Unify TDCALL/SEAMCALL and TDVMCALL assembly
Subsystems (2)
v3 cover Unify TDCALL/SEAMCALL and TDVMCALL assembly
Hi Peter, Kirill, all,
This series unifies the assembly code for TDCALL/SEAMCALL and TDVMCALL. Now all of them use one singe TDX_MODULE_CALL asm macro. More information please see cover letter of v2 (see link below).
This version mainly addressed Peter's comment to add patch to adjust 'struct tdx_module_args' to match KVM's "vcpu::regs".
Tested by booting TDX guest, initializing TDX module, and running TDX guest successfully, all with this series applied.
------- Histroy --------
v2 -> v3:
- New patch (patch 12) to adjust 'struct tdx_module_args' layout to match KVM's "vcpu::regs[]" for VP.ENTER. (Peter)
- Added __seamcall_saved_ret() wrapper to support VP.ENTER (merged to patch 10).
- Fixed a 'noinstr' check build regression found by LKP (patch 7).
- Rebased to latest Linus's tree (6.5-rc3 + 2 commits).
v2: https://lore.kernel.org/lkml/[email protected]/T/
v1 -> v2:
- Rebased to 6.5-rc2.
- Fixed comments from Peter and others.
- Split patch "x86/tdx: Unify TDX_HYPERCALL and TDX_MODULE_CALL assembly" into three smaller patches for better review.
- A new patch to skip saving output registers when SEAMCALL fails due to VMFailInvalid.
- Removed patch "x86/tdx: Use cmovc to save a label in TDX_MODULE_CALL asm"
- Merged patch "x86/tdx: Move FRAME_BEGIN/END to TDX_MODULE_CALL asm macro" to the new patch mentioned above.
v1: https://lore.kernel.org/lkml/[email protected]/T/
Kai Huang (12):
x86/tdx: Zero out the missing RSI in TDX_HYPERCALL macro x86/tdx: Skip saving output regs when SEAMCALL fails with
VMFailInvalid
x86/tdx: Make macros of TDCALLs consistent with the spec x86/tdx: Rename __tdx_module_call() to __tdcall()
x86/tdx: Pass TDCALL/SEAMCALL input/output registers via a structure x86/tdx: Extend TDX_MODULE_CALL to support more TDCALL/SEAMCALL leafs x86/tdx: Make TDX_HYPERCALL asm similar to TDX_MODULE_CALL x86/tdx: Reimplement __tdx_hypercall() using TDX_MODULE_CALL asm x86/tdx: Remove 'struct tdx_hypercall_args'
x86/virt/tdx: Wire up basic SEAMCALL functions x86/virt/tdx: Allow SEAMCALL to handle #UD and #GP x86/virt/tdx: Adjust 'struct tdx_module_args' to use x86 "register
index" layout
arch/x86/Kconfig | 12 ++ arch/x86/Makefile | 2 + arch/x86/boot/compressed/tdx.c | 6 +- arch/x86/coco/tdx/tdcall.S | 231 ++++-------------------------- arch/x86/coco/tdx/tdx-shared.c | 28 +++- arch/x86/coco/tdx/tdx.c | 69 +++++---- arch/x86/include/asm/shared/tdx.h | 92 +++++++----- arch/x86/include/asm/tdx.h | 11 ++ arch/x86/kernel/asm-offsets.c | 33 ++--- arch/x86/virt/Makefile | 2 + arch/x86/virt/vmx/Makefile | 2 + arch/x86/virt/vmx/tdx/Makefile | 2 + arch/x86/virt/vmx/tdx/seamcall.S | 61 ++++++++ arch/x86/virt/vmx/tdx/tdxcall.S | 227 ++++++++++++++++++++++------- 14 files changed, 433 insertions(+), 345 deletions(-) create mode 100644 arch/x86/virt/Makefile create mode 100644 arch/x86/virt/vmx/Makefile create mode 100644 arch/x86/virt/vmx/tdx/Makefile create mode 100644 arch/x86/virt/vmx/tdx/seamcall.S
base-commit: 18b44bc5a67275641fb26f2c54ba7eef80ac5950
Signature · 1 line
2.41.0
PATCH v3 1/12 x86/tdx: Zero out the missing RSI in TDX_HYPERCALL macro
In the TDX_HYPERCALL asm, after the TDCALL instruction returns from the untrusted VMM, the registers that the TDX guest shares to the VMM need to be cleared to avoid speculative execution of VMM-provided values.
RSI is specified in the bitmap of those registers, but it is missing when zeroing out those registers in the current TDX_HYPERCALL.
It was there when it was originally added in commit 752d13305c78 ("x86/tdx: Expand __tdx_hypercall() to handle more arguments"), but was later removed in commit 1e70c680375a ("x86/tdx: Do not corrupt frame-pointer in __tdx_hypercall()"), which was correct because %rsi is later restored in the "pop %rsi". However a later commit 7a3a401874be ("x86/tdx: Drop flags from __tdx_hypercall()") removed that "pop %rsi" but forgot to add the "xor %rsi, %rsi" back.
Fix by adding it back.
--- arch/x86/coco/tdx/tdcall.S | 1 + 1 file changed, 1 insertion(+)
arch/x86/coco/tdx/tdcall.S +1 -0
@@ -195,6 +195,7 @@ SYM_FUNC_END(__tdx_module_call) xor %r10d, %r10d xor %r11d, %r11d xor %rdi, %rdi + xor %rsi, %rsi xor %rdx, %rdx /* Restore callee-saved GPRs as mandated by the x86_64 ABI */
Signature · 1 line
2.41.0
PATCH v3 2/12 x86/tdx: Skip saving output regs when SEAMCALL fails with VMFailInvalid
In reply to Kai HuangIf SEAMCALL fails with VMFailInvalid, the SEAM software (e.g., the TDX module) won't have chance to set any output register. Skip saving the output registers to the structure in this case.
Also, as '.Lno_output_struct' is the very last symbol before RET, rename it to '.Lout' to make it short.
Opportunistically make the asm directives unindented.
---
v2 -> v3:
- No change.
v1 -> v2:
- A new patch to improve SEAMCALL VMFailInvalid failure, with v1 patch "x86/tdx: Move FRAME_BEGIN/END to TDX_MODULE_CALL asm macro" merged.
--- arch/x86/coco/tdx/tdcall.S | 3 --- arch/x86/virt/vmx/tdx/tdxcall.S | 29 ++++++++++++++++++++--------- 2 files changed, 20 insertions(+), 12 deletions(-)
arch/x86/coco/tdx/tdcall.S +0 -3
@@ -78,10 +78,7 @@ * Return status of TDCALL via RAX. */ SYM_FUNC_START(__tdx_module_call) - FRAME_BEGIN TDX_MODULE_CALL host=0 - FRAME_END - RET SYM_FUNC_END(__tdx_module_call) /*
arch/x86/virt/vmx/tdx/tdxcall.S +20 -9
@@ -1,5 +1,6 @@ /* SPDX-License-Identifier: GPL-2.0 */ #include <asm/asm-offsets.h> +#include <asm/frame.h> #include <asm/tdx.h> /*
@@ -18,6 +19,7 @@ * TDX module. */ .macro TDX_MODULE_CALL host:req + FRAME_BEGIN /* * R12 will be used as temporary storage for struct tdx_module_output * pointer. Since R12-R15 registers are not used by TDCALL/SEAMCALL
@@ -44,7 +46,7 @@ mov %rsi, %rcx /* Leave input param 2 in RDX */ - .if \host +.if \host seamcall /* * SEAMCALL instruction is essentially a VMExit from VMX root
@@ -57,13 +59,10 @@ * This value will never be used as actual SEAMCALL error code as * it is from the Reserved status code class. */ - jnc .Lno_vmfailinvalid - mov $TDX_SEAMCALL_VMFAILINVALID, %rax -.Lno_vmfailinvalid: - - .else + jc .Lseamcall_vmfailinvalid +.else tdcall - .endif +.endif /* * Fetch output pointer from stack to R12 (It is used
@@ -80,7 +79,7 @@ * Other registers may contain details of the failure. */ test %r12, %r12 - jz .Lno_output_struct + jz .Lout /* Copy result registers to output struct: */ movq %rcx, TDX_MODULE_rcx(%r12)
@@ -90,7 +89,19 @@ movq %r10, TDX_MODULE_r10(%r12) movq %r11, TDX_MODULE_r11(%r12) -.Lno_output_struct: +.Lout: /* Restore the state of R12 register */ pop %r12 + + FRAME_END + RET + +.if \host +.Lseamcall_vmfailinvalid: + mov $TDX_SEAMCALL_VMFAILINVALID, %rax + /* pop the unused output pointer back to %r9 */ + pop %r9 + jmp .Lout +.endif /* \host */ + .endm
Signature · 1 line
2.41.0
PATCH v3 3/12 x86/tdx: Make macros of TDCALLs consistent with the spec
In reply to Kai HuangThe TDX spec names all TDCALLs with prefix "TDG". Currently, the kernel doesn't follow such convention for the macros of those TDCALLs but uses prefix "TDX_" for all of them. Although it's arguable whether the TDX spec names those TDCALLs properly, it's better for the kernel to follow the spec when naming those macros.
Change all macros of TDCALLs to make them consistent with the spec. As a bonus, they get distinguished easily from the host-side SEAMCALLs, which all have prefix "TDH".
No functional change intended.
---
v2 -> v3:
- No change.
v1 -> v2:
- Rebase to 6.5-rc2.
--- arch/x86/coco/tdx/tdx-shared.c | 4 ++-- arch/x86/coco/tdx/tdx.c | 8 ++++---- arch/x86/include/asm/shared/tdx.h | 10 +++++----- 3 files changed, 11 insertions(+), 11 deletions(-)
arch/x86/coco/tdx/tdx-shared.c +2 -2
@@ -35,7 +35,7 @@ static unsigned long try_accept_one(phys_addr_t start, unsigned long len, } tdcall_rcx = start | page_size; -if (__tdx_module_call (TDX_ACCEPT_PAGE , tdcall_rcx,0 ,0 ,0 ,NULL )) +if (__tdx_module_call (TDG_MEM_PAGE_ACCEPT , tdcall_rcx,0 ,0 ,0 ,NULL ))return 0 ;return accept_size;
@@ -45,7 +45,7 @@ bool tdx_accept_memory(phys_addr_t start, phys_addr_t end) {/* * For shared->private conversion, accept the page using -* TDX_ACCEPT_PAGE TDX module call. +* TDG_MEM_PAGE_ACCEPT TDX module call. */ while (start < end) {unsigned long len = end - start;
arch/x86/coco/tdx/tdx.c +4 -4
@@ -91,7 +91,7 @@ int tdx_mcall_get_report0(u8 *reportdata, u8 *tdreport) {u64 ret; - ret =__tdx_module_call (TDX_GET_REPORT ,virt_to_phys (tdreport), + ret =__tdx_module_call (TDG_MR_REPORT ,virt_to_phys (tdreport),virt_to_phys (reportdata),TDREPORT_SUBTYPE_0 ,0 ,NULL );if (ret) {
@@ -152,7 +152,7 @@ static void tdx_parse_tdinfo(u64 *cc_mask)* Guest-Host-Communication Interface (GHCI), section 2.4.2 TDCALL * [TDG.VP.INFO]. */ -tdx_module_call (TDX_GET_INFO ,0 ,0 ,0 ,0 , &out); +tdx_module_call (TDG_VP_INFO ,0 ,0 ,0 ,0 , &out);/* * The highest bit of a guest physical address is the "sharing" bit.
@@ -594,7 +594,7 @@ void tdx_get_ve_info(struct ve_info *ve)* Note, the TDX module treats virtual NMIs as inhibited if the #VE * valid flag is set. It means that NMI=>#VE will not result in a #DF. */ -tdx_module_call (TDX_GET_VEINFO ,0 ,0 ,0 ,0 , &out); +tdx_module_call (TDG_VP_VEINFO_GET ,0 ,0 ,0 ,0 , &out);/* Transfer the output parameters */ ve->exit_reason = out.rcx;
@@ -774,7 +774,7 @@ void __init tdx_early_init(void)cc_set_mask (cc_mask);/* Kernel does not use NOTIFY_ENABLES and does not need random #VEs */ -tdx_module_call (TDX_WR ,0 ,TDCS_NOTIFY_ENABLES ,0 ,-1ULL ,NULL ); +tdx_module_call (TDG_VM_WR ,0 ,TDCS_NOTIFY_ENABLES ,0 ,-1ULL ,NULL );/* * All bits above GPA width are reserved and kernel treats shared bit
arch/x86/include/asm/shared/tdx.h +5 -5
@@ -11,11 +11,11 @@#define TDX_IDENT "IntelTDX "/* TDX module Call Leaf IDs */ -#define TDX_GET_INFO 1 -#define TDX_GET_VEINFO 3 -#define TDX_GET_REPORT 4 -#define TDX_ACCEPT_PAGE 6 -#define TDX_WR 8 +#define TDG_VP_INFO 1 +#define TDG_VP_VEINFO_GET 3 +#define TDG_MR_REPORT 4 +#define TDG_MEM_PAGE_ACCEPT 6 +#define TDG_VM_WR 8/* TDCS fields. To be used by TDG.VM.WR and TDG.VM.RD module calls */ #define TDCS_NOTIFY_ENABLES 0x9100000000000010
Signature · 1 line
2.41.0
PATCH v3 4/12 x86/tdx: Rename __tdx_module_call() to __tdcall()
In reply to Kai Huang__tdx_module_call() is only used by the TDX guest to issue TDCALL to the TDX module. Rename it to __tdcall() to match its behaviour, e.g., it cannot be used to make host-side SEAMCALL.
Also rename tdx_module_call() which is a wrapper of __tdx_module_call() to tdcall().
No functional change intended.
---
v2 -> v3:
- No change.
v1 -> v2:
- Rebase to 6.5-rc2.
--- arch/x86/coco/tdx/tdcall.S | 10 +++++----- arch/x86/coco/tdx/tdx-shared.c | 2 +- arch/x86/coco/tdx/tdx.c | 18 +++++++++--------- arch/x86/include/asm/shared/tdx.h | 4 ++-- 4 files changed, 17 insertions(+), 17 deletions(-)
arch/x86/coco/tdx/tdcall.S +5 -5
@@ -40,8 +40,8 @@ .section .noinstr.text, "ax" /* - * __tdx_module_call() - Used by TDX guests to request services from - * the TDX module (does not include VMM services) using TDCALL instruction. + * __tdcall() - Used by TDX guests to request services from the TDX + * module (does not include VMM services) using TDCALL instruction. * * Transforms function call register arguments into the TDCALL register ABI. * After TDCALL operation, TDX module output is saved in @out (if it is
@@ -62,7 +62,7 @@ * *------------------------------------------------------------------------- * - * __tdx_module_call() function ABI: + * __tdcall() function ABI: * * @fn (RDI) - TDCALL Leaf ID, moved to RAX * @rcx (RSI) - Input parameter 1, moved to RCX
@@ -77,9 +77,9 @@ * * Return status of TDCALL via RAX. */ -SYM_FUNC_START(__tdx_module_call) +SYM_FUNC_START(__tdcall) TDX_MODULE_CALL host=0 -SYM_FUNC_END(__tdx_module_call) +SYM_FUNC_END(__tdcall) /* * TDX_HYPERCALL - Make hypercalls to a TDX VMM using TDVMCALL leaf of TDCALL
arch/x86/coco/tdx/tdx-shared.c +1 -1
@@ -35,7 +35,7 @@ static unsigned long try_accept_one(phys_addr_t start, unsigned long len, } tdcall_rcx = start | page_size; -if (__tdx_module_call (TDG_MEM_PAGE_ACCEPT , tdcall_rcx,0 ,0 ,0 ,NULL )) +if (__tdcall (TDG_MEM_PAGE_ACCEPT , tdcall_rcx,0 ,0 ,0 ,NULL ))return 0 ;return accept_size;
arch/x86/coco/tdx/tdx.c +9 -9
@@ -66,10 +66,10 @@ EXPORT_SYMBOL_GPL(tdx_kvm_hypercall);* should only be used for calls that have no legitimate reason to fail * or where the kernel can not survive the call failing. */ -static inline void tdx_module_call (u64 fn,u64 rcx,u64 rdx,u64 r8,u64 r9, -struct tdx_module_output *out) +static inline void tdcall (u64 fn,u64 rcx,u64 rdx,u64 r8,u64 r9, +struct tdx_module_output *out) { -if (__tdx_module_call (fn, rcx, rdx, r8, r9, out)) +if (__tdcall (fn, rcx, rdx, r8, r9, out))panic ("TDCALL %lld failed (Buggy TDX module!)\n" , fn); }
@@ -91,9 +91,9 @@ int tdx_mcall_get_report0(u8 *reportdata, u8 *tdreport) {u64 ret; - ret =__tdx_module_call (TDG_MR_REPORT ,virt_to_phys (tdreport), -virt_to_phys (reportdata),TDREPORT_SUBTYPE_0 , -0 ,NULL ); + ret =__tdcall (TDG_MR_REPORT ,virt_to_phys (tdreport), +virt_to_phys (reportdata),TDREPORT_SUBTYPE_0 , +0 ,NULL );if (ret) {if (TDCALL_RETURN_CODE (ret) ==TDCALL_INVALID_OPERAND )return -EINVAL ;
@@ -152,7 +152,7 @@ static void tdx_parse_tdinfo(u64 *cc_mask)* Guest-Host-Communication Interface (GHCI), section 2.4.2 TDCALL * [TDG.VP.INFO]. */ -tdx_module_call (TDG_VP_INFO ,0 ,0 ,0 ,0 , &out); +tdcall (TDG_VP_INFO ,0 ,0 ,0 ,0 , &out);/* * The highest bit of a guest physical address is the "sharing" bit.
