The old code restored all the registers with movq instead of pop.
In theory, this was done because some CPUs have higher movq
throughput, but any gain there would be tiny and is almost certainly
outweighed by the higher text size.
This saves 96 bytes of text.
Signed-off-by: Andy Lutomirski <l...@kernel.org>
---
arch/x86/entry/calling.h | 21 +++++++++++++++++++++
arch/x86/entry/entry_64.S | 12 ++++++------
2 files changed, 27 insertions(+), 6 deletions(-)
diff --git a/arch/x86/entry/calling.h b/arch/x86/entry/calling.h
index 640aafebdc00..0b9dd8123701 100644
--- a/arch/x86/entry/calling.h
+++ b/arch/x86/entry/calling.h
@@ -151,6 +151,27 @@ For 32-bit we have the following conventions - kernel is
built with
UNWIND_HINT_REGS offset=\offset extra=0
.endm
+ .macro POP_EXTRA_REGS
+ popq %r15
+ popq %r14
+ popq %r13
+ popq %r12
+ popq %rbp
+ popq %rbx
+ .endm
+
+ .macro POP_C_REGS
+ popq %r11
+ popq %r10
+ popq %r9
+ popq %r8
+ popq %rax
+ popq %rcx
+ popq %rdx
+ popq %rsi
+ popq %rdi
+ .endm
+
.macro RESTORE_C_REGS_HELPER rstor_rax=1, rstor_rcx=1, rstor_r11=1,
rstor_r8910=1, rstor_rdx=1
.if \rstor_r11
movq 6*8(%rsp), %r11
diff --git a/arch/x86/entry/entry_64.S b/arch/x86/entry/entry_64.S
index 1909a4e42b81..4ad40067162a 100644
--- a/arch/x86/entry/entry_64.S
+++ b/arch/x86/entry/entry_64.S
@@ -615,9 +615,9 @@ GLOBAL(swapgs_restore_regs_and_return_to_usermode)
1:
#endif
SWAPGS
- RESTORE_EXTRA_REGS
- RESTORE_C_REGS
- REMOVE_PT_GPREGS_FROM_STACK 8
+ POP_EXTRA_REGS
+ POP_C_REGS
+ addq $8, %rsp
INTERRUPT_RETURN
@@ -646,9 +646,9 @@ GLOBAL(restore_regs_and_return_to_kernel)
ud2
1:
#endif
- RESTORE_EXTRA_REGS
- RESTORE_C_REGS
- REMOVE_PT_GPREGS_FROM_STACK 8
+ POP_EXTRA_REGS
+ POP_C_REGS
+ addq $8, %rsp
INTERRUPT_RETURN
ENTRY(native_iret)
--
2.13.6
