[PATCH v2 0/5] x86: Switch X86_FEATURE_XENPV to cpu_feature_enabled() use

[PATCH v2 0/5] x86: Switch X86_FEATURE_XENPV to cpu_feature_enabled() use

[Juergen Gross]

Make especially kernels without CONFIG_XEN_PV more efficient by
using cpu_feature_enabled(X86_FEATURE_XENPV) instead of boot_cpu_has()
and friends.

Changes in V2:
- new patch 4

Juergen Gross (5):
  x86: add X86_FEATURE_XENPV to disabled-features.h
  x86: remove unneeded 64-bit dependency in arch_enter_from_user_mode()
  x86: drop 32-bit Xen PV guest code in update_task_stack()
  x86: remove X86_FEATURE_XENPV usage in setup_cpu_entry_area()
  x86: switch to cpu_feature_enabled() for X86_FEATURE_XENPV

 arch/x86/include/asm/disabled-features.h | 8 +++++++-
 arch/x86/include/asm/entry-common.h      | 4 ++--
 arch/x86/include/asm/switch_to.h         | 7 ++-----
 arch/x86/kernel/cpu/amd.c                | 2 +-
 arch/x86/kernel/cpu/bugs.c               | 2 +-
 arch/x86/kernel/cpu/hygon.c              | 2 +-
 arch/x86/kernel/process_64.c             | 4 ++--
 arch/x86/kernel/topology.c               | 2 +-
 arch/x86/mm/cpu_entry_area.c             | 8 ++------
 9 files changed, 19 insertions(+), 20 deletions(-)

-- 
2.35.3

[PATCH v2 1/5] x86: add X86_FEATURE_XENPV to disabled-features.h

[Juergen Gross]

Add X86_FEATURE_XENPV to the features handled specially in
disabled-features.h.

Signed-off-by: Juergen Gross <jgross@suse.com>
---
 arch/x86/include/asm/disabled-features.h | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)

diff --git a/arch/x86/include/asm/disabled-features.h b/arch/x86/include/asm/disabled-features.h
index 33d2cd04d254..c862552d7d6d 100644
--- a/arch/x86/include/asm/disabled-features.h
+++ b/arch/x86/include/asm/disabled-features.h
@@ -81,6 +81,12 @@
 # define DISABLE_SGX	(1 << (X86_FEATURE_SGX & 31))
 #endif
 
+#ifdef CONFIG_XEN_PV
+# define DISABLE_XENPV		0
+#else
+# define DISABLE_XENPV		(1 << (X86_FEATURE_XENPV & 31))
+#endif
+
 #ifdef CONFIG_INTEL_TDX_GUEST
 # define DISABLE_TDX_GUEST	0
 #else
@@ -98,7 +104,7 @@
 #define DISABLED_MASK5	0
 #define DISABLED_MASK6	0
 #define DISABLED_MASK7	(DISABLE_PTI)
-#define DISABLED_MASK8	(DISABLE_TDX_GUEST)
+#define DISABLED_MASK8	(DISABLE_XENPV|DISABLE_TDX_GUEST)
 #define DISABLED_MASK9	(DISABLE_SGX)
 #define DISABLED_MASK10	0
 #define DISABLED_MASK11	(DISABLE_RETPOLINE|DISABLE_RETHUNK|DISABLE_UNRET)
-- 
2.35.3

Re: [PATCH v2 1/5] x86: add X86_FEATURE_XENPV to disabled-features.h

[Dave Hansen]

On 11/4/22 00:26, Juergen Gross wrote:
> Add X86_FEATURE_XENPV to the features handled specially in
> disabled-features.h.

Acked-by: Dave Hansen <dave.hansen@linux.intel.com>

[tip: x86/cpu] x86/cpufeatures: Add X86_FEATURE_XENPV to disabled-features.h

[tip-bot2 for Juergen Gross]

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

Commit-ID:     15e15d64bd8e12d835f6bb1b1ce3ffa13fa03a66
Gitweb:        https://git.kernel.org/tip/15e15d64bd8e12d835f6bb1b1ce3ffa13fa03a66
Author:        Juergen Gross <jgross@suse.com>
AuthorDate:    Fri, 04 Nov 2022 08:26:57 +01:00
Committer:     Borislav Petkov <bp@suse.de>
CommitterDate: Tue, 22 Nov 2022 15:42:33 +01:00

x86/cpufeatures: Add X86_FEATURE_XENPV to disabled-features.h

Add X86_FEATURE_XENPV to the features handled specially in
disabled-features.h.

