/* nto-tdep.c - general QNX Neutrino target functionality.
Copyright (C) 2003-2020 Free Software Foundation, Inc.
Contributed by QNX Software Systems Ltd.
This file is part of GDB.
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation; either version 3 of the License, or
(at your option) any later version.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see . */
#include "defs.h"
#include
#include "nto-tdep.h"
#include "top.h"
#include "inferior.h"
#include "infrun.h"
#include "gdbarch.h"
#include "bfd.h"
#include "elf-bfd.h"
#include "solib-svr4.h"
#include "gdbcore.h"
#include "objfiles.h"
#include "source.h"
#include "gdbsupport/pathstuff.h"
#define QNX_NOTE_NAME "QNX"
#define QNX_INFO_SECT_NAME "QNX_info"
#ifdef __CYGWIN__
#include
#endif
#ifdef __CYGWIN__
static char default_nto_target[] = "C:\\QNXsdk\\target\\qnx6";
#elif defined(__sun__) || defined(linux)
static char default_nto_target[] = "/opt/QNXsdk/target/qnx6";
#else
static char default_nto_target[] = "";
#endif
struct nto_target_ops current_nto_target;
static const struct inferior_key
nto_inferior_data_reg;
static char *
nto_target (void)
{
char *p = getenv ("QNX_TARGET");
#ifdef __CYGWIN__
static char buf[PATH_MAX];
if (p)
cygwin_conv_path (CCP_WIN_A_TO_POSIX, p, buf, PATH_MAX);
else
cygwin_conv_path (CCP_WIN_A_TO_POSIX, default_nto_target, buf, PATH_MAX);
return buf;
#else
return p ? p : default_nto_target;
#endif
}
/* Take a string such as i386, rs6000, etc. and map it onto CPUTYPE_X86,
CPUTYPE_PPC, etc. as defined in nto-share/dsmsgs.h. */
int
nto_map_arch_to_cputype (const char *arch)
{
if (!strcmp (arch, "i386") || !strcmp (arch, "x86"))
return CPUTYPE_X86;
if (!strcmp (arch, "rs6000") || !strcmp (arch, "powerpc"))
return CPUTYPE_PPC;
if (!strcmp (arch, "mips"))
return CPUTYPE_MIPS;
if (!strcmp (arch, "arm"))
return CPUTYPE_ARM;
if (!strcmp (arch, "sh"))
return CPUTYPE_SH;
return CPUTYPE_UNKNOWN;
}
int
nto_find_and_open_solib (const char *solib, unsigned o_flags,
gdb::unique_xmalloc_ptr *temp_pathname)
{
char *buf, *arch_path, *nto_root;
const char *endian;
const char *base;
const char *arch;
int arch_len, len, ret;
#define PATH_FMT \
"%s/lib:%s/usr/lib:%s/usr/photon/lib:%s/usr/photon/dll:%s/lib/dll"
nto_root = nto_target ();
if (strcmp (gdbarch_bfd_arch_info (target_gdbarch ())->arch_name, "i386") == 0)
{
arch = "x86";
endian = "";
}
else if (strcmp (gdbarch_bfd_arch_info (target_gdbarch ())->arch_name,
"rs6000") == 0
|| strcmp (gdbarch_bfd_arch_info (target_gdbarch ())->arch_name,
"powerpc") == 0)
{
arch = "ppc";
endian = "be";
}
else
{
arch = gdbarch_bfd_arch_info (target_gdbarch ())->arch_name;
endian = gdbarch_byte_order (target_gdbarch ())
== BFD_ENDIAN_BIG ? "be" : "le";
}
/* In case nto_root is short, add strlen(solib)
so we can reuse arch_path below. */
arch_len = (strlen (nto_root) + strlen (arch) + strlen (endian) + 2
+ strlen (solib));
arch_path = (char *) alloca (arch_len);
