40c174b38b
This fixes compilation with musl libc. This approach was already used in process.c, so I assume it is safe. * signal.c: Move [HAVE_LINUX_PTRACE_H] code out of [HAVE_SYS_REG_H] check. * syscall.c: Likewise. * util.c: Likewise. Signed-off-by: Maarten ter Huurne <maarten@treewalker.org>
1386 lines
34 KiB
C
1386 lines
34 KiB
C
/*
|
|
* Copyright (c) 1991, 1992 Paul Kranenburg <pk@cs.few.eur.nl>
|
|
* Copyright (c) 1993 Branko Lankester <branko@hacktic.nl>
|
|
* Copyright (c) 1993, 1994, 1995, 1996 Rick Sladkey <jrs@world.std.com>
|
|
* Copyright (c) 1996-1999 Wichert Akkerman <wichert@cistron.nl>
|
|
* Copyright (c) 1999 IBM Deutschland Entwicklung GmbH, IBM Corporation
|
|
* Linux for s390 port by D.J. Barrow
|
|
* <barrow_dj@mail.yahoo.com,djbarrow@de.ibm.com>
|
|
* All rights reserved.
|
|
*
|
|
* 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. The name of the author may not be used to endorse or promote products
|
|
* derived from this software without specific prior written permission.
|
|
*
|
|
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``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 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 "defs.h"
|
|
#include <sys/user.h>
|
|
#include <fcntl.h>
|
|
|
|
#ifdef HAVE_SYS_REG_H
|
|
# include <sys/reg.h>
|
|
#endif
|
|
|
|
#ifdef HAVE_LINUX_PTRACE_H
|
|
# undef PTRACE_SYSCALL
|
|
# ifdef HAVE_STRUCT_IA64_FPREG
|
|
# define ia64_fpreg XXX_ia64_fpreg
|
|
# endif
|
|
# ifdef HAVE_STRUCT_PT_ALL_USER_REGS
|
|
# define pt_all_user_regs XXX_pt_all_user_regs
|
|
# endif
|
|
# ifdef HAVE_STRUCT_PTRACE_PEEKSIGINFO_ARGS
|
|
# define ptrace_peeksiginfo_args XXX_ptrace_peeksiginfo_args
|
|
# endif
|
|
# include <linux/ptrace.h>
|
|
# undef ptrace_peeksiginfo_args
|
|
# undef ia64_fpreg
|
|
# undef pt_all_user_regs
|
|
#endif
|
|
|
|
#ifdef IA64
|
|
# include <asm/ptrace_offsets.h>
|
|
#endif
|
|
|
|
#if defined(SPARC) || defined(SPARC64) || defined(MIPS)
|
|
typedef struct {
|
|
struct pt_regs si_regs;
|
|
int si_mask;
|
|
} m_siginfo_t;
|
|
#elif defined HAVE_ASM_SIGCONTEXT_H
|
|
# if !defined(IA64) && !defined(X86_64) && !defined(X32)
|
|
# include <asm/sigcontext.h>
|
|
# endif
|
|
#else /* !HAVE_ASM_SIGCONTEXT_H */
|
|
# if defined M68K && !defined HAVE_STRUCT_SIGCONTEXT
|
|
struct sigcontext {
|
|
unsigned long sc_mask;
|
|
unsigned long sc_usp;
|
|
unsigned long sc_d0;
|
|
unsigned long sc_d1;
|
|
unsigned long sc_a0;
|
|
unsigned long sc_a1;
|
|
unsigned short sc_sr;
|
|
unsigned long sc_pc;
|
|
unsigned short sc_formatvec;
|
|
};
|
|
# endif /* M68K */
|
|
#endif /* !HAVE_ASM_SIGCONTEXT_H */
|
|
|
|
#ifndef NSIG
|
|
# warning: NSIG is not defined, using 32
|
|
# define NSIG 32
|
|
#elif NSIG < 32
|
|
# error: NSIG < 32
|
|
#endif
|
|
|
|
#ifdef HAVE_SIGACTION
|
|
|
|
/* The libc headers do not define this constant since it should only be
|
|
used by the implementation. So we define it here. */
|
|
#ifndef SA_RESTORER
|
|
# ifdef ASM_SA_RESTORER
|
|
# define SA_RESTORER ASM_SA_RESTORER
|
|
# endif
|
|
#endif
|
|
|
|
/* Some arches define this in their headers, but don't actually have it,
|
|
so we have to delete the define. */
|
|
#if defined(HPPA) || defined(IA64)
|
|
# undef SA_RESTORER
|
|
#endif
|
|
|
|
#include "xlat/sigact_flags.h"
|
|
#include "xlat/sigprocmaskcmds.h"
|
|
|
|
#endif /* HAVE_SIGACTION */
|
|
|
|
/* Anonymous realtime signals. */
|
|
/* Under glibc 2.1, SIGRTMIN et al are functions, but __SIGRTMIN is a
|
|
constant. This is what we want. Otherwise, just use SIGRTMIN. */
|
|
#ifdef SIGRTMIN
|
|
#ifndef __SIGRTMIN
|
|
#define __SIGRTMIN SIGRTMIN
|
|
#define __SIGRTMAX SIGRTMAX /* likewise */
|
|
#endif
|
|
#endif
|
|
|
|
/* Note on the size of sigset_t:
|
|
*
|
|
* In glibc, sigset_t is an array with space for 1024 bits (!),
|
|
* even though all arches supported by Linux have only 64 signals
|
|
* except MIPS, which has 128. IOW, it is 128 bytes long.
|
|
*
|
|
* In-kernel sigset_t is sized correctly (it is either 64 or 128 bit long).
|
|
* However, some old syscall return only 32 lower bits (one word).
|
|
* Example: sys_sigpending vs sys_rt_sigpending.
|
|
*
|
|
* Be aware of this fact when you try to
|
|
* memcpy(&tcp->u_arg[1], &something, sizeof(sigset_t))
|
|
* - sizeof(sigset_t) is much bigger than you think,
|
|
* it may overflow tcp->u_arg[] array, and it may try to copy more data
|
|
* than is really available in <something>.
|
|
* Similarly,
|
|
* umoven(tcp, addr, sizeof(sigset_t), &sigset)
|
|
* may be a bad idea: it'll try to read much more data than needed
|
|
* to fetch a sigset_t.
|
|
* Use (NSIG / 8) as a size instead.
