From: Vipin Sharma <vipinsh@google.com>
To: maz@kernel.org, oliver.upton@linux.dev, james.morse@arm.com,
suzuki.poulose@arm.com, yuzenghui@huawei.com,
catalin.marinas@arm.com, will@kernel.org, chenhuacai@kernel.org,
aleksandar.qemu.devel@gmail.com, tsbogend@alpha.franken.de,
anup@brainfault.org, atishp@atishpatra.org,
paul.walmsley@sifive.com, palmer@dabbelt.com,
aou@eecs.berkeley.edu, seanjc@google.com, pbonzini@redhat.com,
dmatlack@google.com, ricarkol@google.com
Cc: linux-arm-kernel@lists.infradead.org, kvmarm@lists.linux.dev,
linux-mips@vger.kernel.org, kvm-riscv@lists.infradead.org,
linux-riscv@lists.infradead.org,
linux-kselftest@vger.kernel.org, kvm@vger.kernel.org,
linux-kernel@vger.kernel.org, Vipin Sharma <vipinsh@google.com>
Subject: [PATCH v2 12/16] KVM: arm64: Retry shared page table walks outside of fault handler
Date: Fri, 2 Jun 2023 09:09:10 -0700 [thread overview]
Message-ID: <20230602160914.4011728-13-vipinsh@google.com> (raw)
In-Reply-To: <20230602160914.4011728-1-vipinsh@google.com>
For a shared page walker which is not fault handler, retry the walk if
walker callback function returns -EAGAIN, or continue to the next SPTE
if callback function return -ENOENT. Update the kvm_pgtable_walk
documentation.
For fault handler logic remains same, i.e. exit the walk and resume the
guest when getting -EAGAIN and -ENOENT errors from walker callback
function.
Currently, there is no page walker which is shared and not a fault
handler, but this will change in future patches when clear-dirty-log
walker will use MMU read lock and run via shared walker.
Signed-off-by: Vipin Sharma <vipinsh@google.com>
---
arch/arm64/include/asm/kvm_pgtable.h | 23 ++++++++++-------
arch/arm64/kvm/hyp/pgtable.c | 38 +++++++++++++++++++++++-----
2 files changed, 46 insertions(+), 15 deletions(-)
diff --git a/arch/arm64/include/asm/kvm_pgtable.h b/arch/arm64/include/asm/kvm_pgtable.h
index 23e7e7851f1d..145be12a5fc2 100644
--- a/arch/arm64/include/asm/kvm_pgtable.h
+++ b/arch/arm64/include/asm/kvm_pgtable.h
@@ -716,15 +716,20 @@ int kvm_pgtable_stage2_split(struct kvm_pgtable *pgt, u64 addr, u64 size,
* type of the walker, i.e. shared walker (vCPU fault handlers) or non-shared
* walker.
*
- * Walker Type | Callback | Walker action
- * -------------|------------------|--------------
- * Non-Shared | 0 | Continue
- * Non-Shared | -EAGAIN | Continue
- * Non-Shared | -ENOENT | Continue
- * Non-Shared | Any other | Exit
- * -------------|------------------|--------------
- * Shared | 0 | Continue
- * Shared | Any other | Exit
+ * Walker Type | Callback | Walker action
+ * -----------------------|------------------|--------------
+ * Non-Shared | 0 | Continue
+ * Non-Shared | -EAGAIN | Continue
+ * Non-Shared | -ENOENT | Continue
+ * Non-Shared | Any other | Exit
+ * -----------------------|------------------|--------------
+ * Shared | 0 | Continue
+ * Shared | -EAGAIN | Retry
+ * Shared | -ENOENT | Continue
+ * Shared | Any other | Exit
+ * -----------------------|------------------|--------------
+ * Shared (Fault Handler) | 0 | Continue
+ * Shared (Fault Handler) | Any other | Exit
*
* Return: 0 on success, negative error code on failure.
*/
diff --git a/arch/arm64/kvm/hyp/pgtable.c b/arch/arm64/kvm/hyp/pgtable.c
index 7f80e953b502..23cda3de2dd4 100644
--- a/arch/arm64/kvm/hyp/pgtable.c
+++ b/arch/arm64/kvm/hyp/pgtable.c
@@ -191,15 +191,21 @@ static bool kvm_pgtable_walk_continue(const struct kvm_pgtable_walker *walker,
* Callbacks can also return ENOENT when PTE which is visited is not
* valid.
*
- * In the context of a shared walker interpret these as a signal
+ * In the context of a fault handler interpret these as a signal
* to retry guest execution.
*
- * Ignore these return codes altogether for walkers outside a fault
- * handler (e.g. write protecting a range of memory) and chug along
+ * In the context of a shared walker which is not fault handler
+ * interpret:
+ * 1. EAGAIN - A signal to retry walk again.
+ * 2. ENOENT - A signal to ignore and move on to next SPTE.
+ *
+ * Ignore these return codes altogether for other walkers and chug along
* with the page table walk.
*/
- if (r == -EAGAIN || r == -ENOENT)
+ if (r == -EAGAIN)
return !(walker->flags & KVM_PGTABLE_WALK_SHARED);
+ if (r == -ENOENT)
+ return !(walker->flags & KVM_PGTABLE_WALK_HANDLE_FAULT);
return !r;
}
@@ -260,24 +266,44 @@ static inline int __kvm_pgtable_visit(struct kvm_pgtable_walk_data *data,
return ret;
}
+static bool kvm_pgtable_walk_retry(const struct kvm_pgtable_walker *walker,
+ int r)
+{
+ /*
+ * All shared page table walks where visitor callbacks return -EAGAIN
+ * should be retried with the exception of fault handler. In case of
+ * fault handler retry is achieved by resuming the guest.
