bpf_prog_pack_free() uses header->size to decide whether the header
should be freed with module_memfree() or the bpf_prog_pack logic.
However, in kvmalloc() failure path of bpf_jit_binary_pack_alloc(),
header->size is not set yet. As a result, bpf_prog_pack_free() may treat
a slice of a pack as a standalone kvmalloc'd header and call
module_memfree() on the whole pack. This in turn causes use-after-free by
other users of the pack.
Fix this by setting ro_header->size before freeing ro_header.
Fixes: 33c9805860e5 ("bpf: Introduce bpf_jit_binary_pack_[alloc|finalize|free]")
Reported-by: syzbot+2f649ec6d2eea1495a8f@xxxxxxxxxxxxxxxxxxxxxxxxx
Reported-by: syzbot+ecb1e7e51c52f68f7481@xxxxxxxxxxxxxxxxxxxxxxxxx
Reported-by: syzbot+87f65c75f4a72db05445@xxxxxxxxxxxxxxxxxxxxxxxxx
Signed-off-by: Song Liu <song@xxxxxxxxxx>
---
kernel/bpf/core.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/kernel/bpf/core.c b/kernel/bpf/core.c
index 44623c9b5bb1..ebb0193d07f0 100644
--- a/kernel/bpf/core.c
+++ b/kernel/bpf/core.c
@@ -1069,6 +1069,7 @@ bpf_jit_binary_pack_alloc(unsigned int proglen, u8 **image_ptr,
*rw_header = kvmalloc(size, GFP_KERNEL);
if (!*rw_header) {
+ bpf_arch_text_copy(&ro_header->size, &size, sizeof(size));
bpf_prog_pack_free(ro_header);
bpf_jit_uncharge_modmem(size);
return NULL;
--
2.30.2
