📄 fsys.s
字号:
1: ld4 r17=[r9] // r17 <- current->thread_info->flags ;; mov ar.ccv=r17 and r18=~_TIF_SIGPENDING,r17 // r18 <- r17 & ~(1 << TIF_SIGPENDING) ;; st8 [r2]=r14 // update current->blocked with new mask cmpxchg4.acq r8=[r9],r18,ar.ccv // current->thread_info->flags <- r18 ;; cmp.ne p6,p0=r17,r8 // update failed?(p6) br.cond.spnt.few 1b // yes -> retry#ifdef CONFIG_SMP st4.rel [r31]=r0 // release the lock#endif ssm psr.i ;; srlz.d // ensure psr.i is set again mov r18=0 // i must not leak kernel bits....store_mask:EX(.fail_efault, (p15) probe.w.fault r34, 3) // verify user has write-access to *osetEX(.fail_efault, (p15) st8 [r34]=r3) mov r2=0 // i must not leak kernel bits... mov r3=0 // i must not leak kernel bits... mov r8=0 // return 0 mov r9=0 // i must not leak kernel bits... mov r14=0 // i must not leak kernel bits... mov r17=0 // i must not leak kernel bits... mov r31=0 // i must not leak kernel bits... FSYS_RETURN.sig_pending:#ifdef CONFIG_SMP st4.rel [r31]=r0 // release the lock#endif ssm psr.i ;; srlz.d br.sptk.many fsys_fallback_syscall // with signal pending, do the heavy-weight syscall#ifdef CONFIG_SMP.lock_contention: /* Rather than spinning here, fall back on doing a heavy-weight syscall. */ ssm psr.i ;; srlz.d br.sptk.many fsys_fallback_syscall#endifEND(fsys_rt_sigprocmask)ENTRY(fsys_fallback_syscall) .prologue .altrp b6 .body /* * We only get here from light-weight syscall handlers. Thus, we already * know that r15 contains a valid syscall number. No need to re-check. */ adds r17=-1024,r15 movl r14=sys_call_table ;; rsm psr.i shladd r18=r17,3,r14 ;; ld8 r18=[r18] // load normal (heavy-weight) syscall entry-point mov r29=psr // read psr (12 cyc load latency) mov r27=ar.rsc mov r21=ar.fpsr mov r26=ar.pfsEND(fsys_fallback_syscall) /* FALL THROUGH */GLOBAL_ENTRY(fsys_bubble_down) .prologue .altrp b6 .body /* * We get here for syscalls that don't have a lightweight * handler. For those, we need to bubble down into the kernel * and that requires setting up a minimal pt_regs structure, * and initializing the CPU state more or less as if an * interruption had occurred. To make syscall-restarts work, * we setup pt_regs such that cr_iip points to the second * instruction in syscall_via_break. Decrementing the IP * hence will restart the syscall via break and not * decrementing IP will return us to the caller, as usual. * Note that we preserve the value of psr.pp rather than * initializing it from dcr.pp. This makes it possible to * distinguish fsyscall execution from other privileged * execution. * * On entry: * - normal fsyscall handler register usage, except * that we also have: * - r18: address of syscall entry point * - r21: ar.fpsr * - r26: ar.pfs * - r27: ar.rsc * - r29: psr * * We used to clear some PSR bits here but that requires slow * serialization. Fortuntely, that isn't really necessary. * The rationale is as follows: we used to clear bits * ~PSR_PRESERVED_BITS in PSR.L. Since * PSR_PRESERVED_BITS==PSR.{UP,MFL,MFH,PK,DT,PP,SP,RT,IC}, we * ended up clearing PSR.{BE,AC,I,DFL,DFH,DI,DB,SI,TB}. * However, * * PSR.BE : already is turned off in __kernel_syscall_via_epc() * PSR.AC : don't care (kernel normally turns PSR.AC on) * PSR.I : already turned off by the time fsys_bubble_down gets * invoked * PSR.DFL: always 0 (kernel never turns it on) * PSR.DFH: don't care --- kernel never touches f32-f127 on its own * initiative * PSR.DI : always 0 (kernel never turns it on) * PSR.SI : always 0 (kernel never turns it on) * PSR.DB : don't care --- kernel never enables kernel-level * breakpoints * PSR.TB : must be 0 already; if it wasn't zero on entry to * __kernel_syscall_via_epc, the branch to fsys_bubble_down * will trigger a taken branch; the taken-trap-handler then * converts the syscall into a break-based system-call. */ /* * Reading