linux/drivers/mmc/host/pxamci.c

824 lines
19 KiB
C
Raw Normal View History

/*
* linux/drivers/mmc/host/pxa.c - PXA MMCI driver
*
* Copyright (C) 2003 Russell King, All Rights Reserved.
*
* 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.
*
* This hardware is really sick:
* - No way to clear interrupts.
* - Have to turn off the clock whenever we touch the device.
* - Doesn't tell you how many data blocks were transferred.
* Yuck!
*
* 1 and 3 byte data transfers not supported
* max block length up to 1023
*/
#include <linux/module.h>
#include <linux/init.h>
#include <linux/ioport.h>
#include <linux/platform_device.h>
#include <linux/delay.h>
#include <linux/interrupt.h>
#include <linux/dmaengine.h>
#include <linux/dma-mapping.h>
#include <linux/clk.h>
#include <linux/err.h>
#include <linux/mmc/host.h>
#include <linux/mmc/slot-gpio.h>
#include <linux/io.h>
#include <linux/regulator/consumer.h>
#include <linux/gpio/consumer.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 16:04:11 +08:00
#include <linux/gfp.h>
#include <linux/of.h>
#include <linux/of_device.h>
#include <asm/sizes.h>
#include <mach/hardware.h>
ARM: pxa: move platform_data definitions Platform data for device drivers should be defined in include/linux/platform_data/*.h, not in the architecture and platform specific directories. This moves such data out of the pxa include directories Signed-off-by: Arnd Bergmann <arnd@arndb.de> Acked-by: Mark Brown <broonie@opensource.wolfsonmicro.com> Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Acked-by: Nicolas Pitre <nico@linaro.org> Acked-by: Mauro Carvalho Chehab <mchehab@redhat.com> Acked-by: Igor Grinberg <grinberg@compulab.co.il> Acked-by: Jeff Garzik <jgarzik@redhat.com> Acked-by: Marek Vasut <marex@denx.de> Acked-by: Robert Jarzmik <robert.jarzmik@free.fr> Acked-by: Paul Parsons <lost.distance@yahoo.com> Acked-by: Vinod Koul <vinod.koul@linux.intel.com> Acked-By: Stefan Schmidt <stefan@openezx.org> Cc: Eric Miao <eric.y.miao@gmail.com> Cc: Haojian Zhuang <haojian.zhuang@gmail.com> Cc: Daniel Ribeiro <drwyrm@gmail.com> Cc: Harald Welte <laforge@openezx.org> Cc: Philipp Zabel <philipp.zabel@gmail.com> Cc: Tomas Cech <sleep_walker@suse.cz> Cc: Sergey Lapin <slapin@ossfans.org> Cc: Jonathan Cameron <jic23@cam.ac.uk> Cc: Dan Williams <djbw@fb.com> Cc: Dmitry Torokhov <dmitry.torokhov@gmail.com> Cc: Chris Ball <cjb@laptop.org> Cc: David Woodhouse <dwmw2@infradead.org> Cc: Samuel Ortiz <samuel@sortiz.org> Cc: Alan Stern <stern@rowland.harvard.edu> Cc: Florian Tobias Schandinat <FlorianSchandinat@gmx.de> Cc: Liam Girdwood <lrg@ti.com> Cc: Jaroslav Kysela <perex@perex.cz> Cc: Takashi Iwai <tiwai@suse.de> Cc: Guennadi Liakhovetski <g.liakhovetski@gmx.de> Cc: Artem Bityutskiy <artem.bityutskiy@linux.intel.com> Cc: openezx-devel@lists.openezx.org
2012-08-24 21:16:48 +08:00
#include <linux/platform_data/mmc-pxamci.h>
#include "pxamci.h"
#define DRIVER_NAME "pxa2xx-mci"
#define NR_SG 1
#define CLKRT_OFF (~0)
#define mmc_has_26MHz() (cpu_is_pxa300() || cpu_is_pxa310() \
|| cpu_is_pxa935())
struct pxamci_host {
struct mmc_host *mmc;
spinlock_t lock;
struct resource *res;
void __iomem *base;
struct clk *clk;
unsigned long clkrate;
unsigned int clkrt;
unsigned int cmdat;
unsigned int imask;
unsigned int power_mode;
unsigned long detect_delay_ms;
bool use_ro_gpio;
struct gpio_desc *power;
struct pxamci_platform_data *pdata;
struct mmc_request *mrq;
struct mmc_command *cmd;
struct mmc_data *data;
struct dma_chan *dma_chan_rx;
struct dma_chan *dma_chan_tx;
dma_cookie_t dma_cookie;
unsigned int dma_len;
unsigned int dma_dir;
};
static int pxamci_init_ocr(struct pxamci_host *host)
{
struct mmc_host *mmc = host->mmc;
int ret;
ret = mmc_regulator_get_supply(mmc);
if (ret < 0)
return ret;
if (IS_ERR(mmc->supply.vmmc)) {
/* fall-back to platform data */
mmc->ocr_avail = host->pdata ?
