Yao Qi [Thu, 3 Mar 2016 09:17:44 +0000 (09:17 +0000)]
Step over fork/vfork syscall insn in gdbserver
We can also extend disp-step-syscall.exp to test GDBserver step over
breakpoint on syscall instruction. That is, we set a breakpoint
with a false condition on syscall instruction, so that GDBserver will
step over it.
This test triggers a GDBserver internal error, which can be fixed by
this series.
(gdb) PASS: gdb.base/disp-step-syscall.exp: fork: break cond on target: break on syscall insns
continue^M
Continuing.^M
Remote connection closed^M
(gdb) FAIL: gdb.base/disp-step-syscall.exp: fork: break cond on target: continue to fork again
In GDBserver, there is an internal error,
/home/yao/SourceCode/gnu/gdb/git/gdb/gdbserver/linux-low.c:1922: A problem internal to GDBserver has been detected.
unsuspend LWP 25554, suspended=-1
the simplified reproducer is like,
$ ./gdb ./testsuite/outputs/gdb.base/disp-step-syscall/disp-step-fork
(gdb) b main
(gdb) c
(gdb) disassemble fork // in order to find the address of insn 'syscall'
....
0x00007ffff7ad6023 <+179>: syscall
(gdb) b *0x00007ffff7ad6023 if main == 0
(gdb) c
gdb/testsuite:
2016-03-03 Yao Qi <yao.qi@linaro.org>
* gdb.base/disp-step-syscall.exp (break_cond_on_syscall): New.
If target supports condition evaluation on target, invoke
break_cond_on_syscall for fork and vfork.
Yao Qi [Thu, 3 Mar 2016 09:17:44 +0000 (09:17 +0000)]
Step over syscalll insn with disp-step on and off
disp-step-syscall.exp was added to test displaced stepping over syscall
instructions, in which we set breakpoint on syscall instruction, and
step over it. In fact, we can extend the test to non-displaced-stepping
case. This patch wraps the test with displaced stepping on and off.
Note that the indentation and format isn't adjusted here to make this
patch easy to read. The following patch will fix the format separately.
gdb/testsuite:
2016-03-03 Yao Qi <yao.qi@linaro.org>
* gdb.base/disp-step-syscall.exp: Don't invoke
support_displaced_stepping.
(disp_step_cross_syscall): Test with displaced stepping off and
on if supported.
Yao Qi [Thu, 3 Mar 2016 09:17:44 +0000 (09:17 +0000)]
Refactor gdb.base/disp-step-syscall.exp for general step over test
This patch moves some code out of disp_step_cross_syscall to a new proc
check_pc_after_cross_syscall and setup. Procedure setup is to start a
fresh GDB and compute the syscall instruction address.
gdb/testsuite:
2016-03-03 Yao Qi <yao.qi@linaro.org>
* gdb.base/disp-step-syscall.exp (check_pc_after_cross_syscall): New
proc.
(setup): New proc.
(disp_step_cross_syscall): Move code to check_pc_after_cross_syscall
and setup.
Yao Qi [Thu, 3 Mar 2016 09:17:44 +0000 (09:17 +0000)]
[GDBserver] Leave child suspended when step over parent
I see the following GDBserver internal error in two cases,
gdb/gdbserver/linux-low.c:1922: A problem internal to GDBserver has been detected.
unsuspend LWP 17200, suspended=-1
1. step over a breakpoint on fork/vfork syscall instruction,
2. step over a breakpoint on clone syscall instruction and child
threads hits a breakpoint,
the stack backtrace is
#0 internal_error (file=file@entry=0x44c4c0 "gdb/gdbserver/linux-low.c", line=line@entry=1922,
fmt=fmt@entry=0x44c7d0 "unsuspend LWP %ld, suspended=%d\n") at gdb/gdbserver/../common/errors.c:51
#1 0x0000000000424014 in lwp_suspended_decr (lwp=<optimised out>, lwp=<optimised out>) at gdb/gdbserver/linux-low.c:1922
#2 0x000000000042403a in unsuspend_one_lwp (entry=<optimised out>, except=0x66e8c0) at gdb/gdbserver/linux-low.c:2885
#3 0x0000000000405f45 in find_inferior (list=<optimised out>, func=func@entry=0x424020 <unsuspend_one_lwp>, arg=arg@entry=0x66e8c0)
at gdb/gdbserver/inferiors.c:243
#4 0x00000000004297de in unsuspend_all_lwps (except=0x66e8c0) at gdb/gdbserver/linux-low.c:2895
#5 linux_wait_1 (ptid=..., ourstatus=ourstatus@entry=0x665ec0 <last_status>, target_options=target_options@entry=0)
at gdb/gdbserver/linux-low.c:3632
#6 0x000000000042a764 in linux_wait (ptid=..., ourstatus=0x665ec0 <last_status>, target_options=0)
at gdb/gdbserver/linux-low.c:3770
#7 0x0000000000411163 in mywait (ptid=..., ourstatus=ourstatus@entry=0x665ec0 <last_status>, options=options@entry=0, connected_wait=connected_wait@entry=1)
at gdb/gdbserver/target.c:214
#8 0x000000000040b1f2 in resume (actions=0x66f800, num_actions=1) at gdb/gdbserver/server.c:2757
#9 0x000000000040f660 in handle_v_cont (own_buf=0x66a630 "vCont;c:p45e9.-1") at gdb/gdbserver/server.c:2719
when GDBserver steps over a thread, other threads have been suspended,
the "stepping" thread may create new thread, but GDBserver doesn't set
it suspend count to 1. When GDBserver unsuspend threads, the child's
suspend count goes to -1, and the assert is triggered. In fact, GDBserver
has already taken care of suspend count of new thread when GDBserver is
suspending all threads except the one GDBserver wants to step over by
https://sourceware.org/ml/gdb-patches/2015-07/msg00946.html
+ /* If we're suspending all threads, leave this one suspended
+ too. */
+ if (stopping_threads == STOPPING_AND_SUSPENDING_THREADS)
+ {
+ if (debug_threads)
+ debug_printf ("HEW: leaving child suspended\n");
+ child_lwp->suspended = 1;
+ }
but that is not enough, because new thread is still can be spawned in
the thread which is being stepped over. This patch extends the
condition that GDBserver set child's suspend count to one if it is
suspending threads or stepping over the thread.
gdb/gdbserver:
2016-03-03 Yao Qi <yao.qi@linaro.org>
PR server/19736
* linux-low.c (handle_extended_wait): Set child suspended
if event_lwp->bp_reinsert isn't zero.
