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, 1007 insertions, 0 deletions

diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/fb/ramgt215.c b/drivers/gpu/drm/nouveau/nvkm/subdev/fb/ramgt215.c
new file mode 100644
index 000000000..bbfde1cb3
--- /dev/null
+++ b/drivers/gpu/drm/nouveau/nvkm/subdev/fb/ramgt215.c
@@ -0,0 +1,1007 @@
+/*
+ * Copyright 2013 Red Hat Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors: Ben Skeggs
+ * 	    Roy Spliet <rspliet@eclipso.eu>
+ */
+#define gt215_ram(p) container_of((p), struct gt215_ram, base)
+#include "ram.h"
+#include "ramfuc.h"
+
+#include <core/memory.h>
+#include <core/option.h>
+#include <subdev/bios.h>
+#include <subdev/bios/M0205.h>
+#include <subdev/bios/rammap.h>
+#include <subdev/bios/timing.h>
+#include <subdev/clk/gt215.h>
+#include <subdev/gpio.h>
+
+struct gt215_ramfuc {
+	struct ramfuc base;
+	struct ramfuc_reg r_0x001610;
+	struct ramfuc_reg r_0x001700;
+	struct ramfuc_reg r_0x002504;
+	struct ramfuc_reg r_0x004000;
+	struct ramfuc_reg r_0x004004;
+	struct ramfuc_reg r_0x004018;
+	struct ramfuc_reg r_0x004128;
+	struct ramfuc_reg r_0x004168;
+	struct ramfuc_reg r_0x100080;
+	struct ramfuc_reg r_0x100200;
+	struct ramfuc_reg r_0x100210;
+	struct ramfuc_reg r_0x100220[9];
+	struct ramfuc_reg r_0x100264;
+	struct ramfuc_reg r_0x1002d0;
+	struct ramfuc_reg r_0x1002d4;
+	struct ramfuc_reg r_0x1002dc;
+	struct ramfuc_reg r_0x10053c;
+	struct ramfuc_reg r_0x1005a0;
+	struct ramfuc_reg r_0x1005a4;
+	struct ramfuc_reg r_0x100700;
+	struct ramfuc_reg r_0x100714;
+	struct ramfuc_reg r_0x100718;
+	struct ramfuc_reg r_0x10071c;
+	struct ramfuc_reg r_0x100720;
+	struct ramfuc_reg r_0x100760;
+	struct ramfuc_reg r_0x1007a0;
+	struct ramfuc_reg r_0x1007e0;
+	struct ramfuc_reg r_0x100da0;
+	struct ramfuc_reg r_0x10f804;
+	struct ramfuc_reg r_0x1110e0;
+	struct ramfuc_reg r_0x111100;
+	struct ramfuc_reg r_0x111104;
+	struct ramfuc_reg r_0x1111e0;
+	struct ramfuc_reg r_0x111400;
+	struct ramfuc_reg r_0x611200;
+	struct ramfuc_reg r_mr[4];
+	struct ramfuc_reg r_gpio[4];
+};
+
+struct gt215_ltrain {
+	enum {
+		NVA3_TRAIN_UNKNOWN,
+		NVA3_TRAIN_UNSUPPORTED,
+		NVA3_TRAIN_ONCE,
+		NVA3_TRAIN_EXEC,
+		NVA3_TRAIN_DONE
+	} state;
+	u32 r_100720;
+	u32 r_1111e0;
+	u32 r_111400;
+	struct nvkm_memory *memory;
+};
+
+struct gt215_ram {
+	struct nvkm_ram base;
+	struct gt215_ramfuc fuc;
+	struct gt215_ltrain ltrain;
+};
+
+static void
+gt215_link_train_calc(u32 *vals, struct gt215_ltrain *train)
+{
+	int i, lo, hi;
+	u8 median[8], bins[4] = {0, 0, 0, 0}, bin = 0, qty = 0;
+
+	for (i = 0; i < 8; i++) {
+		for (lo = 0; lo < 0x40; lo++) {
+			if (!(vals[lo] & 0x80000000))
+				continue;
+			if (vals[lo] & (0x101 << i))
+				break;
+		}
+
+		if (lo == 0x40)
+			return;
+
+		for (hi = lo + 1; hi < 0x40; hi++) {
+			if (!(vals[lo] & 0x80000000))
+				continue;
+			if (!(vals[hi] & (0x101 << i))) {
+				hi--;
+				break;
+			}
+		}
+
+		median[i] = ((hi - lo) >> 1) + lo;
+		bins[(median[i] & 0xf0) >> 4]++;
+		median[i] += 0x30;
+	}
+
+	/* Find the best value for 0x1111e0 */
+	for (i = 0; i < 4; i++) {
+		if (bins[i] > qty) {
+			bin = i + 3;
+			qty = bins[i];
+		}
+	}
+
+	train->r_100720 = 0;
+	for (i = 0; i < 8; i++) {
+		median[i] = max(median[i], (u8) (bin << 4));
+		median[i] = min(median[i], (u8) ((bin << 4) | 0xf));
+
+		train->r_100720 |= ((median[i] & 0x0f) << (i << 2));
+	}
+
+	train->r_1111e0 = 0x02000000 | (bin * 0x101);
+	train->r_111400 = 0x0;
+}
+
+/*