|
|
*/
|
|
|
|
const char *
|
|
signame(const int sig)
|
|
{
|
|
static char buf[sizeof("SIGRT_%u") + sizeof(int)*3];
|
|
|
|
if (sig >= 0) {
|
|
const unsigned int s = sig;
|
|
|
|
if (s < nsignals)
|
|
return signalent[s];
|
|
#ifdef SIGRTMIN
|
|
if (s >= __SIGRTMIN && s <= __SIGRTMAX) {
|
|
sprintf(buf, "SIGRT_%u", s - __SIGRTMIN);
|
|
return buf;
|
|
}
|
|
#endif
|
|
}
|
|
sprintf(buf, "%d", sig);
|
|
return buf;
|
|
}
|
|
|
|
static unsigned int
|
|
popcount32(const uint32_t *a, unsigned int size)
|
|
{
|
|
unsigned int count = 0;
|
|
|
|
for (; size; ++a, --size) {
|
|
uint32_t x = *a;
|
|
|
|
#ifdef HAVE___BUILTIN_POPCOUNT
|
|
count += __builtin_popcount(x);
|
|
#else
|
|
for (; x; ++count)
|
|
x &= x - 1;
|
|
#endif
|
|
}
|
|
|
|
return count;
|
|
}
|
|
|
|
static const char *
|
|
sprintsigmask_n(const char *prefix, const void *sig_mask, unsigned int bytes)
|
|
{
|
|
/*
|
|
* The maximum number of signal names to be printed is NSIG * 2 / 3.
|
|
* Most of signal names have length 7,
|
|
* average length of signal names is less than 7.
|
|
* The length of prefix string does not exceed 16.
|
|
*/
|
|
static char outstr[128 + 8 * (NSIG * 2 / 3)];
|
|
|
|
char *s;
|
|
const uint32_t *mask;
|
|
uint32_t inverted_mask[NSIG / 32];
|
|
unsigned int size;
|
|
int i;
|
|
char sep;
|
|
|
|
s = stpcpy(outstr, prefix);
|
|
|
|
mask = sig_mask;
|
|
/* length of signal mask in 4-byte words */
|
|
size = (bytes >= NSIG / 8) ? NSIG / 32 : (bytes + 3) / 4;
|
|
|
|
/* check whether 2/3 or more bits are set */
|
|
if (popcount32(mask, size) >= size * 32 * 2 / 3) {
|
|
/* show those signals that are NOT in the mask */
|
|
unsigned int j;
|
|
for (j = 0; j < size; ++j)
|
|
inverted_mask[j] = ~mask[j];
|
|
mask = inverted_mask;
|
|
*s++ = '~';
|
|
}
|
|
|
|
sep = '[';
|
|
for (i = 0; (i = next_set_bit(mask, i, size * 32)) >= 0; ) {
|
|
++i;
|
|
*s++ = sep;
|
|
if ((unsigned) i < nsignals) {
|
|
s = stpcpy(s, signalent[i] + 3);
|
|
}
|
|
#ifdef SIGRTMIN
|
|
else if (i >= __SIGRTMIN && i <= __SIGRTMAX) {
|
|
s += sprintf(s, "RT_%u", i - __SIGRTMIN);
|
|
}
|
|
#endif
|
|
else {
|
|
s += sprintf(s, "%u", i);
|
|
}
|
|
sep = ' ';
|
|
}
|
|
if (sep == '[')
|
|
*s++ = sep;
|
|
*s++ = ']';
|
|
*s = '\0';
|
|
return outstr;
|
|
}
|
|
|
|
#define tprintsigmask_addr(prefix, mask) \
|
|
tprints(sprintsigmask_n((prefix), (mask), sizeof(mask)))
|
|
|
|
#define sprintsigmask_val(prefix, mask) \
|
|
sprintsigmask_n((prefix), &(mask), sizeof(mask))
|
|
|
|
#define tprintsigmask_val(prefix, mask) \
|
|
tprints(sprintsigmask_n((prefix), &(mask), sizeof(mask)))
|
|
|
|
void
|
|
printsignal(int nr)
|
|
{
|
|
tprints(signame(nr));
|
|
}
|
|
|
|
void
|
|
print_sigset_addr_len(struct tcb *tcp, long addr, long len)
|
|
{
|
|
char mask[NSIG / 8];
|
|
|
|
if (!addr) {
|
|
tprints("NULL");
|
|
return;
|
|
}
|
|
/* Here len is usually equals NSIG / 8 or current_wordsize.
|
|
* But we code this defensively:
|
|
*/
|
|
if (len < 0) {
|
|
bad:
|
|
tprintf("%#lx", addr);
|
|
return;
|
|
}
|
|
if (len >= NSIG / 8)
|
|
len = NSIG / 8;
|
|
else
|
|
len = (len + 3) & ~3;
|
|
|
|
if (umoven(tcp, addr, len, mask) < 0)
|
|
goto bad;
|
|
tprints(sprintsigmask_n("", mask, len));
|
|
}
|
|
|
|
#ifndef ILL_ILLOPC
|
|
#define ILL_ILLOPC 1 /* illegal opcode */
|
|
#define ILL_ILLOPN 2 /* illegal operand */
|
|
#define ILL_ILLADR 3 /* illegal addressing mode */
|
|
#define ILL_ILLTRP 4 /* illegal trap */
|
|
#define ILL_PRVOPC 5 /* privileged opcode */
|
|
#define ILL_PRVREG 6 /* privileged register */
|
|
#define ILL_COPROC 7 /* coprocessor error */
|
|
#define ILL_BADSTK 8 /* internal stack error */
|
|
#define FPE_INTDIV 1 /* integer divide by zero */
|
|
#define FPE_INTOVF 2 /* integer overflow */
|
|
#define FPE_FLTDIV 3 /* floating point divide by zero */
|
|
#define FPE_FLTOVF 4 /* floating point overflow */
|
|
#define FPE_FLTUND 5 /* floating point underflow */
|
|
#define FPE_FLTRES 6 /* floating point inexact result */
|
|
#define FPE_FLTINV 7 /* floating point invalid operation */
|
|
#define FPE_FLTSUB 8 /* subscript out of range */
|
|
#define SEGV_MAPERR 1 /* address not mapped to object */
|
|
#define SEGV_ACCERR 2 /* invalid permissions for mapped object */
|
|
#define BUS_ADRALN 1 /* invalid address alignment */
|
|
#define BUS_ADRERR 2 /* non-existant physical address */
|
|
#define BUS_OBJERR 3 /* object specific hardware error */
|
|
#define SYS_SECCOMP 1 /* seccomp triggered */
|
|
#define TRAP_BRKPT 1 /* process breakpoint */
|
|
#define TRAP_TRACE 2 /* process trace trap */
|
|
#define CLD_EXITED 1 /* child has exited */
|
|
#define CLD_KILLED 2 /* child was killed */
|
|
#define CLD_DUMPED 3 /* child terminated abnormally */
|
|
#define CLD_TRAPPED 4 /* traced child has trapped */
|
|
#define CLD_STOPPED 5 /* child has stopped */
|
|
#define CLD_CONTINUED 6 /* stopped child has continued */
|
|
#define POLL_IN 1 /* data input available */
|
|
#define POLL_OUT 2 /* output buffers available */
|
|
#define POLL_MSG 3 /* input message available */
|
|
#define POLL_ERR 4 /* i/o error */
|
|
#define POLL_PRI 5 /* high priority input available */
|
|
#define POLL_HUP 6 /* device disconnected */
|
|
#define SI_KERNEL 0x80 /* sent by kernel */
|
|
