X-Git-Url: http://git.lttng.org/?a=blobdiff_plain;f=lib%2Fringbuffer%2Fring_buffer_backend.c;h=8207f764377a34f5cb0fda0e41da8e56985dbd45;hb=1f0ab1eb0409d23de5f67cc588c3ea4cee4d10e0;hp=332a1d23ae46fef3484670d9b4c8c7e6ad252b31;hpb=2fb46300208fedb9d1a6e93de359674d2a02a5a7;p=lttng-modules.git diff --git a/lib/ringbuffer/ring_buffer_backend.c b/lib/ringbuffer/ring_buffer_backend.c index 332a1d23..8207f764 100644 --- a/lib/ringbuffer/ring_buffer_backend.c +++ b/lib/ringbuffer/ring_buffer_backend.c @@ -1,9 +1,8 @@ -/* - * ring_buffer_backend.c +/* SPDX-License-Identifier: (GPL-2.0 OR LGPL-2.1) * - * Copyright (C) 2005-2010 - Mathieu Desnoyers + * ring_buffer_backend.c * - * Dual LGPL v2.1/GPL v2 license. + * Copyright (C) 2005-2012 Mathieu Desnoyers */ #include @@ -13,13 +12,15 @@ #include #include #include +#include #include #include +#include -#include "../../wrapper/vmalloc.h" /* for wrapper_vmalloc_sync_all() */ -#include "../../wrapper/ringbuffer/config.h" -#include "../../wrapper/ringbuffer/backend.h" -#include "../../wrapper/ringbuffer/frontend.h" +#include /* for wrapper_vmalloc_sync_all() */ +#include +#include +#include /** * lib_ring_buffer_backend_allocate - allocate a channel buffer @@ -40,10 +41,26 @@ int lib_ring_buffer_backend_allocate(const struct lib_ring_buffer_config *config unsigned long subbuf_size, mmap_offset = 0; unsigned long num_subbuf_alloc; struct page **pages; - void **virt; unsigned long i; num_pages = size >> PAGE_SHIFT; + + /* + * Verify that the number of pages requested for that buffer is smaller + * than the number of available pages on the system. si_mem_available() + * returns an _estimate_ of the number of available pages. + */ + if (num_pages > si_mem_available()) + goto not_enough_pages; + + /* + * Set the current user thread as the first target of the OOM killer. + * If the estimate received by si_mem_available() was off, and we do + * end up running out of memory because of this buffer allocation, we + * want to kill the offending app first. + */ + set_current_oom_origin(); + num_pages_per_subbuf = num_pages >> get_count_order(num_subbuf); subbuf_size = chanb->subbuf_size; num_subbuf_alloc = num_subbuf; @@ -53,53 +70,49 @@ int lib_ring_buffer_backend_allocate(const struct lib_ring_buffer_config *config num_subbuf_alloc++; } - pages = kmalloc_node(ALIGN(sizeof(*pages) * num_pages, + pages = vmalloc_node(ALIGN(sizeof(*pages) * num_pages, 1 << INTERNODE_CACHE_SHIFT), - GFP_KERNEL, cpu_to_node(max(bufb->cpu, 0))); + cpu_to_node(max(bufb->cpu, 0))); if (unlikely(!pages)) goto pages_error; - virt = kmalloc_node(ALIGN(sizeof(*virt) * num_pages, - 1 << INTERNODE_CACHE_SHIFT), - GFP_KERNEL, cpu_to_node(max(bufb->cpu, 0))); - if (unlikely(!virt)) - goto virt_error; - - bufb->array = kmalloc_node(ALIGN(sizeof(*bufb->array) + bufb->array = lttng_kvmalloc_node(ALIGN(sizeof(*bufb->array) * num_subbuf_alloc, 1 << INTERNODE_CACHE_SHIFT), - GFP_KERNEL, cpu_to_node(max(bufb->cpu, 0))); + GFP_KERNEL | __GFP_NOWARN, + cpu_to_node(max(bufb->cpu, 0))); if (unlikely(!bufb->array)) goto array_error; for (i = 0; i < num_pages; i++) { pages[i] = alloc_pages_node(cpu_to_node(max(bufb->cpu, 0)), - GFP_KERNEL | __GFP_ZERO, 0); + GFP_KERNEL | __GFP_NOWARN | __GFP_ZERO, 0); if (unlikely(!pages[i])) goto depopulate; - virt[i] = page_address(pages[i]); } bufb->num_pages_per_subbuf = num_pages_per_subbuf; /* Allocate backend pages array elements */ for (i = 0; i < num_subbuf_alloc; i++) { bufb->array[i] = - kzalloc_node(ALIGN( + lttng_kvzalloc_node(ALIGN( sizeof(struct lib_ring_buffer_backend_pages) + sizeof(struct lib_ring_buffer_backend_page) * num_pages_per_subbuf, 1 << INTERNODE_CACHE_SHIFT), - GFP_KERNEL, cpu_to_node(max(bufb->cpu, 0))); + GFP_KERNEL | __GFP_NOWARN, + cpu_to_node(max(bufb->cpu, 0))); if (!bufb->array[i]) goto free_array; } /* Allocate write-side subbuffer table */ - bufb->buf_wsb = kzalloc_node(ALIGN( + bufb->buf_wsb = lttng_kvzalloc_node(ALIGN( sizeof(struct lib_ring_buffer_backend_subbuffer) * num_subbuf, 1 << INTERNODE_CACHE_SHIFT), - GFP_KERNEL, cpu_to_node(max(bufb->cpu, 0))); + GFP_KERNEL | __GFP_NOWARN, + cpu_to_node(max(bufb->cpu, 0))); if (unlikely(!bufb->buf_wsb)) goto free_array; @@ -113,12 +126,22 @@ int lib_ring_buffer_backend_allocate(const struct lib_ring_buffer_config *config else bufb->buf_rsb.id = subbuffer_id(config, 0, 1, 0); + /* Allocate subbuffer packet counter table */ + bufb->buf_cnt = lttng_kvzalloc_node(ALIGN( + sizeof(struct lib_ring_buffer_backend_counts) + * num_subbuf, + 1 << INTERNODE_CACHE_SHIFT), + GFP_KERNEL | __GFP_NOWARN, + cpu_to_node(max(bufb->cpu, 0))); + if (unlikely(!bufb->buf_cnt)) + goto free_wsb; + /* Assign pages to page index */ for (i = 0; i < num_subbuf_alloc; i++) { for (j = 0; j < num_pages_per_subbuf; j++) { CHAN_WARN_ON(chanb, page_idx > num_pages); - bufb->array[i]->p[j].virt = virt[page_idx]; - bufb->array[i]->p[j].page = pages[page_idx]; + bufb->array[i]->p[j].virt = page_address(pages[page_idx]); + bufb->array[i]->p[j].pfn = page_to_pfn(pages[page_idx]); page_idx++; } if (config->output == RING_BUFFER_MMAP) { @@ -132,30 +155,32 @@ int lib_ring_buffer_backend_allocate(const struct lib_ring_buffer_config *config * will not fault. */ wrapper_vmalloc_sync_all(); - kfree(virt); - kfree(pages); + clear_current_oom_origin(); + vfree(pages); return 0; +free_wsb: + lttng_kvfree(bufb->buf_wsb); free_array: for (i = 0; (i < num_subbuf_alloc && bufb->array[i]); i++) - kfree(bufb->array[i]); + lttng_kvfree(bufb->array[i]); depopulate: /* Free all allocated pages */ for (i = 0; (i < num_pages && pages[i]); i++) __free_page(pages[i]); - kfree(bufb->array); + lttng_kvfree(bufb->array); array_error: - kfree(virt); -virt_error: - kfree(pages); + vfree(pages); pages_error: + clear_current_oom_origin(); +not_enough_pages: return -ENOMEM; } int lib_ring_buffer_backend_create(struct lib_ring_buffer_backend *bufb, struct channel_backend *chanb, int cpu) { - const struct lib_ring_buffer_config *config = chanb->config; + const struct lib_ring_buffer_config *config = &chanb->config; bufb->chan = container_of(chanb, struct channel, backend); bufb->cpu = cpu; @@ -174,20 +199,21 @@ void lib_ring_buffer_backend_free(struct lib_ring_buffer_backend *bufb) if (chanb->extra_reader_sb) num_subbuf_alloc++; - kfree(bufb->buf_wsb); + lttng_kvfree(bufb->buf_wsb); + lttng_kvfree(bufb->buf_cnt); for (i = 0; i < num_subbuf_alloc; i++) { for (j = 0; j < bufb->num_pages_per_subbuf; j++) - __free_page(bufb->array[i]->p[j].page); - kfree(bufb->array[i]); + __free_page(pfn_to_page(bufb->array[i]->p[j].pfn)); + lttng_kvfree(bufb->array[i]); } - kfree(bufb->array); + lttng_kvfree(bufb->array); bufb->allocated = 0; } void lib_ring_buffer_backend_reset(struct lib_ring_buffer_backend *bufb) { struct channel_backend *chanb = &bufb->chan->backend; - const struct lib_ring_buffer_config *config = chanb->config; + const struct lib_ring_buffer_config *config = &chanb->config; unsigned long num_subbuf_alloc; unsigned int i; @@ -221,7 +247,7 @@ void lib_ring_buffer_backend_reset(struct lib_ring_buffer_backend *bufb) void channel_backend_reset(struct channel_backend *chanb) { struct channel *chan = container_of(chanb, struct channel, backend); - const struct lib_ring_buffer_config *config = chanb->config; + const struct