Merge tag 'net-next-6.3' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net-nextgrafted

Pull networking updates from Jakub Kicinski:
 "Core:

   - Add dedicated kmem_cache for typical/small skb->head, avoid having
     to access struct page at kfree time, and improve memory use.

   - Introduce sysctl to set default RPS configuration for new netdevs.

   - Define Netlink protocol specification format which can be used to
     describe messages used by each family and auto-generate parsers.
     Add tools for generating kernel data structures and uAPI headers.

   - Expose all net/core sysctls inside netns.

   - Remove 4s sleep in netpoll if carrier is instantly detected on
     boot.

   - Add configurable limit of MDB entries per port, and port-vlan.

   - Continue populating drop reasons throughout the stack.

   - Retire a handful of legacy Qdiscs and classifiers.

  Protocols:

   - Support IPv4 big TCP (TSO frames larger than 64kB).

   - Add IP_LOCAL_PORT_RANGE socket option, to control local port range
     on socket by socket basis.

   - Track and report in procfs number of MPTCP sockets used.

   - Support mixing IPv4 and IPv6 flows in the in-kernel MPTCP path
     manager.

   - IPv6: don't check net.ipv6.route.max_size and rely on garbage
     collection to free memory (similarly to IPv4).

   - Support Penultimate Segment Pop (PSP) flavor in SRv6 (RFC8986).

   - ICMP: add per-rate limit counters.

   - Add support for user scanning requests in ieee802154.

   - Remove static WEP support.

   - Support minimal Wi-Fi 7 Extremely High Throughput (EHT) rate
     reporting.

   - WiFi 7 EHT channel puncturing support (client & AP).

  BPF:

   - Add a rbtree data structure following the "next-gen data structure"
     precedent set by recently added linked list, that is, by using
     kfunc + kptr instead of adding a new BPF map type.

   - Expose XDP hints via kfuncs with initial support for RX hash and
     timestamp metadata.

   - Add BPF_F_NO_TUNNEL_KEY extension to bpf_skb_set_tunnel_key to
     better support decap on GRE tunnel devices not operating in collect
     metadata.

   - Improve x86 JIT's codegen for PROBE_MEM runtime error checks.

   - Remove the need for trace_printk_lock for bpf_trace_printk and
     bpf_trace_vprintk helpers.

   - Extend libbpf's bpf_tracing.h support for tracing arguments of
     kprobes/uprobes and syscall as a special case.

   - Significantly reduce the search time for module symbols by
     livepatch and BPF.

   - Enable cpumasks to be used as kptrs, which is useful for tracing
     programs tracking which tasks end up running on which CPUs in
     different time intervals.

   - Add support for BPF trampoline on s390x and riscv64.

   - Add capability to export the XDP features supported by the NIC.

   - Add __bpf_kfunc tag for marking kernel functions as kfuncs.

   - Add cgroup.memory=nobpf kernel parameter option to disable BPF
     memory accounting for container environments.

  Netfilter:

   - Remove the CLUSTERIP target. It has been marked as obsolete for
     years, and we still have WARN splats wrt races of the out-of-band
     /proc interface installed by this target.

   - Add 'destroy' commands to nf_tables. They are identical to the
     existing 'delete' commands, but do not return an error if the
     referenced object (set, chain, rule...) did not exist.

  Driver API:

   - Improve cpumask_local_spread() locality to help NICs set the right
     IRQ affinity on AMD platforms.

   - Separate C22 and C45 MDIO bus transactions more clearly.

   - Introduce new DCB table to control DSCP rewrite on egress.

   - Support configuration of Physical Layer Collision Avoidance (PLCA)
     Reconciliation Sublayer (RS) (802.3cg-2019). Modern version of
     shared medium Ethernet.

   - Support for MAC Merge layer (IEEE 802.3-2018 clause 99). Allowing
     preemption of low priority frames by high priority frames.

   - Add support for controlling MACSec offload using netlink SET.

   - Rework devlink instance refcounts to allow registration and
     de-registration under the instance lock. Split the code into
     multiple files, drop some of the unnecessarily granular locks and
     factor out common parts of netlink operation handling.

   - Add TX frame aggregation parameters (for USB drivers).

   - Add a new attr TCA_EXT_WARN_MSG to report TC (offload) warning
     messages with notifications for debug.

   - Allow offloading of UDP NEW connections via act_ct.

   - Add support for per action HW stats in TC.

   - Support hardware miss to TC action (continue processing in SW from
     a specific point in the action chain).

   - Warn if old Wireless Extension user space interface is used with
     modern cfg80211/mac80211 drivers. Do not support Wireless
     Extensions for Wi-Fi 7 devices at all. Everyone should switch to
     using nl80211 interface instead.

   - Improve the CAN bit timing configuration. Use extack to return
     error messages directly to user space, update the SJW handling,
     including the definition of a new default value that will benefit
     CAN-FD controllers, by increasing their oscillator tolerance.

