diff mbox series

[v4] RISC-V: enable XIP

Message ID 20210306121438.12827-1-vitaly.wool@konsulko.com (mailing list archive)
State New, archived
Headers show
Series [v4] RISC-V: enable XIP | expand

Commit Message

Vitaly Wool March 6, 2021, 12:14 p.m. UTC
Introduce XIP (eXecute In Place) support for RISC-V platforms.
It allows code to be executed directly from non-volatile storage
directly addressable by the CPU, such as QSPI NOR flash which can
be found on many RISC-V platforms. This makes way for significant
optimization of RAM footprint. The XIP kernel is not compressed
since it has to run directly from flash, so it will occupy more
space on the non-volatile storage to The physical flash address
used to link the kernel object files and for storing it has to
be known at compile time and is represented by a Kconfig option.

XIP on RISC-V will currently only work on MMU-enabled kernels.

Signed-off-by: Vitaly Wool <vitaly.wool@konsulko.com>

---

Changed in v2:
- dedicated macro for XIP address fixup when MMU is not enabled yet
  = both for 32-bit and 64-bit RISC-V
- SP is explicitly set to a safe place in RAM before __copy_data call
- removed redundant alignment requirements in vmlinux-xip.lds.S
- changed long -> uintptr_t typecast in __XIP_FIXUP macro.

Changed in v3:
- rebased against latest for-next
- XIP address fixup macro now takes an argument
- SMP related fixes

Changes in v4:
- rebased against the current for-next
- less #ifdef's in C/ASM code
- dedicated XIP_FIXUP_OFFSET assembler macro in head.S
- C-specific definitions moved into #ifndef __ASSEMBLY__
- Fixed multi-core boot

 arch/riscv/Kconfig                  |  44 +++++++++-
 arch/riscv/Makefile                 |   8 +-
 arch/riscv/boot/Makefile            |  13 +++
 arch/riscv/include/asm/pgtable.h    |  65 ++++++++++++--
 arch/riscv/kernel/cpu_ops_sbi.c     |  10 ++-
 arch/riscv/kernel/head.S            |  59 ++++++++++++-
 arch/riscv/kernel/head.h            |   3 +
 arch/riscv/kernel/setup.c           |   8 +-
 arch/riscv/kernel/vmlinux-xip.lds.S | 132 ++++++++++++++++++++++++++++
 arch/riscv/kernel/vmlinux.lds.S     |   6 ++
 arch/riscv/mm/init.c                | 100 +++++++++++++++++++--
 11 files changed, 430 insertions(+), 18 deletions(-)
 create mode 100644 arch/riscv/kernel/vmlinux-xip.lds.S

Comments

kernel test robot March 7, 2021, 9:40 a.m. UTC | #1
Hi Vitaly,

I love your patch! Yet something to improve:

[auto build test ERROR on linus/master]
[also build test ERROR on v5.12-rc2 next-20210305]
[cannot apply to linux/master]
[If your patch is applied to the wrong git tree, kindly drop us a note.
And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch]

url:    https://github.com/0day-ci/linux/commits/Vitaly-Wool/RISC-V-enable-XIP/20210307-103904
base:   https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git 280d542f6ffac0e6d65dc267f92191d509b13b64
config: riscv-defconfig (attached as .config)
compiler: riscv64-linux-gcc (GCC) 9.3.0
reproduce (this is a W=1 build):
        wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross
        chmod +x ~/bin/make.cross
        # https://github.com/0day-ci/linux/commit/c9cb15092dcf53140fba57780677a9dcaef70612
        git remote add linux-review https://github.com/0day-ci/linux
        git fetch --no-tags linux-review Vitaly-Wool/RISC-V-enable-XIP/20210307-103904
        git checkout c9cb15092dcf53140fba57780677a9dcaef70612
        # save the attached .config to linux build tree
        COMPILER_INSTALL_PATH=$HOME/0day COMPILER=gcc-9.3.0 make.cross ARCH=riscv 

If you fix the issue, kindly add following tag as appropriate
Reported-by: kernel test robot <lkp@intel.com>

All errors (new ones prefixed by >>):

   arch/riscv/kernel/cpu_ops_sbi.c: In function 'get_secondary_start_phys':
>> arch/riscv/kernel/cpu_ops_sbi.c:59:3: error: implicit declaration of function 'XIP_PHYS_ADDR' [-Werror=implicit-function-declaration]
      59 |   XIP_PHYS_ADDR(secondary_start_sbi) :
         |   ^~~~~~~~~~~~~
   cc1: some warnings being treated as errors


vim +/XIP_PHYS_ADDR +59 arch/riscv/kernel/cpu_ops_sbi.c

    55	
    56	static inline unsigned long get_secondary_start_phys(void)
    57	{
    58		return IS_ENABLED(CONFIG_XIP_KERNEL) ?
  > 59			XIP_PHYS_ADDR(secondary_start_sbi) :
    60			__pa_symbol(secondary_start_sbi);
    61	}
    62	

---
0-DAY CI Kernel Test Service, Intel Corporation
https://lists.01.org/hyperkitty/list/kbuild-all@lists.01.org
kernel test robot March 8, 2021, 3:58 a.m. UTC | #2
Hi Vitaly,

I love your patch! Yet something to improve:

[auto build test ERROR on linus/master]
[also build test ERROR on v5.12-rc2]
[cannot apply to linux/master]
[If your patch is applied to the wrong git tree, kindly drop us a note.
And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch]

url:    https://github.com/0day-ci/linux/commits/Vitaly-Wool/RISC-V-enable-XIP/20210307-103904
base:   https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git 280d542f6ffac0e6d65dc267f92191d509b13b64
config: riscv-allmodconfig (attached as .config)
compiler: riscv64-linux-gcc (GCC) 9.3.0
reproduce (this is a W=1 build):
        wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross
        chmod +x ~/bin/make.cross
        # https://github.com/0day-ci/linux/commit/c9cb15092dcf53140fba57780677a9dcaef70612
        git remote add linux-review https://github.com/0day-ci/linux
        git fetch --no-tags linux-review Vitaly-Wool/RISC-V-enable-XIP/20210307-103904
        git checkout c9cb15092dcf53140fba57780677a9dcaef70612
        # save the attached .config to linux build tree
        COMPILER_INSTALL_PATH=$HOME/0day COMPILER=gcc-9.3.0 make.cross ARCH=riscv 

