| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-ID: <op.2lw8gfg2wjvjmi@hhuan26-mobl.amr.corp.intel.com>
Date: Mon, 08 Apr 2024 13:03:41 -0500
From: "Haitao Huang" <haitao.huang@...ux.intel.com>
To: "hpa@...or.com" <hpa@...or.com>, "tim.c.chen@...ux.intel.com"
<tim.c.chen@...ux.intel.com>, "linux-sgx@...r.kernel.org"
<linux-sgx@...r.kernel.org>, "x86@...nel.org" <x86@...nel.org>,
"dave.hansen@...ux.intel.com" <dave.hansen@...ux.intel.com>,
"jarkko@...nel.org" <jarkko@...nel.org>, "cgroups@...r.kernel.org"
<cgroups@...r.kernel.org>, "linux-kernel@...r.kernel.org"
<linux-kernel@...r.kernel.org>, "mkoutny@...e.com" <mkoutny@...e.com>,
"tglx@...utronix.de" <tglx@...utronix.de>, "Mehta, Sohil"
<sohil.mehta@...el.com>, "tj@...nel.org" <tj@...nel.org>, "mingo@...hat.com"
<mingo@...hat.com>, "bp@...en8.de" <bp@...en8.de>, "Huang, Kai"
<kai.huang@...el.com>
Cc: "mikko.ylinen@...ux.intel.com" <mikko.ylinen@...ux.intel.com>,
"seanjc@...gle.com" <seanjc@...gle.com>, "anakrish@...rosoft.com"
<anakrish@...rosoft.com>, "Zhang, Bo" <zhanb@...rosoft.com>,
"kristen@...ux.intel.com" <kristen@...ux.intel.com>, "yangjie@...rosoft.com"
<yangjie@...rosoft.com>, "Li, Zhiquan1" <zhiquan1.li@...el.com>,
"chrisyan@...rosoft.com" <chrisyan@...rosoft.com>
Subject: Re: [PATCH v10 12/14] x86/sgx: Turn on per-cgroup EPC reclamation
On Mon, 08 Apr 2024 07:20:23 -0500, Huang, Kai <kai.huang@...el.com> wrote:
>
>> --- a/arch/x86/kernel/cpu/sgx/epc_cgroup.h
>> +++ b/arch/x86/kernel/cpu/sgx/epc_cgroup.h
>> @@ -28,6 +28,10 @@ static inline int sgx_cgroup_try_charge(struct
>> sgx_cgroup *sgx_cg, enum sgx_recl
>> static inline void sgx_cgroup_uncharge(struct sgx_cgroup *sgx_cg) { }
>>
>> static inline void sgx_cgroup_init(void) { }
>> +
>> +static inline void sgx_cgroup_reclaim_pages(struct misc_cg *root,
>> struct mm_struct *charge_mm)
>> +{
>> +}
>> #else
>> struct sgx_cgroup {
>> struct misc_cg *cg;
>> @@ -65,6 +69,9 @@ static inline void sgx_put_cg(struct sgx_cgroup
>> *sgx_cg)
>>
>> int sgx_cgroup_try_charge(struct sgx_cgroup *sgx_cg, enum sgx_reclaim
>> reclaim);
>> void sgx_cgroup_uncharge(struct sgx_cgroup *sgx_cg);
>> +bool sgx_cgroup_lru_empty(struct misc_cg *root);
>> +bool sgx_cgroup_should_reclaim(struct sgx_cgroup *sgx_cg);
>> +void sgx_cgroup_reclaim_pages(struct misc_cg *root, struct mm_struct
>> *charge_mm);
>
> Seems the decision to choose whether to implement a stub function for
> some
> function is quite random to me.
>
> My impression is people in general don't like #ifdef in the C file but
> prefer to
> implementing it in the header in some helper function.
>
> I guess you might want to just implement a stub function for each of the
> 3
> functions exposed, so that we can eliminate some #ifdefs in the
> sgx/main.c (see
> below).
