GitLab

Fix following error with gcc 4.9:

mmc_cmds.c:1918:9: error: missing braces around initializer [-Werror=missing-braces]
  struct rpmb_frame frame_status = {0};

Fixes:
 - http://autobuild.buildroot.org/results/bf3b6f9f6ef39b99842b3c92495b7bf359c68158

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Link: https://lore.kernel.org/r/20190928185107.21125-1-fontaine.fabrice@gmail.com

Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>

This patch adds support to enable auto(AUTO_EN) and
manual(MANUAL_EN) in BKOPS_EN register. Auto bkops
can only be used on eMMC 5.0 or newer.
Signed-off-by: Lars Pedersen <lapeddk@gmail.com>
Reviewed-by: Avri Altman <avri.altman@wdc.com>
Link: https://lore.kernel.org/r/20191111110051.16490-1-lapeddk@gmail.com

Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>

This patch updates 'mmc writeprotect boot set' with a few more optional
parameters, so that it can be used to enable permanent write-protection
and so that the two boot partitions can be protected independently. It
also splits protection information output by 'mmc writeprotect boot get'
by partition.

(Note: eMMC boot partitions are named "Area 1" and "Area 2" by the eMMC
spec, but mmcblk0boot0 and mmcblk0boot1 by Linux. To avoid confusion
between the two numbering schemes, this patch uses 0 and 1 throughout,
even when defining EXT_CSD register bits.)
Signed-off-by: Julius Werner <jwerner@chromium.org>
Link: https://lore.kernel.org/r/20200316202221.107714-1-jwerner@chromium.org

Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>

JESD84-B51 7.4.30 CACHE_SIZE [252:249] states that "the size is
indicated as multiple of kilobits". This is also supported by Table 39,
"e.MMC internal sizes and related Units / Granularities" which lists
"32Kb (=4KB)" as the cache size granularity for 4KiB native devices.
Signed-off-by: Patrick Oppenlander <patrick.oppenlander@gmail.com>
Reviewed-by: Avri Altman <avri.altman@wdc.com>
Link: https://lore.kernel.org/r/20200401221453.267360-1-patrick.oppenlander@gmail.com

Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>

On our board we would occationally see what looked like a
kernel hang while doing RPMB operations, some additional
printk debugging revealed that in the trouble case data
such as this was being sent to the kernel:

    [  109.076981] idata[0]->ic->opcode = 0x19
    [  109.077027] idata[0]->ic->arg = 0x0
    [  109.080954] idata[0]->ic->write_flag = 0x1
    [  109.084691] idata[0]->ic->is_acmd = 0x0
    [  109.089007] idata[0]->ic->flags = 0xb5
    [  109.092949] idata[0]->ic->blocks = 0x1
    [  109.096865] idata[0]->ic->blksz = 0x200
    [  109.100791] idata[0]->ic->postsleep_min_us = 0x7473696e
    [  109.104745] idata[0]->ic->postsleep_max_us = 0x34383370
    [  109.110167] idata[0]->ic->data_timeout_ns = 0x7265632d
    [  109.115494] idata[0]->ic->data_ptr = 00000000e7ddd4cd
    [  109.120832] idata[0]->rpmb = 00000000ffab5640
    [  109.126077] idata[1]->ic->opcode = 0x12
    [  109.130622] idata[1]->ic->arg = 0x0
    [  109.135451] idata[1]->ic->write_flag = 0x0
    [  109.139153] idata[1]->ic->is_acmd = 0x0
    [  109.143437] idata[1]->ic->flags = 0xb5
    [  109.147553] idata[1]->ic->blocks = 0x1
    [  109.151420] idata[1]->ic->blksz = 0x200
    [  109.155251] idata[1]->ic->postsleep_min_us = 0x652d6873
    [  109.159166] idata[1]->ic->postsleep_max_us = 0x35353264
    [  109.164561] idata[1]->ic->data_timeout_ns = 0x632d3931
    [  109.169915] idata[1]->ic->data_ptr = 00000000bc641ed4
    [  109.175152] idata[1]->rpmb = 00000000ffab5640

Therefor, zero-init the MMC_IOC_MULTI_CMD structures to
avoid the obscene sleep values.
Signed-off-by: John Ernberg <john.ernberg@actia.se>
Link: https://lore.kernel.org/r/20201216093146.6218-1-john.ernberg@actia.se

Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>

As per specification, the host can use either CMD24 or CMD25 in
closed-ended or open-ended way.
CMD25 is better option as it can flash the firmware image in one go.
Signed-off-by: Shivamurthy Shastri <sshivamurthy@micron.com>
Reviewed-by: Avri Altman <avri.altman@wdc.com>
Reviewed-by: Bean Huo <beanhuo@micron.com>
Tested-by: Bean Huo <beanhuo@micron.com>
Link: https://lore.kernel.org/r/MN2PR08MB63973BF8F3C6959487BF5707B86A9@MN2PR08MB6397.namprd08.prod.outlook.com

Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>

Signed-off-by: Michael Heimpold <michael.heimpold@i2se.com>
Cc: Michael Heimpold <mhei@heimpold.de>
Reviewed-by: Avri Altman <avri.altman@wdc.com>
Signed-off-by: Chris Ball <chris@printf.net>

Appending multiple times to same string is slow since strcat() needs
to determine the end during each run. So manually maintain a pointer
to the end to speed-up things.
Signed-off-by: Michael Heimpold <michael.heimpold@i2se.com>
Cc: Michael Heimpold <mhei@heimpold.de>
Reviewed-by: Avri Altman <avri.altman@wdc.com>
Signed-off-by: Chris Ball <chris@printf.net>

When a hex-digit > 'a' or 'A' is read, we have to add an offset of 10
to access the valid symbol in our mapping table.
Signed-off-by: Michael Heimpold <michael.heimpold@i2se.com>
Cc: Michael Heimpold <mhei@heimpold.de>
Reviewed-by: Avri Altman <avri.altman@wdc.com>
Signed-off-by: Chris Ball <chris@printf.net>

In case that we leave due to malformed string,
free the allocated memory before returning.
Signed-off-by: Michael Heimpold <michael.heimpold@i2se.com>
Cc: Michael Heimpold <mhei@heimpold.de>
Reviewed-by: Avri Altman <avri.altman@wdc.com>
Signed-off-by: Chris Ball <chris@printf.net>

If calloc fails, bail out immediately instead of trying to
use the NULL pointer.
Signed-off-by: Michael Heimpold <michael.heimpold@i2se.com>
Cc: Michael Heimpold <mhei@heimpold.de>
Reviewed-by: Avri Altman <avri.altman@wdc.com>
Signed-off-by: Chris Ball <chris@printf.net>

Display secure removal type when printing Extended CSD
Example:
	# mmc extcsd read /dev/mmcblk0
	...
        Secure Removal Type [SECURE_REMOVAL_TYPE]: 0x39
         information is configured to be removed using a vendor defined
         Supported Secure Removal Type:
          information removed by an erase of the physical memory
          information removed using a vendor defined
Signed-off-by: Jungseung Lee <js07.lee@samsung.com>
Reviewed-by: Avri Altman <Avri.Altman@wdc.com>
Signed-off-by: Chris Ball <chris@printf.net>

On some systems which use MMC as a main storage device
it is possible that RPMB commands are mixed with
generic MMC access commands which invalidates RPMB.

This patch uses MMC_IOC_MULTI_CMD.
Signed-off-by: Clément Péron <peron.clem@gmail.com>
Reviewed-by: Avri Altman <avri.altman@wdc.com>
Signed-off-by: Chris Ball <chris@printf.net>

The JEDEC standard is confusing. The number of max blocks for reading
RPMB is determined by CMD23 which can hold an unsigned int and not only
u16. It is true that the current maximum is 64K of blocks, yet this may
be extended in the future. Let's not apply a limit here which should be
checked by the card.
Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: Chris Ball <chris@printf.net>

I got a compile error with GCC7. When trimming white spaces from strings
lsmmc uses strncpy with overlapping memory areas. This is not allowed.
In addition, the implementation was not efficient with calling strlen
and strncpy once per iteration. Refactor the code to be valid and more
effective.
Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Signed-off-by: Chris Ball <chris@printf.net>

Fix enh_start_addr overflow when converting emmc enh_area which will
cause do_enh_area_set failure.
Signed-off-by: jason.zeng <jasonzvz@gmail.com>
Signed-off-by: Chris Ball <chris@printf.net>

According to man-pages(7) the TH macro takes the following parameters:

	.TH title section date source manual

with some recommendations for the individual parameters. Adapt the usage
of TH accordingly. The motivating change is to drop "man" in the title
in favor of "MMC".
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Chris Ball <chris@printf.net>

I didn't find any functions used for which the documentation specifies to
use (at least) one of the dropped headers.

The only Linux specific header <linux/fs.h> is needed for BLKGETSIZE,
document that.
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Chris Ball <chris@printf.net>

offsetof isn't mmc specific, so remove it from mmc.h. As there is only a
single user define it there.
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Chris Ball <chris@printf.net>

Instead of adding .o and .o.d files for the files that are not
explicitly mentioned, add a pattern matching also all future files of
these types. This makes git status stop to mention

	3rdparty/hmac_sha/.hmac_sha2.o.d
	3rdparty/hmac_sha/.sha2.o.d
	3rdparty/hmac_sha/hmac_sha2.o
	3rdparty/hmac_sha/sha2.o
	.lsmmc.o.d
	lsmmc.o
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Chris Ball <chris@printf.net>

While these dependency information was generated from the start of mmc-utils
it was never used.

Using these informations results in mmc being rebuild when e.g. mmc.h
was touched.
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Chris Ball <chris@printf.net>

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Chris Ball <chris@printf.net>

This macro saves a few lines of code but is harder to read than the
equivalent code spelled out. IMHO the latter is more important, so
expand the macro everywhere and drop it.

While touching this also unbreak the strings used there for better
grepability.
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Chris Ball <chris@printf.net>

Signed-off-by: Alexander Stein <alexander.stein@systec-electronic.com>
Signed-off-by: Chris Ball <chris@printf.net>

Signed-off-by: Alexander Stein <alexander.stein@systec-electronic.com>
Signed-off-by: Chris Ball <chris@printf.net>

Signed-off-by: Boris Schmidt <boris.schmidt@siemens.com>

Fixes upstream-merge in 752072da


Write_reliability -c addition from patch was missing.
Signed-off-by: Tomas Melin <tomas.melin@vaisala.com>
Signed-off-by: Chris Ball <chris@printf.net>

Test was wrong way around. Function returns non-zero value on fail.
Signed-off-by: Tomas Melin <tomas.melin@vaisala.com>
Signed-off-by: Chris Ball <chris@printf.net>

-c option added to gp create, enh_area set and write_reliability set
commands.

Signed-off by: Tomas Melin <tomas.melin@vaisala.com>
Signed-off-by: Chris Ball <chris@printf.net>

Adding 'continue' option to:
gp create
enh_area set
write_reliability set

Partitioning commands are connected, register PARTITION_SETTING_COMPLETED
should not be written until all settings have been done.
The continue option enables writing more than one partitioning setting before
sealing settings.
Signed-off-by: Tomas Melin <tomas.melin@vaisala.com>
Signed-off-by: Chris Ball <chris@printf.net>

Signed-off-by: Tomas Melin <tomas.melin@vaisala.com>
Signed-off-by: Chris Ball <chris@printf.net>

Block-addressed devices should have address multiplied with sector size.

Clarify with comment how is_blockaddressed() is calculated.
Signed-off-by: Tomas Melin <tomas.melin@vaisala.com>
Signed-off-by: Chris Ball <chris@printf.net>

Display Command Queue information when printing Extended CSD
Example:

	# mmc extcsd read /dev/mmcblk0 | grep CMDQ
	Command Queue Support [CMDQ_SUPPORT]: 0x01
	Command Queue Depth [CMDQ_DEPTH]: 16
	Command Enabled [CMDQ_MODE_EN]: 0x01
Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Reviewed-by: Ulf Hansson <ulf.hansson@linaro.org>
Signed-off-by: Chris Ball <chris@printf.net>

Add commands to get and set write protect modes for the specified
areas of the user partition. The ability to set permanent write
protect is #ifdef'd with "DANGEROUS_COMMANDS_ENABLED" because
it has the ability to make the eMMC device and possibly the system
permanently unusable.
Signed-off-by: Al Cooper <alcooperx@gmail.com>
Signed-off-by: Chris Ball <chris@printf.net>

Modified 'bootpart enable' function to disable partition boot
by setting the <partition> parameter to 0.
Signed-off-by: Markus Schuetterle <markus.schuetterle@intel.com>
Signed-off-by: Chris Ball <chris@printf.net>

Adding support for field firmware update over multiple command ioctl.
As multiple command ioctl is supported only from kernel 4.4, this patch
should be used against kernel 4.4 and above.

Known issues:
- There is no support for Multiple Block write commands (CMD25) in existing
  IOCTL implementation
- In case MODE_OPERATION_CODES field is not supported by the device
  manual reset of the device/platform is required.
  The reset issue discussed in another email thread - " [RFC 0/6] mmc:
  Field Firmware Update"
Signed-off-by: Yaniv Agman <yaniv.agman@sandisk.com>
Signed-off-by: Avi Shchislowski <avi.shchislowski@sandisk.com>
Signed-off-by: Chris Ball <chris@printf.net>

The lsmmc tools contains an extensive parser of the CID, CSD, SCR
registers from userspace. The utility works as-is and uses sysfs
to read the register values.

The original code is created by Sebastian Rasmussen and still lives
in an attachment in the mail archive of linux-mmc. It need to be merged
into mmc-utils repository, which is convenient for testing MMC device
from userspace.

Change since v3:
 - Remove the unused EXT_CSD parser in lsmmc.c file.
Signed-off-by: Sebastian Rasmussen <sebras@gmail.com>
Signed-off-by: Chris Ball <chris@printf.net>
Signed-off-by: Baolin Wang <baolin.wang@linaro.org>

Signed-off-by: Marcus Folkesson <marcus.folkesson@gmail.com>

Signed-off-by: Marcus Folkesson <marcus.folkesson@gmail.com>

The MMC4.5 specification added an optional cache for eMMC devices.
This change will add the ability to enable/disable the cache.
Signed-off-by: Al Cooper <acooperx@gmail.com>
Signed-off-by: Chris Ball <chris@printf.net>