linux_misc.c revision 161317
19313Ssos/*- 2133749Stjr * Copyright (c) 2002 Doug Rabson 39313Ssos * Copyright (c) 1994-1995 S�ren Schmidt 49313Ssos * All rights reserved. 59313Ssos * 69313Ssos * Redistribution and use in source and binary forms, with or without 79313Ssos * modification, are permitted provided that the following conditions 89313Ssos * are met: 99313Ssos * 1. Redistributions of source code must retain the above copyright 1014331Speter * notice, this list of conditions and the following disclaimer 119313Ssos * in this position and unchanged. 129313Ssos * 2. Redistributions in binary form must reproduce the above copyright 139313Ssos * notice, this list of conditions and the following disclaimer in the 149313Ssos * documentation and/or other materials provided with the distribution. 159313Ssos * 3. The name of the author may not be used to endorse or promote products 1697748Sschweikh * derived from this software without specific prior written permission 179313Ssos * 189313Ssos * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR 199313Ssos * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES 209313Ssos * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. 219313Ssos * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, 229313Ssos * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT 239313Ssos * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, 249313Ssos * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY 259313Ssos * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 269313Ssos * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF 279313Ssos * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 289313Ssos */ 299313Ssos 30116173Sobrien#include <sys/cdefs.h> 31116173Sobrien__FBSDID("$FreeBSD: head/sys/compat/linux/linux_misc.c 161317 2006-08-15 15:15:17Z netchild $"); 32116173Sobrien 33156874Sru#include "opt_compat.h" 34101189Srwatson#include "opt_mac.h" 3549842Smarcel 369313Ssos#include <sys/param.h> 37102954Sbde#include <sys/blist.h> 3876166Smarkm#include <sys/fcntl.h> 39158415Snetchild#if defined(__i386__) 4076166Smarkm#include <sys/imgact_aout.h> 41133816Stjr#endif 4291385Srobert#include <sys/jail.h> 4312458Sbde#include <sys/kernel.h> 44114216Skan#include <sys/limits.h> 4576166Smarkm#include <sys/lock.h> 46101189Srwatson#include <sys/mac.h> 47102954Sbde#include <sys/malloc.h> 489313Ssos#include <sys/mman.h> 499313Ssos#include <sys/mount.h> 5076166Smarkm#include <sys/mutex.h> 519313Ssos#include <sys/namei.h> 5276166Smarkm#include <sys/proc.h> 5372538Sjlemon#include <sys/reboot.h> 549313Ssos#include <sys/resourcevar.h> 5576166Smarkm#include <sys/signalvar.h> 569313Ssos#include <sys/stat.h> 57102814Siedowse#include <sys/syscallsubr.h> 5812458Sbde#include <sys/sysctl.h> 5976166Smarkm#include <sys/sysproto.h> 60102954Sbde#include <sys/systm.h> 6176166Smarkm#include <sys/time.h> 6280180Spirzyk#include <sys/vmmeter.h> 639313Ssos#include <sys/vnode.h> 649313Ssos#include <sys/wait.h> 659313Ssos 6612652Sbde#include <vm/vm.h> 6712689Speter#include <vm/pmap.h> 6812458Sbde#include <vm/vm_kern.h> 6912689Speter#include <vm/vm_map.h> 7012842Sbde#include <vm/vm_extern.h> 7180180Spirzyk#include <vm/vm_object.h> 7280180Spirzyk#include <vm/swap_pager.h> 7312458Sbde 7449849Smarcel#include <posix4/sched.h> 7549849Smarcel 76143197Ssobomax#include <compat/linux/linux_sysproto.h> 77161310Snetchild#include <compat/linux/linux_emul.h> 78143197Ssobomax 79140214Sobrien#ifdef COMPAT_LINUX32 80140214Sobrien#include <machine/../linux32/linux.h> 81140214Sobrien#include <machine/../linux32/linux32_proto.h> 82140214Sobrien#else 8364909Smarcel#include <machine/../linux/linux.h> 8468583Smarcel#include <machine/../linux/linux_proto.h> 85133816Stjr#endif 86102954Sbde 8764909Smarcel#include <compat/linux/linux_mib.h> 8864909Smarcel#include <compat/linux/linux_util.h> 8964909Smarcel 90118149Sdes#ifdef __i386__ 91118149Sdes#include <machine/cputypes.h> 92118149Sdes#endif 93118149Sdes 9451793Smarcel#define BSD_TO_LINUX_SIGNAL(sig) \ 9551793Smarcel (((sig) <= LINUX_SIGTBLSZ) ? bsd_to_linux_signal[_SIG_IDX(sig)] : sig) 9651793Smarcel 97161310Snetchildextern struct sx emul_shared_lock; 98161310Snetchildextern struct sx emul_lock; 99161310Snetchild 10083221Smarcelstatic unsigned int linux_to_bsd_resource[LINUX_RLIM_NLIMITS] = { 10183221Smarcel RLIMIT_CPU, RLIMIT_FSIZE, RLIMIT_DATA, RLIMIT_STACK, 10283221Smarcel RLIMIT_CORE, RLIMIT_RSS, RLIMIT_NPROC, RLIMIT_NOFILE, 10383221Smarcel RLIMIT_MEMLOCK, -1 10449626Smarcel}; 10549626Smarcel 10683221Smarcelstruct l_sysinfo { 10783221Smarcel l_long uptime; /* Seconds since boot */ 10883221Smarcel l_ulong loads[3]; /* 1, 5, and 15 minute load averages */ 109122802Ssobomax#define LINUX_SYSINFO_LOADS_SCALE 65536 11083221Smarcel l_ulong totalram; /* Total usable main memory size */ 11183221Smarcel l_ulong freeram; /* Available memory size */ 11283221Smarcel l_ulong sharedram; /* Amount of shared memory */ 11383221Smarcel l_ulong bufferram; /* Memory used by buffers */ 11483221Smarcel l_ulong totalswap; /* Total swap space size */ 11583221Smarcel l_ulong freeswap; /* swap space still available */ 11683221Smarcel l_ushort procs; /* Number of current processes */ 117122802Ssobomax l_ulong totalbig; 118122802Ssobomax l_ulong freebig; 119122802Ssobomax l_uint mem_unit; 120122802Ssobomax char _f[6]; /* Pads structure to 64 bytes */ 12180180Spirzyk}; 1229313Ssosint 12383366Sjulianlinux_sysinfo(struct thread *td, struct linux_sysinfo_args *args) 12480180Spirzyk{ 12583221Smarcel struct l_sysinfo sysinfo; 12683221Smarcel vm_object_t object; 127117723Sphk int i, j; 12883221Smarcel struct timespec ts; 12980180Spirzyk 13083221Smarcel getnanouptime(&ts); 131147816Sjhb if (ts.tv_nsec != 0) 132147816Sjhb ts.tv_sec++; 133147816Sjhb sysinfo.uptime = ts.tv_sec; 13480180Spirzyk 13583221Smarcel /* Use the information from the mib to get our load averages */ 13683221Smarcel for (i = 0; i < 3; i++) 137122802Ssobomax sysinfo.loads[i] = averunnable.ldavg[i] * 138122802Ssobomax LINUX_SYSINFO_LOADS_SCALE / averunnable.fscale; 13980180Spirzyk 14083221Smarcel sysinfo.totalram = physmem * PAGE_SIZE; 14183221Smarcel sysinfo.freeram = sysinfo.totalram - cnt.v_wire_count * PAGE_SIZE; 14280180Spirzyk 14383221Smarcel sysinfo.sharedram = 0; 144124082Salc mtx_lock(&vm_object_list_mtx); 145124082Salc TAILQ_FOREACH(object, &vm_object_list, object_list) 14683221Smarcel if (object->shadow_count > 1) 14783221Smarcel sysinfo.sharedram += object->resident_page_count; 148124082Salc mtx_unlock(&vm_object_list_mtx); 14980180Spirzyk 15083221Smarcel sysinfo.sharedram *= PAGE_SIZE; 15183221Smarcel sysinfo.bufferram = 0; 15280180Spirzyk 153117723Sphk swap_pager_status(&i, &j); 154117723Sphk sysinfo.totalswap= i * PAGE_SIZE; 155117723Sphk sysinfo.freeswap = (i - j) * PAGE_SIZE; 15680180Spirzyk 157122802Ssobomax sysinfo.procs = nprocs; 15880180Spirzyk 159122802Ssobomax /* The following are only present in newer Linux kernels. */ 160122802Ssobomax sysinfo.totalbig = 0; 161122802Ssobomax sysinfo.freebig = 0; 162122802Ssobomax