lib_ring_buffer_config *config = &chanb->config; /* * Don't reset buf_size, subbuf_size, subbuf_size_order, @@ -231,7 +257,42 @@ void channel_backend_reset(struct channel_backend *chanb) chanb->start_tsc = config->cb.ring_buffer_clock_read(chan); } +#if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,10,0)) + +/* + * No need to implement a "dead" callback to do a buffer switch here, + * because it will happen when tracing is stopped, or will be done by + * switch timer CPU DEAD callback. + * We don't free buffers when CPU go away, because it would make trace + * data vanish, which is unwanted. + */ +int lttng_cpuhp_rb_backend_prepare(unsigned int cpu, + struct lttng_cpuhp_node *node) +{ + struct channel_backend *chanb = container_of(node, + struct channel_backend, cpuhp_prepare); + const struct lib_ring_buffer_config *config = &chanb->config; + struct lib_ring_buffer *buf; + int ret; + + CHAN_WARN_ON(chanb, config->alloc == RING_BUFFER_ALLOC_GLOBAL); + + buf = per_cpu_ptr(chanb->buf, cpu); + ret = lib_ring_buffer_create(buf, chanb, cpu); + if (ret) { + printk(KERN_ERR + "ring_buffer_cpu_hp_callback: cpu %d " + "buffer creation failed\n", cpu); + return ret; + } + return 0; +} +EXPORT_SYMBOL_GPL(lttng_cpuhp_rb_backend_prepare); + +#else /* #if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,10,0)) */ + #ifdef CONFIG_HOTPLUG_CPU + /** * lib_ring_buffer_cpu_hp_callback - CPU hotplug callback * @nb: notifier block @@ -241,14 +302,14 @@ void channel_backend_reset(struct channel_backend *chanb) * Returns the success/failure of the operation. (%NOTIFY_OK, %NOTIFY_BAD) */ static -int __cpuinit lib_ring_buffer_cpu_hp_callback(struct notifier_block *nb, +int lib_ring_buffer_cpu_hp_callback(struct notifier_block *nb, unsigned long action, void *hcpu) { unsigned int cpu = (unsigned long)hcpu; struct channel_backend *chanb = container_of(nb, struct channel_backend, cpu_hp_notifier); - const struct lib_ring_buffer_config *config = chanb->config; + const struct lib_ring_buffer_config *config = &chanb->config; struct lib_ring_buffer *buf; int ret; @@ -275,8 +336,11 @@ int __cpuinit lib_ring_buffer_cpu_hp_callback(struct notifier_block *nb, } return NOTIFY_OK; } + #endif +#endif /* #else #if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,10,0)) */ + /** * channel_backend_init - initialize a channel backend * @chanb: channel backend @@ -307,18 +371,24 @@ int channel_backend_init(struct channel_backend *chanb, if (!name) return -EPERM; - if (!(subbuf_size && num_subbuf)) - return -EPERM; - /* Check that the subbuffer size is larger than a page. */ if (subbuf_size < PAGE_SIZE) return -EINVAL; /* - * Make sure the number of subbuffers and subbuffer size are power of 2. + * Make sure the number of subbuffers and subbuffer size are + * power of 2 and nonzero. */ - CHAN_WARN_ON(chanb, hweight32(subbuf_size) != 1); - CHAN_WARN_ON(chanb, hweight32(num_subbuf) != 1); + if (!subbuf_size || (subbuf_size & (subbuf_size - 1))) + return -EINVAL; + if (!num_subbuf || (num_subbuf & (num_subbuf - 1))) + return -EINVAL; + /* + * Overwrite mode buffers require at least 2 subbuffers per + * buffer. + */ + if (config->mode == RING_BUFFER_OVERWRITE && num_subbuf < 2) + return -EINVAL; ret = subbuffer_id_check_index(config, num_subbuf); if (ret) @@ -334,7 +404,7 @@ int channel_backend_init(struct channel_backend *chanb, (config->mode == RING_BUFFER_OVERWRITE) ? 