From ddfb7d9db843fd4fbdff81fb8a8743f865c3dd96 Mon Sep 17 00:00:00 2001 From: Benjamin Gray Date: Wed, 11 Oct 2023 16:37:03 +1100 Subject: powerpc: Use NULL instead of 0 for null pointers Sparse reports several uses of 0 for pointer arguments and comparisons. Replace with NULL to better convey the intent. Remove entirely if a comparison to follow the kernel style of implicit boolean conversions. Signed-off-by: Benjamin Gray Signed-off-by: Michael Ellerman Link: https://msgid.link/20231011053711.93427-5-bgray@linux.ibm.com --- arch/powerpc/perf/imc-pmu.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/powerpc/perf') diff --git a/arch/powerpc/perf/imc-pmu.c b/arch/powerpc/perf/imc-pmu.c index 9d229ef7f86e..eba2baabccb0 100644 --- a/arch/powerpc/perf/imc-pmu.c +++ b/arch/powerpc/perf/imc-pmu.c @@ -544,7 +544,7 @@ static int nest_imc_event_init(struct perf_event *event) break; } pcni++; - } while (pcni->vbase != 0); + } while (pcni->vbase); if (!flag) return -ENODEV; -- cgit v1.2.3 From 2b4a6cc9a1a7cf6958c8b11f94e61c8e81b60b88 Mon Sep 17 00:00:00 2001 From: Benjamin Gray Date: Wed, 11 Oct 2023 16:37:05 +1100 Subject: powerpc: Annotate endianness of various variables and functions Sparse reports several endianness warnings on variables and functions that are consistently treated as big endian. There are no multi-endianness shenanigans going on here so fix these low hanging fruit up in one patch. All changes are just type annotations; no endianness switching operations are introduced by this patch. Signed-off-by: Benjamin Gray Signed-off-by: Michael Ellerman Link: https://msgid.link/20231011053711.93427-7-bgray@linux.ibm.com --- arch/powerpc/include/asm/book3s/64/pgtable.h | 2 +- arch/powerpc/include/asm/imc-pmu.h | 16 ++++++++-------- arch/powerpc/kernel/prom_init.c | 2 +- arch/powerpc/kexec/core_64.c | 4 ++-- arch/powerpc/kexec/file_load_64.c | 6 +++--- arch/powerpc/mm/drmem.c | 2 +- arch/powerpc/perf/hv-24x7.c | 2 +- arch/powerpc/perf/imc-pmu.c | 9 +++++---- arch/powerpc/platforms/powermac/feature.c | 3 ++- arch/powerpc/platforms/pseries/hotplug-memory.c | 3 ++- 10 files changed, 26 insertions(+), 23 deletions(-) (limited to 'arch/powerpc/perf') diff --git a/arch/powerpc/include/asm/book3s/64/pgtable.h b/arch/powerpc/include/asm/book3s/64/pgtable.h index 751b01227e36..cb77eddca54b 100644 --- a/arch/powerpc/include/asm/book3s/64/pgtable.h +++ b/arch/powerpc/include/asm/book3s/64/pgtable.h @@ -626,7 +626,7 @@ static inline pte_t pte_mkdevmap(pte_t pte) */ static inline int pte_devmap(pte_t pte) { - u64 mask = cpu_to_be64(_PAGE_DEVMAP | _PAGE_PTE); + __be64 mask = cpu_to_be64(_PAGE_DEVMAP | _PAGE_PTE); return (pte_raw(pte) & mask) == mask; } diff --git a/arch/powerpc/include/asm/imc-pmu.h b/arch/powerpc/include/asm/imc-pmu.h index 699a88584ae1..a656635df386 100644 --- a/arch/powerpc/include/asm/imc-pmu.h +++ b/arch/powerpc/include/asm/imc-pmu.h @@ -74,14 +74,14 @@ struct imc_events { * The following is the data structure to hold trace imc data. */ struct trace_imc_data { - u64 tb1; - u64 ip; - u64 val; - u64 cpmc1; - u64 cpmc2; - u64 cpmc3; - u64 cpmc4; - u64 tb2; + __be64 tb1; + __be64 ip; + __be64 val; + __be64 cpmc1; + __be64 cpmc2; + __be64 cpmc3; + __be64 cpmc4; + __be64 tb2; }; /* Event attribute array index */ diff --git a/arch/powerpc/kernel/prom_init.c b/arch/powerpc/kernel/prom_init.c index d464ba412084..e67effdba85c 100644 --- a/arch/powerpc/kernel/prom_init.c +++ b/arch/powerpc/kernel/prom_init.c @@ -947,7 +947,7 @@ struct option_vector7 { } __packed; struct ibm_arch_vec { - struct { u32 mask, val; } pvrs[14]; + struct { __be32 mask, val; } pvrs[14]; u8 num_vectors; diff --git a/arch/powerpc/kexec/core_64.c b/arch/powerpc/kexec/core_64.c index a79e28c91e2b..0bee7ca9a77c 100644 --- a/arch/powerpc/kexec/core_64.c +++ b/arch/powerpc/kexec/core_64.c @@ -379,8 +379,8 @@ void default_machine_kexec(struct kimage *image) #ifdef CONFIG_PPC_64S_HASH_MMU /* Values we need to export to the second kernel via the device tree. */ -static unsigned long htab_base; -static unsigned long htab_size; +static __be64 htab_base; +static __be64 htab_size; static struct property htab_base_prop = { .name = "linux,htab-base", diff --git a/arch/powerpc/kexec/file_load_64.c b/arch/powerpc/kexec/file_load_64.c index 09187aca3d1f..961a6dd67365 100644 --- a/arch/powerpc/kexec/file_load_64.c +++ b/arch/powerpc/kexec/file_load_64.c @@ -32,7 +32,7 @@ #include struct umem_info { - u64 *buf; /* data buffer for usable-memory property */ + __be64 *buf; /* data buffer for usable-memory property */ u32 size; /* size allocated for the data buffer */ u32 max_entries; /* maximum no. of entries */ u32 idx; /* index of current entry */ @@ -443,10 +443,10 @@ static int locate_mem_hole_bottom_up_ppc64(struct kexec_buf *kbuf, * * Returns buffer on success, NULL on error. */ -static u64 *check_realloc_usable_mem(struct umem_info *um_info, int cnt) +static __be64 *check_realloc_usable_mem(struct umem_info *um_info, int cnt) { u32 new_size; - u64 *tbuf; + __be64 *tbuf; if ((um_info->idx + cnt) <= um_info->max_entries) return um_info->buf; diff --git a/arch/powerpc/mm/drmem.c b/arch/powerpc/mm/drmem.c index 2369d1bf2411..fde7790277f7 100644 --- a/arch/powerpc/mm/drmem.c +++ b/arch/powerpc/mm/drmem.c @@ -67,7 +67,7 @@ static int drmem_update_dt_v1(struct device_node *memory, struct property *new_prop; struct of_drconf_cell_v1 *dr_cell; struct drmem_lmb *lmb; - u32 *p; + __be32 *p; new_prop = clone_property(prop, prop->length); if (!new_prop) diff --git a/arch/powerpc/perf/hv-24x7.c b/arch/powerpc/perf/hv-24x7.c index 3449be7c0d51..057ec2e3451d 100644 --- a/arch/powerpc/perf/hv-24x7.c +++ b/arch/powerpc/perf/hv-24x7.c @@ -1338,7 +1338,7 @@ static int get_count_from_result(struct perf_event *event, for (i = count = 0, element_data = res->elements + data_offset; i < num_elements; i++, element_data += data_size + data_offset) - count += be64_to_cpu(*((u64 *) element_data)); + count += be64_to_cpu(*((__be64 *)element_data)); *countp = count; diff --git a/arch/powerpc/perf/imc-pmu.c b/arch/powerpc/perf/imc-pmu.c index eba2baabccb0..2016aee27037 100644 --- a/arch/powerpc/perf/imc-pmu.c +++ b/arch/powerpc/perf/imc-pmu.c @@ -1025,16 +1025,16 @@ static bool is_thread_imc_pmu(struct perf_event *event) return false; } -static u64 * get_event_base_addr(struct perf_event *event) +static __be64 *get_event_base_addr(struct perf_event *event) { u64 addr; if (is_thread_imc_pmu(event)) { addr = (u64)per_cpu(thread_imc_mem, smp_processor_id()); - return (u64 *)(addr + (event->attr.config & IMC_EVENT_OFFSET_MASK)); + return (__be64 *)(addr + (event->attr.config & IMC_EVENT_OFFSET_MASK)); } - return (u64 *)event->hw.event_base; + return (__be64 *)event->hw.event_base; } static void thread_imc_pmu_start_txn(struct pmu *pmu, @@ -1058,7 +1058,8 @@ static int thread_imc_pmu_commit_txn(struct pmu *pmu) static u64 imc_read_counter(struct perf_event *event) { - u64 *addr, data; + __be64 *addr; + u64 data; /* * In-Memory Collection (IMC) counters are free flowing counters. diff --git a/arch/powerpc/platforms/powermac/feature.c b/arch/powerpc/platforms/powermac/feature.c index ae62d432db8b..81c9fbae88b1 100644 --- a/arch/powerpc/platforms/powermac/feature.c +++ b/arch/powerpc/platforms/powermac/feature.c @@ -2614,7 +2614,8 @@ static void __init probe_one_macio(const char *name, const char *compat, int typ struct device_node* node; int i; volatile u32 __iomem *base; - const u32 *addrp, *revp; + const __be32 *addrp; + const u32 *revp; phys_addr_t addr; u64 size; diff --git a/arch/powerpc/platforms/pseries/hotplug-memory.c b/arch/powerpc/platforms/pseries/hotplug-memory.c index aa4042dcd6d4..a43bfb01720a 100644 --- a/arch/powerpc/platforms/pseries/hotplug-memory.c +++ b/arch/powerpc/platforms/pseries/hotplug-memory.c @@ -55,7 +55,8 @@ static bool find_aa_index(struct device_node *dr_node, struct property *ala_prop, const u32 *lmb_assoc, u32 *aa_index) { - u32 *assoc_arrays, new_prop_size; + __be32 *assoc_arrays; + u32 new_prop_size; struct property *new_prop; int aa_arrays, aa_array_entries, aa_array_sz; int i, index; -- cgit v1.2.3 From e08c43e6c3eb5d805b61d981f1e8286ee0dc6d1a Mon Sep 17 00:00:00 2001 From: Kuan-Wei Chiu Date: Sat, 14 Oct 2023 01:57:14 +0800 Subject: powerpc/perf: Optimize find_alternatives_list() using binary search This patch improves the performance of event alternative lookup by replacing the previous linear search with a more efficient binary search. This change reduces the time complexity for the search process from O(n) to O(log(n)). A pre-sorted table of event values and their corresponding indices has been introduced to expedite the search process. Signed-off-by: Kuan-Wei Chiu [mpe: Call the array "presort*ed*_event_table", minor formatting] Signed-off-by: Michael Ellerman Link: https://msgid.link/20231013175714.2142775-1-visitorckw@gmail.com --- arch/powerpc/perf/power6-pmu.c | 46 ++++++++++++++++++++++++++---------------- 1 file changed, 29 insertions(+), 17 deletions(-) (limited to 'arch/powerpc/perf') diff --git a/arch/powerpc/perf/power6-pmu.c b/arch/powerpc/perf/power6-pmu.c index 5729b6e059de..9f720b522e17 100644 --- a/arch/powerpc/perf/power6-pmu.c +++ b/arch/powerpc/perf/power6-pmu.c @@ -335,26 +335,38 @@ static const unsigned int event_alternatives[][MAX_ALT] = { { 0x3000fe, 0x400056 }, /* PM_DATA_FROM_L3MISS */ }; -/* - * This could be made more efficient with a binary search on - * a presorted list, if necessary - */ static int find_alternatives_list(u64 event) { - int i, j; - unsigned int alt; - - for (i = 0; i < ARRAY_SIZE(event_alternatives); ++i) { - if (event < event_alternatives[i][0]) - return -1; - for (j = 0; j < MAX_ALT; ++j) { - alt = event_alternatives[i][j]; - if (!alt || event < alt) - break; - if (event == alt) - return i; - } + const unsigned int presorted_event_table[] = { + 0x0130e8, 0x080080, 0x080088, 0x10000a, 0x10000b, 0x10000d, 0x10000e, + 0x100010, 0x10001a, 0x100026, 0x100054, 0x100056, 0x1000f0, 0x1000f8, + 0x1000fc, 0x200008, 0x20000e, 0x200010, 0x200012, 0x200054, 0x2000f0, + 0x2000f2, 0x2000f4, 0x2000f5, 0x2000f6, 0x2000f8, 0x2000fc, 0x2000fe, + 0x2d0030, 0x30000a, 0x30000c, 0x300010, 0x300012, 0x30001a, 