| 161efa6f4e7ea6657918a2f1736a692eea2f56c0 |
08-Apr-2014 |
Christopher Ferris <cferris@google.com> |
Merge "Re-add stpcpy call."
|
| 855b8a20be0cff41dbfbb245b91fe417acf194cf |
08-Apr-2014 |
Mohamad Ayyash <mkayyash@google.com> |
Merge "ltrace: Fix the mips build."
|
| 27e80c93b4eadfc54c2bd2196028ed7ac1cc13f6 |
08-Apr-2014 |
Mohamad Ayyash <mkayyash@google.com> |
ltrace: Fix the mips build.
1) Fix missing error.h in plt.c.
2) Add a TODO to fix missing relplt_count in mips plt.c.
3) Enforce operator precedence between * and != which is apparently incorrect for the mips compiler.
Change-Id: I2d5d925566daf81a0380724bc103570c80e381e7
ips/plt.c
|
| 6bc89fea4e132d3937dd5aafbabca0d12e3af22f |
08-Apr-2014 |
Christopher Ferris <cferris@google.com> |
Re-add stpcpy call.
Change-Id: I1e80c3ee1693452f8f1669027fcd0a225e986eb2
ooks.c
|
| 1031614c095f00454326deb460a83381e98d5166 |
04-Apr-2014 |
Mohamad Ayyash <mkayyash@google.com> |
ltrace: Get it to build on Android (Android.mk and missing functions)
- Add Android.mk.
- Work around missing stpcpy() and rindex().
Change-Id: Iffdd707b22a84b1a1c6a802574ded05b289cee0a
ooks.c
|
| 6bacfc7b34be8414d01c201f1c6086e3edaa8b62 |
12-Mar-2014 |
dann frazier <dannf@debian.org> |
Add missing #include stdio.h
Include stdio.h in files that use fprintf()
Signed-off-by: dann frazier <dannf@debian.org>
arch64/regs.c
arch64/trace.c
|
| 2e8ee3bb400006df046e432bc9dde1c0ce12b831 |
12-Mar-2014 |
dann frazier <dannf@debian.org> |
add missing <stdint.h> include
aarch64's fetch.c uses the uintptr_t typedef defined by stdint.h without
including it. This doesn't currently cause a build failure because stdint.h
is indirectly included via proc.h.
Signed-off-by: dann frazier <dannf@debian.org>
arch64/fetch.c
|
| 912a0f75b3521803fa724a55f0e883c134c7b4e9 |
24-Jan-2014 |
Petr Machata <pmachata@redhat.com> |
Implement aarch64 support
- IFUNC support is not implemented, the rest works well. The only
other failure is in wide char functions, and that occurs on x86_64
as well.
akefile.am
arch64/Makefile.am
arch64/arch.h
arch64/fetch.c
arch64/plt.c
arch64/ptrace.h
arch64/regs.c
arch64/signalent.h
arch64/syscallent.h
arch64/trace.c
|
| a0093ca43cf40d7e5f6cebeb64156062d2de46d9 |
10-Jan-2014 |
Petr Machata <pmachata@redhat.com> |
Don't crash untraced calls via PLT in prelinked PPC64 binaries
In prelinked binaries, ltrace has to unprelinks PLT slots in order to
catch calls done through PLT. This makes the calls done through these
slots invalid, because the special first PLT slot is not initialized,
and dynamic linker SIGSEGVs because of this. Ltrace relies on
arranging breakpoints such that the dynamic linker is not actually
entered, and moves PC around itself to simulate the effects of a call
through PLT.
Originally, arch_elf_add_plt_entry was called only for symbols that
were actually traced. Later this was changed and it's now called for
all PLT entries, and the resulting candidate list is filtered
afterwards. This gives backends a chance to rename the symbol, as is
useful with IRELATIVE PLT calls, where symbol name may not be
available at all. But the PPC backend was never updated to reflect
this, and unresolved all symbols for which arch_elf_add_plt_entry was
called, thus rendering _all_ PLT slots invalid, even those that
weren't later procted by breakpoints. Thus calls done through any
untraced slots failed.
This patch fixes this problem by deferring the unprelinking of PLT
slots into the on_install hook of breakpoints.
pc/arch.h
pc/plt.c
|
| 79e64bcb176442b0eef89764d5ea7f07972dd240 |
10-Jan-2014 |
Petr Machata <pmachata@redhat.com> |
Type umovebytes properly
race.c
|
| b882ba7bd0840db234eb84c27665da4b3f4b42b8 |
09-Jan-2014 |
Mark Wielaard <mjw@redhat.com> |
Initialize libsym early in trace.c to help the compiler.
GCC 4.4.7 isn't smart enough to realize own_libsym will always be zero
when it sees the goto done which might jump over the initialization of
libsym. And so will produce a warning like:
cc1: warnings being treated as errors
trace.c: In function ‘ifunc_ret_hit’:
trace.c:1433: error: ‘libsym’ may be used uninitialized in this function
race.c
|
| 35742523e3daa0e59de0c1c3fdd8e5ff52891967 |
09-Jan-2014 |
Petr Machata <pmachata@redhat.com> |
Fix a problem in tracing across fork on PPC64
In order to avoid single-stepping through large portions of the
dynamic linker, ltrace remembers at which address the instruction that
resolved a PLT slot is. It then puts a breakpoint to this address so
that it can fast-forward to that address next time it needs to catch a
PLT slot being resolved.
When a process is cloned, the pointer to this breakpoint is simply
copied over to the new process, instead of being looked up in the new
process structures. This patches fixes this.
pc/plt.c
|
| f7864d60f0c8180806ed9b7d8dac26b250bd7120 |
27-Nov-2013 |
Petr Machata <pmachata@redhat.com> |
Support IRELATIVE relocations on s390
390/arch.h
390/plt.c
|
| 42cccc91299718b95eecffa6941e29c33d5616b3 |
27-Nov-2013 |
Petr Machata <pmachata@redhat.com> |
On s390{,x}, r2 is scrambled after syscall entry
This was caught by system_call_params.exp test case:
exe->mount("source", "target", "filesystemtype", 0, nil <unfinished ...>
mount@SYS("", "target", "filesystemtype", 0, nil) = -2
<... mount resumed> = -1
Note how the first parameter disappears--r2 now holds syscall number
(21 in this case), and the original value is stored in orig_gpr2 in
save area.
390/fetch.c
|
| 7b0088dbf6861987f13ad1e0740d633069fdb9d7 |
22-Nov-2013 |
Petr Machata <pmachata@redhat.com> |
On x86{,_64}, PLT entries may not be ordered by their relocation
- In general they are. But IRELATIVE relocations are sorted to come
last, and PLT entries are not sorted accordingly.
86/arch.h
86/plt.c
|
| 89396203c9de0d60374e1b298529afcf0bfbf5ed |
06-Nov-2013 |
Petr Machata <pmachata@redhat.com> |
Plug a leak
ooks.c
|
| 9e33f5ac1037adeb32e9d693e6555967e9be68a6 |
05-Nov-2013 |
Petr Machata <pmachata@redhat.com> |
Fix compilation on ARM
- This was broken several commits back by removing pltrel from
struct ltelf.
rm/arch.h
rm/plt.c
|
| 1383e5bd876baa123b39210c2210d256bcfa5bec |
05-Nov-2013 |
Petr Machata <pmachata@redhat.com> |
In ARM backend, move hooks for ltrace_elf and library to plt.c
- That's the customary location for backend hooks.
rm/fetch.c
rm/plt.c
|
| 4f2f66e6abc7fedf3a5d04fab7cc00e5f82b37cf |
05-Nov-2013 |
Petr Machata <pmachata@redhat.com> |
Move load_dynamic_entry from PPC backend to ltrace-elf.c/.h
pc/plt.c
|
| fa844db00b61b9f61c9ae8c6f4165aa5fff3a5d7 |
05-Nov-2013 |
Petr Machata <pmachata@redhat.com> |
Full support of IFUNC tracing on PPC32
pc/arch.h
pc/plt.c
race.c
race.h
|
| fec0b3b1b8323220255f9ebdcac7db2b689783ef |
04-Nov-2013 |
Petr Machata <pmachata@redhat.com> |
Move code for tracing IFUNC symbols
- ... from sysdeps/linux-gnu/hooks.c to sysdeps/linux-gnu/trace.c
where it fits better.
ooks.c
race.c
|
| 54bb64cf2eae7a0daa4d17e980b743b8ae69413b |
04-Nov-2013 |
Petr Machata <pmachata@redhat.com> |
Partial support of IFUNC tracing on PPC32
- Just like on PPC64, we need to move IRELATIVE relocations from
.rela.dyn to the vector of PLT relocations
- Just like on PPC64, we may need to rename the symbol--except the
symbol to looked is the one with the address of PLT slot address,
not of rela addend.
pc/plt.c
race.c
race.h
|
| ba36f0ac51d194c599fd56457796e33e62c3220b |
04-Nov-2013 |
Petr Machata <pmachata@redhat.com> |
Cleanups
vents.c
pc/plt.c
|
| 0ed5399f5cbb2b505a95bcac634c0ad5d7bbd6b8 |
04-Nov-2013 |
Petr Machata <pmachata@redhat.com> |
Properly publish read_relplt interface, rename it to elf_read_relocs
pc/plt.c
|
| 73b85aadbf377541ac336914e5ff8ec521226a97 |
30-Oct-2013 |
Petr Machata <pmachata@redhat.com> |
Support tracing P_PPC64_JMP_IREL slots
pc/arch.h
pc/plt.c
|
| b061bae322edd4894f14ea2aea6baec36d32eda8 |
25-Oct-2013 |
Petr Machata <pmachata@redhat.com> |
Split linux_elf_find_irelative_name out of linux_elf_add_plt_entry_irelative
race.c
race.h
|
| 673ff510953b65b844a58478aa434120f457c014 |
25-Oct-2013 |
Petr Machata <pmachata@redhat.com> |
Keep PLT relocations in a vector
- That means we have to copy them out of ELF ahead of time instead of
referencing them from inside ELF on demand. But this way we can keep
one grand vector of all PLT-like relocations. On PPC, this makes
a difference: some PLT-like relocations (R_PPC64_JMP_IREL in
particular) are stored in .rela.dyn, not .rela.plt.
ips/plt.c
|
| 481a1eb415945720c0e1c545bd7679fd4cdb48ff |
25-Oct-2013 |
Petr Machata <pmachata@redhat.com> |
OPD-transate the address that comes back from IFUNC resolver
ooks.c
|
| 5069ef8f498e5189de0789d79485f39b76c621d4 |
24-Oct-2013 |
Petr Machata <pmachata@redhat.com> |
Fix fetching of system call arguments on i386
86/fetch.c
|
| 88070a85ddaf5d2f4c6e89b5fbb5db5124049b4b |
24-Oct-2013 |
Petr Machata <pmachata@redhat.com> |
sysdeps/linux-gnu/trace.c misses #include <stdint.h>
race.c
|
| b3d6180e6d1cfc188a44b60dda6a7ac4faf0fdce |
16-Oct-2013 |
Petr Machata <pmachata@redhat.com> |
Linux backend now supports tracing of IFUNC symbols
ooks.c
s.h
|
| 1e4b8c8d93606721c8728c6cacb8c9921353049b |
15-Oct-2013 |
Petr Machata <pmachata@redhat.com> |
Enable IRELATIVE tracing on x86 and x86_64
86/arch.h
86/plt.c
|
| b420a226cd2fc5d6028adcaf236c512a1f1fb437 |
15-Oct-2013 |
Petr Machata <pmachata@redhat.com> |
Add support for tracing of IRELATIVE PLT entries
- Because the IRELATIVE entries have no associated symbol name, we
need to allow arch_elf_add_plt_entry to override the name. This is
done by that callback returning PLT_OK and returning the new symbol
via libsym-chain return argument. Filtering is postponed until we
have that symbol, and the filter is applied to the whole returned
chain.
- Add linux_elf_add_plt_entry_irelative to support proper naming of
IRELATIVE PLT entries. This needs to be called from arch backend,
as the numbers of IRELATIVE relocations differ per-architecture.
race.c
race.h
|
| b94407705ecddffd7820ee61bcad3f0d72ee87c1 |
14-Oct-2013 |
Petr Machata <pmachata@redhat.com> |
Split part of delete_breakpoint into delete_breakpoint_at
race.c
|
| aa3db6b1234da0e542ba7782849cf200d0d91c1c |
23-Oct-2013 |
Petr Machata <pmachata@redhat.com> |
Replace os_get_ltrace_conf_filename with os_get_ltrace_conf_filenames
- I.e. allow returning a list of filenames, as opposed to a single
home directory name. This is to implement exactly the legacy config
file behavior on Linux.
ooks.c
|
| 552d75e2a226782cc9ecf966e6e343af8f51031f |
14-Oct-2013 |
Petr Machata <pmachata@redhat.com> |
Sanitize interface of arch_get_sym_info/elf_get_sym_info
- <0 for errors, ==0 for OK, 1 for skip. Errors are now propagated
and can be handled as necessary without cancelling tracing.
- Clarify the comment at arch_get_sym_info.
ips/plt.c
|
| 02a796e5e49c147982020c78b0066930e979f3e4 |
11-Oct-2013 |
Petr Machata <pmachata@redhat.com> |
Rename insert_breakpoint to insert_breakpoint_at
pc/plt.c
roc.c
race.c
|
| 7287166e8fd5949ffcf8eb1f3d378b5ea538915e |
11-Oct-2013 |
Petr Machata <pmachata@redhat.com> |
arch_library_init and arch_library_clone should be able to signal failure
- This then trickles to library_init, which needs to return int
- Callers of library_init were updated to unroll their transactions
upon failure
- PPC and ARM backends were updated to match the updated interface
rm/fetch.c
pc/plt.c
roc.c
|
| c391cea99973d5fac82bf9b78ff72dd428875721 |
26-Sep-2013 |
Peter Wu <lekensteyn@gmail.com> |
Fix missing 'metag' in distribution tarball
akefile.am
|
| 20a4315a72ea4df9fe808a5efe374f12f24e2cd1 |
19-Sep-2013 |
Petr Machata <pmachata@redhat.com> |
Fix a comment typo
ooks.c
|
| cb58fdb5a072ad70087a07f67d0c094ff77ca4b7 |
26-Mar-2013 |
Petr Machata <pmachata@redhat.com> |
s390 set_instruction_pointer: Set highest bit in 31-bit tracer
- The highest bit is used to distinguish between 24-bit and 31-bit
addressing modes. Up until about 3.1, Linux did this automatically,
but stopped doing so with the following commit:
https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=d4e81b35b882d96f059afdb0f98e5b6025973b09
390/regs.c
|
| 09489a36a9f0b7fda7bf9d0806301e8d712001b9 |
26-Mar-2013 |
Petr Machata <pmachata@redhat.com> |
Coding style in s390/regs.c
- This changes uses of void* to arch_addr_t, and adds tags for future
conversion of arch_addr_t to an integral type.
390/regs.c
|
| a193452188e51f233677e3f2607d9a61a789a1df |
22-Mar-2013 |
Markos Chandras <markos.chandras@imgtec.com> |
ltrace: Add support for Imagination Technologies Meta
This patchset adds support for Imagination's Meta architecture.
The Meta Linux kernel port will be included in the Linux Kernel
v3.9. It also uses the generic system call numbers.
Signed-off-by: Markos Chandras <markos.chandras@imgtec.com>
etag/Makefile.am
etag/arch.h
etag/plt.c
etag/ptrace.h
etag/regs.c
etag/signalent.h
etag/syscallent.h
etag/trace.c
|
| f25af517a06521f69fbc53eaf3c245a7a2298c75 |
06-Feb-2013 |
Petr Machata <pmachata@redhat.com> |
ARM cleanups
rm/regs.c
rm/regs.h
rm/trace.c
|
| 8a7a9f59e75c8605ca8206294b7729d0fd70f11f |
06-Feb-2013 |
Petr Machata <pmachata@redhat.com> |
ARM: Stack slot is always aligned to at least 4 bytes
rm/fetch.c
|
| 88a0fe50a3fad351cf28ef3902dbd0dd3540735c |
06-Feb-2013 |
Petr Machata <pmachata@redhat.com> |
VFP shouldn't be used for parameter passing in vararg functions on ARM
rm/arch.h
rm/fetch.c
|
| 2a102987aa3583449e6bf56060bf5c8129ee8cd2 |
06-Feb-2013 |
Petr Machata <pmachata@redhat.com> |
Simplify the code for ARM parameter passing
rm/fetch.c
|
| 1c8596d41dde12f9af6a21035fac0c64f428ab76 |
06-Feb-2013 |
Petr Machata <pmachata@redhat.com> |
Implement VFP parameter passing for ARM backend
rm/fetch.c
|
| 982cbca34b2b49a158086ff5f43eb9bba89edead |
06-Feb-2013 |
Petr Machata <pmachata@redhat.com> |
Move get_hfa_type from IA64 backend to type.c, name it type_get_hfa_type
a64/fetch.c
|
| 62fc7747d37cacfb21381961674c07ebab5f2fb9 |
05-Feb-2013 |
Petr Machata <pmachata@redhat.com> |
Implement Base Procedure Call Standard for ARM
arch_fetch_arg_next still doesn't support "hardfp" extensions, but much of
the test suite is now clean, with only 16 failures left.
rm/fetch.c
|
| 97a25160b0fe646d9c567e12c5abefe8e59873c2 |
05-Feb-2013 |
Petr Machata <pmachata@redhat.com> |
Add ARM fetch backend
This only properly decodes float and double return values. More to come.
rm/fetch.c
rm/trace.c
|
| df7d23111899c1e1aa16bad83e0db0a8334b11d9 |
05-Feb-2013 |
Petr Machata <pmachata@redhat.com> |
Implement decoding ARM attribute section
This to determine whether hardfp is used in the process.
rm/Makefile.am
rm/arch.h
rm/fetch.c
|
| 427a43ee9972763993a1241815d56921f30623ef |
05-Feb-2013 |
Petr Machata <pmachata@redhat.com> |
Add arch_type_sizeof, arch_type_alignof to ARM backend
rm/arch.h
rm/trace.c
|
| 99b19eb1bb736d7066026894aa69e70de8a03094 |
05-Feb-2013 |
Petr Machata <pmachata@redhat.com> |
elf_get_section_* should return 0 if section wasn't found
Otherwise it won't be possible to differentiate between an error and
not-found condition. It seems some callers (MIPS, PPC backends) already
count on that behavior anyway.
pc/plt.c
|
| f9d93c50bd246ea7fd42e0c8ad24aa01467e76ac |
30-Jan-2013 |
Petr Machata <pmachata@redhat.com> |
Fix tracing Thumb mode on ARM
- Drop what amounts to global variable thumb_mode from struct process
- Thumb mode is encoded to address. That means that the code in
callstack_push_symfunc and insert_breakpoint has to assume that
what comes from get_return_address is mangled, and needs to go
through creating the breakpoint and then looking at its address.
