kernel-aes67/arch/powerpc/platforms/iseries/setup.c

1003 lines
25 KiB
C
Raw Normal View History

/*
* Copyright (c) 2000 Mike Corrigan <mikejc@us.ibm.com>
* Copyright (c) 1999-2000 Grant Erickson <grant@lcse.umn.edu>
*
* Description:
* Architecture- / platform-specific boot-time initialization code for
* the IBM iSeries LPAR. Adapted from original code by Grant Erickson and
* code by Gary Thomas, Cort Dougan <cort@fsmlabs.com>, and Dan Malek
* <dan@net4x.com>.
*
* This program is free software; you can redistribute it and/or
* modify it under the terms of the GNU General Public License
* as published by the Free Software Foundation; either version
* 2 of the License, or (at your option) any later version.
*/
#undef DEBUG
#include <linux/config.h>
#include <linux/init.h>
#include <linux/threads.h>
#include <linux/smp.h>
#include <linux/param.h>
#include <linux/string.h>
#include <linux/initrd.h>
#include <linux/seq_file.h>
#include <linux/kdev_t.h>
#include <linux/major.h>
#include <linux/root_dev.h>
#include <linux/kernel.h>
#include <asm/processor.h>
#include <asm/machdep.h>
#include <asm/page.h>
#include <asm/mmu.h>
#include <asm/pgtable.h>
#include <asm/mmu_context.h>
#include <asm/cputable.h>
#include <asm/sections.h>
#include <asm/iommu.h>
#include <asm/firmware.h>
#include <asm/system.h>
#include <asm/time.h>
#include <asm/paca.h>
#include <asm/cache.h>
#include <asm/sections.h>
#include <asm/abs_addr.h>
#include <asm/iseries/hv_lp_config.h>
#include <asm/iseries/hv_call_event.h>
#include <asm/iseries/hv_call_xm.h>
#include <asm/iseries/it_lp_queue.h>
#include <asm/iseries/mf.h>
#include <asm/iseries/hv_lp_event.h>
#include <asm/iseries/lpar_map.h>
#include <asm/udbg.h>
#include "naca.h"
#include "setup.h"
#include "irq.h"
#include "vpd_areas.h"
#include "processor_vpd.h"
#include "main_store.h"
#include "call_sm.h"
#include "call_hpt.h"
#ifdef DEBUG
#define DBG(fmt...) udbg_printf(fmt)
#else
#define DBG(fmt...)
#endif
/* Function Prototypes */
static unsigned long build_iSeries_Memory_Map(void);
static void iseries_shared_idle(void);
static void iseries_dedicated_idle(void);
#ifdef CONFIG_PCI
extern void iSeries_pci_final_fixup(void);
#else
static void iSeries_pci_final_fixup(void) { }
#endif
/* Global Variables */
int piranha_simulator;
extern int rd_size; /* Defined in drivers/block/rd.c */
extern unsigned long embedded_sysmap_start;
extern unsigned long embedded_sysmap_end;
extern unsigned long iSeries_recal_tb;
extern unsigned long iSeries_recal_titan;
static int mf_initialized;
static unsigned long cmd_mem_limit;
struct MemoryBlock {
unsigned long absStart;
unsigned long absEnd;
unsigned long logicalStart;
unsigned long logicalEnd;
};
/*
* Process the main store vpd to determine where the holes in memory are
* and return the number of physical blocks and fill in the array of
* block data.
*/
static unsigned long iSeries_process_Condor_mainstore_vpd(
struct MemoryBlock *mb_array, unsigned long max_entries)
{
unsigned long holeFirstChunk, holeSizeChunks;
unsigned long numMemoryBlocks = 1;
struct IoHriMainStoreSegment4 *msVpd =
(struct IoHriMainStoreSegment4 *)xMsVpd;
unsigned long holeStart = msVpd->nonInterleavedBlocksStartAdr;
unsigned long holeEnd = msVpd->nonInterleavedBlocksEndAdr;
unsigned long holeSize = holeEnd - holeStart;
printk("Mainstore_VPD: Condor\n");
/*
* Determine if absolute memory has any
* holes so that we can interpret the
* access map we get back from the hypervisor
* correctly.
