- Kernel - mailweb.openeuler.org

[PATCH OLK-6.6] tcp/dccp: Add another way to allocate local ports in connect()
by Zhengchao Shao 29 Nov '23

29 Nov '23

From: Lu Wei <luwei32(a)huawei.com> hulk inclusion category: bugfix bugzilla: https://gitee.com/openeuler/kernel/issues/I8KBYH -------------------------------- Commit 07f4c90062f8 ("tcp/dccp: try to not exhaust ip_local_port_range in connect()") allocates even ports for connect() first while leaving odd ports for bind() and this works well in busy servers. But this strategy causes severe performance degradation in busy clients. when a client has used more than half of the local ports setted in proc/sys/net/ipv4/ip_local_port_range, if this client trys to connect to a server again, the connect time increases rapidly since it will traverse all the even ports though they are exhausted. So this path provides another strategy by introducing a system option: local_port_allocation. If it is a busy client, users should set it to 1 to use sequential allocation while it should be set to 0 in other situations. Its default value is 0. Signed-off-by: Lu Wei <luwei32(a)huawei.com> Signed-off-by: Liu Jian <liujian56(a)huawei.com> Conflicts: include/net/tcp.h net/ipv4/inet_hashtables.c net/ipv4/sysctl_net_ipv4.c Signed-off-by: Zhengchao Shao <shaozhengchao(a)huawei.com> --- include/net/tcp.h | 1 + net/ipv4/inet_hashtables.c | 11 ++++++++--- net/ipv4/sysctl_net_ipv4.c | 8 ++++++++ 3 files changed, 17 insertions(+), 3 deletions(-) diff --git a/include/net/tcp.h b/include/net/tcp.h index 0239e815edf7..e9d387fffe22 100644 --- a/include/net/tcp.h +++ b/include/net/tcp.h @@ -258,6 +258,7 @@ DECLARE_PER_CPU(int, tcp_memory_per_cpu_fw_alloc); extern struct percpu_counter tcp_sockets_allocated; extern unsigned long tcp_memory_pressure; +extern int sysctl_local_port_allocation; /* optimized version of sk_under_memory_pressure() for TCP sockets */ static inline bool tcp_under_memory_pressure(const struct sock *sk) diff --git a/net/ipv4/inet_hashtables.c b/net/ipv4/inet_hashtables.c index 598c1b114d2c..919f0f869118 100644 --- a/net/ipv4/inet_hashtables.c +++ b/net/ipv4/inet_hashtables.c @@ -1011,7 +1011,7 @@ int __inet_hash_connect(struct inet_timewait_death_row *death_row, struct inet_bind_bucket *tb; bool tb_created = false; u32 remaining, offset; - int ret, i, low, high; + int ret, i, low, high, span_size; int l3mdev; u32 index; @@ -1021,6 +1021,11 @@ int __inet_hash_connect(struct inet_timewait_death_row *death_row, local_bh_enable(); return ret; } + /* local_port_allocation 0 means even and odd port allocation strategy + * will be applied, so span size is 2; otherwise sequential allocation + * will be used and span size is 1. Default value is 0. + */ + span_size = sysctl_local_port_allocation ? 1 : 2; l3mdev = inet_sk_bound_l3mdev(sk); @@ -1043,7 +1048,7 @@ int __inet_hash_connect(struct inet_timewait_death_row *death_row, offset &= ~1U; other_parity_scan: port = low + offset; - for (i = 0; i < remaining; i += 2, port += 2) { + for (i = 0; i < remaining; i += span_size, port += span_size) { if (unlikely(port >= high)) port -= remaining; if (inet_is_local_reserved_port(net, port)) @@ -1084,7 +1089,7 @@ int __inet_hash_connect(struct inet_timewait_death_row *death_row, } offset++; - if ((offset & 1) && remaining > 1) + if ((offset & 1) && remaining > 1 && span_size == 2) goto other_parity_scan; return -EADDRNOTAVAIL; diff --git a/net/ipv4/sysctl_net_ipv4.c b/net/ipv4/sysctl_net_ipv4.c index 6ac890b4073f..b17eb28a9690 100644 --- a/net/ipv4/sysctl_net_ipv4.c +++ b/net/ipv4/sysctl_net_ipv4.c @@ -39,6 +39,7 @@ static unsigned long ip_ping_group_range_min[] = { 0, 0 }; static unsigned long ip_ping_group_range_max[] = { GID_T_MAX, GID_T_MAX }; static u32 u32_max_div_HZ = UINT_MAX / HZ; static int one_day_secs = 24 * 3600; +int sysctl_local_port_allocation; static u32 fib_multipath_hash_fields_all_mask __maybe_unused = FIB_MULTIPATH_HASH_FIELD_ALL_MASK; static unsigned int tcp_child_ehash_entries_max = 16 * 1024 * 1024; @@ -579,6 +580,13 @@ static struct ctl_table ipv4_table[] = { .extra1 = &sysctl_fib_sync_mem_min, .extra2 = &sysctl_fib_sync_mem_max, }, + { + .procname = "local_port_allocation", + .data = &sysctl_local_port_allocation, + .maxlen = sizeof(int), + .mode = 0644, + .proc_handler = proc_dointvec, + }, { } }; -- 2.34.1

2 1

[PACH OLK-5.10 v5 00/20] psi irq in cgroupv1 and psi fine grained
by Lu Jialin 29 Nov '23

