Commit graph

861 commits

Author SHA1 Message Date
Ye Li
1468a1cc72 usb: ci_udc: Add function to remove usb device
When unregister gadget driver in ci_udc, the usb device is not
removed or stop. This causes next "usb start" fails to work.

Add a new interface "usb_remove_ehci_gadget" in usb-uclass to
remove the usb device for DM driver. Using "usb_lowlevel_stop" for
non-DM driver.

Signed-off-by: Ye Li <ye.li@nxp.com>
Signed-off-by: Peng Fan <peng.fan@nxp.com>
2020-09-01 14:47:43 +02:00
Masahiro Yamada
702e57e113 treewide: convert devfdt_get_addr_ptr() to dev_read_addr_ptr()
When you enable CONFIG_OF_LIVE, you will end up with a lot of
conversions.

To help this tedious work, this commit converts devfdt_get_addr_ptr()
to dev_read_addr_ptr() by coccinelle. I also removed redundant casts
because dev_read_addr_ptr() returns an opaque pointer.

To generate this commit, I ran the following semantic patch
excluding include/dm/.

  <smpl>
  @@
  type T;
  expression dev;
  @@
  -(T *)devfdt_get_addr_ptr(dev)
  +dev_read_addr_ptr(dev)
  @@
  expression dev;
  @@
  -devfdt_get_addr_ptr(dev)
  +dev_read_addr_ptr(dev)
  </smpl>

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
2020-08-22 08:53:37 -06:00
Patrice Chotard
f3858ce029 usb: host: dwc3-sti-glue: Fix ofnode_valid() parameter
node varaible is used as iterator into ofnode_for_each_subnode()
loop, when exiting of it, node is no more a valid ofnode.
Use dwc3_node instead as parameter of ofnode_valid()

Fixes: ac28e59a57 ("usb: Migrate to support live DT for some driver")
Signed-off-by: Patrice Chotard <patrice.chotard@st.com>
Cc: Kever Yang <kever.yang@rock-chips.com>
2020-08-13 09:19:18 +02:00
Stefan Roese
b5152a653d usb: xhci: Add virt_to_phys() to support mapped platforms
Some platforms, like MIPS Octeon, use mapped addresses (virtual address
!= physical address). On these platforms we need to make sure, that the
local virtual addresses are converted to physical (DMA) addresses for
the xHCI controller. This patch adds the missing virt_to_phys() calls,
so that the correct addresses are used.

Signed-off-by: Stefan Roese <sr@denx.de>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Tested-by: Bin Meng <bmeng.cn@gmail.com>
Cc: Bin Meng <bmeng.cn@gmail.com>
Cc: Marek Vasut <marex@denx.de>
2020-08-05 09:30:46 +02:00
Stefan Roese
d96f6e1c0b usb: usb-uclass.c: Drop le16_to_cpu() as values are already swapped
These values are already swapped to CPU endianess, so swapping them
again is a bug. Let's remove the swap here instead.

Signed-off-by: Stefan Roese <sr@denx.de>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Tested-by: Bin Meng <bmeng.cn@gmail.com>
Cc: Bin Meng <bmeng.cn@gmail.com>
Cc: Marek Vasut <marex@denx.de>
2020-08-05 09:30:46 +02:00
Stefan Roese
61a1acb55e usb: xhci: xhci_mem_init: Use cpu_to_le64() and not xhci_writeq()
xhci_writeq() makes the CPU->LE swapping only when addressing registers
in the xHCI controller address range and not in the local memory (RAM).
We need to use cpu_to_le64() here to ensure that the conversion is done
correctly.

Signed-off-by: Stefan Roese <sr@denx.de>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Tested-by: Bin Meng <bmeng.cn@gmail.com>
Cc: Bin Meng <bmeng.cn@gmail.com>
Cc: Marek Vasut <marex@denx.de>
2020-08-05 09:30:46 +02:00
Stefan Roese
543eb12ecd usb: xhci: Add missing endian conversions (cpu_to_leXX / leXX_to_cpu)
While trying to use the U-Boot xHCI driver on the MIPS Octeon platform,
which is big endian, I noticed that the driver is missing a few endian
conversion calls. This patch adds these missing endian conversion
calls.

