[thirdparty/kernel/linux.git] / io_uring / fdinfo.c

// SPDX-License-Identifier: GPL-2.0
#include <linux/kernel.h>
#include <linux/errno.h>
#include <linux/fs.h>
#include <linux/file.h>
#include <linux/proc_fs.h>
#include <linux/seq_file.h>
#include <linux/io_uring.h>

#include <uapi/linux/io_uring.h>

#include "io_uring.h"
#include "sqpoll.h"
#include "fdinfo.h"
#include "cancel.h"
#include "rsrc.h"

#ifdef CONFIG_PROC_FS
static __cold int io_uring_show_cred(struct seq_file *m, unsigned int id,
		const struct cred *cred)
{
	struct user_namespace *uns = seq_user_ns(m);
	struct group_info *gi;
	kernel_cap_t cap;
	int g;

	seq_printf(m, "%5d\n", id);
	seq_put_decimal_ull(m, "\tUid:\t", from_kuid_munged(uns, cred->uid));
	seq_put_decimal_ull(m, "\t\t", from_kuid_munged(uns, cred->euid));
	seq_put_decimal_ull(m, "\t\t", from_kuid_munged(uns, cred->suid));
	seq_put_decimal_ull(m, "\t\t", from_kuid_munged(uns, cred->fsuid));
	seq_put_decimal_ull(m, "\n\tGid:\t", from_kgid_munged(uns, cred->gid));
	seq_put_decimal_ull(m, "\t\t", from_kgid_munged(uns, cred->egid));
	seq_put_decimal_ull(m, "\t\t", from_kgid_munged(uns, cred->sgid));
	seq_put_decimal_ull(m, "\t\t", from_kgid_munged(uns, cred->fsgid));
	seq_puts(m, "\n\tGroups:\t");
	gi = cred->group_info;
	for (g = 0; g < gi->ngroups; g++) {
		seq_put_decimal_ull(m, g ? " " : "",
					from_kgid_munged(uns, gi->gid[g]));
	}
	seq_puts(m, "\n\tCapEff:\t");
	cap = cred->cap_effective;
	seq_put_hex_ll(m, NULL, cap.val, 16);
	seq_putc(m, '\n');
	return 0;
}

/*
 * Caller holds a reference to the file already, we don't need to do
 * anything else to get an extra reference.
 */
__cold void io_uring_show_fdinfo(struct seq_file *m, struct file *f)
{
	struct io_ring_ctx *ctx = f->private_data;
	struct io_overflow_cqe *ocqe;
	struct io_rings *r = ctx->rings;
	struct rusage sq_usage;
	unsigned int sq_mask = ctx->sq_entries - 1, cq_mask = ctx->cq_entries - 1;
	unsigned int sq_head = READ_ONCE(r->sq.head);
	unsigned int sq_tail = READ_ONCE(r->sq.tail);
	unsigned int cq_head = READ_ONCE(r->cq.head);
	unsigned int cq_tail = READ_ONCE(r->cq.tail);
	unsigned int cq_shift = 0;
	unsigned int sq_shift = 0;
	unsigned int sq_entries, cq_entries;
	int sq_pid = -1, sq_cpu = -1;
	u64 sq_total_time = 0, sq_work_time = 0;
	bool has_lock;
	unsigned int i;

	if (ctx->flags & IORING_SETUP_CQE32)
		cq_shift = 1;
	if (ctx->flags & IORING_SETUP_SQE128)
		sq_shift = 1;

