On 22 Jun 2018, at 21:04, Lam, Tiago wrote:
On 18/06/2018 12:50, Eelco Chaudron wrote:On 11 Jun 2018, at 18:21, Tiago Lam wrote:Most helper functions in dp-packet assume that the data held by a dp_packet is contiguous, and perform operations such as pointer arithmetic under that assumption. However, with the introduction of multi-segment mbufs, where data is non-contiguous, such assumptions are no longer possible. Some examples of Such helper functions are dp_packet_tail(), dp_packet_tailroom(), dp_packet_end(), dp_packet_get_allocated() and dp_packet_at(). Thus, instead of assuming contiguous data in dp_packet, they now iterate over the (non-contiguous) data in mbufs to perform their calculations. Finally, dp_packet_use__() has also been modified to perform the initialisation of the packet (and setting the source) before continuing to set its size and data length, which now depends on the type of packet. Co-authored-by: Mark Kavanagh <[email protected]> Signed-off-by: Mark Kavanagh <[email protected]> Signed-off-by: Tiago Lam <[email protected]> --- lib/dp-packet.c | 4 +- lib/dp-packet.h | 242 +++++++++++++++++++++++++++++++++++++++++++++----------- 2 files changed, 197 insertions(+), 49 deletions(-) diff --git a/lib/dp-packet.c b/lib/dp-packet.c index 782e7c2..2aaeaae 100644 --- a/lib/dp-packet.c +++ b/lib/dp-packet.c @@ -41,11 +41,11 @@ static void dp_packet_use__(struct dp_packet *b, void *base, size_t allocated, enum dp_packet_source source) { + dp_packet_init__(b, allocated, source); + dp_packet_set_base(b, base); dp_packet_set_data(b, base); dp_packet_set_size(b, 0); - - dp_packet_init__(b, allocated, source); } /* Initializes 'b' as an empty dp_packet that contains the 'allocated' bytes of diff --git a/lib/dp-packet.h b/lib/dp-packet.h index c301ed5..272597f 100644 --- a/lib/dp-packet.h +++ b/lib/dp-packet.h @@ -133,6 +133,10 @@ static inline void *dp_packet_at(const struct dp_packet *, size_t offset, size_t size); static inline void *dp_packet_at_assert(const struct dp_packet *,size_t offset, size_t size);+#ifdef DPDK_NETDEV +static inline void * dp_packet_at_offset(const struct dp_packet *b, + size_t offset); +#endif static inline void *dp_packet_tail(const struct dp_packet *); static inline void *dp_packet_end(const struct dp_packet *); @@ -180,40 +184,6 @@ dp_packet_delete(struct dp_packet *b) } }-/* If 'b' contains at least 'offset + size' bytes of data, returns apointer to - * byte 'offset'. Otherwise, returns a null pointer. */ -static inline void * -dp_packet_at(const struct dp_packet *b, size_t offset, size_t size) -{ - return offset + size <= dp_packet_size(b) - ? (char *) dp_packet_data(b) + offset - : NULL; -} - -/* Returns a pointer to byte 'offset' in 'b', which must contain at least - * 'offset + size' bytes of data. */ -static inline void *-dp_packet_at_assert(const struct dp_packet *b, size_t offset, size_tsize) -{ - ovs_assert(offset + size <= dp_packet_size(b)); - return ((char *) dp_packet_data(b)) + offset; -} - -/* Returns a pointer to byte following the last byte of data in use in 'b'. */ -static inline void * -dp_packet_tail(const struct dp_packet *b) -{ - return (char *) dp_packet_data(b) + dp_packet_size(b); -} - -/* Returns a pointer to byte following the last byte allocated for use (but - * not necessarily in use) in 'b'. */ -static inline void * -dp_packet_end(const struct dp_packet *b) -{ - return (char *) dp_packet_base(b) + dp_packet_get_allocated(b); -} - /* Returns the number of bytes of headroom in 'b', that is, the number of bytes * of unused space in dp_packet 'b' before the data that is in use. (Most * commonly, the data in a dp_packet is at its beginning, and thus the @@ -229,6 +199,14 @@ dp_packet_headroom(const struct dp_packet *b) static inline size_t dp_packet_tailroom(const struct dp_packet *b) { +#ifdef DPDK_NETDEV + if (b->source == DPBUF_DPDK) { + struct rte_mbuf *tmbuf = CONST_CAST(struct rte_mbuf *, &b->mbuf); + tmbuf = rte_pktmbuf_lastseg(tmbuf); +What if the last two chains contain no user data, we have more tailroom?!? Should we not call (tmbuf)?I've covered this in my reply to the cover letter, hopefully it is more clear now. This shouldn't be a possibility in the implementation. Also,we do call rte_pktmbuf_tailroom(), just below.
