* setpag
*
*/
-#include "../afs/param.h"
-#include "../afs/sysincludes.h"
-#include "../afs/afsincludes.h"
-#include "../afs/afs_stats.h" /* statistics */
-#ifdef AFS_LINUX22_ENV
-#include "../h/smp_lock.h"
+#include <afsconfig.h>
+#include "afs/param.h"
+#ifdef LINUX_KEYRING_SUPPORT
+#include <linux/seq_file.h>
#endif
-static int afs_getgroups(cred_t *cr, gid_t *groups);
-static int afs_setgroups(cred_t **cr, int ngroups, gid_t *gidset, int change_parent);
-/* Only propogate the PAG to the parent process. Unix's propogate to
- * all processes sharing the cred.
- */
-int set_pag_in_parent(int pag, int g0, int g1)
-{
- gid_t *gp = current->p_pptr->groups;
- int ngroups;
+#include "afs/sysincludes.h"
+#include "afsincludes.h"
+#include "afs/afs_stats.h" /* statistics */
+#include "afs/nfsclient.h"
+#include "osi_compat.h"
+
+#ifdef AFS_LINUX26_ONEGROUP_ENV
+# define NUMPAGGROUPS 1
+
+static afs_uint32
+afs_linux_pag_from_groups(struct group_info *group_info) {
+ afs_uint32 g0 = 0;
+ afs_uint32 i;
+
+ if (group_info->ngroups < NUMPAGGROUPS)
+ return NOPAG;
+
+ for (i = 0; (i < group_info->ngroups &&
+ (g0 = GROUP_AT(group_info, i)) != (gid_t) NOGROUP); i++) {
+ if (((g0 >> 24) & 0xff) == 'A')
+ return g0;
+ }
+ return NOPAG;
+}
+
+static inline void
+afs_linux_pag_to_groups(afs_uint32 newpag,
+ struct group_info *old, struct group_info **new) {
+ int need_space = 0;
int i;
+ int j;
-
- ngroups = current->p_pptr->ngroups;
- gp = current->p_pptr->groups;
+ if (afs_linux_pag_from_groups(old) == NOPAG)
+ need_space = NUMPAGGROUPS;
+ *new = groups_alloc(old->ngroups + need_space);
- if (afs_get_pag_from_groups(gp[0], gp[1]) == NOPAG) {
- /* We will have to shift grouplist to make room for pag */
- if (ngroups + 2 > NGROUPS) {
- return EINVAL;
+ for (i = 0, j = 0; i < old->ngroups; ++i) {
+ int ths = GROUP_AT(old, i);
+ int last = i > 0 ? GROUP_AT(old, i-1) : 0;
+ if ((ths >> 24) == 'A')
+ continue;
+ if (last <= newpag && ths > newpag) {
+ GROUP_AT(*new, j) = newpag;
+ j++;
}
- for (i = ngroups-1; i >= 0; i--) {
- gp[i+2] = gp[i];
- }
- ngroups += 2;
+ GROUP_AT(*new, j) = ths;
+ j++;
+ }
+ if (j != i + need_space)
+ GROUP_AT(*new, j) = newpag;
+}
+
+#else
+# define NUMPAGGROUPS 2
+
+static inline afs_uint32
+afs_linux_pag_from_groups(struct group_info *group_info) {
+
+ if (group_info->ngroups < NUMPAGGROUPS)
+ return NOPAG;
+
+ return afs_get_pag_from_groups(GROUP_AT(group_info, 0), GROUP_AT(group_info, 1));
+}
+
+static inline void
+afs_linux_pag_to_groups(afs_uint32 newpag,
+ struct group_info *old, struct group_info **new) {
+ int need_space = 0;
+ int i;
+ gid_t g0;
+ gid_t g1;
+
+ if (afs_linux_pag_from_groups(old) == NOPAG)
+ need_space = NUMPAGGGROUPS;
+
+ *new = groups_alloc(old->ngroups + need_space);
+
+ for (i = 0; i < old->ngroups; ++i)
+ GROUP_AT(new, i + need_space) = GROUP_AT(old, i);
+
+ afs_get_groups_from_pag(newpag, &g0, g1);
+ GROUP_AT(new, 0) = g0;
+ GROUP_AT(new, 1) = g1;
+}
+#endif
+
+afs_int32
+osi_get_group_pag(afs_ucred_t *cred) {
+ return afs_linux_pag_from_groups(afs_cr_group_info(cred));
+}
+
+
+static int
+afs_setgroups(cred_t **cr, struct group_info *group_info, int change_parent)
+{
+ struct group_info *old_info;