*/
mb_array[0].logicalStart = 0;
mb_array[0].logicalEnd = 0x100000000;
mb_array[0].absStart = 0;
mb_array[0].absEnd = 0x100000000;
if (holeSize) {
numMemoryBlocks = 2;
holeStart = holeStart & 0x000fffffffffffff;
holeStart = addr_to_chunk(holeStart);
holeFirstChunk = holeStart;
holeSize = addr_to_chunk(holeSize);
holeSizeChunks = holeSize;
printk( "Main store hole: start chunk = %0lx, size = %0lx chunks\n",
holeFirstChunk, holeSizeChunks );
mb_array[0].logicalEnd = holeFirstChunk;
mb_array[0].absEnd = holeFirstChunk;
mb_array[1].logicalStart = holeFirstChunk;
mb_array[1].logicalEnd = 0x100000000 - holeSizeChunks;
mb_array[1].absStart = holeFirstChunk + holeSizeChunks;
mb_array[1].absEnd = 0x100000000;
}
return numMemoryBlocks;
}
#define MaxSegmentAreas 32
#define MaxSegmentAdrRangeBlocks 128
#define MaxAreaRangeBlocks 4
static unsigned long iSeries_process_Regatta_mainstore_vpd(
struct MemoryBlock *mb_array, unsigned long max_entries)
{
struct IoHriMainStoreSegment5 *msVpdP =
(struct IoHriMainStoreSegment5 *)xMsVpd;
unsigned long numSegmentBlocks = 0;
u32 existsBits = msVpdP->msAreaExists;
unsigned long area_num;
printk("Mainstore_VPD: Regatta\n");
for (area_num = 0; area_num < MaxSegmentAreas; ++area_num ) {
unsigned long numAreaBlocks;
struct IoHriMainStoreArea4 *currentArea;
if (existsBits & 0x80000000) {
unsigned long block_num;
currentArea = &msVpdP->msAreaArray[area_num];
numAreaBlocks = currentArea->numAdrRangeBlocks;
printk("ms_vpd: processing area %2ld blocks=%ld",
area_num, numAreaBlocks);
for (block_num = 0; block_num < numAreaBlocks;
++block_num ) {
/* Process an address range block */
struct MemoryBlock tempBlock;
unsigned long i;
tempBlock.absStart =
(unsigned long)currentArea->xAdrRangeBlock[block_num].blockStart;
tempBlock.absEnd =
(unsigned long)currentArea->xAdrRangeBlock[block_num].blockEnd;
tempBlock.logicalStart = 0;
tempBlock.logicalEnd = 0;
printk("\n block %ld absStart=%016lx absEnd=%016lx",
block_num, tempBlock.absStart,
tempBlock.absEnd);
for (i = 0; i < numSegmentBlocks; ++i) {
if (mb_array[i].absStart ==
tempBlock.absStart)
break;
}
if (i == numSegmentBlocks) {
if (numSegmentBlocks == max_entries)
panic("iSeries_process_mainstore_vpd: too many memory blocks");
mb_array[numSegmentBlocks] = tempBlock;
++numSegmentBlocks;
} else
printk(" (duplicate)");
}
printk("\n");
}
existsBits <<= 1;
}
/* Now sort the blocks found into ascending sequence */
if (numSegmentBlocks > 1) {
unsigned long m, n;
for (m = 0; m < numSegmentBlocks - 1; ++m) {
for (n = numSegmentBlocks - 1; m < n; --n) {
if (mb_array[n].absStart <
mb_array[n-1].absStart) {
struct MemoryBlock tempBlock;
tempBlock = mb_array[n];
mb_array[n] = mb_array[n-1];
mb_array[n-1] = tempBlock;
}
}
}
}
/*
* Assign "logical" addresses to each block. These
* addresses correspond to the hypervisor "bitmap" space.
* Convert all addresses into units of 256K chunks.