That's awkward, and we should come up with a better API.
rm/breakpoint.c
rm/regs.c
|
| 44ae188c8dcd2727e383dcac0d46c3f784d0da5f |
30-Jan-2013 |
Petr Machata <pmachata@redhat.com> |
Drop set_return_addr
- This is currently not used anywhere, so it is safe to assume that it is
broken anyway.
lpha/regs.c
rm/regs.c
a64/regs.c
68k/regs.c
ips/regs.c
pc/regs.c
390/regs.c
parc/regs.c
86/regs.c
|
| 7fc98ee3538bdcc898b785c26d275b0661bfdc8b |
29-Jan-2013 |
Petr Machata <pmachata@redhat.com> |
Instead of smuggling the vfork return breakpoint back, ref it once more
- Before we would let the breakpoint be removed and later reinserted it.
But that's unnecessarily cumbersome, and besides breaks on ARM Thumb,
where the reinserted breakpoint would be ARM breakpoint instead of
Thumb. This is easier and fixes that issue.
race.c
|
| 037f37c72e252b05c59e53826d5007d24d54ac74 |
23-Jan-2013 |
Petr Machata <pmachata@redhat.com> |
More ARM Thumb support
Still not all of the GDB code ported. Some threading test cases still
fail.
rm/regs.h
rm/trace.c
|
| d4a5447503b7bddfb6d7667747ee1eedd20ae491 |
21-Jan-2013 |
Petr Machata <pmachata@redhat.com> |
Early support for Thumb
This ignores 32-bit Thumb-2 instructions, and is very limited in what
it supports in Thumb, but it's a start.
rm/regs.h
rm/trace.c
|
| 6e5e2de1728215679e83df6943c34573c1384e32 |
21-Jan-2013 |
Petr Machata <pmachata@redhat.com> |
GNU/Linux software singlestepping: store breakpoints, not addresses
The reason being that the actual breakpoint address might be different
from the address requested. One example is Thumb, where Thumb
breakpoints are requested for address+1 (lowest bit set). Whether
this is clean solution is very questionable though.
race.c
race.h
|
| 7aa7b404588184ef9a45eeafbfc5ef4afc7fbf82 |
21-Jan-2013 |
Petr Machata <pmachata@redhat.com> |
More ARM software singlestepping support
This should be about done, sans bugfixes. Thumb to come next.
rm/trace.c
|
| 6b9664af2e87596055eaab59492e79bfd1527dab |
21-Jan-2013 |
Petr Machata <pmachata@redhat.com> |
ptrace takes void* arguments
This might be relevant on Android.
rm/regs.c
|
| 9fa2b1850e664aa70a9559e41befca1a5975f08c |
18-Jan-2013 |
Petr Machata <pmachata@redhat.com> |
Port more of the GDB code for ARM singlestepping
- Not even whole ARM is supported yet, and what is there isn't tested.
Thumb is not supported at all as of now.
rm/Makefile.am
rm/regs.c
rm/regs.h
rm/trace.c
|
| 19e74b5815987bb4260351bc1e5ef4a1ea08b1b4 |
17-Jan-2013 |
Petr Machata <pmachata@redhat.com> |
Initial support for software singlestepping on ARM
- This only knows about branch instructions, it doesn't handle long
branches.
- This also doesn't handle thumb mode at all.
rm/arch.h
rm/trace.c
|
| 273890e09616b8fe0b9b015ea8edda12c52d4ba2 |
08-Mar-2013 |
Petr Machata <pmachata@redhat.com> |
Port MIPS backend to the new dict code
ips/trace.c
|
| 84c8c722654548ed1cef5ba183a2198aad089836 |
09-Jan-2013 |
Petr Machata <pmachata@redhat.com> |
GNU/Linux x86 backend: allocate_stack_slot needs non-NULL valuep
86/fetch.c
|
| 52173f05eada6fab209e4f890b507ebaec32c163 |
09-Jan-2013 |
Petr Machata <pmachata@redhat.com> |
GNU/Linux x86 backend: don't pass valuep to functions that don't need it
86/fetch.c
|
| 6bf1e6abb46255456d307bb73432d8733978e5fd |
09-Jan-2013 |
Petr Machata <pmachata@redhat.com> |
GNU/Linux: close opendir stream in error handling branches of process_tasks
roc.c
|
| 754ce881453ee3b6c6a93bbdaef7c645f43d49f6 |
08-Jan-2013 |
Petr Machata <pmachata@redhat.com> |
GNU/Linux backend: guard writing through task_info pointer against NULL
race.c
|
| de98cb698d69ccbbb4388a6401a27897ba02371a |
08-Jan-2013 |
Petr Machata <pmachata@redhat.com> |
Convert one void* to arch_addr_t in x86 backend
86/fetch.c
|
| 33f0ca5350ccc9ded3a2bfa82a988e9bd640233f |
07-Jan-2013 |
Petr Machata <pmachata@redhat.com> |
Streamline some os_get_config_dirs for GNU/Linux
- Extract strlen($HOME) computation out of if branches
- Replace stpcpy+strcpy with a clearer call to sprintf
ooks.c
|
| aacb95e73bd532bb11fbe4005fe28286ecb2d38d |
07-Jan-2013 |
Petr Machata <pmachata@redhat.com> |
Improve error message in process_status backend call for GNU/Linux
roc.c
|
| 4e553c020b0e621bc15d76296990d06e47cf29f9 |
07-Jan-2013 |
Petr Machata <pmachata@redhat.com> |
Tolerate the failure to add _r_debug breakpoint
roc.c
|
| 364753a1cfe46998946a42badd9099591a00325a |
06-Dec-2012 |
Petr Machata <pmachata@redhat.com> |
New backend hook os_get_ltrace_conf_filename and Linux implementation
ooks.c
|
| 414989183b268337a4f7e329237f3c1fd726d823 |
04-Dec-2012 |
Petr Machata <pmachata@redhat.com> |
Implement os_get_config_dirs for gnu-linux backend
akefile.am
ooks.c
|
| 98ff309cdc98857eb30992f108439cb7d7673598 |
08-Mar-2013 |
Petr Machata <pmachata@redhat.com> |
Change DICT_FIND to DICT_FIND_{REF,VAL}
- In many places, DICT_FIND was awkward to use, as it required the use
of pointers even in cases that the dictionary already stored
pointers. For those cases, add DICT_FIND_VAL.
- Rename the original DICT_FIND to DICT_FIND_REF to make the
disctinction clear. That is useful if the dictionary holds actual
structures.
race.c
|
| d7e4ca82e1cf20bb2605befb1da74dd1688c706e |
28-Nov-2012 |
Petr Machata <pmachata@redhat.com> |
Rewrite dict module to be more in line with vect
It's now a bit more strongly typed, can grow and shrink on demand, and has
a more complete interface.
It uses open addressing scheme to store hashes.
ips/trace.c
race.c
|
| a9d6a6732c5b0b692a40ac073695332078a73738 |
05-Mar-2013 |
Petr Machata <pmachata@redhat.com> |
Remove hacks for SPARC code
Simply continuing on SPARC is probably broken. The breakpoint isn't
disabled at that point, and continuing just hits it again.
That __ia64___ in ifdef is never defined even on an IA64 machine.
It's a typo, note the three underscores.
race.c
|
| bc0de433c25d224cf5f4743ba4a78131267c532d |
16-Feb-2013 |
Oliver Spornitz <ospornit@rz-online.de> |
mips: Move mipsel to mips (3/3)
- move files from sysdeps/linux-gnu/mipsel to sysdeps/linux-gnu/mips
- move mipsel.h to mips.h
ips
ips/Doxyfile
ips/Makefile.am
ips/arch.h
ips/mips.h
ips/plt.c
ips/ptrace.h
ips/regs.c
ips/signalent.h
ips/syscallent.h
ips/trace.c
ipsel/Doxyfile
ipsel/Makefile.am
ipsel/arch.h
ipsel/mipsel.h
ipsel/plt.c
ipsel/ptrace.h
ipsel/regs.c
ipsel/signalent.h
ipsel/syscallent.h
ipsel/trace.c
|
| eb7e05a8d48939a5feb84ae6befac8d66fbd8bed |
16-Feb-2013 |
Oliver Spornitz <ospornit@rz-online.de> |
mips: Move mipsel to mips (2/3)
- trailing whitespace cleanup in files to move
ipsel/Doxyfile
ipsel/mipsel.h
|
| 32ea1b1eaad1a07a120811bca9dd9f3524cc9256 |
16-Feb-2013 |
Oliver Spornitz <ospornit@rz-online.de> |
mips: Move mipsel to mips (1/3)
- move references from mipsel to mips
- move references from mipsel.h to mips.h
akefile.am
ipsel/Doxyfile
ipsel/Makefile.am
ipsel/mipsel.h
ipsel/plt.c
ipsel/regs.c
ipsel/trace.c
|
| 4c81bf1e99bfc0b148f7a50b4aeeb7d43b878b2f |
04-Feb-2013 |
Oliver Spornitz <ospornit@rz-online.de> |
Fix SIGSEGV on MIPS big endian
ipsel/arch.h
|
| 057caa59fc0063bd73568b0ae19bbf668b572737 |
30-Jan-2013 |
Petr Machata <pmachata@redhat.com> |
Introduce continue_after_exec
This needs to be handled differently on non-Linux platforms.
race.c
|
| 653085a32cdc89f5215c2d70249b58c9fe6aebb7 |
15-Jan-2013 |
Petr Machata <pmachata@redhat.com> |
Move byte-reading routines from PPC backend to generic code
- Use backend hook umovebytes to read the bytes from inferior.
pc/plt.c
pc/trace.c
|
| 643c6460b64e37af58d8259d86689d414c6b4052 |
15-Jan-2013 |
Petr Machata <pmachata@redhat.com> |
Update call to arch_sw_singlestep
- After the refactoring, the prototype changed and it doesn't
return int anymore. Of course C happily implicitly converts
this. Ho hum.
race.c
|
| 693dfad9c1b121cf079a3082866daa2225df1797 |
14-Jan-2013 |
Petr Machata <pmachata@redhat.com> |
Rename atomic singlestep to software singlestep
- The old name came from PPC implementation, where the feature was used for
skipping atomic instructions. But this is more useful, was reused for
fully software singlestepping on MIPS, and will be similarly used for ARM.
So rename the interface, and make the prototype more explicit.
- Also document the method in backend.h.
- And convert several void*'s to arch_addr_t's while there.
ipsel/arch.h
ipsel/trace.c
pc/arch.h
pc/trace.c
race.c
race.h
|
| efd12cfb10ccd2c612838c0e22069554ce60637c |
02-Jan-2013 |
Petr Machata <pmachata@redhat.com> |
In Linux backend, initialize linkmap struct
GCC 4.7.2 -m32 -Os gives the following (spurious) warning:
proc.c: In function 'crawl_linkmap':
proc.c:544:25: error: 'rlm.l_addr' may be used uninitialized in this
roc.c
|
| 71f25e25a4081b726e75026727483b6959dd8661 |
17-Dec-2012 |
Petr Machata <pmachata@redhat.com> |
Make enum ecb_status enumerators uppercase
vents.c
vents.h
race.c
|
| 96cb8e3111924780ed1d757e865a83f26421e616 |
17-Dec-2012 |
Petr Machata <pmachata@redhat.com> |
Make enumerators of process stopping status uppercase
race.c
race.h
|
| ade3b9798fbc62becbe1b4854f7a2d106498167a |
17-Dec-2012 |
Petr Machata <pmachata@redhat.com> |
Make enum plt_status enumerators uppercase
ipsel/plt.c
pc/plt.c
|
| 6dfc544b6aa6703d2292be34470aebf874d78452 |
17-Dec-2012 |
Petr Machata <pmachata@redhat.com> |
Make enum process_status enumerators uppercase
Rationale: coding style consistency.
vents.c
roc.c
race.c
|
| 929bd57ca202fd2f2e8485ebf65d683e664f67b5 |
17-Dec-2012 |
Petr Machata <pmachata@redhat.com> |
struct Process becomes struct process
This is for consistency with other structures, and ultimately with
Linux coding style. The typedef ("Process") was dropped as well for
this reason.
This opportunity was used to fix coding style around the impacted
lines.
lpha/plt.c
lpha/regs.c
lpha/trace.c
rm/breakpoint.c
rm/plt.c
rm/regs.c
rm/trace.c
reakpoint.c
ris/plt.c
ris/regs.c
ris/trace.c
vents.c
vents.h
a64/fetch.c
a64/plt.c
a64/regs.c
a64/trace.c
68k/fetch.c
68k/plt.c
68k/regs.c
68k/trace.c
ipsel/plt.c
ipsel/regs.c
ipsel/trace.c
pc/fetch.c
pc/plt.c
pc/regs.c
pc/trace.c
roc.c
390/fetch.c
390/plt.c
390/regs.c
390/trace.c
parc/plt.c
parc/regs.c
parc/trace.c
race.c
race.h
86/fetch.c
86/plt.c
86/regs.c
86/trace.c
|
| c1990eb670b2786ffabe712fdb9bf3512dd1501c |
03-Dec-2012 |
Gustavo Zacarias <gustavo@zacarias.com.ar> |
Fix build failure on ppc
ppc/trace.c is using waitstatus bits without including the appropiate
headers, leading to a build failure:
libtool: link:
/home/gustavoz/b/test/output/host/usr/bin/powerpc-buildroot-linux-uclibc-gcc
-Wall -Wsign-compare -Wfloat-equal -Wformat-security -pipe -Os -o ltrace
main.o ./.libs/libltrace.a -lelf
./.libs/libltrace.a(lt1-trace.o): In function `syscall_p':
trace.c:(.text+0x28): undefined reference to `WIFSTOPPED'
trace.c:(.text+0x40): undefined reference to `WSTOPSIG'
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
pc/trace.c
|
| b2cfac8ace230d3cfafd7f3c15d5610bc2e6e044 |
30-Nov-2012 |
Petr Machata <pmachata@redhat.com> |
Linux VDSO can be also named linux-vdso{32,64}.so.1
roc.c
|
| 15595a3158e0efa20c6f357be526d54aa01393b7 |
12-Nov-2012 |
Petr Machata <pmachata@redhat.com> |
Better detection of VDSO in linkmap
roc.c
|
| ccbdb91fe405910f630ee36781f7617a8b99a76b |
09-Nov-2012 |
Andreas Schwab <schwab@linux-m68k.org> |
Update noinst_HEADERS
pc/Makefile.am
|
| 4754ea02d067a88e87d4cc67500419f20781de39 |
08-Nov-2012 |
Petr Machata <pmachata@redhat.com> |
Fix handling error cases in process_get_entry
roc.c
|
| eb584971e63ce7a866c376d87ae8cb9088dd69be |
08-Nov-2012 |
Petr Machata <pmachata@redhat.com> |
Initialize a variable in process_get_entry
GCC 4.7.2-RC on MIPS warns that 'entry.a_un.a_val' may be used
uninitialized. Personally I think this is invalid and the value
is properly initialized, but there's no harm initializing it
explicitly.
roc.c
|
| 1584cfcf6366645087d3d712d0bc5d19356bbb65 |
08-Nov-2012 |
Petr Machata <pmachata@redhat.com> |
Use arch_translate_address_dyn for ia64 backend
We don't generally need to translate symbol addresses on ia64, but we do
need to convert the breakpoint address in _r_debug structure from OPD to
entry point.
a64/arch.h
a64/plt.c
|
| 85465a6248ad4f97f133803a19726eb9049f8394 |
08-Nov-2012 |
Petr Machata <pmachata@redhat.com> |
Coding style nits in ia64 backend
a64/plt.c
|
| af766ab1e37231fe1d7dc5e4b598b9c92f78ecd4 |
07-Nov-2012 |
Petr Machata <pmachata@redhat.com> |
Move default PTRACE_ constants into new file trace-defs.h
Include it also in events.c, where we use them as well.
akefile.am
vents.c
race-defs.h
race.c
|
| 3a9a0ceefd46feb7d88b89be5cce5c4b190b3119 |
07-Nov-2012 |
Petr Machata <pmachata@redhat.com> |
Include cris among subdirs in sysdeps/linux-gnu/Makefile.am
akefile.am
|
| 03f4b61154d2bfb87530bdc79f548c6374ba0db7 |
07-Nov-2012 |
Petr Machata <pmachata@redhat.com> |
Fix includes in PPC fetch backend
pc/fetch.c
|
| 7e26bd32a1a2269c193a6ef100ed8cd3127f3dc2 |
07-Nov-2012 |
Petr Machata <pmachata@redhat.com> |
Fix detection of resolved-ness for PPC32 BSS PLT slots
This fixes the remaining test suite failures on PPC.
pc/arch.h
pc/plt.c
|
| b5fd53993b71d0301b3547287f1a978679b21be2 |
07-Nov-2012 |
Petr Machata <pmachata@redhat.com> |
Coding style in PPC backend
Don't use error. The rest of ltrace should use fprintf for portability,
use it here as well (even though portability is not an issue in
Linux-specific backend).
pc/plt.c
|
| cf2f9227459448daf136e9f79c2181ac85d067be |
07-Nov-2012 |
Petr Machata <pmachata@redhat.com> |
In PPC's PLT handling, inline two trivial functions with one caller only
pc/plt.c
|
| 13e68683414450f3bf7256e1cf09c0220325399f |
07-Nov-2012 |
Petr Machata <pmachata@redhat.com> |
Move one function up in the PPC's PLT handling
This will be useful later in arch_dynlink_done.
pc/plt.c
|
| 8ddf19a62a4521561ae4e4f118f0c364774ee2db |
07-Nov-2012 |
Petr Machata <pmachata@redhat.com> |
Convert enablement of PPC breakpoints from callbacks to iteration
The original code was not terribly hard to understand, but this resembles
a straightforward iteration (which it is) much more. Too bad it's not
possible to wrap this nicely into a for loop.
pc/plt.c
|
| 305945b2523ea87907952eefd7300cfb9f1a0235 |
07-Nov-2012 |
Petr Machata <pmachata@redhat.com> |
Move PPC insn masks and branch address computation to a separate file
pc/insn.h
pc/trace.c
|
| b8deb4d589d57c49d510b1b463b3904cb4964557 |
31-Oct-2012 |
Petr Machata <pmachata@redhat.com> |
Use symbol delaying to implement PPC32 BSS handling
- this fixes tracing of -l symbols and fixes most of the test suite
- attach-process.exp broken for PPC32 BSS tracee
- trace-clone.exp broken for PPC32 ltrace and PPC32 BSS tracee
- PPC32 secure and PPC64 test suites come out clean
pc/plt.c
|
| c06f23d2803c7744f3d2cd1fb3e54886d9e41ddf |
30-Oct-2012 |
Petr Machata <pmachata@redhat.com> |
Fix three calls to debug()
debug() lines shouldn't include \n.
a64/plt.c
roc.c
86/trace.c
|
| 0ffa9f3f82febbcd475318ac282c456adb76c73c |
30-Oct-2012 |
Petr Machata <pmachata@redhat.com> |
Call value_in_inferior instead of setting WHERE manually
On x86_64, this plugs a leak.
