From f653e29bc2d94ae6b6f470145c630d1ba0d856d2 Mon Sep 17 00:00:00 2001 From: Heiko Carstens Date: Mon, 28 Oct 2019 10:11:41 +0100 Subject: s390/time: remove monotonic_clock() Remove unused monotonic_clock() function. Signed-off-by: Heiko Carstens Signed-off-by: Vasily Gorbik --- arch/s390/kernel/time.c | 9 --------- 1 file changed, 9 deletions(-) (limited to 'arch/s390/kernel') diff --git a/arch/s390/kernel/time.c b/arch/s390/kernel/time.c index e8766beee5ad..f9d070d016e3 100644 --- a/arch/s390/kernel/time.c +++ b/arch/s390/kernel/time.c @@ -110,15 +110,6 @@ unsigned long long notrace sched_clock(void) } NOKPROBE_SYMBOL(sched_clock); -/* - * Monotonic_clock - returns # of nanoseconds passed since time_init() - */ -unsigned long long monotonic_clock(void) -{ - return sched_clock(); -} -EXPORT_SYMBOL(monotonic_clock); - static void ext_to_timespec64(unsigned char *clk, struct timespec64 *xt) { unsigned long long high, low, rem, sec, nsec; -- cgit From d3baaeb5ae0870973bfe0f6aaed887b894d5e560 Mon Sep 17 00:00:00 2001 From: Vasily Gorbik Date: Tue, 22 Oct 2019 17:37:51 +0200 Subject: s390: avoid double handling of "noexec" option "noexec" option is already parsed during startup and its value is exposed via noexec_disabled variable. Simply reuse that value during machine facilities detection. Suggested-by: Ilya Leoshkevich Reviewed-by: Heiko Carstens Signed-off-by: Vasily Gorbik --- arch/s390/kernel/early.c | 17 +---------------- 1 file changed, 1 insertion(+), 16 deletions(-) (limited to 'arch/s390/kernel') diff --git a/arch/s390/kernel/early.c b/arch/s390/kernel/early.c index b432d63d0b37..2e99e01e4f62 100644 --- a/arch/s390/kernel/early.c +++ b/arch/s390/kernel/early.c @@ -238,7 +238,7 @@ static __init void detect_machine_facilities(void) S390_lowcore.machine_flags |= MACHINE_FLAG_VX; __ctl_set_bit(0, 17); } - if (test_facility(130)) { + if (test_facility(130) && !noexec_disabled) { S390_lowcore.machine_flags |= MACHINE_FLAG_NX; __ctl_set_bit(0, 20); } @@ -268,21 +268,6 @@ static int __init disable_vector_extension(char *str) } early_param("novx", disable_vector_extension); -static int __init noexec_setup(char *str) -{ - bool enabled; - int rc; - - rc = kstrtobool(str, &enabled); - if (!rc && !enabled) { - /* Disable no-execute support */ - S390_lowcore.machine_flags &= ~MACHINE_FLAG_NX; - __ctl_clear_bit(0, 20); - } - return rc; -} -early_param("noexec", noexec_setup); - static int __init cad_setup(char *str) { bool enabled; -- cgit From 6756dd9b890fe50c01a6e7546bd498d57ddb98ae Mon Sep 17 00:00:00 2001 From: Vasily Gorbik Date: Mon, 28 Oct 2019 15:17:42 +0100 Subject: s390/process: avoid custom stack unwinding in get_wchan Currently get_wchan uses custom stack unwinding implementation which relies on back_chain presence. Replace it with more abstract stack unwinding api usage. Suggested-by: Ilya Leoshkevich Reviewed-by: Heiko Carstens Signed-off-by: Vasily Gorbik --- arch/s390/kernel/process.c | 36 ++++++++++++++++-------------------- 1 file changed, 16 insertions(+), 20 deletions(-) (limited to 'arch/s390/kernel') diff --git a/arch/s390/kernel/process.c b/arch/s390/kernel/process.c index b0afec673f77..6ccef5f29761 100644 --- a/arch/s390/kernel/process.c +++ b/arch/s390/kernel/process.c @@ -40,6 +40,7 @@ #include #include #include +#include #include "entry.h" asmlinkage void ret_from_fork(void) asm ("ret_from_fork"); @@ -178,9 +179,8 @@ EXPORT_SYMBOL(dump_fpu); unsigned long get_wchan(struct task_struct *p) { - struct stack_frame *sf, *low, *high; - unsigned long return_address; - int count; + struct unwind_state state; + unsigned long ip = 0; if (!p || p == current || p->state == TASK_RUNNING || !task_stack_page(p)) return 0; @@ -188,26 +188,22 @@ unsigned long get_wchan(struct task_struct *p) if (!try_get_task_stack(p)) return 0; - low = task_stack_page(p); - high = (struct stack_frame *) task_pt_regs(p); - sf = (struct stack_frame *) p->thread.ksp; - if (sf <= low || sf > high) { - return_address = 0; - goto out; - } - for (count = 0; count < 16; count++) { - sf = (struct stack_frame *)READ_ONCE_NOCHECK(sf->back_chain); - if (sf <= low || sf > high) { - return_address = 0; - goto out; + unwind_for_each_frame(&state, p, NULL, 0) { + if (state.stack_info.type != STACK_TYPE_TASK) { + ip = 0; + break; } - return_address = READ_ONCE_NOCHECK(sf->gprs[8]); - if (!in_sched_functions(return_address)) - goto out; + + ip = unwind_get_return_address(&state); + if (!ip) + break; + + if (!in_sched_functions(ip)) + break; } -out: + put_task_stack(p); - return return_address; + return ip; } unsigned long arch_align_stack(unsigned long sp) -- cgit From effb83ccc83a97dbbe5214f4c443522719f05f3a Mon Sep 17 00:00:00 2001 From: Ilya Leoshkevich Date: Wed, 30 Oct 2019 14:20:32 +0100 Subject: s390: add error handling to perf_callchain_kernel perf_callchain_kernel stops neither when it encounters a garbage address, nor when it runs out of space. Fix both