DPDK-dev Archive on lore.kernel.org
 help / color / Atom feed
From: "Burakov, Anatoly" <anatoly.burakov@intel.com>
To: Yasufumi Ogawa <yasufum.o@gmail.com>,
	David Marchand <david.marchand@redhat.com>
Cc: "Ananyev, Konstantin" <konstantin.ananyev@intel.com>,
	dev <dev@dpdk.org>,
	 Yasufumi Ogawa <ogawa.yasufumi@lab.ntt.co.jp>
Subject: Re: [dpdk-dev] [PATCH v7 1/1] fbarray: fix duplicated fbarray file in secondary
Date: Mon, 2 Dec 2019 10:43:51 +0000
Message-ID: <a64bafdd-1836-f47d-cf41-e7d8ccb3bfab@intel.com> (raw)
In-Reply-To: <4144a7d7-f6af-8ebf-e572-d907e1975837@gmail.com>

On 29-Nov-19 5:44 AM, Yasufumi Ogawa wrote:
> Hi Anatoly,
> 
> On 2019/11/27 19:26, Burakov, Anatoly wrote:
>> On 26-Nov-19 7:40 PM, Yasufumi Ogawa wrote:
>>> Hi David,
>>>
>>> Sorry for slow reply.
>>>
>>> On 2019/11/14 21:27, David Marchand wrote:
>>>> On Thu, Nov 14, 2019 at 12:42 PM Yasufumi Ogawa 
>>>> <yasufum.o@gmail.com> wrote:
>>>>>
>>>>> On 2019/11/14 2:01, Burakov, Anatoly wrote:
>>>>>> On 13-Nov-19 9:43 PM, yasufum.o@gmail.com wrote:
>>>>>>> From: Yasufumi Ogawa <ogawa.yasufumi@lab.ntt.co.jp>
>>>>>>>
>>>>>>> In secondary_msl_create_walk(), it creates a file for fbarrays 
>>>>>>> with its
>>>>>>> PID for reserving unique name among secondary processes. However, it
>>>>>>> does not work if several secondaries run as app containers 
>>>>>>> because each
>>>>>>> of containerized secondary has PID 1, and failed to reserve 
>>>>>>> unique name
>>>>>>> other than first one. To reserve unique name in each of 
>>>>>>> containers, use
>>>>>>> hostname in addition to PID.
>>>>>>>
>>>>>>> Cc: stable@dpdk.org
>>>>>>>
>>>>>>> Signed-off-by: Yasufumi Ogawa <yasufum.o@gmail.com>
>>>>>>> ---
>>>>>>>    lib/librte_eal/linux/eal/eal_memalloc.c | 16 +++++++++++++---
>>>>>>>    1 file changed, 13 insertions(+), 3 deletions(-)
>>>>>>>
>>>>>>> diff --git a/lib/librte_eal/linux/eal/eal_memalloc.c
>>>>>>> b/lib/librte_eal/linux/eal/eal_memalloc.c
>>>>>>> index af6d0d023..11de6d4d6 100644
>>>>>>> --- a/lib/librte_eal/linux/eal/eal_memalloc.c
>>>>>>> +++ b/lib/librte_eal/linux/eal/eal_memalloc.c
>>>>>>> @@ -1365,6 +1365,12 @@ secondary_msl_create_walk(const struct
>>>>>>> rte_memseg_list *msl,
>>>>>>>        struct rte_memseg_list *primary_msl, *local_msl;
>>>>>>>        char name[PATH_MAX];
>>>>>>>        int msl_idx, ret;
>>>>>>> +    char hostname[HOST_NAME_MAX+1] = { 0 };
>>>>>>> +    /* filename of secondary's fbarray is defined such as