+
+ AFS_STATCNT(afs_setgroups);
+
+ old_info = afs_cr_group_info(*cr);
+ get_group_info(group_info);
+ afs_set_cr_group_info(*cr, group_info);
+ put_group_info(old_info);
+
+ crset(*cr);
+
+#if defined(STRUCT_TASK_STRUCT_HAS_PARENT) && !defined(STRUCT_TASK_STRUCT_HAS_CRED)
+ if (change_parent) {
+ old_info = current->parent->group_info;
+ get_group_info(group_info);
+ current->parent->group_info = group_info;
+ put_group_info(old_info);
+ }
+#endif
+
+ return (0);
+}
+
+int
+__setpag(cred_t **cr, afs_uint32 pagvalue, afs_uint32 *newpag,
+ int change_parent, struct group_info **old_groups)
+{
+ struct group_info *group_info;
+ struct group_info *tmp;
+
+ get_group_info(afs_cr_group_info(*cr));
+ group_info = afs_cr_group_info(*cr);
+
+ *newpag = (pagvalue == -1 ? genpag() : pagvalue);
+ afs_linux_pag_to_groups(*newpag, group_info, &tmp);
+
+ if (old_groups) {
+ *old_groups = group_info;
+ } else {
+ put_group_info(group_info);
+ group_info = NULL;
}
- gp[0] = g0;
- gp[1] = g1;
- if (ngroups < NGROUPS)
- gp[ngroups] = NOGROUP;
- current->p_pptr->ngroups = ngroups;
+ afs_setgroups(cr, tmp, change_parent);
+
+ put_group_info(tmp);
+
return 0;
}
-int setpag(cred_t **cr, afs_uint32 pagvalue, afs_uint32 *newpag, int change_parent)
+#ifdef LINUX_KEYRING_SUPPORT
+extern struct key_type key_type_keyring __attribute__((weak));
+static struct key_type *__key_type_keyring = &key_type_keyring;
+
+/* install_session_keyring returns negative error values */
+static int
+install_session_keyring(struct key *keyring)
{
- gid_t *gidset;
- afs_int32 ngroups, code = 0;
- int j;
+ struct key *old;
+ char desc[20];
+ int code = -EINVAL;
+ unsigned long flags;
+
+ if (!__key_type_keyring)
+ return code;
+
+ if (!keyring) {
+
+ /* create an empty session keyring */
+ sprintf(desc, "_ses.%u", current->tgid);
+
+ /* if we're root, don't count the keyring against our quota. This
+ * avoids starvation issues when dealing with PAM modules that always
+ * setpag() as root */
+ if (current_uid() == 0)
+ flags = KEY_ALLOC_NOT_IN_QUOTA;
+ else
+ flags = KEY_ALLOC_IN_QUOTA;
+
+ keyring = afs_linux_key_alloc(
+ __key_type_keyring, desc,
+ current_uid(), current_gid(),
+ (KEY_POS_ALL & ~KEY_POS_SETATTR) | KEY_USR_ALL,
+ flags);
+
+ if (IS_ERR(keyring)) {
+ code = PTR_ERR(keyring);
+ goto out;
+ }
+ }
+
+ code = key_instantiate_and_link(keyring, NULL, 0, NULL, NULL);
+ if (code < 0) {
+ key_put(keyring);
+ goto out;
+ }
+
+ /* install the keyring */
+ old = afs_set_session_keyring(keyring);
+ if (old)
+ key_put(old);
+
+out:
+ return code;
+}
+#endif /* LINUX_KEYRING_SUPPORT */
+
+/* Error codes from setpag must be positive, otherwise they don't
+ * make it back into userspace properly. Error codes from the
+ * Linux keyring utilities, and from install_session_keyring()
+ * are negative. So we need to be careful to convert them correctly
+ * here
+ */
+int
+setpag(cred_t **cr, afs_uint32 pagvalue, afs_uint32 *newpag,
+ int change_parent)
+{
+ int code;
+ struct group_info *old_groups = NULL;
AFS_STATCNT(setpag);
- gidset = (gid_t *) osi_Alloc(NGROUPS*sizeof(gidset[0]));
- ngroups = afs_getgroups(*cr, gidset);
+ code = __setpag(cr, pagvalue, newpag, change_parent, &old_groups);
