maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity
llvm-project/clang/test/Driver/pic.c

263 lines
14 KiB
C
Raw Normal View History

Teach Clang about PIE compilations. This is the first step of PR12380. First, this patch cleans up the parsing of the PIC and PIE family of options in the driver. The existing logic failed to claim arguments all over the place resulting in kludges that marked the options as unused. Instead actually walk all of the arguments and claim them properly. We now treat -f{,no-}{pic,PIC,pie,PIE} as a single set, accepting the last one on the commandline. Previously there were lots of ordering bugs that could creep in due to the nature of the parsing. Let me know if folks would like weird things such as "-fPIE -fno-pic" to turn on PIE, but disable full PIC. This doesn't make any sense to me, but we could in theory support it. Options that seem to have intentional "trump" status (-static, -mkernel, etc) continue to do so and are commented as such. Next, a -pie-level flag is threaded into the frontend, rigged to a language option, and handled preprocessor, setting up the appropriate defines. We'll now have the correct defines when compiling with -fpie. The one place outside of the preprocessor that was inspecting the PIC level (as opposed to the relocation model, which is set and handled separately, yay!) is in the GNU ObjC runtime. I changed it to exactly preserve existing behavior. If folks want to change its behavior in the face of PIE, they can do that in a separate patch. Essentially the only functionality changed here is the preprocessor defines and bug-fixes to the argument management. Tests have been updated and extended to test all of this a bit more thoroughly. llvm-svn: 154291
2012-04-09 00:40:35 +08:00
// Test the driver's control over the PIC behavior. These consist of tests of
// the relocation model flags and the pic level flags passed to CC1.
//
// CHECK-NO-PIC: "-mrelocation-model" "static"
// CHECK-NO-PIC-NOT: "-pic-level"
// CHECK-NO-PIC-NOT: "-pie-level"
//
Completely re-work how the Clang driver interprets PIC and PIE options. There were numerous issues here that were all entangled, and so I've tried to do a general simplification of the logic. 1) The logic was mimicing actual GCC bugs, rather than "features". These have been fixed in trunk GCC, and this fixes Clang as well. Notably, the logic was always intended to be last-match-wins like any other flag. 2) The logic for handling '-mdynamic-no-pic' was preposterously unclear. It also allowed the use of this flag on non-Darwin platforms where it has no actual meaning. Now this option is handled directly based on tests of how llvm-gcc behaves, and it is only supported on Darwin. 3) The APIs for the Driver's ToolChains had the implementation ugliness of dynamic-no-pic leaking through them. They also had the implementation details of the LLVM relocation model flag names leaking through. 4) The actual results of passing these flags was incorrect on Darwin in many cases. For example, Darwin *always* uses PIC level 2 if it uses in PIC level, and Darwin *always* uses PIC on 64-bit regardless of the flags specified, including -fPIE. Darwin never compiles in PIE mode, but it can *link* in PIE mode. 5) Also, PIC was not always being enabled even when PIE was. This isn't a supported mode at all and may have caused some fallout in builds with complex PIC and PIE interactions. The result is (I hope) cleaner and clearer for readers. I've also left comments and tests about some of the truly strage behavior that is observed on Darwin platforms. We have no real testing of Windows platforms and PIC, but I don't have the tools handy to figure that out. Hopefully others can beef up our testing here. Unfortunately, I can't test this for every platform. =/ If folks have dependencies on these flags that aren't covered by tests, they may break. I've audited and ensured that all the changes in behavior of the existing tests are intentional and good. In particular I've tried to make sure the Darwin behavior (which is more suprising than the Linux behavior) also matches that of 'gcc' on my mac. llvm-svn: 168297
2012-11-19 11:52:03 +08:00
// CHECK-PIC1: "-mrelocation-model" "pic"
Teach Clang about PIE compilations. This is the first step of PR12380. First, this patch cleans up the parsing of the PIC and PIE family of options in the driver. The existing logic failed to claim arguments all over the place resulting in kludges that marked the options as unused. Instead actually walk all of the arguments and claim them properly. We now treat -f{,no-}{pic,PIC,pie,PIE} as a single set, accepting the last one on the commandline. Previously there were lots of ordering bugs that could creep in due to the nature of the parsing. Let me know if folks would like weird things such as "-fPIE -fno-pic" to turn on PIE, but disable full PIC. This doesn't make any sense to me, but we could in theory support it. Options that seem to have intentional "trump" status (-static, -mkernel, etc) continue to do so and are commented as such. Next, a -pie-level flag is threaded into the frontend, rigged to a language option, and handled preprocessor, setting up the appropriate defines. We'll now have the correct defines when compiling with -fpie. The one place outside of the preprocessor that was inspecting the PIC level (as opposed to the relocation model, which is set and handled separately, yay!) is in the GNU ObjC runtime. I changed it to exactly preserve existing behavior. If folks want to change its behavior in the face of PIE, they can do that in a separate patch. Essentially the only functionality changed here is the preprocessor defines and bug-fixes to the argument management. Tests have been updated and extended to test all of this a bit more thoroughly. llvm-svn: 154291
2012-04-09 00:40:35 +08:00
// CHECK-PIC1: "-pic-level" "1"
//
Completely re-work how the Clang driver interprets PIC and PIE options. There were numerous issues here that were all entangled, and so I've tried to do a general simplification of the logic. 1) The logic was mimicing actual GCC bugs, rather than "features". These have been fixed in trunk GCC, and this fixes Clang as well. Notably, the logic was always intended to be last-match-wins like any other flag. 2) The logic for handling '-mdynamic-no-pic' was preposterously unclear. It also allowed the use of this flag on non-Darwin platforms where it has no actual meaning. Now this option is handled directly based on tests of how llvm-gcc behaves, and it is only supported on Darwin. 3) The APIs for the Driver's ToolChains had the implementation ugliness of dynamic-no-pic leaking through them. They also had the implementation details of the LLVM relocation model flag names leaking through. 4) The actual results of passing these flags was incorrect on Darwin in many cases. For example, Darwin *always* uses PIC level 2 if it uses in PIC level, and Darwin *always* uses PIC on 64-bit regardless of the flags specified, including -fPIE. Darwin never compiles in PIE mode, but it can *link* in PIE mode. 5) Also, PIC was not always being enabled even when PIE was. This isn't a supported mode at all and may have caused some fallout in builds with complex PIC and PIE interactions. The result is (I hope) cleaner and clearer for readers. I've also left comments and tests about some of the truly strage behavior that is observed on Darwin platforms. We have no real testing of Windows platforms and PIC, but I don't have the tools handy to figure that out. Hopefully others can beef up our testing here. Unfortunately, I can't test this for every platform. =/ If folks have dependencies on these flags that aren't covered by tests, they may break. I've audited and ensured that all the changes in behavior of the existing tests are intentional and good. In particular I've tried to make sure the Darwin behavior (which is more suprising than the Linux behavior) also matches that of 'gcc' on my mac. llvm-svn: 168297
2012-11-19 11:52:03 +08:00
// CHECK-PIC2: "-mrelocation-model" "pic"
Teach Clang about PIE compilations. This is the first step of PR12380. First, this patch cleans up the parsing of the PIC and PIE family of options in the driver. The existing logic failed to claim arguments all over the place resulting in kludges that marked the options as unused. Instead actually walk all of the arguments and claim them properly. We now treat -f{,no-}{pic,PIC,pie,PIE} as a single set, accepting the last one on the commandline. Previously there were lots of ordering bugs that could creep in due to the nature of the parsing. Let me know if folks would like weird things such as "-fPIE -fno-pic" to turn on PIE, but disable full PIC. This doesn't make any sense to me, but we could in theory support it. Options that seem to have intentional "trump" status (-static, -mkernel, etc) continue to do so and are commented as such. Next, a -pie-level flag is threaded into the frontend, rigged to a language option, and handled preprocessor, setting up the appropriate defines. We'll now have the correct defines when compiling with -fpie. The one place outside of the preprocessor that was inspecting the PIC level (as opposed to the relocation model, which is set and handled separately, yay!) is in the GNU ObjC runtime. I changed it to exactly preserve existing behavior. If folks want to change its behavior in the face of PIE, they can do that in a separate patch. Essentially the only functionality changed here is the preprocessor defines and bug-fixes to the argument management. Tests have been updated and extended to test all of this a bit more thoroughly. llvm-svn: 154291