	/*
	 * we may get imprecise sqe and cqe info if uring is actively running
	 * since we get cached_sq_head and cached_cq_tail without uring_lock
	 * and sq_tail and cq_head are changed by userspace. But it's ok since
	 * we usually use these info when it is stuck.
	 */
	seq_printf(m, "SqMask:\t0x%x\n", sq_mask);
	seq_printf(m, "SqHead:\t%u\n", sq_head);
	seq_printf(m, "SqTail:\t%u\n", sq_tail);
	seq_printf(m, "CachedSqHead:\t%u\n", ctx->cached_sq_head);
	seq_printf(m, "CqMask:\t0x%x\n", cq_mask);
	seq_printf(m, "CqHead:\t%u\n", cq_head);
	seq_printf(m, "CqTail:\t%u\n", cq_tail);
	seq_printf(m, "CachedCqTail:\t%u\n", ctx->cached_cq_tail);
	seq_printf(m, "SQEs:\t%u\n", sq_tail - sq_head);
	sq_entries = min(sq_tail - sq_head, ctx->sq_entries);
	for (i = 0; i < sq_entries; i++) {
		unsigned int entry = i + sq_head;
		struct io_uring_sqe *sqe;
		unsigned int sq_idx;

		if (ctx->flags & IORING_SETUP_NO_SQARRAY)
			break;
		sq_idx = READ_ONCE(ctx->sq_array[entry & sq_mask]);
		if (sq_idx > sq_mask)
			continue;
		sqe = &ctx->sq_sqes[sq_idx << sq_shift];
		seq_printf(m, "%5u: opcode:%s, fd:%d, flags:%x, off:%llu, "
			      "addr:0x%llx, rw_flags:0x%x, buf_index:%d "
			      "user_data:%llu",
			   sq_idx, io_uring_get_opcode(sqe->opcode), sqe->fd,
			   sqe->flags, (unsigned long long) sqe->off,
			   (unsigned long long) sqe->addr, sqe->rw_flags,
			   sqe->buf_index, sqe->user_data);
		if (sq_shift) {
			u64 *sqeb = (void *) (sqe + 1);
			int size = sizeof(struct io_uring_sqe) / sizeof(u64);
			int j;

			for (j = 0; j < size; j++) {
				seq_printf(m, ", e%d:0x%llx", j,
						(unsigned long long) *sqeb);
				sqeb++;
			}
		}
		seq_printf(m, "\n");
	}
	seq_printf(m, "CQEs:\t%u\n", cq_tail - cq_head);
	cq_entries = min(cq_tail - cq_head, ctx->cq_entries);
	for (i = 0; i < cq_entries; i++) {
		unsigned int entry = i + cq_head;
		struct io_uring_cqe *cqe = &r->cqes[(entry & cq_mask) << cq_shift];

		seq_printf(m, "%5u: user_data:%llu, res:%d, flag:%x",
			   entry & cq_mask, cqe->user_data, cqe->res,
			   cqe->flags);
		if (cq_shift)
			seq_printf(m, ", extra1:%llu, extra2:%llu\n",
					cqe->big_cqe[0], cqe->big_cqe[1]);
		seq_printf(m, "\n");
	}

	/*
	 * Avoid ABBA deadlock between the seq lock and the io_uring mutex,
	 * since fdinfo case grabs it in the opposite direction of normal use
	 * cases. If we fail to get the lock, we just don't iterate any
	 * structures that could be going away outside the io_uring mutex.
	 */
	has_lock = mutex_trylock(&ctx->uring_lock);

	if (has_lock && (ctx->flags & IORING_SETUP_SQPOLL)) {
		struct io_sq_data *sq = ctx->sq_data;

		/*
		 * sq->thread might be NULL if we raced with the sqpoll
		 * thread termination.
		 */
		if (sq->thread) {
			sq_pid = sq->task_pid;
			sq_cpu = sq->sq_cpu;
			getrusage(sq->thread, RUSAGE_SELF, &sq_usage);
			sq_total_time = (sq_usage.ru_stime.tv_sec * 1000000
					 + sq_usage.ru_stime.tv_usec);
			sq_work_time = sq->work_time;
		}
	}