*/
{
unsigned long i, nextBitmapAddress;
printk("ms_vpd: %ld sorted memory blocks\n", numSegmentBlocks);
nextBitmapAddress = 0;
for (i = 0; i < numSegmentBlocks; ++i) {
unsigned long length = mb_array[i].absEnd -
mb_array[i].absStart;
mb_array[i].logicalStart = nextBitmapAddress;
mb_array[i].logicalEnd = nextBitmapAddress + length;
nextBitmapAddress += length;
printk(" Bitmap range: %016lx - %016lx\n"
" Absolute range: %016lx - %016lx\n",
mb_array[i].logicalStart,
mb_array[i].logicalEnd,
mb_array[i].absStart, mb_array[i].absEnd);
mb_array[i].absStart = addr_to_chunk(mb_array[i].absStart &
0x000fffffffffffff);
mb_array[i].absEnd = addr_to_chunk(mb_array[i].absEnd &
0x000fffffffffffff);
mb_array[i].logicalStart =
addr_to_chunk(mb_array[i].logicalStart);
mb_array[i].logicalEnd = addr_to_chunk(mb_array[i].logicalEnd);
}
}
return numSegmentBlocks;
}
static unsigned long iSeries_process_mainstore_vpd(struct MemoryBlock *mb_array,
unsigned long max_entries)
{
unsigned long i;
unsigned long mem_blocks = 0;
if (cpu_has_feature(CPU_FTR_SLB))
mem_blocks = iSeries_process_Regatta_mainstore_vpd(mb_array,
max_entries);
else
mem_blocks = iSeries_process_Condor_mainstore_vpd(mb_array,
max_entries);
printk("Mainstore_VPD: numMemoryBlocks = %ld \n", mem_blocks);
for (i = 0; i < mem_blocks; ++i) {
printk("Mainstore_VPD: block %3ld logical chunks %016lx - %016lx\n"
" abs chunks %016lx - %016lx\n",
i, mb_array[i].logicalStart, mb_array[i].logicalEnd,
mb_array[i].absStart, mb_array[i].absEnd);
}
return mem_blocks;
}
static void __init iSeries_get_cmdline(void)
{
char *p, *q;
/* copy the command line parameter from the primary VSP */
HvCallEvent_dmaToSp(cmd_line, 2 * 64* 1024, 256,
HvLpDma_Direction_RemoteToLocal);
p = cmd_line;
q = cmd_line + 255;
while(p < q) {
if (!*p || *p == '\n')
break;
++p;
}
*p = 0;
}
static void __init iSeries_init_early(void)
{
DBG(" -> iSeries_init_early()\n");
ppc64_firmware_features = FW_FEATURE_ISERIES;
ppc64_interrupt_controller = IC_ISERIES;
#if defined(CONFIG_BLK_DEV_INITRD)
/*
* If the init RAM disk has been configured and there is
* a non-zero starting address for it, set it up
*/
if (naca.xRamDisk) {
initrd_start = (unsigned long)__va(naca.xRamDisk);
initrd_end = initrd_start + naca.xRamDiskSize * HW_PAGE_SIZE;
initrd_below_start_ok = 1; // ramdisk in kernel space
ROOT_DEV = Root_RAM0;
if (((rd_size * 1024) / HW_PAGE_SIZE) < naca.xRamDiskSize)
rd_size = (naca.xRamDiskSize * HW_PAGE_SIZE) / 1024;
} else
#endif /* CONFIG_BLK_DEV_INITRD */
{
/* ROOT_DEV = MKDEV(VIODASD_MAJOR, 1); */
}
iSeries_recal_tb = get_tb();
iSeries_recal_titan = HvCallXm_loadTod();
/*
* Initialize the hash table management pointers
*/
hpte_init_iSeries();
/*
* Initialize the DMA/TCE management
*/
iommu_init_early_iSeries();
/* Initialize machine-dependency vectors */
#ifdef CONFIG_SMP
smp_init_iSeries();
#endif
if (itLpNaca.xPirEnvironMode == 0)
piranha_simulator = 1;
/* Associate Lp Event Queue 0 with processor 0 */
HvCallEvent_setLpEventQueueInterruptProc(0, 0);
mf_init();
mf_initialized = 1;
mb();
/* If we were passed an initrd, set the ROOT_DEV properly if the values
* look sensible. If not, clear initrd reference.
*/
#ifdef CONFIG_BLK_DEV_INITRD
if (initrd_start >= KERNELBASE && initrd_end >= KERNELBASE &&
initrd_end > initrd_start)
ROOT_DEV = Root_RAM0;
else
initrd_start = initrd_end = 0;
#endif /* CONFIG_BLK_DEV_INITRD */
DBG(" <- iSeries_init_early()\n");
}
struct mschunks_map mschunks_map = {
/* XXX We don't use these, but Piranha might need them. */
.chunk_size = MSCHUNKS_CHUNK_SIZE,
.chunk_shift = MSCHUNKS_CHUNK_SHIFT,
.chunk_mask = MSCHUNKS_OFFSET_MASK,
};
EXPORT_SYMBOL(mschunks_map);
void mschunks_alloc(unsigned long num_chunks)
{
klimit = _ALIGN(klimit, sizeof(u32));
mschunks_map.mapping = (u32 *)klimit;
klimit += num_chunks * sizeof(u32);
mschunks_map.num_chunks = num_chunks;
}
/*
* The iSeries may have very large memories ( > 128 GB ) and a partition
* may get memory in "chunks" that may be anywhere in the 2**52 real
* address space. The chunks are 256K in size. To map this to the
* memory model Linux expects, the AS/400 specific code builds a
* translation table to translate what Linux thinks are "physical"
* addresses to the actual real addresses. This allows us to make
* it appear to Linux that we have contiguous memory starting at
* physical address zero while in fact this could be far from the truth.
