Re: [PATCH v3 15/22] arm64: capabilities: Change scope of VHE to Boot CPU feature

From: Suzuki K Poulose
Date: Thu Mar 08 2018 - 07:10:32 EST


On 12/02/18 17:17, Dave Martin wrote:
On Fri, Feb 09, 2018 at 05:54:59PM +0000, Suzuki K Poulose wrote:
We expect all CPUs to be running at the same EL inside the kernel
with or without VHE enabled and we have strict checks to ensure
that any mismatch triggers a kernel panic. If VHE is enabled,
we use the feature based on the boot CPU and all other CPUs
should follow. This makes it a perfect candidate for a cpability

capability

based on the boot CPU, which should be matched by all the CPUs
(both when is ON and OFF). This saves us some not-so-pretty
hooks and special code, just for verifying the conflict.

Cc: Marc Zyngier <marc.zyngier@xxxxxxx>
Cc: Dave Martin <dave.martin@xxxxxxx>
Cc: Will Deacon <will.deacon@xxxxxxx>
Signed-off-by: Suzuki K Poulose <suzuki.poulose@xxxxxxx>
---
arch/arm64/include/asm/cpufeature.h | 7 +++++++
arch/arm64/include/asm/virt.h | 6 ------
arch/arm64/kernel/cpufeature.c | 5 +++--
arch/arm64/kernel/smp.c | 38 -------------------------------------
4 files changed, 10 insertions(+), 46 deletions(-)

diff --git a/arch/arm64/include/asm/cpufeature.h b/arch/arm64/include/asm/cpufeature.h
index 5f56a8342065..dfce93f79ae7 100644
--- a/arch/arm64/include/asm/cpufeature.h
+++ b/arch/arm64/include/asm/cpufeature.h
@@ -276,6 +276,13 @@ extern struct arm64_ftr_reg arm64_ftr_reg_ctrel0;
(ARM64_CPUCAP_SCOPE_LOCAL_CPU | \
ARM64_CPUCAP_OPTIONAL_FOR_LATE_CPU)
+/*
+ * Critical CPU feature used early in the boot based on the boot CPU.
+ * The feature should be matched by all booting CPU (both miss and hit
+ * cases).
+ */
+#define ARM64_CPUCAP_CRITICAL_BOOT_CPU_FEATURE ARM64_CPUCAP_SCOPE_BOOT_CPU
+

Nit: would it be consistent with the uses we already have for the word
"strict" to use that word here? i.e.,
ARM64_CPUCAP_STRICT_BOOT_CPU_FEATURE.
Or do you think that would be more confusing?

We don't use the "STRICT" tag anymore. Moreover, I used CRITICAL to indicate
that it is special in a way that all the "late" CPUs (in this case all
secondaries) should match the "state" of the capability (i.e, both ON and OFF)
as that of the boot CPU. I am OK to change it to STRICT.


Otherwise, "critical" sounds a bit like we depend on the capability
being available.

If "strict" doesn't fit though and no other option suggests itself,
it's probably not worth changing this.

struct arm64_cpu_capabilities {
const char *desc;
u16 capability;
diff --git a/arch/arm64/include/asm/virt.h b/arch/arm64/include/asm/virt.h
index c5f89442785c..9d1e24e030b3 100644
--- a/arch/arm64/include/asm/virt.h
+++ b/arch/arm64/include/asm/virt.h
@@ -102,12 +102,6 @@ static inline bool has_vhe(void)
return false;
}
-#ifdef CONFIG_ARM64_VHE
-extern void verify_cpu_run_el(void);
-#else
-static inline void verify_cpu_run_el(void) {}
-#endif
-
#endif /* __ASSEMBLY__ */
#endif /* ! __ASM__VIRT_H */
diff --git a/arch/arm64/kernel/cpufeature.c b/arch/arm64/kernel/cpufeature.c
index 7625e2962e2b..f66e66c79916 100644
--- a/arch/arm64/kernel/cpufeature.c
+++ b/arch/arm64/kernel/cpufeature.c
@@ -1016,11 +1016,13 @@ static const struct arm64_cpu_capabilities arm64_features[] = {
},
#endif /* CONFIG_ARM64_PAN */
{
+#ifdef CONFIG_ARM64_VHE
.desc = "Virtualization Host Extensions",
.capability = ARM64_HAS_VIRT_HOST_EXTN,
- .type = ARM64_CPUCAP_SYSTEM_FEATURE,
+ .type = ARM64_CPUCAP_CRITICAL_BOOT_CPU_FEATURE,
.matches = runs_at_el2,
.cpu_enable = cpu_copy_el2regs,
+#endif

Shouldn't the #ifdef...#endif be outside the { ... },?

Otherwise this yields an empty block that will truncate the list in the
CONFIG_ARM64_VHE case...

Good catch. You're right, I will fix it.



Removal of this block for !CONFIG_ARM64_VHE is a change rather than just
refactoring, so the commit message should explain it.

Ok.

Cheers
Suzuki