  New hardware / drivers:

   - Ethernet:
      - nVidia BlueField-3 support (control traffic driver)
      - Ethernet support for imx93 SoCs
      - Motorcomm yt8531 gigabit Ethernet PHY
      - onsemi NCN26000 10BASE-T1S PHY (with support for PLCA)
      - Microchip LAN8841 PHY (incl. cable diagnostics and PTP)
      - Amlogic gxl MDIO mux

   - WiFi:
      - RealTek RTL8188EU (rtl8xxxu)
      - Qualcomm Wi-Fi 7 devices (ath12k)

   - CAN:
      - Renesas R-Car V4H

  Drivers:

   - Bluetooth:
      - Set Per Platform Antenna Gain (PPAG) for Intel controllers.

   - Ethernet NICs:
      - Intel (1G, igc):
         - support TSN / Qbv / packet scheduling features of i226 model
      - Intel (100G, ice):
         - use GNSS subsystem instead of TTY
         - multi-buffer XDP support
         - extend support for GPIO pins to E823 devices
      - nVidia/Mellanox:
         - update the shared buffer configuration on PFC commands
         - implement PTP adjphase function for HW offset control
         - TC support for Geneve and GRE with VF tunnel offload
         - more efficient crypto key management method
         - multi-port eswitch support
      - Netronome/Corigine:
         - add DCB IEEE support
         - support IPsec offloading for NFP3800
      - Freescale/NXP (enetc):
         - support XDP_REDIRECT for XDP non-linear buffers
         - improve reconfig, avoid link flap and waiting for idle
         - support MAC Merge layer
      - Other NICs:
         - sfc/ef100: add basic devlink support for ef100
         - ionic: rx_push mode operation (writing descriptors via MMIO)
         - bnxt: use the auxiliary bus abstraction for RDMA
         - r8169: disable ASPM and reset bus in case of tx timeout
         - cpsw: support QSGMII mode for J721e CPSW9G
         - cpts: support pulse-per-second output
         - ngbe: add an mdio bus driver
         - usbnet: optimize usbnet_bh() by avoiding unnecessary queuing
         - r8152: handle devices with FW with NCM support
         - amd-xgbe: support 10Mbps, 2.5GbE speeds and rx-adaptation
         - virtio-net: support multi buffer XDP
         - virtio/vsock: replace virtio_vsock_pkt with sk_buff
         - tsnep: XDP support

   - Ethernet high-speed switches:
      - nVidia/Mellanox (mlxsw):
         - add support for latency TLV (in FW control messages)
      - Microchip (sparx5):
         - separate explicit and implicit traffic forwarding rules, make
           the implicit rules always active
         - add support for egress DSCP rewrite
         - IS0 VCAP support (Ingress Classification)
         - IS2 VCAP filters (protos, L3 addrs, L4 ports, flags, ToS
           etc.)
         - ES2 VCAP support (Egress Access Control)
         - support for Per-Stream Filtering and Policing (802.1Q,
           8.6.5.1)

   - Ethernet embedded switches:
      - Marvell (mv88e6xxx):
         - add MAB (port auth) offload support
         - enable PTP receive for mv88e6390
      - NXP (ocelot):
         - support MAC Merge layer
         - support for the the vsc7512 internal copper phys
      - Microchip:
         - lan9303: convert to PHYLINK
         - lan966x: support TC flower filter statistics
         - lan937x: PTP support for KSZ9563/KSZ8563 and LAN937x
         - lan937x: support Credit Based Shaper configuration
         - ksz9477: support Energy Efficient Ethernet
      - other:
         - qca8k: convert to regmap read/write API, use bulk operations
         - rswitch: Improve TX timestamp accuracy

   - Intel WiFi (iwlwifi):
      - EHT (Wi-Fi 7) rate reporting
      - STEP equalizer support: transfer some STEP (connection to radio
        on platforms with integrated wifi) related parameters from the
        BIOS to the firmware.

   - Qualcomm 802.11ax WiFi (ath11k):
      - IPQ5018 support
      - Fine Timing Measurement (FTM) responder role support
      - channel 177 support

   - MediaTek WiFi (mt76):
      - per-PHY LED support
      - mt7996: EHT (Wi-Fi 7) support
      - Wireless Ethernet Dispatch (WED) reset support
      - switch to using page pool allocator

   - RealTek WiFi (rtw89):
      - support new version of Bluetooth co-existance

   - Mobile:
      - rmnet: support TX aggregation"

* tag 'net-next-6.3' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net-next: (1872 commits)
  page_pool: add a comment explaining the fragment counter usage
  net: ethtool: fix __ethtool_dev_mm_supported() implementation
  ethtool: pse-pd: Fix double word in comments
  xsk: add linux/vmalloc.h to xsk.c
  sefltests: netdevsim: wait for devlink instance after netns removal
  selftest: fib_tests: Always cleanup before exit
  net/mlx5e: Align IPsec ASO result memory to be as required by hardware
  net/mlx5e: TC, Set CT miss to the specific ct action instance
  net/mlx5e: Rename CHAIN_TO_REG to MAPPED_OBJ_TO_REG
  net/mlx5: Refactor tc miss handling to a single function
  net/mlx5: Kconfig: Make tc offload depend on tc skb extension
  net/sched: flower: Support hardware miss to tc action
  net/sched: flower: Move filter handle initialization earlier
  net/sched: cls_api: Support hardware miss to tc action
  net/sched: Rename user cookie and act cookie
  sfc: fix builds without CONFIG_RTC_LIB
  sfc: clean up some inconsistent indentings
  net/mlx4_en: Introduce flexible array to silence overflow warning
  net: lan966x: Fix possible deadlock inside PTP
  net/ulp: Remove redundant ->clone() test in inet_clone_ulp().
  ...