+
+#ifdef LINUX_KEYRING_SUPPORT
+ if (code == 0 && afs_cr_rgid(*cr) != NFSXLATOR_CRED) {
+ code = install_session_keyring(NULL);
+ if (code == 0 && current_session_keyring()) {
+ struct key *key;
+ key_perm_t perm;
+
+ perm = KEY_POS_VIEW | KEY_POS_SEARCH;
+ perm |= KEY_USR_VIEW | KEY_USR_SEARCH;
- if (afs_get_pag_from_groups(gidset[0], gidset[1]) == NOPAG) {
- /* We will have to shift grouplist to make room for pag */
- if (ngroups + 2 > NGROUPS) {
- osi_Free((char *)gidset, NGROUPS*sizeof(int));
- return EINVAL;
+ key = afs_linux_key_alloc(&key_type_afs_pag, "_pag", 0, 0, perm, KEY_ALLOC_NOT_IN_QUOTA);
+
+ if (!IS_ERR(key)) {
+ key_instantiate_and_link(key, (void *) newpag, sizeof(afs_uint32),
+ current_session_keyring(), NULL);
+ key_put(key);
+ } else {
+ code = PTR_ERR(key);
+ }
}
- for (j = ngroups - 1; j >= 0; j--) {
- gidset[j+2] = gidset[j];
- }
- ngroups += 2;
+ if (code)
+ code = -code;
}
- *newpag = (pagvalue == -1 ? genpag(): pagvalue);
- afs_get_groups_from_pag(*newpag, &gidset[0], &gidset[1]);
- code = afs_setgroups(cr, ngroups, gidset, change_parent);
+#endif /* LINUX_KEYRING_SUPPORT */
- /* If change_parent is set, then we should set the pag in the parent as
- * well.
- */
- if (change_parent && !code) {
- code = set_pag_in_parent(*newpag, gidset[0], gidset[1]);
+ if (code) {
+ if (old_groups) {
+ afs_setgroups(cr, old_groups, change_parent);
+ put_group_info(old_groups);
+ old_groups = NULL;
+ }
+ if (*newpag > -1) {
+ afs_MarkUserExpired(*newpag);
+ *newpag = -1;
+ }
}
- osi_Free((char *)gidset, NGROUPS*sizeof(int));
return code;
}
/* Intercept the standard system call. */
-extern int (*sys_setgroupsp)(int gidsetsize, gid_t *grouplist);
-asmlinkage int afs_xsetgroups(int gidsetsize, gid_t *grouplist)
+extern asmlinkage long (*sys_setgroupsp) (int gidsetsize, gid_t * grouplist);
+asmlinkage long
+afs_xsetgroups(int gidsetsize, gid_t * grouplist)
{
- int code;
+ long code;
cred_t *cr = crref();
afs_uint32 junk;
int old_pag;
- lock_kernel();
old_pag = PagInCred(cr);
crfree(cr);
- unlock_kernel();
- code = (*sys_setgroupsp)(gidsetsize, grouplist);
+ code = (*sys_setgroupsp) (gidsetsize, grouplist);
if (code) {
return code;
}
- lock_kernel();
cr = crref();
if (old_pag != NOPAG && PagInCred(cr) == NOPAG) {
/* re-install old pag if there's room. */
- code = setpag(&cr, old_pag, &junk, 0);
+ code = __setpag(&cr, old_pag, &junk, 0, NULL);
}
crfree(cr);
- unlock_kernel();
- return code;
+ /* Linux syscall ABI returns errno as negative */
+ return (-code);
}
-#if defined(AFS_LINUX24_ENV)
/* Intercept the standard uid32 system call. */
-extern int (*sys_setgroups32p)(int gidsetsize, gid_t *grouplist);
-asmlinkage int afs_xsetgroups32(int gidsetsize, gid_t *grouplist)
+extern asmlinkage int (*sys_setgroups32p) (int gidsetsize,
+ __kernel_gid32_t * grouplist);
+asmlinkage long
+afs_xsetgroups32(int gidsetsize, gid_t * grouplist)
{
- int code;
+ long code;
+ cred_t *cr = crref();
+ afs_uint32 junk;
+ int old_pag;
+
+ old_pag = PagInCred(cr);
+ crfree(cr);
+
+ code = (*sys_setgroups32p) (gidsetsize, grouplist);
+
+ if (code) {
+ return code;
+ }
+
+ cr = crref();
+ if (old_pag != NOPAG && PagInCred(cr) == NOPAG) {
