Skip to contentSkip to conversation

cleanup setup.c

PATCH In review Franck Bui-Huu 3 participants 17 messages Aug 8, 2006 – Aug 11, 2006 MIPS

lore
Share & tools
lore
#0 lore raw

cover cleanup setup.c

Here's a patchset that clean up arch/mips/kernel/setup.c file.

NOTE ! The semantic for initrd command line argument has been changed mainly because it is now a lot simpler to parse. Before we were using "rd_start=xxx rd_size=yyy", now the patch uses "initrd=yyy@xxx". It seems that no default config files use the old semantic but I'm not sure for bootloader users for example...

Am I allowed to do that ?

Franck

Overall diffstats:

 arch/mips/kernel/setup.c |  429 +++++++++++++++++++---------------------------
 1 files changed, 177 insertions(+), 252 deletions(-)
#1 lore raw

PATCH 1/6 setup.c: cleanup bootmem_init()

This function although doing simple thing is hard to follow. It's mainly due to:

- a lot of #ifdef
- bad local names
- redundant tests

So this patch try to address these issues. It also do not use max_pfn global which is marked as an unused exported symbol.

As a bonus side, it's now really easy to see what part of the code is for no-numa system.

There's also no point to make this function inline.

---
 arch/mips/kernel/setup.c |  135 +++++++++++++++++++---------------------------
 1 files changed, 57 insertions(+), 78 deletions(-)
arch/mips/kernel/setup.c +57 -78
index 8c2b596..e331e63 100644
@@ -96,6 +96,12 @@ void __init add_memory_region(phys_t sta
 	int x = boot_mem_map.nr_map;
 	struct boot_mem_map_entry *prev = boot_mem_map.map + x - 1;
 
+	/* Sanity check */
+	if (start + size < start) {
+		printk("Trying to add an invalid memory region, skipped\n");
+		return;
+	}
+
 	/*
 	 * Try to merge with previous entry if any.  This is far less than
 	 * perfect but is sufficient for most real world cases.
@@ -257,15 +263,16 @@ #endif
 	return 0;
 }
 
-#define MAXMEM		HIGHMEM_START
-#define MAXMEM_PFN	PFN_DOWN(MAXMEM)
-
-static inline void bootmem_init(void)
+/*
+ * Initialize the bootmem allocator. It also setup initrd related data
+ * if needed.
+ */
+static void __init bootmem_init(void)
 {
-	unsigned long start_pfn;
 	unsigned long reserved_end = (unsigned long)&_end;
 #ifndef CONFIG_SGI_IP27
-	unsigned long first_usable_pfn;
+	unsigned long highest = 0;
+	unsigned long mapstart = -1UL;
 	unsigned long bootmap_size;
 	int i;
 #endif
@@ -281,7 +288,7 @@ #ifdef CONFIG_BLK_DEV_INITRD
 		unsigned long tmp;
 		u32 *initrd_header;
 
-		tmp = ((reserved_end + PAGE_SIZE-1) & PAGE_MASK) - sizeof(u32) * 2;
+		tmp = PAGE_ALIGN(reserved_end) - sizeof(u32) * 2;
 		if (tmp < reserved_end)
 			tmp += PAGE_SIZE;
 		initrd_header = (u32 *)tmp;
@@ -294,16 +301,15 @@ #ifdef CONFIG_BLK_DEV_INITRD
 	}
 #endif	/* CONFIG_BLK_DEV_INITRD */
 
+#ifndef CONFIG_SGI_IP27
 	/*
-	 * Partially used pages are not usable - thus
-	 * we are rounding upwards.
+	 * reserved_end is now a pfn
 	 */
-	start_pfn = PFN_UP(CPHYSADDR(reserved_end));
+	reserved_end = PFN_UP(CPHYSADDR(reserved_end));
 
-#ifndef CONFIG_SGI_IP27
-	/* Find the highest page frame number we have available.  */
-	max_pfn = 0;
-	first_usable_pfn = -1UL;
+	/*
+	 * Find the highest page frame number we have available.
+	 */
 	for (i = 0; i < boot_mem_map.nr_map; i++) {
 		unsigned long start, end;
 
@@ -312,56 +318,38 @@ #ifndef CONFIG_SGI_IP27
 
 		start = PFN_UP(boot_mem_map.map[i].addr);
 		end = PFN_DOWN(boot_mem_map.map[i].addr
-		      + boot_mem_map.map[i].size);
+				+ boot_mem_map.map[i].size);
 
-		if (start >= end)
+		if (end > highest)
+			highest = end;
+		if (end <= reserved_end)
 			continue;
-		if (end > max_pfn)
-			max_pfn = end;
-		if (start < first_usable_pfn) {
-			if (start > start_pfn) {
-				first_usable_pfn = start;
-			} else if (end > start_pfn) {
-				first_usable_pfn = start_pfn;
-			}
-		}
+		if (start >= mapstart)
+			continue;
+		mapstart = max(reserved_end, start);
 	}
 
