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


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

FreeBSD/Linux Kernel Cross Reference
sys/amd64/linux32/linux32_sysvec.c

Version: -  FREEBSD  -  FREEBSD-12-STABLE  -  FREEBSD-12-0  -  FREEBSD-11-STABLE  -  FREEBSD-11-2  -  FREEBSD-11-1  -  FREEBSD-11-0  -  FREEBSD-10-STABLE  -  FREEBSD-10-4  -  FREEBSD-10-3  -  FREEBSD-10-2  -  FREEBSD-10-1  -  FREEBSD-10-0  -  FREEBSD-9-STABLE  -  FREEBSD-9-3  -  FREEBSD-9-2  -  FREEBSD-9-1  -  FREEBSD-9-0  -  FREEBSD-8-STABLE  -  FREEBSD-8-4  -  FREEBSD-8-3  -  FREEBSD-8-2  -  FREEBSD-8-1  -  FREEBSD-8-0  -  FREEBSD-7-STABLE  -  FREEBSD-7-4  -  FREEBSD-7-3  -  FREEBSD-7-2  -  FREEBSD-7-1  -  FREEBSD-7-0  -  FREEBSD-6-STABLE  -  FREEBSD-6-4  -  FREEBSD-6-3  -  FREEBSD-6-2  -  FREEBSD-6-1  -  FREEBSD-6-0  -  FREEBSD-5-STABLE  -  FREEBSD-5-5  -  FREEBSD-5-4  -  FREEBSD-5-3  -  FREEBSD-5-2  -  FREEBSD-5-1  -  FREEBSD-5-0  -  FREEBSD-4-STABLE  -  FREEBSD-3-STABLE  -  FREEBSD22  -  linux-2.6  -  linux-2.4.22  -  MK83  -  MK84  -  PLAN9  -  DFBSD  -  NETBSD  -  NETBSD5  -  NETBSD4  -  NETBSD3  -  NETBSD20  -  OPENBSD  -  xnu-517  -  xnu-792  -  xnu-792.6.70  -  xnu-1228  -  xnu-1456.1.26  -  xnu-1699.24.8  -  xnu-2050.18.24  -  OPENSOLARIS  -  minix-3-1-1 
SearchContext: -  none  -  3  -  10 

    1 /*-
    2  * SPDX-License-Identifier: BSD-3-Clause
    3  *
    4  * Copyright (c) 2004 Tim J. Robbins
    5  * Copyright (c) 2003 Peter Wemm
    6  * Copyright (c) 2002 Doug Rabson
    7  * Copyright (c) 1998-1999 Andrew Gallatin
    8  * Copyright (c) 1994-1996 Søren Schmidt
    9  * All rights reserved.
   10  *
   11  * Redistribution and use in source and binary forms, with or without
   12  * modification, are permitted provided that the following conditions
   13  * are met:
   14  * 1. Redistributions of source code must retain the above copyright
   15  *    notice, this list of conditions and the following disclaimer
   16  *    in this position and unchanged.
   17  * 2. Redistributions in binary form must reproduce the above copyright
   18  *    notice, this list of conditions and the following disclaimer in the
   19  *    documentation and/or other materials provided with the distribution.
   20  * 3. The name of the author may not be used to endorse or promote products
   21  *    derived from this software without specific prior written permission
   22  *
   23  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
   24  * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
   25  * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
   26  * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
   27  * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
   28  * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
   29  * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
   30  * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
   31  * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
   32  * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
   33  */
   34 
   35 #include "opt_compat.h"
   36 
   37 #include <sys/cdefs.h>
   38 __FBSDID("$FreeBSD: releng/12.0/sys/amd64/linux32/linux32_sysvec.c 340258 2018-11-08 19:56:29Z tijl $");
   39 
   40 #ifndef COMPAT_FREEBSD32
   41 #error "Unable to compile Linux-emulator due to missing COMPAT_FREEBSD32 option!"
   42 #endif
   43 
   44 #define __ELF_WORD_SIZE 32
   45 
   46 #include <sys/param.h>
   47 #include <sys/systm.h>
   48 #include <sys/exec.h>
   49 #include <sys/fcntl.h>
   50 #include <sys/imgact.h>
   51 #include <sys/imgact_elf.h>
   52 #include <sys/kernel.h>
   53 #include <sys/lock.h>
   54 #include <sys/malloc.h>
   55 #include <sys/module.h>
   56 #include <sys/mutex.h>
   57 #include <sys/proc.h>
   58 #include <sys/resourcevar.h>
   59 #include <sys/signalvar.h>
   60 #include <sys/syscallsubr.h>
   61 #include <sys/sysctl.h>
   62 #include <sys/sysent.h>
   63 #include <sys/sysproto.h>
   64 #include <sys/vnode.h>
   65 #include <sys/eventhandler.h>
   66 
   67 #include <vm/vm.h>
   68 #include <vm/pmap.h>
   69 #include <vm/vm_extern.h>
   70 #include <vm/vm_map.h>
   71 #include <vm/vm_object.h>
   72 #include <vm/vm_page.h>
   73 #include <vm/vm_param.h>
   74 
   75 #include <machine/cpu.h>
   76 #include <machine/md_var.h>
   77 #include <machine/pcb.h>
   78 #include <machine/specialreg.h>
   79 #include <machine/trap.h>
   80 
   81 #include <amd64/linux32/linux.h>
   82 #include <amd64/linux32/linux32_proto.h>
   83 #include <compat/linux/linux_emul.h>
   84 #include <compat/linux/linux_futex.h>
   85 #include <compat/linux/linux_ioctl.h>
   86 #include <compat/linux/linux_mib.h>
   87 #include <compat/linux/linux_misc.h>
   88 #include <compat/linux/linux_signal.h>
   89 #include <compat/linux/linux_util.h>
   90 #include <compat/linux/linux_vdso.h>
   91 
   92 MODULE_VERSION(linux, 1);
   93 
   94 /*
   95  * Allow the sendsig functions to use the ldebug() facility even though they
   96  * are not syscalls themselves.  Map them to syscall 0.  This is slightly less
   97  * bogus than using ldebug(sigreturn).