xsnprintf (arch_path, arch_len, "%s/%s%s", nto_root, arch, endian);
len = strlen (PATH_FMT) + strlen (arch_path) * 5 + 1;
buf = (char *) alloca (len);
xsnprintf (buf, len, PATH_FMT, arch_path, arch_path, arch_path, arch_path,
arch_path);
base = lbasename (solib);
ret = openp (buf, OPF_TRY_CWD_FIRST | OPF_RETURN_REALPATH, base, o_flags,
temp_pathname);
if (ret < 0 && base != solib)
{
xsnprintf (arch_path, arch_len, "/%s", solib);
ret = open (arch_path, o_flags, 0);
if (temp_pathname)
{
if (ret >= 0)
*temp_pathname = gdb_realpath (arch_path);
else
temp_pathname->reset (NULL);
}
}
return ret;
}
void
nto_init_solib_absolute_prefix (void)
{
char buf[PATH_MAX * 2], arch_path[PATH_MAX];
char *nto_root;
const char *endian;
const char *arch;
nto_root = nto_target ();
if (strcmp (gdbarch_bfd_arch_info (target_gdbarch ())->arch_name, "i386") == 0)
{
arch = "x86";
endian = "";
}
else if (strcmp (gdbarch_bfd_arch_info (target_gdbarch ())->arch_name,
"rs6000") == 0
|| strcmp (gdbarch_bfd_arch_info (target_gdbarch ())->arch_name,
"powerpc") == 0)
{
arch = "ppc";
endian = "be";
}
else
{
arch = gdbarch_bfd_arch_info (target_gdbarch ())->arch_name;
endian = gdbarch_byte_order (target_gdbarch ())
== BFD_ENDIAN_BIG ? "be" : "le";
}
xsnprintf (arch_path, sizeof (arch_path), "%s/%s%s", nto_root, arch, endian);
xsnprintf (buf, sizeof (buf), "set solib-absolute-prefix %s", arch_path);
execute_command (buf, 0);
}
char **
nto_parse_redirection (char *pargv[], const char **pin, const char **pout,
const char **perr)
{
char **argv;
const char *in, *out, *err, *p;
int argc, i, n;
for (n = 0; pargv[n]; n++);
if (n == 0)
return NULL;
in = "";
out = "";
err = "";
argv = XCNEWVEC (char *, n + 1);
argc = n;
for (i = 0, n = 0; n < argc; n++)
{
p = pargv[n];
if (*p == '>')
{
p++;
if (*p)
out = p;
else
out = pargv[++n];
}
else if (*p == '<')
{
p++;
if (*p)
in = p;
else
in = pargv[++n];
}
else if (*p++ == '2' && *p++ == '>')
{
if (*p == '&' && *(p + 1) == '1')
err = out;
else if (*p)
err = p;
else
err = pargv[++n];
}
else
argv[i++] = pargv[n];
}
*pin = in;
*pout = out;
*perr = err;
return argv;
}
static CORE_ADDR
lm_addr (struct so_list *so)
{
lm_info_svr4 *li = (lm_info_svr4 *) so->lm_info;
return li->l_addr;
}
static CORE_ADDR
nto_truncate_ptr (CORE_ADDR addr)
{
if (gdbarch_ptr_bit (target_gdbarch ()) == sizeof (CORE_ADDR) * 8)
/* We don't need to truncate anything, and the bit twiddling below
will fail due to overflow problems. */
return addr;
else
return addr & (((CORE_ADDR) 1 << gdbarch_ptr_bit (target_gdbarch ())) - 1);
}
static Elf_Internal_Phdr *
find_load_phdr (bfd *abfd)
{
Elf_Internal_Phdr *phdr;
unsigned int i;
if (!elf_tdata (abfd))
return NULL;
phdr = elf_tdata (abfd)->phdr;
for (i = 0; i < elf_elfheader (abfd)->e_phnum; i++, phdr++)
{
if (phdr->p_type == PT_LOAD && (phdr->p_flags & PF_X))
return phdr;
}
return NULL;
}
void
