[PATCH] ppc64: Abolish ioremap_mm

Currently ppc64 has two mm_structs for the kernel, init_mm and also
ioremap_mm.  The latter really isn't necessary: this patch abolishes it,
instead restricting vmallocs to the lower 1TB of the init_mm's range and
placing io mappings in the upper 1TB.  This simplifies the code in a number
of places and eliminates an unecessary set of pagetables.  It also tweaks
the unmap/free path a little, allowing us to remove the unmap_im_area() set
of page table walkers, replacing them with unmap_vm_area().

Signed-off-by: David Gibson <dwg@au1.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

arch/ppc64/kernel/eeh.c

@@ -505,7 +505,7 @@ static inline unsigned long eeh_token_to_phys(unsigned long token)
 	pte_t *ptep;
 	unsigned long pa;
 
-	ptep = find_linux_pte(ioremap_mm.pgd, token);
+	ptep = find_linux_pte(init_mm.pgd, token);
 	if (!ptep)
 		return token;
 	pa = pte_pfn(*ptep) << PAGE_SHIFT;

arch/ppc64/kernel/head.S

@@ -2121,10 +2121,6 @@ empty_zero_page:
 swapper_pg_dir:
 	.space	4096
 
-	.globl	ioremap_dir
-ioremap_dir:
-	.space	4096
-
 #ifdef CONFIG_SMP
 /* 1 page segment table per cpu (max 48, cpu0 allocated at STAB0_PHYS_ADDR) */
 	.globl	stab_array

arch/ppc64/kernel/process.c

@@ -58,14 +58,6 @@ struct task_struct *last_task_used_math = NULL;
 struct task_struct *last_task_used_altivec = NULL;
 #endif
 
-struct mm_struct ioremap_mm = {
-	.pgd		= ioremap_dir,
-	.mm_users	= ATOMIC_INIT(2),
-	.mm_count	= ATOMIC_INIT(1),
-	.cpu_vm_mask	= CPU_MASK_ALL,
-	.page_table_lock = SPIN_LOCK_UNLOCKED,
-};
-
 /*
  * Make sure the floating-point register state in the
  * the thread_struct is up to date for task tsk.

arch/ppc64/mm/hash_utils.c

@@ -310,10 +310,6 @@ int hash_page(unsigned long ea, unsigned long access, unsigned long trap)
 
 		vsid = get_vsid(mm->context.id, ea);
 		break;
-	case IO_REGION_ID:
-		mm = &ioremap_mm;
-		vsid = get_kernel_vsid(ea);
-		break;
 	case VMALLOC_REGION_ID:
 		mm = &init_mm;
 		vsid = get_kernel_vsid(ea);

arch/ppc64/mm/imalloc.c

@@ -15,6 +15,7 @@
 #include <asm/pgtable.h>
 #include <asm/semaphore.h>
 #include <asm/imalloc.h>
+#include <asm/cacheflush.h>
 
 static DECLARE_MUTEX(imlist_sem);
 struct vm_struct * imlist = NULL;
@@ -285,29 +286,32 @@ struct vm_struct * im_get_area(unsigned long v_addr, unsigned long size,
 	return area;
 }
 
-unsigned long im_free(void * addr)
+void im_free(void * addr)
 {
 	struct vm_struct **p, *tmp;
-	unsigned long ret_size = 0;
   
 	if (!addr)
-		return ret_size;
+		return;
-	if ((PAGE_SIZE-1) & (unsigned long) addr) {
+	if ((unsigned long) addr & ~PAGE_MASK) {
 		printk(KERN_ERR "Trying to %s bad address (%p)\n", __FUNCTION__,			addr);
-		return ret_size;
+		return;
 	}
 	down(&imlist_sem);
 	for (p = &imlist ; (tmp = *p) ; p = &tmp->next) {
 		if (tmp->addr == addr) {
-			ret_size = tmp->size;
 			*p = tmp->next;
+
+			/* XXX: do we need the lock? */
+			spin_lock(&init_mm.page_table_lock);
+			unmap_vm_area(tmp);
+			spin_unlock(&init_mm.page_table_lock);
+
 			kfree(tmp);
 			up(&imlist_sem);
-			return ret_size;
+			return;
 		}
 	}
 	up(&imlist_sem);
 	printk(KERN_ERR "Trying to %s nonexistent area (%p)\n", __FUNCTION__,
 			addr);
-	return ret_size;
 }

arch/ppc64/mm/init.c

@@ -73,9 +73,6 @@ static unsigned long phbs_io_bot = PHBS_IO_BASE;
 extern pgd_t swapper_pg_dir[];
 extern struct task_struct *current_set[NR_CPUS];
 
