Haibo Chen [Tue, 26 May 2020 10:22:02 +0000 (18:22 +0800)]
mmc: sdhci-esdhc-imx: disable the CMD CRC check for standard tuning
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.
Haibo Chen [Tue, 26 May 2020 10:22:01 +0000 (18:22 +0800)]
mmc: sdhci-esdhc-imx: fix the mask for tuning start point
According the RM, the bit[6~0] of register ESDHC_TUNING_CTRL is
TUNING_START_TAP, bit[7] of this register is to disable the command
CRC check for standard tuning. So fix it here.
Haibo Chen [Wed, 27 May 2020 02:39:35 +0000 (10:39 +0800)]
mmc: host: sdhci-esdhc-imx: add wakeup feature for GPIO CD pin
When use the specific GPIO to detect the card insert/remove, we can
also add the GPIO as a wakeup source. When system suspend, insert or
remove the card can wakeup the system.
Ludovic Barre [Tue, 26 May 2020 15:51:03 +0000 (17:51 +0200)]
mmc: mmci_sdmmc: fix DMA API warning max segment size
Turning on CONFIG_DMA_API_DEBUG_SG results in the following warning:
WARNING: CPU: 1 PID: 85 at kernel/dma/debug.c:1302
debug_dma_map_sg+0x2a0/0x3cc mmci-pl18x 58005000.sdmmc: DMA-API: mapping sg
segment longer than device claims to support [len=126976] [max=65536]
dma api debug checks and compares the segment size to dma_get_max_seg_size
(dev->dma_parms->max_segment_size), the sdmmc variant has an internal DMA
and should define its max_segment_size constraint to avoid this warning.
This Patch defines the dev->dma_parms->max_segment_size with the constraint
already set for mmc core (host->mmc->max_seg_size).
Ludovic Barre [Tue, 26 May 2020 15:51:02 +0000 (17:51 +0200)]
mmc: mmci_sdmmc: fix DMA API warning overlapping mappings
Turning on CONFIG_DMA_API_DEBUG_SG results in the following warning:
WARNING: CPU: 1 PID: 20 at kernel/dma/debug.c:500 add_dma_entry+0x16c/0x17c
DMA-API: exceeded 7 overlapping mappings of cacheline 0x031d2645
Modules linked in:
CPU: 1 PID: 20 Comm: kworker/1:1 Not tainted 5.5.0-rc2-00021-gdeda30999c2b-dirty #49
Hardware name: STM32 (Device Tree Support)
Workqueue: events_freezable mmc_rescan
[<c03138c0>] (unwind_backtrace) from [<c030d760>] (show_stack+0x10/0x14)
[<c030d760>] (show_stack) from [<c0f2eb28>] (dump_stack+0xc0/0xd4)
[<c0f2eb28>] (dump_stack) from [<c034a14c>] (__warn+0xd0/0xf8)
[<c034a14c>] (__warn) from [<c034a530>] (warn_slowpath_fmt+0x94/0xb8)
[<c034a530>] (warn_slowpath_fmt) from [<c03bca0c>] (add_dma_entry+0x16c/0x17c)
[<c03bca0c>] (add_dma_entry) from [<c03bdf54>] (debug_dma_map_sg+0xe4/0x3d4)
[<c03bdf54>] (debug_dma_map_sg) from [<c0d09244>] (sdmmc_idma_prep_data+0x94/0xf8)
[<c0d09244>] (sdmmc_idma_prep_data) from [<c0d05a2c>] (mmci_prep_data+0x2c/0xb0)
[<c0d05a2c>] (mmci_prep_data) from [<c0d073ec>] (mmci_start_data+0x134/0x2f0)
[<c0d073ec>] (mmci_start_data) from [<c0d078d0>] (mmci_request+0xe8/0x154)
[<c0d078d0>] (mmci_request) from [<c0cecb44>] (mmc_start_request+0x94/0xbc)
DMA api debug brings to light leaking dma-mappings, dma_map_sg and
dma_unmap_sg are not correctly balanced.
If a request is prepared, the dma_map/unmap are done in asynchronous call
pre_req (prep_data) and post_req (unprep_data). In this case the
dma-mapping is right balanced.
But if the request was not prepared, the data->host_cookie is define to
zero and the dma_map/unmap must be done in the request. The dma_map is
called by mmci_dma_start (prep_data), but there is no dma_unmap in this
case.
This patch adds dma_unmap_sg when the dma is finalized and the data cookie
is zero (request not prepared).
Accordingly to Cadence documentation, PHY write procedure is:
1. Software sets the PHY Register Address (HRS04[5:0]) and the
PHY Write Data (HRS04[15:8]) fields.
2. Software sets the PHY Write Transaction Request (HRS04[24]) field to 1.
3. Software waits as the PHY Write Transaction Acknowledge (HRS04[26])
field is equal to 0.
4. Hardware performs the write transaction to PHY register where
HRS04[15:8] is a data written to register under HRS04[5:0] address.
5. Hardware sets the PHY Transaction Acknowledge (HRS04[26]) to 1 when
transaction is completed.
6. Software clears the PHY Write Transaction Request (HRS04[24]) to 1
after noticing that the PHY Write Transaction Acknowledge (HRS04[26])
field is equal to 1.
7. Software waits for the PHY Acknowledge Register (HRS04[26]) field is
equal to 0.
Add missing steps 3 and 7. Lack of these steps causes
integrity errors detested by hardware.
Pali Rohár [Fri, 22 May 2020 14:44:11 +0000 (16:44 +0200)]
mmc: sdio: Fix Cypress SDIO IDs macros in common include file
All macro names for SDIO device IDs are prefixed by vendor name to which
device ID belongs. So for consistency add Broadcom string vendor prefix to
all Cypress macro names as they belong to SDIO Broadcom vendor ID.
Change also Cypress 43012 value from decimal do hexadecimal notation to be
consistent with all other values.
Pali Rohár [Fri, 22 May 2020 14:44:04 +0000 (16:44 +0200)]
mmc: sdio: Move SDIO IDs from mwifiex driver to common include file
Add _WLAN suffix to macro names for consistency with other Marvell macros.
These IDs represents wlan function of combo bt/wlan cards. Other functions
of these cards have different IDs.
Signed-off-by: Pali Rohár <pali@kernel.org> Acked-by: Kalle Valo <kvalo@codeaurora.org> Link: https://lore.kernel.org/r/20200522144412.19712-4-pali@kernel.org Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org> Acked-by: Ganapathi Bhat <ganapathi.bhat@nxp.com>
Pali Rohár [Fri, 22 May 2020 14:44:02 +0000 (16:44 +0200)]
mmc: sdio: Fix macro name for Marvell device with ID 0x9134
Marvell SDIO device ID 0x9134 is used in SDIO Common CIS (Card Information
Structure) and not in SDIO wlan function (with ID 1). SDIO Common CIS is
accessed by function ID 0.
So change this misleading macro name to SDIO_DEVICE_ID_MARVELL_8887_F0 as
it does not refer to wlan function. It refers to function 0.
Wlan module on this SDIO card is available at function ID 1 and is
identified by different SDIO device ID 0x9135. Kernel quirks for SDIO
devices are matched against device ID from SDIO Common CIS. Therefore
device ID used in quirk is correct, just has misleading name.
Yangbo Lu [Fri, 22 May 2020 03:12:56 +0000 (11:12 +0800)]
mmc: sdhci-of-esdhc: exit HS400 properly before setting any speed mode
The eSDHC HS400 timing requires many specific registers setting,
unlike other speed modes which need to set only host controller 2
register. When driver needs to downgrade HS400 mode to other speed
mode, the controller have to exit HS400 timing properly first.
This patch is to support the procedure of HS400 exiting at the
beginning of esdhc_set_uhs_signaling.
Sarthak Garg [Fri, 22 May 2020 09:32:28 +0000 (15:02 +0530)]
mmc: sdhci-msm: Read and use DLL Config property from device tree file
Certain platforms require different settings in the
SDCC_HC_REG_DLL_CONFIG register. This setting can change from platform
to platform. So the driver should check whether a particular platform
require a different setting by reading the DT file and use it.
Also use msm_cm_dll_set_freq only when DLL not supplied.
Sarthak Garg [Fri, 22 May 2020 09:32:27 +0000 (15:02 +0530)]
mmc: sdhci-msm: Update DDR_CONFIG as per device tree file
Certain platforms require different settings in the
SDCC_HC_REG_DDR_CONFIG register. This setting can change from platform
to platform. So the driver should check whether a particular platform
require a different setting by reading the device tree file and use it.
mmc: host: sdhci-msm: Configure dll-user-control in dll init sequence
With SDCC v5.1.0, additional setting needed for enabling DLL output.
The dll-user-control register need to be configured during dll
initialization for getting proper dll output.
Without this configuration, we don't get the DLL lock status properly.
Also update the DLL register settings according to the SDCC Hardware
Programming Guide.
sdhci: tegra: Avoid reading autocal timeout values when not applicable
When auto calibration timeouts, calibration is disabled and fail-safe
drive strength values are programmed based on the signal voltage.
Different fail-safe drive strength values based on voltage are
applicable only for SoCs supporting 3V3 and 1V8 pad controls.
So, this patch avoids reading these properties from the device tree
for SoCs not using pad controls and the warning of missing properties
will not show up on these SoC platforms.
