[klibc] [PATCH] mips/mips64: Get rid of delay slot hacks

Ben Hutchings ben at decadent.org.uk
Sun Aug 28 13:43:23 PDT 2022

On Sat, 2022-08-13 at 17:45 +0100, Jiaxun Yang wrote:
> There are some MIPS assembly code trying to take advantage of
> delay slots. However nowadays MIPS Release 6 removed delay slots
> so they become an ordering problem.

It didn't remove delay slots.  It added "compact" branch instructions
that don't have them, but after this patch the new instructions still
aren't used.

> Moreover, nowadays assmbler are smart enough to fill delay slots
> properly. All those hacks can actually be archived by binutils.

I can see that it does move instructions into the delay slot for a "j"
instruction.  However, in this case:

> --- a/usr/klibc/arch/mips/syscall.S
> +++ b/usr/klibc/arch/mips/syscall.S
> @@ -1,14 +1,10 @@
>  #include <machine/asm.h>
>  #include <asm/unistd.h>
> -	.set noreorder
>  LEAF(__syscall_common)
>  	syscall
>          beqz    a3, 1f
> -	# sw is actually two instructions; the first one goes
> -	# in the branch delay slot
> -	# XXX: Break this up manually; as it is now it generates warnings.
>          sw      v0, errno
>          li      v0, -1
>  1:      jr      ra

the delay slot gets filled with a nop.  I don't think that matters
much, but this means the commit message is not really accurate.

I think I would be happy to accept this as cleanup if you correct the
commit message.


Ben Hutchings
It is a miracle that curiosity survives formal education.
                                                      - Albert Einstein
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: This is a digitally signed message part
URL: <https://lists.zytor.com/archives/klibc/attachments/20220828/ceecd99f/attachment.sig>

More information about the klibc mailing list