Message ID | 20180312203957.2025833-2-songliubraving@fb.com |
---|---|
State | Changes Requested, archived |
Delegated to: | BPF Maintainers |
Headers | show |
Series | bpf stackmap with build_id+offset | expand |
On 3/12/18 1:39 PM, Song Liu wrote:
> + page = find_get_page(vma->vm_file->f_mapping, 0);
did you test it with config_debug_atomic_sleep ?
it should have complained...
> On Mar 12, 2018, at 2:00 PM, Alexei Starovoitov <ast@fb.com> wrote: > > On 3/12/18 1:39 PM, Song Liu wrote: >> + page = find_get_page(vma->vm_file->f_mapping, 0); > > did you test it with config_debug_atomic_sleep ? > it should have complained... Yeah, I have CONFIG_DEBUG_ATOMIC_SLEEP=y. I think find_get_page() will not sleep. The variation find_get_page_flags() may sleep with flag FGP_CREAT. Song
On 3/12/18 2:12 PM, Song Liu wrote: > >> On Mar 12, 2018, at 2:00 PM, Alexei Starovoitov <ast@fb.com> wrote: >> >> On 3/12/18 1:39 PM, Song Liu wrote: >>> + page = find_get_page(vma->vm_file->f_mapping, 0); >> >> did you test it with config_debug_atomic_sleep ? >> it should have complained... > > Yeah, I have CONFIG_DEBUG_ATOMIC_SLEEP=y. > > I think find_get_page() will not sleep. The variation find_get_page_flags() > may sleep with flag FGP_CREAT. I see. gfp_mask == 0 and no locks. should work indeed. curious how perf report looks like for heavy bpf_get_stackid() usage?
On Mon, Mar 12, 2018 at 01:39:56PM -0700, Song Liu wrote: > +static void stack_map_get_build_id_offset(struct bpf_map *map, > + struct stack_map_bucket *bucket, > + u64 *ips, u32 trace_nr) > +{ > + int i; > + struct vm_area_struct *vma; > + struct bpf_stack_build_id *id_offs; > + > + bucket->nr = trace_nr; > + id_offs = (struct bpf_stack_build_id *)bucket->data; > + > + if (!current || !current->mm || > + down_read_trylock(¤t->mm->mmap_sem) == 0) { You probably want an in_nmi() before the down_read_trylock(). Doing up_read() is an absolute no-no from NMI context. And IIUC its 'trivial' to use this stuff with hardware counters. > + /* cannot access current->mm, fall back to ips */ > + for (i = 0; i < trace_nr; i++) { > + id_offs[i].status = BPF_STACK_BUILD_ID_IP; > + id_offs[i].ip = ips[i]; > + } > + return; > + } > + > + for (i = 0; i < trace_nr; i++) { > + vma = find_vma(current->mm, ips[i]); > + if (!vma || stack_map_get_build_id(vma, id_offs[i].build_id)) { > + /* per entry fall back to ips */ > + id_offs[i].status = BPF_STACK_BUILD_ID_IP; > + id_offs[i].ip = ips[i]; > + continue; > + } > + id_offs[i].offset = (vma->vm_pgoff << PAGE_SHIFT) + ips[i] > + - vma->vm_start; > + id_offs[i].status = BPF_STACK_BUILD_ID_VALID; > + } > + up_read(¤t->mm->mmap_sem); > +}
> On Mar 12, 2018, at 2:31 PM, Alexei Starovoitov <ast@fb.com> wrote: > > On 3/12/18 2:12 PM, Song Liu wrote: >> >>> On Mar 12, 2018, at 2:00 PM, Alexei Starovoitov <ast@fb.com> wrote: >>> >>> On 3/12/18 1:39 PM, Song Liu wrote: >>>> + page = find_get_page(vma->vm_file->f_mapping, 0); >>> >>> did you test it with config_debug_atomic_sleep ? >>> it should have complained... >> >> Yeah, I have CONFIG_DEBUG_ATOMIC_SLEEP=y. >> >> I think find_get_page() will not sleep. The variation find_get_page_flags() >> may sleep with flag FGP_CREAT. > > I see. gfp_mask == 0 and no locks. should work indeed. > curious how