* Thu Nov 23 2023 Miroslav Rezanina <mrezanin@redhat.com> - 6.2.0-42

- kvm-target-s390x-dump-Remove-unneeded-dump-info-function.patch [RHEL-16696]
- kvm-dump-Add-arch-cleanup-function.patch [RHEL-16696]
- kvm-target-s390x-arch_dump-Add-arch-cleanup-function-for.patch [RHEL-16696]
- Resolves: RHEL-16696
  (RHEL8 - KVM : Secure execution guest remains in "paused" state, post "virsh dump" failure (qemu-kvm))
This commit is contained in:
Miroslav Rezanina 2023-11-23 03:46:38 -05:00
parent 095070217f
commit a32e67a8e7
4 changed files with 223 additions and 1 deletions

View File

@ -0,0 +1,69 @@
From 837e09b1a8a38b53488f59aad090fbe6bb94e257 Mon Sep 17 00:00:00 2001
From: Thomas Huth <thuth@redhat.com>
Date: Fri, 17 Nov 2023 11:32:37 +0100
Subject: [PATCH 2/3] dump: Add arch cleanup function
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
RH-Author: Thomas Huth <thuth@redhat.com>
RH-MergeRequest: 323: Fix problem that secure execution guest might remain in "paused" state after failed dump
RH-Jira: RHEL-16696
RH-Acked-by: Marc-André Lureau <marcandre.lureau@redhat.com>
RH-Acked-by: Cédric Le Goater <clg@redhat.com>
RH-Commit: [2/3] b70f406dec88ffd4877f3d5d580fc8f821bdb252
JIRA: https://issues.redhat.com/browse/RHEL-16696
commit e72629e5149aba6f44122ea6d2a803ef136a0c6b
Author: Janosch Frank <frankja@linux.ibm.com>
Date: Thu Nov 9 12:04:42 2023 +0000
dump: Add arch cleanup function
Some architectures (s390x) need to cleanup after a failed dump to be
able to continue to run the vm. Add a cleanup function pointer and
call it if it's set.
Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-ID: <20231109120443.185979-3-frankja@linux.ibm.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
---
dump/dump.c | 4 ++++
include/sysemu/dump-arch.h | 1 +
2 files changed, 5 insertions(+)
diff --git a/dump/dump.c b/dump/dump.c
index 5dee060b73..93edb89547 100644
--- a/dump/dump.c
+++ b/dump/dump.c
@@ -100,6 +100,10 @@ uint64_t cpu_to_dump64(DumpState *s, uint64_t val)
static int dump_cleanup(DumpState *s)
{
+ if (s->dump_info.arch_cleanup_fn) {
+ s->dump_info.arch_cleanup_fn(s);
+ }
+
guest_phys_blocks_free(&s->guest_phys_blocks);
memory_mapping_list_free(&s->list);
close(s->fd);
diff --git a/include/sysemu/dump-arch.h b/include/sysemu/dump-arch.h
index 59bbc9be38..743916e46c 100644
--- a/include/sysemu/dump-arch.h
+++ b/include/sysemu/dump-arch.h
@@ -24,6 +24,7 @@ typedef struct ArchDumpInfo {
void (*arch_sections_add_fn)(DumpState *s);
uint64_t (*arch_sections_write_hdr_fn)(DumpState *s, uint8_t *buff);
int (*arch_sections_write_fn)(DumpState *s, uint8_t *buff);
+ void (*arch_cleanup_fn)(DumpState *s);
} ArchDumpInfo;
struct GuestPhysBlockList; /* memory_mapping.h */
--
2.39.3