+ * Link training for (at least) DDR3
+ */
+static int
+gt215_link_train(struct gt215_ram *ram)
+{
+	struct gt215_ltrain *train = &ram->ltrain;
+	struct gt215_ramfuc *fuc = &ram->fuc;
+	struct nvkm_subdev *subdev = &ram->base.fb->subdev;
+	struct nvkm_device *device = subdev->device;
+	struct nvkm_bios *bios = device->bios;
+	struct nvkm_clk *clk = device->clk;
+	u32 *result, r1700;
+	int ret, i;
+	struct nvbios_M0205T M0205T = { 0 };
+	u8 ver, hdr, cnt, len, snr, ssz;
+	unsigned int clk_current;
+	unsigned long flags;
+	unsigned long *f = &flags;
+
+	if (nvkm_boolopt(device->cfgopt, "NvMemExec", true) != true)
+		return -ENOSYS;
+
+	/* XXX: Multiple partitions? */
+	result = kmalloc_array(64, sizeof(u32), GFP_KERNEL);
+	if (!result)
+		return -ENOMEM;
+
+	train->state = NVA3_TRAIN_EXEC;
+
+	/* Clock speeds for training and back */
+	nvbios_M0205Tp(bios, &ver, &hdr, &cnt, &len, &snr, &ssz, &M0205T);
+	if (M0205T.freq == 0) {
+		kfree(result);
+		return -ENOENT;
+	}
+
+	clk_current = nvkm_clk_read(clk, nv_clk_src_mem);
+
+	ret = gt215_clk_pre(clk, f);
+	if (ret)
+		goto out;
+
+	/* First: clock up/down */
+	ret = ram->base.func->calc(&ram->base, (u32) M0205T.freq * 1000);
+	if (ret)
+		goto out;
+
+	/* Do this *after* calc, eliminates write in script */
+	nvkm_wr32(device, 0x111400, 0x00000000);
+	/* XXX: Magic writes that improve train reliability? */
+	nvkm_mask(device, 0x100674, 0x0000ffff, 0x00000000);
+	nvkm_mask(device, 0x1005e4, 0x0000ffff, 0x00000000);
+	nvkm_mask(device, 0x100b0c, 0x000000ff, 0x00000000);
+	nvkm_wr32(device, 0x100c04, 0x00000400);
+
+	/* Now the training script */
+	r1700 = ram_rd32(fuc, 0x001700);
+
+	ram_mask(fuc, 0x100200, 0x00000800, 0x00000000);
+	ram_wr32(fuc, 0x611200, 0x3300);
+	ram_wait_vblank(fuc);
+	ram_wait(fuc, 0x611200, 0x00000003, 0x00000000, 500000);
+	ram_mask(fuc, 0x001610, 0x00000083, 0x00000003);
+	ram_mask(fuc, 0x100080, 0x00000020, 0x00000000);
+	ram_mask(fuc, 0x10f804, 0x80000000, 0x00000000);
+	ram_wr32(fuc, 0x001700, 0x00000000);
+
+	ram_train(fuc);
+
+	/* Reset */
+	ram_mask(fuc, 0x10f804, 0x80000000, 0x80000000);
+	ram_wr32(fuc, 0x10053c, 0x0);
+	ram_wr32(fuc, 0x100720, train->r_100720);
+	ram_wr32(fuc, 0x1111e0, train->r_1111e0);
+	ram_wr32(fuc, 0x111400, train->r_111400);
+	ram_nuke(fuc, 0x100080);
+	ram_mask(fuc, 0x100080, 0x00000020, 0x00000020);
+	ram_nsec(fuc, 1000);
+
+	ram_wr32(fuc, 0x001700, r1700);
+	ram_mask(fuc, 0x001610, 0x00000083, 0x00000080);
+	ram_wr32(fuc, 0x611200, 0x3330);
+	ram_mask(fuc, 0x100200, 0x00000800, 0x00000800);
+
+	ram_exec(fuc, true);
+
+	ram->base.func->calc(&ram->base, clk_current);
+	ram_exec(fuc, true);
+
+	/* Post-processing, avoids flicker */
+	nvkm_mask(device, 0x616308, 0x10, 0x10);
+	nvkm_mask(device, 0x616b08, 0x10, 0x10);
+
+	gt215_clk_post(clk, f);
+
+	ram_train_result(ram->base.fb, result, 64);
+	for (i = 0; i < 64; i++)
+		nvkm_debug(subdev, "Train: %08x", result[i]);
+	gt215_link_train_calc(result, train);
+
+	nvkm_debug(subdev, "Train: %08x %08x %08x", train->r_100720,
+		   train->r_1111e0, train->r_111400);
+
+	kfree(result);
+
+	train->state = NVA3_TRAIN_DONE;
+
+	return ret;
+
+out:
+	if(ret == -EBUSY)
+		f = NULL;
+
+	train->state = NVA3_TRAIN_UNSUPPORTED;
+
+	gt215_clk_post(clk, f);
+	kfree(result);
+	return ret;
+}
+
+static int
+gt215_link_train_init(struct gt215_ram *ram)
+{
+	static const u32 pattern[16] = {
+		0xaaaaaaaa, 0xcccccccc, 0xdddddddd, 0xeeeeeeee,
+		0x00000000, 0x11111111, 0x44444444, 0xdddddddd,