H.J. Lu [Tue, 1 Mar 2016 20:39:49 +0000 (12:39 -0800)]
Update PR ld/12365 test for GCC 6
PR ld/12365 test is to verify that linker catches the reference to
undefined symbol error caused by a GCC bug. Since the GCC bug has
been fixed in GCC 6, update the test to verify that there is no
reference to undefined symbol in executable
* testsuite/ld-plugin/lto.exp: Update PR ld/12365 test for GCC 6.
Pedro Alves [Tue, 1 Mar 2016 17:42:47 +0000 (17:42 +0000)]
Fix gdb.trace/ftrace-lock.c compilation
Fixes, on F23:
.../src/gdb/testsuite/gdb.trace/ftrace-lock.c: In function 'gdb_agent_gdb_collect':
.../src/gdb/testsuite/gdb.trace/ftrace-lock.c:50:3: warning: implicit declaration of function 'sleep' [-Wimplicit-function-declaration]
sleep (1);
^
gdb/testsuite/ChangeLog:
2016-03-01 Pedro Alves <palves@redhat.com>
Pedro Alves [Tue, 1 Mar 2016 17:25:56 +0000 (17:25 +0000)]
Fix gdb.threads/watchpoint-fork*.c compilation
This testcase currently fails to compile on Fedora 23:
.../src/gdb/testsuite/gdb.threads/watchpoint-fork-mt.c: In function 'start':
.../src/gdb/testsuite/gdb.threads/watchpoint-fork-mt.c:70:11: warning: implicit declaration of function 'pthread_yield' [-Wimplicit-function-declaration]
i = pthread_yield ();
^
.../src/gdb/testsuite/gdb.threads/watchpoint-fork-child.c: In function 'forkoff':
.../src/gdb/testsuite/gdb.threads/watchpoint-fork-child.c:114:8: warning: implicit declaration of function 'pthread_yield' [-Wimplicit-function-declaratio
n]
i = pthread_yield ();
^
/tmp/ccUkNIsI.o: In function `start':
.../src/gdb/testsuite/gdb.threads/watchpoint-fork-mt.c:70: undefined reference to `pthread_yield'
(...)
collect2: error: ld returned 1 exit status
"You should define these macros by using ‘#define’ preprocessor
directives at the top of your source code files. These directives must
come before any #include of a system header file."
I instead put it in the header all the .c files of the testcase must
include anyway.
gdb/testsuite/ChangeLog:
2016-03-01 Pedro Alves <palves@redhat.com>
* gdb.threads/watchpoint-fork-child.c: Include "watchpoint-fork.h"
before anything else.
* gdb.threads/watchpoint-fork-mt.c: Likewise. Don't define
_GNU_SOURCE here.
* gdb.threads/watchpoint-fork-st.c: Include "watchpoint-fork.h"
before anything else.
* gdb.threads/watchpoint-fork.h: Define _GNU_SOURCE.
Yao Qi [Tue, 1 Mar 2016 15:15:58 +0000 (15:15 +0000)]
Fix output path for arm-disp-step.exp
This patch fixes the following error,
ERROR: (/scratch/yao/gdb/build-git/arm-linux-gnueabihf/gdb/testsuite/outputs/gdb.arch/arm-disp-step/arm-disp-step) No such file or directory
FAIL: gdb.arch/arm-disp-step.exp: Can't run to main
gdb/testsuite:
2016-03-01 Yao Qi <yao.qi@linaro.org>
* gdb.arch/arm-disp-step.exp: Use standard_testfile and
prepare_for_testing.
Yao Qi [Tue, 1 Mar 2016 15:07:34 +0000 (15:07 +0000)]
Compile gdb.arch/arm-neon.c with "quiet"
When we compile gdb.arch/arm-neon.c with options that don't enable NEON,
there are many error/warnings emitted into gdb.sum, which is annoying.
This patch fixes it by passing quiet to prepare_for_testing.
gdb/testsuite:
2016-03-01 Yao Qi <yao.qi@linaro.org>
* gdb.arch/arm-neon.exp: Pass quiet to prepare_for_testing.
Andreas Arnez [Tue, 1 Mar 2016 11:43:55 +0000 (12:43 +0100)]
S390: Fix output path for s390-tdbregs test case
Since test artifacts are always organized in a directory hierarchy, the
s390-tdbregs test case is not executed correctly any more. This is
because it uses an obsolete way of constructing the executable's path.
This change invokes prepare_for_testing instead.
gdb/testsuite/ChangeLog:
* gdb.arch/s390-tdbregs.exp: Use prepare_for_testing instead of
manually constructing the output path.
Yao Qi [Fri, 19 Feb 2016 15:54:03 +0000 (15:54 +0000)]
aarch64-linux process record: more syscalls
The last patch supports several syscalls in linux-record.c, so now
GDB aarch64-linux backend can return these canonicalized syscall numbers
per aarch64 syscall number.
This patch fixes the following fails,
Process record and replay target doesn't support syscall number 59^M
Process record: failed to record execution log.^M
^M
Program stopped.^M
0x00000020000eab28 in pipe () from /lib/aarch64-linux-gnu/libc.so.6^M
(gdb) FAIL: gdb.reverse/pipe-reverse.exp: continue to breakpoint: marker2
Process record and replay target doesn't support syscall number 59^M
Process record: failed to record execution log.^M
^M
Program stopped.^M
0x00000020000eab28 in pipe () from /lib/aarch64-linux-gnu/libc.so.6^M
(gdb) FAIL: gdb.reverse/readv-reverse.exp: continue to breakpoint: marker2
gdb:
2016-02-29 Yao Qi <yao.qi@linaro.org>
* aarch64-linux-tdep.c (aarch64_canonicalize_syscall): Support
eventfd2, eventfd2, dup3, inotify_init1, fallocate and pipe2.