#define SI_USER 0 /* sent by kill, sigsend, raise */
|
|
#define SI_QUEUE -1 /* sent by sigqueue */
|
|
#define SI_TIMER -2 /* sent by timer expiration */
|
|
#define SI_MESGQ -3 /* sent by real time mesq state change */
|
|
#define SI_ASYNCIO -4 /* sent by AIO completion */
|
|
#define SI_SIGIO -5 /* sent by SIGIO */
|
|
#define SI_TKILL -6 /* sent by tkill */
|
|
#define SI_DETHREAD -7 /* sent by execve killing subsidiary threads */
|
|
#define SI_ASYNCNL -60 /* sent by asynch name lookup completion */
|
|
#endif
|
|
|
|
#ifndef SI_FROMUSER
|
|
# define SI_FROMUSER(sip) ((sip)->si_code <= 0)
|
|
#endif
|
|
|
|
#include "xlat/siginfo_codes.h"
|
|
#include "xlat/sigill_codes.h"
|
|
#include "xlat/sigfpe_codes.h"
|
|
#include "xlat/sigtrap_codes.h"
|
|
#include "xlat/sigchld_codes.h"
|
|
#include "xlat/sigpoll_codes.h"
|
|
#include "xlat/sigprof_codes.h"
|
|
|
|
#ifdef SIGEMT
|
|
#include "xlat/sigemt_codes.h"
|
|
#endif
|
|
|
|
#include "xlat/sigsegv_codes.h"
|
|
#include "xlat/sigbus_codes.h"
|
|
|
|
#ifndef SYS_SECCOMP
|
|
# define SYS_SECCOMP 1
|
|
#endif
|
|
#include "xlat/sigsys_codes.h"
|
|
|
|
static void
|
|
printsigsource(const siginfo_t *sip)
|
|
{
|
|
tprintf(", si_pid=%lu, si_uid=%lu",
|
|
(unsigned long) sip->si_pid,
|
|
(unsigned long) sip->si_uid);
|
|
}
|
|
|
|
static void
|
|
printsigval(const siginfo_t *sip, int verbose)
|
|
{
|
|
if (!verbose)
|
|
tprints(", ...");
|
|
else
|
|
tprintf(", si_value={int=%u, ptr=%#lx}",
|
|
sip->si_int,
|
|
(unsigned long) sip->si_ptr);
|
|
}
|
|
|
|
void
|
|
printsiginfo(siginfo_t *sip, int verbose)
|
|
{
|
|
const char *code;
|
|
|
|
if (sip->si_signo == 0) {
|
|
tprints("{}");
|
|
return;
|
|
}
|
|
tprints("{si_signo=");
|
|
printsignal(sip->si_signo);
|
|
code = xlookup(siginfo_codes, sip->si_code);
|
|
if (!code) {
|
|
switch (sip->si_signo) {
|
|
case SIGTRAP:
|
|
code = xlookup(sigtrap_codes, sip->si_code);
|
|
break;
|
|
case SIGCHLD:
|
|
code = xlookup(sigchld_codes, sip->si_code);
|
|
break;
|
|
case SIGPOLL:
|
|
code = xlookup(sigpoll_codes, sip->si_code);
|
|
break;
|
|
case SIGPROF:
|
|
code = xlookup(sigprof_codes, sip->si_code);
|
|
break;
|
|
case SIGILL:
|
|
code = xlookup(sigill_codes, sip->si_code);
|
|
break;
|
|
#ifdef SIGEMT
|
|
case SIGEMT:
|
|
code = xlookup(sigemt_codes, sip->si_code);
|
|
break;
|
|
#endif
|
|
case SIGFPE:
|
|
code = xlookup(sigfpe_codes, sip->si_code);
|
|
break;
|
|
case SIGSEGV:
|
|
code = xlookup(sigsegv_codes, sip->si_code);
|
|
break;
|
|
case SIGBUS:
|
|
code = xlookup(sigbus_codes, sip->si_code);
|
|
break;
|
|
case SIGSYS:
|
|
code = xlookup(sigsys_codes, sip->si_code);
|
|
break;
|
|
}
|
|
}
|
|
if (code)
|
|
tprintf(", si_code=%s", code);
|
|
else
|
|
tprintf(", si_code=%#x", sip->si_code);
|
|
#ifdef SI_NOINFO
|
|
if (sip->si_code != SI_NOINFO)
|
|
#endif
|
|
{
|
|
if (sip->si_errno) {
|
|
if (sip->si_errno < 0 || (unsigned) sip->si_errno >= nerrnos)
|
|
tprintf(", si_errno=%d", sip->si_errno);
|
|
else
|
|
tprintf(", si_errno=%s",
|
|
errnoent[sip->si_errno]);
|
|
}
|
|
#ifdef SI_FROMUSER
|
|
if (SI_FROMUSER(sip)) {
|
|
switch (sip->si_code) {
|
|
#ifdef SI_USER
|
|
case SI_USER:
|
|
printsigsource(sip);
|
|
break;
|
|
#endif
|
|
#ifdef SI_TKILL
|
|
case SI_TKILL:
|
|
printsigsource(sip);
|
|
break;
|
|
#endif
|
|
#ifdef SI_TIMER
|
|
case SI_TIMER:
|
|
tprintf(", si_timerid=%#x, si_overrun=%d",
|
|
sip->si_timerid, sip->si_overrun);
|
|
printsigval(sip, verbose);
|
|
break;
|
|
#endif
|
|
default:
|
|
printsigsource(sip);
|
|
if (sip->si_ptr)
|
|
printsigval(sip, verbose);
|
|
break;
|
|
}
|
|
}
|
|
else
|
|
#endif /* SI_FROMUSER */
|
|
{
|
|
switch (sip->si_signo) {
|
|
case SIGCHLD:
|
|
printsigsource(sip);
|
|
tprints(", si_status=");
|
|
if (sip->si_code == CLD_EXITED)
|
|
tprintf("%d", sip->si_status);
|
|
else
|
|
printsignal(sip->si_status);
|
|
if (!verbose)
|
|
tprints(", ...");
|
|
else
|
|
tprintf(", si_utime=%llu, si_stime=%llu",
|
|
(unsigned long long) sip->si_utime,
|
|
(unsigned long long) sip->si_stime);
|
|
break;
|
|
case SIGILL: case SIGFPE:
|
|
case SIGSEGV: case SIGBUS:
|
|
tprintf(", si_addr=%#lx",
|
|
(unsigned long) sip->si_addr);
|
|
break;
|
|
case SIGPOLL:
|
|
switch (sip->si_code) {
|
|
case POLL_IN: case POLL_OUT: case POLL_MSG:
|
|
tprintf(", si_band=%ld",
|
|
(long) sip->si_band);
|
|
break;
|
|
}
|
|
break;
|
|
#ifdef HAVE_SIGINFO_T_SI_SYSCALL
|
|
case SIGSYS:
|
|
tprintf(", si_call_addr=%#lx, si_syscall=%d, si_arch=%u",
|
|
(unsigned long) sip->si_call_addr,
|
|
sip->si_syscall, sip->si_arch);
|
|
break;
|
|
#endif
|
|
default:
|
|
if (sip->si_pid || sip->si_uid)
|
|
printsigsource(sip);
|
|
if (sip->si_ptr)
|
|
printsigval(sip, verbose);
|
|
}
|
|
}
|
|
}
|
|
tprints("}");
|
|
}
|
|
|
|
void
|
|
printsiginfo_at(struct tcb *tcp, long addr)
|
|
{
|
|
siginfo_t si;
|
|
if (!addr) {
|
|
tprints("NULL");
|
|
return;
|
|
}
|
|
if (syserror(tcp)) {
|
|
tprintf("%#lx", addr);
|
|
return;
|
|
}
|
|
if (umove(tcp, addr, &si) < 0) {
|
|
tprints("{???}");
|
|
return;
|
|
}
|
|
printsiginfo(&si, verbose(tcp));
|
|
}
|
|
|
|
int
|
|
sys_sigsetmask(struct tcb *tcp)
|
|
{
|
|
if (entering(tcp)) {
|
|
tprintsigmask_val("", tcp->u_arg[0]);
|
|
}
|
|
else if (!syserror(tcp)) {
|
|
tcp->auxstr = sprintsigmask_val("old mask ", tcp->u_rval);
