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, 641 insertions, 0 deletions

diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/fb/ramnv50.c b/drivers/gpu/drm/nouveau/nvkm/subdev/fb/ramnv50.c
new file mode 100644
index 000000000..7b1eb44ff
--- /dev/null
+++ b/drivers/gpu/drm/nouveau/nvkm/subdev/fb/ramnv50.c
@@ -0,0 +1,641 @@
+/*
+ * 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
+ */
+#define nv50_ram(p) container_of((p), struct nv50_ram, base)
+#include "ram.h"
+#include "ramseq.h"
+#include "nv50.h"
+
+#include <core/option.h>
+#include <subdev/bios.h>
+#include <subdev/bios/perf.h>
+#include <subdev/bios/pll.h>
+#include <subdev/bios/rammap.h>
+#include <subdev/bios/timing.h>
+#include <subdev/clk/pll.h>
+#include <subdev/gpio.h>
+
+struct nv50_ramseq {
+	struct hwsq base;
+	struct hwsq_reg r_0x002504;
+	struct hwsq_reg r_0x004008;
+	struct hwsq_reg r_0x00400c;
+	struct hwsq_reg r_0x00c040;
+	struct hwsq_reg r_0x100200;
+	struct hwsq_reg r_0x100210;
+	struct hwsq_reg r_0x10021c;
+	struct hwsq_reg r_0x1002d0;
+	struct hwsq_reg r_0x1002d4;
+	struct hwsq_reg r_0x1002dc;
+	struct hwsq_reg r_0x10053c;
+	struct hwsq_reg r_0x1005a0;
+	struct hwsq_reg r_0x1005a4;
+	struct hwsq_reg r_0x100710;
+	struct hwsq_reg r_0x100714;
+	struct hwsq_reg r_0x100718;
+	struct hwsq_reg r_0x10071c;
+	struct hwsq_reg r_0x100da0;
+	struct hwsq_reg r_0x100e20;
+	struct hwsq_reg r_0x100e24;
+	struct hwsq_reg r_0x611200;
+	struct hwsq_reg r_timing[9];
+	struct hwsq_reg r_mr[4];
+	struct hwsq_reg r_gpio[4];
+};
+
+struct nv50_ram {
+	struct nvkm_ram base;
+	struct nv50_ramseq hwsq;
+};
+
+#define T(t) cfg->timing_10_##t
+static int
+nv50_ram_timing_calc(struct nv50_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;
+	u32 cur2, cur4, cur7, cur8;
+	u8 unkt3b;
+
+	cur2 = nvkm_rd32(device, 0x100228);
+	cur4 = nvkm_rd32(device, 0x100230);
+	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;
+	}
+
+	/* XXX: N=1 is not proper statistics */
+	if (device->chipset == 0xa0) {
+		unkt3b = 0x19 + ram->base.next->bios.rammap_00_16_40;
+		timing[6] = (0x2d + T(CL) - T(CWL) +
+				ram->base.next->bios.rammap_00_16_40) << 16 |
+			    T(CWL) << 8 |
+			    (0x2f + T(CL) - T(CWL));
+	} else {
+		unkt3b = 0x16;
+		timing[6] = (0x2b + T(CL) - T(CWL)) << 16 |
+			    max_t(s8, T(CWL) - 2, 1) << 8 |
+			    (0x2e + T(CL) - T(CWL));
+	}
+
+	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 |
+		    (3 + T(CL) - T(CWL));
