linux/drivers/dma/dmatest.c

988 lines
26 KiB
C
Raw Normal View History

/*
* DMA Engine test module
*
* Copyright (C) 2007 Atmel Corporation
dmatest: run test via debugfs Instead of doing modprobe dmatest ... modprobe -r dmatest we allow user to run tests interactively. The dmatest could be built as module or inside kernel. Let's consider those cases. 1. When dmatest is built as a module... After mounting debugfs and loading the module, the /sys/kernel/debug/dmatest folder with nodes will be created. They are the same as module parameters with addition of the 'run' node that controls run and stop phases of the test. Note that in this case test will not run on load automatically. Example of usage: % echo dma0chan0 > /sys/kernel/debug/dmatest/channel % echo 2000 > /sys/kernel/debug/dmatest/timeout % echo 1 > /sys/kernel/debug/dmatest/iterations % echo 1 > /sys/kernel/debug/dmatest/run After a while you will start to get messages about current status or error like in the original code. Note that running a new test will stop any in progress test. 2. When built-in in the kernel... The module parameters that is supplied to the kernel command line will be used for the first performed test. After user gets a control, the test could be interrupted or re-run with same or different parameters. For the details see the above section "1. When dmatest is built as a module..." In both cases the module parameters are used as initial values for the test case. You always could check them at run-time by running % grep -H . /sys/module/dmatest/parameters/* Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2013-03-04 09:09:30 +00:00
* Copyright (C) 2013 Intel Corporation
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License version 2 as
* published by the Free Software Foundation.
*/
#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
#include <linux/delay.h>
#include <linux/dma-mapping.h>
#include <linux/dmaengine.h>
#include <linux/freezer.h>
#include <linux/init.h>
#include <linux/kthread.h>
#include <linux/module.h>
#include <linux/moduleparam.h>
#include <linux/random.h>
include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h percpu.h is included by sched.h and module.h and thus ends up being included when building most .c files. percpu.h includes slab.h which in turn includes gfp.h making everything defined by the two files universally available and complicating inclusion dependencies. percpu.h -> slab.h dependency is about to be removed. Prepare for this change by updating users of gfp and slab facilities include those headers directly instead of assuming availability. As this conversion needs to touch large number of source files, the following script is used as the basis of conversion. http://userweb.kernel.org/~tj/misc/slabh-sweep.py The script does the followings. * Scan files for gfp and slab usages and update includes such that only the necessary includes are there. ie. if only gfp is used, gfp.h, if slab is used, slab.h. * When the script inserts a new include, it looks at the include blocks and try to put the new include such that its order conforms to its surrounding. It's put in the include block which contains core kernel includes, in the same order that the rest are ordered - alphabetical, Christmas tree, rev-Xmas-tree or at the end if there doesn't seem to be any matching order. * If the script can't find a place to put a new include (mostly because the file doesn't have fitting include block), it prints out an error message indicating which .h file needs to be added to the file. The conversion was done in the following steps. 1. The initial automatic conversion of all .c files updated slightly over 4000 files, deleting around 700 includes and adding ~480 gfp.h and ~3000 slab.h inclusions. The script emitted errors for ~400 files. 2. Each error was manually checked. Some didn't need the inclusion, some needed manual addition while adding it to implementation .h or embedding .c file was more appropriate for others. This step added inclusions to around 150 files. 3. The script was run again and the output was compared to the edits from #2 to make sure no file was left behind. 4. Several build tests were done and a couple of problems were fixed. e.g. lib/decompress_*.c used malloc/free() wrappers around slab APIs requiring slab.h to be added manually. 5. The script was run on all .h files but without automatically editing them as sprinkling gfp.h and slab.h inclusions around .h files could easily lead to inclusion dependency hell. Most gfp.h inclusion directives were ignored as stuff from gfp.h was usually wildly available and often used in preprocessor macros. Each slab.h inclusion directive was examined and added manually as necessary. 6. percpu.h was updated not to include slab.h. 7. Build test were done on the following configurations and failures were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my distributed build env didn't work with gcov compiles) and a few more options had to be turned off depending on archs to make things build (like ipr on powerpc/64 which failed due to missing writeq). * x86 and x86_64 UP and SMP allmodconfig and a custom test config. * powerpc and powerpc64 SMP allmodconfig * sparc and sparc64 SMP allmodconfig * ia64 SMP allmodconfig * s390 SMP allmodconfig * alpha SMP allmodconfig * um on x86_64 SMP allmodconfig 8. percpu.h modifications were reverted so that it could be applied as a separate patch and serve as bisection point. Given the fact that I had only a couple of failures from tests on step 6, I'm fairly confident about the coverage of this conversion patch. If there is a breakage, it's likely to be something in one of the arch headers which should be easily discoverable easily on most builds of the specific arch. Signed-off-by: Tejun Heo <tj@kernel.org> Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org> Cc: Ingo Molnar <mingo@redhat.com> Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
2010-03-24 08:04:11 +00:00
#include <linux/slab.h>
#include <linux/wait.h>
static unsigned int test_buf_size = 16384;
module_param(test_buf_size, uint, S_IRUGO | S_IWUSR);
MODULE_PARM_DESC(test_buf_size, "Size of the memcpy test buffer");
static char test_channel[20];
module_param_string(channel, test_channel, sizeof(test_channel),
S_IRUGO | S_IWUSR);
MODULE_PARM_DESC(channel, "Bus ID of the channel to test (default: any)");
static char test_device[32];
module_param_string(device, test_device, sizeof(test_device),
S_IRUGO | S_IWUSR);
MODULE_PARM_DESC(device, "Bus ID of the DMA Engine to test (default: any)");
static unsigned int threads_per_chan = 1;
module_param(threads_per_chan, uint, S_IRUGO | S_IWUSR);
MODULE_PARM_DESC(threads_per_chan,
"Number of threads to start per channel (default: 1)");
static unsigned int max_channels;
module_param(max_channels, uint, S_IRUGO | S_IWUSR);
MODULE_PARM_DESC(max_channels,
"Maximum number of channels to use (default: all)");
static unsigned int iterations;
module_param(iterations, uint, S_IRUGO | S_IWUSR);
MODULE_PARM_DESC(iterations,
"Iterations before stopping test (default: infinite)");
static unsigned int xor_sources = 3;
module_param(xor_sources, uint, S_IRUGO | S_IWUSR);
MODULE_PARM_DESC(xor_sources,
"Number of xor source buffers (default: 3)");
static unsigned int pq_sources = 3;
module_param(pq_sources, uint, S_IRUGO | S_IWUSR);
MODULE_PARM_DESC(pq_sources,
"Number of p+q source buffers (default: 3)");
static int timeout = 3000;
module_param(timeout, uint, S_IRUGO | S_IWUSR);
MODULE_PARM_DESC(timeout, "Transfer Timeout in msec (default: 3000), "
"Pass -1 for infinite timeout");
static bool noverify;
module_param(noverify, bool, S_IRUGO | S_IWUSR);
MODULE_PARM_DESC(noverify, "Disable random data setup and verification");
static bool verbose;
module_param(verbose, bool, S_IRUGO | S_IWUSR);
MODULE_PARM_DESC(verbose, "Enable \"success\" result messages (default: off)");
/**
* struct dmatest_params - test parameters.
* @buf_size: size of the memcpy test buffer
* @channel: bus ID of the channel to test
* @device: bus ID of the DMA Engine to test
* @threads_per_chan: number of threads to start per channel
* @max_channels: maximum number of channels to use
* @iterations: iterations before stopping test
* @xor_sources: number of xor source buffers
* @pq_sources: number of p+q source buffers
* @timeout: transfer timeout in msec, -1 for infinite timeout
*/
struct dmatest_params {
unsigned int buf_size;
char channel[20];
char device[32];
unsigned int threads_per_chan;
unsigned int max_channels;
unsigned int iterations;
unsigned int xor_sources;
unsigned int pq_sources;
int timeout;
bool noverify;
};
/**
* struct dmatest_info - test information.
