X-Git-Url: https://gerrit.opnfv.org/gerrit/gitweb?a=blobdiff_plain;f=src%2Fceph%2Fsrc%2Fcrush%2Fmapper.c;fp=src%2Fceph%2Fsrc%2Fcrush%2Fmapper.c;h=e15039ed944b9b880647e73d8cafd38d9f6c4a46;hb=812ff6ca9fcd3e629e49d4328905f33eee8ca3f5;hp=0000000000000000000000000000000000000000;hpb=15280273faafb77777eab341909a3f495cf248d9;p=stor4nfv.git diff --git a/src/ceph/src/crush/mapper.c b/src/ceph/src/crush/mapper.c new file mode 100644 index 0000000..e15039e --- /dev/null +++ b/src/ceph/src/crush/mapper.c @@ -0,0 +1,1088 @@ +/* + * Ceph - scalable distributed file system + * + * Copyright (C) 2015 Intel Corporation All Rights Reserved + * + * This is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License version 2.1, as published by the Free Software + * Foundation. See file COPYING. + * + */ + +#ifdef __KERNEL__ +# include +# include +# include +# include +# include +# include +#else +# include "crush_compat.h" +# include "crush.h" +# include "hash.h" +#endif +#include "crush_ln_table.h" +#include "mapper.h" + +#define dprintk(args...) /* printf(args) */ + +/* + * Implement the core CRUSH mapping algorithm. + */ + +/** + * crush_find_rule - find a crush_rule id for a given ruleset, type, and size. + * @map: the crush_map + * @ruleset: the storage ruleset id (user defined) + * @type: storage ruleset type (user defined) + * @size: output set size + */ +int crush_find_rule(const struct crush_map *map, int ruleset, int type, int size) +{ + __u32 i; + + for (i = 0; i < map->max_rules; i++) { + if (map->rules[i] && + map->rules[i]->mask.ruleset == ruleset && + map->rules[i]->mask.type == type && + map->rules[i]->mask.min_size <= size && + map->rules[i]->mask.max_size >= size) + return i; + } + return -1; +} + +/* + * bucket choose methods + * + * For each bucket algorithm, we have a "choose" method that, given a + * crush input @x and replica position (usually, position in output set) @r, + * will produce an item in the bucket. + */ + +/* + * Choose based on a random permutation of the bucket. + * + * We used to use some prime number arithmetic to do this, but it + * wasn't very random, and had some other bad behaviors. Instead, we + * calculate an actual random permutation of the bucket members. + * Since this is expensive, we optimize for the r=0 case, which + * captures the vast majority of calls. + */ +static int bucket_perm_choose(const struct crush_bucket *bucket, + struct crush_work_bucket *work, + int x, int r) +{ + unsigned int pr = r % bucket->size; + unsigned int i, s; + + /* start a new permutation if @x has changed */ + if (work->perm_x != (__u32)x || work->perm_n == 0) { + dprintk("bucket %d new x=%d\n", bucket->id, x); + work->perm_x = x; + + /* optimize common r=0 case */ + if (pr == 0) { + s = crush_hash32_3(bucket->hash, x, bucket->id, 0) % + bucket->size; + work->perm[0] = s; + work->perm_n = 0xffff; /* magic value, see below */ + goto out; + } + + for (i = 0; i < bucket->size; i++) + work->perm[i] = i; + work->perm_n = 0; + } else if (work->perm_n == 0xffff) { + /* clean up after the r=0 case above */ + for (i = 1; i < bucket->size; i++) + work->perm[i] = i; + work->perm[work->perm[0]] = 0; + work->perm_n = 1; + } + + /* calculate permutation up to pr */ + for (i = 0; i < work->perm_n; i++) + dprintk(" perm_choose have %d: %d\n", i, work->perm[i]); + while (work->perm_n <= pr) { + unsigned int p = work->perm_n; + /* no point in swapping the final entry */ + if (p < bucket->size - 1) { + i = crush_hash32_3(bucket->hash, x, bucket->id, p) % + (bucket->size - p); + if (i) { + unsigned int t = work->perm[p + i]; + work->perm[p + i] = work->perm[p]; + work->perm[p] = t; + } + dprintk(" perm_choose swap %d with %d\n", p, p+i); + } + work->perm_n++; + } + for (i = 0; i < bucket->size; i++) + dprintk(" perm_choose %d: %d\n", i, work->perm[i]); + + s = work->perm[pr]; +out: + dprintk(" perm_choose %d sz=%d x=%d r=%d (%d) s=%d\n", bucket->id, + bucket->size, x, r, pr, s); + return bucket->items[s]; +} + +/* uniform */ +static int bucket_uniform_choose(const struct crush_bucket_uniform *bucket, + struct crush_work_bucket *work, int x, int r) +{ + return bucket_perm_choose(&bucket->h, work, x, r); +} + +/* list */ +static int bucket_list_choose(const struct crush_bucket_list *bucket, + int x, int r) +{ + int i; + + for (i = bucket->h.size-1; i >= 0; i--) { + __u64 w = crush_hash32_4(bucket->h.hash, x, bucket->h.items[i], + r, bucket->h.id); + w &= 0xffff; + dprintk("list_choose i=%d x=%d r=%d item %d weight %x " + "sw %x rand %llx", + i, x, r, bucket->h.items[i], bucket->item_weights[i], + bucket->sum_weights[i], w); + w *= bucket->sum_weights[i]; + w = w >> 16; + /*dprintk(" scaled %llx\n", w);*/ + if (w < bucket->item_weights[i]) { + return bucket->h.items[i]; + } + } + + dprintk("bad list sums for bucket %d\n", bucket->h.id); + return bucket->h.items[0]; +} + + +/* (binary) tree */ +static int height(int n) +{ + int h = 0; + while ((n & 1) == 0) { + h++; + n = n >> 1; + } + return h; +} + +static int left(int x) +{ + int h = height(x); + return x - (1 << (h-1)); +} + +static int right(int x) +{ + int h = height(x); + return x + (1 << (h-1)); +} + +static int terminal(int x) +{ + return x & 1; +} + +static int bucket_tree_choose(const struct crush_bucket_tree *bucket, + int x, int r) +{ + int n; + __u32 w; + __u64 t; + + /* start at root */ + n = bucket->num_nodes >> 1; + + while (!terminal(n)) { + int l; + /* pick point in [0, w) */ + w = bucket->node_weights[n]; + t = (__u64)crush_hash32_4(bucket->h.hash, x, n, r, + bucket->h.id) * (__u64)w; + t = t >> 32; + + /* descend to the left or right? */ + l = left(n); + if (t < bucket->node_weights[l]) + n = l; + else + n = right(n); + } + + return bucket->h.items[n >> 1]; +} + + +/* straw */ + +static int bucket_straw_choose(const struct crush_bucket_straw *bucket, + int x, int r) +{ + __u32 i; + int high = 0; + __u64 high_draw = 0; + __u64 draw; + + for (i = 0; i < bucket->h.size; i++) { + draw = crush_hash32_3(bucket->h.hash, x, bucket->h.items[i], r); + draw &= 0xffff; + draw *= bucket->straws[i]; + if (i == 0 || draw > high_draw) { + high = i; + high_draw = draw; + } + } + return bucket->h.items[high]; +} + +/* compute 2^44*log2(input+1) */ +static __u64 crush_ln(unsigned int xin) +{ + unsigned int x = xin; + int iexpon, index1, index2; + __u64 RH, LH, LL, xl64, result; + + x++; + + /* normalize input */ + iexpon = 15; + + // figure out number of bits we need to shift and + // do it in one step instead of iteratively + if (!(x & 0x18000)) { + int bits = __builtin_clz(x & 0x1FFFF) - 16; + x <<= bits; + iexpon = 15 - bits; + } + + index1 = (x >> 8) << 1; + /* RH ~ 2^56/index1 */ + RH = __RH_LH_tbl[index1 - 256]; + /* LH ~ 2^48 * log2(index1/256) */ + LH = __RH_LH_tbl[index1 + 1 - 256]; + + /* RH*x ~ 2^48 * (2^15 + xf), xf<2^8 */ + xl64 = (__s64)x * RH; + xl64 >>= 48; + + result = iexpon; + result <<= (12 + 32); + + index2 = xl64 & 0xff; + /* LL ~ 2^48*log2(1.0+index2/2^15) */ + LL = __LL_tbl[index2]; + + LH = LH + LL; + + LH >>= (48 - 12 - 32); + result += LH; + + return result; +} + + +/* + * straw2 + * + * for reference, see: + * + * http://en.wikipedia.org/wiki/Exponential_distribution#Distribution_of_the_minimum_of_exponential_random_variables + * + */ + +static inline __u32 *get_choose_arg_weights(const struct crush_bucket_straw2 *bucket, + const struct crush_choose_arg *arg, + int position) +{ + if ((arg == NULL) || + (arg->weight_set == NULL)) + return bucket->item_weights; + if (position >= arg->weight_set_size) + position = arg->weight_set_size - 1; + return arg->weight_set[position].weights; +} + +static inline __s32 *get_choose_arg_ids(const struct crush_bucket_straw2 *bucket, + const struct crush_choose_arg *arg) +{ + if ((arg == NULL) || (arg->ids == NULL)) + return bucket->h.items; + return arg->ids; +} + +static int bucket_straw2_choose(const struct crush_bucket_straw2 *bucket, + int x, int r, const struct crush_choose_arg *arg, + int position) +{ + unsigned int i, high = 0; + unsigned int u; + __s64 ln, draw, high_draw = 0; + __u32 *weights = get_choose_arg_weights(bucket, arg, position); + __s32 *ids = get_choose_arg_ids(bucket, arg); + for (i = 0; i < bucket->h.size; i++) { + dprintk("weight 0x%x item %d\n", weights[i], ids[i]); + if (weights[i]) { + u = crush_hash32_3(bucket->h.hash, x, ids[i], r); + u &= 0xffff; + + /* + * for some reason slightly less than 0x10000 produces + * a slightly more accurate distribution... probably a + * rounding effect. + * + * the natural log lookup table maps [0,0xffff] + * (corresponding to real numbers [1/0x10000, 1] to + * [0, 0xffffffffffff] (corresponding to real numbers + * [-11.090355,0]). + */ + ln = crush_ln(u) - 0x1000000000000ll; + + /* + * divide by 16.16 fixed-point weight. note + * that the ln value is negative, so a larger + * weight means a larger (less negative) value + * for draw. + */ + draw = div64_s64(ln, weights[i]); + } else { + draw = S64_MIN; + } + + if (i == 0 || draw > high_draw) { + high = i; + high_draw = draw; + } + } + + return bucket->h.items[high]; +} + + +static int crush_bucket_choose(const struct crush_bucket *in, + struct crush_work_bucket *work, + int x, int r, + const struct crush_choose_arg *arg, + int position) +{ + dprintk(" crush_bucket_choose %d x=%d r=%d\n", in->id, x, r); + BUG_ON(in->size == 0); + switch (in->alg) { + case CRUSH_BUCKET_UNIFORM: + return bucket_uniform_choose( + (const struct crush_bucket_uniform *)in, + work, x, r); + case CRUSH_BUCKET_LIST: + return bucket_list_choose((const struct crush_bucket_list *)in, + x, r); + case CRUSH_BUCKET_TREE: + return bucket_tree_choose((const struct crush_bucket_tree *)in, + x, r); + case CRUSH_BUCKET_STRAW: + return bucket_straw_choose( + (const struct crush_bucket_straw *)in, + x, r); + case CRUSH_BUCKET_STRAW2: + return bucket_straw2_choose( + (const struct crush_bucket_straw2 *)in, + x, r, arg, position); + default: + dprintk("unknown bucket %d alg %d\n", in->id, in->alg); + return in->items[0]; + } +} + +/* + * true if device is marked "out" (failed, fully offloaded) + * of the cluster + */ +static int is_out(const struct crush_map *map, + const __u32 *weight, int weight_max, + int item, int x) +{ + if (item >= weight_max) + return 1; + if (weight[item] >= 0x10000) + return 0; + if (weight[item] == 0) + return 1; + if ((crush_hash32_2(CRUSH_HASH_RJENKINS1, x, item) & 0xffff) + < weight[item]) + return 0; + return 1; +} + +/** + * crush_choose_firstn - choose numrep distinct items of given type + * @map: the crush_map + * @bucket: the bucket we are choose an item from + * @x: crush input value + * @numrep: the number of items to choose + * @type: the type of item to choose + * @out: pointer to output vector + * @outpos: our position in that vector + * @out_size: size of the out vector + * @tries: number of attempts to make + * @recurse_tries: number of attempts to have recursive chooseleaf make + * @local_retries: localized retries + * @local_fallback_retries: localized fallback retries + * @recurse_to_leaf: true if we want one device under each item of given type (chooseleaf instead of