All of lore.kernel.org
 help / color / mirror / Atom feed
From: Catalin Marinas <catalin.marinas@arm.com>
To: Suzuki K Poulose <suzuki.poulose@arm.com>
Cc: "Ryan Roberts" <ryan.roberts@arm.com>,
	"Will Deacon" <will@kernel.org>,
	"Mikołaj Lenczewski" <miko.lenczewski@arm.com>,
	yang@os.amperecomputing.com, corbet@lwn.net,
	jean-philippe@linaro.org, robin.murphy@arm.com, joro@8bytes.org,
	akpm@linux-foundation.org, paulmck@kernel.org,
	mark.rutland@arm.com, joey.gouly@arm.com, maz@kernel.org,
	james.morse@arm.com, broonie@kernel.org, oliver.upton@linux.dev,
	baohua@kernel.org, david@redhat.com, ioworker0@gmail.com,
	jgg@ziepe.ca, nicolinc@nvidia.com, mshavit@google.com,
	jsnitsel@redhat.com, smostafa@google.com, kevin.tian@intel.com,
	linux-doc@vger.kernel.org, linux-kernel@vger.kernel.org,
	linux-arm-kernel@lists.infradead.org, iommu@lists.linux.dev
Subject: Re: [RESEND PATCH v6 1/3] arm64: Add BBM Level 2 cpu feature
Date: Wed, 14 May 2025 13:05:37 +0100	[thread overview]
Message-ID: <aCSHESk1DzShD4vt@arm.com> (raw)
In-Reply-To: <c5a74dfe-68e2-48f1-9bbb-06db8e62ffea@arm.com>

On Tue, May 13, 2025 at 10:15:49AM +0100, Suzuki K Poulose wrote:
> On 12/05/2025 17:33, Catalin Marinas wrote:
> > Stepping back a bit, we know that the MIDR allow-list implies
> > BBML2_NOABORT (and at least BBML2 as in the ID regs). In theory, we need
> 
> Please be aware that BBML2_NOABORT midr list may not always imply BBLM2 in
> ID registers (e.g., AmpereOne. But the plan is to fixup the per cpu
> ID register - struct cpuinfo_arm64 - for such cores at early boot,
> individually, before it is used for sanitisation of the system wide
> copy).

Ah, good point. We can then ignore BBML2 ID regs and only rely on MIDR
(and some future BBML3).

> > So how about we introduce a WEAK_BOOT_CPU_FEATURE which gets enabled by
> > the boot CPU if it has it _but_ cleared by any secondary early CPU if it
> > doesn't (and never enabled by secondary CPUs). When the features are
> > finalised, we know if all early CPUs had it. In combination with
> > PERMITTED_FOR_LATE_CPU, we'd reject late CPUs that don't have it.
> 
> That could work, but it introduces this "clearing" a capability, which
> we don't do at the moment.
> 
> We had an offline discussion about this some time ago, with Mark
> Rutland. The best way to deal with this is to change the way we compute
> capabilities. i.e.,
> 
> 
> 1. Each boot CPU run through all the capabilities and maintain a per-cpu
>    copy of the state.
> 2. System wide capabilities can then be constructed from the all early
>    boot CPU capability state (e.g., ANDing all the state from all CPUs
>    for SCOPE_SYSTEM or ORing for LOCAL_CPU).
> 
> But this requires a drastic change to the infrastructure.

I think it's a lot simpler to achieve the ANDing - set the (system)
capability if detected on the boot CPU, only clear it if missing on
subsequent CPUs. See below on an attempt to introduce this. For lack of
inspiration, I called it ARM64_CPUCAP_GLOBAL_CPU_FEATURE which has both
SCOPE_LOCAL and SCOPE_SYSTEM. It's permitted for late CPUs but not
optional if already enabled. The advantage of having both local&system
is that the match function will be called for both scopes. I added a
mask in to cpucap_default_scope() when calling matches() since so far
no cap had both.

---------------------8<-----------------------------------------
diff --git a/arch/arm64/include/asm/cpufeature.h b/arch/arm64/include/asm/cpufeature.h
index c4326f1cb917..0b0b26a6f27b 100644
--- a/arch/arm64/include/asm/cpufeature.h
+++ b/arch/arm64/include/asm/cpufeature.h
@@ -331,6 +331,15 @@ extern struct arm64_ftr_reg arm64_ftr_reg_ctrel0;
 #define ARM64_CPUCAP_BOOT_CPU_FEATURE                  \
 	(ARM64_CPUCAP_SCOPE_BOOT_CPU | ARM64_CPUCAP_PERMITTED_FOR_LATE_CPU)
 
+/*
+ * CPU feature detected at boot time based on all CPUs. It is safe for a late
+ * CPU to have this feature even though the system hasn't enabled it, although
+ * the feature will not be used by Linux in this case. If the system has
+ * enabled this feature already, then every late CPU must have it.
+ */
+#define ARM64_CPUCAP_GLOBAL_CPU_FEATURE			\
+	 (ARM64_CPUCAP_SCOPE_LOCAL_CPU | ARM64_CPUCAP_SYSTEM_FEATURE)
+
 struct arm64_cpu_capabilities {
 	const char *desc;
 	u16 capability;
@@ -391,6 +400,11 @@ static inline int cpucap_default_scope(const struct arm64_cpu_capabilities *cap)
 	return cap->type & ARM64_CPUCAP_SCOPE_MASK;
 }
 
