Merge bk://gkernel.bkbits.net/libata-2.6

into ppc970.osdl.org:/home/torvalds/v2.6/linux

arch/mips/kernel/irixelf.c

@@ -127,7 +127,9 @@ static void set_brk(unsigned long start, unsigned long end)
 	end = PAGE_ALIGN(end);
 	if (end <= start)
 		return;
+	down_write(&current->mm->mmap_sem);
 	do_brk(start, end - start);
+	up_write(&current->mm->mmap_sem);
 }
 
 
@@ -375,7 +377,9 @@ static unsigned int load_irix_interp(struct elfhdr * interp_elf_ex,
 
 	/* Map the last of the bss segment */
 	if (last_bss > len) {
+		down_write(&current->mm->mmap_sem);
 		do_brk(len, (last_bss - len));
+		up_write(&current->mm->mmap_sem);
 	}
 	kfree(elf_phdata);
 
@@ -562,7 +566,9 @@ void irix_map_prda_page (void)
 	unsigned long v;
 	struct prda *pp;
 
+	down_write(&current->mm->mmap_sem);
 	v =  do_brk (PRDA_ADDRESS, PAGE_SIZE);
+	up_write(&current->mm->mmap_sem);
 
 	if (v < 0)
 		return;
@@ -852,8 +858,11 @@ static int load_irix_library(struct file *file)
 
 	len = (elf_phdata->p_filesz + elf_phdata->p_vaddr+ 0xfff) & 0xfffff000;
 	bss = elf_phdata->p_memsz + elf_phdata->p_vaddr;
-	if (bss > len)
+	if (bss > len) {
+	  down_write(&current->mm->mmap_sem);
 	  do_brk(len, bss-len);
+	  up_write(&current->mm->mmap_sem);
+	}
 	kfree(elf_phdata);
 	return 0;
 }

arch/sparc64/kernel/binfmt_aout32.c

@@ -49,7 +49,9 @@ static void set_brk(unsigned long start, unsigned long end)
 	end = PAGE_ALIGN(end);
 	if (end <= start)
 		return;
+	down_write(&current->mm->mmap_sem);
 	do_brk(start, end - start);
+	up_write(&current->mm->mmap_sem);
 }
 
 /*
@@ -246,10 +248,14 @@ static int load_aout32_binary(struct linux_binprm * bprm, struct pt_regs * regs)
 	if (N_MAGIC(ex) == NMAGIC) {
 		loff_t pos = fd_offset;
 		/* Fuck me plenty... */
+		down_write(&current->mm->mmap_sem);	
 		error = do_brk(N_TXTADDR(ex), ex.a_text);
+		up_write(&current->mm->mmap_sem);
 		bprm->file->f_op->read(bprm->file, (char __user *)N_TXTADDR(ex),
 			  ex.a_text, &pos);
+		down_write(&current->mm->mmap_sem);
 		error = do_brk(N_DATADDR(ex), ex.a_data);
+		up_write(&current->mm->mmap_sem);
 		bprm->file->f_op->read(bprm->file, (char __user *)N_DATADDR(ex),
 			  ex.a_data, &pos);
 		goto beyond_if;
@@ -257,8 +263,10 @@ static int load_aout32_binary(struct linux_binprm * bprm, struct pt_regs * regs)
 
 	if (N_MAGIC(ex) == OMAGIC) {
 		loff_t pos = fd_offset;
+		down_write(&current->mm->mmap_sem);
 		do_brk(N_TXTADDR(ex) & PAGE_MASK,
 			ex.a_text+ex.a_data + PAGE_SIZE - 1);
+		up_write(&current->mm->mmap_sem);
 		bprm->file->f_op->read(bprm->file, (char __user *)N_TXTADDR(ex),
 			  ex.a_text+ex.a_data, &pos);
 	} else {
@@ -272,7 +280,9 @@ static int load_aout32_binary(struct linux_binprm * bprm, struct pt_regs * regs)
 