@@ -594,7 +594,7 @@ void tdx_get_ve_info(struct ve_info *ve)* Note, the TDX module treats virtual NMIs as inhibited if the #VE * valid flag is set. It means that NMI=>#VE will not result in a #DF. */ -tdx_module_call (TDG_VP_VEINFO_GET ,0 ,0 ,0 ,0 , &out); +tdcall (TDG_VP_VEINFO_GET ,0 ,0 ,0 ,0 , &out);/* Transfer the output parameters */ ve->exit_reason = out.rcx;
@@ -774,7 +774,7 @@ void __init tdx_early_init(void)cc_set_mask (cc_mask);/* Kernel does not use NOTIFY_ENABLES and does not need random #VEs */ -tdx_module_call (TDG_VM_WR ,0 ,TDCS_NOTIFY_ENABLES ,0 ,-1ULL ,NULL ); +tdcall (TDG_VM_WR ,0 ,TDCS_NOTIFY_ENABLES ,0 ,-1ULL ,NULL );/* * All bits above GPA width are reserved and kernel treats shared bit
arch/x86/include/asm/shared/tdx.h +2 -2
@@ -88,8 +88,8 @@ struct tdx_module_output { };/* Used to communicate with the TDX module */ -u64 __tdx_module_call (u64 fn,u64 rcx,u64 rdx,u64 r8,u64 r9, -struct tdx_module_output *out); +u64 __tdcall (u64 fn,u64 rcx,u64 rdx,u64 r8,u64 r9, +struct tdx_module_output *out);bool tdx_accept_memory (phys_addr_t start,phys_addr_t end);
Signature · 1 line
2.41.0
PATCH v3 5/12 x86/tdx: Pass TDCALL/SEAMCALL input/output registers via a structure
In reply to Kai HuangCurrently, the TDX_MODULE_CALL asm macro, which handles both TDCALL and SEAMCALL, takes one parameter for each input register and an optional 'struct tdx_module_output' (a collection of output registers) as output. This is different from the TDX_HYPERCALL macro which uses a single 'struct tdx_hypercall_args' to carry all input/output registers.
The newer TDX versions introduce more TDCALLs/SEAMCALLs which use more input/output registers. Also, the TDH.VP.ENTER (which isn't covered by the current TDX_MODULE_CALL macro) basically can use all registers that the TDX_HYPERCALL does. The current TDX_MODULE_CALL macro isn't extendible to cover those cases.
Similar to the TDX_HYPERCALL macro, simplify the TDX_MODULE_CALL macro to use a single structure 'struct tdx_module_args' to carry all the input/output registers.
Currently, the TDX_MODULE_CALL macro depends on the caller to pass a non-NULL 'struct tdx_module_output' to get additional output registers. Similar to the TDX_HYPERCALL macro, change the TDX_MODULE_CALL macro to take a new 'ret' macro argument to indicate whether to save the output registers to the 'struct tdx_module_args'. Also introduce a new __tdcall_ret() for that purpose, similar to the __tdx_hypercall_ret().
Note the tdcall(), which is a wrapper of __tdcall(), is called by three
callers: tdx_parse_tdinfo(), tdx_get_ve_info() and tdx_early_init(). The former two need the additional output but the last one doesn't. For simplicity, make tdcall() always call __tdcall_ret() to avoid another "_ret()" wrapper. The last caller tdx_early_init() isn't performance critical anyway.
---
v2 -> v3:
- No change.
v1 -> v2:
- Code change due to patch 02.
- Minor improvement in comments and changelog.
--- arch/x86/coco/tdx/tdcall.S | 47 ++++++---------- arch/x86/coco/tdx/tdx-shared.c | 6 +- arch/x86/coco/tdx/tdx.c | 44 ++++++++------- arch/x86/include/asm/shared/tdx.h | 10 ++-- arch/x86/kernel/asm-offsets.c | 12 ++-- arch/x86/virt/vmx/tdx/tdxcall.S | 93 ++++++++++++------------------- 6 files changed, 95 insertions(+), 117 deletions(-)
arch/x86/coco/tdx/tdcall.S +18 -29
@@ -43,37 +43,10 @@ * __tdcall() - Used by TDX guests to request services from the TDX * module (does not include VMM services) using TDCALL instruction. * - * Transforms function call register arguments into the TDCALL register ABI. - * After TDCALL operation, TDX module output is saved in @out (if it is - * provided by the user). - * - *------------------------------------------------------------------------- - * TDCALL ABI: - *------------------------------------------------------------------------- - * Input Registers: - * - * RAX - TDCALL Leaf number. - * RCX,RDX,R8-R9 - TDCALL Leaf specific input registers. - * - * Output Registers: - * - * RAX - TDCALL instruction error code. - * RCX,RDX,R8-R11 - TDCALL Leaf specific output registers. - * - *------------------------------------------------------------------------- - * * __tdcall() function ABI: * - * @fn (RDI) - TDCALL Leaf ID, moved to RAX - * @rcx (RSI) - Input parameter 1, moved to RCX - * @rdx (RDX) - Input parameter 2, moved to RDX - * @r8 (RCX) - Input parameter 3, moved to R8 - * @r9 (R8) - Input parameter 4, moved to R9 - * - * @out (R9) - struct tdx_module_output pointer - * stored temporarily in R12 (not - * shared with the TDX module). It - * can be NULL. + * @fn (RDI) - TDCALL Leaf ID, moved to RAX + * @args (RSI) - struct tdx_module_args for input * * Return status of TDCALL via RAX. */
@@ -81,6 +54,22 @@ SYM_FUNC_START(__tdcall) TDX_MODULE_CALL host=0 SYM_FUNC_END(__tdcall) +/* + * __tdcall_ret() - Used by TDX guests to request services from the TDX + * module (does not include VMM services) using TDCALL instruction, with + * saving output registers to the 'struct tdx_module_args' used as input. + * + * __tdcall_ret() function ABI: + * + * @fn (RDI) - TDCALL Leaf ID, moved to RAX + * @args (RSI) - struct tdx_module_args for input and output + * + * Return status of TDCALL via RAX. + */ +SYM_FUNC_START(__tdcall_ret) + TDX_MODULE_CALL host=0 ret=1 +SYM_FUNC_END(__tdcall_ret) + /* * TDX_HYPERCALL - Make hypercalls to a TDX VMM using TDVMCALL leaf of TDCALL * instruction
arch/x86/coco/tdx/tdx-shared.c +3 -3
@@ -5,7 +5,7 @@ static unsigned long try_accept_one(phys_addr_t start, unsigned long len, enum pg_level pg_level) {unsigned long accept_size =page_level_size (pg_level); -u64 tdcall_rcx; +struct tdx_module_args args = {};u8 page_size;if (!IS_ALIGNED (start, accept_size))
@@ -34,8 +34,8 @@ static unsigned long try_accept_one(phys_addr_t start, unsigned long len,return 0 ; } - tdcall_rcx = start | page_size; -if (__tdcall (TDG_MEM_PAGE_ACCEPT , tdcall_rcx,0 ,0 ,0 ,NULL )) + args.rcx = start | page_size; +if (__tdcall (TDG_MEM_PAGE_ACCEPT , &args))return 0 ;return accept_size;
arch/x86/coco/tdx/tdx.c +25 -19
109 lines not on this page · raw message
arch/x86/include/asm/shared/tdx.h +6 -4
@@ -74,22 +74,24 @@ static inline u64 _tdx_hypercall(u64 fn, u64 r12, u64 r13, u64 r14, u64 r15)void __tdx_hypercall_failed (void );/* -* Used in __tdx_module_call() to gather the output registers' values of the +* Used in __tdcall*() to gather the input/output registers' values of the * TDCALL instruction when requesting services from the TDX module. This is a * software only structure and not part of the TDX module/VMM ABI */ -struct tdx_module_output { +struct tdx_module_args { +/* input/output */ u64 rcx;u64 rdx;u64 r8;u64 r9; +/* additional output */ u64 r10;u64 r11; };/* Used to communicate with the TDX module */ -u64 __tdcall (u64 fn,u64 rcx,u64 rdx,u64 r8,u64 r9, -struct tdx_module_output *out); +u64 __tdcall (u64 fn,struct tdx_module_args *args); +u64 __tdcall_ret (u64 fn,struct tdx_module_args *args);bool tdx_accept_memory (phys_addr_t start,phys_addr_t end);
arch/x86/kernel/asm-offsets.c +6 -6
@@ -68,12 +68,12 @@ static void __used common(void)#endif BLANK (); -OFFSET (TDX_MODULE_rcx, tdx_module_output, rcx); -OFFSET (TDX_MODULE_rdx, tdx_module_output, rdx); -OFFSET (TDX_MODULE_r8, tdx_module_output, r8); -OFFSET (TDX_MODULE_r9, tdx_module_output, r9); -OFFSET (TDX_MODULE_r10, tdx_module_output, r10); -OFFSET (TDX_MODULE_r11, tdx_module_output, r11); +OFFSET (TDX_MODULE_rcx, tdx_module_args, rcx); +OFFSET (TDX_MODULE_rdx, tdx_module_args, rdx); +OFFSET (TDX_MODULE_r8, tdx_module_args, r8); +OFFSET (TDX_MODULE_r9, tdx_module_args, r9); +OFFSET (TDX_MODULE_r10, tdx_module_args, r10); +OFFSET (TDX_MODULE_r11, tdx_module_args, r11);BLANK ();OFFSET (TDX_HYPERCALL_r8, tdx_hypercall_args, r8);
arch/x86/virt/vmx/tdx/tdxcall.S +37 -56
120 lines not on this page · raw message
Signature · 1 line
2.41.0
PATCH v3 6/12 x86/tdx: Extend TDX_MODULE_CALL to support more TDCALL/SEAMCALL leafs
In reply to Kai HuangThe TDX guest live migration support (TDX 1.5) adds new TDCALL/SEAMCALL leaf functions. Those new TDCALLs/SEAMCALLs take additional registers for input (R10-R13) and output (R12-R13). TDG.SERVTD.RD is an example.
Also, the current TDX_MODULE_CALL doesn't aim to handle TDH.VP.ENTER SEAMCALL, which monitors the TDG.VP.VMCALL in input/output registers when it returns in case of VMCALL from TDX guest.
With those new TDCALLs/SEAMCALLs and the TDH.VP.ENTER covered, the TDX_MODULE_CALL macro basically needs to handle the same input/output registers as the TDX_HYPERCALL does. And as a result, they also share similar logic in the assembly, thus should be unified to use one common assembly.
Extend the TDX_MODULE_CALL asm to support the new TDCALLs/SEAMCALLs and also the TDH.VP.ENTER SEAMCALL. Eventually it will be unified with the TDX_HYPERCALL.
The new input/output registers fit with the "callee-saved" registers in the x86 calling convention. Add a new "saved" parameter to support those new TDCALLs/SEAMCALLs and TDH.VP.ENTER and keep the existing TDCALLs/SEAMCALLs minimally impacted.
For TDH.VP.ENTER, after it returns the registers shared by the guest contain guest's values. Explicitly clear them to prevent speculative use of guest's values.
Note most TDX live migration related SEAMCALLs may also clobber AVX* state ("AVX, AVX2 and AVX512 state: may be reset to the architectural INIT state" -- see TDH.EXPORT.MEM for example). And TDH.VP.ENTER also clobbers XMM0-XMM15 when the corresponding bit is set in RCX. Don't handle them in the TDX_MODULE_CALL macro but let the caller save and restore when needed.
This is basically based on Peter's code.
---
v2 -> v3:
- No change.
v1 -> v2:
- xorq %rcx, %rcx -> xorl %eax, %eax to save the REX prefix
- Minor improvement in comments and changelog.
--- arch/x86/coco/tdx/tdcall.S | 4 + arch/x86/include/asm/shared/tdx.h | 12 ++- arch/x86/kernel/asm-offsets.c | 7 ++ arch/x86/virt/vmx/tdx/tdxcall.S | 126 ++++++++++++++++++++++++++++-- 4 files changed, 141 insertions(+), 8 deletions(-)
arch/x86/coco/tdx/tdcall.S +4 -0
@@ -48,6 +48,8 @@ * @fn (RDI) - TDCALL Leaf ID, moved to RAX * @args (RSI) - struct tdx_module_args for input * + * Only RCX/RDX/R8-R11 are used as input registers. + * * Return status of TDCALL via RAX. */ SYM_FUNC_START(__tdcall)
@@ -64,6 +66,8 @@ SYM_FUNC_END(__tdcall) * @fn (RDI) - TDCALL Leaf ID, moved to RAX * @args (RSI) - struct tdx_module_args for input and output * + * Only RCX/RDX/R8-R11 are used as input/output registers. + * * Return status of TDCALL via RAX. */ SYM_FUNC_START(__tdcall_ret)
arch/x86/include/asm/shared/tdx.h +10 -2
@@ -79,14 +79,22 @@ void __tdx_hypercall_failed(void);* software only structure and not part of the TDX module/VMM ABI */ struct tdx_module_args { -/* input/output */ +/* callee-clobbered */ u64 rcx;u64 rdx;u64 r8;u64 r9; -/* additional output */ +/* extra callee-clobbered */ u64 r10;u64 r11; +/* callee-saved + rdi/rsi */ +u64 r12; +u64 r13; +u64 r14; +u64 r15; +u64 rbx; +u64 rdi; +u64 rsi; };/* Used to communicate with the TDX module */
arch/x86/kernel/asm-offsets.c +7 -0
@@ -74,6 +74,13 @@ static void __used common(void)OFFSET (TDX_MODULE_r9, tdx_module_args, r9);OFFSET (TDX_MODULE_r10, tdx_module_args, r10);OFFSET (TDX_MODULE_r11, tdx_module_args, r11); +OFFSET (TDX_MODULE_r12, tdx_module_args, r12); +OFFSET (TDX_MODULE_r13, tdx_module_args, r13); +OFFSET (TDX_MODULE_r14, tdx_module_args, r14); +OFFSET (TDX_MODULE_r15, tdx_module_args, r15); +OFFSET (TDX_MODULE_rbx, tdx_module_args, rbx); +OFFSET (TDX_MODULE_rdi, tdx_module_args, rdi); +OFFSET (TDX_MODULE_rsi, tdx_module_args, rsi);BLANK ();OFFSET (TDX_HYPERCALL_r8, tdx_hypercall_args, r8);
arch/x86/virt/vmx/tdx/tdxcall.S +120 -6
166 lines not on this page · raw message
Signature · 1 line
2.41.0
PATCH v3 7/12 x86/tdx: Make TDX_HYPERCALL asm similar to TDX_MODULE_CALL
In reply to Kai HuangNow the 'struct tdx_hypercall_args' and 'struct tdx_module_args' are almost the same, and the TDX_HYPERCALL and TDX_MODULE_CALL asm macro share similar code pattern too. The __tdx_hypercall() and __tdcall() should be unified to use the same assembly code.
As a preparation to unify them, simplify the TDX_HYPERCALL to make it more like the TDX_MODULE_CALL.
The TDX_HYPERCALL takes the pointer of 'struct tdx_hypercall_args' as function call argument, and does below extra things comparing to the TDX_MODULE_CALL:
1) It sets RAX to 0 (TDG.VP.VMCALL leaf) internally;
2) It sets RCX to the (fixed) bitmap of shared registers internally;
3) It calls __tdx_hypercall_failed() internally (and panics) when the TDCALL instruction itself fails;
4) After TDCALL, it moves R10 to RAX to return the return code of the VMCALL leaf, regardless the '\ret' asm macro argument;
Firstly, change the TDX_HYPERCALL to take the same function call arguments as the TDX_MODULE_CALL does: TDCALL leaf ID, and the pointer to 'struct tdx_module_args'. Then 1) and 2) can be moved to the caller:
- TDG.VP.VMCALL leaf ID can be passed via the function call argument;
- 'struct tdx_module_args' is 'struct tdx_hypercall_args' + RCX, thus the bitmap of shared registers can be passed via RCX in the structure.
Secondly, to move 3) and 4) out of assembly, make the TDX_HYPERCALL always save output registers to the structure. The caller then can:
- Call __tdx_hypercall_failed() when TDX_HYPERCALL returns error;
- Return R10 in the structure as the return code of the VMCALL leaf;
With above changes, change the asm function from __tdx_hypercall() to __tdcall_hypercall(), and reimplement __tdx_hypercall() as the C wrapper of it. This avoids having to add another wrapper of __tdx_hypercall() (_tdx_hypercall() is already taken).
The __tdcall_hypercall() will be replaced with a __tdcall() variant using TDX_MODULE_CALL in a later commit as the final goal is to have one assembly to handle both TDCALL and TDVMCALL.
Currently, the __tdx_hypercall() asm is in '.noinstr.text'. To keep this unchanged, annotate __tdx_hypercall(), which is a C function now, as 'noinstr'.
Remove the __tdx_hypercall_ret() as __tdx_hypercall() already does so.
Implement __tdx_hypercall() in tdx-shared.c so it can be shared with the compressed code.
Opportunistically fix a checkpatch error complaining using space around parenthesis '(' and ')' while moving the bitmap of shared registers to <asm/shared/tdx.h>.
---
v2 -> v3:
- Fixed 'noinstr' check build regression by annotating __tdx_hypercall() as 'noinstr'.
- Slight update in changelog accordingly.
v1 -> v2:
- The first patch (major) split from v1 "x86/tdx: Unify TDX_HYPERCALL and TDX_MODULE_CALL assembly"
- Rebase to 6.5-rc2.
- Removed 'noinstr' and 'instrumentation_begin()' in __tdx_hypercall_failed().
- Minor improvement in comments and changelog.