   98  */
   99 #define LINUX32_SYS_linux_rt_sendsig    0
  100 #define LINUX32_SYS_linux_sendsig       0
  101 
  102 const char *linux_kplatform;
  103 static int linux_szsigcode;
  104 static vm_object_t linux_shared_page_obj;
  105 static char *linux_shared_page_mapping;
  106 extern char _binary_linux32_locore_o_start;
  107 extern char _binary_linux32_locore_o_end;
  108 
  109 extern struct sysent linux32_sysent[LINUX32_SYS_MAXSYSCALL];
  110 
  111 SET_DECLARE(linux_ioctl_handler_set, struct linux_ioctl_handler);
  112 
  113 static int      linux_fixup_elf(register_t **stack_base,
  114                     struct image_params *iparams);
  115 static register_t *linux_copyout_strings(struct image_params *imgp);
  116 static void     linux_sendsig(sig_t catcher, ksiginfo_t *ksi, sigset_t *mask);
  117 static void     linux_exec_setregs(struct thread *td,
  118                                    struct image_params *imgp, u_long stack);
  119 static void     linux32_fixlimit(struct rlimit *rl, int which);
  120 static bool     linux32_trans_osrel(const Elf_Note *note, int32_t *osrel);
  121 static void     linux_vdso_install(void *param);
  122 static void     linux_vdso_deinstall(void *param);
  123 
  124 #define LINUX_T_UNKNOWN  255
  125 static int _bsd_to_linux_trapcode[] = {
  126         LINUX_T_UNKNOWN,        /* 0 */
  127         6,                      /* 1  T_PRIVINFLT */
  128         LINUX_T_UNKNOWN,        /* 2 */
  129         3,                      /* 3  T_BPTFLT */
  130         LINUX_T_UNKNOWN,        /* 4 */
  131         LINUX_T_UNKNOWN,        /* 5 */
  132         16,                     /* 6  T_ARITHTRAP */
  133         254,                    /* 7  T_ASTFLT */
  134         LINUX_T_UNKNOWN,        /* 8 */
  135         13,                     /* 9  T_PROTFLT */
  136         1,                      /* 10 T_TRCTRAP */
  137         LINUX_T_UNKNOWN,        /* 11 */
  138         14,                     /* 12 T_PAGEFLT */
  139         LINUX_T_UNKNOWN,        /* 13 */
  140         17,                     /* 14 T_ALIGNFLT */
  141         LINUX_T_UNKNOWN,        /* 15 */
  142         LINUX_T_UNKNOWN,        /* 16 */
  143         LINUX_T_UNKNOWN,        /* 17 */
  144         0,                      /* 18 T_DIVIDE */
  145         2,                      /* 19 T_NMI */
  146         4,                      /* 20 T_OFLOW */
  147         5,                      /* 21 T_BOUND */
  148         7,                      /* 22 T_DNA */
  149         8,                      /* 23 T_DOUBLEFLT */
  150         9,                      /* 24 T_FPOPFLT */
  151         10,                     /* 25 T_TSSFLT */
  152         11,                     /* 26 T_SEGNPFLT */
  153         12,                     /* 27 T_STKFLT */
  154         18,                     /* 28 T_MCHK */
  155         19,                     /* 29 T_XMMFLT */
  156         15                      /* 30 T_RESERVED */
  157 };
  158 #define bsd_to_linux_trapcode(code) \
  159     ((code)<nitems(_bsd_to_linux_trapcode)? \
  160      _bsd_to_linux_trapcode[(code)]: \
  161      LINUX_T_UNKNOWN)
  162 
  163 struct linux32_ps_strings {
  164         u_int32_t ps_argvstr;   /* first of 0 or more argument strings */
  165         u_int ps_nargvstr;      /* the number of argument strings */
  166         u_int32_t ps_envstr;    /* first of 0 or more environment strings */
  167         u_int ps_nenvstr;       /* the number of environment strings */
  168 };
  169 
  170 LINUX_VDSO_SYM_INTPTR(linux32_sigcode);
  171 LINUX_VDSO_SYM_INTPTR(linux32_rt_sigcode);
  172 LINUX_VDSO_SYM_INTPTR(linux32_vsyscall);
  173 LINUX_VDSO_SYM_CHAR(linux_platform);
  174 
  175 /*
  176  * If FreeBSD & Linux have a difference of opinion about what a trap
  177  * means, deal with it here.
  178  *
  179  * MPSAFE
  180  */
  181 static int
  182 linux_translate_traps(int signal, int trap_code)
  183 {
  184         if (signal != SIGBUS)
  185                 return (signal);
  186         switch (trap_code) {
  187         case T_PROTFLT:
  188         case T_TSSFLT:
  189         case T_DOUBLEFLT:
  190         case T_PAGEFLT:
  191                 return (SIGSEGV);
  192         default:
  193                 return (signal);
  194         }
  195 }
  196 
  197 static int
  198 linux_fixup_elf(register_t **stack_base, struct image_params *imgp)
  199 {
  200         Elf32_Auxargs *args;
  201         Elf32_Auxinfo *argarray, *pos;
  202         Elf32_Addr *auxbase, *base;
  203         struct linux32_ps_strings *arginfo;
  204         int error, issetugid;
  205 
  206         arginfo = (struct linux32_ps_strings *)LINUX32_PS_STRINGS;
  207 
  208         KASSERT(curthread->td_proc == imgp->proc,
  209             ("unsafe linux_fixup_elf(), should be curproc"));
  210         base = (Elf32_Addr *)*stack_base;
  211         args = (Elf32_Auxargs *)imgp->auxargs;
  212         auxbase = base + (imgp->args->argc + 1 + imgp->args->envc + 1);
  213         argarray = pos = malloc(LINUX_AT_COUNT * sizeof(*pos), M_TEMP,
  214             M_WAITOK | M_ZERO);
  215 
  216         issetugid = imgp->proc->p_flag & P_SUGID ? 1 : 0;
  217         AUXARGS_ENTRY(pos, LINUX_AT_SYSINFO_EHDR,
  218             imgp->proc->p_sysent->sv_shared_page_base);
  219         AUXARGS_ENTRY(pos, LINUX_AT_SYSINFO, linux32_vsyscall);
  220         AUXARGS_ENTRY(pos, LINUX_AT_HWCAP, cpu_feature);
  221 
  222         /*
  223          * Do not export AT_CLKTCK when emulating Linux kernel prior to 2.4.0,
  224          * as it has appeared in the 2.4.0-rc7 first time.
  225          * Being exported, AT_CLKTCK is returned by sysconf(_SC_CLK_TCK),
  226          * glibc falls back to the hard-coded CLK_TCK value when aux entry
  227          * is not present.
  228          * Also see linux_times() implementation.
