DPDK patches and discussions
 help / color / mirror / Atom feed
From: Olivier Matz <olivier.matz@6wind.com>
To: Stephen Hemminger <stephen@networkplumber.org>
Cc: dev@dpdk.org, stable@dpdk.org
Subject: Re: [dpdk-dev] [PATCH v3] mbuf: fix dynamic flags lookup from secondary process
Date: Mon, 26 Oct 2020 11:39:35 +0100	[thread overview]
Message-ID: <20201026103935.GL1898@platinum> (raw)
In-Reply-To: <20201024004331.25043-1-stephen@networkplumber.org>

Hi Stephen,

On Fri, Oct 23, 2020 at 05:43:31PM -0700, Stephen Hemminger wrote:
> The dynamic flag management is broken if rte_mbuf_dynflag_lookup()
> is done in a secondary process because the local pointer to
> the memzone is not ever initialized.
> 
> Fix it by using the same checks as dynfield_register().
> I.e if shared memory zone has not been looked up already,
> then discover it.
> 
> Fixes: 4958ca3a443a ("mbuf: support dynamic fields and flags")
> Cc: olivier.matz@6wind.com
> Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
> ---
> 
> v3 - change title, fix one extra whitespace 
> 
>  lib/librte_mbuf/rte_mbuf_dyn.c | 20 ++++++++------------
>  1 file changed, 8 insertions(+), 12 deletions(-)
> 
> diff --git a/lib/librte_mbuf/rte_mbuf_dyn.c b/lib/librte_mbuf/rte_mbuf_dyn.c
> index 538a43f6959f..554ec5a1ca4f 100644
> --- a/lib/librte_mbuf/rte_mbuf_dyn.c
> +++ b/lib/librte_mbuf/rte_mbuf_dyn.c
> @@ -185,13 +185,11 @@ rte_mbuf_dynfield_lookup(const char *name, struct rte_mbuf_dynfield *params)
>  {
>  	struct mbuf_dynfield_elt *mbuf_dynfield;
>  
> -	if (shm == NULL) {
> -		rte_errno = ENOENT;
> -		return -1;
> -	}
> -
>  	rte_mcfg_tailq_read_lock();
> -	mbuf_dynfield = __mbuf_dynfield_lookup(name);
> +	if (shm == NULL && init_shared_mem() < 0)
> +		mbuf_dynfield = NULL;
> +	else
> +		mbuf_dynfield = __mbuf_dynfield_lookup(name);
>  	rte_mcfg_tailq_read_unlock();
>  
>  	if (mbuf_dynfield == NULL) {
>  		rte_errno = ENOENT;
>  		return -1;

There is still a small corner case here: on a primary process,
init_shared_mem() can return -1 in case rte_memzone_reserve_aligned()
returns a NULL memzone. In this situation, rte_errno is set by the
memzone layer by overriden to ENOENT.

Maybe something like this is better, what do you think?

@@ -172,7 +172,7 @@ __mbuf_dynfield_lookup(const char *name)
 			break;
 	}
 
-	if (te == NULL) {
+	if (te == NULL || mbuf_dynfield == NULL) {
 		rte_errno = ENOENT;
 		return NULL;
 	}
@@ -185,19 +185,15 @@ rte_mbuf_dynfield_lookup(const char *name, struct rte_mbuf_dynfield *params)
 {
 	struct mbuf_dynfield_elt *mbuf_dynfield;
 
-	if (shm == NULL) {
-		rte_errno = ENOENT;
-		return -1;
-	}
-
 	rte_mcfg_tailq_read_lock();
-	mbuf_dynfield = __mbuf_dynfield_lookup(name);
+	if (shm == NULL && init_shared_mem() < 0)
+		mbuf_dynfield = NULL;
+	else
+		mbuf_dynfield = __mbuf_dynfield_lookup(name);
 	rte_mcfg_tailq_read_unlock();
 
-	if (mbuf_dynfield == NULL) {
-		rte_errno = ENOENT;
+	if (mbuf_dynfield == NULL)
 		return -1;
-	}
 
 	if (params != NULL)
 		memcpy(params, &mbuf_dynfield->params, sizeof(*params));



Thanks,
Olivier


> @@ -384,13 +382,11 @@ rte_mbuf_dynflag_lookup(const char *name,
>  {
>  	struct mbuf_dynflag_elt *mbuf_dynflag;
>  
> -	if (shm == NULL) {
> -		rte_errno = ENOENT;
> -		return -1;
> -	}
> -
>  	rte_mcfg_tailq_read_lock();
> -	mbuf_dynflag = __mbuf_dynflag_lookup(name);
> +	if (shm == NULL && init_shared_mem() < 0)
> +		mbuf_dynflag = NULL;
> +	else
> +		mbuf_dynflag = __mbuf_dynflag_lookup(name);
>  	rte_mcfg_tailq_read_unlock();
>  
>  	if (mbuf_dynflag == NULL) {
> -- 
> 2.27.0
> 

  reply	other threads:[~2020-10-26 10:39 UTC|newest]

Thread overview: 12+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-10-15 17:20 [dpdk-dev] [PATCH] mbuf: allow dynamic flags to be used by " Stephen Hemminger
2020-10-20 12:18 ` Olivier Matz
2020-10-20 20:58 ` [dpdk-dev] [PATCH v2] " Stephen Hemminger
2020-10-24  0:43 ` [dpdk-dev] [PATCH v3] mbuf: fix dynamic flags lookup from " Stephen Hemminger
2020-10-26 10:39   ` Olivier Matz [this message]
2020-10-26 14:49     ` Stephen Hemminger
2020-11-03 21:02       ` Thomas Monjalon
2020-11-04  5:53   ` [dpdk-dev] [PATCH v4] mbuf: allow dynamic flags to be used by " Stephen Hemminger
2020-11-04  8:17     ` Olivier Matz
2020-11-04 16:20   ` [dpdk-dev] [PATCH v5] " Stephen Hemminger
2020-11-04 16:26     ` Olivier Matz
2020-11-04 17:41       ` David Marchand

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=20201026103935.GL1898@platinum \
    --to=olivier.matz@6wind.com \
    --cc=dev@dpdk.org \
    --cc=stable@dpdk.org \
    --cc=stephen@networkplumber.org \
    /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
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).