--- arch/x86/boot/compressed/tdx.c | 2 +- arch/x86/coco/tdx/tdcall.S | 148 ++++++++++-------------------- arch/x86/coco/tdx/tdx-shared.c | 43 +++++++++ arch/x86/coco/tdx/tdx.c | 9 +- arch/x86/include/asm/shared/tdx.h | 86 +++++++++++------ arch/x86/kernel/asm-offsets.c | 14 --- 6 files changed, 156 insertions(+), 146 deletions(-)
arch/x86/boot/compressed/tdx.c +1 -1
@@ -26,7 +26,7 @@ static inline unsigned int tdx_io_in(int size, u16 port) .r14 = port, }; -if (__tdx_hypercall_ret (&args)) +if (__tdx_hypercall (&args))return UINT_MAX ;return args.r11;
arch/x86/coco/tdx/tdcall.S +50 -98
196 lines not on this page · raw message
arch/x86/coco/tdx/tdx-shared.c +43 -0
46 lines not on this page · raw message
arch/x86/coco/tdx/tdx.c +5 -4
@@ -285,7 +285,7 @@ static int read_msr(struct pt_regs *regs, struct ve_info *ve)* can be found in TDX Guest-Host-Communication Interface * (GHCI), section titled "TDG.VP.VMCALL<Instruction.RDMSR>". */ -if (__tdx_hypercall_ret (&args)) +if (__tdx_hypercall (&args))return -EIO ; regs->ax =lower_32_bits (args.r11);
@@ -339,7 +339,7 @@ static int handle_cpuid(struct pt_regs *regs, struct ve_info *ve)* ABI can be found in TDX Guest-Host-Communication Interface * (GHCI), section titled "VP.VMCALL<Instruction.CPUID>". */ -if (__tdx_hypercall_ret (&args)) +if (__tdx_hypercall (&args))return -EIO ;/*
@@ -366,8 +366,9 @@ static bool mmio_read(int size, unsigned long addr, unsigned long *val) .r15 = *val, }; -if (__tdx_hypercall_ret (&args)) +if (__tdx_hypercall (&args))return false; + *val = args.r11;return true; }
@@ -500,7 +501,7 @@ static bool handle_in(struct pt_regs *regs, int size, int port)* in TDX Guest-Host-Communication Interface (GHCI) section titled * "TDG.VP.VMCALL<Instruction.IO>". */ - success = !__tdx_hypercall_ret (&args); + success = !__tdx_hypercall (&args);/* Update part of the register affected by the emulated instruction */ regs->ax &= ~mask;
arch/x86/include/asm/shared/tdx.h +57 -29
113 lines not on this page · raw message
arch/x86/kernel/asm-offsets.c +0 -14
@@ -82,20 +82,6 @@ static void __used common(void)OFFSET (TDX_MODULE_rdi, tdx_module_args, rdi);OFFSET (TDX_MODULE_rsi, tdx_module_args, rsi); -BLANK (); -OFFSET (TDX_HYPERCALL_r8, tdx_hypercall_args, r8); -OFFSET (TDX_HYPERCALL_r9, tdx_hypercall_args, r9); -OFFSET (TDX_HYPERCALL_r10, tdx_hypercall_args, r10); -OFFSET (TDX_HYPERCALL_r11, tdx_hypercall_args, r11); -OFFSET (TDX_HYPERCALL_r12, tdx_hypercall_args, r12); -OFFSET (TDX_HYPERCALL_r13, tdx_hypercall_args, r13); -OFFSET (TDX_HYPERCALL_r14, tdx_hypercall_args, r14); -OFFSET (TDX_HYPERCALL_r15, tdx_hypercall_args, r15); -OFFSET (TDX_HYPERCALL_rdi, tdx_hypercall_args, rdi); -OFFSET (TDX_HYPERCALL_rsi, tdx_hypercall_args, rsi); -OFFSET (TDX_HYPERCALL_rbx, tdx_hypercall_args, rbx); -OFFSET (TDX_HYPERCALL_rdx, tdx_hypercall_args, rdx); -BLANK ();OFFSET (BP_scratch, boot_params, scratch);OFFSET (BP_secure_boot, boot_params, secure_boot);
Signature · 1 line
2.41.0
PATCH v3 8/12 x86/tdx: Reimplement __tdx_hypercall() using TDX_MODULE_CALL asm
In reply to Kai HuangNow the TDX_HYPERCALL asm is basically identical to the TDX_MODULE_CALL with both '\saved' and '\ret' enabled, with two minor things though:
1) The way to restore the structure pointer is different
The TDX_HYPERCALL uses RCX as spare to restore the structure pointer, but the TDX_MODULE_CALL assumes no spare register can be used. In other words, TDX_MODULE_CALL already covers what TDX_HYPERCALL does.
2) TDX_MODULE_CALL only clears shared registers for TDH.VP.ENTER
For this just need to make that code available for the non-host case.
Thus, remove the TDX_HYPERCALL and reimplement the __tdx_hypercall() using the TDX_MODULE_CALL.
Extend the TDX_MODULE_CALL to cover "clear shared registers" for TDG.VP.VMCALL. Introduce a new __tdcall_saved_ret() to replace the temporary __tdcall_hypercall().
The __tdcall_saved_ret() can also be used for those new TDCALLs which require more input/output registers than the basic TDCALLs do.
---
v2 -> v3:
- No update
v1 -> v2:
- The second patch (new) split from v1 "x86/tdx: Unify TDX_HYPERCALL and TDX_MODULE_CALL assembly"
- Rebase to 6.5-rc2.
--- arch/x86/coco/tdx/tdcall.S | 132 ++---------------------------- arch/x86/coco/tdx/tdx-shared.c | 4 +- arch/x86/include/asm/shared/tdx.h | 2 +- arch/x86/virt/vmx/tdx/tdxcall.S | 8 +- 4 files changed, 15 insertions(+), 131 deletions(-)
arch/x86/coco/tdx/tdcall.S +8 -124
@@ -48,135 +48,19 @@ SYM_FUNC_START(__tdcall_ret) SYM_FUNC_END(__tdcall_ret) /* - * TDX_HYPERCALL - Make hypercalls to a TDX VMM using TDVMCALL leaf of TDCALL - * instruction - * - * Transforms values in function call argument struct tdx_module_args @args - * into the TDCALL register ABI. After TDCALL operation, VMM output is saved - * back in @args, if \ret is 1. - * - * Depends on the caller to pass TDG.VP.VMCALL as the TDCALL leaf, and set - * @args::rcx to TDVMCALL_EXPOSE_REGS_MASK. - * - *------------------------------------------------------------------------- - * TD VMCALL ABI: - *------------------------------------------------------------------------- - * - * Input Registers: - * - * RAX - TDCALL instruction leaf number (0 - TDG.VP.VMCALL) - * RCX - BITMAP which controls which part of TD Guest GPR - * is passed as-is to the VMM and back. - * R10 - Set 0 to indicate TDCALL follows standard TDX ABI - * specification. Non zero value indicates vendor - * specific ABI. - * R11 - VMCALL sub function number - * RBX, RDX, RDI, RSI - Used to pass VMCALL sub function specific arguments. - * R8-R9, R12-R15 - Same as above. - * - * Output Registers: - * - * RAX - TDCALL instruction status (Not related to hypercall - * output). - * RBX, RDX, RDI, RSI - Hypercall sub function specific output values. - * R8-R15 - Same as above. - * - */ -.macro TDX_HYPERCALL - FRAME_BEGIN - - /* Save callee-saved GPRs as mandated by the x86_64 ABI */ - push %r15 - push %r14 - push %r13 - push %r12 - push %rbx - - /* Move Leaf ID to RAX */ - movq %rdi, %rax - - /* Move bitmap of shared registers to RCX */ - movq TDX_MODULE_rcx(%rsi), %rcx - - /* Copy hypercall registers from arg struct: */ - movq TDX_MODULE_r8(%rsi), %r8 - movq TDX_MODULE_r9(%rsi), %r9 - movq TDX_MODULE_r10(%rsi), %r10 - movq TDX_MODULE_r11(%rsi), %r11 - movq TDX_MODULE_r12(%rsi), %r12 - movq TDX_MODULE_r13(%rsi), %r13 - movq TDX_MODULE_r14(%rsi), %r14 - movq TDX_MODULE_r15(%rsi), %r15 - movq TDX_MODULE_rdi(%rsi), %rdi - movq TDX_MODULE_rbx(%rsi), %rbx - movq TDX_MODULE_rdx(%rsi), %rdx - - pushq %rsi - movq TDX_MODULE_rsi(%rsi), %rsi - - tdcall - - /* - * Restore the pointer of the structure to save output registers. - * - * RCX is used as bitmap of shared registers and doesn't hold any - * value provided by the VMM, thus it can be used as spare to - * restore the structure pointer. - */ - popq %rcx - movq %rsi, TDX_MODULE_rsi(%rcx) - movq %rcx, %rsi - - movq %r8, TDX_MODULE_r8(%rsi) - movq %r9, TDX_MODULE_r9(%rsi) - movq %r10, TDX_MODULE_r10(%rsi) - movq %r11, TDX_MODULE_r11(%rsi) - movq %r12, TDX_MODULE_r12(%rsi) - movq %r13, TDX_MODULE_r13(%rsi) - movq %r14, TDX_MODULE_r14(%rsi) - movq %r15, TDX_MODULE_r15(%rsi) - movq %rdi, TDX_MODULE_rdi(%rsi) - movq %rbx, TDX_MODULE_rbx(%rsi) - movq %rdx, TDX_MODULE_rdx(%rsi) - - /* - * Zero out registers exposed to the VMM to avoid speculative execution - * with VMM-controlled values. This needs to include all registers - * present in TDVMCALL_EXPOSE_REGS_MASK, except RBX, and R12-R15 which - * will be restored. - */ - xor %r8d, %r8d - xor %r9d, %r9d - xor %r10d, %r10d - xor %r11d, %r11d - xor %rdi, %rdi - xor %rsi, %rsi - xor %rdx, %rdx - - /* Restore callee-saved GPRs as mandated by the x86_64 ABI */ - pop %rbx - pop %r12 - pop %r13 - pop %r14 - pop %r15 - - FRAME_END - - RET -.endm - -/* + * __tdcall_saved_ret() - Used by TDX guests to request services from the + * TDX module (including VMM services) using TDCALL instruction, with + * saving output registers to the 'struct tdx_module_args' used as input. * - * __tdcall_hypercall() function ABI: + * __tdcall_saved_ret() function ABI: * * @fn (RDI) - TDCALL leaf ID, moved to RAX * @args (RSI) - struct tdx_module_args for input/output * - * @fn and @args::rcx from the caller must be TDG_VP_VMCALL and - * TDVMCALL_EXPOSE_REGS_MASK respectively. + * All registers in @args are used as input/output registers. * * On successful completion, return the hypercall error code. */ -SYM_FUNC_START(__tdcall_hypercall) - TDX_HYPERCALL -SYM_FUNC_END(__tdcall_hypercall) +SYM_FUNC_START(__tdcall_saved_ret) + TDX_MODULE_CALL host=0 ret=1 saved=1 +SYM_FUNC_END(__tdcall_saved_ret)
arch/x86/coco/tdx/tdx-shared.c +2 -2
@@ -89,11 +89,11 @@ noinstr u64 __tdx_hypercall(struct tdx_hypercall_args *args) };/* -* Failure of __tdcall_hypercall() indicates a failure of the TDVMCALL +* Failure of __tdcall_saved_ret() indicates a failure of the TDVMCALL * mechanism itself and that something has gone horribly wrong with * the TDX module. __tdx_hypercall_failed() never returns. */ -if (__tdcall_hypercall (TDG_VP_VMCALL , &margs)) +if (__tdcall_saved_ret (TDG_VP_VMCALL , &margs))__tdx_hypercall_failed (); args->r8 = margs.r8;
arch/x86/include/asm/shared/tdx.h +1 -1
@@ -81,6 +81,7 @@ struct tdx_module_args {/* Used to communicate with the TDX module */ u64 __tdcall (u64 fn,struct tdx_module_args *args);u64 __tdcall_ret (u64 fn,struct tdx_module_args *args); +u64 __tdcall_saved_ret (u64 fn,struct tdx_module_args *args);/* * Used in __tdx_hypercall() to pass down and get back registers' values of
@@ -104,7 +105,6 @@ struct tdx_hypercall_args { };/* Used to request services from the VMM */ -u64 __tdcall_hypercall (u64 fn,struct tdx_module_args *args);u64 __tdx_hypercall (struct tdx_hypercall_args *args);/*
arch/x86/virt/vmx/tdx/tdxcall.S +4 -4
@@ -143,10 +143,10 @@ movq %r11, TDX_MODULE_r11(%rsi) .endif /* \ret */ -.if \host && \saved && \ret +.if \saved && \ret /* - * Clear registers shared by guest for VP.ENTER to prevent - * speculative use of guest's values, including those are + * Clear registers shared by guest for VP.VMCALL/VP.ENTER to prevent + * speculative use of guest's/VMM's values, including those are * restored from the stack. * * See arch/x86/kvm/vmx/vmenter.S:
@@ -171,7 +171,7 @@ xorl %r15d, %r15d xorl %ebx, %ebx xorl %edi, %edi -.endif /* \host && \ret && \host */ +.endif /* \ret && \host */ .if \host .Lout\@:
Signature · 1 line
2.41.0
PATCH v3 9/12 x86/tdx: Remove 'struct tdx_hypercall_args'
In reply to Kai HuangNow 'struct tdx_hypercall_args' is basically 'struct tdx_module_args' minus RCX. Although from __tdx_hypercall()'s perspective RCX isn't used as shared register thus not part of input/output registers, it's not worth to have a separate structure just due to one register.
Remove the 'struct tdx_hypercall_args' and use 'struct tdx_module_args' instead in __tdx_hypercall() related code. This also saves the memory copy between the two structures within __tdx_hypercall().
---
v2 -> v3:
- No change.
v1 -> v2:
- The third patch (new) split from v1 "x86/tdx: Unify TDX_HYPERCALL and TDX_MODULE_CALL assembly"
- Rebase to 6.5-rc2.
--- arch/x86/boot/compressed/tdx.c | 4 ++-- arch/x86/coco/tdx/tdx-shared.c | 37 ++++++------------------------- arch/x86/coco/tdx/tdx.c | 16 ++++++------- arch/x86/include/asm/shared/tdx.h | 25 ++------------------- 4 files changed, 19 insertions(+), 63 deletions(-)
arch/x86/boot/compressed/tdx.c +2 -2
@@ -18,7 +18,7 @@ void __tdx_hypercall_failed(void)static inline unsigned int tdx_io_in (int size,u16 port) { -struct tdx_hypercall_args args = { +struct tdx_module_args args = { .r10 =TDX_HYPERCALL_STANDARD , .r11 =hcall_func (EXIT_REASON_IO_INSTRUCTION ), .r12 = size,
@@ -34,7 +34,7 @@ static inline unsigned int tdx_io_in(int size, u16 port)static inline void tdx_io_out (int size,u16 port,u32 value) { -struct tdx_hypercall_args args = { +struct tdx_module_args args = { .r10 =TDX_HYPERCALL_STANDARD , .r11 =hcall_func (EXIT_REASON_IO_INSTRUCTION ), .r12 = size,
arch/x86/coco/tdx/tdx-shared.c +7 -30
@@ -70,45 +70,22 @@ bool tdx_accept_memory(phys_addr_t start, phys_addr_t end)return true; } -noinstr u64 __tdx_hypercall(struct tdx_hypercall_args *args) +noinstr u64 __tdx_hypercall(struct tdx_module_args *args) { -struct tdx_module_args margs = { - .rcx =TDVMCALL_EXPOSE_REGS_MASK , - .rdx = args->rdx, - .r8 = args->r8, - .r9 = args->r9, - .r10 = args->r10, - .r11 = args->r11, - .r12 = args->r12, - .r13 = args->r13, - .r14 = args->r14, - .r15 = args->r15, - .rbx = args->rbx, - .rdi = args->rdi, - .rsi = args->rsi, - }; +/* +* For TDVMCALL explicitly set RCX to the bitmap of shared registers. +* The caller isn't expected to set @args->rcx anyway. +*/ + args->rcx =TDVMCALL_EXPOSE_REGS_MASK ;/* * Failure of __tdcall_saved_ret() indicates a failure of the TDVMCALL * mechanism itself and that something has gone horribly wrong with * the TDX module. __tdx_hypercall_failed() never returns. */ -if (__tdcall_saved_ret (TDG_VP_VMCALL , &margs)) +if (__tdcall_saved_ret (TDG_VP_VMCALL , args))__tdx_hypercall_failed (); - args->r8 = margs.r8; - args->r9 = margs.r9; - args->r10 = margs.r10; - args->r11 = margs.r11; - args->r12 = margs.r12; - args->r13 = margs.r13; - args->r14 = margs.r14; - args->r15 = margs.r15; - args->rdi = margs.rdi; - args->rsi = margs.rsi; - args->rbx = margs.rbx; - args->rdx = margs.rdx; -/* TDVMCALL leaf return code is in R10 */ return args->r10; }
arch/x86/coco/tdx/tdx.c +8 -8
@@ -48,7 +48,7 @@ noinstr void __tdx_hypercall_failed(void)long tdx_kvm_hypercall (unsigned int nr,unsigned long p1,unsigned long p2,unsigned long p3,unsigned long p4) { -struct tdx_hypercall_args args = { +struct tdx_module_args args = { .r10 = nr, .r11 = p1, .r12 = p2,
@@ -108,7 +108,7 @@ EXPORT_SYMBOL_GPL(tdx_mcall_get_report0);static void __noreturn tdx_panic (const char *msg) { -struct tdx_hypercall_args args = { +struct tdx_module_args args = { .r10 =TDX_HYPERCALL_STANDARD , .r11 =TDVMCALL_REPORT_FATAL_ERROR , .r12 =0 ,/* Error code: 0 is Panic */
@@ -230,7 +230,7 @@ static int ve_instr_len(struct ve_info *ve) static u64 __cpuidle __halt(const bool irq_disabled) { -struct tdx_hypercall_args args = { +struct tdx_module_args args = { .r10 =TDX_HYPERCALL_STANDARD , .r11 =hcall_func (EXIT_REASON_HLT ), .r12 = irq_disabled,
@@ -274,7 +274,7 @@ void __cpuidle tdx_safe_halt(void)static int read_msr (struct pt_regs *regs,struct ve_info *ve) { -struct tdx_hypercall_args args = { +struct tdx_module_args args = { .r10 =TDX_HYPERCALL_STANDARD , .r11 =hcall_func (EXIT_REASON_MSR_READ ), .r12 = regs->cx,
@@ -295,7 +295,7 @@ static int read_msr(struct pt_regs *regs, struct ve_info *ve)static int write_msr (struct pt_regs *regs,struct ve_info *ve) { -struct tdx_hypercall_args args = { +struct tdx_module_args args = { .r10 =TDX_HYPERCALL_STANDARD , .r11 =hcall_func (EXIT_REASON_MSR_WRITE ), .r12 = regs->cx,
@@ -315,7 +315,7 @@ static int write_msr(struct pt_regs *regs, struct ve_info *ve)static int handle_cpuid (struct pt_regs *regs,struct ve_info *ve) { -struct tdx_hypercall_args args = { +struct tdx_module_args args = { .r10 =TDX_HYPERCALL_STANDARD , .r11 =hcall_func (EXIT_REASON_CPUID ), .r12 = regs->ax,
@@ -357,7 +357,7 @@ static int handle_cpuid(struct pt_regs *regs, struct ve_info *ve)static bool mmio_read (int size,unsigned long addr,unsigned long *val) { -struct tdx_hypercall_args args = { +struct tdx_module_args args = { .r10 =TDX_HYPERCALL_STANDARD , .r11 =hcall_func (EXIT_REASON_EPT_VIOLATION ), .r12 = size,
@@ -486,7 +486,7 @@ static int handle_mmio(struct pt_regs *regs, struct ve_info *ve)static bool handle_in (struct pt_regs *regs,int size,int port) { -struct tdx_hypercall_args args = { +struct tdx_module_args args = { .r10 =TDX_HYPERCALL_STANDARD , .r11 =hcall_func (EXIT_REASON_IO_INSTRUCTION ), .r12 = size,
arch/x86/include/asm/shared/tdx.h +2 -23
@@ -83,29 +83,8 @@ u64 __tdcall(u64 fn, struct tdx_module_args *args);u64 __tdcall_ret (u64 fn,struct tdx_module_args *args);u64 __tdcall_saved_ret (u64 fn,struct tdx_module_args *args); -/* -* Used in __tdx_hypercall() to pass down and get back registers' values of -* the TDCALL instruction when requesting services from the VMM. -* -* This is a software only structure and not part of the TDX module/VMM ABI. -*/ -struct tdx_hypercall_args { -u64 r8; -u64 r9; -u64 r10; -u64 r11; -u64 r12; -u64 r13; -u64 r14; -u64 r15; -u64 rdi; -u64 rsi; -u64 rbx; -u64 rdx; -}; -/* Used to request services from the VMM */ -u64 __tdx_hypercall (struct tdx_hypercall_args *args); +u64 __tdx_hypercall (struct tdx_module_args *args);/* * Wrapper for standard use of __tdx_hypercall with no output aside from
@@ -113,7 +92,7 @@ u64 __tdx_hypercall(struct tdx_hypercall_args *args);*/ static inline u64 _tdx_hypercall (u64 fn,u64 r12,u64 r13,u64 r14,u64 r15) { -struct tdx_hypercall_args args = { +struct tdx_module_args args = { .r10 =TDX_HYPERCALL_STANDARD , .r11 = fn, .r12 = r12,
Signature · 1 line
2.41.0
PATCH v3 10/12 x86/virt/tdx: Wire up basic SEAMCALL functions
In reply to Kai HuangIntel Trust Domain Extensions (TDX) protects guest VMs from malicious host and certain physical attacks. A CPU-attested software module called 'the TDX module' runs inside a new isolated memory range as a trusted hypervisor to manage and run protected VMs.
