static void bts_update(struct bts_ctx *bts)
{
int cpu = raw_smp_processor_id();
struct debug_store *ds = per_cpu(cpu_hw_events, cpu).ds;
struct bts_buffer *buf = perf_get_aux(&bts->handle);
unsigned long index = ds->bts_index - ds->bts_buffer_base, old, head;
if (!buf)
return;
head = index + bts_buffer_offset(buf, buf->cur_buf);
old = local_xchg(&buf->head, head);
if (!buf->snapshot) {
if (old == head)
return;
if (ds->bts_index >= ds->bts_absolute_maximum)
local_inc(&buf->lost);
/*
* old and head are always in the same physical buffer, so we
* can subtract them to get the data size.
*/
local_add(head - old, &buf->data_size);
} else {
local_set(&buf->data_size, head);
}
}
void ctype_list_init(char *module)
{
Module = tdrpStrDup(module);
ListBuf = tdrpBufCreate();
/*
* add the names of the known types
*/
local_add("boolean", "tdrp_bool_t");
local_add("char*", "char*");
local_add("string", "char*");
local_add("int", "int");
local_add("long", "long");
local_add("float", "float");
local_add("double", "double");
}
static void etb_update_buffer(struct coresight_device *csdev,
struct perf_output_handle *handle,
void *sink_config)
{
bool lost = false;
int i, cur;
u8 *buf_ptr;
const u32 *barrier;
u32 read_ptr, write_ptr, capacity;
u32 status, read_data, to_read;
unsigned long offset;
struct cs_buffers *buf = sink_config;
struct etb_drvdata *drvdata = dev_get_drvdata(csdev->dev.parent);
if (!buf)
return;
capacity = drvdata->buffer_depth * ETB_FRAME_SIZE_WORDS;
etb_disable_hw(drvdata);
CS_UNLOCK(drvdata->base);
/* unit is in words, not bytes */
read_ptr = readl_relaxed(drvdata->base + ETB_RAM_READ_POINTER);
write_ptr = readl_relaxed(drvdata->base + ETB_RAM_WRITE_POINTER);
/*
* Entries should be aligned to the frame size. If they are not
* go back to the last alignment point to give decoding tools a
* chance to fix things.
*/
if (write_ptr % ETB_FRAME_SIZE_WORDS) {
dev_err(drvdata->dev,
"write_ptr: %lu not aligned to formatter frame size\n",
(unsigned long)write_ptr);
write_ptr &= ~(ETB_FRAME_SIZE_WORDS - 1);
lost = true;
}
/*
* Get a hold of the status register and see if a wrap around
* has occurred. If so adjust things accordingly. Otherwise
* start at the beginning and go until the write pointer has
* been reached.
*/
status = readl_relaxed(drvdata->base + ETB_STATUS_REG);
if (status & ETB_STATUS_RAM_FULL) {
lost = true;
to_read = capacity;
read_ptr = write_ptr;
} else {
to_read = CIRC_CNT(write_ptr, read_ptr, drvdata->buffer_depth);
to_read *= ETB_FRAME_SIZE_WORDS;
}
/*
* Make sure we don't overwrite data that hasn't been consumed yet.
* It is entirely possible that the HW buffer has more data than the
* ring buffer can currently handle. If so adjust the start address
* to take only the last traces.
*
* In snapshot mode we are looking to get the latest traces only and as
* such, we don't care about not overwriting data that hasn't been
* processed by user space.
*/
if (!buf->snapshot && to_read > handle->size) {
u32 mask = ~(ETB_FRAME_SIZE_WORDS - 1);
/* The new read pointer must be frame size aligned */
to_read = handle->size & mask;
/*
* Move the RAM read pointer up, keeping in mind that
* everything is in frame size units.
*/
read_ptr = (write_ptr + drvdata->buffer_depth) -
to_read / ETB_FRAME_SIZE_WORDS;
/* Wrap around if need be*/
if (read_ptr > (drvdata->buffer_depth - 1))
read_ptr -= drvdata->buffer_depth;
/* let the decoder know we've skipped ahead */
lost = true;
}
if (lost)
perf_aux_output_flag(handle, PERF_AUX_FLAG_TRUNCATED);
/* finally tell HW where we want to start reading from */
writel_relaxed(read_ptr, drvdata->base + ETB_RAM_READ_POINTER);
cur = buf->cur;
offset = buf->offset;
barrier = barrier_pkt;
for (i = 0; i < to_read; i += 4) {
buf_ptr = buf->data_pages[cur] + offset;
read_data = readl_relaxed(drvdata->base +
ETB_RAM_READ_DATA_REG);
if (lost && *barrier) {
read_data = *barrier;
barrier++;
}
*(u32 *)buf_ptr = read_data;
buf_ptr += 4;
offset += 4;
if (offset >= PAGE_SIZE) {
offset = 0;
cur++;
/* wrap around at the end of the buffer */
cur &= buf->nr_pages - 1;
}
}
/* reset ETB buffer for next run */
writel_relaxed(0x0, drvdata->base + ETB_RAM_READ_POINTER);
writel_relaxed(0x0, drvdata->base + ETB_RAM_WRITE_POINTER);
/*
* In snapshot mode all we have to do is communicate to
* perf_aux_output_end() the address of the current head. In full
* trace mode the same function expects a size to move rb->aux_head
* forward.
