linux/drivers/macintosh/rack-meter.c

618 lines
15 KiB
C
Raw Normal View History

// SPDX-License-Identifier: GPL-2.0-only
/*
* RackMac vu-meter driver
*
* (c) Copyright 2006 Benjamin Herrenschmidt, IBM Corp.
* <benh@kernel.crashing.org>
*
* Support the CPU-meter LEDs of the Xserve G5
*
* TODO: Implement PWM to do variable intensity and provide userland
* interface for fun. Also, the CPU-meter could be made nicer by being
* a bit less "immediate" but giving instead a more average load over
* time. Patches welcome :-)
*/
#undef DEBUG
#include <linux/types.h>
#include <linux/kernel.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/device.h>
#include <linux/interrupt.h>
#include <linux/module.h>
#include <linux/pci.h>
#include <linux/dma-mapping.h>
#include <linux/kernel_stat.h>
#include <linux/of_address.h>
#include <linux/of_irq.h>
#include <asm/io.h>
#include <asm/machdep.h>
#include <asm/pmac_feature.h>
#include <asm/dbdma.h>
#include <asm/macio.h>
#include <asm/keylargo.h>
/* Number of samples in a sample buffer */
#define SAMPLE_COUNT 256
/* CPU meter sampling rate in ms */
#define CPU_SAMPLING_RATE 250
struct rackmeter_dma {
struct dbdma_cmd cmd[4] ____cacheline_aligned;
u32 mark ____cacheline_aligned;
u32 buf1[SAMPLE_COUNT] ____cacheline_aligned;
u32 buf2[SAMPLE_COUNT] ____cacheline_aligned;
} ____cacheline_aligned;
struct rackmeter_cpu {
struct delayed_work sniffer;
struct rackmeter *rm;
u64 prev_wall;
u64 prev_idle;
int zero;
} ____cacheline_aligned;
struct rackmeter {
struct macio_dev *mdev;
unsigned int irq;
struct device_node *i2s;
u8 *ubuf;
struct dbdma_regs __iomem *dma_regs;
void __iomem *i2s_regs;
dma_addr_t dma_buf_p;
struct rackmeter_dma *dma_buf_v;
int stale_irq;
struct rackmeter_cpu cpu[2];
int paused;
struct mutex sem;
};
/* To be set as a tunable */
static int rackmeter_ignore_nice;
/* This GPIO is whacked by the OS X driver when initializing */
#define RACKMETER_MAGIC_GPIO 0x78
/* This is copied from cpufreq_ondemand, maybe we should put it in
* a common header somewhere
*/
static inline u64 get_cpu_idle_time(unsigned int cpu)
{
struct kernel_cpustat *kcpustat = &kcpustat_cpu(cpu);
u64 retval;
retval = kcpustat->cpustat[CPUTIME_IDLE] +
kcpustat->cpustat[CPUTIME_IOWAIT];
if (rackmeter_ignore_nice)
retval += kcpustat_field(kcpustat, CPUTIME_NICE, cpu);
return retval;
}
static void rackmeter_setup_i2s(struct rackmeter *rm)
{
struct macio_chip *macio = rm->mdev->bus->chip;
/* First whack magic GPIO */
pmac_call_feature(PMAC_FTR_WRITE_GPIO, NULL, RACKMETER_MAGIC_GPIO, 5);
/* Call feature code to enable the sound channel and the proper
* clock sources
*/
pmac_call_feature(PMAC_FTR_SOUND_CHIP_ENABLE, rm->i2s, 0, 1);
/* Power i2s and stop i2s clock. We whack MacIO FCRs directly for now.
