The Design and Implementation of the FreeBSD Operating System, Second Edition
Now available: The Design and Implementation of the FreeBSD Operating System (Second Edition)


[ source navigation ] [ diff markup ] [ identifier search ] [ freetext search ] [ file search ] [ list types ] [ track identifier ]

FreeBSD/Linux Kernel Cross Reference
sys/compat/linux/linux_misc.c

Version: -  FREEBSD  -  FREEBSD-13-STABLE  -  FREEBSD-13-0  -  FREEBSD-12-STABLE  -  FREEBSD-12-0  -  FREEBSD-11-STABLE  -  FREEBSD-11-0  -  FREEBSD-10-STABLE  -  FREEBSD-10-0  -  FREEBSD-9-STABLE  -  FREEBSD-9-0  -  FREEBSD-8-STABLE  -  FREEBSD-8-0  -  FREEBSD-7-STABLE  -  FREEBSD-7-0  -  FREEBSD-6-STABLE  -  FREEBSD-6-0  -  FREEBSD-5-STABLE  -  FREEBSD-5-0  -  FREEBSD-4-STABLE  -  FREEBSD-3-STABLE  -  FREEBSD22  -  l41  -  OPENBSD  -  linux-2.6  -  MK84  -  PLAN9  -  xnu-8792 
SearchContext: -  none  -  3  -  10 

    1 /*-
    2  * Copyright (c) 2002 Doug Rabson
    3  * Copyright (c) 1994-1995 Søren Schmidt
    4  * All rights reserved.
    5  *
    6  * Redistribution and use in source and binary forms, with or without
    7  * modification, are permitted provided that the following conditions
    8  * are met:
    9  * 1. Redistributions of source code must retain the above copyright
   10  *    notice, this list of conditions and the following disclaimer
   11  *    in this position and unchanged.
   12  * 2. Redistributions in binary form must reproduce the above copyright
   13  *    notice, this list of conditions and the following disclaimer in the
   14  *    documentation and/or other materials provided with the distribution.
   15  * 3. The name of the author may not be used to endorse or promote products
   16  *    derived from this software without specific prior written permission
   17  *
   18  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
   19  * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
   20  * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
   21  * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
   22  * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
   23  * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
   24  * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
   25  * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
   26  * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
   27  * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
   28  */
   29 
   30 #include <sys/cdefs.h>
   31 __FBSDID("$FreeBSD$");
   32 
   33 #include "opt_compat.h"
   34 #include "opt_mac.h"
   35 
   36 #include <sys/param.h>
   37 #include <sys/blist.h>
   38 #include <sys/fcntl.h>
   39 #if defined(__i386__)
   40 #include <sys/imgact_aout.h>
   41 #endif
   42 #include <sys/jail.h>
   43 #include <sys/kernel.h>
   44 #include <sys/limits.h>
   45 #include <sys/lock.h>
   46 #include <sys/malloc.h>
   47 #include <sys/mman.h>
   48 #include <sys/mount.h>
   49 #include <sys/mutex.h>
   50 #include <sys/namei.h>
   51 #include <sys/priv.h>
   52 #include <sys/proc.h>
   53 #include <sys/reboot.h>
   54 #include <sys/resourcevar.h>
   55 #include <sys/sched.h>
   56 #include <sys/signalvar.h>
   57 #include <sys/stat.h>
   58 #include <sys/syscallsubr.h>
   59 #include <sys/sysctl.h>
   60 #include <sys/sysproto.h>
   61 #include <sys/systm.h>
   62 #include <sys/time.h>
   63 #include <sys/vmmeter.h>
   64 #include <sys/vnode.h>
   65 #include <sys/wait.h>
   66 #include <sys/cpuset.h>
   67 
   68 #include <security/mac/mac_framework.h>
   69 
   70 #include <vm/vm.h>
   71 #include <vm/pmap.h>
   72 #include <vm/vm_kern.h>
   73 #include <vm/vm_map.h>
   74 #include <vm/vm_extern.h>
   75 #include <vm/vm_object.h>
   76 #include <vm/swap_pager.h>
   77 
   78 #include <compat/linux/linux_sysproto.h>
   79 #include <compat/linux/linux_emul.h>
   80 #include <compat/linux/linux_misc.h>
   81 
   82 #ifdef COMPAT_LINUX32
   83 #include <machine/../linux32/linux.h>
   84 #include <machine/../linux32/linux32_proto.h>
   85 #else
   86 #include <machine/../linux/linux.h>
   87 #include <machine/../linux/linux_proto.h>
   88 #endif
   89 
   90 #include <compat/linux/linux_mib.h>
   91 #include <compat/linux/linux_signal.h>
   92 #include <compat/linux/linux_util.h>
   93 
   94 #define BSD_TO_LINUX_SIGNAL(sig)        \
   95         (((sig) <= LINUX_SIGTBLSZ) ? bsd_to_linux_signal[_SIG_IDX(sig)] : sig)
   96 
   97 static unsigned int linux_to_bsd_resource[LINUX_RLIM_NLIMITS] = {
   98         RLIMIT_CPU, RLIMIT_FSIZE, RLIMIT_DATA, RLIMIT_STACK,
   99         RLIMIT_CORE, RLIMIT_RSS, RLIMIT_NPROC, RLIMIT_NOFILE,
  100         RLIMIT_MEMLOCK, RLIMIT_AS 
  101 };
  102 
  103 struct l_sysinfo {
  104         l_long          uptime;         /* Seconds since boot */
  105         l_ulong         loads[3];       /* 1, 5, and 15 minute load averages */
  106 #define LINUX_SYSINFO_LOADS_SCALE 65536
  107         l_ulong         totalram;       /* Total usable main memory size */
  108         l_ulong         freeram;        /* Available memory size */
  109         l_ulong         sharedram;      /* Amount of shared memory */
  110         l_ulong         bufferram;      /* Memory used by buffers */
  111         l_ulong         totalswap;      /* Total swap space size */
  112         l_ulong         freeswap;       /* swap space still available */
  113         l_ushort        procs;          /* Number of current processes */
  114         l_ushort        pads;
  115         l_ulong         totalbig;
  116         l_ulong         freebig;
  117         l_uint          mem_unit;
  118         char            _f[20-2*sizeof(l_long)-sizeof(l_int)];  /* padding */
  119 };
  120 int
  121 linux_sysinfo(struct thread *td, struct linux_sysinfo_args *args)
  122 {
  123         struct l_sysinfo sysinfo;
  124         vm_object_t object;
  125         int i, j;
  126         struct timespec ts;
  127 
  128         getnanouptime(&ts);
  129         if (ts.tv_nsec != 0)
  130                 ts.tv_sec++;
  131         sysinfo.uptime = ts.tv_sec;
  132 
  133         /* Use the information from the mib to get our load averages */
  134         for (i = 0; i < 3; i++)
  135                 sysinfo.loads[i] = averunnable.ldavg[i] *
  136                     LINUX_SYSINFO_LOADS_SCALE / averunnable.fscale;
  137 
  138         sysinfo.totalram = physmem * PAGE_SIZE;
  139         sysinfo.freeram = sysinfo.totalram - cnt.v_wire_count * PAGE_SIZE;
  140 
  141         sysinfo.sharedram = 0;
  142         mtx_lock(&vm_object_list_mtx);
  143         TAILQ_FOREACH(object, &vm_object_list, object_list)
  144                 if (object->shadow_count > 1)
  145                         sysinfo.sharedram += object->resident_page_count;
  146         mtx_unlock(&vm_object_list_mtx);
  147 
  148         sysinfo.sharedram *= PAGE_SIZE;
  149         sysinfo.bufferram = 0;
  150 
  151         swap_pager_status(&i, &j);
  152         sysinfo.totalswap = i * PAGE_SIZE;
  153         sysinfo.freeswap = (i - j) * PAGE_SIZE;
  154 
  155         sysinfo.procs = nprocs;
  156 
  157         /* The following are only present in newer Linux kernels. */
  158         sysinfo.totalbig = 0;
  159         sysinfo.freebig = 0;
  160         sysinfo.mem_unit = 1;
  161 
  162         return copyout(&sysinfo, args->info, sizeof(sysinfo));
  163 }
  164 
  165 int
  166 linux_alarm(struct thread *td, struct linux_alarm_args *args)
  167 {
  168         struct itimerval it, old_it;
  169         int error;
  170 
  171 #ifdef DEBUG
  172         if (ldebug(alarm))
  173                 printf(ARGS(alarm, "%u"), args->secs);
  174 #endif
  175 
  176         if (args->secs > 100000000)
  177                 return (EINVAL);
  178 
  179         it.it_value.tv_sec = (long)args->secs;
  180         it.it_value.tv_usec = 0;
  181         it.it_interval.tv_sec = 0;
  182         it.it_interval.tv_usec = 0;
  183         error = kern_setitimer(td, ITIMER_REAL, &it, &old_it);
  184         if (error)
  185                 return (error);
  186         if (timevalisset(&old_it.it_value)) {
  187                 if (old_it.it_value.tv_usec != 0)
  188                         old_it.it_value.tv_sec++;
  189                 td->td_retval[0] = old_it.it_value.tv_sec;
  190         }
  191         return (0);
  192 }
  193 
  194 int
  195 linux_brk(struct thread *td, struct linux_brk_args *args)
  196 {
  197         struct vmspace *vm = td->td_proc->p_vmspace;
  198         vm_offset_t new, old;
  199         struct obreak_args /* {
  200                 char * nsize;
  201         } */ tmp;