 		if (!bprm->file->f_op->mmap) {
 			loff_t pos = fd_offset;
+			down_write(&current->mm->mmap_sem);
 			do_brk(0, ex.a_text+ex.a_data);
+			up_write(&current->mm->mmap_sem);
 			bprm->file->f_op->read(bprm->file,
 				  (char __user *)N_TXTADDR(ex),
 				  ex.a_text+ex.a_data, &pos);
@@ -389,7 +399,9 @@ static int load_aout32_library(struct file *file)
 	len = PAGE_ALIGN(ex.a_text + ex.a_data);
 	bss = ex.a_text + ex.a_data + ex.a_bss;
 	if (bss > len) {
+		down_write(&current->mm->mmap_sem);
 		error = do_brk(start_addr + len, bss - len);
+		up_write(&current->mm->mmap_sem);
 		retval = error;
 		if (error != start_addr + len)
 			goto out;

arch/x86_64/ia32/ia32_aout.c

@@ -115,7 +115,9 @@ static void set_brk(unsigned long start, unsigned long end)
 	end = PAGE_ALIGN(end);
 	if (end <= start)
 		return;
+	down_write(&current->mm->mmap_sem);
 	do_brk(start, end - start);
+	up_write(&current->mm->mmap_sem);
 }
 
 #if CORE_DUMP
@@ -325,7 +327,10 @@ static int load_aout_binary(struct linux_binprm * bprm, struct pt_regs * regs)
 		pos = 32;
 		map_size = ex.a_text+ex.a_data;
 
+		down_write(&current->mm->mmap_sem);
 		error = do_brk(text_addr & PAGE_MASK, map_size);
+		up_write(&current->mm->mmap_sem);
+
 		if (error != (text_addr & PAGE_MASK)) {
 			send_sig(SIGKILL, current, 0);
 			return error;
@@ -361,7 +366,9 @@ static int load_aout_binary(struct linux_binprm * bprm, struct pt_regs * regs)
 
 		if (!bprm->file->f_op->mmap||((fd_offset & ~PAGE_MASK) != 0)) {
 			loff_t pos = fd_offset;
+			down_write(&current->mm->mmap_sem);
 			do_brk(N_TXTADDR(ex), ex.a_text+ex.a_data);
+			up_write(&current->mm->mmap_sem);
 			bprm->file->f_op->read(bprm->file,(char *)N_TXTADDR(ex),
 					ex.a_text+ex.a_data, &pos);
 			flush_icache_range((unsigned long) N_TXTADDR(ex),
@@ -469,8 +476,9 @@ static int load_aout_library(struct file *file)
 			error_time = jiffies;
 		}
 #endif
-
+		down_write(&current->mm->mmap_sem);
 		do_brk(start_addr, ex.a_text + ex.a_data + ex.a_bss);
+		up_write(&current->mm->mmap_sem);
 		
 		file->f_op->read(file, (char *)start_addr,
 			ex.a_text + ex.a_data, &pos);
@@ -494,7 +502,9 @@ static int load_aout_library(struct file *file)
 	len = PAGE_ALIGN(ex.a_text + ex.a_data);
 	bss = ex.a_text + ex.a_data + ex.a_bss;
 	if (bss > len) {
+		down_write(&current->mm->mmap_sem);
 		error = do_brk(start_addr + len, bss - len);
+		up_write(&current->mm->mmap_sem);
 		retval = error;
 		if (error != start_addr + len)
 			goto out;

arch/x86_64/mm/k8topology.c

@@ -47,6 +47,10 @@ int __init k8_scan_nodes(unsigned long start, unsigned long end)
 	int nodeid, i, nb; 
 	int found = 0;
 	u32 reg;
+	unsigned numnodes;
+	nodemask_t nodes_parsed;
+
+	nodes_clear(nodes_parsed);
 
 	nb = find_northbridge(); 
 	if (nb < 0) 
@@ -55,10 +59,9 @@ int __init k8_scan_nodes(unsigned long start, unsigned long end)
 	printk(KERN_INFO "Scanning NUMA topology in Northbridge %d\n", nb); 
 
 	reg = read_pci_config(0, nb, 0, 0x60); 
-	for (i = 0; i <= ((reg >> 4) & 7); i++)
-		node_set_online(i);
+	numnodes = ((reg >> 4) & 7) + 1;
 
