linux/drivers/dma/at_hdmac.c

2163 lines
58 KiB
C
Raw Normal View History

// SPDX-License-Identifier: GPL-2.0-or-later
/*
* Driver for the Atmel AHB DMA Controller (aka HDMA or DMAC on AT91 systems)
*
* Copyright (C) 2008 Atmel Corporation
*
* This supports the Atmel AHB DMA Controller found in several Atmel SoCs.
* The only Atmel DMA Controller that is not covered by this driver is the one
* found on AT91SAM9263.
*/
#include <dt-bindings/dma/at91.h>
#include <linux/clk.h>
#include <linux/dmaengine.h>
#include <linux/dma-mapping.h>
#include <linux/dmapool.h>
#include <linux/interrupt.h>
#include <linux/module.h>
#include <linux/platform_device.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>
#include <linux/of.h>
#include <linux/of_device.h>
#include <linux/of_dma.h>
#include "at_hdmac_regs.h"
#include "dmaengine.h"
/*
* Glossary
* --------
*
* at_hdmac : Name of the ATmel AHB DMA Controller
* at_dma_ / atdma : ATmel DMA controller entity related
* atc_ / atchan : ATmel DMA Channel entity related
*/
#define ATC_DEFAULT_CFG (ATC_FIFOCFG_HALFFIFO)
#define ATC_DEFAULT_CTRLB (ATC_SIF(AT_DMA_MEM_IF) \
|ATC_DIF(AT_DMA_MEM_IF))
#define ATC_DMA_BUSWIDTHS\
(BIT(DMA_SLAVE_BUSWIDTH_UNDEFINED) |\
BIT(DMA_SLAVE_BUSWIDTH_1_BYTE) |\
BIT(DMA_SLAVE_BUSWIDTH_2_BYTES) |\
BIT(DMA_SLAVE_BUSWIDTH_4_BYTES))
dmaengine: at_hdmac: fix residue computation As claimed by the programmer datasheet and confirmed by the IP designer, the Block Transfer Size (BTSIZE) bitfield of the Channel x Control A Register (CTRLAx) always refers to a number of Source Width (SRC_WIDTH) transfers. Both the SRC_WIDTH and BTSIZE bitfields can be extacted from the CTRLAx register to compute the DMA residue. So the 'tx_width' field is useless and can be removed from the struct at_desc. Before this patch, atc_prep_slave_sg() was not consistent: BTSIZE was correctly initialized according to the SRC_WIDTH but 'tx_width' was always set to reg_width, which was incorrect for MEM_TO_DEV transfers. It led to bad DMA residue when 'tx_width' != SRC_WIDTH. Also the 'tx_width' field was mostly set only in the first and last descriptors. Depending on the kind of DMA transfer, this field remained uninitialized for intermediate descriptors. The accurate DMA residue was computed only when the currently processed descriptor was the first or the last of the chain. This algorithm was a little bit odd. An accurate DMA residue can always be computed using the SRC_WIDTH and BTSIZE bitfields in the CTRLAx register. Finally, the test to check whether the currently processed descriptor is the last of the chain was wrong: for cyclic transfer, last_desc->lli.dscr is NOT equal to zero, since set_desc_eol() is never called, but logically equal to first_desc->txd.phys. This bug has a side effect on the drivers/tty/serial/atmel_serial.c driver, which uses cyclic DMA transfer to receive data. Since the DMA residue was wrong each time the DMA transfer reaches the second (and last) period of the transfer, no more data were received by the USART driver till the cyclic DMA transfer loops back to the first period. Signed-off-by: Cyrille Pitchen <cyrille.pitchen@atmel.com> Acked-by: Torsten Fleischer <torfl6749@gmail.com> Tested-by: Jirí Prchal <jiri.prchal@aksignal.cz> Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2015-06-18 19:25:41 +08:00
#define ATC_MAX_DSCR_TRIALS 10
/*
* Initial number of descriptors to allocate for each channel. This could
* be increased during dma usage.
*/
static unsigned int init_nr_desc_per_channel = 64;
module_param(init_nr_desc_per_channel, uint, 0644);
MODULE_PARM_DESC(init_nr_desc_per_channel,
"initial descriptors per channel (default: 64)");
/* prototypes */
static dma_cookie_t atc_tx_submit(struct dma_async_tx_descriptor *tx);
static void atc_issue_pending(struct dma_chan *chan);
/*----------------------------------------------------------------------*/
static inline unsigned int atc_get_xfer_width(dma_addr_t src, dma_addr_t dst,
size_t len)
{
unsigned int width;
if (!((src | dst | len) & 3))
width = 2;
else if (!((src | dst | len) & 1))
width = 1;
else
width = 0;
return width;
}
static struct at_desc *atc_first_active(struct at_dma_chan *atchan)
{
return list_first_entry(&atchan->active_list,
struct at_desc, desc_node);
}
static struct at_desc *atc_first_queued(struct at_dma_chan *atchan)
{
return list_first_entry(&atchan->queue,
struct at_desc, desc_node);
}
/**
* atc_alloc_descriptor - allocate and return an initialized descriptor
* @chan: the channel to allocate descriptors for
* @gfp_flags: GFP allocation flags
*
* Note: The ack-bit is positioned in the descriptor flag at creation time
* to make initial allocation more convenient. This bit will be cleared
* and control will be given to client at usage time (during
* preparation functions).
*/
static struct at_desc *atc_alloc_descriptor(struct dma_chan *chan,
gfp_t gfp_flags)
{
struct at_desc *desc = NULL;
struct at_dma *atdma = to_at_dma(chan->device);
dma_addr_t phys;
desc = dma_pool_zalloc(atdma->dma_desc_pool, gfp_flags, &phys);
if (desc) {
INIT_LIST_HEAD(&desc->tx_list);
dma_async_tx_descriptor_init(&desc->txd, chan);
/* txd.flags will be overwritten in prep functions */
desc->txd.flags = DMA_CTRL_ACK;
desc->txd.tx_submit = atc_tx_submit;
desc->txd.phys = phys;
}
return desc;
}
/**
* atc_desc_get - get an unused descriptor from free_list
* @atchan: channel we want a new descriptor for
*/
static struct at_desc *atc_desc_get(struct at_dma_chan *atchan)
{
struct at_desc *desc, *_desc;
struct at_desc *ret = NULL;
unsigned long flags;
unsigned int i = 0;
spin_lock_irqsave(&atchan->lock, flags);
list_for_each_entry_safe(desc, _desc, &atchan->free_list, desc_node) {
i++;
if (async_tx_test_ack(&desc->txd)) {
list_del(&desc->desc_node);
ret = desc;
break;
}
dev_dbg(chan2dev(&atchan->chan_common),
"desc %p not ACKed\n", desc);
}
spin_unlock_irqrestore(&atchan->lock, flags);
dev_vdbg(chan2dev(&atchan->chan_common),
"scanned %u descriptors on freelist\n", i);
/* no more descriptor available in initial pool: create one more */
if (!ret)
ret = atc_alloc_descriptor(&atchan->chan_common, GFP_NOWAIT);
return ret;
}
/**
* atc_desc_put - move a descriptor, including any children, to the free list
* @atchan: channel we work on
* @desc: descriptor, at the head of a chain, to move to free list
*/
static void atc_desc_put(struct at_dma_chan *atchan, struct at_desc *desc)
{
if (desc) {
struct at_desc *child;
unsigned long flags;
spin_lock_irqsave(&atchan->lock, flags);
list_for_each_entry(child, &desc->tx_list, desc_node)
dev_vdbg(chan2dev(&atchan->chan_common),
"moving child desc %p to freelist\n",
child);
list_splice_init(&desc->tx_list, &atchan->free_list);
dev_vdbg(chan2dev(&atchan->chan_common),
"moving desc %p to freelist\n", desc);
list_add(&desc->desc_node, &atchan->free_list);
spin_unlock_irqrestore(&atchan->lock, flags);
}
}
/**
* atc_desc_chain - build chain adding a descriptor
* @first: address of first descriptor of the chain
* @prev: address of previous descriptor of the chain
* @desc: descriptor to queue
*
* Called from prep_* functions
*/
static void atc_desc_chain(struct at_desc **first, struct at_desc **prev,
struct at_desc *desc)
{
if (!(*first)) {
*first = desc;
} else {
/* inform the HW lli about chaining */
(*prev)->lli.dscr = desc->txd.phys;
/* insert the link descriptor to the LD ring */
list_add_tail(&desc->desc_node,
&(*first)->tx_list);
}
*prev = desc;
}
/**
* atc_dostart - starts the DMA engine for real
* @atchan: the channel we want to start
* @first: first descriptor in the list we want to begin with
*
* Called with atchan->lock held and bh disabled
*/
static void atc_dostart(struct at_dma_chan *atchan, struct at_desc *first)
{
struct at_dma *atdma = to_at_dma(atchan->chan_common.device);
/* ASSERT: channel is idle */
if (atc_chan_is_enabled(atchan)) {
dev_err(chan2dev(&atchan->chan_common),
"BUG: Attempted to start non-idle channel\n");
dev_err(chan2dev(&atchan->chan_common),
" channel: s0x%x d0x%x ctrl0x%x:0x%x l0x%x\n",
channel_readl(atchan, SADDR),
channel_readl(atchan, DADDR),
channel_readl(atchan, CTRLA),
channel_readl(atchan, CTRLB),
channel_readl(atchan, DSCR));
/* The tasklet will hopefully advance the queue... */
return;
}
vdbg_dump_regs(atchan);
channel_writel(atchan, SADDR, 0);
channel_writel(atchan, DADDR, 0);
channel_writel(atchan, CTRLA, 0);
channel_writel(atchan, CTRLB, 0);
channel_writel(atchan, DSCR, first->txd.phys);
channel_writel(atchan, SPIP, ATC_SPIP_HOLE(first->src_hole) |
ATC_SPIP_BOUNDARY(first->boundary));
channel_writel(atchan, DPIP, ATC_DPIP_HOLE(first->dst_hole) |
ATC_DPIP_BOUNDARY(first->boundary));
dma_writel(atdma, CHER, atchan->mask);
vdbg_dump_regs(atchan);
}
/*
dmaengine: at_hdmac: Fix calculation of the residual bytes This patch fixes the following issues regarding to the calculation of the residue: 1. The residue is always calculated for the current transfer even if the cookie is associated to a pending transfer. 2. For scatter/gather DMA the calculation of the residue for the current transfer doesn't include the bytes of the child descriptors that are already transferred. It only calculates the difference between the transfer's total length minus the number of bytes that are already transferred for the current child descriptor. For example: There is a scatter/gather DMA transfer with a total length of 1 MByte. Getting the residue several times while the transfer is running shows something like that: 1: residue = 975584 2: residue = 1002766 3: residue = 992627 4: residue = 983767 5: residue = 985694 6: residue = 1008094 7: residue = 1009741 8: residue = 1011195 3. The driver stores the residue but never resets it when starting a new transfer. For example: If there are two subsequent DMA transfers. The first one with a total length of 1 MByte and the second one with a total length of 1 kByte. Getting the residue for both transfers shows something like that: transfer 1: residue = 975584 transfer 2: residue = 1048380 Changes from V1: * Fixed coding style of the multi-line comments. * Improved accuracy of the residue calculation when the transfer for the first descriptor is active. Changes from V2: * Member 'tx_width' of 'struct at_desc' restored, because the transfer width can't be derived from the source width when using "slave_sg". The transfer width is needed for the calculation of the residue if either the transfer of the first or the last descriptor is in progress. In the case of a "memory_to_memory_sg" transfer (part of this patch series) the transfer width of both descriptors may differ. Thus it is required to additionally set 'tx_width' of the last descriptor. * Added functions for multiply used calculations. Signed-off-by: Torsten Fleischer <torfl6749@gmail.com> Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2015-02-24 00:54:10 +08:00
* atc_get_desc_by_cookie - get the descriptor of a cookie
* @atchan: the DMA channel
* @cookie: the cookie to get the descriptor for
*/
dmaengine: at_hdmac: Fix calculation of the residual bytes This patch fixes the following issues regarding to the calculation of the residue: 1. The residue is always calculated for the current transfer even if the cookie is associated to a pending transfer. 2. For scatter/gather DMA the calculation of the residue for the current transfer doesn't include the bytes of the child descriptors that are already transferred. It only calculates the difference between the transfer's total length minus the number of bytes that are already transferred for the current child descriptor. For example: There is a scatter/gather DMA transfer with a total length of 1 MByte. Getting the residue several times while the transfer is running shows something like that: 1: residue = 975584 2: residue = 1002766 3: residue = 992627 4: residue = 983767 5: residue = 985694 6: residue = 1008094 7: residue = 1009741 8: residue = 1011195 3. The driver stores the residue but never resets it when starting a new transfer. For example: If there are two subsequent DMA transfers. The first one with a total length of 1 MByte and the second one with a total length of 1 kByte. Getting the residue for both transfers shows something like that: transfer 1: residue = 975584 transfer 2: residue = 1048380 Changes from V1: * Fixed coding style of the multi-line comments. * Improved accuracy of the residue calculation when the transfer for the first descriptor is active. Changes from V2: * Member 'tx_width' of 'struct at_desc' restored, because the transfer width can't be derived from the source width when using "slave_sg". The transfer width is needed for the calculation of the residue if either the transfer of the first or the last descriptor is in progress. In the case of a "memory_to_memory_sg" transfer (part of this patch series) the transfer width of both descriptors may differ. Thus it is required to additionally set 'tx_width' of the last descriptor. * Added functions for multiply used calculations. Signed-off-by: Torsten Fleischer <torfl6749@gmail.com> Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2015-02-24 00:54:10 +08:00
static struct at_desc *atc_get_desc_by_cookie(struct at_dma_chan *atchan,
dma_cookie_t cookie)
{
dmaengine: at_hdmac: Fix calculation of the residual bytes This patch fixes the following issues regarding to the calculation of the residue: 1. The residue is always calculated for the current transfer even if the cookie is associated to a pending transfer. 2. For scatter/gather DMA the calculation of the residue for the current transfer doesn't include the bytes of the child descriptors that are already transferred. It only calculates the difference between the transfer's total length minus the number of bytes that are already transferred for the current child descriptor. For example: There is a scatter/gather DMA transfer with a total length of 1 MByte. Getting the residue several times while the transfer is running shows something like that: 1: residue = 975584 2: residue = 1002766 3: residue = 992627 4: residue = 983767 5: residue = 985694 6: residue = 1008094 7: residue = 1009741 8: residue = 1011195 3. The driver stores the residue but never resets it when starting a new transfer. For example: If there are two subsequent DMA transfers. The first one with a total length of 1 MByte and the second one with a total length of 1 kByte. Getting the residue for both transfers shows something like that: transfer 1: residue = 975584 transfer 2: residue = 1048380 Changes from V1: * Fixed coding style of the multi-line comments. * Improved accuracy of the residue calculation when the transfer for the first descriptor is active. Changes from V2: * Member 'tx_width' of 'struct at_desc' restored, because the transfer width can't be derived from the source width when using "slave_sg". The transfer width is needed for the calculation of the residue if either the transfer of the first or the last descriptor is in progress. In the case of a "memory_to_memory_sg" transfer (part of this patch series) the transfer width of both descriptors may differ. Thus it is required to additionally set 'tx_width' of the last descriptor. * Added functions for multiply used calculations. Signed-off-by: Torsten Fleischer <torfl6749@gmail.com> Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2015-02-24 00:54:10 +08:00
struct at_desc *desc, *_desc;
dmaengine: at_hdmac: Fix calculation of the residual bytes This patch fixes the following issues regarding to the calculation of the residue: 1. The residue is always calculated for the current transfer even if the cookie is associated to a pending transfer. 2. For scatter/gather DMA the calculation of the residue for the current transfer doesn't include the bytes of the child descriptors that are already transferred. It only calculates the difference between the transfer's total length minus the number of bytes that are already transferred for the current child descriptor. For example: There is a scatter/gather DMA transfer with a total length of 1 MByte. Getting the residue several times while the transfer is running shows something like that: 1: residue = 975584 2: residue = 1002766 3: residue = 992627 4: residue = 983767 5: residue = 985694 6: residue = 1008094 7: residue = 1009741 8: residue = 1011195 3. The driver stores the residue but never resets it when starting a new transfer. For example: If there are two subsequent DMA transfers. The first one with a total length of 1 MByte and the second one with a total length of 1 kByte. Getting the residue for both transfers shows something like that: transfer 1: residue = 975584 transfer 2: residue = 1048380 Changes from V1: * Fixed coding style of the multi-line comments. * Improved accuracy of the residue calculation when the transfer for the first descriptor is active. Changes from V2: * Member 'tx_width' of 'struct at_desc' restored, because the transfer width can't be derived from the source width when using "slave_sg". The transfer width is needed for the calculation of the residue if either the transfer of the first or the last descriptor is in progress. In the case of a "memory_to_memory_sg" transfer (part of this patch series) the transfer width of both descriptors may differ. Thus it is required to additionally set 'tx_width' of the last descriptor. * Added functions for multiply used calculations. Signed-off-by: Torsten Fleischer <torfl6749@gmail.com> Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2015-02-24 00:54:10 +08:00
list_for_each_entry_safe(desc, _desc, &atchan->queue, desc_node) {
if (desc->txd.cookie == cookie)
return desc;
}
dmaengine: at_hdmac: Fix calculation of the residual bytes This patch fixes the following issues regarding to the calculation of the residue: 1. The residue is always calculated for the current transfer even if the cookie is associated to a pending transfer. 2. For scatter/gather DMA the calculation of the residue for the current transfer doesn't include the bytes of the child descriptors that are already transferred. It only calculates the difference between the transfer's total length minus the number of bytes that are already transferred for the current child descriptor. For example: There is a scatter/gather DMA transfer with a total length of 1 MByte. Getting the residue several times while the transfer is running shows something like that: 1: residue = 975584 2: residue = 1002766 3: residue = 992627 4: residue = 983767 5: residue = 985694 6: residue = 1008094 7: residue = 1009741 8: residue = 1011195 3. The driver stores the residue but never resets it when starting a new transfer. For example: If there are two subsequent DMA transfers. The first one with a total length of 1 MByte and the second one with a total length of 1 kByte. Getting the residue for both transfers shows something like that: transfer 1: residue = 975584 transfer 2: residue = 1048380 Changes from V1: * Fixed coding style of the multi-line comments. * Improved accuracy of the residue calculation when the transfer for the first descriptor is active. Changes from V2: * Member 'tx_width' of 'struct at_desc' restored, because the transfer width can't be derived from the source width when using "slave_sg". The transfer width is needed for the calculation of the residue if either the transfer of the first or the last descriptor is in progress. In the case of a "memory_to_memory_sg" transfer (part of this patch series) the transfer width of both descriptors may differ. Thus it is required to additionally set 'tx_width' of the last descriptor. * Added functions for multiply used calculations. Signed-off-by: Torsten Fleischer <torfl6749@gmail.com> Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2015-02-24 00:54:10 +08:00
list_for_each_entry_safe(desc, _desc, &atchan->active_list, desc_node) {
if (desc->txd.cookie == cookie)
return desc;
}
dmaengine: at_hdmac: Fix calculation of the residual bytes This patch fixes the following issues regarding to the calculation of the residue: 1. The residue is always calculated for the current transfer even if the cookie is associated to a pending transfer. 2. For scatter/gather DMA the calculation of the residue for the current transfer doesn't include the bytes of the child descriptors that are already transferred. It only calculates the difference between the transfer's total length minus the number of bytes that are already transferred for the current child descriptor. For example: There is a scatter/gather DMA transfer with a total length of 1 MByte. Getting the residue several times while the transfer is running shows something like that: 1: residue = 975584 2: residue = 1002766 3: residue = 992627 4: residue = 983767 5: residue = 985694 6: residue = 1008094 7: residue = 1009741 8: residue = 1011195 3. The driver stores the residue but never resets it when starting a new transfer. For example: If there are two subsequent DMA transfers. The first one with a total length of 1 MByte and the second one with a total length of 1 kByte. Getting the residue for both transfers shows something like that: transfer 1: residue = 975584 transfer 2: residue = 1048380 Changes from V1: * Fixed coding style of the multi-line comments. * Improved accuracy of the residue calculation when the transfer for the first descriptor is active. Changes from V2: * Member 'tx_width' of 'struct at_desc' restored, because the transfer width can't be derived from the source width when using "slave_sg". The transfer width is needed for the calculation of the residue if either the transfer of the first or the last descriptor is in progress. In the case of a "memory_to_memory_sg" transfer (part of this patch series) the transfer width of both descriptors may differ. Thus it is required to additionally set 'tx_width' of the last descriptor. * Added functions for multiply used calculations. Signed-off-by: Torsten Fleischer <torfl6749@gmail.com> Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2015-02-24 00:54:10 +08:00
return NULL;
}
dmaengine: at_hdmac: Fix calculation of the residual bytes This patch fixes the following issues regarding to the calculation of the residue: 1. The residue is always calculated for the current transfer even if the cookie is associated to a pending transfer. 2. For scatter/gather DMA the calculation of the residue for the current transfer doesn't include the bytes of the child descriptors that are already transferred. It only calculates the difference between the transfer's total length minus the number of bytes that are already transferred for the current child descriptor. For example: There is a scatter/gather DMA transfer with a total length of 1 MByte. Getting the residue several times while the transfer is running shows something like that: 1: residue = 975584 2: residue = 1002766 3: residue = 992627 4: residue = 983767 5: residue = 985694 6: residue = 1008094 7: residue = 1009741 8: residue = 1011195 3. The driver stores the residue but never resets it when starting a new transfer. For example: If there are two subsequent DMA transfers. The first one with a total length of 1 MByte and the second one with a total length of 1 kByte. Getting the residue for both transfers shows something like that: transfer 1: residue = 975584 transfer 2: residue = 1048380 Changes from V1: * Fixed coding style of the multi-line comments. * Improved accuracy of the residue calculation when the transfer for the first descriptor is active. Changes from V2: * Member 'tx_width' of 'struct at_desc' restored, because the transfer width can't be derived from the source width when using "slave_sg". The transfer width is needed for the calculation of the residue if either the transfer of the first or the last descriptor is in progress. In the case of a "memory_to_memory_sg" transfer (part of this patch series) the transfer width of both descriptors may differ. Thus it is required to additionally set 'tx_width' of the last descriptor. * Added functions for multiply used calculations. Signed-off-by: Torsten Fleischer <torfl6749@gmail.com> Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2015-02-24 00:54:10 +08:00
/**
* atc_calc_bytes_left - calculates the number of bytes left according to the
* value read from CTRLA.