  202 
  203 #ifdef DEBUG
  204         if (ldebug(brk))
  205                 printf(ARGS(brk, "%p"), (void *)(uintptr_t)args->dsend);
  206 #endif
  207         old = (vm_offset_t)vm->vm_daddr + ctob(vm->vm_dsize);
  208         new = (vm_offset_t)args->dsend;
  209         tmp.nsize = (char *)new;
  210         if (((caddr_t)new > vm->vm_daddr) && !obreak(td, &tmp))
  211                 td->td_retval[0] = (long)new;
  212         else
  213                 td->td_retval[0] = (long)old;
  214 
  215         return 0;
  216 }
  217 
  218 #if defined(__i386__)
  219 /* XXX: what about amd64/linux32? */
  220 
  221 int
  222 linux_uselib(struct thread *td, struct linux_uselib_args *args)
  223 {
  224         struct nameidata ni;
  225         struct vnode *vp;
  226         struct exec *a_out;
  227         struct vattr attr;
  228         vm_offset_t vmaddr;
  229         unsigned long file_offset;
  230         vm_offset_t buffer;
  231         unsigned long bss_size;
  232         char *library;
  233         int error;
  234         int locked, vfslocked;
  235 
  236         LCONVPATHEXIST(td, args->library, &library);
  237 
  238 #ifdef DEBUG
  239         if (ldebug(uselib))
  240                 printf(ARGS(uselib, "%s"), library);
  241 #endif
  242 
  243         a_out = NULL;
  244         vfslocked = 0;
  245         locked = 0;
  246         vp = NULL;
  247 
  248         NDINIT(&ni, LOOKUP, ISOPEN | FOLLOW | LOCKLEAF | MPSAFE | AUDITVNODE1,
  249             UIO_SYSSPACE, library, td);
  250         error = namei(&ni);
  251         LFREEPATH(library);
  252         if (error)
  253                 goto cleanup;
  254 
  255         vp = ni.ni_vp;
  256         vfslocked = NDHASGIANT(&ni);
  257         NDFREE(&ni, NDF_ONLY_PNBUF);
  258 
  259         /*
  260          * From here on down, we have a locked vnode that must be unlocked.
  261          * XXX: The code below largely duplicates exec_check_permissions().
  262          */
  263         locked = 1;
  264 
  265         /* Writable? */
  266         if (vp->v_writecount) {
  267                 error = ETXTBSY;
  268                 goto cleanup;
  269         }
  270 
  271         /* Executable? */
  272         error = VOP_GETATTR(vp, &attr, td->td_ucred, td);
  273         if (error)
  274                 goto cleanup;
  275 
  276         if ((vp->v_mount->mnt_flag & MNT_NOEXEC) ||
  277             ((attr.va_mode & 0111) == 0) || (attr.va_type != VREG)) {
  278                 /* EACCESS is what exec(2) returns. */
  279                 error = ENOEXEC;
  280                 goto cleanup;
  281         }
  282 
  283         /* Sensible size? */
  284         if (attr.va_size == 0) {
  285                 error = ENOEXEC;
  286                 goto cleanup;
  287         }
  288 
  289         /* Can we access it? */
  290         error = VOP_ACCESS(vp, VEXEC, td->td_ucred, td);
  291         if (error)
  292                 goto cleanup;
  293 
  294         /*
  295          * XXX: This should use vn_open() so that it is properly authorized,
  296          * and to reduce code redundancy all over the place here.
  297          * XXX: Not really, it duplicates far more of exec_check_permissions()
  298          * than vn_open().
  299          */
  300 #ifdef MAC
  301         error = mac_check_vnode_open(td->td_ucred, vp, FREAD);
  302         if (error)
  303                 goto cleanup;
  304 #endif
  305         error = VOP_OPEN(vp, FREAD, td->td_ucred, td, NULL);
  306         if (error)
  307                 goto cleanup;
  308 
  309         /* Pull in executable header into kernel_map */
  310         error = vm_mmap(kernel_map, (vm_offset_t *)&a_out, PAGE_SIZE,
  311             VM_PROT_READ, VM_PROT_READ, 0, OBJT_VNODE, vp, 0);
  312         if (error)
  313                 goto cleanup;
  314 
  315         /* Is it a Linux binary ? */
  316         if (((a_out->a_magic >> 16) & 0xff) != 0x64) {
  317                 error = ENOEXEC;
  318                 goto cleanup;
  319         }
  320 
  321         /*
  322          * While we are here, we should REALLY do some more checks
  323          */
  324 
  325         /* Set file/virtual offset based on a.out variant. */
  326         switch ((int)(a_out->a_magic & 0xffff)) {
  327         case 0413:                      /* ZMAGIC */
  328                 file_offset = 1024;
  329                 break;
  330         case 0314:                      /* QMAGIC */
  331                 file_offset = 0;
  332                 break;
  333         default:
  334                 error = ENOEXEC;
  335                 goto cleanup;
  336         }
  337 
  338         bss_size = round_page(a_out->a_bss);
  339 
  340         /* Check various fields in header for validity/bounds. */
  341         if (a_out->a_text & PAGE_MASK || a_out->a_data & PAGE_MASK) {
  342                 error = ENOEXEC;
  343                 goto cleanup;
  344         }
  345 
  346         /* text + data can't exceed file size */
  347         if (a_out->a_data + a_out->a_text > attr.va_size) {
  348                 error = EFAULT;
  349                 goto cleanup;
  350         }
  351 
  352         /*
  353          * text/data/bss must not exceed limits
  354          * XXX - this is not complete. it should check current usage PLUS
  355          * the resources needed by this library.
  356          */
  357         PROC_LOCK(td->td_proc);
  358         if (a_out->a_text > maxtsiz ||
  359             a_out->a_data + bss_size > lim_cur(td->td_proc, RLIMIT_DATA)) {
  360                 PROC_UNLOCK(td->td_proc);
  361                 error = ENOMEM;
  362                 goto cleanup;
  363         }
  364         PROC_UNLOCK(td->td_proc);
  365 
  366         /*
  367          * Prevent more writers.
  368          * XXX: Note that if any of the VM operations fail below we don't
  369          * clear this flag.
  370          */
  371         vp->v_vflag |= VV_TEXT;
  372 
  373         /*
  374          * Lock no longer needed
  375          */
  376         locked = 0;
  377         VOP_UNLOCK(vp, 0, td);
  378         VFS_UNLOCK_GIANT(vfslocked);
  379 
  380         /*
  381          * Check if file_offset page aligned. Currently we cannot handle
  382          * misalinged file offsets, and so we read in the entire image
  383          * (what a waste).
  384          */
  385         if (file_offset & PAGE_MASK) {
  386 #ifdef DEBUG
  387                 printf("uselib: Non page aligned binary %lu\n", file_offset);
  388 #endif
  389                 /* Map text+data read/write/execute */
  390 
  391                 /* a_entry is the load address and is page aligned */
  392                 vmaddr = trunc_page(a_out->a_entry);
  393 
  394                 /* get anon user mapping, read+write+execute */
  395                 error = vm_map_find(&td->td_proc->p_vmspace->vm_map, NULL, 0,
  396                     &vmaddr, a_out->a_text + a_out->a_data, FALSE, VM_PROT_ALL,
  397                     VM_PROT_ALL, 0);
  398                 if (error)
  399                         goto cleanup;
  400 
  401                 /* map file into kernel_map */
  402                 error = vm_mmap(kernel_map, &buffer,
  403                     round_page(a_out->a_text + a_out->a_data + file_offset),
  404                     VM_PROT_READ, VM_PROT_READ, 0, OBJT_VNODE, vp,
  405                     trunc_page(file_offset));
  406                 if (error)
  407                         goto cleanup;
  408 
  409                 /* copy from kernel VM space to user space */
  410                 error = copyout(PTRIN(buffer + file_offset),
  411                     (void *)vmaddr, a_out->a_text + a_out->a_data);
  412 
  413                 /* release temporary kernel space */
  414                 vm_map_remove(kernel_map, buffer, buffer +
  415                     round_page(a_out->a_text + a_out->a_data + file_offset));
  416 
  417                 if (error)
  418                         goto cleanup;
  419         } else {
  420 #ifdef DEBUG
  421                 printf("uselib: Page aligned binary %lu\n", file_offset);
  422 #endif
  423                 /*
  424                  * for QMAGIC, a_entry is 20 bytes beyond the load address
  425                  * to skip the executable header
  426                  */
  427                 vmaddr = trunc_page(a_out->a_entry);
  428 
  429                 /*
  430                  * Map it all into the process's space as a single
  431                  * copy-on-write "data" segment.
