public inbox for linuxppc-dev@ozlabs.org 
 help / color / mirror / Atom feed
From: Alistair Popple <alistair@popple•id.au>
To: Alexey Kardashevskiy <aik@ozlabs•ru>
Cc: "Jose Ricardo Ziviani" <joserz@linux•ibm.com>,
	"Sam Bobroff" <sbobroff@linux•ibm.com>,
	linuxppc-dev@lists•ozlabs.org,
	"Alex Williamson" <alex.williamson@redhat•com>,
	kvm-ppc@vger•kernel.org,
	"Piotr Jaroszynski" <pjaroszynski@nvidia•com>,
	"Oliver O'Halloran" <oohall@gmail•com>,
	"Andrew Donnellan" <andrew.donnellan@au1•ibm.com>,
	"Leonardo Augusto Guimarães Garcia" <lagarcia@br•ibm.com>,
	"Reza Arbab" <arbab@linux•ibm.com>,
	"David Gibson" <david@gibson•dropbear.id.au>
Subject: Re: [PATCH kernel v3 06/22] powerpc/powernv: Detach npu struct from pnv_phb
Date: Wed, 14 Nov 2018 15:28:36 +1100	[thread overview]
Message-ID: <1686227.ptHs0ZEO6U@townsend> (raw)
In-Reply-To: <20181113082823.2440-7-aik@ozlabs.ru>

Hi Alexey,

On Tuesday, 13 November 2018 7:28:07 PM AEDT Alexey Kardashevskiy wrote:
>  static struct npu *npdev_to_npu(struct pci_dev *npdev)
>  {
> -	struct pnv_phb *nphb;
> +	struct pci_controller *hose = pci_bus_to_host(npdev->bus);
> +	struct npu *npu;
> 
> -	nphb = pci_bus_to_host(npdev->bus)->private_data;
> +	list_for_each_entry(npu, &npu2_devices, next)

This is called from the ATSD path which is (or at least has been) quite a 
performance critical path so searching through all the NPUs in a list may be 
problematic.

I guess currently it wont make any practical difference as we only ever have 2 
NPUs, but in future they may get divided into more logical NPUs. Would it be 
possible to store a back-pointer somewhere so we can avoid the lookup?