+		0x33333333, 0x55555555, 0x77777777, 0x66666666,
+		0x99999999, 0x88888888, 0xeeeeeeee, 0xbbbbbbbb,
+	};
+	struct gt215_ltrain *train = &ram->ltrain;
+	struct nvkm_device *device = ram->base.fb->subdev.device;
+	struct nvkm_bios *bios = device->bios;
+	struct nvbios_M0205E M0205E;
+	u8 ver, hdr, cnt, len;
+	u32 r001700;
+	u64 addr;
+	int ret, i = 0;
+
+	train->state = NVA3_TRAIN_UNSUPPORTED;
+
+	/* We support type "5"
+	 * XXX: training pattern table appears to be unused for this routine */
+	if (!nvbios_M0205Ep(bios, i, &ver, &hdr, &cnt, &len, &M0205E))
+		return -ENOENT;
+
+	if (M0205E.type != 5)
+		return 0;
+
+	train->state = NVA3_TRAIN_ONCE;
+
+	ret = nvkm_ram_get(device, NVKM_RAM_MM_NORMAL, 0x01, 16, 0x8000,
+			   true, true, &ram->ltrain.memory);
+	if (ret)
+		return ret;
+
+	addr = nvkm_memory_addr(ram->ltrain.memory);
+
+	nvkm_wr32(device, 0x100538, 0x10000000 | (addr >> 16));
+	nvkm_wr32(device, 0x1005a8, 0x0000ffff);
+	nvkm_mask(device, 0x10f800, 0x00000001, 0x00000001);
+
+	for (i = 0; i < 0x30; i++) {
+		nvkm_wr32(device, 0x10f8c0, (i << 8) | i);
+		nvkm_wr32(device, 0x10f900, pattern[i % 16]);
+	}
+
+	for (i = 0; i < 0x30; i++) {
+		nvkm_wr32(device, 0x10f8e0, (i << 8) | i);
+		nvkm_wr32(device, 0x10f920, pattern[i % 16]);
+	}
+
+	/* And upload the pattern */
+	r001700 = nvkm_rd32(device, 0x1700);
+	nvkm_wr32(device, 0x1700, addr >> 16);
+	for (i = 0; i < 16; i++)
+		nvkm_wr32(device, 0x700000 + (i << 2), pattern[i]);
+	for (i = 0; i < 16; i++)
+		nvkm_wr32(device, 0x700100 + (i << 2), pattern[i]);
+	nvkm_wr32(device, 0x1700, r001700);
+
+	train->r_100720 = nvkm_rd32(device, 0x100720);
+	train->r_1111e0 = nvkm_rd32(device, 0x1111e0);
+	train->r_111400 = nvkm_rd32(device, 0x111400);
+	return 0;
+}
+
+static void
+gt215_link_train_fini(struct gt215_ram *ram)
+{
+	nvkm_memory_unref(&ram->ltrain.memory);
+}
+
+/*
+ * RAM reclocking
+ */
+#define T(t) cfg->timing_10_##t
+static int
+gt215_ram_timing_calc(struct gt215_ram *ram, u32 *timing)
+{
+	struct nvbios_ramcfg *cfg = &ram->base.target.bios;
+	struct nvkm_subdev *subdev = &ram->base.fb->subdev;
+	struct nvkm_device *device = subdev->device;
+	int tUNK_base, tUNK_40_0, prevCL;
+	u32 cur2, cur3, cur7, cur8;
+
+	cur2 = nvkm_rd32(device, 0x100228);
+	cur3 = nvkm_rd32(device, 0x10022c);
+	cur7 = nvkm_rd32(device, 0x10023c);
+	cur8 = nvkm_rd32(device, 0x100240);
+
+
+	switch ((!T(CWL)) * ram->base.type) {
+	case NVKM_RAM_TYPE_DDR2:
+		T(CWL) = T(CL) - 1;
+		break;
+	case NVKM_RAM_TYPE_GDDR3:
+		T(CWL) = ((cur2 & 0xff000000) >> 24) + 1;
+		break;
+	}
+
+	prevCL = (cur3 & 0x000000ff) + 1;
+	tUNK_base = ((cur7 & 0x00ff0000) >> 16) - prevCL;
+
+	timing[0] = (T(RP) << 24 | T(RAS) << 16 | T(RFC) << 8 | T(RC));
+	timing[1] = (T(WR) + 1 + T(CWL)) << 24 |
+		    max_t(u8,T(18), 1) << 16 |
+		    (T(WTR) + 1 + T(CWL)) << 8 |
+		    (5 + T(CL) - T(CWL));
+	timing[2] = (T(CWL) - 1) << 24 |
+		    (T(RRD) << 16) |
+		    (T(RCDWR) << 8) |
+		    T(RCDRD);
+	timing[3] = (cur3 & 0x00ff0000) |
+		    (0x30 + T(CL)) << 24 |
+		    (0xb + T(CL)) << 8 |
+		    (T(CL) - 1);
+	timing[4] = T(20) << 24 |
+		    T(21) << 16 |
+		    T(13) << 8 |
+		    T(13);
+	timing[5] = T(RFC) << 24 |
+		    max_t(u8,T(RCDRD), T(RCDWR)) << 16 |
+		    max_t(u8, (T(CWL) + 6), (T(CL) + 2)) << 8 |
+		    T(RP);
+	timing[6] = (0x5a + T(CL)) << 16 |
+		    max_t(u8, 1, (6 - T(CL) + T(CWL))) << 8 |
+		    (0x50 + T(CL) - T(CWL));
+	timing[7] = (cur7 & 0xff000000) |
+		    ((tUNK_base + T(CL)) << 16) |
+		    0x202;