*/
if (buf->snapshot)
local_set(&buf->data_size, (cur * PAGE_SIZE) + offset);
else
local_add(to_read, &buf->data_size);
etb_enable_hw(drvdata);
CS_LOCK(drvdata->base);
}
int perf_output_begin(struct perf_output_handle *handle,
struct perf_event *event, unsigned int size)
{
struct ring_buffer *rb;
unsigned long tail, offset, head;
int have_lost;
struct perf_sample_data sample_data;
struct {
struct perf_event_header header;
u64 id;
u64 lost;
} lost_event;
rcu_read_lock();
/*
* For inherited events we send all the output towards the parent.
*/
if (event->parent)
event = event->parent;
rb = rcu_dereference(event->rb);
if (!rb)
goto out;
handle->rb = rb;
handle->event = event;
if (!rb->nr_pages)
goto out;
have_lost = local_read(&rb->lost);
if (have_lost) {
lost_event.header.size = sizeof(lost_event);
perf_event_header__init_id(&lost_event.header, &sample_data,
event);
size += lost_event.header.size;
}
perf_output_get_handle(handle);
do {
/*
* Userspace could choose to issue a mb() before updating the
* tail pointer. So that all reads will be completed before the
* write is issued.
*/
tail = ACCESS_ONCE(rb->user_page->data_tail);
smp_rmb();
offset = head = local_read(&rb->head);
head += size;
if (unlikely(!perf_output_space(rb, tail, offset, head)))
goto fail;
} while (local_cmpxchg(&rb->head, offset, head) != offset);
if (head - local_read(&rb->wakeup) > rb->watermark)
local_add(rb->watermark, &rb->wakeup);
handle->page = offset >> (PAGE_SHIFT + page_order(rb));
handle->page &= rb->nr_pages - 1;
handle->size = offset & ((PAGE_SIZE << page_order(rb)) - 1);
handle->addr = rb->data_pages[handle->page];
handle->addr += handle->size;
handle->size = (PAGE_SIZE << page_order(rb)) - handle->size;
if (have_lost) {
lost_event.header.type = PERF_RECORD_LOST;
lost_event.header.misc = 0;
lost_event.id = event->id;
lost_event.lost = local_xchg(&rb->lost, 0);
perf_output_put(handle, lost_event);
perf_event__output_id_sample(event, handle, &sample_data);
}
return 0;
fail:
local_inc(&rb->lost);
perf_output_put_handle(handle);
out:
rcu_read_unlock();
return -ENOSPC;
}
static void tmc_update_etf_buffer(struct coresight_device *csdev,
struct perf_output_handle *handle,
void *sink_config)
{
int i, cur;
u32 *buf_ptr;
u32 read_ptr, write_ptr;
u32 status, to_read;
unsigned long offset;
struct cs_buffers *buf = sink_config;
struct tmc_drvdata *drvdata = dev_get_drvdata(csdev->dev.parent);
if (!buf)
return;
/* This shouldn't happen */
if (WARN_ON_ONCE(local_read(&drvdata->mode) != CS_MODE_PERF))
return;
CS_UNLOCK(drvdata->base);
tmc_flush_and_stop(drvdata);
read_ptr = readl_relaxed(drvdata->base + TMC_RRP);
write_ptr = readl_relaxed(drvdata->base + TMC_RWP);
/*
* Get a hold of the status register and see if a wrap around
* has occurred. If so adjust things accordingly.
*/
status = readl_relaxed(drvdata->base + TMC_STS);
if (status & TMC_STS_FULL) {
local_inc(&buf->lost);
to_read = drvdata->size;
} else {
to_read = CIRC_CNT(write_ptr, read_ptr, drvdata->size);
}
/*
* The TMC RAM buffer may be bigger than the space available in the
* perf ring buffer (handle->size). If so advance the RRP so that we
* get the latest trace data.
*/
if (to_read > handle->size) {
u32 mask = 0;
/*
* The value written to RRP must be byte-address aligned to
* the width of the trace memory databus _and_ to a frame
* boundary (16 byte), whichever is the biggest. For example,
* for 32-bit, 64-bit and 128-bit wide trace memory, the four
* LSBs must be 0s. For 256-bit wide trace memory, the five
* LSBs must be 0s.
*/
switch (drvdata->memwidth) {
case TMC_MEM_INTF_WIDTH_32BITS:
case TMC_MEM_INTF_WIDTH_64BITS:
case TMC_MEM_INTF_WIDTH_128BITS:
mask = GENMASK(31, 5);
break;
case TMC_MEM_INTF_WIDTH_256BITS:
mask = GENMASK(31, 6);
break;
}
/*
* Make sure the new size is aligned in accordance with the
* requirement explained above.
*/
to_read = handle->size & mask;
/* Move the RAM read pointer up */
read_ptr = (write_ptr + drvdata->size) - to_read;
/* Make sure we are still within our limits */
if (read_ptr > (drvdata->size - 1))
read_ptr -= drvdata->size;
/* Tell the HW */
writel_relaxed(read_ptr, drvdata->base + TMC_RRP);
local_inc(&buf->lost);
}
cur = buf->cur;
offset = buf->offset;
/* for every byte to read */
for (i = 0; i < to_read; i += 4) {
buf_ptr = buf->data_pages[cur] + offset;
*buf_ptr = readl_relaxed(drvdata->base + TMC_RRD);
offset += 4;
if (offset >= PAGE_SIZE) {
offset = 0;
cur++;
/* wrap around at the end of the buffer */
cur &= buf->nr_pages - 1;
}
}
/*
* In snapshot mode all we have to do is communicate to
* perf_aux_output_end() the address of the current head. In full
* trace mode the same function expects a size to move rb->aux_head
* forward.
*/
if (buf->snapshot)
local_set(&buf->data_size, (cur * PAGE_SIZE) + offset);
else
local_add(to_read, &buf->data_size);
CS_LOCK(drvdata->base);
}