* @params: test parameters
dmatest: run test via debugfs Instead of doing modprobe dmatest ... modprobe -r dmatest we allow user to run tests interactively. The dmatest could be built as module or inside kernel. Let's consider those cases. 1. When dmatest is built as a module... After mounting debugfs and loading the module, the /sys/kernel/debug/dmatest folder with nodes will be created. They are the same as module parameters with addition of the 'run' node that controls run and stop phases of the test. Note that in this case test will not run on load automatically. Example of usage: % echo dma0chan0 > /sys/kernel/debug/dmatest/channel % echo 2000 > /sys/kernel/debug/dmatest/timeout % echo 1 > /sys/kernel/debug/dmatest/iterations % echo 1 > /sys/kernel/debug/dmatest/run After a while you will start to get messages about current status or error like in the original code. Note that running a new test will stop any in progress test. 2. When built-in in the kernel... The module parameters that is supplied to the kernel command line will be used for the first performed test. After user gets a control, the test could be interrupted or re-run with same or different parameters. For the details see the above section "1. When dmatest is built as a module..." In both cases the module parameters are used as initial values for the test case. You always could check them at run-time by running % grep -H . /sys/module/dmatest/parameters/* Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2013-03-04 09:09:30 +00:00
* @lock: access protection to the fields of this structure
*/
static struct dmatest_info {
/* Test parameters */
struct dmatest_params params;
/* Internal state */
struct list_head channels;
unsigned int nr_channels;
dmatest: run test via debugfs Instead of doing modprobe dmatest ... modprobe -r dmatest we allow user to run tests interactively. The dmatest could be built as module or inside kernel. Let's consider those cases. 1. When dmatest is built as a module... After mounting debugfs and loading the module, the /sys/kernel/debug/dmatest folder with nodes will be created. They are the same as module parameters with addition of the 'run' node that controls run and stop phases of the test. Note that in this case test will not run on load automatically. Example of usage: % echo dma0chan0 > /sys/kernel/debug/dmatest/channel % echo 2000 > /sys/kernel/debug/dmatest/timeout % echo 1 > /sys/kernel/debug/dmatest/iterations % echo 1 > /sys/kernel/debug/dmatest/run After a while you will start to get messages about current status or error like in the original code. Note that running a new test will stop any in progress test. 2. When built-in in the kernel... The module parameters that is supplied to the kernel command line will be used for the first performed test. After user gets a control, the test could be interrupted or re-run with same or different parameters. For the details see the above section "1. When dmatest is built as a module..." In both cases the module parameters are used as initial values for the test case. You always could check them at run-time by running % grep -H . /sys/module/dmatest/parameters/* Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2013-03-04 09:09:30 +00:00
struct mutex lock;
bool did_init;
} test_info = {
.channels = LIST_HEAD_INIT(test_info.channels),
.lock = __MUTEX_INITIALIZER(test_info.lock),
};
dmatest: run test via debugfs Instead of doing modprobe dmatest ... modprobe -r dmatest we allow user to run tests interactively. The dmatest could be built as module or inside kernel. Let's consider those cases. 1. When dmatest is built as a module... After mounting debugfs and loading the module, the /sys/kernel/debug/dmatest folder with nodes will be created. They are the same as module parameters with addition of the 'run' node that controls run and stop phases of the test. Note that in this case test will not run on load automatically. Example of usage: % echo dma0chan0 > /sys/kernel/debug/dmatest/channel % echo 2000 > /sys/kernel/debug/dmatest/timeout % echo 1 > /sys/kernel/debug/dmatest/iterations % echo 1 > /sys/kernel/debug/dmatest/run After a while you will start to get messages about current status or error like in the original code. Note that running a new test will stop any in progress test. 2. When built-in in the kernel... The module parameters that is supplied to the kernel command line will be used for the first performed test. After user gets a control, the test could be interrupted or re-run with same or different parameters. For the details see the above section "1. When dmatest is built as a module..." In both cases the module parameters are used as initial values for the test case. You always could check them at run-time by running % grep -H . /sys/module/dmatest/parameters/* Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2013-03-04 09:09:30 +00:00
static int dmatest_run_set(const char *val, const struct kernel_param *kp);
static int dmatest_run_get(char *val, const struct kernel_param *kp);
static const struct kernel_param_ops run_ops = {
.set = dmatest_run_set,
.get = dmatest_run_get,
};
static bool dmatest_run;
module_param_cb(run, &run_ops, &dmatest_run, S_IRUGO | S_IWUSR);
MODULE_PARM_DESC(run, "Run the test (default: false)");
/* Maximum amount of mismatched bytes in buffer to print */
#define MAX_ERROR_COUNT 32
/*
* Initialization patterns. All bytes in the source buffer has bit 7
* set, all bytes in the destination buffer has bit 7 cleared.
*
* Bit 6 is set for all bytes which are to be copied by the DMA
* engine. Bit 5 is set for all bytes which are to be overwritten by
* the DMA engine.
*
* The remaining bits are the inverse of a counter which increments by
* one for each byte address.
*/
#define PATTERN_SRC 0x80
#define PATTERN_DST 0x00
#define PATTERN_COPY 0x40
#define PATTERN_OVERWRITE 0x20
#define PATTERN_COUNT_MASK 0x1f
dmatest: run test via debugfs Instead of doing modprobe dmatest ... modprobe -r dmatest we allow user to run tests interactively. The dmatest could be built as module or inside kernel. Let's consider those cases. 1. When dmatest is built as a module... After mounting debugfs and loading the module, the /sys/kernel/debug/dmatest folder with nodes will be created. They are the same as module parameters with addition of the 'run' node that controls run and stop phases of the test. Note that in this case test will not run on load automatically. Example of usage: % echo dma0chan0 > /sys/kernel/debug/dmatest/channel % echo 2000 > /sys/kernel/debug/dmatest/timeout % echo 1 > /sys/kernel/debug/dmatest/iterations % echo 1 > /sys/kernel/debug/dmatest/run After a while you will start to get messages about current status or error like in the original code. Note that running a new test will stop any in progress test. 2. When built-in in the kernel... The module parameters that is supplied to the kernel command line will be used for the first performed test. After user gets a control, the test could be interrupted or re-run with same or different parameters. For the details see the above section "1. When dmatest is built as a module..." In both cases the module parameters are used as initial values for the test case. You always could check them at run-time by running % grep -H . /sys/module/dmatest/parameters/* Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2013-03-04 09:09:30 +00:00
struct dmatest_thread {
struct list_head node;
struct dmatest_info *info;
struct task_struct *task;
struct dma_chan *chan;
u8 **srcs;
u8 **dsts;
enum dma_transaction_type type;
bool done;
};
struct dmatest_chan {
struct list_head node;
struct dma_chan *chan;
struct list_head threads;
};
static DECLARE_WAIT_QUEUE_HEAD(thread_wait);
static bool wait;
static bool is_threaded_test_run(struct dmatest_info *info)
{
struct dmatest_chan *dtc;
list_for_each_entry(dtc, &info->channels, node) {
struct dmatest_thread *thread;
list_for_each_entry(thread, &dtc->threads, node) {
if (!thread->done)
return true;
}
}
return false;
}
static int dmatest_wait_get(char *val, const struct kernel_param *kp)
{
struct dmatest_info *info = &test_info;
struct dmatest_params *params = &info->params;
if (params->iterations)
wait_event(thread_wait, !is_threaded_test_run(info));
wait = true;
return param_get_bool(val, kp);
}
static const struct kernel_param_ops wait_ops = {
.get = dmatest_wait_get,
