prom_init.c

linux 内核源代码

	prom_debug("  alloc_top_hi : %x\n", RELOC(alloc_top_high));
	prom_debug("  rmo_top      : %x\n", RELOC(rmo_top));
	prom_debug("  ram_top      : %x\n", RELOC(ram_top));

	return addr;
}

/*
 * Parse a "reg" cell
 */
static unsigned long __init prom_next_cell(int s, cell_t **cellp)
{
	cell_t *p = *cellp;
	unsigned long r = 0;

	/* Ignore more than 2 cells */
	while (s > sizeof(unsigned long) / 4) {
		p++;
		s--;
	}
	r = *p++;
#ifdef CONFIG_PPC64
	if (s > 1) {
		r <<= 32;
		r |= *(p++);
	}
#endif
	*cellp = p;
	return r;
}

/*
 * Very dumb function for adding to the memory reserve list, but
 * we don't need anything smarter at this point
 *
 * XXX Eventually check for collisions.  They should NEVER happen.
 * If problems seem to show up, it would be a good start to track
 * them down.
 */
static void __init reserve_mem(u64 base, u64 size)
{
	u64 top = base + size;
	unsigned long cnt = RELOC(mem_reserve_cnt);

	if (size == 0)
		return;

	/* We need to always keep one empty entry so that we
	 * have our terminator with "size" set to 0 since we are
	 * dumb and just copy this entire array to the boot params
	 */
	base = _ALIGN_DOWN(base, PAGE_SIZE);
	top = _ALIGN_UP(top, PAGE_SIZE);
	size = top - base;

	if (cnt >= (MEM_RESERVE_MAP_SIZE - 1))
		prom_panic("Memory reserve map exhausted !\n");
	RELOC(mem_reserve_map)[cnt].base = base;
	RELOC(mem_reserve_map)[cnt].size = size;
	RELOC(mem_reserve_cnt) = cnt + 1;
}

/*
 * Initialize memory allocation mechanism, parse "memory" nodes and
 * obtain that way the top of memory and RMO to setup out local allocator
 */
static void __init prom_init_mem(void)
{
	phandle node;
	char *path, type[64];
	unsigned int plen;
	cell_t *p, *endp;
	struct prom_t *_prom = &RELOC(prom);
	u32 rac, rsc;

	/*
	 * We iterate the memory nodes to find
	 * 1) top of RMO (first node)
	 * 2) top of memory
	 */
	rac = 2;
	prom_getprop(_prom->root, "#address-cells", &rac, sizeof(rac));
	rsc = 1;
	prom_getprop(_prom->root, "#size-cells", &rsc, sizeof(rsc));
	prom_debug("root_addr_cells: %x\n", (unsigned long) rac);
	prom_debug("root_size_cells: %x\n", (unsigned long) rsc);

	prom_debug("scanning memory:\n");
	path = RELOC(prom_scratch);

	for (node = 0; prom_next_node(&node); ) {
		type[0] = 0;
		prom_getprop(node, "device_type", type, sizeof(type));

		if (type[0] == 0) {
			/*
			 * CHRP Longtrail machines have no device_type
			 * on the memory node, so check the name instead...
			 */
			prom_getprop(node, "name", type, sizeof(type));
		}
		if (strcmp(type, RELOC("memory")))
			continue;

		plen = prom_getprop(node, "reg", RELOC(regbuf), sizeof(regbuf));
		if (plen > sizeof(regbuf)) {
			prom_printf("memory node too large for buffer !\n");
			plen = sizeof(regbuf);
		}
		p = RELOC(regbuf);
		endp = p + (plen / sizeof(cell_t));

#ifdef DEBUG_PROM
		memset(path, 0, PROM_SCRATCH_SIZE);
		call_prom("package-to-path", 3, 1, node, path, PROM_SCRATCH_SIZE-1);
		prom_debug("  node %s :\n", path);
#endif /* DEBUG_PROM */

		while ((endp - p) >= (rac + rsc)) {
			unsigned long base, size;

			base = prom_next_cell(rac, &p);
			size = prom_next_cell(rsc, &p);

			if (size == 0)
				continue;
			prom_debug("    %x %x\n", base, size);
			if (base == 0 && (RELOC(of_platform) & PLATFORM_LPAR))
				RELOC(rmo_top) = size;
			if ((base + size) > RELOC(ram_top))
				RELOC(ram_top) = base + size;
		}
	}

