- Kernel - mailweb.openeuler.org

[PATCH OLK-5.10 00/19] x86/speculation: Add force option to GDS mitigation
by Zeng Heng 01 Sep '23

01 Sep '23

Borislav Petkov (1): x86/cpu: Init AP exception handling from cpu_init_secondary() Daniel Sneddon (3): x86/speculation: Add force option to GDS mitigation x86/speculation: Add Kconfig option for GDS KVM: Add GDS_NO support to KVM Dave Hansen (1): Documentation/x86: Fix backwards on/off logic about YMM support Juergen Gross (2): x86/xen: Fix secondary processors' FPU initialization x86/mm: fix poking_init() for Xen PV guests Peter Zijlstra (3): x86/mm: Use mm_alloc() in poking_init() mm: Move mm_cachep initialization to mm_init() x86/mm: Initialize text poking earlier Thomas Gleixner (9): init: Provide arch_cpu_finalize_init() x86/cpu: Switch to arch_cpu_finalize_init() ARM: cpu: Switch to arch_cpu_finalize_init() init: Remove check_bugs() leftovers init: Invoke arch_cpu_finalize_init() earlier init, x86: Move mem_encrypt_init() into arch_cpu_finalize_init() x86/fpu: Remove cpuinfo argument from init functions x86/fpu: Mark init functions __init x86/fpu: Move FPU initialization into arch_cpu_finalize_init() .../hw-vuln/gather_data_sampling.rst | 18 ++- .../admin-guide/kernel-parameters.txt | 8 +- arch/Kconfig | 3 + arch/alpha/include/asm/bugs.h | 20 ---- arch/arm/Kconfig | 1 + arch/arm/include/asm/bugs.h | 4 - arch/arm/kernel/bugs.c | 3 +- arch/parisc/include/asm/bugs.h | 20 ---- arch/powerpc/include/asm/bugs.h | 15 --- arch/x86/Kconfig | 20 ++++ arch/x86/include/asm/bugs.h | 2 - arch/x86/include/asm/fpu/api.h | 2 +- arch/x86/include/asm/mem_encrypt.h | 7 +- arch/x86/include/asm/processor.h | 1 + arch/x86/kernel/cpu/bugs.c | 82 ++++++-------- arch/x86/kernel/cpu/common.c | 104 +++++++++++++++--- arch/x86/kernel/cpu/cpu.h | 1 + arch/x86/kernel/fpu/init.c | 8 +- arch/x86/kernel/smpboot.c | 3 +- arch/x86/kernel/traps.c | 4 +- arch/x86/kvm/x86.c | 7 +- arch/x86/mm/init.c | 7 +- arch/x86/xen/smp_pv.c | 2 + arch/xtensa/include/asm/bugs.h | 18 --- include/asm-generic/bugs.h | 11 -- include/linux/cpu.h | 6 + include/linux/sched/task.h | 2 +- init/main.c | 21 +--- kernel/fork.c | 37 +++---- 29 files changed, 222 insertions(+), 215 deletions(-) delete mode 100644 arch/alpha/include/asm/bugs.h delete mode 100644 arch/parisc/include/asm/bugs.h delete mode 100644 arch/powerpc/include/asm/bugs.h delete mode 100644 arch/xtensa/include/asm/bugs.h delete mode 100644 include/asm-generic/bugs.h -- 2.25.1

2 20

[PATCH openEuler-1.0-LTS] fuse: revalidate: don't invalidate if interrupted
by ZhaoLong Wang 31 Aug '23