* To avoid confusion, I'll let the words physical and/or real address
* apply to the Linux addresses while I'll use "absolute address" to
* refer to the actual hardware real address.
*
* build_iSeries_Memory_Map gets information from the Hypervisor and
* looks at the Main Store VPD to determine the absolute addresses
* of the memory that has been assigned to our partition and builds
* a table used to translate Linux's physical addresses to these
* absolute addresses. Absolute addresses are needed when
* communicating with the hypervisor (e.g. to build HPT entries)
*
* Returns the physical memory size
*/
static unsigned long __init build_iSeries_Memory_Map(void)
{
u32 loadAreaFirstChunk, loadAreaLastChunk, loadAreaSize;
u32 nextPhysChunk;
u32 hptFirstChunk, hptLastChunk, hptSizeChunks, hptSizePages;
u32 totalChunks,moreChunks;
u32 currChunk, thisChunk, absChunk;
u32 currDword;
u32 chunkBit;
u64 map;
struct MemoryBlock mb[32];
unsigned long numMemoryBlocks, curBlock;
/* Chunk size on iSeries is 256K bytes */
totalChunks = (u32)HvLpConfig_getMsChunks();
mschunks_alloc(totalChunks);
/*
* Get absolute address of our load area
* and map it to physical address 0
* This guarantees that the loadarea ends up at physical 0
* otherwise, it might not be returned by PLIC as the first
* chunks
*/
loadAreaFirstChunk = (u32)addr_to_chunk(itLpNaca.xLoadAreaAddr);
loadAreaSize = itLpNaca.xLoadAreaChunks;
/*
* Only add the pages already mapped here.
* Otherwise we might add the hpt pages
* The rest of the pages of the load area
* aren't in the HPT yet and can still
* be assigned an arbitrary physical address
*/
if ((loadAreaSize * 64) > HvPagesToMap)
loadAreaSize = HvPagesToMap / 64;
loadAreaLastChunk = loadAreaFirstChunk + loadAreaSize - 1;
/*
* TODO Do we need to do something if the HPT is in the 64MB load area?
* This would be required if the itLpNaca.xLoadAreaChunks includes
* the HPT size
*/
printk("Mapping load area - physical addr = 0000000000000000\n"
" absolute addr = %016lx\n",
chunk_to_addr(loadAreaFirstChunk));
printk("Load area size %dK\n", loadAreaSize * 256);
for (nextPhysChunk = 0; nextPhysChunk < loadAreaSize; ++nextPhysChunk)
mschunks_map.mapping[nextPhysChunk] =
loadAreaFirstChunk + nextPhysChunk;
/*
* Get absolute address of our HPT and remember it so
* we won't map it to any physical address
*/
hptFirstChunk = (u32)addr_to_chunk(HvCallHpt_getHptAddress());
hptSizePages = (u32)HvCallHpt_getHptPages();
hptSizeChunks = hptSizePages >>
(MSCHUNKS_CHUNK_SHIFT - HW_PAGE_SHIFT);
hptLastChunk = hptFirstChunk + hptSizeChunks - 1;
printk("HPT absolute addr = %016lx, size = %dK\n",
chunk_to_addr(hptFirstChunk), hptSizeChunks * 256);
/*
* Determine if absolute memory has any
* holes so that we can interpret the
* access map we get back from the hypervisor
* correctly.
*/
numMemoryBlocks = iSeries_process_mainstore_vpd(mb, 32);
/*
* Process the main store access map from the hypervisor
* to build up our physical -> absolute translation table
*/
curBlock = 0;
currChunk = 0;
currDword = 0;
moreChunks = totalChunks;
while (moreChunks) {
map = HvCallSm_get64BitsOfAccessMap(itLpNaca.xLpIndex,
currDword);
thisChunk = currChunk;
while (map) {
chunkBit = map >> 63;
map <<= 1;
if (chunkBit) {
--moreChunks;
while (thisChunk >= mb[curBlock].logicalEnd) {
++curBlock;
if (curBlock >= numMemoryBlocks)
panic("out of memory blocks");
}
if (thisChunk < mb[curBlock].logicalStart)
panic("memory block error");
absChunk = mb[curBlock].absStart +
(thisChunk - mb[curBlock].logicalStart);
if (((absChunk < hptFirstChunk) ||
(absChunk > hptLastChunk)) &&
((absChunk < loadAreaFirstChunk) ||
(absChunk > loadAreaLastChunk))) {
mschunks_map.mapping[nextPhysChunk] =
absChunk;
++nextPhysChunk;
}
}
++thisChunk;
}
++currDword;
currChunk += 64;
}
/*
* main store size (in chunks) is
* totalChunks - hptSizeChunks
* which should be equal to
* nextPhysChunk
*/
return chunk_to_addr(nextPhysChunk);
}
/*
* Document me.