	RELOC(alloc_bottom) = PAGE_ALIGN((unsigned long)&RELOC(_end) + 0x4000);

	/* Check if we have an initrd after the kernel, if we do move our bottom
	 * point to after it
	 */
	if (RELOC(prom_initrd_start)) {
		if (RELOC(prom_initrd_end) > RELOC(alloc_bottom))
			RELOC(alloc_bottom) = PAGE_ALIGN(RELOC(prom_initrd_end));
	}

	/*
	 * Setup our top alloc point, that is top of RMO or top of
	 * segment 0 when running non-LPAR.
	 * Some RS64 machines have buggy firmware where claims up at
	 * 1GB fail.  Cap at 768MB as a workaround.
	 * Since 768MB is plenty of room, and we need to cap to something
	 * reasonable on 32-bit, cap at 768MB on all machines.
	 */
	if (!RELOC(rmo_top))
		RELOC(rmo_top) = RELOC(ram_top);
	RELOC(rmo_top) = min(0x30000000ul, RELOC(rmo_top));
	RELOC(alloc_top) = RELOC(rmo_top);
	RELOC(alloc_top_high) = RELOC(ram_top);

	prom_printf("memory layout at init:\n");
	prom_printf("  alloc_bottom : %x\n", RELOC(alloc_bottom));
	prom_printf("  alloc_top    : %x\n", RELOC(alloc_top));
	prom_printf("  alloc_top_hi : %x\n", RELOC(alloc_top_high));
	prom_printf("  rmo_top      : %x\n", RELOC(rmo_top));
	prom_printf("  ram_top      : %x\n", RELOC(ram_top));
}


/*
 * Allocate room for and instantiate RTAS
 */
static void __init prom_instantiate_rtas(void)
{
	phandle rtas_node;
	ihandle rtas_inst;
	u32 base, entry = 0;
	u32 size = 0;

	prom_debug("prom_instantiate_rtas: start...\n");

	rtas_node = call_prom("finddevice", 1, 1, ADDR("/rtas"));
	prom_debug("rtas_node: %x\n", rtas_node);
	if (!PHANDLE_VALID(rtas_node))
		return;

	prom_getprop(rtas_node, "rtas-size", &size, sizeof(size));
	if (size == 0)
		return;

	base = alloc_down(size, PAGE_SIZE, 0);
	if (base == 0) {
		prom_printf("RTAS allocation failed !\n");
		return;
	}

	rtas_inst = call_prom("open", 1, 1, ADDR("/rtas"));
	if (!IHANDLE_VALID(rtas_inst)) {
		prom_printf("opening rtas package failed (%x)\n", rtas_inst);
		return;
	}

	prom_printf("instantiating rtas at 0x%x ...", base);

	if (call_prom_ret("call-method", 3, 2, &entry,
			  ADDR("instantiate-rtas"),
			  rtas_inst, base) != 0
	    || entry == 0) {
		prom_printf(" failed\n");
		return;
	}
	prom_printf(" done\n");

	reserve_mem(base, size);

	prom_setprop(rtas_node, "/rtas", "linux,rtas-base",
		     &base, sizeof(base));
	prom_setprop(rtas_node, "/rtas", "linux,rtas-entry",
		     &entry, sizeof(entry));

	prom_debug("rtas base     = 0x%x\n", base);
	prom_debug("rtas entry    = 0x%x\n", entry);
	prom_debug("rtas size     = 0x%x\n", (long)size);

	prom_debug("prom_instantiate_rtas: end...\n");
}

#ifdef CONFIG_PPC64
/*
 * Allocate room for and initialize TCE tables
 */
static void __init prom_initialize_tce_table(void)
{
	phandle node;
	ihandle phb_node;
	char compatible[64], type[64], model[64];
	char *path = RELOC(prom_scratch);
	u64 base, align;
	u32 minalign, minsize;
	u64 tce_entry, *tce_entryp;
	u64 local_alloc_top, local_alloc_bottom;
	u64 i;

	if (RELOC(prom_iommu_off))
		return;

	prom_debug("starting prom_initialize_tce_table\n");

	/* Cache current top of allocs so we reserve a single block */
	local_alloc_top = RELOC(alloc_top_high);
	local_alloc_bottom = local_alloc_top;