+static inline bool cpucap_global_scope(const struct arm64_cpu_capabilities *cap)
+{
+	return (cap->type & SCOPE_LOCAL_CPU) && (cap->type & SCOPE_SYSTEM);
+}
+
 /*
  * Generic helper for handling capabilities with multiple (match,enable) pairs
  * of call backs, sharing the same capability bit.
diff --git a/arch/arm64/kernel/cpufeature.c b/arch/arm64/kernel/cpufeature.c
index 5ba149c0c2ac..1a5a51090c0e 100644
--- a/arch/arm64/kernel/cpufeature.c
+++ b/arch/arm64/kernel/cpufeature.c
@@ -3359,13 +3381,47 @@ static void update_cpu_capabilities(u16 scope_mask)
 
 	scope_mask &= ARM64_CPUCAP_SCOPE_MASK;
 	for (i = 0; i < ARM64_NCAPS; i++) {
+		bool global_cap = false;
+
 		caps = cpucap_ptrs[i];
-		if (!caps || !(caps->type & scope_mask) ||
-		    cpus_have_cap(caps->capability) ||
-		    !caps->matches(caps, cpucap_default_scope(caps)))
+		if (!caps || !(caps->type & scope_mask))
 			continue;
 
-		if (caps->desc && !caps->cpus)
+		global_cap = cpucap_global_scope(caps);
+
+		/*
+		 * If it's not a global CPU capability, avoid probing if
+		 * already detected.
+		 */
+		if (!global_cap && cpus_have_cap(caps->capability))
+			continue;
+
+		/*
+		 * Pass the actual scope we are probing to the match function.
+		 * This is important for the global CPU capabilities that are
+		 * checked both as a local CPU feature and as a system one.
+		 */
+		if (!caps->matches(caps,
+				   cpucap_default_scope(caps) & scope_mask)) {
+			/* All CPUs must have the global capability */
+			if (global_cap)
+				__clear_bit(caps->capability, system_cpucaps);
+			continue;
+		}
+
+		/*
+		 * A global capability is only set when probing the boot CPU.
+		 * It may be cleared subsequently if not detected on secondary
+		 * ones.
+		 */
+		if (global_cap && !(scope_mask & SCOPE_BOOT_CPU))
+			continue;
+
+		/*
+		 * Global CPU capabilities are logged later when the system
+		 * capabilities are finalised.
+		 */
+		if (!global_cap && caps->desc && !caps->cpus)
 			pr_info("detected: %s\n", caps->desc);
 
 		__set_bit(caps->capability, system_cpucaps);
@@ -3771,17 +3827,24 @@ static void __init setup_system_capabilities(void)
 	enable_cpu_capabilities(SCOPE_ALL & ~SCOPE_BOOT_CPU);
 	apply_alternatives_all();
 
-	/*
-	 * Log any cpucaps with a cpumask as these aren't logged by
-	 * update_cpu_capabilities().
-	 */
 	for (int i = 0; i < ARM64_NCAPS; i++) {
 		const struct arm64_cpu_capabilities *caps = cpucap_ptrs[i];
 
-		if (caps && caps->cpus && caps->desc &&
-			cpumask_any(caps->cpus) < nr_cpu_ids)
+		if (!caps || !caps->desc)
+			continue;
+
+		/*
+		 * Log any cpucaps with a cpumask as these aren't logged by
+		 * update_cpu_capabilities().
+		 */
+		if (caps->cpus && cpumask_any(caps->cpus) < nr_cpu_ids)
 			pr_info("detected: %s on CPU%*pbl\n",
 				caps->desc, cpumask_pr_args(caps->cpus));
+
+		/* Log global CPU capabilities */
+		if (cpucap_global_scope(caps) &&
+		    cpus_have_cap(caps->capability))
+			pr_info("detected: %s\n", caps->desc);
 	}
 
 	/*
---------------------8<-----------------------------------------

And an dummy test:

---------------------8<-----------------------------------------
diff --git a/arch/arm64/include/asm/cpufeature.h b/arch/arm64/include/asm/cpufeature.h
index 5ba149c0c2ac..1a5a51090c0e 100644
--- a/arch/arm64/kernel/cpufeature.c
+++ b/arch/arm64/kernel/cpufeature.c
@@ -2480,6 +2480,21 @@ test_has_mpam_hcr(const struct arm64_cpu_capabilities *entry, int scope)
 	return idr & MPAMIDR_EL1_HAS_HCR;
 }
 
+static void
+cpu_enable_dummy_global(const struct arm64_cpu_capabilities *entry)
+{
+	pr_info("%s: %s: smp_processor_id() = %d", __func__, entry->desc, smp_processor_id());
+}
+
+static bool
+has_dummy_global(const struct arm64_cpu_capabilities *entry, int scope)
+{
+	pr_info("%s: %s: scope = %x smp_processor_id() = %d", __func__, entry->desc, scope, smp_processor_id());
+	if (smp_processor_id() < 4)
+		return true;
+	return false;
+}
+
 static const struct arm64_cpu_capabilities arm64_features[] = {
 	{
 		.capability = ARM64_ALWAYS_BOOT,
@@ -3050,6 +3065,13 @@ static const struct arm64_cpu_capabilities arm64_features[] = {
 		.matches = has_pmuv3,
 	},
 #endif
+	{
+		.desc = "Dummy test for global CPU feature",
+		.capability = ARM64_HAS_GLOBAL_CPU_TEST,
+		.type = ARM64_CPUCAP_GLOBAL_CPU_FEATURE,
+		.cpu_enable = cpu_enable_dummy_global,
+		.matches = has_dummy_global,
+	},
 	{},
 };
 
diff --git a/arch/arm64/tools/cpucaps b/arch/arm64/tools/cpucaps
index 772c1b008e43..dbc5a3eb5b3d 100644
--- a/arch/arm64/tools/cpucaps
+++ b/arch/arm64/tools/cpucaps
@@ -37,6 +37,7 @@ HAS_GENERIC_AUTH_IMP_DEF
 HAS_GIC_CPUIF_SYSREGS
 HAS_GIC_PRIO_MASKING
 HAS_GIC_PRIO_RELAXED_SYNC
+HAS_GLOBAL_CPU_TEST
 HAS_HCR_NV1
 HAS_HCX
 HAS_LDAPR
---------------------8<-----------------------------------------

-- 
Catalin


  reply	other threads:[~2025-05-14 14:19 UTC|newest]

Thread overview: 28+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-04-28 15:35 [RESEND PATCH v6 0/3] Initial BBML2 support for contpte_convert() Mikołaj Lenczewski
2025-04-28 15:35 ` [RESEND PATCH v6 1/3] arm64: Add BBM Level 2 cpu feature Mikołaj Lenczewski
2025-04-28 17:55   ` ALOK TIWARI
2025-05-06  8:36     ` Mikołaj Lenczewski
2025-05-06 14:25   ` Will Deacon
2025-05-06 14:51     ` Marc Zyngier
2025-05-06 14:57       ` Will Deacon
2025-05-06 14:52     ` Ryan Roberts
2025-05-09 13:49       ` Will Deacon
2025-05-09 14:16         ` Ryan Roberts
2025-05-09 14:28           ` Will Deacon
2025-05-09 14:58             ` Ryan Roberts
2025-05-09 15:59             ` Catalin Marinas
2025-05-09 16:04         ` Catalin Marinas
2025-05-12 13:07           ` Ryan Roberts
2025-05-12 13:24             ` Suzuki K Poulose
2025-05-12 13:35               ` Ryan Roberts
2025-05-12 16:33                 ` Catalin Marinas
2025-05-13  9:15                   ` Suzuki K Poulose
2025-05-14 12:05                     ` Catalin Marinas [this message]
2025-05-19  9:45                       ` Suzuki K Poulose
2025-05-22 15:23                         ` Catalin Marinas
2025-05-22 16:29                           ` Suzuki K Poulose
2025-05-12 17:17                 ` Suzuki K Poulose
2025-04-28 15:35 ` [RESEND PATCH v6 2/3] iommu/arm: Add BBM Level 2 smmu feature Mikołaj Lenczewski
2025-05-06 14:19   ` Will Deacon
2025-04-28 15:35 ` [RESEND PATCH v6 3/3] arm64/mm: Elide tlbi in contpte_convert() under BBML2 Mikołaj Lenczewski
2025-04-28 16:17   ` Ryan Roberts

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=aCSHESk1DzShD4vt@arm.com \
    --to=catalin.marinas@arm.com \
    --cc=akpm@linux-foundation.org \
    --cc=baohua@kernel.org \
    --cc=broonie@kernel.org \
    --cc=corbet@lwn.net \
    --cc=david@redhat.com \
    --cc=iommu@lists.linux.dev \
    --cc=ioworker0@gmail.com \
    --cc=james.morse@arm.com \
    --cc=jean-philippe@linaro.org \
    --cc=jgg@ziepe.ca \
    --cc=joey.gouly@arm.com \
    --cc=joro@8bytes.org \
    --cc=jsnitsel@redhat.com \
    --cc=kevin.tian@intel.com \
    --cc=linux-arm-kernel@lists.infradead.org \
    --cc=linux-doc@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mark.rutland@arm.com \
    --cc=maz@kernel.org \
    --cc=miko.lenczewski@arm.com \
    --cc=mshavit@google.com \
    --cc=nicolinc@nvidia.com \
    --cc=oliver.upton@linux.dev \
    --cc=paulmck@kernel.org \
    --cc=robin.murphy@arm.com \
    --cc=ryan.roberts@arm.com \
    --cc=smostafa@google.com \
    --cc=suzuki.poulose@arm.com \
    --cc=will@kernel.org \
    --cc=yang@os.amperecomputing.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.