1 files changed, 802 insertions, 0 deletions

diff --git a/arch/alpha/kernel/core_titan.c b/arch/alpha/kernel/core_titan.c
new file mode 100644
index 000000000..77f5d68ed
--- /dev/null
+++ b/arch/alpha/kernel/core_titan.c
@@ -0,0 +1,802 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ *	linux/arch/alpha/kernel/core_titan.c
+ *
+ * Code common to all TITAN core logic chips.
+ */
+
+#define __EXTERN_INLINE inline
+#include <asm/io.h>
+#include <asm/core_titan.h>
+#undef __EXTERN_INLINE
+
+#include <linux/module.h>
+#include <linux/types.h>
+#include <linux/pci.h>
+#include <linux/sched.h>
+#include <linux/init.h>
+#include <linux/vmalloc.h>
+#include <linux/memblock.h>
+
+#include <asm/ptrace.h>
+#include <asm/smp.h>
+#include <asm/tlbflush.h>
+#include <asm/vga.h>
+
+#include "proto.h"
+#include "pci_impl.h"
+
+/* Save Titan configuration data as the console had it set up.  */
+
+struct
+{
+	unsigned long wsba[4];
+	unsigned long wsm[4];
+	unsigned long tba[4];
+} saved_config[4] __attribute__((common));
+
+/*
+ * Is PChip 1 present? No need to query it more than once.
+ */
+static int titan_pchip1_present;
+
+/*
+ * BIOS32-style PCI interface:
+ */
+
+#define DEBUG_CONFIG 0
+
+#if DEBUG_CONFIG
+# define DBG_CFG(args)	printk args
+#else
+# define DBG_CFG(args)
+#endif
+
+
+/*
+ * Routines to access TIG registers.
+ */
+static inline volatile unsigned long *
+mk_tig_addr(int offset)
+{
+	return (volatile unsigned long *)(TITAN_TIG_SPACE + (offset << 6));
+}
+
+static inline u8 
+titan_read_tig(int offset, u8 value)
+{
+	volatile unsigned long *tig_addr = mk_tig_addr(offset);
+	return (u8)(*tig_addr & 0xff);
+}
+
+static inline void 
+titan_write_tig(int offset, u8 value)
+{
+	volatile unsigned long *tig_addr = mk_tig_addr(offset);
+	*tig_addr = (unsigned long)value;
+}
+
+
+/*
+ * Given a bus, device, and function number, compute resulting
+ * configuration space address
+ * accordingly.  It is therefore not safe to have concurrent
+ * invocations to configuration space access routines, but there
+ * really shouldn't be any need for this.
+ *
+ * Note that all config space accesses use Type 1 address format.
+ *
+ * Note also that type 1 is determined by non-zero bus number.
+ *
+ * Type 1:
+ *
+ *  3 3|3 3 2 2|2 2 2 2|2 2 2 2|1 1 1 1|1 1 1 1|1 1 
+ *  3 2|1 0 9 8|7 6 5 4|3 2 1 0|9 8 7 6|5 4 3 2|1 0 9 8|7 6 5 4|3 2 1 0
+ * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+ * | | | | | | | | | | |B|B|B|B|B|B|B|B|D|D|D|D|D|F|F|F|R|R|R|R|R|R|0|1|
+ * +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+ *
+ *	31:24	reserved
+ *	23:16	bus number (8 bits = 128 possible buses)
+ *	15:11	Device number (5 bits)
+ *	10:8	function number
+ *	 7:2	register number
+ *  
+ * Notes:
+ *	The function number selects which function of a multi-function device 
+ *	(e.g., SCSI and Ethernet).
+ * 
+ *	The register selects a DWORD (32 bit) register offset.  Hence it
+ *	doesn't get shifted by 2 bits as we want to "drop" the bottom two
+ *	bits.
+ */
+
+static int
+mk_conf_addr(struct pci_bus *pbus, unsigned int device_fn, int where,
+	     unsigned long *pci_addr, unsigned char *type1)
+{
+	struct pci_controller *hose = pbus->sysdata;
+	unsigned long addr;
+	u8 bus = pbus->number;
+
+	DBG_CFG(("mk_conf_addr(bus=%d ,device_fn=0x%x, where=0x%x, "
+		 "pci_addr=0x%p, type1=0x%p)\n",
+		 bus, device_fn, where, pci_addr, type1));
+
+	if (!pbus->parent) /* No parent means peer PCI bus. */
+		bus = 0;
+        *type1 = (bus != 0);
+
+        addr = (bus << 16) | (device_fn << 8) | where;