If you fix the issue, kindly add following tag as appropriate
Reported-by: kernel test robot <lkp@intel.com>

All errors (new ones prefixed by >>):

>> riscv64-linux-ld: section .data LMA [0000000000800000,0000000000a90eaf] overlaps section .text LMA [00000000001095a0,00000000014ae61b]
>> riscv64-linux-ld: section .rodata VMA [ffffffe000cae640,ffffffe001042f0f] overlaps section .bss VMA [ffffffe00059d000,ffffffe001a54f6f]
   riscv64-linux-ld: arch/riscv/mm/init.o: in function `.L122':
>> init.c:(.text+0x288): undefined reference to `__init_text_begin'
   riscv64-linux-ld: arch/riscv/mm/init.o: in function `.L123':
>> init.c:(.text+0x290): undefined reference to `__init_data_begin'
   riscv64-linux-ld: arch/riscv/mm/init.o: in function `.L0 ':
   init.c:(.text+0x2ea): undefined reference to `__init_text_begin'
>> riscv64-linux-ld: init.c:(.text+0x302): undefined reference to `__init_data_begin'

---
0-DAY CI Kernel Test Service, Intel Corporation
https://lists.01.org/hyperkitty/list/kbuild-all@lists.01.org
Guenter Roeck March 21, 2021, 8:21 p.m. UTC | #3
On Sat, Mar 06, 2021 at 02:14:38PM +0200, Vitaly Wool wrote:
> Introduce XIP (eXecute In Place) support for RISC-V platforms.
> It allows code to be executed directly from non-volatile storage
> directly addressable by the CPU, such as QSPI NOR flash which can
> be found on many RISC-V platforms. This makes way for significant
> optimization of RAM footprint. The XIP kernel is not compressed
> since it has to run directly from flash, so it will occupy more
> space on the non-volatile storage to The physical flash address
> used to link the kernel object files and for storing it has to
> be known at compile time and is represented by a Kconfig option.
> 
> XIP on RISC-V will currently only work on MMU-enabled kernels.
> 
> Signed-off-by: Vitaly Wool <vitaly.wool@konsulko.com>
> ---

The impact of this patch on RISC-V images running in qemu is quite fatal.

[    0.000000] Linux version 5.12.0-rc3-next-20210319 (groeck@saturn.roeck-us.net) (riscv64-linux-gcc (GCC) 9.3.0, GNU ld (GNU Binutils) 2.34) #1 SMP Fri Mar 19 04:01:41 PDT 2021
[    0.000000] OF: fdt: Ignoring memory range 0x80000000 - 0x80200000
[    0.000000] Machine model: riscv-virtio,qemu
[    0.000000] earlycon: uart8250 at MMIO 0x0000000010000000 (options '115200')
[    0.000000] printk: bootconsole [uart8250] enabled
[    0.000000] efi: UEFI not found.
[    0.000000] Unable to handle kernel paging request at virtual address 0000000040000001
[    0.000000] Oops [#1]
[    0.000000] Modules linked in:
[    0.000000] CPU: 0 PID: 0 Comm: swapper Not tainted 5.12.0-rc3-next-20210319 #1
[    0.000000] Hardware name: riscv-virtio,qemu (DT)
[    0.000000] epc : fdt_check_header+0x0/0x1fc
[    0.000000]  ra : early_init_dt_verify+0x16/0x6e
[    0.000000] epc : ffffffe00049c04a ra : ffffffe000a26144 sp : ffffffe001603f10
[    0.000000]  gp : ffffffe0017d04b0 tp : ffffffe001611280 t0 : ffffffffffffffff
[    0.000000]  t1 : 0000000012adc51a t2 : 000000000004ffeb s0 : ffffffe001603f30
[    0.000000]  s1 : 0000000040000000 a0 : 0000000040000000 a1 : ffffffe001617ae8
[    0.000000]  a2 : 0000000000000000 a3 : 0000000000000001 a4 : 0000000000000018
[    0.000000]  a5 : ffffffe000c0f8f0 a6 : ffffffe01fdfee70 a7 : 0000000000000000
[    0.000000]  s2 : 0000000080000200 s3 : 0000000000000fff s4 : ffffffe000c0ec60
[    0.000000]  s5 : 0000000000000006 s6 : 00000000000001c0 s7 : ffffffe01fdfee40
[    0.000000]  s8 : 0000000081000200 s9 : 0000000080000200 s10: ffffffe000c01000
[    0.000000]  s11: 0000000000000fff t3 : 000000000000000e t4 : ffffffffa6000000
[    0.000000]  t5 : ffffffffffffffff t6 : 0000000000000015
[    0.000000] status: 0000000000000100 badaddr: 0000000040000001 cause: 000000000000000d
[    0.000000] Call Trace:
[    0.000000] [<ffffffe00049c04a>] fdt_check_header+0x0/0x1fc
[    0.000000] [<ffffffe000a02c84>] setup_arch+0x3a8/0x414
[    0.000000] [<ffffffe000a006fa>] start_kernel+0x8e/0x62c
[    0.000000] random: get_random_bytes called from print_oops_end_marker+0x22/0x44 with crng_init=0
[    0.000000] ---[ end trace 0000000000000000 ]---
[    0.000000] Kernel panic - not syncing: Attempted to kill the idle task!
[    0.000000] ---[ end Kernel panic - not syncing: Attempted to kill the idle task! ]---

This is with riscv:defconfig. Reverting this patch fixes the problem.

Bisect log is attached.

