The current dm9000x driver accesses its memory mapped registers directly
instead of using the standard I/O accessors. This can cause problems on
Blackfin 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: Mike Frysinger <vapier@gentoo.org>
Tested-by: Hoan Hoang <hnhoan@i-syst.com>
Signed-off-by: Ben Warren <biggerbadderben@gmail.com>
The current OUTW function is always defined as a 16bit function, but this
doesn't work correctly when using the 32bit access mode. So define it as
a 32bit function when in 32bit mode so things work correctly on Blackfin
32bit LE systems.
Signed-off-by: Hoan Hoang <hnhoan@i-syst.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Ben Warren <biggerbadderben@gmail.com>
Some places in the current code equate the Marvell 88E1111 PHY as the family
when in reality it's a subpart of the Alaska family. So once we generalize
that, add support for the 88E1118 PHY.
Signed-off-by: Hoan Hoang <hnhoan@i-syst.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Ben Warren <biggerbadderben@gmail.com>
The Ethernet initialization functions are supposed to return the number of
devices initialized, so fix tsec_eth_init() so that they returns the number of
TSECs initialized, instead of just zero. This is safe because the return value
is currently ignored by all callers, but now they don't have to ignore it.
In general, if an function initializes only one device, then it should return
a negative number if there's an error. If it initializes more than one device,
then it should never return a negative number.
Signed-off-by: Timur Tabi <timur@freescale.com>
Signed-off-by: Ben Warren <biggerbadderben@gmail.com>
Device names should not contain non-printable characters like newlines.
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Ben Warren <biggerbadderben@gmail.com>
No compiled code change here, just drop the local PHY defines in favor of
the common standard ones.
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Ben Warren <biggerbadderben@gmail.com>
Rather than hardcode specific phy addresses, search the possible phy
address space to find the first available phy. Also respect the normal
CONFIG_PHY_ADDR option for board porters to pick a specific address.
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Ben Warren <biggerbadderben@gmail.com>
DRAM window mapping uses kirkwood-provided functions instead
of global gd as do other drivers--fix this.
Also, fix a typo in a comment
Signed-off-by: Albert Aribaud <albert.aribaud@free.fr>
Acked-by: Prafulla Wadaskar <prafulla@marvell.com>
Signed-off-by: Ben Warren <biggerbadderben@gmail.com>
Designware network driver support added.
This is a Synopsys ethernet controller
Signed-off-by: Vipin Kumar <vipin.kumar@st.com>
Signed-off-by: Ben Warren <biggerbadderben@gmail.com>
Insert isb() sequence points to ensure DMA descriptors
are filled in and set up before actual DMA occurs.
Signed-off-by: Albert Aribaud <albert.aribaud@free.fr>
Acked-by: Prafulla Wadaskar <prafulla@marvell.com>
Signed-off-by: Ben Warren <biggerbadderben@gmail.com>
By defining CONFIG_M88E1111_DISABLE_FIBER boards can configure the
M88E1111 PYH to disable fiber. This is needed for an upcoming PPC460GT
based board, which has fiber/copper auto-selection enabled by default.
This doesn't seem to work. So we disable fiber in the PHY register.
Signed-off-by: Stefan Roese <sr@denx.de>
Signed-off-by: Ben Warren <biggerbadderben@gmail.com>
The DM9000 in/out helper functions were casting the register address when
it was accessing things directly (pre commit a45dde2293). But
when it was changed to using the in/out helpers, those casts were dropped
because those functions don't take pointers. Even more recently, those
functions were then changed to use the read/write helpers, but the casts
were not re-added. This is necessary because the read/write helpers do
take pointers. Otherwise we get a lot of warnings like:
dm9000x.c: In function 'dm9000_inblk_8bit':
dm9000x.c:172: warning: passing argument 1 of 'readb'
makes pointer from integer without a cast
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Tested-by: Thomas Weber <weber@corscience.de>
Signed-off-by: Ben Warren <biggerbadderben@gmail.com>
The hush shell dynamically allocates (and re-allocates) memory for the
argument strings in the "char *argv[]" argument vector passed to
commands. Any code that modifies these pointers will cause serious
corruption of the malloc data structures and crash U-Boot, so make
sure the compiler can check that no such modifications are being done
by changing the code into "char * const argv[]".
This modification is the result of debugging a strange crash caused
after adding a new command, which used the following argument
processing code which has been working perfectly fine in all Unix
systems since version 6 - but not so in U-Boot:
int main (int argc, char **argv)
{
while (--argc > 0 && **++argv == '-') {
/* ====> */ while (*++*argv) {
switch (**argv) {
case 'd':
debug++;
break;
...
default:
usage ();
}
}
}
...
