[IA64] add hpzx1_swiotlb machine-vector

I was in need of a 3.3V/dual-voltage-capable PCI sound-card and, as
luck would have it, the only card of that sort in the local computer
store was one that _still_ has a DMA engine that cannot even DMA to
all 32 bits (it's limited to 28 bits).  Hard to believe, but true (the
card in question is a "SoundBlaster Live! 24-bit" with a CA0106 chip;
stay away from that one if you can...).

Anyhow, since I don't like it when PCI cards don't work in my machine,
I created the attached patch which adds a new machine-vector to enable
support of such broken cards.

With the patch applied, you can either configure the kernel for
"HP-zx1/sx1000+swiotlb" or configure for "generic" and boot with
option "machvec=hpzx1_swiotlb" to enable support for broken PCI
devices.

The patch works as follows: the new machvec implements a I/O MMU which
will use the hardware I/O MMU whenever possible but fall back on the
software I/O TLB when encountering a device that can't be supported by
the hardware I/O MMU.  Fortunately, we don't have to mess with
MAX_DMA_ADDRESS or create a new zone: the software I/O TLB allocates
its memory as low as possible and early in the boot-process, so on any
machine with low memory, we're pretty much guaranteed that we'll get a
reasonable amount of low memory, which is all we need to properly support
broken PCI cards.

Note that I made a small change to swiotlb.c: I added a
swiotlb_init_with_default_size() function to let the new I/O MMU
initialize the software I/O TLB with less than 64MB (which is way too
much for the limited uses we'll see for the broken PCI devices; for
example, the CA0106 chip allocates only coherent buffers of about
128KB).

The patch has been tested on a zx1 machine in the generic, hpzx1, and
hpzx1_swiotlb configuration.

Thanks to Alex Williamson for the suggestion of doing this via a
completely separate machvec.
Signed-off-by: David Mosberger-Tang <davidm@hpl.hp.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>

Documentation/kernel-parameters.txt

@@ -673,7 +673,11 @@ running once the system is up.
 
 	mac53c9x=	[HW,SCSI]
 			Format: <num_esps>,<disconnect>,<nosync>,<can_queue>,<cmd_per_lun>,<sg_tablesize>,<hostid>,<use_tags>
-	
+
+	machvec=	[IA64]
+			Force the use of a particular machine-vector (machvec) in a generic
+			kernel.  Example: machvec=hpzx1_swiotlb
+
 	mad16=		[HW,OSS]
 			Format: <io>,<irq>,<dma>,<dma16>,<mpu_io>,<mpu_irq>,<joystick>
 

arch/ia64/Kconfig

@@ -61,11 +61,12 @@ config IA64_GENERIC
 	  will run on any supported IA-64 system.  However, if you configure
 	  a kernel for your specific system, it will be faster and smaller.
 
-	  generic	 For any supported IA-64 system
-	  DIG-compliant	 For DIG ("Developer's Interface Guide") compliant systems
-	  HP-zx1/sx1000	 For HP systems
-	  SGI-SN2	 For SGI Altix systems
-	  Ski-simulator  For the HP simulator <http://www.hpl.hp.com/research/linux/ski/>
+	  generic		For any supported IA-64 system
+	  DIG-compliant		For DIG ("Developer's Interface Guide") compliant systems
+	  HP-zx1/sx1000		For HP systems
+	  HP-zx1/sx1000+swiotlb	For HP systems with (broken) DMA-constrained devices.
+	  SGI-SN2		For SGI Altix systems
+	  Ski-simulator		For the HP simulator <http://www.hpl.hp.com/research/linux/ski/>
 
 	  If you don't know what to do, choose "generic".
 
@@ -78,6 +79,15 @@ config IA64_HP_ZX1
 	  Build a kernel that runs on HP zx1 and sx1000 systems.  This adds
 	  support for the HP I/O MMU.
 
+config IA64_HP_ZX1_SWIOTLB
+	bool "HP-zx1/sx1000 with software I/O TLB"
+	help
+	  Build a kernel that runs on HP zx1 and sx1000 systems even when they
+	  have broken PCI devices which cannot DMA to full 32 bits.  Apart
+	  from support for the HP I/O MMU, this includes support for the software
+	  I/O TLB, which allows supporting the broken devices at the expense of
+	  wasting some kernel memory (about 2MB by default).
+
 config IA64_SGI_SN2
 	bool "SGI-SN2"
 	help
@@ -184,7 +194,7 @@ config VIRTUAL_MEM_MAP
 