Guenter

---
# bad: [f00397ee41c79b6155b9b44abd0055b2c0621349] Add linux-next specific files for 20210319
# good: [1e28eed17697bcf343c6743f0028cc3b5dd88bf0] Linux 5.12-rc3
git bisect start 'HEAD' 'v5.12-rc3'
# bad: [728bc19f9c531acd94e4139e8258a1577f17d2ff] Merge remote-tracking branch 'drm/drm-next'
git bisect bad 728bc19f9c531acd94e4139e8258a1577f17d2ff
# bad: [ad5d10f254a7d91a9fadadcf2b109795544e9dcb] Merge remote-tracking branch 'hid/for-next'
git bisect bad ad5d10f254a7d91a9fadadcf2b109795544e9dcb
# good: [163518c802260a16b5bca8bd7fe0a9335a6860e1] Merge remote-tracking branch 'qcom/for-next'
git bisect good 163518c802260a16b5bca8bd7fe0a9335a6860e1
# bad: [a9fdd30a39510f783a38fe5c30b2d888397b35d5] Merge remote-tracking branch 'btrfs/for-next'
git bisect bad a9fdd30a39510f783a38fe5c30b2d888397b35d5
# good: [3f26d646c9c30116057e2cc34091e7d3fd894863] Merge remote-tracking branch 'microblaze/next'
git bisect good 3f26d646c9c30116057e2cc34091e7d3fd894863
# good: [f0fd694b8c3eba5651fa9eed185cc1c39c1a3d64] Merge branch 'ext/qu/subpage-write-meta-v2' into for-next-next-v5.12-20210317
git bisect good f0fd694b8c3eba5651fa9eed185cc1c39c1a3d64
# bad: [762f6038cff9e5658c29936a0ff0398fb2b77dc9] Merge remote-tracking branch 'sh/for-next'
git bisect bad 762f6038cff9e5658c29936a0ff0398fb2b77dc9
# good: [fb009cbdd0693bd633f11e99526617b3d392cfad] firmware: bcm47xx_nvram: rename finding function and its variables
git bisect good fb009cbdd0693bd633f11e99526617b3d392cfad
# good: [10b3a0a8f783dc5491921f0236f1b756cbe1b12d] Merge remote-tracking branch 'parisc-hd/for-next'
git bisect good 10b3a0a8f783dc5491921f0236f1b756cbe1b12d
# good: [2882b7626f4903a8e9250b328cdf7396a6deecac] sh: kernel: traps: remove unused variable
git bisect good 2882b7626f4903a8e9250b328cdf7396a6deecac
# bad: [06c7c914de26c5a4f1418fd54e4dfd0be4215de6] RISC-V: enable XIP
git bisect bad 06c7c914de26c5a4f1418fd54e4dfd0be4215de6
# good: [9530141455c968938a913d602a236c2a7b0322e1] riscv: Add ARCH_HAS_FORTIFY_SOURCE
git bisect good 9530141455c968938a913d602a236c2a7b0322e1
# good: [2f100585d04506004b8027ec9bbaee26940a769f] riscv: Enable generic clockevent broadcast
git bisect good 2f100585d04506004b8027ec9bbaee26940a769f
# first bad commit: [06c7c914de26c5a4f1418fd54e4dfd0be4215de6] RISC-V: enable XIP
diff mbox series

Patch

diff --git a/arch/riscv/Kconfig b/arch/riscv/Kconfig
index 85d626b8ce5e..59fb945a900e 100644
--- a/arch/riscv/Kconfig
+++ b/arch/riscv/Kconfig
@@ -438,7 +438,7 @@  config EFI_STUB
 
 config EFI
 	bool "UEFI runtime support"
-	depends on OF
+	depends on OF && !XIP_KERNEL
 	select LIBFDT
 	select UCS2_STRING
 	select EFI_PARAMS_FROM_FDT
@@ -462,11 +462,51 @@  config STACKPROTECTOR_PER_TASK
 	def_bool y
 	depends on STACKPROTECTOR && CC_HAVE_STACKPROTECTOR_TLS
 
+config XIP_KERNEL
+	bool "Kernel Execute-In-Place from ROM"
+	depends on MMU
+	help
+	  Execute-In-Place allows the kernel to run from non-volatile storage
+	  directly addressable by the CPU, such as NOR flash. This saves RAM
+	  space since the text section of the kernel is not loaded from flash
+	  to RAM.  Read-write sections, such as the data section and stack,
+	  are still copied to RAM.  The XIP kernel is not compressed since
+	  it has to run directly from flash, so it will take more space to
+	  store it.  The flash address used to link the kernel object files,
+	  and for storing it, is configuration dependent. Therefore, if you
+	  say Y here, you must know the proper physical address where to
+	  store the kernel image depending on your own flash memory usage.
+
+	  Also note that the make target becomes "make xipImage" rather than
+	  "make zImage" or "make Image".  The final kernel binary to put in
+	  ROM memory will be arch/riscv/boot/xipImage.
+
+	  If unsure, say N.
+
+config XIP_PHYS_ADDR
+	hex "XIP Kernel Physical Location"
+	depends on XIP_KERNEL
+	default "0x21000000"
+	help
+	  This is the physical address in your flash memory the kernel will
+	  be linked for and stored to.  This address is dependent on your
+	  own flash usage.
+
+config XIP_PHYS_RAM_BASE
+	hex "Platform Physical RAM address"
+	depends on XIP_KERNEL
+	default "0x80000000"
+	help
+	  This is the physical address of RAM in the system. It has to be
+	  explicitly specified to run early relocations of read-write data
+	  from flash to RAM.
+
 endmenu
 
 config BUILTIN_DTB
-	def_bool n
+	bool
 	depends on OF
+	default y if XIP_KERNEL
 
 menu "Power management options"
 
diff --git a/arch/riscv/Makefile b/arch/riscv/Makefile
index 1368d943f1f3..8fcbec03974d 100644
--- a/arch/riscv/Makefile
+++ b/arch/riscv/Makefile
@@ -82,7 +82,11 @@  CHECKFLAGS += -D__riscv -D__riscv_xlen=$(BITS)
 
 # Default target when executing plain make
 boot		:= arch/riscv/boot
+ifeq ($(CONFIG_XIP_KERNEL),y)
+KBUILD_IMAGE := $(boot)/xipImage
+else
 KBUILD_IMAGE	:= $(boot)/Image.gz
+endif
 
 head-y := arch/riscv/kernel/head.o
 
@@ -95,12 +99,14 @@  PHONY += vdso_install
 vdso_install:
 	$(Q)$(MAKE) $(build)=arch/riscv/kernel/vdso $@
 