TDX introduces a new CPU mode: Secure Arbitration Mode (SEAM). This mode runs only the TDX module itself or other code to load the TDX module.
The host kernel communicates with SEAM software via a new SEAMCALL instruction. This is conceptually similar to a guest->host hypercall, except it is made from the host to SEAM software instead. The TDX module establishes a new SEAMCALL ABI which allows the host to initialize the module and to manage VMs.
The SEAMCALL ABI is very similar to the TDCALL ABI and leverages much TDCALL infrastructure. Wire up basic functions to make SEAMCALLs for the basic support of running TDX guests: __seamcall(), __seamcall_ret(), and __seamcall_saved_ret() for TDH.VP.ENTER. All SEAMCALLs involved in the basic TDX support don't use "callee-saved" registers as input and output, except the TDH.VP.ENTER.
To start to support TDX, create a new arch/x86/virt/vmx/tdx/tdx.c for TDX host kernel support. Add a new Kconfig option CONFIG_INTEL_TDX_HOST to opt-in TDX host kernel support (to distinguish with TDX guest kernel support). So far only KVM uses TDX. Make the new config option depend on KVM_INTEL.
---
v2 -> v3:
- Added __seamcall_saved_ret() back for TDH.VP.ENTER, given the new patch to adjust 'struct tdx_module_args' layout.
v1 -> v2:
- Removed __seamcall_saved_ret() and leave it to KVM TDX patches.
--- arch/x86/Kconfig | 12 +++++++ arch/x86/Makefile | 2 ++ arch/x86/include/asm/tdx.h | 7 ++++ arch/x86/virt/Makefile | 2 ++ arch/x86/virt/vmx/Makefile | 2 ++ arch/x86/virt/vmx/tdx/Makefile | 2 ++ arch/x86/virt/vmx/tdx/seamcall.S | 61 ++++++++++++++++++++++++++++++++ 7 files changed, 88 insertions(+) create mode 100644 arch/x86/virt/Makefile create mode 100644 arch/x86/virt/vmx/Makefile create mode 100644 arch/x86/virt/vmx/tdx/Makefile create mode 100644 arch/x86/virt/vmx/tdx/seamcall.S
arch/x86/Kconfig +12 -0
@@ -1949,6 +1949,18 @@ config X86_SGX If unsure, say N. +config INTEL_TDX_HOST + bool "Intel Trust Domain Extensions (TDX) host support" + depends on CPU_SUP_INTEL + depends on X86_64 + depends on KVM_INTEL + help + Intel Trust Domain Extensions (TDX) protects guest VMs from malicious + host and certain physical attacks. This option enables necessary TDX + support in the host kernel to run confidential VMs. + + If unsure, say N. + config EFI bool "EFI runtime service support" depends on ACPI
arch/x86/Makefile +2 -0
@@ -252,6 +252,8 @@ archheaders: libs-y += arch/x86/lib/ +core-y += arch/x86/virt/ + # drivers-y are linked after core-y drivers-$(CONFIG_MATH_EMULATION) += arch/x86/math-emu/ drivers-$(CONFIG_PCI) += arch/x86/pci/
arch/x86/include/asm/tdx.h +7 -0
@@ -72,5 +72,12 @@ static inline long tdx_kvm_hypercall(unsigned int nr, unsigned long p1,return -ENODEV ; }#endif /* CONFIG_INTEL_TDX_GUEST && CONFIG_KVM_GUEST */ + +#ifdef CONFIG_INTEL_TDX_HOST +u64 __seamcall (u64 fn,struct tdx_module_args *args); +u64 __seamcall_ret (u64 fn,struct tdx_module_args *args); +u64 __seamcall_saved_ret (u64 fn,struct tdx_module_args *args); +#endif /* CONFIG_INTEL_TDX_HOST */ +#endif /* !__ASSEMBLY__ */ #endif /* _ASM_X86_TDX_H */
arch/x86/virt/Makefile new file +2 -0
@@ -0,0 +1,2 @@ +# SPDX-License-Identifier: GPL-2.0-only +obj-y += vmx/
arch/x86/virt/vmx/Makefile new file +2 -0
@@ -0,0 +1,2 @@ +# SPDX-License-Identifier: GPL-2.0-only +obj-$(CONFIG_INTEL_TDX_HOST) += tdx/
arch/x86/virt/vmx/tdx/Makefile new file +2 -0
@@ -0,0 +1,2 @@ +# SPDX-License-Identifier: GPL-2.0-only +obj-y += seamcall.o
arch/x86/virt/vmx/tdx/seamcall.S new file +61 -0
61 lines not on this page · raw message
Signature · 1 line
2.41.0
PATCH v3 11/12 x86/virt/tdx: Allow SEAMCALL to handle #UD and #GP
In reply to Kai HuangTDX memory has integrity and confidentiality protections. Violations of this integrity protection are supposed to only affect TDX operations and are never supposed to affect the host kernel itself. In other words, the host kernel should never, itself, see machine checks induced by the TDX integrity hardware.
Alas, the first few generations of TDX hardware have an erratum. A partial write to a TDX private memory cacheline will silently "poison" the line. Subsequent reads will consume the poison and generate a machine check. According to the TDX hardware spec, neither of these things should have happened.
On the platform with this erratum, it would be nice if the #MC handler could print additional information to show the #MC was TDX private memory error due to possible kernel bug.
To do that, the machine check handler needs to use SEAMCALL to query page type of the error memory from the TDX module, because there's no existing infrastructure to track TDX private pages.
SEAMCALL instruction causes #UD if CPU isn't in VMX operation. In #MC handler, it is legal that CPU isn't in VMX operation when making this SEAMCALL. Extend the TDX_MODULE_CALL macro to handle #UD so the SEAMCALL can return error code instead of Oops in the #MC handler. Opportunistically handles #GP too since they share the same code.
A bonus is when kernel mistakenly calls SEAMCALL when CPU isn't in VMX operation, or when TDX isn't enabled by the BIOS, or when the BIOS is buggy, the kernel can get a nicer error message rather than a less understandable Oops.
This is basically based on Peter's code.
---
v2 -> v3:
- Added more material to the changelog to explain the TDX erratum (hoping it can be merged together with this series).
- Changed to use %rdi to hold $TDX_SW_ERROR instead of %r12 as the latter is callee-saved.
v1 -> v2:
- Skip saving output registers when SEAMCALL #UD/#GP
--- arch/x86/include/asm/tdx.h | 4 ++++ arch/x86/virt/vmx/tdx/tdxcall.S | 19 +++++++++++++++++++ 2 files changed, 23 insertions(+)
arch/x86/include/asm/tdx.h +4 -0
@@ -8,6 +8,7 @@#include <asm/errno.h> #include <asm/ptrace.h> +#include <asm/trapnr.h> #include <asm/shared/tdx.h> /*
@@ -20,6 +21,9 @@#define TDX_SW_ERROR (TDX_ERROR | GENMASK_ULL(47, 40))#define TDX_SEAMCALL_VMFAILINVALID (TDX_SW_ERROR | _UL(0xFFFF0000)) +#define TDX_SEAMCALL_GP (TDX_SW_ERROR | X86_TRAP_GP) +#define TDX_SEAMCALL_UD (TDX_SW_ERROR | X86_TRAP_UD) +#ifndef __ASSEMBLY__/*
arch/x86/virt/vmx/tdx/tdxcall.S +19 -0
@@ -1,6 +1,7 @@ /* SPDX-License-Identifier: GPL-2.0 */ #include <asm/asm-offsets.h> #include <asm/frame.h> +#include <asm/asm.h> #include <asm/tdx.h> /*
@@ -85,6 +86,7 @@ .endif /* \saved */ .if \host +.Lseamcall\@: seamcall /* * SEAMCALL instruction is essentially a VMExit from VMX root
@@ -192,11 +194,28 @@ .if \host .Lseamcall_vmfailinvalid\@: mov $TDX_SEAMCALL_VMFAILINVALID, %rax + jmp .Lseamcall_fail\@ + +.Lseamcall_trap\@: + /* + * SEAMCALL caused #GP or #UD. By reaching here %eax contains + * the trap number. Convert the trap number to the TDX error + * code by setting TDX_SW_ERROR to the high 32-bits of %rax. + * + * Note cannot OR TDX_SW_ERROR directly to %rax as OR instruction + * only accepts 32-bit immediate at most. + */ + movq $TDX_SW_ERROR, %rdi + orq %rdi, %rax + +.Lseamcall_fail\@: .if \ret && \saved /* pop the unused structure pointer back to %rsi */ popq %rsi .endif jmp .Lout\@ + + _ASM_EXTABLE_FAULT(.Lseamcall\@, .Lseamcall_trap\@) .endif /* \host */ .endm
Signature · 1 line
2.41.0
PATCH v3 12/12 x86/virt/tdx: Adjust 'struct tdx_module_args' to use x86 "register index" layout
In reply to Kai HuangFor TDX guest, KVM needs to call __seamcall_saved_ret() to make the TDH.VP.ENTER SEAMCALL to enter the guest, possibly taking all registers in 'struct tdx_module_args' as input/output.
KVM caches guest's GPRs in 'kvm_vcpu_arch::regs[]', which follows the "register index" hardware layout of x86 GPRs. On the other hand, the __seamcall_saved_ret() takes the pointer of 'struct tdx_module_args' as argument, thus there's a mismatch.
KVM could choose to copy input registers from 'vcpu::regs[]' to a 'struct tdx_module_args' and use that as argument to make the SEAMCALL, but such memory copy isn't desired and should be avoided if possible.
It's not feasible to change KVM's 'vcpu::regs[]' layout due to various reasons (e.g., emulation code uses decoded register index as array index to access the register). Therefore, adjust 'struct tdx_module_args' to match KVM's 'vcpu::regs[]' layout so that KVM can simply do below:
__seamcall_saved_ret(TDH_VP_ENTER, (struct tdx_module_args *)vcpu->arch.regs);
Note RAX/RSP/RBP are not used by the TDX_MODULE_CALL assembly, but they are necessary in order match the layout of 'struct tdx_module_args' to KVM's 'vcpu::regs[]'. Thus add them to the structure, but name them as *_unused. Also don't include them to asm-offset.c so that any misuse of them in the assembly would result in build error.
---
v2 -> v3:
- New patch
--- arch/x86/include/asm/shared/tdx.h | 19 +++++++++++++------ arch/x86/kernel/asm-offsets.c | 6 +++--- 2 files changed, 16 insertions(+), 9 deletions(-)
arch/x86/include/asm/shared/tdx.h +13 -6
@@ -58,24 +58,31 @@* Used in __tdcall*() to gather the input/output registers' values of the * TDCALL instruction when requesting services from the TDX module. This is a * software only structure and not part of the TDX module/VMM ABI +* +* Note those *_unused are not used by the TDX_MODULE_CALL assembly. +* The layout of this structure also matches KVM's kvm_vcpu_arch::regs[] +* layout, which follows the "register index" order of x86 GPRs. KVM +* then can simply type cast kvm_vcpu_arch::regs[] to this structure to +* avoid the extra memory copy between two structures when making +* TDH.VP.ENTER SEAMCALL. */ struct tdx_module_args { -/* callee-clobbered */ +u64 rax_unused;u64 rcx;u64 rdx; +u64 rbx; +u64 rsp_unused; +u64 rbp_unused; +u64 rsi; +u64 rdi;u64 r8;u64 r9; -/* extra callee-clobbered */ u64 r10;u64 r11; -/* callee-saved + rdi/rsi */ u64 r12;u64 r13;u64 r14;u64 r15; -u64 rbx; -u64 rdi; -u64 rsi; };/* Used to communicate with the TDX module */
arch/x86/kernel/asm-offsets.c +3 -3
@@ -70,6 +70,9 @@ static void __used common(void)BLANK ();OFFSET (TDX_MODULE_rcx, tdx_module_args, rcx);OFFSET (TDX_MODULE_rdx, tdx_module_args, rdx); +OFFSET (TDX_MODULE_rbx, tdx_module_args, rbx); +OFFSET (TDX_MODULE_rsi, tdx_module_args, rsi); +OFFSET (TDX_MODULE_rdi, tdx_module_args, rdi);OFFSET (TDX_MODULE_r8, tdx_module_args, r8);OFFSET (TDX_MODULE_r9, tdx_module_args, r9);OFFSET (TDX_MODULE_r10, tdx_module_args, r10);
@@ -78,9 +81,6 @@ static void __used common(void)OFFSET (TDX_MODULE_r13, tdx_module_args, r13);OFFSET (TDX_MODULE_r14, tdx_module_args, r14);OFFSET (TDX_MODULE_r15, tdx_module_args, r15); -OFFSET (TDX_MODULE_rbx, tdx_module_args, rbx); -OFFSET (TDX_MODULE_rdi, tdx_module_args, rdi); -OFFSET (TDX_MODULE_rsi, tdx_module_args, rsi);BLANK ();OFFSET (BP_scratch, boot_params, scratch);
Signature · 1 line
2.41.0
Re: [PATCH v3 01/12] x86/tdx: Zero out the missing RSI in TDX_HYPERCALL macro
In reply to Kai HuangQuoted 15 lines · Kai Huang
In the
TDX_HYPERCALLasm, after the TDCALL instruction returns from the untrusted VMM, the registers that the TDX guest shares to the VMM need to be cleared to avoid speculative execution of VMM-provided values.RSI is specified in the bitmap of those registers, but it is missing when zeroing out those registers in the current
TDX_HYPERCALL.It was there when it was originally added in commit
752d13305c78("x86/tdx: Expand__tdx_hypercall()to handle more arguments"), but was later removed in commit1e70c680375a("x86/tdx: Do not corrupt frame-pointer in__tdx_hypercall()"), which was correct because %rsi is later restored in the "pop %rsi". However a later commit7a3a401874be("x86/tdx: Drop flags from__tdx_hypercall()") removed that "pop %rsi" but forgot to add the "xor %rsi, %rsi" back.
Quoted 5 lines
Fix by adding it back.
Fixes 7a3a401874be ("x86/tdx: Drop flags from __tdx_hypercall()")Signed-off-by Kai Huang <[email protected]>Reviewed-by Kuppuswamy Sathyanarayanan <[email protected]>
Signature · 1 line
Kiryl Shutsemau / Kirill A. Shutemov
Re: [PATCH v3 02/12] x86/tdx: Skip saving output regs when SEAMCALL fails with VMFailInvalid
In reply to Kai HuangQuoted 33 lines · Kai Huang
If SEAMCALL fails with VMFailInvalid, the SEAM software (e.g., the TDX module) won't have chance to set any output register. Skip saving the output registers to the structure in this case.
Also, as '.Lno_output_struct' is the very last symbol before RET, rename it to '.Lout' to make it short.
Opportunistically make the asm directives unindented.
Cc Kirill A. Shutemov <[email protected]>Cc Dave Hansen <[email protected]>Cc Peter Zijlstra <[email protected]>Suggested-by Peter Zijlstra <[email protected]>Signed-off-by Kai Huang <[email protected]>---
v2 -> v3:
- No change.v1 -> v2:
- A new patch to improve SEAMCALL VMFailInvalid failure, with v1 patch "x86/tdx: Move FRAME_BEGIN/END toTDX_MODULE_CALLasm macro" merged.--- arch/x86/coco/tdx/tdcall.S | 3 --- arch/x86/virt/vmx/tdx/tdxcall.S | 29 ++++++++++++++++++++--------- 2 files changed, 20 insertions(+), 12 deletions(-)arch/x86/coco/tdx/tdcall.S +0
-0
arch/x86/coco/tdx/tdcall.S line 83
Quoted 9 lines
arch/x86/coco/tdx/tdcall.S +0
-3@@ -78,10 +78,7 @@ * Return status of TDCALL via RAX. */ SYM_FUNC_START(__tdx_module_call) - FRAME_BEGIN TDX_MODULE_CALL host=0 - FRAME_END - RET SYM_FUNC_END(__tdx_module_call)
Do we still need to include asm/frame.h after the change?
Signature · 1 line
Kiryl Shutsemau / Kirill A. Shutemov
Re: [PATCH v3 03/12] x86/tdx: Make macros of TDCALLs consistent with the spec
In reply to Kai HuangQuoted 6 lines · Kai Huang
The TDX spec names all TDCALLs with prefix "TDG". Currently, the kernel doesn't follow such convention for the macros of those TDCALLs but uses prefix "
TDX_" for all of them. Although it's arguable whether the TDX spec names those TDCALLs properly, it's better for the kernel to follow the spec when naming those macros.
Quoted 7 lines
Change all macros of TDCALLs to make them consistent with the spec. As a bonus, they get distinguished easily from the host-side SEAMCALLs, which all have prefix "TDH".
