Index: head/sys/amd64/linux32/linux.h =================================================================== --- head/sys/amd64/linux32/linux.h (revision 283390) +++ head/sys/amd64/linux32/linux.h (revision 283391) @@ -1,795 +1,796 @@ /*- * Copyright (c) 2004 Tim J. Robbins * Copyright (c) 2001 Doug Rabson * Copyright (c) 1994-1996 Søren Schmidt * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer * in this position and unchanged. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. The name of the author may not be used to endorse or promote products * derived from this software without specific prior written permission * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. * * $FreeBSD$ */ #ifndef _AMD64_LINUX_H_ #define _AMD64_LINUX_H_ #include /* * debugging support */ extern u_char linux_debug_map[]; #define ldebug(name) isclr(linux_debug_map, LINUX_SYS_linux_ ## name) #define ARGS(nm, fmt) "linux(%ld/%ld): "#nm"("fmt")\n", \ (long)td->td_proc->p_pid, (long)td->td_tid #define LMSG(fmt) "linux(%ld/%ld): "fmt"\n", \ (long)td->td_proc->p_pid, (long)td->td_tid #define LINUX_DTRACE linuxulator32 #ifdef MALLOC_DECLARE MALLOC_DECLARE(M_LINUX); #endif #define LINUX32_MAXUSER ((1ul << 32) - PAGE_SIZE) #define LINUX32_SHAREDPAGE (LINUX32_MAXUSER - PAGE_SIZE) #define LINUX32_USRSTACK LINUX32_SHAREDPAGE /* XXX 16 = sizeof(linux32_ps_strings) */ #define LINUX32_PS_STRINGS (LINUX32_USRSTACK - 16) #define LINUX32_MAXDSIZ (512 * 1024 * 1024) /* 512MB */ #define LINUX32_MAXSSIZ (64 * 1024 * 1024) /* 64MB */ #define LINUX32_MAXVMEM 0 /* Unlimited */ #define PTRIN(v) (void *)(uintptr_t)(v) #define PTROUT(v) (l_uintptr_t)(uintptr_t)(v) #define CP(src,dst,fld) do { (dst).fld = (src).fld; } while (0) #define CP2(src,dst,sfld,dfld) do { (dst).dfld = (src).sfld; } while (0) #define PTRIN_CP(src,dst,fld) \ do { (dst).fld = PTRIN((src).fld); } while (0) /* * Provide a separate set of types for the Linux types. */ typedef int l_int; typedef int32_t l_long; typedef int64_t l_longlong; typedef short l_short; typedef unsigned int l_uint; typedef uint32_t l_ulong; typedef uint64_t l_ulonglong; typedef unsigned short l_ushort; typedef l_ulong l_uintptr_t; typedef l_long l_clock_t; typedef l_int l_daddr_t; typedef l_ushort l_dev_t; typedef l_uint l_gid_t; typedef l_ushort l_gid16_t; typedef l_ulong l_ino_t; typedef l_int l_key_t; typedef l_longlong l_loff_t; typedef l_ushort l_mode_t; typedef l_long l_off_t; typedef l_int l_pid_t; typedef l_uint l_size_t; typedef l_long l_suseconds_t; typedef l_long l_time_t; typedef l_uint l_uid_t; typedef l_ushort l_uid16_t; typedef l_int l_timer_t; typedef l_int l_mqd_t; typedef struct { l_int val[2]; } __packed l_fsid_t; typedef struct { l_time_t tv_sec; l_suseconds_t tv_usec; } l_timeval; #define l_fd_set fd_set /* * Miscellaneous */ #define LINUX_AT_COUNT 16 /* Count of used aux entry types. * Keep this synchronized with * elf_linux_fixup() code. */ struct l___sysctl_args { l_uintptr_t name; l_int nlen; l_uintptr_t oldval; l_uintptr_t oldlenp; l_uintptr_t newval; l_size_t newlen; l_ulong __spare[4]; } __packed; /* Resource limits */ #define LINUX_RLIMIT_CPU 0 #define LINUX_RLIMIT_FSIZE 1 #define LINUX_RLIMIT_DATA 2 #define LINUX_RLIMIT_STACK 3 #define LINUX_RLIMIT_CORE 4 #define LINUX_RLIMIT_RSS 5 #define LINUX_RLIMIT_NPROC 6 #define LINUX_RLIMIT_NOFILE 7 #define LINUX_RLIMIT_MEMLOCK 8 #define LINUX_RLIMIT_AS 9 /* Address space limit */ #define LINUX_RLIM_NLIMITS 10 struct l_rlimit { l_ulong rlim_cur; l_ulong rlim_max; } __packed; struct l_rusage { l_timeval ru_utime; l_timeval ru_stime; l_long ru_maxrss; l_long ru_ixrss; l_long ru_idrss; l_long ru_isrss; l_long ru_minflt; l_long ru_majflt; l_long ru_nswap; l_long ru_inblock; l_long ru_oublock; l_long ru_msgsnd; l_long ru_msgrcv; l_long ru_nsignals; l_long ru_nvcsw; l_long ru_nivcsw; } __packed; /* mmap options */ #define LINUX_MAP_SHARED 0x0001 #define LINUX_MAP_PRIVATE 0x0002 #define LINUX_MAP_FIXED 0x0010 #define LINUX_MAP_ANON 0x0020 #define LINUX_MAP_GROWSDOWN 0x0100 struct l_mmap_argv { l_uintptr_t addr; l_size_t len; l_int prot; l_int flags; l_int fd; l_ulong pgoff; }; /* * stat family of syscalls */ struct l_timespec { l_time_t tv_sec; l_long tv_nsec; } __packed; struct l_newstat { l_ushort st_dev; l_ushort __pad1; l_ulong st_ino; l_ushort st_mode; l_ushort st_nlink; l_ushort st_uid; l_ushort st_gid; l_ushort st_rdev; l_ushort __pad2; l_ulong st_size; l_ulong st_blksize; l_ulong st_blocks; struct l_timespec st_atim; struct l_timespec st_mtim; struct l_timespec st_ctim; l_ulong __unused4; l_ulong __unused5; } __packed; struct l_stat { l_ushort st_dev; l_ulong st_ino; l_ushort st_mode; l_ushort st_nlink; l_ushort st_uid; l_ushort st_gid; l_ushort st_rdev; l_long st_size; struct l_timespec st_atim; struct l_timespec st_mtim; struct l_timespec st_ctim; l_long st_blksize; l_long st_blocks; l_ulong st_flags; l_ulong st_gen; }; struct l_stat64 { l_ushort st_dev; u_char __pad0[10]; l_ulong __st_ino; l_uint st_mode; l_uint st_nlink; l_ulong st_uid; l_ulong st_gid; l_ushort st_rdev; u_char __pad3[10]; l_longlong st_size; l_ulong st_blksize; l_ulong st_blocks; l_ulong __pad4; struct l_timespec st_atim; struct l_timespec st_mtim; struct l_timespec st_ctim; l_ulonglong st_ino; } __packed; struct l_statfs64 { l_int f_type; l_int f_bsize; uint64_t f_blocks; uint64_t f_bfree; uint64_t f_bavail; uint64_t f_files; uint64_t f_ffree; l_fsid_t f_fsid; l_int f_namelen; l_int f_spare[6]; } __packed; /* * Signalling */ #define LINUX_SIGHUP 1 #define LINUX_SIGINT 2 #define LINUX_SIGQUIT 3 #define LINUX_SIGILL 4 #define LINUX_SIGTRAP 5 #define LINUX_SIGABRT 6 #define LINUX_SIGIOT LINUX_SIGABRT #define LINUX_SIGBUS 7 #define LINUX_SIGFPE 8 #define LINUX_SIGKILL 9 #define LINUX_SIGUSR1 10 #define LINUX_SIGSEGV 11 #define LINUX_SIGUSR2 12 #define LINUX_SIGPIPE 13 #define LINUX_SIGALRM 14 #define LINUX_SIGTERM 15 #define LINUX_SIGSTKFLT 16 #define LINUX_SIGCHLD 17 #define LINUX_SIGCONT 18 #define LINUX_SIGSTOP 19 #define LINUX_SIGTSTP 20 #define LINUX_SIGTTIN 21 #define LINUX_SIGTTOU 22 #define LINUX_SIGURG 23 #define LINUX_SIGXCPU 24 #define LINUX_SIGXFSZ 25 #define LINUX_SIGVTALRM 26 #define LINUX_SIGPROF 27 #define LINUX_SIGWINCH 28 #define LINUX_SIGIO 29 #define LINUX_SIGPOLL LINUX_SIGIO #define LINUX_SIGPWR 30 #define LINUX_SIGSYS 31 #define LINUX_SIGRTMIN 32 #define LINUX_SIGTBLSZ 31 #define LINUX_NSIG_WORDS 2 #define LINUX_NBPW 32 #define LINUX_NSIG (LINUX_NBPW * LINUX_NSIG_WORDS) /* sigaction flags */ #define LINUX_SA_NOCLDSTOP 0x00000001 #define LINUX_SA_NOCLDWAIT 0x00000002 #define LINUX_SA_SIGINFO 0x00000004 #define LINUX_SA_RESTORER 0x04000000 #define LINUX_SA_ONSTACK 0x08000000 #define LINUX_SA_RESTART 0x10000000 #define LINUX_SA_INTERRUPT 0x20000000 #define LINUX_SA_NOMASK 0x40000000 #define LINUX_SA_ONESHOT 0x80000000 /* sigprocmask actions */ #define LINUX_SIG_BLOCK 0 #define LINUX_SIG_UNBLOCK 1 #define LINUX_SIG_SETMASK 2 /* sigset_t macros */ #define LINUX_SIGEMPTYSET(set) (set).__bits[0] = (set).__bits[1] = 0 #define LINUX_SIGISMEMBER(set, sig) SIGISMEMBER(set, sig) #define LINUX_SIGADDSET(set, sig) SIGADDSET(set, sig) /* sigaltstack */ #define LINUX_MINSIGSTKSZ 2048 #define LINUX_SS_ONSTACK 1 #define LINUX_SS_DISABLE 2 int linux_to_bsd_sigaltstack(int lsa); int bsd_to_linux_sigaltstack(int bsa); typedef l_uintptr_t l_handler_t; typedef l_ulong l_osigset_t; typedef struct { l_uint __bits[LINUX_NSIG_WORDS]; } __packed l_sigset_t; typedef struct { l_handler_t lsa_handler; l_osigset_t lsa_mask; l_ulong lsa_flags; l_uintptr_t lsa_restorer; } __packed l_osigaction_t; typedef struct { l_handler_t lsa_handler; l_ulong lsa_flags; l_uintptr_t lsa_restorer; l_sigset_t lsa_mask; } __packed l_sigaction_t; typedef struct { l_uintptr_t ss_sp; l_int ss_flags; l_size_t ss_size; } __packed l_stack_t; /* The Linux sigcontext, pretty much a standard 386 trapframe. */ struct l_sigcontext { l_uint sc_gs; l_uint sc_fs; l_uint sc_es; l_uint sc_ds; l_uint sc_edi; l_uint sc_esi; l_uint sc_ebp; l_uint sc_esp; l_uint sc_ebx; l_uint sc_edx; l_uint sc_ecx; l_uint sc_eax; l_uint sc_trapno; l_uint sc_err; l_uint sc_eip; l_uint sc_cs; l_uint sc_eflags; l_uint sc_esp_at_signal; l_uint sc_ss; l_uint sc_387; l_uint sc_mask; l_uint sc_cr2; } __packed; struct l_ucontext { l_ulong uc_flags; l_uintptr_t uc_link; l_stack_t uc_stack; struct l_sigcontext uc_mcontext; l_sigset_t uc_sigmask; } __packed; #define LINUX_SI_MAX_SIZE 128 #define LINUX_SI_PAD_SIZE ((LINUX_SI_MAX_SIZE/sizeof(l_int)) - 3) typedef union l_sigval { l_int sival_int; l_uintptr_t sival_ptr; } l_sigval_t; typedef struct l_siginfo { l_int lsi_signo; l_int lsi_errno; l_int lsi_code; union { l_int _pad[LINUX_SI_PAD_SIZE]; struct { l_pid_t _pid; l_uid_t _uid; } __packed _kill; struct { l_timer_t _tid; l_int _overrun; char _pad[sizeof(l_uid_t) - sizeof(l_int)]; l_sigval_t _sigval; l_int _sys_private; } __packed _timer; struct { l_pid_t _pid; /* sender's pid */ l_uid_t _uid; /* sender's uid */ l_sigval_t _sigval; } __packed _rt; struct { l_pid_t _pid; /* which child */ l_uid_t _uid; /* sender's uid */ l_int _status; /* exit code */ l_clock_t _utime; l_clock_t _stime; } __packed _sigchld; struct { l_uintptr_t _addr; /* Faulting insn/memory ref. */ } __packed _sigfault; struct { l_long _band; /* POLL_IN,POLL_OUT,POLL_MSG */ l_int _fd; } __packed _sigpoll; } _sifields; } __packed l_siginfo_t; #define lsi_pid _sifields._kill._pid #define lsi_uid _sifields._kill._uid #define lsi_tid _sifields._timer._tid #define lsi_overrun _sifields._timer._overrun #define lsi_sys_private _sifields._timer._sys_private #define lsi_status _sifields._sigchld._status #define lsi_utime _sifields._sigchld._utime #define lsi_stime _sifields._sigchld._stime #define lsi_value _sifields._rt._sigval #define lsi_int _sifields._rt._sigval.sival_int #define lsi_ptr _sifields._rt._sigval.sival_ptr #define lsi_addr _sifields._sigfault._addr #define lsi_band _sifields._sigpoll._band #define lsi_fd _sifields._sigpoll._fd struct l_fpreg { u_int16_t significand[4]; u_int16_t exponent; } __packed; struct l_fpxreg { u_int16_t significand[4]; u_int16_t exponent; u_int16_t padding[3]; } __packed; struct l_xmmreg { u_int32_t element[4]; } __packed; struct l_fpstate { /* Regular FPU environment */ u_int32_t cw; u_int32_t sw; u_int32_t tag; u_int32_t ipoff; u_int32_t cssel; u_int32_t dataoff; u_int32_t datasel; struct l_fpreg _st[8]; u_int16_t status; u_int16_t magic; /* 0xffff = regular FPU data */ /* FXSR FPU environment */ u_int32_t _fxsr_env[6]; /* env is ignored. */ u_int32_t mxcsr; u_int32_t reserved; struct l_fpxreg _fxsr_st[8]; /* reg data is ignored. */ struct l_xmmreg _xmm[8]; u_int32_t padding[56]; } __packed; /* * We make the stack look like Linux expects it when calling a signal * handler, but use the BSD way of calling the handler and sigreturn(). * This means that we need to pass the pointer to the handler too. * It is appended to the frame to not interfere with the rest of it. */ struct l_sigframe { l_int sf_sig; struct l_sigcontext sf_sc; struct l_fpstate sf_fpstate; l_uint sf_extramask[LINUX_NSIG_WORDS-1]; l_handler_t sf_handler; } __packed; struct l_rt_sigframe { l_int sf_sig; l_uintptr_t sf_siginfo; l_uintptr_t sf_ucontext; l_siginfo_t sf_si; struct l_ucontext sf_sc; l_handler_t sf_handler; } __packed; extern struct sysentvec elf_linux_sysvec; /* * open/fcntl flags */ #define LINUX_O_RDONLY 00000000 #define LINUX_O_WRONLY 00000001 #define LINUX_O_RDWR 00000002 #define LINUX_O_ACCMODE 00000003 #define LINUX_O_CREAT 00000100 #define LINUX_O_EXCL 00000200 #define LINUX_O_NOCTTY 00000400 #define LINUX_O_TRUNC 00001000 #define LINUX_O_APPEND 00002000 #define LINUX_O_NONBLOCK 00004000 #define LINUX_O_NDELAY LINUX_O_NONBLOCK #define LINUX_O_SYNC 00010000 #define LINUX_FASYNC 00020000 #define LINUX_O_DIRECT 00040000 /* Direct disk access hint */ #define LINUX_O_LARGEFILE 00100000 #define LINUX_O_DIRECTORY 00200000 /* Must be a directory */ #define LINUX_O_NOFOLLOW 00400000 /* Do not follow links */ #define LINUX_O_NOATIME 01000000 #define LINUX_O_CLOEXEC 02000000 #define LINUX_F_DUPFD 0 #define LINUX_F_GETFD 1 #define LINUX_F_SETFD 2 #define LINUX_F_GETFL 3 #define LINUX_F_SETFL 4 #define LINUX_F_GETLK 5 #define LINUX_F_SETLK 6 #define LINUX_F_SETLKW 7 #define LINUX_F_SETOWN 8 #define LINUX_F_GETOWN 9 #define LINUX_F_GETLK64 12 #define LINUX_F_SETLK64 13 #define LINUX_F_SETLKW64 14 #define LINUX_F_RDLCK 0 #define LINUX_F_WRLCK 1 #define LINUX_F_UNLCK 2 union l_semun { l_int val; l_uintptr_t buf; l_uintptr_t array; l_uintptr_t __buf; l_uintptr_t __pad; } __packed; /* * Socket defines */ #define LINUX_SOL_SOCKET 1 #define LINUX_SOL_IP 0 #define LINUX_SOL_IPX 256 #define LINUX_SOL_AX25 257 #define LINUX_SOL_TCP 6 #define LINUX_SOL_UDP 17 #define LINUX_SO_DEBUG 1 #define LINUX_SO_REUSEADDR 2 #define LINUX_SO_TYPE 3 #define LINUX_SO_ERROR 4 #define LINUX_SO_DONTROUTE 5 #define LINUX_SO_BROADCAST 6 #define LINUX_SO_SNDBUF 7 #define LINUX_SO_RCVBUF 8 #define LINUX_SO_KEEPALIVE 9 #define LINUX_SO_OOBINLINE 10 #define LINUX_SO_NO_CHECK 11 #define LINUX_SO_PRIORITY 12 #define LINUX_SO_LINGER 13 #define LINUX_SO_PEERCRED 17 #define LINUX_SO_RCVLOWAT 18 #define LINUX_SO_SNDLOWAT 19 #define LINUX_SO_RCVTIMEO 20 #define LINUX_SO_SNDTIMEO 21 #define LINUX_SO_TIMESTAMP 29 #define LINUX_SO_ACCEPTCONN 30 struct l_sockaddr { l_ushort sa_family; char sa_data[14]; } __packed; struct l_msghdr { l_uintptr_t msg_name; l_int msg_namelen; l_uintptr_t msg_iov; l_size_t msg_iovlen; l_uintptr_t msg_control; l_size_t msg_controllen; l_uint msg_flags; }; struct l_cmsghdr { l_size_t cmsg_len; l_int cmsg_level; l_int cmsg_type; }; struct l_ifmap { l_ulong mem_start; l_ulong mem_end; l_ushort base_addr; u_char irq; u_char dma; u_char port; } __packed; #define LINUX_IFHWADDRLEN 6 #define LINUX_IFNAMSIZ 16 struct l_ifreq { union { char ifrn_name[LINUX_IFNAMSIZ]; } ifr_ifrn; union { struct l_sockaddr ifru_addr; struct l_sockaddr ifru_dstaddr; struct l_sockaddr ifru_broadaddr; struct l_sockaddr ifru_netmask; struct l_sockaddr ifru_hwaddr; l_short ifru_flags[1]; l_int ifru_metric; l_int ifru_mtu; struct l_ifmap ifru_map; char ifru_slave[LINUX_IFNAMSIZ]; l_uintptr_t ifru_data; } ifr_ifru; } __packed; #define ifr_name ifr_ifrn.ifrn_name /* Interface name */ #define ifr_hwaddr ifr_ifru.ifru_hwaddr /* MAC address */ struct l_ifconf { int ifc_len; union { l_uintptr_t ifcu_buf; l_uintptr_t ifcu_req; } ifc_ifcu; } __packed; #define ifc_buf ifc_ifcu.ifcu_buf #define ifc_req ifc_ifcu.ifcu_req /* * poll() */ #define LINUX_POLLIN 0x0001 #define LINUX_POLLPRI 0x0002 #define LINUX_POLLOUT 0x0004 #define LINUX_POLLERR 0x0008 #define LINUX_POLLHUP 0x0010 #define LINUX_POLLNVAL 0x0020 #define LINUX_POLLRDNORM 0x0040 #define LINUX_POLLRDBAND 0x0080 #define LINUX_POLLWRNORM 0x0100 #define LINUX_POLLWRBAND 0x0200 #define LINUX_POLLMSG 0x0400 struct l_pollfd { l_int fd; l_short events; l_short revents; } __packed; struct l_user_desc { l_uint entry_number; l_uint base_addr; l_uint limit; l_uint seg_32bit:1; l_uint contents:2; l_uint read_exec_only:1; l_uint limit_in_pages:1; l_uint seg_not_present:1; l_uint useable:1; }; #define LINUX_LOWERWORD 0x0000ffff /* * Macros which does the same thing as those in Linux include/asm-um/ldt-i386.h. * These convert Linux user space descriptor to machine one. */ #define LINUX_LDT_entry_a(info) \ ((((info)->base_addr & LINUX_LOWERWORD) << 16) | \ ((info)->limit & LINUX_LOWERWORD)) #define LINUX_ENTRY_B_READ_EXEC_ONLY 9 #define LINUX_ENTRY_B_CONTENTS 10 #define LINUX_ENTRY_B_SEG_NOT_PRESENT 15 #define LINUX_ENTRY_B_BASE_ADDR 16 #define LINUX_ENTRY_B_USEABLE 20 #define LINUX_ENTRY_B_SEG32BIT 22 #define LINUX_ENTRY_B_LIMIT 23 #define LINUX_LDT_entry_b(info) \ (((info)->base_addr & 0xff000000) | \ ((info)->limit & 0xf0000) | \ ((info)->contents << LINUX_ENTRY_B_CONTENTS) | \ (((info)->seg_not_present == 0) << LINUX_ENTRY_B_SEG_NOT_PRESENT) | \ (((info)->base_addr & 0x00ff0000) >> LINUX_ENTRY_B_BASE_ADDR) | \ (((info)->read_exec_only == 0) << LINUX_ENTRY_B_READ_EXEC_ONLY) | \ ((info)->seg_32bit << LINUX_ENTRY_B_SEG32BIT) | \ ((info)->useable << LINUX_ENTRY_B_USEABLE) | \ ((info)->limit_in_pages << LINUX_ENTRY_B_LIMIT) | 0x7000) #define LINUX_LDT_empty(info) \ ((info)->base_addr == 0 && \ (info)->limit == 0 && \ (info)->contents == 0 && \ (info)->seg_not_present == 1 && \ (info)->read_exec_only == 1 && \ (info)->seg_32bit == 0 && \ (info)->limit_in_pages == 0 && \ (info)->useable == 0) /* * Macros for converting segments. * They do the same as those in arch/i386/kernel/process.c in Linux. */ #define LINUX_GET_BASE(desc) \ ((((desc)->a >> 16) & LINUX_LOWERWORD) | \ (((desc)->b << 16) & 0x00ff0000) | \ ((desc)->b & 0xff000000)) #define LINUX_GET_LIMIT(desc) \ (((desc)->a & LINUX_LOWERWORD) | \ ((desc)->b & 0xf0000)) #define LINUX_GET_32BIT(desc) \ (((desc)->b >> LINUX_ENTRY_B_SEG32BIT) & 1) #define LINUX_GET_CONTENTS(desc) \ (((desc)->b >> LINUX_ENTRY_B_CONTENTS) & 3) #define LINUX_GET_WRITABLE(desc) \ (((desc)->b >> LINUX_ENTRY_B_READ_EXEC_ONLY) & 1) #define LINUX_GET_LIMIT_PAGES(desc) \ (((desc)->b >> LINUX_ENTRY_B_LIMIT) & 1) #define LINUX_GET_PRESENT(desc) \ (((desc)->b >> LINUX_ENTRY_B_SEG_NOT_PRESENT) & 1) #define LINUX_GET_USEABLE(desc) \ (((desc)->b >> LINUX_ENTRY_B_USEABLE) & 1) struct iovec; struct l_iovec32 { uint32_t iov_base; l_size_t iov_len; }; int linux32_copyiniov(struct l_iovec32 *iovp32, l_ulong iovcnt, struct iovec **iovp, int error); +int linux_copyout_rusage(struct rusage *ru, void *uaddr); /* robust futexes */ struct linux_robust_list { l_uintptr_t next; }; struct linux_robust_list_head { struct linux_robust_list list; l_long futex_offset; l_uintptr_t pending_list; }; #endif /* !