  229          */
  230         if (linux_kernver(curthread) >= LINUX_KERNVER_2004000)
  231                 AUXARGS_ENTRY(pos, LINUX_AT_CLKTCK, stclohz);
  232         AUXARGS_ENTRY(pos, AT_PHDR, args->phdr);
  233         AUXARGS_ENTRY(pos, AT_PHENT, args->phent);
  234         AUXARGS_ENTRY(pos, AT_PHNUM, args->phnum);
  235         AUXARGS_ENTRY(pos, AT_PAGESZ, args->pagesz);
  236         AUXARGS_ENTRY(pos, AT_FLAGS, args->flags);
  237         AUXARGS_ENTRY(pos, AT_ENTRY, args->entry);
  238         AUXARGS_ENTRY(pos, AT_BASE, args->base);
  239         AUXARGS_ENTRY(pos, LINUX_AT_SECURE, issetugid);
  240         AUXARGS_ENTRY(pos, AT_UID, imgp->proc->p_ucred->cr_ruid);
  241         AUXARGS_ENTRY(pos, AT_EUID, imgp->proc->p_ucred->cr_svuid);
  242         AUXARGS_ENTRY(pos, AT_GID, imgp->proc->p_ucred->cr_rgid);
  243         AUXARGS_ENTRY(pos, AT_EGID, imgp->proc->p_ucred->cr_svgid);
  244         AUXARGS_ENTRY(pos, LINUX_AT_PLATFORM, PTROUT(linux_platform));
  245         AUXARGS_ENTRY(pos, LINUX_AT_RANDOM, PTROUT(imgp->canary));
  246         if (imgp->execpathp != 0)
  247                 AUXARGS_ENTRY(pos, LINUX_AT_EXECFN, PTROUT(imgp->execpathp));
  248         if (args->execfd != -1)
  249                 AUXARGS_ENTRY(pos, AT_EXECFD, args->execfd);
  250         AUXARGS_ENTRY(pos, AT_NULL, 0);
  251 
  252         free(imgp->auxargs, M_TEMP);
  253         imgp->auxargs = NULL;
  254         KASSERT(pos - argarray <= LINUX_AT_COUNT, ("Too many auxargs"));
  255 
  256         error = copyout(&argarray[0], auxbase,
  257             sizeof(*argarray) * LINUX_AT_COUNT);
  258         free(argarray, M_TEMP);
  259         if (error != 0)
  260                 return (error);
  261 
  262         base--;
  263         if (suword32(base, (uint32_t)imgp->args->argc) == -1)
  264                 return (EFAULT);
  265         *stack_base = (register_t *)base;
  266         return (0);
  267 }
  268 
  269 static void
  270 linux_rt_sendsig(sig_t catcher, ksiginfo_t *ksi, sigset_t *mask)
  271 {
  272         struct thread *td = curthread;
  273         struct proc *p = td->td_proc;
  274         struct sigacts *psp;
  275         struct trapframe *regs;
  276         struct l_rt_sigframe *fp, frame;
  277         int oonstack;
  278         int sig;
  279         int code;
  280 
  281         sig = ksi->ksi_signo;
  282         code = ksi->ksi_code;
  283         PROC_LOCK_ASSERT(p, MA_OWNED);
  284         psp = p->p_sigacts;
  285         mtx_assert(&psp->ps_mtx, MA_OWNED);
  286         regs = td->td_frame;
  287         oonstack = sigonstack(regs->tf_rsp);
  288 
  289 #ifdef DEBUG
  290         if (ldebug(rt_sendsig))
  291                 printf(ARGS(rt_sendsig, "%p, %d, %p, %u"),
  292                     catcher, sig, (void*)mask, code);
  293 #endif
  294         /* Allocate space for the signal handler context. */
  295         if ((td->td_pflags & TDP_ALTSTACK) && !oonstack &&
  296             SIGISMEMBER(psp->ps_sigonstack, sig)) {
  297                 fp = (struct l_rt_sigframe *)((uintptr_t)td->td_sigstk.ss_sp +
  298                     td->td_sigstk.ss_size - sizeof(struct l_rt_sigframe));
  299         } else
  300                 fp = (struct l_rt_sigframe *)regs->tf_rsp - 1;
  301         mtx_unlock(&psp->ps_mtx);
  302 
  303         /* Build the argument list for the signal handler. */
  304         sig = bsd_to_linux_signal(sig);
  305 
  306         bzero(&frame, sizeof(frame));
  307 
  308         frame.sf_handler = PTROUT(catcher);
  309         frame.sf_sig = sig;
  310         frame.sf_siginfo = PTROUT(&fp->sf_si);
  311         frame.sf_ucontext = PTROUT(&fp->sf_sc);
  312 
  313         /* Fill in POSIX parts. */
  314         ksiginfo_to_lsiginfo(ksi, &frame.sf_si, sig);
  315 
  316         /*
  317          * Build the signal context to be used by sigreturn and libgcc unwind.
  318          */
  319         frame.sf_sc.uc_flags = 0;               /* XXX ??? */
  320         frame.sf_sc.uc_link = 0;                /* XXX ??? */
  321 
  322         frame.sf_sc.uc_stack.ss_sp = PTROUT(td->td_sigstk.ss_sp);
  323         frame.sf_sc.uc_stack.ss_size = td->td_sigstk.ss_size;
  324         frame.sf_sc.uc_stack.ss_flags = (td->td_pflags & TDP_ALTSTACK)
  325             ? ((oonstack) ? LINUX_SS_ONSTACK : 0) : LINUX_SS_DISABLE;
  326         PROC_UNLOCK(p);
  327 
  328         bsd_to_linux_sigset(mask, &frame.sf_sc.uc_sigmask);
  329 
  330         frame.sf_sc.uc_mcontext.sc_mask   = frame.sf_sc.uc_sigmask.__mask;
  331         frame.sf_sc.uc_mcontext.sc_edi    = regs->tf_rdi;
  332         frame.sf_sc.uc_mcontext.sc_esi    = regs->tf_rsi;
  333         frame.sf_sc.uc_mcontext.sc_ebp    = regs->tf_rbp;
  334         frame.sf_sc.uc_mcontext.sc_ebx    = regs->tf_rbx;
  335         frame.sf_sc.uc_mcontext.sc_esp    = regs->tf_rsp;
  336         frame.sf_sc.uc_mcontext.sc_edx    = regs->tf_rdx;
  337         frame.sf_sc.uc_mcontext.sc_ecx    = regs->tf_rcx;
  338         frame.sf_sc.uc_mcontext.sc_eax    = regs->tf_rax;
  339         frame.sf_sc.uc_mcontext.sc_eip    = regs->tf_rip;
  340         frame.sf_sc.uc_mcontext.sc_cs     = regs->tf_cs;
  341         frame.sf_sc.uc_mcontext.sc_gs     = regs->tf_gs;
  342         frame.sf_sc.uc_mcontext.sc_fs     = regs->tf_fs;
  343         frame.sf_sc.uc_mcontext.sc_es     = regs->tf_es;
  344         frame.sf_sc.uc_mcontext.sc_ds     = regs->tf_ds;
  345         frame.sf_sc.uc_mcontext.sc_eflags = regs->tf_rflags;
  346         frame.sf_sc.uc_mcontext.sc_esp_at_signal = regs->tf_rsp;
  347         frame.sf_sc.uc_mcontext.sc_ss     = regs->tf_ss;
  348         frame.sf_sc.uc_mcontext.sc_err    = regs->tf_err;
  349         frame.sf_sc.uc_mcontext.sc_cr2    = (u_int32_t)(uintptr_t)ksi->ksi_addr;
  350         frame.sf_sc.uc_mcontext.sc_trapno = bsd_to_linux_trapcode(code);
  351 
  352 #ifdef DEBUG
  353         if (ldebug(rt_sendsig))
  354                 printf(LMSG("rt_sendsig flags: 0x%x, sp: %p, ss: 0x%lx, mask: 0x%x"),
  355                     frame.sf_sc.uc_stack.ss_flags, td->td_sigstk.ss_sp,
  356                     td->td_sigstk.ss_size, frame.sf_sc.uc_mcontext.sc_mask);
  357 #endif
  358 
  359         if (copyout(&frame, fp, sizeof(frame)) != 0) {
  360                 /*
  361                  * Process has trashed its stack; give it an illegal
  362                  * instruction to halt it in its tracks.