68k/fetch.c
pc/fetch.c
390/fetch.c
86/fetch.c
|
| 6c2e57ab25263946a09f42aa84f82678bdbfdc4e |
27-Oct-2012 |
Petr Machata <pmachata@redhat.com> |
Coding style in sysdeps/linux-gnu/trace.c
This is all now code coming from the threads branch. Hopefully no patches
against this are in the wild.
race.c
|
| 47d70f6efd145113ad555dd2ba542d1050bbb784 |
26-Oct-2012 |
Petr Machata <pmachata@redhat.com> |
Use os_process_data for keeping debug status
This also takes care of a leak caused by malloc'ing struct debug. That
malloc is not done anymore, because the data is embedded in struct Process
by way of struct os_process_data.
s.h
roc.c
|
| 0f6e6d9861ce790cb8b7d07aca28a5878d705359 |
26-Oct-2012 |
Petr Machata <pmachata@redhat.com> |
Add os_process_data and related
This is meant to be the same for per-OS configuration as arch_process_data
is for per-arch configuration.
akefile.am
s.h
|
| e99af270a60891e68d465c4cd97dbe29cd1a05e4 |
26-Oct-2012 |
Petr Machata <pmachata@redhat.com> |
Update copyright information and license headers
akefile.am
lpha/Makefile.am
lpha/plt.c
lpha/ptrace.h
lpha/regs.c
lpha/signalent.h
lpha/syscallent.h
lpha/trace.c
rch_mksyscallent
rm/Makefile.am
rm/arch_syscallent.h
rm/plt.c
rm/ptrace.h
rm/regs.c
rm/signalent.h
rm/syscallent.h
rm/trace.c
ris/Makefile.am
ris/plt.c
ris/ptrace.h
ris/regs.c
ris/signalent.h
ris/syscallent.h
ris/trace.c
a64/Makefile.am
a64/breakpoint.c
a64/plt.c
a64/ptrace.h
a64/regs.c
a64/signalent.h
a64/syscallent.h
68k/Makefile.am
68k/plt.c
68k/ptrace.h
68k/regs.c
68k/signalent.h
68k/syscallent.h
68k/trace.c
ipsel/Doxyfile
ipsel/Makefile.am
ipsel/mipsel.h
ipsel/plt.c
ipsel/ptrace.h
ipsel/regs.c
ipsel/signalent.h
ipsel/syscallent.h
ipsel/trace.c
ksignalent
ksyscallent
ksyscallent_mips
ksyscallent_s390
pc/Makefile.am
pc/plt.c
pc/ptrace.h
pc/regs.c
pc/signalent.h
pc/syscallent.h
390/Makefile.am
390/plt.c
390/ptrace.h
390/regs.c
390/signalent.h
390/signalent1.h
390/syscallent.h
390/syscallent1.h
390/syscalls31.h
390/syscalls64.h
parc/Makefile.am
parc/plt.c
parc/ptrace.h
parc/regs.c
parc/signalent.h
parc/syscallent.h
parc/trace.c
86/Makefile.am
86/plt.c
86/ptrace.h
86/signalent.h
86/signalent1.h
86/syscallent.h
86/syscallent1.h
|
| b8fd68d15b1f631e38148ac5dab39d7e857fbc92 |
18-Oct-2012 |
Petr Machata <pmachata@redhat.com> |
Improve a debug message at linkmap_init
roc.c
|
| 0475ac36f22c9531cd0e0ba7d02cee5a1c5c67ea |
18-Oct-2012 |
Petr Machata <pmachata@redhat.com> |
Implement process_get_entry for linux-gnu
roc.c
|
| 5a19e67c9f66c47260310223aeae04d2a5191eac |
09-Oct-2012 |
Edgar E. Iglesias <edgar@axis.com> |
mipsel: Add support for CPIC main programs
Add support for CPIC (Calls PIC) only binaries. These ELF
files are not PIC themselves (as ordinary o32 binaries).
External calls via function pointers don't work yet.
Signed-off-by: Edgar E. Iglesias <edgar@axis.com>
ipsel/arch.h
ipsel/plt.c
|
| f14df2387a0312533b1a1bb3718e9a0d7339fb53 |
09-Oct-2012 |
Edgar E. Iglesias <edgar@axis.com> |
mipsel: Add a MIPS specific find_dl_debug
MIPS dynamic loaders put the address of their debug areas
into an DT_MIPS_RLD_MAP entry instead of into the DT_DEBUG
entry.
Signed-off-by: Edgar E. Iglesias <edgar@axis.com>
ipsel/arch.h
ipsel/plt.c
|
| 11c3c3e28461886bb16c69c745b1fefa946c4d9f |
09-Oct-2012 |
Edgar E. Iglesias <edgar@axis.com> |
Make finding the dynamic linkers debug area arch specfic
Make it possible for archs to override the lookup of the
dynamic linkers debug area.
Signed-off-by: Edgar E. Iglesias <edgar@axis.com>
roc.c
|
| fd4b4ef0d4fbaf2039fac61764421e47cac2ca39 |
09-Oct-2012 |
Edgar E. Iglesias <edgar@axis.com> |
Allow extension tags in proc_find_dynamic_entry_addr
Remove the check for generic tags allowing the caller
to pass extension tags. I couldn't find a way to strictly
validate the tag nr, furthermore the seek tag is internally
generated so I simply dropped the check (beyond checking for
positive values).
One way possible could be to check for:
(tag >= DT_NULL && tag < DT_NUM) || (tag >= DT_LOOS && tag <= DT_HIPROC)
But I'm not sure thats OK.
Signed-off-by: Edgar E. Iglesias <edgar@axis.com>
roc.c
|
| cc77b0e0e735d598f4bc746f8e79b6cd342bd606 |
09-Oct-2012 |
Edgar E. Iglesias <edgar@axis.com> |
Export find_dynamic_entry_addr
Export find_dynamic_entry_addr as proc_find_dynamic_entry_addr.
Signed-off-by: Edgar E. Iglesias <edgar@axis.com>
roc.c
|
| 5b5c256d0e9076e745908a43acc46353265a5608 |
04-Oct-2012 |
Edgar E. Iglesias <edgar@axis.com> |
Abort on unsupported value types in the backends
Makes it easier to add new types.
Signed-off-by: Edgar E. Iglesias <edgar@axis.com>
a64/fetch.c
68k/fetch.c
pc/fetch.c
pc/trace.c
390/fetch.c
390/trace.c
86/fetch.c
86/trace.c
|
| f97b187598192f4c1f0ea47369f4e1dd4c08c392 |
01-Oct-2012 |
Edgar E. Iglesias <edgar.iglesias@gmail.com> |
mipsel: Replace LS_TOPLT_GOTONLY with delayed syms
Signed-off-by: Edgar E. Iglesias <edgar@axis.com>
ipsel/arch.h
ipsel/plt.c
|
| 0b02a93aff5808b7640299faedfeb697485fd170 |
01-Oct-2012 |
Edgar E. Iglesias <edgar@axis.com> |
mipsel: Avoid space before parenthesis in func calls
Signed-off-by: Edgar E. Iglesias <edgar@axis.com>
ipsel/trace.c
|
| a35cc7ad54c7ba11d113f729d40857f59bf01b05 |
01-Oct-2012 |
Edgar E. Iglesias <edgar@axis.com> |
mipsel: Always add/remove breakpoints to/from the leader
Signed-off-by: Edgar E. Iglesias <edgar@axis.com>
ipsel/plt.c
ipsel/trace.c
|
| c44ef1ea84927851ddd74abff9a6374a62978cae |
01-Oct-2012 |
Edgar E. Iglesias <edgar@axis.com> |
mipsel: Improve singlestepping over branch always
When singlestepping over BEQ issuing a compare between
a single register (i.e a branch always), only follow the
branch path.
This catches the case for functions that end with loops
and avoids us putting breakpoints at the begining of the
next function (where there likely already is a bp).
Signed-off-by: Edgar E. Iglesias <edgar@axis.com>
ipsel/trace.c
|
| 3a1806db0ecaa4364e51b4822d66d05f4acd87a9 |
27-Sep-2012 |
Edgar E. Iglesias <edgar@axis.com> |
mipsel: Update breakpoints when functions return
When functions return we check if the symbol went from
unresolved to resolved and if the resolved address
differs from the unresolved one. If so, we add a new
breakpoint at the resolved address.
This is not thread-safe.
Signed-off-by: Edgar E. Iglesias <edgar@axis.com>
ipsel/arch.h
ipsel/plt.c
|
| addae05da54ac16c55133e1c384d74e03346f634 |
27-Sep-2012 |
Edgar E. Iglesias <edgar@axis.com> |
mipsel: Singlestep over breakpoints
Signed-off-by: Edgar E. Iglesias <edgar@axis.com>
ipsel/arch.h
ipsel/trace.c
race.c
|
| 6ef7b2597cd3f4edf851a7b3516aef7449e554ad |
27-Sep-2012 |
Edgar E. Iglesias <edgar@axis.com> |
mipsel: Add mips specific symbol info loading
MIPS needs a backend specific way to load symbol info.
We add fields to the symbol representation to keep track
of the state of the dynamic symbol.
At arch_dynlink_done we go through the symbols that are
connected to a GOT entry but that where not resolved at
startup time (e.g function pointers to external syms).
Signed-off-by: Edgar E. Iglesias <edgar@axis.com>
ipsel/arch.h
ipsel/plt.c
|
| 9f274ffc8d38bd73c31c2b181f25349b813cb9dd |
27-Sep-2012 |
Edgar E. Iglesias <edgar@axis.com> |
mipsel: Conditionalize GOT indirection in sym2addr
Signed-off-by: Edgar E. Iglesias <edgar@axis.com>
ipsel/plt.c
|
| 9eb1c0695cf95966a4b27355527691a2b7b7fe92 |
25-Sep-2012 |
Petr Machata <pmachata@redhat.com> |
Drop FORK_EXEC_SYSCALLS that is not used anymore these days
86/arch.h
|
| 78ed40f161c102a10c6033c28ad9a80e5ffe9550 |
22-Sep-2012 |
Petr Machata <pmachata@redhat.com> |
Fix passing struct(float,struct(float,float)) on x86_64
The problem was that we assumed that structure elements never overlap
eightbyte boundary. This assumption is violated by the above layout,
where the first two floats should be passed in %xmm0 together.
This case is covered by the Itanium HFA tests func_hfa_f3 and func_hfa_f4.
86/fetch.c
|
| a400a7c858f81b3be051a37011d88f8639bee598 |
13-Sep-2012 |
Petr Machata <pmachata@redhat.com> |
Fix passing of vararg parameters on Itanium
This fixes two failures in parameter.exp. That test case is now clean.
a64/arch.h
a64/fetch.c
|
| b50f0e143c6d0b99031b880b8ec85c0995a06033 |
10-Sep-2012 |
Petr Machata <pmachata@redhat.com> |
Style nit
a64/fetch.c
|
| bd3752795ee71477adb76c1f1af17b9da3964004 |
08-Sep-2012 |
Petr Machata <pmachata@redhat.com> |
Add fetch back end for ia64
This passes most tests in parameters.exp with the exception of:
FAIL: printf(\"sotnuh %d %ld %g %c.n\", 5, 6, 1.500*, 'X')
FAIL: printf(\"many_args[...]
These two cases are probably hitting some ABI special case for variable
argument lists.
a64/Makefile.am
a64/arch.h
a64/fetch.c
a64/trace.c
|
| a2e16c5cc8cc2343d40a44ac8b4d2230de9fc985 |
21-Sep-2012 |
Petr Machata <pmachata@redhat.com> |
Don't include arch.h, include sysdep.h instead
arch.h is supposed to be a simple configuration file and few have inclusion
guards. sysdep.h is the proper interface to use.
reakpoint.c
|
| 5371a73782447ffa19c0183ad55b6d0936c19e8a |
13-Sep-2012 |
Andreas Schwab <schwab@linux-m68k.org> |
Add fetch back end for m68k
This ports the m68k backend to ARCH_HAVE_FETCH_ARG, which fixes all
remaining testsuite failures.
68k/Makefile.am
68k/arch.h
68k/fetch.c
68k/trace.c
|
| 0fd388f31ce345fd575b750f7fbd2b4831e97070 |
12-Sep-2012 |
Andreas Schwab <schwab@linux-m68k.org> |
Update m68k syscall table
Regenerated from linux 3.6-rc.
68k/syscallent.h
|
| 17e330d4d500e880499bee92875dd55427e16c4a |
12-Sep-2012 |
Edgar E. Iglesias <edgar@axis.com> |
cris: autotoolization
Signed-off-by: Edgar E. Iglesias <edgar@axis.com>
ris/Makefile
ris/Makefile.am
|
| 135ac50e0e326f46942826fe08f64daaa0e3135a |
12-Sep-2012 |
Edgar E. Iglesias <edgar@axis.com> |
cris: Compile fix
Include missing library.h
Signed-off-by: Edgar E. Iglesias <edgar@axis.com>
ris/plt.c
|
| 563bca967f5c6c520643ff50aafb7d3b90e3507e |
05-Sep-2012 |
Sedat Dilek <sedat.dilek@gmail.com> |
mips: Fix prototype of arch_elf_init() in plt.c
arch_elf_init prototype was wrong with respect to its declaration
in backend.h.
Signed-off-by: Sedat Dilek <sedat.dilek@gmail.com>
ipsel/plt.c
|
| 49c93487406e5cc134b78b5c664ec8f77771d03a |
04-Sep-2012 |
Sedat Dilek <sedat.dilek@gmail.com> |
Remove PLTs_INIT_BY_HERE and E_ENTRY_NAME defines
Signed-off-by: Sedat Dilek <sedat.dilek@gmail.com>
ipsel/arch.h
|
| d57c9f8a8515781d687ce57eadc371038cd01526 |
03-Sep-2012 |
Petr Machata <pmachata@redhat.com> |
Remove remnants of PLT_REINITIALISATION_BP
This was only ever used for PowerPC, which doesn't need it anymore.
pc/arch.h
|
| a7e7bea0205959a2bdfe4c2049ebb01d3e3b23f7 |
01-Sep-2012 |
Petr Machata <pmachata@redhat.com> |
Update syscallent.h on MIPS from latest kernel
ipsel/syscallent.h
|
| 3b3b0d3a24c1b6c374cc2338d9791518fffa29be |
01-Sep-2012 |
Petr Machata <pmachata@redhat.com> |
Fix compilation on MIPS
ipsel/plt.c
ipsel/trace.c
|
| 5b4f4673ac4baf825eb5250049b97c9de1a1530a |
31-Aug-2012 |
Petr Machata <pmachata@redhat.com> |
Fix includes
a64/breakpoint.c
|
| 70747c3890f71f80ff6fafacc990293c91dcff77 |
31-Aug-2012 |
Petr Machata <pmachata@redhat.com> |
Use assert instead of printing an error that should never happen
a64/breakpoint.c
|
| 7bd88f51d4305f093ec45f557a0bde69ce5a7d11 |
31-Aug-2012 |
Petr Machata <pmachata@redhat.com> |
Fix formatting string
a64/plt.c
|
| 5e2cfb11c420af3352a62ea197f37f85c00e024c |
31-Aug-2012 |
Andreas Schwab <schwab@linux-m68k.org> |
Fix arch_plt_sym_val for m68k
68k/plt.c
|
| fcf256ceeab4b0b74cf1e18122e894aafce94fdc |
30-Aug-2012 |
Petr Machata <pmachata@redhat.com> |
PPC64 passes floating point equivalent structures in registers as well
Extract the structure unpacking code from s390 back end into the generic
type module. Call this from both s390 and ppc back ends.
pc/fetch.c
390/fetch.c
|
| 12d64f8bbae6245146900e8ee2292d854a22f26d |
30-May-2012 |
Petr Machata <pmachata@redhat.com> |
Move a comment in s390 backend
390/fetch.c
|
| df59c75ee9fa70343c9af929403cd3d371443714 |
30-May-2012 |
Petr Machata <pmachata@redhat.com> |
Support passing 8-byte structures via register pair on s390
390/fetch.c
|
| 0b926f68098c4dd0df922ba2c6214ed902cd4cd6 |
30-May-2012 |
Petr Machata <pmachata@redhat.com> |
In s390 fetch backend, compute type size in advance, pass in argument list
390/fetch.c
|
| 8d30fd9f26ccd15f0fa27e09d1fd99deddb17d36 |
30-May-2012 |
Petr Machata <pmachata@redhat.com> |
Move a comment where it belongs
390/fetch.c
|
| 232fb3cf407d139a475b296ac5e95ae344771786 |
30-May-2012 |
Petr Machata <pmachata@redhat.com> |
Drop gimme_arg from s390 back end
390/trace.c
|
| 917ef246757d89b74eac58e0664b3fe2146a6fdb |
30-May-2012 |
Petr Machata <pmachata@redhat.com> |
Implement arch_type_{sizeof,alignof} for s390
This fixes s390x/s390 tracing scenario.
390/arch.h
390/trace.c
|
| 789ad44c2fa11ea4e6508fd2feb98025954cf01d |
30-May-2012 |
Petr Machata <pmachata@redhat.com> |
Clean up s390/arch.h
Drop FORK_EXEC_SYSCALLS that is not used anymore these days.
Remove duplicity in defining LT_ELFCLASS and friends. We don't require
LT_ELFCLASS to be always defined anymore.
