minix/servers/avfs/misc.c
Thomas Veerman ddbdca6cdb Add support for survival of crashed FSs
When an FS crashes, VFS will clean up resources tied to that FS:
 - Pending requests to the FS are canceled (i.e., fail with EIO)
 - Threads waiting for a reply are stopped (i.e., fail with EIO)
 - Open files are marked invalid. Future operations on a file descriptor
   will cause EBADF errors.
 - vmnt entry is cleared, so in-flight system calls that got past the
   file descriptor check but not yet talking to the crashed FS, will
   fail with EIO.
 - The reference counter of the mount point is decreased, effectively
   removing the crashed FS from the file system tree. Descendants of
   this part of the tree are unreachable by means of a path, but can
   still be unmounted by feeding the block special file to unmount(2).

This patch also gets rid of the "not a known driver endpoint" messages
during shutdown.
2012-01-19 14:21:46 +00:00

662 lines
19 KiB
C

/* This file contains a collection of miscellaneous procedures. Some of them
* perform simple system calls. Some others do a little part of system calls
* that are mostly performed by the Memory Manager.
*
* The entry points into this file are
* do_dup: perform the DUP system call
* do_fcntl: perform the FCNTL system call
* do_sync: perform the SYNC system call
* do_fsync: perform the FSYNC system call
* pm_reboot: sync disks and prepare for shutdown
* pm_fork: adjust the tables after PM has performed a FORK system call
* do_exec: handle files with FD_CLOEXEC on after PM has done an EXEC
* do_exit: a process has exited; note that in the tables
* do_set: set uid or gid for some process
* do_revive: revive a process that was waiting for something (e.g. TTY)
* do_svrctl: file system control
* do_getsysinfo: request copy of FS data structure
* pm_dumpcore: create a core dump
*/
#include "fs.h"
#include <fcntl.h>
#include <assert.h>
#include <unistd.h>
#include <string.h>
#include <minix/callnr.h>
#include <minix/safecopies.h>
#include <minix/endpoint.h>
#include <minix/com.h>
#include <minix/sysinfo.h>
#include <minix/u64.h>
#include <sys/ptrace.h>
#include <sys/svrctl.h>
#include "file.h"
#include "fproc.h"
#include "scratchpad.h"
#include "dmap.h"
#include <minix/vfsif.h>
#include "vnode.h"
#include "vmnt.h"
#include "param.h"
#define CORE_NAME "core"
#define CORE_MODE 0777 /* mode to use on core image files */
#if ENABLE_SYSCALL_STATS
PUBLIC unsigned long calls_stats[NCALLS];
#endif
FORWARD _PROTOTYPE( void free_proc, (struct fproc *freed, int flags) );
/*
FORWARD _PROTOTYPE( int dumpcore, (int proc_e, struct mem_map *seg_ptr) );
FORWARD _PROTOTYPE( int write_bytes, (struct inode *rip, off_t off,
char *buf, size_t bytes) );
FORWARD _PROTOTYPE( int write_seg, (struct inode *rip, off_t off, int proc_e,
int seg, off_t seg_off, phys_bytes seg_bytes) );
*/
/*===========================================================================*
* do_getsysinfo *
*===========================================================================*/
PUBLIC int do_getsysinfo()
{
vir_bytes src_addr, dst_addr;
size_t len;
/* Only su may call do_getsysinfo. This call may leak information (and is not
* stable enough to be part of the API/ABI). In the future, requests from
* non-system processes should be denied.
*/
if (!super_user) return(EPERM);
switch(m_in.SI_WHAT) {
case SI_PROC_TAB:
src_addr = (vir_bytes) fproc;
len = sizeof(struct fproc) * NR_PROCS;
break;
case SI_DMAP_TAB:
src_addr = (vir_bytes) dmap;
len = sizeof(struct dmap) * NR_DEVICES;
break;
#if ENABLE_SYSCALL_STATS
case SI_CALL_STATS:
src_addr = (vir_bytes) calls_stats;
len = sizeof(calls_stats);
break;
#endif
default:
return(EINVAL);
}
if (len != m_in.SI_SIZE)
return(EINVAL);
dst_addr = (vir_bytes) m_in.SI_WHERE;
return sys_datacopy(SELF, src_addr, who_e, dst_addr, len);
}
/*===========================================================================*
* do_dup *
*===========================================================================*/
PUBLIC int do_dup()
{
/* Perform the dup(fd) or dup2(fd,fd2) system call. These system calls are
* obsolete. In fact, it is not even possible to invoke them using the
* current library because the library routines call fcntl(). They are
* provided to permit old binary programs to continue to run.