psr.l gives us only bits 0-31, psr.it, and psr.mc. * The rest we have to synthesize. */# define PSR_ONE_BITS ((3 << IA64_PSR_CPL0_BIT) \ | (0x1 << IA64_PSR_RI_BIT) \ | IA64_PSR_BN | IA64_PSR_I) invala // M0|1 movl r14=ia64_ret_from_syscall // X nop.m 0 movl r28=__kernel_syscall_via_break // X create cr.iip ;; mov r2=r16 // A get task addr to addl-addressable register adds r16=IA64_TASK_THREAD_ON_USTACK_OFFSET,r16 // A mov r31=pr // I0 save pr (2 cyc) ;; st1 [r16]=r0 // M2|3 clear current->thread.on_ustack flag addl r22=IA64_RBS_OFFSET,r2 // A compute base of RBS add r3=TI_FLAGS+IA64_TASK_SIZE,r2 // A ;; ld4 r3=[r3] // M0|1 r3 = current_thread_info()->flags lfetch.fault.excl.nt1 [r22] // M0|1 prefetch register backing-store nop.i 0 ;; mov ar.rsc=0 // M2 set enforced lazy mode, pl 0, LE, loadrs=0 nop.m 0 nop.i 0 ;; mov r23=ar.bspstore // M2 (12 cyc) save ar.bspstore mov.m r24=ar.rnat // M2 (5 cyc) read ar.rnat (dual-issues!) nop.i 0 ;; mov ar.bspstore=r22 // M2 (6 cyc) switch to kernel RBS movl r8=PSR_ONE_BITS // X ;; mov r25=ar.unat // M2 (5 cyc) save ar.unat mov r19=b6 // I0 save b6 (2 cyc) mov r20=r1 // A save caller's gp in r20 ;; or r29=r8,r29 // A construct cr.ipsr value to save mov b6=r18 // I0 copy syscall entry-point to b6 (7 cyc) addl r1=IA64_STK_OFFSET-IA64_PT_REGS_SIZE,r2 // A compute base of memory stack mov r18=ar.bsp // M2 save (kernel) ar.bsp (12 cyc) cmp.ne pKStk,pUStk=r0,r0 // A set pKStk <- 0, pUStk <- 1 br.call.sptk.many b7=ia64_syscall_setup // B ;; mov ar.rsc=0x3 // M2 set eager mode, pl 0, LE, loadrs=0 mov rp=r14 // I0 set the real return addr and r3=_TIF_SYSCALL_TRACEAUDIT,r3 // A ;; ssm psr.i // M2 we're on kernel stacks now, reenable irqs cmp.eq p8,p0=r3,r0 // A(p10) br.cond.spnt.many ia64_ret_from_syscall // B return if bad call-frame or r15 is a NaT nop.m 0(p8) br.call.sptk.many b6=b6 // B (ignore return address) br.cond.spnt ia64_trace_syscall // BEND(fsys_bubble_down) .rodata .align 8 .globl fsyscall_table data8 fsys_bubble_downfsyscall_table: data8 fsys_ni_syscall data8 0 // exit // 1025 data8 0 // read data8 0 // write data8 0 // open data8 0 // close data8 0 // creat // 1030 data8 0 // link data8 0 // unlink data8 0 // execve data8 0 // chdir data8 0 // fchdir // 1035 data8 0 // utimes data8 0 // mknod data8 0 // chmod data8 0 // chown data8 0 // lseek // 1040 data8 fsys_getpid // getpid data8 fsys_getppid // getppid data8 0 // mount data8 0 // umount data8 0 // setuid // 1045 data8 0 // getuid data8 0 // geteuid data8 0 // ptrace data8 0 // access data8 0 // sync // 1050 data8 0 // fsync data8 0 // fdatasync data8 0 // kill data8 0 // rename data8 0 // mkdir // 1055 data8 0 // rmdir data8 0 // dup data8 0 // pipe data8 0 // times data8 0 // brk // 1060 data8 0 // setgid data8 0 // getgid data8 0 // getegid data8 0 // acct data8 0 // ioctl // 1065 data8 0 // fcntl data8 0 // umask data8 0 // chroot data8 0 // ustat data8 0 // dup2 // 1070 data8 0 // setreuid data8 0 // setregid data8 0 // getresuid data8 0 // setresuid data8 0 // getresgid // 1075 data8 0 // setresgid data8 0 // getgroups data8 0 // setgroups data8 0 // getpgid data8 0 // setpgid // 1080 data8 0 // setsid data8 0 // getsid data8 0 // sethostname data8 0 // setrlimit data8 0 // getrlimit // 1085 data8 0 // getrusage data8 fsys_gettimeofday // gettimeofday data8 0 // settimeofday data8 0 // select data8 0 // poll // 1090 data8 0 // symlink data8 0 // readlink data8 0 // uselib data8 0 // swapon data8 0 // swapoff // 1095 data8 0 // reboot data8 0 // truncate data8 0 // ftruncate data8 0 // fchmod data8 0 // fchown // 1100 data8 0 // getpriority data8 0 // setpriority data8 0 // statfs data8 0 // fstatfs data8 0 // gettid // 1105 data8 0 // semget data8 0 // semop data8 0 // semctl data8 0 // msgget data8 0 // msgsnd // 1110 data8 0 // msgrcv data8 0 // msgctl