sysinfo.mem_unit = 1; 163122802Ssobomax 164111797Sdes return copyout(&sysinfo, args->info, sizeof(sysinfo)); 16580180Spirzyk} 16680180Spirzyk 16780180Spirzykint 16883366Sjulianlinux_alarm(struct thread *td, struct linux_alarm_args *args) 1699313Ssos{ 17083221Smarcel struct itimerval it, old_it; 171141467Sjhb int error; 1729313Ssos 1739313Ssos#ifdef DEBUG 17472543Sjlemon if (ldebug(alarm)) 17572543Sjlemon printf(ARGS(alarm, "%u"), args->secs); 1769313Ssos#endif 17783221Smarcel 17883221Smarcel if (args->secs > 100000000) 179141467Sjhb return (EINVAL); 18083221Smarcel 18183221Smarcel it.it_value.tv_sec = (long)args->secs; 18283221Smarcel it.it_value.tv_usec = 0; 18383221Smarcel it.it_interval.tv_sec = 0; 18483221Smarcel it.it_interval.tv_usec = 0; 185141467Sjhb error = kern_setitimer(td, ITIMER_REAL, &it, &old_it); 186141467Sjhb if (error) 187141467Sjhb return (error); 188141467Sjhb if (timevalisset(&old_it.it_value)) { 18983221Smarcel if (old_it.it_value.tv_usec != 0) 19083221Smarcel old_it.it_value.tv_sec++; 19183366Sjulian td->td_retval[0] = old_it.it_value.tv_sec; 19283221Smarcel } 193141467Sjhb return (0); 1949313Ssos} 1959313Ssos 1969313Ssosint 19783366Sjulianlinux_brk(struct thread *td, struct linux_brk_args *args) 1989313Ssos{ 19983366Sjulian struct vmspace *vm = td->td_proc->p_vmspace; 20083221Smarcel vm_offset_t new, old; 20183221Smarcel struct obreak_args /* { 20283221Smarcel char * nsize; 20383221Smarcel } */ tmp; 2049313Ssos 2059313Ssos#ifdef DEBUG 20672543Sjlemon if (ldebug(brk)) 207133845Sobrien printf(ARGS(brk, "%p"), (void *)(uintptr_t)args->dsend); 2089313Ssos#endif 20983221Smarcel old = (vm_offset_t)vm->vm_daddr + ctob(vm->vm_dsize); 21083221Smarcel new = (vm_offset_t)args->dsend; 21183221Smarcel tmp.nsize = (char *) new; 21283366Sjulian if (((caddr_t)new > vm->vm_daddr) && !obreak(td, &tmp)) 21383366Sjulian td->td_retval[0] = (long)new; 21483221Smarcel else 21583366Sjulian td->td_retval[0] = (long)old; 2169313Ssos 21783221Smarcel return 0; 2189313Ssos} 2199313Ssos 220158415Snetchild#if defined(__i386__) 221158415Snetchild/* XXX: what about amd64/linux32? */ 222133816Stjr 2239313Ssosint 22483366Sjulianlinux_uselib(struct thread *td, struct linux_uselib_args *args) 2259313Ssos{ 22683221Smarcel struct nameidata ni; 22783221Smarcel struct vnode *vp; 22883221Smarcel struct exec *a_out; 22983221Smarcel struct vattr attr; 23083221Smarcel vm_offset_t vmaddr; 23183221Smarcel unsigned long file_offset; 23283221Smarcel vm_offset_t buffer; 23383221Smarcel unsigned long bss_size; 234102814Siedowse char *library; 23583221Smarcel int error; 236160555Sjhb int locked, vfslocked; 2379313Ssos 238102814Siedowse LCONVPATHEXIST(td, args->library, &library); 23914331Speter 2409313Ssos#ifdef DEBUG 24172543Sjlemon if (ldebug(uselib)) 242102814Siedowse printf(ARGS(uselib, "%s"), library); 2439313Ssos#endif 2449313Ssos 24583221Smarcel a_out = NULL; 246160555Sjhb vfslocked = 0; 24783221Smarcel locked = 0; 24883221Smarcel vp = NULL; 24914114Speter 250160555Sjhb NDINIT(&ni, LOOKUP, ISOPEN | FOLLOW | LOCKLEAF | MPSAFE | AUDITVNODE1, 251160555Sjhb UIO_SYSSPACE, library, td); 25283221Smarcel error = namei(&ni); 253102814Siedowse LFREEPATH(library); 25483221Smarcel if (error) 25583221Smarcel goto cleanup; 2569313Ssos 25783221Smarcel vp = ni.ni_vp; 258160555Sjhb vfslocked = NDHASGIANT(&ni); 25983221Smarcel NDFREE(&ni, NDF_ONLY_PNBUF); 2609313Ssos 26183221Smarcel /* 26283221Smarcel * From here on down, we have a locked vnode that must be unlocked. 263160555Sjhb * XXX: The code below largely duplicates exec_check_permissions(). 26483221Smarcel */ 265160555Sjhb locked = 1; 26614114Speter 26783221Smarcel /* Writable? */ 26883221Smarcel if (vp->v_writecount) { 26983221Smarcel error = ETXTBSY; 27083221Smarcel goto cleanup; 27183221Smarcel } 2729313Ssos 27383221Smarcel /* Executable? */ 27491406Sjhb error = VOP_GETATTR(vp, &attr, td->td_ucred, td); 27583221Smarcel if (error) 27683221Smarcel goto cleanup; 2779313Ssos 27883221Smarcel if ((vp->v_mount->mnt_flag & MNT_NOEXEC) || 27983221Smarcel ((attr.va_mode & 0111) == 0) || (attr.va_type != VREG)) { 280160555Sjhb /* EACCESS is what exec(2) returns. */ 28183221Smarcel error = ENOEXEC; 28283221Smarcel goto cleanup; 28383221Smarcel } 2849313Ssos 28583221Smarcel /* Sensible size? */ 28683221Smarcel if (attr.va_size == 0) { 28783221Smarcel error = ENOEXEC; 28883221Smarcel goto cleanup; 28983221Smarcel } 2909313Ssos 29183221Smarcel /* Can we access it? */ 29291406Sjhb error = VOP_ACCESS(vp, VEXEC, td->td_ucred, td); 29383221Smarcel if (error) 29483221Smarcel goto cleanup; 2959313Ssos 29698209Srwatson /* 29798209Srwatson * XXX: This should use vn_open() so that it is properly authorized, 29898209Srwatson * and to reduce code redundancy all over the place here. 299160555Sjhb * XXX: Not really, it duplicates far more of exec_check_permissions() 300160555Sjhb * than vn_open(). 30198209Srwatson */ 302101189Srwatson#ifdef MAC 303101189Srwatson error = mac_check_vnode_open(td->td_ucred, vp, FREAD); 304101189Srwatson if (error) 305101189Srwatson goto cleanup; 306101189Srwatson#endif 307118047Sphk error = VOP_OPEN(vp, FREAD, td->td_ucred, td, -1); 30883221Smarcel if (error) 30983221Smarcel goto cleanup; 3109313Ssos 311103941Sjeff /* Pull in executable header into kernel_map */ 312103941Sjeff error = vm_mmap(kernel_map, (vm_offset_t *)&a_out, PAGE_SIZE, 313144501Sjhb VM_PROT_READ, VM_PROT_READ, 0, OBJT_VNODE, vp, 0); 31483221Smarcel if (error) 31583221Smarcel goto cleanup; 3169313Ssos 31783221Smarcel /* Is it a Linux binary ? */ 31883221Smarcel if (((a_out->a_magic >> 16) & 0xff) != 0x64) { 31983221Smarcel error = ENOEXEC; 32083221Smarcel goto cleanup; 32183221Smarcel } 3229313Ssos 32383221Smarcel /* 32483221Smarcel * While we are here, we should REALLY do some more checks 32583221Smarcel */ 32614114Speter 32783221Smarcel /* Set file/virtual offset based on a.out variant. */ 32883221Smarcel switch ((int)(a_out->a_magic & 0xffff)) { 32983221Smarcel case 0413: /* ZMAGIC */ 33083221Smarcel file_offset = 1024; 33183221Smarcel break; 33283221Smarcel case 0314: /* QMAGIC */ 33383221Smarcel file_offset = 0; 33483221Smarcel break; 33583221Smarcel default: 33683221Smarcel error = ENOEXEC; 33783221Smarcel goto cleanup; 33883221Smarcel } 3399313Ssos 34083221Smarcel bss_size = round_page(a_out->a_bss); 34114114Speter 34283221Smarcel /* Check various fields in header for validity/bounds. */ 34383221Smarcel if (a_out->a_text & PAGE_MASK || a_out->a_data & PAGE_MASK) { 34483221Smarcel error = ENOEXEC; 34583221Smarcel goto cleanup; 34683221Smarcel } 34714114Speter 34883221Smarcel /* text + data can't exceed file size */ 34983221Smarcel if (a_out->a_data + a_out->a_text > attr.va_size) { 35083221Smarcel error = EFAULT; 35183221Smarcel goto cleanup; 35283221Smarcel } 35314114Speter 35483221Smarcel /* 35583221Smarcel * text/data/bss must not exceed limits 35683221Smarcel * XXX - this is not complete. it should check current usage PLUS 35783221Smarcel * the resources needed by this library. 