In may, Steven sent a patch deleting the bounce buffer handling
and the CONFIG_MMC_BLOCK_BOUNCE option.
I chose the less invasive path of making it a runtime config
option, and we merged that successfully for kernel v4.12.
The code is however just standing in the way and taking up
space for seemingly no gain on any systems in wide use today.
Pierre says the code was there to improve speed on TI SDHCI
controllers on certain HP laptops and possibly some Ricoh
controllers as well. Early SDHCI controllers lacked the
scatter-gather feature, which made software bounce buffers
a significant speed boost.
We are clearly talking about the list of SDHCI PCI-based
MMC/SD card readers found in the pci_ids[] list in
drivers/mmc/host/sdhci-pci-core.c.
The TI SDHCI derivative is not supported by the upstream
kernel. This leaves the Ricoh.
What we can however notice is that the x86 defconfigs in the
kernel did not enable CONFIG_MMC_BLOCK_BOUNCE option, which
means that any such laptop would have to have a custom
configured kernel to actually take advantage of this
bounce buffer speed-up. It simply seems like there was
a speed optimization for the Ricoh controllers that noone
was using. (I have not checked the distro defconfigs but
I am pretty sure the situation is the same there.)
Bounce buffers increased performance on the OMAP HSMMC
at one point, and was part of the original submission in
commit a45c6cb816 ("[ARM] 5369/1: omap mmc: Add new
omap hsmmc controller for 2430 and 34xx, v3")
This optimization was removed in
commit 0ccd76d4c2 ("omap_hsmmc: Implement scatter-gather
emulation")
which found that scatter-gather emulation provided even
better performance.
The same was introduced for SDHCI in
commit 2134a922c6 ("sdhci: scatter-gather (ADMA) support")
I am pretty positively convinced that software
scatter-gather emulation will do for any host controller what
the bounce buffers were doing. Essentially, the bounce buffer
was a reimplementation of software scatter-gather-emulation in
the MMC subsystem, and it should be done away with.
Cc: Pierre Ossman <pierre@ossman.eu>
Cc: Juha Yrjola <juha.yrjola@solidboot.com>
Cc: Steven J. Hill <Steven.Hill@cavium.com>
Cc: Shawn Lin <shawn.lin@rock-chips.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Suggested-by: Steven J. Hill <Steven.Hill@cavium.com>
Suggested-by: Shawn Lin <shawn.lin@rock-chips.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Pull MMC updates from Ulf Hansson:
"MMC core:
- Continue to refactor the mmc block code to prepare for blkmq
- Move mmc block debugfs into block module
- Next step for eMMC CMDQ by adding a new mmc host interface for it
- Move Kconfig option MMC_DEBUG from core to host
- Some additional minor improvements
MMC host:
- Declare structs as const when applicable
- Explicitly request exclusive reset control when applicable
- Improve some error paths and other various cleanups
- sdhci: Preparations to support SDHCI OMAP
- sdhci: Improve some PM related code
- sdhci: Re-factoring and modernizations
- sdhci-xenon: Add runtime PM and system sleep support
- sdhci-xenon: Add support for eMMC HS400 Enhanced Strobe
- sdhci-cadence: Add system sleep support
- sdhci-of-at91: Improve system sleep support
- dw_mmc: Add support for Hisilicon hi3660
- sunxi: Add support for A83T eMMC
- sunxi: Add support for DDR52 mode
- meson-gx: Add support for UHS-I SD-cards
- meson-gx: Cleanups and improvements
- tmio: Fix CMD12 (STOP) handling
- tmio: Cleanups and improvements
- renesas_sdhi: Add r8a7743/5 support
- renesas-sdhi: Add support for R-Car Gen3 SDHI DMAC
- renesas_sdhi: Cleanups and improvements"
* tag 'mmc-v4.14' of git://git.kernel.org/pub/scm/linux/kernel/git/ulfh/mmc: (145 commits)
mmc: renesas_sdhi: Add r8a7743/5 support
mmc: meson-gx: fix __ffsdi2 undefined on arm32
mmc: sdhci-xenon: add runtime pm support and reimplement standby
mmc: core: Move mmc_start_areq() declaration
mmc: mmci: stop building qcom dml as module
mmc: sunxi: Reset the device at probe time
clk: sunxi-ng: Provide a default reset hook
mmc: meson-gx: rework tuning function
mmc: meson-gx: change default tx phase
mmc: meson-gx: implement voltage switch callback
mmc: meson-gx: use CCF to handle the clock phases
mmc: meson-gx: implement card_busy callback
mmc: meson-gx: simplify interrupt handler
mmc: meson-gx: work around clk-stop issue
mmc: meson-gx: fix dual data rate mode frequencies
mmc: meson-gx: rework clock init function
mmc: meson-gx: rework clk_set function
mmc: meson-gx: rework set_ios function
mmc: meson-gx: cfg init overwrite values
mmc: meson-gx: initialize sane clk default before clock register
...
