Merge inotify strcpy hardening.

2024-12-28 16:52:18 +00:00 · 2024-12-18 11:35:20 +01:00 · 2024-12-18 11:35:20 +01:00 · 71358f64c4
commit 71358f64c4
parent 3f1063103c b8f2688258
385 changed files with 5297 additions and 3216 deletions
--- a/Documentation/ABI/testing/sysfs-class-watchdog
+++ b/Documentation/ABI/testing/sysfs-class-watchdog
@ -76,7 +76,7 @@ Description:
 		timeout when the pretimeout interrupt is delivered.  Pretimeout
 		is an optional feature.

-What:		/sys/class/watchdog/watchdogn/pretimeout_avaialable_governors
+What:		/sys/class/watchdog/watchdogn/pretimeout_available_governors
 Date:		February 2017
 Contact:	Wim Van Sebroeck <wim@iguana.be>
 Description:
--- a/Documentation/arch/arm64/silicon-errata.rst
+++ b/Documentation/arch/arm64/silicon-errata.rst
@ -255,8 +255,9 @@ stable kernels.
 +----------------+-----------------+-----------------+-----------------------------+
 | Hisilicon      | Hip08 SMMU PMCG | #162001800      | N/A                         |
 +----------------+-----------------+-----------------+-----------------------------+
-| Hisilicon      | Hip{08,09,10,10C| #162001900      | N/A                         |
-|                | ,11} SMMU PMCG  |                 |                             |
+| Hisilicon      | Hip{08,09,09A,10| #162001900      | N/A                         |
+|                | ,10C,11}        |                 |                             |
+|                | SMMU PMCG       |                 |                             |
 +----------------+-----------------+-----------------+-----------------------------+
 | Hisilicon      | Hip09           | #162100801      | HISILICON_ERRATUM_162100801 |
 +----------------+-----------------+-----------------+-----------------------------+
--- a/Documentation/devicetree/bindings/power/mediatek,power-controller.yaml
+++ b/Documentation/devicetree/bindings/power/mediatek,power-controller.yaml
@ -55,6 +55,10 @@ patternProperties:
            patternProperties:
              "^power-domain@[0-9a-f]+$":
                $ref: "#/$defs/power-domain-node"
+                patternProperties:
+                  "^power-domain@[0-9a-f]+$":
+                    $ref: "#/$defs/power-domain-node"
+                    unevaluatedProperties: false
                unevaluatedProperties: false
            unevaluatedProperties: false
        unevaluatedProperties: false
--- a/Documentation/devicetree/bindings/watchdog/airoha,en7581-wdt.yaml
+++ b/Documentation/devicetree/bindings/watchdog/airoha,en7581-wdt.yaml
@ -0,0 +1,47 @@
+# SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause)
+%YAML 1.2
+---
+$id: http://devicetree.org/schemas/watchdog/airoha,en7581-wdt.yaml#
+$schema: http://devicetree.org/meta-schemas/core.yaml#
+
+title: Airoha EN7581 Watchdog Timer
+
+maintainers:
+  - Christian Marangi <ansuelsmth@gmail.com>
+
+allOf:
+  - $ref: watchdog.yaml#
+
+properties:
+  compatible:
+    const: airoha,en7581-wdt
+
+  reg:
+    maxItems: 1
+
+  clocks:
+    description: BUS clock (timer ticks at half the BUS clock)
+    maxItems: 1
+
+  clock-names:
+    const: bus
+
+required:
+  - compatible
+  - reg
+  - clocks
+  - clock-names
+
+unevaluatedProperties: false
+
+examples:
+  - |
+    #include <dt-bindings/clock/en7523-clk.h>
+
+    watchdog@1fbf0100 {
+        compatible = "airoha,en7581-wdt";
+        reg = <0x1fbf0100 0x3c>;
+
+        clocks = <&scuclk EN7523_CLK_BUS>;
+        clock-names = "bus";
+    };
--- a/Documentation/devicetree/bindings/watchdog/fsl-imx-wdt.yaml
+++ b/Documentation/devicetree/bindings/watchdog/fsl-imx-wdt.yaml
@ -48,6 +48,8 @@ properties:
  clocks:
    maxItems: 1

+  big-endian: true
+
  fsl,ext-reset-output:
    $ref: /schemas/types.yaml#/definitions/flag
    description: |
@ -93,6 +95,18 @@ allOf:
      properties:
        fsl,suspend-in-wait: false

+  - if:
+      not:
+        properties:
+          compatible:
+            contains:
+              enum:
+                - fsl,ls1012a-wdt
+                - fsl,ls1043a-wdt
+    then:
+      properties:
+        big-endian: false
+
 unevaluatedProperties: false

 examples:
--- a/Documentation/devicetree/bindings/watchdog/qcom-wdt.yaml
+++ b/Documentation/devicetree/bindings/watchdog/qcom-wdt.yaml
@ -26,6 +26,8 @@ properties:
              - qcom,apss-wdt-msm8994
              - qcom,apss-wdt-qcm2290
              - qcom,apss-wdt-qcs404
+              - qcom,apss-wdt-qcs615
+              - qcom,apss-wdt-qcs8300
              - qcom,apss-wdt-sa8255p
              - qcom,apss-wdt-sa8775p
              - qcom,apss-wdt-sc7180
--- a/Documentation/devicetree/bindings/watchdog/samsung-wdt.yaml
+++ b/Documentation/devicetree/bindings/watchdog/samsung-wdt.yaml
@ -26,6 +26,7 @@ properties:
          - samsung,exynos7-wdt                   # for Exynos7
          - samsung,exynos850-wdt                 # for Exynos850
          - samsung,exynosautov9-wdt              # for Exynosautov9
+          - samsung,exynosautov920-wdt            # for Exynosautov920
      - items:
          - enum:
              - tesla,fsd-wdt
@ -77,6 +78,7 @@ allOf:
              - samsung,exynos7-wdt
              - samsung,exynos850-wdt
              - samsung,exynosautov9-wdt
+              - samsung,exynosautov920-wdt
    then:
      required:
        - samsung,syscon-phandle
@ -88,6 +90,7 @@ allOf:
              - google,gs101-wdt
              - samsung,exynos850-wdt
              - samsung,exynosautov9-wdt
+              - samsung,exynosautov920-wdt
    then:
      properties:
        clocks:
--- a/Documentation/networking/bareudp.rst
+++ b/Documentation/networking/bareudp.rst
@ -6,16 +6,17 @@ Bare UDP Tunnelling Module Documentation

 There are various L3 encapsulation standards using UDP being discussed to
 leverage the UDP based load balancing capability of different networks.
-MPLSoUDP (__ https://tools.ietf.org/html/rfc7510) is one among them.
+MPLSoUDP (https://tools.ietf.org/html/rfc7510) is one among them.

 The Bareudp tunnel module provides a generic L3 encapsulation support for
 tunnelling different L3 protocols like MPLS, IP, NSH etc. inside a UDP tunnel.

 Special Handling
 ----------------
+
 The bareudp device supports special handling for MPLS & IP as they can have
 multiple ethertypes.
-MPLS procotcol can have ethertypes ETH_P_MPLS_UC  (unicast) & ETH_P_MPLS_MC (multicast).
+The MPLS protocol can have ethertypes ETH_P_MPLS_UC (unicast) & ETH_P_MPLS_MC (multicast).
 IP protocol can have ethertypes ETH_P_IP (v4) & ETH_P_IPV6 (v6).
 This special handling can be enabled only for ethertypes ETH_P_IP & ETH_P_MPLS_UC
 with a flag called multiproto mode.
@ -52,7 +53,7 @@ be enabled explicitly with the "multiproto" flag.
 3) Device Usage

 The bareudp device could be used along with OVS or flower filter in TC.
-The OVS or TC flower layer must set the tunnel information in SKB dst field before
-sending packet buffer to the bareudp device for transmission. On reception the
-bareudp device extracts and stores the tunnel information in SKB dst field before
+The OVS or TC flower layer must set the tunnel information in the SKB dst field before
+sending the packet buffer to the bareudp device for transmission. On reception, the
+bareUDP device extracts and stores the tunnel information in the SKB dst field before
 passing the packet buffer to the network stack.
--- a/Documentation/watchdog/watchdog-parameters.rst
+++ b/Documentation/watchdog/watchdog-parameters.rst
@ -120,16 +120,6 @@ coh901327_wdt:

 -------------------------------------------------

-cpu5wdt:
-    port:
-	base address of watchdog card, default is 0x91
-    verbose:
-	be verbose, default is 0 (no)
-    ticks:
-	count down ticks, default is 10000
-
-------------------------------------------------
-
 cpwd:
    wd0_timeout:
 	Default watchdog0 timeout in 1/10secs
--- a/5
+++ b/5
@ -3376,6 +3376,8 @@ S:	Maintained
 T:	git git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux.git
 F:	Documentation/arch/arm64/
 F:	arch/arm64/
+F:	drivers/virt/coco/arm-cca-guest/
+F:	drivers/virt/coco/pkvm-guest/
 F:	tools/testing/selftests/arm64/
 X:	arch/arm64/boot/dts/

@ -16267,6 +16269,7 @@ F:	Documentation/devicetree/bindings/net/
 F:	Documentation/networking/net_cachelines/net_device.rst
 F:	drivers/connector/
 F:	drivers/net/
+F:	drivers/ptp/
 F:	include/dt-bindings/net/
 F:	include/linux/cn_proc.h
 F:	include/linux/etherdevice.h
@ -22407,7 +22410,7 @@ F:	drivers/char/hw_random/jh7110-trng.c

 STARFIVE WATCHDOG DRIVER
 M:	Xingyu Wu <xingyu.wu@starfivetech.com>
-M:	Samin Guo <samin.guo@starfivetech.com>
+M:	Ziv Xu <ziv.xu@starfivetech.com>
 S:	Supported
 F:	Documentation/devicetree/bindings/watchdog/starfive*
 F:	drivers/watchdog/starfive-wdt.c
--- a/2
+++ b/2
@ -2,7 +2,7 @@
 VERSION = 6
 PATCHLEVEL = 13
 SUBLEVEL = 0
-EXTRAVERSION = -rc1
+EXTRAVERSION = -rc2
 NAME = Baby Opossum Posse

 # *DOCUMENTATION*
--- a/arch/arm64/include/asm/cpucaps.h
+++ b/arch/arm64/include/asm/cpucaps.h
@ -44,6 +44,8 @@ cpucap_is_possible(const unsigned int cap)
 		return IS_ENABLED(CONFIG_ARM64_TLB_RANGE);
 	case ARM64_HAS_S1POE:
 		return IS_ENABLED(CONFIG_ARM64_POE);
+	case ARM64_HAS_GCS:
+		return IS_ENABLED(CONFIG_ARM64_GCS);
 	case ARM64_UNMAP_KERNEL_AT_EL0:
 		return IS_ENABLED(CONFIG_UNMAP_KERNEL_AT_EL0);
 	case ARM64_WORKAROUND_843419:
--- a/arch/arm64/include/asm/cpufeature.h
+++ b/arch/arm64/include/asm/cpufeature.h
@ -847,8 +847,7 @@ static inline bool system_supports_poe(void)

 static inline bool system_supports_gcs(void)
 {
-	return IS_ENABLED(CONFIG_ARM64_GCS) &&
-		alternative_has_cap_unlikely(ARM64_HAS_GCS);
+	return alternative_has_cap_unlikely(ARM64_HAS_GCS);
 }

 static inline bool system_supports_haft(void)
--- a/arch/arm64/include/asm/mman.h
+++ b/arch/arm64/include/asm/mman.h
@ -7,6 +7,7 @@
 #ifndef BUILD_VDSO
 #include <linux/compiler.h>
 #include <linux/fs.h>
+#include <linux/hugetlb.h>
 #include <linux/shmem_fs.h>
 #include <linux/types.h>

@ -44,7 +45,7 @@ static inline unsigned long arch_calc_vm_flag_bits(struct file *file,
 	if (system_supports_mte()) {
 		if (flags & (MAP_ANONYMOUS | MAP_HUGETLB))
 			return VM_MTE_ALLOWED;
-		if (shmem_file(file))
+		if (shmem_file(file) || is_file_hugepages(file))
 			return VM_MTE_ALLOWED;
 	}

--- a/arch/arm64/kernel/patching.c
+++ b/arch/arm64/kernel/patching.c
@ -30,20 +30,17 @@ static bool is_image_text(unsigned long addr)

 static void __kprobes *patch_map(void *addr, int fixmap)
 {
-	unsigned long uintaddr = (uintptr_t) addr;
-	bool image = is_image_text(uintaddr);
-	struct page *page;
+	phys_addr_t phys;

-	if (image)
-		page = phys_to_page(__pa_symbol(addr));
-	else if (IS_ENABLED(CONFIG_EXECMEM))
-		page = vmalloc_to_page(addr);
-	else
-		return addr;
+	if (is_image_text((unsigned long)addr)) {
+		phys = __pa_symbol(addr);
+	} else {
+		struct page *page = vmalloc_to_page(addr);
+		BUG_ON(!page);
+		phys = page_to_phys(page) + offset_in_page(addr);
+	}

-	BUG_ON(!page);
-	return (void *)set_fixmap_offset(fixmap, page_to_phys(page) +
-			(uintaddr & ~PAGE_MASK));
+	return (void *)set_fixmap_offset(fixmap, phys);
 }

 static void __kprobes patch_unmap(int fixmap)
--- a/arch/arm64/kernel/ptrace.c
+++ b/arch/arm64/kernel/ptrace.c
@ -720,6 +720,8 @@ static int fpmr_set(struct task_struct *target, const struct user_regset *regset
 	if (!system_supports_fpmr())
 		return -EINVAL;

+	fpmr = target->thread.uw.fpmr;
+
 	ret = user_regset_copyin(&pos, &count, &kbuf, &ubuf, &fpmr, 0, count);
 	if (ret)
 		return ret;
@ -1427,7 +1429,7 @@ static int tagged_addr_ctrl_get(struct task_struct *target,
 {
 	long ctrl = get_tagged_addr_ctrl(target);

-	if (IS_ERR_VALUE(ctrl))
+	if (WARN_ON_ONCE(IS_ERR_VALUE(ctrl)))
 		return ctrl;

 	return membuf_write(&to, &ctrl, sizeof(ctrl));
@ -1441,6 +1443,10 @@ static int tagged_addr_ctrl_set(struct task_struct *target, const struct
 	int ret;
 	long ctrl;

+	ctrl = get_tagged_addr_ctrl(target);
+	if (WARN_ON_ONCE(IS_ERR_VALUE(ctrl)))
+		return ctrl;
+
 	ret = user_regset_copyin(&pos, &count, &kbuf, &ubuf, &ctrl, 0, -1);
 	if (ret)
 		return ret;
@ -1472,6 +1478,8 @@ static int poe_set(struct task_struct *target, const struct
 	if (!system_supports_poe())
 		return -EINVAL;

+	ctrl = target->thread.por_el0;
+
 	ret = user_regset_copyin(&pos, &count, &kbuf, &ubuf, &ctrl, 0, -1);
 	if (ret)
 		return ret;
@ -1483,6 +1491,22 @@ static int poe_set(struct task_struct *target, const struct
 #endif

 #ifdef CONFIG_ARM64_GCS
+static void task_gcs_to_user(struct user_gcs *user_gcs,
+			     const struct task_struct *target)
+{
+	user_gcs->features_enabled = target->thread.gcs_el0_mode;
+	user_gcs->features_locked = target->thread.gcs_el0_locked;
+	user_gcs->gcspr_el0 = target->thread.gcspr_el0;
+}
+
+static void task_gcs_from_user(struct task_struct *target,
+			       const struct user_gcs *user_gcs)
+{
+	target->thread.gcs_el0_mode = user_gcs->features_enabled;
+	target->thread.gcs_el0_locked = user_gcs->features_locked;
+	target->thread.gcspr_el0 = user_gcs->gcspr_el0;
+}
+
 static int gcs_get(struct task_struct *target,
 		   const struct user_regset *regset,
 		   struct membuf to)
@ -1495,9 +1519,7 @@ static int gcs_get(struct task_struct *target,
 	if (target == current)
 		gcs_preserve_current_state();

-	user_gcs.features_enabled = target->thread.gcs_el0_mode;
-	user_gcs.features_locked = target->thread.gcs_el0_locked;
-	user_gcs.gcspr_el0 = target->thread.gcspr_el0;
+	task_gcs_to_user(&user_gcs, target);

 	return membuf_write(&to, &user_gcs, sizeof(user_gcs));
 }
@ -1513,6 +1535,8 @@ static int gcs_set(struct task_struct *target, const struct
 	if (!system_supports_gcs())
 		return -EINVAL;

+	task_gcs_to_user(&user_gcs, target);
+
 	ret = user_regset_copyin(&pos, &count, &kbuf, &ubuf, &user_gcs, 0, -1);
 	if (ret)
 		return ret;
@ -1520,9 +1544,7 @@ static int gcs_set(struct task_struct *target, const struct
 	if (user_gcs.features_enabled & ~PR_SHADOW_STACK_SUPPORTED_STATUS_MASK)
 		return -EINVAL;

-	target->thread.gcs_el0_mode = user_gcs.features_enabled;
-	target->thread.gcs_el0_locked = user_gcs.features_locked;
-	target->thread.gcspr_el0 = user_gcs.gcspr_el0;
+	task_gcs_from_user(target, &user_gcs);

 	return 0;
 }
--- a/arch/arm64/mm/context.c
+++ b/arch/arm64/mm/context.c
@ -32,9 +32,9 @@ static unsigned long nr_pinned_asids;
 static unsigned long *pinned_asid_map;

 #define ASID_MASK		(~GENMASK(asid_bits - 1, 0))
-#define ASID_FIRST_VERSION	(1UL << asid_bits)
+#define ASID_FIRST_VERSION	(1UL << 16)

-#define NUM_USER_ASIDS		ASID_FIRST_VERSION
+#define NUM_USER_ASIDS		(1UL << asid_bits)
 #define ctxid2asid(asid)	((asid) & ~ASID_MASK)
 #define asid2ctxid(asid, genid)	((asid) | (genid))

--- a/arch/arm64/mm/copypage.c
+++ b/arch/arm64/mm/copypage.c
@ -30,11 +30,13 @@ void copy_highpage(struct page *to, struct page *from)
 	if (!system_supports_mte())
 		return;

-	if (folio_test_hugetlb(src) &&
-	    folio_test_hugetlb_mte_tagged(src)) {
-		if (!folio_try_hugetlb_mte_tagging(dst))
+	if (folio_test_hugetlb(src)) {
+		if (!folio_test_hugetlb_mte_tagged(src) ||
+		    from != folio_page(src, 0))
 			return;

+		WARN_ON_ONCE(!folio_try_hugetlb_mte_tagging(dst));
+
 		/*
 		 * Populate tags for all subpages.
 		 *
--- a/arch/arm64/mm/init.c
+++ b/arch/arm64/mm/init.c
@ -117,15 +117,6 @@ static void __init arch_reserve_crashkernel(void)

 static phys_addr_t __init max_zone_phys(phys_addr_t zone_limit)
 {
-	/**
-	 * Information we get from firmware (e.g. DT dma-ranges) describe DMA
-	 * bus constraints. Devices using DMA might have their own limitations.
-	 * Some of them rely on DMA zone in low 32-bit memory. Keep low RAM
-	 * DMA zone on platforms that have RAM there.
-	 */
-	if (memblock_start_of_DRAM() < U32_MAX)
-		zone_limit = min(zone_limit, U32_MAX);
-
 	return min(zone_limit, memblock_end_of_DRAM() - 1) + 1;
 }

@ -141,6 +132,14 @@ static void __init zone_sizes_init(void)
 	acpi_zone_dma_limit = acpi_iort_dma_get_max_cpu_address();
 	dt_zone_dma_limit = of_dma_get_max_cpu_address(NULL);
 	zone_dma_limit = min(dt_zone_dma_limit, acpi_zone_dma_limit);
+	/*
+	 * Information we get from firmware (e.g. DT dma-ranges) describe DMA
+	 * bus constraints. Devices using DMA might have their own limitations.
+	 * Some of them rely on DMA zone in low 32-bit memory. Keep low RAM
+	 * DMA zone on platforms that have RAM there.
+	 */
+	if (memblock_start_of_DRAM() < U32_MAX)
+		zone_dma_limit = min(zone_dma_limit, U32_MAX);
 	arm64_dma_phys_limit = max_zone_phys(zone_dma_limit);
 	max_zone_pfns[ZONE_DMA] = PFN_DOWN(arm64_dma_phys_limit);
 #endif
--- a/arch/loongarch/include/asm/hugetlb.h
+++ b/arch/loongarch/include/asm/hugetlb.h
@ -24,6 +24,16 @@ static inline int prepare_hugepage_range(struct file *file,
 	return 0;
 }

+#define __HAVE_ARCH_HUGE_PTE_CLEAR
+static inline void huge_pte_clear(struct mm_struct *mm, unsigned long addr,
+				  pte_t *ptep, unsigned long sz)
+{
+	pte_t clear;
+
+	pte_val(clear) = (unsigned long)invalid_pte_table;
+	set_pte_at(mm, addr, ptep, clear);
+}
+
 #define __HAVE_ARCH_HUGE_PTEP_GET_AND_CLEAR
 static inline pte_t huge_ptep_get_and_clear(struct mm_struct *mm,
 					    unsigned long addr, pte_t *ptep)
--- a/arch/loongarch/include/asm/inst.h
+++ b/arch/loongarch/include/asm/inst.h
@ -683,7 +683,17 @@ DEF_EMIT_REG2I16_FORMAT(blt, blt_op)
 DEF_EMIT_REG2I16_FORMAT(bge, bge_op)
 DEF_EMIT_REG2I16_FORMAT(bltu, bltu_op)
 DEF_EMIT_REG2I16_FORMAT(bgeu, bgeu_op)
-DEF_EMIT_REG2I16_FORMAT(jirl, jirl_op)
+
+static inline void emit_jirl(union loongarch_instruction *insn,
+			     enum loongarch_gpr rd,
+			     enum loongarch_gpr rj,
+			     int offset)
+{
+	insn->reg2i16_format.opcode = jirl_op;
+	insn->reg2i16_format.immediate = offset;
+	insn->reg2i16_format.rd = rd;
+	insn->reg2i16_format.rj = rj;
+}

 #define DEF_EMIT_REG2BSTRD_FORMAT(NAME, OP)				\
 static inline void emit_##NAME(union loongarch_instruction *insn,	\
--- a/arch/loongarch/kernel/efi.c
+++ b/arch/loongarch/kernel/efi.c
@ -95,7 +95,7 @@ static void __init init_screen_info(void)
 	memset(si, 0, sizeof(*si));
 	early_memunmap(si, sizeof(*si));

-	memblock_reserve(screen_info.lfb_base, screen_info.lfb_size);
+	memblock_reserve(__screen_info_lfb_base(&screen_info), screen_info.lfb_size);
 }

 void __init efi_init(void)
--- a/arch/loongarch/kernel/inst.c
+++ b/arch/loongarch/kernel/inst.c
@ -332,7 +332,7 @@ u32 larch_insn_gen_jirl(enum loongarch_gpr rd, enum loongarch_gpr rj, int imm)
 		return INSN_BREAK;
 	}

-	emit_jirl(&insn, rj, rd, imm >> 2);
+	emit_jirl(&insn, rd, rj, imm >> 2);

 	return insn.word;
 }
--- a/arch/loongarch/kernel/smp.c
+++ b/arch/loongarch/kernel/smp.c
@ -82,7 +82,7 @@ void show_ipi_list(struct seq_file *p, int prec)
 	for (i = 0; i < NR_IPI; i++) {
 		seq_printf(p, "%*s%u:%s", prec - 1, "IPI", i, prec >= 4 ? " " : "");
 		for_each_online_cpu(cpu)
-			seq_printf(p, "%10u ", per_cpu(irq_stat, cpu).ipi_irqs[i]);
+			seq_put_decimal_ull_width(p, " ", per_cpu(irq_stat, cpu).ipi_irqs[i], 10);
 		seq_printf(p, " LoongArch  %d  %s\n", i + 1, ipi_types[i]);
 	}
 }
--- a/arch/loongarch/kvm/exit.c
+++ b/arch/loongarch/kvm/exit.c
@ -156,7 +156,7 @@ static int kvm_handle_csr(struct kvm_vcpu *vcpu, larch_inst inst)

 int kvm_emu_iocsr(larch_inst inst, struct kvm_run *run, struct kvm_vcpu *vcpu)
 {
-	int ret;
+	int idx, ret;
 	unsigned long *val;
 	u32 addr, rd, rj, opcode;

@ -167,7 +167,6 @@ int kvm_emu_iocsr(larch_inst inst, struct kvm_run *run, struct kvm_vcpu *vcpu)
 	rj = inst.reg2_format.rj;
 	opcode = inst.reg2_format.opcode;
 	addr = vcpu->arch.gprs[rj];
-	ret = EMULATE_DO_IOCSR;
 	run->iocsr_io.phys_addr = addr;
 	run->iocsr_io.is_write = 0;
 	val = &vcpu->arch.gprs[rd];
@ -207,20 +206,28 @@ int kvm_emu_iocsr(larch_inst inst, struct kvm_run *run, struct kvm_vcpu *vcpu)
 	}

 	if (run->iocsr_io.is_write) {
-		if (!kvm_io_bus_write(vcpu, KVM_IOCSR_BUS, addr, run->iocsr_io.len, val))
+		idx = srcu_read_lock(&vcpu->kvm->srcu);
+		ret = kvm_io_bus_write(vcpu, KVM_IOCSR_BUS, addr, run->iocsr_io.len, val);
+		srcu_read_unlock(&vcpu->kvm->srcu, idx);
+		if (ret == 0)
 			ret = EMULATE_DONE;
-		else
+		else {
+			ret = EMULATE_DO_IOCSR;
 			/* Save data and let user space to write it */
 			memcpy(run->iocsr_io.data, val, run->iocsr_io.len);
-
+		}
 		trace_kvm_iocsr(KVM_TRACE_IOCSR_WRITE, run->iocsr_io.len, addr, val);
 	} else {
-		if (!kvm_io_bus_read(vcpu, KVM_IOCSR_BUS, addr, run->iocsr_io.len, val))
+		idx = srcu_read_lock(&vcpu->kvm->srcu);
+		ret = kvm_io_bus_read(vcpu, KVM_IOCSR_BUS, addr, run->iocsr_io.len, val);
+		srcu_read_unlock(&vcpu->kvm->srcu, idx);
+		if (ret == 0)
 			ret = EMULATE_DONE;
-		else
+		else {
+			ret = EMULATE_DO_IOCSR;
 			/* Save register id for iocsr read completion */
 			vcpu->arch.io_gpr = rd;
-
+		}
 		trace_kvm_iocsr(KVM_TRACE_IOCSR_READ, run->iocsr_io.len, addr, NULL);
 	}