 	/*
 	 * Determine low and high memory ranges
 	 */
-	max_low_pfn = max_pfn;
-	if (max_low_pfn > MAXMEM_PFN) {
-		max_low_pfn = MAXMEM_PFN;
-#ifndef CONFIG_HIGHMEM
-		/* Maximum memory usable is what is directly addressable */
-		printk(KERN_WARNING "Warning only %ldMB will be used.\n",
-		       MAXMEM >> 20);
-		printk(KERN_WARNING "Use a HIGHMEM enabled kernel.\n");
+	if (highest > PFN_DOWN(HIGHMEM_START)) {
+#ifdef CONFIG_HIGHMEM
+		highstart_pfn = PFN_DOWN(HIGHMEM_START);
+		highend_pfn = highest;
 #endif
+		highest = PFN_DOWN(HIGHMEM_START);
 	}
 
-#ifdef CONFIG_HIGHMEM
 	/*
-	 * Crude, we really should make a better attempt at detecting
-	 * highstart_pfn
+	 * Initialize the boot-time allocator with low memory only.
 	 */
-	highstart_pfn = highend_pfn = max_pfn;
-	if (max_pfn > MAXMEM_PFN) {
-		highstart_pfn = MAXMEM_PFN;
-		printk(KERN_NOTICE "%ldMB HIGHMEM available.\n",
-		       (highend_pfn - highstart_pfn) >> (20 - PAGE_SHIFT));
-	}
-#endif
-
-	/* Initialize the boot-time allocator with low memory only.  */
-	bootmap_size = init_bootmem(first_usable_pfn, max_low_pfn);
+	bootmap_size = init_bootmem(mapstart, highest);
 
 	/*
 	 * Register fully available low RAM pages with the bootmem allocator.
 	 */
 	for (i = 0; i < boot_mem_map.nr_map; i++) {
-		unsigned long curr_pfn, last_pfn, size;
+		unsigned long start, end, size;
 
 		/*
 		 * Reserve usable memory.
@@ -369,49 +357,37 @@ #endif
 		if (boot_mem_map.map[i].type != BOOT_MEM_RAM)
 			continue;
 
-		/*
-		 * We are rounding up the start address of usable memory:
-		 */
-		curr_pfn = PFN_UP(boot_mem_map.map[i].addr);
-		if (curr_pfn >= max_low_pfn)
-			continue;
-		if (curr_pfn < start_pfn)
-			curr_pfn = start_pfn;
-
-		/*
-		 * ... and at the end of the usable range downwards:
-		 */
-		last_pfn = PFN_DOWN(boot_mem_map.map[i].addr
+		start = PFN_UP(boot_mem_map.map[i].addr);
+		end   = PFN_DOWN(boot_mem_map.map[i].addr
 				    + boot_mem_map.map[i].size);
-
-		if (last_pfn > max_low_pfn)
-			last_pfn = max_low_pfn;
-
 		/*
-		 * Only register lowmem part of lowmem segment with bootmem.
+		 * We are rounding up the start address of usable memory 
+		 * and at the end of the usable range downwards.
 		 */
-		size = last_pfn - curr_pfn;
-		if (curr_pfn > PFN_DOWN(HIGHMEM_START))
-			continue;
-		if (curr_pfn + size - 1 > PFN_DOWN(HIGHMEM_START))
-			size = PFN_DOWN(HIGHMEM_START) - curr_pfn;
-		if (!size)
+		if (start >= max_low_pfn)
 			continue;
+		if (start < reserved_end)
+			start = reserved_end;
+		if (end > max_low_pfn)
+			end = max_low_pfn;
 
 		/*
-		 * ... finally, did all the rounding and playing
-		 * around just make the area go away?
+		 * ... finally, is the area going away?
 		 */
-		if (last_pfn <= curr_pfn)
+		if (end <= start)
 			continue;
+		size = end - start;
 
 		/* Register lowmem ranges */
-		free_bootmem(PFN_PHYS(curr_pfn), PFN_PHYS(size));
-		memory_present(0, curr_pfn, curr_pfn + size - 1);
+		free_bootmem(PFN_PHYS(start), size << PAGE_SHIFT);
+		memory_present(0, start, end);
 	}
 
-	/* Reserve the bootmap memory.  */
-	reserve_bootmem(PFN_PHYS(first_usable_pfn), bootmap_size);
+	/*
+	 * Reserve the bootmap memory.
+	 */
+	reserve_bootmem(PFN_PHYS(mapstart), bootmap_size);
+
 #endif /* CONFIG_SGI_IP27 */
 
 #ifdef CONFIG_BLK_DEV_INITRD
@@ -483,6 +459,9 @@ static void __init arch_mem_init(char **
 	paging_init();
 }
 
+#define MAXMEM		HIGHMEM_START
+#define MAXMEM_PFN	PFN_DOWN(MAXMEM)
+
 static inline void resource_init(void)
 {
 	int i;
Signature · 1 line
1.4.2.rc2
#2 lore raw

PATCH 2/6 setup.c: move initrd code inside dedicated functions

In reply to Franck Bui-Huu

NUMA specific code could rely on them too.

