minix/servers/vfs/protect.c
David van Moolenbroek 723e51327f VFS: worker thread model overhaul
The main purpose of this patch is to fix handling of unpause calls
from PM while another call is ongoing. The solution to this problem
sparked a full revision of the threading model, consisting of a large
number of related changes:

- all active worker threads are now always associated with a process,
  and every process has at most one active thread working for it;
- the process lock is always held by a process's worker thread;
- a process can now have both normal work and postponed PM work
  associated to it;
- timer expiry and non-postponed PM work is done from the main thread;
- filp garbage collection is done from a thread associated with VFS;
- reboot calls from PM are now done from a thread associated with PM;
- the DS events handler is protected from starting multiple threads;
- support for a system worker thread has been removed;
- the deadlock recovery thread has been replaced by a parameter to the
  worker_start() function; the number of worker threads has
  consequently been increased by one;
- saving and restoring of global but per-thread variables is now
  centralized in worker_suspend() and worker_resume(); err_code is now
  saved and restored in all cases;
- the concept of jobs has been removed, and job_m_in now points to a
  message stored in the worker thread structure instead;
- the PM lock has been removed;
- the separate exec lock has been replaced by a lock on the VM
  process, which was already being locked for exec calls anyway;
- PM_UNPAUSE is now processed as a postponed PM request, from a thread
  associated with the target process;
- the FP_DROP_WORK flag has been removed, since it is no longer more
  than just an optimization and only applied to processes operating on
  a pipe when getting killed;
- assignment to "fp" now takes place only when obtaining new work in
  the main thread or a worker thread, when resuming execution of a
  thread, and in the special case of exiting processes during reboot;
- there are no longer special cases where the yield() call is used to
  force a thread to run.

Change-Id: I7a97b9b95c2450454a9b5318dfa0e6150d4e6858
2014-02-18 11:25:03 +01:00

312 lines
8.9 KiB
C

/* This file deals with protection in the file system. It contains the code
* for four system calls that relate to protection.
*
* The entry points into this file are
* do_chmod: perform the CHMOD and FCHMOD system calls
* do_chown: perform the CHOWN and FCHOWN system calls
* do_umask: perform the UMASK system call
* do_access: perform the ACCESS system call
*/
#include "fs.h"
#include <sys/stat.h>
#include <unistd.h>
#include <assert.h>
#include <minix/callnr.h>
#include "file.h"
#include "path.h"
#include "param.h"
#include <minix/vfsif.h>
#include "vnode.h"
#include "vmnt.h"
/*===========================================================================*
* do_chmod *
*===========================================================================*/
int do_chmod(message *UNUSED(m_out))
{
/* Perform the chmod(name, mode) and fchmod(fd, mode) system calls.
* syscall might provide 'name' embedded in the message.
*/
struct filp *flp;
struct vnode *vp;
struct vmnt *vmp;
int r, rfd;
mode_t result_mode;
char fullpath[PATH_MAX];
struct lookup resolve;
vir_bytes vname;
size_t vname_length;
mode_t new_mode;
flp = NULL;
vname = (vir_bytes) job_m_in.name;
vname_length = (size_t) job_m_in.name_length;
rfd = job_m_in.fd;
new_mode = (mode_t) job_m_in.mode;
lookup_init(&resolve, fullpath, PATH_NOFLAGS, &vmp, &vp);
resolve.l_vmnt_lock = VMNT_READ;
resolve.l_vnode_lock = VNODE_WRITE;
if (job_call_nr == CHMOD) {
/* Temporarily open the file */
if (copy_name(vname_length, fullpath) != OK) {
/* Direct copy failed, try fetching from user space */
if (fetch_name(vname, vname_length, fullpath) != OK)
return(err_code);
}
if ((vp = eat_path(&resolve, fp)) == NULL) return(err_code);
} else { /* call_nr == FCHMOD */
/* File is already opened; get a pointer to vnode from filp. */
if ((flp = get_filp(rfd, VNODE_WRITE)) == NULL) return(err_code);
vp = flp->filp_vno;
assert(vp);
dup_vnode(vp);
}
assert(vp);
/* Only the owner or the super_user may change the mode of a file.
