[PATCH v4 0/2] x86/tdx: Use TDVMCALLs directly for earlyprintk

[PATCH v4 0/2] x86/tdx: Use TDVMCALLs directly for earlyprintk

[Vishal Verma]

Background
==========

TDX guests have early_printk support, but it is through a roundabout way
using a #VE exception as a functional mechanism. The exception handler
has to use TDG.VP.VEINFO.GET to find out what faulted, and then issue
the TDVMCALL. Simplify this by using a TDVMCALL from the guest which can
directly perform the port I/O.

Patch details
=============

Patch 1 is a cleanup that moves the port I/O direction constants to
arch/x86/include/asm/shared/tdx.h.

Patch 2: Add a new earlyprintk=tdx option, which calls the new
tdx_serial_in() / tdx_serial_out() helpers to perform the port I/O
directly using TDVMCALLs.

The serial earlyprintk path is left as is - TD guests using this will
still take a #VE exception, and the exception handler will perform the
I/O.

Testing
=======

Added a counter for the number of COM1 port I/O #VEs handled:

  no earlyprintk     -  ~63000
  earlyprintk=ttyS0  -  ~61000
  earlyprintk=tdx    -  ~35000

The new tdx option nearly halves the number of #VEs handled.

Additionally, ran the usual suite of CI testing, which covers KVM self
tests, KVM unit tests, and Avocado KVM tests without any regressions.

Signed-off-by: Vishal Verma <vishal.l.verma@intel.com>
---
Changes in v4:
- Rename the tdx_inb()/tdx_outb() helpers directly to the serial
  accessors with __noendbr annotations, and remove the config ifdefs
  from early_printk.c. (Kiryl)
- Rebase to v7.3-rc6
- Link to v3: https://patch.msgid.link/20260922-b4-tdx_earlyprintk_tdcalls-v3-0-6ebaa599df10@intel.com

Changes in v3:
- Drop the ttySN and baudrate option parsing for earlyprintk=tdx (Dave Hansen)
- Link to v2: https://patch.msgid.link/20260921-b4-tdx_earlyprintk_tdcalls-v2-0-ad1fde3f25fd@intel.com

Changes in v2:
- Rebase to v7.3-rc4
- Move the TDX hypercall helpers into coco/tdx/tdx.c (Kiryl)
- Add a new earlyprintk=tdx option that implements the direct TDVMCALL
  path (Dave Hansen)
- Link to v1: https://patch.msgid.link/20260910-b4-tdx_earlyprintk_tdcalls-v1-0-4b2b1bf9001b@intel.com

---
Vishal Verma (2):
      x86/tdx: Move port I/O definitions to a shared header
      x86/early_printk: Add earlyprintk=tdx to drive the UART with TDVMCALLs

 Documentation/admin-guide/kernel-parameters.txt |  8 ++++++
 arch/x86/include/asm/shared/tdx.h               |  6 ++++
 arch/x86/include/asm/tdx.h                      |  4 +++
 arch/x86/boot/compressed/tdx.c                  |  4 +--
 arch/x86/coco/tdx/tdx.c                         | 37 +++++++++++++++++++++----
 arch/x86/kernel/early_printk.c                  | 20 +++++++++++++
 6 files changed, 71 insertions(+), 8 deletions(-)
---
base-commit: a90ee4305c4a5df72c11b31dacfdc76e00fcf78a
change-id: 20260903-b4-tdx_earlyprintk_tdcalls-40d1b21a8ba2

Best regards,
--  
Vishal Verma <vishal.l.verma@intel.com>

[PATCH v4 1/2] x86/tdx: Move port I/O definitions to a shared header

[Vishal Verma]

In preparation for adding a 'tdx' option for early_printk which uses
direct TDVMCALLs for I/O, factor out the PORT_READ and PORT_WRITE
definitions from arch/x86/coco/tdx/tdx.c into
arch/x86/include/asm/shared/tdx.h, and namespace them with 'TDVMCALL_'.

The decompressor also used 0/1 literals in its tdx_io_{in,out} helpers -
switch those to use the new definitions.

