1
0
Fork 0
mirror of https://github.com/tobast/libunwind-eh_elf.git synced 2024-12-23 12:03:41 +01:00
Commit graph

3118 commits

Author SHA1 Message Date
Felipe Cerqueira
152a96e291 core: Fix memory leak in libunwind-coredump.
_UCD_destroy() was not freeing all variables from UCD_info.
2017-08-14 18:39:37 -07:00
Stephen Chen
0314ff8522 aarch64: Use PTRACE_GETREGSET if available
In remote ptrace mode, we currently use PTRACE_PEEKUSER to read the registers.

PTRACE_PEEKUSER only works on x86 or arm 32 bit compatibility mode on linux. On aarch64 system,
it always return -EIO. https://github.com/torvalds/linux/blob/master/kernel/ptrace.c#L885-L1102

PTRACE_GETREGSET is the newer and more supported way of reading registers. Use that if it's available.
2017-08-07 09:33:23 -07:00
Paul Pluzhnikov
cc0c170f53 tests: Fix heap overflow in coredump-unwind test
We've tried to run slightly modified test-coredump-unwind.c built with
tcmalloc, and it promptly crashed.  Attached patch fixes the heap buffer
overflow bug which caused it.
2017-08-07 09:07:20 -07:00
Doug Moore
23c3f24ea6 On aarch64, it is the X30 register, not the PC register, that is associated with the program counter. 2017-08-07 09:05:34 -07:00
Doug Moore
6150f2f426 Better imitate the choice of which register to identify as ip
for tilegx.
2017-08-07 09:05:27 -07:00
Doug Moore
e360c3ddbd For architectures that initialize the dwarf ip cache, but do not
update when the ip register is set, add those updates.
2017-08-07 09:05:17 -07:00
Doug Moore
93f319466f ppc: Add register state and iterate functions
Add register state and state iterate functions to ppc32 and ppc64
directories.  They were added to other arch directories in change
502ba27753... but not these, for some reason.
2017-07-25 11:33:23 -07:00
Sergei Trofimovich
85ce08b64a ia64: : disable dwarf-specific code (#35)
Commit dbce594d33
added unconditional dwarf usage into unw_get_proc_name.
Unfortunately ia64 is the only architecture that
does not support it in libunwind (configure.ac):

```
if test x$target_arch != xia64; then
  use_dwarf=yes
else
  use_dwarf=no
fi
```

As a result build fails on ia64 as:

```
ia64-unknown-linux-gnu-gcc ... -c mi/Lget_proc_name.c ...
In file included from mi/Lget_proc_name.c:4:0:
mi/Gget_proc_name.c: In function '_ULia64_get_proc_name':
mi/Gget_proc_name.c:107:8: error: 'struct cursor' has no member named 'dwarf'
   if (c->dwarf.use_prev_instr)
        ^~
mi/Gget_proc_name.c:111:8: error: 'struct cursor' has no member named 'dwarf'
   if (c->dwarf.use_prev_instr && offp != NULL && error == 0)
        ^~
```

Signed-off-by: Sergei Trofimovich <slyfox@gentoo.org>
2017-07-18 10:17:00 -04:00
Doug Moore
55eb47d6e5 dwarf: fix single_fde
Usage of the single_fde field in cb_data suggests that it should be
set only when dwarf_extract_proc_info_from_fde has completed successfully,
but instead it is set before the linear search for the matching ip has
begun.  Set it only when that search has completed successfully, and
has thus extracted the proc_info.
2017-07-10 10:18:14 -07:00
Doug Moore
90d0d15c4e dwarf: fix synthetic eh_frame_hdr
Ben Avison (bavison@riscopen.org) has observed that when a synthetic
eh_frame_hdr is generated, there is no space in it for the eh_frame,
so the eh_frame value is written to, and later read from, memory that
is not assigned to this purpose, with unpredictable results.

This change adds a new field to the dwarf_eh_frame_hdr type, to
make room for that value, and adds the (packed) attribute to the
struct defintion to avoid a problem with unused space in the struct.
2017-07-06 09:15:08 -07:00
Romain Naour
ca6b6f3ad9 libunwind-arm: fix build failure due to asm()
mesa3d on ARM build with libunwind support enabled fail to build due to asm()
function used when building with -std=c99.
The gcc documentation [1] suggest to use __asm__ instead of asm.

Fixes:
https://urldefense.proofpoint.com/v2/url?u=http-3A__autobuild.buildroot.net_results_3ef_3efe156b6494e4392b6c31de447ee2c72acc1a53&d=DwICAg&c=5VD0RTtNlTh3ycd41b3MUw&r=vou6lT5jmE_fWQWZZgNrsMWu4RT87QAB9V07tPHlP5U&m=BlAszRQ0vewy5vW7raCh9FmNOACKez_juz55zoiNfUs&s=4sXL6_rFriQz7qi5ygKXBIVHMc7YSdCBnkkHoi347CU&e=

[1] https://gcc.gnu.org/onlinedocs/gcc/Alternate-Keywords.html#Alternate-Keywords

Signed-off-by: Romain Naour <romain.naour@smile.fr>
Cc: Bernd Kuhls <bernd.kuhls@t-online.de>
2017-07-05 10:07:28 -07:00
Doug Moore
38b2bdab33 Add a missing semicolon. 2017-07-05 09:53:46 -07:00
Stephen Chen
2934cf4052 aarch64: reduce UNW_TDEP_CURSOR_LEN to 512 on aarch64.
This allows libunwind to work in situations with limited stack size (ie. fibers). 512 is still more than enough for storing the cursor with some slack. (#25)
2017-06-20 09:49:16 -07:00
Doug Moore
0e74e583ae arm: Use dwarf_find_proc_info for arm dwarf processing
Rather than using a copy of dwarf_find_proc_info that differs from it slightly.
By using dwarf_find_proc_info, a potential search of the di table is
allowed, where it is omitted now.  Also, for ARM, avoid runtime
checks about which kind of unwind table search to use after dl_iterate_phdr.

A couple of Debug() warnings about ip lookup failure are lost here.
The dwarf callback struct defintion is moved to Gfind_proc_info-lsb.c,
which becomes the only source file that needs it.
2017-06-20 09:47:54 -07:00
Doug Moore
4f14d9714f Don't give up on proc finding if dwarf_callback fails, but
don't bother completing the dwarf search work either.  Move on
to the arm method.
2017-06-16 08:54:48 -07:00
Doug Moore
be06809b5f Arm code uses dwarf before arm for finding proc info, but unlike
dwarf code, ignores error return from dwarf_callback.  Change arm
code to behave like dwarf code in that case.
2017-06-16 08:54:47 -07:00
Doug Moore
0b51f5892d Dwarf cache nodes are allocated in a round-robin fashion, despite the
'lru' prefix used in several data fields.  Drop the unnecessary fields,
and just use a simple counter to track the next cache entry to be recycled.
2017-06-16 08:53:36 -07:00
Doug Moore
5c0e619f84 test: Add a test like test-mem, but using reg_states_iterate to find the state and
apply_reg_state to use it to complete the unw_step.
2017-06-07 08:36:23 -07:00
Doug Moore
27f5f9fa0b Leave ret_addr_column out of the data that gets copied when pushing/popping
data on/off the register state stack.
2017-05-20 14:36:25 -05:00
Doug Moore
c66661f73c Drop reference to dwarf.ret_addr_column. 2017-05-19 19:19:12 -05:00
Doug Moore
afb2491ccb Have dwarf_step return 0 or 1 for success, according to whether the
previous stack frame was the last, just as unw_step does.  For x86_64,
drop the null check for ret_addr_column, since that check is made already
in apply_reg_state.
2017-05-19 18:57:14 -05:00
Doug Moore
f3be411bc8 Use the register_state version of ret_addr_column, rather than the
cursor version, for frame-stashing.  They have the same value in master.
2017-05-19 18:22:08 -05:00
Doug Moore
7634874591 Move the change to c->ret_addr_column to the end of dwarf_step, to make sure
that it gets set in the cache hit case.
2017-05-19 17:59:08 -05:00
Doug Moore
ec1a021243 Move the ret_addr_column field from dwarf_reg_cache_entry to dwarf_reg_state,
so that it will get saved and restored with the register state.  Initialize
the rs_state version of ret_addr_column at the some time the dwarf_cursor
version is initialized, and don't bother copying ret_addr_column explicitly
from cursor to cache since it's copied implicitly as part of reg_state.
Use the reg_state version in apply_reg_state, instead of the cursor version.

Which brings up the question: why do we have ret_addr_column in the dwarf_cursor?
We call find_reg_state before calling apply_reg_state, so the value of ret_addr_column
in the cursor when dwarf_step gets called gets overwritten before it is used.  So
it's initial value doesn't matter.  But some architectures do funky things with
cursor->ret_addr_column, even though I don't see how they matter.

