linux.git/arch/x86/kernel/paravirt_patch.c, branch v5.5

x86/paravirt: Standardize 'insn_buff' variable names

2019-04-29T14:05:49+00:00

We currently have 6 (!) separate naming variants to name temporary instruction
buffers that are used for code patching:

 - insnbuf
 - insnbuff
 - insn_buff
 - insn_buffer
 - ibuf
 - ibuffer

These are used as local variables, percpu fields and function parameters.

Standardize all the names to a single variant: 'insn_buff'.

Cc: Andy Lutomirski 
Cc: Borislav Petkov 
Cc: Brian Gerst 
Cc: Dave Hansen 
Cc: Denys Vlasenko 
Cc: H. Peter Anvin 
Cc: Juergen Gross 
Cc: Linus Torvalds 
Cc: Peter Zijlstra 
Cc: Rik van Riel 
Cc: Thomas Gleixner 
Signed-off-by: Ingo Molnar

x86/paravirt: Match paravirt patchlet field definition ordering to initialization ordering

2019-04-25T10:00:44+00:00

Here's the objdump -D output of the PATCH_XXL data table:

0000000000000010 :
  10:   fa                      cli
  11:   fb                      sti
  12:   57                      push   %rdi
  13:   9d                      popfq
  14:   9c                      pushfq
  15:   58                      pop    %rax
  16:   0f 20 d0                mov    %cr2,%rax
  19:   0f 20 d8                mov    %cr3,%rax
  1c:   0f 22 df                mov    %rdi,%cr3
  1f:   0f 09                   wbinvd
  21:   0f 01 f8                swapgs
  24:   48 0f 07                sysretq
  27:   0f 01 f8                swapgs
  2a:   48 89 f8                mov    %rdi,%rax

Note how this doesn't match up to the source code:

static const struct patch_xxl patch_data_xxl = {
        .irq_irq_disable        = { 0xfa },             // cli
        .irq_irq_enable         = { 0xfb },             // sti
        .irq_save_fl            = { 0x9c, 0x58 },       // pushf; pop %[re]ax
        .mmu_read_cr2           = { 0x0f, 0x20, 0xd0 }, // mov %cr2, %[re]ax
        .mmu_read_cr3           = { 0x0f, 0x20, 0xd8 }, // mov %cr3, %[re]ax
        .irq_restore_fl         = { 0x57, 0x9d },       // push %rdi; popfq
        .mmu_write_cr3          = { 0x0f, 0x22, 0xdf }, // mov %rdi, %cr3
        .cpu_wbinvd             = { 0x0f, 0x09 },       // wbinvd
        .cpu_usergs_sysret64    = { 0x0f, 0x01, 0xf8,
                                    0x48, 0x0f, 0x07 }, // swapgs; sysretq
        .cpu_swapgs             = { 0x0f, 0x01, 0xf8 }, // swapgs
        .mov64                  = { 0x48, 0x89, 0xf8 }, // mov %rdi, %rax
        .irq_restore_fl         = { 0x50, 0x9d },       // push %eax; popf
        .mmu_write_cr3          = { 0x0f, 0x22, 0xd8 }, // mov %eax, %cr3
        .cpu_iret               = { 0xcf },             // iret
};

Note how they are reordered: in the generated code .irq_restore_fl comes
before .irq_save_fl, etc. This is because the field ordering in struct
patch_xxl does not match the initialization ordering of patch_data_xxl.

Match up the initialization order with the definition order - this makes
the disassembly easily reviewable:

0000000000000010 :
  10:   fa                      cli
  11:   fb                      sti
  12:   9c                      pushfq
  13:   58                      pop    %rax
  14:   0f 20 d0                mov    %cr2,%rax
  17:   0f 20 d8                mov    %cr3,%rax
  1a:   0f 22 df                mov    %rdi,%cr3
  1d:   57                      push   %rdi
  1e:   9d                      popfq
  1f:   0f 09                   wbinvd
  21:   0f 01 f8                swapgs
  24:   48 0f 07                sysretq
  27:   0f 01 f8                swapgs
  2a:   48 89 f8                mov    %rdi,%rax

Cc: Andy Lutomirski 
Cc: Borislav Petkov 
Cc: Brian Gerst 
Cc: Dave Hansen 
Cc: Denys Vlasenko 
Cc: H. Peter Anvin 
Cc: Juergen Gross 
Cc: Linus Torvalds 
Cc: Peter Zijlstra 
Cc: Rik van Riel 
Cc: Thomas Gleixner 
Link: http://lkml.kernel.org/r/20190425081012.GA115378@gmail.com
Signed-off-by: Ingo Molnar

x86/paravirt: Replace the paravirt patch asm magic

2019-04-25T10:00:44+00:00

The magic macro DEF_NATIVE() in the paravirt patching code uses inline
assembly to generate a data table for patching in the native instructions.

While clever this is falling apart with LTO and even aside of LTO the
construct is just working by chance according to GCC folks.

Aside of that the tables are constant data and not some form of magic
text.

As these constructs are not subject to frequent changes it is not a
maintenance issue to convert them to regular data tables which are
initialized with hex bytes.

Create a new set of macros and data structures to store the instruction
sequences and convert the code over.

Reported-by: Andi Kleen 
Signed-off-by: Thomas Gleixner 
Cc: Andy Lutomirski 
Cc: Borislav Petkov 
Cc: Brian Gerst 
Cc: Dave Hansen 
Cc: Denys Vlasenko 
Cc: H. Peter Anvin 
Cc: Juergen Gross 
Cc: Linus Torvalds 
Cc: Peter Zijlstra 
Cc: Rik van Riel 
Link: http://lkml.kernel.org/r/20190424134223.690835713@linutronix.de
Signed-off-by: Ingo Molnar

x86/paravirt: Unify the 32/64 bit paravirt patching code

2019-04-25T10:00:44+00:00

Large parts of these two files are identical. Merge them together.

Signed-off-by: Thomas Gleixner 
Cc: Andy Lutomirski 
Cc: Borislav Petkov 
Cc: Brian Gerst 
Cc: Dave Hansen 
Cc: Denys Vlasenko 
Cc: H. Peter Anvin 
Cc: Juergen Gross 
Cc: Linus Torvalds 
Cc: Peter Zijlstra 
Cc: Rik van Riel 
Link: http://lkml.kernel.org/r/20190424134223.603491680@linutronix.de
Signed-off-by: Ingo Molnar