>>>>>>> +     * "fbarray_memseg-1048576k-0-0_PID_HOSTNAME" and length of PID
>>>>>>> +     * can be 7 digits maximumly.
>>>>>>> +     */
>>>>>>> +    int fbarray_sec_name_len = 32 + 7 + 1 + HOST_NAME_MAX + 1;
>>>>>>
>>>>>> What does 32 stand for? Maybe #define both 32 and 7 values?
>>>>> Hi Anatoly,
>>>>>
>>>>> Thank you for your comments! If my understanding is correct, the 
>>>>> prefix
>>>>> "fbarray_memseg-1048576k-0-0_" is 28 digits and it could be larger if
>>>>> using the size of hugepage or the number of NUMA nodes are larger
>>>>> possibly. However, I think 32 digits is still enough.
>>>>>
>>>>>   > Maybe #define both 32 and 7 values?
>>>>> Yes. I think it should be better to use #define if this values are
>>>>> referred several times.
>>>>
>>>>
>>>> We can truncate to RTE_FBARRAY_NAME_LEN in all cases.
>>>> And iiuc, rte_fbarray_init will refuse any longer name anyway.
>>> Could I confirm the issue? I've understood that it is failed to 
>>> validate the name of fbarray in fully_validate() at 
>>> "lib/librte_eal/common/eal_common_fbarray.c:697".
>>>
>>> static int
>>> fully_validate(const char *name, unsigned int elt_sz, unsigned int len)
>>> {
>>>          if (name == NULL || elt_sz == 0 || len == 0 || len > INT_MAX) {
>>>                  rte_errno = EINVAL;
>>>                  return -1;
>>>          }
>>>
>>>          if (strnlen(name, RTE_FBARRAY_NAME_LEN) == 
>>> RTE_FBARRAY_NAME_LEN) {
>>>                  rte_errno = ENAMETOOLONG;
>>>                  return -1;
>>>          }
>>>          return 0;
>>> }
>>>
>>> I should overwrite the definition of RTE_FBARRAY_NAME_LEN as previous 
>>> patch in this case, and it causes an ABI breakage, right? If so, I 
>>> would like to make the change and give up to update stable release.
>>>
>>> Thanks,
>>> Yasufumi
>>>
>>
>> It seems we're getting into bikeshedding...
>>
>> We can do this without ABI breakage. You could have just used 
>> RTE_FBARRAY_NAME_LEN as max fbarray name length for 
>> fbarray_sec_name_len (i.e. that would include hostname + pid + 
>> whatever else there is). The name, as David has pointed out, would be 
>> truncated to RTE_FBARRAY_NAME_LEN anyway (or, more precisely, it will 
>> be refused if it's longer than that), so this is the most you can have 
>> - so you can just use that as the maximum.
> I sent v8 patch to change the size of RTE_FBARRAY_NAME_LEN itself to be 
> allowed the size of secondary's fbarray over 64 bytes. I appreciate if 
> you agree that.
> 

