[Bug default/27552] New: libandroid.so missing symbol type information

[Bug default/27552] New: libandroid.so missing symbol type information

[gprocida+abigail at google dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=27552

            Bug ID: 27552
           Summary: libandroid.so missing symbol type information
           Product: libabigail
           Version: unspecified
            Status: UNCONFIRMED
          Severity: normal
          Priority: P2
         Component: default
          Assignee: dodji at redhat dot com
          Reporter: gprocida+abigail at google dot com
                CC: libabigail at sourceware dot org
  Target Milestone: ---

The libandroid.so library is now part of the libabigail test suite, as is an
ABI extracted from this.

I noticed that the names used of the ELF symbols and the names of declared
variables and functions, as found in the ABI, are disjoint.

I verified that:
1. if a symbol is added or removed, this is reported
2. if any type information changes, this is not reported

An examination of libandroid.so with nm -o and dwarfdump confirms that symbol
names are properly present in the DWARF.

For example, from the dwarfdump:

< 1><0x00024873>    DW_TAG_subprogram
                      DW_AT_low_pc                0x0000ce62
                      DW_AT_high_pc               <offset-from-lowpc>38
                      DW_AT_frame_base            len 0x0001: 0x5d: 
                          DW_OP_reg13
                      DW_AT_GNU_all_call_sites    yes(1)
                      DW_AT_name                  AAssetDir_close
                      DW_AT_decl_file             0x00000034
frameworks/base/native/android/asset_manager.cpp
                      DW_AT_decl_line             0x0000008c
                      DW_AT_external              yes(1)
< 2><0x00024885>      DW_TAG_formal_parameter
                        DW_AT_location              0x00001332
      .debug_loc      offset  : 0x00001332
      <loclist at offset 0x00001332 with 4 entries follows>
   [ 0]<base address       0x0000ce62>
   [ 1]<low,hi addrs       0x0000ce62 , 0x0000ce68>
                            DW_OP_reg0
   [ 2]<low,hi addrs       0x0000ce68 , 0x0000ce82>
                            DW_OP_reg4
   [ 3]<low,hi addrs       0x0000ce86 , 0x0000ce88>
                            DW_OP_reg0
                        DW_AT_name                  assetDir
                        DW_AT_decl_file             0x00000034
frameworks/base/native/android/asset_manager.cpp
                        DW_AT_decl_line             0x0000008c
                        DW_AT_type                  <0x00024e40>
< 2><0x00024894>      DW_TAG_inlined_subroutine
...


If I discover anything further, I'll follow up here.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug default/27552] libandroid.so missing symbol type information

[gprocida+abigail at google dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=27552

--- Comment #1 from Giuliano Procida <gprocida+abigail at google dot com> ---
This looks like an ELF library bug and not a libabigail bug.

The addresses in DWARF are off-by-one with respect to the ELF symbol table
addresses. LLVM and GNU nm disagree about the addresses as well.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug default/27552] libandroid.so missing symbol type information

[mark at klomp dot org]

https://sourceware.org/bugzilla/show_bug.cgi?id=27552

Mark Wielaard <mark at klomp dot org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |mark at klomp dot org

--- Comment #2 from Mark Wielaard <mark at klomp dot org> ---
So this is ./tests/data/test-read-dwarf/test-libandroid.so
Which is a 32bit ARM binary.

Those are "special" :)

https://static.docs.arm.com/ihi0044/g/aaelf32.pdf

Section 5.5.3 Symbol Values

•  If the symbol addresses a Thumb instruction, its value is the address of the
instruction with bit zero set (in a relocatable object, the section offset with
bit zero set).

 This allows a linker to distinguish Arm and Thumb code symbols without having
to refer to the map.  An Arm symbol will always have an even value, while a
Thumb symbol will always have an odd value.

So it depends on when which tool strips off the zero-bit of the value.

It looks like AAssetDir_close is a Thumb function, so its address is 0000ce62,
but the value in the symbol table is 0000ce63.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug default/27552] libabigail needs to interpret ARM32 symbol addresses specially

[gprocida+abigail at google dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=27552

Giuliano Procida <gprocida+abigail at google dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|libandroid.so missing       |libabigail needs to
                   |symbol type information     |interpret ARM32 symbol
                   |                            |addresses specially

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug default/27552] libabigail needs to interpret ARM32 symbol addresses specially

[dodji at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=27552

dodji at redhat dot com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|---                         |FIXED

--- Comment #3 from dodji at redhat dot com ---
This should now be fixed by commit
https://sourceware.org/git/?p=libabigail.git;a=commit;h=32c7829e4156666e1975329fb1b3201c43b5f84f
and should be available in libabigail 2.0.

-- 
You are receiving this mail because:
You are on the CC list for the bug.