2012-04-09 00:40:35 +08:00
// CHECK-PIC2: "-pic-level" "2"
//
Stop treating -static as overriding -fPIC: they are distinct. For some reason, clang had been treating a command like: clang -static -fPIC foo.c as if it should be compiled without the PIC relocation model. This was incorrect: -static should be affecting only the linking model, and -fPIC only the compilation. This new behavior also matches GCC. This is a follow-up from a review comment on r245447. Differential Revision: http://reviews.llvm.org/D12208 llvm-svn: 245667
2015-08-21 12:14:55 +08:00
// CHECK-STATIC: "-static"
Do not force "-static" for aarch64 iOS kernel/kext assembly code. Somehow a check for aarch64 was added to the Darwin toolchain's isKernelStatic function as part of the initial commit for Apple's arm64 target (r205100). That check was not in any of Apple's internal code and no one here knows where it came from. It has been harmless because "-static" does not change much, if anything, for arm64 iOS code, but it makes no sense to keep this check. llvm-svn: 228673
2015-02-10 14:33:04 +08:00
// CHECK-NO-STATIC-NOT: "-static"
//
Completely re-work how the Clang driver interprets PIC and PIE options. There were numerous issues here that were all entangled, and so I've tried to do a general simplification of the logic. 1) The logic was mimicing actual GCC bugs, rather than "features". These have been fixed in trunk GCC, and this fixes Clang as well. Notably, the logic was always intended to be last-match-wins like any other flag. 2) The logic for handling '-mdynamic-no-pic' was preposterously unclear. It also allowed the use of this flag on non-Darwin platforms where it has no actual meaning. Now this option is handled directly based on tests of how llvm-gcc behaves, and it is only supported on Darwin. 3) The APIs for the Driver's ToolChains had the implementation ugliness of dynamic-no-pic leaking through them. They also had the implementation details of the LLVM relocation model flag names leaking through. 4) The actual results of passing these flags was incorrect on Darwin in many cases. For example, Darwin *always* uses PIC level 2 if it uses in PIC level, and Darwin *always* uses PIC on 64-bit regardless of the flags specified, including -fPIE. Darwin never compiles in PIE mode, but it can *link* in PIE mode. 5) Also, PIC was not always being enabled even when PIE was. This isn't a supported mode at all and may have caused some fallout in builds with complex PIC and PIE interactions. The result is (I hope) cleaner and clearer for readers. I've also left comments and tests about some of the truly strage behavior that is observed on Darwin platforms. We have no real testing of Windows platforms and PIC, but I don't have the tools handy to figure that out. Hopefully others can beef up our testing here. Unfortunately, I can't test this for every platform. =/ If folks have dependencies on these flags that aren't covered by tests, they may break. I've audited and ensured that all the changes in behavior of the existing tests are intentional and good. In particular I've tried to make sure the Darwin behavior (which is more suprising than the Linux behavior) also matches that of 'gcc' on my mac. llvm-svn: 168297
2012-11-19 11:52:03 +08:00
// CHECK-PIE1: "-mrelocation-model" "pic"
// CHECK-PIE1: "-pic-level" "1"
Teach Clang about PIE compilations. This is the first step of PR12380. First, this patch cleans up the parsing of the PIC and PIE family of options in the driver. The existing logic failed to claim arguments all over the place resulting in kludges that marked the options as unused. Instead actually walk all of the arguments and claim them properly. We now treat -f{,no-}{pic,PIC,pie,PIE} as a single set, accepting the last one on the commandline. Previously there were lots of ordering bugs that could creep in due to the nature of the parsing. Let me know if folks would like weird things such as "-fPIE -fno-pic" to turn on PIE, but disable full PIC. This doesn't make any sense to me, but we could in theory support it. Options that seem to have intentional "trump" status (-static, -mkernel, etc) continue to do so and are commented as such. Next, a -pie-level flag is threaded into the frontend, rigged to a language option, and handled preprocessor, setting up the appropriate defines. We'll now have the correct defines when compiling with -fpie. The one place outside of the preprocessor that was inspecting the PIC level (as opposed to the relocation model, which is set and handled separately, yay!) is in the GNU ObjC runtime. I changed it to exactly preserve existing behavior. If folks want to change its behavior in the face of PIE, they can do that in a separate patch. Essentially the only functionality changed here is the preprocessor defines and bug-fixes to the argument management. Tests have been updated and extended to test all of this a bit more thoroughly. llvm-svn: 154291
2012-04-09 00:40:35 +08:00
// CHECK-PIE1: "-pie-level" "1"
//
Completely re-work how the Clang driver interprets PIC and PIE options. There were numerous issues here that were all entangled, and so I've tried to do a general simplification of the logic. 1) The logic was mimicing actual GCC bugs, rather than "features". These have been fixed in trunk GCC, and this fixes Clang as well. Notably, the logic was always intended to be last-match-wins like any other flag. 2) The logic for handling '-mdynamic-no-pic' was preposterously unclear. It also allowed the use of this flag on non-Darwin platforms where it has no actual meaning. Now this option is handled directly based on tests of how llvm-gcc behaves, and it is only supported on Darwin. 3) The APIs for the Driver's ToolChains had the implementation ugliness of dynamic-no-pic leaking through them. They also had the implementation details of the LLVM relocation model flag names leaking through. 4) The actual results of passing these flags was incorrect on Darwin in many cases. For example, Darwin *always* uses PIC level 2 if it uses in PIC level, and Darwin *always* uses PIC on 64-bit regardless of the flags specified, including -fPIE. Darwin never compiles in PIE mode, but it can *link* in PIE mode. 5) Also, PIC was not always being enabled even when PIE was. This isn't a supported mode at all and may have caused some fallout in builds with complex PIC and PIE interactions. The result is (I hope) cleaner and clearer for readers. I've also left comments and tests about some of the truly strage behavior that is observed on Darwin platforms. We have no real testing of Windows platforms and PIC, but I don't have the tools handy to figure that out. Hopefully others can beef up our testing here. Unfortunately, I can't test this for every platform. =/ If folks have dependencies on these flags that aren't covered by tests, they may break. I've audited and ensured that all the changes in behavior of the existing tests are intentional and good. In particular I've tried to make sure the Darwin behavior (which is more suprising than the Linux behavior) also matches that of 'gcc' on my mac. llvm-svn: 168297
2012-11-19 11:52:03 +08:00
// CHECK-PIE2: "-mrelocation-model" "pic"
// CHECK-PIE2: "-pic-level" "2"
Teach Clang about PIE compilations. This is the first step of PR12380. First, this patch cleans up the parsing of the PIC and PIE family of options in the driver. The existing logic failed to claim arguments all over the place resulting in kludges that marked the options as unused. Instead actually walk all of the arguments and claim them properly. We now treat -f{,no-}{pic,PIC,pie,PIE} as a single set, accepting the last one on the commandline. Previously there were lots of ordering bugs that could creep in due to the nature of the parsing. Let me know if folks would like weird things such as "-fPIE -fno-pic" to turn on PIE, but disable full PIC. This doesn't make any sense to me, but we could in theory support it. Options that seem to have intentional "trump" status (-static, -mkernel, etc) continue to do so and are commented as such. Next, a -pie-level flag is threaded into the frontend, rigged to a language option, and handled preprocessor, setting up the appropriate defines. We'll now have the correct defines when compiling with -fpie. The one place outside of the preprocessor that was inspecting the PIC level (as opposed to the relocation model, which is set and handled separately, yay!) is in the GNU ObjC runtime. I changed it to exactly preserve existing behavior. If folks want to change its behavior in the face of PIE, they can do that in a separate patch. Essentially the only functionality changed here is the preprocessor defines and bug-fixes to the argument management. Tests have been updated and extended to test all of this a bit more thoroughly. llvm-svn: 154291
2012-04-09 00:40:35 +08:00
// CHECK-PIE2: "-pie-level" "2"
//
