[LLVMbugs] [Bug 17505] New: libcxxabi tests all failing on powerpc-darwin8, suspect unwind info incompatibility
bugzilla-daemon at llvm.org
bugzilla-daemon at llvm.org
Mon Oct 7 18:35:01 PDT 2013
http://llvm.org/bugs/show_bug.cgi?id=17505
Bug ID: 17505
Summary: libcxxabi tests all failing on powerpc-darwin8,
suspect unwind info incompatibility
Product: new-bugs
Version: trunk
Hardware: Macintosh
OS: MacOS X
Status: NEW
Severity: normal
Priority: P
Component: new bugs
Assignee: unassignedbugs at nondot.org
Reporter: fang at csl.cornell.edu
CC: kledzik at apple.com, llvmbugs at cs.uiuc.edu
Classification: Unclassified
This bug is for tracking the long-standing issue of all libcxxabi tests failing
(Abort trap) on powerpc-darwin8 (latent, not recent).
Theory: suspect that unwind info emitted by CodeGen is not compatible with the
libunwind that ships with /usr/bin/libgcc_s.
I'm probably going to need some help digging through eh-frame dumps to figure
out what's wrong.
First example is unwind_01.cpp.
I'm using the clang++ from a stage1 build of the powerpc-darwin8 branch, which
tracks master.
I build all test executables with -O0 -no-integrated-as.
The clang-built unwind_01 executable gives the following dwarfdump
(--eh-frame):
=========== *PASTE* ============
Exception handling frame information for section __eh_frame
0x00000000: CIE
length: 0x00000018
CIE_id: 0xffffffff
version: 0x01
augmentation: "zPLR"
code_align: 1
data_align: -4
ra_register: 0x41
Initial Inst: DW_CFA_def_cfa (1, 0)
0x0000001c: FDE
length: 0x00000030
CIE_pointer: 0x00000000
start_addr: 0xfffff28c
range_size: 0x00000088
Instructions: DW_CFA_advance_loc4 (16)
DW_CFA_def_cfa_offset (96)
DW_CFA_advance_loc4 (0)
DW_CFA_offset (31, -4)
DW_CFA_advance_loc4 (0)
DW_CFA_offset_extended_sf (65, 8)
DW_CFA_advance_loc4 (4)
DW_CFA_def_cfa_register (31)
DW_CFA_nop
DW_CFA_nop
0x00000050: CIE
length: 0x00000010
CIE_id: 0xffffffff
version: 0x01
augmentation: "zR"
code_align: 1
data_align: -4
ra_register: 0x41
Initial Inst: DW_CFA_def_cfa (1, 0)
0x00000064: FDE
length: 0x0000002c
CIE_pointer: 0x00000050
start_addr: 0xfffff674
range_size: 0x0000003c
Instructions: DW_CFA_advance_loc4 (16)
DW_CFA_def_cfa_offset (80)
DW_CFA_advance_loc4 (0)
DW_CFA_offset (31, -4)
DW_CFA_advance_loc4 (0)
DW_CFA_offset_extended_sf (65, 8)
DW_CFA_advance_loc4 (4)
DW_CFA_def_cfa_register (31)
DW_CFA_nop
DW_CFA_nop
0x00000094: FDE
length: 0x0000002c
CIE_pointer: 0x00000050
start_addr: 0xfffff684
range_size: 0x0000003c
Instructions: DW_CFA_advance_loc4 (16)
DW_CFA_def_cfa_offset (80)
DW_CFA_advance_loc4 (0)
DW_CFA_offset (31, -4)
DW_CFA_advance_loc4 (0)
DW_CFA_offset_extended_sf (65, 8)
DW_CFA_advance_loc4 (4)
DW_CFA_def_cfa_register (31)
DW_CFA_nop
DW_CFA_nop
0x000000c4: FDE
length: 0x00000030
CIE_pointer: 0x00000000
start_addr: 0xfffff274
range_size: 0x00000094
Instructions: DW_CFA_advance_loc4 (16)
DW_CFA_def_cfa_offset (96)
DW_CFA_advance_loc4 (0)
DW_CFA_offset (31, -4)
DW_CFA_advance_loc4 (0)
DW_CFA_offset_extended_sf (65, 8)
DW_CFA_advance_loc4 (4)
DW_CFA_def_cfa_register (31)
DW_CFA_nop
DW_CFA_nop
0x000000f8: FDE
length: 0x0000002c
CIE_pointer: 0x00000050
start_addr: 0xfffff6e0
range_size: 0x0000003c
Instructions: DW_CFA_advance_loc4 (16)
DW_CFA_def_cfa_offset (80)
DW_CFA_advance_loc4 (0)
DW_CFA_offset (31, -4)
DW_CFA_advance_loc4 (0)
DW_CFA_offset_extended_sf (65, 8)
DW_CFA_advance_loc4 (4)
DW_CFA_def_cfa_register (31)
DW_CFA_nop
DW_CFA_nop
0x00000128: FDE
length: 0x00000030
CIE_pointer: 0x00000000
start_addr: 0xfffff2b0
range_size: 0x000002f8
Instructions: DW_CFA_advance_loc4 (16)
DW_CFA_def_cfa_offset (128)
DW_CFA_advance_loc4 (0)
DW_CFA_offset (31, -4)
DW_CFA_advance_loc4 (0)
DW_CFA_offset_extended_sf (65, 8)
DW_CFA_advance_loc4 (4)
DW_CFA_def_cfa_register (31)
DW_CFA_nop
0x0000015c: FDE
length: 0x00000014
CIE_pointer: 0x00000000
start_addr: 0xfffff72c
range_size: 0x00000098
Instructions: DW_CFA_nop
DW_CFA_nop
DW_CFA_nop
0x00000174: FDE
length: 0x00000014
CIE_pointer: 0x00000000
start_addr: 0xfffff7f4
range_size: 0x00000098
Instructions: DW_CFA_nop
DW_CFA_nop
DW_CFA_nop
0x0000018c: FDE
length: 0x00000014
CIE_pointer: 0x00000000
start_addr: 0xfffff87c
range_size: 0x00000098
Instructions: DW_CFA_nop
DW_CFA_nop
DW_CFA_nop
=========== end *PASTE* ============
For comparison, I then build unwind_01-copy with g++ (system gcc 4.0.1, no
options), whose executable gives the following dwarfdump --eh-frame:
=========== *PASTE* ============
Exception handling frame information for section __eh_frame
0x00000000: CIE
length: 0x00000018
CIE_id: 0xffffffff
version: 0x01
augmentation: "zPLR"
code_align: 1
data_align: -4
ra_register: 0x41
Initial Inst: DW_CFA_def_cfa (1, 0)
0x0000001c: FDE
length: 0x00000024
CIE_pointer: 0x00000000
start_addr: 0xfffffd00
range_size: 0x0000005c
Instructions: DW_CFA_advance_loc4 (8)
DW_CFA_def_cfa_offset (48)
DW_CFA_offset (31, -4)
DW_CFA_offset (30, -8)
DW_CFA_advance_loc4 (4)
DW_CFA_def_cfa_register (30)
DW_CFA_nop
0x00000044: FDE
length: 0x00000028
CIE_pointer: 0x00000000
start_addr: 0xfffffd34
range_size: 0x00000098
Instructions: DW_CFA_advance_loc4 (16)
DW_CFA_def_cfa_offset (80)
DW_CFA_offset (31, -4)
DW_CFA_offset (30, -8)
DW_CFA_offset_extended_sf (65, 8)
DW_CFA_advance_loc4 (4)
DW_CFA_def_cfa_register (30)
DW_CFA_nop
DW_CFA_nop
0x00000070: FDE
length: 0x00000024
CIE_pointer: 0x00000000
start_addr: 0xfffffda0
range_size: 0x0000005c
Instructions: DW_CFA_advance_loc4 (8)
DW_CFA_def_cfa_offset (48)
DW_CFA_offset (31, -4)
DW_CFA_offset (30, -8)
DW_CFA_advance_loc4 (4)
DW_CFA_def_cfa_register (30)
DW_CFA_nop
0x00000098: FDE
length: 0x00000028
CIE_pointer: 0x00000000
start_addr: 0xfffffdd4
range_size: 0x00000098
Instructions: DW_CFA_advance_loc4 (16)
DW_CFA_def_cfa_offset (80)
DW_CFA_offset (31, -4)
DW_CFA_offset (30, -8)
DW_CFA_offset_extended_sf (65, 8)
DW_CFA_advance_loc4 (4)
DW_CFA_def_cfa_register (30)
DW_CFA_nop
DW_CFA_nop
0x000000c4: FDE
length: 0x00000024
CIE_pointer: 0x00000000
start_addr: 0xfffffe40
range_size: 0x0000005c
Instructions: DW_CFA_advance_loc4 (8)
DW_CFA_def_cfa_offset (48)
DW_CFA_offset (31, -4)
DW_CFA_offset (30, -8)
DW_CFA_advance_loc4 (4)
DW_CFA_def_cfa_register (30)
DW_CFA_nop
0x000000ec: FDE
length: 0x00000028
CIE_pointer: 0x00000000
start_addr: 0xfffffe74
range_size: 0x00000098
Instructions: DW_CFA_advance_loc4 (16)
DW_CFA_def_cfa_offset (80)
DW_CFA_offset (31, -4)
DW_CFA_offset (30, -8)
DW_CFA_offset_extended_sf (65, 8)
DW_CFA_advance_loc4 (4)
DW_CFA_def_cfa_register (30)
DW_CFA_nop
DW_CFA_nop
0x00000118: FDE
length: 0x00000028
CIE_pointer: 0x00000000
start_addr: 0xfffff340
range_size: 0x00000090
Instructions: DW_CFA_advance_loc4 (16)
DW_CFA_def_cfa_offset (96)
DW_CFA_offset (31, -4)
DW_CFA_offset (30, -8)
DW_CFA_offset (29, -12)
DW_CFA_offset_extended_sf (65, 8)
DW_CFA_advance_loc4 (4)
DW_CFA_def_cfa_register (30)
0x00000144: FDE
length: 0x00000028
CIE_pointer: 0x00000000
start_addr: 0xfffff3a4
range_size: 0x000000ac
Instructions: DW_CFA_advance_loc4 (16)
DW_CFA_def_cfa_offset (96)
DW_CFA_offset (31, -4)
DW_CFA_offset (30, -8)
DW_CFA_offset (29, -12)
DW_CFA_offset_extended_sf (65, 8)
DW_CFA_advance_loc4 (4)
DW_CFA_def_cfa_register (30)
0x00000170: FDE
length: 0x00000028
CIE_pointer: 0x00000000
start_addr: 0xfffff424
range_size: 0x000002b4
Instructions: DW_CFA_advance_loc4 (16)
DW_CFA_def_cfa_offset (96)
DW_CFA_offset (31, -4)
DW_CFA_offset (30, -8)
DW_CFA_offset (29, -12)
DW_CFA_offset_extended_sf (65, 8)
DW_CFA_advance_loc4 (4)
DW_CFA_def_cfa_register (30)
=========== end *PASTE* ============
In summary, they are quite different. I just don't know how different they can
be, and still be functional/compatible.
Any hints on how to debug further?
--
You are receiving this mail because:
You are on the CC list for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-bugs/attachments/20131008/c5b2472e/attachment.html>
More information about the llvm-bugs
mailing list