  432                  */
  433                 error = vm_mmap(&td->td_proc->p_vmspace->vm_map, &vmaddr,
  434                     a_out->a_text + a_out->a_data, VM_PROT_ALL, VM_PROT_ALL,
  435                     MAP_PRIVATE | MAP_FIXED, OBJT_VNODE, vp, file_offset);
  436                 if (error)
  437                         goto cleanup;
  438         }
  439 #ifdef DEBUG
  440         printf("mem=%08lx = %08lx %08lx\n", (long)vmaddr, ((long *)vmaddr)[0],
  441             ((long *)vmaddr)[1]);
  442 #endif
  443         if (bss_size != 0) {
  444                 /* Calculate BSS start address */
  445                 vmaddr = trunc_page(a_out->a_entry) + a_out->a_text +
  446                     a_out->a_data;
  447 
  448                 /* allocate some 'anon' space */
  449                 error = vm_map_find(&td->td_proc->p_vmspace->vm_map, NULL, 0,
  450                     &vmaddr, bss_size, FALSE, VM_PROT_ALL, VM_PROT_ALL, 0);
  451                 if (error)
  452                         goto cleanup;
  453         }
  454 
  455 cleanup:
  456         /* Unlock vnode if needed */
  457         if (locked) {
  458                 VOP_UNLOCK(vp, 0, td);
  459                 VFS_UNLOCK_GIANT(vfslocked);
  460         }
  461 
  462         /* Release the kernel mapping. */
  463         if (a_out)
  464                 vm_map_remove(kernel_map, (vm_offset_t)a_out,
  465                     (vm_offset_t)a_out + PAGE_SIZE);
  466 
  467         return error;
  468 }
  469 
  470 #endif  /* __i386__ */
  471 
  472 int
  473 linux_select(struct thread *td, struct linux_select_args *args)
  474 {
  475         l_timeval ltv;
  476         struct timeval tv0, tv1, utv, *tvp;
  477         int error;
  478 
  479 #ifdef DEBUG
  480         if (ldebug(select))
  481                 printf(ARGS(select, "%d, %p, %p, %p, %p"), args->nfds,
  482                     (void *)args->readfds, (void *)args->writefds,
  483                     (void *)args->exceptfds, (void *)args->timeout);
  484 #endif
  485 
  486         /*
  487          * Store current time for computation of the amount of
  488          * time left.
  489          */
  490         if (args->timeout) {
  491                 if ((error = copyin(args->timeout, &ltv, sizeof(ltv))))
  492                         goto select_out;
  493                 utv.tv_sec = ltv.tv_sec;
  494                 utv.tv_usec = ltv.tv_usec;
  495 #ifdef DEBUG
  496                 if (ldebug(select))
  497                         printf(LMSG("incoming timeout (%jd/%ld)"),
  498                             (intmax_t)utv.tv_sec, utv.tv_usec);
  499 #endif
  500 
  501                 if (itimerfix(&utv)) {
  502                         /*
  503                          * The timeval was invalid.  Convert it to something
  504                          * valid that will act as it does under Linux.
  505                          */
  506                         utv.tv_sec += utv.tv_usec / 1000000;
  507                         utv.tv_usec %= 1000000;
  508                         if (utv.tv_usec < 0) {
  509                                 utv.tv_sec -= 1;
  510                                 utv.tv_usec += 1000000;
  511                         }
  512                         if (utv.tv_sec < 0)
  513                                 timevalclear(&utv);
  514                 }
  515                 microtime(&tv0);
  516                 tvp = &utv;
  517         } else
  518                 tvp = NULL;
  519 
  520         error = kern_select(td, args->nfds, args->readfds, args->writefds,
  521             args->exceptfds, tvp);
  522 
  523 #ifdef DEBUG
  524         if (ldebug(select))
  525                 printf(LMSG("real select returns %d"), error);
  526 #endif
  527         if (error) {
  528                 /*
  529                  * See fs/select.c in the Linux kernel.  Without this,
  530                  * Maelstrom doesn't work.
  531                  */
  532                 if (error == ERESTART)
  533                         error = EINTR;
  534                 goto select_out;
  535         }
  536 
  537         if (args->timeout) {
  538                 if (td->td_retval[0]) {
  539                         /*
  540                          * Compute how much time was left of the timeout,
  541                          * by subtracting the current time and the time
  542                          * before we started the call, and subtracting
  543                          * that result from the user-supplied value.
  544                          */
  545                         microtime(&tv1);
  546                         timevalsub(&tv1, &tv0);
  547                         timevalsub(&utv, &tv1);
  548                         if (utv.tv_sec < 0)
  549                                 timevalclear(&utv);
  550                 } else
  551                         timevalclear(&utv);
  552 #ifdef DEBUG
  553                 if (ldebug(select))
  554                         printf(LMSG("outgoing timeout (%jd/%ld)"),
  555                             (intmax_t)utv.tv_sec, utv.tv_usec);
  556 #endif
  557                 ltv.tv_sec = utv.tv_sec;
  558                 ltv.tv_usec = utv.tv_usec;
  559                 if ((error = copyout(&ltv, args->timeout, sizeof(ltv))))
  560                         goto select_out;
  561         }
  562 
  563 select_out:
  564 #ifdef DEBUG
  565         if (ldebug(select))
  566                 printf(LMSG("select_out -> %d"), error);
  567 #endif
  568         return error;
  569 }
  570 
  571 int
  572 linux_mremap(struct thread *td, struct linux_mremap_args *args)
  573 {
  574         struct munmap_args /* {
  575                 void *addr;
  576                 size_t len;
  577         } */ bsd_args;
  578         int error = 0;
  579 
  580 #ifdef DEBUG
  581         if (ldebug(mremap))
  582                 printf(ARGS(mremap, "%p, %08lx, %08lx, %08lx"),
  583                     (void *)(uintptr_t)args->addr,
  584                     (unsigned long)args->old_len,
  585                     (unsigned long)args->new_len,
  586                     (unsigned long)args->flags);
  587 #endif
  588 
  589         if (args->flags & ~(LINUX_MREMAP_FIXED | LINUX_MREMAP_MAYMOVE)) {
  590                 td->td_retval[0] = 0;
  591                 return (EINVAL);
  592         }
  593 
  594         /*
  595          * Check for the page alignment.
  596          * Linux defines PAGE_MASK to be FreeBSD ~PAGE_MASK.