@ -359,7 +366,7 @@ static int kvm_handle_gspr(struct kvm_vcpu *vcpu)

 int kvm_emu_mmio_read(struct kvm_vcpu *vcpu, larch_inst inst)
 {
-	int ret;
+	int idx, ret;
 	unsigned int op8, opcode, rd;
 	struct kvm_run *run = vcpu->run;

@ -464,8 +471,10 @@ int kvm_emu_mmio_read(struct kvm_vcpu *vcpu, larch_inst inst)
 		 * it need not return to user space to handle the mmio
 		 * exception.
 		 */
+		idx = srcu_read_lock(&vcpu->kvm->srcu);
 		ret = kvm_io_bus_read(vcpu, KVM_MMIO_BUS, vcpu->arch.badv,
 				      run->mmio.len, &vcpu->arch.gprs[rd]);
+		srcu_read_unlock(&vcpu->kvm->srcu, idx);
 		if (!ret) {
 			update_pc(&vcpu->arch);
 			vcpu->mmio_needed = 0;
@ -531,7 +540,7 @@ int kvm_complete_mmio_read(struct kvm_vcpu *vcpu, struct kvm_run *run)

 int kvm_emu_mmio_write(struct kvm_vcpu *vcpu, larch_inst inst)
 {
-	int ret;
+	int idx, ret;
 	unsigned int rd, op8, opcode;
 	unsigned long curr_pc, rd_val = 0;
 	struct kvm_run *run = vcpu->run;
@ -631,7 +640,9 @@ int kvm_emu_mmio_write(struct kvm_vcpu *vcpu, larch_inst inst)
 		 * it need not return to user space to handle the mmio
 		 * exception.
 		 */
+		idx = srcu_read_lock(&vcpu->kvm->srcu);
 		ret = kvm_io_bus_write(vcpu, KVM_MMIO_BUS, vcpu->arch.badv, run->mmio.len, data);
+		srcu_read_unlock(&vcpu->kvm->srcu, idx);
 		if (!ret)
 			return EMULATE_DONE;

--- a/arch/loongarch/kvm/intc/ipi.c
+++ b/arch/loongarch/kvm/intc/ipi.c
@ -98,7 +98,7 @@ static void write_mailbox(struct kvm_vcpu *vcpu, int offset, uint64_t data, int

 static int send_ipi_data(struct kvm_vcpu *vcpu, gpa_t addr, uint64_t data)
 {
-	int i, ret;
+	int i, idx, ret;
 	uint32_t val = 0, mask = 0;

 	/*
@ -107,7 +107,9 @@ static int send_ipi_data(struct kvm_vcpu *vcpu, gpa_t addr, uint64_t data)
 	 */
 	if ((data >> 27) & 0xf) {
 		/* Read the old val */
+		idx = srcu_read_lock(&vcpu->kvm->srcu);
 		ret = kvm_io_bus_read(vcpu, KVM_IOCSR_BUS, addr, sizeof(val), &val);
+		srcu_read_unlock(&vcpu->kvm->srcu, idx);
 		if (unlikely(ret)) {
 			kvm_err("%s: : read date from addr %llx failed\n", __func__, addr);
 			return ret;
@ -121,7 +123,9 @@ static int send_ipi_data(struct kvm_vcpu *vcpu, gpa_t addr, uint64_t data)
 		val &= mask;
 	}
 	val |= ((uint32_t)(data >> 32) & ~mask);
+	idx = srcu_read_lock(&vcpu->kvm->srcu);
 	ret = kvm_io_bus_write(vcpu, KVM_IOCSR_BUS, addr, sizeof(val), &val);
+	srcu_read_unlock(&vcpu->kvm->srcu, idx);
 	if (unlikely(ret))
 		kvm_err("%s: : write date to addr %llx failed\n", __func__, addr);

--- a/arch/loongarch/kvm/vcpu.c
+++ b/arch/loongarch/kvm/vcpu.c
@ -240,7 +240,7 @@ static void kvm_late_check_requests(struct kvm_vcpu *vcpu)
 */
 static int kvm_enter_guest_check(struct kvm_vcpu *vcpu)
 {
-	int ret;
+	int idx, ret;

 	/*
 	 * Check conditions before entering the guest
@ -249,7 +249,9 @@ static int kvm_enter_guest_check(struct kvm_vcpu *vcpu)
 	if (ret < 0)
 		return ret;

+	idx = srcu_read_lock(&vcpu->kvm->srcu);
 	ret = kvm_check_requests(vcpu);
+	srcu_read_unlock(&vcpu->kvm->srcu, idx);

 	return ret;
 }
--- a/arch/loongarch/net/bpf_jit.c
+++ b/arch/loongarch/net/bpf_jit.c
@ -181,13 +181,13 @@ static void __build_epilogue(struct jit_ctx *ctx, bool is_tail_call)
 		/* Set return value */
 		emit_insn(ctx, addiw, LOONGARCH_GPR_A0, regmap[BPF_REG_0], 0);
 		/* Return to the caller */
-		emit_insn(ctx, jirl, LOONGARCH_GPR_RA, LOONGARCH_GPR_ZERO, 0);
+		emit_insn(ctx, jirl, LOONGARCH_GPR_ZERO, LOONGARCH_GPR_RA, 0);
 	} else {
 		/*
 		 * Call the next bpf prog and skip the first instruction
 		 * of TCC initialization.
 		 */
-		emit_insn(ctx, jirl, LOONGARCH_GPR_T3, LOONGARCH_GPR_ZERO, 1);
+		emit_insn(ctx, jirl, LOONGARCH_GPR_ZERO, LOONGARCH_GPR_T3, 1);
 	}
 }

@ -904,7 +904,7 @@ static int build_insn(const struct bpf_insn *insn, struct jit_ctx *ctx, bool ext
 			return ret;

 		move_addr(ctx, t1, func_addr);
-		emit_insn(ctx, jirl, t1, LOONGARCH_GPR_RA, 0);
+		emit_insn(ctx, jirl, LOONGARCH_GPR_RA, t1, 0);
 		move_reg(ctx, regmap[BPF_REG_0], LOONGARCH_GPR_A0);
 		break;

--- a/arch/x86/include/asm/pgtable_types.h
+++ b/arch/x86/include/asm/pgtable_types.h
@ -36,10 +36,12 @@
 #define _PAGE_BIT_DEVMAP	_PAGE_BIT_SOFTW4

 #ifdef CONFIG_X86_64
-#define _PAGE_BIT_SAVED_DIRTY	_PAGE_BIT_SOFTW5 /* Saved Dirty bit */
+#define _PAGE_BIT_SAVED_DIRTY	_PAGE_BIT_SOFTW5 /* Saved Dirty bit (leaf) */
+#define _PAGE_BIT_NOPTISHADOW	_PAGE_BIT_SOFTW5 /* No PTI shadow (root PGD) */
 #else
 /* Shared with _PAGE_BIT_UFFD_WP which is not supported on 32 bit */
-#define _PAGE_BIT_SAVED_DIRTY	_PAGE_BIT_SOFTW2 /* Saved Dirty bit */
+#define _PAGE_BIT_SAVED_DIRTY	_PAGE_BIT_SOFTW2 /* Saved Dirty bit (leaf) */
+#define _PAGE_BIT_NOPTISHADOW	_PAGE_BIT_SOFTW2 /* No PTI shadow (root PGD) */
 #endif

 /* If _PAGE_BIT_PRESENT is clear, we use these: */
@ -139,6 +141,8 @@

 #define _PAGE_PROTNONE	(_AT(pteval_t, 1) << _PAGE_BIT_PROTNONE)

+#define _PAGE_NOPTISHADOW (_AT(pteval_t, 1) << _PAGE_BIT_NOPTISHADOW)
+
 /*
 * Set of bits not changed in pte_modify.  The pte's
 * protection key is treated like _PAGE_RW, for
--- a/arch/x86/kernel/cpu/amd.c
+++ b/arch/x86/kernel/cpu/amd.c
@ -1065,7 +1065,7 @@ static void init_amd(struct cpuinfo_x86 *c)
 	 */
 	if (spectre_v2_in_eibrs_mode(spectre_v2_enabled) &&
 	    cpu_has(c, X86_FEATURE_AUTOIBRS))
-		WARN_ON_ONCE(msr_set_bit(MSR_EFER, _EFER_AUTOIBRS));
+		WARN_ON_ONCE(msr_set_bit(MSR_EFER, _EFER_AUTOIBRS) < 0);

 	/* AMD CPUs don't need fencing after x2APIC/TSC_DEADLINE MSR writes. */
 	clear_cpu_cap(c, X86_FEATURE_APIC_MSRS_FENCE);
--- a/arch/x86/kernel/cpu/cacheinfo.c
+++ b/arch/x86/kernel/cpu/cacheinfo.c
@ -178,8 +178,6 @@ struct _cpuid4_info_regs {
 	struct amd_northbridge *nb;
 };

-static unsigned short num_cache_leaves;
-
 /* AMD doesn't have CPUID4. Emulate it here to report the same
   information to the user.  This makes some assumptions about the machine:
   L2 not shared, no SMT etc. that is currently true on AMD CPUs.
@ -717,20 +715,23 @@ void cacheinfo_hygon_init_llc_id(struct cpuinfo_x86 *c)

 void init_amd_cacheinfo(struct cpuinfo_x86 *c)
 {
+	struct cpu_cacheinfo *ci = get_cpu_cacheinfo(c->cpu_index);

 	if (boot_cpu_has(X86_FEATURE_TOPOEXT)) {
-		num_cache_leaves = find_num_cache_leaves(c);
+		ci->num_leaves = find_num_cache_leaves(c);
 	} else if (c->extended_cpuid_level >= 0x80000006) {
 		if (cpuid_edx(0x80000006) & 0xf000)
-			num_cache_leaves = 4;
+			ci->num_leaves = 4;
 		else
-			num_cache_leaves = 3;
+			ci->num_leaves = 3;
 	}
 }

 void init_hygon_cacheinfo(struct cpuinfo_x86 *c)
 {
-	num_cache_leaves = find_num_cache_leaves(c);
+	struct cpu_cacheinfo *ci = get_cpu_cacheinfo(c->cpu_index);
+
+	ci->num_leaves = find_num_cache_leaves(c);
 }

 void init_intel_cacheinfo(struct cpuinfo_x86 *c)
@ -740,21 +741,21 @@ void init_intel_cacheinfo(struct cpuinfo_x86 *c)
 	unsigned int new_l1d = 0, new_l1i = 0; /* Cache sizes from cpuid(4) */
 	unsigned int new_l2 = 0, new_l3 = 0, i; /* Cache sizes from cpuid(4) */
 	unsigned int l2_id = 0, l3_id = 0, num_threads_sharing, index_msb;
+	struct cpu_cacheinfo *ci = get_cpu_cacheinfo(c->cpu_index);

 	if (c->cpuid_level > 3) {
-		static int is_initialized;
-
-		if (is_initialized == 0) {
-			/* Init num_cache_leaves from boot CPU */
-			num_cache_leaves = find_num_cache_leaves(c);
-			is_initialized++;
-		}
+		/*
+		 * There should be at least one leaf. A non-zero value means
+		 * that the number of leaves has been initialized.
+		 */
+		if (!ci->num_leaves)
+			ci->num_leaves = find_num_cache_leaves(c);

 		/*
 		 * Whenever possible use cpuid(4), deterministic cache
 		 * parameters cpuid leaf to find the cache details
 		 */
-		for (i = 0; i < num_cache_leaves; i++) {
+		for (i = 0; i < ci->num_leaves; i++) {
 			struct _cpuid4_info_regs this_leaf = {};
 			int retval;

@ -790,14 +791,14 @@ void init_intel_cacheinfo(struct cpuinfo_x86 *c)
 	 * Don't use cpuid2 if cpuid4 is supported. For P4, we use cpuid2 for
 	 * trace cache
 	 */
-	if ((num_cache_leaves == 0 || c->x86 == 15) && c->cpuid_level > 1) {
+	if ((!ci->num_leaves || c->x86 == 15) && c->cpuid_level > 1) {
 		/* supports eax=2  call */
 		int j, n;
 		unsigned int regs[4];
 		unsigned char *dp = (unsigned char *)regs;
 		int only_trace = 0;

-		if (num_cache_leaves != 0 && c->x86 == 15)
+		if (ci->num_leaves && c->x86 == 15)
 			only_trace = 1;

 		/* Number of times to iterate */
@ -991,14 +992,12 @@ static void ci_leaf_init(struct cacheinfo *this_leaf,

 int init_cache_level(unsigned int cpu)
 {
-	struct cpu_cacheinfo *this_cpu_ci = get_cpu_cacheinfo(cpu);
+	struct cpu_cacheinfo *ci = get_cpu_cacheinfo(cpu);

-	if (!num_cache_leaves)
+	/* There should be at least one leaf. */
+	if (!ci->num_leaves)
 		return -ENOENT;
-	if (!this_cpu_ci)
-		return -EINVAL;
-	this_cpu_ci->num_levels = 3;
-	this_cpu_ci->num_leaves = num_cache_leaves;
+
 	return 0;
 }

--- a/arch/x86/kernel/cpu/intel.c
+++ b/arch/x86/kernel/cpu/intel.c
@ -555,7 +555,9 @@ static void init_intel(struct cpuinfo_x86 *c)
 	     c->x86_vfm == INTEL_WESTMERE_EX))
 		set_cpu_bug(c, X86_BUG_CLFLUSH_MONITOR);

-	if (boot_cpu_has(X86_FEATURE_MWAIT) && c->x86_vfm == INTEL_ATOM_GOLDMONT)
+	if (boot_cpu_has(X86_FEATURE_MWAIT) &&
+	    (c->x86_vfm == INTEL_ATOM_GOLDMONT ||
+	     c->x86_vfm == INTEL_LUNARLAKE_M))
 		set_cpu_bug(c, X86_BUG_MONITOR);

 #ifdef CONFIG_X86_64
--- a/arch/x86/kernel/cpu/topology.c
+++ b/arch/x86/kernel/cpu/topology.c
@ -428,8 +428,8 @@ void __init topology_apply_cmdline_limits_early(void)
 {
 	unsigned int possible = nr_cpu_ids;

-	/* 'maxcpus=0' 'nosmp' 'nolapic' 'disableapic' 'noapic' */
-	if (!setup_max_cpus || ioapic_is_disabled || apic_is_disabled)
+	/* 'maxcpus=0' 'nosmp' 'nolapic' 'disableapic' */
+	if (!setup_max_cpus || apic_is_disabled)
 		possible = 1;

 	/* 'possible_cpus=N' */
@ -443,7 +443,7 @@ void __init topology_apply_cmdline_limits_early(void)

 static __init bool restrict_to_up(void)
 {
-	if (!smp_found_config || ioapic_is_disabled)
+	if (!smp_found_config)
 		return true;
 	/*
 	 * XEN PV is special as it does not advertise the local APIC
--- a/arch/x86/kernel/fpu/signal.c
+++ b/arch/x86/kernel/fpu/signal.c
@ -63,16 +63,6 @@ static inline bool check_xstate_in_sigframe(struct fxregs_state __user *fxbuf,
 	return true;
 }

-/*
- * Update the value of PKRU register that was already pushed onto the signal frame.
- */
-static inline int update_pkru_in_sigframe(struct xregs_state __user *buf, u32 pkru)
-{
-	if (unlikely(!cpu_feature_enabled(X86_FEATURE_OSPKE)))
-		return 0;
-	return __put_user(pkru, (unsigned int __user *)get_xsave_addr_user(buf, XFEATURE_PKRU));
-}
-
 /*
 * Signal frame handlers.
 */
@ -168,14 +158,8 @@ static inline bool save_xstate_epilog(void __user *buf, int ia32_frame,

 static inline int copy_fpregs_to_sigframe(struct xregs_state __user *buf, u32 pkru)
 {
-	int err = 0;
-
-	if (use_xsave()) {
-		err = xsave_to_user_sigframe(buf);
-		if (!err)
-			err = update_pkru_in_sigframe(buf, pkru);
-		return err;
-	}
+	if (use_xsave())
+		return xsave_to_user_sigframe(buf, pkru);

 	if (use_fxsr())
 		return fxsave_to_user_sigframe((struct fxregs_state __user *) buf);
--- a/arch/x86/kernel/fpu/xstate.h
+++ b/arch/x86/kernel/fpu/xstate.h
@ -69,6 +69,28 @@ static inline u64 xfeatures_mask_independent(void)
 	return fpu_kernel_cfg.independent_features;
 }

+/*
+ * Update the value of PKRU register that was already pushed onto the signal frame.
+ */
+static inline int update_pkru_in_sigframe(struct xregs_state __user *buf, u64 mask, u32 pkru)
+{
+	u64 xstate_bv;
+	int err;
+
+	if (unlikely(!cpu_feature_enabled(X86_FEATURE_OSPKE)))
+		return 0;
+
+	/* Mark PKRU as in-use so that it is restored correctly. */
+	xstate_bv = (mask & xfeatures_in_use()) | XFEATURE_MASK_PKRU;
+
+	err =  __put_user(xstate_bv, &buf->header.xfeatures);
+	if (err)
+		return err;
+
+	/* Update PKRU value in the userspace xsave buffer. */
+	return __put_user(pkru, (unsigned int __user *)get_xsave_addr_user(buf, XFEATURE_PKRU));
+}
+
 /* XSAVE/XRSTOR wrapper functions */

 #ifdef CONFIG_X86_64
@ -256,7 +278,7 @@ static inline u64 xfeatures_need_sigframe_write(void)
 * The caller has to zero buf::header before calling this because XSAVE*
 * does not touch the reserved fields in the header.
 */
-static inline int xsave_to_user_sigframe(struct xregs_state __user *buf)
+static inline int xsave_to_user_sigframe(struct xregs_state __user *buf, u32 pkru)
 {
 	/*
 	 * Include the features which are not xsaved/rstored by the kernel
@ -281,6 +303,9 @@ static inline int xsave_to_user_sigframe(struct xregs_state __user *buf)
 	XSTATE_OP(XSAVE, buf, lmask, hmask, err);
 	clac();

+	if (!err)
+		err = update_pkru_in_sigframe(buf, mask, pkru);
+
 	return err;
 }

--- a/arch/x86/kernel/relocate_kernel_64.S
+++ b/arch/x86/kernel/relocate_kernel_64.S
@ -242,6 +242,13 @@ SYM_CODE_START_LOCAL_NOALIGN(virtual_mapped)
 	movq	CR0(%r8), %r8
 	movq	%rax, %cr3
 	movq	%r8, %cr0
+
+#ifdef CONFIG_KEXEC_JUMP
+	/* Saved in save_processor_state. */
+	movq    $saved_context, %rax
+	lgdt    saved_context_gdt_desc(%rax)
+#endif
+
 	movq	%rbp, %rax

 	popf
--- a/arch/x86/mm/ident_map.c
+++ b/arch/x86/mm/ident_map.c
@ -174,7 +174,7 @@ static int ident_p4d_init(struct x86_mapping_info *info, p4d_t *p4d_page,
 		if (result)
 			return result;

-		set_p4d(p4d, __p4d(__pa(pud) | info->kernpg_flag));
+		set_p4d(p4d, __p4d(__pa(pud) | info->kernpg_flag | _PAGE_NOPTISHADOW));
 	}

 	return 0;
@ -218,14 +218,14 @@ int kernel_ident_mapping_init(struct x86_mapping_info *info, pgd_t *pgd_page,
 		if (result)
 			return result;
 		if (pgtable_l5_enabled()) {
-			set_pgd(pgd, __pgd(__pa(p4d) | info->kernpg_flag));
+			set_pgd(pgd, __pgd(__pa(p4d) | info->kernpg_flag | _PAGE_NOPTISHADOW));
 		} else {
 			/*
 			 * With p4d folded, pgd is equal to p4d.
 			 * The pgd entry has to point to the pud page table in this case.
 			 */
 			pud_t *pud = pud_offset(p4d, 0);
-			set_pgd(pgd, __pgd(__pa(pud) | info->kernpg_flag));
+			set_pgd(pgd, __pgd(__pa(pud) | info->kernpg_flag | _PAGE_NOPTISHADOW));
 		}
 	}

--- a/arch/x86/mm/pti.c
+++ b/arch/x86/mm/pti.c
@ -132,7 +132,7 @@ pgd_t __pti_set_user_pgtbl(pgd_t *pgdp, pgd_t pgd)
 	 * Top-level entries added to init_mm's usermode pgd after boot
 	 * will not be automatically propagated to other mms.
 	 */
-	if (!pgdp_maps_userspace(pgdp))
+	if (!pgdp_maps_userspace(pgdp) || (pgd.pgd & _PAGE_NOPTISHADOW))
 		return pgd;

 	/*
--- a/block/blk-mq.c
+++ b/block/blk-mq.c
@ -43,6 +43,7 @@

 static DEFINE_PER_CPU(struct llist_head, blk_cpu_done);
 static DEFINE_PER_CPU(call_single_data_t, blk_cpu_csd);
+static DEFINE_MUTEX(blk_mq_cpuhp_lock);

 static void blk_mq_insert_request(struct request *rq, blk_insert_t flags);
 static void blk_mq_request_bypass_insert(struct request *rq,
@ -3739,13 +3740,91 @@ static int blk_mq_hctx_notify_dead(unsigned int cpu, struct hlist_node *node)
 	return 0;
 }

-static void blk_mq_remove_cpuhp(struct blk_mq_hw_ctx *hctx)
+static void __blk_mq_remove_cpuhp(struct blk_mq_hw_ctx *hctx)
 {
-	if (!(hctx->flags & BLK_MQ_F_STACKING))
+	lockdep_assert_held(&blk_mq_cpuhp_lock);
+
+	if (!(hctx->flags & BLK_MQ_F_STACKING) &&
+	    !hlist_unhashed(&hctx->cpuhp_online)) {
 		cpuhp_state_remove_instance_nocalls(CPUHP_AP_BLK_MQ_ONLINE,
 						    &hctx->cpuhp_online);
-	cpuhp_state_remove_instance_nocalls(CPUHP_BLK_MQ_DEAD,
-					    &hctx->cpuhp_dead);
+		INIT_HLIST_NODE(&hctx->cpuhp_online);
+	}
+
+	if (!hlist_unhashed(&hctx->cpuhp_dead)) {
+		cpuhp_state_remove_instance_nocalls(CPUHP_BLK_MQ_DEAD,
+						    &hctx->cpuhp_dead);
+		INIT_HLIST_NODE(&hctx->cpuhp_dead);
+	}
+}
+
+static void blk_mq_remove_cpuhp(struct blk_mq_hw_ctx *hctx)
+{
+	mutex_lock(&blk_mq_cpuhp_lock);
+	__blk_mq_remove_cpuhp(hctx);
+	mutex_unlock(&blk_mq_cpuhp_lock);
+}
+
+static void __blk_mq_add_cpuhp(struct blk_mq_hw_ctx *hctx)
+{
+	lockdep_assert_held(&blk_mq_cpuhp_lock);
+
+	if (!(hctx->flags & BLK_MQ_F_STACKING) &&
+	    hlist_unhashed(&hctx->cpuhp_online))
+		cpuhp_state_add_instance_nocalls(CPUHP_AP_BLK_MQ_ONLINE,
+				&hctx->cpuhp_online);
+
+	if (hlist_unhashed(&hctx->cpuhp_dead))
+		cpuhp_state_add_instance_nocalls(CPUHP_BLK_MQ_DEAD,
+				&hctx->cpuhp_dead);
+}
+
+static void __blk_mq_remove_cpuhp_list(struct list_head *head)
+{
+	struct blk_mq_hw_ctx *hctx;
+
+	lockdep_assert_held(&blk_mq_cpuhp_lock);
+
+	list_for_each_entry(hctx, head, hctx_list)
+		__blk_mq_remove_cpuhp(hctx);
+}
+
+/*
+ * Unregister cpuhp callbacks from exited hw queues
+ *
+ * Safe to call if this `request_queue` is live
+ */
+static void blk_mq_remove_hw_queues_cpuhp(struct request_queue *q)
+{
+	LIST_HEAD(hctx_list);
+
+	spin_lock(&q->unused_hctx_lock);
+	list_splice_init(&q->unused_hctx_list, &hctx_list);
+	spin_unlock(&q->unused_hctx_lock);
+
+	mutex_lock(&blk_mq_cpuhp_lock);
+	__blk_mq_remove_cpuhp_list(&hctx_list);
+	mutex_unlock(&blk_mq_cpuhp_lock);
+
+	spin_lock(&q->unused_hctx_lock);
+	list_splice(&hctx_list, &q->unused_hctx_list);
+	spin_unlock(&q->unused_hctx_lock);
+}
+
+/*
+ * Register cpuhp callbacks from all hw queues
+ *
+ * Safe to call if this `request_queue` is live
+ */
+static void blk_mq_add_hw_queues_cpuhp(struct request_queue *q)
+{
+	struct blk_mq_hw_ctx *hctx;
+	unsigned long i;
+
+	mutex_lock(&blk_mq_cpuhp_lock);
+	queue_for_each_hw_ctx(q, hctx, i)
+		__blk_mq_add_cpuhp(hctx);
+	mutex_unlock(&blk_mq_cpuhp_lock);
 }

 /*
@ -3796,8 +3875,6 @@ static void blk_mq_exit_hctx(struct request_queue *q,
 	if (set->ops->exit_hctx)
 		set->ops->exit_hctx(hctx, hctx_idx);

-	blk_mq_remove_cpuhp(hctx);
-
 	xa_erase(&q->hctx_table, hctx_idx);

 	spin_lock(&q->unused_hctx_lock);
@ -3814,6 +3891,7 @@ static void blk_mq_exit_hw_queues(struct request_queue *q,
 	queue_for_each_hw_ctx(q, hctx, i) {
 		if (i == nr_queue)
 			break;
+		blk_mq_remove_cpuhp(hctx);
 		blk_mq_exit_hctx(q, set, hctx, i);
 	}
 }
@ -3824,16 +3902,11 @@ static int blk_mq_init_hctx(struct request_queue *q,
 {
 	hctx->queue_num = hctx_idx;

-	if (!(hctx->flags & BLK_MQ_F_STACKING))
-		cpuhp_state_add_instance_nocalls(CPUHP_AP_BLK_MQ_ONLINE,
-				&hctx->cpuhp_online);
-	cpuhp_state_add_instance_nocalls(CPUHP_BLK_MQ_DEAD, &hctx->cpuhp_dead);
-
 	hctx->tags = set->tags[hctx_idx];

 	if (set->ops->init_hctx &&
 	    set->ops->init_hctx(hctx, set->driver_data, hctx_idx))
-		goto unregister_cpu_notifier;
+		goto fail;

 	if (blk_mq_init_request(set, hctx->fq->flush_rq, hctx_idx,
 				hctx->numa_node))
@ -3850,8 +3923,7 @@ static int blk_mq_init_hctx(struct request_queue *q,
 exit_hctx:
 	if (set->ops->exit_hctx)
 		set->ops->exit_hctx(hctx, hctx_idx);
- unregister_cpu_notifier:
-	blk_mq_remove_cpuhp(hctx);
+ fail:
 	return -1;
 }

