Merge branch 'for-linus' of git://git390.osdl.marist.edu/pub/scm/linux-2.6

* 'for-linus' of git://git390.osdl.marist.edu/pub/scm/linux-2.6:
  [S390] Poison init section before freeing it.
  [S390] Use add_active_range() and free_area_init_nodes().
  [S390] Virtual memmap for s390.
  [S390] Update documentation for dynamic subchannel mapping.
  [S390] Use dev->groups for adding/removing the subchannel attribute group.
  [S390] Support for disconnected devices reappearing on another subchannel.
  [S390] subchannel lock conversion.
  [S390] Some preparations for the dynamic subchannel mapping patch.
  [S390] runtime switch for qdio performance statistics
  [S390] New DASD feature for ERP related logging
  [S390] add reset call handler to the ap bus.
  [S390] more workqueue fixes.
  [S390] workqueue fixes.
  [S390] uaccess_pt: add missing down_read() and convert to is_init().

Documentation/s390/driver-model.txt

@@ -18,11 +18,18 @@ devices/
 	   - 0.0.0002/
 	   - 0.1.0000/0.1.1234/
 	   ...
+	   - defunct/
 
 In this example, device 0815 is accessed via subchannel 0 in subchannel set 0,
 device 4711 via subchannel 1 in subchannel set 0, and subchannel 2 is a non-I/O
 subchannel. Device 1234 is accessed via subchannel 0 in subchannel set 1.
 
+The subchannel named 'defunct' does not represent any real subchannel on the
+system; it is a pseudo subchannel where disconnnected ccw devices are moved to
+if they are displaced by another ccw device becoming operational on their
+former subchannel. The ccw devices will be moved again to a proper subchannel
+if they become operational again on that subchannel.
+
 You should address a ccw device via its bus id (e.g. 0.0.4711); the device can
 be found under bus/ccw/devices/.
 

arch/s390/Kconfig

@@ -241,8 +241,14 @@ config WARN_STACK_SIZE
 	  This allows you to specify the maximum frame size a function may
 	  have without the compiler complaining about it.
 
+config ARCH_POPULATES_NODE_MAP
+	def_bool y
+
 source "mm/Kconfig"
 
+config HOLES_IN_ZONE
+	def_bool y
+
 comment "I/O subsystem configuration"
 
 config MACHCHK_WARNING
@@ -266,14 +272,6 @@ config QDIO
 
 	  If unsure, say Y.
 
-config QDIO_PERF_STATS
-	bool "Performance statistics in /proc"
-	depends on QDIO
-	help
-	  Say Y here to get performance statistics in /proc/qdio_perf
-
-	  If unsure, say N.
-
 config QDIO_DEBUG
 	bool "Extended debugging information"
 	depends on QDIO

arch/s390/defconfig

@@ -134,7 +134,6 @@ CONFIG_RESOURCES_64BIT=y
 #
 CONFIG_MACHCHK_WARNING=y
 CONFIG_QDIO=y
-# CONFIG_QDIO_PERF_STATS is not set
 # CONFIG_QDIO_DEBUG is not set
 
 #

arch/s390/kernel/setup.c

@@ -64,9 +64,8 @@ unsigned int console_devno = -1;
 unsigned int console_irq = -1;
 unsigned long machine_flags = 0;
 
-struct mem_chunk memory_chunk[MEMORY_CHUNKS];
+struct mem_chunk __initdata memory_chunk[MEMORY_CHUNKS];
 volatile int __cpu_logical_map[NR_CPUS]; /* logical cpu to cpu address */
-unsigned long __initdata zholes_size[MAX_NR_ZONES];
 static unsigned long __initdata memory_end;
 
 /*
@@ -354,21 +353,6 @@ void machine_power_off(void)
  */
 void (*pm_power_off)(void) = machine_power_off;
 
-static void __init
-add_memory_hole(unsigned long start, unsigned long end)
-{
-	unsigned long dma_pfn = MAX_DMA_ADDRESS >> PAGE_SHIFT;
-
-	if (end <= dma_pfn)
-		zholes_size[ZONE_DMA] += end - start + 1;
-	else if (start > dma_pfn)
-		zholes_size[ZONE_NORMAL] += end - start + 1;
-	else {
-		zholes_size[ZONE_DMA] += dma_pfn - start + 1;
-		zholes_size[ZONE_NORMAL] += end - dma_pfn;
-	}
-}
-
 static int __init early_parse_mem(char *p)
 {
 	memory_end = memparse(p, &p);
@@ -521,7 +505,6 @@ setup_memory(void)
 {
         unsigned long bootmap_size;
 	unsigned long start_pfn, end_pfn, init_pfn;
-	unsigned long last_rw_end;
 	int i;
 
 	/*
@@ -577,39 +560,27 @@ setup_memory(void)
 	/*
 	 * Register RAM areas with the bootmem allocator.
 	 */
-	last_rw_end = start_pfn;
 
 	for (i = 0; i < MEMORY_CHUNKS && memory_chunk[i].size > 0; i++) {
-		unsigned long start_chunk, end_chunk;
+		unsigned long start_chunk, end_chunk, pfn;
 
 		if (memory_chunk[i].type != CHUNK_READ_WRITE)
 			continue;
-		start_chunk = (memory_chunk[i].addr + PAGE_SIZE - 1);
-		start_chunk >>= PAGE_SHIFT;
-		end_chunk = (memory_chunk[i].addr + memory_chunk[i].size);
-		end_chunk >>= PAGE_SHIFT;
-		if (start_chunk < start_pfn)
-			start_chunk = start_pfn;
-		if (end_chunk > end_pfn)
-			end_chunk = end_pfn;
-		if (start_chunk < end_chunk) {
-			/* Initialize storage key for RAM pages */
-			for (init_pfn = start_chunk ; init_pfn < end_chunk;
-			     init_pfn++)
-				page_set_storage_key(init_pfn << PAGE_SHIFT,
-						     PAGE_DEFAULT_KEY);
-			free_bootmem(start_chunk << PAGE_SHIFT,
-				     (end_chunk - start_chunk) << PAGE_SHIFT);
-			if (last_rw_end < start_chunk)
-				add_memory_hole(last_rw_end, start_chunk - 1);
-			last_rw_end = end_chunk;
-		}
+		start_chunk = PFN_DOWN(memory_chunk[i].addr);
+		end_chunk = start_chunk + PFN_DOWN(memory_chunk[i].size) - 1;
+		end_chunk = min(end_chunk, end_pfn);
+		if (start_chunk >= end_chunk)
+			continue;
+		add_active_range(0, start_chunk, end_chunk);
+		pfn = max(start_chunk, start_pfn);
+		for (; pfn <= end_chunk; pfn++)
+			page_set_storage_key(PFN_PHYS(pfn), PAGE_DEFAULT_KEY);
 	}
 
 	psw_set_key(PAGE_DEFAULT_KEY);
 
-	if (last_rw_end < end_pfn - 1)
-		add_memory_hole(last_rw_end, end_pfn - 1);
+	free_bootmem_with_active_regions(0, max_pfn);
+	reserve_bootmem(0, PFN_PHYS(start_pfn));
 
 	/*
 	 * Reserve the bootmem bitmap itself as well. We do this in two

arch/s390/lib/uaccess_pt.c

@@ -8,8 +8,8 @@
  */
 
 #include <linux/errno.h>
-#include <asm/uaccess.h>
 #include <linux/mm.h>
+#include <asm/uaccess.h>
 #include <asm/futex.h>
 
 static inline int __handle_fault(struct mm_struct *mm, unsigned long address,
@@ -60,8 +60,9 @@ out:
 
 out_of_memory:
 	up_read(&mm->mmap_sem);
-	if (current->pid == 1) {
+	if (is_init(current)) {
 		yield();
+		down_read(&mm->mmap_sem);
 		goto survive;
 	}
 	printk("VM: killing process %s\n", current->comm);

arch/s390/mm/Makefile

@@ -2,6 +2,6 @@
 # Makefile for the linux s390-specific parts of the memory manager.
 #
 
-obj-y	 := init.o fault.o ioremap.o extmem.o mmap.o
+obj-y	 := init.o fault.o ioremap.o extmem.o mmap.o vmem.o
 obj-$(CONFIG_CMM) += cmm.o
 

arch/s390/mm/extmem.c

@@ -16,6 +16,7 @@
 #include <linux/bootmem.h>
 #include <linux/ctype.h>
 #include <asm/page.h>
+#include <asm/pgtable.h>
 #include <asm/ebcdic.h>
 #include <asm/errno.h>
 #include <asm/extmem.h>
@@ -237,65 +238,6 @@ query_segment_type (struct dcss_segment *seg)
 	return rc;
 }
 
-/*
- * check if the given segment collides with guest storage.
- * returns 1 if this is the case, 0 if no collision was found
- */
-static int
-segment_overlaps_storage(struct dcss_segment *seg)
-{
-	int i;
-
-	for (i = 0; i < MEMORY_CHUNKS && memory_chunk[i].size > 0; i++) {
-		if (memory_chunk[i].type != CHUNK_READ_WRITE)
-			continue;
-		if ((memory_chunk[i].addr >> 20) > (seg->end >> 20))
-			continue;
-		if (((memory_chunk[i].addr + memory_chunk[i].size - 1) >> 20)
-				< (seg->start_addr >> 20))
-			continue;
-		return 1;
-	}
-	return 0;
-}
-
-/*
- * check if segment collides with other segments that are currently loaded
- * returns 1 if this is the case, 0 if no collision was found
- */
-static int
-segment_overlaps_others (struct dcss_segment *seg)
-{
-	struct list_head *l;
-	struct dcss_segment *tmp;
-
-	BUG_ON(!mutex_is_locked(&dcss_lock));
-	list_for_each(l, &dcss_list) {
-		tmp = list_entry(l, struct dcss_segment, list);
-		if ((tmp->start_addr >> 20) > (seg->end >> 20))
-			continue;
-		if ((tmp->end >> 20) < (seg->start_addr >> 20))
-			continue;
-		if (seg == tmp)
-			continue;
-		return 1;
-	}
-	return 0;
-}
-
-/*
- * check if segment exceeds the kernel mapping range (detected or set via mem=)
- * returns 1 if this is the case, 0 if segment fits into the range
- */
-static inline int
-segment_exceeds_range (struct dcss_segment *seg)
-{
-	int seg_last_pfn = (seg->end) >> PAGE_SHIFT;
-	if (seg_last_pfn > max_pfn)
-		return 1;
-	return 0;
-}
-
 /*
  * get info about a segment
  * possible return values:
@@ -341,24 +283,26 @@ __segment_load (char *name, int do_nonshared, unsigned long *addr, unsigned long
 	rc = query_segment_type (seg);
 	if (rc < 0)
 		goto out_free;
-	if (segment_exceeds_range(seg)) {
-		PRINT_WARN ("segment_load: not loading segment %s - exceeds"
-				" kernel mapping range\n",name);
-		rc = -ERANGE;
+
+	rc = add_shared_memory(seg->start_addr, seg->end - seg->start_addr + 1);
+
+	switch (rc) {
+	case 0:
+		break;
+	case -ENOSPC:
+		PRINT_WARN("segment_load: not loading segment %s - overlaps "
+			   "storage/segment\n", name);
 		goto out_free;
-	}
-	if (segment_overlaps_storage(seg)) {
-		PRINT_WARN ("segment_load: not loading segment %s - overlaps"
-				" storage\n",name);
-		rc = -ENOSPC;
+	case -ERANGE:
+		PRINT_WARN("segment_load: not loading segment %s - exceeds "
+			   "kernel mapping range\n", name);
 		goto out_free;
-	}
-	if (segment_overlaps_others(seg)) {
-		PRINT_WARN ("segment_load: not loading segment %s - overlaps"
-				" other segments\n",name);
-		rc = -EBUSY;
+	default:
+		PRINT_WARN("segment_load: not loading segment %s (rc: %d)\n",
+			   name, rc);
 		goto out_free;
 	}
+
 	if (do_nonshared)
 		dcss_command = DCSS_LOADNSR;
 	else
@@ -372,7 +316,7 @@ __segment_load (char *name, int do_nonshared, unsigned long *addr, unsigned long
 		rc = dcss_diag_translate_rc (seg->end);
 		dcss_diag(DCSS_PURGESEG, seg->dcss_name,
 				&seg->start_addr, &seg->end);
-		goto out_free;
+		goto out_shared;
 	}
 	seg->do_nonshared = do_nonshared;
 	atomic_set(&seg->ref_count, 1);
@@ -391,6 +335,8 @@ __segment_load (char *name, int do_nonshared, unsigned long *addr, unsigned long
 				(void*)seg->start_addr, (void*)seg->end,
 				segtype_string[seg->vm_segtype]);
 	goto out;
+ out_shared:
+	remove_shared_memory(seg->start_addr, seg->end - seg->start_addr + 1);
  out_free:
 	kfree(seg);
  out:
@@ -530,12 +476,12 @@ segment_unload(char *name)
 				"please report to linux390@de.ibm.com\n",name);
 		goto out_unlock;
 	}
-	if (atomic_dec_return(&seg->ref_count) == 0) {
+	if (atomic_dec_return(&seg->ref_count) != 0)
+		goto out_unlock;
+	remove_shared_memory(seg->start_addr, seg->end - seg->start_addr + 1);
 	list_del(&seg->list);
-		dcss_diag(DCSS_PURGESEG, seg->dcss_name,
-			  &dummy, &dummy);
+	dcss_diag(DCSS_PURGESEG, seg->dcss_name, &dummy, &dummy);
 	kfree(seg);
-	}
 out_unlock:
 	mutex_unlock(&dcss_lock);
 }