*/
register int rfd;
register struct filp *f;
int r = OK;
/* Is the file descriptor valid? */
rfd = m_in.fd & ~DUP_MASK; /* kill off dup2 bit, if on */
if ((f = get_filp(rfd, VNODE_READ)) == NULL) return(err_code);
/* Distinguish between dup and dup2. */
if (m_in.fd == rfd) { /* bit not on */
/* dup(fd) */
r = get_fd(0, 0, &m_in.fd2, NULL);
} else {
/* dup2(old_fd, new_fd) */
if (m_in.fd2 < 0 || m_in.fd2 >= OPEN_MAX) {
r = EBADF;
} else if (rfd == m_in.fd2) { /* ignore the call: dup2(x, x) */
r = m_in.fd2;
} else {
/* All is fine, close new_fd if necessary */
m_in.fd = m_in.fd2; /* prepare to close fd2 */
unlock_filp(f); /* or it might deadlock on do_close */
(void) do_close(); /* cannot fail */
f = get_filp(rfd, VNODE_READ); /* lock old_fd again */
}
}
if (r == OK) {
/* Success. Set up new file descriptors. */
f->filp_count++;
fp->fp_filp[m_in.fd2] = f;
FD_SET(m_in.fd2, &fp->fp_filp_inuse);
r = m_in.fd2;
}
unlock_filp(f);
return(r);
}
/*===========================================================================*
* do_fcntl *
*===========================================================================*/
PUBLIC int do_fcntl()
{
/* Perform the fcntl(fd, request, ...) system call. */
register struct filp *f;
int new_fd, fl, r = OK;
tll_access_t locktype;
scratch(fp).file.fd_nr = m_in.fd;
scratch(fp).io.io_buffer = m_in.buffer;
scratch(fp).io.io_nbytes = m_in.nbytes; /* a.k.a. m_in.request */
/* Is the file descriptor valid? */
locktype = (m_in.request == F_FREESP) ? VNODE_WRITE : VNODE_READ;
if ((f = get_filp(scratch(fp).file.fd_nr, locktype)) == NULL)
return(err_code);
switch (m_in.request) {
case F_DUPFD:
/* This replaces the old dup() system call. */
if (m_in.addr < 0 || m_in.addr >= OPEN_MAX) r = EINVAL;
else if ((r = get_fd(m_in.addr, 0, &new_fd, NULL)) == OK) {
f->filp_count++;
fp->fp_filp[new_fd] = f;
FD_SET(new_fd, &fp->fp_filp_inuse);
r = new_fd;
}
break;
case F_GETFD:
/* Get close-on-exec flag (FD_CLOEXEC in POSIX Table 6-2). */
r = 0;
if (FD_ISSET(scratch(fp).file.fd_nr, &fp->fp_cloexec_set))
r = FD_CLOEXEC;
break;
case F_SETFD:
/* Set close-on-exec flag (FD_CLOEXEC in POSIX Table 6-2). */
if(m_in.addr & FD_CLOEXEC)
FD_SET(scratch(fp).file.fd_nr, &fp->fp_cloexec_set);
else
FD_CLR(scratch(fp).file.fd_nr, &fp->fp_cloexec_set);
break;
case F_GETFL:
/* Get file status flags (O_NONBLOCK and O_APPEND). */
fl = f->filp_flags & (O_NONBLOCK | O_APPEND | O_ACCMODE);
r = fl;
break;
case F_SETFL:
/* Set file status flags (O_NONBLOCK and O_APPEND). */
fl = O_NONBLOCK | O_APPEND | O_REOPEN;
f->filp_flags = (f->filp_flags & ~fl) | (m_in.addr & fl);
break;
case F_GETLK:
case F_SETLK:
case F_SETLKW:
/* Set or clear a file lock. */
r = lock_op(f, m_in.request);
break;
case F_FREESP:
{
/* Free a section of a file */
off_t start, end;
struct flock flock_arg;
signed long offset;
/* Check if it's a regular file. */
if ((f->filp_vno->v_mode & I_TYPE) != I_REGULAR) r = EINVAL;
else if (!(f->filp_mode & W_BIT)) r = EBADF;
else
/* Copy flock data from userspace. */
r = sys_datacopy(who_e, (vir_bytes) m_in.name1, SELF,
(vir_bytes) &flock_arg,
(phys_bytes) sizeof(flock_arg));
if (r != OK) break;
/* Convert starting offset to signed. */
offset = (signed long) flock_arg.l_start;
/* Figure out starting position base. */
switch(flock_arg.l_whence) {
case SEEK_SET: start = 0; break;
case SEEK_CUR:
if (ex64hi(f->filp_pos) != 0)
panic("do_fcntl: position in file too high");
start = ex64lo(f->filp_pos);
break;
case SEEK_END: start = f->filp_vno->v_size; break;