}
The line marked "====>" will corrupt the malloc data structures and
usually cause U-Boot to crash when the next command gets executed by
the shell. With the modification, the compiler will prevent this with
an
error: increment of read-only location '*argv'
N.B.: The code above can be trivially rewritten like this:
while (--argc > 0 && **++argv == '-') {
char *arg = *argv;
while (*++arg) {
switch (*arg) {
...
Signed-off-by: Wolfgang Denk <wd@denx.de>
Acked-by: Mike Frysinger <vapier@gentoo.org>
At the moment, the default SPI flash subsystem is quite terse. Errors and
successes both result in a generic message. So move the useful errors and
useful successes to printf output by default.
While we're here, also convert the messages to use print_size().
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Some old STMicro parts do not support JEDEC ID (0x9f). This patch
uses RES (0xab) to get Electronic ID and translates it to JEDEC ID.
Signed-off-by: Thomas Chou <thomas@wytron.com.tw>
Acked-by: Mike Frysinger <vapier@gentoo.org>
This avoids a build warning that you see if anyone in the
header chain has included io.h (which is coming shortly). The previous
code redefined readl/writel; this patch renames it to be specific to
ohci. The defines are also moved from ohci-hcd.c to ohci.h.
Signed-off-by: Becky Bruce <beckyb@kernel.crashing.org>
MUSB PHY on OMAP3EVM Rev >= E uses external Vbus supply to support
500mA of power.We need to program MUSB PHY to use external Vbus
for this purpose.
Adding 'extvbus' member in musb_config structure which should be set
by all the boards where MUSB interface is using external Vbus supply.
Also added ULPI bus control register read/write abstraction for
Blackfin processor as it doesn't have ULPI registers.
CC: Remy Bohmer <linux@bohmer.net>
Signed-off-by: Ajay Kumar Gupta <ajay.gupta@ti.com>
Acked-by: Mike Frysinger <vapier@gentoo.org>
Changed musb_config initialization for omap3.c, davinci.c
and da8xx.c using name of structure fields. This would cause
the uninitialized field to be null by default and thus would
help in avoiding to init some flags required to be set only
for a few selected platforms.
CC: Remy Bohmer <linux@bohmer.net>
Signed-off-by: Ajay Kumar Gupta <ajay.gupta@ti.com>
Add USB OHCI support for at91sam9g45ekes/at91sam9m10g45ek boards.
Note that according to errata from Atmel, OHCI is not operational
on the first revision of at91sam9g45 chip. So this patch enables
OHCI support for later revisions.
Signed-off-by: Sergey Matyukevich <geomatsi@gmail.com>
ehci_submit_async() doesn't really zero out the QH transfer overlay (as the EHCI
specification suggests) which leads to the controller seeing the "token" field
as the previous call has left it, i.e.:
- if a timeout occured on the previous call (Active bit left as 1), controller
incorrectly tries to complete a previous transaction on a newly programmed
endpoint;
- if a halt occured on the previous call (Halted bit set to 1), controller just
ignores the newly programmed TD(s) and the function then keeps returning error
ad infinitum.
This turned out to be caused by the wrong orger of the arguments to the memset()
call in ehci_alloc(), so the allocated TDs weren't cleared either.
While at it, stop needlessly initializing the alternate next TD pointer in the
QH transfer overlay...
Signed-off-by: Sergei Shtylyov <sshtylyov@ru.mvista.com>
Acked-by: Remy Bohmer <linux@bohmer.net>
On P2020RDB eTSEC2 is connected to Vitesse VSC8221 PHY via SGMII.
Current TBI PHY settings for SGMII mode cause link problems on
this platform, link never comes up.
Fix this by making TBI PHY settings configurable and add a working
configuration for P2020RDB.
Signed-off-by: Felix Radensky <felix@embedded-sol.com>
Acked-by: Andy Fleming <afleming@freescale.com>
Acked-by: Peter Tyser <ptyser@xes-inc.com>
Tested-by: Peter Tyser <ptyser@xes-inc.com>
The AmigaOneG3SE board has been orphaned or a very long time, and
broken for more than 12 releases resp. more than 3 years. As nobody
seems to be interested any more in this stuff we may as well ged rid
of it, especially as it clutters many areas of the code so it is a
continuous pain for all kinds of ongoing work.