* No one may change the mode of a file on a read-only file system.
*/
if (vp->v_uid != fp->fp_effuid && fp->fp_effuid != SU_UID)
r = EPERM;
else
r = read_only(vp);
if (r == OK) {
/* Now make the change. Clear setgid bit if file is not in caller's
* group */
if (fp->fp_effuid != SU_UID && vp->v_gid != fp->fp_effgid)
new_mode &= ~I_SET_GID_BIT;
r = req_chmod(vp->v_fs_e, vp->v_inode_nr, new_mode, &result_mode);
if (r == OK)
vp->v_mode = result_mode;
}
if (job_call_nr == CHMOD) {
unlock_vnode(vp);
unlock_vmnt(vmp);
} else { /* FCHMOD */
unlock_filp(flp);
}
put_vnode(vp);
return(r);
}
/*===========================================================================*
* do_chown *
*===========================================================================*/
int do_chown(message *UNUSED(m_out))
{
/* Perform the chown(path, owner, group) and fchmod(fd, owner, group) system
* calls. */
struct filp *flp;
struct vnode *vp;
struct vmnt *vmp;
int r, rfd;
uid_t uid, new_uid;
gid_t gid, new_gid;
mode_t new_mode;
char fullpath[PATH_MAX];
struct lookup resolve;
vir_bytes vname1;
size_t vname1_length;
flp = NULL;
vname1 = (vir_bytes) job_m_in.name1;
vname1_length = (size_t) job_m_in.name1_length;
rfd = job_m_in.fd;
uid = job_m_in.owner;
gid = job_m_in.group;
lookup_init(&resolve, fullpath, PATH_NOFLAGS, &vmp, &vp);
resolve.l_vmnt_lock = VMNT_READ;
resolve.l_vnode_lock = VNODE_WRITE;
if (job_call_nr == CHOWN) {
/* Temporarily open the file. */
if (fetch_name(vname1, vname1_length, fullpath) != OK)
return(err_code);
if ((vp = eat_path(&resolve, fp)) == NULL) return(err_code);
} else { /* call_nr == FCHOWN */
/* File is already opened; get a pointer to the vnode from filp. */
if ((flp = get_filp(rfd, VNODE_WRITE)) == NULL)
return(err_code);
vp = flp->filp_vno;
dup_vnode(vp);
}
r = read_only(vp);
if (r == OK) {
/* FS is R/W. Whether call is allowed depends on ownership, etc. */
/* The super user can do anything, so check permissions only if we're
a regular user. */
if (fp->fp_effuid != SU_UID) {
/* Regular users can only change groups of their own files. */
if (vp->v_uid != fp->fp_effuid) r = EPERM;
if (vp->v_uid != uid) r = EPERM; /* no giving away */
if (fp->fp_effgid != gid) r = EPERM;
}
}
if (r == OK) {
/* Do not change uid/gid if new uid/gid is -1. */
new_uid = (uid == (uid_t)-1 ? vp->v_uid : uid);
new_gid = (gid == (gid_t)-1 ? vp->v_gid : gid);
if (new_uid > UID_MAX || new_gid > GID_MAX)
r = EINVAL;
else if ((r = req_chown(vp->v_fs_e, vp->v_inode_nr, new_uid, new_gid,
&new_mode)) == OK) {
vp->v_uid = new_uid;
vp->v_gid = new_gid;
vp->v_mode = new_mode;
}
}
if (job_call_nr == CHOWN) {
unlock_vnode(vp);
unlock_vmnt(vmp);
} else { /* FCHOWN */
unlock_filp(flp);
}
put_vnode(vp);
return(r);
}
/*===========================================================================*
* do_umask *
*===========================================================================*/
int do_umask(message *UNUSED(m_out))
{
/* Perform the umask(co_mode) system call. */
mode_t complement, new_umask;
new_umask = job_m_in.co_mode;
complement = ~fp->fp_umask; /* set 'r' to complement of old mask */
fp->fp_umask = ~(new_umask & RWX_MODES);
return(complement); /* return complement of old mask */
}
/*===========================================================================*
* do_access *
*===========================================================================*/
int do_access(message *UNUSED(m_out))
{
/* Perform the access(name, mode) system call.