  597          */
  598         if (args->addr & PAGE_MASK) {
  599                 td->td_retval[0] = 0;
  600                 return (EINVAL);
  601         }
  602 
  603         args->new_len = round_page(args->new_len);
  604         args->old_len = round_page(args->old_len);
  605 
  606         if (args->new_len > args->old_len) {
  607                 td->td_retval[0] = 0;
  608                 return ENOMEM;
  609         }
  610 
  611         if (args->new_len < args->old_len) {
  612                 bsd_args.addr =
  613                     (caddr_t)((uintptr_t)args->addr + args->new_len);
  614                 bsd_args.len = args->old_len - args->new_len;
  615                 error = munmap(td, &bsd_args);
  616         }
  617 
  618         td->td_retval[0] = error ? 0 : (uintptr_t)args->addr;
  619         return error;
  620 }
  621 
  622 #define LINUX_MS_ASYNC       0x0001
  623 #define LINUX_MS_INVALIDATE  0x0002
  624 #define LINUX_MS_SYNC        0x0004
  625 
  626 int
  627 linux_msync(struct thread *td, struct linux_msync_args *args)
  628 {
  629         struct msync_args bsd_args;
  630 
  631         bsd_args.addr = (caddr_t)(uintptr_t)args->addr;
  632         bsd_args.len = (uintptr_t)args->len;
  633         bsd_args.flags = args->fl & ~LINUX_MS_SYNC;
  634 
  635         return msync(td, &bsd_args);
  636 }
  637 
  638 int
  639 linux_time(struct thread *td, struct linux_time_args *args)
  640 {
  641         struct timeval tv;
  642         l_time_t tm;
  643         int error;
  644 
  645 #ifdef DEBUG
  646         if (ldebug(time))
  647                 printf(ARGS(time, "*"));
  648 #endif
  649 
  650         microtime(&tv);
  651         tm = tv.tv_sec;
  652         if (args->tm && (error = copyout(&tm, args->tm, sizeof(tm))))
  653                 return error;
  654         td->td_retval[0] = tm;
  655         return 0;
  656 }
  657 
  658 struct l_times_argv {
  659         l_long  tms_utime;
  660         l_long  tms_stime;
  661         l_long  tms_cutime;
  662         l_long  tms_cstime;
  663 };
  664 
  665 #define CLK_TCK 100                     /* Linux uses 100 */
  666 
  667 #define CONVTCK(r)      (r.tv_sec * CLK_TCK + r.tv_usec / (1000000 / CLK_TCK))
  668 
  669 int
  670 linux_times(struct thread *td, struct linux_times_args *args)
  671 {
  672         struct timeval tv, utime, stime, cutime, cstime;
  673         struct l_times_argv tms;
  674         struct proc *p;
  675         int error;
  676 
  677 #ifdef DEBUG
  678         if (ldebug(times))
  679                 printf(ARGS(times, "*"));
  680 #endif
  681 
  682         if (args->buf != NULL) {
  683                 p = td->td_proc;
  684                 PROC_LOCK(p);
  685                 PROC_SLOCK(p);
  686                 calcru(p, &utime, &stime);
  687                 PROC_SUNLOCK(p);
  688                 calccru(p, &cutime, &cstime);
  689                 PROC_UNLOCK(p);
  690 
  691                 tms.tms_utime = CONVTCK(utime);
  692                 tms.tms_stime = CONVTCK(stime);
  693 
  694                 tms.tms_cutime = CONVTCK(cutime);
  695                 tms.tms_cstime = CONVTCK(cstime);
  696 
  697                 if ((error = copyout(&tms, args->buf, sizeof(tms))))
  698                         return error;
  699         }
  700 
  701         microuptime(&tv);
  702         td->td_retval[0] = (int)CONVTCK(tv);
  703         return 0;
  704 }
  705 
  706 int
  707 linux_newuname(struct thread *td, struct linux_newuname_args *args)
  708 {
  709         struct l_new_utsname utsname;
  710         char osname[LINUX_MAX_UTSNAME];
  711         char osrelease[LINUX_MAX_UTSNAME];
  712         char *p;
  713 
  714 #ifdef DEBUG
  715         if (ldebug(newuname))
  716                 printf(ARGS(newuname, "*"));
  717 #endif
  718 
  719         linux_get_osname(td, osname);
  720         linux_get_osrelease(td, osrelease);
  721 
  722         bzero(&utsname, sizeof(utsname));
  723         strlcpy(utsname.sysname, osname, LINUX_MAX_UTSNAME);
  724         getcredhostname(td->td_ucred, utsname.nodename, LINUX_MAX_UTSNAME);
  725         strlcpy(utsname.release, osrelease, LINUX_MAX_UTSNAME);
  726         strlcpy(utsname.version, version, LINUX_MAX_UTSNAME);
  727         for (p = utsname.version; *p != '\0'; ++p)
  728                 if (*p == '\n') {
  729                         *p = '\0';
  730                         break;
  731                 }
  732         strlcpy(utsname.machine, linux_platform, LINUX_MAX_UTSNAME);
  733 
  734         strlcpy(utsname.domainname, domainname, LINUX_MAX_UTSNAME);
  735 
  736         return (copyout(&utsname, args->buf, sizeof(utsname)));
  737 }
  738 
  739 #if defined(__i386__) || (defined(__amd64__) && defined(COMPAT_LINUX32))
  740 struct l_utimbuf {
  741         l_time_t l_actime;
  742         l_time_t l_modtime;
  743 };
  744 
  745 int
  746 linux_utime(struct thread *td, struct linux_utime_args *args)
  747 {
  748         struct timeval tv[2], *tvp;
  749         struct l_utimbuf lut;
  750         char *fname;
  751         int error;
  752 
  753         LCONVPATHEXIST(td, args->fname, &fname);
  754 
  755 #ifdef DEBUG
  756         if (ldebug(utime))
  757                 printf(ARGS(utime, "%s, *"), fname);
  758 #endif
  759 
  760         if (args->times) {
  761                 if ((error = copyin(args->times, &lut, sizeof lut))) {
  762                         LFREEPATH(fname);
  763                         return error;
  764                 }
  765                 tv[0].tv_sec = lut.l_actime;
  766                 tv[0].tv_usec = 0;
  767                 tv[1].tv_sec = lut.l_modtime;
  768                 tv[1].tv_usec = 0;
  769                 tvp = tv;
  770         } else
  771                 tvp = NULL;
  772 
  773         error = kern_utimes(td, fname, UIO_SYSSPACE, tvp, UIO_SYSSPACE);
  774         LFREEPATH(fname);
  775         return (error);
  776 }
  777 
  778 int
  779 linux_utimes(struct thread *td, struct linux_utimes_args *args)
  780 {
  781         l_timeval ltv[2];
  782         struct timeval tv[2], *tvp = NULL;
  783         char *fname;
  784         int error;
  785 
  786         LCONVPATHEXIST(td, args->fname, &fname);
  787 
  788 #ifdef DEBUG
  789         if (ldebug(utimes))
  790                 printf(ARGS(utimes, "%s, *"), fname);
  791 #endif
  792 
  793         if (args->tptr != NULL) {
  794                 if ((error = copyin(args->tptr, ltv, sizeof ltv))) {
  795                         LFREEPATH(fname);
  796                         return (error);
  797                 }
  798                 tv[0].tv_sec = ltv[0].tv_sec;
  799                 tv[0].tv_usec = ltv[0].tv_usec;
  800                 tv[1].tv_sec = ltv[1].tv_sec;
  801                 tv[1].tv_usec = ltv[1].tv_usec;
  802                 tvp = tv;
  803         }
  804 
  805         error = kern_utimes(td, fname, UIO_SYSSPACE, tvp, UIO_SYSSPACE);
  806         LFREEPATH(fname);
  807         return (error);
  808 }
  809 #endif /* __i386__ || (__amd64__ && COMPAT_LINUX32) */
  810 
  811 #define __WCLONE 0x80000000
  812 
  813 int
  814 linux_waitpid(struct thread *td, struct linux_waitpid_args *args)
  815 {
  816         int error, options, tmpstat;
  817 
  818 #ifdef DEBUG
  819         if (ldebug(waitpid))
  820                 printf(ARGS(waitpid, "%d, %p, %d"),
  821                     args->pid, (void *)args->status, args->options);
  822 #endif
  823         /*
  824          * this is necessary because the test in kern_wait doesn't work
  825          * because we mess with the options here
  826          */
  827         if (args->options & ~(WUNTRACED | WNOHANG | WCONTINUED | __WCLONE))
  828                 return (EINVAL);
  829 
  830         options = (args->options & (WNOHANG | WUNTRACED));
  831         /* WLINUXCLONE should be equal to __WCLONE, but we make sure */
  832         if (args->options & __WCLONE)
  833                 options |= WLINUXCLONE;
  834 
  835         error = kern_wait(td, args->pid, &tmpstat, options, NULL);
  836         if (error)
  837                 return error;
  838 
  839         if (args->status) {
  840                 tmpstat &= 0xffff;
  841                 if (WIFSIGNALED(tmpstat))
  842                         tmpstat = (tmpstat & 0xffffff80) |
  843                             BSD_TO_LINUX_SIGNAL(WTERMSIG(tmpstat));
  844                 else if (WIFSTOPPED(tmpstat))
  845                         tmpstat = (tmpstat & 0xffff00ff) |
  846                             (BSD_TO_LINUX_SIGNAL(WSTOPSIG(tmpstat)) << 8);
  847                 return copyout(&tmpstat, args->status, sizeof(int));
  848         }
  849 
  850         return 0;
  851 }
  852 
  853 int
  854 linux_wait4(struct thread *td, struct linux_wait4_args *args)
  855 {
  856         int error, options, tmpstat;
