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, 185 insertions, 0 deletions

diff --git a/arch/powerpc/boot/libfdt-wrapper.c b/arch/powerpc/boot/libfdt-wrapper.c
new file mode 100644
index 000000000..217d0d7a6
--- /dev/null
+++ b/arch/powerpc/boot/libfdt-wrapper.c
@@ -0,0 +1,185 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * This file does the necessary interface mapping between the bootwrapper
+ * device tree operations and the interface provided by shared source
+ * files flatdevicetree.[ch].
+ *
+ * Copyright 2007 David Gibson, IBM Corporation.
+ */
+
+#include <stddef.h>
+#include <stdio.h>
+#include <page.h>
+#include <libfdt.h>
+#include "ops.h"
+
+#define DEBUG	0
+#define BAD_ERROR(err)	(((err) < 0) \
+			 && ((err) != -FDT_ERR_NOTFOUND) \
+			 && ((err) != -FDT_ERR_EXISTS))
+
+#define check_err(err) \
+	({ \
+		if (BAD_ERROR(err) || ((err < 0) && DEBUG)) \
+			printf("%s():%d  %s\n\r", __func__, __LINE__, \
+			       fdt_strerror(err)); \
+		if (BAD_ERROR(err)) \
+			exit(); \
+		(err < 0) ? -1 : 0; \
+	})
+
+#define offset_devp(off)	\
+	({ \
+		unsigned long _offset = (off); \
+		check_err(_offset) ? NULL : (void *)(_offset+1); \
+	})
+
+#define devp_offset_find(devp)	(((unsigned long)(devp))-1)
+#define devp_offset(devp)	(devp ? ((unsigned long)(devp))-1 : 0)
+
+static void *fdt;
+static void *buf; /* = NULL */
+
+#define EXPAND_GRANULARITY	1024
+
+static void expand_buf(int minexpand)
+{
+	int size = fdt_totalsize(fdt);
+	int rc;
+
+	size = _ALIGN(size + minexpand, EXPAND_GRANULARITY);
+	buf = platform_ops.realloc(buf, size);
+	if (!buf)
+		fatal("Couldn't find %d bytes to expand device tree\n\r", size);
+	rc = fdt_open_into(fdt, buf, size);
+	if (rc != 0)
+		fatal("Couldn't expand fdt into new buffer: %s\n\r",
+		      fdt_strerror(rc));
+
+	fdt = buf;
+}
+
+static void *fdt_wrapper_finddevice(const char *path)
+{
+	return offset_devp(fdt_path_offset(fdt, path));
+}
+
+static int fdt_wrapper_getprop(const void *devp, const char *name,
+			       void *buf, const int buflen)
+{
+	const void *p;
+	int len;
+
+	p = fdt_getprop(fdt, devp_offset(devp), name, &len);
+	if (!p)
+		return check_err(len);
+	memcpy(buf, p, min(len, buflen));
+	return len;
+}
+
+static int fdt_wrapper_setprop(const void *devp, const char *name,
+			       const void *buf, const int len)
+{
+	int rc;
+
+	rc = fdt_setprop(fdt, devp_offset(devp), name, buf, len);
+	if (rc == -FDT_ERR_NOSPACE) {
+		expand_buf(len + 16);
+		rc = fdt_setprop(fdt, devp_offset(devp), name, buf, len);
+	}
+
+	return check_err(rc);
+}
+
+static int fdt_wrapper_del_node(const void *devp)
+{
+	return fdt_del_node(fdt, devp_offset(devp));
+}
+
+static void *fdt_wrapper_get_parent(const void *devp)
+{
+	return offset_devp(fdt_parent_offset(fdt, devp_offset(devp)));
+}
+
+static void *fdt_wrapper_create_node(const void *devp, const char *name)
+{
+	int offset;
+
+	offset = fdt_add_subnode(fdt, devp_offset(devp), name);
+	if (offset == -FDT_ERR_NOSPACE) {
+		expand_buf(strlen(name) + 16);
+		offset = fdt_add_subnode(fdt, devp_offset(devp), name);
+	}
+
+	return offset_devp(offset);
+}
+
+static void *fdt_wrapper_find_node_by_prop_value(const void *prev,
+						 const char *name,
+						 const char *val,
+						 int len)
+{
+	int offset = fdt_node_offset_by_prop_value(fdt, devp_offset_find(prev),
+	                                           name, val, len);
+	return offset_devp(offset);
+}
+
+static void *fdt_wrapper_find_node_by_compatible(const void *prev,
+						 const char *val)
+{
+	int offset = fdt_node_offset_by_compatible(fdt, devp_offset_find(prev),
+	                                           val);
+	return offset_devp(offset);
+}
+
+static char *fdt_wrapper_get_path(const void *devp, char *buf, int len)
+{
+	int rc;
+
+	rc = fdt_get_path(fdt, devp_offset(devp), buf, len);
+	if (check_err(rc))
+		return NULL;
+	return buf;
+}
+
+static unsigned long fdt_wrapper_finalize(void)
+{
+	int rc;
+
+	rc = fdt_pack(fdt);
+	if (rc != 0)
+		fatal("Couldn't pack flat tree: %s\n\r",
+		      fdt_strerror(rc));
+	return (unsigned long)fdt;
+}
+
+void fdt_init(void *blob)
+{
+	int err;
+	int bufsize;
+
+	dt_ops.finddevice = fdt_wrapper_finddevice;
+	dt_ops.getprop = fdt_wrapper_getprop;
+	dt_ops.setprop = fdt_wrapper_setprop;
+	dt_ops.get_parent = fdt_wrapper_get_parent;
+	dt_ops.create_node = fdt_wrapper_create_node;
+	dt_ops.find_node_by_prop_value = fdt_wrapper_find_node_by_prop_value;
+	dt_ops.find_node_by_compatible = fdt_wrapper_find_node_by_compatible;
+	dt_ops.del_node = fdt_wrapper_del_node;
+	dt_ops.get_path = fdt_wrapper_get_path;
+	dt_ops.finalize = fdt_wrapper_finalize;
+
+	/* Make sure the dt blob is the right version and so forth */
+	fdt = blob;
+	bufsize = fdt_totalsize(fdt) + EXPAND_GRANULARITY;
+	buf = malloc(bufsize);
+	if(!buf)
+		fatal("malloc failed. can't relocate the device tree\n\r");
+
+	err = fdt_open_into(fdt, buf, bufsize);
+
+	if (err != 0)
+		fatal("fdt_init(): %s\n\r", fdt_strerror(err));
+
+	fdt = buf;
+}