shadow/libmisc/find_new_uid.c

499 lines
14 KiB
C
Raw Normal View History

/*
* Copyright (c) 1991 - 1994, Julianne Frances Haugh
* Copyright (c) 2008 - 2011, Nicolas François
Make the group and user allocation more effective. Previously, the allocation was optimized for an outdated deployment style (that of /etc/group alongside nss_db). The issue here is that this results in extremely poor performance when using SSSD, Winbind or nss_ldap. There were actually two serious bugs here that have been addressed: 1) Running getgrent() loops won't work in most SSSD or Winbind environments, as full group enumeration is disabled by default. This could easily result in auto-allocating a group that was already in use. (This might result in a security issue as well, if the shared GID is a privileged group). 2) For system groups, the loop was always iterating through the complete SYS_GID_MIN->SYS_GID_MAX range. On SSSD and Winbind, this means hundreds of round-trips to LDAP (unless the GIDs were specifically configured to be ignored by the SSSD or winbindd). To a user with a slow connection to their LDAP server, this would appear as if groupadd -r was hung. (Though it would eventually complete). This patch changes the algorithm to be more favorable for LDAP environments, at the expense of some performance when using nss_db. Given that the DB is a local service, this should have a negligible effect from a user's perspective. With the new algorithm, we simply first iterate through all entries in the local database with gr_next(), recording the IDs that are in use. We then start from the highest presumed-available entry and call getgrgid() to see if it is available. We continue this until we come to the first unused GID. We then select that and return it. If we make it through all the remaining IDs without finding a free one, we start over from the beginning of the range and try to find room in one of the gaps in the range. The patch was originally written by Stephen Gallagher and applied identically also to the user allocation by Tomáš Mráz. Signed-off-by: Serge Hallyn <serge@hallyn.com>
2016-11-15 20:13:54 +05:30
* Copyright (c) 2014, Red Hat, Inc.
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. The name of the copyright holders or contributors may not be used to
* endorse or promote products derived from this software without
* specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
* ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
* LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A
* PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
* HOLDERS OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
* SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
* DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
* THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
* OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
#include <config.h>
#include <assert.h>
#include <stdio.h>
#include <errno.h>
#include "prototypes.h"
#include "pwio.h"
#include "getdef.h"
Make the group and user allocation more effective. Previously, the allocation was optimized for an outdated deployment style (that of /etc/group alongside nss_db). The issue here is that this results in extremely poor performance when using SSSD, Winbind or nss_ldap. There were actually two serious bugs here that have been addressed: 1) Running getgrent() loops won't work in most SSSD or Winbind environments, as full group enumeration is disabled by default. This could easily result in auto-allocating a group that was already in use. (This might result in a security issue as well, if the shared GID is a privileged group). 2) For system groups, the loop was always iterating through the complete SYS_GID_MIN->SYS_GID_MAX range. On SSSD and Winbind, this means hundreds of round-trips to LDAP (unless the GIDs were specifically configured to be ignored by the SSSD or winbindd). To a user with a slow connection to their LDAP server, this would appear as if groupadd -r was hung. (Though it would eventually complete). This patch changes the algorithm to be more favorable for LDAP environments, at the expense of some performance when using nss_db. Given that the DB is a local service, this should have a negligible effect from a user's perspective. With the new algorithm, we simply first iterate through all entries in the local database with gr_next(), recording the IDs that are in use. We then start from the highest presumed-available entry and call getgrgid() to see if it is available. We continue this until we come to the first unused GID. We then select that and return it. If we make it through all the remaining IDs without finding a free one, we start over from the beginning of the range and try to find room in one of the gaps in the range. The patch was originally written by Stephen Gallagher and applied identically also to the user allocation by Tomáš Mráz. Signed-off-by: Serge Hallyn <serge@hallyn.com>
2016-11-15 20:13:54 +05:30
/*
* get_ranges - Get the minimum and maximum ID ranges for the search
*
* This function will return the minimum and maximum ranges for IDs
*
* 0: The function completed successfully
* EINVAL: The provided ranges are impossible (such as maximum < minimum)
*
* preferred_min: The special-case minimum value for a specifically-
* requested ID, which may be lower than the standard min_id
*/
static int get_ranges (bool sys_user, uid_t *min_id, uid_t *max_id,
uid_t *preferred_min)
{
uid_t uid_def_max = 0;
if (sys_user) {
/* System users */
/* A requested ID is allowed to be below the autoselect range */
*preferred_min = (uid_t) 1;
/* Get the minimum ID range from login.defs or default to 101 */
*min_id = (uid_t) getdef_ulong ("SYS_UID_MIN", 101UL);
/*
* If SYS_UID_MAX is unspecified, we should assume it to be one
* less than the UID_MIN (which is reserved for non-system accounts)
*/
uid_def_max = (uid_t) getdef_ulong ("UID_MIN", 1000UL) - 1;
*max_id = (uid_t) getdef_ulong ("SYS_UID_MAX",
(unsigned long) uid_def_max);
/* Check that the ranges make sense */
if (*max_id < *min_id) {
(void) fprintf (shadow_logfd,
Make the group and user allocation more effective. Previously, the allocation was optimized for an outdated deployment style (that of /etc/group alongside nss_db). The issue here is that this results in extremely poor performance when using SSSD, Winbind or nss_ldap. There were actually two serious bugs here that have been addressed: 1) Running getgrent() loops won't work in most SSSD or Winbind environments, as full group enumeration is disabled by default. This could easily result in auto-allocating a group that was already in use. (This might result in a security issue as well, if the shared GID is a privileged group). 2) For system groups, the loop was always iterating through the complete SYS_GID_MIN->SYS_GID_MAX range. On SSSD and Winbind, this means hundreds of round-trips to LDAP (unless the GIDs were specifically configured to be ignored by the SSSD or winbindd). To a user with a slow connection to their LDAP server, this would appear as if groupadd -r was hung. (Though it would eventually complete). This patch changes the algorithm to be more favorable for LDAP environments, at the expense of some performance when using nss_db. Given that the DB is a local service, this should have a negligible effect from a user's perspective. With the new algorithm, we simply first iterate through all entries in the local database with gr_next(), recording the IDs that are in use. We then start from the highest presumed-available entry and call getgrgid() to see if it is available. We continue this until we come to the first unused GID. We then select that and return it. If we make it through all the remaining IDs without finding a free one, we start over from the beginning of the range and try to find room in one of the gaps in the range. The patch was originally written by Stephen Gallagher and applied identically also to the user allocation by Tomáš Mráz. Signed-off-by: Serge Hallyn <serge@hallyn.com>
2016-11-15 20:13:54 +05:30
_("%s: Invalid configuration: SYS_UID_MIN (%lu), "
"UID_MIN (%lu), SYS_UID_MAX (%lu)\n"),
Prog, (unsigned long) *min_id,
getdef_ulong ("UID_MIN", 1000UL),
(unsigned long) *max_id);
return EINVAL;
}
} else {
/* Non-system users */
/* Get the values from login.defs or use reasonable defaults */
*min_id = (uid_t) getdef_ulong ("UID_MIN", 1000UL);
*max_id = (uid_t) getdef_ulong ("UID_MAX", 60000UL);
/*
* The preferred minimum should match the standard ID minimum
* for non-system users.
