public inbox for kvm@vger.kernel.org
 help / color / mirror / Atom feed
From: Lai Jiangshan <jiangshanlai@gmail.com>
To: linux-kernel@vger.kernel.org
Cc: Hou Wenlong <houwenlong.hwl@antgroup.com>,
	Lai Jiangshan <jiangshan.ljs@antgroup.com>,
	Linus Torvalds <torvalds@linux-foundation.org>,
	Peter Zijlstra <peterz@infradead.org>,
	Sean Christopherson <seanjc@google.com>,
	Thomas Gleixner <tglx@linutronix.de>,
	Borislav Petkov <bp@alien8.de>, Ingo Molnar <mingo@redhat.com>,
	kvm@vger.kernel.org, Paolo Bonzini <pbonzini@redhat.com>,
	x86@kernel.org, Kees Cook <keescook@chromium.org>,
	Juergen Gross <jgross@suse.com>,
	Dave Hansen <dave.hansen@linux.intel.com>,
	"H. Peter Anvin" <hpa@zytor.com>,
	Petr Pavlu <petr.pavlu@suse.com>,
	Josh Poimboeuf <jpoimboe@kernel.org>,
	Nick Desaulniers <ndesaulniers@google.com>
Subject: [RFC PATCH 52/73] x86/boot: Allow to do relocation for uncompressed kernel
Date: Mon, 26 Feb 2024 22:36:09 +0800	[thread overview]
Message-ID: <20240226143630.33643-53-jiangshanlai@gmail.com> (raw)
In-Reply-To: <20240226143630.33643-1-jiangshanlai@gmail.com>

From: Hou Wenlong <houwenlong.hwl@antgroup.com>

Relocation is currently only performed during the uncompression process.
However, in some situations, such as with security containers, the
uncompressed kernel can be booted directly. Therefore, it is useful to
allow for relocation of the uncompressed kernel. Taking inspiration from
the implementation in MIPS, a new section named ".data.relocs" is
reserved for relocations. The relocs tool can then append the
relocations into this section. Additionally, a helper function is
introduced to perform relocations during booting, similar to the
relocations in the bootloader. For PVH entry, relocation for the
pre-constructed page table should not be performed; otherwise, booting
will fail.

Signed-off-by: Hou Wenlong <houwenlong.hwl@antgroup.com>
Signed-off-by: Lai Jiangshan <jiangshan.ljs@antgroup.com>
---
 arch/x86/Kconfig                  | 20 +++++++++
 arch/x86/Makefile.postlink        |  9 +++-
 arch/x86/kernel/head64_identity.c | 70 +++++++++++++++++++++++++++++++
 arch/x86/kernel/vmlinux.lds.S     | 18 ++++++++
 4 files changed, 116 insertions(+), 1 deletion(-)

diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig
index a53b65499951..d02ef3bdb171 100644
--- a/arch/x86/Kconfig
+++ b/arch/x86/Kconfig
@@ -2183,6 +2183,26 @@ config RELOCATABLE
 	  it has been loaded at and the compile time physical address
 	  (CONFIG_PHYSICAL_START) is used as the minimum location.
 
+config RELOCATABLE_UNCOMPRESSED_KERNEL
+	bool
+	depends on RELOCATABLE
+	help
+	  A table of relocation data will be appended to the uncompressed
+	  kernel binary and parsed at boot to fix up the relocated kernel.
+
+config RELOCATION_TABLE_SIZE
+	hex "Relocation table size"
+	depends on RELOCATABLE_UNCOMPRESSED_KERNEL
+	range 0x0 0x01000000
+	default "0x00200000"
+	help
+	  This option allows the amount of space reserved for the table to be
+	  adjusted, although the default of 1Mb should be ok in most cases.
+
+	  The build will fail and a valid size suggested if this is too small.
+
+	  If unsure, leave at the default value.
+
 config X86_PIE
 	bool "Build a PIE kernel"
 	default n
diff --git a/arch/x86/Makefile.postlink b/arch/x86/Makefile.postlink
index fef2e977cc7d..c115692b67b2 100644
--- a/arch/x86/Makefile.postlink
+++ b/arch/x86/Makefile.postlink
@@ -4,7 +4,8 @@
 # ===========================================================================
 #
 # 1. Separate relocations from vmlinux into vmlinux.relocs.
-# 2. Strip relocations from vmlinux.
+# 2. Insert relocations table into vmlinux
+# 3. Strip relocations from vmlinux.
 