@ -3877,6 +3949,8 @@ blk_mq_alloc_hctx(struct request_queue *q, struct blk_mq_tag_set *set,
 	INIT_DELAYED_WORK(&hctx->run_work, blk_mq_run_work_fn);
 	spin_lock_init(&hctx->lock);
 	INIT_LIST_HEAD(&hctx->dispatch);
+	INIT_HLIST_NODE(&hctx->cpuhp_dead);
+	INIT_HLIST_NODE(&hctx->cpuhp_online);
 	hctx->queue = q;
 	hctx->flags = set->flags & ~BLK_MQ_F_TAG_QUEUE_SHARED;

@ -4415,6 +4489,12 @@ static void blk_mq_realloc_hw_ctxs(struct blk_mq_tag_set *set,
 	xa_for_each_start(&q->hctx_table, j, hctx, j)
 		blk_mq_exit_hctx(q, set, hctx, j);
 	mutex_unlock(&q->sysfs_lock);
+
+	/* unregister cpuhp callbacks for exited hctxs */
+	blk_mq_remove_hw_queues_cpuhp(q);
+
+	/* register cpuhp for new initialized hctxs */
+	blk_mq_add_hw_queues_cpuhp(q);
 }

 int blk_mq_init_allocated_queue(struct blk_mq_tag_set *set,
--- a/drivers/acpi/arm64/iort.c
+++ b/drivers/acpi/arm64/iort.c
@ -1716,6 +1716,8 @@ static struct acpi_platform_list pmcg_plat_info[] __initdata = {
 	/* HiSilicon Hip09 Platform */
 	{"HISI  ", "HIP09   ", 0, ACPI_SIG_IORT, greater_than_or_equal,
 	 "Erratum #162001900", IORT_SMMU_V3_PMCG_HISI_HIP09},
+	{"HISI  ", "HIP09A  ", 0, ACPI_SIG_IORT, greater_than_or_equal,
+	 "Erratum #162001900", IORT_SMMU_V3_PMCG_HISI_HIP09},
 	/* HiSilicon Hip10/11 Platform uses the same SMMU IP with Hip09 */
 	{"HISI  ", "HIP10   ", 0, ACPI_SIG_IORT, greater_than_or_equal,
 	 "Erratum #162001900", IORT_SMMU_V3_PMCG_HISI_HIP09},
--- a/drivers/base/arch_numa.c
+++ b/drivers/base/arch_numa.c
@ -208,6 +208,10 @@ static int __init numa_register_nodes(void)
 {
 	int nid;

+	/* Check the validity of the memblock/node mapping */
+	if (!memblock_validate_numa_coverage(0))
+		return -EINVAL;
+
 	/* Finally register nodes. */
 	for_each_node_mask(nid, numa_nodes_parsed) {
 		unsigned long start_pfn, end_pfn;
--- a/drivers/base/cacheinfo.c
+++ b/drivers/base/cacheinfo.c
@ -58,7 +58,7 @@ bool last_level_cache_is_valid(unsigned int cpu)
 {
 	struct cacheinfo *llc;

-	if (!cache_leaves(cpu))
+	if (!cache_leaves(cpu) || !per_cpu_cacheinfo(cpu))
 		return false;

 	llc = per_cpu_cacheinfo_idx(cpu, cache_leaves(cpu) - 1);
@ -458,11 +458,9 @@ int __weak populate_cache_leaves(unsigned int cpu)
 	return -ENOENT;
 }

-static inline
-int allocate_cache_info(int cpu)
+static inline int allocate_cache_info(int cpu)
 {
-	per_cpu_cacheinfo(cpu) = kcalloc(cache_leaves(cpu),
-					 sizeof(struct cacheinfo), GFP_ATOMIC);
+	per_cpu_cacheinfo(cpu) = kcalloc(cache_leaves(cpu), sizeof(struct cacheinfo), GFP_ATOMIC);
 	if (!per_cpu_cacheinfo(cpu)) {
 		cache_leaves(cpu) = 0;
 		return -ENOMEM;
@ -534,7 +532,11 @@ static inline int init_level_allocate_ci(unsigned int cpu)
 	 */
 	ci_cacheinfo(cpu)->early_ci_levels = false;

-	if (cache_leaves(cpu) <= early_leaves)
+	/*
+	 * Some architectures (e.g., x86) do not use early initialization.
+	 * Allocate memory now in such case.
+	 */
+	if (cache_leaves(cpu) <= early_leaves && per_cpu_cacheinfo(cpu))
 		return 0;

 	kfree(per_cpu_cacheinfo(cpu));
--- a/drivers/base/regmap/regmap.c
+++ b/drivers/base/regmap/regmap.c
@ -598,6 +598,17 @@ int regmap_attach_dev(struct device *dev, struct regmap *map,
 }
 EXPORT_SYMBOL_GPL(regmap_attach_dev);

+static int dev_get_regmap_match(struct device *dev, void *res, void *data);
+
+static int regmap_detach_dev(struct device *dev, struct regmap *map)
+{
+	if (!dev)
+		return 0;
+
+	return devres_release(dev, dev_get_regmap_release,
+			      dev_get_regmap_match, (void *)map->name);
+}
+
 static enum regmap_endian regmap_get_reg_endian(const struct regmap_bus *bus,
 					const struct regmap_config *config)
 {
@ -1052,13 +1063,13 @@ struct regmap *__regmap_init(struct device *dev,

 		/* Sanity check */
 		if (range_cfg->range_max < range_cfg->range_min) {
-			dev_err(map->dev, "Invalid range %d: %d < %d\n", i,
+			dev_err(map->dev, "Invalid range %d: %u < %u\n", i,
 				range_cfg->range_max, range_cfg->range_min);
 			goto err_range;
 		}

 		if (range_cfg->range_max > map->max_register) {
-			dev_err(map->dev, "Invalid range %d: %d > %d\n", i,
+			dev_err(map->dev, "Invalid range %d: %u > %u\n", i,
 				range_cfg->range_max, map->max_register);
 			goto err_range;
 		}
@ -1445,6 +1456,7 @@ void regmap_exit(struct regmap *map)
 {
 	struct regmap_async *async;

+	regmap_detach_dev(map->dev, map);
 	regcache_exit(map);

 	regmap_debugfs_exit(map);
--- a/drivers/block/rnull.rs
+++ b/drivers/block/rnull.rs
@ -28,6 +28,7 @@
    type: NullBlkModule,
    name: "rnull_mod",
    author: "Andreas Hindborg",
+    description: "Rust implementation of the C null block driver",
    license: "GPL v2",
 }

--- a/drivers/block/virtio_blk.c
+++ b/drivers/block/virtio_blk.c
@ -1586,9 +1586,12 @@ static void virtblk_remove(struct virtio_device *vdev)
 static int virtblk_freeze(struct virtio_device *vdev)
 {
 	struct virtio_blk *vblk = vdev->priv;
+	struct request_queue *q = vblk->disk->queue;

 	/* Ensure no requests in virtqueues before deleting vqs. */
-	blk_mq_freeze_queue(vblk->disk->queue);
+	blk_mq_freeze_queue(q);
+	blk_mq_quiesce_queue_nowait(q);
+	blk_mq_unfreeze_queue(q);

 	/* Ensure we don't receive any more interrupts */
 	virtio_reset_device(vdev);
@ -1612,8 +1615,8 @@ static int virtblk_restore(struct virtio_device *vdev)
 		return ret;

 	virtio_device_ready(vdev);
+	blk_mq_unquiesce_queue(vblk->disk->queue);

-	blk_mq_unfreeze_queue(vblk->disk->queue);
 	return 0;
 }
 #endif
--- a/drivers/dma-buf/dma-fence-array.c
+++ b/drivers/dma-buf/dma-fence-array.c
@ -103,10 +103,36 @@ static bool dma_fence_array_enable_signaling(struct dma_fence *fence)
 static bool dma_fence_array_signaled(struct dma_fence *fence)
 {
 	struct dma_fence_array *array = to_dma_fence_array(fence);
+	int num_pending;
+	unsigned int i;

-	if (atomic_read(&array->num_pending) > 0)
+	/*
+	 * We need to read num_pending before checking the enable_signal bit
+	 * to avoid racing with the enable_signaling() implementation, which
+	 * might decrement the counter, and cause a partial check.
+	 * atomic_read_acquire() pairs with atomic_dec_and_test() in
+	 * dma_fence_array_enable_signaling()
+	 *
+	 * The !--num_pending check is here to account for the any_signaled case
+	 * if we race with enable_signaling(), that means the !num_pending check
+	 * in the is_signalling_enabled branch might be outdated (num_pending
+	 * might have been decremented), but that's fine. The user will get the
+	 * right value when testing again later.
+	 */
+	num_pending = atomic_read_acquire(&array->num_pending);
+	if (test_bit(DMA_FENCE_FLAG_ENABLE_SIGNAL_BIT, &array->base.flags)) {
+		if (num_pending <= 0)
+			goto signal;
 		return false;
+	}

+	for (i = 0; i < array->num_fences; ++i) {
+		if (dma_fence_is_signaled(array->fences[i]) && !--num_pending)
+			goto signal;
+	}
+	return false;
+
+signal:
 	dma_fence_array_clear_pending_error(array);
 	return true;
 }
--- a/drivers/dma-buf/dma-fence-unwrap.c
+++ b/drivers/dma-buf/dma-fence-unwrap.c
@ -12,6 +12,7 @@
 #include <linux/dma-fence-chain.h>
 #include <linux/dma-fence-unwrap.h>
 #include <linux/slab.h>
+#include <linux/sort.h>

 /* Internal helper to start new array iteration, don't use directly */
 static struct dma_fence *
@ -59,6 +60,25 @@ struct dma_fence *dma_fence_unwrap_next(struct dma_fence_unwrap *cursor)
 }
 EXPORT_SYMBOL_GPL(dma_fence_unwrap_next);

+
+static int fence_cmp(const void *_a, const void *_b)
+{
+	struct dma_fence *a = *(struct dma_fence **)_a;
+	struct dma_fence *b = *(struct dma_fence **)_b;
+
+	if (a->context < b->context)
+		return -1;
+	else if (a->context > b->context)
+		return 1;
+
+	if (dma_fence_is_later(b, a))
+		return 1;
+	else if (dma_fence_is_later(a, b))
+		return -1;
+
+	return 0;
+}
+
 /* Implementation for the dma_fence_merge() marco, don't use directly */
 struct dma_fence *__dma_fence_unwrap_merge(unsigned int num_fences,
 					   struct dma_fence **fences,
@ -67,8 +87,7 @@ struct dma_fence *__dma_fence_unwrap_merge(unsigned int num_fences,
 	struct dma_fence_array *result;
 	struct dma_fence *tmp, **array;
 	ktime_t timestamp;
-	unsigned int i;
-	size_t count;
+	int i, j, count;

 	count = 0;
 	timestamp = ns_to_ktime(0);
@ -96,78 +115,55 @@ struct dma_fence *__dma_fence_unwrap_merge(unsigned int num_fences,
 	if (!array)
 		return NULL;

-	/*
-	 * This trashes the input fence array and uses it as position for the
-	 * following merge loop. This works because the dma_fence_merge()
-	 * wrapper macro is creating this temporary array on the stack together
-	 * with the iterators.
-	 */
-	for (i = 0; i < num_fences; ++i)
-		fences[i] = dma_fence_unwrap_first(fences[i], &iter[i]);
-
 	count = 0;
-	do {
-		unsigned int sel;
-
-restart:
-		tmp = NULL;
-		for (i = 0; i < num_fences; ++i) {
-			struct dma_fence *next;
-
-			while (fences[i] && dma_fence_is_signaled(fences[i]))
-				fences[i] = dma_fence_unwrap_next(&iter[i]);
-
-			next = fences[i];
-			if (!next)
-				continue;
-
-			/*
-			 * We can't guarantee that inpute fences are ordered by
-			 * context, but it is still quite likely when this
-			 * function is used multiple times. So attempt to order
-			 * the fences by context as we pass over them and merge
-			 * fences with the same context.
-			 */
-			if (!tmp || tmp->context > next->context) {
-				tmp = next;
-				sel = i;
-
-			} else if (tmp->context < next->context) {
-				continue;
-
-			} else if (dma_fence_is_later(tmp, next)) {
-				fences[i] = dma_fence_unwrap_next(&iter[i]);
-				goto restart;
+	for (i = 0; i < num_fences; ++i) {
+		dma_fence_unwrap_for_each(tmp, &iter[i], fences[i]) {
+			if (!dma_fence_is_signaled(tmp)) {
+				array[count++] = dma_fence_get(tmp);
 			} else {
-				fences[sel] = dma_fence_unwrap_next(&iter[sel]);
-				goto restart;
+				ktime_t t = dma_fence_timestamp(tmp);
+
+				if (ktime_after(t, timestamp))
+					timestamp = t;
 			}
 		}
+	}

-		if (tmp) {
-			array[count++] = dma_fence_get(tmp);
-			fences[sel] = dma_fence_unwrap_next(&iter[sel]);
+	if (count == 0 || count == 1)
+		goto return_fastpath;
+
+	sort(array, count, sizeof(*array), fence_cmp, NULL);
+
+	/*
+	 * Only keep the most recent fence for each context.
+	 */
+	j = 0;
+	for (i = 1; i < count; i++) {
+		if (array[i]->context == array[j]->context)
+			dma_fence_put(array[i]);
+		else
+			array[++j] = array[i];
+	}
+	count = ++j;
+
+	if (count > 1) {
+		result = dma_fence_array_create(count, array,
+						dma_fence_context_alloc(1),
+						1, false);
+		if (!result) {
+			for (i = 0; i < count; i++)
+				dma_fence_put(array[i]);
+			tmp = NULL;
+			goto return_tmp;
 		}
-	} while (tmp);
-
-	if (count == 0) {
-		tmp = dma_fence_allocate_private_stub(ktime_get());
-		goto return_tmp;
+		return &result->base;
 	}

-	if (count == 1) {
+return_fastpath:
+	if (count == 0)
+		tmp = dma_fence_allocate_private_stub(timestamp);
+	else
 		tmp = array[0];
-		goto return_tmp;
-	}
-
-	result = dma_fence_array_create(count, array,
-					dma_fence_context_alloc(1),
-					1, false);
-	if (!result) {
-		tmp = NULL;
-		goto return_tmp;
-	}
-	return &result->base;

 return_tmp:
 	kfree(array);
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
@ -145,7 +145,7 @@ const char *amdgpu_asic_name[] = {
 	"LAST",
 };

-#define AMDGPU_IP_BLK_MASK_ALL GENMASK(AMDGPU_MAX_IP_NUM - 1, 0)
+#define AMDGPU_IP_BLK_MASK_ALL GENMASK(AMDGPU_MAX_IP_NUM, 0)
 /*
 * Default init level where all blocks are expected to be initialized. This is
 * the level of initialization expected by default and also after a full reset
@ -3670,9 +3670,11 @@ static int amdgpu_device_ip_reinit_early_sriov(struct amdgpu_device *adev)
 				continue;

 			r = block->version->funcs->hw_init(&adev->ip_blocks[i]);
-			DRM_INFO("RE-INIT-early: %s %s\n", block->version->funcs->name, r?"failed":"succeeded");
-			if (r)
+			if (r) {
+				dev_err(adev->dev, "RE-INIT-early: %s failed\n",
+					 block->version->funcs->name);
 				return r;
+			}
 			block->status.hw = true;
 		}
 	}
@ -3682,7 +3684,8 @@ static int amdgpu_device_ip_reinit_early_sriov(struct amdgpu_device *adev)

 static int amdgpu_device_ip_reinit_late_sriov(struct amdgpu_device *adev)
 {
-	int i, r;
+	struct amdgpu_ip_block *block;
+	int i, r = 0;

 	static enum amd_ip_block_type ip_order[] = {
 		AMD_IP_BLOCK_TYPE_SMC,
@ -3697,34 +3700,28 @@ static int amdgpu_device_ip_reinit_late_sriov(struct amdgpu_device *adev)
 	};

 	for (i = 0; i < ARRAY_SIZE(ip_order); i++) {
-		int j;
-		struct amdgpu_ip_block *block;
+		block = amdgpu_device_ip_get_ip_block(adev, ip_order[i]);

-		for (j = 0; j < adev->num_ip_blocks; j++) {
-			block = &adev->ip_blocks[j];
-
-			if (block->version->type != ip_order[i] ||
-				!block->status.valid ||
-				block->status.hw)
-				continue;
+		if (!block)
+			continue;

+		if (block->status.valid && !block->status.hw) {
 			if (block->version->type == AMD_IP_BLOCK_TYPE_SMC) {
-				r = amdgpu_ip_block_resume(&adev->ip_blocks[i]);
-				if (r)
-					return r;
+				r = amdgpu_ip_block_resume(block);
 			} else {
-				r = block->version->funcs->hw_init(&adev->ip_blocks[i]);
-				if (r) {
-					DRM_ERROR("hw_init of IP block <%s> failed %d\n",
-						  adev->ip_blocks[i].version->funcs->name, r);
-					return r;
-				}
-				block->status.hw = true;
+				r = block->version->funcs->hw_init(block);
 			}
+
+			if (r) {
+				dev_err(adev->dev, "RE-INIT-late: %s failed\n",
+					 block->version->funcs->name);
+				break;
+			}
+			block->status.hw = true;
 		}
 	}

-	return 0;
+	return r;
 }

 /**
@ -3765,7 +3762,7 @@ static int amdgpu_device_ip_resume_phase1(struct amdgpu_device *adev)
 *
 * @adev: amdgpu_device pointer
 *
- * First resume function for hardware IPs.  The list of all the hardware
+ * Second resume function for hardware IPs.  The list of all the hardware
 * IPs that make up the asic is walked and the resume callbacks are run for
 * all blocks except COMMON, GMC, and IH.  resume puts the hardware into a
 * functional state after a suspend and updates the software state as
@ -3783,6 +3780,7 @@ static int amdgpu_device_ip_resume_phase2(struct amdgpu_device *adev)
 		if (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_COMMON ||
 		    adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_GMC ||
 		    adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_IH ||
+		    adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_DCE ||
 		    adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_PSP)
 			continue;
 		r = amdgpu_ip_block_resume(&adev->ip_blocks[i]);
@ -3793,6 +3791,36 @@ static int amdgpu_device_ip_resume_phase2(struct amdgpu_device *adev)
 	return 0;
 }

+/**
+ * amdgpu_device_ip_resume_phase3 - run resume for hardware IPs
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Third resume function for hardware IPs.  The list of all the hardware
+ * IPs that make up the asic is walked and the resume callbacks are run for
+ * all DCE.  resume puts the hardware into a functional state after a suspend
+ * and updates the software state as necessary.  This function is also used
+ * for restoring the GPU after a GPU reset.
+ *
+ * Returns 0 on success, negative error code on failure.
+ */
+static int amdgpu_device_ip_resume_phase3(struct amdgpu_device *adev)
+{
+	int i, r;
+
+	for (i = 0; i < adev->num_ip_blocks; i++) {
+		if (!adev->ip_blocks[i].status.valid || adev->ip_blocks[i].status.hw)
+			continue;
+		if (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_DCE) {
+			r = amdgpu_ip_block_resume(&adev->ip_blocks[i]);
+			if (r)
+				return r;
+		}
+	}
+
+	return 0;
+}
+
 /**
 * amdgpu_device_ip_resume - run resume for hardware IPs
 *
@ -3822,6 +3850,13 @@ static int amdgpu_device_ip_resume(struct amdgpu_device *adev)
 	if (adev->mman.buffer_funcs_ring->sched.ready)
 		amdgpu_ttm_set_buffer_funcs_status(adev, true);

+	if (r)
+		return r;
+
+	amdgpu_fence_driver_hw_init(adev);
+
+	r = amdgpu_device_ip_resume_phase3(adev);
+
 	return r;
 }

@ -4902,7 +4937,6 @@ int amdgpu_device_resume(struct drm_device *dev, bool notify_clients)
 		dev_err(adev->dev, "amdgpu_device_ip_resume failed (%d).\n", r);
 		goto exit;
 	}
-	amdgpu_fence_driver_hw_init(adev);

 	if (!adev->in_s0ix) {
 		r = amdgpu_amdkfd_resume(adev, adev->in_runpm);
@ -5487,6 +5521,10 @@ int amdgpu_device_reinit_after_reset(struct amdgpu_reset_context *reset_context)
 				if (tmp_adev->mman.buffer_funcs_ring->sched.ready)
 					amdgpu_ttm_set_buffer_funcs_status(tmp_adev, true);

+				r = amdgpu_device_ip_resume_phase3(tmp_adev);
+				if (r)
+					goto out;
+
 				if (vram_lost)
 					amdgpu_device_fill_reset_magic(tmp_adev);

--- a/drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c
@ -40,10 +40,12 @@
 static void hdp_v4_0_flush_hdp(struct amdgpu_device *adev,
 				struct amdgpu_ring *ring)
 {
-	if (!ring || !ring->funcs->emit_wreg)
+	if (!ring || !ring->funcs->emit_wreg) {
 		WREG32((adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >> 2, 0);
-	else
+		RREG32((adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >> 2);
+	} else {
 		amdgpu_ring_emit_wreg(ring, (adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >> 2, 0);
+	}
 }

 static void hdp_v4_0_invalidate_hdp(struct amdgpu_device *adev,
@ -54,11 +56,13 @@ static void hdp_v4_0_invalidate_hdp(struct amdgpu_device *adev,
 	    amdgpu_ip_version(adev, HDP_HWIP, 0) == IP_VERSION(4, 4, 5))
 		return;

-	if (!ring || !ring->funcs->emit_wreg)
+	if (!ring || !ring->funcs->emit_wreg) {
 		WREG32_SOC15_NO_KIQ(HDP, 0, mmHDP_READ_CACHE_INVALIDATE, 1);
-	else
+		RREG32_SOC15_NO_KIQ(HDP, 0, mmHDP_READ_CACHE_INVALIDATE);
+	} else {
 		amdgpu_ring_emit_wreg(ring, SOC15_REG_OFFSET(
 			HDP, 0, mmHDP_READ_CACHE_INVALIDATE), 1);
+	}
 }

 static void hdp_v4_0_query_ras_error_count(struct amdgpu_device *adev,
--- a/drivers/gpu/drm/amd/amdgpu/hdp_v5_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/hdp_v5_0.c
@ -31,10 +31,12 @@
 static void hdp_v5_0_flush_hdp(struct amdgpu_device *adev,
 				struct amdgpu_ring *ring)
 {
-	if (!ring || !ring->funcs->emit_wreg)
+	if (!ring || !ring->funcs->emit_wreg) {
 		WREG32((adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >> 2, 0);
-	else
+		RREG32((adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >> 2);
+	} else {
 		amdgpu_ring_emit_wreg(ring, (adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >> 2, 0);
+	}
 }

 static void hdp_v5_0_invalidate_hdp(struct amdgpu_device *adev,
@ -42,6 +44,7 @@ static void hdp_v5_0_invalidate_hdp(struct amdgpu_device *adev,
 {
 	if (!ring || !ring->funcs->emit_wreg) {
 		WREG32_SOC15_NO_KIQ(HDP, 0, mmHDP_READ_CACHE_INVALIDATE, 1);
+		RREG32_SOC15_NO_KIQ(HDP, 0, mmHDP_READ_CACHE_INVALIDATE);
 	} else {
 		amdgpu_ring_emit_wreg(ring, SOC15_REG_OFFSET(
 					HDP, 0, mmHDP_READ_CACHE_INVALIDATE), 1);
--- a/drivers/gpu/drm/amd/amdgpu/hdp_v5_2.c
+++ b/drivers/gpu/drm/amd/amdgpu/hdp_v5_2.c
@ -31,13 +31,15 @@
 static void hdp_v5_2_flush_hdp(struct amdgpu_device *adev,
 				struct amdgpu_ring *ring)
 {
-	if (!ring || !ring->funcs->emit_wreg)
+	if (!ring || !ring->funcs->emit_wreg) {
 		WREG32_NO_KIQ((adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >> 2,
 			0);
-	else
+		RREG32_NO_KIQ((adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >> 2);
+	} else {
 		amdgpu_ring_emit_wreg(ring,
 			(adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >> 2,
 			0);
+	}
 }

 static void hdp_v5_2_update_mem_power_gating(struct amdgpu_device *adev,
--- a/drivers/gpu/drm/amd/amdgpu/hdp_v6_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/hdp_v6_0.c
@ -34,10 +34,12 @@
 static void hdp_v6_0_flush_hdp(struct amdgpu_device *adev,
 				struct amdgpu_ring *ring)
 {
-	if (!ring || !ring->funcs->emit_wreg)
+	if (!ring || !ring->funcs->emit_wreg) {
 		WREG32((adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >> 2, 0);
-	else
+		RREG32((adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >> 2);
+	} else {
 		amdgpu_ring_emit_wreg(ring, (adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >> 2, 0);
+	}
 }

 static void hdp_v6_0_update_clock_gating(struct amdgpu_device *adev,
--- a/drivers/gpu/drm/amd/amdgpu/hdp_v7_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/hdp_v7_0.c
@ -31,10 +31,12 @@
 static void hdp_v7_0_flush_hdp(struct amdgpu_device *adev,
 				struct amdgpu_ring *ring)
 {
-	if (!ring || !ring->funcs->emit_wreg)
+	if (!ring || !ring->funcs->emit_wreg) {
 		WREG32((adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >> 2, 0);
-	else
+		RREG32((adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >> 2);
+	} else {
 		amdgpu_ring_emit_wreg(ring, (adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >> 2, 0);
+	}
 }

 static void hdp_v7_0_update_clock_gating(struct amdgpu_device *adev,
--- a/drivers/gpu/drm/amd/amdgpu/jpeg_v1_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v1_0.c
@ -604,7 +604,7 @@ static void jpeg_v1_0_set_irq_funcs(struct amdgpu_device *adev)
 static void jpeg_v1_0_ring_begin_use(struct amdgpu_ring *ring)
 {
 	struct	amdgpu_device *adev = ring->adev;
-	bool	set_clocks = !cancel_delayed_work_sync(&adev->jpeg.idle_work);
+	bool	set_clocks = !cancel_delayed_work_sync(&adev->vcn.idle_work);
 	int		cnt = 0;

 	mutex_lock(&adev->vcn.vcn1_jpeg1_workaround);
--- a/drivers/gpu/drm/amd/amdkfd/kfd_crat.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_crat.c
@ -1510,6 +1510,8 @@ static int kfd_fill_gpu_cache_info_from_gfx_config_v2(struct kfd_dev *kdev,
 	if (adev->gfx.config.gc_tcp_size_per_cu) {
 		pcache_info[i].cache_size = adev->gfx.config.gc_tcp_size_per_cu;
 		pcache_info[i].cache_level = 1;
+		/* Cacheline size not available in IP discovery for gc943,gc944 */
+		pcache_info[i].cache_line_size = 128;
 		pcache_info[i].flags = (CRAT_CACHE_FLAGS_ENABLED |
 					CRAT_CACHE_FLAGS_DATA_CACHE |
 					CRAT_CACHE_FLAGS_SIMD_CACHE);
@ -1521,6 +1523,7 @@ static int kfd_fill_gpu_cache_info_from_gfx_config_v2(struct kfd_dev *kdev,
 		pcache_info[i].cache_size =
 			adev->gfx.config.gc_l1_instruction_cache_size_per_sqc;
 		pcache_info[i].cache_level = 1;
+		pcache_info[i].cache_line_size = 64;
 		pcache_info[i].flags = (CRAT_CACHE_FLAGS_ENABLED |
 					CRAT_CACHE_FLAGS_INST_CACHE |
 					CRAT_CACHE_FLAGS_SIMD_CACHE);
@ -1531,6 +1534,7 @@ static int kfd_fill_gpu_cache_info_from_gfx_config_v2(struct kfd_dev *kdev,
 	if (adev->gfx.config.gc_l1_data_cache_size_per_sqc) {
 		pcache_info[i].cache_size = adev->gfx.config.gc_l1_data_cache_size_per_sqc;
 		pcache_info[i].cache_level = 1;
+		pcache_info[i].cache_line_size = 64;
 		pcache_info[i].flags = (CRAT_CACHE_FLAGS_ENABLED |
 					CRAT_CACHE_FLAGS_DATA_CACHE |
 					CRAT_CACHE_FLAGS_SIMD_CACHE);
@ -1541,6 +1545,7 @@ static int kfd_fill_gpu_cache_info_from_gfx_config_v2(struct kfd_dev *kdev,
 	if (adev->gfx.config.gc_tcc_size) {
 		pcache_info[i].cache_size = adev->gfx.config.gc_tcc_size;
 		pcache_info[i].cache_level = 2;
+		pcache_info[i].cache_line_size = 128;
 		pcache_info[i].flags = (CRAT_CACHE_FLAGS_ENABLED |
 					CRAT_CACHE_FLAGS_DATA_CACHE |
 					CRAT_CACHE_FLAGS_SIMD_CACHE);
@ -1551,6 +1556,7 @@ static int kfd_fill_gpu_cache_info_from_gfx_config_v2(struct kfd_dev *kdev,
 	if (adev->gmc.mall_size) {
 		pcache_info[i].cache_size = adev->gmc.mall_size / 1024;
 		pcache_info[i].cache_level = 3;
+		pcache_info[i].cache_line_size = 64;
 		pcache_info[i].flags = (CRAT_CACHE_FLAGS_ENABLED |
 					CRAT_CACHE_FLAGS_DATA_CACHE |
 					CRAT_CACHE_FLAGS_SIMD_CACHE);
--- a/drivers/gpu/drm/amd/amdkfd/kfd_device.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_device.c
@ -235,6 +235,9 @@ static void kfd_device_info_init(struct kfd_dev *kfd,
 			 */
 			kfd->device_info.needs_pci_atomics = true;
 			kfd->device_info.no_atomic_fw_version = kfd->adev->gfx.rs64_enable ? 509 : 0;
+		} else if (gc_version < IP_VERSION(13, 0, 0)) {
+			kfd->device_info.needs_pci_atomics = true;
+			kfd->device_info.no_atomic_fw_version = 2090;
 		} else {
 			kfd->device_info.needs_pci_atomics = true;
 		}
--- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
@ -3481,6 +3481,8 @@ static void update_connector_ext_caps(struct amdgpu_dm_connector *aconnector)
 		caps->aux_support = false;
 	else if (amdgpu_backlight == 1)
 		caps->aux_support = true;
+	if (caps->aux_support)
+		aconnector->dc_link->backlight_control_type = BACKLIGHT_CONTROL_AMD_AUX;

 	luminance_range = &conn_base->display_info.luminance_range;

--- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_helpers.c
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_helpers.c
@ -907,14 +907,14 @@ dm_helpers_probe_acpi_edid(void *data, u8 *buf, unsigned int block, size_t len)
 	struct drm_connector *connector = data;
 	struct acpi_device *acpidev = ACPI_COMPANION(connector->dev->dev);
 	unsigned char start = block * EDID_LENGTH;
-	void *edid;
+	struct edid *edid;
 	int r;

 	if (!acpidev)
 		return -ENODEV;

 	/* fetch the entire edid from BIOS */
-	r = acpi_video_get_edid(acpidev, ACPI_VIDEO_DISPLAY_LCD, -1, &edid);
+	r = acpi_video_get_edid(acpidev, ACPI_VIDEO_DISPLAY_LCD, -1, (void *)&edid);
 	if (r < 0) {
 		drm_dbg(connector->dev, "Failed to get EDID from ACPI: %d\n", r);
 		return r;
@ -924,7 +924,14 @@ dm_helpers_probe_acpi_edid(void *data, u8 *buf, unsigned int block, size_t len)
 		goto cleanup;
 	}

-	memcpy(buf, edid + start, len);
+	/* sanity check */
+	if (edid->revision < 4 || !(edid->input & DRM_EDID_INPUT_DIGITAL) ||
+	    (edid->input & DRM_EDID_DIGITAL_TYPE_MASK) == DRM_EDID_DIGITAL_TYPE_UNDEF) {
+		r = -EINVAL;
+		goto cleanup;
+	}
+
+	memcpy(buf, (void *)edid + start, len);
 	r = 0;

 cleanup:
--- a/drivers/gpu/drm/amd/display/dc/core/dc.c
+++ b/drivers/gpu/drm/amd/display/dc/core/dc.c
@ -6109,3 +6109,21 @@ struct dc_power_profile dc_get_power_profile_for_dc_state(const struct dc_state
 		profile.power_level = dc->res_pool->funcs->get_power_profile(context);
 	return profile;
 }
+
+/*
+ **********************************************************************************
+ * dc_get_det_buffer_size_from_state() - extracts detile buffer size from dc state
+ *
+ * Called when DM wants to log detile buffer size from dc_state
+ *
+ **********************************************************************************
+ */
+unsigned int dc_get_det_buffer_size_from_state(const struct dc_state *context)
+{
+	struct dc *dc = context->clk_mgr->ctx->dc;
+
+	if (dc->res_pool->funcs->get_det_buffer_size)
+		return dc->res_pool->funcs->get_det_buffer_size(context);
+	else
+		return 0;
+}
--- a/drivers/gpu/drm/amd/display/dc/core/dc_resource.c
+++ b/drivers/gpu/drm/amd/display/dc/core/dc_resource.c
@ -2094,7 +2094,8 @@ int resource_get_odm_slice_dst_width(struct pipe_ctx *otg_master,
 	count = resource_get_odm_slice_count(otg_master);
 	h_active = timing->h_addressable +
 			timing->h_border_left +
-			timing->h_border_right;
+			timing->h_border_right +
+			otg_master->hblank_borrow;
 	width = h_active / count;

 	if (otg_master->stream_res.tg)
@ -4026,6 +4027,41 @@ enum dc_status dc_validate_with_context(struct dc *dc,
 	return res;
 }

+/**
+ * decide_hblank_borrow - Decides the horizontal blanking borrow value for a given pipe context.
+ * @pipe_ctx: Pointer to the pipe context structure.
+ *
+ * This function calculates the horizontal blanking borrow value for a given pipe context based on the
+ * display stream compression (DSC) configuration. If the horizontal active pixels (hactive) are less
+ * than the total width of the DSC slices, it sets the hblank_borrow value to the difference. If the
+ * total horizontal timing minus the hblank_borrow value is less than 32, it resets the hblank_borrow
+ * value to 0.
+ */
+static void decide_hblank_borrow(struct pipe_ctx *pipe_ctx)
+{
+	uint32_t hactive;
+	uint32_t ceil_slice_width;
+	struct dc_stream_state *stream = NULL;
+
+	if (!pipe_ctx)
+		return;
+
+	stream = pipe_ctx->stream;
+
+	if (stream->timing.flags.DSC) {
+		hactive = stream->timing.h_addressable + stream->timing.h_border_left + stream->timing.h_border_right;
+
+		/* Assume if determined slices does not divide Hactive evenly, Hborrow is needed for padding*/
+		if (hactive % stream->timing.dsc_cfg.num_slices_h != 0) {
+			ceil_slice_width = (hactive / stream->timing.dsc_cfg.num_slices_h) + 1;
+			pipe_ctx->hblank_borrow = ceil_slice_width * stream->timing.dsc_cfg.num_slices_h - hactive;
+
+			if (stream->timing.h_total - hactive - pipe_ctx->hblank_borrow < 32)
+				pipe_ctx->hblank_borrow = 0;
+		}
+	}
+}
+
 /**
 * dc_validate_global_state() - Determine if hardware can support a given state
 *
@ -4064,6 +4100,10 @@ enum dc_status dc_validate_global_state(
 			if (pipe_ctx->stream != stream)
 				continue;

+			/* Decide whether hblank borrow is needed and save it in pipe_ctx */
+			if (dc->debug.enable_hblank_borrow)
+				decide_hblank_borrow(pipe_ctx);
+
 			if (dc->res_pool->funcs->patch_unknown_plane_state &&
 					pipe_ctx->plane_state &&
 					pipe_ctx->plane_state->tiling_info.gfx9.swizzle == DC_SW_UNKNOWN) {
--- a/drivers/gpu/drm/amd/display/dc/dc.h
+++ b/drivers/gpu/drm/amd/display/dc/dc.h
@ -290,6 +290,7 @@ struct dc_caps {
 	uint16_t subvp_vertical_int_margin_us;
 	bool seamless_odm;
 	uint32_t max_v_total;
+	bool vtotal_limited_by_fp2;
 	uint32_t max_disp_clock_khz_at_vmin;
 	uint8_t subvp_drr_vblank_start_margin_us;
 	bool cursor_not_scaled;
@ -1068,6 +1069,7 @@ struct dc_debug_options {
 	unsigned int scale_to_sharpness_policy;
 	bool skip_full_updated_if_possible;
 	unsigned int enable_oled_edp_power_up_opt;
+	bool enable_hblank_borrow;
 };


@ -2550,6 +2552,8 @@ struct dc_power_profile {

 struct dc_power_profile dc_get_power_profile_for_dc_state(const struct dc_state *context);

+unsigned int dc_get_det_buffer_size_from_state(const struct dc_state *context);
+
 /* DSC Interfaces */
 #include "dc_dsc.h"

--- a/drivers/gpu/drm/amd/display/dc/dc_spl_translate.c
+++ b/drivers/gpu/drm/amd/display/dc/dc_spl_translate.c
@ -120,7 +120,7 @@ void translate_SPL_in_params_from_pipe_ctx(struct pipe_ctx *pipe_ctx, struct spl
 	spl_in->odm_slice_index = resource_get_odm_slice_index(pipe_ctx);
 	// Make spl input basic out info output_size width point to stream h active
 	spl_in->basic_out.output_size.width =
-		stream->timing.h_addressable + stream->timing.h_border_left + stream->timing.h_border_right;
+		stream->timing.h_addressable + stream->timing.h_border_left + stream->timing.h_border_right + pipe_ctx->hblank_borrow;
 	// Make spl input basic out info output_size height point to v active
 	spl_in->basic_out.output_size.height =
 		stream->timing.v_addressable + stream->timing.v_border_bottom + stream->timing.v_border_top;
--- a/drivers/gpu/drm/amd/display/dc/dml2/display_mode_core.c
+++ b/drivers/gpu/drm/amd/display/dc/dml2/display_mode_core.c
@ -1222,6 +1222,7 @@ static dml_bool_t CalculatePrefetchSchedule(struct display_mode_lib_scratch_st *
 	s->dst_y_prefetch_oto = s->Tvm_oto_lines + 2 * s->Tr0_oto_lines + s->Lsw_oto;

 	s->dst_y_prefetch_equ = p->VStartup - (*p->TSetup + dml_max(p->TWait + p->TCalc, *p->Tdmdl)) / s->LineTime - (*p->DSTYAfterScaler + (dml_float_t) *p->DSTXAfterScaler / (dml_float_t)p->myPipe->HTotal);
+	s->dst_y_prefetch_equ = dml_min(s->dst_y_prefetch_equ, 63.75); // limit to the reg limit of U6.2 for DST_Y_PREFETCH

 #ifdef __DML_VBA_DEBUG__
 	dml_print("DML::%s: HTotal = %u\n", __func__, p->myPipe->HTotal);
--- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_translation_helper.c
+++ b/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_translation_helper.c
@ -339,11 +339,22 @@ void dml21_apply_soc_bb_overrides(struct dml2_initialize_instance_in_out *dml_in
 	// }
 }

+static unsigned int calc_max_hardware_v_total(const struct dc_stream_state *stream)
+{
+	unsigned int max_hw_v_total = stream->ctx->dc->caps.max_v_total;
+
+	if (stream->ctx->dc->caps.vtotal_limited_by_fp2) {
+		max_hw_v_total -= stream->timing.v_front_porch + 1;
+	}
+
+	return max_hw_v_total;
+}
+
 static void populate_dml21_timing_config_from_stream_state(struct dml2_timing_cfg *timing,
 		struct dc_stream_state *stream,
 		struct dml2_context *dml_ctx)
 {
-	unsigned int hblank_start, vblank_start;
+	unsigned int hblank_start, vblank_start, min_hardware_refresh_in_uhz;

 	timing->h_active = stream->timing.h_addressable + stream->timing.h_border_left + stream->timing.h_border_right;
 	timing->v_active = stream->timing.v_addressable + stream->timing.v_border_bottom + stream->timing.v_border_top;
@ -371,11 +382,23 @@ static void populate_dml21_timing_config_from_stream_state(struct dml2_timing_cf
 		- stream->timing.v_border_top - stream->timing.v_border_bottom;

 	timing->drr_config.enabled = stream->ignore_msa_timing_param;
-	timing->drr_config.min_refresh_uhz = stream->timing.min_refresh_in_uhz;
 	timing->drr_config.drr_active_variable = stream->vrr_active_variable;
 	timing->drr_config.drr_active_fixed = stream->vrr_active_fixed;
 	timing->drr_config.disallowed = !stream->allow_freesync;

+	/* limit min refresh rate to DC cap */
+	min_hardware_refresh_in_uhz = stream->timing.min_refresh_in_uhz;
+	if (stream->ctx->dc->caps.max_v_total != 0) {
+		min_hardware_refresh_in_uhz = div64_u64((stream->timing.pix_clk_100hz * 100000000ULL),
+				(stream->timing.h_total * (long long)calc_max_hardware_v_total(stream)));
+	}
+
+	if (stream->timing.min_refresh_in_uhz > min_hardware_refresh_in_uhz) {
+		timing->drr_config.min_refresh_uhz = stream->timing.min_refresh_in_uhz;
+	} else {
+		timing->drr_config.min_refresh_uhz = min_hardware_refresh_in_uhz;
+	}
+
 	if (dml_ctx->config.callbacks.get_max_flickerless_instant_vtotal_increase &&
 			stream->ctx->dc->config.enable_fpo_flicker_detection == 1)
 		timing->drr_config.max_instant_vtotal_delta = dml_ctx->config.callbacks.get_max_flickerless_instant_vtotal_increase(stream, false);
@ -422,6 +445,21 @@ static void populate_dml21_timing_config_from_stream_state(struct dml2_timing_cf
 	timing->vblank_nom = timing->v_total - timing->v_active;
 }

+/**
+ * adjust_dml21_hblank_timing_config_from_pipe_ctx - Adjusts the horizontal blanking timing configuration
+ *                                                   based on the pipe context.
+ * @timing: Pointer to the dml2_timing_cfg structure to be adjusted.
+ * @pipe: Pointer to the pipe_ctx structure containing the horizontal blanking borrow value.
+ *
+ * This function modifies the horizontal active and blank end timings by adding and subtracting
+ * the horizontal blanking borrow value from the pipe context, respectively.
+ */
+static void adjust_dml21_hblank_timing_config_from_pipe_ctx(struct dml2_timing_cfg *timing, struct pipe_ctx *pipe)
+{
+	timing->h_active += pipe->hblank_borrow;
+	timing->h_blank_end -= pipe->hblank_borrow;
+}
+
 static void populate_dml21_output_config_from_stream_state(struct dml2_link_output_cfg *output,
 		struct dc_stream_state *stream, const struct pipe_ctx *pipe)
 {
@ -709,6 +747,7 @@ static const struct scaler_data *get_scaler_data_for_plane(
 			temp_pipe->plane_state = pipe->plane_state;
 			temp_pipe->plane_res.scl_data.taps = pipe->plane_res.scl_data.taps;
 			temp_pipe->stream_res = pipe->stream_res;
+			temp_pipe->hblank_borrow = pipe->hblank_borrow;
 			dml_ctx->config.callbacks.build_scaling_params(temp_pipe);
 			break;
 		}
@ -973,6 +1012,7 @@ bool dml21_map_dc_state_into_dml_display_cfg(const struct dc *in_dc, struct dc_s

 		ASSERT(disp_cfg_stream_location >= 0 && disp_cfg_stream_location <= __DML2_WRAPPER_MAX_STREAMS_PLANES__);
 		populate_dml21_timing_config_from_stream_state(&dml_dispcfg->stream_descriptors[disp_cfg_stream_location].timing, context->streams[stream_index], dml_ctx);
+		adjust_dml21_hblank_timing_config_from_pipe_ctx(&dml_dispcfg->stream_descriptors[disp_cfg_stream_location].timing, &context->res_ctx.pipe_ctx[stream_index]);
 		populate_dml21_output_config_from_stream_state(&dml_dispcfg->stream_descriptors[disp_cfg_stream_location].output, context->streams[stream_index], &context->res_ctx.pipe_ctx[stream_index]);
 		populate_dml21_stream_overrides_from_stream_state(&dml_dispcfg->stream_descriptors[disp_cfg_stream_location], context->streams[stream_index]);

@ -1111,12 +1151,12 @@ void dml21_populate_pipe_ctx_dlg_params(struct dml2_context *dml_ctx, struct dc_
 	struct dc_crtc_timing *timing = &pipe_ctx->stream->timing;
 	union dml2_global_sync_programming *global_sync = &stream_programming->global_sync;

-	hactive = timing->h_addressable + timing->h_border_left + timing->h_border_right;
+	hactive = timing->h_addressable + timing->h_border_left + timing->h_border_right + pipe_ctx->hblank_borrow;
 	vactive = timing->v_addressable + timing->v_border_bottom + timing->v_border_top;
 	hblank_start = pipe_ctx->stream->timing.h_total - pipe_ctx->stream->timing.h_front_porch;
 	vblank_start = pipe_ctx->stream->timing.v_total - pipe_ctx->stream->timing.v_front_porch;

-	hblank_end = hblank_start - timing->h_addressable - timing->h_border_left - timing->h_border_right;
+	hblank_end = hblank_start - timing->h_addressable - timing->h_border_left - timing->h_border_right - pipe_ctx->hblank_borrow;
 	vblank_end = vblank_start - timing->v_addressable - timing->v_border_top - timing->v_border_bottom;

 	if (dml_ctx->config.svp_pstate.callbacks.get_pipe_subvp_type(context, pipe_ctx) == SUBVP_PHANTOM) {
--- a/drivers/gpu/drm/amd/display/dc/hwss/dcn32/dcn32_hwseq.c
+++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn32/dcn32_hwseq.c
@ -1049,7 +1049,8 @@ void dcn32_update_dsc_on_stream(struct pipe_ctx *pipe_ctx, bool enable)
 		}

 		/* Enable DSC hw block */
-		dsc_cfg.pic_width = (stream->timing.h_addressable + stream->timing.h_border_left + stream->timing.h_border_right) / opp_cnt;
+		dsc_cfg.pic_width = (stream->timing.h_addressable + pipe_ctx->hblank_borrow +
+				stream->timing.h_border_left + stream->timing.h_border_right) / opp_cnt;
 		dsc_cfg.pic_height = stream->timing.v_addressable + stream->timing.v_border_top + stream->timing.v_border_bottom;
 		dsc_cfg.pixel_encoding = stream->timing.pixel_encoding;
 		dsc_cfg.color_depth = stream->timing.display_color_depth;
--- a/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c
+++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c
@ -820,6 +820,7 @@ enum dc_status dcn401_enable_stream_timing(
 	int opp_cnt = 1;
 	int opp_inst[MAX_PIPES] = {0};
 	struct pipe_ctx *opp_heads[MAX_PIPES] = {0};
+	struct dc_crtc_timing patched_crtc_timing = stream->timing;
 	bool manual_mode;
 	unsigned int tmds_div = PIXEL_RATE_DIV_NA;
 	unsigned int unused_div = PIXEL_RATE_DIV_NA;
@ -874,9 +875,13 @@ enum dc_status dcn401_enable_stream_timing(
 	if (dc->hwseq->funcs.PLAT_58856_wa && (!dc_is_dp_signal(stream->signal)))
 		dc->hwseq->funcs.PLAT_58856_wa(context, pipe_ctx);

+	/* if we are borrowing from hblank, h_addressable needs to be adjusted */
+	if (dc->debug.enable_hblank_borrow)
+		patched_crtc_timing.h_addressable = patched_crtc_timing.h_addressable + pipe_ctx->hblank_borrow;
+
 	pipe_ctx->stream_res.tg->funcs->program_timing(
 			pipe_ctx->stream_res.tg,
-			&stream->timing,
+			&patched_crtc_timing,
 			pipe_ctx->pipe_dlg_param.vready_offset,
 			pipe_ctx->pipe_dlg_param.vstartup_start,
 			pipe_ctx->pipe_dlg_param.vupdate_offset,
--- a/drivers/gpu/drm/amd/display/dc/inc/core_types.h
+++ b/drivers/gpu/drm/amd/display/dc/inc/core_types.h
@ -219,6 +219,7 @@ struct resource_funcs {
 	 * Get indicator of power from a context that went through full validation
 	 */
 	int (*get_power_profile)(const struct dc_state *context);
+	unsigned int (*get_det_buffer_size)(const struct dc_state *context);
 };

 struct audio_support{
@ -477,6 +478,8 @@ struct pipe_ctx {
 	/* subvp_index: only valid if the pipe is a SUBVP_MAIN*/
 	uint8_t subvp_index;
 	struct pixel_rate_divider pixel_rate_divider;
+	/* pixels borrowed from hblank to hactive */
+	uint8_t hblank_borrow;
 };

 /* Data used for dynamic link encoder assignment.
--- a/drivers/gpu/drm/amd/display/dc/link/link_dpms.c
+++ b/drivers/gpu/drm/amd/display/dc/link/link_dpms.c
@ -808,7 +808,8 @@ void link_set_dsc_on_stream(struct pipe_ctx *pipe_ctx, bool enable)
 		enum optc_dsc_mode optc_dsc_mode;

 		/* Enable DSC hw block */
-		dsc_cfg.pic_width = (stream->timing.h_addressable + stream->timing.h_border_left + stream->timing.h_border_right) / opp_cnt;
+		dsc_cfg.pic_width = (stream->timing.h_addressable + pipe_ctx->hblank_borrow +
+				stream->timing.h_border_left + stream->timing.h_border_right) / opp_cnt;
 		dsc_cfg.pic_height = stream->timing.v_addressable + stream->timing.v_border_top + stream->timing.v_border_bottom;
 		dsc_cfg.pixel_encoding = stream->timing.pixel_encoding;
 		dsc_cfg.color_depth = stream->timing.display_color_depth;
--- a/drivers/gpu/drm/amd/display/dc/resource/dcn20/dcn20_resource.c
+++ b/drivers/gpu/drm/amd/display/dc/resource/dcn20/dcn20_resource.c
@ -1510,6 +1510,7 @@ bool dcn20_split_stream_for_odm(

 	if (prev_odm_pipe->plane_state) {
 		struct scaler_data *sd = &prev_odm_pipe->plane_res.scl_data;
+		struct output_pixel_processor *opp = next_odm_pipe->stream_res.opp;
 		int new_width;

 		/* HACTIVE halved for odm combine */
@ -1543,7 +1544,28 @@ bool dcn20_split_stream_for_odm(
 		sd->viewport_c.x += dc_fixpt_floor(dc_fixpt_mul_int(
 				sd->ratios.horz_c, sd->h_active - sd->recout.x));
 		sd->recout.x = 0;
+
+		/*
+		 * When odm is used in YcbCr422 or 420 colour space, a split screen
+		 * will be seen with the previous calculations since the extra left
+		 *  edge pixel is accounted for in fmt but not in viewport.
+		 *
+		 * Below are calculations which fix the split by fixing the calculations
+		 * if there is an extra left edge pixel.
+		 */
+		if (opp && opp->funcs->opp_get_left_edge_extra_pixel_count
+				&& opp->funcs->opp_get_left_edge_extra_pixel_count(
+					opp, next_odm_pipe->stream->timing.pixel_encoding,
+					resource_is_pipe_type(next_odm_pipe, OTG_MASTER)) == 1) {
+			sd->h_active += 1;
+			sd->recout.width += 1;
+			sd->viewport.x -= dc_fixpt_ceil(dc_fixpt_mul_int(sd->ratios.horz, 1));
+			sd->viewport_c.x -= dc_fixpt_ceil(dc_fixpt_mul_int(sd->ratios.horz, 1));
+			sd->viewport_c.width += dc_fixpt_ceil(dc_fixpt_mul_int(sd->ratios.horz, 1));
+			sd->viewport.width += dc_fixpt_ceil(dc_fixpt_mul_int(sd->ratios.horz, 1));
+		}
 	}
+
 	if (!next_odm_pipe->top_pipe)
 		next_odm_pipe->stream_res.opp = pool->opps[next_odm_pipe->pipe_idx];
 	else
@ -2132,6 +2154,7 @@ bool dcn20_fast_validate_bw(
 			ASSERT(0);
 		}
 	}
+
 	/* Actual dsc count per stream dsc validation*/
 	if (!dcn20_validate_dsc(dc, context)) {
 		context->bw_ctx.dml.vba.ValidationStatus[context->bw_ctx.dml.vba.soc.num_states] =
--- a/drivers/gpu/drm/amd/display/dc/resource/dcn30/dcn30_resource.c
+++ b/drivers/gpu/drm/amd/display/dc/resource/dcn30/dcn30_resource.c
@ -2353,6 +2353,7 @@ static bool dcn30_resource_construct(

 	dc->caps.dp_hdmi21_pcon_support = true;
 	dc->caps.max_v_total = (1 << 15) - 1;
+	dc->caps.vtotal_limited_by_fp2 = true;

 	/* read VBIOS LTTPR caps */
 	{
--- a/drivers/gpu/drm/amd/display/dc/resource/dcn302/dcn302_resource.c
+++ b/drivers/gpu/drm/amd/display/dc/resource/dcn302/dcn302_resource.c
@ -1233,6 +1233,7 @@ static bool dcn302_resource_construct(
 	dc->caps.extended_aux_timeout_support = true;
 	dc->caps.dmcub_support = true;
 	dc->caps.max_v_total = (1 << 15) - 1;
+	dc->caps.vtotal_limited_by_fp2 = true;

