/* Machine independent support for Solaris /proc (process file system) for GDB. Copyright (C) 1999-2020 Free Software Foundation, Inc. Written by Michael Snyder at Cygnus Solutions. Based on work by Fred Fish, Stu Grossman, Geoff Noer, and others. This file is part of GDB. This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 3 of the License, or (at your option) any later version. This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. You should have received a copy of the GNU General Public License along with this program. If not, see . */ #include "defs.h" #include "inferior.h" #include "infrun.h" #include "target.h" #include "gdbcore.h" #include "elf-bfd.h" /* for elfcore_write_* */ #include "gdbcmd.h" #include "gdbthread.h" #include "regcache.h" #include "inf-child.h" #include "nat/fork-inferior.h" #include "gdbarch.h" #define _STRUCTURED_PROC 1 /* Should be done by configure script. */ #include #include #include #include "gdbsupport/gdb_wait.h" #include #include #include "gdb_bfd.h" #include "inflow.h" #include "auxv.h" #include "procfs.h" #include "observable.h" #include "gdbsupport/scoped_fd.h" #include "gdbsupport/pathstuff.h" /* This module provides the interface between GDB and the /proc file system, which is used on many versions of Unix as a means for debuggers to control other processes. /proc works by imitating a file system: you open a simulated file that represents the process you wish to interact with, and perform operations on that "file" in order to examine or change the state of the other process. The most important thing to know about /proc and this module is that there are two very different interfaces to /proc: One that uses the ioctl system call, and another that uses read and write system calls. This module supports only the Solaris version of the read/write interface. */ #include #include /* opendir/readdir, for listing the LWP's */ #include /* for O_RDONLY */ #include /* for "X_OK" */ #include /* for struct stat */ /* Note: procfs-utils.h must be included after the above system header files, because it redefines various system calls using macros. This may be incompatible with the prototype declarations. */ #include "proc-utils.h" /* Prototypes for supply_gregset etc. */ #include "gregset.h" /* =================== TARGET_OPS "MODULE" =================== */ /* This module defines the GDB target vector and its methods. */ static enum target_xfer_status procfs_xfer_memory (gdb_byte *, const gdb_byte *, ULONGEST, ULONGEST, ULONGEST *); class procfs_target final : public inf_child_target { public: void create_inferior (const char *, const std::string &, char **, int) override; void kill () override; void mourn_inferior () override; void attach (const char *, int) override; void detach (inferior *inf, int) override; void resume (ptid_t, int, enum gdb_signal) override; ptid_t wait (ptid_t, struct target_waitstatus *, int) override; void fetch_registers (struct regcache *, int) override; void store_registers (struct regcache *, int) override; enum target_xfer_status xfer_partial (enum target_object object, const char *annex, gdb_byte *readbuf, const gdb_byte *writebuf, ULONGEST offset, ULONGEST len, ULONGEST *xfered_len) override; void pass_signals (gdb::array_view) override; void files_info () override; void update_thread_list () override; bool thread_alive (ptid_t ptid) override; std::string pid_to_str (ptid_t) override; char *pid_to_exec_file (int pid) override; thread_control_capabilities get_thread_control_capabilities () override { return tc_schedlock; } /* find_memory_regions support method for gcore */ int find_memory_regions (find_memory_region_ftype func, void *data) override; char *make_corefile_notes (bfd *, int *) override; bool info_proc (const char *, enum info_proc_what) override; #if PR_MODEL_NATIVE == PR_MODEL_LP64 int auxv_parse (gdb_byte **readptr, gdb_byte *endptr, CORE_ADDR *typep, CORE_ADDR *valp) override; #endif bool stopped_by_watchpoint () override; int insert_watchpoint (CORE_ADDR, int, enum target_hw_bp_type, struct expression *) override; int remove_watchpoint (CORE_ADDR, int, enum target_hw_bp_type, struct expression *) override; int region_ok_for_hw_watchpoint (CORE_ADDR, int) override; int can_use_hw_breakpoint (enum bptype, int, int) override; bool stopped_data_address (CORE_ADDR *) override; }; static procfs_target the_procfs_target; #if PR_MODEL_NATIVE == PR_MODEL_LP64 /* When GDB is built as 64-bit application on Solaris, the auxv data is presented in 64-bit format. We need to provide a custom parser to handle that. */ int procfs_target::auxv_parse (gdb_byte **readptr, gdb_byte *endptr, CORE_ADDR *typep, CORE_ADDR *valp) { enum bfd_endian byte_order = gdbarch_byte_order (target_gdbarch ()); gdb_byte *ptr = *readptr; if (endptr == ptr) return 0; if (endptr - ptr < 8 * 2) return -1; *typep = extract_unsigned_integer (ptr, 4, byte_order); ptr += 8; /* The size of data is always 64-bit. If the application is 32-bit, it will be zero extended, as expected. */ *valp = extract_unsigned_integer (ptr, 8, byte_order); ptr += 8; *readptr = ptr; return 1; } #endif /* =================== END, TARGET_OPS "MODULE" =================== */ /* World Unification: Put any typedefs, defines etc. here that are required for the unification of code that handles different versions of /proc. */ enum { READ_WATCHFLAG = WA_READ, WRITE_WATCHFLAG = WA_WRITE, EXEC_WATCHFLAG = WA_EXEC, AFTER_WATCHFLAG = WA_TRAPAFTER }; /* =================== STRUCT PROCINFO "MODULE" =================== */ /* FIXME: this comment will soon be out of date W.R.T. threads. */ /* The procinfo struct is a wrapper to hold all the state information concerning a /proc process. There should be exactly one procinfo for each process, and since GDB currently can debug only one process at a time, that means there should be only one procinfo. All of the LWP's of a process can be accessed indirectly thru the single process procinfo. However, against the day when GDB may debug more than one process, this data structure is kept in a list (which for now will hold no more than one member), and many functions will have a pointer to a procinfo as an argument. There will be a separate procinfo structure for use by the (not yet implemented) "info proc" command, so that we can print useful information about any random process without interfering with the inferior's procinfo information. */ /* format strings for /proc paths */ #define MAIN_PROC_NAME_FMT "/proc/%d" #define CTL_PROC_NAME_FMT "/proc/%d/ctl" #define AS_PROC_NAME_FMT "/proc/%d/as" #define MAP_PROC_NAME_FMT "/proc/%d/map" #define STATUS_PROC_NAME_FMT "/proc/%d/status" #define MAX_PROC_NAME_SIZE sizeof("/proc/999999/lwp/0123456789/lwpstatus") typedef struct procinfo { struct procinfo *next; int pid; /* Process ID */ int tid; /* Thread/LWP id */ /* process state */ int was_stopped; int ignore_next_sigstop; int ctl_fd; /* File descriptor for /proc control file */ int status_fd; /* File descriptor for /proc status file */ int as_fd; /* File descriptor for /proc as file */ char pathname[MAX_PROC_NAME_SIZE]; /* Pathname to /proc entry */ fltset_t saved_fltset; /* Saved traced hardware fault set */ sigset_t saved_sigset; /* Saved traced signal set */ sigset_t saved_sighold; /* Saved held signal set */ sysset_t *saved_exitset; /* Saved traced system call exit set */ sysset_t *saved_entryset; /* Saved traced system call entry set */ pstatus_t prstatus; /* Current process status info */ struct procinfo *thread_list; int status_valid : 1; int gregs_valid : 1; int fpregs_valid : 1; int threads_valid: 1; } procinfo; static char errmsg[128]; /* shared error msg buffer */ /* Function prototypes for procinfo module: */ static procinfo *find_procinfo_or_die (int pid, int tid); static procinfo *find_procinfo (int pid, int tid); static procinfo *create_procinfo (int pid, int tid); static void destroy_procinfo (procinfo *p); static void dead_procinfo (procinfo *p, const char *msg, int killp); static int open_procinfo_files (procinfo *p, int which); static void close_procinfo_files (procinfo *p); static int iterate_over_mappings (procinfo *pi, find_memory_region_ftype child_func, void *data, int (*func) (struct prmap *map, find_memory_region_ftype child_func, void *data)); /* The head of the procinfo list: */ static procinfo *procinfo_list; /* Search the procinfo list. Return a pointer to procinfo, or NULL if not found. */ static procinfo * find_procinfo (int pid, int tid) { procinfo *pi; for (pi = procinfo_list; pi; pi = pi->next) if (pi->pid == pid) break; if (pi) if (tid) { /* Don't check threads_valid. If we're updating the thread_list, we want to find whatever threads are already here. This means that in general it is the caller's responsibility to check threads_valid and update before calling find_procinfo, if the caller wants to find a new thread. */ for (pi = pi->thread_list; pi; pi = pi->next) if (pi->tid == tid) break; } return pi; } /* Calls find_procinfo, but errors on failure. */ static procinfo * find_procinfo_or_die (int pid, int tid) { procinfo *pi = find_procinfo (pid, tid); if (pi == NULL) { if (tid) error (_("procfs: couldn't find pid %d " "(kernel thread %d) in procinfo list."), pid, tid); else error (_("procfs: couldn't find pid %d in procinfo list."), pid); } return pi; } /* Wrapper for `open'. The appropriate open call is attempted; if unsuccessful, it will be retried as many times as needed for the EAGAIN and EINTR conditions. For other conditions, retry the open a limited number of times. In addition, a short sleep is imposed prior to retrying the open. The reason for this sleep is to give the kernel a chance to catch up and create the file in question in the event that GDB "wins" the race to open a file before the kernel has created it. */ static int open_with_retry (const char *pathname, int flags) { int retries_remaining, status; retries_remaining = 2; while (1) { status = open (pathname, flags); if (status >= 0 || retries_remaining == 0) break; else if (errno != EINTR && errno != EAGAIN) { retries_remaining--; sleep (1); } } return status; } /* Open the file descriptor for the process or LWP. We only open the control file descriptor; the others are opened lazily as needed. Returns the file descriptor, or zero for failure. */ enum { FD_CTL, FD_STATUS, FD_AS }; static int open_procinfo_files (procinfo *pi, int which) { char tmp[MAX_PROC_NAME_SIZE]; int fd; /* This function is getting ALMOST long enough to break up into several. Here is some rationale: There are several file descriptors that may need to be open for any given process or LWP. The ones we're interested in are: - control (ctl) write-only change the state - status (status) read-only query the state - address space (as) read/write access memory - map (map) read-only virtual addr map Most of these are opened lazily as they are needed. The pathnames for the 'files' for an LWP look slightly different from those of a first-class process: Pathnames for a process (): /proc//ctl /proc//status /proc//as /proc//map Pathnames for an LWP (lwp-id): /proc//lwp//lwpctl /proc//lwp//lwpstatus An LWP has no map or address space file descriptor, since the memory map and address space are shared by all LWPs. */ /* In this case, there are several different file descriptors that we might be asked to open. The control file descriptor will be opened early, but the others will be opened lazily as they are needed. */ strcpy (tmp, pi->pathname); switch (which) { /* Which file descriptor to open? */ case FD_CTL: if (pi->tid) strcat (tmp, "/lwpctl"); else strcat (tmp, "/ctl"); fd = open_with_retry (tmp, O_WRONLY); if (fd < 0) return 0; /* fail */ pi->ctl_fd = fd; break; case FD_AS: if (pi->tid) return 0; /* There is no 'as' file descriptor for an lwp. */ strcat (tmp, "/as"); fd = open_with_retry (tmp, O_RDWR); if (fd < 0) return 0; /* fail */ pi->as_fd = fd; break; case FD_STATUS: if (pi->tid) strcat (tmp, "/lwpstatus"); else strcat (tmp, "/status"); fd = open_with_retry (tmp, O_RDONLY); if (fd < 0) return 0; /* fail */ pi->status_fd = fd; break; default: return 0; /* unknown file descriptor */ } return 1; /* success */ } /* Allocate a data structure and link it into the procinfo list. First tries to find a pre-existing one (FIXME: why?). Returns the pointer to new procinfo struct. */ static procinfo * create_procinfo (int pid, int tid) { procinfo *pi, *parent = NULL; pi = find_procinfo (pid, tid); if (pi != NULL) return pi; /* Already exists, nothing to do. */ /* Find parent before doing malloc, to save having to cleanup. */ if (tid != 0) parent = find_procinfo_or_die (pid, 0); /* FIXME: should I create it if it doesn't exist yet? */ pi = XNEW (procinfo); memset (pi, 0, sizeof (procinfo)); pi->pid = pid; pi->tid = tid; pi->saved_entryset = XNEW (sysset_t); pi->saved_exitset = XNEW (sysset_t); /* Chain into list. */ if (tid == 0) { xsnprintf (pi->pathname, sizeof (pi->pathname), MAIN_PROC_NAME_FMT, pid); pi->next = procinfo_list; procinfo_list = pi; } else { xsnprintf (pi->pathname, sizeof (pi->pathname), "/proc/%d/lwp/%d", pid, tid); pi->next = parent->thread_list; parent->thread_list = pi; } return pi; } /* Close all file descriptors associated with the procinfo. */ static void close_procinfo_files (procinfo *pi) { if (pi->ctl_fd > 0) close (pi->ctl_fd); if (pi->as_fd > 0) close (pi->as_fd); if (pi->status_fd > 0) close (pi->status_fd); pi->ctl_fd = pi->as_fd = pi->status_fd = 0; } /* Destructor function. Close, unlink and deallocate the object. */ static void destroy_one_procinfo (procinfo **list, procinfo *pi) { procinfo *ptr; /* Step one: unlink the procinfo from its list. */ if (pi == *list) *list = pi->next; else for (ptr = *list; ptr; ptr = ptr->next) if (ptr->next == pi) { ptr->next = pi->next; break; } /* Step two: close any open file descriptors. */ close_procinfo_files (pi); /* Step three: free the memory. */ xfree (pi->saved_entryset); xfree (pi->saved_exitset); xfree (pi); } static void destroy_procinfo (procinfo *pi) { procinfo *tmp; if (pi->tid != 0) /* Destroy a thread procinfo. */ { tmp = find_procinfo (pi->pid, 0); /* Find the parent process. */ destroy_one_procinfo (&tmp->thread_list, pi); } else /* Destroy a process procinfo and all its threads. */ { /* First destroy the children, if any; */ while (pi->thread_list != NULL) destroy_one_procinfo (&pi->thread_list, pi->thread_list); /* Then destroy the parent. Genocide!!! */ destroy_one_procinfo (&procinfo_list, pi); } } /* A deleter that calls destroy_procinfo. */ struct procinfo_deleter { void operator() (procinfo *pi) const { destroy_procinfo (pi); } }; typedef std::unique_ptr procinfo_up; enum { NOKILL, KILL }; /* To be called on a non_recoverable error for a procinfo. Prints error messages, optionally sends a SIGKILL to the process, then destroys the data structure. */ static void dead_procinfo (procinfo *pi, const char *msg, int kill_p) { char procfile[80]; if (pi->pathname) print_sys_errmsg (pi->pathname, errno); else { xsnprintf (procfile, sizeof (procfile), "process %d", pi->pid); print_sys_errmsg (procfile, errno); } if (kill_p == KILL) kill (pi->pid, SIGKILL); destroy_procinfo (pi); error ("%s", msg); } /* =================== END, STRUCT PROCINFO "MODULE" =================== */ /* =================== /proc "MODULE" =================== */ /* This "module" is the interface layer between the /proc system API and the gdb target vector functions. This layer consists of access functions that encapsulate each of the basic operations that we need to use from the /proc API. The main motivation for this layer is to hide the fact that there are two very different implementations of the /proc API. Rather than have a bunch of #ifdefs all thru the gdb target vector functions, we do our best to hide them all in here. */ static long proc_flags (procinfo *pi); static int proc_why (procinfo *pi); static int proc_what (procinfo *pi); static int proc_set_current_signal (procinfo *pi, int signo); static int proc_get_current_thread (procinfo *pi); static int proc_iterate_over_threads (procinfo *pi, int (*func) (procinfo *, procinfo *, void *), void *ptr); static void proc_warn (procinfo *pi, const char *func, int line) { xsnprintf (errmsg, sizeof (errmsg), "procfs: %s line %d, %s", func, line, pi->pathname); print_sys_errmsg (errmsg, errno); } static void proc_error (procinfo *pi, const char *func, int line) { xsnprintf (errmsg, sizeof (errmsg), "procfs: %s line %d, %s", func, line, pi->pathname); perror_with_name (errmsg); } /* Updates the status struct in the procinfo. There is a 'valid' flag, to let other functions know when this function needs to be called (so the status is only read when it is needed). The status file descriptor is also only opened when it is needed. Returns non-zero for success, zero for failure. */ static int proc_get_status (procinfo *pi) { /* Status file descriptor is opened "lazily". */ if (pi->status_fd == 0 && open_procinfo_files (pi, FD_STATUS) == 0) { pi->status_valid = 0; return 0; } if (lseek (pi->status_fd, 0, SEEK_SET) < 0) pi->status_valid = 0; /* fail */ else { /* Sigh... I have to read a different data structure, depending on whether this is a main process or an LWP. */ if (pi->tid) pi->status_valid = (read (pi->status_fd, (char *) &pi->prstatus.pr_lwp, sizeof (lwpstatus_t)) == sizeof (lwpstatus_t)); else { pi->status_valid = (read (pi->status_fd, (char *) &pi->prstatus, sizeof (pstatus_t)) == sizeof (pstatus_t)); } } if (pi->status_valid) { PROC_PRETTYFPRINT_STATUS (proc_flags (pi), proc_why (pi), proc_what (pi), proc_get_current_thread (pi)); } /* The status struct includes general regs, so mark them valid too. */ pi->gregs_valid = pi->status_valid; /* In the read/write multiple-fd model, the status struct includes the fp regs too, so mark them valid too. */ pi->fpregs_valid = pi->status_valid; return pi->status_valid; /* True if success, false if failure. */ } /* Returns the process flags (pr_flags field). */ static long proc_flags (procinfo *pi) { if (!pi->status_valid) if (!proc_get_status (pi)) return 0; /* FIXME: not a good failure value (but what is?) */ return pi->prstatus.pr_lwp.pr_flags; } /* Returns the pr_why field (why the process stopped). */ static int proc_why (procinfo *pi) { if (!pi->status_valid) if (!proc_get_status (pi)) return 0; /* FIXME: not a good failure value (but what is?) */ return pi->prstatus.pr_lwp.pr_why; } /* Returns the pr_what field (details of why the process stopped). */ static int proc_what (procinfo *pi) { if (!pi->status_valid) if (!proc_get_status (pi)) return 0; /* FIXME: not a good failure value (but what is?) */ return pi->prstatus.pr_lwp.pr_what; } /* This function is only called when PI is stopped by a watchpoint. Assuming the OS supports it, write to *ADDR the data address which triggered it and return 1. Return 0 if it is not possible to know the address. */ static int proc_watchpoint_address (procinfo *pi, CORE_ADDR *addr) { if (!pi->status_valid) if (!proc_get_status (pi)) return 0; *addr = (CORE_ADDR) gdbarch_pointer_to_address (target_gdbarch (), builtin_type (target_gdbarch ())->builtin_data_ptr, (gdb_byte *) &pi->prstatus.pr_lwp.pr_info.si_addr); return 1; } /* Returns the pr_nsysarg field (number of args to the current syscall). */ static int proc_nsysarg (procinfo *pi) { if (!pi->status_valid) if (!proc_get_status (pi)) return 0; return pi->prstatus.pr_lwp.pr_nsysarg; } /* Returns the pr_sysarg field (pointer to the arguments of current syscall). */ static long * proc_sysargs (procinfo *pi) { if (!pi->status_valid) if (!proc_get_status (pi)) return NULL; return (long *) &pi->prstatus.pr_lwp.pr_sysarg; } /* Set or reset any of the following process flags: PR_FORK -- forked child will inherit trace flags PR_RLC -- traced process runs when last /proc file closed. PR_KLC -- traced process is killed when last /proc file closed. PR_ASYNC -- LWP's get to run/stop independently. This function is done using read/write [PCSET/PCRESET/PCUNSET]. Arguments: pi -- the procinfo flag -- one of PR_FORK, PR_RLC, or PR_ASYNC mode -- 1 for set, 0 for reset. Returns non-zero for success, zero for failure. */ enum { FLAG_RESET, FLAG_SET }; static int proc_modify_flag (procinfo *pi, long flag, long mode) { long win = 0; /* default to fail */ /* These operations affect the process as a whole, and applying them to an individual LWP has the same meaning as applying them to the main process. Therefore, if we're ever called with a pointer to an LWP's procinfo, let's substitute the process's procinfo and avoid opening the LWP's file descriptor unnecessarily. */ if (pi->pid != 0) pi = find_procinfo_or_die (pi->pid, 0); procfs_ctl_t arg[2]; if (mode == FLAG_SET) /* Set the flag (RLC, FORK, or ASYNC). */ arg[0] = PCSET; else /* Reset the flag. */ arg[0] = PCUNSET; arg[1] = flag; win = (write (pi->ctl_fd, (void *) &arg, sizeof (arg)) == sizeof (arg)); /* The above operation renders the procinfo's cached pstatus obsolete. */ pi->status_valid = 0; if (!win) warning (_("procfs: modify_flag failed to turn %s %s"), flag == PR_FORK ? "PR_FORK" : flag == PR_RLC ? "PR_RLC" : flag == PR_ASYNC ? "PR_ASYNC" : flag == PR_KLC ? "PR_KLC" : "", mode == FLAG_RESET ? "off" : "on"); return win; } /* Set the run_on_last_close flag. Process with all threads will become runnable when debugger closes all /proc fds. Returns non-zero for success, zero for failure. */ static int proc_set_run_on_last_close (procinfo *pi) { return proc_modify_flag (pi, PR_RLC, FLAG_SET); } /* Reset the run_on_last_close flag. The process will NOT become runnable when debugger closes its file handles. Returns non-zero for success, zero for failure. */ static int proc_unset_run_on_last_close (procinfo *pi) { return proc_modify_flag (pi, PR_RLC, FLAG_RESET); } /* Reset inherit_on_fork flag. If the process forks a child while we are registered for events in the parent, then we will NOT receive events from the child. Returns non-zero for success, zero for failure. */ static int proc_unset_inherit_on_fork (procinfo *pi) { return proc_modify_flag (pi, PR_FORK, FLAG_RESET); } /* Set PR_ASYNC flag. If one LWP stops because of a debug event (signal etc.), the remaining LWPs will continue to run. Returns non-zero for success, zero for failure. */ static int proc_set_async (procinfo *pi) { return proc_modify_flag (pi, PR_ASYNC, FLAG_SET); } /* Reset PR_ASYNC flag. If one LWP stops because of a debug event (signal etc.), then all other LWPs will stop as well. Returns non-zero for success, zero for failure. */ static int proc_unset_async (procinfo *pi) { return proc_modify_flag (pi, PR_ASYNC, FLAG_RESET); } /* Request the process/LWP to stop. Does not wait. Returns non-zero for success, zero for failure. */ static int proc_stop_process (procinfo *pi) { int win; /* We might conceivably apply this operation to an LWP, and the LWP's ctl file descriptor might not be open. */ if (pi->ctl_fd == 0 && open_procinfo_files (pi, FD_CTL) == 0) return 0; else { procfs_ctl_t cmd = PCSTOP; win = (write (pi->ctl_fd, (char *) &cmd, sizeof (cmd)) == sizeof (cmd)); } return win; } /* Wait for the process or LWP to stop (block until it does). Returns non-zero for success, zero for failure. */ static int proc_wait_for_stop (procinfo *pi) { int win; /* We should never have to apply this operation to any procinfo except the one for the main process. If that ever changes for any reason, then take out the following clause and replace it with one that makes sure the ctl_fd is open. */ if (pi->tid != 0) pi = find_procinfo_or_die (pi->pid, 0); procfs_ctl_t cmd = PCWSTOP; set_sigint_trap (); win = (write (pi->ctl_fd, (char *) &cmd, sizeof (cmd)) == sizeof (cmd)); clear_sigint_trap (); /* We been runnin' and we stopped -- need to update status. */ pi->status_valid = 0; return win; } /* Make the process or LWP runnable. Options (not all are implemented): - single-step - clear current fault - clear current signal - abort the current system call - stop as soon as finished with system call - (ioctl): set traced signal set - (ioctl): set held signal set - (ioctl): set traced fault set - (ioctl): set start pc (vaddr) Always clears the current fault. PI is the process or LWP to operate on. If STEP is true, set the process or LWP to trap after one instruction. If SIGNO is zero, clear the current signal if any; if non-zero, set the current signal to this one. Returns non-zero for success, zero for failure. */ static int proc_run_process (procinfo *pi, int step, int signo) { int win; int runflags; /* We will probably have to apply this operation to individual threads, so make sure the control file descriptor is open. */ if (pi->ctl_fd == 0 && open_procinfo_files (pi, FD_CTL) == 0) return 0; runflags = PRCFAULT; /* Always clear current fault. */ if (step) runflags |= PRSTEP; if (signo == 0) runflags |= PRCSIG; else if (signo != -1) /* -1 means do nothing W.R.T. signals. */ proc_set_current_signal (pi, signo); procfs_ctl_t cmd[2]; cmd[0] = PCRUN; cmd[1] = runflags; win = (write (pi->ctl_fd, (char *) &cmd, sizeof (cmd)) == sizeof (cmd)); return win; } /* Register to trace signals in the process or LWP. Returns non-zero for success, zero for failure. */ static int proc_set_traced_signals (procinfo *pi, sigset_t *sigset) { int win; /* We should never have to apply this operation to any procinfo except the one for the main process. If that ever changes for any reason, then take out the following clause and replace