*/
*preferred_min = *min_id;
/* Check that the ranges make sense */
if (*max_id < *min_id) {
(void) fprintf (shadow_logfd,
Make the group and user allocation more effective. Previously, the allocation was optimized for an outdated deployment style (that of /etc/group alongside nss_db). The issue here is that this results in extremely poor performance when using SSSD, Winbind or nss_ldap. There were actually two serious bugs here that have been addressed: 1) Running getgrent() loops won't work in most SSSD or Winbind environments, as full group enumeration is disabled by default. This could easily result in auto-allocating a group that was already in use. (This might result in a security issue as well, if the shared GID is a privileged group). 2) For system groups, the loop was always iterating through the complete SYS_GID_MIN->SYS_GID_MAX range. On SSSD and Winbind, this means hundreds of round-trips to LDAP (unless the GIDs were specifically configured to be ignored by the SSSD or winbindd). To a user with a slow connection to their LDAP server, this would appear as if groupadd -r was hung. (Though it would eventually complete). This patch changes the algorithm to be more favorable for LDAP environments, at the expense of some performance when using nss_db. Given that the DB is a local service, this should have a negligible effect from a user's perspective. With the new algorithm, we simply first iterate through all entries in the local database with gr_next(), recording the IDs that are in use. We then start from the highest presumed-available entry and call getgrgid() to see if it is available. We continue this until we come to the first unused GID. We then select that and return it. If we make it through all the remaining IDs without finding a free one, we start over from the beginning of the range and try to find room in one of the gaps in the range. The patch was originally written by Stephen Gallagher and applied identically also to the user allocation by Tomáš Mráz. Signed-off-by: Serge Hallyn <serge@hallyn.com>
2016-11-15 20:13:54 +05:30
_("%s: Invalid configuration: UID_MIN (%lu), "
"UID_MAX (%lu)\n"),
Prog, (unsigned long) *min_id,
(unsigned long) *max_id);
return EINVAL;
}
}
return 0;
}
/*
* check_uid - See if the requested UID is available
*
* On success, return 0
* If the ID is in use, return EEXIST
* If the ID is outside the range, return ERANGE
* In other cases, return errno from getpwuid()
*/
static int check_uid(const uid_t uid,
const uid_t uid_min,
const uid_t uid_max,
bool *used_uids)
{
/* First test that the preferred ID is in the range */
if (uid < uid_min || uid > uid_max) {
return ERANGE;
}
/*
* Check whether we already detected this UID
* using the pw_next() loop
*/
if (used_uids != NULL && used_uids[uid]) {
return EEXIST;
}
/* Check if the UID exists according to NSS */
errno = 0;
2016-05-15 19:19:39 +05:30
if (prefix_getpwuid(uid) != NULL) {
Make the group and user allocation more effective. Previously, the allocation was optimized for an outdated deployment style (that of /etc/group alongside nss_db). The issue here is that this results in extremely poor performance when using SSSD, Winbind or nss_ldap. There were actually two serious bugs here that have been addressed: 1) Running getgrent() loops won't work in most SSSD or Winbind environments, as full group enumeration is disabled by default. This could easily result in auto-allocating a group that was already in use. (This might result in a security issue as well, if the shared GID is a privileged group). 2) For system groups, the loop was always iterating through the complete SYS_GID_MIN->SYS_GID_MAX range. On SSSD and Winbind, this means hundreds of round-trips to LDAP (unless the GIDs were specifically configured to be ignored by the SSSD or winbindd). To a user with a slow connection to their LDAP server, this would appear as if groupadd -r was hung. (Though it would eventually complete). This patch changes the algorithm to be more favorable for LDAP environments, at the expense of some performance when using nss_db. Given that the DB is a local service, this should have a negligible effect from a user's perspective. With the new algorithm, we simply first iterate through all entries in the local database with gr_next(), recording the IDs that are in use. We then start from the highest presumed-available entry and call getgrgid() to see if it is available. We continue this until we come to the first unused GID. We then select that and return it. If we make it through all the remaining IDs without finding a free one, we start over from the beginning of the range and try to find room in one of the gaps in the range. The patch was originally written by Stephen Gallagher and applied identically also to the user allocation by Tomáš Mráz. Signed-off-by: Serge Hallyn <serge@hallyn.com>
2016-11-15 20:13:54 +05:30
return EEXIST;
} else {
/* getpwuid() was NULL
* we have to ignore errors as temporary
* failures of remote user identity services
* would completely block user/group creation
*/
}
/* If we've made it here, the UID must be available */
return 0;
}
/*
* find_new_uid - Find a new unused UID.
*
* If successful, find_new_uid provides an unused user ID in the
* [UID_MIN:UID_MAX] range.
* This ID should be higher than all the used UID, but if not possible,
* the lowest unused ID in the range will be returned.
*
* Return 0 on success, -1 if no unused UIDs are available.
*/
Make the group and user allocation more effective. Previously, the allocation was optimized for an outdated deployment style (that of /etc/group alongside nss_db). The issue here is that this results in extremely poor performance when using SSSD, Winbind or nss_ldap. There were actually two serious bugs here that have been addressed: 1) Running getgrent() loops won't work in most SSSD or Winbind environments, as full group enumeration is disabled by default. This could easily result in auto-allocating a group that was already in use. (This might result in a security issue as well, if the shared GID is a privileged group). 2) For system groups, the loop was always iterating through the complete SYS_GID_MIN->SYS_GID_MAX range. On SSSD and Winbind, this means hundreds of round-trips to LDAP (unless the GIDs were specifically configured to be ignored by the SSSD or winbindd). To a user with a slow connection to their LDAP server, this would appear as if groupadd -r was hung. (Though it would eventually complete). This patch changes the algorithm to be more favorable for LDAP environments, at the expense of some performance when using nss_db. Given that the DB is a local service, this should have a negligible effect from a user's perspective. With the new algorithm, we simply first iterate through all entries in the local database with gr_next(), recording the IDs that are in use. We then start from the highest presumed-available entry and call getgrgid() to see if it is available. We continue this until we come to the first unused GID. We then select that and return it. If we make it through all the remaining IDs without finding a free one, we start over from the beginning of the range and try to find room in one of the gaps in the range. The patch was originally written by Stephen Gallagher and applied identically also to the user allocation by Tomáš Mráz. Signed-off-by: Serge Hallyn <serge@hallyn.com>
2016-11-15 20:13:54 +05:30
int find_new_uid(bool sys_user,
uid_t *uid,
/*@null@*/uid_t const *preferred_uid)
{
bool *used_uids;
Make the group and user allocation more effective. Previously, the allocation was optimized for an outdated deployment style (that of /etc/group alongside nss_db). The issue here is that this results in extremely poor performance when using SSSD, Winbind or nss_ldap. There were actually two serious bugs here that have been addressed: 1) Running getgrent() loops won't work in most SSSD or Winbind environments, as full group enumeration is disabled by default. This could easily result in auto-allocating a group that was already in use. (This might result in a security issue as well, if the shared GID is a privileged group). 2) For system groups, the loop was always iterating through the complete SYS_GID_MIN->SYS_GID_MAX range. On SSSD and Winbind, this means hundreds of round-trips to LDAP (unless the GIDs were specifically configured to be ignored by the SSSD or winbindd). To a user with a slow connection to their LDAP server, this would appear as if groupadd -r was hung. (Though it would eventually complete). This patch changes the algorithm to be more favorable for LDAP environments, at the expense of some performance when using nss_db. Given that the DB is a local service, this should have a negligible effect from a user's perspective. With the new algorithm, we simply first iterate through all entries in the local database with gr_next(), recording the IDs that are in use. We then start from the highest presumed-available entry and call getgrgid() to see if it is available. We continue this until we come to the first unused GID. We then select that and return it. If we make it through all the remaining IDs without finding a free one, we start over from the beginning of the range and try to find room in one of the gaps in the range. The patch was originally written by Stephen Gallagher and applied identically also to the user allocation by Tomáš Mráz. Signed-off-by: Serge Hallyn <serge@hallyn.com>