* This is a bit racy, thus we should add new platform functions to
* handle that. snd-aoa needs that too
*/
MACIO_BIS(KEYLARGO_FCR1, KL1_I2S0_ENABLE);
MACIO_BIC(KEYLARGO_FCR1, KL1_I2S0_CLK_ENABLE_BIT);
(void)MACIO_IN32(KEYLARGO_FCR1);
udelay(10);
/* Then setup i2s. For now, we use the same magic value that
* the OS X driver seems to use. We might want to play around
* with the clock divisors later
*/
out_le32(rm->i2s_regs + 0x10, 0x01fa0000);
(void)in_le32(rm->i2s_regs + 0x10);
udelay(10);
/* Fully restart i2s*/
MACIO_BIS(KEYLARGO_FCR1, KL1_I2S0_CELL_ENABLE |
KL1_I2S0_CLK_ENABLE_BIT);
(void)MACIO_IN32(KEYLARGO_FCR1);
udelay(10);
}
static void rackmeter_set_default_pattern(struct rackmeter *rm)
{
int i;
for (i = 0; i < 16; i++) {
if (i < 8)
rm->ubuf[i] = (i & 1) * 255;
else
rm->ubuf[i] = ((~i) & 1) * 255;
}
}
static void rackmeter_do_pause(struct rackmeter *rm, int pause)
{
struct rackmeter_dma *rdma = rm->dma_buf_v;
pr_debug("rackmeter: %s\n", pause ? "paused" : "started");
rm->paused = pause;
if (pause) {
DBDMA_DO_STOP(rm->dma_regs);
return;
}
memset(rdma->buf1, 0, sizeof(rdma->buf1));
memset(rdma->buf2, 0, sizeof(rdma->buf2));
rm->dma_buf_v->mark = 0;
mb();
out_le32(&rm->dma_regs->cmdptr_hi, 0);
out_le32(&rm->dma_regs->cmdptr, rm->dma_buf_p);
out_le32(&rm->dma_regs->control, (RUN << 16) | RUN);
}
static void rackmeter_setup_dbdma(struct rackmeter *rm)
{
struct rackmeter_dma *db = rm->dma_buf_v;
struct dbdma_cmd *cmd = db->cmd;
/* Make sure dbdma is reset */
DBDMA_DO_RESET(rm->dma_regs);
pr_debug("rackmeter: mark offset=0x%zx\n",
offsetof(struct rackmeter_dma, mark));
pr_debug("rackmeter: buf1 offset=0x%zx\n",
offsetof(struct rackmeter_dma, buf1));
pr_debug("rackmeter: buf2 offset=0x%zx\n",
offsetof(struct rackmeter_dma, buf2));
/* Prepare 4 dbdma commands for the 2 buffers */
memset(cmd, 0, 4 * sizeof(struct dbdma_cmd));
cmd->req_count = cpu_to_le16(4);
cmd->command = cpu_to_le16(STORE_WORD | INTR_ALWAYS | KEY_SYSTEM);
cmd->phy_addr = cpu_to_le32(rm->dma_buf_p +
offsetof(struct rackmeter_dma, mark));
cmd->cmd_dep = cpu_to_le32(0x02000000);
cmd++;
cmd->req_count = cpu_to_le16(SAMPLE_COUNT * 4);
cmd->command = cpu_to_le16(OUTPUT_MORE);
cmd->phy_addr = cpu_to_le32(rm->dma_buf_p +
offsetof(struct rackmeter_dma, buf1));
cmd++;
cmd->req_count = cpu_to_le16(4);
cmd->command = cpu_to_le16(STORE_WORD | INTR_ALWAYS | KEY_SYSTEM);
cmd->phy_addr = cpu_to_le32(rm->dma_buf_p +
offsetof(struct rackmeter_dma, mark));
cmd->cmd_dep = cpu_to_le32(0x01000000);
cmd++;
cmd->req_count = cpu_to_le16(SAMPLE_COUNT * 4);
cmd->command = cpu_to_le16(OUTPUT_MORE | BR_ALWAYS);
cmd->phy_addr = cpu_to_le32(rm->dma_buf_p +
offsetof(struct rackmeter_dma, buf2));
cmd->cmd_dep = cpu_to_le32(rm->dma_buf_p);
rackmeter_do_pause(rm, 0);
}
static void rackmeter_do_timer(struct work_struct *work)
{
struct rackmeter_cpu *rcpu =
container_of(work, struct rackmeter_cpu, sniffer.work);
struct rackmeter *rm = rcpu->rm;
unsigned int cpu = smp_processor_id();
u64 cur_nsecs, total_idle_nsecs;
u64 total_nsecs, idle_nsecs;
int i, offset, load, cumm, pause;
cur_nsecs = jiffies64_to_nsecs(get_jiffies_64());
total_nsecs = cur_nsecs - rcpu->prev_wall;
rcpu->prev_wall = cur_nsecs;
total_idle_nsecs = get_cpu_idle_time(cpu);
idle_nsecs = total_idle_nsecs - rcpu->prev_idle;