Return gdb_sys_epoll_create1 instead of gdb_sys_epoll_create
for aarch64_sys_epoll_create1.
bfd/
2016-02-29 Cupertino Miranda <Cupertino.Miranda@synopsys.com>
* elf32-arc.c (arc_elf_final_write_processing): Add condition to
the flag change.
(elf_arc_relocate_section): Fixes and conditions to support PIE.
Assert for code sections dynamic relocs.
Iain Buclaw [Sun, 28 Feb 2016 09:25:55 +0000 (10:25 +0100)]
Don't recursively look for a symbol in all imports of imported modules.
Given two or more modules that import each other's scope, the current symbol
lookup routines would go round in circles looking through each import from
each module, possibly checking the same module twice or more until all possible
paths are marked as "searched".
Given enough modules, this causes an exponential slowdown in time taken to find
symbols that do exist, and infinite recursion when they don't.
gdb/ChangeLog:
* d-namespace.c (d_lookup_symbol_imports): Avoid recursive lookups from
cyclic imports.
gdb/testsuite/ChangeLog:
* gdb.dlang/circular.c: New file.
* gdb.dlang/circular.exp: New file.
Keith Seitz [Fri, 26 Feb 2016 20:22:24 +0000 (12:22 -0800)]
Fix CXX invalid cast from void *.
This is an obvious patch to fix the following build error seen with
--enable-build-with-cxx:
../../src/gdb/rs6000-tdep.c: In function ‘rs6000_frame_cache* rs6000_frame_cache(frame_info*, void**)’:
../../src/gdb/rs6000-tdep.c:3242:15: error: invalid conversion from ‘void*’ to ‘rs6000_frame_cache*’ [-fpermissive]
return (*this_cache);
~^~~~~~~~~~~~
gdb/ChangeLog
* rs6000-tdep.c (rs6000_frame_cache): Explicitly cast return result
to avoid invalid conversion from void *.
For these three relocations, 17 bit signed value should be used, instead of
16 bit. The bitsize field is changed from 16 to 17, this field in aarch64
backend is used for overflow check only.
bfd/
2016-02-26 Renlin Li <renlin.li@arm.com>
* elfnn-aarch64.c (elfNN_aarch64_howto_table): Fix signed overflow
check for MOVW_SABS_G0, MOVW_SABS_G1, MOVW_SABS_G2.
ld/
2016-02-26 Renlin Li <renlin.li@arm.com>
* testsuite/ld-aarch64/aarch64-elf.exp: Run new testcases.
* testsuite/ld-aarch64/emit-relocs-270.d: Update to use new boundary.
* testsuite/ld-aarch64/emit-relocs-271.d: Likewise.
* testsuite/ld-aarch64/emit-relocs-272.d: Likewise.
* testsuite/ld-aarch64/emit-relocs-270-overflow.d: New.
* testsuite/ld-aarch64/emit-relocs-270-overflow.s: New.
* testsuite/ld-aarch64/emit-relocs-271-overflow.d: New.
* testsuite/ld-aarch64/emit-relocs-271-overflow.s: New.
* testsuite/ld-aarch64/emit-relocs-272-overflow.d: New.
* testsuite/ld-aarch64/emit-relocs-272-overflow.s: New.
H.J. Lu [Fri, 26 Feb 2016 17:38:08 +0000 (09:38 -0800)]
Optimize x86 GOT32X/GOTPCRELX relocations
R_386_GOT32X, R_X86_64_GOTPCRELX and R_X86_64_REX_GOTPCRELX relocations
retrieve the symbol address via its GOT slot. If the symbol address is
known at the link-time, we can use it directly by changing instruction
encoding. Indirect branch can only be converted to PC relative direct
branch. MOV can be changed to LEA or encoded differently with signed
address. The subset of binary operations can be encoded only with
signed address.
If undefined weak symbol is resolved to zero link-time, we can use it
as address. Zero addresss can't used with PC relative direct branch
when PIC is true since the current PC is unknown. In 64-bit, 32-bit
relocation for PC relatiave direct branch to zero may also overflow.
If this optimization causes relocation overflow, --no-relax can be used
to work around it.
bfd/
PR ld/19609
* elf32-i386.c (elf_i386_convert_load): Convert to R_386_32 for
load with locally bound symbols if PIC is false or there is no
base register. Optimize branch to 0 if PIC is false.
(elf_i386_relocate_section): Don't generate dynamic relocations
against undefined weak symbols if PIC is false.
* elf64-x86-64.c (elf_x86_64_convert_load): Disable optimization
if we can't estimate relocation overflow with --no-relax.
Convert to R_X86_64_32S/R_X86_64_32 for load with locally bound
symbols if PIC is false. Optimize branch to 0 if PIC is false.
(elf_x86_64_relocate_section): Don't generate dynamic relocations
against undefined weak symbols if PIC is false.
Yao Qi [Fri, 26 Feb 2016 15:00:36 +0000 (15:00 +0000)]
Fix various bugs in arm_record_exreg_ld_st_insn
This patch fixes various bugs in arm_record_exreg_ld_st_insn, and use
gdb.reverse/insn-reverse.c to test more arm instructions.
- Set flag SINGLE_REG correctly. In the arch reference manual,
SING_REG is true when the bit 8 of instruction is zero.
- Record the right D registers for instructions changing S registers.
- Fix the order of length and address in record_buf_mem array.
- Shift the offset by 2 instead of by 24.
This patch also fixes one internal error,
(gdb) PASS: gdb.reverse/finish-precsave.exp: BP at end of main
continue^M
Continuing.^M
../../binutils-gdb/gdb/utils.c:1072: internal-error: virtual memory exhausted.^M
A problem internal to GDB has been detected,FAIL: gdb.reverse/finish-precsave.exp: run to end of main (GDB internal error)
gdb:
2016-02-26 Yao Qi <yao.qi@linaro.org>
* arm-tdep.c (arm_record_exreg_ld_st_insn): Set 'single_reg'
per bit 8. Check bit 20 instead of bit 4 for VMOV
instruction. Record D registers for instructions changing
S registers. Change of the order of length and address
in record_buf_mem array.