perf report looks like for heavy bpf_get_stackid() usage? I modified samples/bpf/sampleip to only call bpf_get_stackid(). The following is captured with bpf_get_stackid() called at 10k Hz. stressapptest is running with 16 threads on a system with 56 cores. Samples: 1M of event 'cycles:pp', Event count (approx.): 628092326243 Overhead Command Shared Object Symbol + 51.61% stressapptest stressapptest [.] AdlerMemcpyC - 20.82% stressapptest [kernel.vmlinux] [k] queued_spin_lock_slowpath - queued_spin_lock_slowpath - 20.80% pcpu_freelist_pop bpf_get_stackid bpf_get_stackid_tp - 0x590c 16.12% AdlerMemcpyC 4.50% OsLayer::CpuStressWorkload + 14.36% stressapptest stressapptest [.] OsLayer::CpuStressWorkload - 8.74% stressapptest [kernel.vmlinux] [k] _raw_spin_lock - _raw_spin_lock - 8.73% bpf_get_stackid bpf_get_stackid_tp + 0x590c - 0.67% stressapptest [kernel.vmlinux] [k] pcpu_freelist_pop - pcpu_freelist_pop - 0.67% bpf_get_stackid bpf_get_stackid_tp + 0x590c Seems lock contention is the dominating overhead here. This should be the same for original stackmap. Song
Hi Song,
Thank you for the patch! Perhaps something to improve:
[auto build test WARNING on next-20180309]
[also build test WARNING on v4.16-rc5]
[cannot apply to linus/master v4.16-rc4 v4.16-rc3 v4.16-rc2]
[if your patch is applied to the wrong git tree, please drop us a note to help improve the system]
url: https://github.com/0day-ci/linux/commits/Song-Liu/bpf-stackmap-with-build_id-offset/20180313-085825
coccinelle warnings: (new ones prefixed by >>)
>> kernel/bpf/stackmap.c:177:2-3: Unneeded semicolon
Please review and possibly fold the followup patch.
---
0-DAY kernel test infrastructure Open Source Technology Center
https://lists.01.org/pipermail/kbuild-all Intel Corporation
> On Mar 12, 2018, at 3:47 PM, Song Liu <songliubraving@fb.com> wrote: > > > >> On Mar 12, 2018, at 2:31 PM, Alexei Starovoitov <ast@fb.com> wrote: >> >> On 3/12/18 2:12 PM, Song Liu wrote: >>> >>>> On Mar 12, 2018, at 2:00 PM, Alexei Starovoitov <ast@fb.com> wrote: >>>> >>>> On 3/12/18 1:39 PM, Song Liu wrote: >>>>> + page = find_get_page(vma->vm_file->f_mapping, 0); >>>> >>>> did you test it with config_debug_atomic_sleep ? >>>> it should have complained... >>> >>> Yeah, I have CONFIG_DEBUG_ATOMIC_SLEEP=y. >>> >>> I think find_get_page() will not sleep. The variation find_get_page_flags() >>> may sleep with flag FGP_CREAT. >> >> I see. gfp_mask == 0 and no locks. should work indeed. >> curious how perf report looks like for heavy bpf_get_stackid() usage? > > I modified samples/bpf/sampleip to only call bpf_get_stackid(). The following > is captured with bpf_get_stackid() called at 10k Hz. stressapptest is running > with 16 threads on a system with 56 cores. > > > Samples: 1M of event 'cycles:pp', Event count (approx.): 628092326243 > Overhead Command Shared Object Symbol > + 51.61% stressapptest stressapptest [.] AdlerMemcpyC > - 20.82% stressapptest [kernel.vmlinux] [k] queued_spin_lock_slowpath > - queued_spin_lock_slowpath > - 20.80% pcpu_freelist_pop > bpf_get_stackid > bpf_get_stackid_tp > - 0x590c > 16.12% AdlerMemcpyC > 4.50% OsLayer::CpuStressWorkload > + 14.36% stressapptest stressapptest [.] OsLayer::CpuStressWorkload > - 8.74% stressapptest [kernel.vmlinux] [k] _raw_spin_lock > - _raw_spin_lock > - 8.73% bpf_get_stackid > bpf_get_stackid_tp > + 0x590c > - 0.67% stressapptest [kernel.vmlinux] [k] pcpu_freelist_pop > - pcpu_freelist_pop > - 0.67% bpf_get_stackid > bpf_get_stackid_tp > + 0x590c > > Seems lock contention is the dominating overhead here. This should be the same > for original stackmap. > > Song Samples: 172K of event 'cycles:pp', Event count (approx.): 102311012653 Overhead Command Shared Object Symbol 78.84% stressapptest stressapptest [.] AdlerMemcpyC 8.78% stressapptest stressapptest [.] OsLayer::CpuStressWorkload 3.14% stressapptest [kernel.vmlinux] [k] _raw_spin_lock 2.56% stressapptest stressapptest [.] WorkerThread::FillPage 0.45% stressapptest [kernel.vmlinux] [k] perf_callchain_user 0.37% stressapptest [kernel.vmlinux] [.] native_irq_return_iret 0.31% stressapptest [kernel.vmlinux] [k] clear_page_erms 0.29% stressapptest [kernel.vmlinux] [k] pcpu_freelist_pop 0.27% stressapptest stressapptest [.] CalculateAdlerChecksum 0.25% stressapptest [kernel.vmlinux] [k] bpf_get_stackid 0.22% swapper [kernel.vmlinux] [k] poll_idle This perf output is taken with stressapptest running on 4 cores. bpf_get_stackid() and pcps_free_list_pop combined about 0.54% of CPU. Song
Hi, Peter, I have a question regarding to one of your comments below. On 3/12/18 3:01 PM, Peter Zijlstra wrote: > On Mon, Mar 12, 2018 at 01:39:56PM -0700, Song Liu wrote: >> +static void stack_map_get_build_id_offset(struct bpf_map *map, >> + struct stack_map_bucket *bucket, >> + u64 *ips, u32 trace_nr) >> +{ >> + int i; >> + struct vm_area_struct *vma; >> + struct bpf_stack_build_id *id_offs; >> + >> + bucket->nr = trace_nr; >> + id_offs = (struct bpf_stack_build_id *)bucket->data; >> + >> + if (!current || !current->mm || >> + down_read_trylock(¤t->mm->mmap_sem) == 0) { > > You probably want an in_nmi() before the down_read_trylock(). Doing > up_read() is an absolute no-no from NMI context. The below is the final code from Song: /* * We cannot do up_read() in nmi context, so build_id lookup is * only supported for non-nmi events. If at some point, it is * possible to run find_vma() without taking the semaphore, we * would like to allow build_id lookup in nmi context. * * Same fallback is used for kernel stack (!user) on a stackmap * with build_id. */ if (!user || !current || !current->mm || in_nmi() || down_read_trylock(¤t->mm->mmap_sem) == 0) { /* cannot access current->mm, fall back to ips */ for (i = 0; i < trace_nr; i++) { id_offs[i].status = BPF_STACK_BUILD_ID_IP; id_offs[i].ip = ips[i]; } return; } .... > > And IIUC its 'trivial' to use this stuff with hardware counters. Here, you mentioned that it was 'trivial' to use buildid thing with hardware counters, if I interpreted correctly. However, the hardware counter overflow will trigger NMI. Based on the above logic, it will default to old IP only behavior. Could you explain a little more how to get buildid with hardware counter overflow events? Thanks! > >> + /* cannot access current->mm, fall back to ips */ >> + for (i = 0; i < trace_nr; i++) { >> + id_offs[i].status = BPF_STACK_BUILD_ID_IP; >> + id_offs[i].ip = ips[i]; >> + } >> + return; >> + } >> + >> + for (i = 0; i < trace_nr; i++) { >> + vma = find_vma(current->mm, ips[i]); >> + if (!vma || stack_map_get_build_id(vma, id_offs[i].build_id)) { >> + /* per entry fall back to ips */ >> + id_offs[i].status = BPF_STACK_BUILD_ID_IP; >> + id_offs[i].ip = ips[i]; >> + continue; >> + } >> + id_offs[i].offset = (vma->vm_pgoff << PAGE_SHIFT) + ips[i] >> + - vma->vm_start; >> + id_offs[i].status = BPF_STACK_BUILD_ID_VALID; >> + } >> + up_read(¤t->mm->mmap_sem); >> +} >
diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h index 2a66769..1e15d17 100644 --- a/include/uapi/linux/bpf.h +++ b/include/uapi/linux/bpf.h @@ -231,6 +231,28 @@ enum bpf_attach_type { #define BPF_F_RDONLY (1U << 3) #define BPF_F_WRONLY (1U << 4) +/* Flag for stack_map, store build_id+offset instead of pointer */ +#define BPF_F_STACK_BUILD_ID (1U << 5) + +enum bpf_stack_build_id_status { + /* user space need an empty entry to identify end of a trace */ + BPF_STACK_BUILD_ID_EMPTY = 0, + /* with valid build_id and offset */ + BPF_STACK_BUILD_ID_VALID = 1, + /* couldn't get build_id, fallback to ip */ + BPF_STACK_BUILD_ID_IP = 2, +}; + +#define BPF_BUILD_ID_SIZE 20 +struct bpf_stack_build_id { + __s32 status; + unsigned char build_id[BPF_BUILD_ID_SIZE]; + union { + __u64 offset; + __u64 ip; + }; +}; + union bpf_attr { struct { /* anonymous struct used by BPF_MAP_CREATE command */ __u32 map_type; /* one of enum bpf_map_type */ diff --git a/kernel/bpf/stackmap.c b/kernel/bpf/stackmap.c index b0ecf43..ec055ea 100644 --- a/kernel/bpf/stackmap.c +++ b/kernel/bpf/stackmap.c @@ -9,16 +9,19 @@ #include <linux/filter.h> #include <linux/stacktrace.h> #include <linux/perf_event.h> +#include <linux/elf.h> +#include <linux/pagemap.h> #include "percpu_freelist.h" -#define STACK_CREATE_FLAG_MASK \ - (BPF_F_NUMA_NODE | BPF_F_RDONLY | BPF_F_WRONLY) +#define STACK_CREATE_FLAG_MASK \ + (BPF_F_NUMA_NODE | BPF_F_RDONLY | BPF_F_WRONLY | \ + BPF_F_STACK_BUILD_ID) struct stack_map_bucket { struct pcpu_freelist_node fnode; u32 hash; u32 nr; - u64 ip[]; + u64 data[]; }; struct bpf_stack_map { @@ -29,6 +32,17 @@ struct bpf_stack_map { struct stack_map_bucket *buckets[]; }; +static inline bool stack_map_use_build_id(struct bpf_map *map) +{ + return (map->map_flags & BPF_F_STACK_BUILD_ID); +} + +static inline int stack_map_data_size(struct bpf_map *map) +{ + return stack_map_use_build_id(map) ? + sizeof(struct bpf_stack_build_id) : sizeof(u64); +} + static int prealloc_elems_and_freelist(struct bpf_stack_map *smap) { u32 elem_size = sizeof(struct stack_map_bucket) + smap->map.value_size; @@ -68,8 +82,16 @@ static struct bpf_map *stack_map_alloc(union bpf_attr *attr) /* check sanity of attributes */ if (attr->max_entries == 0 || attr->key_size != 4 || - value_size < 8 || value_size % 8 || - value_size / 8 > sysctl_perf_event_max_stack) + value_size < 8 || value_size % 8) + return ERR_PTR(-EINVAL); + + BUILD_BUG_ON(sizeof(struct bpf_stack_build_id) % sizeof(u64)); + if (attr->map_flags & BPF_F_STACK_BUILD_ID) { + if (value_size % sizeof(struct bpf_stack_build_id) || + value_size / sizeof(struct bpf_stack_build_id) + > sysctl_perf_event_max_stack) + return ERR_PTR(-EINVAL); + } else if (value_size / 8 > sysctl_perf_event_max_stack) return ERR_PTR(-EINVAL); /* hash table size must be power of 2 */ @@ -114,13 +136,175 @@ static