There are options for listing the current device/hw tree and memory
regions, but no way to find out at run time all the current mappings.
So add a new --map-info option akin to the --memory-info option which
displays all the current mappings.
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
When the sim is configured with profile support disabled, the build fails:
./../common/sim-profile.c: In function 'profile_option_handler':
./../common/sim-profile.c:337:6: warning: implicit declaration of function 'PROFILE_PC_FREQ'
./../common/sim-profile.c:337:6: error: lvalue required as left operand of assignment
./../common/sim-profile.c:351:6: warning: implicit declaration of function 'PROFILE_PC_NR_BUCKETS'
./../common/sim-profile.c:351:6: error: lvalue required as left operand of assignment
./../common/sim-profile.c:381:6: warning: implicit declaration of function 'PROFILE_PC_SHIFT'
./../common/sim-profile.c:381:6: error: lvalue required as left operand of assignment
./../common/sim-profile.c:405:8: warning: implicit declaration of function 'PROFILE_PC_START'
./../common/sim-profile.c:405:8: error: lvalue required as left operand of assignment
./../common/sim-profile.c:406:8: warning: implicit declaration of function 'PROFILE_PC_END'
./../common/sim-profile.c:406:8: error: lvalue required as left operand of assignment
./../common/sim-profile.c: In function 'profile_uninstall':
./../common/sim-profile.c:1299:7: warning: implicit declaration of function 'PROFILE_INSN_COUNT'
./../common/sim-profile.c:1299:37: warning: comparison between pointer and integer
./../common/sim-profile.c:1300:2: warning: passing argument 1 of 'zfree' makes pointer from integer without a cast
../common/sim-utils.h:30:6: note: expected 'void *' but argument is of type 'int'
make[2]: *** [sim-profile.o] Error 1
So add some stubs similar to how some of the other subsystems are
stubbed out so things build correctly.
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
When building with device and hw support, the sim-core code generates a
lot of build time warnings such as:
./../common/sim-core.c: In function 'sim_core_map_attach':
./../common/sim-core.c:198:7: warning: passing argument 1 of 'device_error' from incompatible pointer type
../common/sim-core.h:347:6: note: expected 'struct device *' but argument is of type 'struct hw *'
./../common/sim-core.c:235:7: warning: passing argument 1 of 'device_error' from incompatible pointer type
../common/sim-core.h:347:6: note: expected 'struct device *' but argument is of type 'struct hw *'
In reality, these two structures get cast back and forth in the core
code already and so are "compatible". So tweak the three functions
that generate all of these warnings to include the casts automatically.
I know this isn't exactly clean, but the current device/hw ifdef
approach is full of landmines itself and I'm not entirely sure how
to unscrew it.
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Rather than having to bang out chunks of data one byte at a time over
the socket interface, add a write variant that accepts an arbitrarily
long buffer. This speeds things up considerably when we have many
chars to send out at once.
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
* rx.h (RX_Operand_Type): Add TwoReg.
(RX_Opcode_ID): Remove ediv and ediv2.
[opcodes]
* rx-decode.opc (SRR): New.
(rx_decode_opcode): Use it for movbi and movbir. Decode NOP2 (mov
r0,r0) and NOP3 (max r0,r0) special cases.
* rx-decode.c: Regenerate.
[sim/rx]
* rx.c (decode_cache_base): New.
(id_names): Remove ediv and edivu.
(optype_names): Add TwoReg.
(maybe_get_mem_page): New.
(rx_get_byte): Call it.
(get_op): Add TwoReg support.
(put_op): Likewise.
(PD, PS, PS2, GD, GS, GS2, DSZ, SSZ, S2SZ, US1, US2, OM): "opcode"
is a pointer now.
(DO_RETURN): New. We use longjmp to return an exception result.
(decode_opcode): Make opcode a pointer to the decode cache. Save
decoded opcode information and re-use. Call DO_RETURN instead of
return throughout. Remove ediv and edivu.
* mem.c (ptdc): New. Adds decode cache.
(rx_mem_ptr): Support it.
(rx_mem_decode_cache): New.
* mem.h (enum mem_ptr_action): add MPA_DECODE_CACHE.
(rx_mem_decode_cache): Declare.
* gdb-if.c (sim_resume): Add decode_opcode's setjmp logic here...
* main.c (main): ...and here. Use a fast loop if neither trace
nor disassemble is given.