+ifneq ($(CONFIG_XIP_KERNEL),y)
 ifeq ($(CONFIG_RISCV_M_MODE)$(CONFIG_SOC_CANAAN),yy)
 KBUILD_IMAGE := $(boot)/loader.bin
 else
 KBUILD_IMAGE := $(boot)/Image.gz
 endif
-BOOT_TARGETS := Image Image.gz loader loader.bin
+endif
+BOOT_TARGETS := Image Image.gz loader loader.bin xipImage
 
 all:	$(notdir $(KBUILD_IMAGE))
 
diff --git a/arch/riscv/boot/Makefile b/arch/riscv/boot/Makefile
index 03404c84f971..6bf299f70c27 100644
--- a/arch/riscv/boot/Makefile
+++ b/arch/riscv/boot/Makefile
@@ -17,8 +17,21 @@ 
 KCOV_INSTRUMENT := n
 
 OBJCOPYFLAGS_Image :=-O binary -R .note -R .note.gnu.build-id -R .comment -S
+OBJCOPYFLAGS_xipImage :=-O binary -R .note -R .note.gnu.build-id -R .comment -S
 
 targets := Image Image.* loader loader.o loader.lds loader.bin
+targets := Image Image.* loader loader.o loader.lds loader.bin xipImage
+
+ifeq ($(CONFIG_XIP_KERNEL),y)
+
+quiet_cmd_mkxip = $(quiet_cmd_objcopy)
+cmd_mkxip = $(cmd_objcopy)
+
+$(obj)/xipImage: vmlinux FORCE
+	$(call if_changed,mkxip)
+	@$(kecho) '  Physical Address of xipImage: $(CONFIG_XIP_PHYS_ADDR)'
+
+endif
 
 $(obj)/Image: vmlinux FORCE
 	$(call if_changed,objcopy)
diff --git a/arch/riscv/include/asm/pgtable.h b/arch/riscv/include/asm/pgtable.h
index ebf817c1bdf4..4c2e6902f2b8 100644
--- a/arch/riscv/include/asm/pgtable.h
+++ b/arch/riscv/include/asm/pgtable.h
@@ -11,6 +11,33 @@ 
 
 #include <asm/pgtable-bits.h>
 
+#ifdef CONFIG_MMU
+
+#define VMALLOC_START    (PAGE_OFFSET - VMALLOC_SIZE)
+
+#ifdef CONFIG_XIP_KERNEL
+#define VMALLOC_SIZE     ((KERN_VIRT_SIZE >> 1) - SZ_16M)
+#define VMALLOC_END      (PAGE_OFFSET - SZ_16M - 1)
+
+#define XIP_OFFSET		SZ_8M
+#define XIP_MASK		(SZ_8M - 1)
+#define XIP_VIRT_ADDR(physaddr)	\
+	(PAGE_OFFSET - XIP_OFFSET + ((physaddr) & XIP_MASK))
+
+#else
+
+#define VMALLOC_SIZE     (KERN_VIRT_SIZE >> 1)
+#define VMALLOC_END      (PAGE_OFFSET - 1)
+
+#endif /* CONFIG_XIP_KERNEL */
+
+#else
+
+#ifdef CONFIG_XIP_KERNEL
+#define XIP_VIRT_ADDR(physaddr) (physaddr)
+#endif /* CONFIG_XIP_KERNEL */
+#endif /* CONFIG_MMU */
+
 #ifndef __ASSEMBLY__
 
 /* Page Upper Directory not used in RISC-V */
@@ -21,9 +48,25 @@ 
 
 #ifdef CONFIG_MMU
 
-#define VMALLOC_SIZE     (KERN_VIRT_SIZE >> 1)
-#define VMALLOC_END      (PAGE_OFFSET - 1)
-#define VMALLOC_START    (PAGE_OFFSET - VMALLOC_SIZE)
+#ifdef CONFIG_XIP_KERNEL
+/*
+ * Since we use sections to map it, this macro replaces the physical address
+ * with its virtual address while keeping offset from the base section.
+ */
+#define XIP_PHYS_ADDR(va)	 \
+	((uintptr_t)(va) - PAGE_OFFSET + XIP_OFFSET + CONFIG_XIP_PHYS_ADDR)
+
+#define XIP_VIRT_ADDR_START	XIP_VIRT_ADDR(CONFIG_XIP_PHYS_ADDR)
+#define __PHRAM_BASE		CONFIG_XIP_PHYS_RAM_BASE
+
+#define XIP_FIXUP(addr)		\
+	(((uintptr_t)(addr) >= CONFIG_XIP_PHYS_ADDR && \
+	  (uintptr_t)(addr) <= CONFIG_XIP_PHYS_ADDR + SZ_16M) ? \
+	(uintptr_t)(addr) - CONFIG_XIP_PHYS_ADDR + __PHRAM_BASE - XIP_OFFSET : \
+	(uintptr_t)(addr))
+#else
+#define XIP_FIXUP(addr)		(addr)
+#endif /* CONFIG_XIP_KERNEL */
 
 #define BPF_JIT_REGION_SIZE	(SZ_128M)
 #define BPF_JIT_REGION_START	(PAGE_OFFSET - BPF_JIT_REGION_SIZE)
@@ -484,8 +527,20 @@  static inline int ptep_clear_flush_young(struct vm_area_struct *vma,
 
 #define kern_addr_valid(addr)   (1) /* FIXME */
 
-extern void *dtb_early_va;
-extern uintptr_t dtb_early_pa;
+extern void *_dtb_early_va;
+extern uintptr_t _dtb_early_pa;
+#if defined(CONFIG_XIP_KERNEL) && defined(CONFIG_MMU)
+
+#define dtb_early_va	(*(void **)XIP_FIXUP(&_dtb_early_va))
+#define dtb_early_pa	(*(uintptr_t *)XIP_FIXUP(&_dtb_early_pa))
+
+#else
+
+#define dtb_early_va	_dtb_early_va
+#define dtb_early_pa	_dtb_early_pa
+
+#endif /* CONFIG_XIP_KERNEL */
+
 void setup_bootmem(void);
 void paging_init(void);
 void misc_mem_init(void);
diff --git a/arch/riscv/kernel/cpu_ops_sbi.c b/arch/riscv/kernel/cpu_ops_sbi.c
index 685fae72b7f5..b557eb983085 100644
--- a/arch/riscv/kernel/cpu_ops_sbi.c
+++ b/arch/riscv/kernel/cpu_ops_sbi.c
@@ -53,10 +53,18 @@  static int sbi_hsm_hart_get_status(unsigned long hartid)
 }
 #endif
 
+static inline unsigned long get_secondary_start_phys(void)
+{
+	return IS_ENABLED(CONFIG_XIP_KERNEL) ?
+		XIP_PHYS_ADDR(secondary_start_sbi) :
+		__pa_symbol(secondary_start_sbi);
+}
+
+
 static int sbi_cpu_start(unsigned int cpuid, struct task_struct *tidle)
 {
 	int rc;
-	unsigned long boot_addr = __pa_symbol(secondary_start_sbi);
+	unsigned long boot_addr = get_secondary_start_phys();
 	int hartid = cpuid_to_hartid_map(cpuid);
 
 	cpu_update_secondary_bootdata(cpuid, tidle);
diff --git a/arch/riscv/kernel/head.S b/arch/riscv/kernel/head.S
index f5a9bad86e58..17ecf6095209 100644
--- a/arch/riscv/kernel/head.S
+++ b/arch/riscv/kernel/head.S
@@ -9,11 +9,33 @@ 
 #include <linux/linkage.h>
 #include <asm/thread_info.h>
 #include <asm/page.h>
+#include <asm/pgtable.h>
 #include <asm/csr.h>
 #include <asm/hwcap.h>
 #include <asm/image.h>
 #include "efi-header.S"
 
+#ifdef CONFIG_XIP_KERNEL
+.macro XIP_FIXUP_OFFSET reg
+#ifdef CONFIG_64BIT
+		ld t0, _xip_fixup
+#else
+		lw t0, _xip_fixup
+#endif
+		add \reg, \reg, t0
+.endm
+
+_xip_fixup:
+#ifdef CONFIG_64BIT
+	.dword CONFIG_XIP_PHYS_RAM_BASE - CONFIG_XIP_PHYS_ADDR - XIP_OFFSET
+#else
+	.word CONFIG_XIP_PHYS_RAM_BASE - CONFIG_XIP_PHYS_ADDR - XIP_OFFSET
+#endif
+#else
+.macro XIP_FIXUP_OFFSET reg
+.endm
+#endif /* CONFIG_XIP_KERNEL */
+
 __HEAD
 ENTRY(_start)
 	/*
@@ -69,7 +91,11 @@  pe_head_start:
 #ifdef CONFIG_MMU
 relocate:
 	/* Relocate return address */
+#ifdef CONFIG_XIP_KERNEL
+	li a1, XIP_VIRT_ADDR(CONFIG_XIP_PHYS_ADDR)
+#else
 	li a1, PAGE_OFFSET
+#endif
 	la a2, _start
 	sub a1, a1, a2
 	add ra, ra, a1
@@ -91,6 +117,7 @@  relocate:
 	 * to ensure the new translations are in use.
 	 */
 	la a0, trampoline_pg_dir
+	XIP_FIXUP_OFFSET a0
 	srl a0, a0, PAGE_SHIFT
 	or a0, a0, a1
 	sfence.vma
@@ -144,7 +171,9 @@  secondary_start_sbi:
 
 	slli a3, a0, LGREG
 	la a4, __cpu_up_stack_pointer
+	XIP_FIXUP_OFFSET a4
 	la a5, __cpu_up_task_pointer
+	XIP_FIXUP_OFFSET a5
 	add a4, a3, a4
 	add a5, a3, a5
 	REG_L sp, (a4)
@@ -156,6 +185,7 @@  secondary_start_common:
 #ifdef CONFIG_MMU
 	/* Enable virtual memory and relocate to virtual address */
 	la a0, swapper_pg_dir
+	XIP_FIXUP_OFFSET a0
 	call relocate
 #endif
 	call setup_trap_vector
@@ -236,12 +266,33 @@  pmp_done:
 .Lgood_cores:
 #endif
 
+#ifndef CONFIG_XIP_KERNEL
 	/* Pick one hart to run the main boot sequence */
 	la a3, hart_lottery
 	li a2, 1
 	amoadd.w a3, a2, (a3)
 	bnez a3, .Lsecondary_start
 
+#else
+	/* hart_lottery in flash contains a magic number */
+	la a3, hart_lottery
+	mv a2, a3
+	XIP_FIXUP_OFFSET a2
+	lw t1, (a3)
+	amoswap.w t0, t1, (a2)
+	/* first time here if hart_lottery in RAM is not set */
+	beq t0, t1, .Lsecondary_start
+
+	la sp, _end + THREAD_SIZE
+	XIP_FIXUP_OFFSET sp
+	mv s0, a0
+	call __copy_data
+
+	/* Restore a0 copy */
+	mv a0, s0
+#endif
+
+#ifndef CONFIG_XIP_KERNEL
 	/* Clear BSS for flat non-ELF images */
 	la a3, __bss_start
 	la a4, __bss_stop
@@ -251,15 +302,18 @@  clear_bss:
 	add a3, a3, RISCV_SZPTR
 	blt a3, a4, clear_bss
 clear_bss_done:
-
+#endif
 	/* Save hart ID and DTB physical address */
 	mv s0, a0
 	mv s1, a1
+
 	la a2, boot_cpu_hartid
+	XIP_FIXUP_OFFSET a2
 	REG_S a0, (a2)
 
 	/* Initialize page tables and relocate to virtual addresses */
 	la sp, init_thread_union + THREAD_SIZE
+	XIP_FIXUP_OFFSET sp
 #ifdef CONFIG_BUILTIN_DTB
 	la a0, __dtb_start
 #else
@@ -268,6 +322,7 @@  clear_bss_done:
 	call setup_vm
 #ifdef CONFIG_MMU
 	la a0, early_pg_dir
+	XIP_FIXUP_OFFSET a0
 	call relocate
 #endif /* CONFIG_MMU */
 