> +		if (hose == npu->hose)
> +			return npu;
> 
> -	return &nphb->npu;
> +	WARN_ON_ONCE(1);
> +	return NULL;
>  }
> 
>  /* Maximum number of nvlinks per npu */
> @@ -505,6 +531,9 @@ static void acquire_atsd_reg(struct npu_context
> *npu_context, continue;
> 
>  			npu = npdev_to_npu(npdev);
> +			if (!npu)
> +				continue;
> +
>  			mmio_atsd_reg[i].npu = npu;
>  			mmio_atsd_reg[i].reg = get_mmio_atsd_reg(npu);
>  			while (mmio_atsd_reg[i].reg < 0) {
> @@ -701,6 +730,8 @@ struct npu_context *pnv_npu2_init_context(struct pci_dev
> *gpdev,
> 
>  	nphb = pci_bus_to_host(npdev->bus)->private_data;
>  	npu = npdev_to_npu(npdev);
> +	if (!npu)
> +		return ERR_PTR(-ENODEV);
> 
>  	/*
>  	 * Setup the NPU context table for a particular GPU. These need to be
> @@ -821,6 +852,8 @@ void pnv_npu2_destroy_context(struct npu_context
> *npu_context,
> 
>  	nphb = pci_bus_to_host(npdev->bus)->private_data;
>  	npu = npdev_to_npu(npdev);
> +	if (!npu)
> +		return;
>  	nvlink_dn = of_parse_phandle(npdev->dev.of_node, "ibm,nvlink", 0);
>  	if (WARN_ON(of_property_read_u32(nvlink_dn, "ibm,npu-link-index",
>  							&nvlink_index)))
> @@ -898,9 +931,15 @@ int pnv_npu2_init(struct pnv_phb *phb)
>  	struct pci_dev *gpdev;
>  	static int npu_index;
>  	uint64_t rc = 0;
> +	struct pci_controller *hose = phb->hose;
> +	struct npu *npu;
> +	int ret;
> 
> -	phb->npu.nmmu_flush =
> -		of_property_read_bool(phb->hose->dn, "ibm,nmmu-flush");
> +	npu = kzalloc(sizeof(*npu), GFP_KERNEL);
> +	if (!npu)
> +		return -ENOMEM;
> +
> +	npu->nmmu_flush = of_property_read_bool(hose->dn, "ibm,nmmu-flush");
>  	for_each_child_of_node(phb->hose->dn, dn) {
>  		gpdev = pnv_pci_get_gpu_dev(get_pci_dev(dn));
>  		if (gpdev) {
> @@ -914,18 +953,31 @@ int pnv_npu2_init(struct pnv_phb *phb)
>  		}
>  	}
> 
> -	for (i = 0; !of_property_read_u64_index(phb->hose->dn, "ibm,mmio-atsd",
> +	for (i = 0; !of_property_read_u64_index(hose->dn, "ibm,mmio-atsd",
>  							i, &mmio_atsd); i++)
> -		phb->npu.mmio_atsd_regs[i] = ioremap(mmio_atsd, 32);
> +		npu->mmio_atsd_regs[i] = ioremap(mmio_atsd, 32);
> 
> -	pr_info("NPU%lld: Found %d MMIO ATSD registers", phb->opal_id, i);
> -	phb->npu.mmio_atsd_count = i;
> -	phb->npu.mmio_atsd_usage = 0;
> +	pr_info("NPU%d: Found %d MMIO ATSD registers", hose->global_number, i);
> +	npu->mmio_atsd_count = i;
> +	npu->mmio_atsd_usage = 0;
>  	npu_index++;
> -	if (WARN_ON(npu_index >= NV_MAX_NPUS))
> -		return -ENOSPC;
> +	if (WARN_ON(npu_index >= NV_MAX_NPUS)) {
> +		ret = -ENOSPC;
> +		goto fail_exit;
> +	}
>  	max_npu2_index = npu_index;
> -	phb->npu.index = npu_index;
> +	npu->index = npu_index;
> +	npu->hose = hose;
> +
> +	list_add(&npu->next, &npu2_devices);

Guess we don't need any locking here as the list gets setup once during boot 
long before loading the driver and is never modified right?

- Alistair

>  	return 0;
> +
> +fail_exit:
> +	for (i = 0; i < npu->mmio_atsd_count; ++i)
> +		iounmap(npu->mmio_atsd_regs[i]);
> +
> +	kfree(npu);
> +
> +	return ret;
>  }



  reply	other threads:[~2018-11-14  4:30 UTC|newest]