Signed-off-by: Juergen Gross <jgross@suse.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Acked-by: Dave Hansen <dave.hansen@linux.intel.com>
Link: https://lore.kernel.org/r/20221104072701.20283-2-jgross@suse.com
---
 arch/x86/include/asm/disabled-features.h | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)

diff --git a/arch/x86/include/asm/disabled-features.h b/arch/x86/include/asm/disabled-features.h
index 33d2cd0..c862552 100644
--- a/arch/x86/include/asm/disabled-features.h
+++ b/arch/x86/include/asm/disabled-features.h
@@ -81,6 +81,12 @@
 # define DISABLE_SGX	(1 << (X86_FEATURE_SGX & 31))
 #endif
 
+#ifdef CONFIG_XEN_PV
+# define DISABLE_XENPV		0
+#else
+# define DISABLE_XENPV		(1 << (X86_FEATURE_XENPV & 31))
+#endif
+
 #ifdef CONFIG_INTEL_TDX_GUEST
 # define DISABLE_TDX_GUEST	0
 #else
@@ -98,7 +104,7 @@
 #define DISABLED_MASK5	0
 #define DISABLED_MASK6	0
 #define DISABLED_MASK7	(DISABLE_PTI)
-#define DISABLED_MASK8	(DISABLE_TDX_GUEST)
+#define DISABLED_MASK8	(DISABLE_XENPV|DISABLE_TDX_GUEST)
 #define DISABLED_MASK9	(DISABLE_SGX)
 #define DISABLED_MASK10	0
 #define DISABLED_MASK11	(DISABLE_RETPOLINE|DISABLE_RETHUNK|DISABLE_UNRET)

[PATCH v2 2/5] x86: remove unneeded 64-bit dependency in arch_enter_from_user_mode()

[Juergen Gross]

The check for 64-bit mode when testing X86_FEATURE_XENPV isn't needed,
as Xen PV guests are no longer supported in 32-bit mode.

While at it switch from boot_cpu_has() to cpu_feature_enabled().

Signed-off-by: Juergen Gross <jgross@suse.com>
---
 arch/x86/include/asm/entry-common.h | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/arch/x86/include/asm/entry-common.h b/arch/x86/include/asm/entry-common.h
index 674ed46d3ced..117903881fe4 100644
--- a/arch/x86/include/asm/entry-common.h
+++ b/arch/x86/include/asm/entry-common.h
@@ -24,8 +24,8 @@ static __always_inline void arch_enter_from_user_mode(struct pt_regs *regs)
 		/*
 		 * For !SMAP hardware we patch out CLAC on entry.
 		 */
-		if (boot_cpu_has(X86_FEATURE_SMAP) ||
-		    (IS_ENABLED(CONFIG_64BIT) && boot_cpu_has(X86_FEATURE_XENPV)))
+		if (cpu_feature_enabled(X86_FEATURE_SMAP) ||
+		    cpu_feature_enabled(X86_FEATURE_XENPV))
 			mask |= X86_EFLAGS_AC;
 
 		WARN_ON_ONCE(flags & mask);
-- 
2.35.3

Re: [PATCH v2 2/5] x86: remove unneeded 64-bit dependency in arch_enter_from_user_mode()

[Dave Hansen]

On 11/4/22 00:26, Juergen Gross wrote:
> The check for 64-bit mode when testing X86_FEATURE_XENPV isn't needed,
> as Xen PV guests are no longer supported in 32-bit mode.

Tiniest nit if you revise this: please call out that this "no longer
supported" thing is enforced in Kconfig.  It's not just some random
edict where we pronounced 32-bit PV guests unsupported.

