[PATCH 0/3] arm64 related PREEMPT_RT patches.

[PATCH 0/3] arm64 related PREEMPT_RT patches.

[Sebastian Andrzej Siewior]

This is what I have for ARM64 in my PREEMPT_RT queue of what I believe
won't change.

#1 + #2 Were posted before and I don't remember the outcome other than
   to opencode the kfree() in #1.

#3 Replaces a warning and Will said that there is a patch in
   "multi-generational LRU series" which gets rid of it. I'm posting it
   again since it isn't -next.

Sebastian



_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

[PATCH 1/3] arm64/sve: Delay freeing memory in fpsimd_flush_thread()

[Sebastian Andrzej Siewior]

fpsimd_flush_thread() invokes kfree() via sve_free()+sme_free() within a
preempt disabled section which is not working on -RT.

Delay freeing of memory until preemption is enabled again.

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
---
 arch/arm64/kernel/fpsimd.c | 17 +++++++++++++++--
 1 file changed, 15 insertions(+), 2 deletions(-)

diff --git a/arch/arm64/kernel/fpsimd.c b/arch/arm64/kernel/fpsimd.c
index 95a733d3b2538..475939beb0167 100644
--- a/arch/arm64/kernel/fpsimd.c
+++ b/arch/arm64/kernel/fpsimd.c
@@ -1562,6 +1562,9 @@ static void fpsimd_flush_thread_vl(enum vec_type type)
 
 void fpsimd_flush_thread(void)
 {
+	void *sve_state = NULL;
+	void *za_state = NULL;
+
 	if (!system_supports_fpsimd())
 		return;
 
@@ -1573,18 +1576,28 @@ void fpsimd_flush_thread(void)
 
 	if (system_supports_sve()) {
 		clear_thread_flag(TIF_SVE);
-		sve_free(current);
+
+		/* Defer kfree() while in atomic context */
+		sve_state = current->thread.sve_state;
+		current->thread.sve_state = NULL;
+
 		fpsimd_flush_thread_vl(ARM64_VEC_SVE);
 	}
 
 	if (system_supports_sme()) {
 		clear_thread_flag(TIF_SME);
-		sme_free(current);
+
+		/* Defer kfree() while in atomic context */
+		za_state = current->thread.za_state;
+		current->thread.za_state = NULL;
+
 		fpsimd_flush_thread_vl(ARM64_VEC_SME);
 		current->thread.svcr = 0;
 	}
 
 	put_cpu_fpsimd_context();
+	kfree(sve_state);
+	kfree(za_state);
 }
 
 /*
-- 
2.36.0


_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

[PATCH 2/3] arm64/sve: Make kernel FPU protection RT friendly

[Sebastian Andrzej Siewior]

Non RT kernels need to protect FPU against preemption and bottom half
processing. This is achieved by disabling bottom halves via
local_bh_disable() which implictly disables preemption.

On RT kernels this protection mechanism is not sufficient because
local_bh_disable() does not disable preemption. It serializes bottom half
related processing via a CPU local lock.

As bottom halves are running always in thread context on RT kernels
disabling preemption is the proper choice as it implicitly prevents bottom
half processing.

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
---
 arch/arm64/kernel/fpsimd.c | 16 ++++++++++++++--
 1 file changed, 14 insertions(+), 2 deletions(-)

diff --git a/arch/arm64/kernel/fpsimd.c b/arch/arm64/kernel/fpsimd.c
index 475939beb0167..ce4ee36b1da88 100644
--- a/arch/arm64/kernel/fpsimd.c
+++ b/arch/arm64/kernel/fpsimd.c
@@ -237,10 +237,19 @@ static void __get_cpu_fpsimd_context(void)
  *
  * The double-underscore version must only be called if you know the task
  * can't be preempted.
+ *
+ * On RT kernels local_bh_disable() is not sufficient because it only
+ * serializes soft interrupt related sections via a local lock, but stays
+ * preemptible. Disabling preemption is the right choice here as bottom
+ * half processing is always in thread context on RT kernels so it
+ * implicitly prevents bottom half processing as well.
  */
 static void get_cpu_fpsimd_context(void)
 {
-	local_bh_disable();
+	if (!IS_ENABLED(CONFIG_PREEMPT_RT))
+		local_bh_disable();
+	else
+		preempt_disable();
 	__get_cpu_fpsimd_context();
 }
 
