[pve-devel] [PATCH-SERIES kernel] backport fix for kvm performance regression with Intel Emerald Rapids

[pve-devel] [PATCH-SERIES kernel] backport fix for kvm performance regression with Intel Emerald Rapids

[Fiona Ebner]

Stumbled across this in the news [0].

Built, installed on a (CPU type 'host') VM and started a nested VM for
both 6.8 and 6.11 as a smoke test. Note I have an AMD TR 7960X, but
AFIACT, the patch affects a general code path not specific to Intel
CPUs.

[0]: https://www.phoronix.com/news/Linux-6.13-rc3-Released


Fiona Ebner (2):
  patches: kernel: switch to using full index for patch files
  backport fix for kvm performance regression with Intel Emerald Rapids

 ...-accept-an-alternate-timestamp-strin.patch |   2 +-
 ...d-Debian-wireless-regdb-certificates.patch |   2 +-
 ...idge-keep-MAC-of-first-assigned-port.patch |   2 +-
 ...ides-for-missing-ACS-capabilities-4..patch |   4 +-
 ...-default-dynamic-halt-polling-growth.patch |   2 +-
 ...de-unregister_netdevice-refcount-lea.patch |   2 +-
 ...fortify-Do-not-cast-to-unsigned-char.patch |   2 +-
 ...sk-out-PKRU-bit-in-xfeatures-if-vCPU.patch |   6 +-
 ...allow-pass-through-on-broken-hardwar.patch |   2 +-
 ...-Advertise-support-for-flush-by-ASID.patch |   2 +-
 ...rove-userspace-warnings-for-missing-.patch |   2 +-
 ...pect-msg_namelen-0-for-recvmsg-calls.patch |   2 +-
 ...ix-pagecache-leak-when-do-writepages.patch |   2 +-
 ...pu-pm-Don-t-use-OD-table-on-Arcturus.patch |   2 +-
 ...UCE-iommu-intel-disable-DMAR-for-SKL.patch |   2 +-
 ...t-EOPNOTSUPP-for-IOCB_NOWAIT-like-EA.patch |   2 +-
 ...et-subreq-iov-iter-before-tail-clean.patch |   2 +-
 ...r-virtualized-VMLOAD-VMSAVE-on-Zen4-.patch |   2 +-
 ...UID.0xD-XSTATE-offsets-sizes-during-.patch | 165 ++++++++++++++++++
 19 files changed, 186 insertions(+), 21 deletions(-)
 create mode 100644 patches/kernel/0019-KVM-x86-Cache-CPUID.0xD-XSTATE-offsets-sizes-during-.patch

-- 
2.39.5



_______________________________________________
pve-devel mailing list
pve-devel@lists.proxmox.com
https://lists.proxmox.com/cgi-bin/mailman/listinfo/pve-devel

[pve-devel] [PATCH master kernel 1/4] patches: kernel: switch to using full index for patch files

[Fiona Ebner]

git will automatically change the length of the index hashes when
formatting a patch depending on what references are present in the
submodule. After pulling in the stable tags today, git wanted to add
a character to all hashes for me. Use --full-index when generating the
patches to avoid such issues in the future.

No functional change intended.

Signed-off-by: Fiona Ebner <f.ebner@proxmox.com>
---
 ...ke-mkcompile_h-accept-an-alternate-timestamp-strin.patch | 2 +-
 ...02-wireless-Add-Debian-wireless-regdb-certificates.patch | 2 +-
 .../0003-bridge-keep-MAC-of-first-assigned-port.patch       | 2 +-
 ...i-Enable-overrides-for-missing-ACS-capabilities-4..patch | 4 ++--
 ...05-kvm-disable-default-dynamic-halt-polling-growth.patch | 2 +-
 ...t-core-downgrade-unregister_netdevice-refcount-lea.patch | 2 +-
 .../0007-Revert-fortify-Do-not-cast-to-unsigned-char.patch  | 2 +-
 ...m-xsave-set-mask-out-PKRU-bit-in-xfeatures-if-vCPU.patch | 6 +++---
 ...low-opt-in-to-allow-pass-through-on-broken-hardwar.patch | 2 +-
 .../0010-KVM-nSVM-Advertise-support-for-flush-by-ASID.patch | 2 +-
 ...vert-memfd-improve-userspace-warnings-for-missing-.patch | 2 +-
 ...12-apparmor-expect-msg_namelen-0-for-recvmsg-calls.patch | 2 +-
 ..._uring-rw-treat-EOPNOTSUPP-for-IOCB_NOWAIT-like-EA.patch | 2 +-
 ...0014-netfs-reset-subreq-iov-iter-before-tail-clean.patch | 2 +-
 ...6-CPU-AMD-Clear-virtualized-VMLOAD-VMSAVE-on-Zen4-.patch | 2 +-
 ...-PCI-Fix-pci_enable_acs-support-for-the-ACS-quirks.patch | 2 +-
 ...vert-UBUNTU-SAUCE-iommu-intel-disable-DMAR-for-SKL.patch | 2 +-
 17 files changed, 20 insertions(+), 20 deletions(-)

diff --git a/patches/kernel/0001-Make-mkcompile_h-accept-an-alternate-timestamp-strin.patch b/patches/kernel/0001-Make-mkcompile_h-accept-an-alternate-timestamp-strin.patch
index 2ba3815..1963808 100644
--- a/patches/kernel/0001-Make-mkcompile_h-accept-an-alternate-timestamp-strin.patch
+++ b/patches/kernel/0001-Make-mkcompile_h-accept-an-alternate-timestamp-strin.patch
@@ -21,7 +21,7 @@ Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
  1 file changed, 1 insertion(+), 1 deletion(-)
 
 diff --git a/init/Makefile b/init/Makefile
-index 10b652d33e87..e4dabde27b90 100644
+index 10b652d33e872658f2d8c8ce3a0b45e9a473fd64..e4dabde27b90c12bd72c6239e15509f9a9448d80 100644
 --- a/init/Makefile
 +++ b/init/Makefile
 @@ -29,7 +29,7 @@ preempt-flag-$(CONFIG_PREEMPT_DYNAMIC)	:= PREEMPT_DYNAMIC
diff --git a/patches/kernel/0002-wireless-Add-Debian-wireless-regdb-certificates.patch b/patches/kernel/0002-wireless-Add-Debian-wireless-regdb-certificates.patch
index 4c0ac0f..274b7ff 100644
--- a/patches/kernel/0002-wireless-Add-Debian-wireless-regdb-certificates.patch
+++ b/patches/kernel/0002-wireless-Add-Debian-wireless-regdb-certificates.patch
@@ -19,7 +19,7 @@ Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
 
 diff --git a/net/wireless/certs/debian.hex b/net/wireless/certs/debian.hex
 new file mode 100644
-index 000000000000..c5ab03f8c500
+index 0000000000000000000000000000000000000000..c5ab03f8c500d2f0e5b7931d5790bd22983c3660
 --- /dev/null
 +++ b/net/wireless/certs/debian.hex
 @@ -0,0 +1,1426 @@
diff --git a/patches/kernel/0003-bridge-keep-MAC-of-first-assigned-port.patch b/patches/kernel/0003-bridge-keep-MAC-of-first-assigned-port.patch
index 135d17c..efa47df 100644
--- a/patches/kernel/0003-bridge-keep-MAC-of-first-assigned-port.patch
+++ b/patches/kernel/0003-bridge-keep-MAC-of-first-assigned-port.patch
@@ -19,7 +19,7 @@ Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
  1 file changed, 1 insertion(+), 4 deletions(-)
 
 diff --git a/net/bridge/br_stp_if.c b/net/bridge/br_stp_if.c
-index 75204d36d7f9..1fb5ff73ec1e 100644
+index 75204d36d7f9062306dfc66c3c35448e16257215..1fb5ff73ec1ef3bd79960182c87a0ba312b3635d 100644
 --- a/net/bridge/br_stp_if.c
 +++ b/net/bridge/br_stp_if.c
 @@ -265,10 +265,7 @@ bool br_stp_recalculate_bridge_id(struct net_bridge *br)
diff --git a/patches/kernel/0004-pci-Enable-overrides-for-missing-ACS-capabilities-4..patch b/patches/kernel/0004-pci-Enable-overrides-for-missing-ACS-capabilities-4..patch
index eac3c16..3d1ae40 100644
--- a/patches/kernel/0004-pci-Enable-overrides-for-missing-ACS-capabilities-4..patch
+++ b/patches/kernel/0004-pci-Enable-overrides-for-missing-ACS-capabilities-4..patch
@@ -55,7 +55,7 @@ Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
  2 files changed, 111 insertions(+)
 
 diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt
-index 89f899cccf38..dfe0d1e400e2 100644
+index 89f899cccf3875e0c085d33d1129a9ec0482a953..dfe0d1e400e2c8f62770510cb7ad35c5060535cf 100644
 --- a/Documentation/admin-guide/kernel-parameters.txt
 +++ b/Documentation/admin-guide/kernel-parameters.txt
 @@ -4422,6 +4422,15 @@
@@ -75,7 +75,7 @@ index 89f899cccf38..dfe0d1e400e2 100644
  				Safety option to keep boot IRQs enabled. This
  				should never be necessary.
 diff --git a/drivers/pci/quirks.c b/drivers/pci/quirks.c
-index ba021ba458a7..bc35fd0e9b4b 100644
+index ba021ba458a7d0e5b92a0a87a50ef18805c48246..bc35fd0e9b4beb10edaf71c6a854be7e01aac336 100644
 --- a/drivers/pci/quirks.c
 +++ b/drivers/pci/quirks.c
 @@ -287,6 +287,106 @@ static int __init pci_apply_final_quirks(void)
diff --git a/patches/kernel/0005-kvm-disable-default-dynamic-halt-polling-growth.patch b/patches/kernel/0005-kvm-disable-default-dynamic-halt-polling-growth.patch
index 0cca29c..715603a 100644
--- a/patches/kernel/0005-kvm-disable-default-dynamic-halt-polling-growth.patch
+++ b/patches/kernel/0005-kvm-disable-default-dynamic-halt-polling-growth.patch
@@ -13,7 +13,7 @@ Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
  1 file changed, 1 insertion(+), 1 deletion(-)
 
 diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c
-index cb2b78e92910..ebfb1def4662 100644
+index cb2b78e92910fb1d6ebed0b73cf359e866f0a06b..ebfb1def4662c4c16f672f8b832ee510fa3770cd 100644
 --- a/virt/kvm/kvm_main.c
 +++ b/virt/kvm/kvm_main.c
 @@ -80,7 +80,7 @@ module_param(halt_poll_ns, uint, 0644);
diff --git a/patches/kernel/0006-net-core-downgrade-unregister_netdevice-refcount-lea.patch b/patches/kernel/0006-net-core-downgrade-unregister_netdevice-refcount-lea.patch
index 958b999..15c64ed 100644
--- a/patches/kernel/0006-net-core-downgrade-unregister_netdevice-refcount-lea.patch
+++ b/patches/kernel/0006-net-core-downgrade-unregister_netdevice-refcount-lea.patch
@@ -14,7 +14,7 @@ Signed-off-by: Fabian Grünbichler <f.gruenbichler@proxmox.com>
  1 file changed, 1 insertion(+), 1 deletion(-)
 
 diff --git a/net/core/dev.c b/net/core/dev.c
-index f66e61407883..07ffd47a2010 100644
+index f66e614078832842e903fc2ee792cd89b8a3572f..07ffd47a2010e7735c72588afc097132b062f764 100644
 --- a/net/core/dev.c
 +++ b/net/core/dev.c
 @@ -10672,7 +10672,7 @@ static struct net_device *netdev_wait_allrefs_any(struct list_head *list)
diff --git a/patches/kernel/0007-Revert-fortify-Do-not-cast-to-unsigned-char.patch b/patches/kernel/0007-Revert-fortify-Do-not-cast-to-unsigned-char.patch
index 29c97d6..94e9f87 100644
--- a/patches/kernel/0007-Revert-fortify-Do-not-cast-to-unsigned-char.patch
+++ b/patches/kernel/0007-Revert-fortify-Do-not-cast-to-unsigned-char.patch
@@ -16,7 +16,7 @@ Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
  1 file changed, 1 insertion(+), 1 deletion(-)
 
 diff --git a/include/linux/fortify-string.h b/include/linux/fortify-string.h
-index 0d99bf11d260..fe04f7f93575 100644
+index 0d99bf11d260a3482bbe46e35c7553c0ccfb8b94..fe04f7f9357506baf21a0c3cc070c37f00a24d5c 100644
 --- a/include/linux/fortify-string.h
 +++ b/include/linux/fortify-string.h
 @@ -62,7 +62,7 @@ void __write_overflow_field(size_t avail, size_t wanted) __compiletime_warning("
diff --git a/patches/kernel/0008-kvm-xsave-set-mask-out-PKRU-bit-in-xfeatures-if-vCPU.patch b/patches/kernel/0008-kvm-xsave-set-mask-out-PKRU-bit-in-xfeatures-if-vCPU.patch
index b14e6da..7997372 100644
--- a/patches/kernel/0008-kvm-xsave-set-mask-out-PKRU-bit-in-xfeatures-if-vCPU.patch
+++ b/patches/kernel/0008-kvm-xsave-set-mask-out-PKRU-bit-in-xfeatures-if-vCPU.patch
@@ -78,7 +78,7 @@ Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
  3 files changed, 21 insertions(+)
 
 diff --git a/arch/x86/kvm/cpuid.c b/arch/x86/kvm/cpuid.c
-index 2617be544480..b6892645c4cd 100644
+index 2617be544480aaf78d21d89f5f0e978a0715e0d6..b6892645c4cde238923062e3f71df21d68cc201d 100644
 --- a/arch/x86/kvm/cpuid.c
 +++ b/arch/x86/kvm/cpuid.c
 @@ -269,6 +269,12 @@ static u64 cpuid_get_supported_xcr0(struct kvm_cpuid_entry2 *entries, int nent)
@@ -95,7 +95,7 @@ index 2617be544480..b6892645c4cd 100644
  				       int nent)
  {
 diff --git a/arch/x86/kvm/cpuid.h b/arch/x86/kvm/cpuid.h
-index 41697cca354e..0e46c4555311 100644
+index 41697cca354e6bc3cb7787602d440eb26e630e5b..0e46c4555311d02b6ec9eba7f33b292c3562b0ee 100644
 --- a/arch/x86/kvm/cpuid.h
 +++ b/arch/x86/kvm/cpuid.h
 @@ -32,6 +32,8 @@ int kvm_vcpu_ioctl_get_cpuid2(struct kvm_vcpu *vcpu,
@@ -108,7 +108,7 @@ index 41697cca354e..0e46c4555311 100644
  
  int cpuid_query_maxphyaddr(struct kvm_vcpu *vcpu);
 diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
-index c983c8e434b8..c7c3d04198c6 100644
+index c983c8e434b8b8eadc8571aebeb06c1b694e134f..c7c3d04198c6c7f88f97e122bfe78b857f26cb26 100644
 --- a/arch/x86/kvm/x86.c
 +++ b/arch/x86/kvm/x86.c
 @@ -5633,6 +5633,19 @@ static int kvm_vcpu_ioctl_x86_set_xsave(struct kvm_vcpu *vcpu,
diff --git a/patches/kernel/0009-allow-opt-in-to-allow-pass-through-on-broken-hardwar.patch b/patches/kernel/0009-allow-opt-in-to-allow-pass-through-on-broken-hardwar.patch
index eac42cd..187c00b 100644
--- a/patches/kernel/0009-allow-opt-in-to-allow-pass-through-on-broken-hardwar.patch
+++ b/patches/kernel/0009-allow-opt-in-to-allow-pass-through-on-broken-hardwar.patch
@@ -11,7 +11,7 @@ Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
  1 file changed, 5 insertions(+), 1 deletion(-)
 
 diff --git a/drivers/iommu/intel/iommu.c b/drivers/iommu/intel/iommu.c
-index 555301f45bec..34c8b02ca787 100644
+index 555301f45bec5ef12290a48095c2cfdaf05602db..34c8b02ca7878e0b1803a07f2f6e5c6c0039a9f8 100644
 --- a/drivers/iommu/intel/iommu.c
 +++ b/drivers/iommu/intel/iommu.c
 @@ -228,6 +228,7 @@ EXPORT_SYMBOL_GPL(intel_iommu_enabled);
diff --git a/patches/kernel/0010-KVM-nSVM-Advertise-support-for-flush-by-ASID.patch b/patches/kernel/0010-KVM-nSVM-Advertise-support-for-flush-by-ASID.patch
index 9596220..34a7e40 100644
--- a/patches/kernel/0010-KVM-nSVM-Advertise-support-for-flush-by-ASID.patch
+++ b/patches/kernel/0010-KVM-nSVM-Advertise-support-for-flush-by-ASID.patch
@@ -24,7 +24,7 @@ Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
  1 file changed, 1 insertion(+)
 
 diff --git a/arch/x86/kvm/svm/svm.c b/arch/x86/kvm/svm/svm.c
-index 5ab2c92c7331..1cd03c985fdc 100644
+index 5ab2c92c7331deba414de1077b3ea0eb58920d66..1cd03c985fdc4f72ea774d8d4457400230409430 100644
 --- a/arch/x86/kvm/svm/svm.c
 +++ b/arch/x86/kvm/svm/svm.c
 @@ -5174,6 +5174,7 @@ static __init void svm_set_cpu_caps(void)
diff --git a/patches/kernel/0011-revert-memfd-improve-userspace-warnings-for-missing-.patch b/patches/kernel/0011-revert-memfd-improve-userspace-warnings-for-missing-.patch
index 4adca03..01aa276 100644
--- a/patches/kernel/0011-revert-memfd-improve-userspace-warnings-for-missing-.patch
+++ b/patches/kernel/0011-revert-memfd-improve-userspace-warnings-for-missing-.patch
@@ -30,7 +30,7 @@ Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
  1 file changed, 1 insertion(+), 1 deletion(-)
 
 diff --git a/mm/memfd.c b/mm/memfd.c
-index e7b7c5294d59..3a1b073ad22c 100644
+index e7b7c5294d59635539ee75a165a3b6ae171a632e..3a1b073ad22c9f64b0dc3ca0823e6569072936d3 100644
 --- a/mm/memfd.c
 +++ b/mm/memfd.c
 @@ -316,7 +316,7 @@ static int check_sysctl_memfd_noexec(unsigned int *flags)
diff --git a/patches/kernel/0012-apparmor-expect-msg_namelen-0-for-recvmsg-calls.patch b/patches/kernel/0012-apparmor-expect-msg_namelen-0-for-recvmsg-calls.patch
index 9bd7c79..36d67f1 100644
--- a/patches/kernel/0012-apparmor-expect-msg_namelen-0-for-recvmsg-calls.patch
+++ b/patches/kernel/0012-apparmor-expect-msg_namelen-0-for-recvmsg-calls.patch
@@ -17,7 +17,7 @@ Signed-off-by: Wolfgang Bumiller <w.bumiller@proxmox.com>
  1 file changed, 1 insertion(+), 1 deletion(-)
 
 diff --git a/security/apparmor/af_inet.c b/security/apparmor/af_inet.c
-index 57b710054a76..35f905d9b960 100644
+index 57b710054a76582346f37671843f3f8d6e99331c..35f905d9b960f62fa2ecb80b5c1a8e9edecd9b5d 100644
 --- a/security/apparmor/af_inet.c
 +++ b/security/apparmor/af_inet.c
 @@ -766,7 +766,7 @@ int aa_inet_msg_perm(const char *op, u32 request, struct socket *sock,
diff --git a/patches/kernel/0013-io_uring-rw-treat-EOPNOTSUPP-for-IOCB_NOWAIT-like-EA.patch b/patches/kernel/0013-io_uring-rw-treat-EOPNOTSUPP-for-IOCB_NOWAIT-like-EA.patch
index e0027ba..6d19b8f 100644
--- a/patches/kernel/0013-io_uring-rw-treat-EOPNOTSUPP-for-IOCB_NOWAIT-like-EA.patch
+++ b/patches/kernel/0013-io_uring-rw-treat-EOPNOTSUPP-for-IOCB_NOWAIT-like-EA.patch
@@ -25,7 +25,7 @@ Signed-off-by: Daniel Kral <d.kral@proxmox.com>
  1 file changed, 8 insertions(+)
 
 diff --git a/io_uring/rw.c b/io_uring/rw.c
-index c004d21e2f12..d85e2d41a992 100644
+index c004d21e2f12e39d336339e0340f431eddcec597..d85e2d41a992b82622571c764325af73c12b1e75 100644
 --- a/io_uring/rw.c
 +++ b/io_uring/rw.c
 @@ -855,6 +855,14 @@ static int __io_read(struct io_kiocb *req, unsigned int issue_flags)
diff --git a/patches/kernel/0014-netfs-reset-subreq-iov-iter-before-tail-clean.patch b/patches/kernel/0014-netfs-reset-subreq-iov-iter-before-tail-clean.patch
index e8ac17d..d8003f9 100644
--- a/patches/kernel/0014-netfs-reset-subreq-iov-iter-before-tail-clean.patch
+++ b/patches/kernel/0014-netfs-reset-subreq-iov-iter-before-tail-clean.patch
@@ -15,7 +15,7 @@ Signed-off-by: Christian Ebner <c.ebner@proxmox.com>
  1 file changed, 1 insertion(+)
 
 diff --git a/fs/netfs/io.c b/fs/netfs/io.c
-index d6ada4eba744..500119285346 100644
+index d6ada4eba74455aad26273a63247356a3910dc4e..500119285346be28a87698dd6ac66b5e276a6c66 100644
 --- a/fs/netfs/io.c
 +++ b/fs/netfs/io.c
 @@ -528,6 +528,7 @@ void netfs_subreq_terminated(struct netfs_io_subrequest *subreq,
diff --git a/patches/kernel/0015-x86-CPU-AMD-Clear-virtualized-VMLOAD-VMSAVE-on-Zen4-.patch b/patches/kernel/0015-x86-CPU-AMD-Clear-virtualized-VMLOAD-VMSAVE-on-Zen4-.patch
index 76957b6..af273c6 100644
--- a/patches/kernel/0015-x86-CPU-AMD-Clear-virtualized-VMLOAD-VMSAVE-on-Zen4-.patch
+++ b/patches/kernel/0015-x86-CPU-AMD-Clear-virtualized-VMLOAD-VMSAVE-on-Zen4-.patch
@@ -21,7 +21,7 @@ Signed-off-by: Fiona Ebner <f.ebner@proxmox.com>
  1 file changed, 11 insertions(+)
 
 diff --git a/arch/x86/kernel/cpu/amd.c b/arch/x86/kernel/cpu/amd.c
-index 1e0fe5f8ab84..ee87f997d31f 100644
+index 1e0fe5f8ab84e478294b62267176ae01fa540f31..ee87f997d31f90793e759a81e7c3049aadd0f782 100644
 --- a/arch/x86/kernel/cpu/amd.c
 +++ b/arch/x86/kernel/cpu/amd.c
 @@ -924,6 +924,17 @@ static void init_amd_zen4(struct cpuinfo_x86 *c)
diff --git a/patches/kernel/0016-PCI-Fix-pci_enable_acs-support-for-the-ACS-quirks.patch b/patches/kernel/0016-PCI-Fix-pci_enable_acs-support-for-the-ACS-quirks.patch
index 8f2a133..4b2bba1 100644
--- a/patches/kernel/0016-PCI-Fix-pci_enable_acs-support-for-the-ACS-quirks.patch
+++ b/patches/kernel/0016-PCI-Fix-pci_enable_acs-support-for-the-ACS-quirks.patch
@@ -33,7 +33,7 @@ Signed-off-by: Stoiko Ivanov <s.ivanov@proxmox.com>
  1 file changed, 9 insertions(+), 5 deletions(-)
 
 diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c
-index ffaaca0978cb..a89436d481f0 100644
+index ffaaca0978cbc95fbb7566e9700b375cdb4940cc..a89436d481f0702d57516b378d282808766410c6 100644
 --- a/drivers/pci/pci.c
 +++ b/drivers/pci/pci.c
 @@ -1067,8 +1067,15 @@ static void pci_std_enable_acs(struct pci_dev *dev, struct pci_acs *caps)
diff --git a/patches/kernel/0017-Revert-UBUNTU-SAUCE-iommu-intel-disable-DMAR-for-SKL.patch b/patches/kernel/0017-Revert-UBUNTU-SAUCE-iommu-intel-disable-DMAR-for-SKL.patch
index 100797a..8d3f744 100644
--- a/patches/kernel/0017-Revert-UBUNTU-SAUCE-iommu-intel-disable-DMAR-for-SKL.patch
+++ b/patches/kernel/0017-Revert-UBUNTU-SAUCE-iommu-intel-disable-DMAR-for-SKL.patch
@@ -14,7 +14,7 @@ This reverts commit b310f5f58c83756fc164f7d391d76f0df9cc65c3.
  1 file changed, 27 deletions(-)
 
 diff --git a/drivers/iommu/intel/iommu.c b/drivers/iommu/intel/iommu.c
-index 34c8b02ca787..a27cbc22148f 100644
+index 34c8b02ca7878e0b1803a07f2f6e5c6c0039a9f8..a27cbc22148fa3123efbe4d0188bccc4731b7b87 100644
 --- a/drivers/iommu/intel/iommu.c
 +++ b/drivers/iommu/intel/iommu.c
 @@ -4766,33 +4766,6 @@ DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, 0x1632, quirk_iommu_igfx);
-- 
2.39.5



_______________________________________________
pve-devel mailing list
pve-devel@lists.proxmox.com
https://lists.proxmox.com/cgi-bin/mailman/listinfo/pve-devel

[pve-devel] [PATCH master kernel 2/4] backport fix for kvm performance regression with Intel Emerald Rapids

[Fiona Ebner]

Adapted to context change in "arch/x86/kvm/cpuid.h", because of the
vcpu_supports_xsave_pkru() function that got added by Proxmox VE
downstream patch "kvm: xsave set: mask-out PKRU bit in xfeatures if
vCPU has no support". But otherwise clean cherry-pick from linux-next,
no functional changes.

Signed-off-by: Fiona Ebner <f.ebner@proxmox.com>
---
 ...UID.0xD-XSTATE-offsets-sizes-during-.patch | 165 ++++++++++++++++++
 1 file changed, 165 insertions(+)
 create mode 100644 patches/kernel/0018-KVM-x86-Cache-CPUID.0xD-XSTATE-offsets-sizes-during-.patch

diff --git a/patches/kernel/0018-KVM-x86-Cache-CPUID.0xD-XSTATE-offsets-sizes-during-.patch b/patches/kernel/0018-KVM-x86-Cache-CPUID.0xD-XSTATE-offsets-sizes-during-.patch
new file mode 100644
index 0000000..5b03995
--- /dev/null
+++ b/patches/kernel/0018-KVM-x86-Cache-CPUID.0xD-XSTATE-offsets-sizes-during-.patch
@@ -0,0 +1,165 @@
+From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
+From: Sean Christopherson <seanjc@google.com>
+Date: Tue, 10 Dec 2024 17:32:58 -0800
+Subject: [PATCH] KVM: x86: Cache CPUID.0xD XSTATE offsets+sizes during module
+ init
+
+Snapshot the output of CPUID.0xD.[1..n] during kvm.ko initiliaization to
+avoid the overead of CPUID during runtime.  The offset, size, and metadata
+for CPUID.0xD.[1..n] sub-leaves does not depend on XCR0 or XSS values, i.e.
+is constant for a given CPU, and thus can be cached during module load.
+
+On Intel's Emerald Rapids, CPUID is *wildly* expensive, to the point where
+recomputing XSAVE offsets and sizes results in a 4x increase in latency of
+nested VM-Enter and VM-Exit (nested transitions can trigger
+xstate_required_size() multiple times per transition), relative to using
+cached values.  The issue is easily visible by running `perf top` while
+triggering nested transitions: kvm_update_cpuid_runtime() shows up at a
+whopping 50%.
+
+As measured via RDTSC from L2 (using KVM-Unit-Test's CPUID VM-Exit test
+and a slightly modified L1 KVM to handle CPUID in the fastpath), a nested
+roundtrip to emulate CPUID on Skylake (SKX), Icelake (ICX), and Emerald
+Rapids (EMR) takes:
+
+  SKX 11650
+  ICX 22350
+  EMR 28850
+
+Using cached values, the latency drops to:
+
+  SKX 6850
+  ICX 9000
+  EMR 7900
+
+The underlying issue is that CPUID itself is slow on ICX, and comically
+slow on EMR.  The problem is exacerbated on CPUs which support XSAVES
+and/or XSAVEC, as KVM invokes xstate_required_size() twice on each
+runtime CPUID update, and because there are more supported XSAVE features
+(CPUID for supported XSAVE feature sub-leafs is significantly slower).
+
+ SKX:
+  CPUID.0xD.2  = 348 cycles
+  CPUID.0xD.3  = 400 cycles
+  CPUID.0xD.4  = 276 cycles
+  CPUID.0xD.5  = 236 cycles
+  <other sub-leaves are similar>
+
+ EMR:
+  CPUID.0xD.2  = 1138 cycles
+  CPUID.0xD.3  = 1362 cycles
+  CPUID.0xD.4  = 1068 cycles
+  CPUID.0xD.5  = 910 cycles
+  CPUID.0xD.6  = 914 cycles
+  CPUID.0xD.7  = 1350 cycles
+  CPUID.0xD.8  = 734 cycles
+  CPUID.0xD.9  = 766 cycles
+  CPUID.0xD.10 = 732 cycles
+  CPUID.0xD.11 = 718 cycles
+  CPUID.0xD.12 = 734 cycles
+  CPUID.0xD.13 = 1700 cycles
+  CPUID.0xD.14 = 1126 cycles
+  CPUID.0xD.15 = 898 cycles
+  CPUID.0xD.16 = 716 cycles
+  CPUID.0xD.17 = 748 cycles
+  CPUID.0xD.18 = 776 cycles
+
+Note, updating runtime CPUID information multiple times per nested
+transition is itself a flaw, especially since CPUID is a mandotory
+intercept on both Intel and AMD.  E.g. KVM doesn't need to ensure emulated
+CPUID state is up-to-date while running L2.  That flaw will be fixed in a
+future patch, as deferring runtime CPUID updates is more subtle than it
+appears at first glance, the benefits aren't super critical to have once
+the XSAVE issue is resolved, and caching CPUID output is desirable even if
+KVM's updates are deferred.
+
+Cc: Jim Mattson <jmattson@google.com>
+Cc: stable@vger.kernel.org
+Signed-off-by: Sean Christopherson <seanjc@google.com>
+Message-ID: <20241211013302.1347853-2-seanjc@google.com>
+Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
+(cherry picked from commit 1201f226c863b7da739f7420ddba818cedf372fc)
+Signed-off-by: Fiona Ebner <f.ebner@proxmox.com>
+---
+ arch/x86/kvm/cpuid.c | 31 ++++++++++++++++++++++++++-----
+ arch/x86/kvm/cpuid.h |  1 +
+ arch/x86/kvm/x86.c   |  2 ++
+ 3 files changed, 29 insertions(+), 5 deletions(-)
+
+diff --git a/arch/x86/kvm/cpuid.c b/arch/x86/kvm/cpuid.c
+index b6892645c4cde238923062e3f71df21d68cc201d..dc73965aa73b21d26b4cf039336da3ca38e89bc6 100644
+--- a/arch/x86/kvm/cpuid.c
++++ b/arch/x86/kvm/cpuid.c
+@@ -36,6 +36,26 @@
+ u32 kvm_cpu_caps[NR_KVM_CPU_CAPS] __read_mostly;
+ EXPORT_SYMBOL_GPL(kvm_cpu_caps);
+ 
++struct cpuid_xstate_sizes {
++	u32 eax;
++	u32 ebx;
++	u32 ecx;
++};
++
++static struct cpuid_xstate_sizes xstate_sizes[XFEATURE_MAX] __ro_after_init;
++
++void __init kvm_init_xstate_sizes(void)
++{
++	u32 ign;
++	int i;
++
++	for (i = XFEATURE_YMM; i < ARRAY_SIZE(xstate_sizes); i++) {
++		struct cpuid_xstate_sizes *xs = &xstate_sizes[i];
++
++		cpuid_count(0xD, i, &xs->eax, &xs->ebx, &xs->ecx, &ign);
++	}
++}
++
+ u32 xstate_required_size(u64 xstate_bv, bool compacted)
+ {
+ 	int feature_bit = 0;
+@@ -44,14 +64,15 @@ u32 xstate_required_size(u64 xstate_bv, bool compacted)
+ 	xstate_bv &= XFEATURE_MASK_EXTEND;
+ 	while (xstate_bv) {
+ 		if (xstate_bv & 0x1) {
+-		        u32 eax, ebx, ecx, edx, offset;
+-		        cpuid_count(0xD, feature_bit, &eax, &ebx, &ecx, &edx);
++			struct cpuid_xstate_sizes *xs = &xstate_sizes[feature_bit];
++			u32 offset;
++
+ 			/* ECX[1]: 64B alignment in compacted form */
+ 			if (compacted)
+-				offset = (ecx & 0x2) ? ALIGN(ret, 64) : ret;
++				offset = (xs->ecx & 0x2) ? ALIGN(ret, 64) : ret;
+ 			else
+-				offset = ebx;
+-			ret = max(ret, offset + eax);
++				offset = xs->ebx;
++			ret = max(ret, offset + xs->eax);
+ 		}
+ 
+ 		xstate_bv >>= 1;
+diff --git a/arch/x86/kvm/cpuid.h b/arch/x86/kvm/cpuid.h
+index 0e46c4555311d02b6ec9eba7f33b292c3562b0ee..6287ec33feb7b143a10a7d8ee38d74381a567d24 100644
+--- a/arch/x86/kvm/cpuid.h
++++ b/arch/x86/kvm/cpuid.h
+@@ -34,6 +34,7 @@ bool kvm_cpuid(struct kvm_vcpu *vcpu, u32 *eax, u32 *ebx,
+ 
+ bool vcpu_supports_xsave_pkru(struct kvm_vcpu *vcpu);
+ 
++void __init kvm_init_xstate_sizes(void);
+ u32 xstate_required_size(u64 xstate_bv, bool compacted);
+ 
+ int cpuid_query_maxphyaddr(struct kvm_vcpu *vcpu);
+diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
+index c7c3d04198c6c7f88f97e122bfe78b857f26cb26..4d10fc1a9b4114d1e2edf133717f307043560263 100644
+--- a/arch/x86/kvm/x86.c
++++ b/arch/x86/kvm/x86.c
+@@ -14061,6 +14061,8 @@ EXPORT_TRACEPOINT_SYMBOL_GPL(kvm_rmp_fault);
+ 
+ static int __init kvm_x86_init(void)
+ {
++	kvm_init_xstate_sizes();
++
+ 	kvm_mmu_x86_module_init();
+ 	mitigate_smt_rsb &= boot_cpu_has_bug(X86_BUG_SMT_RSB) && cpu_smt_possible();
+ 	return 0;
-- 
2.39.5



_______________________________________________
pve-devel mailing list
pve-devel@lists.proxmox.com
https://lists.proxmox.com/cgi-bin/mailman/listinfo/pve-devel

[pve-devel] [PATCH bookworm-6.8 kernel 3/4] patches: kernel: switch to using full index for patch files

[Fiona Ebner]

git will automatically change the length of the index hashes when
formatting a patch depending on what references are present in the
submodule. After pulling in the stable tags today, git wanted to add
a character to all hashes for me. Use --full-index when generating the
patches to avoid such issues in the future.

No functional change intended.

Signed-off-by: Fiona Ebner <f.ebner@proxmox.com>
---
 ...ke-mkcompile_h-accept-an-alternate-timestamp-strin.patch | 2 +-
 ...02-wireless-Add-Debian-wireless-regdb-certificates.patch | 2 +-
 .../0003-bridge-keep-MAC-of-first-assigned-port.patch       | 2 +-
 ...i-Enable-overrides-for-missing-ACS-capabilities-4..patch | 4 ++--
 ...05-kvm-disable-default-dynamic-halt-polling-growth.patch | 2 +-
 ...t-core-downgrade-unregister_netdevice-refcount-lea.patch | 2 +-
 .../0007-Revert-fortify-Do-not-cast-to-unsigned-char.patch  | 2 +-
 ...m-xsave-set-mask-out-PKRU-bit-in-xfeatures-if-vCPU.patch | 6 +++---
 ...low-opt-in-to-allow-pass-through-on-broken-hardwar.patch | 2 +-
 .../0010-KVM-nSVM-Advertise-support-for-flush-by-ASID.patch | 2 +-
 ...vert-memfd-improve-userspace-warnings-for-missing-.patch | 2 +-
 ...12-apparmor-expect-msg_namelen-0-for-recvmsg-calls.patch | 2 +-
 .../0013-cifs-fix-pagecache-leak-when-do-writepages.patch   | 2 +-
 .../0014-drm-amdgpu-pm-Don-t-use-OD-table-on-Arcturus.patch | 2 +-
 ...vert-UBUNTU-SAUCE-iommu-intel-disable-DMAR-for-SKL.patch | 2 +-
 ..._uring-rw-treat-EOPNOTSUPP-for-IOCB_NOWAIT-like-EA.patch | 2 +-
 ...0017-netfs-reset-subreq-iov-iter-before-tail-clean.patch | 2 +-
 ...6-CPU-AMD-Clear-virtualized-VMLOAD-VMSAVE-on-Zen4-.patch | 2 +-
 18 files changed, 21 insertions(+), 21 deletions(-)

diff --git a/patches/kernel/0001-Make-mkcompile_h-accept-an-alternate-timestamp-strin.patch b/patches/kernel/0001-Make-mkcompile_h-accept-an-alternate-timestamp-strin.patch
index 93dba9f..e626b45 100644
--- a/patches/kernel/0001-Make-mkcompile_h-accept-an-alternate-timestamp-strin.patch
+++ b/patches/kernel/0001-Make-mkcompile_h-accept-an-alternate-timestamp-strin.patch
@@ -21,7 +21,7 @@ Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
  1 file changed, 1 insertion(+), 1 deletion(-)
 
 diff --git a/init/Makefile b/init/Makefile
-index cbac576c57d6..479b1253fcbe 100644
+index cbac576c57d63f20fda8720762f80225e0bcb542..479b1253fcbe2438afed4632882ff9a958472e2b 100644
 --- a/init/Makefile
 +++ b/init/Makefile
 @@ -29,7 +29,7 @@ preempt-flag-$(CONFIG_PREEMPT_DYNAMIC)	:= PREEMPT_DYNAMIC
diff --git a/patches/kernel/0002-wireless-Add-Debian-wireless-regdb-certificates.patch b/patches/kernel/0002-wireless-Add-Debian-wireless-regdb-certificates.patch
index 4c0ac0f..274b7ff 100644
--- a/patches/kernel/0002-wireless-Add-Debian-wireless-regdb-certificates.patch
+++ b/patches/kernel/0002-wireless-Add-Debian-wireless-regdb-certificates.patch
@@ -19,7 +19,7 @@ Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
 
 diff --git a/net/wireless/certs/debian.hex b/net/wireless/certs/debian.hex
 new file mode 100644
-index 000000000000..c5ab03f8c500
+index 0000000000000000000000000000000000000000..c5ab03f8c500d2f0e5b7931d5790bd22983c3660
 --- /dev/null
 +++ b/net/wireless/certs/debian.hex
 @@ -0,0 +1,1426 @@
diff --git a/patches/kernel/0003-bridge-keep-MAC-of-first-assigned-port.patch b/patches/kernel/0003-bridge-keep-MAC-of-first-assigned-port.patch
index 135d17c..efa47df 100644
--- a/patches/kernel/0003-bridge-keep-MAC-of-first-assigned-port.patch
+++ b/patches/kernel/0003-bridge-keep-MAC-of-first-assigned-port.patch
@@ -19,7 +19,7 @@ Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
  1 file changed, 1 insertion(+), 4 deletions(-)
 
 diff --git a/net/bridge/br_stp_if.c b/net/bridge/br_stp_if.c
-index 75204d36d7f9..1fb5ff73ec1e 100644
+index 75204d36d7f9062306dfc66c3c35448e16257215..1fb5ff73ec1ef3bd79960182c87a0ba312b3635d 100644
 --- a/net/bridge/br_stp_if.c
 +++ b/net/bridge/br_stp_if.c
 @@ -265,10 +265,7 @@ bool br_stp_recalculate_bridge_id(struct net_bridge *br)
diff --git a/patches/kernel/0004-pci-Enable-overrides-for-missing-ACS-capabilities-4..patch b/patches/kernel/0004-pci-Enable-overrides-for-missing-ACS-capabilities-4..patch
index 02c6d64..d239cbf 100644
--- a/patches/kernel/0004-pci-Enable-overrides-for-missing-ACS-capabilities-4..patch
+++ b/patches/kernel/0004-pci-Enable-overrides-for-missing-ACS-capabilities-4..patch
@@ -55,7 +55,7 @@ Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
  2 files changed, 111 insertions(+)
 
 diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt
-index e9be08602d4d..2ed9b047bbd8 100644
+index e9be08602d4db4e9eeca3d3d8cc9ee2e8a330ee8..2ed9b047bbd8fab74596fbc3ccb035fdbb63aa4e 100644
 --- a/Documentation/admin-guide/kernel-parameters.txt
 +++ b/Documentation/admin-guide/kernel-parameters.txt
 @@ -4397,6 +4397,15 @@
@@ -75,7 +75,7 @@ index e9be08602d4d..2ed9b047bbd8 100644
  				Safety option to keep boot IRQs enabled. This
  				should never be necessary.
 diff --git a/drivers/pci/quirks.c b/drivers/pci/quirks.c
-index ce469d84ebae..4f163ef55e7b 100644
+index ce469d84ebaec642fdcffc66334b5d0c4ebb672c..4f163ef55e7b3272dc31fdb16aa9e16e46965496 100644
 --- a/drivers/pci/quirks.c
 +++ b/drivers/pci/quirks.c
 @@ -287,6 +287,106 @@ static int __init pci_apply_final_quirks(void)
diff --git a/patches/kernel/0005-kvm-disable-default-dynamic-halt-polling-growth.patch b/patches/kernel/0005-kvm-disable-default-dynamic-halt-polling-growth.patch
index 8d590eb..e85f2ec 100644
--- a/patches/kernel/0005-kvm-disable-default-dynamic-halt-polling-growth.patch
+++ b/patches/kernel/0005-kvm-disable-default-dynamic-halt-polling-growth.patch
@@ -13,7 +13,7 @@ Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
  1 file changed, 1 insertion(+), 1 deletion(-)
 
 diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c
-index 6a56de7ff82e..96bd40a73e0e 100644
+index 6a56de7ff82e707035a1ade82c7f5fafba4b1f26..96bd40a73e0ed5872eca156ef1f61ab4bda6661f 100644
 --- a/virt/kvm/kvm_main.c
 +++ b/virt/kvm/kvm_main.c
 @@ -82,7 +82,7 @@ module_param(halt_poll_ns, uint, 0644);
diff --git a/patches/kernel/0006-net-core-downgrade-unregister_netdevice-refcount-lea.patch b/patches/kernel/0006-net-core-downgrade-unregister_netdevice-refcount-lea.patch
index 50b02ce..a75076e 100644
--- a/patches/kernel/0006-net-core-downgrade-unregister_netdevice-refcount-lea.patch
+++ b/patches/kernel/0006-net-core-downgrade-unregister_netdevice-refcount-lea.patch
@@ -14,7 +14,7 @@ Signed-off-by: Fabian Grünbichler <f.gruenbichler@proxmox.com>
  1 file changed, 1 insertion(+), 1 deletion(-)
 
 diff --git a/net/core/dev.c b/net/core/dev.c
-index a32811aebde5..15078ab81ec8 100644
+index a32811aebde5979c5ac2023b3925f6f076a42563..15078ab81ec824c56307b77f784fbe30b66d303e 100644
 --- a/net/core/dev.c
 +++ b/net/core/dev.c
 @@ -10471,7 +10471,7 @@ static struct net_device *netdev_wait_allrefs_any(struct list_head *list)
diff --git a/patches/kernel/0007-Revert-fortify-Do-not-cast-to-unsigned-char.patch b/patches/kernel/0007-Revert-fortify-Do-not-cast-to-unsigned-char.patch
index f6186d1..2d4fd6f 100644
--- a/patches/kernel/0007-Revert-fortify-Do-not-cast-to-unsigned-char.patch
+++ b/patches/kernel/0007-Revert-fortify-Do-not-cast-to-unsigned-char.patch
@@ -16,7 +16,7 @@ Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
  1 file changed, 1 insertion(+), 1 deletion(-)
 