it with one that makes sure the ctl_fd is open. */ if (pi->tid != 0) pi = find_procinfo_or_die (pi->pid, 0); struct { procfs_ctl_t cmd; /* Use char array to avoid alignment issues. */ char sigset[sizeof (sigset_t)]; } arg; arg.cmd = PCSTRACE; memcpy (&arg.sigset, sigset, sizeof (sigset_t)); win = (write (pi->ctl_fd, (char *) &arg, sizeof (arg)) == sizeof (arg)); /* The above operation renders the procinfo's cached pstatus obsolete. */ pi->status_valid = 0; if (!win) warning (_("procfs: set_traced_signals failed")); return win; } /* Register to trace hardware faults in the process or LWP. Returns non-zero for success, zero for failure. */ static int proc_set_traced_faults (procinfo *pi, fltset_t *fltset) { int win; /* We should never have to apply this operation to any procinfo except the one for the main process. If that ever changes for any reason, then take out the following clause and replace it with one that makes sure the ctl_fd is open. */ if (pi->tid != 0) pi = find_procinfo_or_die (pi->pid, 0); struct { procfs_ctl_t cmd; /* Use char array to avoid alignment issues. */ char fltset[sizeof (fltset_t)]; } arg; arg.cmd = PCSFAULT; memcpy (&arg.fltset, fltset, sizeof (fltset_t)); win = (write (pi->ctl_fd, (char *) &arg, sizeof (arg)) == sizeof (arg)); /* The above operation renders the procinfo's cached pstatus obsolete. */ pi->status_valid = 0; return win; } /* Register to trace entry to system calls in the process or LWP. Returns non-zero for success, zero for failure. */ static int proc_set_traced_sysentry (procinfo *pi, sysset_t *sysset) { int win; /* We should never have to apply this operation to any procinfo except the one for the main process. If that ever changes for any reason, then take out the following clause and replace it with one that makes sure the ctl_fd is open. */ if (pi->tid != 0) pi = find_procinfo_or_die (pi->pid, 0); struct { procfs_ctl_t cmd; /* Use char array to avoid alignment issues. */ char sysset[sizeof (sysset_t)]; } arg; arg.cmd = PCSENTRY; memcpy (&arg.sysset, sysset, sizeof (sysset_t)); win = (write (pi->ctl_fd, (char *) &arg, sizeof (arg)) == sizeof (arg)); /* The above operation renders the procinfo's cached pstatus obsolete. */ pi->status_valid = 0; return win; } /* Register to trace exit from system calls in the process or LWP. Returns non-zero for success, zero for failure. */ static int proc_set_traced_sysexit (procinfo *pi, sysset_t *sysset) { int win; /* We should never have to apply this operation to any procinfo except the one for the main process. If that ever changes for any reason, then take out the following clause and replace it with one that makes sure the ctl_fd is open. */ if (pi->tid != 0) pi = find_procinfo_or_die (pi->pid, 0); struct gdb_proc_ctl_pcsexit { procfs_ctl_t cmd; /* Use char array to avoid alignment issues. */ char sysset[sizeof (sysset_t)]; } arg; arg.cmd = PCSEXIT; memcpy (&arg.sysset, sysset, sizeof (sysset_t)); win = (write (pi->ctl_fd, (char *) &arg, sizeof (arg)) == sizeof (arg)); /* The above operation renders the procinfo's cached pstatus obsolete. */ pi->status_valid = 0; return win; } /* Specify the set of blocked / held signals in the process or LWP. Returns non-zero for success, zero for failure. */ static int proc_set_held_signals (procinfo *pi, sigset_t *sighold) { int win; /* We should never have to apply this operation to any procinfo except the one for the main process. If that ever changes for any reason, then take out the following clause and replace it with one that makes sure the ctl_fd is open. */ if (pi->tid != 0) pi = find_procinfo_or_die (pi->pid, 0); struct { procfs_ctl_t cmd; /* Use char array to avoid alignment issues. */ char hold[sizeof (sigset_t)]; } arg; arg.cmd = PCSHOLD; memcpy (&arg.hold, sighold, sizeof (sigset_t)); win = (write (pi->ctl_fd, (void *) &arg, sizeof (arg)) == sizeof (arg)); /* The above operation renders the procinfo's cached pstatus obsolete. */ pi->status_valid = 0; return win; } /* Returns the set of signals that are held / blocked. Will also copy the sigset if SAVE is non-zero. */ static sigset_t * proc_get_held_signals (procinfo *pi, sigset_t *save) { sigset_t *ret = NULL; /* We should never have to apply this operation to any procinfo except the one for the main process. If that ever changes for any reason, then take out the following clause and replace it with one that makes sure the ctl_fd is open. */ if (pi->tid != 0) pi = find_procinfo_or_die (pi->pid, 0); if (!pi->status_valid) if (!proc_get_status (pi)) return NULL; ret = &pi->prstatus.pr_lwp.pr_lwphold; if (save && ret) memcpy (save, ret, sizeof (sigset_t)); return ret; } /* Returns the set of signals that are traced / debugged. Will also copy the sigset if SAVE is non-zero. */ static sigset_t * proc_get_traced_signals (procinfo *pi, sigset_t *save) { sigset_t *ret = NULL; /* We should never have to apply this operation to any procinfo except the one for the main process. If that ever changes for any reason, then take out the following clause and replace it with one that makes sure the ctl_fd is open. */ if (pi->tid != 0) pi = find_procinfo_or_die (pi->pid, 0); if (!pi->status_valid) if (!proc_get_status (pi)) return NULL; ret = &pi->prstatus.pr_sigtrace; if (save && ret) memcpy (save, ret, sizeof (sigset_t)); return ret; } /* Returns the set of hardware faults that are traced /debugged. Will also copy the faultset if SAVE is non-zero. */ static fltset_t * proc_get_traced_faults (procinfo *pi, fltset_t *save) { fltset_t *ret = NULL; /* We should never have to apply this operation to any procinfo except the one for the main process. If that ever changes for any reason, then take out the following clause and replace it with one that makes sure the ctl_fd is open. */ if (pi->tid != 0) pi = find_procinfo_or_die (pi->pid, 0); if (!pi->status_valid) if (!proc_get_status (pi)) return NULL; ret = &pi->prstatus.pr_flttrace; if (save && ret) memcpy (save, ret, sizeof (fltset_t)); return ret; } /* Returns the set of syscalls that are traced /debugged on entry. Will also copy the syscall set if SAVE is non-zero. */ static sysset_t * proc_get_traced_sysentry (procinfo *pi, sysset_t *save) { sysset_t *ret = NULL; /* We should never have to apply this operation to any procinfo except the one for the main process. If that ever changes for any reason, then take out the following clause and replace it with one that makes sure the ctl_fd is open. */ if (pi->tid != 0) pi = find_procinfo_or_die (pi->pid, 0); if (!pi->status_valid) if (!proc_get_status (pi)) return NULL; ret = &pi->prstatus.pr_sysentry; if (save && ret) memcpy (save, ret, sizeof (sysset_t)); return ret; } /* Returns the set of syscalls that are traced /debugged on exit. Will also copy the syscall set if SAVE is non-zero. */ static sysset_t * proc_get_traced_sysexit (procinfo *pi, sysset_t *save) { sysset_t *ret = NULL; /* We should never have to apply this operation to any procinfo except the one for the main process. If that ever changes for any reason, then take out the following clause and replace it with one that makes sure the ctl_fd is open. */ if (pi->tid != 0) pi = find_procinfo_or_die (pi->pid, 0); if (!pi->status_valid) if (!proc_get_status (pi)) return NULL; ret = &pi->prstatus.pr_sysexit; if (save && ret) memcpy (save, ret, sizeof (sysset_t)); return ret; } /* The current fault (if any) is cleared; the associated signal will not be sent to the process or LWP when it resumes. Returns non-zero for success, zero for failure. */ static int proc_clear_current_fault (procinfo *pi) { int win; /* We should never have to apply this operation to any procinfo except the one for the main process. If that ever changes for any reason, then take out the following clause and replace it with one that makes sure the ctl_fd is open. */ if (pi->tid != 0) pi = find_procinfo_or_die (pi->pid, 0); procfs_ctl_t cmd = PCCFAULT; win = (write (pi->ctl_fd, (void *) &cmd, sizeof (cmd)) == sizeof (cmd)); return win; } /* Set the "current signal" that will be delivered next to the process. NOTE: semantics are different from those of KILL. This signal will be delivered to the process or LWP immediately when it is resumed (even if the signal is held/blocked); it will NOT immediately cause another event of interest, and will NOT first trap back to the debugger. Returns non-zero for success, zero for failure. */ static int proc_set_current_signal (procinfo *pi, int signo) { int win; struct { procfs_ctl_t cmd; /* Use char array to avoid alignment issues. */ char sinfo[sizeof (siginfo_t)]; } arg; siginfo_t mysinfo; ptid_t wait_ptid; struct target_waitstatus wait_status; /* We should never have to apply this operation to any procinfo except the one for the main process. If that ever changes for any reason, then take out the following clause and replace it with one that makes sure the ctl_fd is open. */ if (pi->tid != 0) pi = find_procinfo_or_die (pi->pid, 0); /* The pointer is just a type alias. */ get_last_target_status (&wait_ptid, &wait_status); if (wait_ptid == inferior_ptid && wait_status.kind == TARGET_WAITKIND_STOPPED && wait_status.value.sig == gdb_signal_from_host (signo) && proc_get_status (pi) && pi->prstatus.pr_lwp.pr_info.si_signo == signo ) /* Use the siginfo associated with the signal being redelivered. */ memcpy (arg.sinfo, &pi->prstatus.pr_lwp.pr_info, sizeof (siginfo_t)); else { mysinfo.si_signo = signo; mysinfo.si_code = 0; mysinfo.si_pid = getpid (); /* ?why? */ mysinfo.si_uid = getuid (); /* ?why? */ memcpy (arg.sinfo, &mysinfo, sizeof (siginfo_t)); } arg.cmd = PCSSIG; win = (write (pi->ctl_fd, (void *) &arg, sizeof (arg)) == sizeof (arg)); return win; } /* The current signal (if any) is cleared, and is not sent to the process or LWP when it resumes. Returns non-zero for success, zero for failure. */ static int proc_clear_current_signal (procinfo *pi) { int win; /* We should never have to apply this operation to any procinfo except the one for the main process. If that ever changes for any reason, then take out the following clause and replace it with one that makes sure the ctl_fd is open. */ if (pi->tid != 0) pi = find_procinfo_or_die (pi->pid, 0); struct { procfs_ctl_t cmd; /* Use char array to avoid alignment issues. */ char sinfo[sizeof (siginfo_t)]; } arg; siginfo_t mysinfo; arg.cmd = PCSSIG; /* The pointer is just a type alias. */ mysinfo.si_signo = 0; mysinfo.si_code = 0; mysinfo.si_errno = 0; mysinfo.si_pid = getpid (); /* ?why? */ mysinfo.si_uid = getuid (); /* ?why? */ memcpy (arg.sinfo, &mysinfo, sizeof (siginfo_t)); win = (write (pi->ctl_fd, (void *) &arg, sizeof (arg)) == sizeof (arg)); return win; } /* Return the general-purpose registers for the process or LWP corresponding to PI. Upon failure, return NULL. */ static gdb_gregset_t * proc_get_gregs (procinfo *pi) { if (!pi->status_valid || !pi->gregs_valid) if (!proc_get_status (pi)) return NULL; return &pi->prstatus.pr_lwp.pr_reg; } /* Return the general-purpose registers for the process or LWP corresponding to PI. Upon failure, return NULL. */ static gdb_fpregset_t * proc_get_fpregs (procinfo *pi) { if (!pi->status_valid || !pi->fpregs_valid) if (!proc_get_status (pi)) return NULL; return &pi->prstatus.pr_lwp.pr_fpreg; } /* Write the general-purpose registers back to the process or LWP corresponding to PI. Return non-zero for success, zero for failure. */ static int proc_set_gregs (procinfo *pi) { gdb_gregset_t *gregs; int win; gregs = proc_get_gregs (pi); if (gregs == NULL) return 0; /* proc_get_regs has already warned. */ if (pi->ctl_fd == 0 && open_procinfo_files (pi, FD_CTL) == 0) return 0; else { struct { procfs_ctl_t cmd; /* Use char array to avoid alignment issues. */ char gregs[sizeof (gdb_gregset_t)]; } arg; arg.cmd = PCSREG; memcpy (&arg.gregs, gregs, sizeof (arg.gregs)); win = (write (pi->ctl_fd, (void *) &arg, sizeof (arg)) == sizeof (arg)); } /* Policy: writing the registers invalidates our cache. */ pi->gregs_valid = 0; return win; } /* Write the floating-pointer registers back to the process or LWP corresponding to PI. Return non-zero for success, zero for failure. */ static int proc_set_fpregs (procinfo *pi) { gdb_fpregset_t *fpregs; int win; fpregs = proc_get_fpregs (pi); if (fpregs == NULL) return 0; /* proc_get_fpregs has already warned. */ if (pi->ctl_fd == 0 && open_procinfo_files (pi, FD_CTL) == 0) return 0; else { struct { procfs_ctl_t cmd; /* Use char array to avoid alignment issues. */ char fpregs[sizeof (gdb_fpregset_t)]; } arg; arg.cmd = PCSFPREG; memcpy (&arg.fpregs, fpregs, sizeof (arg.fpregs)); win = (write (pi->ctl_fd, (void *) &arg, sizeof (arg)) == sizeof (arg)); } /* Policy: writing the registers invalidates our cache. */ pi->fpregs_valid = 