Otherwise:

Acked-by: Dave Hansen <dave.hansen@linux.intel.com>

[tip: x86/cpu] x86/cpu: Remove unneeded 64-bit dependency in arch_enter_from_user_mode()

[tip-bot2 for Juergen Gross]

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

Commit-ID:     0bafc51babe2344e9b0ff6629a4f044727728349
Gitweb:        https://git.kernel.org/tip/0bafc51babe2344e9b0ff6629a4f044727728349
Author:        Juergen Gross <jgross@suse.com>
AuthorDate:    Fri, 04 Nov 2022 08:26:58 +01:00
Committer:     Borislav Petkov <bp@suse.de>
CommitterDate: Tue, 22 Nov 2022 16:11:57 +01:00

x86/cpu: Remove unneeded 64-bit dependency in arch_enter_from_user_mode()

The check for 64-bit mode when testing X86_FEATURE_XENPV isn't needed,
as Xen PV guests are no longer supported in 32-bit mode, see

  a13f2ef168cb ("x86/xen: remove 32-bit Xen PV guest support").

While at it switch from boot_cpu_has() to cpu_feature_enabled().

Signed-off-by: Juergen Gross <jgross@suse.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Acked-by: Dave Hansen <dave.hansen@linux.intel.com>
Link: https://lore.kernel.org/r/20221104072701.20283-3-jgross@suse.com
---
 arch/x86/include/asm/entry-common.h | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/arch/x86/include/asm/entry-common.h b/arch/x86/include/asm/entry-common.h
index 674ed46..1179038 100644
--- a/arch/x86/include/asm/entry-common.h
+++ b/arch/x86/include/asm/entry-common.h
@@ -24,8 +24,8 @@ static __always_inline void arch_enter_from_user_mode(struct pt_regs *regs)
 		/*
 		 * For !SMAP hardware we patch out CLAC on entry.
 		 */
-		if (boot_cpu_has(X86_FEATURE_SMAP) ||
-		    (IS_ENABLED(CONFIG_64BIT) && boot_cpu_has(X86_FEATURE_XENPV)))
+		if (cpu_feature_enabled(X86_FEATURE_SMAP) ||
+		    cpu_feature_enabled(X86_FEATURE_XENPV))
 			mask |= X86_EFLAGS_AC;
 
 		WARN_ON_ONCE(flags & mask);

[PATCH v2 3/5] x86: drop 32-bit Xen PV guest code in update_task_stack()

[Juergen Gross]

Testing for Xen PV guest mode in a 32-bit only code section can be
dropped, as Xen PV guests are supported in 64-bit mode only.

While at it switch from boot_cpu_has() to cpu_feature_enabled() in the
64-bit part of the code.

Signed-off-by: Juergen Gross <jgross@suse.com>
---
 arch/x86/include/asm/switch_to.h | 7 ++-----
 1 file changed, 2 insertions(+), 5 deletions(-)

diff --git a/arch/x86/include/asm/switch_to.h b/arch/x86/include/asm/switch_to.h
index c08eb0fdd11f..5c91305d09d2 100644
--- a/arch/x86/include/asm/switch_to.h
+++ b/arch/x86/include/asm/switch_to.h
@@ -66,13 +66,10 @@ static inline void update_task_stack(struct task_struct *task)
 {
 	/* sp0 always points to the entry trampoline stack, which is constant: */
 #ifdef CONFIG_X86_32
-	if (static_cpu_has(X86_FEATURE_XENPV))
-		load_sp0(task->thread.sp0);
-	else
-		this_cpu_write(cpu_tss_rw.x86_tss.sp1, task->thread.sp0);
+	this_cpu_write(cpu_tss_rw.x86_tss.sp1, task->thread.sp0);
 #else
 	/* Xen PV enters the kernel on the thread stack. */
-	if (static_cpu_has(X86_FEATURE_XENPV))
+	if (cpu_feature_enabled(X86_FEATURE_XENPV))
 		load_sp0(task_top_of_stack(task));
 #endif
 }
-- 
2.35.3

Re: [PATCH v2 3/5] x86: drop 32-bit Xen PV guest code in update_task_stack()

[Dave Hansen]

On 11/4/22 00:26, Juergen Gross wrote:
> Testing for Xen PV guest mode in a 32-bit only code section can be
> dropped, as Xen PV guests are supported in 64-bit mode only.
> 
> While at it switch from boot_cpu_has() to cpu_feature_enabled() in the
> 64-bit part of the code.