Signed-off-by: Stefan Roese <sr@denx.de>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Tested-by: Bin Meng <bmeng.cn@gmail.com>
Cc: Bin Meng <bmeng.cn@gmail.com>
Cc: Marek Vasut <marex@denx.de>
2020-08-05 09:30:46 +02:00
Masahiro Yamada
2548493ab4 treewide: convert devfdt_get_addr() to dev_read_addr()
When you enable CONFIG_OF_LIVE, you will end up with a lot of
conversions.

To generate this commit, I used coccinelle excluding drivers/core/,
include/dm/, and test/

The semantic patch that makes this change is as follows:

  <smpl>
  @@
  expression dev;
  @@
  -devfdt_get_addr(dev)
  +dev_read_addr(dev)
  </smpl>

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
2020-07-25 14:46:57 -06:00
Masahiro Yamada
8613c8d897 treewide: convert (void *)devfdt_get_addr() to dev_read_addr_ptr()
Use the _ptr suffixed variant instead of casting. Also, convert it to
dev_read_addr_ptr(), which is safe to CONFIG_OF_LIVE.

One curious part is an error check like follows in
drivers/watchdog/omap_wdt.c:

    priv->regs = (struct wd_timer *)devfdt_get_addr(dev);
    if (!priv->regs)
            return -EINVAL;

devfdt_get_addr() returns FDT_ADDR_T_NONE (i.e. -1) on error.
So, this code does not catch any error in DT parsing.

dev_read_addr_ptr() returns NULL on error, so this error check
will work.

I generated this commit by the following command:

 $ find . -name .git -prune -o -name '*.[ch]' -type f -print | \
   xargs sed -i -e 's/([^*)]*\*)devfdt_get_addr(/dev_read_addr_ptr(/'

I manually fixed drivers/usb/host/ehci-mx6.c

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
2020-07-25 14:46:57 -06:00
Tom Rini
7208396bbf Revert "Merge tag 'dm-pull-20jul20' of git://git.denx.de/u-boot-dm"
This reverts commit 5d3a21df66, reversing
changes made to 56d37f1c56.

Unfortunately this is causing CI failures:
https://travis-ci.org/github/trini/u-boot/jobs/711313649

Signed-off-by: Tom Rini <trini@konsulko.com>
2020-07-24 08:42:06 -04:00
Masahiro Yamada
60e7fa8b3b treewide: convert devfdt_get_addr() to dev_read_addr()
When you enable CONFIG_OF_LIVE, you will end up with a lot of
conversions.

To generate this commit, I used coccinelle excluding drivers/core/,
include/dm/, and test/

The semantic patch that makes this change is as follows:

  <smpl>
  @@
  expression dev;
  @@
  -devfdt_get_addr(dev)
  +dev_read_addr(dev)
  </smpl>

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
2020-07-20 11:37:47 -06:00
Masahiro Yamada
3c12c62ba5 treewide: convert (void *)devfdt_get_addr() to dev_read_addr_ptr()
Use the _ptr suffixed variant instead of casting. Also, convert it to
dev_read_addr_ptr(), which is safe to CONFIG_OF_LIVE.

One curious part is an error check like follows in
drivers/watchdog/omap_wdt.c:

    priv->regs = (struct wd_timer *)devfdt_get_addr(dev);
    if (!priv->regs)
            return -EINVAL;

devfdt_get_addr() returns FDT_ADDR_T_NONE (i.e. -1) on error.
So, this code does not catch any error in DT parsing.

dev_read_addr_ptr() returns NULL on error, so this error check
will work.

I generated this commit by the following command:

 $ find . -name .git -prune -o -name '*.[ch]' -type f -print | \
   xargs sed -i -e 's/([^*)]*\*)devfdt_get_addr(/dev_read_addr_ptr(/'

I manually fixed drivers/usb/host/ehci-mx6.c

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
2020-07-20 11:37:47 -06:00
Nicolas Saenz Julienne
0b80371b35 usb: xhci: Add reset controller support
Some atypical users of xhci might need to manually reset their xHCI
controller before starting the HCD setup. Check if a reset controller
device is available to the PCI bus and trigger a reset.