+	timing[8] = cur8 & 0xffffff00;
+
+	switch (ram->base.type) {
+	case NVKM_RAM_TYPE_DDR2:
+	case NVKM_RAM_TYPE_GDDR3:
+		tUNK_40_0 = prevCL - (cur8 & 0xff);
+		if (tUNK_40_0 > 0)
+			timing[8] |= T(CL);
+		break;
+	default:
+		break;
+	}
+
+	nvkm_debug(subdev, "Entry: 220: %08x %08x %08x %08x\n",
+		   timing[0], timing[1], timing[2], timing[3]);
+	nvkm_debug(subdev, "  230: %08x %08x %08x %08x\n",
+		   timing[4], timing[5], timing[6], timing[7]);
+	nvkm_debug(subdev, "  240: %08x\n", timing[8]);
+	return 0;
+}
+#undef T
+
+static void
+nvkm_sddr2_dll_reset(struct gt215_ramfuc *fuc)
+{
+	ram_mask(fuc, mr[0], 0x100, 0x100);
+	ram_nsec(fuc, 1000);
+	ram_mask(fuc, mr[0], 0x100, 0x000);
+	ram_nsec(fuc, 1000);
+}
+
+static void
+nvkm_sddr3_dll_disable(struct gt215_ramfuc *fuc, u32 *mr)
+{
+	u32 mr1_old = ram_rd32(fuc, mr[1]);
+
+	if (!(mr1_old & 0x1)) {
+		ram_wr32(fuc, 0x1002d4, 0x00000001);
+		ram_wr32(fuc, mr[1], mr[1]);
+		ram_nsec(fuc, 1000);
+	}
+}
+
+static void
+nvkm_gddr3_dll_disable(struct gt215_ramfuc *fuc, u32 *mr)
+{
+	u32 mr1_old = ram_rd32(fuc, mr[1]);
+
+	if (!(mr1_old & 0x40)) {
+		ram_wr32(fuc, mr[1], mr[1]);
+		ram_nsec(fuc, 1000);
+	}
+}
+
+static void
+gt215_ram_lock_pll(struct gt215_ramfuc *fuc, struct gt215_clk_info *mclk)
+{
+	ram_wr32(fuc, 0x004004, mclk->pll);
+	ram_mask(fuc, 0x004000, 0x00000001, 0x00000001);
+	ram_mask(fuc, 0x004000, 0x00000010, 0x00000000);
+	ram_wait(fuc, 0x004000, 0x00020000, 0x00020000, 64000);
+	ram_mask(fuc, 0x004000, 0x00000010, 0x00000010);
+}
+
+static void
+gt215_ram_gpio(struct gt215_ramfuc *fuc, u8 tag, u32 val)
+{
+	struct nvkm_gpio *gpio = fuc->base.fb->subdev.device->gpio;
+	struct dcb_gpio_func func;
+	u32 reg, sh, gpio_val;
+	int ret;
+
+	if (nvkm_gpio_get(gpio, 0, tag, DCB_GPIO_UNUSED) != val) {
+		ret = nvkm_gpio_find(gpio, 0, tag, DCB_GPIO_UNUSED, &func);
+		if (ret)
+			return;
+
+		reg = func.line >> 3;
+		sh = (func.line & 0x7) << 2;
+		gpio_val = ram_rd32(fuc, gpio[reg]);
+		if (gpio_val & (8 << sh))
+			val = !val;
+		if (!(func.log[1] & 1))
+			val = !val;
+
+		ram_mask(fuc, gpio[reg], (0x3 << sh), ((val | 0x2) << sh));
+		ram_nsec(fuc, 20000);
+	}
+}
+
+static int
+gt215_ram_calc(struct nvkm_ram *base, u32 freq)
+{
+	struct gt215_ram *ram = gt215_ram(base);
+	struct gt215_ramfuc *fuc = &ram->fuc;
+	struct gt215_ltrain *train = &ram->ltrain;
+	struct nvkm_subdev *subdev = &ram->base.fb->subdev;
+	struct nvkm_device *device = subdev->device;
+	struct nvkm_bios *bios = device->bios;
+	struct gt215_clk_info mclk;
+	struct nvkm_gpio *gpio = device->gpio;
+	struct nvkm_ram_data *next;
+	u8  ver, hdr, cnt, len, strap;
+	u32 data;
+	u32 r004018, r100760, r100da0, r111100, ctrl;
+	u32 unk714, unk718, unk71c;
+	int ret, i;
+	u32 timing[9];
+	bool pll2pll;
+
+	next = &ram->base.target;
+	next->freq = freq;
+	ram->base.next = next;
+
+	if (ram->ltrain.state == NVA3_TRAIN_ONCE)
+		gt215_link_train(ram);
+
+	/* lookup memory config data relevant to the target frequency */
+	data = nvbios_rammapEm(bios, freq / 1000, &ver, &hdr, &cnt, &len,
+			       &next->bios);
+	if (!data || ver != 0x10 || hdr < 0x05) {
+		nvkm_error(subdev, "invalid/missing rammap entry\n");
+		return -EINVAL;
+	}
+
+	/* locate specific data set for the attached memory */
+	strap = nvbios_ramcfg_index(subdev);
+	if (strap >= cnt) {
+		nvkm_error(subdev, "invalid ramcfg strap\n");
+		return -EINVAL;
+	}
+
+	data = nvbios_rammapSp(bios, data, ver, hdr, cnt, len, strap,
+			       &ver, &hdr, &next->bios);