1 : 0; chanb->num_subbuf = num_subbuf; strlcpy(chanb->name, name, NAME_MAX); - chanb->config = config; + memcpy(&chanb->config, config, sizeof(chanb->config)); if (config->alloc == RING_BUFFER_ALLOC_PER_CPU) { if (!zalloc_cpumask_var(&chanb->cpumask, GFP_KERNEL)) @@ -347,39 +417,50 @@ int channel_backend_init(struct channel_backend *chanb, if (!chanb->buf) goto free_cpumask; - /* - * In case of non-hotplug cpu, if the ring-buffer is allocated - * in early initcall, it will not be notified of secondary cpus. - * In that off case, we need to allocate for all possible cpus. - */ +#if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,10,0)) + chanb->cpuhp_prepare.component = LTTNG_RING_BUFFER_BACKEND; + ret = cpuhp_state_add_instance(lttng_rb_hp_prepare, + &chanb->cpuhp_prepare.node); + if (ret) + goto free_bufs; +#else /* #if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,10,0)) */ + + { + /* + * In case of non-hotplug cpu, if the ring-buffer is allocated + * in early initcall, it will not be notified of secondary cpus. + * In that off case, we need to allocate for all possible cpus. + */ #ifdef CONFIG_HOTPLUG_CPU - /* - * buf->backend.allocated test takes care of concurrent CPU - * hotplug. - * Priority higher than frontend, so we create the ring buffer - * before we start the timer. - */ - chanb->cpu_hp_notifier.notifier_call = - lib_ring_buffer_cpu_hp_callback; - chanb->cpu_hp_notifier.priority = 5; - register_hotcpu_notifier(&chanb->cpu_hp_notifier); - - get_online_cpus(); - for_each_online_cpu(i) { - ret = lib_ring_buffer_create(per_cpu_ptr(chanb->buf, i), - chanb, i); - if (ret) - goto free_bufs; /* cpu hotplug locked */ - } - put_online_cpus(); + /* + * buf->backend.allocated test takes care of concurrent CPU + * hotplug. + * Priority higher than frontend, so we create the ring buffer + * before we start the timer. + */ + chanb->cpu_hp_notifier.notifier_call = + lib_ring_buffer_cpu_hp_callback; + chanb->cpu_hp_notifier.priority = 5; + register_hotcpu_notifier(&chanb->cpu_hp_notifier); + + get_online_cpus(); + for_each_online_cpu(i) { + ret = lib_ring_buffer_create(per_cpu_ptr(chanb->buf, i), + chanb, i); + if (ret) + goto free_bufs; /* cpu hotplug locked */ + } + put_online_cpus(); #else - for_each_possible_cpu(i) { - ret = lib_ring_buffer_create(per_cpu_ptr(chanb->buf, i), - chanb, i); - if (ret) - goto free_bufs; /* cpu hotplug locked */ - } + for_each_possible_cpu(i) { + ret = lib_ring_buffer_create(per_cpu_ptr(chanb->buf, i), + chanb, i); + if (ret) + goto free_bufs; + } #endif + } +#endif /* #else #if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,10,0)) */ } else { chanb->buf = kzalloc(sizeof(struct lib_ring_buffer), GFP_KERNEL); if (!chanb->buf) @@ -394,16 +475,26 @@ int channel_backend_init(struct channel_backend *chanb, free_bufs: if (config->alloc == RING_BUFFER_ALLOC_PER_CPU) { +#if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,10,0)) + /* + * Teardown of lttng_rb_hp_prepare instance + * on "add" error is handled within cpu hotplug, + * no teardown to do from the caller. + */ +#else /* #if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,10,0)) */ +#ifdef CONFIG_HOTPLUG_CPU + put_online_cpus(); + unregister_hotcpu_notifier(&chanb->cpu_hp_notifier); +#endif +#endif /* #else #if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,10,0)) */ for_each_possible_cpu(i) { - struct lib_ring_buffer *buf = per_cpu_ptr(chanb->buf, i); + struct lib_ring_buffer *buf = + per_cpu_ptr(chanb->buf, i); if (!buf->backend.allocated) continue; lib_ring_buffer_free(buf); } -#ifdef CONFIG_HOTPLUG_CPU - put_online_cpus(); -#endif free_percpu(chanb->buf); } else kfree(chanb->buf); @@ -421,10 +512,19 @@ free_cpumask: */ void channel_backend_unregister_notifiers(struct channel_backend *chanb) { - const struct lib_ring_buffer_config *config = chanb->config; + const struct lib_ring_buffer_config *config = &chanb->config; - if (config->alloc == RING_BUFFER_ALLOC_PER_CPU) + if (config->alloc == RING_BUFFER_ALLOC_PER_CPU) { +#if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,10,0)) + int ret; + + ret = cpuhp_state_remove_instance(lttng_rb_hp_prepare, + &chanb->cpuhp_prepare.node); + WARN_ON(ret); +#else /* #if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,10,0)) */ unregister_hotcpu_notifier(&chanb->cpu_hp_notifier); +#endif /* #else #if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,10,0)) */ + } } /** @@ -435,7 +535,7 @@ void channel_backend_unregister_notifiers(struct channel_backend *chanb) */ void channel_backend_free(struct channel_backend *chanb) { - const struct lib_ring_buffer_config *config = chanb->config; + const struct lib_ring_buffer_config *config = &chanb->config; unsigned int i; if (config->alloc == RING_BUFFER_ALLOC_PER_CPU) { @@ -466,10 +566,10 @@ void channel_backend_free(struct channel_backend *chanb) * @pagecpy : page size copied so far */ void _lib_ring_buffer_write(struct lib_ring_buffer_backend *bufb, size_t offset, - const void *src, size_t len, ssize_t pagecpy) + const void *src, size_t len, size_t pagecpy) { struct channel_backend *chanb = &bufb->chan->backend; - const struct lib_ring_buffer_config *config = chanb->config; + const struct lib_ring_buffer_config *config = &chanb->config; size_t sbidx, index; struct lib_ring_buffer_backend_pages *rpages; unsigned long sb_bindex, id; @@ -501,6 +601,272 @@ void _lib_ring_buffer_write(struct lib_ring_buffer_backend *bufb, size_t offset, } EXPORT_SYMBOL_GPL(_lib_ring_buffer_write); + +/** + * lib_ring_buffer_memset - write len bytes of c to a ring_buffer buffer. + * @bufb : buffer backend + * @offset : offset within the buffer + * @c : the byte to write + * @len : length to write + * @pagecpy : page size copied so far + */ +void _lib_ring_buffer_memset(struct lib_ring_buffer_backend *bufb, + size_t offset, + int c, size_t len, size_t pagecpy) +{ + struct channel_backend *chanb = &bufb->chan->backend; + const struct lib_ring_buffer_config *config = &chanb->config; + size_t sbidx, index; + struct lib_ring_buffer_backend_pages *rpages; + unsigned long sb_bindex, id; + + do { + len -= pagecpy; + offset += pagecpy; + sbidx = offset >> chanb->subbuf_size_order; + index = (offset & (chanb->subbuf_size - 1)) >> PAGE_SHIFT; + + /* + * Underlying layer should never ask for writes across + * subbuffers. + */ + CHAN_WARN_ON(chanb, offset >= chanb->buf_size); + + pagecpy = min_t(size_t, len, PAGE_SIZE - (offset & ~PAGE_MASK)); + id = bufb->buf_wsb[sbidx].id; + sb_bindex = subbuffer_id_get_index(config, id); + rpages = bufb->array[sb_bindex]; + CHAN_WARN_ON(chanb, config->mode == RING_BUFFER_OVERWRITE + && subbuffer_id_is_noref(config, id)); + lib_ring_buffer_do_memset(rpages->p[index].virt + + (offset & ~PAGE_MASK), + c, pagecpy); + } while (unlikely(len != pagecpy)); +} +EXPORT_SYMBOL_GPL(_lib_ring_buffer_memset); + +/** + * lib_ring_buffer_strcpy - write string data to a ring_buffer buffer. + * @bufb : buffer backend + * @offset : offset within the buffer + * @src : source address + * @len : length to write + * @pagecpy : page size copied so far + * @pad : character to use for padding + */ +void _lib_ring_buffer_strcpy(struct lib_ring_buffer_backend *bufb, + size_t offset, const char *src, size_t len, + size_t pagecpy, int pad) +{ + struct channel_backend *chanb = &bufb->chan->backend; + const struct lib_ring_buffer_config *config = &chanb->config; + size_t sbidx, index; + struct lib_ring_buffer_backend_pages *rpages; + unsigned long sb_bindex, id; + int src_terminated = 0; + + CHAN_WARN_ON(chanb, !len); + offset += pagecpy; + do { + len -= pagecpy; + if (!src_terminated) + src += pagecpy; + sbidx = offset >> chanb->subbuf_size_order; + index = (offset & (chanb->subbuf_size - 1)) >> PAGE_SHIFT; + + /* + * Underlying layer should never ask for writes across + * subbuffers. + */ + CHAN_WARN_ON(chanb, offset >= chanb->buf_size); + + pagecpy = min_t(size_t, len, PAGE_SIZE - (offset & ~PAGE_MASK)); + id = bufb->buf_wsb[sbidx].id; + sb_bindex = subbuffer_id_get_index(config, id); + rpages = bufb->array[sb_bindex]; + CHAN_WARN_ON(chanb, config->mode == RING_BUFFER_OVERWRITE + && subbuffer_id_is_noref(config, id)); + + if (likely(!src_terminated)) { + size_t count, to_copy; + + to_copy = pagecpy; + if (pagecpy == len) + to_copy--; /* Final '\0' */ + count = lib_ring_buffer_do_strcpy(config, + rpages->p[index].virt + + (offset & ~PAGE_MASK), + src, to_copy); + offset += count; + /* Padding */ + if (unlikely(count < to_copy)) { + size_t pad_len = to_copy - count; + + /* Next pages will have padding */ + src_terminated = 1; + lib_ring_buffer_do_memset(rpages->p[index].virt + + (offset & ~PAGE_MASK), + pad, pad_len); + offset += pad_len; + } + } else { + size_t pad_len; + + pad_len = pagecpy; + if (pagecpy == len) + pad_len--; /* Final '\0' */ + lib_ring_buffer_do_memset(rpages->p[index].virt + + (offset & ~PAGE_MASK), + pad, pad_len); + offset += pad_len; + } + } while (unlikely(len != pagecpy)); + /* Ending '\0' */ + lib_ring_buffer_do_memset(rpages->p[index].virt + (offset & ~PAGE_MASK), + '\0', 1); +} +EXPORT_SYMBOL_GPL(_lib_ring_buffer_strcpy); + +/** + * lib_ring_buffer_copy_from_user_inatomic - write user data to a ring_buffer buffer. + * @bufb : buffer backend + * @offset : offset within the buffer + * @src : source address + * @len : length to write + * @pagecpy : page size copied so far + * + * This function deals with userspace pointers, it should never be called + * directly without having the src pointer checked with access_ok() + * previously. + */ +void _lib_ring_buffer_copy_from_user_inatomic(struct lib_ring_buffer_backend *bufb, + size_t offset, + const void __user *src, size_t len, + size_t pagecpy) +{ + struct channel_backend *chanb = &bufb->chan->backend; + const struct lib_ring_buffer_config *config = &chanb->config; + size_t sbidx, index; + struct lib_ring_buffer_backend_pages *rpages; + unsigned long sb_bindex, id; + int ret; + + do { + len -= pagecpy; + src += pagecpy; + offset += pagecpy; + sbidx = offset >> chanb->subbuf_size_order; + index = (offset & (chanb->subbuf_size - 1)) >> PAGE_SHIFT; + + /* + * Underlying layer should never ask for writes across + * subbuffers. + */ + CHAN_WARN_ON(chanb, offset >= chanb->buf_size); + + pagecpy = min_t(size_t, len, PAGE_SIZE - (offset & ~PAGE_MASK)); + id = bufb->buf_wsb[sbidx].id; + sb_bindex = subbuffer_id_get_index(config, id); + rpages = bufb->array[sb_bindex]; + CHAN_WARN_ON(chanb, config->mode == RING_BUFFER_OVERWRITE + && subbuffer_id_is_noref(config, id)); + ret = lib_ring_buffer_do_copy_from_user_inatomic(rpages->p[index].virt + + (offset & ~PAGE_MASK), + src, pagecpy) != 0; + if (ret > 0) { + /* Copy failed. */ + _lib_ring_buffer_memset(bufb, offset, 0, len, 0); + break; /* stop copy */ + } + } while (unlikely(len != pagecpy)); +} +EXPORT_SYMBOL_GPL(_lib_ring_buffer_copy_from_user_inatomic); + +/** + * lib_ring_buffer_strcpy_from_user_inatomic - write userspace string data to a ring_buffer buffer. + * @bufb : buffer backend + * @offset : offset within the buffer + * @src : source address + * @len : length to write + * @pagecpy : page size copied so far + * @pad : character to use for padding + * + * This function