fs/pidfs.c | 70 +++++++++++++++- include/uapi/linux/pidfd.h | 24 ++++++ .../testing/selftests/pidfd/pidfd_open_test.c | 81 ++++++++++++++++++- 3 files changed, 171 insertions(+), 4 deletions(-)
From: Luca Boccassi <luca.boccassi@gmail.com>
A common pattern when using pid fds is having to get information
about the process, which currently requires /proc being mounted,
resolving the fd to a pid, and then do manual string parsing of
/proc/N/status and friends. This needs to be reimplemented over
and over in all userspace projects (e.g.: I have reimplemented
resolving in systemd, dbus, dbus-daemon, polkit so far), and
requires additional care in checking that the fd is still valid
after having parsed the data, to avoid races.
Having a programmatic API that can be used directly removes all
these requirements, including having /proc mounted.
As discussed at LPC24, add an ioctl with an extensible struct
so that more parameters can be added later if needed. Start with
returning pid/tgid/ppid and creds unconditionally, and cgroupid
optionally.
Signed-off-by: Luca Boccassi <luca.boccassi@gmail.com>
---
v4: fix arg check in pidfd_ioctl() by moving it after the new call
v3: switch from pid_vnr() to task_pid_vnr()
v2: Apply comments from Christian, apart from the one about pid namespaces
as I need additional hints on how to implement it.
Drop the security_context string as it is not the appropriate
metadata to give userspace these days.
fs/pidfs.c | 70 +++++++++++++++-
include/uapi/linux/pidfd.h | 24 ++++++
.../testing/selftests/pidfd/pidfd_open_test.c | 81 ++++++++++++++++++-
3 files changed, 171 insertions(+), 4 deletions(-)
diff --git a/fs/pidfs.c b/fs/pidfs.c
index 7ffdc88dfb52..9ef63c1053d4 100644
--- a/fs/pidfs.c
+++ b/fs/pidfs.c
@@ -2,6 +2,7 @@
#include <linux/anon_inodes.h>
#include <linux/file.h>
#include <linux/fs.h>
+#include <linux/cgroup.h>
#include <linux/magic.h>
#include <linux/mount.h>
#include <linux/pid.h>
@@ -114,6 +115,65 @@ static __poll_t pidfd_poll(struct file *file, struct poll_table_struct *pts)
return poll_flags;
}
+static long pidfd_info(struct task_struct *task, unsigned int cmd, unsigned long arg)
+{
+ struct pidfd_info __user *uinfo = (struct pidfd_info __user *)arg;
+ size_t usize = _IOC_SIZE(cmd);
+ struct pidfd_info kinfo = {};
+ struct user_namespace *user_ns;
+ const struct cred *c;
+ __u64 request_mask;
+
+ if (!uinfo)
+ return -EINVAL;
+ if (usize < sizeof(struct pidfd_info))
+ return -EINVAL; /* First version, no smaller struct possible */
+
+ if (copy_from_user(&request_mask, &uinfo->request_mask, sizeof(request_mask)))
+ return -EFAULT;
+
+ c = get_task_cred(task);
+ if (!c)
+ return -ESRCH;
+
+ /* Unconditionally return identifiers and credentials, the rest only on request */
+
+ kinfo.pid = task_pid_vnr(task);
+ kinfo.tgid = task_tgid_vnr(task);
+ kinfo.ppid = task_ppid_nr_ns(task, task_active_pid_ns(task));
+
+ user_ns = current_user_ns();
+ kinfo.ruid = from_kuid_munged(user_ns, c->uid);
+ kinfo.rgid = from_kgid_munged(user_ns, c->gid);
+ kinfo.euid = from_kuid_munged(user_ns, c->euid);
+ kinfo.egid = from_kgid_munged(user_ns, c->egid);
+ kinfo.suid = from_kuid_munged(user_ns, c->suid);
+ kinfo.sgid = from_kgid_munged(user_ns, c->sgid);
+ kinfo.fsuid = from_kuid_munged(user_ns, c->fsuid);
+ kinfo.fsgid = from_kgid_munged(user_ns, c->fsgid);
+
+ if (request_mask & PIDFD_INFO_CGROUPID) {
+ struct cgroup *cgrp = task_css_check(task, pids_cgrp_id, 1)->cgroup;
+ if (!cgrp)
+ return -ENODEV;
+
+ kinfo.cgroupid = cgroup_id(cgrp);
+ kinfo.result_mask |= PIDFD_INFO_CGROUPID;
+ }
+
+ /*
+ * If userspace and the kernel have the same struct size it can just
+ * be copied. If userspace provides an older struct, only the bits that
+ * userspace knows about will be copied. If userspace provides a new
+ * struct, only the bits that the kernel knows about will be copied and
+ * the size value will be set to the size the kernel knows about.
