DPDK patches and discussions
 help / color / mirror / Atom feed
From: "Michał Krawczyk" <mk@semihalf.com>
To: Ferruh Yigit <ferruh.yigit@intel.com>
Cc: dev <dev@dpdk.org>, "Dagan, Noam" <ndagan@amazon.com>,
	"Tzalik, Guy" <gtzalik@amazon.com>,
	 "Chauskin, Igor" <igorch@amazon.com>,
	upstream@semihalf.com,  Stanislaw Kardach <kda@semihalf.com>,
	Shay Agroskin <shayagr@amazon.com>
Subject: Re: [dpdk-dev] [PATCH v3 19/22] net/ena: make ethdev references smp safe
Date: Sun, 9 May 2021 16:40:28 +0200	[thread overview]
Message-ID: <CAJMMOfMFkPcF3C_a5_PYb9JE7ppVddjNinWWAWKsOLadRabXHQ@mail.gmail.com> (raw)
In-Reply-To: <53b53264-f440-a377-6a78-d8a179754045@intel.com>

pt., 7 maj 2021 o 18:49 Ferruh Yigit <ferruh.yigit@intel.com> napisał(a):
>
> On 5/6/2021 3:25 PM, Michal Krawczyk wrote:
> > From: Stanislaw Kardach <kda@semihalf.com>
> >
> > rte_pci_device and rte_eth_dev are process-local structures. Therefore
> > ena_adapter::pdev and ena_adapter::rte_dev cannot be used universally.
> > Switch this to extracting those structures via rte_eth_devices indexing
> > and remove pdev since it's not used outside of init.
> >
> > Signed-off-by: Stanislaw Kardach <kda@semihalf.com>
> > Reviewed-by: Michal Krawczyk <mk@semihalf.com>
> > Reviewed-by: Igor Chauskin <igorch@amazon.com>
> > Reviewed-by: Shay Agroskin <shayagr@amazon.com>
>
> <...>
>
> > @@ -1634,7 +1633,7 @@ static void ena_timer_wd_callback(__rte_unused struct rte_timer *timer,
> >                                 void *arg)
> >  {
> >       struct ena_adapter *adapter = arg;
> > -     struct rte_eth_dev *dev = adapter->rte_dev;
> > +     struct rte_eth_dev *dev = &rte_eth_devices[adapter->port_id];
> >
>
> I think better to try to avoid using 'rte_eth_devices' global variable as much
> as possible, although it may not be possible always, but for this case it seems
> it can be done.
>
> Why not just pass "struct rte_eth_dev" as 'arg', instead of "struct ena_adapter"?

Good suggestion, it will be changed in the v4.

>
> >       check_for_missing_keep_alive(adapter);
> >       check_for_admin_com_state(adapter);
> > @@ -1819,11 +1818,10 @@ static int eth_ena_dev_init(struct rte_eth_dev *eth_dev)
> >       memset(adapter, 0, sizeof(struct ena_adapter));
> >       ena_dev = &adapter->ena_dev;
> >
> > -     adapter->rte_eth_dev_data = eth_dev->data;
> > -     adapter->rte_dev = eth_dev;
> > +     adapter->edev_data = eth_dev->data;
> > +     adapter->port_id = eth_dev->data->port_id;
> >
> >       pci_dev = RTE_ETH_DEV_TO_PCI(eth_dev);
> > -     adapter->pdev = pci_dev;
> >
> >       PMD_INIT_LOG(INFO, "Initializing %x:%x:%x.%d",
> >                    pci_dev->addr.domain,
> > @@ -1843,7 +1841,8 @@ static int eth_ena_dev_init(struct rte_eth_dev *eth_dev)
> >       }
> >
> >       ena_dev->reg_bar = adapter->regs;
> > -     ena_dev->dmadev = adapter->pdev;
> > +     /* This is a dummy pointer for ena_com functions. */
> > +     ena_dev->dmadev = adapter;