+	addr |= hose->config_space_base;
+		
+	*pci_addr = addr;
+	DBG_CFG(("mk_conf_addr: returning pci_addr 0x%lx\n", addr));
+	return 0;
+}
+
+static int
+titan_read_config(struct pci_bus *bus, unsigned int devfn, int where,
+		  int size, u32 *value)
+{
+	unsigned long addr;
+	unsigned char type1;
+
+	if (mk_conf_addr(bus, devfn, where, &addr, &type1))
+		return PCIBIOS_DEVICE_NOT_FOUND;
+
+	switch (size) {
+	case 1:
+		*value = __kernel_ldbu(*(vucp)addr);
+		break;
+	case 2:
+		*value = __kernel_ldwu(*(vusp)addr);
+		break;
+	case 4:
+		*value = *(vuip)addr;
+		break;
+	}
+
+	return PCIBIOS_SUCCESSFUL;
+}
+
+static int 
+titan_write_config(struct pci_bus *bus, unsigned int devfn, int where,
+		   int size, u32 value)
+{
+	unsigned long addr;
+	unsigned char type1;
+
+	if (mk_conf_addr(bus, devfn, where, &addr, &type1))
+		return PCIBIOS_DEVICE_NOT_FOUND;
+
+	switch (size) {
+	case 1:
+		__kernel_stb(value, *(vucp)addr);
+		mb();
+		__kernel_ldbu(*(vucp)addr);
+		break;
+	case 2:
+		__kernel_stw(value, *(vusp)addr);
+		mb();
+		__kernel_ldwu(*(vusp)addr);
+		break;
+	case 4:
+		*(vuip)addr = value;
+		mb();
+		*(vuip)addr;
+		break;
+	}
+
+	return PCIBIOS_SUCCESSFUL;
+}
+
+struct pci_ops titan_pci_ops = 
+{
+	.read =		titan_read_config,
+	.write =	titan_write_config,
+};
+
+
+void
+titan_pci_tbi(struct pci_controller *hose, dma_addr_t start, dma_addr_t end)
+{
+	titan_pachip *pachip = 
+	  (hose->index & 1) ? TITAN_pachip1 : TITAN_pachip0;
+	titan_pachip_port *port;
+	volatile unsigned long *csr;
+	unsigned long value;
+
+	/* Get the right hose.  */
+	port = &pachip->g_port;
+	if (hose->index & 2) 
+		port = &pachip->a_port;
+
+	/* We can invalidate up to 8 tlb entries in a go.  The flush
+	   matches against <31:16> in the pci address.  
+	   Note that gtlbi* and atlbi* are in the same place in the g_port
+	   and a_port, respectively, so the g_port offset can be used
+	   even if hose is an a_port */
+	csr = &port->port_specific.g.gtlbia.csr;
+	if (((start ^ end) & 0xffff0000) == 0)
+		csr = &port->port_specific.g.gtlbiv.csr;
+
+	/* For TBIA, it doesn't matter what value we write.  For TBI, 
+	   it's the shifted tag bits.  */
+	value = (start & 0xffff0000) >> 12;
+
+	wmb();
+	*csr = value;
+	mb();
+	*csr;
+}
+
+static int
+titan_query_agp(titan_pachip_port *port)
+{
+	union TPAchipPCTL pctl;
+
+	/* set up APCTL */
+	pctl.pctl_q_whole = port->pctl.csr;
+
+	return pctl.pctl_r_bits.apctl_v_agp_present;
+
+}
+
+static void __init
+titan_init_one_pachip_port(titan_pachip_port *port, int index)
+{
+	struct pci_controller *hose;
+
+	hose = alloc_pci_controller();
+	if (index == 0)
+		pci_isa_hose = hose;
+	hose->io_space = alloc_resource();
+	hose->mem_space = alloc_resource();
+
+	/*
+	 * This is for userland consumption.  The 40-bit PIO bias that we 
+	 * use in the kernel through KSEG doesn't work in the page table 
+	 * based user mappings. (43-bit KSEG sign extends the physical
+	 * address from bit 40 to hit the I/O bit - mapped addresses don't).
+	 * So make sure we get the 43-bit PIO bias.  
+	 */
+	hose->sparse_mem_base = 0;
+	hose->sparse_io_base = 0;
+	hose->dense_mem_base
+	  = (TITAN_MEM(index) & 0xffffffffffUL) | 0x80000000000UL;
+	hose->dense_io_base
+	  = (TITAN_IO(index) & 0xffffffffffUL) | 0x80000000000UL;
+
+	hose->config_space_base = TITAN_CONF(index);
+	hose->index = index;
+
+	hose->io_space->start = TITAN_IO(index) - TITAN_IO_BIAS;
+	hose->io_space->end = hose->io_space->start + TITAN_IO_SPACE - 1;
+	hose->io_space->name = pci_io_names[index];