  363                  */
  364 #ifdef DEBUG
  365                 if (ldebug(rt_sendsig))
  366                         printf(LMSG("rt_sendsig: bad stack %p, oonstack=%x"),
  367                             fp, oonstack);
  368 #endif
  369                 PROC_LOCK(p);
  370                 sigexit(td, SIGILL);
  371         }
  372 
  373         /* Build context to run handler in. */
  374         regs->tf_rsp = PTROUT(fp);
  375         regs->tf_rip = linux32_rt_sigcode;
  376         regs->tf_rflags &= ~(PSL_T | PSL_D);
  377         regs->tf_cs = _ucode32sel;
  378         regs->tf_ss = _udatasel;
  379         regs->tf_ds = _udatasel;
  380         regs->tf_es = _udatasel;
  381         regs->tf_fs = _ufssel;
  382         regs->tf_gs = _ugssel;
  383         regs->tf_flags = TF_HASSEGS;
  384         set_pcb_flags(td->td_pcb, PCB_FULL_IRET);
  385         PROC_LOCK(p);
  386         mtx_lock(&psp->ps_mtx);
  387 }
  388 
  389 
  390 /*
  391  * Send an interrupt to process.
  392  *
  393  * Stack is set up to allow sigcode stored
  394  * in u. to call routine, followed by kcall
  395  * to sigreturn routine below.  After sigreturn
  396  * resets the signal mask, the stack, and the
  397  * frame pointer, it returns to the user
  398  * specified pc, psl.
  399  */
  400 static void
  401 linux_sendsig(sig_t catcher, ksiginfo_t *ksi, sigset_t *mask)
  402 {
  403         struct thread *td = curthread;
  404         struct proc *p = td->td_proc;
  405         struct sigacts *psp;
  406         struct trapframe *regs;
  407         struct l_sigframe *fp, frame;
  408         l_sigset_t lmask;
  409         int oonstack;
  410         int sig, code;
  411 
  412         sig = ksi->ksi_signo;
  413         code = ksi->ksi_code;
  414         PROC_LOCK_ASSERT(p, MA_OWNED);
  415         psp = p->p_sigacts;
  416         mtx_assert(&psp->ps_mtx, MA_OWNED);
  417         if (SIGISMEMBER(psp->ps_siginfo, sig)) {
  418                 /* Signal handler installed with SA_SIGINFO. */
  419                 linux_rt_sendsig(catcher, ksi, mask);
  420                 return;
  421         }
  422 
  423         regs = td->td_frame;
  424         oonstack = sigonstack(regs->tf_rsp);
  425 
  426 #ifdef DEBUG
  427         if (ldebug(sendsig))
  428                 printf(ARGS(sendsig, "%p, %d, %p, %u"),
  429                     catcher, sig, (void*)mask, code);
  430 #endif
  431 
  432         /* Allocate space for the signal handler context. */
  433         if ((td->td_pflags & TDP_ALTSTACK) && !oonstack &&
  434             SIGISMEMBER(psp->ps_sigonstack, sig)) {
  435                 fp = (struct l_sigframe *)((uintptr_t)td->td_sigstk.ss_sp +
  436                     td->td_sigstk.ss_size - sizeof(struct l_sigframe));
  437         } else
  438                 fp = (struct l_sigframe *)regs->tf_rsp - 1;
  439         mtx_unlock(&psp->ps_mtx);
  440         PROC_UNLOCK(p);
  441 
  442         /* Build the argument list for the signal handler. */
  443         sig = bsd_to_linux_signal(sig);
  444 
  445         bzero(&frame, sizeof(frame));
  446 
  447         frame.sf_handler = PTROUT(catcher);
  448         frame.sf_sig = sig;
  449 
  450         bsd_to_linux_sigset(mask, &lmask);
  451 
  452         /* Build the signal context to be used by sigreturn. */
  453         frame.sf_sc.sc_mask   = lmask.__mask;
  454         frame.sf_sc.sc_gs     = regs->tf_gs;
  455         frame.sf_sc.sc_fs     = regs->tf_fs;
  456         frame.sf_sc.sc_es     = regs->tf_es;
  457         frame.sf_sc.sc_ds     = regs->tf_ds;
  458         frame.sf_sc.sc_edi    = regs->tf_rdi;
  459         frame.sf_sc.sc_esi    = regs->tf_rsi;
  460         frame.sf_sc.sc_ebp    = regs->tf_rbp;
  461         frame.sf_sc.sc_ebx    = regs->tf_rbx;
  462         frame.sf_sc.sc_esp    = regs->tf_rsp;
  463         frame.sf_sc.sc_edx    = regs->tf_rdx;
  464         frame.sf_sc.sc_ecx    = regs->tf_rcx;
  465         frame.sf_sc.sc_eax    = regs->tf_rax;
  466         frame.sf_sc.sc_eip    = regs->tf_rip;
  467         frame.sf_sc.sc_cs     = regs->tf_cs;
  468         frame.sf_sc.sc_eflags = regs->tf_rflags;
  469         frame.sf_sc.sc_esp_at_signal = regs->tf_rsp;
  470         frame.sf_sc.sc_ss     = regs->tf_ss;
  471         frame.sf_sc.sc_err    = regs->tf_err;
  472         frame.sf_sc.sc_cr2    = (u_int32_t)(uintptr_t)ksi->ksi_addr;
  473         frame.sf_sc.sc_trapno = bsd_to_linux_trapcode(code);
  474 
  475         frame.sf_extramask[0] = lmask.__mask;
  476 
  477         if (copyout(&frame, fp, sizeof(frame)) != 0) {
  478                 /*
  479                  * Process has trashed its stack; give it an illegal
  480                  * instruction to halt it in its tracks.
  481                  */
  482                 PROC_LOCK(p);
  483                 sigexit(td, SIGILL);
  484         }
  485 
  486         /* Build context to run handler in. */
  487         regs->tf_rsp = PTROUT(fp);
  488         regs->tf_rip = linux32_sigcode;
  489         regs->tf_rflags &= ~(PSL_T | PSL_D);
  490         regs->tf_cs = _ucode32sel;
  491         regs->tf_ss = _udatasel;
  492         regs->tf_ds = _udatasel;
  493         regs->tf_es = _udatasel;
  494         regs->tf_fs = _ufssel;
  495         regs->tf_gs = _ugssel;
  496         regs->tf_flags = TF_HASSEGS;
  497         set_pcb_flags(td->td_pcb, PCB_FULL_IRET);
  498         PROC_LOCK(p);
  499         mtx_lock(&psp->ps_mtx);
  500 }
  501 
  502 /*
  503  * System call to cleanup state after a signal
  504  * has been taken.  Reset signal mask and
  505  * stack state from context left by sendsig (above).
  506  * Return to previous pc and psl as specified by
  507  * context left by sendsig. Check carefully to
  508  * make sure that the user has not modified the
  509  * psl to gain improper privileges or to cause
  510  * a machine fault.