choose) + * @stable: stable mode starts rep=0 in the recursive call for all replicas + * @vary_r: pass r to recursive calls + * @out2: second output vector for leaf items (if @recurse_to_leaf) + * @parent_r: r value passed from the parent + */ +static int crush_choose_firstn(const struct crush_map *map, + struct crush_work *work, + const struct crush_bucket *bucket, + const __u32 *weight, int weight_max, + int x, int numrep, int type, + int *out, int outpos, + int out_size, + unsigned int tries, + unsigned int recurse_tries, + unsigned int local_retries, + unsigned int local_fallback_retries, + int recurse_to_leaf, + unsigned int vary_r, + unsigned int stable, + int *out2, + int parent_r, + const struct crush_choose_arg *choose_args) +{ + int rep; + unsigned int ftotal, flocal; + int retry_descent, retry_bucket, skip_rep; + const struct crush_bucket *in = bucket; + int r; + int i; + int item = 0; + int itemtype; + int collide, reject; + int count = out_size; + + dprintk("CHOOSE%s bucket %d x %d outpos %d numrep %d tries %d \ +recurse_tries %d local_retries %d local_fallback_retries %d \ +parent_r %d stable %d\n", + recurse_to_leaf ? "_LEAF" : "", + bucket->id, x, outpos, numrep, + tries, recurse_tries, local_retries, local_fallback_retries, + parent_r, stable); + + for (rep = stable ? 0 : outpos; rep < numrep && count > 0 ; rep++) { + /* keep trying until we get a non-out, non-colliding item */ + ftotal = 0; + skip_rep = 0; + do { + retry_descent = 0; + in = bucket; /* initial bucket */ + + /* choose through intervening buckets */ + flocal = 0; + do { + collide = 0; + retry_bucket = 0; + r = rep + parent_r; + /* r' = r + f_total */ + r += ftotal; + + /* bucket choose */ + if (in->size == 0) { + reject = 1; + goto reject; + } + if (local_fallback_retries > 0 && + flocal >= (in->size>>1) && + flocal > local_fallback_retries) + item = bucket_perm_choose( + in, work->work[-1-in->id], + x, r); + else + item = crush_bucket_choose( + in, work->work[-1-in->id], + x, r, + (choose_args ? &choose_args[-1-in->id] : 0), + outpos); + if (item >= map->max_devices) { + dprintk(" bad item %d\n", item); + skip_rep = 1; + break; + } + + /* desired type? */ + if (item < 0) + itemtype = map->buckets[-1-item]->type; + else + itemtype = 0; + dprintk(" item %d type %d\n", item, itemtype); + + /* keep going? */ + if (itemtype != type) { + if (item >= 0 || + (-1-item) >= map->max_buckets) { + dprintk(" bad item type %d\n", type); + skip_rep = 1; + break; + } + in = map->buckets[-1-item]; + retry_bucket = 1; + continue; + } + + /* collision? */ + for (i = 0; i < outpos; i++) { + if (out[i] == item) { + collide = 1; + break; + } + } + + reject = 0; + if (!collide && recurse_to_leaf) { + if (item < 0) { + int sub_r; + if (vary_r) + sub_r = r >> (vary_r-1); + else + sub_r = 0; + if (crush_choose_firstn( + map, + work, + map->buckets[-1-item], + weight, weight_max, + x, stable ? 1 : outpos+1, 0, + out2, outpos, count, + recurse_tries, 0, + local_retries, + local_fallback_retries, + 0, + vary_r, + stable, + NULL, + sub_r, + choose_args) <= outpos) + /* didn't get leaf */ + reject = 1; + } else { + /* we already have a leaf! */ + out2[outpos] = item; + } + } + + if (!reject && !collide) { + /* out? */ + if (itemtype == 0) + reject = is_out(map, weight, + weight_max, + item, x); + } + +reject: + if (reject || collide) { + ftotal++; + flocal++; + + if (collide && flocal <= local_retries) + /* retry locally a few times */ + retry_bucket = 1; + else if (local_fallback_retries > 0 && + flocal <= in->size + local_fallback_retries) + /* exhaustive bucket search */ + retry_bucket = 1; + else if (ftotal < tries) + /* then retry descent */ + retry_descent = 1; + else + /* else