0; return win; } /* Send a signal to the proc or lwp with the semantics of "kill()". Returns non-zero for success, zero for failure. */ static int proc_kill (procinfo *pi, int signo) { int win; /* We might conceivably apply this operation to an LWP, and the LWP's ctl file descriptor might not be open. */ if (pi->ctl_fd == 0 && open_procinfo_files (pi, FD_CTL) == 0) return 0; else { procfs_ctl_t cmd[2]; cmd[0] = PCKILL; cmd[1] = signo; win = (write (pi->ctl_fd, (char *) &cmd, sizeof (cmd)) == sizeof (cmd)); } return win; } /* Find the pid of the process that started this one. Returns the parent process pid, or zero. */ static int proc_parent_pid (procinfo *pi) { /* We should never have to apply this operation to any procinfo except the one for the main process. If that ever changes for any reason, then take out the following clause and replace it with one that makes sure the ctl_fd is open. */ if (pi->tid != 0) pi = find_procinfo_or_die (pi->pid, 0); if (!pi->status_valid) if (!proc_get_status (pi)) return 0; return pi->prstatus.pr_ppid; } /* Convert a target address (a.k.a. CORE_ADDR) into a host address (a.k.a void pointer)! */ static void * procfs_address_to_host_pointer (CORE_ADDR addr) { struct type *ptr_type = builtin_type (target_gdbarch ())->builtin_data_ptr; void *ptr; gdb_assert (sizeof (ptr) == TYPE_LENGTH (ptr_type)); gdbarch_address_to_pointer (target_gdbarch (), ptr_type, (gdb_byte *) &ptr, addr); return ptr; } static int proc_set_watchpoint (procinfo *pi, CORE_ADDR addr, int len, int wflags) { struct { procfs_ctl_t cmd; char watch[sizeof (prwatch_t)]; } arg; prwatch_t pwatch; /* NOTE: cagney/2003-02-01: Even more horrible hack. Need to convert a target address into something that can be stored in a native data structure. */ pwatch.pr_vaddr = (uintptr_t) procfs_address_to_host_pointer (addr); pwatch.pr_size = len; pwatch.pr_wflags = wflags; arg.cmd = PCWATCH; memcpy (arg.watch, &pwatch, sizeof (prwatch_t)); return (write (pi->ctl_fd, &arg, sizeof (arg)) == sizeof (arg)); } #if (defined(__i386__) || defined(__x86_64__)) && defined (sun) #include /* The KEY is actually the value of the lower 16 bits of the GS register for the LWP that we're interested in. Returns the matching ssh struct (LDT entry). */ static struct ssd * proc_get_LDT_entry (procinfo *pi, int key) /* ARI: editCase function */ { static struct ssd *ldt_entry = NULL; char pathname[MAX_PROC_NAME_SIZE]; /* Allocate space for one LDT entry. This alloc must persist, because we return a pointer to it. */ if (ldt_entry == NULL) ldt_entry = XNEW (struct ssd); /* Open the file descriptor for the LDT table. */ xsnprintf (pathname, sizeof (pathname), "/proc/%d/ldt", pi->pid); scoped_fd fd (open_with_retry (pathname, O_RDONLY)); if (fd.get () < 0) { proc_warn (pi, "proc_get_LDT_entry (open)", __LINE__); return NULL; } /* Now 'read' thru the table, find a match and return it. */ while (read (fd.get (), ldt_entry, sizeof (struct ssd)) == sizeof (struct ssd)) { if (ldt_entry->sel == 0 && ldt_entry->bo == 0 && ldt_entry->acc1 == 0 && ldt_entry->acc2 == 0) break; /* end of table */ /* If key matches, return this entry. */ if (ldt_entry->sel == key) return ldt_entry; } /* Loop ended, match not found. */ return NULL; } /* Returns the pointer to the LDT entry of PTID. */ struct ssd * procfs_find_LDT_entry (ptid_t ptid) /* ARI: editCase function */ { gdb_gregset_t *gregs; int key; procinfo *pi; /* Find procinfo for the lwp. */ pi = find_procinfo (ptid.pid (), ptid.lwp ()); if (pi == NULL) { warning (_("procfs_find_LDT_entry: could not find procinfo for %d:%ld."), ptid.pid (), ptid.lwp ()); return NULL; } /* get its general registers. */ gregs = proc_get_gregs (pi); if (gregs == NULL) { warning (_("procfs_find_LDT_entry: could not read gregs for %d:%ld."), ptid.pid (), ptid.lwp ()); return NULL; } /* Now extract the GS register's lower 16 bits. */ key = (*gregs)[GS] & 0xffff; /* Find the matching entry and return it. */ return proc_get_LDT_entry (pi, key); } #endif /* =============== END, non-thread part of /proc "MODULE" =============== */ /* =================== Thread "MODULE" =================== */ /* NOTE: you'll see more ifdefs and duplication of functions here, since there is a different way to do threads on every OS. */ /* Returns the number of threads for the process. */ static int proc_get_nthreads (procinfo *pi) { if (!pi->status_valid) if (!proc_get_status (pi)) return 0; /* Only works for the process procinfo, because the LWP procinfos do not get prstatus filled in. */ if (pi->tid != 0) /* Find the parent process procinfo. */ pi = find_procinfo_or_die (pi->pid, 0); return pi->prstatus.pr_nlwp; } /* LWP version. Return the ID of the thread that had an event of interest. (ie. the one that hit a breakpoint or other traced event). All other things being equal, this should be the ID of a thread that is currently executing. */ static int proc_get_current_thread (procinfo *pi) { /* Note: this should be applied to the root procinfo for the process, not to the procinfo for an LWP. If applied to the procinfo for an LWP, it will simply return that LWP's ID. In that case, find the parent process procinfo. */ if (pi->tid != 0) pi = find_procinfo_or_die (pi->pid, 0); if (!pi->status_valid) if (!proc_get_status (pi)) return 0; return pi->prstatus.pr_lwp.pr_lwpid; } /* Discover the IDs of all the threads within the process, and create a procinfo for each of them (chained to the parent). This unfortunately requires a different method on every OS. Returns non-zero for success, zero for failure. */ static int proc_delete_dead_threads (procinfo *parent, procinfo *thread, void *ignore) { if (thread && parent) /* sanity */ { thread->status_valid = 0; if (!proc_get_status (thread)) destroy_one_procinfo (&parent->thread_list, thread); } return 0; /* keep iterating */ } static int proc_update_threads (procinfo *pi) { char pathname[MAX_PROC_NAME_SIZE + 16]; struct dirent *direntry; procinfo *thread; gdb_dir_up dirp; int lwpid; /* We should never have to apply this operation to any procinfo except the one for the main process. If that ever changes for any reason, then take out the following clause and replace it with one that makes sure the ctl_fd is open. */ if (pi->tid != 0) pi = find_procinfo_or_die (pi->pid, 0); proc_iterate_over_threads (pi, proc_delete_dead_threads, NULL); /* Note: this brute-force method was originally devised for Unixware (support removed since), and will also work on Solaris 2.6 and 2.7. The original comment mentioned the existence of a much simpler and more elegant way to do this on Solaris, but didn't point out what that was. */ strcpy (pathname, pi->pathname); strcat (pathname, "/lwp"); dirp.reset (opendir (pathname)); if (dirp == NULL) proc_error (pi, "update_threads, opendir", __LINE__); while ((direntry = readdir (dirp.get ())) != NULL) if (direntry->d_name[0] != '.') /* skip '.' and '..' */ { lwpid = atoi (&direntry->d_name[0]); thread = create_procinfo (pi->pid, lwpid); if (thread == NULL) proc_error (pi, "update_threads, create_procinfo", __LINE__); } pi->threads_valid = 1; return 1; } /* Given a pointer to a function, call that function once for each lwp in the procinfo list, until the function returns non-zero, in which event return the value returned by the function. Note: this function does NOT call update_threads. If you want to discover new threads first, you must call that function explicitly. This function just makes a quick pass over the currently-known procinfos. PI is the parent process procinfo. FUNC is the per-thread function. PTR is an opaque parameter for function. Returns the first non-zero return value from the callee, or zero. */ static int proc_iterate_over_threads (procinfo *pi, int (*func) (procinfo *, procinfo *, void *), void *ptr) { procinfo *thread, *next; int retval = 0; /* We should never have to apply this operation to any procinfo except the one for the main process. If that ever changes for any reason, then take out the following clause and replace it with one that makes sure the ctl_fd is open. */ if (pi->tid != 0) pi = find_procinfo_or_die (pi->pid, 0); for (thread = pi->thread_list; thread != NULL; thread = next) { next = thread->next; /* In case thread is destroyed. */ retval = (*func) (pi, thread, ptr); if (retval != 0) break; } return retval; } /* =================== END, Thread "MODULE" =================== */ /* =================== END, /proc "MODULE" =================== */ /* =================== GDB "MODULE" =================== */ /* Here are all of the gdb target vector functions and their friends. */ static ptid_t do_attach (ptid_t ptid); static void do_detach (); static void proc_trace_syscalls_1 (procinfo *pi, int syscallnum, int entry_or_exit, int mode, int from_tty); /* Sets up the inferior to be debugged. Registers to trace signals, hardware faults, and syscalls. Note: does not set RLC flag: caller may want to customize that. Returns zero for success (note! unlike most functions in this module); on failure, returns the LINE NUMBER where it failed! */ static int procfs_debug_inferior (procinfo *pi) { fltset_t traced_faults; sigset_t traced_signals; sysset_t *traced_syscall_entries; sysset_t *traced_syscall_exits; int status; /* Register to trace hardware faults in the child. */ prfillset (&traced_faults); /* trace all faults... */ prdelset (&traced_faults, FLTPAGE); /* except page fault. */ if (!proc_set_traced_faults (pi, &traced_faults)) return __LINE__; /* Initially, register to trace all signals in the child. */ prfillset (&traced_signals); if (!proc_set_traced_signals (pi, &traced_signals)) return __LINE__; /* Register to trace the 'exit' system call (on entry). */ traced_syscall_entries = XNEW (sysset_t); premptyset (traced_syscall_entries); praddset (traced_syscall_entries, SYS_exit); praddset (traced_syscall_entries, SYS_lwp_exit); status = proc_set_traced_sysentry (pi, traced_syscall_entries); xfree (traced_syscall_entries); if (!status) return __LINE__; /* Method for tracing exec syscalls. */ /* GW: Rationale... Not all systems with /proc have all the exec* syscalls with the same names. On the SGI, for example, there is no SYS_exec, but there *is* a SYS_execv. So, we try to account for that. */ traced_syscall_exits = XNEW (sysset_t); premptyset (traced_syscall_exits); #ifdef SYS_exec praddset (traced_syscall_exits, SYS_exec); #endif praddset (traced_syscall_exits, SYS_execve); praddset (traced_syscall_exits, SYS_lwp_create); praddset (traced_syscall_exits, SYS_lwp_exit); status = proc_set_traced_sysexit (pi, traced_syscall_exits); xfree (traced_syscall_exits); if (!status) return __LINE__; return 0; } void procfs_target::attach (const char *args, int from_tty) { int pid; pid = parse_pid_to_attach (args); if (pid == getpid ()) error (_("Attaching GDB to itself is not a good idea...")); if (from_tty) { const char *exec_file = get_exec_file (0); if (exec_file) printf_filtered (_("Attaching to program `%s', %s\n"), exec_file, target_pid_to_str (ptid_t (pid)).c_str ()); else printf_filtered (_("Attaching to %s\n"), target_pid_to_str (ptid_t (pid)).c_str ()); fflush (stdout); } inferior_ptid = do_attach (ptid_t (pid)); if (!target_is_pushed (this)) push_target (this); } void procfs_target::detach (inferior *inf, int from_tty) { int pid = inferior_ptid.pid (); if (from_tty) { const char *exec_file; exec_file = get_exec_file (0); if (exec_file == NULL) exec_file = ""; printf_filtered (_("Detaching from program: %s, %s\n"), exec_file, target_pid_to_str (ptid_t (pid)).c_str ()); } do_detach (); inferior_ptid = null_ptid; detach_inferior (inf); maybe_unpush_target (); } static ptid_t do_attach (ptid_t ptid) { procinfo *pi; struct inferior *inf; int fail; int lwpid; pi = create_procinfo (ptid.pid (), 0); if (pi == NULL) perror (_("procfs: out of memory in 'attach'")); if (!open_procinfo_files (pi, FD_CTL)) { fprintf_filtered (gdb_stderr, "procfs:%d -- ", __LINE__); xsnprintf (errmsg, sizeof (errmsg), "do_attach: couldn't open /proc file for process %d", ptid.pid ()); dead_procinfo (pi, errmsg, NOKILL); } /* Stop the process (if it isn't already stopped). */ if (proc_flags (pi) & (PR_STOPPED | PR_ISTOP)) { pi->was_stopped = 1; proc_prettyprint_why (proc_why (pi), proc_what (pi), 1); } else { pi->was_stopped = 0; /* Set the process to run again when we close it. */ if (!proc_set_run_on_last_close (pi)) dead_procinfo (pi, "do_attach: couldn't set RLC.", NOKILL); /* Now stop the process. */ if (!proc_stop_process (pi)) dead_procinfo (pi, "do_attach: couldn't stop the process.", NOKILL); pi->ignore_next_sigstop = 1; } /* Save some of the /proc state to be restored if we detach. */ if (!proc_get_traced_faults (pi, &pi->saved_fltset)) dead_procinfo (pi, "do_attach: couldn't save traced faults.", NOKILL); if (!proc_get_traced_signals (pi, &pi->saved_sigset)) dead_procinfo (pi, "do_attach: couldn't save traced signals.", NOKILL); if (!proc_get_traced_sysentry (pi, pi->saved_entryset)) dead_procinfo (pi, "do_attach: couldn't save traced syscall entries.", NOKILL); if (!proc_get_traced_sysexit (pi, pi->saved_exitset)) dead_procinfo (pi, "do_attach: couldn't save traced syscall exits.", NOKILL); if (!proc_get_held_signals (pi, &pi->saved_sighold)) dead_procinfo (pi, "do_attach: couldn't save held signals.", NOKILL); fail = procfs_debug_inferior (pi); if (fail != 0) dead_procinfo (pi, "do_attach: failed in procfs_debug_inferior", NOKILL); inf = current_inferior (); inferior_appeared (inf, pi->pid); /* Let GDB know that the inferior was attached. */ inf->attach_flag = 1; /* Create a procinfo for the current lwp. */ lwpid = proc_get_current_thread (pi); create_procinfo (pi->pid, lwpid); /* Add it to gdb's thread list. */ ptid = ptid_t (pi->pid, lwpid, 0); add_thread (ptid); return ptid; } static void do_detach () { procinfo *pi; /* Find procinfo for the main process. */ pi = find_procinfo_or_die (inferior_ptid.pid (), 0); /* FIXME: threads */ if (!proc_set_traced_signals (pi, &pi->saved_sigset)) proc_warn (pi, "do_detach, set_traced_signal", __LINE__); if (!proc_set_traced_faults (pi, &pi->saved_fltset)) proc_warn (pi, "do_detach, set_traced_faults", __LINE__); if (!proc_set_traced_sysentry (pi, pi->saved_entryset)) proc_warn (pi, "do_detach, set_traced_sysentry", __LINE__); if (!proc_set_traced_sysexit (pi, pi->saved_exitset)) proc_warn (pi, "do_detach, set_traced_sysexit", __LINE__); if (!proc_set_held_signals (pi, &pi->saved_sighold)) proc_warn (pi, "do_detach, set_held_signals", __LINE__); if (proc_flags (pi) & (PR_STOPPED | PR_ISTOP)) if (!