+	hose->io_space->flags = IORESOURCE_IO;
+
+	hose->mem_space->start = TITAN_MEM(index) - TITAN_MEM_BIAS;
+	hose->mem_space->end = hose->mem_space->start + 0xffffffff;
+	hose->mem_space->name = pci_mem_names[index];
+	hose->mem_space->flags = IORESOURCE_MEM;
+
+	if (request_resource(&ioport_resource, hose->io_space) < 0)
+		printk(KERN_ERR "Failed to request IO on hose %d\n", index);
+	if (request_resource(&iomem_resource, hose->mem_space) < 0)
+		printk(KERN_ERR "Failed to request MEM on hose %d\n", index);
+
+	/*
+	 * Save the existing PCI window translations.  SRM will 
+	 * need them when we go to reboot.
+	 */
+	saved_config[index].wsba[0] = port->wsba[0].csr;
+	saved_config[index].wsm[0]  = port->wsm[0].csr;
+	saved_config[index].tba[0]  = port->tba[0].csr;
+
+	saved_config[index].wsba[1] = port->wsba[1].csr;
+	saved_config[index].wsm[1]  = port->wsm[1].csr;
+	saved_config[index].tba[1]  = port->tba[1].csr;
+
+	saved_config[index].wsba[2] = port->wsba[2].csr;
+	saved_config[index].wsm[2]  = port->wsm[2].csr;
+	saved_config[index].tba[2]  = port->tba[2].csr;
+
+	saved_config[index].wsba[3] = port->wsba[3].csr;
+	saved_config[index].wsm[3]  = port->wsm[3].csr;
+	saved_config[index].tba[3]  = port->tba[3].csr;
+
+	/*
+	 * Set up the PCI to main memory translation windows.
+	 *
+	 * Note: Window 3 on Titan is Scatter-Gather ONLY.
+	 *
+	 * Window 0 is scatter-gather 8MB at 8MB (for isa)
+	 * Window 1 is direct access 1GB at 2GB
+	 * Window 2 is scatter-gather 1GB at 3GB
+	 */
+	hose->sg_isa = iommu_arena_new(hose, 0x00800000, 0x00800000,
+				       SMP_CACHE_BYTES);
+	hose->sg_isa->align_entry = 8; /* 64KB for ISA */
+
+	hose->sg_pci = iommu_arena_new(hose, 0xc0000000, 0x40000000,
+				       SMP_CACHE_BYTES);
+	hose->sg_pci->align_entry = 4; /* Titan caches 4 PTEs at a time */
+
+	port->wsba[0].csr = hose->sg_isa->dma_base | 3;
+	port->wsm[0].csr  = (hose->sg_isa->size - 1) & 0xfff00000;
+	port->tba[0].csr  = virt_to_phys(hose->sg_isa->ptes);
+
+	port->wsba[1].csr = __direct_map_base | 1;
+	port->wsm[1].csr  = (__direct_map_size - 1) & 0xfff00000;
+	port->tba[1].csr  = 0;
+
+	port->wsba[2].csr = hose->sg_pci->dma_base | 3;
+	port->wsm[2].csr  = (hose->sg_pci->size - 1) & 0xfff00000;
+	port->tba[2].csr  = virt_to_phys(hose->sg_pci->ptes);
+
+	port->wsba[3].csr = 0;
+
+	/* Enable the Monster Window to make DAC pci64 possible.  */
+	port->pctl.csr |= pctl_m_mwin;
+
+	/*
+	 * If it's an AGP port, initialize agplastwr.
+	 */
+	if (titan_query_agp(port)) 
+		port->port_specific.a.agplastwr.csr = __direct_map_base;
+
+	titan_pci_tbi(hose, 0, -1);
+}
+
+static void __init
+titan_init_pachips(titan_pachip *pachip0, titan_pachip *pachip1)
+{
+	titan_pchip1_present = TITAN_cchip->csc.csr & 1L<<14;
+
+	/* Init the ports in hose order... */
+	titan_init_one_pachip_port(&pachip0->g_port, 0);	/* hose 0 */
+	if (titan_pchip1_present)
+		titan_init_one_pachip_port(&pachip1->g_port, 1);/* hose 1 */
+	titan_init_one_pachip_port(&pachip0->a_port, 2);	/* hose 2 */
+	if (titan_pchip1_present)
+		titan_init_one_pachip_port(&pachip1->a_port, 3);/* hose 3 */
+}
+
+void __init
+titan_init_arch(void)
+{
+#if 0
+	printk("%s: titan_init_arch()\n", __func__);
+	printk("%s: CChip registers:\n", __func__);
+	printk("%s: CSR_CSC 0x%lx\n", __func__, TITAN_cchip->csc.csr);
+	printk("%s: CSR_MTR 0x%lx\n", __func__, TITAN_cchip->mtr.csr);
+	printk("%s: CSR_MISC 0x%lx\n", __func__, TITAN_cchip->misc.csr);
+	printk("%s: CSR_DIM0 0x%lx\n", __func__, TITAN_cchip->dim0.csr);
