At 2020-10-23 22:46:42, "Ananyev, Konstantin" <konstantin.anan...@intel.com>
wrote:
>> From: yang_y_yi <yang_y...@163.com>
>> Sent: Friday, October 23, 2020 2:18 PM
>> To: Ananyev, Konstantin <konstantin.anan...@intel.com>
>> Cc: dev@dpdk.org; Hu, Jiayu <jiayu...@intel.com>; techbo...@dpdk.org;
>> tho...@monjalon.net; yangy...@inspur.com
>> Subject: Re:RE: [PATCH v2] gso: fix free issue of mbuf gso segments attach to
>>
>> Konstantin, thank you so much for comments, my replies inline, please check
>> them.
>> At 2020-10-22 21:16:43, "Ananyev, Konstantin"
>> <mailto:konstantin.anan...@intel.com> wrote:
>> >
>> >>
>> >> rte_gso_segment decreased refcnt of pkt by one, but
>> >> it is wrong if pkt is external mbuf, pkt won't be
>> >> freed because of incorrect refcnt, the result is
>> >> application can't allocate mbuf from mempool because
>> >> mbufs in mempool are run out of.
>> >>
>> >> One correct way is application should call
>> >> rte_pktmbuf_free after calling rte_gso_segment to free
>> >> pkt explicitly. rte_gso_segment mustn't handle it, this
>> >> should be responsibility of application.
>> >
>> >Probably needs to be stated clearly:
>> >It is a change in functional behaviour.
>> >Without deprecation note in advance.
>>
>> Ok, I'll add such statement in next version.
>>
>> >TB members: please provide your opinion on that patch.
>> >
>> >>
>> >> Fixes: 119583797b6a ("gso: support TCP/IPv4 GSO")
>> >> Signed-off-by: Yi Yang <mailto:yangy...@inspur.com>
>> >> ---
>> >> Changelog:
>> >>
>> >> v1->v2:
>> >> - update description of rte_gso_segment().
>> >> - change code which calls rte_gso_segment() to
>> >> fix free issue.
>> >>
>> >> ---
>> >> app/test-pmd/csumonly.c | 3 ++-
>> >> doc/guides/prog_guide/generic_segmentation_offload_lib.rst | 7 +++++--
>> >
>> >I think release notes also have to be updated.
>>
>> Ok, also will update it to reflect this change.
>>
>> >
>> >> lib/librte_gso/rte_gso.c | 9 +--------
>> >> lib/librte_gso/rte_gso.h | 7 +++++--
>> >> 4 files changed, 13 insertions(+), 13 deletions(-)
>> >>
>> >> diff --git a/app/test-pmd/csumonly.c b/app/test-pmd/csumonly.c
>> >> index 3d7d244..829e07f 100644
>> >> --- a/app/test-pmd/csumonly.c
>> >> +++ b/app/test-pmd/csumonly.c
>> >> @@ -1080,11 +1080,12 @@ struct simple_gre_hdr {
>> >> ret = rte_gso_segment(pkts_burst[i], gso_ctx,
>> >> &gso_segments[nb_segments],
>> >> GSO_MAX_PKT_BURST - nb_segments);
>> >> + /* pkts_burst[i] can be freed safely here. */
>> >> + rte_pktmbuf_free(pkts_burst[i]);
>> >
>> >It doesn't look correct to me.
>> >I think it should be:
>> >If (ret > 1) rte_pktmbuf_free(pkts_burst[i]);
>>
>> No, in original implementation, if gso failed, application will free it,
>> otherwise rte_gso_segment will free it (i.e. refcnt update -1 in
>> rte_gso_segment), this change will change previous behavior. application
>> will free it for both cases.
>
>
>That's the point - with current implementation:
>If ret == 1, then you shouldn't free input packet.
>Because in that case:
>input_pkt == output_pkt[0]
>
>And if you'll free it, you can't use it after it.
>In that particular case, you can't TX it.
I checked gso code again, there are two cases even if ret == 1, one case is it
isn't segmented, the other is it is segmented but gso_do_segment returns 1, for
case #1, we can handle it as you said, but for case #2, we can't handle it as
you said because it has been segmented in fact. So I think we should return 0
foe case #1 and don't do assignment "pkts_out[0] = pkt;", we should handle
case #2 as before, right?
>
>>
>> >
>> >> if (ret >= 0)
>> >> nb_segments += ret;
>> >> else {
>> >> TESTPMD_LOG(DEBUG, "Unable to segment packet");
>> >> - rte_pktmbuf_free(pkts_burst[i]);
>> >> }
>> >> }
>> >
>> >
>> >About drivers/net/tap/rte_eth_tap.c:
>> >I think it has to be modified too, as here:
>> >
>> >/* free original mbuf */
>> > rte_pktmbuf_free(mbuf_in);
>> > /* free tso mbufs */
>> > if (num_tso_mbufs > 0)
>> > rte_pktmbuf_free_bulk(mbuf, num_tso_mbufs);
>> >
>> >if mbuf[0] == mbuf_in
>> >Will have a double free() for the same mbuf.