issues using x86 version as an inspiration. Signed-off-by: Ilya Leoshkevich Signed-off-by: Vasily Gorbik --- arch/s390/kernel/perf_event.c | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-) (limited to 'arch/s390/kernel') diff --git a/arch/s390/kernel/perf_event.c b/arch/s390/kernel/perf_event.c index fcb6c2e92b07..1e75cc983546 100644 --- a/arch/s390/kernel/perf_event.c +++ b/arch/s390/kernel/perf_event.c @@ -224,9 +224,13 @@ void perf_callchain_kernel(struct perf_callchain_entry_ctx *entry, struct pt_regs *regs) { struct unwind_state state; + unsigned long addr; - unwind_for_each_frame(&state, current, regs, 0) - perf_callchain_store(entry, state.ip); + unwind_for_each_frame(&state, current, regs, 0) { + addr = unwind_get_return_address(&state); + if (!addr || perf_callchain_store(entry, addr)) + return; + } } /* Perf definitions for PMU event attributes in sysfs */ -- cgit From c2f2093e149d3109f1457deac4909191d9aca323 Mon Sep 17 00:00:00 2001 From: Miroslav Benes Date: Tue, 29 Oct 2019 15:39:02 +0100 Subject: s390/unwind: drop unnecessary code around calling ftrace_graph_ret_addr() The current code around calling ftrace_graph_ret_addr() is ifdeffed and also tests if ftrace redirection is present on stack. ftrace_graph_ret_addr() however performs the test internally and there is a version for !CONFIG_FUNCTION_GRAPH_TRACER as well. The unnecessary code can thus be dropped. Link: http://lkml.kernel.org/r/20191029143904.24051-2-mbenes@suse.cz Signed-off-by: Miroslav Benes Signed-off-by: Heiko Carstens Signed-off-by: Vasily Gorbik --- arch/s390/kernel/unwind_bc.c | 14 ++------------ 1 file changed, 2 insertions(+), 12 deletions(-) (limited to 'arch/s390/kernel') diff --git a/arch/s390/kernel/unwind_bc.c b/arch/s390/kernel/unwind_bc.c index 8fc9daae47a2..5699e820c621 100644 --- a/arch/s390/kernel/unwind_bc.c +++ b/arch/s390/kernel/unwind_bc.c @@ -80,12 +80,7 @@ bool unwind_next_frame(struct unwind_state *state) } } -#ifdef CONFIG_FUNCTION_GRAPH_TRACER - /* Decode any ftrace redirection */ - if (ip == (unsigned long) return_to_handler) - ip = ftrace_graph_ret_addr(state->task, &state->graph_idx, - ip, (void *) sp); -#endif + ip = ftrace_graph_ret_addr(state->task, &state->graph_idx, ip, (void *) sp); /* Update unwind state */ state->sp = sp; @@ -140,12 +135,7 @@ void __unwind_start(struct unwind_state *state, struct task_struct *task, reliable = false; } -#ifdef CONFIG_FUNCTION_GRAPH_TRACER - /* Decode any ftrace redirection */ - if (ip == (unsigned long) return_to_handler) - ip = ftrace_graph_ret_addr(state->task, &state->graph_idx, - ip, NULL); -#endif + ip = ftrace_graph_ret_addr(state->task, &state->graph_idx, ip, NULL); /* Update unwind state */ state->sp = sp; -- cgit From d98b5d0728d6d9f49f370d1846dbac30ec596a42 Mon Sep 17 00:00:00 2001 From: Thomas Richter Date: Thu, 31 Oct 2019 09:54:31 +0100 Subject: s390/cpum_sf: Use consistant debug print format for sampling Use consistant debug print format of the form variable blank value. Also add leading 0x for all hex values. Signed-off-by: Thomas Richter Signed-off-by: Vasily Gorbik --- arch/s390/kernel/perf_cpum_sf.c | 91 ++++++++++++++++++++--------------------- 1 file changed, 44 insertions(+), 47 deletions(-) (limited to 'arch/s390/kernel') diff --git a/arch/s390/kernel/perf_cpum_sf.c b/arch/s390/kernel/perf_cpum_sf.c index 3d8b12a9a6ff..504dac2be4dd 100644 --- a/arch/s390/kernel/perf_cpum_sf.c +++ b/arch/s390/kernel/perf_cpum_sf.c @@ -157,7 +157,7 @@ static void free_sampling_buffer(struct sf_buffer *sfb) } debug_sprintf_event(sfdbg, 5, - "free_sampling_buffer: freed sdbt=%p\n", sfb->sdbt); + "free_sampling_buffer: freed sdbt %p\n", sfb->sdbt); memset(sfb, 0, sizeof(*sfb)); } @@ -213,9 +213,9 @@ static int realloc_sampling_buffer(struct sf_buffer *sfb, */ if (sfb->sdbt != get_next_sdbt(tail)) { debug_sprintf_event(sfdbg, 3, "realloc_sampling_buffer: " - "sampling buffer is not linked: origin=%p" - "tail=%p\n", - (void *) sfb->sdbt, (void *) tail); + "sampling buffer is not linked: origin %p" + " tail %p\n", + (void *)sfb->sdbt, (void *)tail); return -EINVAL; } @@ -252,7 +252,7 @@ static int realloc_sampling_buffer(struct sf_buffer *sfb, sfb->tail = tail; debug_sprintf_event(sfdbg, 4, "realloc_sampling_buffer: new buffer" - " settings: sdbt=%lu sdb=%lu\n", + " settings: sdbt %lu sdb %lu\n", sfb->num_sdbt, sfb->num_sdb); return rc; } @@ -293,11 +293,11 @@ static int alloc_sampling_buffer(struct sf_buffer *sfb, unsigned long num_sdb) if (rc) { free_sampling_buffer(sfb); debug_sprintf_event(sfdbg, 4, "alloc_sampling_buffer: " - "realloc_sampling_buffer failed with rc=%i\n", rc); + "realloc_sampling_buffer failed with rc %i\n", rc); } else debug_sprintf_event(sfdbg, 4, - "alloc_sampling_buffer: tear=%p dear=%p\n", - sfb->sdbt, (void *) *sfb->sdbt); + "alloc_sampling_buffer: tear %p dear %p\n", + sfb->sdbt, (void *)*sfb->sdbt); return rc; } @@ -404,8 +404,8 @@ static int allocate_buffers(struct cpu_hw_sf *cpuhw, struct hw_perf_event *hwc) return 0; debug_sprintf_event(sfdbg, 3, - "allocate_buffers: rate=%lu f=%lu sdb=%lu/%lu" - " sample_size=%lu cpuhw=%p\n", + "allocate_buffers: rate %lu f %lu sdb %lu/%lu" + " sample_size %lu cpuhw %p\n", SAMPL_RATE(hwc), freq, n_sdb, sfb_max_limit(hwc), sample_size, cpuhw); @@ -465,8 +465,8 @@ static void sfb_account_overflows(struct cpu_hw_sf *cpuhw, if (num) sfb_account_allocs(num, hwc); - debug_sprintf_event(sfdbg, 5, "sfb: overflow: overflow=%llu ratio=%lu" - " num=%lu\n", OVERFLOW_REG(hwc), ratio, num); + debug_sprintf_event(sfdbg, 5, "sfb: overflow: overflow %llu ratio %lu" + " num %lu\n", OVERFLOW_REG(hwc), ratio, num); OVERFLOW_REG(hwc) = 0; } @@ -505,11 +505,11 @@ static void extend_sampling_buffer(struct sf_buffer *sfb, rc = realloc_sampling_buffer(sfb, num, GFP_ATOMIC); if (rc) debug_sprintf_event(sfdbg, 5, "sfb: extend: realloc " - "failed with rc=%i\n", rc); + "failed with rc %i\n", rc); if (sfb_has_pending_allocs(sfb, hwc)) debug_sprintf_event(sfdbg, 5, "sfb: extend: " - "req=%lu alloc=%lu remaining=%lu\n", + "req %lu alloc %lu remaining %lu\n", num, sfb->num_sdb - num_old, sfb_pending_allocs(sfb, hwc)); } @@ -538,20 +538,22 @@ static void setup_pmc_cpu(void *flags) err = sf_disable(); if (err) pr_err("Switching off the sampling facility failed " - "with rc=%i\n", err); + "with rc %i\n", err); debug_sprintf_event(sfdbg, 5, - "setup_pmc_cpu: initialized: cpuhw=%p\n", cpusf); + "setup_pmc_cpu: initialized: cpuhw %p\n", + cpusf); break; case PMC_RELEASE: cpusf->flags &= ~PMU_F_RESERVED; err = sf_disable(); if (err) { pr_err("Switching off the sampling facility failed " - "with rc=%i\n", err); + "with rc %i\n", err); } else deallocate_buffers(cpusf); debug_sprintf_event(sfdbg, 5, - "setup_pmc_cpu: released: cpuhw=%p\n", cpusf); + "setup_pmc_cpu: released: cpuhw %p\n", + cpusf); break; } if (err) @@ -744,7 +746,7 @@ static int __hw_perf_event_init_rate(struct perf_event *event, SAMPL_RATE(hwc) = rate; hw_init_period(hwc, SAMPL_RATE(hwc)); debug_sprintf_event(sfdbg, 4, "__hw_perf_event_init_rate:" - "cpu:%d period:%llx freq:%d,%#lx\n", event->cpu, + "cpu:%d period:%#llx freq:%d,%#lx\n", event->cpu, event->attr.sample_period, event->attr.freq, SAMPLE_FREQ_MODE(hwc)); return 0; @@ -963,7 +965,7 @@ static void cpumsf_pmu_enable(struct pmu *pmu) err = lsctl(&cpuhw->lsctl); if (err) { cpuhw->flags &= ~PMU_F_ENABLED; - pr_err("Loading sampling controls failed: op=%i err=%i\n", + pr_err("Loading sampling controls failed: op %i err %i\n", 1, err); return; } @@ -971,8 +973,8 @@ static void cpumsf_pmu_enable(struct pmu *pmu) /* Load current program parameter */ lpp(&S390_lowcore.lpp); - debug_sprintf_event(sfdbg, 6, "pmu_enable: es=%i cs=%i ed=%i cd=%i " - "interval:%lx tear=%p dear=%p\n", + debug_sprintf_event(sfdbg, 6, "pmu_enable: es %i cs %i ed %i cd %i " + "interval %#lx tear %p dear %p\n", cpuhw->lsctl.es, cpuhw->lsctl.cs, cpuhw->lsctl.ed, cpuhw->lsctl.cd, cpuhw->lsctl.interval, (void *) cpuhw->lsctl.tear, @@ -999,7 +1001,7 @@ static void cpumsf_pmu_disable(struct pmu *pmu) err = lsctl(&inactive); if (err) { - pr_err("Loading sampling controls failed: op=%i err=%i\n", + pr_err("Loading sampling controls failed: op %i err %i\n", 2, err); return; } @@ -1017,7 +1019,7 @@ static void cpumsf_pmu_disable(struct pmu *pmu) } } else debug_sprintf_event(sfdbg, 3, "cpumsf_pmu_disable: " - "qsi() failed with err=%i\n", err); + "qsi() failed with err %i\n", err); cpuhw->flags &= ~PMU_F_ENABLED; } @@ -1130,15 +1132,6 @@ static void perf_event_count_update(struct perf_event *event, u64 count) local64_add(count, &event->count); } -static void debug_sample_entry(struct hws_basic_entry *sample, - struct hws_trailer_entry *te) -{ - debug_sprintf_event(sfdbg, 4, "hw_collect_samples: Found unknown " - "sampling data entry: te->f=%i basic.def=%04x " - "(%p)\n", - te->f, sample->def, sample); -} - /* hw_collect_samples() - Walk through a sample-data-block and collect samples * @event: The perf event * @sdbt: Sample-data-block table @@ -1192,7 +1185,11 @@ static void hw_collect_samples(struct perf_event *event, unsigned long *sdbt, /* Count discarded samples */ *overflow += 1; } else { - debug_sample_entry(sample, te); + debug_sprintf_event(sfdbg, 4, + "hw_collect_samples: Found unknown" + " sampling data entry: te->f %i" + " basic.def %#4x (%p)\n", + te->f, sample->def, sample); /* Sample slot is not yet written or other record. * * This condition can occur if the buffer was reused @@ -1267,8 +1264,8 @@ static void hw_perf_event_update(struct perf_event *event, int flush_all) sampl_overflow += te->overflow; /* Timestamps are valid for full sample-data-blocks only */ - debug_sprintf_event(sfdbg, 6, "hw_perf_event_update: sdbt=%p " - "overflow=%llu timestamp=%#llx\n", + debug_sprintf_event(sfdbg, 6, "hw_perf_event_update: sdbt %p " + "overflow %llu timestamp %#llx\n", sdbt, te->overflow, (te->f) ? trailer_timestamp(te) : 