31 Aug '23

From: Miklos Szeredi <mszeredi(a)redhat.com> stable inclusion from stable-v4.19.201 commit 00260209e140aae7f1a274b27e29fa6b4e986f95 category: bugfix bugzilla: https://gitee.com/openeuler/kernel/issues/I7XMB3 CVE: NA Reference: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id… -------------------------------- commit a9d1c4c6df0e568207907c04aed9e7beb1294c42 upstream. If the LOOKUP request triggered from fuse_dentry_revalidate() is interrupted, then the dentry will be invalidated, possibly resulting in submounts being unmounted. Reported-by: Xu Rongbo <xurongbo(a)baidu.com> Closes: https://lore.kernel.org/all/CAJfpegswN_CJJ6C3RZiaK6rpFmNyWmXfaEpnQUJ42KCwNF… Fixes: 9e6268db496a ("[PATCH] FUSE - read-write operations") Cc: <stable(a)vger.kernel.org> Signed-off-by: Miklos Szeredi <mszeredi(a)redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh(a)linuxfoundation.org> Signed-off-by: ZhaoLong Wang <wangzhaolong1(a)huawei.com> --- fs/fuse/dir.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/fs/fuse/dir.c b/fs/fuse/dir.c index b468ccb296c3..6068955995da 100644 --- a/fs/fuse/dir.c +++ b/fs/fuse/dir.c @@ -232,7 +232,7 @@ static int fuse_dentry_revalidate(struct dentry *entry, unsigned int flags) spin_unlock(&fc->lock); } kfree(forget); - if (ret == -ENOMEM) + if (ret == -ENOMEM || ret == -EINTR) goto out; if (ret || fuse_invalid_attr(&outarg.attr) || (outarg.attr.mode ^ inode->i_mode) & S_IFMT) -- 2.34.3

2 1

[PATCH OLK-5.10 0/2] zs_malloc: return ERR_PTR on failure
by Jinjiang Tu 31 Aug '23

31 Aug '23

1. zs_malloc() returns error code instead of 0 when error happens. 2. check parameter handle in zs_free(). Hui Zhu (1): zsmalloc: zs_malloc: return ERR_PTR on failure Sergey Senozhatsky (1): mm/zsmalloc: do not attempt to free IS_ERR handle drivers/block/zram/zram_drv.c | 10 +++++----- mm/zsmalloc.c | 15 +++++++++------ 2 files changed, 14 insertions(+), 11 deletions(-) -- 2.25.1

2 3

[PATCH v2 openEuler-23.09 0/2] openeuler_defconfig: Enable SDEI Watchdog
by Wupeng Ma 31 Aug '23

31 Aug '23

From: Ma Wupeng <mawupeng1(a)huawei.com> Tweak gmem related configs 1. Remove select config GMEM_DEV for gmem. 2. Enable gmem related configs. Changelog since v1: 1. add license for gmem_dev. Ma Wupeng (2): mm: gmem: Remove select config GMEM_DEV for gmem openeuler_defconfig: Enable gmem related configs arch/arm64/configs/openeuler_defconfig | 2 ++ arch/x86/configs/openeuler_defconfig | 2 ++ drivers/char/gmem_dev.c | 1 + mm/Kconfig | 1 - 4 files changed, 5 insertions(+), 1 deletion(-) -- 2.25.1

2 3

[PATCH v6 openEuler-23.09 00/25] backport etmem feature from OLK-5.10
by liubo 31 Aug '23