---
 arch/mips/kernel/setup.c |  145 ++++++++++++++++++++++++++++------------------
 1 files changed, 87 insertions(+), 58 deletions(-)
arch/mips/kernel/setup.c +87 -58
index e331e63..45cba98 100644
@@ -200,7 +200,12 @@ static inline void parse_cmdline_early(v
 	}
 }
 
-static inline int parse_rd_cmdline(unsigned long* rd_start, unsigned long* rd_end)
+/*
+ * Manage initrd
+ */
+#ifdef CONFIG_BLK_DEV_INITRD
+
+static int __init parse_rd_cmdline(unsigned long *rd_start, unsigned long *rd_end)
 {
 	/*
 	 * "rd_start=0xNNNNNNNN" defines the memory address of an initrd
@@ -263,49 +268,94 @@ #endif
 	return 0;
 }
 
+static unsigned long __init init_initrd(void)
+{
+	unsigned long tmp, end;
+	u32 *initrd_header;
+
+	ROOT_DEV = Root_RAM0;
+	if (parse_rd_cmdline(&initrd_start, &initrd_end))
+		return initrd_end;
+
+	/* 
+	 * Board specific code should have set up initrd_start 
+	 * and initrd_end...
+	 */
+	end = (unsigned long)&_end;
+	tmp = PAGE_ALIGN(end) - sizeof(u32) * 2;
+	if (tmp < end)
+		tmp += PAGE_SIZE;
+	
+	initrd_header = (u32 *)tmp;
+	if (initrd_header[0] == 0x494E5244) {
+		initrd_start = (unsigned long)&initrd_header[2];
+		initrd_end = initrd_start + initrd_header[1];
+	}
+	return initrd_end;
+}
+
+static void __init finalize_initrd(void)
+{
+	unsigned long initrd_size = 
+		(unsigned long)initrd_end - (unsigned long)initrd_start;
+
+	if (initrd_size == 0) {
+		printk(KERN_INFO "Initrd not found or empty");
+		goto check_ko;
+	}
+	if (CPHYSADDR(initrd_end) > PFN_PHYS(max_low_pfn)) {
+		printk("Initrd extends beyond end of memory");
+		goto check_ko;
+	}
+
+	reserve_bootmem(CPHYSADDR(initrd_start), initrd_size);
+	initrd_below_start_ok = 1;
+
+	printk(KERN_INFO "Initial ramdisk at: 0x%p (%lu bytes)\n",
+	       (void *)initrd_start, initrd_size);
+	return;
+check_ko:
+	printk(" - disabling initrd\n");
+	initrd_start = 0;
+	initrd_end = 0;
+}
+
+#else  /* !CONFIG_BLK_DEV_INITRD */
+
+#define init_initrd()		0
+#define finalize_initrd()	do {} while (0)
+
+#endif
+
 /*
  * Initialize the bootmem allocator. It also setup initrd related data
  * if needed.
  */
+#ifdef CONFIG_SGI_IP27
+
 static void __init bootmem_init(void)
 {
-	unsigned long reserved_end = (unsigned long)&_end;
-#ifndef CONFIG_SGI_IP27
+	init_initrd();
+	finalize_initrd();
+}
+
+#else  /* !CONFIG_SGI_IP27 */
+
+static void __init bootmem_init(void)
+{
+	unsigned long reserved_end;
 	unsigned long highest = 0;
 	unsigned long mapstart = -1UL;
 	unsigned long bootmap_size;
 	int i;
-#endif
-#ifdef CONFIG_BLK_DEV_INITRD
-	int initrd_reserve_bootmem = 0;
-
-	/* Board specific code should have set up initrd_start and initrd_end */
- 	ROOT_DEV = Root_RAM0;
-	if (parse_rd_cmdline(&initrd_start, &initrd_end)) {
-		reserved_end = max(reserved_end, initrd_end);
-		initrd_reserve_bootmem = 1;
-	} else {
-		unsigned long tmp;
-		u32 *initrd_header;
-
-		tmp = PAGE_ALIGN(reserved_end) - sizeof(u32) * 2;
-		if (tmp < reserved_end)
-			tmp += PAGE_SIZE;
-		initrd_header = (u32 *)tmp;
-		if (initrd_header[0] == 0x494E5244) {
-			initrd_start = (unsigned long)&initrd_header[2];
-			initrd_end = initrd_start + initrd_header[1];
-			reserved_end = max(reserved_end, initrd_end);
-			initrd_reserve_bootmem = 1;
-		}
-	}
-#endif	/* CONFIG_BLK_DEV_INITRD */
 
-#ifndef CONFIG_SGI_IP27
 	/*
-	 * reserved_end is now a pfn
+	 * Init any data related to initrd. It's a nop if INITRD is
+	 * not selected. Once that done we can determine the low bound
+	 * of usable memory.
 	 */
-	reserved_end = PFN_UP(CPHYSADDR(reserved_end));
+	reserved_end = init_initrd();
+	reserved_end = PFN_UP(CPHYSADDR(max(reserved_end, (unsigned long)&_end)));
 