|
|
return RVAL_HEX | RVAL_STR;
|
|
}
|
|
return 0;
|
|
}
|
|
|
|
#ifdef HAVE_SIGACTION
|
|
|
|
struct old_sigaction {
|
|
/* sa_handler may be a libc #define, need to use other name: */
|
|
#ifdef MIPS
|
|
unsigned int sa_flags;
|
|
void (*__sa_handler)(int);
|
|
/* Kernel treats sa_mask as an array of longs. */
|
|
unsigned long sa_mask[NSIG / sizeof(long) ? NSIG / sizeof(long) : 1];
|
|
#else
|
|
void (*__sa_handler)(int);
|
|
unsigned long sa_mask;
|
|
unsigned long sa_flags;
|
|
#endif /* !MIPS */
|
|
#ifdef SA_RESTORER
|
|
void (*sa_restorer)(void);
|
|
#endif
|
|
};
|
|
|
|
struct old_sigaction32 {
|
|
/* sa_handler may be a libc #define, need to use other name: */
|
|
uint32_t __sa_handler;
|
|
uint32_t sa_mask;
|
|
uint32_t sa_flags;
|
|
#ifdef SA_RESTORER
|
|
uint32_t sa_restorer;
|
|
#endif
|
|
};
|
|
|
|
static void
|
|
decode_old_sigaction(struct tcb *tcp, long addr)
|
|
{
|
|
struct old_sigaction sa;
|
|
int r;
|
|
|
|
if (!addr) {
|
|
tprints("NULL");
|
|
return;
|
|
}
|
|
if (!verbose(tcp) || (exiting(tcp) && syserror(tcp))) {
|
|
tprintf("%#lx", addr);
|
|
return;
|
|
}
|
|
|
|
#if SUPPORTED_PERSONALITIES > 1 && SIZEOF_LONG > 4
|
|
if (current_wordsize != sizeof(sa.__sa_handler) && current_wordsize == 4) {
|
|
struct old_sigaction32 sa32;
|
|
r = umove(tcp, addr, &sa32);
|
|
if (r >= 0) {
|
|
memset(&sa, 0, sizeof(sa));
|
|
sa.__sa_handler = (void*)(uintptr_t)sa32.__sa_handler;
|
|
sa.sa_flags = sa32.sa_flags;
|
|
#ifdef SA_RESTORER
|
|
sa.sa_restorer = (void*)(uintptr_t)sa32.sa_restorer;
|
|
#endif
|
|
sa.sa_mask = sa32.sa_mask;
|
|
}
|
|
} else
|
|
#endif
|
|
{
|
|
r = umove(tcp, addr, &sa);
|
|
}
|
|
if (r < 0) {
|
|
tprints("{...}");
|
|
return;
|
|
}
|
|
|
|
/* Architectures using function pointers, like
|
|
* hppa, may need to manipulate the function pointer
|
|
* to compute the result of a comparison. However,
|
|
* the __sa_handler function pointer exists only in
|
|
* the address space of the traced process, and can't
|
|
* be manipulated by strace. In order to prevent the
|
|
* compiler from generating code to manipulate
|
|
* __sa_handler we cast the function pointers to long. */
|
|
if ((long)sa.__sa_handler == (long)SIG_ERR)
|
|
tprints("{SIG_ERR, ");
|
|
else if ((long)sa.__sa_handler == (long)SIG_DFL)
|
|
tprints("{SIG_DFL, ");
|
|
else if ((long)sa.__sa_handler == (long)SIG_IGN)
|
|
tprints("{SIG_IGN, ");
|
|
else
|
|
tprintf("{%#lx, ", (long) sa.__sa_handler);
|
|
#ifdef MIPS
|
|
tprintsigmask_addr("", sa.sa_mask);
|
|
#else
|
|
tprintsigmask_val("", sa.sa_mask);
|
|
#endif
|
|
tprints(", ");
|
|
printflags(sigact_flags, sa.sa_flags, "SA_???");
|
|
#ifdef SA_RESTORER
|
|
if (sa.sa_flags & SA_RESTORER)
|
|
tprintf(", %p", sa.sa_restorer);
|
|
#endif
|
|
tprints("}");
|
|
}
|
|
|
|
int
|
|
sys_sigaction(struct tcb *tcp)
|
|
{
|
|
if (entering(tcp)) {
|
|
printsignal(tcp->u_arg[0]);
|
|
tprints(", ");
|
|
decode_old_sigaction(tcp, tcp->u_arg[1]);
|
|
tprints(", ");
|
|
} else
|
|
decode_old_sigaction(tcp, tcp->u_arg[2]);
|
|
return 0;
|
|
}
|
|
|
|
int
|
|
sys_signal(struct tcb *tcp)
|
|
{
|
|
if (entering(tcp)) {
|
|
printsignal(tcp->u_arg[0]);
|
|
tprints(", ");
|
|
switch (tcp->u_arg[1]) {
|
|
case (long) SIG_ERR:
|
|
tprints("SIG_ERR");
|
|
break;
|
|
case (long) SIG_DFL:
|
|
tprints("SIG_DFL");
|
|
break;
|
|
case (long) SIG_IGN:
|
|
tprints("SIG_IGN");
|
|
break;
|
|
default:
|
|
tprintf("%#lx", tcp->u_arg[1]);
|
|
}
|
|
return 0;
|
|
}
|
|
else if (!syserror(tcp)) {
|
|
switch (tcp->u_rval) {
|
|
case (long) SIG_ERR:
|
|
tcp->auxstr = "SIG_ERR"; break;
|
|
case (long) SIG_DFL:
|
|
tcp->auxstr = "SIG_DFL"; break;
|
|
case (long) SIG_IGN:
|
|
tcp->auxstr = "SIG_IGN"; break;
|
|
default:
|
|
tcp->auxstr = NULL;
|
|
}
|
|
return RVAL_HEX | RVAL_STR;
|
|
}
|
|
return 0;
|
|
}
|
|
|
|
#endif /* HAVE_SIGACTION */
|
|
|
|
int
|
|
sys_sigreturn(struct tcb *tcp)
|
|
{
|
|
#if defined(ARM)
|
|
if (entering(tcp)) {
|
|
struct arm_sigcontext {
|
|
unsigned long trap_no;
|
|
unsigned long error_code;
|
|
unsigned long oldmask;
|
|
unsigned long arm_r0;
|
|
unsigned long arm_r1;
|
|
unsigned long arm_r2;
|
|
unsigned long arm_r3;
|
|
unsigned long arm_r4;
|
|
unsigned long arm_r5;
|
|
unsigned long arm_r6;
|
|
unsigned long arm_r7;
|
|
unsigned long arm_r8;
|
|
unsigned long arm_r9;
|
|
unsigned long arm_r10;
|
|
unsigned long arm_fp;
|
|
unsigned long arm_ip;
|
|
unsigned long arm_sp;
|
|
unsigned long arm_lr;
|
|
unsigned long arm_pc;
|
|
unsigned long arm_cpsr;
|
|
unsigned long fault_address;
|
|
};
|
|
struct arm_ucontext {
|
|
unsigned long uc_flags;
|
|
unsigned long uc_link; /* struct ucontext* */
|
|
/* The next three members comprise stack_t struct: */
|
|
unsigned long ss_sp; /* void* */
|
|
unsigned long ss_flags; /* int */
|
|
unsigned long ss_size; /* size_t */
|
|
struct arm_sigcontext sc;
|
|
/* These two members are sigset_t: */
|
|
unsigned long uc_sigmask[2];
|
|
/* more fields follow, which we aren't interested in */
|
|
};
|
|
struct arm_ucontext uc;
|
|
if (umove(tcp, arm_regs.ARM_sp, &uc) < 0)
|
|
return 0;
|
|
/*
|
|
* Kernel fills out uc.sc.oldmask too when it sets up signal stack,
|
|
* but for sigmask restore, sigreturn syscall uses uc.uc_sigmask instead.