2016-11-15 20:13:54 +05:30
const struct passwd *pwd;
uid_t uid_min, uid_max, preferred_min;
uid_t id;
Make the group and user allocation more effective. Previously, the allocation was optimized for an outdated deployment style (that of /etc/group alongside nss_db). The issue here is that this results in extremely poor performance when using SSSD, Winbind or nss_ldap. There were actually two serious bugs here that have been addressed: 1) Running getgrent() loops won't work in most SSSD or Winbind environments, as full group enumeration is disabled by default. This could easily result in auto-allocating a group that was already in use. (This might result in a security issue as well, if the shared GID is a privileged group). 2) For system groups, the loop was always iterating through the complete SYS_GID_MIN->SYS_GID_MAX range. On SSSD and Winbind, this means hundreds of round-trips to LDAP (unless the GIDs were specifically configured to be ignored by the SSSD or winbindd). To a user with a slow connection to their LDAP server, this would appear as if groupadd -r was hung. (Though it would eventually complete). This patch changes the algorithm to be more favorable for LDAP environments, at the expense of some performance when using nss_db. Given that the DB is a local service, this should have a negligible effect from a user's perspective. With the new algorithm, we simply first iterate through all entries in the local database with gr_next(), recording the IDs that are in use. We then start from the highest presumed-available entry and call getgrgid() to see if it is available. We continue this until we come to the first unused GID. We then select that and return it. If we make it through all the remaining IDs without finding a free one, we start over from the beginning of the range and try to find room in one of the gaps in the range. The patch was originally written by Stephen Gallagher and applied identically also to the user allocation by Tomáš Mráz. Signed-off-by: Serge Hallyn <serge@hallyn.com>
2016-11-15 20:13:54 +05:30
uid_t lowest_found, highest_found;
int result;
int nospam = 0;
assert (uid != NULL);
Make the group and user allocation more effective. Previously, the allocation was optimized for an outdated deployment style (that of /etc/group alongside nss_db). The issue here is that this results in extremely poor performance when using SSSD, Winbind or nss_ldap. There were actually two serious bugs here that have been addressed: 1) Running getgrent() loops won't work in most SSSD or Winbind environments, as full group enumeration is disabled by default. This could easily result in auto-allocating a group that was already in use. (This might result in a security issue as well, if the shared GID is a privileged group). 2) For system groups, the loop was always iterating through the complete SYS_GID_MIN->SYS_GID_MAX range. On SSSD and Winbind, this means hundreds of round-trips to LDAP (unless the GIDs were specifically configured to be ignored by the SSSD or winbindd). To a user with a slow connection to their LDAP server, this would appear as if groupadd -r was hung. (Though it would eventually complete). This patch changes the algorithm to be more favorable for LDAP environments, at the expense of some performance when using nss_db. Given that the DB is a local service, this should have a negligible effect from a user's perspective. With the new algorithm, we simply first iterate through all entries in the local database with gr_next(), recording the IDs that are in use. We then start from the highest presumed-available entry and call getgrgid() to see if it is available. We continue this until we come to the first unused GID. We then select that and return it. If we make it through all the remaining IDs without finding a free one, we start over from the beginning of the range and try to find room in one of the gaps in the range. The patch was originally written by Stephen Gallagher and applied identically also to the user allocation by Tomáš Mráz. Signed-off-by: Serge Hallyn <serge@hallyn.com>
2016-11-15 20:13:54 +05:30
/*
* First, figure out what ID range is appropriate for
* automatic assignment
*/
result = get_ranges (sys_user, &uid_min, &uid_max, &preferred_min);
if (result == EINVAL) {
return -1;
}
/* Check if the preferred UID is available */
if (preferred_uid) {
result = check_uid (*preferred_uid, preferred_min, uid_max, NULL);
if (result == 0) {
/*
* Make sure the UID isn't queued for use already
*/
if (pw_locate_uid (*preferred_uid) == NULL) {
*uid = *preferred_uid;
return 0;
}
/*
* pw_locate_uid() found the UID in an as-yet uncommitted
* entry. We'll proceed below and auto-set an UID.
*/
} else if (result == EEXIST || result == ERANGE) {
/*
* Continue on below. At this time, we won't
* treat these two cases differently.
*/
} else {
/*
* An unexpected error occurred. We should report
* this and fail the user creation.
* This differs from the automatic creation
* behavior below, since if a specific UID was
* requested and generated an error, the user is
* more likely to want to stop and address the
* issue.
*/
fprintf (shadow_logfd,
Make the group and user allocation more effective. Previously, the allocation was optimized for an outdated deployment style (that of /etc/group alongside nss_db). The issue here is that this results in extremely poor performance when using SSSD, Winbind or nss_ldap. There were actually two serious bugs here that have been addressed: 1) Running getgrent() loops won't work in most SSSD or Winbind environments, as full group enumeration is disabled by default. This could easily result in auto-allocating a group that was already in use. (This might result in a security issue as well, if the shared GID is a privileged group). 2) For system groups, the loop was always iterating through the complete SYS_GID_MIN->SYS_GID_MAX range. On SSSD and Winbind, this means hundreds of round-trips to LDAP (unless the GIDs were specifically configured to be ignored by the SSSD or winbindd). To a user with a slow connection to their LDAP server, this would appear as if groupadd -r was hung. (Though it would eventually complete). This patch changes the algorithm to be more favorable for LDAP environments, at the expense of some performance when using nss_db. Given that the DB is a local service, this should have a negligible effect from a user's perspective. With the new algorithm, we simply first iterate through all entries in the local database with gr_next(), recording the IDs that are in use. We then start from the highest presumed-available entry and call getgrgid() to see if it is available. We continue this until we come to the first unused GID. We then select that and return it. If we make it through all the remaining IDs without finding a free one, we start over from the beginning of the range and try to find room in one of the gaps in the range. The patch was originally written by Stephen Gallagher and applied identically also to the user allocation by Tomáš Mráz. Signed-off-by: Serge Hallyn <serge@hallyn.com>
2016-11-15 20:13:54 +05:30
_("%s: Encountered error attempting to use "
"preferred UID: %s\n"),
Prog, strerror (result));
return -1;
}
}
Make the group and user allocation more effective. Previously, the allocation was optimized for an outdated deployment style (that of /etc/group alongside nss_db). The issue here is that this results in extremely poor performance when using SSSD, Winbind or nss_ldap. There were actually two serious bugs here that have been addressed: 1) Running getgrent() loops won't work in most SSSD or Winbind environments, as full group enumeration is disabled by default. This could easily result in auto-allocating a group that was already in use. (This might result in a security issue as well, if the shared GID is a privileged group). 2) For system groups, the loop was always iterating through the complete SYS_GID_MIN->SYS_GID_MAX range. On SSSD and Winbind, this means hundreds of round-trips to LDAP (unless the GIDs were specifically configured to be ignored by the SSSD or winbindd). To a user with a slow connection to their LDAP server, this would appear as if groupadd -r was hung. (Though it would eventually complete). This patch changes the algorithm to be more favorable for LDAP environments, at the expense of some performance when using nss_db. Given that the DB is a local service, this should have a negligible effect from a user's perspective. With the new algorithm, we simply first iterate through all entries in the local database with gr_next(), recording the IDs that are in use. We then start from the highest presumed-available entry and call getgrgid() to see if it is available. We continue this until we come to the first unused GID. We then select that and return it. If we make it through all the remaining IDs without finding a free one, we start over from the beginning of the range and try to find room in one of the gaps in the range. The patch was originally written by Stephen Gallagher and applied identically also to the user allocation by Tomáš Mráz. Signed-off-by: Serge Hallyn <serge@hallyn.com>
2016-11-15 20:13:54 +05:30
/*
* Search the entire passwd file,
* looking for the next unused value.
