linux/arch/arm/plat-omap/dma.c

1471 lines
34 KiB
C
Raw Normal View History

/*
* linux/arch/arm/plat-omap/dma.c
*
* Copyright (C) 2003 - 2008 Nokia Corporation
* Author: Juha Yrjölä <juha.yrjola@nokia.com>
* DMA channel linking for 1610 by Samuel Ortiz <samuel.ortiz@nokia.com>
* Graphics DMA and LCD DMA graphics tranformations
* by Imre Deak <imre.deak@nokia.com>
* OMAP2/3 support Copyright (C) 2004-2007 Texas Instruments, Inc.
* Merged to support both OMAP1 and OMAP2 by Tony Lindgren <tony@atomide.com>
* Some functions based on earlier dma-omap.c Copyright (C) 2001 RidgeRun, Inc.
*
* Copyright (C) 2009 Texas Instruments
* Added OMAP4 support - Santosh Shilimkar <santosh.shilimkar@ti.com>
*
* Support functions for the OMAP internal DMA channels.
*
* Copyright (C) 2010 Texas Instruments Incorporated - http://www.ti.com/
* Converted DMA library into DMA platform driver.
* - G, Manjunath Kondaiah <manjugk@ti.com>
*
* 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.
*
*/
#include <linux/module.h>
#include <linux/init.h>
#include <linux/sched.h>
#include <linux/spinlock.h>
#include <linux/errno.h>
#include <linux/interrupt.h>
#include <linux/irq.h>
#include <linux/io.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/slab.h>
OMAP3: DMA: Errata i541: sDMA FIFO draining does not finish Implement the suggested workaround for OMAP3 regarding to sDMA draining issue, when the channel is disabled on the fly. This errata affects the following configuration: sDMA transfer is source synchronized Buffering is enabled SmartStandby is selected. The issue can be easily reproduced by creating overrun situation while recording audio. Either introduce load to the CPU: nice -19 arecord -D hw:0 -M -B 10000 -F 5000 -f dat > /dev/null & \ dd if=/dev/urandom of=/dev/null or suspending the arecord, and resuming it: arecord -D hw:0 -M -B 10000 -F 5000 -f dat > /dev/null CTRL+Z; fg; CTRL+Z; fg; ... In case of overrun audio stops DMA, and restarts it (without reseting the sDMA channel). When we hit this errata in stop case (sDMA drain did not complete), at the coming start the sDMA will not going to be operational (it is still draining). This leads to DMA stall condition. On OMAP3 we can recover with sDMA channel reset, it has been observed that by introducing unrelated sDMA activity might also help (reading from MMC for example). The same errata exists for OMAP2, where the suggestion is to disable the buffering to avoid this type of error. On OMAP3 the suggestion is to set sDMA to NoStandby before disabling the channel, and wait for the drain to finish, than configure sDMA to SmartStandby again. Signed-off-by: Peter Ujfalusi <peter.ujfalusi@nokia.com> Acked-by: Jarkko Nikula <jhnikula@gmail.com> Acked-by : Santosh Shilimkar <santosh.shilimkar@ti.com> Acked-by : Manjunath Kondaiah G <manjugk@ti.com> Signed-off-by: Tony Lindgren <tony@atomide.com>
2010-10-12 05:18:56 +08:00
#include <linux/delay.h>
ARM: OMAP: Move plat-omap/dma-omap.h to include/linux/omap-dma.h Based on earlier discussions[1] we attempted to find a suitable location for the omap DMA header in commit 2b6c4e73 (ARM: OMAP: DMA: Move plat/dma.h to plat-omap/dma-omap.h) until the conversion to dmaengine is complete. Unfortunately that was before I was able to try to test compile of the ARM multiplatform builds for omap2+, and the end result was not very good. So I'm creating yet another all over the place patch to cut the last dependency for building omap2+ for ARM multiplatform. After this, we have finally removed the driver dependencies to the arch/arm code, except for few drivers that are being worked on. The other option was to make the <plat-omap/dma-omap.h> path to work, but we'd have to add some new header directory to for multiplatform builds. Or we would have to manually include arch/arm/plat-omap/include again from arch/arm/Makefile for omap2+. Neither of these alternatives sound appealing as they will likely lead addition of various other headers exposed to the drivers, which we want to avoid for the multiplatform kernels. Since we already have a minimal include/linux/omap-dma.h, let's just use that instead and add a note to it to not use the custom omap DMA functions any longer where possible. Note that converting omap DMA to dmaengine depends on dmaengine supporting automatically incrementing the FIFO address at the device end, and converting all the remaining legacy drivers. So it's going to be few more merge windows. [1] https://patchwork.kernel.org/patch/1519591/# cc: Russell King <linux@arm.linux.org.uk> cc: Kevin Hilman <khilman@ti.com> cc: "Benoît Cousson" <b-cousson@ti.com> cc: Herbert Xu <herbert@gondor.apana.org.au> cc: "David S. Miller" <davem@davemloft.net> cc: Vinod Koul <vinod.koul@intel.com> cc: Dan Williams <djbw@fb.com> cc: Mauro Carvalho Chehab <mchehab@infradead.org> cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com> cc: Guennadi Liakhovetski <g.liakhovetski@gmx.de> cc: David Woodhouse <dwmw2@infradead.org> cc: Kyungmin Park <kyungmin.park@samsung.com> cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> cc: Tomi Valkeinen <tomi.valkeinen@ti.com> cc: Florian Tobias Schandinat <FlorianSchandinat@gmx.de> cc: Hans Verkuil <hans.verkuil@cisco.com> cc: Vaibhav Hiremath <hvaibhav@ti.com> cc: Lokesh Vutla <lokeshvutla@ti.com> cc: Rusty Russell <rusty@rustcorp.com.au> cc: Artem Bityutskiy <artem.bityutskiy@linux.intel.com> cc: Afzal Mohammed <afzal@ti.com> cc: linux-crypto@vger.kernel.org cc: linux-media@vger.kernel.org cc: linux-mtd@lists.infradead.org cc: linux-usb@vger.kernel.org cc: linux-fbdev@vger.kernel.org Acked-by: Felipe Balbi <balbi@ti.com> Signed-off-by: Tony Lindgren <tony@atomide.com>
2012-12-01 00:41:50 +08:00
#include <linux/omap-dma.h>
#ifdef CONFIG_ARCH_OMAP1
#include <mach/soc.h>
#endif
/*
* MAX_LOGICAL_DMA_CH_COUNT: the maximum number of logical DMA
* channels that an instance of the SDMA IP block can support. Used
* to size arrays. (The actual maximum on a particular SoC may be less
* than this -- for example, OMAP1 SDMA instances only support 17 logical
* DMA channels.)
