[llvm] ef28379 - [llvm-objcopy] Fix file offsets when PT_INTERP/PT_LOAD offsets are equal (#80562)
via llvm-commits
llvm-commits at lists.llvm.org
Tue Feb 20 09:26:09 PST 2024
Author: Fangrui Song
Date: 2024-02-20T09:26:04-08:00
New Revision: ef28379022121ce32cf93d7c3a7571a73ab5f801
URL: https://github.com/llvm/llvm-project/commit/ef28379022121ce32cf93d7c3a7571a73ab5f801
DIFF: https://github.com/llvm/llvm-project/commit/ef28379022121ce32cf93d7c3a7571a73ab5f801.diff
LOG: [llvm-objcopy] Fix file offsets when PT_INTERP/PT_LOAD offsets are equal (#80562)
(#79887) When the offset of a PT_INTERP segment equals the offset of a
PT_LOAD segment, we consider that the parent of the PT_LOAD segment is
the PT_INTERP segment. In `layoutSegments`, we place both segments to be
after the current `Offset`, ignoring the PT_LOAD alignment.
This scenario is possible with fixed section addresses, but doesn't
happen with default linker layouts (.interp precedes other sections and
is part of a PT_LOAD segment containing the ELF header and program
headers).
```
% cat a.s
.globl _start; _start: ret
.rodata; .byte 0
.tdata; .balign 4096; .byte 0
% clang -fuse-ld=lld a.s -o a -nostdlib -no-pie -z separate-loadable-segments -Wl,-Ttext=0x201000,--section-start=.interp=0x202000,--section-start=.rodata=0x202020,-z,nognustack
% llvm-objcopy a a2
% llvm-readelf -l a2 # incorrect offset(PT_LOAD)
Type Offset VirtAddr PhysAddr FileSiz MemSiz Flg Align
PHDR 0x000040 0x0000000000200040 0x0000000000200040 0x0001c0 0x0001c0 R 0x8
INTERP 0x001001 0x0000000000202000 0x0000000000202000 0x00001c 0x00001c R 0x1
[Requesting program interpreter: /lib64/ld-linux-x86-64.so.2]
LOAD 0x000000 0x0000000000200000 0x0000000000200000 0x000200 0x000200 R 0x1000
LOAD 0x001000 0x0000000000201000 0x0000000000201000 0x000001 0x000001 R E 0x1000
//// incorrect offset
LOAD 0x001001 0x0000000000202000 0x0000000000202000 0x000021 0x000021 R 0x1000
LOAD 0x002000 0x0000000000203000 0x0000000000203000 0x000001 0x001000 RW 0x1000
TLS 0x002000 0x0000000000203000 0x0000000000203000 0x000001 0x000001 R 0x1000
GNU_RELRO 0x002000 0x0000000000203000 0x0000000000203000 0x000001 0x001000 R 0x1000
```
The same issue occurs for PT_TLS/PT_GNU_RELRO if we PT_TLS's alignment
is smaller and we place the PT_LOAD after PT_TLS/PT_GNU_RELRO segments
(not linker default, but possible with a `PHDRS` linker script command).
Fix #79887: when two segments have the same offset, order the one with a
larger alignment first. In the previous case, the PT_LOAD segment will
go before the PT_INTERP segment. In case of equal alignments, it doesn't
matter which segment is treated as the parent segment.
Added:
llvm/test/tools/llvm-objcopy/ELF/non-load-at-load-start.test
Modified:
llvm/lib/ObjCopy/ELF/ELFObject.cpp
Removed:
################################################################################
diff --git a/llvm/lib/ObjCopy/ELF/ELFObject.cpp b/llvm/lib/ObjCopy/ELF/ELFObject.cpp
index b9b91670b9983c..9547cc10d2a0ba 100644
--- a/llvm/lib/ObjCopy/ELF/ELFObject.cpp
+++ b/llvm/lib/ObjCopy/ELF/ELFObject.cpp
@@ -1234,6 +1234,12 @@ static bool compareSegmentsByOffset(const Segment *A, const Segment *B) {
return true;
if (A->OriginalOffset > B->OriginalOffset)
return false;
+ // If alignments are
diff erent, the one with a smaller alignment cannot be the
+ // parent; otherwise, layoutSegments will not respect the larger alignment
+ // requirement. This rule ensures that PT_LOAD/PT_INTERP/PT_GNU_RELRO/PT_TLS
+ // segments at the same offset will be aligned correctly.
+ if (A->Align != B->Align)
+ return A->Align > B->Align;
return A->Index < B->Index;
}
diff --git a/llvm/test/tools/llvm-objcopy/ELF/non-load-at-load-start.test b/llvm/test/tools/llvm-objcopy/ELF/non-load-at-load-start.test
new file mode 100644
index 00000000000000..60404a139d507a
--- /dev/null
+++ b/llvm/test/tools/llvm-objcopy/ELF/non-load-at-load-start.test
@@ -0,0 +1,118 @@
+## When the offset of a non-PT_LOAD segment (e.g. PT_INTERP) equals the offset
+## of a PT_LOAD segment, set the parent of the segment with a smaller alignment
+## to the segment with a larger alignment, ensuring that the offset is correctly
+## aligned.
