In current code, we add 1ms dealy after each tuning command for standard
tuning method. Adding this 1ms dealy is because USDHC default check the
CMD CRC and DATA line. If detect the CMD CRC, USDHC standard tuning
IC logic do not wait for the tuning data sending out by the card, trigger
the buffer read ready interrupt immediately, and step to next cycle. So
when next time the new tuning command send out by USDHC, card may still
not send out the tuning data of the upper command，then some eMMC cards
may stuck, can't response to any command, block the whole tuning procedure.
If do not check the CMD CRC for tuning, then do not has this issue. USDHC
will wait for the tuning data of each tuning command and check them. If the
tuning data pass the check, it also means the CMD line also okay for tuning.
So this patch disable the CMD CRC check for tuning, save some time for the
whole tuning procedure.
Signed-off-by: Haibo Chen <haibo.chen@nxp.com>
This patch try to avoids eviction of dirty lines during DMA
transfer. The code right now execute the following step:
- allocate the buffer
- start a dma operation using the non-coherent dma buffer
- invalidate cache lines associated with the buffer
- read the buffer
This can lead to reading back not valid information, because the cache
controller could evict dirty cache lines belonging to the buffer *after*
the DMA operation has started to fill the DRAM.
In order to avoid this, a new invalidation is required *before* starting
the DMA operation. The patch just adds an invalidation before submitting
the DMA command.
Example below shows the nvme disk scan result without the following
patch
=> nvme scan
nvme_get_info_from_identify: nn = 544502629, vwc = 100,
sn = dev_0T, mn = `�\�, fr = t_part, mdts = 105
So, invalidating the cache before submitting the admin command,
fix the cpu read.
Cc: André Przywara <andre.przywara@arm.com>
Reported-by: Suniel Mahesh <sunil@amarulasolutions.com>
Signed-off-by: Michael Trimarchi <michael@amarulasolutions.com>
Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Tested-by: Suniel Mahesh <sunil@amarulasolutions.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Since the i.MX8 GPIO banks are indexed from 0 not 1 on other i.MX
platforms, so we have to adjust the index accordingly.
Signed-off-by: Adrian Alonso <adrian.alonso@nxp.com>
Signed-off-by: Ye Li <ye.li@nxp.com>
As we move towards driver model, it is required to let the FEC driver
know how to properly deal with an Ethernet PHY subnode in the device tree.
For example:
&fec {
pinctrl-names = "default";
pinctrl-0 = <&pinctrl_microsom_enet_ar8035>;
phy-handle = <&phy>;
phy-mode = "rgmii-id";
phy-reset-duration = <2>;
phy-reset-gpios = <&gpio4 15 GPIO_ACTIVE_LOW>;
status = "okay";
mdio {
#address-cells = <1>;
#size-cells = <0>;
phy: ethernet-phy@0 {
reg = <0>;
qca,clk-out-frequency = <125000000>;
};
};
};
Currently the PHY node pointer is incorrectly associated with the
Ethernel controller instead of the PHY node itself.
This causes the PHY properties, such as "qca,clk-out-frequency" in
the example above to not get parsed.
Fix this problem by populating the phy_of_node node.
Suggested-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: Fabio Estevam <festevam@gmail.com>
Tested-by: Tom Rini <trini@konsulko.com>
The clock ouput frequency is calculated incorrectly for AR8035 due to
wrong masking of priv->clk_25m_reg and priv->clk_25m_mask.
This same issue has been already fixed in the kernel by:
commit b1f4c209d84057b6d40b939b6e4404854271d797
Author: Oleksij Rempel <o.rempel@pengutronix.de>
Date: Wed Apr 1 11:57:32 2020 +0200
net: phy: at803x: fix clock sink configuration on ATH8030 and ATH8035
The masks in priv->clk_25m_reg and priv->clk_25m_mask are one-bits-set
for the values that comprise the fields, not zero-bits-set.
This patch fixes the clock frequency configuration for ATH8030 and
ATH8035 Atheros PHYs by removing the erroneous "~".
To reproduce this bug, configure the PHY with the device tree binding
"qca,clk-out-frequency" and remove the machine specific PHY fixups.
Fixes: 2f664823a47021 ("net: phy: at803x: add device tree binding")
Signed-off-by: Oleksij Rempel <o.rempel@pengutronix.de>
Reported-by: Russell King <rmk+kernel@armlinux.org.uk>
Reviewed-by: Russell King <rmk+kernel@armlinux.org.uk>
Tested-by: Russell King <rmk+kernel@armlinux.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
Apply the same fix in the U-Boot driver.
Tested on a i.MX6 Hummingboard.
Signed-off-by: Fabio Estevam <festevam@gmail.com>
Reviewed-by: Michael Walle <michael@walle.cc>
Tested-by: Tom Rini <trini@konsulko.com>
- migrate remaining imx5/imx6qd boards to DM_VIDEO
- use 'vidconsole' stdout in nitrogen6x upgrade script
- add option for selection of the vidconsole commands
lcdputs and setcurs
- allow building ipuv3 driver with disabled CONFIG_PANEL
- make backlight and panel drivers optional
- remove useless code from ipuv3 driver
- extend existing DM_VIDEO work-around for console name
to support 'vga'
- remove non-DM code in ipuv3
When CONFIG_OF_LIVE is enabled, dev_of_offset() cannot be used and
if used returns an invalid offset. This causes the call to
fdt_stringlist_get() in the psci_probe() to fail to read the 'method'
property from the PSCI node for the device and hence prevents PSCI
from working. Fix this by using the ofnode_read_string() API instead
of the fdt_stringlist_get() because this will handle reading the
property both when CONFIG_OF_LIVE is enabled or disabled.