390/arch.h
|
| 82b1977bfdc0c31be26454257db28a1cb497fbf2 |
30-May-2012 |
Petr Machata <pmachata@redhat.com> |
Support s390/s390 tracing
s390x/s390 is not supported yet
390/fetch.c
|
| ecb2e82fe81684fc10876db6b73de8cefad06d9a |
29-May-2012 |
Petr Machata <pmachata@redhat.com> |
Implement fetch back end for s390x
390/Makefile.am
390/arch.h
390/fetch.c
|
| bac2da505ee174b7fb984b975c5938f88f0dbab2 |
29-May-2012 |
Petr Machata <pmachata@redhat.com> |
Move target_address_t to arch.h, rename to arch_addr_t
Note that the placement under arch.h is currently conceptual only. Since
no backend defines ARCH_HAVE_ADDRESS_TYPES, it's actually in sysdep.h in
all cases.
rm/breakpoint.c
pc/fetch.c
pc/plt.c
pc/trace.c
roc.c
race.c
86/fetch.c
86/regs.c
|
| 1b4d3bec9409b4c9a5cbc1be9faa31ea988c6ed3 |
18-May-2012 |
Petr Machata <pmachata@redhat.com> |
We need to fetch floating registers on pure i386 as well
Also, remove i386 from the fail spec of parameters.exp, as now it's
expected to pass.
86/fetch.c
|
| 67154e908e059729405c4479354b3216b3d62f90 |
17-May-2012 |
Petr Machata <pmachata@redhat.com> |
Align x86 syscallent tables, add new syscalls
86/syscallent.h
86/syscallent1.h
|
| ddd96a3fbc7c54146d1d831810fc9e29c9bc3c76 |
17-May-2012 |
Petr Machata <pmachata@redhat.com> |
Merge i386 and x86_64 back ends into a general x86
The reason being that x86_64 needs to handle i386 anyway, and keeping the
two together might reduce code duplicity.
akefile.am
386/Makefile.am
386/arch.h
386/plt.c
386/ptrace.h
386/regs.c
386/signalent.h
386/syscallent.h
386/trace.c
86/Makefile.am
86/arch.h
86/fetch.c
86/plt.c
86/ptrace.h
86/regs.c
86/signalent.h
86/signalent1.h
86/syscallent.h
86/syscallent1.h
86/trace.c
86_64/Makefile.am
86_64/arch.h
86_64/fetch.c
86_64/plt.c
86_64/ptrace.h
86_64/regs.c
86_64/signalent.h
86_64/signalent1.h
86_64/syscallent.h
86_64/syscallent1.h
86_64/trace.c
|
| 2d9be647dd7402402ad3d2d2f114632aba21eaca |
17-May-2012 |
Petr Machata <pmachata@redhat.com> |
Rename snip_small_int to align_small_int, move the function up
pc/fetch.c
|
| 36718b01e510ff3973e14b2bad0dbad3177b9943 |
17-May-2012 |
Petr Machata <pmachata@redhat.com> |
Fix compilation on PPC32
The test suite comes out clean in PPC32/PPC32 case as well.
pc/fetch.c
|
| dbcb32a5ab5a1c9e2d8cb70b07320db1ae8c234b |
17-May-2012 |
Petr Machata <pmachata@redhat.com> |
FP registers are not backed by GPR's on PPC32
This fixes func_many_args, printf/sotnuh and printf/many tests in
parameters.exp. The test suite is now clean in PPC64/PPC32.
pc/fetch.c
|
| 95792af484c635071ed96107d34a3762c4ca2ef2 |
17-May-2012 |
Petr Machata <pmachata@redhat.com> |
Size should be computed from the type stored at valuep
That because that type could have been changed above for passing structs
by reference on PPC32. This fixes the three func_struct_large* tests in
parameters.exp.
pc/fetch.c
|
| ebd9191418bd73796533d6e21a2536906459ec1f |
17-May-2012 |
Petr Machata <pmachata@redhat.com> |
Fixes for PPC fetch back end
This takes care of func_short, func_ushort, printf/sotnuh1, func_lens
and func_negative_enum tests in parameters.exp
pc/fetch.c
|
| b52f89c1302256bae003edc44c6e3788833629f1 |
17-May-2012 |
Petr Machata <pmachata@redhat.com> |
Elementary support for fetch_* from PPC32 processes
Many test case failures in parameters.exp, but elementary fetching now works.
pc/fetch.c
|
| d0a4cf7dc388729bce29fc8b1459e0a0007478f6 |
17-May-2012 |
Petr Machata <pmachata@redhat.com> |
Remove unused code
pc/ptrace.h
pc/trace.c
|
| af9e30ec852a24f52a36775154ff79cc4be0193d |
17-May-2012 |
Petr Machata <pmachata@redhat.com> |
Implement arch_type_sizeof and arch_type_alignof for PPC back end
pc/arch.h
pc/trace.c
|
| d5d70af126d79150ba62a471af08119f9cd44ed0 |
16-May-2012 |
Petr Machata <pmachata@redhat.com> |
Resolve the remaining problems in parameters.exp (PPC64 only)
pc/fetch.c
|
| 7d1e7926b92403ce03d1c2002c80937b280fa415 |
16-May-2012 |
Petr Machata <pmachata@redhat.com> |
Fix computation of address of the first used stack slot
This fixes func_struct_large in parameters.exp
pc/fetch.c
|
| f7537b62af3507de87d6fa645ef20fe61e3f3c09 |
16-May-2012 |
Petr Machata <pmachata@redhat.com> |
Implement passing and returning structures by value on PPC64
This fixes tests func_struct_2, func_struct_large2 and func_struct_large3
in parameters.exp.
pc/fetch.c
|
| 54737daf1c51ec826fcabc8189a7ff891f29d59b |
05-May-2012 |
Petr Machata <pmachata@redhat.com> |
First stab at fetch interface for PPC
This doesn't support passing structures by value and returning structures
is partially broken.
pc/Makefile.am
pc/arch.h
pc/fetch.c
pc/trace.c
|
| a7ec2d840a68bf103aacd240555406955c397ecf |
04-May-2012 |
Petr Machata <pmachata@redhat.com> |
Move code related to fetch.h from trace.c to fetch.c in x86_64 back end
86_64/Makefile.am
86_64/fetch.c
86_64/trace.c
|
| dcf518472b0f8581bac5af8a55d3a3c23472aaca |
04-May-2012 |
Petr Machata <pmachata@redhat.com> |
Use struct Process.e_machine in conditions, not .mask_32bit
Though what we need in fact is a field determining ABI of the process.
e_machine is just one aspect of the used ABI, it doesn't include ELF class,
which we need to tell x86_64 binary from x32 binary.
86_64/trace.c
|
| 2fb192b7bb49b659343f6dec89931168de00660c |
04-May-2012 |
Petr Machata <pmachata@redhat.com> |
Add interface value_pass_by_reference, use in x86_64 back end
86_64/trace.c
|
| 42332ad028c8616c6d8155520bdc7293b53af138 |
04-May-2012 |
Petr Machata <pmachata@redhat.com> |
Drop fpregs_copy from PPC back end
pc/ptrace.h
pc/trace.c
|
| 98d319eb9654ce903c751405535ebcd7077be887 |
04-May-2012 |
Petr Machata <pmachata@redhat.com> |
Fix includes in PPC back end
pc/plt.c
|
| 8ae3673880056b5da0bf9a5fbf9b37276adae56e |
30-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Fix memory errors
race.c
|
| ba1664b062414481d0f37d06bb01a19874c8d481 |
28-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Fix compilation
vents.c
roc.c
race.c
race.h
86_64/trace.c
|
| c6a1463684a03af0fd74bf438f85c26bb6c7a76c |
22-Feb-2012 |
Petr Machata <pmachata@redhat.com> |
PPC backend needs to include backend.h
pc/trace.c
|
| c07a26a8d0f86bc4137eb7b7fea8bbf9021275f1 |
09-Jan-2012 |
Petr Machata <pmachata@redhat.com> |
Add enum lens, drop ARGTYPE_ENUM
86_64/trace.c
|
| d5285b23e7e515ccb6afdb82fae32a608bb28a2f |
09-Jan-2012 |
Petr Machata <pmachata@redhat.com> |
Declare endianess in arch.h of each architecture
lpha/arch.h
rm/arch.h
ris/arch.h
386/arch.h
a64/arch.h
68k/arch.h
ipsel/arch.h
pc/arch.h
390/arch.h
parc/arch.h
86_64/arch.h
|
| 541cdc7736ffa770958208995fae71cbd1d50bcf |
09-Jan-2012 |
Petr Machata <pmachata@redhat.com> |
New x86_64 back end
86_64/arch.h
86_64/ptrace.h
86_64/trace.c
|
| 642626096a694c6af279d25d2b1b2fba5b10ddfb |
07-Jan-2012 |
Petr Machata <pmachata@redhat.com> |
Add backend.h for backend interfaces
reakpoint.c
vents.c
vents.h
roc.c
race.c
|
| cd972587a4c0d14d764b3d0a43d75f5941db14cc |
07-Jan-2012 |
Petr Machata <pmachata@redhat.com> |
Move event-que declarations to sysdeps, new backend interface process_removed
- the reason being that the que-handling was declared on global level, but
was only implemented in one back end. If this is deemed generally useful,
it should all be moved to front end, but as things are it's all the same,
so I'm preferring the less invasive change
akefile.am
vents.c
vents.h
roc.c
race.c
|
| e6ee77bc6197b11355ebb911f4e09d4ff95286d9 |
07-Jan-2012 |
Petr Machata <pmachata@redhat.com> |
Tweak x86_64 backend to fix the most glaring errors
- better back end is on the way
86_64/trace.c
|
| 6f9f03b3108b68e790eca48fcf053f74ff433094 |
06-Jan-2012 |
Petr Machata <pmachata@redhat.com> |
Drop struct arg_type_info.u.{float,double}_info
- this is not generally useful, for proper ABI support each backend has
to implement some sort of register-allocating state machine anyway (like
is currently done for ppc).
- ia64 was the only backend making use of this, fix that part of the code
a64/trace.c
|
| 97f13e3fca546aa292ba5d8aa6011338e64f2f7a |
06-Jan-2012 |
Petr Machata <pmachata@redhat.com> |
Drop unused interfaces save_register_args, struct stack_element.arch_ptr
lpha/trace.c
rm/trace.c
ris/trace.c
386/trace.c
a64/trace.c
68k/trace.c
ipsel/trace.c
pc/trace.c
390/trace.c
parc/trace.c
86_64/trace.c
|
| 46a41f7c72828f2f43a640e0ce1fbb41d7a196e9 |
06-Jan-2012 |
Petr Machata <pmachata@redhat.com> |
Drop umovestr, umovelong that are not used anymore
pc/trace.c
race.c
|
| 000e31195ad4ad30a0c80c93ab57a424e7d8d918 |
03-Jan-2012 |
Petr Machata <pmachata@redhat.com> |
Add type.c/.h, value.c/.h, value_dict.c/.h
- none of this is plugged in yet
- expr.c/.h is not in yet, so we still express array length with len_spec
integer. Some function are mocked out so the thing compiles
lpha/trace.c
rm/trace.c
ris/trace.c
386/trace.c
a64/trace.c
68k/trace.c
ipsel/trace.c
pc/trace.c
390/trace.c
parc/trace.c
race.c
86_64/trace.c
|
| f48031c60b05113194ad6dc2163ce3895a6b8d80 |
07-Feb-2012 |
Petr Machata <pmachata@redhat.com> |
Create breakpoint.h with decls related to breakpoints
reakpoint.c
|
| 5f8a5e78e9d410b8cb10401e463f9ffc9e062505 |
18-May-2012 |
Petr Machata <pmachata@redhat.com> |
Fix a s390 build failure
390/plt.c
|
| c897cb796dc4a7d256cbfbf0137ef7cdff9e8ece |
05-May-2012 |
Petr Machata <pmachata@redhat.com> |
Fix build on android, which requires that ptrace data and addr are void *
Also, for several requests that ignore addr, we pass 1.
Change these back to 0.
This patch is largely based on work by Florian Echtler <floe@butterbrot.org>.
rm/breakpoint.c
rm/regs.c
rm/trace.c
race.c
|
| c0ab28673d03bdf8f264556bde0d3806ea062322 |
05-May-2012 |
Petr Machata <pmachata@redhat.com> |
Fix compilation of ARM back end
rm/breakpoint.c
rm/plt.c
|
| 2eb941ddadeec2382a45a1e4e80003dc8abe59d0 |
05-May-2012 |
Petr Machata <pmachata@redhat.com> |
Drop process_get_entry, which is not useful right now
roc.c
|
| a82d32209f276e10dd1f437b38b4b6e5de32b980 |
01-May-2012 |
Petr Machata <pmachata@redhat.com> |
Fix 32-bit builds
pc/plt.c
|
| 6ca19a3c08bc75fb451df30f4c6e6c1a128731fe |
30-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Use read_target_8 in arch_translate_address_dyn
pc/plt.c
|
| 8bfb5734dfc5a82addc9db9d4c7642173bb4206b |
30-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Fix arch_translate_address on 32-bit PPCs
pc/plt.c
|
| b1492dfaca6882fa0798b549e0557c7dec6b7e9c |
30-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Split arch_translate_address into this and arch_translate_address_dyn
The former is called when ltelf is available. The latter in dynamic context
when ltelf is not necessary anymore (or if the data is necessary, it will
have been copied out to struct library).
pc/arch.h
pc/plt.c
roc.c
|
| 917725c482d4857561fbb1e6a9ca8a964549a22b |
26-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Add a commentary about a correctness issue to detach_task_cb
pc/plt.c
|
| cc0e1e4b83d69441cc5f61ea87eda5458ee9fae3 |
25-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Replace some uses of error with fprintf
error is not standard so it has no business being used in generic code.
The linux-gnu back end is useful for android, and that doesn't have that
interface either.
reakpoint.c
roc.c
race.c
|
| fd2641c0d27705c89d224e3205b1296110b6598c |
24-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Fix double frees, remove_process now releases memory
Presence in singly-linked list strongly indicates that process is heap
allocated, and that if we are about to stop tracing it, we should also
reclaim the memory.
pc/plt.c
|
| d94108be1ae41b0ead839fb01cb0c1fd384ea65f |
24-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Use on_retract to fix detach on PPC64 stripped
There should still be a small window where this can fail, namely when we
catch IP in a stub, after address is read from .plt and moved to CTR, but
before it jumps there (which would get the process to breakpoint address,
which we now notice). We may need to either fixup the .plt[0], or do
code reading to figure out whether we are in a stub and compare CTR.
pc/plt.c
|
| 86d3828129e70222abdb77bdda6f31a7f1eafd5a |
24-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Add on_retract breakpoint callback
race.c
|
| 622581f4db4ae4ef8932984a4d1c4935ca7f61ef |
23-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Use the right process to insert breakpoints for stepping over atomic block
race.c
|
| e577a10e73b41e2c68fff7e8d89c22dfb9378b99 |
23-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Don't assume that ORIG_RAX/ORIG_EAX always contains original syscall number
Sometimes it's set to -1, apparently in cases where the system call
shouldn't be restarted. It that's the case, just assume that it's return
from the topmost system call. This gets rid of some "unexpected breakpoint"
messages on x86_64 and i386.
386/trace.c
86_64/trace.c
|
| 02a41a5a996ffb774078cafe6311150720d2fd5b |
20-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Fix attach on PPC32 BSS
pc/plt.c
|
| a753c98f0957df9d12d5ecf0607eef7033ab7745 |
20-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Fix compilation on PPC32
pc/plt.c
|
| b556058c2126fd69dfe03d1187429d98b5636d61 |
19-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Fixes for PPC32 BSS unprelinked
The testsuite now mostly passes, the outstanding problem that's specific
to that configuration is with attach and vfork-threads test cases.
pc/plt.c
|
| 77e8fa8d24af41ec57635d3e73726d3930d95830 |
19-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Reinsert vfork return breakpoint in handler
Previously we assumed that that breakpoint would still be around in the
breakpoint table, but now that they are actually erased on removal, it
doesn't hold anymore. If it should hold (which might happen), then
insert_breakpoint knows to just piggy-back on an existing breakpoint.
race.c
|
| 24a47043b182f7fec6670d5a4d0e485e94a53ac1 |
19-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
When removing atomic breakpoints, fetch data from leader, not current proc
race.c
|
| d2fc09dccfc18680209a918dc8cbcc1f75e41118 |
19-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
On PPC32 BSS unprelinked, watch both .plt slots for changes
pc/plt.c
|
| 42748aca73359c83881556c8b28f6cda4f1c143b |
19-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Fixes for atomic block support
- Support branches in atomic blocks
- Fix interaction between atomic skip and single stepping
pc/plt.c
pc/trace.c
race.c
race.h
|
| f1fd7cd52e3829f0073c2d740b3d1998dcfd4923 |
19-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Fix post_singlestep for cases where event was already sunk
race.c
|
| 9e1e969838d9cc520abb96038aa98520c08c80b9 |
19-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Fix rebase fallout
pc/trace.c
race.c
|
| 8454bd71f00aba4b383d42a8fec44695c9b0f018 |
17-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Don't emit newline in one debug message
roc.c
|
| f685a3d0b3aaf9cf2bda30937ddc1a55009ee01a |
17-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Proper tracing of PPC32 BSS unprelinked binaries
pc/plt.c
|
| b04b64b7285183ad5fbc011ee381613359c35a2b |
17-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Debugging output in PPC back end
pc/plt.c
|
| 585f60f7b171a391fbd5149f3d397d192168a67c |
17-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Rename PPC symbol states to reflect that they are used for PPC32 as well
pc/arch.h
pc/plt.c
|
| 8d930e8f095d04d62705f35f5362a537b73ffefe |
17-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
We also need to mark PLT BP as resolved after post-update BP hits
pc/plt.c
|
| 8557b4a1a6f2a6d12867ee57b417e3e941963721 |
17-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Extract marking as resolved into own function
pc/plt.c
|
| 9a45d22e3b67a0d611675630cb7cd38e3895a1ff |
17-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Partial support for PPC32 BSS unprelinked case
Turs out on PPC32 with BSS PLT, .plt changes after the first call as well.
pc/plt.c
|
| 72b5ee8b5c795e88e090be6bc64492ec216d40ce |
17-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Actually unresolve .plt slot when the post-update breakpoint hits
The clever optimization from two days ago actually broke things as my test
case was doing a lot of one-off calls and I never noticed that it breaks
the common case.
pc/plt.c
|
| d95733284377c0b186ba0c81a1158edc2b913e45 |
17-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Re-enable breakpoints for PPC32 BSS unprelinked case
Unfortunately in this case, the dynamic linker additionally overwrites
the PLT entry on first call. We will have to handle this similarly to
what we do for PPC64 unprelinked stripped case.