-	printk(KERN_INFO "Number of nodes %d (%x)\n", num_online_nodes(), reg);
+	printk(KERN_INFO "Number of nodes %d\n", numnodes);
 
 	memset(&nodes,0,sizeof(nodes)); 
 	prevbase = 0;
@@ -70,11 +73,11 @@ int __init k8_scan_nodes(unsigned long start, unsigned long end)
 
 		nodeid = limit & 7; 
 		if ((base & 3) == 0) { 
-			if (i < num_online_nodes())
+			if (i < numnodes)
 				printk("Skipping disabled node %d\n", i); 
 			continue;
 		} 
-		if (nodeid >= num_online_nodes()) {
+		if (nodeid >= numnodes) {
 			printk("Ignoring excess node %d (%lx:%lx)\n", nodeid,
 			       base, limit); 
 			continue;
@@ -90,7 +93,7 @@ int __init k8_scan_nodes(unsigned long start, unsigned long end)
 			       nodeid, (base>>8)&3, (limit>>8) & 3); 
 			return -1; 
 		}	
-		if (node_online(nodeid)) { 
+		if (node_isset(nodeid, nodes_parsed)) { 
 			printk(KERN_INFO "Node %d already present. Skipping\n", 
 			       nodeid);
 			continue;
@@ -138,12 +141,14 @@ int __init k8_scan_nodes(unsigned long start, unsigned long end)
 		nodes[nodeid].end = limit;
 
 		prevbase = base;
+
+		node_set(nodeid, nodes_parsed);
 	} 
 
 	if (!found)
 		return -1; 
 
-	memnode_shift = compute_hash_shift(nodes);
+	memnode_shift = compute_hash_shift(nodes, numnodes);
 	if (memnode_shift < 0) { 
 		printk(KERN_ERR "No NUMA node hash function found. Contact maintainer\n"); 
 		return -1; 
@@ -154,8 +159,8 @@ int __init k8_scan_nodes(unsigned long start, unsigned long end)
 		if (nodes[i].start != nodes[i].end) { 
 			/* assume 1:1 NODE:CPU */
 			cpu_to_node[i] = i; 
-		setup_node_bootmem(i, nodes[i].start, nodes[i].end); 
-	} 
+			setup_node_bootmem(i, nodes[i].start, nodes[i].end); 
+		} 
 	}
 
 	numa_init_array();

arch/x86_64/mm/numa.c

@@ -34,9 +34,7 @@ cpumask_t     node_to_cpumask[MAX_NUMNODES];
 
 int numa_off __initdata;
 
-unsigned long nodes_present; 
-
-int __init compute_hash_shift(struct node *nodes)
+int __init compute_hash_shift(struct node *nodes, int numnodes)
 {
 	int i; 
 	int shift = 24;
@@ -45,7 +43,7 @@ int __init compute_hash_shift(struct node *nodes)
 	/* When in doubt use brute force. */
 	while (shift < 48) { 
 		memset(memnodemap,0xff,sizeof(*memnodemap) * NODEMAPSIZE); 
-		for_each_online_node(i) {
+		for (i = 0; i < numnodes; i++) {
 			if (nodes[i].start == nodes[i].end) 
 				continue;
 			for (addr = nodes[i].start; 
@@ -197,7 +195,7 @@ static int numa_emulation(unsigned long start_pfn, unsigned long end_pfn)
  		       (nodes[i].end - nodes[i].start) >> 20);
 		node_set_online(i);
  	}
- 	memnode_shift = compute_hash_shift(nodes);
+ 	memnode_shift = compute_hash_shift(nodes, numa_fake);
  	if (memnode_shift < 0) {
  		memnode_shift = 0;
  		printk(KERN_ERR "No NUMA hash function found. Emulation disabled.\n");
@@ -274,7 +272,7 @@ void __init paging_init(void)
 /* [numa=off] */
 __init int numa_setup(char *opt) 
 { 
-	if (!strcmp(opt,"off"))
+	if (!strncmp(opt,"off",3))
 		numa_off = 1;
 #ifdef CONFIG_NUMA_EMU
 	if(!strncmp(opt, "fake=", 5)) {

arch/x86_64/mm/srat.c

@@ -20,17 +20,20 @@
 
 static struct acpi_table_slit *acpi_slit;
 