35883221Smarcel */ 359125454Sjhb PROC_LOCK(td->td_proc); 36084783Sps if (a_out->a_text > maxtsiz || 361125454Sjhb a_out->a_data + bss_size > lim_cur(td->td_proc, RLIMIT_DATA)) { 362125454Sjhb PROC_UNLOCK(td->td_proc); 36383221Smarcel error = ENOMEM; 36483221Smarcel goto cleanup; 36583221Smarcel } 366125454Sjhb PROC_UNLOCK(td->td_proc); 36714114Speter 368160555Sjhb /* 369160555Sjhb * Prevent more writers. 370160555Sjhb * XXX: Note that if any of the VM operations fail below we don't 371160555Sjhb * clear this flag. 372160555Sjhb */ 373101308Sjeff vp->v_vflag |= VV_TEXT; 37414114Speter 37583221Smarcel /* 376160555Sjhb * Lock no longer needed 377160555Sjhb */ 378160555Sjhb locked = 0; 379160555Sjhb VOP_UNLOCK(vp, 0, td); 380160555Sjhb VFS_UNLOCK_GIANT(vfslocked); 381160555Sjhb 382160555Sjhb /* 38383221Smarcel * Check if file_offset page aligned. Currently we cannot handle 38483221Smarcel * misalinged file offsets, and so we read in the entire image 38583221Smarcel * (what a waste). 38683221Smarcel */ 38783221Smarcel if (file_offset & PAGE_MASK) { 3889313Ssos#ifdef DEBUG 38983221Smarcel printf("uselib: Non page aligned binary %lu\n", file_offset); 3909313Ssos#endif 39183221Smarcel /* Map text+data read/write/execute */ 39214114Speter 39383221Smarcel /* a_entry is the load address and is page aligned */ 39483221Smarcel vmaddr = trunc_page(a_out->a_entry); 39514114Speter 39683221Smarcel /* get anon user mapping, read+write+execute */ 39783366Sjulian error = vm_map_find(&td->td_proc->p_vmspace->vm_map, NULL, 0, 39883366Sjulian &vmaddr, a_out->a_text + a_out->a_data, FALSE, VM_PROT_ALL, 39983221Smarcel VM_PROT_ALL, 0); 40083221Smarcel if (error) 40183221Smarcel goto cleanup; 4029313Ssos 40383221Smarcel /* map file into kernel_map */ 40483221Smarcel error = vm_mmap(kernel_map, &buffer, 40583221Smarcel round_page(a_out->a_text + a_out->a_data + file_offset), 406144501Sjhb VM_PROT_READ, VM_PROT_READ, 0, OBJT_VNODE, vp, 40783221Smarcel trunc_page(file_offset)); 40883221Smarcel if (error) 40983221Smarcel goto cleanup; 4109313Ssos 41183221Smarcel /* copy from kernel VM space to user space */ 412133816Stjr error = copyout(PTRIN(buffer + file_offset), 413111797Sdes (void *)vmaddr, a_out->a_text + a_out->a_data); 4149313Ssos 41583221Smarcel /* release temporary kernel space */ 41683221Smarcel vm_map_remove(kernel_map, buffer, buffer + 41783221Smarcel round_page(a_out->a_text + a_out->a_data + file_offset)); 4189313Ssos 41983221Smarcel if (error) 42083221Smarcel goto cleanup; 42183221Smarcel } else { 4229313Ssos#ifdef DEBUG 42383221Smarcel printf("uselib: Page aligned binary %lu\n", file_offset); 4249313Ssos#endif 42583221Smarcel /* 42683221Smarcel * for QMAGIC, a_entry is 20 bytes beyond the load address 42783221Smarcel * to skip the executable header 42883221Smarcel */ 42983221Smarcel vmaddr = trunc_page(a_out->a_entry); 43014114Speter 43183221Smarcel /* 43283221Smarcel * Map it all into the process's space as a single 43383221Smarcel * copy-on-write "data" segment. 43483221Smarcel */ 43583366Sjulian error = vm_mmap(&td->td_proc->p_vmspace->vm_map, &vmaddr, 43683221Smarcel a_out->a_text + a_out->a_data, VM_PROT_ALL, VM_PROT_ALL, 437144501Sjhb MAP_PRIVATE | MAP_FIXED, OBJT_VNODE, vp, file_offset); 43883221Smarcel if (error) 43983221Smarcel goto cleanup; 44083221Smarcel } 4419313Ssos#ifdef DEBUG 44283221Smarcel printf("mem=%08lx = %08lx %08lx\n", (long)vmaddr, ((long*)vmaddr)[0], 44383221Smarcel ((long*)vmaddr)[1]); 4449313Ssos#endif 44583221Smarcel if (bss_size != 0) { 44683221Smarcel /* Calculate BSS start address */ 44783221Smarcel vmaddr = trunc_page(a_out->a_entry) + a_out->a_text + 44883221Smarcel a_out->a_data; 44914114Speter 45083221Smarcel /* allocate some 'anon' space */ 45183366Sjulian error = vm_map_find(&td->td_proc->p_vmspace->vm_map, NULL, 0, 45283366Sjulian &vmaddr, bss_size, FALSE, VM_PROT_ALL, VM_PROT_ALL, 0); 45383221Smarcel if (error) 45483221Smarcel goto cleanup; 45583221Smarcel } 45614114Speter 45714114Spetercleanup: 45883221Smarcel /* Unlock vnode if needed */ 459160555Sjhb if (locked) { 46083366Sjulian VOP_UNLOCK(vp, 0, td); 461160555Sjhb VFS_UNLOCK_GIANT(vfslocked); 462160555Sjhb } 46314114Speter 46483221Smarcel /* Release the kernel mapping. */ 46583221Smarcel if (a_out) 46683221Smarcel vm_map_remove(kernel_map, (vm_offset_t)a_out, 46783221Smarcel (vm_offset_t)a_out + PAGE_SIZE); 46814114Speter 46983221Smarcel return error; 4709313Ssos} 4719313Ssos 472158415Snetchild#endif /* __i386__ */ 473133816Stjr 4749313Ssosint 47583366Sjulianlinux_select(struct thread *td, struct linux_select_args *args) 47614331Speter{ 477133747Stjr l_timeval ltv; 47883221Smarcel struct timeval tv0, tv1, utv, *tvp; 47983221Smarcel int error; 48014331Speter 4819313Ssos#ifdef DEBUG 48283221Smarcel if (ldebug(select)) 48383221Smarcel printf(ARGS(select, "%d, %p, %p, %p, %p"), args->nfds, 48483221Smarcel (void *)args->readfds, (void *)args->writefds, 48583221Smarcel (void *)args->exceptfds, (void *)args->timeout); 4869313Ssos#endif 48714331Speter 48883221Smarcel /* 48983221Smarcel * Store current time for computation of the amount of 49083221Smarcel * time left. 49183221Smarcel */ 49283221Smarcel if (args->timeout) { 493133747Stjr if ((error = copyin(args->timeout, <v, sizeof(ltv)))) 49483221Smarcel goto select_out; 495133747Stjr utv.tv_sec = ltv.tv_sec; 496133747Stjr utv.tv_usec = ltv.tv_usec; 49714331Speter#ifdef DEBUG 49883221Smarcel if (ldebug(select)) 499153775Strhodes printf(LMSG("incoming timeout (%jd/%ld)"), 500153775Strhodes (intmax_t)utv.tv_sec, utv.tv_usec); 50114331Speter#endif 50283221Smarcel 50383221Smarcel if (itimerfix(&utv)) { 50483221Smarcel /* 50583221Smarcel * The timeval was invalid. Convert it to something 50683221Smarcel * valid that will act as it does under Linux. 50783221Smarcel */ 50883221Smarcel utv.tv_sec += utv.tv_usec / 1000000; 50983221Smarcel utv.tv_usec %= 1000000; 51083221Smarcel if (utv.tv_usec < 0) { 51183221Smarcel utv.tv_sec -= 1; 51283221Smarcel utv.tv_usec += 1000000; 51383221Smarcel } 51483221Smarcel if (utv.tv_sec < 0) 51583221Smarcel timevalclear(&utv); 51683221Smarcel } 51783221Smarcel microtime(&tv0); 518102814Siedowse tvp = &utv; 519102814Siedowse } else 520102814Siedowse tvp = NULL; 52114331Speter 522102814Siedowse error = kern_select(td, args->nfds, args->readfds, args->writefds, 523102814Siedowse args->exceptfds, tvp); 524102814Siedowse 52514331Speter#ifdef DEBUG 52683221Smarcel if (ldebug(select)) 52772543Sjlemon printf(LMSG("real select returns %d"), error); 52814331Speter#endif 52983221Smarcel if (error) { 53083221Smarcel /* 53183221Smarcel * See fs/select.c in the Linux kernel. Without this, 53283221Smarcel * Maelstrom doesn't work. 53383221Smarcel */ 53483221Smarcel if (error == ERESTART) 53583221Smarcel error = EINTR; 53683221Smarcel goto select_out; 53783221Smarcel } 53814331Speter 53983221Smarcel if (args->timeout) { 54083366Sjulian if (td->td_retval[0]) { 54183221Smarcel /* 54283221Smarcel * Compute how much time was left of the timeout, 54383221Smarcel * by subtracting the current time and the time 54483221Smarcel * before we started the call, and subtracting 54583221Smarcel * that result from the user-supplied value. 