> >
> >       adapter->id_number = adapters_found;
> >
> > @@ -1857,7 +1856,7 @@ static int eth_ena_dev_init(struct rte_eth_dev *eth_dev)
> >       }
> >
> >       /* device specific initialization routine */
> > -     rc = ena_device_init(ena_dev, &get_feat_ctx, &wd_state);
> > +     rc = ena_device_init(ena_dev, pci_dev, &get_feat_ctx, &wd_state);
> >       if (rc) {
> >               PMD_INIT_LOG(CRIT, "Failed to init ENA device");
> >               goto err;
> > @@ -2716,7 +2715,7 @@ static int ena_xstats_get_names(struct rte_eth_dev *dev,
> >                               struct rte_eth_xstat_name *xstats_names,
> >                               unsigned int n)
> >  {
> > -     unsigned int xstats_count = ena_xstats_calc_num(dev);
> > +     unsigned int xstats_count = ena_xstats_calc_num(dev->data);
> >       unsigned int stat, i, count = 0;
> >
> >       if (n < xstats_count || !xstats_names)
> > @@ -2765,7 +2764,7 @@ static int ena_xstats_get(struct rte_eth_dev *dev,
> >                         unsigned int n)
> >  {
> >       struct ena_adapter *adapter = dev->data->dev_private;
> > -     unsigned int xstats_count = ena_xstats_calc_num(dev);
> > +     unsigned int xstats_count = ena_xstats_calc_num(dev->data);
> >       unsigned int stat, i, count = 0;
> >       int stat_offset;
> >       void *stats_begin;
> > @@ -2997,7 +2996,7 @@ static void ena_update_on_link_change(void *adapter_data,
> >
> >       adapter = adapter_data;
> >       aenq_link_desc = (struct ena_admin_aenq_link_change_desc *)aenq_e;
> > -     eth_dev = adapter->rte_dev;
> > +     eth_dev = &rte_eth_devices[adapter->port_id];
> >
>
> Similarly, instead of passing "struct ena_adapter" as "void *adapter_data", it
> can be possible to pass "struct rte_eth_dev", as far as I can that chain has
> more layers but still possible.
>

Yes, it's being executed directly in the callback associated with the
admin interrupt, so as long as the interrupt callbacks are being
executed in the same process which registered them (and from what I
read in the documentation, it should be so, but please correct me if
I'm wrong), the "struct rte_eth_dev" pointer should be valid.

>
> Meanhile, you can use "void *process_private;" field of "struct rte_eth_dev" to
> keep process private eth_dev data, if you want.
>

  reply	other threads:[~2021-05-09 14:40 UTC|newest]

Thread overview: 121+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-04-30 12:57 [dpdk-dev] [PATCH 00/22] net/ena: update ENA PMD to v2.3.0 Michal Krawczyk
2021-04-30 12:57 ` [dpdk-dev] [PATCH 01/22] net/ena: switch memcpy to dpdk-optimized version Michal Krawczyk
2021-04-30 12:57 ` [dpdk-dev] [PATCH 02/22] net/ena/base: unify arg names for the functions Michal Krawczyk
2021-04-30 12:57 ` [dpdk-dev] [PATCH 03/22] net/ena/base: add dev arg to the logging macros Michal Krawczyk
2021-04-30 12:57 ` [dpdk-dev] [PATCH 04/22] net/ena/base: typos, style and comments improvements Michal Krawczyk