	seq_printf(m, "SqThread:\t%d\n", sq_pid);
	seq_printf(m, "SqThreadCpu:\t%d\n", sq_cpu);
	seq_printf(m, "SqTotalTime:\t%llu\n", sq_total_time);
	seq_printf(m, "SqWorkTime:\t%llu\n", sq_work_time);
	seq_printf(m, "UserFiles:\t%u\n", ctx->nr_user_files);
	for (i = 0; has_lock && i < ctx->nr_user_files; i++) {
		struct file *f = io_file_from_index(&ctx->file_table, i);

		if (f)
			seq_printf(m, "%5u: %s\n", i, file_dentry(f)->d_iname);
		else
			seq_printf(m, "%5u: <none>\n", i);
	}
	seq_printf(m, "UserBufs:\t%u\n", ctx->nr_user_bufs);
	for (i = 0; has_lock && i < ctx->nr_user_bufs; i++) {
		struct io_mapped_ubuf *buf = ctx->user_bufs[i];
		unsigned int len = buf->ubuf_end - buf->ubuf;

		seq_printf(m, "%5u: 0x%llx/%u\n", i, buf->ubuf, len);
	}
	if (has_lock && !xa_empty(&ctx->personalities)) {
		unsigned long index;
		const struct cred *cred;

		seq_printf(m, "Personalities:\n");
		xa_for_each(&ctx->personalities, index, cred)
			io_uring_show_cred(m, index, cred);
	}

	seq_puts(m, "PollList:\n");
	for (i = 0; i < (1U << ctx->cancel_table.hash_bits); i++) {
		struct io_hash_bucket *hb = &ctx->cancel_table.hbs[i];
		struct io_hash_bucket *hbl = &ctx->cancel_table_locked.hbs[i];
		struct io_kiocb *req;

		spin_lock(&hb->lock);
		hlist_for_each_entry(req, &hb->list, hash_node)
			seq_printf(m, "  op=%d, task_works=%d\n", req->opcode,
					task_work_pending(req->task));
		spin_unlock(&hb->lock);

		if (!has_lock)
			continue;
		hlist_for_each_entry(req, &hbl->list, hash_node)
			seq_printf(m, "  op=%d, task_works=%d\n", req->opcode,
					task_work_pending(req->task));
	}

	if (has_lock)
		mutex_unlock(&ctx->uring_lock);

	seq_puts(m, "CqOverflowList:\n");
	spin_lock(&ctx->completion_lock);
	list_for_each_entry(ocqe, &ctx->cq_overflow_list, list) {
		struct io_uring_cqe *cqe = &ocqe->cqe;

		seq_printf(m, "  user_data=%llu, res=%d, flags=%x\n",
			   cqe->user_data, cqe->res, cqe->flags);