+ /* re-install old pag if there's room. */
+ code = __setpag(&cr, old_pag, &junk, 0, NULL);
+ }
+ crfree(cr);
+
+ /* Linux syscall ABI returns errno as negative */
+ return (-code);
+}
+
+#if defined(AFS_PPC64_LINUX20_ENV)
+/* Intercept the uid16 system call as used by 32bit programs. */
+extern asmlinkage long (*sys32_setgroupsp)(int gidsetsize, gid_t *grouplist);
+asmlinkage long afs32_xsetgroups(int gidsetsize, gid_t *grouplist)
+{
+ long code;
+ cred_t *cr = crref();
+ afs_uint32 junk;
+ int old_pag;
+
+ old_pag = PagInCred(cr);
+ crfree(cr);
+
+ code = (*sys32_setgroupsp)(gidsetsize, grouplist);
+ if (code) {
+ return code;
+ }
+
+ cr = crref();
+ if (old_pag != NOPAG && PagInCred(cr) == NOPAG) {
+ /* re-install old pag if there's room. */
+ code = __setpag(&cr, old_pag, &junk, 0, NULL);
+ }
+ crfree(cr);
+
+ /* Linux syscall ABI returns errno as negative */
+ return (-code);
+}
+#endif
+
+#if defined(AFS_SPARC64_LINUX20_ENV) || defined(AFS_AMD64_LINUX20_ENV)
+/* Intercept the uid16 system call as used by 32bit programs. */
+#ifdef AFS_AMD64_LINUX20_ENV
+extern asmlinkage long (*sys32_setgroupsp) (int gidsetsize, u16 * grouplist);
+#endif /* AFS_AMD64_LINUX20_ENV */
+#ifdef AFS_SPARC64_LINUX26_ENV
+extern asmlinkage int (*sys32_setgroupsp) (int gidsetsize,
+ __kernel_gid32_t * grouplist);
+#endif /* AFS_SPARC64_LINUX26_ENV */
+asmlinkage long
+afs32_xsetgroups(int gidsetsize, u16 * grouplist)
+{
+ long code;
cred_t *cr = crref();
afs_uint32 junk;
int old_pag;
- lock_kernel();
old_pag = PagInCred(cr);
crfree(cr);
- unlock_kernel();
+
+ code = (*sys32_setgroupsp) (gidsetsize, grouplist);
+ if (code) {
+ return code;
+ }
+
+ cr = crref();
+ if (old_pag != NOPAG && PagInCred(cr) == NOPAG) {
+ /* re-install old pag if there's room. */
+ code = __setpag(&cr, old_pag, &junk, 0, NULL);
+ }
+ crfree(cr);
+
+ /* Linux syscall ABI returns errno as negative */
+ return (-code);
+}
+
+/* Intercept the uid32 system call as used by 32bit programs. */
+#ifdef AFS_AMD64_LINUX20_ENV
+extern asmlinkage long (*sys32_setgroups32p) (int gidsetsize, gid_t * grouplist);
+#endif /* AFS_AMD64_LINUX20_ENV */
+#ifdef AFS_SPARC64_LINUX26_ENV
+extern asmlinkage int (*sys32_setgroups32p) (int gidsetsize,
+ __kernel_gid32_t * grouplist);
+#endif /* AFS_SPARC64_LINUX26_ENV */
+asmlinkage long
+afs32_xsetgroups32(int gidsetsize, gid_t * grouplist)
+{
+ long code;
+ cred_t *cr = crref();
+ afs_uint32 junk;
+ int old_pag;
+
+ old_pag = PagInCred(cr);
+ crfree(cr);
- code = (*sys_setgroups32p)(gidsetsize, grouplist);
+ code = (*sys32_setgroups32p) (gidsetsize, grouplist);
if (code) {
return code;
}
- lock_kernel();
cr = crref();
if (old_pag != NOPAG && PagInCred(cr) == NOPAG) {
/* re-install old pag if there's room. */
- code = setpag(&cr, old_pag, &junk, 0);
+ code = __setpag(&cr, old_pag, &junk, 0, NULL);
}
crfree(cr);
- unlock_kernel();
- return code;
+ /* Linux syscall ABI returns errno as negative */
+ return (-code);
}
#endif
-#if defined(AFS_SPARC64_LINUX20_ENV)
-asmlinkage int afs32_xsetgroups(int gidsetsize, __kernel_gid_t32 *grouplist)
-{
- gid_t gl[NGROUPS];
- int ret, i;
- mm_segment_t old_fs = get_fs ();
-