Yao Qi [Fri, 26 Feb 2016 15:00:36 +0000 (15:00 +0000)]
Record right reg num of thumb special data instructions
When GDB decodes these thumb special data instructions, such as 'mov sp, r7'
the Rd is got incorrectly. According to the arch reference manual, the Rd
is DN:Rdn, in which DN is bit 7 and Rdn is bits 0 to 2. This patch fixes it.
gdb:
2016-02-26 Yao Qi <yao.qi@linaro.org>
* arm-tdep.c (thumb_record_ld_st_reg_offset): Fix the register
number of Rd.
Yao Qi [Fri, 26 Feb 2016 15:00:36 +0000 (15:00 +0000)]
Rename gdb.reverse/aarch64.{exp,c} to gdb.reverse/insn-reverse.{exp,c}
gdb/testsuite:
2016-02-26 Yao Qi <yao.qi@linaro.org>
* gdb.reverse/aarch64.c: Rename to ...
* gdb.reverse/insn-reverse.c: ... it.
* gdb.reverse/aarch64.exp: Rename to ...
* gdb.reverse/insn-reverse.exp: ... it.
Yao Qi [Fri, 26 Feb 2016 15:00:36 +0000 (15:00 +0000)]
Generalize gdb.reverse/aarch64.exp
I said we can generialize gdb.reverse/aarch64.exp for other
architectures https://sourceware.org/ml/gdb-patches/2015-05/msg00482.html
and here is the patch to change aarch64.exp so that it can be used to
test for other architectures as well.
gdb/testsuite:
2016-02-26 Yao Qi <yao.qi@linaro.org>
* gdb.reverse/aarch64.c: [__aarch64__] Include arm_neon.h.
(testcase_ftype): New.
(testcases): New array.
(n_testcases): New.
(main): Call each element in testcases.
* gdb.reverse/aarch64.exp: Remove is_aarch64_target check.
(read_testcase): New.
Do the tests in a loop.
H.J. Lu [Mon, 22 Feb 2016 17:18:52 +0000 (09:18 -0800)]
Properly implement STT_COMMON
The BFD configure option, --enable-elf-stt-common, can't be to used to
verify STT_COMMON implementation with the normal binutils build. Instead,
this patch removes it from BFD. It adds --elf-stt-common=[no|yes] to ELF
assembler/objcopy and adds -z common/-z nocommon to ld.
A configure option, --enable-elf-stt-common, is added to gas to specify
whether ELF assembler should generate common symbols with the STT_COMMON
type by default.
Since BSF_KEEP_G is never used, it is renamed to BSF_ELF_COMMON for ELF
common symbols.
bfd/
PR ld/19645
* bfd.c (bfd): Change flags to 20 bits.
(BFD_CONVERT_ELF_COMMON): New.
(BFD_USE_ELF_STT_COMMON): Likewise.
(BFD_FLAGS_SAVED): Add BFD_CONVERT_ELF_COMMON and
BFD_USE_ELF_STT_COMMON.
(BFD_FLAGS_FOR_BFD_USE_MASK): Likewise.
* configure.ac: Remove --enable-elf-stt-common.
* elf.c (swap_out_syms): Choose STT_COMMON or STT_OBJECT for
common symbol depending on BFD_CONVERT_ELF_COMMON and
BFD_USE_ELF_STT_COMMON.
* elfcode.h (elf_slurp_symbol_table): Set BSF_ELF_COMMON for
STT_COMMON.
* elflink.c (bfd_elf_link_mark_dynamic_symbol): Also check
STT_COMMON.
(elf_link_convert_common_type): New function.
(elf_link_output_extsym): Choose STT_COMMON or STT_OBJECT for
common symbol depending on BFD_CONVERT_ELF_COMMON and
BFD_USE_ELF_STT_COMMON. Set sym.st_info after sym.st_shndx.
* elfxx-target.h (TARGET_BIG_SYM): Add BFD_CONVERT_ELF_COMMON
and BFD_USE_ELF_STT_COMMON to object_flags.
(TARGET_LITTLE_SYM): Likewise.
* syms.c (BSF_KEEP_G): Renamed to ...
(BSF_ELF_COMMON): This.
* bfd-in2.h: Regenerated.
* config.in: Likewise.
* configure: Likewise.
binutils/
PR ld/19645
* NEWS: Mention --elf-stt-common= for objcopy.
* doc/binutils.texi: Document --elf-stt-common= for objcopy.
* objcopy.c (do_elf_stt_common): New.
(command_line_switch): Add OPTION_ELF_STT_COMMON.
(copy_options): Add --elf-stt-common=.
(copy_usage): Add --elf-stt-common=.
(copy_object): Also check do_elf_stt_common for ELF targets.
(copy_file): Handle do_elf_stt_common.
(copy_main): Handle OPTION_ELF_STT_COMMON.
* readelf.c (apply_relocations): Support STT_COMMON.
* testsuite/binutils-all/common-1.s: New file.
* testsuite/binutils-all/common-1a.d: Likewise.
* testsuite/binutils-all/common-1b.d: Likewise.
* testsuite/binutils-all/common-1c.d: Likewise.
* testsuite/binutils-all/common-1d.d: Likewise.
* testsuite/binutils-all/common-1e.d: Likewise.
* testsuite/binutils-all/common-1f.d: Likewise.
* testsuite/binutils-all/common-2.s: Likewise.
* testsuite/binutils-all/common-2a.d: Likewise.
* testsuite/binutils-all/common-2b.d: Likewise.
* testsuite/binutils-all/common-2c.d: Likewise.
* testsuite/binutils-all/common-2d.d: Likewise.
* testsuite/binutils-all/common-2e.d: Likewise.
* testsuite/binutils-all/common-2f.d: Likewise.
