ftrace: Also allocate and copy hash for reading of filter files

Currently the reader of set_ftrace_filter and set_ftrace_notrace just adds the pointer to the global tracer hash to its iterator. Unlike the writer that allocates a copy of the hash, the reader keeps the pointer to the filter hashes. This is problematic because this pointer is static across function calls that release the locks that can update the global tracer hashes. This can cause UAF and similar bugs. Allocate and copy the hash for reading the filter files like it is done for the writers. This not only fixes UAF bugs, but also makes the code a bit simpler as it doesn't have to differentiate when to free the iterator's hash between writers and readers.

Steven Rostedt committed 1 year ago

Commit bfb336c1 parent 4013aef1 file · +10 −9Browse files

Showing the changes to kernel/trace (1 of 1 file). Show all

Unified · Split

kernel/trace/ftrace.c+10−9

@@ -4661,13 +4661,17 @@ ftrace_regex_open(struct ftrace_ops *ops, int flag, · 4660 unmodified lines
4661	4661	} else {
4662	4662	iter->hash = alloc_and_copy_ftrace_hash(size_bits, hash);
4663	4663	}
	4664	+ } else {
	4665	+ if (hash)
	4666	+ iter->hash = alloc_and_copy_ftrace_hash(hash->size_bits, hash);
	4667	+ else
	4668	+ iter->hash = EMPTY_HASH;
	4669	+ }
4664	4670
4665		− if (!iter->hash) {
4666		− trace_parser_put(&iter->parser);
4667		− goto out_unlock;
4668		− }
4669		− } else
4670		− iter->hash = hash;
	4671	+ if (!iter->hash) {
	4672	+ trace_parser_put(&iter->parser);
	4673	+ goto out_unlock;
	4674	+ }
4671	4675
4672	4676	ret = 0;
4673	4677
@@ -6543,9 +6547,6 @@ int ftrace_regex_release(struct inode inode, struct file file) · 1869 unmodified lines
6543	6547	ftrace_hash_move_and_update_ops(iter->ops, orig_hash,
6544	6548	iter->hash, filter_hash);
6545	6549	mutex_unlock(&ftrace_lock);
6546		− } else {
6547		− /* For read only, the hash is the ops hash */
6548		− iter->hash = NULL;
6549	6550	}
6550	6551
6551	6552	mutex_unlock(&iter->ops->func_hash->regex_lock);