+	if (!data || ver != 0x10 || hdr < 0x09) {
+		nvkm_error(subdev, "invalid/missing ramcfg entry\n");
+		return -EINVAL;
+	}
+
+	/* lookup memory timings, if bios says they're present */
+	if (next->bios.ramcfg_timing != 0xff) {
+		data = nvbios_timingEp(bios, next->bios.ramcfg_timing,
+				       &ver, &hdr, &cnt, &len,
+				       &next->bios);
+		if (!data || ver != 0x10 || hdr < 0x17) {
+			nvkm_error(subdev, "invalid/missing timing entry\n");
+			return -EINVAL;
+		}
+	}
+
+	ret = gt215_pll_info(device->clk, 0x12, 0x4000, freq, &mclk);
+	if (ret < 0) {
+		nvkm_error(subdev, "failed mclk calculation\n");
+		return ret;
+	}
+
+	gt215_ram_timing_calc(ram, timing);
+
+	ret = ram_init(fuc, ram->base.fb);
+	if (ret)
+		return ret;
+
+	/* Determine ram-specific MR values */
+	ram->base.mr[0] = ram_rd32(fuc, mr[0]);
+	ram->base.mr[1] = ram_rd32(fuc, mr[1]);
+	ram->base.mr[2] = ram_rd32(fuc, mr[2]);
+
+	switch (ram->base.type) {
+	case NVKM_RAM_TYPE_DDR2:
+		ret = nvkm_sddr2_calc(&ram->base);
+		break;
+	case NVKM_RAM_TYPE_DDR3:
+		ret = nvkm_sddr3_calc(&ram->base);
+		break;
+	case NVKM_RAM_TYPE_GDDR3:
+		ret = nvkm_gddr3_calc(&ram->base);
+		break;
+	default:
+		ret = -ENOSYS;
+		break;
+	}
+
+	if (ret)
+		return ret;
+
+	/* XXX: 750MHz seems rather arbitrary */
+	if (freq <= 750000) {
+		r004018 = 0x10000000;
+		r100760 = 0x22222222;
+		r100da0 = 0x00000010;
+	} else {
+		r004018 = 0x00000000;
+		r100760 = 0x00000000;
+		r100da0 = 0x00000000;
+	}
+
+	if (!next->bios.ramcfg_DLLoff)
+		r004018 |= 0x00004000;
+
+	/* pll2pll requires to switch to a safe clock first */
+	ctrl = ram_rd32(fuc, 0x004000);
+	pll2pll = (!(ctrl & 0x00000008)) && mclk.pll;
+
+	/* Pre, NVIDIA does this outside the script */
+	if (next->bios.ramcfg_10_02_10) {
+		ram_mask(fuc, 0x111104, 0x00000600, 0x00000000);
+	} else {
+		ram_mask(fuc, 0x111100, 0x40000000, 0x40000000);
+		ram_mask(fuc, 0x111104, 0x00000180, 0x00000000);
+	}
+	/* Always disable this bit during reclock */
+	ram_mask(fuc, 0x100200, 0x00000800, 0x00000000);
+
+	/* If switching from non-pll to pll, lock before disabling FB */
+	if (mclk.pll && !pll2pll) {
+		ram_mask(fuc, 0x004128, 0x003f3141, mclk.clk | 0x00000101);
+		gt215_ram_lock_pll(fuc, &mclk);
+	}
+
+	/* Start with disabling some CRTCs and PFIFO? */
+	ram_wait_vblank(fuc);
+	ram_wr32(fuc, 0x611200, 0x3300);
+	ram_mask(fuc, 0x002504, 0x1, 0x1);
+	ram_nsec(fuc, 10000);
+	ram_wait(fuc, 0x002504, 0x10, 0x10, 20000); /* XXX: or longer? */
+	ram_block(fuc);
+	ram_nsec(fuc, 2000);
+
+	if (!next->bios.ramcfg_10_02_10) {
+		if (ram->base.type == NVKM_RAM_TYPE_GDDR3)
+			ram_mask(fuc, 0x111100, 0x04020000, 0x00020000);
+		else
+			ram_mask(fuc, 0x111100, 0x04020000, 0x04020000);
+	}
+
+	/* If we're disabling the DLL, do it now */
+	switch (next->bios.ramcfg_DLLoff * ram->base.type) {
+	case NVKM_RAM_TYPE_DDR3:
+		nvkm_sddr3_dll_disable(fuc, ram->base.mr);
+		break;
+	case NVKM_RAM_TYPE_GDDR3:
+		nvkm_gddr3_dll_disable(fuc, ram->base.mr);
+		break;
+	}
+
+	if (next->bios.timing_10_ODT)
+		gt215_ram_gpio(fuc, 0x2e, 1);
+
+	/* Brace RAM for impact */
+	ram_wr32(fuc, 0x1002d4, 0x00000001);
+	ram_wr32(fuc, 0x1002d0, 0x00000001);
+	ram_wr32(fuc, 0x1002d0, 0x00000001);
+	ram_wr32(fuc, 0x100210, 0x00000000);
+	ram_wr32(fuc, 0x1002dc, 0x00000001);
+	ram_nsec(fuc, 2000);
+
+	if (device->chipset == 0xa3 && freq <= 500000)
+		ram_mask(fuc, 0x100700, 0x00000006, 0x00000006);
+
+	/* Alter FBVDD/Q, apparently must be done with PLL disabled, thus
