indx)) != 0)
goto exec_abort;
#ifdef DIAGNOSTIC
if (indx != i)
panic("sys_execve: falloc indx != i");
#endif
if ((error = cdevvp(getnulldev(), &vp)) != 0) {
fdremove(p->p_fd, indx);
closef(fp, p);
goto exec_abort;
}
if ((error = VOP_OPEN(vp, flags, p->p_ucred, p)) != 0) {
fdremove(p->p_fd, indx);
closef(fp, p);
vrele(vp);
goto exec_abort;
}
if (flags & FWRITE)
vp->v_writecount++;
fp->f_flag = flags;
fp->f_type = DTYPE_VNODE;
fp->f_ops = &vnops;
fp->f_data = (caddr_t)vp;
FILE_SET_MATURE(fp);
}
}
} else
atomic_clearbits_int(&pr->ps_flags, PS_SUGID);
p->p_cred->p_svuid = p->p_ucred->cr_uid;
p->p_cred->p_svgid = p->p_ucred->cr_gid;
if (pr->ps_flags & PS_SUGIDEXEC) {
int i, s = splclock();
timeout_del(&p->p_realit_to);
timerclear(&p->p_realtimer.it_interval);
timerclear(&p->p_realtimer.it_value);
for (i = 0; i < sizeof(p->p_stats->p_timer) /
sizeof(p->p_stats->p_timer[0]); i++) {
timerclear(&p->p_stats->p_timer[i].it_interval);
timerclear(&p->p_stats->p_timer[i].it_value);
}
splx(s);
}
uvm_km_free_wakeup(exec_map, (vaddr_t) argp, NCARGS);
pool_put(&namei_pool, nid.ni_cnd.cn_pnbuf);
vn_close(pack.ep_vp, FREAD, cred, p);
/*
* notify others that we exec'd
*/
KNOTE(&pr->ps_klist, NOTE_EXEC);
/* setup new registers and do misc. setup. */
if (pack.ep_emul->e_fixup != NULL) {
if ((*pack.ep_emul->e_fixup)(p, &pack) != 0)
goto free_pack_abort;
}
#ifdef MACHINE_STACK_GROWS_UP
(*pack.ep_emul->e_setregs)(p, &pack, (u_long)stack + slen, retval);
#else
(*pack.ep_emul->e_setregs)(p, &pack, (u_long)stack, retval);
#endif
/* map the process's signal trampoline code */
if (exec_sigcode_map(p, pack.ep_emul))
goto free_pack_abort;
#ifdef __HAVE_EXEC_MD_MAP
/* perform md specific mappings that process might need */
if (exec_md_map(p, &pack))
goto free_pack_abort;
#endif
if (p->p_flag & P_TRACED)
psignal(p, SIGTRAP);
free(pack.ep_hdr, M_EXEC);
/*
* Call emulation specific exec hook. This can setup per-process
* p->p_emuldata or do any other per-process stuff an emulation needs.
*
* If we are executing process of different emulation than the
* original forked process, call e_proc_exit() of the old emulation
* first, then e_proc_exec() of new emulation. If the emulation is
* same, the exec hook code should deallocate any old emulation
* resources held previously by this process.
*/
if (p->p_emul && p->p_emul->e_proc_exit &&
p->p_emul != pack.ep_emul)
(*p->p_emul->e_proc_exit)(p);
p->p_descfd = 255;
if ((pack.ep_flags & EXEC_HASFD) && pack.ep_fd < 255)
p->p_descfd = pack.ep_fd;
/*
* Call exec hook. Emulation code may NOT store reference to anything
* from &pack.
*/
if (pack.ep_emul->e_proc_exec)
(*pack.ep_emul->e_proc_exec)(p, &pack);
/* update p_emul, the old value is no longer needed */
p->p_emul = pack.ep_emul;
#ifdef KTRACE
if (KTRPOINT(p, KTR_EMUL))
ktremul(p, p->p_emul->e_name);
#endif
atomic_clearbits_int(&p->p_flag, P_INEXEC);
#if NSYSTRACE > 0
if (ISSET(p->p_flag, P_SYSTRACE) &&
wassugid && !ISSET(pr->ps_flags, PS_SUGID | PS_SUGIDEXEC))
systrace_execve1(pathbuf, p);
#endif
if (pathbuf != NULL)
pool_put(&namei_pool, pathbuf);
return (0);
bad:
/* free the vmspace-creation commands, and release their references */
kill_vmcmds(&pack.ep_vmcmds);
/* kill any opened file descriptor, if necessary */
if (pack.ep_flags & EXEC_HASFD) {
pack.ep_flags &= ~EXEC_HASFD;
(void) fdrelease(p, pack.ep_fd);
}
if (pack.ep_interp != NULL)
pool_put(&namei_pool, pack.ep_interp);
if (pack.ep_emul_arg != NULL)
free(pack.ep_emul_arg, M_TEMP);
/* close and put the exec'd file */
vn_close(pack.ep_vp, FREAD, cred, p);
pool_put(&namei_pool, nid.ni_cnd.cn_pnbuf);
uvm_km_free_wakeup(exec_map, (vaddr_t) argp, NCARGS);
freehdr:
free(pack.ep_hdr, M_EXEC);
#if NSYSTRACE > 0
clrflag:
#endif
atomic_clearbits_int(&p->p_flag, P_INEXEC);
if (pathbuf != NULL)
pool_put(&namei_pool, pathbuf);
return (error);
exec_abort:
/*
* the old process doesn't exist anymore. exit gracefully.
* get rid of the (new) address space we have created, if any, get rid
* of our namei data and vnode, and exit noting failure
*/
uvm_deallocate(&vm->vm_map, VM_MIN_ADDRESS,
VM_MAXUSER_ADDRESS - VM_MIN_ADDRESS);
if (pack.ep_interp != NULL)
pool_put(&namei_pool, pack.ep_interp);
if (pack.ep_emul_arg != NULL)
free(pack.ep_emul_arg, M_TEMP);
pool_put(&namei_pool, nid.ni_cnd.cn_pnbuf);
vn_close(pack.ep_vp, FREAD, cred, p);
uvm_km_free_wakeup(exec_map, (vaddr_t) argp, NCARGS);
free_pack_abort:
free(pack.ep_hdr, M_EXEC);
exit1(p, W_EXITCODE(0, SIGABRT), EXIT_NORMAL);
/* NOTREACHED */
atomic_clearbits_int(&p->p_flag, P_INEXEC);
if (pathbuf != NULL)
pool_put(&namei_pool, pathbuf);
return (0);
}