- if ((unsigned) gidsetsize > NGROUPS)
- return -EINVAL;
- for (i = 0; i < gidsetsize; i++, grouplist++)
- if (__get_user (gl[i], grouplist))
- return -EFAULT;
- set_fs (KERNEL_DS);
- ret = afs_xsetgroups(gidsetsize, gl);
- set_fs (old_fs);
- return ret;
-}
-#ifdef AFS_LINUX24_ENV
-asmlinkage int afs32_xsetgroups32(int gidsetsize, __kernel_gid_t32 *grouplist)
-{
- gid_t gl[NGROUPS];
- int ret, i;
- mm_segment_t old_fs = get_fs ();
-
- if ((unsigned) gidsetsize > NGROUPS)
- return -EINVAL;
- for (i = 0; i < gidsetsize; i++, grouplist++)
- if (__get_user (gl[i], grouplist))
- return -EFAULT;
- set_fs (KERNEL_DS);
- ret = afs_xsetgroups32(gidsetsize, gl);
- set_fs (old_fs);
- return ret;
+
+
+#ifdef LINUX_KEYRING_SUPPORT
+static void afs_pag_describe(const struct key *key, struct seq_file *m)
+{
+ seq_puts(m, key->description);
+
+ seq_printf(m, ": %u", key->datalen);
}
+
+#if defined(STRUCT_KEY_TYPE_HAS_PREPARSE)
+static int afs_pag_instantiate(struct key *key, struct key_preparsed_payload *prep)
+#else
+static int afs_pag_instantiate(struct key *key, const void *data, size_t datalen)
+#endif
+{
+ int code;
+ afs_uint32 *userpag, pag = NOPAG;
+
+ if (key->uid != 0 || key->gid != 0)
+ return -EPERM;
+
+ code = -EINVAL;
+ get_group_info(current_group_info());
+
+#if defined(STRUCT_KEY_TYPE_HAS_PREPARSE)
+ if (prep->datalen != sizeof(afs_uint32) || !prep->data)
+#else
+ if (datalen != sizeof(afs_uint32) || !data)
#endif
+ goto error;
+
+ /* ensure key being set matches current pag */
+ pag = afs_linux_pag_from_groups(current_group_info());
+
+ if (pag == NOPAG)
+ goto error;
+
+#if defined(STRUCT_KEY_TYPE_HAS_PREPARSE)
+ userpag = (afs_uint32 *)prep->data;
+#else
+ userpag = (afs_uint32 *)data;
#endif
+ if (*userpag != pag)
+ goto error;
+
+ key->payload.value = (unsigned long) *userpag;
+ key->datalen = sizeof(afs_uint32);
+ code = 0;
+
+error:
+ put_group_info(current_group_info());
+ return code;
+}
-static int afs_setgroups(cred_t **cr, int ngroups, gid_t *gidset, int change_parent)
+static int afs_pag_match(const struct key *key, const void *description)
{
- int ngrps;
- int i;
- gid_t *gp;
+ return strcmp(key->description, description) == 0;
+}
- AFS_STATCNT(afs_setgroups);
+static void afs_pag_destroy(struct key *key)
+{
+ afs_uint32 pag = key->payload.value;
+ int locked = ISAFS_GLOCK();
+
+ if (!locked)
+ AFS_GLOCK();
- if (ngroups > NGROUPS)
- return EINVAL;
+ afs_MarkUserExpired(pag);
- gp = (*cr)->cr_groups;
- if (ngroups < NGROUPS)
- gp[ngroups] = (gid_t)NOGROUP;
+ if (!locked)
+ AFS_GUNLOCK();
+}
- for (i = ngroups; i > 0; i--) {
- *gp++ = *gidset++;
+struct key_type key_type_afs_pag =
+{
+ .name = "afs_pag",
+ .describe = afs_pag_describe,
+ .instantiate = afs_pag_instantiate,
+ .match = afs_pag_match,
+ .destroy = afs_pag_destroy,
+};
+
+#ifdef EXPORTED_TASKLIST_LOCK
+extern rwlock_t tasklist_lock __attribute__((weak));
+#endif
+
+void osi_keyring_init(void)
+{
+#if !defined(EXPORTED_KEY_TYPE_KEYRING)
+ struct task_struct *p;
+
+ /* If we can't lock the tasklist, either with its explicit lock,
+ * or by using the RCU lock, then we can't safely work out the
+ * type of a keyring. So, we have to rely on the weak reference.