+ */
+ if (copy_to_user(uinfo, &kinfo, min(usize, sizeof(kinfo))))
+ return -EFAULT;
+
+ return 0;
+}
+
static long pidfd_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
{
struct task_struct *task __free(put_task) = NULL;
@@ -121,13 +181,17 @@ static long pidfd_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
struct pid *pid = pidfd_pid(file);
struct ns_common *ns_common = NULL;
- if (arg)
- return -EINVAL;
-
task = get_pid_task(pid, PIDTYPE_PID);
if (!task)
return -ESRCH;
+ /* Extensible IOCTL that does not open namespace FDs, take a shortcut */
+ if (_IOC_NR(cmd) == _IOC_NR(PIDFD_GET_INFO))
+ return pidfd_info(task, cmd, arg);
+
+ if (arg)
+ return -EINVAL;
+
scoped_guard(task_lock, task) {
nsp = task->nsproxy;
if (nsp)
diff --git a/include/uapi/linux/pidfd.h b/include/uapi/linux/pidfd.h
index 565fc0629fff..f278db1a3fe8 100644
--- a/include/uapi/linux/pidfd.h
+++ b/include/uapi/linux/pidfd.h
@@ -16,6 +16,29 @@
#define PIDFD_SIGNAL_THREAD_GROUP (1UL << 1)
#define PIDFD_SIGNAL_PROCESS_GROUP (1UL << 2)
+/* Flags for pidfd_info. */
+#define PIDFD_INFO_CGROUPID (1UL << 0)
+
+struct pidfd_info {
+ /* Let userspace request expensive stuff explictly. */
+ __u64 request_mask;
+ /* And let the kernel indicate whether it knows about it. */
+ __u64 result_mask;
+ __u64 cgroupid;
+ __u32 pid;
+ __u32 tgid;
+ __u32 ppid;
+ __u32 ruid;
+ __u32 rgid;
+ __u32 euid;
+ __u32 egid;
+ __u32 suid;
+ __u32 sgid;
+ __u32 fsuid;
+ __u32 fsgid;
+ __u32 spare0[1];
+};
+
#define PIDFS_IOCTL_MAGIC 0xFF
#define PIDFD_GET_CGROUP_NAMESPACE _IO(PIDFS_IOCTL_MAGIC, 1)
@@ -28,5 +51,6 @@
#define PIDFD_GET_TIME_FOR_CHILDREN_NAMESPACE _IO(PIDFS_IOCTL_MAGIC, 8)
#define PIDFD_GET_USER_NAMESPACE _IO(PIDFS_IOCTL_MAGIC, 9)
#define PIDFD_GET_UTS_NAMESPACE _IO(PIDFS_IOCTL_MAGIC, 10)
+#define PIDFD_GET_INFO _IOWR(PIDFS_IOCTL_MAGIC, 11, struct pidfd_info)
#endif /* _UAPI_LINUX_PIDFD_H */
diff --git a/tools/testing/selftests/pidfd/pidfd_open_test.c b/tools/testing/selftests/pidfd/pidfd_open_test.c
index c62564c264b1..30c50a8ae10b 100644
--- a/tools/testing/selftests/pidfd/pidfd_open_test.c
+++ b/tools/testing/selftests/pidfd/pidfd_open_test.c
@@ -13,6 +13,7 @@
#include <stdlib.h>
#include <string.h>
#include <syscall.h>
+#include <sys/ioctl.h>
#include <sys/mount.h>
#include <sys/prctl.h>
#include <sys/wait.h>
@@ -21,6 +22,35 @@
#include "pidfd.h"
#include "../kselftest.h"
+#ifndef PIDFS_IOCTL_MAGIC
+#define PIDFS_IOCTL_MAGIC 0xFF
+#endif
+
+#ifndef PIDFD_GET_INFO
+#define PIDFD_GET_INFO _IOWR(PIDFS_IOCTL_MAGIC, 11, struct pidfd_info)
+#define PIDFD_INFO_CGROUPID (1UL << 0)
+
+struct pidfd_info {
+ /* Let userspace request expensive stuff explictly. */
+ __u64 request_mask;
+ /* And let the kernel indicate whether it knows about it. */
+ __u64 result_mask;
+ __u64 cgroupid;
+ __u32 pid;
+ __u32 tgid;
+ __u32 ppid;
+ __u32 ruid;
+ __u32 rgid;
+ __u32 euid;
+ __u32 egid;
+ __u32 suid;
+ __u32 sgid;
+ __u32 fsuid;
+ __u32 fsgid;
+ __u32 spare0[1];
+};
+#endif
+
static int safe_int(const char *numstr, int *converted)
{
char *err = NULL;
@@ -120,10 +150,13 @@ static pid_t get_pid_from_fdinfo_file(int pidfd, const char *key, size_t keylen)
int main(int argc, char **argv)
{
+ struct pidfd_info info = {
+ .request_mask = PIDFD_INFO_CGROUPID,
+ };
int pidfd = -1, ret = 1;
pid_t pid;
- ksft_set_plan(3);