+	timing[2] = (T(CWL) - 1) << 24 |
+		    (T(RRD) << 16) |
+		    (T(RCDWR) << 8) |
+		    T(RCDRD);
+	timing[3] = (unkt3b - 2 + T(CL)) << 24 |
+		    unkt3b << 16 |
+		    (T(CL) - 1) << 8 |
+		    (T(CL) - 1);
+	timing[4] = (cur4 & 0xffff0000) |
+		    T(13) << 8 |
+		    T(13);
+	timing[5] = T(RFC) << 24 |
+		    max_t(u8, T(RCDRD), T(RCDWR)) << 16 |
+		    T(RP);
+	/* Timing 6 is already done above */
+	timing[7] = (cur7 & 0xff00ffff) | (T(CL) - 1) << 16;
+	timing[8] = (cur8 & 0xffffff00);
+
+	/* XXX: P.version == 1 only has DDR2 and GDDR3? */
+	if (ram->base.type == NVKM_RAM_TYPE_DDR2) {
+		timing[5] |= (T(CL) + 3) << 8;
+		timing[8] |= (T(CL) - 4);
+	} else
+	if (ram->base.type == NVKM_RAM_TYPE_GDDR3) {
+		timing[5] |= (T(CL) + 2) << 8;
+		timing[8] |= (T(CL) - 2);
+	}
+
+	nvkm_debug(subdev, " 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;
+}
+
+static int
+nv50_ram_timing_read(struct nv50_ram *ram, u32 *timing)
+{
+	unsigned int i;
+	struct nvbios_ramcfg *cfg = &ram->base.target.bios;
+	struct nvkm_subdev *subdev = &ram->base.fb->subdev;
+	struct nvkm_device *device = subdev->device;
+
+	for (i = 0; i <= 8; i++)
+		timing[i] = nvkm_rd32(device, 0x100220 + (i * 4));
+
+	/* Derive the bare minimum for the MR calculation to succeed */
+	cfg->timing_ver = 0x10;
+	T(CL) = (timing[3] & 0xff) + 1;
+
+	switch (ram->base.type) {
+	case NVKM_RAM_TYPE_DDR2:
+		T(CWL) = T(CL) - 1;
+		break;
+	case NVKM_RAM_TYPE_GDDR3:
+		T(CWL) = ((timing[2] & 0xff000000) >> 24) + 1;
+		break;
+	default:
+		return -ENOSYS;
+	}
+
+	T(WR) = ((timing[1] >> 24) & 0xff) - 1 - T(CWL);
+
+	return 0;
+}
+#undef T
+
+static void
+nvkm_sddr2_dll_reset(struct nv50_ramseq *hwsq)
+{
+	ram_mask(hwsq, mr[0], 0x100, 0x100);
+	ram_mask(hwsq, mr[0], 0x100, 0x000);
+	ram_nsec(hwsq, 24000);
+}
+
+static void
+nv50_ram_gpio(struct nv50_ramseq *hwsq, u8 tag, u32 val)
+{
+	struct nvkm_gpio *gpio = hwsq->base.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(hwsq, gpio[reg]);
+
+		if (gpio_val & (8 << sh))
+			val = !val;
+		if (!(func.log[1] & 1))
+			val = !val;
+
+		ram_mask(hwsq, gpio[reg], (0x3 << sh), ((val | 0x2) << sh));
+		ram_nsec(hwsq, 20000);
+	}
+}
+
+static int
+nv50_ram_calc(struct nvkm_ram *base, u32 freq)
+{
+	struct nv50_ram *ram = nv50_ram(base);
+	struct nv50_ramseq *hwsq = &ram->hwsq;
+	struct nvkm_subdev *subdev = &ram->base.fb->subdev;
+	struct nvkm_bios *bios = subdev->device->bios;
+	struct nvbios_perfE perfE;
+	struct nvbios_pll mpll;