Acked-by: Dave Hansen <dave.hansen@linux.intel.com>

[tip: x86/cpu] x86/cpu: Drop 32-bit Xen PV guest code in update_task_stack()

[tip-bot2 for Juergen Gross]

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

Commit-ID:     dfbd9e4059c4edad4d92ef7f3deb4954f76e4ba0
Gitweb:        https://git.kernel.org/tip/dfbd9e4059c4edad4d92ef7f3deb4954f76e4ba0
Author:        Juergen Gross <jgross@suse.com>
AuthorDate:    Fri, 04 Nov 2022 08:26:59 +01:00
Committer:     Borislav Petkov <bp@suse.de>
CommitterDate: Tue, 22 Nov 2022 16:14:15 +01:00

x86/cpu: Drop 32-bit Xen PV guest code in update_task_stack()

Testing for Xen PV guest mode in a 32-bit only code section can be
dropped, as Xen PV guests are supported in 64-bit mode only.

While at it, switch from boot_cpu_has() to cpu_feature_enabled() in the
64-bit part of the code.

Signed-off-by: Juergen Gross <jgross@suse.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Acked-by: Dave Hansen <dave.hansen@linux.intel.com>
Link: https://lore.kernel.org/r/20221104072701.20283-4-jgross@suse.com
---
 arch/x86/include/asm/switch_to.h | 7 ++-----
 1 file changed, 2 insertions(+), 5 deletions(-)

diff --git a/arch/x86/include/asm/switch_to.h b/arch/x86/include/asm/switch_to.h
index c08eb0f..5c91305 100644
--- a/arch/x86/include/asm/switch_to.h
+++ b/arch/x86/include/asm/switch_to.h
@@ -66,13 +66,10 @@ static inline void update_task_stack(struct task_struct *task)
 {
 	/* sp0 always points to the entry trampoline stack, which is constant: */
 #ifdef CONFIG_X86_32
-	if (static_cpu_has(X86_FEATURE_XENPV))
-		load_sp0(task->thread.sp0);
-	else
-		this_cpu_write(cpu_tss_rw.x86_tss.sp1, task->thread.sp0);
+	this_cpu_write(cpu_tss_rw.x86_tss.sp1, task->thread.sp0);
 #else
 	/* Xen PV enters the kernel on the thread stack. */
-	if (static_cpu_has(X86_FEATURE_XENPV))
+	if (cpu_feature_enabled(X86_FEATURE_XENPV))
 		load_sp0(task_top_of_stack(task));
 #endif
 }

[PATCH v2 4/5] x86: remove X86_FEATURE_XENPV usage in setup_cpu_entry_area()

[Juergen Gross]

Testing of X86_FEATURE_XENPV in setup_cpu_entry_area() can be removed,
as this code path is 32-bit only, and Xen PV guests are not supporting
32-bit mode.

Signed-off-by: Juergen Gross <jgross@suse.com>
---
V2:
- new patch
---
 arch/x86/mm/cpu_entry_area.c | 8 ++------
 1 file changed, 2 insertions(+), 6 deletions(-)

diff --git a/arch/x86/mm/cpu_entry_area.c b/arch/x86/mm/cpu_entry_area.c
index 6c2f1b76a0b6..42cd96e7d733 100644
--- a/arch/x86/mm/cpu_entry_area.c
+++ b/arch/x86/mm/cpu_entry_area.c
@@ -138,17 +138,13 @@ static void __init setup_cpu_entry_area(unsigned int cpu)
 	pgprot_t tss_prot = PAGE_KERNEL_RO;
 #else
 	/*
-	 * On native 32-bit systems, the GDT cannot be read-only because
+	 * On 32-bit systems, the GDT cannot be read-only because
 	 * our double fault handler uses a task gate, and entering through
 	 * a task gate needs to change an available TSS to busy.  If the
 	 * GDT is read-only, that will triple fault.  The TSS cannot be
 	 * read-only because the CPU writes to it on task switches.
-	 *
-	 * On Xen PV, the GDT must be read-only because the hypervisor
-	 * requires it.
 	 */
-	pgprot_t gdt_prot = boot_cpu_has(X86_FEATURE_XENPV) ?
-		PAGE_KERNEL_RO : PAGE_KERNEL;
+	pgprot_t gdt_prot = PAGE_KERNEL;
 	pgprot_t tss_prot = PAGE_KERNEL;
 #endif
 
-- 
2.35.3

Re: [PATCH v2 4/5] x86: remove X86_FEATURE_XENPV usage in setup_cpu_entry_area()

[Dave Hansen]

On 11/4/22 00:27, Juergen Gross wrote:
> Testing of X86_FEATURE_XENPV in setup_cpu_entry_area() can be removed,
> as this code path is 32-bit only, and Xen PV guests are not supporting
> 32-bit mode.