default: r = EINVAL;
}
if (r != OK) break;
/* Check for overflow or underflow. */
if (offset > 0 && start + offset < start) r = EINVAL;
else if (offset < 0 && start + offset > start) r = EINVAL;
else {
start += offset;
if (start < 0) r = EINVAL;
}
if (r != OK) break;
if (flock_arg.l_len != 0) {
if (start >= f->filp_vno->v_size) r = EINVAL;
else if ((end = start + flock_arg.l_len) <= start) r = EINVAL;
else if (end > f->filp_vno->v_size) end = f->filp_vno->v_size;
} else {
end = 0;
}
if (r != OK) break;
r = req_ftrunc(f->filp_vno->v_fs_e, f->filp_vno->v_inode_nr,start,end);
if (r == OK && flock_arg.l_len == 0)
f->filp_vno->v_size = start;
break;
}
default:
r = EINVAL;
}
unlock_filp(f);
return(r);
}
/*===========================================================================*
* do_sync *
*===========================================================================*/
PUBLIC int do_sync()
{
struct vmnt *vmp;
int r = OK;
for (vmp = &vmnt[0]; vmp < &vmnt[NR_MNTS]; ++vmp) {
if (vmp->m_dev != NO_DEV && vmp->m_fs_e != NONE &&
vmp->m_root_node != NULL) {
if ((r = lock_vmnt(vmp, VMNT_EXCL)) != OK)
break;
req_sync(vmp->m_fs_e);
unlock_vmnt(vmp);
}
}
return(r);
}
/*===========================================================================*
* do_fsync *
*===========================================================================*/
PUBLIC int do_fsync()
{
/* Perform the fsync() system call. */
struct filp *rfilp;
struct vmnt *vmp;
dev_t dev;
int r = OK;
if ((rfilp = get_filp(m_in.m1_i1, VNODE_READ)) == NULL) return(err_code);
dev = rfilp->filp_vno->v_dev;
for (vmp = &vmnt[0]; vmp < &vmnt[NR_MNTS]; ++vmp) {
if (vmp->m_dev != NO_DEV && vmp->m_dev == dev &&
vmp->m_fs_e != NONE && vmp->m_root_node != NULL) {
if ((r = lock_vmnt(vmp, VMNT_EXCL)) != OK)
break;
req_sync(vmp->m_fs_e);
unlock_vmnt(vmp);
}
}
unlock_filp(rfilp);
return(r);
}
/*===========================================================================*
* pm_reboot *
*===========================================================================*/
PUBLIC void pm_reboot()
{
/* Perform the VFS side of the reboot call. */
int i;
struct fproc *rfp;
do_sync();
/* Do exit processing for all leftover processes and servers,
* but don't actually exit them (if they were really gone, PM
* will tell us about it).
*/
for (i = 0; i < NR_PROCS; i++) {
rfp = &fproc[i];
if (rfp->fp_endpoint == NONE) continue;
/* Don't just free the proc right away, but let it finish what it was
* doing first */
lock_proc(rfp, 0);
free_proc(rfp, 0);
unlock_proc(rfp);
}
do_sync();
unmount_all();
}
/*===========================================================================*
* pm_fork *
*===========================================================================*/
PUBLIC void pm_fork(pproc, cproc, cpid)
int pproc; /* Parent process */
int cproc; /* Child process */
int cpid; /* Child process id */
{
/* Perform those aspects of the fork() system call that relate to files.
* In particular, let the child inherit its parent's file descriptors.
* The parent and child parameters tell who forked off whom. The file
* system uses the same slot numbers as the kernel. Only PM makes this call.
*/
register struct fproc *cp, *pp;
int i, parentno, childno;
mutex_t c_fp_lock;
/* Check up-to-dateness of fproc. */
okendpt(pproc, &parentno);
/* PM gives child endpoint, which implies process slot information.
* Don't call isokendpt, because that will verify if the endpoint
* number is correct in fproc, which it won't be.