pc/arch.h
pc/plt.c
|
| 05058b79e6f995e4d328952ecd73cf090d89cb13 |
17-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Streamline continue of PPC64 unresolved PLT breakpoint
pc/plt.c
|
| ea8eb9a606096fd072abe104b4239f67883ad39a |
17-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Add casts where we convert from wide integer to narrow pointer
pc/plt.c
roc.c
|
| 1a53e0ab6ee4e53dfa6e81239deeaefdcd639082 |
17-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Fix compilation of linux-gnu/i386/plt.c
386/plt.c
|
| fbd9742d03154ca842eeae8f6a32e35c1e3c8326 |
16-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Fix nits in linux-ppc back end
pc/arch.h
pc/plt.c
|
| 19c0f295a1a923b923e1083e480496c1454a2e0e |
15-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Commentary updates to PPC PLTs
pc/plt.c
|
| e5ebe21369b45ce997b74c04e14f06c67b9b70c1 |
15-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
In PPC back end, unresolve_plt_slot may fail
pc/plt.c
|
| 6b314183200b1462ef4aad6e04fda72f3f6b0d87 |
15-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Improve tracing speed in PPC64 unprelinked stripped case
After we learn where the after-update instruction is, we install breakpoint
and use that as a door-stop instead of singlestepping all the way. This
improves the tracing speed of a synthetic test case that doesn't does many
one-off library calls by a factor of 25x.
pc/arch.h
pc/plt.c
|
| 5d1c5fc1474aed4a02103bb34b7532b9d26c6f3d |
15-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Call breakpoint handler if that's what we hit during stopping on linux
race.c
|
| 1e2a4dd274ebb612e0cfa5c83ef081e5846aa949 |
15-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Expose on_all_stopped hooks for linux stopping handler
race.c
race.h
|
| abfe118271cd45581f33f71b0cf0fddc3a2e1b05 |
15-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Nits in PPC back end
pc/arch.h
|
| 24c6e9daa8d89e13c19fc3f9a475ba7913d7d9c8 |
15-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Add arch_library_symbol arch hooks, call properly
pc/arch.h
pc/plt.c
|
| e8d9076a97f6617868466a99bd18e11e3f6389ac |
15-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
library_symbol_init may fail
pc/plt.c
|
| 218c5ff26841f5bbd188c42ccbd67422a7a20556 |
15-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Nits
race.h
|
| 9ace5742e1d61c88db8adefdeb9e49d4c3a9ff5d |
14-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Updates to debugging messages
race.c
|
| d3cc9889fdfe2e523e99ca5f664f8ae4b3936612 |
13-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Fix breakpoint cloning
rm/breakpoint.c
pc/plt.c
|
| f0f90edc2b0746e0c9f05f6224e38de6fbfed9d2 |
13-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Fix debugging output
reakpoint.c
|
| 18bd8ffa003c680e3d44d1b05fc3a7b43e6a25ce |
10-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Add auxv reader
We should be able to instrument dynamic linker as well, if that is
requested. Unfortunately we may still need artificial _start probe,
because we don't know _r_debug address until the dynamic linker fills
the DT_DEBUG entry. (Though we are likely to find _r_debug in symbol
table if we look.)
roc.c
|
| 18c801c3f29081d9de517815df89bc1bbf8e2188 |
07-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Drop the rest of LS_TOPLT_POINT support
If any other architecture needs it, we'll add it back in some form. But
now that we have callbacks on breakpoints, it might be possible to do it
differently, without explicit support.
pc/plt.c
|
| bc6c466bb7a758b12da073aae4553e46b64ae472 |
06-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Update syscallent.h on PPC linux
pc/syscallent.h
|
| 3ed2a4227d5f2a8d44f3c6ffd704a9d2e867c591 |
06-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Drop PID parameter from trace_set_options
race.c
|
| 50969624d377d485c6d9638a4b00a0b02022d635 |
06-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Add asserts to ppc64_plt_bp_continue
pc/plt.c
|
| 8b00d5bb6a0925ece06aad0d9df0a85e8dbd7b57 |
06-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Drop leftover fprintf's
- convert some to debug or error calls
pc/plt.c
roc.c
|
| 052b5f191ec00b7677c43278cb2ce886af1d0ef1 |
06-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Ignore non-PLT symbols on PPC's arch_breakpoint_init
pc/plt.c
|
| 1b4c76e739f36a01ca4cc07a1df5aa5e970279e6 |
06-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Fix a commentary nit
pc/arch.h
|
| 029171fffcf6328768866cf510763b2eb927f1bd |
05-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Drop debug_state from struc process, use custom process.debug for that
- ideally we would define arch_proc_data in sysdeps/linux-gnu/os.h or some
such, same as we do with arch_breakpoint_data and others, but that's
currently not implemented, so do it this way.
roc.c
|
| b5f80ac8982c40f79915ce1e1cb9bf8650ac5fe7 |
04-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Don't create PLT symbols that don't match filtering criteria
- instead of creating them, keeping them around, and then just not enabling
breakpoints for them
roc.c
|
| 74d80544c9ad0b1c5ff94dda639e12ddf1d2ca1c |
03-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Nits
roc.c
|
| 89ac0395a9e018f4d4dbda9d3e27159419a92da2 |
03-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Differentiate fresh libraries from old by "key", not by base address
- that because base is unknown at the time that library is mapped in, only
bias is known. We'd have to read the elf file to determine whether this
is the right library or not. Instead we assume that the dynamic linker
keeps linked list, and use address of internal structure as the key
roc.c
|
| 6ef5c9e302885ac85a009e5ba000825199813b81 |
03-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Drop debugging output
pc/plt.c
|
| 050b0a6fd01fc01952c0ab8dbb84d6eba65c71c9 |
03-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Coding style
reakpoint.c
|
| 949a56a0e1d4a13dc437621a991e993d6623aef5 |
03-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Fix PLT calls on PPC64 unprelinked stripped
reakpoint.c
pc/plt.c
race.c
|
| 6a7997de9eabf9b0c6e32309ca08870198eb864b |
29-Mar-2012 |
Petr Machata <pmachata@redhat.com> |
Support cross tracing from libraries
- PPC64 and other arches where OPD is used may be broken at the moment
roc.c
|
| 12affff3c88731a0880690442485494e540f7a58 |
29-Mar-2012 |
Petr Machata <pmachata@redhat.com> |
Make several breakpoint callback names more systematic
roc.c
|
| 52dbfb161efeab85bddc880966db2f7af9b9cf9a |
29-Mar-2012 |
Petr Machata <pmachata@redhat.com> |
Shuffle breakpoint functions around
pc/plt.c
roc.c
|
| cb9a28da448439eab4bf554810fd1004fbc00885 |
28-Mar-2012 |
Petr Machata <pmachata@redhat.com> |
Configurable ugly workaround invocation
- doesn't help to fix the weird singlestepping problem on ppc64 though
pc/plt.c
race.c
race.h
|
| c67a6e62bc7e95f7f181a51c5b314ef10a6e231e |
28-Mar-2012 |
Petr Machata <pmachata@redhat.com> |
Add missing implementations of default callback
pc/arch.h
|
| ce40b3974d485837d130afc6b8034a83b3d03c5b |
23-Feb-2012 |
Petr Machata <pmachata@redhat.com> |
Whitespace fix
roc.c
|
| 17476b7bfac0e1cb68f9a1dbdb65dab2ac2160c6 |
23-Feb-2012 |
Petr Machata <pmachata@redhat.com> |
Drop ELF_MAX_SEGMENTS, simplify iteration through dynamic entries
roc.c
|
| 9df15016447915a61526af8cb81c588913bcf44c |
20-Feb-2012 |
Petr Machata <pmachata@redhat.com> |
Remove "enable" flag from insert_breakpoint
roc.c
race.c
|
| 58b2d0fd5deed5f33fbd47a6b6e5c109f43908b5 |
28-Mar-2012 |
Petr Machata <pmachata@redhat.com> |
Somewhat working tracing of stripped unprelinked PPC64 binaries
- this seems like it shoud work, and indeed it does, mostly. But after the
stopping handler is uninstalled, we get series of apparent singlestep
events, even though we just call PTRACE_SYSCALL time after time. If
there is a kernel bug, we may need to employ some sort of code reading
after all.
pc/arch.h
pc/plt.c
|
| 36f40e7d46838ec41ea03a2b9b748536d8fd57e2 |
28-Mar-2012 |
Petr Machata <pmachata@redhat.com> |
Further generalize the process stopping handler
race.c
race.h
|
| 9282254e29896366ea354b930b06f9defebbd041 |
28-Mar-2012 |
Petr Machata <pmachata@redhat.com> |
Move declarations related to process stopping handler to a new file trace.h
akefile.am
race.c
race.h
|
| c1aa9588f3347c285038c2e45c01bbd2e7086a9f |
27-Mar-2012 |
Petr Machata <pmachata@redhat.com> |
Generalize the stopping handler so that it can be reused
- though more generalizing may be needed
race.c
|
| f04d032e298c410e45553223a979c1f26f280259 |
27-Mar-2012 |
Petr Machata <pmachata@redhat.com> |
Error reporting in PPC back end
pc/regs.c
|
| b64b5c7b6f4a368ccaf60507090192845221a3be |
27-Mar-2012 |
Petr Machata <pmachata@redhat.com> |
Support tracing prelinked stripped PPC64 binaries
- distinguish stub breakpoints from unresolved PLT breakpoints. The latter
have special on_continue handler that just moves IP
- unprelinked stripped case not supported yet
pc/arch.h
pc/plt.c
|
| d1746d17eda0c4d2c1004c9deb8b229eb6fb1c78 |
27-Mar-2012 |
Petr Machata <pmachata@redhat.com> |
Generalize load_ppcgot in PPC back end
pc/plt.c
|
| 8cce1193ebd35cb5a8b288bc7325cdda1b8ffe50 |
25-Mar-2012 |
Petr Machata <pmachata@redhat.com> |
Add arch_breakpoint_destroy as a counterpart to arch_breakpoint_init
- even though we don't destroy breakpoints at all right now
rm/breakpoint.c
|
| 7b36114b46cf93aa2828de055e71ec57a801dfc1 |
24-Mar-2012 |
Petr Machata <pmachata@redhat.com> |
Better handling of errors in the code that looks for PPC64 stub symbols
pc/plt.c
|
| 37d368e49f2d757484252a060d3021de96998e0d |
24-Mar-2012 |
Petr Machata <pmachata@redhat.com> |
Support for tracing on PPC64 with stub symbols in symbol table
pc/plt.c
|
| 50b96039ebfc60825bb7afb8ab8f56bb87aeb67b |
24-Mar-2012 |
Petr Machata <pmachata@redhat.com> |
Make MIPS and PPC arch.h into full-fledged include file, glue in sysdep.h
ipsel/arch.h
pc/arch.h
|
| 4d9a91c5c677d6a6b2db21f00385bce5167373c4 |
24-Mar-2012 |
Petr Machata <pmachata@redhat.com> |
Add arch_elf_destroy
ipsel/plt.c
pc/plt.c
|
| e6523e69b0230fef8dddbf628e0511a004203eb8 |
24-Mar-2012 |
Petr Machata <pmachata@redhat.com> |
Don't require target library as part of symbol init
roc.c
|
| 644d669f96c0fe261fe938cecda41938e804c7d9 |
24-Mar-2012 |
Petr Machata <pmachata@redhat.com> |
Drop recently introduced arch_elf_dynamic_tag
ipsel/arch.h
ipsel/plt.c
pc/arch.h
pc/plt.c
|
| be04d6b478ecc51902f34672e4c3c1516d502c23 |
24-Mar-2012 |
Petr Machata <pmachata@redhat.com> |
Drop library_symbol.is_weak
roc.c
|
| 4e2073f64f9db2974d89064dcdc49b2ed7aa9006 |
21-Mar-2012 |
Petr Machata <pmachata@redhat.com> |
Support tracing PPC32 processes with both BSS and secure PLTs
- only entry breakpoint is realized at the beginning
- when it's hit, it's deleted, and all enabled breakpoints are realized
- all PLTs are LS_TOPLT_EXEC. The difference is in the way the PLT entry
address is computed
pc/arch.h
pc/plt.c
|
| b120fdfda45e49d082ab2aa5caadf030f50cb4ed |
21-Mar-2012 |
Petr Machata <pmachata@redhat.com> |
Change interface of ltelf_read_library
- base->bias, pass proc
roc.c
|
| 81c65272697a63d81e6baece69bef6ba8f55932e |
21-Mar-2012 |
Petr Machata <pmachata@redhat.com> |
Error handling
reakpoint.c
|
| e67635d6dcecb0f44448a5329d69fd0de74ebaba |
21-Mar-2012 |
Petr Machata <pmachata@redhat.com> |
Move arch-specific bits from ltrace-elf.c to PPC and MIPS back ends
ipsel/arch.h
ipsel/plt.c
pc/arch.h
pc/plt.c
|
| d3dc17b5e87068819a780af434afaf69e63d2b50 |
21-Mar-2012 |
Petr Machata <pmachata@redhat.com> |
Fixes in includes
reakpoint.c
race.c
|
| 74132a447b18c588c698ecfb8cf4ad4b16360f2f |
16-Mar-2012 |
Petr Machata <pmachata@redhat.com> |
Change iterators each_task, each_process, library_each_symbol
... now the restarts are supported by start_after instead of start.
Otherwise the restart scenario was something like:
for (it *sym = NULL; (sym = next(sym)) != NULL; )
if ((sym = sym->next) == NULL)
break;
which just seems too convoluted.
race.c
|
| 522a6ca083c8b9e964548b0e79a4bdc8095d6e2e |
16-Mar-2012 |
Petr Machata <pmachata@redhat.com> |
library_symbol_init takes target_address_t instead of GElf_Addr
roc.c
|
| c805c624d4cd23674bbc18f9d0f97c5d8dcdff7c |
02-Mar-2012 |
Petr Machata <pmachata@redhat.com> |
wait_for_proc may fail, and should simply waitpid instead of ptracing
- which means that we need to continue the process after starting it,
the same as we do when attaching
race.c
|
| 29add4fdf852b10ddd22cac0d1390f6d01577bc2 |
18-Feb-2012 |
Petr Machata <pmachata@redhat.com> |
Support bias, display library name on breakpoint
- syscall support very hacky right now
roc.c
|
| 2b46cfc1127d390eddd9593fe5ce5399c1f68130 |
18-Feb-2012 |
Petr Machata <pmachata@redhat.com> |
The first crude version of tracing across libraries
- the patch will be sliced later
rm/arch.h
rm/breakpoint.c
rm/regs.c
reakpoint.c
vents.c
pc/plt.c
roc.c
race.c
86_64/plt.c
|
| 366c2f46d844f040458df9b7e35fc3b8527ed2d3 |
09-Feb-2012 |
Petr Machata <pmachata@redhat.com> |
Move process-related functions and structures to proc.h
lpha/plt.c
lpha/regs.c
lpha/trace.c
rm/plt.c
rm/regs.c
rm/trace.c
reakpoint.c
ris/plt.c
ris/regs.c
ris/trace.c
vents.c
386/plt.c
386/regs.c
386/trace.c
a64/plt.c
a64/regs.c
a64/trace.c
68k/plt.c
68k/regs.c
68k/trace.c
ipsel/plt.c
ipsel/regs.c
ipsel/trace.c
pc/plt.c
pc/regs.c
pc/trace.c
roc.c
390/plt.c
390/regs.c
390/trace.c
parc/plt.c
parc/regs.c
parc/trace.c
race.c
86_64/plt.c
86_64/regs.c
86_64/trace.c
|
| bc37326ace5c70e57928c000162cffbcca9afb77 |
07-Feb-2012 |
Petr Machata <pmachata@redhat.com> |
Remove Breakpoint typedef
rm/breakpoint.c
reakpoint.c
a64/breakpoint.c
race.c
|
| 61196a4a81e77322bf1f3dc609007f5d35a5103a |
07-Feb-2012 |
Petr Machata <pmachata@redhat.com> |
Drop Process::breakpoints_enabled
vents.c
|
| 02648a119092bb5b64918063521237f257283c72 |
07-Feb-2012 |
Petr Machata <pmachata@redhat.com> |
Drop most of start-up magic and use startup callback instead
vents.c
|
| a611fc8f887c099b04cb764f477b9289a4c21d6e |
07-Feb-2012 |
Petr Machata <pmachata@redhat.com> |
Cleanup
roc.c
|
| 9294d82f67e20f5f2b61f317ad04f5cb717c7d27 |
07-Feb-2012 |
Petr Machata <pmachata@redhat.com> |
Create breakpoint.h with decls related to breakpoints
reakpoint.c
vents.c
roc.c
race.c
|
| a266acb9c3bbde884a32268f164de62d03aa04d0 |
12-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Add code for stepping over atomic instruction sequence on PPC
pc/arch.h
pc/trace.c
race.c
|
| ffe4cd25089680daf1bd1ec0114d177ec3e0cf95 |
11-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Handle detach from sleeping or unresponsive processes
vents.c
race.c
|
| cec06ec8282c538a40bde968ae36fe8356daffaa |
10-Apr-2012 |
Petr Machata <pmachata@redhat.com> |
Warn when we fail to trace and SELinux boolean deny_ptrace is in effect
race.c
|
| b4f9e0c0413c03ad897d7aad320a4a1cd49c467a |
07-Feb-2012 |
Petr Machata <pmachata@redhat.com> |
Fix a race between fork/trace_me and first ptrace request
- That first ptrace may fail if ltrace is too quick
race.c
|
| 97b208405bcbf2fc75a70fba7d094740f590cb90 |
22-Nov-2011 |
Petr Machata <pmachata@redhat.com> |
Turn in a workaround for seeing a process as R after waitpid (should be T)
vents.c
|
| 12387c439d08755c0d8edf2516677d3f7b1f6ef0 |
22-Nov-2011 |
Petr Machata <pmachata@redhat.com> |
Be more fault tolerant on ia64
- this fixes some rare bugs that were observed on ia64
a64/regs.c
a64/trace.c
|
| 9d29b3e7d2d7f7ccb1891dc56b03e91a0319cf42 |
09-Nov-2011 |
Petr Machata <pmachata@redhat.com> |
Only delay continuing the process if we are actually mid-_stopping_
- The check fired also on vfork handling, and this hung the process
- Add a test case exercising this code, courtesy of Michal Nowak and
Denys Vlasenko
race.c
|
| 43d2fe5436d39da0b1ff6648fc0dfd766d28243e |
02-Nov-2011 |
Petr Machata <pmachata@redhat.com> |
Don't queue sysret events
- That's because on s390x, we use process call stack to actually figure out
whether it's a syscall/sysret event, and if it's sysret, then what syscall
it returns from. So we need to keep the stack in sync with reality.
race.c
|
| 06986d5432c92337dc2fc0bd81fa43640b48d1e2 |
02-Nov-2011 |
Petr Machata <pmachata@redhat.com> |
More robust fix of signal-before-singlestep problem
- On s390x, when we PTRACE_CONT the process, it will actually continue,
even though we ordered PTRACE_SINGLESTEP before. So always issue
PTRACE_SINGLESTEP when we get signal before singlestep.
