Openwall GNU/*/Linux - a small security-enhanced Linux distro for servers
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Tue, 17 Jul 2018 22:35:26 -0400
From: Rich Felker <dalias@...c.org>
To: musl@...ts.openwall.com
Subject: Re: [PATCH] arm asm for vfork

On Wed, Jul 18, 2018 at 12:20:00PM +1000, Patrick Oppenlander wrote:
> Hi Rich,
> 
> I saw another thread where it was mentioned you may be doing a 1.20
> release some time soon.
> 
> Is there any chance this could get merged in time? I've been running
> it for months without any issues now.
> 
> Attached is an updated patch using svc rather than swi.

Thanks. One detail:

> diff --git a/src/process/arm/vfork.s b/src/process/arm/vfork.s
> new file mode 100644
> index 00000000..db4e7b43
> --- /dev/null
> +++ b/src/process/arm/vfork.s
> @@ -0,0 +1,12 @@
> +.syntax unified
> +.global __vfork
> +.weak vfork
> +.type __vfork,%function
> +.type vfork,%function
> +__vfork:
> +vfork:
> +	mov ip, r7
> +	mov r7, 190
> +	svc 0
> +	mov r7, ip
> +	b __syscall_ret
> -- 

I think there needs to be a ".hidden __syscall_ret" (by de facto musl
convention, on the line before it's used) here. It *might* be ok
having the reference omit .hidden as long as the definition is hidden
at link-time (which it is), but I'm not convinced the tooling won't
complain about a branch to a destination that's not known to be
link-time constant displacement.

If you have no other changes or comments I'm happy to just --amend
that into the patch when I commit it.

Rich

Powered by blists - more mailing lists

Your e-mail address:

Powered by Openwall GNU/*/Linux - Powered by OpenVZ