MINOR: mux-quic: define rxbuf wrapper

Define a new type qc_stream_rxbuf. This is used as a wrapper around QCS
Rx buffer with encapsulation of the ncbuf storage. It is allocated via a
new pool. Several functions are adapted to be able to deal with
qc_stream_rxbuf as a wrapper instead of the previous plain ncbuf
instance.

No functional change should happen with this patch. For now, only a
single qc_stream_rxbuf can be instantiated per QCS. However, this new
type will be useful to implement multiple Rx buffer storage in a future
commit.
This commit is contained in:
Amaury Denoyelle 2025-02-24 16:22:22 +01:00
parent 4b1e63d191
commit cc3c2d1f12
2 changed files with 71 additions and 24 deletions

View File

@ -134,6 +134,13 @@ enum qcs_state {
QC_SS_CLO, /* closed */
} __attribute__((packed));
/* STREAM receive buffer. Can handle out-of-order storage. */
struct qc_stream_rxbuf {
struct ncbuf ncb; /* data storage with support for out of order offset */
uint64_t off; /* base offset of current buffer */
uint64_t off_end; /* first offset directly outside of current buffer */
};
struct qcs {
struct qcc *qcc;
struct sedesc *sd;
@ -142,9 +149,9 @@ struct qcs {
void *ctx; /* app-ops context */
struct {
uint64_t offset; /* absolute current base offset of ncbuf */
uint64_t offset; /* read offset */
uint64_t offset_max; /* maximum absolute offset received */
struct ncbuf ncbuf; /* receive buffer - can handle out-of-order offset frames */
struct qc_stream_rxbuf *buf; /* single receive buffer */
struct buffer app_buf; /* receive buffer used by stconn layer */
uint64_t msd; /* current max-stream-data limit to enforce */
uint64_t msd_init; /* initial max-stream-data */