Signed-off-by: Wolfgang Denk <wd@denx.de>
UBI: initialise update marker
The in kernel copy of a volume's update marker is not initialised from the
volume table. This means that volumes where an update was unfinnished will
not be treated as "forbidden to use". This is basically that the update
functionality was broken.
Signed-off-by: Peter Horton <zero@colonel-panic.org>
Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>
Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
Acked-by: Stefan Roese <sr@denx.de>
The following restructuring and optimisations increase the SPI
read performance from 1.3MiB/s (on da850) to 2.87MiB/s (on da830):
Remove continual revaluation of driver state from the core of the
copy loop. State can not change during the copy loop, so it is
possible to move these evaluations to before the copy loop.
Cost is more code space as loop variants are required for each set
of possible configurations. The loops are simpler however, so the
extra is only 128bytes on da830 with CONFIG_SPI_HALF_DUPLEX
defined.
Unrolling the first copy loop iteration allows the TX buffer to be
pre-loaded reducing SPI clock starvation.
Unrolling the last copy loop iteration removes testing for the
final loop iteration every time round the loop.
Using the RX buffer empty flag as a transfer throttle allows the
assumption that it is always safe to write to the TX buffer, so
polling of TX buffer full flag can be removed.
Signed-off-by: Nick Thompson <nick.thompson@ge.com>
Signed-off-by: Sandeep Paulraj <s-paulraj@ti.com>
When set to PULL_NONE, gpio_set_pull function is returned without write the register.
This patch fixed it.
Signed-off-by: Minkyu Kang <mk7.kang@samsung.com>
For boards using sm501/sm502 on PCI bus some driver
functions normaly defined in the board code are not
needed and empty. Provide weak default functions for
them and do not enforce board code to define empty
functions.
Signed-off-by: Anatolij Gustschin <agust@denx.de>
Do not enforce drivers to provide empty video_set_lut()
if they do not implement indexed color (8 bpp) frame
buffer support. Add default function to the cfb_console
driver and remove empty video_set_lut() functions.
Signed-off-by: Anatolij Gustschin <agust@denx.de>
This patch enables PXAMCI support on PXA3xx CPUs. This patch only enables MMC1
though, MMC2 and PXA31x MMC3 will need further patch to be operational.
Signed-off-by: Marek Vasut <marek.vasut@gmail.com>
In case the delays were set to 10000, the MMC card on PXA27X boards (and PXA3xx
boards) didn't initialize on first try. Increasing the delays and leaving just
those for PXA25x and 26x (that is 200000) fixes this problem.
Signed-off-by: Marek Vasut <marek.vasut@gmail.com>
I have updated this patch based on the comments [1] by Wolfgang Denk and
removed unused variables.
[1][http://lists.denx.de/pipermail/u-boot/2010-May/071728.html]
Reduce the number of reads per byte transferred on the BUF register from 2 to 1 and
take advantage of the TX buffer in the SPI module. On LogicPD OMAP-L138 EVM,
SPI read throughput goes up from ~0.8Mbyte/s to ~1.3Mbyte/s. Tested with a 2Mbyte image file.
Remove unused variables in the spi_xfer() function.
Signed-off-by: Delio Brignoli <dbrignoli@audioscience.com>
Tested-by: Ben Gardiner <bengardiner@nanometrics.ca>
Signed-off-by: Sandeep Paulraj <s-paulraj@ti.com>
This patch adds the driver of altera spi controller, which is
used as epcs/spi flash controller. It also works with mmc_spi
driver.
This driver support more than one spi bus, with base list declared
#define CONFIG_SYS_ALTERA_SPI_LIST { BASE_0,BASE_1,... }
Signed-off-by: Thomas Chou <thomas@wytron.com.tw>
Tested-by: Ian Abbott <abbotti@mev.co.uk>
Signed-off-by: Scott McNutt <smcnutt@psyent.com>
This patch adds a status led driver followed the GPIO access
conventions of Linux. The led mask is used to specify the gpio pin.
Signed-off-by: Thomas Chou <thomas@wytron.com.tw>
Tested-by: Ian Abbott <abbotti@mev.co.uk>
Signed-off-by: Scott McNutt <smcnutt@psyent.com>
The current Blackfin nand write function fills up the write buffer but
returns before it has had a chance to drain. On faster systems, this
isn't a problem as the operation finishes before the ECC registers are
read, but on slower systems the ECC may be incomplete when the core tries
to read it.
So wait for the buffer to drain once we're done writing to it.
Signed-off-by: Andrew Caldwell <Andrew.Caldwell@analog.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
