Merge tag 'x86-fpu-2024-09-17' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip

Pull x86 fpu updates from Thomas Gleixner:
 "Provide FPU buffer layout in core dumps:

  Debuggers have guess the FPU buffer layout in core dumps, which is
  error prone. This is because AMD and Intel layouts differ.

  To avoid buggy heuristics add a ELF section which describes the buffer
  layout which can be retrieved by tools"

* tag 'x86-fpu-2024-09-17' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/elf: Add a new FPU buffer layout info to x86 core files

arch/x86/Kconfig

@@ -107,6 +107,7 @@ config X86
 	select ARCH_HAS_DEBUG_WX
 	select ARCH_HAS_ZONE_DMA_SET if EXPERT
 	select ARCH_HAVE_NMI_SAFE_CMPXCHG
+	select ARCH_HAVE_EXTRA_ELF_NOTES
 	select ARCH_MHP_MEMMAP_ON_MEMORY_ENABLE
 	select ARCH_MIGHT_HAVE_ACPI_PDC		if ACPI
 	select ARCH_MIGHT_HAVE_PC_PARPORT

arch/x86/include/uapi/asm/elf.h

+/* SPDX-License-Identifier: GPL-2.0 WITH Linux-syscall-note */
+#ifndef _UAPI_ASM_X86_ELF_H
+#define _UAPI_ASM_X86_ELF_H
+
+#include <linux/types.h>
+
+struct x86_xfeat_component {
+	__u32 type;
+	__u32 size;
+	__u32 offset;
+	__u32 flags;
+} __packed;
+
+_Static_assert(sizeof(struct x86_xfeat_component) % 4 == 0, "x86_xfeat_component is not aligned");
+
+#endif /* _UAPI_ASM_X86_ELF_H */

arch/x86/kernel/fpu/xstate.c

@@ -13,6 +13,7 @@
 #include <linux/seq_file.h>
 #include <linux/proc_fs.h>
 #include <linux/vmalloc.h>
+#include <linux/coredump.h>
 
 #include <asm/fpu/api.h>
 #include <asm/fpu/regset.h>
@@ -23,6 +24,8 @@
 #include <asm/prctl.h>
 #include <asm/elf.h>
 
+#include <uapi/asm/elf.h>
+
 #include "context.h"
 #include "internal.h"
 #include "legacy.h"
@@ -1841,3 +1844,89 @@ int proc_pid_arch_status(struct seq_file *m, struct pid_namespace *ns,
 	return 0;
 }
 #endif /* CONFIG_PROC_PID_ARCH_STATUS */
+
+#ifdef CONFIG_COREDUMP
+static const char owner_name[] = "LINUX";
+
+/*
+ * Dump type, size, offset and flag values for every xfeature that is present.
+ */
+static int dump_xsave_layout_desc(struct coredump_params *cprm)
+{
+	int num_records = 0;
+	int i;
+
+	for_each_extended_xfeature(i, fpu_user_cfg.max_features) {
+		struct x86_xfeat_component xc = {
+			.type   = i,
+			.size   = xstate_sizes[i],
+			.offset = xstate_offsets[i],
+			/* reserved for future use */
+			.flags  = 0,
+		};
+
+		if (!dump_emit(cprm, &xc, sizeof(xc)))
+			return 0;
+
+		num_records++;
+	}
+	return num_records;
+}
+
+static u32 get_xsave_desc_size(void)
+{
+	u32 cnt = 0;
+	u32 i;
+
+	for_each_extended_xfeature(i, fpu_user_cfg.max_features)
+		cnt++;
+
+	return cnt * (sizeof(struct x86_xfeat_component));
+}
+
+int elf_coredump_extra_notes_write(struct coredump_params *cprm)
+{
+	int num_records = 0;
+	struct elf_note en;
+
+	if (!fpu_user_cfg.max_features)
+		return 0;
+
+	en.n_namesz = sizeof(owner_name);
+	en.n_descsz = get_xsave_desc_size();
+	en.n_type = NT_X86_XSAVE_LAYOUT;
+
+	if (!dump_emit(cprm, &en, sizeof(en)))
+		return 1;
+	if (!dump_emit(cprm, owner_name, en.n_namesz))
+		return 1;
+	if (!dump_align(cprm, 4))
+		return 1;
+
+	num_records = dump_xsave_layout_desc(cprm);
+	if (!num_records)
+		return 1;
+
+	/* Total size should be equal to the number of records */
+	if ((sizeof(struct x86_xfeat_component) * num_records) != en.n_descsz)
+		return 1;
+
+	return 0;
+}
+
+int elf_coredump_extra_notes_size(void)
+{
+	int size;
+
+	if (!fpu_user_cfg.max_features)
+		return 0;
+
+	/* .note header */
+	size  = sizeof(struct elf_note);
+	/*  Name plus alignment to 4 bytes */
+	size += roundup(sizeof(owner_name), 4);
+	size += get_xsave_desc_size();
+
+	return size;
+}
+#endif /* CONFIG_COREDUMP */

fs/binfmt_elf.c

@@ -2039,7 +2039,7 @@ static int elf_core_dump(struct coredump_params *cprm)
 	{
 		size_t sz = info.size;
 
-		/* For cell spufs */
+		/* For cell spufs and x86 xstate */
 		sz += elf_coredump_extra_notes_size();
 
 		phdr4note = kmalloc(sizeof(*phdr4note), GFP_KERNEL);
@@ -2103,7 +2103,7 @@ static int elf_core_dump(struct coredump_params *cprm)
 	if (!write_note_info(&info, cprm))
 		goto end_coredump;
 
-	/* For cell spufs */
+	/* For cell spufs and x86 xstate */
 	if (elf_coredump_extra_notes_write(cprm))
 		goto end_coredump;
 

include/uapi/linux/elf.h

@@ -411,6 +411,7 @@ typedef struct elf64_shdr {
 #define NT_X86_XSTATE	0x202		/* x86 extended state using xsave */
 /* Old binutils treats 0x203 as a CET state */
 #define NT_X86_SHSTK	0x204		/* x86 SHSTK state */
+#define NT_X86_XSAVE_LAYOUT	0x205	/* XSAVE layout description */
 #define NT_S390_HIGH_GPRS	0x300	/* s390 upper register halves */
 #define NT_S390_TIMER	0x301		/* s390 timer register */
 #define NT_S390_TODCMP	0x302		/* s390 TOD clock comparator register */