>>
>> Here I'm not clear, is this code ok before I change? It looks like there is
>> same issue there before.
>>
>> >
>> >>
>> >> diff --git a/doc/guides/prog_guide/generic_segmentation_offload_lib.rst
>> b/doc/guides/prog_guide/generic_segmentation_offload_lib.rst
>> >> index 205cb8a..8577572 100644
>> >> --- a/doc/guides/prog_guide/generic_segmentation_offload_lib.rst
>> >> +++ b/doc/guides/prog_guide/generic_segmentation_offload_lib.rst
>> >> @@ -25,8 +25,9 @@ Bearing that in mind, the GSO library enables DPDK
>> >> applications to segment
>> >> packets in software. Note however, that GSO is implemented as a
>> >> standalone
>> >> library, and not via a 'fallback' mechanism (i.e. for when TSO is
>> >> unsupported
>> >> in the underlying hardware); that is, applications must explicitly
>> >> invoke the
>> >> -GSO library to segment packets. The size of GSO segments ``(segsz)`` is
>> >> -configurable by the application.
>> >> +GSO library to segment packets, they also must call
>> >> ``rte_pktmbuf_free()`` to
>> >> +free mbuf GSO segments attach to after calling ``rte_gso_segment()``.
>> >
>> >Probably worth to mention that if return code == 1, then
>> >output mbuf will point to input mbuf and extra care with free() is required.
>>
>> Ok.
>>
>> >
>> >> The size
>> >> +of GSO segments ``(segsz)`` is configurable by the application.
>> >>
>> >> Limitations
>> >> -----------
>> >> @@ -233,6 +234,8 @@ To segment an outgoing packet, an application must:
>> >>
>> >> #. Invoke the GSO segmentation API, ``rte_gso_segment()``.
>> >>
>> >> +#. Call ``rte_pktmbuf_free()`` to free mbuf ``rte_gso_segment()``
>> >> segments.
>> >> +
>> >> #. If required, update the L3 and L4 checksums of the newly-created
>> >> segments.
>> >> For tunneled packets, the outer IPv4 headers' checksums should also be
>> >> updated. Alternatively, the application may offload checksum
>> >> calculation
>> >> diff --git a/lib/librte_gso/rte_gso.c b/lib/librte_gso/rte_gso.c
>> >> index 751b5b6..0d6cae5 100644
>> >> --- a/lib/librte_gso/rte_gso.c
>> >> +++ b/lib/librte_gso/rte_gso.c
>> >> @@ -30,7 +30,6 @@
>> >> uint16_t nb_pkts_out)
>> >> {
>> >> struct rte_mempool *direct_pool, *indirect_pool;
>> >> - struct rte_mbuf *pkt_seg;
>> >> uint64_t ol_flags;
>> >> uint16_t gso_size;
>> >> uint8_t ipid_delta;
>> >> @@ -80,13 +79,7 @@
>> >> return 1;
>> >> }
>> >>
>> >> - if (ret > 1) {
>> >> - pkt_seg = pkt;
>> >> - while (pkt_seg) {
>> >> - rte_mbuf_refcnt_update(pkt_seg, -1);
>> >> - pkt_seg = pkt_seg->next;
>> >> - }
>> >> - } else if (ret < 0) {
>> >> + if (ret < 0) {
>> >> /* Revert the ol_flags in the event of failure. */
>> >> pkt->ol_flags = ol_flags;
>> >> }
>> >> diff --git a/lib/librte_gso/rte_gso.h b/lib/librte_gso/rte_gso.h
>> >> index 3aab297..f6694f9 100644
>> >> --- a/lib/librte_gso/rte_gso.h
>> >> +++ b/lib/librte_gso/rte_gso.h
>> >> @@ -89,8 +89,11 @@ struct rte_gso_ctx {
>> >> * the GSO segments are sent to should support transmission of
>> >> multi-segment
>> >> * packets.
>> >> *
>> >> - * If the input packet is GSO'd, its mbuf refcnt reduces by 1. Therefore,
>> >> - * when all GSO segments are freed, the input packet is freed
>> >> automatically.
>> >> + * If the input packet is GSO'd, all the indirect segments are attached
>> >> to the
>> >> + * input packet.
>> >> + *
>> >> + * rte_gso_segment() will not free the input packet no matter whether it
>> >> is
>> >> + * GSO'd or not, the application should free it after call
>> >> rte_gso_segment().
>> >> *
>> >> * If the memory space in pkts_out or MBUF pools is insufficient, this
>> >> * function fails, and it returns (-1) * errno. Otherwise, GSO succeeds,
>> >> --
>> >> 1.8.3.1
>>
>>
>>