idle_nsecs = min(idle_nsecs, total_nsecs);
rcpu->prev_idle = total_idle_nsecs;
/* We do a very dumb calculation to update the LEDs for now,
* we'll do better once we have actual PWM implemented
*/
load = div64_u64(9 * (total_nsecs - idle_nsecs), total_nsecs);
offset = cpu << 3;
cumm = 0;
for (i = 0; i < 8; i++) {
u8 ub = (load > i) ? 0xff : 0;
rm->ubuf[i + offset] = ub;
cumm |= ub;
}
rcpu->zero = (cumm == 0);
/* Now check if LEDs are all 0, we can stop DMA */
pause = (rm->cpu[0].zero && rm->cpu[1].zero);
if (pause != rm->paused) {
mutex_lock(&rm->sem);
pause = (rm->cpu[0].zero && rm->cpu[1].zero);
rackmeter_do_pause(rm, pause);
mutex_unlock(&rm->sem);
}
schedule_delayed_work_on(cpu, &rcpu->sniffer,
msecs_to_jiffies(CPU_SAMPLING_RATE));
}
static void rackmeter_init_cpu_sniffer(struct rackmeter *rm)
{
unsigned int cpu;
/* This driver works only with 1 or 2 CPUs numbered 0 and 1,
* but that's really all we have on Apple Xserve. It doesn't
* play very nice with CPU hotplug neither but we don't do that
* on those machines yet
*/
rm->cpu[0].rm = rm;
INIT_DELAYED_WORK(&rm->cpu[0].sniffer, rackmeter_do_timer);
rm->cpu[1].rm = rm;
INIT_DELAYED_WORK(&rm->cpu[1].sniffer, rackmeter_do_timer);
for_each_online_cpu(cpu) {
struct rackmeter_cpu *rcpu;
if (cpu > 1)
continue;
rcpu = &rm->cpu[cpu];
rcpu->prev_idle = get_cpu_idle_time(cpu);
rcpu->prev_wall = jiffies64_to_nsecs(get_jiffies_64());
schedule_delayed_work_on(cpu, &rm->cpu[cpu].sniffer,
msecs_to_jiffies(CPU_SAMPLING_RATE));
}
}
static void rackmeter_stop_cpu_sniffer(struct rackmeter *rm)
{
cancel_delayed_work_sync(&rm->cpu[0].sniffer);
cancel_delayed_work_sync(&rm->cpu[1].sniffer);
}
static int rackmeter_setup(struct rackmeter *rm)
{
pr_debug("rackmeter: setting up i2s..\n");
rackmeter_setup_i2s(rm);
pr_debug("rackmeter: setting up default pattern..\n");
rackmeter_set_default_pattern(rm);
pr_debug("rackmeter: setting up dbdma..\n");
rackmeter_setup_dbdma(rm);
pr_debug("rackmeter: start CPU measurements..\n");
rackmeter_init_cpu_sniffer(rm);
printk(KERN_INFO "RackMeter initialized\n");
return 0;
}
/* XXX FIXME: No PWM yet, this is 0/1 */
static u32 rackmeter_calc_sample(struct rackmeter *rm, unsigned int index)
{
int led;
u32 sample = 0;
for (led = 0; led < 16; led++) {
sample >>= 1;
sample |= ((rm->ubuf[led] >= 0x80) << 15);
}
return (sample << 17) | (sample >> 15);
}
static irqreturn_t rackmeter_irq(int irq, void *arg)
{
struct rackmeter *rm = arg;
struct rackmeter_dma *db = rm->dma_buf_v;
unsigned int mark, i;
u32 *buf;
/* Flush PCI buffers with an MMIO read. Maybe we could actually
* check the status one day ... in case things go wrong, though
* this never happened to me
*/
(void)in_le32(&rm->dma_regs->status);
/* Make sure the CPU gets us in order */
rmb();
/* Read mark */
mark = db->mark;
if (mark != 1 && mark != 2) {
printk(KERN_WARNING "rackmeter: Incorrect DMA mark 0x%08x\n",
mark);
/* We allow for 3 errors like that (stale DBDMA irqs) */
if (++rm->stale_irq > 3) {
printk(KERN_ERR "rackmeter: Too many errors,"
" stopping DMA\n");
DBDMA_DO_RESET(rm->dma_regs);
}
return IRQ_HANDLED;
}
/* Next buffer we need to fill is mark value */
buf = mark == 1 ? db->buf1 : db->buf2;
/* Fill it now. This routine converts the 8 bits depth sample array
* into the PWM bitmap for each LED.