Acked-by: Dave Hansen <dave.hansen@linux.intel.com>

[tip: x86/cpu] x86/cpu: Remove X86_FEATURE_XENPV usage in setup_cpu_entry_area()

[tip-bot2 for Juergen Gross]

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

Commit-ID:     d76c4f7a610ac56c5b06e34258859945e77d190c
Gitweb:        https://git.kernel.org/tip/d76c4f7a610ac56c5b06e34258859945e77d190c
Author:        Juergen Gross <jgross@suse.com>
AuthorDate:    Fri, 04 Nov 2022 08:27:00 +01:00
Committer:     Borislav Petkov <bp@suse.de>
CommitterDate: Tue, 22 Nov 2022 16:16:25 +01:00

x86/cpu: Remove X86_FEATURE_XENPV usage in setup_cpu_entry_area()

Testing of X86_FEATURE_XENPV in setup_cpu_entry_area() can be removed,
as this code path is 32-bit only, and Xen PV guests are 64-bit only.

Signed-off-by: Juergen Gross <jgross@suse.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Acked-by: Dave Hansen <dave.hansen@linux.intel.com>
Link: https://lore.kernel.org/r/20221104072701.20283-5-jgross@suse.com
---
 arch/x86/mm/cpu_entry_area.c | 8 ++------
 1 file changed, 2 insertions(+), 6 deletions(-)

diff --git a/arch/x86/mm/cpu_entry_area.c b/arch/x86/mm/cpu_entry_area.c
index 6c2f1b7..42cd96e 100644
--- a/arch/x86/mm/cpu_entry_area.c
+++ b/arch/x86/mm/cpu_entry_area.c
@@ -138,17 +138,13 @@ static void __init setup_cpu_entry_area(unsigned int cpu)
 	pgprot_t tss_prot = PAGE_KERNEL_RO;
 #else
 	/*
-	 * On native 32-bit systems, the GDT cannot be read-only because
+	 * On 32-bit systems, the GDT cannot be read-only because
 	 * our double fault handler uses a task gate, and entering through
 	 * a task gate needs to change an available TSS to busy.  If the
 	 * GDT is read-only, that will triple fault.  The TSS cannot be
 	 * read-only because the CPU writes to it on task switches.
-	 *
-	 * On Xen PV, the GDT must be read-only because the hypervisor
-	 * requires it.
 	 */
-	pgprot_t gdt_prot = boot_cpu_has(X86_FEATURE_XENPV) ?
-		PAGE_KERNEL_RO : PAGE_KERNEL;
+	pgprot_t gdt_prot = PAGE_KERNEL;
 	pgprot_t tss_prot = PAGE_KERNEL;
 #endif
 

[PATCH v2 5/5] x86: switch to cpu_feature_enabled() for X86_FEATURE_XENPV

[Juergen Gross]

Convert the remaining cases of static_cpu_has(X86_FEATURE_XENPV) and
boot_cpu_has(X86_FEATURE_XENPV) to use cpu_feature_enabled(), allowing
more efficient code in case the kernel is configured without
CONFIG_XEN_PV.

