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, 315 insertions, 0 deletions

diff --git a/init/do_mounts_rd.c b/init/do_mounts_rd.c
new file mode 100644
index 000000000..48bfab2fc
--- /dev/null
+++ b/init/do_mounts_rd.c
@@ -0,0 +1,315 @@
+// SPDX-License-Identifier: GPL-2.0
+#include <linux/kernel.h>
+#include <linux/fs.h>
+#include <linux/minix_fs.h>
+#include <linux/ext2_fs.h>
+#include <linux/romfs_fs.h>
+#include <uapi/linux/cramfs_fs.h>
+#include <linux/initrd.h>
+#include <linux/string.h>
+#include <linux/string_choices.h>
+#include <linux/slab.h>
+
+#include "do_mounts.h"
+#include "../fs/squashfs/squashfs_fs.h"
+
+#include <linux/decompress/generic.h>
+
+static struct file *in_file, *out_file;
+static loff_t in_pos, out_pos;
+
+int __initdata rd_image_start;		/* starting block # of image */
+
+static int __init ramdisk_start_setup(char *str)
+{
+	pr_warn("ramdisk_start= option is deprecated and will be removed soon\n");
+	return kstrtoint(str, 0, &rd_image_start) == 0;
+}
+__setup("ramdisk_start=", ramdisk_start_setup);
+
+static int __init crd_load(decompress_fn deco);
+
+/*
+ * This routine tries to find a RAM disk image to load, and returns the
+ * number of blocks to read for a non-compressed image, 0 if the image
+ * is a compressed image, and -1 if an image with the right magic
+ * numbers could not be found.
+ *
+ * We currently check for the following magic numbers:
+ *	minix
+ *	ext2
+ *	romfs
+ *	cramfs
+ *	squashfs
+ *	gzip
+ *	bzip2
+ *	lzma
+ *	xz
+ *	lzo
+ *	lz4
+ */
+static int __init
+identify_ramdisk_image(struct file *file, loff_t pos,
+		decompress_fn *decompressor)
+{
+	const int size = 512;
+	struct minix_super_block *minixsb;
+	struct romfs_super_block *romfsb;
+	struct cramfs_super *cramfsb;
+	struct squashfs_super_block *squashfsb;
+	int nblocks = -1;
+	unsigned char *buf;
+	const char *compress_name;
+	unsigned long n;
+	int start_block = rd_image_start;
+
+	buf = kmalloc(size, GFP_KERNEL);
+	if (!buf)
+		return -ENOMEM;
+
+	minixsb = (struct minix_super_block *) buf;
+	romfsb = (struct romfs_super_block *) buf;
+	cramfsb = (struct cramfs_super *) buf;
+	squashfsb = (struct squashfs_super_block *) buf;
+	memset(buf, 0xe5, size);
+
+	/*
+	 * Read block 0 to test for compressed kernel
+	 */
+	pos = start_block * BLOCK_SIZE;
+	kernel_read(file, buf, size, &pos);
+
+	*decompressor = decompress_method(buf, size, &compress_name);
+	if (compress_name) {
+		printk(KERN_NOTICE "RAMDISK: %s image found at block %d\n",
+		       compress_name, start_block);
+		if (!*decompressor)
+			printk(KERN_EMERG
+			       "RAMDISK: %s decompressor not configured!\n",
+			       compress_name);
+		nblocks = 0;
+		goto done;
+	}
+
+	/* romfs is at block zero too */
+	if (romfsb->word0 == ROMSB_WORD0 &&
+	    romfsb->word1 == ROMSB_WORD1) {
+		printk(KERN_NOTICE
+		       "RAMDISK: romfs filesystem found at block %d\n",
+		       start_block);
+		nblocks = (ntohl(romfsb->size)+BLOCK_SIZE-1)>>BLOCK_SIZE_BITS;
+		goto done;
+	}
+
+	if (cramfsb->magic == CRAMFS_MAGIC) {
+		printk(KERN_NOTICE
+		       "RAMDISK: cramfs filesystem found at block %d\n",
+		       start_block);
+		nblocks = (cramfsb->size + BLOCK_SIZE - 1) >> BLOCK_SIZE_BITS;
+		goto done;
+	}
+
+	/* squashfs is at block zero too */
+	if (le32_to_cpu(squashfsb->s_magic) == SQUASHFS_MAGIC) {
+		printk(KERN_NOTICE
+		       "RAMDISK: squashfs filesystem found at block %d\n",
+		       start_block);
+		nblocks = (le64_to_cpu(squashfsb->bytes_used) + BLOCK_SIZE - 1)
+			 >> BLOCK_SIZE_BITS;
+		goto done;
+	}
+
+	/*
+	 * Read 512 bytes further to check if cramfs is padded
+	 */
+	pos = start_block * BLOCK_SIZE + 0x200;
+	kernel_read(file, buf, size, &pos);
+
+	if (cramfsb->magic == CRAMFS_MAGIC) {
+		printk(KERN_NOTICE
+		       "RAMDISK: cramfs filesystem found at block %d\n",
+		       start_block);
+		nblocks = (cramfsb->size + BLOCK_SIZE - 1) >> BLOCK_SIZE_BITS;
+		goto done;
+	}
+
+	/*
+	 * Read block 1 to test for minix and ext2 superblock
+	 */
+	pos = (start_block + 1) * BLOCK_SIZE;
+	kernel_read(file, buf, size, &pos);