  857         struct rusage ru, *rup;
  858         struct proc *p;
  859 
  860 #ifdef DEBUG
  861         if (ldebug(wait4))
  862                 printf(ARGS(wait4, "%d, %p, %d, %p"),
  863                     args->pid, (void *)args->status, args->options,
  864                     (void *)args->rusage);
  865 #endif
  866 
  867         options = (args->options & (WNOHANG | WUNTRACED));
  868         /* WLINUXCLONE should be equal to __WCLONE, but we make sure */
  869         if (args->options & __WCLONE)
  870                 options |= WLINUXCLONE;
  871 
  872         if (args->rusage != NULL)
  873                 rup = &ru;
  874         else
  875                 rup = NULL;
  876         error = kern_wait(td, args->pid, &tmpstat, options, rup);
  877         if (error)
  878                 return error;
  879 
  880         p = td->td_proc;
  881         PROC_LOCK(p);
  882         sigqueue_delete(&p->p_sigqueue, SIGCHLD);
  883         PROC_UNLOCK(p);
  884 
  885         if (args->status) {
  886                 tmpstat &= 0xffff;
  887                 if (WIFSIGNALED(tmpstat))
  888                         tmpstat = (tmpstat & 0xffffff80) |
  889                             BSD_TO_LINUX_SIGNAL(WTERMSIG(tmpstat));
  890                 else if (WIFSTOPPED(tmpstat))
  891                         tmpstat = (tmpstat & 0xffff00ff) |
  892                             (BSD_TO_LINUX_SIGNAL(WSTOPSIG(tmpstat)) << 8);
  893                 error = copyout(&tmpstat, args->status, sizeof(int));
  894         }
  895         if (args->rusage != NULL && error == 0)
  896                 error = copyout(&ru, args->rusage, sizeof(ru));
  897 
  898         return (error);
  899 }
  900 
  901 int
  902 linux_mknod(struct thread *td, struct linux_mknod_args *args)
  903 {
  904         char *path;
  905         int error;
  906 
  907         LCONVPATHCREAT(td, args->path, &path);
  908 
  909 #ifdef DEBUG
  910         if (ldebug(mknod))
  911                 printf(ARGS(mknod, "%s, %d, %d"), path, args->mode, args->dev);
  912 #endif
  913 
  914         switch (args->mode & S_IFMT) {
  915         case S_IFIFO:
  916         case S_IFSOCK:
  917                 error = kern_mkfifo(td, path, UIO_SYSSPACE, args->mode);
  918                 break;
  919 
  920         case S_IFCHR:
  921         case S_IFBLK:
  922                 error = kern_mknod(td, path, UIO_SYSSPACE, args->mode,
  923                     args->dev);
  924                 break;
  925 
  926         case S_IFDIR:
  927                 error = EPERM;
  928                 break;
  929 
  930         case 0:
  931                 args->mode |= S_IFREG;
  932                 /* FALLTHROUGH */
  933         case S_IFREG:
  934                 error = kern_open(td, path, UIO_SYSSPACE,
  935                     O_WRONLY | O_CREAT | O_TRUNC, args->mode);
  936                 break;
  937 
  938         default:
  939                 error = EINVAL;
  940                 break;
  941         }
  942         LFREEPATH(path);
  943         return (error);
  944 }
  945 
  946 /*
  947  * UGH! This is just about the dumbest idea I've ever heard!!
  948  */
  949 int
  950 linux_personality(struct thread *td, struct linux_personality_args *args)
  951 {
  952 #ifdef DEBUG
  953         if (ldebug(personality))
  954                 printf(ARGS(personality, "%lu"), (unsigned long)args->per);
  955 #endif
  956         if (args->per != 0)
  957                 return EINVAL;
  958 
  959         /* Yes Jim, it's still a Linux... */
  960         td->td_retval[0] = 0;
  961         return 0;
  962 }
  963 
  964 struct l_itimerval {
  965         l_timeval it_interval;
  966         l_timeval it_value;
  967 };
  968 
  969 #define B2L_ITIMERVAL(bip, lip)                                         \
  970         (bip)->it_interval.tv_sec = (lip)->it_interval.tv_sec;          \
  971         (bip)->it_interval.tv_usec = (lip)->it_interval.tv_usec;        \
  972         (bip)->it_value.tv_sec = (lip)->it_value.tv_sec;                \
  973         (bip)->it_value.tv_usec = (lip)->it_value.tv_usec;
  974 
  975 int
  976 linux_setitimer(struct thread *td, struct linux_setitimer_args *uap)
  977 {
  978         int error;
  979         struct l_itimerval ls;
  980         struct itimerval aitv, oitv;
  981 
  982 #ifdef DEBUG
  983         if (ldebug(setitimer))
  984                 printf(ARGS(setitimer, "%p, %p"),
  985                     (void *)uap->itv, (void *)uap->oitv);
  986 #endif
  987 
  988         if (uap->itv == NULL) {
  989                 uap->itv = uap->oitv;
  990                 return (linux_getitimer(td, (struct linux_getitimer_args *)uap));
  991         }
  992 
  993         error = copyin(uap->itv, &ls, sizeof(ls));
  994         if (error != 0)
  995                 return (error);
  996         B2L_ITIMERVAL(&aitv, &ls);
  997 #ifdef DEBUG
  998         if (ldebug(setitimer)) {
  999                 printf("setitimer: value: sec: %jd, usec: %ld\n",
 1000                     (intmax_t)aitv.it_value.tv_sec, aitv.it_value.tv_usec);
 1001                 printf("setitimer: interval: sec: %jd, usec: %ld\n",
 1002                     (intmax_t)aitv.it_interval.tv_sec, aitv.it_interval.tv_usec);
 1003         }
 1004 #endif
 1005         error = kern_setitimer(td, uap->which, &aitv, &oitv);
 1006         if (error != 0 || uap->oitv == NULL)
 1007                 return (error);
 1008         B2L_ITIMERVAL(&ls, &oitv);
 1009 
 1010         return (copyout(&ls, uap->oitv, sizeof(ls)));
 1011 }
 1012 
 1013 int
 1014 linux_getitimer(struct thread *td, struct linux_getitimer_args *uap)
 1015 {
 1016         int error;
 1017         struct l_itimerval ls;
 1018         struct itimerval aitv;
 1019 
 1020 #ifdef DEBUG
 1021         if (ldebug(getitimer))
 1022                 printf(ARGS(getitimer, "%p"), (void *)uap->itv);
 1023 #endif
 1024         error = kern_getitimer(td, uap->which, &aitv);
 1025         if (error != 0)
 1026                 return (error);
 1027         B2L_ITIMERVAL(&ls, &aitv);
 1028         return (copyout(&ls, uap->itv, sizeof(ls)));
 1029 }
 1030 
 1031 int
 1032 linux_nice(struct thread *td, struct linux_nice_args *args)
 1033 {
 1034         struct setpriority_args bsd_args;
 1035 
 1036         bsd_args.which = PRIO_PROCESS;
 1037         bsd_args.who = 0;               /* current process */
 1038         bsd_args.prio = args->inc;
 1039         return setpriority(td, &bsd_args);
 1040 }
 1041 
 1042 int
 1043 linux_setgroups(struct thread *td, struct linux_setgroups_args *args)
 1044 {
 1045         struct ucred *newcred, *oldcred;
 1046         l_gid_t linux_gidset[NGROUPS];
 1047         gid_t *bsd_gidset;
 1048         int ngrp, error;
 1049         struct proc *p;
 1050 
 1051         ngrp = args->gidsetsize;
 1052         if (ngrp < 0 || ngrp >= NGROUPS)
 1053                 return (EINVAL);
 1054         error = copyin(args->grouplist, linux_gidset, ngrp * sizeof(l_gid_t));
 1055         if (error)
 1056                 return (error);
 1057         newcred = crget();
 1058         p = td->td_proc;
 1059         PROC_LOCK(p);
 1060         oldcred = p->p_ucred;
 1061 
 1062         /*
 1063          * cr_groups[0] holds egid. Setting the whole set from
 1064          * the supplied set will cause egid to be changed too.
 1065          * Keep cr_groups[0] unchanged to prevent that.
 1066          */
 1067 
 1068         if ((error = priv_check_cred(oldcred, PRIV_CRED_SETGROUPS, 0)) != 0) {
 1069                 PROC_UNLOCK(p);
 1070                 crfree(newcred);
 1071                 return (error);
 1072         }
 1073 
 1074         crcopy(newcred, oldcred);
 1075         if (ngrp > 0) {
 1076                 newcred->cr_ngroups = ngrp + 1;
 1077 
 1078                 bsd_gidset = newcred->cr_groups;
 1079                 ngrp--;
 1080                 while (ngrp >= 0) {
 1081                         bsd_gidset[ngrp + 1] = linux_gidset[ngrp];
 1082                         ngrp--;
 1083                 }
 1084         } else
 1085                 newcred->cr_ngroups = 1;
 1086 
 1087         setsugid(p);
 1088         p->p_ucred = newcred;
 1089         PROC_UNLOCK(p);
 1090         crfree(oldcred);
 1091         return (0);
 1092 }
 1093 
 1094 int
 1095 linux_getgroups(struct thread *td, struct linux_getgroups_args *args)
 1096 {
 1097         struct ucred *cred;
 1098         l_gid_t linux_gidset[NGROUPS];
 1099         gid_t *bsd_gidset;
 1100         int bsd_gidsetsz, ngrp, error;
 1101 
 1102         cred = td->td_ucred;
 1103         bsd_gidset = cred->cr_groups;
 1104         bsd_gidsetsz = cred->cr_ngroups - 1;
 1105 
 1106         /*
 1107          * cr_groups[0] holds egid. Returning the whole set
 1108          * here will cause a duplicate. Exclude cr_groups[0]
 1109          * to prevent that.
