This reverts commit 02f3029f18.
This patch add 3 times retry to CMD8 because the Marvell mmc controller
doesn't obey the power ramp up process in the SD specification 6.4.1.
(Please refer to figure 6.1 and 6.2 in the specification.)
The CMD0 should be send after power ramp up has been finished.
However, the Marvell mmc contorller must do power ramp up after the
first CMD0 command has been send.
This patch also affect existing platforms like Nokia N900 and other
platforms.
Signed-off-by: Macpaul Lin <macpaul@andestech.com>
Acked-by: Lei Wen <leiwen@marvell.com>
Acked-by: Stephen Warren <swarren@nvidia.com>
Tested-by: Stephen Warren <swarren@nvidia.com>
As per JEDEC document JESD84-A441 (page 105) response for CMD7
(MMC_CMD_SELECT_CARD) response should be R1 instead of R1b. In uboot we
never take MMC to disconnected state and on powerup its always ideal
state which later goes to stand-by state.
from document footnote:
R1 while selecting from Stand-By State to Transfer State; R1b while
selecting from Disconnected State to Programming State.
Signed-off-by: Ajay Bhargav <ajay.bhargav@einfochips.com>
For we don't know mmc bus width from reading registers, the only way
to check is to test.
Current compare offset is:
EXT_CSD_PARTITIONING_SUPPORT
EXT_CSD_ERASE_GROUP_DEF
EXT_CSD_REV
EXT_CSD_HC_ERASE_GRP_SIZE
EXT_CSD_SEC_CNT
Signed-off-by: Lei Wen <leiwen@marvell.com>
Previous magic number is hard to parse its meaning, change it to
respective macro definition
Signed-off-by: Lei Wen <leiwen@marvell.com>
Acked-by: WOlfgang Denk <wd@denx.de>
For some controller it has dynamic clock gating, and only toggle out clk
when the first cmd0 send out, while some card strictly obey the 74
clocks rule, the interval may not be sufficient between the cmd0 and
this cmd8, retry to fulfil the clock requirement.
Signed-off-by: Lei Wen <leiwen@marvell.com>
Tested-by: Marek Vasut <marek.vasut@gmail.com>
Currently the mmc_change_freq and mmc_startup functions allocates
buffers on the stack that are passed down to the MMC device driver.
These buffers could be unaligned to the L1 dcache line size. This
causes problems when using DMA and with caches enabled.
This patch correctly cache alignes the buffers used for reading the
ext_csd data from an MMC device.
Signed-off-by: Anton Staaf <robotboy@chromium.org>
Cc: Lukasz Majewski <l.majewski@samsung.com>
Cc: Mike Frysinger <vapier@gentoo.org>
Cc: Albert ARIBAUD <albert.u.boot@aribaud.net>
Currently the sd_change_freq function allocates two buffers on the
stack that it passes down to the MMC device driver. These buffers
could be unaligned to the L1 dcache line size. This causes problems
when using DMA and with caches enabled.
This patch correctly cache alignes the buffers used for reading the
scr register and switch status values from an MMC device.
Change-Id: Ifa8414f572ef907681bd2d5ff3950285a215357d
Signed-off-by: Anton Staaf <robotboy@chromium.org>
Cc: Lukasz Majewski <l.majewski@samsung.com>
Cc: Mike Frysinger <vapier@gentoo.org>
Cc: Albert ARIBAUD <albert.u.boot@aribaud.net>
Acked-by: Mike Frysinger <vapier@gentoo.org>
The mmc_send_status() function sets cmd.arg = 0. That's incorrect, so fix it.
Signed-off-by: Marek Vasut <marek.vasut@gmail.com>
Acked-by: Andy Fleming <afleming@freescale.com>
Tested-by: Lei Wen <adrian.wenl@gmail.com>
If we don't want to build support for any partition types we can now
add #undef CONFIG_PARTITIONS in a board config file to keep this from
being compiled in. Otherwise boards assume this is compiled in by
default
Signed-off-by: Matthew McClintock <msm@freescale.com>
This patch provides handling of the two way handshake when SEND_OP_COND
(CMD1) is send to mmc card. It is necessary to inform eMMC card if the
host can work with high capacity cards (Jedec JESD84-A441, point 7.4.3).
The extra flag MMC_MODE_HC (high capacity) is added to indicate if the
host is capable of handling the high capacity eMMC cards.
Since this change is added to the generic mmc framework, then it requires
other boards to indicate if their mmc controllers can handle high capacity
cards. As it is now - the old behaviour of the framework is preserved.
Signed-off-by: Lukasz Majewski <l.majewski@samsung.com>
Signed-off-by: Andy Fleming <afleming@freescale.com>
Fix the problem that if we use the chip of MMC version 4 and
the capacity is smaller than 2GB or equal, the mmc->capacity is
invalid. According to the JEDEC Standard, the value of ext_csd's
capacity is valid if the value is more than 2GB.