+	 * set it to bypass */
+	if (nvkm_gpio_get(gpio, 0, 0x18, DCB_GPIO_UNUSED) ==
+			next->bios.ramcfg_FBVDDQ) {
+		data = ram_rd32(fuc, 0x004000) & 0x9;
+
+		if (data == 0x1)
+			ram_mask(fuc, 0x004000, 0x8, 0x8);
+		if (data & 0x1)
+			ram_mask(fuc, 0x004000, 0x1, 0x0);
+
+		gt215_ram_gpio(fuc, 0x18, !next->bios.ramcfg_FBVDDQ);
+
+		if (data & 0x1)
+			ram_mask(fuc, 0x004000, 0x1, 0x1);
+	}
+
+	/* Fiddle with clocks */
+	/* There's 4 scenario's
+	 * pll->pll: first switch to a 324MHz clock, set up new PLL, switch
+	 * clk->pll: Set up new PLL, switch
+	 * pll->clk: Set up clock, switch
+	 * clk->clk: Overwrite ctrl and other bits, switch */
+
+	/* Switch to regular clock - 324MHz */
+	if (pll2pll) {
+		ram_mask(fuc, 0x004000, 0x00000004, 0x00000004);
+		ram_mask(fuc, 0x004168, 0x003f3141, 0x00083101);
+		ram_mask(fuc, 0x004000, 0x00000008, 0x00000008);
+		ram_mask(fuc, 0x1110e0, 0x00088000, 0x00088000);
+		ram_wr32(fuc, 0x004018, 0x00001000);
+		gt215_ram_lock_pll(fuc, &mclk);
+	}
+
+	if (mclk.pll) {
+		ram_mask(fuc, 0x004000, 0x00000105, 0x00000105);
+		ram_wr32(fuc, 0x004018, 0x00001000 | r004018);
+		ram_wr32(fuc, 0x100da0, r100da0);
+	} else {
+		ram_mask(fuc, 0x004168, 0x003f3141, mclk.clk | 0x00000101);
+		ram_mask(fuc, 0x004000, 0x00000108, 0x00000008);
+		ram_mask(fuc, 0x1110e0, 0x00088000, 0x00088000);
+		ram_wr32(fuc, 0x004018, 0x00009000 | r004018);
+		ram_wr32(fuc, 0x100da0, r100da0);
+	}
+	ram_nsec(fuc, 20000);
+
+	if (next->bios.rammap_10_04_08) {
+		ram_wr32(fuc, 0x1005a0, next->bios.ramcfg_10_06 << 16 |
+					next->bios.ramcfg_10_05 << 8 |
+					next->bios.ramcfg_10_05);
+		ram_wr32(fuc, 0x1005a4, next->bios.ramcfg_10_08 << 8 |
+					next->bios.ramcfg_10_07);
+		ram_wr32(fuc, 0x10f804, next->bios.ramcfg_10_09_f0 << 20 |
+					next->bios.ramcfg_10_03_0f << 16 |
+					next->bios.ramcfg_10_09_0f |
+					0x80000000);
+		ram_mask(fuc, 0x10053c, 0x00001000, 0x00000000);
+	} else {
+		if (train->state == NVA3_TRAIN_DONE) {
+			ram_wr32(fuc, 0x100080, 0x1020);
+			ram_mask(fuc, 0x111400, 0xffffffff, train->r_111400);
+			ram_mask(fuc, 0x1111e0, 0xffffffff, train->r_1111e0);
+			ram_mask(fuc, 0x100720, 0xffffffff, train->r_100720);
+		}
+		ram_mask(fuc, 0x10053c, 0x00001000, 0x00001000);
+		ram_mask(fuc, 0x10f804, 0x80000000, 0x00000000);
+		ram_mask(fuc, 0x100760, 0x22222222, r100760);
+		ram_mask(fuc, 0x1007a0, 0x22222222, r100760);
+		ram_mask(fuc, 0x1007e0, 0x22222222, r100760);
+	}
+
+	if (device->chipset == 0xa3 && freq > 500000) {
+		ram_mask(fuc, 0x100700, 0x00000006, 0x00000000);
+	}
+
+	/* Final switch */
+	if (mclk.pll) {
+		ram_mask(fuc, 0x1110e0, 0x00088000, 0x00011000);
+		ram_mask(fuc, 0x004000, 0x00000008, 0x00000000);
+	}
+
+	ram_wr32(fuc, 0x1002dc, 0x00000000);
+	ram_wr32(fuc, 0x1002d4, 0x00000001);
+	ram_wr32(fuc, 0x100210, 0x80000000);
+	ram_nsec(fuc, 2000);
+
+	/* Set RAM MR parameters and timings */
+	for (i = 2; i >= 0; i--) {
+		if (ram_rd32(fuc, mr[i]) != ram->base.mr[i]) {
+			ram_wr32(fuc, mr[i], ram->base.mr[i]);
+			ram_nsec(fuc, 1000);
+		}
+	}
+
+	ram_wr32(fuc, 0x100220[3], timing[3]);
+	ram_wr32(fuc, 0x100220[1], timing[1]);
+	ram_wr32(fuc, 0x100220[6], timing[6]);
+	ram_wr32(fuc, 0x100220[7], timing[7]);
+	ram_wr32(fuc, 0x100220[2], timing[2]);
+	ram_wr32(fuc, 0x100220[4], timing[4]);
+	ram_wr32(fuc, 0x100220[5], timing[5]);
+	ram_wr32(fuc, 0x100220[0], timing[0]);
+	ram_wr32(fuc, 0x100220[8], timing[8]);
+
+	/* Misc */
+	ram_mask(fuc, 0x100200, 0x00001000, !next->bios.ramcfg_10_02_08 << 12);
