x86/startup: Let the caller retrieve encryption mask
PATCH 1/1 x86/startup: Let the caller retrieve encryption mask
From: Khalid Ali <[email protected]>
Don't return encryption mask . This makes __startup_64() to handle
doing the actual work including encrypting the kernel. The caller has already access to encryption mask and can directly retrieve it. On C
code, the caller can call sme_get_me_mask() and include /arch/x86/include/asm/mem_encrypt.h directly while on assembly functions like startup_64 the "sme_me_mask"
is directly accessible to them if CONFIG_AMD_MEM_ENCRYPT is set. This
also makes consistent with the way secondary_startup_64_no_verify label
is handled. On intel CPUs this is not even neccessary, so we should
retrieve the mask only if CONFIG_AMD_MEM_ENCRYPT is set.
--- arch/x86/boot/startup/map_kernel.c | 11 +++-------- arch/x86/include/asm/setup.h | 2 +- arch/x86/kernel/head_64.S | 12 +++++------- 3 files changed, 9 insertions(+), 16 deletions(-)
arch/x86/boot/startup/map_kernel.c +3 -8
@@ -30,7 +30,7 @@ static inline bool check_la57_support(void)return true; } -static unsigned long __head sme_postprocess_startup (struct boot_params *bp, +static void __head sme_postprocess_startup(struct boot_params *bp,pmdval_t *pmd,unsigned long p2v_offset) {
@@ -68,11 +68,6 @@ static unsigned long __head sme_postprocess_startup(struct boot_params *bp, } } -/* -* Return the SME encryption mask (if SME is active) to be used as a -* modifier for the initial pgdir entry programmed into CR3. -*/ -return sme_get_me_mask (); }/*
@@ -84,7 +79,7 @@ static unsigned long __head sme_postprocess_startup(struct boot_params *bp,* the 1:1 mapping of memory. Kernel virtual addresses can be determined by * subtracting p2v_offset from the RIP-relative address. */ -unsigned long __head __startup_64 (unsigned long p2v_offset, +void __head __startup_64(unsigned long p2v_offset,struct boot_params *bp) {pmd_t (*early_pgts)[PTRS_PER_PMD ] =rip_rel_ptr (early_dynamic_pgts);
@@ -213,5 +208,5 @@ unsigned long __head __startup_64(unsigned long p2v_offset,for (; i <PTRS_PER_PMD ; i++) pmd[i] &= ~_PAGE_PRESENT; -return sme_postprocess_startup (bp, pmd, p2v_offset); +sme_postprocess_startup (bp, pmd, p2v_offset); }
arch/x86/include/asm/setup.h +1 -1
@@ -50,7 +50,7 @@ extern unsigned long acpi_realmode_flags;extern void reserve_standard_io_resources (void );extern void i386_reserve_resources (void ); -extern unsigned long __startup_64 (unsigned long p2v_offset,struct boot_params *bp); +extern void __startup_64 (unsigned long p2v_offset,struct boot_params *bp);extern void startup_64_setup_gdt_idt (void );extern void startup_64_load_idt (void *vc_handler);extern void early_setup_idt (void );
arch/x86/kernel/head_64.S +5 -7
@@ -106,18 +106,16 @@ SYM_CODE_START_NOALIGN(startup_64) /* * Perform pagetable fixups. Additionally, if SME is active, encrypt - * the kernel and retrieve the modifier (SME encryption mask if SME - * is active) to be added to the initial pgdir entry that will be - * programmed into CR3. - */ + * the kernel. + / movq %r15, %rsi call __startup_64 /* Form the CR3 value being sure to include the CR3 modifier */ - leaq early_top_pgt(%rip), %rcx - addq %rcx, %rax - + leaq early_top_pgt(%rip), %rax + #ifdef CONFIG_AMD_MEM_ENCRYPT + addq sme_me_mask(%rip), %rax mov %rax, %rdi /*
Signature · 1 line
2.49.0
Borislav Petkov
maintainerQuoted 11 lines · Khalid Ali
From: Khalid Ali <[email protected]>
Don't return encryption mask . This makes
__startup_64()to handle
doing the actual work including encrypting the kernel. The caller has already access to encryption mask and can directly retrieve it. On C
code, the caller can callsme_get_me_mask()and include /arch/x86/include/asm/mem_encrypt.h directly while on assembly functions likestartup_64the "sme_me_mask"
is directly accessible to them ifCONFIG_AMD_MEM_ENCRYPTis set. This
also makes consistent with the waysecondary_startup_64_no_verifylabel
is handled. On intel CPUs this is not even neccessary, so we should
retrieve the mask only ifCONFIG_AMD_MEM_ENCRYPTis set.