host->pdata->ocr_mask :
MMC_VDD_32_33 | MMC_VDD_33_34;
}
return 0;
}
static inline int pxamci_set_power(struct pxamci_host *host,
unsigned char power_mode,
unsigned int vdd)
{
struct mmc_host *mmc = host->mmc;
struct regulator *supply = mmc->supply.vmmc;
if (!IS_ERR(supply))
return mmc_regulator_set_ocr(mmc, supply, vdd);
if (host->power) {
bool on = !!((1 << vdd) & host->pdata->ocr_mask);
gpiod_set_value(host->power, on);
}
if (host->pdata && host->pdata->setpower)
return host->pdata->setpower(mmc_dev(host->mmc), vdd);
return 0;
}
static void pxamci_stop_clock(struct pxamci_host *host)
{
if (readl(host->base + MMC_STAT) & STAT_CLK_EN) {
unsigned long timeout = 10000;
unsigned int v;
writel(STOP_CLOCK, host->base + MMC_STRPCL);
do {
v = readl(host->base + MMC_STAT);
if (!(v & STAT_CLK_EN))
break;
udelay(1);
} while (timeout--);
if (v & STAT_CLK_EN)
dev_err(mmc_dev(host->mmc), "unable to stop clock\n");
}
}
static void pxamci_enable_irq(struct pxamci_host *host, unsigned int mask)
{
unsigned long flags;
spin_lock_irqsave(&host->lock, flags);
host->imask &= ~mask;
writel(host->imask, host->base + MMC_I_MASK);
spin_unlock_irqrestore(&host->lock, flags);
}
static void pxamci_disable_irq(struct pxamci_host *host, unsigned int mask)
{
unsigned long flags;
spin_lock_irqsave(&host->lock, flags);
host->imask |= mask;
writel(host->imask, host->base + MMC_I_MASK);
spin_unlock_irqrestore(&host->lock, flags);
}
static void pxamci_dma_irq(void *param);
static void pxamci_setup_data(struct pxamci_host *host, struct mmc_data *data)
{
struct dma_async_tx_descriptor *tx;
enum dma_transfer_direction direction;
struct dma_slave_config config;
struct dma_chan *chan;
unsigned int nob = data->blocks;
unsigned long long clks;
unsigned int timeout;
int ret;
host->data = data;
writel(nob, host->base + MMC_NOB);
writel(data->blksz, host->base + MMC_BLKLEN);
clks = (unsigned long long)data->timeout_ns * host->clkrate;
do_div(clks, 1000000000UL);
timeout = (unsigned int)clks + (data->timeout_clks << host->clkrt);
writel((timeout + 255) / 256, host->base + MMC_RDTO);
memset(&config, 0, sizeof(config));
config.src_addr_width = DMA_SLAVE_BUSWIDTH_1_BYTE;
config.dst_addr_width = DMA_SLAVE_BUSWIDTH_1_BYTE;
config.src_addr = host->res->start + MMC_RXFIFO;
config.dst_addr = host->res->start + MMC_TXFIFO;
config.src_maxburst = 32;
config.dst_maxburst = 32;
if (data->flags & MMC_DATA_READ) {
host->dma_dir = DMA_FROM_DEVICE;
direction = DMA_DEV_TO_MEM;
chan = host->dma_chan_rx;
} else {
host->dma_dir = DMA_TO_DEVICE;
direction = DMA_MEM_TO_DEV;
chan = host->dma_chan_tx;
}
config.direction = direction;
ret = dmaengine_slave_config(chan, &config);
if (ret < 0) {
dev_err(mmc_dev(host->mmc), "dma slave config failed\n");
return;
}
host->dma_len = dma_map_sg(chan->device->dev, data->sg, data->sg_len,
host->dma_dir);
tx = dmaengine_prep_slave_sg(chan, data->sg, host->dma_len, direction,
DMA_PREP_INTERRUPT);
if (!tx) {
dev_err(mmc_dev(host->mmc), "prep_slave_sg() failed\n");
return;
}
if (!(data->flags & MMC_DATA_READ)) {
tx->callback = pxamci_dma_irq;
tx->callback_param = host;
}
host->dma_cookie = dmaengine_submit(tx);
/*
* workaround for erratum #91:
* only start DMA now if we are doing a read,
* otherwise we wait until CMD/RESP has finished
* before starting DMA.