give up */ + skip_rep = 1; + dprintk(" reject %d collide %d " + "ftotal %u flocal %u\n", + reject, collide, ftotal, + flocal); + } + } while (retry_bucket); + } while (retry_descent); + + if (skip_rep) { + dprintk("skip rep\n"); + continue; + } + + dprintk("CHOOSE got %d\n", item); + out[outpos] = item; + outpos++; + count--; +#ifndef __KERNEL__ + if (map->choose_tries && ftotal <= map->choose_total_tries) + map->choose_tries[ftotal]++; +#endif + } + + dprintk("CHOOSE returns %d\n", outpos); + return outpos; +} + + +/** + * crush_choose_indep: alternative breadth-first positionally stable mapping + * + */ +static void crush_choose_indep(const struct crush_map *map, + struct crush_work *work, + const struct crush_bucket *bucket, + const __u32 *weight, int weight_max, + int x, int left, int numrep, int type, + int *out, int outpos, + unsigned int tries, + unsigned int recurse_tries, + int recurse_to_leaf, + int *out2, + int parent_r, + const struct crush_choose_arg *choose_args) +{ + const struct crush_bucket *in = bucket; + int endpos = outpos + left; + int rep; + unsigned int ftotal; + int r; + int i; + int item = 0; + int itemtype; + int collide; + + dprintk("CHOOSE%s INDEP bucket %d x %d outpos %d numrep %d\n", recurse_to_leaf ? "_LEAF" : "", + bucket->id, x, outpos, numrep); + + /* initially my result is undefined */ + for (rep = outpos; rep < endpos; rep++) { + out[rep] = CRUSH_ITEM_UNDEF; + if (out2) + out2[rep] = CRUSH_ITEM_UNDEF; + } + + for (ftotal = 0; left > 0 && ftotal < tries; ftotal++) { +#ifdef DEBUG_INDEP + if (out2 && ftotal) { + dprintk("%u %d a: ", ftotal, left); + for (rep = outpos; rep < endpos; rep++) { + dprintk(" %d", out[rep]); + } + dprintk("\n"); + dprintk("%u %d b: ", ftotal, left); + for (rep = outpos; rep < endpos; rep++) { + dprintk(" %d", out2[rep]); + } + dprintk("\n"); + } +#endif + for (rep = outpos; rep < endpos; rep++) { + if (out[rep] != CRUSH_ITEM_UNDEF) + continue; + + in = bucket; /* initial bucket */ + + /* choose through intervening buckets */ + for (;;) { + /* note: we base the choice on the position + * even in the nested call. that means that + * if the first layer chooses the same bucket + * in a different position, we will tend to + * choose a different item in that bucket. + * this will involve more devices in data + * movement and tend to distribute the load. + */ + r = rep + parent_r; + + /* be careful */ + if (in->alg == CRUSH_BUCKET_UNIFORM && + in->size % numrep == 0) + /* r'=r+(n+1)*f_total */ + r += (numrep+1) * ftotal; + else + /* r' = r + n*f_total */ + r += numrep * ftotal; + + /* bucket choose */ + if (in->size == 0) { + dprintk(" empty bucket\n"); + break; + } + + item = crush_bucket_choose( + in, work->work[-1-in->id], + x, r, + (choose_args ? &choose_args[-1-in->id] : 0), + outpos); + if (item >= map->max_devices) { + dprintk(" bad item %d\n", item); + out[rep] = CRUSH_ITEM_NONE; + if (out2) + out2[rep] = CRUSH_ITEM_NONE; + left--; + break; + } + + /* desired type? */ + if (item < 0) + itemtype = map->buckets[-1-item]->type; + else + itemtype = 0; + dprintk(" item %d type %d\n", item, itemtype); + + /* keep going? */ + if (itemtype != type) { + if (item >= 0 || + (-1-item) >= map->max_buckets) { + dprintk(" bad item type %d\n", type); + out[rep] = CRUSH_ITEM_NONE; + if (out2) + out2[rep] = + CRUSH_ITEM_NONE; + left--; + break; + } + in = map->buckets[-1-item]; + continue; + } + + /* collision? */ + collide = 0; + for (i = outpos; i < endpos; i++) { + if (out[i] == item) { + collide = 1; + break; + } + } + if (collide) + break; + + if (recurse_to_leaf) { + if (item < 0) { + crush_choose_indep( + map, + work, + map->buckets[-1-item], + weight, weight_max, + x, 1, numrep, 