Signed-off-by: Yoshihiro Shimoda <yoshihiro.shimoda.uh@renesas.com>
Acked-by: Andy Fleming <afleming@freescale.com>
Erase is a very basic function since the begin of sd specification is
announced. Although we could write a bulk of full 0xff memory to the
range to take place of erase, it is more convenient and safe to
implement the erase function itself.
Signed-off-by: Lei Wen <leiwen@marvell.com>
Signed-off-by: Andy Fleming <afleming@freescale.com>
Acked-by: Mike Frysinger <vapier@gentoo.org>
For emmc, it may have up to 7 partitions: two boot partitions, one
user partition, one RPMB partition and four general purpose partitions.
(Refer to JESD84-A44.pdf/page 154)
As bootloader may need to read out or reflashing images on those
different partitions, it is better to enable the partition switch with
console command support.
Also for partition would be restore to user partition(part 0) when CMD0
is used, so change mmc_init routine to perform normal initialization
only once for each slot, unless use the rescan command to force init
again.
Signed-off-by: Lei Wen <leiwen@marvell.com>
Acked-by: Andy Fleming <afleming@freescale.com>
Make existing field b_max field in struct mmc unconditional
and use it instead of CONFIG_SYS_MMC_MAX_BLK_COUNT in mmc_bread
and mmc_bwrite.
Initialize b_max to CONFIG_SYS_MMC_MAX_BLK_COUNT in mmc_register
if it has not been initialized by the hw driver.
Initialize b_max to 0 in all callers to mmc_register.
Signed-off-by: John Rigby <john.rigby@linaro.org>
Signed-off-by: Andy Fleming <afleming@freescale.com>
Hi Terry,
> So I guess:
> mmc_init calls mmc_send_op_cond that set high_capacity,
> than it calls mmc_startup, that, with MMC_CMD_SEND_CSD command, set
> the capacity, using values in CSD register.
> So I guess that mmc_change_freq should not recalculate high_capacity.
>
> It seems better, isn't it?
>
> Regards,
> Raffaele
>
Finally I think that it is enough to apply the following patch in order
to fix the issue.
Regards,
Raffaele
Signed-off-by: Andy Fleming <afleming@freescale.com>
Defining CONFIG_MMC_TRACE in the include board file it is possible to activate
a tracing support.
This code helps in case of eMMC hw failure or to investigate possible eMMC
initialization issues.
Signed-off-by: Raffaele Recalcati <raffaele.recalcati@bticino.it>
Signed-off-by: Andy Fleming <afleming@freescale.com>
The first SEND_OP_COND (CMD1) command added is used to ask card capabilities.
After it an AND operation is done between card capabilities and host
capabilities (at the moment only for the voltage field).
Finally the correct value is sent to the MMC, waiting that the card
exits from busy state.
Signed-off-by: Raffaele Recalcati <raffaele.recalcati@bticino.it>
Signed-off-by: Andy Fleming <afleming@freescale.com>
It is recommended to check card status after these kind of commands.
This is done using CMD13 (SEND_STATUS) JEDEC command until
the card is ready.
In case of error the card status field is displayed.
Signed-off-by: Raffaele Recalcati <raffaele.recalcati@bticino.it>
Signed-off-by: Andy Fleming <afleming@freescale.com>
This patch supports mmc/sd card with spi interface. It is based on
the generic mmc framework. It works with SDHC and supports multi
blocks read/write.
The crc checksum on data packet is enabled with the def,
There is a subcomamnd "mmc_spi" to setup spi bus and cs at run time.
Signed-off-by: Thomas Chou <thomas@wytron.com.tw>
Signed-off-by: Andy Fleming <afleming@freescale.com>
These local vars need not be writable nor exported.
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Andy Fleming <afleming@freescale.com>
The maximum blocks value was hardcoded to 65535 due to a 16 bit
register length. The value can change for different platforms.
This patch makes the default the current value of 65535, but it
is configurable for other platforms.
Signed-off-by: Matt Waddel <matt.waddel@linaro.org>
Signed-off-by: Andy Fleming <afleming@freescale.com>
This patch adds multi-block read support for the generic MMC
driver. Large reads are broken into chunks of 65535 blocks to
ensure that the code works with controllers having a 16 bit block counter.
This patch results in a significant performance improvement.
Time to read a 45 MB file went from 36 seconds to 9 seconds on Overo
Signed-off-by: Steve Sakoman <steve.sakoman@linaro.org>
Tested-by: Steve Sakoman <steve.sakoman@linaro.org>
Signed-off-by: Sandeep Paulraj <s-paulraj@ti.com>
The current mmc write implementation is type ulong, but returns int values.
Some of the printf's are terminated with /n/r, one has none.
This patch fixes these issues and also removes some unnecessary local
variables.
Signed-off-by: Steve Sakoman <steve.sakoman@linaro.org>
Signed-off-by: Sandeep Paulraj <s-paulraj@ti.com>
Constraint the mmc framework to only send no more than 65535
blocks in one go during the multi-write command. This constraint
comes due to the limitation of 16bit width block counter register
at some hardware.
Signed-off-by: Lei Wen <leiwen@marvell.com>
Cc: Reinhard Meyer <u-boot@emk-elektronik.de>
The current mmc driver returns erroneous capacity information for
eMMC. The capacity of eMMC devices is available only in the ext-CSD
register. This patch add code to read the ext-CDSD register and
correctly calculate eMMC capacity.
Signed-off-by: Sukumar Ghorai <s-ghorai@ti.com>
Acked-by: Steve Sakoman <steve@sakoman.com>
Commit d2bf29e3 caused a number of compiler warnings:
mmc.c: In function 'mmc_bwrite':
mmc.c:97: warning: format '%x' expects type 'unsigned int', but argument 2 has type 'long unsigned int'
mmc.c:97: warning: format '%x' expects type 'unsigned int', but argument 3 has type 'lbaint_t'
mmc.c: In function 'mmc_bread':
mmc.c:229: warning: format '%x' expects type 'unsigned int', but argument 2 has type 'long unsigned int'
mmc.c:229: warning: format '%x' expects type 'unsigned int', but argument 3 has type 'lbaint_t'
Fix these.
Signed-off-by: Wolfgang Denk <wd@denx.de>
Cc: Lei Wen <leiwen@marvell.com>
There is be a path through mmc_read in drivers/mmc/mmc.c where
malloc'd memory is not freed before exiting mmc_read: it occurs if
mmc_set_blocklen() returns a non-zero value.
Reported-by: Quentin Armitage <Quentin@Armitage.org.uk>
Signed-off-by: Wolfgang Denk <wd@denx.de>
Most controllers can check if there is a card in the slot.
However, they require pins that could be not available because
required by other functions and the detection of a card must
be performed in another way. This patch adds a weak function
that a board can implement to add its internal custom way
to check the presence of a MMC/SD card.
Signed-off-by: Stefano Babic <sbabic@denx.de>
Most cards do not answer if some reserved bits
in the ocr are set. However, some controllers
can set bit 7 (reserved for low voltages), but
how to manage low voltages SD card is not yet
specified.
Signed-off-by: Stefano Babic <sbabic@denx.de>
We need to switch back to 1-bit before initialization or SD 2.0 cards
will fail to send SCR if we've switched to 4-bit already.
Signed-off-by: Ilya Yanok <yanok@emcraft.com>
SCR & switch data are read from card as big-endian words and should be
converted to CPU byte order.
Signed-off-by: Yauhen Kharuzhy <jekhor@gmail.com>
Signed-off-by: Andy Fleming <afleming@freescale.com>
Cards which are not compatible with SD 2.0 standard, may return response
for CMD8 command, but it will be invalid in terms of SD 2.0. We should
accept this case as admissible, just like Linux does.
Signed-off-by: Yauhen Kharuzhy <jekhor@gmail.com>
Signed-off-by: Andy Fleming <afleming@freescale.com>
The mmc code defines the response as an array of chars. However, it
access the response bytes both as (i) an array of four uints (with
casts) and (ii) as individual chars. The former case is used more
often, including by the driver when it assigns the response.
The char-wise accesses are broken on little endian systems because they
assume that the bytes in the uints are in big endian byte order.
This patch fixes this by changing the response to be an array of four
uints and replacing the char-wise accesses with equivalent uint-wise
accesses.
Signed-off-by: Rabin Vincent <rabin@rab.in>
The generic MMC core uses direct long long divisions, which do not build
with ARM EABI toolchains. Use lldiv() instead, which works everywhere.
Signed-off-by: Rabin Vincent <rabin@rab.in>
find_mmc_device returns NULL if an invalid device number is specified.
Check for this to avoid dereferencing NULL pointers.
Signed-off-by: Rabin Vincent <rabin@rab.in>
__attribute__ follows gcc's documented syntax and is generally more
common than __attribute. This change is only asthetic and should not
affect functionality.
Signed-off-by: Peter Tyser <ptyser@xes-inc.com>
Here's a new framework (based roughly off the linux one) for managing
MMC controllers. It handles all of the standard SD/MMC transactions,
leaving the host drivers to implement only what is necessary to
deal with their specific hardware.
This also hooks the infrastructure into the PowerPC board code
(similar to how the ethernet infrastructure now hooks in)
Some of this code was contributed by Dave Liu <daveliu@freescale.com>
Signed-off-by: Andy Fleming <afleming@freescale.com>