54683221Smarcel */ 54783221Smarcel microtime(&tv1); 54883221Smarcel timevalsub(&tv1, &tv0); 54983221Smarcel timevalsub(&utv, &tv1); 55083221Smarcel if (utv.tv_sec < 0) 55183221Smarcel timevalclear(&utv); 55283221Smarcel } else 55383221Smarcel timevalclear(&utv); 55414331Speter#ifdef DEBUG 55583221Smarcel if (ldebug(select)) 556153775Strhodes printf(LMSG("outgoing timeout (%jd/%ld)"), 557153775Strhodes (intmax_t)utv.tv_sec, utv.tv_usec); 55814331Speter#endif 559133747Stjr ltv.tv_sec = utv.tv_sec; 560133747Stjr ltv.tv_usec = utv.tv_usec; 561133747Stjr if ((error = copyout(<v, args->timeout, sizeof(ltv)))) 56283221Smarcel goto select_out; 56383221Smarcel } 56414331Speter 56514331Speterselect_out: 56614331Speter#ifdef DEBUG 56783221Smarcel if (ldebug(select)) 56883221Smarcel printf(LMSG("select_out -> %d"), error); 56914331Speter#endif 57083221Smarcel return error; 5719313Ssos} 5729313Ssos 573111798Sdesint 57483366Sjulianlinux_mremap(struct thread *td, struct linux_mremap_args *args) 57537548Sjkh{ 57637548Sjkh struct munmap_args /* { 57737548Sjkh void *addr; 57837548Sjkh size_t len; 579111798Sdes } */ bsd_args; 58037548Sjkh int error = 0; 581111798Sdes 58237548Sjkh#ifdef DEBUG 58372543Sjlemon if (ldebug(mremap)) 58472543Sjlemon printf(ARGS(mremap, "%p, %08lx, %08lx, %08lx"), 585133845Sobrien (void *)(uintptr_t)args->addr, 586111798Sdes (unsigned long)args->old_len, 58772543Sjlemon (unsigned long)args->new_len, 58872543Sjlemon (unsigned long)args->flags); 58937548Sjkh#endif 59037548Sjkh args->new_len = round_page(args->new_len); 59137548Sjkh args->old_len = round_page(args->old_len); 59237548Sjkh 59337548Sjkh if (args->new_len > args->old_len) { 59483366Sjulian td->td_retval[0] = 0; 59537548Sjkh return ENOMEM; 59637548Sjkh } 59737548Sjkh 59837548Sjkh if (args->new_len < args->old_len) { 599133816Stjr bsd_args.addr = 600133816Stjr (caddr_t)((uintptr_t)args->addr + args->new_len); 60137548Sjkh bsd_args.len = args->old_len - args->new_len; 60283366Sjulian error = munmap(td, &bsd_args); 60337548Sjkh } 60437548Sjkh 605102963Sbde td->td_retval[0] = error ? 0 : (uintptr_t)args->addr; 60637548Sjkh return error; 60737548Sjkh} 60837548Sjkh 609103652Smdodd#define LINUX_MS_ASYNC 0x0001 610103652Smdodd#define LINUX_MS_INVALIDATE 0x0002 611103652Smdodd#define LINUX_MS_SYNC 0x0004 612103652Smdodd 61314331Speterint 61483366Sjulianlinux_msync(struct thread *td, struct linux_msync_args *args) 61514331Speter{ 61614331Speter struct msync_args bsd_args; 6179313Ssos 618133816Stjr bsd_args.addr = (caddr_t)(uintptr_t)args->addr; 619133816Stjr bsd_args.len = (uintptr_t)args->len; 620103652Smdodd bsd_args.flags = args->fl & ~LINUX_MS_SYNC; 62114331Speter 62283366Sjulian return msync(td, &bsd_args); 62314331Speter} 62414331Speter 6259313Ssosint 62683366Sjulianlinux_time(struct thread *td, struct linux_time_args *args) 6279313Ssos{ 62883221Smarcel struct timeval tv; 62983221Smarcel l_time_t tm; 63083221Smarcel int error; 6319313Ssos 6329313Ssos#ifdef DEBUG 63372543Sjlemon if (ldebug(time)) 63472543Sjlemon printf(ARGS(time, "*")); 6359313Ssos#endif 63683221Smarcel 63783221Smarcel microtime(&tv); 63883221Smarcel tm = tv.tv_sec; 639111797Sdes if (args->tm && (error = copyout(&tm, args->tm, sizeof(tm)))) 64083221Smarcel return error; 64183366Sjulian td->td_retval[0] = tm; 64283221Smarcel return 0; 6439313Ssos} 6449313Ssos 64583221Smarcelstruct l_times_argv { 64683221Smarcel l_long tms_utime; 64783221Smarcel l_long tms_stime; 64883221Smarcel l_long tms_cutime; 64983221Smarcel l_long tms_cstime; 6509313Ssos}; 6519313Ssos 65214381Speter#define CLK_TCK 100 /* Linux uses 100 */ 65374701Sgallatin 65414381Speter#define CONVTCK(r) (r.tv_sec * CLK_TCK + r.tv_usec / (1000000 / CLK_TCK)) 65514381Speter 6569313Ssosint 65783366Sjulianlinux_times(struct thread *td, struct linux_times_args *args) 6589313Ssos{ 659136152Sjhb struct timeval tv, utime, stime, cutime, cstime; 66083221Smarcel struct l_times_argv tms; 661136152Sjhb struct proc *p; 66283221Smarcel int error; 6639313Ssos 6649313Ssos#ifdef DEBUG 66572543Sjlemon if (ldebug(times)) 66672543Sjlemon printf(ARGS(times, "*")); 6679313Ssos#endif 66814381Speter 669159896Snetchild if (args->buf != NULL) { 670159896Snetchild p = td->td_proc; 671159896Snetchild PROC_LOCK(p); 672159896Snetchild calcru(p, &utime, &stime); 673159896Snetchild calccru(p, &cutime, &cstime); 674159896Snetchild PROC_UNLOCK(p); 67514381Speter 676159896Snetchild tms.tms_utime = CONVTCK(utime); 677159896Snetchild tms.tms_stime = CONVTCK(stime); 67814381Speter 679159896Snetchild tms.tms_cutime = CONVTCK(cutime); 680159896Snetchild tms.tms_cstime = CONVTCK(cstime); 68114381Speter 682159896Snetchild if ((error = copyout(&tms, args->buf, sizeof(tms)))) 683159896Snetchild return error; 684159896Snetchild } 68583221Smarcel 68683221Smarcel microuptime(&tv); 68783366Sjulian td->td_retval[0] = (int)CONVTCK(tv); 68883221Smarcel return 0; 6899313Ssos} 6909313Ssos 6919313Ssosint 69283366Sjulianlinux_newuname(struct thread *td, struct linux_newuname_args *args) 6939313Ssos{ 69483221Smarcel struct l_new_utsname utsname; 69587275Srwatson char osname[LINUX_MAX_UTSNAME]; 69687275Srwatson char osrelease[LINUX_MAX_UTSNAME]; 697118149Sdes char *p; 6989313Ssos 6999313Ssos#ifdef DEBUG 70072543Sjlemon if (ldebug(newuname)) 70172543Sjlemon printf(ARGS(newuname, "*")); 7029313Ssos#endif 70350345Smarcel 704112206Sjhb linux_get_osname(td, osname); 705112206Sjhb linux_get_osrelease(td, osrelease); 70650465Smarcel 70783221Smarcel bzero(&utsname, sizeof(utsname)); 708105359Srobert strlcpy(utsname.sysname, osname, LINUX_MAX_UTSNAME); 709105359Srobert getcredhostname(td->td_ucred, utsname.nodename, LINUX_MAX_UTSNAME); 710105359Srobert strlcpy(utsname.release, osrelease, LINUX_MAX_UTSNAME); 711105359Srobert strlcpy(utsname.version, version, LINUX_MAX_UTSNAME); 712118149Sdes for (p = utsname.version; *p != '\0'; ++p) 713118149Sdes if (*p == '\n') { 714118149Sdes *p = '\0'; 715118149Sdes break; 716118149Sdes } 717118149Sdes#ifdef __i386__ 718118149Sdes { 719118149Sdes const char *class; 720118149Sdes switch (cpu_class) { 721118149Sdes case CPUCLASS_686: 722118149Sdes class = "i686"; 723118149Sdes break; 724118149Sdes case CPUCLASS_586: 725118149Sdes class = "i586"; 726118149Sdes break; 727118149Sdes case CPUCLASS_486: 728118149Sdes class = "i486"; 729118149Sdes break; 730118149Sdes default: 731118149Sdes class = "i386"; 732118149Sdes } 733118149Sdes strlcpy(utsname.machine, class, LINUX_MAX_UTSNAME); 734118149Sdes } 735140213Sobrien#elif defined(__amd64__) /* XXX: Linux can change 'personality'. */ 736140213Sobrien#ifdef COMPAT_LINUX32 737140213Sobrien strlcpy(utsname.machine, "i686", LINUX_MAX_UTSNAME); 738118149Sdes#else 739140213Sobrien strlcpy(utsname.machine, "x86_64", LINUX_MAX_UTSNAME); 740140213Sobrien#endif /* COMPAT_LINUX32 */ 741140213Sobrien#else /* something other than i386 or amd64 - assume we and Linux agree */ 742105359Srobert