DPDK patches and discussions
 help / color / mirror / Atom feed
From: "Ananyev, Konstantin" <konstantin.ananyev@intel.com>
To: yang_y_yi <yang_y_yi@163.com>
Cc: "dev@dpdk.org" <dev@dpdk.org>, "Hu, Jiayu" <jiayu.hu@intel.com>,
	"techboard@dpdk.org" <techboard@dpdk.org>,
	"thomas@monjalon.net" <thomas@monjalon.net>,
	"yangyi01@inspur.com" <yangyi01@inspur.com>
Subject: Re: [dpdk-dev] [PATCH v2] gso: fix free issue of mbuf gso segments attach to
Date: Fri, 23 Oct 2020 14:46:42 +0000
Message-ID: <BYAPR11MB3301CB4522A5FF243B1FFFDE9A1A0@BYAPR11MB3301.namprd11.prod.outlook.com> (raw)
In-Reply-To: <3dcc6d13.74bb.175559c0cf4.Coremail.yang_y_yi@163.com>

> From: yang_y_yi <yang_y_yi@163.com>
> Sent: Friday, October 23, 2020 2:18 PM
> To: Ananyev, Konstantin <konstantin.ananyev@intel.com>
> Cc: dev@dpdk.org; Hu, Jiayu <jiayu.hu@intel.com>; techboard@dpdk.org; thomas@monjalon.net; yangyi01@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.ananyev@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:yangyi01@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.
 
> 
> >
> >>  			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
> 
> 
>

  reply	other threads:[~2020-10-23 14:47 UTC|newest]

Thread overview: 12+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-10-22  6:51 yang_y_yi
2020-10-22 13:16 ` Ananyev, Konstantin
2020-10-22 15:33   ` Ananyev, Konstantin
2020-10-23  0:57     ` Hu, Jiayu
2020-10-23 13:23       ` yang_y_yi
2020-10-23 13:21     ` yang_y_yi
2020-10-23 13:18   ` yang_y_yi
2020-10-23 14:46     ` Ananyev, Konstantin [this message]
2020-10-26  0:57       ` yang_y_yi
2020-10-26  2:06         ` Jiayu Hu
2020-10-26  2:12           ` yang_y_yi
2020-10-26  6:16             ` Jiayu Hu

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=BYAPR11MB3301CB4522A5FF243B1FFFDE9A1A0@BYAPR11MB3301.namprd11.prod.outlook.com \
    --to=konstantin.ananyev@intel.com \
    --cc=dev@dpdk.org \
    --cc=jiayu.hu@intel.com \
    --cc=techboard@dpdk.org \
    --cc=thomas@monjalon.net \
    --cc=yang_y_yi@163.com \
    --cc=yangyi01@inspur.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link

DPDK patches and discussions

This inbox may be cloned and mirrored by anyone:

	git clone --mirror https://inbox.dpdk.org/dev/0 dev/git/0.git

	# If you have public-inbox 1.1+ installed, you may
	# initialize and index your mirror using the following commands:
	public-inbox-init -V2 dev dev/ https://inbox.dpdk.org/dev \
		dev@dpdk.org
	public-inbox-index dev

Example config snippet for mirrors.
Newsgroup available over NNTP:
	nntp://inbox.dpdk.org/inbox.dpdk.dev


AGPL code for this site: git clone https://public-inbox.org/public-inbox.git