@@ -292,7 +347,9 @@  clear_bss_done:
 
 	slli a3, a0, LGREG
 	la a1, __cpu_up_stack_pointer
+	XIP_FIXUP_OFFSET a1
 	la a2, __cpu_up_task_pointer
+	XIP_FIXUP_OFFSET a2
 	add a1, a3, a1
 	add a2, a3, a2
 
diff --git a/arch/riscv/kernel/head.h b/arch/riscv/kernel/head.h
index b48dda3d04f6..aabbc3ac3e48 100644
--- a/arch/riscv/kernel/head.h
+++ b/arch/riscv/kernel/head.h
@@ -12,6 +12,9 @@  extern atomic_t hart_lottery;
 
 asmlinkage void do_page_fault(struct pt_regs *regs);
 asmlinkage void __init setup_vm(uintptr_t dtb_pa);
+#ifdef CONFIG_XIP_KERNEL
+asmlinkage void __init __copy_data(void);
+#endif
 
 extern void *__cpu_up_stack_pointer[];
 extern void *__cpu_up_task_pointer[];
diff --git a/arch/riscv/kernel/setup.c b/arch/riscv/kernel/setup.c
index e85bacff1b50..29956c41a9cc 100644
--- a/arch/riscv/kernel/setup.c
+++ b/arch/riscv/kernel/setup.c
@@ -50,7 +50,11 @@  struct screen_info screen_info __section(".data") = {
  * This is used before the kernel initializes the BSS so it can't be in the
  * BSS.
  */
-atomic_t hart_lottery __section(".sdata");
+atomic_t hart_lottery __section(".sdata")
+#ifdef CONFIG_XIP_KERNEL
+= ATOMIC_INIT(0xC001BEEF)
+#endif
+;
 unsigned long boot_cpu_hartid;
 static DEFINE_PER_CPU(struct cpu, cpu_devices);
 
@@ -254,7 +258,7 @@  void __init setup_arch(char **cmdline_p)
 #if IS_ENABLED(CONFIG_BUILTIN_DTB)
 	unflatten_and_copy_device_tree();
 #else
-	if (early_init_dt_verify(__va(dtb_early_pa)))
+	if (early_init_dt_verify(dtb_early_va))
 		unflatten_device_tree();
 	else
 		pr_err("No DTB found in kernel mappings\n");
diff --git a/arch/riscv/kernel/vmlinux-xip.lds.S b/arch/riscv/kernel/vmlinux-xip.lds.S
new file mode 100644
index 000000000000..9f0f08c34cd3
--- /dev/null
+++ b/arch/riscv/kernel/vmlinux-xip.lds.S
@@ -0,0 +1,132 @@ 
+/* SPDX-License-Identifier: GPL-2.0-only */
+/*
+ * Copyright (C) 2012 Regents of the University of California
+ * Copyright (C) 2017 SiFive
+ * Copyright (C) 2020 Vitaly Wool, Konsulko AB
+ */
+
+#define LOAD_OFFSET XIP_VIRT_ADDR(CONFIG_XIP_PHYS_ADDR)
+/* No __ro_after_init data in the .rodata section - which will always be ro */
+#define RO_AFTER_INIT_DATA
+
+#include <asm/vmlinux.lds.h>
+#include <asm/page.h>
+#include <asm/pgtable.h>
+#include <asm/cache.h>
+#include <asm/thread_info.h>
+
+OUTPUT_ARCH(riscv)
+ENTRY(_start)
+
+jiffies = jiffies_64;
+
+SECTIONS
+{
+	/* Beginning of code and text segment */
+	. = XIP_VIRT_ADDR(CONFIG_XIP_PHYS_ADDR);
+	_xiprom = .;
+	_start = .;
+	HEAD_TEXT_SECTION
+	INIT_TEXT_SECTION(PAGE_SIZE)
+	/* we have to discard exit text and such at runtime, not link time */
+	.exit.text :
+	{
+		EXIT_TEXT
+	}
+
+	.text : {
+		_text = .;
+		_stext = .;
+		TEXT_TEXT
+		SCHED_TEXT
+		CPUIDLE_TEXT
+		LOCK_TEXT
+		KPROBES_TEXT
+		ENTRY_TEXT
+		IRQENTRY_TEXT
+		SOFTIRQENTRY_TEXT
+		*(.fixup)
+		_etext = .;
+	}
+	RO_DATA(L1_CACHE_BYTES)
+	.srodata : {
+		*(.srodata*)
+	}
+	.init.rodata : {
+		INIT_SETUP(16)
+		INIT_CALLS
+		CON_INITCALL
+		INIT_RAM_FS
+	}
+	_exiprom = .;			/* End of XIP ROM area */
+
+
+/*
+ * From this point, stuff is considered writable and will be copied to RAM
+ */
+	__data_loc = ALIGN(16);		/* location in file */
+	. = PAGE_OFFSET;		/* location in memory */
+
+	_sdata = .;			/* Start of data section */
+	_data = .;
+	RW_DATA(L1_CACHE_BYTES, PAGE_SIZE, THREAD_SIZE)
+	_edata = .;
+	__start_ro_after_init = .;
+	.data.ro_after_init : AT(ADDR(.data.ro_after_init) - LOAD_OFFSET) {
+		*(.data..ro_after_init)
+	}
+	__end_ro_after_init = .;
+
+	. = ALIGN(PAGE_SIZE);
+	__init_begin = .;
+	.init.data : {
+		INIT_DATA
+	}
+	.exit.data : {
+		EXIT_DATA
+	}
+	. = ALIGN(8);
+	__soc_early_init_table : {
+		__soc_early_init_table_start = .;
+		KEEP(*(__soc_early_init_table))
+		__soc_early_init_table_end = .;
+	}
+	__soc_builtin_dtb_table : {
+		__soc_builtin_dtb_table_start = .;
+		KEEP(*(__soc_builtin_dtb_table))
+		__soc_builtin_dtb_table_end = .;
+	}
+	PERCPU_SECTION(L1_CACHE_BYTES)
+
+	. = ALIGN(PAGE_SIZE);
+	__init_end = .;
+
+	.sdata : {
+		__global_pointer$ = . + 0x800;
+		*(.sdata*)
+		*(.sbss*)
+	}
+
+	BSS_SECTION(PAGE_SIZE, PAGE_SIZE, 0)
+	EXCEPTION_TABLE(0x10)
+
+	.rel.dyn : AT(ADDR(.rel.dyn) - LOAD_OFFSET) {
+		*(.rel.dyn*)
+	}
+
+	/*
+	 * End of copied data. We need a dummy section to get its LMA.
+	 * Also located before final ALIGN() as trailing padding is not stored
+	 * in the resulting binary file and useless to copy.
+	 */
+	.data.endmark : AT(ADDR(.data.endmark) - LOAD_OFFSET) { }
+	_edata_loc = LOADADDR(.data.endmark);
+
+	. = ALIGN(PAGE_SIZE);
+	_end = .;
+
+	STABS_DEBUG
+	DWARF_DEBUG
+
+	DISCARDS
+}
diff --git a/arch/riscv/kernel/vmlinux.lds.S b/arch/riscv/kernel/vmlinux.lds.S
index de03cb22d0e9..6745ec325930 100644
--- a/arch/riscv/kernel/vmlinux.lds.S
+++ b/arch/riscv/kernel/vmlinux.lds.S
@@ -4,7 +4,12 @@ 
  * Copyright (C) 2017 SiFive
  */
 