|
|
*/
|
|
tprintsigmask_addr(") (mask ", uc.uc_sigmask);
|
|
}
|
|
#elif defined(S390) || defined(S390X)
|
|
if (entering(tcp)) {
|
|
long usp;
|
|
struct sigcontext sc;
|
|
if (upeek(tcp->pid, PT_GPR15, &usp) < 0)
|
|
return 0;
|
|
if (umove(tcp, usp + __SIGNAL_FRAMESIZE, &sc) < 0)
|
|
return 0;
|
|
tprintsigmask_addr(") (mask ", sc.oldmask);
|
|
}
|
|
#elif defined(I386) || defined(X86_64)
|
|
# if defined(X86_64)
|
|
if (current_personality == 0) /* 64-bit */
|
|
return 0;
|
|
# endif
|
|
if (entering(tcp)) {
|
|
struct i386_sigcontext_struct {
|
|
uint16_t gs, __gsh;
|
|
uint16_t fs, __fsh;
|
|
uint16_t es, __esh;
|
|
uint16_t ds, __dsh;
|
|
uint32_t edi;
|
|
uint32_t esi;
|
|
uint32_t ebp;
|
|
uint32_t esp;
|
|
uint32_t ebx;
|
|
uint32_t edx;
|
|
uint32_t ecx;
|
|
uint32_t eax;
|
|
uint32_t trapno;
|
|
uint32_t err;
|
|
uint32_t eip;
|
|
uint16_t cs, __csh;
|
|
uint32_t eflags;
|
|
uint32_t esp_at_signal;
|
|
uint16_t ss, __ssh;
|
|
uint32_t i387;
|
|
uint32_t oldmask;
|
|
uint32_t cr2;
|
|
};
|
|
struct i386_fpstate {
|
|
uint32_t cw;
|
|
uint32_t sw;
|
|
uint32_t tag;
|
|
uint32_t ipoff;
|
|
uint32_t cssel;
|
|
uint32_t dataoff;
|
|
uint32_t datasel;
|
|
uint8_t st[8][10]; /* 8*10 bytes: FP regs */
|
|
uint16_t status;
|
|
uint16_t magic;
|
|
uint32_t fxsr_env[6];
|
|
uint32_t mxcsr;
|
|
uint32_t reserved;
|
|
uint8_t stx[8][16]; /* 8*16 bytes: FP regs, each padded to 16 bytes */
|
|
uint8_t xmm[8][16]; /* 8 XMM regs */
|
|
uint32_t padding1[44];
|
|
uint32_t padding2[12]; /* union with struct _fpx_sw_bytes */
|
|
};
|
|
struct {
|
|
struct i386_sigcontext_struct sc;
|
|
struct i386_fpstate fp;
|
|
uint32_t extramask[1];
|
|
} signal_stack;
|
|
/* On i386, sc is followed on stack by struct fpstate
|
|
* and after it an additional u32 extramask[1] which holds
|
|
* upper half of the mask.
|
|
*/
|
|
uint32_t sigmask[2];
|
|
if (umove(tcp, *i386_esp_ptr, &signal_stack) < 0)
|
|
return 0;
|
|
sigmask[0] = signal_stack.sc.oldmask;
|
|
sigmask[1] = signal_stack.extramask[0];
|
|
tprintsigmask_addr(") (mask ", sigmask);
|
|
}
|
|
#elif defined(IA64)
|
|
if (entering(tcp)) {
|
|
struct sigcontext sc;
|
|
long sp;
|
|
/* offset of sigcontext in the kernel's sigframe structure: */
|
|
# define SIGFRAME_SC_OFFSET 0x90
|
|
if (upeek(tcp->pid, PT_R12, &sp) < 0)
|
|
return 0;
|
|
if (umove(tcp, sp + 16 + SIGFRAME_SC_OFFSET, &sc) < 0)
|
|
return 0;
|
|
tprintsigmask_val(") (mask ", sc.sc_mask);
|
|
}
|
|
#elif defined(POWERPC)
|
|
if (entering(tcp)) {
|
|
long esp;
|
|
struct sigcontext sc;
|
|
|
|
esp = ppc_regs.gpr[1];
|
|
|
|
/* Skip dummy stack frame. */
|
|
#ifdef POWERPC64
|
|
if (current_personality == 0)
|
|
esp += 128;
|
|
else
|
|
esp += 64;
|
|
#else
|
|
esp += 64;
|
|
#endif
|
|
if (umove(tcp, esp, &sc) < 0)
|
|
return 0;
|
|
tprintsigmask_val(") (mask ", sc.oldmask);
|
|
}
|
|
#elif defined(M68K)
|
|
if (entering(tcp)) {
|
|
long usp;
|
|
struct sigcontext sc;
|
|
if (upeek(tcp->pid, 4*PT_USP, &usp) < 0)
|
|
return 0;
|
|
if (umove(tcp, usp, &sc) < 0)
|
|
return 0;
|
|
tprintsigmask_val(") (mask ", sc.sc_mask);
|
|
}
|
|
#elif defined(ALPHA)
|
|
if (entering(tcp)) {
|
|
long fp;
|
|
struct sigcontext sc;
|
|
if (upeek(tcp->pid, REG_FP, &fp) < 0)
|
|
return 0;
|
|
if (umove(tcp, fp, &sc) < 0)
|
|
return 0;
|
|
tprintsigmask_val(") (mask ", sc.sc_mask);
|
|
}
|
|
#elif defined(SPARC) || defined(SPARC64)
|
|
if (entering(tcp)) {
|
|
long i1;
|
|
m_siginfo_t si;
|
|
i1 = sparc_regs.u_regs[U_REG_O1];
|
|
if (umove(tcp, i1, &si) < 0) {
|
|
perror_msg("sigreturn: umove");
|
|
return 0;
|
|
}
|
|
tprintsigmask_val(") (mask ", si.si_mask);
|
|
}
|
|
#elif defined(LINUX_MIPSN32) || defined(LINUX_MIPSN64)
|
|
/* This decodes rt_sigreturn. The 64-bit ABIs do not have
|
|
sigreturn. */
|
|
if (entering(tcp)) {
|
|
long sp;
|
|
struct ucontext uc;
|
|
if (upeek(tcp->pid, REG_SP, &sp) < 0)
|
|
return 0;
|
|
/* There are six words followed by a 128-byte siginfo. */
|
|
sp = sp + 6 * 4 + 128;
|
|
if (umove(tcp, sp, &uc) < 0)
|
|
return 0;
|
|
tprintsigmask_val(") (mask ", uc.uc_sigmask);
|
|
}
|
|
#elif defined(MIPS)
|
|
if (entering(tcp)) {
|
|
long sp;
|
|
struct pt_regs regs;
|
|
m_siginfo_t si;
|
|
if (ptrace(PTRACE_GETREGS, tcp->pid, (char *)®s, 0) < 0) {
|
|
perror_msg("sigreturn: PTRACE_GETREGS");
|
|
return 0;
|
|
}
|
|
sp = regs.regs[29];
|
|
if (umove(tcp, sp, &si) < 0)
|
|
return 0;
|
|
tprintsigmask_val(") (mask ", si.si_mask);
|
|
}
|
|
#elif defined(CRISV10) || defined(CRISV32)
|
|
if (entering(tcp)) {
|
|
struct sigcontext sc;
|
|
long regs[PT_MAX+1];
|
|
if (ptrace(PTRACE_GETREGS, tcp->pid, NULL, (long)regs) < 0) {
|
|
perror_msg("sigreturn: PTRACE_GETREGS");
|
|
return 0;
|
|
}
|
|
if (umove(tcp, regs[PT_USP], &sc) < 0)
|
|
return 0;
|
|
tprintsigmask_val(") (mask ", sc.oldmask);
|
|
}
|
|
#elif defined(TILE)
|
|
if (entering(tcp)) {
|
|
struct ucontext uc;
|
|
|
|
/* offset of ucontext in the kernel's sigframe structure */
|
|
# define SIGFRAME_UC_OFFSET C_ABI_SAVE_AREA_SIZE + sizeof(siginfo_t)
|
|
if (umove(tcp, tile_regs.sp + SIGFRAME_UC_OFFSET, &uc) < 0)
|
|
return 0;
|
|
tprintsigmask_val(") (mask ", uc.uc_sigmask);
|
|
}
|
|
#elif defined(MICROBLAZE)
|
|
/* TODO: Verify that this is correct... */
|
|
if (entering(tcp)) {
|
|
struct sigcontext sc;
|
|
long sp;
|
|
/* Read r1, the stack pointer. */
|
|
if (upeek(tcp->pid, 1 * 4, &sp) < 0)
|
|
return 0;
|
|
if (umove(tcp, sp, &sc) < 0)
|
|
return 0;
|
|
tprintsigmask_val(") (mask ", sc.oldmask);
|
|
}
|
|
#elif defined(XTENSA)
|
|
/* Xtensa only has rt_sys_sigreturn */
|
|
#elif defined(ARC)
|
|
/* ARC syscall ABI only supports rt_sys_sigreturn */
|
|
#else
|
|
# warning No sys_sigreturn() for this architecture
|
|
# warning (no problem, just a reminder :-)
|
|
#endif
|
|
return 0;
|
|
}
|
|
|
|
int
|
|
sys_siggetmask(struct tcb *tcp)
|
|
{
|
|
if (exiting(tcp)) {
|
|
tcp->auxstr = sprintsigmask_val("mask ", tcp->u_rval);
|
|
}
|
|
return RVAL_HEX | RVAL_STR;
|
|
}
|
|
|
|
int
|
|
sys_sigsuspend(struct tcb *tcp)
|
|
{
|
|
if (entering(tcp)) {
|
|
tprintsigmask_val("", tcp->u_arg[2]);
|
|
}
|
|
return 0;
|
|
}
|
|
|
|
#if !defined SS_ONSTACK
|
|
#define SS_ONSTACK 1
|
|
#define SS_DISABLE 2
|
|
#endif
|
|
|
|
#include "xlat/sigaltstack_flags.h"
|
|
|
|
static void
|
|
print_stack_t(struct tcb *tcp, unsigned long addr)
|
|
{
|
|
stack_t ss;
|
|
int r;
|
|
|
|
if (!addr) {
|
|
tprints("NULL");
|
|
return;
|
|
}
|
|
|
|
#if SUPPORTED_PERSONALITIES > 1 && SIZEOF_LONG > 4
|
|
if (current_wordsize != sizeof(ss.ss_sp) && current_wordsize == 4) {
|
|
struct {
|
|
uint32_t ss_sp;
|
|
int32_t ss_flags;
|
|
uint32_t ss_size;
|
|
} ss32;
|
|
r = umove(tcp, addr, &ss32);
|
|
if (r >= 0) {
|
|
memset(&ss, 0, sizeof(ss));
|
|
ss.ss_sp = (void*)(unsigned long) ss32.ss_sp;
|
|
ss.ss_flags = ss32.ss_flags;
|
|
ss.ss_size = (unsigned long) ss32.ss_size;
|
|
}
|
|
} else
|
|
#endif
|
|
{
|
|
r = umove(tcp, addr, &ss);
|
|
}
|
|
if (r < 0) {
|
|
tprintf("%#lx", addr);
|
|
} else {
|
|
tprintf("{ss_sp=%#lx, ss_flags=", (unsigned long) ss.ss_sp);
|
|
printflags(sigaltstack_flags, ss.ss_flags, "SS_???");
|
|
tprintf(", ss_size=%lu}", (unsigned long) ss.ss_size);
|
|
}
|
|
}
|
|
|
|
int
|
|
sys_sigaltstack(struct tcb *tcp)
|
|
{
|
|
if (entering(tcp)) {
|
|
print_stack_t(tcp, tcp->u_arg[0]);
|
|
}
|
|
else {
|
|
tprints(", ");
|
|
print_stack_t(tcp, tcp->u_arg[1]);
|
|
}
|
|
return 0;
|
|
}
|
|
|
|
#ifdef HAVE_SIGACTION
|
|
|
|
/* "Old" sigprocmask, which operates with word-sized signal masks */
|
|
int
|
|
sys_sigprocmask(struct tcb *tcp)
|
|
{
|
|
# ifdef ALPHA
|
|
if (entering(tcp)) {
|
|
/*
|
|
* Alpha/OSF is different: it doesn't pass in two pointers,
|
|
* but rather passes in the new bitmask as an argument and
|
|
* then returns the old bitmask. This "works" because we
|
|
* only have 64 signals to worry about. If you want more,
|
|
* use of the rt_sigprocmask syscall is required.
|
|
* Alpha:
|
|
* old = osf_sigprocmask(how, new);
|
|
* Everyone else:
|
|
* ret = sigprocmask(how, &new, &old, ...);
|
|
*/
|
|
printxval(sigprocmaskcmds, tcp->u_arg[0], "SIG_???");
|
|
tprintsigmask_val(", ", tcp->u_arg[1]);
|
|
}
|
|
else if (!syserror(tcp)) {
|
|
tcp->auxstr = sprintsigmask_val("old mask ", tcp->u_rval);
|
|
return RVAL_HEX | RVAL_STR;
|
|
}
|
|
# else /* !ALPHA */
|
|
if (entering(tcp)) {
|
|
printxval(sigprocmaskcmds, tcp->u_arg[0], "SIG_???");
|
|
tprints(", ");
|
|
print_sigset_addr_len(tcp, tcp->u_arg[1], current_wordsize);
|
|
tprints(", ");
|
|
}
|
|
else {
|
|
if (syserror(tcp))
|
|
tprintf("%#lx", tcp->u_arg[2]);
|
|
else
|
|
print_sigset_addr_len(tcp, tcp->u_arg[2], current_wordsize);
|
|
}
|
|
# endif /* !ALPHA */
|
|
return 0;
|
|
}
|
|
|
|
#endif /* HAVE_SIGACTION */
|
|
|
|
int
|
|
sys_kill(struct tcb *tcp)
|
|
{
|
|
if (entering(tcp)) {
|
|
tprintf("%ld, %s",
|
|
widen_to_long(tcp->u_arg[0]),
|
|
signame(tcp->u_arg[1])
|
|
);
|
|
}
|
|
return 0;
|
|
}
|
|
|
|
int
|
|
sys_tgkill(struct tcb *tcp)
|
|
{
|
|
if (entering(tcp)) {
|
|
tprintf("%ld, %ld, %s",
|
|
widen_to_long(tcp->u_arg[0]),
|
|
widen_to_long(tcp->u_arg[1]),
|
|
signame(tcp->u_arg[2])
|
|
);
|
|
}
|
|
return 0;
|
|
}
|
|
|
|
int
|
|
sys_sigpending(struct tcb *tcp)
|
|
{
|
|
if (exiting(tcp)) {
|
|
if (syserror(tcp))
|
|
tprintf("%#lx", tcp->u_arg[0]);
|
|
else
|
|
print_sigset_addr_len(tcp, tcp->u_arg[0], current_wordsize);
|
|
}
|
|
return 0;
|
|
}
|
|
|
|
int
|
|
sys_rt_sigprocmask(struct tcb *tcp)
|
|
{
|
|
/* Note: arg[3] is the length of the sigset. Kernel requires NSIG / 8 */
|
|
if (entering(tcp)) {
|
|
printxval(sigprocmaskcmds, tcp->u_arg[0], "SIG_???");
|
|
tprints(", ");
|
|
print_sigset_addr_len(tcp, tcp->u_arg[1], tcp->u_arg[3]);
|
|
tprints(", ");
|
|
}
|
|
else {
|
|
if (syserror(tcp))
|
|
tprintf("%#lx", tcp->u_arg[2]);
|
|
else
|
|
print_sigset_addr_len(tcp, tcp->u_arg[2], tcp->u_arg[3]);
|
|
tprintf(", %lu", tcp->u_arg[3]);
|
|
}
|
|
return 0;
|
|
}
|
|
|
|
/* Structure describing the action to be taken when a signal arrives. */
|
|
struct new_sigaction
|
|
{
|
|
/* sa_handler may be a libc #define, need to use other name: */
|
|
#ifdef MIPS
|
|
unsigned int sa_flags;
|
|
void (*__sa_handler)(int);
|
|
#else
|
|
void (*__sa_handler)(int);
|
|
unsigned long sa_flags;
|
|
#endif /* !MIPS */
|
|
#ifdef SA_RESTORER
|
|
void (*sa_restorer)(void);
|
|
#endif
|
|
/* Kernel treats sa_mask as an array of longs. */
|
|
unsigned long sa_mask[NSIG / sizeof(long) ? NSIG / sizeof(long) : 1];
|
|
};
|
|
/* Same for i386-on-x86_64 and similar cases */
|
|
struct new_sigaction32
|
|
{
|
|
uint32_t __sa_handler;
|
|
uint32_t sa_flags;
|
|
#ifdef SA_RESTORER
|
|
uint32_t sa_restorer;
|
|
#endif
|
|
uint32_t sa_mask[2 * (NSIG / sizeof(long) ? NSIG / sizeof(long) : 1)];
|
|
};
|
|
|
|
static void
|
|
decode_new_sigaction(struct tcb *tcp, long addr)
|
|
{
|
|
struct new_sigaction sa;
|
|
int r;
|
|
|
|
if (!addr) {
|
|
tprints("NULL");
|
|
return;
|
|
}
|
|
if (!verbose(tcp) || (exiting(tcp) && syserror(tcp))) {
|
|
tprintf("%#lx", addr);
|
|
return;
|
|
}
|
|
#if SUPPORTED_PERSONALITIES > 1 && SIZEOF_LONG > 4
|
|
if (current_wordsize != sizeof(sa.sa_flags) && current_wordsize == 4) {
|
|
struct new_sigaction32 sa32;
|
|
r = umove(tcp, addr, &sa32);
|
|
if (r >= 0) {
|
|
memset(&sa, 0, sizeof(sa));
|
|
sa.__sa_handler = (void*)(unsigned long)sa32.__sa_handler;
|
|
sa.sa_flags = sa32.sa_flags;
|
|
#ifdef SA_RESTORER
|
|
sa.sa_restorer = (void*)(unsigned long)sa32.sa_restorer;
|
|
#endif
|
|
/* Kernel treats sa_mask as an array of longs.
|
|
* For 32-bit process, "long" is uint32_t, thus, for example,
|
|
* 32th bit in sa_mask will end up as bit 0 in sa_mask[1].
|
|
* But for (64-bit) kernel, 32th bit in sa_mask is
|
|
* 32th bit in 0th (64-bit) long!
|
|
* For little-endian, it's the same.
|
|
* For big-endian, we swap 32-bit words.
|
|
*/
|
|
sa.sa_mask[0] = sa32.sa_mask[0] + ((long)(sa32.sa_mask[1]) << 32);
|
|
}
|
|
} else
|
|
#endif
|
|
{
|
|
r = umove(tcp, addr, &sa);
|
|
}
|
|
if (r < 0) {
|
|
tprints("{...}");
|
|
return;
|
|
}
|
|
/* Architectures using function pointers, like
|
|
* hppa, may need to manipulate the function pointer
|
|
* to compute the result of a comparison. However,
|
|
* the __sa_handler function pointer exists only in
|
|
* the address space of the traced process, and can't
|
|
* be manipulated by strace. In order to prevent the
|
|
* compiler from generating code to manipulate
|
|
* __sa_handler we cast the function pointers to long. */
|
|
if ((long)sa.__sa_handler == (long)SIG_ERR)
|
|
tprints("{SIG_ERR, ");
|
|
else if ((long)sa.__sa_handler == (long)SIG_DFL)
|
|
tprints("{SIG_DFL, ");
|
|
else if ((long)sa.__sa_handler == (long)SIG_IGN)
|
|
tprints("{SIG_IGN, ");
|
|
else
|
|
tprintf("{%#lx, ", (long) sa.__sa_handler);
|
|
/*
|
|
* Sigset size is in tcp->u_arg[4] (SPARC)
|
|
* or in tcp->u_arg[3] (all other),
|
|
* but kernel won't handle sys_rt_sigaction
|
|
* with wrong sigset size (just returns EINVAL instead).
|
|
* We just fetch the right size, which is NSIG / 8.
|
|
*/
|
|
tprintsigmask_val("", sa.sa_mask);
|
|
tprints(", ");
|
|
|
|
printflags(sigact_flags, sa.sa_flags, "SA_???");
|
|
#ifdef SA_RESTORER
|
|
if (sa.sa_flags & SA_RESTORER)
|
|
tprintf(", %p", sa.sa_restorer);
|
|
#endif
|
|
tprints("}");
|
|
}
|
|
|
|
int
|
|
sys_rt_sigaction(struct tcb *tcp)
|
|
{
|
|
if (entering(tcp)) {
|
|
printsignal(tcp->u_arg[0]);
|
|
tprints(", ");
|
|
decode_new_sigaction(tcp, tcp->u_arg[1]);
|
|
tprints(", ");
|
|
} else {
|
|
decode_new_sigaction(tcp, tcp->u_arg[2]);
|
|
#if defined(SPARC) || defined(SPARC64)
|
|
tprintf(", %#lx, %lu", tcp->u_arg[3], tcp->u_arg[4]);
|
|
#elif defined(ALPHA)
|
|
tprintf(", %lu, %#lx", tcp->u_arg[3], tcp->u_arg[4]);
|
|
#else
|
|
tprintf(", %lu", tcp->u_arg[3]);
|
|
#endif
|
|
}
|
|
return 0;
|
|
}
|
|
|
|
int
|
|
sys_rt_sigpending(struct tcb *tcp)
|
|
{
|
|
if (exiting(tcp)) {
|
|
/*
|
|
* One of the few syscalls where sigset size (arg[1])
|
|
* is allowed to be <= NSIG / 8, not strictly ==.
|
|
* This allows non-rt sigpending() syscall
|
|
* to reuse rt_sigpending() code in kernel.
|
|
*/
|
|
if (syserror(tcp))
|
|
tprintf("%#lx", tcp->u_arg[0]);
|
|
else
|
|
print_sigset_addr_len(tcp, tcp->u_arg[0], tcp->u_arg[1]);
|
|
tprintf(", %lu", tcp->u_arg[1]);
|
|
}
|
|
return 0;
|
|
}
|
|
|
|
int
|
|
sys_rt_sigsuspend(struct tcb *tcp)
|
|
{
|
|
if (entering(tcp)) {
|
|
/* NB: kernel requires arg[1] == NSIG / 8 */
|
|
print_sigset_addr_len(tcp, tcp->u_arg[0], tcp->u_arg[1]);
|
|
tprintf(", %lu", tcp->u_arg[1]);
|
|
}
|
|
return 0;
|
|
}
|
|
|
|
static void
|
|
print_sigqueueinfo(struct tcb *tcp, int sig, unsigned long uinfo)
|
|
{
|
|
printsignal(sig);
|
|
tprints(", ");
|
|
printsiginfo_at(tcp, uinfo);
|
|
}
|
|
|
|
int
|
|
sys_rt_sigqueueinfo(struct tcb *tcp)
|
|
{
|
|
if (entering(tcp)) {
|
|
tprintf("%lu, ", tcp->u_arg[0]);
|
|
print_sigqueueinfo(tcp, tcp->u_arg[1], tcp->u_arg[2]);
|
|
}
|
|
return 0;
|
|
}
|
|
|
|
int
|
|
sys_rt_tgsigqueueinfo(struct tcb *tcp)
|
|
{
|
|
if (entering(tcp)) {
|
|
tprintf("%lu, %lu, ", tcp->u_arg[0], tcp->u_arg[1]);
|
|
print_sigqueueinfo(tcp, tcp->u_arg[2], tcp->u_arg[3]);
|
|
}
|
|
return 0;
|
|
}
|
|
|
|
int sys_rt_sigtimedwait(struct tcb *tcp)
|
|
{
|
|
/* NB: kernel requires arg[3] == NSIG / 8 */
|
|
if (entering(tcp)) {
|
|
print_sigset_addr_len(tcp, tcp->u_arg[0], tcp->u_arg[3]);
|
|
tprints(", ");
|
|
/* This is the only "return" parameter, */
|
|
if (tcp->u_arg[1] != 0)
|
|
return 0;
|
|
/* ... if it's NULL, can decode all on entry */
|
|
tprints("NULL, ");
|
|
}
|
|
else if (tcp->u_arg[1] != 0) {
|
|
/* syscall exit, and u_arg[1] wasn't NULL */
|
|
printsiginfo_at(tcp, tcp->u_arg[1]);
|
|
tprints(", ");
|
|
}
|
|
else {
|
|
/* syscall exit, and u_arg[1] was NULL */
|
|
return 0;
|
|
}
|
|
print_timespec(tcp, tcp->u_arg[2]);
|
|
tprintf(", %lu", tcp->u_arg[3]);
|
|
return 0;
|
|
};
|
|
|
|
int
|
|
sys_restart_syscall(struct tcb *tcp)
|
|
{
|
|
if (entering(tcp))
|
|
tprints("<... resuming interrupted call ...>");
|
|
return 0;
|
|
}
|
|
|
|
static int
|
|
do_signalfd(struct tcb *tcp, int flags_arg)
|
|
{
|
|
/* NB: kernel requires arg[2] == NSIG / 8 */
|
|
if (entering(tcp)) {
|
|
printfd(tcp, tcp->u_arg[0]);
|
|
tprints(", ");
|
|
print_sigset_addr_len(tcp, tcp->u_arg[1], tcp->u_arg[2]);
|
|
tprintf(", %lu", tcp->u_arg[2]);
|
|
if (flags_arg >= 0) {
|
|
tprints(", ");
|
|
printflags(open_mode_flags, tcp->u_arg[flags_arg], "O_???");
|
|
}
|
|
}
|
|
return 0;
|
|
}
|
|
|
|
int
|
|
sys_signalfd(struct tcb *tcp)
|
|
{
|
|
return do_signalfd(tcp, -1);
|
|
}
|
|
|
|
int
|
|
sys_signalfd4(struct tcb *tcp)
|
|
{
|
|
return do_signalfd(tcp, 3);
|
|
}
|