2021-04-30 12:57 ` [dpdk-dev] [PATCH 05/22] net/ena/base: fix issues from the static code scan Michal Krawczyk
2021-04-30 12:57 ` [dpdk-dev] [PATCH 06/22] net/ena/base: destroy multiple "wait events" Michal Krawczyk
2021-04-30 12:57 ` [dpdk-dev] [PATCH 07/22] net/ena/base: remove indir table from ENA feat ctx Michal Krawczyk
2021-04-30 12:57 ` [dpdk-dev] [PATCH 08/22] net/ena/base: remove mutable RSS from the host info Michal Krawczyk
2021-04-30 12:57 ` [dpdk-dev] [PATCH 09/22] net/ena/base: update generation date and commit Michal Krawczyk
2021-04-30 12:57 ` [dpdk-dev] [PATCH 10/22] net/ena/base: use rte_prefetch0_write Michal Krawczyk
2021-04-30 12:57 ` [dpdk-dev] [PATCH 11/22] net/ena/base: adjust changes to lastest ena-com Michal Krawczyk
2021-04-30 12:57 ` [dpdk-dev] [PATCH 12/22] net/ena: fix parsing of large_llq_hdr argument Michal Krawczyk
2021-04-30 12:57 ` [dpdk-dev] [PATCH 13/22] net/ena: terminate devargs allowed keys with NULL Michal Krawczyk
2021-04-30 12:57 ` [dpdk-dev] [PATCH 14/22] net/ena: indicate Rx RSS hash presence Michal Krawczyk
2021-04-30 12:57 ` [dpdk-dev] [PATCH 15/22] net/ena: remove endian swap functions Michal Krawczyk
2021-04-30 12:57 ` [dpdk-dev] [PATCH 16/22] net/ena: handle spurious wakeups in ENA_WAIT_EVENT Michal Krawczyk
2021-04-30 12:57 ` [dpdk-dev] [PATCH 17/22] net/ena: support SMP for mz alloc counter Michal Krawczyk
2021-04-30 12:57 ` [dpdk-dev] [PATCH 18/22] net/ena: move default RSS key to shared mem Michal Krawczyk
2021-04-30 12:57 ` [dpdk-dev] [PATCH 19/22] net/ena: make ethdev references smp safe Michal Krawczyk
2021-04-30 12:57 ` [dpdk-dev] [PATCH 20/22] net/ena: disable dev_ops not supported in SMP Michal Krawczyk
2021-04-30 12:57 ` [dpdk-dev] [PATCH 21/22] net/ena: report default ring size Michal Krawczyk
2021-04-30 12:57 ` [dpdk-dev] [PATCH 22/22] net/ena: update version to v2.3.0 Michal Krawczyk
2021-05-05  7:33 ` [dpdk-dev] [PATCH v2 00/22] net/ena: update ENA PMD " Michal Krawczyk
2021-05-05  7:33   ` [dpdk-dev] [PATCH v2 01/22] net/ena: switch memcpy to dpdk-optimized version Michal Krawczyk
2021-05-05  7:33   ` [dpdk-dev] [PATCH v2 02/22] net/ena/base: unify arg names for the functions Michal Krawczyk
2021-05-05  7:33   ` [dpdk-dev] [PATCH v2 03/22] net/ena/base: add dev arg to the logging macros Michal Krawczyk
2021-05-05  7:33   ` [dpdk-dev] [PATCH v2 04/22] net/ena/base: typos, style and comments improvements Michal Krawczyk
2021-05-05  7:33   ` [dpdk-dev] [PATCH v2 05/22] net/ena/base: fix issues from the static code scan Michal Krawczyk
2021-05-05  7:33   ` [dpdk-dev] [PATCH v2 06/22] net/ena/base: destroy multiple "wait events" Michal Krawczyk
2021-05-05  7:33   ` [dpdk-dev] [PATCH v2 07/22] net/ena/base: remove indir table from ENA feat ctx Michal Krawczyk
2021-05-05  7:33   ` [dpdk-dev] [PATCH v2 08/22] net/ena/base: remove mutable RSS from the host info Michal Krawczyk