*
* We first check the local database with pw_rewind/pw_next to find
* all local values that are in use.
*
* We then compare the next free value to all databases (local and
* remote) and iterate until we find a free one. If there are free
* values beyond the lowest (system users) or highest (non-system
* users), we will prefer those and avoid potentially reclaiming a
* deleted user (which can be a security issue, since it may grant
* access to files belonging to that former user).
*
* If there are no UIDs available at the end of the search, we will
* have no choice but to iterate through the range looking for gaps.
*
*/
/* Create an array to hold all of the discovered UIDs */
used_uids = malloc (sizeof (bool) * (uid_max +1));
if (NULL == used_uids) {
fprintf (shadow_logfd,
Make the group and user allocation more effective. Previously, the allocation was optimized for an outdated deployment style (that of /etc/group alongside nss_db). The issue here is that this results in extremely poor performance when using SSSD, Winbind or nss_ldap. There were actually two serious bugs here that have been addressed: 1) Running getgrent() loops won't work in most SSSD or Winbind environments, as full group enumeration is disabled by default. This could easily result in auto-allocating a group that was already in use. (This might result in a security issue as well, if the shared GID is a privileged group). 2) For system groups, the loop was always iterating through the complete SYS_GID_MIN->SYS_GID_MAX range. On SSSD and Winbind, this means hundreds of round-trips to LDAP (unless the GIDs were specifically configured to be ignored by the SSSD or winbindd). To a user with a slow connection to their LDAP server, this would appear as if groupadd -r was hung. (Though it would eventually complete). This patch changes the algorithm to be more favorable for LDAP environments, at the expense of some performance when using nss_db. Given that the DB is a local service, this should have a negligible effect from a user's perspective. With the new algorithm, we simply first iterate through all entries in the local database with gr_next(), recording the IDs that are in use. We then start from the highest presumed-available entry and call getgrgid() to see if it is available. We continue this until we come to the first unused GID. We then select that and return it. If we make it through all the remaining IDs without finding a free one, we start over from the beginning of the range and try to find room in one of the gaps in the range. The patch was originally written by Stephen Gallagher and applied identically also to the user allocation by Tomáš Mráz. Signed-off-by: Serge Hallyn <serge@hallyn.com>
2016-11-15 20:13:54 +05:30
_("%s: failed to allocate memory: %s\n"),
Prog, strerror (errno));
return -1;
}
memset (used_uids, false, sizeof (bool) * (uid_max + 1));
Make the group and user allocation more effective. Previously, the allocation was optimized for an outdated deployment style (that of /etc/group alongside nss_db). The issue here is that this results in extremely poor performance when using SSSD, Winbind or nss_ldap. There were actually two serious bugs here that have been addressed: 1) Running getgrent() loops won't work in most SSSD or Winbind environments, as full group enumeration is disabled by default. This could easily result in auto-allocating a group that was already in use. (This might result in a security issue as well, if the shared GID is a privileged group). 2) For system groups, the loop was always iterating through the complete SYS_GID_MIN->SYS_GID_MAX range. On SSSD and Winbind, this means hundreds of round-trips to LDAP (unless the GIDs were specifically configured to be ignored by the SSSD or winbindd). To a user with a slow connection to their LDAP server, this would appear as if groupadd -r was hung. (Though it would eventually complete). This patch changes the algorithm to be more favorable for LDAP environments, at the expense of some performance when using nss_db. Given that the DB is a local service, this should have a negligible effect from a user's perspective. With the new algorithm, we simply first iterate through all entries in the local database with gr_next(), recording the IDs that are in use. We then start from the highest presumed-available entry and call getgrgid() to see if it is available. We continue this until we come to the first unused GID. We then select that and return it. If we make it through all the remaining IDs without finding a free one, we start over from the beginning of the range and try to find room in one of the gaps in the range. The patch was originally written by Stephen Gallagher and applied identically also to the user allocation by Tomáš Mráz. Signed-off-by: Serge Hallyn <serge@hallyn.com>
2016-11-15 20:13:54 +05:30
/* First look for the lowest and highest value in the local database */
(void) pw_rewind ();
highest_found = uid_min;
lowest_found = uid_max;
while ((pwd = pw_next ()) != NULL) {
/*
* Does this entry have a lower UID than the lowest we've found
* so far?
*/
if ((pwd->pw_uid <= lowest_found) && (pwd->pw_uid >= uid_min)) {
lowest_found = pwd->pw_uid - 1;
}
Make the group and user allocation more effective. Previously, the allocation was optimized for an outdated deployment style (that of /etc/group alongside nss_db). The issue here is that this results in extremely poor performance when using SSSD, Winbind or nss_ldap. There were actually two serious bugs here that have been addressed: 1) Running getgrent() loops won't work in most SSSD or Winbind environments, as full group enumeration is disabled by default. This could easily result in auto-allocating a group that was already in use. (This might result in a security issue as well, if the shared GID is a privileged group). 2) For system groups, the loop was always iterating through the complete SYS_GID_MIN->SYS_GID_MAX range. On SSSD and Winbind, this means hundreds of round-trips to LDAP (unless the GIDs were specifically configured to be ignored by the SSSD or winbindd). To a user with a slow connection to their LDAP server, this would appear as if groupadd -r was hung. (Though it would eventually complete). This patch changes the algorithm to be more favorable for LDAP environments, at the expense of some performance when using nss_db. Given that the DB is a local service, this should have a negligible effect from a user's perspective. With the new algorithm, we simply first iterate through all entries in the local database with gr_next(), recording the IDs that are in use. We then start from the highest presumed-available entry and call getgrgid() to see if it is available. We continue this until we come to the first unused GID. We then select that and return it. If we make it through all the remaining IDs without finding a free one, we start over from the beginning of the range and try to find room in one of the gaps in the range. The patch was originally written by Stephen Gallagher and applied identically also to the user allocation by Tomáš Mráz. Signed-off-by: Serge Hallyn <serge@hallyn.com>
2016-11-15 20:13:54 +05:30
/*
* Does this entry have a higher UID than the highest we've found
* so far?