	}

	spin_unlock(&ctx->completion_lock);
}
#endif
Commit	Line	Data
a4ad4f74 JA	1	// SPDX-License-Identifier: GPL-2.0
	2	#include <linux/kernel.h>
	3	#include <linux/errno.h>
	4	#include <linux/fs.h>
	5	#include <linux/file.h>
	6	#include <linux/proc_fs.h>
	7	#include <linux/seq_file.h>
	8	#include <linux/io_uring.h>
	9
	10	#include <uapi/linux/io_uring.h>
	11
a4ad4f74 JA	12	#include "io_uring.h"
	13	#include "sqpoll.h"
	14	#include "fdinfo.h"
38513c46	15	#include "cancel.h"
ad163a7e	16	#include "rsrc.h"
a4ad4f74 JA	17
	18	#ifdef CONFIG_PROC_FS
	19	static __cold int io_uring_show_cred(struct seq_file *m, unsigned int id,
	20	const struct cred *cred)
	21	{
	22	struct user_namespace *uns = seq_user_ns(m);
	23	struct group_info *gi;
	24	kernel_cap_t cap;
a4ad4f74 JA	25	int g;
	26
	27	seq_printf(m, "%5d\n", id);
	28	seq_put_decimal_ull(m, "\tUid:\t", from_kuid_munged(uns, cred->uid));
	29	seq_put_decimal_ull(m, "\t\t", from_kuid_munged(uns, cred->euid));
	30	seq_put_decimal_ull(m, "\t\t", from_kuid_munged(uns, cred->suid));
	31	seq_put_decimal_ull(m, "\t\t", from_kuid_munged(uns, cred->fsuid));
	32	seq_put_decimal_ull(m, "\n\tGid:\t", from_kgid_munged(uns, cred->gid));
	33	seq_put_decimal_ull(m, "\t\t", from_kgid_munged(uns, cred->egid));
	34	seq_put_decimal_ull(m, "\t\t", from_kgid_munged(uns, cred->sgid));
	35	seq_put_decimal_ull(m, "\t\t", from_kgid_munged(uns, cred->fsgid));
	36	seq_puts(m, "\n\tGroups:\t");
	37	gi = cred->group_info;
	38	for (g = 0; g < gi->ngroups; g++) {
	39	seq_put_decimal_ull(m, g ? " " : "",
	40	from_kgid_munged(uns, gi->gid[g]));
	41	}
	42	seq_puts(m, "\n\tCapEff:\t");
	43	cap = cred->cap_effective;
f122a08b	44	seq_put_hex_ll(m, NULL, cap.val, 16);
a4ad4f74 JA	45	seq_putc(m, '\n');
	46	return 0;
	47	}
	48
3aaf22b6 JA	49	/*
	50	* Caller holds a reference to the file already, we don't need to do
	51	* anything else to get an extra reference.
	52	*/
	53	__cold void io_uring_show_fdinfo(struct seq_file m, struct file f)
a4ad4f74	54	{
3aaf22b6	55	struct io_ring_ctx *ctx = f->private_data;
a4ad4f74 JA	56	struct io_overflow_cqe *ocqe;
a4ad4f74 JA	57	struct io_rings *r = ctx->rings;
3fcb9d17	58	struct rusage sq_usage;
a4ad4f74 JA	59	unsigned int sq_mask = ctx->sq_entries - 1, cq_mask = ctx->cq_entries - 1;
	60	unsigned int sq_head = READ_ONCE(r->sq.head);
	61	unsigned int sq_tail = READ_ONCE(r->sq.tail);
	62	unsigned int cq_head = READ_ONCE(r->cq.head);
	63	unsigned int cq_tail = READ_ONCE(r->cq.tail);
	64	unsigned int cq_shift = 0;
3b8fdd1d	65	unsigned int sq_shift = 0;
a4ad4f74	66	unsigned int sq_entries, cq_entries;
7644b1a1	67	int sq_pid = -1, sq_cpu = -1;
3fcb9d17	68	u64 sq_total_time = 0, sq_work_time = 0;
a4ad4f74	69	bool has_lock;
a4ad4f74 JA	70	unsigned int i;
a4ad4f74 JA	71
4f731705	72	if (ctx->flags & IORING_SETUP_CQE32)
a4ad4f74	73	cq_shift = 1;
3b8fdd1d JA	74	if (ctx->flags & IORING_SETUP_SQE128)
3b8fdd1d JA	75	sq_shift = 1;