-static DECLARE_BITMAP(nodes_parsed, MAX_NUMNODES) __initdata;
+static nodemask_t nodes_parsed __initdata;
+static nodemask_t nodes_found __initdata;
 static struct node nodes[MAX_NUMNODES] __initdata;
 static __u8  pxm2node[256] __initdata = { [0 ... 255] = 0xff };
 
 static __init int setup_node(int pxm)
 {
-	if (pxm2node[pxm] == 0xff) {
-		if (num_online_nodes() >= MAX_NUMNODES)
+	unsigned node = pxm2node[pxm];
+	if (node == 0xff) {
+		if (nodes_weight(nodes_found) >= MAX_NUMNODES)
 			return -1;
-		pxm2node[pxm] = num_online_nodes();
-		node_set_online(num_online_nodes());
+		node = first_unset_node(nodes_found); 
+		node_set(node, nodes_found);
+		pxm2node[pxm] = node;
 	}
 	return pxm2node[pxm];
 }
@@ -140,7 +143,7 @@ acpi_numa_memory_affinity_init(struct acpi_table_memory_affinity *ma)
 		return;
 	}
 	nd = &nodes[node];
-	if (!test_and_set_bit(node, &nodes_parsed)) {
+	if (!node_test_and_set(node, nodes_parsed)) {
 		nd->start = start;
 		nd->end = end;
 	} else {
@@ -163,7 +166,7 @@ int __init acpi_scan_nodes(unsigned long start, unsigned long end)
 	int i;
 	if (acpi_numa <= 0)
 		return -1;
-	memnode_shift = compute_hash_shift(nodes);
+	memnode_shift = compute_hash_shift(nodes, nodes_weight(nodes_parsed));
 	if (memnode_shift < 0) {
 		printk(KERN_ERR
 		     "SRAT: No NUMA node hash function found. Contact maintainer\n");
@@ -171,7 +174,7 @@ int __init acpi_scan_nodes(unsigned long start, unsigned long end)
 		return -1;
 	}
 	for (i = 0; i < MAX_NUMNODES; i++) {
-		if (!test_bit(i, &nodes_parsed))
+		if (!node_isset(i, nodes_parsed))
 			continue;
 		cutoff_node(i, start, end);
 		if (nodes[i].start == nodes[i].end)

fs/binfmt_aout.c

@@ -50,7 +50,10 @@ static int set_brk(unsigned long start, unsigned long end)
 	start = PAGE_ALIGN(start);
 	end = PAGE_ALIGN(end);
 	if (end > start) {
-		unsigned long addr = do_brk(start, end - start);
+		unsigned long addr;
+		down_write(&current->mm->mmap_sem);
+		addr = do_brk(start, end - start);
+		up_write(&current->mm->mmap_sem);
 		if (BAD_ADDR(addr))
 			return addr;
 	}
@@ -323,10 +326,14 @@ static int load_aout_binary(struct linux_binprm * bprm, struct pt_regs * regs)
 		loff_t pos = fd_offset;
 		/* Fuck me plenty... */
 		/* <AOL></AOL> */
+		down_write(&current->mm->mmap_sem);	
 		error = do_brk(N_TXTADDR(ex), ex.a_text);
+		up_write(&current->mm->mmap_sem);
 		bprm->file->f_op->read(bprm->file, (char *) N_TXTADDR(ex),
 			  ex.a_text, &pos);
+		down_write(&current->mm->mmap_sem);
 		error = do_brk(N_DATADDR(ex), ex.a_data);
+		up_write(&current->mm->mmap_sem);
 		bprm->file->f_op->read(bprm->file, (char *) N_DATADDR(ex),
 			  ex.a_data, &pos);
 		goto beyond_if;
@@ -346,8 +353,9 @@ static int load_aout_binary(struct linux_binprm * bprm, struct pt_regs * regs)
 		pos = 32;
 		map_size = ex.a_text+ex.a_data;
 #endif
-
+		down_write(&current->mm->mmap_sem);
 		error = do_brk(text_addr & PAGE_MASK, map_size);
+		up_write(&current->mm->mmap_sem);
 		if (error != (text_addr & PAGE_MASK)) {
 			send_sig(SIGKILL, current, 0);
 			return error;
@@ -382,7 +390,9 @@ static int load_aout_binary(struct linux_binprm * bprm, struct pt_regs * regs)
 
 		if (!bprm->file->f_op->mmap||((fd_offset & ~PAGE_MASK) != 0)) {
 			loff_t pos = fd_offset;
+			down_write(&current->mm->mmap_sem);
 			do_brk(N_TXTADDR(ex), ex.a_text+ex.a_data);
+			up_write(&current->mm->mmap_sem);
 			bprm->file->f_op->read(bprm->file,
 					(char __user *)N_TXTADDR(ex),
 					ex.a_text+ex.a_data, &pos);
@@ -487,8 +497,9 @@ static int load_aout_library(struct file *file)
 			       file->f_dentry->d_name.name);
 			error_time = jiffies;
 		}
-
+		down_write(&current->mm->mmap_sem);
 		do_brk(start_addr, ex.a_text + ex.a_data + ex.a_bss);
+		up_write(&current->mm->mmap_sem);
 		
 		file->f_op->read(file, (char __user *)start_addr,
 			ex.a_text + ex.a_data, &pos);

fs/binfmt_elf.c

@@ -88,7 +88,10 @@ static int set_brk(unsigned long start, unsigned long end)
 	start = ELF_PAGEALIGN(start);
 	end = ELF_PAGEALIGN(end);
 	if (end > start) {
-		unsigned long addr = do_brk(start, end - start);
+		unsigned long addr;
+		down_write(&current->mm->mmap_sem);
+		addr = do_brk(start, end - start);
+		up_write(&current->mm->mmap_sem);
 		if (BAD_ADDR(addr))
 			return addr;
 	}
@@ -409,7 +412,9 @@ static unsigned long load_elf_interp(struct elfhdr * interp_elf_ex,
 
 	/* Map the last of the bss segment */
 	if (last_bss > elf_bss) {
+		down_write(&current->mm->mmap_sem);
 		error = do_brk(elf_bss, last_bss - elf_bss);
+		up_write(&current->mm->mmap_sem);
 		if (BAD_ADDR(error))
 			goto out_close;
 	}
@@ -449,7 +454,9 @@ static unsigned long load_aout_interp(struct exec * interp_ex,
 		goto out;
 	}
 
+	down_write(&current->mm->mmap_sem);	
 	do_brk(0, text_data);
+	up_write(&current->mm->mmap_sem);
 	if (!interpreter->f_op || !interpreter->f_op->read)
 		goto out;
 	if (interpreter->f_op->read(interpreter, addr, text_data, &offset) < 0)
@@ -457,8 +464,11 @@ static unsigned long load_aout_interp(struct exec * interp_ex,
 	flush_icache_range((unsigned long)addr,
 	                   (unsigned long)addr + text_data);
 
+
+	down_write(&current->mm->mmap_sem);	
 	do_brk(ELF_PAGESTART(text_data + ELF_MIN_ALIGN - 1),
 		interp_ex->a_bss);
+	up_write(&current->mm->mmap_sem);
 	elf_entry = interp_ex->a_entry;
 
 out:

include/asm-x86_64/numa.h

@@ -8,7 +8,7 @@ struct node {
 	u64 start,end; 
 };
 
-extern int compute_hash_shift(struct node *nodes);
+extern int compute_hash_shift(struct node *nodes, int numnodes);
 
 #define ZONE_ALIGN (1UL << (MAX_ORDER+PAGE_SHIFT))
 

include/linux/nodemask.h

@@ -38,6 +38,8 @@
  *
  * int first_node(mask)			Number lowest set bit, or MAX_NUMNODES
  * int next_node(node, mask)		Next node past 'node', or MAX_NUMNODES
+ * int first_unset_node(mask)		First node not set in mask, or 
+ *					MAX_NUMNODES.
  *
  * nodemask_t nodemask_of_node(node)	Return nodemask with bit 'node' set
  * NODE_MASK_ALL			Initializer - all bits set
@@ -235,6 +237,13 @@ static inline int __next_node(int n, const nodemask_t *srcp, int nbits)
 	m;								\
 })
 