0ULL); @@ -1314,7 +1311,7 @@ static void hw_perf_event_update(struct perf_event *event, int flush_all) sampl_overflow, 1 + num_sdb); if (sampl_overflow || event_overflow) debug_sprintf_event(sfdbg, 4, "hw_perf_event_update: " - "overflow stats: sample=%llu event=%llu\n", + "overflow stats: sample %llu event %llu\n", sampl_overflow, event_overflow); } @@ -1368,7 +1365,7 @@ static void aux_output_end(struct perf_output_handle *handle) te = aux_sdb_trailer(aux, aux->alert_mark); te->flags &= ~SDB_TE_ALERT_REQ_MASK; - debug_sprintf_event(sfdbg, 6, "aux_output_end: collect %lx SDBs\n", i); + debug_sprintf_event(sfdbg, 6, "aux_output_end: collect %#lx SDBs\n", i); } /* @@ -1428,8 +1425,8 @@ static int aux_output_begin(struct perf_output_handle *handle, debug_sprintf_event(sfdbg, 6, "aux_output_begin: " "head->alert_mark->empty_mark (num_alert, range)" - "[%lx -> %lx -> %lx] (%lx, %lx) " - "tear index %lx, tear %lx dear %lx\n", + "[%#lx -> %#lx -> %#lx] (%#lx, %#lx) " + "tear index %#lx, tear %#lx dear %#lx\n", aux->head, aux->alert_mark, aux->empty_mark, AUX_SDB_NUM_ALERT(aux), range, head / CPUM_SF_SDB_PER_TABLE, @@ -1596,13 +1593,13 @@ static void hw_collect_aux(struct cpu_hw_sf *cpuhw) perf_aux_output_end(&cpuhw->handle, size); pr_err("Sample data caused the AUX buffer with %lu " "pages to overflow\n", num_sdb); - debug_sprintf_event(sfdbg, 1, "head %lx range %lx " - "overflow %llx\n", + debug_sprintf_event(sfdbg, 1, "head %#lx range %#lx " + "overflow %#llx\n", aux->head, range, overflow); } else { size = AUX_SDB_NUM_ALERT(aux) << PAGE_SHIFT; perf_aux_output_end(&cpuhw->handle, size); - debug_sprintf_event(sfdbg, 6, "head %lx alert %lx " + debug_sprintf_event(sfdbg, 6, "head %#lx alert %#lx " "already full, try another\n", aux->head, aux->alert_mark); } @@ -1610,7 +1607,7 @@ static void hw_collect_aux(struct cpu_hw_sf *cpuhw) if (done) debug_sprintf_event(sfdbg, 6, "aux_reset_buffer: " - "[%lx -> %lx -> %lx] (%lx, %lx)\n", + "[%#lx -> %#lx -> %#lx] (%#lx, %#lx)\n", aux->head, aux->alert_mark, aux->empty_mark, AUX_SDB_NUM_ALERT(aux), range); } @@ -1800,7 +1797,7 @@ static int cpumsf_pmu_check_period(struct perf_event *event, u64 value) SAMPL_RATE(&event->hw) = rate; hw_init_period(&event->hw, SAMPL_RATE(&event->hw)); debug_sprintf_event(sfdbg, 4, "cpumsf_pmu_check_period:" - "cpu:%d value:%llx period:%llx freq:%d\n", + "cpu:%d value:%#llx period:%#llx freq:%d\n", event->cpu, value, event->attr.sample_period, do_freq); return 0; @@ -2111,7 +2108,7 @@ static int param_set_sfb_size(const char *val, const struct kernel_param *kp) sfb_set_limits(min, max); pr_info("The sampling buffer limits have changed to: " - "min=%lu max=%lu (diag=x%lu)\n", + "min %lu max %lu (diag %lu)\n", CPUM_SF_MIN_SDB, CPUM_SF_MAX_SDB, CPUM_SF_SDB_DIAG_FACTOR); return 0; } @@ -2129,7 +2126,7 @@ static const struct kernel_param_ops param_ops_sfb_size = { static void __init pr_cpumsf_err(unsigned int reason) { pr_err("Sampling facility support for perf is not available: " - "reason=%04x\n", reason); + "reason %#x\n", reason); } static int __init init_cpum_sampling_pmu(void) -- cgit From c18388340c4165c53f82d30569e8a5fc2f8a850e Mon Sep 17 00:00:00 2001 From: Thomas Richter Date: Thu, 31 Oct 2019 10:21:05 +0100 Subject: s390/cpum_sf: Replace function name in debug statements Replace hard coded function names in debug statements by the "%s ...", __func__ construct suggested by checkpatch.pl script. Signed-off-by: Thomas Richter Signed-off-by: Vasily Gorbik --- arch/s390/kernel/perf_cpum_sf.c | 30 +++++++++++++++--------------- 1 file changed, 15 insertions(+), 15 deletions(-) (limited to 'arch/s390/kernel') diff --git a/arch/s390/kernel/perf_cpum_sf.c b/arch/s390/kernel/perf_cpum_sf.c index 504dac2be4dd..e7587343e272 100644 --- a/arch/s390/kernel/perf_cpum_sf.c +++ b/arch/s390/kernel/perf_cpum_sf.c @@ -156,8 +156,8 @@ static void free_sampling_buffer(struct sf_buffer *sfb) } } - debug_sprintf_event(sfdbg, 5, - "free_sampling_buffer: freed sdbt %p\n", sfb->sdbt); + debug_sprintf_event(sfdbg, 5, "%s freed sdbt %p\n", __func__, + sfb->sdbt); memset(sfb, 0, sizeof(*sfb)); } @@ -212,9 +212,9 @@ static int realloc_sampling_buffer(struct sf_buffer *sfb, * the sampling buffer origin. */ if (sfb->sdbt != get_next_sdbt(tail)) { - debug_sprintf_event(sfdbg, 3, "realloc_sampling_buffer: " + debug_sprintf_event(sfdbg, 3, "%s: " "sampling buffer is not linked: origin %p" - " tail %p\n", + " tail %p\n", __func__, (void *)sfb->sdbt, (void *)tail); return -EINVAL; } @@ -404,8 +404,8 @@ static int allocate_buffers(struct cpu_hw_sf *cpuhw, struct hw_perf_event *hwc) return 0; debug_sprintf_event(sfdbg, 3, - "allocate_buffers: rate %lu f %lu sdb %lu/%lu" - " sample_size %lu cpuhw %p\n", + "%s: rate %lu f %lu sdb %lu/%lu" + " sample_size %lu cpuhw %p\n", __func__, SAMPL_RATE(hwc), freq, n_sdb, sfb_max_limit(hwc), sample_size, cpuhw); @@ -540,7 +540,7 @@ static void