linux-kernel - Re: [PATCH v2 00/16] usb: dwc3: add support for AMD NL SoC

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:	Fri, 17 Oct 2014 10:10:26 -0500
From:	Felipe Balbi <balbi@...com>
To:	Huang Rui <ray.huang@....com>
CC:	Felipe Balbi <balbi@...com>,
	Alan Stern <stern@...land.harvard.edu>,
	Bjorn Helgaas <bhelgaas@...gle.com>,
	Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
	Paul Zimmerman <Paul.Zimmerman@...opsys.com>,
	Heikki Krogerus <heikki.krogerus@...ux.intel.com>,
	Vincent Wan <vincent.wan@....com>, Tony Li <tony.li@....com>,
	<linux-usb@...r.kernel.org>, <linux-pci@...r.kernel.org>,
	<linux-kernel@...r.kernel.org>
Subject: Re: [PATCH v2 00/16] usb: dwc3: add support for AMD NL SoC

Hi,

On Fri, Oct 17, 2014 at 04:53:25PM +0800, Huang Rui wrote:
> The series of patches add AMD NL SoC support for DesignWare USB3 OTG
> IP with PCI bus glue layer. This controller supported hibernation, LPM
> erratum and used the 2.80a IP version and amd own phy. Current
> implementation support both simulation and SoC platform. And already
> tested with gadget zero and msc tool.  It works well on file storage
> gadget.

patches look much, much nicer there are still a few things to fix. A
global set of issues which I see:

1) Let's get confirmation that all those quirks will be needed in
production as well, I have a feeling quite a few of them won't be.

2) All quirks should become 1-bit fields insteads of single-bits on a
32-bit variable.

3) All quirks should have DeviceTree counterparts. They should all
become boolean properties should we can:

	dwc->tx_deemphasis_quirk = of_property_read_bool(node,
		"snps,tx_deemphasis_quirk");

> These patches are generated on balbi/testing/next
> 
> Changes from v1 -> v2
> - remove dual role function temporarily
> - add pci quirk to avoid to bind with xhci driver
> - distinguish between simulation board and soc
> - break down all the special quirks
> 
> 
> Patch 1:
> - add PCI device id into pci bus glue

this guy should be the last in the series, with all AMD quirks being
enabled at once. This will avoid bisection points where AMD's platforms
don't work.

> Patch 2:

this should become as patch one :-)

> - add PCI quirk to avoid to bind with xhci
> 
> Patch 3:
> - enable hibernation
> 
> Patch 4:
> - distinguish between simulation board and soc
> 
> Patch 5:
> - add quirks flag to be compatible for kinds of soc
> 
> Patch 6 - 16:
> - as felipe's suggestion, break down all the special quirks of amd nl
> 
> 
> Patch set already passed all the MSC testing, detailed result is below:
> 
> root@...ub:/home/ray/felipe/usb-tools# ./msc.sh -o /dev/sdb1
> test 0e: simple 64k read/write
> test  0: sent      62.50 MB read      22.00 MB/s write      16.33 MB/s ... success

Are you still running with VERBOSE_DEBUG on USB2 ? Here's what I get on
USB2 connected to my PC with DWC3 running on a single-core cortex-a9
board:

$ ./msc -t0 -o /dev/sdh -s 65536 -c 5000
test  0: sent     312.50 MB read      31.63 MB/s write      29.10 MB/s ... success

And with RAM as backend:

$ ./msc -t0 -o /dev/sdh -s 65536 -c 5000
test  0: sent     312.50 MB read      31.64 MB/s write      29.04 MB/s ... success

cheers

-- 
balbi

Download attachment "signature.asc" of type "application/pgp-signature" (820 bytes)