deals with userspace pointers, it should never be called + * directly without having the src pointer checked with access_ok() + * previously. + */ +void _lib_ring_buffer_strcpy_from_user_inatomic(struct lib_ring_buffer_backend *bufb, + size_t offset, const char __user *src, size_t len, + size_t pagecpy, int pad) +{ + struct channel_backend *chanb = &bufb->chan->backend; + const struct lib_ring_buffer_config *config = &chanb->config; + size_t sbidx, index; + struct lib_ring_buffer_backend_pages *rpages; + unsigned long sb_bindex, id; + int src_terminated = 0; + + offset += pagecpy; + do { + len -= pagecpy; + if (!src_terminated) + src += pagecpy; + sbidx = offset >> chanb->subbuf_size_order; + index = (offset & (chanb->subbuf_size - 1)) >> PAGE_SHIFT; + + /* + * Underlying layer should never ask for writes across + * subbuffers. + */ + CHAN_WARN_ON(chanb, offset >= chanb->buf_size); + + pagecpy = min_t(size_t, len, PAGE_SIZE - (offset & ~PAGE_MASK)); + id = bufb->buf_wsb[sbidx].id; + sb_bindex = subbuffer_id_get_index(config, id); + rpages = bufb->array[sb_bindex]; + CHAN_WARN_ON(chanb, config->mode == RING_BUFFER_OVERWRITE + && subbuffer_id_is_noref(config, id)); + + if (likely(!src_terminated)) { + size_t count, to_copy; + + to_copy = pagecpy; + if (pagecpy == len) + to_copy--; /* Final '\0' */ + count = lib_ring_buffer_do_strcpy_from_user_inatomic(config, + rpages->p[index].virt + + (offset & ~PAGE_MASK), + src, to_copy); + offset += count; + /* Padding */ + if (unlikely(count < to_copy)) { + size_t pad_len = to_copy - count; + + /* Next pages will have padding */ + src_terminated = 1; + lib_ring_buffer_do_memset(rpages->p[index].virt + + (offset & ~PAGE_MASK), + pad, pad_len); + offset += pad_len; + } + } else { + size_t pad_len; + + pad_len = pagecpy; + if (pagecpy == len) + pad_len--; /* Final '\0' */ + lib_ring_buffer_do_memset(rpages->p[index].virt + + (offset & ~PAGE_MASK), + pad, pad_len); + offset += pad_len; + } + } while (unlikely(len != pagecpy)); + /* Ending '\0' */ + lib_ring_buffer_do_memset(rpages->p[index].virt + (offset & ~PAGE_MASK), + '\0', 1); +} +EXPORT_SYMBOL_GPL(_lib_ring_buffer_strcpy_from_user_inatomic); + /** * lib_ring_buffer_read - read data from ring_buffer_buffer. * @bufb : buffer backend @@ -515,9 +881,8 @@ size_t lib_ring_buffer_read(struct lib_ring_buffer_backend *bufb, size_t offset, void *dest, size_t len) { struct channel_backend *chanb = &bufb->chan->backend; - const struct lib_ring_buffer_config *config = chanb->config; - size_t index; - ssize_t pagecpy, orig_len; + const struct lib_ring_buffer_config *config = &chanb->config; + size_t index, pagecpy, orig_len; struct lib_ring_buffer_backend_pages *rpages; unsigned long sb_bindex, id; @@ -567,13 +932,12 @@ int __lib_ring_buffer_copy_to_user(struct lib_ring_buffer_backend *bufb, size_t offset, void __user *dest, size_t len) { struct channel_backend *chanb = &bufb->chan->backend; - const struct lib_ring_buffer_config *config = chanb->config; + const struct lib_ring_buffer_config *config = &chanb->config; size_t index; - ssize_t pagecpy, orig_len; + ssize_t pagecpy; struct lib_ring_buffer_backend_pages *rpages; unsigned long sb_bindex, id; - orig_len = len; offset &= chanb->buf_size - 1; index = (offset & (chanb->subbuf_size - 1)) >> PAGE_SHIFT; if (unlikely(!len)) @@ -612,14 +976,15 @@ EXPORT_SYMBOL_GPL(__lib_ring_buffer_copy_to_user); * @dest : destination address * @len : destination's length * - * return string's length + * Return string's length, or -EINVAL on error. * Should be protected by get_subbuf/put_subbuf. + * Destination length should be at least 1 to hold '\0'. */ int