*
* @current_len: the number of bytes left before reading CTRLA
* @ctrla: the value of CTRLA
*/
dmaengine: at_hdmac: fix residue computation As claimed by the programmer datasheet and confirmed by the IP designer, the Block Transfer Size (BTSIZE) bitfield of the Channel x Control A Register (CTRLAx) always refers to a number of Source Width (SRC_WIDTH) transfers. Both the SRC_WIDTH and BTSIZE bitfields can be extacted from the CTRLAx register to compute the DMA residue. So the 'tx_width' field is useless and can be removed from the struct at_desc. Before this patch, atc_prep_slave_sg() was not consistent: BTSIZE was correctly initialized according to the SRC_WIDTH but 'tx_width' was always set to reg_width, which was incorrect for MEM_TO_DEV transfers. It led to bad DMA residue when 'tx_width' != SRC_WIDTH. Also the 'tx_width' field was mostly set only in the first and last descriptors. Depending on the kind of DMA transfer, this field remained uninitialized for intermediate descriptors. The accurate DMA residue was computed only when the currently processed descriptor was the first or the last of the chain. This algorithm was a little bit odd. An accurate DMA residue can always be computed using the SRC_WIDTH and BTSIZE bitfields in the CTRLAx register. Finally, the test to check whether the currently processed descriptor is the last of the chain was wrong: for cyclic transfer, last_desc->lli.dscr is NOT equal to zero, since set_desc_eol() is never called, but logically equal to first_desc->txd.phys. This bug has a side effect on the drivers/tty/serial/atmel_serial.c driver, which uses cyclic DMA transfer to receive data. Since the DMA residue was wrong each time the DMA transfer reaches the second (and last) period of the transfer, no more data were received by the USART driver till the cyclic DMA transfer loops back to the first period. Signed-off-by: Cyrille Pitchen <cyrille.pitchen@atmel.com> Acked-by: Torsten Fleischer <torfl6749@gmail.com> Tested-by: Jirí Prchal <jiri.prchal@aksignal.cz> Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2015-06-18 19:25:41 +08:00
static inline int atc_calc_bytes_left(int current_len, u32 ctrla)
dmaengine: at_hdmac: Fix calculation of the residual bytes This patch fixes the following issues regarding to the calculation of the residue: 1. The residue is always calculated for the current transfer even if the cookie is associated to a pending transfer. 2. For scatter/gather DMA the calculation of the residue for the current transfer doesn't include the bytes of the child descriptors that are already transferred. It only calculates the difference between the transfer's total length minus the number of bytes that are already transferred for the current child descriptor. For example: There is a scatter/gather DMA transfer with a total length of 1 MByte. Getting the residue several times while the transfer is running shows something like that: 1: residue = 975584 2: residue = 1002766 3: residue = 992627 4: residue = 983767 5: residue = 985694 6: residue = 1008094 7: residue = 1009741 8: residue = 1011195 3. The driver stores the residue but never resets it when starting a new transfer. For example: If there are two subsequent DMA transfers. The first one with a total length of 1 MByte and the second one with a total length of 1 kByte. Getting the residue for both transfers shows something like that: transfer 1: residue = 975584 transfer 2: residue = 1048380 Changes from V1: * Fixed coding style of the multi-line comments. * Improved accuracy of the residue calculation when the transfer for the first descriptor is active. Changes from V2: * Member 'tx_width' of 'struct at_desc' restored, because the transfer width can't be derived from the source width when using "slave_sg". The transfer width is needed for the calculation of the residue if either the transfer of the first or the last descriptor is in progress. In the case of a "memory_to_memory_sg" transfer (part of this patch series) the transfer width of both descriptors may differ. Thus it is required to additionally set 'tx_width' of the last descriptor. * Added functions for multiply used calculations. Signed-off-by: Torsten Fleischer <torfl6749@gmail.com> Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2015-02-24 00:54:10 +08:00
{
dmaengine: at_hdmac: fix residue computation As claimed by the programmer datasheet and confirmed by the IP designer, the Block Transfer Size (BTSIZE) bitfield of the Channel x Control A Register (CTRLAx) always refers to a number of Source Width (SRC_WIDTH) transfers. Both the SRC_WIDTH and BTSIZE bitfields can be extacted from the CTRLAx register to compute the DMA residue. So the 'tx_width' field is useless and can be removed from the struct at_desc. Before this patch, atc_prep_slave_sg() was not consistent: BTSIZE was correctly initialized according to the SRC_WIDTH but 'tx_width' was always set to reg_width, which was incorrect for MEM_TO_DEV transfers. It led to bad DMA residue when 'tx_width' != SRC_WIDTH. Also the 'tx_width' field was mostly set only in the first and last descriptors. Depending on the kind of DMA transfer, this field remained uninitialized for intermediate descriptors. The accurate DMA residue was computed only when the currently processed descriptor was the first or the last of the chain. This algorithm was a little bit odd. An accurate DMA residue can always be computed using the SRC_WIDTH and BTSIZE bitfields in the CTRLAx register. Finally, the test to check whether the currently processed descriptor is the last of the chain was wrong: for cyclic transfer, last_desc->lli.dscr is NOT equal to zero, since set_desc_eol() is never called, but logically equal to first_desc->txd.phys. This bug has a side effect on the drivers/tty/serial/atmel_serial.c driver, which uses cyclic DMA transfer to receive data. Since the DMA residue was wrong each time the DMA transfer reaches the second (and last) period of the transfer, no more data were received by the USART driver till the cyclic DMA transfer loops back to the first period. Signed-off-by: Cyrille Pitchen <cyrille.pitchen@atmel.com> Acked-by: Torsten Fleischer <torfl6749@gmail.com> Tested-by: Jirí Prchal <jiri.prchal@aksignal.cz> Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2015-06-18 19:25:41 +08:00
u32 btsize = (ctrla & ATC_BTSIZE_MAX);
u32 src_width = ATC_REG_TO_SRC_WIDTH(ctrla);
dmaengine: at_hdmac: Fix calculation of the residual bytes This patch fixes the following issues regarding to the calculation of the residue: 1. The residue is always calculated for the current transfer even if the cookie is associated to a pending transfer. 2. For scatter/gather DMA the calculation of the residue for the current transfer doesn't include the bytes of the child descriptors that are already transferred. It only calculates the difference between the transfer's total length minus the number of bytes that are already transferred for the current child descriptor. For example: There is a scatter/gather DMA transfer with a total length of 1 MByte. Getting the residue several times while the transfer is running shows something like that: 1: residue = 975584 2: residue = 1002766 3: residue = 992627 4: residue = 983767 5: residue = 985694 6: residue = 1008094 7: residue = 1009741 8: residue = 1011195 3. The driver stores the residue but never resets it when starting a new transfer. For example: If there are two subsequent DMA transfers. The first one with a total length of 1 MByte and the second one with a total length of 1 kByte. Getting the residue for both transfers shows something like that: transfer 1: residue = 975584 transfer 2: residue = 1048380 Changes from V1: * Fixed coding style of the multi-line comments. * Improved accuracy of the residue calculation when the transfer for the first descriptor is active. Changes from V2: * Member 'tx_width' of 'struct at_desc' restored, because the transfer width can't be derived from the source width when using "slave_sg". The transfer width is needed for the calculation of the residue if either the transfer of the first or the last descriptor is in progress. In the case of a "memory_to_memory_sg" transfer (part of this patch series) the transfer width of both descriptors may differ. Thus it is required to additionally set 'tx_width' of the last descriptor. * Added functions for multiply used calculations. Signed-off-by: Torsten Fleischer <torfl6749@gmail.com> Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2015-02-24 00:54:10 +08:00
dmaengine: at_hdmac: fix residue computation As claimed by the programmer datasheet and confirmed by the IP designer, the Block Transfer Size (BTSIZE) bitfield of the Channel x Control A Register (CTRLAx) always refers to a number of Source Width (SRC_WIDTH) transfers. Both the SRC_WIDTH and BTSIZE bitfields can be extacted from the CTRLAx register to compute the DMA residue. So the 'tx_width' field is useless and can be removed from the struct at_desc. Before this patch, atc_prep_slave_sg() was not consistent: BTSIZE was correctly initialized according to the SRC_WIDTH but 'tx_width' was always set to reg_width, which was incorrect for MEM_TO_DEV transfers. It led to bad DMA residue when 'tx_width' != SRC_WIDTH. Also the 'tx_width' field was mostly set only in the first and last descriptors. Depending on the kind of DMA transfer, this field remained uninitialized for intermediate descriptors. The accurate DMA residue was computed only when the currently processed descriptor was the first or the last of the chain. This algorithm was a little bit odd. An accurate DMA residue can always be computed using the SRC_WIDTH and BTSIZE bitfields in the CTRLAx register. Finally, the test to check whether the currently processed descriptor is the last of the chain was wrong: for cyclic transfer, last_desc->lli.dscr is NOT equal to zero, since set_desc_eol() is never called, but logically equal to first_desc->txd.phys. This bug has a side effect on the drivers/tty/serial/atmel_serial.c driver, which uses cyclic DMA transfer to receive data. Since the DMA residue was wrong each time the DMA transfer reaches the second (and last) period of the transfer, no more data were received by the USART driver till the cyclic DMA transfer loops back to the first period. Signed-off-by: Cyrille Pitchen <cyrille.pitchen@atmel.com> Acked-by: Torsten Fleischer <torfl6749@gmail.com> Tested-by: Jirí Prchal <jiri.prchal@aksignal.cz> Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2015-06-18 19:25:41 +08:00
/*
* According to the datasheet, when reading the Control A Register
* (ctrla), the Buffer Transfer Size (btsize) bitfield refers to the
* number of transfers completed on the Source Interface.
* So btsize is always a number of source width transfers.
*/
return current_len - (btsize << src_width);
dmaengine: at_hdmac: Fix calculation of the residual bytes This patch fixes the following issues regarding to the calculation of the residue: 1. The residue is always calculated for the current transfer even if the cookie is associated to a pending transfer. 2. For scatter/gather DMA the calculation of the residue for the current transfer doesn't include the bytes of the child descriptors that are already transferred. It only calculates the difference between the transfer's total length minus the number of bytes that are already transferred for the current child descriptor. For example: There is a scatter/gather DMA transfer with a total length of 1 MByte. Getting the residue several times while the transfer is running shows something like that: 1: residue = 975584 2: residue = 1002766 3: residue = 992627 4: residue = 983767 5: residue = 985694 6: residue = 1008094 7: residue = 1009741 8: residue = 1011195 3. The driver stores the residue but never resets it when starting a new transfer. For example: If there are two subsequent DMA transfers. The first one with a total length of 1 MByte and the second one with a total length of 1 kByte. Getting the residue for both transfers shows something like that: transfer 1: residue = 975584 transfer 2: residue = 1048380 Changes from V1: * Fixed coding style of the multi-line comments. * Improved accuracy of the residue calculation when the transfer for the first descriptor is active. Changes from V2: * Member 'tx_width' of 'struct at_desc' restored, because the transfer width can't be derived from the source width when using "slave_sg". The transfer width is needed for the calculation of the residue if either the transfer of the first or the last descriptor is in progress. In the case of a "memory_to_memory_sg" transfer (part of this patch series) the transfer width of both descriptors may differ. Thus it is required to additionally set 'tx_width' of the last descriptor. * Added functions for multiply used calculations. Signed-off-by: Torsten Fleischer <torfl6749@gmail.com> Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2015-02-24 00:54:10 +08:00
}
/**
* atc_get_bytes_left - get the number of bytes residue for a cookie
* @chan: DMA channel
* @cookie: transaction identifier to check status of
*/
dmaengine: at_hdmac: Fix calculation of the residual bytes This patch fixes the following issues regarding to the calculation of the residue: 1. The residue is always calculated for the current transfer even if the cookie is associated to a pending transfer. 2. For scatter/gather DMA the calculation of the residue for the current transfer doesn't include the bytes of the child descriptors that are already transferred. It only calculates the difference between the transfer's total length minus the number of bytes that are already transferred for the current child descriptor. For example: There is a scatter/gather DMA transfer with a total length of 1 MByte. Getting the residue several times while the transfer is running shows something like that: 1: residue = 975584 2: residue = 1002766 3: residue = 992627 4: residue = 983767 5: residue = 985694 6: residue = 1008094 7: residue = 1009741 8: residue = 1011195 3. The driver stores the residue but never resets it when starting a new transfer. For example: If there are two subsequent DMA transfers. The first one with a total length of 1 MByte and the second one with a total length of 1 kByte. Getting the residue for both transfers shows something like that: transfer 1: residue = 975584 transfer 2: residue = 1048380 Changes from V1: * Fixed coding style of the multi-line comments. * Improved accuracy of the residue calculation when the transfer for the first descriptor is active. Changes from V2: * Member 'tx_width' of 'struct at_desc' restored, because the transfer width can't be derived from the source width when using "slave_sg". The transfer width is needed for the calculation of the residue if either the transfer of the first or the last descriptor is in progress. In the case of a "memory_to_memory_sg" transfer (part of this patch series) the transfer width of both descriptors may differ. Thus it is required to additionally set 'tx_width' of the last descriptor. * Added functions for multiply used calculations. Signed-off-by: Torsten Fleischer <torfl6749@gmail.com> Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2015-02-24 00:54:10 +08:00
static int atc_get_bytes_left(struct dma_chan *chan, dma_cookie_t cookie)
{
struct at_dma_chan *atchan = to_at_dma_chan(chan);
struct at_desc *desc_first = atc_first_active(atchan);
dmaengine: at_hdmac: Fix calculation of the residual bytes This patch fixes the following issues regarding to the calculation of the residue: 1. The residue is always calculated for the current transfer even if the cookie is associated to a pending transfer. 2. For scatter/gather DMA the calculation of the residue for the current transfer doesn't include the bytes of the child descriptors that are already transferred. It only calculates the difference between the transfer's total length minus the number of bytes that are already transferred for the current child descriptor. For example: There is a scatter/gather DMA transfer with a total length of 1 MByte. Getting the residue several times while the transfer is running shows something like that: 1: residue = 975584 2: residue = 1002766 3: residue = 992627 4: residue = 983767 5: residue = 985694 6: residue = 1008094 7: residue = 1009741 8: residue = 1011195 3. The driver stores the residue but never resets it when starting a new transfer. For example: If there are two subsequent DMA transfers. The first one with a total length of 1 MByte and the second one with a total length of 1 kByte. Getting the residue for both transfers shows something like that: transfer 1: residue = 975584 transfer 2: residue = 1048380 Changes from V1: * Fixed coding style of the multi-line comments. * Improved accuracy of the residue calculation when the transfer for the first descriptor is active. Changes from V2: * Member 'tx_width' of 'struct at_desc' restored, because the transfer width can't be derived from the source width when using "slave_sg". The transfer width is needed for the calculation of the residue if either the transfer of the first or the last descriptor is in progress. In the case of a "memory_to_memory_sg" transfer (part of this patch series) the transfer width of both descriptors may differ. Thus it is required to additionally set 'tx_width' of the last descriptor. * Added functions for multiply used calculations. Signed-off-by: Torsten Fleischer <torfl6749@gmail.com> Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2015-02-24 00:54:10 +08:00
struct at_desc *desc;
int ret;
dmaengine: at_hdmac: fix residue computation As claimed by the programmer datasheet and confirmed by the IP designer, the Block Transfer Size (BTSIZE) bitfield of the Channel x Control A Register (CTRLAx) always refers to a number of Source Width (SRC_WIDTH) transfers. Both the SRC_WIDTH and BTSIZE bitfields can be extacted from the CTRLAx register to compute the DMA residue. So the 'tx_width' field is useless and can be removed from the struct at_desc. Before this patch, atc_prep_slave_sg() was not consistent: BTSIZE was correctly initialized according to the SRC_WIDTH but 'tx_width' was always set to reg_width, which was incorrect for MEM_TO_DEV transfers. It led to bad DMA residue when 'tx_width' != SRC_WIDTH. Also the 'tx_width' field was mostly set only in the first and last descriptors. Depending on the kind of DMA transfer, this field remained uninitialized for intermediate descriptors. The accurate DMA residue was computed only when the currently processed descriptor was the first or the last of the chain. This algorithm was a little bit odd. An accurate DMA residue can always be computed using the SRC_WIDTH and BTSIZE bitfields in the CTRLAx register. Finally, the test to check whether the currently processed descriptor is the last of the chain was wrong: for cyclic transfer, last_desc->lli.dscr is NOT equal to zero, since set_desc_eol() is never called, but logically equal to first_desc->txd.phys. This bug has a side effect on the drivers/tty/serial/atmel_serial.c driver, which uses cyclic DMA transfer to receive data. Since the DMA residue was wrong each time the DMA transfer reaches the second (and last) period of the transfer, no more data were received by the USART driver till the cyclic DMA transfer loops back to the first period. Signed-off-by: Cyrille Pitchen <cyrille.pitchen@atmel.com> Acked-by: Torsten Fleischer <torfl6749@gmail.com> Tested-by: Jirí Prchal <jiri.prchal@aksignal.cz> Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2015-06-18 19:25:41 +08:00
u32 ctrla, dscr, trials;
/*
dmaengine: at_hdmac: Fix calculation of the residual bytes This patch fixes the following issues regarding to the calculation of the residue: 1. The residue is always calculated for the current transfer even if the cookie is associated to a pending transfer. 2. For scatter/gather DMA the calculation of the residue for the current transfer doesn't include the bytes of the child descriptors that are already transferred. It only calculates the difference between the transfer's total length minus the number of bytes that are already transferred for the current child descriptor. For example: There is a scatter/gather DMA transfer with a total length of 1 MByte. Getting the residue several times while the transfer is running shows something like that: 1: residue = 975584 2: residue = 1002766 3: residue = 992627 4: residue = 983767 5: residue = 985694 6: residue = 1008094 7: residue = 1009741 8: residue = 1011195 3. The driver stores the residue but never resets it when starting a new transfer. For example: If there are two subsequent DMA transfers. The first one with a total length of 1 MByte and the second one with a total length of 1 kByte. Getting the residue for both transfers shows something like that: transfer 1: residue = 975584 transfer 2: residue = 1048380 Changes from V1: * Fixed coding style of the multi-line comments. * Improved accuracy of the residue calculation when the transfer for the first descriptor is active. Changes from V2: * Member 'tx_width' of 'struct at_desc' restored, because the transfer width can't be derived from the source width when using "slave_sg". The transfer width is needed for the calculation of the residue if either the transfer of the first or the last descriptor is in progress. In the case of a "memory_to_memory_sg" transfer (part of this patch series) the transfer width of both descriptors may differ. Thus it is required to additionally set 'tx_width' of the last descriptor. * Added functions for multiply used calculations. Signed-off-by: Torsten Fleischer <torfl6749@gmail.com> Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2015-02-24 00:54:10 +08:00
* If the cookie doesn't match to the currently running transfer then
* we can return the total length of the associated DMA transfer,
* because it is still queued.
