https://sourceware.org/bugzilla/show_bug.cgi?id=34887
Bug ID: 34887
Summary: bjdump --dwarf=frames: .eh_frame CIE ID and CIE
pointer read as 8 bytes after a 64-bit length
Product: binutils
Version: 2.47
Status: UNCONFIRMED
Severity: normal
Priority: P2
Component: binutils
Assignee: unassigned at sourceware dot org
Reporter: i at mi11ion dot io
Target Milestone: ---
Host: aarch64-apple-darwin
Created attachment 17206
--> https://sourceware.org/bugzilla/attachment.cgi?id=17206&action=edit
dwarf64.o (+ .s): Mach-O object whose __eh_frame CIE and FDE use 64-bit
lengths; dwarf32.o: same with 32-bit length
Environment: GNU objdump (GNU Binutils) 2.47.20260726, Homebrew binutils 2.47
on macOS 27.0.1 (arm64). The code is unchanged on master at 1c79f853160d.
Reproducer: the attached eh-frame-64.tar.gz holds dwarf64.o (504 bytes, Mach-O
arm64 object, assembled from dwarf64.s with "as -arch arm64"). Its .eh_frame is
64 bytes: a CIE and an FDE, each with Length 0xffffffff and an 8-byte Extended
Length, then the 4-byte CIE ID / CIE pointer:
ff ff ff ff 0c 00 00 00 00 00 00 00 00 00 00 00
01 00 01 78 1e 0c 1f 00 ff ff ff ff 1c 00 00 00
00 00 00 00 24 00 00 00 00 10 00 00 00 00 00 00
40 00 00 00 00 00 00 00 44 0e 10 00 00 00 00 00
dwarf32.o holds the same records with 32-bit lengths, and objdump prints them
as expected.
Command: objdump --dwarf=frames dwarf64.o
Actual:
00000000 000000000000000c 7801000100000000 FDE cie=invalid
pc=0000000000000000..0000000000000000
00000018 000000000000001c 0000100000000024 FDE cie=invalid
pc=0000004000000000..00100e8400000000
DW_CFA_nop
DW_CFA_nop
DW_CFA_nop
DW_CFA_nop
with "Warning: Encoded value extends past end of section".
Expected:
00000000 000000000000000c 00000000 CIE
Version: 1
Augmentation: ""
Code alignment factor: 1
Data alignment factor: -8
Return address column: 30
DW_CFA_def_cfa: r31 (sp) ofs 0
00000018 000000000000001c 00000024 FDE cie=00000000
pc=0000000000001000..0000000000001040
DW_CFA_advance_loc: 4 to 0000000000001004
DW_CFA_def_cfa_offset: 16
DW_CFA_nop
...
llvm-dwarfdump --eh-frame prints this CIE and FDE (pc=00001000...00001040), and
the macOS libunwind accepts the FDE through __register_frame with the range
0x1000..0x1040.
--
You are receiving this mail because:
You are on the CC list for the bug.