*/
static void __init iSeries_setup_arch(void)
{
if (get_lppaca()->shared_proc) {
ppc_md.idle_loop = iseries_shared_idle;
printk(KERN_INFO "Using shared processor idle loop\n");
} else {
ppc_md.idle_loop = iseries_dedicated_idle;
printk(KERN_INFO "Using dedicated idle loop\n");
}
/* Setup the Lp Event Queue */
setup_hvlpevent_queue();
printk("Max logical processors = %d\n",
itVpdAreas.xSlicMaxLogicalProcs);
printk("Max physical processors = %d\n",
itVpdAreas.xSlicMaxPhysicalProcs);
}
static void iSeries_show_cpuinfo(struct seq_file *m)
{
seq_printf(m, "machine\t\t: 64-bit iSeries Logical Partition\n");
}
/*
* Document me.
*/
static void iSeries_restart(char *cmd)
{
mf_reboot();
}
/*
* Document me.
*/
static void iSeries_power_off(void)
{
mf_power_off();
}
/*
* Document me.
*/
static void iSeries_halt(void)
{
mf_power_off();
}
static void __init iSeries_progress(char * st, unsigned short code)
{
printk("Progress: [%04x] - %s\n", (unsigned)code, st);
if (!piranha_simulator && mf_initialized) {
if (code != 0xffff)
mf_display_progress(code);
else
mf_clear_src();
}
}
static void __init iSeries_fixup_klimit(void)
{
/*
* Change klimit to take into account any ram disk
* that may be included
*/
if (naca.xRamDisk)
klimit = KERNELBASE + (u64)naca.xRamDisk +
(naca.xRamDiskSize * HW_PAGE_SIZE);
else {
/*
* No ram disk was included - check and see if there
* was an embedded system map. Change klimit to take
* into account any embedded system map
*/
if (embedded_sysmap_end)
klimit = KERNELBASE + ((embedded_sysmap_end + 4095) &
0xfffffffffffff000);
}
}
static int __init iSeries_src_init(void)
{
/* clear the progress line */
ppc_md.progress(" ", 0xffff);
return 0;
}
late_initcall(iSeries_src_init);
static inline void process_iSeries_events(void)
{
asm volatile ("li 0,0x5555; sc" : : : "r0", "r3");
}
static void yield_shared_processor(void)
{
unsigned long tb;
HvCall_setEnabledInterrupts(HvCall_MaskIPI |
HvCall_MaskLpEvent |
HvCall_MaskLpProd |
HvCall_MaskTimeout);
tb = get_tb();
/* Compute future tb value when yield should expire */
HvCall_yieldProcessor(HvCall_YieldTimed, tb+tb_ticks_per_jiffy);
/*
* The decrementer stops during the yield. Force a fake decrementer
* here and let the timer_interrupt code sort out the actual time.
*/
get_lppaca()->int_dword.fields.decr_int = 1;
ppc64_runlatch_on();
process_iSeries_events();
}
static void iseries_shared_idle(void)
{
while (1) {
while (!need_resched() && !hvlpevent_is_pending()) {
local_irq_disable();
ppc64_runlatch_off();
/* Recheck with irqs off */
if (!need_resched() && !hvlpevent_is_pending())
yield_shared_processor();
HMT_medium();
local_irq_enable();
}
ppc64_runlatch_on();
if (hvlpevent_is_pending())
process_iSeries_events();
preempt_enable_no_resched();
schedule();
preempt_disable();
}
}
static void iseries_dedicated_idle(void)
{
[PATCH] sched: resched and cpu_idle rework Make some changes to the NEED_RESCHED and POLLING_NRFLAG to reduce confusion, and make their semantics rigid. Improves efficiency of resched_task and some cpu_idle routines. * In resched_task: - TIF_NEED_RESCHED is only cleared with the task's runqueue lock held, and as we hold it during resched_task, then there is no need for an atomic test and set there. The only other time this should be set is when the task's quantum expires, in the timer interrupt - this is protected against because the rq lock is irq-safe. - If TIF_NEED_RESCHED is set, then we don't need to do anything. It won't get unset until the task get's schedule()d off. - If we are running on the same CPU as the task we resched, then set TIF_NEED_RESCHED and no further action is required. - If we are running on another CPU, and TIF_POLLING_NRFLAG is *not* set after TIF_NEED_RESCHED has been set, then we need to send an IPI. Using these rules, we are able to remove the test and set operation in resched_task, and make clear the previously vague semantics of POLLING_NRFLAG. * In idle routines: - Enter cpu_idle with preempt disabled. When the need_resched() condition becomes true, explicitly call schedule(). This makes things a bit clearer (IMO), but haven't updated all architectures yet. - Many do a test and clear of TIF_NEED_RESCHED for some reason. According to the resched_task rules, this isn't needed (and actually breaks the assumption that TIF_NEED_RESCHED is only cleared with the runqueue lock held). So remove that. Generally one less locked memory op when switching to the idle thread. - Many idle routines clear TIF_POLLING_NRFLAG, and only set it in the inner most polling idle loops. The above resched_task semantics allow it to be set until before the last time need_resched() is checked before going into a halt requiring interrupt wakeup. Many idle routines simply never enter such a halt, and so POLLING_NRFLAG can be always left set, completely eliminating resched IPIs when rescheduling the idle task. POLLING_NRFLAG width can be increased, to reduce the chance of resched IPIs. Signed-off-by: Nick Piggin <npiggin@suse.de> Cc: Ingo Molnar <mingo@elte.hu> Cc: Con Kolivas <kernel@kolivas.org> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-11-09 00:39:04 -05:00
set_thread_flag(TIF_POLLING_NRFLAG);
while (1) {
[PATCH] sched: resched and cpu_idle rework Make some changes to the NEED_RESCHED and POLLING_NRFLAG to reduce confusion, and make their semantics rigid. Improves efficiency of resched_task and some cpu_idle routines. * In resched_task: - TIF_NEED_RESCHED is only cleared with the task's runqueue lock held, and as we hold it during resched_task, then there is no need for an atomic test and set there. The only other time this should be set is when the task's quantum expires, in the timer interrupt - this is protected against because the rq lock is irq-safe. - If TIF_NEED_RESCHED is set, then we don't need to do anything. It won't get unset until the task get's schedule()d off. - If we are running on the same CPU as the task we resched, then set TIF_NEED_RESCHED and no further action is required. - If we are running on another CPU, and TIF_POLLING_NRFLAG is *not* set after TIF_NEED_RESCHED has been set, then we need to send an IPI. Using these rules, we are able to remove the test and set operation in resched_task, and make clear the previously vague semantics of POLLING_NRFLAG. * In idle routines: - Enter cpu_idle with