View File

@ -34,6 +34,7 @@
DECLARE_POOL(pool_head_qcc, "qcc", sizeof(struct qcc));
DECLARE_POOL(pool_head_qcs, "qcs", sizeof(struct qcs));
DECLARE_STATIC_POOL(pool_head_qc_stream_rxbuf, "qc_stream_rxbuf", sizeof(struct qc_stream_rxbuf));
static void qmux_ctrl_send(struct qc_stream_desc *, uint64_t data, uint64_t offset);
static void qmux_ctrl_room(struct qc_stream_desc *, uint64_t room);
@ -44,23 +45,30 @@ static int qcc_is_pacing_active(const struct connection *conn)
return !(quic_tune.options & QUIC_TUNE_NO_PACING);
}
static void qcs_free_ncbuf(struct qcs *qcs, struct ncbuf *ncbuf)
/* Free <rxbuf> instance and its inner data storage attached to <qcs> stream. */
static void qcs_free_rxbuf(struct qcs *qcs, struct qc_stream_rxbuf *rxbuf)
{
struct ncbuf *ncbuf;
struct buffer buf;
if (ncb_is_null(ncbuf))
if (!rxbuf)
return;
ncbuf = &rxbuf->ncb;
if (!ncb_is_null(ncbuf)) {
buf = b_make(ncbuf->area, ncbuf->size, 0, 0);
b_free(&buf);
offer_buffers(NULL, 1);
*ncbuf = NCBUF_NULL;
}
/* Reset DEM_FULL as buffer is released. This ensures mux is not woken
* up from rcv_buf stream callback when demux was previously blocked.
*/
qcs->flags &= ~QC_SF_DEM_FULL;
pool_free(pool_head_qc_stream_rxbuf, rxbuf);
qcs->rx.buf = NULL;
}
/* Free <qcs> instance. This function is reserved for internal usage : it must
@ -97,7 +105,7 @@ static void qcs_free(struct qcs *qcs)
}
/* Free Rx buffer. */
qcs_free_ncbuf(qcs, &qcs->rx.ncbuf);
qcs_free_rxbuf(qcs, qcs->rx.buf);
/* Remove qcs from qcc tree. */
eb64_delete(&qcs->by_id);
@ -152,7 +160,7 @@ static struct qcs *qcs_new(struct qcc *qcc, uint64_t id, enum qcs_type type)
qfctl_init(&qcs->tx.fc, 0);
}
qcs->rx.ncbuf = NCBUF_NULL;
qcs->rx.buf = NULL;
qcs->rx.app_buf = BUF_NULL;
qcs->rx.offset = qcs->rx.offset_max = 0;
@ -1074,10 +1082,22 @@ int qcc_get_qcs(struct qcc *qcc, uint64_t id, int receive_only, int send_only,
return 1;
}
/* Simple function to duplicate a buffer */
static inline struct buffer qcs_b_dup(const struct ncbuf *b)
/* Convert <b> out-of-order storage into a contiguous buffer. */
static inline struct buffer qcs_b_dup(const struct qc_stream_rxbuf *b)
{
return b_make(ncb_orig(b), b->size, b->head, ncb_data(b, 0));
if (b) {
const struct ncbuf *ncb = &b->ncb;
return b_make(ncb_orig(ncb), ncb->size, ncb->head, ncb_data(ncb, 0));
}
else {
return BUF_NULL;
}
}
/* Returns the amount of data readable at <qcs> stream current offset. */
static ncb_sz_t qcs_rx_avail_data(struct qcs *qcs)
{
return qcs->rx.buf ? ncb_data(&qcs->rx.buf->ncb, 0) : 0;
}
/* Remove <bytes> from <qcs> Rx buffer. Flow-control for received offsets may
@ -1087,18 +1107,18 @@ static void qcs_consume(struct qcs *qcs, uint64_t bytes)
{
struct qcc *qcc = qcs->qcc;
struct quic_frame *frm;
struct ncbuf *buf = &qcs->rx.ncbuf;
struct ncbuf *ncbuf = &qcs->rx.buf->ncb;
enum ncb_ret ret;
TRACE_ENTER(QMUX_EV_QCS_RECV, qcc->conn, qcs);
ret = ncb_advance(buf, bytes);
ret = ncb_advance(ncbuf, bytes);
if (ret) {
ABORT_NOW(); /* should not happens because removal only in data */
}
if (ncb_is_empty(buf))
qcs_free_ncbuf(qcs, buf);
if (ncb_is_empty(ncbuf))
qcs_free_rxbuf(qcs, qcs->rx.buf);
qcs->rx.offset += bytes;
/* Not necessary to emit a MAX_STREAM_DATA if all data received. */
@ -1157,7 +1177,7 @@ static int qcc_decode_qcs(struct qcc *qcc, struct qcs *qcs)
TRACE_ENTER(QMUX_EV_QCS_RECV, qcc->conn, qcs);
b = qcs_b_dup(&qcs->rx.ncbuf);
b = qcs_b_dup(qcs->rx.buf);
/* Signal FIN to application if STREAM FIN received with all data. */
if (qcs_is_close_remote(qcs))
@ -1500,6 +1520,7 @@ int qcc_recv(struct qcc *qcc, uint64_t id, uint64_t len, uint64_t offset,
char fin, char *data)
{
const int fin_standalone = (!len && fin);
struct ncbuf *ncbuf;
struct qcs *qcs;
enum ncb_ret ret;
@ -1586,13 +1607,32 @@ int qcc_recv(struct qcc *qcc, uint64_t id, uint64_t len, uint64_t offset,
}
if (len) {
if (!qcs_get_ncbuf(qcs, &qcs->rx.ncbuf) || ncb_is_null(&qcs->rx.ncbuf)) {
if (!qcs->rx.buf) {
struct qc_stream_rxbuf *buf;
buf = pool_alloc(pool_head_qc_stream_rxbuf);
if (!buf) {
TRACE_ERROR("rxbuf alloc failure", QMUX_EV_QCC_RECV|QMUX_EV_QCS_RECV, qcc->conn, qcs);
qcc_set_error(qcc, QC_ERR_INTERNAL_ERROR, 0);
goto err;
}
buf->ncb = NCBUF_NULL;
buf->off = qcs->rx.offset;
buf->off_end = buf->off + qmux_stream_rx_bufsz();
qcs->rx.buf = buf;
ncbuf = &buf->ncb;
}
else {
ncbuf = &qcs->rx.buf->ncb;
}
if (!qcs_get_ncbuf(qcs, ncbuf) || ncb_is_null(ncbuf)) {
TRACE_ERROR("receive ncbuf alloc failure", QMUX_EV_QCC_RECV|QMUX_EV_QCS_RECV, qcc->conn, qcs);
qcc_set_error(qcc, QC_ERR_INTERNAL_ERROR, 0);
goto err;
}
ret = ncb_add(&qcs->rx.ncbuf, offset - qcs->rx.offset, data, len, NCB_ADD_COMPARE);
ret = ncb_add(ncbuf, offset - qcs->rx.offset, data, len, NCB_ADD_COMPARE);
switch (ret) {
case NCB_RET_OK:
break;
@ -1624,11 +1664,11 @@ int qcc_recv(struct qcc *qcc, uint64_t id, uint64_t len, uint64_t offset,
qcs->flags |= QC_SF_SIZE_KNOWN;
if (qcs->flags & QC_SF_SIZE_KNOWN &&
qcs->rx.offset_max == qcs->rx.offset + ncb_data(&qcs->rx.ncbuf, 0)) {
qcs->rx.offset_max == qcs->rx.offset + qcs_rx_avail_data(qcs)) {
qcs_close_remote(qcs);
}
if ((ncb_data(&qcs->rx.ncbuf, 0) && !(qcs->flags & QC_SF_DEM_FULL)) ||
if ((qcs_rx_avail_data(qcs) && !(qcs->flags & QC_SF_DEM_FULL)) ||
unlikely(fin_standalone && qcs_is_close_remote(qcs))) {
qcc_decode_qcs(qcc, qcs);
LIST_DEL_INIT(&qcs->el_recv);
@ -1805,7 +1845,7 @@ int qcc_recv_reset_stream(struct qcc *qcc, uint64_t id, uint64_t err, uint64_t f
*/
qcs->flags |= QC_SF_SIZE_KNOWN|QC_SF_RECV_RESET;
qcs_close_remote(qcs);
qcs_free_ncbuf(qcs, &qcs->rx.ncbuf);
qcs_free_rxbuf(qcs, qcs->rx.buf);
out:
if (qcc->glitches != prev_glitches)
@ -3345,7 +3385,7 @@ static size_t qmux_strm_rcv_buf(struct stconn *sc, struct buffer *buf,
/* Ensure DEM_FULL is only set if there is available data to
* ensure we never do unnecessary wakeup here.
*/
BUG_ON(!ncb_data(&qcs->rx.ncbuf, 0));
BUG_ON(!qcs_rx_avail_data(qcs));
qcs->flags &= ~QC_SF_DEM_FULL;
if (!(qcc->flags & QC_CF_ERRL)) {