  511  */
  512 int
  513 linux_sigreturn(struct thread *td, struct linux_sigreturn_args *args)
  514 {
  515         struct l_sigframe frame;
  516         struct trapframe *regs;
  517         sigset_t bmask;
  518         l_sigset_t lmask;
  519         int eflags;
  520         ksiginfo_t ksi;
  521 
  522         regs = td->td_frame;
  523 
  524 #ifdef DEBUG
  525         if (ldebug(sigreturn))
  526                 printf(ARGS(sigreturn, "%p"), (void *)args->sfp);
  527 #endif
  528         /*
  529          * The trampoline code hands us the sigframe.
  530          * It is unsafe to keep track of it ourselves, in the event that a
  531          * program jumps out of a signal handler.
  532          */
  533         if (copyin(args->sfp, &frame, sizeof(frame)) != 0)
  534                 return (EFAULT);
  535 
  536         /* Check for security violations. */
  537 #define EFLAGS_SECURE(ef, oef)  ((((ef) ^ (oef)) & ~PSL_USERCHANGE) == 0)
  538         eflags = frame.sf_sc.sc_eflags;
  539         if (!EFLAGS_SECURE(eflags, regs->tf_rflags))
  540                 return(EINVAL);
  541 
  542         /*
  543          * Don't allow users to load a valid privileged %cs.  Let the
  544          * hardware check for invalid selectors, excess privilege in
  545          * other selectors, invalid %eip's and invalid %esp's.
  546          */
  547 #define CS_SECURE(cs)   (ISPL(cs) == SEL_UPL)
  548         if (!CS_SECURE(frame.sf_sc.sc_cs)) {
  549                 ksiginfo_init_trap(&ksi);
  550                 ksi.ksi_signo = SIGBUS;
  551                 ksi.ksi_code = BUS_OBJERR;
  552                 ksi.ksi_trapno = T_PROTFLT;
  553                 ksi.ksi_addr = (void *)regs->tf_rip;
  554                 trapsignal(td, &ksi);
  555                 return(EINVAL);
  556         }
  557 
  558         lmask.__mask = frame.sf_sc.sc_mask;
  559         lmask.__mask = frame.sf_extramask[0];
  560         linux_to_bsd_sigset(&lmask, &bmask);
  561         kern_sigprocmask(td, SIG_SETMASK, &bmask, NULL, 0);
  562 
  563         /* Restore signal context. */
  564         regs->tf_rdi    = frame.sf_sc.sc_edi;
  565         regs->tf_rsi    = frame.sf_sc.sc_esi;
  566         regs->tf_rbp    = frame.sf_sc.sc_ebp;
  567         regs->tf_rbx    = frame.sf_sc.sc_ebx;
  568         regs->tf_rdx    = frame.sf_sc.sc_edx;
  569         regs->tf_rcx    = frame.sf_sc.sc_ecx;
  570         regs->tf_rax    = frame.sf_sc.sc_eax;
  571         regs->tf_rip    = frame.sf_sc.sc_eip;
  572         regs->tf_cs     = frame.sf_sc.sc_cs;
  573         regs->tf_ds     = frame.sf_sc.sc_ds;
  574         regs->tf_es     = frame.sf_sc.sc_es;
  575         regs->tf_fs     = frame.sf_sc.sc_fs;
  576         regs->tf_gs     = frame.sf_sc.sc_gs;
  577         regs->tf_rflags = eflags;
  578         regs->tf_rsp    = frame.sf_sc.sc_esp_at_signal;
  579         regs->tf_ss     = frame.sf_sc.sc_ss;
  580         set_pcb_flags(td->td_pcb, PCB_FULL_IRET);
  581 
  582         return (EJUSTRETURN);
  583 }
  584 
  585 /*
  586  * System call to cleanup state after a signal
  587  * has been taken.  Reset signal mask and
  588  * stack state from context left by rt_sendsig (above).
  589  * Return to previous pc and psl as specified by
  590  * context left by sendsig. Check carefully to
  591  * make sure that the user has not modified the
  592  * psl to gain improper privileges or to cause
  593  * a machine fault.
  594  */
  595 int
  596 linux_rt_sigreturn(struct thread *td, struct linux_rt_sigreturn_args *args)
  597 {
  598         struct l_ucontext uc;
  599         struct l_sigcontext *context;
  600         sigset_t bmask;
  601         l_stack_t *lss;
  602         stack_t ss;
  603         struct trapframe *regs;
  604         int eflags;
  605         ksiginfo_t ksi;
  606 
  607         regs = td->td_frame;
  608 
  609 #ifdef DEBUG
  610         if (ldebug(rt_sigreturn))
  611                 printf(ARGS(rt_sigreturn, "%p"), (void *)args->ucp);
  612 #endif
  613         /*
  614          * The trampoline code hands us the ucontext.
  615          * It is unsafe to keep track of it ourselves, in the event that a
  616          * program jumps out of a signal handler.
  617          */
  618         if (copyin(args->ucp, &uc, sizeof(uc)) != 0)
  619                 return (EFAULT);
  620 
  621         context = &uc.uc_mcontext;
  622 
  623         /* Check for security violations. */
  624 #define EFLAGS_SECURE(ef, oef)  ((((ef) ^ (oef)) & ~PSL_USERCHANGE) == 0)
  625         eflags = context->sc_eflags;
  626         if (!EFLAGS_SECURE(eflags, regs->tf_rflags))
  627                 return(EINVAL);
  628 
  629         /*
  630          * Don't allow users to load a valid privileged %cs.  Let the
  631          * hardware check for invalid selectors, excess privilege in
  632          * other selectors, invalid %eip's and invalid %esp's.
  633          */
  634 #define CS_SECURE(cs)   (ISPL(cs) == SEL_UPL)
  635         if (!CS_SECURE(context->sc_cs)) {
  636                 ksiginfo_init_trap(&ksi);
  637                 ksi.ksi_signo = SIGBUS;
  638                 ksi.ksi_code = BUS_OBJERR;
  639                 ksi.ksi_trapno = T_PROTFLT;
  640                 ksi.ksi_addr = (void *)regs->tf_rip;
  641                 trapsignal(td, &ksi);
  642                 return(EINVAL);
  643         }
  644 
  645         linux_to_bsd_sigset(&uc.uc_sigmask, &bmask);
  646         kern_sigprocmask(td, SIG_SETMASK, &bmask, NULL, 0);
  647 
  648         /*
  649          * Restore signal context
  650          */
  651         regs->tf_gs     = context->sc_gs;
  652         regs->tf_fs     = context->sc_fs;
  653         regs->tf_es     = context->sc_es;
  654         regs->tf_ds     = context->sc_ds;
  655         regs->tf_rdi    = context->sc_edi;
  656         regs->tf_rsi    = context->sc_esi;
  657         regs->tf_rbp    = context->sc_ebp;
  658         regs->tf_rbx    = context->sc_ebx;
  659         regs->tf_rdx    = context->sc_edx;
  660         regs->tf_rcx    = context->sc_ecx;
  661         regs->tf_rax    = context->sc_eax;
  662         regs->tf_rip    = context->sc_eip;
  663         regs->tf_cs     = context->sc_cs;
  664         regs->tf_rflags = eflags;
  665         regs->tf_rsp    = context->sc_esp_at_signal;
  666         regs->tf_ss     = context->sc_ss;
  667         set_pcb_flags(td->td_pcb, PCB_FULL_IRET);
  668 
  669         /*
  670          * call sigaltstack & ignore results..