Signed-off-by: Juergen Gross <jgross@suse.com>
---
 arch/x86/kernel/cpu/amd.c    | 2 +-
 arch/x86/kernel/cpu/bugs.c   | 2 +-
 arch/x86/kernel/cpu/hygon.c  | 2 +-
 arch/x86/kernel/process_64.c | 4 ++--
 arch/x86/kernel/topology.c   | 2 +-
 5 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/arch/x86/kernel/cpu/amd.c b/arch/x86/kernel/cpu/amd.c
index 860b60273df3..697fe881e967 100644
--- a/arch/x86/kernel/cpu/amd.c
+++ b/arch/x86/kernel/cpu/amd.c
@@ -985,7 +985,7 @@ static void init_amd(struct cpuinfo_x86 *c)
 			set_cpu_cap(c, X86_FEATURE_3DNOWPREFETCH);
 
 	/* AMD CPUs don't reset SS attributes on SYSRET, Xen does. */
-	if (!cpu_has(c, X86_FEATURE_XENPV))
+	if (!cpu_feature_enabled(X86_FEATURE_XENPV))
 		set_cpu_bug(c, X86_BUG_SYSRET_SS_ATTRS);
 
 	/*
diff --git a/arch/x86/kernel/cpu/bugs.c b/arch/x86/kernel/cpu/bugs.c
index da7c361f47e0..7f78e1527c5e 100644
--- a/arch/x86/kernel/cpu/bugs.c
+++ b/arch/x86/kernel/cpu/bugs.c
@@ -1302,7 +1302,7 @@ static enum spectre_v2_mitigation_cmd __init spectre_v2_parse_cmdline(void)
 		return SPECTRE_V2_CMD_AUTO;
 	}
 
-	if (cmd == SPECTRE_V2_CMD_IBRS && boot_cpu_has(X86_FEATURE_XENPV)) {
+	if (cmd == SPECTRE_V2_CMD_IBRS && cpu_feature_enabled(X86_FEATURE_XENPV)) {
 		pr_err("%s selected but running as XenPV guest. Switching to AUTO select\n",
 		       mitigation_options[i].option);
 		return SPECTRE_V2_CMD_AUTO;
diff --git a/arch/x86/kernel/cpu/hygon.c b/arch/x86/kernel/cpu/hygon.c
index 21fd425088fe..1c27645fd429 100644
--- a/arch/x86/kernel/cpu/hygon.c
+++ b/arch/x86/kernel/cpu/hygon.c
@@ -339,7 +339,7 @@ static void init_hygon(struct cpuinfo_x86 *c)
 	set_cpu_cap(c, X86_FEATURE_ARAT);
 
 	/* Hygon CPUs don't reset SS attributes on SYSRET, Xen does. */
-	if (!cpu_has(c, X86_FEATURE_XENPV))
+	if (!cpu_feature_enabled(X86_FEATURE_XENPV))
 		set_cpu_bug(c, X86_BUG_SYSRET_SS_ATTRS);
 
 	check_null_seg_clears_base(c);
diff --git a/arch/x86/kernel/process_64.c b/arch/x86/kernel/process_64.c
index 6b3418bff326..e2f469175be8 100644
--- a/arch/x86/kernel/process_64.c
+++ b/arch/x86/kernel/process_64.c
@@ -165,7 +165,7 @@ static noinstr unsigned long __rdgsbase_inactive(void)
 
 	lockdep_assert_irqs_disabled();
 
-	if (!static_cpu_has(X86_FEATURE_XENPV)) {
+	if (!cpu_feature_enabled(X86_FEATURE_XENPV)) {
 		native_swapgs();
 		gsbase = rdgsbase();
 		native_swapgs();
@@ -190,7 +190,7 @@ static noinstr void __wrgsbase_inactive(unsigned long gsbase)
 {
 	lockdep_assert_irqs_disabled();
 
-	if (!static_cpu_has(X86_FEATURE_XENPV)) {
+	if (!cpu_feature_enabled(X86_FEATURE_XENPV)) {
 		native_swapgs();
 		wrgsbase(gsbase);
 		native_swapgs();
diff --git a/arch/x86/kernel/topology.c b/arch/x86/kernel/topology.c
index 8617d1ed9d31..1b83377274b8 100644
--- a/arch/x86/kernel/topology.c
+++ b/arch/x86/kernel/topology.c
@@ -106,7 +106,7 @@ int arch_register_cpu(int num)
 	 * Xen PV guests don't support CPU0 hotplug at all.
 	 */
 	if (c->x86_vendor != X86_VENDOR_INTEL ||
-	    boot_cpu_has(X86_FEATURE_XENPV))
+	    cpu_feature_enabled(X86_FEATURE_XENPV))
 		cpu0_hotpluggable = 0;
 
 	/*
-- 
2.35.3

Re: [PATCH v2 5/5] x86: switch to cpu_feature_enabled() for X86_FEATURE_XENPV

[Dave Hansen]

On 11/4/22 00:27, Juergen Gross wrote:
> Convert the remaining cases of static_cpu_has(X86_FEATURE_XENPV) and
> boot_cpu_has(X86_FEATURE_XENPV) to use cpu_feature_enabled(), allowing
> more efficient code in case the kernel is configured without
> CONFIG_XEN_PV.