2021-05-05  7:33   ` [dpdk-dev] [PATCH v2 09/22] net/ena/base: update generation date and commit Michal Krawczyk
2021-05-05  7:33   ` [dpdk-dev] [PATCH v2 10/22] net/ena/base: use rte_prefetch0_write Michal Krawczyk
2021-05-05  7:33   ` [dpdk-dev] [PATCH v2 11/22] net/ena/base: adjust changes to lastest ena-com Michal Krawczyk
2021-05-05  7:33   ` [dpdk-dev] [PATCH v2 12/22] net/ena: fix parsing of large_llq_hdr argument Michal Krawczyk
2021-05-05  7:33   ` [dpdk-dev] [PATCH v2 13/22] net/ena: terminate devargs allowed keys with NULL Michal Krawczyk
2021-05-05  7:33   ` [dpdk-dev] [PATCH v2 14/22] net/ena: indicate Rx RSS hash presence Michal Krawczyk
2021-05-05  7:33   ` [dpdk-dev] [PATCH v2 15/22] net/ena: remove endian swap functions Michal Krawczyk
2021-05-05  7:33   ` [dpdk-dev] [PATCH v2 16/22] net/ena: handle spurious wakeups in ENA_WAIT_EVENT Michal Krawczyk
2021-05-05  7:33   ` [dpdk-dev] [PATCH v2 17/22] net/ena: support SMP for mz alloc counter Michal Krawczyk
2021-05-05  7:33   ` [dpdk-dev] [PATCH v2 18/22] net/ena: move default RSS key to shared mem Michal Krawczyk
2021-05-05  7:33   ` [dpdk-dev] [PATCH v2 19/22] net/ena: make ethdev references smp safe Michal Krawczyk
2021-05-05  7:33   ` [dpdk-dev] [PATCH v2 20/22] net/ena: disable dev_ops not supported in SMP Michal Krawczyk
2021-05-05  7:33   ` [dpdk-dev] [PATCH v2 21/22] net/ena: report default ring size Michal Krawczyk
2021-05-05  7:33   ` [dpdk-dev] [PATCH v2 22/22] net/ena: update version to v2.3.0 Michal Krawczyk
2021-05-06 14:25   ` [dpdk-dev] [PATCH v3 00/22] net/ena: update ENA PMD " Michal Krawczyk
2021-05-06 14:25     ` [dpdk-dev] [PATCH v3 01/22] net/ena: switch memcpy to dpdk-optimized version Michal Krawczyk
2021-05-06 14:25     ` [dpdk-dev] [PATCH v3 02/22] net/ena/base: unify arg names for the functions Michal Krawczyk
2021-05-06 14:25     ` [dpdk-dev] [PATCH v3 03/22] net/ena/base: add dev arg to the logging macros Michal Krawczyk
2021-05-06 14:25     ` [dpdk-dev] [PATCH v3 04/22] net/ena/base: typos, style and comments improvements Michal Krawczyk
2021-05-06 14:25     ` [dpdk-dev] [PATCH v3 05/22] net/ena/base: fix issues from the static code scan Michal Krawczyk
2021-05-06 14:25     ` [dpdk-dev] [PATCH v3 06/22] net/ena/base: destroy multiple "wait events" Michal Krawczyk
2021-05-07 16:47       ` Ferruh Yigit
2021-05-09 14:17         ` Michał Krawczyk
2021-05-06 14:25     ` [dpdk-dev] [PATCH v3 07/22] net/ena/base: remove indir table from ENA feat ctx Michal Krawczyk
2021-05-06 14:25     ` [dpdk-dev] [PATCH v3 08/22] net/ena/base: remove mutable RSS from the host info Michal Krawczyk
2021-05-06 14:25     ` [dpdk-dev] [PATCH v3 09/22] net/ena/base: update generation date and commit Michal Krawczyk
2021-05-06 14:25     ` [dpdk-dev] [PATCH v3 10/22] net/ena/base: use rte_prefetch0_write Michal Krawczyk
2021-05-06 14:25     ` [dpdk-dev] [PATCH v3 11/22] net/ena/base: adjust changes to lastest ena-com Michal Krawczyk
