ARM: 9485/1: mm: acquire mmap write lock around show_pte() for user faultsgrafted

When CONFIG_DEBUG_USER=y, and cmdline "user_debug=31" is set,
a user fault may trigger show_pte() without any lock.
If another thread in the same process concurrently calls munmap(),
the page table pages may be freed while show_pte() is still traversing
them, causing a use-after-free in show_pte().

If CONFIG_ARM_LPAE=y, this may cause a kernel panic if the pages table
of PMD are freed when show_pte() is running.

Acquire mmap_write_lock() around show_pte() for user faults to fix the
contention.

For user faults, additionally restrict that show_pte() is called only
when the addr is a user-space address (addr < TASK_SIZE). This is because
the lock of tsk->mm only protects the virtual memory of user address space,
furthermore, dumping the page tables of a kernel-space address for user
faults is unnecessary and may have security implications.

Keep everything unchanged for kernel faults, because the kernel is
already in the "oops" state, acquiring a lock may risk a deadlock.

Co-developed-by: Qi Xi <xiqi2@huawei.com>

Fixes: 6d021b724481 ("ARM: dump pgd, pmd and pte states on unhandled data abort faults")
Link: https://lore.kernel.org/20260716014022.2823-1-xieyuanbin1@huawei.com
Acked-by: Lorenzo Stoakes (ARM) <ljs@kernel.org>
Reviewed-by: Linus Walleij <linusw@kernel.org>
Signed-off-by: Qi Xi <xiqi2@huawei.com>
Signed-off-by: Xie Yuanbin <xieyuanbin1@huawei.com>
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>

1 files changed, 170 insertions, 0 deletions

diff --git a/tools/perf/util/strbuf.c b/tools/perf/util/strbuf.c
new file mode 100644
index 000000000..a64a37628
--- /dev/null
+++ b/tools/perf/util/strbuf.c
@@ -0,0 +1,170 @@
+// SPDX-License-Identifier: GPL-2.0
+#include "cache.h"
+#include "debug.h"
+#include "strbuf.h"
+#include <linux/kernel.h>
+#include <linux/string.h>
+#include <linux/zalloc.h>
+#include <errno.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <unistd.h>
+
+/*
+ * Used as the default ->buf value, so that people can always assume
+ * buf is non NULL and ->buf is NUL terminated even for a freshly
+ * initialized strbuf.
+ */
+char strbuf_slopbuf[1];
+
+int strbuf_init(struct strbuf *sb, ssize_t hint)
+{
+	sb->alloc = sb->len = 0;
+	sb->buf = strbuf_slopbuf;
+	if (hint)
+		return strbuf_grow(sb, hint);
+	return 0;
+}
+
+void strbuf_release(struct strbuf *sb)
+{
+	if (sb->alloc) {
+		zfree(&sb->buf);
+		strbuf_init(sb, 0);
+	}
+}
+
+char *strbuf_detach(struct strbuf *sb, size_t *sz)
+{
+	char *res = sb->alloc ? sb->buf : NULL;
+	if (sz)
+		*sz = sb->len;
+	strbuf_init(sb, 0);
+	return res;
+}
+
+int strbuf_grow(struct strbuf *sb, size_t extra)
+{
+	char *buf;
+	size_t nr = sb->len + extra + 1;
+
+	if (nr < sb->alloc)
+		return 0;
+
+	if (nr <= sb->len)
+		return -E2BIG;
+
+	if (alloc_nr(sb->alloc) > nr)
+		nr = alloc_nr(sb->alloc);
+
+	/*
+	 * Note that sb->buf == strbuf_slopbuf if sb->alloc == 0, and it is
+	 * a static variable. Thus we have to avoid passing it to realloc.
+	 */
+	buf = realloc(sb->alloc ? sb->buf : NULL, nr * sizeof(*buf));
+	if (!buf)
+		return -ENOMEM;
+
+	sb->buf = buf;
+	sb->alloc = nr;
+	return 0;
+}
+
+int strbuf_addch(struct strbuf *sb, int c)
+{
+	int ret = strbuf_grow(sb, 1);
+	if (ret)
+		return ret;
+
+	sb->buf[sb->len++] = c;
+	sb->buf[sb->len] = '\0';
+	return 0;
+}
+
+int strbuf_add(struct strbuf *sb, const void *data, size_t len)
+{
+	int ret = strbuf_grow(sb, len);
+	if (ret)
+		return ret;
+
+	memcpy(sb->buf + sb->len, data, len);
+	return strbuf_setlen(sb, sb->len + len);
+}
+
+static int strbuf_addv(struct strbuf *sb, const char *fmt, va_list ap)
+{
+	int len, ret;
+	va_list ap_saved;
+
+	if (!strbuf_avail(sb)) {
+		ret = strbuf_grow(sb, 64);
+		if (ret)
+			return ret;
+	}
+
+	va_copy(ap_saved, ap);
+	len = vsnprintf(sb->buf + sb->len, sb->alloc - sb->len, fmt, ap);
+	if (len < 0) {
+		va_end(ap_saved);
+		return len;
+	}
+	if (len > strbuf_avail(sb)) {
+		ret = strbuf_grow(sb, len);
+		if (ret) {
+			va_end(ap_saved);
+			return ret;
+		}
+		len = vsnprintf(sb->buf + sb->len, sb->alloc - sb->len, fmt, ap_saved);
+		if (len > strbuf_avail(sb)) {
+			pr_debug("this should not happen, your vsnprintf is broken");
+			va_end(ap_saved);
+			return -EINVAL;
+		}
+	}
+	va_end(ap_saved);
+	return strbuf_setlen(sb, sb->len + len);
+}
+
+int strbuf_addf(struct strbuf *sb, const char *fmt, ...)
+{
+	va_list ap;
+	int ret;
+
+	va_start(ap, fmt);
+	ret = strbuf_addv(sb, fmt, ap);
+	va_end(ap);
+	return ret;
+}
+
+ssize_t strbuf_read(struct strbuf *sb, int fd, ssize_t hint)
+{
+	size_t oldlen = sb->len;
+	size_t oldalloc = sb->alloc;
+	int ret;
+
+	ret = strbuf_grow(sb, hint ? hint : 8192);
+	if (ret)
+		return ret;
+
+	for (;;) {
+		ssize_t cnt;
+
+		cnt = read(fd, sb->buf + sb->len, sb->alloc - sb->len - 1);
+		if (cnt < 0) {
+			if (oldalloc == 0)
+				strbuf_release(sb);
+			else
+				strbuf_setlen(sb, oldlen);
+			return cnt;
+		}
+		if (!cnt)
+			break;
+		sb->len += cnt;
+		ret = strbuf_grow(sb, 8192);
+		if (ret)
+			return ret;
+	}
+
+	sb->buf[sb->len] = '\0';
+	return sb->len - oldlen;
+}