setup_pmc_cpu(void *flags) pr_err("Switching off the sampling facility failed " "with rc %i\n", err); debug_sprintf_event(sfdbg, 5, - "setup_pmc_cpu: initialized: cpuhw %p\n", + "%s: initialized: cpuhw %p\n", __func__, cpusf); break; case PMC_RELEASE: @@ -552,7 +552,7 @@ static void setup_pmc_cpu(void *flags) } else deallocate_buffers(cpusf); debug_sprintf_event(sfdbg, 5, - "setup_pmc_cpu: released: cpuhw %p\n", + "%s: released: cpuhw %p\n", __func__, cpusf); break; } @@ -1186,9 +1186,9 @@ static void hw_collect_samples(struct perf_event *event, unsigned long *sdbt, *overflow += 1; } else { debug_sprintf_event(sfdbg, 4, - "hw_collect_samples: Found unknown" + "%s: Found unknown" " sampling data entry: te->f %i" - " basic.def %#4x (%p)\n", + " basic.def %#4x (%p)\n", __func__, te->f, sample->def, sample); /* Sample slot is not yet written or other record. * @@ -1264,9 +1264,9 @@ static void hw_perf_event_update(struct perf_event *event, int flush_all) sampl_overflow += te->overflow; /* Timestamps are valid for full sample-data-blocks only */ - debug_sprintf_event(sfdbg, 6, "hw_perf_event_update: sdbt %p " + debug_sprintf_event(sfdbg, 6, "%s: sdbt %p " "overflow %llu timestamp %#llx\n", - sdbt, te->overflow, + __func__, sdbt, te->overflow, (te->f) ? trailer_timestamp(te) : 0ULL); /* Collect all samples from a single sample-data-block and @@ -1310,9 +1310,9 @@ static void hw_perf_event_update(struct perf_event *event, int flush_all) OVERFLOW_REG(hwc) = DIV_ROUND_UP(OVERFLOW_REG(hwc) + sampl_overflow, 1 + num_sdb); if (sampl_overflow || event_overflow) - debug_sprintf_event(sfdbg, 4, "hw_perf_event_update: " + debug_sprintf_event(sfdbg, 4, "%s: " "overflow stats: sample %llu event %llu\n", - sampl_overflow, event_overflow); + __func__, sampl_overflow, event_overflow); } #define AUX_SDB_INDEX(aux, i) ((i) % aux->sfb.num_sdb) @@ -1365,7 +1365,7 @@ static void aux_output_end(struct perf_output_handle *handle) te = aux_sdb_trailer(aux, aux->alert_mark); te->flags &= ~SDB_TE_ALERT_REQ_MASK; - debug_sprintf_event(sfdbg, 6, "aux_output_end: collect %#lx SDBs\n", i); + debug_sprintf_event(sfdbg, 6, "%s: collect %#lx SDBs\n", __func__, i); } /* -- cgit From 72fbcd057fa1d25da70e33312a0c8401d4295599 Mon Sep 17 00:00:00 2001 From: Thomas Richter Date: Thu, 31 Oct 2019 10:38:53 +0100 Subject: s390/cpum_sf: Assign error value to err variable When starting the CPU Measurement sampling facility using qsi() function, this function may return an error value. This error value is referenced in the else part of the if statement to dump its value in a debug statement. Right now this value is always zero because it has not been assigned a value. Signed-off-by: Thomas Richter Signed-off-by: Vasily Gorbik --- arch/s390/kernel/perf_cpum_sf.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'arch/s390/kernel') diff --git a/arch/s390/kernel/perf_cpum_sf.c b/arch/s390/kernel/perf_cpum_sf.c index e7587343e272..69506fdbd9a1 100644 --- a/arch/s390/kernel/perf_cpum_sf.c +++ b/arch/s390/kernel/perf_cpum_sf.c @@ -1007,7 +1007,8 @@ static void cpumsf_pmu_disable(struct pmu *pmu) } /* Save state of TEAR and DEAR register contents */ - if (!qsi(&si)) { + err = qsi(&si); + if (!err) { /* TEAR/DEAR values are valid only if the sampling facility is * enabled. Note that cpumsf_pmu_disable() might be called even * for a disabled sampling facility because cpumsf_pmu_enable() -- cgit From 544f1d62e3e6c6e6d17a5e56f6139208acb5ff46 Mon Sep 17 00:00:00 2001 From: Ilya Leoshkevich Date: Thu, 31 Oct 2019 18:25:16 +0100 Subject: s390/disassembler: don't hide instruction addresses Due to kptr_restrict, JITted BPF code is now displayed like this: 000000000b6ed1b2: ebdff0800024 stmg %r13,%r15,128(%r15) 000000004cde2ba0: 41d0f040 la %r13,64(%r15) 00000000fbad41b0: a7fbffa0 aghi %r15,-96 Leaking kernel addresses to dmesg is not a concern in this case, because this happens only when JIT debugging is explicitly activated, which only root can do. Use %px in this particular instance, and also to print an instruction address in show_code and PCREL (e.g. brasl) arguments in print_insn. While at present functionally equivalent to %016lx, %px is recommended by Documentation/core-api/printk-formats.rst for such cases. Signed-off-by: Ilya Leoshkevich Reviewed-by: Vasily Gorbik Signed-off-by: Vasily Gorbik --- arch/s390/kernel/dis.c | 13 +++++++------ 1 file changed, 7 insertions(+), 6 deletions(-) (limited to 'arch/s390/kernel') diff --git a/arch/s390/kernel/dis.c b/arch/s390/kernel/dis.c index 7abe6ae261b4..f304802ecf7b 100644 --- a/arch/s390/kernel/dis.c +++ b/arch/s390/kernel/dis.c @@ -461,10 +461,11 @@ static int print_insn(char *buffer, unsigned char *code, unsigned long addr) ptr += sprintf(ptr, "%%c%i", value); else if (operand->flags & OPERAND_VR) ptr += sprintf(ptr, "%%v%i", value); - else if (operand->flags & OPERAND_PCREL) - ptr += sprintf(ptr, "%lx", (signed int) value - + addr); - else