nto_relocate_section_addresses (struct so_list *so, struct target_section *sec)
{
/* Neutrino treats the l_addr base address field in link.h as different than
the base address in the System V ABI and so the offset needs to be
calculated and applied to relocations. */
Elf_Internal_Phdr *phdr = find_load_phdr (sec->the_bfd_section->owner);
unsigned vaddr = phdr ? phdr->p_vaddr : 0;
sec->addr = nto_truncate_ptr (sec->addr + lm_addr (so) - vaddr);
sec->endaddr = nto_truncate_ptr (sec->endaddr + lm_addr (so) - vaddr);
}
/* This is cheating a bit because our linker code is in libc.so. If we
ever implement lazy linking, this may need to be re-examined. */
int
nto_in_dynsym_resolve_code (CORE_ADDR pc)
{
if (in_plt_section (pc))
return 1;
return 0;
}
void
nto_dummy_supply_regset (struct regcache *regcache, char *regs)
{
/* Do nothing. */
}
static void
nto_sniff_abi_note_section (bfd *abfd, asection *sect, void *obj)
{
const char *sectname;
unsigned int sectsize;
/* Buffer holding the section contents. */
char *note;
unsigned int namelen;
const char *name;
const unsigned sizeof_Elf_Nhdr = 12;
sectname = bfd_section_name (sect);
sectsize = bfd_section_size (sect);
if (sectsize > 128)
sectsize = 128;
if (sectname != NULL && strstr (sectname, QNX_INFO_SECT_NAME) != NULL)
*(enum gdb_osabi *) obj = GDB_OSABI_QNXNTO;
else if (sectname != NULL && strstr (sectname, "note") != NULL
&& sectsize > sizeof_Elf_Nhdr)
{
note = XNEWVEC (char, sectsize);
bfd_get_section_contents (abfd, sect, note, 0, sectsize);
namelen = (unsigned int) bfd_h_get_32 (abfd, note);
name = note + sizeof_Elf_Nhdr;
if (sectsize >= namelen + sizeof_Elf_Nhdr
&& namelen == sizeof (QNX_NOTE_NAME)
&& 0 == strcmp (name, QNX_NOTE_NAME))
*(enum gdb_osabi *) obj = GDB_OSABI_QNXNTO;
XDELETEVEC (note);
}
}
enum gdb_osabi
nto_elf_osabi_sniffer (bfd *abfd)
{
enum gdb_osabi osabi = GDB_OSABI_UNKNOWN;
bfd_map_over_sections (abfd,
nto_sniff_abi_note_section,
&osabi);
return osabi;
}
static const char *nto_thread_state_str[] =
{
"DEAD", /* 0 0x00 */
"RUNNING", /* 1 0x01 */
"READY", /* 2 0x02 */
"STOPPED", /* 3 0x03 */
"SEND", /* 4 0x04 */
"RECEIVE", /* 5 0x05 */
"REPLY", /* 6 0x06 */
"STACK", /* 7 0x07 */
"WAITTHREAD", /* 8 0x08 */
"WAITPAGE", /* 9 0x09 */
"SIGSUSPEND", /* 10 0x0a */
"SIGWAITINFO", /* 11 0x0b */
"NANOSLEEP", /* 12 0x0c */
"MUTEX", /* 13 0x0d */
"CONDVAR", /* 14 0x0e */
"JOIN", /* 15 0x0f */
"INTR", /* 16 0x10 */
"SEM", /* 17 0x11 */
"WAITCTX", /* 18 0x12 */
"NET_SEND", /* 19 0x13 */
"NET_REPLY" /* 20 0x14 */
};
const char *
nto_extra_thread_info (struct target_ops *self, struct thread_info *ti)
{
if (ti != NULL && ti->priv != NULL)
{
nto_thread_info *priv = get_nto_thread_info (ti);
if (priv->state < ARRAY_SIZE (nto_thread_state_str))
return nto_thread_state_str [priv->state];
}
return "";
}
void
nto_initialize_signals (void)
{
/* We use SIG45 for pulses, or something, so nostop, noprint
and pass them. */
signal_stop_update (gdb_signal_from_name ("SIG45"), 0);
signal_print_update (gdb_signal_from_name ("SIG45"), 0);
signal_pass_update (gdb_signal_from_name ("SIG45"), 1);
/* By default we don't want to stop on these two, but we do want to pass. */
#if defined(SIGSELECT)
signal_stop_update (SIGSELECT, 0);
signal_print_update (SIGSELECT, 0);
signal_pass_update (SIGSELECT, 1);
#endif
#if defined(SIGPHOTON)
signal_stop_update (SIGPHOTON, 0);
signal_print_update (SIGPHOTON, 0);
signal_pass_update (SIGPHOTON, 1);
#endif
}
/* Read AUXV from initial_stack. */
LONGEST
nto_read_auxv_from_initial_stack (CORE_ADDR initial_stack, gdb_byte *readbuf,
LONGEST len, size_t sizeof_auxv_t)
{
gdb_byte targ32[4]; /* For 32 bit target values. */
gdb_byte targ64[8]; /* For 64 bit target values. */
CORE_ADDR data_ofs = 0;
ULONGEST anint;
LONGEST len_read = 0;
gdb_byte *buff;
enum bfd_endian byte_order;
int ptr_size;
if (sizeof_auxv_t == 16)
ptr_size = 8;
else
ptr_size = 4;
/* Skip over argc, argv and envp... Comment from ldd.c:
The startup frame is set-up so that we have:
auxv
NULL
...
envp2
envp1 <----- void *frame + (argc + 2) * sizeof(char *)
NULL
...
argv2
argv1
argc <------ void * frame
On entry to ldd, frame gives the address of argc on the stack. */
/* Read argc. 4 bytes on both 64 and 32 bit arches and luckily little
* endian. So we just read first 4 bytes. */
if (target_read_memory (initial_stack + data_ofs, targ32, 4) != 0)
return 0;
byte_order = gdbarch_byte_order (target_gdbarch ());
anint = extract_unsigned_integer (targ32, sizeof (targ32), byte_order);
/* Size of pointer is assumed to be 4 bytes (32 bit arch.) */
data_ofs += (anint + 2) * ptr_size; /* + 2 comes from argc itself and
NULL terminating pointer in
argv. */
/* Now loop over env table: */
anint = 0;
while (target_read_memory (initial_stack + data_ofs, targ64, ptr_size)
== 0)
{
if (extract_unsigned_integer (targ64, ptr_size, byte_order) == 0)
anint = 1; /* Keep looping until non-null entry is found. */
else if (anint)
break;
data_ofs += ptr_size;
}
initial_stack += data_ofs;
memset (readbuf, 0, len);
buff = readbuf;
while (len_read <= len-sizeof_auxv_t)
{
if (target_read_memory (initial_stack + len_read, buff, sizeof_auxv_t)
== 0)
{
/* Both 32 and 64 bit structures have int as the first field. */
const ULONGEST a_type
= extract_unsigned_integer (buff, sizeof (targ32), byte_order);
if (a_type == AT_NULL)
break;
buff += sizeof_auxv_t;
len_read += sizeof_auxv_t;
}
else
break;
}
return len_read;
}
/* Return nto_inferior_data for the given INFERIOR. If not yet created,
construct it. */
struct nto_inferior_data *
nto_inferior_data (struct inferior *const inferior)
{
struct inferior *const inf = inferior ? inferior : current_inferior ();
struct nto_inferior_data *inf_data;
gdb_assert (inf != NULL);
inf_data = nto_inferior_data_reg.get (inf);
if (inf_data == NULL)
inf_data = nto_inferior_data_reg.emplace (inf);
return inf_data;
}