0x300056, + 0x3000f0, 0x3000f2, 0x3000f6, 0x3000f8, 0x3000fc, 0x3000fe, 0x400006, + 0x400007, 0x40000a, 0x40000e, 0x400010, 0x400018, 0x400056, 0x4000f0, + 0x4000f8, 0x600005 + }; + const unsigned int event_index_table[] = { + 0, 1, 2, 3, 4, 1, 5, 6, 7, 8, 9, 10, 11, 12, 13, 12, 14, + 7, 15, 2, 9, 16, 3, 4, 0, 17, 10, 18, 19, 20, 1, 17, 15, 19, + 18, 2, 16, 21, 8, 0, 22, 13, 14, 11, 21, 5, 20, 22, 1, 6, 3 + }; + int hi = ARRAY_SIZE(presorted_event_table) - 1; + int lo = 0; + + while (lo <= hi) { + int mid = lo + (hi - lo) / 2; + unsigned int alt = presorted_event_table[mid]; + + if (alt < event) + lo = mid + 1; + else if (alt > event) + hi = mid - 1; + else + return event_index_table[mid]; } + return -1; } -- cgit v1.2.3 From 007240d59c11f87ac4f6cfc6a1d116630b6b634c Mon Sep 17 00:00:00 2001 From: Sebastian Andrzej Siewior Date: Thu, 9 Mar 2023 14:48:31 +0100 Subject: powerpc/imc-pmu: Use the correct spinlock initializer. The macro __SPIN_LOCK_INITIALIZER() is implementation specific. Users that desire to initialize a spinlock in a struct must use __SPIN_LOCK_UNLOCKED(). Use __SPIN_LOCK_UNLOCKED() for the spinlock_t in imc_global_refc. Fixes: 76d588dddc459 ("powerpc/imc-pmu: Fix use of mutex in IRQs disabled section") Signed-off-by: Sebastian Andrzej Siewior Signed-off-by: Michael Ellerman Link: https://msgid.link/20230309134831.Nz12nqsU@linutronix.de --- arch/powerpc/perf/imc-pmu.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/powerpc/perf') diff --git a/arch/powerpc/perf/imc-pmu.c b/arch/powerpc/perf/imc-pmu.c index 2016aee27037..5d12ca386c1f 100644 --- a/arch/powerpc/perf/imc-pmu.c +++ b/arch/powerpc/perf/imc-pmu.c @@ -51,7 +51,7 @@ static int trace_imc_mem_size; * core and trace-imc */ static struct imc_pmu_ref imc_global_refc = { - .lock = __SPIN_LOCK_INITIALIZER(imc_global_refc.lock), + .lock = __SPIN_LOCK_UNLOCKED(imc_global_refc.lock), .id = 0, .refc = 0, }; -- cgit v1.2.3 From ea142e590aec55ba40c5affb4d49e68c713c63dc Mon Sep 17 00:00:00 2001 From: Nicholas Piggin Date: Thu, 19 Oct 2023 01:34:23 +1000 Subject: powerpc/perf: Fix disabling BHRB and instruction sampling When the PMU is disabled, MMCRA is not updated to disable BHRB and instruction sampling. This can lead to those features remaining enabled, which can slow down a real or emulated CPU. Fixes: 1cade527f6e9 ("powerpc/perf: BHRB control to disable BHRB logic when not used") Cc: stable@vger.kernel.org # v5.9+ Signed-off-by: Nicholas Piggin Signed-off-by: Michael Ellerman Link: https://msgid.link/20231018153423.298373-1-npiggin@gmail.com --- arch/powerpc/perf/core-book3s.c | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-) (limited to 'arch/powerpc/perf') diff --git a/arch/powerpc/perf/core-book3s.c b/arch/powerpc/perf/core-book3s.c index 8c1f7def596e..10b946e9c6e7 100644 --- a/arch/powerpc/perf/core-book3s.c +++ b/arch/powerpc/perf/core-book3s.c @@ -1371,8 +1371,7 @@ static void power_pmu_disable(struct pmu *pmu) /* * Disable instruction sampling if it was enabled */ - if (cpuhw->mmcr.mmcra & MMCRA_SAMPLE_ENABLE) - val &= ~MMCRA_SAMPLE_ENABLE; + val &= ~MMCRA_SAMPLE_ENABLE; /* Disable BHRB via mmcra (BHRBRD) for p10 */ if (ppmu->flags & PPMU_ARCH_31) @@ -1383,7 +1382,7 @@ static void power_pmu_disable(struct pmu *pmu) * instruction sampling or BHRB. */ if (val != mmcra) { - mtspr(SPRN_MMCRA, mmcra); + mtspr(SPRN_MMCRA, val); mb(); isync(); } -- cgit v1.2.3