29 Nov '23

Chen Wandun (3): mm: disable psi cgroup v1 by default mm: add config isolation for psi under cgroup v1 psi: dont alloc memory for psi by default Chengming Zhou (3): sched/psi: Fix periodic aggregation shut off sched/psi: Optimize task switch inside shared cgroups again sched/psi: Add PSI_IRQ to track IRQ/SOFTIRQ pressure Hao Jia (1): sched/psi: Zero the memory of struct psi_group Johannes Weiner (1): sched/psi: Remove NR_ONCPU task accounting Lu Jialin (11): psi: support irq.pressure under cgroup v1 psi: enable CONFIG_PSI_CGROUP_V1 in openeuler_defconfig psi: update psi irqtime when the irq delta is nozero memcg: split async memcg reclaim from reclaim_high psi: add struct psi_group_ext PSI: Introduce fine grained stall time collect for cgroup reclaim PSI: Introduce avgs and total calculation for cgroup reclaim PSI: Introduce pressure.stat in psi PSI: add more memory fine grained stall tracking in pressure.stat add cpu fine grained stall tracking in pressure.stat PSI: enable CONFIG_PSI_FINE_GRAINED in openeuler_defconfig Suren Baghdasaryan (1): psi: Fix "defined but not used" warnings when CONFIG_PROC_FS=n Documentation/admin-guide/cgroup-v2.rst | 6 + arch/arm64/configs/openeuler_defconfig | 2 + arch/x86/configs/openeuler_defconfig | 2 + block/blk-cgroup.c | 2 +- block/blk-core.c | 2 +- include/linux/cgroup-defs.h | 6 +- include/linux/cgroup.h | 2 +- include/linux/psi.h | 9 +- include/linux/psi_types.h | 99 +++- include/linux/sched.h | 4 + init/Kconfig | 20 + kernel/cgroup/cgroup.c | 64 ++- kernel/sched/core.c | 2 + kernel/sched/cpuacct.c | 6 +- kernel/sched/fair.c | 6 - kernel/sched/psi.c | 691 ++++++++++++++++++++---- kernel/sched/stats.h | 9 + mm/compaction.c | 2 +- mm/filemap.c | 4 +- mm/memcontrol.c | 92 ++-- mm/page_alloc.c | 6 + mm/page_io.c | 3 + mm/vmscan.c | 5 +- 23 files changed, 873 insertions(+), 171 deletions(-) -- 2.34.1

1 20

[PATCH OLK-5.10] livepatch/powerpc: Fix issue that miss one layer on stack checking
by Zheng Yejian 29 Nov '23