(pi->was_stopped) || query (_("Was stopped when attached, make it runnable again? "))) { /* Clear any pending signal. */ if (!proc_clear_current_fault (pi)) proc_warn (pi, "do_detach, clear_current_fault", __LINE__); if (!proc_clear_current_signal (pi)) proc_warn (pi, "do_detach, clear_current_signal", __LINE__); if (!proc_set_run_on_last_close (pi)) proc_warn (pi, "do_detach, set_rlc", __LINE__); } destroy_procinfo (pi); } /* Fetch register REGNUM from the inferior. If REGNUM is -1, do this for all registers. ??? Is the following note still relevant? We can't get individual registers with the PT_GETREGS ptrace(2) request either, yet we don't bother with caching at all in that case. NOTE: Since the /proc interface cannot give us individual registers, we pay no attention to REGNUM, and just fetch them all. This results in the possibility that we will do unnecessarily many fetches, since we may be called repeatedly for individual registers. So we cache the results, and mark the cache invalid when the process is resumed. */ void procfs_target::fetch_registers (struct regcache *regcache, int regnum) { gdb_gregset_t *gregs; procinfo *pi; ptid_t ptid = regcache->ptid (); int pid = ptid.pid (); int tid = ptid.lwp (); struct gdbarch *gdbarch = regcache->arch (); pi = find_procinfo_or_die (pid, tid); if (pi == NULL) error (_("procfs: fetch_registers failed to find procinfo for %s"), target_pid_to_str (ptid).c_str ()); gregs = proc_get_gregs (pi); if (gregs == NULL) proc_error (pi, "fetch_registers, get_gregs", __LINE__); supply_gregset (regcache, (const gdb_gregset_t *) gregs); if (gdbarch_fp0_regnum (gdbarch) >= 0) /* Do we have an FPU? */ { gdb_fpregset_t *fpregs; if ((regnum >= 0 && regnum < gdbarch_fp0_regnum (gdbarch)) || regnum == gdbarch_pc_regnum (gdbarch) || regnum == gdbarch_sp_regnum (gdbarch)) return; /* Not a floating point register. */ fpregs = proc_get_fpregs (pi); if (fpregs == NULL) proc_error (pi, "fetch_registers, get_fpregs", __LINE__); supply_fpregset (regcache, (const gdb_fpregset_t *) fpregs); } } /* Store register REGNUM back into the inferior. If REGNUM is -1, do this for all registers. NOTE: Since the /proc interface will not read individual registers, we will cache these requests until the process is resumed, and only then write them back to the inferior process. FIXME: is that a really bad idea? Have to think about cases where writing one register might affect the value of others, etc. */ void procfs_target::store_registers (struct regcache *regcache, int regnum) { gdb_gregset_t *gregs; procinfo *pi; ptid_t ptid = regcache->ptid (); int pid = ptid.pid (); int tid = ptid.lwp (); struct gdbarch *gdbarch = regcache->arch (); pi = find_procinfo_or_die (pid, tid); if (pi == NULL) error (_("procfs: store_registers: failed to find procinfo for %s"), target_pid_to_str (ptid).c_str ()); gregs = proc_get_gregs (pi); if (gregs == NULL) proc_error (pi, "store_registers, get_gregs", __LINE__); fill_gregset (regcache, gregs, regnum); if (!proc_set_gregs (pi)) proc_error (pi, "store_registers, set_gregs", __LINE__); if (gdbarch_fp0_regnum (gdbarch) >= 0) /* Do we have an FPU? */ { gdb_fpregset_t *fpregs; if ((regnum >= 0 && regnum < gdbarch_fp0_regnum (gdbarch)) || regnum == gdbarch_pc_regnum (gdbarch) || regnum == gdbarch_sp_regnum (gdbarch)) return; /* Not a floating point register. */ fpregs = proc_get_fpregs (pi); if (fpregs == NULL) proc_error (pi, "store_registers, get_fpregs", __LINE__); fill_fpregset (regcache, fpregs, regnum); if (!proc_set_fpregs (pi)) proc_error (pi, "store_registers, set_fpregs", __LINE__); } } static int syscall_is_lwp_exit (procinfo *pi, int scall) { if (scall == SYS_lwp_exit) return 1; return 0; } static int syscall_is_exit (procinfo *pi, int scall) { if (scall == SYS_exit) return 1; return 0; } static int syscall_is_exec (procinfo *pi, int scall) { #ifdef SYS_exec if (scall == SYS_exec) return 1; #endif if (scall == SYS_execve) return 1; return 0; } static int syscall_is_lwp_create (procinfo *pi, int scall) { if (scall == SYS_lwp_create) return 1; return 0; } /* Retrieve the next stop event from the child process. If child has not stopped yet, wait for it to stop. Translate /proc eventcodes (or possibly wait eventcodes) into gdb internal event codes. Returns the id of process (and possibly thread) that incurred the event. Event codes are returned through a pointer parameter. */ ptid_t procfs_target::wait (ptid_t ptid, struct target_waitstatus *status, int options) { /* First cut: loosely based on original version 2.1. */ procinfo *pi; int wstat; int temp_tid; ptid_t retval, temp_ptid; int why, what, flags; int retry = 0; wait_again: retry++; wstat = 0; retval = ptid_t (-1); /* Find procinfo for main process. */ pi = find_procinfo_or_die (inferior_ptid.pid (), 0); if (pi) { /* We must assume that the status is stale now... */ pi->status_valid = 0; pi->gregs_valid = 0; pi->fpregs_valid = 0; #if 0 /* just try this out... */ flags = proc_flags (pi); why = proc_why (pi); if ((flags & PR_STOPPED) && (why == PR_REQUESTED)) pi->status_valid = 0; /* re-read again, IMMEDIATELY... */ #endif /* If child is not stopped, wait for it to stop. */ if (!(proc_flags (pi) & (PR_STOPPED | PR_ISTOP)) && !proc_wait_for_stop (pi)) { /* wait_for_stop failed: has the child terminated? */ if (errno == ENOENT) { int wait_retval; /* /proc file not found; presumably child has terminated. */ wait_retval = ::wait (&wstat); /* "wait" for the child's exit. */ /* Wrong child? */ if (wait_retval != inferior_ptid.pid ()) error (_("procfs: couldn't stop " "process %d: wait returned %d."), inferior_ptid.pid (), wait_retval); /* FIXME: might I not just use waitpid? Or try find_procinfo to see if I know about this child? */ retval = ptid_t (wait_retval); } else if (errno == EINTR) goto wait_again; else { /* Unknown error from wait_for_stop. */ proc_error (pi, "target_wait (wait_for_stop)", __LINE__); } } else { /* This long block is reached if either: a) the child was already stopped, or b) we successfully waited for the child with wait_for_stop. This block will analyze the /proc status, and translate it into a waitstatus for GDB. If we actually had to call wait because the /proc file is gone (child terminated), then we skip this block, because we already have a waitstatus. */ flags = proc_flags (pi); why = proc_why (pi); what = proc_what (pi); if (flags & (PR_STOPPED | PR_ISTOP)) { /* If it's running async (for single_thread control), set it back to normal again. */ if (flags & PR_ASYNC) if (!proc_unset_async (pi)) proc_error (pi, "target_wait, unset_async", __LINE__); if (info_verbose) proc_prettyprint_why (why, what, 1); /* The 'pid' we will return to GDB is composed of the process ID plus the lwp ID. */ retval = ptid_t (pi->pid, proc_get_current_thread (pi), 0); switch (why) { case PR_SIGNALLED: wstat = (what << 8) | 0177; break; case PR_SYSENTRY: if (syscall_is_lwp_exit (pi, what)) { if (print_thread_events) printf_unfiltered (_("[%s exited]\n"), target_pid_to_str (retval).c_str ()); delete_thread (find_thread_ptid (retval)); status->kind = TARGET_WAITKIND_SPURIOUS; return retval; } else if (syscall_is_exit (pi, what)) { struct inferior *inf; /* Handle SYS_exit call only. */ /* Stopped at entry to SYS_exit. Make it runnable, resume it, then use the wait system call to get its exit code. Proc_run_process always clears the current fault and signal. Then return its exit status. */ pi->status_valid = 0; wstat = 0; /* FIXME: what we should do is return TARGET_WAITKIND_SPURIOUS. */ if (!proc_run_process (pi, 0, 0)) proc_error (pi, "target_wait, run_process", __LINE__); inf = find_inferior_pid (pi->pid); if (inf->attach_flag) { /* Don't call wait: simulate waiting for exit, return a "success" exit code. Bogus: what if it returns something else? */ wstat = 0; retval = inferior_ptid; /* ? ? ? */ } else { int temp = ::wait (&wstat); /* FIXME: shouldn't I make sure I get the right event from the right process? If (for instance) I have killed an earlier inferior process but failed to clean up after it somehow, I could get its termination event here. */ /* If wait returns -1, that's what we return to GDB. */ if (temp < 0) retval = ptid_t (temp); } } else { printf_filtered (_("procfs: trapped on entry to ")); proc_prettyprint_syscall (proc_what (pi), 0); printf_filtered ("\n"); long i, nsysargs, *sysargs; nsysargs = proc_nsysarg (pi); sysargs = proc_sysargs (pi); if (nsysargs > 0 && sysargs != NULL) { printf_filtered (_("%ld syscall arguments:\n"), nsysargs); for (i = 0; i < nsysargs; i++) printf_filtered ("#%ld: 0x%08lx\n", i, sysargs[i]); } if (status) { /* How to exit gracefully, returning "unknown event". */ status->kind = TARGET_WAITKIND_SPURIOUS; return inferior_ptid; } else { /* How to keep going without returning to wfi: */ target_continue_no_signal (ptid); goto wait_again; } } break; case PR_SYSEXIT: if (syscall_is_exec (pi, what)) { /* Hopefully this is our own "fork-child" execing the real child. Hoax this event into a trap, and GDB will see the child about to execute its start address. */ wstat = (SIGTRAP << 8) | 0177; } else if (syscall_is_lwp_create (pi, what)) { /* This syscall is somewhat like fork/exec. We will get the event twice: once for the parent LWP, and once for the child. We should already know about the parent LWP, but the child will be new to us. So, whenever we get this event, if it represents a new thread, simply add the thread to the list. */ /* If not in procinfo list, add it. */ temp_tid = proc_get_current_thread (pi); if (!find_procinfo (pi->pid, temp_tid)) create_procinfo (pi->pid, temp_tid); temp_ptid = ptid_t (pi->pid, temp_tid, 0); /* If not in GDB's thread list, add it. */ if (!in_thread_list (temp_ptid)) add_thread (temp_ptid); /* Return to WFI, but tell it to immediately resume. */ status->kind = TARGET_WAITKIND_SPURIOUS; return inferior_ptid; } else if (syscall_is_lwp_exit (pi, what)) { if (print_thread_events) printf_unfiltered (_("[%s exited]\n"), target_pid_to_str (retval).c_str ()); delete_thread (find_thread_ptid (retval)); status->kind = TARGET_WAITKIND_SPURIOUS; return retval; } else if (0) { /* FIXME: Do we need to handle SYS_sproc, SYS_fork, or SYS_vfork here? The old procfs seemed to use this event to handle threads on older (non-LWP) systems, where I'm assuming that threads were actually separate processes. Irix, maybe? Anyway, low priority for now. */ } else { printf_filtered (_("procfs: trapped on exit from ")); proc_prettyprint_syscall (proc_what (pi), 0); printf_filtered ("\n"); long i, nsysargs, *sysargs; nsysargs = proc_nsysarg (pi); sysargs = proc_sysargs (pi); if (nsysargs > 0 && sysargs != NULL) { printf_filtered (_("%ld syscall arguments:\n"), nsysargs); for (i = 0; i < nsysargs; i++) printf_filtered ("#%ld: 0x%08lx\n", i, sysargs[i]); } status->kind = TARGET_WAITKIND_SPURIOUS; return inferior_ptid; } break; case PR_REQUESTED: #if 0 /* FIXME */ wstat = (SIGSTOP << 8) | 0177; break; #else if (retry < 5) { printf_filtered (_("Retry #%d:\n"), retry); pi->status_valid = 0; goto wait_again; } else { /* If not in procinfo list, add it. */ temp_tid = proc_get_current_thread (pi); if (!find_procinfo (pi->pid, temp_tid)) create_procinfo (pi->pid, temp_tid); /* If not in GDB's thread list, add it. */ temp_ptid = ptid_t (pi->pid, temp_tid, 0); if (!in_thread_list (temp_ptid)) add_thread (temp_ptid); status->kind = TARGET_WAITKIND_STOPPED; status->value.sig = GDB_SIGNAL_0; return retval; } #endif case PR_JOBCONTROL: wstat = (what << 8) | 0177; break; case PR_FAULTED: { int signo = pi->prstatus.pr_lwp.pr_info.si_signo; if (signo != 0) wstat = (signo << 8) | 0177; } break; default: /* switch (why) unmatched */ printf_filtered ("procfs:%d -- ", __LINE__); printf_filtered (_("child stopped for unknown reason:\n")); proc_prettyprint_why (why, what, 1); error (_("... giving up...")); break; } /* Got this far without error: If retval isn't in the threads database, add it. */ if (retval.pid () > 0 && retval != inferior_ptid && !in_thread_list (retval)) { /* We have a new thread. We need to add it both to GDB's list and to our own. If we don't create a procinfo, resume may be unhappy later. */ add_thread (retval); if (find_procinfo (retval.pid (), retval.lwp ()) == NULL) create_procinfo (retval.pid (), retval.lwp ()); } } else /* Flags do not indicate STOPPED. */ { /* surely this can't happen... */ printf_filtered ("procfs:%d -- process not stopped.