Signed-off-by: Nicolas Saenz Julienne <nsaenzjulienne@suse.de>
[mb: squash fix to only build xhci_reset_hw() if CONFIG_DM_BUS]
Signed-off-by: Matthias Brugger <mbrugger@suse.com>
2020-07-10 11:49:28 +02:00
Sylwester Nawrocki
61293f519e usb: xhci: Add missing cache flush in the scratchpad array initialization
In current code there is no cache flush after initializing the scratchpad
buffer array with the scratchpad buffer pointers. This leads to a failure
of the "slot enable" command on the rpi4 board (Broadcom STB PCIe
controller + VL805 USB hub) - the very first TRB transfer on the command
ring fails and there is a timeout while waiting for the command completion
event. After adding the missing cache flush everything seems to be working
as expected.

Signed-off-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Nicolas Saenz Julienne <nsaenzjulienne@suse.de>
Signed-off-by: Matthias Brugger <mbrugger@suse.com>
2020-07-09 15:46:12 +02:00
Michael Walle
c03b7612ea usb: provide a device tree node to USB devices
It is possible to specify a device tree node for an USB device. This is
useful if you have a static USB setup and want to use aliases which
point to these nodes, like on the Raspberry Pi.
The nodes are matched against their hub port number, the compatible
strings are not matched for now.

Signed-off-by: Michael Walle <michael@walle.cc>
Reviewed-by: Marek Vasut <marex@denx.de>
Reviewed-by: Simon Glass <sjg@chromium.org>
2020-06-11 20:52:11 -06:00
Simon Glass
72be237561 usb: ohci: Add header file for ARCH_DMA_MINALIGN
This is defined in the asm/cache.h header file. Update this header file to
include it so it gets the same value consistently across U-Boot.

This fixes 'usb host' on omapl138_lcdk.

Fixes: 90526e9fba ("common: Drop net.h from common header")
Reported-by: Adam Ford <aford173@gmail.com>
Signed-off-by: Simon Glass <sjg@chromium.org>
Tested-by: Adam Ford <aford173@gmail.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
2020-06-02 13:06:10 -04:00
Tom Rini
9452b7496f Merge tag 'u-boot-rockchip-20200531' of https://gitlab.denx.de/u-boot/custodians/u-boot-rockchip
- Fix mmc of path after syncfrom kernel dts;
- Add dwc3 host support with DM for rk3399;
- Add usb2phy and typec phy for rockchip platform;
- Migrate board list doc to rockchip.rst;
- Add rk3399 Pinebook Pro board support;
- Update dram_init in board_init and add memory node in SPL;
2020-05-31 20:07:39 -04:00
Frank Wang
604ab3ecda driver: usb: drop legacy rockchip xhci driver
We have changed to use dwc3 generic driver for usb3.0 host, so the
legacy Rockchip's xHCI driver is not needed, and drop it.

Signed-off-by: Frank Wang <frank.wang@rock-chips.com>
Reviewed-by: Jagan Teki <jagan@amarulasolutions.com>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
2020-05-30 07:38:50 +08:00
Marek Vasut
73021d11d4 usb: ehci-mx6: Print error code on failure
Print the error code if the regulator enable fails, otherwise the error
message is rather useless and confusing.

Signed-off-by: Marek Vasut <marex@denx.de>
2020-05-29 19:23:36 +02:00
Marek Vasut
10bcafb8ac usb: ehci-mx6: Handle fixed regulators correctly
The regulator-fixed would return -ENOSYS when enabled/disabled,
because this operation is not supported, but this is not an error
e.g. on systems where the VBUS cannot be controlled, so if this
is the error code reported by the regulator core, consider it a
success and continue.

Signed-off-by: Marek Vasut <marex@denx.de>
2020-05-29 19:18:55 +02:00
Chunfeng Yun
58221d7e87 usb: dwc3: fix NULL pointer issue
The phy_bulk pointer *usb_phys is used before allocated,
fix it by using a phy_bulk variable instead in
xhci_dwc3_platdata struct

Signed-off-by: Chunfeng Yun <chunfeng.yun@mediatek.com>
2020-05-22 15:22:36 +02:00
Adam Ford
94ed66194f usb: ehci-omap: Add Support for DM_USB and OF_CONTROL
The omap3.dtsi file shows the usbhshost node with two sub-nodes
for ohci and ehci.  This patch file creates the usbhshost, and
pulls the portX-mode information.  It then locates the EHCI
sub-node, and initializes the EHCI controller with the info
pulled from the usbhshost node.