29 Nov '23

hulk inclusion category: bugfix bugzilla: https://gitee.com/openeuler/kernel/issues/I7BMC9 CVE: NA -------------------------------- Suppose task 'T' is running in function 'B', and user is doing livepatch for function 'A', and function 'A' is calling 'B'. Then task 'T' will be interrupted by stop_machine, and suppose that 'B' has not saved the link register (which point to address in 'A') into stack memory, and data in link register position may be invalid, then 'A' would be missed on stack checking. To find a solution, we list following cases (suppose calling chain: P -> A -> B, A runs `bl B`, 'nip' means instruction pointer found in interrupt frame, 'lr' means link register found in interrupt frame, 'frame->pc' means link register saved in first stack frame): 1. 'nip' is in 'A', `bl B` is not executed, then: 'lr' is in 'P', 'frame->pc' should be ignored; 2. 'nip' is in some ppc64_stub or ppc32_plt which also not save link register, then: 'lr' is in 'A', 'frame->pc' should be ignored; 3. 'nip' is in 'B', but 'B' has not saved link register, then: 'lr' is in 'A', 'frame->pc' should be ignored; 4. 'nip' is in 'B', 'B' has saved link register, then: 'lr' is in 'A', 'frame->pc' is also in 'A'; 5. 'nip' is in 'A', 'B' has returned but its stack pointer is not moved, then 'lr' is in 'A', 'frame->pc' is also in 'A'; 6. 'nip' is in 'A', 'B' has returned and its stack pointer is moved, then 'lr' is in 'A', 'frame->pc' is in 'P'. As a conclusion, we need to: 1. check 'nip' or 'lr' or both if they are not in same function; 2. ignore 'frame->pc' if 'nip' and 'lr' are not in same function. Signed-off-by: Zheng Yejian <zhengyejian1(a)huawei.com> --- arch/powerpc/include/asm/livepatch.h | 11 ++++++++++ arch/powerpc/kernel/livepatch.c | 28 +++++++++++++++++++++++-- arch/powerpc/kernel/livepatch_32.c | 31 +++++++++++++++++++--------- arch/powerpc/kernel/livepatch_64.c | 31 +++++++++++++++++++--------- 4 files changed, 79 insertions(+), 22 deletions(-) diff --git a/arch/powerpc/include/asm/livepatch.h b/arch/powerpc/include/asm/livepatch.h index ae674ea59ab3..285602e637f1 100644 --- a/arch/powerpc/include/asm/livepatch.h +++ b/arch/powerpc/include/asm/livepatch.h @@ -121,9 +121,20 @@ struct arch_klp_data { #define KLP_MAX_REPLACE_SIZE sizeof_field(struct arch_klp_data, old_insns) struct stackframe { + /* stack frame to be unwinded */ unsigned long sp; + /* link register saved in last stack frame */ unsigned long pc; + /* instruction register saved in pt_regs */ unsigned long nip; + /* link register saved in pt_regs */ + unsigned long link; + /* stack frame pointer (r1 register) saved in pt_regs */ + unsigned long sfp; + /* check if nip and link are in same function */ + unsigned int nip_link_in_same_func; + /* check if it is top frame before interrupt */ + unsigned int is_top_frame; }; #ifdef PPC64_ELF_ABI_v1 diff --git a/arch/powerpc/kernel/livepatch.c b/arch/powerpc/kernel/livepatch.c index d568e8c8b16b..5ba38c2c7c5c 100644 --- a/arch/powerpc/kernel/livepatch.c +++ b/arch/powerpc/kernel/livepatch.c @@ -23,6 +23,7 @@ #include <linux/module.h> #include <linux/ftrace.h> #include <linux/livepatch.h> +#include <linux/kallsyms.h> #include <asm/probes.h> #include <asm/livepatch.h> #include <asm/code-patching.h> @@ -67,6 +68,22 @@ int klp_brk_handler(struct pt_regs *regs) return 1; } +static int check_addr_in_same_func(unsigned long addr1, unsigned long addr2) +{ + unsigned long size = 0; + unsigned long offset = 0; + unsigned long start; + + if (addr1 == 0 || addr2 == 0) + return 0; + if (addr1 == addr2) + return 1; + if (!kallsyms_lookup_size_offset(addr1, &size, &offset)) + return 0; + start = addr1 - offset; + return (addr2 >= start) && (addr2 - start < size); +} + int klp_unwind_frame(struct task_struct *tsk, struct stackframe *frame) { unsigned long *stack; @@ -79,7 +96,10 @@ int klp_unwind_frame(struct task_struct *tsk, struct stackframe *frame) if (frame->nip != 0) frame->nip = 0; + if (frame->link != 0) + frame->link = 0; + frame->is_top_frame = (frame->sfp == frame->sp); stack = (unsigned long *)frame->sp; /* @@ -94,10 +114,14 @@ int klp_unwind_frame(struct task_struct *tsk, struct stackframe *frame) struct pt_regs *regs = (struct pt_regs *) (frame->sp + STACK_FRAME_OVERHEAD); frame->nip = regs->nip; - pr_debug("--- interrupt: task = %d/%s, trap %lx at NIP=x%lx/%pS, LR=0x%lx/%pS\n", + frame->link = regs->link; + frame->sfp = regs->gpr[PT_R1]; + frame->nip_link_in_same_func = check_addr_in_same_func(frame->nip, frame->link); + pr_debug("--- interrupt: task = %d/%s, trap %lx at NIP=0x%lx/%pS, LR=0x%lx/%pS, SFP=0x%lx, nip_link_in_same_func=%u\n", tsk->pid, tsk->comm, regs->trap, regs->nip, (void *)regs->nip, - regs->link, (void *)regs->link); + regs->link, (void *)regs->link, + frame->sfp, frame->nip_link_in_same_func); } frame->sp = stack[0]; diff --git a/arch/powerpc/kernel/livepatch_32.c b/arch/powerpc/kernel/livepatch_32.c index 3fe4f3c5790b..134b08e12e74 100644 --- a/arch/powerpc/kernel/livepatch_32.c +++ b/arch/powerpc/kernel/livepatch_32.c @@ -272,13 +272,21 @@ static int klp_check_jump_func(struct stackframe *frame, void *data) struct walk_stackframe_args *args = data; struct klp_func_list *check_funcs = args->check_funcs; - /* check the PC first */ - if (!check_func_list(check_funcs, &args->ret, frame->pc)) - return args->ret; - /* check NIP when the exception stack switching */ if (frame->nip && !check_func_list(check_funcs, &args->ret, frame->nip)) return args->ret; + if (frame->link && !frame->nip_link_in_same_func && + !check_func_list(check_funcs, &args->ret, frame->link)) + return args->ret; + /* + * There are two cases that frame->pc is reliable: + * 1. frame->pc is not in top frame before interrupt; + * 2. nip and link are in same function; + */ + if (!frame->is_top_frame || frame->nip_link_in_same_func) { + if (!check_func_list(check_funcs, &args->ret, frame->pc)) + return args->ret; + } return 0; } @@ -298,10 +306,11 @@ static int do_check_calltrace(struct walk_stackframe_args *args, int (*fn)(struct stackframe *, void *)) { struct task_struct *g, *t; - struct stackframe frame; unsigned long *stack; for_each_process_thread(g, t) { + struct stackframe frame = { 0 }; + if (t == current) { /* * Handle the current carefully on each CPUs, we shouldn't @@ -329,7 +338,6 @@ static int do_check_calltrace(struct walk_stackframe_args *args, frame.sp = (unsigned long)stack; frame.pc = stack[STACK_FRAME_LR_SAVE]; - frame.nip = 0; klp_walk_stackframe(&frame, fn, t, args); if (args->ret) { pr_info("PID: %d Comm: %.20s\n", t->pid, t->comm); @@ -368,13 +376,16 @@ static int check_module_calltrace(struct stackframe *frame, void *data) { struct walk_stackframe_args *args = data; - /* check the PC first */ - if (within_module_core(frame->pc, args->mod)) - goto err_out; - /* check NIP when the exception stack switching */ if (frame->nip && within_module_core(frame->nip, args->mod)) goto err_out; + if (frame->link && !frame->nip_link_in_same_func && + within_module_core(frame->link, args->mod)) + goto err_out; + if (!frame->is_top_frame || frame->nip_link_in_same_func) { + if (within_module_core(frame->pc, args->mod)) + goto err_out; + } return 0; diff --git a/arch/powerpc/kernel/livepatch_64.c b/arch/powerpc/kernel/livepatch_64.c index a2ec7c8c1bad..b33839b5916a 100644 --- a/arch/powerpc/kernel/livepatch_64.c +++ b/arch/powerpc/kernel/livepatch_64.c @@ -273,13 +273,21 @@ static int klp_check_jump_func(struct stackframe *frame, void *data) struct walk_stackframe_args *args = data; struct klp_func_list *check_funcs = args->check_funcs; - /* check the PC first */ - if (!check_func_list(check_funcs, &args->ret, frame->pc)) - return args->ret; - /* check NIP when the exception stack switching */ if (frame->nip && !check_func_list(check_funcs, &args->ret, frame->nip)) return args->ret; + if (frame->link && !frame->nip_link_in_same_func && + !check_func_list(check_funcs, &args->ret, frame->link)) + return args->ret; + /* + * There are two cases that frame->pc is reliable: + * 1. frame->pc is not in top frame before interrupt; + * 2. nip and link are in same function; + */ + if (!frame->is_top_frame || frame->nip_link_in_same_func) { + if (!check_func_list(check_funcs, &args->ret, frame->pc)) + return args->ret; + } return 0; } @@ -299,10 +307,11 @@ static int do_check_calltrace(struct walk_stackframe_args *args, int (*fn)(struct stackframe *, void *)) { struct task_struct *g, *t; - struct stackframe frame; unsigned long *stack; for_each_process_thread(g, t) { + struct stackframe frame = { 0 }; + if (t == current) { /* * Handle the current carefully on each CPUs, @@ -332,7 +341,6 @@ static int do_check_calltrace(struct walk_stackframe_args *args, frame.sp = (unsigned long)stack; frame.pc = stack[STACK_FRAME_LR_SAVE]; - frame.nip = 0; klp_walk_stackframe(&frame, fn, t, args); if (args->ret) { pr_debug("%s FAILED when %s\n", __func__, @@ -373,13 +381,16 @@ static int check_module_calltrace(struct stackframe *frame, void *data) { struct walk_stackframe_args *args = data; - /* check the PC first */ - if (within_module_core(frame->pc, args->mod)) - goto err_out; - /* check NIP when the exception stack switching */ if (frame->nip && within_module_core(frame->nip, args->mod)) goto err_out; + if (frame->link && !frame->nip_link_in_same_func && + within_module_core(frame->link, args->mod)) + goto err_out; + if (!frame->is_top_frame || frame->nip_link_in_same_func) { + if (within_module_core(frame->pc, args->mod)) + goto err_out; + } return 0; -- 2.25.1