No functional change intended.
Signed-off-by Kai Huang <[email protected]>
I remember we had few back-and-forth on naming this defines. I think matching names to the spec is helpful for future readers:
Signature · 1 line
Kiryl Shutsemau / Kirill A. Shutemov
Re: [PATCH v3 04/12] x86/tdx: Rename __tdx_module_call() to __tdcall()
In reply to Kai HuangQuoted 10 lines · Kai Huang
__tdx_module_call()is only used by the TDX guest to issue TDCALL to the TDX module. Rename it to__tdcall()to match its behaviour, e.g., it cannot be used to make host-side SEAMCALL.Also rename
tdx_module_call()which is a wrapper of__tdx_module_call()totdcall().No functional change intended.
Signed-off-by Kai Huang <[email protected]>
Fair enough:
Signature · 1 line
Kiryl Shutsemau / Kirill A. Shutemov
Re: [PATCH v3 05/12] x86/tdx: Pass TDCALL/SEAMCALL input/output registers via a structure
In reply to Kai HuangQuoted 13 lines · Kai Huang
arch/x86/virt/vmx/tdx/tdxcall.S +5
-0@@ -17,34 +17,33 @@ * TDX module and hypercalls to the VMM. * SEAMCALL - used by TDX hosts to make requests to the * TDX module. + * + *------------------------------------------------------------------------- + * TDCALL/SEAMCALL ABI: + *------------------------------------------------------------------------- + * Input Registers:
arch/x86/virt/vmx/tdx/tdxcall.S line 34
Quoted 10 lines
arch/x86/virt/vmx/tdx/tdxcall.S +10
-0+ * + * RAX - TDCALL/SEAMCALL Leaf number. + * RCX,RDX,R8-R9 - TDCALL/SEAMCALL Leaf specific input registers. + * + * Output Registers: + * + * RAX - TDCALL/SEAMCALL instruction error code. + * RCX,RDX,R8-R11 - TDCALL/SEAMCALL Leaf specific output registers. + * + *-------------------------------------------------------------------------
So, you keep the existing asymetry in IN and OUT registers. R10 and R11 are OUT-only registers. It can be confusing for user since it is the same structure now. I guess it changes in the following patches, but I would prefer to make them even here if there's no good reason not to.
Signature · 1 line
Kiryl Shutsemau / Kirill A. Shutemov
Re: [PATCH v3 06/12] x86/tdx: Extend TDX_MODULE_CALL to support more TDCALL/SEAMCALL leafs
In reply to Kai HuangQuoted 13 lines · Kai Huang
arch/x86/virt/vmx/tdx/tdxcall.S +9
-0@@ -64,6 +103,37 @@ .endif .if \ret +.if \saved + /* + * Restore the structure from stack to save the output registers + * + * In case of VP.ENTER returns due to TDVMCALL, all registers are + * valid thus no register can be used as spare to restore the + * structure from the stack (see "TDH.VP.ENTER Output Operands + * Definition on TDCALL(TDG.VP.VMCALL) Following a TD Entry"). + * For this case, need to make one register as spare by saving it
arch/x86/virt/vmx/tdx/tdxcall.S line 124
Quoted 10 lines
arch/x86/virt/vmx/tdx/tdxcall.S +10
-0+ * to the stack and then manually load the structure pointer to + * the spare register. + * + * Note for other TDCALLs/SEAMCALLs there are spare registers + * thus no need for such hack but just use this for all. + */ + pushq %rax /* save the TDCALL/SEAMCALL return code */ + movq 8(%rsp), %rax /* restore the structure pointer */ + movq %rsi, TDX_MODULE_rsi(%rax) /* save %rsi */ + movq %rax, %rsi /* use %rsi as structure pointer */
This looks redundant. You get struct in RSI with popq two lines below.
And please use upper case for registers: RSI instead of %rsi.
Quoted 5 lines
arch/x86/virt/vmx/tdx/tdxcall.S +5
-0+ popq %rax /* restore the return code */ + popq %rsi /* pop the structure pointer */ + + /* Copy additional output regs to the structure */ + movq %r12, TDX_MODULE_r12(%rsi)
arch/x86/virt/vmx/tdx/tdxcall.S line 139
Quoted 10 lines
arch/x86/virt/vmx/tdx/tdxcall.S +7
-0+ movq %r13, TDX_MODULE_r13(%rsi) + movq %r14, TDX_MODULE_r14(%rsi) + movq %r15, TDX_MODULE_r15(%rsi) + movq %rbx, TDX_MODULE_rbx(%rsi) + movq %rdi, TDX_MODULE_rdi(%rsi) +.endif /* \saved */ + /* Copy output registers to the structure */ movq %rcx, TDX_MODULE_rcx(%rsi) movq %rdx, TDX_MODULE_rdx(%rsi)
Otherwise, looks sane:
Signature · 1 line
Kiryl Shutsemau / Kirill A. Shutemov
Re: [PATCH v3 07/12] x86/tdx: Make TDX_HYPERCALL asm similar to TDX_MODULE_CALL
In reply to Kai HuangQuoted 2 lines · Kai Huang
Remove the
__tdx_hypercall_ret()as__tdx_hypercall()already does so.
Hm. So we now update struct on all VMCALLs. Is it a good idea? We give more control to VMM where it is not needed. I would rather keep the struct read-only where possible.
Signature · 1 line
Kiryl Shutsemau / Kirill A. Shutemov
Re: [PATCH v3 05/12] x86/tdx: Pass TDCALL/SEAMCALL input/output registers via a structure
In reply to [email protected]Quoted 21 lines · On Thu, 2023-07-27 at 19:36 +0300, [email protected] wrote:
Quoted 20 lines · Kai Huang
arch/x86/virt/vmx/tdx/tdxcall.S +12
-0@@ -17,34 +17,33 @@ * TDX module and hypercalls to the VMM. * SEAMCALL - used by TDX hosts to make requests to the * TDX module. + * + *------------------------------------------------------------------------- + * TDCALL/SEAMCALL ABI: + *------------------------------------------------------------------------- + * Input Registers: + * + * RAX - TDCALL/SEAMCALL Leaf number. + * RCX,RDX,R8-R9 - TDCALL/SEAMCALL Leaf specific input registers. + * + * Output Registers: + * + * RAX - TDCALL/SEAMCALL instruction error code.
Quoted 9 lines
arch/x86/virt/vmx/tdx/tdxcall.S line 34
Quoted 3 lines
arch/x86/virt/vmx/tdx/tdxcall.S +3
-0+ * RCX,RDX,R8-R11 - TDCALL/SEAMCALL Leaf specific output registers. + * + *-------------------------------------------------------------------------So, you keep the existing asymetry in IN and OUT registers. R10 and R11 are OUT-only registers. It can be confusing for user since it is the same structure now. I guess it changes in the following patches, but I would prefer to make them even here if there's no good reason not to.
Do you mean you prefer to use R10/R11 as input too even in this patch?
I think _logically_ it should be part of the next patch, because w/o extending TDX_MODULE_CALL to support additional TDCALLs/SEAMCALLs, we don't need R10/R11 as input. This patch only changes to take a structure as function argument, rather than taking individual registers as argument.
Also, we have a comment to say this around the structure too:
/*
- * Used in __tdx_module_call() to gather the output registers' values of the
+ * Used in __tdcall*() to gather the input/output registers' values of the
* TDCALL instruction when requesting services from the TDX module. This is a
* software only structure and not part of the TDX module/VMM ABI */
-struct tdx_module_output {
+struct tdx_module_args {
+ /* input/output */
u64 rcx;
u64 rdx;
u64 r8;
u64 r9;
+ /* additional output */
u64 r10;
u64 r11;
};
So to me there should be no confusion.
Even consider a theoretical case: someone wants to backport this patch to an old kernel w/o further patches, then it makes little sense to do R10/R11 in TDX_MODULE_CALL here in this patch
:-)
Re: [PATCH v3 02/12] x86/tdx: Skip saving output regs when SEAMCALL fails with VMFailInvalid
In reply to [email protected]Quoted 29 lines · On Thu, 2023-07-27 at 15:52 +0300, [email protected] wrote:
Quoted 28 lines · Kai Huang
If SEAMCALL fails with VMFailInvalid, the SEAM software (e.g., the TDX module) won't have chance to set any output register. Skip saving the output registers to the structure in this case.
Also, as '.Lno_output_struct' is the very last symbol before RET, rename it to '.Lout' to make it short.
Opportunistically make the asm directives unindented.
Cc Kirill A. Shutemov <[email protected]>Cc Dave Hansen <[email protected]>Cc Peter Zijlstra <[email protected]>Suggested-by Peter Zijlstra <[email protected]>Signed-off-by Kai Huang <[email protected]>---
v2 -> v3:
- No change.v1 -> v2:
- A new patch to improve SEAMCALL VMFailInvalid failure, with v1 patch "x86/tdx: Move FRAME_BEGIN/END toTDX_MODULE_CALLasm macro" merged.--- arch/x86/coco/tdx/tdcall.S | 3 --- arch/x86/virt/vmx/tdx/tdxcall.S | 29 ++++++++++++++++++++--------- 2 files changed, 20 insertions(+), 12 deletions(-)
Quoted 17 lines
Quoted 5 lines
arch/x86/coco/tdx/tdcall.S +0
-0arch/x86/coco/tdx/tdcall.S line 83
Quoted 9 lines
arch/x86/coco/tdx/tdcall.S +0
-3@@ -78,10 +78,7 @@ * Return status of TDCALL via RAX. */ SYM_FUNC_START(__tdx_module_call) - FRAME_BEGIN TDX_MODULE_CALL host=0 - FRAME_END - RET SYM_FUNC_END(__tdx_module_call)Do we still need to include
asm/frame.hafter the change?
For this patch yes because we still have TDX_HYPERCALL macro here.
We probably can remove it in the patch where TDX_HYPERCALL gets removed.
Re: [PATCH v3 06/12] x86/tdx: Extend TDX_MODULE_CALL to support more TDCALL/SEAMCALL leafs
In reply to [email protected]Quoted 18 lines · On Thu, 2023-07-27 at 19:50 +0300, [email protected] wrote:
Quoted 17 lines · Kai Huang
arch/x86/virt/vmx/tdx/tdxcall.S +13
-0@@ -64,6 +103,37 @@ .endif .if \ret +.if \saved + /* + * Restore the structure from stack to save the output registers + * + * In case of VP.ENTER returns due to TDVMCALL, all registers are + * valid thus no register can be used as spare to restore the + * structure from the stack (see "TDH.VP.ENTER Output Operands + * Definition on TDCALL(TDG.VP.VMCALL) Following a TD Entry"). + * For this case, need to make one register as spare by saving it + * to the stack and then manually load the structure pointer to + * the spare register. + * + * Note for other TDCALLs/SEAMCALLs there are spare registers
Quoted 8 lines
arch/x86/virt/vmx/tdx/tdxcall.S line 124
Quoted 6 lines
arch/x86/virt/vmx/tdx/tdxcall.S +6
-0+ * thus no need for such hack but just use this for all. + */ + pushq %rax /* save the TDCALL/SEAMCALL return code */ + movq 8(%rsp), %rax /* restore the structure pointer */ + movq %rsi, TDX_MODULE_rsi(%rax) /* save %rsi */ + movq %rax, %rsi /* use %rsi as structure pointer */This looks redundant. You get struct in RSI with popq two lines below.
Yes I guess so. Will remove and have a test.
Quoted 2 lines
And please use upper case for registers: RSI instead of %rsi.
In the comments? Will do.
Quoted 19 lines
Quoted 6 lines
arch/x86/virt/vmx/tdx/tdxcall.S +5
-0+ popq %rax /* restore the return code */ + popq %rsi /* pop the structure pointer */ + + /* Copy additional output regs to the structure */ + movq %r12, TDX_MODULE_r12(%rsi)arch/x86/virt/vmx/tdx/tdxcall.S line 139
Quoted 10 lines
arch/x86/virt/vmx/tdx/tdxcall.S +7
-0+ movq %r13, TDX_MODULE_r13(%rsi) + movq %r14, TDX_MODULE_r14(%rsi) + movq %r15, TDX_MODULE_r15(%rsi) + movq %rbx, TDX_MODULE_rbx(%rsi) + movq %rdi, TDX_MODULE_rdi(%rsi) +.endif /* \saved */ + /* Copy output registers to the structure */ movq %rcx, TDX_MODULE_rcx(%rsi) movq %rdx, TDX_MODULE_rdx(%rsi)Otherwise, looks sane:
Quoted 2 lines
Reviewed-by Kirill A. Shutemov <[email protected]>
Thanks.
Re: [PATCH v3 07/12] x86/tdx: Make TDX_HYPERCALL asm similar to TDX_MODULE_CALL
In reply to [email protected]Quoted 6 lines · On Thu, 2023-07-27 at 20:10 +0300, [email protected] wrote:
Quoted 2 lines · Kai Huang
Remove the
__tdx_hypercall_ret()as__tdx_hypercall()already does so.Hm. So we now update struct on all VMCALLs. Is it a good idea?
Do you mean we "unconditionally save output registers to the structure", right?
Quoted 3 lines
We give
more control to VMM where it is not needed.
I don't quite follow this. Can you elaborate?
Do you worry about VMM being malicious and putting malicious values to the registers?
Quoted 3 lines
I would rather keep the struct
read-only where possible.
We can achieve this if there's a clean way to do, but I don't see that.
Re: [PATCH v3 03/12] x86/tdx: Make macros of TDCALLs consistent with the spec
In reply to Kai HuangQuoted 6 lines · Kai Huang
The TDX spec names all TDCALLs with prefix "TDG". Currently, the kernel doesn't follow such convention for the macros of those TDCALLs but uses prefix "
TDX_" for all of them. Although it's arguable whether the TDX spec names those TDCALLs properly, it's better for the kernel to follow the spec when naming those macros.
Quoted 6 lines
Change all macros of TDCALLs to make them consistent with the spec. As a bonus, they get distinguished easily from the host-side SEAMCALLs, which all have prefix "TDH".
No functional change intended.
When upstreaming the TDX guest patches, there was a discussion about using TDG vs TDX. Final agreement is to use TDX_ prefix. I think it makes sense to align with the spec, but it is up to the maintainer.
What about the function name prefix? Are you planning to change them to tdg_*?
Quoted 98 lines
Signed-off-by Kai Huang <[email protected]>---
v2 -> v3:
- No change.v1 -> v2:
- Rebase to 6.5-rc2.--- arch/x86/coco/tdx/tdx-shared.c | 4 ++-- arch/x86/coco/tdx/tdx.c | 8 ++++---- arch/x86/include/asm/shared/tdx.h | 10 +++++----- 3 files changed, 11 insertions(+), 11 deletions(-)arch/x86/coco/tdx/tdx-shared.c +2
-2@@ -35,7 +35,7 @@ static unsigned long try_accept_one(phys_addr_t start, unsigned long len, } tdcall_rcx = start | page_size; -if (__tdx_module_call (TDX_ACCEPT_PAGE , tdcall_rcx,0 ,0 ,0 ,NULL )) +if (__tdx_module_call (TDG_MEM_PAGE_ACCEPT , tdcall_rcx,0 ,0 ,0 ,NULL ))return 0 ;return accept_size;@@ -45,7 +45,7 @@ bool tdx_accept_memory(phys_addr_t start, phys_addr_t end) {/* * For shared->private conversion, accept the page using -* TDX_ACCEPT_PAGE TDX module call. +* TDG_MEM_PAGE_ACCEPT TDX module call. */ while (start < end) {unsigned long len = end - start;arch/x86/coco/tdx/tdx.c +4
-4@@ -91,7 +91,7 @@ int tdx_mcall_get_report0(u8 *reportdata, u8 *tdreport) {u64 ret; - ret =__tdx_module_call (TDX_GET_REPORT ,virt_to_phys (tdreport), + ret =__tdx_module_call (TDG_MR_REPORT ,virt_to_phys (tdreport),virt_to_phys (reportdata),TDREPORT_SUBTYPE_0 ,0 ,NULL );if (ret) {@@ -152,7 +152,7 @@ static void tdx_parse_tdinfo(u64 *cc_mask)* Guest-Host-Communication Interface (GHCI), section 2.4.2 TDCALL * [TDG.VP.INFO]. */ -tdx_module_call (TDX_GET_INFO ,0 ,0 ,0 ,0 , &out); +tdx_module_call (TDG_VP_INFO ,0 ,0 ,0 ,0 , &out);/* * The highest bit of a guest physical address is the "sharing" bit. @@ -594,7 +594,7 @@ void tdx_get_ve_info(struct ve_info *ve)* Note, the TDX module treats virtual NMIs as inhibited if the #VE * valid flag is set. It means that NMI=>#VE will not result in a #DF. */ -tdx_module_call (TDX_GET_VEINFO ,0 ,0 ,0 ,0 , &out); +tdx_module_call (TDG_VP_VEINFO_GET ,0 ,0 ,0 ,0 , &out);/* Transfer the output parameters */ ve->exit_reason = out.rcx;@@ -774,7 +774,7 @@ void __init tdx_early_init(void)cc_set_mask (cc_mask);/* Kernel does not use NOTIFY_ENABLES and does not need random #VEs */ -tdx_module_call (TDX_WR ,0 ,TDCS_NOTIFY_ENABLES ,0 ,-1ULL ,NULL ); +tdx_module_call (TDG_VM_WR ,0 ,TDCS_NOTIFY_ENABLES ,0 ,-1ULL ,NULL );/* * All bits above GPA width are reserved and kernel treats shared bit arch/x86/include/asm/shared/tdx.h +5
-5@@ -11,11 +11,11 @@#define TDX_IDENT "IntelTDX "/* TDX module Call Leaf IDs */ -#define TDX_GET_INFO 1 -#define TDX_GET_VEINFO 3 -#define TDX_GET_REPORT 4 -#define TDX_ACCEPT_PAGE 6 -#define TDX_WR 8 +#define TDG_VP_INFO 1 +#define TDG_VP_VEINFO_GET 3 +#define TDG_MR_REPORT 4 +#define TDG_MEM_PAGE_ACCEPT 6 +#define TDG_VM_WR 8/* TDCS fields. To be used by TDG.VM.WR and TDG.VM.RD module calls */ #define TDCS_NOTIFY_ENABLES 0x9100000000000010
Signature · 2 lines
Sathyanarayanan Kuppuswamy Linux Kernel Developer
Re: [PATCH v3 03/12] x86/tdx: Make macros of TDCALLs consistent with the spec
Quoted 12 lines · Sathyanarayanan Kuppuswamy
Quoted 10 lines · Kai Huang
The TDX spec names all TDCALLs with prefix "TDG". Currently, the kernel doesn't follow such convention for the macros of those TDCALLs but uses prefix "
TDX_" for all of them. Although it's arguable whether the TDX spec names those TDCALLs properly, it's better for the kernel to follow the spec when naming those macros.Change all macros of TDCALLs to make them consistent with the spec. As a bonus, they get distinguished easily from the host-side SEAMCALLs, which all have prefix "TDH".