*/
dmaengine: at_hdmac: Fix calculation of the residual bytes This patch fixes the following issues regarding to the calculation of the residue: 1. The residue is always calculated for the current transfer even if the cookie is associated to a pending transfer. 2. For scatter/gather DMA the calculation of the residue for the current transfer doesn't include the bytes of the child descriptors that are already transferred. It only calculates the difference between the transfer's total length minus the number of bytes that are already transferred for the current child descriptor. For example: There is a scatter/gather DMA transfer with a total length of 1 MByte. Getting the residue several times while the transfer is running shows something like that: 1: residue = 975584 2: residue = 1002766 3: residue = 992627 4: residue = 983767 5: residue = 985694 6: residue = 1008094 7: residue = 1009741 8: residue = 1011195 3. The driver stores the residue but never resets it when starting a new transfer. For example: If there are two subsequent DMA transfers. The first one with a total length of 1 MByte and the second one with a total length of 1 kByte. Getting the residue for both transfers shows something like that: transfer 1: residue = 975584 transfer 2: residue = 1048380 Changes from V1: * Fixed coding style of the multi-line comments. * Improved accuracy of the residue calculation when the transfer for the first descriptor is active. Changes from V2: * Member 'tx_width' of 'struct at_desc' restored, because the transfer width can't be derived from the source width when using "slave_sg". The transfer width is needed for the calculation of the residue if either the transfer of the first or the last descriptor is in progress. In the case of a "memory_to_memory_sg" transfer (part of this patch series) the transfer width of both descriptors may differ. Thus it is required to additionally set 'tx_width' of the last descriptor. * Added functions for multiply used calculations. Signed-off-by: Torsten Fleischer <torfl6749@gmail.com> Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2015-02-24 00:54:10 +08:00
desc = atc_get_desc_by_cookie(atchan, cookie);
if (desc == NULL)
return -EINVAL;
else if (desc != desc_first)
return desc->total_len;
dmaengine: at_hdmac: Fix calculation of the residual bytes This patch fixes the following issues regarding to the calculation of the residue: 1. The residue is always calculated for the current transfer even if the cookie is associated to a pending transfer. 2. For scatter/gather DMA the calculation of the residue for the current transfer doesn't include the bytes of the child descriptors that are already transferred. It only calculates the difference between the transfer's total length minus the number of bytes that are already transferred for the current child descriptor. For example: There is a scatter/gather DMA transfer with a total length of 1 MByte. Getting the residue several times while the transfer is running shows something like that: 1: residue = 975584 2: residue = 1002766 3: residue = 992627 4: residue = 983767 5: residue = 985694 6: residue = 1008094 7: residue = 1009741 8: residue = 1011195 3. The driver stores the residue but never resets it when starting a new transfer. For example: If there are two subsequent DMA transfers. The first one with a total length of 1 MByte and the second one with a total length of 1 kByte. Getting the residue for both transfers shows something like that: transfer 1: residue = 975584 transfer 2: residue = 1048380 Changes from V1: * Fixed coding style of the multi-line comments. * Improved accuracy of the residue calculation when the transfer for the first descriptor is active. Changes from V2: * Member 'tx_width' of 'struct at_desc' restored, because the transfer width can't be derived from the source width when using "slave_sg". The transfer width is needed for the calculation of the residue if either the transfer of the first or the last descriptor is in progress. In the case of a "memory_to_memory_sg" transfer (part of this patch series) the transfer width of both descriptors may differ. Thus it is required to additionally set 'tx_width' of the last descriptor. * Added functions for multiply used calculations. Signed-off-by: Torsten Fleischer <torfl6749@gmail.com> Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2015-02-24 00:54:10 +08:00
/* cookie matches to the currently running transfer */
ret = desc_first->total_len;
dmaengine: at_hdmac: Fix calculation of the residual bytes This patch fixes the following issues regarding to the calculation of the residue: 1. The residue is always calculated for the current transfer even if the cookie is associated to a pending transfer. 2. For scatter/gather DMA the calculation of the residue for the current transfer doesn't include the bytes of the child descriptors that are already transferred. It only calculates the difference between the transfer's total length minus the number of bytes that are already transferred for the current child descriptor. For example: There is a scatter/gather DMA transfer with a total length of 1 MByte. Getting the residue several times while the transfer is running shows something like that: 1: residue = 975584 2: residue = 1002766 3: residue = 992627 4: residue = 983767 5: residue = 985694 6: residue = 1008094 7: residue = 1009741 8: residue = 1011195 3. The driver stores the residue but never resets it when starting a new transfer. For example: If there are two subsequent DMA transfers. The first one with a total length of 1 MByte and the second one with a total length of 1 kByte. Getting the residue for both transfers shows something like that: transfer 1: residue = 975584 transfer 2: residue = 1048380 Changes from V1: * Fixed coding style of the multi-line comments. * Improved accuracy of the residue calculation when the transfer for the first descriptor is active. Changes from V2: * Member 'tx_width' of 'struct at_desc' restored, because the transfer width can't be derived from the source width when using "slave_sg". The transfer width is needed for the calculation of the residue if either the transfer of the first or the last descriptor is in progress. In the case of a "memory_to_memory_sg" transfer (part of this patch series) the transfer width of both descriptors may differ. Thus it is required to additionally set 'tx_width' of the last descriptor. * Added functions for multiply used calculations. Signed-off-by: Torsten Fleischer <torfl6749@gmail.com> Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2015-02-24 00:54:10 +08:00
if (desc_first->lli.dscr) {
/* hardware linked list transfer */
/*
* Calculate the residue by removing the length of the child
* descriptors already transferred from the total length.
* To get the current child descriptor we can use the value of
* the channel's DSCR register and compare it against the value
* of the hardware linked list structure of each child
* descriptor.
dmaengine: at_hdmac: fix residue computation As claimed by the programmer datasheet and confirmed by the IP designer, the Block Transfer Size (BTSIZE) bitfield of the Channel x Control A Register (CTRLAx) always refers to a number of Source Width (SRC_WIDTH) transfers. Both the SRC_WIDTH and BTSIZE bitfields can be extacted from the CTRLAx register to compute the DMA residue. So the 'tx_width' field is useless and can be removed from the struct at_desc. Before this patch, atc_prep_slave_sg() was not consistent: BTSIZE was correctly initialized according to the SRC_WIDTH but 'tx_width' was always set to reg_width, which was incorrect for MEM_TO_DEV transfers. It led to bad DMA residue when 'tx_width' != SRC_WIDTH. Also the 'tx_width' field was mostly set only in the first and last descriptors. Depending on the kind of DMA transfer, this field remained uninitialized for intermediate descriptors. The accurate DMA residue was computed only when the currently processed descriptor was the first or the last of the chain. This algorithm was a little bit odd. An accurate DMA residue can always be computed using the SRC_WIDTH and BTSIZE bitfields in the CTRLAx register. Finally, the test to check whether the currently processed descriptor is the last of the chain was wrong: for cyclic transfer, last_desc->lli.dscr is NOT equal to zero, since set_desc_eol() is never called, but logically equal to first_desc->txd.phys. This bug has a side effect on the drivers/tty/serial/atmel_serial.c driver, which uses cyclic DMA transfer to receive data. Since the DMA residue was wrong each time the DMA transfer reaches the second (and last) period of the transfer, no more data were received by the USART driver till the cyclic DMA transfer loops back to the first period. Signed-off-by: Cyrille Pitchen <cyrille.pitchen@atmel.com> Acked-by: Torsten Fleischer <torfl6749@gmail.com> Tested-by: Jirí Prchal <jiri.prchal@aksignal.cz> Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2015-06-18 19:25:41 +08:00
*
* The CTRLA register provides us with the amount of data
* already read from the source for the current child
* descriptor. So we can compute a more accurate residue by also
* removing the number of bytes corresponding to this amount of
* data.
*
* However, the DSCR and CTRLA registers cannot be read both
* atomically. Hence a race condition may occur: the first read
* register may refer to one child descriptor whereas the second
* read may refer to a later child descriptor in the list
* because of the DMA transfer progression inbetween the two
* reads.
*
* One solution could have been to pause the DMA transfer, read
* the DSCR and CTRLA then resume the DMA transfer. Nonetheless,
* this approach presents some drawbacks:
* - If the DMA transfer is paused, RX overruns or TX underruns
* are more likey to occur depending on the system latency.
* Taking the USART driver as an example, it uses a cyclic DMA
* transfer to read data from the Receive Holding Register
* (RHR) to avoid RX overruns since the RHR is not protected
* by any FIFO on most Atmel SoCs. So pausing the DMA transfer
* to compute the residue would break the USART driver design.
* - The atc_pause() function masks interrupts but we'd rather
* avoid to do so for system latency purpose.
*
* Then we'd rather use another solution: the DSCR is read a
* first time, the CTRLA is read in turn, next the DSCR is read
* a second time. If the two consecutive read values of the DSCR
* are the same then we assume both refers to the very same
* child descriptor as well as the CTRLA value read inbetween
* does. For cyclic tranfers, the assumption is that a full loop
* is "not so fast".
* If the two DSCR values are different, we read again the CTRLA
* then the DSCR till two consecutive read values from DSCR are
* equal or till the maxium trials is reach.
* This algorithm is very unlikely not to find a stable value for
* DSCR.
dmaengine: at_hdmac: Fix calculation of the residual bytes This patch fixes the following issues regarding to the calculation of the residue: 1. The residue is always calculated for the current transfer even if the cookie is associated to a pending transfer. 2. For scatter/gather DMA the calculation of the residue for the current transfer doesn't include the bytes of the child descriptors that are already transferred. It only calculates the difference between the transfer's total length minus the number of bytes that are already transferred for the current child descriptor. For example: There is a scatter/gather DMA transfer with a total length of 1 MByte. Getting the residue several times while the transfer is running shows something like that: 1: residue = 975584 2: residue = 1002766 3: residue = 992627 4: residue = 983767 5: residue = 985694 6: residue = 1008094 7: residue = 1009741 8: residue = 1011195 3. The driver stores the residue but never resets it when starting a new transfer. For example: If there are two subsequent DMA transfers. The first one with a total length of 1 MByte and the second one with a total length of 1 kByte. Getting the residue for both transfers shows something like that: transfer 1: residue = 975584 transfer 2: residue = 1048380 Changes from V1: * Fixed coding style of the multi-line comments. * Improved accuracy of the residue calculation when the transfer for the first descriptor is active. Changes from V2: * Member 'tx_width' of 'struct at_desc' restored, because the transfer width can't be derived from the source width when using "slave_sg". The transfer width is needed for the calculation of the residue if either the transfer of the first or the last descriptor is in progress. In the case of a "memory_to_memory_sg" transfer (part of this patch series) the transfer width of both descriptors may differ. Thus it is required to additionally set 'tx_width' of the last descriptor. * Added functions for multiply used calculations. Signed-off-by: Torsten Fleischer <torfl6749@gmail.com> Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2015-02-24 00:54:10 +08:00
*/
dscr = channel_readl(atchan, DSCR);
dmaengine: at_hdmac: fix residue computation As claimed by the programmer datasheet and confirmed by the IP designer, the Block Transfer Size (BTSIZE) bitfield of the Channel x Control A Register (CTRLAx) always refers to a number of Source Width (SRC_WIDTH) transfers. Both the SRC_WIDTH and BTSIZE bitfields can be extacted from the CTRLAx register to compute the DMA residue. So the 'tx_width' field is useless and can be removed from the struct at_desc. Before this patch, atc_prep_slave_sg() was not consistent: BTSIZE was correctly initialized according to the SRC_WIDTH but 'tx_width' was always set to reg_width, which was incorrect for MEM_TO_DEV transfers. It led to bad DMA residue when 'tx_width' != SRC_WIDTH. Also the 'tx_width' field was mostly set only in the first and last descriptors. Depending on the kind of DMA transfer, this field remained uninitialized for intermediate descriptors. The accurate DMA residue was computed only when the currently processed descriptor was the first or the last of the chain. This algorithm was a little bit odd. An accurate DMA residue can always be computed using the SRC_WIDTH and BTSIZE bitfields in the CTRLAx register. Finally, the test to check whether the currently processed descriptor is the last of the chain was wrong: for cyclic transfer, last_desc->lli.dscr is NOT equal to zero, since set_desc_eol() is never called, but logically equal to first_desc->txd.phys. This bug has a side effect on the drivers/tty/serial/atmel_serial.c driver, which uses cyclic DMA transfer to receive data. Since the DMA residue was wrong each time the DMA transfer reaches the second (and last) period of the transfer, no more data were received by the USART driver till the cyclic DMA transfer loops back to the first period. Signed-off-by: Cyrille Pitchen <cyrille.pitchen@atmel.com> Acked-by: Torsten Fleischer <torfl6749@gmail.com> Tested-by: Jirí Prchal <jiri.prchal@aksignal.cz> Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2015-06-18 19:25:41 +08:00
rmb(); /* ensure DSCR is read before CTRLA */
ctrla = channel_readl(atchan, CTRLA);
for (trials = 0; trials < ATC_MAX_DSCR_TRIALS; ++trials) {
u32 new_dscr;
rmb(); /* ensure DSCR is read after CTRLA */
new_dscr = channel_readl(atchan, DSCR);
/*
* If the DSCR register value has not changed inside the
* DMA controller since the previous read, we assume
* that both the dscr and ctrla values refers to the
* very same descriptor.
*/
if (likely(new_dscr == dscr))
break;
/*
* DSCR has changed inside the DMA controller, so the
* previouly read value of CTRLA may refer to an already
* processed descriptor hence could be outdated.
* We need to update ctrla to match the current
* descriptor.
*/
dscr = new_dscr;
rmb(); /* ensure DSCR is read before CTRLA */
ctrla = channel_readl(atchan, CTRLA);
}
if (unlikely(trials >= ATC_MAX_DSCR_TRIALS))
return -ETIMEDOUT;
dmaengine: at_hdmac: Fix calculation of the residual bytes This patch fixes the following issues regarding to the calculation of the residue: 1. The residue is always calculated for the current transfer even if the cookie is associated to a pending transfer. 2. For scatter/gather DMA the calculation of the residue for the current transfer doesn't include the bytes of the child descriptors that are already transferred. It only calculates the difference between the transfer's total length minus the number of bytes that are already transferred for the current child descriptor. For example: There is a scatter/gather DMA transfer with a total length of 1 MByte. Getting the residue several times while the transfer is running shows something like that: 1: residue = 975584 2: residue = 1002766 3: residue = 992627 4: residue = 983767 5: residue = 985694 6: residue = 1008094 7: residue = 1009741 8: residue = 1011195 3. The driver stores the residue but never resets it when starting a new transfer. For example: If there are two subsequent DMA transfers. The first one with a total length of 1 MByte and the second one with a total length of 1 kByte. Getting the residue for both transfers shows something like that: transfer 1: residue = 975584 transfer 2: residue = 1048380 Changes from V1: * Fixed coding style of the multi-line comments. * Improved accuracy of the residue calculation when the transfer for the first descriptor is active. Changes from V2: * Member 'tx_width' of 'struct at_desc' restored, because the transfer width can't be derived from the source width when using "slave_sg". The transfer width is needed for the calculation of the residue if either the transfer of the first or the last descriptor is in progress. In the case of a "memory_to_memory_sg" transfer (part of this patch series) the transfer width of both descriptors may differ. Thus it is required to additionally set 'tx_width' of the last descriptor. * Added functions for multiply used calculations. Signed-off-by: Torsten Fleischer <torfl6749@gmail.com> Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2015-02-24 00:54:10 +08:00
/* for the first descriptor we can be more accurate */
if (desc_first->lli.dscr == dscr)
dmaengine: at_hdmac: fix residue computation As claimed by the programmer datasheet and confirmed by the IP designer, the Block Transfer Size (BTSIZE) bitfield of the Channel x Control A Register (CTRLAx) always refers to a number of Source Width (SRC_WIDTH) transfers. Both the SRC_WIDTH and BTSIZE bitfields can be extacted from the CTRLAx register to compute the DMA residue. So the 'tx_width' field is useless and can be removed from the struct at_desc. Before this patch, atc_prep_slave_sg() was not consistent: BTSIZE was correctly initialized according to the SRC_WIDTH but 'tx_width' was always set to reg_width, which was incorrect for MEM_TO_DEV transfers. It led to bad DMA residue when 'tx_width' != SRC_WIDTH. Also the 'tx_width' field was mostly set only in the first and last descriptors. Depending on the kind of DMA transfer, this field remained uninitialized for intermediate descriptors. The accurate DMA residue was computed only when the currently processed descriptor was the first or the last of the chain. This algorithm was a little bit odd. An accurate DMA residue can always be computed using the SRC_WIDTH and BTSIZE bitfields in the CTRLAx register. Finally, the test to check whether the currently processed descriptor is the last of the chain was wrong: for cyclic transfer, last_desc->lli.dscr is NOT equal to zero, since set_desc_eol() is never called, but logically equal to first_desc->txd.phys. This bug has a side effect on the drivers/tty/serial/atmel_serial.c driver, which uses cyclic DMA transfer to receive data. Since the DMA residue was wrong each time the DMA transfer reaches the second (and last) period of the transfer, no more data were received by the USART driver till the cyclic DMA transfer loops back to the first period. Signed-off-by: Cyrille Pitchen <cyrille.pitchen@atmel.com> Acked-by: Torsten Fleischer <torfl6749@gmail.com> Tested-by: Jirí Prchal <jiri.prchal@aksignal.cz> Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2015-06-18 19:25:41 +08:00
return atc_calc_bytes_left(ret, ctrla);
dmaengine: at_hdmac: Fix calculation of the residual bytes This patch fixes the following issues regarding to the calculation of the residue: 1. The residue is always calculated for the current transfer even if the cookie is associated to a pending transfer. 2. For scatter/gather DMA the calculation of the residue for the current transfer doesn't include the bytes of the child descriptors that are already transferred. It only calculates the difference between the transfer's total length minus the number of bytes that are already transferred for the current child descriptor. For example: There is a scatter/gather DMA transfer with a total length of 1 MByte. Getting the residue several times while the transfer is running shows something like that: 1: residue = 975584 2: residue = 1002766 3: residue = 992627 4: residue = 983767 5: residue = 985694 6: residue = 1008094 7: residue = 1009741 8: residue = 1011195 3. The driver stores the residue but never resets it when starting a new transfer. For example: If there are two subsequent DMA transfers. The first one with a total length of 1 MByte and the second one with a total length of 1 kByte. Getting the residue for both transfers shows something like that: transfer 1: residue = 975584 transfer 2: residue = 1048380 Changes from V1: * Fixed coding style of the multi-line comments. * Improved accuracy of the residue calculation when the transfer for the first descriptor is active. Changes from V2: * Member 'tx_width' of 'struct at_desc' restored, because the transfer width can't be derived from the source width when using "slave_sg". The transfer width is needed for the calculation of the residue if either the transfer of the first or the last descriptor is in progress. In the case of a "memory_to_memory_sg" transfer (part of this patch series) the transfer width of both descriptors may differ. Thus it is required to additionally set 'tx_width' of the last descriptor. * Added functions for multiply used calculations. Signed-off-by: Torsten Fleischer <torfl6749@gmail.com> Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2015-02-24 00:54:10 +08:00
ret -= desc_first->len;
list_for_each_entry(desc, &desc_first->tx_list, desc_node) {
if (desc->lli.dscr == dscr)
break;
ret -= desc->len;
}
/*
dmaengine: at_hdmac: fix residue computation As claimed by the programmer datasheet and confirmed by the IP designer, the Block Transfer Size (BTSIZE) bitfield of the Channel x Control A Register (CTRLAx) always refers to a number of Source Width (SRC_WIDTH) transfers. Both the SRC_WIDTH and BTSIZE bitfields can be extacted from the CTRLAx register to compute the DMA residue. So the 'tx_width' field is useless and can be removed from the struct at_desc. Before this patch, atc_prep_slave_sg() was not consistent: BTSIZE was correctly initialized according to the SRC_WIDTH but 'tx_width' was always set to reg_width, which was incorrect for MEM_TO_DEV transfers. It led to bad DMA residue when 'tx_width' != SRC_WIDTH. Also the 'tx_width' field was mostly set only in the first and last descriptors. Depending on the kind of DMA transfer, this field remained uninitialized for intermediate descriptors. The accurate DMA residue was computed only when the currently processed descriptor was the first or the last of the chain. This algorithm was a little bit odd. An accurate DMA residue can always be computed using the SRC_WIDTH and BTSIZE bitfields in the CTRLAx register. Finally, the test to check whether the currently processed descriptor is the last of the chain was wrong: for cyclic transfer, last_desc->lli.dscr is NOT equal to zero, since set_desc_eol() is never called, but logically equal to first_desc->txd.phys. This bug has a side effect on the drivers/tty/serial/atmel_serial.c driver, which uses cyclic DMA transfer to receive data. Since the DMA residue was wrong each time the DMA transfer reaches the second (and last) period of the transfer, no more data were received by the USART driver till the cyclic DMA transfer loops back to the first period. Signed-off-by: Cyrille Pitchen <cyrille.pitchen@atmel.com> Acked-by: Torsten Fleischer <torfl6749@gmail.com> Tested-by: Jirí Prchal <jiri.prchal@aksignal.cz> Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2015-06-18 19:25:41 +08:00
* For the current descriptor in the chain we can calculate
dmaengine: at_hdmac: Fix calculation of the residual bytes This patch fixes the following issues regarding to the calculation of the residue: 1. The residue is always calculated for the current transfer even if the cookie is associated to a pending transfer. 2. For scatter/gather DMA the calculation of the residue for the current transfer doesn't include the bytes of the child descriptors that are already transferred. It only calculates the difference between the transfer's total length minus the number of bytes that are already transferred for the current child descriptor. For example: There is a scatter/gather DMA transfer with a total length of 1 MByte. Getting the residue several times while the transfer is running shows something like that: 1: residue = 975584 2: residue = 1002766 3: residue = 992627 4: residue = 983767 5: residue = 985694 6: residue = 1008094 7: residue = 1009741 8: residue = 1011195 3. The driver stores the residue but never resets it when starting a new transfer. For example: If there are two subsequent DMA transfers. The first one with a total length of 1 MByte and the second one with a total length of 1 kByte. Getting the residue for both transfers shows something like that: transfer 1: residue = 975584 transfer 2: residue = 1048380 Changes from V1: * Fixed coding style of the multi-line comments. * Improved accuracy of the residue calculation when the transfer for the first descriptor is active. Changes from V2: * Member 'tx_width' of 'struct at_desc' restored, because the transfer width can't be derived from the source width when using "slave_sg". The transfer width is needed for the calculation of the residue if either the transfer of the first or the last descriptor is in progress. In the case of a "memory_to_memory_sg" transfer (part of this patch series) the transfer width of both descriptors may differ. Thus it is required to additionally set 'tx_width' of the last descriptor. * Added functions for multiply used calculations. Signed-off-by: Torsten Fleischer <torfl6749@gmail.com> Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2015-02-24 00:54:10 +08:00
* the remaining bytes using the channel's register.
*/
dmaengine: at_hdmac: fix residue computation As claimed by the programmer datasheet and confirmed by the IP designer, the Block Transfer Size (BTSIZE) bitfield of the Channel x Control A Register (CTRLAx) always refers to a number of Source Width (SRC_WIDTH) transfers. Both the SRC_WIDTH and BTSIZE bitfields can be extacted from the CTRLAx register to compute the DMA residue. So the 'tx_width' field is useless and can be removed from the struct at_desc. Before this patch, atc_prep_slave_sg() was not consistent: BTSIZE was correctly initialized according to the SRC_WIDTH but 'tx_width' was always set to reg_width, which was incorrect for MEM_TO_DEV transfers. It led to bad DMA residue when 'tx_width' != SRC_WIDTH. Also the 'tx_width' field was mostly set only in the first and last descriptors. Depending on the kind of DMA transfer, this field remained uninitialized for intermediate descriptors. The accurate DMA residue was computed only when the currently processed descriptor was the first or the last of the chain. This algorithm was a little bit odd. An accurate DMA residue can always be computed using the SRC_WIDTH and BTSIZE bitfields in the CTRLAx register. Finally, the test to check whether the currently processed descriptor is the last of the chain was wrong: for cyclic transfer, last_desc->lli.dscr is NOT equal to zero, since set_desc_eol() is never called, but logically equal to first_desc->txd.phys. This bug has a side effect on the drivers/tty/serial/atmel_serial.c driver, which uses cyclic DMA transfer to receive data. Since the DMA residue was wrong each time the DMA transfer reaches the second (and last) period of the transfer, no more data were received by the USART driver till the cyclic DMA transfer loops back to the first period. Signed-off-by: Cyrille Pitchen <cyrille.pitchen@atmel.com> Acked-by: Torsten Fleischer <torfl6749@gmail.com> Tested-by: Jirí Prchal <jiri.prchal@aksignal.cz> Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2015-06-18 19:25:41 +08:00
ret = atc_calc_bytes_left(ret, ctrla);
dmaengine: at_hdmac: Fix calculation of the residual bytes This patch fixes the following issues regarding to the calculation of the residue: 1. The residue is always calculated for the current transfer even if the cookie is associated to a pending transfer. 2. For scatter/gather DMA the calculation of the residue for the current transfer doesn't include the bytes of the child descriptors that are already transferred. It only calculates the difference between the transfer's total length minus the number of bytes that are already transferred for the current child descriptor. For example: There is a scatter/gather DMA transfer with a total length of 1 MByte. Getting the residue several times while the transfer is running shows something like that: 1: residue = 975584 2: residue = 1002766 3: residue = 992627 4: residue = 983767 5: residue = 985694 6: residue = 1008094 7: residue = 1009741 8: residue = 1011195 3. The driver stores the residue but never resets it when starting a new transfer. For example: If there are two subsequent DMA transfers. The first one with a total length of 1 MByte and the second one with a total length of 1 kByte. Getting the residue for both transfers shows something like that: transfer 1: residue = 975584 transfer 2: residue = 1048380 Changes from V1: * Fixed coding style of the multi-line comments. * Improved accuracy of the residue calculation when the transfer for the first descriptor is active. Changes from V2: * Member 'tx_width' of 'struct at_desc' restored, because the transfer width can't be derived from the source width when using "slave_sg". The transfer width is needed for the calculation of the residue if either the transfer of the first or the last descriptor is in progress. In the case of a "memory_to_memory_sg" transfer (part of this patch series) the transfer width of both descriptors may differ. Thus it is required to additionally set 'tx_width' of the last descriptor. * Added functions for multiply used calculations. Signed-off-by: Torsten Fleischer <torfl6749@gmail.com> Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2015-02-24 00:54:10 +08:00
} else {
/* single transfer */
dmaengine: at_hdmac: fix residue computation As claimed by the programmer datasheet and confirmed by the IP designer, the Block Transfer Size (BTSIZE) bitfield of the Channel x Control A Register (CTRLAx) always refers to a number of Source Width (SRC_WIDTH) transfers. Both the SRC_WIDTH and BTSIZE bitfields can be extacted from the CTRLAx register to compute the DMA residue. So the 'tx_width' field is useless and can be removed from the struct at_desc. Before this patch, atc_prep_slave_sg() was not consistent: BTSIZE was correctly initialized according to the SRC_WIDTH but 'tx_width' was always set to reg_width, which was incorrect for MEM_TO_DEV transfers. It led to bad DMA residue when 'tx_width' != SRC_WIDTH. Also the 'tx_width' field was mostly set only in the first and last descriptors. Depending on the kind of DMA transfer, this field remained uninitialized for intermediate descriptors. The accurate DMA residue was computed only when the currently processed descriptor was the first or the last of the chain. This algorithm was a little bit odd. An accurate DMA residue can always be computed using the SRC_WIDTH and BTSIZE bitfields in the CTRLAx register. Finally, the test to check whether the currently processed descriptor is the last of the chain was wrong: for cyclic transfer, last_desc->lli.dscr is NOT equal to zero, since set_desc_eol() is never called, but logically equal to first_desc->txd.phys. This bug has a side effect on the drivers/tty/serial/atmel_serial.c driver, which uses cyclic DMA transfer to receive data. Since the DMA residue was wrong each time the DMA transfer reaches the second (and last) period of the transfer, no more data were received by the USART driver till the cyclic DMA transfer loops back to the first period. Signed-off-by: Cyrille Pitchen <cyrille.pitchen@atmel.com> Acked-by: Torsten Fleischer <torfl6749@gmail.com> Tested-by: Jirí Prchal <jiri.prchal@aksignal.cz> Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2015-06-18 19:25:41 +08:00
ctrla = channel_readl(atchan, CTRLA);
ret = atc_calc_bytes_left(ret, ctrla);
}
return ret;
}
/**
* atc_chain_complete - finish work for one transaction chain
* @atchan: channel we work on
* @desc: descriptor at the head of the chain we want do complete
*/
static void
atc_chain_complete(struct at_dma_chan *atchan, struct at_desc *desc)
{
struct dma_async_tx_descriptor *txd = &desc->txd;
struct at_dma *atdma = to_at_dma(atchan->chan_common.device);
unsigned long flags;
dev_vdbg(chan2dev(&atchan->chan_common),
"descriptor %u complete\n", txd->cookie);
spin_lock_irqsave(&atchan->lock, flags);
/* mark the descriptor as complete for non cyclic cases only */
if (!atc_chan_is_cyclic(atchan))
dma_cookie_complete(txd);
/* If the transfer was a memset, free our temporary buffer */
if (desc->memset_buffer) {
dma_pool_free(atdma->memset_pool, desc->memset_vaddr,
desc->memset_paddr);
desc->memset_buffer = false;
}
/* move children to free_list */
list_splice_init(&desc->tx_list, &atchan->free_list);
/* move myself to free_list */
list_move(&desc->desc_node, &atchan->free_list);
spin_unlock_irqrestore(&atchan->lock, flags);
dma_descriptor_unmap(txd);
/* for cyclic transfers,
* no need to replay callback function while stopping */
if (!atc_chan_is_cyclic(atchan))
dmaengine_desc_get_callback_invoke(txd, NULL);
dma_run_dependencies(txd);
}
/**
* atc_complete_all - finish work for all transactions
* @atchan: channel to complete transactions for
*
* Eventually submit queued descriptors if any
*
* Assume channel is idle while calling this function
* Called with atchan->lock held and bh disabled
*/
static void atc_complete_all(struct at_dma_chan *atchan)
{
struct at_desc *desc, *_desc;
LIST_HEAD(list);
unsigned long flags;
dev_vdbg(chan2dev(&atchan->chan_common), "complete all\n");
spin_lock_irqsave(&atchan->lock, flags);
/*
* Submit queued descriptors ASAP, i.e. before we go through
* the completed ones.
*/
if (!list_empty(&atchan->queue))
atc_dostart(atchan, atc_first_queued(atchan));
/* empty active_list now it is completed */
list_splice_init(&atchan->active_list, &list);
/* empty queue list by moving descriptors (if any) to active_list */
list_splice_init(&atchan->queue, &atchan->active_list);
spin_unlock_irqrestore(&atchan->lock, flags);
list_for_each_entry_safe(desc, _desc, &list, desc_node)
atc_chain_complete(atchan, desc);
}
/**
* atc_advance_work - at the end of a transaction, move forward
* @atchan: channel where the transaction ended
*/
static void atc_advance_work(struct at_dma_chan *atchan)
{
unsigned long flags;
int ret;
dev_vdbg(chan2dev(&atchan->chan_common), "advance_work\n");
spin_lock_irqsave(&atchan->lock, flags);
ret = atc_chan_is_enabled(atchan);
spin_unlock_irqrestore(&atchan->lock, flags);
if (ret)
return;
if (list_empty(&atchan->active_list) ||
list_is_singular(&atchan->active_list))
return atc_complete_all(atchan);
atc_chain_complete(atchan, atc_first_active(atchan));
/* advance work */
spin_lock_irqsave(&atchan->lock, flags);
atc_dostart(atchan, atc_first_active(atchan));
spin_unlock_irqrestore(&atchan->lock, flags);
}
/**
* atc_handle_error - handle errors reported by DMA controller
* @atchan: channel where error occurs
*/
static void atc_handle_error(struct at_dma_chan *atchan)
{
struct at_desc *bad_desc;
struct at_desc *child;
unsigned long flags;
spin_lock_irqsave(&atchan->lock, flags);
/*
* The descriptor currently at the head of the active list is
* broked. Since we don't have any way to report errors, we'll
* just have to scream loudly and try to carry on.