+ ksft_set_plan(4);
pidfd = sys_pidfd_open(-1, 0);
if (pidfd >= 0) {
@@ -153,6 +186,52 @@ int main(int argc, char **argv)
pid = get_pid_from_fdinfo_file(pidfd, "Pid:", sizeof("Pid:") - 1);
ksft_print_msg("pidfd %d refers to process with pid %d\n", pidfd, pid);
+ if (ioctl(pidfd, PIDFD_GET_INFO, &info) < 0) {
+ ksft_print_msg("%s - failed to get info from pidfd\n", strerror(errno));
+ goto on_error;
+ }
+ if (info.pid != pid) {
+ ksft_print_msg("pid from fdinfo file %d does not match pid from ioctl %d\n",
+ pid, info.pid);
+ goto on_error;
+ }
+ if (info.ppid != getppid()) {
+ ksft_print_msg("ppid %d does not match ppid from ioctl %d\n",
+ pid, info.pid);
+ goto on_error;
+ }
+ if (info.ruid != getuid()) {
+ ksft_print_msg("uid %d does not match uid from ioctl %d\n",
+ getuid(), info.ruid);
+ goto on_error;
+ }
+ if (info.rgid != getgid()) {
+ ksft_print_msg("gid %d does not match gid from ioctl %d\n",
+ getgid(), info.rgid);
+ goto on_error;
+ }
+ if (info.euid != geteuid()) {
+ ksft_print_msg("euid %d does not match euid from ioctl %d\n",
+ geteuid(), info.euid);
+ goto on_error;
+ }
+ if (info.egid != getegid()) {
+ ksft_print_msg("egid %d does not match egid from ioctl %d\n",
+ getegid(), info.egid);
+ goto on_error;
+ }
+ if (info.suid != geteuid()) {
+ ksft_print_msg("suid %d does not match suid from ioctl %d\n",
+ geteuid(), info.suid);
+ goto on_error;
+ }
+ if (info.sgid != getegid()) {
+ ksft_print_msg("sgid %d does not match sgid from ioctl %d\n",
+ getegid(), info.sgid);
+ goto on_error;
+ }
+ ksft_test_result_pass("get info from pidfd test: passed\n");
+
ret = 0;
on_error:
base-commit: 9852d85ec9d492ebef56dc5f229416c925758edc
--
2.45.2Hi,
kernel test robot noticed the following build errors:
[auto build test ERROR on 9852d85ec9d492ebef56dc5f229416c925758edc]
url: https://github.com/intel-lab-lkp/linux/commits/luca-boccassi-gmail-com/pidfd-add-ioctl-to-retrieve-pid-info/20241005-041639
base: 9852d85ec9d492ebef56dc5f229416c925758edc
patch link: https://lore.kernel.org/r/20241004194751.215507-1-luca.boccassi%40gmail.com
patch subject: [PATCH v4] pidfd: add ioctl to retrieve pid info
config: arm-randconfig-002-20241006 (https://download.01.org/0day-ci/archive/20241006/202410060624.Z84m82vK-lkp@intel.com/config)
compiler: clang version 14.0.6 (https://github.com/llvm/llvm-project f28c006a5895fc0e329fe15fead81e37457cb1d1)
reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20241006/202410060624.Z84m82vK-lkp@intel.com/reproduce)
If you fix the issue in a separate patch/commit (i.e. not just a new version of
the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <lkp@intel.com>
| Closes: https://lore.kernel.org/oe-kbuild-all/202410060624.Z84m82vK-lkp@intel.com/
All errors (new ones prefixed by >>):
>> fs/pidfs.c:156:25: error: implicit declaration of function 'task_css_check' is invalid in C99 [-Werror,-Wimplicit-function-declaration]
struct cgroup *cgrp = task_css_check(task, pids_cgrp_id, 1)->cgroup;
^
fs/pidfs.c:156:64: error: member reference type 'int' is not a pointer
struct cgroup *cgrp = task_css_check(task, pids_cgrp_id, 1)->cgroup;
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^
fs/pidfs.c:156:46: error: use of undeclared identifier 'pids_cgrp_id'
struct cgroup *cgrp = task_css_check(task, pids_cgrp_id, 1)->cgroup;
^
3 errors generated.
