[PATCH 0/2] x86/fred: Reconstruct fault state for rejected INT instructions

[PATCH 0/2] x86/fred: Reconstruct fault state for rejected INT instructions

[Matthew Schwartz]

This series reconstructs the #GP signal context for rejected INT
instructions with FRED enabled. The incorrect error code and saved IP
break Wine's handling of INT 0x2d, causing Elden Ring to fail with a
spurious access violation on Panther Lake.

The software event flag handling follows the same reasoning as the
existing sigreturn fix in prevent_single_step_upon_eretu(). The second
patch adds 32-bit and 64-bit signal-context and ptrace resume coverage.

Both selftest variants pass all 29 checks on patched FRED-enabled
Panther Lake and on a non-FRED AMD host. With the same binaries on
unpatched Panther Lake, 16 signal-context checks and the first ptrace IP
check fail. The two dependent ptrace resume checks are not reached.

Separate continuation testing with the fix matched an IDT baseline on
another system in both bitnesses: unchanged IP retries the INT whether
RF is set or clear, and advancing IP resumes at the next instruction.

One observed ptrace divergence remains separate: after rewriting IP at
the stop from raise(SIGSTOP), PTRACE_SINGLESTEP traps before the target
instruction on FRED, whereas IDT executes it. The selftest starts from
an INT3 stop to isolate rejected-INT behavior. The syscall-stop case is
a follow-up investigation outside this series.

Matthew Schwartz (2):
  x86/fred: Reconstruct the #GP context for rejected INT instructions
  selftests/x86: Check signal state for rejected software interrupts

 arch/x86/entry/entry_fred.c              |  11 +-
 tools/testing/selftests/x86/Makefile     |   2 +-
 tools/testing/selftests/x86/int_signal.c | 311 +++++++++++++++++++++++
 3 files changed, 322 insertions(+), 2 deletions(-)
 create mode 100644 tools/testing/selftests/x86/int_signal.c


base-commit: b5a051f6b840d48f159166ef073d3021989bfb50
-- 
2.55.0

[PATCH 1/2] x86/fred: Reconstruct the #GP context for rejected INT instructions

[Matthew Schwartz]

FRED event delivery does not use the IDT, so the gate DPL check that
rejects a user INT n falls to software (Intel FRED specification [1],
section 8.3). fred_intx() rejects the same vectors as IDT delivery, but
reports a zero error code and the IP after the INT. This breaks the
signal ABI. Wine uses the error code to recognize INT 0x2d, so the
changed context turns a handled breakpoint into an access violation in
Elden Ring.

Rewind IP using the instruction length in the augmented SS and
synthesize the IDT selector error code, (vector << 3) | 2. Set RF in the
saved flags, as the CPU does for a #GP fault. Section 5.2.1 defines the
saved vector, instruction length and RF state. The supplied length
handles prefixes without reading user memory. Limit the changes to
already-rejected software interrupts, preserving the accepted INT3, INT4
and enabled INT80 paths and hardware exceptions. With IA32 emulation
disabled, INT 0x80 now reports the same #GP as the DPL 0 gate IDT
installs there. The rewound IP also stops fixup_iopl_exception() from
inspecting the byte after the INT.

Also clear the software event flag. Section 6.2.3 specifies that ERETU
with this flag and TF set traps before executing any user instruction. A
tracer that suppresses SIGSEGV and resumes with TF set expects the next
instruction to run first, as after IRET. The sigreturn path clears the
same flag for this reason in prevent_single_step_upon_eretu().

[1] Intel Flexible Return and Event Delivery (FRED) Specification,
revision 9.0 (346446-009US), sections 5.2.1, 6.2.3 and 8.3.

Fixes: 14619d912b65 ("x86/fred: FRED entry/exit and dispatch code")
Reported-by: Paul Gofman <pgofman@codeweavers.com>
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/work_items/15745
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/work_items/16132
Signed-off-by: Matthew Schwartz <matthew.schwartz@linux.dev>
Link: https://cdrdv2.intel.com/v1/dl/getContent/678938 # [1]
---
 arch/x86/entry/entry_fred.c | 11 ++++++++++-
 1 file changed, 10 insertions(+), 1 deletion(-)

diff --git a/arch/x86/entry/entry_fred.c b/arch/x86/entry/entry_fred.c
index fb3594ddf731..854899bfec5d 100644
--- a/arch/x86/entry/entry_fred.c
+++ b/arch/x86/entry/entry_fred.c
@@ -10,6 +10,7 @@
 #include <asm/desc.h>
 #include <asm/fred.h>
 #include <asm/idtentry.h>
+#include <asm/processor-flags.h>
 #include <asm/syscall.h>
 #include <asm/trapnr.h>
 #include <asm/traps.h>
@@ -71,7 +72,15 @@ static noinstr void fred_intx(struct pt_regs *regs)
 #endif
 
 	default:
-		return exc_general_protection(regs, 0);
+		/*
+		 * Reconstruct the #GP fault state that IDT delivery would produce.
+		 * Clear the software event flag so ERETU with TF set does not trap
+		 * before the resumed instruction. See prevent_single_step_upon_eretu().
+		 */
+		regs->ip -= regs->fred_ss.insnlen;
+		regs->flags |= X86_EFLAGS_RF;
+		regs->fred_ss.swevent = 0;
+		return exc_general_protection(regs, (regs->fred_ss.vector << 3) | 2);
 	}
 }
 
-- 
2.55.0

Re: [PATCH 1/2] x86/fred: Reconstruct the #GP context for rejected INT instructions

[H. Peter Anvin]