Completely re-work how the Clang driver interprets PIC and PIE options. There were numerous issues here that were all entangled, and so I've tried to do a general simplification of the logic. 1) The logic was mimicing actual GCC bugs, rather than "features". These have been fixed in trunk GCC, and this fixes Clang as well. Notably, the logic was always intended to be last-match-wins like any other flag. 2) The logic for handling '-mdynamic-no-pic' was preposterously unclear. It also allowed the use of this flag on non-Darwin platforms where it has no actual meaning. Now this option is handled directly based on tests of how llvm-gcc behaves, and it is only supported on Darwin. 3) The APIs for the Driver's ToolChains had the implementation ugliness of dynamic-no-pic leaking through them. They also had the implementation details of the LLVM relocation model flag names leaking through. 4) The actual results of passing these flags was incorrect on Darwin in many cases. For example, Darwin *always* uses PIC level 2 if it uses in PIC level, and Darwin *always* uses PIC on 64-bit regardless of the flags specified, including -fPIE. Darwin never compiles in PIE mode, but it can *link* in PIE mode. 5) Also, PIC was not always being enabled even when PIE was. This isn't a supported mode at all and may have caused some fallout in builds with complex PIC and PIE interactions. The result is (I hope) cleaner and clearer for readers. I've also left comments and tests about some of the truly strage behavior that is observed on Darwin platforms. We have no real testing of Windows platforms and PIC, but I don't have the tools handy to figure that out. Hopefully others can beef up our testing here. Unfortunately, I can't test this for every platform. =/ If folks have dependencies on these flags that aren't covered by tests, they may break. I've audited and ensured that all the changes in behavior of the existing tests are intentional and good. In particular I've tried to make sure the Darwin behavior (which is more suprising than the Linux behavior) also matches that of 'gcc' on my mac. llvm-svn: 168297
2012-11-19 11:52:03 +08:00
// CHECK-PIE-LD: "{{.*}}ld{{(.exe)?}}"
// CHECK-PIE-LD: "-pie"
// CHECK-PIE-LD: "Scrt1.o" "crti.o" "crtbeginS.o"
// CHECK-PIE-LD: "crtendS.o" "crtn.o"
//
Default to PIE and using stack protection on OpenBSD. Patch by Matthew Dempsky. llvm-svn: 183295
2013-06-05 12:28:55 +08:00
// CHECK-NOPIE-LD: "-nopie"
//
Completely re-work how the Clang driver interprets PIC and PIE options. There were numerous issues here that were all entangled, and so I've tried to do a general simplification of the logic. 1) The logic was mimicing actual GCC bugs, rather than "features". These have been fixed in trunk GCC, and this fixes Clang as well. Notably, the logic was always intended to be last-match-wins like any other flag. 2) The logic for handling '-mdynamic-no-pic' was preposterously unclear. It also allowed the use of this flag on non-Darwin platforms where it has no actual meaning. Now this option is handled directly based on tests of how llvm-gcc behaves, and it is only supported on Darwin. 3) The APIs for the Driver's ToolChains had the implementation ugliness of dynamic-no-pic leaking through them. They also had the implementation details of the LLVM relocation model flag names leaking through. 4) The actual results of passing these flags was incorrect on Darwin in many cases. For example, Darwin *always* uses PIC level 2 if it uses in PIC level, and Darwin *always* uses PIC on 64-bit regardless of the flags specified, including -fPIE. Darwin never compiles in PIE mode, but it can *link* in PIE mode. 5) Also, PIC was not always being enabled even when PIE was. This isn't a supported mode at all and may have caused some fallout in builds with complex PIC and PIE interactions. The result is (I hope) cleaner and clearer for readers. I've also left comments and tests about some of the truly strage behavior that is observed on Darwin platforms. We have no real testing of Windows platforms and PIC, but I don't have the tools handy to figure that out. Hopefully others can beef up our testing here. Unfortunately, I can't test this for every platform. =/ If folks have dependencies on these flags that aren't covered by tests, they may break. I've audited and ensured that all the changes in behavior of the existing tests are intentional and good. In particular I've tried to make sure the Darwin behavior (which is more suprising than the Linux behavior) also matches that of 'gcc' on my mac. llvm-svn: 168297
2012-11-19 11:52:03 +08:00
// CHECK-DYNAMIC-NO-PIC-32: "-mrelocation-model" "dynamic-no-pic"
// CHECK-DYNAMIC-NO-PIC-32-NOT: "-pic-level"
// CHECK-DYNAMIC-NO-PIC-32-NOT: "-pie-level"
//
// CHECK-DYNAMIC-NO-PIC-64: "-mrelocation-model" "dynamic-no-pic"
// CHECK-DYNAMIC-NO-PIC-64: "-pic-level" "2"
// CHECK-DYNAMIC-NO-PIC-64-NOT: "-pie-level"
//
// CHECK-NON-DARWIN-DYNAMIC-NO-PIC: error: unsupported option '-mdynamic-no-pic' for target 'i386-unknown-unknown'
Link to crtend.S when PIE in the FreeBSD driver. Patch by Brooks Davis! llvm-svn: 163388
2012-09-07 21:36:21 +08:00
//
Do not pass -pie flag to linker if -shared specified. This matches the gcc driver and makes it possible to add -pie to $CC or similar and have it apply in the right places. llvm-svn: 172753
2013-01-18 04:17:16 +08:00
// CHECK-NO-PIE-NOT: "-pie"
//
One more follow-up to r179082 - parse PIC/PIE arguments even on platfroms that force default PIC (like Darwin x86-64), otherwise specifying -fPIC will produce bogus unused argument warning llvm-svn: 179092
2013-04-09 20:28:19 +08:00
// CHECK-NO-UNUSED-ARG-NOT: argument unused during compilation
//
Teach Clang about PIE compilations. This is the first step of PR12380. First, this patch cleans up the parsing of the PIC and PIE family of options in the driver. The existing logic failed to claim arguments all over the place resulting in kludges that marked the options as unused. Instead actually walk all of the arguments and claim them properly. We now treat -f{,no-}{pic,PIC,pie,PIE} as a single set, accepting the last one on the commandline. Previously there were lots of ordering bugs that could creep in due to the nature of the parsing. Let me know if folks would like weird things such as "-fPIE -fno-pic" to turn on PIE, but disable full PIC. This doesn't make any sense to me, but we could in theory support it. Options that seem to have intentional "trump" status (-static, -mkernel, etc) continue to do so and are commented as such. Next, a -pie-level flag is threaded into the frontend, rigged to a language option, and handled preprocessor, setting up the appropriate defines. We'll now have the correct defines when compiling with -fpie. The one place outside of the preprocessor that was inspecting the PIC level (as opposed to the relocation model, which is set and handled separately, yay!) is in the GNU ObjC runtime. I changed it to exactly preserve existing behavior. If folks want to change its behavior in the face of PIE, they can do that in a separate patch. Essentially the only functionality changed here is the preprocessor defines and bug-fixes to the argument management. Tests have been updated and extended to test all of this a bit more thoroughly. llvm-svn: 154291
2012-04-09 00:40:35 +08:00
// RUN: %clang -c %s -target i386-unknown-unknown -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-NO-PIC
// RUN: %clang -c %s -target i386-unknown-unknown -fpic -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-PIC1
// RUN: %clang -c %s -target i386-unknown-unknown -fPIC -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-PIC2
// RUN: %clang -c %s -target i386-unknown-unknown -fpie -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-PIE1
// RUN: %clang -c %s -target i386-unknown-unknown -fPIE -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-PIE2
Completely re-work how the Clang driver interprets PIC and PIE options. There were numerous issues here that were all entangled, and so I've tried to do a general simplification of the logic. 1) The logic was mimicing actual GCC bugs, rather than "features". These have been fixed in trunk GCC, and this fixes Clang as well. Notably, the logic was always intended to be last-match-wins like any other flag. 2) The logic for handling '-mdynamic-no-pic' was preposterously unclear. It also allowed the use of this flag on non-Darwin platforms where it has no actual meaning. Now this option is handled directly based on tests of how llvm-gcc behaves, and it is only supported on Darwin. 3) The APIs for the Driver's ToolChains had the implementation ugliness of dynamic-no-pic leaking through them. They also had the implementation details of the LLVM relocation model flag names leaking through. 4) The actual results of passing these flags was incorrect on Darwin in many cases. For example, Darwin *always* uses PIC level 2 if it uses in PIC level, and Darwin *always* uses PIC on 64-bit regardless of the flags specified, including -fPIE. Darwin never compiles in PIE mode, but it can *link* in PIE mode. 5) Also, PIC was not always being enabled even when PIE was. This isn't a supported mode at all and may have caused some fallout in builds with complex PIC and PIE interactions. The result is (I hope) cleaner and clearer for readers. I've also left comments and tests about some of the truly strage behavior that is observed on Darwin platforms. We have no real testing of Windows platforms and PIC, but I don't have the tools handy to figure that out. Hopefully others can beef up our testing here. Unfortunately, I can't test this for every platform. =/ If folks have dependencies on these flags that aren't covered by tests, they may break. I've audited and ensured that all the changes in behavior of the existing tests are intentional and good. In particular I've tried to make sure the Darwin behavior (which is more suprising than the Linux behavior) also matches that of 'gcc' on my mac. llvm-svn: 168297
2012-11-19 11:52:03 +08:00
//
// Check that PIC and PIE flags obey last-match-wins. If the last flag is
// a no-* variant, regardless of which variant or which flags precede it, we
// get no PIC.