What Thomas told you about structuring commit messages:
https://lore.kernel.org/r/875xgziprs.ffs@tglx
Do it please and then send patches.
Thx.
Signature · 4 lines
Regards/Gruss,
Boris.
https://people.kernel.org/tglx/notes-about-netiquetteQuoted 7 lines
What Thomas told you about structuring commit messages:
https://lore.kernel.org/r/875xgziprs.ffs@tglx
Do it please and then send patches.
Thx.
I apologies, i was aware, however English isn't my native language. I was grasping the doc as best i could. I sent v2 already.
Thanks
Khalid Ali
kernel test robot
build botkernel test robot noticed the following build warnings:
Hi Khalid,
kernel test robot noticed the following build warnings:
[auto build test WARNING on tip/x86/core]
[also build test WARNING on tip/master linus/master v6.16-rc2 next-20250616] [cannot apply to tip/auto-latest]
[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#_base_tree_information]
url: https://github.com/intel-lab-lkp/linux/commits/Khalid-Ali/x86-startup-Let-the-caller-retrieve-encryption-mask/20250616-204024 base: tip/x86/core patch link: https://lore.kernel.org/r/20250616123605.927-1-khaliidcaliy%40gmail.com patch subject: [PATCH] x86/startup: Let the caller retrieve encryption mask config: x86_64-buildonly-randconfig-001-20250617 (https://download.01.org/0day-ci/archive/20250617/[email protected]/config) compiler: clang version 20.1.2 (https://github.com/llvm/llvm-project 58df0ef89dd64126512e4ee27b4ac3fd8ddf6247) reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20250617/[email protected]/reproduce)
If you fix the issue in a separate patch/commit (i.e. not just a new version of the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <[email protected]>
| Closes: https://lore.kernel.org/oe-kbuild-all/[email protected]/
All warnings (new ones prefixed by >>):
>> arch/x86/kernel/head_64.S:114:2: warning: '/*' within block comment [-Wcomment]
114 | /* Form the CR3 value being sure to include the CR3 modifier */
| ^1 warning generated.
vim +114 arch/x86/kernel/head_64.S
bcce829083339b arch/x86/kernel/head_64.S Michael Roth 2022-02-09 96 04633df0c43d71 arch/x86/kernel/head_64.S Borislav Petkov 2015-11-05 97 /* Sanitize CPU configuration */ 04633df0c43d71 arch/x86/kernel/head_64.S Borislav Petkov 2015-11-05 98 call verify_cpu 04633df0c43d71 arch/x86/kernel/head_64.S Borislav Petkov 2015-11-05 99 093562198e1a63 arch/x86/kernel/head_64.S Ard Biesheuvel 2024-12-05 100 /* 093562198e1a63 arch/x86/kernel/head_64.S Ard Biesheuvel 2024-12-05 101 * Derive the kernel's physical-to-virtual offset from the physical and 093562198e1a63 arch/x86/kernel/head_64.S Ard Biesheuvel 2024-12-05 102 * virtual addresses of common_startup_64(). 