 PHONY := __archpost
 __archpost:
@@ -20,6 +21,9 @@ quiet_cmd_relocs = RELOCS  $(OUT_RELOCS)/$@.relocs
 	$(CMD_RELOCS) $@ > $(OUT_RELOCS)/$@.relocs; \
 	$(CMD_RELOCS) --abs-relocs $@
 
+quiet_cmd_insert_relocs = RELOCS  $@
+      cmd_insert_relocs = $(CMD_RELOCS) --keep $@
+
 quiet_cmd_strip_relocs = RSTRIP  $@
       cmd_strip_relocs = \
 	$(OBJCOPY) --remove-section='.rel.*' --remove-section='.rel__*' \
@@ -29,6 +33,9 @@ quiet_cmd_strip_relocs = RSTRIP  $@
 
 vmlinux: FORCE
 	@true
+ifeq ($(CONFIG_RELOCATABLE_UNCOMPRESSED_KERNEL),y)
+	$(call cmd,insert_relocs)
+endif
 ifeq ($(CONFIG_X86_NEED_RELOCS),y)
 	$(call cmd,relocs)
 	$(call cmd,strip_relocs)
diff --git a/arch/x86/kernel/head64_identity.c b/arch/x86/kernel/head64_identity.c
index ecac6e704868..4548ad615ecf 100644
--- a/arch/x86/kernel/head64_identity.c
+++ b/arch/x86/kernel/head64_identity.c
@@ -315,3 +315,73 @@ void __head startup_64_setup_env(void)
 
 	startup_64_load_idt();
 }
+
+#ifdef CONFIG_RELOCATABLE_UNCOMPRESSED_KERNEL
+extern u8 __relocation_end[];
+
+static bool __head is_in_pvh_pgtable(unsigned long ptr)
+{
+#ifdef CONFIG_PVH
+	if (ptr >= (unsigned long)init_top_pgt &&
+	    ptr < (unsigned long)init_top_pgt + PAGE_SIZE)
+		return true;
+	if (ptr >= (unsigned long)level3_ident_pgt &&
+	    ptr < (unsigned long)level3_ident_pgt + PAGE_SIZE)
+		return true;
+#endif
+	return false;
+}
+
+void __head __relocate_kernel(unsigned long physbase, unsigned long virtbase)
+{
+	int *reloc = (int *)__relocation_end;
+	unsigned long ptr;
+	unsigned long delta = virtbase - __START_KERNEL_map;
+	unsigned long map = physbase - __START_KERNEL;
+	long extended;
+
+	/*
+	 * Relocation had happended in bootloader,
+	 * don't do it again.
+	 */
+	if (SYM_ABS_VA(_text) != __START_KERNEL)
+		return;
+
+	if (!delta)
+		return;
+
+	/*
+	 * Format is:
+	 *
+	 * kernel bits...
+	 * 0 - zero terminator for 64 bit relocations
+	 * 64 bit relocation repeated
+	 * 0 - zero terminator for inverse 32 bit relocations
+	 * 32 bit inverse relocation repeated
+	 * 0 - zero terminator for 32 bit relocations
+	 * 32 bit relocation repeated
+	 *
+	 * So we work backwards from the end of .data.relocs section, see
+	 * handle_relocations() in arch/x86/boot/compressed/misc.c.
+	 */
+	while (*--reloc) {
+		extended = *reloc;
+		ptr = (unsigned long)(extended + map);
+		*(uint32_t *)ptr += delta;
+	}
+
+	while (*--reloc) {
+		extended = *reloc;
+		ptr = (unsigned long)(extended + map);
+		*(int32_t *)ptr -= delta;
+	}
+
+	while (*--reloc) {
+		extended = *reloc;
+		ptr = (unsigned long)(extended + map);
+		if (is_in_pvh_pgtable(ptr))
+			continue;
+		*(uint64_t *)ptr += delta;
+	}
+}
+#endif
diff --git a/arch/x86/kernel/vmlinux.lds.S b/arch/x86/kernel/vmlinux.lds.S
index 834c68b45f15..3b05807fe1dc 100644
--- a/arch/x86/kernel/vmlinux.lds.S
+++ b/arch/x86/kernel/vmlinux.lds.S
@@ -339,6 +339,24 @@ SECTIONS
 	}
 #endif
 