2 1

[PACH OLK-5.10 v5 00/20] psi irq in cgroupv1 and psi fine grained
by Lu Jialin 29 Nov '23

29 Nov '23

Chen Wandun (3): mm: disable psi cgroup v1 by default mm: add config isolation for psi under cgroup v1 psi: dont alloc memory for psi by default Chengming Zhou (3): sched/psi: Fix periodic aggregation shut off sched/psi: Optimize task switch inside shared cgroups again sched/psi: Add PSI_IRQ to track IRQ/SOFTIRQ pressure Hao Jia (1): sched/psi: Zero the memory of struct psi_group Johannes Weiner (1): sched/psi: Remove NR_ONCPU task accounting Lu Jialin (11): psi: support irq.pressure under cgroup v1 psi: enable CONFIG_PSI_CGROUP_V1 in openeuler_defconfig psi: update psi irqtime when the irq delta is nozero memcg: split async memcg reclaim from reclaim_high psi: add struct psi_group_ext PSI: Introduce fine grained stall time collect for cgroup reclaim PSI: Introduce avgs and total calculation for cgroup reclaim PSI: Introduce pressure.stat in psi PSI: add more memory fine grained stall tracking in pressure.stat add cpu fine grained stall tracking in pressure.stat PSI: enable CONFIG_PSI_FINE_GRAINED in openeuler_defconfig Suren Baghdasaryan (1): psi: Fix "defined but not used" warnings when CONFIG_PROC_FS=n Documentation/admin-guide/cgroup-v2.rst | 6 + arch/arm64/configs/openeuler_defconfig | 2 + arch/x86/configs/openeuler_defconfig | 2 + block/blk-cgroup.c | 2 +- block/blk-core.c | 2 +- include/linux/cgroup-defs.h | 6 +- include/linux/cgroup.h | 2 +- include/linux/psi.h | 9 +- include/linux/psi_types.h | 99 +++- include/linux/sched.h | 4 + init/Kconfig | 20 + kernel/cgroup/cgroup.c | 64 ++- kernel/sched/core.c | 2 + kernel/sched/cpuacct.c | 6 +- kernel/sched/fair.c | 6 - kernel/sched/psi.c | 691 ++++++++++++++++++++---- kernel/sched/stats.h | 9 + mm/compaction.c | 2 +- mm/filemap.c | 4 +- mm/memcontrol.c | 92 ++-- mm/page_alloc.c | 6 + mm/page_io.c | 3 + mm/vmscan.c | 5 +- 23 files changed, 873 insertions(+), 171 deletions(-) -- 2.34.1

1 20

[PATCH v2 OLK-5.10 0/3] support page eject
by Wupeng Ma 29 Nov '23

29 Nov '23

From: Ma Wupeng <mawupeng1(a)huawei.com> Page ejection is achieved by soft_offline_page and unpoison_memory. A linked list is maintained to log the pfns which are offlined by this module. Only the pfns present in the list is allowed to go online. The follow sysfs is introduced to online/offline page via this module, since there is a internal linked list to record all offlined pages, remove_page is introduce to remove this page from list. - /sys/kernel/page_eject/online_page - /sys/kernel/page_eject/offline_page - /sys/kernel/page_eject/remove_page Changelog since v1: - remove mce related patches temporarily. Ma Wupeng (3): mm/hwpoison: Export symbol soft_offline_page mm: page_eject: Introuduce page ejection config: update defconfig for PAGE_EJECT arch/arm64/configs/openeuler_defconfig | 1 + arch/x86/configs/openeuler_defconfig | 1 + drivers/ras/Kconfig | 10 + drivers/ras/Makefile | 1 + drivers/ras/page_eject.c | 248 +++++++++++++++++++++++++ mm/memory-failure.c | 1 + 6 files changed, 262 insertions(+) create mode 100644 drivers/ras/page_eject.c -- 2.25.1

