Netdev Archive on lore.kernel.org
help / color / mirror / Atom feed
From: Dongdong Liu <liudongdong3@huawei.com>
To: Bjorn Helgaas <helgaas@kernel.org>
Cc: <hch@infradead.org>, <kw@linux.com>, <linux-pci@vger.kernel.org>,
	<rajur@chelsio.com>, <hverkuil-cisco@xs4all.nl>,
	<linux-media@vger.kernel.org>, <netdev@vger.kernel.org>,
	Logan Gunthorpe <logang@deltatee.com>
Subject: Re: [PATCH V5 4/6] PCI: Enable 10-Bit tag support for PCIe Endpoint devices
Date: Fri, 16 Jul 2021 19:12:16 +0800	[thread overview]
Message-ID: <db506d81-3cb9-4cdc-fb4a-f2d28587b9b2@huawei.com> (raw)
In-Reply-To: <20210715172336.GA1972959@bjorn-Precision-5520>

Hi Bjorn

Many thanks for your review.

On 2021/7/16 1:23, Bjorn Helgaas wrote:
> [+cc Logan]
>
> On Mon, Jun 21, 2021 at 06:27:20PM +0800, Dongdong Liu wrote:
>> 10-Bit Tag capability, introduced in PCIe-4.0 increases the total Tag
>> field size from 8 bits to 10 bits.
>>
>> For platforms where the RC supports 10-Bit Tag Completer capability,
>> it is highly recommended for platform firmware or operating software
>
> Recommended by whom?  If the spec recommends it, we should provide the
> citation.
PCIe spec 5.0 r1.0 section 2.2.6.2 IMPLEMENTATION NOTE says that.
Will fix.
>
>> that configures PCIe hierarchies to Set the 10-Bit Tag Requester Enable
>> bit automatically in Endpoints with 10-Bit Tag Requester capability. This
>> enables the important class of 10-Bit Tag capable adapters that send
>> Memory Read Requests only to host memory.
>
> What is the implication for P2PDMA?  What happens if we enable 10-bit
> tags for device A, and A generates Mem Read Requests to device B,
> which does not support 10-bit tags?
PCIe spec 5.0 r1.0 section 2.2.6.2 says
If an Endpoint supports sending Requests to other Endpoints (as opposed 
to host memory), the Endpoint must not send 10-Bit Tag Requests to 
another given Endpoint unless an implementation-specific mechanism 
determines that the Endpoint supports 10-Bit Tag Completer capability. 
Not sending 10-Bit Tag Requests to other Endpoints at all
may be acceptable for some implementations. More sophisticated 
mechanisms are outside the scope of this specification.

Not sending 10-Bit Tag Requests to other Endpoints at all seems simple.
Add kernel parameter pci=pcie_bus_peer2peer when boot kernel with 
P2PDMA, then do not config 10-BIT Tag.

if (pcie_bus_config != PCIE_BUS_PEER2PEER)
	pci_configure_10bit_tags(dev);

Bjorn and Logan, any suggestion?
>
>> Signed-off-by: Dongdong Liu <liudongdong3@huawei.com>
>> Reviewed-by: Christoph Hellwig <hch@lst.de>
>> ---
>>  drivers/pci/probe.c | 33 +++++++++++++++++++++++++++++++++
>>  include/linux/pci.h |  2 ++
>>  2 files changed, 35 insertions(+)
>>
>> diff --git a/drivers/pci/probe.c b/drivers/pci/probe.c
>> index 0208865..33241fb 100644
>> --- a/drivers/pci/probe.c
>> +++ b/drivers/pci/probe.c
>> @@ -2048,6 +2048,38 @@ int pci_configure_extended_tags(struct pci_dev *dev, void *ign)
>>  	return 0;
>>  }
>>
>> +static void pci_configure_10bit_tags(struct pci_dev *dev)
>> +{
>> +	struct pci_dev *bridge;
>> +
>> +	if (!(dev->pcie_devcap2 & PCI_EXP_DEVCAP2_10BIT_TAG_COMP))
>> +		return;
>> +
>> +	if (pci_pcie_type(dev) == PCI_EXP_TYPE_ROOT_PORT) {
>> +		dev->ext_10bit_tag = 1;
>> +		return;
>> +	}
>> +
>> +	bridge = pci_upstream_bridge(dev);
>> +	if (bridge && bridge->ext_10bit_tag)
>> +		dev->ext_10bit_tag = 1;
>> +
>> +	/*
>> +	 * 10-Bit Tag Requester Enable in Device Control 2 Register is RsvdP
>> +	 * for VF.
>> +	 */
>> +	if (dev->is_virtfn)
>> +		return;
>> +
>> +	if (pci_pcie_type(dev) == PCI_EXP_TYPE_ENDPOINT &&
>> +	    dev->ext_10bit_tag == 1 &&
>> +	    (dev->pcie_devcap2 & PCI_EXP_DEVCAP2_10BIT_TAG_REQ)) {
>> +		pci_dbg(dev, "enabling 10-Bit Tag Requester\n");
>> +		pcie_capability_set_word(dev, PCI_EXP_DEVCTL2,
>> +					PCI_EXP_DEVCTL2_10BIT_TAG_REQ_EN);
>> +	}
>> +}
>> +
>>  /**
>>   * pcie_relaxed_ordering_enabled - Probe for PCIe relaxed ordering enable
>>   * @dev: PCI device to query
>> @@ -2184,6 +2216,7 @@ static void pci_configure_device(struct pci_dev *dev)
>>  {
>>  	pci_configure_mps(dev);
>>  	pci_configure_extended_tags(dev, NULL);
>> +	pci_configure_10bit_tags(dev);
>
> I think 10-bit tag support should be integrated with extended (8-bit)
> tag support instead of having two separate functions.
>
> If we have "no_ext_tags" set because some device doesn't support 8-bit
> tags correctly, we probably shouldn't try to enable 10-bit tags
> either.
Looks good, will fix.