strlcpy(utsname.machine, machine, LINUX_MAX_UTSNAME); 743140213Sobrien#endif /* __i386__ */ 744105359Srobert strlcpy(utsname.domainname, domainname, LINUX_MAX_UTSNAME); 74550345Smarcel 746111797Sdes return (copyout(&utsname, args->buf, sizeof(utsname))); 7479313Ssos} 7489313Ssos 749140214Sobrien#if defined(__i386__) || (defined(__amd64__) && defined(COMPAT_LINUX32)) 75083221Smarcelstruct l_utimbuf { 75183221Smarcel l_time_t l_actime; 75283221Smarcel l_time_t l_modtime; 75314381Speter}; 7549313Ssos 7559313Ssosint 75683366Sjulianlinux_utime(struct thread *td, struct linux_utime_args *args) 7579313Ssos{ 75883221Smarcel struct timeval tv[2], *tvp; 75983221Smarcel struct l_utimbuf lut; 760102814Siedowse char *fname; 76183221Smarcel int error; 7629313Ssos 763102814Siedowse LCONVPATHEXIST(td, args->fname, &fname); 76414331Speter 7659313Ssos#ifdef DEBUG 76672543Sjlemon if (ldebug(utime)) 767102814Siedowse printf(ARGS(utime, "%s, *"), fname); 7689313Ssos#endif 76914381Speter 77083221Smarcel if (args->times) { 771111797Sdes if ((error = copyin(args->times, &lut, sizeof lut))) { 772102814Siedowse LFREEPATH(fname); 77383221Smarcel return error; 774102814Siedowse } 77583221Smarcel tv[0].tv_sec = lut.l_actime; 77683221Smarcel tv[0].tv_usec = 0; 77783221Smarcel tv[1].tv_sec = lut.l_modtime; 77883221Smarcel tv[1].tv_usec = 0; 779102814Siedowse tvp = tv; 78083221Smarcel } else 781102814Siedowse tvp = NULL; 78283221Smarcel 783102814Siedowse error = kern_utimes(td, fname, UIO_SYSSPACE, tvp, UIO_SYSSPACE); 784102814Siedowse LFREEPATH(fname); 785102814Siedowse return (error); 7869313Ssos} 787133816Stjr#endif /* __i386__ || (__amd64__ && COMPAT_LINUX32) */ 7889313Ssos 78944384Sjulian#define __WCLONE 0x80000000 79044384Sjulian 7919313Ssosint 79283366Sjulianlinux_waitpid(struct thread *td, struct linux_waitpid_args *args) 7939313Ssos{ 794127140Sjhb int error, options, tmpstat; 7959313Ssos 7969313Ssos#ifdef DEBUG 79772543Sjlemon if (ldebug(waitpid)) 79872543Sjlemon printf(ARGS(waitpid, "%d, %p, %d"), 79972543Sjlemon args->pid, (void *)args->status, args->options); 8009313Ssos#endif 8019313Ssos 802127140Sjhb options = (args->options & (WNOHANG | WUNTRACED)); 80383221Smarcel /* WLINUXCLONE should be equal to __WCLONE, but we make sure */ 80483221Smarcel if (args->options & __WCLONE) 805127140Sjhb options |= WLINUXCLONE; 80643208Sjulian 807127140Sjhb error = kern_wait(td, args->pid, &tmpstat, options, NULL); 808127140Sjhb if (error) 80983221Smarcel return error; 81083221Smarcel 81183221Smarcel if (args->status) { 81283221Smarcel tmpstat &= 0xffff; 81383221Smarcel if (WIFSIGNALED(tmpstat)) 81483221Smarcel tmpstat = (tmpstat & 0xffffff80) | 81583221Smarcel BSD_TO_LINUX_SIGNAL(WTERMSIG(tmpstat)); 81683221Smarcel else if (WIFSTOPPED(tmpstat)) 81783221Smarcel tmpstat = (tmpstat & 0xffff00ff) | 81883221Smarcel (BSD_TO_LINUX_SIGNAL(WSTOPSIG(tmpstat)) << 8); 819111797Sdes return copyout(&tmpstat, args->status, sizeof(int)); 82083221Smarcel } 82183221Smarcel 82214331Speter return 0; 8239313Ssos} 8249313Ssos 82514331Speterint 82683366Sjulianlinux_wait4(struct thread *td, struct linux_wait4_args *args) 8279313Ssos{ 828127140Sjhb int error, options, tmpstat; 829136152Sjhb struct rusage ru, *rup; 830113613Sjhb struct proc *p; 8319313Ssos 8329313Ssos#ifdef DEBUG 83372543Sjlemon if (ldebug(wait4)) 83472543Sjlemon printf(ARGS(wait4, "%d, %p, %d, %p"), 83572543Sjlemon args->pid, (void *)args->status, args->options, 83672543Sjlemon (void *)args->rusage); 8379313Ssos#endif 8389313Ssos 839127140Sjhb options = (args->options & (WNOHANG | WUNTRACED)); 84083221Smarcel /* WLINUXCLONE should be equal to __WCLONE, but we make sure */ 84183221Smarcel if (args->options & __WCLONE) 842127140Sjhb options |= WLINUXCLONE; 84314331Speter 844136152Sjhb if (args->rusage != NULL) 845136152Sjhb rup = &ru; 846136152Sjhb else 847136152Sjhb rup = NULL; 848136152Sjhb error = kern_wait(td, args->pid, &tmpstat, options, rup); 849127140Sjhb if (error) 85083221Smarcel return error; 85114331Speter 852113613Sjhb p = td->td_proc; 853113613Sjhb PROC_LOCK(p); 854151316Sdavidxu sigqueue_delete(&p->p_sigqueue, SIGCHLD); 855113613Sjhb PROC_UNLOCK(p); 85683221Smarcel 85783221Smarcel if (args->status) { 85883221Smarcel tmpstat &= 0xffff; 85983221Smarcel if (WIFSIGNALED(tmpstat)) 86083221Smarcel tmpstat = (tmpstat & 0xffffff80) | 86183221Smarcel BSD_TO_LINUX_SIGNAL(WTERMSIG(tmpstat)); 86283221Smarcel else if (WIFSTOPPED(tmpstat)) 86383221Smarcel tmpstat = (tmpstat & 0xffff00ff) | 86483221Smarcel (BSD_TO_LINUX_SIGNAL(WSTOPSIG(tmpstat)) << 8); 865127140Sjhb error = copyout(&tmpstat, args->status, sizeof(int)); 86683221Smarcel } 867127140Sjhb if (args->rusage != NULL && error == 0) 868127140Sjhb error = copyout(&ru, args->rusage, sizeof(ru)); 86983221Smarcel 870127140Sjhb return (error); 8719313Ssos} 87213420Ssos 87314331Speterint 87483366Sjulianlinux_mknod(struct thread *td, struct linux_mknod_args *args) 87513420Ssos{ 876102814Siedowse char *path; 877102814Siedowse int error; 87814331Speter 879102814Siedowse LCONVPATHCREAT(td, args->path, &path); 88014331Speter 88114331Speter#ifdef DEBUG 88272543Sjlemon if (ldebug(mknod)) 883102814Siedowse printf(ARGS(mknod, "%s, %d, %d"), path, args->mode, args->dev); 88414331Speter#endif 88514331Speter 886102814Siedowse if (args->mode & S_IFIFO) 887102814Siedowse error = kern_mkfifo(td, path, UIO_SYSSPACE, args->mode); 888102814Siedowse else 889102814Siedowse error = kern_mknod(td, path, UIO_SYSSPACE, args->mode, 890102814Siedowse args->dev); 891102814Siedowse LFREEPATH(path); 892102814Siedowse return (error); 89313420Ssos} 89414331Speter 89514331Speter/* 89614331Speter * UGH! This is just about the dumbest idea I've ever heard!! 89714331Speter */ 89814331Speterint 89983366Sjulianlinux_personality(struct thread *td, struct linux_personality_args *args) 90014331Speter{ 90114331Speter#ifdef DEBUG 90272543Sjlemon if (ldebug(personality)) 903113579Sjhb printf(ARGS(personality, "%lu"), (unsigned long)args->per); 90414331Speter#endif 90514331Speter if (args->per != 0) 90614331Speter return EINVAL; 90714331Speter 90814331Speter /* Yes Jim, it's still a Linux... */ 90983366Sjulian td->td_retval[0] = 0; 91014331Speter return 0; 91114331Speter} 91214331Speter 913133749Stjrstruct l_itimerval { 914133749Stjr l_timeval it_interval; 915133749Stjr l_timeval it_value; 916133749Stjr}; 917133749Stjr 918140832Ssobomax#define B2L_ITIMERVAL(bip, lip) \ 919140832Ssobomax (bip)->it_interval.tv_sec = (lip)->it_interval.tv_sec; \ 920140832Ssobomax (bip)->it_interval.tv_usec = (lip)->it_interval.tv_usec; \ 921140832Ssobomax (bip)->it_value.tv_sec = (lip)->it_value.tv_sec; \ 922140832Ssobomax (bip)->it_value.tv_usec = (lip)->it_value.tv_usec; 923140832Ssobomax 92414331Speterint 925133749Stjrlinux_setitimer(struct thread *td, struct linux_setitimer_args *uap) 92614331Speter{ 92714331Speter int error; 928140832Ssobomax struct l_itimerval ls; 929140832Ssobomax struct itimerval aitv, oitv; 93014331Speter 93114331Speter#ifdef DEBUG 93272543Sjlemon if (ldebug(setitimer)) 