\n", __LINE__); proc_prettyprint_flags (flags, 1); error (_("procfs: ...giving up...")); } } if (status) store_waitstatus (status, wstat); } return retval; } /* Perform a partial transfer to/from the specified object. For memory transfers, fall back to the old memory xfer functions. */ enum target_xfer_status procfs_target::xfer_partial (enum target_object object, const char *annex, gdb_byte *readbuf, const gdb_byte *writebuf, ULONGEST offset, ULONGEST len, ULONGEST *xfered_len) { switch (object) { case TARGET_OBJECT_MEMORY: return procfs_xfer_memory (readbuf, writebuf, offset, len, xfered_len); case TARGET_OBJECT_AUXV: return memory_xfer_auxv (this, object, annex, readbuf, writebuf, offset, len, xfered_len); default: return this->beneath ()->xfer_partial (object, annex, readbuf, writebuf, offset, len, xfered_len); } } /* Helper for procfs_xfer_partial that handles memory transfers. Arguments are like target_xfer_partial. */ static enum target_xfer_status procfs_xfer_memory (gdb_byte *readbuf, const gdb_byte *writebuf, ULONGEST memaddr, ULONGEST len, ULONGEST *xfered_len) { procinfo *pi; int nbytes; /* Find procinfo for main process. */ pi = find_procinfo_or_die (inferior_ptid.pid (), 0); if (pi->as_fd == 0 && open_procinfo_files (pi, FD_AS) == 0) { proc_warn (pi, "xfer_memory, open_proc_files", __LINE__); return TARGET_XFER_E_IO; } if (lseek (pi->as_fd, (off_t) memaddr, SEEK_SET) != (off_t) memaddr) return TARGET_XFER_E_IO; if (writebuf != NULL) { PROCFS_NOTE ("write memory:\n"); nbytes = write (pi->as_fd, writebuf, len); } else { PROCFS_NOTE ("read memory:\n"); nbytes = read (pi->as_fd, readbuf, len); } if (nbytes <= 0) return TARGET_XFER_E_IO; *xfered_len = nbytes; return TARGET_XFER_OK; } /* Called by target_resume before making child runnable. Mark cached registers and status's invalid. If there are "dirty" caches that need to be written back to the child process, do that. File descriptors are also cached. As they are a limited resource, we cannot hold onto them indefinitely. However, as they are expensive to open, we don't want to throw them away indiscriminately either. As a compromise, we will keep the file descriptors for the parent process, but discard any file descriptors we may have accumulated for the threads. As this function is called by iterate_over_threads, it always returns zero (so that iterate_over_threads will keep iterating). */ static int invalidate_cache (procinfo *parent, procinfo *pi, void *ptr) { /* About to run the child; invalidate caches and do any other cleanup. */ #if 0 if (pi->gregs_dirty) if (parent == NULL || proc_get_current_thread (parent) != pi->tid) if (!proc_set_gregs (pi)) /* flush gregs cache */ proc_warn (pi, "target_resume, set_gregs", __LINE__); if (gdbarch_fp0_regnum (target_gdbarch ()) >= 0) if (pi->fpregs_dirty) if (parent == NULL || proc_get_current_thread (parent) != pi->tid) if (!proc_set_fpregs (pi)) /* flush fpregs cache */ proc_warn (pi, "target_resume, set_fpregs", __LINE__); #endif if (parent != NULL) { /* The presence of a parent indicates that this is an LWP. Close any file descriptors that it might have open. We don't do this to the master (parent) procinfo. */ close_procinfo_files (pi); } pi->gregs_valid = 0; pi->fpregs_valid = 0; #if 0 pi->gregs_dirty = 0; pi->fpregs_dirty = 0; #endif pi->status_valid = 0; pi->threads_valid = 0; return 0; } #if 0 /* A callback function for iterate_over_threads. Find the asynchronous signal thread, and make it runnable. See if that helps matters any. */ static int make_signal_thread_runnable (procinfo *process, procinfo *pi, void *ptr) { #ifdef PR_ASLWP if (proc_flags (pi) & PR_ASLWP) { if (!proc_run_process (pi, 0, -1)) proc_error (pi, "make_signal_thread_runnable", __LINE__); return 1; } #endif return 0; } #endif /* Make the child process runnable. Normally we will then call procfs_wait and wait for it to stop again (unless gdb is async). If STEP is true, then arrange for the child to stop again after executing a single instruction. If SIGNO is zero, then cancel any pending signal; if non-zero, then arrange for the indicated signal to be delivered to the child when it runs. If PID is -1, then allow any child thread to run; if non-zero, then allow only the indicated thread to run. (not implemented yet). */ void procfs_target::resume (ptid_t ptid, int step, enum gdb_signal signo) { procinfo *pi, *thread; int native_signo; /* 2.1: prrun.prflags |= PRSVADDR; prrun.pr_vaddr = $PC; set resume address prrun.prflags |= PRSTRACE; trace signals in pr_trace (all) prrun.prflags |= PRSFAULT; trace faults in pr_fault (all but PAGE) prrun.prflags |= PRCFAULT; clear current fault. PRSTRACE and PRSFAULT can be done by other means (proc_trace_signals, proc_trace_faults) PRSVADDR is unnecessary. PRCFAULT may be replaced by a PIOCCFAULT call (proc_clear_current_fault) This basically leaves PRSTEP and PRCSIG. PRCSIG is like PIOCSSIG (proc_clear_current_signal). So basically PR_STEP is the sole argument that must be passed to proc_run_process (for use in the prrun struct by ioctl). */ /* Find procinfo for main process. */ pi = find_procinfo_or_die (inferior_ptid.pid (), 0); /* First cut: ignore pid argument. */ errno = 0; /* Convert signal to host numbering. */ if (signo == 0 || (signo == GDB_SIGNAL_STOP && pi->ignore_next_sigstop)) native_signo = 0; else native_signo = gdb_signal_to_host (signo); pi->ignore_next_sigstop = 0; /* Running the process voids all cached registers and status. */ /* Void the threads' caches first. */ proc_iterate_over_threads (pi, invalidate_cache, NULL); /* Void the process procinfo's caches. */ invalidate_cache (NULL, pi, NULL); if (ptid.pid () != -1) { /* Resume a specific thread, presumably suppressing the others. */ thread = find_procinfo (ptid.pid (), ptid.lwp ()); if (thread != NULL) { if (thread->tid != 0) { /* We're to resume a specific thread, and not the others. Set the child process's PR_ASYNC flag. */ if (!proc_set_async (pi)) proc_error (pi, "target_resume, set_async", __LINE__); #if 0 proc_iterate_over_threads (pi, make_signal_thread_runnable, NULL); #endif pi = thread; /* Substitute the thread's procinfo for run. */ } } } if (!proc_run_process (pi, step, native_signo)) { if (errno == EBUSY) warning (_("resume: target already running. " "Pretend to resume, and hope for the best!")); else proc_error (pi, "target_resume", __LINE__); } } /* Set up to trace signals in the child process. */ void procfs_target::pass_signals (gdb::array_view pass_signals) { sigset_t signals; procinfo *pi = find_procinfo_or_die (inferior_ptid.pid (), 0); int signo; prfillset (&signals); for (signo = 0; signo < NSIG; signo++) { int target_signo = gdb_signal_from_host (signo); if (target_signo < pass_signals.size () && pass_signals[target_signo]) prdelset (&signals, signo); } if (!proc_set_traced_signals (pi, &signals)) proc_error (pi, "pass_signals", __LINE__); } /* Print status information about the child process. */ void procfs_target::files_info () { struct inferior *inf = current_inferior (); printf_filtered (_("\tUsing the running image of %s %s via /proc.\n"), inf->attach_flag? "attached": "child", target_pid_to_str (inferior_ptid).c_str ()); } /* Make it die. Wait for it to die. Clean up after it. Note: this should only be applied to the real process, not to an LWP, because of the check for parent-process. If we need this to work for an LWP, it needs some more logic. */ static void unconditionally_kill_inferior (procinfo *pi) { int parent_pid; parent_pid = proc_parent_pid (pi); if (!proc_kill (pi, SIGKILL)) proc_error (pi, "unconditionally_kill, proc_kill", __LINE__); destroy_procinfo (pi); /* If pi is GDB's child, wait for it to die. */ if (parent_pid == getpid ()) /* FIXME: should we use waitpid to make sure we get the right event? Should we check the returned event? */ { #if 0 int status, ret; ret = waitpid (pi->pid, &status, 0); #else wait (NULL); #endif } } /* We're done debugging it, and we want it to go away. Then we want GDB to forget all about it. */ void procfs_target::kill () { if (inferior_ptid != null_ptid) /* ? */ { /* Find procinfo for main process. */ procinfo *pi = find_procinfo (inferior_ptid.pid (), 0); if (pi) unconditionally_kill_inferior (pi); target_mourn_inferior (inferior_ptid); } } /* Forget we ever debugged this thing! */ void procfs_target::mourn_inferior () { procinfo *pi; if (inferior_ptid != null_ptid) { /* Find procinfo for main process. */ pi = find_procinfo (inferior_ptid.pid (), 0); if (pi) destroy_procinfo (pi); } generic_mourn_inferior (); maybe_unpush_target (); } /* When GDB forks to create a runnable inferior process, this function is called on the parent side of the fork. It's job is to do whatever is necessary to make the child ready to be debugged, and then wait for the child to synchronize. */ static void procfs_init_inferior (struct target_ops *ops, int pid) { procinfo *pi; int fail; int lwpid; /* This routine called on the parent side (GDB side) after GDB forks the inferior. */ if (!target_is_pushed (ops)) push_target (ops); pi = create_procinfo (pid, 0); if (pi == NULL) perror (_("procfs: out of memory in 'init_inferior'")); if (!open_procinfo_files (pi, FD_CTL)) proc_error (pi, "init_inferior, open_proc_files", __LINE__); /* xmalloc // done open_procinfo_files // done link list // done prfillset (trace) procfs_notice_signals prfillset (fault) prdelset (FLTPAGE) PIOCWSTOP PIOCSFAULT */ /* If not stopped yet, wait for it to stop. */ if (!(proc_flags (pi) & PR_STOPPED) && !