>
>> void sgx_cgroup_init(void);
>>
>> #endif
>> diff --git a/arch/x86/kernel/cpu/sgx/main.c
>> b/arch/x86/kernel/cpu/sgx/main.c
>> index 7f92455d957d..68f28ff2d5ef 100644
>> --- a/arch/x86/kernel/cpu/sgx/main.c
>> +++ b/arch/x86/kernel/cpu/sgx/main.c
>> @@ -34,6 +34,16 @@ static struct sgx_epc_lru_list sgx_global_lru;
>>
>> static inline struct sgx_epc_lru_list *sgx_lru_list(struct
>> sgx_epc_page *epc_page)
>> {
>> +#ifdef CONFIG_CGROUP_SGX_EPC
>> + if (epc_page->sgx_cg)
>> + return &epc_page->sgx_cg->lru;
>> +
>> + /*
>> + * This should not happen when cgroup is enabled: Every page belongs
>> + * to a cgroup, or the root by default.
>> + */
>> + WARN_ON_ONCE(1);
>
> In the case MISC cgroup is enabled in Kconfig but disabled by command
> line, I
> think this becomes legal now?
>
I'm not sure actually. Never saw the warning even if I set
"cgroup_disable=misc" in commandlibe. Tried both v1 and v2. Anyway, I
think we can remove this warning and we handle the NULL sgx_cg now.
>> +#endif
>> return &sgx_global_lru;
>> }
>>
>> @@ -42,7 +52,11 @@ static inline struct sgx_epc_lru_list
>> *sgx_lru_list(struct sgx_epc_page *epc_pag
>> */
>> static inline bool sgx_can_reclaim(void)
>> {
>> +#ifdef CONFIG_CGROUP_SGX_EPC
>> + return !sgx_cgroup_lru_empty(misc_cg_root());
>> +#else
>> return !list_empty(&sgx_global_lru.reclaimable);
>> +#endif
>> }
>>
>
> Here you are using #ifdef CONFIG_CGRUP_SGX_EPC, but ...
>
>> static atomic_long_t sgx_nr_free_pages = ATOMIC_LONG_INIT(0);
>> @@ -404,7 +418,10 @@ static bool sgx_should_reclaim(unsigned long
>> watermark)
>>
>> static void sgx_reclaim_pages_global(struct mm_struct *charge_mm)
>> {
>> - sgx_reclaim_pages(&sgx_global_lru, charge_mm);
>> + if (IS_ENABLED(CONFIG_CGROUP_SGX_EPC))
>> + sgx_cgroup_reclaim_pages(misc_cg_root(), charge_mm);
>> + else
>> + sgx_reclaim_pages(&sgx_global_lru, charge_mm);
>> }
>
> ... here you are using IS_ENABLED(CONFIG_CGROUP_SGX_EPC).
>
> Any reason they are not consistent?
I was hesitant to expose sgx_global_lru needed for implementing
sgx_cgroup_lru_empty() stub which would also be a random decision and
overkill to just remove couple of #ifdefs in short functions.
>
> Also, in the case where MISC cgroup is disabled via commandline, I think
> it
> won't work, because misc_cg_root() should be NULL in this case while
> IS_ENABLED(CONFIG_CGROUP_SGX_EPC) is true.
>
>>
The misc root cgroup is a static similar to sgx_cg_root. So
misc_cg_root() won't be NULL
However, based on how css_misc() was check NULL, I suppose
sgx_get_current_cg() may be NULL when cgroup is disabled (again not 100%
sure but we handle it anyway)
>> /*
>> @@ -414,6 +431,16 @@ static void sgx_reclaim_pages_global(struct
>> mm_struct *charge_mm)
>> */
>> void sgx_reclaim_direct(void)
>> {
>> +#ifdef CONFIG_CGROUP_SGX_EPC
>> + struct sgx_cgroup *sgx_cg = sgx_get_current_cg();
>> +
>> + /* Make sure there are some free pages at cgroup level */
>> + if (sgx_cg && sgx_cgroup_should_reclaim(sgx_cg)) {
>> + sgx_cgroup_reclaim_pages(sgx_cg->cg, current->mm);
>> + sgx_put_cg(sgx_cg);
>> + }
>> +#endif
>
> This #ifdef CONFIG_CGROUP_SGX_EPC can be removed if we implement a stub
> function
> for sgx_cgroup_should_reclaim().
>
Yes.
>> + /* Make sure there are some free pages at global level */
>> if (sgx_should_reclaim(SGX_NR_LOW_PAGES))
>> sgx_reclaim_pages_global(current->mm);
>> }
>
Thanks
Haitao
Powered by blists - more mailing lists