+
+	/* Try minix */
+	if (minixsb->s_magic == MINIX_SUPER_MAGIC ||
+	    minixsb->s_magic == MINIX_SUPER_MAGIC2) {
+		printk(KERN_NOTICE
+		       "RAMDISK: Minix filesystem found at block %d\n",
+		       start_block);
+		nblocks = minixsb->s_nzones << minixsb->s_log_zone_size;
+		goto done;
+	}
+
+	/* Try ext2 */
+	n = ext2_image_size(buf);
+	if (n) {
+		printk(KERN_NOTICE
+		       "RAMDISK: ext2 filesystem found at block %d\n",
+		       start_block);
+		nblocks = n;
+		goto done;
+	}
+
+	printk(KERN_NOTICE
+	       "RAMDISK: Couldn't find valid RAM disk image starting at %d.\n",
+	       start_block);
+
+done:
+	kfree(buf);
+	return nblocks;
+}
+
+static unsigned long nr_blocks(struct file *file)
+{
+	struct inode *inode = file->f_mapping->host;
+
+	if (!S_ISBLK(inode->i_mode))
+		return 0;
+	return i_size_read(inode) >> 10;
+}
+
+int __init rd_load_image(void)
+{
+	int res = 0;
+	unsigned long rd_blocks, devblocks, nr_disks;
+	int nblocks, i;
+	char *buf = NULL;
+	unsigned short rotate = 0;
+	decompress_fn decompressor = NULL;
+	char rotator[4] = { '|' , '/' , '-' , '\\' };
+
+	out_file = filp_open("/dev/ram", O_RDWR, 0);
+	if (IS_ERR(out_file))
+		goto out;
+
+	in_file = filp_open("/initrd.image", O_RDONLY, 0);
+	if (IS_ERR(in_file))
+		goto noclose_input;
+
+	in_pos = rd_image_start * BLOCK_SIZE;
+	nblocks = identify_ramdisk_image(in_file, in_pos, &decompressor);
+	if (nblocks < 0)
+		goto done;
+
+	if (nblocks == 0) {
+		if (crd_load(decompressor) == 0)
+			goto successful_load;
+		goto done;
+	}
+
+	/*
+	 * NOTE NOTE: nblocks is not actually blocks but
+	 * the number of kibibytes of data to load into a ramdisk.
+	 */
+	rd_blocks = nr_blocks(out_file);
+	if (nblocks > rd_blocks) {
+		printk("RAMDISK: image too big! (%dKiB/%ldKiB)\n",
+		       nblocks, rd_blocks);
+		goto done;
+	}
+
+	/*
+	 * OK, time to copy in the data
+	 */
+	devblocks = nblocks;
+
+	if (devblocks == 0) {
+		printk(KERN_ERR "RAMDISK: could not determine device size\n");
+		goto done;
+	}
+
+	buf = kmalloc(BLOCK_SIZE, GFP_KERNEL);
+	if (!buf) {
+		printk(KERN_ERR "RAMDISK: could not allocate buffer\n");
+		goto done;
+	}
+
+	nr_disks = (nblocks - 1) / devblocks + 1;
+	pr_notice("RAMDISK: Loading %dKiB [%ld disk%s] into ram disk... ",
+		  nblocks, nr_disks, str_plural(nr_disks));
+	for (i = 0; i < nblocks; i++) {
+		if (i && (i % devblocks == 0)) {
+			pr_cont("done disk #1.\n");
+			rotate = 0;
+			fput(in_file);
+			break;
+		}
+		kernel_read(in_file, buf, BLOCK_SIZE, &in_pos);
+		kernel_write(out_file, buf, BLOCK_SIZE, &out_pos);
+		if (!IS_ENABLED(CONFIG_S390) && !(i % 16)) {
+			pr_cont("%c\b", rotator[rotate & 0x3]);
+			rotate++;
+		}
+	}
+	pr_cont("done.\n");
+
+successful_load:
+	res = 1;
+done:
+	fput(in_file);
+noclose_input:
+	fput(out_file);
+out:
+	kfree(buf);
+	init_unlink("/dev/ram");
+	return res;
+}
+
+static int exit_code;
+static int decompress_error;
+
+static long __init compr_fill(void *buf, unsigned long len)
+{
+	long r = kernel_read(in_file, buf, len, &in_pos);
+	if (r < 0)
+		printk(KERN_ERR "RAMDISK: error while reading compressed data");
+	else if (r == 0)
+		printk(KERN_ERR "RAMDISK: EOF while reading compressed data");
+	return r;
+}
+
+static long __init compr_flush(void *window, unsigned long outcnt)
+{
+	long written = kernel_write(out_file, window, outcnt, &out_pos);
+	if (written != outcnt) {
+		if (decompress_error == 0)
+			printk(KERN_ERR
+			       "RAMDISK: incomplete write (%ld != %ld)\n",
+			       written, outcnt);
+		decompress_error = 1;
+		return -1;
+	}
+	return outcnt;
+}
+
+static void __init error(char *x)
+{
+	printk(KERN_ERR "%s\n", x);
+	exit_code = 1;
+	decompress_error = 1;
+}
+
+static int __init crd_load(decompress_fn deco)
+{
+	int result;
+
+	if (!deco) {
+		pr_emerg("Invalid ramdisk decompression routine.  "
+			 "Select appropriate config option.\n");
+		panic("Could not decompress initial ramdisk image.");
+	}
+
+	result = deco(NULL, 0, compr_fill, compr_flush, NULL, NULL, error);
+	if (decompress_error)
+		result = 1;
+	return result;
+}