* syscall might provide 'name' embedded in the message.
*/
int r;
struct vnode *vp;
struct vmnt *vmp;
char fullpath[PATH_MAX];
struct lookup resolve;
vir_bytes vname;
size_t vname_length;
mode_t access;
vname = (vir_bytes) job_m_in.name;
vname_length = (size_t) job_m_in.name_length;
access = job_m_in.mode;
lookup_init(&resolve, fullpath, PATH_NOFLAGS, &vmp, &vp);
resolve.l_vmnt_lock = VMNT_READ;
resolve.l_vnode_lock = VNODE_READ;
/* First check to see if the mode is correct. */
if ( (access & ~(R_OK | W_OK | X_OK)) != 0 && access != F_OK)
return(EINVAL);
/* Temporarily open the file. */
if (copy_name(vname_length, fullpath) != OK) {
/* Direct copy failed, try fetching from user space */
if (fetch_name(vname, vname_length, fullpath) != OK)
return(err_code);
}
if ((vp = eat_path(&resolve, fp)) == NULL) return(err_code);
r = forbidden(fp, vp, access);
unlock_vnode(vp);
unlock_vmnt(vmp);
put_vnode(vp);
return(r);
}
/*===========================================================================*
* forbidden *
*===========================================================================*/
int forbidden(struct fproc *rfp, struct vnode *vp, mode_t access_desired)
{
/* Given a pointer to an vnode, 'vp', and the access desired, determine
* if the access is allowed, and if not why not. The routine looks up the
* caller's uid in the 'fproc' table. If access is allowed, OK is returned
* if it is forbidden, EACCES is returned.
*/
register mode_t bits, perm_bits;
uid_t uid;
gid_t gid;
int r, shift;
if (vp->v_uid == (uid_t) -1 || vp->v_gid == (gid_t) -1) return(EACCES);
/* Isolate the relevant rwx bits from the mode. */
bits = vp->v_mode;
uid = (job_call_nr == ACCESS ? rfp->fp_realuid : rfp->fp_effuid);
gid = (job_call_nr == ACCESS ? rfp->fp_realgid : rfp->fp_effgid);
if (uid == SU_UID) {
/* Grant read and write permission. Grant search permission for
* directories. Grant execute permission (for non-directories) if
* and only if one of the 'X' bits is set.
*/
if ( S_ISDIR(bits) || bits & ((X_BIT << 6) | (X_BIT << 3) | X_BIT))
perm_bits = R_BIT | W_BIT | X_BIT;
else
perm_bits = R_BIT | W_BIT;
} else {
if (uid == vp->v_uid) shift = 6; /* owner */
else if (gid == vp->v_gid) shift = 3; /* group */
else if (in_group(fp, vp->v_gid) == OK) shift = 3; /* suppl. groups */
else shift = 0; /* other */
perm_bits = (bits >> shift) & (R_BIT | W_BIT | X_BIT);
}
/* If access desired is not a subset of what is allowed, it is refused. */
r = OK;
if ((perm_bits | access_desired) != perm_bits) r = EACCES;
/* Check to see if someone is trying to write on a file system that is
* mounted read-only.
*/
if (r == OK)
if (access_desired & W_BIT)
r = read_only(vp);
return(r);
}
/*===========================================================================*
* read_only *
*===========================================================================*/
int read_only(vp)
struct vnode *vp; /* ptr to inode whose file sys is to be cked */
{
/* Check to see if the file system on which the inode 'ip' resides is mounted
* read only. If so, return EROFS, else return OK.
*/
assert(vp);
return(vp->v_vmnt && (vp->v_vmnt->m_flags & VMNT_READONLY) ? EROFS : OK);
}