lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
Open Source and information security mailing list archives
| ||
|
Message-ID: <11755fdb-4a28-0ea5-89a4-d51b2715f8c2@samba.org> Date: Fri, 21 Oct 2022 16:03:54 +0200 From: Stefan Metzmacher <metze@...ba.org> To: Pavel Begunkov <asml.silence@...il.com>, io-uring <io-uring@...r.kernel.org>, Jens Axboe <axboe@...nel.dk> Cc: Jakub Kicinski <kuba@...nel.org>, netdev <netdev@...r.kernel.org>, Dylan Yudaken <dylany@...com> Subject: Re: IORING_SEND_NOTIF_REPORT_USAGE (was Re: IORING_CQE_F_COPIED) Am 21.10.22 um 13:09 schrieb Pavel Begunkov: > On 10/21/22 10:36, Stefan Metzmacher wrote: >> Hi Pavel, > [...] >>> Right, I'm just tired of back porting patches by hand :) >> >> ok, I just assumed it would be 6.1 only. > > I'm fine with 6.1 only, it'd make things easier. I thought from > your first postings you wanted it 6.0. Then we don't need to care > about the placing of the copied/used flags. > >>>> Otherwise we could have IORING_CQE_F_COPIED by default without opt-in >>>> flag... >> >> Do you still want an opt-in flag to get IORING_CQE_F_COPIED? >> If so what name do you want it to be? > > Ala a IORING_SEND_* flag? Yes please. > > *_REPORT_USAGE was fine but I'd make it IORING_SEND_ZC_REPORT_USAGE. > And can be extended if there is more info needed in the future. > > And I don't mind using a bit in cqe->res, makes cflags less polluted. So no worries about the delayed/skip sendmsg completion anymore? Should I define it like this, ok? #define IORING_NOTIF_USAGE_ZC_COPIED (1U << 31) See the full patch below... metze diff --git a/include/uapi/linux/io_uring.h b/include/uapi/linux/io_uring.h index d69ae7eba773..32e1f2a55b70 100644 --- a/include/uapi/linux/io_uring.h +++ b/include/uapi/linux/io_uring.h @@ -296,10 +296,24 @@ enum io_uring_op { * * IORING_RECVSEND_FIXED_BUF Use registered buffers, the index is stored in * the buf_index field. + + * IORING_SEND_NOTIF_REPORT_USAGE + * If SEND[MSG]_ZC should report + * the zerocopy usage in cqe.res + * for the IORING_CQE_F_NOTIF cqe. + * IORING_NOTIF_USAGE_ZC_COPIED if data was copied + * (at least partially). */ #define IORING_RECVSEND_POLL_FIRST (1U << 0) #define IORING_RECV_MULTISHOT (1U << 1) #define IORING_RECVSEND_FIXED_BUF (1U << 2) +#define IORING_SEND_ZC_REPORT_USAGE (1U << 3) + +/* + * cqe.res for IORING_CQE_F_NOTIF if + * IORING_SEND_ZC_REPORT_USAGE was requested + */ +#define IORING_NOTIF_USAGE_ZC_COPIED (1U << 31) /* * accept flags stored in sqe->ioprio diff --git a/io_uring/net.c b/io_uring/net.c index 56078f47efe7..1aa3b50b3e82 100644 --- a/io_uring/net.c +++ b/io_uring/net.c @@ -939,7 +939,8 @@ int io_send_zc_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe) zc->flags = READ_ONCE(sqe->ioprio); if (zc->flags & ~(IORING_RECVSEND_POLL_FIRST | - IORING_RECVSEND_FIXED_BUF)) + IORING_RECVSEND_FIXED_BUF | + IORING_SEND_ZC_REPORT_USAGE)) return -EINVAL; notif = zc->notif = io_alloc_notif(ctx); if (!notif) @@ -957,6 +958,9 @@ int io_send_zc_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe) req->imu = READ_ONCE(ctx->user_bufs[idx]); io_req_set_rsrc_node(notif, ctx, 0); } + if (zc->flags & IORING_SEND_ZC_REPORT_USAGE) { + io_notif_to_data(notif)->zc_report = true; + } if (req->opcode == IORING_OP_SEND_ZC) { if (READ_ONCE(sqe->__pad3[0])) diff --git a/io_uring/notif.c b/io_uring/notif.c index e37c6569d82e..4bfef10161fa 100644 --- a/io_uring/notif.c +++ b/io_uring/notif.c @@ -18,6 +18,10 @@ static void __io_notif_complete_tw(struct io_kiocb *notif, bool *locked) __io_unaccount_mem(ctx->user, nd->account_pages); nd->account_pages = 0; } + + if (nd->zc_report && (nd->zc_copied || !nd->zc_used)) + notif->cqe.res |= IORING_NOTIF_USAGE_ZC_COPIED; + io_req_task_complete(notif, locked); } @@ -28,6 +32,13 @@ static void io_uring_tx_zerocopy_callback(struct sk_buff *skb, struct io_notif_data *nd = container_of(uarg, struct io_notif_data, uarg); struct io_kiocb *notif = cmd_to_io_kiocb(nd); + if (nd->zc_report) { + if (success && !nd->zc_used && skb) + WRITE_ONCE(nd->zc_used, true); + else if (!success && !nd->zc_copied) + WRITE_ONCE(nd->zc_copied, true); + } + if (refcount_dec_and_test(&uarg->refcnt)) { notif->io_task_work.func = __io_notif_complete_tw; io_req_task_work_add(notif); @@ -55,6 +66,7 @@ struct io_kiocb *io_alloc_notif(struct io_ring_ctx *ctx) nd->account_pages = 0; nd->uarg.flags = SKBFL_ZEROCOPY_FRAG | SKBFL_DONT_ORPHAN; nd->uarg.callback = io_uring_tx_zerocopy_callback; + nd->zc_report = nd->zc_used = nd->zc_copied = false; refcount_set(&nd->uarg.refcnt, 1); return notif; } diff --git a/io_uring/notif.h b/io_uring/notif.h index e4fbcae0f3fd..6be2e5ae8581 100644 --- a/io_uring/notif.h +++ b/io_uring/notif.h @@ -15,6 +15,9 @@ struct io_notif_data { struct file *file; struct ubuf_info uarg; unsigned long account_pages; + bool zc_report; + bool zc_used; + bool zc_copied; }; void io_notif_flush(struct io_kiocb *notif);
Powered by blists - more mailing lists