 	/*
 	 * Find the highest page frame number we have available.
@@ -388,35 +438,14 @@ #endif
 	 */
 	reserve_bootmem(PFN_PHYS(mapstart), bootmap_size);
 
-#endif /* CONFIG_SGI_IP27 */
-
-#ifdef CONFIG_BLK_DEV_INITRD
-	initrd_below_start_ok = 1;
-	if (initrd_start) {
-		unsigned long initrd_size = ((unsigned char *)initrd_end) -
-			((unsigned char *)initrd_start);
-		const int width = sizeof(long) * 2;
-
-		printk("Initial ramdisk at: 0x%p (%lu bytes)\n",
-		       (void *)initrd_start, initrd_size);
-
-		if (CPHYSADDR(initrd_end) > PFN_PHYS(max_low_pfn)) {
-			printk("initrd extends beyond end of memory "
-			       "(0x%0*Lx > 0x%0*Lx)\ndisabling initrd\n",
-			       width,
-			       (unsigned long long) CPHYSADDR(initrd_end),
-			       width,
-			       (unsigned long long) PFN_PHYS(max_low_pfn));
-			initrd_start = initrd_end = 0;
-			initrd_reserve_bootmem = 0;
-		}
-
-		if (initrd_reserve_bootmem)
-			reserve_bootmem(CPHYSADDR(initrd_start), initrd_size);
-	}
-#endif /* CONFIG_BLK_DEV_INITRD  */
+	/*
+	 * Reserve initrd memory if needed.
+	 */
+	finalize_initrd();
 }
 
+#endif	/* CONFIG_SGI_IP27 */
+
 /*
  * arch_mem_init - initialize memory managment subsystem
  *
Signature · 1 line
1.4.2.rc2
#3 lore raw

PATCH 3/6 setup.c: remove useless includes.

In reply to Franck Bui-Huu
---
 arch/mips/kernel/setup.c |   14 --------------
 1 files changed, 0 insertions(+), 14 deletions(-)
arch/mips/kernel/setup.c +0 -14
index 45cba98..49a466c 100644
@@ -10,29 +10,15 @@
  * Copyright (C) 1999 Silicon Graphics, Inc.
  * Copyright (C) 2000 2001, 2002  Maciej W. Rozycki
  */
-#include <linux/errno.h>
 #include <linux/init.h>
 #include <linux/ioport.h>
-#include <linux/sched.h>
-#include <linux/kernel.h>
-#include <linux/mm.h>
 #include <linux/module.h>
-#include <linux/stddef.h>
-#include <linux/string.h>
-#include <linux/unistd.h>
-#include <linux/slab.h>
-#include <linux/user.h>
-#include <linux/utsname.h>
-#include <linux/a.out.h>
 #include <linux/screen_info.h>
 #include <linux/bootmem.h>
 #include <linux/initrd.h>
-#include <linux/major.h>
-#include <linux/kdev_t.h>
 #include <linux/root_dev.h>
 #include <linux/highmem.h>
 #include <linux/console.h>
-#include <linux/mmzone.h>
 #include <linux/pfn.h>
 
 #include <asm/addrspace.h>
Signature · 1 line
1.4.2.rc2
#4 lore raw

PATCH 4/6 setup.c: do not inline functions

In reply to Franck Bui-Huu

There's no point to inline any functions in setup.c. Let's GCC doing its job, it's good enough for that now.

---
 arch/mips/kernel/setup.c |    4 ++--
 1 files changed, 2 insertions(+), 2 deletions(-)
arch/mips/kernel/setup.c +2 -2
index 49a466c..47395dd 100644
@@ -135,7 +135,7 @@ static void __init print_memory_map(void
 	}
 }
 
-static inline void parse_cmdline_early(void)
+static void __init parse_cmdline_early(void)
 {
 	char c = ' ', *to = command_line, *from = saved_command_line;
 	unsigned long start_at, mem_size;
@@ -477,7 +477,7 @@ static void __init arch_mem_init(char **
 #define MAXMEM		HIGHMEM_START
 #define MAXMEM_PFN	PFN_DOWN(MAXMEM)
 
-static inline void resource_init(void)
+static void __init resource_init(void)
 {
 	int i;
 
Signature · 1 line
1.4.2.rc2
#5 lore raw

PATCH 5/6 setup.c: remove MAXMEM macro

In reply to Franck Bui-Huu

It doens't improve readability.