+	struct nvkm_ram_data *next;
+	u8  ver, hdr, cnt, len, strap, size;
+	u32 data;
+	u32 r100da0, r004008, unk710, unk714, unk718, unk71c;
+	int N1, M1, N2, M2, P;
+	int ret, i;
+	u32 timing[9];
+
+	next = &ram->base.target;
+	next->freq = freq;
+	ram->base.next = next;
+
+	/* lookup closest matching performance table entry for frequency */
+	i = 0;
+	do {
+		data = nvbios_perfEp(bios, i++, &ver, &hdr, &cnt,
+				     &size, &perfE);
+		if (!data || (ver < 0x25 || ver >= 0x40) ||
+		    (size < 2)) {
+			nvkm_error(subdev, "invalid/missing perftab entry\n");
+			return -EINVAL;
+		}
+	} while (perfE.memory < freq);
+
+	nvbios_rammapEp_from_perf(bios, data, hdr, &next->bios);
+
+	/* 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_from_perf(bios, data + hdr, size, strap,
+			&next->bios);
+	if (!data) {
+		nvkm_error(subdev, "invalid/missing rammap entry ");
+		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 < 0x12) {
+			nvkm_error(subdev, "invalid/missing timing entry "
+				 "%02x %04x %02x %02x\n",
+				 strap, data, ver, hdr);
+			return -EINVAL;
+		}
+		nv50_ram_timing_calc(ram, timing);
+	} else {
+		nv50_ram_timing_read(ram, timing);
+	}
+
+	ret = ram_init(hwsq, subdev);
+	if (ret)
+		return ret;
+
+	/* Determine ram-specific MR values */
+	ram->base.mr[0] = ram_rd32(hwsq, mr[0]);
+	ram->base.mr[1] = ram_rd32(hwsq, mr[1]);
+	ram->base.mr[2] = ram_rd32(hwsq, mr[2]);
+
+	switch (ram->base.type) {
+	case NVKM_RAM_TYPE_GDDR3:
+		ret = nvkm_gddr3_calc(&ram->base);
+		break;
+	default:
+		ret = -ENOSYS;
+		break;
+	}
+
+	if (ret) {
+		nvkm_error(subdev, "Could not calculate MR\n");
+		return ret;
+	}
+
+	if (subdev->device->chipset <= 0x96 && !next->bios.ramcfg_00_03_02)
+		ram_mask(hwsq, 0x100710, 0x00000200, 0x00000000);
+
+	/* Always disable this bit during reclock */
+	ram_mask(hwsq, 0x100200, 0x00000800, 0x00000000);
+
+	ram_wait_vblank(hwsq);
+	ram_wr32(hwsq, 0x611200, 0x00003300);
+	ram_wr32(hwsq, 0x002504, 0x00000001); /* block fifo */
+	ram_nsec(hwsq, 8000);
+	ram_setf(hwsq, 0x10, 0x00); /* disable fb */
+	ram_wait(hwsq, 0x00, 0x01); /* wait for fb disabled */
+	ram_nsec(hwsq, 2000);
+
+	if (next->bios.timing_10_ODT)
+		nv50_ram_gpio(hwsq, 0x2e, 1);
+
+	ram_wr32(hwsq, 0x1002d4, 0x00000001); /* precharge */
+	ram_wr32(hwsq, 0x1002d0, 0x00000001); /* refresh */
+	ram_wr32(hwsq, 0x1002d0, 0x00000001); /* refresh */
+	ram_wr32(hwsq, 0x100210, 0x00000000); /* disable auto-refresh */