(proc_wait_for_stop (pi))) dead_procinfo (pi, "init_inferior: wait_for_stop failed", KILL); /* Save some of the /proc state to be restored if we detach. */ /* FIXME: Why? In case another debugger was debugging it? We're it's parent, for Ghu's sake! */ if (!proc_get_traced_signals (pi, &pi->saved_sigset)) proc_error (pi, "init_inferior, get_traced_signals", __LINE__); if (!proc_get_held_signals (pi, &pi->saved_sighold)) proc_error (pi, "init_inferior, get_held_signals", __LINE__); if (!proc_get_traced_faults (pi, &pi->saved_fltset)) proc_error (pi, "init_inferior, get_traced_faults", __LINE__); if (!proc_get_traced_sysentry (pi, pi->saved_entryset)) proc_error (pi, "init_inferior, get_traced_sysentry", __LINE__); if (!proc_get_traced_sysexit (pi, pi->saved_exitset)) proc_error (pi, "init_inferior, get_traced_sysexit", __LINE__); fail = procfs_debug_inferior (pi); if (fail != 0) proc_error (pi, "init_inferior (procfs_debug_inferior)", fail); /* FIXME: logically, we should really be turning OFF run-on-last-close, and possibly even turning ON kill-on-last-close at this point. But I can't make that change without careful testing which I don't have time to do right now... */ /* Turn on run-on-last-close flag so that the child will die if GDB goes away for some reason. */ if (!proc_set_run_on_last_close (pi)) proc_error (pi, "init_inferior, set_RLC", __LINE__); /* We now have have access to the lwpid of the main thread/lwp. */ lwpid = proc_get_current_thread (pi); /* Create a procinfo for the main lwp. */ create_procinfo (pid, lwpid); /* We already have a main thread registered in the thread table at this point, but it didn't have any lwp info yet. Notify the core about it. This changes inferior_ptid as well. */ thread_change_ptid (ptid_t (pid), ptid_t (pid, lwpid, 0)); gdb_startup_inferior (pid, START_INFERIOR_TRAPS_EXPECTED); } /* When GDB forks to create a new process, this function is called on the child side of the fork before GDB exec's the user program. Its job is to make the child minimally debuggable, so that the parent GDB process can connect to the child and take over. This function should do only the minimum to make that possible, and to synchronize with the parent process. The parent process should take care of the details. */ static void procfs_set_exec_trap (void) { /* This routine called on the child side (inferior side) after GDB forks the inferior. It must use only local variables, because it may be sharing data space with its parent. */ procinfo *pi; sysset_t *exitset; pi = create_procinfo (getpid (), 0); if (pi == NULL) perror_with_name (_("procfs: create_procinfo failed in child.")); if (open_procinfo_files (pi, FD_CTL) == 0) { proc_warn (pi, "set_exec_trap, open_proc_files", __LINE__); gdb_flush (gdb_stderr); /* No need to call "dead_procinfo", because we're going to exit. */ _exit (127); } /* Method for tracing exec syscalls. */ /* GW: Rationale... Not all systems with /proc have all the exec* syscalls with the same names. On the SGI, for example, there is no SYS_exec, but there *is* a SYS_execv. So, we try to account for that. */ exitset = XNEW (sysset_t); premptyset (exitset); #ifdef SYS_exec praddset (exitset, SYS_exec); #endif praddset (exitset, SYS_execve); if (!proc_set_traced_sysexit (pi, exitset)) { proc_warn (pi, "set_exec_trap, set_traced_sysexit", __LINE__); gdb_flush (gdb_stderr); _exit (127); } /* FIXME: should this be done in the parent instead? */ /* Turn off inherit on fork flag so that all grand-children of gdb start with tracing flags cleared. */ if (!proc_unset_inherit_on_fork (pi)) proc_warn (pi, "set_exec_trap, unset_inherit", __LINE__); /* Turn off run on last close flag, so that the child process cannot run away just because we close our handle on it. We want it to wait for the parent to attach. */ if (!proc_unset_run_on_last_close (pi)) proc_warn (pi, "set_exec_trap, unset_RLC", __LINE__); /* FIXME: No need to destroy the procinfo -- we have our own address space, and we're about to do an exec! */ /*destroy_procinfo (pi);*/ } /* This function is called BEFORE gdb forks the inferior process. Its only real responsibility is to set things up for the fork, and tell GDB which two functions to call after the fork (one for the parent, and one for the child). This function does a complicated search for a unix shell program, which it then uses to parse arguments and environment variables to be sent to the child. I wonder whether this code could not be abstracted out and shared with other unix targets such as inf-ptrace? */ void procfs_target::create_inferior (const char *exec_file, const std::string &allargs, char **env, int from_tty) { const char *shell_file = get_shell (); char *tryname; int pid; if (strchr (shell_file, '/') == NULL) { /* We will be looking down the PATH to find shell_file. If we just do this the normal way (via execlp, which operates by attempting an exec for each element of the PATH until it finds one which succeeds), then there will be an exec for each failed attempt, each of which will cause a PR_SYSEXIT stop, and we won't know how to distinguish the PR_SYSEXIT's for these failed execs with the ones for successful execs (whether the exec has succeeded is stored at that time in the carry bit or some such architecture-specific and non-ABI-specified place). So I can't think of anything better than to search the PATH now. This has several disadvantages: (1) There is a race condition; if we find a file now and it is deleted before we exec it, we lose, even if the deletion leaves a valid file further down in the PATH, (2) there is no way to know exactly what an executable (in the sense of "capable of being exec'd") file is. Using access() loses because it may lose if the caller is the superuser; failing to use it loses if there are ACLs or some such. */ const char *p; const char *p1; /* FIXME-maybe: might want "set path" command so user can change what path is used from within GDB. */ const char *path = getenv ("PATH"); int len; struct stat statbuf; if (path == NULL) path = "/bin:/usr/bin"; tryname = (char *) alloca (strlen (path) + strlen (shell_file) + 2); for (p = path; p != NULL; p = p1 ? p1 + 1: NULL) { p1 = strchr (p, ':'); if (p1 != NULL) len = p1 - p; else len = strlen (p); strncpy (tryname, p, len); tryname[len] = '\0'; strcat (tryname, "/"); strcat (tryname, shell_file); if (access (tryname, X_OK) < 0) continue; if (stat (tryname, &statbuf) < 0) continue; if (!S_ISREG (statbuf.st_mode)) /* We certainly need to reject directories. I'm not quite as sure about FIFOs, sockets, etc., but I kind of doubt that people want to exec() these things. */ continue; break; } if (p == NULL) /* Not found. This must be an error rather than merely passing the file to execlp(), because execlp() would try all the exec()s, causing GDB to get confused. */ error (_("procfs:%d -- Can't find shell %s in PATH"), __LINE__, shell_file); shell_file = tryname; } pid = fork_inferior (exec_file, allargs, env, procfs_set_exec_trap, NULL, NULL, shell_file, NULL); /* We have something that executes now. We'll be running through the shell at this point (if startup-with-shell is true), but the pid shouldn't change. */ add_thread_silent (ptid_t (pid)); procfs_init_inferior (this, pid); } /* An observer for the "inferior_created" event. */ static void procfs_inferior_created (struct target_ops *ops, int from_tty) { } /* Callback for update_thread_list. Calls "add_thread". */ static int procfs_notice_thread (procinfo *pi, procinfo *thread, void *ptr) { ptid_t gdb_threadid = ptid_t (pi->pid, thread->tid, 0); thread_info *thr = find_thread_ptid (gdb_threadid); if (thr == NULL || thr->state == THREAD_EXITED) add_thread (gdb_threadid); return 0; } /* Query all the threads that the target knows about, and give them back to GDB to add to its list. */ void procfs_target::update_thread_list () { procinfo *pi; prune_threads (); /* Find procinfo for main process. */ pi = find_procinfo_or_die (inferior_ptid.pid (), 0); proc_update_threads (pi); proc_iterate_over_threads (pi, procfs_notice_thread, NULL); } /* Return true if the thread is still 'alive'. This guy doesn't really seem to be doing his job. Got to investigate how to tell when a thread is really gone. */ bool procfs_target::thread_alive (ptid_t ptid) { int proc, thread; procinfo *pi; proc = ptid.pid (); thread = ptid.lwp (); /* If I don't know it, it ain't alive! */ pi = find_procinfo (proc, thread); if (pi == NULL) return false; /* If I can't get its status, it ain't alive! What's more, I need to forget about it! */ if (!proc_get_status (pi)) { destroy_procinfo (pi); return false; } /* I couldn't have got its status if it weren't alive, so it's alive. */ return true; } /* Convert PTID to a string. */ std::string procfs_target::pid_to_str (ptid_t ptid) { if (ptid.lwp () == 0) return string_printf ("process %d", ptid.pid ()); else return string_printf ("LWP %ld", ptid.lwp ()); } /* Accepts an integer PID; Returns a string representing a file that can be opened to get the symbols for the child process. */ char * procfs_target::pid_to_exec_file (int pid) { static char buf[PATH_MAX]; char name[PATH_MAX]; /* Solaris 11 introduced /proc//execname. */ xsnprintf (name, sizeof (name), "/proc/%d/execname", pid); scoped_fd fd (gdb_open_cloexec (name, O_RDONLY, 0)); if (fd.get () < 0 || read (fd.get (), buf, PATH_MAX - 1) < 0) { /* If that fails, fall back to /proc//path/a.out introduced in Solaris 10. */ ssize_t len; xsnprintf (name, sizeof (name), "/proc/%d/path/a.out", pid); len = readlink (name, buf, PATH_MAX - 1); if (len <= 0) strcpy (buf, name); else buf[len] = '\0'; } return buf; } /* Insert a watchpoint. */ static int procfs_set_watchpoint (ptid_t ptid, CORE_ADDR addr, int len, int rwflag, int after) { int pflags = 0; procinfo *pi; pi = find_procinfo_or_die (ptid.pid () == -1 ? inferior_ptid.pid () : ptid.pid (), 0); /* Translate from GDB's flags to /proc's. */ if (len > 0) /* len == 0 means delete watchpoint. */ { switch (rwflag) { /* FIXME: need an enum! */ case hw_write: /* default watchpoint (write) */ pflags = WRITE_WATCHFLAG; break; case hw_read: /* read watchpoint */ pflags = READ_WATCHFLAG; break; case hw_access: /* access watchpoint */ pflags = READ_WATCHFLAG | WRITE_WATCHFLAG; break; case hw_execute: /* execution HW breakpoint */ pflags = EXEC_WATCHFLAG; break; default: /* Something weird. Return error. */ return -1; } if (after) /* Stop after r/w access is completed. */ pflags |= AFTER_WATCHFLAG; } if (!proc_set_watchpoint (pi, addr, len, pflags)) { if (errno == E2BIG) /* Typical error for no resources. */ return -1; /* fail */ /* GDB may try to remove the same watchpoint twice. If a remove request returns no match, don't error. */ if (errno == ESRCH && len == 0) return 0; /* ignore */ proc_error (pi, "set_watchpoint", __LINE__); } return 0; } /* Return non-zero if we can set a hardware watchpoint of type TYPE. TYPE is one of bp_hardware_watchpoint, bp_read_watchpoint, bp_write_watchpoint, or bp_hardware_watchpoint. CNT is the number of watchpoints used so far. Note: procfs_can_use_hw_breakpoint() is not yet used by all procfs.c targets due to the fact that some of them still define target_can_use_hardware_watchpoint. */ int procfs_target::can_use_hw_breakpoint (enum bptype type, int cnt, int othertype) { /* Due to the way that proc_set_watchpoint() is implemented, host and target pointers must be of the same size. If they are not, we can't use hardware watchpoints. This limitation is due to the fact that proc_set_watchpoint() calls procfs_address_to_host_pointer(); a close inspection of procfs_address_to_host_pointer will reveal that an internal error will be generated when the host and target pointer sizes are different. */ struct type *ptr_type = builtin_type (target_gdbarch ())->builtin_data_ptr; if (sizeof (void *) != TYPE_LENGTH (ptr_type)) return 0; /* Other tests here??? */ return 1; } /* Returns non-zero if process is stopped on a hardware watchpoint fault, else returns zero. */ bool procfs_target::stopped_by_watchpoint () { procinfo *pi; pi = find_procinfo_or_die (inferior_ptid.pid (), 0); if (proc_flags (pi) & (PR_STOPPED | PR_ISTOP)) if (proc_why (pi) == PR_FAULTED) if (proc_what (pi) == FLTWATCH) return true; return false; } /* Returns 1 if the OS knows the position of the triggered watchpoint, and sets *ADDR to that address. Returns 0 if OS cannot report that address. This function is only called if procfs_stopped_by_watchpoint returned 1, thus no further checks are done. The function also assumes that ADDR is not NULL. */ bool procfs_target::stopped_data_address (CORE_ADDR *addr) { procinfo *pi; pi = find_procinfo_or_die (inferior_ptid.pid (), 0); return proc_watchpoint_address (pi, addr); } int procfs_target::insert_watchpoint (CORE_ADDR addr, int len, enum target_hw_bp_type type, struct expression *cond) { if (!target_have_steppable_watchpoint && !gdbarch_have_nonsteppable_watchpoint (target_gdbarch ())) /* When a hardware watchpoint fires off the PC will be left at the instruction following the one which caused the watchpoint. It will *NOT* be necessary for GDB to step over the watchpoint. */ return procfs_set_watchpoint (inferior_ptid, addr, len, type, 1); else /* When a hardware watchpoint fires off the PC will be left at the instruction which caused the watchpoint. It will be necessary for GDB to step over the watchpoint. */ return procfs_set_watchpoint (inferior_ptid, addr, len, type, 0); } int procfs_target::remove_watchpoint (CORE_ADDR addr, int len, enum target_hw_bp_type type, struct expression *cond) { return procfs_set_watchpoint (inferior_ptid, addr, 0, 0, 0); } int procfs_target::region_ok_for_hw_watchpoint (CORE_ADDR addr, int len) { /* The man page for proc(4) on Solaris 2.6 and up says that the system can support "thousands" of hardware watchpoints, but gives no method for finding out how many; It doesn't say anything about the allowed size for the watched area either. So we just tell GDB 'yes'. */ return 1; } /* Memory Mappings Functions: */ /* Call a callback function once for each mapping, passing it the mapping, an optional secondary callback function, and some optional opaque data. Quit and return the first non-zero value returned from the callback. PI is the procinfo struct for the process to be mapped. FUNC is the callback function to be called by this iterator. DATA is the optional opaque data