Teach Clang about PIE compilations. This is the first step of PR12380. First, this patch cleans up the parsing of the PIC and PIE family of options in the driver. The existing logic failed to claim arguments all over the place resulting in kludges that marked the options as unused. Instead actually walk all of the arguments and claim them properly. We now treat -f{,no-}{pic,PIC,pie,PIE} as a single set, accepting the last one on the commandline. Previously there were lots of ordering bugs that could creep in due to the nature of the parsing. Let me know if folks would like weird things such as "-fPIE -fno-pic" to turn on PIE, but disable full PIC. This doesn't make any sense to me, but we could in theory support it. Options that seem to have intentional "trump" status (-static, -mkernel, etc) continue to do so and are commented as such. Next, a -pie-level flag is threaded into the frontend, rigged to a language option, and handled preprocessor, setting up the appropriate defines. We'll now have the correct defines when compiling with -fpie. The one place outside of the preprocessor that was inspecting the PIC level (as opposed to the relocation model, which is set and handled separately, yay!) is in the GNU ObjC runtime. I changed it to exactly preserve existing behavior. If folks want to change its behavior in the face of PIE, they can do that in a separate patch. Essentially the only functionality changed here is the preprocessor defines and bug-fixes to the argument management. Tests have been updated and extended to test all of this a bit more thoroughly. llvm-svn: 154291
2012-04-09 00:40:35 +08:00
// RUN: %clang -c %s -target i386-unknown-unknown -fpic -fno-pic -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-NO-PIC
Completely re-work how the Clang driver interprets PIC and PIE options. There were numerous issues here that were all entangled, and so I've tried to do a general simplification of the logic. 1) The logic was mimicing actual GCC bugs, rather than "features". These have been fixed in trunk GCC, and this fixes Clang as well. Notably, the logic was always intended to be last-match-wins like any other flag. 2) The logic for handling '-mdynamic-no-pic' was preposterously unclear. It also allowed the use of this flag on non-Darwin platforms where it has no actual meaning. Now this option is handled directly based on tests of how llvm-gcc behaves, and it is only supported on Darwin. 3) The APIs for the Driver's ToolChains had the implementation ugliness of dynamic-no-pic leaking through them. They also had the implementation details of the LLVM relocation model flag names leaking through. 4) The actual results of passing these flags was incorrect on Darwin in many cases. For example, Darwin *always* uses PIC level 2 if it uses in PIC level, and Darwin *always* uses PIC on 64-bit regardless of the flags specified, including -fPIE. Darwin never compiles in PIE mode, but it can *link* in PIE mode. 5) Also, PIC was not always being enabled even when PIE was. This isn't a supported mode at all and may have caused some fallout in builds with complex PIC and PIE interactions. The result is (I hope) cleaner and clearer for readers. I've also left comments and tests about some of the truly strage behavior that is observed on Darwin platforms. We have no real testing of Windows platforms and PIC, but I don't have the tools handy to figure that out. Hopefully others can beef up our testing here. Unfortunately, I can't test this for every platform. =/ If folks have dependencies on these flags that aren't covered by tests, they may break. I've audited and ensured that all the changes in behavior of the existing tests are intentional and good. In particular I've tried to make sure the Darwin behavior (which is more suprising than the Linux behavior) also matches that of 'gcc' on my mac. llvm-svn: 168297
2012-11-19 11:52:03 +08:00
// RUN: %clang -c %s -target i386-unknown-unknown -fPIC -fno-pic -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-NO-PIC
// RUN: %clang -c %s -target i386-unknown-unknown -fpie -fno-pic -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-NO-PIC
// RUN: %clang -c %s -target i386-unknown-unknown -fPIE -fno-pic -### 2>&1 \
Teach Clang about PIE compilations. This is the first step of PR12380. First, this patch cleans up the parsing of the PIC and PIE family of options in the driver. The existing logic failed to claim arguments all over the place resulting in kludges that marked the options as unused. Instead actually walk all of the arguments and claim them properly. We now treat -f{,no-}{pic,PIC,pie,PIE} as a single set, accepting the last one on the commandline. Previously there were lots of ordering bugs that could creep in due to the nature of the parsing. Let me know if folks would like weird things such as "-fPIE -fno-pic" to turn on PIE, but disable full PIC. This doesn't make any sense to me, but we could in theory support it. Options that seem to have intentional "trump" status (-static, -mkernel, etc) continue to do so and are commented as such. Next, a -pie-level flag is threaded into the frontend, rigged to a language option, and handled preprocessor, setting up the appropriate defines. We'll now have the correct defines when compiling with -fpie. The one place outside of the preprocessor that was inspecting the PIC level (as opposed to the relocation model, which is set and handled separately, yay!) is in the GNU ObjC runtime. I changed it to exactly preserve existing behavior. If folks want to change its behavior in the face of PIE, they can do that in a separate patch. Essentially the only functionality changed here is the preprocessor defines and bug-fixes to the argument management. Tests have been updated and extended to test all of this a bit more thoroughly. llvm-svn: 154291
2012-04-09 00:40:35 +08:00
// RUN: | FileCheck %s --check-prefix=CHECK-NO-PIC
// RUN: %clang -c %s -target i386-unknown-unknown -fpic -fno-PIC -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-NO-PIC
Completely re-work how the Clang driver interprets PIC and PIE options. There were numerous issues here that were all entangled, and so I've tried to do a general simplification of the logic. 1) The logic was mimicing actual GCC bugs, rather than "features". These have been fixed in trunk GCC, and this fixes Clang as well. Notably, the logic was always intended to be last-match-wins like any other flag. 2) The logic for handling '-mdynamic-no-pic' was preposterously unclear. It also allowed the use of this flag on non-Darwin platforms where it has no actual meaning. Now this option is handled directly based on tests of how llvm-gcc behaves, and it is only supported on Darwin. 3) The APIs for the Driver's ToolChains had the implementation ugliness of dynamic-no-pic leaking through them. They also had the implementation details of the LLVM relocation model flag names leaking through. 4) The actual results of passing these flags was incorrect on Darwin in many cases. For example, Darwin *always* uses PIC level 2 if it uses in PIC level, and Darwin *always* uses PIC on 64-bit regardless of the flags specified, including -fPIE. Darwin never compiles in PIE mode, but it can *link* in PIE mode. 5) Also, PIC was not always being enabled even when PIE was. This isn't a supported mode at all and may have caused some fallout in builds with complex PIC and PIE interactions. The result is (I hope) cleaner and clearer for readers. I've also left comments and tests about some of the truly strage behavior that is observed on Darwin platforms. We have no real testing of Windows platforms and PIC, but I don't have the tools handy to figure that out. Hopefully others can beef up our testing here. Unfortunately, I can't test this for every platform. =/ If folks have dependencies on these flags that aren't covered by tests, they may break. I've audited and ensured that all the changes in behavior of the existing tests are intentional and good. In particular I've tried to make sure the Darwin behavior (which is more suprising than the Linux behavior) also matches that of 'gcc' on my mac. llvm-svn: 168297
2012-11-19 11:52:03 +08:00
// RUN: %clang -c %s -target i386-unknown-unknown -fPIC -fno-PIC -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-NO-PIC
// RUN: %clang -c %s -target i386-unknown-unknown -fpie -fno-PIC -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-NO-PIC
// RUN: %clang -c %s -target i386-unknown-unknown -fPIE -fno-PIC -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-NO-PIC
// RUN: %clang -c %s -target i386-unknown-unknown -fpic -fno-pie -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-NO-PIC
// RUN: %clang -c %s -target i386-unknown-unknown -fPIC -fno-pie -### 2>&1 \
Teach Clang about PIE compilations. This is the first step of PR12380. First, this patch cleans up the parsing of the PIC and PIE family of options in the driver. The existing logic failed to claim arguments all over the place resulting in kludges that marked the options as unused. Instead actually walk all of the arguments and claim them properly. We now treat -f{,no-}{pic,PIC,pie,PIE} as a single set, accepting the last one on the commandline. Previously there were lots of ordering bugs that could creep in due to the nature of the parsing. Let me know if folks would like weird things such as "-fPIE -fno-pic" to turn on PIE, but disable full PIC. This doesn't make any sense to me, but we could in theory support it. Options that seem to have intentional "trump" status (-static, -mkernel, etc) continue to do so and are commented as such. Next, a -pie-level flag is threaded into the frontend, rigged to a language option, and handled preprocessor, setting up the appropriate defines. We'll now have the correct defines when compiling with -fpie. The one place outside of the preprocessor that was inspecting the PIC level (as opposed to the relocation model, which is set and handled separately, yay!) is in the GNU ObjC runtime. I changed it to exactly preserve existing behavior. If folks want to change its behavior in the face of PIE, they can do that in a separate patch. Essentially the only functionality changed here is the preprocessor defines and bug-fixes to the argument management. Tests have been updated and extended to test all of this a bit more thoroughly. llvm-svn: 154291
2012-04-09 00:40:35 +08:00
// RUN: | FileCheck %s --check-prefix=CHECK-NO-PIC
// RUN: %clang -c %s -target i386-unknown-unknown -fpie -fno-pie -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-NO-PIC
