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>

1 files changed, 151 insertions, 0 deletions

diff --git a/arch/powerpc/boot/mktree.c b/arch/powerpc/boot/mktree.c
new file mode 100644
index 000000000..dc603f3c1
--- /dev/null
+++ b/arch/powerpc/boot/mktree.c
@@ -0,0 +1,151 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Makes a tree bootable image for IBM Evaluation boards.
+ * Basically, just take a zImage, skip the ELF header, and stuff
+ * a 32 byte header on the front.
+ *
+ * We use htonl, which is a network macro, to make sure we're doing
+ * The Right Thing on an LE machine.  It's non-obvious, but it should
+ * work on anything BSD'ish.
+ */
+
+#include <fcntl.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <sys/stat.h>
+#include <unistd.h>
+#include <netinet/in.h>
+#ifdef __sun__
+#include <inttypes.h>
+#else
+#include <stdint.h>
+#endif
+
+/* This gets tacked on the front of the image.  There are also a few
+ * bytes allocated after the _start label used by the boot rom (see
+ * head.S for details).
+ */
+typedef struct boot_block {
+	uint32_t bb_magic;		/* 0x0052504F */
+	uint32_t bb_dest;		/* Target address of the image */
+	uint32_t bb_num_512blocks;	/* Size, rounded-up, in 512 byte blks */
+	uint32_t bb_debug_flag;	/* Run debugger or image after load */
+	uint32_t bb_entry_point;	/* The image address to start */
+	uint32_t bb_checksum;	/* 32 bit checksum including header */
+	uint32_t reserved[2];
+} boot_block_t;
+
+#define IMGBLK	512
+unsigned int	tmpbuf[IMGBLK / sizeof(unsigned int)];
+
+int main(int argc, char *argv[])
+{
+	int	in_fd, out_fd;
+	int	nblks, i;
+	unsigned int	cksum, *cp;
+	struct	stat	st;
+	boot_block_t	bt;
+
+	if (argc < 5) {
+		fprintf(stderr, "usage: %s <zImage-file> <boot-image> <load address> <entry point>\n",argv[0]);
+		exit(1);
+	}
+
+	if (stat(argv[1], &st) < 0) {
+		perror("stat");
+		exit(2);
+	}
+
+	nblks = (st.st_size + IMGBLK) / IMGBLK;
+
+	bt.bb_magic = htonl(0x0052504F);
+
+	/* If we have the optional entry point parameter, use it */
+	bt.bb_dest = htonl(strtoul(argv[3], NULL, 0));
+	bt.bb_entry_point = htonl(strtoul(argv[4], NULL, 0));
+
+	/* We know these from the linker command.
+	 * ...and then move it up into memory a little more so the
+	 * relocation can happen.
+	 */
+	bt.bb_num_512blocks = htonl(nblks);
+	bt.bb_debug_flag = 0;
+
+	bt.bb_checksum = 0;
+
+	/* To be neat and tidy :-).
+	*/
+	bt.reserved[0] = 0;
+	bt.reserved[1] = 0;
+
+	if ((in_fd = open(argv[1], O_RDONLY)) < 0) {
+		perror("zImage open");
+		exit(3);
+	}
+
+	if ((out_fd = open(argv[2], (O_RDWR | O_CREAT | O_TRUNC), 0666)) < 0) {
+		perror("bootfile open");
+		exit(3);
+	}
+
+	cksum = 0;
+	cp = (void *)&bt;
+	for (i = 0; i < sizeof(bt) / sizeof(unsigned int); i++)
+		cksum += *cp++;
+
+	/* Assume zImage is an ELF file, and skip the 64K header.
+	*/
+	if (read(in_fd, tmpbuf, sizeof(tmpbuf)) != sizeof(tmpbuf)) {
+		fprintf(stderr, "%s is too small to be an ELF image\n",
+				argv[1]);
+		exit(4);
+	}
+
+	if (tmpbuf[0] != htonl(0x7f454c46)) {
+		fprintf(stderr, "%s is not an ELF image\n", argv[1]);
+		exit(4);
+	}
+
+	if (lseek(in_fd, (64 * 1024), SEEK_SET) < 0) {
+		fprintf(stderr, "%s failed to seek in ELF image\n", argv[1]);
+		exit(4);
+	}
+
+	nblks -= (64 * 1024) / IMGBLK;
+
+	/* And away we go......
+	*/
+	if (write(out_fd, &bt, sizeof(bt)) != sizeof(bt)) {
+		perror("boot-image write");
+		exit(5);
+	}
+
+	while (nblks-- > 0) {
+		if (read(in_fd, tmpbuf, sizeof(tmpbuf)) < 0) {
+			perror("zImage read");
+			exit(5);
+		}
+		cp = tmpbuf;
+		for (i = 0; i < sizeof(tmpbuf) / sizeof(unsigned int); i++)
+			cksum += *cp++;
+		if (write(out_fd, tmpbuf, sizeof(tmpbuf)) != sizeof(tmpbuf)) {
+			perror("boot-image write");
+			exit(5);
+		}
+	}
+
+	/* rewrite the header with the computed checksum.
+	*/
+	bt.bb_checksum = htonl(cksum);
+	if (lseek(out_fd, 0, SEEK_SET) < 0) {
+		perror("rewrite seek");
+		exit(1);
+	}
+	if (write(out_fd, &bt, sizeof(bt)) != sizeof(bt)) {
+		perror("boot-image rewrite");
+		exit(1);
+	}
+
+	exit(0);
+}