_AMD64_LINUX_H_ */ Index: head/sys/amd64/linux32/linux32_genassym.c =================================================================== --- head/sys/amd64/linux32/linux32_genassym.c (revision 283390) +++ head/sys/amd64/linux32/linux32_genassym.c (revision 283391) @@ -1,14 +1,15 @@ #include __FBSDID("$FreeBSD$"); #include #include +#include #include #include ASSYM(LINUX_SIGF_HANDLER, offsetof(struct l_sigframe, sf_handler)); ASSYM(LINUX_SIGF_SC, offsetof(struct l_sigframe, sf_sc)); ASSYM(LINUX_RT_SIGF_HANDLER, offsetof(struct l_rt_sigframe, sf_handler)); ASSYM(LINUX_RT_SIGF_UC, offsetof(struct l_rt_sigframe, sf_sc)); ASSYM(LINUX_RT_SIGF_SC, offsetof(struct l_ucontext, uc_mcontext)); Index: head/sys/amd64/linux32/linux32_machdep.c =================================================================== --- head/sys/amd64/linux32/linux32_machdep.c (revision 283390) +++ head/sys/amd64/linux32/linux32_machdep.c (revision 283391) @@ -1,1060 +1,1035 @@ /*- * Copyright (c) 2004 Tim J. Robbins * Copyright (c) 2002 Doug Rabson * Copyright (c) 2000 Marcel Moolenaar * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer * in this position and unchanged. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. The name of the author may not be used to endorse or promote products * derived from this software without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include +static void bsd_to_linux_rusage(struct rusage *ru, struct l_rusage *lru); + struct l_old_select_argv { l_int nfds; l_uintptr_t readfds; l_uintptr_t writefds; l_uintptr_t exceptfds; l_uintptr_t timeout; } __packed; int linux_to_bsd_sigaltstack(int lsa) { int bsa = 0; if (lsa & LINUX_SS_DISABLE) bsa |= SS_DISABLE; if (lsa & LINUX_SS_ONSTACK) bsa |= SS_ONSTACK; return (bsa); } static int linux_mmap_common(struct thread *td, l_uintptr_t addr, l_size_t len, l_int prot, l_int flags, l_int fd, l_loff_t pos); int bsd_to_linux_sigaltstack(int bsa) { int lsa = 0; if (bsa & SS_DISABLE) lsa |= LINUX_SS_DISABLE; if (bsa & SS_ONSTACK) lsa |= LINUX_SS_ONSTACK; return (lsa); } static void bsd_to_linux_rusage(struct rusage *ru, struct l_rusage *lru) { lru->ru_utime.tv_sec = ru->ru_utime.tv_sec; lru->ru_utime.tv_usec = ru->ru_utime.tv_usec; lru->ru_stime.tv_sec = ru->ru_stime.tv_sec; lru->ru_stime.tv_usec = ru->ru_stime.tv_usec; lru->ru_maxrss = ru->ru_maxrss; lru->ru_ixrss = ru->ru_ixrss; lru->ru_idrss = ru->ru_idrss; lru->ru_isrss = ru->ru_isrss; lru->ru_minflt = ru->ru_minflt; lru->ru_majflt = ru->ru_majflt; lru->ru_nswap = ru->ru_nswap; lru->ru_inblock = ru->ru_inblock; lru->ru_oublock = ru->ru_oublock; lru->ru_msgsnd = ru->ru_msgsnd; lru->ru_msgrcv = ru->ru_msgrcv; lru->ru_nsignals = ru->ru_nsignals; lru->ru_nvcsw = ru->ru_nvcsw; lru->ru_nivcsw = ru->ru_nivcsw; } int +linux_copyout_rusage(struct rusage *ru, void *uaddr) +{ + struct l_rusage lru; + + bsd_to_linux_rusage(ru, &lru); + + return (copyout(&lru, uaddr, sizeof(struct l_rusage))); +} + +int linux_execve(struct thread *td, struct linux_execve_args *args) { struct image_args eargs; struct vmspace *oldvmspace; char *path; int error; LCONVPATHEXIST(td, args->path, &path); #ifdef DEBUG if (ldebug(execve)) printf(ARGS(execve, "%s"), path); #endif error = pre_execve(td, &oldvmspace); if (error != 0) { free(path, M_TEMP); return (error); } error = freebsd32_exec_copyin_args(&eargs, path, UIO_SYSSPACE, args->argp, args->envp); free(path, M_TEMP); if (error == 0) error = kern_execve(td, &eargs, NULL); if (error == 0) error = linux_common_execve(td, &eargs); post_execve(td, error, oldvmspace); return (error); } CTASSERT(sizeof(struct l_iovec32) == 8); static int linux32_copyinuio(struct l_iovec32 *iovp, l_ulong iovcnt, struct uio **uiop) { struct l_iovec32 iov32; struct iovec *iov; struct uio *uio; uint32_t iovlen; int error, i; *uiop = NULL; if (iovcnt > UIO_MAXIOV) return (EINVAL); iovlen = iovcnt * sizeof(struct iovec); uio = malloc(iovlen + sizeof(*uio), M_IOV, M_WAITOK); iov = (struct iovec *)(uio + 1); for (i = 0; i < iovcnt; i++) { error = copyin(&iovp[i], &iov32, sizeof(struct l_iovec32)); if (error) { free(uio, M_IOV); return (error); } iov[i].iov_base = PTRIN(iov32.iov_base); iov[i].iov_len = iov32.iov_len; } uio->uio_iov = iov; uio->uio_iovcnt = iovcnt; uio->uio_segflg = UIO_USERSPACE; uio->uio_offset = -1; uio->uio_resid = 0; for (i = 0; i < iovcnt; i++) { if (iov->iov_len > INT_MAX - uio->uio_resid) { free(uio, M_IOV); return (EINVAL); } uio->uio_resid += iov->iov_len; iov++; } *uiop = uio; return (0); } int linux32_copyiniov(struct l_iovec32 *iovp32, l_ulong iovcnt, struct iovec **iovp, int error) { struct l_iovec32 iov32; struct iovec *iov; uint32_t iovlen; int i; *iovp = NULL; if (iovcnt > UIO_MAXIOV) return (error); iovlen = iovcnt * sizeof(struct iovec); iov = malloc(iovlen, M_IOV, M_WAITOK); for (i = 0; i < iovcnt; i++) { error = copyin(&iovp32[i], &iov32, sizeof(struct l_iovec32)); if (error) { free(iov, M_IOV); return (error); } iov[i].iov_base = PTRIN(iov32.iov_base); iov[i].iov_len = iov32.iov_len; } *iovp = iov; return(0); } int linux_readv(struct thread *td, struct linux_readv_args *uap) { struct uio *auio; int error; error = linux32_copyinuio(uap->iovp, uap->iovcnt, &auio); if (error) return (error); error = kern_readv(td, uap->fd, auio); free(auio, M_IOV); return (error); } int linux_writev(struct thread *td, struct linux_writev_args *uap) { struct uio *auio; int error; error = linux32_copyinuio(uap->iovp, uap->iovcnt, &auio); if (error) return (error); error = kern_writev(td, uap->fd, auio); free(auio, M_IOV); return (error); } struct l_ipc_kludge { l_uintptr_t msgp; l_long msgtyp; } __packed; int linux_ipc(struct thread *td, struct linux_ipc_args *args) { switch (args->what & 0xFFFF) { case LINUX_SEMOP: { struct linux_semop_args a; a.semid = args->arg1; a.tsops = args->ptr; a.nsops = args->arg2; return (linux_semop(td, &a)); } case LINUX_SEMGET: { struct linux_semget_args a; a.key = args->arg1; a.nsems = args->arg2; a.semflg = args->arg3; return (linux_semget(td, &a)); } case LINUX_SEMCTL: { struct linux_semctl_args a; int error; a.semid = args->arg1; a.semnum = args->arg2; a.cmd = args->arg3; error = copyin(args->ptr, &a.arg, sizeof(a.arg)); if (error) return (error); return (linux_semctl(td, &a)); } case LINUX_MSGSND: { struct linux_msgsnd_args a; a.msqid = args->arg1; a.msgp = args->ptr; a.msgsz = args->arg2; a.msgflg = args->arg3; return (linux_msgsnd(td, &a)); } case LINUX_MSGRCV: { struct linux_msgrcv_args a; a.msqid = args->arg1; a.msgsz = args->arg2; a.msgflg = args->arg3; if ((args->what >> 16) == 0) { struct l_ipc_kludge tmp; int error; if (args->ptr == 0) return (EINVAL); error = copyin(args->ptr, &tmp, sizeof(tmp)); if (error) return (error); a.msgp = PTRIN(tmp.msgp); a.msgtyp = tmp.msgtyp; } else { a.msgp = args->ptr; a.msgtyp = args->arg5; } return (linux_msgrcv(td, &a)); } case LINUX_MSGGET: { struct linux_msgget_args a; a.key = args->arg1; a.msgflg = args->arg2; return (linux_msgget(td, &a)); } case LINUX_MSGCTL: { struct linux_msgctl_args a; a.msqid = args->arg1; a.cmd = args->arg2; a.buf = args->ptr; return (linux_msgctl(td, &a)); } case LINUX_SHMAT: { struct linux_shmat_args a; a.shmid = args->arg1; a.shmaddr = args->ptr; a.shmflg = args->arg2; a.raddr = PTRIN((l_uint)args->arg3); return (linux_shmat(td, &a)); } case LINUX_SHMDT: { struct linux_shmdt_args a; a.shmaddr = args->ptr; return (linux_shmdt(td, &a)); } case LINUX_SHMGET: { struct linux_shmget_args a; a.key = args->arg1; a.size = args->arg2; a.shmflg = args->arg3; return (linux_shmget(td, &a)); } case LINUX_SHMCTL: { struct linux_shmctl_args a; a.shmid = args->arg1; a.cmd = args->arg2; a.buf = args->ptr; return (linux_shmctl(td, &a)); } default: break; } return (EINVAL); } int linux_old_select(struct thread *td, struct linux_old_select_args *args) { struct l_old_select_argv linux_args; struct linux_select_args newsel; int error; #ifdef DEBUG if (ldebug(old_select)) printf(ARGS(old_select, "%p"), args->ptr); #endif error = copyin(args->ptr, &linux_args, sizeof(linux_args)); if (error) return (error); newsel.nfds = linux_args.nfds; newsel.readfds = PTRIN(linux_args.readfds); newsel.writefds = PTRIN(linux_args.writefds); newsel.exceptfds = PTRIN(linux_args.exceptfds); newsel.timeout = PTRIN(linux_args.timeout); return (linux_select(td, &newsel)); } int linux_set_cloned_tls(struct thread *td, void *desc) { struct user_segment_descriptor sd; struct l_user_desc info; struct pcb *pcb; int error; int a[2]; error = copyin(desc, &info, sizeof(struct l_user_desc)); if (error) { printf(LMSG("copyin failed!")); } else { /* We might copy out the entry_number as GUGS32_SEL. */ info.entry_number = GUGS32_SEL; error = copyout(&info, desc, sizeof(struct l_user_desc)); if (error) printf(LMSG("copyout failed!")); a[0] = LINUX_LDT_entry_a(&info); a[1] = LINUX_LDT_entry_b(&info); memcpy(&sd, &a, sizeof(a)); #ifdef DEBUG if (ldebug(clone)) printf("Segment created in clone with " "CLONE_SETTLS: lobase: %x, hibase: %x, " "lolimit: %x, hilimit: %x, type: %i, " "dpl: %i, p: %i, xx: %i, long: %i, " "def32: %i, gran: %i\n", sd.sd_lobase, sd.sd_hibase, sd.sd_lolimit, sd.sd_hilimit, sd.sd_type, sd.sd_dpl, sd.sd_p, sd.sd_xx, sd.sd_long, sd.sd_def32, sd.sd_gran); #endif pcb = td->td_pcb; pcb->pcb_gsbase = (register_t)info.base_addr; /* XXXKIB pcb->pcb_gs32sd = sd; */ td->td_frame->tf_gs = GSEL(GUGS32_SEL, SEL_UPL); set_pcb_flags(pcb, PCB_32BIT); } return (error); } int linux_set_upcall_kse(struct thread *td, register_t stack) { if (stack) td->td_frame->tf_rsp = stack; /* * The newly created Linux thread returns * to the user space by the same path that a parent do. */ td->td_frame->tf_rax = 0; return (0); } #define STACK_SIZE (2 * 1024 * 1024) #define GUARD_SIZE (4 * PAGE_SIZE) int linux_mmap2(struct thread *td, struct linux_mmap2_args *args) { #ifdef DEBUG if (ldebug(mmap2)) printf(ARGS(mmap2, "0x%08x, %d, %d, 0x%08x, %d, %d"), args->addr, args->len, args->prot, args->flags, args->fd, args->pgoff); #endif return (linux_mmap_common(td, PTROUT(args->addr), args->len, args->prot, args->flags, args->fd, (uint64_t)(uint32_t)args->pgoff * PAGE_SIZE)); } int linux_mmap(struct thread *td, struct linux_mmap_args *args) { int error; struct l_mmap_argv linux_args; error = copyin(args->ptr, &linux_args, sizeof(linux_args)); if (error) return (error); #ifdef DEBUG if (ldebug(mmap)) printf(ARGS(mmap, "0x%08x, %d, %d, 0x%08x, %d, %d"), linux_args.addr, linux_args.len, linux_args.prot, linux_args.flags, linux_args.fd, linux_args.pgoff); #endif return (linux_mmap_common(td, linux_args.addr, linux_args.len, linux_args.prot, linux_args.flags, linux_args.fd, (uint32_t)linux_args.pgoff)); } static int linux_mmap_common(struct thread *td, l_uintptr_t addr, l_size_t len, l_int prot, l_int flags, l_int fd, l_loff_t pos) { struct proc *p = td->td_proc; struct mmap_args /* { caddr_t addr; size_t len; int prot; int flags; int fd; long pad; off_t pos; } */ bsd_args; int error; struct file *fp; cap_rights_t rights; error = 0; bsd_args.flags = 0; fp = NULL; /* * Linux mmap(2): * You must specify exactly one of MAP_SHARED and MAP_PRIVATE */ if (!((flags & LINUX_MAP_SHARED) ^ (flags & LINUX_MAP_PRIVATE))) return (EINVAL); if (flags & LINUX_MAP_SHARED) bsd_args.flags |= MAP_SHARED; if (flags & LINUX_MAP_PRIVATE) bsd_args.flags |= MAP_PRIVATE; if (flags & LINUX_MAP_FIXED) bsd_args.flags |= MAP_FIXED; if (flags & LINUX_MAP_ANON) { /* Enforce pos to be on page boundary, then ignore. */ if ((pos & PAGE_MASK) != 0) return (EINVAL); pos = 0; bsd_args.flags |= MAP_ANON; } else bsd_args.flags |= MAP_NOSYNC; if (flags & LINUX_MAP_GROWSDOWN) bsd_args.flags |= MAP_STACK; /* * PROT_READ, PROT_WRITE, or PROT_EXEC implies PROT_READ and PROT_EXEC * on Linux/i386. We do this to ensure maximum compatibility. * Linux/ia64 does the same in i386 emulation mode. */ bsd_args.prot = prot; if (bsd_args.prot & (PROT_READ | PROT_WRITE | PROT_EXEC)) bsd_args.prot |= PROT_READ | PROT_EXEC; /* Linux does not check file descriptor when MAP_ANONYMOUS is set. */ bsd_args.fd = (bsd_args.flags & MAP_ANON) ? -1 : fd; if (bsd_args.fd != -1) { /* * Linux follows Solaris mmap(2) description: * The file descriptor fildes is opened with * read permission, regardless of the * protection options specified. */ error = fget(td, bsd_args.fd, cap_rights_init(&rights, CAP_MMAP), &fp); if (error != 0) return (error); if (fp->f_type != DTYPE_VNODE) { fdrop(fp, td); return (EINVAL); } /* Linux mmap() just fails for O_WRONLY files */ if (!