Completely re-work how the Clang driver interprets PIC and PIE options. There were numerous issues here that were all entangled, and so I've tried to do a general simplification of the logic. 1) The logic was mimicing actual GCC bugs, rather than "features". These have been fixed in trunk GCC, and this fixes Clang as well. Notably, the logic was always intended to be last-match-wins like any other flag. 2) The logic for handling '-mdynamic-no-pic' was preposterously unclear. It also allowed the use of this flag on non-Darwin platforms where it has no actual meaning. Now this option is handled directly based on tests of how llvm-gcc behaves, and it is only supported on Darwin. 3) The APIs for the Driver's ToolChains had the implementation ugliness of dynamic-no-pic leaking through them. They also had the implementation details of the LLVM relocation model flag names leaking through. 4) The actual results of passing these flags was incorrect on Darwin in many cases. For example, Darwin *always* uses PIC level 2 if it uses in PIC level, and Darwin *always* uses PIC on 64-bit regardless of the flags specified, including -fPIE. Darwin never compiles in PIE mode, but it can *link* in PIE mode. 5) Also, PIC was not always being enabled even when PIE was. This isn't a supported mode at all and may have caused some fallout in builds with complex PIC and PIE interactions. The result is (I hope) cleaner and clearer for readers. I've also left comments and tests about some of the truly strage behavior that is observed on Darwin platforms. We have no real testing of Windows platforms and PIC, but I don't have the tools handy to figure that out. Hopefully others can beef up our testing here. Unfortunately, I can't test this for every platform. =/ If folks have dependencies on these flags that aren't covered by tests, they may break. I've audited and ensured that all the changes in behavior of the existing tests are intentional and good. In particular I've tried to make sure the Darwin behavior (which is more suprising than the Linux behavior) also matches that of 'gcc' on my mac. llvm-svn: 168297
2012-11-19 11:52:03 +08:00
// RUN: %clang -c %s -target i386-unknown-unknown -fPIE -fno-pie -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-NO-PIC
// RUN: %clang -c %s -target i386-unknown-unknown -fpic -fno-PIE -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-NO-PIC
// RUN: %clang -c %s -target i386-unknown-unknown -fPIC -fno-PIE -### 2>&1 \
Teach Clang about PIE compilations. This is the first step of PR12380. First, this patch cleans up the parsing of the PIC and PIE family of options in the driver. The existing logic failed to claim arguments all over the place resulting in kludges that marked the options as unused. Instead actually walk all of the arguments and claim them properly. We now treat -f{,no-}{pic,PIC,pie,PIE} as a single set, accepting the last one on the commandline. Previously there were lots of ordering bugs that could creep in due to the nature of the parsing. Let me know if folks would like weird things such as "-fPIE -fno-pic" to turn on PIE, but disable full PIC. This doesn't make any sense to me, but we could in theory support it. Options that seem to have intentional "trump" status (-static, -mkernel, etc) continue to do so and are commented as such. Next, a -pie-level flag is threaded into the frontend, rigged to a language option, and handled preprocessor, setting up the appropriate defines. We'll now have the correct defines when compiling with -fpie. The one place outside of the preprocessor that was inspecting the PIC level (as opposed to the relocation model, which is set and handled separately, yay!) is in the GNU ObjC runtime. I changed it to exactly preserve existing behavior. If folks want to change its behavior in the face of PIE, they can do that in a separate patch. Essentially the only functionality changed here is the preprocessor defines and bug-fixes to the argument management. Tests have been updated and extended to test all of this a bit more thoroughly. llvm-svn: 154291
2012-04-09 00:40:35 +08:00
// RUN: | FileCheck %s --check-prefix=CHECK-NO-PIC
// RUN: %clang -c %s -target i386-unknown-unknown -fpie -fno-PIE -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-NO-PIC
Completely re-work how the Clang driver interprets PIC and PIE options. There were numerous issues here that were all entangled, and so I've tried to do a general simplification of the logic. 1) The logic was mimicing actual GCC bugs, rather than "features". These have been fixed in trunk GCC, and this fixes Clang as well. Notably, the logic was always intended to be last-match-wins like any other flag. 2) The logic for handling '-mdynamic-no-pic' was preposterously unclear. It also allowed the use of this flag on non-Darwin platforms where it has no actual meaning. Now this option is handled directly based on tests of how llvm-gcc behaves, and it is only supported on Darwin. 3) The APIs for the Driver's ToolChains had the implementation ugliness of dynamic-no-pic leaking through them. They also had the implementation details of the LLVM relocation model flag names leaking through. 4) The actual results of passing these flags was incorrect on Darwin in many cases. For example, Darwin *always* uses PIC level 2 if it uses in PIC level, and Darwin *always* uses PIC on 64-bit regardless of the flags specified, including -fPIE. Darwin never compiles in PIE mode, but it can *link* in PIE mode. 5) Also, PIC was not always being enabled even when PIE was. This isn't a supported mode at all and may have caused some fallout in builds with complex PIC and PIE interactions. The result is (I hope) cleaner and clearer for readers. I've also left comments and tests about some of the truly strage behavior that is observed on Darwin platforms. We have no real testing of Windows platforms and PIC, but I don't have the tools handy to figure that out. Hopefully others can beef up our testing here. Unfortunately, I can't test this for every platform. =/ If folks have dependencies on these flags that aren't covered by tests, they may break. I've audited and ensured that all the changes in behavior of the existing tests are intentional and good. In particular I've tried to make sure the Darwin behavior (which is more suprising than the Linux behavior) also matches that of 'gcc' on my mac. llvm-svn: 168297
2012-11-19 11:52:03 +08:00
// RUN: %clang -c %s -target i386-unknown-unknown -fPIE -fno-PIE -### 2>&1 \
Teach Clang about PIE compilations. This is the first step of PR12380. First, this patch cleans up the parsing of the PIC and PIE family of options in the driver. The existing logic failed to claim arguments all over the place resulting in kludges that marked the options as unused. Instead actually walk all of the arguments and claim them properly. We now treat -f{,no-}{pic,PIC,pie,PIE} as a single set, accepting the last one on the commandline. Previously there were lots of ordering bugs that could creep in due to the nature of the parsing. Let me know if folks would like weird things such as "-fPIE -fno-pic" to turn on PIE, but disable full PIC. This doesn't make any sense to me, but we could in theory support it. Options that seem to have intentional "trump" status (-static, -mkernel, etc) continue to do so and are commented as such. Next, a -pie-level flag is threaded into the frontend, rigged to a language option, and handled preprocessor, setting up the appropriate defines. We'll now have the correct defines when compiling with -fpie. The one place outside of the preprocessor that was inspecting the PIC level (as opposed to the relocation model, which is set and handled separately, yay!) is in the GNU ObjC runtime. I changed it to exactly preserve existing behavior. If folks want to change its behavior in the face of PIE, they can do that in a separate patch. Essentially the only functionality changed here is the preprocessor defines and bug-fixes to the argument management. Tests have been updated and extended to test all of this a bit more thoroughly. llvm-svn: 154291
2012-04-09 00:40:35 +08:00
// RUN: | FileCheck %s --check-prefix=CHECK-NO-PIC
//
Completely re-work how the Clang driver interprets PIC and PIE options. There were numerous issues here that were all entangled, and so I've tried to do a general simplification of the logic. 1) The logic was mimicing actual GCC bugs, rather than "features". These have been fixed in trunk GCC, and this fixes Clang as well. Notably, the logic was always intended to be last-match-wins like any other flag. 2) The logic for handling '-mdynamic-no-pic' was preposterously unclear. It also allowed the use of this flag on non-Darwin platforms where it has no actual meaning. Now this option is handled directly based on tests of how llvm-gcc behaves, and it is only supported on Darwin. 3) The APIs for the Driver's ToolChains had the implementation ugliness of dynamic-no-pic leaking through them. They also had the implementation details of the LLVM relocation model flag names leaking through. 4) The actual results of passing these flags was incorrect on Darwin in many cases. For example, Darwin *always* uses PIC level 2 if it uses in PIC level, and Darwin *always* uses PIC on 64-bit regardless of the flags specified, including -fPIE. Darwin never compiles in PIE mode, but it can *link* in PIE mode. 5) Also, PIC was not always being enabled even when PIE was. This isn't a supported mode at all and may have caused some fallout in builds with complex PIC and PIE interactions. The result is (I hope) cleaner and clearer for readers. I've also left comments and tests about some of the truly strage behavior that is observed on Darwin platforms. We have no real testing of Windows platforms and PIC, but I don't have the tools handy to figure that out. Hopefully others can beef up our testing here. Unfortunately, I can't test this for every platform. =/ If folks have dependencies on these flags that aren't covered by tests, they may break. I've audited and ensured that all the changes in behavior of the existing tests are intentional and good. In particular I've tried to make sure the Darwin behavior (which is more suprising than the Linux behavior) also matches that of 'gcc' on my mac. llvm-svn: 168297
2012-11-19 11:52:03 +08:00
// Last-match-wins where both pic and pie are specified.
This patch makes the behavior of clang consistent with the behavior of gcc 4.6 in cases where both -fPIC and -fPIE is used. - Separately check if -fPIE was specified in the command line and define both __PIC__ and __PIE__ when -fPIE is used. We need to check this separately because -fPIE will infer -fPIC even if its not explicitly used. - Fixed existing tests. - Added new tests for cases where both -fPIC and -fPIE is used. Author: Tareq A. Siraj <tareq.a.siraj@intel.com> Fixes: PR13221 Review: http://llvm-reviews.chandlerc.com/D94 llvm-svn: 167846
2012-11-13 23:32:35 +08:00
// RUN: %clang -c %s -target i386-unknown-unknown -fpie -fpic -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-PIC1