.set = param_set_bool,
};
module_param_cb(wait, &wait_ops, &wait, S_IRUGO);
MODULE_PARM_DESC(wait, "Wait for tests to complete (default: false)");
static bool dmatest_match_channel(struct dmatest_params *params,
struct dma_chan *chan)
{
if (params->channel[0] == '\0')
return true;
return strcmp(dma_chan_name(chan), params->channel) == 0;
}
static bool dmatest_match_device(struct dmatest_params *params,
struct dma_device *device)
{
if (params->device[0] == '\0')
return true;
return strcmp(dev_name(device->dev), params->device) == 0;
}
static unsigned long dmatest_random(void)
{
unsigned long buf;
prandom_bytes(&buf, sizeof(buf));
return buf;
}
static void dmatest_init_srcs(u8 **bufs, unsigned int start, unsigned int len,
unsigned int buf_size)
{
unsigned int i;
u8 *buf;
for (; (buf = *bufs); bufs++) {
for (i = 0; i < start; i++)
buf[i] = PATTERN_SRC | (~i & PATTERN_COUNT_MASK);
for ( ; i < start + len; i++)
buf[i] = PATTERN_SRC | PATTERN_COPY
| (~i & PATTERN_COUNT_MASK);
for ( ; i < buf_size; i++)
buf[i] = PATTERN_SRC | (~i & PATTERN_COUNT_MASK);
buf++;
}
}
static void dmatest_init_dsts(u8 **bufs, unsigned int start, unsigned int len,
unsigned int buf_size)
{
unsigned int i;
u8 *buf;
for (; (buf = *bufs); bufs++) {
for (i = 0; i < start; i++)
buf[i] = PATTERN_DST | (~i & PATTERN_COUNT_MASK);
for ( ; i < start + len; i++)
buf[i] = PATTERN_DST | PATTERN_OVERWRITE
| (~i & PATTERN_COUNT_MASK);
for ( ; i < buf_size; i++)
buf[i] = PATTERN_DST | (~i & PATTERN_COUNT_MASK);
}
}
static void dmatest_mismatch(u8 actual, u8 pattern, unsigned int index,
unsigned int counter, bool is_srcbuf)
{
u8 diff = actual ^ pattern;
u8 expected = pattern | (~counter & PATTERN_COUNT_MASK);
const char *thread_name = current->comm;
if (is_srcbuf)
pr_warn("%s: srcbuf[0x%x] overwritten! Expected %02x, got %02x\n",
thread_name, index, expected, actual);
else if ((pattern & PATTERN_COPY)
&& (diff & (PATTERN_COPY | PATTERN_OVERWRITE)))
pr_warn("%s: dstbuf[0x%x] not copied! Expected %02x, got %02x\n",
thread_name, index, expected, actual);
else if (diff & PATTERN_SRC)
pr_warn("%s: dstbuf[0x%x] was copied! Expected %02x, got %02x\n",
thread_name, index, expected, actual);
else
pr_warn("%s: dstbuf[0x%x] mismatch! Expected %02x, got %02x\n",
thread_name, index, expected, actual);
}
static unsigned int dmatest_verify(u8 **bufs, unsigned int start,
unsigned int end, unsigned int counter, u8 pattern,
bool is_srcbuf)
{
unsigned int i;
unsigned int error_count = 0;
u8 actual;
u8 expected;
u8 *buf;
unsigned int counter_orig = counter;
for (; (buf = *bufs); bufs++) {
counter = counter_orig;
for (i = start; i < end; i++) {
actual = buf[i];
expected = pattern | (~counter & PATTERN_COUNT_MASK);
if (actual != expected) {
if (error_count < MAX_ERROR_COUNT)
dmatest_mismatch(actual, pattern, i,
counter, is_srcbuf);
error_count++;
}
counter++;
}
}
if (error_count > MAX_ERROR_COUNT)
pr_warn("%s: %u errors suppressed\n",
current->comm, error_count - MAX_ERROR_COUNT);
return error_count;
}
dmatest: don't use set_freezable_with_signal() Commit 981ed70d8e (dmatest: make dmatest threads freezable) made dmatest kthread use set_freezable_with_signal(); however, the interface is scheduled to be removed in the next merge window. The problem is that unlike userland tasks there's no default place which handles signal pending state and it isn't clear who owns and/or is responsible for clearing TIF_SIGPENDING. For example, in the current code, try_to_freeze() clears TIF_SIGPENDING but it isn't sure whether it actually owns the TIF_SIGPENDING nor is it race-free - ie. the task may continue to run with TIF_SIGPENDING set after the freezable section. Unfortunately, we don't have wait_for_completion_freezable_timeout(). This patch open codes it and uses wait_event_freezable_timeout() instead and removes timeout reloading - wait_event_freezable_timeout() won't return across freezing events (currently racy but fix scheduled) and timer doesn't decrement while the task is in freezer. Although this does lose timer-reset-over-freezing, given that timeout is supposed to be long enough and failure to finish inside is considered irrecoverable, I don't think this is worth the complexity. While at it, move completion to outer scope and explain that we're ignoring dangling pointer problem after timeout. This should give slightly better chance at avoiding oops after timeout. Signed-off-by: Tejun Heo <tj@kernel.org> Acked-by: Dan Williams <dan.j.williams@intel.com> Cc: Guennadi Liakhovetski <g.liakhovetski@gmx.de> Cc: Nicolas Ferre <nicolas.ferre@atmel.com>
2011-11-23 17:28:16 +00:00
/* poor man's completion - we want to use wait_event_freezable() on it */
struct dmatest_done {
bool done;
wait_queue_head_t *wait;
};
static void dmatest_callback(void *arg)
{
dmatest: don't use set_freezable_with_signal() Commit 981ed70d8e (dmatest: make dmatest threads freezable) made dmatest kthread use set_freezable_with_signal(); however, the interface is scheduled to be removed in the next merge window. The problem is that unlike userland tasks there's no default place which handles signal pending state and it isn't clear who owns and/or is responsible for clearing TIF_SIGPENDING. For example, in the current code, try_to_freeze() clears TIF_SIGPENDING but it isn't sure whether it actually owns the TIF_SIGPENDING nor is it race-free - ie. the task may continue to run with TIF_SIGPENDING set after the freezable section. Unfortunately, we don't have wait_for_completion_freezable_timeout(). This patch open codes it and uses wait_event_freezable_timeout() instead and removes timeout reloading - wait_event_freezable_timeout() won't return across freezing events (currently racy but fix scheduled) and timer doesn't decrement while the task is in freezer. Although this does lose timer-reset-over-freezing, given that timeout is supposed to be long enough and failure to finish inside is considered irrecoverable, I don't think this is worth the complexity. While at it, move completion to outer scope and explain that we're ignoring dangling pointer problem after timeout. This should give slightly better chance at avoiding oops after timeout. Signed-off-by: Tejun Heo <tj@kernel.org> Acked-by: Dan Williams <dan.j.williams@intel.com> Cc: Guennadi Liakhovetski <g.liakhovetski@gmx.de> Cc: Nicolas Ferre <nicolas.ferre@atmel.com>
2011-11-23 17:28:16 +00:00
struct dmatest_done *done = arg;
done->done = true;
wake_up_all(done->wait);
}
static unsigned int min_odd(unsigned int x, unsigned int y)
{
unsigned int val = min(x, y);
return val % 2 ? val : val - 1;
}
static void result(const char *err, unsigned int n, unsigned int src_off,
unsigned int dst_off, unsigned int len, unsigned long data)
{
pr_info("%s: result #%u: '%s' with src_off=0x%x dst_off=0x%x len=0x%x (%lu)\n",
current->comm, n, err, src_off, dst_off, len, data);
}
static void dbg_result(const char *err, unsigned int n, unsigned int src_off,
unsigned int dst_off, unsigned int len,
unsigned long data)
{
pr_debug("%s: result #%u: '%s' with src_off=0x%x dst_off=0x%x len=0x%x (%lu)\n",
current->comm, n, err, src_off, dst_off, len, data);
}
#define verbose_result(err, n, src_off, dst_off, len, data) ({ \
if (verbose) \
result(err, n, src_off, dst_off, len, data); \
else \
dbg_result(err, n, src_off, dst_off, len, data);\
})
static unsigned long long dmatest_persec(s64 runtime, unsigned int val)
{
unsigned long long per_sec = 1000000;
if (runtime <= 0)
return 0;
/* drop precision until runtime is 32-bits */
while (runtime > UINT_MAX) {
runtime >>= 1;
per_sec <<= 1;
}
per_sec *= val;
do_div(per_sec, runtime);
return per_sec;
}
static unsigned long long dmatest_KBs(s64 runtime, unsigned long long len)
{
return dmatest_persec(runtime, len >> 10);
}
/*
* This function repeatedly tests DMA transfers of various lengths and
* offsets for a given operation type until it is told to exit by
* kthread_stop(). There may be multiple threads running this function
* in parallel for a single channel, and there may be multiple channels
* being tested in parallel.
*
* Before each test, the source and destination buffer is initialized
* with a known pattern. This pattern is different depending on
* whether it's in an area which is supposed to be copied or
* overwritten, and different in the source and destination buffers.
* So if the DMA engine doesn't copy exactly what we tell it to copy,
* we'll notice.