data8 0 // shmget data8 0 // shmat data8 0 // shmdt // 1115 data8 0 // shmctl data8 0 // syslog data8 0 // setitimer data8 0 // getitimer data8 0 // 1120 data8 0 data8 0 data8 0 // vhangup data8 0 // lchown data8 0 // remap_file_pages // 1125 data8 0 // wait4 data8 0 // sysinfo data8 0 // clone data8 0 // setdomainname data8 0 // newuname // 1130 data8 0 // adjtimex data8 0 data8 0 // init_module data8 0 // delete_module data8 0 // 1135 data8 0 data8 0 // quotactl data8 0 // bdflush data8 0 // sysfs data8 0 // personality // 1140 data8 0 // afs_syscall data8 0 // setfsuid data8 0 // setfsgid data8 0 // getdents data8 0 // flock // 1145 data8 0 // readv data8 0 // writev data8 0 // pread64 data8 0 // pwrite64 data8 0 // sysctl // 1150 data8 0 // mmap data8 0 // munmap data8 0 // mlock data8 0 // mlockall data8 0 // mprotect // 1155 data8 0 // mremap data8 0 // msync data8 0 // munlock data8 0 // munlockall data8 0 // sched_getparam // 1160 data8 0 // sched_setparam data8 0 // sched_getscheduler data8 0 // sched_setscheduler data8 0 // sched_yield data8 0 // sched_get_priority_max // 1165 data8 0 // sched_get_priority_min data8 0 // sched_rr_get_interval data8 0 // nanosleep data8 0 // nfsservctl data8 0 // prctl // 1170 data8 0 // getpagesize data8 0 // mmap2 data8 0 // pciconfig_read data8 0 // pciconfig_write data8 0 // perfmonctl // 1175 data8 0 // sigaltstack data8 0 // rt_sigaction data8 0 // rt_sigpending data8 fsys_rt_sigprocmask // rt_sigprocmask data8 0 // rt_sigqueueinfo // 1180 data8 0 // rt_sigreturn data8 0 // rt_sigsuspend data8 0 // rt_sigtimedwait data8 0 // getcwd data8 0 // capget // 1185 data8 0 // capset data8 0 // sendfile data8 0 data8 0 data8 0 // socket // 1190 data8 0 // bind data8 0 // connect data8 0 // listen data8 0 // accept data8 0 // getsockname // 1195 data8 0 // getpeername data8 0 // socketpair data8 0 // send data8 0 // sendto data8 0 // recv // 1200 data8 0 // recvfrom data8 0 // shutdown data8 0 // setsockopt data8 0 // getsockopt data8 0 // sendmsg // 1205 data8 0 // recvmsg data8 0 // pivot_root data8 0 // mincore data8 0 // madvise data8 0 // newstat // 1210 data8 0 // newlstat data8 0 // newfstat data8 0 // clone2 data8 0 // getdents64 data8 0 // getunwind // 1215 data8 0 // readahead data8 0 // setxattr data8 0 // lsetxattr data8 0 // fsetxattr data8 0 // getxattr // 1220 data8 0 // lgetxattr data8 0 // fgetxattr data8 0 // listxattr data8 0 // llistxattr data8 0 // flistxattr // 1225 data8 0 // removexattr data8 0 // lremovexattr data8 0 // fremovexattr data8 0 // tkill data8 0 // futex // 1230 data8 0 // sched_setaffinity data8 0 // sched_getaffinity data8 fsys_set_tid_address // set_tid_address data8 0 // fadvise64_64 data8 0 // tgkill // 1235 data8 0 // exit_group data8 0 // lookup_dcookie data8 0 // io_setup data8 0 // io_destroy data8 0 // io_getevents // 1240 data8 0 // io_submit data8 0 // io_cancel data8 0 // epoll_create data8 0 // epoll_ctl data8 0 // epoll_wait // 1245 data8 0 // restart_syscall data8 0 // semtimedop data8 0 // timer_create data8 0 // timer_settime data8 0 // timer_gettime // 1250 data8 0 // timer_getoverrun data8 0 // timer_delete data8 0 // clock_settime data8 fsys_clock_gettime // clock_gettime data8 0 // clock_getres // 1255 data8 0 // clock_nanosleep data8 0 // fstatfs64 data8 0 // statfs64 data8 0 data8 0 // 1260 data8 0 data8 0 // mq_open data8 0 // mq_unlink data8 0 // mq_timedsend data8 0 // mq_timedreceive // 1265 data8 0 // mq_notify data8 0 // mq_getsetattr data8 0 // kexec_load data8 0 data8 0 // 1270 data8 0 data8 0 data8 0 data8 0 data8 0 // 1275 data8 0 data8 0 data8 0 data8 0 .org fsyscall_table + 8*NR_syscalls // guard against failures to increase NR_syscalls⌨️ 快捷键说明
复制代码
Ctrl + C
搜索代码
Ctrl + F
全屏模式
F11
切换主题
Ctrl + Shift + D
显示快捷键
?
增大字号
Ctrl + =
减小字号
Ctrl + -