2 4

[PATCH v2 OLK-5.10 0/3] support page eject
by Wupeng Ma 29 Nov '23

29 Nov '23

From: Ma Wupeng <mawupeng1(a)huawei.com> Page ejection is achieved by soft_offline_page and unpoison_memory. A linked list is maintained to log the pfns which are offlined by this module. Only the pfns present in the list is allowed to go online. The follow sysfs is introduced to online/offline page via this module, since there is a internal linked list to record all offlined pages, remove_page is introduce to remove this page from list. - /sys/kernel/page_eject/online_page - /sys/kernel/page_eject/offline_page - /sys/kernel/page_eject/remove_page Besides pag eject, machine check safe is enhanced as follow: During soft offline page, page is copied to new page in kernel. If the origin page has UCE, there will lead to kernel panic. In order to solve this problem, use machine check safe to catch this error which can be achieved by using copy_mc_to_kernel to replace copy_page. Signal SIGBUS will be send to user task if this UCE is consumed by this situation to avoid kernel panic. Ma Wupeng (3): mm/hwpoison: Export symbol soft_offline_page mm: page_eject: Introuduce page ejection config: update defconfig for PAGE_EJECT arch/arm64/configs/openeuler_defconfig | 1 + arch/x86/configs/openeuler_defconfig | 1 + drivers/ras/Kconfig | 10 + drivers/ras/Makefile | 1 + drivers/ras/page_eject.c | 248 +++++++++++++++++++++++++ mm/memory-failure.c | 1 + 6 files changed, 262 insertions(+) create mode 100644 drivers/ras/page_eject.c -- 2.25.1

2 4

[PATCH OLK-5.10] driver/perf: Not enabled ARM64_BRBE by default
by Junhao He 29 Nov '23

29 Nov '23

driver inclusion category: bugfix bugzilla: https://gitee.com/openeuler/kernel/issues/I8EC9K CVE: NA -------------------------------- The BRBE feature depend on ARMV8 PMU hardware support for branch stack sampling. Not enabled ARM64_BRBE by default. Signed-off-by: Junhao He <hejunhao3(a)huawei.com> --- drivers/perf/Kconfig | 1 - 1 file changed, 1 deletion(-) diff --git a/drivers/perf/Kconfig b/drivers/perf/Kconfig index 25a577530da4..67ad53cde11f 100644 --- a/drivers/perf/Kconfig +++ b/drivers/perf/Kconfig @@ -133,7 +133,6 @@ config ARM_SPE_PMU config ARM64_BRBE bool "Enable support for Branch Record Buffer Extension (BRBE)" depends on PERF_EVENTS && ARM64 && ARM_PMU - default y help Enable perf support for Branch Record Buffer Extension (BRBE) which records all branches taken in an execution path. This supports some -- 2.33.0

2 1

[PACH OLK-5.10 v5 00/20] psi irq in cgroupv1 and psi fine grained
by Lu Jialin 29 Nov '23

29 Nov '23

Chen Wandun (3): mm: disable psi cgroup v1 by default mm: add config isolation for psi under cgroup v1 psi: dont alloc memory for psi by default Chengming Zhou (3): sched/psi: Fix periodic aggregation shut off sched/psi: Optimize task switch inside shared cgroups again sched/psi: Add PSI_IRQ to track IRQ/SOFTIRQ pressure Hao Jia (1): sched/psi: Zero the memory of struct psi_group Johannes Weiner (1): sched/psi: Remove NR_ONCPU task accounting Lu Jialin (11): psi: support irq.pressure under cgroup v1 psi: enable CONFIG_PSI_CGROUP_V1 in openeuler_defconfig psi: update psi irqtime when the irq delta is nozero memcg: split async memcg reclaim from reclaim_high psi: add struct psi_group_ext PSI: Introduce fine grained stall time collect for cgroup reclaim PSI: Introduce avgs and total calculation for cgroup reclaim PSI: Introduce pressure.stat in psi PSI: add more memory fine grained stall tracking in pressure.stat add cpu fine grained stall tracking in pressure.stat PSI: enable CONFIG_PSI_FINE_GRAINED in openeuler_defconfig Suren Baghdasaryan (1): psi: Fix "defined but not used" warnings when CONFIG_PROC_FS=n Documentation/admin-guide/cgroup-v2.rst | 6 + arch/arm64/configs/openeuler_defconfig | 2 + arch/x86/configs/openeuler_defconfig | 2 + block/blk-cgroup.c | 2 +- block/blk-core.c | 2 +- include/linux/cgroup-defs.h | 6 +- include/linux/cgroup.h | 2 +- include/linux/psi.h | 9 +- include/linux/psi_types.h | 99 +++- include/linux/sched.h | 4 + init/Kconfig | 20 + kernel/cgroup/cgroup.c | 64 ++- kernel/sched/core.c | 2 + kernel/sched/cpuacct.c | 6 +- kernel/sched/fair.c | 6 - kernel/sched/psi.c | 691 ++++++++++++++++++++---- kernel/sched/stats.h | 9 + mm/compaction.c | 2 +- mm/filemap.c | 4 +- mm/memcontrol.c | 92 ++-- mm/page_alloc.c | 6 + mm/page_io.c | 3 + mm/vmscan.c | 5 +- 23 files changed, 873 insertions(+), 171 deletions(-) -- 2.34.1

