csu: Implement _start using as to satisfy unwinders on i386

[dchagin]

Summary

The right unwinding stop indicator should be CFI-undefined PC.
https://dwarfstd.org/doc/Dwarf3.pdf - page 118:
If a Return Address register is defined in the virtual unwind table,
and its rule is undefined (for example, by DW_CFA_undefined), then
there is no return address and no call address, and the virtual
unwind of stack activations is complete.

The hack localizing _start1 symbol removed.

[dchagin]

https://reviews.freebsd.org/D40623

[dchagin]

I suggest don't waste time with rewriting _start1 on i386

[kib]

In D40624#928890, @dchagin wrote:

I suggest don't waste time with rewriting _start1 on i386

What do you mean? i386 userspace, unlike kernel, is there with us for long time.

[dchagin]

In D40624#928940, @kib wrote:

In D40624#928890, @dchagin wrote:

I suggest don't waste time with rewriting _start1 on i386

What do you mean? i386 userspace, unlike kernel, is there with us for long time.

Sure, I want to finish our arm64 and libunwind before 14.0 is out to release a libunwind too, and after that I want to finish i386,
but I'm not sure I have enough time for i386 due to arm64 has much more unresolved issues

btw, with latest changes libunwind x86_64 make check is green, thank you

[dchagin]

Implement _start using as

[dchagin]

cfi_offsett for %ebp set to -4, ie to 0

[kib]

lib/csu/Makefile.inc

47

The main reason why I mostly agree with the change, is due to removal of the hack for _start1.

lib/csu/i386/crt1_s.S

81

This comment should go into the #else block.

[dchagin]

csu: Implement _start using as to satisfy unwinders on i386

The right unwinding stop indicator should be CFI-undefined PC.
https://dwarfstd.org/doc/Dwarf3.pdf - page 118:
If a Return Address register is defined in the virtual unwind table,
and its rule is undefined (for example, by DW_CFA_undefined), then
there is no return address and no call address, and the virtual
unwind of stack activations is complete.

The hack localizing _start1 symbol removed.

[kib]

I suppose the latest comment is really a commit message.

lib/csu/i386/crt1_s.S

57

why not use the local 1: label and then calll 1f?

[dchagin]

.L was used due to it not entered into the symbol table, replaced by numbered labels.