+	printk("%s: CSR_DIM1 0x%lx\n", __func__, TITAN_cchip->dim1.csr);
+	printk("%s: CSR_DIR0 0x%lx\n", __func__, TITAN_cchip->dir0.csr);
+	printk("%s: CSR_DIR1 0x%lx\n", __func__, TITAN_cchip->dir1.csr);
+	printk("%s: CSR_DRIR 0x%lx\n", __func__, TITAN_cchip->drir.csr);
+
+	printk("%s: DChip registers:\n", __func__);
+	printk("%s: CSR_DSC 0x%lx\n", __func__, TITAN_dchip->dsc.csr);
+	printk("%s: CSR_STR 0x%lx\n", __func__, TITAN_dchip->str.csr);
+	printk("%s: CSR_DREV 0x%lx\n", __func__, TITAN_dchip->drev.csr);
+#endif
+
+	boot_cpuid = __hard_smp_processor_id();
+
+	/* With multiple PCI busses, we play with I/O as physical addrs.  */
+	ioport_resource.end = ~0UL;
+	iomem_resource.end = ~0UL;
+
+	/* PCI DMA Direct Mapping is 1GB at 2GB.  */
+	__direct_map_base = 0x80000000;
+	__direct_map_size = 0x40000000;
+
+	/* Init the PA chip(s).  */
+	titan_init_pachips(TITAN_pachip0, TITAN_pachip1);
+
+	/* Check for graphic console location (if any).  */
+	find_console_vga_hose();
+}
+
+static void
+titan_kill_one_pachip_port(titan_pachip_port *port, int index)
+{
+	port->wsba[0].csr = saved_config[index].wsba[0];
+	port->wsm[0].csr  = saved_config[index].wsm[0];
+	port->tba[0].csr  = saved_config[index].tba[0];
+
+	port->wsba[1].csr = saved_config[index].wsba[1];
+	port->wsm[1].csr  = saved_config[index].wsm[1];
+	port->tba[1].csr  = saved_config[index].tba[1];
+
+	port->wsba[2].csr = saved_config[index].wsba[2];
+	port->wsm[2].csr  = saved_config[index].wsm[2];
+	port->tba[2].csr  = saved_config[index].tba[2];
+
+	port->wsba[3].csr = saved_config[index].wsba[3];
+	port->wsm[3].csr  = saved_config[index].wsm[3];
+	port->tba[3].csr  = saved_config[index].tba[3];
+}
+
+static void
+titan_kill_pachips(titan_pachip *pachip0, titan_pachip *pachip1)
+{
+	if (titan_pchip1_present) {
+		titan_kill_one_pachip_port(&pachip1->g_port, 1);
+		titan_kill_one_pachip_port(&pachip1->a_port, 3);
+	}
+	titan_kill_one_pachip_port(&pachip0->g_port, 0);
+	titan_kill_one_pachip_port(&pachip0->a_port, 2);
+}
+
+void
+titan_kill_arch(int mode)
+{
+	titan_kill_pachips(TITAN_pachip0, TITAN_pachip1);
+}
+
+
+/*
+ * IO map support.
+ */
+
+void __iomem *
+titan_ioportmap(unsigned long addr)
+{
+	FIXUP_IOADDR_VGA(addr);
+	return (void __iomem *)(addr + TITAN_IO_BIAS);
+}
+
+
+void __iomem *
+titan_ioremap(unsigned long addr, unsigned long size)
+{
+	int h = (addr & TITAN_HOSE_MASK) >> TITAN_HOSE_SHIFT;
+	unsigned long baddr = addr & ~TITAN_HOSE_MASK;
+	unsigned long last = baddr + size - 1;
+	struct pci_controller *hose;	
+	struct vm_struct *area;
+	unsigned long vaddr;
+	unsigned long *ptes;
+	unsigned long pfn;
+
+#ifdef CONFIG_VGA_HOSE
+	/*
+	 * Adjust the address and hose, if necessary.
+	 */ 
+	if (pci_vga_hose && __is_mem_vga(addr)) {
+		h = pci_vga_hose->index;
+		addr += pci_vga_hose->mem_space->start;
+	}
+#endif
+
+	/*
+	 * Find the hose.
+	 */
+	for (hose = hose_head; hose; hose = hose->next)
+		if (hose->index == h)
+			break;
+	if (!hose)
+		return NULL;
+
+	/*
+	 * Is it direct-mapped?
+	 */
+	if ((baddr >= __direct_map_base) && 
+	    ((baddr + size - 1) < __direct_map_base + __direct_map_size)) {
+		vaddr = addr - __direct_map_base + TITAN_MEM_BIAS;
+		return (void __iomem *) vaddr;
+	}
+
+	/* 
+	 * Check the scatter-gather arena.
+	 */
+	if (hose->sg_pci &&
+	    baddr >= (unsigned long)hose->sg_pci->dma_base &&
+	    last < (unsigned long)hose->sg_pci->dma_base + hose->sg_pci->size){
+
+		/*
+		 * Adjust the limits (mappings must be page aligned)
+		 */
+		baddr -= hose->sg_pci->dma_base;
