gentoo-overlay/sys-process/audit/files/audit-3.0.8-linux-headers-5.17.patch

Upstream rejected a workaround/fix at https://github.com/linux-audit/audit-userspace/pull/253
/ https://github.com/linux-audit/audit-userspace/issues/252#issuecomment-1078595249.

Instead, in Fedora (same maintainer as upstream), they're patching the headers then unpatching before install.

Apparently the swig bindings are on their way out but I'm not convinced that's going to be a quick migration given the API will.. surely change?

It's not ideal but let's take the patch slyfox ended up using in nixpkgs anyway.

https://bugs.gentoo.org/836702

From beed138222421a2eb4212d83cb889404bd7efc49 Mon Sep 17 00:00:00 2001
From: Sergei Trofimovich <slyich@gmail.com>
Date: Wed, 23 Mar 2022 07:27:05 +0000
Subject: [PATCH] auditswig.i: avoid setter generation for audit_rule_data::buf

As it's a flexible array generated code was never safe to use.
With kernel's https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=ed98ea2128b6fd83bce13716edf8f5fe6c47f574
change it's a build failure now:

    audit> audit_wrap.c:5010:15: error: invalid use of flexible array member
    audit>  5010 |     arg1->buf = (char [])(char *)memcpy(malloc((size)*sizeof(char)), (const char *)(arg2), sizeof(char)*(size));
    audit>       |               ^

Let's avoid setter generation entirely.

Closes: https://github.com/linux-audit/audit-userspace/issues/252
--- a/bindings/swig/src/auditswig.i
+++ b/bindings/swig/src/auditswig.i
@@ -39,6 +39,10 @@ signed
 #define __attribute(X) /*nothing*/
 typedef unsigned __u32;
 typedef unsigned uid_t;
+/* Sidestep SWIG's limitation of handling c99 Flexible arrays by not:
+ * generating setters against them: https://github.com/swig/swig/issues/1699
+ */
+%ignore audit_rule_data::buf;
 %include "/usr/include/linux/audit.h"
 #define __extension__ /*nothing*/
 %include <stdint.i>