- Also, simply queue the signal as usual.
race.c
|
| d5d93c492fd7cec016f09883f1e9c09cf97c6fb5 |
21-Oct-2011 |
Petr Machata <pmachata@redhat.com> |
Signal event may arrive before we finish singlestepping
race.c
|
| 36d1982d45d54e455413a5c0959f87b1857940d6 |
21-Oct-2011 |
Petr Machata <pmachata@redhat.com> |
Remove "default" branch in the switch in task_stopped
- and a couple commentary changes
race.c
|
| 3797cd6f378d647583f15e2134cbcf9143f57f9c |
03-Oct-2011 |
Petr Machata <pmachata@redhat.com> |
We have to insert breakpoint into parent, not into leader
- leader might be dead by the point this gets executed
race.c
|
| ba9911f3603473791163bf11dd1c59503faa4ff1 |
27-Sep-2011 |
Petr Machata <pmachata@redhat.com> |
Fix interplay between vfork and threads
race.c
|
| 134a108693f8155d2167675ad8127ff6def1425e |
27-Sep-2011 |
Petr Machata <pmachata@redhat.com> |
Trace vfork return in parent as well as in child
race.c
|
| cbe29c6c0ad01839a81272c4715ea73d17e89611 |
27-Sep-2011 |
Petr Machata <pmachata@redhat.com> |
Basic support for tracing vfork
- allow even individual tasks to have their event handlers. These are
called in precedence to the thread group event handlers.
- distinguish CLONE and VFORK events
- add methods for dynamic changes of leader
- add new process status ps_sleeping
vents.c
roc.c
race.c
|
| f296e0141c062cc0fd8e3301460ba62c49829973 |
03-Sep-2011 |
Petr Machata <pmachata@redhat.com> |
Drop unused variable
race.c
|
| 73894bdb7ef4ea81a44a7b8b3ab896f6b3cab344 |
20-Aug-2011 |
Petr Machata <pmachata@redhat.com> |
Add explanatory comment to the ugly workaround
race.c
|
| 590c80862c009608fd70e976a282473234d92265 |
20-Aug-2011 |
Petr Machata <pmachata@redhat.com> |
Rewrite the whole detach logic to use stopping handler
race.c
|
| ec85e84d2d3dc441dc2f1a6b403ed495cefd2c5b |
20-Aug-2011 |
Petr Machata <pmachata@redhat.com> |
Don't get stuck on detach waiting for SIGSTOP to arrive to stopped process
race.c
|
| 3c9b6295c4d4f7424e84c2fbef2fc98bda5e672c |
20-Aug-2011 |
Petr Machata <pmachata@redhat.com> |
More detach bugfixes
race.c
|
| 46d66ab6d7ef5d12e4d3dd3671450f77a776ea1b |
20-Aug-2011 |
Petr Machata <pmachata@redhat.com> |
Work around some odd behavior of PTRACE_DETACH after PTRACE_SINGLESTEP
- this might turn out to be a bug in the kernel that I'm using right now.
The thing that when task is detached right after it was singlestepped
(and the singlestep is caught by waitpid), it's immediately killed by
the SIGTRAP signal. This goes away if we insert breakpoint right under
the IP, CONT the process, catch the breakpoint event, retract the
breakpoint, and detach _then_.
race.c
|
| 4b9f4d956e125360d4860d6a8d7574a271c799a0 |
20-Aug-2011 |
Petr Machata <pmachata@redhat.com> |
Refine the logic for detach after singlestep
race.c
|
| 13d5df7743a79c8bceeb9be50d57f9b0f76fc4e8 |
19-Aug-2011 |
Petr Machata <pmachata@redhat.com> |
Undo breakpoints also for unqueued events
race.c
|
| 68abfea2bb3c5e57214929f773755dfcef63fe90 |
19-Aug-2011 |
Petr Machata <pmachata@redhat.com> |
Fix one source of SEGVs/TRAPs on detach
- we can't cont the process if the IP is mid-instruction due to the process
having hit a breakpoint recently
race.c
|
| 1974dbccbf10a2dd8e06724d2fb4eb61fd91076a |
19-Aug-2011 |
Petr Machata <pmachata@redhat.com> |
Fix serious race in attach to many-threaded process
- the gist is that tasks continue running while we attach, so more tasks
come into existence, or the ones that we didn't attach to yet disappear,
etc.
- besides, we really can't enable breakpoints before we are done attaching,
otherwise the still-running tasks risk running into them and dying of
SIGTRAP.
roc.c
race.c
|
| 3c516d5ced53508d8df7d82914a3190a3235b62d |
18-Aug-2011 |
Petr Machata <pmachata@redhat.com> |
Handle the race between reading list of tasks and attaching to those tasks
- For processes with many threads, we still get SIGTRAP on startup. This
only takes care of ltrace SIGSEGVing and/or giving out copious "cannot
attach" messages.
roc.c
|
| 28e80f65e2c808f715e466cdb8c0c53cfe78c756 |
09-Aug-2011 |
Petr Machata <pmachata@redhat.com> |
Move break_insn to function that's its sole consumer
reakpoint.c
|
| e21264ed3759f02fa4577bf9ac319d8dc36e248a |
06-Oct-2011 |
Petr Machata <pmachata@redhat.com> |
Fix a race between disabling breakpoint and stopping all threads
race.c
|
| 18c970711cd2597d52c2f146d97350d62980ec2c |
06-Oct-2011 |
Petr Machata <pmachata@redhat.com> |
We need to check for exit earlier
race.c
|
| 750ca8c13f402a40b01802dcb5c88e092ff68125 |
06-Oct-2011 |
Petr Machata <pmachata@redhat.com> |
Fix a race between waiting for SIGSTOP delivery and task exit
roc.c
race.c
|
| 617ff0bb228c64df4bb6682321338aecfec2d170 |
06-Oct-2011 |
Petr Machata <pmachata@redhat.com> |
Type the "process_status" interface properly
- and in implementation for linux-gnu, reflect that some kernels don't
distinguish job-control stop and tracing stop by different letter.
roc.c
race.c
|
| f2d2ba5ce96587f4cc0975441a5da4c6e672e8fa |
09-Jul-2011 |
Petr Machata <pmachata@redhat.com> |
Fixes for process exit before doing checks for syscalls etc.
vents.c
86_64/trace.c
|
| 602330f458e8049ec07685c4e02f0368821a2d2c |
09-Jul-2011 |
Petr Machata <pmachata@redhat.com> |
Handle multi-threaded attach/detach gracefully
race.c
|
| 98f0992f9c3fd5da41a4928407251d9f5a837702 |
09-Jul-2011 |
Petr Machata <pmachata@redhat.com> |
Use custom event handler for implementation of stop-the-world
- ... which itself is necessary for breakpoint re-enablement without races
race.c
|
| 4007d74c2435ce4b50305d64ffe831627f989335 |
09-Jul-2011 |
Petr Machata <pmachata@redhat.com> |
Facility for custom event handler
vents.c
|
| 69a03e6f8c15fb0272089e387a658acad887fb9c |
09-Jul-2011 |
Petr Machata <pmachata@redhat.com> |
Event queue
vents.c
|
| 9a5420c82a2fd81681572a2e3859ea1671c3bded |
09-Jul-2011 |
Petr Machata <pmachata@redhat.com> |
Add a concept of tasks and leader thread
vents.c
roc.c
race.c
|
| cebb88491e68bdf4d466b03970c9ca77f65615f8 |
09-Jul-2011 |
Petr Machata <pmachata@redhat.com> |
Conceal the list of processes behind an interface
vents.c
|
| c7585b60235268e1f62bd91c5f040ef6df6e0ef3 |
08-Jul-2011 |
Petr Machata <pmachata@redhat.com> |
Add argument that defines whether we should enable breakpoints
roc.c
|
| 6901b7e429f7e92440f3501df864ad74b39f66c2 |
09-Jul-2011 |
Petr Machata <pmachata@redhat.com> |
Cleanups in next_event
- always fetch instruction_pointer eagerly to simplify the code
- clarify that we can indeed get nen-SIGTRAP signals in certain cases
vents.c
|
| f789c9c0d1f3301afad66e5f0520b9093665f242 |
09-Jul-2011 |
Petr Machata <pmachata@redhat.com> |
Pass Process * instead of pid to a couple functions
reakpoint.c
race.c
|
| 2662768efe599f6bb43c4310177e30f56b601bb7 |
08-Jul-2011 |
Petr Machata <pmachata@redhat.com> |
Coding style nits
reakpoint.c
vents.c
roc.c
race.c
|
| c46448f4e5a4c124fbc75ca9b14697212e676893 |
07-Mar-2011 |
Michael K. Edwards <m.k.edwards@gmail.com> |
fix type punning in ARM arch_(dis|en)able_breakpoint
rm/breakpoint.c
|
| 9bc4a9b46340209f774181f7e55ece1c47bd59d6 |
06-Mar-2011 |
Michael K. Edwards <m.k.edwards@gmail.com> |
Make PPC64 long-to-int conversions conditional on SIZEOF_LONG == 8
pc/plt.c
pc/trace.c
|
| e276cf41371cd5c5a722043658ccc15321aa15ac |
08-May-2011 |
Andreas Schwab <schwab@linux-m68k.org> |
Correct fetching arguments beyond eighth arg on ppc/ppc64
The parameter save area starts at sp+8 on ppc32 and at sp+112 on ppc64.
Signed-off-by: Andreas Schwab <schwab@linux-m68k.org>
pc/trace.c
|
| 08677d0557284d36d3c898b1d876e65411069912 |
19-May-2011 |
Petr Machata <pmachata@redhat.com> |
Fix a buffer overrun bug
roc.c
|
| 3ba522fed3d34f476090da9349be96e15e0652cb |
15-Dec-2010 |
Zachary T Welch <zwelch@codesourcery.com> |
Fix build for x86_64
Some ELF-related variables differ depending on the native host, so
this patch uses conditional compilation tricks to ensure everything
works out on both 32- and 64-bit hosts.
Signed-off-by: Zachary T Welch <zwelch@codesourcery.com>
roc.c
|
| ba6aca21bd9e0e66dac17b1828cf6b6e7377b983 |
09-Dec-2010 |
Zachary T Welch <zwelch@codesourcery.com> |
Fix building with --enable-debug
Numerous warnings had snuck into the build, preventing ltrace from
building with --enable-debug (which uses -Wall, -Werror, and more).
This patch fixes the problems, mostly caused by -Wsign-compare.
Signed-off-by: Zachary T Welch <zwelch@codesourcery.com>
rm/trace.c
roc.c
race.c
|
| a2ad619c24b3e9cf3edd8f3181a97022f0799fab |
07-Dec-2010 |
Zachary T Welch <zwelch@codesourcery.com> |
Remove unused variable
Eliminate an unused variable to quench the compiler's warning.
Signed-off-by: Zachary T Welch <zwelch@codesourcery.com>
race.c
|
| bfb26c745705a839f1ef5fc45e1b32135b9bc3e7 |
07-Dec-2010 |
Zachary T Welch <zwelch@codesourcery.com> |
Fix formatting specifiers for printing GElf_Addr
The use of GElf_Addr types was causing warnings when printed with %lx
format specifiers. This patch fixes the problem by using the portable
<inttypes.h> header and PRIx64 macro.
In a few places, added missing alternate form specifier ('#') to ensure
0x prefix is printed for the addresses in question.
Signed-off-by: Zachary T Welch <zwelch@codesourcery.com>
roc.c
|
| f0bd98b3e6753d8609a3054a61f2df6f9cdac10a |
09-Nov-2010 |
Joe Damato <ice799@gmail.com> |
Add support for tracing functions in libraries loaded at runtime (via libdl).
vents.c
roc.c
|
| 47cae1e251a212425cae5ac6e736de0cb0bbf714 |
09-Nov-2010 |
Joe Damato <ice799@gmail.com> |
Create a function find_dynamic_entry_addr.
This functions searches the DYNAMIC vector of an inferior process searching for
the first entry whose d_tag matches. An out parameter is provided to copy out
address stored in the matching entry.
roc.c
|
| dfa3fa3879b12ca7307b42a3feb79f42e60a90b0 |
09-Nov-2010 |
Joe Damato <ice799@gmail.com> |
Add umovebytes to move a range of bytes from the traced process to ltrace
race.c
|
| 87e1a5cc080cdab488a6c7076aa5c86dfe1c26f0 |
18-Nov-2010 |
Petr Machata <pmachata@redhat.com> |
Fix compilation issue on ppc32
pc/ptrace.h
pc/trace.c
|
| fac99be277bd370852faadc7fdd8754e2efa26f6 |
25-Oct-2010 |
Petr Machata <pmachata@redhat.com> |
Fix save_register_args and related on ppc64
pc/ptrace.h
pc/trace.c
|
| 211f0886c5aeceebacb44e3ae84d6b8f931cedf2 |
03-Nov-2010 |
Petr Machata <pmachata@redhat.com> |
Fix output of return arguments in nested calls on x86_64
86_64/ptrace.h
86_64/trace.c
|
| 41fb5a60ec30a5bcb48c624a96a55abe7cddf79a |
03-Nov-2010 |
Petr Machata <pmachata@redhat.com> |
Fix passing and returning float and double values
86_64/ptrace.h
86_64/trace.c
|
| 5117edacf42e945992def8adb9f7110d14b1fc1f |
03-Feb-2010 |
Marc Kleine-Budde <mkl@pengutronix.de> |
autotoolization: remove to-be generated file
This patch remove all files that will be autogenerated by the autoools
from the tree.
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
akefile
lpha/Makefile
rm/Makefile
386/Makefile
a64/Makefile
68k/Makefile
ipsel/Makefile
pc/Makefile
390/Makefile
parc/Makefile
86_64/Makefile
86_64/ffcheck.c
|
| 584929c4409d55c3483caca7bdf6ba85f5b20f14 |
03-Feb-2010 |
Marc Kleine-Budde <mkl@pengutronix.de> |
autotoolization: add basic autotools infrastructure
This patch add a basic autotools infrastructure to ltrace. Native
compilation on i386, x86_64 and ppc. Cross compilation has been tested
against arm.
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
akefile.am
lpha/Makefile.am
rch_syscallent.h
rm/Makefile.am
386/Makefile.am
a64/Makefile.am
68k/Makefile.am
ipsel/Makefile.am
pc/Makefile.am
390/Makefile.am
ignalent1.h
parc/Makefile.am
yscallent1.h
86_64/Makefile.am
|
| 8ce5c492d5230e41e9f9800f76ab3b3971120647 |
31-Oct-2010 |
Petr Machata <pmachata@redhat.com> |
More cleanup from Fedora:
- Implement fetching of 5th and further function arguments on s390 (contributed
by Supriya Kannery)
- Support fork/exec syscalls on 31-bit s390
- Nit
390/arch.h
390/trace.c
|
| d568f51920d5b2e0813cb5bb7e3cca10fc404623 |
08-Oct-2010 |
Zach Welch <zwelch@codesourcery.com> |
Fix ARM PLT lookups.
An ARM PLT may not contain fixed-length records, because Thumb stubs may
be added to provide interworking. These changes ensure the breakpoint
gets set on the ARM portion of PLT entries while skipping any Thumb
stubs that may be present. This patch makes the testsuite work when
built for ARMv7a and Thumb-2.
Signed-off-by: Zachary T Welch <zwelch@codesourcery.com>
rm/plt.c
|
| ee87a9a06960ba44b2ff980a366b222ae0e3b6b3 |
08-Oct-2010 |
Zachary T Welch <zwelch@codesourcery.com> |
Ensure ARM PC sets Thumb mode.
The low bit of the PC must contain the Thumb mode state.
Signed-off-by: Zachary T Welch <zwelch@codesourcery.com>
rm/regs.c
|
| b1dda4bc2c78429823a5ca7699f3207a7aea2371 |
19-Jul-2010 |
Edgar E. Iglesias <edgar.iglesias@gmail.com> |
Add support for cris-*-linux-gnu (v32) arch.
Signed-off-by: Edgar E. Iglesias <edgar@axis.com>
ris/Makefile
ris/arch.h
ris/plt.c
ris/ptrace.h
ris/regs.c
ris/signalent.h
ris/syscallent.h
ris/trace.c
|
| e280a5e51b72825ab3234c532f832cd64695bc1e |
01-Oct-2010 |
Zach Welch <zwelch@codesourcery.com> |
Identify ARM syscalls in Thumb-2 mode. Tested on IGEP v2 board.
Fixes LP:639796.
Signed-off-by: Zach Welch <zwelch@codesourcery.com>
rm/trace.c
|
| 60b3b838ef3fa5fb62845945498e4151ccc4005c |
08-Jan-2010 |
Arnaud Patard <apatard@mandriva.com> |
mipsel.h: add fp regs offset
Add fpregs offsets to be able to handle functions with fp args
Signed-off-by: Arnaud Patard <apatard@mandriva.com>
ipsel/mipsel.h
|
| 7a919b9236b06215af87eda135b71f69478b5c15 |
08-Jan-2010 |
Arnaud Patard <apatard@mandriva.com> |
x86_64/{trace.c, ptrace.h}: fix arguments handling
- handle floats
- fix arguments in function return. regs may have changed and stack
handling may (imho) become a nightmare so backup integer and fp regs
at function/syscall entry with save_register_args
- As we're now getting regs in get_arch_dep, remove other ptrace()
calls used to get registers
Signed-off-by: Arnaud Patard <apatard@mandriva.com>
86_64/ptrace.h
86_64/trace.c
|
| b2cf42cd6ef3beb471008da67e6fb681732b6586 |
08-Jan-2010 |
Arnaud Patard <apatard@mandriva.com> |
Fix mips
Makefile are using uname -m which return mips for mips and mipsel.