So I'm not deleting dwarf_cursor->ret_addr_column, even though I suspect this
patch makes it useless.
2017-05-16 22:00:38 -05:00
Doug Moore
028a60f064 Change dwarf_reg_state from an array of pairs of differently-sized objects
to a pair of arrays, to reduce internal fragmentation.  Reduces storage
use by 37.5% on x86_64.
2017-05-16 17:40:58 -05:00
Doug Moore
2b692db23f In dwarf_reg_states_table_iterate, the rs stack has to persist from
one run_cfi_program call to the next, so pull the stack out of run_cfi_program
and make the caller responsible for cleanup at the end.
2017-05-16 01:04:49 -05:00
Doug Moore
bb61e0bc2b Bury the last_ip field until it can reappear in version 2.0. 2017-05-12 23:45:56 -05:00
Doug Moore
1568d7afa2 Add documentation. 2017-05-12 22:24:00 -05:00
Doug Moore
fef5de6c45 Move ret_addr_column and signal_frame from dwarf_reg_state to
dwarf_reg_cache_entry, leaving in dwarf_reg_state only what
apply_reg_state needs.
2017-05-12 22:24:00 -05:00
Doug Moore
a7c65f5c3e Remove next field from dwarf_reg_state. Create new struct
that includes next field and dwarf_reg_state, and use that
strictly for stack push/pop in run_cfi_program.
2017-05-12 22:24:00 -05:00
Doug Moore
bfb246a499 Add new names to check-namespace. 2017-05-12 22:24:00 -05:00
Doug Moore
4184bb478a Restore a dwarf-free "apply_reg_state" function. 2017-05-12 22:24:00 -05:00
Doug Moore
3888b2bbcb In dwarf_apply_reg_state change from dewarf_reg_state_t back to struct dwarf_reg_state. 2017-05-12 22:24:00 -05:00
Doug Moore
502ba27753 Add a function to capture the dwarf_reg_states that occur in processing
the dwarf code for a procedure, and a function to apply a captured
dwarf_reg_state later.
2017-05-12 22:24:00 -05:00
Doug Moore
bbdc4b12da Capture the address of the first byte after the instructionless gap
that follows the procedure.
2017-05-12 22:24:00 -05:00
Doug Moore
06d230d732 Merge dwarf_find_save_locs into dwarf_step. 2017-05-12 22:24:00 -05:00
Doug Moore
1602987b5f Extract 'find_reg_state' from dwarf_find_save_locs. 2017-05-12 22:23:59 -05:00
Doug Moore
04db2b7cea The call to put_unwind_info is what makes pi invalid, so clear the
valid bit here, and soon stop bothering to clear it in other places.
2017-05-12 22:23:59 -05:00
Doug Moore
1c76d41939 Pass in a pointer to the address where run_cfi_program will start,
or resume, its decoding, rather than always starting at the procedure
start, so that resuming a cfi_program becomes possible.
2017-05-12 22:23:59 -05:00
Doug Moore
7a962b951b Create setup_fde, and put all the one-time parts of fde processing there. 2017-05-12 22:23:59 -05:00
Doug Moore
50a457abb0 Pull cache-related fields out of reg_state, and put them into reg_cache struct. 2017-05-12 22:23:59 -05:00
Doug Moore
59ecb24ca2 dwarf:Drop dwarf_create_state_record.
In dwarf_make_proc_info, fix a leak in the case that create_state_record fails.
2017-04-28 09:59:22 -07:00
Doug Moore
3775cdd997 dwarf: Avoid a leak for a no-cache failure in create_state_record_for.
Avoid a duplicate copy for a cache miss.
2017-04-28 09:58:40 -07:00
Doug Moore
24ac32b480 run_cfi patch for libunwind
dwarf: If the dwarf_readu8 call to set op fails, and if there are register
states pushed onto the stack, the stack is not emptied before the function
returns. This change addresses that.

Most of the rest is eliminating ‘goto fail’ from the code.
2017-04-26 08:11:34 -07:00
Doug Moore
24f27f02ac dwarf: error in unmap size arguments
This patch fixes an error in the unmapping of
hash table memory when the dwarf hash table is resized.
2017-04-17 10:28:48 -07:00
Doug Moore
0f24c6dec8 dwarf: Cleanup dwarf cache hint
0-valued hints are used when they are just initial values with no use as
hints. And I think there were some other problems as well.
This patch cleans up and stores hint values with 1 added, so that 0-valued
hints can be ignored.
2017-04-14 08:02:10 -07:00
Paul Pluzhnikov
f819f1b87c ppc: fix inverted check 2017-04-05 12:00:21 -07:00
Paul Pluzhnikov
7a7833ee0a ppc: return UNW_INVALID instead of assert on unaligned addresses
In src/ppc64/Gstep.c, we use fetch32 to fetch instruction from the
inferior process. In  UNW_REMOTE case, fetch32 asserts that the
address we are fetching from is aligned.

But if the inferior is corrupt, we can get unaligned IP, and hit the assert.

Attached patch removes the assert, and makes fetch32 (and fetch16)
return -UNW_EINVAL instead.
2017-04-05 11:14:04 -07:00
Dave Watson
982e934951 travis: run unittests for x86_64
some tests are failing with clang currently.
2017-04-04 12:41:03 -07:00
Dave Watson
aee8bbe795 dwarf: fix unw_get_proc_info
unw_get_proc_info must always load the unwind info so that unw_resume
works with GNU_args_size expressions, but must not update
use_prev_expr unless we are unw_step()ing.

blame rev: 4b63a536ee

reported-by: Doug Moore <dougm@rice.edu>
2017-04-04 11:15:23 -07:00