* cpu.h (RX_MAKE_STEPPED, RX_MAKE_HIT_BREAK, RX_MAKE_EXITED,
RX_MAKE_STOPPED, RX_EXITED, RX_STOPPED): Adjust so that 0 is not a
valid code for anything.
gdb/
* ser-tcp.c (net_open): Check error return from socket() call by its
equality to -1 not by it being negative.
(net_close): Likewise.
gdb/gdbserver/
* gdbreplay.c (remote_open): Check error return from socket() call by
its equality to -1 not by it being negative.
* remote-utils.c (remote_open): Likewise.
sim/arm/
* communicate.c (MYread_char): Check error return from accept() call
by its equality to -1 not by it being negative.
(MYread_charwait): Likewise.
* main.c (main): Likewise for both socket() and accept() calls.
sim/common/
* dv-sockser.c (dv_sockser_init): Check error return from socket()
call by its equality to -1 not by it being negative.
(connected_p): Likewise for accept() call.
sim/cris/
* dv-rv.c (hw_rv_init_socket): Check error return from socket() call
by its equality to -1 not by it being negative.
(hw_rv_write): Likewise.
(hw_rv_handle_incoming): Likewise.
(hw_rv_poll_once): Likewise.
* rvdummy.c (setupsocket): Likewise.
(main): Likewise for accept() call as returned from setupsocket().
sim/m32c/
* main.c (setup_tcp_console): Check error return from socket() call
by its equality to -1 not by it being negative.
* erc32.c (sis_memory_write): Change prototype to const unsigned char *.
* func.c (exec_cmd, event, advance_time, wait_for_irq): Use uint64
for counts.
* interf.c (run_sim): Change icount to uint64_t. Use strtol directly.
(sim_resume): Specify maximum run time as uint64.
* sis.c (run_sim): Change icount to uint64_t.
* sis.h: Define uint64 as uint64_t. Change various fields and
prototypes to uint64 to support longer simulations.
The testsuite subdir has a note about unifying the target->subdir logic,
so do just that. The end goal here is to have `make check` work out of
the box without having to delve into dejagnu internals.
The target-specific logic is split out of the top level configure.ac file
and into a dedicated configure.tgt similar to other subprojects (gdb and
ld and etc...) with the difference that this file has to be included at
the m4 level instead of the shell level. This is necessary only because
autoconf requires AC_CONFIG_SUBDIRS be given a string literal and not a
variable value.
Then the toplevel and the testsuite configure files pull this in, the sim
subdir gets expanded into testsuite/site.exp, and the default sim run code
uses this info to set the sim path to the local compiled run file if it
hasn't already been specified.
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
The current fpu code with externals enabled results in the warnings:
common/sim-fpu.c:2437: warning: missing initializer
common/sim-fpu.c:2437: warning: (near initialization for 'sim_fpu_zero.sign')
common/sim-fpu.c:2440: warning: missing initializer
common/sim-fpu.c:2440: warning: (near initialization for 'sim_fpu_qnan.sign')
So tweak the old style initializers to avoid these.
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
The common/ code uses sim_cpu rather than SIM_CPU to avoid inter-header
dependency issues, so follow convention to fix building some targets.
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
The profile code already has options and handling for accepting a file to
write the profile output like the trace code, but it doesn't actually use
it. At the moment, it simply opens the file at the start and closes it at
the end. So add two new local functions the way the trace code is doing
it and have them figure out whether to write the output to stdout or the
specified file. Then convert all existing output in the profile code to
use these helpers.
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
There is an architecture-info flag for listing possible arch values, but
there is on equivalent for listing possible model values. So add the
equivalent for models.
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
As pointed out by Sandra Loosemore, a bunch of targets define sim_write
themselves instead of using the common/ code. So constify them too.
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
A few ports rely on internal dv-sockser state in order to detect whether
a connection has been made (look for 'extern sockser_addr'). Rather than
continuing that tradition, extend the existing status function to return
the socket connection status.
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Most the sim write functions declare their source buffer const because
they only ever read from it. The global sim_write() function does not
follow this convention though which causes some warnings when trying to
pass it const strings or buffers.
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
I found the documentation lacking in many places, so I tried filling in a
lot of holes that I personally fell into.
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
The profile code has a lot of helper macros already, but none yet for the
branch profiling code. So add ones for the basic functions -- taken and
untaken branches.
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
The sim code has a lot of static initializer for options and devices, but
since they aren't using newer struct style, they have to specify a value
for every option otherwise gcc spits a lot of warnings about "missing
initializer". So add NULL/0 stubs for pointers/values.