Easy fix: use a symlink. Better fix: git mv
Signed-off-by: Arnaud Patard <apatard@mandriva.com>
ips
|
| 8a115b6cedaf530e4589e6dc190bfb70b2c2619a |
08-Jan-2010 |
Arnaud Patard <apatard@mandriva.com> |
mips: fix gimme_arg
- handle functions with fp regs as args
- handle arguments value at return
- try to handle more than 4 (int) args
Signed-off-by: Arnaud Patard <apatard@mandriva.com>
ipsel/trace.c
|
| f3d1c535dd4d20d2fb7589ec019c4d3af24bb52a |
08-Jan-2010 |
Arnaud Patard <apatard@mandriva.com> |
Don't use PTRACE_SINGLESTEP on mips
PTRACE_SINGLESTEP is not supported on mips
Signed-off-by: Arnaud Patard <apatard@mandriva.com>
race.c
|
| 32e8e76f492a48ee3e6284e237f958b71d4d3e71 |
08-Jan-2010 |
Arnaud Patard <apatard@mandriva.com> |
arm/trace.c: syscall_p fix return value if unexpected instruction
In the case of unexpected instruction, syscall_p was returning -1 which
means errors. Unfornatunately, one may get there for valid reasons (bug
triggered by attach_process test case). Better return 0 (not a syscall)
Signed-off-by: Arnaud Patard <apatard@mandriva.com>
rm/trace.c
|
| 634a9462c212effe3e5a2cea9180fc9b89851b9d |
08-Jan-2010 |
Arnaud Patard <apatard@mandriva.com> |
plt, regs: fix includes
Fix build due to broken includes
Signed-off-by: Arnaud Patard <apatard@mandriva.com>
ipsel/plt.c
ipsel/regs.c
|
| 2f1959ffbd213e61e96fffe9979443f873418301 |
08-Jan-2010 |
Arnaud Patard <apatard@mandriva.com> |
syscallent.h/signalent.h: update
update syscallent.h/signalent.h to match newer systems
Signed-off-by: Arnaud Patard <apatard@mandriva.com>
ipsel/signalent.h
ipsel/syscallent.h
|
| f16fcff75f62d44a49ad9c020622956bd2b8df45 |
08-Jan-2010 |
Arnaud Patard <apatard@mandriva.com> |
trace.c: fix (partially) umovelong
umovelong was always returning the full value read in ptrace'd process
memory (eg long) even if the type was not a long (eg int).
This patch add handling for at least int case.
[ found out with parameters test case on x86_64 as int is 32bit and long
64bit ]
Signed-off-by: Arnaud Patard <apatard@mandriva.com>
race.c
|
| 0aac3525a48a5485ee82f966e7a0af40420b99bf |
08-Jan-2010 |
Arnaud Patard <apatard@mandriva.com> |
Add a script to create syscallent.h for mips
Add a modified version of mksyscallent which works on mips o32
Signed-off-by: Arnaud Patard <apatard@mandriva.com>
ksyscallent_mips
|
| a7af00db2231e99a4506e4f5587f9dd00b9d1175 |
26-Jul-2009 |
Juan Cespedes <cespedes@thehackers.org> |
Fixed compilation problems in armel, ia64 and powerpc
rm/breakpoint.c
reakpoint.c
a64/breakpoint.c
a64/regs.c
pc/plt.c
|
| 3458456ffbf0b6d85e1818dea7cbeacc8b1d2298 |
25-Jul-2009 |
Juan Cespedes <cespedes@coco.thehackers.org> |
Make it work again in sparc
akefile
parc/ptrace.h
parc/regs.c
parc/trace.c
|
| f670eea50e959eeb9da53d70cad8d43c19494ef0 |
25-Jul-2009 |
Juan Cespedes <cespedes@coco.thehackers.org> |
Finally got rid of GNU's Autoconf stuff
lpha/regs.c
lpha/trace.c
rm/regs.c
rm/trace.c
reakpoint.c
vents.c
386/regs.c
386/trace.c
a64/regs.c
a64/trace.c
68k/regs.c
68k/trace.c
ipsel/regs.c
ipsel/trace.c
pc/regs.c
pc/trace.c
roc.c
390/regs.c
390/trace.c
parc/regs.c
parc/trace.c
86_64/regs.c
86_64/trace.c
|
| 427b7811598f905b583ac86b35c72228fd415476 |
06-Jul-2009 |
Juan Cespedes <cespedes@coco.thehackers.org> |
Indentation, comments
vents.c
|
| 2a61d19bd244dadcde5009f1632cf14b95623e3d |
04-Jul-2009 |
Juan Cespedes <cespedes@coco.thehackers.org> |
Added set_return_addr() to sysdeps/linux-gnu/*/regs.c
lpha/plt.c
lpha/regs.c
rm/plt.c
rm/regs.c
reakpoint.c
386/plt.c
386/regs.c
a64/plt.c
a64/regs.c
68k/plt.c
68k/regs.c
ipsel/plt.c
ipsel/regs.c
pc/plt.c
pc/regs.c
390/plt.c
390/regs.c
parc/plt.c
parc/regs.c
86_64/plt.c
86_64/regs.c
|
| 61da33723c5fb09762e38bd39a26ee15d62ffebc |
03-Jul-2009 |
Juan Cespedes <cespedes@coco.thehackers.org> |
Added different callback handlers for each event
reakpoint.c
vents.c
pc/regs.c
|
| 8d1b92ba755f6d6229f5e230fc43d958b13836f8 |
03-Jul-2009 |
Juan Cespedes <cespedes@coco.thehackers.org> |
Some minor fixes
reakpoint.c
vents.c
race.c
|
| 03192f890af95bf9abf5ea1c3dc0998f6b12550d |
03-Jul-2009 |
Juan Cespedes <cespedes@coco.thehackers.org> |
fixed configure.ac, renamed process_event -> handle_event
ipsel/regs.c
|
| f728123bd75a65a6a1536e198c3c30719e494e71 |
25-Jun-2009 |
Juan Cespedes <cespedes@debian.org> |
Re-organize file names
lpha/plt.c
lpha/regs.c
lpha/trace.c
rm/plt.c
rm/regs.c
rm/trace.c
vents.c
386/plt.c
386/regs.c
386/trace.c
a64/plt.c
a64/regs.c
a64/trace.c
68k/plt.c
68k/regs.c
68k/trace.c
ipsel/plt.c
ipsel/regs.c
ipsel/trace.c
pc/plt.c
pc/regs.c
pc/trace.c
390/plt.c
390/regs.c
390/trace.c
parc/plt.c
parc/ptrace.h
parc/regs.c
parc/trace.c
race.c
86_64/plt.c
86_64/regs.c
86_64/trace.c
|
| 3df476b28e4a9cdb43cf29fff8e89481310eb30d |
28-May-2009 |
Juan Cespedes <cespedes@debian.org> |
Created library "libltrace.a" and a simple program that calls it
lpha/plt.c
lpha/regs.c
lpha/trace.c
rm/plt.c
rm/regs.c
rm/trace.c
vents.c
386/plt.c
386/regs.c
386/trace.c
a64/plt.c
a64/regs.c
a64/trace.c
68k/plt.c
68k/regs.c
68k/trace.c
ipsel/plt.c
ipsel/regs.c
ipsel/trace.c
pc/plt.c
pc/regs.c
pc/trace.c
390/plt.c
390/regs.c
390/trace.c
parc/plt.c
parc/ptrace.h
parc/regs.c
parc/trace.c
race.c
86_64/plt.c
86_64/regs.c
86_64/trace.c
|
| 6be8028492b29f68126ce413b7ef4b875c7bb017 |
28-May-2009 |
Juan Cespedes <cespedes@debian.org> |
wait for clone()s as well as for fork()s
vents.c
|
| fea4a1285dc7bea02b0203845ad43117f091fc3e |
28-May-2009 |
Juan Cespedes <cespedes@debian.org> |
misc typos
a64/plt.c
a64/trace.c
ipsel/plt.c
ipsel/regs.c
ipsel/trace.c
race.c
|
| 3e94cbf0d388a5b5b49f8dbc60f4b24900de89ae |
22-May-2009 |
Juan Cespedes <cespedes@coco.thehackers.org> |
See syscalls as syscalls and not sysrets when we are a clone
lpha/trace.c
386/trace.c
a64/trace.c
68k/trace.c
ipsel/trace.c
pc/trace.c
parc/trace.c
86_64/trace.c
|
| 30439b42294d678f2066522aecdb664132170a41 |
22-May-2009 |
Juan Cespedes <cespedes@coco.thehackers.org> |
Make clone() work when child starts after parent finishes
386/trace.c
|
| e0660df7f943cc832694db3b8b6778c0ff65860d |
21-May-2009 |
Juan Cespedes <cespedes@debian.org> |
Make exec() work again using PTRACE_O_TRACEEXEC
vents.c
roc.c
race.c
|
| 2721e6adbc7399e1b621b896b4f5b864082fbf03 |
21-May-2009 |
Juan Cespedes <cespedes@debian.org> |
new fork() and clone() approach
used PTRACE_O_TRACE{FORK,VFORK,CLONE} to trace newly created
processes instead of figuring it out whether a given syscall
would create another process or not
vents.c
race.c
|
| 85f7d7616f2736de79f6b7f77bceedbc97c83406 |
14-May-2009 |
Juan Cespedes <cespedes@debian.org> |
updated man page
race.c
|
| cd8976dbee947f152c3a322503a1063c6359da76 |
14-May-2009 |
Juan Cespedes <cespedes@debian.org> |
better debugging support (-d option)
reakpoint.c
vents.c
race.c
|
| 8f6d1ecd5f4301f899927a553572c5089fd29bcf |
07-May-2009 |
Juan Cespedes <cespedes@debian.org> |
New event: EVENT_NEW (received from a new process)
vents.c
|
| 393f1d00c582058078cc238547f65202c438fb19 |
07-May-2009 |
Juan Cespedes <cespedes@debian.org> |
struct event -> Event
vents.c
|
| cde582642f959f69d436d435b504f50470af5db9 |
07-May-2009 |
Juan Cespedes <cespedes@debian.org> |
struct function -> Function
vents.c
|
| 1dec217e47f998c03c642561d98753c32683985c |
07-May-2009 |
Juan Cespedes <cespedes@debian.org> |
struct breakpoint -> Breakpoint
rm/breakpoint.c
reakpoint.c
a64/breakpoint.c
race.c
|
| a8909f71e1421949c960f287217be6c42c286c0f |
28-Apr-2009 |
Juan Cespedes <cespedes@coco.thehackers.org> |
struct process -> Process
lpha/plt.c
lpha/regs.c
lpha/trace.c
rm/plt.c
rm/regs.c
rm/trace.c
vents.c
386/plt.c
386/regs.c
386/trace.c
a64/plt.c
a64/regs.c
a64/trace.c
68k/plt.c
68k/regs.c
68k/trace.c
ipsel/plt.c
ipsel/regs.c
ipsel/trace.c
pc/plt.c
pc/regs.c
pc/trace.c
390/plt.c
390/regs.c
390/trace.c
parc/plt.c
parc/regs.c
parc/trace.c
race.c
86_64/plt.c
86_64/regs.c
86_64/trace.c
|
| 1e583132c95aa211fab9e0e0869b237adfa78304 |
07-Apr-2009 |
Juan Cespedes <cespedes@debian.org> |
Beginning support of PTRACE_O_TRACE* (doesn't work yet)
vents.c
race.c
|
| da9b953284966b8c52de43713ebb5b503756e468 |
07-Apr-2009 |
Juan Cespedes <cespedes@debian.org> |
More fields to struct "options"
(opt_c, opt_d, opt_A)
pc/plt.c
|
| 714ee9df1696e696d5bd2e9d95c13e9f5051eb73 |
07-Apr-2009 |
Juan Cespedes <cespedes@debian.org> |
wait for new children to stop before assuming they have stopped
race.c
|
| e2023f721b66db48cc8a0f4ab6519460093c2cf1 |
07-Apr-2009 |
Juan Cespedes <cespedes@debian.org> |
wait_for_something.c -> sysdeps/linux-gnu/events.c
akefile
vents.c
ipsel/trace.c
|
| e74c80de19a628cab68f10a1e7c92bef2988aafb |
11-Feb-2009 |
Juan Cespedes <cespedes@coco.thehackers.org> |
Fixed segfault in traced process when receiving a signal
Specifically, when a signal was received while enabling a breakpoint
race.c
|
| b65bdc581f1ff747a53a1e2494d1cebb56453cb2 |
16-Dec-2008 |
Juan Cespedes <cespedes@debian.org> |
More fields to struct "options"
Substituted opt_* with options.indent, options.output
parc/trace.c
|
| f13505251e6402460f6cc7ec84e0d8ca91607b4f |
16-Dec-2008 |
Juan Cespedes <cespedes@debian.org> |
Re-indent entry points to functions
lpha/plt.c
lpha/regs.c
lpha/trace.c
rm/breakpoint.c
rm/plt.c
rm/regs.c
rm/trace.c
reakpoint.c
386/plt.c
386/regs.c
386/trace.c
a64/breakpoint.c
a64/plt.c
a64/regs.c
a64/trace.c
68k/plt.c
68k/regs.c
68k/trace.c
ipsel/plt.c
ipsel/regs.c
ipsel/trace.c
pc/plt.c
pc/regs.c
pc/trace.c
roc.c
390/plt.c
390/regs.c
390/trace.c
parc/plt.c
parc/regs.c
parc/trace.c
race.c
86_64/plt.c
86_64/regs.c
86_64/trace.c
|
| 63184be8c577f5799e44db2a4e312a8240ad7751 |
10-Dec-2008 |
Juan Cespedes <cespedes@debian.org> |
Patches from Anderson Lizardo and Riku Voipio to reach Debian version 0.5-3.1
* Add generic support for arm targets
* Save funtion arguments on arm
* Add thumb instruction support
* Add basic arm/eabi support
* fix exec() testcase cleanup
* fix memory corruption in clone() test
* fix tracing child with "-p" option
akefile
rch_mksyscallent
rm/Makefile
rm/arch.h
rm/arch_syscallent.h
rm/breakpoint.c
rm/ptrace.h
rm/regs.c
rm/trace.c
|
| 55c5feba87bbb8d9190a71ec5ddc38380a384bf7 |
12-Mar-2008 |
Luis Machado <luisgpm@br.ibm.com> |
PPC argument handling improvements
* sysdeps/linux-gnu/ppc/trace.c (arch_umovelong): New function.
* sysdeps/linux-gnu/ppc/regs.c (get_instruction): New function.
(get_count_register): New function.
* sysdeps/linux-gnu/ppc/arch.h (ARCH_HAVE_UMOVELONG): New define.
* sysdeps/linux-gnu/trace.c (umovelong): Create arch-specific
variant.
* ltrace.h (umovelong): Change prototype.
* process_event.c (process_breakpoint): Handle specifics of ppc3
* display_args.c: Call umovelong with info parameter.
pc/arch.h
pc/regs.c
pc/trace.c
race.c
|
| a413e5b8880de643a83ad124d078091c0956fe1d |
04-Sep-2007 |
Juan Cespedes <cespedes@debian.org> |
Take arg_num out of arg_type_info
lpha/trace.c
rm/trace.c
386/trace.c
a64/trace.c
68k/trace.c
ipsel/trace.c
pc/trace.c
390/trace.c
parc/trace.c
86_64/trace.c
|
| aee093126654f722523b47848a0c5449e39cf4bb |
31-Aug-2007 |
Juan Cespedes <cespedes@debian.org> |
Some random fixes
* General: Small fixes (indentation, typos, clean-up of code)
* ltrace.c: Close output file on exit
* ltrace.c: use getenf("HOME") instead of getpwuid(geteuid())->pw_dir
* read_config_file.c, display_args.c: remove "ignore" argtype;
that's what "void" is for
* packaging/debian/: misc fixes, sync with version 0.5-1
* etc/ltrace.conf: added more system calls
* testsuite/ltrace.minor/trace-clone.c: sleep(1) to avoid earlier
termination of process
* sysdeps/linux-gnu/trace.c: trace_pid(): reverted Petr's patch
to wait for child to stop, as it stopped following clone()
* process_event.c: Disable breakpoints before doing fork() (again!),
to make children work as expected
race.c
|
| 55ed83b24df9c6d671091a8c75caab33ffecd40e |
17-May-2007 |
Petr Machata <pmachata@redhat.com> |
Fixes from Supriya Kannery <supriyak@in.ibm.com>
* wait_for_something.c, process_event.c: Tracing across exec.
* sysdeps/linux-gnu/trace.c, ltrace.h: New interface was_exec.
* testsuite/ltrace.minor/trace-exec.c,
testsuite/ltrace.minor/trace-exec.exp,
testsuite/ltrace.minor/trace-exec1.c: Testcase for same.
race.c
|
| 89a536028ab3b699b7f953b6b0fd7607917fc303 |
25-Jan-2007 |
Petr Machata <pmachata@redhat.com> |
Fix -p and -L behavior
* wait for process to stop after PTRACE_ATTACH
* always initialize the breakpoint dictionary,
and thus avoid NULL dereference.
race.c
|
| 1228a91e6560c5e5ac4fdd051adc9b34bf9fc047 |
28-Dec-2006 |
Eric Vaitl <evaitl@cisco.com> |
Added mipsel support
ipsel/Doxyfile
ipsel/Makefile
ipsel/arch.h
ipsel/mipsel.h
ipsel/plt.c
ipsel/ptrace.h
ipsel/regs.c
ipsel/signalent.h
ipsel/syscallent.h
ipsel/trace.c
|
| b3f8fef12fccb0914b7b28725f42192c279d31c9 |
30-Nov-2006 |
Petr Machata <pmachata@redhat.com> |
Fix .opd handling on IA64, few cleanups.
a64/arch.h
pc/arch.h
|
| 65b53df7fa2577c4138aef86c115873eab684a0a |
25-Sep-2006 |
Steve Fink <sphink@gmail.com> |
Large IA64 fix...
* display_args.c: store arg_num in arg_type_info
* display_args.c: support 'double' parameters
* display_args.c: fix implementation of float,double params for ia64
* output.c, process_event.c: store arg_num in arg_type_info
* read_config_file.c: support 'double' parameters
* read_config_file.c: store arg_num in arg_type_info, and as a result,
stop using singleton objects for any of the arg_type_info's.
* read_config_file.c: improve support for struct field alignments
* read_config_file.c: count floating-point parameters to support ia64
float parameter passing
* sysdeps/README, sysdeps/linux-gnu/*/trace.c: pass in the full
arg_type_info to gimme_arg rather than just the arg_num (necessary
for float params on some architectures)
* sysdeps/linux-gnu/ia64/trace.c: accommodate register renaming when
fetching the parameters of a function after it has returned
* sysdeps/linux-gnu/ia64/trace.c: support floating point parameters
lpha/trace.c
rm/trace.c
386/trace.c
a64/trace.c
68k/trace.c
pc/trace.c
390/trace.c
parc/trace.c
86_64/trace.c
|
| a841f6592e75d78620ee9e645af22a58c9e3bbf9 |
15-Sep-2006 |
Olaf Hering <olh@suse.de> |
Some fixes...
* Makefile.in : allow installation as non-root user, print out
some debugging information before running test suite.