+		last -= hose->sg_pci->dma_base;
+		baddr &= PAGE_MASK;
+		size = PAGE_ALIGN(last) - baddr;
+
+		/*
+		 * Map it
+		 */
+		area = get_vm_area(size, VM_IOREMAP);
+		if (!area) {
+			printk("ioremap failed... no vm_area...\n");
+			return NULL;
+		}
+
+		ptes = hose->sg_pci->ptes;
+		for (vaddr = (unsigned long)area->addr; 
+		    baddr <= last; 
+		    baddr += PAGE_SIZE, vaddr += PAGE_SIZE) {
+			pfn = ptes[baddr >> PAGE_SHIFT];
+			if (!(pfn & 1)) {
+				printk("ioremap failed... pte not valid...\n");
+				vfree(area->addr);
+				return NULL;
+			}
+			pfn >>= 1;	/* make it a true pfn */
+			
+			if (__alpha_remap_area_pages(vaddr,
+						     pfn << PAGE_SHIFT, 
+						     PAGE_SIZE, 0)) {
+				printk("FAILED to remap_area_pages...\n");
+				vfree(area->addr);
+				return NULL;
+			}
+		}
+
+		flush_tlb_all();
+
+		vaddr = (unsigned long)area->addr + (addr & ~PAGE_MASK);
+		return (void __iomem *) vaddr;
+	}
+
+	/* Assume a legacy (read: VGA) address, and return appropriately. */
+	return (void __iomem *)(addr + TITAN_MEM_BIAS);
+}
+
+void
+titan_iounmap(volatile void __iomem *xaddr)
+{
+	unsigned long addr = (unsigned long) xaddr;
+	if (addr >= VMALLOC_START)
+		vfree((void *)(PAGE_MASK & addr)); 
+}
+
+int
+titan_is_mmio(const volatile void __iomem *xaddr)
+{
+	unsigned long addr = (unsigned long) xaddr;
+
+	if (addr >= VMALLOC_START)
+		return 1;
+	else
+		return (addr & 0x100000000UL) == 0;
+}
+
+#ifndef CONFIG_ALPHA_GENERIC
+EXPORT_SYMBOL(titan_ioportmap);
+EXPORT_SYMBOL(titan_ioremap);
+EXPORT_SYMBOL(titan_iounmap);
+EXPORT_SYMBOL(titan_is_mmio);
+#endif
+
+/*
+ * AGP GART Support.
+ */
+#include <linux/agp_backend.h>
+#include <asm/agp_backend.h>
+#include <linux/slab.h>
+#include <linux/delay.h>
+
+struct titan_agp_aperture {
+	struct pci_iommu_arena *arena;
+	long pg_start;
+	long pg_count;
+};
+
+static int
+titan_agp_setup(alpha_agp_info *agp)
+{
+	struct titan_agp_aperture *aper;
+
+	if (!alpha_agpgart_size)
+		return -ENOMEM;
+
+	aper = kmalloc(sizeof(struct titan_agp_aperture), GFP_KERNEL);
+	if (aper == NULL)
+		return -ENOMEM;
+
+	aper->arena = agp->hose->sg_pci;
+	aper->pg_count = alpha_agpgart_size / PAGE_SIZE;
+	aper->pg_start = iommu_reserve(aper->arena, aper->pg_count,
+				       aper->pg_count - 1);
+	if (aper->pg_start < 0) {
+		printk(KERN_ERR "Failed to reserve AGP memory\n");
+		kfree(aper);
+		return -ENOMEM;
+	}
+
+	agp->aperture.bus_base = 
+		aper->arena->dma_base + aper->pg_start * PAGE_SIZE;
+	agp->aperture.size = aper->pg_count * PAGE_SIZE;
+	agp->aperture.sysdata = aper;
+
+	return 0;
+}
+
+static void
+titan_agp_cleanup(alpha_agp_info *agp)
+{
+	struct titan_agp_aperture *aper = agp->aperture.sysdata;
+	int status;
+
+	status = iommu_release(aper->arena, aper->pg_start, aper->pg_count);
+	if (status == -EBUSY) {
+		printk(KERN_WARNING 
+		       "Attempted to release bound AGP memory - unbinding\n");
+		iommu_unbind(aper->arena, aper->pg_start, aper->pg_count);
+		status = iommu_release(aper->arena, aper->pg_start, 
+				       aper->pg_count);
+	}
+	if (status < 0)
+		printk(KERN_ERR "Failed to release AGP memory\n");
+
+	kfree(aper);
+	kfree(agp);
+}
+
+static int
+titan_agp_configure(alpha_agp_info *agp)
+{
+	union TPAchipPCTL pctl;
+	titan_pachip_port *port = agp->private;
+	pctl.pctl_q_whole = port->pctl.csr;
+
+	/* Side-Band Addressing? */
+	pctl.pctl_r_bits.apctl_v_agp_sba_en = agp->mode.bits.sba;
+
+	/* AGP Rate? */
+	pctl.pctl_r_bits.apctl_v_agp_rate = 0;		/* 1x */
+	if (agp->mode.bits.rate & 2) 
