summaryrefslogtreecommitdiffstats
path: root/samples/pidfd
diff options
context:
space:
mode:
authorTejun Heo <tj@kernel.org>2026-09-14 22:12:34 -1000
committerTejun Heo <tj@kernel.org>2026-09-15 06:57:20 -1000
commita9e3760b0838299649c0d57cca44daaf40ba3c33 (patch)
tree95807ff0e511c05602ef34808cfc660303309057 /samples/pidfd
downloadlinux-stable-a9e3760b0838299649c0d57cca44daaf40ba3c33.tar.gz
linux-stable-a9e3760b0838299649c0d57cca44daaf40ba3c33.zip
sched_ext: Maintain an online cid mask in the scheduler arenagrafted
Schedulers on the default cid mapping treat [0, nr_online_cids) as the online set and restart on hotplug. Schedulers that install their own mapping with scx_bpf_cid_override() have no way to learn which cids are online: the count no longer identifies members and the CPU-form cpumask is unusable from cid programs. This is an obvious hole in the cid API. Add scx_bpf_online_cmask(), a kernel-maintained cmask in the scheduler's arena, allocated alongside the per-CPU scratch masks and populated after the cid mapping is finalized and before ops.init(), for child schedulers too. The pointer stays valid through ops.exit() with no reference to take. It is the arena offset as a void pointer, the same form struct_ops arena arguments arrive in. The verifier types the void return as a scalar for the program's arena cast. The mask follows the SCX hotplug notifications: seeded from cpu_active_mask and updated before ops.cid_online/offline() runs, so it lags cpu_online_mask only inside a hotplug transition. Updates walk the scheduler list under the lock that also serializes unlinking. Reads are live, not atomic snapshots. Root initialization excludes hotplug. v2: Reworded the getter kerneldoc (Andrea Righi). Signed-off-by: Tejun Heo <tj@kernel.org> Reviewed-by: Andrea Righi <arighi@nvidia.com>
Diffstat (limited to 'samples/pidfd')
-rw-r--r--samples/pidfd/.gitignore2
-rw-r--r--samples/pidfd/Makefile4
-rw-r--r--samples/pidfd/pidfd-metadata.c120
3 files changed, 126 insertions, 0 deletions
diff --git a/samples/pidfd/.gitignore b/samples/pidfd/.gitignore
new file mode 100644
index 000000000..d4cfa3176
--- /dev/null
+++ b/samples/pidfd/.gitignore
@@ -0,0 +1,2 @@
+# SPDX-License-Identifier: GPL-2.0-only
+/pidfd-metadata
diff --git a/samples/pidfd/Makefile b/samples/pidfd/Makefile
new file mode 100644
index 000000000..9754e2d81
--- /dev/null
+++ b/samples/pidfd/Makefile
@@ -0,0 +1,4 @@
+# SPDX-License-Identifier: GPL-2.0
+usertprogs-always-y += pidfd-metadata
+
+userccflags += -I usr/include
diff --git a/samples/pidfd/pidfd-metadata.c b/samples/pidfd/pidfd-metadata.c
new file mode 100644
index 000000000..c459155da
--- /dev/null
+++ b/samples/pidfd/pidfd-metadata.c
@@ -0,0 +1,120 @@
+// SPDX-License-Identifier: GPL-2.0
+
+#define _GNU_SOURCE
+#include <err.h>
+#include <errno.h>
+#include <fcntl.h>
+#include <inttypes.h>
+#include <limits.h>
+#include <sched.h>
+#include <signal.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <sys/stat.h>
+#include <sys/syscall.h>
+#include <sys/types.h>
+#include <sys/wait.h>
+#include <unistd.h>
+
+#ifndef CLONE_PIDFD
+#define CLONE_PIDFD 0x00001000
+#endif
+
+#ifndef __NR_pidfd_send_signal
+#define __NR_pidfd_send_signal -1
+#endif
+
+static int do_child(void *args)
+{
+ printf("%d\n", getpid());
+ _exit(EXIT_SUCCESS);
+}
+
+static pid_t pidfd_clone(int flags, int *pidfd)
+{
+ size_t stack_size = 1024;
+ char *stack[1024] = { 0 };
+
+#ifdef __ia64__
+ return __clone2(do_child, stack, stack_size, flags | SIGCHLD, NULL, pidfd);
+#else
+ return clone(do_child, stack + stack_size, flags | SIGCHLD, NULL, pidfd);
+#endif
+}
+
+static inline int sys_pidfd_send_signal(int pidfd, int sig, siginfo_t *info,
+ unsigned int flags)
+{
+ return syscall(__NR_pidfd_send_signal, pidfd, sig, info, flags);
+}
+
+static int pidfd_metadata_fd(pid_t pid, int pidfd)
+{
+ int procfd, ret;
+ char path[100];
+
+ snprintf(path, sizeof(path), "/proc/%d", pid);
+ procfd = open(path, O_DIRECTORY | O_RDONLY | O_CLOEXEC);
+ if (procfd < 0) {
+ warn("Failed to open %s\n", path);
+ return -1;
+ }
+
+ /*
+ * Verify that the pid has not been recycled and our /proc/<pid> handle
+ * is still valid.
+ */
+ ret = sys_pidfd_send_signal(pidfd, 0, NULL, 0);
+ if (ret < 0) {
+ switch (errno) {
+ case EPERM:
+ /* Process exists, just not allowed to signal it. */
+ break;
+ default:
+ warn("Failed to signal process\n");
+ close(procfd);
+ procfd = -1;
+ }
+ }
+
+ return procfd;
+}
+
+int main(int argc, char *argv[])
+{
+ int pidfd = -1, ret = EXIT_FAILURE;
+ char buf[4096] = { 0 };
+ pid_t pid;
+ int procfd, statusfd;
+ ssize_t bytes;
+
+ pid = pidfd_clone(CLONE_PIDFD, &pidfd);
+ if (pid < 0)
+ err(ret, "CLONE_PIDFD");
+ if (pidfd == -1) {
+ warnx("CLONE_PIDFD is not supported by the kernel");
+ goto out;
+ }
+
+ procfd = pidfd_metadata_fd(pid, pidfd);
+ close(pidfd);
+ if (procfd < 0)
+ goto out;
+
+ statusfd = openat(procfd, "status", O_RDONLY | O_CLOEXEC);
+ close(procfd);
+ if (statusfd < 0)
+ goto out;
+
+ bytes = read(statusfd, buf, sizeof(buf));
+ if (bytes > 0)
+ bytes = write(STDOUT_FILENO, buf, bytes);
+ close(statusfd);
+ ret = EXIT_SUCCESS;
+
+out:
+ (void)wait(NULL);
+
+ exit(ret);
+}