+#ifdef CONFIG_RELOCATABLE_UNCOMPRESSED_KERNEL
+	. = ALIGN(4);
+	.data.reloc : AT(ADDR(.data.reloc) - LOAD_OFFSET) {
+		__relocation_start = .;
+		/*
+		 * Space for relocation table
+		 * This needs to be filled so that the
+		 * relocs tool can overwrite the content.
+		 * Put a dummy data item at the start to
+		 * avoid to generate NOBITS section.
+		 */
+		LONG(0);
+		FILL(0);
+		. += CONFIG_RELOCATION_TABLE_SIZE - 4;
+		__relocation_end = .;
+	}
+#endif
+
 	/*
 	 * struct alt_inst entries. From the header (alternative.h):
 	 * "Alternative instructions for different CPU types or capabilities"
-- 
2.19.1.6.gb485710b


  parent reply	other threads:[~2024-02-26 14:37 UTC|newest]

Thread overview: 82+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-02-26 14:35 [RFC PATCH 00/73] KVM: x86/PVM: Introduce a new hypervisor Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 01/73] KVM: Documentation: Add the specification for PVM Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 02/73] x86/ABI/PVM: Add PVM-specific ABI header file Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 03/73] x86/entry: Implement switcher for PVM VM enter/exit Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 04/73] x86/entry: Implement direct switching for the switcher Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 05/73] KVM: x86: Set 'vcpu->arch.exception.injected' as true before vendor callback Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 06/73] KVM: x86: Move VMX interrupt/nmi handling into kvm.ko Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 07/73] KVM: x86/mmu: Adapt shadow MMU for PVM Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 08/73] KVM: x86: Allow hypercall handling to not skip the instruction Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 09/73] KVM: x86: Add PVM virtual MSRs into emulated_msrs_all[] Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 10/73] KVM: x86: Introduce vendor feature to expose vendor-specific CPUID Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 11/73] KVM: x86: Implement gpc refresh for guest usage Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 12/73] KVM: x86: Add NR_VCPU_SREG in SREG enum Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 13/73] KVM: x86/emulator: Reinject #GP if instruction emulation failed for PVM Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 14/73] KVM: x86: Create stubs for PVM module as a new vendor Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 15/73] mm/vmalloc: Add a helper to reserve a contiguous and aligned kernel virtual area Lai Jiangshan
2024-02-27 14:56   ` Christoph Hellwig
2024-02-27 17:07     ` Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 16/73] KVM: x86/PVM: Implement host mmu initialization Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 17/73] KVM: x86/PVM: Implement module initialization related callbacks Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 18/73] KVM: x86/PVM: Implement VM/VCPU " Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 19/73] x86/entry: Export 32-bit ignore syscall entry and __ia32_enabled variable Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 20/73] KVM: x86/PVM: Implement vcpu_load()/vcpu_put() related callbacks Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 21/73] KVM: x86/PVM: Implement vcpu_run() callbacks Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 22/73] KVM: x86/PVM: Handle some VM exits before enable interrupts Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 23/73] KVM: x86/PVM: Handle event handling related MSR read/write operation Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 24/73] KVM: x86/PVM: Introduce PVM mode switching Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 25/73] KVM: x86/PVM: Implement APIC emulation related callbacks Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 26/73] KVM: x86/PVM: Implement event delivery flags " Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 27/73] KVM: x86/PVM: Implement event injection " Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 28/73] KVM: x86/PVM: Handle syscall from user mode Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 29/73] KVM: x86/PVM: Implement allowed range checking for #PF Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 30/73] KVM: x86/PVM: Implement segment related callbacks Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 31/73] KVM: x86/PVM: Implement instruction emulation for #UD and #GP Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 32/73] KVM: x86/PVM: Enable guest debugging functions Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 33/73] KVM: x86/PVM: Handle VM-exit due to hardware exceptions Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 34/73] KVM: x86/PVM: Handle ERETU/ERETS synthetic instruction Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 35/73] KVM: x86/PVM: Handle PVM_SYNTHETIC_CPUID " Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 36/73] KVM: x86/PVM: Handle KVM