maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity
llvm-project/lld/test/ELF/arm-tls-norelax-gd-ie.s

31 lines
1.0 KiB
ArmAsm
Raw Normal View History

[ELF] Move `// REQUIRES:` line to the top llvm-svn: 335676
2018-06-27 06:20:04 +08:00
// REQUIRES: arm
[ARM] ARM TLS shouldn't use relaxations The ARM TLS relocations are placed on literal data and not the code-sequence, it is therefore not possible to implement the relaxTls* functions. This change updates handleMipsTlsRelocation() to handleNoRelaxTlsRelocation() and incorporates ARM as well as Mips. The ARM support in handleNoRelaxTlsRelocation() currently needs to ouput the module index dynamic relocation in all cases as it is relying on the dynamic linker to set the module index in the got. Should address PR30218 Differential Revision: https://reviews.llvm.org/D24827 llvm-svn: 282250
2016-09-23 21:54:48 +08:00
// RUN: llvm-mc -filetype=obj -triple=armv7a-none-linux-gnueabi %p/Inputs/arm-tls-get-addr.s -o %t1
// RUN: ld.lld %t1 --shared -o %t1.so
// RUN: llvm-mc %s -o %t.o -filetype=obj -triple=armv7a-linux-gnueabi
[ELF] - Do --hash-style=both by default. Its PR34712, GNU linkers recently changed default values to "both" of "sysv". Patch do the same for all targets except MIPS, where .gnu.hash section is not yet supported. Code suggested by Rui Ueyama. Differential revision: https://reviews.llvm.org/D38407 llvm-svn: 315051
2017-10-06 17:37:44 +08:00
// RUN: ld.lld --hash-style=sysv %t1.so %t.o -o %t
[test] Change llvm-readobj -long-option to --long-option or well-known short options. NFC Also change some options that have different semantics (cause confusion) in llvm-readelf mode: -s => -S -t => --symbols -sd => --section-data llvm-svn: 359651
2019-05-01 13:49:01 +08:00
// RUN: llvm-readobj -S --dyn-relocations %t | FileCheck %s
[ARM] ARM TLS shouldn't use relaxations The ARM TLS relocations are placed on literal data and not the code-sequence, it is therefore not possible to implement the relaxTls* functions. This change updates handleMipsTlsRelocation() to handleNoRelaxTlsRelocation() and incorporates ARM as well as Mips. The ARM support in handleNoRelaxTlsRelocation() currently needs to ouput the module index dynamic relocation in all cases as it is relying on the dynamic linker to set the module index in the got. Should address PR30218 Differential Revision: https://reviews.llvm.org/D24827 llvm-svn: 282250
2016-09-23 21:54:48 +08:00
// This tls global-dynamic sequence is with respect to a preemptible symbol but
// is in an application so a relaxation to Initial Exec would normally be
// possible. This would result in an assertion failure on ARM as the
// relaxation functions can't be implemented on ARM. Check that the sequence
// is handled as global dynamic
.text
.syntax unified
.globl func
.p2align 2
.type func,%function
func:
.L0:
.globl __tls_get_addr
bl __tls_get_addr
bx lr
.p2align 2
.Lt0: .word y(TLSGD) + (. - .L0 - 8)
// CHECK: Dynamic Relocations {
[ELF] Simplify RelRo, TLS, NOBITS section ranks and make RW PT_LOAD start with RelRo Old: PT_LOAD(.data | PT_GNU_RELRO(.data.rel.ro .bss.rel.ro) | .bss) New: PT_LOAD(PT_GNU_RELRO(.data.rel.ro .bss.rel.ro) | .data .bss) The placement of | indicates page alignment caused by PT_GNU_RELRO. The new layout has simpler rules and saves space for many cases. Old size: roundup(.data) + roundup(.data.rel.ro) New size: roundup(.data.rel.ro + .bss.rel.ro) + .data Other advantages: * At runtime the 3 memory mappings decrease to 2. * start(PT_TLS) = start(PT_GNU_RELRO) = start(RW PT_LOAD). This simplifies binary manipulation tools. GNU strip before 2.31 discards PT_GNU_RELRO if its address is not equal to the start of its associated PT_LOAD. This has been fixed by https://sourceware.org/git/gitweb.cgi?p=binutils-gdb.git;h=f2731e0c374e5323ce4cdae2bcc7b7fe22da1a6f But with this change, we will be compatible with GNU strip before 2.31 * Before, .got.plt (non-relro by default) was placed before .got (relro by default), which made it impossible to have _GLOBAL_OFFSET_TABLE_ (start of .got.plt on x86-64) equal to the end of .got (R_GOT*_FROM_END) (https://bugs.llvm.org/show_bug.cgi?id=36555). With the new ordering, we can improve on this regard if we'd like to. Reviewers: ruiu, espindola, pcc Subscribers: emaste, arichardson, llvm-commits, joerg, jdoerfert Differential Revision: https://reviews.llvm.org/D56828 llvm-svn: 356117
2019-03-14 11:47:45 +08:00
// CHECK-NEXT: 0x12078 R_ARM_TLS_DTPMOD32 y
// CHECK-NEXT: 0x1207C R_ARM_TLS_DTPOFF32 y
// CHECK-NEXT: 0x1300C R_ARM_JUMP_SLOT __tls_get_addr