From: Patrick Roy <roypat@amazon.co.uk>
To: <seanjc@google.com>, <pbonzini@redhat.com>,
<akpm@linux-foundation.org>, <dwmw@amazon.co.uk>,
<rppt@kernel.org>, <david@redhat.com>
Cc: Patrick Roy <roypat@amazon.co.uk>, <tglx@linutronix.de>,
<mingo@redhat.com>, <bp@alien8.de>, <dave.hansen@linux.intel.com>,
<x86@kernel.org>, <hpa@zytor.com>, <willy@infradead.org>,
<graf@amazon.com>, <derekmn@amazon.com>, <kalyazin@amazon.com>,
<kvm@vger.kernel.org>, <linux-kernel@vger.kernel.org>,
<linux-mm@kvack.org>, <dmatlack@google.com>, <tabba@google.com>,
<chao.p.peng@linux.intel.com>, <xmarcalx@amazon.co.uk>
Subject: [RFC PATCH 3/8] kvm: pfncache: enlighten about gmem
Date: Tue, 9 Jul 2024 14:20:31 +0100 [thread overview]
Message-ID: <20240709132041.3625501-4-roypat@amazon.co.uk> (raw)
In-Reply-To: <20240709132041.3625501-1-roypat@amazon.co.uk>
KVM uses gfn_to_pfn_caches to cache translations from gfn all the way to
the pfn (for example, kvm-clock caches the page storing the page used
for guest/host communication this way). Unlike the gfn_to_hva_cache,
where no equivalent caching semantics were possible to gmem-backed gfns
(see also 858e8068a750 ("kvm: pfncache: enlighten about gmem")), here it
is possible to simply cache the pfn returned by `kvm_gmem_get_pfn`.
Additionally, gfn_to_pfn_caches now invalidate whenever a cached gfn's
attributes are flipped from shared to private (or vice-versa).
Signed-off-by: Patrick Roy <roypat@amazon.co.uk>
---
include/linux/kvm_types.h | 1 +
virt/kvm/pfncache.c | 41 +++++++++++++++++++++++++++++++++------
2 files changed, 36 insertions(+), 6 deletions(-)
diff --git a/include/linux/kvm_types.h b/include/linux/kvm_types.h
index 827ecc0b7e10..8f85f01f6bb0 100644
--- a/include/linux/kvm_types.h
+++ b/include/linux/kvm_types.h
@@ -70,6 +70,7 @@ struct gfn_to_pfn_cache {
kvm_pfn_t pfn;
bool active;
bool valid;
+ bool is_private;
};
#ifdef KVM_ARCH_NR_OBJS_PER_MEMORY_CACHE
diff --git a/virt/kvm/pfncache.c b/virt/kvm/pfncache.c
index f0039efb9e1e..6430e0a49558 100644
--- a/virt/kvm/pfncache.c
+++ b/virt/kvm/pfncache.c
@@ -90,6 +90,9 @@ bool kvm_gpc_check(struct gfn_to_pfn_cache *gpc, unsigned long len)
if (!kvm_gpc_is_valid_len(gpc->gpa, gpc->uhva, len))
return false;
+ if (gpc->is_private != kvm_mem_is_private(gpc->kvm, gpa_to_gfn(gpc->gpa)))
+ return false;
+
if (!gpc->valid)
return false;
@@ -159,6 +162,7 @@ static kvm_pfn_t hva_to_pfn_retry(struct gfn_to_pfn_cache *gpc)
kvm_pfn_t new_pfn = KVM_PFN_ERR_FAULT;
void *new_khva = NULL;
unsigned long mmu_seq;
+ gfn_t gfn;
lockdep_assert_held(&gpc->refresh_lock);
@@ -173,6 +177,7 @@ static kvm_pfn_t hva_to_pfn_retry(struct gfn_to_pfn_cache *gpc)
do {
mmu_seq = gpc->kvm->mmu_invalidate_seq;
+ gfn = gpa_to_gfn(gpc->gpa);
smp_rmb();
write_unlock_irq(&gpc->lock);
@@ -197,10 +202,19 @@ static kvm_pfn_t hva_to_pfn_retry(struct gfn_to_pfn_cache *gpc)
cond_resched();
}
- /* We always request a writeable mapping */
- new_pfn = hva_to_pfn(gpc->uhva, false, false, NULL, true, NULL);
- if (is_error_noslot_pfn(new_pfn))
- goto out_error;
+ if (gpc->is_private) {
+ int r = kvm_gmem_get_pfn(gpc->kvm, gfn_to_memslot(gpc->kvm, gfn), gfn,
+ &new_pfn, NULL);
+
+ if (r)
+ goto out_error;
+ } else {
+ /* We always request a writeable mapping */
+ new_pfn = hva_to_pfn(gpc->uhva, false, false, NULL,
+ true, NULL);
+ if (is_error_noslot_pfn(new_pfn))
+ goto out_error;
+ }
/*
* Obtain a new kernel mapping if KVM itself will access the
@@ -252,6 +266,7 @@ static int __kvm_gpc_refresh(struct gfn_to_pfn_cache *gpc, gpa_t gpa, unsigned l
unsigned long old_uhva;
kvm_pfn_t old_pfn;
bool hva_change = false;
+ bool old_private;
void *old_khva;
int ret;
@@ -271,8 +286,21 @@ static int __kvm_gpc_refresh(struct gfn_to_pfn_cache *gpc, gpa_t gpa, unsigned l
old_pfn = gpc->pfn;
old_khva = (void *)PAGE_ALIGN_DOWN((uintptr_t)gpc->khva);
old_uhva = PAGE_ALIGN_DOWN(gpc->uhva);
+ old_private = gpc->is_private;
+
+ gpc->is_private = kvm_mem_is_private(gpc->kvm, gpa_to_gfn(gpa));
+
+ if (gpc->is_private && !kvm_can_access_gmem(gpc->kvm)) {
+ ret = -EFAULT;
+ goto out_unlock;
+ }
if (kvm_is_error_gpa(gpa)) {
+ if (WARN_ON_ONCE(gpc->is_private)) {
+ ret = -EINVAL;
+ goto out_unlock;
+ }
+
page_offset = offset_in_page(uhva);
gpc->gpa = INVALID_GPA;
@@ -316,9 +344,10 @@ static int __kvm_gpc_refresh(struct gfn_to_pfn_cache *gpc, gpa_t gpa, unsigned l
/*
* If the userspace HVA changed or the PFN was already invalid,
- * drop the lock and do the HVA to PFN lookup again.