+ * If that's not available, then keyring based PAGs won't work.
+ */
+
+#if defined(EXPORTED_TASKLIST_LOCK) || (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) && defined(HAVE_LINUX_RCU_READ_LOCK))
+ if (__key_type_keyring == NULL) {
+# ifdef EXPORTED_TASKLIST_LOCK
+ if (&tasklist_lock)
+ read_lock(&tasklist_lock);
+# endif
+# if (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) && defined(HAVE_LINUX_RCU_READ_LOCK))
+# ifdef EXPORTED_TASKLIST_LOCK
+ else
+# endif
+ rcu_read_lock();
+# endif
+#if defined(HAVE_LINUX_FIND_TASK_BY_PID)
+ p = find_task_by_pid(1);
+#else
+ p = find_task_by_vpid(1);
+#endif
+ if (p && task_user(p)->session_keyring)
+ __key_type_keyring = task_user(p)->session_keyring->type;
+# ifdef EXPORTED_TASKLIST_LOCK
+ if (&tasklist_lock)
+ read_unlock(&tasklist_lock);
+# endif
+# if (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) && defined(HAVE_LINUX_RCU_READ_LOCK))
+# ifdef EXPORTED_TASKLIST_LOCK
+ else
+# endif
+ rcu_read_unlock();
+# endif
}
+#endif
+#endif
- (*cr)->cr_ngroups = ngroups;
- crset(*cr);
- return (0);
+ register_key_type(&key_type_afs_pag);
}
-/* Returns number of groups. And we trust groups to be large enough to
- * hold all the groups.
- */
-static int afs_getgroups(cred_t *cr, gid_t *groups)
+void osi_keyring_shutdown(void)
{
- int i;
- gid_t *gp = cr->cr_groups;
- int n = cr->cr_ngroups;
- AFS_STATCNT(afs_getgroups);
+ unregister_key_type(&key_type_afs_pag);
+}
- for (i = 0; (i < n) && (*gp != (gid_t)NOGROUP); i++) {
- *groups++ = *gp++;
+afs_int32
+osi_get_keyring_pag(afs_ucred_t *cred)
+{
+ struct key *key;
+ afs_uint32 newpag;
+ afs_int32 keyring_pag = NOPAG;
+
+ if (afs_cr_rgid(cred) != NFSXLATOR_CRED) {
+ key = afs_linux_search_keyring(cred, &key_type_afs_pag);
+
+ if (!IS_ERR(key)) {
+ if (key_validate(key) == 0 && key->uid == 0) { /* also verify in the session keyring? */
+ keyring_pag = key->payload.value;
+ /* Only set PAG in groups if needed,
+ * and the creds are from the current process */
+ if (afs_linux_cred_is_current(cred) &&
+ ((keyring_pag >> 24) & 0xff) == 'A' &&
+ keyring_pag != afs_linux_pag_from_groups(current_group_info())) {
+ __setpag(&cred, keyring_pag, &newpag, 0, NULL);
+ }
+ }
+ key_put(key);
+ }
}
- return i;
+ return keyring_pag;
}
+#else
+void osi_keyring_init(void)
+{
+ return;
+}
+
+void osi_keyring_shutdown(void)
+{
+ return;
+}
+#endif
git://git.openafs.org / openafs.git / blobdiff