	/* Search all nodes looking for PHBs. */
	for (node = 0; prom_next_node(&node); ) {
		compatible[0] = 0;
		type[0] = 0;
		model[0] = 0;
		prom_getprop(node, "compatible",
			     compatible, sizeof(compatible));
		prom_getprop(node, "device_type", type, sizeof(type));
		prom_getprop(node, "model", model, sizeof(model));

		if ((type[0] == 0) || (strstr(type, RELOC("pci")) == NULL))
			continue;

		/* Keep the old logic intact to avoid regression. */
		if (compatible[0] != 0) {
			if ((strstr(compatible, RELOC("python")) == NULL) &&
			    (strstr(compatible, RELOC("Speedwagon")) == NULL) &&
			    (strstr(compatible, RELOC("Winnipeg")) == NULL))
				continue;
		} else if (model[0] != 0) {
			if ((strstr(model, RELOC("ython")) == NULL) &&
			    (strstr(model, RELOC("peedwagon")) == NULL) &&
			    (strstr(model, RELOC("innipeg")) == NULL))
				continue;
		}

		if (prom_getprop(node, "tce-table-minalign", &minalign,
				 sizeof(minalign)) == PROM_ERROR)
			minalign = 0;
		if (prom_getprop(node, "tce-table-minsize", &minsize,
				 sizeof(minsize)) == PROM_ERROR)
			minsize = 4UL << 20;

		/*
		 * Even though we read what OF wants, we just set the table
		 * size to 4 MB.  This is enough to map 2GB of PCI DMA space.
		 * By doing this, we avoid the pitfalls of trying to DMA to
		 * MMIO space and the DMA alias hole.
		 *
		 * On POWER4, firmware sets the TCE region by assuming
		 * each TCE table is 8MB. Using this memory for anything
		 * else will impact performance, so we always allocate 8MB.
		 * Anton
		 */
		if (__is_processor(PV_POWER4) || __is_processor(PV_POWER4p))
			minsize = 8UL << 20;
		else
			minsize = 4UL << 20;

		/* Align to the greater of the align or size */
		align = max(minalign, minsize);
		base = alloc_down(minsize, align, 1);
		if (base == 0)
			prom_panic("ERROR, cannot find space for TCE table.\n");
		if (base < local_alloc_bottom)
			local_alloc_bottom = base;

		/* It seems OF doesn't null-terminate the path :-( */
		memset(path, 0, PROM_SCRATCH_SIZE);
		/* Call OF to setup the TCE hardware */
		if (call_prom("package-to-path", 3, 1, node,
			      path, PROM_SCRATCH_SIZE-1) == PROM_ERROR) {
			prom_printf("package-to-path failed\n");
		}

		/* Save away the TCE table attributes for later use. */
		prom_setprop(node, path, "linux,tce-base", &base, sizeof(base));
		prom_setprop(node, path, "linux,tce-size", &minsize, sizeof(minsize));

		prom_debug("TCE table: %s\n", path);
		prom_debug("\tnode = 0x%x\n", node);
		prom_debug("\tbase = 0x%x\n", base);
		prom_debug("\tsize = 0x%x\n", minsize);

		/* Initialize the table to have a one-to-one mapping
		 * over the allocated size.
		 */
		tce_entryp = (unsigned long *)base;
		for (i = 0; i < (minsize >> 3) ;tce_entryp++, i++) {
			tce_entry = (i << PAGE_SHIFT);
			tce_entry |= 0x3;
			*tce_entryp = tce_entry;
		}

		prom_printf("opening PHB %s", path);
		phb_node = call_prom("open", 1, 1, path);
		if (phb_node == 0)
			prom_printf("... failed\n");
		else
			prom_printf("... done\n");

		call_prom("call-method", 6, 0, ADDR("set-64-bit-addressing"),
			  phb_node, -1, minsize,
			  (u32) base, (u32) (base >> 32));
		call_prom("close", 1, 0, phb_node);
	}

	reserve_mem(local_alloc_bottom, local_alloc_top - local_alloc_bottom);

	/* These are only really needed if there is a memory limit in
	 * effect, but we don't know so export them always. */
	RELOC(prom_tce_alloc_start) = local_alloc_bottom;
	RELOC(prom_tce_alloc_end) = local_alloc_top;

	/* Flag the first invalid entry */
	prom_debug("ending prom_initialize_tce_table\n");
}
#endif