Completely re-work how the Clang driver interprets PIC and PIE options. There were numerous issues here that were all entangled, and so I've tried to do a general simplification of the logic. 1) The logic was mimicing actual GCC bugs, rather than "features". These have been fixed in trunk GCC, and this fixes Clang as well. Notably, the logic was always intended to be last-match-wins like any other flag. 2) The logic for handling '-mdynamic-no-pic' was preposterously unclear. It also allowed the use of this flag on non-Darwin platforms where it has no actual meaning. Now this option is handled directly based on tests of how llvm-gcc behaves, and it is only supported on Darwin. 3) The APIs for the Driver's ToolChains had the implementation ugliness of dynamic-no-pic leaking through them. They also had the implementation details of the LLVM relocation model flag names leaking through. 4) The actual results of passing these flags was incorrect on Darwin in many cases. For example, Darwin *always* uses PIC level 2 if it uses in PIC level, and Darwin *always* uses PIC on 64-bit regardless of the flags specified, including -fPIE. Darwin never compiles in PIE mode, but it can *link* in PIE mode. 5) Also, PIC was not always being enabled even when PIE was. This isn't a supported mode at all and may have caused some fallout in builds with complex PIC and PIE interactions. The result is (I hope) cleaner and clearer for readers. I've also left comments and tests about some of the truly strage behavior that is observed on Darwin platforms. We have no real testing of Windows platforms and PIC, but I don't have the tools handy to figure that out. Hopefully others can beef up our testing here. Unfortunately, I can't test this for every platform. =/ If folks have dependencies on these flags that aren't covered by tests, they may break. I've audited and ensured that all the changes in behavior of the existing tests are intentional and good. In particular I've tried to make sure the Darwin behavior (which is more suprising than the Linux behavior) also matches that of 'gcc' on my mac. llvm-svn: 168297
2012-11-19 11:52:03 +08:00
// RUN: %clang -c %s -target i386-unknown-unknown -fPIE -fpic -### 2>&1 \
This patch makes the behavior of clang consistent with the behavior of gcc 4.6 in cases where both -fPIC and -fPIE is used. - Separately check if -fPIE was specified in the command line and define both __PIC__ and __PIE__ when -fPIE is used. We need to check this separately because -fPIE will infer -fPIC even if its not explicitly used. - Fixed existing tests. - Added new tests for cases where both -fPIC and -fPIE is used. Author: Tareq A. Siraj <tareq.a.siraj@intel.com> Fixes: PR13221 Review: http://llvm-reviews.chandlerc.com/D94 llvm-svn: 167846
2012-11-13 23:32:35 +08:00
// RUN: | FileCheck %s --check-prefix=CHECK-PIC1
// RUN: %clang -c %s -target i386-unknown-unknown -fpie -fPIC -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-PIC2
Completely re-work how the Clang driver interprets PIC and PIE options. There were numerous issues here that were all entangled, and so I've tried to do a general simplification of the logic. 1) The logic was mimicing actual GCC bugs, rather than "features". These have been fixed in trunk GCC, and this fixes Clang as well. Notably, the logic was always intended to be last-match-wins like any other flag. 2) The logic for handling '-mdynamic-no-pic' was preposterously unclear. It also allowed the use of this flag on non-Darwin platforms where it has no actual meaning. Now this option is handled directly based on tests of how llvm-gcc behaves, and it is only supported on Darwin. 3) The APIs for the Driver's ToolChains had the implementation ugliness of dynamic-no-pic leaking through them. They also had the implementation details of the LLVM relocation model flag names leaking through. 4) The actual results of passing these flags was incorrect on Darwin in many cases. For example, Darwin *always* uses PIC level 2 if it uses in PIC level, and Darwin *always* uses PIC on 64-bit regardless of the flags specified, including -fPIE. Darwin never compiles in PIE mode, but it can *link* in PIE mode. 5) Also, PIC was not always being enabled even when PIE was. This isn't a supported mode at all and may have caused some fallout in builds with complex PIC and PIE interactions. The result is (I hope) cleaner and clearer for readers. I've also left comments and tests about some of the truly strage behavior that is observed on Darwin platforms. We have no real testing of Windows platforms and PIC, but I don't have the tools handy to figure that out. Hopefully others can beef up our testing here. Unfortunately, I can't test this for every platform. =/ If folks have dependencies on these flags that aren't covered by tests, they may break. I've audited and ensured that all the changes in behavior of the existing tests are intentional and good. In particular I've tried to make sure the Darwin behavior (which is more suprising than the Linux behavior) also matches that of 'gcc' on my mac. llvm-svn: 168297
2012-11-19 11:52:03 +08:00
// RUN: %clang -c %s -target i386-unknown-unknown -fPIE -fPIC -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-PIC2
// RUN: %clang -c %s -target i386-unknown-unknown -fpic -fpie -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-PIE1
This patch makes the behavior of clang consistent with the behavior of gcc 4.6 in cases where both -fPIC and -fPIE is used. - Separately check if -fPIE was specified in the command line and define both __PIC__ and __PIE__ when -fPIE is used. We need to check this separately because -fPIE will infer -fPIC even if its not explicitly used. - Fixed existing tests. - Added new tests for cases where both -fPIC and -fPIE is used. Author: Tareq A. Siraj <tareq.a.siraj@intel.com> Fixes: PR13221 Review: http://llvm-reviews.chandlerc.com/D94 llvm-svn: 167846
2012-11-13 23:32:35 +08:00
// RUN: %clang -c %s -target i386-unknown-unknown -fPIC -fpie -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-PIE1
Completely re-work how the Clang driver interprets PIC and PIE options. There were numerous issues here that were all entangled, and so I've tried to do a general simplification of the logic. 1) The logic was mimicing actual GCC bugs, rather than "features". These have been fixed in trunk GCC, and this fixes Clang as well. Notably, the logic was always intended to be last-match-wins like any other flag. 2) The logic for handling '-mdynamic-no-pic' was preposterously unclear. It also allowed the use of this flag on non-Darwin platforms where it has no actual meaning. Now this option is handled directly based on tests of how llvm-gcc behaves, and it is only supported on Darwin. 3) The APIs for the Driver's ToolChains had the implementation ugliness of dynamic-no-pic leaking through them. They also had the implementation details of the LLVM relocation model flag names leaking through. 4) The actual results of passing these flags was incorrect on Darwin in many cases. For example, Darwin *always* uses PIC level 2 if it uses in PIC level, and Darwin *always* uses PIC on 64-bit regardless of the flags specified, including -fPIE. Darwin never compiles in PIE mode, but it can *link* in PIE mode. 5) Also, PIC was not always being enabled even when PIE was. This isn't a supported mode at all and may have caused some fallout in builds with complex PIC and PIE interactions. The result is (I hope) cleaner and clearer for readers. I've also left comments and tests about some of the truly strage behavior that is observed on Darwin platforms. We have no real testing of Windows platforms and PIC, but I don't have the tools handy to figure that out. Hopefully others can beef up our testing here. Unfortunately, I can't test this for every platform. =/ If folks have dependencies on these flags that aren't covered by tests, they may break. I've audited and ensured that all the changes in behavior of the existing tests are intentional and good. In particular I've tried to make sure the Darwin behavior (which is more suprising than the Linux behavior) also matches that of 'gcc' on my mac. llvm-svn: 168297
2012-11-19 11:52:03 +08:00
// RUN: %clang -c %s -target i386-unknown-unknown -fpic -fPIE -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-PIE2
This patch makes the behavior of clang consistent with the behavior of gcc 4.6 in cases where both -fPIC and -fPIE is used. - Separately check if -fPIE was specified in the command line and define both __PIC__ and __PIE__ when -fPIE is used. We need to check this separately because -fPIE will infer -fPIC even if its not explicitly used. - Fixed existing tests. - Added new tests for cases where both -fPIC and -fPIE is used. Author: Tareq A. Siraj <tareq.a.siraj@intel.com> Fixes: PR13221 Review: http://llvm-reviews.chandlerc.com/D94 llvm-svn: 167846
2012-11-13 23:32:35 +08:00
// RUN: %clang -c %s -target i386-unknown-unknown -fPIC -fPIE -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-PIE2
Completely re-work how the Clang driver interprets PIC and PIE options. There were numerous issues here that were all entangled, and so I've tried to do a general simplification of the logic. 1) The logic was mimicing actual GCC bugs, rather than "features". These have been fixed in trunk GCC, and this fixes Clang as well. Notably, the logic was always intended to be last-match-wins like any other flag. 2) The logic for handling '-mdynamic-no-pic' was preposterously unclear. It also allowed the use of this flag on non-Darwin platforms where it has no actual meaning. Now this option is handled directly based on tests of how llvm-gcc behaves, and it is only supported on Darwin. 3) The APIs for the Driver's ToolChains had the implementation ugliness of dynamic-no-pic leaking through them. They also had the implementation details of the LLVM relocation model flag names leaking through. 4) The actual results of passing these flags was incorrect on Darwin in many cases. For example, Darwin *always* uses PIC level 2 if it uses in PIC level, and Darwin *always* uses PIC on 64-bit regardless of the flags specified, including -fPIE. Darwin never compiles in PIE mode, but it can *link* in PIE mode. 5) Also, PIC was not always being enabled even when PIE was. This isn't a supported mode at all and may have caused some fallout in builds with complex PIC and PIE interactions. The result is (I hope) cleaner and clearer for readers. I've also left comments and tests about some of the truly strage behavior that is observed on Darwin platforms. We have no real testing of Windows platforms and PIC, but I don't have the tools handy to figure that out. Hopefully others can beef up our testing here. Unfortunately, I can't test this for every platform. =/ If folks have dependencies on these flags that aren't covered by tests, they may break. I've audited and ensured that all the changes in behavior of the existing tests are intentional and good. In particular I've tried to make sure the Darwin behavior (which is more suprising than the Linux behavior) also matches that of 'gcc' on my mac. llvm-svn: 168297
2012-11-19 11:52:03 +08:00
//
// Last-match-wins when selecting level 1 vs. level 2.
// RUN: %clang -c %s -target i386-unknown-unknown -fpic -fPIC -### 2>&1 \
This patch makes the behavior of clang consistent with the behavior of gcc 4.6 in cases where both -fPIC and -fPIE is used. - Separately check if -fPIE was specified in the command line and define both __PIC__ and __PIE__ when -fPIE is used. We need to check this separately because -fPIE will infer -fPIC even if its not explicitly used. - Fixed existing tests. - Added new tests for cases where both -fPIC and -fPIE is used. Author: Tareq A. Siraj <tareq.a.siraj@intel.com> Fixes: PR13221 Review: http://llvm-reviews.chandlerc.com/D94 llvm-svn: 167846
2012-11-13 23:32:35 +08:00
// RUN: | FileCheck %s --check-prefix=CHECK-PIC2
Completely re-work how the Clang driver interprets PIC and PIE options. There were numerous issues here that were all entangled, and so I've tried to do a general simplification of the logic. 1) The logic was mimicing actual GCC bugs, rather than "features". These have been fixed in trunk GCC, and this fixes Clang as well. Notably, the logic was always intended to be last-match-wins like any other flag. 2) The logic for handling '-mdynamic-no-pic' was preposterously unclear. It also allowed the use of this flag on non-Darwin platforms where it has no actual meaning. Now this option is handled directly based on tests of how llvm-gcc behaves, and it is only supported on Darwin. 3) The APIs for the Driver's ToolChains had the implementation ugliness of dynamic-no-pic leaking through them. They also had the implementation details of the LLVM relocation model flag names leaking through. 4) The actual results of passing these flags was incorrect on Darwin in many cases. For example, Darwin *always* uses PIC level 2 if it uses in PIC level, and Darwin *always* uses PIC on 64-bit regardless of the flags specified, including -fPIE. Darwin never compiles in PIE mode, but it can *link* in PIE mode. 5) Also, PIC was not always being enabled even when PIE was. This isn't a supported mode at all and may have caused some fallout in builds with complex PIC and PIE interactions. The result is (I hope) cleaner and clearer for readers. I've also left comments and tests about some of the truly strage behavior that is observed on Darwin platforms. We have no real testing of Windows platforms and PIC, but I don't have the tools handy to figure that out. Hopefully others can beef up our testing here. Unfortunately, I can't test this for every platform. =/ If folks have dependencies on these flags that aren't covered by tests, they may break. I've audited and ensured that all the changes in behavior of the existing tests are intentional and good. In particular I've tried to make sure the Darwin behavior (which is more suprising than the Linux behavior) also matches that of 'gcc' on my mac. llvm-svn: 168297
2012-11-19 11:52:03 +08:00
// RUN: %clang -c %s -target i386-unknown-unknown -fPIC -fpic -### 2>&1 \
This patch makes the behavior of clang consistent with the behavior of gcc 4.6 in cases where both -fPIC and -fPIE is used. - Separately check if -fPIE was specified in the command line and define both __PIC__ and __PIE__ when -fPIE is used. We need to check this separately because -fPIE will infer -fPIC even if its not explicitly used. - Fixed existing tests. - Added new tests for cases where both -fPIC and -fPIE is used. Author: Tareq A. Siraj <tareq.a.siraj@intel.com> Fixes: PR13221 Review: http://llvm-reviews.chandlerc.com/D94 llvm-svn: 167846
2012-11-13 23:32:35 +08:00
// RUN: | FileCheck %s --check-prefix=CHECK-PIC1
Completely re-work how the Clang driver interprets PIC and PIE options. There were numerous issues here that were all entangled, and so I've tried to do a general simplification of the logic. 1) The logic was mimicing actual GCC bugs, rather than "features". These have been fixed in trunk GCC, and this fixes Clang as well. Notably, the logic was always intended to be last-match-wins like any other flag. 2) The logic for handling '-mdynamic-no-pic' was preposterously unclear. It also allowed the use of this flag on non-Darwin platforms where it has no actual meaning. Now this option is handled directly based on tests of how llvm-gcc behaves, and it is only supported on Darwin. 3) The APIs for the Driver's ToolChains had the implementation ugliness of dynamic-no-pic leaking through them. They also had the implementation details of the LLVM relocation model flag names leaking through. 4) The actual results of passing these flags was incorrect on Darwin in many cases. For example, Darwin *always* uses PIC level 2 if it uses in PIC level, and Darwin *always* uses PIC on 64-bit regardless of the flags specified, including -fPIE. Darwin never compiles in PIE mode, but it can *link* in PIE mode. 5) Also, PIC was not always being enabled even when PIE was. This isn't a supported mode at all and may have caused some fallout in builds with complex PIC and PIE interactions. The result is (I hope) cleaner and clearer for readers. I've also left comments and tests about some of the truly strage behavior that is observed on Darwin platforms. We have no real testing of Windows platforms and PIC, but I don't have the tools handy to figure that out. Hopefully others can beef up our testing here. Unfortunately, I can't test this for every platform. =/ If folks have dependencies on these flags that aren't covered by tests, they may break. I've audited and ensured that all the changes in behavior of the existing tests are intentional and good. In particular I've tried to make sure the Darwin behavior (which is more suprising than the Linux behavior) also matches that of 'gcc' on my mac. llvm-svn: 168297
2012-11-19 11:52:03 +08:00
// RUN: %clang -c %s -target i386-unknown-unknown -fpic -fPIE -fpie -### 2>&1 \
This patch makes the behavior of clang consistent with the behavior of gcc 4.6 in cases where both -fPIC and -fPIE is used. - Separately check if -fPIE was specified in the command line and define both __PIC__ and __PIE__ when -fPIE is used. We need to check this separately because -fPIE will infer -fPIC even if its not explicitly used. - Fixed existing tests. - Added new tests for cases where both -fPIC and -fPIE is used. Author: Tareq A. Siraj <tareq.a.siraj@intel.com> Fixes: PR13221 Review: http://llvm-reviews.chandlerc.com/D94 llvm-svn: 167846
2012-11-13 23:32:35 +08:00
// RUN: | FileCheck %s --check-prefix=CHECK-PIE1
Completely re-work how the Clang driver interprets PIC and PIE options. There were numerous issues here that were all entangled, and so I've tried to do a general simplification of the logic. 1) The logic was mimicing actual GCC bugs, rather than "features". These have been fixed in trunk GCC, and this fixes Clang as well. Notably, the logic was always intended to be last-match-wins like any other flag. 2) The logic for handling '-mdynamic-no-pic' was preposterously unclear. It also allowed the use of this flag on non-Darwin platforms where it has no actual meaning. Now this option is handled directly based on tests of how llvm-gcc behaves, and it is only supported on Darwin. 3) The APIs for the Driver's ToolChains had the implementation ugliness of dynamic-no-pic leaking through them. They also had the implementation details of the LLVM relocation model flag names leaking through. 4) The actual results of passing these flags was incorrect on Darwin in many cases. For example, Darwin *always* uses PIC level 2 if it uses in PIC level, and Darwin *always* uses PIC on 64-bit regardless of the flags specified, including -fPIE. Darwin never compiles in PIE mode, but it can *link* in PIE mode. 5) Also, PIC was not always being enabled even when PIE was. This isn't a supported mode at all and may have caused some fallout in builds with complex PIC and PIE interactions. The result is (I hope) cleaner and clearer for readers. I've also left comments and tests about some of the truly strage behavior that is observed on Darwin platforms. We have no real testing of Windows platforms and PIC, but I don't have the tools handy to figure that out. Hopefully others can beef up our testing here. Unfortunately, I can't test this for every platform. =/ If folks have dependencies on these flags that aren't covered by tests, they may break. I've audited and ensured that all the changes in behavior of the existing tests are intentional and good. In particular I've tried to make sure the Darwin behavior (which is more suprising than the Linux behavior) also matches that of 'gcc' on my mac. llvm-svn: 168297
2012-11-19 11:52:03 +08:00
// RUN: %clang -c %s -target i386-unknown-unknown -fpie -fPIC -fPIE -### 2>&1 \
This patch makes the behavior of clang consistent with the behavior of gcc 4.6 in cases where both -fPIC and -fPIE is used. - Separately check if -fPIE was specified in the command line and define both __PIC__ and __PIE__ when -fPIE is used. We need to check this separately because -fPIE will infer -fPIC even if its not explicitly used. - Fixed existing tests. - Added new tests for cases where both -fPIC and -fPIE is used. Author: Tareq A. Siraj <tareq.a.siraj@intel.com> Fixes: PR13221 Review: http://llvm-reviews.chandlerc.com/D94 llvm-svn: 167846
2012-11-13 23:32:35 +08:00
// RUN: | FileCheck %s --check-prefix=CHECK-PIE2
//
Link to crtend.S when PIE in the FreeBSD driver. Patch by Brooks Davis! llvm-svn: 163388
2012-09-07 21:36:21 +08:00
// Make sure -pie is passed to along to ld and that the right *crt* files
// are linked in.