31 Aug '23

This patch set is used to backport etmem feature patches to version 23.09. etmem is a tiered memory extension technology that uses DRAM+memory compression/high-performance storage media to form tiered memory storage. Memory data is tiered, and cold data is migrated from memory media to high-performance storage media to release memory space and reduce memory costs. v5 -> v6 - Change the value of sysctl_hugetlb_* to static and delete unnecessary external declarations. v4 -> v5 - fix some review comment during https://gitee.com/openeuler/kernel/pulls/1772 patch v3 -> v4 - fix the commit message format error v2 -> v3 - fix the build problem in riscv64 arch. v1 -> v2 - some cleanup fix Kemeng Shi (11): etmem: add ioctl for mm idle scan etmem: x86: support scan hugetlb of vm etmem_scan: x86: support scan 4 level ept under 5 level host page table etmem scan: fix memleak in vm_idle_read etmem: fix concurrent access to export file operations etmem: fix potential UAF when walk ept page table etmem_scan: release CPU after scan walk_step size etmem_scan: add pte_hole callback acpi/numa: memorize node type from SRAT table hugepage: add sysctl for hugepage alloc and mig x86: hugepage: use nt copy hugepage to AEP in x86 liubo (14): memig: add memig-scan feature to openEuler memig: add memig-swap feature to openEuler memig: fix compile error when CONFIG_NUMA is turned off etmem: Modify the memig feature name to etmem move ETMEM feature CONFIG to mm/Kconfig and add architecture dependency etmem: etmem scan module Replace WARN_ONCE() with debug_printk for "nothing read" add ETMEM feature CONFIG to mm/Kconfig config: enable CONFIG_ETMEM by default etmem: add CONFIG_ETMEM macro definition for etmem feature etmem: add original kernel swap enabled options etmem: add swapcache reclaim to etmem etmem: Add a scan flag to support specified page swap-out etmem: fix the div 0 problem in swapcache reclaim process etmem: fix some problem during backport etmem patches arch/arm64/configs/openeuler_defconfig | 3 + arch/x86/configs/openeuler_defconfig | 3 + arch/x86/include/asm/page_64.h | 7 + arch/x86/lib/Makefile | 1 + arch/x86/lib/copy_highpages.c | 107 ++ arch/x86/lib/copy_page_64.S | 73 ++ drivers/acpi/numa/srat.c | 5 + fs/proc/Makefile | 2 + fs/proc/base.c | 8 + fs/proc/etmem_scan.c | 1382 ++++++++++++++++++++++++ fs/proc/etmem_scan.h | 149 +++ fs/proc/etmem_swap.c | 282 +++++ fs/proc/internal.h | 4 + fs/proc/task_mmu.c | 194 ++++ include/linux/highmem.h | 17 + include/linux/list.h | 17 + include/linux/mm.h | 4 + include/linux/mm_types.h | 17 +- include/linux/numa.h | 12 + include/linux/swap.h | 39 + include/uapi/asm-generic/mman-common.h | 4 + mm/Kconfig | 25 + mm/Makefile | 1 + mm/etmem.c | 383 +++++++ mm/huge_memory.c | 1 + mm/hugetlb.c | 27 +- mm/internal.h | 1 - mm/madvise.c | 17 +- mm/page_alloc.c | 13 + mm/pagewalk.c | 1 + mm/swap_state.c | 3 + mm/util.c | 11 +- mm/vmscan.c | 18 + virt/kvm/kvm_main.c | 6 + 34 files changed, 2824 insertions(+), 13 deletions(-) create mode 100644 arch/x86/lib/copy_highpages.c create mode 100644 fs/proc/etmem_scan.c create mode 100644 fs/proc/etmem_scan.h create mode 100644 fs/proc/etmem_swap.c create mode 100644 mm/etmem.c -- 2.27.0

2 26

[PATCH openEuler-23.09 0/2] Tweak gmem related configs
by Wupeng Ma 31 Aug '23

31 Aug '23

From: Ma Wupeng <mawupeng1(a)huawei.com> 1. Remove select config GMEM_DEV for gmem. 2. Enable gmem related configs. Ma Wupeng (2): mm: gmem: Remove select config GMEM_DEV for gmem openeuler_defconfig: Enable gmem related configs arch/arm64/configs/openeuler_defconfig | 2 ++ arch/x86/configs/openeuler_defconfig | 2 ++ mm/Kconfig | 1 - 3 files changed, 4 insertions(+), 1 deletion(-) -- 2.25.1

2 3

[PATCH v5 openEuler-23.09 00/25] backport etmem feature from OLK-5.10
by liubo 31 Aug '23