On 2026-09-17 16:09, Matthew Schwartz wrote:
> FRED event delivery does not use the IDT, so the gate DPL check that
> rejects a user INT n falls to software (Intel FRED specification [1],
> section 8.3). fred_intx() rejects the same vectors as IDT delivery, but
> reports a zero error code and the IP after the INT. This breaks the
> signal ABI. Wine uses the error code to recognize INT 0x2d, so the
> changed context turns a handled breakpoint into an access violation in
> Elden Ring.
> 
> Rewind IP using the instruction length in the augmented SS and
> synthesize the IDT selector error code, (vector << 3) | 2. Set RF in the
> saved flags, as the CPU does for a #GP fault. Section 5.2.1 defines the
> saved vector, instruction length and RF state. The supplied length
> handles prefixes without reading user memory. Limit the changes to
> already-rejected software interrupts, preserving the accepted INT3, INT4
> and enabled INT80 paths and hardware exceptions. With IA32 emulation
> disabled, INT 0x80 now reports the same #GP as the DPL 0 gate IDT
> installs there. The rewound IP also stops fixup_iopl_exception() from
> inspecting the byte after the INT.
> 
> Also clear the software event flag. Section 6.2.3 specifies that ERETU
> with this flag and TF set traps before executing any user instruction. A
> tracer that suppresses SIGSEGV and resumes with TF set expects the next
> instruction to run first, as after IRET. The sigreturn path clears the
> same flag for this reason in prevent_single_step_upon_eretu().
> 
> [1] Intel Flexible Return and Event Delivery (FRED) Specification,
> revision 9.0 (346446-009US), sections 5.2.1, 6.2.3 and 8.3.
> 
> Fixes: 14619d912b65 ("x86/fred: FRED entry/exit and dispatch code")
> Reported-by: Paul Gofman <pgofman@codeweavers.com>
> Closes: https://gitlab.freedesktop.org/mesa/mesa/-/work_items/15745
> Closes: https://gitlab.freedesktop.org/mesa/mesa/-/work_items/16132
> Signed-off-by: Matthew Schwartz <matthew.schwartz@linux.dev>
> Link: https://cdrdv2.intel.com/v1/dl/getContent/678938 # [1]

Reviewed-by: H. Peter Anvin <hpa@zytor.com>

This really should go into -stable.

	-hpa

Re: [PATCH 1/2] x86/fred: Reconstruct the #GP context for rejected INT instructions

[Peter Zijlstra]

On Thu, Sep 17, 2026 at 05:07:45PM -0700, H. Peter Anvin wrote:
> On 2026-09-17 16:09, Matthew Schwartz wrote:
> > FRED event delivery does not use the IDT, so the gate DPL check that
> > rejects a user INT n falls to software (Intel FRED specification [1],
> > section 8.3). fred_intx() rejects the same vectors as IDT delivery, but
> > reports a zero error code and the IP after the INT. This breaks the
> > signal ABI. Wine uses the error code to recognize INT 0x2d, so the
> > changed context turns a handled breakpoint into an access violation in
> > Elden Ring.
> > 
> > Rewind IP using the instruction length in the augmented SS and
> > synthesize the IDT selector error code, (vector << 3) | 2. Set RF in the
> > saved flags, as the CPU does for a #GP fault. Section 5.2.1 defines the
> > saved vector, instruction length and RF state. The supplied length
> > handles prefixes without reading user memory. Limit the changes to
> > already-rejected software interrupts, preserving the accepted INT3, INT4
> > and enabled INT80 paths and hardware exceptions. With IA32 emulation
> > disabled, INT 0x80 now reports the same #GP as the DPL 0 gate IDT
> > installs there. The rewound IP also stops fixup_iopl_exception() from
> > inspecting the byte after the INT.
> > 
> > Also clear the software event flag. Section 6.2.3 specifies that ERETU
> > with this flag and TF set traps before executing any user instruction. A
> > tracer that suppresses SIGSEGV and resumes with TF set expects the next
> > instruction to run first, as after IRET. The sigreturn path clears the
> > same flag for this reason in prevent_single_step_upon_eretu().
> > 
> > [1] Intel Flexible Return and Event Delivery (FRED) Specification,
> > revision 9.0 (346446-009US), sections 5.2.1, 6.2.3 and 8.3.
> > 
> > Fixes: 14619d912b65 ("x86/fred: FRED entry/exit and dispatch code")
> > Reported-by: Paul Gofman <pgofman@codeweavers.com>
> > Closes: https://gitlab.freedesktop.org/mesa/mesa/-/work_items/15745
> > Closes: https://gitlab.freedesktop.org/mesa/mesa/-/work_items/16132
> > Signed-off-by: Matthew Schwartz <matthew.schwartz@linux.dev>
> > Link: https://cdrdv2.intel.com/v1/dl/getContent/678938 # [1]
> 
> Reviewed-by: H. Peter Anvin <hpa@zytor.com>
> 
> This really should go into -stable.

It has a Fixes tag, afaik they really good at picking those up. Let me
go queue them in x86/urgent.

[tip: x86/urgent] x86/fred: Reconstruct the #GP context for rejected INT instructions

[tip-bot2 for Matthew Schwartz]

The following commit has been merged into the x86/urgent branch of tip:

Commit-ID:     93f53499d0b945e8ae447f497faf743d60069f61
Gitweb:        https://git.kernel.org/tip/93f53499d0b945e8ae447f497faf743d60069f61
Author:        Matthew Schwartz <matthew.schwartz@linux.dev>
AuthorDate:    Thu, 17 Sep 2026 16:09:06 -07:00
Committer:     Peter Zijlstra <peterz@infradead.org>
CommitterDate: Fri, 18 Sep 2026 12:33:17 +02:00

x86/fred: Reconstruct the #GP context for rejected INT instructions

FRED event delivery does not use the IDT, so the gate DPL check that
rejects a user INT n falls to software (Intel FRED specification [1],
section 8.3). fred_intx() rejects the same vectors as IDT delivery, but
reports a zero error code and the IP after the INT. This breaks the
signal ABI. Wine uses the error code to recognize INT 0x2d, so the
changed context turns a handled breakpoint into an access violation in
Elden Ring.

Rewind IP using the instruction length in the augmented SS and
synthesize the IDT selector error code, (vector << 3) | 2. Set RF in the
saved flags, as the CPU does for a #GP fault. Section 5.2.1 defines the
saved vector, instruction length and RF state. The supplied length
handles prefixes without reading user memory. Limit the changes to
already-rejected software interrupts, preserving the accepted INT3, INT4
and enabled INT80 paths and hardware exceptions. With IA32 emulation
disabled, INT 0x80 now reports the same #GP as the DPL 0 gate IDT
installs there. The rewound IP also stops fixup_iopl_exception() from
inspecting the byte after the INT.