-extern pgd_t ioremap_dir[];
-pgd_t * ioremap_pgd = (pgd_t *)&ioremap_dir;
-
 unsigned long klimit = (unsigned long)_end;
 
 unsigned long _SDR1=0;
@@ -137,69 +134,6 @@ void iounmap(volatile void __iomem *addr)
 
 #else
 
-static void unmap_im_area_pte(pmd_t *pmd, unsigned long addr,
-				  unsigned long end)
-{
-	pte_t *pte;
-
-	pte = pte_offset_kernel(pmd, addr);
-	do {
-		pte_t ptent = ptep_get_and_clear(&ioremap_mm, addr, pte);
-		WARN_ON(!pte_none(ptent) && !pte_present(ptent));
-	} while (pte++, addr += PAGE_SIZE, addr != end);
-}
-
-static inline void unmap_im_area_pmd(pud_t *pud, unsigned long addr,
-				     unsigned long end)
-{
-	pmd_t *pmd;
-	unsigned long next;
-
-	pmd = pmd_offset(pud, addr);
-	do {
-		next = pmd_addr_end(addr, end);
-		if (pmd_none_or_clear_bad(pmd))
-			continue;
-		unmap_im_area_pte(pmd, addr, next);
-	} while (pmd++, addr = next, addr != end);
-}
-
-static inline void unmap_im_area_pud(pgd_t *pgd, unsigned long addr,
-				     unsigned long end)
-{
-	pud_t *pud;
-	unsigned long next;
-
-	pud = pud_offset(pgd, addr);
-	do {
-		next = pud_addr_end(addr, end);
-		if (pud_none_or_clear_bad(pud))
-			continue;
-		unmap_im_area_pmd(pud, addr, next);
-	} while (pud++, addr = next, addr != end);
-}
-
-static void unmap_im_area(unsigned long addr, unsigned long end)
-{
-	struct mm_struct *mm = &ioremap_mm;
-	unsigned long next;
-	pgd_t *pgd;
-
-	spin_lock(&mm->page_table_lock);
-
-	pgd = pgd_offset_i(addr);
-	flush_cache_vunmap(addr, end);
-	do {
-		next = pgd_addr_end(addr, end);
-		if (pgd_none_or_clear_bad(pgd))
-			continue;
-		unmap_im_area_pud(pgd, addr, next);
-	} while (pgd++, addr = next, addr != end);
-	flush_tlb_kernel_range(start, end);
-
-	spin_unlock(&mm->page_table_lock);
-}
-
 /*
  * map_io_page currently only called by __ioremap
  * map_io_page adds an entry to the ioremap page table
@@ -214,21 +148,21 @@ static int map_io_page(unsigned long ea, unsigned long pa, int flags)
 	unsigned long vsid;
 
 	if (mem_init_done) {
-		spin_lock(&ioremap_mm.page_table_lock);
+		spin_lock(&init_mm.page_table_lock);
-		pgdp = pgd_offset_i(ea);
+		pgdp = pgd_offset_k(ea);
-		pudp = pud_alloc(&ioremap_mm, pgdp, ea);
+		pudp = pud_alloc(&init_mm, pgdp, ea);
 		if (!pudp)
 			return -ENOMEM;
-		pmdp = pmd_alloc(&ioremap_mm, pudp, ea);
+		pmdp = pmd_alloc(&init_mm, pudp, ea);
 		if (!pmdp)
 			return -ENOMEM;
-		ptep = pte_alloc_kernel(&ioremap_mm, pmdp, ea);
+		ptep = pte_alloc_kernel(&init_mm, pmdp, ea);
 		if (!ptep)
 			return -ENOMEM;
 		pa = abs_to_phys(pa);
-		set_pte_at(&ioremap_mm, ea, ptep, pfn_pte(pa >> PAGE_SHIFT,
+		set_pte_at(&init_mm, ea, ptep, pfn_pte(pa >> PAGE_SHIFT,
 							  __pgprot(flags)));
-		spin_unlock(&ioremap_mm.page_table_lock);
+		spin_unlock(&init_mm.page_table_lock);
 	} else {
 		unsigned long va, vpn, hash, hpteg;
 
@@ -267,13 +201,9 @@ static void __iomem * __ioremap_com(unsigned long addr, unsigned long pa,
 
 	for (i = 0; i < size; i += PAGE_SIZE)
 		if (map_io_page(ea+i, pa+i, flags))
-			goto failure;
+			return NULL;
 
 	return (void __iomem *) (ea + (addr & ~PAGE_MASK));
- failure:
-	if (mem_init_done)
-		unmap_im_area(ea, ea + size);
-	return NULL;
 }
 
 
@@ -381,19 +311,14 @@ int __ioremap_explicit(unsigned long pa, unsigned long ea,
  */
 void iounmap(volatile void __iomem *token)
 {
-	unsigned long address, size;
 	void *addr;
 
 	if (!mem_init_done)
 		return;
 	
 	addr = (void *) ((unsigned long __force) token & PAGE_MASK);
-	
-	if ((size = im_free(addr)) == 0)
-		return;
 