2021-05-06 14:25     ` [dpdk-dev] [PATCH v3 12/22] net/ena: fix parsing of large_llq_hdr argument Michal Krawczyk
2021-05-06 14:25     ` [dpdk-dev] [PATCH v3 13/22] net/ena: terminate devargs allowed keys with NULL Michal Krawczyk
2021-05-07 16:47       ` [dpdk-dev] [dpdk-stable] " Ferruh Yigit
2021-05-06 14:25     ` [dpdk-dev] [PATCH v3 14/22] net/ena: indicate Rx RSS hash presence Michal Krawczyk
2021-05-07 16:47       ` [dpdk-dev] [dpdk-stable] " Ferruh Yigit
2021-05-09 14:20         ` Michał Krawczyk
2021-05-06 14:25     ` [dpdk-dev] [PATCH v3 15/22] net/ena: remove endian swap functions Michal Krawczyk
2021-05-06 14:25     ` [dpdk-dev] [PATCH v3 16/22] net/ena: handle spurious wakeups in ENA_WAIT_EVENT Michal Krawczyk
2021-05-07 16:55       ` Ferruh Yigit
2021-05-06 14:25     ` [dpdk-dev] [PATCH v3 17/22] net/ena: support SMP for mz alloc counter Michal Krawczyk
2021-05-07 16:48       ` Ferruh Yigit
2021-05-07 17:18         ` Stanislaw Kardach
2021-05-09 13:41           ` Ferruh Yigit
2021-05-10  8:18             ` Stanislaw Kardach
2021-05-06 14:25     ` [dpdk-dev] [PATCH v3 18/22] net/ena: move default RSS key to shared mem Michal Krawczyk
2021-05-10  8:27       ` Stanislaw Kardach
2021-05-06 14:25     ` [dpdk-dev] [PATCH v3 19/22] net/ena: make ethdev references smp safe Michal Krawczyk
2021-05-07 16:49       ` Ferruh Yigit
2021-05-09 14:40         ` Michał Krawczyk [this message]
2021-05-06 14:25     ` [dpdk-dev] [PATCH v3 20/22] net/ena: disable dev_ops not supported in SMP Michal Krawczyk
2021-05-06 14:25     ` [dpdk-dev] [PATCH v3 21/22] net/ena: report default ring size Michal Krawczyk
2021-05-06 14:25     ` [dpdk-dev] [PATCH v3 22/22] net/ena: update version to v2.3.0 Michal Krawczyk
2021-05-07 16:50       ` Ferruh Yigit
2021-05-09 14:33         ` Michał Krawczyk
2021-05-07 14:59     ` [dpdk-dev] [PATCH v3 00/22] net/ena: update ENA PMD " Ferruh Yigit
2021-05-07 15:07       ` Ferruh Yigit
2021-05-10 12:02       ` Michał Krawczyk
2021-05-10 12:27         ` Ferruh Yigit
2021-05-11  6:45           ` [dpdk-dev] [PATCH v4 00/19] " Michal Krawczyk
2021-05-11  6:45             ` [dpdk-dev] [PATCH v4 01/19] net/ena: switch memcpy to dpdk-optimized version Michal Krawczyk
2021-05-11  6:45             ` [dpdk-dev] [PATCH v4 02/19] net/ena/base: unify arg names for the functions Michal Krawczyk
2021-05-11  6:45             ` [dpdk-dev] [PATCH v4 03/19] net/ena/base: add dev arg to the logging macros Michal Krawczyk
2021-05-11  6:45             ` [dpdk-dev] [PATCH v4 04/19] net/ena/base: style and comments improvements Michal Krawczyk
2021-05-11  6:45             ` [dpdk-dev] [PATCH v4 05/19] net/ena/base: fix issues from the static code scan Michal Krawczyk
2021-05-11  6:45             ` [dpdk-dev] [PATCH v4 06/19] net/ena/base: destroy multiple "wait events" Michal Krawczyk
2021-05-11  6:45             ` [dpdk-dev] [PATCH v4 07/19] net/ena/base: remove indir table from ENA feat ctx Michal Krawczyk
