1 # This test has a subtle purpose. Because of the specifics of the layout 2 # algorithm if we're not careful cycles can occur when we resolve the order 3 # that we should layout segments in. In this test we're making sure that the 4 # PT_LOAD segment won't set the PT_NOTE segment as the first to be laid out 5 # while the PT_NOTE segment sets the PT_LOAD load to be the first to be laid 6 # out. This problem becomes visible if the layout of the file changes due to 7 # the removal of a section. We use -O binary here because removing a section 8 # in this way won't cause the location of the PT_LOAD segment to change if we 9 # don't. 10 11 # RUN: yaml2obj %s -o %t 12 # RUN: llvm-objcopy -R .note -O binary %t %t2 13 # RUN: od -Ax -t x1 %t2 | FileCheck %s 14 15 !ELF 16 FileHeader: 17 Class: ELFCLASS64 18 Data: ELFDATA2LSB 19 Type: ET_EXEC 20 Machine: EM_X86_64 21 Sections: 22 - Name: .note 23 Type: SHT_PROGBITS 24 Flags: [ SHF_ALLOC ] 25 Address: 0x1000 26 AddressAlign: 0x1000 27 Content: "32323232" 28 Size: 32 29 - Name: .rodata 30 Flags: [ SHF_ALLOC ] 31 Type: SHT_PROGBITS 32 Address: 0x1020 33 Size: 4064 34 Content: "DEADBEEF" 35 ProgramHeaders: 36 - Type: PT_LOAD 37 Flags: [ PF_R ] 38 VAddr: 0x1000 39 PAddr: 0x1000 40 Sections: 41 - Section: .note 42 - Section: .rodata 43 - Type: PT_NOTE 44 Flags: [ PF_R ] 45 VAddr: 0x1000 46 PAddr: 0x1000 47 Sections: 48 - Section: .note 49 50 # CHECK: 000000 de ad be ef 51