+	ram_wr32(hwsq, 0x1002dc, 0x00000001); /* enable self-refresh */
+
+	ret = nvbios_pll_parse(bios, 0x004008, &mpll);
+	mpll.vco2.max_freq = 0;
+	if (ret >= 0) {
+		ret = nv04_pll_calc(subdev, &mpll, freq,
+				    &N1, &M1, &N2, &M2, &P);
+		if (ret <= 0)
+			ret = -EINVAL;
+	}
+
+	if (ret < 0)
+		return ret;
+
+	/* XXX: 750MHz seems rather arbitrary */
+	if (freq <= 750000) {
+		r100da0 = 0x00000010;
+		r004008 = 0x90000000;
+	} else {
+		r100da0 = 0x00000000;
+		r004008 = 0x80000000;
+	}
+
+	r004008 |= (mpll.bias_p << 19) | (P << 22) | (P << 16);
+
+	ram_mask(hwsq, 0x00c040, 0xc000c000, 0x0000c000);
+	/* XXX: Is rammap_00_16_40 the DLL bit we've seen in GT215? Why does
+	 * it have a different rammap bit from DLLoff? */
+	ram_mask(hwsq, 0x004008, 0x00004200, 0x00000200 |
+			next->bios.rammap_00_16_40 << 14);
+	ram_mask(hwsq, 0x00400c, 0x0000ffff, (N1 << 8) | M1);
+	ram_mask(hwsq, 0x004008, 0x91ff0000, r004008);
+
+	/* XXX: GDDR3 only? */
+	if (subdev->device->chipset >= 0x92)
+		ram_wr32(hwsq, 0x100da0, r100da0);
+
+	nv50_ram_gpio(hwsq, 0x18, !next->bios.ramcfg_FBVDDQ);
+	ram_nsec(hwsq, 64000); /*XXX*/
+	ram_nsec(hwsq, 32000); /*XXX*/
+
+	ram_mask(hwsq, 0x004008, 0x00002200, 0x00002000);
+
+	ram_wr32(hwsq, 0x1002dc, 0x00000000); /* disable self-refresh */
+	ram_wr32(hwsq, 0x1002d4, 0x00000001); /* disable self-refresh */
+	ram_wr32(hwsq, 0x100210, 0x80000000); /* enable auto-refresh */
+
+	ram_nsec(hwsq, 12000);
+
+	switch (ram->base.type) {
+	case NVKM_RAM_TYPE_DDR2:
+		ram_nuke(hwsq, mr[0]); /* force update */
+		ram_mask(hwsq, mr[0], 0x000, 0x000);
+		break;
+	case NVKM_RAM_TYPE_GDDR3:
+		ram_nuke(hwsq, mr[1]); /* force update */
+		ram_wr32(hwsq, mr[1], ram->base.mr[1]);
+		ram_nuke(hwsq, mr[0]); /* force update */
+		ram_wr32(hwsq, mr[0], ram->base.mr[0]);
+		break;
+	default:
+		break;
+	}
+
+	ram_mask(hwsq, timing[3], 0xffffffff, timing[3]);
+	ram_mask(hwsq, timing[1], 0xffffffff, timing[1]);
+	ram_mask(hwsq, timing[6], 0xffffffff, timing[6]);
+	ram_mask(hwsq, timing[7], 0xffffffff, timing[7]);
+	ram_mask(hwsq, timing[8], 0xffffffff, timing[8]);
+	ram_mask(hwsq, timing[0], 0xffffffff, timing[0]);
+	ram_mask(hwsq, timing[2], 0xffffffff, timing[2]);
+	ram_mask(hwsq, timing[4], 0xffffffff, timing[4]);
+	ram_mask(hwsq, timing[5], 0xffffffff, timing[5]);
+
+	if (!next->bios.ramcfg_00_03_02)
+		ram_mask(hwsq, 0x10021c, 0x00010000, 0x00000000);
+	ram_mask(hwsq, 0x100200, 0x00001000, !next->bios.ramcfg_00_04_02 << 12);
+
