Re: [PATCH v2 1/4] x86/resctrl: Enable non-contiguous bits in Intel CAT

[Moger, Babu]

Hi Maciej,

On 9/28/23 02:06, Maciej Wieczór-Retman wrote:
> Hi, thanks for reviewing the series,
> 
> On 2023-09-27 at 17:34:27 -0500, Moger, Babu wrote:
>> Hi Maciej,
>>
>> How about this subject line?
>>
>> x86/resctrl: Enable non-contiguous CBMs on Intel CAT
> 
> Changing "bits" to "CBMs" does indeed seem sensible so I'll do that if
> there are no objections.
> 
> But I'm not sure the preposition collocation change from "in" to "on"
> would be grammatical (at least from what I've read in docs about Intel
> CAT so far).
> 
>> On 9/22/2023 3:48 AM, Maciej Wieczor-Retman wrote:
>>> The setting for non-contiguous 1s support in Intel CAT is
>>
>>> hardcoded to false. On these systems, writing non-contiguous
>>> 1s into the schemata file will fail before resctrl passes
>>> the value to the hardware.
>>>
>>> In Intel CAT CPUID.0x10.1:ECX[3] and CPUID.0x10.2:ECX[3] stopped
>>> being reserved and now carry information about non-contiguous 1s
>>> value support for L3 and L2 cache respectively. The CAT
>>> capacity bitmask (CBM) supports a non-contiguous 1s value if
>>> the bit is set.
>>>
>>> Replace the hardcoded non-contiguous support value with
>>> the support learned from the hardware. Add hardcoded non-contiguous
>>> support value to Haswell probe since it can't make use of CPUID for
>>> Cache allocation.
>>>
>>> Originally-by: Fenghua Yu <fenghua.yu@xxxxxxxxx>
>>> Signed-off-by: Maciej Wieczor-Retman <maciej.wieczor-retman@xxxxxxxxx>
>>> ---
>>> Changelog v2:
>>> - Rewrite part of a comment concerning Haswell. (Reinette)
>>>
>>>   arch/x86/kernel/cpu/resctrl/core.c        |  9 ++++++---
>>>   arch/x86/kernel/cpu/resctrl/ctrlmondata.c | 10 ++++++----
>>>   arch/x86/kernel/cpu/resctrl/internal.h    |  9 +++++++++
>>>   3 files changed, 21 insertions(+), 7 deletions(-)
>>>
>>> diff --git a/arch/x86/kernel/cpu/resctrl/core.c b/arch/x86/kernel/cpu/resctrl/core.c
>>> index 030d3b409768..c783a873147c 100644
>>> --- a/arch/x86/kernel/cpu/resctrl/core.c
>>> +++ b/arch/x86/kernel/cpu/resctrl/core.c
>>> @@ -152,6 +152,7 @@ static inline void cache_alloc_hsw_probe(void)
>>>   	r->cache.cbm_len = 20;
>>>   	r->cache.shareable_bits = 0xc0000;
>>>   	r->cache.min_cbm_bits = 2;
>>> +	r->cache.arch_has_sparse_bitmaps = false;
>>
>> Is this change required?
>>
>> This is always set to false in rdt_init_res_defs_intel().
> 
> The logic behind moving this variable initialization from
> rdt_init_res_defs_intel() into both cache_alloc_hsw_probe() and
> rdt_get_cache_alloc_cfg() is that the variable doesn't really have a
> default value anymore. It used to when the CPUID.0x10.1:ECX[3] and
> CPUID.0x10.2:ECX[3] bits were reserved.
> 
> Now for the general case the variable is dependent on CPUID output.
> And only for Haswell case it needs to be hardcoded to "false", so the
> assignment makes more sense in Haswell probe rather than in the default
> section.

Here is the current sequence order with your change.

1.
resctrl_late_init -> check_quirks -> __check_quirks_intel ->
cache_alloc_hsw_probe
   r->cache.arch_has_sparse_bitmaps = false; (new code)

2. resctrl_late_init -> rdt_init_res_defs -> rdt_init_res_defs_intel
   r->cache.arch_has_sparse_bitmaps = false; (old code)

3. resctrl_late_init -> get_rdt_resources -> get_rdt_alloc_resources ->
rdt_get_cache_alloc_cfg
   r->cache.arch_has_sparse_bitmaps = ecx.split.noncont; (new code)

The code in (3) is going to overwrite whatever is set in (1) or (2).

I would say you can just remove initialization in both (1) and (2). That
makes the code clearer to me. I assume reserved bits in Intel is always 0.

Thanks
Babu


> 
>>>   	r->alloc_capable = true;
>>>   	rdt_alloc_capable = true;
>>> @@ -267,15 +268,18 @@ static void rdt_get_cache_alloc_cfg(int idx, struct rdt_resource *r)
>>>   {
>>>   	struct rdt_hw_resource *hw_res = resctrl_to_arch_res(r);
>>>   	union cpuid_0x10_1_eax eax;
>>> +	union cpuid_0x10_x_ecx ecx;
>>>   	union cpuid_0x10_x_edx edx;
>>> -	u32 ebx, ecx;
>>> +	u32 ebx;
>>> -	cpuid_count(0x00000010, idx, &eax.full, &ebx, &ecx, &edx.full);
>>> +	cpuid_count(0x00000010, idx, &eax.full, &ebx, &ecx.full, &edx.full);
>>>   	hw_res->num_closid = edx.split.cos_max + 1;
>>>   	r->cache.cbm_len = eax.split.cbm_len + 1;
>>>   	r->default_ctrl = BIT_MASK(eax.split.cbm_len + 1) - 1;
>>>   	r->cache.shareable_bits = ebx & r->default_ctrl;
>>>   	r->data_width = (r->cache.cbm_len + 3) / 4;
>>> +	if (boot_cpu_data.x86_vendor == X86_VENDOR_INTEL)
>>> +		r->cache.arch_has_sparse_bitmaps = ecx.split.noncont;
>>>   	r->alloc_capable = true;
>>>   }
>>> @@ -872,7 +876,6 @@ static __init void rdt_init_res_defs_intel(void)
>>>   		if (r->rid == RDT_RESOURCE_L3 ||
>>>   		    r->rid == RDT_RESOURCE_L2) {
>>> -			r->cache.arch_has_sparse_bitmaps = false;
>>
>> Why do you have to remove this one here?   This seems like a right place to
>> initialize.
> 
> Look at the previous comment.
> 

-- 
Thanks
Babu Moger