Quoted 8 lines
Quoted 2 lines
No functional change intended.
When upstreaming the TDX guest patches, there was a discussion about using TDG vs TDX. Final agreement is to use
TDX_prefix. I think it makes sense to align with the spec, but it is up to the maintainer.What about the function name prefix? Are you planning to change them to tdg_*?
I changed to __tdcall() in the next patch.
Quoted 2 lines
Reviewed-by Kuppuswamy Sathyanarayanan <[email protected]>
Thanks.
Re: [PATCH v3 04/12] x86/tdx: Rename __tdx_module_call() to __tdcall()
In reply to Kai HuangQuoted 11 lines · Kai Huang
__tdx_module_call()is only used by the TDX guest to issue TDCALL to the TDX module. Rename it to__tdcall()to match its behaviour, e.g., it cannot be used to make host-side SEAMCALL.Also rename
tdx_module_call()which is a wrapper of__tdx_module_call()totdcall().No functional change intended.
Signed-off-by Kai Huang <[email protected]>---
Looks fine to me.
Quoted 136 lines
v2 -> v3:
- No change.v1 -> v2:
- Rebase to 6.5-rc2.--- arch/x86/coco/tdx/tdcall.S | 10 +++++----- arch/x86/coco/tdx/tdx-shared.c | 2 +- arch/x86/coco/tdx/tdx.c | 18 +++++++++--------- arch/x86/include/asm/shared/tdx.h | 4 ++-- 4 files changed, 17 insertions(+), 17 deletions(-)arch/x86/coco/tdx/tdcall.S +5
-5@@ -40,8 +40,8 @@ .section .noinstr.text, "ax" /* - * __tdx_module_call() - Used by TDX guests to request services from - * the TDX module (does not include VMM services) using TDCALL instruction. + * __tdcall() - Used by TDX guests to request services from the TDX + * module (does not include VMM services) using TDCALL instruction. * * Transforms function call register arguments into the TDCALL register ABI. * After TDCALL operation, TDX module output is saved in @out (if it is@@ -62,7 +62,7 @@ * *------------------------------------------------------------------------- * - * __tdx_module_call() function ABI: + * __tdcall() function ABI: * * @fn (RDI) - TDCALL Leaf ID, moved to RAX * @rcx (RSI) - Input parameter 1, moved to RCX@@ -77,9 +77,9 @@ * * Return status of TDCALL via RAX. */ -SYM_FUNC_START(__tdx_module_call) +SYM_FUNC_START(__tdcall) TDX_MODULE_CALL host=0 -SYM_FUNC_END(__tdx_module_call) +SYM_FUNC_END(__tdcall) /* * TDX_HYPERCALL - Make hypercalls to a TDX VMM using TDVMCALL leaf of TDCALLarch/x86/coco/tdx/tdx-shared.c +1
-1@@ -35,7 +35,7 @@ static unsigned long try_accept_one(phys_addr_t start, unsigned long len, } tdcall_rcx = start | page_size; -if (__tdx_module_call (TDG_MEM_PAGE_ACCEPT , tdcall_rcx,0 ,0 ,0 ,NULL )) +if (__tdcall (TDG_MEM_PAGE_ACCEPT , tdcall_rcx,0 ,0 ,0 ,NULL ))return 0 ;return accept_size;arch/x86/coco/tdx/tdx.c +9
-9@@ -66,10 +66,10 @@ EXPORT_SYMBOL_GPL(tdx_kvm_hypercall);* should only be used for calls that have no legitimate reason to fail * or where the kernel can not survive the call failing. */ -static inline void tdx_module_call (u64 fn,u64 rcx,u64 rdx,u64 r8,u64 r9, -struct tdx_module_output *out) +static inline void tdcall (u64 fn,u64 rcx,u64 rdx,u64 r8,u64 r9, +struct tdx_module_output *out) { -if (__tdx_module_call (fn, rcx, rdx, r8, r9, out)) +if (__tdcall (fn, rcx, rdx, r8, r9, out))panic ("TDCALL %lld failed (Buggy TDX module!)\n" , fn); }@@ -91,9 +91,9 @@ int tdx_mcall_get_report0(u8 *reportdata, u8 *tdreport) {u64 ret; - ret =__tdx_module_call (TDG_MR_REPORT ,virt_to_phys (tdreport), -virt_to_phys (reportdata),TDREPORT_SUBTYPE_0 , -0 ,NULL ); + ret =__tdcall (TDG_MR_REPORT ,virt_to_phys (tdreport), +virt_to_phys (reportdata),TDREPORT_SUBTYPE_0 , +0 ,NULL );if (ret) {if (TDCALL_RETURN_CODE (ret) ==TDCALL_INVALID_OPERAND )return -EINVAL ;@@ -152,7 +152,7 @@ static void tdx_parse_tdinfo(u64 *cc_mask)* Guest-Host-Communication Interface (GHCI), section 2.4.2 TDCALL * [TDG.VP.INFO]. */ -tdx_module_call (TDG_VP_INFO ,0 ,0 ,0 ,0 , &out); +tdcall (TDG_VP_INFO ,0 ,0 ,0 ,0 , &out);/* * The highest bit of a guest physical address is the "sharing" bit. @@ -594,7 +594,7 @@ void tdx_get_ve_info(struct ve_info *ve)* Note, the TDX module treats virtual NMIs as inhibited if the #VE * valid flag is set. It means that NMI=>#VE will not result in a #DF. */ -tdx_module_call (TDG_VP_VEINFO_GET ,0 ,0 ,0 ,0 , &out); +tdcall (TDG_VP_VEINFO_GET ,0 ,0 ,0 ,0 , &out);/* Transfer the output parameters */ ve->exit_reason = out.rcx;@@ -774,7 +774,7 @@ void __init tdx_early_init(void)cc_set_mask (cc_mask);/* Kernel does not use NOTIFY_ENABLES and does not need random #VEs */ -tdx_module_call (TDG_VM_WR ,0 ,TDCS_NOTIFY_ENABLES ,0 ,-1ULL ,NULL ); +tdcall (TDG_VM_WR ,0 ,TDCS_NOTIFY_ENABLES ,0 ,-1ULL ,NULL );/* * All bits above GPA width are reserved and kernel treats shared bit arch/x86/include/asm/shared/tdx.h +2
-2@@ -88,8 +88,8 @@ struct tdx_module_output { };/* Used to communicate with the TDX module */ -u64 __tdx_module_call (u64 fn,u64 rcx,u64 rdx,u64 r8,u64 r9, -struct tdx_module_output *out); +u64 __tdcall (u64 fn,u64 rcx,u64 rdx,u64 r8,u64 r9, +struct tdx_module_output *out);bool tdx_accept_memory (phys_addr_t start,phys_addr_t end);
Signature · 2 lines
Sathyanarayanan Kuppuswamy Linux Kernel Developer
Re: [PATCH v3 12/12] x86/virt/tdx: Adjust 'struct tdx_module_args' to use x86 "register index" layout
In reply to Kai HuangQuoted 113 lines · Kai Huang
For TDX guest, KVM needs to call
__seamcall_saved_ret()to make the TDH.VP.ENTER SEAMCALL to enter the guest, possibly taking all registers in 'structtdx_module_args' as input/output.KVM caches guest's GPRs in 'kvm_vcpu_arch::regs[]', which follows the "register index" hardware layout of x86 GPRs. On the other hand, the
__seamcall_saved_ret()takes the pointer of 'structtdx_module_args' as argument, thus there's a mismatch.KVM could choose to copy input registers from 'vcpu::regs[]' to a 'struct
tdx_module_args' and use that as argument to make the SEAMCALL, but such memory copy isn't desired and should be avoided if possible.It's not feasible to change KVM's 'vcpu::regs[]' layout due to various reasons (e.g., emulation code uses decoded register index as array index to access the register). Therefore, adjust 'struct
tdx_module_args' to match KVM's 'vcpu::regs[]' layout so that KVM can simply do below:__seamcall_saved_ret(TDH_VP_ENTER, (struct tdx_module_args *)vcpu->arch.regs);Note RAX/RSP/RBP are not used by the
TDX_MODULE_CALLassembly, but they are necessary in order match the layout of 'structtdx_module_args' to KVM's 'vcpu::regs[]'. Thus add them to the structure, but name them as*_unused. Also don't include them toasm-offset.cso that any misuse of them in the assembly would result in build error.Cc Kirill A. Shutemov <[email protected]>Cc Dave Hansen <[email protected]>Cc Peter Zijlstra <[email protected]>Cc Sean Christopherson <[email protected]>Cc Paolo Bonzini <[email protected]>Cc Isaku Yamahata <[email protected]>Suggested-by Peter Zijlstra <[email protected]>Signed-off-by Kai Huang <[email protected]>---
v2 -> v3:
- New patch--- arch/x86/include/asm/shared/tdx.h | 19 +++++++++++++------ arch/x86/kernel/asm-offsets.c | 6 +++--- 2 files changed, 16 insertions(+), 9 deletions(-)arch/x86/include/asm/shared/tdx.h +13
-6@@ -58,24 +58,31 @@* Used in __tdcall*() to gather the input/output registers' values of the * TDCALL instruction when requesting services from the TDX module. This is a * software only structure and not part of the TDX module/VMM ABI +* +* Note those *_unused are not used by the TDX_MODULE_CALL assembly. +* The layout of this structure also matches KVM's kvm_vcpu_arch::regs[] +* layout, which follows the "register index" order of x86 GPRs. KVM +* then can simply type cast kvm_vcpu_arch::regs[] to this structure to +* avoid the extra memory copy between two structures when making +* TDH.VP.ENTER SEAMCALL. */ struct tdx_module_args { -/* callee-clobbered */ +u64 rax_unused;u64 rcx;u64 rdx; +u64 rbx; +u64 rsp_unused; +u64 rbp_unused; +u64 rsi; +u64 rdi;u64 r8;u64 r9; -/* extra callee-clobbered */ u64 r10;u64 r11; -/* callee-saved + rdi/rsi */ u64 r12;u64 r13;u64 r14;u64 r15; -u64 rbx; -u64 rdi; -u64 rsi; };/* Used to communicate with the TDX module */ arch/x86/kernel/asm-offsets.c +3
-3@@ -70,6 +70,9 @@ static void __used common(void)BLANK ();OFFSET (TDX_MODULE_rcx, tdx_module_args, rcx);OFFSET (TDX_MODULE_rdx, tdx_module_args, rdx); +OFFSET (TDX_MODULE_rbx, tdx_module_args, rbx); +OFFSET (TDX_MODULE_rsi, tdx_module_args, rsi); +OFFSET (TDX_MODULE_rdi, tdx_module_args, rdi);OFFSET (TDX_MODULE_r8, tdx_module_args, r8);OFFSET (TDX_MODULE_r9, tdx_module_args, r9);OFFSET (TDX_MODULE_r10, tdx_module_args, r10);@@ -78,9 +81,6 @@ static void __used common(void)OFFSET (TDX_MODULE_r13, tdx_module_args, r13);OFFSET (TDX_MODULE_r14, tdx_module_args, r14);OFFSET (TDX_MODULE_r15, tdx_module_args, r15); -OFFSET (TDX_MODULE_rbx, tdx_module_args, rbx); -OFFSET (TDX_MODULE_rdi, tdx_module_args, rdi); -OFFSET (TDX_MODULE_rsi, tdx_module_args, rsi);BLANK ();OFFSET (BP_scratch, boot_params, scratch);Signature · 1 line
2.41.0
I replaced the current TDX KVM TDH.VP.ENTER with this function and it worked.
Signature · 1 line
Isaku Yamahata <[email protected]>
Re: [PATCH v3 12/12] x86/virt/tdx: Adjust 'struct tdx_module_args' to use x86 "register index" layout
In reply to Kai HuangQuoted 18 lines · Kai Huang
For TDX guest, KVM needs to call
__seamcall_saved_ret()to make the TDH.VP.ENTER SEAMCALL to enter the guest, possibly taking all registers in 'structtdx_module_args' as input/output.KVM caches guest's GPRs in 'kvm_vcpu_arch::regs[]', which follows the "register index" hardware layout of x86 GPRs. On the other hand, the
__seamcall_saved_ret()takes the pointer of 'structtdx_module_args' as argument, thus there's a mismatch.KVM could choose to copy input registers from 'vcpu::regs[]' to a 'struct
tdx_module_args' and use that as argument to make the SEAMCALL, but such memory copy isn't desired and should be avoided if possible.It's not feasible to change KVM's 'vcpu::regs[]' layout due to various reasons (e.g., emulation code uses decoded register index as array index to access the register). Therefore, adjust 'struct
tdx_module_args' to match KVM's 'vcpu::regs[]' layout so that KVM can simply do below:
Quoted 8 lines
__seamcall_saved_ret(TDH_VP_ENTER, (struct tdx_module_args *)vcpu->arch.regs);Note RAX/RSP/RBP are not used by the
TDX_MODULE_CALLassembly, but they are necessary in order match the layout of 'structtdx_module_args' to KVM's 'vcpu::regs[]'. Thus add them to the structure, but name them as*_unused. Also don't include them toasm-offset.cso that any misuse of them in the assembly would result in build error.
Maybe we can have static check if the offsets match. e.g. BUILD_BUG_ON(__VCPU_REGS_RAX * 8 != TDX_MODULE_rax); etc...
Anyway, I can have such a patch in TDX KVM side when I use this function for TDH.VP.ENTER.
Thansk,
Quoted 88 lines
Cc Kirill A. Shutemov <[email protected]>Cc Dave Hansen <[email protected]>Cc Peter Zijlstra <[email protected]>Cc Sean Christopherson <[email protected]>Cc Paolo Bonzini <[email protected]>Cc Isaku Yamahata <[email protected]>Suggested-by Peter Zijlstra <[email protected]>Signed-off-by Kai Huang <[email protected]>---
v2 -> v3:
- New patch--- arch/x86/include/asm/shared/tdx.h | 19 +++++++++++++------ arch/x86/kernel/asm-offsets.c | 6 +++--- 2 files changed, 16 insertions(+), 9 deletions(-)arch/x86/include/asm/shared/tdx.h +13
-6@@ -58,24 +58,31 @@* Used in __tdcall*() to gather the input/output registers' values of the * TDCALL instruction when requesting services from the TDX module. This is a * software only structure and not part of the TDX module/VMM ABI +* +* Note those *_unused are not used by the TDX_MODULE_CALL assembly. +* The layout of this structure also matches KVM's kvm_vcpu_arch::regs[] +* layout, which follows the "register index" order of x86 GPRs. KVM +* then can simply type cast kvm_vcpu_arch::regs[] to this structure to +* avoid the extra memory copy between two structures when making +* TDH.VP.ENTER SEAMCALL. */ struct tdx_module_args { -/* callee-clobbered */ +u64 rax_unused;u64 rcx;u64 rdx; +u64 rbx; +u64 rsp_unused; +u64 rbp_unused; +u64 rsi; +u64 rdi;u64 r8;u64 r9; -/* extra callee-clobbered */ u64 r10;u64 r11; -/* callee-saved + rdi/rsi */ u64 r12;u64 r13;u64 r14;u64 r15; -u64 rbx; -u64 rdi; -u64 rsi; };/* Used to communicate with the TDX module */ arch/x86/kernel/asm-offsets.c +3
-3@@ -70,6 +70,9 @@ static void __used common(void)BLANK ();OFFSET (TDX_MODULE_rcx, tdx_module_args, rcx);OFFSET (TDX_MODULE_rdx, tdx_module_args, rdx); +OFFSET (TDX_MODULE_rbx, tdx_module_args, rbx); +OFFSET (TDX_MODULE_rsi, tdx_module_args, rsi); +OFFSET (TDX_MODULE_rdi, tdx_module_args, rdi);OFFSET (TDX_MODULE_r8, tdx_module_args, r8);OFFSET (TDX_MODULE_r9, tdx_module_args, r9);OFFSET (TDX_MODULE_r10, tdx_module_args, r10);@@ -78,9 +81,6 @@ static void __used common(void)OFFSET (TDX_MODULE_r13, tdx_module_args, r13);OFFSET (TDX_MODULE_r14, tdx_module_args, r14);OFFSET (TDX_MODULE_r15, tdx_module_args, r15); -OFFSET (TDX_MODULE_rbx, tdx_module_args, rbx); -OFFSET (TDX_MODULE_rdi, tdx_module_args, rdi); -OFFSET (TDX_MODULE_rsi, tdx_module_args, rsi);BLANK ();OFFSET (BP_scratch, boot_params, scratch);Signature · 1 line
2.41.0
Signature · 1 line
Isaku Yamahata <[email protected]>
Re: [PATCH v3 12/12] x86/virt/tdx: Adjust 'struct tdx_module_args' to use x86 "register index" layout
In reply to Isaku YamahataQuoted 5 lines
I replaced the current TDX KVM TDH.VP.ENTER with this function and it worked.
I suppose you mean:
Tested-by: ... :-)
Anyway I only sent this series out after I got confirmation from you that VP.ENTER worked using the SEAMCALL function in this series, but I forgot to add your Tested-by before sending out this series. Will add. Thanks!
Re: [PATCH v3 05/12] x86/tdx: Pass TDCALL/SEAMCALL input/output registers via a structure
In reply to Huang, KaiQuoted 33 lines · Kai
Quoted 21 lines · On Thu, 2023-07-27 at 19:36 +0300, [email protected] wrote:
Quoted 20 lines · Kai Huang
arch/x86/virt/vmx/tdx/tdxcall.S +12
-0@@ -17,34 +17,33 @@ * TDX module and hypercalls to the VMM. * SEAMCALL - used by TDX hosts to make requests to the * TDX module. + * + *------------------------------------------------------------------------- + * TDCALL/SEAMCALL ABI: + *------------------------------------------------------------------------- + * Input Registers: + * + * RAX - TDCALL/SEAMCALL Leaf number. + * RCX,RDX,R8-R9 - TDCALL/SEAMCALL Leaf specific input registers. + * + * Output Registers: + * + * RAX - TDCALL/SEAMCALL instruction error code.Quoted 9 lines
arch/x86/virt/vmx/tdx/tdxcall.S line 34
Quoted 3 lines
arch/x86/virt/vmx/tdx/tdxcall.S +3
-0+ * RCX,RDX,R8-R11 - TDCALL/SEAMCALL Leaf specific output registers. + * + *-------------------------------------------------------------------------So, you keep the existing asymetry in IN and OUT registers. R10 and R11 are OUT-only registers. It can be confusing for user since it is the same structure now. I guess it changes in the following patches, but I would prefer to make them even here if there's no good reason not to.
