These CPU models support the split lock detect feature. Add them to
the x86_cpu_ids match table.

Signed-off-by: Fenghua Yu <fenghua.yu@intel.com>
Reviewed-by: Tony Luck <tony.luck@intel.com>
---
This patch supersedes the earlier:
https://lore.kernel.org/lkml/20220609193549.2725612-1-fenghua.yu@intel.com/
that just added Raptor Lake P.

 arch/x86/kernel/cpu/intel.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/arch/x86/kernel/cpu/intel.c b/arch/x86/kernel/cpu/intel.c
index fd5dead8371c..5afb7d3d248d 100644
--- a/arch/x86/kernel/cpu/intel.c
+++ b/arch/x86/kernel/cpu/intel.c
@@ -1290,6 +1290,8 @@ static const struct x86_cpu_id split_lock_cpu_ids[] __initconst = {
 	X86_MATCH_INTEL_FAM6_MODEL(ALDERLAKE,		1),
 	X86_MATCH_INTEL_FAM6_MODEL(ALDERLAKE_L,		1),
 	X86_MATCH_INTEL_FAM6_MODEL(RAPTORLAKE,		1),
+	X86_MATCH_INTEL_FAM6_MODEL(RAPTORLAKE_P,	1),
+	X86_MATCH_INTEL_FAM6_MODEL(ALDERLAKE_N,		1),
 	{}
 };
 
-- 
2.32.0

I have an alternative patch.

Let's just consider the MSR_IA32_CORE_CAPS_SPLIT_LOCK_DETECT bit to be
architectural.  Are there *ACTUAL* implementations where this is wrong?
 If not, let's just make it part of the architecture.  Intel can then
keep it true forever.

Some documentation which says:

	All processors that enumerate support for MSR_IA32_CORE_CAPS and
	set MSR_IA32_CORE_CAPS_SPLIT_LOCK_DETECT support split lock 	
	detection.

would suffice, I think.  Just look at the diffstat.  What's not to love?

On 7/1/22 06:19, Fenghua Yu wrote:
> These CPU models support the split lock detect feature. Add them to
> the x86_cpu_ids match table.

When is there going to be architectural detection for this feature?  Or,
is the plan to just keep adding CPUs to this table as they are available.

From: Dave Hansen
> Sent: 01 July 2022 15:20
> 
> On 7/1/22 06:19, Fenghua Yu wrote:
> > These CPU models support the split lock detect feature. Add them to
> > the x86_cpu_ids match table.
> 
> When is there going to be architectural detection for this feature?  Or,
> is the plan to just keep adding CPUs to this table as they are available.

I saw that as well, it really doesn't scale at all.

Isn't it usual for the flag to not be settable on old cpu?

In any case the flag could be set and then a 'split lock'
transfer and see if it faults.

OTOH why bother with the test?
Unless you want to report that the 'split lock' detection
doesn't work?

Or did Intel really screw this up and use a flag that has
some other undesirable effect on (at least some) older cpu.

	David

-
Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK
Registration No: 1397386 (Wales)