093562198e1a63 arch/x86/kernel/head_64.S Ard Biesheuvel 2024-12-05 103 */ 093562198e1a63 arch/x86/kernel/head_64.S Ard Biesheuvel 2024-12-05 104 leaq common_startup_64(%rip), %rdi 093562198e1a63 arch/x86/kernel/head_64.S Ard Biesheuvel 2024-12-05 105 subq .Lcommon_startup_64(%rip), %rdi 093562198e1a63 arch/x86/kernel/head_64.S Ard Biesheuvel 2024-12-05 106 5868f3651fa0df arch/x86/kernel/head_64.S Tom Lendacky 2017-07-17 107 /* 5868f3651fa0df arch/x86/kernel/head_64.S Tom Lendacky 2017-07-17 108 * Perform pagetable fixups. Additionally, if SME is active, encrypt 8b4b3620a6b25c arch/x86/kernel/head_64.S Khalid Ali 2025-06-16 109 * the kernel. 8b4b3620a6b25c arch/x86/kernel/head_64.S Khalid Ali 2025-06-16 110 / 2f69a81ad68732 arch/x86/kernel/head_64.S Ard Biesheuvel 2023-08-07 111 movq %r15, %rsi c88d71508e36b5 arch/x86/kernel/head_64.S Kirill A. Shutemov 2017-06-06 112 call __startup_64 ^1da177e4c3f41 arch/x86_64/kernel/head.S Linus Torvalds 2005-04-16 113 5868f3651fa0df arch/x86/kernel/head_64.S Tom Lendacky 2017-07-17 @114 /* Form the CR3 value being sure to include the CR3 modifier */ 8b4b3620a6b25c arch/x86/kernel/head_64.S Khalid Ali 2025-06-16 115 leaq early_top_pgt(%rip), %rax 30579c8baa5b4b arch/x86/kernel/head_64.S Borislav Petkov (AMD 2023-11-30 116) 30579c8baa5b4b arch/x86/kernel/head_64.S Borislav Petkov (AMD 2023-11-30 117) #ifdef CONFIG_AMD_MEM_ENCRYPT 8b4b3620a6b25c arch/x86/kernel/head_64.S Khalid Ali 2025-06-16 118 addq sme_me_mask(%rip), %rax 30579c8baa5b4b arch/x86/kernel/head_64.S Borislav Petkov (AMD 2023-11-30 119) mov %rax, %rdi 30579c8baa5b4b arch/x86/kernel/head_64.S Borislav Petkov (AMD 2023-11-30 120) 30579c8baa5b4b arch/x86/kernel/head_64.S Borislav Petkov (AMD 2023-11-30 121) /* 30579c8baa5b4b arch/x86/kernel/head_64.S Borislav Petkov (AMD 2023-11-30 122) * For SEV guests: Verify that the C-bit is correct. A malicious 30579c8baa5b4b arch/x86/kernel/head_64.S Borislav Petkov (AMD 2023-11-30 123) * hypervisor could lie about the C-bit position to perform a ROP 30579c8baa5b4b arch/x86/kernel/head_64.S Borislav Petkov (AMD 2023-11-30 124) * attack on the guest by writing to the unencrypted stack and wait for 30579c8baa5b4b arch/x86/kernel/head_64.S Borislav Petkov (AMD 2023-11-30 125) * the next RET instruction. 30579c8baa5b4b arch/x86/kernel/head_64.S Borislav Petkov (AMD 2023-11-30 126) */ 30579c8baa5b4b arch/x86/kernel/head_64.S Borislav Petkov (AMD 2023-11-30 127) call sev_verify_cbit 30579c8baa5b4b arch/x86/kernel/head_64.S Borislav Petkov (AMD 2023-11-30 128) #endif 30579c8baa5b4b arch/x86/kernel/head_64.S Borislav Petkov (AMD 2023-11-30 129) 828263957611c2 arch/x86/kernel/head_64.S Ard Biesheuvel 2024-02-27 130 /* 828263957611c2 arch/x86/kernel/head_64.S Ard Biesheuvel 2024-02-27 131 * Switch to early_top_pgt which still has the identity mappings 828263957611c2 arch/x86/kernel/head_64.S Ard Biesheuvel 2024-02-27 132 * present. 