+#ifdef CONFIG_XIP_KERNEL
+#include "vmlinux-xip.lds.S"
+#else
+
 #define LOAD_OFFSET PAGE_OFFSET
+
 #include <asm/vmlinux.lds.h>
 #include <asm/page.h>
 #include <asm/cache.h>
@@ -132,3 +137,4 @@  SECTIONS
 
 	DISCARDS
 }
+#endif /* CONFIG_XIP_KERNEL */
diff --git a/arch/riscv/mm/init.c b/arch/riscv/mm/init.c
index 067583ab1bd7..b5ac455cb159 100644
--- a/arch/riscv/mm/init.c
+++ b/arch/riscv/mm/init.c
@@ -31,8 +31,8 @@  EXPORT_SYMBOL(empty_zero_page);
 
 extern char _start[];
 #define DTB_EARLY_BASE_VA      PGDIR_SIZE
-void *dtb_early_va __initdata;
-uintptr_t dtb_early_pa __initdata;
+void *_dtb_early_va __initdata;
+uintptr_t _dtb_early_pa __initdata;
 
 struct pt_alloc_ops {
 	pte_t *(*get_pte_virt)(phys_addr_t pa);
@@ -88,6 +88,10 @@  static void print_vm_layout(void)
 		  (unsigned long)VMALLOC_END);
 	print_mlm("lowmem", (unsigned long)PAGE_OFFSET,
 		  (unsigned long)high_memory);
+#ifdef CONFIG_XIP_KERNEL
+	print_mlm("xip", (unsigned long)XIP_VIRT_ADDR_START,
+		  (unsigned long)XIP_VIRT_ADDR_START + SZ_16M);
+#endif /* CONFIG_XIP_KERNEL */
 }
 #else
 static void print_vm_layout(void) { }
@@ -113,6 +117,10 @@  void __init setup_bootmem(void)
 	phys_addr_t dram_end = memblock_end_of_DRAM();
 	phys_addr_t max_mapped_addr = __pa(~(ulong)0);
 
+#ifdef CONFIG_XIP_KERNEL
+	vmlinux_start = __pa_symbol(&_sdata);
+#endif
+
 	/* The maximal physical memory size is -PAGE_OFFSET. */
 	memblock_enforce_memory_limit(-PAGE_OFFSET);
 
@@ -148,11 +156,27 @@  void __init setup_bootmem(void)
 	memblock_allow_resize();
 }
 
+#ifdef CONFIG_XIP_KERNEL
+
+extern char _xiprom[], _exiprom[];
+extern char _sdata[], _edata[];
+
+#endif /* CONFIG_XIP_KERNEL */
+
 #ifdef CONFIG_MMU
-static struct pt_alloc_ops pt_ops;
+static struct pt_alloc_ops _pt_ops;
+
+#ifdef CONFIG_XIP_KERNEL
+#define pt_ops (*(struct pt_alloc_ops *)XIP_FIXUP(&_pt_ops))
+#else
+#define pt_ops	_pt_ops
+#endif
 
 unsigned long va_pa_offset;
 EXPORT_SYMBOL(va_pa_offset);
+#ifdef CONFIG_XIP_KERNEL
+#define va_pa_offset	(*((unsigned long *)XIP_FIXUP(&va_pa_offset)))
+#endif
 unsigned long pfn_base;
 EXPORT_SYMBOL(pfn_base);
 
@@ -162,6 +186,12 @@  pte_t fixmap_pte[PTRS_PER_PTE] __page_aligned_bss;
 
 pgd_t early_pg_dir[PTRS_PER_PGD] __initdata __aligned(PAGE_SIZE);
 
+#ifdef CONFIG_XIP_KERNEL
+#define trampoline_pg_dir	((pgd_t *)XIP_FIXUP(trampoline_pg_dir))
+#define fixmap_pte		((pte_t *)XIP_FIXUP(fixmap_pte))
+#define early_pg_dir		((pgd_t *)XIP_FIXUP(early_pg_dir))
+#endif /* CONFIG_XIP_KERNEL */
+
 void __set_fixmap(enum fixed_addresses idx, phys_addr_t phys, pgprot_t prot)
 {
 	unsigned long addr = __fix_to_virt(idx);
@@ -237,6 +267,15 @@  pmd_t fixmap_pmd[PTRS_PER_PMD] __page_aligned_bss;
 pmd_t early_pmd[PTRS_PER_PMD] __initdata __aligned(PAGE_SIZE);
 pmd_t early_dtb_pmd[PTRS_PER_PMD] __initdata __aligned(PAGE_SIZE);
 