Thread overview: 42+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2018-11-13  8:28 [PATCH kernel v3 00/22] powerpc/powernv/npu, vfio: NVIDIA V100 + P9 passthrough Alexey Kardashevskiy
2018-11-13  8:28 ` [PATCH kernel v3 01/22] powerpc/ioda/npu: Call skiboot's hot reset hook when disabling NPU2 Alexey Kardashevskiy
2018-11-13  8:28 ` [PATCH kernel v3 02/22] powerpc/mm/iommu/vfio_spapr_tce: Change mm_iommu_get to reference a region Alexey Kardashevskiy
2018-11-15  5:32   ` David Gibson
2018-11-13  8:28 ` [PATCH kernel v3 03/22] powerpc/mm/iommu: Make mm_iommu_new() fail on existing regions Alexey Kardashevskiy
2018-11-15  5:38   ` David Gibson
2018-11-13  8:28 ` [PATCH kernel v3 04/22] powerpc/vfio/iommu/kvm: Do not pin device memory Alexey Kardashevskiy
2018-11-16  3:11   ` David Gibson
2018-11-13  8:28 ` [PATCH kernel v3 05/22] powerpc/powernv/npu: Add helper to access struct npu for NPU device Alexey Kardashevskiy
2018-11-14  3:42   ` Alistair Popple
2018-11-13  8:28 ` [PATCH kernel v3 06/22] powerpc/powernv: Detach npu struct from pnv_phb Alexey Kardashevskiy
2018-11-14  4:28   ` Alistair Popple [this message]
2018-11-19  7:18     ` Alexey Kardashevskiy
2018-11-13  8:28 ` [PATCH kernel v3 07/22] powerpc/powernv/npu: Move OPAL calls away from context manipulation Alexey Kardashevskiy
2018-11-14  4:57   ` Alistair Popple
2018-11-13  8:28 ` [PATCH kernel v3 08/22] powerpc/pseries/iommu: Allow dynamic window to start from zero Alexey Kardashevskiy
2018-11-13  8:28 ` [PATCH kernel v3 09/22] powerpc/pseries/iommu: Force default DMA window removal Alexey Kardashevskiy
2018-11-16  4:54   ` David Gibson
2018-11-19  7:28     ` Alexey Kardashevskiy
2018-11-13  8:28 ` [PATCH kernel v3 10/22] powerpc/pseries/iommu: Use memory@ nodes in max RAM address calculation Alexey Kardashevskiy
2018-11-16  5:23   ` David Gibson
2018-11-19  7:43     ` Alexey Kardashevskiy
2018-11-13  8:28 ` [PATCH kernel v3 11/22] powerpc/pseries/npu: Enable platform support Alexey Kardashevskiy
2018-11-16  5:25   ` David Gibson
2018-11-19  7:50     ` Alexey Kardashevskiy
2018-11-13  8:28 ` [PATCH kernel v3 12/22] powerpc/pseries: Remove IOMMU API support for non-LPAR systems Alexey Kardashevskiy
2018-11-13  8:28 ` [PATCH kernel v3 13/22] powerpc/powernv/pseries: Rework device adding to IOMMU groups Alexey Kardashevskiy
2018-11-13  8:28 ` [PATCH kernel v3 14/22] powerpc/iommu_api: Move IOMMU groups setup to a single place Alexey Kardashevskiy
2018-11-19  0:15   ` David Gibson
2018-11-13  8:28 ` [PATCH kernel v3 15/22] powerpc/powernv: Reference iommu_table while it is linked to a group Alexey Kardashevskiy
2018-11-19  0:20   ` David Gibson
2018-11-13  8:28 ` [PATCH kernel v3 16/22] powerpc/powernv: Add purge cache OPAL call Alexey Kardashevskiy
2018-11-19  0:21   ` David Gibson
2018-11-13  8:28 ` [PATCH kernel v3 17/22] powerpc/powernv/npu: Convert NPU IOMMU helpers to iommu_table_group_ops Alexey Kardashevskiy
2018-11-19  0:24   ` David Gibson
2018-11-13  8:28 ` [PATCH kernel v3 18/22] powerpc/powernv/npu: Add compound IOMMU groups Alexey Kardashevskiy
2018-11-19  1:12   ` David Gibson
2018-11-19  2:29     ` Alexey Kardashevskiy
2018-11-13  8:28 ` [PATCH kernel v3 19/22] powerpc/powernv/npu: Add release_ownership hook Alexey Kardashevskiy
2018-11-13  8:28 ` [PATCH kernel v3 20/22] vfio_pci: Allow mapping extra regions Alexey Kardashevskiy
2018-11-13  8:28 ` [PATCH kernel v3 21/22] vfio_pci: Allow regions to add own capabilities Alexey Kardashevskiy
2018-11-13  8:28 ` [PATCH kernel v3 22/22] vfio_pci: Add NVIDIA GV100GL [Tesla V100 SXM2] [10de:1db1] subdriver Alexey Kardashevskiy

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=1686227.ptHs0ZEO6U@townsend \
    --to=alistair@popple$(echo .)id.au \
    --cc=aik@ozlabs$(echo .)ru \
    --cc=alex.williamson@redhat$(echo .)com \
    --cc=andrew.donnellan@au1$(echo .)ibm.com \
    --cc=arbab@linux$(echo .)ibm.com \
    --cc=david@gibson$(echo .)dropbear.id.au \
    --cc=joserz@linux$(echo .)ibm.com \
    --cc=kvm-ppc@vger$(echo .)kernel.org \
    --cc=lagarcia@br$(echo .)ibm.com \
    --cc=linuxppc-dev@lists$(echo .)ozlabs.org \
    --cc=oohall@gmail$(echo .)com \
    --cc=pjaroszynski@nvidia$(echo .)com \
    --cc=sbobroff@linux$(echo .)ibm.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html
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