if (operand->flags & OPERAND_SIGNED) + else if (operand->flags & OPERAND_PCREL) { + void *pcrel = (void *)((int)value + addr); + + ptr += sprintf(ptr, "%px", pcrel); + } else if (operand->flags & OPERAND_SIGNED) ptr += sprintf(ptr, "%i", value); else ptr += sprintf(ptr, "%u", value); @@ -536,7 +537,7 @@ void show_code(struct pt_regs *regs) else *ptr++ = ' '; addr = regs->psw.addr + start - 32; - ptr += sprintf(ptr, "%016lx: ", addr); + ptr += sprintf(ptr, "%px: ", (void *)addr); if (start + opsize >= end) break; for (i = 0; i < opsize; i++) @@ -564,7 +565,7 @@ void print_fn_code(unsigned char *code, unsigned long len) opsize = insn_length(*code); if (opsize > len) break; - ptr += sprintf(ptr, "%p: ", code); + ptr += sprintf(ptr, "%px: ", code); for (i = 0; i < opsize; i++) ptr += sprintf(ptr, "%02x", code[i]); *ptr++ = '\t'; -- cgit From c02ee6a16a260ae00a403be440e229fd8618486b Mon Sep 17 00:00:00 2001 From: Vasily Gorbik Date: Tue, 12 Nov 2019 17:35:28 +0100 Subject: s390/early: move control registers setup in C code Reviewed-by: Heiko Carstens Signed-off-by: Vasily Gorbik --- arch/s390/kernel/early.c | 12 ++++++++++++ arch/s390/kernel/head64.S | 6 ------ 2 files changed, 12 insertions(+), 6 deletions(-) (limited to 'arch/s390/kernel') diff --git a/arch/s390/kernel/early.c b/arch/s390/kernel/early.c index 2e99e01e4f62..0ed6ae6a8bde 100644 --- a/arch/s390/kernel/early.c +++ b/arch/s390/kernel/early.c @@ -260,6 +260,17 @@ static inline void save_vector_registers(void) #endif } +static inline void setup_control_registers(void) +{ + unsigned long reg; + + __ctl_store(reg, 0, 0); + reg |= CR0_LOW_ADDRESS_PROTECTION; + reg |= CR0_EMERGENCY_SIGNAL_SUBMASK; + reg |= CR0_EXTERNAL_CALL_SUBMASK; + __ctl_load(reg, 0, 0); +} + static int __init disable_vector_extension(char *str) { S390_lowcore.machine_flags &= ~MACHINE_FLAG_VX; @@ -317,5 +328,6 @@ void __init startup_init(void) save_vector_registers(); setup_topology(); sclp_early_detect(); + setup_control_registers(); lockdep_on(); } diff --git a/arch/s390/kernel/head64.S b/arch/s390/kernel/head64.S index 0d9ee198f4eb..5b21702ea080 100644 --- a/arch/s390/kernel/head64.S +++ b/arch/s390/kernel/head64.S @@ -42,12 +42,6 @@ ENTRY(startup_continue) # brasl %r14,startup_init -# check control registers - stctg %c0,%c15,0(%r15) - oi 6(%r15),0x60 # enable sigp emergency & external call - oi 4(%r15),0x10 # switch on low address proctection - lctlg %c0,%c15,0(%r15) - lam 0,15,.Laregs-.LPG1(%r13) # load acrs needed by uaccess brasl %r14,start_kernel # go to C code # -- cgit From b8ce1fa4892cccff0576827c8fd7fcad1698de3b Mon Sep 17 00:00:00 2001 From: Vasily Gorbik Date: Wed, 13 Nov 2019 11:22:30 +0100 Subject: s390/head64: remove unnecessary vdso_per_cpu_data setup vdso_per_cpu_data lowcore value is only needed for fully functional exception handlers, which are activated in setup_lowcore_dat_off. The same function does init vdso_per_cpu_data via vdso_alloc_boot_cpu. Reviewed-by: Heiko Carstens Signed-off-by: Vasily Gorbik --- arch/s390/kernel/head64.S | 2 -- 1 file changed, 2 deletions(-) (limited to 'arch/s390/kernel') diff --git a/arch/s390/kernel/head64.S b/arch/s390/kernel/head64.S index 5b21702ea080..9d649abddffd 100644 --- a/arch/s390/kernel/head64.S +++ b/arch/s390/kernel/head64.S @@ -26,8 +26,6 @@ ENTRY(startup_continue) 0: larl %r1,tod_clock_base mvc 0(16,%r1),__LC_BOOT_CLOCK larl %r13,.LPG1 # get base - larl %r0,boot_vdso_data - stg %r0,__LC_VDSO_PER_CPU # # Setup stack # -- cgit From c2313594216b3fde9559e502bb36d14e9d601a56 Mon Sep 17 00:00:00 2001 From: Vasily Gorbik Date: Thu, 14 Nov 2019 14:08:28 +0100 Subject: s390/early: move access registers setup in C code Reviewed-by: Heiko Carstens Signed-off-by: Vasily Gorbik --- arch/s390/kernel/early.c | 9 +++++++++ arch/s390/kernel/head64.S | 10 ++-------- 2 files changed, 11 insertions(+), 8 deletions(-) (limited to 'arch/s390/kernel') diff --git a/arch/s390/kernel/early.c b/arch/s390/kernel/early.c index 0ed6ae6a8bde..db32a55daaec 100644 --- a/arch/s390/kernel/early.c +++ b/arch/s390/kernel/early.c @@ -30,6 +30,7 @@ #include #include #include +#include #include "entry.h" static void __init reset_tod_clock(void) @@ -271,6 +272,13 @@ static inline void setup_control_registers(void) __ctl_load(reg, 0, 0); } +static inline void setup_access_registers(void) +{ + unsigned int acrs[NUM_ACRS] = { 0 }; + + restore_access_regs(acrs); +} + static int __init disable_vector_extension(char *str) { S390_lowcore.machine_flags &= ~MACHINE_FLAG_VX; @@ -329,5 +337,6 @@ void __init startup_init(void) setup_topology(); sclp_early_detect(); setup_control_registers(); + setup_access_registers(); lockdep_on(); } diff --git a/arch/s390/kernel/head64.S b/arch/s390/kernel/head64.S index 9d649abddffd..b9e585f528a6 100644 --- a/arch/s390/kernel/head64.S +++ b/arch/s390/kernel/head64.S @@ -35,13 +35,8 @@ ENTRY(startup_continue) #ifdef CONFIG_KASAN brasl %r14,kasan_early_init #endif -# -# Early machine initialization and detection functions. -# - brasl %r14,startup_init - - lam 0,15,.Laregs-.LPG1(%r13) # load acrs needed by uaccess - brasl %r14,start_kernel # go to C code + brasl %r14,startup_init # s390 specific early init + brasl %r14,start_kernel # common init code # # We returned from start_kernel ?!? PANIK # @@ -51,4 +46,3 @@ ENTRY(startup_continue) .align 16 .LPG1: .Ldw: .quad 0x0002000180000000,0x0000000000000000 -.Laregs:.long 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 -- cgit From 72a81ad9d6d62dcb79f7e8ad66ffd1c768b72026 Mon Sep 17 00:00:00 2001 From: Heiko Carstens Date: Sun, 17 Nov 2019 14:55:38 +0100 Subject: s390/smp: fix physical to logical CPU map for SMT If an SMT capable system is not IPL'ed from the first CPU the setup of the physical to logical CPU mapping is broken: the IPL core gets CPU number 0, but then the next core gets CPU number 1. Correct would be that all SMT threads of CPU 0 get the subsequent logical CPU numbers. This is important since a lot of code (like e.g. the CPU topology code) assumes that CPU maps are setup like this. If the mapping is broken the system will not IPL due to broken topology masks: [ 1.716341] BUG: arch topology broken [ 1.716342] the SMT domain not a subset of the MC domain [ 1.716343] BUG: arch topology broken [ 1.716344] the MC domain not a subset of the BOOK domain This scenario can usually not happen since LPARs are always IPL'ed from CPU 0 and also re-IPL is intiated from CPU 0. However older kernels did initiate re-IPL on an arbitrary CPU. If therefore a re-IPL from an old kernel into a new kernel is initiated this may lead to crash. Fix this by setting up the physical to logical CPU mapping correctly. Signed-off-by: Heiko Carstens Signed-off-by: Vasily Gorbik --- arch/s390/kernel/smp.c | 80 ++++++++++++++++++++++++++++++++++---------------- 1 file changed, 54 insertions(+), 26 deletions(-) (limited to 'arch/s390/kernel') diff --git a/arch/s390/kernel/smp.c b/arch/s390/kernel/smp.c index 44974654cbd0..6acdcf1d4074 100644 --- a/arch/s390/kernel/smp.c +++ b/arch/s390/kernel/smp.c @@ -724,39 +724,67 @@ static void __ref smp_get_core_info(struct sclp_core_info *info, int early) static int smp_add_present_cpu(int cpu); -static int __smp_rescan_cpus(struct sclp_core_info *info, int sysfs_add) +static int smp_add_core(struct sclp_core_entry *core, cpumask_t *avail, + bool configured, bool early) { struct pcpu *pcpu; - cpumask_t avail; - int cpu, nr, i, j; + int cpu, nr, i; u16 address; nr = 0; - cpumask_xor(&avail, cpu_possible_mask, cpu_present_mask); - cpu = cpumask_first(&avail); - for (i = 0; (i < info->combined) && (cpu < nr_cpu_ids); i++) { - if (sclp.has_core_type && info->core[i].type != boot_core_type) + if (sclp.has_core_type && core->type != boot_core_type) + return nr; + cpu = cpumask_first(avail); + address = core->core_id << smp_cpu_mt_shift; + for (i = 0; (i <= smp_cpu_mtid) && (cpu < nr_cpu_ids); i++) { + if (pcpu_find_address(cpu_present_mask, address + i)) continue; - address = info->core[i].core_id << smp_cpu_mt_shift; - for (j = 0; j <= smp_cpu_mtid; j++) { - if (pcpu_find_address(cpu_present_mask, address + j)) - continue; - pcpu = pcpu_devices + cpu; - pcpu->address = address + j; - pcpu->state = - (cpu >= info->configured*(smp_cpu_mtid + 1)) ? - CPU_STATE_STANDBY : CPU_STATE_CONFIGURED; - smp_cpu_set_polarization(cpu, POLARIZATION_UNKNOWN); - set_cpu_present(cpu, true); - if (sysfs_add && smp_add_present_cpu(cpu) != 0) - set_cpu_present(cpu, false); - else - nr++; - cpu = cpumask_next(cpu, &avail); - if (cpu >= nr_cpu_ids) + pcpu = pcpu_devices + cpu; + pcpu->address = address + i; + if (configured) + pcpu->state = CPU_STATE_CONFIGURED; + else + pcpu->state = CPU_STATE_STANDBY; + smp_cpu_set_polarization(cpu, POLARIZATION_UNKNOWN); + set_cpu_present(cpu, true); + if (!early && smp_add_present_cpu(cpu) != 0) + set_cpu_present(cpu, false); + else + nr++; + cpumask_clear_cpu(cpu, avail); + cpu = cpumask_next(cpu, avail); + } + return nr; +} + +static int __smp_rescan_cpus(struct sclp_core_info *info, bool early) +{ + struct sclp_core_entry *core; + cpumask_t avail; + bool configured; + u16 core_id; + int nr, i; + + nr = 0; + cpumask_xor(&avail, cpu_possible_mask, cpu_present_mask); + /* + * Add IPL core first (which got logical CPU number 0) to make sure + * that all SMT threads get subsequent logical CPU numbers. + */ + if (early) { + core_id = pcpu_devices[0].address >> smp_cpu_mt_shift; + for (i = 0; i < info->configured; i++) { + core = &info->core[i]; + if (core->core_id == core_id) { + nr += smp_add_core(core, &avail, true, early); break; + } } } + for (i = 0; i < info->combined; i++) { + configured = i < info->configured; + nr += smp_add_core(&info->core[i], &avail, configured, early); + } return nr; } @@ -805,7 +833,7 @@ void __init smp_detect_cpus(void) /* Add CPUs present at boot */ get_online_cpus(); - __smp_rescan_cpus(info, 