Thread (28 messages) 28 messages, 7 authors, 2022-08-10

Re: [PATCH v9 03/10] btf: Handle dynamic pointer parameter in kfuncs

From: Daniel Borkmann <daniel@iogearbox.net>
Date: 2022-08-09 22:13:20
Also in: bpf, keyrings, linux-doc, linux-kselftest, lkml
Subsystem: networking [general], the rest · Maintainers: "David S. Miller", Eric Dumazet, Jakub Kicinski, Paolo Abeni, Linus Torvalds

On 8/9/22 3:45 PM, Roberto Sassu wrote:
quoted hunk ↗ jump to hunk
Allow the bpf_dynptr_kern parameter to be specified in kfuncs. Also, ensure
that the dynamic pointer is valid and initialized.

Cc: Joanne Koong <redacted>
Signed-off-by: Roberto Sassu <roberto.sassu@huawei.com>
---
  include/linux/bpf_verifier.h |  3 +++
  kernel/bpf/btf.c             | 17 +++++++++++++++++
  kernel/bpf/verifier.c        |  4 ++--
  3 files changed, 22 insertions(+), 2 deletions(-)
diff --git a/include/linux/bpf_verifier.h b/include/linux/bpf_verifier.h
index 2e3bad8640dc..55876fbdbae2 100644
--- a/include/linux/bpf_verifier.h
+++ b/include/linux/bpf_verifier.h
@@ -560,6 +560,9 @@ int check_kfunc_mem_size_reg(struct bpf_verifier_env *env, struct bpf_reg_state
  			     u32 regno);
  int check_mem_reg(struct bpf_verifier_env *env, struct bpf_reg_state *reg,
  		   u32 regno, u32 mem_size);
+bool is_dynptr_reg_valid_init(struct bpf_verifier_env *env,
+			      struct bpf_reg_state *reg,
+			      enum bpf_arg_type arg_type);
  
  /* this lives here instead of in bpf.h because it needs to dereference tgt_prog */
  static inline u64 bpf_trampoline_compute_key(const struct bpf_prog *tgt_prog,
diff --git a/kernel/bpf/btf.c b/kernel/bpf/btf.c
index 67dfc728fbf8..17cca396c89f 100644
--- a/kernel/bpf/btf.c
+++ b/kernel/bpf/btf.c
@@ -6363,6 +6363,8 @@ static int btf_check_func_arg_match(struct bpf_verifier_env *env,
  
  			if (is_kfunc) {
  				bool arg_mem_size = i + 1 < nargs && is_kfunc_arg_mem_size(btf, &args[i + 1], &regs[regno + 1]);
+				bool arg_dynptr = btf_type_is_struct(ref_t) &&
+						  !strcmp(ref_tname, "bpf_dynptr_kern");
For the strcmp(), could we add some BUILD_BUG_ON() if "bpf_dynptr_kern" ever
gets renamed? I played a bit with the below compile-tested toy example. If we
rename foo_bar into something else, we then get:

   [...]
   CC      net/core/dev.o
   In file included from <command-line>:
   net/core/dev.c: In function ‘net_dev_init’:
   net/core/dev.c:11376:25: error: invalid application of ‘sizeof’ to incomplete type ‘struct foo_bar’
   11376 |  ({ BUILD_BUG_ON(sizeof(struct x) < 0); \
         |                         ^~~~~~
   [...]
diff --git a/net/core/dev.c b/net/core/dev.c
index 716df64fcfa5..a2ed271f7ded 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -11368,16 +11368,30 @@ static struct pernet_operations __net_initdata default_device_ops = {
   *
   */

+struct foo_bar {
+       int a;
+};
+
+#define stringify_struct(x)                    \
+       ({ BUILD_BUG_ON(sizeof(struct x) < 0);  \
+       __stringify(x); })
+
  /*
   *       This is called single threaded during boot, so no need
   *       to take the rtnl semaphore.
   */
  static int __init net_dev_init(void)
  {
+       const char *ref_tname = "abc";
         int i, rc = -ENOMEM;

         BUG_ON(!dev_boot_phase);

+       printk("%s\n", stringify_struct(foo_bar));
+
+       if (!strcmp(ref_tname, stringify_struct(foo_bar)))
+               goto out;
+
         if (dev_proc_init())
                 goto out;
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help