[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[PATCH RFC v2 14/14] XSM: avoid fragile assumptions in xsm_fixup_ops()


  • To: "xen-devel@xxxxxxxxxxxxxxxxxxxx" <xen-devel@xxxxxxxxxxxxxxxxxxxx>
  • From: Jan Beulich <jbeulich@xxxxxxxx>
  • Date: Mon, 17 Aug 2026 10:58:47 +0200
  • Authentication-results: eu.smtp.expurgate.cloud; dkim=pass header.s=google header.d=suse.com header.i="@suse.com" header.h="Content-Transfer-Encoding:Content-Type:In-Reply-To:Autocrypt:Content-Language:References:Cc:To:From:Subject:User-Agent:MIME-Version:Date:Message-ID"
  • Autocrypt: addr=jbeulich@xxxxxxxx; keydata= xsDiBFk3nEQRBADAEaSw6zC/EJkiwGPXbWtPxl2xCdSoeepS07jW8UgcHNurfHvUzogEq5xk hu507c3BarVjyWCJOylMNR98Yd8VqD9UfmX0Hb8/BrA+Hl6/DB/eqGptrf4BSRwcZQM32aZK 7Pj2XbGWIUrZrd70x1eAP9QE3P79Y2oLrsCgbZJfEwCgvz9JjGmQqQkRiTVzlZVCJYcyGGsD /0tbFCzD2h20ahe8rC1gbb3K3qk+LpBtvjBu1RY9drYk0NymiGbJWZgab6t1jM7sk2vuf0Py O9Hf9XBmK0uE9IgMaiCpc32XV9oASz6UJebwkX+zF2jG5I1BfnO9g7KlotcA/v5ClMjgo6Gl MDY4HxoSRu3i1cqqSDtVlt+AOVBJBACrZcnHAUSuCXBPy0jOlBhxPqRWv6ND4c9PH1xjQ3NP nxJuMBS8rnNg22uyfAgmBKNLpLgAGVRMZGaGoJObGf72s6TeIqKJo/LtggAS9qAUiuKVnygo 3wjfkS9A3DRO+SpU7JqWdsveeIQyeyEJ/8PTowmSQLakF+3fote9ybzd880fSmFuIEJldWxp Y2ggPGpiZXVsaWNoQHN1c2UuY29tPsJgBBMRAgAgBQJZN5xEAhsDBgsJCAcDAgQVAggDBBYC AwECHgECF4AACgkQoDSui/t3IH4J+wCfQ5jHdEjCRHj23O/5ttg9r9OIruwAn3103WUITZee e7Sbg12UgcQ5lv7SzsFNBFk3nEQQCACCuTjCjFOUdi5Nm244F+78kLghRcin/awv+IrTcIWF hUpSs1Y91iQQ7KItirz5uwCPlwejSJDQJLIS+QtJHaXDXeV6NI0Uef1hP20+y8qydDiVkv6l IreXjTb7DvksRgJNvCkWtYnlS3mYvQ9NzS9PhyALWbXnH6sIJd2O9lKS1Mrfq+y0IXCP10eS FFGg+Av3IQeFatkJAyju0PPthyTqxSI4lZYuJVPknzgaeuJv/2NccrPvmeDg6Coe7ZIeQ8Yj t0ARxu2xytAkkLCel1Lz1WLmwLstV30g80nkgZf/wr+/BXJW/oIvRlonUkxv+IbBM3dX2OV8 AmRv1ySWPTP7AAMFB/9PQK/VtlNUJvg8GXj9ootzrteGfVZVVT4XBJkfwBcpC/XcPzldjv+3 HYudvpdNK3lLujXeA5fLOH+Z/G9WBc5pFVSMocI71I8bT8lIAzreg0WvkWg5V2WZsUMlnDL9 mpwIGFhlbM3gfDMs7MPMu8YQRFVdUvtSpaAs8OFfGQ0ia3LGZcjA6Ik2+xcqscEJzNH+qh8V m5jjp28yZgaqTaRbg3M/+MTbMpicpZuqF4rnB0AQD12/3BNWDR6bmh+EkYSMcEIpQmBM51qM EKYTQGybRCjpnKHGOxG0rfFY1085mBDZCH5Kx0cl0HVJuQKC+dV2ZY5AqjcKwAxpE75MLFkr wkkEGBECAAkFAlk3nEQCGwwACgkQoDSui/t3IH7nnwCfcJWUDUFKdCsBH/E5d+0ZnMQi+G0A nAuWpQkjM1ASeQwSHEeAWPgskBQL
  • Cc: Daniel Smith <dpsmith@xxxxxxxxxxxxxxxxxxxx>, Andrew Cooper <andrew.cooper3@xxxxxxxxxx>
  • Delivery-date: Mon, 17 Aug 2026 08:58:49 +0000
  • List-id: Xen developer discussion <xen-devel.lists.xenproject.org>

Using the recently introduced hook machinery, the assumptions made can be
avoided, at the expense of a significant code size increase.