0); + __smp_rescan_cpus(info, true); put_online_cpus(); memblock_free_early((unsigned long)info, sizeof(*info)); } @@ -1148,7 +1176,7 @@ int __ref smp_rescan_cpus(void) smp_get_core_info(info, 0); get_online_cpus(); mutex_lock(&smp_cpu_state_mutex); - nr = __smp_rescan_cpus(info, 1); + nr = __smp_rescan_cpus(info, false); mutex_unlock(&smp_cpu_state_mutex); put_online_cpus(); kfree(info); -- cgit From 6a82e23f45fe0aa821e7a935e39d0acb20c275c0 Mon Sep 17 00:00:00 2001 From: Thomas Richter Date: Fri, 15 Nov 2019 12:50:46 +0100 Subject: s390/cpumf: Adjust registration of s390 PMU device drivers Linux-next commit titled "perf/core: Optimize perf_init_event()" changed the semantics of PMU device driver registration. It was done to speed up the lookup/handling of PMU device driver specific events. It also enforces that only one PMU device driver will be registered of type PERF_EVENT_RAW. This change added these line in function perf_pmu_register(): ... + ret = idr_alloc(&pmu_idr, pmu, max, 0, GFP_KERNEL); + if (ret < 0) goto free_pdc; + + WARN_ON(type >= 0 && ret != type); The warn_on generates a message. We have 3 PMU device drivers, each registered as type PERF_TYPE_RAW. The cf_diag device driver (arch/s390/kernel/perf_cpumf_cf_diag.c) always hits the WARN_ON because it is the second PMU device driver (after sampling device driver arch/s390/kernel/perf_cpumf_sf.c) which is registered as type 4 (PERF_TYPE_RAW). So when the sampling device driver is registered, ret has value 4. When cf_diag device driver is registered with type 4, ret has value of 5 and WARN_ON fires. Adjust the PMU device drivers for s390 to support the new semantics required by perf_pmu_register(). Signed-off-by: Thomas Richter Signed-off-by: Vasily Gorbik --- arch/s390/kernel/perf_cpum_cf.c | 21 ++++++++++----------- arch/s390/kernel/perf_cpum_cf_diag.c | 10 +++++----- 2 files changed, 15 insertions(+), 16 deletions(-) (limited to 'arch/s390/kernel') diff --git a/arch/s390/kernel/perf_cpum_cf.c b/arch/s390/kernel/perf_cpum_cf.c index 48d48b6187c0..0eb1d1cc53a8 100644 --- a/arch/s390/kernel/perf_cpum_cf.c +++ b/arch/s390/kernel/perf_cpum_cf.c @@ -199,7 +199,7 @@ static const int cpumf_generic_events_user[] = { [PERF_COUNT_HW_BUS_CYCLES] = -1, }; -static int __hw_perf_event_init(struct perf_event *event) +static int __hw_perf_event_init(struct perf_event *event, unsigned int type) { struct perf_event_attr *attr = &event->attr; struct hw_perf_event *hwc = &event->hw; @@ -207,7 +207,7 @@ static int __hw_perf_event_init(struct perf_event *event) int err = 0; u64 ev; - switch (attr->type) { + switch (type) { case PERF_TYPE_RAW: /* Raw events are used to access counters directly, * hence do not permit excludes */ @@ -294,17 +294,16 @@ static int __hw_perf_event_init(struct perf_event *event) static int cpumf_pmu_event_init(struct perf_event *event) { + unsigned int type = event->attr.type; int err; - switch (event->attr.type) { - case PERF_TYPE_HARDWARE: - case PERF_TYPE_HW_CACHE: - case PERF_TYPE_RAW: - err = __hw_perf_event_init(event); - break; - default: + if (type == PERF_TYPE_HARDWARE || type == PERF_TYPE_RAW) + err = __hw_perf_event_init(event, type); + else if (event->pmu->type == type) + /* Registered as unknown PMU */ + err = __hw_perf_event_init(event, PERF_TYPE_RAW); + else return -ENOENT; - } if (unlikely(err) && event->destroy) event->destroy(event); @@ -553,7 +552,7 @@ static int __init cpumf_pmu_init(void) return -ENODEV; cpumf_pmu.attr_groups = cpumf_cf_event_group(); - rc = perf_pmu_register(&cpumf_pmu, "cpum_cf", PERF_TYPE_RAW); + rc = perf_pmu_register(&cpumf_pmu, "cpum_cf", -1); if (rc) pr_err("Registering the cpum_cf PMU failed with rc=%i\n", rc); return rc; diff --git a/arch/s390/kernel/perf_cpum_cf_diag.c b/arch/s390/kernel/perf_cpum_cf_diag.c index 2654e348801a..e949ab832ed7 100644 --- a/arch/s390/kernel/perf_cpum_cf_diag.c +++ b/arch/s390/kernel/perf_cpum_cf_diag.c @@ -243,13 +243,13 @@ static int cf_diag_event_init(struct perf_event *event) int err = -ENOENT; debug_sprintf_event(cf_diag_dbg, 5, - "%s event %p cpu %d config %#llx " + "%s event %p cpu %d config %#llx type:%u " "sample_type %#llx cf_diag_events %d\n", __func__, - event, event->cpu, attr->config, attr->sample_type, - atomic_read(&cf_diag_events)); + event, event->cpu, attr->config, event->pmu->type, + attr->sample_type, atomic_read(&cf_diag_events)); if (event->attr.config != PERF_EVENT_CPUM_CF_DIAG || - event->attr.type != PERF_TYPE_RAW) + event->attr.type != event->pmu->type) goto out; /* Raw events are used to access counters directly, @@ -693,7 +693,7 @@ static int __init cf_diag_init(void) } debug_register_view(cf_diag_dbg, &debug_sprintf_view); - rc = perf_pmu_register(&cf_diag, "cpum_cf_diag", PERF_TYPE_RAW); + rc = perf_pmu_register(&cf_diag, "cpum_cf_diag", -1); if (rc) { debug_unregister_view(cf_diag_dbg, &debug_sprintf_view); debug_unregister(cf_diag_dbg); -- cgit