[9fans] another kernel question

[9fans] another kernel question

[Steve Simon]

I have found another piece of code I don't understand in the kernel.

syscalls are all fed through a single trap, and the common code which
processes them performs a waserror():

	/sys/src/9/pc/trap.c:694

A few lines down this function (after the system call has been
executed up->nerrlab is checked to ensure we have matching
poperror()s for each waserror() during the execution of said call.

this is fine for most calls, however rfork() explicitly sets up->nerrlab
to zero rather than copying it and memmove()ing up->errlab from the parent
proc to the child:

	 /sys/src/9/port/sysproc.c:90

Surely this means that rfork will always fail with a "bad errstack [19]: -1 extra"
error?

Clearly this doesn't happen but I doin't understand how why it works.

anyone?

-Steve

Re: [9fans] another kernel question

[Russ Cox]

> this is fine for most calls, however rfork() explicitly sets up->nerrlab
> to zero rather than copying it and memmove()ing up->errlab from the parent
> proc to the child:
>
> 	 /sys/src/9/port/sysproc.c:90
>
> Surely this means that rfork will always fail with a "bad errstack [19]: -1 extra"
> error?

long
sysrfork(ulong *arg)
{
	...

	/* Craft a return frame which will cause the child to pop out of
	 * the scheduler in user mode with the return register zero
	 */
	forkchild(p, up->dbgreg);
	...
	ready(p);
	...
}

void
forkchild(Proc *p, Ureg *ureg)
{
	Ureg *cureg;

	/*
	 * Add 2*BY2WD to the stack to account for
	 *  - the return PC
	 *  - trap's argument (ur)
	 */
	p->sched.sp = (ulong)p->kstack+KSTACK-(sizeof(Ureg)+2*BY2WD);
	p->sched.pc = (ulong)forkret;

	cureg = (Ureg*)(p->sched.sp+2*BY2WD);
	memmove(cureg, ureg, sizeof(Ureg));
	/* return value of syscall in child */
	cureg->ax = 0;

	/* Things from bottom of syscall which were never executed */
	p->psstate = 0;
	p->insyscall = 0;
}

TEXT forkret(SB), $0
	POPL	AX
	POPAL
	POPL	GS
	POPL	FS
	POPL	ES
	POPL	DS
	ADDL	$8, SP			/* pop error code and trap type */
	IRETL

Re: [9fans] another kernel question

[Steve Simon]

> TEXT forkret(SB), $0

very cunning...

thanks russ, I get it now.

-Steve