arch/s390/mm/init.c

@@ -24,6 +24,7 @@
 #include <linux/pagemap.h>
 #include <linux/bootmem.h>
 #include <linux/pfn.h>
+#include <linux/poison.h>
 
 #include <asm/processor.h>
 #include <asm/system.h>
@@ -69,6 +70,8 @@ void show_mem(void)
         printk("Free swap:       %6ldkB\n", nr_swap_pages<<(PAGE_SHIFT-10));
         i = max_mapnr;
         while (i-- > 0) {
+		if (!pfn_valid(i))
+			continue;
 		page = pfn_to_page(i);
                 total++;
 		if (PageReserved(page))
@@ -84,150 +87,52 @@ void show_mem(void)
         printk("%d pages swap cached\n",cached);
 }
 
-extern unsigned long __initdata zholes_size[];
-/*
- * paging_init() sets up the page tables
- */
-
-#ifndef CONFIG_64BIT
-void __init paging_init(void)
+static void __init setup_ro_region(void)
 {
-        pgd_t * pg_dir;
-        pte_t * pg_table;
-        pte_t   pte;
-	int     i;
-        unsigned long tmp;
-        unsigned long pfn = 0;
-        unsigned long pgdir_k = (__pa(swapper_pg_dir) & PAGE_MASK) | _KERNSEG_TABLE;
-        static const int ssm_mask = 0x04000000L;
-	unsigned long ro_start_pfn, ro_end_pfn;
-	unsigned long zones_size[MAX_NR_ZONES];
-
-	ro_start_pfn = PFN_DOWN((unsigned long)&__start_rodata);
-	ro_end_pfn = PFN_UP((unsigned long)&__end_rodata);
-
-	memset(zones_size, 0, sizeof(zones_size));
-	zones_size[ZONE_DMA] = max_low_pfn;
-	free_area_init_node(0, &contig_page_data, zones_size,
-			    __pa(PAGE_OFFSET) >> PAGE_SHIFT,
-			    zholes_size);
-
-	/* unmap whole virtual address space */
-	
-        pg_dir = swapper_pg_dir;
-
-	for (i = 0; i < PTRS_PER_PGD; i++)
-		pmd_clear((pmd_t *) pg_dir++);
-		
-	/*
-	 * map whole physical memory to virtual memory (identity mapping) 
-	 */
-
-        pg_dir = swapper_pg_dir;
-
-        while (pfn < max_low_pfn) {
-                /*
-                 * pg_table is physical at this point
-                 */
-		pg_table = (pte_t *) alloc_bootmem_pages(PAGE_SIZE);
-
-		pmd_populate_kernel(&init_mm, (pmd_t *) pg_dir, pg_table);
-                pg_dir++;
-
-                for (tmp = 0 ; tmp < PTRS_PER_PTE ; tmp++,pg_table++) {
-			if (pfn >= ro_start_pfn && pfn < ro_end_pfn)
-				pte = pfn_pte(pfn, __pgprot(_PAGE_RO));
-			else
-				pte = pfn_pte(pfn, PAGE_KERNEL);
-                        if (pfn >= max_low_pfn)
-				pte_val(pte) = _PAGE_TYPE_EMPTY;
-			set_pte(pg_table, pte);
-                        pfn++;
+	pgd_t *pgd;
+	pmd_t *pmd;
+	pte_t *pte;
+	pte_t new_pte;
+	unsigned long address, end;
+
+	address = ((unsigned long)&__start_rodata) & PAGE_MASK;
+	end = PFN_ALIGN((unsigned long)&__end_rodata);
+
+	for (; address < end; address += PAGE_SIZE) {
+		pgd = pgd_offset_k(address);
+		pmd = pmd_offset(pgd, address);
+		pte = pte_offset_kernel(pmd, address);
+		new_pte = mk_pte_phys(address, __pgprot(_PAGE_RO));
+		set_pte(pte, new_pte);
 	}
-        }
-
-	S390_lowcore.kernel_asce = pgdir_k;
-
-        /* enable virtual mapping in kernel mode */
-	__ctl_load(pgdir_k, 1, 1);
-	__ctl_load(pgdir_k, 7, 7);
-	__ctl_load(pgdir_k, 13, 13);
-	__raw_local_irq_ssm(ssm_mask);
-
-        local_flush_tlb();
 }
 
-#else /* CONFIG_64BIT */
+extern void vmem_map_init(void);
 
+/*
+ * paging_init() sets up the page tables
+ */
 void __init paging_init(void)
 {
-        pgd_t * pg_dir;
-	pmd_t * pm_dir;
-        pte_t * pt_dir;
-        pte_t   pte;
-	int     i,j,k;
-        unsigned long pfn = 0;
-        unsigned long pgdir_k = (__pa(swapper_pg_dir) & PAGE_MASK) |
-          _KERN_REGION_TABLE;
+	pgd_t *pg_dir;
+	int i;
+	unsigned long pgdir_k;
 	static const int ssm_mask = 0x04000000L;
-	unsigned long zones_size[MAX_NR_ZONES];
-	unsigned long dma_pfn, high_pfn;
-	unsigned long ro_start_pfn, ro_end_pfn;
-
-	memset(zones_size, 0, sizeof(zones_size));
-	dma_pfn = MAX_DMA_ADDRESS >> PAGE_SHIFT;
-	high_pfn = max_low_pfn;
-	ro_start_pfn = PFN_DOWN((unsigned long)&__start_rodata);
-	ro_end_pfn = PFN_UP((unsigned long)&__end_rodata);
-
-	if (dma_pfn > high_pfn)
-		zones_size[ZONE_DMA] = high_pfn;
-	else {
-		zones_size[ZONE_DMA] = dma_pfn;
-		zones_size[ZONE_NORMAL] = high_pfn - dma_pfn;
-	}
-
-	/* Initialize mem_map[].  */
-	free_area_init_node(0, &contig_page_data, zones_size,
-			    __pa(PAGE_OFFSET) >> PAGE_SHIFT, zholes_size);
-
-	/*
-	 * map whole physical memory to virtual memory (identity mapping) 
-	 */
+	unsigned long max_zone_pfns[MAX_NR_ZONES];
 
 	pg_dir = swapper_pg_dir;
 	
-        for (i = 0 ; i < PTRS_PER_PGD ; i++,pg_dir++) {
-	
-                if (pfn >= max_low_pfn) {
-                        pgd_clear(pg_dir);
-                        continue;
-                }          
-        
-		pm_dir = (pmd_t *) alloc_bootmem_pages(PAGE_SIZE * 4);
-                pgd_populate(&init_mm, pg_dir, pm_dir);
-
-                for (j = 0 ; j < PTRS_PER_PMD ; j++,pm_dir++) {
-                        if (pfn >= max_low_pfn) {
-                                pmd_clear(pm_dir);
-                                continue; 
-                        }          
-                        
-			pt_dir = (pte_t *) alloc_bootmem_pages(PAGE_SIZE);
-                        pmd_populate_kernel(&init_mm, pm_dir, pt_dir);
-	
-                        for (k = 0 ; k < PTRS_PER_PTE ; k++,pt_dir++) {
-				if (pfn >= ro_start_pfn && pfn < ro_end_pfn)
-					pte = pfn_pte(pfn, __pgprot(_PAGE_RO));
-				else
-					pte = pfn_pte(pfn, PAGE_KERNEL);
-				if (pfn >= max_low_pfn)
-					pte_val(pte) = _PAGE_TYPE_EMPTY;
-                                set_pte(pt_dir, pte);
-                                pfn++;
-                        }
-                }
-        }
+#ifdef CONFIG_64BIT
+	pgdir_k = (__pa(swapper_pg_dir) & PAGE_MASK) | _KERN_REGION_TABLE;
+	for (i = 0; i < PTRS_PER_PGD; i++)
+		pgd_clear(pg_dir + i);
+#else
+	pgdir_k = (__pa(swapper_pg_dir) & PAGE_MASK) | _KERNSEG_TABLE;
+	for (i = 0; i < PTRS_PER_PGD; i++)
+		pmd_clear((pmd_t *)(pg_dir + i));
+#endif
+	vmem_map_init();
+	setup_ro_region();
 
 	S390_lowcore.kernel_asce = pgdir_k;
 
@@ -237,9 +142,11 @@ void __init paging_init(void)
 	__ctl_load(pgdir_k, 13, 13);
 	__raw_local_irq_ssm(ssm_mask);
 
