From: Hans de Goede <hdegoede@redhat.com>
To: Tom Yan <tom.ty89@gmail.com>, Alan Stern <stern@rowland.harvard.edu>
Cc: Christoph Hellwig <hch@lst.de>,
Greg KH <gregkh@linuxfoundation.org>,
linux-usb <linux-usb@vger.kernel.org>,
Mathias Nyman <mathias.nyman@intel.com>,
Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
linux-pci@vger.kernel.org, Lu Baolu <baolu.lu@linux.intel.com>
Subject: Re: [PATCH 1/2] uas: revert from scsi_add_host_with_dma() to scsi_add_host()
Date: Tue, 1 Dec 2020 12:09:54 +0100 [thread overview]
Message-ID: <899cea7f-b306-0ce4-06e9-aa5d083573cb@redhat.com> (raw)
In-Reply-To: <CAGnHSE=sS7tvttuTwE_s+QbCUVCfhmHnuXQp1g1AkZ=JEoxmQA@mail.gmail.com>
Hi,
On 11/30/20 8:30 PM, Tom Yan wrote:
> Hmm, I wonder if I/we wrongly assumed that the dma_dev used for the
> hw_max_sectors clamping in __scsi_init_queue() is wrong.
>
> So instead of adding a fallback else-clause here or using "sysdev" as
> dma_dev like in the current upstream code, maybe we should actually do
> a three-way min: the "changed" hw_max_sectors, dma_max_mapping_size of
> dma_dev("dev") and dma_max_mapping_size of sysdev...?
So both here and in the 2/2 patch thread there are lots of open questions,
which to me suggests that for 5.10 we really should just go with the 3 reverts
which I suggested earlier.
Regards,
Hans
>
> On Mon, 30 Nov 2020 at 17:48, Hans de Goede <hdegoede@redhat.com> wrote:
>>
>> Hi,
>>
>> On 11/28/20 4:48 PM, Tom Yan wrote:
>>> Apparently the former (with the chosen dma_dev) may cause problem in certain
>>> case (e.g. where thunderbolt dock and intel iommu are involved). The error
>>> observed was:
>>>
>>> XHCI swiotlb buffer is full / DMAR: Device bounce map failed
>>>
>>> For now we retain the clamp for hw_max_sectors against the dma_max_mapping_size.
>>> Since the device/size for the clamp that is applied when the scsi request queue
>>> is initialized/allocated is different than the one used here, we invalidate the
>>> early clamping by making a fallback blk_queue_max_hw_sectors() call.
>>>
>>> Signed-off-by: Tom Yan <tom.ty89@gmail.com>
>>
>> I can confirm that this fixes the network performance on a Lenovo Thunderbolt
>> dock generation 2, which uses an USB attach NIC.
>>
>> With this patch added on top of 5.10-rc5 scp performance to another machine
>> on the local gbit LAN goes back from the regressed 1 MB/s to its original 100MB/s
>> as it should be:
>>
>> Tested-by: Hans de Goede <hdegoede@redhat.com>
>>
>> Regards,
>>
>> Hans
>>
>>
>>> ---
>>> drivers/usb/storage/uas.c | 11 +++++++----
>>> 1 file changed, 7 insertions(+), 4 deletions(-)
>>>
>>> diff --git a/drivers/usb/storage/uas.c b/drivers/usb/storage/uas.c
>>> index c8a577309e8f..5db1325cea20 100644
>>> --- a/drivers/usb/storage/uas.c
>>> +++ b/drivers/usb/storage/uas.c
>>> @@ -843,18 +843,21 @@ static int uas_slave_alloc(struct scsi_device *sdev)
>>> static int uas_slave_configure(struct scsi_device *sdev)
>>> {
>>> struct uas_dev_info *devinfo = sdev->hostdata;
>>> - struct device *dev = sdev->host->dma_dev;
>>> + struct usb_device *udev = devinfo->udev;
>>>
>>> if (devinfo->flags & US_FL_MAX_SECTORS_64)
>>> blk_queue_max_hw_sectors(sdev->request_queue, 64);
>>> else if (devinfo->flags & US_FL_MAX_SECTORS_240)