 config DISCONTIGMEM
 	bool "Discontiguous memory support"
-	depends on (IA64_DIG || IA64_SGI_SN2 || IA64_GENERIC || IA64_HP_ZX1) && NUMA && VIRTUAL_MEM_MAP
+	depends on (IA64_DIG || IA64_SGI_SN2 || IA64_GENERIC || IA64_HP_ZX1 || IA64_HP_ZX1_SWIOTLB) && NUMA && VIRTUAL_MEM_MAP
 	default y if (IA64_SGI_SN2 || IA64_GENERIC) && NUMA
 	help
 	  Say Y to support efficient handling of discontiguous physical memory,
@@ -321,7 +331,7 @@ menu "Power management and ACPI"
 
 config PM
 	bool "Power Management support"
-	depends on IA64_GENERIC || IA64_DIG || IA64_HP_ZX1
+	depends on IA64_GENERIC || IA64_DIG || IA64_HP_ZX1 || IA64_HP_ZX1_SWIOTLB
 	default y
 	help
 	  "Power Management" means that parts of your computer are shut

arch/ia64/Kconfig.debug

@@ -16,7 +16,7 @@ config IA64_GRANULE_16MB
 
 config IA64_GRANULE_64MB
 	bool "64MB"
-	depends on !(IA64_GENERIC || IA64_HP_ZX1 || IA64_SGI_SN2)
+	depends on !(IA64_GENERIC || IA64_HP_ZX1 || IA64_HP_ZX1_SWIOTLB || IA64_SGI_SN2)
 
 endchoice
 

arch/ia64/Makefile

@@ -57,11 +57,13 @@ core-$(CONFIG_IA32_SUPPORT)	+= arch/ia64/ia32/
 core-$(CONFIG_IA64_DIG) 	+= arch/ia64/dig/
 core-$(CONFIG_IA64_GENERIC) 	+= arch/ia64/dig/
 core-$(CONFIG_IA64_HP_ZX1)	+= arch/ia64/dig/
+core-$(CONFIG_IA64_HP_ZX1_SWIOTLB) += arch/ia64/dig/
 core-$(CONFIG_IA64_SGI_SN2)	+= arch/ia64/sn/
 
 drivers-$(CONFIG_PCI)		+= arch/ia64/pci/
 drivers-$(CONFIG_IA64_HP_SIM)	+= arch/ia64/hp/sim/
 drivers-$(CONFIG_IA64_HP_ZX1)	+= arch/ia64/hp/common/ arch/ia64/hp/zx1/
+drivers-$(CONFIG_IA64_HP_ZX1_SWIOTLB) += arch/ia64/hp/common/ arch/ia64/hp/zx1/
 drivers-$(CONFIG_IA64_GENERIC)	+= arch/ia64/hp/common/ arch/ia64/hp/zx1/ arch/ia64/hp/sim/ arch/ia64/sn/
 drivers-$(CONFIG_OPROFILE)	+= arch/ia64/oprofile/
 

arch/ia64/hp/common/Makefile

@@ -6,3 +6,5 @@
 #
 
 obj-y := sba_iommu.o
+obj-$(CONFIG_IA64_HP_ZX1_SWIOTLB) += hwsw_iommu.o
+obj-$(CONFIG_IA64_GENERIC) += hwsw_iommu.o

arch/ia64/hp/zx1/Makefile

@@ -5,4 +5,4 @@
 # Copyright (C) Alex Williamson (alex_williamson@hp.com)
 #
 
-obj-$(CONFIG_IA64_GENERIC) += hpzx1_machvec.o
+obj-$(CONFIG_IA64_GENERIC) += hpzx1_machvec.o hpzx1_swiotlb_machvec.o

arch/ia64/kernel/Makefile

@@ -12,6 +12,7 @@ obj-y := acpi.o entry.o efi.o efi_stub.o gate-data.o fsys.o ia64_ksyms.o irq.o i
 obj-$(CONFIG_IA64_BRL_EMU)	+= brl_emu.o
 obj-$(CONFIG_IA64_GENERIC)	+= acpi-ext.o
 obj-$(CONFIG_IA64_HP_ZX1)	+= acpi-ext.o
+obj-$(CONFIG_IA64_HP_ZX1_SWIOTLB) += acpi-ext.o
 obj-$(CONFIG_IA64_PALINFO)	+= palinfo.o
 obj-$(CONFIG_IOSAPIC)		+= iosapic.o
 obj-$(CONFIG_MODULES)		+= module.o

arch/ia64/kernel/acpi.c

@@ -113,6 +113,8 @@ acpi_get_sysname (void)
 	return "hpsim";
 # elif defined (CONFIG_IA64_HP_ZX1)
 	return "hpzx1";
+# elif defined (CONFIG_IA64_HP_ZX1_SWIOTLB)
+	return "hpzx1_swiotlb";
 # elif defined (CONFIG_IA64_SGI_SN2)
 	return "sn2";
 # elif defined (CONFIG_IA64_DIG)

arch/ia64/kernel/setup.c

 /*
  * Architecture-specific setup.
  *
- * Copyright (C) 1998-2001, 2003 Hewlett-Packard Co
+ * Copyright (C) 1998-2001, 2003-2004 Hewlett-Packard Co
  *	David Mosberger-Tang <davidm@hpl.hp.com>
  *	Stephane Eranian <eranian@hpl.hp.com>
  * Copyright (C) 2000, Rohit Seth <rohit.seth@intel.com>
@@ -312,7 +312,28 @@ setup_arch (char **cmdline_p)
 	io_port_init();
 