---
 arch/mips/kernel/setup.c |   12 +++---------
 1 files changed, 3 insertions(+), 9 deletions(-)
arch/mips/kernel/setup.c +3 -9
index 47395dd..e835737 100644
@@ -474,9 +474,6 @@ static void __init arch_mem_init(char **
 	paging_init();
 }
 
-#define MAXMEM		HIGHMEM_START
-#define MAXMEM_PFN	PFN_DOWN(MAXMEM)
-
 static void __init resource_init(void)
 {
 	int i;
@@ -498,10 +495,10 @@ static void __init resource_init(void)
 
 		start = boot_mem_map.map[i].addr;
 		end = boot_mem_map.map[i].addr + boot_mem_map.map[i].size - 1;
-		if (start >= MAXMEM)
+		if (start >= HIGHMEM_START)
 			continue;
-		if (end >= MAXMEM)
-			end = MAXMEM - 1;
+		if (end >= HIGHMEM_START)
+			end = HIGHMEM_START - 1;
 
 		res = alloc_bootmem(sizeof(struct resource));
 		switch (boot_mem_map.map[i].type) {
@@ -530,9 +527,6 @@ static void __init resource_init(void)
 	}
 }
 
-#undef MAXMEM
-#undef MAXMEM_PFN
-
 void __init setup_arch(char **cmdline_p)
 {
 	cpu_probe();
Signature · 1 line
1.4.2.rc2
#6 lore raw

PATCH 6/6 setup.c: use early_param() for early command line parsing

In reply to Franck Bui-Huu

There's no point to rewrite some logic to parse command line to pass initrd parameters or to declare a user memory area. We could use instead parse_early_param() that does the same thing.

NOTE ! This patch also changes the initrd semantic. Old code was expecting "rd_start=xxx rd_size=xxx" which uses two parameters. Now the code expects "initrd=xxx@yyy" which is really simpler to parse and to use. No default config files use these parameters anyways but not sure for bootloader's users...

---
 arch/mips/kernel/setup.c |  173 +++++++++++++++-------------------------------
 1 files changed, 55 insertions(+), 118 deletions(-)
arch/mips/kernel/setup.c +55 -118
index e835737..ec459a1 100644
@@ -135,124 +135,29 @@ static void __init print_memory_map(void
 	}
 }
 
-static void __init parse_cmdline_early(void)
-{
-	char c = ' ', *to = command_line, *from = saved_command_line;
-	unsigned long start_at, mem_size;
-	int len = 0;
-	int usermem = 0;
-
-	printk("Determined physical RAM map:\n");
-	print_memory_map();
-
-	for (;;) {
-		/*
-		 * "mem=XXX[kKmM]" defines a memory region from
-		 * 0 to <XXX>, overriding the determined size.
-		 * "mem=XXX[KkmM]@YYY[KkmM]" defines a memory region from
-		 * <YYY> to <YYY>+<XXX>, overriding the determined size.
-		 */
-		if (c == ' ' && !memcmp(from, "mem=", 4)) {
-			if (to != command_line)
-				to--;
-			/*
-			 * If a user specifies memory size, we
-			 * blow away any automatically generated
-			 * size.
-			 */
-			if (usermem == 0) {
-				boot_mem_map.nr_map = 0;
-				usermem = 1;
-			}
-			mem_size = memparse(from + 4, &from);
-			if (*from == '@')
-				start_at = memparse(from + 1, &from);
-			else
-				start_at = 0;
-			add_memory_region(start_at, mem_size, BOOT_MEM_RAM);
-		}
-		c = *(from++);
-		if (!c)
-			break;
-		if (CL_SIZE <= ++len)
-			break;
-		*(to++) = c;
-	}
-	*to = '\0';
-
-	if (usermem) {
-		printk("User-defined physical RAM map:\n");
-		print_memory_map();
-	}
-}
-
 /*
  * Manage initrd
  */
 #ifdef CONFIG_BLK_DEV_INITRD
 
-static int __init parse_rd_cmdline(unsigned long *rd_start, unsigned long *rd_end)
+static int __init initrd_early(char *p)
 {
-	/*
-	 * "rd_start=0xNNNNNNNN" defines the memory address of an initrd
-	 * "rd_size=0xNN" it's size
-	 */
-	unsigned long start = 0;
-	unsigned long size = 0;
-	unsigned long end;
-	char cmd_line[CL_SIZE];
-	char *start_str;
-	char *size_str;
-	char *tmp;
-
-	strcpy(cmd_line, command_line);
-	*command_line = 0;
-	tmp = cmd_line;
-	/* Ignore "rd_start=" strings in other parameters. */
-	start_str = strstr(cmd_line, "rd_start=");
-	if (start_str && start_str != cmd_line && *(start_str - 1) != ' ')
-		start_str = strstr(start_str, " rd_start=");
-	while (start_str) {
-		if (start_str != cmd_line)
-			strncat(command_line, tmp, start_str - tmp);
-		start = memparse(start_str + 9, &start_str);
-		tmp = start_str + 1;
-		start_str = strstr(start_str, " rd_start=");
-	}
-	if (*tmp)
-		strcat(command_line, tmp);
-
-	strcpy(cmd_line, command_line);
-	*command_line = 0;
-	tmp = cmd_line;
-	/* Ignore "rd_size" strings in other parameters. */
-	size_str = strstr(cmd_line, "rd_size=");
-	if (size_str && size_str != cmd_line && *(size_str - 1) != ' ')
-		size_str = strstr(size_str, " rd_size=");
-	while (size_str) {
-		if (size_str != cmd_line)
-			strncat(command_line, tmp, size_str - tmp);
-		size = memparse(size_str + 8, &size_str);
-		tmp = size_str + 1;
-		size_str = strstr(size_str, " rd_size=");
-	}
-	if (*tmp)
-		strcat(command_line, tmp);
+	unsigned long start, size;
 
+	size = memparse(p, &p);
+	if (size && *p == '@') {
+		start = memparse(p + 1, &p);
 #ifdef CONFIG_64BIT
-	/* HACK: Guess if the sign extension was forgotten */
-	if (start > 0x0000000080000000 && start < 0x00000000ffffffff)
-		start |= 0xffffffff00000000UL;
+		/* HACK: Guess if the sign extension was forgotten */
+		if (start > 0x0000000080000000 && start < 0x00000000ffffffff)
+			start |= 0xffffffff00000000UL;
 #endif
-
-	end = start + size;
-	if (start && end) {
-		*rd_start = start;
-		*rd_end = end;
-		return 1;
+		initrd_start = start;
+		initrd_end = start + size;
 	}
 	return 0;
 }
+early_param("initrd", initrd_early);
 
 static unsigned long __init init_initrd(void)
 {
@@ -260,7 +165,7 @@ static unsigned long __init init_initrd(
 	u32 *initrd_header;
 
 	ROOT_DEV = Root_RAM0;
-	if (parse_rd_cmdline(&initrd_start, &initrd_end))
+	if (initrd_end)
 		return initrd_end;
 
 	/* 
@@ -282,25 +187,25 @@ static unsigned long __init init_initrd(
 
 static void __init finalize_initrd(void)
 {
-	unsigned long initrd_size = 
+	unsigned long size =
 		(unsigned long)initrd_end - (unsigned long)initrd_start;
 
-	if (initrd_size == 0) {
+	if (size == 0) {
 		printk(KERN_INFO "Initrd not found or empty");
-		goto check_ko;
+		goto disable;
 	}
 	if (CPHYSADDR(initrd_end) > PFN_PHYS(max_low_pfn)) {
 		printk("Initrd extends beyond end of memory");
-		goto check_ko;
+		goto disable;
 	}
 
-	reserve_bootmem(CPHYSADDR(initrd_start), initrd_size);
+	reserve_bootmem(CPHYSADDR(initrd_start), size);
 	initrd_below_start_ok = 1;
 
 	printk(KERN_INFO "Initial ramdisk at: 0x%p (%lu bytes)\n",
-	       (void *)initrd_start, initrd_size);
+	       (void *)initrd_start, size);
 	return;
-check_ko:
+disable:
 	printk(" - disabling initrd\n");
 	initrd_start = 0;
 	initrd_end = 0;
@@ -437,8 +342,6 @@ #endif	/* CONFIG_SGI_IP27 */
  *
  *  o plat_mem_setup() detects the memory configuration and will record detected
  *    memory areas using add_memory_region.
- *  o parse_cmdline_early() parses the command line for mem= options which,
- *    iff detected, will override the results of the automatic detection.
  *
  * At this stage the memory configuration of the system is known to the
  * kernel but generic memory managment system is still entirely uninitialized.
@@ -456,19 +359,53 @@ #endif	/* CONFIG_SGI_IP27 */
  * initialization hook for anything else was introduced.
  */
 
-extern void plat_mem_setup(void);
+static int usermem __initdata = 0;
+
+static int __init early_parse_mem(char *p)
+{
+	unsigned long start, size;
+
+	/*
+	 * If a user specifies memory size, we
+	 * blow away any automatically generated
+	 * size.
+	 */
+	if (usermem == 0) {
+		boot_mem_map.nr_map = 0;
+		usermem = 1;
+ 	}
+	start = 0;
+	size = memparse(p, &p);
+	if (*p == '@')
+		start = memparse(p + 1, &p);
+
+	add_memory_region(start, size, BOOT_MEM_RAM);
+	return 0;
+}
+early_param("mem", early_parse_mem);
 
 static void __init arch_mem_init(char **cmdline_p)
 {
+	extern void plat_mem_setup(void);
+
 	/* call board setup routine */
 	plat_mem_setup();
 
+	printk("Determined physical RAM map:\n");
+	print_memory_map();
+
 	strlcpy(command_line, arcs_cmdline, sizeof(command_line));
 	strlcpy(saved_command_line, command_line, COMMAND_LINE_SIZE);
 
 	*cmdline_p = command_line;
 
-	parse_cmdline_early();
+	parse_early_param();
+
+	if (usermem) {
+		printk("User-defined physical RAM map:\n");
+		print_memory_map();
+	}
+
 	bootmem_init();
 	sparse_init();
 	paging_init();
Signature · 1 line
1.4.2.rc2
#7 lore raw

Re: [PATCH 6/6] setup.c: use early_param() for early command line parsing

Quoted 11 lines · Franck Bui-Huu

There's no point to rewrite some logic to parse command line to pass initrd parameters or to declare a user memory area. We could use instead parse_early_param() that does the same thing.