+	/* XXX: A lot of this could be "chipset"/"ram type" specific stuff */
+	unk710  = ram_rd32(hwsq, 0x100710) & ~0x00000100;
+	unk714  = ram_rd32(hwsq, 0x100714) & ~0xf0000020;
+	unk718  = ram_rd32(hwsq, 0x100718) & ~0x00000100;
+	unk71c  = ram_rd32(hwsq, 0x10071c) & ~0x00000100;
+	if (subdev->device->chipset <= 0x96) {
+		unk710 &= ~0x0000006e;
+		unk714 &= ~0x00000100;
+
+		if (!next->bios.ramcfg_00_03_08)
+			unk710 |= 0x00000060;
+		if (!next->bios.ramcfg_FBVDDQ)
+			unk714 |= 0x00000100;
+		if ( next->bios.ramcfg_00_04_04)
+			unk710 |= 0x0000000e;
+	} else {
+		unk710 &= ~0x00000001;
+
+		if (!next->bios.ramcfg_00_03_08)
+			unk710 |= 0x00000001;
+	}
+
+	if ( next->bios.ramcfg_00_03_01)
+		unk71c |= 0x00000100;
+	if ( next->bios.ramcfg_00_03_02)
+		unk710 |= 0x00000100;
+	if (!next->bios.ramcfg_00_03_08)
+		unk714 |= 0x00000020;
+	if ( next->bios.ramcfg_00_04_04)
+		unk714 |= 0x70000000;
+	if ( next->bios.ramcfg_00_04_20)
+		unk718 |= 0x00000100;
+
+	ram_mask(hwsq, 0x100714, 0xffffffff, unk714);
+	ram_mask(hwsq, 0x10071c, 0xffffffff, unk71c);
+	ram_mask(hwsq, 0x100718, 0xffffffff, unk718);
+	ram_mask(hwsq, 0x100710, 0xffffffff, unk710);
+
+	/* XXX: G94 does not even test these regs in trace. Harmless we do it,
+	 * but why is it omitted? */
+	if (next->bios.rammap_00_16_20) {
+		ram_wr32(hwsq, 0x1005a0, next->bios.ramcfg_00_07 << 16 |
+					 next->bios.ramcfg_00_06 << 8 |
+					 next->bios.ramcfg_00_05);
+		ram_wr32(hwsq, 0x1005a4, next->bios.ramcfg_00_09 << 8 |
+					 next->bios.ramcfg_00_08);
+		ram_mask(hwsq, 0x10053c, 0x00001000, 0x00000000);
+	} else {
+		ram_mask(hwsq, 0x10053c, 0x00001000, 0x00001000);
+	}
+	ram_mask(hwsq, mr[1], 0xffffffff, ram->base.mr[1]);
+
+	if (!next->bios.timing_10_ODT)
+		nv50_ram_gpio(hwsq, 0x2e, 0);
+
+	/* Reset DLL */
+	if (!next->bios.ramcfg_DLLoff)
+		nvkm_sddr2_dll_reset(hwsq);
+
+	ram_setf(hwsq, 0x10, 0x01); /* enable fb */
+	ram_wait(hwsq, 0x00, 0x00); /* wait for fb enabled */
+	ram_wr32(hwsq, 0x611200, 0x00003330);
+	ram_wr32(hwsq, 0x002504, 0x00000000); /* un-block fifo */
+
+	if (next->bios.rammap_00_17_02)
+		ram_mask(hwsq, 0x100200, 0x00000800, 0x00000800);
+	if (!next->bios.rammap_00_16_40)
+		ram_mask(hwsq, 0x004008, 0x00004000, 0x00000000);
+	if (next->bios.ramcfg_00_03_02)
+		ram_mask(hwsq, 0x10021c, 0x00010000, 0x00010000);
+	if (subdev->device->chipset <= 0x96 && next->bios.ramcfg_00_03_02)
+		ram_mask(hwsq, 0x100710, 0x00000200, 0x00000200);
+
+	return 0;
+}
+
+static int