* testsuite/binutils-all/objcopy.exp
(objcopy_test_elf_common_symbols): New proc.
Run objcopy_test_elf_common_symbols for ELF targets
gas/
PR ld/19645
* NEWS: Mention --enable-elf-stt-common and --elf-stt-common=
for ELF assemblers.
* as.c (flag_use_elf_stt_common): New.
(show_usage): Add --elf-stt-common=.
(option_values): Add OPTION_ELF_STT_COMMON.
(std_longopts): Add --elf-stt-common=.
(parse_args): Handle --elf-stt-common=.
* as.h (flag_use_elf_stt_common): New.
* config.in: Regenerated.
* configure: Likewise.
* configure.ac: Add --enable-elf-stt-common and define
DEFAULT_GENERATE_ELF_STT_COMMON.
* gas/write.c (write_object_file): Set BFD_CONVERT_ELF_COMMON
and BFD_USE_ELF_STT_COMMON if flag_use_elf_stt_common is set.
* doc/as.texinfo: Document --elf-stt-common=.
* testsuite/gas/elf/common3.s: New file.
* testsuite/gas/elf/common3a.d: Likewise.
* testsuite/gas/elf/common3b.d: Likewise.
* testsuite/gas/elf/common4.s: Likewise.
* testsuite/gas/elf/common4a.d: Likewise.
* testsuite/gas/elf/common4b.d: Likewise.
* testsuite/gas/i386/dw2-compress-3b.d: Likewise.
* testsuite/gas/i386/dw2-compressed-3b.d: Likewise.
* testsuite/gas/elf/elf.exp: Run common3a, common3b, common4a
and common4b.
* testsuite/gas/i386/dw2-compress-3.d: Renamed to ...
* testsuite/gas/i386/dw2-compress-3a.d: This. Pass
--elf-stt-common=no to as.
* testsuite/gas/i386/dw2-compressed-3.d: Renamed to ...
* testsuite/gas/i386/dw2-compressed-3a.d: This. Pass
--elf-stt-common=no to as.
* testsuite/gas/i386/i386.exp: Run dw2-compress-3a,
dw2-compress-3b, dw2-compressed-3a and dw2-compressed-3b instead
of dw2-compress-3 and dw2-compressed-3.
include/
PR ld/19645
* bfdlink.h (bfd_link_elf_stt_common): New enum.
(bfd_link_info): Add elf_stt_common.
ld/
PR ld/19645
* NEWS: Mention -z common/-z nocommon for ELF targets.
* emultempl/elf32.em (gld${EMULATION_NAME}_handle_option): Handle
-z common and -z nocommon.
* ld.texinfo: Document -z common/-z nocommon.
* lexsup.c (elf_shlib_list_options): Add -z common/-z nocommon.
* testsuite/ld-elf/tls_common.exp: Test --elf-stt-common=no and
--elf-stt-common=yes with assembler.
* testsuite/ld-elfcomm/common-1.s: New file.
* testsuite/ld-elfcomm/common-1a.d: Likewise.
* testsuite/ld-elfcomm/common-1b.d: Likewise.
* testsuite/ld-elfcomm/common-1c.d: Likewise.
* testsuite/ld-elfcomm/common-1d.d: Likewise.
* testsuite/ld-elfcomm/common-1e.d: Likewise.
* testsuite/ld-elfcomm/common-1f.d: Likewise.
* testsuite/ld-elfcomm/common-2.s: Likewise.
* testsuite/ld-elfcomm/common-2a.d: Likewise.
* testsuite/ld-elfcomm/common-2b.d: Likewise.
* testsuite/ld-elfcomm/common-2c.d: Likewise.
* testsuite/ld-elfcomm/common-2d.d: Likewise.
* testsuite/ld-elfcomm/common-2e.d: Likewise.
* testsuite/ld-elfcomm/common-2f.d: Likewise.
* testsuite/ld-elfcomm/common-3a.rd: Likewise.
* testsuite/ld-elfcomm/common-3b.rd: Likewise.
* testsuite/ld-i386/pr19645.d: Likewise.
* testsuite/ld-i386/pr19645.s: Likewise.
* testsuite/ld-x86-64/largecomm-1.s: Likewise.
* testsuite/ld-x86-64/largecomm-1a.d: Likewise.
* testsuite/ld-x86-64/largecomm-1b.d: Likewise.
* testsuite/ld-x86-64/largecomm-1c.d: Likewise.
* testsuite/ld-x86-64/largecomm-1d.d: Likewise.
* testsuite/ld-x86-64/largecomm-1e.d: Likewise.
* testsuite/ld-x86-64/largecomm-1f.d: Likewise.
* testsuite/ld-x86-64/pr19645.d: Likewise.
* testsuite/ld-x86-64/pr19645.s: Likewise.
* testsuite/ld-elfcomm/elfcomm.exp: Test --elf-stt-common=yes
with assembler.
(assembler_generates_commons): Removed.
Run -z common/-z nocommon tests. Run *.d tests.
* testsuite/ld-i386/i386.exp: Run pr19645.
* testsuite/ld-x86-64/x86-64.exp: Likewise.
* testsuite/ld-x86-64/dwarfreloc.exp: Test --elf-stt-common with
assembler. Test STT_COMMON with readelf.
H.J. Lu [Fri, 26 Feb 2016 12:16:15 +0000 (04:16 -0800)]
[x86] Resolve non-PIC undefweak symbols in executable
For i386 and x86-64, non-PIC references to undefined weak symbols are
resolved without dynamic relocation when creating executable. Resolved
undefined weak symbols are removed from the dynamic symbol table in
executable. One exception is on i386, we need resolved undefined weak
symbols in the dynamic symbol table in PIE if input relocatable files
contain branchs without PLT so that we can branch to 0 with dynamic
relocation in text section.
This makes behaviors of dynamic executable and position independent
executable predictable with mixed PIC and non-PIC references to undefined
weak symbols. If all references to undefined weak symbols are PIC,
dynamic relocations against undefined weak symbols will be generated
in executable unless -z nodynamic-undefined-weak is passed to linker.
bfd/
PR ld/19636
PR ld/19704
PR ld/19719
* elf32-i386.c (UNDEFINED_WEAK_RESOLVED_TO_ZERO): New.