I know but I do not see the reason for not calling rte_pktmbuf_tailroom() on the mbuf directly. This will work for all cases, add’s less complexity, even in the future if we move to a design where we will support multiple buffers. Trying to “limit” the usual mbuf cases does not sound right, and it will confuse people.
+ return rte_pktmbuf_tailroom(tmbuf); + } +#endif return (char *) dp_packet_end(b) - (char *) dp_packet_tail(b); } @@ -236,8 +214,13 @@ dp_packet_tailroom(const struct dp_packet *b) static inline void dp_packet_clear(struct dp_packet *b) { +#ifdef DPDK_NETDEV + dp_packet_set_size(b, 0); + rte_pktmbuf_reset(&b->mbuf);This would probably assert on none DPBUF_DPDK, do we need a “if (b->source == DPBUF_DPDK)” here also?We do, I'll add it to next iteration. It only asserts when using DPDK with `RTE_LIBRTE_MBUF_DEBUG`, but still...
Yes, but you do not want this assert when you are trying to debug something else ;)
+#else dp_packet_set_data(b, dp_packet_base(b)); dp_packet_set_size(b, 0); +#endif }/* Removes 'size' bytes from the head end of 'b', which must containat least@@ -252,12 +235,32 @@ dp_packet_pull(struct dp_packet *b, size_t size)return data; } +#ifdef DPDK_NETDEV +/* Similar to dp_packet_try_pull() but doesn't actually pull any code, only+ * returns true or false if it would be possible to actually pull anycode. + * Valid for dp_packets carrying mbufs only. */ +static inline bool +dp_packet_mbuf_may_pull(const struct dp_packet *b, size_t size) { + if (size > b->mbuf.data_len) { + return false; + } + + return true; +} +#endif +/* If 'b' has at least 'size' bytes of data, removes that many bytesfrom the * head end of 'b' and returns the first byte removed. Otherwise, returns a * null pointer without modifying 'b'. */ static inline void * dp_packet_try_pull(struct dp_packet *b, size_t size) { +#ifdef DPDK_NETDEV + if (!dp_packet_mbuf_may_pull(b, size)) { + return NULL; + } +#endif + return dp_packet_size(b) - dp_packet_l2_pad_size(b) >= size ? dp_packet_pull(b, size) : NULL; } @@ -311,6 +314,12 @@ dp_packet_set_l2_pad_size(struct dp_packet *b, uint8_t pad_size) static inline void * dp_packet_l2_5(const struct dp_packet *b) { +#ifdef DPDK_NETDEV + if (!dp_packet_mbuf_may_pull(b, b->l2_5_ofs)) { + return NULL; + } +#endif + return b->l2_5_ofs != UINT16_MAX ? (char *) dp_packet_data(b) + b->l2_5_ofs : NULL; @@ -327,6 +336,12 @@ dp_packet_set_l2_5(struct dp_packet *b, void *l2_5) static inline void * dp_packet_l3(const struct dp_packet *b) { +#ifdef DPDK_NETDEV + if (!dp_packet_mbuf_may_pull(b, b->l3_ofs)) { + return NULL; + } +#endif +Can we not use the dp_packet_at_offset() function here (and the similarfunctions)? And include the above and below checks there?They are aimed at different purposes. dp_packet_at_offset() should walkthe chain of mbufs and return a pointer to the offset, while here we actually want to know if the offset is within the first mbuf. If theseheaders are not within the first mbuf, we don't return it as it crossesmbufs.