+		pctl.pctl_r_bits.apctl_v_agp_rate = 1;	/* 2x */
+#if 0
+	if (agp->mode.bits.rate & 4) 
+		pctl.pctl_r_bits.apctl_v_agp_rate = 2;	/* 4x */
+#endif
+	
+	/* RQ Depth? */
+	pctl.pctl_r_bits.apctl_v_agp_hp_rd = 2;
+	pctl.pctl_r_bits.apctl_v_agp_lp_rd = 7;
+
+	/*
+	 * AGP Enable.
+	 */
+	pctl.pctl_r_bits.apctl_v_agp_en = agp->mode.bits.enable;
+
+	/* Tell the user.  */
+	printk("Enabling AGP: %dX%s\n", 
+	       1 << pctl.pctl_r_bits.apctl_v_agp_rate,
+	       pctl.pctl_r_bits.apctl_v_agp_sba_en ? " - SBA" : "");
+	       
+	/* Write it.  */
+	port->pctl.csr = pctl.pctl_q_whole;
+	
+	/* And wait at least 5000 66MHz cycles (per Titan spec).  */
+	udelay(100);
+
+	return 0;
+}
+
+static int 
+titan_agp_bind_memory(alpha_agp_info *agp, off_t pg_start, struct agp_memory *mem)
+{
+	struct titan_agp_aperture *aper = agp->aperture.sysdata;
+	return iommu_bind(aper->arena, aper->pg_start + pg_start, 
+			  mem->page_count, mem->pages);
+}
+
+static int 
+titan_agp_unbind_memory(alpha_agp_info *agp, off_t pg_start, struct agp_memory *mem)
+{
+	struct titan_agp_aperture *aper = agp->aperture.sysdata;
+	return iommu_unbind(aper->arena, aper->pg_start + pg_start,
+			    mem->page_count);
+}
+
+static unsigned long
+titan_agp_translate(alpha_agp_info *agp, dma_addr_t addr)
+{
+	struct titan_agp_aperture *aper = agp->aperture.sysdata;
+	unsigned long baddr = addr - aper->arena->dma_base;
+	unsigned long pte;
+
+	if (addr < agp->aperture.bus_base ||
+	    addr >= agp->aperture.bus_base + agp->aperture.size) {
+		printk("%s: addr out of range\n", __func__);
+		return -EINVAL;
+	}
+
+	pte = aper->arena->ptes[baddr >> PAGE_SHIFT];
+	if (!(pte & 1)) {
+		printk("%s: pte not valid\n", __func__);
+		return -EINVAL;
+	}
+
+	return (pte >> 1) << PAGE_SHIFT;
+}
+
+struct alpha_agp_ops titan_agp_ops =
+{
+	.setup		= titan_agp_setup,
+	.cleanup	= titan_agp_cleanup,
+	.configure	= titan_agp_configure,
+	.bind		= titan_agp_bind_memory,
+	.unbind		= titan_agp_unbind_memory,
+	.translate	= titan_agp_translate
+};
+
+alpha_agp_info *
+titan_agp_info(void)
+{
+	alpha_agp_info *agp;
+	struct pci_controller *hose;
+	titan_pachip_port *port;
+	int hosenum = -1;
+	union TPAchipPCTL pctl;
+
+	/*
+	 * Find the AGP port.
+	 */
+	port = &TITAN_pachip0->a_port;
+	if (titan_query_agp(port))
+		hosenum = 2;
+	if (hosenum < 0 && 
+	    titan_pchip1_present &&
+	    titan_query_agp(port = &TITAN_pachip1->a_port)) 
+		hosenum = 3;
+	
+	/*
+	 * Find the hose the port is on.
+	 */
+	for (hose = hose_head; hose; hose = hose->next)
+		if (hose->index == hosenum)
+			break;
+
+	if (!hose || !hose->sg_pci)
+		return NULL;
+
+	/*
+	 * Allocate the info structure.
+	 */
+	agp = kmalloc(sizeof(*agp), GFP_KERNEL);
+	if (!agp)
+		return NULL;
+
+	/*
+	 * Fill it in.
+	 */
+	agp->hose = hose;
+	agp->private = port;
+	agp->ops = &titan_agp_ops;
+
+	/*
+	 * Aperture - not configured until ops.setup().
+	 *
+	 * FIXME - should we go ahead and allocate it here?
+	 */
+	agp->aperture.bus_base = 0;
+	agp->aperture.size = 0;
+	agp->aperture.sysdata = NULL;
+
+	/*
+	 * Capabilities.
+	 */
+	agp->capability.lw = 0;
+	agp->capability.bits.rate = 3; 	/* 2x, 1x */
+	agp->capability.bits.sba = 1;
+	agp->capability.bits.rq = 7;	/* 8 - 1 */
+
+	/*
+	 * Mode.
+	 */
+	pctl.pctl_q_whole = port->pctl.csr;
+	agp->mode.lw = 0;
+	agp->mode.bits.rate = 1 << pctl.pctl_r_bits.apctl_v_agp_rate;
+	agp->mode.bits.sba = pctl.pctl_r_bits.apctl_v_agp_sba_en;
+	agp->mode.bits.rq = 7;	/* RQ Depth? */
+	agp->mode.bits.enable = pctl.pctl_r_bits.apctl_v_agp_en;
+
+	return agp;
+}