DPDK patches and discussions
 help / color / mirror / Atom feed
From: "Gaëtan Rivet" <grive@u256.net>
To: Sunil Kumar Kori <skori@marvell.com>
Cc: stephen@networkplumber.org, david.marchand@redhat.com,
	jerinj@marvell.com, dev@dpdk.org
Subject: Re: [dpdk-dev] [PATCH v4 1/1] bus/pci: optimise scanning with whitelist/blacklist
Date: Fri, 1 May 2020 23:00:24 +0200	[thread overview]
Message-ID: <20200501210013.hbkcm6pcsyxscyvc@u256.net> (raw)
In-Reply-To: <20200501113909.26355-1-skori@marvell.com>

Hello Sunil,

It's pretty close, thanks. Unfortunately, I just have a few nits
remaining.

On 01/05/20 17:09 +0530, Sunil Kumar Kori wrote:
> rte_bus_scan API scans all the available PCI devices irrespective of white
> or black listing parameters then further devices are probed based on white
> or black listing parameters. So unnecessary CPU cycles are wasted during
> rte_pci_scan.
> 
> For Octeontx2 platform with core frequency 2.4 Ghz, rte_bus_scan consumes
> around 26ms to scan around 90 PCI devices but all may not be used by the
> application. So for the application which uses 2 NICs, rte_bus_scan
> consumes few microseconds and rest time is saved with this patch.
> 
> Patch restricts devices to be scanned as per below mentioned conditions:
>  - All devices will be scanned if no parameters are passed.
>  - Only white listed devices will be scanned if white list is available.
>  - All devices, except black listed, will be scanned if black list is
>    available.
> 
> Signed-off-by: Sunil Kumar Kori <skori@marvell.com>
> ---
> v4:
>  - Review comments incorporated (Gaeten and David).
>  - Rebased on top of tree.
> v3:
>  - Remove __rte_experimental from private function.
>  - Remove entry from map file too.
> v2:
>  - Added function to validate ignorance of device based on PCI address.
>  - Marked device validation function as experimental.
> 
>  drivers/bus/pci/bsd/pci.c    | 12 +++++++++++-
>  drivers/bus/pci/linux/pci.c  |  3 +++
>  drivers/bus/pci/pci_common.c | 33 ++++++++++++---------------------
>  drivers/bus/pci/private.h    | 11 +++++++++++
>  4 files changed, 37 insertions(+), 22 deletions(-)
> 
> diff --git a/drivers/bus/pci/bsd/pci.c b/drivers/bus/pci/bsd/pci.c
> index ebbfeb13a..709a1e7e5 100644
> --- a/drivers/bus/pci/bsd/pci.c
> +++ b/drivers/bus/pci/bsd/pci.c
> @@ -338,6 +338,7 @@ rte_pci_scan(void)
>  			.match_buf_len = sizeof(matches),
>  			.matches = &matches[0],
>  	};
> +	struct rte_pci_addr pci_addr;
>  
>  	/* for debug purposes, PCI can be disabled */
>  	if (!rte_eal_has_pci())
> @@ -357,9 +358,18 @@ rte_pci_scan(void)
>  			goto error;
>  		}
>  
> -		for (i = 0; i < conf_io.num_matches; i++)
> +		for (i = 0; i < conf_io.num_matches; i++) {
> +			pci_addr.domain = matches[i].pc_sel.pc_domain;
> +			pci_addr.bus = matches[i].pc_sel.pc_bus;
> +			pci_addr.devid = matches[i].pc_sel.pc_dev;
> +			pci_addr.function = matches[i].pc_sel.pc_func;
> +
> +			if (rte_pci_ignore_device_addr(&pci_addr))
> +				continue;
> +
>  			if (pci_scan_one(fd, &matches[i]) < 0)
>  				goto error;
> +		}
>  
>  		dev_count += conf_io.num_matches;
>  	} while(conf_io.status == PCI_GETCONF_MORE_DEVS);
> diff --git a/drivers/bus/pci/linux/pci.c b/drivers/bus/pci/linux/pci.c
> index ca783b157..ec347eff3 100644
> --- a/drivers/bus/pci/linux/pci.c
> +++ b/drivers/bus/pci/linux/pci.c
> @@ -487,6 +487,9 @@ rte_pci_scan(void)
>  		if (parse_pci_addr_format(e->d_name, sizeof(e->d_name), &addr) != 0)
>  			continue;
>  
> +		if (rte_pci_ignore_device_addr(&addr))
> +			continue;
> +
>  		snprintf(dirname, sizeof(dirname), "%s/%s",
>  				rte_pci_get_sysfs_path(), e->d_name);
>  
> diff --git a/drivers/bus/pci/pci_common.c b/drivers/bus/pci/pci_common.c
> index 3f5542076..d34e59536 100644
> --- a/drivers/bus/pci/pci_common.c
> +++ b/drivers/bus/pci/pci_common.c
> @@ -42,14 +42,17 @@ const char *rte_pci_get_sysfs_path(void)
>  	return path;
>  }
>  
> -static struct rte_devargs *pci_devargs_lookup(struct rte_pci_device *dev)
> +static struct rte_devargs *
> +pci_devargs_lookup(const struct rte_pci_addr *pci_addr)
>  {
>  	struct rte_devargs *devargs;
>  	struct rte_pci_addr addr;
>  
>  	RTE_EAL_DEVARGS_FOREACH("pci", devargs) {
> -		devargs->bus->parse(devargs->name, &addr);
> -		if (!rte_pci_addr_cmp(&dev->addr, &addr))
> +		if (rte_pci_addr_parse(devargs->name, &addr) < 0)
> +			continue;
> +
> +		if (!rte_pci_addr_cmp(pci_addr, &addr))