My bad, I was referring to dp_packet_at(), which probably also need the may_pull() if there are control protocols with jumbo packets?
return b->l3_ofs != UINT16_MAX ? (char *) dp_packet_data(b) + b->l3_ofs : NULL; @@ -341,6 +356,12 @@ dp_packet_set_l3(struct dp_packet *b, void *l3) static inline void * dp_packet_l4(const struct dp_packet *b) { +#ifdef DPDK_NETDEV + if (!dp_packet_mbuf_may_pull(b, b->l4_ofs)) { + return NULL; + } +#endif + return b->l4_ofs != UINT16_MAX ? (char *) dp_packet_data(b) + b->l4_ofs : NULL;@@ -355,10 +376,37 @@ dp_packet_set_l4(struct dp_packet *b, void *l4)static inline size_t dp_packet_l4_size(const struct dp_packet *b) { +#ifdef DPDK_NETDEV + if (b->l4_ofs == UINT16_MAX) { + return 0; + } ++ struct rte_mbuf *hmbuf = CONST_CAST(struct rte_mbuf *, &b->mbuf);+ struct rte_mbuf *tmbuf = dp_packet_tail(b); + + size_t l4_size = 0; + if (hmbuf == tmbuf) {Do not think this optimisation help much, as the dp_packet_tail(b) doesthe same walk as in the "while (hmbuf)" clause below, so why not just keet the else case?Yeah, good point. I'll change that in the next iteration.+ l4_size = hmbuf->data_len - b->l4_ofs; + } else { + l4_size = hmbuf->data_len - b->l4_ofs; + hmbuf = hmbuf->next; + + while (hmbuf) { + l4_size += tmbuf->data_len; + + hmbuf = hmbuf->next; + } + } + + l4_size -= dp_packet_l2_pad_size(b); + + return l4_size; +#else return b->l4_ofs != UINT16_MAX ? (const char *)dp_packet_tail(b) - (const char *)dp_packet_l4(b) - dp_packet_l2_pad_size(b) : 0; +#endif } static inline const void *@@ -491,7 +539,7 @@ __packet_set_data(struct dp_packet *b, uint16_t v)static inline uint16_t dp_packet_get_allocated(const struct dp_packet *b) { - return b->mbuf.buf_len; + return b->mbuf.nb_segs * b->mbuf.buf_len; } static inline void @@ -499,7 +547,107 @@ dp_packet_set_allocated(struct dp_packet *b, uint16_t s) { b->mbuf.buf_len = s; } + +static inline void * +dp_packet_at_offset(const struct dp_packet *b, size_t offset) +{ + if (b->source == DPBUF_DPDK) { + struct rte_mbuf *buf = CONST_CAST(struct rte_mbuf *, &b->mbuf); + + while (buf && offset > buf->data_len) { + offset -= buf->data_len; + + buf = buf->next; + } + return buf ? rte_pktmbuf_mtod_offset(buf, char *, offset) : NULL; + } else { + return (char *) dp_packet_data(b) + offset; + } +} ++/* If 'b' contains at least 'offset + size' bytes of data, returns apointer to + * byte 'offset'. Otherwise, returns a null pointer. */ +static inline void * +dp_packet_at(const struct dp_packet *b, size_t offset, size_t size) +{ + return offset + size <= dp_packet_size(b) + ? dp_packet_at_offset(b, offset) + : NULL; +} + +/* Returns a pointer to byte 'offset' in 'b', which must contain at least + * 'offset + size' bytes of data. */ +static inline void *+dp_packet_at_assert(const struct dp_packet *b, size_t offset, size_tsize) +{ + ovs_assert(offset + size <= dp_packet_size(b)); + return dp_packet_at_offset(b, offset); +} + +/* Returns a pointer to byte following the last byte of data in use in 'b'. */ +static inline void * +dp_packet_tail(const struct dp_packet *b) +{ + struct rte_mbuf *buf = CONST_CAST(struct rte_mbuf *, &b->mbuf); + if (b->source == DPBUF_DPDK) {+ /* Find last segment where data ends, meaning the tail of thechained + mbufs is there */ + buf = rte_pktmbuf_lastseg(buf); + } +What if the last two chains contains no user data, we should find the last buffer with data. Looks like we assume only the last in the chain contains room, and weachieve this by stripping off buffers (but we never add them again whenneeded, so something is wrong in this implementation).Does my reply to the cover letter help here as well? That's a validpoint, that if the size is decreased and mbufs are freed then mbufs arenot allocated again if size needs to be incremented. But as explained,this is to enforce that data is contiguous and so the tail points to thelast mbuf. I haven't yet encountered cases where this would lead to complications, but maybe you have something in mind?
I feel like we should not design the general APIs to force the current assumed limitation. As rte_pktmbuf_lastseg() is already walking the three, we could do it here instead and checking for the mailroom.
_______________________________________________ dev mailing list [email protected] https://mail.openvswitch.org/mailman/listinfo/ovs-dev