View File

@ -0,0 +1,84 @@
From 4aa08999f8502e9d6869352db89081319c2d7119 Mon Sep 17 00:00:00 2001
From: Thomas Huth <thuth@redhat.com>
Date: Fri, 17 Nov 2023 11:32:37 +0100
Subject: [PATCH 3/3] target/s390x/arch_dump: Add arch cleanup function for PV
dumps
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
RH-Author: Thomas Huth <thuth@redhat.com>
RH-MergeRequest: 323: Fix problem that secure execution guest might remain in "paused" state after failed dump
RH-Jira: RHEL-16696
RH-Acked-by: Marc-André Lureau <marcandre.lureau@redhat.com>
RH-Acked-by: Cédric Le Goater <clg@redhat.com>
RH-Commit: [3/3] 0bb389c9339b95f7ff6dc284526b0c8d5ef736b4
JIRA: https://issues.redhat.com/browse/RHEL-16696
commit d12a91e0baafce7b1cbacff7cf9339eeb0011732
Author: Janosch Frank <frankja@linux.ibm.com>
Date: Thu Nov 9 12:04:43 2023 +0000
target/s390x/arch_dump: Add arch cleanup function for PV dumps
PV dumps block vcpu runs until dump end is reached. If there's an
error between PV dump init and PV dump end the vm will never be able
to run again. One example of such an error is insufficient disk space
for the dump file.
Let's add a cleanup function that tries to do a dump end. The dump
completion data is discarded but there's no point in writing it to a
file anyway if there's a possibility that other PV dump data is
missing.
Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Claudio Imbrenda <imbrenda@linux.ibm.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-ID: <20231109120443.185979-4-frankja@linux.ibm.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
---
target/s390x/arch_dump.c | 17 +++++++++++++++++
1 file changed, 17 insertions(+)
diff --git a/target/s390x/arch_dump.c b/target/s390x/arch_dump.c
index 7cdd4b7167..3b1f178dc3 100644
--- a/target/s390x/arch_dump.c
+++ b/target/s390x/arch_dump.c
@@ -439,6 +439,22 @@ static int arch_sections_write(DumpState *s, uint8_t *buff)
return 0;
}
+static void arch_cleanup(DumpState *s)
+{
+ g_autofree uint8_t *buff = NULL;
+ int rc;
+
+ if (!pv_dump_initialized) {
+ return;
+ }
+
+ buff = g_malloc(kvm_s390_pv_dmp_get_size_completion_data());
+ rc = kvm_s390_dump_completion_data(buff);
+ if (!rc) {
+ pv_dump_initialized = false;
+ }
+}
+
int cpu_get_dump_info(ArchDumpInfo *info,
const struct GuestPhysBlockList *guest_phys_blocks)
{
@@ -454,6 +470,7 @@ int cpu_get_dump_info(ArchDumpInfo *info,
info->arch_sections_add_fn = *arch_sections_add;
info->arch_sections_write_hdr_fn = *arch_sections_write_hdr;
info->arch_sections_write_fn = *arch_sections_write;
+ info->arch_cleanup_fn = *arch_cleanup;
}
return 0;
}
--
2.39.3

View File

@ -0,0 +1,56 @@
From f647258696cbdce78316b2d9ae513f9ae6f4a0b5 Mon Sep 17 00:00:00 2001
From: Thomas Huth <thuth@redhat.com>
Date: Fri, 17 Nov 2023 11:32:37 +0100
Subject: [PATCH 1/3] target/s390x/dump: Remove unneeded dump info function
pointer init
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
RH-Author: Thomas Huth <thuth@redhat.com>
RH-MergeRequest: 323: Fix problem that secure execution guest might remain in "paused" state after failed dump
RH-Jira: RHEL-16696
RH-Acked-by: Marc-André Lureau <marcandre.lureau@redhat.com>
RH-Acked-by: Cédric Le Goater <clg@redhat.com>
RH-Commit: [1/3] e3b0697ec76274f778fc523efb72f0cbca25cd77
JIRA: https://issues.redhat.com/browse/RHEL-16696
commit 816644b1219900875f47d7adf9bfb283f1b29aa0
Author: Janosch Frank <frankja@linux.ibm.com>
Date: Thu Nov 9 12:04:41 2023 +0000
target/s390x/dump: Remove unneeded dump info function pointer init
dump_state_prepare() now sets the function pointers to NULL so we only
need to touch them if we're going to use them.
Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-ID: <20231109120443.185979-2-frankja@linux.ibm.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
---
target/s390x/arch_dump.c | 4 ----
1 file changed, 4 deletions(-)
diff --git a/target/s390x/arch_dump.c b/target/s390x/arch_dump.c
index a7c44ba49d..7cdd4b7167 100644
--- a/target/s390x/arch_dump.c
+++ b/target/s390x/arch_dump.c
@@ -454,10 +454,6 @@ int cpu_get_dump_info(ArchDumpInfo *info,
info->arch_sections_add_fn = *arch_sections_add;
info->arch_sections_write_hdr_fn = *arch_sections_write_hdr;
info->arch_sections_write_fn = *arch_sections_write;
- } else {
- info->arch_sections_add_fn = NULL;
- info->arch_sections_write_hdr_fn = NULL;
- info->arch_sections_write_fn = NULL;
}
return 0;
}
--
2.39.3