828263957611c2 arch/x86/kernel/head_64.S Ard Biesheuvel 2024-02-27 133 */ 828263957611c2 arch/x86/kernel/head_64.S Ard Biesheuvel 2024-02-27 134 movq %rax, %cr3 828263957611c2 arch/x86/kernel/head_64.S Ard Biesheuvel 2024-02-27 135 828263957611c2 arch/x86/kernel/head_64.S Ard Biesheuvel 2024-02-27 136 /* Branch to the common startup code at its kernel virtual address */ 828263957611c2 arch/x86/kernel/head_64.S Ard Biesheuvel 2024-02-27 137 ANNOTATE_RETPOLINE_SAFE 093562198e1a63 arch/x86/kernel/head_64.S Ard Biesheuvel 2024-12-05 138 jmp *.Lcommon_startup_64(%rip) 37818afd15fe72 arch/x86/kernel/head_64.S Jiri Slaby 2019-10-11 139 SYM_CODE_END(startup_64) 37818afd15fe72 arch/x86/kernel/head_64.S Jiri Slaby 2019-10-11 140 828263957611c2 arch/x86/kernel/head_64.S Ard Biesheuvel 2024-02-27 141 __INITRODATA 093562198e1a63 arch/x86/kernel/head_64.S Ard Biesheuvel 2024-12-05 142 SYM_DATA_LOCAL(.Lcommon_startup_64, .quad common_startup_64) 828263957611c2 arch/x86/kernel/head_64.S Ard Biesheuvel 2024-02-27 143 828263957611c2 arch/x86/kernel/head_64.S Ard Biesheuvel 2024-02-27 144 .text bc7b11c04ee9c9 arch/x86/kernel/head_64.S Jiri Slaby 2019-10-11 145 SYM_CODE_START(secondary_startup_64) fb799447ae2974 arch/x86/kernel/head_64.S Josh Poimboeuf 2023-03-01 146 UNWIND_HINT_END_OF_STACK 3e3f069504344c arch/x86/kernel/head_64.S Peter Zijlstra 2022-03-08 147 ANNOTATE_NOENDBR 1ab60e0f72f71e arch/x86_64/kernel/head.S Vivek Goyal 2007-05-02 148 /* 1256276c98dbcf arch/x86/kernel/head_64.S Konrad Rzeszutek Wilk 2013-02-25 149 * At this point the CPU runs in 64bit mode CS.L = 1 CS.D = 0, 1ab60e0f72f71e arch/x86_64/kernel/head.S Vivek Goyal 2007-05-02 150 * and someone has loaded a mapped page table. 1ab60e0f72f71e arch/x86_64/kernel/head.S Vivek Goyal 2007-05-02 151 * 1ab60e0f72f71e arch/x86_64/kernel/head.S Vivek Goyal 2007-05-02 152 * We come here either from startup_64 (using physical addresses) 1ab60e0f72f71e arch/x86_64/kernel/head.S Vivek Goyal 2007-05-02 153 * or from trampoline.S (using virtual addresses). 1ab60e0f72f71e arch/x86_64/kernel/head.S Vivek Goyal 2007-05-02 154 * 1ab60e0f72f71e arch/x86_64/kernel/head.S Vivek Goyal 2007-05-02 155 * Using virtual addresses from trampoline.S removes the need 1ab60e0f72f71e arch/x86_64/kernel/head.S Vivek Goyal 2007-05-02 156 * to have any identity mapped pages in the kernel page table 1ab60e0f72f71e arch/x86_64/kernel/head.S Vivek Goyal 2007-05-02 157 * after the boot processor executes this code. ^1da177e4c3f41 arch/x86_64/kernel/head.S Linus Torvalds 2005-04-16 158 */ ^1da177e4c3f41 arch/x86_64/kernel/head.S Linus Torvalds 2005-04-16 159 04633df0c43d71 arch/x86/kernel/head_64.S Borislav Petkov 2015-11-05 160 /* Sanitize CPU configuration */ 04633df0c43d71 arch/x86/kernel/head_64.S Borislav Petkov 2015-11-05 161 call verify_cpu 04633df0c43d71 arch/x86/kernel/head_64.S Borislav Petkov 2015-11-05 162 3ecacdbd23956a arch/x86/kernel/head_64.S Joerg Roedel 2020-09-07 163 /* 3ecacdbd23956a arch/x86/kernel/head_64.S Joerg Roedel 2020-09-07 164 * The secondary_startup_64_no_verify entry point is only used by 3ecacdbd23956a arch/x86/kernel/head_64.S Joerg Roedel 2020-09-07 165 * SEV-ES guests. In those guests the call to verify_cpu() would cause 3ecacdbd23956a arch/x86/kernel/head_64.S Joerg Roedel 2020-09-07 166 * #VC exceptions which can not be handled at this stage of secondary 3ecacdbd23956a arch/x86/kernel/head_64.S Joerg Roedel 2020-09-07 167 * CPU bringup. 