*/
for (i = 0; i < SAMPLE_COUNT; i++)
buf[i] = rackmeter_calc_sample(rm, i);
return IRQ_HANDLED;
}
static int rackmeter_probe(struct macio_dev* mdev,
const struct of_device_id *match)
{
struct device_node *i2s = NULL, *np = NULL;
struct rackmeter *rm = NULL;
struct resource ri2s, rdma;
int rc = -ENODEV;
pr_debug("rackmeter_probe()\n");
/* Get i2s-a node */
for_each_child_of_node(mdev->ofdev.dev.of_node, i2s)
if (of_node_name_eq(i2s, "i2s-a"))
break;
if (i2s == NULL) {
pr_debug(" i2s-a child not found\n");
goto bail;
}
/* Get lightshow or virtual sound */
for_each_child_of_node(i2s, np) {
if (of_node_name_eq(np, "lightshow"))
break;
if (of_node_name_eq(np, "sound") &&
of_property_present(np, "virtual"))
break;
}
if (np == NULL) {
pr_debug(" lightshow or sound+virtual child not found\n");
goto bail;
}
/* Create and initialize our instance data */
rm = kzalloc(sizeof(*rm), GFP_KERNEL);
if (rm == NULL) {
printk(KERN_ERR "rackmeter: failed to allocate memory !\n");
rc = -ENOMEM;
goto bail_release;
}
rm->mdev = mdev;
rm->i2s = i2s;
mutex_init(&rm->sem);
dev_set_drvdata(&mdev->ofdev.dev, rm);
/* Check resources availability. We need at least resource 0 and 1 */
#if 0 /* Use that when i2s-a is finally an mdev per-se */
if (macio_resource_count(mdev) < 2 || macio_irq_count(mdev) < 2) {
printk(KERN_ERR
"rackmeter: found match but lacks resources: %pOF"
" (%d resources, %d interrupts)\n",
mdev->ofdev.dev.of_node);
rc = -ENXIO;
goto bail_free;
}
if (macio_request_resources(mdev, "rackmeter")) {
printk(KERN_ERR
"rackmeter: failed to request resources: %pOF\n",
mdev->ofdev.dev.of_node);
rc = -EBUSY;
goto bail_free;
}
rm->irq = macio_irq(mdev, 1);
#else
rm->irq = irq_of_parse_and_map(i2s, 1);
if (!rm->irq ||
of_address_to_resource(i2s, 0, &ri2s) ||
of_address_to_resource(i2s, 1, &rdma)) {
printk(KERN_ERR
"rackmeter: found match but lacks resources: %pOF",
mdev->ofdev.dev.of_node);
rc = -ENXIO;
goto bail_free;
}
#endif
pr_debug(" i2s @0x%08x\n", (unsigned int)ri2s.start);
pr_debug(" dma @0x%08x\n", (unsigned int)rdma.start);
pr_debug(" irq %d\n", rm->irq);
rm->ubuf = (u8 *)__get_free_page(GFP_KERNEL);
if (rm->ubuf == NULL) {
printk(KERN_ERR
"rackmeter: failed to allocate samples page !\n");
rc = -ENOMEM;
goto bail_release;
}
rm->dma_buf_v = dma_alloc_coherent(&macio_get_pci_dev(mdev)->dev,
sizeof(struct rackmeter_dma),
&rm->dma_buf_p, GFP_KERNEL);
if (rm->dma_buf_v == NULL) {
printk(KERN_ERR
"rackmeter: failed to allocate dma buffer !\n");
rc = -ENOMEM;
goto bail_free_samples;
}
#if 0
rm->i2s_regs = ioremap(macio_resource_start(mdev, 0), 0x1000);
#else
rm->i2s_regs = ioremap(ri2s.start, 0x1000);
#endif
if (rm->i2s_regs == NULL) {
printk(KERN_ERR
"rackmeter: failed to map i2s registers !\n");
rc = -ENXIO;
goto bail_free_dma;
}
#if 0
rm->dma_regs = ioremap(macio_resource_start(mdev, 1), 0x100);