to be passed to the callback function. CHILD_FUNC is the optional secondary function pointer to be passed to the child function. Returns the first non-zero return value from the callback function, or zero. */ static int iterate_over_mappings (procinfo *pi, find_memory_region_ftype child_func, void *data, int (*func) (struct prmap *map, find_memory_region_ftype child_func, void *data)) { char pathname[MAX_PROC_NAME_SIZE]; struct prmap *prmaps; struct prmap *prmap; int funcstat; int nmap; struct stat sbuf; /* Get the number of mappings, allocate space, and read the mappings into prmaps. */ /* Open map fd. */ xsnprintf (pathname, sizeof (pathname), "/proc/%d/map", pi->pid); scoped_fd map_fd (open (pathname, O_RDONLY)); if (map_fd.get () < 0) proc_error (pi, "iterate_over_mappings (open)", __LINE__); /* Use stat to determine the file size, and compute the number of prmap_t objects it contains. */ if (fstat (map_fd.get (), &sbuf) != 0) proc_error (pi, "iterate_over_mappings (fstat)", __LINE__); nmap = sbuf.st_size / sizeof (prmap_t); prmaps = (struct prmap *) alloca ((nmap + 1) * sizeof (*prmaps)); if (read (map_fd.get (), (char *) prmaps, nmap * sizeof (*prmaps)) != (nmap * sizeof (*prmaps))) proc_error (pi, "iterate_over_mappings (read)", __LINE__); for (prmap = prmaps; nmap > 0; prmap++, nmap--) { funcstat = (*func) (prmap, child_func, data); if (funcstat != 0) return funcstat; } return 0; } /* Implements the to_find_memory_regions method. Calls an external function for each memory region. Returns the integer value returned by the callback. */ static int find_memory_regions_callback (struct prmap *map, find_memory_region_ftype func, void *data) { return (*func) ((CORE_ADDR) map->pr_vaddr, map->pr_size, (map->pr_mflags & MA_READ) != 0, (map->pr_mflags & MA_WRITE) != 0, (map->pr_mflags & MA_EXEC) != 0, 1, /* MODIFIED is unknown, pass it as true. */ data); } /* External interface. Calls a callback function once for each mapped memory region in the child process, passing as arguments: CORE_ADDR virtual_address, unsigned long size, int read, TRUE if region is readable by the child int write, TRUE if region is writable by the child int execute TRUE if region is executable by the child. Stops iterating and returns the first non-zero value returned by the callback. */ int procfs_target::find_memory_regions (find_memory_region_ftype func, void *data) { procinfo *pi = find_procinfo_or_die (inferior_ptid.pid (), 0); return iterate_over_mappings (pi, func, data, find_memory_regions_callback); } /* Returns an ascii representation of a memory mapping's flags. */ static char * mappingflags (long flags) { static char asciiflags[8]; strcpy (asciiflags, "-------"); if (flags & MA_STACK) asciiflags[1] = 's'; if (flags & MA_BREAK) asciiflags[2] = 'b'; if (flags & MA_SHARED) asciiflags[3] = 's'; if (flags & MA_READ) asciiflags[4] = 'r'; if (flags & MA_WRITE) asciiflags[5] = 'w'; if (flags & MA_EXEC) asciiflags[6] = 'x'; return (asciiflags); } /* Callback function, does the actual work for 'info proc mappings'. */ static int info_mappings_callback (struct prmap *map, find_memory_region_ftype ignore, void *unused) { unsigned int pr_off; pr_off = (unsigned int) map->pr_offset; if (gdbarch_addr_bit (target_gdbarch ()) == 32) printf_filtered ("\t%#10lx %#10lx %#10lx %#10x %7s\n", (unsigned long) map->pr_vaddr, (unsigned long) map->pr_vaddr + map->pr_size - 1, (unsigned long) map->pr_size, pr_off, mappingflags (map->pr_mflags)); else printf_filtered (" %#18lx %#18lx %#10lx %#10x %7s\n", (unsigned long) map->pr_vaddr, (unsigned long) map->pr_vaddr + map->pr_size - 1, (unsigned long) map->pr_size, pr_off, mappingflags (map->pr_mflags)); return 0; } /* Implement the "info proc mappings" subcommand. */ static void info_proc_mappings (procinfo *pi, int summary) { if (summary) return; /* No output for summary mode. */ printf_filtered (_("Mapped address spaces:\n\n")); if (gdbarch_ptr_bit (target_gdbarch ()) == 32) printf_filtered ("\t%10s %10s %10s %10s %7s\n", "Start Addr", " End Addr", " Size", " Offset", "Flags"); else printf_filtered (" %18s %18s %10s %10s %7s\n", "Start Addr", " End Addr", " Size", " Offset", "Flags"); iterate_over_mappings (pi, NULL, NULL, info_mappings_callback); printf_filtered ("\n"); } /* Implement the "info proc" command. */ bool procfs_target::info_proc (const char *args, enum info_proc_what what) { procinfo *process = NULL; procinfo *thread = NULL; char *tmp = NULL; int pid = 0; int tid = 0; int mappings = 0; switch (what) { case IP_MINIMAL: break; case IP_MAPPINGS: case IP_ALL: mappings = 1; break; default: error (_("Not supported on this target.")); } gdb_argv built_argv (args); for (char *arg : built_argv) { if (isdigit (arg[0])) { pid = strtoul (arg, &tmp, 10); if (*tmp == '/') tid = strtoul (++tmp, NULL, 10); } else if (arg[0] == '/') { tid = strtoul (arg + 1, NULL, 10); } } procinfo_up temporary_procinfo; if (pid == 0) pid = inferior_ptid.pid (); if (pid == 0) error (_("No current process: you must name one.")); else { /* Have pid, will travel. First see if it's a process we're already debugging. */ process = find_procinfo (pid, 0); if (process == NULL) { /* No. So open a procinfo for it, but remember to close it again when finished. */ process = create_procinfo (pid, 0); temporary_procinfo.reset (process); if (!open_procinfo_files (process, FD_CTL)) proc_error (process, "info proc, open_procinfo_files", __LINE__); } } if (tid != 0) thread = create_procinfo (pid, tid); if (process) { printf_filtered (_("process %d flags:\n"), process->pid); proc_prettyprint_flags (proc_flags (process), 1); if (proc_flags (process) & (PR_STOPPED | PR_ISTOP)) proc_prettyprint_why (proc_why (process), proc_what (process), 1); if (proc_get_nthreads (process) > 1) printf_filtered ("Process has %d threads.\n", proc_get_nthreads (process)); } if (thread) { printf_filtered (_("thread %d flags:\n"), thread->tid); proc_prettyprint_flags (proc_flags (thread), 1); if (proc_flags (thread) & (PR_STOPPED | PR_ISTOP)) proc_prettyprint_why (proc_why (thread), proc_what (thread), 1); } if (mappings) info_proc_mappings (process, 0); return true; } /* Modify the status of the system call identified by SYSCALLNUM in the set of syscalls that are currently traced/debugged. If ENTRY_OR_EXIT is set to PR_SYSENTRY, then the entry syscalls set will be updated. Otherwise, the exit syscalls set will be updated. If MODE is FLAG_SET, then traces will be enabled. Otherwise, they will be disabled. */ static void proc_trace_syscalls_1 (procinfo *pi, int syscallnum, int entry_or_exit, int mode, int from_tty) { sysset_t *sysset; if (entry_or_exit == PR_SYSENTRY) sysset = proc_get_traced_sysentry (pi, NULL); else sysset = proc_get_traced_sysexit (pi, NULL); if (sysset == NULL) proc_error (pi, "proc-trace, get_traced_sysset", __LINE__); if (mode == FLAG_SET) praddset (sysset, syscallnum); else prdelset (sysset, syscallnum); if (entry_or_exit == PR_SYSENTRY) { if (!proc_set_traced_sysentry (pi, sysset)) proc_error (pi, "proc-trace, set_traced_sysentry", __LINE__); } else { if (!proc_set_traced_sysexit (pi, sysset)) proc_error (pi, "proc-trace, set_traced_sysexit", __LINE__); } } static void proc_trace_syscalls (const char *args, int from_tty, int entry_or_exit, int mode) { procinfo *pi; if (inferior_ptid.pid () <= 0) error (_("you must be debugging a process to use this command.")); if (args == NULL || args[0] == 0) error_no_arg (_("system call to trace")); pi = find_procinfo_or_die (inferior_ptid.pid (), 0); if (isdigit (args[0])) { const int syscallnum = atoi (args); proc_trace_syscalls_1 (pi, syscallnum, entry_or_exit, mode, from_tty); } } static void proc_trace_sysentry_cmd (const char *args, int from_tty) { proc_trace_syscalls (args, from_tty, PR_SYSENTRY, FLAG_SET); } static void proc_trace_sysexit_cmd (const char *args, int from_tty) { proc_trace_syscalls (args, from_tty, PR_SYSEXIT, FLAG_SET); } static void proc_untrace_sysentry_cmd (const char *args, int from_tty) { proc_trace_syscalls (args, from_tty, PR_SYSENTRY, FLAG_RESET); } static void proc_untrace_sysexit_cmd (const char *args, int from_tty) { proc_trace_syscalls (args, from_tty, PR_SYSEXIT, FLAG_RESET); } void _initialize_procfs (void) { gdb::observers::inferior_created.attach (procfs_inferior_created); add_com ("proc-trace-entry", no_class, proc_trace_sysentry_cmd, _("Give a trace of entries into the syscall.")); add_com ("proc-trace-exit", no_class, proc_trace_sysexit_cmd, _("Give a trace of exits from the syscall.")); add_com ("proc-untrace-entry", no_class, proc_untrace_sysentry_cmd, _("Cancel a trace of entries into the syscall.")); add_com ("proc-untrace-exit", no_class, proc_untrace_sysexit_cmd, _("Cancel a trace of exits from the syscall.")); add_inf_child_target (&the_procfs_target); } /* =================== END, GDB "MODULE" =================== */ /* miscellaneous stubs: */ /* The following satisfy a few random symbols mostly created by the solaris threads implementation, which I will chase down later. */ /* Return a pid for which we guarantee we will be able to find a 'live' procinfo. */ ptid_t procfs_first_available (void) { return ptid_t (procinfo_list ? procinfo_list->pid : -1); } /* =================== GCORE .NOTE "MODULE" =================== */ static char * procfs_do_thread_registers (bfd *obfd, ptid_t ptid, char *note_data, int *note_size, enum gdb_signal stop_signal) { struct regcache *regcache = get_thread_regcache (ptid); gdb_gregset_t gregs; gdb_fpregset_t fpregs; unsigned long merged_pid; merged_pid = ptid.lwp () << 16 | ptid.pid (); /* This part is the old method for fetching registers. It should be replaced by the newer one using regsets once it is implemented in this platform: gdbarch_iterate_over_regset_sections(). */ scoped_restore save_inferior_ptid = make_scoped_restore (&inferior_ptid); inferior_ptid = ptid; target_fetch_registers (regcache, -1); fill_gregset (regcache, &gregs, -1); note_data = (char *) elfcore_write_lwpstatus (obfd, note_data, note_size, merged_pid, stop_signal, &gregs); fill_fpregset (regcache, &fpregs, -1); note_data = (char *) elfcore_write_prfpreg (obfd, note_data, note_size, &fpregs, sizeof (fpregs)); return note_data; } struct procfs_corefile_thread_data { bfd *obfd; char *note_data; int *note_size; enum gdb_signal stop_signal; }; static int procfs_corefile_thread_callback (procinfo *pi, procinfo *thread, void *data) { struct procfs_corefile_thread_data *args = (struct procfs_corefile_thread_data *) data; if (pi != NULL) { ptid_t ptid = ptid_t (pi->pid, thread->tid, 0); args->note_data = procfs_do_thread_registers (args->obfd, ptid, args->note_data, args->note_size, args->stop_signal); } return 0; } static int find_signalled_thread (struct thread_info *info, void *data) { if (info->suspend.stop_signal != GDB_SIGNAL_0 && info->ptid.pid () == inferior_ptid.pid ()) return 1; return 0; } static enum gdb_signal find_stop_signal (void) { struct thread_info *info = iterate_over_threads (find_signalled_thread, NULL); if (info) return info->suspend.stop_signal; else return GDB_SIGNAL_0; } char * procfs_target::make_corefile_notes (bfd *obfd, int *note_size) { gdb_gregset_t gregs; char fname[16] = {'\0'}; char psargs[80] = {'\0'}; procinfo *pi = find_procinfo_or_die (inferior_ptid.pid (), 0); char *note_data = NULL; const char *inf_args; struct procfs_corefile_thread_data thread_args; enum gdb_signal stop_signal; if (get_exec_file (0)) { strncpy (fname, lbasename (get_exec_file (0)), sizeof (fname)); fname[sizeof (fname) - 1] = 0; strncpy (psargs, get_exec_file (0), sizeof (psargs)); psargs[sizeof (psargs) - 1] = 0; inf_args = get_inferior_args (); if (inf_args && *inf_args && (strlen (inf_args) < ((int) sizeof (psargs) - (int) strlen (psargs)))) { strncat (psargs, " ", sizeof (psargs) - strlen (psargs)); strncat (psargs, inf_args, sizeof (psargs) - strlen (psargs)); } } note_data = (char *) elfcore_write_prpsinfo (obfd, note_data, note_size, fname, psargs); stop_signal = find_stop_signal (); fill_gregset (get_current_regcache (), &gregs, -1); note_data = elfcore_write_pstatus (obfd, note_data, note_size, inferior_ptid.pid (), stop_signal, &gregs); thread_args.obfd = obfd; thread_args.note_data = note_data; thread_args.note_size = note_size; thread_args.stop_signal = stop_signal; proc_iterate_over_threads (pi, procfs_corefile_thread_callback, &thread_args); note_data = thread_args.note_data; gdb::optional auxv = target_read_alloc (current_top_target (), TARGET_OBJECT_AUXV, NULL); if (auxv && !auxv->empty ()) note_data = elfcore_write_note (obfd, note_data, note_size, "CORE", NT_AUXV, auxv->data (), auxv->size ()); return note_data; } /* =================== END GCORE .NOTE "MODULE" =================== */