31 Aug '23

This patch set is used to backport etmem feature patches to version 23.09. etmem is a tiered memory extension technology that uses DRAM+memory compression/high-performance storage media to form tiered memory storage. Memory data is tiered, and cold data is migrated from memory media to high-performance storage media to release memory space and reduce memory costs. v4 -> v5 - fix some review comment during https://gitee.com/openeuler/kernel/pulls/1772 patch v3 -> v4 - fix the commit message format error v2 -> v3 - fix the build problem in riscv64 arch. v1 -> v2 - some cleanup fix Kemeng Shi (11): etmem: add ioctl for mm idle scan etmem: x86: support scan hugetlb of vm etmem_scan: x86: support scan 4 level ept under 5 level host page table etmem scan: fix memleak in vm_idle_read etmem: fix concurrent access to export file operations etmem: fix potential UAF when walk ept page table etmem_scan: release CPU after scan walk_step size etmem_scan: add pte_hole callback acpi/numa: memorize node type from SRAT table hugepage: add sysctl for hugepage alloc and mig x86: hugepage: use nt copy hugepage to AEP in x86 liubo (14): memig: add memig-scan feature to openEuler memig: add memig-swap feature to openEuler memig: fix compile error when CONFIG_NUMA is turned off etmem: Modify the memig feature name to etmem move ETMEM feature CONFIG to mm/Kconfig and add architecture dependency etmem: etmem scan module Replace WARN_ONCE() with debug_printk for "nothing read" add ETMEM feature CONFIG to mm/Kconfig config: enable CONFIG_ETMEM by default etmem: add CONFIG_ETMEM macro definition for etmem feature etmem: add original kernel swap enabled options etmem: add swapcache reclaim to etmem etmem: Add a scan flag to support specified page swap-out etmem: fix the div 0 problem in swapcache reclaim process etmem: fix some problem during backport etmem patches arch/arm64/configs/openeuler_defconfig | 3 + arch/x86/configs/openeuler_defconfig | 3 + arch/x86/include/asm/page_64.h | 7 + arch/x86/lib/Makefile | 1 + arch/x86/lib/copy_highpages.c | 107 ++ arch/x86/lib/copy_page_64.S | 73 ++ drivers/acpi/numa/srat.c | 5 + fs/proc/Makefile | 2 + fs/proc/base.c | 8 + fs/proc/etmem_scan.c | 1382 ++++++++++++++++++++++++ fs/proc/etmem_scan.h | 149 +++ fs/proc/etmem_swap.c | 282 +++++ fs/proc/internal.h | 4 + fs/proc/task_mmu.c | 194 ++++ include/linux/highmem.h | 17 + include/linux/hugetlb.h | 6 + include/linux/list.h | 17 + include/linux/mm.h | 4 + include/linux/mm_types.h | 17 +- include/linux/numa.h | 12 + include/linux/swap.h | 39 + include/uapi/asm-generic/mman-common.h | 4 + mm/Kconfig | 25 + mm/Makefile | 1 + mm/etmem.c | 383 +++++++ mm/huge_memory.c | 1 + mm/hugetlb.c | 27 +- mm/internal.h | 1 - mm/madvise.c | 17 +- mm/page_alloc.c | 13 + mm/pagewalk.c | 1 + mm/swap_state.c | 3 + mm/util.c | 11 +- mm/vmscan.c | 19 + virt/kvm/kvm_main.c | 6 + 35 files changed, 2831 insertions(+), 13 deletions(-) create mode 100644 arch/x86/lib/copy_highpages.c create mode 100644 fs/proc/etmem_scan.c create mode 100644 fs/proc/etmem_scan.h create mode 100644 fs/proc/etmem_swap.c create mode 100644 mm/etmem.c -- 2.27.0

2 26

[PATCH v3 openEuler-23.09 00/46] ima: Add IMA digest lists extension
by Zhou Shuiqing 31 Aug '23