1 20

[PATCH OLK-6.6 v2 00/31] Backport ARM64-ILP32 patches
by Jinjie Ruan 29 Nov '23

29 Nov '23

Backport ILP32 for ARM64 patches to OLK6.6 form openEuler-23.09. Andrew Pinski (3): arm64: rename COMPAT to AARCH32_EL0 arm64: uapi: set __BITS_PER_LONG correctly for ILP32 and LP64 arm64:ilp32: add ARM64_ILP32 to Kconfig Chen Jiahao (4): arm64: fix AUDIT_ARCH_AARCH64ILP32 bug on audit subsystem arm64: fix address limit problem with TASK_SIZE_MAX arm64: set 32-bit compatible TASK_SIZE_MAX to fix U32 libc_write_01 error arm64: fix image size inflation with CONFIG_COMPAT_TASK_SIZE Dave Martin (1): arm64: signal: Make parse_user_sigframe() independent of rt_sigframe layout James Morse (1): ptrace: Add compat PTRACE_{G,S}ETSIGMASK handlers Philipp Tomsich (1): arm64:ilp32: add vdso-ilp32 and use for signal return Xiongfeng Wang (6): arm64: rename functions that reference compat term arm64: secomp: fix the secure computing mode 1 syscall check for ilp32 ilp32: avoid clearing upper 32 bits of syscall return value for ilp32 ilp32: skip ARM erratum 1418040 for ilp32 application arm64: fix abi change caused by ILP32 ilp32: fix compile problem when ARM64_ILP32 and UBSAN are both enabled Yury Norov (14): thread: move thread bits accessors to separated file arm64: ilp32: add documentation on the ILP32 ABI for ARM64 arm64: introduce is_a32_compat_{task,thread} for AArch32 compat arm64: ilp32: add is_ilp32_compat_{task,thread} and TIF_32BIT_AARCH64 arm64: introduce AUDIT_ARCH_AARCH64ILP32 for ilp32 arm64: introduce binfmt_elf32.c arm64: change compat_elf_hwcap and compat_elf_hwcap2 prefix to a32 arm64: ilp32: introduce binfmt_ilp32.c arm64: ilp32: share aarch32 syscall handlers arm64: ilp32: introduce syscall table for ILP32 arm64: signal: share lp64 signal structures and routines to ilp32 arm64: signal32: move ilp32 and aarch32 common code to separated file arm64: ilp32: introduce ilp32-specific sigframe and ucontext arm64: ptrace: handle ptrace_request differently for aarch32 and ilp32 Zhen Lei (1): arm64: replace is_compat_task() with is_ilp32_compat_task() in TASK_SIZE_MAX Documentation/arm64/ilp32.txt | 52 +++ arch/arm64/Kconfig | 26 +- arch/arm64/Makefile | 3 + arch/arm64/include/asm/arch_timer.h | 4 +- arch/arm64/include/asm/compat.h | 18 +- arch/arm64/include/asm/elf.h | 32 +- arch/arm64/include/asm/fpsimd.h | 2 +- arch/arm64/include/asm/ftrace.h | 2 +- arch/arm64/include/asm/hwcap.h | 8 +- arch/arm64/include/asm/is_compat.h | 83 ++++ arch/arm64/include/asm/processor.h | 24 +- arch/arm64/include/asm/ptrace.h | 16 +- arch/arm64/include/asm/seccomp.h | 32 +- arch/arm64/include/asm/signal32.h | 31 +- arch/arm64/include/asm/signal32_common.h | 13 + arch/arm64/include/asm/signal_common.h | 391 +++++++++++++++++ arch/arm64/include/asm/signal_ilp32.h | 23 + arch/arm64/include/asm/syscall.h | 14 +- arch/arm64/include/asm/thread_info.h | 4 +- arch/arm64/include/asm/unistd.h | 5 + arch/arm64/include/asm/vdso.h | 9 + arch/arm64/include/uapi/asm/bitsperlong.h | 9 +- arch/arm64/include/uapi/asm/unistd.h | 15 +- arch/arm64/kernel/Makefile | 8 +- arch/arm64/kernel/armv8_deprecated.c | 14 +- arch/arm64/kernel/asm-offsets.c | 13 +- arch/arm64/kernel/binfmt_elf32.c | 27 ++ arch/arm64/kernel/binfmt_ilp32.c | 89 ++++ arch/arm64/kernel/compat_alignment.c | 2 +- arch/arm64/kernel/cpufeature.c | 36 +- arch/arm64/kernel/cpuinfo.c | 18 +- arch/arm64/kernel/debug-monitors.c | 4 +- arch/arm64/kernel/entry-common.c | 6 +- arch/arm64/kernel/hw_breakpoint.c | 8 +- arch/arm64/kernel/perf_callchain.c | 28 +- arch/arm64/kernel/perf_regs.c | 4 +- arch/arm64/kernel/process.c | 15 +- arch/arm64/kernel/proton-pack.c | 2 +- arch/arm64/kernel/ptrace.c | 38 +- arch/arm64/kernel/signal.c | 397 +++--------------- arch/arm64/kernel/signal32.c | 97 ++--- arch/arm64/kernel/signal32_common.c | 37 ++ arch/arm64/kernel/signal_ilp32.c | 67 +++ arch/arm64/kernel/sys32.c | 104 +---- arch/arm64/kernel/sys32_common.c | 106 +++++ arch/arm64/kernel/sys_compat.c | 12 +- arch/arm64/kernel/sys_ilp32.c | 82 ++++ arch/arm64/kernel/syscall.c | 37 +- arch/arm64/kernel/traps.c | 5 +- arch/arm64/kernel/vdso-ilp32/.gitignore | 2 + arch/arm64/kernel/vdso-ilp32/Makefile | 111 +++++ arch/arm64/kernel/vdso-ilp32/vdso-ilp32.S | 22 + arch/arm64/kernel/vdso-ilp32/vdso-ilp32.lds.S | 88 ++++ arch/arm64/kernel/vdso.c | 46 +- arch/arm64/mm/fault.c | 2 +- include/linux/sched.h | 1 + include/linux/thread_bits.h | 87 ++++ include/linux/thread_info.h | 73 +--- kernel/ptrace.c | 52 ++- 59 files changed, 1790 insertions(+), 766 deletions(-) create mode 100644 Documentation/arm64/ilp32.txt create mode 100644 arch/arm64/include/asm/is_compat.h create mode 100644 arch/arm64/include/asm/signal32_common.h create mode 100644 arch/arm64/include/asm/signal_common.h create mode 100644 arch/arm64/include/asm/signal_ilp32.h create mode 100644 arch/arm64/kernel/binfmt_elf32.c create mode 100644 arch/arm64/kernel/binfmt_ilp32.c create mode 100644 arch/arm64/kernel/signal32_common.c create mode 100644 arch/arm64/kernel/signal_ilp32.c create mode 100644 arch/arm64/kernel/sys32_common.c create mode 100644 arch/arm64/kernel/sys_ilp32.c create mode 100644 arch/arm64/kernel/vdso-ilp32/.gitignore create mode 100644 arch/arm64/kernel/vdso-ilp32/Makefile create mode 100644 arch/arm64/kernel/vdso-ilp32/vdso-ilp32.S create mode 100644 arch/arm64/kernel/vdso-ilp32/vdso-ilp32.lds.S create mode 100644 include/linux/thread_bits.h -- 2.34.1