+ */
+ if (r == -EAGAIN)
+ return (walker->flags & KVM_PGTABLE_WALK_SHARED) &&
+ !(walker->flags & KVM_PGTABLE_WALK_HANDLE_FAULT);
+
+ return !r;
+}
+
static int __kvm_pgtable_walk(struct kvm_pgtable_walk_data *data,
struct kvm_pgtable_mm_ops *mm_ops, kvm_pteref_t pgtable, u32 level)
{
u32 idx;
int ret = 0;
+ kvm_pteref_t pteref;
if (WARN_ON_ONCE(level >= KVM_PGTABLE_MAX_LEVELS))
return -EINVAL;
for (idx = kvm_pgtable_idx(data, level); idx < PTRS_PER_PTE; ++idx) {
- kvm_pteref_t pteref = &pgtable[idx];
+retry:
+ pteref = &pgtable[idx];
if (data->addr >= data->end)
break;
ret = __kvm_pgtable_visit(data, mm_ops, pteref, level);
- if (ret)
+ if (ret) {
+ if (kvm_pgtable_walk_retry(data->walker, ret))
+ goto retry;
break;
+ }
}
return ret;
--
2.41.0.rc0.172.g3f132b7071-goog
_______________________________________________
linux-riscv mailing list
linux-riscv@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-riscv
next prev parent reply other threads:[~2023-06-02 16:34 UTC|newest]
Thread overview: 21+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-06-02 16:08 [PATCH v2 00/16] Use MMU read lock for clear-dirty-log Vipin Sharma
2023-06-02 16:08 ` [PATCH v2 01/16] KVM: selftests: Clear dirty logs in user defined chunks sizes in dirty_log_perf_test Vipin Sharma
2023-06-02 16:09 ` [PATCH v2 02/16] KVM: selftests: Add optional delay between consecutive clear-dirty-log calls Vipin Sharma
2023-06-02 16:09 ` [PATCH v2 03/16] KVM: selftests: Pass the count of read and write accesses from guest to host Vipin Sharma
2023-06-02 16:09 ` [PATCH v2 04/16] KVM: selftests: Print read-write progress by vCPUs in dirty_log_perf_test Vipin Sharma
2023-06-02 16:09 ` [PATCH v2 05/16] KVM: selftests: Allow independent execution of " Vipin Sharma
2023-06-02 16:09 ` [PATCH v2 06/16] KVM: arm64: Correct the kvm_pgtable_stage2_flush() documentation Vipin Sharma
2023-06-02 16:09 ` [PATCH v2 07/16] KVM: mmu: Move mmu lock/unlock to arch code for clear dirty log Vipin Sharma
2023-06-02 16:09 ` [PATCH v2 08/16] KMV: arm64: Pass page table walker flags to stage2_apply_range_*() Vipin Sharma
2023-06-02 16:09 ` [PATCH v2 09/16] KVM: arm64: Document the page table walker actions based on the callback's return value Vipin Sharma
2023-06-05 14:35 ` Zhi Wang
2023-06-06 17:30 ` Vipin Sharma
2023-06-07 12:37 ` Zhi Wang
2023-06-08 20:17 ` Vipin Sharma
2023-06-02 16:09 ` [PATCH v2 10/16] KVM: arm64: Return -ENOENT if PTE is not valid in stage2_attr_walker Vipin Sharma
2023-06-02 16:09 ` [PATCH v2 11/16] KVM: arm64: Use KVM_PGTABLE_WALK_SHARED flag instead of KVM_PGTABLE_WALK_HANDLE_FAULT Vipin Sharma
2023-06-02 16:09 ` Vipin Sharma [this message]
2023-06-02 16:09 ` [PATCH v2 13/16] KVM: arm64: Run clear-dirty-log under MMU read lock Vipin Sharma
2023-06-02 16:09 ` [PATCH v2 14/16] KVM: arm64: Pass page walker flags from callers of stage 2 split walker Vipin Sharma
2023-06-02 16:09 ` [PATCH v2 15/16] KVM: arm64: Provide option to pass page walker flag for huge page splits Vipin Sharma
2023-06-02 16:09 ` [PATCH v2 16/16] KVM: arm64: Split huge pages during clear-dirty-log under MMU read lock Vipin Sharma
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=20230602160914.4011728-13-vipinsh@google.com \
--to=vipinsh@google.com \
--cc=aleksandar.qemu.devel@gmail.com \
--cc=anup@brainfault.org \
--cc=aou@eecs.berkeley.edu \
--cc=atishp@atishpatra.org \
--cc=catalin.marinas@arm.com \
--cc=chenhuacai@kernel.org \
--cc=dmatlack@google.com \
--cc=james.morse@arm.com \
--cc=kvm-riscv@lists.infradead.org \
--cc=kvm@vger.kernel.org \
--cc=kvmarm@lists.linux.dev \
--cc=linux-arm-kernel@lists.infradead.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-kselftest@vger.kernel.org \
--cc=linux-mips@vger.kernel.org \
--cc=linux-riscv@lists.infradead.org \
--cc=maz@kernel.org \
--cc=oliver.upton@linux.dev \
--cc=palmer@dabbelt.com \
--cc=paul.walmsley@sifive.com \
--cc=pbonzini@redhat.com \
--cc=ricarkol@google.com \
--cc=seanjc@google.com \
--cc=suzuki.poulose@arm.com \
--cc=tsbogend@alpha.franken.de \
--cc=will@kernel.org \
--cc=yuzenghui@huawei.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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).