*/
static int dmatest_func(void *data)
{
dmatest: don't use set_freezable_with_signal() Commit 981ed70d8e (dmatest: make dmatest threads freezable) made dmatest kthread use set_freezable_with_signal(); however, the interface is scheduled to be removed in the next merge window. The problem is that unlike userland tasks there's no default place which handles signal pending state and it isn't clear who owns and/or is responsible for clearing TIF_SIGPENDING. For example, in the current code, try_to_freeze() clears TIF_SIGPENDING but it isn't sure whether it actually owns the TIF_SIGPENDING nor is it race-free - ie. the task may continue to run with TIF_SIGPENDING set after the freezable section. Unfortunately, we don't have wait_for_completion_freezable_timeout(). This patch open codes it and uses wait_event_freezable_timeout() instead and removes timeout reloading - wait_event_freezable_timeout() won't return across freezing events (currently racy but fix scheduled) and timer doesn't decrement while the task is in freezer. Although this does lose timer-reset-over-freezing, given that timeout is supposed to be long enough and failure to finish inside is considered irrecoverable, I don't think this is worth the complexity. While at it, move completion to outer scope and explain that we're ignoring dangling pointer problem after timeout. This should give slightly better chance at avoiding oops after timeout. Signed-off-by: Tejun Heo <tj@kernel.org> Acked-by: Dan Williams <dan.j.williams@intel.com> Cc: Guennadi Liakhovetski <g.liakhovetski@gmx.de> Cc: Nicolas Ferre <nicolas.ferre@atmel.com>
2011-11-23 17:28:16 +00:00
DECLARE_WAIT_QUEUE_HEAD_ONSTACK(done_wait);
struct dmatest_thread *thread = data;
dmatest: don't use set_freezable_with_signal() Commit 981ed70d8e (dmatest: make dmatest threads freezable) made dmatest kthread use set_freezable_with_signal(); however, the interface is scheduled to be removed in the next merge window. The problem is that unlike userland tasks there's no default place which handles signal pending state and it isn't clear who owns and/or is responsible for clearing TIF_SIGPENDING. For example, in the current code, try_to_freeze() clears TIF_SIGPENDING but it isn't sure whether it actually owns the TIF_SIGPENDING nor is it race-free - ie. the task may continue to run with TIF_SIGPENDING set after the freezable section. Unfortunately, we don't have wait_for_completion_freezable_timeout(). This patch open codes it and uses wait_event_freezable_timeout() instead and removes timeout reloading - wait_event_freezable_timeout() won't return across freezing events (currently racy but fix scheduled) and timer doesn't decrement while the task is in freezer. Although this does lose timer-reset-over-freezing, given that timeout is supposed to be long enough and failure to finish inside is considered irrecoverable, I don't think this is worth the complexity. While at it, move completion to outer scope and explain that we're ignoring dangling pointer problem after timeout. This should give slightly better chance at avoiding oops after timeout. Signed-off-by: Tejun Heo <tj@kernel.org> Acked-by: Dan Williams <dan.j.williams@intel.com> Cc: Guennadi Liakhovetski <g.liakhovetski@gmx.de> Cc: Nicolas Ferre <nicolas.ferre@atmel.com>
2011-11-23 17:28:16 +00:00
struct dmatest_done done = { .wait = &done_wait };
struct dmatest_info *info;
struct dmatest_params *params;
struct dma_chan *chan;
struct dma_device *dev;
unsigned int error_count;
unsigned int failed_tests = 0;
unsigned int total_tests = 0;
dma_cookie_t cookie;
enum dma_status status;
enum dma_ctrl_flags flags;
u8 *pq_coefs = NULL;
int ret;
int src_cnt;
int dst_cnt;
int i;
ktime_t ktime;
s64 runtime = 0;
unsigned long long total_len = 0;
dmatest: don't use set_freezable_with_signal() Commit 981ed70d8e (dmatest: make dmatest threads freezable) made dmatest kthread use set_freezable_with_signal(); however, the interface is scheduled to be removed in the next merge window. The problem is that unlike userland tasks there's no default place which handles signal pending state and it isn't clear who owns and/or is responsible for clearing TIF_SIGPENDING. For example, in the current code, try_to_freeze() clears TIF_SIGPENDING but it isn't sure whether it actually owns the TIF_SIGPENDING nor is it race-free - ie. the task may continue to run with TIF_SIGPENDING set after the freezable section. Unfortunately, we don't have wait_for_completion_freezable_timeout(). This patch open codes it and uses wait_event_freezable_timeout() instead and removes timeout reloading - wait_event_freezable_timeout() won't return across freezing events (currently racy but fix scheduled) and timer doesn't decrement while the task is in freezer. Although this does lose timer-reset-over-freezing, given that timeout is supposed to be long enough and failure to finish inside is considered irrecoverable, I don't think this is worth the complexity. While at it, move completion to outer scope and explain that we're ignoring dangling pointer problem after timeout. This should give slightly better chance at avoiding oops after timeout. Signed-off-by: Tejun Heo <tj@kernel.org> Acked-by: Dan Williams <dan.j.williams@intel.com> Cc: Guennadi Liakhovetski <g.liakhovetski@gmx.de> Cc: Nicolas Ferre <nicolas.ferre@atmel.com>
2011-11-23 17:28:16 +00:00
set_freezable();
ret = -ENOMEM;
smp_rmb();
info = thread->info;
params = &info->params;
chan = thread->chan;
dev = chan->device;
if (thread->type == DMA_MEMCPY)
src_cnt = dst_cnt = 1;
else if (thread->type == DMA_XOR) {
/* force odd to ensure dst = src */
src_cnt = min_odd(params->xor_sources | 1, dev->max_xor);
dst_cnt = 1;
} else if (thread->type == DMA_PQ) {
/* force odd to ensure dst = src */
src_cnt = min_odd(params->pq_sources | 1, dma_maxpq(dev, 0));
dst_cnt = 2;
pq_coefs = kmalloc(params->pq_sources+1, GFP_KERNEL);
if (!pq_coefs)
goto err_thread_type;
for (i = 0; i < src_cnt; i++)
pq_coefs[i] = 1;
} else
goto err_thread_type;
thread->srcs = kcalloc(src_cnt+1, sizeof(u8 *), GFP_KERNEL);
if (!thread->srcs)
goto err_srcs;
for (i = 0; i < src_cnt; i++) {
thread->srcs[i] = kmalloc(params->buf_size, GFP_KERNEL);
if (!thread->srcs[i])
goto err_srcbuf;
}
thread->srcs[i] = NULL;
thread->dsts = kcalloc(dst_cnt+1, sizeof(u8 *), GFP_KERNEL);
if (!thread->dsts)
goto err_dsts;
for (i = 0; i < dst_cnt; i++) {
thread->dsts[i] = kmalloc(params->buf_size, GFP_KERNEL);
if (!thread->dsts[i])
goto err_dstbuf;
}
thread->dsts[i] = NULL;
set_user_nice(current, 10);
/*
* src and dst buffers are freed by ourselves below
*/
flags = DMA_CTRL_ACK | DMA_PREP_INTERRUPT;
ktime = ktime_get();
while (!kthread_should_stop()
&& !(params->iterations && total_tests >= params->iterations)) {
struct dma_async_tx_descriptor *tx = NULL;
struct dmaengine_unmap_data *um;
dma_addr_t srcs[src_cnt];
dma_addr_t *dsts;
unsigned int src_off, dst_off, len;
u8 align = 0;
total_tests++;
/* honor alignment restrictions */
if (thread->type == DMA_MEMCPY)
align = dev->copy_align;
else if (thread->type == DMA_XOR)
align = dev->xor_align;
else if (thread->type == DMA_PQ)
align = dev->pq_align;
if (1 << align > params->buf_size) {
pr_err("%u-byte buffer too small for %d-byte alignment\n",
params->buf_size, 1 << align);
break;
}
if (params->noverify)
len = params->buf_size;
else
len = dmatest_random() % params->buf_size + 1;
len = (len >> align) << align;
if (!len)
len = 1 << align;
total_len += len;
if (params->noverify) {
src_off = 0;
dst_off = 0;
} else {
src_off = dmatest_random() % (params->buf_size - len + 1);
dst_off = dmatest_random() % (params->buf_size - len + 1);
src_off = (src_off >> align) << align;