0, + out2, rep, + recurse_tries, 0, + 0, NULL, r, choose_args); + if (out2[rep] == CRUSH_ITEM_NONE) { + /* placed nothing; no leaf */ + break; + } + } else { + /* we already have a leaf! */ + out2[rep] = item; + } + } + + /* out? */ + if (itemtype == 0 && + is_out(map, weight, weight_max, item, x)) + break; + + /* yay! */ + out[rep] = item; + left--; + break; + } + } + } + for (rep = outpos; rep < endpos; rep++) { + if (out[rep] == CRUSH_ITEM_UNDEF) { + out[rep] = CRUSH_ITEM_NONE; + } + if (out2 && out2[rep] == CRUSH_ITEM_UNDEF) { + out2[rep] = CRUSH_ITEM_NONE; + } + } +#ifndef __KERNEL__ + if (map->choose_tries && ftotal <= map->choose_total_tries) + map->choose_tries[ftotal]++; +#endif +#ifdef DEBUG_INDEP + if (out2) { + dprintk("%u %d a: ", ftotal, left); + for (rep = outpos; rep < endpos; rep++) { + dprintk(" %d", out[rep]); + } + dprintk("\n"); + dprintk("%u %d b: ", ftotal, left); + for (rep = outpos; rep < endpos; rep++) { + dprintk(" %d", out2[rep]); + } + dprintk("\n"); + } +#endif +} + + +/* This takes a chunk of memory and sets it up to be a shiny new + working area for a CRUSH placement computation. It must be called + on any newly allocated memory before passing it in to + crush_do_rule. It may be used repeatedly after that, so long as the + map has not changed. If the map /has/ changed, you must make sure + the working size is no smaller than what was allocated and re-run + crush_init_workspace. + + If you do retain the working space between calls to crush, make it + thread-local. If you reinstitute the locking I've spent so much + time getting rid of, I will be very unhappy with you. */ + +void crush_init_workspace(const struct crush_map *m, void *v) { + /* We work by moving through the available space and setting + values and pointers as we go. + + It's a bit like Forth's use of the 'allot' word since we + set the pointer first and then reserve the space for it to + point to by incrementing the point. */ + struct crush_work *w = (struct crush_work *)v; + char *point = (char *)v; + __s32 b; + point += sizeof(struct crush_work); + w->work = (struct crush_work_bucket **)point; + point += m->max_buckets * sizeof(struct crush_work_bucket *); + for (b = 0; b < m->max_buckets; ++b) { + if (m->buckets[b] == 0) + continue; + + w->work[b] = (struct crush_work_bucket *) point; + switch (m->buckets[b]->alg) { + default: + point += sizeof(struct crush_work_bucket); + break; + } + w->work[b]->perm_x = 0; + w->work[b]->perm_n = 0; + w->work[b]->perm = (__u32 *)point; + point += m->buckets[b]->size * sizeof(__u32); + } + BUG_ON((char *)point - (char *)w != m->working_size); +} + +/** + * crush_do_rule - calculate a mapping with the given input and rule + * @map: the crush_map + * @ruleno: the rule id + * @x: hash input + * @result: pointer to result vector + * @result_max: maximum result size + * @weight: weight vector (for map leaves) + * @weight_max: size of weight vector + * @cwin: Pointer to at least map->working_size bytes of memory or NULL. + */ +int crush_do_rule(const struct crush_map *map, + int ruleno, int x, int *result, int result_max, + const __u32 *weight, int weight_max, + void *cwin, const struct crush_choose_arg *choose_args) +{ + int result_len; + struct crush_work *cw = cwin; + int *a = (int *)((char *)cw + map->working_size); + int *b = a + result_max; + int *c = b + result_max; + int *w = a; + int *o = b; + int recurse_to_leaf; + int wsize = 0; + int osize; + int *tmp; + const struct crush_rule *rule; + __u32 step; + int i, j; + int numrep; + int out_size; + /* + * the original choose_total_tries value was off by one (it + * counted "retries" and not "tries"). add one. + */ + int choose_tries = map->choose_total_tries + 1; + int choose_leaf_tries = 0; + /* + * the local tries values were counted as "retries", though, + * and need no adjustment + */ + int choose_local_retries = map->choose_local_tries; + int choose_local_fallback_retries = map->choose_local_fallback_tries; + + int vary_r = map->chooseleaf_vary_r; + int stable = map->chooseleaf_stable; + + if ((__u32)ruleno >= map->max_rules) { + dprintk(" bad ruleno %d\n", ruleno); + return 0; + } + + rule = map->rules[ruleno]; + result_len = 0; + + for (step = 0; step < rule->len; step++) { + int firstn = 0; + const struct crush_rule_step *curstep = &rule->steps[step]; + + switch (curstep->op) { + case CRUSH_RULE_TAKE: + if ((curstep->arg1 >= 0 && + curstep->arg1 < map->max_devices) || + (-1-curstep->arg1 >= 0 && + -1-curstep->arg1 < map->max_buckets && + map->buckets[-1-curstep->arg1])) { + w[0] = curstep->arg1; + wsize = 1; + } else { + dprintk(" bad take value %d\n", curstep->arg1); + } + break; + + case CRUSH_RULE_SET_CHOOSE_TRIES: + if (curstep->arg1 > 0) + choose_tries = curstep->arg1; + break; + + case CRUSH_RULE_SET_CHOOSELEAF_TRIES: + if (curstep->arg1 > 0) + choose_leaf_tries = curstep->arg1; + break; + + case CRUSH_RULE_SET_CHOOSE_LOCAL_TRIES: + if (curstep->arg1 >= 0) + choose_local_retries = curstep->arg1; + break; + + case CRUSH_RULE_SET_CHOOSE_LOCAL_FALLBACK_TRIES: + if (curstep->arg1 >= 0) + choose_local_fallback_retries = curstep->arg1; + break; + + case CRUSH_RULE_SET_CHOOSELEAF_VARY_R: + if (curstep->arg1 >= 0) + vary_r = curstep->arg1; + break; + + case CRUSH_RULE_SET_CHOOSELEAF_STABLE: + if (curstep->arg1 >= 0) + stable = curstep->arg1; + break; + + case CRUSH_RULE_CHOOSELEAF_FIRSTN: + case CRUSH_RULE_CHOOSE_FIRSTN: + firstn = 1; + /* fall through */ + case CRUSH_RULE_CHOOSELEAF_INDEP: + case CRUSH_RULE_CHOOSE_INDEP: + if (wsize == 0) + break; + + recurse_to_leaf = + curstep->op == + CRUSH_RULE_CHOOSELEAF_FIRSTN || + curstep->op == + CRUSH_RULE_CHOOSELEAF_INDEP; + + /* reset output */ + osize = 0; + + for (i = 0; i < wsize; i++) { + int bno; + numrep = curstep->arg1; + if (numrep <= 0) { + numrep += result_max; + if (numrep <= 0) + continue; + } + j = 0; + /* make sure bucket id is valid */ + bno = -1 - w[i]; + if (bno < 0 || bno >= map->max_buckets) { + // w[i] is probably CRUSH_ITEM_NONE + dprintk(" bad w[i] %d\n", w[i]); + continue; + } + if (firstn) { + int recurse_tries; + if (choose_leaf_tries) + recurse_tries = + choose_leaf_tries; + else if (map->chooseleaf_descend_once) + recurse_tries = 1; + else + recurse_tries = choose_tries; + osize += crush_choose_firstn( + map, + cw, + map->buckets[bno], + weight, weight_max, + x, numrep, + curstep->arg2, + o+osize, j, + result_max-osize, + choose_tries, + recurse_tries, + choose_local_retries, + choose_local_fallback_retries, + recurse_to_leaf, + vary_r, + stable, + c+osize, + 0, + choose_args); + } else { + out_size = ((numrep < (result_max-osize)) ? + numrep : (result_max-osize)); + crush_choose_indep( + map, + cw, + map->buckets[bno], + weight, weight_max, + x, out_size, numrep, + curstep->arg2, + o+osize, j, + choose_tries, + choose_leaf_tries ? + choose_leaf_tries : 1, + recurse_to_leaf, + c+osize, + 0, + choose_args); + osize += out_size; + } + } + + if (recurse_to_leaf) + /* copy final _leaf_ values to output set */ + memcpy(o, c, osize*sizeof(*o)); + + /* swap o and w arrays */ + tmp = o; + o = w; + w = tmp; + wsize = osize; + break; + + + case CRUSH_RULE_EMIT: + for (i = 0; i < wsize && result_len < result_max; i++) { + result[result_len] = w[i]; + result_len++; + } + wsize = 0; + break; + + default: + dprintk(" unknown op %d at step %d\n", + curstep->op, step); + break; + } + } + + return result_len; +}