*/
if (!cpu_is_pxa27x() || data->flags & MMC_DATA_READ)
dma_async_issue_pending(chan);
}
static void pxamci_start_cmd(struct pxamci_host *host, struct mmc_command *cmd, unsigned int cmdat)
{
WARN_ON(host->cmd != NULL);
host->cmd = cmd;
if (cmd->flags & MMC_RSP_BUSY)
cmdat |= CMDAT_BUSY;
#define RSP_TYPE(x) ((x) & ~(MMC_RSP_BUSY|MMC_RSP_OPCODE))
switch (RSP_TYPE(mmc_resp_type(cmd))) {
case RSP_TYPE(MMC_RSP_R1): /* r1, r1b, r6, r7 */
cmdat |= CMDAT_RESP_SHORT;
break;
case RSP_TYPE(MMC_RSP_R3):
cmdat |= CMDAT_RESP_R3;
break;
case RSP_TYPE(MMC_RSP_R2):
cmdat |= CMDAT_RESP_R2;
break;
default:
break;
}
writel(cmd->opcode, host->base + MMC_CMD);
writel(cmd->arg >> 16, host->base + MMC_ARGH);
writel(cmd->arg & 0xffff, host->base + MMC_ARGL);
writel(cmdat, host->base + MMC_CMDAT);
writel(host->clkrt, host->base + MMC_CLKRT);
writel(START_CLOCK, host->base + MMC_STRPCL);
pxamci_enable_irq(host, END_CMD_RES);
}
static void pxamci_finish_request(struct pxamci_host *host, struct mmc_request *mrq)
{
host->mrq = NULL;
host->cmd = NULL;
host->data = NULL;
mmc_request_done(host->mmc, mrq);
}
static int pxamci_cmd_done(struct pxamci_host *host, unsigned int stat)
{
struct mmc_command *cmd = host->cmd;
int i;
u32 v;
if (!cmd)
return 0;
host->cmd = NULL;
/*
* Did I mention this is Sick. We always need to
* discard the upper 8 bits of the first 16-bit word.
*/
v = readl(host->base + MMC_RES) & 0xffff;
for (i = 0; i < 4; i++) {
u32 w1 = readl(host->base + MMC_RES) & 0xffff;
u32 w2 = readl(host->base + MMC_RES) & 0xffff;
cmd->resp[i] = v << 24 | w1 << 8 | w2 >> 8;
v = w2;
}
if (stat & STAT_TIME_OUT_RESPONSE) {
cmd->error = -ETIMEDOUT;
} else if (stat & STAT_RES_CRC_ERR && cmd->flags & MMC_RSP_CRC) {
/*
* workaround for erratum #42:
* Intel PXA27x Family Processor Specification Update Rev 001
* A bogus CRC error can appear if the msb of a 136 bit
* response is a one.
*/
if (cpu_is_pxa27x() &&
(cmd->flags & MMC_RSP_136 && cmd->resp[0] & 0x80000000))
pr_debug("ignoring CRC from command %d - *risky*\n", cmd->opcode);
else
cmd->error = -EILSEQ;
}
pxamci_disable_irq(host, END_CMD_RES);
if (host->data && !cmd->error) {
pxamci_enable_irq(host, DATA_TRAN_DONE);
/*
* workaround for erratum #91, if doing write
* enable DMA late
*/
if (cpu_is_pxa27x() && host->data->flags & MMC_DATA_WRITE)
dma_async_issue_pending(host->dma_chan_tx);
} else {
pxamci_finish_request(host, host->mrq);
}
return 1;
}
static int pxamci_data_done(struct pxamci_host *host, unsigned int stat)
{
struct mmc_data *data = host->data;
struct dma_chan *chan;
if (!data)
return 0;
if (data->flags & MMC_DATA_READ)
chan = host->dma_chan_rx;
else
chan = host->dma_chan_tx;
dma_unmap_sg(chan->device->dev,
data->sg, data->sg_len, host->dma_dir);
if (stat & STAT_READ_TIME_OUT)
data->error = -ETIMEDOUT;
else if (stat & (STAT_CRC_READ_ERROR|STAT_CRC_WRITE_ERROR))
data->error = -EILSEQ;
/*
* There appears to be a hardware design bug here. There seems to
* be no way to find out how much data was transferred to the card.
* This means that if there was an error on any block, we mark all
* data blocks as being in error.
*/
if (!data->error)
data->bytes_xfered = data->blocks * data->blksz;
else
data->bytes_xfered = 0;
pxamci_disable_irq(host, DATA_TRAN_DONE);
host->data = NULL;
if (host->mrq->stop) {
pxamci_stop_clock(host);
pxamci_start_cmd(host, host->mrq->stop, host->cmdat);
} else {
pxamci_finish_request(host, host->mrq);
}
return 1;
}
IRQ: Maintain regs pointer globally rather than passing to IRQ handlers Maintain a per-CPU global "struct pt_regs *" variable which can be used instead of passing regs around manually through all ~1800 interrupt handlers in the Linux kernel. The regs pointer is used in few places, but it potentially costs both stack space and code to pass it around. On the FRV arch, removing the regs parameter from all the genirq function results in a 20% speed up of the IRQ exit path (ie: from leaving timer_interrupt() to leaving do_IRQ()). Where appropriate, an arch may override the generic storage facility and do something different with the variable. On FRV, for instance, the address is maintained in GR28 at all times inside the kernel as part of general exception handling. Having looked over the code, it appears that the parameter may be handed down through up to twenty or so layers of functions. Consider a USB character device attached to a USB hub, attached to a USB controller that posts its interrupts through a cascaded auxiliary interrupt controller. A character device driver may want to pass regs to the sysrq handler through the input layer which adds another few layers of parameter passing. I've build this code with allyesconfig for x86_64 and i386. I've runtested the main part of the code on FRV and i386, though I can't test most of the drivers. I've also done partial conversion for powerpc and MIPS - these at least compile with minimal configurations. This will affect all archs. Mostly the changes should be relatively easy. Take do_IRQ(), store the regs pointer at the beginning, saving the old one: struct pt_regs *old_regs = set_irq_regs(regs); And put the old one back at the end: set_irq_regs(old_regs); Don't pass regs through to generic_handle_irq() or __do_IRQ(). In timer_interrupt(), this sort of change will be necessary: - update_process_times(user_mode(regs)); - profile_tick(CPU_PROFILING, regs); + update_process_times(user_mode(get_irq_regs())); + profile_tick(CPU_PROFILING); I'd like to move update_process_times()'s use of get_irq_regs() into itself, except that i386, alone of the archs, uses something other than user_mode(). Some notes on the interrupt handling in the drivers: (*) input_dev() is now gone entirely. The regs pointer is no longer stored in the input_dev struct. (*) finish_unlinks() in drivers/usb/host/ohci-q.c needs checking. It does something different depending on whether it's been supplied with a regs pointer or not. (*) Various IRQ handler function pointers have been moved to type irq_handler_t. Signed-Off-By: David Howells <dhowells@redhat.com> (cherry picked from 1b16e7ac850969f38b375e511e3fa2f474a33867 commit)
2006-10-05 21:55:46 +08:00
static irqreturn_t pxamci_irq(int irq, void *devid)
{
struct pxamci_host *host = devid;
unsigned int ireg;
int handled = 0;
ireg = readl(host->base + MMC_I_REG) & ~readl(host->base + MMC_I_MASK);
if (ireg) {
unsigned stat = readl(host->base + MMC_STAT);
pr_debug("PXAMCI: irq %08x stat %08x\n", ireg, stat);
if (ireg & END_CMD_RES)
handled |= pxamci_cmd_done(host, stat);
if (ireg & DATA_TRAN_DONE)
handled |= pxamci_data_done(host, stat);
if (ireg & SDIO_INT) {
mmc_signal_sdio_irq(host->mmc);
handled = 1;
}
}
return IRQ_RETVAL(handled);
}
static void pxamci_request(struct mmc_host *mmc, struct mmc_request *mrq)
{
struct pxamci_host *host = mmc_priv(mmc);
unsigned int cmdat;
WARN_ON(host->mrq != NULL);
host->mrq = mrq;
pxamci_stop_clock(host);
cmdat = host->cmdat;
host->cmdat &= ~CMDAT_INIT;
if (mrq->data) {
pxamci_setup_data(host, mrq->data);
cmdat &= ~CMDAT_BUSY;
cmdat |= CMDAT_DATAEN | CMDAT_DMAEN;
if (mrq->data->flags & MMC_DATA_WRITE)
cmdat |= CMDAT_WRITE;
}
pxamci_start_cmd(host, mrq->cmd, cmdat);
}
static int pxamci_get_ro(struct mmc_host *mmc)
{
struct pxamci_host *host = mmc_priv(mmc);
if (host->use_ro_gpio)
return mmc_gpio_get_ro(mmc);
if (host->pdata && host->pdata->get_ro)
return !!host->pdata->get_ro(mmc_dev(mmc));
/*
* Board doesn't support read only detection; let the mmc core
* decide what to do.
*/
return -ENOSYS;
}
static void pxamci_set_ios(struct mmc_host *mmc, struct mmc_ios *ios)
{
struct pxamci_host *host = mmc_priv(mmc);
if (ios->clock) {
unsigned long rate = host->clkrate;
unsigned int clk = rate / ios->clock;
if (host->clkrt == CLKRT_OFF)
clk_prepare_enable(host->clk);
if (ios->clock == 26000000) {
/* to support 26MHz */
host->clkrt = 7;
} else {
/* to handle (19.5MHz, 26MHz) */
if (!clk)
clk = 1;
/*
* clk might result in a lower divisor than we
* desire. check for that condition and adjust
* as appropriate.
*/
if (rate / clk > ios->clock)
clk <<= 1;
host->clkrt = fls(clk) - 1;
}
/*
* we write clkrt on the next command
*/
} else {
pxamci_stop_clock(host);
if (host->clkrt != CLKRT_OFF) {
host->clkrt = CLKRT_OFF;
clk_disable_unprepare(host->clk);
}
}
if (host->power_mode != ios->power_mode) {
int ret;
host->power_mode = ios->power_mode;
ret = pxamci_set_power(host, ios->power_mode, ios->vdd);
if (ret) {
dev_err(mmc_dev(mmc), "unable to set power\n");
/*
* The .set_ios() function in the mmc_host_ops
* struct return void, and failing to set the
* power should be rare so we print an error and
* return here.
*/
return;
}
if (ios->power_mode == MMC_POWER_ON)
host->cmdat |= CMDAT_INIT;
}
if (ios->bus_width == MMC_BUS_WIDTH_4)
host->cmdat |= CMDAT_SD_4DAT;
else
host->cmdat &= ~CMDAT_SD_4DAT;
dev_dbg(mmc_dev(mmc), "PXAMCI: clkrt = %x cmdat = %x\n",
host->clkrt, host->cmdat);
}
static void pxamci_enable_sdio_irq(struct mmc_host *host, int enable)
{
struct pxamci_host *pxa_host = mmc_priv(host);
if (enable)
pxamci_enable_irq(pxa_host, SDIO_INT);
else
pxamci_disable_irq(pxa_host, SDIO_INT);
}
static const struct mmc_host_ops pxamci_ops = {
.request = pxamci_request,
.get_cd = mmc_gpio_get_cd,
.get_ro = pxamci_get_ro,
.set_ios = pxamci_set_ios,
.enable_sdio_irq = pxamci_enable_sdio_irq,
};
static void pxamci_dma_irq(void *param)
{
struct pxamci_host *host = param;
struct dma_tx_state state;
enum dma_status status;
struct dma_chan *chan;
unsigned long flags;
spin_lock_irqsave(&host->lock, flags);
if (!host->data)
goto out_unlock;
if (host->data->flags & MMC_DATA_READ)
chan = host->dma_chan_rx;
else
chan = host->dma_chan_tx;
status = dmaengine_tx_status(chan, host->dma_cookie, &state);
if (likely(status == DMA_COMPLETE)) {
writel(BUF_PART_FULL, host->base + MMC_PRTBUF);
} else {
pr_err("%s: DMA error on %s channel\n", mmc_hostname(host->mmc),
host->data->flags & MMC_DATA_READ ? "rx" : "tx");
host->data->error = -EIO;
pxamci_data_done(host, 0);
}
out_unlock:
spin_unlock_irqrestore(&host->lock, flags);
}
IRQ: Maintain regs pointer globally rather than passing to IRQ handlers Maintain a per-CPU global "struct pt_regs *" variable which can be used instead of passing regs around manually through all ~1800 interrupt handlers in the Linux kernel. The regs pointer is used in few places, but it potentially costs both stack space and code to pass it around. On the FRV arch, removing the regs parameter from all the genirq function results in a 20% speed up of the IRQ exit path (ie: from leaving timer_interrupt() to leaving do_IRQ()). Where appropriate, an arch may override the generic storage facility and do something different with the variable. On FRV, for instance, the address is maintained in GR28 at all times inside the kernel as part of general exception handling. Having looked over the code, it appears that the parameter may be handed down through up to twenty or so layers of functions. Consider a USB character device attached to a USB hub, attached to a USB controller that posts its interrupts through a cascaded auxiliary interrupt controller. A character device driver may want to pass regs to the sysrq handler through the input layer which adds another few layers of parameter passing. I've build this code with allyesconfig for x86_64 and i386. I've runtested the main part of the code on FRV and i386, though I can't test most of the drivers. I've also done partial conversion for powerpc and MIPS - these at least compile with minimal configurations. This will affect all archs. Mostly the changes should be relatively easy. Take do_IRQ(), store the regs pointer at the beginning, saving the old one: struct pt_regs *old_regs = set_irq_regs(regs); And put the old one back at the end: set_irq_regs(old_regs); Don't pass regs through to generic_handle_irq() or __do_IRQ(). In timer_interrupt(), this sort of change will be necessary: - update_process_times(user_mode(regs)); - profile_tick(CPU_PROFILING, regs); + update_process_times(user_mode(get_irq_regs())); + profile_tick(CPU_PROFILING); I'd like to move update_process_times()'s use of get_irq_regs() into itself, except that i386, alone of the archs, uses something other than user_mode(). Some notes on the interrupt handling in the drivers: (*) input_dev() is now gone entirely. The regs pointer is no longer stored in the input_dev struct. (*) finish_unlinks() in drivers/usb/host/ohci-q.c needs checking. It does something different depending on whether it's been supplied with a regs pointer or not. (*) Various IRQ handler function pointers have been moved to type irq_handler_t. Signed-Off-By: David Howells <dhowells@redhat.com> (cherry picked from 1b16e7ac850969f38b375e511e3fa2f474a33867 commit)
2006-10-05 21:55:46 +08:00
static irqreturn_t pxamci_detect_irq(int irq, void *devid)
{
struct pxamci_host *host = mmc_priv(devid);
mmc_detect_change(devid, msecs_to_jiffies(host->detect_delay_ms));
return IRQ_HANDLED;
}
#ifdef CONFIG_OF
static const struct of_device_id pxa_mmc_dt_ids[] = {
{ .compatible = "marvell,pxa-mmc" },
{ }
};
MODULE_DEVICE_TABLE(of, pxa_mmc_dt_ids);
static int pxamci_of_init(struct platform_device *pdev,
struct mmc_host *mmc)
{
struct device_node *np = pdev->dev.of_node;
struct pxamci_host *host = mmc_priv(mmc);
u32 tmp;
int ret;
if (!np)
return 0;
/* pxa-mmc specific */
if (of_property_read_u32(np, "pxa-mmc,detect-delay-ms", &tmp) == 0)
host->detect_delay_ms = tmp;
ret = mmc_of_parse(mmc);
if (ret < 0)
return ret;
return 0;
}
#else
static int pxamci_of_init(struct platform_device *pdev,
struct mmc_host *mmc)
{
return 0;
}
#endif
static int pxamci_probe(struct platform_device *pdev)
{
struct mmc_host *mmc;
struct pxamci_host *host = NULL;
struct device *dev = &pdev->dev;
struct resource *r;
int ret, irq;
r = platform_get_resource(pdev, IORESOURCE_MEM, 0);
irq = platform_get_irq(pdev, 0);
if (irq < 0)
return irq;
mmc = mmc_alloc_host(sizeof(struct pxamci_host), dev);
if (!mmc) {
ret = -ENOMEM;
goto out;
}
mmc->ops = &pxamci_ops;
/*
* We can do SG-DMA, but we don't because we never know how much
* data we successfully wrote to the card.
*/
mmc->max_segs = NR_SG;
/*
* Our hardware DMA can handle a maximum of one page per SG entry.
*/
mmc->max_seg_size = PAGE_SIZE;
/*
* Block length register is only 10 bits before PXA27x.
*/
mmc->max_blk_size = cpu_is_pxa25x() ? 1023 : 2048;
/*
* Block count register is 16 bits.
*/
mmc->max_blk_count = 65535;
ret = pxamci_of_init(pdev, mmc);
if (ret)
return ret;
host = mmc_priv(mmc);
host->mmc = mmc;
host->pdata = pdev->dev.platform_data;
host->clkrt = CLKRT_OFF;
host->clk = devm_clk_get(dev, NULL);
if (IS_ERR(host->clk)) {
ret = PTR_ERR(host->clk);
host->clk = NULL;
goto out;
}
host->clkrate = clk_get_rate(host->clk);
/*
* Calculate minimum clock rate, rounding up.
*/
mmc->f_min = (host->clkrate + 63) / 64;
mmc->f_max = (mmc_has_26MHz()) ? 26000000 : host->clkrate;
ret = pxamci_init_ocr(host);
if (ret < 0)
return ret;
mmc: Delete bounce buffer handling 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 a45c6cb81647 ("[ARM] 5369/1: omap mmc: Add new omap hsmmc controller for 2430 and 34xx, v3") This optimization was removed in commit 0ccd76d4c236 ("omap_hsmmc: Implement scatter-gather emulation") which found that scatter-gather emulation provided even better performance. The same was introduced for SDHCI in commit 2134a922c6e7 ("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>
2017-09-20 16:56:14 +08:00
mmc->caps = 0;
host->cmdat = 0;
if (!cpu_is_pxa25x()) {
mmc->caps |= MMC_CAP_4_BIT_DATA | MMC_CAP_SDIO_IRQ;
host->cmdat |= CMDAT_SDIO_INT_EN;
if (mmc_has_26MHz())
mmc->caps |= MMC_CAP_MMC_HIGHSPEED |
MMC_CAP_SD_HIGHSPEED;
}
spin_lock_init(&host->lock);
host->res = r;
host->imask = MMC_I_MASK_ALL;
host->base = devm_ioremap_resource(dev, r);
if (IS_ERR(host->base)) {
ret = PTR_ERR(host->base);
goto out;
}
/*
* Ensure that the host controller is shut down, and setup
* with our defaults.