a4ad4f74 JA	76
	77	/*
	78	* we may get imprecise sqe and cqe info if uring is actively running
	79	* since we get cached_sq_head and cached_cq_tail without uring_lock
	80	* and sq_tail and cq_head are changed by userspace. But it's ok since
	81	* we usually use these info when it is stuck.
	82	*/
	83	seq_printf(m, "SqMask:\t0x%x\n", sq_mask);
	84	seq_printf(m, "SqHead:\t%u\n", sq_head);
	85	seq_printf(m, "SqTail:\t%u\n", sq_tail);
	86	seq_printf(m, "CachedSqHead:\t%u\n", ctx->cached_sq_head);
	87	seq_printf(m, "CqMask:\t0x%x\n", cq_mask);
	88	seq_printf(m, "CqHead:\t%u\n", cq_head);
	89	seq_printf(m, "CqTail:\t%u\n", cq_tail);
	90	seq_printf(m, "CachedCqTail:\t%u\n", ctx->cached_cq_tail);
3b8fdd1d	91	seq_printf(m, "SQEs:\t%u\n", sq_tail - sq_head);
a4ad4f74 JA	92	sq_entries = min(sq_tail - sq_head, ctx->sq_entries);
	93	for (i = 0; i < sq_entries; i++) {
	94	unsigned int entry = i + sq_head;
a4ad4f74	95	struct io_uring_sqe *sqe;
3b8fdd1d	96	unsigned int sq_idx;
a4ad4f74	97
32f5dea0 JA	98	if (ctx->flags & IORING_SETUP_NO_SQARRAY)
32f5dea0 JA	99	break;
3b8fdd1d	100	sq_idx = READ_ONCE(ctx->sq_array[entry & sq_mask]);
a4ad4f74 JA	101	if (sq_idx > sq_mask)
a4ad4f74 JA	102	continue;
00927931	103	sqe = &ctx->sq_sqes[sq_idx << sq_shift];
3b8fdd1d JA	104	seq_printf(m, "%5u: opcode:%s, fd:%d, flags:%x, off:%llu, "
	105	"addr:0x%llx, rw_flags:0x%x, buf_index:%d "
	106	"user_data:%llu",
	107	sq_idx, io_uring_get_opcode(sqe->opcode), sqe->fd,
	108	sqe->flags, (unsigned long long) sqe->off,
	109	(unsigned long long) sqe->addr, sqe->rw_flags,
	110	sqe->buf_index, sqe->user_data);
	111	if (sq_shift) {
	112	u64 sqeb = (void ) (sqe + 1);
	113	int size = sizeof(struct io_uring_sqe) / sizeof(u64);
	114	int j;
	115
	116	for (j = 0; j < size; j++) {
	117	seq_printf(m, ", e%d:0x%llx", j,
	118	(unsigned long long) *sqeb);
	119	sqeb++;
	120	}
	121	}
	122	seq_printf(m, "\n");
a4ad4f74 JA	123	}
	124	seq_printf(m, "CQEs:\t%u\n", cq_tail - cq_head);
	125	cq_entries = min(cq_tail - cq_head, ctx->cq_entries);
	126	for (i = 0; i < cq_entries; i++) {
	127	unsigned int entry = i + cq_head;
	128	struct io_uring_cqe *cqe = &r->cqes[(entry & cq_mask) << cq_shift];
	129
4f731705	130	seq_printf(m, "%5u: user_data:%llu, res:%d, flag:%x",
a4ad4f74 JA	131	entry & cq_mask, cqe->user_data, cqe->res,
a4ad4f74 JA	132	cqe->flags);
4f731705 JA	133	if (cq_shift)
	134	seq_printf(m, ", extra1:%llu, extra2:%llu\n",
	135	cqe->big_cqe[0], cqe->big_cqe[1]);
	136	seq_printf(m, "\n");
a4ad4f74 JA	137	}
	138
	139	/*
	140	* Avoid ABBA deadlock between the seq lock and the io_uring mutex,
	141	* since fdinfo case grabs it in the opposite direction of normal use
	142	* cases. If we fail to get the lock, we just don't iterate any
	143	* structures that could be going away outside the io_uring mutex.
	144	*/
	145	has_lock = mutex_trylock(&ctx->uring_lock);
	146
	147	if (has_lock && (ctx->flags & IORING_SETUP_SQPOLL)) {