+
+	/* XXX: A lot of "chipset"/"ram type" specific stuff...? */
+	unk714  = ram_rd32(fuc, 0x100714) & ~0xf0000130;
+	unk718  = ram_rd32(fuc, 0x100718) & ~0x00000100;
+	unk71c  = ram_rd32(fuc, 0x10071c) & ~0x00000100;
+	r111100 = ram_rd32(fuc, 0x111100) & ~0x3a800000;
+
+	/* NVA8 seems to skip various bits related to ramcfg_10_02_04 */
+	if (device->chipset == 0xa8) {
+		r111100 |= 0x08000000;
+		if (!next->bios.ramcfg_10_02_04)
+			unk714  |= 0x00000010;
+	} else {
+		if (next->bios.ramcfg_10_02_04) {
+			switch (ram->base.type) {
+			case NVKM_RAM_TYPE_DDR2:
+			case NVKM_RAM_TYPE_DDR3:
+				r111100 &= ~0x00000020;
+				if (next->bios.ramcfg_10_02_10)
+					r111100 |= 0x08000004;
+				else
+					r111100 |= 0x00000024;
+				break;
+			default:
+				break;
+			}
+		} else {
+			switch (ram->base.type) {
+			case NVKM_RAM_TYPE_DDR2:
+			case NVKM_RAM_TYPE_DDR3:
+				r111100 &= ~0x00000024;
+				r111100 |=  0x12800000;
+
+				if (next->bios.ramcfg_10_02_10)
+					r111100 |= 0x08000000;
+				unk714  |= 0x00000010;
+				break;
+			case NVKM_RAM_TYPE_GDDR3:
+				r111100 |= 0x30000000;
+				unk714  |= 0x00000020;
+				break;
+			default:
+				break;
+			}
+		}
+	}
+
+	unk714 |= (next->bios.ramcfg_10_04_01) << 8;
+
+	if (next->bios.ramcfg_10_02_20)
+		unk714 |= 0xf0000000;
+	if (next->bios.ramcfg_10_02_02)
+		unk718 |= 0x00000100;
+	if (next->bios.ramcfg_10_02_01)
+		unk71c |= 0x00000100;
+	if (next->bios.timing_10_24 != 0xff) {
+		unk718 &= ~0xf0000000;
+		unk718 |= next->bios.timing_10_24 << 28;
+	}
+	if (next->bios.ramcfg_10_02_10)
+		r111100 &= ~0x04020000;
+
+	ram_mask(fuc, 0x100714, 0xffffffff, unk714);
+	ram_mask(fuc, 0x10071c, 0xffffffff, unk71c);
+	ram_mask(fuc, 0x100718, 0xffffffff, unk718);
+	ram_mask(fuc, 0x111100, 0xffffffff, r111100);
+
+	if (!next->bios.timing_10_ODT)
+		gt215_ram_gpio(fuc, 0x2e, 0);
+
+	/* Reset DLL */
+	if (!next->bios.ramcfg_DLLoff)
+		nvkm_sddr2_dll_reset(fuc);
+
+	if (ram->base.type == NVKM_RAM_TYPE_GDDR3) {
+		ram_nsec(fuc, 31000);
+	} else {
+		ram_nsec(fuc, 14000);
+	}
+
+	if (ram->base.type == NVKM_RAM_TYPE_DDR3) {
+		ram_wr32(fuc, 0x100264, 0x1);
+		ram_nsec(fuc, 2000);
+	}
+
+	ram_nuke(fuc, 0x100700);
+	ram_mask(fuc, 0x100700, 0x01000000, 0x01000000);
+	ram_mask(fuc, 0x100700, 0x01000000, 0x00000000);
+
+	/* Re-enable FB */
+	ram_unblock(fuc);
+	ram_wr32(fuc, 0x611200, 0x3330);
+
+	/* Post fiddlings */
+	if (next->bios.rammap_10_04_02)
+		ram_mask(fuc, 0x100200, 0x00000800, 0x00000800);
+	if (next->bios.ramcfg_10_02_10) {
+		ram_mask(fuc, 0x111104, 0x00000180, 0x00000180);
+		ram_mask(fuc, 0x111100, 0x40000000, 0x00000000);
+	} else {
+		ram_mask(fuc, 0x111104, 0x00000600, 0x00000600);
+	}
+
+	if (mclk.pll) {
+		ram_mask(fuc, 0x004168, 0x00000001, 0x00000000);
+		ram_mask(fuc, 0x004168, 0x00000100, 0x00000000);
+	} else {
+		ram_mask(fuc, 0x004000, 0x00000001, 0x00000000);
+		ram_mask(fuc, 0x004128, 0x00000001, 0x00000000);
+		ram_mask(fuc, 0x004128, 0x00000100, 0x00000000);
+	}
+
+	return 0;
+}
+
+static int
+gt215_ram_prog(struct nvkm_ram *base)
+{
+	struct gt215_ram *ram = gt215_ram(base);
+	struct gt215_ramfuc *fuc = &ram->fuc;
+	struct nvkm_device *device = ram->base.fb->subdev.device;
+	bool exec = nvkm_boolopt(device->cfgopt, "NvMemExec", true);
+
+	if (exec) {
+		nvkm_mask(device, 0x001534, 0x2, 0x2);
+
+		ram_exec(fuc, true);
+
+		/* Post-processing, avoids flicker */
+		nvkm_mask(device, 0x002504, 0x1, 0x0);
+		nvkm_mask(device, 0x001534, 0x2, 0x0);