(elf_i386_link_hash_entry): Add has_got_reloc and
has_non_got_reloc.
(elf_i386_link_hash_table): Add interp.
(elf_i386_link_hash_newfunc): Initialize has_got_reloc and
has_non_got_reloc.
(elf_i386_copy_indirect_symbol): Copy has_got_reloc and
has_non_got_reloc.
(elf_i386_check_relocs): Set has_got_reloc and has_non_got_reloc.
(elf_i386_fixup_symbol): New function.
(elf_i386_pie_finish_undefweak_symbol): Likewise.
(elf_i386_allocate_dynrelocs): Don't allocate space for dynamic
relocations and discard relocations against resolved undefined
weak symbols in executable. Don't make resolved undefined weak
symbols in executable dynamic. Keep dynamic non-GOT/non-PLT
relocation against undefined weak symbols in PIE.
(elf_i386_size_dynamic_sections): Set interp to .interp section.
(elf_i386_relocate_section): Don't generate dynamic relocations
against resolved undefined weak symbols in PIE, except for
R_386_PC32.
(elf_i386_finish_dynamic_symbol): Keep PLT/GOT entries without
dynamic PLT/GOT relocations for resolved undefined weak symbols.
Don't generate dynamic relocation against resolved undefined weak
symbol in executable.
(elf_i386_finish_dynamic_sections): Call
elf_i386_pie_finish_undefweak_symbol on all symbols in PIE.
(elf_backend_fixup_symbol): New.
* elf64-x86-64.c (UNDEFINED_WEAK_RESOLVED_TO_ZERO): New.
(elf_x86_64_link_hash_entry): Add has_got_reloc and
has_non_got_reloc.
(elf_x86_64_link_hash_table): Add interp.
(elf_x86_64_link_hash_newfunc): Initialize has_got_reloc and
has_non_got_reloc.
(elf_x86_64_copy_indirect_symbol): Copy has_got_reloc and
has_non_got_reloc.
(elf_x86_64_check_relocs): Set has_got_reloc and
has_non_got_reloc.
(elf_x86_64_fixup_symbol): New function.
(elf_x86_64_pie_finish_undefweak_symbol): Likewise.
(elf_x86_64_allocate_dynrelocs): Don't allocate space for dynamic
relocations and discard relocations against resolved undefined
weak symbols in executable. Don't make resolved undefined weak
symbols in executable dynamic.
(elf_x86_64_size_dynamic_sections): Set interp to .interp section.
(elf_x86_64_relocate_section): Check relocation overflow for
dynamic relocations against unresolved weak undefined symbols.
Don't generate dynamic relocations against resolved weak
undefined symbols in PIE.
(elf_x86_64_finish_dynamic_symbol): Keep PLT/GOT entries without
dynamic PLT/GOT relocations for resolved undefined weak symbols.
Don't generate dynamic relocation against resolved undefined weak
symbol in executable.
(elf_x86_64_finish_dynamic_sections): Call
elf_x86_64_pie_finish_undefweak_symbol on all symbols in PIE.
(elf_backend_fixup_symbol): New.
Alan Modra [Fri, 26 Feb 2016 12:30:33 +0000 (23:00 +1030)]
Fix powerpc64 -r --save-restore-funcs
* elf64-ppc.c (create_linkage_sections): Create sfpr when
save_restore_funcs, rest of sections when not relocatable.
(ppc64_elf_init_stub_bfd): Always call create_linkage_sections.
(sfpr_define): Define all symbols on emitted code.
(ppc64_elf_func_desc_adjust): Adjust for sfpr now being created
when relocatable. Move sfpr_define loop earlier.
gdbserver/s390: Enable high GPRs, VX, TDB with 31-bit gdbserver.
Currently, 31-bit gdbserver doesn't support collecting/supplying high
GPRs, VX registers, and TDB data. This is not much of a problem now,
since machines that have them usually have a 64-bit gdbserver that can
be used to debug 31-bit targets just fine. However, with fast
tracepoints, it's not possible to use a 64-bit gdbserver with a 31-bit
IPA (and thus a 31-bit target), so 31-bit gdbserver has to be used
for 31-bit targets. Thus, this patch is needed to allow collecting
high GPRs and VX registers on 31-bit targets via fast tracepoints.
gdb/gdbserver/ChangeLog:
* linux-s390-low.c (s390_num_regs_3264): Define on 31-bit too.
(s390_regmap_3264) [!__s390x__]: New global.
(s390_collect_ptrace_register): Skip map entries containing -1.
(s390_supply_ptrace_register): Ditto.
(s390_fill_gprs_high): New function.
(s390_store_gprs_high): New function.
(s390_regsets): Add NT_S390_HIGH_GPRS.
(s390_get_hwcap): Enable on 31-bit.
(have_hwcap_s390_high_gprs): Enable on 31-bit.
(s390_arch_setup): Enable detection of high GPRs, TDB, VX on 31-bit.
Detect NT_S390_HIGH_GPRS.
(s390_usrregs_info_3264): Enable on 31-bit.
(s390_regs_info): Enable regs_info_3264 on 31-bit.
(initialize_low_arch): Initialize s390_regsets_info_3264 on 31-bit.
Thu Apr 16 10:52:34 1998 John Metzler <jmetzler@cygnus.com>
* gdb.base/branches.c: Code with lots of loops and
subroutines. Used to test gdbs ability to single step through PC
changes, especially to test mips-tdep.c:mips_next_pc
[PR gdb/13808] gdb.trace: Pass tdesc selected in gdbserver to IPA.
If gdbserver and IPA are using different tdesc, they will disagree
about 'R' trace packet size. This results in mangled traces.
To make sure they pick the same tdesc, gdbserver pokes the tdesc
(specified as an index in a target-specific list) into a global
variable in IPA. In theory, IPA could find out the tdesc on its
own, but that may be complex (in particular, I don't know how to
tell whether we have LAST_BREAK on s390 without messing with ptrace),
and we'd have to duplicate the logic.