*/
if ((pwd->pw_uid >= highest_found) && (pwd->pw_uid <= uid_max)) {
highest_found = pwd->pw_uid + 1;
}
/* create index of used UIDs */
if (pwd->pw_uid >= uid_min
&& pwd->pw_uid <= uid_max) {
used_uids[pwd->pw_uid] = true;
}
}
if (sys_user) {
Make the group and user allocation more effective. Previously, the allocation was optimized for an outdated deployment style (that of /etc/group alongside nss_db). The issue here is that this results in extremely poor performance when using SSSD, Winbind or nss_ldap. There were actually two serious bugs here that have been addressed: 1) Running getgrent() loops won't work in most SSSD or Winbind environments, as full group enumeration is disabled by default. This could easily result in auto-allocating a group that was already in use. (This might result in a security issue as well, if the shared GID is a privileged group). 2) For system groups, the loop was always iterating through the complete SYS_GID_MIN->SYS_GID_MAX range. On SSSD and Winbind, this means hundreds of round-trips to LDAP (unless the GIDs were specifically configured to be ignored by the SSSD or winbindd). To a user with a slow connection to their LDAP server, this would appear as if groupadd -r was hung. (Though it would eventually complete). This patch changes the algorithm to be more favorable for LDAP environments, at the expense of some performance when using nss_db. Given that the DB is a local service, this should have a negligible effect from a user's perspective. With the new algorithm, we simply first iterate through all entries in the local database with gr_next(), recording the IDs that are in use. We then start from the highest presumed-available entry and call getgrgid() to see if it is available. We continue this until we come to the first unused GID. We then select that and return it. If we make it through all the remaining IDs without finding a free one, we start over from the beginning of the range and try to find room in one of the gaps in the range. The patch was originally written by Stephen Gallagher and applied identically also to the user allocation by Tomáš Mráz. Signed-off-by: Serge Hallyn <serge@hallyn.com>
2016-11-15 20:13:54 +05:30
/*
* For system users, we want to start from the
* top of the range and work downwards.
*/
Make the group and user allocation more effective. Previously, the allocation was optimized for an outdated deployment style (that of /etc/group alongside nss_db). The issue here is that this results in extremely poor performance when using SSSD, Winbind or nss_ldap. There were actually two serious bugs here that have been addressed: 1) Running getgrent() loops won't work in most SSSD or Winbind environments, as full group enumeration is disabled by default. This could easily result in auto-allocating a group that was already in use. (This might result in a security issue as well, if the shared GID is a privileged group). 2) For system groups, the loop was always iterating through the complete SYS_GID_MIN->SYS_GID_MAX range. On SSSD and Winbind, this means hundreds of round-trips to LDAP (unless the GIDs were specifically configured to be ignored by the SSSD or winbindd). To a user with a slow connection to their LDAP server, this would appear as if groupadd -r was hung. (Though it would eventually complete). This patch changes the algorithm to be more favorable for LDAP environments, at the expense of some performance when using nss_db. Given that the DB is a local service, this should have a negligible effect from a user's perspective. With the new algorithm, we simply first iterate through all entries in the local database with gr_next(), recording the IDs that are in use. We then start from the highest presumed-available entry and call getgrgid() to see if it is available. We continue this until we come to the first unused GID. We then select that and return it. If we make it through all the remaining IDs without finding a free one, we start over from the beginning of the range and try to find room in one of the gaps in the range. The patch was originally written by Stephen Gallagher and applied identically also to the user allocation by Tomáš Mráz. Signed-off-by: Serge Hallyn <serge@hallyn.com>
2016-11-15 20:13:54 +05:30
/*
* At the conclusion of the pw_next() search, we will either
* have a presumed-free UID or we will be at UID_MIN - 1.
*/
if (lowest_found < uid_min) {
/*
* In this case, an UID is in use at UID_MIN.
*
* We will reset the search to UID_MAX and proceed down
* through all the UIDs (skipping those we detected with
* used_uids) for a free one. It is a known issue that
* this may result in reusing a previously-deleted UID,
* so administrators should be instructed to use this
* auto-detection with care (and prefer to assign UIDs
* explicitly).
*/
lowest_found = uid_max;
}
Make the group and user allocation more effective. Previously, the allocation was optimized for an outdated deployment style (that of /etc/group alongside nss_db). The issue here is that this results in extremely poor performance when using SSSD, Winbind or nss_ldap. There were actually two serious bugs here that have been addressed: 1) Running getgrent() loops won't work in most SSSD or Winbind environments, as full group enumeration is disabled by default. This could easily result in auto-allocating a group that was already in use. (This might result in a security issue as well, if the shared GID is a privileged group). 2) For system groups, the loop was always iterating through the complete SYS_GID_MIN->SYS_GID_MAX range. On SSSD and Winbind, this means hundreds of round-trips to LDAP (unless the GIDs were specifically configured to be ignored by the SSSD or winbindd). To a user with a slow connection to their LDAP server, this would appear as if groupadd -r was hung. (Though it would eventually complete). This patch changes the algorithm to be more favorable for LDAP environments, at the expense of some performance when using nss_db. Given that the DB is a local service, this should have a negligible effect from a user's perspective. With the new algorithm, we simply first iterate through all entries in the local database with gr_next(), recording the IDs that are in use. We then start from the highest presumed-available entry and call getgrgid() to see if it is available. We continue this until we come to the first unused GID. We then select that and return it. If we make it through all the remaining IDs without finding a free one, we start over from the beginning of the range and try to find room in one of the gaps in the range. The patch was originally written by Stephen Gallagher and applied identically also to the user allocation by Tomáš Mráz. Signed-off-by: Serge Hallyn <serge@hallyn.com>
2016-11-15 20:13:54 +05:30
/* Search through all of the IDs in the range */
for (id = lowest_found; id >= uid_min; id--) {
result = check_uid (id, uid_min, uid_max, used_uids);
if (result == 0) {
/* This UID is available. Return it. */
*uid = id;
free (used_uids);
return 0;
} else if (result == EEXIST) {
/* This UID is in use, we'll continue to the next */
} else {
/*
* An unexpected error occurred.
*
* Only report it the first time to avoid spamming
* the logs
*
*/
if (!nospam) {
fprintf (shadow_logfd,
Make the group and user allocation more effective. Previously, the allocation was optimized for an outdated deployment style (that of /etc/group alongside nss_db). The issue here is that this results in extremely poor performance when using SSSD, Winbind or nss_ldap. There were actually two serious bugs here that have been addressed: 1) Running getgrent() loops won't work in most SSSD or Winbind environments, as full group enumeration is disabled by default. This could easily result in auto-allocating a group that was already in use. (This might result in a security issue as well, if the shared GID is a privileged group). 2) For system groups, the loop was always iterating through the complete SYS_GID_MIN->SYS_GID_MAX range. On SSSD and Winbind, this means hundreds of round-trips to LDAP (unless the GIDs were specifically configured to be ignored by the SSSD or winbindd). To a user with a slow connection to their LDAP server, this would appear as if groupadd -r was hung. (Though it would eventually complete). This patch changes the algorithm to be more favorable for LDAP environments, at the expense of some performance when using nss_db. Given that the DB is a local service, this should have a negligible effect from a user's perspective. With the new algorithm, we simply first iterate through all entries in the local database with gr_next(), recording the IDs that are in use. We then start from the highest presumed-available entry and call getgrgid() to see if it is available. We continue this until we come to the first unused GID. We then select that and return it. If we make it through all the remaining IDs without finding a free one, we start over from the beginning of the range and try to find room in one of the gaps in the range. The patch was originally written by Stephen Gallagher and applied identically also to the user allocation by Tomáš Mráz. Signed-off-by: Serge Hallyn <serge@hallyn.com>
2016-11-15 20:13:54 +05:30
_("%s: Can't get unique system UID (%s). "
"Suppressing additional messages.\n"),
Prog, strerror (result));
SYSLOG ((LOG_ERR,
"Error checking available UIDs: %s",
strerror (result)));
nospam = 1;
}
/*
* We will continue anyway. Hopefully a later UID
* will work properly.
