Re: [PATCH] x86/boot: Clean up handling of boot_params pointer

[Brian Gerst]

On Fri, Mar 31, 2023 at 9:00 PM H. Peter Anvin <hpa@xxxxxxxxx> wrote:
>
> On March 31, 2023 11:28:39 AM PDT, Brian Gerst <brgerst@xxxxxxxxx> wrote:
> >On entry from the bootloader, RSI contains the pointer to the
> >boot_params data structure.  Since the RSI register can be clobbered
> >when calling C functions, it is saved and restored around every call.
> >Instead, move it to the R12 register, which is preserved across calls.
> >
> >Signed-off-by: Brian Gerst <brgerst@xxxxxxxxx>
> >---
> > arch/x86/kernel/head_64.S | 29 ++++++++++-------------------
> > 1 file changed, 10 insertions(+), 19 deletions(-)
> >
> >diff --git a/arch/x86/kernel/head_64.S b/arch/x86/kernel/head_64.S
> >index a5df3e994f04..0d130ca2e0a3 100644
> >--- a/arch/x86/kernel/head_64.S
> >+++ b/arch/x86/kernel/head_64.S
> >@@ -49,8 +49,6 @@ SYM_CODE_START_NOALIGN(startup_64)
> >        * for us.  These identity mapped page tables map all of the
> >        * kernel pages and possibly all of memory.
> >        *
> >-       * %rsi holds a physical pointer to real_mode_data.
> >-       *
> >        * We come here either directly from a 64bit bootloader, or from
> >        * arch/x86/boot/compressed/head_64.S.
> >        *
> >@@ -61,6 +59,12 @@ SYM_CODE_START_NOALIGN(startup_64)
> >        * tables and then reload them.
> >        */
> >
> >+      /*
> >+       * RSI holds a physical pointer to real_mode_data.  Move it to R12,
> >+       * which is preserved across C function calls.
> >+       */
> >+      movq    %rsi, %r12
> >+
> >       /* Set up the stack for verify_cpu() */
> >       leaq    (__end_init_task - PTREGS_SIZE)(%rip), %rsp
> >
> >@@ -73,9 +77,7 @@ SYM_CODE_START_NOALIGN(startup_64)
> >       shrq    $32,  %rdx
> >       wrmsr
> >
> >-      pushq   %rsi
> >       call    startup_64_setup_env
> >-      popq    %rsi
> >
> > #ifdef CONFIG_AMD_MEM_ENCRYPT
> >       /*
> >@@ -84,10 +86,8 @@ SYM_CODE_START_NOALIGN(startup_64)
> >        * which needs to be done before any CPUID instructions are executed in
> >        * subsequent code.
> >        */
> >-      movq    %rsi, %rdi
> >-      pushq   %rsi
> >+      movq    %r12, %rdi
> >       call    sme_enable
> >-      popq    %rsi
> > #endif
> >
> >       /* Now switch to __KERNEL_CS so IRET works reliably */
> >@@ -109,9 +109,8 @@ SYM_CODE_START_NOALIGN(startup_64)
> >        * programmed into CR3.
> >        */
> >       leaq    _text(%rip), %rdi
> >-      pushq   %rsi
> >+      movq    %r12, %rsi
> >       call    __startup_64
> >-      popq    %rsi
> >
> >       /* Form the CR3 value being sure to include the CR3 modifier */
> >       addq    $(early_top_pgt - __START_KERNEL_map), %rax
> >@@ -125,8 +124,6 @@ SYM_CODE_START(secondary_startup_64)
> >        * At this point the CPU runs in 64bit mode CS.L = 1 CS.D = 0,
> >        * and someone has loaded a mapped page table.
> >        *
> >-       * %rsi holds a physical pointer to real_mode_data.
> >-       *
> >        * We come here either from startup_64 (using physical addresses)
> >        * or from trampoline.S (using virtual addresses).
> >        *
> >@@ -197,13 +194,9 @@ SYM_INNER_LABEL(secondary_startup_64_no_verify, SYM_L_GLOBAL)
> >        * hypervisor could lie about the C-bit position to perform a ROP
> >        * attack on the guest by writing to the unencrypted stack and wait for
> >        * the next RET instruction.
> >-       * %rsi carries pointer to realmode data and is callee-clobbered. Save
> >-       * and restore it.
> >        */
> >-      pushq   %rsi
> >       movq    %rax, %rdi
> >       call    sev_verify_cbit
> >-      popq    %rsi
> >
> >       /*
> >        * Switch to new page-table
> >@@ -294,9 +287,7 @@ SYM_INNER_LABEL(secondary_startup_64_no_verify, SYM_L_GLOBAL)
> >       wrmsr
> >
> >       /* Setup and Load IDT */
> >-      pushq   %rsi
> >       call    early_setup_idt
> >-      popq    %rsi
> >
> >       /* Check if nx is implemented */
> >       movl    $0x80000001, %eax
> >@@ -332,9 +323,9 @@ SYM_INNER_LABEL(secondary_startup_64_no_verify, SYM_L_GLOBAL)
> >       pushq $0
> >       popfq
> >
> >-      /* rsi is pointer to real mode structure with interesting info.
> >+      /* R12 is pointer to real mode structure with interesting info.
> >          pass it to C */
> >-      movq    %rsi, %rdi
> >+      movq    %r12, %rdi
> >
> > .Ljump_to_C_code:
> >       /*
>
> Would it not make more sense to write it into a memory variable and accessing that variable from the C code by name?

I think ideally we'd want to copy the real mode data as early as
possible.  However I don't know how that would interact with memory
encryption.  By reading the code, I think it would work, but I don't
have the hardware to test it.

--
Brian Gerst