Signed-off-by: Vishal Verma <vishal.l.verma@intel.com>
Reviewed-by: Kiryl Shutsemau (Meta) <kas@kernel.org>
Cc: Rick Edgecombe <rick.p.edgecombe@intel.com>
---
 arch/x86/include/asm/shared/tdx.h | 6 ++++++
 arch/x86/boot/compressed/tdx.c    | 4 ++--
 arch/x86/coco/tdx/tdx.c           | 8 ++------
 3 files changed, 10 insertions(+), 8 deletions(-)

diff --git a/arch/x86/include/asm/shared/tdx.h b/arch/x86/include/asm/shared/tdx.h
index f20e91d7ac35..ecaf965f059b 100644
--- a/arch/x86/include/asm/shared/tdx.h
+++ b/arch/x86/include/asm/shared/tdx.h
@@ -84,6 +84,12 @@
 #define TDVMCALL_STATUS_ALIGN_ERROR	0x8000000000000002ULL
 #define TDVMCALL_STATUS_SUBFUNC_UNSUPPORTED	0x8000000000000003ULL
 
+/*
+ * TDG.VP.VMCALL<Instruction.IO> direction (passed in R13)
+ */
+#define TDVMCALL_PORT_READ		0
+#define TDVMCALL_PORT_WRITE		1
+
 /*
  * Bitmasks of exposed registers (with VMM).
  */
diff --git a/arch/x86/boot/compressed/tdx.c b/arch/x86/boot/compressed/tdx.c
index 8451d6a1030c..ed278bac3c93 100644
--- a/arch/x86/boot/compressed/tdx.c
+++ b/arch/x86/boot/compressed/tdx.c
@@ -22,7 +22,7 @@ static inline unsigned int tdx_io_in(int size, u16 port)
 		.r10 = TDX_HYPERCALL_STANDARD,
 		.r11 = hcall_func(EXIT_REASON_IO_INSTRUCTION),
 		.r12 = size,
-		.r13 = 0,
+		.r13 = TDVMCALL_PORT_READ,
 		.r14 = port,
 	};
 
@@ -38,7 +38,7 @@ static inline void tdx_io_out(int size, u16 port, u32 value)
 		.r10 = TDX_HYPERCALL_STANDARD,
 		.r11 = hcall_func(EXIT_REASON_IO_INSTRUCTION),
 		.r12 = size,
-		.r13 = 1,
+		.r13 = TDVMCALL_PORT_WRITE,
 		.r14 = port,
 		.r15 = value,
 	};
diff --git a/arch/x86/coco/tdx/tdx.c b/arch/x86/coco/tdx/tdx.c
index f904a636d449..7d1a93ee2534 100644
--- a/arch/x86/coco/tdx/tdx.c
+++ b/arch/x86/coco/tdx/tdx.c
@@ -24,10 +24,6 @@
 #define EPT_READ	0
 #define EPT_WRITE	1
 
-/* Port I/O direction */
-#define PORT_READ	0
-#define PORT_WRITE	1
-
 /* See Exit Qualification for I/O Instructions in VMX documentation */
 #define VE_IS_IO_IN(e)		((e) & BIT(3))
 #define VE_GET_IO_SIZE(e)	(((e) & GENMASK(2, 0)) + 1)
@@ -691,7 +687,7 @@ static bool handle_in(struct pt_regs *regs, int size, int port)
 		.r10 = TDX_HYPERCALL_STANDARD,
 		.r11 = hcall_func(EXIT_REASON_IO_INSTRUCTION),
 		.r12 = size,
-		.r13 = PORT_READ,
+		.r13 = TDVMCALL_PORT_READ,
 		.r14 = port,
 	};
 	bool success;
@@ -720,7 +716,7 @@ static bool handle_out(struct pt_regs *regs, int size, int port)
 	 * "TDG.VP.VMCALL<Instruction.IO>".
 	 */
 	return !_tdx_hypercall(hcall_func(EXIT_REASON_IO_INSTRUCTION), size,
-			       PORT_WRITE, port, regs->ax & mask);
+			       TDVMCALL_PORT_WRITE, port, regs->ax & mask);
 }
 
 /*

-- 
2.55.0

[PATCH v4 2/2] x86/early_printk: Add earlyprintk=tdx to drive the UART with TDVMCALLs

[Vishal Verma]

A TDX guest cannot execute port I/O instructions directly, but
earlyprintk's serial console still issues plain inb()/outb() and lets
each one fault into the #VE handler to be emulated as a TDVMCALL.