lib_ring_buffer_read_cstr(struct lib_ring_buffer_backend *bufb, size_t offset, void *dest, size_t len) { struct channel_backend *chanb = &bufb->chan->backend; - const struct lib_ring_buffer_config *config = chanb->config; + const struct lib_ring_buffer_config *config = &chanb->config; size_t index; ssize_t pagecpy, pagelen, strpagelen, orig_offset; char *str; @@ -629,6 +994,8 @@ int lib_ring_buffer_read_cstr(struct lib_ring_buffer_backend *bufb, size_t offse offset &= chanb->buf_size - 1; index = (offset & (chanb->subbuf_size - 1)) >> PAGE_SHIFT; orig_offset = offset; + if (unlikely(!len)) + return -EINVAL; for (;;) { id = bufb->buf_rsb.id; sb_bindex = subbuffer_id_get_index(config, id); @@ -663,21 +1030,21 @@ int lib_ring_buffer_read_cstr(struct lib_ring_buffer_backend *bufb, size_t offse EXPORT_SYMBOL_GPL(lib_ring_buffer_read_cstr); /** - * lib_ring_buffer_read_get_page - Get a whole page to read from + * lib_ring_buffer_read_get_pfn - Get a page frame number to read from * @bufb : buffer backend * @offset : offset within the buffer * @virt : pointer to page address (output) * * Should be protected by get_subbuf/put_subbuf. - * Returns the pointer to the page struct pointer. + * Returns the pointer to the page frame number unsigned long. */ -struct page **lib_ring_buffer_read_get_page(struct lib_ring_buffer_backend *bufb, +unsigned long *lib_ring_buffer_read_get_pfn(struct lib_ring_buffer_backend *bufb, size_t offset, void ***virt) { size_t index; struct lib_ring_buffer_backend_pages *rpages; struct channel_backend *chanb = &bufb->chan->backend; - const struct lib_ring_buffer_config *config = chanb->config; + const struct lib_ring_buffer_config *config = &chanb->config; unsigned long sb_bindex, id; offset &= chanb->buf_size - 1; @@ -688,9 +1055,9 @@ struct page **lib_ring_buffer_read_get_page(struct lib_ring_buffer_backend *bufb CHAN_WARN_ON(chanb, config->mode == RING_BUFFER_OVERWRITE && subbuffer_id_is_noref(config, id)); *virt = &rpages->p[index].virt; - return &rpages->p[index].page; + return &rpages->p[index].pfn; } -EXPORT_SYMBOL_GPL(lib_ring_buffer_read_get_page); +EXPORT_SYMBOL_GPL(lib_ring_buffer_read_get_pfn); /** * lib_ring_buffer_read_offset_address - get address of a buffer location @@ -699,8 +1066,9 @@ EXPORT_SYMBOL_GPL(lib_ring_buffer_read_get_page); * * Return the address where a given offset is located (for read). * Should be used to get the current subbuffer header pointer. Given we know - * it's never on a page boundary, it's safe to write directly to this address, - * as long as the write is never bigger than a page size. + * it's never on a page boundary, it's safe to read/write directly + * from/to this address, as long as the read/write is never bigger than a + * page size. */ void *lib_ring_buffer_read_offset_address(struct lib_ring_buffer_backend *bufb, size_t offset) @@ -708,7 +1076,7 @@ void *lib_ring_buffer_read_offset_address(struct lib_ring_buffer_backend *bufb, size_t index; struct lib_ring_buffer_backend_pages *rpages; struct channel_backend *chanb = &bufb->chan->backend; - const struct lib_ring_buffer_config *config = chanb->config; + const struct lib_ring_buffer_config *config = &chanb->config; unsigned long sb_bindex, id; offset &= chanb->buf_size - 1; @@ -738,7 +1106,7 @@ void *lib_ring_buffer_offset_address(struct lib_ring_buffer_backend *bufb, size_t sbidx, index; struct lib_ring_buffer_backend_pages *rpages; struct channel_backend *chanb = &bufb->chan->backend; - const struct lib_ring_buffer_config *config = chanb->config; + const struct lib_ring_buffer_config *config = &chanb->config; unsigned long sb_bindex, id; offset &= chanb->buf_size - 1;