dst_off = (dst_off >> align) << align;
dmatest_init_srcs(thread->srcs, src_off, len,
params->buf_size);
dmatest_init_dsts(thread->dsts, dst_off, len,
params->buf_size);
}
um = dmaengine_get_unmap_data(dev->dev, src_cnt+dst_cnt,
GFP_KERNEL);
if (!um) {
failed_tests++;
result("unmap data NULL", total_tests,
src_off, dst_off, len, ret);
continue;
}
um->len = params->buf_size;
for (i = 0; i < src_cnt; i++) {
void *buf = thread->srcs[i];
struct page *pg = virt_to_page(buf);
unsigned pg_off = (unsigned long) buf & ~PAGE_MASK;
um->addr[i] = dma_map_page(dev->dev, pg, pg_off,
um->len, DMA_TO_DEVICE);
srcs[i] = um->addr[i] + src_off;
ret = dma_mapping_error(dev->dev, um->addr[i]);
if (ret) {
dmaengine_unmap_put(um);
result("src mapping error", total_tests,
src_off, dst_off, len, ret);
failed_tests++;
continue;
}
um->to_cnt++;
}
/* map with DMA_BIDIRECTIONAL to force writeback/invalidate */
dsts = &um->addr[src_cnt];
for (i = 0; i < dst_cnt; i++) {
void *buf = thread->dsts[i];
struct page *pg = virt_to_page(buf);
unsigned pg_off = (unsigned long) buf & ~PAGE_MASK;
dsts[i] = dma_map_page(dev->dev, pg, pg_off, um->len,
DMA_BIDIRECTIONAL);
ret = dma_mapping_error(dev->dev, dsts[i]);
if (ret) {
dmaengine_unmap_put(um);
result("dst mapping error", total_tests,
src_off, dst_off, len, ret);
failed_tests++;
continue;
}
um->bidi_cnt++;
}
if (thread->type == DMA_MEMCPY)
tx = dev->device_prep_dma_memcpy(chan,
dsts[0] + dst_off,
srcs[0], len, flags);
else if (thread->type == DMA_XOR)
tx = dev->device_prep_dma_xor(chan,
dsts[0] + dst_off,
srcs, src_cnt,
len, flags);
else if (thread->type == DMA_PQ) {
dma_addr_t dma_pq[dst_cnt];
for (i = 0; i < dst_cnt; i++)
dma_pq[i] = dsts[i] + dst_off;
tx = dev->device_prep_dma_pq(chan, dma_pq, srcs,
src_cnt, pq_coefs,
len, flags);
}
if (!tx) {
dmaengine_unmap_put(um);
result("prep error", total_tests, src_off,
dst_off, len, ret);
msleep(100);
failed_tests++;
continue;
}
dmatest: don't use set_freezable_with_signal() Commit 981ed70d8e (dmatest: make dmatest threads freezable) made dmatest kthread use set_freezable_with_signal(); however, the interface is scheduled to be removed in the next merge window. The problem is that unlike userland tasks there's no default place which handles signal pending state and it isn't clear who owns and/or is responsible for clearing TIF_SIGPENDING. For example, in the current code, try_to_freeze() clears TIF_SIGPENDING but it isn't sure whether it actually owns the TIF_SIGPENDING nor is it race-free - ie. the task may continue to run with TIF_SIGPENDING set after the freezable section. Unfortunately, we don't have wait_for_completion_freezable_timeout(). This patch open codes it and uses wait_event_freezable_timeout() instead and removes timeout reloading - wait_event_freezable_timeout() won't return across freezing events (currently racy but fix scheduled) and timer doesn't decrement while the task is in freezer. Although this does lose timer-reset-over-freezing, given that timeout is supposed to be long enough and failure to finish inside is considered irrecoverable, I don't think this is worth the complexity. While at it, move completion to outer scope and explain that we're ignoring dangling pointer problem after timeout. This should give slightly better chance at avoiding oops after timeout. Signed-off-by: Tejun Heo <tj@kernel.org> Acked-by: Dan Williams <dan.j.williams@intel.com> Cc: Guennadi Liakhovetski <g.liakhovetski@gmx.de> Cc: Nicolas Ferre <nicolas.ferre@atmel.com>
2011-11-23 17:28:16 +00:00
done.done = false;
tx->callback = dmatest_callback;
dmatest: don't use set_freezable_with_signal() Commit 981ed70d8e (dmatest: make dmatest threads freezable) made dmatest kthread use set_freezable_with_signal(); however, the interface is scheduled to be removed in the next merge window. The problem is that unlike userland tasks there's no default place which handles signal pending state and it isn't clear who owns and/or is responsible for clearing TIF_SIGPENDING. For example, in the current code, try_to_freeze() clears TIF_SIGPENDING but it isn't sure whether it actually owns the TIF_SIGPENDING nor is it race-free - ie. the task may continue to run with TIF_SIGPENDING set after the freezable section. Unfortunately, we don't have wait_for_completion_freezable_timeout(). This patch open codes it and uses wait_event_freezable_timeout() instead and removes timeout reloading - wait_event_freezable_timeout() won't return across freezing events (currently racy but fix scheduled) and timer doesn't decrement while the task is in freezer. Although this does lose timer-reset-over-freezing, given that timeout is supposed to be long enough and failure to finish inside is considered irrecoverable, I don't think this is worth the complexity. While at it, move completion to outer scope and explain that we're ignoring dangling pointer problem after timeout. This should give slightly better chance at avoiding oops after timeout. Signed-off-by: Tejun Heo <tj@kernel.org> Acked-by: Dan Williams <dan.j.williams@intel.com> Cc: Guennadi Liakhovetski <g.liakhovetski@gmx.de> Cc: Nicolas Ferre <nicolas.ferre@atmel.com>
2011-11-23 17:28:16 +00:00
tx->callback_param = &done;
cookie = tx->tx_submit(tx);
if (dma_submit_error(cookie)) {
dmaengine_unmap_put(um);
result("submit error", total_tests, src_off,
dst_off, len, ret);
msleep(100);
failed_tests++;
continue;
}
dma_async_issue_pending(chan);
wait_event_freezable_timeout(done_wait, done.done,
msecs_to_jiffies(params->timeout));
status = dma_async_is_tx_complete(chan, cookie, NULL, NULL);
dmatest: don't use set_freezable_with_signal() Commit 981ed70d8e (dmatest: make dmatest threads freezable) made dmatest kthread use set_freezable_with_signal(); however, the interface is scheduled to be removed in the next merge window. The problem is that unlike userland tasks there's no default place which handles signal pending state and it isn't clear who owns and/or is responsible for clearing TIF_SIGPENDING. For example, in the current code, try_to_freeze() clears TIF_SIGPENDING but it isn't sure whether it actually owns the TIF_SIGPENDING nor is it race-free - ie. the task may continue to run with TIF_SIGPENDING set after the freezable section. Unfortunately, we don't have wait_for_completion_freezable_timeout(). This patch open codes it and uses wait_event_freezable_timeout() instead and removes timeout reloading - wait_event_freezable_timeout() won't return across freezing events (currently racy but fix scheduled) and timer doesn't decrement while the task is in freezer. Although this does lose timer-reset-over-freezing, given that timeout is supposed to be long enough and failure to finish inside is considered irrecoverable, I don't think this is worth the complexity. While at it, move completion to outer scope and explain that we're ignoring dangling pointer problem after timeout. This should give slightly better chance at avoiding oops after timeout. Signed-off-by: Tejun Heo <tj@kernel.org> Acked-by: Dan Williams <dan.j.williams@intel.com> Cc: Guennadi Liakhovetski <g.liakhovetski@gmx.de> Cc: Nicolas Ferre <nicolas.ferre@atmel.com>
2011-11-23 17:28:16 +00:00
if (!done.done) {
/*
* We're leaving the timed out dma operation with
* dangling pointer to done_wait. To make this
* correct, we'll need to allocate wait_done for
* each test iteration and perform "who's gonna
* free it this time?" dancing. For now, just
* leave it dangling.
*/
dmaengine_unmap_put(um);
result("test timed out", total_tests, src_off, dst_off,
len, 0);
failed_tests++;
continue;
} else if (status != DMA_COMPLETE) {
dmaengine_unmap_put(um);
result(status == DMA_ERROR ?