7644b1a1 JA	148	struct io_sq_data *sq = ctx->sq_data;
7644b1a1 JA	149
606559dc GKB	150	/*
	151	* sq->thread might be NULL if we raced with the sqpoll
	152	* thread termination.
	153	*/
	154	if (sq->thread) {
	155	sq_pid = sq->task_pid;
	156	sq_cpu = sq->sq_cpu;
	157	getrusage(sq->thread, RUSAGE_SELF, &sq_usage);
	158	sq_total_time = (sq_usage.ru_stime.tv_sec * 1000000
	159	+ sq_usage.ru_stime.tv_usec);
	160	sq_work_time = sq->work_time;
	161	}
a4ad4f74 JA	162	}
a4ad4f74 JA	163
7644b1a1 JA	164	seq_printf(m, "SqThread:\t%d\n", sq_pid);
7644b1a1 JA	165	seq_printf(m, "SqThreadCpu:\t%d\n", sq_cpu);
3fcb9d17 XL	166	seq_printf(m, "SqTotalTime:\t%llu\n", sq_total_time);
3fcb9d17 XL	167	seq_printf(m, "SqWorkTime:\t%llu\n", sq_work_time);
a4ad4f74 JA	168	seq_printf(m, "UserFiles:\t%u\n", ctx->nr_user_files);
	169	for (i = 0; has_lock && i < ctx->nr_user_files; i++) {
	170	struct file *f = io_file_from_index(&ctx->file_table, i);
	171
	172	if (f)
	173	seq_printf(m, "%5u: %s\n", i, file_dentry(f)->d_iname);
	174	else
	175	seq_printf(m, "%5u: <none>\n", i);
	176	}
	177	seq_printf(m, "UserBufs:\t%u\n", ctx->nr_user_bufs);
	178	for (i = 0; has_lock && i < ctx->nr_user_bufs; i++) {
	179	struct io_mapped_ubuf *buf = ctx->user_bufs[i];
	180	unsigned int len = buf->ubuf_end - buf->ubuf;
	181
	182	seq_printf(m, "%5u: 0x%llx/%u\n", i, buf->ubuf, len);
	183	}
	184	if (has_lock && !xa_empty(&ctx->personalities)) {
	185	unsigned long index;
	186	const struct cred *cred;
	187
	188	seq_printf(m, "Personalities:\n");
	189	xa_for_each(&ctx->personalities, index, cred)
	190	io_uring_show_cred(m, index, cred);
	191	}
a4ad4f74 JA	192
a4ad4f74 JA	193	seq_puts(m, "PollList:\n");
e6f89be6 PB	194	for (i = 0; i < (1U << ctx->cancel_table.hash_bits); i++) {
e6f89be6 PB	195	struct io_hash_bucket *hb = &ctx->cancel_table.hbs[i];
ea97cbeb	196	struct io_hash_bucket *hbl = &ctx->cancel_table_locked.hbs[i];
a4ad4f74 JA	197	struct io_kiocb *req;
a4ad4f74 JA	198
38513c46 HX	199	spin_lock(&hb->lock);
38513c46 HX	200	hlist_for_each_entry(req, &hb->list, hash_node)
a4ad4f74 JA	201	seq_printf(m, " op=%d, task_works=%d\n", req->opcode,
a4ad4f74 JA	202	task_work_pending(req->task));
38513c46	203	spin_unlock(&hb->lock);
ea97cbeb JA	204
	205	if (!has_lock)
	206	continue;
	207	hlist_for_each_entry(req, &hbl->list, hash_node)
	208	seq_printf(m, " op=%d, task_works=%d\n", req->opcode,
	209	task_work_pending(req->task));
a4ad4f74 JA	210	}
a4ad4f74 JA	211
ea97cbeb JA	212	if (has_lock)
	213	mutex_unlock(&ctx->uring_lock);
	214
a4ad4f74	215	seq_puts(m, "CqOverflowList:\n");
38513c46	216	spin_lock(&ctx->completion_lock);
a4ad4f74 JA	217	list_for_each_entry(ocqe, &ctx->cq_overflow_list, list) {
	218	struct io_uring_cqe *cqe = &ocqe->cqe;
	219
	220	seq_printf(m, " user_data=%llu, res=%d, flags=%x\n",
	221	cqe->user_data, cqe->res, cqe->flags);
	222
	223	}
	224
	225	spin_unlock(&ctx->completion_lock);
	226	}
a4ad4f74	227	#endif