*/
pxamci_stop_clock(host);
writel(0, host->base + MMC_SPI);
writel(64, host->base + MMC_RESTO);
writel(host->imask, host->base + MMC_I_MASK);
ret = devm_request_irq(dev, irq, pxamci_irq, 0,
DRIVER_NAME, host);
if (ret)
goto out;
platform_set_drvdata(pdev, mmc);
host->dma_chan_rx = dma_request_slave_channel(dev, "rx");
if (host->dma_chan_rx == NULL) {
dev_err(dev, "unable to request rx dma channel\n");
ret = -ENODEV;
goto out;
}
host->dma_chan_tx = dma_request_slave_channel(dev, "tx");
if (host->dma_chan_tx == NULL) {
dev_err(dev, "unable to request tx dma channel\n");
ret = -ENODEV;
goto out;
}
if (host->pdata) {
host->detect_delay_ms = host->pdata->detect_delay_ms;
host->power = devm_gpiod_get_optional(dev, "power", GPIOD_OUT_LOW);
if (IS_ERR(host->power)) {
dev_err(dev, "Failed requesting gpio_power\n");
goto out;
}
/* FIXME: should we pass detection delay to debounce? */
ret = mmc_gpiod_request_cd(mmc, "cd", 0, false, 0, NULL);
if (ret && ret != -ENOENT) {
dev_err(dev, "Failed requesting gpio_cd\n");
goto out;
}
ret = mmc_gpiod_request_ro(mmc, "wp", 0, 0, NULL);
if (ret && ret != -ENOENT) {
dev_err(dev, "Failed requesting gpio_ro\n");
goto out;
}
if (!ret) {
host->use_ro_gpio = true;
mmc->caps2 |= host->pdata->gpio_card_ro_invert ?
0 : MMC_CAP2_RO_ACTIVE_HIGH;
}
if (host->pdata->init)
host->pdata->init(dev, pxamci_detect_irq, mmc);
if (host->power && host->pdata->setpower)
dev_warn(dev, "gpio_power and setpower() both defined\n");
if (host->use_ro_gpio && host->pdata->get_ro)
dev_warn(dev, "gpio_ro and get_ro() both defined\n");
}
mmc_add_host(mmc);
return 0;
out:
if (host) {
if (host->dma_chan_rx)
dma_release_channel(host->dma_chan_rx);
if (host->dma_chan_tx)
dma_release_channel(host->dma_chan_tx);
}
if (mmc)
mmc_free_host(mmc);
return ret;
}
static int pxamci_remove(struct platform_device *pdev)
{
struct mmc_host *mmc = platform_get_drvdata(pdev);
if (mmc) {
struct pxamci_host *host = mmc_priv(mmc);
[ARM] pxamci: call mmc_remove_host() before freeing resources mmc_remove_host() will cause the mmc core to switch off the bus power by eventually calling pxamci_set_ios(). This function uses the regulator or the GPIO which have been freed already. This causes the following Oops on module unload. [ 49.519649] Unable to handle kernel paging request at virtual address 30303a70 [ 49.526878] pgd = c7084000 [ 49.529563] [30303a70] *pgd=00000000 [ 49.533136] Internal error: Oops: 5 [#1] [ 49.537025] last sysfs file: /sys/devices/platform/pxa27x-ohci/usb1/1-1/1-1:1.0/host0/target0:0:0/0:0:0:0/scsi_level [ 49.547471] Modules linked in: pxamci(-) eeti_ts [ 49.552061] CPU: 0 Not tainted (2.6.32-rc8 #322) [ 49.557001] PC is at regulator_is_enabled+0x3c/0xbc [ 49.561846] LR is at regulator_is_enabled+0x30/0xbc [ 49.566691] pc : [<c01a2448>] lr : [<c01a243c>] psr: 60000013 [ 49.566702] sp : c7083e70 ip : 30303a30 fp : 00000000 [ 49.578093] r10: c705e200 r9 : c7082000 r8 : c705e2e0 [ 49.583280] r7 : c7061340 r6 : c7061340 r5 : c7083e70 r4 : 00000000 [ 49.589759] r3 : c04dc434 r2 : c04dc434 r1 : c03eecea r0 : 00000047 [ 49.596241] Flags: nZCv IRQs on FIQs on Mode SVC_32 ISA ARM Segment user [ 49.603329] Control: 0000397f Table: a7084018 DAC: 00000015 [ 49.609031] Process rmmod (pid: 1101, stack limit = 0xc7082278) [ 49.614908] Stack: (0xc7083e70 to 0xc7084000) [ 49.619238] 3e60: c7082000 c703c4f8 c705ea00 c04f4074 [ 49.627366] 3e80: 00000000 c705e3a0 ffffffff c0247ddc c70361a0 00000000 c705e3a0 ffffffff [ 49.635499] 3ea0: c705e200 bf006400 c78c4f00 c705e200 c705e3a0 ffffffff c705e200 ffffffff [ 49.643633] 3ec0: c04d8ac8 c02476d0 ffffffff c0247c60 c705e200 c0248678 c705e200 c0249064 [ 49.651765] 3ee0: ffffffff bf006204 c04d8ad0 c04d8ad0 c04d8ac8 bf007490 00000880 c00440c4 [ 49.659898] 3f00: 0000b748 c01c5708 bf007490 c01c44c8 c04d8ac8 c04d8afc bf007490 c01c4570 [ 49.668031] 3f20: bf007490 bf00750c c04f4258 c01c37a4 00000000 bf00750c c7083f44 c007b014 [ 49.676162] 3f40: 4000d000 6d617870 08006963 00000001 00000000 c7085000 00000001 00000000 [ 49.684287] 3f60: 4000d000 c7083f8c 00000001 bea01a54 00005401 c7ab1400 c00440c4 00082000 [ 49.692420] 3f80: bf00750c 00000880 c7083f8c 00000000 4000cfa8 00000000 00000880 bea01cc8 [ 49.700552] 3fa0: 00000081 c0043f40 00000000 00000880 bea01cc8 00000880 00000006 00000000 [ 49.708677] 3fc0: 00000000 00000880 bea01cc8 00000081 00000097 0000cca4 0000b748 00000000 [ 49.716802] 3fe0: 4001a4f0 bea01cc0 00018bf4 4001a4fc 20000010 bea01cc8 a063e021 a063e421 [ 49.724958] [<c01a2448>] (regulator_is_enabled+0x3c/0xbc) from [<c0247ddc>] (mmc_regulator_set_ocr+0x14/0xd8) [ 49.734836] [<c0247ddc>] (mmc_regulator_set_ocr+0x14/0xd8) from [<bf006400>] (pxamci_set_ios+0xd8/0x17c [pxamci]) [ 49.745044] [<bf006400>] (pxamci_set_ios+0xd8/0x17c [pxamci]) from [<c02476d0>] (mmc_power_off+0x50/0x58) [ 49.754555] [<c02476d0>] (mmc_power_off+0x50/0x58) from [<c0247c60>] (mmc_detach_bus+0x68/0xc4) [ 49.763207] [<c0247c60>] (mmc_detach_bus+0x68/0xc4) from [<c0248678>] (mmc_stop_host+0xd4/0x1bc) [ 49.771944] [<c0248678>] (mmc_stop_host+0xd4/0x1bc) from [<c0249064>] (mmc_remove_host+0xc/0x20) [ 49.780681] [<c0249064>] (mmc_remove_host+0xc/0x20) from [<bf006204>] (pxamci_remove+0xc8/0x174 [pxamci]) [ 49.790211] [<bf006204>] (pxamci_remove+0xc8/0x174 [pxamci]) from [<c01c5708>] (platform_drv_remove+0x1c/0x24) [ 49.800164] [<c01c5708>] (platform_drv_remove+0x1c/0x24) from [<c01c44c8>] (__device_release_driver+0x7c/0xc4) [ 49.810110] [<c01c44c8>] (__device_release_driver+0x7c/0xc4) from [<c01c4570>] (driver_detach+0x60/0x8c) [ 49.819535] [<c01c4570>] (driver_detach+0x60/0x8c) from [<c01c37a4>] (bus_remove_driver+0x90/0xcc) [ 49.828452] [<c01c37a4>] (bus_remove_driver+0x90/0xcc) from [<c007b014>] (sys_delete_module+0x1d8/0x254) [ 49.837891] [<c007b014>] (sys_delete_module+0x1d8/0x254) from [<c0043f40>] (ret_fast_syscall+0x0/0x28) [ 49.847145] Code: eb06c53a e596c030 e1a0500d e59f106c (e59c0040) [ 49.853566] ---[ end trace b5fa66a00cea142f ]--- Signed-off-by: Daniel Mack <daniel@caiaq.de> Reported-by: Sven Neumann <s.neumann@raumfeld.com> Cc: Pierre Ossman <pierre@ossman.eu> Cc: linux-mmc@vger.kernel.org Cc: linux-arm-kernel@lists.infradead.org Cc: stable@kernel.org Signed-off-by: Eric Miao <eric.y.miao@gmail.com>
2009-12-02 01:17:18 +08:00
mmc_remove_host(mmc);
if (host->pdata && host->pdata->exit)
host->pdata->exit(&pdev->dev, mmc);
pxamci_stop_clock(host);
writel(TXFIFO_WR_REQ|RXFIFO_RD_REQ|CLK_IS_OFF|STOP_CMD|
END_CMD_RES|PRG_DONE|DATA_TRAN_DONE,
host->base + MMC_I_MASK);
dmaengine_terminate_all(host->dma_chan_rx);
dmaengine_terminate_all(host->dma_chan_tx);
dma_release_channel(host->dma_chan_rx);
dma_release_channel(host->dma_chan_tx);
mmc_free_host(mmc);
}
return 0;
}
static struct platform_driver pxamci_driver = {
.probe = pxamci_probe,
.remove = pxamci_remove,
.driver = {
.name = DRIVER_NAME,
.of_match_table = of_match_ptr(pxa_mmc_dt_ids),
},
};
module_platform_driver(pxamci_driver);
MODULE_DESCRIPTION("PXA Multimedia Card Interface Driver");
MODULE_LICENSE("GPL");
MODULE_ALIAS("platform:pxa2xx-mci");