 diff --git a/include/linux/fortify-string.h b/include/linux/fortify-string.h
-index e99dbc052575..9e9cdb198b82 100644
+index e99dbc0525751dacf0bcd49c208e45f59312f986..9e9cdb198b825ff315a336af36e4fc3b6c09fb7d 100644
 --- a/include/linux/fortify-string.h
 +++ b/include/linux/fortify-string.h
 @@ -18,7 +18,7 @@ void __write_overflow_field(size_t avail, size_t wanted) __compiletime_warning("
diff --git a/patches/kernel/0008-kvm-xsave-set-mask-out-PKRU-bit-in-xfeatures-if-vCPU.patch b/patches/kernel/0008-kvm-xsave-set-mask-out-PKRU-bit-in-xfeatures-if-vCPU.patch
index 3ef9c52..9621750 100644
--- a/patches/kernel/0008-kvm-xsave-set-mask-out-PKRU-bit-in-xfeatures-if-vCPU.patch
+++ b/patches/kernel/0008-kvm-xsave-set-mask-out-PKRU-bit-in-xfeatures-if-vCPU.patch
@@ -78,7 +78,7 @@ Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
  3 files changed, 21 insertions(+)
 
 diff --git a/arch/x86/kvm/cpuid.c b/arch/x86/kvm/cpuid.c
-index ce1499732cb8..d68c04bde5ed 100644
+index ce1499732cb86f7a547ec319ef3430f6e463acdc..d68c04bde5ededcd2ffb24300646f4a9d9c1bc9a 100644
 --- a/arch/x86/kvm/cpuid.c
 +++ b/arch/x86/kvm/cpuid.c
 @@ -262,6 +262,12 @@ static u64 cpuid_get_supported_xcr0(struct kvm_cpuid_entry2 *entries, int nent)
@@ -95,7 +95,7 @@ index ce1499732cb8..d68c04bde5ed 100644
  				       int nent)
  {
 diff --git a/arch/x86/kvm/cpuid.h b/arch/x86/kvm/cpuid.h
-index 23dbb9eb277c..07da153802e4 100644
+index 23dbb9eb277c7465f19bc5af137b79d5a2b894d1..07da153802e4d4f75219d77596e323347393eab7 100644
 --- a/arch/x86/kvm/cpuid.h
 +++ b/arch/x86/kvm/cpuid.h
 @@ -32,6 +32,8 @@ int kvm_vcpu_ioctl_get_cpuid2(struct kvm_vcpu *vcpu,
@@ -108,7 +108,7 @@ index 23dbb9eb277c..07da153802e4 100644
  
  int cpuid_query_maxphyaddr(struct kvm_vcpu *vcpu);
 diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
-index e36bf6162fc8..08ef6f01bf7e 100644
+index e36bf6162fc801d5ad505f0fd28baf5c7c2278ce..08ef6f01bf7e8c9c6cdb842f23adac09195c1aff 100644
 --- a/arch/x86/kvm/x86.c
 +++ b/arch/x86/kvm/x86.c
 @@ -5580,6 +5580,19 @@ static int kvm_vcpu_ioctl_x86_set_xsave(struct kvm_vcpu *vcpu,
diff --git a/patches/kernel/0009-allow-opt-in-to-allow-pass-through-on-broken-hardwar.patch b/patches/kernel/0009-allow-opt-in-to-allow-pass-through-on-broken-hardwar.patch
index c75297a..f308440 100644
--- a/patches/kernel/0009-allow-opt-in-to-allow-pass-through-on-broken-hardwar.patch
+++ b/patches/kernel/0009-allow-opt-in-to-allow-pass-through-on-broken-hardwar.patch
@@ -11,7 +11,7 @@ Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
  1 file changed, 5 insertions(+), 1 deletion(-)
 
 diff --git a/drivers/iommu/intel/iommu.c b/drivers/iommu/intel/iommu.c
-index 94518306c6f9..d9cc1264e40a 100644
+index 94518306c6f96fdf590cdf638e7f1e73c176906a..d9cc1264e40afa9250305b960419185e23863ccc 100644
 --- a/drivers/iommu/intel/iommu.c
 +++ b/drivers/iommu/intel/iommu.c
 @@ -234,6 +234,7 @@ static int dmar_map_gfx = 1;
diff --git a/patches/kernel/0010-KVM-nSVM-Advertise-support-for-flush-by-ASID.patch b/patches/kernel/0010-KVM-nSVM-Advertise-support-for-flush-by-ASID.patch
index c4b0186..d9c1f87 100644
--- a/patches/kernel/0010-KVM-nSVM-Advertise-support-for-flush-by-ASID.patch
+++ b/patches/kernel/0010-KVM-nSVM-Advertise-support-for-flush-by-ASID.patch
@@ -24,7 +24,7 @@ Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
  1 file changed, 1 insertion(+)
 
 diff --git a/arch/x86/kvm/svm/svm.c b/arch/x86/kvm/svm/svm.c
-index 9d070f7da796..2ac30170f887 100644
+index 9d070f7da796ebc18b04481d84a05586be726725..2ac30170f887caa2c191db46678cb3ad678b9568 100644
 --- a/arch/x86/kvm/svm/svm.c
 +++ b/arch/x86/kvm/svm/svm.c
 @@ -5114,6 +5114,7 @@ static __init void svm_set_cpu_caps(void)
diff --git a/patches/kernel/0011-revert-memfd-improve-userspace-warnings-for-missing-.patch b/patches/kernel/0011-revert-memfd-improve-userspace-warnings-for-missing-.patch
index 5df98ce..59fc777 100644
--- a/patches/kernel/0011-revert-memfd-improve-userspace-warnings-for-missing-.patch
+++ b/patches/kernel/0011-revert-memfd-improve-userspace-warnings-for-missing-.patch
@@ -30,7 +30,7 @@ Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
  1 file changed, 1 insertion(+), 1 deletion(-)
 
 diff --git a/mm/memfd.c b/mm/memfd.c
-index d3a1ba4208c9..6a9de5d9105e 100644
+index d3a1ba4208c90139f5c172751b7ca7320b2f54a9..6a9de5d9105ea5b2b41be8c9549dc98624a8290d 100644
 --- a/mm/memfd.c
 +++ b/mm/memfd.c
 @@ -282,7 +282,7 @@ static int check_sysctl_memfd_noexec(unsigned int *flags)
diff --git a/patches/kernel/0012-apparmor-expect-msg_namelen-0-for-recvmsg-calls.patch b/patches/kernel/0012-apparmor-expect-msg_namelen-0-for-recvmsg-calls.patch
index 9bd7c79..36d67f1 100644
--- a/patches/kernel/0012-apparmor-expect-msg_namelen-0-for-recvmsg-calls.patch
+++ b/patches/kernel/0012-apparmor-expect-msg_namelen-0-for-recvmsg-calls.patch
@@ -17,7 +17,7 @@ Signed-off-by: Wolfgang Bumiller <w.bumiller@proxmox.com>
  1 file changed, 1 insertion(+), 1 deletion(-)
 
 diff --git a/security/apparmor/af_inet.c b/security/apparmor/af_inet.c
-index 57b710054a76..35f905d9b960 100644
+index 57b710054a76582346f37671843f3f8d6e99331c..35f905d9b960f62fa2ecb80b5c1a8e9edecd9b5d 100644
 --- a/security/apparmor/af_inet.c
 +++ b/security/apparmor/af_inet.c
 @@ -766,7 +766,7 @@ int aa_inet_msg_perm(const char *op, u32 request, struct socket *sock,
diff --git a/patches/kernel/0013-cifs-fix-pagecache-leak-when-do-writepages.patch b/patches/kernel/0013-cifs-fix-pagecache-leak-when-do-writepages.patch
index 79ae5aa..52369af 100644
--- a/patches/kernel/0013-cifs-fix-pagecache-leak-when-do-writepages.patch
+++ b/patches/kernel/0013-cifs-fix-pagecache-leak-when-do-writepages.patch
@@ -47,7 +47,7 @@ Signed-off-by: Fiona Ebner <f.ebner@proxmox.com>
  1 file changed, 13 insertions(+), 3 deletions(-)
 
 diff --git a/fs/smb/client/file.c b/fs/smb/client/file.c
-index d495e3511014..e755ae643c09 100644
+index d495e3511014b9e628d0592d328f63b4a680de97..e755ae643c090e595bf2431e471eddff7b868e9f 100644
 --- a/fs/smb/client/file.c
 +++ b/fs/smb/client/file.c
 @@ -2845,17 +2845,21 @@ static ssize_t cifs_write_back_from_locked_folio(struct address_space *mapping,
diff --git a/patches/kernel/0014-drm-amdgpu-pm-Don-t-use-OD-table-on-Arcturus.patch b/patches/kernel/0014-drm-amdgpu-pm-Don-t-use-OD-table-on-Arcturus.patch
index 086a9cb..1df03d6 100644
--- a/patches/kernel/0014-drm-amdgpu-pm-Don-t-use-OD-table-on-Arcturus.patch
+++ b/patches/kernel/0014-drm-amdgpu-pm-Don-t-use-OD-table-on-Arcturus.patch
@@ -16,7 +16,7 @@ Signed-off-by: Fiona Ebner <f.ebner@proxmox.com>
  1 file changed, 5 insertions(+), 28 deletions(-)
 
 diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu11/arcturus_ppt.c b/drivers/gpu/drm/amd/pm/swsmu/smu11/arcturus_ppt.c
-index 437ac31b1673..415032f58522 100644
+index 437ac31b1673986f96a5a714e3e017d06e053dc9..415032f58522f8eefcfe1ddad5dc992f6f3e2e3d 100644
 --- a/drivers/gpu/drm/amd/pm/swsmu/smu11/arcturus_ppt.c
 +++ b/drivers/gpu/drm/amd/pm/swsmu/smu11/arcturus_ppt.c
 @@ -1283,11 +1283,8 @@ static int arcturus_get_power_limit(struct smu_context *smu,
diff --git a/patches/kernel/0015-Revert-UBUNTU-SAUCE-iommu-intel-disable-DMAR-for-SKL.patch b/patches/kernel/0015-Revert-UBUNTU-SAUCE-iommu-intel-disable-DMAR-for-SKL.patch
index b273b4a..534e836 100644
--- a/patches/kernel/0015-Revert-UBUNTU-SAUCE-iommu-intel-disable-DMAR-for-SKL.patch
+++ b/patches/kernel/0015-Revert-UBUNTU-SAUCE-iommu-intel-disable-DMAR-for-SKL.patch
@@ -14,7 +14,7 @@ This reverts commit b310f5f58c83756fc164f7d391d76f0df9cc65c3.
  1 file changed, 27 deletions(-)
 
 diff --git a/drivers/iommu/intel/iommu.c b/drivers/iommu/intel/iommu.c
-index d9cc1264e40a..89fdcaad20e1 100644
+index d9cc1264e40afa9250305b960419185e23863ccc..89fdcaad20e1e8251fe379cded7701c762628871 100644
 --- a/drivers/iommu/intel/iommu.c
 +++ b/drivers/iommu/intel/iommu.c
 @@ -5034,33 +5034,6 @@ DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, 0x1632, quirk_iommu_igfx);
diff --git a/patches/kernel/0016-io_uring-rw-treat-EOPNOTSUPP-for-IOCB_NOWAIT-like-EA.patch b/patches/kernel/0016-io_uring-rw-treat-EOPNOTSUPP-for-IOCB_NOWAIT-like-EA.patch
index 7187c94..1f320cf 100644
--- a/patches/kernel/0016-io_uring-rw-treat-EOPNOTSUPP-for-IOCB_NOWAIT-like-EA.patch
+++ b/patches/kernel/0016-io_uring-rw-treat-EOPNOTSUPP-for-IOCB_NOWAIT-like-EA.patch
@@ -25,7 +25,7 @@ Signed-off-by: Daniel Kral <d.kral@proxmox.com>
  1 file changed, 8 insertions(+)
 
 diff --git a/io_uring/rw.c b/io_uring/rw.c
-index c3c154790e45..ed7f67097572 100644
+index c3c154790e45230a1d503173e287c6305ab22916..ed7f6709757268faacd69ed4c83a17fa155b3c98 100644
 --- a/io_uring/rw.c
 +++ b/io_uring/rw.c
 @@ -825,6 +825,14 @@ static int __io_read(struct io_kiocb *req, unsigned int issue_flags)
diff --git a/patches/kernel/0017-netfs-reset-subreq-iov-iter-before-tail-clean.patch b/patches/kernel/0017-netfs-reset-subreq-iov-iter-before-tail-clean.patch
index cc4104b..4ecd9a9 100644
--- a/patches/kernel/0017-netfs-reset-subreq-iov-iter-before-tail-clean.patch
+++ b/patches/kernel/0017-netfs-reset-subreq-iov-iter-before-tail-clean.patch
@@ -15,7 +15,7 @@ Signed-off-by: Christian Ebner <c.ebner@proxmox.com>
  1 file changed, 1 insertion(+)
 
 diff --git a/fs/netfs/io.c b/fs/netfs/io.c
-index aaff3844e9b7..e8a884040e2b 100644
+index aaff3844e9b708e20b885cfdcc378f91d3335d4c..e8a884040e2b377b48d6da9f1b5096f8fcd99f92 100644
 --- a/fs/netfs/io.c
 +++ b/fs/netfs/io.c
 @@ -517,6 +517,7 @@ void netfs_subreq_terminated(struct netfs_io_subrequest *subreq,
diff --git a/patches/kernel/0018-x86-CPU-AMD-Clear-virtualized-VMLOAD-VMSAVE-on-Zen4-.patch b/patches/kernel/0018-x86-CPU-AMD-Clear-virtualized-VMLOAD-VMSAVE-on-Zen4-.patch
index 51ed9b9..94ce254 100644
--- a/patches/kernel/0018-x86-CPU-AMD-Clear-virtualized-VMLOAD-VMSAVE-on-Zen4-.patch
+++ b/patches/kernel/0018-x86-CPU-AMD-Clear-virtualized-VMLOAD-VMSAVE-on-Zen4-.patch
@@ -21,7 +21,7 @@ Signed-off-by: Fiona Ebner <f.ebner@proxmox.com>
  1 file changed, 11 insertions(+)
 
 diff --git a/arch/x86/kernel/cpu/amd.c b/arch/x86/kernel/cpu/amd.c
-index 809a66d522c1..edc1b32b5021 100644
+index 809a66d522c1131c850f7fda26b61752f122e192..edc1b32b502106089833c1f327768848a5145081 100644
 --- a/arch/x86/kernel/cpu/amd.c
 +++ b/arch/x86/kernel/cpu/amd.c
 @@ -1061,6 +1061,17 @@ static void init_amd_zen4(struct cpuinfo_x86 *c)
-- 
2.39.5



_______________________________________________
pve-devel mailing list
pve-devel@lists.proxmox.com
https://lists.proxmox.com/cgi-bin/mailman/listinfo/pve-devel

[pve-devel] [PATCH bookworm-6.8 kernel 4/4] backport fix for kvm performance regression with Intel Emerald Rapids

[Fiona Ebner]

Adapted to context change in "arch/x86/kvm/cpuid.h", because of the
vcpu_supports_xsave_pkru() function that got added by Proxmox VE
downstream patch "kvm: xsave set: mask-out PKRU bit in xfeatures if
vCPU has no support". But otherwise clean cherry-pick from linux-next,
no functional changes.

Signed-off-by: Fiona Ebner <f.ebner@proxmox.com>
---
 ...UID.0xD-XSTATE-offsets-sizes-during-.patch | 165 ++++++++++++++++++
 1 file changed, 165 insertions(+)
 create mode 100644 patches/kernel/0019-KVM-x86-Cache-CPUID.0xD-XSTATE-offsets-sizes-during-.patch