While that works, it is a roundabout way to get a character out.
early_serial_putc() polls the LSR, and then writes a byte, but since the
TDX guest can't directly do port I/O, a #VE exception is raised. The #VE
handler must call TDG.VP.VEINFO.GET to find out what faulted, and then
it can issue the TDVMCALL that does the actual work.

This makes #VE a functional mechanism for doing I/O, which is not
desirable, is unnecessarily complicated and fragile, and results in
twice the number of calls into the TDX module.

Instead, add an earlyprintk=tdx option, which can issue the TDVMCALLs
directly. Add a pair of tdx_serial_in() and tdx_serial_out() accessors
and set them up in this case.

Note that the output does not appear any earlier - "earlyprintk=" is an
early_param(), so the console is still registered from
parse_early_param(). This only changes how the bytes leave the guest
once it is up.

LLMs were used under supervision to create this patch, to help
understand the scope and mechanisms, create testing instrumentation
(throwaway) to count #VEs in the serial vs tdx earlyprintk setups,
and to drive lab machines to do this testing.

Signed-off-by: Vishal Verma <vishal.l.verma@intel.com>
---
 Documentation/admin-guide/kernel-parameters.txt |  8 +++++++
 arch/x86/include/asm/tdx.h                      |  4 ++++
 arch/x86/coco/tdx/tdx.c                         | 29 +++++++++++++++++++++++++
 arch/x86/kernel/early_printk.c                  | 20 +++++++++++++++++
 4 files changed, 61 insertions(+)

diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt
index e75344f4e0cd..3e43e7c7385c 100644
--- a/Documentation/admin-guide/kernel-parameters.txt
+++ b/Documentation/admin-guide/kernel-parameters.txt
@@ -1546,6 +1546,7 @@ Kernel parameters
 			earlyprintk=serial[,ttySn[,baudrate]]
 			earlyprintk=serial[,0x...[,baudrate]]
 			earlyprintk=ttySn[,baudrate]
+			earlyprintk=tdx
 			earlyprintk=dbgp[debugController#]
 			earlyprintk=mmio32,membase[,{nocfg|baudrate}]
 			earlyprintk=pciserial[,force],bus:device.function[,{nocfg|baudrate}]
@@ -1559,6 +1560,13 @@ Kernel parameters
 			Use "nocfg" to skip UART configuration, assume
 			BIOS/firmware has configured UART correctly.
 
+			On x86, "tdx" is a fixed COM1 console, and takes no
+			options. In a TDX guest the port is driven with
+			TDVMCALLs rather than port I/O instructions, avoiding
+			the #VE exception that would otherwise be taken to
+			emulate every access. Outside a TDX guest it falls back
+			to port I/O, behaving like "earlyprintk=serial".
+
 			Append ",keep" to not disable it when the real console
 			takes over.
 
diff --git a/arch/x86/include/asm/tdx.h b/arch/x86/include/asm/tdx.h
index 89e97d5761d8..eac298979498 100644
--- a/arch/x86/include/asm/tdx.h
+++ b/arch/x86/include/asm/tdx.h
@@ -95,6 +95,10 @@ static inline bool tdx_early_handle_ve(struct pt_regs *regs) { return false; }
 
 #endif /* CONFIG_INTEL_TDX_GUEST */
 
+/* Only referenced when X86_FEATURE_TDX_GUEST is enabled, no stubs needed */
+__noendbr unsigned int tdx_serial_in(unsigned long addr, int offset);
+__noendbr void tdx_serial_out(unsigned long addr, int offset, int value);
+
 #if defined(CONFIG_KVM_GUEST) && defined(CONFIG_INTEL_TDX_GUEST)
 long tdx_kvm_hypercall(unsigned int nr, unsigned long p1, unsigned long p2,
 		       unsigned long p3, unsigned long p4);
diff --git a/arch/x86/coco/tdx/tdx.c b/arch/x86/coco/tdx/tdx.c
index 7d1a93ee2534..c5e16ea3844d 100644
--- a/arch/x86/coco/tdx/tdx.c
+++ b/arch/x86/coco/tdx/tdx.c
@@ -194,6 +194,35 @@ u64 tdx_hcall_get_quote(u8 *buf, size_t size)
 }
 EXPORT_SYMBOL_GPL(tdx_hcall_get_quote);
 