NOTE ! This patch also changes the initrd semantic. Old code was expecting "rd_start=xxx rd_size=xxx" which uses two parameters. Now the code expects "initrd=xxx@yyy" which is really simpler to parse and to use. No default config files use these parameters anyways but not sure for bootloader's users...

This code is there precisely because most mips bootloaders use rd_start/rd_size. It also is IMHO a bad idea to overload the semantics of initrd= with both file names and memory locations.

Thiemo

#8 lore raw

Re: [PATCH 6/6] setup.c: use early_param() for early command line parsing

Quoted 13 lines · Thiemo Seufer
Quoted 11 lines · Franck Bui-Huu

There's no point to rewrite some logic to parse command line to pass initrd parameters or to declare a user memory area. We could use instead parse_early_param() that does the same thing.

NOTE ! This patch also changes the initrd semantic. Old code was expecting "rd_start=xxx rd_size=xxx" which uses two parameters. Now the code expects "initrd=xxx@yyy" which is really simpler to parse and to use. No default config files use these parameters anyways but not sure for bootloader's users...

Quoted 2 lines

This code is there precisely because most mips bootloaders use rd_start/rd_size.

OK, I guess we have to stick with this weird semantic...

Quoted 3 lines

It also is IMHO a bad idea to overload the
semantics of initrd= with both file names and memory locations.

I wasn't aware of any file name usages. Can you give a pointer ?

Thanks

Franck

#9 lore raw

Re: [PATCH 6/6] setup.c: use early_param() for early command line parsing

Quoted 17 lines · Franck Bui-Huu
Quoted 13 lines · Thiemo Seufer
Quoted 11 lines · Franck Bui-Huu

There's no point to rewrite some logic to parse command line to pass initrd parameters or to declare a user memory area. We could use instead parse_early_param() that does the same thing.

NOTE ! This patch also changes the initrd semantic. Old code was expecting "rd_start=xxx rd_size=xxx" which uses two parameters. Now the code expects "initrd=xxx@yyy" which is really simpler to parse and to use. No default config files use these parameters anyways but not sure for bootloader's users...

Quoted 2 lines

This code is there precisely because most mips bootloaders use rd_start/rd_size.

Quoted 6 lines

OK, I guess we have to stick with this weird semantic...

Quoted 2 lines

It also is IMHO a bad idea to overload the
semantics of initrd= with both file names and memory locations.

I wasn't aware of any file name usages. Can you give a pointer ?

Documentation/initrd.txt Documentation/filesystems/ramfs-rootfs-initramfs.txt

Thiemo

#10 lore raw

Re: [PATCH 6/6] setup.c: use early_param() for early command line parsing

Quoted 16 lines · Thiemo Seufer
Quoted 15 lines · Franck Bui-Huu
Quoted 14 lines · Thiemo Seufer
Quoted 11 lines · Franck Bui-Huu

There's no point to rewrite some logic to parse command line to pass initrd parameters or to declare a user memory area. We could use instead parse_early_param() that does the same thing.

NOTE ! This patch also changes the initrd semantic. Old code was expecting "rd_start=xxx rd_size=xxx" which uses two parameters. Now the code expects "initrd=xxx@yyy" which is really simpler to parse and to use. No default config files use these parameters anyways but not sure for bootloader's users...

This code is there precisely because most mips bootloaders use rd_start/rd_size.

Quoted 9 lines
Quoted 5 lines

OK, I guess we have to stick with this weird semantic...

Quoted 2 lines

It also is IMHO a bad idea to overload the
semantics of initrd= with both file names and memory locations.

I wasn't aware of any file name usages. Can you give a pointer ?

Documentation/initrd.txt Documentation/filesystems/ramfs-rootfs-initramfs.txt

I was asking for pointers on MIPS bootloaders which use initrd=/path/to/initrd...

Anyways, you're talking about specific bootloader's parameters, aren't you ? I don't know any MIPS bootloaders, but I wouldn't expect them to pass their own parameters to the kernel, that would be surprising...

What are you suggesting ? kernel_initrd ?

BTW, what do you think about rd_start/rd_size names ?

thanks

Franck

#11 lore raw

Re: [PATCH 6/6] setup.c: use early_param() for early command line parsing

Franck Bui-Huu wrote:
[snip]

Quoted 10 lines
Quoted 7 lines
Quoted 3 lines
Quoted 2 lines

It also is IMHO a bad idea to overload the
semantics of initrd= with both file names and memory locations.

I wasn't aware of any file name usages. Can you give a pointer ?

Documentation/initrd.txt Documentation/filesystems/ramfs-rootfs-initramfs.txt

I was asking for pointers on MIPS bootloaders which use initrd=/path/to/initrd...