Also clear the software event flag. Section 6.2.3 specifies that ERETU
with this flag and TF set traps before executing any user instruction. A
tracer that suppresses SIGSEGV and resumes with TF set expects the next
instruction to run first, as after IRET. The sigreturn path clears the
same flag for this reason in prevent_single_step_upon_eretu().

[1] Intel Flexible Return and Event Delivery (FRED) Specification,
revision 9.0 (346446-009US), sections 5.2.1, 6.2.3 and 8.3.

Fixes: 14619d912b65 ("x86/fred: FRED entry/exit and dispatch code")
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/work_items/15745
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/work_items/16132
Reported-by: Paul Gofman <pgofman@codeweavers.com>
Signed-off-by: Matthew Schwartz <matthew.schwartz@linux.dev>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: H. Peter Anvin <hpa@zytor.com>
Link: https://cdrdv2.intel.com/v1/dl/getContent/678938 # [1]
Link: https://patch.msgid.link/20260917230907.2080792-2-matthew.schwartz@linux.dev
---
 arch/x86/entry/entry_fred.c | 11 ++++++++++-
 1 file changed, 10 insertions(+), 1 deletion(-)

diff --git a/arch/x86/entry/entry_fred.c b/arch/x86/entry/entry_fred.c
index fb3594d..854899b 100644
--- a/arch/x86/entry/entry_fred.c
+++ b/arch/x86/entry/entry_fred.c
@@ -10,6 +10,7 @@
 #include <asm/desc.h>
 #include <asm/fred.h>
 #include <asm/idtentry.h>
+#include <asm/processor-flags.h>
 #include <asm/syscall.h>
 #include <asm/trapnr.h>
 #include <asm/traps.h>
@@ -71,7 +72,15 @@ static noinstr void fred_intx(struct pt_regs *regs)
 #endif
 
 	default:
-		return exc_general_protection(regs, 0);
+		/*
+		 * Reconstruct the #GP fault state that IDT delivery would produce.
+		 * Clear the software event flag so ERETU with TF set does not trap
+		 * before the resumed instruction. See prevent_single_step_upon_eretu().
+		 */
+		regs->ip -= regs->fred_ss.insnlen;
+		regs->flags |= X86_EFLAGS_RF;
+		regs->fred_ss.swevent = 0;
+		return exc_general_protection(regs, (regs->fred_ss.vector << 3) | 2);
 	}
 }
 

[PATCH 2/2] selftests/x86: Check signal state for rejected software interrupts

[Matthew Schwartz]

Add a test of the signal ABI for INT instructions in both 32-bit and
64-bit processes. Check the signal number, trap number, error code,
si_code, si_addr, instruction pointer and RF/TF state against legacy
IDT behavior. Include a 15-byte prefixed INT to check that IP uses the
hardware instruction length. Exercise both INT3 encodings, INT4, UD2
and HLT to cover the unchanged trap and fault paths.

Run each instruction with TF clear and set. Resume at a known NOP after
handling the signal and check that single-stepping traps after the NOP.

Also drive INT 0x2d under ptrace, which resumes through the fault frame
rather than sigreturn and so exposes a stale FRED software event flag.
Start from an INT3 stop, whose FRED frame has no software event flag,
instead of the syscall frame of raise(SIGSTOP). Single-step into the INT
and check that the fault reports its address. Then suppress SIGSEGV and
resume at the NOP, once with PTRACE_SINGLESTEP and once with PTRACE_CONT
and TF set. Section 6.2.3 of the Intel FRED specification [1] specifies
the immediate single-step trap caused by returning with both that flag
and TF set. Check that each trap occurs after the NOP, rather than at
its address.

Report whether the CPU supports FRED, since a pass looks the same on
either entry path. INT 0x80 with IA32 emulation disabled and a 64-bit
tracer of a 32-bit tracee are not covered.

Both variants pass all 29 checks on a non-FRED AMD host and on Panther
Lake with FRED enabled and the fix applied. With the same binaries on
unpatched Panther Lake, 16 signal-context checks fail and the first
ptrace check reports the IP after the INT. The two dependent ptrace
resume checks are not reached.

[1] Intel Flexible Return and Event Delivery (FRED) Specification,
revision 9.0 (346446-009US), section 6.2.3.

Signed-off-by: Matthew Schwartz <matthew.schwartz@linux.dev>
Link: https://cdrdv2.intel.com/v1/dl/getContent/678938 # [1]
---
 tools/testing/selftests/x86/Makefile     |   2 +-
 tools/testing/selftests/x86/int_signal.c | 311 +++++++++++++++++++++++
 2 files changed, 312 insertions(+), 1 deletion(-)
 create mode 100644 tools/testing/selftests/x86/int_signal.c