+/*
+ * early_printk accessors for a TDX guest. Ask the VMM to do the port I/O with
+ * TDG.VP.VMCALL<Instruction.IO> instead of executing IN/OUT and having the
+ * resulting #VE emulate it.
+ */
+__noendbr unsigned int tdx_serial_in(unsigned long addr, int offset)
+{
+	struct tdx_module_args args = {
+		.r10 = TDX_HYPERCALL_STANDARD,
+		.r11 = hcall_func(EXIT_REASON_IO_INSTRUCTION),
+		.r12 = 1,
+		.r13 = TDVMCALL_PORT_READ,
+		.r14 = addr + offset,
+	};
+
+	if (__tdx_hypercall(&args))
+		return 0xFF;
+
+	return args.r11;
+}
+ANNOTATE_NOENDBR_SYM(tdx_serial_in);
+
+__noendbr void tdx_serial_out(unsigned long addr, int offset, int value)
+{
+	_tdx_hypercall(hcall_func(EXIT_REASON_IO_INSTRUCTION), 1,
+		       TDVMCALL_PORT_WRITE, addr + offset, value);
+}
+ANNOTATE_NOENDBR_SYM(tdx_serial_out);
+
 static void __noreturn tdx_panic(const char *msg)
 {
 	struct tdx_module_args args = {
diff --git a/arch/x86/kernel/early_printk.c b/arch/x86/kernel/early_printk.c
index cba75306e5b6..04dadcd73b85 100644
--- a/arch/x86/kernel/early_printk.c
+++ b/arch/x86/kernel/early_printk.c
@@ -21,6 +21,7 @@
 #include <linux/usb/xhci-dbgp.h>
 #include <asm/pci_x86.h>
 #include <linux/static_call.h>
+#include <asm/tdx.h>
 
 /* Simple VGA output */
 #define VGABASE		(__ISA_IO_base + 0xb8000)
@@ -196,6 +197,20 @@ static __init void early_serial_init(char *s)
 	early_serial_hw_init(divisor);
 }
 
+/*
+ * Use early_serial_base (COM1) and same default baud rate as
+ * "earlyprintk=serial"
+ */
+static __init void early_serial_tdx_init(void)
+{
+	if (cpu_feature_enabled(X86_FEATURE_TDX_GUEST)) {
+		static_call_update(serial_in, tdx_serial_in);
+		static_call_update(serial_out, tdx_serial_out);
+	}
+
+	early_serial_hw_init(115200 / DEFAULT_BAUD);
+}
+
 static __noendbr void mem32_serial_out(unsigned long addr, int offset, int value)
 {
 	u32 __iomem *vaddr = (u32 __iomem *)addr;
@@ -414,6 +429,11 @@ static int __init setup_early_printk(char *buf)
 			early_serial_init(buf + 4);
 			early_console_register(&early_serial_console, keep);
 		}
+		if (!strncmp(buf, "tdx", 3)) {
+			early_serial_tdx_init();
+			early_console_register(&early_serial_console, keep);
+			break;
+		}
 #ifdef CONFIG_PCI
 		if (!strncmp(buf, "pciserial", 9)) {
 			buf += 9; /* Keep from match the above "pciserial" */

-- 
2.55.0

Re: [PATCH v4 2/2] x86/early_printk: Add earlyprintk=tdx to drive the UART with TDVMCALLs

[Kiryl Shutsemau]