1 31

[openEuler-20.03-LTS-SP4 PATCH] arm64: cpufeature: Extract capped perfmon fields
by Kunkun Jiang 29 Nov '23

29 Nov '23

From: zhangyuan <zhangyuan162(a)huawei.com> KVM: arm64: limit PMU version to PMUv3 for ARMv8.1 Backport patches from upstream to fix the issue of PMU. Signed-off-by: Yuan Zhang <zhangyuan162(a)huawei.com> --- kernel.spec | 5 +- ...eature-Extract-capped-perfmon-fields.patch | 80 +++++++++++++++++ ...mit-PMU-version-to-PMUv3-for-ARMv8.1.patch | 90 +++++++++++++++++++ series.conf | 2 + 4 files changed, 176 insertions(+), 1 deletion(-) create mode 100644 patches/0758-arm64-cpufeature-Extract-capped-perfmon-fields.patch create mode 100644 patches/0759-KVM-arm64-limit-PMU-version-to-PMUv3-for-ARMv8.1.patch diff --git a/kernel.spec b/kernel.spec index 0ddbd93..7a06c2c 100644 --- a/kernel.spec +++ b/kernel.spec @@ -32,7 +32,7 @@ Name: kernel Version: 4.19.90 -Release: %{hulkrelease}.0252 +Release: %{hulkrelease}.0253 Summary: Linux Kernel License: GPLv2 URL: http://www.kernel.org/ @@ -849,6 +849,9 @@ fi %endif %changelog +* Tue Nov 28 2023 Yuan Zhang <zhangyuan162(a)huawei.com> - 4.19.90-2311.5.0.0253 +- arm64: cpufeature: Extract capped perfmon fields +- KVM: arm64: limit PMU version to PMUv3 for ARMv8.1 * Tue Nov 28 2023 Li Xiaodong <lixiaodong67(a)huawei.com> - 4.19.90-2311.5.0.0252 - !3000 [openEuler-1.0-LTS] add Phytium drivers CONFIG diff --git a/patches/0758-arm64-cpufeature-Extract-capped-perfmon-fields.patch b/patches/0758-arm64-cpufeature-Extract-capped-perfmon-fields.patch new file mode 100644 index 0000000..de1db70 --- /dev/null +++ b/patches/0758-arm64-cpufeature-Extract-capped-perfmon-fields.patch @@ -0,0 +1,80 @@ +From a207eaeac98757a86cce704be6c4e097439503c5 Mon Sep 17 00:00:00 2001 +From: Andrew Murray <andrew.murray(a)arm.com> +Date: Tue, 28 Nov 2023 15:46:32 +0800 +Subject: [PATCH 1/2] arm64: cpufeature: Extract capped perfmon fields + +mainline inclusion +from mainline-v5.7-rc1 +commit 8e35aa642ee4dab01b16cc4b2df59d1936f3b3c2 +category: bugfix +bugzilla: https://gitee.com/openeuler/kernel/issues/I8K8XV +CVE: NA + +Reference: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?… + +----------------------------------- + +commit 8e35aa642ee4dab01b16cc4b2df59d1936f3b3c2 upstream. + +When emulating ID registers there is often a need to cap the version +bits of a feature such that the guest will not use features that the +host is not aware of. For example, when KVM mediates access to the PMU +by emulating register accesses. + +Let's add a helper that extracts a performance monitors ID field and +caps the version to a given value. + +Fields that identify the version of the Performance Monitors Extension +do not follow the standard ID scheme, and instead follow the scheme +described in ARM DDI 0487E.a page D13-2825 "Alternative ID scheme used +for the Performance Monitors Extension version". The value 0xF means an +IMPLEMENTATION DEFINED PMU is present, and values 0x0-OxE can be treated +the same as an unsigned field with 0x0 meaning no PMU is present. + +Signed-off-by: Andrew Murray <andrew.murray(a)arm.com> +Reviewed-by: Suzuki K Poulose <suzuki.poulose(a)arm.com> +[Mark: rework to handle perfmon fields] +Signed-off-by: Mark Rutland <mark.rutland(a)arm.com> +Signed-off-by: Will Deacon <will(a)kernel.org> +Signed-off-by: zhangyuan <zhangyuan162(a)huawei.com> +--- + arch/arm64/include/asm/cpufeature.h | 23 +++++++++++++++++++++++ + 1 file changed, 23 insertions(+) + +diff --git a/arch/arm64/include/asm/cpufeature.h b/arch/arm64/include/asm/cpufeature.h +index eef5a9c9b823..e2ba573644d5 100644 +--- a/arch/arm64/include/asm/cpufeature.h ++++ b/arch/arm64/include/asm/cpufeature.h +@@ -449,6 +449,29 @@ cpuid_feature_extract_unsigned_field(u64 features, int field) + return cpuid_feature_extract_unsigned_field_width(features, field, 4); + } + ++/* ++ * Fields that identify the version of the Performance Monitors Extension do ++ * not follow the standard ID scheme. See ARM DDI 0487E.a page D13-2825, ++ * "Alternative ID scheme used for the Performance Monitors Extension version". ++ */ ++static inline u64 __attribute_const__ ++cpuid_feature_cap_perfmon_field(u64 features, int field, u64 cap) ++{ ++ u64 val = cpuid_feature_extract_unsigned_field(features, field); ++ u64 mask = GENMASK_ULL(field + 3, field); ++ ++ /* Treat IMPLEMENTATION DEFINED functionality as unimplemented */ ++ if (val == 0xf) ++ val = 0; ++ ++ if (val > cap) { ++ features &= ~mask; ++ features |= (cap << field) & mask; ++ } ++ ++ return features; ++} ++ + static inline u64 arm64_ftr_mask(const struct arm64_ftr_bits *ftrp) + { + return (u64)GENMASK(ftrp->shift + ftrp->width - 1, ftrp->shift); +-- +2.33.0 + diff --git a/patches/0759-KVM-arm64-limit-PMU-version-to-PMUv3-for-ARMv8.1.patch b/patches/0759-KVM-arm64-limit-PMU-version-to-PMUv3-for-ARMv8.1.patch new file mode 100644 index 0000000..1f72f92 --- /dev/null +++ b/patches/0759-KVM-arm64-limit-PMU-version-to-PMUv3-for-ARMv8.1.patch @@ -0,0 +1,90 @@ +From 6a416c5efa3ce99f5f0f7e9c603b42b0e047481c Mon Sep 17 00:00:00 2001 +From: Andrew Murray <andrew.murray(a)arm.com> +Date: Tue, 28 Nov 2023 15:46:33 +0800 +Subject: [PATCH 2/2] KVM: arm64: limit PMU version to PMUv3 for ARMv8.1 + +mainline inclusion +from mainline-v5.7-rc1 +commit c854188ea01062f5a5fd7f05658feb1863774eaa +category: bugfix +bugzilla: https://gitee.com/openeuler/kernel/issues/I8K8XV +CVE: NA + +Reference: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?… + +----------------------------------- + +commit c854188ea01062f5a5fd7f05658feb1863774eaa upstream. + +We currently expose the PMU version of the host to the guest via +emulation of the DFR0_EL1 and AA64DFR0_EL1 debug feature registers. +However many of the features offered beyond PMUv3 for 8.1 are not +supported in KVM. Examples of this include support for the PMMIR +registers (added in PMUv3 for ARMv8.4) and 64-bit event counters +added in (PMUv3 for ARMv8.5). + +Let's trap the Debug Feature Registers in order to limit +PMUVer/PerfMon in the Debug Feature Registers to PMUv3 for ARMv8.1 +to avoid unexpected behaviour. + +Both ID_AA64DFR0.PMUVer and ID_DFR0.PerfMon follow the "Alternative ID +scheme used for the Performance Monitors Extension version" where 0xF +means an IMPLEMENTATION DEFINED PMU is implemented, and values 0x0-0xE +are treated as with an unsigned field (with 0x0 meaning no PMU is +present). As we don't expect to expose an IMPLEMENTATION DEFINED PMU, +and our cap is below 0xF, we can treat these fields as unsigned when +applying the cap. + +Signed-off-by: Andrew Murray <andrew.murray(a)arm.com> +Reviewed-by: Suzuki K Poulose <suzuki.poulose(a)arm.com> +[Mark: make field names consistent, use perfmon cap] +Signed-off-by: Mark Rutland <mark.rutland(a)arm.com> +Signed-off-by: Will Deacon <will(a)kernel.org> +[zhangyuan162(a)huawei.com: adjust the context in read_id_reg()] +Signed-off-by: zhangyuan <zhangyuan162(a)huawei.com> +--- + arch/arm64/include/asm/sysreg.h | 6 ++++++ + arch/arm64/kvm/sys_regs.c | 10 ++++++++++ + 2 files changed, 16 insertions(+) + +diff --git a/arch/arm64/include/asm/sysreg.h b/arch/arm64/include/asm/sysreg.h +index de4c31f073bb..041100c28cab 100644 +--- a/arch/arm64/include/asm/sysreg.h ++++ b/arch/arm64/include/asm/sysreg.h +@@ -637,6 +637,12 @@ + #define ID_AA64DFR0_TRACEVER_SHIFT 4 + #define ID_AA64DFR0_DEBUGVER_SHIFT 0 + ++#define ID_AA64DFR0_PMUVER_8_1 0x4 ++ ++#define ID_DFR0_PERFMON_SHIFT 24 ++ ++#define ID_DFR0_PERFMON_8_1 0x4 ++ + #define ID_AA64DFR0_PMSVER_8_2 0x1 + #define ID_AA64DFR0_PMSVER_8_3 0x2 + +diff --git a/arch/arm64/kvm/sys_regs.c b/arch/arm64/kvm/sys_regs.c +index 25caa9ec95e4..cc824a208684 100644 +--- a/arch/arm64/kvm/sys_regs.c ++++ b/arch/arm64/kvm/sys_regs.c +@@ -1082,6 +1082,16 @@ static u64 read_id_reg(struct kvm_vcpu *vcpu, + kvm_debug("LORegions unsupported for guests, suppressing\n"); + + val &= ~(0xfUL << ID_AA64MMFR1_LOR_SHIFT); ++ } else if (id == SYS_ID_AA64DFR0_EL1) { ++ /* Limit guests to PMUv3 for ARMv8.1 */ ++ val = cpuid_feature_cap_perfmon_field(val, ++ ID_AA64DFR0_PMUVER_SHIFT, ++ ID_AA64DFR0_PMUVER_8_1); ++ } else if (id == SYS_ID_DFR0_EL1) { ++ /* Limit guests to PMUv3 for ARMv8.1 */ ++ val = cpuid_feature_cap_perfmon_field(val, ++ ID_DFR0_PERFMON_SHIFT, ++ ID_DFR0_PERFMON_8_1); + } + + return val; +-- +2.33.0 + diff --git a/series.conf b/series.conf index f5f2522..67b9be0 100644 --- a/series.conf +++ b/series.conf @@ -758,3 +758,5 @@ patches/0754-scsi-hisi_sas_v3_hw-Remove-extra-function-calls-for-.patch patches/0755-config-arm64-Enable-dubugfs-config-of-hisi-sas.patch patches/0756-crypto-hisilicon-Add-value-profile-support-for-kerne.patch patches/0757-Revert-genirq-Increase-the-number-of-IRQ-descriptors.patch +patches/arm64-cpufeature-Extract-capped-perfmon-fields.patch +patches/KVM-arm64-limit-PMU-version-to-PMUv3-for-ARMv8.1.patch -- 2.33.0

1 0