diff --git a/patches/kernel/0019-KVM-x86-Cache-CPUID.0xD-XSTATE-offsets-sizes-during-.patch b/patches/kernel/0019-KVM-x86-Cache-CPUID.0xD-XSTATE-offsets-sizes-during-.patch
new file mode 100644
index 0000000..beda78e
--- /dev/null
+++ b/patches/kernel/0019-KVM-x86-Cache-CPUID.0xD-XSTATE-offsets-sizes-during-.patch
@@ -0,0 +1,165 @@
+From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
+From: Sean Christopherson <seanjc@google.com>
+Date: Tue, 10 Dec 2024 17:32:58 -0800
+Subject: [PATCH] KVM: x86: Cache CPUID.0xD XSTATE offsets+sizes during module
+ init
+
+Snapshot the output of CPUID.0xD.[1..n] during kvm.ko initiliaization to
+avoid the overead of CPUID during runtime.  The offset, size, and metadata
+for CPUID.0xD.[1..n] sub-leaves does not depend on XCR0 or XSS values, i.e.
+is constant for a given CPU, and thus can be cached during module load.
+
+On Intel's Emerald Rapids, CPUID is *wildly* expensive, to the point where
+recomputing XSAVE offsets and sizes results in a 4x increase in latency of
+nested VM-Enter and VM-Exit (nested transitions can trigger
+xstate_required_size() multiple times per transition), relative to using
+cached values.  The issue is easily visible by running `perf top` while
+triggering nested transitions: kvm_update_cpuid_runtime() shows up at a
+whopping 50%.
+
+As measured via RDTSC from L2 (using KVM-Unit-Test's CPUID VM-Exit test
+and a slightly modified L1 KVM to handle CPUID in the fastpath), a nested
+roundtrip to emulate CPUID on Skylake (SKX), Icelake (ICX), and Emerald
+Rapids (EMR) takes:
+
+  SKX 11650
+  ICX 22350
+  EMR 28850
+
+Using cached values, the latency drops to:
+
+  SKX 6850
+  ICX 9000
+  EMR 7900
+
+The underlying issue is that CPUID itself is slow on ICX, and comically
+slow on EMR.  The problem is exacerbated on CPUs which support XSAVES
+and/or XSAVEC, as KVM invokes xstate_required_size() twice on each
+runtime CPUID update, and because there are more supported XSAVE features
+(CPUID for supported XSAVE feature sub-leafs is significantly slower).
+
+ SKX:
+  CPUID.0xD.2  = 348 cycles
+  CPUID.0xD.3  = 400 cycles
+  CPUID.0xD.4  = 276 cycles
+  CPUID.0xD.5  = 236 cycles
+  <other sub-leaves are similar>
+
+ EMR:
+  CPUID.0xD.2  = 1138 cycles
+  CPUID.0xD.3  = 1362 cycles
+  CPUID.0xD.4  = 1068 cycles
+  CPUID.0xD.5  = 910 cycles
+  CPUID.0xD.6  = 914 cycles
+  CPUID.0xD.7  = 1350 cycles
+  CPUID.0xD.8  = 734 cycles
+  CPUID.0xD.9  = 766 cycles
+  CPUID.0xD.10 = 732 cycles
+  CPUID.0xD.11 = 718 cycles
+  CPUID.0xD.12 = 734 cycles
+  CPUID.0xD.13 = 1700 cycles
+  CPUID.0xD.14 = 1126 cycles
+  CPUID.0xD.15 = 898 cycles
+  CPUID.0xD.16 = 716 cycles
+  CPUID.0xD.17 = 748 cycles
+  CPUID.0xD.18 = 776 cycles
+
+Note, updating runtime CPUID information multiple times per nested
+transition is itself a flaw, especially since CPUID is a mandotory
+intercept on both Intel and AMD.  E.g. KVM doesn't need to ensure emulated
+CPUID state is up-to-date while running L2.  That flaw will be fixed in a
+future patch, as deferring runtime CPUID updates is more subtle than it
+appears at first glance, the benefits aren't super critical to have once
+the XSAVE issue is resolved, and caching CPUID output is desirable even if
+KVM's updates are deferred.
+
+Cc: Jim Mattson <jmattson@google.com>
+Cc: stable@vger.kernel.org
+Signed-off-by: Sean Christopherson <seanjc@google.com>
+Message-ID: <20241211013302.1347853-2-seanjc@google.com>
+Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
+(cherry picked from commit 1201f226c863b7da739f7420ddba818cedf372fc)
+Signed-off-by: Fiona Ebner <f.ebner@proxmox.com>
+---
+ arch/x86/kvm/cpuid.c | 31 ++++++++++++++++++++++++++-----
+ arch/x86/kvm/cpuid.h |  1 +
+ arch/x86/kvm/x86.c   |  2 ++
+ 3 files changed, 29 insertions(+), 5 deletions(-)
+
+diff --git a/arch/x86/kvm/cpuid.c b/arch/x86/kvm/cpuid.c
+index d68c04bde5ededcd2ffb24300646f4a9d9c1bc9a..be6e9995fb6dcd84aa21ff7a324b585d0a4a008f 100644
+--- a/arch/x86/kvm/cpuid.c
++++ b/arch/x86/kvm/cpuid.c
+@@ -36,6 +36,26 @@
+ u32 kvm_cpu_caps[NR_KVM_CPU_CAPS] __read_mostly;
+ EXPORT_SYMBOL_GPL(kvm_cpu_caps);
+ 
++struct cpuid_xstate_sizes {
++	u32 eax;
++	u32 ebx;
++	u32 ecx;
++};
++
++static struct cpuid_xstate_sizes xstate_sizes[XFEATURE_MAX] __ro_after_init;
++
++void __init kvm_init_xstate_sizes(void)
++{
++	u32 ign;
++	int i;
++
++	for (i = XFEATURE_YMM; i < ARRAY_SIZE(xstate_sizes); i++) {
++		struct cpuid_xstate_sizes *xs = &xstate_sizes[i];
++
++		cpuid_count(0xD, i, &xs->eax, &xs->ebx, &xs->ecx, &ign);
++	}
++}
++
+ u32 xstate_required_size(u64 xstate_bv, bool compacted)
+ {
+ 	int feature_bit = 0;
+@@ -44,14 +64,15 @@ u32 xstate_required_size(u64 xstate_bv, bool compacted)
+ 	xstate_bv &= XFEATURE_MASK_EXTEND;
+ 	while (xstate_bv) {
+ 		if (xstate_bv & 0x1) {
+-		        u32 eax, ebx, ecx, edx, offset;
+-		        cpuid_count(0xD, feature_bit, &eax, &ebx, &ecx, &edx);
++			struct cpuid_xstate_sizes *xs = &xstate_sizes[feature_bit];
++			u32 offset;
++
+ 			/* ECX[1]: 64B alignment in compacted form */
+ 			if (compacted)
+-				offset = (ecx & 0x2) ? ALIGN(ret, 64) : ret;
++				offset = (xs->ecx & 0x2) ? ALIGN(ret, 64) : ret;
+ 			else
+-				offset = ebx;
+-			ret = max(ret, offset + eax);
++				offset = xs->ebx;
++			ret = max(ret, offset + xs->eax);
+ 		}
+ 
+ 		xstate_bv >>= 1;
+diff --git a/arch/x86/kvm/cpuid.h b/arch/x86/kvm/cpuid.h
+index 07da153802e4d4f75219d77596e323347393eab7..a50b57e5d40073027901249a355b65b9a8b54a74 100644
+--- a/arch/x86/kvm/cpuid.h
++++ b/arch/x86/kvm/cpuid.h
+@@ -34,6 +34,7 @@ bool kvm_cpuid(struct kvm_vcpu *vcpu, u32 *eax, u32 *ebx,
+ 
+ bool vcpu_supports_xsave_pkru(struct kvm_vcpu *vcpu);
+ 
++void __init kvm_init_xstate_sizes(void);
+ u32 xstate_required_size(u64 xstate_bv, bool compacted);
+ 
+ int cpuid_query_maxphyaddr(struct kvm_vcpu *vcpu);
+diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
+index 08ef6f01bf7e8c9c6cdb842f23adac09195c1aff..558e267ae0def34227f6f854a0955a26766a6e2c 100644
+--- a/arch/x86/kvm/x86.c
++++ b/arch/x86/kvm/x86.c
+@@ -13930,6 +13930,8 @@ EXPORT_TRACEPOINT_SYMBOL_GPL(kvm_vmgexit_msr_protocol_exit);
+ 
+ static int __init kvm_x86_init(void)
+ {
++	kvm_init_xstate_sizes();
++
+ 	kvm_mmu_x86_module_init();
+ 	mitigate_smt_rsb &= boot_cpu_has_bug(X86_BUG_SMT_RSB) && cpu_smt_possible();
+ 	return 0;
-- 
2.39.5



_______________________________________________
pve-devel mailing list
pve-devel@lists.proxmox.com
https://lists.proxmox.com/cgi-bin/mailman/listinfo/pve-devel

Re: [pve-devel] [PATCH master kernel 1/4] patches: kernel: switch to using full index for patch files

[Fiona Ebner]

Am 19.12.24 um 15:29 schrieb Fiona Ebner:
> git will automatically change the length of the index hashes when
> formatting a patch depending on what references are present in the
> submodule. After pulling in the stable tags today, git wanted to add
> a character to all hashes for me. Use --full-index when generating the
> patches to avoid such issues in the future.
> 
> No functional change intended.
> 
> Signed-off-by: Fiona Ebner <f.ebner@proxmox.com>
> ---
>  ...ke-mkcompile_h-accept-an-alternate-timestamp-strin.patch | 2 +-
>  ...02-wireless-Add-Debian-wireless-regdb-certificates.patch | 2 +-
>  .../0003-bridge-keep-MAC-of-first-assigned-port.patch       | 2 +-
>  ...i-Enable-overrides-for-missing-ACS-capabilities-4..patch | 4 ++--
>  ...05-kvm-disable-default-dynamic-halt-polling-growth.patch | 2 +-
>  ...t-core-downgrade-unregister_netdevice-refcount-lea.patch | 2 +-
>  .../0007-Revert-fortify-Do-not-cast-to-unsigned-char.patch  | 2 +-
>  ...m-xsave-set-mask-out-PKRU-bit-in-xfeatures-if-vCPU.patch | 6 +++---
>  ...low-opt-in-to-allow-pass-through-on-broken-hardwar.patch | 2 +-
>  .../0010-KVM-nSVM-Advertise-support-for-flush-by-ASID.patch | 2 +-
>  ...vert-memfd-improve-userspace-warnings-for-missing-.patch | 2 +-
>  ...12-apparmor-expect-msg_namelen-0-for-recvmsg-calls.patch | 2 +-
>  ..._uring-rw-treat-EOPNOTSUPP-for-IOCB_NOWAIT-like-EA.patch | 2 +-
>  ...0014-netfs-reset-subreq-iov-iter-before-tail-clean.patch | 2 +-
>  ...6-CPU-AMD-Clear-virtualized-VMLOAD-VMSAVE-on-Zen4-.patch | 2 +-
>  ...-PCI-Fix-pci_enable_acs-support-for-the-ACS-quirks.patch | 2 +-
>  ...vert-UBUNTU-SAUCE-iommu-intel-disable-DMAR-for-SKL.patch | 2 +-
>  17 files changed, 20 insertions(+), 20 deletions(-)
> 

Fabian pointed out we have:
https://git.proxmox.com/?p=pve-kernel.git;a=blob;f=debian/scripts/export-patchqueue;h=bc3cea8d0fac054ca83637299a9d93e6ac54079c;hb=refs/heads/master
that should be adapted too ;)

Well I had used this command manually since Thomas gave it to me once
when I started working on QEMU (or at least I forgot about the kernel
scripts existence since then :P).


_______________________________________________
pve-devel mailing list
pve-devel@lists.proxmox.com
https://lists.proxmox.com/cgi-bin/mailman/listinfo/pve-devel