There is still more to do since there isn't an actual link
between the 'phys' reference and the corresponding phy driver,
and there is no nop-xceiv driver yet.

In the meantime, the older style reference to
CONFIG_OMAP_EHCI_PHYx_RESET_GPIO is still needed to pull
the phy out of reset until the phy driver is completed and the
phandle reference is made.

Signed-off-by: Adam Ford <aford173@gmail.com>
2020-05-22 15:22:35 +02:00
Simon Glass
cd93d625fd common: Drop linux/bitops.h from common header
Move this uncommon header out of the common header.

Signed-off-by: Simon Glass <sjg@chromium.org>
2020-05-18 21:19:23 -04:00
Simon Glass
c05ed00afb common: Drop linux/delay.h from common header
Move this uncommon header out of the common header.

Signed-off-by: Simon Glass <sjg@chromium.org>
2020-05-18 21:19:23 -04:00
Simon Glass
eb41d8a1be common: Drop linux/bug.h from common header
Move this uncommon header out of the common header.

Signed-off-by: Simon Glass <sjg@chromium.org>
2020-05-18 21:19:23 -04:00
Simon Glass
f7ae49fc4f common: Drop log.h from common header
Move this header out of the common header.

Signed-off-by: Simon Glass <sjg@chromium.org>
2020-05-18 21:19:18 -04:00
Simon Glass
691d719db7 common: Drop init.h from common header
Move this uncommon header out of the common header.

Signed-off-by: Simon Glass <sjg@chromium.org>
2020-05-18 17:33:33 -04:00
Simon Glass
90526e9fba common: Drop net.h from common header
Move this header out of the common header. Network support is used in
quite a few places but it still does not warrant blanket inclusion.

Note that this net.h header itself has quite a lot in it. It could be
split into the driver-mode support, functions, structures, checksumming,
etc.

Signed-off-by: Simon Glass <sjg@chromium.org>
2020-05-18 17:33:31 -04:00
Trevor Woerner
18138ab203 rename symbol: CONFIG_TEGRA -> CONFIG_ARCH_TEGRA
Have this symbol follow the pattern of all other such symbols.

Signed-off-by: Trevor Woerner <twoerner@gmail.com>
2020-05-15 14:47:35 -04:00
Trevor Woerner
bb0fb4c0f4 rename symbol: CONFIG_KIRKWOOD -> CONFIG_ARCH_KIRKWOOD
Have this symbol follow the pattern of all other such symbols.
This patch also removes a TODO from the code.

Reviewed-by: Chris Packham <judge.packham@gmail.com>
Signed-off-by: Trevor Woerner <twoerner@gmail.com>
2020-05-15 14:47:35 -04:00
Trevor Woerner
b16a331697 rename symbol: CONFIG_ORION5X -> CONFIG_ARCH_ORION5X
Have this symbol follow the pattern of all other such symbols.
This patch removes a TODO from the code.

Signed-off-by: Trevor Woerner <twoerner@gmail.com>
2020-05-15 14:47:35 -04:00
Peng Fan
0bd3d911df usb: mx6: add fuse check
Add fuse check for USB. If the fuse indicates the module
will not work in the SoC, let's fail the initialization.

Signed-off-by: Peng Fan <peng.fan@nxp.com>
2020-05-10 13:21:27 +02:00
Chunfeng Yun
7410283415 xhci: mediatek: Add support for MTK xHCI host controller
This patch is used to support the on-chip xHCI controller on
MediaTek SoCs, currently control/bulk/interrupt transfers are
supported.

Signed-off-by: Chunfeng Yun <chunfeng.yun@mediatek.com>
Signed-off-by: Frank Wunderlich <frank-w@public-files.de>
Reviewed-by: Weijie Gao <weijie.gao@mediatek.com>
Reviewed-by: Jagan Teki <jagan@amarulasolutions.com>
2020-05-02 12:32:28 +02:00
Chunfeng Yun
6dfb8a8052 usb: dwc3: use the phy bulk API to get phys
Get a group of phys by the phy bulk API

Signed-off-by: Chunfeng Yun <chunfeng.yun@mediatek.com>
Signed-off-by: Frank Wunderlich <frank-w@public-files.de>
Reviewed-by: Weijie Gao <weijie.gao@mediatek.com>
2020-05-02 12:32:28 +02:00
Patrice Chotard
f3bc736e41 usb: host: dwc3-sti-glue: Use UCLASS_NOP instead of UCLASS_MISC
dwc3-sti-glue has been broken since MISC uclass has been
modified to scan DT sub-nodes after bind.
Fixing it by a using the no-op uclass.