I'm really sorry, I was overzealous on your v4. I thought using
devargs->bus->parse() was from your patch, but it was already there.

It's even more shameful as it was me who wrote this.

Anyway, it's much better looking this way, but it should be a separate
patch.

>  			return devargs;
>  	}
>  	return NULL;
> @@ -63,7 +66,7 @@ pci_name_set(struct rte_pci_device *dev)
>  	/* Each device has its internal, canonical name set. */
>  	rte_pci_device_name(&dev->addr,
>  			dev->name, sizeof(dev->name));
> -	devargs = pci_devargs_lookup(dev);
> +	devargs = pci_devargs_lookup(&dev->addr);
>  	dev->device.devargs = devargs;
>  	/* In blacklist mode, if the device is not blacklisted, no
>  	 * rte_devargs exists for it.
> @@ -293,23 +296,12 @@ rte_pci_probe(void)
>  {
>  	struct rte_pci_device *dev = NULL;
>  	size_t probed = 0, failed = 0;
> -	struct rte_devargs *devargs;
> -	int probe_all = 0;
>  	int ret = 0;
>  
> -	if (rte_pci_bus.bus.conf.scan_mode != RTE_BUS_SCAN_WHITELIST)
> -		probe_all = 1;
> -
>  	FOREACH_DEVICE_ON_PCIBUS(dev) {
>  		probed++;
>  
> -		devargs = dev->device.devargs;
> -		/* probe all or only whitelisted devices */
> -		if (probe_all)
> -			ret = pci_probe_all_drivers(dev);
> -		else if (devargs != NULL &&
> -			devargs->policy == RTE_DEV_WHITELISTED)
> -			ret = pci_probe_all_drivers(dev);
> +		ret = pci_probe_all_drivers(dev);
>  		if (ret < 0) {
>  			if (ret != -EEXIST) {
>  				RTE_LOG(ERR, EAL, "Requested device "
> @@ -589,10 +581,10 @@ pci_dma_unmap(struct rte_device *dev, void *addr, uint64_t iova, size_t len)
>  	return -1;
>  }
>  
> -static bool
> -pci_ignore_device(const struct rte_pci_device *dev)
> +bool
> +rte_pci_ignore_device_addr(const struct rte_pci_addr *pci_addr)

I'd prefer naming this function rte_pci_ignore_device(),
given that pci_ignore_device() disappears.