Quoted 1 line
Do you mean you prefer to use R10/R11 as input too even in this patch?
Yes.
Quoted 4 lines
I think _logically_ it should be part of the next patch, because w/o extending
TDX_MODULE_CALLto support additional TDCALLs/SEAMCALLs, we don't need R10/R11 as input. This patch only changes to take a structure as function argument, rather than taking individual registers as argument.
As a user, if I see a struct used for in and out, I would expect that all fields have the same rules.
Quoted 20 lines
Also, we have a comment to say this around the structure too:
/*
- * Used in__tdx_module_call()to gather the output registers' values of the
+ * Used in __tdcall*() to gather the input/output registers' values of the
* TDCALL instruction when requesting services from the TDX module. This is a
* software only structure and not part of the TDX module/VMM ABI */arch/x86/include/asm/shared/tdx.h +3
-1-struct tdx_module_output { +struct tdx_module_args { +/* input/output */ u64 rcx;u64 rdx;u64 r8;u64 r9; +/* additional output */ u64 r10;u64 r11; };
Quoted 1 line
So to me there should be no confusion.
Do you always read documentation? :P Maybe it is only me...
Quoted 5 lines
Even consider a theoretical case: someone wants to backport this patch to an old kernel w/o further patches, then it makes little sense to do R10/R11 in
TDX_MODULE_CALLhere in this patch:-)
Consider the case whe the patch was (wrongly) backported to use new call that uses R10 as input.
I realize that all my objections are rather hand-wavy. I would like to have in/out symmetry here. But I would not NAK patch over this.
Signature · 1 line
Kiryl Shutsemau / Kirill A. Shutemov
Re: [PATCH v3 05/12] x86/tdx: Pass TDCALL/SEAMCALL input/output registers via a structure
Quoted 74 lines · On Thu, 2023-08-03 at 13:58 +0300, [email protected] wrote:
Quoted 33 lines · Kai
Quoted 21 lines · On Thu, 2023-07-27 at 19:36 +0300, [email protected] wrote:
Quoted 20 lines · Kai Huang
arch/x86/virt/vmx/tdx/tdxcall.S +12
-0@@ -17,34 +17,33 @@ * TDX module and hypercalls to the VMM. * SEAMCALL - used by TDX hosts to make requests to the * TDX module. + * + *------------------------------------------------------------------------- + * TDCALL/SEAMCALL ABI: + *------------------------------------------------------------------------- + * Input Registers: + * + * RAX - TDCALL/SEAMCALL Leaf number. + * RCX,RDX,R8-R9 - TDCALL/SEAMCALL Leaf specific input registers. + * + * Output Registers: + * + * RAX - TDCALL/SEAMCALL instruction error code.Quoted 9 lines
arch/x86/virt/vmx/tdx/tdxcall.S line 34
Quoted 3 lines
arch/x86/virt/vmx/tdx/tdxcall.S +3
-0+ * RCX,RDX,R8-R11 - TDCALL/SEAMCALL Leaf specific output registers. + * + *-------------------------------------------------------------------------So, you keep the existing asymetry in IN and OUT registers. R10 and R11 are OUT-only registers. It can be confusing for user since it is the same structure now. I guess it changes in the following patches, but I would prefer to make them even here if there's no good reason not to.
Quoted 1 line
Do you mean you prefer to use R10/R11 as input too even in this patch?
Yes.
Quoted 4 lines
I think _logically_ it should be part of the next patch, because w/o extending
TDX_MODULE_CALLto support additional TDCALLs/SEAMCALLs, we don't need R10/R11 as input. This patch only changes to take a structure as function argument, rather than taking individual registers as argument.As a user, if I see a struct used for in and out, I would expect that all fields have the same rules.
Quoted 20 lines
Also, we have a comment to say this around the structure too:
/*
- * Used in__tdx_module_call()to gather the output registers' values of the
+ * Used in __tdcall*() to gather the input/output registers' values of the
* TDCALL instruction when requesting services from the TDX module. This is a
* software only structure and not part of the TDX module/VMM ABI */arch/x86/include/asm/shared/tdx.h +3
-1-struct tdx_module_output { +struct tdx_module_args { +/* input/output */ u64 rcx;u64 rdx;u64 r8;u64 r9; +/* additional output */ u64 r10;u64 r11; };Quoted 1 line
So to me there should be no confusion.
Do you always read documentation? :P Maybe it is only me...
Quoted 4 lines
Even consider a theoretical case: someone wants to backport this patch to an old kernel w/o further patches, then it makes little sense to do R10/R11 in
TDX_MODULE_CALLhere in this patch
Quoted 8 lines
Quoted 1 line
:-)
Consider the case whe the patch was (wrongly) backported to use new call that uses R10 as input.
I realize that all my objections are rather hand-wavy. I would like to have in/out symmetry here. But I would not NAK patch over this.
The only concern is I don't particularly like to add additional logic to this patch. Anyway not big deal to me. I can do what you said if I don't see Peter's (or other maintainers') comment on this.
Btw, should I say something like below in the changelog to justify this additional logic:
Also use R10/R11 as input registers too to make the input/output registers symmetric, although currently no TDCALL/SEAMCALL use them as input registers.
?
Re: [PATCH v3 07/12] x86/tdx: Make TDX_HYPERCALL asm similar to TDX_MODULE_CALL
In reply to Huang, KaiQuoted 10 lines · Kai
Quoted 6 lines · On Thu, 2023-07-27 at 20:10 +0300, [email protected] wrote:
Quoted 2 lines · Kai Huang
Remove the
__tdx_hypercall_ret()as__tdx_hypercall()already does so.Hm. So we now update struct on all VMCALLs. Is it a good idea?
Do you mean we "unconditionally save output registers to the structure", right?
Quoted 8 lines
Quoted 3 lines
We give
more control to VMM where it is not needed.I don't quite follow this. Can you elaborate?
Do you worry about VMM being malicious and putting malicious values to the registers?
Yes. Caller of the hypercall may expect that the register is in-only and re-use the field for other stuff. And it would work until VMM decide otherwise.
Quoted 5 lines
Quoted 3 lines
I would rather keep the struct
read-only where possible.We can achieve this if there's a clean way to do, but I don't see that.
Keep _ret() and non-_ret() versions?
Signature · 1 line
Kiryl Shutsemau / Kirill A. Shutemov
Re: [PATCH v3 05/12] x86/tdx: Pass TDCALL/SEAMCALL input/output registers via a structure
In reply to Huang, KaiQuoted 8 lines · Kai
Btw, should I say something like below in the changelog to justify this additional logic:
Also use R10/R11 as input registers too to make the input/output registers symmetric, although currently no TDCALL/SEAMCALL use them as input registers.
?
LGTM.
Signature · 1 line
Kiryl Shutsemau / Kirill A. Shutemov
Re: [PATCH v3 07/12] x86/tdx: Make TDX_HYPERCALL asm similar to TDX_MODULE_CALL
In reply to [email protected]Quoted 15 lines · On Thu, 2023-08-03 at 14:45 +0300, [email protected] wrote:
Quoted 14 lines · Kai
Quoted 6 lines · On Thu, 2023-07-27 at 20:10 +0300, [email protected] wrote:
Quoted 2 lines · Kai Huang
Remove the
__tdx_hypercall_ret()as__tdx_hypercall()already does so.Hm. So we now update struct on all VMCALLs. Is it a good idea?
Do you mean we "unconditionally save output registers to the structure", right?
Quoted 3 lines
We give
more control to VMM where it is not needed.
Quoted 8 lines
Quoted 4 lines
I don't quite follow this. Can you elaborate?
Do you worry about VMM being malicious and putting malicious values to the registers?
Yes. Caller of the hypercall may expect that the register is in-only and re-use the field for other stuff. And it would work until VMM decide otherwise.
I would argue from this way: in TDX_HYPERCALL the guest has shared all registers to the VMM, thus the caller of hypercall cannot assume those registers is in- only.
Quoted 8 lines
Quoted 6 lines
Quoted 4 lines
I would rather keep the struct
read-only where possible.We can achieve this if there's a clean way to do, but I don't see that.
Keep
_ret()and non-_ret() versions?
The problem is the assembly needs to always turn on the "\ret" so that the R10 (used as VP.VMCALL leaf return code) can be saved to the structure. Otherwise we are not able to return VP.VMCALL leaf return code.
Can you elaborate how to do?
Re: [PATCH v3 07/12] x86/tdx: Make TDX_HYPERCALL asm similar to TDX_MODULE_CALL
Quoted 5 lines · Kai
Quoted 4 lines · On Thu, 2023-08-03 at 14:45 +0300, [email protected] wrote:
Quoted 4 lines
Quoted 3 lines
I would rather keep the struct
read-only where possible.
Quoted 7 lines
Quoted 3 lines
Quoted 1 line
We can achieve this if there's a clean way to do, but I don't see that.
Keep
_ret()and non-_ret() versions?The problem is the assembly needs to always turn on the "\ret" so that the R10 (used as VP.VMCALL leaf return code) can be saved to the structure. Otherwise we are not able to return VP.VMCALL leaf return code.
Yeah. This is downside of single assembly macro for all calls.
One possible way is to make it in C: non-_ret() version pass to the assembly helper copy of the caller's struct, keeping original intact. But, yeah, it is ugly.
Signature · 1 line
Kiryl Shutsemau / Kirill A. Shutemov
Re: [PATCH v3 07/12] x86/tdx: Make TDX_HYPERCALL asm similar to TDX_MODULE_CALL
Quoted 14 lines · On Thu, 2023-08-03 at 15:12 +0300, [email protected] wrote:
Quoted 5 lines · Kai
Quoted 4 lines · On Thu, 2023-08-03 at 14:45 +0300, [email protected] wrote:
Quoted 4 lines
Quoted 3 lines
I would rather keep the struct
read-only where possible.Quoted 7 lines
Quoted 3 lines
Quoted 1 line
We can achieve this if there's a clean way to do, but I don't see that.
Keep
_ret()and non-_ret() versions?The problem is the assembly needs to always turn on the "\ret" so that the R10 (used as VP.VMCALL leaf return code) can be saved to the structure. Otherwise we are not able to return VP.VMCALL leaf return code.
Quoted 6 lines
Yeah. This is downside of single assembly macro for all calls.
One possible way is to make it in C: non-_ret() version pass to the assembly helper copy of the caller's struct, keeping original intact. But, yeah, it is ugly.
You sure you want to do this? :-)
Re: [PATCH v3 07/12] x86/tdx: Make TDX_HYPERCALL asm similar to TDX_MODULE_CALL
Quoted 15 lines · Kai
Quoted 14 lines · On Thu, 2023-08-03 at 15:12 +0300, [email protected] wrote:
Quoted 12 lines · Kai
Quoted 7 lines · On Thu, 2023-08-03 at 14:45 +0300, [email protected] wrote:
Quoted 5 lines
Quoted 3 lines
I would rather keep the struct
read-only where possible.We can achieve this if there's a clean way to do, but I don't see that.
Keep
_ret()and non-_ret() versions?The problem is the assembly needs to always turn on the "\ret" so that the R10 (used as VP.VMCALL leaf return code) can be saved to the structure. Otherwise we are not able to return VP.VMCALL leaf return code.
Quoted 8 lines
Quoted 6 lines
Yeah. This is downside of single assembly macro for all calls.
One possible way is to make it in C: non-_ret() version pass to the assembly helper copy of the caller's struct, keeping original intact. But, yeah, it is ugly.
You sure you want to do this? :-)
No, I am not.
Maybe somebody else has better ideas.
Signature · 1 line
Kiryl Shutsemau / Kirill A. Shutemov
Re: [PATCH v3 07/12] x86/tdx: Make TDX_HYPERCALL asm similar to TDX_MODULE_CALL
Quoted 23 lines · On Thu, 2023-08-03 at 16:47 +0300, [email protected] wrote:
Quoted 22 lines · Kai
Quoted 20 lines · On Thu, 2023-08-03 at 15:12 +0300, [email protected] wrote:
Quoted 5 lines · Kai
Quoted 4 lines · On Thu, 2023-08-03 at 14:45 +0300, [email protected] wrote:
Quoted 4 lines
Quoted 3 lines
I would rather keep the struct
read-only where possible.Quoted 7 lines
Quoted 3 lines
Quoted 1 line
We can achieve this if there's a clean way to do, but I don't see that.
Keep
_ret()and non-_ret() versions?The problem is the assembly needs to always turn on the "\ret" so that the R10 (used as VP.VMCALL leaf return code) can be saved to the structure. Otherwise we are not able to return VP.VMCALL leaf return code.
Yeah. This is downside of single assembly macro for all calls.
One possible way is to make it in C: non-_ret() version pass to the assembly helper copy of the caller's struct, keeping original intact. But, yeah, it is ugly.
Quoted 6 lines
Quoted 1 line
You sure you want to do this? :-)
No, I am not.
Maybe somebody else has better ideas.
Anyway thanks for feedback. So I don't see there's any real problem with this patch, thus I will just keep the current way.
Re: [PATCH v3 08/12] x86/tdx: Reimplement __tdx_hypercall() using TDX_MODULE_CALL asm
In reply to Kai HuangQuoted 20 lines · Kai Huang
Now the
TDX_HYPERCALLasm is basically identical to theTDX_MODULE_CALLwith both '\saved' and '\ret' enabled, with two minor things though:1) The way to restore the structure pointer is different
The
TDX_HYPERCALLuses RCX as spare to restore the structure pointer, but theTDX_MODULE_CALLassumes no spare register can be used. In other words,TDX_MODULE_CALLalready covers whatTDX_HYPERCALLdoes.2)
TDX_MODULE_CALLonly clears shared registers for TDH.VP.ENTERFor this just need to make that code available for the non-host case.
Thus, remove the
TDX_HYPERCALLand reimplement the__tdx_hypercall()using theTDX_MODULE_CALL.Extend the
TDX_MODULE_CALLto cover "clear shared registers" for TDG.VP.VMCALL. Introduce a new__tdcall_saved_ret()to replace the temporary__tdcall_hypercall().
Quoted 8 lines
The
__tdcall_saved_ret()can also be used for those new TDCALLs which require more input/output registers than the basic TDCALLs do.Cc Kirill A. Shutemov <[email protected]>Cc Dave Hansen <[email protected]>Cc Peter Zijlstra <[email protected]>Suggested-by Peter Zijlstra <[email protected]>Signed-off-by Kai Huang <[email protected]>
Signature · 1 line
Kiryl Shutsemau / Kirill A. Shutemov
Re: [PATCH v3 09/12] x86/tdx: Remove 'struct tdx_hypercall_args'
In reply to Kai HuangQuoted 9 lines · Kai Huang
Now 'struct
tdx_hypercall_args' is basically 'structtdx_module_args' minus RCX. Although from __tdx_hypercall()'s perspective RCX isn't used as shared register thus not part of input/output registers, it's not worth to have a separate structure just due to one register.Remove the 'struct
tdx_hypercall_args' and use 'structtdx_module_args' instead in__tdx_hypercall()related code. This also saves the memory copy between the two structures within__tdx_hypercall().
Quoted 5 lines
Cc Kirill A. Shutemov <[email protected]>Cc Dave Hansen <[email protected]>Cc Peter Zijlstra <[email protected]>Suggested-by Peter Zijlstra <[email protected]>Signed-off-by Kai Huang <[email protected]>
Signature · 1 line
Kiryl Shutsemau / Kirill A. Shutemov
Re: [PATCH v3 10/12] x86/virt/tdx: Wire up basic SEAMCALL functions
In reply to Kai HuangQuoted 57 lines · Kai Huang
Intel Trust Domain Extensions (TDX) protects guest VMs from malicious host and certain physical attacks. A CPU-attested software module called 'the TDX module' runs inside a new isolated memory range as a trusted hypervisor to manage and run protected VMs.
TDX introduces a new CPU mode: Secure Arbitration Mode (SEAM). This mode runs only the TDX module itself or other code to load the TDX module.
The host kernel communicates with SEAM software via a new SEAMCALL instruction. This is conceptually similar to a
guest->hosthypercall, except it is made from the host to SEAM software instead. The TDX module establishes a new SEAMCALL ABI which allows the host to initialize the module and to manage VMs.The SEAMCALL ABI is very similar to the TDCALL ABI and leverages much TDCALL infrastructure. Wire up basic functions to make SEAMCALLs for the basic support of running TDX guests:
__seamcall(),__seamcall_ret(), and__seamcall_saved_ret()for TDH.VP.ENTER. All SEAMCALLs involved in the basic TDX support don't use "callee-saved" registers as input and output, except the TDH.VP.ENTER.To start to support TDX, create a new
arch/x86/virt/vmx/tdx/tdx.cfor TDX host kernel support. Add a newKconfigoptionCONFIG_INTEL_TDX_HOSTto opt-in TDX host kernel support (to distinguish with TDX guest kernel support). So far only KVM uses TDX. Make the new config option depend onKVM_INTEL.Signed-off-by Kai Huang <[email protected]>---
v2 -> v3:
- Added__seamcall_saved_ret()back for TDH.VP.ENTER, given the new patch to adjust 'structtdx_module_args' layout.v1 -> v2:
- Removed__seamcall_saved_ret()and leave it to KVM TDX patches.--- arch/x86/Kconfig | 12 +++++++ arch/x86/Makefile | 2 ++ arch/x86/include/asm/tdx.h | 7 ++++ arch/x86/virt/Makefile | 2 ++ arch/x86/virt/vmx/Makefile | 2 ++ arch/x86/virt/vmx/tdx/Makefile | 2 ++ arch/x86/virt/vmx/tdx/seamcall.S | 61 ++++++++++++++++++++++++++++++++ 7 files changed, 88 insertions(+) create mode 100644 arch/x86/virt/Makefile create mode 100644 arch/x86/virt/vmx/Makefile create mode 100644 arch/x86/virt/vmx/tdx/Makefile create mode 100644 arch/x86/virt/vmx/tdx/seamcall.Sarch/x86/Kconfig +0
-0
arch/x86/Kconfig config X86_SGX · line 1956
Quoted 8 lines
arch/x86/Kconfig +5
-0@@ -1949,6 +1949,18 @@ config X86_SGX If unsure, say N. +config INTEL_TDX_HOST + bool "Intel Trust Domain Extensions (TDX) host support" + depends on CPU_SUP_INTEL + depends on X86_64 + depends on KVM_INTEL
Hm. I expected KVM_INTEL to depend on CPU_SUP_INTEL, but apparently no. Any reasons why?