>>> blk_queue_max_hw_sectors(sdev->request_queue, 240);
>>> - else if (devinfo->udev->speed >= USB_SPEED_SUPER)
>>> + else if (udev->speed >= USB_SPEED_SUPER)
>>> blk_queue_max_hw_sectors(sdev->request_queue, 2048);
>>> + else
>>> + blk_queue_max_hw_sectors(sdev->request_queue,
>>> + SCSI_DEFAULT_MAX_SECTORS);
>>>
>>> blk_queue_max_hw_sectors(sdev->request_queue,
>>> min_t(size_t, queue_max_hw_sectors(sdev->request_queue),
>>> - dma_max_mapping_size(dev) >> SECTOR_SHIFT));
>>> + dma_max_mapping_size(udev->bus->sysdev) >> SECTOR_SHIFT));
>>>
>>> if (devinfo->flags & US_FL_NO_REPORT_OPCODES)
>>> sdev->no_report_opcodes = 1;
>>> @@ -1040,7 +1043,7 @@ static int uas_probe(struct usb_interface *intf, const struct usb_device_id *id)
>>> shost->can_queue = devinfo->qdepth - 2;
>>>
>>> usb_set_intfdata(intf, shost);
>>> - result = scsi_add_host_with_dma(shost, &intf->dev, udev->bus->sysdev);
>>> + result = scsi_add_host(shost, &intf->dev);
>>> if (result)
>>> goto free_streams;
>>>
>>>
>>
>
next prev parent reply other threads:[~2020-12-01 11:12 UTC|newest]
Thread overview: 37+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-11-10 11:36 5.10 regression, many XHCI swiotlb buffer is full / DMAR: Device bounce map failed errors on thunderbolt connected XHCI controller Hans de Goede
2020-11-18 21:43 ` Hans de Goede
2020-11-23 14:49 ` Hans de Goede
2020-11-24 10:27 ` Christoph Hellwig
2020-11-24 10:31 ` Hans de Goede
2020-11-24 12:17 ` Mathias Nyman
2020-11-27 11:41 ` Hans de Goede
2020-11-27 12:32 ` 5.10 regression caused by: "uas: fix sdev->host->dma_dev": " Hans de Goede
2020-11-27 16:19 ` Christoph Hellwig
2020-11-27 18:12 ` Hans de Goede
2020-11-28 1:25 ` Tom Yan
2020-11-28 10:43 ` Hans de Goede
2020-11-28 15:48 ` [PATCH 1/2] uas: revert from scsi_add_host_with_dma() to scsi_add_host() Tom Yan
2020-11-28 15:48 ` [PATCH 2/2] usb-storage: " Tom Yan
2020-11-30 9:50 ` Hans de Goede
2020-11-30 12:58 ` Tom Yan
2020-11-30 13:23 ` Hans de Goede
2020-11-30 13:30 ` Greg KH
2020-11-30 13:36 ` Hans de Goede
2020-11-30 13:53 ` Greg KH
2020-11-30 13:55 ` Hans de Goede
2020-12-04 15:02 ` Greg KH
2020-11-30 17:20 ` Alan Stern
2020-11-30 17:24 ` Christoph Hellwig
2020-11-30 18:18 ` Hans de Goede
2020-11-30 18:57 ` Tom Yan
2020-11-30 19:01 ` Tom Yan
2020-11-30 20:36 ` Alan Stern
2021-02-25 16:35 ` Alan Stern
2021-02-26 5:53 ` Christoph Hellwig
2021-03-01 15:59 ` Alan Stern
2020-11-30 14:39 ` Tom Yan
2020-11-30 9:48 ` [PATCH 1/2] uas: " Hans de Goede
2020-11-30 19:30 ` Tom Yan
2020-12-01 11:09 ` Hans de Goede [this message]
2020-11-28 17:15 ` 5.10 regression caused by: "uas: fix sdev->host->dma_dev": many XHCI swiotlb buffer is full / DMAR: Device bounce map failed errors on thunderbolt connected XHCI controller Christoph Hellwig
2020-11-30 8:43 ` Hans de Goede
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=899cea7f-b306-0ce4-06e9-aa5d083573cb@redhat.com \
--to=hdegoede@redhat.com \
--cc=baolu.lu@linux.intel.com \
--cc=gregkh@linuxfoundation.org \
--cc=hch@lst.de \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-pci@vger.kernel.org \
--cc=linux-usb@vger.kernel.org \
--cc=mathias.nyman@intel.com \
--cc=stern@rowland.harvard.edu \
--cc=tom.ty89@gmail.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).