[RFC PATCH 4/4] bpf, x64: Enable unprivlidged jit in vmalloc

[Rick Edgecombe <rick.p.edgecombe@intel.com>]

This enables unprivlidged JIT allocations to be made in vmalloc space
when the bpf jit limit is exceeded.

The logic is we use module space unless it is full or we are not
CAP_SYS_ADMIN and bpf_jit_limit is exceeded, in which case we use vmalloc
space. So vmalloc is only used when either the insertion would fail, or
BPF would fallback to the interpreter.

In the case of using vmalloc, it is not charged against bpf_jit_limit.

Cc: Daniel Borkmann <daniel@iogearbox.net>
Cc: Alexei Starovoitov <ast@fb.com>
Signed-off-by: Rick Edgecombe <rick.p.edgecombe@intel.com>
---
 arch/x86/net/bpf_jit_comp.c | 32 ++++++++++++++++++++++++++++++++
 1 file changed, 32 insertions(+)

diff --git a/arch/x86/net/bpf_jit_comp.c b/arch/x86/net/bpf_jit_comp.c
index c9781d471e31..66d2b32a1db1 100644
--- a/arch/x86/net/bpf_jit_comp.c
+++ b/arch/x86/net/bpf_jit_comp.c
@@ -1118,6 +1118,38 @@ struct x64_jit_data {
 	struct jit_context ctx;
 };
 
+void *bpf_jit_alloc_exec(unsigned long size)
+{
+	void *ret;
+	u32 pages = size / PAGE_SIZE;
+
+	/*
+	 * The logic is we use module space unless it is full or we are not
+	 * CAP_SYS_ADMIN and bpf_jit_limit is exceeded, in which case we use
+	 * vmalloc space.
+	 */
+	if (bpf_jit_charge_modmem(pages))
+		return vmalloc_exec(size);
+
+	ret = module_alloc(size);
+
+	if (!ret) {
+		bpf_jit_uncharge_modmem(pages);
+		/* If module space is full, try vmalloc */
+		return vmalloc_exec(size);
+	}
+
+	return ret;
+}
+
+void bpf_jit_free_exec(void *addr)
+{
+	if (is_vmalloc_addr(addr))
+		vfree(addr);
+	else
+		module_memfree(addr);
+}
+
 struct bpf_prog *bpf_int_jit_compile(struct bpf_prog *prog)
 {
 	struct bpf_binary_header *header = NULL;
-- 
2.17.1