  671          */
  672         lss = &uc.uc_stack;
  673         ss.ss_sp = PTRIN(lss->ss_sp);
  674         ss.ss_size = lss->ss_size;
  675         ss.ss_flags = linux_to_bsd_sigaltstack(lss->ss_flags);
  676 
  677 #ifdef DEBUG
  678         if (ldebug(rt_sigreturn))
  679                 printf(LMSG("rt_sigret flags: 0x%x, sp: %p, ss: 0x%lx, mask: 0x%x"),
  680                     ss.ss_flags, ss.ss_sp, ss.ss_size, context->sc_mask);
  681 #endif
  682         (void)kern_sigaltstack(td, &ss, NULL);
  683 
  684         return (EJUSTRETURN);
  685 }
  686 
  687 static int
  688 linux32_fetch_syscall_args(struct thread *td)
  689 {
  690         struct proc *p;
  691         struct trapframe *frame;
  692         struct syscall_args *sa;
  693 
  694         p = td->td_proc;
  695         frame = td->td_frame;
  696         sa = &td->td_sa;
  697 
  698         sa->args[0] = frame->tf_rbx;
  699         sa->args[1] = frame->tf_rcx;
  700         sa->args[2] = frame->tf_rdx;
  701         sa->args[3] = frame->tf_rsi;
  702         sa->args[4] = frame->tf_rdi;
  703         sa->args[5] = frame->tf_rbp;    /* Unconfirmed */
  704         sa->code = frame->tf_rax;
  705 
  706         if (sa->code >= p->p_sysent->sv_size)
  707                 /* nosys */
  708                 sa->callp = &p->p_sysent->sv_table[p->p_sysent->sv_size - 1];
  709         else
  710                 sa->callp = &p->p_sysent->sv_table[sa->code];
  711         sa->narg = sa->callp->sy_narg;
  712 
  713         td->td_retval[0] = 0;
  714         td->td_retval[1] = frame->tf_rdx;
  715 
  716         return (0);
  717 }
  718 
  719 /*
  720  * Clear registers on exec
  721  * XXX copied from ia32_signal.c.
  722  */
  723 static void
  724 linux_exec_setregs(struct thread *td, struct image_params *imgp, u_long stack)
  725 {
  726         struct trapframe *regs = td->td_frame;
  727         struct pcb *pcb = td->td_pcb;
  728         register_t saved_rflags;
  729 
  730         regs = td->td_frame;
  731         pcb = td->td_pcb;
  732 
  733         if (td->td_proc->p_md.md_ldt != NULL)
  734                 user_ldt_free(td);
  735 
  736         critical_enter();
  737         wrmsr(MSR_FSBASE, 0);
  738         wrmsr(MSR_KGSBASE, 0);  /* User value while we're in the kernel */
  739         pcb->pcb_fsbase = 0;
  740         pcb->pcb_gsbase = 0;
  741         critical_exit();
  742         pcb->pcb_initial_fpucw = __LINUX_NPXCW__;
  743 
  744         saved_rflags = regs->tf_rflags & PSL_T;
  745         bzero((char *)regs, sizeof(struct trapframe));
  746         regs->tf_rip = imgp->entry_addr;
  747         regs->tf_rsp = stack;
  748         regs->tf_rflags = PSL_USER | saved_rflags;
  749         regs->tf_gs = _ugssel;
  750         regs->tf_fs = _ufssel;
  751         regs->tf_es = _udatasel;
  752         regs->tf_ds = _udatasel;
  753         regs->tf_ss = _udatasel;
  754         regs->tf_flags = TF_HASSEGS;
  755         regs->tf_cs = _ucode32sel;
  756         regs->tf_rbx = imgp->ps_strings;
  757 
  758         fpstate_drop(td);
  759 
  760         /* Do full restore on return so that we can change to a different %cs */
  761         set_pcb_flags(pcb, PCB_32BIT | PCB_FULL_IRET);
  762 }
  763 
  764 /*
  765  * XXX copied from ia32_sysvec.c.
  766  */
  767 static register_t *
  768 linux_copyout_strings(struct image_params *imgp)
  769 {
  770         int argc, envc;
  771         u_int32_t *vectp;
  772         char *stringp, *destp;
  773         u_int32_t *stack_base;
  774         struct linux32_ps_strings *arginfo;
  775         char canary[LINUX_AT_RANDOM_LEN];
  776         size_t execpath_len;
  777 
  778         /* Calculate string base and vector table pointers. */
  779         if (imgp->execpath != NULL && imgp->auxargs != NULL)
  780                 execpath_len = strlen(imgp->execpath) + 1;
  781         else
  782                 execpath_len = 0;
  783 
  784         arginfo = (struct linux32_ps_strings *)LINUX32_PS_STRINGS;
  785         destp = (caddr_t)arginfo - SPARE_USRSPACE -
  786             roundup(sizeof(canary), sizeof(char *)) -
  787             roundup(execpath_len, sizeof(char *)) -
  788             roundup(ARG_MAX - imgp->args->stringspace, sizeof(char *));
  789 
  790         if (execpath_len != 0) {
  791                 imgp->execpathp = (uintptr_t)arginfo - execpath_len;
  792                 copyout(imgp->execpath, (void *)imgp->execpathp, execpath_len);
  793         }
  794 
  795         /* Prepare the canary for SSP. */
  796         arc4rand(canary, sizeof(canary), 0);
  797         imgp->canary = (uintptr_t)arginfo -
  798             roundup(execpath_len, sizeof(char *)) -
  799             roundup(sizeof(canary), sizeof(char *));
  800         copyout(canary, (void *)imgp->canary, sizeof(canary));
  801 
  802         vectp = (uint32_t *)destp;
  803         if (imgp->auxargs) {
  804                 /*
  805                  * Allocate room on the stack for the ELF auxargs
  806                  * array.  It has LINUX_AT_COUNT entries.
  807                  */
  808                 vectp -= howmany(LINUX_AT_COUNT * sizeof(Elf32_Auxinfo),
  809                     sizeof(*vectp));
  810         }
  811 
  812         /*
  813          * Allocate room for the argv[] and env vectors including the
  814          * terminating NULL pointers.
