From mboxrd@z Thu Jan 1 00:00:00 1970 From: santosh Subject: Re: [RFC v2 01/17] mempool: fix phys contig check if populate default skipped Date: Thu, 1 Feb 2018 16:03:32 +0530 Message-ID: <44211f68-9a31-8025-afd4-43f482bd4e87@caviumnetworks.com> References: <1511539591-20966-1-git-send-email-arybchenko@solarflare.com> <1516713372-10572-1-git-send-email-arybchenko@solarflare.com> <1516713372-10572-2-git-send-email-arybchenko@solarflare.com> <20180131164504.cnfgfwo2x3ftxnaj@platinum> <90005aad-1508-44bc-92a4-7c5d5f7b0246@caviumnetworks.com> <53796b3f-b2c2-dcfc-a10b-71ce1a68d3d2@caviumnetworks.com> <618e4eb2-e66c-592c-2065-f86328916d5d@solarflare.com> <20180201101423.k5yqxd5a2dk55qqw@platinum> Mime-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: 8bit Cc: dev@dpdk.org, stable@dpdk.org, jerin.jacob@caviumnetworks.com To: Olivier Matz , Andrew Rybchenko Return-path: In-Reply-To: <20180201101423.k5yqxd5a2dk55qqw@platinum> Content-Language: en-US List-Id: DPDK patches and discussions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: dev-bounces@dpdk.org Sender: "dev" On Thursday 01 February 2018 03:44 PM, Olivier Matz wrote: > On Thu, Feb 01, 2018 at 01:00:12PM +0300, Andrew Rybchenko wrote: >> On 02/01/2018 12:30 PM, santosh wrote: >>> On Thursday 01 February 2018 02:48 PM, Andrew Rybchenko wrote: >>>> On 02/01/2018 12:09 PM, santosh wrote: >>>>> On Thursday 01 February 2018 12:24 PM, Andrew Rybchenko wrote: >>>>>> On 02/01/2018 08:05 AM, santosh wrote: >>>>>>> On Wednesday 31 January 2018 10:15 PM, Olivier Matz wrote: >>>>>>>> On Tue, Jan 23, 2018 at 01:15:56PM +0000, Andrew Rybchenko wrote: >>>>>>>>> There is not specified dependency between rte_mempool_populate_default() >>>>>>>>> and rte_mempool_populate_iova(). So, the second should not rely on the >>>>>>>>> fact that the first adds capability flags to the mempool flags. >>>>>>>>> >>>>>>>>> Fixes: 65cf769f5e6a ("mempool: detect physical contiguous objects") >>>>>>>>> Cc: stable@dpdk.org >>>>>>>>> >>>>>>>>> Signed-off-by: Andrew Rybchenko >>>>>>>> Looks good to me. I agree it's strange that the mp->flags are >>>>>>>> updated with capabilities only in rte_mempool_populate_default(). >>>>>>>> I see that this behavior is removed later in the patchset since the >>>>>>>> get_capa() is removed! >>>>>>>> >>>>>>>> However maybe this single patch could go in 18.02. >>>>>>>> +Santosh +Jerin since it's mostly about Octeon. >>>>>>> rte_mempool_xmem_size should return correct size if MEMPOOL_F_CAPA_BLK_ALIGNED_OBJECTS flag >>>>>>> is set in 'mp->flags'. Thats why _ops_get_capabilities() called in _populate_default() but not >>>>>>> at _populate_iova(). >>>>>>> I think, this 'alone' patch may break octeontx mempool. >>>>>> The patch does not touch rte_mempool_populate_default(). >>>>>> _ops_get_capabilities() is still called there before >>>>>> rte_mempool_xmem_size(). The theoretical problem which >>>>>> the patch tries to fix is the case when >>>>>> rte_mempool_populate_default() is not called at all. I.e. application >>>>>> calls _ops_get_capabilities() to get flags, then, together with >>>>>> mp->flags, calls rte_mempool_xmem_size() directly, allocates >>>>>> calculated amount of memory and calls _populate_iova(). >>>>>> >>>>> In that case, Application does like below: >>>>> >>>>>     /* Get mempool capabilities */ >>>>>     mp_flags = 0; >>>>>     ret = rte_mempool_ops_get_capabilities(mp, &mp_flags); >>>>>     if ((ret < 0) && (ret != -ENOTSUP)) >>>>>         return ret; >>>>> >>>>>     /* update mempool capabilities */ >>>>>     mp->flags |= mp_flags; >>>> Above line is not mandatory. "mp->flags | mp_flags" could be simply >>>> passed to  rte_mempool_xmem_size() below. >>>> >>> That depends and again upto application requirement, if app further down >>> wants to refer mp->flags for _align/_contig then better update to mp->flags. >>> >>> But that wasn't the point of discussion, I'm trying to understand that >>> w/o this patch, whats could be the application level problem? >> The problem that it is fragile. If application does not use >> rte_mempool_populate_default() it has to care about addition >> of mempool capability flags into mempool flags. If it is not done, >> rte_mempool_populate_iova/virt/iova_tab() functions will work >> incorrectly since F_CAPA_PHYS_CONTIG and >> F_CAPA_BLK_ALIGNED_OBJECTS are missing. >> >> The idea of the patch is to make it a bit more robust. I have no >> idea how it can break something. If capability flags are already >> there - no problem. If no, just make sure that we locally have them. > The example given by Santosh will work, but it is *not* the role of the > application to update the mempool flags. And nothing says that it is mandatory > to call rte_mempool_ops_get_capabilities() before the populate functions. > > For instance, in testpmd it calls rte_mempool_populate_anon() when using > anonymous memory. The capabilities will never be updated in mp->flags. Valid case and I agree with your example and explanation. With nits change: mp->flags |= mp_capa_flags; Acked-by: Santosh Shukla