*/
bad_desc = atc_first_active(atchan);
list_del_init(&bad_desc->desc_node);
/* As we are stopped, take advantage to push queued descriptors
* in active_list */
list_splice_init(&atchan->queue, atchan->active_list.prev);
/* Try to restart the controller */
if (!list_empty(&atchan->active_list))
atc_dostart(atchan, atc_first_active(atchan));
/*
* KERN_CRITICAL may seem harsh, but since this only happens
* when someone submits a bad physical address in a
* descriptor, we should consider ourselves lucky that the
* controller flagged an error instead of scribbling over
* random memory locations.
*/
dev_crit(chan2dev(&atchan->chan_common),
"Bad descriptor submitted for DMA!\n");
dev_crit(chan2dev(&atchan->chan_common),
" cookie: %d\n", bad_desc->txd.cookie);
atc_dump_lli(atchan, &bad_desc->lli);
list_for_each_entry(child, &bad_desc->tx_list, desc_node)
atc_dump_lli(atchan, &child->lli);
spin_unlock_irqrestore(&atchan->lock, flags);
/* Pretend the descriptor completed successfully */
atc_chain_complete(atchan, bad_desc);
}
/**
* atc_handle_cyclic - at the end of a period, run callback function
* @atchan: channel used for cyclic operations
*/
static void atc_handle_cyclic(struct at_dma_chan *atchan)
{
struct at_desc *first = atc_first_active(atchan);
struct dma_async_tx_descriptor *txd = &first->txd;
dev_vdbg(chan2dev(&atchan->chan_common),
"new cyclic period llp 0x%08x\n",
channel_readl(atchan, DSCR));
dmaengine_desc_get_callback_invoke(txd, NULL);
}
/*-- IRQ & Tasklet ---------------------------------------------------*/
static void atc_tasklet(struct tasklet_struct *t)
{
struct at_dma_chan *atchan = from_tasklet(atchan, t, tasklet);
if (test_and_clear_bit(ATC_IS_ERROR, &atchan->status))
return atc_handle_error(atchan);
if (atc_chan_is_cyclic(atchan))
return atc_handle_cyclic(atchan);
atc_advance_work(atchan);
}
static irqreturn_t at_dma_interrupt(int irq, void *dev_id)
{
struct at_dma *atdma = (struct at_dma *)dev_id;
struct at_dma_chan *atchan;
int i;
u32 status, pending, imr;
int ret = IRQ_NONE;
do {
imr = dma_readl(atdma, EBCIMR);
status = dma_readl(atdma, EBCISR);
pending = status & imr;
if (!pending)
break;
dev_vdbg(atdma->dma_common.dev,
"interrupt: status = 0x%08x, 0x%08x, 0x%08x\n",
status, imr, pending);
for (i = 0; i < atdma->dma_common.chancnt; i++) {
atchan = &atdma->chan[i];
if (pending & (AT_DMA_BTC(i) | AT_DMA_ERR(i))) {
if (pending & AT_DMA_ERR(i)) {
/* Disable channel on AHB error */
dma_writel(atdma, CHDR,
AT_DMA_RES(i) | atchan->mask);
/* Give information to tasklet */
set_bit(ATC_IS_ERROR, &atchan->status);
}
tasklet_schedule(&atchan->tasklet);
ret = IRQ_HANDLED;
}
}
} while (pending);
return ret;
}
/*-- DMA Engine API --------------------------------------------------*/
/**
* atc_tx_submit - set the prepared descriptor(s) to be executed by the engine
dmaengine: at_hdmac: Repair parameter misspelling and demote non-kerneldoc headers Fixes the following W=1 kernel build warning(s): drivers/dma/at_hdmac.c:666: warning: Function parameter or member 'tx' not described in 'atc_tx_submit' drivers/dma/at_hdmac.c:666: warning: Excess function parameter 'desc' description in 'atc_tx_submit' drivers/dma/at_hdmac.c:1206: warning: Function parameter or member 'reg_width' not described in 'atc_dma_cyclic_check_values' drivers/dma/at_hdmac.c:1206: warning: Function parameter or member 'buf_addr' not described in 'atc_dma_cyclic_check_values' drivers/dma/at_hdmac.c:1206: warning: Function parameter or member 'period_len' not described in 'atc_dma_cyclic_check_values' drivers/dma/at_hdmac.c:1228: warning: Function parameter or member 'chan' not described in 'atc_dma_cyclic_fill_desc' drivers/dma/at_hdmac.c:1228: warning: Function parameter or member 'desc' not described in 'atc_dma_cyclic_fill_desc' drivers/dma/at_hdmac.c:1228: warning: Function parameter or member 'period_index' not described in 'atc_dma_cyclic_fill_desc' drivers/dma/at_hdmac.c:1228: warning: Function parameter or member 'buf_addr' not described in 'atc_dma_cyclic_fill_desc' drivers/dma/at_hdmac.c:1228: warning: Function parameter or member 'reg_width' not described in 'atc_dma_cyclic_fill_desc' drivers/dma/at_hdmac.c:1228: warning: Function parameter or member 'period_len' not described in 'atc_dma_cyclic_fill_desc' drivers/dma/at_hdmac.c:1228: warning: Function parameter or member 'direction' not described in 'atc_dma_cyclic_fill_desc' Signed-off-by: Lee Jones <lee.jones@linaro.org> Cc: Ludovic Desroches <ludovic.desroches@microchip.com> Cc: Nicolas Ferre <nicolas.ferre@atmel.com> Link: https://lore.kernel.org/r/20200714111546.1755231-9-lee.jones@linaro.org Signed-off-by: Vinod Koul <vkoul@kernel.org>
2020-07-14 19:15:37 +08:00
* @tx: descriptor at the head of the transaction chain
*
* Queue chain if DMA engine is working already
*
* Cookie increment and adding to active_list or queue must be atomic
*/
static dma_cookie_t atc_tx_submit(struct dma_async_tx_descriptor *tx)
{
struct at_desc *desc = txd_to_at_desc(tx);
struct at_dma_chan *atchan = to_at_dma_chan(tx->chan);
dma_cookie_t cookie;
unsigned long flags;
spin_lock_irqsave(&atchan->lock, flags);
cookie = dma_cookie_assign(tx);
if (list_empty(&atchan->active_list)) {
dev_vdbg(chan2dev(tx->chan), "tx_submit: started %u\n",
desc->txd.cookie);
atc_dostart(atchan, desc);
list_add_tail(&desc->desc_node, &atchan->active_list);
} else {
dev_vdbg(chan2dev(tx->chan), "tx_submit: queued %u\n",
desc->txd.cookie);
list_add_tail(&desc->desc_node, &atchan->queue);
}
spin_unlock_irqrestore(&atchan->lock, flags);
return cookie;
}
/**
* atc_prep_dma_interleaved - prepare memory to memory interleaved operation
* @chan: the channel to prepare operation on
* @xt: Interleaved transfer template
* @flags: tx descriptor status flags
*/
static struct dma_async_tx_descriptor *
atc_prep_dma_interleaved(struct dma_chan *chan,
struct dma_interleaved_template *xt,
unsigned long flags)
{
struct at_dma_chan *atchan = to_at_dma_chan(chan);
struct data_chunk *first;
struct at_desc *desc = NULL;
size_t xfer_count;
unsigned int dwidth;
u32 ctrla;
u32 ctrlb;
size_t len = 0;
int i;
if (unlikely(!xt || xt->numf != 1 || !xt->frame_size))
return NULL;
first = xt->sgl;
dev_info(chan2dev(chan),
"%s: src=%pad, dest=%pad, numf=%d, frame_size=%d, flags=0x%lx\n",
__func__, &xt->src_start, &xt->dst_start, xt->numf,
xt->frame_size, flags);
/*
* The controller can only "skip" X bytes every Y bytes, so we
* need to make sure we are given a template that fit that
* description, ie a template with chunks that always have the
* same size, with the same ICGs.
*/
for (i = 0; i < xt->frame_size; i++) {
struct data_chunk *chunk = xt->sgl + i;
if ((chunk->size != xt->sgl->size) ||
(dmaengine_get_dst_icg(xt, chunk) != dmaengine_get_dst_icg(xt, first)) ||
(dmaengine_get_src_icg(xt, chunk) != dmaengine_get_src_icg(xt, first))) {
dev_err(chan2dev(chan),
"%s: the controller can transfer only identical chunks\n",
__func__);
return NULL;
}
len += chunk->size;
}
dwidth = atc_get_xfer_width(xt->src_start,
xt->dst_start, len);
xfer_count = len >> dwidth;
if (xfer_count > ATC_BTSIZE_MAX) {
dev_err(chan2dev(chan), "%s: buffer is too big\n", __func__);
return NULL;
}
ctrla = ATC_SRC_WIDTH(dwidth) |
ATC_DST_WIDTH(dwidth);
ctrlb = ATC_DEFAULT_CTRLB | ATC_IEN
| ATC_SRC_ADDR_MODE_INCR
| ATC_DST_ADDR_MODE_INCR
| ATC_SRC_PIP
| ATC_DST_PIP
| ATC_FC_MEM2MEM;
/* create the transfer */
desc = atc_desc_get(atchan);
if (!desc) {
dev_err(chan2dev(chan),
"%s: couldn't allocate our descriptor\n", __func__);
return NULL;
}
desc->lli.saddr = xt->src_start;
desc->lli.daddr = xt->dst_start;
desc->lli.ctrla = ctrla | xfer_count;
desc->lli.ctrlb = ctrlb;
desc->boundary = first->size >> dwidth;
desc->dst_hole = (dmaengine_get_dst_icg(xt, first) >> dwidth) + 1;
desc->src_hole = (dmaengine_get_src_icg(xt, first) >> dwidth) + 1;
desc->txd.cookie = -EBUSY;
desc->total_len = desc->len = len;
/* set end-of-link to the last link descriptor of list*/
set_desc_eol(desc);
desc->txd.flags = flags; /* client is in control of this ack */
return &desc->txd;
}
/**
* atc_prep_dma_memcpy - prepare a memcpy operation
* @chan: the channel to prepare operation on
* @dest: operation virtual destination address
* @src: operation virtual source address
* @len: operation length
* @flags: tx descriptor status flags
*/
static struct dma_async_tx_descriptor *
atc_prep_dma_memcpy(struct dma_chan *chan, dma_addr_t dest, dma_addr_t src,
size_t len, unsigned long flags)
{
struct at_dma_chan *atchan = to_at_dma_chan(chan);
struct at_desc *desc = NULL;
struct at_desc *first = NULL;
struct at_desc *prev = NULL;
size_t xfer_count;
size_t offset;
unsigned int src_width;
unsigned int dst_width;
u32 ctrla;
u32 ctrlb;
dev_vdbg(chan2dev(chan), "prep_dma_memcpy: d%pad s%pad l0x%zx f0x%lx\n",
&dest, &src, len, flags);
if (unlikely(!len)) {
dev_dbg(chan2dev(chan), "prep_dma_memcpy: length is zero!\n");
return NULL;
}
ctrlb = ATC_DEFAULT_CTRLB | ATC_IEN
| ATC_SRC_ADDR_MODE_INCR
| ATC_DST_ADDR_MODE_INCR
| ATC_FC_MEM2MEM;
/*
* We can be a lot more clever here, but this should take care
* of the most common optimization.
*/
src_width = dst_width = atc_get_xfer_width(src, dest, len);
ctrla = ATC_SRC_WIDTH(src_width) |
ATC_DST_WIDTH(dst_width);
for (offset = 0; offset < len; offset += xfer_count << src_width) {
xfer_count = min_t(size_t, (len - offset) >> src_width,
ATC_BTSIZE_MAX);
desc = atc_desc_get(atchan);
if (!desc)
goto err_desc_get;
desc->lli.saddr = src + offset;
desc->lli.daddr = dest + offset;
desc->lli.ctrla = ctrla | xfer_count;
desc->lli.ctrlb = ctrlb;
desc->txd.cookie = 0;
dmaengine: at_hdmac: Fix calculation of the residual bytes This patch fixes the following issues regarding to the calculation of the residue: 1. The residue is always calculated for the current transfer even if the cookie is associated to a pending transfer. 2. For scatter/gather DMA the calculation of the residue for the current transfer doesn't include the bytes of the child descriptors that are already transferred. It only calculates the difference between the transfer's total length minus the number of bytes that are already transferred for the current child descriptor. For example: There is a scatter/gather DMA transfer with a total length of 1 MByte. Getting the residue several times while the transfer is running shows something like that: 1: residue = 975584 2: residue = 1002766 3: residue = 992627 4: residue = 983767 5: residue = 985694 6: residue = 1008094 7: residue = 1009741 8: residue = 1011195 3. The driver stores the residue but never resets it when starting a new transfer. For example: If there are two subsequent DMA transfers. The first one with a total length of 1 MByte and the second one with a total length of 1 kByte. Getting the residue for both transfers shows something like that: transfer 1: residue = 975584 transfer 2: residue = 1048380 Changes from V1: * Fixed coding style of the multi-line comments. * Improved accuracy of the residue calculation when the transfer for the first descriptor is active. Changes from V2: * Member 'tx_width' of 'struct at_desc' restored, because the transfer width can't be derived from the source width when using "slave_sg". The transfer width is needed for the calculation of the residue if either the transfer of the first or the last descriptor is in progress. In the case of a "memory_to_memory_sg" transfer (part of this patch series) the transfer width of both descriptors may differ. Thus it is required to additionally set 'tx_width' of the last descriptor. * Added functions for multiply used calculations. Signed-off-by: Torsten Fleischer <torfl6749@gmail.com> Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2015-02-24 00:54:10 +08:00
desc->len = xfer_count << src_width;
atc_desc_chain(&first, &prev, desc);
}
/* First descriptor of the chain embedds additional information */
first->txd.cookie = -EBUSY;
dmaengine: at_hdmac: Fix calculation of the residual bytes This patch fixes the following issues regarding to the calculation of the residue: 1. The residue is always calculated for the current transfer even if the cookie is associated to a pending transfer. 2. For scatter/gather DMA the calculation of the residue for the current transfer doesn't include the bytes of the child descriptors that are already transferred. It only calculates the difference between the transfer's total length minus the number of bytes that are already transferred for the current child descriptor. For example: There is a scatter/gather DMA transfer with a total length of 1 MByte. Getting the residue several times while the transfer is running shows something like that: 1: residue = 975584 2: residue = 1002766 3: residue = 992627 4: residue = 983767 5: residue = 985694 6: residue = 1008094 7: residue = 1009741 8: residue = 1011195 3. The driver stores the residue but never resets it when starting a new transfer. For example: If there are two subsequent DMA transfers. The first one with a total length of 1 MByte and the second one with a total length of 1 kByte. Getting the residue for both transfers shows something like that: transfer 1: residue = 975584 transfer 2: residue = 1048380 Changes from V1: * Fixed coding style of the multi-line comments. * Improved accuracy of the residue calculation when the transfer for the first descriptor is active. Changes from V2: * Member 'tx_width' of 'struct at_desc' restored, because the transfer width can't be derived from the source width when using "slave_sg". The transfer width is needed for the calculation of the residue if either the transfer of the first or the last descriptor is in progress. In the case of a "memory_to_memory_sg" transfer (part of this patch series) the transfer width of both descriptors may differ. Thus it is required to additionally set 'tx_width' of the last descriptor. * Added functions for multiply used calculations. Signed-off-by: Torsten Fleischer <torfl6749@gmail.com> Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2015-02-24 00:54:10 +08:00
first->total_len = len;
/* set end-of-link to the last link descriptor of list*/
set_desc_eol(desc);
first->txd.flags = flags; /* client is in control of this ack */
return &first->txd;
err_desc_get:
atc_desc_put(atchan, first);
return NULL;
}
static struct at_desc *atc_create_memset_desc(struct dma_chan *chan,
dma_addr_t psrc,
dma_addr_t pdst,
size_t len)
{
struct at_dma_chan *atchan = to_at_dma_chan(chan);
struct at_desc *desc;
size_t xfer_count;
u32 ctrla = ATC_SRC_WIDTH(2) | ATC_DST_WIDTH(2);
u32 ctrlb = ATC_DEFAULT_CTRLB | ATC_IEN |
ATC_SRC_ADDR_MODE_FIXED |
ATC_DST_ADDR_MODE_INCR |
ATC_FC_MEM2MEM;
xfer_count = len >> 2;
if (xfer_count > ATC_BTSIZE_MAX) {
dev_err(chan2dev(chan), "%s: buffer is too big\n",
__func__);
return NULL;
}
desc = atc_desc_get(atchan);
if (!desc) {
dev_err(chan2dev(chan), "%s: can't get a descriptor\n",
__func__);
return NULL;
}
desc->lli.saddr = psrc;
desc->lli.daddr = pdst;
desc->lli.ctrla = ctrla | xfer_count;
desc->lli.ctrlb = ctrlb;
desc->txd.cookie = 0;
desc->len = len;
return desc;
}
/**
* atc_prep_dma_memset - prepare a memcpy operation
* @chan: the channel to prepare operation on
* @dest: operation virtual destination address
* @value: value to set memory buffer to
* @len: operation length
* @flags: tx descriptor status flags
*/
static struct dma_async_tx_descriptor *
atc_prep_dma_memset(struct dma_chan *chan, dma_addr_t dest, int value,
size_t len, unsigned long flags)
{
struct at_dma *atdma = to_at_dma(chan->device);
struct at_desc *desc;
void __iomem *vaddr;
dma_addr_t paddr;
dev_vdbg(chan2dev(chan), "%s: d%pad v0x%x l0x%zx f0x%lx\n", __func__,
&dest, value, len, flags);
if (unlikely(!len)) {
dev_dbg(chan2dev(chan), "%s: length is zero!\n", __func__);
return NULL;
}
if (!is_dma_fill_aligned(chan->device, dest, 0, len)) {
dev_dbg(chan2dev(chan), "%s: buffer is not aligned\n",
__func__);
return NULL;
}
vaddr = dma_pool_alloc(atdma->memset_pool, GFP_NOWAIT, &paddr);
if (!vaddr) {
dev_err(chan2dev(chan), "%s: couldn't allocate buffer\n",
__func__);
return NULL;
}
*(u32*)vaddr = value;
desc = atc_create_memset_desc(chan, paddr, dest, len);
if (!desc) {
dev_err(chan2dev(chan), "%s: couldn't get a descriptor\n",
__func__);
goto err_free_buffer;
}
desc->memset_paddr = paddr;
desc->memset_vaddr = vaddr;
desc->memset_buffer = true;
desc->txd.cookie = -EBUSY;
desc->total_len = len;
/* set end-of-link on the descriptor */
set_desc_eol(desc);
desc->txd.flags = flags;
return &desc->txd;
err_free_buffer:
dma_pool_free(atdma->memset_pool, vaddr, paddr);
return NULL;
}
static struct dma_async_tx_descriptor *
atc_prep_dma_memset_sg(struct dma_chan *chan,
struct scatterlist *sgl,
unsigned int sg_len, int value,
unsigned long flags)
{
struct at_dma_chan *atchan = to_at_dma_chan(chan);
struct at_dma *atdma = to_at_dma(chan->device);
struct at_desc *desc = NULL, *first = NULL, *prev = NULL;
struct scatterlist *sg;
void __iomem *vaddr;
dma_addr_t paddr;
size_t total_len = 0;
int i;
dev_vdbg(chan2dev(chan), "%s: v0x%x l0x%zx f0x%lx\n", __func__,
value, sg_len, flags);
if (unlikely(!sgl || !sg_len)) {
dev_dbg(chan2dev(chan), "%s: scatterlist is empty!\n",
__func__);
return NULL;
}
vaddr = dma_pool_alloc(atdma->memset_pool, GFP_NOWAIT, &paddr);
if (!vaddr) {
dev_err(chan2dev(chan), "%s: couldn't allocate buffer\n",
__func__);
return NULL;