Tested on i386 and x86_64. On i386, it fixes two FAILs in ftrace.exp.
On x86_64, these failures have been KFAILed - one of them works now,
but the other now fails due to an unrelated reason (ugh).
gdb/gdbserver/ChangeLog:
PR gdb/13808
* Makefile.in: Add i386-*-linux-ipa.o and amd64-*-linux-ipa.o.
* configure.srv: Ditto.
* linux-aarch64-ipa.c (get_ipa_tdesc): New function.
(initialize_low_tracepoint): Remove ipa_tdesc assignment.
* linux-amd64-ipa.c: Add "linux-x86-tdesc.h" include.
(init_registers_amd64_linux): Remove prototype.
(tdesc_amd64_linux): Remove declaration.
(get_ipa_tdesc): New function.
(initialize_low_tracepoint): Remove ipa_tdesc assignment,
initialize remaining tdescs.
* linux-i386-ipa.c: Add "linux-x86-tdesc.h" include.
(init_registers_i386_linux): Remove prototype.
(tdesc_i386_linux): Remove declaration.
(get_ipa_tdesc): New function.
(initialize_low_tracepoint): Remove ipa_tdesc assignment,
initialize remaining tdescs.
* linux-low.c (linux_get_ipa_tdesc_idx): New function.
(linux_target_ops): wire in linux_get_ipa_tdesc_idx.
* linux-low.h (struct linux_target_ops): Add get_ipa_tdesc_idx.
* linux-x86-low.c: Move tdesc declarations to linux-x86-tdesc.h.
(x86_get_ipa_tdesc_idx): New function.
(the_low_target): Wire in x86_get_ipa_tdesc_idx.
* linux-x86-tdesc.h: New file.
* target.h (struct target_ops): Add get_ipa_tdesc_idx.
(target_get_ipa_tdesc_idx): New macro.
* tracepoint.c (ipa_tdesc_idx): New macro.
(struct ipa_sym_addresses): Add addr_ipa_tdesc_idx.
(symbol_list): Add ipa_tdesc_idx.
(cmd_qtstart): Write ipa_tdesc_idx in the target.
(ipa_tdesc): Remove.
(ipa_tdesc_idx): New variable.
(get_context_regcache): Use get_ipa_tdesc.
(gdb_collect): Ditto.
(gdb_probe): Ditto.
* tracepoint.h (get_ipa_tdesc): New prototype.
(ipa_tdesc): Remove.
Simon Marchi [Thu, 25 Feb 2016 16:03:22 +0000 (11:03 -0500)]
Initialize variables in i386_linux_handle_segmentation_fault
We see this error when building with gcc 4.3.
../../gdb/i386-linux-tdep.c: In function ‘i386_linux_handle_segmentation_fault’:
../../gdb/i386-linux-tdep.c:399: error: ‘access’ may be used uninitialized in this function
../../gdb/i386-linux-tdep.c:399: error: ‘upper_bound’ may be used uninitialized in this function
../../gdb/i386-linux-tdep.c:399: error: ‘lower_bound’ may be used uninitialized in this function
It's a false positive, since the variables will always get initialized
in the TRY clause, and the CATCH returns.
gdb.trace: Remove unnecessary target check from ftrace.exp.
The check used hardcoded targets and wasn't doing anything useful anyway,
since unsupported architectures blow up on link due to missing the IPA
library before they ever get to that check.
gdb.trace: Surround $call_insn with \y in entry-values.exp
The PPC64 tracepoint patch added \y at the end of the call_insn pattern -
without that, it embarassed itself and matched the 'bl' in "Dump of
assem*bl*er code for function" as the powerpc call opcode. Since that
sounds like a generally good idea, I've added \y before and after
call_insn for every target. As a result, I had to change x86_64's mnemonic
to 'callq'.
gdb/testsuite/ChangeLog:
* gdb.trace/entry-values.exp: Surround $call_insn with '\y',
change x86_64 call_insn to 'callq'.
Antoine Tremblay [Wed, 24 Feb 2016 20:00:17 +0000 (15:00 -0500)]
Map registers to remote numbers when encoding an ax_reg or ax_reg_mask operation
When encoding the agent expression operation ax_reg or ax_reg_mask, the
register number used is internal to GDB. However GDBServer expects a tdesc
based number.
This usually does not cause a problem since at the moment, for raw
registers GDBServer R trace action ignores the register mask and just
collects all registers.
It can be a problem, however with pseudo registers on some platforms if the
tdesc number doesn't match the GDB internal register number.
This is the case with ARM, the upcoming ARM tracepoint support, fails
these test cases without this patch:
Andrew Burgess [Tue, 23 Feb 2016 00:05:27 +0000 (00:05 +0000)]
gas: Update tests for big-endian arc targets
Some gas tests are excluded for arc targets using the pattern arc-*-*,
however this only covers little endian arc targets. This commit extends
the patter to cover big endian arc targets, the pattern is now arc*-*-*.
The MIPS kernel reports SI_KERNEL for all kernel generated traps,
instead of TRAP_BRKPT / TRAP_HWBKPT, but GDB isn't aware of this.
Basically, this commit:
- Folds watchpoints logic into check_stopped_by_breakpoint, and
renames it to save_stop_reason.
- Adds GDB_ARCH_IS_TRAP_HWBKPT.
- Makes MIPS set both GDB_ARCH_IS_TRAP_BRPT and
GDB_ARCH_IS_TRAP_HWBKPT to SI_KERNEL. In save_stop_reason, we
handle the case of the same si_code returning true for both
TRAP_BRPT and TRAP_HWBKPT by looking at what the debug registers
say.
Tested on x86-64 Fedora 20, native and gdbserver.
gdb/ChangeLog:
2016-02-24 Pedro Alves <palves@redhat.com>
* linux-nat.c (save_sigtrap) Delete.
(stop_wait_callback): Call save_stop_reason instead of
save_sigtrap.
(check_stopped_by_breakpoint): Rename to ...