31 Aug '23

This patchset supports IMA digest lists for the kernel. v2: -remove unused variable in ima_main.c v3: -modify patch header information David Howells (4): PGPLIB: PGP definitions (RFC 4880) PGPLIB: Basic packet parser KEYS: Provide PGP key description autogeneration KEYS: Provide a function to load keys from a PGP keyring blob Mimi Zohar (1): initramfs: add file metadata Roberto Sassu (32): initramfs: read metadata from special file METADATA!!! gen_init_cpio: add support for file metadata init: Add kernel option to force usage of tmpfs for rootfs ima: Add enforce-evm and log-evm modes to strictly check EVM status ima: Generalize ima_read_policy() ima: Generalize ima_write_policy() and raise uploaded data size limit ima: Generalize policy file operations ima: Use ima_show_htable_value to show violations and hash table data ima: Add parser of compact digest list ima: Prevent usage of digest lists not measured or appraised ima: Introduce new securityfs files ima: Introduce new hook DIGEST_LIST_CHECK ima: Load all digest lists from a directory at boot time ima: Add support for measurement with digest lists ima: Add support for appraisal with digest lists evm: Add support for digest lists of metadata ima: Introduce exec_tcb policy ima: Introduce appraise_exec_tcb policy ima: Introduce appraise_exec_immutable policy ima: Add Documentation/security/IMA-digest-lists.txt mpi: introduce mpi_key_length() rsa: add parser of raw format KEYS: PGP data parser KEYS: Introduce load_pgp_public_keyring() certs: Introduce search_trusted_key() ima: Search key in the built-in keyrings ima: Allow direct upload of digest lists to securityfs ima: Add parser keyword to the policy evm: Extend evm= with x509. allow_metadata_writes and complete values ima: Execute parser to upload digest lists not recognizable by the kernel evm: Propagate choice of HMAC algorithm in evm_crypto.c config: add digest list options for arm64 and x86 Zhang Tianxing (5): ima: fix a memory leak in ima_del_digest_data_entry ima: Add max size for IMA digest database ima: don't allow control characters in policy path ima: fix CONFIG_IMA_DIGEST_DB_MEGABYTES in openeuler_defconfig ima: fix db size overflow and Kconfig issues Zheng Zengkai (1): Revert "evm: Refuse EVM_ALLOW_METADATA_WRITES only if an HMAC key is loaded" Zhou Shuiqing (2): ima: Allow choice of file hash algorithm for measurement and audit ima: Add meta_immutable appraisal type shenxiangwei (1): ima: bugfix for digest lists importing Documentation/ABI/testing/evm | 4 +- .../admin-guide/kernel-parameters.txt | 49 +- Documentation/security/IMA-digest-lists.txt | 259 ++++++++++ arch/arm64/configs/openeuler_defconfig | 11 + arch/x86/configs/openeuler_defconfig | 31 +- certs/Kconfig | 7 + certs/Makefile | 7 + certs/system_certificates.S | 18 + certs/system_keyring.c | 44 ++ crypto/asymmetric_keys/Kconfig | 25 + crypto/asymmetric_keys/Makefile | 10 + crypto/asymmetric_keys/pgp_library.c | 281 +++++++++++ crypto/asymmetric_keys/pgp_parser.h | 23 + crypto/asymmetric_keys/pgp_preload.c | 119 +++++ crypto/asymmetric_keys/pgp_public_key.c | 383 +++++++++++++++ crypto/rsa.c | 14 +- crypto/rsa_helper.c | 69 +++ include/crypto/internal/rsa.h | 6 + include/linux/initramfs.h | 21 + include/linux/kernel_read_file.h | 1 + include/linux/mpi.h | 2 + include/linux/pgp.h | 220 +++++++++ include/linux/pgplib.h | 48 ++ include/linux/verification.h | 4 + init/do_mounts.c | 11 +- init/initramfs.c | 138 +++++- lib/mpi/mpicoder.c | 33 +- security/integrity/digsig_asymmetric.c | 10 + security/integrity/evm/Kconfig | 32 ++ security/integrity/evm/evm.h | 1 + security/integrity/evm/evm_crypto.c | 24 +- security/integrity/evm/evm_main.c | 103 +++- security/integrity/evm/evm_secfs.c | 2 +- security/integrity/iint.c | 2 + security/integrity/ima/Kconfig | 49 ++ security/integrity/ima/Makefile | 1 + security/integrity/ima/ima.h | 39 +- security/integrity/ima/ima_api.c | 42 +- security/integrity/ima/ima_appraise.c | 117 ++++- security/integrity/ima/ima_digest_list.c | 465 ++++++++++++++++++ security/integrity/ima/ima_digest_list.h | 54 ++ security/integrity/ima/ima_fs.c | 294 ++++++++--- security/integrity/ima/ima_init.c | 2 +- security/integrity/ima/ima_main.c | 125 ++++- security/integrity/ima/ima_policy.c | 117 ++++- security/integrity/integrity.h | 47 +- usr/Kconfig | 8 + usr/Makefile | 4 +- usr/gen_init_cpio.c | 136 ++++- usr/gen_initramfs.sh | 7 +- 50 files changed, 3300 insertions(+), 219 deletions(-) create mode 100644 Documentation/security/IMA-digest-lists.txt create mode 100644 crypto/asymmetric_keys/pgp_library.c create mode 100644 crypto/asymmetric_keys/pgp_parser.h create mode 100644 crypto/asymmetric_keys/pgp_preload.c create mode 100644 crypto/asymmetric_keys/pgp_public_key.c create mode 100644 include/linux/initramfs.h create mode 100644 include/linux/pgp.h create mode 100644 include/linux/pgplib.h create mode 100644 security/integrity/ima/ima_digest_list.c create mode 100644 security/integrity/ima/ima_digest_list.h -- 2.33.0