*/
}
}
Make the group and user allocation more effective. Previously, the allocation was optimized for an outdated deployment style (that of /etc/group alongside nss_db). The issue here is that this results in extremely poor performance when using SSSD, Winbind or nss_ldap. There were actually two serious bugs here that have been addressed: 1) Running getgrent() loops won't work in most SSSD or Winbind environments, as full group enumeration is disabled by default. This could easily result in auto-allocating a group that was already in use. (This might result in a security issue as well, if the shared GID is a privileged group). 2) For system groups, the loop was always iterating through the complete SYS_GID_MIN->SYS_GID_MAX range. On SSSD and Winbind, this means hundreds of round-trips to LDAP (unless the GIDs were specifically configured to be ignored by the SSSD or winbindd). To a user with a slow connection to their LDAP server, this would appear as if groupadd -r was hung. (Though it would eventually complete). This patch changes the algorithm to be more favorable for LDAP environments, at the expense of some performance when using nss_db. Given that the DB is a local service, this should have a negligible effect from a user's perspective. With the new algorithm, we simply first iterate through all entries in the local database with gr_next(), recording the IDs that are in use. We then start from the highest presumed-available entry and call getgrgid() to see if it is available. We continue this until we come to the first unused GID. We then select that and return it. If we make it through all the remaining IDs without finding a free one, we start over from the beginning of the range and try to find room in one of the gaps in the range. The patch was originally written by Stephen Gallagher and applied identically also to the user allocation by Tomáš Mráz. Signed-off-by: Serge Hallyn <serge@hallyn.com>
2016-11-15 20:13:54 +05:30
/*
* If we get all the way through the loop, try again from UID_MAX,
* unless that was where we previously started. (NOTE: the worst-case
* scenario here is that we will run through (UID_MAX - UID_MIN - 1)
* cycles *again* if we fall into this case with lowest_found as
* UID_MAX - 1, all users in the range in use and maintained by
* network services such as LDAP.)
*/
if (lowest_found != uid_max) {
for (id = uid_max; id >= uid_min; id--) {
result = check_uid (id, uid_min, uid_max, used_uids);
if (result == 0) {
/* This UID is available. Return it. */
*uid = id;
free (used_uids);
return 0;
} else if (result == EEXIST) {
/* This UID is in use, we'll continue to the next */
} else {
/*
* An unexpected error occurred.
*
* Only report it the first time to avoid spamming
* the logs
*
*/
if (!nospam) {
fprintf (shadow_logfd,
Make the group and user allocation more effective. Previously, the allocation was optimized for an outdated deployment style (that of /etc/group alongside nss_db). The issue here is that this results in extremely poor performance when using SSSD, Winbind or nss_ldap. There were actually two serious bugs here that have been addressed: 1) Running getgrent() loops won't work in most SSSD or Winbind environments, as full group enumeration is disabled by default. This could easily result in auto-allocating a group that was already in use. (This might result in a security issue as well, if the shared GID is a privileged group). 2) For system groups, the loop was always iterating through the complete SYS_GID_MIN->SYS_GID_MAX range. On SSSD and Winbind, this means hundreds of round-trips to LDAP (unless the GIDs were specifically configured to be ignored by the SSSD or winbindd). To a user with a slow connection to their LDAP server, this would appear as if groupadd -r was hung. (Though it would eventually complete). This patch changes the algorithm to be more favorable for LDAP environments, at the expense of some performance when using nss_db. Given that the DB is a local service, this should have a negligible effect from a user's perspective. With the new algorithm, we simply first iterate through all entries in the local database with gr_next(), recording the IDs that are in use. We then start from the highest presumed-available entry and call getgrgid() to see if it is available. We continue this until we come to the first unused GID. We then select that and return it. If we make it through all the remaining IDs without finding a free one, we start over from the beginning of the range and try to find room in one of the gaps in the range. The patch was originally written by Stephen Gallagher and applied identically also to the user allocation by Tomáš Mráz. Signed-off-by: Serge Hallyn <serge@hallyn.com>
2016-11-15 20:13:54 +05:30
_("%s: Can't get unique system UID (%s). "
"Suppressing additional messages.\n"),
Prog, strerror (result));
SYSLOG((LOG_ERR,
"Error checking available UIDs: %s",
strerror (result)));
nospam = 1;
}
/*
* We will continue anyway. Hopefully a later UID
* will work properly.
*/
}
}
}
Make the group and user allocation more effective. Previously, the allocation was optimized for an outdated deployment style (that of /etc/group alongside nss_db). The issue here is that this results in extremely poor performance when using SSSD, Winbind or nss_ldap. There were actually two serious bugs here that have been addressed: 1) Running getgrent() loops won't work in most SSSD or Winbind environments, as full group enumeration is disabled by default. This could easily result in auto-allocating a group that was already in use. (This might result in a security issue as well, if the shared GID is a privileged group). 2) For system groups, the loop was always iterating through the complete SYS_GID_MIN->SYS_GID_MAX range. On SSSD and Winbind, this means hundreds of round-trips to LDAP (unless the GIDs were specifically configured to be ignored by the SSSD or winbindd). To a user with a slow connection to their LDAP server, this would appear as if groupadd -r was hung. (Though it would eventually complete). This patch changes the algorithm to be more favorable for LDAP environments, at the expense of some performance when using nss_db. Given that the DB is a local service, this should have a negligible effect from a user's perspective. With the new algorithm, we simply first iterate through all entries in the local database with gr_next(), recording the IDs that are in use. We then start from the highest presumed-available entry and call getgrgid() to see if it is available. We continue this until we come to the first unused GID. We then select that and return it. If we make it through all the remaining IDs without finding a free one, we start over from the beginning of the range and try to find room in one of the gaps in the range. The patch was originally written by Stephen Gallagher and applied identically also to the user allocation by Tomáš Mráz. Signed-off-by: Serge Hallyn <serge@hallyn.com>
2016-11-15 20:13:54 +05:30
} else { /* !sys_user */
/*
* For non-system users, we want to start from the
* bottom of the range and work upwards.
*/
Make the group and user allocation more effective. Previously, the allocation was optimized for an outdated deployment style (that of /etc/group alongside nss_db). The issue here is that this results in extremely poor performance when using SSSD, Winbind or nss_ldap. There were actually two serious bugs here that have been addressed: 1) Running getgrent() loops won't work in most SSSD or Winbind environments, as full group enumeration is disabled by default. This could easily result in auto-allocating a group that was already in use. (This might result in a security issue as well, if the shared GID is a privileged group). 2) For system groups, the loop was always iterating through the complete SYS_GID_MIN->SYS_GID_MAX range. On SSSD and Winbind, this means hundreds of round-trips to LDAP (unless the GIDs were specifically configured to be ignored by the SSSD or winbindd). To a user with a slow connection to their LDAP server, this would appear as if groupadd -r was hung. (Though it would eventually complete). This patch changes the algorithm to be more favorable for LDAP environments, at the expense of some performance when using nss_db. Given that the DB is a local service, this should have a negligible effect from a user's perspective. With the new algorithm, we simply first iterate through all entries in the local database with gr_next(), recording the IDs that are in use. We then start from the highest presumed-available entry and call getgrgid() to see if it is available. We continue this until we come to the first unused GID. We then select that and return it. If we make it through all the remaining IDs without finding a free one, we start over from the beginning of the range and try to find room in one of the gaps in the range. The patch was originally written by Stephen Gallagher and applied identically also to the user allocation by Tomáš Mráz. Signed-off-by: Serge Hallyn <serge@hallyn.com>
2016-11-15 20:13:54 +05:30
/*
* At the conclusion of the pw_next() search, we will either
* have a presumed-free UID or we will be at UID_MAX + 1.
*/
if (highest_found > uid_max) {
/*
* In this case, a UID is in use at UID_MAX.