-        local_flush_tlb();
+	memset(max_zone_pfns, 0, sizeof(max_zone_pfns));
+	max_zone_pfns[ZONE_DMA] = PFN_DOWN(MAX_DMA_ADDRESS);
+	max_zone_pfns[ZONE_NORMAL] = max_low_pfn;
+	free_area_init_nodes(max_zone_pfns);
 }
-#endif /* CONFIG_64BIT */
 
 void __init mem_init(void)
 {
@@ -269,6 +176,8 @@ void __init mem_init(void)
 	printk("Write protected kernel read-only data: %#lx - %#lx\n",
 	       (unsigned long)&__start_rodata,
 	       PFN_ALIGN((unsigned long)&__end_rodata) - 1);
+	printk("Virtual memmap size: %ldk\n",
+	       (max_pfn * sizeof(struct page)) >> 10);
 }
 
 void free_initmem(void)
@@ -279,6 +188,7 @@ void free_initmem(void)
         for (; addr < (unsigned long)(&__init_end); addr += PAGE_SIZE) {
 		ClearPageReserved(virt_to_page(addr));
 		init_page_count(virt_to_page(addr));
+		memset((void *)addr, POISON_FREE_INITMEM, PAGE_SIZE);
 		free_page(addr);
 		totalram_pages++;
         }

arch/s390/mm/vmem.c

+/*
+ *  arch/s390/mm/vmem.c
+ *
+ *    Copyright IBM Corp. 2006
+ *    Author(s): Heiko Carstens <heiko.carstens@de.ibm.com>
+ */
+
+#include <linux/bootmem.h>
+#include <linux/pfn.h>
+#include <linux/mm.h>
+#include <linux/module.h>
+#include <linux/list.h>
+#include <asm/pgalloc.h>
+#include <asm/pgtable.h>
+#include <asm/setup.h>
+#include <asm/tlbflush.h>
+
+unsigned long vmalloc_end;
+EXPORT_SYMBOL(vmalloc_end);
+
+static struct page *vmem_map;
+static DEFINE_MUTEX(vmem_mutex);
+
+struct memory_segment {
+	struct list_head list;
+	unsigned long start;
+	unsigned long size;
+};
+
+static LIST_HEAD(mem_segs);
+
+void memmap_init(unsigned long size, int nid, unsigned long zone,
+		 unsigned long start_pfn)
+{
+	struct page *start, *end;
+	struct page *map_start, *map_end;
+	int i;
+
+	start = pfn_to_page(start_pfn);
+	end = start + size;
+
+	for (i = 0; i < MEMORY_CHUNKS && memory_chunk[i].size > 0; i++) {
+		unsigned long cstart, cend;
+
+		cstart = PFN_DOWN(memory_chunk[i].addr);
+		cend = cstart + PFN_DOWN(memory_chunk[i].size);
+
+		map_start = mem_map + cstart;
+		map_end = mem_map + cend;
+
+		if (map_start < start)
+			map_start = start;
+		if (map_end > end)
+			map_end = end;
+
+		map_start -= ((unsigned long) map_start & (PAGE_SIZE - 1))
+			/ sizeof(struct page);
+		map_end += ((PFN_ALIGN((unsigned long) map_end)
+			     - (unsigned long) map_end)
+			    / sizeof(struct page));
+
+		if (map_start < map_end)
+			memmap_init_zone((unsigned long)(map_end - map_start),
+					 nid, zone, page_to_pfn(map_start));
+	}
+}
+
+static inline void *vmem_alloc_pages(unsigned int order)
+{
+	if (slab_is_available())
+		return (void *)__get_free_pages(GFP_KERNEL, order);
+	return alloc_bootmem_pages((1 << order) * PAGE_SIZE);
+}
+
+static inline pmd_t *vmem_pmd_alloc(void)
+{
+	pmd_t *pmd;
+	int i;
+
+	pmd = vmem_alloc_pages(PMD_ALLOC_ORDER);
+	if (!pmd)
+		return NULL;
+	for (i = 0; i < PTRS_PER_PMD; i++)
+		pmd_clear(pmd + i);
+	return pmd;
+}
+
+static inline pte_t *vmem_pte_alloc(void)
+{
+	pte_t *pte;
+	pte_t empty_pte;
+	int i;
+
+	pte = vmem_alloc_pages(PTE_ALLOC_ORDER);
+	if (!pte)
+		return NULL;
+	pte_val(empty_pte) = _PAGE_TYPE_EMPTY;
+	for (i = 0; i < PTRS_PER_PTE; i++)
+		set_pte(pte + i, empty_pte);
+	return pte;
+}
+
+/*
+ * Add a physical memory range to the 1:1 mapping.
+ */
+static int vmem_add_range(unsigned long start, unsigned long size)
+{
+	unsigned long address;
+	pgd_t *pg_dir;
+	pmd_t *pm_dir;
+	pte_t *pt_dir;
+	pte_t  pte;
+	int ret = -ENOMEM;
+
+	for (address = start; address < start + size; address += PAGE_SIZE) {
+		pg_dir = pgd_offset_k(address);
+		if (pgd_none(*pg_dir)) {
+			pm_dir = vmem_pmd_alloc();
+			if (!pm_dir)
+				goto out;
+			pgd_populate(&init_mm, pg_dir, pm_dir);
+		}
+
+		pm_dir = pmd_offset(pg_dir, address);
+		if (pmd_none(*pm_dir)) {
+			pt_dir = vmem_pte_alloc();
+			if (!pt_dir)
+				goto out;
+			pmd_populate_kernel(&init_mm, pm_dir, pt_dir);
+		}
+
+		pt_dir = pte_offset_kernel(pm_dir, address);
+		pte = pfn_pte(address >> PAGE_SHIFT, PAGE_KERNEL);
+		set_pte(pt_dir, pte);
+	}
+	ret = 0;
+out:
+	flush_tlb_kernel_range(start, start + size);
+	return ret;
+}
+
+/*
+ * Remove a physical memory range from the 1:1 mapping.
+ * Currently only invalidates page table entries.
+ */
+static void vmem_remove_range(unsigned long start, unsigned long size)
+{
+	unsigned long address;
+	pgd_t *pg_dir;
+	pmd_t *pm_dir;
+	pte_t *pt_dir;
+	pte_t  pte;
+
+	pte_val(pte) = _PAGE_TYPE_EMPTY;
+	for (address = start; address < start + size; address += PAGE_SIZE) {
+		pg_dir = pgd_offset_k(address);
+		if (pgd_none(*pg_dir))
+			continue;
+		pm_dir = pmd_offset(pg_dir, address);
+		if (pmd_none(*pm_dir))
+			continue;
+		pt_dir = pte_offset_kernel(pm_dir, address);
+		set_pte(pt_dir, pte);
+	}
+	flush_tlb_kernel_range(start, start + size);
+}
+
+/*
+ * Add a backed mem_map array to the virtual mem_map array.
+ */
+static int vmem_add_mem_map(unsigned long start, unsigned long size)
+{
+	unsigned long address, start_addr, end_addr;
+	struct page *map_start, *map_end;
+	pgd_t *pg_dir;
+	pmd_t *pm_dir;
+	pte_t *pt_dir;
+	pte_t  pte;
+	int ret = -ENOMEM;
+
+	map_start = vmem_map + PFN_DOWN(start);
+	map_end	= vmem_map + PFN_DOWN(start + size);
+
+	start_addr = (unsigned long) map_start & PAGE_MASK;
+	end_addr = PFN_ALIGN((unsigned long) map_end);
+
+	for (address = start_addr; address < end_addr; address += PAGE_SIZE) {
+		pg_dir = pgd_offset_k(address);
+		if (pgd_none(*pg_dir)) {
+			pm_dir = vmem_pmd_alloc();
+			if (!pm_dir)
+				goto out;
+			pgd_populate(&init_mm, pg_dir, pm_dir);
+		}
+
+		pm_dir = pmd_offset(pg_dir, address);
+		if (pmd_none(*pm_dir)) {
+			pt_dir = vmem_pte_alloc();
+			if (!pt_dir)
+				goto out;
+			pmd_populate_kernel(&init_mm, pm_dir, pt_dir);
+		}
+
+		pt_dir = pte_offset_kernel(pm_dir, address);
+		if (pte_none(*pt_dir)) {
+			unsigned long new_page;
+
+			new_page =__pa(vmem_alloc_pages(0));
+			if (!new_page)
+				goto out;
+			pte = pfn_pte(new_page >> PAGE_SHIFT, PAGE_KERNEL);
+			set_pte(pt_dir, pte);
+		}
+	}
+	ret = 0;
+out:
+	flush_tlb_kernel_range(start_addr, end_addr);
+	return ret;
+}
+
+static int vmem_add_mem(unsigned long start, unsigned long size)
+{
+	int ret;
+
+	ret = vmem_add_range(start, size);
+	if (ret)
+		return ret;
+	return vmem_add_mem_map(start, size);
+}
+
+/*
+ * Add memory segment to the segment list if it doesn't overlap with
+ * an already present segment.
+ */
+static int insert_memory_segment(struct memory_segment *seg)
+{
+	struct memory_segment *tmp;
+
+	if (PFN_DOWN(seg->start + seg->size) > max_pfn ||
+	    seg->start + seg->size < seg->start)
+		return -ERANGE;
+
+	list_for_each_entry(tmp, &mem_segs, list) {
+		if (seg->start >= tmp->start + tmp->size)
+			continue;
+		if (seg->start + seg->size <= tmp->start)
+			continue;
+		return -ENOSPC;
+	}
+	list_add(&seg->list, &mem_segs);
+	return 0;
+}
+
+/*
+ * Remove memory segment from the segment list.
+ */
+static void remove_memory_segment(struct memory_segment *seg)
+{
+	list_del(&seg->list);
+}
+
+static void __remove_shared_memory(struct memory_segment *seg)
+{
+	remove_memory_segment(seg);
+	vmem_remove_range(seg->start, seg->size);
+}
+
+int remove_shared_memory(unsigned long start, unsigned long size)
+{
+	struct memory_segment *seg;
+	int ret;
+
+	mutex_lock(&vmem_mutex);
+
+	ret = -ENOENT;
+	list_for_each_entry(seg, &mem_segs, list) {
+		if (seg->start == start && seg->size == size)
+			break;
+	}
+
+	if (seg->start != start || seg->size != size)
+		goto out;
+
+	ret = 0;
+	__remove_shared_memory(seg);
+	kfree(seg);
+out:
+	mutex_unlock(&vmem_mutex);
+	return ret;
+}
+
+int add_shared_memory(unsigned long start, unsigned long size)
+{
+	struct memory_segment *seg;
+	struct page *page;
+	unsigned long pfn, num_pfn, end_pfn;
+	int ret;
+
+	mutex_lock(&vmem_mutex);
+	ret = -ENOMEM;
+	seg = kzalloc(sizeof(*seg), GFP_KERNEL);
+	if (!seg)
+		goto out;
+	seg->start = start;
+	seg->size = size;
+
+	ret = insert_memory_segment(seg);
+	if (ret)
+		goto out_free;
+
+	ret = vmem_add_mem(start, size);
+	if (ret)
+		goto out_remove;
+
+	pfn = PFN_DOWN(start);
+	num_pfn = PFN_DOWN(size);
+	end_pfn = pfn + num_pfn;
+
+	page = pfn_to_page(pfn);
+	memset(page, 0, num_pfn * sizeof(struct page));
+
+	for (; pfn < end_pfn; pfn++) {
+		page = pfn_to_page(pfn);
+		init_page_count(page);
+		reset_page_mapcount(page);
+		SetPageReserved(page);
+		INIT_LIST_HEAD(&page->lru);
+	}
+	goto out;
+
+out_remove:
+	__remove_shared_memory(seg);
+out_free:
+	kfree(seg);
+out:
+	mutex_unlock(&vmem_mutex);
+	return ret;
+}
+
+/*
+ * map whole physical memory to virtual memory (identity mapping)
+ */
+void __init vmem_map_init(void)
+{
+	unsigned long map_size;
+	int i;
+
+	map_size = ALIGN(max_low_pfn, MAX_ORDER_NR_PAGES) * sizeof(struct page);
+	vmalloc_end = PFN_ALIGN(VMALLOC_END_INIT) - PFN_ALIGN(map_size);
+	vmem_map = (struct page *) vmalloc_end;
+	NODE_DATA(0)->node_mem_map = vmem_map;
+
+	for (i = 0; i < MEMORY_CHUNKS && memory_chunk[i].size > 0; i++)
+		vmem_add_mem(memory_chunk[i].addr, memory_chunk[i].size);
+}
+
+/*
+ * Convert memory chunk array to a memory segment list so there is a single
+ * list that contains both r/w memory and shared memory segments.
+ */
+static int __init vmem_convert_memory_chunk(void)
+{
+	struct memory_segment *seg;
+	int i;
+
+	mutex_lock(&vmem_mutex);
+	for (i = 0; i < MEMORY_CHUNKS && memory_chunk[i].size > 0; i++) {
+		if (!memory_chunk[i].size)
+			continue;
+		seg = kzalloc(sizeof(*seg), GFP_KERNEL);
+		if (!seg)
+			panic("Out of memory...\n");
+		seg->start = memory_chunk[i].addr;
+		seg->size = memory_chunk[i].size;
+		insert_memory_segment(seg);
+	}
+	mutex_unlock(&vmem_mutex);
+	return 0;
+}
+
+core_initcall(vmem_convert_memory_chunk);

drivers/s390/block/dasd.c

@@ -1050,10 +1050,10 @@ dasd_int_handler(struct ccw_device *cdev, unsigned long intparm,
 		}
 	} else {		/* error */
 		memcpy(&cqr->irb, irb, sizeof (struct irb));
-#ifdef ERP_DEBUG
+		if (device->features & DASD_FEATURE_ERPLOG) {
 			/* dump sense data */
 			dasd_log_sense(cqr, irb);
-#endif
+		}
 		switch (era) {
 		case dasd_era_fatal:
 			cqr->status = DASD_CQR_FAILED;

drivers/s390/block/dasd_3990_erp.c

@@ -2641,14 +2641,12 @@ dasd_3990_erp_action(struct dasd_ccw_req * cqr)
 	struct dasd_ccw_req *erp = NULL;
 	struct dasd_device *device = cqr->device;
 	__u32 cpa = cqr->irb.scsw.cpa;
+	struct dasd_ccw_req *temp_erp = NULL;
 