 #ifdef CONFIG_IA64_GENERIC
-	machvec_init(acpi_get_sysname());
+	{
+		const char *mvec_name = strstr (*cmdline_p, "machvec=");
+		char str[64];
+
+		if (mvec_name) {
+			const char *end;
+			size_t len;
+
+			mvec_name += 8;
+			end = strchr (mvec_name, ' ');
+			if (end)
+				len = end - mvec_name;
+			else
+				len = strlen (mvec_name);
+			len = min(len, sizeof (str) - 1);
+			strncpy (str, mvec_name, len);
+			str[len] = '\0';
+			mvec_name = str;
+		} else
+			mvec_name = acpi_get_sysname();
+		machvec_init(mvec_name);
+	}
 #endif
 
 	if (early_console_setup(*cmdline_p) == 0)

arch/ia64/lib/swiotlb.c

@@ -61,9 +61,8 @@ static char *io_tlb_start, *io_tlb_end;
 /*
  * The number of IO TLB blocks (in groups of 64) betweeen io_tlb_start and
  * io_tlb_end.  This is command line adjustable via setup_io_tlb_npages.
- * Default to 64MB.
  */
-static unsigned long io_tlb_nslabs = 32768;
+static unsigned long io_tlb_nslabs;
 
 /*
  * When the IOMMU overflows we return a fallback buffer. This sets the size.
@@ -113,10 +112,15 @@ __setup("swiotlb=", setup_io_tlb_npages);
  * structures for the software IO TLB used to implement the PCI DMA API.
  */
 void
-swiotlb_init(void)
+swiotlb_init_with_default_size (size_t default_size)
 {
 	unsigned long i;
 
+	if (!io_tlb_nslabs) {
+		io_tlb_nslabs = (default_size >> PAGE_SHIFT);
+		io_tlb_nslabs = ALIGN(io_tlb_nslabs, IO_TLB_SEGSIZE);
+	}
+
 	/*
 	 * Get IO TLB memory from the low pages
 	 */
@@ -145,6 +149,12 @@ swiotlb_init(void)
 	       virt_to_phys(io_tlb_start), virt_to_phys(io_tlb_end));
 }
 
+void
+swiotlb_init (void)
+{
+	swiotlb_init_with_default_size(64 * (1<<20));	/* default to 64MB */
+}
+
 static inline int
 address_needs_mapping(struct device *hwdev, dma_addr_t addr)
 {

drivers/char/agp/Kconfig

@@ -144,7 +144,7 @@ config AGP_I460
 
 config AGP_HP_ZX1
 	tristate "HP ZX1 chipset AGP support"
-	depends on AGP && (IA64_HP_ZX1 || IA64_GENERIC)
+	depends on AGP && (IA64_HP_ZX1 || IA64_HP_ZX1_SWIOTLB || IA64_GENERIC)
 	help
 	  This option gives you AGP GART support for the HP ZX1 chipset
 	  for IA64 processors.

include/asm-ia64/machvec.h

@@ -94,6 +94,8 @@ extern void machvec_tlb_migrate_finish (struct mm_struct *);
 #  include <asm/machvec_dig.h>
 # elif defined (CONFIG_IA64_HP_ZX1)
 #  include <asm/machvec_hpzx1.h>
+# elif defined (CONFIG_IA64_HP_ZX1_SWIOTLB)
+#  include <asm/machvec_hpzx1_swiotlb.h>
 # elif defined (CONFIG_IA64_SGI_SN2)
 #  include <asm/machvec_sn2.h>
 # elif defined (CONFIG_IA64_GENERIC)

include/asm-ia64/numnodes.h

@@ -4,7 +4,7 @@
 #ifdef CONFIG_IA64_DIG
 /* Max 8 Nodes */
 #define NODES_SHIFT	3
-#elif defined(CONFIG_IA64_HP_ZX1)
+#elif defined(CONFIG_IA64_HP_ZX1) || defined(CONFIG_IA64_HP_ZX1_SWIOTLB)
 /* Max 32 Nodes */
 #define NODES_SHIFT	5
 #elif defined(CONFIG_IA64_SGI_SN2) || defined(CONFIG_IA64_GENERIC)