 1110          */
 1111 
 1112         if ((ngrp = args->gidsetsize) == 0) {
 1113                 td->td_retval[0] = bsd_gidsetsz;
 1114                 return (0);
 1115         }
 1116 
 1117         if (ngrp < bsd_gidsetsz)
 1118                 return (EINVAL);
 1119 
 1120         ngrp = 0;
 1121         while (ngrp < bsd_gidsetsz) {
 1122                 linux_gidset[ngrp] = bsd_gidset[ngrp + 1];
 1123                 ngrp++;
 1124         }
 1125 
 1126         if ((error = copyout(linux_gidset, args->grouplist,
 1127             ngrp * sizeof(l_gid_t))))
 1128                 return (error);
 1129 
 1130         td->td_retval[0] = ngrp;
 1131         return (0);
 1132 }
 1133 
 1134 int
 1135 linux_setrlimit(struct thread *td, struct linux_setrlimit_args *args)
 1136 {
 1137         struct rlimit bsd_rlim;
 1138         struct l_rlimit rlim;
 1139         u_int which;
 1140         int error;
 1141 
 1142 #ifdef DEBUG
 1143         if (ldebug(setrlimit))
 1144                 printf(ARGS(setrlimit, "%d, %p"),
 1145                     args->resource, (void *)args->rlim);
 1146 #endif
 1147 
 1148         if (args->resource >= LINUX_RLIM_NLIMITS)
 1149                 return (EINVAL);
 1150 
 1151         which = linux_to_bsd_resource[args->resource];
 1152         if (which == -1)
 1153                 return (EINVAL);
 1154 
 1155         error = copyin(args->rlim, &rlim, sizeof(rlim));
 1156         if (error)
 1157                 return (error);
 1158 
 1159         bsd_rlim.rlim_cur = (rlim_t)rlim.rlim_cur;
 1160         bsd_rlim.rlim_max = (rlim_t)rlim.rlim_max;
 1161         return (kern_setrlimit(td, which, &bsd_rlim));
 1162 }
 1163 
 1164 int
 1165 linux_old_getrlimit(struct thread *td, struct linux_old_getrlimit_args *args)
 1166 {
 1167         struct l_rlimit rlim;
 1168         struct proc *p = td->td_proc;
 1169         struct rlimit bsd_rlim;
 1170         u_int which;
 1171 
 1172 #ifdef DEBUG
 1173         if (ldebug(old_getrlimit))
 1174                 printf(ARGS(old_getrlimit, "%d, %p"),
 1175                     args->resource, (void *)args->rlim);
 1176 #endif
 1177 
 1178         if (args->resource >= LINUX_RLIM_NLIMITS)
 1179                 return (EINVAL);
 1180 
 1181         which = linux_to_bsd_resource[args->resource];
 1182         if (which == -1)
 1183                 return (EINVAL);
 1184 
 1185         PROC_LOCK(p);
 1186         lim_rlimit(p, which, &bsd_rlim);
 1187         PROC_UNLOCK(p);
 1188 
 1189 #ifdef COMPAT_LINUX32
 1190         rlim.rlim_cur = (unsigned int)bsd_rlim.rlim_cur;
 1191         if (rlim.rlim_cur == UINT_MAX)
 1192                 rlim.rlim_cur = INT_MAX;
 1193         rlim.rlim_max = (unsigned int)bsd_rlim.rlim_max;
 1194         if (rlim.rlim_max == UINT_MAX)
 1195                 rlim.rlim_max = INT_MAX;
 1196 #else
 1197         rlim.rlim_cur = (unsigned long)bsd_rlim.rlim_cur;
 1198         if (rlim.rlim_cur == ULONG_MAX)
 1199                 rlim.rlim_cur = LONG_MAX;
 1200         rlim.rlim_max = (unsigned long)bsd_rlim.rlim_max;
 1201         if (rlim.rlim_max == ULONG_MAX)
 1202                 rlim.rlim_max = LONG_MAX;
 1203 #endif
 1204         return (copyout(&rlim, args->rlim, sizeof(rlim)));
 1205 }
 1206 
 1207 int
 1208 linux_getrlimit(struct thread *td, struct linux_getrlimit_args *args)
 1209 {
 1210         struct l_rlimit rlim;
 1211         struct proc *p = td->td_proc;
 1212         struct rlimit bsd_rlim;
 1213         u_int which;
 1214 
 1215 #ifdef DEBUG
 1216         if (ldebug(getrlimit))
 1217                 printf(ARGS(getrlimit, "%d, %p"),
 1218                     args->resource, (void *)args->rlim);
 1219 #endif
 1220 
 1221         if (args->resource >= LINUX_RLIM_NLIMITS)
 1222                 return (EINVAL);
 1223 
 1224         which = linux_to_bsd_resource[args->resource];
 1225         if (which == -1)
 1226                 return (EINVAL);
 1227 
 1228         PROC_LOCK(p);
 1229         lim_rlimit(p, which, &bsd_rlim);
 1230         PROC_UNLOCK(p);
 1231 
 1232         rlim.rlim_cur = (l_ulong)bsd_rlim.rlim_cur;
 1233         rlim.rlim_max = (l_ulong)bsd_rlim.rlim_max;
 1234         return (copyout(&rlim, args->rlim, sizeof(rlim)));
 1235 }
 1236 
 1237 int
 1238 linux_sched_setscheduler(struct thread *td,
 1239     struct linux_sched_setscheduler_args *args)
 1240 {
 1241         struct sched_setscheduler_args bsd;
 1242 
 1243 #ifdef DEBUG
 1244         if (ldebug(sched_setscheduler))
 1245                 printf(ARGS(sched_setscheduler, "%d, %d, %p"),
 1246                     args->pid, args->policy, (const void *)args->param);
 1247 #endif
 1248 
 1249         switch (args->policy) {
 1250         case LINUX_SCHED_OTHER:
 1251                 bsd.policy = SCHED_OTHER;
 1252                 break;
 1253         case LINUX_SCHED_FIFO:
 1254                 bsd.policy = SCHED_FIFO;
 1255                 break;
 1256         case LINUX_SCHED_RR:
 1257                 bsd.policy = SCHED_RR;
 1258                 break;
 1259         default:
 1260                 return EINVAL;
 1261         }
 1262 
 1263         bsd.pid = args->pid;
 1264         bsd.param = (struct sched_param *)args->param;
 1265         return sched_setscheduler(td, &bsd);
 1266 }
 1267 
 1268 int
 1269 linux_sched_getscheduler(struct thread *td,
 1270     struct linux_sched_getscheduler_args *args)
 1271 {
 1272         struct sched_getscheduler_args bsd;
 1273         int error;
 1274 
 1275 #ifdef DEBUG
 1276         if (ldebug(sched_getscheduler))
 1277                 printf(ARGS(sched_getscheduler, "%d"), args->pid);
 1278 #endif
 1279 
 1280         bsd.pid = args->pid;
 1281         error = sched_getscheduler(td, &bsd);
 1282 
 1283         switch (td->td_retval[0]) {
 1284         case SCHED_OTHER:
 1285                 td->td_retval[0] = LINUX_SCHED_OTHER;
 1286                 break;
 1287         case SCHED_FIFO:
 1288                 td->td_retval[0] = LINUX_SCHED_FIFO;
 1289                 break;
 1290         case SCHED_RR:
 1291                 td->td_retval[0] = LINUX_SCHED_RR;
 1292                 break;
 1293         }
 1294 
 1295         return error;
 1296 }
 1297 
 1298 int
 1299 linux_sched_get_priority_max(struct thread *td,
 1300     struct linux_sched_get_priority_max_args *args)
 1301 {
 1302         struct sched_get_priority_max_args bsd;
 1303 
 1304 #ifdef DEBUG
 1305         if (ldebug(sched_get_priority_max))
 1306                 printf(ARGS(sched_get_priority_max, "%d"), args->policy);
 1307 #endif
 1308 
 1309         switch (args->policy) {
 1310         case LINUX_SCHED_OTHER:
 1311                 bsd.policy = SCHED_OTHER;
 1312                 break;
 1313         case LINUX_SCHED_FIFO:
 1314                 bsd.policy = SCHED_FIFO;
 1315                 break;
 1316         case LINUX_SCHED_RR:
 1317                 bsd.policy = SCHED_RR;
 1318                 break;
 1319         default:
 1320                 return EINVAL;
 1321         }
 1322         return sched_get_priority_max(td, &bsd);
 1323 }
 1324 
 1325 int
 1326 linux_sched_get_priority_min(struct thread *td,
 1327     struct linux_sched_get_priority_min_args *args)
 1328 {
 1329         struct sched_get_priority_min_args bsd;
 1330 
 1331 #ifdef DEBUG
 1332         if (ldebug(sched_get_priority_min))
 1333                 printf(ARGS(sched_get_priority_min, "%d"), args->policy);
 1334 #endif
 1335 
 1336         switch (args->policy) {
 1337         case LINUX_SCHED_OTHER:
 1338                 bsd.policy = SCHED_OTHER;
 1339                 break;
 1340         case LINUX_SCHED_FIFO:
 1341                 bsd.policy = SCHED_FIFO;
 1342                 break;
 1343         case LINUX_SCHED_RR:
 1344                 bsd.policy = SCHED_RR;
 1345                 break;
 1346         default:
 1347                 return EINVAL;
 1348         }
 1349         return sched_get_priority_min(td, &bsd);
 1350 }
 1351 
 1352 #define REBOOT_CAD_ON   0x89abcdef
 1353 #define REBOOT_CAD_OFF  0
 1354 #define REBOOT_HALT     0xcdef0123