@@ -261,7 +270,10 @@ static void __put_cpu_fpsimd_context(void)
 static void put_cpu_fpsimd_context(void)
 {
 	__put_cpu_fpsimd_context();
-	local_bh_enable();
+	if (!IS_ENABLED(CONFIG_PREEMPT_RT))
+		local_bh_enable();
+	else
+		preempt_enable();
 }
 
 static bool have_cpu_fpsimd_context(void)
-- 
2.36.0


_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

[PATCH 3/3] arm64: mm: Make arch_faults_on_old_pte() check for migratability

[Sebastian Andrzej Siewior]

From: Valentin Schneider <valentin.schneider@arm.com>

arch_faults_on_old_pte() relies on the calling context being
non-preemptible. CONFIG_PREEMPT_RT turns the PTE lock into a sleepable
spinlock, which doesn't disable preemption once acquired, triggering the
warning in arch_faults_on_old_pte().

It does however disable migration, ensuring the task remains on the same
CPU during the entirety of the critical section, making the read of
cpu_has_hw_af() safe and stable.

Make arch_faults_on_old_pte() check cant_migrate() instead of preemptible().

Cc: Valentin Schneider <vschneid@redhat.com>
Suggested-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Valentin Schneider <valentin.schneider@arm.com>
Link: https://lore.kernel.org/r/20220127192437.1192957-1-valentin.schneider@arm.com
Acked-by: Catalin Marinas <catalin.marinas@arm.com>
---
 arch/arm64/include/asm/pgtable.h | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/arch/arm64/include/asm/pgtable.h b/arch/arm64/include/asm/pgtable.h
index 8ebf1cec5d901..5a53558578ab3 100644
--- a/arch/arm64/include/asm/pgtable.h
+++ b/arch/arm64/include/asm/pgtable.h
@@ -1018,7 +1018,8 @@ static inline void update_mmu_cache(struct vm_area_struct *vma,
  */
 static inline bool arch_faults_on_old_pte(void)
 {
-	WARN_ON(preemptible());
+	/* The register read below requires a stable CPU to make any sense */
+	cant_migrate();
 
 	return !cpu_has_hw_af();
 }
-- 
2.36.0


_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

Re: [PATCH 1/3] arm64/sve: Delay freeing memory in fpsimd_flush_thread()

[Mark Brown]

[-- Attachment #1.1: Type: text/plain, Size: 418 bytes --]

On Thu, May 05, 2022 at 06:32:05PM +0200, Sebastian Andrzej Siewior wrote:
> fpsimd_flush_thread() invokes kfree() via sve_free()+sme_free() within a
> preempt disabled section which is not working on -RT.

Reviewed-by: Mark Brown <broonie@kernel.org>

It'd be better to copy people actively working on the specific code
you're patching as well as the architecture maintainers, it helps things
get noticed for review.

[-- Attachment #1.2: signature.asc --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

[-- Attachment #2: Type: text/plain, Size: 176 bytes --]

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

Re: [PATCH 2/3] arm64/sve: Make kernel FPU protection RT friendly

[Mark Brown]

[-- Attachment #1.1: Type: text/plain, Size: 310 bytes --]

On Thu, May 05, 2022 at 06:32:06PM +0200, Sebastian Andrzej Siewior wrote:
> Non RT kernels need to protect FPU against preemption and bottom half
> processing. This is achieved by disabling bottom halves via
> local_bh_disable() which implictly disables preemption.

Acked-by: Mark Brown <broonie@kernel.org>

[-- Attachment #1.2: signature.asc --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

[-- Attachment #2: Type: text/plain, Size: 176 bytes --]

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

Re: [PATCH 0/3] arm64 related PREEMPT_RT patches.

[Catalin Marinas]

On Thu, 5 May 2022 18:32:04 +0200, Sebastian Andrzej Siewior wrote:
> This is what I have for ARM64 in my PREEMPT_RT queue of what I believe
> won't change.
> 
> #1 + #2 Were posted before and I don't remember the outcome other than
>    to opencode the kfree() in #1.
> 
> #3 Replaces a warning and Will said that there is a patch in
>    "multi-generational LRU series" which gets rid of it. I'm posting it
>    again since it isn't -next.
> 
> [...]

Applied to arm64 (for-next/sme), thanks!

[1/3] arm64/sve: Delay freeing memory in fpsimd_flush_thread()
      https://git.kernel.org/arm64/c/a1259dd80719
[2/3] arm64/sve: Make kernel FPU protection RT friendly
      https://git.kernel.org/arm64/c/696207d4258b

I appled patch 3 as well to for-next/misc but b4 cannot cope with
applying two subsets of a series.

-- 
Catalin


_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel