linux/drivers/mmc/core/host.c

434 lines
11 KiB
C
Raw Normal View History

/*
* linux/drivers/mmc/core/host.c
*
* Copyright (C) 2003 Russell King, All Rights Reserved.
* Copyright (C) 2007-2008 Pierre Ossman
* Copyright (C) 2010 Linus Walleij
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License version 2 as
* published by the Free Software Foundation.
*
* MMC host class device management
*/
#include <linux/device.h>
#include <linux/err.h>
#include <linux/idr.h>
#include <linux/pagemap.h>
#include <linux/export.h>
#include <linux/leds.h>
include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h percpu.h is included by sched.h and module.h and thus ends up being included when building most .c files. percpu.h includes slab.h which in turn includes gfp.h making everything defined by the two files universally available and complicating inclusion dependencies. percpu.h -> slab.h dependency is about to be removed. Prepare for this change by updating users of gfp and slab facilities include those headers directly instead of assuming availability. As this conversion needs to touch large number of source files, the following script is used as the basis of conversion. http://userweb.kernel.org/~tj/misc/slabh-sweep.py The script does the followings. * Scan files for gfp and slab usages and update includes such that only the necessary includes are there. ie. if only gfp is used, gfp.h, if slab is used, slab.h. * When the script inserts a new include, it looks at the include blocks and try to put the new include such that its order conforms to its surrounding. It's put in the include block which contains core kernel includes, in the same order that the rest are ordered - alphabetical, Christmas tree, rev-Xmas-tree or at the end if there doesn't seem to be any matching order. * If the script can't find a place to put a new include (mostly because the file doesn't have fitting include block), it prints out an error message indicating which .h file needs to be added to the file. The conversion was done in the following steps. 1. The initial automatic conversion of all .c files updated slightly over 4000 files, deleting around 700 includes and adding ~480 gfp.h and ~3000 slab.h inclusions. The script emitted errors for ~400 files. 2. Each error was manually checked. Some didn't need the inclusion, some needed manual addition while adding it to implementation .h or embedding .c file was more appropriate for others. This step added inclusions to around 150 files. 3. The script was run again and the output was compared to the edits from #2 to make sure no file was left behind. 4. Several build tests were done and a couple of problems were fixed. e.g. lib/decompress_*.c used malloc/free() wrappers around slab APIs requiring slab.h to be added manually. 5. The script was run on all .h files but without automatically editing them as sprinkling gfp.h and slab.h inclusions around .h files could easily lead to inclusion dependency hell. Most gfp.h inclusion directives were ignored as stuff from gfp.h was usually wildly available and often used in preprocessor macros. Each slab.h inclusion directive was examined and added manually as necessary. 6. percpu.h was updated not to include slab.h. 7. Build test were done on the following configurations and failures were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my distributed build env didn't work with gcov compiles) and a few more options had to be turned off depending on archs to make things build (like ipr on powerpc/64 which failed due to missing writeq). * x86 and x86_64 UP and SMP allmodconfig and a custom test config. * powerpc and powerpc64 SMP allmodconfig * sparc and sparc64 SMP allmodconfig * ia64 SMP allmodconfig * s390 SMP allmodconfig * alpha SMP allmodconfig * um on x86_64 SMP allmodconfig 8. percpu.h modifications were reverted so that it could be applied as a separate patch and serve as bisection point. Given the fact that I had only a couple of failures from tests on step 6, I'm fairly confident about the coverage of this conversion patch. If there is a breakage, it's likely to be something in one of the arch headers which should be easily discoverable easily on most builds of the specific arch. Signed-off-by: Tejun Heo <tj@kernel.org> Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org> Cc: Ingo Molnar <mingo@redhat.com> Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
2010-03-24 08:04:11 +00:00
#include <linux/slab.h>
#include <linux/suspend.h>
#include <linux/mmc/host.h>
#include <linux/mmc/card.h>
#include "core.h"
#include "host.h"
#define cls_dev_to_mmc_host(d) container_of(d, struct mmc_host, class_dev)
static void mmc_host_classdev_release(struct device *dev)
{
struct mmc_host *host = cls_dev_to_mmc_host(dev);
kfree(host);
}
static struct class mmc_host_class = {
.name = "mmc_host",
.dev_release = mmc_host_classdev_release,
};
int mmc_register_host_class(void)
{
return class_register(&mmc_host_class);
}
void mmc_unregister_host_class(void)
{
class_unregister(&mmc_host_class);
}
static DEFINE_IDR(mmc_host_idr);
static DEFINE_SPINLOCK(mmc_host_lock);
#ifdef CONFIG_MMC_CLKGATE
mmc: core: Use delayed work in clock gating framework Current clock gating framework disables the MCI clock as soon as the request is completed and enables it when a request arrives. This aggressive clock gating framework, when enabled, cause following issues: When there are back-to-back requests from the Queue layer, we unnecessarily end up disabling and enabling the clocks between these requests since 8MCLK clock cycles is a very short duration compared to the time delay between back to back requests reaching the MMC layer. This overhead can effect the overall performance depending on how long the clock enable and disable calls take which is platform dependent. For example on some platforms we can have clock control not on the local processor, but on a different subsystem and the time taken to perform the clock enable/disable can add significant overhead. Also if the host controller driver decides to disable the host clock too when mmc_set_ios function is called with ios.clock=0, it adds additional delay and it is highly possible that the next request had already arrived and unnecessarily blocked in enabling the clocks. This is seen frequently when the processor is executing at high speeds and in multi-core platforms thus reduces the overall throughput compared to if clock gating is disabled. Fix this by delaying turning off the clocks by posting request on delayed workqueue. Also cancel the unscheduled pending work, if any, when there is access to card. sysfs entry is provided to tune the delay as needed, default value set to 200ms. Signed-off-by: Sujit Reddy Thumma <sthumma@codeaurora.org> Acked-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: Chris Ball <cjb@laptop.org>
2011-11-14 08:23:29 +00:00
static ssize_t clkgate_delay_show(struct device *dev,
struct device_attribute *attr, char *buf)
{
struct mmc_host *host = cls_dev_to_mmc_host(dev);
return snprintf(buf, PAGE_SIZE, "%lu\n", host->clkgate_delay);
mmc: core: Use delayed work in clock gating framework Current clock gating framework disables the MCI clock as soon as the request is completed and enables it when a request arrives. This aggressive clock gating framework, when enabled, cause following issues: When there are back-to-back requests from the Queue layer, we unnecessarily end up disabling and enabling the clocks between these requests since 8MCLK clock cycles is a very short duration compared to the time delay between back to back requests reaching the MMC layer. This overhead can effect the overall performance depending on how long the clock enable and disable calls take which is platform dependent. For example on some platforms we can have clock control not on the local processor, but on a different subsystem and the time taken to perform the clock enable/disable can add significant overhead. Also if the host controller driver decides to disable the host clock too when mmc_set_ios function is called with ios.clock=0, it adds additional delay and it is highly possible that the next request had already arrived and unnecessarily blocked in enabling the clocks. This is seen frequently when the processor is executing at high speeds and in multi-core platforms thus reduces the overall throughput compared to if clock gating is disabled. Fix this by delaying turning off the clocks by posting request on delayed workqueue. Also cancel the unscheduled pending work, if any, when there is access to card. sysfs entry is provided to tune the delay as needed, default value set to 200ms. Signed-off-by: Sujit Reddy Thumma <sthumma@codeaurora.org> Acked-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: Chris Ball <cjb@laptop.org>
2011-11-14 08:23:29 +00:00
}
static ssize_t clkgate_delay_store(struct device *dev,
struct device_attribute *attr, const char *buf, size_t count)
{
struct mmc_host *host = cls_dev_to_mmc_host(dev);
unsigned long flags, value;
if (kstrtoul(buf, 0, &value))
return -EINVAL;
spin_lock_irqsave(&host->clk_lock, flags);
host->clkgate_delay = value;
spin_unlock_irqrestore(&host->clk_lock, flags);
return count;
}
/*
* Enabling clock gating will make the core call out to the host
* once up and once down when it performs a request or card operation
* intermingled in any fashion. The driver will see this through
* set_ios() operations with ios.clock field set to 0 to gate (disable)
* the block clock, and to the old frequency to enable it again.
*/
static void mmc_host_clk_gate_delayed(struct mmc_host *host)
{
unsigned long tick_ns;
unsigned long freq = host->ios.clock;
unsigned long flags;
if (!freq) {
pr_debug("%s: frequency set to 0 in disable function, "
"this means the clock is already disabled.\n",
mmc_hostname(host));
return;
}
/*
* New requests may have appeared while we were scheduling,
* then there is no reason to delay the check before
* clk_disable().
*/
spin_lock_irqsave(&host->clk_lock, flags);
/*
* Delay n bus cycles (at least 8 from MMC spec) before attempting
* to disable the MCI block clock. The reference count may have
* gone up again after this delay due to rescheduling!
*/
if (!host->clk_requests) {
spin_unlock_irqrestore(&host->clk_lock, flags);
tick_ns = DIV_ROUND_UP(1000000000, freq);
ndelay(host->clk_delay * tick_ns);
} else {
/* New users appeared while waiting for this work */
spin_unlock_irqrestore(&host->clk_lock, flags);
return;
}
mutex_lock(&host->clk_gate_mutex);
spin_lock_irqsave(&host->clk_lock, flags);
if (!host->clk_requests) {
spin_unlock_irqrestore(&host->clk_lock, flags);
/* This will set host->ios.clock to 0 */
mmc_gate_clock(host);
spin_lock_irqsave(&host->clk_lock, flags);
pr_debug("%s: gated MCI clock\n", mmc_hostname(host));
}
spin_unlock_irqrestore(&host->clk_lock, flags);
mutex_unlock(&host->clk_gate_mutex);
}
/*
* Internal work. Work to disable the clock at some later point.
*/
static void mmc_host_clk_gate_work(struct work_struct *work)
{
struct mmc_host *host = container_of(work, struct mmc_host,
mmc: core: Use delayed work in clock gating framework Current clock gating framework disables the MCI clock as soon as the request is completed and enables it when a request arrives. This aggressive clock gating framework, when enabled, cause following issues: When there are back-to-back requests from the Queue layer, we unnecessarily end up disabling and enabling the clocks between these requests since 8MCLK clock cycles is a very short duration compared to the time delay between back to back requests reaching the MMC layer. This overhead can effect the overall performance depending on how long the clock enable and disable calls take which is platform dependent. For example on some platforms we can have clock control not on the local processor, but on a different subsystem and the time taken to perform the clock enable/disable can add significant overhead. Also if the host controller driver decides to disable the host clock too when mmc_set_ios function is called with ios.clock=0, it adds additional delay and it is highly possible that the next request had already arrived and unnecessarily blocked in enabling the clocks. This is seen frequently when the processor is executing at high speeds and in multi-core platforms thus reduces the overall throughput compared to if clock gating is disabled. Fix this by delaying turning off the clocks by posting request on delayed workqueue. Also cancel the unscheduled pending work, if any, when there is access to card. sysfs entry is provided to tune the delay as needed, default value set to 200ms. Signed-off-by: Sujit Reddy Thumma <sthumma@codeaurora.org> Acked-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: Chris Ball <cjb@laptop.org>
2011-11-14 08:23:29 +00:00
clk_gate_work.work);
mmc_host_clk_gate_delayed(host);
}
/**
* mmc_host_clk_hold - ungate hardware MCI clocks
* @host: host to ungate.
*
* Makes sure the host ios.clock is restored to a non-zero value
* past this call. Increase clock reference count and ungate clock
* if we're the first user.
*/
void mmc_host_clk_hold(struct mmc_host *host)
{
unsigned long flags;
mmc: core: Use delayed work in clock gating framework Current clock gating framework disables the MCI clock as soon as the request is completed and enables it when a request arrives. This aggressive clock gating framework, when enabled, cause following issues: When there are back-to-back requests from the Queue layer, we unnecessarily end up disabling and enabling the clocks between these requests since 8MCLK clock cycles is a very short duration compared to the time delay between back to back requests reaching the MMC layer. This overhead can effect the overall performance depending on how long the clock enable and disable calls take which is platform dependent. For example on some platforms we can have clock control not on the local processor, but on a different subsystem and the time taken to perform the clock enable/disable can add significant overhead. Also if the host controller driver decides to disable the host clock too when mmc_set_ios function is called with ios.clock=0, it adds additional delay and it is highly possible that the next request had already arrived and unnecessarily blocked in enabling the clocks. This is seen frequently when the processor is executing at high speeds and in multi-core platforms thus reduces the overall throughput compared to if clock gating is disabled. Fix this by delaying turning off the clocks by posting request on delayed workqueue. Also cancel the unscheduled pending work, if any, when there is access to card. sysfs entry is provided to tune the delay as needed, default value set to 200ms. Signed-off-by: Sujit Reddy Thumma <sthumma@codeaurora.org> Acked-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: Chris Ball <cjb@laptop.org>
2011-11-14 08:23:29 +00:00
/* cancel any clock gating work scheduled by mmc_host_clk_release() */
cancel_delayed_work_sync(&host->clk_gate_work);
mutex_lock(&host->clk_gate_mutex);
spin_lock_irqsave(&host->clk_lock, flags);
if (host->clk_gated) {
spin_unlock_irqrestore(&host->clk_lock, flags);
mmc_ungate_clock(host);
spin_lock_irqsave(&host->clk_lock, flags);
pr_debug("%s: ungated MCI clock\n", mmc_hostname(host));
}
host->clk_requests++;
spin_unlock_irqrestore(&host->clk_lock, flags);
mutex_unlock(&host->clk_gate_mutex);
}
/**
* mmc_host_may_gate_card - check if this card may be gated
* @card: card to check.
*/
static bool mmc_host_may_gate_card(struct mmc_card *card)
{
/* If there is no card we may gate it */
if (!card)
return true;
/*
* Don't gate SDIO cards! These need to be clocked at all times
* since they may be independent systems generating interrupts
* and other events. The clock requests counter from the core will
* go down to zero since the core does not need it, but we will not
* gate the clock, because there is somebody out there that may still
* be using it.
*/
return !(card->quirks & MMC_QUIRK_BROKEN_CLK_GATING);
}
/**
* mmc_host_clk_release - gate off hardware MCI clocks
* @host: host to gate.
*
* Calls the host driver with ios.clock set to zero as often as possible
* in order to gate off hardware MCI clocks. Decrease clock reference
* count and schedule disabling of clock.
*/
void mmc_host_clk_release(struct mmc_host *host)
{
unsigned long flags;
spin_lock_irqsave(&host->clk_lock, flags);
host->clk_requests--;
if (mmc_host_may_gate_card(host->card) &&
!host->clk_requests)
mmc: core: Use delayed work in clock gating framework Current clock gating framework disables the MCI clock as soon as the request is completed and enables it when a request arrives. This aggressive clock gating framework, when enabled, cause following issues: When there are back-to-back requests from the Queue layer, we unnecessarily end up disabling and enabling the clocks between these requests since 8MCLK clock cycles is a very short duration compared to the time delay between back to back requests reaching the MMC layer. This overhead can effect the overall performance depending on how long the clock enable and disable calls take which is platform dependent. For example on some platforms we can have clock control not on the local processor, but on a different subsystem and the time taken to perform the clock enable/disable can add significant overhead. Also if the host controller driver decides to disable the host clock too when mmc_set_ios function is called with ios.clock=0, it adds additional delay and it is highly possible that the next request had already arrived and unnecessarily blocked in enabling the clocks. This is seen frequently when the processor is executing at high speeds and in multi-core platforms thus reduces the overall throughput compared to if clock gating is disabled. Fix this by delaying turning off the clocks by posting request on delayed workqueue. Also cancel the unscheduled pending work, if any, when there is access to card. sysfs entry is provided to tune the delay as needed, default value set to 200ms. Signed-off-by: Sujit Reddy Thumma <sthumma@codeaurora.org> Acked-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: Chris Ball <cjb@laptop.org>
2011-11-14 08:23:29 +00:00
queue_delayed_work(system_nrt_wq, &host->clk_gate_work,
msecs_to_jiffies(host->clkgate_delay));
spin_unlock_irqrestore(&host->clk_lock, flags);
}
/**
* mmc_host_clk_rate - get current clock frequency setting
* @host: host to get the clock frequency for.
*
* Returns current clock frequency regardless of gating.
*/
unsigned int mmc_host_clk_rate(struct mmc_host *host)
{
unsigned long freq;
unsigned long flags;
spin_lock_irqsave(&host->clk_lock, flags);
if (host->clk_gated)
freq = host->clk_old;
else
freq = host->ios.clock;
spin_unlock_irqrestore(&host->clk_lock, flags);
return freq;
}
/**
* mmc_host_clk_init - set up clock gating code
* @host: host with potential clock to control
*/
static inline void mmc_host_clk_init(struct mmc_host *host)
{
host->clk_requests = 0;
/* Hold MCI clock for 8 cycles by default */
host->clk_delay = 8;
mmc: core: Use delayed work in clock gating framework Current clock gating framework disables the MCI clock as soon as the request is completed and enables it when a request arrives. This aggressive clock gating framework, when enabled, cause following issues: When there are back-to-back requests from the Queue layer, we unnecessarily end up disabling and enabling the clocks between these requests since 8MCLK clock cycles is a very short duration compared to the time delay between back to back requests reaching the MMC layer. This overhead can effect the overall performance depending on how long the clock enable and disable calls take which is platform dependent. For example on some platforms we can have clock control not on the local processor, but on a different subsystem and the time taken to perform the clock enable/disable can add significant overhead. Also if the host controller driver decides to disable the host clock too when mmc_set_ios function is called with ios.clock=0, it adds additional delay and it is highly possible that the next request had already arrived and unnecessarily blocked in enabling the clocks. This is seen frequently when the processor is executing at high speeds and in multi-core platforms thus reduces the overall throughput compared to if clock gating is disabled. Fix this by delaying turning off the clocks by posting request on delayed workqueue. Also cancel the unscheduled pending work, if any, when there is access to card. sysfs entry is provided to tune the delay as needed, default value set to 200ms. Signed-off-by: Sujit Reddy Thumma <sthumma@codeaurora.org> Acked-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: Chris Ball <cjb@laptop.org>
2011-11-14 08:23:29 +00:00
/*
* Default clock gating delay is 0ms to avoid wasting power.
mmc: core: Use delayed work in clock gating framework Current clock gating framework disables the MCI clock as soon as the request is completed and enables it when a request arrives. This aggressive clock gating framework, when enabled, cause following issues: When there are back-to-back requests from the Queue layer, we unnecessarily end up disabling and enabling the clocks between these requests since 8MCLK clock cycles is a very short duration compared to the time delay between back to back requests reaching the MMC layer. This overhead can effect the overall performance depending on how long the clock enable and disable calls take which is platform dependent. For example on some platforms we can have clock control not on the local processor, but on a different subsystem and the time taken to perform the clock enable/disable can add significant overhead. Also if the host controller driver decides to disable the host clock too when mmc_set_ios function is called with ios.clock=0, it adds additional delay and it is highly possible that the next request had already arrived and unnecessarily blocked in enabling the clocks. This is seen frequently when the processor is executing at high speeds and in multi-core platforms thus reduces the overall throughput compared to if clock gating is disabled. Fix this by delaying turning off the clocks by posting request on delayed workqueue. Also cancel the unscheduled pending work, if any, when there is access to card. sysfs entry is provided to tune the delay as needed, default value set to 200ms. Signed-off-by: Sujit Reddy Thumma <sthumma@codeaurora.org> Acked-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: Chris Ball <cjb@laptop.org>
2011-11-14 08:23:29 +00:00
* This value can be tuned by writing into sysfs entry.
*/
host->clkgate_delay = 0;
host->clk_gated = false;
mmc: core: Use delayed work in clock gating framework Current clock gating framework disables the MCI clock as soon as the request is completed and enables it when a request arrives. This aggressive clock gating framework, when enabled, cause following issues: When there are back-to-back requests from the Queue layer, we unnecessarily end up disabling and enabling the clocks between these requests since 8MCLK clock cycles is a very short duration compared to the time delay between back to back requests reaching the MMC layer. This overhead can effect the overall performance depending on how long the clock enable and disable calls take which is platform dependent. For example on some platforms we can have clock control not on the local processor, but on a different subsystem and the time taken to perform the clock enable/disable can add significant overhead. Also if the host controller driver decides to disable the host clock too when mmc_set_ios function is called with ios.clock=0, it adds additional delay and it is highly possible that the next request had already arrived and unnecessarily blocked in enabling the clocks. This is seen frequently when the processor is executing at high speeds and in multi-core platforms thus reduces the overall throughput compared to if clock gating is disabled. Fix this by delaying turning off the clocks by posting request on delayed workqueue. Also cancel the unscheduled pending work, if any, when there is access to card. sysfs entry is provided to tune the delay as needed, default value set to 200ms. Signed-off-by: Sujit Reddy Thumma <sthumma@codeaurora.org> Acked-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: Chris Ball <cjb@laptop.org>
2011-11-14 08:23:29 +00:00
INIT_DELAYED_WORK(&host->clk_gate_work, mmc_host_clk_gate_work);
spin_lock_init(&host->clk_lock);
mutex_init(&host->clk_gate_mutex);
}
/**
* mmc_host_clk_exit - shut down clock gating code
* @host: host with potential clock to control
*/
static inline void mmc_host_clk_exit(struct mmc_host *host)
{
/*
* Wait for any outstanding gate and then make sure we're
* ungated before exiting.
*/
mmc: core: Use delayed work in clock gating framework Current clock gating framework disables the MCI clock as soon as the request is completed and enables it when a request arrives. This aggressive clock gating framework, when enabled, cause following issues: When there are back-to-back requests from the Queue layer, we unnecessarily end up disabling and enabling the clocks between these requests since 8MCLK clock cycles is a very short duration compared to the time delay between back to back requests reaching the MMC layer. This overhead can effect the overall performance depending on how long the clock enable and disable calls take which is platform dependent. For example on some platforms we can have clock control not on the local processor, but on a different subsystem and the time taken to perform the clock enable/disable can add significant overhead. Also if the host controller driver decides to disable the host clock too when mmc_set_ios function is called with ios.clock=0, it adds additional delay and it is highly possible that the next request had already arrived and unnecessarily blocked in enabling the clocks. This is seen frequently when the processor is executing at high speeds and in multi-core platforms thus reduces the overall throughput compared to if clock gating is disabled. Fix this by delaying turning off the clocks by posting request on delayed workqueue. Also cancel the unscheduled pending work, if any, when there is access to card. sysfs entry is provided to tune the delay as needed, default value set to 200ms. Signed-off-by: Sujit Reddy Thumma <sthumma@codeaurora.org> Acked-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: Chris Ball <cjb@laptop.org>
2011-11-14 08:23:29 +00:00
if (cancel_delayed_work_sync(&host->clk_gate_work))
mmc_host_clk_gate_delayed(host);
if (host->clk_gated)
mmc_host_clk_hold(host);
/* There should be only one user now */
WARN_ON(host->clk_requests > 1);
}
mmc: core: Use delayed work in clock gating framework Current clock gating framework disables the MCI clock as soon as the request is completed and enables it when a request arrives. This aggressive clock gating framework, when enabled, cause following issues: When there are back-to-back requests from the Queue layer, we unnecessarily end up disabling and enabling the clocks between these requests since 8MCLK clock cycles is a very short duration compared to the time delay between back to back requests reaching the MMC layer. This overhead can effect the overall performance depending on how long the clock enable and disable calls take which is platform dependent. For example on some platforms we can have clock control not on the local processor, but on a different subsystem and the time taken to perform the clock enable/disable can add significant overhead. Also if the host controller driver decides to disable the host clock too when mmc_set_ios function is called with ios.clock=0, it adds additional delay and it is highly possible that the next request had already arrived and unnecessarily blocked in enabling the clocks. This is seen frequently when the processor is executing at high speeds and in multi-core platforms thus reduces the overall throughput compared to if clock gating is disabled. Fix this by delaying turning off the clocks by posting request on delayed workqueue. Also cancel the unscheduled pending work, if any, when there is access to card. sysfs entry is provided to tune the delay as needed, default value set to 200ms. Signed-off-by: Sujit Reddy Thumma <sthumma@codeaurora.org> Acked-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: Chris Ball <cjb@laptop.org>
2011-11-14 08:23:29 +00:00
static inline void mmc_host_clk_sysfs_init(struct mmc_host *host)
{
host->clkgate_delay_attr.show = clkgate_delay_show;
host->clkgate_delay_attr.store = clkgate_delay_store;
sysfs_attr_init(&host->clkgate_delay_attr.attr);
host->clkgate_delay_attr.attr.name = "clkgate_delay";
host->clkgate_delay_attr.attr.mode = S_IRUGO | S_IWUSR;
if (device_create_file(&host->class_dev, &host->clkgate_delay_attr))
pr_err("%s: Failed to create clkgate_delay sysfs entry\n",
mmc_hostname(host));
}
#else
static inline void mmc_host_clk_init(struct mmc_host *host)
{
}
static inline void mmc_host_clk_exit(struct mmc_host *host)
{
}
mmc: core: Use delayed work in clock gating framework Current clock gating framework disables the MCI clock as soon as the request is completed and enables it when a request arrives. This aggressive clock gating framework, when enabled, cause following issues: When there are back-to-back requests from the Queue layer, we unnecessarily end up disabling and enabling the clocks between these requests since 8MCLK clock cycles is a very short duration compared to the time delay between back to back requests reaching the MMC layer. This overhead can effect the overall performance depending on how long the clock enable and disable calls take which is platform dependent. For example on some platforms we can have clock control not on the local processor, but on a different subsystem and the time taken to perform the clock enable/disable can add significant overhead. Also if the host controller driver decides to disable the host clock too when mmc_set_ios function is called with ios.clock=0, it adds additional delay and it is highly possible that the next request had already arrived and unnecessarily blocked in enabling the clocks. This is seen frequently when the processor is executing at high speeds and in multi-core platforms thus reduces the overall throughput compared to if clock gating is disabled. Fix this by delaying turning off the clocks by posting request on delayed workqueue. Also cancel the unscheduled pending work, if any, when there is access to card. sysfs entry is provided to tune the delay as needed, default value set to 200ms. Signed-off-by: Sujit Reddy Thumma <sthumma@codeaurora.org> Acked-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: Chris Ball <cjb@laptop.org>
2011-11-14 08:23:29 +00:00
static inline void mmc_host_clk_sysfs_init(struct mmc_host *host)
{
}
#endif
/**
* mmc_alloc_host - initialise the per-host structure.
* @extra: sizeof private data structure
* @dev: pointer to host device model structure
*
* Initialise the per-host structure.
*/
struct mmc_host *mmc_alloc_host(int extra, struct device *dev)
{
int err;
struct mmc_host *host;
if (!idr_pre_get(&mmc_host_idr, GFP_KERNEL))
return NULL;
host = kzalloc(sizeof(struct mmc_host) + extra, GFP_KERNEL);
if (!host)
return NULL;
spin_lock(&mmc_host_lock);
err = idr_get_new(&mmc_host_idr, host, &host->index);
spin_unlock(&mmc_host_lock);
if (err)
goto free;
dev_set_name(&host->class_dev, "mmc%d", host->index);
host->parent = dev;
host->class_dev.parent = dev;
host->class_dev.class = &mmc_host_class;
device_initialize(&host->class_dev);
mmc_host_clk_init(host);
spin_lock_init(&host->lock);
init_waitqueue_head(&host->wq);
INIT_DELAYED_WORK(&host->detect, mmc_rescan);
INIT_DELAYED_WORK_DEFERRABLE(&host->disable, mmc_host_deeper_disable);
#ifdef CONFIG_PM
host->pm_notify.notifier_call = mmc_pm_notify;
#endif
/*
* By default, hosts do not support SGIO or large requests.
* They have to set these according to their abilities.
*/
host->max_segs = 1;
host->max_seg_size = PAGE_CACHE_SIZE;
host->max_req_size = PAGE_CACHE_SIZE;
host->max_blk_size = 512;
host->max_blk_count = PAGE_CACHE_SIZE / 512;
return host;
free:
kfree(host);
return NULL;
}
EXPORT_SYMBOL(mmc_alloc_host);
/**
* mmc_add_host - initialise host hardware
* @host: mmc host
*
* Register the host with the driver model. The host must be
* prepared to start servicing requests before this function
* completes.
*/
int mmc_add_host(struct mmc_host *host)
{
int err;
WARN_ON((host->caps & MMC_CAP_SDIO_IRQ) &&
!host->ops->enable_sdio_irq);
err = device_add(&host->class_dev);
if (err)
return err;
led_trigger_register_simple(dev_name(&host->class_dev), &host->led);
#ifdef CONFIG_DEBUG_FS
mmc_add_host_debugfs(host);
#endif
mmc: core: Use delayed work in clock gating framework Current clock gating framework disables the MCI clock as soon as the request is completed and enables it when a request arrives. This aggressive clock gating framework, when enabled, cause following issues: When there are back-to-back requests from the Queue layer, we unnecessarily end up disabling and enabling the clocks between these requests since 8MCLK clock cycles is a very short duration compared to the time delay between back to back requests reaching the MMC layer. This overhead can effect the overall performance depending on how long the clock enable and disable calls take which is platform dependent. For example on some platforms we can have clock control not on the local processor, but on a different subsystem and the time taken to perform the clock enable/disable can add significant overhead. Also if the host controller driver decides to disable the host clock too when mmc_set_ios function is called with ios.clock=0, it adds additional delay and it is highly possible that the next request had already arrived and unnecessarily blocked in enabling the clocks. This is seen frequently when the processor is executing at high speeds and in multi-core platforms thus reduces the overall throughput compared to if clock gating is disabled. Fix this by delaying turning off the clocks by posting request on delayed workqueue. Also cancel the unscheduled pending work, if any, when there is access to card. sysfs entry is provided to tune the delay as needed, default value set to 200ms. Signed-off-by: Sujit Reddy Thumma <sthumma@codeaurora.org> Acked-by: Linus Walleij <linus.walleij@linaro.org> Signed-off-by: Chris Ball <cjb@laptop.org>
2011-11-14 08:23:29 +00:00
mmc_host_clk_sysfs_init(host);
mmc_start_host(host);
register_pm_notifier(&host->pm_notify);
return 0;
}
EXPORT_SYMBOL(mmc_add_host);
/**
* mmc_remove_host - remove host hardware
* @host: mmc host
*
* Unregister and remove all cards associated with this host,
* and power down the MMC bus. No new requests will be issued
* after this function has returned.
*/
void mmc_remove_host(struct mmc_host *host)
{
unregister_pm_notifier(&host->pm_notify);
mmc_stop_host(host);
#ifdef CONFIG_DEBUG_FS
mmc_remove_host_debugfs(host);
#endif
device_del(&host->class_dev);
led_trigger_unregister_simple(host->led);
mmc_host_clk_exit(host);
}
EXPORT_SYMBOL(mmc_remove_host);
/**
* mmc_free_host - free the host structure
* @host: mmc host
*
* Free the host once all references to it have been dropped.
*/
void mmc_free_host(struct mmc_host *host)
{
spin_lock(&mmc_host_lock);
idr_remove(&mmc_host_idr, host->index);
spin_unlock(&mmc_host_lock);
put_device(&host->class_dev);
}
EXPORT_SYMBOL(mmc_free_host);