AFAIR arcboot does.

Quoted 8 lines

Anyways, you're talking about specific bootloader's parameters, aren't you ? I don't know any MIPS bootloaders, but I wouldn't expect them to pass their own parameters to the kernel, that would be surprising...

What are you suggesting ? kernel_initrd ?

BTW, what do you think about rd_start/rd_size names ?

Is there a good reason to change it?

Thiemo

#12 lore raw

Re: [PATCH 6/6] setup.c: use early_param() for early command line parsing

In reply to Franck Bui-Huu
Quoted 11 lines · Franck Bui-Huu
Quoted 9 lines
Quoted 6 lines

NOTE ! This patch also changes the initrd semantic. Old code was expecting "rd_start=xxx rd_size=xxx" which uses two parameters. Now the code expects "initrd=xxx@yyy" which is really simpler to parse and to use. No default config files use these parameters anyways but not sure for bootloader's users...

This code is there precisely because most mips bootloaders use rd_start/rd_size.

OK, I guess we have to stick with this weird semantic...

Maybe you can add something like "initrdmem=xxx@yyy", keeping "rd_start" and "rd_size" for the backward compatibility. Just a thought.

---
Atsushi Nemoto

#13 lore raw

Re: [PATCH 6/6] setup.c: use early_param() for early command line parsing

Hi Atsushi,

Quoted 7 lines · Atsushi Nemoto
Quoted 6 lines · Franck Bui-Huu
Quoted 6 lines
Quoted 6 lines

NOTE ! This patch also changes the initrd semantic. Old code was expecting "rd_start=xxx rd_size=xxx" which uses two parameters. Now the code expects "initrd=xxx@yyy" which is really simpler to parse and to use. No default config files use these parameters anyways but not sure for bootloader's users...

Quoted 8 lines
Quoted 3 lines
Quoted 2 lines

This code is there precisely because most mips bootloaders use rd_start/rd_size.

OK, I guess we have to stick with this weird semantic...

Maybe you can add something like "initrdmem=xxx@yyy", keeping "rd_start" and "rd_size" for the backward compatibility. Just a thought.

Well that what I was planning when writing this patch but I didn't. I think that we will end up with two different semantics and the old one never replaced by the new one... Except if we mark them as deprecated by showing a warning at boot. What do you think ?

Franck

#14 lore raw

Re: [PATCH 6/6] setup.c: use early_param() for early command line parsing

Quoted 8 lines · Franck Bui-Huu
Quoted 3 lines

Maybe you can add something like "initrdmem=xxx@yyy", keeping "rd_start" and "rd_size" for the backward compatibility. Just a thought.

Well that what I was planning when writing this patch but I didn't. I think that we will end up with two different semantics and the old one never replaced by the new one... Except if we mark them as deprecated by showing a warning at boot. What do you think ?

While the kernel command line is very limited resource (only 256 chars), I prefer a single short option to specify initrd range, if available.

But nothing wrong with rd_start and rd_size, and it seems there are some boot loader expected them already, so removing them would not be good (especially without some grace period).

I don't care if there were two way to specify initrd range. It would be somewhat redundant, but that is usual on "Backword compatibility" issue, isn't it? ;-)

---
Atsushi Nemoto

#15 lore raw

Re: [PATCH 6/6] setup.c: use early_param() for early command line parsing

Quoted 13 lines · Atsushi Nemoto
Quoted 7 lines · Franck Bui-Huu
Quoted 3 lines

Maybe you can add something like "initrdmem=xxx@yyy", keeping "rd_start" and "rd_size" for the backward compatibility. Just a thought.

Well that what I was planning when writing this patch but I didn't. I think that we will end up with two different semantics and the old one never replaced by the new one... Except if we mark them as deprecated by showing a warning at boot. What do you think ?

While the kernel command line is very limited resource (only 256 chars), I prefer a single short option to specify initrd range, if available.

Quoted 8 lines

But nothing wrong with rd_start and rd_size, and it seems there are some boot loader expected them already, so removing them would not be good (especially without some grace period).

I don't care if there were two way to specify initrd range. It would be somewhat redundant, but that is usual on "Backword compatibility" issue, isn't it? ;-)

Well, I resent a new version (take #2) of the patchset that uses _only_ "rd_xxx" semantic. I prefer not add some code which isn't going to be used. Mainly because only bootloaders use this parameter and I guess they never change the way they pass initrd address. And there won't be a lot of new bootloaders anyways.

Thanks
Franck

#16 lore raw

Re: [PATCH 6/6] setup.c: use early_param() for early command line parsing

Quoted 5 lines · Franck Bui-Huu

Well, I resent a new version (take #2) of the patchset that uses _only_ "rd_xxx" semantic. I prefer not add some code which isn't going to be used. Mainly because only bootloaders use this parameter and I guess they never change the way they pass initrd address. And there won't be a lot of new bootloaders anyways.

I see. OK for me.
---
Atsushi Nemoto