+#define first_unset_node(mask) __first_unset_node(&(mask))
+static inline int __first_unset_node(const nodemask_t *maskp)
+{
+	return min_t(int,MAX_NUMNODES,
+			find_first_zero_bit(maskp->bits, MAX_NUMNODES));
+}
+
 #define NODE_MASK_LAST_WORD BITMAP_LAST_WORD_MASK(MAX_NUMNODES)
 
 #if MAX_NUMNODES <= BITS_PER_LONG

mm/mmap.c

@@ -1475,7 +1475,6 @@ static int acct_stack_growth(struct vm_area_struct * vma, unsigned long size, un
 int expand_stack(struct vm_area_struct * vma, unsigned long address)
 {
 	int error;
-	unsigned long size, grow;
 
 	if (!(vma->vm_flags & VM_GROWSUP))
 		return -EFAULT;
@@ -1495,12 +1494,19 @@ int expand_stack(struct vm_area_struct * vma, unsigned long address)
 	 */
 	address += 4 + PAGE_SIZE - 1;
 	address &= PAGE_MASK;
-	size = address - vma->vm_start;
-	grow = (address - vma->vm_end) >> PAGE_SHIFT;
+	error = 0;
 
-	error = acct_stack_growth(vma, size, grow);
-	if (!error)
-		vma->vm_end = address;
+	/* Somebody else might have raced and expanded it already */
+	if (address > vma->vm_end) {
+		unsigned long size, grow;
+
+		size = address - vma->vm_start;
+		grow = (address - vma->vm_end) >> PAGE_SHIFT;
+
+		error = acct_stack_growth(vma, size, grow);
+		if (!error)
+			vma->vm_end = address;
+	}
 	anon_vma_unlock(vma);
 	return error;
 }
@@ -1528,7 +1534,6 @@ find_extend_vma(struct mm_struct *mm, unsigned long addr)
 int expand_stack(struct vm_area_struct *vma, unsigned long address)
 {
 	int error;
-	unsigned long size, grow;
 
 	/*
 	 * We must make sure the anon_vma is allocated
@@ -1544,13 +1549,20 @@ int expand_stack(struct vm_area_struct *vma, unsigned long address)
 	 * anon_vma lock to serialize against concurrent expand_stacks.
 	 */
 	address &= PAGE_MASK;
-	size = vma->vm_end - address;
-	grow = (vma->vm_start - address) >> PAGE_SHIFT;
+	error = 0;
+
+	/* Somebody else might have raced and expanded it already */
+	if (address < vma->vm_start) {
+		unsigned long size, grow;
 
-	error = acct_stack_growth(vma, size, grow);
-	if (!error) {
-		vma->vm_start = address;
-		vma->vm_pgoff -= grow;
+		size = vma->vm_end - address;
+		grow = (vma->vm_start - address) >> PAGE_SHIFT;
+
+		error = acct_stack_growth(vma, size, grow);
+		if (!error) {
+			vma->vm_start = address;
+			vma->vm_pgoff -= grow;
+		}
 	}
 	anon_vma_unlock(vma);
 	return error;
@@ -1858,6 +1870,16 @@ asmlinkage long sys_munmap(unsigned long addr, size_t len)
 	return ret;
 }
 
+static inline void verify_mm_writelocked(struct mm_struct *mm)
+{
+#ifdef CONFIG_DEBUG_KERNEL
+	if (unlikely(down_read_trylock(&mm->mmap_sem))) {
+		WARN_ON(1);
+		up_read(&mm->mmap_sem);
+	}
+#endif
+}
+
 /*
  *  this is really a simplified "do_mmap".  it only handles
  *  anonymous maps.  eventually we may be able to do some
@@ -1890,6 +1912,12 @@ unsigned long do_brk(unsigned long addr, unsigned long len)
 			return -EAGAIN;
 	}
 
+	/*
+	 * mm->mmap_sem is required to protect against another thread
+	 * changing the mappings in case we sleep.
+	 */
+	verify_mm_writelocked(mm);
+
 	/*
 	 * Clear old maps.  this also does some error checking for us
 	 */