powerpc: syscall_dotrace() && retcode (Was: powerpc: fork && stepping)

[Oleg Nesterov]

On 11/30, Benjamin Herrenschmidt wrote:
>
> On Sun, 2009-11-29 at 22:07 +0100, Oleg Nesterov wrote:
> >
> > If I change the test-case to use NEWVAL == 1000 (or any other value
> > greater than NR_syscalls), then the tracee sees ENOSYS and this is
> > correct too.
> >
> > But I do not see how it is possible to change the retcode on powerpc.
> > Unlike x86, powepc doesn't set -ENOSYS "in advance", before doing
> > do_syscall_trace_enter() logic. This means that if the tracer "cancels"
> > syscall, r3 will be overwritten by syscall_enosys.
> >
> > This probably means the kernel should be fixed too, but I am not
> > brave enough to change the asm which I can't understand ;)
>
> Yes, the asm should be changed. I suppose we could check if the result
> of do_syscall_trace_enter is negative, and if it is, branch to the exit
> path using r3 as the error code. Would that be ok ?
>
> Something like this:
>
> --- a/arch/powerpc/kernel/entry_64.S
> +++ b/arch/powerpc/kernel/entry_64.S
> @@ -240,6 +240,9 @@ syscall_dotrace:
>  	bl	.save_nvgprs
>  	addi	r3,r1,STACK_FRAME_OVERHEAD
>  	bl	.do_syscall_trace_enter
> +	cmpdi	cr0,r3,0
> +	blt	syscall_exit
> +

Yes, but this doesn't allow to a) cancel this syscall and b) make it
return a non-negative result to the tracee.

Perhaps poweprc should set pt_regs->result = -ENOSYS before calling
do_syscall_trace_enter() like x86 does ? (in this case syscall_exit()
shouldn't change RESULT(r1)). This way the tracer can change both
pt_regs->result and gpr[0] independently.

Oleg.