#else
rm->dma_regs = ioremap(rdma.start, 0x100);
#endif
if (rm->dma_regs == NULL) {
printk(KERN_ERR
"rackmeter: failed to map dma registers !\n");
rc = -ENXIO;
goto bail_unmap_i2s;
}
rc = rackmeter_setup(rm);
if (rc) {
printk(KERN_ERR
"rackmeter: failed to initialize !\n");
rc = -ENXIO;
goto bail_unmap_dma;
}
rc = request_irq(rm->irq, rackmeter_irq, 0, "rackmeter", rm);
if (rc != 0) {
printk(KERN_ERR
"rackmeter: failed to request interrupt !\n");
goto bail_stop_dma;
}
of_node_put(np);
return 0;
bail_stop_dma:
DBDMA_DO_RESET(rm->dma_regs);
bail_unmap_dma:
iounmap(rm->dma_regs);
bail_unmap_i2s:
iounmap(rm->i2s_regs);
bail_free_dma:
dma_free_coherent(&macio_get_pci_dev(mdev)->dev,
sizeof(struct rackmeter_dma),
rm->dma_buf_v, rm->dma_buf_p);
bail_free_samples:
free_page((unsigned long)rm->ubuf);
bail_release:
#if 0
macio_release_resources(mdev);
#endif
bail_free:
kfree(rm);
bail:
of_node_put(i2s);
of_node_put(np);
dev_set_drvdata(&mdev->ofdev.dev, NULL);
return rc;
}
static void rackmeter_remove(struct macio_dev *mdev)
{
struct rackmeter *rm = dev_get_drvdata(&mdev->ofdev.dev);
/* Stop CPU sniffer timer & work queues */
rackmeter_stop_cpu_sniffer(rm);
/* Clear reference to private data */
dev_set_drvdata(&mdev->ofdev.dev, NULL);
/* Stop/reset dbdma */
DBDMA_DO_RESET(rm->dma_regs);
/* Release the IRQ */
free_irq(rm->irq, rm);
/* Unmap registers */
iounmap(rm->dma_regs);
iounmap(rm->i2s_regs);
/* Free DMA */
dma_free_coherent(&macio_get_pci_dev(mdev)->dev,
sizeof(struct rackmeter_dma),
rm->dma_buf_v, rm->dma_buf_p);
/* Free samples */
free_page((unsigned long)rm->ubuf);
#if 0
/* Release resources */
macio_release_resources(mdev);
#endif
/* Get rid of me */
kfree(rm);
}
static int rackmeter_shutdown(struct macio_dev* mdev)
{
struct rackmeter *rm = dev_get_drvdata(&mdev->ofdev.dev);
if (rm == NULL)
return -ENODEV;
/* Stop CPU sniffer timer & work queues */
rackmeter_stop_cpu_sniffer(rm);
/* Stop/reset dbdma */
DBDMA_DO_RESET(rm->dma_regs);
return 0;
}
static const struct of_device_id rackmeter_match[] = {
{ .name = "i2s" },
{ }
};
MODULE_DEVICE_TABLE(of, rackmeter_match);
static struct macio_driver rackmeter_driver = {
.driver = {
.name = "rackmeter",
.owner = THIS_MODULE,
.of_match_table = rackmeter_match,
},
.probe = rackmeter_probe,
.remove = rackmeter_remove,
.shutdown = rackmeter_shutdown,
};
static int __init rackmeter_init(void)
{
pr_debug("rackmeter_init()\n");
return macio_register_driver(&rackmeter_driver);
}
static void __exit rackmeter_exit(void)
{
pr_debug("rackmeter_exit()\n");
macio_unregister_driver(&rackmeter_driver);
}
module_init(rackmeter_init);
module_exit(rackmeter_exit);
MODULE_LICENSE("GPL");
MODULE_AUTHOR("Benjamin Herrenschmidt <benh@kernel.crashing.org>");
MODULE_DESCRIPTION("RackMeter: Support vu-meter on XServe front panel");