+#ifdef CONFIG_XIP_KERNEL
+pmd_t xip_pmd[PTRS_PER_PMD] __page_aligned_bss;
+
+#define trampoline_pmd	((pmd_t *)XIP_FIXUP(trampoline_pmd))
+#define fixmap_pmd	((pmd_t *)XIP_FIXUP(fixmap_pmd))
+#define xip_pmd		((pmd_t *)XIP_FIXUP(xip_pmd))
+#define early_pmd	((pmd_t *)XIP_FIXUP(early_pmd))
+#endif /* CONFIG_XIP_KERNEL */
+
 static pmd_t *__init get_pmd_virt_early(phys_addr_t pa)
 {
 	/* Before MMU is enabled */
@@ -353,6 +392,19 @@  static uintptr_t __init best_map_size(phys_addr_t base, phys_addr_t size)
 	return PMD_SIZE;
 }
 
+#ifdef CONFIG_XIP_KERNEL
+/* called from head.S with MMU off */
+asmlinkage void __init __copy_data(void)
+{
+	void *from = (void *)(&_sdata);
+	void *end = (void *)(&_end);
+	void *to = (void *)CONFIG_XIP_PHYS_RAM_BASE;
+	size_t sz = (size_t)(end - from);
+
+	memcpy(to, from, sz);
+}
+#endif
+
 /*
  * setup_vm() is called from head.S with MMU-off.
  *
@@ -373,7 +425,8 @@  static uintptr_t __init best_map_size(phys_addr_t base, phys_addr_t size)
 
 asmlinkage void __init setup_vm(uintptr_t dtb_pa)
 {
-	uintptr_t va, pa, end_va;
+	uintptr_t va, end_va;
+	uintptr_t __maybe_unused pa;
 	uintptr_t load_pa = (uintptr_t)(&_start);
 	uintptr_t load_sz = (uintptr_t)(&_end) - load_pa;
 	uintptr_t map_size;
@@ -381,6 +434,13 @@  asmlinkage void __init setup_vm(uintptr_t dtb_pa)
 	pmd_t fix_bmap_spmd, fix_bmap_epmd;
 #endif
 
+#ifdef CONFIG_XIP_KERNEL
+	uintptr_t xiprom = (uintptr_t)CONFIG_XIP_PHYS_ADDR;
+	uintptr_t xiprom_sz = (uintptr_t)(&_exiprom) - (uintptr_t)(&_xiprom);
+
+	load_pa = (uintptr_t)CONFIG_XIP_PHYS_RAM_BASE;
+	load_sz = (uintptr_t)(&_end) - (uintptr_t)(&_sdata);
+#endif
 	va_pa_offset = PAGE_OFFSET - load_pa;
 	pfn_base = PFN_DOWN(load_pa);
 
@@ -419,6 +479,21 @@  asmlinkage void __init setup_vm(uintptr_t dtb_pa)
 			   load_pa, PGDIR_SIZE, PAGE_KERNEL_EXEC);
 #endif
 
+#ifdef CONFIG_XIP_KERNEL
+	create_pgd_mapping(trampoline_pg_dir, XIP_VIRT_ADDR_START,
+			   (uintptr_t)xip_pmd, PGDIR_SIZE, PAGE_TABLE);
+	for (va = XIP_VIRT_ADDR_START;
+	     va < XIP_VIRT_ADDR_START + xiprom_sz;
+	     va += PMD_SIZE) {
+		create_pmd_mapping(xip_pmd, va,
+				   xiprom + (va - XIP_VIRT_ADDR_START),
+				   PMD_SIZE, PAGE_KERNEL_EXEC);
+	}
+
+	create_pgd_mapping(early_pg_dir, XIP_VIRT_ADDR_START,
+			   (uintptr_t)xip_pmd, PGDIR_SIZE, PAGE_TABLE);
+#endif
+
 	/*
 	 * Setup early PGD covering entire kernel which will allows
 	 * us to reach paging_init(). We map all memory banks later
@@ -443,7 +518,7 @@  asmlinkage void __init setup_vm(uintptr_t dtb_pa)
 			   pa + PMD_SIZE, PMD_SIZE, PAGE_KERNEL);
 	dtb_early_va = (void *)DTB_EARLY_BASE_VA + (dtb_pa & (PMD_SIZE - 1));
 #else /* CONFIG_BUILTIN_DTB */
-	dtb_early_va = __va(dtb_pa);
+	dtb_early_va = __va(XIP_FIXUP(dtb_pa));
 #endif /* CONFIG_BUILTIN_DTB */
 #else
 #ifndef CONFIG_BUILTIN_DTB
@@ -455,7 +530,7 @@  asmlinkage void __init setup_vm(uintptr_t dtb_pa)
 			   pa + PGDIR_SIZE, PGDIR_SIZE, PAGE_KERNEL);
 	dtb_early_va = (void *)DTB_EARLY_BASE_VA + (dtb_pa & (PGDIR_SIZE - 1));
 #else /* CONFIG_BUILTIN_DTB */
-	dtb_early_va = __va(dtb_pa);
+	dtb_early_va = __va(XIP_FIXUP(dtb_pa));
 #endif /* CONFIG_BUILTIN_DTB */
 #endif
 	dtb_early_pa = dtb_pa;
@@ -496,6 +571,9 @@  static void __init setup_vm_final(void)
 	uintptr_t va, map_size;
 	phys_addr_t pa, start, end;
 	u64 i;
+#ifdef CONFIG_XIP_KERNEL
+	uintptr_t xiprom_sz = (uintptr_t)(&_exiprom) - (uintptr_t)(&_xiprom);
+#endif
 
 	/**
 	 * MMU is enabled at this point. But page table setup is not complete yet.
@@ -527,6 +605,16 @@  static void __init setup_vm_final(void)
 					   map_size, PAGE_KERNEL_EXEC);
 		}
 	}
+#ifdef CONFIG_XIP_KERNEL
+	map_size = best_map_size(CONFIG_XIP_PHYS_ADDR, xiprom_sz);
+	for (va = XIP_VIRT_ADDR_START;
+	     va < XIP_VIRT_ADDR_START + xiprom_sz;
+	     va += map_size)
+		create_pgd_mapping(swapper_pg_dir, va,
+				   CONFIG_XIP_PHYS_ADDR + (va - XIP_VIRT_ADDR_START),
+				   map_size, PAGE_KERNEL_EXEC);
+
+#endif
 
 	/* Clear fixmap PTE and PMD mappings */
 	clear_fixmap(FIX_PTE);