"completion error status" :
"completion busy status", total_tests, src_off,
dst_off, len, ret);
failed_tests++;
continue;
}
dmaengine_unmap_put(um);
if (params->noverify) {
verbose_result("test passed", total_tests, src_off,
dst_off, len, 0);
continue;
}
pr_debug("%s: verifying source buffer...\n", current->comm);
error_count = dmatest_verify(thread->srcs, 0, src_off,
0, PATTERN_SRC, true);
error_count += dmatest_verify(thread->srcs, src_off,
src_off + len, src_off,
PATTERN_SRC | PATTERN_COPY, true);
error_count += dmatest_verify(thread->srcs, src_off + len,
params->buf_size, src_off + len,
PATTERN_SRC, true);
pr_debug("%s: verifying dest buffer...\n", current->comm);
error_count += dmatest_verify(thread->dsts, 0, dst_off,
0, PATTERN_DST, false);
error_count += dmatest_verify(thread->dsts, dst_off,
dst_off + len, src_off,
PATTERN_SRC | PATTERN_COPY, false);
error_count += dmatest_verify(thread->dsts, dst_off + len,
params->buf_size, dst_off + len,
PATTERN_DST, false);
if (error_count) {
result("data error", total_tests, src_off, dst_off,
len, error_count);
failed_tests++;
} else {
verbose_result("test passed", total_tests, src_off,
dst_off, len, 0);
}
}
runtime = ktime_us_delta(ktime_get(), ktime);
ret = 0;
err_dstbuf:
for (i = 0; thread->dsts[i]; i++)
kfree(thread->dsts[i]);
kfree(thread->dsts);
err_dsts:
err_srcbuf:
for (i = 0; thread->srcs[i]; i++)
kfree(thread->srcs[i]);
kfree(thread->srcs);
err_srcs:
kfree(pq_coefs);
err_thread_type:
pr_info("%s: summary %u tests, %u failures %llu iops %llu KB/s (%d)\n",
current->comm, total_tests, failed_tests,
dmatest_persec(runtime, total_tests),
dmatest_KBs(runtime, total_len), ret);
/* terminate all transfers on specified channels */
if (ret)
dmaengine_terminate_all(chan);
thread->done = true;
wake_up(&thread_wait);
return ret;
}
static void dmatest_cleanup_channel(struct dmatest_chan *dtc)
{
struct dmatest_thread *thread;
struct dmatest_thread *_thread;
int ret;
list_for_each_entry_safe(thread, _thread, &dtc->threads, node) {
ret = kthread_stop(thread->task);
pr_debug("thread %s exited with status %d\n",
thread->task->comm, ret);
list_del(&thread->node);
put_task_struct(thread->task);
kfree(thread);
}
/* terminate all transfers on specified channels */
dmaengine_terminate_all(dtc->chan);
kfree(dtc);
}
static int dmatest_add_threads(struct dmatest_info *info,
struct dmatest_chan *dtc, enum dma_transaction_type type)
{
struct dmatest_params *params = &info->params;
struct dmatest_thread *thread;
struct dma_chan *chan = dtc->chan;
char *op;
unsigned int i;
if (type == DMA_MEMCPY)
op = "copy";
else if (type == DMA_XOR)
op = "xor";
else if (type == DMA_PQ)
op = "pq";
else
return -EINVAL;
for (i = 0; i < params->threads_per_chan; i++) {
thread = kzalloc(sizeof(struct dmatest_thread), GFP_KERNEL);
if (!thread) {
pr_warn("No memory for %s-%s%u\n",
dma_chan_name(chan), op, i);
break;
}
thread->info = info;
thread->chan = dtc->chan;
thread->type = type;
smp_wmb();
thread->task = kthread_create(dmatest_func, thread, "%s-%s%u",
dma_chan_name(chan), op, i);
if (IS_ERR(thread->task)) {
pr_warn("Failed to create thread %s-%s%u\n",
dma_chan_name(chan), op, i);
kfree(thread);
break;
}
/* srcbuf and dstbuf are allocated by the thread itself */
get_task_struct(thread->task);
list_add_tail(&thread->node, &dtc->threads);
wake_up_process(thread->task);
}
return i;
}
static int dmatest_add_channel(struct dmatest_info *info,
struct dma_chan *chan)
{
struct dmatest_chan *dtc;
struct dma_device *dma_dev = chan->device;
unsigned int thread_count = 0;
int cnt;
dtc = kmalloc(sizeof(struct dmatest_chan), GFP_KERNEL);
if (!dtc) {
pr_warn("No memory for %s\n", dma_chan_name(chan));
return -ENOMEM;
}
dtc->chan = chan;
INIT_LIST_HEAD(&dtc->threads);
if (dma_has_cap(DMA_MEMCPY, dma_dev->cap_mask)) {
cnt = dmatest_add_threads(info, dtc, DMA_MEMCPY);
thread_count += cnt > 0 ? cnt : 0;
}
if (dma_has_cap(DMA_XOR, dma_dev->cap_mask)) {
cnt = dmatest_add_threads(info, dtc, DMA_XOR);
thread_count += cnt > 0 ? cnt : 0;
}
if (dma_has_cap(DMA_PQ, dma_dev->cap_mask)) {
cnt = dmatest_add_threads(info, dtc, DMA_PQ);
thread_count += cnt > 0 ? cnt : 0;
}
pr_info("Started %u threads using %s\n",
thread_count, dma_chan_name(chan));
list_add_tail(&dtc->node, &info->channels);
info->nr_channels++;
return 0;
}
static bool filter(struct dma_chan *chan, void *param)
{
struct dmatest_params *params = param;
if (!dmatest_match_channel(params, chan) ||
!dmatest_match_device(params, chan->device))
return false;
else
return true;
}
static void request_channels(struct dmatest_info *info,
enum dma_transaction_type type)
{
dma_cap_mask_t mask;
dma_cap_zero(mask);
dma_cap_set(type, mask);
for (;;) {
struct dmatest_params *params = &info->params;
struct dma_chan *chan;
chan = dma_request_channel(mask, filter, params);
if (chan) {
if (dmatest_add_channel(info, chan)) {
dma_release_channel(chan);
break; /* add_channel failed, punt */
}
} else
break; /* no more channels available */
if (params->max_channels &&
info->nr_channels >= params->max_channels)
break; /* we have all we need */
}
}
static void run_threaded_test(struct dmatest_info *info)
dmatest: run test via debugfs Instead of doing modprobe dmatest ... modprobe -r dmatest we allow user to run tests interactively. The dmatest could be built as module or inside kernel. Let's consider those cases. 1. When dmatest is built as a module... After mounting debugfs and loading the module, the /sys/kernel/debug/dmatest folder with nodes will be created. They are the same as module parameters with addition of the 'run' node that controls run and stop phases of the test. Note that in this case test will not run on load automatically. Example of usage: % echo dma0chan0 > /sys/kernel/debug/dmatest/channel % echo 2000 > /sys/kernel/debug/dmatest/timeout % echo 1 > /sys/kernel/debug/dmatest/iterations % echo 1 > /sys/kernel/debug/dmatest/run After a while you will start to get messages about current status or error like in the original code. Note that running a new test will stop any in progress test. 2. When built-in in the kernel... The module parameters that is supplied to the kernel command line will be used for the first performed test. After user gets a control, the test could be interrupted or re-run with same or different parameters. For the details see the above section "1. When dmatest is built as a module..." In both cases the module parameters are used as initial values for the test case. You always could check them at run-time by running % grep -H . /sys/module/dmatest/parameters/* Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2013-03-04 09:09:30 +00:00
{
struct dmatest_params *params = &info->params;
dmatest: run test via debugfs Instead of doing modprobe dmatest ... modprobe -r dmatest we allow user to run tests interactively. The dmatest could be built as module or inside kernel. Let's consider those cases. 1. When dmatest is built as a module... After mounting debugfs and loading the module, the /sys/kernel/debug/dmatest folder with nodes will be created. They are the same as module parameters with addition of the 'run' node that controls run and stop phases of the test. Note that in this case test will not run on load automatically. Example of usage: % echo dma0chan0 > /sys/kernel/debug/dmatest/channel % echo 2000 > /sys/kernel/debug/dmatest/timeout % echo 1 > /sys/kernel/debug/dmatest/iterations % echo 1 > /sys/kernel/debug/dmatest/run After a while you will start to get messages about current status or error like in the original code. Note that running a new test will stop any in progress test. 2. When built-in in the kernel... The module parameters that is supplied to the kernel command line will be used for the first performed test. After user gets a control, the test could be interrupted or re-run with same or different parameters. For the details see the above section "1. When dmatest is built as a module..." In both cases the module parameters are used as initial values for the test case. You always could check them at run-time by running % grep -H . /sys/module/dmatest/parameters/* Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2013-03-04 09:09:30 +00:00
/* Copy test parameters */
params->buf_size = test_buf_size;
strlcpy(params->channel, strim(test_channel), sizeof(params->channel));
strlcpy(params->device, strim(test_device), sizeof(params->device));
params->threads_per_chan = threads_per_chan;
params->max_channels = max_channels;
params->iterations = iterations;
params->xor_sources = xor_sources;
params->pq_sources = pq_sources;
params->timeout = timeout;
params->noverify = noverify;
request_channels(info, DMA_MEMCPY);
request_channels(info, DMA_XOR);
request_channels(info, DMA_PQ);
dmatest: run test via debugfs Instead of doing modprobe dmatest ... modprobe -r dmatest we allow user to run tests interactively. The dmatest could be built as module or inside kernel. Let's consider those cases. 1. When dmatest is built as a module... After mounting debugfs and loading the module, the /sys/kernel/debug/dmatest folder with nodes will be created. They are the same as module parameters with addition of the 'run' node that controls run and stop phases of the test. Note that in this case test will not run on load automatically. Example of usage: % echo dma0chan0 > /sys/kernel/debug/dmatest/channel % echo 2000 > /sys/kernel/debug/dmatest/timeout % echo 1 > /sys/kernel/debug/dmatest/iterations % echo 1 > /sys/kernel/debug/dmatest/run After a while you will start to get messages about current status or error like in the original code. Note that running a new test will stop any in progress test. 2. When built-in in the kernel... The module parameters that is supplied to the kernel command line will be used for the first performed test. After user gets a control, the test could be interrupted or re-run with same or different parameters. For the details see the above section "1. When dmatest is built as a module..." In both cases the module parameters are used as initial values for the test case. You always could check them at run-time by running % grep -H . /sys/module/dmatest/parameters/* Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2013-03-04 09:09:30 +00:00
}
static void stop_threaded_test(struct dmatest_info *info)
{
struct dmatest_chan *dtc, *_dtc;
struct dma_chan *chan;
list_for_each_entry_safe(dtc, _dtc, &info->channels, node) {
list_del(&dtc->node);
chan = dtc->chan;
dmatest_cleanup_channel(dtc);
pr_debug("dropped channel %s\n", dma_chan_name(chan));
dma_release_channel(chan);
}
info->nr_channels = 0;
}
static void restart_threaded_test(struct dmatest_info *info, bool run)
