[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <cfa7ef63-feb2-4afd-b05b-76335af888dd@amd.com>
Date: Mon, 21 Oct 2024 09:40:04 -0500
From: "Moger, Babu" <babu.moger@....com>
To: Reinette Chatre <reinette.chatre@...el.com>, corbet@....net,
fenghua.yu@...el.com, tglx@...utronix.de, mingo@...hat.com, bp@...en8.de,
dave.hansen@...ux.intel.com
Cc: x86@...nel.org, hpa@...or.com, paulmck@...nel.org, rdunlap@...radead.org,
tj@...nel.org, peterz@...radead.org, yanjiewtw@...il.com,
kim.phillips@....com, lukas.bulwahn@...il.com, seanjc@...gle.com,
jmattson@...gle.com, leitao@...ian.org, jpoimboe@...nel.org,
rick.p.edgecombe@...el.com, kirill.shutemov@...ux.intel.com,
jithu.joseph@...el.com, kai.huang@...el.com, kan.liang@...ux.intel.com,
daniel.sneddon@...ux.intel.com, pbonzini@...hat.com, sandipan.das@....com,
ilpo.jarvinen@...ux.intel.com, peternewman@...gle.com,
maciej.wieczor-retman@...el.com, linux-doc@...r.kernel.org,
linux-kernel@...r.kernel.org, eranian@...gle.com, james.morse@....com
Subject: Re: [PATCH v8 17/25] x86/resctrl: Add the interface to assign/update
counter assignment
Hi Reinette,
On 10/18/24 10:59, Reinette Chatre wrote:
> Hi Babu,
>
> On 10/17/24 3:56 PM, Moger, Babu wrote:
>> On 10/15/2024 10:25 PM, Reinette Chatre wrote:
>>> On 10/9/24 10:39 AM, Babu Moger wrote:
>
>>>> + */
>>>> +int rdtgroup_assign_cntr_event(struct rdt_resource *r, struct rdtgroup *rdtgrp,
>>>> + struct rdt_mon_domain *d, enum resctrl_event_id evtid)
>>>> +{
>>>> + int index = MBM_EVENT_ARRAY_INDEX(evtid);
>>>> + int cntr_id = rdtgrp->mon.cntr_id[index];
>>>> + int ret;
>>>> +
>>>> + /*
>>>> + * Allocate a new counter id to the event if the counter is not
>>>> + * assigned already.
>>>> + */
>>>> + if (cntr_id == MON_CNTR_UNSET) {
>>>> + cntr_id = mbm_cntr_alloc(r);
>>>> + if (cntr_id < 0) {
>>>> + rdt_last_cmd_puts("Out of MBM assignable counters\n");
>>>> + return -ENOSPC;
>>>> + }
>>>> + rdtgrp->mon.cntr_id[index] = cntr_id;
>>>> + }
>>>> +
>>>> + if (!d) {
>>>> + list_for_each_entry(d, &r->mon_domains, hdr.list) {
>>>> + ret = resctrl_arch_config_cntr(r, d, evtid, rdtgrp->mon.rmid,
>>>> + rdtgrp->closid, cntr_id, true);
>>>> + if (ret)
>>>> + goto out_done_assign;
>>>> +
>>>> + set_bit(cntr_id, d->mbm_cntr_map);
>>>
>>> The code pattern above is repeated four times in this work, twice in
>>> rdtgroup_assign_cntr_event() and twice in rdtgroup_unassign_cntr_event(). This
>>> duplication should be avoided. It can be done in a function that also resets
>>> the architectural state.
>>
>> Are you suggesting to combine rdtgroup_assign_cntr_event() and rdtgroup_unassign_cntr_event()?
>
> No. My comment was about the following pattern that is repeated four times:
> ...
> ret = resctrl_arch_config_cntr(...)
> if (ret)
> ...
> set_bit()/clear_bit()
> ...
>
ok.
>> It can be done. We need a flag to tell if it is a assign or unassign.
>
> There is already a flag that is used by resctrl_arch_config_cntr(), the same parameters
> as resctrl_arch_config_cntr() can be used for a wrapper that just calls
> resctrl_arch_config_cntr() directly and uses that same flag to
> select between set_bit() and clear_bit(). This wrapper can then also include
> the reset of architectural state.
ok. Got it, It will look like this.
+/*
+ * Wrapper to configure the counter in a domain.
+ */
+static int rdtgroup_config_cntr(struct rdt_resource *r,struct
rdt_mon_domain *d,
+ enum resctrl_event_id evtid, u32 rmid, u32
closid,
+ u32 cntr_id, bool assign)
+{
+ int ret;
+
+ ret = resctrl_arch_config_cntr(r, d, evtid, rmid, closid, cntr_id,
assign);
+ if (ret)
+ return ret;
+
+ if (assign)
+ __set_bit(cntr_id, d->mbm_cntr_map);
+ else
+ __clear_bit(cntr_id, d->mbm_cntr_map);
+
+ /*
+ * Reset the architectural state so that reading of hardware
+ * counter is not considered as an overflow in next update.
+ */
+ resctrl_arch_reset_rmid(r, d, closid, rmid, evtid);
+
+ return ret;
+}
+
>
> Also, I do not think we need atomic bitops here so these can be __set_bit()
> and __clear_bit() that also matches how bits of mbm_cntr_free_map are managed
> earlier in series.
>
--
Thanks
Babu Moger
Powered by blists - more mailing lists