  815          */
  816         vectp -= imgp->args->argc + 1 + imgp->args->envc + 1;
  817 
  818         /* vectp also becomes our initial stack base. */
  819         stack_base = vectp;
  820 
  821         stringp = imgp->args->begin_argv;
  822         argc = imgp->args->argc;
  823         envc = imgp->args->envc;
  824         /* Copy out strings - arguments and environment. */
  825         copyout(stringp, destp, ARG_MAX - imgp->args->stringspace);
  826 
  827         /* Fill in "ps_strings" struct for ps, w, etc. */
  828         suword32(&arginfo->ps_argvstr, (uint32_t)(intptr_t)vectp);
  829         suword32(&arginfo->ps_nargvstr, argc);
  830 
  831         /* Fill in argument portion of vector table. */
  832         for (; argc > 0; --argc) {
  833                 suword32(vectp++, (uint32_t)(intptr_t)destp);
  834                 while (*stringp++ != 0)
  835                         destp++;
  836                 destp++;
  837         }
  838 
  839         /* A null vector table pointer separates the argp's from the envp's. */
  840         suword32(vectp++, 0);
  841 
  842         suword32(&arginfo->ps_envstr, (uint32_t)(intptr_t)vectp);
  843         suword32(&arginfo->ps_nenvstr, envc);
  844 
  845         /* Fill in environment portion of vector table. */
  846         for (; envc > 0; --envc) {
  847                 suword32(vectp++, (uint32_t)(intptr_t)destp);
  848                 while (*stringp++ != 0)
  849                         destp++;
  850                 destp++;
  851         }
  852 
  853         /* The end of the vector table is a null pointer. */
  854         suword32(vectp, 0);
  855 
  856         return ((register_t *)stack_base);
  857 }
  858 
  859 static SYSCTL_NODE(_compat, OID_AUTO, linux32, CTLFLAG_RW, 0,
  860     "32-bit Linux emulation");
  861 
  862 static u_long   linux32_maxdsiz = LINUX32_MAXDSIZ;
  863 SYSCTL_ULONG(_compat_linux32, OID_AUTO, maxdsiz, CTLFLAG_RW,
  864     &linux32_maxdsiz, 0, "");
  865 static u_long   linux32_maxssiz = LINUX32_MAXSSIZ;
  866 SYSCTL_ULONG(_compat_linux32, OID_AUTO, maxssiz, CTLFLAG_RW,
  867     &linux32_maxssiz, 0, "");
  868 static u_long   linux32_maxvmem = LINUX32_MAXVMEM;
  869 SYSCTL_ULONG(_compat_linux32, OID_AUTO, maxvmem, CTLFLAG_RW,
  870     &linux32_maxvmem, 0, "");
  871 
  872 #if defined(DEBUG)
  873 SYSCTL_PROC(_compat_linux32, OID_AUTO, debug, CTLTYPE_STRING | CTLFLAG_RW, 0, 0,
  874     linux_sysctl_debug, "A", "Linux debugging control");
  875 #endif
  876 
  877 static void
  878 linux32_fixlimit(struct rlimit *rl, int which)
  879 {
  880 
  881         switch (which) {
  882         case RLIMIT_DATA:
  883                 if (linux32_maxdsiz != 0) {
  884                         if (rl->rlim_cur > linux32_maxdsiz)
  885                                 rl->rlim_cur = linux32_maxdsiz;
  886                         if (rl->rlim_max > linux32_maxdsiz)
  887                                 rl->rlim_max = linux32_maxdsiz;
  888                 }
  889                 break;
  890         case RLIMIT_STACK:
  891                 if (linux32_maxssiz != 0) {
  892                         if (rl->rlim_cur > linux32_maxssiz)
  893                                 rl->rlim_cur = linux32_maxssiz;
  894                         if (rl->rlim_max > linux32_maxssiz)
  895                                 rl->rlim_max = linux32_maxssiz;
  896                 }
  897                 break;
  898         case RLIMIT_VMEM:
  899                 if (linux32_maxvmem != 0) {
  900                         if (rl->rlim_cur > linux32_maxvmem)
  901                                 rl->rlim_cur = linux32_maxvmem;
  902                         if (rl->rlim_max > linux32_maxvmem)
  903                                 rl->rlim_max = linux32_maxvmem;
  904                 }
  905                 break;
  906         }
  907 }
  908 
  909 struct sysentvec elf_linux_sysvec = {
  910         .sv_size        = LINUX32_SYS_MAXSYSCALL,
  911         .sv_table       = linux32_sysent,
  912         .sv_mask        = 0,
  913         .sv_errsize     = ELAST + 1,
  914         .sv_errtbl      = linux_errtbl,
  915         .sv_transtrap   = linux_translate_traps,
  916         .sv_fixup       = linux_fixup_elf,
  917         .sv_sendsig     = linux_sendsig,
  918         .sv_sigcode     = &_binary_linux32_locore_o_start,
  919         .sv_szsigcode   = &linux_szsigcode,
  920         .sv_name        = "Linux ELF32",
  921         .sv_coredump    = elf32_coredump,
  922         .sv_imgact_try  = linux_exec_imgact_try,
  923         .sv_minsigstksz = LINUX_MINSIGSTKSZ,
  924         .sv_pagesize    = PAGE_SIZE,
  925         .sv_minuser     = VM_MIN_ADDRESS,
  926         .sv_maxuser     = LINUX32_MAXUSER,
  927         .sv_usrstack    = LINUX32_USRSTACK,
  928         .sv_psstrings   = LINUX32_PS_STRINGS,
  929         .sv_stackprot   = VM_PROT_ALL,
  930         .sv_copyout_strings = linux_copyout_strings,
  931         .sv_setregs     = linux_exec_setregs,
  932         .sv_fixlimit    = linux32_fixlimit,
  933         .sv_maxssiz     = &linux32_maxssiz,
  934         .sv_flags       = SV_ABI_LINUX | SV_ILP32 | SV_IA32 | SV_SHP,
  935         .sv_set_syscall_retval = cpu_set_syscall_retval,
  936         .sv_fetch_syscall_args = linux32_fetch_syscall_args,
  937         .sv_syscallnames = NULL,
  938         .sv_shared_page_base = LINUX32_SHAREDPAGE,
  939         .sv_shared_page_len = PAGE_SIZE,
  940         .sv_schedtail   = linux_schedtail,
  941         .sv_thread_detach = linux_thread_detach,
  942         .sv_trap        = NULL,
  943 };
  944 
  945 static void
  946 linux_vdso_install(void *param)
  947 {
  948 
  949         linux_szsigcode = (&_binary_linux32_locore_o_end -
  950             &_binary_linux32_locore_o_start);
  951 
  952         if (linux_szsigcode > elf_linux_sysvec.sv_shared_page_len)
  953                 panic("Linux invalid vdso size\n");
  954 
  955         __elfN(linux_vdso_fixup)(&elf_linux_sysvec);
  956 
  957         linux_shared_page_obj = __elfN(linux_shared_page_init)
  958             (&linux_shared_page_mapping);
  959 
  960         __elfN(linux_vdso_reloc)(&elf_linux_sysvec);
  961 
  962         bcopy(elf_linux_sysvec.sv_sigcode, linux_shared_page_mapping,
  963             linux_szsigcode);
  964         elf_linux_sysvec.sv_shared_page_obj = linux_shared_page_obj;