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
The dup_arg_p function is only given const strings, and it maintains a
local table of those const strings. So constify the whole thing to fix
the GCC warnings:
common/sim-options.c: In function 'sim_parse_args':
common/sim-options.c:559: warning: passing argument 1 of 'dup_arg_p'
discards qualifiers from pointer target type
common/sim-options.c: In function 'print_help':
common/sim-options.c:675: warning: passing argument 1 of 'dup_arg_p'
discards qualifiers from pointer target type
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
I've committed the following patch as obvious. The local "prefix"
variable is only assigned const strings, and only passed to printf()
functions, so add "const" to avoid gcc warnings:
common/sim-fpu.c: In function 'sim_fpu_print_status':
common/sim-fpu.c:2508: warning: initialization discards qualifiers
from pointer target type
common/sim-fpu.c:2566: warning: assignment discards qualifiers
from pointer target type
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
The local save_data() function takes a pointer to a buffer and only uses it as
the source to the memcpy() function. Since it is given const strings, GCC
likes to spit out warnings:
common/sim-trace.c: In function 'trace_prefix':
common/sim-trace.c:697: warning: passing argument 5 of 'save_data' discards
qualifiers from pointer target type
So I've committed this as obvious.
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
The sim-options code assigns a const string to "char *" in decoding the
current environment settings, and only uses it to pass to a printf. GCC
outputs the warnings:
common/sim-options.c: In function 'standard_option_handler':
common/sim-options.c:271: warning: assignment discards qualifiers from pointer
target type
common/sim-options.c:272: warning: assignment discards qualifiers from pointer
target type
common/sim-options.c:273: warning: assignment discards qualifiers from pointer
target type
So I've committed this as "obvious".
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
The accept() function takes a socklen_t, not an int. Using an int causes:
dv-sockser.c: In function 'connected_p':
dv-sockser.c:273: warning: pointer targets in passing argument 3
of 'accept' differ in signedness
So use the same socklen_t detection code as gdb and convert the accept().
This is the normal "passing a buffer straight to printf makes GCC warn".
While we know this particular case is fine, the trend in this source tree
is to fix this anyways to avoid the warnings.
GCC issues warnings because const strings like "foo" are passed as char*.
sim-watch.c: In function 'watchpoint_type_to_str':
sim-watch.c:120: warning: return discards qualifiers from pointer target type
The trace_option_handler() function only uses cpu_nr when the
SIM_HAVE_ADDR_RANGE define is enabled. So move the decl down
to where the code exists. Otherwise GCC warns:
sim-trace.c: In function 'trace_option_handler':
sim-trace.c:236: warning: unused variable 'cpu_nr'
The device_error() takes a printf style string, so update the prototype
accordingly. The message should be const and it should use an attribute.
This fixes gcc warnings like:
sim-core.c: In function 'sim_core_map_attach':
sim-core.c:200: warning: passing argument 2 of 'device_error' discards qualifiers from pointer target type
sim-core.c:237: warning: passing argument 2 of 'device_error' discards qualifiers from pointer target type
sim-core.c: In function 'sim_core_attach':
sim-core.c:304: warning: passing argument 2 of 'device_error' discards qualifiers from pointer target type
sim-core.c:314: warning: passing argument 2 of 'device_error' discards qualifiers from pointer target type
sim-core.c:335: warning: passing argument 2 of 'device_error' discards qualifiers from pointer target type
sim-core.c:348: warning: passing argument 2 of 'device_error' discards qualifiers from pointer target type
The gdb code always passes down -E <little|big> to the sim core when using
the sim target. But the sim core only recognizes this option when the sim
supports big endian systems. So for little endian simulators, any attempt
to use the sim target fails with:
(gdb) target sim
gdbsim: invalid option -- 'E'
unable to create simulator instance
Since always respecting the option doesn't cause any problems, do just
that. If someone tries to use an invalid endian, they'll get an error
anyways.
The common code sets up an autoconf option --enable-sim-trace which adds
-DTRACE= to CPPFLAGS. This causes warnings in the building of some files
that declare a local TRACE() helper macro. So punt it from hw-ports.c
(since it isn't actually used) and convert hw-properties.c to HW_TRACE().
Including stdbool.h before hw-properties.h results in a build error due
to the hw_add_boolean_property function having an argument named "bool"
in its prototype. The source file has already be renamed to not use
this ("boolean" instead), so match the header to the source.
* configure: Regenerate.
* config.in: Regenerate.
* main.c: Include config.h.
Use HAVE_STDLIB_H, HAVE_UNISTD_H, HAVE_GETOPT_H.
Include getopt.h in case HAVE_GETOPT_H is defined.
* compile.c(fetch_1): Fix pre-dec, pre-inc, post-dec and post-inc.
Index registers not masked memory areas.
Only simply increment or decrement.
* compile.c(store_1): Ditto.
netinet/in.h, and netinet/tcp.h.
* configure: Regenerate.
* config.in: Add those headers.
* main.c: Check for them.
(setup_tcp_console): Disable if no networking.
(main): Note missing networking or termios.
* mem.c: Check for those headers.
(stdin_ready): Disable if no termios.
(m32c_sim_restore_console): Disable if no termios.
(mem_get_byte): Disable console input if no termios.
identifiers for file types rather than encoding internal
implementation details of cgen.sh.
(CGEN_CPU_WRITE,CGEN_CPU_SEM,CGEN_CPU_SEMSW): Ditto.
* cgen.sh: Add support for parallel makes.
Rewrite cpu/decode handling to avoid generating "extrafiles" twice.
2009-05-18 Jon Beniston <jon@beniston.com>
* MAINTAINERS: Add lm32 target.
* Makefile.in: Add lm32 dependencies.
* NEWS: Indicate lm32 is a new target.
* configure.tgt: Add lm32 targets.
* lm32-tdep.c: New file.
gdb/testsuite
2009-05-18 Jon Beniston <jon@beniston.com>
* gdb.asm/asm-source.exp: Add lm32 target.
include/gdb/
2009-05-18 Jon Beniston <jon@beniston.com>
* sim-lm32.h: New file.
sim/
2009-05-18 Jon Beniston <jon@beniston.com>
* MAINTAINERS: Add Jon Beniston as maintainer of lm32 sim.
* configure.ac: Add lm32 target.
* lm32: New directory.
sim/common
2009-05-18 Jon Beniston <jon@beniston.com>
* gennltvals.sh: Add lm32 target.
* nltvals.def: Add lm32 syscall definitions.
sim/lm32/
2009-05-18 Jon Beniston <jon@beniston.com>
* Makefile.in: New file.
* arch.c: New file.
* arch.h: New file.
* config.in: New file.
* configure: New file.
* configure.ac: New file.
* cpu.c: New file.
* cpu.h: New file.
* cpuall.h: New file.
* decode.c: New file.
* decode.h: New file.
* dv-lm32cpu.c: New file.
* dv-lm32timer.c: New file.
* dv-lm32uart.c: New file.
* lm32.c: New file.
* lm32-sim.h: New file.
* mloop.in: New file.
* model.c: New file.
* sem.c: New file.
* sem-switch.c: New file.
* sim-if.c: New file.
* sim-main.c: New file.
* tconfig.in: New file.
* traps.c: New file.
* user.c: New file.
m32c_signal_to_target. Change signal return values from SIGILL,
SIGTRAP, SIGSEGV, etc. to TARGET_SIGNAL_ILL, TARGET_SIGNAL_TRAP,
TARGET_SIGNAL_SEGV, etc. Fix all callers.
(cris_start_address, cris_program_offset): New variables.
(OPTION_CRIS_PROGRAM_OFFSET, OPTION_CRIS_STARTADDR): New option
enums.
(cris_options): New options --cris-program-offset and
--cris-start-address.
(cris_option_handler): Handle new options.
(cris_program_offset_write, cris_set_section_offset_iterator)
(cris_offset_sections, cris_offset_sections): New functions.
(sim_load): Use cris_program_offset_write as function argument to
cris_load_elf_file, not sim_write.
(struct offsetinfo): New struct.
(cris_handle_interpreter): Fix typo in comment.
(sim_open): Call cris_offset_sections as soon as the bfd of the
infile is available. Gate bfd validity checks on abfd non-NULL.
(sim_create_inferior): Let cris_start_address when != -1 override
other start-address choices.
(MULHI, NEGHI, ABSHI, ADDSI, SUBSI, MULSI, NEGSI, ABSSI, ADDDI)
(SUBDI, MULDI, NEGDI, ABSDI): Cast arguments to the unsigned type
variant; UQI, UHI, USI, UDI, and cast the result to the signed
type, QI, HI, SI, DI.
(MULHI, NEGHI, ABSHI, ADDSI, SUBSI, MULSI, NEGSI, ABSSI, ADDDI)
(SUBDI, MULDI, NEGDI, ABSDI): Cast arguments to the unsigned type
variant; UQI, UHI, USI, UDI, and cast the result to the signed
type, QI, HI, SI, DI.
(create_map): Make -1 imply a non-fixed address, not 0. All
callers changed. Only prefer the next higher unmapped address if
the last mapped address is no less than 0x40000000. Check that
the address to be mapped is not already mapped. Update head
comment.
(unmap_pages): Don't call abort when recursive call fails, just
note and return an error if a page in the range couldn't be unmapped.
(cris_bmod_handler, h_supr_set_handler, h_supr_get_handler)
(schedule, make_first_thread, cris_pipe_empty): New local variable sd.
(cris_break_13_handler) <case TARGET_SYS_mmap2>: Handle
non-MAP_FIXED argument overlapping existing map. For MAP_FIXED,
don't abort on page not being mapped. Handle non-anon filemap
with length padded to pagesize.
set_target_thread_data.
* cris/crisv32f.c (CRIS_TLS_REGISTER): Define.
* cris/crisv10f.c: Ditto.
* cris/cris-tmpl.c (MY (set_target_thread_data)): New function.
(MY (f_specific_init)): Set new _sim_cpu member to new function.
* cris/traps.c (TARGET_SYS_set_thread_area): Define.
(cris_break_13_handler) <case TARGET_SYS_set_thread_area>: New
case.
(TARGET_EPOCH): Update to match TARGET_UTSNAME. Correct comment.
(cris_break_13_handler) <case TARGET_SYS_uname>: Update to
2.6.27. Set machine field to the BFD printable name of the
machine.
sim/cris/c/badldso2.c, sim/cris/c/badldso3.c,
sim/cris/c/helloaout.c, sim/cris/c/hellodyn.c,
sim/cris/c/hellodyn2.c, sim/cris/c/writev1.c,
sim/cris/c/writev2.c: New tests.
* sim/cris/c/c.exp: If compiler links libc.so when attempting to
link dynamically, create symlink named "lib" to the directory
where it is found. Handle new test-case option "dynamic".
(is_mapped_only, cris_dump_map): New functions.
(cris_break_13_handler) <case TARGET_SYS_mmap2>: Handle more flags
and prot combinations and a non-zero page-offset. If
TARGET_MAP_FIXED, unmap pages before mapping them.
<case TARGET_SYS_mprotect>: When checking, allow any length
argument. Don't actually do anything.
<case TARGET_SYS_writev>: New case.
* cris/sim-if.c: Include elf-bfd.h.
(struct progbounds): New members end_loadmem, start_nonloadmem.
(xprintf, eprintf): New functions, copied from common/sim-load.c.
(cris_load_elf_file, sim_load, cris_get_progbounds): New functions.
(get_progbounds_iterator): Renamed from get_progbounds. Make
static. Update head comment. Set new struct progbounds members.
(exec_load_addr, interp_load_addr, interp_start_addr): New static
variables.
(aux_ent_phdr, aux_ent_phent, aux_ent_phnum, aux_ent_base)
(aux_ent_entry, cris_write_interp, cris_handle_interpreter): New
functions.
(sim_open): New constant array auxv_entries. Rewrite AUX_ENT
handling to use auxv_entries. Improve error message and checking
for invalid programs. Use new variable abfd for the program
instead of for each access reaching into sd to get it.
(sim_create_inferior): If non-zero, use interp_start_addr instead
of the program start address.
(cris_disassemble_insn): Remove incorrect and unclear, supposedly
stale comment. Always specify little-endian.
* configure: Regenerated.
* configure.ac: Add test for System V shared memory and semaphore.
* debug.c, debug.h: Add trace support for new devices.
* hw_sem.c, hw_shm.c: New files.
* Makefile.in: Add hw_sem.c and hw_shm.c.
command interpreter to use to run the input file. This is
necessary because otherwise SHELL is taken from the user's
environment, and not from the makefile that invoked this script
and the user might not be running an sh-like shell.
* cris/Makefile.in: Pass -shell parameter to genmloop.sh.
* fr30/Makefile.in: Likewise.
* frv/Makefile.in: Likewise.
* i960/Makefile.in: Likewise.
* iq2000/Makefile.in: Likewise.
* m32r/Makefile.in: Likewise.
* frv/mloop.in: Add missing start of line comment marker.
* main.c (main): Add option to set raw console.
* mem.h (m32c_use_raw_console): Declare.
* mem.c (m32c_sim_restore_console): Only restore console if it's
been previously set.
(m32c_use_raw_console): Define.
(mem_get_byte): Set raw console if m32c_use_raw_console is set.