As with the rest:

Acked-by: Dave Hansen <dave.hansen@linux.intel.com>

Seems like a very straightforward, sane set of cleanups.

[tip: x86/cpu] x86/cpu: Switch to cpu_feature_enabled() for X86_FEATURE_XENPV

[tip-bot2 for Juergen Gross]

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

Commit-ID:     6007878a782eb96f50a71c3a06cf3e931cf8aac1
Gitweb:        https://git.kernel.org/tip/6007878a782eb96f50a71c3a06cf3e931cf8aac1
Author:        Juergen Gross <jgross@suse.com>
AuthorDate:    Fri, 04 Nov 2022 08:27:01 +01:00
Committer:     Borislav Petkov <bp@suse.de>
CommitterDate: Tue, 22 Nov 2022 16:18:19 +01:00

x86/cpu: Switch to cpu_feature_enabled() for X86_FEATURE_XENPV

Convert the remaining cases of static_cpu_has(X86_FEATURE_XENPV) and
boot_cpu_has(X86_FEATURE_XENPV) to use cpu_feature_enabled(), allowing
more efficient code in case the kernel is configured without
CONFIG_XEN_PV.

Signed-off-by: Juergen Gross <jgross@suse.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Acked-by: Dave Hansen <dave.hansen@linux.intel.com>
Link: https://lore.kernel.org/r/20221104072701.20283-6-jgross@suse.com
---
 arch/x86/kernel/cpu/amd.c    | 2 +-
 arch/x86/kernel/cpu/bugs.c   | 2 +-
 arch/x86/kernel/cpu/hygon.c  | 2 +-
 arch/x86/kernel/process_64.c | 4 ++--
 arch/x86/kernel/topology.c   | 2 +-
 5 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/arch/x86/kernel/cpu/amd.c b/arch/x86/kernel/cpu/amd.c
index 860b602..697fe88 100644
--- a/arch/x86/kernel/cpu/amd.c
+++ b/arch/x86/kernel/cpu/amd.c
@@ -985,7 +985,7 @@ static void init_amd(struct cpuinfo_x86 *c)
 			set_cpu_cap(c, X86_FEATURE_3DNOWPREFETCH);
 
 	/* AMD CPUs don't reset SS attributes on SYSRET, Xen does. */
-	if (!cpu_has(c, X86_FEATURE_XENPV))
+	if (!cpu_feature_enabled(X86_FEATURE_XENPV))
 		set_cpu_bug(c, X86_BUG_SYSRET_SS_ATTRS);
 
 	/*
diff --git a/arch/x86/kernel/cpu/bugs.c b/arch/x86/kernel/cpu/bugs.c
index e254f07..9e84b68 100644
--- a/arch/x86/kernel/cpu/bugs.c
+++ b/arch/x86/kernel/cpu/bugs.c
@@ -1302,7 +1302,7 @@ static enum spectre_v2_mitigation_cmd __init spectre_v2_parse_cmdline(void)
 		return SPECTRE_V2_CMD_AUTO;
 	}
 
-	if (cmd == SPECTRE_V2_CMD_IBRS && boot_cpu_has(X86_FEATURE_XENPV)) {
+	if (cmd == SPECTRE_V2_CMD_IBRS && cpu_feature_enabled(X86_FEATURE_XENPV)) {
 		pr_err("%s selected but running as XenPV guest. Switching to AUTO select\n",
 		       mitigation_options[i].option);
 		return SPECTRE_V2_CMD_AUTO;
diff --git a/arch/x86/kernel/cpu/hygon.c b/arch/x86/kernel/cpu/hygon.c
index 21fd425..1c27645 100644
--- a/arch/x86/kernel/cpu/hygon.c
+++ b/arch/x86/kernel/cpu/hygon.c
@@ -339,7 +339,7 @@ static void init_hygon(struct cpuinfo_x86 *c)
 	set_cpu_cap(c, X86_FEATURE_ARAT);
 
 	/* Hygon CPUs don't reset SS attributes on SYSRET, Xen does. */
-	if (!cpu_has(c, X86_FEATURE_XENPV))
+	if (!cpu_feature_enabled(X86_FEATURE_XENPV))
 		set_cpu_bug(c, X86_BUG_SYSRET_SS_ATTRS);
 
 	check_null_seg_clears_base(c);
diff --git a/arch/x86/kernel/process_64.c b/arch/x86/kernel/process_64.c
index 6b3418b..e2f4691 100644
--- a/arch/x86/kernel/process_64.c
+++ b/arch/x86/kernel/process_64.c
@@ -165,7 +165,7 @@ static noinstr unsigned long __rdgsbase_inactive(void)
 
 	lockdep_assert_irqs_disabled();
 
-	if (!static_cpu_has(X86_FEATURE_XENPV)) {
+	if (!cpu_feature_enabled(X86_FEATURE_XENPV)) {
 		native_swapgs();
 		gsbase = rdgsbase();
 		native_swapgs();
@@ -190,7 +190,7 @@ static noinstr void __wrgsbase_inactive(unsigned long gsbase)
 {
 	lockdep_assert_irqs_disabled();
 
-	if (!static_cpu_has(X86_FEATURE_XENPV)) {
+	if (!cpu_feature_enabled(X86_FEATURE_XENPV)) {
 		native_swapgs();
 		wrgsbase(gsbase);
 		native_swapgs();
diff --git a/arch/x86/kernel/topology.c b/arch/x86/kernel/topology.c
index 8617d1e..1b83377 100644
--- a/arch/x86/kernel/topology.c
+++ b/arch/x86/kernel/topology.c
@@ -106,7 +106,7 @@ int arch_register_cpu(int num)
 	 * Xen PV guests don't support CPU0 hotplug at all.
 	 */
 	if (c->x86_vendor != X86_VENDOR_INTEL ||
-	    boot_cpu_has(X86_FEATURE_XENPV))
+	    cpu_feature_enabled(X86_FEATURE_XENPV))
 		cpu0_hotpluggable = 0;
 
 	/*

Re: [PATCH v2 0/5] x86: Switch X86_FEATURE_XENPV to cpu_feature_enabled() use

[Juergen Gross]

[-- Attachment #1.1.1: Type: text/plain, Size: 1265 bytes --]

On 04.11.22 08:26, Juergen Gross wrote:
> Make especially kernels without CONFIG_XEN_PV more efficient by
> using cpu_feature_enabled(X86_FEATURE_XENPV) instead of boot_cpu_has()
> and friends.
> 
> Changes in V2:
> - new patch 4
> 
> Juergen Gross (5):
>    x86: add X86_FEATURE_XENPV to disabled-features.h
>    x86: remove unneeded 64-bit dependency in arch_enter_from_user_mode()
>    x86: drop 32-bit Xen PV guest code in update_task_stack()
>    x86: remove X86_FEATURE_XENPV usage in setup_cpu_entry_area()
>    x86: switch to cpu_feature_enabled() for X86_FEATURE_XENPV
> 
>   arch/x86/include/asm/disabled-features.h | 8 +++++++-
>   arch/x86/include/asm/entry-common.h      | 4 ++--
>   arch/x86/include/asm/switch_to.h         | 7 ++-----
>   arch/x86/kernel/cpu/amd.c                | 2 +-
>   arch/x86/kernel/cpu/bugs.c               | 2 +-
>   arch/x86/kernel/cpu/hygon.c              | 2 +-
>   arch/x86/kernel/process_64.c             | 4 ++--
>   arch/x86/kernel/topology.c               | 2 +-
>   arch/x86/mm/cpu_entry_area.c             | 8 ++------
>   9 files changed, 19 insertions(+), 20 deletions(-)
> 

The patches have all an "Ack" by Dave. I think this series should go in
via the tip tree, no?


Juergen

[-- Attachment #1.1.2: OpenPGP public key --]
[-- Type: application/pgp-keys, Size: 3149 bytes --]

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 495 bytes --]