struct bpf_map *stack_map_alloc(union bpf_attr *attr) return ERR_PTR(err); } +#define BPF_BUILD_ID 3 +/* + * Parse build id from the note segment. This logic can be shared between + * 32-bit and 64-bit system, because Elf32_Nhdr and Elf64_Nhdr are + * identical. + */ +static inline int stack_map_parse_build_id(void *page_addr, + unsigned char *build_id, + void *note_start, + Elf32_Word note_size) +{ + Elf32_Word note_offs = 0, new_offs; + + /* check for overflow */ + if (note_start < page_addr || note_start + note_size < note_start) + return -EINVAL; + + /* only supports note that fits in the first page */ + if (note_start + note_size > page_addr + PAGE_SIZE) + return -EINVAL; + + while (note_offs + sizeof(Elf32_Nhdr) < note_size) { + Elf32_Nhdr *nhdr = (Elf32_Nhdr *)(note_start + note_offs); + + if (nhdr->n_type == BPF_BUILD_ID && + nhdr->n_namesz == sizeof("GNU") && + nhdr->n_descsz == BPF_BUILD_ID_SIZE) { + memcpy(build_id, + note_start + note_offs + + ALIGN(sizeof("GNU"), 4) + sizeof(Elf32_Nhdr), + BPF_BUILD_ID_SIZE); + return 0; + } + new_offs = note_offs + sizeof(Elf32_Nhdr) + + ALIGN(nhdr->n_namesz, 4) + ALIGN(nhdr->n_descsz, 4); + if (new_offs <= note_offs) /* overflow */ + break; + note_offs = new_offs; + }; + return -EINVAL; +} + +/* Parse build ID from 32-bit ELF */ +static int stack_map_get_build_id_32(void *page_addr, + unsigned char *build_id) +{ + Elf32_Ehdr *ehdr = (Elf32_Ehdr *)page_addr; + Elf32_Phdr *phdr; + int i; + + /* only supports phdr that fits in one page */ + if (ehdr->e_phnum > + (PAGE_SIZE - sizeof(Elf32_Ehdr)) / sizeof(Elf32_Phdr)) + return -EINVAL; + + phdr = (Elf32_Phdr *)(page_addr + sizeof(Elf32_Ehdr)); + + for (i = 0; i < ehdr->e_phnum; ++i) + if (phdr[i].p_type == PT_NOTE) + return stack_map_parse_build_id(page_addr, build_id, + page_addr + phdr[i].p_offset, + phdr[i].p_filesz); + return -EINVAL; +} + +/* Parse build ID from 64-bit ELF */ +static int stack_map_get_build_id_64(void *page_addr, + unsigned char *build_id) +{ + Elf64_Ehdr *ehdr = (Elf64_Ehdr *)page_addr; + Elf64_Phdr *phdr; + int i; + + /* only supports phdr that fits in one page */ + if (ehdr->e_phnum > + (PAGE_SIZE - sizeof(Elf64_Ehdr)) / sizeof(Elf64_Phdr)) + return -EINVAL; + + phdr = (Elf64_Phdr *)(page_addr + sizeof(Elf64_Ehdr)); + + for (i = 0; i < ehdr->e_phnum; ++i) + if (phdr[i].p_type == PT_NOTE) + return stack_map_parse_build_id(page_addr, build_id, + page_addr + phdr[i].p_offset, + phdr[i].p_filesz); + return -EINVAL; +} + +/* Parse build ID of ELF file mapped to vma */ +static int stack_map_get_build_id(struct vm_area_struct *vma, + unsigned char *build_id) +{ + Elf32_Ehdr *ehdr; + struct page *page; + void *page_addr; + int ret; + + /* only works for page backed storage */ + if (!vma->vm_file) + return -EINVAL; + + page = find_get_page(vma->vm_file->f_mapping, 0); + if (!page) + return -EFAULT; /* page not mapped */ + + ret = -EINVAL; + page_addr = page_address(page); + ehdr = (Elf32_Ehdr *)page_addr; + + /* compare magic x7f "ELF" */ + if (memcmp(ehdr->e_ident, ELFMAG, SELFMAG) != 0) + goto out; + + /* only support executable file and shared object file */ + if (ehdr->e_type != ET_EXEC && ehdr->e_type != ET_DYN) + goto out; + + if (ehdr->e_ident[EI_CLASS] == ELFCLASS32) + ret = stack_map_get_build_id_32(page_addr, build_id); + else if (ehdr->e_ident[EI_CLASS] == ELFCLASS64) + ret = stack_map_get_build_id_64(page_addr, build_id); +out: + put_page(page); + return ret; +} + +static void stack_map_get_build_id_offset(struct bpf_map *map, + struct stack_map_bucket *bucket, + u64 *ips, u32 trace_nr) +{ + int i; + struct vm_area_struct *vma; + struct bpf_stack_build_id *id_offs; + + bucket->nr = trace_nr; + id_offs = (struct bpf_stack_build_id *)bucket->data; + + if (!current || !current->mm || + down_read_trylock(¤t->mm->mmap_sem) == 0) { + /* cannot access current->mm, fall back to ips */ + for (i = 0; i < trace_nr; i++) { + id_offs[i].status = BPF_STACK_BUILD_ID_IP; + id_offs[i].ip = ips[i]; + } + return; + } + + for (i = 0; i < trace_nr; i++) { + vma = find_vma(current->mm, ips[i]); + if (!vma || stack_map_get_build_id(vma, id_offs[i].build_id)) { + /* per entry fall back to ips */ + id_offs[i].status = BPF_STACK_BUILD_ID_IP; + id_offs[i].ip = ips[i]; + continue; + } + id_offs[i].offset = (vma->vm_pgoff << PAGE_SHIFT) + ips[i] + - vma->vm_start; + id_offs[i].status = BPF_STACK_BUILD_ID_VALID; + } + up_read(¤t->mm->mmap_sem); +} + BPF_CALL_3(bpf_get_stackid, struct pt_regs *, regs, struct bpf_map *, map, u64, flags) { struct bpf_stack_map *smap = container_of(map, struct bpf_stack_map, map); struct perf_callchain_entry *trace; struct stack_map_bucket *bucket, *new_bucket, *old_bucket; - u32 max_depth = map->value_size / 8; + u32 max_depth = map->value_size / stack_map_data_size(map); /* stack_map_alloc() checks that max_depth <= sysctl_perf_event_max_stack */ u32 init_nr = sysctl_perf_event_max_stack - max_depth; u32 skip = flags & BPF_F_SKIP_FIELD_MASK; @@ -128,11 +312,16 @@ BPF_CALL_3(bpf_get_stackid, struct pt_regs *, regs, struct bpf_map *, map, bool user = flags & BPF_F_USER_STACK; bool kernel = !user; u64 *ips; + bool hash_matches; if (unlikely(flags & ~(BPF_F_SKIP_FIELD_MASK | BPF_F_USER_STACK | BPF_F_FAST_STACK_CMP | BPF_F_REUSE_STACKID))) return -EINVAL; + /* build_id+offset stack map only supports user stack */ + if (stack_map_use_build_id(map) && !user) + return -EINVAL; + trace = get_perf_callchain(regs, init_nr, kernel, user, sysctl_perf_event_max_stack, false, false); @@ -156,24 +345,42 @@ BPF_CALL_3(bpf_get_stackid, struct pt_regs *, regs, struct bpf_map *, map, id = hash & (smap->n_buckets - 1); bucket = READ_ONCE(smap->buckets[id]); - if (bucket && bucket->hash == hash) { - if (flags & BPF_F_FAST_STACK_CMP) + hash_matches = bucket && bucket->hash == hash; + /* fast cmp */ + if (hash_matches && flags & BPF_F_FAST_STACK_CMP) + return id; + + if (stack_map_use_build_id(map)) { + /* for build_id+offset, pop a bucket before slow cmp */ + new_bucket = (struct stack_map_bucket *) + pcpu_freelist_pop(&smap->freelist); + if (unlikely(!new_bucket)) + return -ENOMEM; + stack_map_get_build_id_offset(map, new_bucket, ips, trace_nr); + trace_len = trace_nr * sizeof(struct bpf_stack_build_id); + if (hash_matches && bucket->nr == trace_nr && + memcmp(bucket->data, new_bucket->data, trace_len) == 0) { + pcpu_freelist_push(&smap->freelist, &new_bucket->fnode); return id; - if (bucket->nr == trace_nr && - memcmp(bucket->ip, ips, trace_len) == 0) + } + if (bucket && !