 	/* Color pipeline capabilities */
 	dc->caps.color.dpp.dcn_arch = 1;
--- a/drivers/gpu/drm/amd/display/dc/resource/dcn303/dcn303_resource.c
+++ b/drivers/gpu/drm/amd/display/dc/resource/dcn303/dcn303_resource.c
@ -1178,6 +1178,7 @@ static bool dcn303_resource_construct(
 	dc->caps.extended_aux_timeout_support = true;
 	dc->caps.dmcub_support = true;
 	dc->caps.max_v_total = (1 << 15) - 1;
+	dc->caps.vtotal_limited_by_fp2 = true;

 	/* Color pipeline capabilities */
 	dc->caps.color.dpp.dcn_arch = 1;
--- a/drivers/gpu/drm/amd/display/dc/resource/dcn31/dcn31_resource.c
+++ b/drivers/gpu/drm/amd/display/dc/resource/dcn31/dcn31_resource.c
@ -1720,6 +1720,12 @@ int dcn31_populate_dml_pipes_from_context(
 	return pipe_cnt;
 }

+unsigned int dcn31_get_det_buffer_size(
+	const struct dc_state *context)
+{
+	return context->bw_ctx.dml.ip.det_buffer_size_kbytes;
+}
+
 void dcn31_calculate_wm_and_dlg(
 		struct dc *dc, struct dc_state *context,
 		display_e2e_pipe_params_st *pipes,
@ -1842,6 +1848,7 @@ static struct resource_funcs dcn31_res_pool_funcs = {
 	.update_bw_bounding_box = dcn31_update_bw_bounding_box,
 	.patch_unknown_plane_state = dcn20_patch_unknown_plane_state,
 	.get_panel_config_defaults = dcn31_get_panel_config_defaults,
+	.get_det_buffer_size = dcn31_get_det_buffer_size,
 };

 static struct clock_source *dcn30_clock_source_create(
--- a/drivers/gpu/drm/amd/display/dc/resource/dcn31/dcn31_resource.h
+++ b/drivers/gpu/drm/amd/display/dc/resource/dcn31/dcn31_resource.h
@ -63,6 +63,9 @@ struct resource_pool *dcn31_create_resource_pool(
 		const struct dc_init_data *init_data,
 		struct dc *dc);

+unsigned int dcn31_get_det_buffer_size(
+	const struct dc_state *context);
+
 /*temp: B0 specific before switch to dcn313 headers*/
 #ifndef regPHYPLLF_PIXCLK_RESYNC_CNTL
 #define regPHYPLLF_PIXCLK_RESYNC_CNTL 0x007e
--- a/drivers/gpu/drm/amd/display/dc/resource/dcn314/dcn314_resource.c
+++ b/drivers/gpu/drm/amd/display/dc/resource/dcn314/dcn314_resource.c
@ -1777,6 +1777,7 @@ static struct resource_funcs dcn314_res_pool_funcs = {
 	.patch_unknown_plane_state = dcn20_patch_unknown_plane_state,
 	.get_panel_config_defaults = dcn314_get_panel_config_defaults,
 	.get_preferred_eng_id_dpia = dcn314_get_preferred_eng_id_dpia,
+	.get_det_buffer_size = dcn31_get_det_buffer_size,
 };

 static struct clock_source *dcn30_clock_source_create(
--- a/drivers/gpu/drm/amd/display/dc/resource/dcn315/dcn315_resource.c
+++ b/drivers/gpu/drm/amd/display/dc/resource/dcn315/dcn315_resource.c
@ -1845,6 +1845,7 @@ static struct resource_funcs dcn315_res_pool_funcs = {
 	.patch_unknown_plane_state = dcn20_patch_unknown_plane_state,
 	.get_panel_config_defaults = dcn315_get_panel_config_defaults,
 	.get_power_profile = dcn315_get_power_profile,
+	.get_det_buffer_size = dcn31_get_det_buffer_size,
 };

 static bool dcn315_resource_construct(
--- a/drivers/gpu/drm/amd/display/dc/resource/dcn316/dcn316_resource.c
+++ b/drivers/gpu/drm/amd/display/dc/resource/dcn316/dcn316_resource.c
@ -1719,6 +1719,7 @@ static struct resource_funcs dcn316_res_pool_funcs = {
 	.update_bw_bounding_box = dcn316_update_bw_bounding_box,
 	.patch_unknown_plane_state = dcn20_patch_unknown_plane_state,
 	.get_panel_config_defaults = dcn316_get_panel_config_defaults,
+	.get_det_buffer_size = dcn31_get_det_buffer_size,
 };

 static bool dcn316_resource_construct(
--- a/drivers/gpu/drm/amd/display/dc/resource/dcn32/dcn32_resource.c
+++ b/drivers/gpu/drm/amd/display/dc/resource/dcn32/dcn32_resource.c
@ -2189,6 +2189,7 @@ static bool dcn32_resource_construct(
 	dc->caps.dmcub_support = true;
 	dc->caps.seamless_odm = true;
 	dc->caps.max_v_total = (1 << 15) - 1;
+	dc->caps.vtotal_limited_by_fp2 = true;

 	/* Color pipeline capabilities */
 	dc->caps.color.dpp.dcn_arch = 1;
@ -2803,6 +2804,7 @@ struct pipe_ctx *dcn32_acquire_free_pipe_as_secondary_opp_head(
 		free_pipe->plane_res.xfm = pool->transforms[free_pipe_idx];
 		free_pipe->plane_res.dpp = pool->dpps[free_pipe_idx];
 		free_pipe->plane_res.mpcc_inst = pool->dpps[free_pipe_idx]->inst;
+		free_pipe->hblank_borrow = otg_master->hblank_borrow;
 		if (free_pipe->stream->timing.flags.DSC == 1) {
 			dcn20_acquire_dsc(free_pipe->stream->ctx->dc,
 					&new_ctx->res_ctx,
--- a/drivers/gpu/drm/amd/display/dc/resource/dcn321/dcn321_resource.c
+++ b/drivers/gpu/drm/amd/display/dc/resource/dcn321/dcn321_resource.c
@ -1742,6 +1742,7 @@ static bool dcn321_resource_construct(
 	dc->caps.extended_aux_timeout_support = true;
 	dc->caps.dmcub_support = true;
 	dc->caps.max_v_total = (1 << 15) - 1;
+	dc->caps.vtotal_limited_by_fp2 = true;

 	/* Color pipeline capabilities */
 	dc->caps.color.dpp.dcn_arch = 1;
--- a/drivers/gpu/drm/amd/display/dc/resource/dcn35/dcn35_resource.c
+++ b/drivers/gpu/drm/amd/display/dc/resource/dcn35/dcn35_resource.c
@ -1778,6 +1778,7 @@ static struct resource_funcs dcn35_res_pool_funcs = {
 	.patch_unknown_plane_state = dcn20_patch_unknown_plane_state,
 	.get_panel_config_defaults = dcn35_get_panel_config_defaults,
 	.get_preferred_eng_id_dpia = dcn35_get_preferred_eng_id_dpia,
+	.get_det_buffer_size = dcn31_get_det_buffer_size,
 };

 static bool dcn35_resource_construct(
@ -1849,6 +1850,7 @@ static bool dcn35_resource_construct(
 	dc->caps.zstate_support = true;
 	dc->caps.ips_support = true;
 	dc->caps.max_v_total = (1 << 15) - 1;
+	dc->caps.vtotal_limited_by_fp2 = true;

 	/* Color pipeline capabilities */
 	dc->caps.color.dpp.dcn_arch = 1;
--- a/drivers/gpu/drm/amd/display/dc/resource/dcn351/dcn351_resource.c
+++ b/drivers/gpu/drm/amd/display/dc/resource/dcn351/dcn351_resource.c
@ -1757,6 +1757,7 @@ static struct resource_funcs dcn351_res_pool_funcs = {
 	.patch_unknown_plane_state = dcn20_patch_unknown_plane_state,
 	.get_panel_config_defaults = dcn35_get_panel_config_defaults,
 	.get_preferred_eng_id_dpia = dcn351_get_preferred_eng_id_dpia,
+	.get_det_buffer_size = dcn31_get_det_buffer_size,
 };

 static bool dcn351_resource_construct(
@ -1828,6 +1829,7 @@ static bool dcn351_resource_construct(
 	dc->caps.zstate_support = true;
 	dc->caps.ips_support = true;
 	dc->caps.max_v_total = (1 << 15) - 1;
+	dc->caps.vtotal_limited_by_fp2 = true;

 	/* Color pipeline capabilities */
 	dc->caps.color.dpp.dcn_arch = 1;
--- a/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.c
+++ b/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.c
@ -1864,6 +1864,7 @@ static bool dcn401_resource_construct(
 	dc->caps.extended_aux_timeout_support = true;
 	dc->caps.dmcub_support = true;
 	dc->caps.max_v_total = (1 << 15) - 1;
+	dc->caps.vtotal_limited_by_fp2 = true;

 	if (ASICREV_IS_GC_12_0_1_A0(dc->ctx->asic_id.hw_internal_rev))
 		dc->caps.dcc_plane_width_limit = 7680;
--- a/drivers/gpu/drm/amd/display/modules/freesync/freesync.c
+++ b/drivers/gpu/drm/amd/display/modules/freesync/freesync.c
@ -122,6 +122,17 @@ static unsigned int calc_duration_in_us_from_v_total(
 	return duration_in_us;
 }

+static unsigned int calc_max_hardware_v_total(const struct dc_stream_state *stream)
+{
+	unsigned int max_hw_v_total = stream->ctx->dc->caps.max_v_total;
+
+	if (stream->ctx->dc->caps.vtotal_limited_by_fp2) {
+		max_hw_v_total -= stream->timing.v_front_porch + 1;
+	}
+
+	return max_hw_v_total;
+}
+
 unsigned int mod_freesync_calc_v_total_from_refresh(
 		const struct dc_stream_state *stream,
 		unsigned int refresh_in_uhz)
@ -1016,7 +1027,7 @@ void mod_freesync_build_vrr_params(struct mod_freesync *mod_freesync,

 	if (stream->ctx->dc->caps.max_v_total != 0 && stream->timing.h_total != 0) {
 		min_hardware_refresh_in_uhz = div64_u64((stream->timing.pix_clk_100hz * 100000000ULL),
-			(stream->timing.h_total * (long long)stream->ctx->dc->caps.max_v_total));
+			(stream->timing.h_total * (long long)calc_max_hardware_v_total(stream)));
 	}
 	/* Limit minimum refresh rate to what can be supported by hardware */
 	min_refresh_in_uhz = min_hardware_refresh_in_uhz > in_config->min_refresh_in_uhz ?
--- a/drivers/gpu/drm/amd/pm/amdgpu_pm.c
+++ b/drivers/gpu/drm/amd/pm/amdgpu_pm.c
@ -1361,7 +1361,11 @@ static ssize_t amdgpu_set_pp_mclk_od(struct device *dev,
 * create a custom set of heuristics, write a string of numbers to the file
 * starting with the number of the custom profile along with a setting
 * for each heuristic parameter.  Due to differences across asic families
- * the heuristic parameters vary from family to family.
+ * the heuristic parameters vary from family to family. Additionally,
+ * you can apply the custom heuristics to different clock domains.  Each
+ * clock domain is considered a distinct operation so if you modify the
+ * gfxclk heuristics and then the memclk heuristics, the all of the
+ * custom heuristics will be retained until you switch to another profile.
 *
 */

--- a/drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c
+++ b/drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c
@ -72,6 +72,10 @@ static int smu_set_power_limit(void *handle, uint32_t limit);
 static int smu_set_fan_speed_rpm(void *handle, uint32_t speed);
 static int smu_set_gfx_cgpg(struct smu_context *smu, bool enabled);
 static int smu_set_mp1_state(void *handle, enum pp_mp1_state mp1_state);
+static void smu_power_profile_mode_get(struct smu_context *smu,
+				       enum PP_SMC_POWER_PROFILE profile_mode);
+static void smu_power_profile_mode_put(struct smu_context *smu,
+				       enum PP_SMC_POWER_PROFILE profile_mode);

 static int smu_sys_get_pp_feature_mask(void *handle,
 				       char *buf)
@ -1259,42 +1263,19 @@ static int smu_sw_init(struct amdgpu_ip_block *ip_block)
 	INIT_WORK(&smu->interrupt_work, smu_interrupt_work_fn);
 	atomic64_set(&smu->throttle_int_counter, 0);
 	smu->watermarks_bitmap = 0;
-	smu->power_profile_mode = PP_SMC_POWER_PROFILE_BOOTUP_DEFAULT;
-	smu->default_power_profile_mode = PP_SMC_POWER_PROFILE_BOOTUP_DEFAULT;
-	smu->user_dpm_profile.user_workload_mask = 0;

 	atomic_set(&smu->smu_power.power_gate.vcn_gated, 1);
 	atomic_set(&smu->smu_power.power_gate.jpeg_gated, 1);
 	atomic_set(&smu->smu_power.power_gate.vpe_gated, 1);
 	atomic_set(&smu->smu_power.power_gate.umsch_mm_gated, 1);

-	smu->workload_priority[PP_SMC_POWER_PROFILE_BOOTUP_DEFAULT] = 0;
-	smu->workload_priority[PP_SMC_POWER_PROFILE_FULLSCREEN3D] = 1;
-	smu->workload_priority[PP_SMC_POWER_PROFILE_POWERSAVING] = 2;
-	smu->workload_priority[PP_SMC_POWER_PROFILE_VIDEO] = 3;
-	smu->workload_priority[PP_SMC_POWER_PROFILE_VR] = 4;
-	smu->workload_priority[PP_SMC_POWER_PROFILE_COMPUTE] = 5;
-	smu->workload_priority[PP_SMC_POWER_PROFILE_CUSTOM] = 6;
-
 	if (smu->is_apu ||
-	    !smu_is_workload_profile_available(smu, PP_SMC_POWER_PROFILE_FULLSCREEN3D)) {
-		smu->driver_workload_mask =
-			1 << smu->workload_priority[PP_SMC_POWER_PROFILE_BOOTUP_DEFAULT];
-	} else {
-		smu->driver_workload_mask =
-			1 << smu->workload_priority[PP_SMC_POWER_PROFILE_FULLSCREEN3D];
-		smu->default_power_profile_mode = PP_SMC_POWER_PROFILE_FULLSCREEN3D;
-	}
+	    !smu_is_workload_profile_available(smu, PP_SMC_POWER_PROFILE_FULLSCREEN3D))
+		smu->power_profile_mode = PP_SMC_POWER_PROFILE_BOOTUP_DEFAULT;
+	else
+		smu->power_profile_mode = PP_SMC_POWER_PROFILE_FULLSCREEN3D;
+	smu_power_profile_mode_get(smu, smu->power_profile_mode);

-	smu->workload_mask = smu->driver_workload_mask |
-							smu->user_dpm_profile.user_workload_mask;
-	smu->workload_setting[0] = PP_SMC_POWER_PROFILE_BOOTUP_DEFAULT;
-	smu->workload_setting[1] = PP_SMC_POWER_PROFILE_FULLSCREEN3D;
-	smu->workload_setting[2] = PP_SMC_POWER_PROFILE_POWERSAVING;
-	smu->workload_setting[3] = PP_SMC_POWER_PROFILE_VIDEO;
-	smu->workload_setting[4] = PP_SMC_POWER_PROFILE_VR;
-	smu->workload_setting[5] = PP_SMC_POWER_PROFILE_COMPUTE;
-	smu->workload_setting[6] = PP_SMC_POWER_PROFILE_CUSTOM;
 	smu->display_config = &adev->pm.pm_display_cfg;

 	smu->smu_dpm.dpm_level = AMD_DPM_FORCED_LEVEL_AUTO;
@ -1347,6 +1328,11 @@ static int smu_sw_fini(struct amdgpu_ip_block *ip_block)
 		return ret;
 	}

+	if (smu->custom_profile_params) {
+		kfree(smu->custom_profile_params);
+		smu->custom_profile_params = NULL;
+	}
+
 	smu_fini_microcode(smu);

 	return 0;
@ -2131,6 +2117,9 @@ static int smu_suspend(struct amdgpu_ip_block *ip_block)
 	if (!ret)
 		adev->gfx.gfx_off_entrycount = count;

+	/* clear this on suspend so it will get reprogrammed on resume */
+	smu->workload_mask = 0;
+
 	return 0;
 }

@ -2243,25 +2232,49 @@ static int smu_enable_umd_pstate(void *handle,
 }

 static int smu_bump_power_profile_mode(struct smu_context *smu,
-					   long *param,
-					   uint32_t param_size)
+				       long *custom_params,
+				       u32 custom_params_max_idx)
 {
-	int ret = 0;
+	u32 workload_mask = 0;
+	int i, ret = 0;
+
+	for (i = 0; i < PP_SMC_POWER_PROFILE_COUNT; i++) {
+		if (smu->workload_refcount[i])
+			workload_mask |= 1 << i;
+	}
+
+	if (smu->workload_mask == workload_mask)
+		return 0;

 	if (smu->ppt_funcs->set_power_profile_mode)
-		ret = smu->ppt_funcs->set_power_profile_mode(smu, param, param_size);
+		ret = smu->ppt_funcs->set_power_profile_mode(smu, workload_mask,
+							     custom_params,
+							     custom_params_max_idx);
+
+	if (!ret)
+		smu->workload_mask = workload_mask;

 	return ret;
 }

+static void smu_power_profile_mode_get(struct smu_context *smu,
+				       enum PP_SMC_POWER_PROFILE profile_mode)
+{
+	smu->workload_refcount[profile_mode]++;
+}
+
+static void smu_power_profile_mode_put(struct smu_context *smu,
+				       enum PP_SMC_POWER_PROFILE profile_mode)
+{
+	if (smu->workload_refcount[profile_mode])
+		smu->workload_refcount[profile_mode]--;
+}
+
 static int smu_adjust_power_state_dynamic(struct smu_context *smu,
 					  enum amd_dpm_forced_level level,
-					  bool skip_display_settings,
-					  bool init)
+					  bool skip_display_settings)
 {
 	int ret = 0;
-	int index = 0;
-	long workload[1];
 	struct smu_dpm_context *smu_dpm_ctx = &(smu->smu_dpm);

 	if (!skip_display_settings) {
@ -2298,14 +2311,8 @@ static int smu_adjust_power_state_dynamic(struct smu_context *smu,
 	}

 	if (smu_dpm_ctx->dpm_level != AMD_DPM_FORCED_LEVEL_MANUAL &&
-		smu_dpm_ctx->dpm_level != AMD_DPM_FORCED_LEVEL_PERF_DETERMINISM) {
-		index = fls(smu->workload_mask);
-		index = index > 0 && index <= WORKLOAD_POLICY_MAX ? index - 1 : 0;
-		workload[0] = smu->workload_setting[index];
-
-		if (init || smu->power_profile_mode != workload[0])
-			smu_bump_power_profile_mode(smu, workload, 0);
-	}
+	    smu_dpm_ctx->dpm_level != AMD_DPM_FORCED_LEVEL_PERF_DETERMINISM)
+		smu_bump_power_profile_mode(smu, NULL, 0);

 	return ret;
 }
@ -2324,13 +2331,13 @@ static int smu_handle_task(struct smu_context *smu,
 		ret = smu_pre_display_config_changed(smu);
 		if (ret)
 			return ret;
-		ret = smu_adjust_power_state_dynamic(smu, level, false, false);
+		ret = smu_adjust_power_state_dynamic(smu, level, false);
 		break;
 	case AMD_PP_TASK_COMPLETE_INIT:
-		ret = smu_adjust_power_state_dynamic(smu, level, true, true);
+		ret = smu_adjust_power_state_dynamic(smu, level, true);
 		break;
 	case AMD_PP_TASK_READJUST_POWER_STATE:
-		ret = smu_adjust_power_state_dynamic(smu, level, true, false);
+		ret = smu_adjust_power_state_dynamic(smu, level, true);
 		break;
 	default:
 		break;
@ -2352,12 +2359,11 @@ static int smu_handle_dpm_task(void *handle,

 static int smu_switch_power_profile(void *handle,
 				    enum PP_SMC_POWER_PROFILE type,
-				    bool en)
+				    bool enable)
 {
 	struct smu_context *smu = handle;
 	struct smu_dpm_context *smu_dpm_ctx = &(smu->smu_dpm);
-	long workload[1];
-	uint32_t index;
+	int ret;

 	if (!smu->pm_enabled || !smu->adev->pm.dpm_enabled)
 		return -EOPNOTSUPP;
@ -2365,24 +2371,21 @@ static int smu_switch_power_profile(void *handle,
 	if (!(type < PP_SMC_POWER_PROFILE_CUSTOM))
 		return -EINVAL;

-	if (!en) {
-		smu->driver_workload_mask &= ~(1 << smu->workload_priority[type]);
-		index = fls(smu->workload_mask);
-		index = index > 0 && index <= WORKLOAD_POLICY_MAX ? index - 1 : 0;
-		workload[0] = smu->workload_setting[index];
-	} else {
-		smu->driver_workload_mask |= (1 << smu->workload_priority[type]);
-		index = fls(smu->workload_mask);
-		index = index <= WORKLOAD_POLICY_MAX ? index - 1 : 0;
-		workload[0] = smu->workload_setting[index];
-	}
-
-	smu->workload_mask = smu->driver_workload_mask |
-						 smu->user_dpm_profile.user_workload_mask;
-
 	if (smu_dpm_ctx->dpm_level != AMD_DPM_FORCED_LEVEL_MANUAL &&
-		smu_dpm_ctx->dpm_level != AMD_DPM_FORCED_LEVEL_PERF_DETERMINISM)
-		smu_bump_power_profile_mode(smu, workload, 0);
+	    smu_dpm_ctx->dpm_level != AMD_DPM_FORCED_LEVEL_PERF_DETERMINISM) {
+		if (enable)
+			smu_power_profile_mode_get(smu, type);
+		else
+			smu_power_profile_mode_put(smu, type);
+		ret = smu_bump_power_profile_mode(smu, NULL, 0);
+		if (ret) {
+			if (enable)
+				smu_power_profile_mode_put(smu, type);
+			else
+				smu_power_profile_mode_get(smu, type);
+			return ret;
+		}
+	}

 	return 0;
 }
@ -3074,21 +3077,33 @@ static int smu_set_power_profile_mode(void *handle,
 				      uint32_t param_size)
 {
 	struct smu_context *smu = handle;
-	int ret;
+	bool custom = false;
+	int ret = 0;

 	if (!smu->pm_enabled || !smu->adev->pm.dpm_enabled ||
 	    !smu->ppt_funcs->set_power_profile_mode)
 		return -EOPNOTSUPP;

-	if (smu->user_dpm_profile.user_workload_mask &
-	   (1 << smu->workload_priority[param[param_size]]))
-	   return 0;
+	if (param[param_size] == PP_SMC_POWER_PROFILE_CUSTOM) {
+		custom = true;
+		/* clear frontend mask so custom changes propogate */
+		smu->workload_mask = 0;
+	}

-	smu->user_dpm_profile.user_workload_mask =
-		(1 << smu->workload_priority[param[param_size]]);
-	smu->workload_mask = smu->user_dpm_profile.user_workload_mask |
-		smu->driver_workload_mask;
-	ret = smu_bump_power_profile_mode(smu, param, param_size);
+	if ((param[param_size] != smu->power_profile_mode) || custom) {
+		/* clear the old user preference */
+		smu_power_profile_mode_put(smu, smu->power_profile_mode);
+		/* set the new user preference */
+		smu_power_profile_mode_get(smu, param[param_size]);
+		ret = smu_bump_power_profile_mode(smu,
+						  custom ? param : NULL,
+						  custom ? param_size : 0);
+		if (ret)
+			smu_power_profile_mode_put(smu, param[param_size]);
+		else
+			/* store the user's preference */
+			smu->power_profile_mode = param[param_size];
+	}

 	return ret;
 }
--- a/drivers/gpu/drm/amd/pm/swsmu/inc/amdgpu_smu.h
+++ b/drivers/gpu/drm/amd/pm/swsmu/inc/amdgpu_smu.h
@ -240,7 +240,6 @@ struct smu_user_dpm_profile {
 	/* user clock state information */
 	uint32_t clk_mask[SMU_CLK_COUNT];
 	uint32_t clk_dependency;
-	uint32_t user_workload_mask;
 };

 #define SMU_TABLE_INIT(tables, table_id, s, a, d)	\
@ -557,12 +556,13 @@ struct smu_context {
 	uint32_t hard_min_uclk_req_from_dal;
 	bool disable_uclk_switch;

+	/* asic agnostic workload mask */
 	uint32_t workload_mask;
-	uint32_t driver_workload_mask;
-	uint32_t workload_priority[WORKLOAD_POLICY_MAX];
-	uint32_t workload_setting[WORKLOAD_POLICY_MAX];
+	/* default/user workload preference */
 	uint32_t power_profile_mode;
-	uint32_t default_power_profile_mode;
+	uint32_t workload_refcount[PP_SMC_POWER_PROFILE_COUNT];
+	/* backend specific custom workload settings */
+	long *custom_profile_params;
 	bool pm_enabled;
 	bool is_apu;

@ -733,9 +733,12 @@ struct pptable_funcs {
 	 * @set_power_profile_mode: Set a power profile mode. Also used to
 	 *                          create/set custom power profile modes.
 	 * &input: Power profile mode parameters.
-	 * &size: Size of &input.
+	 * &workload_mask: mask of workloads to enable
+	 * &custom_params: custom profile parameters
+	 * &custom_params_max_idx: max valid idx into custom_params
 	 */
-	int (*set_power_profile_mode)(struct smu_context *smu, long *input, uint32_t size);
+	int (*set_power_profile_mode)(struct smu_context *smu, u32 workload_mask,
+				      long *custom_params, u32 custom_params_max_idx);

 	/**
 	 * @dpm_set_vcn_enable: Enable/disable VCN engine dynamic power
--- a/drivers/gpu/drm/amd/pm/swsmu/smu11/arcturus_ppt.c
+++ b/drivers/gpu/drm/amd/pm/swsmu/smu11/arcturus_ppt.c
@ -1445,97 +1445,120 @@ static int arcturus_get_power_profile_mode(struct smu_context *smu,
 	return size;
 }

-static int arcturus_set_power_profile_mode(struct smu_context *smu,
-					   long *input,
-					   uint32_t size)
+#define ARCTURUS_CUSTOM_PARAMS_COUNT 10
+#define ARCTURUS_CUSTOM_PARAMS_CLOCK_COUNT 2
+#define ARCTURUS_CUSTOM_PARAMS_SIZE (ARCTURUS_CUSTOM_PARAMS_CLOCK_COUNT * ARCTURUS_CUSTOM_PARAMS_COUNT * sizeof(long))
+
+static int arcturus_set_power_profile_mode_coeff(struct smu_context *smu,
+						 long *input)
 {
 	DpmActivityMonitorCoeffInt_t activity_monitor;
-	int workload_type = 0;
-	uint32_t profile_mode = input[size];
-	int ret = 0;
+	int ret, idx;