+
+		nvkm_mask(device, 0x616308, 0x10, 0x10);
+		nvkm_mask(device, 0x616b08, 0x10, 0x10);
+	} else {
+		ram_exec(fuc, false);
+	}
+	return 0;
+}
+
+static void
+gt215_ram_tidy(struct nvkm_ram *base)
+{
+	struct gt215_ram *ram = gt215_ram(base);
+	ram_exec(&ram->fuc, false);
+}
+
+static int
+gt215_ram_init(struct nvkm_ram *base)
+{
+	struct gt215_ram *ram = gt215_ram(base);
+	gt215_link_train_init(ram);
+	return 0;
+}
+
+static void *
+gt215_ram_dtor(struct nvkm_ram *base)
+{
+	struct gt215_ram *ram = gt215_ram(base);
+	gt215_link_train_fini(ram);
+	return ram;
+}
+
+static const struct nvkm_ram_func
+gt215_ram_func = {
+	.dtor = gt215_ram_dtor,
+	.init = gt215_ram_init,
+	.calc = gt215_ram_calc,
+	.prog = gt215_ram_prog,
+	.tidy = gt215_ram_tidy,
+};
+
+int
+gt215_ram_new(struct nvkm_fb *fb, struct nvkm_ram **pram)
+{
+	struct gt215_ram *ram;
+	int ret, i;
+
+	if (!(ram = kzalloc(sizeof(*ram), GFP_KERNEL)))
+		return -ENOMEM;
+	*pram = &ram->base;
+
+	ret = nv50_ram_ctor(&gt215_ram_func, fb, &ram->base);
+	if (ret)
+		return ret;
+
+	ram->fuc.r_0x001610 = ramfuc_reg(0x001610);
+	ram->fuc.r_0x001700 = ramfuc_reg(0x001700);
+	ram->fuc.r_0x002504 = ramfuc_reg(0x002504);
+	ram->fuc.r_0x004000 = ramfuc_reg(0x004000);
+	ram->fuc.r_0x004004 = ramfuc_reg(0x004004);
+	ram->fuc.r_0x004018 = ramfuc_reg(0x004018);
+	ram->fuc.r_0x004128 = ramfuc_reg(0x004128);
+	ram->fuc.r_0x004168 = ramfuc_reg(0x004168);
+	ram->fuc.r_0x100080 = ramfuc_reg(0x100080);
+	ram->fuc.r_0x100200 = ramfuc_reg(0x100200);
+	ram->fuc.r_0x100210 = ramfuc_reg(0x100210);
+	for (i = 0; i < 9; i++)
+		ram->fuc.r_0x100220[i] = ramfuc_reg(0x100220 + (i * 4));
+	ram->fuc.r_0x100264 = ramfuc_reg(0x100264);
+	ram->fuc.r_0x1002d0 = ramfuc_reg(0x1002d0);
+	ram->fuc.r_0x1002d4 = ramfuc_reg(0x1002d4);
+	ram->fuc.r_0x1002dc = ramfuc_reg(0x1002dc);
+	ram->fuc.r_0x10053c = ramfuc_reg(0x10053c);
+	ram->fuc.r_0x1005a0 = ramfuc_reg(0x1005a0);
+	ram->fuc.r_0x1005a4 = ramfuc_reg(0x1005a4);
+	ram->fuc.r_0x100700 = ramfuc_reg(0x100700);
+	ram->fuc.r_0x100714 = ramfuc_reg(0x100714);
+	ram->fuc.r_0x100718 = ramfuc_reg(0x100718);
+	ram->fuc.r_0x10071c = ramfuc_reg(0x10071c);
+	ram->fuc.r_0x100720 = ramfuc_reg(0x100720);
+	ram->fuc.r_0x100760 = ramfuc_stride(0x100760, 4, ram->base.part_mask);
+	ram->fuc.r_0x1007a0 = ramfuc_stride(0x1007a0, 4, ram->base.part_mask);
+	ram->fuc.r_0x1007e0 = ramfuc_stride(0x1007e0, 4, ram->base.part_mask);
+	ram->fuc.r_0x100da0 = ramfuc_stride(0x100da0, 4, ram->base.part_mask);
+	ram->fuc.r_0x10f804 = ramfuc_reg(0x10f804);
+	ram->fuc.r_0x1110e0 = ramfuc_stride(0x1110e0, 4, ram->base.part_mask);
+	ram->fuc.r_0x111100 = ramfuc_reg(0x111100);
+	ram->fuc.r_0x111104 = ramfuc_reg(0x111104);
+	ram->fuc.r_0x1111e0 = ramfuc_reg(0x1111e0);
+	ram->fuc.r_0x111400 = ramfuc_reg(0x111400);
+	ram->fuc.r_0x611200 = ramfuc_reg(0x611200);
+
+	if (ram->base.ranks > 1) {
+		ram->fuc.r_mr[0] = ramfuc_reg2(0x1002c0, 0x1002c8);
+		ram->fuc.r_mr[1] = ramfuc_reg2(0x1002c4, 0x1002cc);
+		ram->fuc.r_mr[2] = ramfuc_reg2(0x1002e0, 0x1002e8);
+		ram->fuc.r_mr[3] = ramfuc_reg2(0x1002e4, 0x1002ec);
+	} else {
+		ram->fuc.r_mr[0] = ramfuc_reg(0x1002c0);
+		ram->fuc.r_mr[1] = ramfuc_reg(0x1002c4);
+		ram->fuc.r_mr[2] = ramfuc_reg(0x1002e0);
+		ram->fuc.r_mr[3] = ramfuc_reg(0x1002e4);
+	}
+	ram->fuc.r_gpio[0] = ramfuc_reg(0x00e104);
+	ram->fuc.r_gpio[1] = ramfuc_reg(0x00e108);
+	ram->fuc.r_gpio[2] = ramfuc_reg(0x00e120);
+	ram->fuc.r_gpio[3] = ramfuc_reg(0x00e124);
+
+	return 0;
+}