diff --git a/tools/testing/selftests/x86/Makefile b/tools/testing/selftests/x86/Makefile
index 434065215d12..d478b13cc8d5 100644
--- a/tools/testing/selftests/x86/Makefile
+++ b/tools/testing/selftests/x86/Makefile
@@ -13,7 +13,7 @@ CAN_BUILD_WITH_NOPIE := $(shell ./check_cc.sh "$(CC)" trivial_program.c -no-pie)
 TARGETS_C_BOTHBITS := single_step_syscall sysret_ss_attrs syscall_nt test_mremap_vdso \
 			check_initial_reg_state sigreturn iopl ioperm \
 			test_vsyscall mov_ss_trap sigtrap_loop \
-			syscall_arg_fault fsgsbase_restore sigaltstack
+			syscall_arg_fault fsgsbase_restore sigaltstack int_signal
 TARGETS_C_BOTHBITS += nx_stack
 TARGETS_C_32BIT_ONLY := entry_from_vm86 test_syscall_vdso unwind_vdso \
 			test_FCMOV test_FCOMI test_FISTTP \
diff --git a/tools/testing/selftests/x86/int_signal.c b/tools/testing/selftests/x86/int_signal.c
new file mode 100644
index 000000000000..22676dac72b5
--- /dev/null
+++ b/tools/testing/selftests/x86/int_signal.c
@@ -0,0 +1,311 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/* Check the signal context for INT instructions with IDT and FRED entry. */
+#define _GNU_SOURCE
+
+#include <cpuid.h>
+#include <errno.h>
+#include <stdbool.h>
+#include <stddef.h>
+#include <stdint.h>
+#include <sys/ptrace.h>
+#include <sys/user.h>
+#include <sys/wait.h>
+#include <unistd.h>
+#include <ucontext.h>
+
+#include "helpers.h"
+
+#ifdef __x86_64__
+#define REG_IP REG_RIP
+#define USER_IP rip
+#define STACK_PTR "%rsp"
+#else
+#define REG_IP REG_EIP
+#define USER_IP eip
+#define STACK_PTR "%esp"
+#endif
+
+/*
+ * Each instruction has normal and single-step entry points. Resume at the
+ * NOP after handling its signal, then expect a trace trap after that NOP
+ * when TF is set. Explicit labels avoid assuming the kernel's saved IP.
+ */
+#define PROBE(name, insn) \
+	extern void name(void); \
+	extern void name##_tf(void); \
+	extern const char name##_end[], name##_step[]; \
+	asm(".pushsection .text\n" \
+	    ".globl " #name "_tf\n" \
+	    ".type " #name "_tf, @function\n" \
+	    #name "_tf:\n" \
+	    "pushf\n" \
+	    "orl $0x100, (" STACK_PTR ")\n" \
+	    "popf\n" \
+	    ".globl " #name "\n" \
+	    ".type " #name ", @function\n" \
+	    #name ":\n" insn "\n" \
+	    ".globl " #name "_end\n" \
+	    #name "_end:\nnop\n" \
+	    ".globl " #name "_step\n" \
+	    #name "_step:\nret\n" \
+	    ".size " #name ", .-" #name "\n" \
+	    ".size " #name "_tf, .-" #name "_tf\n" \
+	    ".popsection\n")
+
+PROBE(int1, ".byte 0xcd, 0x01");
+PROBE(int29, ".byte 0xcd, 0x29");
+PROBE(int2c, ".byte 0xcd, 0x2c");
+PROBE(int2d, ".byte 0xcd, 0x2d");
+PROBE(prefixed_int2d, ".byte 0x66, 0xcd, 0x2d");
+PROBE(long_int2d, ".fill 13, 1, 0x2e\n.byte 0xcd, 0x2d");
+PROBE(int81, ".byte 0xcd, 0x81");
+PROBE(intff, ".byte 0xcd, 0xff");
+PROBE(short_int3, ".byte 0xcc");
+PROBE(long_int3, ".byte 0xcd, 0x03");
+PROBE(int4, ".byte 0xcd, 0x04");
+PROBE(ud2, ".byte 0x0f, 0x0b");
+PROBE(hlt, ".byte 0xf4");
+
+struct test {
+	const char *name;
+	void (*run)(void);
+	void (*run_tf)(void);
+	const char *end, *step;
+	int signo, trap, error, ip_offset, flags, code;
+};
+
+#define TEST(name, sig, trap, error, offset, flags, code) \
+	{ #name, name, name##_tf, name##_end, name##_step, \
+	  sig, trap, error, offset, flags, code }
+
+#define GP(name, error) \
+	TEST(name, SIGSEGV, 13, error, 0, X86_EFLAGS_RF, SI_KERNEL)
+
+static const struct test tests[] = {
+	GP(int1, 0x00a),
+	GP(int29, 0x14a),
+	GP(int2c, 0x162),
+	GP(int2d, 0x16a),
+	GP(prefixed_int2d, 0x16a),
+	GP(long_int2d, 0x16a),
+	GP(int81, 0x40a),
+	GP(intff, 0x7fa),
+	GP(hlt, 0),
+	TEST(short_int3, SIGTRAP, 3, 0, 1, 0, SI_KERNEL),
+	TEST(long_int3, SIGTRAP, 3, 0, 2, 0, SI_KERNEL),
+	TEST(int4, SIGSEGV, 4, 0, 2, 0, SI_KERNEL),
+	TEST(ud2, SIGILL, 6, 0, 0, X86_EFLAGS_RF, ILL_ILLOPN),
+};
+
+static const struct test *active;
+static volatile sig_atomic_t seen, signo, trap, error, ip_offset, flags;
+static volatile sig_atomic_t code, addr_ok, single_step, stepped, step_ok;
+
+static void handler(int sig, siginfo_t *info, void *context)
+{
+	ucontext_t *uc = context;
+	uintptr_t ip = uc->uc_mcontext.gregs[REG_IP];
+	uintptr_t start = (uintptr_t)active->run;
+	uintptr_t end = (uintptr_t)active->end;
+
+	if (seen && single_step && sig == SIGTRAP) {
+		if (stepped++) {
+			ksft_print_msg("%s: second trace trap at %#lx\n",
+				       active->name, (unsigned long)ip);
+			_exit(KSFT_FAIL);
+		}
+		step_ok = ip == (uintptr_t)active->step &&
+			uc->uc_mcontext.gregs[REG_TRAPNO] == 1 &&
+			info->si_code == TRAP_TRACE;
+		uc->uc_mcontext.gregs[REG_EFL] &= ~X86_EFLAGS_TF;
+		return;
+	}
+
+	if (seen || ip < start || ip > end) {
+		ksft_print_msg("%s: unexpected signal %d at %#lx\n",
+			       active->name, sig, (unsigned long)ip);
+		_exit(KSFT_FAIL);
+	}
+
+	signo = sig;
+	trap = uc->uc_mcontext.gregs[REG_TRAPNO];
+	error = uc->uc_mcontext.gregs[REG_ERR];
+	ip_offset = ip - start;
+	flags = uc->uc_mcontext.gregs[REG_EFL] & (X86_EFLAGS_RF | X86_EFLAGS_TF);
+	code = info->si_code;
+	/* force_sig() reports no address, force_sig_fault() reports the IP. */
+	addr_ok = info->si_addr == (code == SI_KERNEL ? NULL : (void *)ip);