*
* We will reset the search to UID_MIN and proceed up
* through all the UIDs (skipping those we detected with
* used_uids) for a free one. It is a known issue that
* this may result in reusing a previously-deleted UID,
* so administrators should be instructed to use this
* auto-detection with care (and prefer to assign UIDs
* explicitly).
*/
highest_found = uid_min;
}
Make the group and user allocation more effective. Previously, the allocation was optimized for an outdated deployment style (that of /etc/group alongside nss_db). The issue here is that this results in extremely poor performance when using SSSD, Winbind or nss_ldap. There were actually two serious bugs here that have been addressed: 1) Running getgrent() loops won't work in most SSSD or Winbind environments, as full group enumeration is disabled by default. This could easily result in auto-allocating a group that was already in use. (This might result in a security issue as well, if the shared GID is a privileged group). 2) For system groups, the loop was always iterating through the complete SYS_GID_MIN->SYS_GID_MAX range. On SSSD and Winbind, this means hundreds of round-trips to LDAP (unless the GIDs were specifically configured to be ignored by the SSSD or winbindd). To a user with a slow connection to their LDAP server, this would appear as if groupadd -r was hung. (Though it would eventually complete). This patch changes the algorithm to be more favorable for LDAP environments, at the expense of some performance when using nss_db. Given that the DB is a local service, this should have a negligible effect from a user's perspective. With the new algorithm, we simply first iterate through all entries in the local database with gr_next(), recording the IDs that are in use. We then start from the highest presumed-available entry and call getgrgid() to see if it is available. We continue this until we come to the first unused GID. We then select that and return it. If we make it through all the remaining IDs without finding a free one, we start over from the beginning of the range and try to find room in one of the gaps in the range. The patch was originally written by Stephen Gallagher and applied identically also to the user allocation by Tomáš Mráz. Signed-off-by: Serge Hallyn <serge@hallyn.com>
2016-11-15 20:13:54 +05:30
/* Search through all of the IDs in the range */
for (id = highest_found; id <= uid_max; id++) {
result = check_uid (id, uid_min, uid_max, used_uids);
if (result == 0) {
/* This UID is available. Return it. */
*uid = id;
free (used_uids);
Make the group and user allocation more effective. Previously, the allocation was optimized for an outdated deployment style (that of /etc/group alongside nss_db). The issue here is that this results in extremely poor performance when using SSSD, Winbind or nss_ldap. There were actually two serious bugs here that have been addressed: 1) Running getgrent() loops won't work in most SSSD or Winbind environments, as full group enumeration is disabled by default. This could easily result in auto-allocating a group that was already in use. (This might result in a security issue as well, if the shared GID is a privileged group). 2) For system groups, the loop was always iterating through the complete SYS_GID_MIN->SYS_GID_MAX range. On SSSD and Winbind, this means hundreds of round-trips to LDAP (unless the GIDs were specifically configured to be ignored by the SSSD or winbindd). To a user with a slow connection to their LDAP server, this would appear as if groupadd -r was hung. (Though it would eventually complete). This patch changes the algorithm to be more favorable for LDAP environments, at the expense of some performance when using nss_db. Given that the DB is a local service, this should have a negligible effect from a user's perspective. With the new algorithm, we simply first iterate through all entries in the local database with gr_next(), recording the IDs that are in use. We then start from the highest presumed-available entry and call getgrgid() to see if it is available. We continue this until we come to the first unused GID. We then select that and return it. If we make it through all the remaining IDs without finding a free one, we start over from the beginning of the range and try to find room in one of the gaps in the range. The patch was originally written by Stephen Gallagher and applied identically also to the user allocation by Tomáš Mráz. Signed-off-by: Serge Hallyn <serge@hallyn.com>
2016-11-15 20:13:54 +05:30
return 0;
} else if (result == EEXIST) {
/* This UID is in use, we'll continue to the next */
} else {
/*
* An unexpected error occurred.
*
* Only report it the first time to avoid spamming
* the logs
*
*/
if (!nospam) {
fprintf (shadow_logfd,
Make the group and user allocation more effective. Previously, the allocation was optimized for an outdated deployment style (that of /etc/group alongside nss_db). The issue here is that this results in extremely poor performance when using SSSD, Winbind or nss_ldap. There were actually two serious bugs here that have been addressed: 1) Running getgrent() loops won't work in most SSSD or Winbind environments, as full group enumeration is disabled by default. This could easily result in auto-allocating a group that was already in use. (This might result in a security issue as well, if the shared GID is a privileged group). 2) For system groups, the loop was always iterating through the complete SYS_GID_MIN->SYS_GID_MAX range. On SSSD and Winbind, this means hundreds of round-trips to LDAP (unless the GIDs were specifically configured to be ignored by the SSSD or winbindd). To a user with a slow connection to their LDAP server, this would appear as if groupadd -r was hung. (Though it would eventually complete). This patch changes the algorithm to be more favorable for LDAP environments, at the expense of some performance when using nss_db. Given that the DB is a local service, this should have a negligible effect from a user's perspective. With the new algorithm, we simply first iterate through all entries in the local database with gr_next(), recording the IDs that are in use. We then start from the highest presumed-available entry and call getgrgid() to see if it is available. We continue this until we come to the first unused GID. We then select that and return it. If we make it through all the remaining IDs without finding a free one, we start over from the beginning of the range and try to find room in one of the gaps in the range. The patch was originally written by Stephen Gallagher and applied identically also to the user allocation by Tomáš Mráz. Signed-off-by: Serge Hallyn <serge@hallyn.com>
2016-11-15 20:13:54 +05:30
_("%s: Can't get unique UID (%s). "
"Suppressing additional messages.\n"),
Prog, strerror (result));
SYSLOG ((LOG_ERR,
"Error checking available UIDs: %s",
strerror (result)));
nospam = 1;
}
/*
* We will continue anyway. Hopefully a later UID
* will work properly.
*/
}
}
Make the group and user allocation more effective. Previously, the allocation was optimized for an outdated deployment style (that of /etc/group alongside nss_db). The issue here is that this results in extremely poor performance when using SSSD, Winbind or nss_ldap. There were actually two serious bugs here that have been addressed: 1) Running getgrent() loops won't work in most SSSD or Winbind environments, as full group enumeration is disabled by default. This could easily result in auto-allocating a group that was already in use. (This might result in a security issue as well, if the shared GID is a privileged group). 2) For system groups, the loop was always iterating through the complete SYS_GID_MIN->SYS_GID_MAX range. On SSSD and Winbind, this means hundreds of round-trips to LDAP (unless the GIDs were specifically configured to be ignored by the SSSD or winbindd). To a user with a slow connection to their LDAP server, this would appear as if groupadd -r was hung. (Though it would eventually complete). This patch changes the algorithm to be more favorable for LDAP environments, at the expense of some performance when using nss_db. Given that the DB is a local service, this should have a negligible effect from a user's perspective. With the new algorithm, we simply first iterate through all entries in the local database with gr_next(), recording the IDs that are in use. We then start from the highest presumed-available entry and call getgrgid() to see if it is available. We continue this until we come to the first unused GID. We then select that and return it. If we make it through all the remaining IDs without finding a free one, we start over from the beginning of the range and try to find room in one of the gaps in the range. The patch was originally written by Stephen Gallagher and applied identically also to the user allocation by Tomáš Mráz. Signed-off-by: Serge Hallyn <serge@hallyn.com>
2016-11-15 20:13:54 +05:30
/*
* If we get all the way through the loop, try again from UID_MIN,
* unless that was where we previously started. (NOTE: the worst-case
* scenario here is that we will run through (UID_MAX - UID_MIN - 1)
* cycles *again* if we fall into this case with highest_found as
* UID_MIN + 1, all users in the range in use and maintained by
* network services such as LDAP.)