1 46

[PATCH openEuler-1.0-LTS] sched: Add 'affinity_preferred_nodes' for smart grid
by Hui Tang 31 Aug '23

31 Aug '23

hulk inclusion category: feature bugzilla: https://gitee.com/openeuler/kernel/issues/I7G6SW CVE: NA ------------------------------- Add sysctl 'affinity_preferred_nodes' to set preferred nodes. For example: echo 0,3 > /proc/sys/kernel/affinity_preferred_nodes The preferred nodes is preferentially selected to init affinity domain if util is lower than 85%. If the util is higher than 85% and less than 15% higher than the non-preferred node, the node is also preferentially selected. Signed-off-by: Hui Tang <tanghui20(a)huawei.com> --- include/linux/sched/sysctl.h | 4 + kernel/sched/fair.c | 177 +++++++++++++++++------------------ kernel/sched/sched.h | 1 + kernel/sysctl.c | 7 ++ 4 files changed, 98 insertions(+), 91 deletions(-) diff --git a/include/linux/sched/sysctl.h b/include/linux/sched/sysctl.h index ad472760e97d7..8475ee46727ac 100644 --- a/include/linux/sched/sysctl.h +++ b/include/linux/sched/sysctl.h @@ -38,6 +38,10 @@ extern int sysctl_sched_util_low_pct; #ifdef CONFIG_QOS_SCHED_SMART_GRID extern int sysctl_affinity_adjust_delay_ms; +extern unsigned long *smart_grid_preferred_nodemask_bits; +extern int proc_cpu_affinity_domain_nodemask(struct ctl_table *table, int write, + void __user *buffer, size_t *lenp, + loff_t *ppos); #endif enum sched_tunable_scaling { diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c index e9afb1e6ca4ca..571123e2a23a6 100644 --- a/kernel/sched/fair.c +++ b/kernel/sched/fair.c @@ -5296,6 +5296,9 @@ static unsigned long weighted_cpuload(struct rq *rq); static inline bool prefer_cpus_valid(struct task_struct *p); int sysctl_affinity_adjust_delay_ms = 5000; +nodemask_t smart_grid_preferred_nodemask; +unsigned long *smart_grid_preferred_nodemask_bits = + nodes_addr(smart_grid_preferred_nodemask); struct static_key __smart_grid_used; @@ -5494,90 +5497,6 @@ void stop_auto_affinity(struct auto_affinity *auto_affi) mutex_unlock(&smart_grid_used_mutex); } -static struct sched_group *sd_find_idlest_group(struct sched_domain *sd) -{ - struct sched_group *idlest = NULL, *group = sd->groups; - unsigned long min_runnable_load = ULONG_MAX; - unsigned long min_avg_load = ULONG_MAX; - int imbalance_scale = 100 + (sd->imbalance_pct-100)/2; - unsigned long imbalance = scale_load_down(NICE_0_LOAD) * - (sd->imbalance_pct-100) / 100; - - do { - unsigned long load, avg_load, runnable_load; - int i; - - avg_load = 0; - runnable_load = 0; - - for_each_cpu(i, sched_group_span(group)) { - load = target_load(i, 0); - runnable_load += load; - avg_load += cfs_rq_load_avg(&cpu_rq(i)->cfs); - } - - avg_load = (avg_load * SCHED_CAPACITY_SCALE) / - group->sgc->capacity; - runnable_load = (runnable_load * SCHED_CAPACITY_SCALE) / - group->sgc->capacity; - - if (min_runnable_load > (runnable_load + imbalance)) { - min_runnable_load = runnable_load; - min_avg_load = avg_load; - idlest = group; - } else if ((runnable_load < (min_runnable_load + imbalance)) && - (100*min_avg_load > imbalance_scale*avg_load)) { - min_avg_load = avg_load; - idlest = group; - } - } while (group = group->next, group != sd->groups); - - return idlest ? idlest : group; -} - -static int group_find_idlest_cpu(struct sched_group *group) -{ - int least_loaded_cpu = cpumask_first(sched_group_span(group)); - unsigned long load, min_load = ULONG_MAX; - unsigned int min_exit_latency = UINT_MAX; - u64 latest_idle_timestamp = 0; - int shallowest_idle_cpu = -1; - int i; - - if (group->group_weight == 1) - return least_loaded_cpu; - - for_each_cpu(i, sched_group_span(group)) { - if (sched_idle_cpu(i)) - return i; - - if (available_idle_cpu(i)) { - struct rq *rq = cpu_rq(i); - struct cpuidle_state *idle = idle_get_state(rq); - - if (idle && idle->exit_latency < min_exit_latency) { - min_exit_latency = idle->exit_latency; - latest_idle_timestamp = rq->idle_stamp; - shallowest_idle_cpu = i; - } else if ((!idle || - idle->exit_latency == min_exit_latency) && - rq->idle_stamp > latest_idle_timestamp) { - latest_idle_timestamp = rq->idle_stamp; - shallowest_idle_cpu = i; - } - } else if (shallowest_idle_cpu == -1) { - load = weighted_cpuload(cpu_rq(i)); - if (load < min_load) { - min_load = load; - least_loaded_cpu = i; - } - } - } - - return shallowest_idle_cpu != -1 ? shallowest_idle_cpu : - least_loaded_cpu; -} - void free_affinity_domains(struct affinity_domain *ad) { int i; @@ -5612,14 +5531,72 @@ static int init_affinity_domains_orig(struct affinity_domain *ad) return -ENOMEM; } +struct nid_stats { + unsigned long util; + unsigned long compute_capacity; +}; + +static inline void update_nid_stats(struct nid_stats *ns, int nid) +{ + int cpu; + + memset(ns, 0, sizeof(*ns)); + for_each_cpu(cpu, cpumask_of_node(nid)) { + ns->compute_capacity += capacity_of(cpu); + ns->util += cpu_util(cpu); + } +} + +static int find_idlest_nid(void) +{ + int nid, imbalance_pct, is_prefer; + unsigned long long util_min = UINT_MAX; + int idlest_is_prefer = 0; + struct nid_stats ns; + int idlest_nid = 0; + + for_each_online_node(nid) { + if (!cpumask_intersects(cpumask_of_node(nid), + housekeeping_cpumask(HK_FLAG_DOMAIN))) + continue; + + update_nid_stats(&ns, nid); + + if (node_isset(nid, smart_grid_preferred_nodemask)) { + if (ns.util * 100 < + ns.compute_capacity * sysctl_sched_util_low_pct) { + idlest_nid = nid; + break; + } + is_prefer = 1; + } else { + is_prefer = 0; + } + + if (is_prefer && !idlest_is_prefer) + imbalance_pct = 117; // higher 15% + else if (!is_prefer && idlest_is_prefer) + imbalance_pct = 85; // lower 15% + else + imbalance_pct = 100; + + if (ns.util * 100 < util_min * imbalance_pct) { + util_min = ns.util * 100 / imbalance_pct; + idlest_nid = nid; + idlest_is_prefer = is_prefer; + } + } + + return idlest_nid; +} + static int init_affinity_domains(struct affinity_domain *ad) { - struct sched_domain *sd = NULL, *tmp; - struct sched_group *idlest = NULL; + struct sched_domain *tmp; + int cpu, idlest_nid; int ret = -ENOMEM; int dcount = 0; int i = 0; - int cpu; for (i = 0; i < AD_LEVEL_MAX; i++) { ad->domains[i] = kmalloc(sizeof(cpumask_t), GFP_KERNEL); @@ -5631,18 +5608,18 @@ static int init_affinity_domains(struct affinity_domain *ad) cpu = cpumask_first_and(cpu_active_mask, housekeeping_cpumask(HK_FLAG_DOMAIN)); for_each_domain(cpu, tmp) { - sd = tmp; dcount++; } - if (!sd || dcount > AD_LEVEL_MAX) { + if (dcount > AD_LEVEL_MAX) { rcu_read_unlock(); ret = -EINVAL; goto err; } - idlest = sd_find_idlest_group(sd); - cpu = group_find_idlest_cpu(idlest); + idlest_nid = find_idlest_nid(); + cpu = cpumask_first_and(cpumask_of_node(idlest_nid), + housekeeping_cpumask(HK_FLAG_DOMAIN)); i = 0; for_each_domain(cpu, tmp) { cpumask_copy(ad->domains[i], sched_domain_span(tmp)); @@ -5711,6 +5688,24 @@ static void destroy_auto_affinity(struct task_group *tg) kfree(tg->auto_affinity); tg->auto_affinity = NULL; } + +int proc_cpu_affinity_domain_nodemask(struct ctl_table *table, int write, + void __user *buffer, size_t *lenp, + loff_t *ppos) +{ + int err; + + mutex_lock(&smart_grid_used_mutex); + + err = proc_do_large_bitmap(table, write, buffer, lenp, ppos); + if (!err && write) + nodes_and(smart_grid_preferred_nodemask, + smart_grid_preferred_nodemask, + node_online_map); + + mutex_unlock(&smart_grid_used_mutex); + return err; +} #else static void destroy_auto_affinity(struct task_group *tg) {} diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h index 1d882a2b8d5fe..ef3f5c84e8b95 100644 --- a/kernel/sched/sched.h +++ b/kernel/sched/sched.h @@ -509,6 +509,7 @@ extern void sched_offline_group(struct task_group *tg); extern void sched_move_task(struct task_struct *tsk); #ifdef CONFIG_QOS_SCHED_SMART_GRID +extern nodemask_t smart_grid_preferred_nodemask; extern void start_auto_affinity(struct auto_affinity *auto_affi); extern void stop_auto_affinity(struct auto_affinity *auto_affi); extern int init_auto_affinity(struct task_group *tg); diff --git a/kernel/sysctl.c b/kernel/sysctl.c index c7064f67f4a53..b0656ef92a2df 100644 --- a/kernel/sysctl.c +++ b/kernel/sysctl.c @@ -1347,6 +1347,13 @@ static struct ctl_table kern_table[] = { .extra1 = &zero, .extra2 = &hundred_thousand, }, + { + .procname = "affinity_preferred_nodes", + .data = &smart_grid_preferred_nodemask_bits, + .maxlen = MAX_NUMNODES, + .mode = 0644, + .proc_handler = proc_cpu_affinity_domain_nodemask, + }, #endif { } }; -- 2.17.1