Ulf Hansson [Tue, 19 May 2020 15:24:45 +0000 (17:24 +0200)]
mmc: tmio: Make sure the PM domain is 'started' while probing
If the tmio device is attached to a genpd (PM domain), that genpd may have
->start|stop() callback assigned to it. To make sure the device is
accessible during ->probe(), genpd's ->start() callback must be invoked,
which is currently managed by tmio_mmc_host_probe(). However, it's likely
that may be too late for some cases, as registers may be read and written
way before that point.
To fix the behaviour, let's move the call to dev_pm_domain_start() from
tmio_mmc_host_probe() into those clients that needs it. From discussions at
linux-mmc mailing list, it turned out that it should be sufficient to do
this for the SDHI renesas variants, hence the call is move to
renesas_sdhi_probe().
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org> Reviewed-by: Wolfram Sang <wsa+renesas@sang-engineering.com> Tested-by: Wolfram Sang <wsa+renesas@sang-engineering.com> Link: https://lore.kernel.org/r/20200519152445.6922-1-ulf.hansson@linaro.org Tested-by: Geert Uytterhoeven <geert+renesas@glider.be>
Ulf Hansson [Tue, 19 May 2020 15:24:34 +0000 (17:24 +0200)]
mmc: tmio: Further fixup runtime PM management at remove
Before calling tmio_mmc_host_probe(), the caller is required to enable
clocks for its device, as to make it accessible when reading/writing
registers during probe.
Therefore, the responsibility to disable these clocks, in the error path of
->probe() and during ->remove(), is better managed outside
tmio_mmc_host_remove(). As a matter of fact, callers of
tmio_mmc_host_remove() already expects this to be the behaviour.
However, there's a problem with tmio_mmc_host_remove() when the Kconfig
option, CONFIG_PM, is set. More precisely, tmio_mmc_host_remove() may then
disable the clock via runtime PM, which leads to clock enable/disable
imbalance problems, when the caller of tmio_mmc_host_remove() also tries to
disable the same clocks.
To solve the problem, let's make sure tmio_mmc_host_remove() leaves the
device with clocks enabled, but also make sure to disable the IRQs, as we
normally do at ->runtime_suspend().
This driver has been developed as a separate module starting
from the similar sdhci-esdhc-imx.c.
Reasons for a separate sdchi-esdhc-mcf driver:
- m68K architecture does not support devicetrees, so modifying
sdhci-of-esdhc.c that is devicetree-related adding platform data
seems not appropriate,
- clock-related part, has to be implemented specifically for
mcf5441x family (see esdhc_mcf_pltfm_set_clock()),
- this is a big endian cpu accessing a big endian controller,
but about sdma, this controller does not support hw swap, which
needs to be handled with specific code,
- some other minor differences but mainly to avoid risks on
tweaking inside largely used imx driver. Adding just a small
size ColdFire-specific driver, with benefits in a further less
risky maintenance.
mmc: meson-mx-sdhc: Don't use literal 0 to initialize structs
Kbuild test robot reports the following warning in lines 56 and 87 of
drivers/mmc/host/meson-mx-sdhc-clkc.c:
Using plain integer as NULL pointer
Drop the integer value from the struct initialization to fix that
warning. This will still ensure that the compiler will zero out the
struct so it's in a well-defined state.
For Meson8 and Meson8b SoCs the vendor driver follows the following
pattern:
- for eMMC and SD cards in .set_pdma it sets:
pdma->rxfifo_manual_flush = 1;
- for SDIO cards in .set_pdma it sets:
pdma->rxfifo_manual_flush = 0;
- before syncing the DMA read buffer is sets:
pdma->rxfifo_manual_flush |= 0x02;
Set the second bit of MESON_SDHC_PDMA_RXFIFO_MANUAL_FLUSH without
clearing the first bit before syncing the DMA read buffer. This fixes a
problem where Meson8 and Meson8b SoCs would read random garbage from SD
cards. It is not clear why it worked for eMMC cards. This manifested in
the following errors when plugging in an SD card:
unrecognised SCR structure version <random number>
mmc: meson-mx-sdhc: Fix building with CONFIG_MMC_MESON_MX_SDHC=m
For an x86_64 allmodconfig build Stephen reports that building
meson-mx-sdhc-clkc.o warns that MODULE_LICENSE is missing and when
linking meson_mx_sdhc_register_clkc cannot be found.
Compile the MMC controller driver together with the build-in clock
controller driver into one module rather than using two separate
modules to fix these issues.
mmc: host: meson-mx-sdhc: new driver for the Amlogic Meson SDHC host
The SDHC MMC host controller on Amlogic SoCs provides an eMMC and MMC
card interface with 1/4/8-bit bus width.
It supports eMMC spec 4.4x/4.5x including HS200 (up to 100MHz clock).
The public S805 datasheet [0] contains a short documentation about the
registers. Unfortunately it does not describe how to use the registers
to make the hardware work. Thus this driver is based on reading (and
understanding) the Amlogic 3.10 GPL kernel code.
Some hardware details are not easy to see. Jianxin Pan was kind enough
to answer my questions:
The hardware has built-in busy timeout support. The maximum timeout is
30 seconds. This is only documented in Amlogic's internal
documentation.
The controller only works with very specific clock configurations. The
details are not part of the public datasheet. In my own words the
supported configurations are:
- 399.812kHz: clkin = 850MHz div = 2126 sd_rx_phase = 63
- 1MHz: clkin = 850MHz div = 850 sd_rx_phase = 55
- 5.986MHz: clkin = 850MHz div = 142 sd_rx_phase = 24
- 25MHz: clkin = 850MHz div = 34 sd_rx_phase = 15
- 47.222MHz: clkin = 850MHz div = 18 sd_rx_phase = 11/15 (SDR50/HS)
- 53.125MHz: clkin = 850MHz div = 16 sd_rx_phase = (tuning)
- 70.833MHz: clkin = 850MHz div = 12 sd_rx_phase = (tuning)
- 85MHz: clkin = 850MHz div = 10 sd_rx_phase = (tuning)
- 94.44MHz: clkin = 850MHz div = 9 sd_rx_phase = (tuning)
- 106.25MHz: clkin = 850MHz div = 8 sd_rx_phase = (tuning)
- 127.5MHz: clkin = 1275MHz div = 10 sd_rx_phase = (tuning)
- 141.667MHz: clkin = 850MHz div = 6 sd_rx_phase = (tuning)
- 159.375MHz: clkin = 1275MHz div = 8 sd_rx_phase = (tuning)
- 212.5MHz: clkin = 1275MHz div = 6 sd_rx_phase = (tuning)
- (sd_tx_phase is always 1, 94.44MHz is not listed in the datasheet
but this is what the 3.10 BSP kernel on Odroid-C1 actually uses)
NOTE: CMD23 support is disabled for now because it results in command
timeouts and thus decreases read performance.
dt-bindings: mmc: Document the Amlogic Meson SDHC MMC host controller
This documents the devicetree bindings for the SDHC MMC host controller
found in Meson6, Meson8, Meson8b and Meson8m2 SoCs. It can use a
bus-width of 1/4/8-bit and it supports eMMC spec 4.4x/4.5x including
HS200 mode (up to 100MHz clock). It embeds an internal clock controller
which outputs four clocks (mod_clk, sd_clk, tx_clk and rx_clk) and is
fed by four external input clocks (clkin[0-3]). "pclk" is the module
register clock, it has to be enabled to access the registers.
Masahiro Yamada [Mon, 11 May 2020 06:21:58 +0000 (15:21 +0900)]
mmc: uniphier-sd: call devm_request_irq() after tmio_mmc_host_probe()
Currently, tmio_mmc_irq() handler is registered before the host is
fully initialized by tmio_mmc_host_probe(). I did not previously notice
this problem.
The boot ROM of a new Socionext SoC unmasks interrupts (CTL_IRQ_MASK)
somehow. The handler is invoked before tmio_mmc_host_probe(), then
emits noisy call trace.
Ulf Hansson [Fri, 8 May 2020 11:29:02 +0000 (13:29 +0200)]
mmc: host: Drop redundant MMC_CAP_ERASE
The MMC_CAP_ERASE bit is no longer used by the mmc core as erase, discard
and trim operations are now always supported. Therefore, drop the bit and
move all mmc hosts away from using it.
Ulf Hansson [Fri, 8 May 2020 11:28:53 +0000 (13:28 +0200)]
mmc: core: Enable erase/discard/trim support for all mmc hosts
Step by step, mmc host drivers and the mmc core have been improved in
regards to support erase/discard/trim operations. We have now reached a
point when it no longer seems reasonable to use an opt-in approach to
enable the functionality. Therefore, let's switch to make the operations
always supported.
Ulf Hansson [Fri, 8 May 2020 09:52:28 +0000 (11:52 +0200)]
mmc: sdricoh_cs: Respect the cmd->busy_timeout from the mmc core
Using a fixed 1s polling timeout for all commands is a bit problematic.
For some commands it means waiting longer than needed for the polling to be
aborted, which may not a big issue, but still. For other commands, like for
an erase (CMD38) that uses a R1B response, may require longer timeouts than
1s. In these cases, we may end up treating the command as it failed, while
it just needed some more time to complete successfully.
Fix the problem by respecting the cmd->busy_timeout, which is provided by
the mmc core.
Note that, even if the sdricoh_cs driver may currently not support HW busy
detection on DAT0, some comments in the code refer to that the HW may
support it. Therefore, it seems better to be proactive in this case.
Ulf Hansson [Fri, 8 May 2020 09:52:18 +0000 (11:52 +0200)]
mmc: sdricoh_cs: Throttle polling rate for commands
Rather than to poll in a busy-loop, let's convert into using
read_poll_timeout() and insert a small delay between each polling attempts.
In particular, this avoids hogging the CPU.
Additionally, to convert to read_poll_timeout() we also need to switch from
using a specific number of polling attempts, into a specific timeout in us
instead. The previous 100000 attempts, is translated into a total timeout
of total 1s, as that seemed like reasonable value to pick.
Ulf Hansson [Fri, 8 May 2020 09:52:10 +0000 (11:52 +0200)]
mmc: sdricoh_cs: Throttle polling rate for data transfers
Rather than to poll in a busy-loop, let's convert into using
read_poll_timeout() and insert a small delay between each polling attempts.
In particular, this avoids hogging the CPU.
Additionally, to convert to read_poll_timeout() we also need to switch from
using a specific number of polling attempts, into a specific timeout in us
instead. The previous 100000 attempts, is translated into a total timeout
of total 1s, as that seemed like reasonable value to pick.
mmc: cb710: Inform the mmc core about the maximum busy timeout
Some commands uses R1B responses, which means the card may assert the DAT0
line to signal busy for a period of time, after it has received the
command. The mmc core normally specifies the busy period for the command in
the cmd->busy_timeout. Ideally the driver should respect it, but that
requires quite some update of the code, so let's defer that to someone with
the HW at hand.
Instead, let's inform the mmc core about the maximum supported busy timeout
in ->max_busy_timeout during ->probe(). This value corresponds to the fixed
~2s timeout of the polling loop, implemented in cb710_wait_for_event(). In
this way, we let the mmc core validate the needed timeout, which may lead
to that it converts from a R1B into a R1 response and then use CMD13 to
poll for busy completion.
In other words, this change enables support for commands with longer busy
periods than 2s, like erase (CMD38) for example.
memstick: Replace zero-length array with flexible-array
The current codebase makes use of the zero-length array language
extension to the C90 standard, but the preferred mechanism to declare
variable-length types such as these ones is a flexible array member[1][2],
introduced in C99:
struct foo {
int stuff;
struct boo array[];
};
By making use of the mechanism above, we will get a compiler warning
in case the flexible array does not occur last in the structure, which
will help us prevent some kind of undefined behavior bugs from being
inadvertently introduced[3] to the codebase from now on.
Also, notice that, dynamic memory allocations won't be affected by
this change:
"Flexible array members have incomplete type, and so the sizeof operator
may not be applied. As a quirk of the original implementation of
zero-length arrays, sizeof evaluates to zero."[1]
sizeof(flexible-array-member) triggers a warning because flexible array
members have incomplete type[1]. There are some instances of code in
which the sizeof operator is being incorrectly/erroneously applied to
zero-length arrays and the result is zero. Such instances may be hiding
some bugs. So, this work (flexible-array member conversions) will also
help to get completely rid of those sorts of issues.
mmc: meson-mx-sdio: trigger a soft reset after a timeout or CRC error
The vendor driver (from the 3.10 kernel) triggers a soft reset every
time before starting a new command. While this fixes a problem where
SDIO cards are not detected at all (because all commands simply
timed out) this hurts SD card read performance a bit (in my tests
between 10% to 20%).
Trigger a soft reset after we got a CRC error or if the previous command
timed out (just like the vendor driver from the same 3.10 kernel for the
newer SDHC controller IP does). This fixes detection of SDIO cards and
doesn't hurt SD card read performance at the same time.
Krishna Konda [Fri, 1 May 2020 13:53:01 +0000 (19:23 +0530)]
mmc: core: expose info about enhanced rpmb support
Following eMMC JEDEC JESD84-B51 standard, an enhanced form of
rpmb is supported. What this enhanced mode supports is in addition
to be able to write one rpmb or two rpmb frames at a time,
32 frames can be written at a time.
Expose this information present in ext csd field so that the
user space application that wants to make use of this can do so.
mmc: sdio: Align the initialization commands in retry path for UHS-I
According to the comment in mmc_sdio_reinit_card(), some SDIO cards may
require a "[CMD5,5,3,7] init sequence", which isn't always obeyed in
mmc_sdio_init_card(). Especially, when we end up retrying the UHS-I
specific initialization, there is a missing CMD5.
Let's update the code to make the behaviour consistent and let's also take
the opportunity to clean up the code a bit, to avoid open coding.
mmc: sdio: Re-use negotiated OCR mask when re-sending CMD8
While initializing an SDIO card in mmc_sdio_init_card(), we may need to
retry the UHS-I specific initialization, in case the first attempt fails.
This leads to resending a CMD8, but also to restart from scratch with the
so called OCR mask negotiations. This is unnecessary as we already have a
negotiated OCR mask, so let's use that instead. In this way, the behaviour
also becomes more consistent with other similar paths.
mmc: sdio: Fix several potential memory leaks in mmc_sdio_init_card()
Over the years, the code in mmc_sdio_init_card() has grown to become quite
messy. Unfortunate this has also lead to that several paths are leaking
memory in form of an allocated struct mmc_card, which includes additional
data, such as initialized struct device for example.
Unfortunate, it's a too complex task find each offending commit. Therefore,
this change fixes all memory leaks at once.
mmc: sdio: Fix potential NULL pointer error in mmc_sdio_init_card()
During some scenarios mmc_sdio_init_card() runs a retry path for the UHS-I
specific initialization, which leads to removal of the previously allocated
card. A new card is then re-allocated while retrying.
However, in one of the corresponding error paths we may end up to remove an
already removed card, which likely leads to a NULL pointer exception. So,
let's fix this.
mmc: sdhci-msm: Fix error handling for dev_pm_opp_of_add_table()
Even though specifying OPP's in device tree is optional, ignoring all errors
reported by dev_pm_opp_of_add_table() means we can't distinguish between a
missing OPP table and a wrong/buggy OPP table. While missing OPP table
(dev_pm_opp_of_add_table() returns a -ENODEV in such case) can be ignored,
a wrong/buggy OPP table in device tree should make the driver error out.
while we fix that, lets also fix the variable names for opp/opp_table to
avoid confusion and name them opp_table/has_opp_table instead.
(Note: for 4tap SoCs, the numbers get divided by 2)
Do not use these TAPs in HS400, and also don't use auto correction but
manual correction.
We check for bad taps in two places:
1) After tuning HS400: Then, we select a neighbouring TAP. One of them
must be good, because there are never three bad taps in a row.
Retuning won't help because we just finished tuning.
2) After a manual correction request: Here, we can't switch to the
requested TAP. But we can retune (if the HS200 tuning was good)
because the environment might have changed since the last tuning.
If not, we stay on the same TAP.
Signed-off-by: Takeshi Saito <takeshi.saito.xv@renesas.com>
[wsa: refactored to match upstream driver, reworded commit msg] Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com> Reviewed-by: Yoshihiro Shimoda <yoshihiro.shimoda.uh@renesas.com> Link: https://lore.kernel.org/r/20200423130432.9990-3-wsa+renesas@sang-engineering.com Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
mmc: mmc_spi: Respect the cmd->busy_timeout from the mmc core
Using a fixed 3s polling timeout for all commands with R1B responses is a
bit problematic.
For some commands it means waiting longer than needed for the polling to be
aborted, which may not a big issue, but still. For other commands, like for
an erase (CMD38), may require longer timeouts than 3s. In these cases, we
may end up treating the command as it failed, while it just needed some
more time to complete successfully.
Fix the problem by respecting the cmd->busy_timeout, which is provided by
the mmc core.
Clarify the use of r1b_timeout, by renaming it to MMC_SPI_R1B_TIMEOUT_MS
and by dropping the corresponding confusing comment about it.
Additionally, let's also add a new define, MMC_SPI_INIT_TIMEOUT_MS and use
it during the initialization. Even if these two defines are given the same
value, the split makes it easier to understand them.
mmc: via-sdmmc: Respect the cmd->busy_timeout from the mmc core
Using a fixed 1s timeout for all commands (and data transfers) is a bit
problematic.
For some commands it means waiting longer than needed for the timer to
expire, which may not a big issue, but still. For other commands, like for
an erase (CMD38) that uses a R1B response, may require longer timeouts than
1s. In these cases, we may end up treating the command as it failed, while
it just needed some more time to complete successfully.
Fix the problem by respecting the cmd->busy_timeout, which is provided by
the mmc core.
mmc: tifm_sd: Inform the mmc core about the maximum busy timeout
Some commands uses R1B responses, which means the card may assert the DAT0
line to signal busy for a period of time, after it has received the
command. The mmc core normally specifies the busy period for the command in
the cmd->busy_timeout. Ideally the driver should respect it, but that
requires quite some update of the code, so let's defer that to someone with
the HW at hand.
Instead, let's inform the mmc core about the maximum supported busy timeout
in ->max_busy_timeout during ->probe(). This value corresponds to the fixed
1s timeout used by tifm_sd. In this way, we let the mmc core validate the
needed timeout, which may lead to that it converts from a R1B into a R1
response and then use CMD13 to poll for busy completion.
In other words, this change enables support for commands with longer busy
periods than 1s, like erase (CMD38) for example.
mmc: owl-mmc: Respect the cmd->busy_timeout from the mmc core
For commands that doesn't involve to prepare a data transfer, owl-mmc is
using a fixed 30s response timeout. This is a bit problematic.
For some commands it means waiting longer than needed for the completion to
expire, which may not a big issue, but still. For other commands, like for
an erase (CMD38) that uses a R1B response, may require longer timeouts than
30s. In these cases, we may end up treating the command as it failed, while
it just needed some more time to complete successfully.
Fix the problem by respecting the cmd->busy_timeout, which is provided by
the mmc core.
mmc: jz4740: Inform the mmc core about the maximum busy timeout
Some commands uses R1B responses, which means the card may assert the DAT0
line to signal busy for a period of time, after it has received the
command. The mmc core normally specifies the busy period for the command in
the cmd->busy_timeout. Ideally the driver should respect it, but that
requires quite some update of the code, so let's defer that to someone with
the HW at hand.
Instead, let's inform the mmc core about the maximum supported busy timeout
in ->max_busy_timeout during ->probe(). This value corresponds to the fixed
5s timeout used by jz4740. In this way, we let the mmc core validate the
needed timeout, which may lead to that it converts from a R1B into a R1
response and then use CMD13 to poll for busy completion.
In other words, this change enables support for commands with longer busy
periods than 5s, like erase (CMD38) for example.
Zou Wei [Thu, 23 Apr 2020 06:43:19 +0000 (14:43 +0800)]
mmc: sdhci-pci-o2micro: Make some symbols static
Fix the following sparse warning:
drivers/mmc/host/sdhci-pci-o2micro.c:497:6: warning: symbol
'sdhci_pci_o2_set_clock' was not declared. Should it be static?
drivers/mmc/host/sdhci-pci-o2micro.c:512:5: warning: symbol
'sdhci_pci_o2_probe_slot' was not declared. Should it be static?
drivers/mmc/host/sdhci-pci-o2micro.c:581:5: warning: symbol
'sdhci_pci_o2_probe' was not declared. Should it be static?
drivers/mmc/host/sdhci-pci-o2micro.c:786:5: warning: symbol
'sdhci_pci_o2_resume' was not declared. Should it be static?
Zou Wei [Thu, 23 Apr 2020 09:15:19 +0000 (17:15 +0800)]
mmc: core: Use DEFINE_DEBUGFS_ATTRIBUTE instead of DEFINE_SIMPLE_ATTRIBUTE
Fixes coccicheck warning:
drivers/mmc/core/debugfs.c:222:0-23: WARNING: mmc_clock_fops should be
defined with DEFINE_DEBUGFS_ATTRIBUTE
Use DEFINE_DEBUGFS_ATTRIBUTE rather than DEFINE_SIMPLE_ATTRIBUTE
for debugfs files
Fixes: b4ae90fa7afc7 ("mmc: add a file to debugfs for changing host clock at runtime") Fixes: afca4e1d4466f ("mmc: core: no need to check return value of debugfs_create functions") Reported-by: Hulk Robot <hulkci@huawei.com> Signed-off-by: Zou Wei <zou_wei@huawei.com> Link: https://lore.kernel.org/r/1587633319-19835-1-git-send-email-zou_wei@huawei.com Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
mmc: mmci_sdmmc: fix power on issue due to pwr_reg initialization
This patch fix a power-on issue, and avoid to retry the power sequence.
In power off sequence: sdmmc must set pwr_reg in "power-cycle" state
(value 0x2), to prevent the card from being supplied through the signal
lines (all the lines are driven low).
In power on sequence: when the power is stable, sdmmc must set pwr_reg
in "power-off" state (value 0x0) to drive all signal to high before to
set "power-on".
To avoid writing the same value to the power register several times, this
register is cached by the pwr_reg variable. At probe pwr_reg is initialized
to 0 by kzalloc of mmc_alloc_host.
Like pwr_reg value is 0 at probing, the power on sequence fail because
the "power-off" state is not writes (value 0x0) and the lines
remain drive to low.
This patch initializes "pwr_reg" variable with power register value.
This it done in sdmmc variant init to not disturb default mmci behavior.
On some qualcomm SoCs we need to vote on a performance state of a power
domain depending on the clock rates. Hence move to using OPP api to set
the clock rate and performance state specified in the OPP table.
On platforms without an OPP table, dev_pm_opp_set_rate() is eqvivalent to
clk_set_rate()
staging: greybus: sdio: Respect the cmd->busy_timeout from the mmc core
Using a fixed 1s timeout for all commands is a bit problematic.
For some commands it means waiting longer than needed for the timeout to
expire, which may not a big issue, but still. For other commands, like for
an erase (CMD38) that uses a R1B response, may require longer timeouts than
1s. In these cases, we may end up treating the command as it failed, while
it just needed some more time to complete successfully.
Fix the problem by respecting the cmd->busy_timeout, which is provided by
the mmc core.
Cc: Rui Miguel Silva <rmfrfs@gmail.com> Cc: Johan Hovold <johan@kernel.org> Cc: Alex Elder <elder@kernel.org> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: greybus-dev@lists.linaro.org Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org> Acked-by: Rui Miguel Silva <rmfrfs@gmail.com> Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Link: https://lore.kernel.org/r/20200414161413.3036-20-ulf.hansson@linaro.org Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
mmc: usdhi6rol0: Inform the mmc core about the maximum busy timeout
Some commands uses R1B responses, which means the card may assert the DAT0
line to signal busy for a period of time, after it has received the
command. The mmc core normally specifies the busy period for the command in
the cmd->busy_timeout. Ideally the driver should respect it, but that
requires quite some update of the code, so let's defer that to someone with
the HW at hand.
Instead, let's inform the mmc core about the maximum supported busy timeout
in ->max_busy_timeout during ->probe(). This value corresponds to the fixed
4s timeout used by usdhi6rol0. In this way, we let the mmc core validate
the needed timeout, which may lead to that it converts from a R1B into a R1
response and then use CMD13 to poll for busy completion.
In other words, this change enables support for commands with longer busy
periods than 4s, like erase (CMD38) for example.
mmc: atmel-mci: Respect the cmd->busy_timeout from the mmc core
Using a fixed 2s timeout for all commands is a bit problematic.
For some commands it means waiting longer than needed for the timer to
expire, which may not a big issue, but still. For other commands, like for
an erase (CMD38) that uses a R1B response, may require longer timeouts than
2s. In these cases, we may end up treating the command as it failed, while
it just needed some more time to complete successfully.
Fix the problem by respecting the cmd->busy_timeout, which is provided by
the mmc core.
mmc: atmel-mci: Keep timer enabled when queuing a next request
When atmci_request_end() is about to finish a request for one slot, there
is a possibility that there is new request queued for another slot. If this
turns out to be the case, the new request is started and the timer is
re-programmed for it.
Although, a few lines below in atmci_request_end(), this timer becomes
deleted, likely corresponding to the other recently completed request. This
looks wrong, so let's fix it.
mmc: sdhci: add spin lock for sdhci_set_default_irqs in sdhci_init
When use one SDIO wifi which enable the runtime PM feature on i.MX6SX,
we meet system hang. This hang happened during the usdhc runtime resume,
in sdhci_init(), when call the sdhci_set_default_irqs. One interrupt
(SDHCI_INT_CARD_INT) triggered just after the host->ier update and before
the write of register SDHCI_SIGNAL_ENABLE. So in sdhci_irq, it will skip
the call of sdio_signal_irq() because current host->ier do not set the
SDHCI_INT_CARD_INT. So this SDIO wifi interrupt always keep triggered,
let the system stuck in irq handle, can't response any other thread.
This patch add spin lock for the sdhci_set_default_irqs to fix this issue.
Baolin Wang [Mon, 13 Apr 2020 02:46:03 +0000 (10:46 +0800)]
mmc: host: Introduce the request_atomic() for the host
The SD host controller can process one request in the atomic context if
the card is nonremovable, which means we can submit next request in the
irq hard handler when using the MMC host software queue to reduce the
latency. Thus this patch adds a new API request_atomic() for the host
controller, as well as adding support for host software queue to submit
a request by the new request_atomic() API.
Moreover there is an unusual case that the card is busy when trying to
send a command, and we can not polling the card status in interrupt
context by using request_atomic() to dispatch requests. Thus we should
queue a work to try again in the non-atomic context in case the host
releases the busy signal later.
Adrian Hunter [Sun, 12 Apr 2020 09:03:49 +0000 (12:03 +0300)]
mmc: sdhci: Reduce maximum time under spinlock in sdhci_send_command()
Spending time under spinlock increases IRQ latencies and also
response times because preemption is disabled.
sdhci_send_command() waits up to 10 ms under spinlock for inhibit bits
to clear. In general inhibit bits will not be set, but there may be
corner cases, especially in the face of errors, where waiting helps.
There might also be dysfunctional hardware that needs the waiting. So
retain the legacy behaviour but do not wait for inhibit bits while under
spinlock. Instead adjust the logic to enable waiting while not under
spinlock. That is mostly straight forward, but in the interrupt handler
it requires deferring an "inhibited" command to the IRQ thread where
sleeping is allowed.
projects / kernel.git / log