hypercall Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 37/73] KVM: x86/PVM: Use host PCID to reduce guest TLB flushing Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 38/73] KVM: x86/PVM: Handle hypercalls for privilege instruction emulation Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 39/73] KVM: x86/PVM: Handle hypercall for CR3 switching Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 40/73] KVM: x86/PVM: Handle hypercall for loading GS selector Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 41/73] KVM: x86/PVM: Allow to load guest TLS in host GDT Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 42/73] KVM: x86/PVM: Support for kvm_exit() tracepoint Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 43/73] KVM: x86/PVM: Enable direct switching Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 44/73] KVM: x86/PVM: Implement TSC related callbacks Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 45/73] KVM: x86/PVM: Add dummy PMU " Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 46/73] KVM: x86/PVM: Support for CPUID faulting Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 47/73] KVM: x86/PVM: Handle the left supported MSRs in msrs_to_save_base[] Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 48/73] KVM: x86/PVM: Implement system registers setting callbacks Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 49/73] KVM: x86/PVM: Implement emulation for non-PVM mode Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 50/73] x86/tools/relocs: Cleanup cmdline options Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 51/73] x86/tools/relocs: Append relocations into input file Lai Jiangshan
2024-02-26 14:36 ` Lai Jiangshan [this message]
2024-02-26 14:36 ` [RFC PATCH 53/73] x86/pvm: Add Kconfig option and the CPU feature bit for PVM guest Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 54/73] x86/pvm: Detect PVM hypervisor support Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 55/73] x86/pvm: Relocate kernel image to specific virtual address range Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 56/73] x86/pvm: Relocate kernel image early in PVH entry Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 57/73] x86/pvm: Make cpu entry area and vmalloc area variable Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 58/73] x86/pvm: Relocate kernel address space layout Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 59/73] x86/pti: Force enabling KPTI for PVM guest Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 60/73] x86/pvm: Add event entry/exit and dispatch code Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 61/73] x86/pvm: Allow to install a system interrupt handler Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 62/73] x86/pvm: Add early kernel event entry and dispatch code Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 63/73] x86/pvm: Add hypercall support Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 64/73] x86/pvm: Enable PVM event delivery Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 65/73] x86/kvm: Patch KVM hypercall as PVM hypercall Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 66/73] x86/pvm: Use new cpu feature to describe XENPV and PVM Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 67/73] x86/pvm: Implement cpu related PVOPS Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 68/73] x86/pvm: Implement irq " Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 69/73] x86/pvm: Implement mmu " Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 70/73] x86/pvm: Don't use SWAPGS for gsbase read/write Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 71/73] x86/pvm: Adapt pushf/popf in this_cpu_cmpxchg16b_emu() Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 72/73] x86/pvm: Use RDTSCP as default in vdso_read_cpunode() Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 73/73] x86/pvm: Disable some unsupported syscalls and features Lai Jiangshan
2024-02-26 14:49 ` [RFC PATCH 00/73] KVM: x86/PVM: Introduce a new hypervisor Paolo Bonzini
2024-02-27 17:27   ` Sean Christopherson
2024-02-29  9:33     ` David Woodhouse
2024-03-01 14:00     ` Lai Jiangshan
2024-02-29 14:55   ` Lai Jiangshan
2024-03-06 11:05 ` Like Xu

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20240226143630.33643-53-jiangshanlai@gmail.com \
    --to=jiangshanlai@gmail.com \
    --cc=bp@alien8.de \
    --cc=dave.hansen@linux.intel.com \
    --cc=houwenlong.hwl@antgroup.com \
    --cc=hpa@zytor.com \
    --cc=jgross@suse.com \
    --cc=jiangshan.ljs@antgroup.com \
    --cc=jpoimboe@kernel.org \
    --cc=keescook@chromium.org \
    --cc=kvm@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mingo@redhat.com \
    --cc=ndesaulniers@google.com \
    --cc=pbonzini@redhat.com \
    --cc=peterz@infradead.org \
    --cc=petr.pavlu@suse.com \
    --cc=seanjc@google.com \
    --cc=tglx@linutronix.de \
    --cc=torvalds@linux-foundation.org \
    --cc=x86@kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox