arch/powerpc/linux_machdep.c

/*	$NetBSD: linux_machdep.c,v 1.41 2010/02/02 15:02:07 wiz Exp $ */

/*-
 * Copyright (c) 1995, 2000, 2001 The NetBSD Foundation, Inc.
 * All rights reserved.
 *
 * This code is derived from software contributed to The NetBSD Foundation
 * by Frank van der Linden and Emmanuel Dreyfus.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions
 * are met:
 * 1. Redistributions of source code must retain the above copyright
 *    notice, this list of conditions and the following disclaimer.
 * 2. Redistributions in binary form must reproduce the above copyright
 *    notice, this list of conditions and the following disclaimer in the
 *    documentation and/or other materials provided with the distribution.
 *
 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
 * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
 * POSSIBILITY OF SUCH DAMAGE.
 */

#include <sys/cdefs.h>
__KERNEL_RCSID(0, "$NetBSD: linux_machdep.c,v 1.41 2010/02/02 15:02:07 wiz Exp $");

#include <sys/param.h>
#include <sys/systm.h>
#include <sys/signalvar.h>
#include <sys/kernel.h>
#include <sys/proc.h>
#include <sys/buf.h>
#include <sys/reboot.h>
#include <sys/conf.h>
#include <sys/exec.h>
#include <sys/file.h>
#include <sys/callout.h>
#include <sys/malloc.h>
#include <sys/mbuf.h>
#include <sys/msgbuf.h>
#include <sys/mount.h>
#include <sys/vnode.h>
#include <sys/device.h>
#include <sys/syscallargs.h>
#include <sys/filedesc.h>
#include <sys/exec_elf.h>
#include <sys/disklabel.h>
#include <sys/ioctl.h>
#include <miscfs/specfs/specdev.h>

#include <compat/linux/common/linux_types.h>
#include <compat/linux/common/linux_signal.h>
#include <compat/linux/common/linux_util.h>
#include <compat/linux/common/linux_ioctl.h>
#include <compat/linux/common/linux_hdio.h>
#include <compat/linux/common/linux_exec.h>
#include <compat/linux/common/linux_machdep.h>

#include <compat/linux/linux_syscallargs.h>

#include <sys/cpu.h>
#include <machine/fpu.h>
#include <machine/psl.h>
#include <machine/reg.h>
#include <machine/vmparam.h>

/*
 * To see whether wscons is configured (for virtual console ioctl calls).
 */
#if defined(_KERNEL_OPT)
#include "wsdisplay.h"
#endif
#if (NWSDISPLAY > 0)
#include <dev/wscons/wsconsio.h>
#include <dev/wscons/wsdisplay_usl_io.h>
#endif

/*
 * Set set up registers on exec.
 * XXX not used at the moment since in sys/kern/exec_conf, LINUX_COMPAT
 * entry uses NetBSD's native setregs instead of linux_setregs
 */
void
linux_setregs(struct lwp *l, struct exec_package *pack, u_long stack)
{
	setregs(l, pack, stack);
}

/*
 * Send an interrupt to process.
 *
 * Adapted from arch/powerpc/powerpc/sig_machdep.c:sendsig and
 * compat/linux/arch/i386/linux_machdep.c:linux_sendsig
 *
 * XXX Does not work well yet with RT signals
 *
 */

void
linux_sendsig(const ksiginfo_t *ksi, const sigset_t *mask)
{
	const int sig = ksi->ksi_signo;
	struct lwp *l = curlwp;
	struct proc *p = l->l_proc;
	struct trapframe *tf;
	sig_t catcher = SIGACTION(p, sig).sa_handler;
	struct linux_sigregs frame;
	struct linux_pt_regs linux_regs;
	struct linux_sigcontext sc;
	register_t fp;
	int onstack, error;
	int i;

	tf = trapframe(l);

	/*
	 * Do we need to jump onto the signal stack?
	 */
	onstack =
	    (l->l_sigstk.ss_flags & (SS_DISABLE | SS_ONSTACK)) == 0 &&
	    (SIGACTION(p, sig).sa_flags & SA_ONSTACK) != 0;

	/*
	 * Signal stack is broken (see at the end of linux_sigreturn), so we do
	 * not use it yet. XXX fix this.
	 */
	onstack=0;

	/*
	 * Allocate space for the signal handler context.
	 */
	if (onstack) {
		fp = (register_t)
		    ((char *)l->l_sigstk.ss_sp +
		    l->l_sigstk.ss_size);
	} else {
		fp = tf->fixreg[1];
	}
#ifdef DEBUG_LINUX
	printf("fp at start of linux_sendsig = %x\n", fp);
#endif
	fp -= sizeof(struct linux_sigregs);
	fp &= ~0xf;

	/*
	 * Prepare a sigcontext for later.
	 */
	memset(&sc, 0, sizeof sc);
	sc.lsignal = (int)native_to_linux_signo[sig];
	sc.lhandler = (unsigned long)catcher;
	native_to_linux_old_extra_sigset(&sc.lmask, &sc._unused[3], mask);
	sc.lregs = (struct linux_pt_regs*)fp;

	/*
	 * Setup the signal stack frame as Linux does it in
	 * arch/ppc/kernel/signal.c:setup_frame()
	 *
	 * Save register context.
	 */
	for (i = 0; i < 32; i++)
		linux_regs.lgpr[i] = tf->fixreg[i];
	linux_regs.lnip = tf->srr0;
	linux_regs.lmsr = tf->srr1 & PSL_USERSRR1;
	linux_regs.lorig_gpr3 = tf->fixreg[3]; /* XXX Is that right? */
	linux_regs.lctr = tf->ctr;
	linux_regs.llink = tf->lr;
	linux_regs.lxer = tf->xer;
	linux_regs.lccr = tf->cr;
	linux_regs.lmq = 0;  			/* Unused, 601 only */
	linux_regs.ltrap = tf->exc;
	linux_regs.ldar = tf->dar;
	linux_regs.ldsisr = tf->dsisr;
	linux_regs.lresult = 0;

	memset(&frame, 0, sizeof(frame));
	memcpy(&frame.lgp_regs, &linux_regs, sizeof(linux_regs));

	save_fpu_lwp(curlwp, FPU_SAVE);
	memcpy(&frame.lfp_regs, curpcb->pcb_fpu.fpreg, sizeof(frame.lfp_regs));

	/*
	 * Copy Linux's signal trampoline on the user stack It should not
	 * be used, but Linux binaries might expect it to be there.
	 */
	frame.ltramp[0] = 0x38997777; /* li r0, 0x7777 */
	frame.ltramp[1] = 0x44000002; /* sc */

	/*
	 * Move it to the user stack
	 * There is a little trick here, about the LINUX_ABIGAP: the
	 * linux_sigreg structure has a 56 int gap to support rs6000/xcoff
	 * binaries. But the Linux kernel seems to do without it, and it
	 * just skip it when building the stack frame. Hence the LINUX_ABIGAP.
	 */
	sendsig_reset(l, sig);
	mutex_exit(p->p_lock);
	error = copyout(&frame, (void *)fp, sizeof (frame) - LINUX_ABIGAP);

	if (error != 0) {
		/*
		 * Process has trashed its stack; give it an illegal
		 * instruction to halt it in its tracks.
		 */
		mutex_enter(p->p_lock);
		sigexit(l, SIGILL);
		/* NOTREACHED */
	}

	/*
	 * Add a sigcontext on the stack
	 */
	fp -= sizeof(struct linux_sigcontext);
	error = copyout(&sc, (void *)fp, sizeof (struct linux_sigcontext));
	mutex_enter(p->p_lock);

	if (error != 0) {
		/*
		 * Process has trashed its stack; give it an illegal
		 * instruction to halt it in its tracks.
		 */
		sigexit(l, SIGILL);
		/* NOTREACHED */
	}

	/*
	 * Set the registers according to how the Linux process expects them.
	 * "Mind the gap" Linux expects a gap here.
	 */
	tf->fixreg[1] = fp - LINUX__SIGNAL_FRAMESIZE;
	tf->lr = (int)catcher;
	tf->fixreg[3] = (int)native_to_linux_signo[sig];
	tf->fixreg[4] = fp;
	tf->srr0 = (int)p->p_sigctx.ps_sigcode;

#ifdef DEBUG_LINUX
	printf("fp at end of linux_sendsig = %x\n", fp);
#endif
	/*
	 * Remember that we're now on the signal stack.
	 */
	if (onstack)
		l->l_sigstk.ss_flags |= SS_ONSTACK;
#ifdef DEBUG_LINUX
	printf("linux_sendsig: exitting. fp=0x%lx\n",(long)fp);
#endif
}

/*
 * System call to cleanup state after a signal
 * has been taken.  Reset signal mask and
 * stack state from context left by sendsig (above).
 * Return to previous pc and psl as specified by
 * context left by sendsig. Check carefully to
 * make sure that the user has not modified the
 * psl to gain improper privileges or to cause
 * a machine fault.
 *
 * XXX not tested
 */
int
linux_sys_rt_sigreturn(struct lwp *l, const struct linux_sys_rt_sigreturn_args *uap, register_t *retval)
{
	/* {
		syscallarg(struct linux_rt_sigframe *) sfp;
	} */
	struct proc *p = l->l_proc;
	struct linux_rt_sigframe *scp, sigframe;
	struct linux_sigregs sregs;
	struct linux_pt_regs *lregs;
	struct trapframe *tf;
	sigset_t mask;
	int i;

	/*
	 * The trampoline code hands us the context.
	 * It is unsafe to keep track of it ourselves, in the event that a
	 * program jumps out of a signal handler.
	 */
	scp = SCARG(uap, sfp);

	/*
	 * Get the context from user stack
	 */
	if (copyin((void *)scp, &sigframe, sizeof(*scp)))
		return (EFAULT);

	/*
	 *  Restore register context.
	 */
	if (copyin((void *)sigframe.luc.luc_context.lregs,
		   &sregs, sizeof(sregs)))
		return (EFAULT);
	lregs = (struct linux_pt_regs *)&sregs.lgp_regs;

	tf = trapframe(l);
#ifdef DEBUG_LINUX
	printf("linux_sys_rt_sigreturn: trapframe=0x%lx scp=0x%lx\n",
	    (unsigned long)tf, (unsigned long)scp);
#endif

	if (!PSL_USEROK_P(lregs->lmsr))
		return (EINVAL);

	for (i = 0; i < 32; i++)
		tf->fixreg[i] = lregs->lgpr[i];
	tf->lr = lregs->llink;
	tf->cr = lregs->lccr;
	tf->xer = lregs->lxer;
	tf->ctr = lregs->lctr;
	tf->srr0 = lregs->lnip;
	tf->srr1 = lregs->lmsr;

	/*
	 * Make sure the fpu state is discarded
	 */
	save_fpu_lwp(curlwp, FPU_DISCARD);

	memcpy(curpcb->pcb_fpu.fpreg, (void *)&sregs.lfp_regs,
	       sizeof(curpcb->pcb_fpu.fpreg));

	mutex_enter(p->p_lock);

	/*
	 * Restore signal stack.
	 *
	 * XXX cannot find the onstack information in Linux sig context.
	 * Is signal stack really supported on Linux?
	 *
	 * It seems to be supported in libc6...
	 */
	/* if (sc.sc_onstack & SS_ONSTACK)
		l->l_sigstk.ss_flags |= SS_ONSTACK;
	else */
		l->l_sigstk.ss_flags &= ~SS_ONSTACK;

	/*
	 * Grab the signal mask
	 */
	linux_to_native_sigset(&mask, &sigframe.luc.luc_sigmask);
	(void) sigprocmask1(l, SIG_SETMASK, &mask, 0);

	mutex_exit(p->p_lock);

	return (EJUSTRETURN);
}


/*
 * The following needs code review for potential security issues
 */
int
linux_sys_sigreturn(struct lwp *l, const struct linux_sys_sigreturn_args *uap, register_t *retval)
{
	/* {
		syscallarg(struct linux_sigcontext *) scp;
	} */
	struct proc *p = l->l_proc;
	struct linux_sigcontext *scp, context;
	struct linux_sigregs sregs;
	struct linux_pt_regs *lregs;
	struct trapframe *tf;
	sigset_t mask;
	int i;

	/*
	 * The trampoline code hands us the context.
	 * It is unsafe to keep track of it ourselves, in the event that a
	 * program jumps out of a signal handler.
	 */
	scp = SCARG(uap, scp);

	/*
	 * Get the context from user stack
	 */
	if (copyin(scp, &context, sizeof(*scp)))
		return (EFAULT);

	/*
	 *  Restore register context.
	 */
	if (copyin((void *)context.lregs, &sregs, sizeof(sregs)))
		return (EFAULT);
	lregs = (struct linux_pt_regs *)&sregs.lgp_regs;

	tf = trapframe(l);
#ifdef DEBUG_LINUX
	printf("linux_sys_sigreturn: trapframe=0x%lx scp=0x%lx\n",
	    (unsigned long)tf, (unsigned long)scp);
#endif

	if (!PSL_USEROK_P(lregs->lmsr))
		return (EINVAL);

	for (i = 0; i < 32; i++)
		tf->fixreg[i] = lregs->lgpr[i];
	tf->lr = lregs->llink;
	tf->cr = lregs->lccr;
	tf->xer = lregs->lxer;
	tf->ctr = lregs->lctr;
	tf->srr0 = lregs->lnip;
	tf->srr1 = lregs->lmsr;

	/*
	 * Make sure the fpu state is discarded
	 */
	save_fpu_lwp(curlwp, FPU_DISCARD);

	memcpy(curpcb->pcb_fpu.fpreg, (void *)&sregs.lfp_regs,
	       sizeof(curpcb->pcb_fpu.fpreg));

	mutex_enter(p->p_lock);

	/*
	 * Restore signal stack.
	 *
	 * XXX cannot find the onstack information in Linux sig context.
	 * Is signal stack really supported on Linux?
	 */
#if 0
	if (sc.sc_onstack & SS_ONSTACK)
		l->l_sigstk.ss_flags |= SS_ONSTACK;
	else
#endif
		l->l_sigstk.ss_flags &= ~SS_ONSTACK;

	/* Restore signal mask. */
	linux_old_extra_to_native_sigset(&mask, &context.lmask,
	    &context._unused[3]);
	(void) sigprocmask1(l, SIG_SETMASK, &mask, 0);

	mutex_exit(p->p_lock);

	return (EJUSTRETURN);
}


#if 0
int
linux_sys_modify_ldt(struct proc *p, void *v, register_t *retval)
{
	/*
	 * This syscall is not implemented in Linux/PowerPC: we should not
	 * be here
	 */
#ifdef DEBUG_LINUX
	printf("linux_sys_modify_ldt: should not be here.\n");
#endif
  return 0;
}
#endif

/*
 * major device numbers remapping
 */
dev_t
linux_fakedev(dev_t dev, int raw)
{
	/* XXX write me */
	return dev;
}

/*
 * We come here in a last attempt to satisfy a Linux ioctl() call
 */
int
linux_machdepioctl(struct lwp *l, const struct linux_sys_ioctl_args *uap, register_t *retval)
{
	/* {
		syscallarg(int) fd;
		syscallarg(u_long) com;
		syscallarg(void *) data;
	} */
	struct sys_ioctl_args bia;
	u_long com;

	SCARG(&bia, fd) = SCARG(uap, fd);
	SCARG(&bia, data) = SCARG(uap, data);
	com = SCARG(uap, com);

	switch (com) {
	default:
		printf("linux_machdepioctl: invalid ioctl %08lx\n", com);
		return EINVAL;
	}
	SCARG(&bia, com) = com;
	/* XXX NJWLWP */
	return sys_ioctl(curlwp, &bia, retval);
}
#if 0
/*
 * Set I/O permissions for a process. Just set the maximum level
 * right away (ignoring the argument), otherwise we would have
 * to rely on I/O permission maps, which are not implemented.
 */
int
linux_sys_iopl(struct lwp *l, const void *v, register_t *retval)
{
	/*
	 * This syscall is not implemented in Linux/PowerPC: we should not be here
	 */
#ifdef DEBUG_LINUX
	printf("linux_sys_iopl: should not be here.\n");
#endif
	return 0;
}
#endif

/*
 * See above. If a root process tries to set access to an I/O port,
 * just let it have the whole range.
 */
int
linux_sys_ioperm(struct lwp *l, const struct linux_sys_ioperm_args *uap, register_t *retval)
{
	/*
	 * This syscall is not implemented in Linux/PowerPC: we should not be here
	 */
#ifdef DEBUG_LINUX
	printf("linux_sys_ioperm: should not be here.\n");
#endif
	return 0;
}

/*
 * wrapper linux_sys_new_uname() -> linux_sys_uname()
 */
int
linux_sys_new_uname(struct lwp *l, const struct linux_sys_new_uname_args *uap, register_t *retval)
{
	return linux_sys_uname(l, (const void *)uap, retval);
}

/*
 * wrapper linux_sys_new_select() -> linux_sys_select()
 */
int
linux_sys_new_select(struct lwp *l, const struct linux_sys_new_select_args *uap, register_t *retval)
{
	return linux_sys_select(l, (const void *)uap, retval);
}

int
linux_usertrap(struct lwp *l, vaddr_t trapaddr, void *arg)
{
	return 0;
}