(fp->f_flag & FREAD)) { fdrop(fp, td); return (EACCES); } fdrop(fp, td); } if (flags & LINUX_MAP_GROWSDOWN) { /* * The Linux MAP_GROWSDOWN option does not limit auto * growth of the region. Linux mmap with this option * takes as addr the inital BOS, and as len, the initial * region size. It can then grow down from addr without * limit. However, Linux threads has an implicit internal * limit to stack size of STACK_SIZE. Its just not * enforced explicitly in Linux. But, here we impose * a limit of (STACK_SIZE - GUARD_SIZE) on the stack * region, since we can do this with our mmap. * * Our mmap with MAP_STACK takes addr as the maximum * downsize limit on BOS, and as len the max size of * the region. It then maps the top SGROWSIZ bytes, * and auto grows the region down, up to the limit * in addr. * * If we don't use the MAP_STACK option, the effect * of this code is to allocate a stack region of a * fixed size of (STACK_SIZE - GUARD_SIZE). */ if ((caddr_t)PTRIN(addr) + len > p->p_vmspace->vm_maxsaddr) { /* * Some Linux apps will attempt to mmap * thread stacks near the top of their * address space. If their TOS is greater * than vm_maxsaddr, vm_map_growstack() * will confuse the thread stack with the * process stack and deliver a SEGV if they * attempt to grow the thread stack past their * current stacksize rlimit. To avoid this, * adjust vm_maxsaddr upwards to reflect * the current stacksize rlimit rather * than the maximum possible stacksize. * It would be better to adjust the * mmap'ed region, but some apps do not check * mmap's return value. */ PROC_LOCK(p); p->p_vmspace->vm_maxsaddr = (char *)LINUX32_USRSTACK - lim_cur(p, RLIMIT_STACK); PROC_UNLOCK(p); } /* * This gives us our maximum stack size and a new BOS. * If we're using VM_STACK, then mmap will just map * the top SGROWSIZ bytes, and let the stack grow down * to the limit at BOS. If we're not using VM_STACK * we map the full stack, since we don't have a way * to autogrow it. */ if (len > STACK_SIZE - GUARD_SIZE) { bsd_args.addr = (caddr_t)PTRIN(addr); bsd_args.len = len; } else { bsd_args.addr = (caddr_t)PTRIN(addr) - (STACK_SIZE - GUARD_SIZE - len); bsd_args.len = STACK_SIZE - GUARD_SIZE; } } else { bsd_args.addr = (caddr_t)PTRIN(addr); bsd_args.len = len; } bsd_args.pos = pos; #ifdef DEBUG if (ldebug(mmap)) printf("-> %s(%p, %d, %d, 0x%08x, %d, 0x%x)\n", __func__, (void *)bsd_args.addr, (int)bsd_args.len, bsd_args.prot, bsd_args.flags, bsd_args.fd, (int)bsd_args.pos); #endif error = sys_mmap(td, &bsd_args); #ifdef DEBUG if (ldebug(mmap)) printf("-> %s() return: 0x%x (0x%08x)\n", __func__, error, (u_int)td->td_retval[0]); #endif return (error); } int linux_mprotect(struct thread *td, struct linux_mprotect_args *uap) { struct mprotect_args bsd_args; bsd_args.addr = uap->addr; bsd_args.len = uap->len; bsd_args.prot = uap->prot; if (bsd_args.prot & (PROT_READ | PROT_WRITE | PROT_EXEC)) bsd_args.prot |= PROT_READ | PROT_EXEC; return (sys_mprotect(td, &bsd_args)); } int linux_iopl(struct thread *td, struct linux_iopl_args *args) { int error; if (args->level < 0 || args->level > 3) return (EINVAL); if ((error = priv_check(td, PRIV_IO)) != 0) return (error); if ((error = securelevel_gt(td->td_ucred, 0)) != 0) return (error); td->td_frame->tf_rflags = (td->td_frame->tf_rflags & ~PSL_IOPL) | (args->level * (PSL_IOPL / 3)); return (0); } int linux_sigaction(struct thread *td, struct linux_sigaction_args *args) { l_osigaction_t osa; l_sigaction_t act, oact; int error; #ifdef DEBUG if (ldebug(sigaction)) printf(ARGS(sigaction, "%d, %p, %p"), args->sig, (void *)args->nsa, (void *)args->osa); #endif if (args->nsa != NULL) { error = copyin(args->nsa, &osa, sizeof(l_osigaction_t)); if (error) return (error); act.lsa_handler = osa.lsa_handler; act.lsa_flags = osa.lsa_flags; act.lsa_restorer = osa.lsa_restorer; LINUX_SIGEMPTYSET(act.lsa_mask); act.lsa_mask.__bits[0] = osa.lsa_mask; } error = linux_do_sigaction(td, args->sig, args->nsa ? &act : NULL, args->osa ? &oact : NULL); if (args->osa != NULL && !error) { osa.lsa_handler = oact.lsa_handler; osa.lsa_flags = oact.lsa_flags; osa.lsa_restorer = oact.lsa_restorer; osa.lsa_mask = oact.lsa_mask.__bits[0]; error = copyout(&osa, args->osa, sizeof(l_osigaction_t)); } return (error); } /* * Linux has two extra args, restart and oldmask. We don't use these, * but it seems that "restart" is actually a context pointer that * enables the signal to happen with a different register set. */ int linux_sigsuspend(struct thread *td, struct linux_sigsuspend_args *args) { sigset_t sigmask; l_sigset_t mask; #ifdef DEBUG if (ldebug(sigsuspend)) printf(ARGS(sigsuspend, "%08lx"), (unsigned long)args->mask); #endif LINUX_SIGEMPTYSET(mask); mask.__bits[0] = args->mask; linux_to_bsd_sigset(&mask, &sigmask); return (kern_sigsuspend(td, sigmask)); } int linux_rt_sigsuspend(struct thread *td, struct linux_rt_sigsuspend_args *uap) { l_sigset_t lmask; sigset_t sigmask; int error; #ifdef DEBUG if (ldebug(rt_sigsuspend)) printf(ARGS(rt_sigsuspend, "%p, %d"), (void *)uap->newset, uap->sigsetsize); #endif if (uap->sigsetsize != sizeof(l_sigset_t)) return (EINVAL); error = copyin(uap->newset, &lmask, sizeof(l_sigset_t)); if (error) return (error); linux_to_bsd_sigset(&lmask, &sigmask); return (kern_sigsuspend(td, sigmask)); } int linux_pause(struct thread *td, struct linux_pause_args *args) { struct proc *p = td->td_proc; sigset_t sigmask; #ifdef DEBUG if (ldebug(pause)) printf(ARGS(pause, "")); #endif PROC_LOCK(p); sigmask = td->td_sigmask; PROC_UNLOCK(p); return (kern_sigsuspend(td, sigmask)); } int linux_sigaltstack(struct thread *td, struct linux_sigaltstack_args *uap) { stack_t ss, oss; l_stack_t lss; int error; #ifdef DEBUG if (ldebug(sigaltstack)) printf(ARGS(sigaltstack, "%p, %p"), uap->uss, uap->uoss); #endif if (uap->uss != NULL) { error = copyin(uap->uss, &lss, sizeof(l_stack_t)); if (error) return (error); ss.ss_sp = PTRIN(lss.ss_sp); ss.ss_size = lss.ss_size; ss.ss_flags = linux_to_bsd_sigaltstack(lss.ss_flags); } error = kern_sigaltstack(td, (uap->uss != NULL) ? &ss : NULL, (uap->uoss != NULL) ? &oss : NULL); if (!error && uap->uoss != NULL) { lss.ss_sp = PTROUT(oss.ss_sp); lss.ss_size = oss.ss_size; lss.ss_flags = bsd_to_linux_sigaltstack(oss.ss_flags); error = copyout(&lss, uap->uoss, sizeof(l_stack_t)); } return (error); } int linux_ftruncate64(struct thread *td, struct linux_ftruncate64_args *args) { struct ftruncate_args sa; #ifdef DEBUG if (ldebug(ftruncate64)) printf(ARGS(ftruncate64, "%u, %jd"), args->fd, (intmax_t)args->length); #endif sa.fd = args->fd; sa.length = args->length; return sys_ftruncate(td, &sa); } int linux_gettimeofday(struct thread *td, struct linux_gettimeofday_args *uap) { struct timeval atv; l_timeval atv32; struct timezone rtz; int error = 0; if (uap->tp) { microtime(&atv); atv32.tv_sec = atv.tv_sec; atv32.tv_usec = atv.tv_usec; error = copyout(&atv32, uap->tp, sizeof(atv32)); } if (error == 0 && uap->tzp != NULL) { rtz.tz_minuteswest = tz_minuteswest; rtz.tz_dsttime = tz_dsttime; error = copyout(&rtz, uap->tzp, sizeof(rtz)); } return (error); } int linux_settimeofday(struct thread *td, struct linux_settimeofday_args *uap) { l_timeval atv32; struct timeval atv, *tvp; struct timezone atz, *tzp; int error; if (uap->tp) { error = copyin(uap->tp, &atv32, sizeof(atv32)); if (error) return (error); atv.tv_sec = atv32.tv_sec; atv.tv_usec = atv32.tv_usec; tvp = &atv; } else tvp = NULL; if (uap->tzp) { error = copyin(uap->tzp, &atz, sizeof(atz)); if (error) return (error); tzp = &atz; } else tzp = NULL; return (kern_settimeofday(td, tvp, tzp)); } int linux_getrusage(struct thread *td, struct linux_getrusage_args *uap) { - struct l_rusage s32; struct rusage s; int error; error = kern_getrusage(td, uap->who, &s); if (error != 0) return (error); - if (uap->rusage != NULL) { - bsd_to_linux_rusage(&s, &s32); - error = copyout(&s32, uap->rusage, sizeof(s32)); - } + if (uap->rusage != NULL) + error = linux_copyout_rusage(&s, uap->rusage); return (error); } int linux_set_thread_area(struct thread *td, struct linux_set_thread_area_args *args) { struct l_user_desc info; struct user_segment_descriptor sd; struct pcb *pcb; int a[2]; int error; error = copyin(args->desc, &info, sizeof(struct l_user_desc)); if (error) return (error); #ifdef DEBUG if (ldebug(set_thread_area)) printf(ARGS(set_thread_area, "%i, %x, %x, %i, %i, %i, " "%i, %i, %i"), info.entry_number, info.base_addr, info.limit, info.seg_32bit, info.contents, info.read_exec_only, info.limit_in_pages, info.seg_not_present, info.useable); #endif /* * Semantics of Linux version: every thread in the system has array * of three TLS descriptors. 1st is GLIBC TLS, 2nd is WINE, 3rd unknown. * This syscall loads one of the selected TLS decriptors with a value * and also loads GDT descriptors 6, 7 and 8 with the content of * the per-thread descriptors. * * Semantics of FreeBSD version: I think we can ignore that Linux has * three per-thread descriptors and use just the first one. * The tls_array[] is used only in [gs]et_thread_area() syscalls and * for loading the GDT descriptors. We use just one GDT descriptor * for TLS, so we will load just one. * * XXX: This doesn't work when a user space process tries to use more * than one TLS segment. Comment in the Linux source says wine might * do this. */ /* * GLIBC reads current %gs and call set_thread_area() with it. * We should let GUDATA_SEL and GUGS32_SEL proceed as well because * we use these segments. */ switch (info.entry_number) { case GUGS32_SEL: case GUDATA_SEL: case 6: case -1: info.entry_number = GUGS32_SEL; break; default: return (EINVAL); } /* * We have to copy out the GDT entry we use. * * XXX: What if a user space program does not check the return value * and tries to use 6, 7 or 8? */ error = copyout(&info, args->desc, sizeof(struct l_user_desc)); if (error) return (error); if (LINUX_LDT_empty(&info)) { a[0] = 0; a[1] = 0; } else { a[0] = LINUX_LDT_entry_a(&info); a[1] = LINUX_LDT_entry_b(&info); } memcpy(&sd, &a, sizeof(a)); #ifdef DEBUG if (ldebug(set_thread_area)) printf("Segment created in set_thread_area: " "lobase: %x, hibase: %x, lolimit: %x, hilimit: %x, " "type: %i, dpl: %i, p: %i, xx: %i, long: %i, " "def32: %i, gran: %i\n", sd.sd_lobase, sd.sd_hibase, sd.sd_lolimit, sd.sd_hilimit, sd.sd_type, sd.sd_dpl, sd.sd_p, sd.sd_xx, sd.sd_long, sd.sd_def32, sd.sd_gran); #endif pcb = td->td_pcb; pcb->pcb_gsbase = (register_t)info.base_addr; set_pcb_flags(pcb, PCB_32BIT); update_gdt_gsbase(td, info.base_addr); return (0); -} - -int -linux_wait4(struct thread *td, struct linux_wait4_args *args) -{ - int error, options; - struct rusage ru, *rup; - struct l_rusage lru; - -#ifdef DEBUG - if (ldebug(wait4)) - printf(ARGS(wait4, "%d, %p, %d, %p"), - args->pid, (void *)args->status, args->options, - (void *)args->rusage); -#endif - - options = (args->options & (WNOHANG | WUNTRACED)); - /* WLINUXCLONE should be equal to __WCLONE, but we make sure */ - if (args->options & __WCLONE) - options |= WLINUXCLONE; - - if (args->rusage != NULL) - rup = &ru; - else - rup = NULL; - error = linux_common_wait(td, args->pid, args->status, options, rup); - if (error) - return (error); - if (args->rusage != NULL) { - bsd_to_linux_rusage(rup, &lru); - error = copyout(&lru, args->rusage, sizeof(lru)); - } - - return (error); } Index: head/sys/compat/linux/linux_misc.c =================================================================== --- head/sys/compat/linux/linux_misc.c (revision 283390) +++ head/sys/compat/linux/linux_misc.c (revision 283391) @@ -1,2029 +1,2058 @@ /*- * Copyright (c) 2002 Doug Rabson * Copyright (c) 1994-1995 Søren Schmidt * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer * in this position and unchanged. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. The name of the author may not be used to endorse or promote products * derived from this software without specific prior written permission * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include __FBSDID("$FreeBSD$"); #include "opt_compat.h" #include #include #include #if defined(__i386__) #include #endif #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef COMPAT_LINUX32 #include #include #else #include #include #endif #include #include #include #include #include #include #include int stclohz; /* Statistics clock frequency */ static unsigned int linux_to_bsd_resource[LINUX_RLIM_NLIMITS] = { RLIMIT_CPU, RLIMIT_FSIZE, RLIMIT_DATA, RLIMIT_STACK, RLIMIT_CORE, RLIMIT_RSS, RLIMIT_NPROC, RLIMIT_NOFILE, RLIMIT_MEMLOCK, RLIMIT_AS }; struct l_sysinfo { l_long uptime; /* Seconds since boot */ l_ulong loads[3]; /* 1, 5, and 15 minute load averages */ #define LINUX_SYSINFO_LOADS_SCALE 65536 l_ulong totalram; /* Total usable main memory size */ l_ulong freeram; /* Available memory size */ l_ulong sharedram; /* Amount of shared memory */ l_ulong bufferram; /* Memory used by buffers */ l_ulong totalswap; /* Total swap space size */ l_ulong freeswap; /* swap space still available */ l_ushort procs; /* Number of current processes */ l_ushort pads; l_ulong totalbig; l_ulong freebig; l_uint mem_unit; char _f[20-2*sizeof(l_long)-sizeof(l_int)]; /* padding */ }; int linux_sysinfo(struct thread *td, struct linux_sysinfo_args *args) { struct l_sysinfo sysinfo; vm_object_t object; int i, j; struct timespec ts; getnanouptime(&ts); if (ts.tv_nsec != 0) ts.tv_sec++; sysinfo.uptime = ts.tv_sec; /* Use the information from the mib to get our load averages */ for (i = 0; i < 3; i++) sysinfo.loads[i] = averunnable.ldavg[i] * LINUX_SYSINFO_LOADS_SCALE / averunnable.fscale; sysinfo.totalram = physmem * PAGE_SIZE; sysinfo.freeram = sysinfo.totalram - vm_cnt.v_wire_count * PAGE_SIZE; sysinfo.sharedram = 0; mtx_lock(&vm_object_list_mtx); TAILQ_FOREACH(object, &vm_object_list, object_list) if (object->shadow_count > 1) sysinfo.sharedram += object->resident_page_count; mtx_unlock(&vm_object_list_mtx); sysinfo.sharedram *= PAGE_SIZE; sysinfo.bufferram = 0; swap_pager_status(&i, &j); sysinfo.totalswap = i * PAGE_SIZE; sysinfo.freeswap = (i - j) * PAGE_SIZE; sysinfo.procs = nprocs; /* The following are only present in newer Linux kernels. */ sysinfo.totalbig = 0; sysinfo.freebig = 0; sysinfo.mem_unit = 1; return (copyout(&sysinfo, args->info, sizeof(sysinfo))); } int linux_alarm(struct thread *td, struct linux_alarm_args *args) { struct itimerval it, old_it; u_int secs; int error; #ifdef DEBUG if (ldebug(alarm)) printf(ARGS(alarm, "%u"), args->secs); #endif secs = args->secs; if (secs > INT_MAX) secs = INT_MAX; it.it_value.tv_sec = (long) secs; it.it_value.tv_usec = 0; it.it_interval.tv_sec = 0; it.it_interval.tv_usec = 0; error = kern_setitimer(td, ITIMER_REAL, &it, &old_it); if (error) return (error); if (timevalisset(&old_it.it_value)) { if (old_it.it_value.tv_usec != 0) old_it.it_value.tv_sec++; td->td_retval[0] = old_it.it_value.tv_sec; } return (0); } int linux_brk(struct thread *td, struct linux_brk_args *args) { struct vmspace *vm = td->td_proc->p_vmspace; vm_offset_t new, old; struct obreak_args /* { char * nsize; } */ tmp; #ifdef DEBUG if (ldebug(brk)) printf(ARGS(brk, "%p"), (void *)(uintptr_t)args->dsend); #endif old = (vm_offset_t)vm->vm_daddr + ctob(vm->vm_dsize); new = (vm_offset_t)args->dsend; tmp.nsize = (char *)new; if (((caddr_t)new > vm->vm_daddr) && !sys_obreak(td, &tmp)) td->td_retval[0] = (long)new; else td->td_retval[0] = (long)old; return (0); } #if defined(__i386__) /* XXX: what about amd64/linux32? */ int linux_uselib(struct thread *td, struct linux_uselib_args *args) { struct nameidata ni; struct vnode *vp; struct exec *a_out; struct vattr attr; vm_offset_t vmaddr; unsigned long file_offset; unsigned long bss_size; char *library; ssize_t aresid; int error, locked, writecount; LCONVPATHEXIST(td, args->library, &library); #ifdef DEBUG if (ldebug(uselib)) printf(ARGS(uselib, "%s"), library); #endif a_out = NULL; locked = 0; vp = NULL; NDINIT(&ni, LOOKUP, ISOPEN | FOLLOW | LOCKLEAF | AUDITVNODE1, UIO_SYSSPACE, library, td); error = namei(&ni); LFREEPATH(library); if (error) goto cleanup; vp = ni.ni_vp; NDFREE(&ni, NDF_ONLY_PNBUF); /* * From here on down, we have a