}
*(u32*)vaddr = value;
for_each_sg(sgl, sg, sg_len, i) {
dma_addr_t dest = sg_dma_address(sg);
size_t len = sg_dma_len(sg);
dev_vdbg(chan2dev(chan), "%s: d%pad, l0x%zx\n",
__func__, &dest, len);
if (!is_dma_fill_aligned(chan->device, dest, 0, len)) {
dev_err(chan2dev(chan), "%s: buffer is not aligned\n",
__func__);
goto err_put_desc;
}
desc = atc_create_memset_desc(chan, paddr, dest, len);
if (!desc)
goto err_put_desc;
atc_desc_chain(&first, &prev, desc);
total_len += len;
}
/*
* Only set the buffer pointers on the last descriptor to
* avoid free'ing while we have our transfer still going
*/
desc->memset_paddr = paddr;
desc->memset_vaddr = vaddr;
desc->memset_buffer = true;
first->txd.cookie = -EBUSY;
first->total_len = total_len;
/* set end-of-link on the descriptor */
set_desc_eol(desc);
first->txd.flags = flags;
return &first->txd;
err_put_desc:
atc_desc_put(atchan, first);
return NULL;
}
/**
* atc_prep_slave_sg - prepare descriptors for a DMA_SLAVE transaction
* @chan: DMA channel
* @sgl: scatterlist to transfer to/from
* @sg_len: number of entries in @scatterlist
* @direction: DMA direction
* @flags: tx descriptor status flags
* @context: transaction context (ignored)
*/
static struct dma_async_tx_descriptor *
atc_prep_slave_sg(struct dma_chan *chan, struct scatterlist *sgl,
unsigned int sg_len, enum dma_transfer_direction direction,
unsigned long flags, void *context)
{
struct at_dma_chan *atchan = to_at_dma_chan(chan);
struct at_dma_slave *atslave = chan->private;
struct dma_slave_config *sconfig = &atchan->dma_sconfig;
struct at_desc *first = NULL;
struct at_desc *prev = NULL;
u32 ctrla;
u32 ctrlb;
dma_addr_t reg;
unsigned int reg_width;
unsigned int mem_width;
unsigned int i;
struct scatterlist *sg;
size_t total_len = 0;
dev_vdbg(chan2dev(chan), "prep_slave_sg (%d): %s f0x%lx\n",
sg_len,
direction == DMA_MEM_TO_DEV ? "TO DEVICE" : "FROM DEVICE",
flags);
if (unlikely(!atslave || !sg_len)) {
dev_dbg(chan2dev(chan), "prep_slave_sg: sg length is zero!\n");
return NULL;
}
ctrla = ATC_SCSIZE(sconfig->src_maxburst)
| ATC_DCSIZE(sconfig->dst_maxburst);
ctrlb = ATC_IEN;
switch (direction) {
case DMA_MEM_TO_DEV:
reg_width = convert_buswidth(sconfig->dst_addr_width);
ctrla |= ATC_DST_WIDTH(reg_width);
ctrlb |= ATC_DST_ADDR_MODE_FIXED
| ATC_SRC_ADDR_MODE_INCR
| ATC_FC_MEM2PER
| ATC_SIF(atchan->mem_if) | ATC_DIF(atchan->per_if);
reg = sconfig->dst_addr;
for_each_sg(sgl, sg, sg_len, i) {
struct at_desc *desc;
u32 len;
u32 mem;
desc = atc_desc_get(atchan);
if (!desc)
goto err_desc_get;
mem = sg_dma_address(sg);
len = sg_dma_len(sg);
if (unlikely(!len)) {
dev_dbg(chan2dev(chan),
"prep_slave_sg: sg(%d) data length is zero\n", i);
goto err;
}
mem_width = 2;
if (unlikely(mem & 3 || len & 3))
mem_width = 0;
desc->lli.saddr = mem;
desc->lli.daddr = reg;
desc->lli.ctrla = ctrla
| ATC_SRC_WIDTH(mem_width)
| len >> mem_width;
desc->lli.ctrlb = ctrlb;
dmaengine: at_hdmac: Fix calculation of the residual bytes This patch fixes the following issues regarding to the calculation of the residue: 1. The residue is always calculated for the current transfer even if the cookie is associated to a pending transfer. 2. For scatter/gather DMA the calculation of the residue for the current transfer doesn't include the bytes of the child descriptors that are already transferred. It only calculates the difference between the transfer's total length minus the number of bytes that are already transferred for the current child descriptor. For example: There is a scatter/gather DMA transfer with a total length of 1 MByte. Getting the residue several times while the transfer is running shows something like that: 1: residue = 975584 2: residue = 1002766 3: residue = 992627 4: residue = 983767 5: residue = 985694 6: residue = 1008094 7: residue = 1009741 8: residue = 1011195 3. The driver stores the residue but never resets it when starting a new transfer. For example: If there are two subsequent DMA transfers. The first one with a total length of 1 MByte and the second one with a total length of 1 kByte. Getting the residue for both transfers shows something like that: transfer 1: residue = 975584 transfer 2: residue = 1048380 Changes from V1: * Fixed coding style of the multi-line comments. * Improved accuracy of the residue calculation when the transfer for the first descriptor is active. Changes from V2: * Member 'tx_width' of 'struct at_desc' restored, because the transfer width can't be derived from the source width when using "slave_sg". The transfer width is needed for the calculation of the residue if either the transfer of the first or the last descriptor is in progress. In the case of a "memory_to_memory_sg" transfer (part of this patch series) the transfer width of both descriptors may differ. Thus it is required to additionally set 'tx_width' of the last descriptor. * Added functions for multiply used calculations. Signed-off-by: Torsten Fleischer <torfl6749@gmail.com> Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2015-02-24 00:54:10 +08:00
desc->len = len;
atc_desc_chain(&first, &prev, desc);
total_len += len;
}
break;
case DMA_DEV_TO_MEM:
reg_width = convert_buswidth(sconfig->src_addr_width);
ctrla |= ATC_SRC_WIDTH(reg_width);
ctrlb |= ATC_DST_ADDR_MODE_INCR
| ATC_SRC_ADDR_MODE_FIXED
| ATC_FC_PER2MEM
| ATC_SIF(atchan->per_if) | ATC_DIF(atchan->mem_if);
reg = sconfig->src_addr;
for_each_sg(sgl, sg, sg_len, i) {
struct at_desc *desc;
u32 len;
u32 mem;
desc = atc_desc_get(atchan);
if (!desc)
goto err_desc_get;
mem = sg_dma_address(sg);
len = sg_dma_len(sg);
if (unlikely(!len)) {
dev_dbg(chan2dev(chan),
"prep_slave_sg: sg(%d) data length is zero\n", i);
goto err;
}
mem_width = 2;
if (unlikely(mem & 3 || len & 3))
mem_width = 0;
desc->lli.saddr = reg;
desc->lli.daddr = mem;
desc->lli.ctrla = ctrla
| ATC_DST_WIDTH(mem_width)
| len >> reg_width;
desc->lli.ctrlb = ctrlb;
dmaengine: at_hdmac: Fix calculation of the residual bytes This patch fixes the following issues regarding to the calculation of the residue: 1. The residue is always calculated for the current transfer even if the cookie is associated to a pending transfer. 2. For scatter/gather DMA the calculation of the residue for the current transfer doesn't include the bytes of the child descriptors that are already transferred. It only calculates the difference between the transfer's total length minus the number of bytes that are already transferred for the current child descriptor. For example: There is a scatter/gather DMA transfer with a total length of 1 MByte. Getting the residue several times while the transfer is running shows something like that: 1: residue = 975584 2: residue = 1002766 3: residue = 992627 4: residue = 983767 5: residue = 985694 6: residue = 1008094 7: residue = 1009741 8: residue = 1011195 3. The driver stores the residue but never resets it when starting a new transfer. For example: If there are two subsequent DMA transfers. The first one with a total length of 1 MByte and the second one with a total length of 1 kByte. Getting the residue for both transfers shows something like that: transfer 1: residue = 975584 transfer 2: residue = 1048380 Changes from V1: * Fixed coding style of the multi-line comments. * Improved accuracy of the residue calculation when the transfer for the first descriptor is active. Changes from V2: * Member 'tx_width' of 'struct at_desc' restored, because the transfer width can't be derived from the source width when using "slave_sg". The transfer width is needed for the calculation of the residue if either the transfer of the first or the last descriptor is in progress. In the case of a "memory_to_memory_sg" transfer (part of this patch series) the transfer width of both descriptors may differ. Thus it is required to additionally set 'tx_width' of the last descriptor. * Added functions for multiply used calculations. Signed-off-by: Torsten Fleischer <torfl6749@gmail.com> Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2015-02-24 00:54:10 +08:00
desc->len = len;
atc_desc_chain(&first, &prev, desc);
total_len += len;
}
break;
default:
return NULL;
}
/* set end-of-link to the last link descriptor of list*/
set_desc_eol(prev);
/* First descriptor of the chain embedds additional information */
first->txd.cookie = -EBUSY;
dmaengine: at_hdmac: Fix calculation of the residual bytes This patch fixes the following issues regarding to the calculation of the residue: 1. The residue is always calculated for the current transfer even if the cookie is associated to a pending transfer. 2. For scatter/gather DMA the calculation of the residue for the current transfer doesn't include the bytes of the child descriptors that are already transferred. It only calculates the difference between the transfer's total length minus the number of bytes that are already transferred for the current child descriptor. For example: There is a scatter/gather DMA transfer with a total length of 1 MByte. Getting the residue several times while the transfer is running shows something like that: 1: residue = 975584 2: residue = 1002766 3: residue = 992627 4: residue = 983767 5: residue = 985694 6: residue = 1008094 7: residue = 1009741 8: residue = 1011195 3. The driver stores the residue but never resets it when starting a new transfer. For example: If there are two subsequent DMA transfers. The first one with a total length of 1 MByte and the second one with a total length of 1 kByte. Getting the residue for both transfers shows something like that: transfer 1: residue = 975584 transfer 2: residue = 1048380 Changes from V1: * Fixed coding style of the multi-line comments. * Improved accuracy of the residue calculation when the transfer for the first descriptor is active. Changes from V2: * Member 'tx_width' of 'struct at_desc' restored, because the transfer width can't be derived from the source width when using "slave_sg". The transfer width is needed for the calculation of the residue if either the transfer of the first or the last descriptor is in progress. In the case of a "memory_to_memory_sg" transfer (part of this patch series) the transfer width of both descriptors may differ. Thus it is required to additionally set 'tx_width' of the last descriptor. * Added functions for multiply used calculations. Signed-off-by: Torsten Fleischer <torfl6749@gmail.com> Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2015-02-24 00:54:10 +08:00
first->total_len = total_len;
/* first link descriptor of list is responsible of flags */
first->txd.flags = flags; /* client is in control of this ack */
return &first->txd;
err_desc_get:
dev_err(chan2dev(chan), "not enough descriptors available\n");
err:
atc_desc_put(atchan, first);
return NULL;
}
dmaengine: at_hdmac: Repair parameter misspelling and demote non-kerneldoc headers Fixes the following W=1 kernel build warning(s): drivers/dma/at_hdmac.c:666: warning: Function parameter or member 'tx' not described in 'atc_tx_submit' drivers/dma/at_hdmac.c:666: warning: Excess function parameter 'desc' description in 'atc_tx_submit' drivers/dma/at_hdmac.c:1206: warning: Function parameter or member 'reg_width' not described in 'atc_dma_cyclic_check_values' drivers/dma/at_hdmac.c:1206: warning: Function parameter or member 'buf_addr' not described in 'atc_dma_cyclic_check_values' drivers/dma/at_hdmac.c:1206: warning: Function parameter or member 'period_len' not described in 'atc_dma_cyclic_check_values' drivers/dma/at_hdmac.c:1228: warning: Function parameter or member 'chan' not described in 'atc_dma_cyclic_fill_desc' drivers/dma/at_hdmac.c:1228: warning: Function parameter or member 'desc' not described in 'atc_dma_cyclic_fill_desc' drivers/dma/at_hdmac.c:1228: warning: Function parameter or member 'period_index' not described in 'atc_dma_cyclic_fill_desc' drivers/dma/at_hdmac.c:1228: warning: Function parameter or member 'buf_addr' not described in 'atc_dma_cyclic_fill_desc' drivers/dma/at_hdmac.c:1228: warning: Function parameter or member 'reg_width' not described in 'atc_dma_cyclic_fill_desc' drivers/dma/at_hdmac.c:1228: warning: Function parameter or member 'period_len' not described in 'atc_dma_cyclic_fill_desc' drivers/dma/at_hdmac.c:1228: warning: Function parameter or member 'direction' not described in 'atc_dma_cyclic_fill_desc' Signed-off-by: Lee Jones <lee.jones@linaro.org> Cc: Ludovic Desroches <ludovic.desroches@microchip.com> Cc: Nicolas Ferre <nicolas.ferre@atmel.com> Link: https://lore.kernel.org/r/20200714111546.1755231-9-lee.jones@linaro.org Signed-off-by: Vinod Koul <vkoul@kernel.org>
2020-07-14 19:15:37 +08:00
/*
* atc_dma_cyclic_check_values
* Check for too big/unaligned periods and unaligned DMA buffer
*/
static int
atc_dma_cyclic_check_values(unsigned int reg_width, dma_addr_t buf_addr,
size_t period_len)
{
if (period_len > (ATC_BTSIZE_MAX << reg_width))
goto err_out;
if (unlikely(period_len & ((1 << reg_width) - 1)))
goto err_out;
if (unlikely(buf_addr & ((1 << reg_width) - 1)))
goto err_out;
return 0;
err_out:
return -EINVAL;
}
dmaengine: at_hdmac: Repair parameter misspelling and demote non-kerneldoc headers Fixes the following W=1 kernel build warning(s): drivers/dma/at_hdmac.c:666: warning: Function parameter or member 'tx' not described in 'atc_tx_submit' drivers/dma/at_hdmac.c:666: warning: Excess function parameter 'desc' description in 'atc_tx_submit' drivers/dma/at_hdmac.c:1206: warning: Function parameter or member 'reg_width' not described in 'atc_dma_cyclic_check_values' drivers/dma/at_hdmac.c:1206: warning: Function parameter or member 'buf_addr' not described in 'atc_dma_cyclic_check_values' drivers/dma/at_hdmac.c:1206: warning: Function parameter or member 'period_len' not described in 'atc_dma_cyclic_check_values' drivers/dma/at_hdmac.c:1228: warning: Function parameter or member 'chan' not described in 'atc_dma_cyclic_fill_desc' drivers/dma/at_hdmac.c:1228: warning: Function parameter or member 'desc' not described in 'atc_dma_cyclic_fill_desc' drivers/dma/at_hdmac.c:1228: warning: Function parameter or member 'period_index' not described in 'atc_dma_cyclic_fill_desc' drivers/dma/at_hdmac.c:1228: warning: Function parameter or member 'buf_addr' not described in 'atc_dma_cyclic_fill_desc' drivers/dma/at_hdmac.c:1228: warning: Function parameter or member 'reg_width' not described in 'atc_dma_cyclic_fill_desc' drivers/dma/at_hdmac.c:1228: warning: Function parameter or member 'period_len' not described in 'atc_dma_cyclic_fill_desc' drivers/dma/at_hdmac.c:1228: warning: Function parameter or member 'direction' not described in 'atc_dma_cyclic_fill_desc' Signed-off-by: Lee Jones <lee.jones@linaro.org> Cc: Ludovic Desroches <ludovic.desroches@microchip.com> Cc: Nicolas Ferre <nicolas.ferre@atmel.com> Link: https://lore.kernel.org/r/20200714111546.1755231-9-lee.jones@linaro.org Signed-off-by: Vinod Koul <vkoul@kernel.org>
2020-07-14 19:15:37 +08:00
/*
* atc_dma_cyclic_fill_desc - Fill one period descriptor
*/
static int
atc_dma_cyclic_fill_desc(struct dma_chan *chan, struct at_desc *desc,
unsigned int period_index, dma_addr_t buf_addr,
unsigned int reg_width, size_t period_len,
enum dma_transfer_direction direction)
{
struct at_dma_chan *atchan = to_at_dma_chan(chan);
struct dma_slave_config *sconfig = &atchan->dma_sconfig;
u32 ctrla;
/* prepare common CRTLA value */
ctrla = ATC_SCSIZE(sconfig->src_maxburst)
| ATC_DCSIZE(sconfig->dst_maxburst)
| ATC_DST_WIDTH(reg_width)
| ATC_SRC_WIDTH(reg_width)
| period_len >> reg_width;
switch (direction) {
case DMA_MEM_TO_DEV:
desc->lli.saddr = buf_addr + (period_len * period_index);
desc->lli.daddr = sconfig->dst_addr;
desc->lli.ctrla = ctrla;
desc->lli.ctrlb = ATC_DST_ADDR_MODE_FIXED
| ATC_SRC_ADDR_MODE_INCR
| ATC_FC_MEM2PER
| ATC_SIF(atchan->mem_if)
| ATC_DIF(atchan->per_if);
dmaengine: at_hdmac: Fix calculation of the residual bytes This patch fixes the following issues regarding to the calculation of the residue: 1. The residue is always calculated for the current transfer even if the cookie is associated to a pending transfer. 2. For scatter/gather DMA the calculation of the residue for the current transfer doesn't include the bytes of the child descriptors that are already transferred. It only calculates the difference between the transfer's total length minus the number of bytes that are already transferred for the current child descriptor. For example: There is a scatter/gather DMA transfer with a total length of 1 MByte. Getting the residue several times while the transfer is running shows something like that: 1: residue = 975584 2: residue = 1002766 3: residue = 992627 4: residue = 983767 5: residue = 985694 6: residue = 1008094 7: residue = 1009741 8: residue = 1011195 3. The driver stores the residue but never resets it when starting a new transfer. For example: If there are two subsequent DMA transfers. The first one with a total length of 1 MByte and the second one with a total length of 1 kByte. Getting the residue for both transfers shows something like that: transfer 1: residue = 975584 transfer 2: residue = 1048380 Changes from V1: * Fixed coding style of the multi-line comments. * Improved accuracy of the residue calculation when the transfer for the first descriptor is active. Changes from V2: * Member 'tx_width' of 'struct at_desc' restored, because the transfer width can't be derived from the source width when using "slave_sg". The transfer width is needed for the calculation of the residue if either the transfer of the first or the last descriptor is in progress. In the case of a "memory_to_memory_sg" transfer (part of this patch series) the transfer width of both descriptors may differ. Thus it is required to additionally set 'tx_width' of the last descriptor. * Added functions for multiply used calculations. Signed-off-by: Torsten Fleischer <torfl6749@gmail.com> Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2015-02-24 00:54:10 +08:00
desc->len = period_len;
break;
case DMA_DEV_TO_MEM:
desc->lli.saddr = sconfig->src_addr;
desc->lli.daddr = buf_addr + (period_len * period_index);
desc->lli.ctrla = ctrla;
desc->lli.ctrlb = ATC_DST_ADDR_MODE_INCR
| ATC_SRC_ADDR_MODE_FIXED
| ATC_FC_PER2MEM
| ATC_SIF(atchan->per_if)
| ATC_DIF(atchan->mem_if);
dmaengine: at_hdmac: Fix calculation of the residual bytes This patch fixes the following issues regarding to the calculation of the residue: 1. The residue is always calculated for the current transfer even if the cookie is associated to a pending transfer. 2. For scatter/gather DMA the calculation of the residue for the current transfer doesn't include the bytes of the child descriptors that are already transferred. It only calculates the difference between the transfer's total length minus the number of bytes that are already transferred for the current child descriptor. For example: There is a scatter/gather DMA transfer with a total length of 1 MByte. Getting the residue several times while the transfer is running shows something like that: 1: residue = 975584 2: residue = 1002766 3: residue = 992627 4: residue = 983767 5: residue = 985694 6: residue = 1008094 7: residue = 1009741 8: residue = 1011195 3. The driver stores the residue but never resets it when starting a new transfer. For example: If there are two subsequent DMA transfers. The first one with a total length of 1 MByte and the second one with a total length of 1 kByte. Getting the residue for both transfers shows something like that: transfer 1: residue = 975584 transfer 2: residue = 1048380 Changes from V1: * Fixed coding style of the multi-line comments. * Improved accuracy of the residue calculation when the transfer for the first descriptor is active. Changes from V2: * Member 'tx_width' of 'struct at_desc' restored, because the transfer width can't be derived from the source width when using "slave_sg". The transfer width is needed for the calculation of the residue if either the transfer of the first or the last descriptor is in progress. In the case of a "memory_to_memory_sg" transfer (part of this patch series) the transfer width of both descriptors may differ. Thus it is required to additionally set 'tx_width' of the last descriptor. * Added functions for multiply used calculations. Signed-off-by: Torsten Fleischer <torfl6749@gmail.com> Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2015-02-24 00:54:10 +08:00
desc->len = period_len;
break;
default:
return -EINVAL;
}
return 0;
}
/**
* atc_prep_dma_cyclic - prepare the cyclic DMA transfer
* @chan: the DMA channel to prepare
* @buf_addr: physical DMA address where the buffer starts
* @buf_len: total number of bytes for the entire buffer
* @period_len: number of bytes for each period
* @direction: transfer direction, to or from device
* @flags: tx descriptor status flags
*/
static struct dma_async_tx_descriptor *
atc_prep_dma_cyclic(struct dma_chan *chan, dma_addr_t buf_addr, size_t buf_len,
size_t period_len, enum dma_transfer_direction direction,
unsigned long flags)
{
struct at_dma_chan *atchan = to_at_dma_chan(chan);
struct at_dma_slave *atslave = chan->private;
struct dma_slave_config *sconfig = &atchan->dma_sconfig;
struct at_desc *first = NULL;
struct at_desc *prev = NULL;
unsigned long was_cyclic;
unsigned int reg_width;
unsigned int periods = buf_len / period_len;
unsigned int i;
dev_vdbg(chan2dev(chan), "prep_dma_cyclic: %s buf@%pad - %d (%d/%d)\n",
direction == DMA_MEM_TO_DEV ? "TO DEVICE" : "FROM DEVICE",
&buf_addr,
periods, buf_len, period_len);
if (unlikely(!atslave || !buf_len || !period_len)) {
dev_dbg(chan2dev(chan), "prep_dma_cyclic: length is zero!\n");
return NULL;
}
was_cyclic = test_and_set_bit(ATC_IS_CYCLIC, &atchan->status);
if (was_cyclic) {
dev_dbg(chan2dev(chan), "prep_dma_cyclic: channel in use!\n");
return NULL;
}
if (unlikely(!is_slave_direction(direction)))
goto err_out;
if (direction == DMA_MEM_TO_DEV)
reg_width = convert_buswidth(sconfig->dst_addr_width);
else
reg_width = convert_buswidth(sconfig->src_addr_width);
/* Check for too big/unaligned periods and unaligned DMA buffer */
if (atc_dma_cyclic_check_values(reg_width, buf_addr, period_len))
goto err_out;
/* build cyclic linked list */
for (i = 0; i < periods; i++) {
struct at_desc *desc;
desc = atc_desc_get(atchan);
if (!desc)
goto err_desc_get;
if (atc_dma_cyclic_fill_desc(chan, desc, i, buf_addr,
reg_width, period_len, direction))
goto err_desc_get;
atc_desc_chain(&first, &prev, desc);
}
/* lets make a cyclic list */
prev->lli.dscr = first->txd.phys;
/* First descriptor of the chain embedds additional information */
first->txd.cookie = -EBUSY;
dmaengine: at_hdmac: Fix calculation of the residual bytes This patch fixes the following issues regarding to the calculation of the residue: 1. The residue is always calculated for the current transfer even if the cookie is associated to a pending transfer. 2. For scatter/gather DMA the calculation of the residue for the current transfer doesn't include the bytes of the child descriptors that are already transferred. It only calculates the difference between the transfer's total length minus the number of bytes that are already transferred for the current child descriptor. For example: There is a scatter/gather DMA transfer with a total length of 1 MByte. Getting the residue several times while the transfer is running shows something like that: 1: residue = 975584 2: residue = 1002766 3: residue = 992627 4: residue = 983767 5: residue = 985694 6: residue = 1008094 7: residue = 1009741 8: residue = 1011195 3. The driver stores the residue but never resets it when starting a new transfer. For example: If there are two subsequent DMA transfers. The first one with a total length of 1 MByte and the second one with a total length of 1 kByte. Getting the residue for both transfers shows something like that: transfer 1: residue = 975584 transfer 2: residue = 1048380 Changes from V1: * Fixed coding style of the multi-line comments. * Improved accuracy of the residue calculation when the transfer for the first descriptor is active. Changes from V2: * Member 'tx_width' of 'struct at_desc' restored, because the transfer width can't be derived from the source width when using "slave_sg". The transfer width is needed for the calculation of the residue if either the transfer of the first or the last descriptor is in progress. In the case of a "memory_to_memory_sg" transfer (part of this patch series) the transfer width of both descriptors may differ. Thus it is required to additionally set 'tx_width' of the last descriptor. * Added functions for multiply used calculations. Signed-off-by: Torsten Fleischer <torfl6749@gmail.com> Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2015-02-24 00:54:10 +08:00
first->total_len = buf_len;
return &first->txd;
err_desc_get:
dev_err(chan2dev(chan), "not enough descriptors available\n");
atc_desc_put(atchan, first);
err_out:
clear_bit(ATC_IS_CYCLIC, &atchan->status);
return NULL;
}
static int atc_config(struct dma_chan *chan,
struct dma_slave_config *sconfig)
{
struct at_dma_chan *atchan = to_at_dma_chan(chan);
dev_vdbg(chan2dev(chan), "%s\n", __func__);
/* Check if it is chan is configured for slave transfers */
if (!chan->private)
return -EINVAL;
memcpy(&atchan->dma_sconfig, sconfig, sizeof(*sconfig));
convert_burst(&atchan->dma_sconfig.src_maxburst);
convert_burst(&atchan->dma_sconfig.dst_maxburst);
return 0;
}
static int atc_pause(struct dma_chan *chan)
{
struct at_dma_chan *atchan = to_at_dma_chan(chan);
struct at_dma *atdma = to_at_dma(chan->device);
int chan_id = atchan->chan_common.chan_id;
unsigned long flags;
dev_vdbg(chan2dev(chan), "%s\n", __func__);
spin_lock_irqsave(&atchan->lock, flags);
dma_writel(atdma, CHER, AT_DMA_SUSP(chan_id));
set_bit(ATC_IS_PAUSED, &atchan->status);
spin_unlock_irqrestore(&atchan->lock, flags);
return 0;
}
static int atc_resume(struct dma_chan *chan)
{
struct at_dma_chan *atchan = to_at_dma_chan(chan);
struct at_dma *atdma = to_at_dma(chan->device);
int chan_id = atchan->chan_common.chan_id;
unsigned long flags;
dev_vdbg(chan2dev(chan), "%s\n", __func__);
if (!atc_chan_is_paused(atchan))
return 0;
spin_lock_irqsave(&atchan->lock, flags);
dma_writel(atdma, CHDR, AT_DMA_RES(chan_id));
clear_bit(ATC_IS_PAUSED, &atchan->status);
spin_unlock_irqrestore(&atchan->lock, flags);
return 0;
}
static int atc_terminate_all(struct dma_chan *chan)
{
struct at_dma_chan *atchan = to_at_dma_chan(chan);
struct at_dma *atdma = to_at_dma(chan->device);
int chan_id = atchan->chan_common.chan_id;
struct at_desc *desc, *_desc;
unsigned long flags;
LIST_HEAD(list);
dev_vdbg(chan2dev(chan), "%s\n", __func__);
/*
* This is only called when something went wrong elsewhere, so
* we don't really care about the data. Just disable the
* channel. We still have to poll the channel enable bit due
* to AHB/HSB limitations.
*/
spin_lock_irqsave(&atchan->lock, flags);
/* disabling channel: must also remove suspend state */
dma_writel(atdma, CHDR, AT_DMA_RES(chan_id) | atchan->mask);
/* confirm that this channel is disabled */
while (dma_readl(atdma, CHSR) & atchan->mask)
cpu_relax();
/* active_list entries will end up before queued entries */
list_splice_init(&atchan->queue, &list);
list_splice_init(&atchan->active_list, &list);
spin_unlock_irqrestore(&atchan->lock, flags);
/* Flush all pending and queued descriptors */
list_for_each_entry_safe(desc, _desc, &list, desc_node)
atc_chain_complete(atchan, desc);
clear_bit(ATC_IS_PAUSED, &atchan->status);
/* if channel dedicated to cyclic operations, free it */
clear_bit(ATC_IS_CYCLIC, &atchan->status);
return 0;
}
/**
* atc_tx_status - poll for transaction completion
* @chan: DMA channel
* @cookie: transaction identifier to check status of
* @txstate: if not %NULL updated with transaction state
*
* If @txstate is passed in, upon return it reflect the driver
* internal state and can be used with dma_async_is_complete() to check
* the status of multiple cookies without re-checking hardware state.
*/
static enum dma_status
atc_tx_status(struct dma_chan *chan,
dma_cookie_t cookie,
struct dma_tx_state *txstate)
{
struct at_dma_chan *atchan = to_at_dma_chan(chan);
unsigned long flags;
enum dma_status ret;
int bytes = 0;
ret = dma_cookie_status(chan, cookie, txstate);
if (ret == DMA_COMPLETE)
return ret;
/*
* There's no point calculating the residue if there's
* no txstate to store the value.
*/
if (!txstate)
return DMA_ERROR;
spin_lock_irqsave(&atchan->lock, flags);
/* Get number of bytes left in the active transactions */
dmaengine: at_hdmac: Fix calculation of the residual bytes This patch fixes the following issues regarding to the calculation of the residue: 1. The residue is always calculated for the current transfer even if the cookie is associated to a pending transfer. 2. For scatter/gather DMA the calculation of the residue for the current transfer doesn't include the bytes of the child descriptors that are already transferred. It only calculates the difference between the transfer's total length minus the number of bytes that are already transferred for the current child descriptor. For example: There is a scatter/gather DMA transfer with a total length of 1 MByte. Getting the residue several times while the transfer is running shows something like that: 1: residue = 975584 2: residue = 1002766 3: residue = 992627 4: residue = 983767 5: residue = 985694 6: residue = 1008094 7: residue = 1009741 8: residue = 1011195 3. The driver stores the residue but never resets it when starting a new transfer. For example: If there are two subsequent DMA transfers. The first one with a total length of 1 MByte and the second one with a total length of 1 kByte. Getting the residue for both transfers shows something like that: transfer 1: residue = 975584 transfer 2: residue = 1048380 Changes from V1: * Fixed coding style of the multi-line comments. * Improved accuracy of the residue calculation when the transfer for the first descriptor is active. Changes from V2: * Member 'tx_width' of 'struct at_desc' restored, because the transfer width can't be derived from the source width when using "slave_sg". The transfer width is needed for the calculation of the residue if either the transfer of the first or the last descriptor is in progress. In the case of a "memory_to_memory_sg" transfer (part of this patch series) the transfer width of both descriptors may differ. Thus it is required to additionally set 'tx_width' of the last descriptor. * Added functions for multiply used calculations. Signed-off-by: Torsten Fleischer <torfl6749@gmail.com> Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2015-02-24 00:54:10 +08:00
bytes = atc_get_bytes_left(chan, cookie);
spin_unlock_irqrestore(&atchan->lock, flags);
if (unlikely(bytes < 0)) {
dev_vdbg(chan2dev(chan), "get residual bytes error\n");
return DMA_ERROR;
} else {
dma_set_residue(txstate, bytes);
}
dev_vdbg(chan2dev(chan), "tx_status %d: cookie = %d residue = %d\n",
ret, cookie, bytes);
return ret;
}
/**
* atc_issue_pending - try to finish work
* @chan: target DMA channel
*/
static void atc_issue_pending(struct dma_chan *chan)
{
struct at_dma_chan *atchan = to_at_dma_chan(chan);
dev_vdbg(chan2dev(chan), "issue_pending\n");
/* Not needed for cyclic transfers */
if (atc_chan_is_cyclic(atchan))
return;
atc_advance_work(atchan);
}
/**
* atc_alloc_chan_resources - allocate resources for DMA channel
* @chan: allocate descriptor resources for this channel
*
* return - the number of allocated descriptors
*/
static int atc_alloc_chan_resources(struct dma_chan *chan)
{
struct at_dma_chan *atchan = to_at_dma_chan(chan);
struct at_dma *atdma = to_at_dma(chan->device);
struct at_desc *desc;
struct at_dma_slave *atslave;
int i;
u32 cfg;
dev_vdbg(chan2dev(chan), "alloc_chan_resources\n");
/* ASSERT: channel is idle */
if (atc_chan_is_enabled(atchan)) {
dev_dbg(chan2dev(chan), "DMA channel not idle ?\n");
return -EIO;
}
if (!list_empty(&atchan->free_list)) {
dev_dbg(chan2dev(chan), "can't allocate channel resources (channel not freed from a previous use)\n");
return -EIO;
}
cfg = ATC_DEFAULT_CFG;
atslave = chan->private;
if (atslave) {
/*
* We need controller-specific data to set up slave
* transfers.
*/
BUG_ON(!atslave->dma_dev || atslave->dma_dev != atdma->dma_common.dev);
/* if cfg configuration specified take it instead of default */
if (atslave->cfg)
cfg = atslave->cfg;
}
/* Allocate initial pool of descriptors */
for (i = 0; i < init_nr_desc_per_channel; i++) {
desc = atc_alloc_descriptor(chan, GFP_KERNEL);
if (!desc) {
dev_err(atdma->dma_common.dev,
"Only %d initial descriptors\n", i);
break;
}
list_add_tail(&desc->desc_node, &atchan->free_list);
}
dma_cookie_init(chan);
/* channel parameters */
channel_writel(atchan, CFG, cfg);
dev_dbg(chan2dev(chan),
"alloc_chan_resources: allocated %d descriptors\n", i);
return i;
}
/**
* atc_free_chan_resources - free all channel resources
* @chan: DMA channel
*/
static void atc_free_chan_resources(struct dma_chan *chan)
{
struct at_dma_chan *atchan = to_at_dma_chan(chan);
struct at_dma *atdma = to_at_dma(chan->device);
struct at_desc *desc, *_desc;
LIST_HEAD(list);
/* ASSERT: channel is idle */
BUG_ON(!list_empty(&atchan->active_list));
BUG_ON(!list_empty(&atchan->queue));
BUG_ON(atc_chan_is_enabled(atchan));
list_for_each_entry_safe(desc, _desc, &atchan->free_list, desc_node) {
dev_vdbg(chan2dev(chan), " freeing descriptor %p\n", desc);
list_del(&desc->desc_node);
/* free link descriptor */
dma_pool_free(atdma->dma_desc_pool, desc, desc->txd.phys);
}
list_splice_init(&atchan->free_list, &list);
atchan->status = 0;
/*
* Free atslave allocated in at_dma_xlate()
*/
kfree(chan->private);
chan->private = NULL;
dev_vdbg(chan2dev(chan), "free_chan_resources: done\n");
}
#ifdef CONFIG_OF
static bool at_dma_filter(struct dma_chan *chan, void *slave)
{
struct at_dma_slave *atslave = slave;
if (atslave->dma_dev == chan->device->dev) {
chan->private = atslave;
return true;
} else {
return false;
}
}
static struct dma_chan *at_dma_xlate(struct of_phandle_args *dma_spec,
struct of_dma *of_dma)
{
struct dma_chan *chan;
struct at_dma_chan *atchan;
struct at_dma_slave *atslave;
dma_cap_mask_t mask;
unsigned int per_id;
struct platform_device *dmac_pdev;
if (dma_spec->args_count != 2)
return NULL;
dmac_pdev = of_find_device_by_node(dma_spec->np);
if (!dmac_pdev)
return NULL;
dma_cap_zero(mask);
dma_cap_set(DMA_SLAVE, mask);
atslave = kmalloc(sizeof(*atslave), GFP_KERNEL);
if (!atslave) {
put_device(&dmac_pdev->dev);
return NULL;
}
atslave->cfg = ATC_DST_H2SEL_HW | ATC_SRC_H2SEL_HW;
/*
* We can fill both SRC_PER and DST_PER, one of these fields will be
* ignored depending on DMA transfer direction.
*/
per_id = dma_spec->args[1] & AT91_DMA_CFG_PER_ID_MASK;
atslave->cfg |= ATC_DST_PER_MSB(per_id) | ATC_DST_PER(per_id)
| ATC_SRC_PER_MSB(per_id) | ATC_SRC_PER(per_id);
/*
* We have to translate the value we get from the device tree since
* the half FIFO configuration value had to be 0 to keep backward
* compatibility.
*/
switch (dma_spec->args[1] & AT91_DMA_CFG_FIFOCFG_MASK) {
case AT91_DMA_CFG_FIFOCFG_ALAP:
atslave->cfg |= ATC_FIFOCFG_LARGESTBURST;
break;
case AT91_DMA_CFG_FIFOCFG_ASAP:
atslave->cfg |= ATC_FIFOCFG_ENOUGHSPACE;
break;
case AT91_DMA_CFG_FIFOCFG_HALF:
default:
atslave->cfg |= ATC_FIFOCFG_HALFFIFO;
}
atslave->dma_dev = &dmac_pdev->dev;
chan = dma_request_channel(mask, at_dma_filter, atslave);
if (!chan) {
put_device(&dmac_pdev->dev);
kfree(atslave);
return NULL;
}
atchan = to_at_dma_chan(chan);
atchan->per_if = dma_spec->args[0] & 0xff;
atchan->mem_if = (dma_spec->args[0] >> 16) & 0xff;
return chan;
}
#else
static struct dma_chan *at_dma_xlate(struct of_phandle_args *dma_spec,
struct of_dma *of_dma)
{
return NULL;
}
#endif
/*-- Module Management -----------------------------------------------*/
/* cap_mask is a multi-u32 bitfield, fill it with proper C code. */
static struct at_dma_platform_data at91sam9rl_config = {
.nr_channels = 2,
};
static struct at_dma_platform_data at91sam9g45_config = {
.nr_channels = 8,
};
#if defined(CONFIG_OF)
static const struct of_device_id atmel_dma_dt_ids[] = {
{
.compatible = "atmel,at91sam9rl-dma",
.data = &at91sam9rl_config,
}, {
.compatible = "atmel,at91sam9g45-dma",
.data = &at91sam9g45_config,
}, {
/* sentinel */
}
};
MODULE_DEVICE_TABLE(of, atmel_dma_dt_ids);
#endif
static const struct platform_device_id atdma_devtypes[] = {
{
.name = "at91sam9rl_dma",
.driver_data = (unsigned long) &at91sam9rl_config,
}, {
.name = "at91sam9g45_dma",
.driver_data = (unsigned long) &at91sam9g45_config,
}, {
/* sentinel */
}
};
static inline const struct at_dma_platform_data * __init at_dma_get_driver_data(
struct platform_device *pdev)
{
if (pdev->dev.of_node) {
const struct of_device_id *match;
match = of_match_node(atmel_dma_dt_ids, pdev->dev.of_node);
if (match == NULL)
return NULL;
return match->data;
}
return (struct at_dma_platform_data *)
platform_get_device_id(pdev)->driver_data;
}
/**
* at_dma_off - disable DMA controller
* @atdma: the Atmel HDAMC device
*/
static void at_dma_off(struct at_dma *atdma)
{
dma_writel(atdma, EN, 0);
/* disable all interrupts */
dma_writel(atdma, EBCIDR, -1L);
/* confirm that all channels are disabled */
while (dma_readl(atdma, CHSR) & atdma->all_chan_mask)
cpu_relax();
}
static int __init at_dma_probe(struct platform_device *pdev)
{
struct resource *io;
struct at_dma *atdma;
size_t size;
int irq;
int err;
int i;
const struct at_dma_platform_data *plat_dat;
/* setup platform data for each SoC */
dma_cap_set(DMA_MEMCPY, at91sam9rl_config.cap_mask);
dma_cap_set(DMA_INTERLEAVE, at91sam9g45_config.cap_mask);
dma_cap_set(DMA_MEMCPY, at91sam9g45_config.cap_mask);
dma_cap_set(DMA_MEMSET, at91sam9g45_config.cap_mask);
dma_cap_set(DMA_MEMSET_SG, at91sam9g45_config.cap_mask);
dma_cap_set(DMA_PRIVATE, at91sam9g45_config.cap_mask);
dma_cap_set(DMA_SLAVE, at91sam9g45_config.cap_mask);
/* get DMA parameters from controller type */
plat_dat = at_dma_get_driver_data(pdev);
if (!plat_dat)
return -ENODEV;
io = platform_get_resource(pdev, IORESOURCE_MEM, 0);
if (!io)
return -EINVAL;
irq = platform_get_irq(pdev, 0);
if (irq < 0)
return irq;
size = sizeof(struct at_dma);
size += plat_dat->nr_channels * sizeof(struct at_dma_chan);
atdma = kzalloc(size, GFP_KERNEL);
if (!atdma)
return -ENOMEM;
/* discover transaction capabilities */
atdma->dma_common.cap_mask = plat_dat->cap_mask;
atdma->all_chan_mask = (1 << plat_dat->nr_channels) - 1;
size = resource_size(io);
if (!request_mem_region(io->start, size, pdev->dev.driver->name)) {
err = -EBUSY;
goto err_kfree;
}
atdma->regs = ioremap(io->start, size);
if (!atdma->regs) {
err = -ENOMEM;
goto err_release_r;
}
atdma->clk = clk_get(&pdev->dev, "dma_clk");
if (IS_ERR(atdma->clk)) {
err = PTR_ERR(atdma->clk);
goto err_clk;
}
err = clk_prepare_enable(atdma->clk);
if (err)
goto err_clk_prepare;
/* force dma off, just in case */
at_dma_off(atdma);
err = request_irq(irq, at_dma_interrupt, 0, "at_hdmac", atdma);
if (err)
goto err_irq;
platform_set_drvdata(pdev, atdma);
/* create a pool of consistent memory blocks for hardware descriptors */
atdma->dma_desc_pool = dma_pool_create("at_hdmac_desc_pool",
&pdev->dev, sizeof(struct at_desc),
4 /* word alignment */, 0);
if (!atdma->dma_desc_pool) {
dev_err(&pdev->dev, "No memory for descriptors dma pool\n");
err = -ENOMEM;
goto err_desc_pool_create;
}
/* create a pool of consistent memory blocks for memset blocks */
atdma->memset_pool = dma_pool_create("at_hdmac_memset_pool",
&pdev->dev, sizeof(int), 4, 0);
if (!atdma->memset_pool) {
dev_err(&pdev->dev, "No memory for memset dma pool\n");
err = -ENOMEM;
goto err_memset_pool_create;
}
/* clear any pending interrupt */
while (dma_readl(atdma, EBCISR))
cpu_relax();
/* initialize channels related values */
INIT_LIST_HEAD(&atdma->dma_common.channels);
for (i = 0; i < plat_dat->nr_channels; i++) {
struct at_dma_chan *atchan = &atdma->chan[i];
atchan->mem_if = AT_DMA_MEM_IF;
atchan->per_if = AT_DMA_PER_IF;
atchan->chan_common.device = &atdma->dma_common;
dma_cookie_init(&atchan->chan_common);
list_add_tail(&atchan->chan_common.device_node,
&atdma->dma_common.channels);
atchan->ch_regs = atdma->regs + ch_regs(i);
spin_lock_init(&atchan->lock);
atchan->mask = 1 << i;
INIT_LIST_HEAD(&atchan->active_list);
INIT_LIST_HEAD(&atchan->queue);
INIT_LIST_HEAD(&atchan->free_list);
tasklet_setup(&atchan->tasklet, atc_tasklet);
atc_enable_chan_irq(atdma, i);
}
/* set base routines */
atdma->dma_common.device_alloc_chan_resources = atc_alloc_chan_resources;
atdma->dma_common.device_free_chan_resources = atc_free_chan_resources;
atdma->dma_common.device_tx_status = atc_tx_status;
atdma->dma_common.device_issue_pending = atc_issue_pending;
atdma->dma_common.dev = &pdev->dev;
/* set prep routines based on capability */
if (dma_has_cap(DMA_INTERLEAVE, atdma->dma_common.cap_mask))
atdma->dma_common.device_prep_interleaved_dma = atc_prep_dma_interleaved;
if (dma_has_cap(DMA_MEMCPY, atdma->dma_common.cap_mask))
atdma->dma_common.device_prep_dma_memcpy = atc_prep_dma_memcpy;
if (dma_has_cap(DMA_MEMSET, atdma->dma_common.cap_mask)) {
atdma->dma_common.device_prep_dma_memset = atc_prep_dma_memset;
atdma->dma_common.device_prep_dma_memset_sg = atc_prep_dma_memset_sg;
atdma->dma_common.fill_align = DMAENGINE_ALIGN_4_BYTES;
}
if (dma_has_cap(DMA_SLAVE, atdma->dma_common.cap_mask)) {
atdma->dma_common.device_prep_slave_sg = atc_prep_slave_sg;
/* controller can do slave DMA: can trigger cyclic transfers */
dma_cap_set(DMA_CYCLIC, atdma->dma_common.cap_mask);
atdma->dma_common.device_prep_dma_cyclic = atc_prep_dma_cyclic;
atdma->dma_common.device_config = atc_config;
atdma->dma_common.device_pause = atc_pause;
atdma->dma_common.device_resume = atc_resume;
atdma->dma_common.device_terminate_all = atc_terminate_all;
atdma->dma_common.src_addr_widths = ATC_DMA_BUSWIDTHS;
atdma->dma_common.dst_addr_widths = ATC_DMA_BUSWIDTHS;
atdma->dma_common.directions = BIT(DMA_DEV_TO_MEM) | BIT(DMA_MEM_TO_DEV);
atdma->dma_common.residue_granularity = DMA_RESIDUE_GRANULARITY_BURST;
}
dma_writel(atdma, EN, AT_DMA_ENABLE);
dev_info(&pdev->dev, "Atmel AHB DMA Controller ( %s%s%s), %d channels\n",
dma_has_cap(DMA_MEMCPY, atdma->dma_common.cap_mask) ? "cpy " : "",
dma_has_cap(DMA_MEMSET, atdma->dma_common.cap_mask) ? "set " : "",
dma_has_cap(DMA_SLAVE, atdma->dma_common.cap_mask) ? "slave " : "",
plat_dat->nr_channels);
dma_async_device_register(&atdma->dma_common);
/*
* Do not return an error if the dmac node is not present in order to
* not break the existing way of requesting channel with
* dma_request_channel().
*/
if (pdev->dev.of_node) {
err = of_dma_controller_register(pdev->dev.of_node,
at_dma_xlate, atdma);
if (err) {
dev_err(&pdev->dev, "could not register of_dma_controller\n");
goto err_of_dma_controller_register;
}
}
return 0;
err_of_dma_controller_register:
dma_async_device_unregister(&atdma->dma_common);
dma_pool_destroy(atdma->memset_pool);
err_memset_pool_create:
dma_pool_destroy(atdma->dma_desc_pool);
err_desc_pool_create:
free_irq(platform_get_irq(pdev, 0), atdma);
err_irq:
clk_disable_unprepare(atdma->clk);
err_clk_prepare:
clk_put(atdma->clk);
err_clk:
iounmap(atdma->regs);
atdma->regs = NULL;
err_release_r:
release_mem_region(io->start, size);
err_kfree:
kfree(atdma);
return err;
}
static int at_dma_remove(struct platform_device *pdev)
{
struct at_dma *atdma = platform_get_drvdata(pdev);
struct dma_chan *chan, *_chan;
struct resource *io;
at_dma_off(atdma);
if (pdev->dev.of_node)
of_dma_controller_free(pdev->dev.of_node);
dma_async_device_unregister(&atdma->dma_common);
dma_pool_destroy(atdma->memset_pool);
dma_pool_destroy(atdma->dma_desc_pool);
free_irq(platform_get_irq(pdev, 0), atdma);
list_for_each_entry_safe(chan, _chan, &atdma->dma_common.channels,
device_node) {
struct at_dma_chan *atchan = to_at_dma_chan(chan);
/* Disable interrupts */
atc_disable_chan_irq(atdma, chan->chan_id);
tasklet_kill(&atchan->tasklet);
list_del(&chan->device_node);
}
clk_disable_unprepare(atdma->clk);
clk_put(atdma->clk);
iounmap(atdma->regs);
atdma->regs = NULL;
io = platform_get_resource(pdev, IORESOURCE_MEM, 0);
release_mem_region(io->start, resource_size(io));
kfree(atdma);
return 0;
}
static void at_dma_shutdown(struct platform_device *pdev)
{
struct at_dma *atdma = platform_get_drvdata(pdev);
at_dma_off(platform_get_drvdata(pdev));
clk_disable_unprepare(atdma->clk);
}
static int at_dma_prepare(struct device *dev)
{
struct at_dma *atdma = dev_get_drvdata(dev);
struct dma_chan *chan, *_chan;
list_for_each_entry_safe(chan, _chan, &atdma->dma_common.channels,
device_node) {
struct at_dma_chan *atchan = to_at_dma_chan(chan);
/* wait for transaction completion (except in cyclic case) */
if (atc_chan_is_enabled(atchan) && !atc_chan_is_cyclic(atchan))
return -EAGAIN;
}
return 0;
}
static void atc_suspend_cyclic(struct at_dma_chan *atchan)
{
struct dma_chan *chan = &atchan->chan_common;
/* Channel should be paused by user
* do it anyway even if it is not done already */
if (!atc_chan_is_paused(atchan)) {
dev_warn(chan2dev(chan),
"cyclic channel not paused, should be done by channel user\n");
atc_pause(chan);
}
/* now preserve additional data for cyclic operations */
/* next descriptor address in the cyclic list */
atchan->save_dscr = channel_readl(atchan, DSCR);
vdbg_dump_regs(atchan);
}
static int at_dma_suspend_noirq(struct device *dev)
{
struct at_dma *atdma = dev_get_drvdata(dev);
struct dma_chan *chan, *_chan;
/* preserve data */
list_for_each_entry_safe(chan, _chan, &atdma->dma_common.channels,
device_node) {
struct at_dma_chan *atchan = to_at_dma_chan(chan);
if (atc_chan_is_cyclic(atchan))
atc_suspend_cyclic(atchan);
atchan->save_cfg = channel_readl(atchan, CFG);
}
atdma->save_imr = dma_readl(atdma, EBCIMR);
/* disable DMA controller */
at_dma_off(atdma);
clk_disable_unprepare(atdma->clk);
return 0;
}
static void atc_resume_cyclic(struct at_dma_chan *atchan)
{
struct at_dma *atdma = to_at_dma(atchan->chan_common.device);
/* restore channel status for cyclic descriptors list:
* next descriptor in the cyclic list at the time of suspend */
channel_writel(atchan, SADDR, 0);
channel_writel(atchan, DADDR, 0);
channel_writel(atchan, CTRLA, 0);
channel_writel(atchan, CTRLB, 0);
channel_writel(atchan, DSCR, atchan->save_dscr);
dma_writel(atdma, CHER, atchan->mask);
/* channel pause status should be removed by channel user
* We cannot take the initiative to do it here */
vdbg_dump_regs(atchan);
}
static int at_dma_resume_noirq(struct device *dev)
{
struct at_dma *atdma = dev_get_drvdata(dev);
struct dma_chan *chan, *_chan;
/* bring back DMA controller */
clk_prepare_enable(atdma->clk);
dma_writel(atdma, EN, AT_DMA_ENABLE);
/* clear any pending interrupt */
while (dma_readl(atdma, EBCISR))
cpu_relax();
/* restore saved data */
dma_writel(atdma, EBCIER, atdma->save_imr);
list_for_each_entry_safe(chan, _chan, &atdma->dma_common.channels,
device_node) {
struct at_dma_chan *atchan = to_at_dma_chan(chan);
channel_writel(atchan, CFG, atchan->save_cfg);
if (atc_chan_is_cyclic(atchan))
atc_resume_cyclic(atchan);
}
return 0;
}
static const struct dev_pm_ops at_dma_dev_pm_ops = {
.prepare = at_dma_prepare,
.suspend_noirq = at_dma_suspend_noirq,
.resume_noirq = at_dma_resume_noirq,
};
static struct platform_driver at_dma_driver = {
.remove = at_dma_remove,
.shutdown = at_dma_shutdown,
.id_table = atdma_devtypes,
.driver = {
.name = "at_hdmac",
.pm = &at_dma_dev_pm_ops,
.of_match_table = of_match_ptr(atmel_dma_dt_ids),
},
};
static int __init at_dma_init(void)
{
return platform_driver_probe(&at_dma_driver, at_dma_probe);
}
subsys_initcall(at_dma_init);
static void __exit at_dma_exit(void)
{
platform_driver_unregister(&at_dma_driver);
}
module_exit(at_dma_exit);
MODULE_DESCRIPTION("Atmel AHB DMA Controller driver");
MODULE_AUTHOR("Nicolas Ferre <nicolas.ferre@atmel.com>");
MODULE_LICENSE("GPL");
MODULE_ALIAS("platform:at_hdmac");