On Mon, Oct 05, 2026 at 04:50:07PM -0600, Vishal Verma wrote:
> A TDX guest cannot execute port I/O instructions directly, but
> earlyprintk's serial console still issues plain inb()/outb() and lets
> each one fault into the #VE handler to be emulated as a TDVMCALL.
> 
> While that works, it is a roundabout way to get a character out.
> early_serial_putc() polls the LSR, and then writes a byte, but since the
> TDX guest can't directly do port I/O, a #VE exception is raised. The #VE
> handler must call TDG.VP.VEINFO.GET to find out what faulted, and then
> it can issue the TDVMCALL that does the actual work.
> 
> This makes #VE a functional mechanism for doing I/O, which is not
> desirable, is unnecessarily complicated and fragile, and results in
> twice the number of calls into the TDX module.
> 
> Instead, add an earlyprintk=tdx option, which can issue the TDVMCALLs
> directly. Add a pair of tdx_serial_in() and tdx_serial_out() accessors
> and set them up in this case.
> 
> Note that the output does not appear any earlier - "earlyprintk=" is an
> early_param(), so the console is still registered from
> parse_early_param(). This only changes how the bytes leave the guest
> once it is up.
> 
> LLMs were used under supervision to create this patch, to help
> understand the scope and mechanisms, create testing instrumentation
> (throwaway) to count #VEs in the serial vs tdx earlyprintk setups,
> and to drive lab machines to do this testing.

Sounds like you are missing "Assisted-by: LLM" tag.

> Signed-off-by: Vishal Verma <vishal.l.verma@intel.com>

Reviewed-by: Kiryl Shutsemau (Meta) <kas@kernel.org>

-- 
  Kiryl Shutsemau / Kirill A. Shutemov

Re: [PATCH v4 1/2] x86/tdx: Move port I/O definitions to a shared header

[Edgecombe, Rick P]

On Mon, 2026-10-05 at 16:50 -0600, Vishal Verma wrote:
> In preparation for adding a 'tdx' option for early_printk which uses
> direct TDVMCALLs for I/O, factor out the PORT_READ and PORT_WRITE
> definitions from arch/x86/coco/tdx/tdx.c into
> arch/x86/include/asm/shared/tdx.h, and namespace them with 'TDVMCALL_'.
> 
> The decompressor also used 0/1 literals in its tdx_io_{in,out} helpers -
> switch those to use the new definitions.
> 
> Signed-off-by: Vishal Verma <vishal.l.verma@intel.com>
> Reviewed-by: Kiryl Shutsemau (Meta) <kas@kernel.org>
> Cc: Rick Edgecombe <rick.p.edgecombe@intel.com>

Reviewed-by: Rick Edgecombe <rick.p.edgecome@intel.com>

Re: [PATCH v4 2/2] x86/early_printk: Add earlyprintk=tdx to drive the UART with TDVMCALLs

[Edgecombe, Rick P]

On Mon, 2026-10-05 at 16:50 -0600, Vishal Verma wrote:
> A TDX guest cannot execute port I/O instructions directly, but
> earlyprintk's serial console still issues plain inb()/outb() and lets
> each one fault into the #VE handler to be emulated as a TDVMCALL.
> 
> While that works, it is a roundabout way to get a character out.
> early_serial_putc() polls the LSR, and then writes a byte, but since the
> TDX guest can't directly do port I/O, a #VE exception is raised. The #VE
> handler must call TDG.VP.VEINFO.GET to find out what faulted, and then
> it can issue the TDVMCALL that does the actual work.
> 
> This makes #VE a functional mechanism for doing I/O, which is not
> desirable, is unnecessarily complicated and fragile, and results in
> twice the number of calls into the TDX module.
> 
> Instead, add an earlyprintk=tdx option, which can issue the TDVMCALLs
> directly. Add a pair of tdx_serial_in() and tdx_serial_out() accessors
> and set them up in this case.
> 
> Note that the output does not appear any earlier - "earlyprintk=" is an
> early_param(), so the console is still registered from
> parse_early_param(). This only changes how the bytes leave the guest
> once it is up.
> 
> LLMs were used under supervision to create this patch, to help
> understand the scope and mechanisms, create testing instrumentation
> (throwaway) to count #VEs in the serial vs tdx earlyprintk setups,
> and to drive lab machines to do this testing.
> 
> Signed-off-by: Vishal Verma <vishal.l.verma@intel.com>

Reviewed-by: Rick Edgecombe <rick.p.edgecombe@intel.com>