Why not just limit the name to RTE_FBARRAY_NAME_LEN instead of changing 
the definition of RTE_FBARRAY_NAME_LEN?

One the other hand, technically, fbarray API is experimental. The only 
structure that uses rte_fbarray is rte_memseg_list, but API's using 
either rte_fbarray or rte_memseg_list are either internal (memory/VFIO 
subsystem), or are marked as experimental (walk functions).

So i *think* we're actually OK with changing the length of 
RTE_FBARRAY_NAME_LEN as far as ABI policy goes: nothing in the stable 
ABI gets affected. David, thoughts?

(i think it's probably time to make experimental memory/fbarray stuff 
stable, but that's a different conversation...)

> Thanks,
> Yasufumi
> 


-- 
Thanks,
Anatoly

  reply index

Thread overview: 62+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-04-16  1:59 [dpdk-dev] [PATCH] fbarray: get fbarrays from containerized secondary ogawa.yasufumi
2019-04-16  3:43 ` [dpdk-dev] [PATCH v2 0/1] Get " ogawa.yasufumi
2019-04-16  3:43   ` [dpdk-dev] [PATCH v2 1/1] fbarray: get " ogawa.yasufumi
2019-07-04 20:17     ` Thomas Monjalon
2019-07-05  8:53     ` Burakov, Anatoly
2019-07-09 10:22       ` Yasufumi Ogawa
2019-07-09 10:24         ` Burakov, Anatoly
2019-07-09 10:26           ` Burakov, Anatoly
2019-07-11  9:37             ` Yasufumi Ogawa
2019-07-11  9:43               ` Burakov, Anatoly
2019-07-11 10:31 ` [dpdk-dev] [PATCH v3 0/1] " yasufum.o
2019-07-11 10:31   ` [dpdk-dev] [PATCH v3 1/1] " yasufum.o
2019-07-11 10:53     ` Burakov, Anatoly
2019-07-11 11:57       ` Yasufumi Ogawa
2019-07-11 13:14         ` Burakov, Anatoly
2019-07-12  2:22           ` Yasufumi Ogawa
2019-07-22  1:06             ` Ogawa Yasufumi
2019-07-22  9:33               ` Burakov, Anatoly
2019-07-22  9:25             ` Burakov, Anatoly
2019-07-24  8:20   ` [dpdk-dev] [PATCH v4 0/1] " yasufum.o
2019-07-24  8:20     ` [dpdk-dev] [PATCH v4 1/1] " yasufum.o
2019-07-24  9:59       ` Burakov, Anatoly
2019-07-30  8:16         ` [dpdk-dev] [dpdk-stable] " Thomas Monjalon
2019-07-30  9:18           ` Burakov, Anatoly
2019-07-31  5:48             ` Yasufumi Ogawa
2019-10-11  9:36       ` [dpdk-dev] " David Marchand
2019-10-25 15:36         ` David Marchand
2019-10-25 19:54           ` Yasufumi Ogawa
2019-10-26 16:15             ` David Marchand
2019-10-26 18:11               ` Yasufumi Ogawa
2019-10-28  8:07     ` [dpdk-dev] [PATCH v5 0/1] fbarray: fix duplicated fbarray file in secondary yasufum.o
2019-10-28  8:07       ` [dpdk-dev] [PATCH v5 1/1] " yasufum.o
2019-10-29 12:03         ` Ananyev, Konstantin
2019-10-30 13:42           ` Yasufumi Ogawa
2019-10-30 19:00             ` Ananyev, Konstantin
2019-10-31 10:03               ` Yasufumi Ogawa
2019-10-31 10:32                 ` Ananyev, Konstantin
2019-11-01  9:04     ` [dpdk-dev] [PATCH v6 0/1] " yasufum.o
2019-11-01  9:04       ` [dpdk-dev] [PATCH v6 1/1] " yasufum.o
2019-11-01 12:01         ` Ananyev, Konstantin
2019-11-04 10:20         ` Burakov, Anatoly
2019-11-05 10:13           ` David Marchand
2019-11-05 11:31             ` Burakov, Anatoly
2019-11-05 11:41               ` Ananyev, Konstantin
2019-11-06 10:37                 ` Burakov, Anatoly
2019-11-08  3:19                   ` Yasufumi Ogawa
2019-11-13 21:43     ` [dpdk-dev] [PATCH v7 0/1] " yasufum.o
2019-11-13 21:43       ` [dpdk-dev] [PATCH v7 1/1] " yasufum.o
2019-11-14 10:01         ` Burakov, Anatoly
2019-11-14 11:42           ` Yasufumi Ogawa
2019-11-14 12:27             ` David Marchand
2019-11-26 19:40               ` Yasufumi Ogawa
2019-11-27 10:26                 ` Burakov, Anatoly
2019-11-29  5:44                   ` Yasufumi Ogawa
2019-12-02 10:43                     ` Burakov, Anatoly [this message]
2019-12-05 20:13                       ` Yasufumi Ogawa
2019-11-14 12:55         ` David Marchand
2019-11-14 17:32         ` Ananyev, Konstantin
2019-11-27  8:48       ` [dpdk-dev] [PATCH v8 0/1] " Yasufumi Ogawa
2019-11-27  8:48         ` [dpdk-dev] [PATCH v8 1/1] " Yasufumi Ogawa
2019-12-06 10:44           ` Burakov, Anatoly
2019-12-06 13:18             ` Yasufumi Ogawa

Reply instructions:

You may reply publically 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=a64bafdd-1836-f47d-cf41-e7d8ccb3bfab@intel.com \
    --to=anatoly.burakov@intel.com \
    --cc=david.marchand@redhat.com \
    --cc=dev@dpdk.org \
    --cc=konstantin.ananyev@intel.com \
    --cc=ogawa.yasufumi@lab.ntt.co.jp \
    --cc=yasufum.o@gmail.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-dev Archive on lore.kernel.org

Archives are clonable:
	git clone --mirror https://lore.kernel.org/dpdk-dev/0 dpdk-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 dpdk-dev dpdk-dev/ https://lore.kernel.org/dpdk-dev \
		dev@dpdk.org
	public-inbox-index dpdk-dev

Example config snippet for mirrors

Newsgroup available over NNTP:
	nntp://nntp.lore.kernel.org/org.dpdk.dev


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