+	seen = 1;
+	uc->uc_mcontext.gregs[REG_IP] = end;
+}
+
+static void wait_for_child(pid_t child, int *status)
+{
+	pid_t ret;
+
+	do {
+		ret = waitpid(child, status, 0);
+	} while (ret < 0 && errno == EINTR);
+	if (ret != child)
+		ksft_exit_fail_perror("waitpid");
+}
+
+/* Resume the tracee and check where the next stop lands. */
+static bool resume_to(pid_t child, int *status, int request, int sig,
+		      const void *ip, const char *what)
+{
+	struct user_regs_struct regs;
+
+	if (ptrace(request, child, 0, 0))
+		return false;
+	wait_for_child(child, status);
+	if (!WIFSTOPPED(*status)) {
+		ksft_print_msg("%s: tracee did not stop\n", what);
+		return false;
+	}
+	if (WSTOPSIG(*status) != sig) {
+		ksft_print_msg("%s: stopped with signal %d, expected %d\n",
+			       what, WSTOPSIG(*status), sig);
+		return false;
+	}
+	if (ptrace(PTRACE_GETREGS, child, 0, &regs))
+		return false;
+	if ((unsigned long)regs.USER_IP != (unsigned long)ip) {
+		ksft_print_msg("%s: stopped at %#lx, expected %#lx\n", what,
+			       (unsigned long)regs.USER_IP, (unsigned long)ip);
+		return false;
+	}
+	return true;
+}
+
+static bool set_ip(pid_t child, const void *ip, bool tf)
+{
+	struct user_regs_struct regs;
+
+	if (ptrace(PTRACE_GETREGS, child, 0, &regs))
+		return false;
+	regs.USER_IP = (unsigned long)ip;
+	if (tf)
+		regs.eflags |= X86_EFLAGS_TF;
+	return !ptrace(PTRACE_SETREGS, child, 0, &regs);
+}
+
+/*
+ * Exercise the tracer paths that resume through the fault frame rather than
+ * sigreturn. A stale FRED software event flag on that frame traps before the
+ * NOP executes instead of after it.
+ */
+static void test_ptrace(void)
+{
+	bool into = false, step = false, cont = false;
+	pid_t child;
+	int status;
+
+	child = fork();
+	if (child < 0)
+		ksft_exit_fail_perror("fork");
+	if (!child) {
+		if (ptrace(PTRACE_TRACEME, 0, 0, 0))
+			_exit(KSFT_FAIL);
+		/* Start from a breakpoint frame, not the syscall frame of raise(). */
+		asm volatile("int3");
+		_exit(KSFT_FAIL);
+	}
+
+	wait_for_child(child, &status);
+	if (!WIFSTOPPED(status) || WSTOPSIG(status) != SIGTRAP)
+		goto out;
+	if (ptrace(PTRACE_SETOPTIONS, child, 0, PTRACE_O_EXITKILL))
+		goto out;
+
+	/* Single-step into the INT. The fault must report the INT's address. */
+	if (!set_ip(child, int2d, false))
+		goto out;
+	into = resume_to(child, &status, PTRACE_SINGLESTEP, SIGSEGV, int2d,
+			 "single-step into INT");
+	if (!into)
+		goto out;
+
+	/* Suppress SIGSEGV and single-step the NOP. */
+	if (!set_ip(child, int2d_end, false))
+		goto out;
+	step = resume_to(child, &status, PTRACE_SINGLESTEP, SIGTRAP, int2d_step,
+			 "single-step after INT");
+	if (!step)
+		goto out;
+
+	/* Fault again, then suppress SIGSEGV and continue with TF set. */
+	if (!set_ip(child, int2d, false))
+		goto out;
+	if (!resume_to(child, &status, PTRACE_CONT, SIGSEGV, int2d,
+		       "continue to INT"))
+		goto out;
+	if (!set_ip(child, int2d_end, true))
+		goto out;
+	cont = resume_to(child, &status, PTRACE_CONT, SIGTRAP, int2d_step,
+			 "continue with TF after INT");
+out:
+	if (WIFSTOPPED(status)) {
+		kill(child, SIGKILL);
+		wait_for_child(child, &status);
+	}
+	ksft_test_result(into, "ptrace single-step into INT faults at the INT\n");
+	ksft_test_result(step, "ptrace single-step after suppressing SIGSEGV\n");
+	ksft_test_result(cont, "ptrace continue with TF after suppressing SIGSEGV\n");
+}
+
+static bool cpu_has_fred(void)
+{
+	unsigned int eax, ebx, ecx, edx;
+
+	if (__get_cpuid_max(0, NULL) < 7)
+		return false;
+	__cpuid_count(7, 1, eax, ebx, ecx, edx);
+	return eax & (1 << 17);
+}
+
+int main(void)
+{
+	unsigned int i, tf;
+	int expected_flags, ok;
+
+	ksft_print_header();
+	ksft_set_plan(2 * ARRAY_SIZE(tests) + 3);
+	ksft_print_msg("CPU %s FRED\n", cpu_has_fred() ? "supports" : "lacks");
+	sethandler(SIGSEGV, handler, 0);
+	sethandler(SIGTRAP, handler, 0);
+	sethandler(SIGILL, handler, 0);
+
+	for (tf = 0; tf < 2; tf++) {
+		for (i = 0; i < ARRAY_SIZE(tests); i++) {
+			active = &tests[i];
+			single_step = tf;
+			seen = signo = trap = error = ip_offset = flags = 0;
+			code = addr_ok = stepped = step_ok = 0;
+			expected_flags = active->flags | (tf ? X86_EFLAGS_TF : 0);
+			if (tf)
+				active->run_tf();
+			else
+				active->run();
+
+			ok = seen && signo == active->signo && trap == active->trap &&
+				error == active->error && ip_offset == active->ip_offset &&
+				flags == expected_flags && code == active->code && addr_ok &&
+				(!tf || (stepped && step_ok));
+			ksft_test_result(ok, "%s%s\n", active->name, tf ? " with TF" : "");
+			if (!ok) {
+				ksft_print_msg("got signal=%d trap=%d error=%#x ip=%d\n",
+					       signo, trap, error, ip_offset);
+				ksft_print_msg("got flags=%#x code=%d addr_ok=%d step_ok=%d\n",
+					       flags, code, addr_ok, step_ok);
+				ksft_print_msg("expected signal=%d trap=%d error=%#x ip=%d\n",
+					       active->signo, active->trap, active->error,
+					       active->ip_offset);
+				ksft_print_msg("expected flags=%#x code=%d\n",
+					       expected_flags, active->code);
+			}
+		}
+	}
+	test_ptrace();
+	ksft_finished();
+}
-- 
2.55.0