preempt disabled. When the need_resched() condition becomes true, explicitly call schedule(). This makes things a bit clearer (IMO), but haven't updated all architectures yet. - Many do a test and clear of TIF_NEED_RESCHED for some reason. According to the resched_task rules, this isn't needed (and actually breaks the assumption that TIF_NEED_RESCHED is only cleared with the runqueue lock held). So remove that. Generally one less locked memory op when switching to the idle thread. - Many idle routines clear TIF_POLLING_NRFLAG, and only set it in the inner most polling idle loops. The above resched_task semantics allow it to be set until before the last time need_resched() is checked before going into a halt requiring interrupt wakeup. Many idle routines simply never enter such a halt, and so POLLING_NRFLAG can be always left set, completely eliminating resched IPIs when rescheduling the idle task. POLLING_NRFLAG width can be increased, to reduce the chance of resched IPIs. Signed-off-by: Nick Piggin <npiggin@suse.de> Cc: Ingo Molnar <mingo@elte.hu> Cc: Con Kolivas <kernel@kolivas.org> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-11-09 00:39:04 -05:00
if (!need_resched()) {
while (!need_resched()) {
ppc64_runlatch_off();
HMT_low();
if (hvlpevent_is_pending()) {
HMT_medium();
ppc64_runlatch_on();
process_iSeries_events();
}
}
HMT_medium();
}
ppc64_runlatch_on();
preempt_enable_no_resched();
schedule();
preempt_disable();
}
}
#ifndef CONFIG_PCI
void __init iSeries_init_IRQ(void) { }
#endif
static int __init iseries_probe(int platform)
{
return PLATFORM_ISERIES_LPAR == platform;
}
struct machdep_calls __initdata iseries_md = {
.setup_arch = iSeries_setup_arch,
.show_cpuinfo = iSeries_show_cpuinfo,
.init_IRQ = iSeries_init_IRQ,
.get_irq = iSeries_get_irq,
.init_early = iSeries_init_early,
.pcibios_fixup = iSeries_pci_final_fixup,
.restart = iSeries_restart,
.power_off = iSeries_power_off,
.halt = iSeries_halt,
.get_boot_time = iSeries_get_boot_time,
.set_rtc_time = iSeries_set_rtc_time,
.get_rtc_time = iSeries_get_rtc_time,
.calibrate_decr = generic_calibrate_decr,
.progress = iSeries_progress,
.probe = iseries_probe,
/* XXX Implement enable_pmcs for iSeries */
};
struct blob {
unsigned char data[PAGE_SIZE];
unsigned long next;
};
struct iseries_flat_dt {
struct boot_param_header header;
u64 reserve_map[2];
struct blob dt;
struct blob strings;
};
struct iseries_flat_dt iseries_dt;
void dt_init(struct iseries_flat_dt *dt)
{
dt->header.off_mem_rsvmap =
offsetof(struct iseries_flat_dt, reserve_map);
dt->header.off_dt_struct = offsetof(struct iseries_flat_dt, dt);
dt->header.off_dt_strings = offsetof(struct iseries_flat_dt, strings);
dt->header.totalsize = sizeof(struct iseries_flat_dt);
dt->header.dt_strings_size = sizeof(struct blob);
/* There is no notion of hardware cpu id on iSeries */
dt->header.boot_cpuid_phys = smp_processor_id();
dt->dt.next = (unsigned long)&dt->dt.data;
dt->strings.next = (unsigned long)&dt->strings.data;
dt->header.magic = OF_DT_HEADER;
dt->header.version = 0x10;
dt->header.last_comp_version = 0x10;
dt->reserve_map[0] = 0;
dt->reserve_map[1] = 0;
}
void dt_check_blob(struct blob *b)
{
if (b->next >= (unsigned long)&b->next) {
DBG("Ran out of space in flat device tree blob!\n");
BUG();
}
}
void dt_push_u32(struct iseries_flat_dt *dt, u32 value)
{
*((u32*)dt->dt.next) = value;
dt->dt.next += sizeof(u32);
dt_check_blob(&dt->dt);
}
void dt_push_u64(struct iseries_flat_dt *dt, u64 value)
{
*((u64*)dt->dt.next) = value;
dt->dt.next += sizeof(u64);
dt_check_blob(&dt->dt);
}
unsigned long dt_push_bytes(struct blob *blob, char *data, int len)
{
unsigned long start = blob->next - (unsigned long)blob->data;
memcpy((char *)blob->next, data, len);
blob->next = _ALIGN(blob->next + len, 4);
dt_check_blob(blob);
return start;
}
void dt_start_node(struct iseries_flat_dt *dt, char *name)
{
dt_push_u32(dt, OF_DT_BEGIN_NODE);
dt_push_bytes(&dt->dt, name, strlen(name) + 1);
}
#define dt_end_node(dt) dt_push_u32(dt, OF_DT_END_NODE)
void dt_prop(struct iseries_flat_dt *dt, char *name, char *data, int len)