*/
childno = _ENDPOINT_P(cproc);
if (childno < 0 || childno >= NR_PROCS)
panic("VFS: bogus child for forking: %d", m_in.child_endpt);
if (fproc[childno].fp_pid != PID_FREE)
panic("VFS: forking on top of in-use child: %d", childno);
/* Copy the parent's fproc struct to the child. */
/* However, the mutex variables belong to a slot and must stay the same. */
c_fp_lock = fproc[childno].fp_lock;
fproc[childno] = fproc[parentno];
fproc[childno].fp_lock = c_fp_lock;
/* Increase the counters in the 'filp' table. */
cp = &fproc[childno];
pp = &fproc[parentno];
for (i = 0; i < OPEN_MAX; i++)
if (cp->fp_filp[i] != NULL) cp->fp_filp[i]->filp_count++;
/* Fill in new process and endpoint id. */
cp->fp_pid = cpid;
cp->fp_endpoint = cproc;
/* A forking process never has an outstanding grant, as it isn't blocking on
* I/O. */
if(GRANT_VALID(pp->fp_grant)) {
panic("VFS: fork: pp (endpoint %d) has grant %d\n", pp->fp_endpoint,
pp->fp_grant);
}
if(GRANT_VALID(cp->fp_grant)) {
panic("VFS: fork: cp (endpoint %d) has grant %d\n", cp->fp_endpoint,
cp->fp_grant);
}
/* A child is not a process leader, not being revived, etc. */
cp->fp_flags = FP_NOFLAGS;
/* Record the fact that both root and working dir have another user. */
if (cp->fp_rd) dup_vnode(cp->fp_rd);
if (cp->fp_wd) dup_vnode(cp->fp_wd);
}
/*===========================================================================*
* free_proc *
*===========================================================================*/
PRIVATE void free_proc(struct fproc *exiter, int flags)
{
int i;
register struct fproc *rfp;
register struct filp *rfilp;
register struct vnode *vp;
dev_t dev;
if (exiter->fp_endpoint == NONE)
panic("free_proc: already free");
if (fp_is_blocked(exiter))
unpause(exiter->fp_endpoint);
/* Loop on file descriptors, closing any that are open. */
for (i = 0; i < OPEN_MAX; i++) {
(void) close_fd(exiter, i);
}
/* Release root and working directories. */
if (exiter->fp_rd) { put_vnode(exiter->fp_rd); exiter->fp_rd = NULL; }
if (exiter->fp_wd) { put_vnode(exiter->fp_wd); exiter->fp_wd = NULL; }
/* The rest of these actions is only done when processes actually exit. */
if (!(flags & FP_EXITING)) return;
/* Check if any process is SUSPENDed on this driver.
* If a driver exits, unmap its entries in the dmap table.
* (unmapping has to be done after the first step, because the
* dmap table is used in the first step.)
*/
unsuspend_by_endpt(exiter->fp_endpoint);
dmap_unmap_by_endpt(exiter->fp_endpoint);
worker_stop_by_endpt(exiter->fp_endpoint); /* Unblock waiting threads */
vmnt_unmap_by_endpt(exiter->fp_endpoint); /* Invalidate open files if this
* was an active FS */
/* Invalidate endpoint number for error and sanity checks. */
exiter->fp_endpoint = NONE;
exiter->fp_flags |= FP_EXITING;
/* If a session leader exits and it has a controlling tty, then revoke
* access to its controlling tty from all other processes using it.