mmc_start_areq() is an internal mmc core API. Move the declaration
accordingly.
Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Most of the information needed to issue requests to a CQE is already in
struct mmc_request and struct mmc_data. Add data block address, some flags,
and the task id (tag), and allow for cmd being NULL which it is for CQE
tasks.
Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Per the spec of JESD84-B51, section 7.3, replace tacc with taac to
fix the obvious typo.
Signed-off-by: Shawn Lin <shawn.lin@rock-chips.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
The intention of this check was to prevent the conflict between
hotplug and removing driver for whatever reason. Currently it
doesn't improve anything and the following rescan process could
still saftly perform the scan flow. So these code seems pointless
now and let's remove them.
Signed-off-by: Shawn Lin <shawn.lin@rock-chips.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Add support for CYW4373 SDIO/USB chipset.
CYW4373 is a 1x1 dual-band 11ac chipset with 20/40/80Mhz channel support.
It's a WiFi/BT combo device.
Signed-off-by: Chi-Hsien Lin <chi-hsien.lin@cypress.com>
Reviewed-by: Arend van Spriel <arend.vanspriel@broadcom.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Add host capability MMC_CAP_CD_WAKE to enable irq wake on the card detect
irq.
Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
The MMC_CAP2_HC_ERASE_SZ is used only by a few mmc host drivers. Its intent
is to enable eMMC's high-capacity erase size, as to improve the behaviour
of the erase operations.
We should strive to avoid software configuration options that aren't
necessary, but instead deploy common behaviours. For these reasons, let's
remove the capability bit for MMC_CAP2_HC_ERASE_SZ and make it the default
behaviour.
Note that this change doesn't affect eMMCs supporting trim/discard, because
these commands operates on sectors and takes precedence over erase
commands.
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Acked-by: Adrian Hunter <adrian.hunter@intel.com>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Reviewed-by: Shawn Lin <shawn.lin@rock-chips.com>
Tested-by: Shawn Lin <shawn.lin@rock-chips.com>
The mmc_queue_req is a per-request state container the MMC core uses
to carry bounce buffers, pointers to asynchronous requests and so on.
Currently allocated as a static array of objects, then as a request
comes in, a mmc_queue_req is assigned to it, and used during the
lifetime of the request.
This is backwards compared to how other block layer drivers work:
they usally let the block core provide a per-request struct that get
allocated right beind the struct request, and which can be obtained
using the blk_mq_rq_to_pdu() helper. (The _mq_ infix in this function
name is misleading: it is used by both the old and the MQ block
layer.)
The per-request struct gets allocated to the size stored in the queue
variable .cmd_size initialized using the .init_rq_fn() and
cleaned up using .exit_rq_fn().
The block layer code makes the MMC core rely on this mechanism to
allocate the per-request mmc_queue_req state container.
Doing this make a lot of complicated queue handling go away. We only
need to keep the .qnct that keeps count of how many request are
currently being processed by the MMC layer. The MQ block layer will
replace also this once we transition to it.
Doing this refactoring is necessary to move the ioctl() operations
into custom block layer requests tagged with REQ_OP_DRV_[IN|OUT]
instead of the custom code using the BigMMCHostLock that we have
today: those require that per-request data be obtainable easily from
a request after creating a custom request with e.g.:
struct request *rq = blk_get_request(q, REQ_OP_DRV_IN, __GFP_RECLAIM);
struct mmc_queue_req *mq_rq = req_to_mq_rq(rq);
And this is not possible with the current construction, as the request
is not immediately assigned the per-request state container, but
instead it gets assigned when the request finally enters the MMC
queue, which is way too late for custom requests.
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
[Ulf: Folded in the fix to drop a call to blk_cleanup_queue()]
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Tested-by: Heiner Kallweit <hkallweit1@gmail.com>
This option is activated by all multiplatform configs and what
not so we almost always have it turned on, and the memory it
saves is negligible, even more so moving forward. The actual
bounce buffer only gets allocated only when used, the only
thing the ifdefs are saving is a little bit of code.
It is highly improper to have this as a Kconfig option that
get turned on by Kconfig, make this a pure runtime-thing and
let the host decide whether we use bounce buffers. We add a
new property "disable_bounce" to the host struct.
Notice that mmc_queue_calc_bouncesz() already disables the
bounce buffers if host->max_segs != 1, so any arch that has a
maximum number of segments higher than 1 will have bounce
buffers disabled.
The option CONFIG_MMC_BLOCK_BOUNCE is default y so the
majority of platforms in the kernel already have it on, and
it then gets turned off at runtime since most of these have
a host->max_segs > 1. The few exceptions that have
host->max_segs == 1 and still turn off the bounce buffering
are those that disable it in their defconfig.
Those are the following:
arch/arm/configs/colibri_pxa300_defconfig
arch/arm/configs/zeus_defconfig
- Uses MMC_PXA, drivers/mmc/host/pxamci.c
- Sets host->max_segs = NR_SG, which is 1
- This needs its bounce buffer deactivated so we set
host->disable_bounce to true in the host driver
arch/arm/configs/davinci_all_defconfig
- Uses MMC_DAVINCI, drivers/mmc/host/davinci_mmc.c
- This driver sets host->max_segs to MAX_NR_SG, which is 16
- That means this driver anyways disabled bounce buffers
- No special action needed for this platform
arch/arm/configs/lpc32xx_defconfig
arch/arm/configs/nhk8815_defconfig
arch/arm/configs/u300_defconfig
- Uses MMC_ARMMMCI, drivers/mmc/host/mmci.[c|h]
- This driver by default sets host->max_segs to NR_SG,
which is 128, unless a DMA engine is used, and in that case
the number of segments are also > 1
- That means this driver already disables bounce buffers
- No special action needed for these platforms
arch/arm/configs/sama5_defconfig
- Uses MMC_SDHCI, MMC_SDHCI_PLTFM, MMC_SDHCI_OF_AT91, MMC_ATMELMCI
- Uses drivers/mmc/host/sdhci.c
- Normally sets host->max_segs to SDHCI_MAX_SEGS which is 128 and
thus disables bounce buffers
- Sets host->max_segs to 1 if SDHCI_USE_SDMA is set
- SDHCI_USE_SDMA is only set by SDHCI on PCI adapers
- That means that for this platform bounce buffers are already
disabled at runtime
- No special action needed for this platform
arch/blackfin/configs/CM-BF533_defconfig
arch/blackfin/configs/CM-BF537E_defconfig
- Uses MMC_SPI (a simple MMC card connected on SPI pins)
- Uses drivers/mmc/host/mmc_spi.c
- Sets host->max_segs to MMC_SPI_BLOCKSATONCE which is 128
- That means this platform already disables bounce buffers at
runtime
- No special action needed for these platforms
arch/mips/configs/cavium_octeon_defconfig
- Uses MMC_CAVIUM_OCTEON, drivers/mmc/host/cavium.c
- Sets host->max_segs to 16 or 1
- Setting host->disable_bounce to be sure for the 1 case
arch/mips/configs/qi_lb60_defconfig
- Uses MMC_JZ4740, drivers/mmc/host/jz4740_mmc.c
- This sets host->max_segs to 128 so bounce buffers are
already runtime disabled
- No action needed for this platform
It would be interesting to come up with a list of the platforms
that actually end up using bounce buffers. I have not been
able to infer such a list, but it occurs when
host->max_segs == 1 and the bounce buffering is not explicitly
disabled.
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
For hosts not supporting MMC_CAP2_SDIO_IRQ_NOTHREAD but MMC_CAP_SDIO_IRQ,
the SDIO IRQs are processed from a dedicated kernel thread. For these
cases, the host calls mmc_signal_sdio_irq() from its ISR to signal a new
SDIO IRQ.
Signaling an SDIO IRQ makes the host's ->enable_sdio_irq() callback to be
invoked to temporary disable the IRQs, before the kernel thread is woken up
to process it. When processing of the IRQs are completed, they are
re-enabled by the kernel thread, again via invoking the host's
->enable_sdio_irq().
The observation from this, is that the execution path is being unnecessary
complex, as the host driver already knows that it needs to temporary
disable the IRQs before signaling a new one. Moreover, replacing the kernel
thread with a work/workqueue would not only greatly simplify the code, but
also make it more robust.
To address the above problems, let's continue to build upon the support for
MMC_CAP2_SDIO_IRQ_NOTHREAD, as it already implements SDIO IRQs to be
processed without using the clumsy kernel thread and without the ping-pong
calls of the host's ->enable_sdio_irq() callback for each processed IRQ.
Therefore, let's add new API sdio_signal_irq(), which enables hosts to
signal/process SDIO IRQs by using a work/workqueue, rather than using the
kernel thread.
Add also a new host callback ->ack_sdio_irq(), which the work invokes when
the SDIO IRQs have been processed. This informs the host about when it
shall re-enable the SDIO IRQs. Potentially, we could re-use the existing
->enable_sdio_irq() callback instead of adding a new one, however it has
turned out that it's more convenient for hosts to get this information via
a separate callback.
Hosts that wants to use this new method to signal/process SDIO IRQs, must
enable MMC_CAP2_SDIO_IRQ_NOTHREAD and implement the ->ack_sdio_irq()
callback.
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Tested-by: Douglas Anderson <dianders@chromium.org>
Reviewed-by: Douglas Anderson <dianders@chromium.org>
Some hosts controllers, like Cavium, needs to know whether the card
operates in byte- or block-address mode. Therefore export a new API,
mmc_card_is_blockaddr(), which provides this information.
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Signed-off-by: Steven J. Hill <Steven.Hill@cavium.com>
Acked-by: David Daney <david.daney@cavium.com>
Normal read and write commands may not be used while the command queue is
enabled. Disable the Command Queue when mmc_test is probed and re-enable it
when it is removed.
Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Reviewed-by: Harjani Ritesh <riteshh@codeaurora.org>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
eMMC can have multiple internal partitions that are represented as separate
disks / queues. However switching between partitions is only done when the
queue is empty. Consequently the array of mmc requests that are queued can
be shared between partitions saving memory.
Keep a pointer to the mmc request queue on the card, and use that instead
of allocating a new one for each partition.
Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Certain 64-bit systems (e.g. Amlogic Meson GX) require buffers to be
used for DMA to be 8-byte-aligned. struct sdio_func has an embedded
small DMA buffer not meeting this requirement.
When testing switching to descriptor chain mode in meson-gx driver
SDIO is broken therefore. Fix this by allocating the small DMA buffer
separately as kmalloc ensures that the returned memory area is
properly aligned for every basic data type.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Tested-by: Helmut Klein <hgkr.klein@gmail.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Pull networking updates from David Miller:
"Highlights:
1) Support TX_RING in AF_PACKET TPACKET_V3 mode, from Sowmini
Varadhan.
2) Simplify classifier state on sk_buff in order to shrink it a bit.
From Willem de Bruijn.
3) Introduce SIPHASH and it's usage for secure sequence numbers and
syncookies. From Jason A. Donenfeld.
4) Reduce CPU usage for ICMP replies we are going to limit or
suppress, from Jesper Dangaard Brouer.
5) Introduce Shared Memory Communications socket layer, from Ursula
Braun.
6) Add RACK loss detection and allow it to actually trigger fast
recovery instead of just assisting after other algorithms have
triggered it. From Yuchung Cheng.
7) Add xmit_more and BQL support to mvneta driver, from Simon Guinot.
8) skb_cow_data avoidance in esp4 and esp6, from Steffen Klassert.
9) Export MPLS packet stats via netlink, from Robert Shearman.
10) Significantly improve inet port bind conflict handling, especially
when an application is restarted and changes it's setting of
reuseport. From Josef Bacik.
11) Implement TX batching in vhost_net, from Jason Wang.
12) Extend the dummy device so that VF (virtual function) features,
such as configuration, can be more easily tested. From Phil
Sutter.
13) Avoid two atomic ops per page on x86 in bnx2x driver, from Eric
Dumazet.
14) Add new bpf MAP, implementing a longest prefix match trie. From
Daniel Mack.
15) Packet sample offloading support in mlxsw driver, from Yotam Gigi.
16) Add new aquantia driver, from David VomLehn.
17) Add bpf tracepoints, from Daniel Borkmann.
18) Add support for port mirroring to b53 and bcm_sf2 drivers, from
Florian Fainelli.
19) Remove custom busy polling in many drivers, it is done in the core
networking since 4.5 times. From Eric Dumazet.
20) Support XDP adjust_head in virtio_net, from John Fastabend.
21) Fix several major holes in neighbour entry confirmation, from
Julian Anastasov.
22) Add XDP support to bnxt_en driver, from Michael Chan.
23) VXLAN offloads for enic driver, from Govindarajulu Varadarajan.
24) Add IPVTAP driver (IP-VLAN based tap driver) from Sainath Grandhi.
25) Support GRO in IPSEC protocols, from Steffen Klassert"
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next: (1764 commits)
Revert "ath10k: Search SMBIOS for OEM board file extension"
net: socket: fix recvmmsg not returning error from sock_error
bnxt_en: use eth_hw_addr_random()
bpf: fix unlocking of jited image when module ronx not set
arch: add ARCH_HAS_SET_MEMORY config
net: napi_watchdog() can use napi_schedule_irqoff()
tcp: Revert "tcp: tcp_probe: use spin_lock_bh()"
net/hsr: use eth_hw_addr_random()
net: mvpp2: enable building on 64-bit platforms
net: mvpp2: switch to build_skb() in the RX path
net: mvpp2: simplify MVPP2_PRS_RI_* definitions
net: mvpp2: fix indentation of MVPP2_EXT_GLOBAL_CTRL_DEFAULT
net: mvpp2: remove unused register definitions
net: mvpp2: simplify mvpp2_bm_bufs_add()
net: mvpp2: drop useless fields in mvpp2_bm_pool and related code
net: mvpp2: remove unused 'tx_skb' field of 'struct mvpp2_tx_queue'
net: mvpp2: release reference to txq_cpu[] entry after unmapping
net: mvpp2: handle too large value in mvpp2_rx_time_coal_set()
net: mvpp2: handle too large value handling in mvpp2_rx_pkts_coal_set()
net: mvpp2: remove useless arguments in mvpp2_rx_{pkts, time}_coal_set
...
Adding dedicated flag for AUTO_BKOPS in card->ext_csd structure.
Read AUTO_BKOPS bit value from the device EXT_CSD and set to the
card->ext_csd structure.
In mmc_decode_ext_csd() add a print message in case the AUTO_BKOPS
is enabled
Signed-off-by: Uri Yanai <uri.yanai@sandisk.com>
Signed-off-by: Alex Lemberg <alex.lemberg@sandisk.com>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
With the coexisting __mmc_start_request(), mmc_start_request()
and __mmc_start_req() it is a bit confusing that mmc_start_req()
actually does not start a normal request, but an asynchronous
request.
Rename it to mmc_start_areq() to make it explicit what the
function is doing, also fix the kerneldoc for this function
while we're at it.
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
According the JEDEC specification an eMMC card supporting 1.8V vccq in DDR
mode should also be capable of 3.3V. However, it's been reported that some
mmc hosts supports 3.3V, but not 1.8V.
Currently the mmc core implements an error handling when the host fails to
set 1.8V for vccq, by falling back to 3.3V. Unfortunate, this seems to be
insufficient for some mmc hosts. To enable these to use eMMC DDR mode let's
invent a new mmc cap, MMC_CAP_3_3V_DDR, which tells whether they support
the eMMC 3.3V DDR mode.
In case MMC_CAP_3_3V_DDR is set, but not MMC_CAP_1_8V_DDR, let's change to
remain on the 3.3V, as it's the default voltage level for vccq, set by the
earlier power up sequence.
As this change introduces MMC_CAP_3_3V_DDR, let's take the opportunity to
do some re-formatting of the related defines in the header file.
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Reviewed-by: Shawn Lin <shawn.lin@rock-chips.com>
Tested-by: Jan Glauber <jglauber@cavium.com>
Tested-by: Stefan Wahren <stefan.wahren@i2se.com>