-#ifdef ERP_DEBUG
+	if (device->features & DASD_FEATURE_ERPLOG) {
 		/* print current erp_chain */
 		DEV_MESSAGE(KERN_ERR, device, "%s",
 			    "ERP chain at BEGINNING of ERP-ACTION");
-	{
-		struct dasd_ccw_req *temp_erp = NULL;
-
 		for (temp_erp = cqr;
 		     temp_erp != NULL; temp_erp = temp_erp->refers) {
 
@@ -2658,7 +2656,6 @@ dasd_3990_erp_action(struct dasd_ccw_req * cqr)
 				    temp_erp->refers);
 		}
 	}
-#endif				/* ERP_DEBUG */
 
 	/* double-check if current erp/cqr was successfull */
 	if ((cqr->irb.scsw.cstat == 0x00) &&
@@ -2695,11 +2692,10 @@ dasd_3990_erp_action(struct dasd_ccw_req * cqr)
 		erp = dasd_3990_erp_handle_match_erp(cqr, erp);
 	}
 
-#ifdef ERP_DEBUG
+	if (device->features & DASD_FEATURE_ERPLOG) {
 		/* print current erp_chain */
-	DEV_MESSAGE(KERN_ERR, device, "%s", "ERP chain at END of ERP-ACTION");
-	{
-		struct dasd_ccw_req *temp_erp = NULL;
+		DEV_MESSAGE(KERN_ERR, device, "%s",
+			    "ERP chain at END of ERP-ACTION");
 		for (temp_erp = erp;
 		     temp_erp != NULL; temp_erp = temp_erp->refers) {
 
@@ -2709,7 +2705,6 @@ dasd_3990_erp_action(struct dasd_ccw_req * cqr)
 				    temp_erp->refers);
 		}
 	}
-#endif				/* ERP_DEBUG */
 
 	if (erp->status == DASD_CQR_FAILED)
 		dasd_log_ccw(erp, 1, cpa);

drivers/s390/block/dasd_devmap.c

@@ -202,6 +202,8 @@ dasd_feature_list(char *str, char **endp)
 			features |= DASD_FEATURE_READONLY;
 		else if (len == 4 && !strncmp(str, "diag", 4))
 			features |= DASD_FEATURE_USEDIAG;
+		else if (len == 6 && !strncmp(str, "erplog", 6))
+			features |= DASD_FEATURE_ERPLOG;
 		else {
 			MESSAGE(KERN_WARNING,
 				"unsupported feature: %*s, "
@@ -709,6 +711,52 @@ dasd_ro_store(struct device *dev, struct device_attribute *attr,
 }
 
 static DEVICE_ATTR(readonly, 0644, dasd_ro_show, dasd_ro_store);
+/*
+ * erplog controls the logging of ERP related data
+ * (e.g. failing channel programs).
+ */
+static ssize_t
+dasd_erplog_show(struct device *dev, struct device_attribute *attr, char *buf)
+{
+	struct dasd_devmap *devmap;
+	int erplog;
+
+	devmap = dasd_find_busid(dev->bus_id);
+	if (!IS_ERR(devmap))
+		erplog = (devmap->features & DASD_FEATURE_ERPLOG) != 0;
+	else
+		erplog = (DASD_FEATURE_DEFAULT & DASD_FEATURE_ERPLOG) != 0;
+	return snprintf(buf, PAGE_SIZE, erplog ? "1\n" : "0\n");
+}
+
+static ssize_t
+dasd_erplog_store(struct device *dev, struct device_attribute *attr,
+	      const char *buf, size_t count)
+{
+	struct dasd_devmap *devmap;
+	int val;
+	char *endp;
+
+	devmap = dasd_devmap_from_cdev(to_ccwdev(dev));
+	if (IS_ERR(devmap))
+		return PTR_ERR(devmap);
+
+	val = simple_strtoul(buf, &endp, 0);
+	if (((endp + 1) < (buf + count)) || (val > 1))
+		return -EINVAL;
+
+	spin_lock(&dasd_devmap_lock);
+	if (val)
+		devmap->features |= DASD_FEATURE_ERPLOG;
+	else
+		devmap->features &= ~DASD_FEATURE_ERPLOG;
+	if (devmap->device)
+		devmap->device->features = devmap->features;
+	spin_unlock(&dasd_devmap_lock);
+	return count;
+}
+
+static DEVICE_ATTR(erplog, 0644, dasd_erplog_show, dasd_erplog_store);
 
 /*
  * use_diag controls whether the driver should use diag rather than ssch
@@ -896,6 +944,7 @@ static struct attribute * dasd_attrs[] = {
 	&dev_attr_uid.attr,
 	&dev_attr_use_diag.attr,
 	&dev_attr_eer_enabled.attr,
+	&dev_attr_erplog.attr,
 	NULL,
 };
 

drivers/s390/block/dasd_int.h

@@ -13,10 +13,6 @@
 
 #ifdef __KERNEL__
 
-/* erp debugging in dasd.c and dasd_3990_erp.c */
-#define ERP_DEBUG
-
-
 /* we keep old device allocation scheme; IOW, minors are still in 0..255 */
 #define DASD_PER_MAJOR (1U << (MINORBITS - DASD_PARTN_BITS))
 #define DASD_PARTN_MASK ((1 << DASD_PARTN_BITS) - 1)

drivers/s390/char/ctrlchar.c

@@ -16,14 +16,15 @@
 
 #ifdef CONFIG_MAGIC_SYSRQ
 static int ctrlchar_sysrq_key;
+static struct tty_struct *sysrq_tty;
 
 static void
-ctrlchar_handle_sysrq(void *tty)
+ctrlchar_handle_sysrq(struct work_struct *work)
 {
-	handle_sysrq(ctrlchar_sysrq_key, (struct tty_struct *) tty);
+	handle_sysrq(ctrlchar_sysrq_key, sysrq_tty);
 }
 
-static DECLARE_WORK(ctrlchar_work, ctrlchar_handle_sysrq, NULL);
+static DECLARE_WORK(ctrlchar_work, ctrlchar_handle_sysrq);
 #endif
 
 
@@ -53,7 +54,7 @@ ctrlchar_handle(const unsigned char *buf, int len, struct tty_struct *tty)
 	/* racy */
 	if (len == 3 && buf[1] == '-') {
 		ctrlchar_sysrq_key = buf[2];
-		ctrlchar_work.data = tty;
+		sysrq_tty = tty;
 		schedule_work(&ctrlchar_work);
 		return CTRLCHAR_SYSRQ;
 	}

drivers/s390/char/tape.h

@@ -179,6 +179,7 @@ struct tape_char_data {
 /* Block Frontend Data */
 struct tape_blk_data
 {
+	struct tape_device *	device;
 	/* Block device request queue. */
 	request_queue_t *	request_queue;
 	spinlock_t		request_queue_lock;
@@ -240,7 +241,7 @@ struct tape_device {
 #endif
 
 	/* Function to start or stop the next request later. */
-	struct work_struct		tape_dnr;
+	struct delayed_work		tape_dnr;
 };
 
 /* Externals from tape_core.c */

drivers/s390/char/tape_34xx.c

@@ -95,6 +95,12 @@ tape_34xx_medium_sense(struct tape_device *device)
 	return rc;
 }
 
+struct tape_34xx_work {
+	struct tape_device	*device;
+	enum tape_op		 op;
+	struct work_struct	 work;
+};
+
 /*
  * These functions are currently used only to schedule a medium_sense for
  * later execution. This is because we get an interrupt whenever a medium
@@ -103,13 +109,10 @@ tape_34xx_medium_sense(struct tape_device *device)
  * interrupt handler.
  */
 static void
-tape_34xx_work_handler(void *data)
+tape_34xx_work_handler(struct work_struct *work)
 {
-	struct {
-		struct tape_device	*device;
-		enum tape_op		 op;
-		struct work_struct	 work;
-	} *p = data;
+	struct tape_34xx_work *p =
+		container_of(work, struct tape_34xx_work, work);
 
 	switch(p->op) {
 		case TO_MSEN:
@@ -126,17 +129,13 @@ tape_34xx_work_handler(void *data)
 static int
 tape_34xx_schedule_work(struct tape_device *device, enum tape_op op)
 {
-	struct {
-		struct tape_device	*device;
-		enum tape_op		 op;
-		struct work_struct	 work;
-	} *p;
+	struct tape_34xx_work *p;
 
 	if ((p = kmalloc(sizeof(*p), GFP_ATOMIC)) == NULL)
 		return -ENOMEM;
 
 	memset(p, 0, sizeof(*p));
-	INIT_WORK(&p->work, tape_34xx_work_handler, p);
+	INIT_WORK(&p->work, tape_34xx_work_handler);
 
 	p->device = tape_get_device_reference(device);
 	p->op     = op;

drivers/s390/char/tape_3590.c

@@ -236,9 +236,10 @@ struct work_handler_data {
 };
 
 static void
-tape_3590_work_handler(void *data)
+tape_3590_work_handler(struct work_struct *work)
 {
-	struct work_handler_data *p = data;
+	struct work_handler_data *p =
+		container_of(work, struct work_handler_data, work);
 
 	switch (p->op) {
 	case TO_MSEN:
@@ -263,7 +264,7 @@ tape_3590_schedule_work(struct tape_device *device, enum tape_op op)
 	if ((p = kzalloc(sizeof(*p), GFP_ATOMIC)) == NULL)
 		return -ENOMEM;
 
-	INIT_WORK(&p->work, tape_3590_work_handler, p);
+	INIT_WORK(&p->work, tape_3590_work_handler);
 
 	p->device = tape_get_device_reference(device);
 	p->op = op;

drivers/s390/char/tape_block.c

@@ -15,6 +15,7 @@
 #include <linux/blkdev.h>
 #include <linux/interrupt.h>
 #include <linux/buffer_head.h>
+#include <linux/kernel.h>
 
 #include <asm/debug.h>
 
@@ -143,7 +144,8 @@ tapeblock_start_request(struct tape_device *device, struct request *req)
  * queue.
  */
 static void
-tapeblock_requeue(void *data) {
+tapeblock_requeue(struct work_struct *work) {
+	struct tape_blk_data *	blkdat;
 	struct tape_device *	device;
 	request_queue_t *	queue;
 	int			nr_queued;
@@ -151,7 +153,8 @@ tapeblock_requeue(void *data) {
 	struct list_head *	l;
 	int			rc;
 
-	device = (struct tape_device *) data;
+	blkdat = container_of(work, struct tape_blk_data, requeue_task);
+	device = blkdat->device;
 	if (!device)
 		return;
 
@@ -212,6 +215,7 @@ tapeblock_setup_device(struct tape_device * device)
 	int			rc;
 
 	blkdat = &device->blk_data;
+	blkdat->device = device;
 	spin_lock_init(&blkdat->request_queue_lock);
 	atomic_set(&blkdat->requeue_scheduled, 0);
 
@@ -255,8 +259,8 @@ tapeblock_setup_device(struct tape_device * device)
 
 	add_disk(disk);
 
-	INIT_WORK(&blkdat->requeue_task, tapeblock_requeue,
-		tape_get_device_reference(device));
+	tape_get_device_reference(device);
+	INIT_WORK(&blkdat->requeue_task, tapeblock_requeue);
 
 	return 0;
 
@@ -271,7 +275,7 @@ void
 tapeblock_cleanup_device(struct tape_device *device)
 {
 	flush_scheduled_work();
-	device->blk_data.requeue_task.data = tape_put_device(device);
+	tape_put_device(device);
 
 	if (!device->blk_data.disk) {
 		PRINT_ERR("(%s): No gendisk to clean up!\n",

drivers/s390/char/tape_core.c

@@ -28,7 +28,7 @@
 #define PRINTK_HEADER "TAPE_CORE: "
 
 static void __tape_do_irq (struct ccw_device *, unsigned long, struct irb *);
-static void tape_delayed_next_request(void * data);
+static void tape_delayed_next_request(struct work_struct *);
 
 /*
  * One list to contain all tape devices of all disciplines, so
@@ -272,7 +272,7 @@ __tape_cancel_io(struct tape_device *device, struct tape_request *request)
 				return 0;
 			case -EBUSY:
 				request->status	= TAPE_REQUEST_CANCEL;
-				schedule_work(&device->tape_dnr);
+				schedule_delayed_work(&device->tape_dnr, 0);
 				return 0;
 			case -ENODEV:
 				DBF_EXCEPTION(2, "device gone, retry\n");
@@ -470,7 +470,7 @@ tape_alloc_device(void)
 	*device->modeset_byte = 0;
 	device->first_minor = -1;
 	atomic_set(&device->ref_count, 1);
-	INIT_WORK(&device->tape_dnr, tape_delayed_next_request, device);
+	INIT_DELAYED_WORK(&device->tape_dnr, tape_delayed_next_request);
 
 	return device;
 }
@@ -724,7 +724,7 @@ __tape_start_io(struct tape_device *device, struct tape_request *request)
 	} else if (rc == -EBUSY) {
 		/* The common I/O subsystem is currently busy. Retry later. */
 		request->status = TAPE_REQUEST_QUEUED;
-		schedule_work(&device->tape_dnr);
+		schedule_delayed_work(&device->tape_dnr, 0);
 		rc = 0;
 	} else {
 		/* Start failed. Remove request and indicate failure. */
@@ -790,11 +790,11 @@ __tape_start_next_request(struct tape_device *device)
 }
 
 static void
-tape_delayed_next_request(void *data)
+tape_delayed_next_request(struct work_struct *work)
 {
-	struct tape_device *	device;
+	struct tape_device *device =
+		container_of(work, struct tape_device, tape_dnr.work);
 
-	device = (struct tape_device *) data;
 	DBF_LH(6, "tape_delayed_next_request(%p)\n", device);
 	spin_lock_irq(get_ccwdev_lock(device->cdev));
 	__tape_start_next_request(device);

drivers/s390/cio/chsc.c

@@ -183,7 +183,7 @@ css_get_ssd_info(struct subchannel *sch)
 	page = (void *)get_zeroed_page(GFP_KERNEL | GFP_DMA);
 	if (!page)
 		return -ENOMEM;
-	spin_lock_irq(&sch->lock);
+	spin_lock_irq(sch->lock);
 	ret = chsc_get_sch_desc_irq(sch, page);
 	if (ret) {
 		static int cio_chsc_err_msg;
@@ -197,7 +197,7 @@ css_get_ssd_info(struct subchannel *sch)
 			cio_chsc_err_msg = 1;
 		}
 	}
-	spin_unlock_irq(&sch->lock);
+	spin_unlock_irq(sch->lock);
 	free_page((unsigned long)page);
 	if (!ret) {
 		int j, chpid, mask;
@@ -233,7 +233,7 @@ s390_subchannel_remove_chpid(struct device *dev, void *data)
 	if (j >= 8)
 		return 0;
 
-	spin_lock_irq(&sch->lock);
+	spin_lock_irq(sch->lock);
 
 	stsch(sch->schid, &schib);
 	if (!schib.pmcw.dnv)
@@ -265,10 +265,10 @@ s390_subchannel_remove_chpid(struct device *dev, void *data)
 	else if (sch->lpm == mask)
 		goto out_unreg;
 out_unlock:
-	spin_unlock_irq(&sch->lock);
+	spin_unlock_irq(sch->lock);
 	return 0;
 out_unreg:
-	spin_unlock_irq(&sch->lock);
+	spin_unlock_irq(sch->lock);
 	sch->lpm = 0;
 	if (css_enqueue_subchannel_slow(sch->schid)) {
 		css_clear_subchannel_slow_list();
@@ -378,12 +378,12 @@ __s390_process_res_acc(struct subchannel_id schid, void *data)
 		/* Check if a subchannel is newly available. */
 		return s390_process_res_acc_new_sch(schid);
 
-	spin_lock_irq(&sch->lock);
+	spin_lock_irq(sch->lock);
 
 	chp_mask = s390_process_res_acc_sch(res_data, sch);
 
 	if (chp_mask == 0) {
-		spin_unlock_irq(&sch->lock);
+		spin_unlock_irq(sch->lock);
 		put_device(&sch->dev);
 		return 0;
 	}
@@ -397,7 +397,7 @@ __s390_process_res_acc(struct subchannel_id schid, void *data)
 	else if (sch->driver && sch->driver->verify)
 		sch->driver->verify(&sch->dev);
 
-	spin_unlock_irq(&sch->lock);
+	spin_unlock_irq(sch->lock);
 	put_device(&sch->dev);
 	return 0;
 }
@@ -635,21 +635,21 @@ __chp_add(struct subchannel_id schid, void *data)
 	if (!sch)
 		/* Check if the subchannel is now available. */
 		return __chp_add_new_sch(schid);
-	spin_lock_irq(&sch->lock);
+	spin_lock_irq(sch->lock);
 	for (i=0; i<8; i++) {
 		mask = 0x80 >> i;
 		if ((sch->schib.pmcw.pim & mask) &&
 		    (sch->schib.pmcw.chpid[i] == chp->id)) {
 			if (stsch(sch->schid, &sch->schib) != 0) {
 				/* Endgame. */
-				spin_unlock_irq(&sch->lock);
+				spin_unlock_irq(sch->lock);
 				return -ENXIO;
 			}
 			break;
 		}
 	}
 	if (i==8) {
-		spin_unlock_irq(&sch->lock);
+		spin_unlock_irq(sch->lock);
 		return 0;
 	}
 	sch->lpm = ((sch->schib.pmcw.pim &
@@ -660,7 +660,7 @@ __chp_add(struct subchannel_id schid, void *data)
 	if (sch->driver && sch->driver->verify)
 		sch->driver->verify(&sch->dev);
 
-	spin_unlock_irq(&sch->lock);
+	spin_unlock_irq(sch->lock);
 	put_device(&sch->dev);
 	return 0;
 }
@@ -750,7 +750,7 @@ __s390_subchannel_vary_chpid(struct subchannel *sch, __u8 chpid, int on)
 	if (!sch->ssd_info.valid)
 		return;
 	
-	spin_lock_irqsave(&sch->lock, flags);
+	spin_lock_irqsave(sch->lock, flags);
 	old_lpm = sch->lpm;
 	for (chp = 0; chp < 8; chp++) {
 		if (sch->ssd_info.chpid[chp] != chpid)
@@ -785,7 +785,7 @@ __s390_subchannel_vary_chpid(struct subchannel *sch, __u8 chpid, int on)
 			sch->driver->verify(&sch->dev);
 		break;
 	}
-	spin_unlock_irqrestore(&sch->lock, flags);
+	spin_unlock_irqrestore(sch->lock, flags);
 }
 
 static int

drivers/s390/cio/cio.c

@@ -143,11 +143,11 @@ cio_tpi(void)
 		return 1;
 	local_bh_disable();
 	irq_enter ();
-	spin_lock(&sch->lock);
+	spin_lock(sch->lock);
 	memcpy (&sch->schib.scsw, &irb->scsw, sizeof (struct scsw));
 	if (sch->driver && sch->driver->irq)
 		sch->driver->irq(&sch->dev);
-	spin_unlock(&sch->lock);
+	spin_unlock(sch->lock);
 	irq_exit ();
 	_local_bh_enable();
 	return 1;
@@ -415,6 +415,8 @@ cio_enable_subchannel (struct subchannel *sch, unsigned int isc)
 	CIO_TRACE_EVENT (2, "ensch");
 	CIO_TRACE_EVENT (2, sch->dev.bus_id);
 
+	if (sch_is_pseudo_sch(sch))
+		return -EINVAL;
 	ccode = stsch (sch->schid, &sch->schib);
 	if (ccode)
 		return -ENODEV;
@@ -462,6 +464,8 @@ cio_disable_subchannel (struct subchannel *sch)
 	CIO_TRACE_EVENT (2, "dissch");
 	CIO_TRACE_EVENT (2, sch->dev.bus_id);
 
+	if (sch_is_pseudo_sch(sch))
+		return 0;
 	ccode = stsch (sch->schid, &sch->schib);
 	if (ccode == 3)		/* Not operational. */
 		return -ENODEV;
@@ -496,6 +500,15 @@ cio_disable_subchannel (struct subchannel *sch)
 	return ret;
 }
 
+int cio_create_sch_lock(struct subchannel *sch)
+{
+	sch->lock = kmalloc(sizeof(spinlock_t), GFP_KERNEL);
+	if (!sch->lock)
+		return -ENOMEM;
+	spin_lock_init(sch->lock);
+	return 0;
+}
+
 /*
  * cio_validate_subchannel()
  *
@@ -513,6 +526,7 @@ cio_validate_subchannel (struct subchannel *sch, struct subchannel_id schid)
 {
 	char dbf_txt[15];
 	int ccode;
+	int err;
 
 	sprintf (dbf_txt, "valsch%x", schid.sch_no);
 	CIO_TRACE_EVENT (4, dbf_txt);
@@ -520,9 +534,15 @@ cio_validate_subchannel (struct subchannel *sch, struct subchannel_id schid)
 	/* Nuke all fields. */
 	memset(sch, 0, sizeof(struct subchannel));
 
-	spin_lock_init(&sch->lock);
+	sch->schid = schid;
+	if (cio_is_console(schid)) {
+		sch->lock = cio_get_console_lock();
+	} else {
+		err = cio_create_sch_lock(sch);
+		if (err)
+			goto out;
+	}
 	mutex_init(&sch->reg_mutex);
-
 	/* Set a name for the subchannel */
 	snprintf (sch->dev.bus_id, BUS_ID_SIZE, "0.%x.%04x", schid.ssid,
 		  schid.sch_no);
@@ -534,10 +554,10 @@ cio_validate_subchannel (struct subchannel *sch, struct subchannel_id schid)
 	 *  is not valid.
 	 */
 	ccode = stsch_err (schid, &sch->schib);
-	if (ccode)
-		return (ccode == 3) ? -ENXIO : ccode;
-
-	sch->schid = schid;
+	if (ccode) {
+		err = (ccode == 3) ? -ENXIO : ccode;
+		goto out;
+	}
 	/* Copy subchannel type from path management control word. */
 	sch->st = sch->schib.pmcw.st;
 
@@ -550,14 +570,16 @@ cio_validate_subchannel (struct subchannel *sch, struct subchannel_id schid)
 			  "non-I/O subchannel type %04X\n",
 			  sch->schid.ssid, sch->schid.sch_no, sch->st);
 		/* We stop here for non-io subchannels. */
-		return sch->st;
+		err = sch->st;
+		goto out;
 	}
 
 	/* Initialization for io subchannels. */
-	if (!sch->schib.pmcw.dnv)
+	if (!sch->schib.pmcw.dnv) {
 		/* io subchannel but device number is invalid. */
-		return -ENODEV;
-
+		err = -ENODEV;
+		goto out;
+	}
 	/* Devno is valid. */
 	if (is_blacklisted (sch->schid.ssid, sch->schib.pmcw.dev)) {
 		/*
@@ -567,7 +589,8 @@ cio_validate_subchannel (struct subchannel *sch, struct subchannel_id schid)
 		CIO_MSG_EVENT(0, "Blacklisted device detected "
 			      "at devno %04X, subchannel set %x\n",
 			      sch->schib.pmcw.dev, sch->schid.ssid);
-		return -ENODEV;
+		err = -ENODEV;
+		goto out;
 	}
 	sch->opm = 0xff;
 	if (!cio_is_console(sch->schid))
@@ -595,6 +618,11 @@ cio_validate_subchannel (struct subchannel *sch, struct subchannel_id schid)
 	if ((sch->lpm & (sch->lpm - 1)) != 0)
 		sch->schib.pmcw.mp = 1;	/* multipath mode */
 	return 0;
+out:
+	if (!cio_is_console(schid))
+		kfree(sch->lock);
+	sch->lock = NULL;
+	return err;
 }
 
 /*
@@ -637,7 +665,7 @@ do_IRQ (struct pt_regs *regs)
 		}
 		sch = (struct subchannel *)(unsigned long)tpi_info->intparm;
 		if (sch)
-			spin_lock(&sch->lock);
+			spin_lock(sch->lock);
 		/* Store interrupt response block to lowcore. */
 		if (tsch (tpi_info->schid, irb) == 0 && sch) {
 			/* Keep subchannel information word up to date. */
@@ -648,7 +676,7 @@ do_IRQ (struct pt_regs *regs)
 				sch->driver->irq(&sch->dev);
 		}
 		if (sch)
-			spin_unlock(&sch->lock);
+			spin_unlock(sch->lock);
 		/*
 		 * Are more interrupts pending?
 		 * If so, the tpi instruction will update the lowcore
@@ -687,10 +715,10 @@ wait_cons_dev (void)
 	__ctl_load (cr6, 6, 6);
 
 	do {
-		spin_unlock(&console_subchannel.lock);
+		spin_unlock(console_subchannel.lock);
 		if (!cio_tpi())
 			cpu_relax();
-		spin_lock(&console_subchannel.lock);
+		spin_lock(console_subchannel.lock);
 	} while (console_subchannel.schib.scsw.actl != 0);
 	/*
 	 * restore previous isc value

drivers/s390/cio/cio.h

@@ -87,7 +87,7 @@ struct orb {
 /* subchannel data structure used by I/O subroutines */
 struct subchannel {
 	struct subchannel_id schid;
-	spinlock_t lock;	/* subchannel lock */
+	spinlock_t *lock;	/* subchannel lock */
 	struct mutex reg_mutex;
 	enum {
 		SUBCHANNEL_TYPE_IO = 0,
@@ -131,15 +131,19 @@ extern int cio_set_options (struct subchannel *, int);
 extern int cio_get_options (struct subchannel *);
 extern int cio_modify (struct subchannel *);
 
+int cio_create_sch_lock(struct subchannel *);
+
 /* Use with care. */
 #ifdef CONFIG_CCW_CONSOLE
 extern struct subchannel *cio_probe_console(void);
 extern void cio_release_console(void);
 extern int cio_is_console(struct subchannel_id);
 extern struct subchannel *cio_get_console_subchannel(void);
+extern spinlock_t * cio_get_console_lock(void);
 #else
 #define cio_is_console(schid) 0
 #define cio_get_console_subchannel() NULL
+#define cio_get_console_lock() NULL;
 #endif
 
 extern int cio_show_msg;

drivers/s390/cio/css.c

@@ -91,9 +91,9 @@ css_free_subchannel(struct subchannel *sch)
 		/* Reset intparm to zeroes. */
 		sch->schib.pmcw.intparm = 0;
 		cio_modify(sch);
+		kfree(sch->lock);
 		kfree(sch);
 	}
-	
 }
 
 static void
@@ -102,8 +102,10 @@ css_subchannel_release(struct device *dev)
 	struct subchannel *sch;
 
 	sch = to_subchannel(dev);
-	if (!cio_is_console(sch->schid))
+	if (!cio_is_console(sch->schid)) {
+		kfree(sch->lock);
 		kfree(sch);
+	}
 }
 
 extern int css_get_ssd_info(struct subchannel *sch);
@@ -135,13 +137,15 @@ css_register_subchannel(struct subchannel *sch)
 	sch->dev.parent = &css[0]->device;
 	sch->dev.bus = &css_bus_type;
 	sch->dev.release = &css_subchannel_release;
+	sch->dev.groups = subch_attr_groups;
 
 	/* make it known to the system */
 	ret = css_sch_device_register(sch);
-	if (ret)
+	if (ret) {
 		printk (KERN_WARNING "%s: could not register %s\n",
 			__func__, sch->dev.bus_id);
-	else
+		return ret;
+	}
 	css_get_ssd_info(sch);
 	return ret;
 }
@@ -201,18 +205,18 @@ static int css_evaluate_known_subchannel(struct subchannel *sch, int slow)
 	unsigned long flags;
 	enum { NONE, UNREGISTER, UNREGISTER_PROBE, REPROBE } action;
 
-	spin_lock_irqsave(&sch->lock, flags);
+	spin_lock_irqsave(sch->lock, flags);
 	disc = device_is_disconnected(sch);
 	if (disc && slow) {
 		/* Disconnected devices are evaluated directly only.*/
-		spin_unlock_irqrestore(&sch->lock, flags);
+		spin_unlock_irqrestore(sch->lock, flags);
 		return 0;
 	}
 	/* No interrupt after machine check - kill pending timers. */
 	device_kill_pending_timer(sch);
 	if (!disc && !slow) {
 		/* Non-disconnected devices are evaluated on the slow path. */
-		spin_unlock_irqrestore(&sch->lock, flags);
+		spin_unlock_irqrestore(sch->lock, flags);
 		return -EAGAIN;
 	}
 	event = css_get_subchannel_status(sch);
@@ -237,9 +241,9 @@ static int css_evaluate_known_subchannel(struct subchannel *sch, int slow)
 		/* Ask driver what to do with device. */
 		action = UNREGISTER;
 		if (sch->driver && sch->driver->notify) {
-			spin_unlock_irqrestore(&sch->lock, flags);
+			spin_unlock_irqrestore(sch->lock, flags);
 			ret = sch->driver->notify(&sch->dev, event);
-			spin_lock_irqsave(&sch->lock, flags);
+			spin_lock_irqsave(sch->lock, flags);
 			if (ret)
 				action = NONE;
 		}
@@ -264,9 +268,9 @@ static int css_evaluate_known_subchannel(struct subchannel *sch, int slow)
 	case UNREGISTER:
 	case UNREGISTER_PROBE:
 		/* Unregister device (will use subchannel lock). */
-		spin_unlock_irqrestore(&sch->lock, flags);
+		spin_unlock_irqrestore(sch->lock, flags);
 		css_sch_device_unregister(sch);
-		spin_lock_irqsave(&sch->lock, flags);
+		spin_lock_irqsave(sch->lock, flags);
 
 		/* Reset intparm to zeroes. */
 		sch->schib.pmcw.intparm = 0;
@@ -278,7 +282,7 @@ static int css_evaluate_known_subchannel(struct subchannel *sch, int slow)
 	default:
 		break;
 	}
-	spin_unlock_irqrestore(&sch->lock, flags);
+	spin_unlock_irqrestore(sch->lock, flags);
 	/* Probe if necessary. */
 	if (action == UNREGISTER_PROBE)
 		ret = css_probe_device(sch->schid);
@@ -573,12 +577,24 @@ css_cm_enable_store(struct device *dev, struct device_attribute *attr,
 
 static DEVICE_ATTR(cm_enable, 0644, css_cm_enable_show, css_cm_enable_store);
 
-static inline void __init
-setup_css(int nr)
+static inline int __init setup_css(int nr)
 {
 	u32 tod_high;
+	int ret;
 
 	memset(css[nr], 0, sizeof(struct channel_subsystem));
+	css[nr]->pseudo_subchannel =
+		kzalloc(sizeof(*css[nr]->pseudo_subchannel), GFP_KERNEL);
+	if (!css[nr]->pseudo_subchannel)
+		return -ENOMEM;
+	css[nr]->pseudo_subchannel->dev.parent = &css[nr]->device;
+	css[nr]->pseudo_subchannel->dev.release = css_subchannel_release;
+	sprintf(css[nr]->pseudo_subchannel->dev.bus_id, "defunct");
+	ret = cio_create_sch_lock(css[nr]->pseudo_subchannel);
+	if (ret) {
+		kfree(css[nr]->pseudo_subchannel);
+		return ret;
+	}
 	mutex_init(&css[nr]->mutex);
 	css[nr]->valid = 1;
 	css[nr]->cssid = nr;
@@ -586,6 +602,7 @@ setup_css(int nr)
 	css[nr]->device.release = channel_subsystem_release;
 	tod_high = (u32) (get_clock() >> 32);
 	css_generate_pgid(css[nr], tod_high);
+	return 0;
 }
 
 /*
@@ -622,10 +639,12 @@ init_channel_subsystem (void)
 			ret = -ENOMEM;
 			goto out_unregister;
 		}
-		setup_css(i);
-		ret = device_register(&css[i]->device);
+		ret = setup_css(i);
 		if (ret)
 			goto out_free;
+		ret = device_register(&css[i]->device);
+		if (ret)
+			goto out_free_all;
 		if (css_characteristics_avail &&
 		    css_chsc_characteristics.secm) {
 			ret = device_create_file(&css[i]->device,
@@ -633,6 +652,9 @@ init_channel_subsystem (void)
 			if (ret)
 				goto out_device;
 		}
+		ret = device_register(&css[i]->pseudo_subchannel->dev);
+		if (ret)
+			goto out_file;
 	}
 	css_init_done = 1;
 
@@ -640,13 +662,19 @@ init_channel_subsystem (void)
 
 	for_each_subchannel(__init_channel_subsystem, NULL);
 	return 0;
+out_file:
+	device_remove_file(&css[i]->device, &dev_attr_cm_enable);
 out_device:
 	device_unregister(&css[i]->device);
+out_free_all:
+	kfree(css[i]->pseudo_subchannel->lock);
+	kfree(css[i]->pseudo_subchannel);
 out_free:
 	kfree(css[i]);
 out_unregister:
 	while (i > 0) {
 		i--;
+		device_unregister(&css[i]->pseudo_subchannel->dev);
 		if (css_characteristics_avail && css_chsc_characteristics.secm)
 			device_remove_file(&css[i]->device,
 					   &dev_attr_cm_enable);
@@ -658,6 +686,11 @@ out:
 	return ret;
 }
 
+int sch_is_pseudo_sch(struct subchannel *sch)
+{
+	return sch == to_css(sch->dev.parent)->pseudo_subchannel;
+}
+
 /*
  * find a driver for a subchannel. They identify by the subchannel
  * type with the exception that the console subchannel driver has its own

drivers/s390/cio/css.h

@@ -73,6 +73,8 @@ struct senseid {
 }  __attribute__ ((packed,aligned(4)));
 
 struct ccw_device_private {
+	struct ccw_device *cdev;
+	struct subchannel *sch;
 	int state;		/* device state */
 	atomic_t onoff;
 	unsigned long registered;
@@ -158,6 +160,8 @@ struct channel_subsystem {
 	int cm_enabled;
 	void *cub_addr1;
 	void *cub_addr2;
+	/* for orphaned ccw devices */
+	struct subchannel *pseudo_subchannel;
 };
 #define to_css(dev) container_of(dev, struct channel_subsystem, device)
 
@@ -185,6 +189,11 @@ void css_clear_subchannel_slow_list(void);
 int css_slow_subchannels_exist(void);
 extern int need_rescan;
 
+int sch_is_pseudo_sch(struct subchannel *);
+
 extern struct workqueue_struct *slow_path_wq;
 extern struct work_struct slow_path_work;
+
+int subchannel_add_files (struct device *);
+extern struct attribute_group *subch_attr_groups[];
 #endif

drivers/s390/cio/device.h

@@ -78,8 +78,10 @@ void io_subchannel_recog_done(struct ccw_device *cdev);
 
 int ccw_device_cancel_halt_clear(struct ccw_device *);
 
-void ccw_device_do_unreg_rereg(void *);
-void ccw_device_call_sch_unregister(void *);
+void ccw_device_do_unreg_rereg(struct work_struct *);
+void ccw_device_call_sch_unregister(struct work_struct *);
+void ccw_device_move_to_orphanage(struct work_struct *);
+int ccw_device_is_orphan(struct ccw_device *);
 
 int ccw_device_recognition(struct ccw_device *);
 int ccw_device_online(struct ccw_device *);

drivers/s390/cio/device_fsm.c

@@ -186,15 +186,14 @@ ccw_device_handle_oper(struct ccw_device *cdev)
 	/*
 	 * Check if cu type and device type still match. If
 	 * not, it is certainly another device and we have to
-	 * de- and re-register. Also check here for non-matching devno.
+	 * de- and re-register.
 	 */
 	if (cdev->id.cu_type != cdev->private->senseid.cu_type ||
 	    cdev->id.cu_model != cdev->private->senseid.cu_model ||
 	    cdev->id.dev_type != cdev->private->senseid.dev_type ||
-	    cdev->id.dev_model != cdev->private->senseid.dev_model ||
-	    cdev->private->dev_id.devno != sch->schib.pmcw.dev) {
+	    cdev->id.dev_model != cdev->private->senseid.dev_model) {
 		PREPARE_WORK(&cdev->private->kick_work,
-			     ccw_device_do_unreg_rereg, cdev);
+			     ccw_device_do_unreg_rereg);
 		queue_work(ccw_device_work, &cdev->private->kick_work);
 		return 0;
 	}
@@ -329,19 +328,21 @@ ccw_device_sense_id_done(struct ccw_device *cdev, int err)
 }
 
 static void
-ccw_device_oper_notify(void *data)
+ccw_device_oper_notify(struct work_struct *work)
 {
+	struct ccw_device_private *priv;
 	struct ccw_device *cdev;
 	struct subchannel *sch;
 	int ret;
 
-	cdev = data;
+	priv = container_of(work, struct ccw_device_private, kick_work);
+	cdev = priv->cdev;
 	sch = to_subchannel(cdev->dev.parent);
 	ret = (sch->driver && sch->driver->notify) ?
 		sch->driver->notify(&sch->dev, CIO_OPER) : 0;
 	if (!ret)
 		/* Driver doesn't want device back. */
-		ccw_device_do_unreg_rereg(cdev);
+		ccw_device_do_unreg_rereg(work);
 	else {
 		/* Reenable channel measurements, if needed. */
 		cmf_reenable(cdev);
@@ -377,8 +378,7 @@ ccw_device_done(struct ccw_device *cdev, int state)
 
 	if (cdev->private->flags.donotify) {
 		cdev->private->flags.donotify = 0;
-		PREPARE_WORK(&cdev->private->kick_work, ccw_device_oper_notify,
-			     cdev);
+		PREPARE_WORK(&cdev->private->kick_work, ccw_device_oper_notify);
 		queue_work(ccw_device_notify_work, &cdev->private->kick_work);
 	}
 	wake_up(&cdev->private->wait_q);
@@ -528,13 +528,15 @@ ccw_device_recog_timeout(struct ccw_device *cdev, enum dev_event dev_event)
 
 
 static void
-ccw_device_nopath_notify(void *data)
+ccw_device_nopath_notify(struct work_struct *work)
 {
+	struct ccw_device_private *priv;
 	struct ccw_device *cdev;
 	struct subchannel *sch;
 	int ret;
 
-	cdev = data;
+	priv = container_of(work, struct ccw_device_private, kick_work);
+	cdev = priv->cdev;
 	sch = to_subchannel(cdev->dev.parent);
 	/* Extra sanity. */
 	if (sch->lpm)
@@ -547,8 +549,7 @@ ccw_device_nopath_notify(void *data)
 			cio_disable_subchannel(sch);
 			if (get_device(&cdev->dev)) {
 				PREPARE_WORK(&cdev->private->kick_work,
-					     ccw_device_call_sch_unregister,
-					     cdev);
+					     ccw_device_call_sch_unregister);
 				queue_work(ccw_device_work,
 					   &cdev->private->kick_work);
 			} else
@@ -607,7 +608,7 @@ ccw_device_verify_done(struct ccw_device *cdev, int err)
 		/* Reset oper notify indication after verify error. */
 		cdev->private->flags.donotify = 0;
 		PREPARE_WORK(&cdev->private->kick_work,
-			     ccw_device_nopath_notify, cdev);
+			     ccw_device_nopath_notify);
 		queue_work(ccw_device_notify_work, &cdev->private->kick_work);
 		ccw_device_done(cdev, DEV_STATE_NOT_OPER);
 		break;
@@ -674,6 +675,10 @@ ccw_device_offline(struct ccw_device *cdev)
 {
 	struct subchannel *sch;
 
+	if (ccw_device_is_orphan(cdev)) {
+		ccw_device_done(cdev, DEV_STATE_OFFLINE);
+		return 0;
+	}
 	sch = to_subchannel(cdev->dev.parent);
 	if (stsch(sch->schid, &sch->schib) || !sch->schib.pmcw.dnv)
 		return -ENODEV;
@@ -738,7 +743,7 @@ ccw_device_offline_notoper(struct ccw_device *cdev, enum dev_event dev_event)
 	sch = to_subchannel(cdev->dev.parent);
 	if (get_device(&cdev->dev)) {
 		PREPARE_WORK(&cdev->private->kick_work,
-			     ccw_device_call_sch_unregister, cdev);
+			     ccw_device_call_sch_unregister);
 		queue_work(ccw_device_work, &cdev->private->kick_work);
 	}
 	wake_up(&cdev->private->wait_q);
@@ -769,7 +774,7 @@ ccw_device_online_notoper(struct ccw_device *cdev, enum dev_event dev_event)
 	}
 	if (get_device(&cdev->dev)) {
 		PREPARE_WORK(&cdev->private->kick_work,
-			     ccw_device_call_sch_unregister, cdev);
+			     ccw_device_call_sch_unregister);
 		queue_work(ccw_device_work, &cdev->private->kick_work);
 	}
 	wake_up(&cdev->private->wait_q);
@@ -874,7 +879,7 @@ ccw_device_online_timeout(struct ccw_device *cdev, enum dev_event dev_event)
 		sch = to_subchannel(cdev->dev.parent);
 		if (!sch->lpm) {
 			PREPARE_WORK(&cdev->private->kick_work,
-				     ccw_device_nopath_notify, cdev);
+				     ccw_device_nopath_notify);
 			queue_work(ccw_device_notify_work,
 				   &cdev->private->kick_work);
 		} else
@@ -969,7 +974,7 @@ ccw_device_killing_irq(struct ccw_device *cdev, enum dev_event dev_event)
 			      ERR_PTR(-EIO));
 	if (!sch->lpm) {
 		PREPARE_WORK(&cdev->private->kick_work,
-			     ccw_device_nopath_notify, cdev);
+			     ccw_device_nopath_notify);
 		queue_work(ccw_device_notify_work, &cdev->private->kick_work);
 	} else if (cdev->private->flags.doverify)
 		/* Start delayed path verification. */
@@ -992,7 +997,7 @@ ccw_device_killing_timeout(struct ccw_device *cdev, enum dev_event dev_event)
 		sch = to_subchannel(cdev->dev.parent);
 		if (!sch->lpm) {
 			PREPARE_WORK(&cdev->private->kick_work,
-				     ccw_device_nopath_notify, cdev);
+				     ccw_device_nopath_notify);
 			queue_work(ccw_device_notify_work,
 				   &cdev->private->kick_work);
 		} else
@@ -1021,7 +1026,7 @@ void device_kill_io(struct subchannel *sch)
 	if (ret == -ENODEV) {
 		if (!sch->lpm) {
 			PREPARE_WORK(&cdev->private->kick_work,
-				     ccw_device_nopath_notify, cdev);
+				     ccw_device_nopath_notify);
 			queue_work(ccw_device_notify_work,
 				   &cdev->private->kick_work);
 		} else
@@ -1033,7 +1038,7 @@ void device_kill_io(struct subchannel *sch)
 			      ERR_PTR(-EIO));
 	if (!sch->lpm) {
 		PREPARE_WORK(&cdev->private->kick_work,
-			     ccw_device_nopath_notify, cdev);
+			     ccw_device_nopath_notify);
 		queue_work(ccw_device_notify_work, &cdev->private->kick_work);
 	} else
 		/* Start delayed path verification. */
@@ -1104,7 +1109,8 @@ device_trigger_reprobe(struct subchannel *sch)
 	/* Update some values. */
 	if (stsch(sch->schid, &sch->schib))
 		return;
-
+	if (!sch->schib.pmcw.dnv)
+		return;
 	/*
 	 * The pim, pam, pom values may not be accurate, but they are the best
 	 * we have before performing device selection :/
@@ -1118,6 +1124,12 @@ device_trigger_reprobe(struct subchannel *sch)
 		sch->schib.pmcw.mp = 1;
 	sch->schib.pmcw.intparm = (__u32)(unsigned long)sch;
 	/* We should also udate ssd info, but this has to wait. */
+	/* Check if this is another device which appeared on the same sch. */
+	if (sch->schib.pmcw.dev != cdev->private->dev_id.devno) {
+		PREPARE_WORK(&cdev->private->kick_work,
+			     ccw_device_move_to_orphanage);
+		queue_work(ccw_device_work, &cdev->private->kick_work);
+	} else
 		ccw_device_start_id(cdev, 0);
 }
 

drivers/s390/cio/device_ops.c

@@ -316,9 +316,9 @@ __ccw_device_retry_loop(struct ccw_device *cdev, struct ccw1 *ccw, long magic, _
 			ccw_device_set_timeout(cdev, 0);
 		if (ret == -EBUSY) {
 			/* Try again later. */
-			spin_unlock_irq(&sch->lock);
+			spin_unlock_irq(sch->lock);
 			msleep(10);
-			spin_lock_irq(&sch->lock);
+			spin_lock_irq(sch->lock);
 			continue;
 		}
 		if (ret != 0)
@@ -326,12 +326,12 @@ __ccw_device_retry_loop(struct ccw_device *cdev, struct ccw1 *ccw, long magic, _
 			break;
 		/* Wait for end of request. */
 		cdev->private->intparm = magic;
-		spin_unlock_irq(&sch->lock);
+		spin_unlock_irq(sch->lock);
 		wait_event(cdev->private->wait_q,
 			   (cdev->private->intparm == -EIO) ||
 			   (cdev->private->intparm == -EAGAIN) ||
 			   (cdev->private->intparm == 0));
-		spin_lock_irq(&sch->lock);
+		spin_lock_irq(sch->lock);
 		/* Check at least for channel end / device end */
 		if (cdev->private->intparm == -EIO) {
 			/* Non-retryable error. */
@@ -342,9 +342,9 @@ __ccw_device_retry_loop(struct ccw_device *cdev, struct ccw1 *ccw, long magic, _
 			/* Success. */
 			break;
 		/* Try again later. */
-		spin_unlock_irq(&sch->lock);
+		spin_unlock_irq(sch->lock);
 		msleep(10);
-		spin_lock_irq(&sch->lock);
+		spin_lock_irq(sch->lock);
 	} while (1);
 
 	return ret;
@@ -389,7 +389,7 @@ read_dev_chars (struct ccw_device *cdev, void **buffer, int length)
 		return ret;
 	}
 