>  {
> -	struct rte_devargs *devargs = dev->device.devargs;
> +	struct rte_devargs *devargs = pci_devargs_lookup(pci_addr);
>  
>  	switch (rte_pci_bus.bus.conf.scan_mode) {
>  	case RTE_BUS_SCAN_WHITELIST:
> @@ -627,8 +619,7 @@ rte_pci_get_iommu_class(void)
>  		if (iommu_no_va == -1)
>  			iommu_no_va = pci_device_iommu_support_va(dev)
>  					? 0 : 1;
> -		if (pci_ignore_device(dev))
> -			continue;
> +
>  		if (dev->kdrv == RTE_KDRV_UNKNOWN ||
>  		    dev->kdrv == RTE_KDRV_NONE)
>  			continue;
> diff --git a/drivers/bus/pci/private.h b/drivers/bus/pci/private.h
> index a205d4d9f..3874219ba 100644
> --- a/drivers/bus/pci/private.h
> +++ b/drivers/bus/pci/private.h
> @@ -42,6 +42,17 @@ int rte_pci_scan(void);
>  void
>  pci_name_set(struct rte_pci_device *dev);
>  
> +/**
> + * Validate whether a device with given pci address should be ignored or not.
> + *
> + * @param pci_addr
> + *	PCI address of device to be validated
> + * @return
> + *	1: if device is to be ignored,
> + *	0: if device is to be scanned,
> + */
> +bool rte_pci_ignore_device_addr(const struct rte_pci_addr *pci_addr);
> +
>  /**
>   * Add a PCI device to the PCI Bus (append to PCI Device list). This function
>   * also updates the bus references of the PCI Device (and the generic device
> -- 
> 2.17.1
> 

Otherwise it looks good to me, almost finished!

-- 
Gaëtan

  parent reply	other threads:[~2020-05-01 21:00 UTC|newest]

Thread overview: 35+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-12-16  7:55 [dpdk-dev] [PATCH] bus/pci: restricted bus scanning to allowed devices Sunil Kumar Kori
2019-12-16 16:13 ` Stephen Hemminger
2019-12-17 11:00   ` [dpdk-dev] [EXT] " Sunil Kumar Kori
2020-01-21  8:39     ` Sunil Kumar Kori
     [not found]       ` <MN2PR18MB327936807460D9F2AE4894F3B40F0@MN2PR18MB3279.namprd18.prod.outlook.com>
2020-02-27  8:30         ` Sunil Kumar Kori
2020-03-09  6:06           ` Sunil Kumar Kori
2020-04-06  9:32             ` Sunil Kumar Kori
2020-04-06 13:21               ` David Marchand
2020-04-07  9:21                 ` Sunil Kumar Kori
2020-04-07  9:28 ` [dpdk-dev] [PATCH v2 1/1] bus/pci: optimise scanning with whitelist/blacklist Sunil Kumar Kori
2020-04-17  8:30   ` Sunil Kumar Kori
2020-04-17  8:44   ` David Marchand
2020-04-17 11:15     ` [dpdk-dev] [EXT] " Sunil Kumar Kori
2020-04-17 13:25       ` David Marchand
2020-04-17 15:12         ` Sunil Kumar Kori
2020-04-17 15:35           ` David Marchand
2020-04-17 16:00             ` Sunil Kumar Kori
2020-04-20  6:59               ` Sunil Kumar Kori
2020-04-20  6:55   ` [dpdk-dev] [PATCH v3 " Sunil Kumar Kori
2020-04-21 15:18     ` Gaëtan Rivet
2020-04-22  6:17       ` [dpdk-dev] [EXT] " Sunil Kumar Kori
2020-04-22  9:38         ` Gaëtan Rivet
2020-04-23  7:47           ` Sunil Kumar Kori
2020-04-27 18:43     ` [dpdk-dev] " Gaëtan Rivet
2020-04-28 13:52       ` [dpdk-dev] [EXT] " Sunil Kumar Kori
2020-05-01 11:39     ` [dpdk-dev] [PATCH v4 " Sunil Kumar Kori
2020-05-01 12:40       ` Sunil Kumar Kori
2020-05-01 21:00       ` Gaëtan Rivet [this message]
2020-05-02  7:20         ` [dpdk-dev] [EXT] " Sunil Kumar Kori
2020-05-02  7:42       ` [dpdk-dev] [PATCH v5 " Sunil Kumar Kori
2020-05-02 11:27         ` Gaëtan Rivet
2020-05-04 14:17         ` David Marchand
2020-05-05  5:57           ` [dpdk-dev] [EXT] " Sunil Kumar Kori
2020-05-06 12:54           ` [dpdk-dev] " David Marchand
2020-05-11 14:59         ` 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=20200501210013.hbkcm6pcsyxscyvc@u256.net \
    --to=grive@u256.net \
    --cc=david.marchand@redhat.com \
    --cc=dev@dpdk.org \
    --cc=jerinj@marvell.com \
    --cc=skori@marvell.com \
    --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).