locked vnode that must be unlocked. * XXX: The code below largely duplicates exec_check_permissions(). */ locked = 1; /* Writable? */ error = VOP_GET_WRITECOUNT(vp, &writecount); if (error != 0) goto cleanup; if (writecount != 0) { error = ETXTBSY; goto cleanup; } /* Executable? */ error = VOP_GETATTR(vp, &attr, td->td_ucred); if (error) goto cleanup; if ((vp->v_mount->mnt_flag & MNT_NOEXEC) || ((attr.va_mode & 0111) == 0) || (attr.va_type != VREG)) { /* EACCESS is what exec(2) returns. */ error = ENOEXEC; goto cleanup; } /* Sensible size? */ if (attr.va_size == 0) { error = ENOEXEC; goto cleanup; } /* Can we access it? */ error = VOP_ACCESS(vp, VEXEC, td->td_ucred, td); if (error) goto cleanup; /* * XXX: This should use vn_open() so that it is properly authorized, * and to reduce code redundancy all over the place here. * XXX: Not really, it duplicates far more of exec_check_permissions() * than vn_open(). */ #ifdef MAC error = mac_vnode_check_open(td->td_ucred, vp, VREAD); if (error) goto cleanup; #endif error = VOP_OPEN(vp, FREAD, td->td_ucred, td, NULL); if (error) goto cleanup; /* Pull in executable header into exec_map */ error = vm_mmap(exec_map, (vm_offset_t *)&a_out, PAGE_SIZE, VM_PROT_READ, VM_PROT_READ, 0, OBJT_VNODE, vp, 0); if (error) goto cleanup; /* Is it a Linux binary ? */ if (((a_out->a_magic >> 16) & 0xff) != 0x64) { error = ENOEXEC; goto cleanup; } /* * While we are here, we should REALLY do some more checks */ /* Set file/virtual offset based on a.out variant. */ switch ((int)(a_out->a_magic & 0xffff)) { case 0413: /* ZMAGIC */ file_offset = 1024; break; case 0314: /* QMAGIC */ file_offset = 0; break; default: error = ENOEXEC; goto cleanup; } bss_size = round_page(a_out->a_bss); /* Check various fields in header for validity/bounds. */ if (a_out->a_text & PAGE_MASK || a_out->a_data & PAGE_MASK) { error = ENOEXEC; goto cleanup; } /* text + data can't exceed file size */ if (a_out->a_data + a_out->a_text > attr.va_size) { error = EFAULT; goto cleanup; } /* * text/data/bss must not exceed limits * XXX - this is not complete. it should check current usage PLUS * the resources needed by this library. */ PROC_LOCK(td->td_proc); if (a_out->a_text > maxtsiz || a_out->a_data + bss_size > lim_cur(td->td_proc, RLIMIT_DATA) || racct_set(td->td_proc, RACCT_DATA, a_out->a_data + bss_size) != 0) { PROC_UNLOCK(td->td_proc); error = ENOMEM; goto cleanup; } PROC_UNLOCK(td->td_proc); /* * Prevent more writers. * XXX: Note that if any of the VM operations fail below we don't * clear this flag. */ VOP_SET_TEXT(vp); /* * Lock no longer needed */ locked = 0; VOP_UNLOCK(vp, 0); /* * Check if file_offset page aligned. Currently we cannot handle * misalinged file offsets, and so we read in the entire image * (what a waste). */ if (file_offset & PAGE_MASK) { #ifdef DEBUG printf("uselib: Non page aligned binary %lu\n", file_offset); #endif /* Map text+data read/write/execute */ /* a_entry is the load address and is page aligned */ vmaddr = trunc_page(a_out->a_entry); /* get anon user mapping, read+write+execute */ error = vm_map_find(&td->td_proc->p_vmspace->vm_map, NULL, 0, &vmaddr, a_out->a_text + a_out->a_data, 0, VMFS_NO_SPACE, VM_PROT_ALL, VM_PROT_ALL, 0); if (error) goto cleanup; error = vn_rdwr(UIO_READ, vp, (void *)vmaddr, file_offset, a_out->a_text + a_out->a_data, UIO_USERSPACE, 0, td->td_ucred, NOCRED, &aresid, td); if (error != 0) goto cleanup; if (aresid != 0) { error = ENOEXEC; goto cleanup; } } else { #ifdef DEBUG printf("uselib: Page aligned binary %lu\n", file_offset); #endif /* * for QMAGIC, a_entry is 20 bytes beyond the load address * to skip the executable header */ vmaddr = trunc_page(a_out->a_entry); /* * Map it all into the process's space as a single * copy-on-write "data" segment. */ error = vm_mmap(&td->td_proc->p_vmspace->vm_map, &vmaddr, a_out->a_text + a_out->a_data, VM_PROT_ALL, VM_PROT_ALL, MAP_PRIVATE | MAP_FIXED, OBJT_VNODE, vp, file_offset); if (error) goto cleanup; } #ifdef DEBUG printf("mem=%08lx = %08lx %08lx\n", (long)vmaddr, ((long *)vmaddr)[0], ((long *)vmaddr)[1]); #endif if (bss_size != 0) { /* Calculate BSS start address */ vmaddr = trunc_page(a_out->a_entry) + a_out->a_text + a_out->a_data; /* allocate some 'anon' space */ error = vm_map_find(&td->td_proc->p_vmspace->vm_map, NULL, 0, &vmaddr, bss_size, 0, VMFS_NO_SPACE, VM_PROT_ALL, VM_PROT_ALL, 0); if (error) goto cleanup; } cleanup: /* Unlock vnode if needed */ if (locked) VOP_UNLOCK(vp, 0); /* Release the temporary mapping. */ if (a_out) kmap_free_wakeup(exec_map, (vm_offset_t)a_out, PAGE_SIZE); return (error); } #endif /* __i386__ */ int linux_select(struct thread *td, struct linux_select_args *args) { l_timeval ltv; struct timeval tv0, tv1, utv, *tvp; int error; #ifdef DEBUG if (ldebug(select)) printf(ARGS(select, "%d, %p, %p, %p, %p"), args->nfds, (void *)args->readfds, (void *)args->writefds, (void *)args->exceptfds, (void *)args->timeout); #endif /* * Store current time for computation of the amount of * time left. */ if (args->timeout) { if ((error = copyin(args->timeout, <v, sizeof(ltv)))) goto select_out; utv.tv_sec = ltv.tv_sec; utv.tv_usec = ltv.tv_usec; #ifdef DEBUG if (ldebug(select)) printf(LMSG("incoming timeout (%jd/%ld)"), (intmax_t)utv.tv_sec, utv.tv_usec); #endif if (itimerfix(&utv)) { /* * The timeval was invalid. Convert it to something * valid that will act as it does under Linux. */ utv.tv_sec += utv.tv_usec / 1000000; utv.tv_usec %= 1000000; if (utv.tv_usec < 0) { utv.tv_sec -= 1; utv.tv_usec += 1000000; } if (utv.tv_sec < 0) timevalclear(&utv); } microtime(&tv0); tvp = &utv; } else tvp = NULL; error = kern_select(td, args->nfds, args->readfds, args->writefds, args->exceptfds, tvp, sizeof(l_int) * 8); #ifdef DEBUG if (ldebug(select)) printf(LMSG("real select returns %d"), error); #endif if (error) goto select_out; if (args->timeout) { if (td->td_retval[0]) { /* * Compute how much time was left of the timeout, * by subtracting the current time and the time * before we started the call, and subtracting * that result from the user-supplied value. */ microtime(&tv1); timevalsub(&tv1, &tv0); timevalsub(&utv, &tv1); if (utv.tv_sec < 0) timevalclear(&utv); } else timevalclear(&utv); #ifdef DEBUG if (ldebug(select)) printf(LMSG("outgoing timeout (%jd/%ld)"), (intmax_t)utv.tv_sec, utv.tv_usec); #endif ltv.tv_sec = utv.tv_sec; ltv.tv_usec = utv.tv_usec; if ((error = copyout(<v, args->timeout, sizeof(ltv)))) goto select_out; } select_out: #ifdef DEBUG if (ldebug(select)) printf(LMSG("select_out -> %d"), error); #endif return (error); } int linux_mremap(struct thread *td, struct linux_mremap_args *args) { struct munmap_args /* { void *addr; size_t len; } */ bsd_args; int error = 0; #ifdef DEBUG if (ldebug(mremap)) printf(ARGS(mremap, "%p, %08lx, %08lx, %08lx"), (void *)(uintptr_t)args->addr, (unsigned long)args->old_len, (unsigned long)args->new_len, (unsigned long)args->flags); #endif if (args->flags & ~(LINUX_MREMAP_FIXED | LINUX_MREMAP_MAYMOVE)) { td->td_retval[0] = 0; return (EINVAL); } /* * Check for the page alignment. * Linux defines PAGE_MASK to be FreeBSD ~PAGE_MASK. */ if (args->addr & PAGE_MASK) { td->td_retval[0] = 0; return (EINVAL); } args->new_len = round_page(args->new_len); args->old_len = round_page(args->old_len); if (args->new_len > args->old_len) { td->td_retval[0] = 0; return (ENOMEM); } if (args->new_len < args->old_len) { bsd_args.addr = (caddr_t)((uintptr_t)args->addr + args->new_len); bsd_args.len = args->old_len - args->new_len; error = sys_munmap(td, &bsd_args); } td->td_retval[0] = error ? 0 : (uintptr_t)args->addr; return (error); } #define LINUX_MS_ASYNC 0x0001 #define LINUX_MS_INVALIDATE 0x0002 #define LINUX_MS_SYNC 0x0004 int linux_msync(struct thread *td, struct linux_msync_args *args) { struct msync_args bsd_args; bsd_args.addr = (caddr_t)(uintptr_t)args->addr; bsd_args.len = (uintptr_t)args->len; bsd_args.flags = args->fl & ~LINUX_MS_SYNC; return (sys_msync(td, &bsd_args)); } int linux_time(struct thread *td, struct linux_time_args *args) { struct timeval tv; l_time_t tm; int error; #ifdef DEBUG if (ldebug(time)) printf(ARGS(time, "*")); #endif microtime(&tv); tm = tv.tv_sec; if (args->tm && (error = copyout(&tm, args->tm, sizeof(tm)))) return (error); td->td_retval[0] = tm; return (0); } struct l_times_argv { l_clock_t tms_utime; l_clock_t tms_stime; l_clock_t tms_cutime; l_clock_t tms_cstime; }; /* * Glibc versions prior to 2.2.1 always use hard-coded CLK_TCK value. * Since 2.2.1 Glibc uses value exported from kernel via AT_CLKTCK * auxiliary vector entry. */ #define CLK_TCK 100 #define CONVOTCK(r) (r.tv_sec * CLK_TCK + r.tv_usec / (1000000 / CLK_TCK)) #define CONVNTCK(r) (r.tv_sec * stclohz + r.tv_usec / (1000000 / stclohz)) #define CONVTCK(r) (linux_kernver(td) >= LINUX_KERNVER_2004000 ? \ CONVNTCK(r) : CONVOTCK(r)) int linux_times(struct thread *td, struct linux_times_args *args) { struct timeval tv, utime, stime, cutime, cstime; struct l_times_argv tms; struct proc *p; int error; #ifdef DEBUG if (ldebug(times)) printf(ARGS(times, "*")); #endif if (args->buf != NULL) { p = td->td_proc; PROC_LOCK(p); PROC_STATLOCK(p); calcru(p, &utime, &stime); PROC_STATUNLOCK(p); calccru(p, &cutime, &cstime); PROC_UNLOCK(p); tms.tms_utime = CONVTCK(utime); tms.tms_stime = CONVTCK(stime); tms.tms_cutime = CONVTCK(cutime); tms.tms_cstime = CONVTCK(cstime); if ((error = copyout(&tms, args->buf, sizeof(tms)))) return (error); } microuptime(&tv); td->td_retval[0] = (int)CONVTCK(tv); return (0); } int linux_newuname(struct thread *td, struct linux_newuname_args *args) { struct l_new_utsname utsname; char osname[LINUX_MAX_UTSNAME]; char osrelease[LINUX_MAX_UTSNAME]; char *p; #ifdef DEBUG if (ldebug(newuname)) printf(ARGS(newuname, "*")); #endif linux_get_osname(td, osname); linux_get_osrelease(td, osrelease); bzero(&utsname, sizeof(utsname)); strlcpy(utsname.sysname, osname, LINUX_MAX_UTSNAME); getcredhostname(td->td_ucred, utsname.nodename, LINUX_MAX_UTSNAME); getcreddomainname(td->td_ucred, utsname.domainname, LINUX_MAX_UTSNAME); strlcpy(utsname.release, osrelease, LINUX_MAX_UTSNAME); strlcpy(utsname.version, version, LINUX_MAX_UTSNAME); for (p = utsname.version; *p != '\0'; ++p) if (*p == '\n') { *p = '\0'; break; } strlcpy(utsname.machine, linux_platform, LINUX_MAX_UTSNAME); return (copyout(&utsname, args->buf, sizeof(utsname))); } #if defined(__i386__) || (defined(__amd64__) && defined(COMPAT_LINUX32)) struct l_utimbuf { l_time_t l_actime; l_time_t l_modtime; }; int linux_utime(struct thread *td, struct linux_utime_args *args) { struct timeval tv[2], *tvp; struct l_utimbuf lut; char *fname; int error; LCONVPATHEXIST(td, args->fname, &fname); #ifdef DEBUG if (ldebug(utime)) printf(ARGS(utime, "%s, *"), fname); #endif if (args->times) { if ((error = copyin(args->times, &lut, sizeof lut))) { LFREEPATH(fname); return (error); } tv[0].tv_sec = lut.l_actime; tv[0].tv_usec = 0; tv[1].tv_sec = lut.l_modtime; tv[1].tv_usec = 0; tvp = tv; } else tvp = NULL; error = kern_utimesat(td, AT_FDCWD, fname, UIO_SYSSPACE, tvp, UIO_SYSSPACE); LFREEPATH(fname); return (error); } int linux_utimes(struct thread *td, struct linux_utimes_args *args) { l_timeval ltv[2]; struct timeval tv[2], *tvp = NULL; char *fname; int error; LCONVPATHEXIST(td, args->fname, &fname); #ifdef DEBUG if (ldebug(utimes)) printf(ARGS(utimes, "%s, *"), fname); #endif if (args->tptr != NULL) { if ((error = copyin(args->tptr, ltv, sizeof ltv))) { LFREEPATH(fname); return (error); } tv[0].tv_sec = ltv[0].tv_sec; tv[0].tv_usec = ltv[0].tv_usec; tv[1].tv_sec = ltv[1].tv_sec; tv[1].tv_usec = ltv[1].tv_usec; tvp = tv; } error = kern_utimesat(td, AT_FDCWD, fname, UIO_SYSSPACE, tvp, UIO_SYSSPACE); LFREEPATH(fname); return (error); } int linux_futimesat(struct thread *td, struct linux_futimesat_args *args) { l_timeval ltv[2]; struct timeval tv[2], *tvp = NULL; char *fname; int error, dfd; dfd = (args->dfd == LINUX_AT_FDCWD) ? AT_FDCWD : args->dfd; LCONVPATHEXIST_AT(td, args->filename, &fname, dfd); #ifdef DEBUG if (ldebug(futimesat)) printf(ARGS(futimesat, "%s, *"), fname); #endif if (args->utimes != NULL) { if ((error = copyin(args->utimes, ltv, sizeof ltv))) { LFREEPATH(fname); return (error); } tv[0].tv_sec = ltv[0].tv_sec; tv[0].tv_usec = ltv[0].tv_usec; tv[1].tv_sec = ltv[1].tv_sec; tv[1].tv_usec = ltv[1].tv_usec; tvp = tv; } error = kern_utimesat(td, dfd, fname, UIO_SYSSPACE, tvp, UIO_SYSSPACE); LFREEPATH(fname); return (error); } #endif /* __i386__ || (__amd64__ && COMPAT_LINUX32) */ int linux_common_wait(struct thread *td, int pid, int *status, int options, struct rusage *ru) { int error, tmpstat; error = kern_wait(td, pid, &tmpstat, options, ru); if (error) return (error); if (status) { tmpstat &= 0xffff; if (WIFSIGNALED(tmpstat)) tmpstat = (tmpstat & 0xffffff80) | BSD_TO_LINUX_SIGNAL(WTERMSIG(tmpstat)); else if (WIFSTOPPED(tmpstat)) tmpstat = (tmpstat & 0xffff00ff) | (BSD_TO_LINUX_SIGNAL(WSTOPSIG(tmpstat)) << 8); error = copyout(&tmpstat, status, sizeof(int)); } return (error); } int linux_waitpid(struct thread *td, struct linux_waitpid_args *args) { int options; #ifdef DEBUG if (ldebug(waitpid)) printf(ARGS(waitpid, "%d, %p, %d"), args->pid, (void *)args->status, args->options); #endif /* * this is necessary because the test in kern_wait doesn't work * because we mess with the options here */ if (args->options & ~(WUNTRACED | WNOHANG | WCONTINUED | __WCLONE)) return (EINVAL); options = (args->options & (WNOHANG | WUNTRACED)); /* WLINUXCLONE should be equal to __WCLONE, but we make sure */ if (args->options & __WCLONE) options |= WLINUXCLONE; return (linux_common_wait(td, args->pid, args->status, options, NULL)); } +int +linux_wait4(struct thread *td, struct linux_wait4_args *args) +{ + int error, options; + struct rusage ru, *rup; + +#ifdef DEBUG + if (ldebug(wait4)) + printf(ARGS(wait4, "%d, %p, %d, %p"), + args->pid, (void *)args->status, args->options, + (void *)args->rusage); +#endif + + options = (args->options & (WNOHANG | WUNTRACED)); + /* WLINUXCLONE should be equal to __WCLONE, but we make sure */ + if (args->options & __WCLONE) + options |= WLINUXCLONE; + + if (args->rusage != NULL) + rup = &ru; + else + rup = NULL; + error = linux_common_wait(td, args->pid, args->status, options, rup); + if (error != 0) + return (error); + if (args->rusage != NULL) + error = linux_copyout_rusage(&ru, args->rusage); + return (error); +} int linux_mknod(struct thread *td, struct linux_mknod_args *args) { char *path; int error; LCONVPATHCREAT(td, args->path, &path); #ifdef DEBUG if (ldebug(mknod)) printf(ARGS(mknod, "%s, %d, %d"), path, args->mode, args->dev); #endif switch (args->mode & S_IFMT) { case S_IFIFO: case S_IFSOCK: error = kern_mkfifoat(td, AT_FDCWD, path, UIO_SYSSPACE, args->mode); break; case S_IFCHR: case S_IFBLK: error = kern_mknodat(td, AT_FDCWD, path, UIO_SYSSPACE, args->mode, args->dev); break; case S_IFDIR: error = EPERM; break; case 0: args->mode |= S_IFREG; /* FALLTHROUGH */ case S_IFREG: error = kern_openat(td, AT_FDCWD, path, UIO_SYSSPACE, O_WRONLY | O_CREAT | O_TRUNC, args->mode); if (error == 0) kern_close(td, td->td_retval[0]); break; default: error = EINVAL; break; } LFREEPATH(path); return (error); } int linux_mknodat(struct thread *td, struct linux_mknodat_args *args) { char *path; int error, dfd; dfd = (args->dfd == LINUX_AT_FDCWD) ? AT_FDCWD : args->dfd; LCONVPATHCREAT_AT(td, args->filename, &path, dfd); #ifdef