3ecacdbd23956a arch/x86/kernel/head_64.S Joerg Roedel 2020-09-07 168 * 3ecacdbd23956a arch/x86/kernel/head_64.S Joerg Roedel 2020-09-07 169 * All non SEV-ES systems, especially Intel systems, need to execute 3ecacdbd23956a arch/x86/kernel/head_64.S Joerg Roedel 2020-09-07 170 * verify_cpu() above to make sure NX is enabled. 3ecacdbd23956a arch/x86/kernel/head_64.S Joerg Roedel 2020-09-07 171 */ 3ecacdbd23956a arch/x86/kernel/head_64.S Joerg Roedel 2020-09-07 172 SYM_INNER_LABEL(secondary_startup_64_no_verify, SYM_L_GLOBAL) fb799447ae2974 arch/x86/kernel/head_64.S Josh Poimboeuf 2023-03-01 173 UNWIND_HINT_END_OF_STACK 3e3f069504344c arch/x86/kernel/head_64.S Peter Zijlstra 2022-03-08 174 ANNOTATE_NOENDBR 3ecacdbd23956a arch/x86/kernel/head_64.S Joerg Roedel 2020-09-07 175 2f69a81ad68732 arch/x86/kernel/head_64.S Ard Biesheuvel 2023-08-07 176 /* Clear %R15 which holds the boot_params pointer on the boot CPU */ 721f791ce1cddf arch/x86/kernel/head_64.S Uros Bizjak 2024-01-24 177 xorl %r15d, %r15d 2f69a81ad68732 arch/x86/kernel/head_64.S Ard Biesheuvel 2023-08-07 178 d6a41f184dcea0 arch/x86/kernel/head_64.S Ard Biesheuvel 2024-02-27 179 /* Derive the runtime physical address of init_top_pgt[] */ d6a41f184dcea0 arch/x86/kernel/head_64.S Ard Biesheuvel 2024-02-27 180 movq phys_base(%rip), %rax d6a41f184dcea0 arch/x86/kernel/head_64.S Ard Biesheuvel 2024-02-27 181 addq $(init_top_pgt - __START_KERNEL_map), %rax d6a41f184dcea0 arch/x86/kernel/head_64.S Ard Biesheuvel 2024-02-27 182 5868f3651fa0df arch/x86/kernel/head_64.S Tom Lendacky 2017-07-17 183 /* 5868f3651fa0df arch/x86/kernel/head_64.S Tom Lendacky 2017-07-17 184 * Retrieve the modifier (SME encryption mask if SME is active) to be 5868f3651fa0df arch/x86/kernel/head_64.S Tom Lendacky 2017-07-17 185 * added to the initial pgdir entry that will be programmed into CR3. 5868f3651fa0df arch/x86/kernel/head_64.S Tom Lendacky 2017-07-17 186 */ 469693d8f62299 arch/x86/kernel/head_64.S Michael Roth 2022-02-09 187 #ifdef CONFIG_AMD_MEM_ENCRYPT d6a41f184dcea0 arch/x86/kernel/head_64.S Ard Biesheuvel 2024-02-27 188 addq sme_me_mask(%rip), %rax 469693d8f62299 arch/x86/kernel/head_64.S Michael Roth 2022-02-09 189 #endif 828263957611c2 arch/x86/kernel/head_64.S Ard Biesheuvel 2024-02-27 190 /* 828263957611c2 arch/x86/kernel/head_64.S Ard Biesheuvel 2024-02-27 191 * Switch to the init_top_pgt here, away from the trampoline_pgd and 828263957611c2 arch/x86/kernel/head_64.S Ard Biesheuvel 2024-02-27 192 * unmap the identity mapped ranges. 828263957611c2 arch/x86/kernel/head_64.S Ard Biesheuvel 2024-02-27 193 */ 828263957611c2 arch/x86/kernel/head_64.S Ard Biesheuvel 2024-02-27 194 movq %rax, %cr3 5868f3651fa0df arch/x86/kernel/head_64.S Tom Lendacky 2017-07-17 195
Signature · 2 lines
0-DAY CI Kernel Test Service https://github.com/intel/lkp-tests/wiki