(save_stop_reason): ... this. Bits of save_sigtrap folded here.
Use GDB_ARCH_IS_TRAP_HWBKPT and handle ambiguous
GDB_ARCH_IS_TRAP_BRKPT / GDB_ARCH_IS_TRAP_HWBKPT. Factor out
common code between the USE_SIGTRAP_SIGINFO and
!USE_SIGTRAP_SIGINFO blocks.
(linux_nat_filter_event): Call save_stop_reason instead of
save_sigtrap.
* nat/linux-ptrace.h: Check for both SI_KERNEL and TRAP_BRKPT
si_code for MIPS.
* nat/linux-ptrace.h: Fix "TRAP_HWBPT" typo in x86 table. Add
comments on MIPS behavior.
(GDB_ARCH_IS_TRAP_HWBKPT): Define for all archs.
gdb/gdbserver/ChangeLog:
2016-02-24 Pedro Alves <palves@redhat.com>
* linux-low.c (check_stopped_by_breakpoint): Rename to ...
(save_stop_reason): ... this. Use GDB_ARCH_IS_TRAP_HWBKPT and
handle ambiguous GDB_ARCH_IS_TRAP_BRKPT / GDB_ARCH_IS_TRAP_HWBKPT.
Factor out common code between the USE_SIGTRAP_SIGINFO and
!USE_SIGTRAP_SIGINFO blocks.
(linux_low_filter_event): Call save_stop_reason instead of
check_stopped_by_breakpoint and check_stopped_by_watchpoint.
Update comments.
(linux_wait_1): Update comments.
H.J. Lu [Wed, 24 Feb 2016 13:42:27 +0000 (05:42 -0800)]
Set plt_got.offset to (bfd_vma) -1
Since plt_got.offset may be updated in x86 allocate_dynrelocs, set
plt_got.offset to (bfd_vma) -1 when setting needs_plt to 0.
* elf32-i386.c (elf_i386_allocate_dynrelocs): Set plt_got.offset
to (bfd_vma) -1 when setting needs_plt to 0.
* elf64-x86-64.c (elf_x86_64_allocate_dynrelocs): Likewise.
Gary Benson [Wed, 24 Feb 2016 11:31:58 +0000 (11:31 +0000)]
Fix logic in exec_file_locate_attach
This commit fixes an error in exec_file_locate_attach where
the main executable could be loaded from outside the sysroot
if a nonempty, non-"target:" sysroot was set but the discovered
executable filename did not exist in that sysroot and did exist
on the main filesystem.
gdb/ChangeLog:
* exec.c (exec_file_locate_attach): Do not attempt to
locate main executable locally if not found in sysroot.
gdb/testsuite/ChangeLog:
* gdb.base/attach-pie-noexec.exp: Do not expect an error
message on attach.
Doug Evans [Tue, 23 Feb 2016 21:25:18 +0000 (13:25 -0800)]
Extend "skip" command to support -file, -gfile, -function, -rfunction.
gdb/ChangeLog:
Extend "skip" command to support -file, -gfile, -function, -rfunction.
* NEWS: Document new features.
* skip.c: #include "fnmatch.h", "gdb_regex.h".
(skiplist_entry) <file>: Renamed from filename.
<function>: Renamed from function_name.
<file_is_glob, function_is_regexp>: New members.
<compiled_function_regexp, compiled_function_regexp_is_valid>:
New members.
(make_skip_entry): New function.
(free_skiplist_entry, free_skiplist_entry_cleanup): New functions.
(make_free_skiplist_entry_cleanup): New function.
(skip_file_command): Update.
(skip_function, skip_function_command): Update.
(compile_skip_regexp): New functions.
(skip_command): Add support for new options.
(skip_info): Update.
(skip_file_p, skip_gfile_p): New functions.
(skip_function_p, skip_rfunction_p): New functions.
(function_name_is_marked_for_skip): Update and simplify.
(_initialize_step_skip): Update.
* symtab.c: #include "fnmatch.h".
(compare_glob_filenames_for_search): New function.
* symtab.h (compare_glob_filenames_for_search): Declare.
* utils.c (count_path_elements): New function.
(strip_leading_path_elements): New function.
* utils.h (count_path_elements): Declare.
(strip_leading_path_elements): Declare.
gdb/doc/ChangeLog:
* gdb.texinfo (Skipping Over Functions and Files): Document new
options to "skip" command. Update docs of output of "info skip".
gdb/testsuite/ChangeLog:
* gdb.base/skip.c (test_skip): New function.
(end_test_skip_file_and_function): New function.
(test_skip_file_and_function): New function.
* gdb.base/skip1.c (test_skip): New function.
(skip1_test_skip_file_and_function): New function.
* gdb.base/skip.exp: Add tests for new skip options.
* gdb.base/skip-solib.exp: Update expected output.
* gdb.perf/skip-command.cc: New file.
* gdb.perf/skip-command.exp: New file.
* gdb.perf/skip-command.py: New file.
H.J. Lu [Tue, 23 Feb 2016 13:18:43 +0000 (05:18 -0800)]
Make linker assigned symbol dynamic only for shared object
Linker assigned symbols should be made dynamic by default only when
creating shared object or relocatable executable.
bfd/
* elflink.c (bfd_elf_record_link_assignment): Check for shared
library, instead of PIC, and don't check PDE when making linker
assigned symbol dynamic.
Yao Qi [Tue, 23 Feb 2016 09:21:09 +0000 (09:21 +0000)]
Process record for aarch64-linux syscall
This patch updates the syscalls in sync with syscalls/aarch64-linux.xml.
Some syscalls are still not supported by gdb/linux-record.c yet. Mark
them UNSUPPORTED_SYSCALL_MAP.
This patch fixes the following test fail,
Process record and replay target doesn't support syscall number 56^M
Process record: failed to record execution log.^M
^M
Program stopped.^M
0x00000020000e9dfc in open () from /lib/aarch64-linux-gnu/libc.so.6^M
(gdb) FAIL: gdb.reverse/fstatat-reverse.exp: continue to breakpoint: marker2