Due to the above problem and since commit 81ea00838c ("efi_loader:
PSCI reset and shutdown") was added, the EFI system reset has been
broken for Tegra210 and Tegra196 platforms. This also fixes the EFI
system reset for these Tegra platforms.
Signed-off-by: Jon Hunter <jonathanh@nvidia.com>
Tested-by: Peter Robinson <pbrobinson@gmail.com>
Pull the common parts of functions out so they can be reused by both
DM and non-DM code paths. The recv() function had to be reworked to
fit into this scheme and this means it now only receives one packet
at a time instead of spinning in an endless loop.
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Cc: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
Cc: Joe Hershberger <joe.hershberger@ni.com>
Instead of using the non-DM-only name and enetaddr in struct eth_device,
add pointers into the private data which can either point to that non-DM
name or a DM one later on.
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Cc: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
Cc: Joe Hershberger <joe.hershberger@ni.com>
Get rid of the global point to private data, and rather pass it
thought dev->priv. Also remove the unnecessary check for lp being
non-NULL, since it is always NULL at this point.
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Cc: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
Cc: Joe Hershberger <joe.hershberger@ni.com>
Instead of using eth_device priv for this PCI devbusfn, free it
so it could be used for driver private data, and wrap devbusfn
into those driver private data.
Note that using the name dev for the variable is a trick left for
later, when DM support is in place, so dm_pci_virt_to_mem() can be
used with minimal ifdeffery.
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Cc: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
Cc: Joe Hershberger <joe.hershberger@ni.com>
The private data allocation does not have to be done every time the
NIC is initialized at run time, move the allocation to initialize
function, which means it will be done only once when the driver
starts.
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Cc: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
Cc: Joe Hershberger <joe.hershberger@ni.com>
The current code is horribly complex. Both the RX and TX buffer
descriptors are 16 bytes in size, the init block is 32 bytes in
size, so simplify the code such that the entire private data of
the driver are allocated cache aligned and the RX and TX buffer
descriptors are part of the private data.
This removes multiple malloc calls and cache flushes.
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Cc: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
Cc: Joe Hershberger <joe.hershberger@ni.com>
Split the common code from the non-DM code, so it can be reused by
the DM code later. As always, the recv() function had to be split
into the actual receiving part and free_pkt part to fit with the
DM.
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
The RX/TX DMA descriptor rings are per-device-instance private data,
so move them into the private data.
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Instead of doing ethernet device lookup by name every time there
is an MDIO access, pass the driver private data via mdiobus priv
to the MDIO bus accessors.
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
This patch replaces the various uses of struct eth_device for accessing
device private data with struct eepro100_priv, which is compatible both
with DM and non-DM operation.
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
This is a trick in preparation for adding DM support. By passing in
the PCI BDF into the bus_to_phys()/phys_to_bus() macros and calling
that dev, we can substitute dev with udevice when DM support lands
and do minor adjustment to the macros to support both DM and non-DM
operation. No functional change.
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Drop the inline keyword from the static functions, the compiler has a
much better overview and can decide how to inline those functions much
better.
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Those macros depended on specific variable names to be declared
at their usage sites, fix this by adding an argument to those
macros and also protect the argument with braces.
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Pull the MII registration code into a separate function. Moreover,
properly free memory in case any of the registration or allocation
functions fail, so this fixes an existing memleak.
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Use this macro to fully fill the PCI device ID table. This is mandatory
for the DM PCI support, which checks all the fields.
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Move the functions around in the driver to prepare it for DM conversion.
Drop forward declarations which are not necessary anymore. No functional
change.
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Remove all the remaining use of the 'volatile' keyword, as this is
no longer required. All the accesses which might have needed this
use of 'volatile' have been repaired properly.
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Add cache invalidation and flushes wherever the DMA descriptors are
written or read, otherwise this driver cannot work reliably on any
systems where caches are enabled.
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
This code is replicated in the driver thrice almost verbatim, factor
it out into a separate function and clean it up. No functional change.
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
The current eepro100 driver accesses its memory mapped registers directly
instead of using the standard I/O accessors. This can cause problems on
some systems as the accesses can get out of order. So convert the direct
volatile dereferences to use the normal in/out macros.
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
This is automated cleanup via checkpatch, no functional change.
./scripts/checkpatch.pl --show-types -f drivers/net/eepro100.c
./scripts/checkpatch.pl --types INDENTED_LABEL -f --fix --fix-inplace drivers/net/eepro100.c
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
This is automated cleanup via checkpatch, no functional change.
./scripts/checkpatch.pl --show-types -f drivers/net/eepro100.c
./scripts/checkpatch.pl -f --fix --fix-inplace drivers/net/eepro100.c
This fixes all the remaining errors except a couple of comments which
are longer than 80 characters, all the volatile misuse and all the
camelcase, that needs a separate patch.
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
This is automated cleanup via checkpatch, no functional change.
./scripts/checkpatch.pl --show-types -f drivers/net/eepro100.c
./scripts/checkpatch.pl --types INDENTED_LABEL -f --fix --fix-inplace drivers/net/eepro100.c
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
This is automated cleanup via checkpatch, no functional change.
./scripts/checkpatch.pl --show-types -f drivers/net/eepro100.c
./scripts/checkpatch.pl --types POINTER_LOCATION -f --fix --fix-inplace drivers/net/eepro100.c
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
This is automated cleanup via checkpatch, no functional change.
./scripts/checkpatch.pl --show-types -f drivers/net/eepro100.c
./scripts/checkpatch.pl --types PARENTHESIS_ALIGNMENT -f --fix --fix-inplace drivers/net/eepro100.c
Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