-	address = (unsigned long)addr; 
+	im_free(addr);
-	unmap_im_area(address, address + size);
 }
 
 static int iounmap_subset_regions(unsigned long addr, unsigned long size)

include/asm-ppc64/imalloc.h

@@ -4,9 +4,9 @@
 /*
  * Define the address range of the imalloc VM area.
  */
-#define PHBS_IO_BASE  	  IOREGIONBASE
+#define PHBS_IO_BASE  	  VMALLOC_END
-#define IMALLOC_BASE      (IOREGIONBASE + 0x80000000ul)	/* Reserve 2 gigs for PHBs */
+#define IMALLOC_BASE      (PHBS_IO_BASE + 0x80000000ul)	/* Reserve 2 gigs for PHBs */
-#define IMALLOC_END       (IOREGIONBASE + EADDR_MASK)
+#define IMALLOC_END       (VMALLOC_START + EADDR_MASK)
 
 
 /* imalloc region types */
@@ -18,7 +18,9 @@
 
 extern struct vm_struct * im_get_free_area(unsigned long size);
 extern struct vm_struct * im_get_area(unsigned long v_addr, unsigned long size,
-			int region_type);
+				      int region_type);
-unsigned long im_free(void *addr);
+extern void im_free(void *addr);
+
+extern unsigned long ioremap_bot;
 
 #endif /* _PPC64_IMALLOC_H */

include/asm-ppc64/page.h

@@ -202,9 +202,7 @@ extern u64 ppc64_pft_size;		/* Log 2 of page table size */
 #define PAGE_OFFSET     ASM_CONST(0xC000000000000000)
 #define KERNELBASE      PAGE_OFFSET
 #define VMALLOCBASE     ASM_CONST(0xD000000000000000)
-#define IOREGIONBASE    ASM_CONST(0xE000000000000000)
 
-#define IO_REGION_ID       (IOREGIONBASE >> REGION_SHIFT)
 #define VMALLOC_REGION_ID  (VMALLOCBASE >> REGION_SHIFT)
 #define KERNEL_REGION_ID   (KERNELBASE >> REGION_SHIFT)
 #define USER_REGION_ID     (0UL)

include/asm-ppc64/pgtable.h

@@ -53,7 +53,8 @@
  * Define the address range of the vmalloc VM area.
  */
 #define VMALLOC_START (0xD000000000000000ul)
-#define VMALLOC_END   (VMALLOC_START + EADDR_MASK)
+#define VMALLOC_SIZE  (0x10000000000UL)
+#define VMALLOC_END   (VMALLOC_START + VMALLOC_SIZE)
 
 /*
  * Bits in a linux-style PTE.  These match the bits in the
@@ -239,9 +240,6 @@ static inline pte_t pfn_pte(unsigned long pfn, pgprot_t pgprot)
 /* This now only contains the vmalloc pages */
 #define pgd_offset_k(address) pgd_offset(&init_mm, address)
 
-/* to find an entry in the ioremap page-table-directory */
-#define pgd_offset_i(address) (ioremap_pgd + pgd_index(address))
-
 /*
  * The following only work if pte_present() is true.
  * Undefined behaviour if not..
@@ -459,15 +457,12 @@ extern pgprot_t phys_mem_access_prot(struct file *file, unsigned long addr,
 #define __HAVE_ARCH_PTE_SAME
 #define pte_same(A,B)	(((pte_val(A) ^ pte_val(B)) & ~_PAGE_HPTEFLAGS) == 0)
 
-extern unsigned long ioremap_bot, ioremap_base;
-
 #define pmd_ERROR(e) \
 	printk("%s:%d: bad pmd %08x.\n", __FILE__, __LINE__, pmd_val(e))
 #define pgd_ERROR(e) \
 	printk("%s:%d: bad pgd %08x.\n", __FILE__, __LINE__, pgd_val(e))
 
 extern pgd_t swapper_pg_dir[];
-extern pgd_t ioremap_dir[];
 
 extern void paging_init(void);
 

include/asm-ppc64/processor.h

@@ -429,16 +429,6 @@ struct thread_struct {
 	.fpexc_mode = MSR_FE0|MSR_FE1, \
 }
 
-/*
- * Note: the vm_start and vm_end fields here should *not*
- * be in kernel space.  (Could vm_end == vm_start perhaps?)
- */
-#define IOREMAP_MMAP { &ioremap_mm, 0, 0x1000, NULL, \
-		    PAGE_SHARED, VM_READ | VM_WRITE | VM_EXEC, \
-		    1, NULL, NULL }
-
-extern struct mm_struct ioremap_mm;
-
 /*
  * Return saved PC of a blocked thread. For now, this is the "user" PC
  */