DPDK patches and discussions
 help / color / mirror / Atom feed
From: Jerin Jacob <jerinjacobk@gmail.com>
To: "Somalapuram, Amaranath" <Amaranath.Somalapuram@amd.com>
Cc: "dev@dpdk.org" <dev@dpdk.org>, "stable@dpdk.org" <stable@dpdk.org>
Subject: Re: [dpdk-dev] [PATCH v1 5/6] crypto/ccp: enable IOMMU for CCP
Date: Tue, 15 Oct 2019 15:13:33 +0530	[thread overview]
Message-ID: <CALBAE1M=eGjtYKAy=GL+TJiR=h4ODN8kC-FNoK5vQK9s+h7kZg@mail.gmail.com> (raw)
In-Reply-To: <DM5PR1201MB247459D2C717D2F89820D937F8930@DM5PR1201MB2474.namprd12.prod.outlook.com>

On Tue, Oct 15, 2019 at 2:05 PM Somalapuram, Amaranath
<Amaranath.Somalapuram@amd.com> wrote:
>
> Problem statement: As of now vdev device do not support IOMMU.
> vdev device used to custom solution, even for software drives like openssl uses vdev.

I spend some time going through the driver.

#The ideal architecture of this driver would have been to introduce a
bus driver(see drivers/bus/ifpga/) which does all the PCI
probes(drivers/crypto/ccp/rte_ccp_pmd.c and
drivers/crypto/ccp/ccp_pci.c)
and arrange the devices on the bus scan and enable "non bus" specific
driver crypto driver. So that no pci probe etc in crypto driver.
# On the upside, You DONT need to give vdev= on eal arguments, on AMD
machines it can probe the crypto devices automatically and probe the
crypto driver
# I dont think, it is specific to vdev IOMMU support, If would have
been PCI/Any bus driver, you would have similar changes. Right?
#  Major portion of this patch does following
+               if (iommu_mode == 2)
+                       pst.src_addr = (phys_addr_t)rte_mem_virt2iova(
+                                       (void *) lsb_buf);
+               else
+                       pst.src_addr = (phys_addr_t)rte_mem_virt2phy(
+                                       (void *) lsb_buf);
+

Since the following check already present common code, Do we need the
above check, just calling rte_mem_virt2iova() is enough. Right?

rte_iova_t
rte_mem_virt2iova(const void *virtaddr)
{
        if (rte_eal_iova_mode() == RTE_IOVA_VA)
                return (uintptr_t)virtaddr;
        return rte_mem_virt2phy(virtaddr);
}










> I feel its not advisable to put iommu in vdev.
> moving the changes to vdev will effect rest of the vdev drives.
> That will be big efforts. Every vdev drivers has their own implementation. Need better design to move it to vdev or common place.
>
> Regards,
> S.Amarnath
>
> -----Original Message-----
> From: Jerin Jacob <jerinjacobk@gmail.com>
> Sent: Tuesday, October 15, 2019 1:47 PM
> To: Somalapuram, Amaranath <Amaranath.Somalapuram@amd.com>
> Cc: dev@dpdk.org; stable@dpdk.org
> Subject: Re: [dpdk-dev] [PATCH v1 5/6] crypto/ccp: enable IOMMU for CCP
>
> [CAUTION: External Email]
>
> On Tue, Oct 15, 2019 at 12:32 PM <asomalap@amd.com> wrote:
> >
> > From: Amaranath Somalapuram <asomalap@amd.com>
> >
> > CCP use vdev framework, and vdev framework don’t support IOMMU.
> > Adding custom IOMMU support for AMD CCP drives.
> > Cc: stable@dpdk.org
> >
> > +       if (iommu_mode == 2)
> > +               pci->kdrv = RTE_KDRV_VFIO;
> > +       else if (iommu_mode == 0)
> > +               pci->kdrv = RTE_KDRV_IGB_UIO;
> > +       else if (iommu_mode == 1)
> > +               pci->kdrv = RTE_KDRV_UIO_GENERIC;
>
> The crypto driver should not have iommu mode-specific handling.
> I am not sure about the problem statement. If the problem is, iommu support for PCI based vdev device then move the solution to common layer so that everyone can use it. If not, please share the problem statement

  reply	other threads:[~2019-10-15  9:43 UTC|newest]

Thread overview: 7+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-10-15  7:02 asomalap
2019-10-15  8:16 ` Jerin Jacob
2019-10-15  8:35   ` Somalapuram, Amaranath
2019-10-15  9:43     ` Jerin Jacob [this message]
2019-10-22  5:13       ` Somalapuram, Amaranath
2019-10-15 11:08 ` Akhil Goyal
2019-10-22  5:14   ` Somalapuram, Amaranath

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='CALBAE1M=eGjtYKAy=GL+TJiR=h4ODN8kC-FNoK5vQK9s+h7kZg@mail.gmail.com' \
    --to=jerinjacobk@gmail.com \
    --cc=Amaranath.Somalapuram@amd.com \
    --cc=dev@dpdk.org \
    --cc=stable@dpdk.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).