Thanks
Dongdong
>
>>  	pci_configure_relaxed_ordering(dev);
>>  	pci_configure_ltr(dev);
>>  	pci_configure_eetlp_prefix(dev);
>> diff --git a/include/linux/pci.h b/include/linux/pci.h
>> index de1fc24..445d102 100644
>> --- a/include/linux/pci.h
>> +++ b/include/linux/pci.h
>> @@ -393,6 +393,8 @@ struct pci_dev {
>>  #endif
>>  	unsigned int	eetlp_prefix_path:1;	/* End-to-End TLP Prefix */
>>
>> +	unsigned int	ext_10bit_tag:1; /* 10-Bit Tag Completer Supported
>> +					    from root to here */
>>  	pci_channel_state_t error_state;	/* Current connectivity state */
>>  	struct device	dev;			/* Generic device interface */
>>
>> --
>> 2.7.4
>>
> .
>

  reply	other threads:[~2021-07-16 11:12 UTC|newest]

Thread overview: 15+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-06-21 10:27 [PATCH V5 0/6] PCI: Enable 10-Bit tag support for PCIe devices Dongdong Liu
2021-06-21 10:27 ` [PATCH V5 1/6] PCI: Use cached Device Capabilities Register Dongdong Liu
2021-06-21 10:27 ` [PATCH V5 2/6] PCI: Use cached Device Capabilities 2 Register Dongdong Liu
2021-06-22  6:17   ` Christoph Hellwig
2021-06-21 10:27 ` [PATCH V5 3/6] PCI: Add 10-Bit Tag register definitions Dongdong Liu
2021-06-21 10:27 ` [PATCH V5 4/6] PCI: Enable 10-Bit tag support for PCIe Endpoint devices Dongdong Liu
2021-07-15 17:23   ` Bjorn Helgaas
2021-07-16 11:12     ` Dongdong Liu [this message]
2021-07-16 14:17       ` Bjorn Helgaas
2021-07-17  8:50         ` Dongdong Liu
2021-07-16 15:51       ` Logan Gunthorpe
2021-07-17  9:41         ` Dongdong Liu
2021-06-21 10:27 ` [PATCH V5 5/6] PCI/IOV: Enable 10-Bit tag support for PCIe VF devices Dongdong Liu
2021-06-21 10:27 ` [PATCH V5 6/6] PCI: Enable 10-Bit tag support for PCIe RP devices Dongdong Liu
2021-06-22  6:19   ` Christoph Hellwig

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=db506d81-3cb9-4cdc-fb4a-f2d28587b9b2@huawei.com \
    --to=liudongdong3@huawei.com \
    --cc=hch@infradead.org \
    --cc=helgaas@kernel.org \
    --cc=hverkuil-cisco@xs4all.nl \
    --cc=kw@linux.com \
    --cc=linux-media@vger.kernel.org \
    --cc=linux-pci@vger.kernel.org \
    --cc=logang@deltatee.com \
    --cc=netdev@vger.kernel.org \
    --cc=rajur@chelsio.com \
    --subject='Re: [PATCH V5 4/6] PCI: Enable 10-Bit tag support for PCIe Endpoint devices' \
    /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

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).