 1355 #define REBOOT_RESTART  0x01234567
 1356 #define REBOOT_RESTART2 0xA1B2C3D4
 1357 #define REBOOT_POWEROFF 0x4321FEDC
 1358 #define REBOOT_MAGIC1   0xfee1dead
 1359 #define REBOOT_MAGIC2   0x28121969
 1360 #define REBOOT_MAGIC2A  0x05121996
 1361 #define REBOOT_MAGIC2B  0x16041998
 1362 
 1363 int
 1364 linux_reboot(struct thread *td, struct linux_reboot_args *args)
 1365 {
 1366         struct reboot_args bsd_args;
 1367 
 1368 #ifdef DEBUG
 1369         if (ldebug(reboot))
 1370                 printf(ARGS(reboot, "0x%x"), args->cmd);
 1371 #endif
 1372 
 1373         if (args->magic1 != REBOOT_MAGIC1)
 1374                 return EINVAL;
 1375 
 1376         switch (args->magic2) {
 1377         case REBOOT_MAGIC2:
 1378         case REBOOT_MAGIC2A:
 1379         case REBOOT_MAGIC2B:
 1380                 break;
 1381         default:
 1382                 return EINVAL;
 1383         }
 1384 
 1385         switch (args->cmd) {
 1386         case REBOOT_CAD_ON:
 1387         case REBOOT_CAD_OFF:
 1388                 return (priv_check(td, PRIV_REBOOT));
 1389         case REBOOT_HALT:
 1390                 bsd_args.opt = RB_HALT;
 1391                 break;
 1392         case REBOOT_RESTART:
 1393         case REBOOT_RESTART2:
 1394                 bsd_args.opt = 0;
 1395                 break;
 1396         case REBOOT_POWEROFF:
 1397                 bsd_args.opt = RB_POWEROFF;
 1398                 break;
 1399         default:
 1400                 return EINVAL;
 1401         }
 1402         return reboot(td, &bsd_args);
 1403 }
 1404 
 1405 
 1406 /*
 1407  * The FreeBSD native getpid(2), getgid(2) and getuid(2) also modify
 1408  * td->td_retval[1] when COMPAT_43 is defined. This clobbers registers that
 1409  * are assumed to be preserved. The following lightweight syscalls fixes
 1410  * this. See also linux_getgid16() and linux_getuid16() in linux_uid16.c
 1411  *
 1412  * linux_getpid() - MP SAFE
 1413  * linux_getgid() - MP SAFE
 1414  * linux_getuid() - MP SAFE
 1415  */
 1416 
 1417 int
 1418 linux_getpid(struct thread *td, struct linux_getpid_args *args)
 1419 {
 1420         struct linux_emuldata *em;
 1421 
 1422 #ifdef DEBUG
 1423         if (ldebug(getpid))
 1424                 printf(ARGS(getpid, ""));
 1425 #endif
 1426 
 1427         if (linux_use26(td)) {
 1428                 em = em_find(td->td_proc, EMUL_DONTLOCK);
 1429                 KASSERT(em != NULL, ("getpid: emuldata not found.\n"));
 1430                 td->td_retval[0] = em->shared->group_pid;
 1431         } else {
 1432                 td->td_retval[0] = td->td_proc->p_pid;
 1433         }
 1434 
 1435         return (0);
 1436 }
 1437 
 1438 int
 1439 linux_gettid(struct thread *td, struct linux_gettid_args *args)
 1440 {
 1441 
 1442 #ifdef DEBUG
 1443         if (ldebug(gettid))
 1444                 printf(ARGS(gettid, ""));
 1445 #endif
 1446 
 1447         td->td_retval[0] = td->td_proc->p_pid;
 1448         return (0);
 1449 }
 1450 
 1451 
 1452 int
 1453 linux_getppid(struct thread *td, struct linux_getppid_args *args)
 1454 {
 1455         struct linux_emuldata *em;
 1456         struct proc *p, *pp;
 1457 
 1458 #ifdef DEBUG
 1459         if (ldebug(getppid))
 1460                 printf(ARGS(getppid, ""));
 1461 #endif
 1462 
 1463         if (!linux_use26(td)) {
 1464                 PROC_LOCK(td->td_proc);
 1465                 td->td_retval[0] = td->td_proc->p_pptr->p_pid;
 1466                 PROC_UNLOCK(td->td_proc);
 1467                 return (0);
 1468         }
 1469 
 1470         em = em_find(td->td_proc, EMUL_DONTLOCK);
 1471 
 1472         KASSERT(em != NULL, ("getppid: process emuldata not found.\n"));
 1473 
 1474         /* find the group leader */
 1475         p = pfind(em->shared->group_pid);
 1476 
 1477         if (p == NULL) {
 1478 #ifdef DEBUG
 1479                 printf(LMSG("parent process not found.\n"));
 1480 #endif
 1481                 return (0);
 1482         }
 1483 
 1484         pp = p->p_pptr;         /* switch to parent */
 1485         PROC_LOCK(pp);
 1486         PROC_UNLOCK(p);
 1487 
 1488         /* if its also linux process */
 1489         if (pp->p_sysent == &elf_linux_sysvec) {
 1490                 em = em_find(pp, EMUL_DONTLOCK);
 1491                 KASSERT(em != NULL, ("getppid: parent emuldata not found.\n"));
 1492 
 1493                 td->td_retval[0] = em->shared->group_pid;
 1494         } else
 1495                 td->td_retval[0] = pp->p_pid;
 1496 
 1497         PROC_UNLOCK(pp);
 1498 
 1499         return (0);
 1500 }
 1501 
 1502 int
 1503 linux_getgid(struct thread *td, struct linux_getgid_args *args)
 1504 {
 1505 
 1506 #ifdef DEBUG
 1507         if (ldebug(getgid))
 1508                 printf(ARGS(getgid, ""));
 1509 #endif
 1510 
 1511         td->td_retval[0] = td->td_ucred->cr_rgid;
 1512         return (0);
 1513 }
 1514 
 1515 int
 1516 linux_getuid(struct thread *td, struct linux_getuid_args *args)
 1517 {
 1518 
 1519 #ifdef DEBUG
 1520         if (ldebug(getuid))
 1521                 printf(ARGS(getuid, ""));
 1522 #endif
 1523 
 1524         td->td_retval[0] = td->td_ucred->cr_ruid;
 1525         return (0);
 1526 }
 1527 
 1528 
 1529 int
 1530 linux_getsid(struct thread *td, struct linux_getsid_args *args)
 1531 {
 1532         struct getsid_args bsd;
 1533 
 1534 #ifdef DEBUG
 1535         if (ldebug(getsid))
 1536                 printf(ARGS(getsid, "%i"), args->pid);
 1537 #endif
 1538 
 1539         bsd.pid = args->pid;
 1540         return getsid(td, &bsd);
 1541 }
 1542 
 1543 int
 1544 linux_nosys(struct thread *td, struct nosys_args *ignore)
 1545 {
 1546 
 1547         return (ENOSYS);
 1548 }
 1549 
 1550 int
 1551 linux_getpriority(struct thread *td, struct linux_getpriority_args *args)
 1552 {
 1553         struct getpriority_args bsd_args;
 1554         int error;
 1555 
 1556 #ifdef DEBUG
 1557         if (ldebug(getpriority))
 1558                 printf(ARGS(getpriority, "%i, %i"), args->which, args->who);
 1559 #endif
 1560 
 1561         bsd_args.which = args->which;
 1562         bsd_args.who = args->who;
 1563         error = getpriority(td, &bsd_args);
 1564         td->td_retval[0] = 20 - td->td_retval[0];
 1565         return error;
 1566 }
 1567 
 1568 int
 1569 linux_sethostname(struct thread *td, struct linux_sethostname_args *args)
 1570 {
 1571         int name[2];
 1572 
 1573 #ifdef DEBUG
 1574         if (ldebug(sethostname))
 1575                 printf(ARGS(sethostname, "*, %i"), args->len);
 1576 #endif
 1577 
 1578         name[0] = CTL_KERN;
 1579         name[1] = KERN_HOSTNAME;
 1580         return (userland_sysctl(td, name, 2, 0, 0, 0, args->hostname,
 1581             args->len, 0, 0));
 1582 }
 1583 
 1584 int
 1585 linux_exit_group(struct thread *td, struct linux_exit_group_args *args)
 1586 {
 1587         struct linux_emuldata *em, *td_em, *tmp_em;
 1588         struct proc *sp;
 1589 
 1590 #ifdef DEBUG
 1591         if (ldebug(exit_group))
 1592                 printf(ARGS(exit_group, "%i"), args->error_code);
 1593 #endif
 1594 
 1595         if (linux_use26(td)) {
 1596                 td_em = em_find(td->td_proc, EMUL_DONTLOCK);
 1597 
 1598                 KASSERT(td_em != NULL, ("exit_group: emuldata not found.\n"));
 1599 
 1600                 EMUL_SHARED_RLOCK(&emul_shared_lock);
 1601                 LIST_FOREACH_SAFE(em, &td_em->shared->threads, threads, tmp_em) {
 1602                         if (em->pid == td_em->pid)
 1603                                 continue;
 1604 
 1605                         sp = pfind(em->pid);
 1606                         psignal(sp, SIGKILL);
 1607                         PROC_UNLOCK(sp);
 1608 #ifdef DEBUG
 1609                         printf(LMSG("linux_sys_exit_group: kill PID %d\n"), em->pid);
 1610 #endif
 1611                 }
 1612 
 1613                 EMUL_SHARED_RUNLOCK(&emul_shared_lock);
 1614         }
 1615         /*
 1616          * XXX: we should send a signal to the parent if
 1617          * SIGNAL_EXIT_GROUP is set. We ignore that (temporarily?)