-	if (profile_mode > PP_SMC_POWER_PROFILE_CUSTOM) {
-		dev_err(smu->adev->dev, "Invalid power profile mode %d\n", profile_mode);
-		return -EINVAL;
-	}
-
-	if ((profile_mode == PP_SMC_POWER_PROFILE_CUSTOM) &&
-	     (smu->smc_fw_version >= 0x360d00)) {
-		if (size != 10)
-			return -EINVAL;
-
-		ret = smu_cmn_update_table(smu,
-				       SMU_TABLE_ACTIVITY_MONITOR_COEFF,
-				       WORKLOAD_PPLIB_CUSTOM_BIT,
-				       (void *)(&activity_monitor),
-				       false);
-		if (ret) {
-			dev_err(smu->adev->dev, "[%s] Failed to get activity monitor!", __func__);
-			return ret;
-		}
-
-		switch (input[0]) {
-		case 0: /* Gfxclk */
-			activity_monitor.Gfx_FPS = input[1];
-			activity_monitor.Gfx_UseRlcBusy = input[2];
-			activity_monitor.Gfx_MinActiveFreqType = input[3];
-			activity_monitor.Gfx_MinActiveFreq = input[4];
-			activity_monitor.Gfx_BoosterFreqType = input[5];
-			activity_monitor.Gfx_BoosterFreq = input[6];
-			activity_monitor.Gfx_PD_Data_limit_c = input[7];
-			activity_monitor.Gfx_PD_Data_error_coeff = input[8];
-			activity_monitor.Gfx_PD_Data_error_rate_coeff = input[9];
-			break;
-		case 1: /* Uclk */
-			activity_monitor.Mem_FPS = input[1];
-			activity_monitor.Mem_UseRlcBusy = input[2];
-			activity_monitor.Mem_MinActiveFreqType = input[3];
-			activity_monitor.Mem_MinActiveFreq = input[4];
-			activity_monitor.Mem_BoosterFreqType = input[5];
-			activity_monitor.Mem_BoosterFreq = input[6];
-			activity_monitor.Mem_PD_Data_limit_c = input[7];
-			activity_monitor.Mem_PD_Data_error_coeff = input[8];
-			activity_monitor.Mem_PD_Data_error_rate_coeff = input[9];
-			break;
-		default:
-			return -EINVAL;
-		}
-
-		ret = smu_cmn_update_table(smu,
-				       SMU_TABLE_ACTIVITY_MONITOR_COEFF,
-				       WORKLOAD_PPLIB_CUSTOM_BIT,
-				       (void *)(&activity_monitor),
-				       true);
-		if (ret) {
-			dev_err(smu->adev->dev, "[%s] Failed to set activity monitor!", __func__);
-			return ret;
-		}
-	}
-
-	/*
-	 * Conv PP_SMC_POWER_PROFILE* to WORKLOAD_PPLIB_*_BIT
-	 * Not all profile modes are supported on arcturus.
-	 */
-	workload_type = smu_cmn_to_asic_specific_index(smu,
-						       CMN2ASIC_MAPPING_WORKLOAD,
-						       profile_mode);
-	if (workload_type < 0) {
-		dev_dbg(smu->adev->dev, "Unsupported power profile mode %d on arcturus\n", profile_mode);
-		return -EINVAL;
-	}
-
-	ret = smu_cmn_send_smc_msg_with_param(smu,
-					  SMU_MSG_SetWorkloadMask,
-					  smu->workload_mask,
-					  NULL);
+	ret = smu_cmn_update_table(smu,
+				   SMU_TABLE_ACTIVITY_MONITOR_COEFF,
+				   WORKLOAD_PPLIB_CUSTOM_BIT,
+				   (void *)(&activity_monitor),
+				   false);
 	if (ret) {
-		dev_err(smu->adev->dev, "Fail to set workload type %d\n", workload_type);
+		dev_err(smu->adev->dev, "[%s] Failed to get activity monitor!", __func__);
 		return ret;
 	}

-	smu_cmn_assign_power_profile(smu);
+	idx = 0 * ARCTURUS_CUSTOM_PARAMS_COUNT;
+	if (input[idx]) {
+		/* Gfxclk */
+		activity_monitor.Gfx_FPS = input[idx + 1];
+		activity_monitor.Gfx_UseRlcBusy = input[idx + 2];
+		activity_monitor.Gfx_MinActiveFreqType = input[idx + 3];
+		activity_monitor.Gfx_MinActiveFreq = input[idx + 4];
+		activity_monitor.Gfx_BoosterFreqType = input[idx + 5];
+		activity_monitor.Gfx_BoosterFreq = input[idx + 6];
+		activity_monitor.Gfx_PD_Data_limit_c = input[idx + 7];
+		activity_monitor.Gfx_PD_Data_error_coeff = input[idx + 8];
+		activity_monitor.Gfx_PD_Data_error_rate_coeff = input[idx + 9];
+	}
+	idx = 1 * ARCTURUS_CUSTOM_PARAMS_COUNT;
+	if (input[idx]) {
+		/* Uclk */
+		activity_monitor.Mem_FPS = input[idx + 1];
+		activity_monitor.Mem_UseRlcBusy = input[idx + 2];
+		activity_monitor.Mem_MinActiveFreqType = input[idx + 3];
+		activity_monitor.Mem_MinActiveFreq = input[idx + 4];
+		activity_monitor.Mem_BoosterFreqType = input[idx + 5];
+		activity_monitor.Mem_BoosterFreq = input[idx + 6];
+		activity_monitor.Mem_PD_Data_limit_c = input[idx + 7];
+		activity_monitor.Mem_PD_Data_error_coeff = input[idx + 8];
+		activity_monitor.Mem_PD_Data_error_rate_coeff = input[idx + 9];
+	}

-	return 0;
+	ret = smu_cmn_update_table(smu,
+				   SMU_TABLE_ACTIVITY_MONITOR_COEFF,
+				   WORKLOAD_PPLIB_CUSTOM_BIT,
+				   (void *)(&activity_monitor),
+				   true);
+	if (ret) {
+		dev_err(smu->adev->dev, "[%s] Failed to set activity monitor!", __func__);
+		return ret;
+	}
+
+	return ret;
+}
+
+static int arcturus_set_power_profile_mode(struct smu_context *smu,
+					   u32 workload_mask,
+					   long *custom_params,
+					   u32 custom_params_max_idx)
+{
+	u32 backend_workload_mask = 0;
+	int ret, idx = -1, i;
+
+	smu_cmn_get_backend_workload_mask(smu, workload_mask,
+					  &backend_workload_mask);
+
+	if (workload_mask & (1 << PP_SMC_POWER_PROFILE_CUSTOM)) {
+		if (smu->smc_fw_version < 0x360d00)
+			return -EINVAL;
+		if (!smu->custom_profile_params) {
+			smu->custom_profile_params =
+				kzalloc(ARCTURUS_CUSTOM_PARAMS_SIZE, GFP_KERNEL);
+			if (!smu->custom_profile_params)
+				return -ENOMEM;
+		}
+		if (custom_params && custom_params_max_idx) {
+			if (custom_params_max_idx != ARCTURUS_CUSTOM_PARAMS_COUNT)
+				return -EINVAL;
+			if (custom_params[0] >= ARCTURUS_CUSTOM_PARAMS_CLOCK_COUNT)
+				return -EINVAL;
+			idx = custom_params[0] * ARCTURUS_CUSTOM_PARAMS_COUNT;
+			smu->custom_profile_params[idx] = 1;
+			for (i = 1; i < custom_params_max_idx; i++)
+				smu->custom_profile_params[idx + i] = custom_params[i];
+		}
+		ret = arcturus_set_power_profile_mode_coeff(smu,
+							    smu->custom_profile_params);
+		if (ret) {
+			if (idx != -1)
+				smu->custom_profile_params[idx] = 0;
+			return ret;
+		}
+	} else if (smu->custom_profile_params) {
+		memset(smu->custom_profile_params, 0, ARCTURUS_CUSTOM_PARAMS_SIZE);
+	}
+
+	ret = smu_cmn_send_smc_msg_with_param(smu,
+					      SMU_MSG_SetWorkloadMask,
+					      backend_workload_mask,
+					      NULL);
+	if (ret) {
+		dev_err(smu->adev->dev, "Failed to set workload mask 0x%08x\n",
+			workload_mask);
+		if (idx != -1)
+			smu->custom_profile_params[idx] = 0;
+		return ret;
+	}
+
+	return ret;
 }

 static int arcturus_set_performance_level(struct smu_context *smu,
--- a/drivers/gpu/drm/amd/pm/swsmu/smu11/navi10_ppt.c
+++ b/drivers/gpu/drm/amd/pm/swsmu/smu11/navi10_ppt.c
@ -2006,90 +2006,122 @@ static int navi10_get_power_profile_mode(struct smu_context *smu, char *buf)
 	return size;
 }

-static int navi10_set_power_profile_mode(struct smu_context *smu, long *input, uint32_t size)
+#define NAVI10_CUSTOM_PARAMS_COUNT 10
+#define NAVI10_CUSTOM_PARAMS_CLOCKS_COUNT 3
+#define NAVI10_CUSTOM_PARAMS_SIZE (NAVI10_CUSTOM_PARAMS_CLOCKS_COUNT * NAVI10_CUSTOM_PARAMS_COUNT * sizeof(long))
+
+static int navi10_set_power_profile_mode_coeff(struct smu_context *smu,
+					       long *input)
 {
 	DpmActivityMonitorCoeffInt_t activity_monitor;
-	int workload_type, ret = 0;
+	int ret, idx;

-	smu->power_profile_mode = input[size];
-
-	if (smu->power_profile_mode > PP_SMC_POWER_PROFILE_CUSTOM) {
-		dev_err(smu->adev->dev, "Invalid power profile mode %d\n", smu->power_profile_mode);
-		return -EINVAL;
+	ret = smu_cmn_update_table(smu,
+				   SMU_TABLE_ACTIVITY_MONITOR_COEFF, WORKLOAD_PPLIB_CUSTOM_BIT,
+				   (void *)(&activity_monitor), false);
+	if (ret) {
+		dev_err(smu->adev->dev, "[%s] Failed to get activity monitor!", __func__);
+		return ret;
 	}

-	if (smu->power_profile_mode == PP_SMC_POWER_PROFILE_CUSTOM) {
-		if (size != 10)
-			return -EINVAL;
-
-		ret = smu_cmn_update_table(smu,
-				       SMU_TABLE_ACTIVITY_MONITOR_COEFF, WORKLOAD_PPLIB_CUSTOM_BIT,
-				       (void *)(&activity_monitor), false);
-		if (ret) {
-			dev_err(smu->adev->dev, "[%s] Failed to get activity monitor!", __func__);
-			return ret;
-		}
-
-		switch (input[0]) {
-		case 0: /* Gfxclk */
-			activity_monitor.Gfx_FPS = input[1];
-			activity_monitor.Gfx_MinFreqStep = input[2];
-			activity_monitor.Gfx_MinActiveFreqType = input[3];
-			activity_monitor.Gfx_MinActiveFreq = input[4];
-			activity_monitor.Gfx_BoosterFreqType = input[5];
-			activity_monitor.Gfx_BoosterFreq = input[6];
-			activity_monitor.Gfx_PD_Data_limit_c = input[7];
-			activity_monitor.Gfx_PD_Data_error_coeff = input[8];
-			activity_monitor.Gfx_PD_Data_error_rate_coeff = input[9];
-			break;
-		case 1: /* Socclk */
-			activity_monitor.Soc_FPS = input[1];
-			activity_monitor.Soc_MinFreqStep = input[2];
-			activity_monitor.Soc_MinActiveFreqType = input[3];
-			activity_monitor.Soc_MinActiveFreq = input[4];
-			activity_monitor.Soc_BoosterFreqType = input[5];
-			activity_monitor.Soc_BoosterFreq = input[6];
-			activity_monitor.Soc_PD_Data_limit_c = input[7];
-			activity_monitor.Soc_PD_Data_error_coeff = input[8];
-			activity_monitor.Soc_PD_Data_error_rate_coeff = input[9];
-			break;
-		case 2: /* Memclk */
-			activity_monitor.Mem_FPS = input[1];
-			activity_monitor.Mem_MinFreqStep = input[2];
-			activity_monitor.Mem_MinActiveFreqType = input[3];
-			activity_monitor.Mem_MinActiveFreq = input[4];
-			activity_monitor.Mem_BoosterFreqType = input[5];
-			activity_monitor.Mem_BoosterFreq = input[6];
-			activity_monitor.Mem_PD_Data_limit_c = input[7];
-			activity_monitor.Mem_PD_Data_error_coeff = input[8];
-			activity_monitor.Mem_PD_Data_error_rate_coeff = input[9];
-			break;
-		default:
-			return -EINVAL;
-		}
-
-		ret = smu_cmn_update_table(smu,
-				       SMU_TABLE_ACTIVITY_MONITOR_COEFF, WORKLOAD_PPLIB_CUSTOM_BIT,
-				       (void *)(&activity_monitor), true);
-		if (ret) {
-			dev_err(smu->adev->dev, "[%s] Failed to set activity monitor!", __func__);
-			return ret;
-		}
+	idx = 0 * NAVI10_CUSTOM_PARAMS_COUNT;
+	if (input[idx]) {
+		/* Gfxclk */
+		activity_monitor.Gfx_FPS = input[idx + 1];
+		activity_monitor.Gfx_MinFreqStep = input[idx + 2];
+		activity_monitor.Gfx_MinActiveFreqType = input[idx + 3];
+		activity_monitor.Gfx_MinActiveFreq = input[idx + 4];
+		activity_monitor.Gfx_BoosterFreqType = input[idx + 5];
+		activity_monitor.Gfx_BoosterFreq = input[idx + 6];
+		activity_monitor.Gfx_PD_Data_limit_c = input[idx + 7];
+		activity_monitor.Gfx_PD_Data_error_coeff = input[idx + 8];
+		activity_monitor.Gfx_PD_Data_error_rate_coeff = input[idx + 9];
+	}
+	idx = 1 * NAVI10_CUSTOM_PARAMS_COUNT;
+	if (input[idx]) {
+		/* Socclk */
+		activity_monitor.Soc_FPS = input[idx + 1];
+		activity_monitor.Soc_MinFreqStep = input[idx + 2];
+		activity_monitor.Soc_MinActiveFreqType = input[idx + 3];
+		activity_monitor.Soc_MinActiveFreq = input[idx + 4];
+		activity_monitor.Soc_BoosterFreqType = input[idx + 5];
+		activity_monitor.Soc_BoosterFreq = input[idx + 6];
+		activity_monitor.Soc_PD_Data_limit_c = input[idx + 7];
+		activity_monitor.Soc_PD_Data_error_coeff = input[idx + 8];
+		activity_monitor.Soc_PD_Data_error_rate_coeff = input[idx + 9];
+	}
+	idx = 2 * NAVI10_CUSTOM_PARAMS_COUNT;
+	if (input[idx]) {
+		/* Memclk */
+		activity_monitor.Mem_FPS = input[idx + 1];
+		activity_monitor.Mem_MinFreqStep = input[idx + 2];
+		activity_monitor.Mem_MinActiveFreqType = input[idx + 3];
+		activity_monitor.Mem_MinActiveFreq = input[idx + 4];
+		activity_monitor.Mem_BoosterFreqType = input[idx + 5];
+		activity_monitor.Mem_BoosterFreq = input[idx + 6];
+		activity_monitor.Mem_PD_Data_limit_c = input[idx + 7];
+		activity_monitor.Mem_PD_Data_error_coeff = input[idx + 8];
+		activity_monitor.Mem_PD_Data_error_rate_coeff = input[idx + 9];
 	}

-	/* conv PP_SMC_POWER_PROFILE* to WORKLOAD_PPLIB_*_BIT */
-	workload_type = smu_cmn_to_asic_specific_index(smu,
-						       CMN2ASIC_MAPPING_WORKLOAD,
-						       smu->power_profile_mode);
-	if (workload_type < 0)
-		return -EINVAL;
+	ret = smu_cmn_update_table(smu,
+				   SMU_TABLE_ACTIVITY_MONITOR_COEFF, WORKLOAD_PPLIB_CUSTOM_BIT,
+				   (void *)(&activity_monitor), true);
+	if (ret) {
+		dev_err(smu->adev->dev, "[%s] Failed to set activity monitor!", __func__);
+		return ret;
+	}
+
+	return ret;
+}
+
+static int navi10_set_power_profile_mode(struct smu_context *smu,
+					 u32 workload_mask,
+					 long *custom_params,
+					 u32 custom_params_max_idx)
+{
+	u32 backend_workload_mask = 0;
+	int ret, idx = -1, i;
+
+	smu_cmn_get_backend_workload_mask(smu, workload_mask,
+					  &backend_workload_mask);
+
+	if (workload_mask & (1 << PP_SMC_POWER_PROFILE_CUSTOM)) {
+		if (!smu->custom_profile_params) {
+			smu->custom_profile_params = kzalloc(NAVI10_CUSTOM_PARAMS_SIZE, GFP_KERNEL);
+			if (!smu->custom_profile_params)
+				return -ENOMEM;
+		}
+		if (custom_params && custom_params_max_idx) {
+			if (custom_params_max_idx != NAVI10_CUSTOM_PARAMS_COUNT)
+				return -EINVAL;
+			if (custom_params[0] >= NAVI10_CUSTOM_PARAMS_CLOCKS_COUNT)
+				return -EINVAL;
+			idx = custom_params[0] * NAVI10_CUSTOM_PARAMS_COUNT;
+			smu->custom_profile_params[idx] = 1;
+			for (i = 1; i < custom_params_max_idx; i++)
+				smu->custom_profile_params[idx + i] = custom_params[i];
+		}
+		ret = navi10_set_power_profile_mode_coeff(smu,
+							  smu->custom_profile_params);
+		if (ret) {
+			if (idx != -1)
+				smu->custom_profile_params[idx] = 0;
+			return ret;
+		}
+	} else if (smu->custom_profile_params) {
+		memset(smu->custom_profile_params, 0, NAVI10_CUSTOM_PARAMS_SIZE);
+	}

 	ret = smu_cmn_send_smc_msg_with_param(smu, SMU_MSG_SetWorkloadMask,
-				    smu->workload_mask, NULL);
-	if (ret)
-		dev_err(smu->adev->dev, "[%s] Failed to set work load mask!", __func__);
-	else
-		smu_cmn_assign_power_profile(smu);
+					      backend_workload_mask, NULL);
+	if (ret) {
+		dev_err(smu->adev->dev, "Failed to set workload mask 0x%08x\n",
+			workload_mask);
+		if (idx != -1)
+			smu->custom_profile_params[idx] = 0;
+		return ret;
+	}

 	return ret;
 }
--- a/drivers/gpu/drm/amd/pm/swsmu/smu11/sienna_cichlid_ppt.c
+++ b/drivers/gpu/drm/amd/pm/swsmu/smu11/sienna_cichlid_ppt.c
@ -1708,93 +1708,126 @@ static int sienna_cichlid_get_power_profile_mode(struct smu_context *smu, char *
 	return size;
 }

-static int sienna_cichlid_set_power_profile_mode(struct smu_context *smu, long *input, uint32_t size)
+#define SIENNA_CICHLID_CUSTOM_PARAMS_COUNT 10
+#define SIENNA_CICHLID_CUSTOM_PARAMS_CLOCK_COUNT 3
+#define SIENNA_CICHLID_CUSTOM_PARAMS_SIZE (SIENNA_CICHLID_CUSTOM_PARAMS_CLOCK_COUNT * SIENNA_CICHLID_CUSTOM_PARAMS_COUNT * sizeof(long))
+
+static int sienna_cichlid_set_power_profile_mode_coeff(struct smu_context *smu,
+						       long *input)
 {

 	DpmActivityMonitorCoeffIntExternal_t activity_monitor_external;
 	DpmActivityMonitorCoeffInt_t *activity_monitor =
 		&(activity_monitor_external.DpmActivityMonitorCoeffInt);
-	int workload_type, ret = 0;
+	int ret, idx;

-	smu->power_profile_mode = input[size];
-
-	if (smu->power_profile_mode > PP_SMC_POWER_PROFILE_CUSTOM) {
-		dev_err(smu->adev->dev, "Invalid power profile mode %d\n", smu->power_profile_mode);
-		return -EINVAL;
+	ret = smu_cmn_update_table(smu,
+				   SMU_TABLE_ACTIVITY_MONITOR_COEFF, WORKLOAD_PPLIB_CUSTOM_BIT,
+				   (void *)(&activity_monitor_external), false);
+	if (ret) {
+		dev_err(smu->adev->dev, "[%s] Failed to get activity monitor!", __func__);
+		return ret;
 	}

-	if (smu->power_profile_mode == PP_SMC_POWER_PROFILE_CUSTOM) {
-		if (size != 10)
-			return -EINVAL;
-
-		ret = smu_cmn_update_table(smu,
-				       SMU_TABLE_ACTIVITY_MONITOR_COEFF, WORKLOAD_PPLIB_CUSTOM_BIT,
-				       (void *)(&activity_monitor_external), false);
-		if (ret) {
-			dev_err(smu->adev->dev, "[%s] Failed to get activity monitor!", __func__);
-			return ret;
-		}
-
-		switch (input[0]) {
-		case 0: /* Gfxclk */
-			activity_monitor->Gfx_FPS = input[1];
-			activity_monitor->Gfx_MinFreqStep = input[2];
-			activity_monitor->Gfx_MinActiveFreqType = input[3];
-			activity_monitor->Gfx_MinActiveFreq = input[4];
-			activity_monitor->Gfx_BoosterFreqType = input[5];
-			activity_monitor->Gfx_BoosterFreq = input[6];
-			activity_monitor->Gfx_PD_Data_limit_c = input[7];
-			activity_monitor->Gfx_PD_Data_error_coeff = input[8];
-			activity_monitor->Gfx_PD_Data_error_rate_coeff = input[9];
-			break;
-		case 1: /* Socclk */
-			activity_monitor->Fclk_FPS = input[1];
-			activity_monitor->Fclk_MinFreqStep = input[2];
-			activity_monitor->Fclk_MinActiveFreqType = input[3];
-			activity_monitor->Fclk_MinActiveFreq = input[4];
-			activity_monitor->Fclk_BoosterFreqType = input[5];
-			activity_monitor->Fclk_BoosterFreq = input[6];
-			activity_monitor->Fclk_PD_Data_limit_c = input[7];
-			activity_monitor->Fclk_PD_Data_error_coeff = input[8];
-			activity_monitor->Fclk_PD_Data_error_rate_coeff = input[9];
-			break;
-		case 2: /* Memclk */
-			activity_monitor->Mem_FPS = input[1];
-			activity_monitor->Mem_MinFreqStep = input[2];
-			activity_monitor->Mem_MinActiveFreqType = input[3];
-			activity_monitor->Mem_MinActiveFreq = input[4];
-			activity_monitor->Mem_BoosterFreqType = input[5];
-			activity_monitor->Mem_BoosterFreq = input[6];
-			activity_monitor->Mem_PD_Data_limit_c = input[7];
-			activity_monitor->Mem_PD_Data_error_coeff = input[8];
-			activity_monitor->Mem_PD_Data_error_rate_coeff = input[9];
-			break;
-		default:
-			return -EINVAL;
-		}
-
-		ret = smu_cmn_update_table(smu,
-				       SMU_TABLE_ACTIVITY_MONITOR_COEFF, WORKLOAD_PPLIB_CUSTOM_BIT,
-				       (void *)(&activity_monitor_external), true);
-		if (ret) {
-			dev_err(smu->adev->dev, "[%s] Failed to set activity monitor!", __func__);
-			return ret;
-		}
+	idx = 0 * SIENNA_CICHLID_CUSTOM_PARAMS_COUNT;
+	if (input[idx]) {
+		/* Gfxclk */
+		activity_monitor->Gfx_FPS = input[idx + 1];
+		activity_monitor->Gfx_MinFreqStep = input[idx + 2];
+		activity_monitor->Gfx_MinActiveFreqType = input[idx + 3];
+		activity_monitor->Gfx_MinActiveFreq = input[idx + 4];
+		activity_monitor->Gfx_BoosterFreqType = input[idx + 5];
+		activity_monitor->Gfx_BoosterFreq = input[idx + 6];
+		activity_monitor->Gfx_PD_Data_limit_c = input[idx + 7];
+		activity_monitor->Gfx_PD_Data_error_coeff = input[idx + 8];
+		activity_monitor->Gfx_PD_Data_error_rate_coeff = input[idx + 9];
+	}
+	idx = 1 * SIENNA_CICHLID_CUSTOM_PARAMS_COUNT;
+	if (input[idx]) {
+		/* Socclk */
+		activity_monitor->Fclk_FPS = input[idx + 1];
+		activity_monitor->Fclk_MinFreqStep = input[idx + 2];
+		activity_monitor->Fclk_MinActiveFreqType = input[idx + 3];
+		activity_monitor->Fclk_MinActiveFreq = input[idx + 4];
+		activity_monitor->Fclk_BoosterFreqType = input[idx + 5];
+		activity_monitor->Fclk_BoosterFreq = input[idx + 6];
+		activity_monitor->Fclk_PD_Data_limit_c = input[idx + 7];
+		activity_monitor->Fclk_PD_Data_error_coeff = input[idx + 8];
+		activity_monitor->Fclk_PD_Data_error_rate_coeff = input[idx + 9];
+	}
+	idx = 2 * SIENNA_CICHLID_CUSTOM_PARAMS_COUNT;
+	if (input[idx]) {
+		/* Memclk */
+		activity_monitor->Mem_FPS = input[idx + 1];
+		activity_monitor->Mem_MinFreqStep = input[idx + 2];
+		activity_monitor->Mem_MinActiveFreqType = input[idx + 3];
+		activity_monitor->Mem_MinActiveFreq = input[idx + 4];
+		activity_monitor->Mem_BoosterFreqType = input[idx + 5];
+		activity_monitor->Mem_BoosterFreq = input[idx + 6];
+		activity_monitor->Mem_PD_Data_limit_c = input[idx + 7];
+		activity_monitor->Mem_PD_Data_error_coeff = input[idx + 8];
+		activity_monitor->Mem_PD_Data_error_rate_coeff = input[idx + 9];
 	}

-	/* conv PP_SMC_POWER_PROFILE* to WORKLOAD_PPLIB_*_BIT */
-	workload_type = smu_cmn_to_asic_specific_index(smu,
-						       CMN2ASIC_MAPPING_WORKLOAD,
-						       smu->power_profile_mode);
-	if (workload_type < 0)
-		return -EINVAL;
+	ret = smu_cmn_update_table(smu,
+				   SMU_TABLE_ACTIVITY_MONITOR_COEFF, WORKLOAD_PPLIB_CUSTOM_BIT,
+				   (void *)(&activity_monitor_external), true);
+	if (ret) {
+		dev_err(smu->adev->dev, "[%s] Failed to set activity monitor!", __func__);
+		return ret;
+	}
+
+	return ret;
+}
+
+static int sienna_cichlid_set_power_profile_mode(struct smu_context *smu,
+						 u32 workload_mask,
+						 long *custom_params,
+						 u32 custom_params_max_idx)
+{
+	u32 backend_workload_mask = 0;
+	int ret, idx = -1, i;
+
+	smu_cmn_get_backend_workload_mask(smu, workload_mask,
+					  &backend_workload_mask);
+
+	if (workload_mask & (1 << PP_SMC_POWER_PROFILE_CUSTOM)) {
+		if (!smu->custom_profile_params) {
+			smu->custom_profile_params =
+				kzalloc(SIENNA_CICHLID_CUSTOM_PARAMS_SIZE, GFP_KERNEL);
+			if (!smu->custom_profile_params)
+				return -ENOMEM;
+		}
+		if (custom_params && custom_params_max_idx) {
+			if (custom_params_max_idx != SIENNA_CICHLID_CUSTOM_PARAMS_COUNT)
+				return -EINVAL;
+			if (custom_params[0] >= SIENNA_CICHLID_CUSTOM_PARAMS_CLOCK_COUNT)
+				return -EINVAL;
+			idx = custom_params[0] * SIENNA_CICHLID_CUSTOM_PARAMS_COUNT;
+			smu->custom_profile_params[idx] = 1;
+			for (i = 1; i < custom_params_max_idx; i++)
+				smu->custom_profile_params[idx + i] = custom_params[i];
+		}
+		ret = sienna_cichlid_set_power_profile_mode_coeff(smu,
+								  smu->custom_profile_params);
+		if (ret) {
+			if (idx != -1)
+				smu->custom_profile_params[idx] = 0;
+			return ret;
+		}
+	} else if (smu->custom_profile_params) {
+		memset(smu->custom_profile_params, 0, SIENNA_CICHLID_CUSTOM_PARAMS_SIZE);
+	}

 	ret = smu_cmn_send_smc_msg_with_param(smu, SMU_MSG_SetWorkloadMask,
-				    smu->workload_mask, NULL);
-	if (ret)
-		dev_err(smu->adev->dev, "[%s] Failed to set work load mask!", __func__);
-	else
-		smu_cmn_assign_power_profile(smu);
+					      backend_workload_mask, NULL);
+	if (ret) {
+		dev_err(smu->adev->dev, "Failed to set workload mask 0x%08x\n",
+			workload_mask);
+		if (idx != -1)
+			smu->custom_profile_params[idx] = 0;
+		return ret;
+	}

 	return ret;
 }
--- a/drivers/gpu/drm/amd/pm/swsmu/smu11/vangogh_ppt.c
+++ b/drivers/gpu/drm/amd/pm/swsmu/smu11/vangogh_ppt.c
@ -1056,42 +1056,27 @@ static int vangogh_get_power_profile_mode(struct smu_context *smu,
 	return size;
 }

-static int vangogh_set_power_profile_mode(struct smu_context *smu, long *input, uint32_t size)
+static int vangogh_set_power_profile_mode(struct smu_context *smu,
+					  u32 workload_mask,
+					  long *custom_params,
+					  u32 custom_params_max_idx)
 {
-	int workload_type, ret;
-	uint32_t profile_mode = input[size];
+	u32 backend_workload_mask = 0;
+	int ret;

-	if (profile_mode >= PP_SMC_POWER_PROFILE_COUNT) {
-		dev_err(smu->adev->dev, "Invalid power profile mode %d\n", profile_mode);
-		return -EINVAL;
-	}
-
-	if (profile_mode == PP_SMC_POWER_PROFILE_BOOTUP_DEFAULT ||
-			profile_mode == PP_SMC_POWER_PROFILE_POWERSAVING)
-		return 0;
-
-	/* conv PP_SMC_POWER_PROFILE* to WORKLOAD_PPLIB_*_BIT */
-	workload_type = smu_cmn_to_asic_specific_index(smu,
-						       CMN2ASIC_MAPPING_WORKLOAD,
-						       profile_mode);
-	if (workload_type < 0) {
-		dev_dbg(smu->adev->dev, "Unsupported power profile mode %d on VANGOGH\n",
-					profile_mode);
-		return -EINVAL;
-	}
+	smu_cmn_get_backend_workload_mask(smu, workload_mask,
+					  &backend_workload_mask);

 	ret = smu_cmn_send_smc_msg_with_param(smu, SMU_MSG_ActiveProcessNotify,
-				    smu->workload_mask,
-				    NULL);
+					      backend_workload_mask,
+					      NULL);
 	if (ret) {
-		dev_err_once(smu->adev->dev, "Fail to set workload type %d\n",
-					workload_type);
+		dev_err_once(smu->adev->dev, "Fail to set workload mask 0x%08x\n",
+			     workload_mask);
 		return ret;
 	}

-	smu_cmn_assign_power_profile(smu);
-
-	return 0;
+	return ret;
 }

 static int vangogh_set_soft_freq_limited_range(struct smu_context *smu,
--- a/drivers/gpu/drm/amd/pm/swsmu/smu12/renoir_ppt.c
+++ b/drivers/gpu/drm/amd/pm/swsmu/smu12/renoir_ppt.c
@ -864,44 +864,27 @@ static int renoir_force_clk_levels(struct smu_context *smu,
 	return ret;
 }

-static int renoir_set_power_profile_mode(struct smu_context *smu, long *input, uint32_t size)
+static int renoir_set_power_profile_mode(struct smu_context *smu,
+					 u32 workload_mask,
+					 long *custom_params,
+					 u32 custom_params_max_idx)
 {
-	int workload_type, ret;
-	uint32_t profile_mode = input[size];
+	int ret;
+	u32 backend_workload_mask = 0;

-	if (profile_mode > PP_SMC_POWER_PROFILE_CUSTOM) {
-		dev_err(smu->adev->dev, "Invalid power profile mode %d\n", profile_mode);
-		return -EINVAL;
-	}
-
-	if (profile_mode == PP_SMC_POWER_PROFILE_BOOTUP_DEFAULT ||
-			profile_mode == PP_SMC_POWER_PROFILE_POWERSAVING)
-		return 0;
-
-	/* conv PP_SMC_POWER_PROFILE* to WORKLOAD_PPLIB_*_BIT */
-	workload_type = smu_cmn_to_asic_specific_index(smu,
-						       CMN2ASIC_MAPPING_WORKLOAD,
-						       profile_mode);
-	if (workload_type < 0) {
-		/*
-		 * TODO: If some case need switch to powersave/default power mode
-		 * then can consider enter WORKLOAD_COMPUTE/WORKLOAD_CUSTOM for power saving.
-		 */
-		dev_dbg(smu->adev->dev, "Unsupported power profile mode %d on RENOIR\n", profile_mode);
-		return -EINVAL;
-	}
+	smu_cmn_get_backend_workload_mask(smu, workload_mask,
+					  &backend_workload_mask);

 	ret = smu_cmn_send_smc_msg_with_param(smu, SMU_MSG_ActiveProcessNotify,
-				    smu->workload_mask,
-				    NULL);
+					      backend_workload_mask,
+					      NULL);
 	if (ret) {
-		dev_err_once(smu->adev->dev, "Fail to set workload type %d\n", workload_type);
+		dev_err_once(smu->adev->dev, "Failed to set workload mask 0x08%x\n",
+			     workload_mask);
 		return ret;
 	}

-	smu_cmn_assign_power_profile(smu);
-
-	return 0;
+	return ret;
 }

 static int renoir_set_peak_clock_by_device(struct smu_context *smu)
--- a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_0_ppt.c
+++ b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_0_ppt.c
@ -2571,82 +2571,76 @@ static int smu_v13_0_0_get_power_profile_mode(struct smu_context *smu,
 	return size;
 }

-static int smu_v13_0_0_set_power_profile_mode(struct smu_context *smu,
-					      long *input,
-					      uint32_t size)
+#define SMU_13_0_0_CUSTOM_PARAMS_COUNT 9
+#define SMU_13_0_0_CUSTOM_PARAMS_CLOCK_COUNT 2
+#define SMU_13_0_0_CUSTOM_PARAMS_SIZE (SMU_13_0_0_CUSTOM_PARAMS_CLOCK_COUNT * SMU_13_0_0_CUSTOM_PARAMS_COUNT * sizeof(long))
+
+static int smu_v13_0_0_set_power_profile_mode_coeff(struct smu_context *smu,
+						    long *input)
 {
 	DpmActivityMonitorCoeffIntExternal_t activity_monitor_external;
 	DpmActivityMonitorCoeffInt_t *activity_monitor =
 		&(activity_monitor_external.DpmActivityMonitorCoeffInt);
-	int workload_type, ret = 0;
-	u32 workload_mask;
+	int ret, idx;

-	smu->power_profile_mode = input[size];
-
-	if (smu->power_profile_mode >= PP_SMC_POWER_PROFILE_COUNT) {
-		dev_err(smu->adev->dev, "Invalid power profile mode %d\n", smu->power_profile_mode);
-		return -EINVAL;
+	ret = smu_cmn_update_table(smu,
+				   SMU_TABLE_ACTIVITY_MONITOR_COEFF,
+				   WORKLOAD_PPLIB_CUSTOM_BIT,
+				   (void *)(&activity_monitor_external),
+				   false);
+	if (ret) {
+		dev_err(smu->adev->dev, "[%s] Failed to get activity monitor!", __func__);
+		return ret;
 	}

-	if (smu->power_profile_mode == PP_SMC_POWER_PROFILE_CUSTOM) {
-		if (size != 9)
-			return -EINVAL;
-
-		ret = smu_cmn_update_table(smu,
-					   SMU_TABLE_ACTIVITY_MONITOR_COEFF,
-					   WORKLOAD_PPLIB_CUSTOM_BIT,
-					   (void *)(&activity_monitor_external),
-					   false);
-		if (ret) {
-			dev_err(smu->adev->dev, "[%s] Failed to get activity monitor!", __func__);
-			return ret;
-		}
-
-		switch (input[0]) {
-		case 0: /* Gfxclk */
-			activity_monitor->Gfx_FPS = input[1];
-			activity_monitor->Gfx_MinActiveFreqType = input[2];
-			activity_monitor->Gfx_MinActiveFreq = input[3];
-			activity_monitor->Gfx_BoosterFreqType = input[4];
-			activity_monitor->Gfx_BoosterFreq = input[5];
-			activity_monitor->Gfx_PD_Data_limit_c = input[6];
-			activity_monitor->Gfx_PD_Data_error_coeff = input[7];
-			activity_monitor->Gfx_PD_Data_error_rate_coeff = input[8];
-			break;
-		case 1: /* Fclk */
-			activity_monitor->Fclk_FPS = input[1];
-			activity_monitor->Fclk_MinActiveFreqType = input[2];
-			activity_monitor->Fclk_MinActiveFreq = input[3];
-			activity_monitor->Fclk_BoosterFreqType = input[4];
-			activity_monitor->Fclk_BoosterFreq = input[5];
-			activity_monitor->Fclk_PD_Data_limit_c = input[6];
-			activity_monitor->Fclk_PD_Data_error_coeff = input[7];
-			activity_monitor->Fclk_PD_Data_error_rate_coeff = input[8];
-			break;
-		default:
-			return -EINVAL;
-		}
-
-		ret = smu_cmn_update_table(smu,
-					   SMU_TABLE_ACTIVITY_MONITOR_COEFF,
-					   WORKLOAD_PPLIB_CUSTOM_BIT,
-					   (void *)(&activity_monitor_external),
-					   true);
-		if (ret) {
-			dev_err(smu->adev->dev, "[%s] Failed to set activity monitor!", __func__);
-			return ret;
-		}
+	idx = 0 * SMU_13_0_0_CUSTOM_PARAMS_COUNT;
+	if (input[idx]) {
+		/* Gfxclk */
+		activity_monitor->Gfx_FPS = input[idx + 1];
+		activity_monitor->Gfx_MinActiveFreqType = input[idx + 2];
+		activity_monitor->Gfx_MinActiveFreq = input[idx + 3];
+		activity_monitor->Gfx_BoosterFreqType = input[idx + 4];
+		activity_monitor->Gfx_BoosterFreq = input[idx + 5];
+		activity_monitor->Gfx_PD_Data_limit_c = input[idx + 6];
+		activity_monitor->Gfx_PD_Data_error_coeff = input[idx + 7];
+		activity_monitor->Gfx_PD_Data_error_rate_coeff = input[idx + 8];
+	}
+	idx = 1 * SMU_13_0_0_CUSTOM_PARAMS_COUNT;
+	if (input[idx]) {
+		/* Fclk */
+		activity_monitor->Fclk_FPS = input[idx + 1];
+		activity_monitor->Fclk_MinActiveFreqType = input[idx + 2];
+		activity_monitor->Fclk_MinActiveFreq = input[idx + 3];
+		activity_monitor->Fclk_BoosterFreqType = input[idx + 4];
+		activity_monitor->Fclk_BoosterFreq = input[idx + 5];
+		activity_monitor->Fclk_PD_Data_limit_c = input[idx + 6];
+		activity_monitor->Fclk_PD_Data_error_coeff = input[idx + 7];
+		activity_monitor->Fclk_PD_Data_error_rate_coeff = input[idx + 8];
 	}

-	/* conv PP_SMC_POWER_PROFILE* to WORKLOAD_PPLIB_*_BIT */
-	workload_type = smu_cmn_to_asic_specific_index(smu,
-						       CMN2ASIC_MAPPING_WORKLOAD,
-						       smu->power_profile_mode);
+	ret = smu_cmn_update_table(smu,
+				   SMU_TABLE_ACTIVITY_MONITOR_COEFF,
+				   WORKLOAD_PPLIB_CUSTOM_BIT,
+				   (void *)(&activity_monitor_external),
+				   true);
+	if (ret) {
+		dev_err(smu->adev->dev, "[%s] Failed to set activity monitor!", __func__);
+		return ret;
+	}

-	if (workload_type < 0)
-		return -EINVAL;
+	return ret;
+}

-	workload_mask = 1 << workload_type;
+static int smu_v13_0_0_set_power_profile_mode(struct smu_context *smu,
+					      u32 workload_mask,
+					      long *custom_params,
+					      u32 custom_params_max_idx)
+{
+	u32 backend_workload_mask = 0;
+	int workload_type, ret, idx = -1, i;
+
+	smu_cmn_get_backend_workload_mask(smu, workload_mask,
+					  &backend_workload_mask);

 	/* Add optimizations for SMU13.0.0/10.  Reuse the power saving profile */
 	if ((amdgpu_ip_version(smu->adev, MP1_HWIP, 0) == IP_VERSION(13, 0, 0) &&
@ -2658,24 +2652,47 @@ static int smu_v13_0_0_set_power_profile_mode(struct smu_context *smu,
 							       CMN2ASIC_MAPPING_WORKLOAD,
 							       PP_SMC_POWER_PROFILE_POWERSAVING);
 		if (workload_type >= 0)
-			workload_mask |= 1 << workload_type;
+			backend_workload_mask |= 1 << workload_type;
 	}

-	smu->workload_mask |= workload_mask;
-	ret = smu_cmn_send_smc_msg_with_param(smu,
-					       SMU_MSG_SetWorkloadMask,
-					       smu->workload_mask,
-					       NULL);
-	if (!ret) {
-		smu_cmn_assign_power_profile(smu);
-		if (smu->power_profile_mode == PP_SMC_POWER_PROFILE_POWERSAVING) {
-			workload_type = smu_cmn_to_asic_specific_index(smu,
-							       CMN2ASIC_MAPPING_WORKLOAD,
-							       PP_SMC_POWER_PROFILE_FULLSCREEN3D);
-			smu->power_profile_mode = smu->workload_mask & (1 << workload_type)
-										? PP_SMC_POWER_PROFILE_FULLSCREEN3D
-										: PP_SMC_POWER_PROFILE_BOOTUP_DEFAULT;
+	if (workload_mask & (1 << PP_SMC_POWER_PROFILE_CUSTOM)) {
+		if (!smu->custom_profile_params) {
+			smu->custom_profile_params =
+				kzalloc(SMU_13_0_0_CUSTOM_PARAMS_SIZE, GFP_KERNEL);
+			if (!smu->custom_profile_params)
+				return -ENOMEM;
 		}
+		if (custom_params && custom_params_max_idx) {
+			if (custom_params_max_idx != SMU_13_0_0_CUSTOM_PARAMS_COUNT)
+				return -EINVAL;
+			if (custom_params[0] >= SMU_13_0_0_CUSTOM_PARAMS_CLOCK_COUNT)
+				return -EINVAL;
+			idx = custom_params[0] * SMU_13_0_0_CUSTOM_PARAMS_COUNT;
+			smu->custom_profile_params[idx] = 1;
+			for (i = 1; i < custom_params_max_idx; i++)
+				smu->custom_profile_params[idx + i] = custom_params[i];
+		}
+		ret = smu_v13_0_0_set_power_profile_mode_coeff(smu,
+							       smu->custom_profile_params);
+		if (ret) {
+			if (idx != -1)
+				smu->custom_profile_params[idx] = 0;
+			return ret;
+		}
+	} else if (smu->custom_profile_params) {
+		memset(smu->custom_profile_params, 0, SMU_13_0_0_CUSTOM_PARAMS_SIZE);
+	}
+
+	ret = smu_cmn_send_smc_msg_with_param(smu,
+					      SMU_MSG_SetWorkloadMask,
+					      backend_workload_mask,
+					      NULL);
+	if (ret) {
+		dev_err(smu->adev->dev, "Failed to set workload mask 0x%08x\n",
+			workload_mask);
+		if (idx != -1)
+			smu->custom_profile_params[idx] = 0;
+		return ret;
 	}

 	return ret;
--- a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_7_ppt.c
+++ b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_7_ppt.c
@ -2530,79 +2530,110 @@ do {													\
 	return result;
 }

-static int smu_v13_0_7_set_power_profile_mode(struct smu_context *smu, long *input, uint32_t size)
+#define SMU_13_0_7_CUSTOM_PARAMS_COUNT 8
+#define SMU_13_0_7_CUSTOM_PARAMS_CLOCK_COUNT 2
+#define SMU_13_0_7_CUSTOM_PARAMS_SIZE (SMU_13_0_7_CUSTOM_PARAMS_CLOCK_COUNT * SMU_13_0_7_CUSTOM_PARAMS_COUNT * sizeof(long))
+
+static int smu_v13_0_7_set_power_profile_mode_coeff(struct smu_context *smu,
+						    long *input)
 {

 	DpmActivityMonitorCoeffIntExternal_t activity_monitor_external;
 	DpmActivityMonitorCoeffInt_t *activity_monitor =
 		&(activity_monitor_external.DpmActivityMonitorCoeffInt);
-	int workload_type, ret = 0;
+	int ret, idx;

-	smu->power_profile_mode = input[size];
-
-	if (smu->power_profile_mode > PP_SMC_POWER_PROFILE_WINDOW3D) {
-		dev_err(smu->adev->dev, "Invalid power profile mode %d\n", smu->power_profile_mode);
-		return -EINVAL;
+	ret = smu_cmn_update_table(smu,
+				   SMU_TABLE_ACTIVITY_MONITOR_COEFF, WORKLOAD_PPLIB_CUSTOM_BIT,
+				   (void *)(&activity_monitor_external), false);
+	if (ret) {
+		dev_err(smu->adev->dev, "[%s] Failed to get activity monitor!", __func__);
+		return ret;
 	}

-	if (smu->power_profile_mode == PP_SMC_POWER_PROFILE_CUSTOM) {
-		if (size != 8)
-			return -EINVAL;
-
-		ret = smu_cmn_update_table(smu,
-				       SMU_TABLE_ACTIVITY_MONITOR_COEFF, WORKLOAD_PPLIB_CUSTOM_BIT,
-				       (void *)(&activity_monitor_external), false);
-		if (ret) {
-			dev_err(smu->adev->dev, "[%s] Failed to get activity monitor!", __func__);
-			return ret;
-		}
-
-		switch (input[0]) {
-		case 0: /* Gfxclk */
-			activity_monitor->Gfx_ActiveHystLimit = input[1];
-			activity_monitor->Gfx_IdleHystLimit = input[2];
-			activity_monitor->Gfx_FPS = input[3];
-			activity_monitor->Gfx_MinActiveFreqType = input[4];
-			activity_monitor->Gfx_BoosterFreqType = input[5];
-			activity_monitor->Gfx_MinActiveFreq = input[6];
-			activity_monitor->Gfx_BoosterFreq = input[7];
-			break;
-		case 1: /* Fclk */
-			activity_monitor->Fclk_ActiveHystLimit = input[1];
-			activity_monitor->Fclk_IdleHystLimit = input[2];
-			activity_monitor->Fclk_FPS = input[3];
-			activity_monitor->Fclk_MinActiveFreqType = input[4];
-			activity_monitor->Fclk_BoosterFreqType = input[5];
-			activity_monitor->Fclk_MinActiveFreq = input[6];
-			activity_monitor->Fclk_BoosterFreq = input[7];
-			break;
-		default:
-			return -EINVAL;
-		}
-
-		ret = smu_cmn_update_table(smu,
-				       SMU_TABLE_ACTIVITY_MONITOR_COEFF, WORKLOAD_PPLIB_CUSTOM_BIT,
-				       (void *)(&activity_monitor_external), true);
-		if (ret) {
-			dev_err(smu->adev->dev, "[%s] Failed to set activity monitor!", __func__);
-			return ret;
-		}
+	idx = 0 * SMU_13_0_7_CUSTOM_PARAMS_COUNT;
+	if (input[idx]) {
+		/* Gfxclk */
+		activity_monitor->Gfx_ActiveHystLimit = input[idx + 1];
+		activity_monitor->Gfx_IdleHystLimit = input[idx + 2];
+		activity_monitor->Gfx_FPS = input[idx + 3];
+		activity_monitor->Gfx_MinActiveFreqType = input[idx + 4];
+		activity_monitor->Gfx_BoosterFreqType = input[idx + 5];
+		activity_monitor->Gfx_MinActiveFreq = input[idx + 6];
+		activity_monitor->Gfx_BoosterFreq = input[idx + 7];
+	}
+	idx = 1 * SMU_13_0_7_CUSTOM_PARAMS_COUNT;
+	if (input[idx]) {
+		/* Fclk */
+		activity_monitor->Fclk_ActiveHystLimit = input[idx + 1];
+		activity_monitor->Fclk_IdleHystLimit = input[idx + 2];
+		activity_monitor->Fclk_FPS = input[idx + 3];
+		activity_monitor->Fclk_MinActiveFreqType = input[idx + 4];
+		activity_monitor->Fclk_BoosterFreqType = input[idx + 5];
+		activity_monitor->Fclk_MinActiveFreq = input[idx + 6];
+		activity_monitor->Fclk_BoosterFreq = input[idx + 7];
 	}

-	/* conv PP_SMC_POWER_PROFILE* to WORKLOAD_PPLIB_*_BIT */
-	workload_type = smu_cmn_to_asic_specific_index(smu,
-						       CMN2ASIC_MAPPING_WORKLOAD,
-						       smu->power_profile_mode);
-	if (workload_type < 0)
-		return -EINVAL;
+	ret = smu_cmn_update_table(smu,
+				   SMU_TABLE_ACTIVITY_MONITOR_COEFF, WORKLOAD_PPLIB_CUSTOM_BIT,
+				   (void *)(&activity_monitor_external), true);
+	if (ret) {
+		dev_err(smu->adev->dev, "[%s] Failed to set activity monitor!", __func__);
+		return ret;
+	}
+
+	return ret;
+}
+
+static int smu_v13_0_7_set_power_profile_mode(struct smu_context *smu,
+					      u32 workload_mask,
+					      long *custom_params,
+					      u32 custom_params_max_idx)
+{
+	u32 backend_workload_mask = 0;
+	int ret, idx = -1, i;
+
+	smu_cmn_get_backend_workload_mask(smu, workload_mask,
+					  &backend_workload_mask);
+
+	if (workload_mask & (1 << PP_SMC_POWER_PROFILE_CUSTOM)) {
+		if (!smu->custom_profile_params) {
+			smu->custom_profile_params =
+				kzalloc(SMU_13_0_7_CUSTOM_PARAMS_SIZE, GFP_KERNEL);
+			if (!smu->custom_profile_params)
+				return -ENOMEM;
+		}
+		if (custom_params && custom_params_max_idx) {
+			if (custom_params_max_idx != SMU_13_0_7_CUSTOM_PARAMS_COUNT)
+				return -EINVAL;
+			if (custom_params[0] >= SMU_13_0_7_CUSTOM_PARAMS_CLOCK_COUNT)
+				return -EINVAL;
+			idx = custom_params[0] * SMU_13_0_7_CUSTOM_PARAMS_COUNT;
+			smu->custom_profile_params[idx] = 1;
+			for (i = 1; i < custom_params_max_idx; i++)
+				smu->custom_profile_params[idx + i] = custom_params[i];
+		}
+		ret = smu_v13_0_7_set_power_profile_mode_coeff(smu,
+							       smu->custom_profile_params);
+		if (ret) {
+			if (idx != -1)
+				smu->custom_profile_params[idx] = 0;
+			return ret;
+		}
+	} else if (smu->custom_profile_params) {
+		memset(smu->custom_profile_params, 0, SMU_13_0_7_CUSTOM_PARAMS_SIZE);
+	}

 	ret = smu_cmn_send_smc_msg_with_param(smu, SMU_MSG_SetWorkloadMask,
-				    smu->workload_mask, NULL);
+					      backend_workload_mask, NULL);

-	if (ret)
-		dev_err(smu->adev->dev, "[%s] Failed to set work load mask!", __func__);
-	else
-		smu_cmn_assign_power_profile(smu);
+	if (ret) {
+		dev_err(smu->adev->dev, "Failed to set workload mask 0x%08x\n",
+			workload_mask);
+		if (idx != -1)
+			smu->custom_profile_params[idx] = 0;
+		return ret;
+	}

 	return ret;
 }
--- a/drivers/gpu/drm/amd/pm/swsmu/smu14/smu_v14_0_2_ppt.c
+++ b/drivers/gpu/drm/amd/pm/swsmu/smu14/smu_v14_0_2_ppt.c
@ -1739,89 +1739,120 @@ static int smu_v14_0_2_get_power_profile_mode(struct smu_context *smu,
 	return size;
 }

-static int smu_v14_0_2_set_power_profile_mode(struct smu_context *smu,
-					      long *input,
-					      uint32_t size)
+#define SMU_14_0_2_CUSTOM_PARAMS_COUNT 9
+#define SMU_14_0_2_CUSTOM_PARAMS_CLOCK_COUNT 2
+#define SMU_14_0_2_CUSTOM_PARAMS_SIZE (SMU_14_0_2_CUSTOM_PARAMS_CLOCK_COUNT * SMU_14_0_2_CUSTOM_PARAMS_COUNT * sizeof(long))
+
+static int smu_v14_0_2_set_power_profile_mode_coeff(struct smu_context *smu,
+						    long *input)
 {
 	DpmActivityMonitorCoeffIntExternal_t activity_monitor_external;
 	DpmActivityMonitorCoeffInt_t *activity_monitor =
 		&(activity_monitor_external.DpmActivityMonitorCoeffInt);
-	int workload_type, ret = 0;
-	uint32_t current_profile_mode = smu->power_profile_mode;
-	smu->power_profile_mode = input[size];
+	int ret, idx;

