https://gcc.gnu.org/bugzilla/show_bug.cgi?id=99836
Bug ID: 99836
Summary: aarch64: -fpatchable-function-entry=N[,0] should place
.cfi_startproc before NOPs
Product: gcc
Version: 11.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: target
Assignee: unassigned at gcc dot gnu.org
Reporter: i at maskray dot me
Target Milestone: ---
Extracted from https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92424#c8
% echo 'int main() {}' > a.c
% clang --target=aarch64 -fpatchable-function-entry=2
-mbranch-protection=standard -S a.c -o -
...
main: // @main
.Lfunc_begin0:
.cfi_startproc
// %bb.0: // %entry
hint #34
.Lpatch0:
nop
nop
%
/tmp/glibc-many/install/compilers/aarch64-linux-gnu/bin/aarch64-glibc-linux-gnu-g++
-fpatchable-function-entry=2 -mbranch-protection=standard -S a.c -o -
.arch armv8-a
.file "a.c"
.text
.align 2
.global main
.type main, %function
main:
hint 34 // bti c
.section __patchable_function_entries,"aw",@progbits
.align 3
.8byte .LPFE1
.text
.LPFE1:
nop
nop
.LFB0:
.cfi_startproc
For -fpatchable-function-entry=N[,0], placing .cfi_startproc before NOPs makes
more sense and can make unwinding work in that region.
For N[,M] where M>0, that is a very narrow use case by the Linux kernel. I
prefer not to place .cfi_startproc above the function label.