dmatest: run test via debugfs Instead of doing modprobe dmatest ... modprobe -r dmatest we allow user to run tests interactively. The dmatest could be built as module or inside kernel. Let's consider those cases. 1. When dmatest is built as a module... After mounting debugfs and loading the module, the /sys/kernel/debug/dmatest folder with nodes will be created. They are the same as module parameters with addition of the 'run' node that controls run and stop phases of the test. Note that in this case test will not run on load automatically. Example of usage: % echo dma0chan0 > /sys/kernel/debug/dmatest/channel % echo 2000 > /sys/kernel/debug/dmatest/timeout % echo 1 > /sys/kernel/debug/dmatest/iterations % echo 1 > /sys/kernel/debug/dmatest/run After a while you will start to get messages about current status or error like in the original code. Note that running a new test will stop any in progress test. 2. When built-in in the kernel... The module parameters that is supplied to the kernel command line will be used for the first performed test. After user gets a control, the test could be interrupted or re-run with same or different parameters. For the details see the above section "1. When dmatest is built as a module..." In both cases the module parameters are used as initial values for the test case. You always could check them at run-time by running % grep -H . /sys/module/dmatest/parameters/* Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2013-03-04 09:09:30 +00:00
{
/* we might be called early to set run=, defer running until all
* parameters have been evaluated
*/
if (!info->did_init)
return;
dmatest: run test via debugfs Instead of doing modprobe dmatest ... modprobe -r dmatest we allow user to run tests interactively. The dmatest could be built as module or inside kernel. Let's consider those cases. 1. When dmatest is built as a module... After mounting debugfs and loading the module, the /sys/kernel/debug/dmatest folder with nodes will be created. They are the same as module parameters with addition of the 'run' node that controls run and stop phases of the test. Note that in this case test will not run on load automatically. Example of usage: % echo dma0chan0 > /sys/kernel/debug/dmatest/channel % echo 2000 > /sys/kernel/debug/dmatest/timeout % echo 1 > /sys/kernel/debug/dmatest/iterations % echo 1 > /sys/kernel/debug/dmatest/run After a while you will start to get messages about current status or error like in the original code. Note that running a new test will stop any in progress test. 2. When built-in in the kernel... The module parameters that is supplied to the kernel command line will be used for the first performed test. After user gets a control, the test could be interrupted or re-run with same or different parameters. For the details see the above section "1. When dmatest is built as a module..." In both cases the module parameters are used as initial values for the test case. You always could check them at run-time by running % grep -H . /sys/module/dmatest/parameters/* Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2013-03-04 09:09:30 +00:00
/* Stop any running test first */
stop_threaded_test(info);
dmatest: run test via debugfs Instead of doing modprobe dmatest ... modprobe -r dmatest we allow user to run tests interactively. The dmatest could be built as module or inside kernel. Let's consider those cases. 1. When dmatest is built as a module... After mounting debugfs and loading the module, the /sys/kernel/debug/dmatest folder with nodes will be created. They are the same as module parameters with addition of the 'run' node that controls run and stop phases of the test. Note that in this case test will not run on load automatically. Example of usage: % echo dma0chan0 > /sys/kernel/debug/dmatest/channel % echo 2000 > /sys/kernel/debug/dmatest/timeout % echo 1 > /sys/kernel/debug/dmatest/iterations % echo 1 > /sys/kernel/debug/dmatest/run After a while you will start to get messages about current status or error like in the original code. Note that running a new test will stop any in progress test. 2. When built-in in the kernel... The module parameters that is supplied to the kernel command line will be used for the first performed test. After user gets a control, the test could be interrupted or re-run with same or different parameters. For the details see the above section "1. When dmatest is built as a module..." In both cases the module parameters are used as initial values for the test case. You always could check them at run-time by running % grep -H . /sys/module/dmatest/parameters/* Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2013-03-04 09:09:30 +00:00
/* Run test with new parameters */
run_threaded_test(info);
dmatest: run test via debugfs Instead of doing modprobe dmatest ... modprobe -r dmatest we allow user to run tests interactively. The dmatest could be built as module or inside kernel. Let's consider those cases. 1. When dmatest is built as a module... After mounting debugfs and loading the module, the /sys/kernel/debug/dmatest folder with nodes will be created. They are the same as module parameters with addition of the 'run' node that controls run and stop phases of the test. Note that in this case test will not run on load automatically. Example of usage: % echo dma0chan0 > /sys/kernel/debug/dmatest/channel % echo 2000 > /sys/kernel/debug/dmatest/timeout % echo 1 > /sys/kernel/debug/dmatest/iterations % echo 1 > /sys/kernel/debug/dmatest/run After a while you will start to get messages about current status or error like in the original code. Note that running a new test will stop any in progress test. 2. When built-in in the kernel... The module parameters that is supplied to the kernel command line will be used for the first performed test. After user gets a control, the test could be interrupted or re-run with same or different parameters. For the details see the above section "1. When dmatest is built as a module..." In both cases the module parameters are used as initial values for the test case. You always could check them at run-time by running % grep -H . /sys/module/dmatest/parameters/* Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2013-03-04 09:09:30 +00:00
}
static int dmatest_run_get(char *val, const struct kernel_param *kp)
dmatest: run test via debugfs Instead of doing modprobe dmatest ... modprobe -r dmatest we allow user to run tests interactively. The dmatest could be built as module or inside kernel. Let's consider those cases. 1. When dmatest is built as a module... After mounting debugfs and loading the module, the /sys/kernel/debug/dmatest folder with nodes will be created. They are the same as module parameters with addition of the 'run' node that controls run and stop phases of the test. Note that in this case test will not run on load automatically. Example of usage: % echo dma0chan0 > /sys/kernel/debug/dmatest/channel % echo 2000 > /sys/kernel/debug/dmatest/timeout % echo 1 > /sys/kernel/debug/dmatest/iterations % echo 1 > /sys/kernel/debug/dmatest/run After a while you will start to get messages about current status or error like in the original code. Note that running a new test will stop any in progress test. 2. When built-in in the kernel... The module parameters that is supplied to the kernel command line will be used for the first performed test. After user gets a control, the test could be interrupted or re-run with same or different parameters. For the details see the above section "1. When dmatest is built as a module..." In both cases the module parameters are used as initial values for the test case. You always could check them at run-time by running % grep -H . /sys/module/dmatest/parameters/* Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2013-03-04 09:09:30 +00:00
{
struct dmatest_info *info = &test_info;
dmatest: run test via debugfs Instead of doing modprobe dmatest ... modprobe -r dmatest we allow user to run tests interactively. The dmatest could be built as module or inside kernel. Let's consider those cases. 1. When dmatest is built as a module... After mounting debugfs and loading the module, the /sys/kernel/debug/dmatest folder with nodes will be created. They are the same as module parameters with addition of the 'run' node that controls run and stop phases of the test. Note that in this case test will not run on load automatically. Example of usage: % echo dma0chan0 > /sys/kernel/debug/dmatest/channel % echo 2000 > /sys/kernel/debug/dmatest/timeout % echo 1 > /sys/kernel/debug/dmatest/iterations % echo 1 > /sys/kernel/debug/dmatest/run After a while you will start to get messages about current status or error like in the original code. Note that running a new test will stop any in progress test. 2. When built-in in the kernel... The module parameters that is supplied to the kernel command line will be used for the first performed test. After user gets a control, the test could be interrupted or re-run with same or different parameters. For the details see the above section "1. When dmatest is built as a module..." In both cases the module parameters are used as initial values for the test case. You always could check them at run-time by running % grep -H . /sys/module/dmatest/parameters/* Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2013-03-04 09:09:30 +00:00
mutex_lock(&info->lock);
if (is_threaded_test_run(info)) {
dmatest_run = true;
} else {
stop_threaded_test(info);
dmatest_run = false;
}
dmatest: run test via debugfs Instead of doing modprobe dmatest ... modprobe -r dmatest we allow user to run tests interactively. The dmatest could be built as module or inside kernel. Let's consider those cases. 1. When dmatest is built as a module... After mounting debugfs and loading the module, the /sys/kernel/debug/dmatest folder with nodes will be created. They are the same as module parameters with addition of the 'run' node that controls run and stop phases of the test. Note that in this case test will not run on load automatically. Example of usage: % echo dma0chan0 > /sys/kernel/debug/dmatest/channel % echo 2000 > /sys/kernel/debug/dmatest/timeout % echo 1 > /sys/kernel/debug/dmatest/iterations % echo 1 > /sys/kernel/debug/dmatest/run After a while you will start to get messages about current status or error like in the original code. Note that running a new test will stop any in progress test. 2. When built-in in the kernel... The module parameters that is supplied to the kernel command line will be used for the first performed test. After user gets a control, the test could be interrupted or re-run with same or different parameters. For the details see the above section "1. When dmatest is built as a module..." In both cases the module parameters are used as initial values for the test case. You always could check them at run-time by running % grep -H . /sys/module/dmatest/parameters/* Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2013-03-04 09:09:30 +00:00
