Skip to contentSkip to conversation

x86/virt/tdx: Enforce no indirect calls of TDX assembly

PATCH In review Kai Huang 2 participants 3 messages Jun 6, 2025 – Jun 9, 2025

Subsystems (2)
lore
Share & tools
lore

Kai Huang

author
#0 lore raw

PATCH v3 1/1 x86/virt/tdx: Enforce no indirect calls of TDX assembly

Two static inline TDX helper functions (sc_retry() and sc_retry_prerr()) pass around function pointers to assembly functions. Normally, the compiler inlines these helper functions, realizes that the function pointer targets are completely static and can be resolved at compile time, and generates direct call instructions.

But, other times (like when CONFIG_CC_OPTIMIZE_FOR_SIZE=y), the compiler will stop inlining and instead generate indirect call instructions.

Indirect calls to assembly functions require special annotation so that both hardware and software implementations of Control Flow Integrity mechanisms can work correctly. But TDX assembly functions are declared as if they are only called directly.

Annotate both sc_retry() and sc_retry_prerr() as __always_inline so that the compiler will always inline these helpers and generate direct call instructions when calling TDX assembly functions (see [1]).

This was found through randconfig testing, presumably setting CONFIG_CC_OPTIMIZE_FOR_SIZE=1 when objtool spewed a bunch of these:

vmlinux.o: warning: objtool: tdh_mem_range_block+0x7e: relocation to !ENDBR: __seamcall_ret+0x0

---

This is updated version (with patch subject updated) of v1 patch:

https://lore.kernel.org/lkml/[email protected]/

v1 -> v3:
- Follow Peter's suggestion to use __always_inline, instead of declaring TDX assembly can be called indirectly.
- Change patch subject accordingly.
- Update changelog accordingly, based on Dave's version.

v2 is similar to v1 and was sent right before Peter suggested to use __always_inline, thus feel free to ignore.

---
 arch/x86/include/asm/tdx.h  | 2 +-
 arch/x86/virt/vmx/tdx/tdx.c | 5 +++--
 2 files changed, 4 insertions(+), 3 deletions(-)
arch/x86/include/asm/tdx.h +1 -1
index 8b19294600c4..7ddef3a69866 100644
@@ -106,7 +106,7 @@ void tdx_init(void);
 
 typedef u64 (*sc_func_t)(u64 fn, struct tdx_module_args *args);
 
-static inline u64 sc_retry(sc_func_t func, u64 fn,
+static __always_inline u64 sc_retry(sc_func_t func, u64 fn,
 			   struct tdx_module_args *args)
 {
 	int retry = RDRAND_RETRY_LOOPS;
arch/x86/virt/vmx/tdx/tdx.c +3 -2
index 2457d13c3f9e..c7a9a087ccaf 100644
@@ -75,8 +75,9 @@ static inline void seamcall_err_ret(u64 fn, u64 err,
 			args->r9, args->r10, args->r11);
 }
 
-static inline int sc_retry_prerr(sc_func_t func, sc_err_func_t err_func,
-				 u64 fn, struct tdx_module_args *args)
+static __always_inline int sc_retry_prerr(sc_func_t func,
+					  sc_err_func_t err_func,
+					  u64 fn, struct tdx_module_args *args)
 {
 	u64 sret = sc_retry(func, fn, args);
 

base-commit: ec7714e4947909190ffb3041a03311a975350fe0

Signature · 1 line
2.49.0
#1 lore raw

It doesn't really "enforce" anything. But, oh well, I'll just fix it up when I apply it early next week. Here's what I'll probably apply:

x86/virt/tdx: Avoid indirect calls to TDX assembly functions

Two 'static inline' TDX helper functions (sc_retry() and sc_retry_prerr()) take function pointer arguments which refer to assembly functions. Normally, the compiler inlines the TDX helper, realizes that the function pointer targets are completely static -- thus can be resolved at compile time -- and generates direct call instructions.

But, other times (like when CONFIG_CC_OPTIMIZE_FOR_SIZE=y), the compiler declines to inline the helpers and will instead generate indirect call instructions.

Indirect calls to assembly functions require special annotation (for various Control Flow Integrity mechanisms). But TDX assembly functions lack the special annotations and can only be called directly.

Annotate both the helpers as '__always_inline' to prod the compiler into maintaining the direct calls. There is no guarantee here, but Peter has volunteered to report the compiler bug if this assumption ever breaks[1].

...

Quoted 7 lines

This was found through randconfig testing, presumably setting CONFIG_CC_OPTIMIZE_FOR_SIZE=1 when objtool spewed a bunch of these:

vmlinux.o: warning: objtool: tdh_mem_range_block+0x7e: relocation to !ENDBR: __seamcall_ret+0x0

author
#2 lore raw
Quoted 2 lines · Dave Hansen

It doesn't really "enforce" anything. But, oh well, I'll just fix it up when I apply it early next week. Here's what I'll probably apply:

Thanks!

Quoted 17 lines

x86/virt/tdx: Avoid indirect calls to TDX assembly functions

Two 'static inline' TDX helper functions (sc_retry() and sc_retry_prerr()) take function pointer arguments which refer to assembly functions. Normally, the compiler inlines the TDX helper, realizes that the function pointer targets are completely static -- thus can be resolved at compile time -- and generates direct call instructions.

But, other times (like when CONFIG_CC_OPTIMIZE_FOR_SIZE=y), the compiler declines to inline the helpers and will instead generate indirect call instructions.

Indirect calls to assembly functions require special annotation (for various Control Flow Integrity mechanisms). But TDX assembly functions lack the special annotations and can only be called directly.

Quoted 13 lines

Annotate both the helpers as '__always_inline' to prod the compiler into maintaining the direct calls. There is no guarantee here, but Peter has volunteered to report the compiler bug if this assumption ever breaks[1].

...

Quoted 7 lines

This was found through randconfig testing, presumably setting CONFIG_CC_OPTIMIZE_FOR_SIZE=1 when objtool spewed a bunch of these:

vmlinux.o: warning: objtool: tdh_mem_range_block+0x7e: relocation to !ENDBR: __seamcall_ret+0x0

And sorry that I somehow missed the Fixes tag here.

Since sc_retry() and sc_retry_prerr() were introduced in two (contiguous) commits, perhaps we need to add two Fixes tags:

Fixes 1e66a7e27539 ("x86/virt/tdx: Handle SEAMCALL no entropy error in

common code")

Fixes df01f5ae07dd ("x86/virt/tdx: Add SEAMCALL error printing for module

initialization")

Please let me know if you need anything more from me.