View File

@ -83,7 +83,7 @@ Obsoletes: %1-rhev <= %{epoch}:%{version}-%{release}
Summary: QEMU is a machine emulator and virtualizer Summary: QEMU is a machine emulator and virtualizer
Name: qemu-kvm Name: qemu-kvm
Version: 6.2.0 Version: 6.2.0
Release: 41%{?rcrel}%{?dist} Release: 42%{?rcrel}%{?dist}
# Epoch because we pushed a qemu-1.0 package. AIUI this can't ever be dropped # Epoch because we pushed a qemu-1.0 package. AIUI this can't ever be dropped
Epoch: 15 Epoch: 15
License: GPLv2 and GPLv2+ and CC-BY License: GPLv2 and GPLv2+ and CC-BY
@ -793,6 +793,12 @@ Patch315: kvm-redhat-Update-linux-headers-for-kvm_s390_vm_cpu_uv_f.patch
Patch316: kvm-target-s390x-kvm-Refactor-AP-functionalities.patch Patch316: kvm-target-s390x-kvm-Refactor-AP-functionalities.patch
# For bz#2111390 - [IBM 8.10 FEAT] KVM: Enable Secure Execution Crypto Passthrough - qemu part # For bz#2111390 - [IBM 8.10 FEAT] KVM: Enable Secure Execution Crypto Passthrough - qemu part
Patch317: kvm-target-s390x-AP-passthrough-for-PV-guests.patch Patch317: kvm-target-s390x-AP-passthrough-for-PV-guests.patch
# For RHEL-16696 - RHEL8 - KVM : Secure execution guest remains in "paused" state, post "virsh dump" failure (qemu-kvm)
Patch318: kvm-target-s390x-dump-Remove-unneeded-dump-info-function.patch
# For RHEL-16696 - RHEL8 - KVM : Secure execution guest remains in "paused" state, post "virsh dump" failure (qemu-kvm)
Patch319: kvm-dump-Add-arch-cleanup-function.patch
# For RHEL-16696 - RHEL8 - KVM : Secure execution guest remains in "paused" state, post "virsh dump" failure (qemu-kvm)
Patch320: kvm-target-s390x-arch_dump-Add-arch-cleanup-function-for.patch
BuildRequires: wget BuildRequires: wget
BuildRequires: rpm-build BuildRequires: rpm-build
@ -1962,6 +1968,13 @@ sh %{_sysconfdir}/sysconfig/modules/kvm.modules &> /dev/null || :
%changelog %changelog
* Thu Nov 23 2023 Miroslav Rezanina <mrezanin@redhat.com> - 6.2.0-42
- kvm-target-s390x-dump-Remove-unneeded-dump-info-function.patch [RHEL-16696]
- kvm-dump-Add-arch-cleanup-function.patch [RHEL-16696]
- kvm-target-s390x-arch_dump-Add-arch-cleanup-function-for.patch [RHEL-16696]
- Resolves: RHEL-16696
(RHEL8 - KVM : Secure execution guest remains in "paused" state, post "virsh dump" failure (qemu-kvm))
* Fri Sep 29 2023 Jon Maloy <jmaloy@redhat.com> - 6.2.0-41 * Fri Sep 29 2023 Jon Maloy <jmaloy@redhat.com> - 6.2.0-41
- kvm-s390x-ap-fix-missing-subsystem-reset-registration.patch [bz#2111390] - kvm-s390x-ap-fix-missing-subsystem-reset-registration.patch [bz#2111390]
- kvm-s390x-do-a-subsystem-reset-before-the-unprotect-on-r.patch [bz#2111390] - kvm-s390x-do-a-subsystem-reset-before-the-unprotect-on-r.patch [bz#2111390]