+ * drop the lock and do the HVA to PFN lookup again. Also
+ * recompute the pfn if the gfn changed from shared to private (or vice-versa).
*/
- if (!gpc->valid || hva_change) {
+ if (!gpc->valid || hva_change || gpc->is_private != old_private) {
ret = hva_to_pfn_retry(gpc);
} else {
/*
--
2.45.2
next prev parent reply other threads:[~2024-07-09 13:21 UTC|newest]
Thread overview: 34+ messages / expand[flat|nested] mbox.gz Atom feed top
2024-07-09 13:20 [RFC PATCH 0/8] Unmapping guest_memfd from Direct Map Patrick Roy
2024-07-09 13:20 ` [RFC PATCH 1/8] kvm: Allow reading/writing gmem using kvm_{read,write}_guest Patrick Roy
2024-07-09 13:20 ` [RFC PATCH 2/8] kvm: use slowpath in gfn_to_hva_cache if memory is private Patrick Roy
2024-07-09 13:20 ` Patrick Roy [this message]
2024-07-09 14:36 ` [RFC PATCH 3/8] kvm: pfncache: enlighten about gmem David Woodhouse
2024-07-10 9:49 ` Patrick Roy
2024-07-10 10:20 ` David Woodhouse
2024-07-10 10:46 ` Patrick Roy
2024-07-10 10:50 ` David Woodhouse
2024-07-09 13:20 ` [RFC PATCH 4/8] kvm: x86: support walking guest page tables in gmem Patrick Roy
2024-07-09 13:20 ` [RFC PATCH 5/8] kvm: gmem: add option to remove guest private memory from direct map Patrick Roy
2024-07-10 7:31 ` Mike Rapoport
2024-07-10 9:50 ` Patrick Roy
2024-07-09 13:20 ` [RFC PATCH 6/8] kvm: gmem: Temporarily restore direct map entries when needed Patrick Roy
2024-07-11 6:25 ` Paolo Bonzini
2024-07-09 13:20 ` [RFC PATCH 7/8] mm: secretmem: use AS_INACCESSIBLE to prohibit GUP Patrick Roy
2024-07-09 21:09 ` David Hildenbrand
2024-07-10 7:32 ` Mike Rapoport
2024-07-10 9:50 ` Patrick Roy
2024-07-10 21:14 ` David Hildenbrand
2024-07-09 13:20 ` [RFC PATCH 8/8] kvm: gmem: Allow restricted userspace mappings Patrick Roy
2024-07-09 14:48 ` Fuad Tabba
2024-07-09 21:13 ` David Hildenbrand
2024-07-10 9:51 ` Patrick Roy
2024-07-10 21:12 ` David Hildenbrand
2024-07-10 21:53 ` Sean Christopherson
2024-07-10 21:56 ` David Hildenbrand
2024-07-12 15:59 ` Patrick Roy
2024-07-30 10:15 ` David Hildenbrand
2024-08-01 10:30 ` Patrick Roy
2024-07-22 12:28 ` [RFC PATCH 0/8] Unmapping guest_memfd from Direct Map Vlastimil Babka (SUSE)
2024-07-26 6:55 ` Patrick Roy
2024-07-30 10:17 ` David Hildenbrand
2024-07-26 16:44 ` Yosry Ahmed
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=20240709132041.3625501-4-roypat@amazon.co.uk \
--to=roypat@amazon.co.uk \
--cc=akpm@linux-foundation.org \
--cc=bp@alien8.de \
--cc=chao.p.peng@linux.intel.com \
--cc=dave.hansen@linux.intel.com \
--cc=david@redhat.com \
--cc=derekmn@amazon.com \
--cc=dmatlack@google.com \
--cc=dwmw@amazon.co.uk \
--cc=graf@amazon.com \
--cc=hpa@zytor.com \
--cc=kalyazin@amazon.com \
--cc=kvm@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=mingo@redhat.com \
--cc=pbonzini@redhat.com \
--cc=rppt@kernel.org \
--cc=seanjc@google.com \
--cc=tabba@google.com \
--cc=tglx@linutronix.de \
--cc=willy@infradead.org \
--cc=x86@kernel.org \
--cc=xmarcalx@amazon.co.uk \
/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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox