freebsd-dev/usr.bin/truss/ia64-fbsd.c
Marcel Moolenaar f2c49dd248 Revamp of the syscall path, exception and context handling. The
prime objectives are:
o  Implement a syscall path based on the epc inststruction (see
   sys/ia64/ia64/syscall.s).
o  Revisit the places were we need to save and restore registers
   and define those contexts in terms of the register sets (see
   sys/ia64/include/_regset.h).

Secundairy objectives:
o  Remove the requirement to use contigmalloc for kernel stacks.
o  Better handling of the high FP registers for SMP systems.
o  Switch to the new cpu_switch() and cpu_throw() semantics.
o  Add a good unwinder to reconstruct contexts for the rare
   cases we need to (see sys/contrib/ia64/libuwx)

Many files are affected by this change. Functionally it boils
down to:
o  The EPC syscall doesn't preserve registers it does not need
   to preserve and places the arguments differently on the stack.
   This affects libc and truss.
o  The address of the kernel page directory (kptdir) had to
   be unstaticized for use by the nested TLB fault handler.
   The name has been changed to ia64_kptdir to avoid conflicts.
   The renaming affects libkvm.
o  The trapframe only contains the special registers and the
   scratch registers. For syscalls using the EPC syscall path
   no scratch registers are saved. This affects all places where
   the trapframe is accessed. Most notably the unaligned access
   handler, the signal delivery code and the debugger.
o  Context switching only partly saves the special registers
   and the preserved registers. This affects cpu_switch() and
   triggered the move to the new semantics, which additionally
   affects cpu_throw().
o  The high FP registers are either in the PCB or on some
   CPU. context switching for them is done lazily. This affects
   trap().
o  The mcontext has room for all registers, but not all of them
   have to be defined in all cases. This mostly affects signal
   delivery code now. The *context syscalls are as of yet still
   unimplemented.

Many details went into the removal of the requirement to use
contigmalloc for kernel stacks. The details are mostly CPU
specific and limited to exception_save() and exception_restore().
The few places where we create, destroy or switch stacks were
mostly simplified by not having to construct physical addresses
and additionally saving the virtual addresses for later use.

Besides more efficient context saving and restoring, which of
course yields a noticable speedup, this also fixes the dreaded
SMP bootup problem as a side-effect. The details of which are
still not fully understood.

This change includes all the necessary backward compatibility
code to have it handle older userland binaries that use the
break instruction for syscalls. Support for break-based syscalls
has been pessimized in favor of a clean implementation. Due to
the overall better performance of the kernel, this will still
be notived as an improvement if it's noticed at all.

Approved by: re@ (jhb)
2003-05-16 21:26:42 +00:00

343 lines
9.6 KiB
C

/*
* Copryight 1997 Sean Eric Fagan
*
* 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.
* 3. All advertising materials mentioning features or use of this software
* must display the following acknowledgement:
* This product includes software developed by Sean Eric Fagan
* 4. Neither the name of the author may be used to endorse or promote
* products derived from this software without specific prior written
* permission.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR 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 AUTHOR 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.
*/
#ifndef lint
static const char rcsid[] =
"$FreeBSD$";
#endif /* not lint */
/*
* FreeBSD/ia64-specific system call handling. This is probably the most
* complex part of the entire truss program, although I've got lots of
* it handled relatively cleanly now. The system call names are generated
* automatically, thanks to /usr/src/sys/kern/syscalls.master. The
* names used for the various structures are confusing, I sadly admit.
*/
#include <sys/types.h>
#include <sys/ioctl.h>
#include <sys/pioctl.h>
#include <sys/syscall.h>
#include <machine/reg.h>
#include <errno.h>
#include <fcntl.h>
#include <signal.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <time.h>
#include <unistd.h>
#include "truss.h"
#include "syscall.h"
#include "extern.h"
static int fd = -1;
static int cpid = -1;
extern int Procfd;
#include "syscalls.h"
static int nsyscalls = sizeof(syscallnames) / sizeof(syscallnames[0]);
/*
* This is what this particular file uses to keep track of a system call.
* It is probably not quite sufficient -- I can probably use the same
* structure for the various syscall personalities, and I also probably
* need to nest system calls (for signal handlers).
*
* 'struct syscall' describes the system call; it may be NULL, however,
* if we don't know about this particular system call yet.
*/
static struct freebsd_syscall {
struct syscall *sc;
const char *name;
int number;
unsigned long *args;
int nargs; /* number of arguments -- *not* number of words! */
char **s_args; /* the printable arguments */
} fsc;
/* Clear up and free parts of the fsc structure. */
static __inline void
clear_fsc(void) {
if (fsc.args) {
free(fsc.args);
}
if (fsc.s_args) {
int i;
for (i = 0; i < fsc.nargs; i++)
if (fsc.s_args[i])
free(fsc.s_args[i]);
free(fsc.s_args);
}
memset(&fsc, 0, sizeof(fsc));
}
/*
* Called when a process has entered a system call. nargs is the
* number of words, not number of arguments (a necessary distinction
* in some cases). Note that if the STOPEVENT() code in ia64/ia64/trap.c
* is ever changed these functions need to keep up.
*/
void
ia64_syscall_entry(struct trussinfo *trussinfo, int nargs) {
char buf[32];
struct reg regs;
int syscall_num;
int i;
unsigned int parm_offset;
struct syscall *sc;
if (fd == -1 || trussinfo->pid != cpid) {
sprintf(buf, "/proc/%d/regs", trussinfo->pid);
fd = open(buf, O_RDWR);
if (fd == -1) {
fprintf(trussinfo->outfile, "-- CANNOT OPEN REGISTERS --\n");
return;
}
cpid = trussinfo->pid;
}
clear_fsc();
lseek(fd, 0L, 0);
if (read(fd, &regs, sizeof(regs)) != sizeof(regs)) {
fprintf(trussinfo->outfile, "-- CANNOT READ REGISTERS --\n");
return;
}
parm_offset = regs.r_special.sp + 16;
/*
* FreeBSD has two special kinds of system call redirctions --
* SYS_syscall, and SYS___syscall. The former is the old syscall()
* routine, basicly; the latter is for quad-aligned arguments.
*/
syscall_num = regs.r_scratch.gr15; /* XXX double-check. */
switch (syscall_num) {
case SYS_syscall:
lseek(Procfd, parm_offset, SEEK_SET);
read(Procfd, &syscall_num, sizeof(int));
parm_offset += sizeof(int);
break;
case SYS___syscall:
lseek(Procfd, parm_offset, SEEK_SET);
read(Procfd, &syscall_num, sizeof(int));
parm_offset += sizeof(quad_t);
break;
}
fsc.number = syscall_num;
fsc.name = (syscall_num < 0 || syscall_num > nsyscalls)
? NULL : syscallnames[syscall_num];
if (!fsc.name) {
fprintf(trussinfo->outfile, "-- UNKNOWN SYSCALL %d --\n", syscall_num);
}
if (fsc.name && (trussinfo->flags & FOLLOWFORKS)
&& ((!strcmp(fsc.name, "fork")
|| !strcmp(fsc.name, "rfork")
|| !strcmp(fsc.name, "vfork"))))
{
trussinfo->in_fork = 1;
}
if (nargs == 0)
return;
fsc.args = malloc((1+nargs) * sizeof(unsigned long));
lseek(Procfd, parm_offset, SEEK_SET);
if (read(Procfd, fsc.args, nargs * sizeof(unsigned long)) == -1)
return;
sc = get_syscall(fsc.name);
if (sc) {
fsc.nargs = sc->nargs;
} else {
#if DEBUG
fprintf(trussinfo->outfile, "unknown syscall %s -- setting args to %d\n",
fsc.name, nargs);
#endif
fsc.nargs = nargs;
}
fsc.s_args = malloc((1+fsc.nargs) * sizeof(char*));
memset(fsc.s_args, 0, fsc.nargs * sizeof(char*));
fsc.sc = sc;
/*
* At this point, we set up the system call arguments.
* We ignore any OUT ones, however -- those are arguments that
* are set by the system call, and so are probably meaningless
* now. This doesn't currently support arguments that are
* passed in *and* out, however.
*/
if (fsc.name) {
#if DEBUG
fprintf(stderr, "syscall %s(", fsc.name);
#endif
for (i = 0; i < fsc.nargs; i++) {
#if DEBUG
fprintf(stderr, "0x%x%s",
sc
? fsc.args[sc->args[i].offset]
: fsc.args[i],
i < (fsc.nargs - 1) ? "," : "");
#endif
if (sc && !(sc->args[i].type & OUT)) {
fsc.s_args[i] = print_arg(Procfd, &sc->args[i], fsc.args);
}
}
#if DEBUG
fprintf(stderr, ")\n");
#endif
}
#if DEBUG
fprintf(trussinfo->outfile, "\n");
#endif
/*
* Some system calls should be printed out before they are done --
* execve() and exit(), for example, never return. Possibly change
* this to work for any system call that doesn't have an OUT
* parameter?
*/
if (!strcmp(fsc.name, "execve") || !strcmp(fsc.name, "exit")) {
/* XXX
* This could be done in a more general
* manner but it still wouldn't be very pretty.
*/
if (!strcmp(fsc.name, "execve")) {
if ((trussinfo->flags & EXECVEARGS) == 0)
if (fsc.s_args[1]) {
free(fsc.s_args[1]);
fsc.s_args[1] = NULL;
}
if ((trussinfo->flags & EXECVEENVS) == 0)
if (fsc.s_args[2]) {
free(fsc.s_args[2]);
fsc.s_args[2] = NULL;
}
}
print_syscall(trussinfo, fsc.name, fsc.nargs, fsc.s_args);
fprintf(trussinfo->outfile, "\n");
}
return;
}
/*
* And when the system call is done, we handle it here.
* Currently, no attempt is made to ensure that the system calls
* match -- this needs to be fixed (and is, in fact, why S_SCX includes
* the sytem call number instead of, say, an error status).
*/
int
ia64_syscall_exit(struct trussinfo *trussinfo, int syscall_num __unused) {
char buf[32];
struct reg regs;
int retval;
int i;
int errorp;
struct syscall *sc;
if (fd == -1 || trussinfo->pid != cpid) {
sprintf(buf, "/proc/%d/regs", trussinfo->pid);
fd = open(buf, O_RDONLY);
if (fd == -1) {
fprintf(trussinfo->outfile, "-- CANNOT OPEN REGISTERS --\n");
return (-1);
}
cpid = trussinfo->pid;
}
lseek(fd, 0L, 0);
if (read(fd, &regs, sizeof(regs)) != sizeof(regs)) {
fprintf(trussinfo->outfile, "-- CANNOT READ REGISTERS --\n");
return (-1);
}
retval = regs.r_scratch.gr8;
errorp = (regs.r_scratch.gr10 != 0) ? 1 : 0;
/*
* This code, while simpler than the initial versions I used, could
* stand some significant cleaning.
*/
sc = fsc.sc;
if (!sc) {
for (i = 0; i < fsc.nargs; i++) {
fsc.s_args[i] = malloc(12);
sprintf(fsc.s_args[i], "0x%lx", fsc.args[i]);
}
} else {
/*
* Here, we only look for arguments that have OUT masked in --
* otherwise, they were handled in the syscall_entry function.
*/
for (i = 0; i < sc->nargs; i++) {
char *temp;
if (sc->args[i].type & OUT) {
/*
* If an error occurred, than don't bothe getting the data;
* it may not be valid.
*/
if (errorp) {
temp = malloc(12);
sprintf(temp, "0x%lx", fsc.args[sc->args[i].offset]);
} else {
temp = print_arg(Procfd, &sc->args[i], fsc.args);
}
fsc.s_args[i] = temp;
}
}
}
/*
* It would probably be a good idea to merge the error handling,
* but that complicates things considerably.
*/
print_syscall_ret(trussinfo, fsc.name, fsc.nargs, fsc.s_args, errorp, retval);
clear_fsc();
return (retval);
}