*/
#define MAX_LOGICAL_DMA_CH_COUNT 32
#undef DEBUG
#ifndef CONFIG_ARCH_OMAP1
enum { DMA_CH_ALLOC_DONE, DMA_CH_PARAMS_SET_DONE, DMA_CH_STARTED,
DMA_CH_QUEUED, DMA_CH_NOTSTARTED, DMA_CH_PAUSED, DMA_CH_LINK_ENABLED
};
enum { DMA_CHAIN_STARTED, DMA_CHAIN_NOTSTARTED };
#endif
#define OMAP_DMA_ACTIVE 0x01
#define OMAP2_DMA_CSR_CLEAR_MASK 0xffffffff
#define OMAP_FUNC_MUX_ARM_BASE (0xfffe1000 + 0xec)
static struct omap_system_dma_plat_info *p;
static struct omap_dma_dev_attr *d;
static void omap_clear_dma(int lch);
static int omap_dma_set_prio_lch(int lch, unsigned char read_prio,
unsigned char write_prio);
static int enable_1510_mode;
static u32 errata;
static struct omap_dma_global_context_registers {
u32 dma_irqenable_l0;
u32 dma_irqenable_l1;
u32 dma_ocp_sysconfig;
u32 dma_gcr;
} omap_dma_global_context;
struct dma_link_info {
int *linked_dmach_q;
int no_of_lchs_linked;
int q_count;
int q_tail;
int q_head;
int chain_state;
int chain_mode;
};
static struct dma_link_info *dma_linked_lch;
#ifndef CONFIG_ARCH_OMAP1
/* Chain handling macros */
#define OMAP_DMA_CHAIN_QINIT(chain_id) \
do { \
dma_linked_lch[chain_id].q_head = \
dma_linked_lch[chain_id].q_tail = \
dma_linked_lch[chain_id].q_count = 0; \
} while (0)
#define OMAP_DMA_CHAIN_QFULL(chain_id) \
(dma_linked_lch[chain_id].no_of_lchs_linked == \
dma_linked_lch[chain_id].q_count)
#define OMAP_DMA_CHAIN_QLAST(chain_id) \
do { \
((dma_linked_lch[chain_id].no_of_lchs_linked-1) == \
dma_linked_lch[chain_id].q_count) \
} while (0)
#define OMAP_DMA_CHAIN_QEMPTY(chain_id) \
(0 == dma_linked_lch[chain_id].q_count)
#define __OMAP_DMA_CHAIN_INCQ(end) \
((end) = ((end)+1) % dma_linked_lch[chain_id].no_of_lchs_linked)
#define OMAP_DMA_CHAIN_INCQHEAD(chain_id) \
do { \
__OMAP_DMA_CHAIN_INCQ(dma_linked_lch[chain_id].q_head); \
dma_linked_lch[chain_id].q_count--; \
} while (0)
#define OMAP_DMA_CHAIN_INCQTAIL(chain_id) \
do { \
__OMAP_DMA_CHAIN_INCQ(dma_linked_lch[chain_id].q_tail); \
dma_linked_lch[chain_id].q_count++; \
} while (0)
#endif
static int dma_lch_count;
static int dma_chan_count;
static int omap_dma_reserve_channels;
static spinlock_t dma_chan_lock;
static struct omap_dma_lch *dma_chan;
static inline void disable_lnk(int lch);
static void omap_disable_channel_irq(int lch);
static inline void omap_enable_channel_irq(int lch);
#define REVISIT_24XX() printk(KERN_ERR "FIXME: no %s on 24xx\n", \
__func__);
#ifdef CONFIG_ARCH_OMAP15XX
/* Returns 1 if the DMA module is in OMAP1510-compatible mode, 0 otherwise */
static int omap_dma_in_1510_mode(void)
{
return enable_1510_mode;
}
#else
#define omap_dma_in_1510_mode() 0
#endif
#ifdef CONFIG_ARCH_OMAP1
static inline void set_gdma_dev(int req, int dev)
{
u32 reg = OMAP_FUNC_MUX_ARM_BASE + ((req - 1) / 5) * 4;
int shift = ((req - 1) % 5) * 6;
u32 l;
l = omap_readl(reg);
l &= ~(0x3f << shift);
l |= (dev - 1) << shift;
omap_writel(l, reg);
}
#else
#define set_gdma_dev(req, dev) do {} while (0)
#define omap_readl(reg) 0
#define omap_writel(val, reg) do {} while (0)
#endif
#ifdef CONFIG_ARCH_OMAP1
void omap_set_dma_priority(int lch, int dst_port, int priority)
{
unsigned long reg;
u32 l;
if (dma_omap1()) {
switch (dst_port) {
case OMAP_DMA_PORT_OCP_T1: /* FFFECC00 */
reg = OMAP_TC_OCPT1_PRIOR;
break;
case OMAP_DMA_PORT_OCP_T2: /* FFFECCD0 */
reg = OMAP_TC_OCPT2_PRIOR;
break;
case OMAP_DMA_PORT_EMIFF: /* FFFECC08 */
reg = OMAP_TC_EMIFF_PRIOR;
break;
case OMAP_DMA_PORT_EMIFS: /* FFFECC04 */
reg = OMAP_TC_EMIFS_PRIOR;
break;
default:
BUG();
return;
}
l = omap_readl(reg);
l &= ~(0xf << 8);
l |= (priority & 0xf) << 8;
omap_writel(l, reg);
}
}
#endif
#ifdef CONFIG_ARCH_OMAP2PLUS
void omap_set_dma_priority(int lch, int dst_port, int priority)
{
u32 ccr;
ccr = p->dma_read(CCR, lch);
if (priority)
ccr |= (1 << 6);
else
ccr &= ~(1 << 6);
p->dma_write(ccr, CCR, lch);
}
#endif
EXPORT_SYMBOL(omap_set_dma_priority);
void omap_set_dma_transfer_params(int lch, int data_type, int elem_count,
int frame_count, int sync_mode,
int dma_trigger, int src_or_dst_synch)
{
u32 l;
l = p->dma_read(CSDP, lch);
l &= ~0x03;
l |= data_type;
p->dma_write(l, CSDP, lch);
if (dma_omap1()) {
u16 ccr;
ccr = p->dma_read(CCR, lch);
ccr &= ~(1 << 5);
if (sync_mode == OMAP_DMA_SYNC_FRAME)
ccr |= 1 << 5;
p->dma_write(ccr, CCR, lch);
ccr = p->dma_read(CCR2, lch);
ccr &= ~(1 << 2);
if (sync_mode == OMAP_DMA_SYNC_BLOCK)
ccr |= 1 << 2;
p->dma_write(ccr, CCR2, lch);
}
if (dma_omap2plus() && dma_trigger) {
u32 val;
val = p->dma_read(CCR, lch);
/* DMA_SYNCHRO_CONTROL_UPPER depends on the channel number */
val &= ~((1 << 23) | (3 << 19) | 0x1f);
val |= (dma_trigger & ~0x1f) << 14;
val |= dma_trigger & 0x1f;
if (sync_mode & OMAP_DMA_SYNC_FRAME)
val |= 1 << 5;
else
val &= ~(1 << 5);
if (sync_mode & OMAP_DMA_SYNC_BLOCK)
val |= 1 << 18;
else
val &= ~(1 << 18);
if (src_or_dst_synch == OMAP_DMA_DST_SYNC_PREFETCH) {
val &= ~(1 << 24); /* dest synch */
val |= (1 << 23); /* Prefetch */
} else if (src_or_dst_synch) {
val |= 1 << 24; /* source synch */
} else {
val &= ~(1 << 24); /* dest synch */
}
p->dma_write(val, CCR, lch);
}
p->dma_write(elem_count, CEN, lch);
p->dma_write(frame_count, CFN, lch);
}
EXPORT_SYMBOL(omap_set_dma_transfer_params);
void omap_set_dma_write_mode(int lch, enum omap_dma_write_mode mode)
{
if (dma_omap2plus()) {
u32 csdp;
csdp = p->dma_read(CSDP, lch);
csdp &= ~(0x3 << 16);
csdp |= (mode << 16);
p->dma_write(csdp, CSDP, lch);
}
}
EXPORT_SYMBOL(omap_set_dma_write_mode);
void omap_set_dma_channel_mode(int lch, enum omap_dma_channel_mode mode)
{
if (dma_omap1() && !dma_omap15xx()) {
u32 l;
l = p->dma_read(LCH_CTRL, lch);
l &= ~0x7;
l |= mode;
p->dma_write(l, LCH_CTRL, lch);
}
}
EXPORT_SYMBOL(omap_set_dma_channel_mode);
/* Note that src_port is only for omap1 */
void omap_set_dma_src_params(int lch, int src_port, int src_amode,
unsigned long src_start,
int src_ei, int src_fi)
{
u32 l;
if (dma_omap1()) {
u16 w;
w = p->dma_read(CSDP, lch);
w &= ~(0x1f << 2);
w |= src_port << 2;
p->dma_write(w, CSDP, lch);
}
l = p->dma_read(CCR, lch);
l &= ~(0x03 << 12);
l |= src_amode << 12;
p->dma_write(l, CCR, lch);
p->dma_write(src_start, CSSA, lch);
p->dma_write(src_ei, CSEI, lch);
p->dma_write(src_fi, CSFI, lch);
}
EXPORT_SYMBOL(omap_set_dma_src_params);
void omap_set_dma_params(int lch, struct omap_dma_channel_params *params)
{
omap_set_dma_transfer_params(lch, params->data_type,
params->elem_count, params->frame_count,
params->sync_mode, params->trigger,
params->src_or_dst_synch);
omap_set_dma_src_params(lch, params->src_port,
params->src_amode, params->src_start,
params->src_ei, params->src_fi);
omap_set_dma_dest_params(lch, params->dst_port,
params->dst_amode, params->dst_start,
params->dst_ei, params->dst_fi);
if (params->read_prio || params->write_prio)
omap_dma_set_prio_lch(lch, params->read_prio,
params->write_prio);
}
EXPORT_SYMBOL(omap_set_dma_params);
void omap_set_dma_src_data_pack(int lch, int enable)
{
u32 l;
l = p->dma_read(CSDP, lch);
l &= ~(1 << 6);
if (enable)
l |= (1 << 6);
p->dma_write(l, CSDP, lch);
}
EXPORT_SYMBOL(omap_set_dma_src_data_pack);
void omap_set_dma_src_burst_mode(int lch, enum omap_dma_burst_mode burst_mode)
{
unsigned int burst = 0;
u32 l;
l = p->dma_read(CSDP, lch);
l &= ~(0x03 << 7);
switch (burst_mode) {
case OMAP_DMA_DATA_BURST_DIS:
break;
case OMAP_DMA_DATA_BURST_4:
if (dma_omap2plus())
burst = 0x1;
else
burst = 0x2;
break;
case OMAP_DMA_DATA_BURST_8:
if (dma_omap2plus()) {
burst = 0x2;
break;
}
/*
* not supported by current hardware on OMAP1
* w |= (0x03 << 7);
* fall through
*/
case OMAP_DMA_DATA_BURST_16:
if (dma_omap2plus()) {
burst = 0x3;
break;
}
/*
* OMAP1 don't support burst 16
* fall through
*/
default:
BUG();
}
l |= (burst << 7);
p->dma_write(l, CSDP, lch);
}
EXPORT_SYMBOL(omap_set_dma_src_burst_mode);
/* Note that dest_port is only for OMAP1 */
void omap_set_dma_dest_params(int lch, int dest_port, int dest_amode,
unsigned long dest_start,
int dst_ei, int dst_fi)
{
u32 l;
if (dma_omap1()) {
l = p->dma_read(CSDP, lch);
l &= ~(0x1f << 9);
l |= dest_port << 9;
p->dma_write(l, CSDP, lch);
}
l = p->dma_read(CCR, lch);
l &= ~(0x03 << 14);
l |= dest_amode << 14;
p->dma_write(l, CCR, lch);
p->dma_write(dest_start, CDSA, lch);
p->dma_write(dst_ei, CDEI, lch);
p->dma_write(dst_fi, CDFI, lch);
}
EXPORT_SYMBOL(omap_set_dma_dest_params);
void omap_set_dma_dest_data_pack(int lch, int enable)
{
u32 l;
l = p->dma_read(CSDP, lch);
l &= ~(1 << 13);
if (enable)
l |= 1 << 13;
p->dma_write(l, CSDP, lch);
}
EXPORT_SYMBOL(omap_set_dma_dest_data_pack);
void omap_set_dma_dest_burst_mode(int lch, enum omap_dma_burst_mode burst_mode)
{
unsigned int burst = 0;
u32 l;
l = p->dma_read(CSDP, lch);
l &= ~(0x03 << 14);
switch (burst_mode) {
case OMAP_DMA_DATA_BURST_DIS:
break;
case OMAP_DMA_DATA_BURST_4:
if (dma_omap2plus())
burst = 0x1;
else
burst = 0x2;
break;
case OMAP_DMA_DATA_BURST_8:
if (dma_omap2plus())
burst = 0x2;
else
burst = 0x3;
break;
case OMAP_DMA_DATA_BURST_16:
if (dma_omap2plus()) {
burst = 0x3;
break;
}
/*
* OMAP1 don't support burst 16
* fall through
*/
default:
printk(KERN_ERR "Invalid DMA burst mode\n");
BUG();
return;
}
l |= (burst << 14);
p->dma_write(l, CSDP, lch);
}
EXPORT_SYMBOL(omap_set_dma_dest_burst_mode);
static inline void omap_enable_channel_irq(int lch)
{
/* Clear CSR */
if (dma_omap1())
p->dma_read(CSR, lch);
else
p->dma_write(OMAP2_DMA_CSR_CLEAR_MASK, CSR, lch);
/* Enable some nice interrupts. */
p->dma_write(dma_chan[lch].enabled_irqs, CICR, lch);
}
static inline void omap_disable_channel_irq(int lch)
{
/* disable channel interrupts */
p->dma_write(0, CICR, lch);
/* Clear CSR */
if (dma_omap1())
p->dma_read(CSR, lch);
else
p->dma_write(OMAP2_DMA_CSR_CLEAR_MASK, CSR, lch);
}
void omap_enable_dma_irq(int lch, u16 bits)
{
dma_chan[lch].enabled_irqs |= bits;
}
EXPORT_SYMBOL(omap_enable_dma_irq);
void omap_disable_dma_irq(int lch, u16 bits)
{
dma_chan[lch].enabled_irqs &= ~bits;
}
EXPORT_SYMBOL(omap_disable_dma_irq);
static inline void enable_lnk(int lch)
{
u32 l;
l = p->dma_read(CLNK_CTRL, lch);
if (dma_omap1())
l &= ~(1 << 14);
/* Set the ENABLE_LNK bits */
if (dma_chan[lch].next_lch != -1)
l = dma_chan[lch].next_lch | (1 << 15);
#ifndef CONFIG_ARCH_OMAP1
if (dma_omap2plus())
if (dma_chan[lch].next_linked_ch != -1)
l = dma_chan[lch].next_linked_ch | (1 << 15);
#endif
p->dma_write(l, CLNK_CTRL, lch);
}
static inline void disable_lnk(int lch)
{
u32 l;
l = p->dma_read(CLNK_CTRL, lch);
/* Disable interrupts */
omap_disable_channel_irq(lch);
if (dma_omap1()) {
/* Set the STOP_LNK bit */
l |= 1 << 14;
}
if (dma_omap2plus()) {
/* Clear the ENABLE_LNK bit */
l &= ~(1 << 15);
}
p->dma_write(l, CLNK_CTRL, lch);
dma_chan[lch].flags &= ~OMAP_DMA_ACTIVE;
}
static inline void omap2_enable_irq_lch(int lch)
{
u32 val;
unsigned long flags;
if (dma_omap1())
return;
spin_lock_irqsave(&dma_chan_lock, flags);
/* clear IRQ STATUS */
p->dma_write(1 << lch, IRQSTATUS_L0, lch);
/* Enable interrupt */
val = p->dma_read(IRQENABLE_L0, lch);
val |= 1 << lch;
p->dma_write(val, IRQENABLE_L0, lch);
spin_unlock_irqrestore(&dma_chan_lock, flags);
}
static inline void omap2_disable_irq_lch(int lch)
{
u32 val;
unsigned long flags;
if (dma_omap1())
return;
spin_lock_irqsave(&dma_chan_lock, flags);
/* Disable interrupt */
val = p->dma_read(IRQENABLE_L0, lch);
val &= ~(1 << lch);
p->dma_write(val, IRQENABLE_L0, lch);
/* clear IRQ STATUS */
p->dma_write(1 << lch, IRQSTATUS_L0, lch);
spin_unlock_irqrestore(&dma_chan_lock, flags);
}
int omap_request_dma(int dev_id, const char *dev_name,
void (*callback)(int lch, u16 ch_status, void *data),
void *data, int *dma_ch_out)
{
int ch, free_ch = -1;
unsigned long flags;
struct omap_dma_lch *chan;
WARN(strcmp(dev_name, "DMA engine"), "Using deprecated platform DMA API - please update to DMA engine");
spin_lock_irqsave(&dma_chan_lock, flags);
for (ch = 0; ch < dma_chan_count; ch++) {
if (free_ch == -1 && dma_chan[ch].dev_id == -1) {
free_ch = ch;
/* Exit after first free channel found */
break;
}
}
if (free_ch == -1) {
spin_unlock_irqrestore(&dma_chan_lock, flags);
return -EBUSY;
}
chan = dma_chan + free_ch;
chan->dev_id = dev_id;
if (p->clear_lch_regs)
p->clear_lch_regs(free_ch);
if (dma_omap2plus())
omap_clear_dma(free_ch);
spin_unlock_irqrestore(&dma_chan_lock, flags);
chan->dev_name = dev_name;
chan->callback = callback;
chan->data = data;
chan->flags = 0;
#ifndef CONFIG_ARCH_OMAP1
if (dma_omap2plus()) {
chan->chain_id = -1;
chan->next_linked_ch = -1;
}
#endif
chan->enabled_irqs = OMAP_DMA_DROP_IRQ | OMAP_DMA_BLOCK_IRQ;
if (dma_omap1())
chan->enabled_irqs |= OMAP1_DMA_TOUT_IRQ;
else if (dma_omap2plus())
chan->enabled_irqs |= OMAP2_DMA_MISALIGNED_ERR_IRQ |
OMAP2_DMA_TRANS_ERR_IRQ;
if (dma_omap16xx()) {
/* If the sync device is set, configure it dynamically. */
if (dev_id != 0) {
set_gdma_dev(free_ch + 1, dev_id);
dev_id = free_ch + 1;
}
/*
* Disable the 1510 compatibility mode and set the sync device
* id.
*/
p->dma_write(dev_id | (1 << 10), CCR, free_ch);
} else if (dma_omap1()) {
p->dma_write(dev_id, CCR, free_ch);
}
if (dma_omap2plus()) {
omap_enable_channel_irq(free_ch);
omap2_enable_irq_lch(free_ch);
}
*dma_ch_out = free_ch;
return 0;
}
EXPORT_SYMBOL(omap_request_dma);
void omap_free_dma(int lch)
{
unsigned long flags;
if (dma_chan[lch].dev_id == -1) {
pr_err("omap_dma: trying to free unallocated DMA channel %d\n",
lch);
return;
}
/* Disable interrupt for logical channel */
if (dma_omap2plus())
omap2_disable_irq_lch(lch);
/* Disable all DMA interrupts for the channel. */
omap_disable_channel_irq(lch);
/* Make sure the DMA transfer is stopped. */
p->dma_write(0, CCR, lch);
/* Clear registers */
if (dma_omap2plus())
omap_clear_dma(lch);
spin_lock_irqsave(&dma_chan_lock, flags);
dma_chan[lch].dev_id = -1;
dma_chan[lch].next_lch = -1;
dma_chan[lch].callback = NULL;
spin_unlock_irqrestore(&dma_chan_lock, flags);
}
EXPORT_SYMBOL(omap_free_dma);
/**
* @brief omap_dma_set_global_params : Set global priority settings for dma
*
* @param arb_rate
* @param max_fifo_depth
* @param tparams - Number of threads to reserve : DMA_THREAD_RESERVE_NORM
* DMA_THREAD_RESERVE_ONET
* DMA_THREAD_RESERVE_TWOT
* DMA_THREAD_RESERVE_THREET
*/
void
omap_dma_set_global_params(int arb_rate, int max_fifo_depth, int tparams)
{
u32 reg;
if (dma_omap1()) {
printk(KERN_ERR "FIXME: no %s on 15xx/16xx\n", __func__);
return;
}
if (max_fifo_depth == 0)
max_fifo_depth = 1;
if (arb_rate == 0)
arb_rate = 1;
reg = 0xff & max_fifo_depth;
reg |= (0x3 & tparams) << 12;
reg |= (arb_rate & 0xff) << 16;
p->dma_write(reg, GCR, 0);
}
EXPORT_SYMBOL(omap_dma_set_global_params);
/**
* @brief omap_dma_set_prio_lch : Set channel wise priority settings
*
* @param lch
* @param read_prio - Read priority
* @param write_prio - Write priority
* Both of the above can be set with one of the following values :
* DMA_CH_PRIO_HIGH/DMA_CH_PRIO_LOW
*/
static int
omap_dma_set_prio_lch(int lch, unsigned char read_prio,
unsigned char write_prio)
{
u32 l;
if (unlikely((lch < 0 || lch >= dma_lch_count))) {
printk(KERN_ERR "Invalid channel id\n");
return -EINVAL;
}
l = p->dma_read(CCR, lch);
l &= ~((1 << 6) | (1 << 26));
if (d->dev_caps & IS_RW_PRIORITY)
l |= ((read_prio & 0x1) << 6) | ((write_prio & 0x1) << 26);
else
l |= ((read_prio & 0x1) << 6);
p->dma_write(l, CCR, lch);
return 0;
}
/*
* Clears any DMA state so the DMA engine is ready to restart with new buffers
* through omap_start_dma(). Any buffers in flight are discarded.
*/
static void omap_clear_dma(int lch)
{
unsigned long flags;
local_irq_save(flags);
p->clear_dma(lch);
local_irq_restore(flags);
}
void omap_start_dma(int lch)
{
u32 l;
/*
* The CPC/CDAC register needs to be initialized to zero
* before starting dma transfer.
*/
if (dma_omap15xx())
p->dma_write(0, CPC, lch);
else
p->dma_write(0, CDAC, lch);
if (!omap_dma_in_1510_mode() && dma_chan[lch].next_lch != -1) {
int next_lch, cur_lch;
char dma_chan_link_map[MAX_LOGICAL_DMA_CH_COUNT];
/* Set the link register of the first channel */
enable_lnk(lch);
memset(dma_chan_link_map, 0, sizeof(dma_chan_link_map));
dma_chan_link_map[lch] = 1;
cur_lch = dma_chan[lch].next_lch;
do {
next_lch = dma_chan[cur_lch].next_lch;
/* The loop case: we've been here already */
if (dma_chan_link_map[cur_lch])
break;
/* Mark the current channel */
dma_chan_link_map[cur_lch] = 1;
enable_lnk(cur_lch);
omap_enable_channel_irq(cur_lch);
cur_lch = next_lch;
} while (next_lch != -1);
} else if (IS_DMA_ERRATA(DMA_ERRATA_PARALLEL_CHANNELS))
p->dma_write(lch, CLNK_CTRL, lch);
omap_enable_channel_irq(lch);
l = p->dma_read(CCR, lch);
if (IS_DMA_ERRATA(DMA_ERRATA_IFRAME_BUFFERING))
l |= OMAP_DMA_CCR_BUFFERING_DISABLE;
l |= OMAP_DMA_CCR_EN;
/*
* As dma_write() uses IO accessors which are weakly ordered, there
* is no guarantee that data in coherent DMA memory will be visible
* to the DMA device. Add a memory barrier here to ensure that any
* such data is visible prior to enabling DMA.
*/
mb();
p->dma_write(l, CCR, lch);
dma_chan[lch].flags |= OMAP_DMA_ACTIVE;
}
EXPORT_SYMBOL(omap_start_dma);
void omap_stop_dma(int lch)
{
u32 l;
/* Disable all interrupts on the channel */
omap_disable_channel_irq(lch);
l = p->dma_read(CCR, lch);
if (IS_DMA_ERRATA(DMA_ERRATA_i541) &&
(l & OMAP_DMA_CCR_SEL_SRC_DST_SYNC)) {
OMAP3: DMA: Errata i541: sDMA FIFO draining does not finish Implement the suggested workaround for OMAP3 regarding to sDMA draining issue, when the channel is disabled on the fly. This errata affects the following configuration: sDMA transfer is source synchronized Buffering is enabled SmartStandby is selected. The issue can be easily reproduced by creating overrun situation while recording audio. Either introduce load to the CPU: nice -19 arecord -D hw:0 -M -B 10000 -F 5000 -f dat > /dev/null & \ dd if=/dev/urandom of=/dev/null or suspending the arecord, and resuming it: arecord -D hw:0 -M -B 10000 -F 5000 -f dat > /dev/null CTRL+Z; fg; CTRL+Z; fg; ... In case of overrun audio stops DMA, and restarts it (without reseting the sDMA channel). When we hit this errata in stop case (sDMA drain did not complete), at the coming start the sDMA will not going to be operational (it is still draining). This leads to DMA stall condition. On OMAP3 we can recover with sDMA channel reset, it has been observed that by introducing unrelated sDMA activity might also help (reading from MMC for example). The same errata exists for OMAP2, where the suggestion is to disable the buffering to avoid this type of error. On OMAP3 the suggestion is to set sDMA to NoStandby before disabling the channel, and wait for the drain to finish, than configure sDMA to SmartStandby again. Signed-off-by: Peter Ujfalusi <peter.ujfalusi@nokia.com> Acked-by: Jarkko Nikula <jhnikula@gmail.com> Acked-by : Santosh Shilimkar <santosh.shilimkar@ti.com> Acked-by : Manjunath Kondaiah G <manjugk@ti.com> Signed-off-by: Tony Lindgren <tony@atomide.com>
2010-10-12 05:18:56 +08:00
int i = 0;
u32 sys_cf;
/* Configure No-Standby */
l = p->dma_read(OCP_SYSCONFIG, lch);
OMAP3: DMA: Errata i541: sDMA FIFO draining does not finish Implement the suggested workaround for OMAP3 regarding to sDMA draining issue, when the channel is disabled on the fly. This errata affects the following configuration: sDMA transfer is source synchronized Buffering is enabled SmartStandby is selected. The issue can be easily reproduced by creating overrun situation while recording audio. Either introduce load to the CPU: nice -19 arecord -D hw:0 -M -B 10000 -F 5000 -f dat > /dev/null & \ dd if=/dev/urandom of=/dev/null or suspending the arecord, and resuming it: arecord -D hw:0 -M -B 10000 -F 5000 -f dat > /dev/null CTRL+Z; fg; CTRL+Z; fg; ... In case of overrun audio stops DMA, and restarts it (without reseting the sDMA channel). When we hit this errata in stop case (sDMA drain did not complete), at the coming start the sDMA will not going to be operational (it is still draining). This leads to DMA stall condition. On OMAP3 we can recover with sDMA channel reset, it has been observed that by introducing unrelated sDMA activity might also help (reading from MMC for example). The same errata exists for OMAP2, where the suggestion is to disable the buffering to avoid this type of error. On OMAP3 the suggestion is to set sDMA to NoStandby before disabling the channel, and wait for the drain to finish, than configure sDMA to SmartStandby again. Signed-off-by: Peter Ujfalusi <peter.ujfalusi@nokia.com> Acked-by: Jarkko Nikula <jhnikula@gmail.com> Acked-by : Santosh Shilimkar <santosh.shilimkar@ti.com> Acked-by : Manjunath Kondaiah G <manjugk@ti.com> Signed-off-by: Tony Lindgren <tony@atomide.com>