Signed-off-by: Jan Beulich <jbeulich@xxxxxxxx>
---
RFC: Of course the significantly increased code size isn't nice. Yet then
     it's all .init.text bloat only.

RFC: The asm() is somewhat like RELOC_HIDE(), with an offset of 0. It
     didn't feel quite right to use that macro here, though.

The generated code I've looked at (both for x86 and Arm64) it is clear
that a similar weakness in codegen exists for the recurring accesses to
_{s,e}text[] by is_kernel_text(): A register variable each would shrink
code size significantly, yet that doesn't appear to happen at -O1. Yet
open-coding isn't an option, imo.
---
v2: New.

--- a/xen/include/xsm/xsm.h
+++ b/xen/include/xsm/xsm.h
@@ -51,15 +51,6 @@ typedef enum xsm_default xsm_default_t;
 #define XSM_MMU_MACHPHYS_UPDATE  8
 #endif /* CONFIG_X86 */
 
-/*
- * !!! WARNING !!!
- *
- * For simplicity, xsm_fixup_ops() expects that this structure is made
- * exclusively of function pointers to non-init functions.  Think carefully
- * before deviating from the pattern.
- *
- * !!! WARNING !!!
- */
 struct xsm_ops {
 
 #define XSM_HOOK0(rtype, name) rtype (*name)(void);
--- a/xen/xsm/dummy.c
+++ b/xen/xsm/dummy.c
@@ -28,29 +28,35 @@ static const struct xsm_ops __initconst_
 
 void __init xsm_fixup_ops(struct xsm_ops *ops)
 {
+    const struct xsm_ops *dops = &dummy_ops;
+
     /*
-     * We make some simplifying assumptions about struct xsm_ops; that it is
-     * made exclusively of function pointers to non-init text.
-     *
-     * This allows us to walk over struct xsm_ops as if it were an array of
-     * unsigned longs.
+     * To limit the size of generated code (by avoiding the compiler using
+     * dummy_ops directly for every access), hide the relationship between
+     * dops and dummy_ops.
      */
-    unsigned long *dst = _p(ops);
-    const unsigned long *src = _p(&dummy_ops);
+    asm ( "" : "+r" (dops) );
 
-    for ( ; dst < (unsigned long *)(ops + 1); src++, dst++ )
-    {
-        /*
-         * If you encounter this BUG(), then you've most likely added a new
-         * XSM hook but failed to provide the default implementation in
-         * dummy_ops.
-         *
-         * If not, then perhaps a function pointer to an init function, or
-         * something which isn't a function pointer at all.
-         */
-        BUG_ON(!is_kernel_text(*src));
+    /*
+     * If you encounter the BUG() below, then you've most likely added a
+     * new XSM hook but failed to provide the default implementation in
+     * dummy_ops.
+     *
+     * If not, then perhaps a function pointer to an init function, or
+     * (less likely) something which isn't a function pointer at all.
+     */
+#define XSM_HOOK0(rtype, name)                    \
+    do {                                          \
+        typeof(xsm_ ## name) *dummy = dops->name; \
+        BUG_ON(!is_kernel_text(dummy));           \
+        if ( !ops->name )                         \
+            ops->name = dummy;                    \
+    } while ( false );
+#define XSM_HOOK1(rtype, name, ...) XSM_HOOK0(rtype, name)
+#define XSM_HOOK2(rtype, name, ...) XSM_HOOK0(rtype, name)
+#define XSM_HOOK3(rtype, name, ...) XSM_HOOK0(rtype, name)
+#define XSM_HOOK4(rtype, name, ...) XSM_HOOK0(rtype, name)
+#define XSM_HOOK5(rtype, name, ...) XSM_HOOK0(rtype, name)
 
-        if ( !*dst )
-            *dst = *src;
-    }
+#include <xsm/hooks.h>
 }




 


Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.