Signed-off-by: Patrice Chotard <patrice.chotard@st.com>
2020-04-28 13:52:52 +02:00
Patrick Delaunay
245847f065 usb: host: dwc2: add trace to have clean usb start
Solve issue for the display of "usb start" command on stm32mp1
because one carriage return is missing in DWC2 probe.

Before the patch:

STM32MP> usb start
starting USB...
Bus usb-otg@49000000:    Bus usbh-ehci@5800d000:   USB EHCI 1.00

after the patch:

STM32MP> usb start
starting USB...
Bus usb-otg@49000000: USB DWC2
Bus usbh-ehci@5800d000: USB EHCI 1.00

Signed-off-by: Patrick Delaunay <patrick.delaunay@st.com>
2020-04-28 13:52:52 +02:00
Patrick Delaunay
6600438166 usb: host: dwc2: force reset assert
Assert reset before deassert in dwc2_reset;
this patch solve issues when the DWC2 registers are already
initialized with value incompatible with host mode.

Force a hardware reset of the IP reset all the DWC2 registers at
default value, the host driver start with a clean state
(Core Soft reset doen in dwc_otg_core_reset is not enought
 to reset all register).

The error can occurs in U-Boot when DWC2 device gadget driver
force device mode (called by ums or dfu command, before to execute
the usb start command).

Signed-off-by: Patrick Delaunay <patrick.delaunay@st.com>
2020-04-28 13:52:52 +02:00
Patrick Delaunay
0bc632c9b1 usb: host: dwc2: add clk support
Add support for clock with driver model.

This patch don't added dependency because when CONFIG_CLK
is not activated the clk function are stubbed.

Reviewed-by: Simon Goldschmidt <simon.k.r.goldschmidt@gmail.com>
Signed-off-by: Patrick Delaunay <patrick.delaunay@st.com>
2020-04-28 13:52:52 +02:00
Patrick Delaunay
e17a4bf198 usb: host: dwc2: add phy support
Use generic phy to initialize the PHY associated to the
DWC2 device and available in the device tree.

This patch don't added dependency because when CONFIG_PHY
is not activated, the generic PHY function are stubbed.

Reviewed-by: Simon Goldschmidt <simon.k.r.goldschmidt@gmail.com>
Signed-off-by: Patrick Delaunay <patrick.delaunay@st.com>
2020-04-28 13:52:52 +02:00
Heinrich Schuchardt
cff0144e4c usb: avoid NULL check before free
The free() function checks if the argument is NULL.
Do not duplicate this check.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
2020-04-28 13:52:52 +02:00
Rayagonda Kokatanur
9cadf05958 drivers: usb: host: Add BRCM xHCI driver
Base driver for Broadcom xHCI controllers

Signed-off-by: Bharat Kumar Reddy Gooty <bharat.gooty@broadcom.com>
Signed-off-by: Rayagonda Kokatanur <rayagonda.kokatanur@broadcom.com>
Signed-off-by: Vladimir Olovyannikov <vladimir.olovyannikov@broadcom.com>
2020-04-13 13:48:41 +02:00
Kever Yang
2517deafc2 usb: Migrate to support live DT for some driver
Use ofnode_ instead of fdt_ APIs so that the drivers can support live DT.
This patch updates usb_get_dr_mode() and usb_get_maximum_speed() to use
ofnode as parameter instead of fdt offset. And all the drivers who use
these APIs update to use live dt APIs at the same time.

Signed-off-by: Kever Yang <kever.yang@rock-chips.com>
2020-04-13 13:48:30 +02:00
Kever Yang
0148bbe8f3 usb: ehci-msm: Use dev interface to get device address
Use dev_read_addr_ptr() instead of devfdt_get_addr() so that we can support
live DT.