  965 
  966         linux_kplatform = linux_shared_page_mapping +
  967             (linux_platform - (caddr_t)elf_linux_sysvec.sv_shared_page_base);
  968 }
  969 SYSINIT(elf_linux_vdso_init, SI_SUB_EXEC, SI_ORDER_ANY,
  970     linux_vdso_install, NULL);
  971 
  972 static void
  973 linux_vdso_deinstall(void *param)
  974 {
  975 
  976         __elfN(linux_shared_page_fini)(linux_shared_page_obj);
  977 }
  978 SYSUNINIT(elf_linux_vdso_uninit, SI_SUB_EXEC, SI_ORDER_FIRST,
  979     linux_vdso_deinstall, NULL);
  980 
  981 static char GNU_ABI_VENDOR[] = "GNU";
  982 static int GNULINUX_ABI_DESC = 0;
  983 
  984 static bool
  985 linux32_trans_osrel(const Elf_Note *note, int32_t *osrel)
  986 {
  987         const Elf32_Word *desc;
  988         uintptr_t p;
  989 
  990         p = (uintptr_t)(note + 1);
  991         p += roundup2(note->n_namesz, sizeof(Elf32_Addr));
  992 
  993         desc = (const Elf32_Word *)p;
  994         if (desc[0] != GNULINUX_ABI_DESC)
  995                 return (false);
  996 
  997         /*
  998          * For Linux we encode osrel using the Linux convention of
  999          *      (version << 16) | (major << 8) | (minor)
 1000          * See macro in linux_mib.h
 1001          */
 1002         *osrel = LINUX_KERNVER(desc[1], desc[2], desc[3]);
 1003 
 1004         return (true);
 1005 }
 1006 
 1007 static Elf_Brandnote linux32_brandnote = {
 1008         .hdr.n_namesz   = sizeof(GNU_ABI_VENDOR),
 1009         .hdr.n_descsz   = 16,   /* XXX at least 16 */
 1010         .hdr.n_type     = 1,
 1011         .vendor         = GNU_ABI_VENDOR,
 1012         .flags          = BN_TRANSLATE_OSREL,
 1013         .trans_osrel    = linux32_trans_osrel
 1014 };
 1015 
 1016 static Elf32_Brandinfo linux_brand = {
 1017         .brand          = ELFOSABI_LINUX,
 1018         .machine        = EM_386,
 1019         .compat_3_brand = "Linux",
 1020         .emul_path      = "/compat/linux",
 1021         .interp_path    = "/lib/ld-linux.so.1",
 1022         .sysvec         = &elf_linux_sysvec,
 1023         .interp_newpath = NULL,
 1024         .brand_note     = &linux32_brandnote,
 1025         .flags          = BI_CAN_EXEC_DYN | BI_BRAND_NOTE
 1026 };
 1027 
 1028 static Elf32_Brandinfo linux_glibc2brand = {
 1029         .brand          = ELFOSABI_LINUX,
 1030         .machine        = EM_386,
 1031         .compat_3_brand = "Linux",
 1032         .emul_path      = "/compat/linux",
 1033         .interp_path    = "/lib/ld-linux.so.2",
 1034         .sysvec         = &elf_linux_sysvec,
 1035         .interp_newpath = NULL,
 1036         .brand_note     = &linux32_brandnote,
 1037         .flags          = BI_CAN_EXEC_DYN | BI_BRAND_NOTE
 1038 };
 1039 
 1040 static Elf32_Brandinfo linux_muslbrand = {
 1041         .brand          = ELFOSABI_LINUX,
 1042         .machine        = EM_386,
 1043         .compat_3_brand = "Linux",
 1044         .emul_path      = "/compat/linux",
 1045         .interp_path    = "/lib/ld-musl-i386.so.1",
 1046         .sysvec         = &elf_linux_sysvec,
 1047         .interp_newpath = NULL,
 1048         .brand_note     = &linux32_brandnote,
 1049         .flags          = BI_CAN_EXEC_DYN | BI_BRAND_NOTE
 1050 };
 1051 
 1052 Elf32_Brandinfo *linux_brandlist[] = {
 1053         &linux_brand,
 1054         &linux_glibc2brand,
 1055         &linux_muslbrand,
 1056         NULL
 1057 };
 1058 
 1059 static int
 1060 linux_elf_modevent(module_t mod, int type, void *data)
 1061 {
 1062         Elf32_Brandinfo **brandinfo;
 1063         int error;
 1064         struct linux_ioctl_handler **lihp;
 1065 
 1066         error = 0;
 1067 
 1068         switch(type) {
 1069         case MOD_LOAD:
 1070                 for (brandinfo = &linux_brandlist[0]; *brandinfo != NULL;
 1071                      ++brandinfo)
 1072                         if (elf32_insert_brand_entry(*brandinfo) < 0)
 1073                                 error = EINVAL;
 1074                 if (error == 0) {
 1075                         SET_FOREACH(lihp, linux_ioctl_handler_set)
 1076                                 linux32_ioctl_register_handler(*lihp);
 1077                         LIST_INIT(&futex_list);
 1078                         mtx_init(&futex_mtx, "ftllk", NULL, MTX_DEF);
 1079                         stclohz = (stathz ? stathz : hz);
 1080                         if (bootverbose)
 1081                                 printf("Linux ELF exec handler installed\n");
 1082                 } else
 1083                         printf("cannot insert Linux ELF brand handler\n");
 1084                 break;
 1085         case MOD_UNLOAD:
 1086                 for (brandinfo = &linux_brandlist[0]; *brandinfo != NULL;
 1087                      ++brandinfo)
 1088                         if (elf32_brand_inuse(*brandinfo))
 1089                                 error = EBUSY;
 1090                 if (error == 0) {
 1091                         for (brandinfo = &linux_brandlist[0];
 1092                              *brandinfo != NULL; ++brandinfo)
 1093                                 if (elf32_remove_brand_entry(*brandinfo) < 0)
 1094                                         error = EINVAL;
 1095                 }
 1096                 if (error == 0) {
 1097                         SET_FOREACH(lihp, linux_ioctl_handler_set)
 1098                                 linux32_ioctl_unregister_handler(*lihp);
 1099                         mtx_destroy(&futex_mtx);
 1100                         if (bootverbose)
 1101                                 printf("Linux ELF exec handler removed\n");
 1102                 } else
 1103                         printf("Could not deinstall ELF interpreter entry\n");
 1104                 break;
 1105         default:
 1106                 return (EOPNOTSUPP);
 1107         }
 1108         return (error);
 1109 }
 1110 
 1111 static moduledata_t linux_elf_mod = {
 1112         "linuxelf",
 1113         linux_elf_modevent,
 1114         0
 1115 };
 1116 
 1117 DECLARE_MODULE_TIED(linuxelf, linux_elf_mod, SI_SUB_EXEC, SI_ORDER_ANY);
 1118 MODULE_DEPEND(linuxelf, linux_common, 1, 1, 1);
 1119 FEATURE(linux, "Linux 32bit support");

Cache object: 85d0ea3448279f369703c9e1e045f586


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


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