DEBUG if (ldebug(mknodat)) printf(ARGS(mknodat, "%s, %d, %d"), path, args->mode, args->dev); #endif switch (args->mode & S_IFMT) { case S_IFIFO: case S_IFSOCK: error = kern_mkfifoat(td, dfd, path, UIO_SYSSPACE, args->mode); break; case S_IFCHR: case S_IFBLK: error = kern_mknodat(td, dfd, path, UIO_SYSSPACE, args->mode, args->dev); break; case S_IFDIR: error = EPERM; break; case 0: args->mode |= S_IFREG; /* FALLTHROUGH */ case S_IFREG: error = kern_openat(td, dfd, path, UIO_SYSSPACE, O_WRONLY | O_CREAT | O_TRUNC, args->mode); if (error == 0) kern_close(td, td->td_retval[0]); break; default: error = EINVAL; break; } LFREEPATH(path); return (error); } /* * UGH! This is just about the dumbest idea I've ever heard!! */ int linux_personality(struct thread *td, struct linux_personality_args *args) { #ifdef DEBUG if (ldebug(personality)) printf(ARGS(personality, "%lu"), (unsigned long)args->per); #endif if (args->per != 0) return (EINVAL); /* Yes Jim, it's still a Linux... */ td->td_retval[0] = 0; return (0); } struct l_itimerval { l_timeval it_interval; l_timeval it_value; }; #define B2L_ITIMERVAL(bip, lip) \ (bip)->it_interval.tv_sec = (lip)->it_interval.tv_sec; \ (bip)->it_interval.tv_usec = (lip)->it_interval.tv_usec; \ (bip)->it_value.tv_sec = (lip)->it_value.tv_sec; \ (bip)->it_value.tv_usec = (lip)->it_value.tv_usec; int linux_setitimer(struct thread *td, struct linux_setitimer_args *uap) { int error; struct l_itimerval ls; struct itimerval aitv, oitv; #ifdef DEBUG if (ldebug(setitimer)) printf(ARGS(setitimer, "%p, %p"), (void *)uap->itv, (void *)uap->oitv); #endif if (uap->itv == NULL) { uap->itv = uap->oitv; return (linux_getitimer(td, (struct linux_getitimer_args *)uap)); } error = copyin(uap->itv, &ls, sizeof(ls)); if (error != 0) return (error); B2L_ITIMERVAL(&aitv, &ls); #ifdef DEBUG if (ldebug(setitimer)) { printf("setitimer: value: sec: %jd, usec: %ld\n", (intmax_t)aitv.it_value.tv_sec, aitv.it_value.tv_usec); printf("setitimer: interval: sec: %jd, usec: %ld\n", (intmax_t)aitv.it_interval.tv_sec, aitv.it_interval.tv_usec); } #endif error = kern_setitimer(td, uap->which, &aitv, &oitv); if (error != 0 || uap->oitv == NULL) return (error); B2L_ITIMERVAL(&ls, &oitv); return (copyout(&ls, uap->oitv, sizeof(ls))); } int linux_getitimer(struct thread *td, struct linux_getitimer_args *uap) { int error; struct l_itimerval ls; struct itimerval aitv; #ifdef DEBUG if (ldebug(getitimer)) printf(ARGS(getitimer, "%p"), (void *)uap->itv); #endif error = kern_getitimer(td, uap->which, &aitv); if (error != 0) return (error); B2L_ITIMERVAL(&ls, &aitv); return (copyout(&ls, uap->itv, sizeof(ls))); } int linux_nice(struct thread *td, struct linux_nice_args *args) { struct setpriority_args bsd_args; bsd_args.which = PRIO_PROCESS; bsd_args.who = 0; /* current process */ bsd_args.prio = args->inc; return (sys_setpriority(td, &bsd_args)); } int linux_setgroups(struct thread *td, struct linux_setgroups_args *args) { struct ucred *newcred, *oldcred; l_gid_t *linux_gidset; gid_t *bsd_gidset; int ngrp, error; struct proc *p; ngrp = args->gidsetsize; if (ngrp < 0 || ngrp >= ngroups_max + 1) return (EINVAL); linux_gidset = malloc(ngrp * sizeof(*linux_gidset), M_TEMP, M_WAITOK); error = copyin(args->grouplist, linux_gidset, ngrp * sizeof(l_gid_t)); if (error) goto out; newcred = crget(); p = td->td_proc; PROC_LOCK(p); oldcred = crcopysafe(p, newcred); /* * cr_groups[0] holds egid. Setting the whole set from * the supplied set will cause egid to be changed too. * Keep cr_groups[0] unchanged to prevent that. */ if ((error = priv_check_cred(oldcred, PRIV_CRED_SETGROUPS, 0)) != 0) { PROC_UNLOCK(p); crfree(newcred); goto out; } if (ngrp > 0) { newcred->cr_ngroups = ngrp + 1; bsd_gidset = newcred->cr_groups; ngrp--; while (ngrp >= 0) { bsd_gidset[ngrp + 1] = linux_gidset[ngrp]; ngrp--; } } else newcred->cr_ngroups = 1; setsugid(p); proc_set_cred(p, newcred); PROC_UNLOCK(p); crfree(oldcred); error = 0; out: free(linux_gidset, M_TEMP); return (error); } int linux_getgroups(struct thread *td, struct linux_getgroups_args *args) { struct ucred *cred; l_gid_t *linux_gidset; gid_t *bsd_gidset; int bsd_gidsetsz, ngrp, error; cred = td->td_ucred; bsd_gidset = cred->cr_groups; bsd_gidsetsz = cred->cr_ngroups - 1; /* * cr_groups[0] holds egid. Returning the whole set * here will cause a duplicate. Exclude cr_groups[0] * to prevent that. */ if ((ngrp = args->gidsetsize) == 0) { td->td_retval[0] = bsd_gidsetsz; return (0); } if (ngrp < bsd_gidsetsz) return (EINVAL); ngrp = 0; linux_gidset = malloc(bsd_gidsetsz * sizeof(*linux_gidset), M_TEMP, M_WAITOK); while (ngrp < bsd_gidsetsz) { linux_gidset[ngrp] = bsd_gidset[ngrp + 1]; ngrp++; } error = copyout(linux_gidset, args->grouplist, ngrp * sizeof(l_gid_t)); free(linux_gidset, M_TEMP); if (error) return (error); td->td_retval[0] = ngrp; return (0); } int linux_setrlimit(struct thread *td, struct linux_setrlimit_args *args) { struct rlimit bsd_rlim; struct l_rlimit rlim; u_int which; int error; #ifdef DEBUG if (ldebug(setrlimit)) printf(ARGS(setrlimit, "%d, %p"), args->resource, (void *)args->rlim); #endif if (args->resource >= LINUX_RLIM_NLIMITS) return (EINVAL); which = linux_to_bsd_resource[args->resource]; if (which == -1) return (EINVAL); error = copyin(args->rlim, &rlim, sizeof(rlim)); if (error) return (error); bsd_rlim.rlim_cur = (rlim_t)rlim.rlim_cur; bsd_rlim.rlim_max = (rlim_t)rlim.rlim_max; return (kern_setrlimit(td, which, &bsd_rlim)); } int linux_old_getrlimit(struct thread *td, struct linux_old_getrlimit_args *args) { struct l_rlimit rlim; struct proc *p = td->td_proc; struct rlimit bsd_rlim; u_int which; #ifdef DEBUG if (ldebug(old_getrlimit)) printf(ARGS(old_getrlimit, "%d, %p"), args->resource, (void *)args->rlim); #endif if (args->resource >= LINUX_RLIM_NLIMITS) return (EINVAL); which = linux_to_bsd_resource[args->resource]; if (which == -1) return (EINVAL); PROC_LOCK(p); lim_rlimit(p, which, &bsd_rlim); PROC_UNLOCK(p); #ifdef COMPAT_LINUX32 rlim.rlim_cur = (unsigned int)bsd_rlim.rlim_cur; if (rlim.rlim_cur == UINT_MAX) rlim.rlim_cur = INT_MAX; rlim.rlim_max = (unsigned int)bsd_rlim.rlim_max; if (rlim.rlim_max == UINT_MAX) rlim.rlim_max = INT_MAX; #else rlim.rlim_cur = (unsigned long)bsd_rlim.rlim_cur; if (rlim.rlim_cur == ULONG_MAX) rlim.rlim_cur = LONG_MAX; rlim.rlim_max = (unsigned long)bsd_rlim.rlim_max; if (rlim.rlim_max == ULONG_MAX) rlim.rlim_max = LONG_MAX; #endif return (copyout(&rlim, args->rlim, sizeof(rlim))); } int linux_getrlimit(struct thread *td, struct linux_getrlimit_args *args) { struct l_rlimit rlim; struct proc *p = td->td_proc; struct rlimit bsd_rlim; u_int which; #ifdef DEBUG if (ldebug(getrlimit)) printf(ARGS(getrlimit, "%d, %p"), args->resource, (void *)args->rlim); #endif if (args->resource >= LINUX_RLIM_NLIMITS) return (EINVAL); which = linux_to_bsd_resource[args->resource]; if (which == -1) return (EINVAL); PROC_LOCK(p); lim_rlimit(p, which, &bsd_rlim); PROC_UNLOCK(p); rlim.rlim_cur = (l_ulong)bsd_rlim.rlim_cur; rlim.rlim_max = (l_ulong)bsd_rlim.rlim_max; return (copyout(&rlim, args->rlim, sizeof(rlim))); } int linux_sched_setscheduler(struct thread *td, struct linux_sched_setscheduler_args *args) { struct sched_param sched_param; struct thread *tdt; int error, policy; #ifdef DEBUG if (ldebug(sched_setscheduler)) printf(ARGS(sched_setscheduler, "%d, %d, %p"), args->pid, args->policy, (const void *)args->param); #endif switch (args->policy) { case LINUX_SCHED_OTHER: policy = SCHED_OTHER; break; case LINUX_SCHED_FIFO: policy = SCHED_FIFO; break; case LINUX_SCHED_RR: policy = SCHED_RR; break; default: return (EINVAL); } error = copyin(args->param, &sched_param, sizeof(sched_param)); if (error) return (error); tdt = linux_tdfind(td, args->pid, -1); if (tdt == NULL) return (ESRCH); error = kern_sched_setscheduler(td, tdt, policy, &sched_param); PROC_UNLOCK(tdt->td_proc); return (error); } int linux_sched_getscheduler(struct thread *td, struct linux_sched_getscheduler_args *args) { struct thread *tdt; int error, policy; #ifdef DEBUG if (ldebug(sched_getscheduler)) printf(ARGS(sched_getscheduler, "%d"), args->pid); #endif tdt = linux_tdfind(td, args->pid, -1); if (tdt == NULL) return (ESRCH); error = kern_sched_getscheduler(td, tdt, &policy); PROC_UNLOCK(tdt->td_proc); switch (policy) { case SCHED_OTHER: td->td_retval[0] = LINUX_SCHED_OTHER; break; case SCHED_FIFO: td->td_retval[0] = LINUX_SCHED_FIFO; break; case SCHED_RR: td->td_retval[0] = LINUX_SCHED_RR; break; } return (error); } int linux_sched_get_priority_max(struct thread *td, struct linux_sched_get_priority_max_args *args) { struct sched_get_priority_max_args bsd; #ifdef DEBUG if (ldebug(sched_get_priority_max)) printf(ARGS(sched_get_priority_max, "%d"), args->policy); #endif switch (args->policy) { case LINUX_SCHED_OTHER: bsd.policy = SCHED_OTHER; break; case LINUX_SCHED_FIFO: bsd.policy = SCHED_FIFO; break; case LINUX_SCHED_RR: bsd.policy = SCHED_RR; break; default: return (EINVAL); } return (sys_sched_get_priority_max(td, &bsd)); } int linux_sched_get_priority_min(struct thread *td, struct linux_sched_get_priority_min_args *args) { struct sched_get_priority_min_args bsd; #ifdef DEBUG if (ldebug(sched_get_priority_min)) printf(ARGS(sched_get_priority_min, "%d"), args->policy); #endif switch (args->policy) { case LINUX_SCHED_OTHER: bsd.policy = SCHED_OTHER; break; case LINUX_SCHED_FIFO: bsd.policy = SCHED_FIFO; break; case LINUX_SCHED_RR: bsd.policy = SCHED_RR; break; default: return (EINVAL); } return (sys_sched_get_priority_min(td, &bsd)); } #define REBOOT_CAD_ON 0x89abcdef #define REBOOT_CAD_OFF 0 #define REBOOT_HALT 0xcdef0123 #define REBOOT_RESTART 0x01234567 #define REBOOT_RESTART2 0xA1B2C3D4 #define REBOOT_POWEROFF 0x4321FEDC #define REBOOT_MAGIC1 0xfee1dead #define REBOOT_MAGIC2 0x28121969 #define REBOOT_MAGIC2A 0x05121996 #define REBOOT_MAGIC2B 0x16041998 int linux_reboot(struct thread *td, struct linux_reboot_args *args) { struct reboot_args bsd_args; #ifdef DEBUG if (ldebug(reboot)) printf(ARGS(reboot, "0x%x"), args->cmd); #endif if (args->magic1 != REBOOT_MAGIC1) return (EINVAL); switch (args->magic2) { case REBOOT_MAGIC2: case REBOOT_MAGIC2A: case REBOOT_MAGIC2B: break; default: return (EINVAL); } switch (args->cmd) { case REBOOT_CAD_ON: case REBOOT_CAD_OFF: return (priv_check(td, PRIV_REBOOT)); case REBOOT_HALT: bsd_args.opt = RB_HALT; break; case REBOOT_RESTART: case REBOOT_RESTART2: bsd_args.opt = 0; break; case REBOOT_POWEROFF: bsd_args.opt = RB_POWEROFF; break; default: return (EINVAL); } return (sys_reboot(td, &bsd_args)); } /* * The FreeBSD native getpid(2), getgid(2) and getuid(2) also modify * td->td_retval[1] when COMPAT_43 is defined. This clobbers registers that * are assumed to be preserved. The following lightweight syscalls fixes * this. See also linux_getgid16() and linux_getuid16() in linux_uid16.c * * linux_getpid() - MP SAFE * linux_getgid() - MP SAFE * linux_getuid() - MP SAFE */ int linux_getpid(struct thread *td, struct linux_getpid_args *args) { #ifdef DEBUG if (ldebug(getpid)) printf(ARGS(getpid, "")); #endif td->td_retval[0] = td->td_proc->p_pid; return (0); } int linux_gettid(struct thread *td, struct linux_gettid_args *args) { struct linux_emuldata *em; #ifdef DEBUG if (ldebug(gettid)) printf(ARGS(gettid, "")); #endif em = em_find(td); KASSERT(em != NULL, ("gettid: emuldata not found.\n")); td->td_retval[0] = em->em_tid; return (0); } int linux_getppid(struct thread *td, struct linux_getppid_args *args) { #ifdef DEBUG if (ldebug(getppid)) printf(ARGS(getppid, "")); #endif PROC_LOCK(td->td_proc); td->td_retval[0] = td->td_proc->p_pptr->p_pid; PROC_UNLOCK(td->td_proc); return (0); } int linux_getgid(struct thread *td, struct linux_getgid_args *args) { #ifdef DEBUG if (ldebug(getgid)) printf(ARGS(getgid, "")); #endif td->td_retval[0] = td->td_ucred->cr_rgid; return (0); } int linux_getuid(struct thread *td, struct linux_getuid_args *args) { #ifdef DEBUG if (ldebug(getuid)) printf(ARGS(getuid, "")); #endif td->td_retval[0] = td->td_ucred->cr_ruid; return (0); } int linux_getsid(struct thread *td, struct linux_getsid_args *args) { struct getsid_args bsd; #ifdef DEBUG if (ldebug(getsid)) printf(ARGS(getsid, "%i"), args->pid); #endif bsd.pid = args->pid; return (sys_getsid(td, &bsd)); } int linux_nosys(struct thread *td, struct nosys_args *ignore) { return (ENOSYS); } int linux_getpriority(struct thread *td, struct linux_getpriority_args *args) { struct getpriority_args bsd_args; int error; #ifdef DEBUG if (ldebug(getpriority)) printf(ARGS(getpriority, "%i, %i"), args->which, args->who); #endif bsd_args.which = args->which; bsd_args.who = args->who; error = sys_getpriority(td, &bsd_args); td->td_retval[0] = 20 - td->td_retval[0]; return (error); } int linux_sethostname(struct thread *td, struct linux_sethostname_args *args) { int name[2]; #ifdef DEBUG if (ldebug(sethostname)) printf(ARGS(sethostname, "*, %i"), args->len); #endif name[0] = CTL_KERN; name[1] = KERN_HOSTNAME; return (userland_sysctl(td, name, 2, 0, 0, 0, args->hostname, args->len, 0, 0)); } int linux_setdomainname(struct thread *td, struct linux_setdomainname_args *args) { int name[2]; #ifdef DEBUG if (ldebug(setdomainname)) printf(ARGS(setdomainname, "*, %i"), args->len); #endif name[0] = CTL_KERN; name[1] = KERN_NISDOMAINNAME; return (userland_sysctl(td, name, 2, 0, 0, 0, args->name, args->len, 0, 0)); } int linux_exit_group(struct thread *td, struct linux_exit_group_args *args) { #ifdef DEBUG if (ldebug(exit_group)) printf(ARGS(exit_group, "%i"), args->error_code); #endif LINUX_CTR2(exit_group, "thread(%d) (%d)", td->td_tid, args->error_code); /* * XXX: we should send a signal to the parent if * SIGNAL_EXIT_GROUP is set. We ignore that (temporarily?) * as it doesnt occur often. */ exit1(td, W_EXITCODE(args->error_code, 0)); /* NOTREACHED */ } #define _LINUX_CAPABILITY_VERSION 0x19980330 struct l_user_cap_header { l_int version; l_int pid; }; struct l_user_cap_data { l_int effective; l_int permitted; l_int inheritable; }; int linux_capget(struct thread *td, struct linux_capget_args *args) { struct l_user_cap_header luch; struct l_user_cap_data lucd; int error; if (args->hdrp == NULL) return (EFAULT); error = copyin(args->hdrp, &luch, sizeof(luch)); if (error != 0) return (error); if (luch.version != _LINUX_CAPABILITY_VERSION) { luch.version = _LINUX_CAPABILITY_VERSION; error = copyout(&luch, args->hdrp, sizeof(luch)); if (error) return (error); return (EINVAL); } if (luch.pid) return (EPERM); if (args->datap) { /* * The current implementation doesn't support setting * a capability (it's essentially a stub) so indicate * that no capabilities are currently set or available * to request. */ bzero (&lucd, sizeof(lucd)); error = copyout(&lucd, args->datap, sizeof(lucd)); } return (error); } int linux_capset(struct thread *td, struct linux_capset_args *args) { struct l_user_cap_header luch; struct l_user_cap_data lucd; int error; if (args->hdrp == NULL || args->datap == NULL) return (EFAULT); error = copyin(args->hdrp, &luch, sizeof(luch)); if (error != 0) return (error); if (luch.version != _LINUX_CAPABILITY_VERSION) { luch.version = _LINUX_CAPABILITY_VERSION; error = copyout(&luch, args->hdrp, sizeof(luch)); if (error) return (error); return (EINVAL); } if (luch.pid) return (EPERM); error = copyin(args->datap, &lucd, sizeof(lucd)); if (error != 0) return (error); /* We currently don't support setting any capabilities. */ if (lucd.effective || lucd.permitted || lucd.inheritable) { linux_msg(td, "capset effective=0x%x, permitted=0x%x, " "inheritable=0x%x is not implemented", (int)lucd.effective, (int)lucd.permitted, (int)lucd.inheritable); return (EPERM); } return (0); } int linux_prctl(struct thread *td, struct linux_prctl_args *args) { int error = 0, max_size; struct proc *p = td->td_proc; char comm[LINUX_MAX_COMM_LEN]; struct linux_emuldata *em; int pdeath_signal; #ifdef DEBUG if (ldebug(prctl)) printf(ARGS(prctl, "%d, %d, %d, %d, %d"), args->option, args->arg2, args->arg3, args->arg4, args->arg5); #endif switch (args->option) { case LINUX_PR_SET_PDEATHSIG: if (!LINUX_SIG_VALID(args->arg2)) return (EINVAL); em = em_find(td); KASSERT(em != NULL, ("prctl: emuldata not found.