2021-05-11  6:45             ` [dpdk-dev] [PATCH v4 08/19] net/ena/base: remove mutable RSS from the host info Michal Krawczyk
2021-05-11  6:45             ` [dpdk-dev] [PATCH v4 09/19] net/ena/base: use rte prefetch0 write Michal Krawczyk
2021-05-11  6:45             ` [dpdk-dev] [PATCH v4 10/19] net/ena/base: adjust changes to lastest ena-com Michal Krawczyk
2021-05-11  6:45             ` [dpdk-dev] [PATCH v4 11/19] net/ena: fix parsing of large LLQ header devarg Michal Krawczyk
2021-05-12 17:48               ` [dpdk-dev] [dpdk-stable] " Thomas Monjalon
2021-05-11  6:45             ` [dpdk-dev] [PATCH v4 12/19] net/ena: terminate devargs allowed keys with null Michal Krawczyk
2021-05-11  6:45             ` [dpdk-dev] [PATCH v4 13/19] net/ena: indicate Rx RSS hash presence Michal Krawczyk
2021-05-11 12:16               ` Ferruh Yigit
2021-05-11  6:45             ` [dpdk-dev] [PATCH v4 14/19] net/ena: remove endian swap functions Michal Krawczyk
2021-05-11  6:45             ` [dpdk-dev] [PATCH v4 15/19] net/ena: handle spurious wakeups in ENA wait event Michal Krawczyk
2021-05-11  6:45             ` [dpdk-dev] [PATCH v4 16/19] net/ena: make ethdev references SMP safe Michal Krawczyk
2021-05-11 12:22               ` Ferruh Yigit
2021-05-11  6:45             ` [dpdk-dev] [PATCH v4 17/19] net/ena: disable dev ops not supported in SMP Michal Krawczyk
2021-05-11 12:29               ` Ferruh Yigit
2021-05-11  6:45             ` [dpdk-dev] [PATCH v4 18/19] net/ena: report default ring size Michal Krawczyk
2021-05-11  6:45             ` [dpdk-dev] [PATCH v4 19/19] net/ena: update version to v2.3.0 Michal Krawczyk
2021-05-11 13:07             ` [dpdk-dev] [PATCH v4 00/19] net/ena: update ENA PMD " Ferruh Yigit
2021-05-11 13:34               ` Michał Krawczyk
2021-05-12 10:13                 ` [dpdk-dev] [PATCH v5 0/3] " Michal Krawczyk
2021-05-12 10:13                   ` [dpdk-dev] [PATCH v5 1/3] net/ena: indicate Rx RSS hash presence Michal Krawczyk
2021-05-12 10:13                   ` [dpdk-dev] [PATCH v5 2/3] net/ena: make ethdev references multiprocess safe Michal Krawczyk
2021-05-12 10:13                   ` [dpdk-dev] [PATCH v5 3/3] net/ena: disable dev ops not supported by secondary process Michal Krawczyk
2021-05-12 12:20                   ` [dpdk-dev] [PATCH v5 0/3] net/ena: update ENA PMD to v2.3.0 Ferruh Yigit
2021-05-07 15:52     ` [dpdk-dev] [PATCH v3 00/22] " Ferruh Yigit
2021-05-09 14:15       ` Michał Krawczyk

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=CAJMMOfMFkPcF3C_a5_PYb9JE7ppVddjNinWWAWKsOLadRabXHQ@mail.gmail.com \
    --to=mk@semihalf.com \
    --cc=dev@dpdk.org \
    --cc=ferruh.yigit@intel.com \
    --cc=gtzalik@amazon.com \
    --cc=igorch@amazon.com \
    --cc=kda@semihalf.com \
    --cc=ndagan@amazon.com \
    --cc=shayagr@amazon.com \
    --cc=upstream@semihalf.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
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).