/*
 * With CHRP SMP we need to use the OF to start the other processors.
 * We can't wait until smp_boot_cpus (the OF is trashed by then)
 * so we have to put the processors into a holding pattern controlled
 * by the kernel (not OF) before we destroy the OF.
 *
 * This uses a chunk of low memory, puts some holding pattern
 * code there and sends the other processors off to there until
 * smp_boot_cpus tells them to do something.  The holding pattern
 * checks that address until its cpu # is there, when it is that
 * cpu jumps to __secondary_start().  smp_boot_cpus() takes care
 * of setting those values.
 *
 * We also use physical address 0x4 here to tell when a cpu
 * is in its holding pattern code.
 *
 * -- Cort
 */
extern void __secondary_hold(void);
extern unsigned long __secondary_hold_spinloop;
extern unsigned long __secondary_hold_acknowledge;

/*
 * We want to reference the copy of __secondary_hold_* in the
 * 0 - 0x100 address range
 */
#define LOW_ADDR(x)	(((unsigned long) &(x)) & 0xff)

static void __init prom_hold_cpus(void)
{
	unsigned long i;
	unsigned int reg;
	phandle node;
	char type[64];
	int cpuid = 0;
	unsigned int interrupt_server[MAX_CPU_THREADS];
	unsigned int cpu_threads, hw_cpu_num;
	int propsize;
	struct prom_t *_prom = &RELOC(prom);
	unsigned long *spinloop
		= (void *) LOW_ADDR(__secondary_hold_spinloop);
	unsigned long *acknowledge
		= (void *) LOW_ADDR(__secondary_hold_acknowledge);
#ifdef CONFIG_PPC64
	/* __secondary_hold is actually a descriptor, not the text address */
	unsigned long secondary_hold
		= __pa(*PTRRELOC((unsigned long *)__secondary_hold));
#else
	unsigned long secondary_hold = LOW_ADDR(__secondary_hold);
#endif

	prom_debug("prom_hold_cpus: start...\n");
	prom_debug("    1) spinloop       = 0x%x\n", (unsigned long)spinloop);
	prom_debug("    1) *spinloop      = 0x%x\n", *spinloop);
	prom_debug("    1) acknowledge    = 0x%x\n",
		   (unsigned long)acknowledge);
	prom_debug("    1) *acknowledge   = 0x%x\n", *acknowledge);
	prom_debug("    1) secondary_hold = 0x%x\n", secondary_hold);

	/* Set the common spinloop variable, so all of the secondary cpus
	 * will block when they are awakened from their OF spinloop.
	 * This must occur for both SMP and non SMP kernels, since OF will
	 * be trashed when we move the kernel.
	 */
	*spinloop = 0;

	/* look for cpus */
	for (node = 0; prom_next_node(&node); ) {
		type[0] = 0;
		prom_getprop(node, "device_type", type, sizeof(type));
		if (strcmp(type, RELOC("cpu")) != 0)
			continue;

		/* Skip non-configured cpus. */
		if (prom_getprop(node, "status", type, sizeof(type)) > 0)
			if (strcmp(type, RELOC("okay")) != 0)
				continue;

		reg = -1;
		prom_getprop(node, "reg", &reg, sizeof(reg));

		prom_debug("\ncpuid        = 0x%x\n", cpuid);
		prom_debug("cpu hw idx   = 0x%x\n", reg);

		/* Init the acknowledge var which will be reset by
		 * the secondary cpu when it awakens from its OF
		 * spinloop.
		 */
		*acknowledge = (unsigned long)-1;

		propsize = prom_getprop(node, "ibm,ppc-interrupt-server#s",
					&interrupt_server,
					sizeof(interrupt_server));
		if (propsize < 0) {
			/* no property.  old hardware has no SMT */
			cpu_threads = 1;
			interrupt_server[0] = reg; /* fake it with phys id */
		} else {
			/* We have a threaded processor */
			cpu_threads = propsize / sizeof(u32);
			if (cpu_threads > MAX_CPU_THREADS) {
				prom_printf("SMT: too many threads!\n"
					    "SMT: found %x, max is %x\n",
					    cpu_threads, MAX_CPU_THREADS);
				cpu_threads = 1; /* ToDo: panic? */
			}
		}

		hw_cpu_num = interrupt_server[0];
		if (hw_cpu_num != _prom->cpu) {