(flags & BPF_F_REUSE_STACKID)) { + pcpu_freelist_push(&smap->freelist, &new_bucket->fnode); + return -EEXIST; + } + } else { + if (hash_matches && bucket->nr == trace_nr && + memcmp(bucket->data, ips, trace_len) == 0) return id; + if (bucket && !(flags & BPF_F_REUSE_STACKID)) + return -EEXIST; + + new_bucket = (struct stack_map_bucket *) + pcpu_freelist_pop(&smap->freelist); + if (unlikely(!new_bucket)) + return -ENOMEM; + memcpy(new_bucket->data, ips, trace_len); } - /* this call stack is not in the map, try to add it */ - if (bucket && !(flags & BPF_F_REUSE_STACKID)) - return -EEXIST; - - new_bucket = (struct stack_map_bucket *) - pcpu_freelist_pop(&smap->freelist); - if (unlikely(!new_bucket)) - return -ENOMEM; - - memcpy(new_bucket->ip, ips, trace_len); new_bucket->hash = hash; new_bucket->nr = trace_nr; @@ -212,8 +419,8 @@ int bpf_stackmap_copy(struct bpf_map *map, void *key, void *value) if (!bucket) return -ENOENT; - trace_len = bucket->nr * sizeof(u64); - memcpy(value, bucket->ip, trace_len); + trace_len = bucket->nr * stack_map_data_size(map); + memcpy(value, bucket->data, trace_len); memset(value + trace_len, 0, map->value_size - trace_len); old_bucket = xchg(&smap->buckets[id], bucket);
Currently, bpf stackmap store address for each entry in the call trace. To map these addresses to user space files, it is necessary to maintain the mapping from these virtual address to symbols in the binary. Usually, the user space profiler (such as perf) has to scan /proc/pid/maps at the beginning of profiling, and monitor mmap2() calls afterwards. Given the cost of maintaining the address map, this solution is not practical for system wide profiling that is always on. This patch tries to solve this problem with a variation of stackmap. This variation is enabled by flag BPF_F_STACK_BUILD_ID, and only works for user stack. Instead of storing addresses, the variation stores ELF file build_id + offset. Build ID is a 20-byte unique identifier for ELF files. The following command shows the Build ID of /bin/bash: [user@]$ readelf -n /bin/bash ... Build ID: XXXXXXXXXX ... With BPF_F_STACK_BUILD_ID, bpf_get_stackid() tries to parse Build ID for each entry in the call trace, and translate it into the following struct: struct bpf_stack_build_id_offset { __s32 status; unsigned char build_id[BPF_BUILD_ID_SIZE]; union { __u64 offset; __u64 ip; }; }; The search of build_id is limited to the first page of the file, and this page should be in page cache. Otherwise, we fallback to store ip for this entry (ip field in struct bpf_stack_build_id_offset). This requires the build_id to be stored in the first page. A quick survey of binary and dynamic library files in a few different systems shows that almost all binary and dynamic library files have build_id in the first page. User space can access struct bpf_stack_build_id_offset with bpf syscall BPF_MAP_LOOKUP_ELEM. It is necessary for user space to maintain mapping from build id to binary files. This mostly static mapping is much easier to maintain than per process address maps. Signed-off-by: Song Liu <songliubraving@fb.com> --- include/uapi/linux/bpf.h | 22 +++++ kernel/bpf/stackmap.c | 251 ++++++++++++++++++++++++++++++++++++++++++----- 2 files changed, 251 insertions(+), 22 deletions(-)