\n")); em->pdeath_signal = args->arg2; break; case LINUX_PR_GET_PDEATHSIG: em = em_find(td); KASSERT(em != NULL, ("prctl: emuldata not found.\n")); pdeath_signal = em->pdeath_signal; error = copyout(&pdeath_signal, (void *)(register_t)args->arg2, sizeof(pdeath_signal)); break; case LINUX_PR_GET_KEEPCAPS: /* * Indicate that we always clear the effective and * permitted capability sets when the user id becomes * non-zero (actually the capability sets are simply * always zero in the current implementation). */ td->td_retval[0] = 0; break; case LINUX_PR_SET_KEEPCAPS: /* * Ignore requests to keep the effective and permitted * capability sets when the user id becomes non-zero. */ break; case LINUX_PR_SET_NAME: /* * To be on the safe side we need to make sure to not * overflow the size a linux program expects. We already * do this here in the copyin, so that we don't need to * check on copyout. */ max_size = MIN(sizeof(comm), sizeof(p->p_comm)); error = copyinstr((void *)(register_t)args->arg2, comm, max_size, NULL); /* Linux silently truncates the name if it is too long. */ if (error == ENAMETOOLONG) { /* * XXX: copyinstr() isn't documented to populate the * array completely, so do a copyin() to be on the * safe side. This should be changed in case * copyinstr() is changed to guarantee this. */ error = copyin((void *)(register_t)args->arg2, comm, max_size - 1); comm[max_size - 1] = '\0'; } if (error) return (error); PROC_LOCK(p); strlcpy(p->p_comm, comm, sizeof(p->p_comm)); PROC_UNLOCK(p); break; case LINUX_PR_GET_NAME: PROC_LOCK(p); strlcpy(comm, p->p_comm, sizeof(comm)); PROC_UNLOCK(p); error = copyout(comm, (void *)(register_t)args->arg2, strlen(comm) + 1); break; default: error = EINVAL; break; } return (error); } int linux_sched_setparam(struct thread *td, struct linux_sched_setparam_args *uap) { struct sched_param sched_param; struct thread *tdt; int error; #ifdef DEBUG if (ldebug(sched_setparam)) printf(ARGS(sched_setparam, "%d, *"), uap->pid); #endif error = copyin(uap->param, &sched_param, sizeof(sched_param)); if (error) return (error); tdt = linux_tdfind(td, uap->pid, -1); if (tdt == NULL) return (ESRCH); error = kern_sched_setparam(td, tdt, &sched_param); PROC_UNLOCK(tdt->td_proc); return (error); } int linux_sched_getparam(struct thread *td, struct linux_sched_getparam_args *uap) { struct sched_param sched_param; struct thread *tdt; int error; #ifdef DEBUG if (ldebug(sched_getparam)) printf(ARGS(sched_getparam, "%d, *"), uap->pid); #endif tdt = linux_tdfind(td, uap->pid, -1); if (tdt == NULL) return (ESRCH); error = kern_sched_getparam(td, tdt, &sched_param); PROC_UNLOCK(tdt->td_proc); if (error == 0) error = copyout(&sched_param, uap->param, sizeof(sched_param)); return (error); } /* * Get affinity of a process. */ int linux_sched_getaffinity(struct thread *td, struct linux_sched_getaffinity_args *args) { int error; struct thread *tdt; struct cpuset_getaffinity_args cga; #ifdef DEBUG if (ldebug(sched_getaffinity)) printf(ARGS(sched_getaffinity, "%d, %d, *"), args->pid, args->len); #endif if (args->len < sizeof(cpuset_t)) return (EINVAL); tdt = linux_tdfind(td, args->pid, -1); if (tdt == NULL) return (ESRCH); PROC_UNLOCK(tdt->td_proc); cga.level = CPU_LEVEL_WHICH; cga.which = CPU_WHICH_TID; cga.id = tdt->td_tid; cga.cpusetsize = sizeof(cpuset_t); cga.mask = (cpuset_t *) args->user_mask_ptr; if ((error = sys_cpuset_getaffinity(td, &cga)) == 0) td->td_retval[0] = sizeof(cpuset_t); return (error); } /* * Set affinity of a process. */ int linux_sched_setaffinity(struct thread *td, struct linux_sched_setaffinity_args *args) { struct cpuset_setaffinity_args csa; struct thread *tdt; #ifdef DEBUG if (ldebug(sched_setaffinity)) printf(ARGS(sched_setaffinity, "%d, %d, *"), args->pid, args->len); #endif if (args->len < sizeof(cpuset_t)) return (EINVAL); tdt = linux_tdfind(td, args->pid, -1); if (tdt == NULL) return (ESRCH); PROC_UNLOCK(tdt->td_proc); csa.level = CPU_LEVEL_WHICH; csa.which = CPU_WHICH_TID; csa.id = tdt->td_tid; csa.cpusetsize = sizeof(cpuset_t); csa.mask = (cpuset_t *) args->user_mask_ptr; return (sys_cpuset_setaffinity(td, &csa)); } int linux_sched_rr_get_interval(struct thread *td, struct linux_sched_rr_get_interval_args *uap) { struct timespec ts; struct l_timespec lts; struct thread *tdt; int error; tdt = linux_tdfind(td, uap->pid, -1); if (tdt == NULL) return (ESRCH); error = kern_sched_rr_get_interval_td(td, tdt, &ts); PROC_UNLOCK(tdt->td_proc); if (error != 0) return (error); lts.tv_sec = ts.tv_sec; lts.tv_nsec = ts.tv_nsec; return (copyout(<s, uap->interval, sizeof(lts))); } /* * In case when the Linux thread is the initial thread in * the thread group thread id is equal to the process id. * Glibc depends on this magic (assert in pthread_getattr_np.c). */ struct thread * linux_tdfind(struct thread *td, lwpid_t tid, pid_t pid) { struct linux_emuldata *em; struct thread *tdt; struct proc *p; tdt = NULL; if (tid == 0 || tid == td->td_tid) { tdt = td; PROC_LOCK(tdt->td_proc); } else if (tid > PID_MAX) tdt = tdfind(tid, pid); else { /* * Initial thread where the tid equal to the pid. */ p = pfind(tid); if (p != NULL) { if (SV_PROC_ABI(p) != SV_ABI_LINUX) { /* * p is not a Linuxulator process. */ PROC_UNLOCK(p); return (NULL); } FOREACH_THREAD_IN_PROC(p, tdt) { em = em_find(tdt); if (tid == em->em_tid) return (tdt); } PROC_UNLOCK(p); } return (NULL); } return (tdt); } void linux_to_bsd_waitopts(int options, int *bsdopts) { if (options & LINUX_WNOHANG) *bsdopts |= WNOHANG; if (options & LINUX_WUNTRACED) *bsdopts |= WUNTRACED; if (options & LINUX_WEXITED) *bsdopts |= WEXITED; if (options & LINUX_WCONTINUED) *bsdopts |= WCONTINUED; if (options & LINUX_WNOWAIT) *bsdopts |= WNOWAIT; if (options & __WCLONE) *bsdopts |= WLINUXCLONE; } Index: head/sys/i386/linux/linux.h =================================================================== --- head/sys/i386/linux/linux.h (revision 283390) +++ head/sys/i386/linux/linux.h (revision 283391) @@ -1,755 +1,757 @@ /*- * Copyright (c) 1994-1996 Søren Schmidt * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer * in this position and unchanged. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. The name of the author may not be used to endorse or promote products * derived from this software without specific prior written permission * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. * * $FreeBSD$ */ #ifndef _I386_LINUX_H_ #define _I386_LINUX_H_ #include /* for sigval union */ #include /* * debugging support */ extern u_char linux_debug_map[]; #define ldebug(name) isclr(linux_debug_map, LINUX_SYS_linux_ ## name) #define ARGS(nm, fmt) "linux(%ld/%ld): "#nm"("fmt")\n", \ (long)td->td_proc->p_pid, (long)td->td_tid #define LMSG(fmt) "linux(%ld/%ld): "fmt"\n", \ (long)td->td_proc->p_pid, (long)td->td_tid #define LINUX_DTRACE linuxulator #ifdef MALLOC_DECLARE MALLOC_DECLARE(M_LINUX); #endif #define LINUX_SHAREDPAGE (VM_MAXUSER_ADDRESS - PAGE_SIZE) #define LINUX_USRSTACK LINUX_SHAREDPAGE #define PTRIN(v) (void *)(v) #define PTROUT(v) (l_uintptr_t)(v) #define CP(src,dst,fld) do { (dst).fld = (src).fld; } while (0) #define CP2(src,dst,sfld,dfld) do { (dst).dfld = (src).sfld; } while (0) #define PTRIN_CP(src,dst,fld) \ do { (dst).fld = PTRIN((src).fld); } while (0) /* * Provide a separate set of types for the Linux types. */ typedef int l_int; typedef int32_t l_long; typedef int64_t l_longlong; typedef short l_short; typedef unsigned int l_uint; typedef uint32_t l_ulong; typedef uint64_t l_ulonglong; typedef unsigned short l_ushort; typedef char *l_caddr_t; typedef l_ulong l_uintptr_t; typedef l_long l_clock_t; typedef l_int l_daddr_t; typedef l_ushort l_dev_t; typedef l_uint l_gid_t; typedef l_ushort l_gid16_t; typedef l_ulong l_ino_t; typedef l_int l_key_t; typedef l_longlong l_loff_t; typedef l_ushort l_mode_t; typedef l_long l_off_t; typedef l_int l_pid_t; typedef l_uint l_size_t; typedef l_long l_suseconds_t; typedef l_long l_time_t; typedef l_uint l_uid_t; typedef l_ushort l_uid16_t; typedef l_int l_timer_t; typedef l_int l_mqd_t; typedef struct { l_int val[2]; } l_fsid_t; typedef struct { l_time_t tv_sec; l_suseconds_t tv_usec; } l_timeval; #define l_fd_set fd_set /* * Miscellaneous */ #define LINUX_AT_COUNT 16 /* Count of used aux entry types. * Keep this synchronized with * elf_linux_fixup() code. */ struct l___sysctl_args { l_int *name; l_int nlen; void *oldval; l_size_t *oldlenp; void *newval; l_size_t newlen; l_ulong __spare[4]; }; /* Resource limits */ #define LINUX_RLIMIT_CPU 0 #define LINUX_RLIMIT_FSIZE 1 #define LINUX_RLIMIT_DATA 2 #define LINUX_RLIMIT_STACK 3 #define LINUX_RLIMIT_CORE 4 #define LINUX_RLIMIT_RSS 5 #define LINUX_RLIMIT_NPROC 6 #define LINUX_RLIMIT_NOFILE 7 #define LINUX_RLIMIT_MEMLOCK 8 #define LINUX_RLIMIT_AS 9 /* Address space limit */ #define LINUX_RLIM_NLIMITS 10 struct l_rlimit { l_ulong rlim_cur; l_ulong rlim_max; }; /* mmap options */ #define LINUX_MAP_SHARED 0x0001 #define LINUX_MAP_PRIVATE 0x0002 #define LINUX_MAP_FIXED 0x0010 #define LINUX_MAP_ANON 0x0020 #define LINUX_MAP_GROWSDOWN 0x0100 struct l_mmap_argv { l_uintptr_t addr; l_size_t len; l_int prot; l_int flags; l_int fd; l_off_t pgoff; } __packed; /* * stat family of syscalls */ struct l_timespec { l_time_t tv_sec; l_long tv_nsec; }; struct l_newstat { l_ushort st_dev; l_ushort __pad1; l_ulong st_ino; l_ushort st_mode; l_ushort st_nlink; l_ushort st_uid; l_ushort st_gid; l_ushort st_rdev; l_ushort __pad2; l_ulong st_size; l_ulong st_blksize; l_ulong st_blocks; struct l_timespec st_atim; struct l_timespec st_mtim; struct l_timespec st_ctim; l_ulong __unused4; l_ulong __unused5; }; struct l_stat { l_ushort st_dev; l_ulong st_ino; l_ushort st_mode; l_ushort st_nlink; l_ushort st_uid; l_ushort st_gid; l_ushort st_rdev; l_long st_size; struct l_timespec st_atim; struct l_timespec st_mtim; struct l_timespec st_ctim; l_long st_blksize; l_long st_blocks; l_ulong st_flags; l_ulong st_gen; }; struct l_stat64 { l_ushort st_dev; u_char __pad0[10]; l_ulong __st_ino; l_uint st_mode; l_uint st_nlink; l_ulong st_uid; l_ulong st_gid; l_ushort st_rdev; u_char __pad3[10]; l_longlong st_size; l_ulong st_blksize; l_ulong st_blocks; l_ulong __pad4; struct l_timespec st_atim; struct l_timespec st_mtim; struct l_timespec st_ctim; l_ulonglong st_ino; }; struct l_statfs64 { l_int f_type; l_int f_bsize; uint64_t f_blocks; uint64_t f_bfree; uint64_t f_bavail; uint64_t f_files; uint64_t f_ffree; l_fsid_t f_fsid; l_int f_namelen; l_int f_spare[6]; }; /* * Signalling */ #define LINUX_SIGHUP 1 #define LINUX_SIGINT 2 #define LINUX_SIGQUIT 3 #define LINUX_SIGILL 4 #define LINUX_SIGTRAP 5 #define LINUX_SIGABRT 6 #define LINUX_SIGIOT LINUX_SIGABRT #define LINUX_SIGBUS 7 #define LINUX_SIGFPE 8 #define LINUX_SIGKILL 9 #define LINUX_SIGUSR1 10 #define LINUX_SIGSEGV 11 #define LINUX_SIGUSR2 12 #define LINUX_SIGPIPE 13 #define LINUX_SIGALRM 14 #define LINUX_SIGTERM 15 #define LINUX_SIGSTKFLT 16 #define LINUX_SIGCHLD 17 #define LINUX_SIGCONT 18 #define LINUX_SIGSTOP 19 #define LINUX_SIGTSTP 20 #define LINUX_SIGTTIN 21 #define LINUX_SIGTTOU 22 #define LINUX_SIGURG 23 #define LINUX_SIGXCPU 24 #define LINUX_SIGXFSZ 25 #define LINUX_SIGVTALRM 26 #define LINUX_SIGPROF 27 #define LINUX_SIGWINCH 28 #define LINUX_SIGIO 29 #define LINUX_SIGPOLL LINUX_SIGIO #define LINUX_SIGPWR 30 #define LINUX_SIGSYS 31 #define LINUX_SIGRTMIN 32 #define LINUX_SIGTBLSZ 31 #define LINUX_NSIG_WORDS 2 #define LINUX_NBPW 32 #define LINUX_NSIG (LINUX_NBPW * LINUX_NSIG_WORDS) /* sigaction flags */ #define LINUX_SA_NOCLDSTOP 0x00000001 #define LINUX_SA_NOCLDWAIT 0x00000002 #define LINUX_SA_SIGINFO 0x00000004 #define LINUX_SA_RESTORER 0x04000000 #define LINUX_SA_ONSTACK 0x08000000 #define LINUX_SA_RESTART 0x10000000 #define LINUX_SA_INTERRUPT 0x20000000 #define LINUX_SA_NOMASK 0x40000000 #define LINUX_SA_ONESHOT 0x80000000 /* sigprocmask actions */ #define LINUX_SIG_BLOCK 0 #define LINUX_SIG_UNBLOCK 1 #define LINUX_SIG_SETMASK 2 /* sigset_t macros */ #define LINUX_SIGEMPTYSET(set) (set).__bits[0] = (set).__bits[1] = 0 #define LINUX_SIGISMEMBER(set, sig) SIGISMEMBER(set, sig) #define LINUX_SIGADDSET(set, sig) SIGADDSET(set, sig) /* sigaltstack */ #define LINUX_MINSIGSTKSZ 2048 #define LINUX_SS_ONSTACK 1 #define LINUX_SS_DISABLE 2 int linux_to_bsd_sigaltstack(int lsa); int bsd_to_linux_sigaltstack(int bsa); typedef void (*l_handler_t)(l_int); typedef l_ulong l_osigset_t; typedef struct { l_uint __bits[LINUX_NSIG_WORDS]; } l_sigset_t; typedef struct { l_handler_t lsa_handler; l_osigset_t lsa_mask; l_ulong lsa_flags; void (*lsa_restorer)(void); } l_osigaction_t; typedef struct { l_handler_t lsa_handler; l_ulong lsa_flags; void (*lsa_restorer)(void); l_sigset_t lsa_mask; } l_sigaction_t; typedef struct { void *ss_sp; l_int ss_flags; l_size_t ss_size; } l_stack_t; /* The Linux sigcontext, pretty much a standard 386 trapframe. */ struct l_sigcontext { l_int sc_gs; l_int sc_fs; l_int sc_es; l_int sc_ds; l_int sc_edi; l_int sc_esi; l_int sc_ebp; l_int sc_esp; l_int sc_ebx; l_int sc_edx; l_int sc_ecx; l_int sc_eax; l_int sc_trapno; l_int sc_err; l_int sc_eip; l_int sc_cs; l_int sc_eflags; l_int sc_esp_at_signal; l_int sc_ss; l_int sc_387; l_int sc_mask; l_int sc_cr2; }; struct l_ucontext { l_ulong uc_flags; void *uc_link; l_stack_t uc_stack; struct l_sigcontext uc_mcontext; l_sigset_t uc_sigmask; }; #define LINUX_SI_MAX_SIZE 128 #define LINUX_SI_PAD_SIZE ((LINUX_SI_MAX_SIZE/sizeof(l_int)) - 3) typedef union l_sigval { l_int sival_int; l_uintptr_t sival_ptr; } l_sigval_t; typedef struct l_siginfo { l_int lsi_signo; l_int lsi_errno; l_int lsi_code; union { l_int _pad[LINUX_SI_PAD_SIZE]; struct { l_pid_t _pid; l_uid_t _uid; } _kill; struct { l_timer_t _tid; l_int _overrun; char _pad[sizeof(l_uid_t) - sizeof(l_int)]; l_sigval_t _sigval; l_int _sys_private; } _timer; struct { l_pid_t _pid; /* sender's pid */ l_uid_t _uid; /* sender's uid */ l_sigval_t _sigval; } _rt; struct { l_pid_t _pid; /* which child */ l_uid_t _uid; /* sender's uid */ l_int _status; /* exit code */ l_clock_t _utime; l_clock_t _stime; } _sigchld; struct { l_uintptr_t _addr; /* Faulting insn/memory ref. */ } _sigfault; struct { l_long _band; /* POLL_IN,POLL_OUT,POLL_MSG */ l_int _fd; } _sigpoll; } _sifields; } l_siginfo_t; #define lsi_pid _sifields._kill._pid #define lsi_uid _sifields._kill._uid #define lsi_tid _sifields._timer._tid #define lsi_overrun _sifields._timer._overrun #define lsi_sys_private _sifields._timer._sys_private #define lsi_status _sifields._sigchld._status #define lsi_utime _sifields._sigchld._utime #define lsi_stime _sifields._sigchld._stime #define lsi_value _sifields._rt._sigval #define lsi_int _sifields._rt._sigval.sival_int #define lsi_ptr _sifields._rt._sigval.sival_ptr #define lsi_addr _sifields._sigfault._addr #define lsi_band _sifields._sigpoll._band #define lsi_fd _sifields._sigpoll._fd struct l_fpreg { u_int16_t significand[4]; u_int16_t exponent; }; struct l_fpxreg { u_int16_t significand[4]; u_int16_t exponent; u_int16_t padding[3]; }; struct l_xmmreg { u_int32_t element[4]; }; struct l_fpstate { /* Regular FPU environment */ u_int32_t cw; u_int32_t sw; u_int32_t tag; u_int32_t ipoff; u_int32_t cssel; u_int32_t dataoff; u_int32_t datasel; struct l_fpreg _st[8]; u_int16_t status; u_int16_t magic; /* 0xffff = regular FPU data */ /* FXSR FPU environment */ u_int32_t _fxsr_env[6]; /* env is ignored. */ u_int32_t mxcsr; u_int32_t reserved; struct l_fpxreg _fxsr_st[8]; /* reg data is ignored. */ struct l_xmmreg _xmm[8]; u_int32_t padding[56]; }; /* * We make the stack look like Linux expects it when calling a signal * handler, but use the BSD way of calling the handler and sigreturn(). * This means that we need to pass the pointer to the handler too. * It is appended to the frame to not interfere with the rest of it. */ struct l_sigframe { l_int sf_sig; struct l_sigcontext sf_sc; struct l_fpstate sf_fpstate; l_uint sf_extramask[LINUX_NSIG_WORDS-1]; l_handler_t sf_handler; }; struct l_rt_sigframe { l_int sf_sig; l_siginfo_t *sf_siginfo; struct l_ucontext *sf_ucontext; l_siginfo_t sf_si; struct l_ucontext sf_sc; l_handler_t sf_handler; }; extern struct sysentvec linux_sysvec; extern struct sysentvec elf_linux_sysvec; /* * open/fcntl flags */ #define LINUX_O_RDONLY 00000000 #define LINUX_O_WRONLY 00000001 #define LINUX_O_RDWR 00000002 #define LINUX_O_ACCMODE 00000003 #define LINUX_O_CREAT 00000100 #define LINUX_O_EXCL 00000200 #define LINUX_O_NOCTTY 00000400 #define LINUX_O_TRUNC 00001000 #define LINUX_O_APPEND 00002000 #define LINUX_O_NONBLOCK 00004000 #define LINUX_O_NDELAY LINUX_O_NONBLOCK #define LINUX_O_SYNC 00010000 #define LINUX_FASYNC 00020000 #define LINUX_O_DIRECT 00040000 /* Direct disk access hint */ #define LINUX_O_LARGEFILE 00100000 #define LINUX_O_DIRECTORY 00200000 /* Must be a directory */ #define LINUX_O_NOFOLLOW 00400000 /* Do not follow links */ #define LINUX_O_NOATIME 01000000 #define LINUX_O_CLOEXEC 02000000 #define LINUX_F_DUPFD 0 #define LINUX_F_GETFD 1 #define LINUX_F_SETFD 2 #define LINUX_F_GETFL 3 #define LINUX_F_SETFL 4 #define LINUX_F_GETLK 5 #define LINUX_F_SETLK 6 #define LINUX_F_SETLKW 7 #define LINUX_F_SETOWN 8 #define LINUX_F_GETOWN 9 #define LINUX_F_GETLK64 12 #define LINUX_F_SETLK64 13 #define LINUX_F_SETLKW64 14 #define LINUX_F_RDLCK 0 #define LINUX_F_WRLCK 1 #define LINUX_F_UNLCK 2 union l_semun { l_int val; struct l_semid_ds *buf; l_ushort *array; struct l_seminfo *__buf; void *__pad; }; /* * Socket defines */ #define LINUX_SOL_SOCKET 1 #define LINUX_SOL_IP 0 #define LINUX_SOL_IPX 256 #define LINUX_SOL_AX25 257 #define LINUX_SOL_TCP 6 #define LINUX_SOL_UDP 17 #define LINUX_SO_DEBUG 1 #define LINUX_SO_REUSEADDR 2 #define LINUX_SO_TYPE 3 #define LINUX_SO_ERROR 4 #define LINUX_SO_DONTROUTE 5 #define LINUX_SO_BROADCAST 6 #define LINUX_SO_SNDBUF 7 #define LINUX_SO_RCVBUF 8 #define LINUX_SO_KEEPALIVE 9 #define LINUX_SO_OOBINLINE 10 #define LINUX_SO_NO_CHECK 11 #define LINUX_SO_PRIORITY 12 #define LINUX_SO_LINGER 13 #define LINUX_SO_PEERCRED 17 #define LINUX_SO_RCVLOWAT 18 #define LINUX_SO_SNDLOWAT 19 #define LINUX_SO_RCVTIMEO 20 #define LINUX_SO_SNDTIMEO 21 #define LINUX_SO_TIMESTAMP 29 #define LINUX_SO_ACCEPTCONN 30 struct l_sockaddr { l_ushort sa_family; char sa_data[14]; }; struct l_msghdr { l_uintptr_t msg_name; l_int msg_namelen; l_uintptr_t msg_iov; l_size_t msg_iovlen; l_uintptr_t msg_control; l_size_t msg_controllen; l_uint msg_flags; }; struct l_cmsghdr { l_size_t cmsg_len; l_int cmsg_level; l_int cmsg_type; }; struct l_ifmap { l_ulong mem_start; l_ulong mem_end; l_ushort base_addr; u_char irq; u_char dma; u_char port; }; #define LINUX_IFHWADDRLEN 6 #define LINUX_IFNAMSIZ 16 struct l_ifreq { union { char ifrn_name[LINUX_IFNAMSIZ]; } ifr_ifrn; union { struct l_sockaddr ifru_addr; struct l_sockaddr ifru_dstaddr; struct l_sockaddr ifru_broadaddr; struct l_sockaddr ifru_netmask; struct l_sockaddr ifru_hwaddr; l_short ifru_flags[1]; l_int ifru_metric; l_int ifru_mtu; struct l_ifmap ifru_map; char ifru_slave[LINUX_IFNAMSIZ]; l_caddr_t ifru_data; } ifr_ifru; }; #define ifr_name ifr_ifrn.ifrn_name /* Interface name */ #define ifr_hwaddr ifr_ifru.ifru_hwaddr /* MAC address */ /* * poll() */ #define LINUX_POLLIN 0x0001 #define LINUX_POLLPRI 0x0002 #define LINUX_POLLOUT 0x0004 #define LINUX_POLLERR 0x0008 #define LINUX_POLLHUP 0x0010 #define LINUX_POLLNVAL 0x0020 #define LINUX_POLLRDNORM 0x0040 #define LINUX_POLLRDBAND 0x0080 #define LINUX_POLLWRNORM 0x0100 #define LINUX_POLLWRBAND 0x0200 #define LINUX_POLLMSG 0x0400 struct l_pollfd { l_int fd; l_short events; l_short revents; }; struct l_user_desc { l_uint entry_number; l_uint base_addr; l_uint limit; l_uint seg_32bit:1; l_uint contents:2; l_uint read_exec_only:1; l_uint limit_in_pages:1; l_uint seg_not_present:1; l_uint useable:1; }; struct l_desc_struct { unsigned long a, b; }; #define LINUX_LOWERWORD 0x0000ffff /* * Macros which does the same thing as those in Linux include/asm-um/ldt-i386.h. * These convert Linux user space descriptor to machine one. */ #define LINUX_LDT_entry_a(info) \ ((((info)->base_addr & LINUX_LOWERWORD) << 16) | \ ((info)->limit & LINUX_LOWERWORD)) #define LINUX_ENTRY_B_READ_EXEC_ONLY 9 #define LINUX_ENTRY_B_CONTENTS 10 #define LINUX_ENTRY_B_SEG_NOT_PRESENT 15 #define LINUX_ENTRY_B_BASE_ADDR 16 #define LINUX_ENTRY_B_USEABLE 20 #define LINUX_ENTRY_B_SEG32BIT 22 #define LINUX_ENTRY_B_LIMIT 23 #define LINUX_LDT_entry_b(info) \ (((info)->base_addr & 0xff000000) | \ ((info)->limit & 0xf0000) | \ ((info)->contents << LINUX_ENTRY_B_CONTENTS) | \ (((info)->seg_not_present == 0) << LINUX_ENTRY_B_SEG_NOT_PRESENT) | \ (((info)->base_addr & 0x00ff0000) >> LINUX_ENTRY_B_BASE_ADDR) | \ (((info)->read_exec_only == 0) << LINUX_ENTRY_B_READ_EXEC_ONLY) | \ ((info)->seg_32bit << LINUX_ENTRY_B_SEG32BIT) | \ ((info)->useable << LINUX_ENTRY_B_USEABLE) | \ ((info)->limit_in_pages << LINUX_ENTRY_B_LIMIT) | 0x7000) #define LINUX_LDT_empty(info) \ ((info)->base_addr == 0 && \ (info)->limit == 0 && \ (info)->contents == 0 && \ (info)->seg_not_present == 1 && \ (info)->read_exec_only == 1 && \ (info)->seg_32bit == 0 && \ (info)->limit_in_pages == 0 && \ (info)->useable == 0) /* * Macros for converting segments. * They do the same as those in arch/i386/kernel/process.c in Linux. */ #define LINUX_GET_BASE(desc) \ ((((desc)->a >> 16) & LINUX_LOWERWORD) | \ (((desc)->b << 16) & 0x00ff0000) | \ ((desc)->b & 0xff000000)) #define LINUX_GET_LIMIT(desc) \ (((desc)->a & LINUX_LOWERWORD) | \ ((desc)->b & 0xf0000)) #define LINUX_GET_32BIT(desc) \ (((desc)->b >> LINUX_ENTRY_B_SEG32BIT) & 1) #define LINUX_GET_CONTENTS(desc) \ (((desc)->b >> LINUX_ENTRY_B_CONTENTS) & 3) #define LINUX_GET_WRITABLE(desc) \ (((desc)->b >> LINUX_ENTRY_B_READ_EXEC_ONLY) & 1) #define LINUX_GET_LIMIT_PAGES(desc) \ (((desc)->b >> LINUX_ENTRY_B_LIMIT) & 1) #define LINUX_GET_PRESENT(desc) \ (((desc)->b >> LINUX_ENTRY_B_SEG_NOT_PRESENT) & 1) #define LINUX_GET_USEABLE(desc) \ (((desc)->b >> LINUX_ENTRY_B_USEABLE) & 1) +#define linux_copyout_rusage(r, u) copyout(r, u, sizeof(*r)) + /* robust futexes */ struct linux_robust_list { struct linux_robust_list *next; }; struct linux_robust_list_head { struct linux_robust_list list; l_long futex_offset; struct linux_robust_list *pending_list; }; #endif /* !_I386_LINUX_H_ */ Index: head/sys/i386/linux/linux_machdep.c =================================================================== --- head/sys/i386/linux/linux_machdep.c (revision 283390) +++ head/sys/i386/linux/linux_machdep.c (revision 283391) @@ -1,1080 +1,1049 @@ /*- * Copyright (c) 2000 Marcel Moolenaar * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer * in this position and unchanged. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. The name of the author may not be used to endorse or promote products * derived from this software without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include /* needed for pcb definition in linux_set_thread_area */ #include "opt_posix.h" extern struct sysentvec elf32_freebsd_sysvec; /* defined in i386/i386/elf_machdep.c */ struct l_descriptor { l_uint entry_number; l_ulong base_addr; l_uint limit; l_uint seg_32bit:1; l_uint contents:2; l_uint read_exec_only:1; l_uint limit_in_pages:1; l_uint seg_not_present:1; l_uint useable:1; }; struct l_old_select_argv { l_int nfds; l_fd_set *readfds; l_fd_set *writefds; l_fd_set *exceptfds; struct l_timeval *timeout; }; static int linux_mmap_common(struct thread *td, l_uintptr_t addr, l_size_t len, l_int prot, l_int flags, l_int fd, l_loff_t pos); int linux_to_bsd_sigaltstack(int lsa) { int bsa = 0; if (lsa & LINUX_SS_DISABLE) bsa |= SS_DISABLE; if (lsa & LINUX_SS_ONSTACK) bsa |= SS_ONSTACK; return (bsa); } int bsd_to_linux_sigaltstack(int bsa) { int lsa = 0; if (bsa & SS_DISABLE) lsa |= LINUX_SS_DISABLE; if (bsa & SS_ONSTACK) lsa |= LINUX_SS_ONSTACK; return (lsa); } int linux_execve(struct thread *td, struct linux_execve_args *args) { struct image_args eargs; struct vmspace *oldvmspace; char *newpath; int error; LCONVPATHEXIST(td, args->path, &newpath); #ifdef DEBUG if (ldebug(execve)) printf(ARGS(execve, "%s"), newpath); #endif error = pre_execve(td, &oldvmspace); if (error != 0) { free(newpath, M_TEMP); return (error); } error = exec_copyin_args(&eargs, newpath, UIO_SYSSPACE, args->argp, args->envp); free(newpath, M_TEMP); if (error == 0) error = kern_execve(td, &eargs, NULL); if (error == 0) error = linux_common_execve(td, &eargs); post_execve(td, error, oldvmspace); return (error); } struct l_ipc_kludge { struct l_msgbuf *msgp; l_long msgtyp; }; int linux_ipc(struct thread *td, struct linux_ipc_args *args) { switch (args->what & 0xFFFF) { case LINUX_SEMOP: { struct linux_semop_args a; a.semid = args->arg1; a.tsops = args->ptr; a.nsops = args->arg2; return (linux_semop(td, &a)); } case LINUX_SEMGET: { struct linux_semget_args a; a.key = args->arg1; a.nsems = args->arg2; a.semflg = args->arg3; return (linux_semget(td, &a)); } case LINUX_SEMCTL: { struct linux_semctl_args a; int error; a.semid = args->arg1; a.semnum = args->arg2; a.cmd = args->arg3; error = copyin(args->ptr, &a.arg, sizeof(a.arg)); if (error) return (error); return (linux_semctl(td, &a)); } case LINUX_MSGSND: { struct linux_msgsnd_args a; a.msqid = args->arg1; a.msgp = args->ptr; a.msgsz = args->arg2; a.msgflg = args->arg3; return (linux_msgsnd(td, &a)); } case LINUX_MSGRCV: { struct linux_msgrcv_args a; a.msqid = args->arg1; a.msgsz = args->arg2; a.msgflg = args->arg3; if ((args->what >> 16) == 0) { struct l_ipc_kludge tmp; int error; if (args->ptr == NULL) return (EINVAL); error = copyin(args->ptr, &tmp, sizeof(tmp)); if (error) return (error); a.msgp = tmp.msgp; a.msgtyp = tmp.msgtyp; } else { a.msgp = args->ptr; a.msgtyp = args->arg5; } return (linux_msgrcv(td, &a)); } case LINUX_MSGGET: { struct linux_msgget_args a; a.key = args->arg1; a.msgflg = args->arg2; return (linux_msgget(td, &a)); } case LINUX_MSGCTL: { struct linux_msgctl_args a; a.msqid = args->arg1; a.cmd = args->arg2; a.buf = args->ptr; return (linux_msgctl(td, &a)); } case LINUX_SHMAT: { struct linux_shmat_args a; a.shmid = args->arg1; a.shmaddr = args->ptr; a.shmflg = args->arg2; a.raddr = (l_ulong *)args->arg3; return (linux_shmat(td, &a)); } case LINUX_SHMDT: { struct linux_shmdt_args a; a.shmaddr = args->ptr; return (linux_shmdt(td, &a)); } case LINUX_SHMGET: { struct linux_shmget_args a; a.key = args->arg1; a.size = args->arg2; a.shmflg = args->arg3; return (linux_shmget(td, &a)); } case LINUX_SHMCTL: { struct linux_shmctl_args a; a.shmid = args->arg1; a.cmd = args->arg2; a.buf = args->ptr; return (linux_shmctl(td, &a)); } default: break; } return (EINVAL); } int linux_old_select(struct thread *td, struct linux_old_select_args *args) { struct l_old_select_argv linux_args; struct linux_select_args newsel; int error; #ifdef DEBUG if (ldebug(old_select)) printf(ARGS(old_select, "%p"), args->ptr); #endif error = copyin(args->ptr, &linux_args, sizeof(linux_args)); if (error) return (error); newsel.nfds = linux_args.nfds; newsel.readfds = linux_args.readfds; newsel.writefds = linux_args.writefds; newsel.exceptfds = linux_args.exceptfds; newsel.timeout = linux_args.timeout; return (linux_select(td, &newsel)); } int linux_set_cloned_tls(struct thread *td, void *desc) { struct segment_descriptor sd; struct l_user_desc info; int idx, error; int a[2]; error = copyin(desc, &info, sizeof(struct l_user_desc)); if (error) { printf(LMSG("copyin failed!")); } else { idx = info.entry_number; /* * looks like we're getting the idx we returned * in the set_thread_area() syscall */ if (idx != 6 && idx != 3) { printf(LMSG("resetting idx!")); idx = 3; } /* this doesnt happen in practice */ if (idx == 6) { /* we might copy out the entry_number as 3 */ info.entry_number = 3; error = copyout(&info, desc, sizeof(struct l_user_desc)); if (error) printf(LMSG("copyout failed!")); } a[0] = LINUX_LDT_entry_a(&info); a[1] = LINUX_LDT_entry_b(&info); memcpy(&sd, &a, sizeof(a)); #ifdef DEBUG if (ldebug(clone)) printf("Segment created in clone with " "CLONE_SETTLS: lobase: %x, hibase: %x, " "lolimit: %x, hilimit: %x, type: %i, " "dpl: %i, p: %i, xx: %i, def32: %i, " "gran: %i\n", sd.sd_lobase, sd.sd_hibase, sd.sd_lolimit, sd.sd_hilimit, sd.sd_type, sd.sd_dpl, sd.sd_p, sd.sd_xx, sd.sd_def32, sd.sd_gran); #endif /* set %gs */ td->td_pcb->pcb_gsd = sd; td->td_pcb->pcb_gs = GSEL(GUGS_SEL, SEL_UPL); } return (error); } int linux_set_upcall_kse(struct thread *td, register_t stack) { if (stack) td->td_frame->tf_esp = stack; /* * The newly created Linux thread returns * to the user space by the same path that a parent do. */ td->td_frame->tf_eax = 0; return (0); } #define STACK_SIZE (2 * 1024 * 1024) #define GUARD_SIZE (4 * PAGE_SIZE) int linux_mmap2(struct thread *td, struct linux_mmap2_args *args) { #ifdef DEBUG if (ldebug(mmap2)) printf(ARGS(mmap2, "%p, %d, %d, 0x%08x, %d, %d"), (void *)args->addr, args->len, args->prot, args->flags, args->fd, args->pgoff); #endif return (linux_mmap_common(td, args->addr, args->len, args->prot, args->flags, args->fd, (uint64_t)(uint32_t)args->pgoff * PAGE_SIZE)); } int linux_mmap(struct thread *td, struct linux_mmap_args *args) { int error; struct l_mmap_argv linux_args; error = copyin(args->ptr, &linux_args, sizeof(linux_args)); if (error) return (error); #ifdef DEBUG if (ldebug(mmap)) printf(ARGS(mmap, "%p, %d, %d, 0x%08x, %d, %d"), (void *)linux_args.addr, linux_args.len, linux_args.prot, linux_args.flags, linux_args.fd, linux_args.pgoff); #endif return (linux_mmap_common(td, linux_args.addr, linux_args.len, linux_args.prot, linux_args.flags, linux_args.fd, (uint32_t)linux_args.pgoff)); } static int linux_mmap_common(struct thread *td, l_uintptr_t addr, l_size_t len, l_int prot, l_int flags, l_int fd, l_loff_t pos) { struct proc *p = td->td_proc; struct mmap_args /* { caddr_t addr; size_t len; int prot; int flags; int fd; long pad; off_t pos; } */ bsd_args; int error; struct file *fp; cap_rights_t rights; error = 0; bsd_args.flags = 0; fp = NULL; /* * Linux mmap(2): * You must specify exactly one of MAP_SHARED and MAP_PRIVATE */ if (!