93372543Sjlemon printf(ARGS(setitimer, "%p, %p"), 934133840Sobrien (void *)uap->itv, (void *)uap->oitv); 93514331Speter#endif 936140832Ssobomax 937140832Ssobomax if (uap->itv == NULL) { 938140832Ssobomax uap->itv = uap->oitv; 939140832Ssobomax return (linux_getitimer(td, (struct linux_getitimer_args *)uap)); 94014331Speter } 941140832Ssobomax 942140832Ssobomax error = copyin(uap->itv, &ls, sizeof(ls)); 943133749Stjr if (error != 0) 944133749Stjr return (error); 945140832Ssobomax B2L_ITIMERVAL(&aitv, &ls); 946140832Ssobomax#ifdef DEBUG 947140832Ssobomax if (ldebug(setitimer)) { 948153775Strhodes printf("setitimer: value: sec: %jd, usec: %ld\n", 949153775Strhodes (intmax_t)aitv.it_value.tv_sec, aitv.it_value.tv_usec); 950153775Strhodes printf("setitimer: interval: sec: %jd, usec: %ld\n", 951153775Strhodes (intmax_t)aitv.it_interval.tv_sec, aitv.it_interval.tv_usec); 952133749Stjr } 953140832Ssobomax#endif 954140832Ssobomax error = kern_setitimer(td, uap->which, &aitv, &oitv); 955140832Ssobomax if (error != 0 || uap->oitv == NULL) 956140832Ssobomax return (error); 957140832Ssobomax B2L_ITIMERVAL(&ls, &oitv); 958140832Ssobomax 959140832Ssobomax return (copyout(&ls, uap->oitv, sizeof(ls))); 96014331Speter} 96114331Speter 96214331Speterint 963133749Stjrlinux_getitimer(struct thread *td, struct linux_getitimer_args *uap) 96414331Speter{ 965133749Stjr int error; 966140832Ssobomax struct l_itimerval ls; 967140832Ssobomax struct itimerval aitv; 968133749Stjr 96914331Speter#ifdef DEBUG 97072543Sjlemon if (ldebug(getitimer)) 971133840Sobrien printf(ARGS(getitimer, "%p"), (void *)uap->itv); 97214331Speter#endif 973140832Ssobomax error = kern_getitimer(td, uap->which, &aitv); 974133749Stjr if (error != 0) 975133749Stjr return (error); 976140832Ssobomax B2L_ITIMERVAL(&ls, &aitv); 977140832Ssobomax return (copyout(&ls, uap->itv, sizeof(ls))); 97814331Speter} 97930837Skato 98030837Skatoint 98183366Sjulianlinux_nice(struct thread *td, struct linux_nice_args *args) 98230837Skato{ 98330837Skato struct setpriority_args bsd_args; 98430837Skato 98530837Skato bsd_args.which = PRIO_PROCESS; 98630837Skato bsd_args.who = 0; /* current process */ 98730837Skato bsd_args.prio = args->inc; 98883366Sjulian return setpriority(td, &bsd_args); 98930837Skato} 99030837Skato 99142185Ssosint 99283366Sjulianlinux_setgroups(struct thread *td, struct linux_setgroups_args *args) 99342185Ssos{ 99477183Srwatson struct ucred *newcred, *oldcred; 99583221Smarcel l_gid_t linux_gidset[NGROUPS]; 99650350Smarcel gid_t *bsd_gidset; 99750350Smarcel int ngrp, error; 99894621Sjhb struct proc *p; 99942185Ssos 100083221Smarcel ngrp = args->gidsetsize; 1001121302Stjr if (ngrp < 0 || ngrp >= NGROUPS) 100294621Sjhb return (EINVAL); 1003111797Sdes error = copyin(args->grouplist, linux_gidset, ngrp * sizeof(l_gid_t)); 100494621Sjhb if (error) 100594621Sjhb return (error); 100694621Sjhb newcred = crget(); 100794621Sjhb p = td->td_proc; 100894621Sjhb PROC_LOCK(p); 100994621Sjhb oldcred = p->p_ucred; 101042185Ssos 101150350Smarcel /* 101250350Smarcel * cr_groups[0] holds egid. Setting the whole set from 101350350Smarcel * the supplied set will cause egid to be changed too. 101450350Smarcel * Keep cr_groups[0] unchanged to prevent that. 101550350Smarcel */ 101642185Ssos 1017132653Scperciva if ((error = suser_cred(oldcred, SUSER_ALLOWJAIL)) != 0) { 101894621Sjhb PROC_UNLOCK(p); 101994621Sjhb crfree(newcred); 102050350Smarcel return (error); 102194621Sjhb } 102242185Ssos 102394621Sjhb crcopy(newcred, oldcred); 102450350Smarcel if (ngrp > 0) { 102577183Srwatson newcred->cr_ngroups = ngrp + 1; 102650350Smarcel 102777183Srwatson bsd_gidset = newcred->cr_groups; 102850350Smarcel ngrp--; 102950350Smarcel while (ngrp >= 0) { 103050350Smarcel bsd_gidset[ngrp + 1] = linux_gidset[ngrp]; 103150350Smarcel ngrp--; 103250350Smarcel } 103350350Smarcel } 103450350Smarcel else 103577183Srwatson newcred->cr_ngroups = 1; 103650350Smarcel 103794621Sjhb setsugid(p); 103894621Sjhb p->p_ucred = newcred; 103994621Sjhb PROC_UNLOCK(p); 104077183Srwatson crfree(oldcred); 104150350Smarcel return (0); 104242185Ssos} 104342185Ssos 104442185Ssosint 104583366Sjulianlinux_getgroups(struct thread *td, struct linux_getgroups_args *args) 104642185Ssos{ 104777183Srwatson struct ucred *cred; 104883221Smarcel l_gid_t linux_gidset[NGROUPS]; 104950350Smarcel gid_t *bsd_gidset; 105050350Smarcel int bsd_gidsetsz, ngrp, error; 105142185Ssos 105294454Sjhb cred = td->td_ucred; 105377183Srwatson bsd_gidset = cred->cr_groups; 105477183Srwatson bsd_gidsetsz = cred->cr_ngroups - 1; 105542185Ssos 105650350Smarcel /* 105750350Smarcel * cr_groups[0] holds egid. Returning the whole set 105850350Smarcel * here will cause a duplicate. Exclude cr_groups[0] 105950350Smarcel * to prevent that. 106050350Smarcel */ 106142185Ssos 106283221Smarcel if ((ngrp = args->gidsetsize) == 0) { 106383366Sjulian td->td_retval[0] = bsd_gidsetsz; 106450350Smarcel return (0); 106550350Smarcel } 106642185Ssos 106750546Smarcel if (ngrp < bsd_gidsetsz) 106850350Smarcel return (EINVAL); 106942185Ssos 107050546Smarcel ngrp = 0; 107150350Smarcel while (ngrp < bsd_gidsetsz) { 107250546Smarcel linux_gidset[ngrp] = bsd_gidset[ngrp + 1]; 107350350Smarcel ngrp++; 107450350Smarcel } 107550350Smarcel 1076111797Sdes if ((error = copyout(linux_gidset, args->grouplist, 107783221Smarcel ngrp * sizeof(l_gid_t)))) 107850350Smarcel return (error); 107950350Smarcel 108083366Sjulian td->td_retval[0] = ngrp; 108150350Smarcel return (0); 108242185Ssos} 108349626Smarcel 108449626Smarcelint 108583366Sjulianlinux_setrlimit(struct thread *td, struct linux_setrlimit_args *args) 108649626Smarcel{ 1087102814Siedowse struct rlimit bsd_rlim; 108883221Smarcel struct l_rlimit rlim; 1089102814Siedowse u_int which; 109065099Smarcel int error; 109149842Smarcel 109249626Smarcel#ifdef DEBUG 109372543Sjlemon if (ldebug(setrlimit)) 109472543Sjlemon printf(ARGS(setrlimit, "%d, %p"), 109583221Smarcel args->resource, (void *)args->rlim); 109649626Smarcel#endif 109749626Smarcel 109883221Smarcel if (args->resource >= LINUX_RLIM_NLIMITS) 109965099Smarcel return (EINVAL); 110049626Smarcel 1101102814Siedowse which = linux_to_bsd_resource[args->resource]; 1102102814Siedowse if (which == -1) 110365099Smarcel return (EINVAL); 110449626Smarcel 1105111797Sdes error = copyin(args->rlim, &rlim, sizeof(rlim)); 110665099Smarcel if (error) 110765099Smarcel return (error); 110849626Smarcel 1109102814Siedowse bsd_rlim.rlim_cur = (rlim_t)rlim.rlim_cur; 1110102814Siedowse bsd_rlim.rlim_max = (rlim_t)rlim.rlim_max; 1111125454Sjhb return (kern_setrlimit(td, which, &bsd_rlim)); 111249626Smarcel} 111349626Smarcel 111449626Smarcelint 111583366Sjulianlinux_old_getrlimit(struct thread *td, struct linux_old_getrlimit_args *args) 111649626Smarcel{ 111783221Smarcel struct l_rlimit rlim; 1118102814Siedowse struct proc *p = td->td_proc; 1119125454Sjhb struct rlimit bsd_rlim; 1120102814Siedowse u_int which; 112149842Smarcel 112249626Smarcel#ifdef DEBUG 112383221Smarcel if (ldebug(old_getrlimit)) 112483221Smarcel