2 1

[PATCH OLK-5.10 00/12] Some misc changes
by Yihang Li 31 Aug '23

31 Aug '23

From: xiabing <xiabing12(a)h-partners.com> Update the Openeuler community patch information to ensure that it is consistent with the Linux mainline community patch information. John Garry (1): scsi: hisi_sas: Revert change to limit max hw sectors for v3 HW Luo Jiaxing (3): scsi: hisi_sas: Increase debugfs_dump_index after dump is completed scsi: libsas: Export sas_phy_enable() scsi: hisi_sas: Disable SATA disk phy for severe I_T nexus reset failure Yihang Li (2): scsi: hisi_sas: Revert "scsi: hisi_sas: Disable SATA disk phy for severe I_T nexus reset failure" scsi: hisi_sas: Configure initial value of some registers according to HBA model xiabing (6): scsi: hisi_sas: Revert "Increase-debugfs_dump_index-after-dump" scsi: libsas: Revert "Export-sas_phy_enable" scsi: hisi_sas: Revert "Revert "scsi: hisi_sas: Disable SATA disk phy for severe I_T nexus reset failure"" scsi: hisi_sas: Revert "Disable-SATA-disk-phy-for-severe-I_T-nexus reset failure" scsi: hisi_sas: Revert "scsi: hisi_sas: Revert "scsi: hisi_sas: Limit max hw sectors for v3 HW"" scsi: hisi_sas: Revert "scsi: hisi_sas: Configure the initialization registers according to HBA model" drivers/scsi/hisi_sas/hisi_sas_v3_hw.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) -- 2.30.0

2 13

2024

2023

2022

2021

2020

2019

Kernel