*/
if (highest_found != uid_min) {
for (id = uid_min; id <= uid_max; id++) {
result = check_uid (id, uid_min, uid_max, used_uids);
if (result == 0) {
/* This UID is available. Return it. */
*uid = id;
free (used_uids);
return 0;
} else if (result == EEXIST) {
/* This UID is in use, we'll continue to the next */
} else {
/*
* An unexpected error occurred.
*
* Only report it the first time to avoid spamming
* the logs
*
*/
if (!nospam) {
fprintf (shadow_logfd,
Make the group and user allocation more effective. Previously, the allocation was optimized for an outdated deployment style (that of /etc/group alongside nss_db). The issue here is that this results in extremely poor performance when using SSSD, Winbind or nss_ldap. There were actually two serious bugs here that have been addressed: 1) Running getgrent() loops won't work in most SSSD or Winbind environments, as full group enumeration is disabled by default. This could easily result in auto-allocating a group that was already in use. (This might result in a security issue as well, if the shared GID is a privileged group). 2) For system groups, the loop was always iterating through the complete SYS_GID_MIN->SYS_GID_MAX range. On SSSD and Winbind, this means hundreds of round-trips to LDAP (unless the GIDs were specifically configured to be ignored by the SSSD or winbindd). To a user with a slow connection to their LDAP server, this would appear as if groupadd -r was hung. (Though it would eventually complete). This patch changes the algorithm to be more favorable for LDAP environments, at the expense of some performance when using nss_db. Given that the DB is a local service, this should have a negligible effect from a user's perspective. With the new algorithm, we simply first iterate through all entries in the local database with gr_next(), recording the IDs that are in use. We then start from the highest presumed-available entry and call getgrgid() to see if it is available. We continue this until we come to the first unused GID. We then select that and return it. If we make it through all the remaining IDs without finding a free one, we start over from the beginning of the range and try to find room in one of the gaps in the range. The patch was originally written by Stephen Gallagher and applied identically also to the user allocation by Tomáš Mráz. Signed-off-by: Serge Hallyn <serge@hallyn.com>
2016-11-15 20:13:54 +05:30
_("%s: Can't get unique UID (%s). "
"Suppressing additional messages.\n"),
Prog, strerror (result));
SYSLOG ((LOG_ERR,
"Error checking available UIDs: %s",
strerror (result)));
nospam = 1;
}
/*
* We will continue anyway. Hopefully a later UID
* will work properly.
*/
}
}
}
}
Make the group and user allocation more effective. Previously, the allocation was optimized for an outdated deployment style (that of /etc/group alongside nss_db). The issue here is that this results in extremely poor performance when using SSSD, Winbind or nss_ldap. There were actually two serious bugs here that have been addressed: 1) Running getgrent() loops won't work in most SSSD or Winbind environments, as full group enumeration is disabled by default. This could easily result in auto-allocating a group that was already in use. (This might result in a security issue as well, if the shared GID is a privileged group). 2) For system groups, the loop was always iterating through the complete SYS_GID_MIN->SYS_GID_MAX range. On SSSD and Winbind, this means hundreds of round-trips to LDAP (unless the GIDs were specifically configured to be ignored by the SSSD or winbindd). To a user with a slow connection to their LDAP server, this would appear as if groupadd -r was hung. (Though it would eventually complete). This patch changes the algorithm to be more favorable for LDAP environments, at the expense of some performance when using nss_db. Given that the DB is a local service, this should have a negligible effect from a user's perspective. With the new algorithm, we simply first iterate through all entries in the local database with gr_next(), recording the IDs that are in use. We then start from the highest presumed-available entry and call getgrgid() to see if it is available. We continue this until we come to the first unused GID. We then select that and return it. If we make it through all the remaining IDs without finding a free one, we start over from the beginning of the range and try to find room in one of the gaps in the range. The patch was originally written by Stephen Gallagher and applied identically also to the user allocation by Tomáš Mráz. Signed-off-by: Serge Hallyn <serge@hallyn.com>
2016-11-15 20:13:54 +05:30
/* The code reached here and found no available IDs in the range */
fprintf (shadow_logfd,
Make the group and user allocation more effective. Previously, the allocation was optimized for an outdated deployment style (that of /etc/group alongside nss_db). The issue here is that this results in extremely poor performance when using SSSD, Winbind or nss_ldap. There were actually two serious bugs here that have been addressed: 1) Running getgrent() loops won't work in most SSSD or Winbind environments, as full group enumeration is disabled by default. This could easily result in auto-allocating a group that was already in use. (This might result in a security issue as well, if the shared GID is a privileged group). 2) For system groups, the loop was always iterating through the complete SYS_GID_MIN->SYS_GID_MAX range. On SSSD and Winbind, this means hundreds of round-trips to LDAP (unless the GIDs were specifically configured to be ignored by the SSSD or winbindd). To a user with a slow connection to their LDAP server, this would appear as if groupadd -r was hung. (Though it would eventually complete). This patch changes the algorithm to be more favorable for LDAP environments, at the expense of some performance when using nss_db. Given that the DB is a local service, this should have a negligible effect from a user's perspective. With the new algorithm, we simply first iterate through all entries in the local database with gr_next(), recording the IDs that are in use. We then start from the highest presumed-available entry and call getgrgid() to see if it is available. We continue this until we come to the first unused GID. We then select that and return it. If we make it through all the remaining IDs without finding a free one, we start over from the beginning of the range and try to find room in one of the gaps in the range. The patch was originally written by Stephen Gallagher and applied identically also to the user allocation by Tomáš Mráz. Signed-off-by: Serge Hallyn <serge@hallyn.com>
2016-11-15 20:13:54 +05:30
_("%s: Can't get unique UID (no more available UIDs)\n"),
Prog);
SYSLOG ((LOG_WARN, "no more available UIDs on the system"));
free (used_uids);
Make the group and user allocation more effective. Previously, the allocation was optimized for an outdated deployment style (that of /etc/group alongside nss_db). The issue here is that this results in extremely poor performance when using SSSD, Winbind or nss_ldap. There were actually two serious bugs here that have been addressed: 1) Running getgrent() loops won't work in most SSSD or Winbind environments, as full group enumeration is disabled by default. This could easily result in auto-allocating a group that was already in use. (This might result in a security issue as well, if the shared GID is a privileged group). 2) For system groups, the loop was always iterating through the complete SYS_GID_MIN->SYS_GID_MAX range. On SSSD and Winbind, this means hundreds of round-trips to LDAP (unless the GIDs were specifically configured to be ignored by the SSSD or winbindd). To a user with a slow connection to their LDAP server, this would appear as if groupadd -r was hung. (Though it would eventually complete). This patch changes the algorithm to be more favorable for LDAP environments, at the expense of some performance when using nss_db. Given that the DB is a local service, this should have a negligible effect from a user's perspective. With the new algorithm, we simply first iterate through all entries in the local database with gr_next(), recording the IDs that are in use. We then start from the highest presumed-available entry and call getgrgid() to see if it is available. We continue this until we come to the first unused GID. We then select that and return it. If we make it through all the remaining IDs without finding a free one, we start over from the beginning of the range and try to find room in one of the gaps in the range. The patch was originally written by Stephen Gallagher and applied identically also to the user allocation by Tomáš Mráz. Signed-off-by: Serge Hallyn <serge@hallyn.com>
2016-11-15 20:13:54 +05:30
return -1;
}