printf(ARGS(old_getrlimit, "%d, %p"), 112583221Smarcel args->resource, (void *)args->rlim); 112649626Smarcel#endif 112749626Smarcel 112883221Smarcel if (args->resource >= LINUX_RLIM_NLIMITS) 112965099Smarcel return (EINVAL); 113049626Smarcel 1131102814Siedowse which = linux_to_bsd_resource[args->resource]; 1132102814Siedowse if (which == -1) 113365099Smarcel return (EINVAL); 113449626Smarcel 1135125454Sjhb PROC_LOCK(p); 1136125454Sjhb lim_rlimit(p, which, &bsd_rlim); 1137125454Sjhb PROC_UNLOCK(p); 1138125454Sjhb 1139140214Sobrien#ifdef COMPAT_LINUX32 1140140214Sobrien rlim.rlim_cur = (unsigned int)bsd_rlim.rlim_cur; 1141140214Sobrien if (rlim.rlim_cur == UINT_MAX) 1142140214Sobrien rlim.rlim_cur = INT_MAX; 1143140214Sobrien rlim.rlim_max = (unsigned int)bsd_rlim.rlim_max; 1144140214Sobrien if (rlim.rlim_max == UINT_MAX) 1145140214Sobrien rlim.rlim_max = INT_MAX; 1146140214Sobrien#else 1147125454Sjhb rlim.rlim_cur = (unsigned long)bsd_rlim.rlim_cur; 114865106Smarcel if (rlim.rlim_cur == ULONG_MAX) 114965106Smarcel rlim.rlim_cur = LONG_MAX; 1150125454Sjhb rlim.rlim_max = (unsigned long)bsd_rlim.rlim_max; 115165106Smarcel if (rlim.rlim_max == ULONG_MAX) 115265106Smarcel rlim.rlim_max = LONG_MAX; 1153133816Stjr#endif 1154111797Sdes return (copyout(&rlim, args->rlim, sizeof(rlim))); 115549626Smarcel} 115683221Smarcel 115783221Smarcelint 115883366Sjulianlinux_getrlimit(struct thread *td, struct linux_getrlimit_args *args) 115983221Smarcel{ 116083221Smarcel struct l_rlimit rlim; 1161102814Siedowse struct proc *p = td->td_proc; 1162125454Sjhb struct rlimit bsd_rlim; 1163102814Siedowse u_int which; 116483221Smarcel 116583221Smarcel#ifdef DEBUG 116683221Smarcel if (ldebug(getrlimit)) 116783221Smarcel printf(ARGS(getrlimit, "%d, %p"), 116883221Smarcel args->resource, (void *)args->rlim); 116983221Smarcel#endif 117083221Smarcel 117183221Smarcel if (args->resource >= LINUX_RLIM_NLIMITS) 117283221Smarcel return (EINVAL); 117383221Smarcel 1174102814Siedowse which = linux_to_bsd_resource[args->resource]; 1175102814Siedowse if (which == -1) 117683221Smarcel return (EINVAL); 117783221Smarcel 1178125454Sjhb PROC_LOCK(p); 1179125454Sjhb lim_rlimit(p, which, &bsd_rlim); 1180125454Sjhb PROC_UNLOCK(p); 1181125454Sjhb 1182125454Sjhb rlim.rlim_cur = (l_ulong)bsd_rlim.rlim_cur; 1183125454Sjhb rlim.rlim_max = (l_ulong)bsd_rlim.rlim_max; 1184111797Sdes return (copyout(&rlim, args->rlim, sizeof(rlim))); 118583221Smarcel} 118649849Smarcel 118749849Smarcelint 118883366Sjulianlinux_sched_setscheduler(struct thread *td, 118983221Smarcel struct linux_sched_setscheduler_args *args) 119049849Smarcel{ 119149849Smarcel struct sched_setscheduler_args bsd; 119249849Smarcel 119349849Smarcel#ifdef DEBUG 119472543Sjlemon if (ldebug(sched_setscheduler)) 119572543Sjlemon printf(ARGS(sched_setscheduler, "%d, %d, %p"), 119683221Smarcel args->pid, args->policy, (const void *)args->param); 119749849Smarcel#endif 119849849Smarcel 119983221Smarcel switch (args->policy) { 120049849Smarcel case LINUX_SCHED_OTHER: 120149849Smarcel bsd.policy = SCHED_OTHER; 120249849Smarcel break; 120349849Smarcel case LINUX_SCHED_FIFO: 120449849Smarcel bsd.policy = SCHED_FIFO; 120549849Smarcel break; 120649849Smarcel case LINUX_SCHED_RR: 120749849Smarcel bsd.policy = SCHED_RR; 120849849Smarcel break; 120949849Smarcel default: 121049849Smarcel return EINVAL; 121149849Smarcel } 121249849Smarcel 121383221Smarcel bsd.pid = args->pid; 121483221Smarcel bsd.param = (struct sched_param *)args->param; 121583366Sjulian return sched_setscheduler(td, &bsd); 121649849Smarcel} 121749849Smarcel 121849849Smarcelint 121983366Sjulianlinux_sched_getscheduler(struct thread *td, 122083221Smarcel struct linux_sched_getscheduler_args *args) 122149849Smarcel{ 122249849Smarcel struct sched_getscheduler_args bsd; 122349849Smarcel int error; 122449849Smarcel 122549849Smarcel#ifdef DEBUG 122672543Sjlemon if (ldebug(sched_getscheduler)) 122783221Smarcel printf(ARGS(sched_getscheduler, "%d"), args->pid); 122849849Smarcel#endif 122949849Smarcel 123083221Smarcel bsd.pid = args->pid; 123183366Sjulian error = sched_getscheduler(td, &bsd); 123249849Smarcel 123383366Sjulian switch (td->td_retval[0]) { 123449849Smarcel case SCHED_OTHER: 123583366Sjulian td->td_retval[0] = LINUX_SCHED_OTHER; 123649849Smarcel break; 123749849Smarcel case SCHED_FIFO: 123883366Sjulian td->td_retval[0] = LINUX_SCHED_FIFO; 123949849Smarcel break; 124049849Smarcel case SCHED_RR: 124183366Sjulian td->td_retval[0] = LINUX_SCHED_RR; 124249849Smarcel break; 124349849Smarcel } 124449849Smarcel 124549849Smarcel return error; 124649849Smarcel} 124772538Sjlemon 124875053Salcint 124983366Sjulianlinux_sched_get_priority_max(struct thread *td, 125083221Smarcel struct linux_sched_get_priority_max_args *args) 125175053Salc{ 125275053Salc struct sched_get_priority_max_args bsd; 125375053Salc 125475053Salc#ifdef DEBUG 125575053Salc if (ldebug(sched_get_priority_max)) 125683221Smarcel printf(ARGS(sched_get_priority_max, "%d"), args->policy); 125775053Salc#endif 125875053Salc 125983221Smarcel switch (args->policy) { 126075053Salc case LINUX_SCHED_OTHER: 126175053Salc bsd.policy = SCHED_OTHER; 126275053Salc break; 126375053Salc case LINUX_SCHED_FIFO: 126475053Salc bsd.policy = SCHED_FIFO; 126575053Salc break; 126675053Salc case LINUX_SCHED_RR: 126775053Salc bsd.policy = SCHED_RR; 126875053Salc break; 126975053Salc default: 127075053Salc return EINVAL; 127175053Salc } 127283366Sjulian return sched_get_priority_max(td, &bsd); 127375053Salc} 127475053Salc 127575053Salcint 127683366Sjulianlinux_sched_get_priority_min(struct thread *td, 127783221Smarcel struct linux_sched_get_priority_min_args *args) 127875053Salc{ 127975053Salc struct sched_get_priority_min_args bsd; 128075053Salc 128175053Salc#ifdef DEBUG 128275053Salc if (ldebug(sched_get_priority_min)) 128383221Smarcel printf(ARGS(sched_get_priority_min, "%d"), args->policy); 128475053Salc#endif 128575053Salc 128683221Smarcel switch (args->policy) { 128775053Salc case LINUX_SCHED_OTHER: 128875053Salc bsd.policy = SCHED_OTHER; 128975053Salc break; 129075053Salc case LINUX_SCHED_FIFO: 129175053Salc bsd.policy = SCHED_FIFO; 129275053Salc break; 129375053Salc case LINUX_SCHED_RR: 129475053Salc bsd.policy = SCHED_RR; 129575053Salc break; 129675053Salc default: 129775053Salc return EINVAL; 129875053Salc } 129983366Sjulian return sched_get_priority_min(td, &bsd); 130075053Salc} 130175053Salc 130272538Sjlemon#define REBOOT_CAD_ON 0x89abcdef 130372538Sjlemon#define REBOOT_CAD_OFF 0 130472538Sjlemon#define REBOOT_HALT 0xcdef0123 130572538Sjlemon 130672538Sjlemonint 130783366Sjulianlinux_reboot(struct thread *td, struct linux_reboot_args *args) 130872538Sjlemon{ 130972538Sjlemon struct reboot_args bsd_args; 131072538Sjlemon 131172538Sjlemon#ifdef DEBUG 131272538Sjlemon if (ldebug(reboot)) 131383221Smarcel printf(ARGS(reboot, "0x%x"), args->cmd); 131472538Sjlemon#endif 131583221Smarcel if (args->cmd == REBOOT_CAD_ON || args->cmd == REBOOT_CAD_OFF) 131672538Sjlemon return (0); 131783221Smarcel bsd_args.opt = (args->cmd == REBOOT_HALT) ? RB_HALT : 0; 131883366Sjulian return (reboot(td, &bsd_args)); 131972538Sjlemon} 132083221Smarcel 132189717Sgallatin 132283221Smarcel/* 132383221Smarcel * The FreeBSD native getpid(2), getgid(2) and getuid(2) also modify 1324130344Sphk * td->td_retval[1] when COMPAT_43 is defined. This 132583221Smarcel * globbers registers that are assumed to be preserved. The following 132683221Smarcel * lightweight syscalls fixes this. See also linux_getgid16() and 132783221Smarcel * linux_getuid16() in linux_uid16.c. 132883221Smarcel * 132983221Smarcel * linux_getpid() - MP SAFE 133083221Smarcel * linux_getgid() - MP SAFE 133183221Smarcel * linux_getuid() - MP SAFE 133283221Smarcel */ 133383221Smarcel 133483221Smarcelint 133583366Sjulianlinux_getpid(struct thread *td, struct linux_getpid_args *args) 133683221Smarcel{ 1337161317Snetchild#ifdef __i386__ 1338161310Snetchild struct linux_emuldata *em; 133983366Sjulian 1340161310Snetchild em = em_find(td->td_proc, EMUL_UNLOCKED); 1341161310Snetchild 1342161310Snetchild KASSERT(em != NULL, ("getpid: emuldata not found.\n")); 1343161310Snetchild 1344161310Snetchild td->td_retval[0] = em->shared->group_pid; 1345161310Snetchild EMUL_UNLOCK(&emul_lock); 1346161317Snetchild#else 1347161317Snetchild td->td_retval[0] = td->td_proc->p_pid; 1348161317Snetchild#endif 1349161310Snetchild return (0); 1350161310Snetchild} 1351161310Snetchild 1352161310Snetchildint 1353161310Snetchildlinux_gettid(struct thread *td, struct linux_gettid_args *args) 1354161310Snetchild{ 1355161310Snetchild#ifdef DEBUG 1356161310Snetchild if (ldebug(gettid)) 1357161310Snetchild printf(ARGS(gettid, "")); 1358161310Snetchild#endif 1359161310Snetchild 136083366Sjulian td->td_retval[0] = td->td_proc->p_pid; 136183221Smarcel return (0); 136283221Smarcel} 136383221Smarcel 1364161310Snetchild 136583221Smarcelint 1366161310Snetchildlinux_getppid(struct thread *td, struct linux_getppid_args *args) 1367161310Snetchild{ 1368161317Snetchild#ifdef __i386__ 1369161310Snetchild struct linux_emuldata *em; 1370161310Snetchild struct proc *p, *pp; 1371161310Snetchild 1372161310Snetchild em = em_find(td->td_proc, EMUL_UNLOCKED); 1373161310Snetchild 1374161310Snetchild KASSERT(em != NULL, ("getppid: process emuldata not found.\n")); 1375161310Snetchild 1376161310Snetchild /* find the group leader */ 1377161310Snetchild p = pfind(em->shared->group_pid); 1378161310Snetchild 1379161310Snetchild if (p == NULL) { 1380161310Snetchild#ifdef DEBUG 1381161310Snetchild printf(LMSG("parent process not found.\n")); 1382161310Snetchild#endif 1383161310Snetchild return (0); 1384161310Snetchild } 1385161310Snetchild 1386161310Snetchild pp = p->p_pptr; /* switch to parent */ 1387161310Snetchild PROC_LOCK(pp); 1388161310Snetchild PROC_UNLOCK(p); 1389161310Snetchild 1390161310Snetchild /* if its also linux process */ 1391161310Snetchild if (pp->p_sysent == &elf_linux_sysvec) { 1392161310Snetchild em = em_find(pp, EMUL_LOCKED); 1393161310Snetchild KASSERT(em != NULL, ("getppid: parent emuldata not found.\n")); 1394161310Snetchild 1395161310Snetchild td->td_retval[0] = em->shared->group_pid; 1396161310Snetchild } else 1397161310Snetchild td->td_retval[0] = pp->p_pid; 1398161310Snetchild 1399161310Snetchild EMUL_UNLOCK(&emul_lock); 1400161310Snetchild PROC_UNLOCK(pp); 1401161317Snetchild#else 1402161317Snetchild return getppid(td, (struct getppid_args *) args); 1403161317Snetchild#endif 1404161310Snetchild return (0); 1405161310Snetchild} 1406161310Snetchild 1407161310Snetchildint 140883366Sjulianlinux_getgid(struct thread *td, struct linux_getgid_args *args) 140983221Smarcel{ 141083366Sjulian 141194454Sjhb td->td_retval[0] = td->td_ucred->cr_rgid; 141283221Smarcel return (0); 141383221Smarcel} 141483221Smarcel 141583221Smarcelint 141683366Sjulianlinux_getuid(struct thread *td, struct linux_getuid_args *args) 141783221Smarcel{ 141883366Sjulian 141994454Sjhb td->td_retval[0] = td->td_ucred->cr_ruid; 142083221Smarcel return (0); 142183221Smarcel} 142283503Smr 142389717Sgallatin 142483503Smrint 142583503Smrlinux_getsid(struct thread *td, struct linux_getsid_args *args) 142683503Smr{ 142783503Smr struct getsid_args bsd; 142883503Smr bsd.pid = args->pid; 142983503Smr return getsid(td, &bsd); 143083503Smr} 1431143197Ssobomax 1432143197Ssobomaxint 1433143197Ssobomaxlinux_nosys(struct thread *td, struct nosys_args *ignore) 1434143197Ssobomax{ 1435143197Ssobomax 1436143197Ssobomax return (ENOSYS); 1437143197Ssobomax} 1438147141Ssobomax 1439147141Ssobomaxint 1440147141Ssobomaxlinux_getpriority(struct thread *td, struct linux_getpriority_args *args) 1441147141Ssobomax{ 1442147141Ssobomax struct getpriority_args bsd_args; 1443147141Ssobomax int error; 1444147141Ssobomax 1445147141Ssobomax bsd_args.which = args->which; 1446147141Ssobomax bsd_args.who = args->who; 1447147141Ssobomax error = getpriority(td, &bsd_args); 1448147141Ssobomax td->td_retval[0] = 20 - td->td_retval[0]; 1449147141Ssobomax return error; 1450147141Ssobomax} 1451156842Snetchild 1452156842Snetchildint 1453156842Snetchildlinux_sethostname(struct thread *td, struct linux_sethostname_args *args) 1454156842Snetchild{ 1455156842Snetchild int name[2]; 1456156842Snetchild int error; 1457156842Snetchild 1458156842Snetchild name[0] = CTL_KERN; 1459156842Snetchild name[1] = KERN_HOSTNAME; 1460157183Sjhb if ((error = suser_cred(td->td_ucred, SUSER_ALLOWJAIL))) 1461156842Snetchild return (error); 1462156842Snetchild return (userland_sysctl(td, name, 2, 0, 0, 0, args->hostname, 1463156842Snetchild args->len, 0, 0)); 1464156842Snetchild} 1465156842Snetchild 1466161310Snetchildint 1467161310Snetchildlinux_exit_group(struct thread *td, struct linux_exit_group_args *args) 1468161310Snetchild{ 1469161310Snetchild struct linux_emuldata *em, *td_em, *tmp_em; 1470161310Snetchild struct proc *sp; 1471161310Snetchild 1472161310Snetchild#ifdef DEBUG 1473161310Snetchild if (ldebug(exit_group)) 1474161310Snetchild printf(ARGS(exit_group, "%i"), args->error_code); 1475161310Snetchild#endif 1476161310Snetchild 1477161310Snetchild td_em = em_find(td->td_proc, EMUL_UNLOCKED); 1478161310Snetchild 1479161310Snetchild KASSERT(td_em != NULL, ("exit_group: emuldata not found.\n")); 1480161310Snetchild 1481161310Snetchild EMUL_SHARED_RLOCK(&emul_shared_lock); 1482161310Snetchild LIST_FOREACH_SAFE(em, &td_em->shared->threads, threads, tmp_em) { 1483161310Snetchild if (em->pid == td_em->pid) 1484161310Snetchild continue; 1485161310Snetchild 1486161310Snetchild sp = pfind(em->pid); 1487161310Snetchild psignal(sp, SIGKILL); 1488161310Snetchild PROC_UNLOCK(sp); 1489161310Snetchild#ifdef DEBUG 1490161310Snetchild printf(LMSG("linux_sys_exit_group: kill PID %d\n"), em->pid); 1491161310Snetchild#endif 1492161310Snetchild } 1493161310Snetchild 1494161310Snetchild EMUL_SHARED_RUNLOCK(&emul_shared_lock); 1495161310Snetchild EMUL_UNLOCK(&emul_lock); 1496161310Snetchild 1497161310Snetchild exit1(td, W_EXITCODE(args->error_code,0)); 1498161310Snetchild 1499161310Snetchild return (0); 1500161310Snetchild} 1501161310Snetchild 1502