+nv50_ram_prog(struct nvkm_ram *base)
+{
+	struct nv50_ram *ram = nv50_ram(base);
+	struct nvkm_device *device = ram->base.fb->subdev.device;
+	ram_exec(&ram->hwsq, nvkm_boolopt(device->cfgopt, "NvMemExec", true));
+	return 0;
+}
+
+static void
+nv50_ram_tidy(struct nvkm_ram *base)
+{
+	struct nv50_ram *ram = nv50_ram(base);
+	ram_exec(&ram->hwsq, false);
+}
+
+static const struct nvkm_ram_func
+nv50_ram_func = {
+	.calc = nv50_ram_calc,
+	.prog = nv50_ram_prog,
+	.tidy = nv50_ram_tidy,
+};
+
+static u32
+nv50_fb_vram_rblock(struct nvkm_ram *ram)
+{
+	struct nvkm_subdev *subdev = &ram->fb->subdev;
+	struct nvkm_device *device = subdev->device;
+	int colbits, rowbitsa, rowbitsb, banks;
+	u64 rowsize, predicted;
+	u32 r0, r4, rt, rblock_size;
+
+	r0 = nvkm_rd32(device, 0x100200);
+	r4 = nvkm_rd32(device, 0x100204);
+	rt = nvkm_rd32(device, 0x100250);
+	nvkm_debug(subdev, "memcfg %08x %08x %08x %08x\n",
+		   r0, r4, rt, nvkm_rd32(device, 0x001540));
+
+	colbits  =  (r4 & 0x0000f000) >> 12;
+	rowbitsa = ((r4 & 0x000f0000) >> 16) + 8;
+	rowbitsb = ((r4 & 0x00f00000) >> 20) + 8;
+	banks    = 1 << (((r4 & 0x03000000) >> 24) + 2);
+
+	rowsize = ram->parts * banks * (1 << colbits) * 8;
+	predicted = rowsize << rowbitsa;
+	if (r0 & 0x00000004)
+		predicted += rowsize << rowbitsb;
+
+	if (predicted != ram->size) {
+		nvkm_warn(subdev, "memory controller reports %d MiB VRAM\n",
+			  (u32)(ram->size >> 20));
+	}
+
+	rblock_size = rowsize;
+	if (rt & 1)
+		rblock_size *= 3;
+
+	nvkm_debug(subdev, "rblock %d bytes\n", rblock_size);
+	return rblock_size;
+}
+
+int
+nv50_ram_ctor(const struct nvkm_ram_func *func,
+	      struct nvkm_fb *fb, struct nvkm_ram *ram)
+{
+	struct nvkm_device *device = fb->subdev.device;
+	struct nvkm_bios *bios = device->bios;
+	const u32 rsvd_head = ( 256 * 1024); /* vga memory */
+	const u32 rsvd_tail = (1024 * 1024); /* vbios etc */
+	u64 size = nvkm_rd32(device, 0x10020c);
+	enum nvkm_ram_type type = NVKM_RAM_TYPE_UNKNOWN;
+	int ret;
+
+	switch (nvkm_rd32(device, 0x100714) & 0x00000007) {
+	case 0: type = NVKM_RAM_TYPE_DDR1; break;
+	case 1:
+		if (nvkm_fb_bios_memtype(bios) == NVKM_RAM_TYPE_DDR3)
+			type = NVKM_RAM_TYPE_DDR3;
+		else
+			type = NVKM_RAM_TYPE_DDR2;
+		break;
+	case 2: type = NVKM_RAM_TYPE_GDDR3; break;
+	case 3: type = NVKM_RAM_TYPE_GDDR4; break;
+	case 4: type = NVKM_RAM_TYPE_GDDR5; break;
+	default:
+		break;
+	}
+
+	size = (size & 0x000000ff) << 32 | (size & 0xffffff00);
+