[tip: x86/urgent] selftests/x86: Check signal state for rejected software interrupts

[tip-bot2 for Matthew Schwartz]

The following commit has been merged into the x86/urgent branch of tip:

Commit-ID:     96443a53bc3ef4b67dab0c497878fc8d56f799f3
Gitweb:        https://git.kernel.org/tip/96443a53bc3ef4b67dab0c497878fc8d56f799f3
Author:        Matthew Schwartz <matthew.schwartz@linux.dev>
AuthorDate:    Thu, 17 Sep 2026 16:09:07 -07:00
Committer:     Peter Zijlstra <peterz@infradead.org>
CommitterDate: Fri, 18 Sep 2026 12:33:18 +02:00

selftests/x86: Check signal state for rejected software interrupts

Add a test of the signal ABI for INT instructions in both 32-bit and
64-bit processes. Check the signal number, trap number, error code,
si_code, si_addr, instruction pointer and RF/TF state against legacy
IDT behavior. Include a 15-byte prefixed INT to check that IP uses the
hardware instruction length. Exercise both INT3 encodings, INT4, UD2
and HLT to cover the unchanged trap and fault paths.

Run each instruction with TF clear and set. Resume at a known NOP after
handling the signal and check that single-stepping traps after the NOP.

Also drive INT 0x2d under ptrace, which resumes through the fault frame
rather than sigreturn and so exposes a stale FRED software event flag.
Start from an INT3 stop, whose FRED frame has no software event flag,
instead of the syscall frame of raise(SIGSTOP). Single-step into the INT
and check that the fault reports its address. Then suppress SIGSEGV and
resume at the NOP, once with PTRACE_SINGLESTEP and once with PTRACE_CONT
and TF set. Section 6.2.3 of the Intel FRED specification [1] specifies
the immediate single-step trap caused by returning with both that flag
and TF set. Check that each trap occurs after the NOP, rather than at
its address.

Report whether the CPU supports FRED, since a pass looks the same on
either entry path. INT 0x80 with IA32 emulation disabled and a 64-bit
tracer of a 32-bit tracee are not covered.

Both variants pass all 29 checks on a non-FRED AMD host and on Panther
Lake with FRED enabled and the fix applied. With the same binaries on
unpatched Panther Lake, 16 signal-context checks fail and the first
ptrace check reports the IP after the INT. The two dependent ptrace
resume checks are not reached.

[1] Intel Flexible Return and Event Delivery (FRED) Specification,
revision 9.0 (346446-009US), section 6.2.3.

Signed-off-by: Matthew Schwartz <matthew.schwartz@linux.dev>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://cdrdv2.intel.com/v1/dl/getContent/678938 # [1]
Link: https://patch.msgid.link/20260917230907.2080792-3-matthew.schwartz@linux.dev
---
 tools/testing/selftests/x86/Makefile     |   2 +-
 tools/testing/selftests/x86/int_signal.c | 311 ++++++++++++++++++++++-
 2 files changed, 312 insertions(+), 1 deletion(-)
 create mode 100644 tools/testing/selftests/x86/int_signal.c