 1618          * as it doesnt occur often.
 1619          */
 1620         exit1(td, W_EXITCODE(args->error_code, 0));
 1621 
 1622         return (0);
 1623 }
 1624 
 1625 int
 1626 linux_prctl(struct thread *td, struct linux_prctl_args *args)
 1627 {
 1628         int error = 0, max_size;
 1629         struct proc *p = td->td_proc;
 1630         char comm[LINUX_MAX_COMM_LEN];
 1631         struct linux_emuldata *em;
 1632         int pdeath_signal;
 1633 
 1634 #ifdef DEBUG
 1635         if (ldebug(prctl))
 1636                 printf(ARGS(prctl, "%d, %d, %d, %d, %d"), args->option,
 1637                     args->arg2, args->arg3, args->arg4, args->arg5);
 1638 #endif
 1639 
 1640         switch (args->option) {
 1641         case LINUX_PR_SET_PDEATHSIG:
 1642                 if (!LINUX_SIG_VALID(args->arg2))
 1643                         return (EINVAL);
 1644                 em = em_find(p, EMUL_DOLOCK);
 1645                 KASSERT(em != NULL, ("prctl: emuldata not found.\n"));
 1646                 em->pdeath_signal = args->arg2;
 1647                 EMUL_UNLOCK(&emul_lock);
 1648                 break;
 1649         case LINUX_PR_GET_PDEATHSIG:
 1650                 em = em_find(p, EMUL_DOLOCK);
 1651                 KASSERT(em != NULL, ("prctl: emuldata not found.\n"));
 1652                 pdeath_signal = em->pdeath_signal;
 1653                 EMUL_UNLOCK(&emul_lock);
 1654                 error = copyout(&pdeath_signal,
 1655                     (void *)(register_t)args->arg2,
 1656                     sizeof(pdeath_signal));
 1657                 break;
 1658         case LINUX_PR_SET_NAME:
 1659                 /*
 1660                  * To be on the safe side we need to make sure to not
 1661                  * overflow the size a linux program expects. We already
 1662                  * do this here in the copyin, so that we don't need to
 1663                  * check on copyout.
 1664                  */
 1665                 max_size = MIN(sizeof(comm), sizeof(p->p_comm));
 1666                 error = copyinstr((void *)(register_t)args->arg2, comm,
 1667                     max_size, NULL);
 1668 
 1669                 /* Linux silently truncates the name if it is too long. */
 1670                 if (error == ENAMETOOLONG) {
 1671                         /*
 1672                          * XXX: copyinstr() isn't documented to populate the
 1673                          * array completely, so do a copyin() to be on the
 1674                          * safe side. This should be changed in case
 1675                          * copyinstr() is changed to guarantee this.
 1676                          */
 1677                         error = copyin((void *)(register_t)args->arg2, comm,
 1678                             max_size - 1);
 1679                         comm[max_size - 1] = '\0';
 1680                 }
 1681                 if (error)
 1682                         return (error);
 1683 
 1684                 PROC_LOCK(p);
 1685                 strlcpy(p->p_comm, comm, sizeof(p->p_comm));
 1686                 PROC_UNLOCK(p);
 1687                 break;
 1688         case LINUX_PR_GET_NAME:
 1689                 PROC_LOCK(p);
 1690                 strlcpy(comm, p->p_comm, sizeof(comm));
 1691                 PROC_UNLOCK(p);
 1692                 error = copyout(comm, (void *)(register_t)args->arg2,
 1693                     strlen(comm) + 1);
 1694                 break;
 1695         default:
 1696                 error = EINVAL;
 1697                 break;
 1698         }
 1699 
 1700         return (error);
 1701 }
 1702 
 1703 /*
 1704  * Get affinity of a process.
 1705  */
 1706 int
 1707 linux_sched_getaffinity(struct thread *td,
 1708     struct linux_sched_getaffinity_args *args)
 1709 {
 1710         int error;
 1711         struct cpuset_getaffinity_args cga;
 1712 
 1713 #ifdef DEBUG
 1714         if (ldebug(sched_getaffinity))
 1715                 printf(ARGS(sched_getaffinity, "%d, %d, *"), args->pid,
 1716                     args->len);
 1717 #endif
 1718         if (args->len < sizeof(cpuset_t))
 1719                 return (EINVAL);
 1720 
 1721         cga.level = CPU_LEVEL_WHICH;
 1722         cga.which = CPU_WHICH_PID;
 1723         cga.id = args->pid;
 1724         cga.cpusetsize = sizeof(cpuset_t);
 1725         cga.mask = (cpuset_t *) args->user_mask_ptr;
 1726 
 1727         if ((error = cpuset_getaffinity(td, &cga)) == 0)
 1728                 td->td_retval[0] = sizeof(cpuset_t);
 1729 
 1730         return (error);
 1731 }
 1732 
 1733 /*
 1734  *  Set affinity of a process.
 1735  */
 1736 int
 1737 linux_sched_setaffinity(struct thread *td,
 1738     struct linux_sched_setaffinity_args *args)
 1739 {
 1740         struct cpuset_setaffinity_args csa;
 1741 
 1742 #ifdef DEBUG
 1743         if (ldebug(sched_setaffinity))
 1744                 printf(ARGS(sched_setaffinity, "%d, %d, *"), args->pid,
 1745                     args->len);
 1746 #endif
 1747         if (args->len < sizeof(cpuset_t))
 1748                 return (EINVAL);
 1749 
 1750         csa.level = CPU_LEVEL_WHICH;
 1751         csa.which = CPU_WHICH_PID;
 1752         csa.id = args->pid;
 1753         csa.cpusetsize = sizeof(cpuset_t);
 1754         csa.mask = (cpuset_t *) args->user_mask_ptr;
 1755 
 1756         return (cpuset_setaffinity(td, &csa));
 1757 }

Cache object: 59d7bb8f1e02c4fc0052013da217f979


[ source navigation ] [ diff markup ] [ identifier search ] [ freetext search ] [ file search ] [ list types ] [ track identifier ]


This page is part of the FreeBSD/Linux Linux Kernel Cross-Reference, and was automatically generated using a modified version of the LXR engine.