*/
if ((exiter->fp_flags & FP_SESLDR) && exiter->fp_tty != 0) {
dev = exiter->fp_tty;
for (rfp = &fproc[0]; rfp < &fproc[NR_PROCS]; rfp++) {
if(rfp->fp_pid == PID_FREE) continue;
if (rfp->fp_tty == dev) rfp->fp_tty = 0;
for (i = 0; i < OPEN_MAX; i++) {
if ((rfilp = rfp->fp_filp[i]) == NULL) continue;
if (rfilp->filp_mode == FILP_CLOSED) continue;
vp = rfilp->filp_vno;
if ((vp->v_mode & I_TYPE) != I_CHAR_SPECIAL) continue;
if ((dev_t) vp->v_sdev != dev) continue;
lock_filp(rfilp, VNODE_READ);
(void) dev_close(dev, rfilp-filp); /* Ignore any errors, even
* SUSPEND. */
rfilp->filp_mode = FILP_CLOSED;
unlock_filp(rfilp);
}
}
}
/* Exit done. Mark slot as free. */
exiter->fp_pid = PID_FREE;
if (exiter->fp_flags & FP_PENDING)
pending--; /* No longer pending job, not going to do it */
exiter->fp_flags = FP_NOFLAGS;
}
/*===========================================================================*
* pm_exit *
*===========================================================================*/
PUBLIC void pm_exit(proc)
int proc;
{
/* Perform the file system portion of the exit(status) system call. */
int exitee_p;
/* Nevertheless, pretend that the call came from the user. */
okendpt(proc, &exitee_p);
fp = &fproc[exitee_p];
free_proc(fp, FP_EXITING);
}
/*===========================================================================*
* pm_setgid *
*===========================================================================*/
PUBLIC void pm_setgid(proc_e, egid, rgid)
int proc_e;
int egid;
int rgid;
{
register struct fproc *tfp;
int slot;
okendpt(proc_e, &slot);
tfp = &fproc[slot];
tfp->fp_effgid = egid;
tfp->fp_realgid = rgid;
}
/*===========================================================================*
* pm_setgroups *
*===========================================================================*/
PUBLIC void pm_setgroups(proc_e, ngroups, groups)
int proc_e;
int ngroups;
gid_t *groups;
{
struct fproc *rfp;
int slot;
okendpt(proc_e, &slot);
rfp = &fproc[slot];
if (ngroups * sizeof(gid_t) > sizeof(rfp->fp_sgroups))
panic("VFS: pm_setgroups: too much data to copy");
if (sys_datacopy(who_e, (vir_bytes) groups, SELF, (vir_bytes) rfp->fp_sgroups,
ngroups * sizeof(gid_t)) == OK) {
rfp->fp_ngroups = ngroups;
} else
panic("VFS: pm_setgroups: datacopy failed");
}
/*===========================================================================*
* pm_setuid *
*===========================================================================*/
PUBLIC void pm_setuid(proc_e, euid, ruid)
int proc_e;
int euid;
int ruid;
{
struct fproc *tfp;
int slot;
okendpt(proc_e, &slot);
tfp = &fproc[slot];
tfp->fp_effuid = euid;
tfp->fp_realuid = ruid;
}
/*===========================================================================*
* do_svrctl *
*===========================================================================*/
PUBLIC int do_svrctl()
{
switch (m_in.svrctl_req) {
/* No control request implemented yet. */
default:
return(EINVAL);
}
}
/*===========================================================================*
* pm_dumpcore *
*===========================================================================*/
PUBLIC int pm_dumpcore(endpoint_t proc_e, int csig, vir_bytes exe_name)
{
int slot, r, core_fd;
struct filp *f;
char core_path[PATH_MAX];
char proc_name[PROC_NAME_LEN];
okendpt(proc_e, &slot);
fp = &fproc[slot];
/* open core file */
snprintf(core_path, PATH_MAX, "%s.%d", CORE_NAME, fp->fp_pid);
core_fd = common_open(core_path, O_WRONLY | O_CREAT | O_TRUNC, CORE_MODE);
if (core_fd < 0) return(core_fd);
/* get process' name */
r = sys_datacopy(PM_PROC_NR, exe_name, VFS_PROC_NR, (vir_bytes) proc_name,
PROC_NAME_LEN);
if (r != OK) return(r);
proc_name[PROC_NAME_LEN - 1] = '\0';
if ((f = get_filp(core_fd, VNODE_WRITE)) == NULL) return(EBADF);
write_elf_core_file(f, csig, proc_name);
unlock_filp(f);
(void) close_fd(fp, core_fd); /* ignore failure, we're exiting anyway */
free_proc(fp, FP_EXITING);
return(OK);
}
/*===========================================================================*
* ds_event *
*===========================================================================*/
PUBLIC void ds_event(void)
{
char key[DS_MAX_KEYLEN];
char *blkdrv_prefix = "drv.blk.";
char *chrdrv_prefix = "drv.chr.";
u32_t value;
int type, r, is_blk;
endpoint_t owner_endpoint;
/* Get the event and the owner from DS. */
while ((r = ds_check(key, &type, &owner_endpoint)) == OK) {
/* Only check for block and character driver up events. */
if (!strncmp(key, blkdrv_prefix, strlen(blkdrv_prefix))) {
is_blk = TRUE;
} else if (!strncmp(key, chrdrv_prefix, strlen(chrdrv_prefix))) {
is_blk = FALSE;
} else {
continue;
}
if ((r = ds_retrieve_u32(key, &value)) != OK) {
printf("VFS: ds_event: ds_retrieve_u32 failed\n");
return;
}
if (value != DS_DRIVER_UP) continue;
/* Perform up. */
dmap_endpt_up(owner_endpoint, is_blk);
}
if (r != ENOENT) printf("VFS: ds_event: ds_check failed: %d\n", r);
}