diff --git a/tools/testing/selftests/x86/Makefile b/tools/testing/selftests/x86/Makefile
index 4340652..d478b13 100644
--- a/tools/testing/selftests/x86/Makefile
+++ b/tools/testing/selftests/x86/Makefile
@@ -13,7 +13,7 @@ CAN_BUILD_WITH_NOPIE := $(shell ./check_cc.sh "$(CC)" trivial_program.c -no-pie)
 TARGETS_C_BOTHBITS := single_step_syscall sysret_ss_attrs syscall_nt test_mremap_vdso \
 			check_initial_reg_state sigreturn iopl ioperm \
 			test_vsyscall mov_ss_trap sigtrap_loop \
-			syscall_arg_fault fsgsbase_restore sigaltstack
+			syscall_arg_fault fsgsbase_restore sigaltstack int_signal
 TARGETS_C_BOTHBITS += nx_stack
 TARGETS_C_32BIT_ONLY := entry_from_vm86 test_syscall_vdso unwind_vdso \
 			test_FCMOV test_FCOMI test_FISTTP \
diff --git a/tools/testing/selftests/x86/int_signal.c b/tools/testing/selftests/x86/int_signal.c
new file mode 100644
index 0000000..22676da
--- /dev/null
+++ b/tools/testing/selftests/x86/int_signal.c
@@ -0,0 +1,311 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/* Check the signal context for INT instructions with IDT and FRED entry. */
+#define _GNU_SOURCE
+
+#include <cpuid.h>
+#include <errno.h>
+#include <stdbool.h>
+#include <stddef.h>
+#include <stdint.h>
+#include <sys/ptrace.h>
+#include <sys/user.h>
+#include <sys/wait.h>
+#include <unistd.h>
+#include <ucontext.h>
+
+#include "helpers.h"
+
+#ifdef __x86_64__
+#define REG_IP REG_RIP
+#define USER_IP rip
+#define STACK_PTR "%rsp"
+#else
+#define REG_IP REG_EIP
+#define USER_IP eip
+#define STACK_PTR "%esp"
+#endif
+
+/*
+ * Each instruction has normal and single-step entry points. Resume at the
+ * NOP after handling its signal, then expect a trace trap after that NOP
+ * when TF is set. Explicit labels avoid assuming the kernel's saved IP.
+ */
+#define PROBE(name, insn) \
+	extern void name(void); \
+	extern void name##_tf(void); \
+	extern const char name##_end[], name##_step[]; \
+	asm(".pushsection .text\n" \
+	    ".globl " #name "_tf\n" \
+	    ".type " #name "_tf, @function\n" \
+	    #name "_tf:\n" \
+	    "pushf\n" \
+	    "orl $0x100, (" STACK_PTR ")\n" \
+	    "popf\n" \
+	    ".globl " #name "\n" \
+	    ".type " #name ", @function\n" \
+	    #name ":\n" insn "\n" \
+	    ".globl " #name "_end\n" \
+	    #name "_end:\nnop\n" \
+	    ".globl " #name "_step\n" \
+	    #name "_step:\nret\n" \
+	    ".size " #name ", .-" #name "\n" \
+	    ".size " #name "_tf, .-" #name "_tf\n" \
+	    ".popsection\n")
+
+PROBE(int1, ".byte 0xcd, 0x01");
+PROBE(int29, ".byte 0xcd, 0x29");
+PROBE(int2c, ".byte 0xcd, 0x2c");
+PROBE(int2d, ".byte 0xcd, 0x2d");
+PROBE(prefixed_int2d, ".byte 0x66, 0xcd, 0x2d");
+PROBE(long_int2d, ".fill 13, 1, 0x2e\n.byte 0xcd, 0x2d");
+PROBE(int81, ".byte 0xcd, 0x81");
+PROBE(intff, ".byte 0xcd, 0xff");
+PROBE(short_int3, ".byte 0xcc");
+PROBE(long_int3, ".byte 0xcd, 0x03");
+PROBE(int4, ".byte 0xcd, 0x04");
+PROBE(ud2, ".byte 0x0f, 0x0b");
+PROBE(hlt, ".byte 0xf4");
+
+struct test {
+	const char *name;
+	void (*run)(void);
+	void (*run_tf)(void);
+	const char *end, *step;
+	int signo, trap, error, ip_offset, flags, code;
+};
+
+#define TEST(name, sig, trap, error, offset, flags, code) \
+	{ #name, name, name##_tf, name##_end, name##_step, \
+	  sig, trap, error, offset, flags, code }
+
+#define GP(name, error) \
+	TEST(name, SIGSEGV, 13, error, 0, X86_EFLAGS_RF, SI_KERNEL)
+
+static const struct test tests[] = {
+	GP(int1, 0x00a),
+	GP(int29, 0x14a),
+	GP(int2c, 0x162),
+	GP(int2d, 0x16a),
+	GP(prefixed_int2d, 0x16a),
+	GP(long_int2d, 0x16a),
+	GP(int81, 0x40a),
+	GP(intff, 0x7fa),
+	GP(hlt, 0),
+	TEST(short_int3, SIGTRAP, 3, 0, 1, 0, SI_KERNEL),
+	TEST(long_int3, SIGTRAP, 3, 0, 2, 0, SI_KERNEL),
+	TEST(int4, SIGSEGV, 4, 0, 2, 0, SI_KERNEL),
+	TEST(ud2, SIGILL, 6, 0, 0, X86_EFLAGS_RF, ILL_ILLOPN),
+};
+
+static const struct test *active;
+static volatile sig_atomic_t seen, signo, trap, error, ip_offset, flags;
+static volatile sig_atomic_t code, addr_ok, single_step, stepped, step_ok;
+
+static void handler(int sig, siginfo_t *info, void *context)
+{
+	ucontext_t *uc = context;
+	uintptr_t ip = uc->uc_mcontext.gregs[REG_IP];
+	uintptr_t start = (uintptr_t)active->run;
+	uintptr_t end = (uintptr_t)active->end;
+
+	if (seen && single_step && sig == SIGTRAP) {
+		if (stepped++) {
+			ksft_print_msg("%s: second trace trap at %#lx\n",
+				       active->name, (unsigned long)ip);
+			_exit(KSFT_FAIL);
+		}
+		step_ok = ip == (uintptr_t)active->step &&
+			uc->uc_mcontext.gregs[REG_TRAPNO] == 1 &&
+			info->si_code == TRAP_TRACE;
+		uc->uc_mcontext.gregs[REG_EFL] &= ~X86_EFLAGS_TF;
+		return;
+	}
+
+	if (seen || ip < start || ip > end) {
+		ksft_print_msg("%s: unexpected signal %d at %#lx\n",
+			       active->name, sig, (unsigned long)ip);
+		_exit(KSFT_FAIL);
+	}
+
+	signo = sig;
+	trap = uc->uc_mcontext.gregs[REG_TRAPNO];
+	error = uc->uc_mcontext.gregs[REG_ERR];
+	ip_offset = ip - start;
+	flags = uc->uc_mcontext.gregs[REG_EFL] & (X86_EFLAGS_RF | X86_EFLAGS_TF);
+	code = info->si_code;
+	/* force_sig() reports no address, force_sig_fault() reports the IP. */
+	addr_ok = info->si_addr == (code == SI_KERNEL ? NULL : (void *)ip);
