1 // SPDX-License-Identifier: GPL-2.0 2 #include <linux/kernel.h> 3 #include <linux/errno.h> 4 #include <linux/file.h> 5 #include <linux/mm.h> 6 #include <linux/slab.h> 7 #include <linux/nospec.h> 8 #include <linux/io_uring.h> 9 10 #include <uapi/linux/io_uring.h> 11 12 #include "io_uring.h" 13 #include "tctx.h" 14 15 static struct io_wq *io_init_wq_offload(struct io_ring_ctx *ctx, 16 struct task_struct *task) 17 { 18 struct io_wq_hash *hash; 19 struct io_wq_data data; 20 unsigned int concurrency; 21 22 mutex_lock(&ctx->uring_lock); 23 hash = ctx->hash_map; 24 if (!hash) { 25 hash = kzalloc(sizeof(*hash), GFP_KERNEL); 26 if (!hash) { 27 mutex_unlock(&ctx->uring_lock); 28 return ERR_PTR(-ENOMEM); 29 } 30 refcount_set(&hash->refs, 1); 31 init_waitqueue_head(&hash->wait); 32 ctx->hash_map = hash; 33 } 34 mutex_unlock(&ctx->uring_lock); 35 36 data.hash = hash; 37 data.task = task; 38 data.free_work = io_wq_free_work; 39 data.do_work = io_wq_submit_work; 40 41 /* Do QD, or 4 * CPUS, whatever is smallest */ 42 concurrency = min(ctx->sq_entries, 4 * num_online_cpus()); 43 44 return io_wq_create(concurrency, &data); 45 } 46 47 void __io_uring_free(struct task_struct *tsk) 48 { 49 struct io_uring_task *tctx = tsk->io_uring; 50 51 WARN_ON_ONCE(!xa_empty(&tctx->xa)); 52 WARN_ON_ONCE(tctx->io_wq); 53 WARN_ON_ONCE(tctx->cached_refs); 54 55 percpu_counter_destroy(&tctx->inflight); 56 kfree(tctx); 57 tsk->io_uring = NULL; 58 } 59 60 __cold int io_uring_alloc_task_context(struct task_struct *task, 61 struct io_ring_ctx *ctx) 62 { 63 struct io_uring_task *tctx; 64 int ret; 65 66 tctx = kzalloc(sizeof(*tctx), GFP_KERNEL); 67 if (unlikely(!tctx)) 68 return -ENOMEM; 69 70 ret = percpu_counter_init(&tctx->inflight, 0, GFP_KERNEL); 71 if (unlikely(ret)) { 72 kfree(tctx); 73 return ret; 74 } 75 76 tctx->io_wq = io_init_wq_offload(ctx, task); 77 if (IS_ERR(tctx->io_wq)) { 78 ret = PTR_ERR(tctx->io_wq); 79 percpu_counter_destroy(&tctx->inflight); 80 kfree(tctx); 81 return ret; 82 } 83 84 xa_init(&tctx->xa); 85 init_waitqueue_head(&tctx->wait); 86 atomic_set(&tctx->in_idle, 0); 87 atomic_set(&tctx->inflight_tracked, 0); 88 task->io_uring = tctx; 89 init_llist_head(&tctx->task_list); 90 init_task_work(&tctx->task_work, tctx_task_work); 91 return 0; 92 } 93 94 static int io_register_submitter(struct io_ring_ctx *ctx) 95 { 96 int ret = 0; 97 98 mutex_lock(&ctx->uring_lock); 99 if (!ctx->submitter_task) 100 ctx->submitter_task = get_task_struct(current); 101 else if (ctx->submitter_task != current) 102 ret = -EEXIST; 103 mutex_unlock(&ctx->uring_lock); 104 105 return ret; 106 } 107 108 int __io_uring_add_tctx_node(struct io_ring_ctx *ctx, bool submitter) 109 { 110 struct io_uring_task *tctx = current->io_uring; 111 struct io_tctx_node *node; 112 int ret; 113 114 if ((ctx->flags & IORING_SETUP_SINGLE_ISSUER) && submitter) { 115 ret = io_register_submitter(ctx); 116 if (ret) 117 return ret; 118 } 119 120 if (unlikely(!tctx)) { 121 ret = io_uring_alloc_task_context(current, ctx); 122 if (unlikely(ret)) 123 return ret; 124 125 tctx = current->io_uring; 126 if (ctx->iowq_limits_set) { 127 unsigned int limits[2] = { ctx->iowq_limits[0], 128 ctx->iowq_limits[1], }; 129 130 ret = io_wq_max_workers(tctx->io_wq, limits); 131 if (ret) 132 return ret; 133 } 134 } 135 if (!xa_load(&tctx->xa, (unsigned long)ctx)) { 136 node = kmalloc(sizeof(*node), GFP_KERNEL); 137 if (!node) 138 return -ENOMEM; 139 node->ctx = ctx; 140 node->task = current; 141 142 ret = xa_err(xa_store(&tctx->xa, (unsigned long)ctx, 143 node, GFP_KERNEL)); 144 if (ret) { 145 kfree(node); 146 return ret; 147 } 148 149 mutex_lock(&ctx->uring_lock); 150 list_add(&node->ctx_node, &ctx->tctx_list); 151 mutex_unlock(&ctx->uring_lock); 152 } 153 if (submitter) 154 tctx->last = ctx; 155 return 0; 156 } 157 158 /* 159 * Remove this io_uring_file -> task mapping. 