Anyway, it is not strictly related to the patch:
Signature · 1 line
Kiryl Shutsemau / Kirill A. Shutemov
Re: [PATCH v3 11/12] x86/virt/tdx: Allow SEAMCALL to handle #UD and #GP
In reply to Kai Huangarch/x86/include/asm/tdx.h line 25
Quoted 6 lines · Kai Huang
arch/x86/include/asm/tdx.h +2
-0@@ -20,6 +21,9 @@#define TDX_SW_ERROR (TDX_ERROR | GENMASK_ULL(47, 40))#define TDX_SEAMCALL_VMFAILINVALID (TDX_SW_ERROR | _UL(0xFFFF0000)) +#define TDX_SEAMCALL_GP (TDX_SW_ERROR | X86_TRAP_GP) +#define TDX_SEAMCALL_UD (TDX_SW_ERROR | X86_TRAP_UD)
Is there any explantion how these error codes got chosen? Looks very arbitrary and may collide with other error codes in the future.
Signature · 1 line
Kiryl Shutsemau / Kirill A. Shutemov
Re: [PATCH v3 12/12] x86/virt/tdx: Adjust 'struct tdx_module_args' to use x86 "register index" layout
In reply to Kai HuangQuoted 4 lines · Kai Huang
For TDX guest, KVM needs to call
__seamcall_saved_ret()to make the TDH.VP.ENTER SEAMCALL to enter the guest, possibly taking all registers in 'structtdx_module_args' as input/output.
Quoted 8 lines
KVM caches guest's GPRs in 'kvm_vcpu_arch::regs[]', which follows the "register index" hardware layout of x86 GPRs. On the other hand, the
__seamcall_saved_ret()takes the pointer of 'structtdx_module_args' as argument, thus there's a mismatch.KVM could choose to copy input registers from 'vcpu::regs[]' to a 'struct
tdx_module_args' and use that as argument to make the SEAMCALL, but such memory copy isn't desired and should be avoided if possible.
I doubt the copy will be visible on any profile.
I personally don't like that kvm implementation detail leaks here. It suppose to be generic TDX code.
Signature · 1 line
Kiryl Shutsemau / Kirill A. Shutemov
Re: [PATCH v3 10/12] x86/virt/tdx: Wire up basic SEAMCALL functions
In reply to [email protected]Quoted 10 lines
arch/x86/Kconfig config X86_SGX · line 1956
Quoted 7 lines
arch/x86/Kconfig +5
-0+config INTEL_TDX_HOST + bool "Intel Trust Domain Extensions (TDX) host support" + depends on CPU_SUP_INTEL + depends on X86_64 + depends on KVM_INTELHm. I expected
KVM_INTELto depend onCPU_SUP_INTEL, but apparently no. Any reasons why?
Hmm.. Not sure :-)
Re: [PATCH v3 11/12] x86/virt/tdx: Allow SEAMCALL to handle #UD and #GP
In reply to [email protected]Quoted 11 lines · On Sun, 2023-08-06 at 14:41 +0300, [email protected] wrote:
Review line 25
Quoted 6 lines · Kai Huang
(quoted diff) +2
-0@@ -20,6 +21,9 @@ #define TDX_SW_ERROR (TDX_ERROR | GENMASK_ULL(47, 40)) #define TDX_SEAMCALL_VMFAILINVALID (TDX_SW_ERROR | _UL(0xFFFF0000)) +#define TDX_SEAMCALL_GP (TDX_SW_ERROR | X86_TRAP_GP) +#define TDX_SEAMCALL_UD (TDX_SW_ERROR | X86_TRAP_UD)Is there any explantion how these error codes got chosen? Looks very arbitrary and may collide with other error codes in the future.
Any error code has TDX_SW_ERROR is reserved to software use so the TDX module can never return any error code which conflicts with those software ones.
For why to choose these two, I believe XOR the TRAP number to TDX_SW_ERROR is the simplest way to achieve: 1) costing minimal assembly code; 2) opportunistically handling #GP too, allowing caller to distinguish the two errors.
I can add this to the changelog.
Btw, as I chatted to you I believe we have another justification to handle #UD/#GP in the assembly: emergency virtualization disable. Thus we can even get rid of the erratum staff in the changelog.
How does below look like?
commit d3ff21da1083a525eb2cac6576490045e22f6f5d
Author: Kai Huang <[email protected]> Date: Mon Jun 26 16:04:08 2023 +1200
x86/virt/tdx: Allow SEAMCALL to handle #UD and #GP
SEAMCALL instruction causes #UD if the CPU isn't in VMX operation. Currently the TDX_MODULE_CALL assembly doesn't handle #UD, thus making SEAMCALL when VMX is disabled would cause Oops.
Unfortunately, there are legal cases that SEAMCALL can be made when VMX is disabled. For instance, VMX can be disabled due to emergency reboot while there are still TDX guest is running.
Extend the TDX_MODULE_CALL assembly to return an error code for #UD to handle this case gracefully, e.g., KVM can then quitely eat all SEAMCALL errors caused by emergency reboot.
SEAMCALL instruction also causes #GP when TDX isn't enabled by the BIOS. Use _ASM_EXTABLE_FAULT() to catch both exceptions with the trap number recorded, and define two new error codes by XORing the trap number to the TDX_SW_ERROR. This opportunistically handles #GP too while using the same simple assembly code.
A bonus is when kernel mistakenly calls SEAMCALL when CPU isn't in VMX operation, or when TDX isn't enabled by the BIOS, or when the BIOS is buggy, the kernel can get a nicer error code rather than a less understandable Oops.
This is basically based on Peter's code.
arch/x86/include/asm/tdx.h +4 -0
@@ -8,6 +8,7 @@#include <asm/errno.h> #include <asm/ptrace.h> +#include <asm/trapnr.h> #include <asm/shared/tdx.h> /*
@@ -20,6 +21,9 @@#define TDX_SW_ERROR (TDX_ERROR | GENMASK_ULL(47, 40))#define TDX_SEAMCALL_VMFAILINVALID (TDX_SW_ERROR | _UL(0xFFFF0000)) +#define TDX_SEAMCALL_GP (TDX_SW_ERROR | X86_TRAP_GP) +#define TDX_SEAMCALL_UD (TDX_SW_ERROR | X86_TRAP_UD) +#ifndef __ASSEMBLY__/*
arch/x86/virt/vmx/tdx/tdxcall.S +19 -0
@@ -1,6 +1,7 @@ /* SPDX-License-Identifier: GPL-2.0 */ #include <asm/asm-offsets.h> #include <asm/frame.h> +#include <asm/asm.h> #include <asm/tdx.h> /*
@@ -85,6 +86,7 @@ .endif /* \saved */ .if \host +.Lseamcall\@: seamcall /* * SEAMCALL instruction is essentially a VMExit from VMX root
@@ -191,11 +193,28 @@ .if \host .Lseamcall_vmfailinvalid\@: mov $TDX_SEAMCALL_VMFAILINVALID, %rax + jmp .Lseamcall_fail\@ + +.Lseamcall_trap\@: + /* + * SEAMCALL caused #GP or #UD. By reaching here RAX contains + * the trap number. Convert the trap number to the TDX error + * code by setting TDX_SW_ERROR to the high 32-bits of RAX. + * + * Note cannot OR TDX_SW_ERROR directly to RAX as OR instruction + * only accepts 32-bit immediate at most. + */ + movq $TDX_SW_ERROR, %rdi + orq %rdi, %rax + +.Lseamcall_fail\@: .if \ret && \saved /* pop the unused structure pointer back to RSI */ popq %rsi .endif jmp .Lout\@ + + _ASM_EXTABLE_FAULT(.Lseamcall\@, .Lseamcall_trap\@) .endif /* \host */ .endm
Re: [PATCH v3 12/12] x86/virt/tdx: Adjust 'struct tdx_module_args' to use x86 "register index" layout
In reply to [email protected]Quoted 10 lines · On Sun, 2023-08-06 at 14:50 +0300, [email protected] wrote:
Quoted 9 lines · Kai Huang
For TDX guest, KVM needs to call
__seamcall_saved_ret()to make the TDH.VP.ENTER SEAMCALL to enter the guest, possibly taking all registers in 'structtdx_module_args' as input/output.KVM caches guest's GPRs in 'kvm_vcpu_arch::regs[]', which follows the "register index" hardware layout of x86 GPRs. On the other hand, the
__seamcall_saved_ret()takes the pointer of 'structtdx_module_args' as argument, thus there's a mismatch.
Quoted 10 lines
Quoted 3 lines
KVM could choose to copy input registers from 'vcpu::regs[]' to a 'struct
tdx_module_args' and use that as argument to make the SEAMCALL, but such memory copy isn't desired and should be avoided if possible.I doubt the copy will be visible on any profile.
I personally don't like that kvm implementation detail leaks here. It suppose to be generic TDX code.
Well I kinda agree with you. But it seems Peter wanted this to be done:
https://lore.kernel.org/lkml/[email protected]/T/#m37f39493e9f2bf0a4c9ccc72aaf4938927375dc1
Re: [PATCH v3 11/12] x86/virt/tdx: Allow SEAMCALL to handle #UD and #GP
In reply to Huang, KaiQuoted 13 lines · Kai
Quoted 11 lines · On Sun, 2023-08-06 at 14:41 +0300, [email protected] wrote:
arch/x86/include/asm/tdx.h line 25
Quoted 6 lines · Kai Huang
arch/x86/include/asm/tdx.h +2
-0@@ -20,6 +21,9 @@#define TDX_SW_ERROR (TDX_ERROR | GENMASK_ULL(47, 40))#define TDX_SEAMCALL_VMFAILINVALID (TDX_SW_ERROR | _UL(0xFFFF0000)) +#define TDX_SEAMCALL_GP (TDX_SW_ERROR | X86_TRAP_GP) +#define TDX_SEAMCALL_UD (TDX_SW_ERROR | X86_TRAP_UD)Is there any explantion how these error codes got chosen? Looks very arbitrary and may collide with other error codes in the future.
Quoted 7 lines
Any error code has
TDX_SW_ERRORis reserved to software use so the TDX module can never return any error code which conflicts with those software ones.For why to choose these two, I believe XOR the TRAP number to
TDX_SW_ERRORis the simplest way to achieve: 1) costing minimal assembly code; 2) opportunistically handling #GP too, allowing caller to distinguish the two errors.
My problem is that it is going to conflict with errno-based errors if we going to take this path in the future. Like these errors are the same as (TDX_SW_ERROR | EACCES) and (TDX_SW_ERROR | ENXIO) respectively.
Signature · 1 line
Kiryl Shutsemau / Kirill A. Shutemov
Re: [PATCH v3 11/12] x86/virt/tdx: Allow SEAMCALL to handle #UD and #GP
Quoted 17 lines · On Mon, 2023-08-07 at 12:53 +0300, [email protected] wrote:
Quoted 16 lines · Kai
Quoted 11 lines · On Sun, 2023-08-06 at 14:41 +0300, [email protected] wrote:
arch/x86/include/asm/tdx.h line 25
Quoted 6 lines · Kai Huang
arch/x86/include/asm/tdx.h +2
-0@@ -20,6 +21,9 @@#define TDX_SW_ERROR (TDX_ERROR | GENMASK_ULL(47, 40))#define TDX_SEAMCALL_VMFAILINVALID (TDX_SW_ERROR | _UL(0xFFFF0000)) +#define TDX_SEAMCALL_GP (TDX_SW_ERROR | X86_TRAP_GP) +#define TDX_SEAMCALL_UD (TDX_SW_ERROR | X86_TRAP_UD)Is there any explantion how these error codes got chosen? Looks very arbitrary and may collide with other error codes in the future.
Any error code has
TDX_SW_ERRORis reserved to software use so the TDX module can never return any error code which conflicts with those software ones.
Quoted 9 lines
Quoted 4 lines
For why to choose these two, I believe XOR the TRAP number to
TDX_SW_ERRORis the simplest way to achieve: 1) costing minimal assembly code; 2) opportunistically handling #GP too, allowing caller to distinguish the two errors.My problem is that it is going to conflict with errno-based errors if we going to take this path in the future. Like these errors are the same as (
TDX_SW_ERROR| EACCES) and (TDX_SW_ERROR| ENXIO) respectively.
Is there any use case that we need those definitions?
Even we have such requirement in the future, we still have many bits available after taking out the bits of TDX_SW_ERROR thus I assume we can do some bit shift when this really happens??
Re: [PATCH v3 11/12] x86/virt/tdx: Allow SEAMCALL to handle #UD and #GP
Quoted 28 lines · Kai
Quoted 17 lines · On Mon, 2023-08-07 at 12:53 +0300, [email protected] wrote:
Quoted 16 lines · Kai
Quoted 11 lines · On Sun, 2023-08-06 at 14:41 +0300, [email protected] wrote:
arch/x86/include/asm/tdx.h line 25
Quoted 6 lines · Kai Huang
arch/x86/include/asm/tdx.h +2
-0@@ -20,6 +21,9 @@#define TDX_SW_ERROR (TDX_ERROR | GENMASK_ULL(47, 40))#define TDX_SEAMCALL_VMFAILINVALID (TDX_SW_ERROR | _UL(0xFFFF0000)) +#define TDX_SEAMCALL_GP (TDX_SW_ERROR | X86_TRAP_GP) +#define TDX_SEAMCALL_UD (TDX_SW_ERROR | X86_TRAP_UD)Is there any explantion how these error codes got chosen? Looks very arbitrary and may collide with other error codes in the future.
Any error code has
TDX_SW_ERRORis reserved to software use so the TDX module can never return any error code which conflicts with those software ones.Quoted 9 lines
Quoted 4 lines
For why to choose these two, I believe XOR the TRAP number to
TDX_SW_ERRORis the simplest way to achieve: 1) costing minimal assembly code; 2) opportunistically handling #GP too, allowing caller to distinguish the two errors.My problem is that it is going to conflict with errno-based errors if we going to take this path in the future. Like these errors are the same as (
TDX_SW_ERROR| EACCES) and (TDX_SW_ERROR| ENXIO) respectively.
Quoted 5 lines
Is there any use case that we need those definitions?
Even we have such requirement in the future, we still have many bits available after taking out the bits of
TDX_SW_ERRORthus I assume we can do some bit shift when this really happens??
Okay, fair enough.
Signature · 1 line
Kiryl Shutsemau / Kirill A. Shutemov
Re: [PATCH v3 10/12] x86/virt/tdx: Wire up basic SEAMCALL functions
In reply to Huang, KaiQuoted 11 lines · Kai Huang
Quoted 10 lines
arch/x86/Kconfig config X86_SGX · line 1956
Quoted 7 lines
arch/x86/Kconfig +5
-0+config INTEL_TDX_HOST + bool "Intel Trust Domain Extensions (TDX) host support" + depends on CPU_SUP_INTEL + depends on X86_64 + depends on KVM_INTELHm. I expected
KVM_INTELto depend onCPU_SUP_INTEL, but apparently no. Any reasons why?
Quoted 1 line
Hmm.. Not sure :-)
Centuar and Zhaoxin CPUs also support VMX.
commit 8f63aaf5c493c6502a058585cdfa3c71cdf8c44a
Author: Sean Christopherson <[email protected]>
Date: Fri Dec 20 20:45:13 2019 -0800
KVM: VMX: Allow KVM_INTEL when building for Centaur and/or Zhaoxin CPUs
Change the dependency for KVM_INTEL, i.e. KVM w/ VMX, from Intel CPUs to any CPU that supports the IA32_FEAT_CTL MSR and thus VMX functionality. This effectively allows building KVM_INTEL for Centaur and Zhaoxin CPUs.
Re: [PATCH v3 11/12] x86/virt/tdx: Allow SEAMCALL to handle #UD and #GP
In reply to [email protected]Quoted 9 lines · On 8/6/23 04:41, [email protected] wrote:
arch/x86/include/asm/tdx.h line 25
Quoted 6 lines · Kai Huang
arch/x86/include/asm/tdx.h +2
-0@@ -20,6 +21,9 @@#define TDX_SW_ERROR (TDX_ERROR | GENMASK_ULL(47, 40))#define TDX_SEAMCALL_VMFAILINVALID (TDX_SW_ERROR | _UL(0xFFFF0000)) +#define TDX_SEAMCALL_GP (TDX_SW_ERROR | X86_TRAP_GP) +#define TDX_SEAMCALL_UD (TDX_SW_ERROR | X86_TRAP_UD)Is there any explantion how these error codes got chosen? Looks very arbitrary and may collide with other error codes in the future.
If they collide, we can just fix it then.
So, please, do comment what the limitations are and what must be avoided in the future, but I don't think we need to go mucking with this at all.
Re: [PATCH v3 10/12] x86/virt/tdx: Wire up basic SEAMCALL functions
In reply to Sean ChristophersonQuoted 20 lines · Sean Christopherson
Quoted 11 lines · Kai Huang
Quoted 10 lines
arch/x86/Kconfig config X86_SGX · line 1956
Quoted 7 lines
arch/x86/Kconfig +5
-0+config INTEL_TDX_HOST + bool "Intel Trust Domain Extensions (TDX) host support" + depends on CPU_SUP_INTEL + depends on X86_64 + depends on KVM_INTELHm. I expected
KVM_INTELto depend onCPU_SUP_INTEL, but apparently no. Any reasons why?Quoted 1 line
Hmm.. Not sure :-)
Centuar and Zhaoxin CPUs also support VMX.
commit
8f63aaf5c493c6502a058585cdfa3c71cdf8c44a
Author: Sean Christopherson <[email protected]>
Date: Fri Dec 20 20:45:13 2019 -0800
Quoted 6 lines
KVM: VMX: Allow
KVM_INTELwhen building for Centaur and/or Zhaoxin CPUsChange the dependency for
KVM_INTEL, i.e. KVM w/ VMX, from Intel CPUs to any CPU that supports theIA32_FEAT_CTLMSR and thus VMX functionality. This effectively allows buildingKVM_INTELfor Centaur and Zhaoxin CPUs.
Ah. Thanks Sean!