+
+# RUN: yaml2obj %s -o %t
+# RUN: llvm-objcopy %t %t2
+# RUN: llvm-readelf -Sl %t2 | FileCheck %s
+
+# CHECK: [Nr] Name Type Address Off Size ES Flg Lk Inf Al
+# CHECK-NEXT: [ 0] NULL 0000000000000000 000000 000000 00 0 0 0
+# CHECK-NEXT: [ 1] .text PROGBITS 0000000000201000 001000 000001 00 0 0 4
+# CHECK-NEXT: [ 2] .interp PROGBITS 0000000000202000 002000 00001c 00 0 0 1
+# CHECK-NEXT: [ 3] .rodata PROGBITS 0000000000202020 002020 000001 00 0 0 1
+# CHECK-NEXT: [ 4] .tdata PROGBITS 0000000000203000 003000 000001 00 0 0 4096
+# CHECK-NEXT: [ 5] .relro_padding NOBITS 0000000000203001 003001 000fff 00 0 0 1
+# CHECK-NEXT: [ 6] .strtab STRTAB 0000000000000000 003001 000001 00 0 0 1
+# CHECK-NEXT: [ 7] .shstrtab STRTAB 0000000000000000 003002 00003f 00 0 0 1
+
+# CHECK: Program Headers:
+# CHECK-NEXT: Type Offset VirtAddr PhysAddr FileSiz MemSiz Flg Align
+# CHECK-NEXT: PHDR 0x000040 0x0000000000200040 0x0000000000200040 0x0001c0 0x0001c0 0x8
+# CHECK-NEXT: INTERP 0x002000 0x0000000000202000 0x0000000000202000 0x00001c 0x00001c 0x1
+# CHECK-NEXT: [Requesting program interpreter: ]
+# CHECK-NEXT: LOAD 0x000000 0x0000000000200000 0x0000000000200000 0x000200 0x000200 0x1000
+# CHECK-NEXT: LOAD 0x001000 0x0000000000201000 0x0000000000201000 0x000001 0x000001 0x1000
+# CHECK-NEXT: LOAD 0x002000 0x0000000000202000 0x0000000000202000 0x000021 0x000021 0x1000
+# CHECK-NEXT: TLS 0x003000 0x0000000000203000 0x0000000000203000 0x000001 0x001000 0x1000
+# CHECK-NEXT: GNU_RELRO 0x003000 0x0000000000203000 0x0000000000203000 0x000001 0x001000 0x1000
+# CHECK-NEXT: LOAD 0x003000 0x0000000000203000 0x0000000000203000 0x000001 0x001000 0x1000
+
+--- !ELF
+FileHeader:
+ Class: ELFCLASS64
+ Data: ELFDATA2LSB
+ Type: ET_EXEC
+ Machine: EM_X86_64
+ProgramHeaders:
+ - Type: PT_PHDR
+ VAddr: 0x200040
+ Align: 0x8
+ Offset: 0x40
+ FileSize: 0x1c0
+ MemSize: 0x1c0
+ - Type: PT_INTERP
+ FirstSec: .interp
+ LastSec: .interp
+ ## The address equals the address of its containing PT_LOAD.
+ VAddr: 0x202000
+ Offset: 0x2000
+ - Type: PT_LOAD
+ VAddr: 0x200000
+ Align: 0x1000
+ Offset: 0x0
+ FileSize: 0x200
+ MemSize: 0x200
+ - Type: PT_LOAD
+ FirstSec: .text
+ LastSec: .text
+ VAddr: 0x201000
+ Align: 0x1000
+ Offset: 0x1000
+ - Type: PT_LOAD
+ FirstSec: .interp
+ LastSec: .rodata
+ VAddr: 0x202000
+ Align: 0x1000
+ Offset: 0x2000
+ ## Intentionally place PT_TLS/PT_GNU_RELRO before PT_LOAD to test that we
+ ## correctly set parent segments.
+ - Type: PT_TLS
+ FirstSec: .tdata
+ LastSec: .relro_padding
+ VAddr: 0x203000
+ Align: 0x1000
+ Offset: 0x3000
+ - Type: PT_GNU_RELRO
+ FirstSec: .tdata
+ LastSec: .relro_padding
+ VAddr: 0x203000
+ Align: 0x1000
+ Offset: 0x3000
+ - Type: PT_LOAD
+ FirstSec: .tdata
+ LastSec: .relro_padding
+ VAddr: 0x203000
+ Align: 0x1000
+ Offset: 0x3000
+Sections:
+ - Name: .text
+ Type: SHT_PROGBITS
+ Address: 0x201000
+ AddressAlign: 0x4
+ Offset: 0x1000
+ Content: C3
+ - Name: .interp
+ Type: SHT_PROGBITS
+ Address: 0x202000
+ AddressAlign: 0x1
+ Offset: 0x2000
+ Size: 0x1C
+ - Name: .rodata
+ Type: SHT_PROGBITS
+ Address: 0x202020
+ AddressAlign: 0x1
+ Offset: 0x2020
+ Size: 1
+ - Name: .tdata
+ Type: SHT_PROGBITS
+ Address: 0x203000
+ AddressAlign: 0x1000
+ Offset: 0x3000
+ Size: 1
+ - Name: .relro_padding
+ Type: SHT_NOBITS
+ Address: 0x203001
+ AddressAlign: 0x1
+ Size: 0xFFF
More information about the llvm-commits
mailing list