Signed-off-by: Kever Yang <kever.yang@rock-chips.com>
Reviewed-by: Ramon Fried <rfried.dev@gmail.com>
2020-04-13 13:48:30 +02:00
Kever Yang
5c17030358 usb: dwc3-of-simple: Drop redundant inclding header file
The fdtdec.h is no use in this file, remove the include code.

Signed-off-by: Kever Yang <kever.yang@rock-chips.com>
2020-04-13 13:48:30 +02:00
Marek Vasut
31232de07e usb: Keep async schedule running only across mass storage xfers
Rather than keeping the asynchronous schedule running always, keep it
running only across USB mass storage transfers for now, as it seems
that keeping it running all the time interferes with certain control
transfers during device enumeration.

Note that running the async schedule all the time should not be an
issue, especially on EHCI HCD, as that one implements most of the
transfers using async schedule.

Note that we have usb_disable_asynch(), which however is utterly broken.
The usb_disable_asynch() blocks the USB core from doing async transfers
by setting a global flag. The async schedule should however be disabled
per USB controller. Moreover, setting a global flag does not prevent the
controller from using the async schedule, which e.g. the EHCI HCD does.

This patch implements additional callback to the controller, which
permits it to lock the async schedule and keep it running across
multiple transfers. Once the schedule is unlocked, it must also be
disabled. This thus prevents the async schedule from running outside
of the USB mass storage transfers.

Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Cc: Lukasz Majewski <lukma@denx.de>
Cc: Tom Rini <trini@konsulko.com>
Tested-by: Tom Rini <trini@konsulko.com> [omap3_beagle, previously failing]
2020-04-09 15:26:59 -04:00
Simon Glass
336d4615f8 dm: core: Create a new header file for 'compat' features
At present dm/device.h includes the linux-compatible features. This
requires including linux/compat.h which in turn includes a lot of headers.
One of these is malloc.h which we thus end up including in every file in
U-Boot. Apart from the inefficiency of this, it is problematic for sandbox
which needs to use the system malloc() in some files.

Move the compatibility features into a separate header file.

Signed-off-by: Simon Glass <sjg@chromium.org>
2020-02-05 19:33:46 -07:00
Simon Glass
61b29b8268 dm: core: Require users of devres to include the header
At present devres.h is included in all files that include dm.h but few
make use of it. Also this pulls in linux/compat which adds several more
headers. Drop the automatic inclusion and require files to include devres
themselves. This provides a good indication of which files use devres.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Anatolij Gustschin <agust@denx.de>
2020-02-05 19:33:46 -07:00
Simon Glass
1eb69ae498 common: Move ARM cache operations out of common.h
These functions are CPU-related and do not use driver model. Move them to
cpu_func.h

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
2019-12-02 18:24:58 -05:00
Chunfeng Yun
e0891bb679 usb: xhci: support 1.1 or later version
The xHCI 1.1 version also need set Transfer Type field

Signed-off-by: Chunfeng Yun <chunfeng.yun@mediatek.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
2019-11-08 11:09:08 +01:00
Marek Vasut
02b0e1a36c usb: ehci-hcd: Keep async schedule running
Profiling the EHCI driver shows a significant performance problem in
ehci_submit_async(). Specifically, this function keeps enabling and
disabling async schedule back and forth for every single transaction.
However, enabling/disabling the async schedule does not take effect
immediatelly, but instead may take up to 1 mS (8 uFrames) to complete.

This impacts USB storage significantly, esp. since the recent reduction
of maximum transfer size to support more USB storage devices. This in
turn results in sharp increase in the number of ehci_submit_async()
calls. Since one USB storage BBB transfer does three such calls and
the maximum transfer size is 120 kiB, the overhead is 6 mS per 120 kiB,
which is unacceptable.

However, this overhead can be removed simply by keeping the async
schedule running. Specifically, the first transfer starts the async
schedule and then each and every subsequent transfer only adds a new
QH into that schedule, waits until the QH is completed and does NOT
disable the async schedule. The async schedule is stopped only by
shutting down the controller, which must happen before moving out
of U-Boot, otherwise the controller will corrupt memory.

Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Cc: Bin Meng <bmeng.cn@gmail.com>
Cc: Simon Glass <sjg@chromium.org>
2019-10-31 12:13:40 +01:00