* summary.c : allow compilation without USE_DEMANGLE
* sysdeps/linux-gnu/ppc/plt.c : fix warning in sym2addr
* sysdeps/linux-gnu/ia64/regs.c : fix warning when finding
instruction slot
* elf.c : fix up error created in 2006-07-26 refactor
a64/regs.c
pc/plt.c
|
| 7bafff09cc66e23519512a54e2d1ebd3664a1a70 |
07-Aug-2006 |
Steve Fink <sphink@gmail.com> |
pass values instead of argument numbers to fetch
race.c
|
| 7003feea8035a3c4ee661dc2556d158e9ba41d39 |
18-Jul-2006 |
Petr Machata <pmachata@redhat.com> |
Compile-time fixes for ppc32.
pc/arch.h
|
| c53309ef99c205a2e9ae7ccbdf6557adcd52d24e |
19-Jun-2006 |
Ian Wienand <ianw@ieee.org> |
Version 0.5
update syscalls to 2.6.17
rm/signalent.h
rm/syscallent.h
386/signalent.h
386/syscallent.h
a64/signalent.h
a64/syscallent.h
68k/signalent.h
68k/syscallent.h
ksyscallent
390/syscalls31.h
390/syscalls64.h
|
| 76c61f15d7989bf7adffed2e46a44c34a80bd927 |
14-Jun-2006 |
Paul Gilliam <pgilliam@us.ibm.com> |
Added Paul Gilliam's secure PLT patch
lpha/plt.c
rm/plt.c
reakpoint.c
386/plt.c
a64/plt.c
68k/plt.c
pc/arch.h
pc/plt.c
390/plt.c
parc/plt.c
86_64/plt.c
|
| dcbfb8ee67323014098df9fb5a535d1099e2e2ec |
11-May-2006 |
Heiko Carstens <heiko.carstens@de.ibm.com> |
update s390 syscalls, add script
EADME
ksyscallent_s390
390/syscalls31.h
390/syscalls64.h
|
| be3207721cb07e7d01a81fb3beecd0327201478c |
24-Apr-2006 |
Paul Gilliam <pgilliam@us.ibm.com> |
Use e_entry if "_start" can't be found on architectures where it is needed
lpha/arch.h
rm/arch.h
386/arch.h
a64/arch.h
68k/arch.h
pc/arch.h
390/arch.h
parc/arch.h
86_64/arch.h
|
| 3f1219f7803599df8556e7fc096a6a239f9a04fd |
24-Apr-2006 |
Paul Gilliam <pgilliam@us.ibm.com> |
Adds some casts that keeps a more picky version of GCC happy.
reakpoint.c
race.c
|
| 2d45b1a8e26a36a9f85dc49e721c4390ca93dc40 |
20-Feb-2006 |
Ian Wienand <ianw@ieee.org> |
run Lindent over source to get everything looking about the same
lpha/plt.c
lpha/regs.c
lpha/signalent.h
lpha/syscallent.h
lpha/trace.c
rm/plt.c
rm/regs.c
rm/signalent.h
rm/syscallent.h
rm/trace.c
reakpoint.c
386/plt.c
386/regs.c
386/signalent.h
386/syscallent.h
386/trace.c
a64/breakpoint.c
a64/plt.c
a64/regs.c
a64/signalent.h
a64/syscallent.h
a64/trace.c
68k/plt.c
68k/regs.c
68k/signalent.h
68k/syscallent.h
68k/trace.c
pc/plt.c
pc/regs.c
pc/signalent.h
pc/syscallent.h
pc/trace.c
roc.c
390/plt.c
390/regs.c
390/signalent.h
390/syscalls31.h
390/syscalls64.h
390/trace.c
parc/plt.c
parc/regs.c
parc/signalent.h
parc/syscallent.h
parc/trace.c
race.c
86_64/plt.c
86_64/regs.c
86_64/signalent.h
86_64/syscallent.h
86_64/trace.c
|
| 9a2ad351a1c3215dc596ff3e2e3fd4bc24445a6b |
20-Feb-2006 |
Ian Wienand <ianw@ieee.org> |
Switched to SVN (initial import)
akefile
lpha/arch.h
lpha/plt.c
lpha/regs.c
lpha/trace.c
rm/arch.h
rm/plt.c
rm/regs.c
rm/trace.c
reakpoint.c
386/arch.h
386/plt.c
386/regs.c
386/syscallent.h
386/trace.c
a64/arch.h
a64/plt.c
a64/syscallent.h
a64/trace.c
68k/arch.h
68k/plt.c
68k/regs.c
68k/trace.c
pc/arch.h
pc/plt.c
pc/regs.c
pc/syscallent.h
pc/trace.c
roc.c
390/arch.h
390/plt.c
390/regs.c
390/signalent.h
390/signalent1.h
390/syscallent.h
390/syscallent1.h
390/syscalls31.h
390/syscalls64.h
390/trace.c
parc/arch.h
parc/plt.c
parc/regs.c
parc/syscallent.h
parc/trace.c
race.c
86_64/arch.h
86_64/plt.c
86_64/regs.c
86_64/signalent1.h
86_64/syscallent.h
86_64/syscallent1.h
86_64/trace.c
|
| 5570a7769869a4df25ef85f302f74a7feb6c0cd3 |
17-Feb-2006 |
Ian Wienand <ianw@ieee.org> |
Version 0.3.38-1
* Added ia64 port
* elf.h: add extra field for PLT size
* elf.c: put in PLT size
* sysdeps/linux-gnu/breakpoint.c: add arch breakpoint override
* sysdeps/linux-gnu/trace.c: don't single step after breakpoin for ia64
* configure.ac: add version to AC_INIT, bump version to 0.3.38
* options.c: use PACKAGE_VERSION
reakpoint.c
a64/Makefile
a64/arch.h
a64/breakpoint.c
a64/plt.c
a64/ptrace.h
a64/regs.c
a64/signalent.h
a64/syscallent.h
a64/trace.c
race.c
|
| 3219f320604810532a4938dda8f9dfadb0e840f3 |
16-Feb-2006 |
Ian Wienand <ianw@ieee.org> |
Some fixes...
* read_config_file.c: initialise pt stack argument to stop warning
* summary.c: make show_summary() obey -C for demangaling function names
lpha/plt.c
lpha/regs.c
lpha/trace.c
rm/plt.c
rm/regs.c
rm/trace.c
reakpoint.c
386/plt.c
386/regs.c
386/trace.c
68k/plt.c
68k/regs.c
68k/trace.c
pc/plt.c
pc/regs.c
pc/trace.c
roc.c
390/plt.c
390/regs.c
390/trace.c
parc/plt.c
parc/regs.c
parc/trace.c
race.c
86_64/plt.c
86_64/regs.c
86_64/trace.c
|
| d914a206a11cc1011a45f00674b1e16988fae77f |
10-Nov-2004 |
Juan Cespedes <cespedes@debian.org> |
Version: 0.3.36
* Fixed "--indent" option
* Got rid of Debian and RedHat specific parts in tarball
akefile
lpha/Makefile
lpha/arch.h
lpha/plt.c
lpha/syscallent.h
rm/Makefile
rm/arch.h
rm/plt.c
386/Makefile
386/arch.h
386/plt.c
386/syscallent.h
68k/Makefile
68k/arch.h
68k/plt.c
pc/Makefile
pc/arch.h
pc/plt.c
pc/syscallent.h
390/Makefile
390/arch.h
390/plt.c
390/syscallent.h
parc/Makefile
parc/arch.h
parc/plt.c
parc/syscallent.h
86_64/Makefile
86_64/arch.h
86_64/ffcheck.c
86_64/plt.c
86_64/syscallent.h
|
| 5464e8d9900fd098f197e58df4a5d99c1217d20c |
16-Jul-2004 |
Juan Cespedes <cespedes@debian.org> |
Version 0.3.35
* Fixed include line in m68k, caused build problems
68k/regs.c
|
| 948a5d077939939dc086c0d0d199ff2f22706eea |
07-Jul-2004 |
Juan Cespedes <cespedes@debian.org> |
Version 0.3.34
* Fixed prototype declaration problem in arm, m68k, powerpc, s390
* Added "amd64" to list of architectures (closes: Bug#252756)
* Sparc port is hopefully working (closes: Bug#35524)
rm/regs.c
68k/regs.c
pc/regs.c
390/regs.c
parc/regs.c
86_64/regs.c
|
| 5c3fe0697b202cc7d95e90459de0fb312b297b27 |
14-Jun-2004 |
Juan Cespedes <cespedes@debian.org> |
Version 0.3.33
* Fixed two bugs, thanks to Mauro Meneghin <G1ld0@lycos.it>:
+ Cope correctly with breakpoint values greater than
sizeof(long) bytes
+ Fixed small bug in option -r (closes: Bug#212792)
* Show help if no (or few) arguments are given, just like
strace and fenris (thanks, Tomasz Wegrzanowski <taw@users.sf.net>)
* Some fixes from Jakub Bogusz <qboosh@pld-linux.org>:
+ Small 64-bit cleanup of code
+ support for more than 6 function arguments on amd64
+ Adapted SPARC port from Jakub Jelinek <jakub@redhat.com>
+ Added alpha support
lpha/Makefile
lpha/arch.h
lpha/ptrace.h
lpha/regs.c
lpha/signalent.h
lpha/syscallent.h
lpha/trace.c
rm/ptrace.h
rm/regs.c
rm/trace.c
reakpoint.c
386/ptrace.h
386/regs.c
386/trace.c
68k/ptrace.h
68k/regs.c
68k/trace.c
pc/ptrace.h
pc/regs.c
pc/trace.c
390/ptrace.h
390/regs.c
390/trace.c
parc/Makefile
parc/arch.h
parc/ptrace.h
parc/regs.c
parc/signalent.h
parc/syscallent.h
parc/trace.c
race.c
86_64/ptrace.h
86_64/regs.c
86_64/trace.c
|
| efe85f0668a077b1e851df4b3f87a380cf2269fd |
04-Apr-2004 |
Juan Cespedes <cespedes@debian.org> |
Version 0.3.32
* Fixed wrong version number
* Removed unused file "opt_c.c"
* Remove error when tracing no calls and doing fork()
* Clean-up of sysdeps/linux-gnu/s390/trace.c
* Clean-up of sysdeps/linux-gnu/ppc/trace.c
* Make `--library' option really work (closes: Bug#232321)
* Merged several patches from SuSE:
+ Added some functions to ltrace.conf
+ Handle 64-big ELF files nicely
+ AMD64 support
+ Updated list of syscalls for S/390
+ Improved some debugging statements
Many thanks to Bernhard Kaindl <bk@suse.de> for his great work
rm/arch.h
reakpoint.c
pc/arch.o
pc/regs.o
pc/trace.c
pc/trace.o
390/Makefile
390/regs.c
390/syscallent.h
390/trace.c
86_64/Makefile
86_64/arch.h
86_64/ffcheck.c
86_64/regs.c
86_64/signalent.h
86_64/syscallent.h
86_64/trace.c
|
| 504a385858a49352bcfceca444ba4f1a7bfd20cd |
04-Feb-2003 |
Juan Cespedes <cespedes@debian.org> |
Version 0.3.31
* Added a lot of functions to ltrace.conf,
thanks to Jakub Jelinek <jakub@redhat.com> (closes: Bug#144518)
* Fixed off-by-one problem in checking syscall number
* Removed some warnings
386/regs.c
386/trace.c
race.c
|
| a0ccf39a68c0fcdf2165bde0f9b70ed12fc61cd8 |
01-Feb-2003 |
Juan Cespedes <cespedes@debian.org> |
Version 0.3.29
* Align return values depending on screen width
* Updated list of syscalls and signals to Linux 2.4.20
* Fixed bug introduced in 0.3.27 which caused -L option to segfault
rm/syscallent.h
386/syscallent.h
68k/syscallent.h
pc/syscallent.h
390/syscallent.h
|
| 5bfb061c4c7cfac6e1882a586cf36c849d95fcea |
31-Mar-2002 |
Juan Cespedes <cespedes@debian.org> |
Version 0.3.25
* Finally added powerpc support (Anton Blanchard <anton@samba.org>)
pc/arch.h
pc/regs.c
pc/syscallent.h
pc/trace.c
|
| f1bfe203f5f1c0e11a614f9d593a68406f5cb47e |
27-Mar-2002 |
Juan Cespedes <cespedes@debian.org> |
Version 0.3.24
* Fixed 2 minor buffer overflows (closes: Bug#130746)
* Obey --prefix, --sysconfdir, --mandir options in configure
* Adding powerpc support (doesn't work yet)
(Anton Blanchard <anton@linuxcare.com.au>)
pc/Makefile
pc/arch.h
pc/arch.o
pc/regs.c
pc/regs.o
pc/signalent.h
pc/syscallent.h
pc/trace.c
pc/trace.o
|
| 8f8282f72eaeadc5419cd5470100e8dcaba5b7fd |
03-Mar-2002 |
Juan Cespedes <cespedes@debian.org> |
Version 0.3.23
* Fixed missing include <unistd.h> in trace.c
* One arch-dependent function less (continue_after_breakpoint)
* Fixed S/390 port (it didn't compile yet...)
rm/regs.c
rm/trace.c
386/regs.c
386/trace.c
68k/regs.c
68k/trace.c
390/regs.c
390/trace.c
race.c
|
| 96c3fd0c58f7f57a0d8b317ba9b43dffc03a18a1 |
03-Mar-2002 |
Juan Cespedes <cespedes@debian.org> |
Version 0.3.22
* S/390: Removed extra target in sysdeps/linux-gnu/s390 which avoided
compiling...
390/Makefile
|
| a4850831bf4b4548af517335081b58cf006fd168 |
03-Mar-2002 |
Juan Cespedes <cespedes@debian.org> |
Version 0.3.21
* Get rid of arch/breakpoint.c; we can do it arch-independent
akefile
rm/Makefile
rm/arch.h
rm/breakpoint.c
reakpoint.c
386/Makefile
386/breakpoint.c
68k/Makefile
68k/arch.h
68k/breakpoint.c
390/arch.h
390/breakpoint.c
|
| b1dd77d66f93eeaa6cbf8b101d3e9cf70981e000 |
03-Mar-2002 |
Juan Cespedes <cespedes@debian.org> |
Version 0.3.20
* Added s390 port (Timothy R. Fesig <slate@us.ibm.com>)
* Modified configure process to use ltrace.spec.in to generate
ltrace.spec (Timothy R. Fesig <slate@us.ibm.com>)
* Fixed some problems using ltrace.spec on Intel platform.
(Timothy R. Fesig <slate@us.ibm.com>)
akefile
rm/trace.c
386/trace.c
68k/trace.c
390/Makefile
390/arch.h
390/breakpoint.c
390/regs.c
390/signalent.h
390/syscallent.h
390/trace.c
|
| 8cc1b9d456c8f41e3a8cc0d3ec0d77eb779cb7bb |
01-Mar-2002 |
Juan Cespedes <cespedes@debian.org> |
Version: 0.3.18
* Simplified arch-dependent stuff
* Updated list of syscalls and signals to Linux 2.4.18
* Unified coding-style of all function declarations
* Do not indent lines indicating signals, exit codes, etc
* Updated description
* fix off-by-one problem in checking syscall number (Tim Waugh
<twaugh@redhat.com> fixed this problem in RedHat two years ago;
thank you for NOT noticing me...)
rm/breakpoint.c
rm/regs.c
rm/syscallent.h
rm/trace.c
386/regs.c
386/syscallent.h
386/trace.c
68k/breakpoint.c
68k/regs.c
68k/trace.c
roc.c
race.c
|
| 5916fda0d07da90cac8a78cbfa73374d81b150be |
25-Feb-2002 |
Juan Cespedes <cespedes@debian.org> |
Version: 0.3.17
* Added a bit more debugging
* Fixed display of return address in nested functions
* Added posibility to exit from a function different from the last
* called
one (this fixes "ltrace gnome-terminal", for example)
386/breakpoint.c
race.c
|
| 21c63a1469bc049d93444b8a15afd685ee251392 |
07-Jul-2001 |
Juan Cespedes <cespedes@debian.org> |
Version: 0.3.14
* Assume a syscall is always immediatly followed by a sysret
in i386 (fixes bug which prevented ltrace to work properly
in any program using signals); I will have to rethink all
this and fix it correctly or port it to non-i386 archs
* Fixed -n option: now it is done in output.c (this still has
problems when tracing more than one process at a time)
386/breakpoint.c
386/trace.c
|
| 1cd999a32728f41208dc30f38a125d7bb7063625 |
03-Jul-2001 |
Juan Cespedes <cespedes@debian.org> |
Version: 0.3.12
* Re-wrote of "elf.c" (Silvio Cesare <silvio@big.net.au>)
* Added "--library" option (Silvio)
* Updated list of syscalls and signals to Linux 2.4.5
* Compile cleanly with gcc-3.0 (thanks to Frederic L. W. Meunier)
rm/signalent.h
rm/syscallent.h
386/signalent.h
386/syscallent.h
68k/signalent.h
68k/syscallent.h
ksyscallent
|
| 5b3ffdf2e696273d38434ff7b3c26349fff5a0ea |
02-Jul-2001 |
Juan Cespedes <cespedes@debian.org> |
Version: 0.3.11
* Clean up lintian bugs
* Fixed small bug reading start of arguments in config file
* Keep a stack of nested calls (Morten Eriksen, 1999-07-04)
* Add "--indent" option (Morten Eriksen, 1999-07-04)
* cleans up connection between a breakpoint address and
a call instance (Morten Eriksen, 1999-07-04)
* New Standards-Version (3.5.5)
rm/breakpoint.c
rm/trace.c
386/breakpoint.c
386/trace.c
68k/breakpoint.c
68k/trace.c
race.c
|
| 1b9cfd6ad305ad909e8ff17139111a7c78f01464 |
30-Aug-1999 |
Juan Cespedes <cespedes@debian.org> |
Version: 0.3.8
* glibc-2.1 does no longer need `_GNU_SOURCE' defined to use <getopt.h>
* Changed description of package; adopted Red Hat's one
(thanks to whoever wrote it)
* Close all the file descriptors used before executing program (close-on-exec)
* Updated copyright file for new location /usr/share/common-licenses/GPL.
* Used GNU autoconf instead of "uname" to guess host system type
* Install man page in /usr/share/man instead of /usr/man
* Added a few functions to /etc/ltrace.conf
* Updated list of syscalls and signals to linux-2.2.12
* Fixed bugs in C++ demangle (Morten Eriksen <mortene@sim.no>)
* New Standards-Version: 3.0.1 (but keeping docs in /usr/doc)
akefile
EADME
rm/Makefile
rm/arch.h
rm/breakpoint.c
rm/regs.c
rm/signalent.h
rm/syscallent.h
rm/trace.c
386/Makefile
386/arch.h
386/breakpoint.c
386/regs.c
386/signalent.h
386/syscallent.h
386/trace.c
68k/Makefile
68k/arch.h
68k/breakpoint.c
68k/regs.c
68k/signalent.h
68k/syscallent.h
68k/trace.c
ksignalent
ksyscallent
roc.c
race.c
|