((flags & LINUX_MAP_SHARED) ^ (flags & LINUX_MAP_PRIVATE))) return (EINVAL); if (flags & LINUX_MAP_SHARED) bsd_args.flags |= MAP_SHARED; if (flags & LINUX_MAP_PRIVATE) bsd_args.flags |= MAP_PRIVATE; if (flags & LINUX_MAP_FIXED) bsd_args.flags |= MAP_FIXED; if (flags & LINUX_MAP_ANON) { /* Enforce pos to be on page boundary, then ignore. */ if ((pos & PAGE_MASK) != 0) return (EINVAL); pos = 0; bsd_args.flags |= MAP_ANON; } else bsd_args.flags |= MAP_NOSYNC; if (flags & LINUX_MAP_GROWSDOWN) bsd_args.flags |= MAP_STACK; /* * PROT_READ, PROT_WRITE, or PROT_EXEC implies PROT_READ and PROT_EXEC * on Linux/i386. We do this to ensure maximum compatibility. * Linux/ia64 does the same in i386 emulation mode. */ bsd_args.prot = prot; if (bsd_args.prot & (PROT_READ | PROT_WRITE | PROT_EXEC)) bsd_args.prot |= PROT_READ | PROT_EXEC; /* Linux does not check file descriptor when MAP_ANONYMOUS is set. */ bsd_args.fd = (bsd_args.flags & MAP_ANON) ? -1 : fd; if (bsd_args.fd != -1) { /* * Linux follows Solaris mmap(2) description: * The file descriptor fildes is opened with * read permission, regardless of the * protection options specified. * * Checking just CAP_MMAP is fine here, since the real work * is done in the FreeBSD mmap(). */ error = fget(td, bsd_args.fd, cap_rights_init(&rights, CAP_MMAP), &fp); if (error != 0) return (error); if (fp->f_type != DTYPE_VNODE) { fdrop(fp, td); return (EINVAL); } /* Linux mmap() just fails for O_WRONLY files */ if (!(fp->f_flag & FREAD)) { fdrop(fp, td); return (EACCES); } fdrop(fp, td); } if (flags & LINUX_MAP_GROWSDOWN) { /* * The Linux MAP_GROWSDOWN option does not limit auto * growth of the region. Linux mmap with this option * takes as addr the inital BOS, and as len, the initial * region size. It can then grow down from addr without * limit. However, linux threads has an implicit internal * limit to stack size of STACK_SIZE. Its just not * enforced explicitly in linux. But, here we impose * a limit of (STACK_SIZE - GUARD_SIZE) on the stack * region, since we can do this with our mmap. * * Our mmap with MAP_STACK takes addr as the maximum * downsize limit on BOS, and as len the max size of * the region. It them maps the top SGROWSIZ bytes, * and auto grows the region down, up to the limit * in addr. * * If we don't use the MAP_STACK option, the effect * of this code is to allocate a stack region of a * fixed size of (STACK_SIZE - GUARD_SIZE). */ if ((caddr_t)PTRIN(addr) + len > p->p_vmspace->vm_maxsaddr) { /* * Some linux apps will attempt to mmap * thread stacks near the top of their * address space. If their TOS is greater * than vm_maxsaddr, vm_map_growstack() * will confuse the thread stack with the * process stack and deliver a SEGV if they * attempt to grow the thread stack past their * current stacksize rlimit. To avoid this, * adjust vm_maxsaddr upwards to reflect * the current stacksize rlimit rather * than the maximum possible stacksize. * It would be better to adjust the * mmap'ed region, but some apps do not check * mmap's return value. */ PROC_LOCK(p); p->p_vmspace->vm_maxsaddr = (char *)USRSTACK - lim_cur(p, RLIMIT_STACK); PROC_UNLOCK(p); } /* * This gives us our maximum stack size and a new BOS. * If we're using VM_STACK, then mmap will just map * the top SGROWSIZ bytes, and let the stack grow down * to the limit at BOS. If we're not using VM_STACK * we map the full stack, since we don't have a way * to autogrow it. */ if (len > STACK_SIZE - GUARD_SIZE) { bsd_args.addr = (caddr_t)PTRIN(addr); bsd_args.len = len; } else { bsd_args.addr = (caddr_t)PTRIN(addr) - (STACK_SIZE - GUARD_SIZE - len); bsd_args.len = STACK_SIZE - GUARD_SIZE; } } else { bsd_args.addr = (caddr_t)PTRIN(addr); bsd_args.len = len; } bsd_args.pos = pos; #ifdef DEBUG if (ldebug(mmap)) printf("-> %s(%p, %d, %d, 0x%08x, %d, 0x%x)\n", __func__, (void *)bsd_args.addr, bsd_args.len, bsd_args.prot, bsd_args.flags, bsd_args.fd, (int)bsd_args.pos); #endif error = sys_mmap(td, &bsd_args); #ifdef DEBUG if (ldebug(mmap)) printf("-> %s() return: 0x%x (0x%08x)\n", __func__, error, (u_int)td->td_retval[0]); #endif return (error); } int linux_mprotect(struct thread *td, struct linux_mprotect_args *uap) { struct mprotect_args bsd_args; bsd_args.addr = uap->addr; bsd_args.len = uap->len; bsd_args.prot = uap->prot; if (bsd_args.prot & (PROT_READ | PROT_WRITE | PROT_EXEC)) bsd_args.prot |= PROT_READ | PROT_EXEC; return (sys_mprotect(td, &bsd_args)); } int linux_ioperm(struct thread *td, struct linux_ioperm_args *args) { int error; struct i386_ioperm_args iia; iia.start = args->start; iia.length = args->length; iia.enable = args->enable; error = i386_set_ioperm(td, &iia); return (error); } int linux_iopl(struct thread *td, struct linux_iopl_args *args) { int error; if (args->level < 0 || args->level > 3) return (EINVAL); if ((error = priv_check(td, PRIV_IO)) != 0) return (error); if ((error = securelevel_gt(td->td_ucred, 0)) != 0) return (error); td->td_frame->tf_eflags = (td->td_frame->tf_eflags & ~PSL_IOPL) | (args->level * (PSL_IOPL / 3)); return (0); } int linux_modify_ldt(struct thread *td, struct linux_modify_ldt_args *uap) { int error; struct i386_ldt_args ldt; struct l_descriptor ld; union descriptor desc; int size, written; switch (uap->func) { case 0x00: /* read_ldt */ ldt.start = 0; ldt.descs = uap->ptr; ldt.num = uap->bytecount / sizeof(union descriptor); error = i386_get_ldt(td, &ldt); td->td_retval[0] *= sizeof(union descriptor); break; case 0x02: /* read_default_ldt = 0 */ size = 5*sizeof(struct l_desc_struct); if (size > uap->bytecount) size = uap->bytecount; for (written = error = 0; written < size && error == 0; written++) error = subyte((char *)uap->ptr + written, 0); td->td_retval[0] = written; break; case 0x01: /* write_ldt */ case 0x11: /* write_ldt */ if (uap->bytecount != sizeof(ld)) return (EINVAL); error = copyin(uap->ptr, &ld, sizeof(ld)); if (error) return (error); ldt.start = ld.entry_number; ldt.descs = &desc; ldt.num = 1; desc.sd.sd_lolimit = (ld.limit & 0x0000ffff); desc.sd.sd_hilimit = (ld.limit & 0x000f0000) >> 16; desc.sd.sd_lobase = (ld.base_addr & 0x00ffffff); desc.sd.sd_hibase = (ld.base_addr & 0xff000000) >> 24; desc.sd.sd_type = SDT_MEMRO | ((ld.read_exec_only ^ 1) << 1) | (ld.contents << 2); desc.sd.sd_dpl = 3; desc.sd.sd_p = (ld.seg_not_present ^ 1); desc.sd.sd_xx = 0; desc.sd.sd_def32 = ld.seg_32bit; desc.sd.sd_gran = ld.limit_in_pages; error = i386_set_ldt(td, &ldt, &desc); break; default: error = ENOSYS; break; } if (error == EOPNOTSUPP) { printf("linux: modify_ldt needs kernel option USER_LDT\n"); error = ENOSYS; } return (error); } int linux_sigaction(struct thread *td, struct linux_sigaction_args *args) { l_osigaction_t osa; l_sigaction_t act, oact; int error; #ifdef DEBUG if (ldebug(sigaction)) printf(ARGS(sigaction, "%d, %p, %p"), args->sig, (void *)args->nsa, (void *)args->osa); #endif if (args->nsa != NULL) { error = copyin(args->nsa, &osa, sizeof(l_osigaction_t)); if (error) return (error); act.lsa_handler = osa.lsa_handler; act.lsa_flags = osa.lsa_flags; act.lsa_restorer = osa.lsa_restorer; LINUX_SIGEMPTYSET(act.lsa_mask); act.lsa_mask.__bits[0] = osa.lsa_mask; } error = linux_do_sigaction(td, args->sig, args->nsa ? &act : NULL, args->osa ? &oact : NULL); if (args->osa != NULL && !error) { osa.lsa_handler = oact.lsa_handler; osa.lsa_flags = oact.lsa_flags; osa.lsa_restorer = oact.lsa_restorer; osa.lsa_mask = oact.lsa_mask.__bits[0]; error = copyout(&osa, args->osa, sizeof(l_osigaction_t)); } return (error); } /* * Linux has two extra args, restart and oldmask. We dont use these, * but it seems that "restart" is actually a context pointer that * enables the signal to happen with a different register set. */ int linux_sigsuspend(struct thread *td, struct linux_sigsuspend_args *args) { sigset_t sigmask; l_sigset_t mask; #ifdef DEBUG if (ldebug(sigsuspend)) printf(ARGS(sigsuspend, "%08lx"), (unsigned long)args->mask); #endif LINUX_SIGEMPTYSET(mask); mask.__bits[0] = args->mask; linux_to_bsd_sigset(&mask, &sigmask); return (kern_sigsuspend(td, sigmask)); } int linux_rt_sigsuspend(struct thread *td, struct linux_rt_sigsuspend_args *uap) { l_sigset_t lmask; sigset_t sigmask; int error; #ifdef DEBUG if (ldebug(rt_sigsuspend)) printf(ARGS(rt_sigsuspend, "%p, %d"), (void *)uap->newset, uap->sigsetsize); #endif if (uap->sigsetsize != sizeof(l_sigset_t)) return (EINVAL); error = copyin(uap->newset, &lmask, sizeof(l_sigset_t)); if (error) return (error); linux_to_bsd_sigset(&lmask, &sigmask); return (kern_sigsuspend(td, sigmask)); } int linux_pause(struct thread *td, struct linux_pause_args *args) { struct proc *p = td->td_proc; sigset_t sigmask; #ifdef DEBUG if (ldebug(pause)) printf(ARGS(pause, "")); #endif PROC_LOCK(p); sigmask = td->td_sigmask; PROC_UNLOCK(p); return (kern_sigsuspend(td, sigmask)); } int linux_sigaltstack(struct thread *td, struct linux_sigaltstack_args *uap) { stack_t ss, oss; l_stack_t lss; int error; #ifdef DEBUG if (ldebug(sigaltstack)) printf(ARGS(sigaltstack, "%p, %p"), uap->uss, uap->uoss); #endif if (uap->uss != NULL) { error = copyin(uap->uss, &lss, sizeof(l_stack_t)); if (error) return (error); ss.ss_sp = lss.ss_sp; ss.ss_size = lss.ss_size; ss.ss_flags = linux_to_bsd_sigaltstack(lss.ss_flags); } error = kern_sigaltstack(td, (uap->uss != NULL) ? &ss : NULL, (uap->uoss != NULL) ? &oss : NULL); if (!error && uap->uoss != NULL) { lss.ss_sp = oss.ss_sp; lss.ss_size = oss.ss_size; lss.ss_flags = bsd_to_linux_sigaltstack(oss.ss_flags); error = copyout(&lss, uap->uoss, sizeof(l_stack_t)); } return (error); } int linux_ftruncate64(struct thread *td, struct linux_ftruncate64_args *args) { struct ftruncate_args sa; #ifdef DEBUG if (ldebug(ftruncate64)) printf(ARGS(ftruncate64, "%u, %jd"), args->fd, (intmax_t)args->length); #endif sa.fd = args->fd; sa.length = args->length; return sys_ftruncate(td, &sa); } int linux_set_thread_area(struct thread *td, struct linux_set_thread_area_args *args) { struct l_user_desc info; int error; int idx; int a[2]; struct segment_descriptor sd; error = copyin(args->desc, &info, sizeof(struct l_user_desc)); if (error) return (error); #ifdef DEBUG if (ldebug(set_thread_area)) printf(ARGS(set_thread_area, "%i, %x, %x, %i, %i, %i, %i, %i, %i\n"), info.entry_number, info.base_addr, info.limit, info.seg_32bit, info.contents, info.read_exec_only, info.limit_in_pages, info.seg_not_present, info.useable); #endif idx = info.entry_number; /* * Semantics of linux version: every thread in the system has array of * 3 tls descriptors. 1st is GLIBC TLS, 2nd is WINE, 3rd unknown. This * syscall loads one of the selected tls decriptors with a value and * also loads GDT descriptors 6, 7 and 8 with the content of the * per-thread descriptors. * * Semantics of fbsd version: I think we can ignore that linux has 3 * per-thread descriptors and use just the 1st one. The tls_array[] * is used only in set/get-thread_area() syscalls and for loading the * GDT descriptors. In fbsd we use just one GDT descriptor for TLS so * we will load just one. * * XXX: this doesn't work when a user space process tries to use more * than 1 TLS segment. Comment in the linux sources says wine might do * this. */ /* * we support just GLIBC TLS now * we should let 3 proceed as well because we use this segment so * if code does two subsequent calls it should succeed */ if (idx != 6 && idx != -1 && idx != 3) return (EINVAL); /* * we have to copy out the GDT entry we use * FreeBSD uses GDT entry #3 for storing %gs so load that * * XXX: what if a user space program doesn't check this value and tries * to use 6, 7 or 8? */ idx = info.entry_number = 3; error = copyout(&info, args->desc, sizeof(struct l_user_desc)); if (error) return (error); if (LINUX_LDT_empty(&info)) { a[0] = 0; a[1] = 0; } else { a[0] = LINUX_LDT_entry_a(&info); a[1] = LINUX_LDT_entry_b(&info); } memcpy(&sd, &a, sizeof(a)); #ifdef DEBUG if (ldebug(set_thread_area)) printf("Segment created in set_thread_area: lobase: %x, hibase: %x, lolimit: %x, hilimit: %x, type: %i, dpl: %i, p: %i, xx: %i, def32: %i, gran: %i\n", sd.sd_lobase, sd.sd_hibase, sd.sd_lolimit, sd.sd_hilimit, sd.sd_type, sd.sd_dpl, sd.sd_p, sd.sd_xx, sd.sd_def32, sd.sd_gran); #endif /* this is taken from i386 version of cpu_set_user_tls() */ critical_enter(); /* set %gs */ td->td_pcb->pcb_gsd = sd; PCPU_GET(fsgs_gdt)[1] = sd; load_gs(GSEL(GUGS_SEL, SEL_UPL)); critical_exit(); return (0); } int linux_get_thread_area(struct thread *td, struct linux_get_thread_area_args *args) { struct l_user_desc info; int error; int idx; struct l_desc_struct desc; struct segment_descriptor sd; #ifdef DEBUG if (ldebug(get_thread_area)) printf(ARGS(get_thread_area, "%p"), args->desc); #endif error = copyin(args->desc, &info, sizeof(struct l_user_desc)); if (error) return (error); idx = info.entry_number; /* XXX: I am not sure if we want 3 to be allowed too. */ if (idx != 6 && idx != 3) return (EINVAL); idx = 3; memset(&info, 0, sizeof(info)); sd = PCPU_GET(fsgs_gdt)[1]; memcpy(&desc, &sd, sizeof(desc)); info.entry_number = idx; info.base_addr = LINUX_GET_BASE(&desc); info.limit = LINUX_GET_LIMIT(&desc); info.seg_32bit = LINUX_GET_32BIT(&desc); info.contents = LINUX_GET_CONTENTS(&desc); info.read_exec_only = !LINUX_GET_WRITABLE(&desc); info.limit_in_pages = LINUX_GET_LIMIT_PAGES(&desc); info.seg_not_present = !LINUX_GET_PRESENT(&desc); info.useable = LINUX_GET_USEABLE(&desc); error = copyout(&info, args->desc, sizeof(struct l_user_desc)); if (error) return (EFAULT); return (0); } /* XXX: this wont work with module - convert it */ int linux_mq_open(struct thread *td, struct linux_mq_open_args *args) { #ifdef P1003_1B_MQUEUE return sys_kmq_open(td, (struct kmq_open_args *) args); #else return (ENOSYS); #endif } int linux_mq_unlink(struct thread *td, struct linux_mq_unlink_args *args) { #ifdef P1003_1B_MQUEUE return sys_kmq_unlink(td, (struct kmq_unlink_args *) args); #else return (ENOSYS); #endif } int linux_mq_timedsend(struct thread *td, struct linux_mq_timedsend_args *args) { #ifdef P1003_1B_MQUEUE return sys_kmq_timedsend(td, (struct kmq_timedsend_args *) args); #else return (ENOSYS); #endif } int linux_mq_timedreceive(struct thread *td, struct linux_mq_timedreceive_args *args) { #ifdef P1003_1B_MQUEUE return sys_kmq_timedreceive(td, (struct kmq_timedreceive_args *) args); #else return (ENOSYS); #endif } int linux_mq_notify(struct thread *td, struct linux_mq_notify_args *args) { #ifdef P1003_1B_MQUEUE return sys_kmq_notify(td, (struct kmq_notify_args *) args); #else return (ENOSYS); #endif } int linux_mq_getsetattr(struct thread *td, struct linux_mq_getsetattr_args *args) { #ifdef P1003_1B_MQUEUE return sys_kmq_setattr(td, (struct kmq_setattr_args *) args); #else return (ENOSYS); #endif } - -int -linux_wait4(struct thread *td, struct linux_wait4_args *args) -{ - int error, options; - struct rusage ru, *rup; - -#ifdef DEBUG - if (ldebug(wait4)) - printf(ARGS(wait4, "%d, %p, %d, %p"), - args->pid, (void *)args->status, args->options, - (void *)args->rusage); -#endif - - options = (args->options & (WNOHANG | WUNTRACED)); - /* WLINUXCLONE should be equal to __WCLONE, but we make sure */ - if (args->options & __WCLONE) - options |= WLINUXCLONE; - - if (args->rusage != NULL) - rup = &ru; - else - rup = NULL; - error = linux_common_wait(td, args->pid, args->status, options, rup); - if (error) - return (error); - if (args->rusage != NULL) - error = copyout(&ru, args->rusage, sizeof(ru)); - - return (error); -}