mutex_unlock(&info->lock);
return param_get_bool(val, kp);
dmatest: run test via debugfs Instead of doing modprobe dmatest ... modprobe -r dmatest we allow user to run tests interactively. The dmatest could be built as module or inside kernel. Let's consider those cases. 1. When dmatest is built as a module... After mounting debugfs and loading the module, the /sys/kernel/debug/dmatest folder with nodes will be created. They are the same as module parameters with addition of the 'run' node that controls run and stop phases of the test. Note that in this case test will not run on load automatically. Example of usage: % echo dma0chan0 > /sys/kernel/debug/dmatest/channel % echo 2000 > /sys/kernel/debug/dmatest/timeout % echo 1 > /sys/kernel/debug/dmatest/iterations % echo 1 > /sys/kernel/debug/dmatest/run After a while you will start to get messages about current status or error like in the original code. Note that running a new test will stop any in progress test. 2. When built-in in the kernel... The module parameters that is supplied to the kernel command line will be used for the first performed test. After user gets a control, the test could be interrupted or re-run with same or different parameters. For the details see the above section "1. When dmatest is built as a module..." In both cases the module parameters are used as initial values for the test case. You always could check them at run-time by running % grep -H . /sys/module/dmatest/parameters/* Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2013-03-04 09:09:30 +00:00
}
static int dmatest_run_set(const char *val, const struct kernel_param *kp)
{
struct dmatest_info *info = &test_info;
int ret;
mutex_lock(&info->lock);
ret = param_set_bool(val, kp);
if (ret) {
dmatest: run test via debugfs Instead of doing modprobe dmatest ... modprobe -r dmatest we allow user to run tests interactively. The dmatest could be built as module or inside kernel. Let's consider those cases. 1. When dmatest is built as a module... After mounting debugfs and loading the module, the /sys/kernel/debug/dmatest folder with nodes will be created. They are the same as module parameters with addition of the 'run' node that controls run and stop phases of the test. Note that in this case test will not run on load automatically. Example of usage: % echo dma0chan0 > /sys/kernel/debug/dmatest/channel % echo 2000 > /sys/kernel/debug/dmatest/timeout % echo 1 > /sys/kernel/debug/dmatest/iterations % echo 1 > /sys/kernel/debug/dmatest/run After a while you will start to get messages about current status or error like in the original code. Note that running a new test will stop any in progress test. 2. When built-in in the kernel... The module parameters that is supplied to the kernel command line will be used for the first performed test. After user gets a control, the test could be interrupted or re-run with same or different parameters. For the details see the above section "1. When dmatest is built as a module..." In both cases the module parameters are used as initial values for the test case. You always could check them at run-time by running % grep -H . /sys/module/dmatest/parameters/* Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2013-03-04 09:09:30 +00:00
mutex_unlock(&info->lock);
return ret;
}
if (is_threaded_test_run(info))
ret = -EBUSY;
else if (dmatest_run)
restart_threaded_test(info, dmatest_run);
dmatest: run test via debugfs Instead of doing modprobe dmatest ... modprobe -r dmatest we allow user to run tests interactively. The dmatest could be built as module or inside kernel. Let's consider those cases. 1. When dmatest is built as a module... After mounting debugfs and loading the module, the /sys/kernel/debug/dmatest folder with nodes will be created. They are the same as module parameters with addition of the 'run' node that controls run and stop phases of the test. Note that in this case test will not run on load automatically. Example of usage: % echo dma0chan0 > /sys/kernel/debug/dmatest/channel % echo 2000 > /sys/kernel/debug/dmatest/timeout % echo 1 > /sys/kernel/debug/dmatest/iterations % echo 1 > /sys/kernel/debug/dmatest/run After a while you will start to get messages about current status or error like in the original code. Note that running a new test will stop any in progress test. 2. When built-in in the kernel... The module parameters that is supplied to the kernel command line will be used for the first performed test. After user gets a control, the test could be interrupted or re-run with same or different parameters. For the details see the above section "1. When dmatest is built as a module..." In both cases the module parameters are used as initial values for the test case. You always could check them at run-time by running % grep -H . /sys/module/dmatest/parameters/* Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2013-03-04 09:09:30 +00:00
mutex_unlock(&info->lock);
dmatest: run test via debugfs Instead of doing modprobe dmatest ... modprobe -r dmatest we allow user to run tests interactively. The dmatest could be built as module or inside kernel. Let's consider those cases. 1. When dmatest is built as a module... After mounting debugfs and loading the module, the /sys/kernel/debug/dmatest folder with nodes will be created. They are the same as module parameters with addition of the 'run' node that controls run and stop phases of the test. Note that in this case test will not run on load automatically. Example of usage: % echo dma0chan0 > /sys/kernel/debug/dmatest/channel % echo 2000 > /sys/kernel/debug/dmatest/timeout % echo 1 > /sys/kernel/debug/dmatest/iterations % echo 1 > /sys/kernel/debug/dmatest/run After a while you will start to get messages about current status or error like in the original code. Note that running a new test will stop any in progress test. 2. When built-in in the kernel... The module parameters that is supplied to the kernel command line will be used for the first performed test. After user gets a control, the test could be interrupted or re-run with same or different parameters. For the details see the above section "1. When dmatest is built as a module..." In both cases the module parameters are used as initial values for the test case. You always could check them at run-time by running % grep -H . /sys/module/dmatest/parameters/* Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2013-03-04 09:09:30 +00:00
return ret;
dmatest: run test via debugfs Instead of doing modprobe dmatest ... modprobe -r dmatest we allow user to run tests interactively. The dmatest could be built as module or inside kernel. Let's consider those cases. 1. When dmatest is built as a module... After mounting debugfs and loading the module, the /sys/kernel/debug/dmatest folder with nodes will be created. They are the same as module parameters with addition of the 'run' node that controls run and stop phases of the test. Note that in this case test will not run on load automatically. Example of usage: % echo dma0chan0 > /sys/kernel/debug/dmatest/channel % echo 2000 > /sys/kernel/debug/dmatest/timeout % echo 1 > /sys/kernel/debug/dmatest/iterations % echo 1 > /sys/kernel/debug/dmatest/run After a while you will start to get messages about current status or error like in the original code. Note that running a new test will stop any in progress test. 2. When built-in in the kernel... The module parameters that is supplied to the kernel command line will be used for the first performed test. After user gets a control, the test could be interrupted or re-run with same or different parameters. For the details see the above section "1. When dmatest is built as a module..." In both cases the module parameters are used as initial values for the test case. You always could check them at run-time by running % grep -H . /sys/module/dmatest/parameters/* Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2013-03-04 09:09:30 +00:00
}
static int __init dmatest_init(void)
{
struct dmatest_info *info = &test_info;
struct dmatest_params *params = &info->params;
if (dmatest_run) {
mutex_lock(&info->lock);
run_threaded_test(info);
mutex_unlock(&info->lock);
}
if (params->iterations && wait)
wait_event(thread_wait, !is_threaded_test_run(info));
/* module parameters are stable, inittime tests are started,
* let userspace take over 'run' control
*/
info->did_init = true;
dmatest: run test via debugfs Instead of doing modprobe dmatest ... modprobe -r dmatest we allow user to run tests interactively. The dmatest could be built as module or inside kernel. Let's consider those cases. 1. When dmatest is built as a module... After mounting debugfs and loading the module, the /sys/kernel/debug/dmatest folder with nodes will be created. They are the same as module parameters with addition of the 'run' node that controls run and stop phases of the test. Note that in this case test will not run on load automatically. Example of usage: % echo dma0chan0 > /sys/kernel/debug/dmatest/channel % echo 2000 > /sys/kernel/debug/dmatest/timeout % echo 1 > /sys/kernel/debug/dmatest/iterations % echo 1 > /sys/kernel/debug/dmatest/run After a while you will start to get messages about current status or error like in the original code. Note that running a new test will stop any in progress test. 2. When built-in in the kernel... The module parameters that is supplied to the kernel command line will be used for the first performed test. After user gets a control, the test could be interrupted or re-run with same or different parameters. For the details see the above section "1. When dmatest is built as a module..." In both cases the module parameters are used as initial values for the test case. You always could check them at run-time by running % grep -H . /sys/module/dmatest/parameters/* Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> Signed-off-by: Vinod Koul <vinod.koul@intel.com>
2013-03-04 09:09:30 +00:00
return 0;
}
/* when compiled-in wait for drivers to load first */
late_initcall(dmatest_init);
static void __exit dmatest_exit(void)
{
struct dmatest_info *info = &test_info;
mutex_lock(&info->lock);
stop_threaded_test(info);
mutex_unlock(&info->lock);
}
module_exit(dmatest_exit);
MODULE_AUTHOR("Haavard Skinnemoen (Atmel)");
MODULE_LICENSE("GPL v2");