niansa-misc/linux
1
0
Fork 0
mirror of synced 2025-03-06 20:59:54 +01:00
Code Issues Projects Releases Packages Wiki Activity

io_uring/register: document io_register_resize_rings() shared mem usage

Browse source 
It can be a bit hard to tell which parts of io_register_resize_rings()
are operating on shared memory, and which ones are not. And anything
reading or writing to those regions should really use the read/write
once primitives.

Hence add those, ensuring sanity in how this memory is accessed, and
helping document the shared nature of it.

Signed-off-by: Jens Axboe <axboe@kernel.dk>
This commit is contained in:
Jens Axboe 2025-01-15 08:23:55 -07:00
parent 8911798d3e
commit 2c5aae129f
1 changed files with 26 additions and 18 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

44
io_uring/register.c
View file

@ -449,10 +449,18 @@ static int io_register_resize_rings(struct io_ring_ctx *ctx, void __user *arg)
if (IS_ERR(n.rings)) if (IS_ERR(n.rings))
return PTR_ERR(n.rings); return PTR_ERR(n.rings);
n.rings->sq_ring_mask = p.sq_entries - 1; /*
n.rings->cq_ring_mask = p.cq_entries - 1; * At this point n.rings is shared with userspace, just like o.rings
n.rings->sq_ring_entries = p.sq_entries; * is as well. While we don't expect userspace to modify it while
n.rings->cq_ring_entries = p.cq_entries; * a resize is in progress, and it's most likely that userspace will
* shoot itself in the foot if it does, we can't always assume good
* intent... Use read/write once helpers from here on to indicate the
* shared nature of it.
*/
WRITE_ONCE(n.rings->sq_ring_mask, p.sq_entries - 1);
WRITE_ONCE(n.rings->cq_ring_mask, p.cq_entries - 1);
WRITE_ONCE(n.rings->sq_ring_entries, p.sq_entries);
WRITE_ONCE(n.rings->cq_ring_entries, p.cq_entries);
if (copy_to_user(arg, &p, sizeof(p))) { if (copy_to_user(arg, &p, sizeof(p))) {
io_register_free_rings(&p, &n); io_register_free_rings(&p, &n);
@ -509,20 +517,20 @@ static int io_register_resize_rings(struct io_ring_ctx *ctx, void __user *arg)
* rings can't hold what is already there, then fail the operation. * rings can't hold what is already there, then fail the operation.
*/ */
n.sq_sqes = ptr; n.sq_sqes = ptr;
tail = o.rings->sq.tail; tail = READ_ONCE(o.rings->sq.tail);
if (tail - o.rings->sq.head > p.sq_entries) if (tail - READ_ONCE(o.rings->sq.head) > p.sq_entries)
goto overflow; goto overflow;
for (i = o.rings->sq.head; i < tail; i++) { for (i = READ_ONCE(o.rings->sq.head); i < tail; i++) {
unsigned src_head = i & (ctx->sq_entries - 1); unsigned src_head = i & (ctx->sq_entries - 1);
unsigned dst_head = i & (p.sq_entries - 1); unsigned dst_head = i & (p.sq_entries - 1);
n.sq_sqes[dst_head] = o.sq_sqes[src_head]; n.sq_sqes[dst_head] = o.sq_sqes[src_head];
} }
n.rings->sq.head = o.rings->sq.head; WRITE_ONCE(n.rings->sq.head, READ_ONCE(o.rings->sq.head));
n.rings->sq.tail = o.rings->sq.tail; WRITE_ONCE(n.rings->sq.tail, READ_ONCE(o.rings->sq.tail));
tail = o.rings->cq.tail; tail = READ_ONCE(o.rings->cq.tail);
if (tail - o.rings->cq.head > p.cq_entries) { if (tail - READ_ONCE(o.rings->cq.head) > p.cq_entries) {
overflow: overflow:
/* restore old rings, and return -EOVERFLOW via cleanup path */ /* restore old rings, and return -EOVERFLOW via cleanup path */
ctx->rings = o.rings; ctx->rings = o.rings;
@ -531,21 +539,21 @@ overflow:
ret = -EOVERFLOW; ret = -EOVERFLOW;
goto out; goto out;
} }
for (i = o.rings->cq.head; i < tail; i++) { for (i = READ_ONCE(o.rings->cq.head); i < tail; i++) {
unsigned src_head = i & (ctx->cq_entries - 1); unsigned src_head = i & (ctx->cq_entries - 1);
unsigned dst_head = i & (p.cq_entries - 1); unsigned dst_head = i & (p.cq_entries - 1);
n.rings->cqes[dst_head] = o.rings->cqes[src_head]; n.rings->cqes[dst_head] = o.rings->cqes[src_head];
} }
n.rings->cq.head = o.rings->cq.head; WRITE_ONCE(n.rings->cq.head, READ_ONCE(o.rings->cq.head));
n.rings->cq.tail = o.rings->cq.tail; WRITE_ONCE(n.rings->cq.tail, READ_ONCE(o.rings->cq.tail));
/* invalidate cached cqe refill */ /* invalidate cached cqe refill */
ctx->cqe_cached = ctx->cqe_sentinel = NULL; ctx->cqe_cached = ctx->cqe_sentinel = NULL;
n.rings->sq_dropped = o.rings->sq_dropped; WRITE_ONCE(n.rings->sq_dropped, READ_ONCE(o.rings->sq_dropped));
n.rings->sq_flags = o.rings->sq_flags; WRITE_ONCE(n.rings->sq_flags, READ_ONCE(o.rings->sq_flags));
n.rings->cq_flags = o.rings->cq_flags; WRITE_ONCE(n.rings->cq_flags, READ_ONCE(o.rings->cq_flags));
n.rings->cq_overflow = o.rings->cq_overflow; WRITE_ONCE(n.rings->cq_overflow, READ_ONCE(o.rings->cq_overflow));
/* all done, store old pointers and assign new ones */ /* all done, store old pointers and assign new ones */
if (!(ctx->flags & IORING_SETUP_NO_SQARRAY)) if (!(ctx->flags & IORING_SETUP_NO_SQARRAY))