+	seen = 1;
+	uc->uc_mcontext.gregs[REG_IP] = end;
+}
+
+static void wait_for_child(pid_t child, int *status)
+{
+	pid_t ret;
+
+	do {
+		ret = waitpid(child, status, 0);
+	} while (ret < 0 && errno == EINTR);
+	if (ret != child)
+		ksft_exit_fail_perror("waitpid");
+}
+
+/* Resume the tracee and check where the next stop lands. */
+static bool resume_to(pid_t child, int *status, int request, int sig,
+		      const void *ip, const char *what)
+{
+	struct user_regs_struct regs;
+
+	if (ptrace(request, child, 0, 0))
+		return false;
+	wait_for_child(child, status);
+	if (!WIFSTOPPED(*status)) {
+		ksft_print_msg("%s: tracee did not stop\n", what);
+		return false;
+	}
+	if (WSTOPSIG(*status) != sig) {
+		ksft_print_msg("%s: stopped with signal %d, expected %d\n",
+			       what, WSTOPSIG(*status), sig);
+		return false;
+	}
+	if (ptrace(PTRACE_GETREGS, child, 0, &regs))
+		return false;
+	if ((unsigned long)regs.USER_IP != (unsigned long)ip) {
+		ksft_print_msg("%s: stopped at %#lx, expected %#lx\n", what,
+			       (unsigned long)regs.USER_IP, (unsigned long)ip);
+		return false;
+	}
+	return true;
+}
+
+static bool set_ip(pid_t child, const void *ip, bool tf)
+{
+	struct user_regs_struct regs;
+
+	if (ptrace(PTRACE_GETREGS, child, 0, &regs))
+		return false;
+	regs.USER_IP = (unsigned long)ip;
+	if (tf)
+		regs.eflags |= X86_EFLAGS_TF;
+	return !ptrace(PTRACE_SETREGS, child, 0, &regs);
+}
+
+/*
+ * Exercise the tracer paths that resume through the fault frame rather than
+ * sigreturn. A stale FRED software event flag on that frame traps before the
+ * NOP executes instead of after it.
+ */
+static void test_ptrace(void)
+{
+	bool into = false, step = false, cont = false;
+	pid_t child;
+	int status;
+
+	child = fork();
+	if (child < 0)
+		ksft_exit_fail_perror("fork");
+	if (!child) {
+		if (ptrace(PTRACE_TRACEME, 0, 0, 0))
+			_exit(KSFT_FAIL);
+		/* Start from a breakpoint frame, not the syscall frame of raise(). */
+		asm volatile("int3");
+		_exit(KSFT_FAIL);
+	}
+
+	wait_for_child(child, &status);
+	if (!WIFSTOPPED(status) || WSTOPSIG(status) != SIGTRAP)
+		goto out;
+	if (ptrace(PTRACE_SETOPTIONS, child, 0, PTRACE_O_EXITKILL))
+		goto out;
+
+	/* Single-step into the INT. The fault must report the INT's address. */
+	if (!set_ip(child, int2d, false))
+		goto out;
+	into = resume_to(child, &status, PTRACE_SINGLESTEP, SIGSEGV, int2d,
+			 "single-step into INT");
+	if (!into)
+		goto out;
+
+	/* Suppress SIGSEGV and single-step the NOP. */
+	if (!set_ip(child, int2d_end, false))
+		goto out;
+	step = resume_to(child, &status, PTRACE_SINGLESTEP, SIGTRAP, int2d_step,
+			 "single-step after INT");
+	if (!step)
+		goto out;
+
+	/* Fault again, then suppress SIGSEGV and continue with TF set. */
+	if (!set_ip(child, int2d, false))
+		goto out;
+	if (!resume_to(child, &status, PTRACE_CONT, SIGSEGV, int2d,
+		       "continue to INT"))
+		goto out;
+	if (!set_ip(child, int2d_end, true))
+		goto out;
+	cont = resume_to(child, &status, PTRACE_CONT, SIGTRAP, int2d_step,
+			 "continue with TF after INT");
+out:
+	if (WIFSTOPPED(status)) {
+		kill(child, SIGKILL);
+		wait_for_child(child, &status);
+	}
+	ksft_test_result(into, "ptrace single-step into INT faults at the INT\n");
+	ksft_test_result(step, "ptrace single-step after suppressing SIGSEGV\n");
+	ksft_test_result(cont, "ptrace continue with TF after suppressing SIGSEGV\n");
+}
+
+static bool cpu_has_fred(void)
+{
+	unsigned int eax, ebx, ecx, edx;
+
+	if (__get_cpuid_max(0, NULL) < 7)
+		return false;
+	__cpuid_count(7, 1, eax, ebx, ecx, edx);
+	return eax & (1 << 17);
+}
+
+int main(void)
+{
+	unsigned int i, tf;
+	int expected_flags, ok;
+
+	ksft_print_header();
+	ksft_set_plan(2 * ARRAY_SIZE(tests) + 3);
+	ksft_print_msg("CPU %s FRED\n", cpu_has_fred() ? "supports" : "lacks");
+	sethandler(SIGSEGV, handler, 0);
+	sethandler(SIGTRAP, handler, 0);
+	sethandler(SIGILL, handler, 0);
+
+	for (tf = 0; tf < 2; tf++) {
+		for (i = 0; i < ARRAY_SIZE(tests); i++) {
+			active = &tests[i];
+			single_step = tf;
+			seen = signo = trap = error = ip_offset = flags = 0;
+			code = addr_ok = stepped = step_ok = 0;
+			expected_flags = active->flags | (tf ? X86_EFLAGS_TF : 0);
+			if (tf)
+				active->run_tf();
+			else
+				active->run();
+
+			ok = seen && signo == active->signo && trap == active->trap &&
+				error == active->error && ip_offset == active->ip_offset &&
+				flags == expected_flags && code == active->code && addr_ok &&
+				(!tf || (stepped && step_ok));
+			ksft_test_result(ok, "%s%s\n", active->name, tf ? " with TF" : "");
+			if (!ok) {
+				ksft_print_msg("got signal=%d trap=%d error=%#x ip=%d\n",
+					       signo, trap, error, ip_offset);
+				ksft_print_msg("got flags=%#x code=%d addr_ok=%d step_ok=%d\n",
+					       flags, code, addr_ok, step_ok);
+				ksft_print_msg("expected signal=%d trap=%d error=%#x ip=%d\n",
+					       active->signo, active->trap, active->error,
+					       active->ip_offset);
+				ksft_print_msg("expected flags=%#x code=%d\n",
+					       expected_flags, active->code);
+			}
+		}
+	}
+	test_ptrace();
+	ksft_finished();
+}