{
unsigned long offset;
dt_push_u32(dt, OF_DT_PROP);
/* Length of the data */
dt_push_u32(dt, len);
/* Put the property name in the string blob. */
offset = dt_push_bytes(&dt->strings, name, strlen(name) + 1);
/* The offset of the properties name in the string blob. */
dt_push_u32(dt, (u32)offset);
/* The actual data. */
dt_push_bytes(&dt->dt, data, len);
}
void dt_prop_str(struct iseries_flat_dt *dt, char *name, char *data)
{
dt_prop(dt, name, data, strlen(data) + 1); /* + 1 for NULL */
}
void dt_prop_u32(struct iseries_flat_dt *dt, char *name, u32 data)
{
dt_prop(dt, name, (char *)&data, sizeof(u32));
}
void dt_prop_u64(struct iseries_flat_dt *dt, char *name, u64 data)
{
dt_prop(dt, name, (char *)&data, sizeof(u64));
}
void dt_prop_u64_list(struct iseries_flat_dt *dt, char *name, u64 *data, int n)
{
dt_prop(dt, name, (char *)data, sizeof(u64) * n);
}
void dt_prop_u32_list(struct iseries_flat_dt *dt, char *name, u32 *data, int n)
{
dt_prop(dt, name, (char *)data, sizeof(u32) * n);
}
void dt_prop_empty(struct iseries_flat_dt *dt, char *name)
{
dt_prop(dt, name, NULL, 0);
}
void dt_cpus(struct iseries_flat_dt *dt)
{
unsigned char buf[32];
unsigned char *p;
unsigned int i, index;
struct IoHriProcessorVpd *d;
u32 pft_size[2];
/* yuck */
snprintf(buf, 32, "PowerPC,%s", cur_cpu_spec->cpu_name);
p = strchr(buf, ' ');
if (!p) p = buf + strlen(buf);
dt_start_node(dt, "cpus");
dt_prop_u32(dt, "#address-cells", 1);
dt_prop_u32(dt, "#size-cells", 0);
pft_size[0] = 0; /* NUMA CEC cookie, 0 for non NUMA */
pft_size[1] = __ilog2(HvCallHpt_getHptPages() * HW_PAGE_SIZE);
for (i = 0; i < NR_CPUS; i++) {
if (lppaca[i].dyn_proc_status >= 2)
continue;
snprintf(p, 32 - (p - buf), "@%d", i);
dt_start_node(dt, buf);
dt_prop_str(dt, "device_type", "cpu");
index = lppaca[i].dyn_hv_phys_proc_index;
d = &xIoHriProcessorVpd[index];
dt_prop_u32(dt, "i-cache-size", d->xInstCacheSize * 1024);
dt_prop_u32(dt, "i-cache-line-size", d->xInstCacheOperandSize);
dt_prop_u32(dt, "d-cache-size", d->xDataL1CacheSizeKB * 1024);
dt_prop_u32(dt, "d-cache-line-size", d->xDataCacheOperandSize);
/* magic conversions to Hz copied from old code */
dt_prop_u32(dt, "clock-frequency",
((1UL << 34) * 1000000) / d->xProcFreq);
dt_prop_u32(dt, "timebase-frequency",
((1UL << 32) * 1000000) / d->xTimeBaseFreq);
dt_prop_u32(dt, "reg", i);
dt_prop_u32_list(dt, "ibm,pft-size", pft_size, 2);
dt_end_node(dt);
}
dt_end_node(dt);
}
void build_flat_dt(struct iseries_flat_dt *dt, unsigned long phys_mem_size)
{
u64 tmp[2];
dt_init(dt);
dt_start_node(dt, "");
dt_prop_u32(dt, "#address-cells", 2);
dt_prop_u32(dt, "#size-cells", 2);
/* /memory */
dt_start_node(dt, "memory@0");
dt_prop_str(dt, "name", "memory");
dt_prop_str(dt, "device_type", "memory");
tmp[0] = 0;
tmp[1] = phys_mem_size;
dt_prop_u64_list(dt, "reg", tmp, 2);
dt_end_node(dt);
/* /chosen */
dt_start_node(dt, "chosen");
dt_prop_u32(dt, "linux,platform", PLATFORM_ISERIES_LPAR);
if (cmd_mem_limit)
dt_prop_u64(dt, "linux,memory-limit", cmd_mem_limit);
dt_end_node(dt);
dt_cpus(dt);
dt_end_node(dt);
dt_push_u32(dt, OF_DT_END);
}
void * __init iSeries_early_setup(void)
{
unsigned long phys_mem_size;
iSeries_fixup_klimit();
/*
* Initialize the table which translate Linux physical addresses to
* AS/400 absolute addresses
*/
phys_mem_size = build_iSeries_Memory_Map();
iSeries_get_cmdline();
/* Save unparsed command line copy for /proc/cmdline */
strlcpy(saved_command_line, cmd_line, COMMAND_LINE_SIZE);
/* Parse early parameters, in particular mem=x */
parse_early_param();
build_flat_dt(&iseries_dt, phys_mem_size);
return (void *) __pa(&iseries_dt);
}
/*
* On iSeries we just parse the mem=X option from the command line.
* On pSeries it's a bit more complicated, see prom_init_mem()
*/
static int __init early_parsemem(char *p)
{
if (p)
cmd_mem_limit = ALIGN(memparse(p, &p), PAGE_SIZE);
return 0;
}
early_param("mem", early_parsemem);
static void hvputc(char c)
{
if (c == '\n')
hvputc('\r');
HvCall_writeLogBuffer(&c, 1);
}
void __init udbg_init_iseries(void)
{
udbg_putc = hvputc;
}