2010-10-12 05:18:56 +08:00
sys_cf = l;
l &= ~DMA_SYSCONFIG_MIDLEMODE_MASK;
l |= DMA_SYSCONFIG_MIDLEMODE(DMA_IDLEMODE_NO_IDLE);
p->dma_write(l , OCP_SYSCONFIG, 0);
OMAP3: DMA: Errata i541: sDMA FIFO draining does not finish Implement the suggested workaround for OMAP3 regarding to sDMA draining issue, when the channel is disabled on the fly. This errata affects the following configuration: sDMA transfer is source synchronized Buffering is enabled SmartStandby is selected. The issue can be easily reproduced by creating overrun situation while recording audio. Either introduce load to the CPU: nice -19 arecord -D hw:0 -M -B 10000 -F 5000 -f dat > /dev/null & \ dd if=/dev/urandom of=/dev/null or suspending the arecord, and resuming it: arecord -D hw:0 -M -B 10000 -F 5000 -f dat > /dev/null CTRL+Z; fg; CTRL+Z; fg; ... In case of overrun audio stops DMA, and restarts it (without reseting the sDMA channel). When we hit this errata in stop case (sDMA drain did not complete), at the coming start the sDMA will not going to be operational (it is still draining). This leads to DMA stall condition. On OMAP3 we can recover with sDMA channel reset, it has been observed that by introducing unrelated sDMA activity might also help (reading from MMC for example). The same errata exists for OMAP2, where the suggestion is to disable the buffering to avoid this type of error. On OMAP3 the suggestion is to set sDMA to NoStandby before disabling the channel, and wait for the drain to finish, than configure sDMA to SmartStandby again. Signed-off-by: Peter Ujfalusi <peter.ujfalusi@nokia.com> Acked-by: Jarkko Nikula <jhnikula@gmail.com> Acked-by : Santosh Shilimkar <santosh.shilimkar@ti.com> Acked-by : Manjunath Kondaiah G <manjugk@ti.com> Signed-off-by: Tony Lindgren <tony@atomide.com>
2010-10-12 05:18:56 +08:00
l = p->dma_read(CCR, lch);
OMAP3: DMA: Errata i541: sDMA FIFO draining does not finish Implement the suggested workaround for OMAP3 regarding to sDMA draining issue, when the channel is disabled on the fly. This errata affects the following configuration: sDMA transfer is source synchronized Buffering is enabled SmartStandby is selected. The issue can be easily reproduced by creating overrun situation while recording audio. Either introduce load to the CPU: nice -19 arecord -D hw:0 -M -B 10000 -F 5000 -f dat > /dev/null & \ dd if=/dev/urandom of=/dev/null or suspending the arecord, and resuming it: arecord -D hw:0 -M -B 10000 -F 5000 -f dat > /dev/null CTRL+Z; fg; CTRL+Z; fg; ... In case of overrun audio stops DMA, and restarts it (without reseting the sDMA channel). When we hit this errata in stop case (sDMA drain did not complete), at the coming start the sDMA will not going to be operational (it is still draining). This leads to DMA stall condition. On OMAP3 we can recover with sDMA channel reset, it has been observed that by introducing unrelated sDMA activity might also help (reading from MMC for example). The same errata exists for OMAP2, where the suggestion is to disable the buffering to avoid this type of error. On OMAP3 the suggestion is to set sDMA to NoStandby before disabling the channel, and wait for the drain to finish, than configure sDMA to SmartStandby again. Signed-off-by: Peter Ujfalusi <peter.ujfalusi@nokia.com> Acked-by: Jarkko Nikula <jhnikula@gmail.com> Acked-by : Santosh Shilimkar <santosh.shilimkar@ti.com> Acked-by : Manjunath Kondaiah G <manjugk@ti.com> Signed-off-by: Tony Lindgren <tony@atomide.com>
2010-10-12 05:18:56 +08:00
l &= ~OMAP_DMA_CCR_EN;
p->dma_write(l, CCR, lch);
OMAP3: DMA: Errata i541: sDMA FIFO draining does not finish Implement the suggested workaround for OMAP3 regarding to sDMA draining issue, when the channel is disabled on the fly. This errata affects the following configuration: sDMA transfer is source synchronized Buffering is enabled SmartStandby is selected. The issue can be easily reproduced by creating overrun situation while recording audio. Either introduce load to the CPU: nice -19 arecord -D hw:0 -M -B 10000 -F 5000 -f dat > /dev/null & \ dd if=/dev/urandom of=/dev/null or suspending the arecord, and resuming it: arecord -D hw:0 -M -B 10000 -F 5000 -f dat > /dev/null CTRL+Z; fg; CTRL+Z; fg; ... In case of overrun audio stops DMA, and restarts it (without reseting the sDMA channel). When we hit this errata in stop case (sDMA drain did not complete), at the coming start the sDMA will not going to be operational (it is still draining). This leads to DMA stall condition. On OMAP3 we can recover with sDMA channel reset, it has been observed that by introducing unrelated sDMA activity might also help (reading from MMC for example). The same errata exists for OMAP2, where the suggestion is to disable the buffering to avoid this type of error. On OMAP3 the suggestion is to set sDMA to NoStandby before disabling the channel, and wait for the drain to finish, than configure sDMA to SmartStandby again. Signed-off-by: Peter Ujfalusi <peter.ujfalusi@nokia.com> Acked-by: Jarkko Nikula <jhnikula@gmail.com> Acked-by : Santosh Shilimkar <santosh.shilimkar@ti.com> Acked-by : Manjunath Kondaiah G <manjugk@ti.com> Signed-off-by: Tony Lindgren <tony@atomide.com>
2010-10-12 05:18:56 +08:00
/* Wait for sDMA FIFO drain */
l = p->dma_read(CCR, lch);
OMAP3: DMA: Errata i541: sDMA FIFO draining does not finish Implement the suggested workaround for OMAP3 regarding to sDMA draining issue, when the channel is disabled on the fly. This errata affects the following configuration: sDMA transfer is source synchronized Buffering is enabled SmartStandby is selected. The issue can be easily reproduced by creating overrun situation while recording audio. Either introduce load to the CPU: nice -19 arecord -D hw:0 -M -B 10000 -F 5000 -f dat > /dev/null & \ dd if=/dev/urandom of=/dev/null or suspending the arecord, and resuming it: arecord -D hw:0 -M -B 10000 -F 5000 -f dat > /dev/null CTRL+Z; fg; CTRL+Z; fg; ... In case of overrun audio stops DMA, and restarts it (without reseting the sDMA channel). When we hit this errata in stop case (sDMA drain did not complete), at the coming start the sDMA will not going to be operational (it is still draining). This leads to DMA stall condition. On OMAP3 we can recover with sDMA channel reset, it has been observed that by introducing unrelated sDMA activity might also help (reading from MMC for example). The same errata exists for OMAP2, where the suggestion is to disable the buffering to avoid this type of error. On OMAP3 the suggestion is to set sDMA to NoStandby before disabling the channel, and wait for the drain to finish, than configure sDMA to SmartStandby again. Signed-off-by: Peter Ujfalusi <peter.ujfalusi@nokia.com> Acked-by: Jarkko Nikula <jhnikula@gmail.com> Acked-by : Santosh Shilimkar <santosh.shilimkar@ti.com> Acked-by : Manjunath Kondaiah G <manjugk@ti.com> Signed-off-by: Tony Lindgren <tony@atomide.com>
2010-10-12 05:18:56 +08:00
while (i < 100 && (l & (OMAP_DMA_CCR_RD_ACTIVE |
OMAP_DMA_CCR_WR_ACTIVE))) {
udelay(5);
i++;
l = p->dma_read(CCR, lch);
OMAP3: DMA: Errata i541: sDMA FIFO draining does not finish Implement the suggested workaround for OMAP3 regarding to sDMA draining issue, when the channel is disabled on the fly. This errata affects the following configuration: sDMA transfer is source synchronized Buffering is enabled SmartStandby is selected. The issue can be easily reproduced by creating overrun situation while recording audio. Either introduce load to the CPU: nice -19 arecord -D hw:0 -M -B 10000 -F 5000 -f dat > /dev/null & \ dd if=/dev/urandom of=/dev/null or suspending the arecord, and resuming it: arecord -D hw:0 -M -B 10000 -F 5000 -f dat > /dev/null CTRL+Z; fg; CTRL+Z; fg; ... In case of overrun audio stops DMA, and restarts it (without reseting the sDMA channel). When we hit this errata in stop case (sDMA drain did not complete), at the coming start the sDMA will not going to be operational (it is still draining). This leads to DMA stall condition. On OMAP3 we can recover with sDMA channel reset, it has been observed that by introducing unrelated sDMA activity might also help (reading from MMC for example). The same errata exists for OMAP2, where the suggestion is to disable the buffering to avoid this type of error. On OMAP3 the suggestion is to set sDMA to NoStandby before disabling the channel, and wait for the drain to finish, than configure sDMA to SmartStandby again. Signed-off-by: Peter Ujfalusi <peter.ujfalusi@nokia.com> Acked-by: Jarkko Nikula <jhnikula@gmail.com> Acked-by : Santosh Shilimkar <santosh.shilimkar@ti.com> Acked-by : Manjunath Kondaiah G <manjugk@ti.com> Signed-off-by: Tony Lindgren <tony@atomide.com>
2010-10-12 05:18:56 +08:00
}
if (i >= 100)
pr_err("DMA drain did not complete on lch %d\n", lch);
OMAP3: DMA: Errata i541: sDMA FIFO draining does not finish Implement the suggested workaround for OMAP3 regarding to sDMA draining issue, when the channel is disabled on the fly. This errata affects the following configuration: sDMA transfer is source synchronized Buffering is enabled SmartStandby is selected. The issue can be easily reproduced by creating overrun situation while recording audio. Either introduce load to the CPU: nice -19 arecord -D hw:0 -M -B 10000 -F 5000 -f dat > /dev/null & \ dd if=/dev/urandom of=/dev/null or suspending the arecord, and resuming it: arecord -D hw:0 -M -B 10000 -F 5000 -f dat > /dev/null CTRL+Z; fg; CTRL+Z; fg; ... In case of overrun audio stops DMA, and restarts it (without reseting the sDMA channel). When we hit this errata in stop case (sDMA drain did not complete), at the coming start the sDMA will not going to be operational (it is still draining). This leads to DMA stall condition. On OMAP3 we can recover with sDMA channel reset, it has been observed that by introducing unrelated sDMA activity might also help (reading from MMC for example). The same errata exists for OMAP2, where the suggestion is to disable the buffering to avoid this type of error. On OMAP3 the suggestion is to set sDMA to NoStandby before disabling the channel, and wait for the drain to finish, than configure sDMA to SmartStandby again. Signed-off-by: Peter Ujfalusi <peter.ujfalusi@nokia.com> Acked-by: Jarkko Nikula <jhnikula@gmail.com> Acked-by : Santosh Shilimkar <santosh.shilimkar@ti.com> Acked-by : Manjunath Kondaiah G <manjugk@ti.com> Signed-off-by: Tony Lindgren <tony@atomide.com>
2010-10-12 05:18:56 +08:00
/* Restore OCP_SYSCONFIG */
p->dma_write(sys_cf, OCP_SYSCONFIG, lch);
OMAP3: DMA: Errata i541: sDMA FIFO draining does not finish Implement the suggested workaround for OMAP3 regarding to sDMA draining issue, when the channel is disabled on the fly. This errata affects the following configuration: sDMA transfer is source synchronized Buffering is enabled SmartStandby is selected. The issue can be easily reproduced by creating overrun situation while recording audio. Either introduce load to the CPU: nice -19 arecord -D hw:0 -M -B 10000 -F 5000 -f dat > /dev/null & \ dd if=/dev/urandom of=/dev/null or suspending the arecord, and resuming it: arecord -D hw:0 -M -B 10000 -F 5000 -f dat > /dev/null CTRL+Z; fg; CTRL+Z; fg; ... In case of overrun audio stops DMA, and restarts it (without reseting the sDMA channel). When we hit this errata in stop case (sDMA drain did not complete), at the coming start the sDMA will not going to be operational (it is still draining). This leads to DMA stall condition. On OMAP3 we can recover with sDMA channel reset, it has been observed that by introducing unrelated sDMA activity might also help (reading from MMC for example). The same errata exists for OMAP2, where the suggestion is to disable the buffering to avoid this type of error. On OMAP3 the suggestion is to set sDMA to NoStandby before disabling the channel, and wait for the drain to finish, than configure sDMA to SmartStandby again. Signed-off-by: Peter Ujfalusi <peter.ujfalusi@nokia.com> Acked-by: Jarkko Nikula <jhnikula@gmail.com> Acked-by : Santosh Shilimkar <santosh.shilimkar@ti.com> Acked-by : Manjunath Kondaiah G <manjugk@ti.com> Signed-off-by: Tony Lindgren <tony@atomide.com>
2010-10-12 05:18:56 +08:00
} else {
l &= ~OMAP_DMA_CCR_EN;
p->dma_write(l, CCR, lch);
OMAP3: DMA: Errata i541: sDMA FIFO draining does not finish Implement the suggested workaround for OMAP3 regarding to sDMA draining issue, when the channel is disabled on the fly. This errata affects the following configuration: sDMA transfer is source synchronized Buffering is enabled SmartStandby is selected. The issue can be easily reproduced by creating overrun situation while recording audio. Either introduce load to the CPU: nice -19 arecord -D hw:0 -M -B 10000 -F 5000 -f dat > /dev/null & \ dd if=/dev/urandom of=/dev/null or suspending the arecord, and resuming it: arecord -D hw:0 -M -B 10000 -F 5000 -f dat > /dev/null CTRL+Z; fg; CTRL+Z; fg; ... In case of overrun audio stops DMA, and restarts it (without reseting the sDMA channel). When we hit this errata in stop case (sDMA drain did not complete), at the coming start the sDMA will not going to be operational (it is still draining). This leads to DMA stall condition. On OMAP3 we can recover with sDMA channel reset, it has been observed that by introducing unrelated sDMA activity might also help (reading from MMC for example). The same errata exists for OMAP2, where the suggestion is to disable the buffering to avoid this type of error. On OMAP3 the suggestion is to set sDMA to NoStandby before disabling the channel, and wait for the drain to finish, than configure sDMA to SmartStandby again. Signed-off-by: Peter Ujfalusi <peter.ujfalusi@nokia.com> Acked-by: Jarkko Nikula <jhnikula@gmail.com> Acked-by : Santosh Shilimkar <santosh.shilimkar@ti.com> Acked-by : Manjunath Kondaiah G <manjugk@ti.com> Signed-off-by: Tony Lindgren <tony@atomide.com>
2010-10-12 05:18:56 +08:00
}
/*
* Ensure that data transferred by DMA is visible to any access
* after DMA has been disabled. This is important for coherent
* DMA regions.
*/
mb();
if (!omap_dma_in_1510_mode() && dma_chan[lch].next_lch != -1) {
int next_lch, cur_lch = lch;
char dma_chan_link_map[MAX_LOGICAL_DMA_CH_COUNT];
memset(dma_chan_link_map, 0, sizeof(dma_chan_link_map));
do {
/* The loop case: we've been here already */
if (dma_chan_link_map[cur_lch])
break;
/* Mark the current channel */
dma_chan_link_map[cur_lch] = 1;
disable_lnk(cur_lch);
next_lch = dma_chan[cur_lch].next_lch;
cur_lch = next_lch;
} while (next_lch != -1);
}
dma_chan[lch].flags &= ~OMAP_DMA_ACTIVE;
}
EXPORT_SYMBOL(omap_stop_dma);
/*
* Allows changing the DMA callback function or data. This may be needed if
* the driver shares a single DMA channel for multiple dma triggers.
*/
int omap_set_dma_callback(int lch,
void (*callback)(int lch, u16 ch_status, void *data),
void *data)
{
unsigned long flags;
if (lch < 0)
return -ENODEV;
spin_lock_irqsave(&dma_chan_lock, flags);
if (dma_chan[lch].dev_id == -1) {
printk(KERN_ERR "DMA callback for not set for free channel\n");
spin_unlock_irqrestore(&dma_chan_lock, flags);
return -EINVAL;
}
dma_chan[lch].callback = callback;
dma_chan[lch].data = data;
spin_unlock_irqrestore(&dma_chan_lock, flags);
return 0;
}
EXPORT_SYMBOL(omap_set_dma_callback);
/*
* Returns current physical source address for the given DMA channel.
* If the channel is running the caller must disable interrupts prior calling
* this function and process the returned value before re-enabling interrupt to
* prevent races with the interrupt handler. Note that in continuous mode there
* is a chance for CSSA_L register overflow between the two reads resulting
* in incorrect return value.
*/
dma_addr_t omap_get_dma_src_pos(int lch)
{
dma_addr_t offset = 0;
if (dma_omap15xx())
offset = p->dma_read(CPC, lch);
else
offset = p->dma_read(CSAC, lch);
if (IS_DMA_ERRATA(DMA_ERRATA_3_3) && offset == 0)
offset = p->dma_read(CSAC, lch);
if (!dma_omap15xx()) {
/*
* CDAC == 0 indicates that the DMA transfer on the channel has
* not been started (no data has been transferred so far).
* Return the programmed source start address in this case.
*/
if (likely(p->dma_read(CDAC, lch)))
offset = p->dma_read(CSAC, lch);
else
offset = p->dma_read(CSSA, lch);
}
if (dma_omap1())
offset |= (p->dma_read(CSSA, lch) & 0xFFFF0000);
return offset;
}
EXPORT_SYMBOL(omap_get_dma_src_pos);
/*
* Returns current physical destination address for the given DMA channel.
* If the channel is running the caller must disable interrupts prior calling
* this function and process the returned value before re-enabling interrupt to
* prevent races with the interrupt handler. Note that in continuous mode there
* is a chance for CDSA_L register overflow between the two reads resulting
* in incorrect return value.
*/
dma_addr_t omap_get_dma_dst_pos(int lch)
{
dma_addr_t offset = 0;
if (dma_omap15xx())
offset = p->dma_read(CPC, lch);
else
offset = p->dma_read(CDAC, lch);
/*
* omap 3.2/3.3 erratum: sometimes 0 is returned if CSAC/CDAC is
* read before the DMA controller finished disabling the channel.
*/
if (!dma_omap15xx() && offset == 0) {
offset = p->dma_read(CDAC, lch);
/*
* CDAC == 0 indicates that the DMA transfer on the channel has
* not been started (no data has been transferred so far).
* Return the programmed destination start address in this case.
*/
if (unlikely(!offset))
offset = p->dma_read(CDSA, lch);
}
if (dma_omap1())
offset |= (p->dma_read(CDSA, lch) & 0xFFFF0000);
return offset;
}
EXPORT_SYMBOL(omap_get_dma_dst_pos);
int omap_get_dma_active_status(int lch)
{
return (p->dma_read(CCR, lch) & OMAP_DMA_CCR_EN) != 0;
}
EXPORT_SYMBOL(omap_get_dma_active_status);
int omap_dma_running(void)
{
int lch;
if (dma_omap1())
if (omap_lcd_dma_running())
return 1;
for (lch = 0; lch < dma_chan_count; lch++)
if (p->dma_read(CCR, lch) & OMAP_DMA_CCR_EN)
return 1;
return 0;
}
/*
* lch_queue DMA will start right after lch_head one is finished.
* For this DMA link to start, you still need to start (see omap_start_dma)
* the first one. That will fire up the entire queue.
*/
void omap_dma_link_lch(int lch_head, int lch_queue)
{
if (omap_dma_in_1510_mode()) {
if (lch_head == lch_queue) {
p->dma_write(p->dma_read(CCR, lch_head) | (3 << 8),
CCR, lch_head);
return;
}
printk(KERN_ERR "DMA linking is not supported in 1510 mode\n");
BUG();
return;
}
if ((dma_chan[lch_head].dev_id == -1) ||
(dma_chan[lch_queue].dev_id == -1)) {
pr_err("omap_dma: trying to link non requested channels\n");
dump_stack();
}
dma_chan[lch_head].next_lch = lch_queue;
}
EXPORT_SYMBOL(omap_dma_link_lch);
/*----------------------------------------------------------------------------*/
#ifdef CONFIG_ARCH_OMAP1
static int omap1_dma_handle_ch(int ch)
{
u32 csr;
if (enable_1510_mode && ch >= 6) {
csr = dma_chan[ch].saved_csr;
dma_chan[ch].saved_csr = 0;
} else
csr = p->dma_read(CSR, ch);
if (enable_1510_mode && ch <= 2 && (csr >> 7) != 0) {
dma_chan[ch + 6].saved_csr = csr >> 7;
csr &= 0x7f;
}
if ((csr & 0x3f) == 0)
return 0;
if (unlikely(dma_chan[ch].dev_id == -1)) {
pr_warn("Spurious interrupt from DMA channel %d (CSR %04x)\n",
ch, csr);
return 0;
}
if (unlikely(csr & OMAP1_DMA_TOUT_IRQ))
pr_warn("DMA timeout with device %d\n", dma_chan[ch].dev_id);
if (unlikely(csr & OMAP_DMA_DROP_IRQ))
pr_warn("DMA synchronization event drop occurred with device %d\n",
dma_chan[ch].dev_id);
if (likely(csr & OMAP_DMA_BLOCK_IRQ))
dma_chan[ch].flags &= ~OMAP_DMA_ACTIVE;
if (likely(dma_chan[ch].callback != NULL))
dma_chan[ch].callback(ch, csr, dma_chan[ch].data);
return 1;
}
static irqreturn_t omap1_dma_irq_handler(int irq, void *dev_id)
{
int ch = ((int) dev_id) - 1;
int handled = 0;
for (;;) {
int handled_now = 0;
handled_now += omap1_dma_handle_ch(ch);
if (enable_1510_mode && dma_chan[ch + 6].saved_csr)
handled_now += omap1_dma_handle_ch(ch + 6);
if (!handled_now)
break;
handled += handled_now;
}
return handled ? IRQ_HANDLED : IRQ_NONE;
}
#else
#define omap1_dma_irq_handler NULL
#endif
#ifdef CONFIG_ARCH_OMAP2PLUS
static int omap2_dma_handle_ch(int ch)
{
u32 status = p->dma_read(CSR, ch);
if (!status) {
if (printk_ratelimit())
pr_warn("Spurious DMA IRQ for lch %d\n", ch);
p->dma_write(1 << ch, IRQSTATUS_L0, ch);
return 0;
}
if (unlikely(dma_chan[ch].dev_id == -1)) {
if (printk_ratelimit())
pr_warn("IRQ %04x for non-allocated DMA channel %d\n",
status, ch);
return 0;
}
if (unlikely(status & OMAP_DMA_DROP_IRQ))
pr_info("DMA synchronization event drop occurred with device %d\n",
dma_chan[ch].dev_id);
if (unlikely(status & OMAP2_DMA_TRANS_ERR_IRQ)) {
printk(KERN_INFO "DMA transaction error with device %d\n",
dma_chan[ch].dev_id);
if (IS_DMA_ERRATA(DMA_ERRATA_i378)) {
u32 ccr;
ccr = p->dma_read(CCR, ch);
ccr &= ~OMAP_DMA_CCR_EN;
p->dma_write(ccr, CCR, ch);
dma_chan[ch].flags &= ~OMAP_DMA_ACTIVE;
}
}
if (unlikely(status & OMAP2_DMA_SECURE_ERR_IRQ))
printk(KERN_INFO "DMA secure error with device %d\n",
dma_chan[ch].dev_id);
if (unlikely(status & OMAP2_DMA_MISALIGNED_ERR_IRQ))
printk(KERN_INFO "DMA misaligned error with device %d\n",
dma_chan[ch].dev_id);
p->dma_write(status, CSR, ch);
p->dma_write(1 << ch, IRQSTATUS_L0, ch);
/* read back the register to flush the write */
p->dma_read(IRQSTATUS_L0, ch);
/* If the ch is not chained then chain_id will be -1 */
if (dma_chan[ch].chain_id != -1) {
int chain_id = dma_chan[ch].chain_id;
dma_chan[ch].state = DMA_CH_NOTSTARTED;
if (p->dma_read(CLNK_CTRL, ch) & (1 << 15))
dma_chan[dma_chan[ch].next_linked_ch].state =
DMA_CH_STARTED;
if (dma_linked_lch[chain_id].chain_mode ==
OMAP_DMA_DYNAMIC_CHAIN)
disable_lnk(ch);
if (!OMAP_DMA_CHAIN_QEMPTY(chain_id))
OMAP_DMA_CHAIN_INCQHEAD(chain_id);
status = p->dma_read(CSR, ch);
p->dma_write(status, CSR, ch);
}
if (likely(dma_chan[ch].callback != NULL))
dma_chan[ch].callback(ch, status, dma_chan[ch].data);
return 0;
}
/* STATUS register count is from 1-32 while our is 0-31 */
static irqreturn_t omap2_dma_irq_handler(int irq, void *dev_id)
{
u32 val, enable_reg;
int i;
val = p->dma_read(IRQSTATUS_L0, 0);
if (val == 0) {
if (printk_ratelimit())
printk(KERN_WARNING "Spurious DMA IRQ\n");
return IRQ_HANDLED;
}
enable_reg = p->dma_read(IRQENABLE_L0, 0);
val &= enable_reg; /* Dispatch only relevant interrupts */
for (i = 0; i < dma_lch_count && val != 0; i++) {
if (val & 1)
omap2_dma_handle_ch(i);
val >>= 1;
}
return IRQ_HANDLED;
}
static struct irqaction omap24xx_dma_irq = {
.name = "DMA",
.handler = omap2_dma_irq_handler,
};
#else
static struct irqaction omap24xx_dma_irq;
#endif
/*----------------------------------------------------------------------------*/
/*
* Note that we are currently using only IRQENABLE_L0 and L1.
* As the DSP may be using IRQENABLE_L2 and L3, let's not
* touch those for now.
*/
void omap_dma_global_context_save(void)
{
omap_dma_global_context.dma_irqenable_l0 =
p->dma_read(IRQENABLE_L0, 0);
omap_dma_global_context.dma_irqenable_l1 =
p->dma_read(IRQENABLE_L1, 0);
omap_dma_global_context.dma_ocp_sysconfig =
p->dma_read(OCP_SYSCONFIG, 0);
omap_dma_global_context.dma_gcr = p->dma_read(GCR, 0);
}
void omap_dma_global_context_restore(void)
{
int ch;
p->dma_write(omap_dma_global_context.dma_gcr, GCR, 0);
p->dma_write(omap_dma_global_context.dma_ocp_sysconfig,
OCP_SYSCONFIG, 0);
p->dma_write(omap_dma_global_context.dma_irqenable_l0,
IRQENABLE_L0, 0);
p->dma_write(omap_dma_global_context.dma_irqenable_l1,
IRQENABLE_L1, 0);
if (IS_DMA_ERRATA(DMA_ROMCODE_BUG))
p->dma_write(0x3 , IRQSTATUS_L0, 0);
for (ch = 0; ch < dma_chan_count; ch++)
if (dma_chan[ch].dev_id != -1)
omap_clear_dma(ch);
}
struct omap_system_dma_plat_info *omap_get_plat_info(void)
{
return p;
}
EXPORT_SYMBOL_GPL(omap_get_plat_info);
static int omap_system_dma_probe(struct platform_device *pdev)
{
int ch, ret = 0;
int dma_irq;
char irq_name[4];
int irq_rel;
p = pdev->dev.platform_data;
if (!p) {
dev_err(&pdev->dev,
"%s: System DMA initialized without platform data\n",
__func__);
return -EINVAL;
}
d = p->dma_attr;
errata = p->errata;
if ((d->dev_caps & RESERVE_CHANNEL) && omap_dma_reserve_channels
&& (omap_dma_reserve_channels < d->lch_count))
d->lch_count = omap_dma_reserve_channels;
dma_lch_count = d->lch_count;
dma_chan_count = dma_lch_count;
enable_1510_mode = d->dev_caps & ENABLE_1510_MODE;
dma_chan = devm_kcalloc(&pdev->dev, dma_lch_count,
sizeof(struct omap_dma_lch), GFP_KERNEL);
if (!dma_chan) {
dev_err(&pdev->dev, "%s: kzalloc fail\n", __func__);
return -ENOMEM;
}
if (dma_omap2plus()) {
dma_linked_lch = kzalloc(sizeof(struct dma_link_info) *
dma_lch_count, GFP_KERNEL);
if (!dma_linked_lch) {
ret = -ENOMEM;
goto exit_dma_lch_fail;
}
}
spin_lock_init(&dma_chan_lock);
for (ch = 0; ch < dma_chan_count; ch++) {
omap_clear_dma(ch);
if (dma_omap2plus())
omap2_disable_irq_lch(ch);
dma_chan[ch].dev_id = -1;
dma_chan[ch].next_lch = -1;
if (ch >= 6 && enable_1510_mode)
continue;
if (dma_omap1()) {
/*
* request_irq() doesn't like dev_id (ie. ch) being
* zero, so we have to kludge around this.
*/
sprintf(&irq_name[0], "%d", ch);
dma_irq = platform_get_irq_byname(pdev, irq_name);
if (dma_irq < 0) {
ret = dma_irq;
goto exit_dma_irq_fail;
}
/* INT_DMA_LCD is handled in lcd_dma.c */
if (dma_irq == INT_DMA_LCD)
continue;
ret = request_irq(dma_irq,
omap1_dma_irq_handler, 0, "DMA",
(void *) (ch + 1));
if (ret != 0)
goto exit_dma_irq_fail;
}
}
if (d->dev_caps & IS_RW_PRIORITY)
omap_dma_set_global_params(DMA_DEFAULT_ARB_RATE,
DMA_DEFAULT_FIFO_DEPTH, 0);
if (dma_omap2plus() && !(d->dev_caps & DMA_ENGINE_HANDLE_IRQ)) {
strcpy(irq_name, "0");
dma_irq = platform_get_irq_byname(pdev, irq_name);
if (dma_irq < 0) {
dev_err(&pdev->dev, "failed: request IRQ %d", dma_irq);
ret = dma_irq;
goto exit_dma_lch_fail;
}
ret = setup_irq(dma_irq, &omap24xx_dma_irq);
if (ret) {
dev_err(&pdev->dev, "set_up failed for IRQ %d for DMA (error %d)\n",
dma_irq, ret);
goto exit_dma_lch_fail;
}
}
/* reserve dma channels 0 and 1 in high security devices on 34xx */
if (d->dev_caps & HS_CHANNELS_RESERVED) {
pr_info("Reserving DMA channels 0 and 1 for HS ROM code\n");
dma_chan[0].dev_id = 0;
dma_chan[1].dev_id = 1;
}
p->show_dma_caps();
return 0;
exit_dma_irq_fail:
dev_err(&pdev->dev, "unable to request IRQ %d for DMA (error %d)\n",
dma_irq, ret);
for (irq_rel = 0; irq_rel < ch; irq_rel++) {
dma_irq = platform_get_irq(pdev, irq_rel);
free_irq(dma_irq, (void *)(irq_rel + 1));
}
exit_dma_lch_fail:
return ret;
}
static int omap_system_dma_remove(struct platform_device *pdev)
{
int dma_irq;
if (dma_omap2plus()) {
char irq_name[4];
strcpy(irq_name, "0");
dma_irq = platform_get_irq_byname(pdev, irq_name);
if (dma_irq >= 0)
remove_irq(dma_irq, &omap24xx_dma_irq);
} else {
int irq_rel = 0;
for ( ; irq_rel < dma_chan_count; irq_rel++) {
dma_irq = platform_get_irq(pdev, irq_rel);
free_irq(dma_irq, (void *)(irq_rel + 1));
}
}
return 0;
}
static struct platform_driver omap_system_dma_driver = {
.probe = omap_system_dma_probe,
.remove = omap_system_dma_remove,
.driver = {
.name = "omap_dma_system"
},
};
static int __init omap_system_dma_init(void)
{
return platform_driver_register(&omap_system_dma_driver);
}
arch_initcall(omap_system_dma_init);
static void __exit omap_system_dma_exit(void)
{
platform_driver_unregister(&omap_system_dma_driver);
}
MODULE_DESCRIPTION("OMAP SYSTEM DMA DRIVER");
MODULE_LICENSE("GPL");
MODULE_ALIAS("platform:" DRIVER_NAME);
MODULE_AUTHOR("Texas Instruments Inc");
/*
* Reserve the omap SDMA channels using cmdline bootarg
* "omap_dma_reserve_ch=". The valid range is 1 to 32
*/
static int __init omap_dma_cmdline_reserve_ch(char *str)
{
if (get_option(&str, &omap_dma_reserve_channels) != 1)
omap_dma_reserve_channels = 0;
return 1;
}
__setup("omap_dma_reserve_ch=", omap_dma_cmdline_reserve_ch);