-	spin_lock_irq(&sch->lock);
+	spin_lock_irq(sch->lock);
 	/* Save interrupt handler. */
 	handler = cdev->handler;
 	/* Temporarily install own handler. */
@@ -406,7 +406,7 @@ read_dev_chars (struct ccw_device *cdev, void **buffer, int length)
 
 	/* Restore interrupt handler. */
 	cdev->handler = handler;
-	spin_unlock_irq(&sch->lock);
+	spin_unlock_irq(sch->lock);
 
 	clear_normalized_cda (rdc_ccw);
 	kfree(rdc_ccw);
@@ -463,7 +463,7 @@ read_conf_data_lpm (struct ccw_device *cdev, void **buffer, int *length, __u8 lp
 	rcd_ccw->count = ciw->count;
 	rcd_ccw->flags = CCW_FLAG_SLI;
 
-	spin_lock_irq(&sch->lock);
+	spin_lock_irq(sch->lock);
 	/* Save interrupt handler. */
 	handler = cdev->handler;
 	/* Temporarily install own handler. */
@@ -480,7 +480,7 @@ read_conf_data_lpm (struct ccw_device *cdev, void **buffer, int *length, __u8 lp
 
 	/* Restore interrupt handler. */
 	cdev->handler = handler;
-	spin_unlock_irq(&sch->lock);
+	spin_unlock_irq(sch->lock);
 
  	/*
  	 * on success we update the user input parms
@@ -537,7 +537,7 @@ ccw_device_stlck(struct ccw_device *cdev)
 		kfree(buf);
 		return -ENOMEM;
 	}
-	spin_lock_irqsave(&sch->lock, flags);
+	spin_lock_irqsave(sch->lock, flags);
 	ret = cio_enable_subchannel(sch, 3);
 	if (ret)
 		goto out_unlock;
@@ -559,9 +559,9 @@ ccw_device_stlck(struct ccw_device *cdev)
 		goto out_unlock;
 	}
 	cdev->private->irb.scsw.actl |= SCSW_ACTL_START_PEND;
-	spin_unlock_irqrestore(&sch->lock, flags);
+	spin_unlock_irqrestore(sch->lock, flags);
 	wait_event(cdev->private->wait_q, cdev->private->irb.scsw.actl == 0);
-	spin_lock_irqsave(&sch->lock, flags);
+	spin_lock_irqsave(sch->lock, flags);
 	cio_disable_subchannel(sch); //FIXME: return code?
 	if ((cdev->private->irb.scsw.dstat !=
 	     (DEV_STAT_CHN_END|DEV_STAT_DEV_END)) ||
@@ -572,7 +572,7 @@ ccw_device_stlck(struct ccw_device *cdev)
 out_unlock:
 	kfree(buf);
 	kfree(buf2);
-	spin_unlock_irqrestore(&sch->lock, flags);
+	spin_unlock_irqrestore(sch->lock, flags);
 	return ret;
 }
 

drivers/s390/cio/qdio.h

@@ -12,10 +12,6 @@
 #endif /* CONFIG_QDIO_DEBUG */
 #define QDIO_USE_PROCESSING_STATE
 
-#ifdef CONFIG_QDIO_PERF_STATS
-#define QDIO_PERFORMANCE_STATS
-#endif /* CONFIG_QDIO_PERF_STATS */
-
 #define QDIO_MINIMAL_BH_RELIEF_TIME 16
 #define QDIO_TIMER_POLL_VALUE 1
 #define IQDIO_TIMER_POLL_VALUE 1
@@ -409,25 +405,23 @@ do_clear_global_summary(void)
 #define CHSC_FLAG_SIGA_SYNC_DONE_ON_THININTS 0x08
 #define CHSC_FLAG_SIGA_SYNC_DONE_ON_OUTB_PCIS 0x04
 
-#ifdef QDIO_PERFORMANCE_STATS
 struct qdio_perf_stats {
-	unsigned int tl_runs;
+	unsigned long tl_runs;
 
-	unsigned int siga_outs;
-	unsigned int siga_ins;
-	unsigned int siga_syncs;
-	unsigned int pcis;
-	unsigned int thinints;
-	unsigned int fast_reqs;
+	unsigned long siga_outs;
+	unsigned long siga_ins;
+	unsigned long siga_syncs;
+	unsigned long pcis;
+	unsigned long thinints;
+	unsigned long fast_reqs;
 
 	__u64 start_time_outbound;
-	unsigned int outbound_cnt;
-	unsigned int outbound_time;
+	unsigned long outbound_cnt;
+	unsigned long outbound_time;
 	__u64 start_time_inbound;
-	unsigned int inbound_cnt;
-	unsigned int inbound_time;
+	unsigned long inbound_cnt;
+	unsigned long inbound_time;
 };
-#endif /* QDIO_PERFORMANCE_STATS */
 
 /* unlikely as the later the better */
 #define SYNC_MEMORY if (unlikely(q->siga_sync)) qdio_siga_sync_q(q)

drivers/s390/crypto/ap_bus.c

@@ -33,6 +33,7 @@
 #include <linux/kthread.h>
 #include <linux/mutex.h>
 #include <asm/s390_rdev.h>
+#include <asm/reset.h>
 
 #include "ap_bus.h"
 
@@ -1128,6 +1129,19 @@ static void ap_poll_thread_stop(void)
 	mutex_unlock(&ap_poll_thread_mutex);
 }
 
+static void ap_reset(void)
+{
+	int i, j;
+
+	for (i = 0; i < AP_DOMAINS; i++)
+		for (j = 0; j < AP_DEVICES; j++)
+			ap_reset_queue(AP_MKQID(j, i));
+}
+
+static struct reset_call ap_reset_call = {
+	.fn = ap_reset,
+};
+
 /**
  * The module initialization code.
  */
@@ -1144,6 +1158,7 @@ int __init ap_module_init(void)
 		printk(KERN_WARNING "AP instructions not installed.\n");
 		return -ENODEV;
 	}
+	register_reset_call(&ap_reset_call);
 
 	/* Create /sys/bus/ap. */
 	rc = bus_register(&ap_bus_type);
@@ -1197,6 +1212,7 @@ out_bus:
 		bus_remove_file(&ap_bus_type, ap_bus_attrs[i]);
 	bus_unregister(&ap_bus_type);
 out:
+	unregister_reset_call(&ap_reset_call);
 	return rc;
 }
 
@@ -1227,6 +1243,7 @@ void ap_module_exit(void)
 	for (i = 0; ap_bus_attrs[i]; i++)
 		bus_remove_file(&ap_bus_type, ap_bus_attrs[i]);
 	bus_unregister(&ap_bus_type);
+	unregister_reset_call(&ap_reset_call);
 }
 
 #ifndef CONFIG_ZCRYPT_MONOLITHIC

include/asm-s390/dasd.h

@@ -69,11 +69,13 @@ typedef struct dasd_information2_t {
  * 0x01: readonly (ro)
  * 0x02: use diag discipline (diag)
  * 0x04: set the device initially online (internal use only)
+ * 0x08: enable ERP related logging
  */
 #define DASD_FEATURE_DEFAULT	     0x00
 #define DASD_FEATURE_READONLY	     0x01
 #define DASD_FEATURE_USEDIAG	     0x02
 #define DASD_FEATURE_INITIAL_ONLINE  0x04
+#define DASD_FEATURE_ERPLOG	     0x08
 
 #define DASD_PARTN_BITS 2
 

include/asm-s390/page.h

@@ -127,6 +127,26 @@ page_get_storage_key(unsigned long addr)
 	return skey;
 }
 
+extern unsigned long max_pfn;
+
+static inline int pfn_valid(unsigned long pfn)
+{
+	unsigned long dummy;
+	int ccode;
+
+	if (pfn >= max_pfn)
+		return 0;
+
+	asm volatile(
+		"	lra	%0,0(%2)\n"
+		"	ipm	%1\n"
+		"	srl	%1,28\n"
+		: "=d" (dummy), "=d" (ccode)
+		: "a" (pfn << PAGE_SHIFT)
+		: "cc");
+	return !ccode;
+}
+
 #endif /* !__ASSEMBLY__ */
 
 /* to align the pointer to the (next) page boundary */
@@ -138,8 +158,6 @@ page_get_storage_key(unsigned long addr)
 #define __va(x)                 (void *)(unsigned long)(x)
 #define virt_to_page(kaddr)	pfn_to_page(__pa(kaddr) >> PAGE_SHIFT)
 #define page_to_phys(page)	(page_to_pfn(page) << PAGE_SHIFT)
-
-#define pfn_valid(pfn)		((pfn) < max_mapnr)
 #define virt_addr_valid(kaddr)	pfn_valid(__pa(kaddr) >> PAGE_SHIFT)
 
 #define VM_DATA_DEFAULT_FLAGS	(VM_READ | VM_WRITE | VM_EXEC | \

include/asm-s390/pgalloc.h

@@ -25,8 +25,11 @@ extern void diag10(unsigned long addr);
  * Page allocation orders.
  */
 #ifndef __s390x__
+# define PTE_ALLOC_ORDER	0
+# define PMD_ALLOC_ORDER	0
 # define PGD_ALLOC_ORDER	1
 #else /* __s390x__ */
+# define PTE_ALLOC_ORDER	0
 # define PMD_ALLOC_ORDER	2
 # define PGD_ALLOC_ORDER	2
 #endif /* __s390x__ */

include/asm-s390/pgtable.h

@@ -107,23 +107,25 @@ extern char empty_zero_page[PAGE_SIZE];
  * The vmalloc() routines leaves a hole of 4kB between each vmalloced
  * area for the same reason. ;)
  */
+extern unsigned long vmalloc_end;
 #define VMALLOC_OFFSET  (8*1024*1024)
 #define VMALLOC_START   (((unsigned long) high_memory + VMALLOC_OFFSET) \
 			 & ~(VMALLOC_OFFSET-1))
+#define VMALLOC_END	vmalloc_end
 
 /*
  * We need some free virtual space to be able to do vmalloc.
  * VMALLOC_MIN_SIZE defines the minimum size of the vmalloc
  * area. On a machine with 2GB memory we make sure that we
  * have at least 128MB free space for vmalloc. On a machine
- * with 4TB we make sure we have at least 1GB.
+ * with 4TB we make sure we have at least 128GB.
  */
 #ifndef __s390x__
 #define VMALLOC_MIN_SIZE	0x8000000UL
-#define VMALLOC_END		0x80000000UL
+#define VMALLOC_END_INIT	0x80000000UL
 #else /* __s390x__ */
-#define VMALLOC_MIN_SIZE	0x40000000UL
-#define VMALLOC_END		0x40000000000UL
+#define VMALLOC_MIN_SIZE	0x2000000000UL
+#define VMALLOC_END_INIT	0x40000000000UL
 #endif /* __s390x__ */
 
 /*
@@ -815,11 +817,17 @@ static inline pte_t mk_swap_pte(unsigned long type, unsigned long offset)
 
 #define kern_addr_valid(addr)   (1)
 
+extern int add_shared_memory(unsigned long start, unsigned long size);
+extern int remove_shared_memory(unsigned long start, unsigned long size);
+
 /*
  * No page table caches to initialise
  */
 #define pgtable_cache_init()	do { } while (0)
 
+#define __HAVE_ARCH_MEMMAP_INIT
+extern void memmap_init(unsigned long, int, unsigned long, unsigned long);
+
 #define __HAVE_ARCH_PTEP_ESTABLISH
 #define __HAVE_ARCH_PTEP_SET_ACCESS_FLAGS
 #define __HAVE_ARCH_PTEP_TEST_AND_CLEAR_YOUNG