All of lore.kernel.org
 help / color / mirror / Atom feed
From: Avi Kivity <avi@qumranet.com>
To: Marcelo Tosatti <mtosatti@redhat.com>
Cc: kvm@vger.kernel.org
Subject: Re: [patch 06/13] KVM: MMU: global page keeping
Date: Sun, 07 Sep 2008 12:16:07 +0300	[thread overview]
Message-ID: <48C39BD7.9010803@qumranet.com> (raw)
In-Reply-To: <20080906192430.958862856@localhost.localdomain>

Marcelo Tosatti wrote:
> Account pages that only have global entries. Such pages need to be
> synced on cr4 writes, but not cr3 writes.
>
> @@ -1147,18 +1148,22 @@ struct page *gva_to_page(struct kvm_vcpu
>  static void mmu_set_spte(struct kvm_vcpu *vcpu, u64 *shadow_pte,
>  			 unsigned pt_access, unsigned pte_access,
>  			 int user_fault, int write_fault, int dirty,
> -			 int *ptwrite, int largepage, gfn_t gfn,
> -			 pfn_t pfn, bool speculative)
> +			 int global, int *ptwrite, int largepage,
> +			 gfn_t gfn, pfn_t pfn, bool speculative)
>   

bool global

One option to stop the ever-increasing m_s_s() parameter list is to put 
it into pte_access.  Not sure if it's a good idea.

> ===================================================================
> --- kvm.orig/include/asm-x86/kvm_host.h
> +++ kvm/include/asm-x86/kvm_host.h
> @@ -199,6 +199,7 @@ struct kvm_mmu_page {
>  		u64 *parent_pte;               /* !multimapped */
>  		struct hlist_head parent_ptes; /* multimapped, kvm_pte_chain */
>  	};
> +	unsigned long flags;
>  };
>   

bool global;

(or: bool global : 1; if we run out of space)

>  int kvm_unmap_hva(struct kvm *kvm, unsigned long hva);
>  int kvm_age_hva(struct kvm *kvm, unsigned long hva);
>  
> +enum kvm_page_flags {
> +	KVM_PG_global,
> +};
> +
> +#define KVMPGFLAG(name)							\
> +static inline int kvm_page_##name(struct kvm_mmu_page *sp)		\
> +	{ return test_bit(KVM_PG_##name, &sp->flags); }			\
> +static inline void kvm_set_pg_##name(struct kvm_mmu_page *sp)		\
> +	{ set_bit(KVM_PG_##name, &sp->flags); }				\
> +static inline void kvm_clear_pg_##name(struct kvm_mmu_page *sp)		\
> +	{ clear_bit(KVM_PG_##name, &sp->flags); }			\
> +static inline int kvm_test_clear_pg_##name(struct kvm_mmu_page *sp)	\
> +	{ return test_and_clear_bit(KVM_PG_##name, &sp->flags); }
> +
> +KVMPGFLAG(global);
> +
>   

Ugh.  Note you're using locked operations even though the structure is 
protected by the mmu lock.

-- 
error compiling committee.c: too many arguments to function


  reply	other threads:[~2008-09-07  9:16 UTC|newest]

Thread overview: 42+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2008-09-06 18:48 [patch 00/13] RFC: out of sync shadow Marcelo Tosatti
2008-09-06 18:48 ` [patch 01/13] x86/mm: get_user_pages_fast_atomic Marcelo Tosatti
2008-09-07  8:42   ` Avi Kivity
2008-09-08  6:10     ` Marcelo Tosatti
2008-09-08 14:20       ` Avi Kivity
2008-09-06 18:48 ` [patch 02/13] KVM: MMU: switch to get_user_pages_fast Marcelo Tosatti
2008-09-07  8:45   ` Avi Kivity
2008-09-07 20:44     ` Marcelo Tosatti
2008-09-08 14:53       ` Avi Kivity
2008-09-09 12:21     ` Andrea Arcangeli
2008-09-09 13:57       ` Avi Kivity
2008-09-06 18:48 ` [patch 03/13] KVM: MMU: gfn_to_page_atomic Marcelo Tosatti
2008-09-06 18:48 ` [patch 04/13] KVM: MMU: switch prefetch_page to gfn_to_page_atomic Marcelo Tosatti
2008-09-06 18:48 ` [patch 05/13] KVM: MMU: do not write-protect large mappings Marcelo Tosatti
2008-09-07  9:04   ` Avi Kivity
2008-09-07 20:54     ` Marcelo Tosatti
2008-09-06 18:48 ` [patch 06/13] KVM: MMU: global page keeping Marcelo Tosatti
2008-09-07  9:16   ` Avi Kivity [this message]
2008-09-06 18:48 ` [patch 07/13] KVM: MMU: mode specific sync_page Marcelo Tosatti
2008-09-07  9:52   ` Avi Kivity
2008-09-08  6:03     ` Marcelo Tosatti
2008-09-08  9:50       ` Avi Kivity
2008-09-06 18:48 ` [patch 08/13] KVM: MMU: record guest root level on struct guest_walker Marcelo Tosatti
2008-09-06 18:48 ` [patch 09/13] KVM: MMU: out of sync shadow core Marcelo Tosatti
2008-09-07 11:01   ` Avi Kivity
2008-09-08  7:19     ` Marcelo Tosatti
2008-09-08 14:51       ` Avi Kivity
2008-09-11  8:19         ` Marcelo Tosatti
2008-09-11 13:15     ` Marcelo Tosatti
2008-09-06 18:48 ` [patch 10/13] KVM: MMU: sync roots on mmu reload Marcelo Tosatti
2008-09-06 18:48 ` [patch 11/13] KVM: MMU: sync global pages on cr0/cr4 writes Marcelo Tosatti
2008-09-06 18:48 ` [patch 12/13] KVM: x86: trap invlpg Marcelo Tosatti
2008-09-07 11:14   ` Avi Kivity
2008-09-06 18:48 ` [patch 13/13] KVM: MMU: ignore multiroot when unsyncing global pages Marcelo Tosatti
2008-09-07 11:22 ` [patch 00/13] RFC: out of sync shadow Avi Kivity
2008-09-08  7:23   ` Marcelo Tosatti
2008-09-08 14:56     ` Avi Kivity
2008-09-12  4:05 ` David S. Ahern
2008-09-12 11:51   ` Marcelo Tosatti
2008-09-12 15:12     ` David S. Ahern
2008-09-12 18:09       ` Marcelo Tosatti
2008-09-12 18:19         ` David S. Ahern

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=48C39BD7.9010803@qumranet.com \
    --to=avi@qumranet.com \
    --cc=kvm@vger.kernel.org \
    --cc=mtosatti@redhat.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.