-	if (smu->power_profile_mode >= PP_SMC_POWER_PROFILE_COUNT) {
-		dev_err(smu->adev->dev, "Invalid power profile mode %d\n", smu->power_profile_mode);
-		return -EINVAL;
+	ret = smu_cmn_update_table(smu,
+				   SMU_TABLE_ACTIVITY_MONITOR_COEFF,
+				   WORKLOAD_PPLIB_CUSTOM_BIT,
+				   (void *)(&activity_monitor_external),
+				   false);
+	if (ret) {
+		dev_err(smu->adev->dev, "[%s] Failed to get activity monitor!", __func__);
+		return ret;
 	}

-	if (smu->power_profile_mode == PP_SMC_POWER_PROFILE_CUSTOM) {
-		if (size != 9)
-			return -EINVAL;
-
-		ret = smu_cmn_update_table(smu,
-					   SMU_TABLE_ACTIVITY_MONITOR_COEFF,
-					   WORKLOAD_PPLIB_CUSTOM_BIT,
-					   (void *)(&activity_monitor_external),
-					   false);
-		if (ret) {
-			dev_err(smu->adev->dev, "[%s] Failed to get activity monitor!", __func__);
-			return ret;
-		}
-
-		switch (input[0]) {
-		case 0: /* Gfxclk */
-			activity_monitor->Gfx_FPS = input[1];
-			activity_monitor->Gfx_MinActiveFreqType = input[2];
-			activity_monitor->Gfx_MinActiveFreq = input[3];
-			activity_monitor->Gfx_BoosterFreqType = input[4];
-			activity_monitor->Gfx_BoosterFreq = input[5];
-			activity_monitor->Gfx_PD_Data_limit_c = input[6];
-			activity_monitor->Gfx_PD_Data_error_coeff = input[7];
-			activity_monitor->Gfx_PD_Data_error_rate_coeff = input[8];
-			break;
-		case 1: /* Fclk */
-			activity_monitor->Fclk_FPS = input[1];
-			activity_monitor->Fclk_MinActiveFreqType = input[2];
-			activity_monitor->Fclk_MinActiveFreq = input[3];
-			activity_monitor->Fclk_BoosterFreqType = input[4];
-			activity_monitor->Fclk_BoosterFreq = input[5];
-			activity_monitor->Fclk_PD_Data_limit_c = input[6];
-			activity_monitor->Fclk_PD_Data_error_coeff = input[7];
-			activity_monitor->Fclk_PD_Data_error_rate_coeff = input[8];
-			break;
-		default:
-			return -EINVAL;
-		}
-
-		ret = smu_cmn_update_table(smu,
-					   SMU_TABLE_ACTIVITY_MONITOR_COEFF,
-					   WORKLOAD_PPLIB_CUSTOM_BIT,
-					   (void *)(&activity_monitor_external),
-					   true);
-		if (ret) {
-			dev_err(smu->adev->dev, "[%s] Failed to set activity monitor!", __func__);
-			return ret;
-		}
+	idx = 0 * SMU_14_0_2_CUSTOM_PARAMS_COUNT;
+	if (input[idx]) {
+		/* Gfxclk */
+		activity_monitor->Gfx_FPS = input[idx + 1];
+		activity_monitor->Gfx_MinActiveFreqType = input[idx + 2];
+		activity_monitor->Gfx_MinActiveFreq = input[idx + 3];
+		activity_monitor->Gfx_BoosterFreqType = input[idx + 4];
+		activity_monitor->Gfx_BoosterFreq = input[idx + 5];
+		activity_monitor->Gfx_PD_Data_limit_c = input[idx + 6];
+		activity_monitor->Gfx_PD_Data_error_coeff = input[idx + 7];
+		activity_monitor->Gfx_PD_Data_error_rate_coeff = input[idx + 8];
+	}
+	idx = 1 * SMU_14_0_2_CUSTOM_PARAMS_COUNT;
+	if (input[idx]) {
+		/* Fclk */
+		activity_monitor->Fclk_FPS = input[idx + 1];
+		activity_monitor->Fclk_MinActiveFreqType = input[idx + 2];
+		activity_monitor->Fclk_MinActiveFreq = input[idx + 3];
+		activity_monitor->Fclk_BoosterFreqType = input[idx + 4];
+		activity_monitor->Fclk_BoosterFreq = input[idx + 5];
+		activity_monitor->Fclk_PD_Data_limit_c = input[idx + 6];
+		activity_monitor->Fclk_PD_Data_error_coeff = input[idx + 7];
+		activity_monitor->Fclk_PD_Data_error_rate_coeff = input[idx + 8];
 	}

-	if (smu->power_profile_mode == PP_SMC_POWER_PROFILE_COMPUTE)
+	ret = smu_cmn_update_table(smu,
+				   SMU_TABLE_ACTIVITY_MONITOR_COEFF,
+				   WORKLOAD_PPLIB_CUSTOM_BIT,
+				   (void *)(&activity_monitor_external),
+				   true);
+	if (ret) {
+		dev_err(smu->adev->dev, "[%s] Failed to set activity monitor!", __func__);
+		return ret;
+	}
+
+	return ret;
+}
+
+static int smu_v14_0_2_set_power_profile_mode(struct smu_context *smu,
+					      u32 workload_mask,
+					      long *custom_params,
+					      u32 custom_params_max_idx)
+{
+	u32 backend_workload_mask = 0;
+	int ret, idx = -1, i;
+
+	smu_cmn_get_backend_workload_mask(smu, workload_mask,
+					  &backend_workload_mask);
+
+	/* disable deep sleep if compute is enabled */
+	if (workload_mask & (1 << PP_SMC_POWER_PROFILE_COMPUTE))
 		smu_v14_0_deep_sleep_control(smu, false);
-	else if (current_profile_mode == PP_SMC_POWER_PROFILE_COMPUTE)
+	else
 		smu_v14_0_deep_sleep_control(smu, true);

-	/* conv PP_SMC_POWER_PROFILE* to WORKLOAD_PPLIB_*_BIT */
-	workload_type = smu_cmn_to_asic_specific_index(smu,
-						       CMN2ASIC_MAPPING_WORKLOAD,
-						       smu->power_profile_mode);
-	if (workload_type < 0)
-		return -EINVAL;
+	if (workload_mask & (1 << PP_SMC_POWER_PROFILE_CUSTOM)) {
+		if (!smu->custom_profile_params) {
+			smu->custom_profile_params =
+				kzalloc(SMU_14_0_2_CUSTOM_PARAMS_SIZE, GFP_KERNEL);
+			if (!smu->custom_profile_params)
+				return -ENOMEM;
+		}
+		if (custom_params && custom_params_max_idx) {
+			if (custom_params_max_idx != SMU_14_0_2_CUSTOM_PARAMS_COUNT)
+				return -EINVAL;
+			if (custom_params[0] >= SMU_14_0_2_CUSTOM_PARAMS_CLOCK_COUNT)
+				return -EINVAL;
+			idx = custom_params[0] * SMU_14_0_2_CUSTOM_PARAMS_COUNT;
+			smu->custom_profile_params[idx] = 1;
+			for (i = 1; i < custom_params_max_idx; i++)
+				smu->custom_profile_params[idx + i] = custom_params[i];
+		}
+		ret = smu_v14_0_2_set_power_profile_mode_coeff(smu,
+							       smu->custom_profile_params);
+		if (ret) {
+			if (idx != -1)
+				smu->custom_profile_params[idx] = 0;
+			return ret;
+		}
+	} else if (smu->custom_profile_params) {
+		memset(smu->custom_profile_params, 0, SMU_14_0_2_CUSTOM_PARAMS_SIZE);
+	}

 	ret = smu_cmn_send_smc_msg_with_param(smu, SMU_MSG_SetWorkloadMask,
-										  smu->workload_mask, NULL);
-
-	if (!ret)
-		smu_cmn_assign_power_profile(smu);
+					      backend_workload_mask, NULL);
+	if (ret) {
+		dev_err(smu->adev->dev, "Failed to set workload mask 0x%08x\n",
+			workload_mask);
+		if (idx != -1)
+			smu->custom_profile_params[idx] = 0;
+		return ret;
+	}

 	return ret;
 }
--- a/drivers/gpu/drm/amd/pm/swsmu/smu_cmn.c
+++ b/drivers/gpu/drm/amd/pm/swsmu/smu_cmn.c
@ -1144,14 +1144,6 @@ int smu_cmn_set_mp1_state(struct smu_context *smu,
 	return ret;
 }

-void smu_cmn_assign_power_profile(struct smu_context *smu)
-{
-	uint32_t index;
-	index = fls(smu->workload_mask);
-	index = index > 0 && index <= WORKLOAD_POLICY_MAX ? index - 1 : 0;
-	smu->power_profile_mode = smu->workload_setting[index];
-}
-
 bool smu_cmn_is_audio_func_enabled(struct amdgpu_device *adev)
 {
 	struct pci_dev *p = NULL;
@ -1229,3 +1221,28 @@ void smu_cmn_generic_plpd_policy_desc(struct smu_dpm_policy *policy)
 {
 	policy->desc = &xgmi_plpd_policy_desc;
 }
+
+void smu_cmn_get_backend_workload_mask(struct smu_context *smu,
+				       u32 workload_mask,
+				       u32 *backend_workload_mask)
+{
+	int workload_type;
+	u32 profile_mode;
+
+	*backend_workload_mask = 0;
+
+	for (profile_mode = 0; profile_mode < PP_SMC_POWER_PROFILE_COUNT; profile_mode++) {
+		if (!(workload_mask & (1 << profile_mode)))
+			continue;
+
+		/* conv PP_SMC_POWER_PROFILE* to WORKLOAD_PPLIB_*_BIT */
+		workload_type = smu_cmn_to_asic_specific_index(smu,
+							       CMN2ASIC_MAPPING_WORKLOAD,
+							       profile_mode);
+
+		if (workload_type < 0)
+			continue;
+
+		*backend_workload_mask |= 1 << workload_type;
+	}
+}
--- a/drivers/gpu/drm/amd/pm/swsmu/smu_cmn.h
+++ b/drivers/gpu/drm/amd/pm/swsmu/smu_cmn.h
@ -130,8 +130,6 @@ void smu_cmn_init_soft_gpu_metrics(void *table, uint8_t frev, uint8_t crev);
 int smu_cmn_set_mp1_state(struct smu_context *smu,
 			  enum pp_mp1_state mp1_state);

-void smu_cmn_assign_power_profile(struct smu_context *smu);
-
 /*
 * Helper function to make sysfs_emit_at() happy. Align buf to
 * the current page boundary and record the offset.
@ -149,5 +147,9 @@ bool smu_cmn_is_audio_func_enabled(struct amdgpu_device *adev);
 void smu_cmn_generic_soc_policy_desc(struct smu_dpm_policy *policy);
 void smu_cmn_generic_plpd_policy_desc(struct smu_dpm_policy *policy);

+void smu_cmn_get_backend_workload_mask(struct smu_context *smu,
+				       u32 workload_mask,
+				       u32 *backend_workload_mask);
+
 #endif
 #endif
--- a/drivers/gpu/drm/display/drm_dp_mst_topology.c
+++ b/drivers/gpu/drm/display/drm_dp_mst_topology.c
@ -320,6 +320,9 @@ static bool drm_dp_decode_sideband_msg_hdr(const struct drm_dp_mst_topology_mgr
 	hdr->broadcast = (buf[idx] >> 7) & 0x1;
 	hdr->path_msg = (buf[idx] >> 6) & 0x1;
 	hdr->msg_len = buf[idx] & 0x3f;
+	if (hdr->msg_len < 1)		/* min space for body CRC */
+		return false;
+
 	idx++;
 	hdr->somt = (buf[idx] >> 7) & 0x1;
 	hdr->eomt = (buf[idx] >> 6) & 0x1;
@ -3697,8 +3700,7 @@ int drm_dp_mst_topology_mgr_set_mst(struct drm_dp_mst_topology_mgr *mgr, bool ms
 		ret = 0;
 		mgr->payload_id_table_cleared = false;

-		memset(&mgr->down_rep_recv, 0, sizeof(mgr->down_rep_recv));
-		memset(&mgr->up_req_recv, 0, sizeof(mgr->up_req_recv));
+		mgr->reset_rx_state = true;
 	}

 out_unlock:
@ -3856,6 +3858,11 @@ int drm_dp_mst_topology_mgr_resume(struct drm_dp_mst_topology_mgr *mgr,
 }
 EXPORT_SYMBOL(drm_dp_mst_topology_mgr_resume);

+static void reset_msg_rx_state(struct drm_dp_sideband_msg_rx *msg)
+{
+	memset(msg, 0, sizeof(*msg));
+}
+
 static bool
 drm_dp_get_one_sb_msg(struct drm_dp_mst_topology_mgr *mgr, bool up,
 		      struct drm_dp_mst_branch **mstb)
@ -3934,6 +3941,34 @@ drm_dp_get_one_sb_msg(struct drm_dp_mst_topology_mgr *mgr, bool up,
 	return true;
 }

+static int get_msg_request_type(u8 data)
+{
+	return data & 0x7f;
+}
+
+static bool verify_rx_request_type(struct drm_dp_mst_topology_mgr *mgr,
+				   const struct drm_dp_sideband_msg_tx *txmsg,
+				   const struct drm_dp_sideband_msg_rx *rxmsg)
+{
+	const struct drm_dp_sideband_msg_hdr *hdr = &rxmsg->initial_hdr;
+	const struct drm_dp_mst_branch *mstb = txmsg->dst;
+	int tx_req_type = get_msg_request_type(txmsg->msg[0]);
+	int rx_req_type = get_msg_request_type(rxmsg->msg[0]);
+	char rad_str[64];
+
+	if (tx_req_type == rx_req_type)
+		return true;
+
+	drm_dp_mst_rad_to_str(mstb->rad, mstb->lct, rad_str, sizeof(rad_str));
+	drm_dbg_kms(mgr->dev,
+		    "Got unexpected MST reply, mstb: %p seqno: %d lct: %d rad: %s rx_req_type: %s (%02x) != tx_req_type: %s (%02x)\n",
+		    mstb, hdr->seqno, mstb->lct, rad_str,
+		    drm_dp_mst_req_type_str(rx_req_type), rx_req_type,
+		    drm_dp_mst_req_type_str(tx_req_type), tx_req_type);
+
+	return false;
+}
+
 static int drm_dp_mst_handle_down_rep(struct drm_dp_mst_topology_mgr *mgr)
 {
 	struct drm_dp_sideband_msg_tx *txmsg;
@ -3949,9 +3984,9 @@ static int drm_dp_mst_handle_down_rep(struct drm_dp_mst_topology_mgr *mgr)

 	/* find the message */
 	mutex_lock(&mgr->qlock);
+
 	txmsg = list_first_entry_or_null(&mgr->tx_msg_downq,
 					 struct drm_dp_sideband_msg_tx, next);
-	mutex_unlock(&mgr->qlock);

 	/* Were we actually expecting a response, and from this mstb? */
 	if (!txmsg || txmsg->dst != mstb) {
@ -3960,6 +3995,15 @@ static int drm_dp_mst_handle_down_rep(struct drm_dp_mst_topology_mgr *mgr)
 		hdr = &msg->initial_hdr;
 		drm_dbg_kms(mgr->dev, "Got MST reply with no msg %p %d %d %02x %02x\n",
 			    mstb, hdr->seqno, hdr->lct, hdr->rad[0], msg->msg[0]);
+
+		mutex_unlock(&mgr->qlock);
+
+		goto out_clear_reply;
+	}
+
+	if (!verify_rx_request_type(mgr, txmsg, msg)) {
+		mutex_unlock(&mgr->qlock);
+
 		goto out_clear_reply;
 	}

@ -3975,20 +4019,15 @@ static int drm_dp_mst_handle_down_rep(struct drm_dp_mst_topology_mgr *mgr)
 			    txmsg->reply.u.nak.nak_data);
 	}

-	memset(msg, 0, sizeof(struct drm_dp_sideband_msg_rx));
-	drm_dp_mst_topology_put_mstb(mstb);
-
-	mutex_lock(&mgr->qlock);
 	txmsg->state = DRM_DP_SIDEBAND_TX_RX;
 	list_del(&txmsg->next);
+
 	mutex_unlock(&mgr->qlock);

 	wake_up_all(&mgr->tx_waitq);

-	return 0;
-
 out_clear_reply:
-	memset(msg, 0, sizeof(struct drm_dp_sideband_msg_rx));
+	reset_msg_rx_state(msg);
 out:
 	if (mstb)
 		drm_dp_mst_topology_put_mstb(mstb);
@ -4070,16 +4109,20 @@ static void drm_dp_mst_up_req_work(struct work_struct *work)
 static int drm_dp_mst_handle_up_req(struct drm_dp_mst_topology_mgr *mgr)
 {
 	struct drm_dp_pending_up_req *up_req;
+	struct drm_dp_mst_branch *mst_primary;
+	int ret = 0;

 	if (!drm_dp_get_one_sb_msg(mgr, true, NULL))
-		goto out;
+		goto out_clear_reply;

 	if (!mgr->up_req_recv.have_eomt)
 		return 0;

 	up_req = kzalloc(sizeof(*up_req), GFP_KERNEL);
-	if (!up_req)
-		return -ENOMEM;
+	if (!up_req) {
+		ret = -ENOMEM;
+		goto out_clear_reply;
+	}

 	INIT_LIST_HEAD(&up_req->next);

@ -4090,10 +4133,19 @@ static int drm_dp_mst_handle_up_req(struct drm_dp_mst_topology_mgr *mgr)
 		drm_dbg_kms(mgr->dev, "Received unknown up req type, ignoring: %x\n",
 			    up_req->msg.req_type);
 		kfree(up_req);
-		goto out;
+		goto out_clear_reply;
 	}

-	drm_dp_send_up_ack_reply(mgr, mgr->mst_primary, up_req->msg.req_type,
+	mutex_lock(&mgr->lock);
+	mst_primary = mgr->mst_primary;
+	if (!mst_primary || !drm_dp_mst_topology_try_get_mstb(mst_primary)) {
+		mutex_unlock(&mgr->lock);
+		kfree(up_req);
+		goto out_clear_reply;
+	}
+	mutex_unlock(&mgr->lock);
+
+	drm_dp_send_up_ack_reply(mgr, mst_primary, up_req->msg.req_type,
 				 false);

 	if (up_req->msg.req_type == DP_CONNECTION_STATUS_NOTIFY) {
@ -4110,13 +4162,13 @@ static int drm_dp_mst_handle_up_req(struct drm_dp_mst_topology_mgr *mgr)
 			    conn_stat->peer_device_type);

 		mutex_lock(&mgr->probe_lock);
-		handle_csn = mgr->mst_primary->link_address_sent;
+		handle_csn = mst_primary->link_address_sent;
 		mutex_unlock(&mgr->probe_lock);

 		if (!handle_csn) {
 			drm_dbg_kms(mgr->dev, "Got CSN before finish topology probing. Skip it.");
 			kfree(up_req);
-			goto out;
+			goto out_put_primary;
 		}
 	} else if (up_req->msg.req_type == DP_RESOURCE_STATUS_NOTIFY) {
 		const struct drm_dp_resource_status_notify *res_stat =
@ -4133,9 +4185,22 @@ static int drm_dp_mst_handle_up_req(struct drm_dp_mst_topology_mgr *mgr)
 	mutex_unlock(&mgr->up_req_lock);
 	queue_work(system_long_wq, &mgr->up_req_work);

-out:
-	memset(&mgr->up_req_recv, 0, sizeof(struct drm_dp_sideband_msg_rx));
-	return 0;
+out_put_primary:
+	drm_dp_mst_topology_put_mstb(mst_primary);
+out_clear_reply:
+	reset_msg_rx_state(&mgr->up_req_recv);
+	return ret;
+}
+
+static void update_msg_rx_state(struct drm_dp_mst_topology_mgr *mgr)
+{
+	mutex_lock(&mgr->lock);
+	if (mgr->reset_rx_state) {
+		mgr->reset_rx_state = false;
+		reset_msg_rx_state(&mgr->down_rep_recv);
+		reset_msg_rx_state(&mgr->up_req_recv);
+	}
+	mutex_unlock(&mgr->lock);
 }

 /**
@ -4172,6 +4237,8 @@ int drm_dp_mst_hpd_irq_handle_event(struct drm_dp_mst_topology_mgr *mgr, const u
 		*handled = true;
 	}

+	update_msg_rx_state(mgr);
+
 	if (esi[1] & DP_DOWN_REP_MSG_RDY) {
 		ret = drm_dp_mst_handle_down_rep(mgr);
 		*handled = true;
--- a/drivers/gpu/drm/sti/sti_mixer.c
+++ b/drivers/gpu/drm/sti/sti_mixer.c
@ -137,7 +137,7 @@ static void mixer_dbg_crb(struct seq_file *s, int val)
 	}
 }

-static void mixer_dbg_mxn(struct seq_file *s, void *addr)
+static void mixer_dbg_mxn(struct seq_file *s, void __iomem *addr)
 {
 	int i;

--- a/drivers/gpu/drm/v3d/v3d_perfmon.c
+++ b/drivers/gpu/drm/v3d/v3d_perfmon.c
@ -254,9 +254,9 @@ void v3d_perfmon_start(struct v3d_dev *v3d, struct v3d_perfmon *perfmon)
 		V3D_CORE_WRITE(0, V3D_V4_PCTR_0_SRC_X(source), channel);
 	}

+	V3D_CORE_WRITE(0, V3D_V4_PCTR_0_EN, mask);
 	V3D_CORE_WRITE(0, V3D_V4_PCTR_0_CLR, mask);
 	V3D_CORE_WRITE(0, V3D_PCTR_0_OVERFLOW, mask);
-	V3D_CORE_WRITE(0, V3D_V4_PCTR_0_EN, mask);

 	v3d->active_perfmon = perfmon;
 }
--- a/drivers/gpu/drm/xe/xe_devcoredump.c
+++ b/drivers/gpu/drm/xe/xe_devcoredump.c
@ -155,36 +155,6 @@ static void xe_devcoredump_snapshot_free(struct xe_devcoredump_snapshot *ss)
 	ss->vm = NULL;
 }

-static void xe_devcoredump_deferred_snap_work(struct work_struct *work)
-{
-	struct xe_devcoredump_snapshot *ss = container_of(work, typeof(*ss), work);
-	struct xe_devcoredump *coredump = container_of(ss, typeof(*coredump), snapshot);
-	struct xe_device *xe = coredump_to_xe(coredump);
-	unsigned int fw_ref;
-
-	xe_pm_runtime_get(xe);
-
-	/* keep going if fw fails as we still want to save the memory and SW data */
-	fw_ref = xe_force_wake_get(gt_to_fw(ss->gt), XE_FORCEWAKE_ALL);
-	if (!xe_force_wake_ref_has_domain(fw_ref, XE_FORCEWAKE_ALL))
-		xe_gt_info(ss->gt, "failed to get forcewake for coredump capture\n");
-	xe_vm_snapshot_capture_delayed(ss->vm);
-	xe_guc_exec_queue_snapshot_capture_delayed(ss->ge);
-	xe_force_wake_put(gt_to_fw(ss->gt), fw_ref);
-
-	xe_pm_runtime_put(xe);
-
-	/* Calculate devcoredump size */
-	ss->read.size = __xe_devcoredump_read(NULL, INT_MAX, coredump);
-
-	ss->read.buffer = kvmalloc(ss->read.size, GFP_USER);
-	if (!ss->read.buffer)
-		return;
-
-	__xe_devcoredump_read(ss->read.buffer, ss->read.size, coredump);
-	xe_devcoredump_snapshot_free(ss);
-}
-
 static ssize_t xe_devcoredump_read(char *buffer, loff_t offset,
 				   size_t count, void *data, size_t datalen)
 {
@ -234,6 +204,45 @@ static void xe_devcoredump_free(void *data)
 		 "Xe device coredump has been deleted.\n");
 }

+static void xe_devcoredump_deferred_snap_work(struct work_struct *work)
+{
+	struct xe_devcoredump_snapshot *ss = container_of(work, typeof(*ss), work);
+	struct xe_devcoredump *coredump = container_of(ss, typeof(*coredump), snapshot);
+	struct xe_device *xe = coredump_to_xe(coredump);
+	unsigned int fw_ref;
+
+	/*
+	 * NB: Despite passing a GFP_ flags parameter here, more allocations are done
+	 * internally using GFP_KERNEL expliictly. Hence this call must be in the worker
+	 * thread and not in the initial capture call.
+	 */
+	dev_coredumpm_timeout(gt_to_xe(ss->gt)->drm.dev, THIS_MODULE, coredump, 0, GFP_KERNEL,
+			      xe_devcoredump_read, xe_devcoredump_free,
+			      XE_COREDUMP_TIMEOUT_JIFFIES);
+
+	xe_pm_runtime_get(xe);
+
+	/* keep going if fw fails as we still want to save the memory and SW data */
+	fw_ref = xe_force_wake_get(gt_to_fw(ss->gt), XE_FORCEWAKE_ALL);
+	if (!xe_force_wake_ref_has_domain(fw_ref, XE_FORCEWAKE_ALL))
+		xe_gt_info(ss->gt, "failed to get forcewake for coredump capture\n");
+	xe_vm_snapshot_capture_delayed(ss->vm);
+	xe_guc_exec_queue_snapshot_capture_delayed(ss->ge);
+	xe_force_wake_put(gt_to_fw(ss->gt), fw_ref);
+
+	xe_pm_runtime_put(xe);
+
+	/* Calculate devcoredump size */
+	ss->read.size = __xe_devcoredump_read(NULL, INT_MAX, coredump);
+
+	ss->read.buffer = kvmalloc(ss->read.size, GFP_USER);
+	if (!ss->read.buffer)
+		return;
+
+	__xe_devcoredump_read(ss->read.buffer, ss->read.size, coredump);
+	xe_devcoredump_snapshot_free(ss);
+}
+
 static void devcoredump_snapshot(struct xe_devcoredump *coredump,
 				 struct xe_sched_job *job)
 {
@ -310,10 +319,6 @@ void xe_devcoredump(struct xe_sched_job *job)
 	drm_info(&xe->drm, "Xe device coredump has been created\n");
 	drm_info(&xe->drm, "Check your /sys/class/drm/card%d/device/devcoredump/data\n",
 		 xe->drm.primary->index);
-
-	dev_coredumpm_timeout(xe->drm.dev, THIS_MODULE, coredump, 0, GFP_KERNEL,
-			      xe_devcoredump_read, xe_devcoredump_free,
-			      XE_COREDUMP_TIMEOUT_JIFFIES);
 }

 static void xe_driver_devcoredump_fini(void *arg)
--- a/Show More
+++ b/Show More