+	ret = nvkm_ram_ctor(func, fb, type, size, ram);
+	if (ret)
+		return ret;
+
+	ram->part_mask = (nvkm_rd32(device, 0x001540) & 0x00ff0000) >> 16;
+	ram->parts = hweight8(ram->part_mask);
+	ram->ranks = (nvkm_rd32(device, 0x100200) & 0x4) ? 2 : 1;
+	nvkm_mm_fini(&ram->vram);
+
+	return nvkm_mm_init(&ram->vram, NVKM_RAM_MM_NORMAL,
+			    rsvd_head >> NVKM_RAM_MM_SHIFT,
+			    (size - rsvd_head - rsvd_tail) >> NVKM_RAM_MM_SHIFT,
+			    nv50_fb_vram_rblock(ram) >> NVKM_RAM_MM_SHIFT);
+}
+
+int
+nv50_ram_new(struct nvkm_fb *fb, struct nvkm_ram **pram)
+{
+	struct nv50_ram *ram;
+	int ret, i;
+
+	if (!(ram = kzalloc(sizeof(*ram), GFP_KERNEL)))
+		return -ENOMEM;
+	*pram = &ram->base;
+
+	ret = nv50_ram_ctor(&nv50_ram_func, fb, &ram->base);
+	if (ret)
+		return ret;
+
+	ram->hwsq.r_0x002504 = hwsq_reg(0x002504);
+	ram->hwsq.r_0x00c040 = hwsq_reg(0x00c040);
+	ram->hwsq.r_0x004008 = hwsq_reg(0x004008);
+	ram->hwsq.r_0x00400c = hwsq_reg(0x00400c);
+	ram->hwsq.r_0x100200 = hwsq_reg(0x100200);
+	ram->hwsq.r_0x100210 = hwsq_reg(0x100210);
+	ram->hwsq.r_0x10021c = hwsq_reg(0x10021c);
+	ram->hwsq.r_0x1002d0 = hwsq_reg(0x1002d0);
+	ram->hwsq.r_0x1002d4 = hwsq_reg(0x1002d4);
+	ram->hwsq.r_0x1002dc = hwsq_reg(0x1002dc);
+	ram->hwsq.r_0x10053c = hwsq_reg(0x10053c);
+	ram->hwsq.r_0x1005a0 = hwsq_reg(0x1005a0);
+	ram->hwsq.r_0x1005a4 = hwsq_reg(0x1005a4);
+	ram->hwsq.r_0x100710 = hwsq_reg(0x100710);
+	ram->hwsq.r_0x100714 = hwsq_reg(0x100714);
+	ram->hwsq.r_0x100718 = hwsq_reg(0x100718);
+	ram->hwsq.r_0x10071c = hwsq_reg(0x10071c);
+	ram->hwsq.r_0x100da0 = hwsq_stride(0x100da0, 4, ram->base.part_mask);
+	ram->hwsq.r_0x100e20 = hwsq_reg(0x100e20);
+	ram->hwsq.r_0x100e24 = hwsq_reg(0x100e24);
+	ram->hwsq.r_0x611200 = hwsq_reg(0x611200);
+
+	for (i = 0; i < 9; i++)
+		ram->hwsq.r_timing[i] = hwsq_reg(0x100220 + (i * 0x04));
+
+	if (ram->base.ranks > 1) {
+		ram->hwsq.r_mr[0] = hwsq_reg2(0x1002c0, 0x1002c8);
+		ram->hwsq.r_mr[1] = hwsq_reg2(0x1002c4, 0x1002cc);
+		ram->hwsq.r_mr[2] = hwsq_reg2(0x1002e0, 0x1002e8);
+		ram->hwsq.r_mr[3] = hwsq_reg2(0x1002e4, 0x1002ec);
+	} else {
+		ram->hwsq.r_mr[0] = hwsq_reg(0x1002c0);
+		ram->hwsq.r_mr[1] = hwsq_reg(0x1002c4);
+		ram->hwsq.r_mr[2] = hwsq_reg(0x1002e0);
+		ram->hwsq.r_mr[3] = hwsq_reg(0x1002e4);
+	}
+
+	ram->hwsq.r_gpio[0] = hwsq_reg(0x00e104);
+	ram->hwsq.r_gpio[1] = hwsq_reg(0x00e108);
+	ram->hwsq.r_gpio[2] = hwsq_reg(0x00e120);
+	ram->hwsq.r_gpio[3] = hwsq_reg(0x00e124);
+
+	return 0;
+}