160 */ 161 __cold void io_uring_del_tctx_node(unsigned long index) 162 { 163 struct io_uring_task *tctx = current->io_uring; 164 struct io_tctx_node *node; 165 166 if (!tctx) 167 return; 168 node = xa_erase(&tctx->xa, index); 169 if (!node) 170 return; 171 172 WARN_ON_ONCE(current != node->task); 173 WARN_ON_ONCE(list_empty(&node->ctx_node)); 174 175 mutex_lock(&node->ctx->uring_lock); 176 list_del(&node->ctx_node); 177 mutex_unlock(&node->ctx->uring_lock); 178 179 if (tctx->last == node->ctx) 180 tctx->last = NULL; 181 kfree(node); 182 } 183 184 __cold void io_uring_clean_tctx(struct io_uring_task *tctx) 185 { 186 struct io_wq *wq = tctx->io_wq; 187 struct io_tctx_node *node; 188 unsigned long index; 189 190 xa_for_each(&tctx->xa, index, node) { 191 io_uring_del_tctx_node(index); 192 cond_resched(); 193 } 194 if (wq) { 195 /* 196 * Must be after io_uring_del_tctx_node() (removes nodes under 197 * uring_lock) to avoid race with io_uring_try_cancel_iowq(). 198 */ 199 io_wq_put_and_exit(wq); 200 tctx->io_wq = NULL; 201 } 202 } 203 204 void io_uring_unreg_ringfd(void) 205 { 206 struct io_uring_task *tctx = current->io_uring; 207 int i; 208 209 for (i = 0; i < IO_RINGFD_REG_MAX; i++) { 210 if (tctx->registered_rings[i]) { 211 fput(tctx->registered_rings[i]); 212 tctx->registered_rings[i] = NULL; 213 } 214 } 215 } 216 217 static int io_ring_add_registered_fd(struct io_uring_task *tctx, int fd, 218 int start, int end) 219 { 220 struct file *file; 221 int offset; 222 223 for (offset = start; offset < end; offset++) { 224 offset = array_index_nospec(offset, IO_RINGFD_REG_MAX); 225 if (tctx->registered_rings[offset]) 226 continue; 227 228 file = fget(fd); 229 if (!file) { 230 return -EBADF; 231 } else if (!io_is_uring_fops(file)) { 232 fput(file); 233 return -EOPNOTSUPP; 234 } 235 tctx->registered_rings[offset] = file; 236 return offset; 237 } 238 239 return -EBUSY; 240 } 241 242 /* 243 * Register a ring fd to avoid fdget/fdput for each io_uring_enter() 244 * invocation. User passes in an array of struct io_uring_rsrc_update 245 * with ->data set to the ring_fd, and ->offset given for the desired 246 * index. If no index is desired, application may set ->offset == -1U 247 * and we'll find an available index. Returns number of entries 248 * successfully processed, or < 0 on error if none were processed. 249 */ 250 int io_ringfd_register(struct io_ring_ctx *ctx, void __user *__arg, 251 unsigned nr_args) 252 { 253 struct io_uring_rsrc_update __user *arg = __arg; 254 struct io_uring_rsrc_update reg; 255 struct io_uring_task *tctx; 256 int ret, i; 257 258 if (!nr_args || nr_args > IO_RINGFD_REG_MAX) 259 return -EINVAL; 260 261 mutex_unlock(&ctx->uring_lock); 262 ret = __io_uring_add_tctx_node(ctx, false); 263 mutex_lock(&ctx->uring_lock); 264 if (ret) 265 return ret; 266 267 tctx = current->io_uring; 268 for (i = 0; i < nr_args; i++) { 269 int start, end; 270 271 if (copy_from_user(®, &arg[i], sizeof(reg))) { 272 ret = -EFAULT; 273 break; 274 } 275 276 if (reg.resv) { 277 ret = -EINVAL; 278 break; 279 } 280 281 if (reg.offset == -1U) { 282 start = 0; 283 end = IO_RINGFD_REG_MAX; 284 } else { 285 if (reg.offset >= IO_RINGFD_REG_MAX) { 286 ret = -EINVAL; 287 break; 288 } 289 start = reg.offset; 290 end = start + 1; 291 } 292 293 ret = io_ring_add_registered_fd(tctx, reg.data, start, end); 294 if (ret < 0) 295 break; 296 297 reg.offset = ret; 298 if (copy_to_user(&arg[i], ®, sizeof(reg))) { 299 fput(tctx->registered_rings[reg.offset]); 300 tctx->registered_rings[reg.offset] = NULL; 301 ret = -EFAULT; 302 break; 303 } 304 } 305 306 return i ? i : ret; 307 } 308 309 int io_ringfd_unregister(struct io_ring_ctx *ctx, void __user *__arg, 310 unsigned nr_args) 311 { 312 struct io_uring_rsrc_update __user *arg = __arg; 313 struct io_uring_task *tctx = current->io_uring; 314 struct io_uring_rsrc_update reg; 315 int ret = 0, i; 316 317 if (!nr_args || nr_args > IO_RINGFD_REG_MAX) 318 return -EINVAL; 319 if (!tctx) 320 return 0; 321 322 for (i = 0; i < nr_args; i++) { 323 if (copy_from_user(®, &arg[i], sizeof(reg))) { 324 ret = -EFAULT; 325 break; 326 } 327 if (reg.resv || reg.data || reg.offset >= IO_RINGFD_REG_MAX) { 328 ret = -EINVAL; 329 break; 330 } 331 332 reg.offset = array_index_nospec(reg.offset, IO_RINGFD_REG_MAX); 333 if (tctx->registered_rings[reg.offset]) { 334 fput(tctx->registered_rings[reg.offset]); 335 tctx->registered_rings[reg.offset] = NULL; 336 } 337 } 338 339 return i ? i : ret; 340 } 341