Kconfig warnings: (for reference only)
WARNING: unmet direct dependencies detected for GET_FREE_REGION
Depends on [n]: SPARSEMEM [=n]
Selected by [m]:
- RESOURCE_KUNIT_TEST [=m] && RUNTIME_TESTING_MENU [=y] && KUNIT [=m]
vim +/task_css_check +156 fs/pidfs.c
117
118 static long pidfd_info(struct task_struct *task, unsigned int cmd, unsigned long arg)
119 {
120 struct pidfd_info __user *uinfo = (struct pidfd_info __user *)arg;
121 size_t usize = _IOC_SIZE(cmd);
122 struct pidfd_info kinfo = {};
123 struct user_namespace *user_ns;
124 const struct cred *c;
125 __u64 request_mask;
126
127 if (!uinfo)
128 return -EINVAL;
129 if (usize < sizeof(struct pidfd_info))
130 return -EINVAL; /* First version, no smaller struct possible */
131
132 if (copy_from_user(&request_mask, &uinfo->request_mask, sizeof(request_mask)))
133 return -EFAULT;
134
135 c = get_task_cred(task);
136 if (!c)
137 return -ESRCH;
138
139 /* Unconditionally return identifiers and credentials, the rest only on request */
140
141 kinfo.pid = task_pid_vnr(task);
142 kinfo.tgid = task_tgid_vnr(task);
143 kinfo.ppid = task_ppid_nr_ns(task, task_active_pid_ns(task));
144
145 user_ns = current_user_ns();
146 kinfo.ruid = from_kuid_munged(user_ns, c->uid);
147 kinfo.rgid = from_kgid_munged(user_ns, c->gid);
148 kinfo.euid = from_kuid_munged(user_ns, c->euid);
149 kinfo.egid = from_kgid_munged(user_ns, c->egid);
150 kinfo.suid = from_kuid_munged(user_ns, c->suid);
151 kinfo.sgid = from_kgid_munged(user_ns, c->sgid);
152 kinfo.fsuid = from_kuid_munged(user_ns, c->fsuid);
153 kinfo.fsgid = from_kgid_munged(user_ns, c->fsgid);
154
155 if (request_mask & PIDFD_INFO_CGROUPID) {
> 156 struct cgroup *cgrp = task_css_check(task, pids_cgrp_id, 1)->cgroup;
157 if (!cgrp)
158 return -ENODEV;
159
160 kinfo.cgroupid = cgroup_id(cgrp);
161 kinfo.result_mask |= PIDFD_INFO_CGROUPID;
162 }
163
164 /*
165 * If userspace and the kernel have the same struct size it can just
166 * be copied. If userspace provides an older struct, only the bits that
167 * userspace knows about will be copied. If userspace provides a new
168 * struct, only the bits that the kernel knows about will be copied and
169 * the size value will be set to the size the kernel knows about.
170 */
171 if (copy_to_user(uinfo, &kinfo, min(usize, sizeof(kinfo))))
172 return -EFAULT;
173
174 return 0;
175 }
176
--
0-DAY CI Kernel Test Service
https://github.com/intel/lkp-tests/wiki
© 2016 - 2024 Red Hat, Inc.