// RUN: %clang %s -target i386-unknown-freebsd -fPIE -pie -### \
Fix clash of gcc toolchains in driver regression tests. For some regression tests the path to the right toolchain is specified using the -sysroot switch. However, if clang was configured with a custom gcc toolchain (either by using GCC_INSTALL_PREFIX in cmake or the equivalent configure command), the path to the custom gcc toolchain path takes precedence to the one specified by sysroot. This causes several regression tests to fail as they will be using an unexpected path. This patch fixes this issue by adding --gcc-toolchain='' to all tests that rely on that. The empty string causes the driver to pick the path from sysroot instead. llvm-svn: 225182
2015-01-06 01:07:42 +08:00
// RUN: --gcc-toolchain="" \
Link to crtend.S when PIE in the FreeBSD driver. Patch by Brooks Davis! llvm-svn: 163388
2012-09-07 21:36:21 +08:00
// RUN: --sysroot=%S/Inputs/basic_freebsd_tree 2>&1 \
Completely re-work how the Clang driver interprets PIC and PIE options. There were numerous issues here that were all entangled, and so I've tried to do a general simplification of the logic. 1) The logic was mimicing actual GCC bugs, rather than "features". These have been fixed in trunk GCC, and this fixes Clang as well. Notably, the logic was always intended to be last-match-wins like any other flag. 2) The logic for handling '-mdynamic-no-pic' was preposterously unclear. It also allowed the use of this flag on non-Darwin platforms where it has no actual meaning. Now this option is handled directly based on tests of how llvm-gcc behaves, and it is only supported on Darwin. 3) The APIs for the Driver's ToolChains had the implementation ugliness of dynamic-no-pic leaking through them. They also had the implementation details of the LLVM relocation model flag names leaking through. 4) The actual results of passing these flags was incorrect on Darwin in many cases. For example, Darwin *always* uses PIC level 2 if it uses in PIC level, and Darwin *always* uses PIC on 64-bit regardless of the flags specified, including -fPIE. Darwin never compiles in PIE mode, but it can *link* in PIE mode. 5) Also, PIC was not always being enabled even when PIE was. This isn't a supported mode at all and may have caused some fallout in builds with complex PIC and PIE interactions. The result is (I hope) cleaner and clearer for readers. I've also left comments and tests about some of the truly strage behavior that is observed on Darwin platforms. We have no real testing of Windows platforms and PIC, but I don't have the tools handy to figure that out. Hopefully others can beef up our testing here. Unfortunately, I can't test this for every platform. =/ If folks have dependencies on these flags that aren't covered by tests, they may break. I've audited and ensured that all the changes in behavior of the existing tests are intentional and good. In particular I've tried to make sure the Darwin behavior (which is more suprising than the Linux behavior) also matches that of 'gcc' on my mac. llvm-svn: 168297
2012-11-19 11:52:03 +08:00
// RUN: | FileCheck %s --check-prefix=CHECK-PIE-LD
Link to crtend.S when PIE in the FreeBSD driver. Patch by Brooks Davis! llvm-svn: 163388
2012-09-07 21:36:21 +08:00
// RUN: %clang %s -target i386-linux-gnu -fPIE -pie -### \
Fix clash of gcc toolchains in driver regression tests. For some regression tests the path to the right toolchain is specified using the -sysroot switch. However, if clang was configured with a custom gcc toolchain (either by using GCC_INSTALL_PREFIX in cmake or the equivalent configure command), the path to the custom gcc toolchain path takes precedence to the one specified by sysroot. This causes several regression tests to fail as they will be using an unexpected path. This patch fixes this issue by adding --gcc-toolchain='' to all tests that rely on that. The empty string causes the driver to pick the path from sysroot instead. llvm-svn: 225182
2015-01-06 01:07:42 +08:00
// RUN: --gcc-toolchain="" \
Link to crtend.S when PIE in the FreeBSD driver. Patch by Brooks Davis! llvm-svn: 163388
2012-09-07 21:36:21 +08:00
// RUN: --sysroot=%S/Inputs/basic_linux_tree 2>&1 \
Completely re-work how the Clang driver interprets PIC and PIE options. There were numerous issues here that were all entangled, and so I've tried to do a general simplification of the logic. 1) The logic was mimicing actual GCC bugs, rather than "features". These have been fixed in trunk GCC, and this fixes Clang as well. Notably, the logic was always intended to be last-match-wins like any other flag. 2) The logic for handling '-mdynamic-no-pic' was preposterously unclear. It also allowed the use of this flag on non-Darwin platforms where it has no actual meaning. Now this option is handled directly based on tests of how llvm-gcc behaves, and it is only supported on Darwin. 3) The APIs for the Driver's ToolChains had the implementation ugliness of dynamic-no-pic leaking through them. They also had the implementation details of the LLVM relocation model flag names leaking through. 4) The actual results of passing these flags was incorrect on Darwin in many cases. For example, Darwin *always* uses PIC level 2 if it uses in PIC level, and Darwin *always* uses PIC on 64-bit regardless of the flags specified, including -fPIE. Darwin never compiles in PIE mode, but it can *link* in PIE mode. 5) Also, PIC was not always being enabled even when PIE was. This isn't a supported mode at all and may have caused some fallout in builds with complex PIC and PIE interactions. The result is (I hope) cleaner and clearer for readers. I've also left comments and tests about some of the truly strage behavior that is observed on Darwin platforms. We have no real testing of Windows platforms and PIC, but I don't have the tools handy to figure that out. Hopefully others can beef up our testing here. Unfortunately, I can't test this for every platform. =/ If folks have dependencies on these flags that aren't covered by tests, they may break. I've audited and ensured that all the changes in behavior of the existing tests are intentional and good. In particular I've tried to make sure the Darwin behavior (which is more suprising than the Linux behavior) also matches that of 'gcc' on my mac. llvm-svn: 168297
2012-11-19 11:52:03 +08:00
// RUN: | FileCheck %s --check-prefix=CHECK-PIE-LD
// RUN: %clang %s -target i386-linux-gnu -fPIC -pie -### \
Fix clash of gcc toolchains in driver regression tests. For some regression tests the path to the right toolchain is specified using the -sysroot switch. However, if clang was configured with a custom gcc toolchain (either by using GCC_INSTALL_PREFIX in cmake or the equivalent configure command), the path to the custom gcc toolchain path takes precedence to the one specified by sysroot. This causes several regression tests to fail as they will be using an unexpected path. This patch fixes this issue by adding --gcc-toolchain='' to all tests that rely on that. The empty string causes the driver to pick the path from sysroot instead. llvm-svn: 225182
2015-01-06 01:07:42 +08:00
// RUN: --gcc-toolchain="" \
Completely re-work how the Clang driver interprets PIC and PIE options. There were numerous issues here that were all entangled, and so I've tried to do a general simplification of the logic. 1) The logic was mimicing actual GCC bugs, rather than "features". These have been fixed in trunk GCC, and this fixes Clang as well. Notably, the logic was always intended to be last-match-wins like any other flag. 2) The logic for handling '-mdynamic-no-pic' was preposterously unclear. It also allowed the use of this flag on non-Darwin platforms where it has no actual meaning. Now this option is handled directly based on tests of how llvm-gcc behaves, and it is only supported on Darwin. 3) The APIs for the Driver's ToolChains had the implementation ugliness of dynamic-no-pic leaking through them. They also had the implementation details of the LLVM relocation model flag names leaking through. 4) The actual results of passing these flags was incorrect on Darwin in many cases. For example, Darwin *always* uses PIC level 2 if it uses in PIC level, and Darwin *always* uses PIC on 64-bit regardless of the flags specified, including -fPIE. Darwin never compiles in PIE mode, but it can *link* in PIE mode. 5) Also, PIC was not always being enabled even when PIE was. This isn't a supported mode at all and may have caused some fallout in builds with complex PIC and PIE interactions. The result is (I hope) cleaner and clearer for readers. I've also left comments and tests about some of the truly strage behavior that is observed on Darwin platforms. We have no real testing of Windows platforms and PIC, but I don't have the tools handy to figure that out. Hopefully others can beef up our testing here. Unfortunately, I can't test this for every platform. =/ If folks have dependencies on these flags that aren't covered by tests, they may break. I've audited and ensured that all the changes in behavior of the existing tests are intentional and good. In particular I've tried to make sure the Darwin behavior (which is more suprising than the Linux behavior) also matches that of 'gcc' on my mac. llvm-svn: 168297
2012-11-19 11:52:03 +08:00
// RUN: --sysroot=%S/Inputs/basic_linux_tree 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-PIE-LD
Link to crtend.S when PIE in the FreeBSD driver. Patch by Brooks Davis! llvm-svn: 163388
2012-09-07 21:36:21 +08:00
//
Teach Clang about PIE compilations. This is the first step of PR12380. First, this patch cleans up the parsing of the PIC and PIE family of options in the driver. The existing logic failed to claim arguments all over the place resulting in kludges that marked the options as unused. Instead actually walk all of the arguments and claim them properly. We now treat -f{,no-}{pic,PIC,pie,PIE} as a single set, accepting the last one on the commandline. Previously there were lots of ordering bugs that could creep in due to the nature of the parsing. Let me know if folks would like weird things such as "-fPIE -fno-pic" to turn on PIE, but disable full PIC. This doesn't make any sense to me, but we could in theory support it. Options that seem to have intentional "trump" status (-static, -mkernel, etc) continue to do so and are commented as such. Next, a -pie-level flag is threaded into the frontend, rigged to a language option, and handled preprocessor, setting up the appropriate defines. We'll now have the correct defines when compiling with -fpie. The one place outside of the preprocessor that was inspecting the PIC level (as opposed to the relocation model, which is set and handled separately, yay!) is in the GNU ObjC runtime. I changed it to exactly preserve existing behavior. If folks want to change its behavior in the face of PIE, they can do that in a separate patch. Essentially the only functionality changed here is the preprocessor defines and bug-fixes to the argument management. Tests have been updated and extended to test all of this a bit more thoroughly. llvm-svn: 154291
2012-04-09 00:40:35 +08:00
// Disregard any of the PIC-specific flags if we have a trump-card flag.
// RUN: %clang -c %s -target i386-unknown-unknown -mkernel -fPIC -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-NO-PIC
Stop treating -static as overriding -fPIC: they are distinct. For some reason, clang had been treating a command like: clang -static -fPIC foo.c as if it should be compiled without the PIC relocation model. This was incorrect: -static should be affecting only the linking model, and -fPIC only the compilation. This new behavior also matches GCC. This is a follow-up from a review comment on r245447. Differential Revision: http://reviews.llvm.org/D12208 llvm-svn: 245667
2015-08-21 12:14:55 +08:00
// The -static argument *doesn't* override PIC: -static only affects
// linking, and -fPIC only affects code generation.
Teach Clang about PIE compilations. This is the first step of PR12380. First, this patch cleans up the parsing of the PIC and PIE family of options in the driver. The existing logic failed to claim arguments all over the place resulting in kludges that marked the options as unused. Instead actually walk all of the arguments and claim them properly. We now treat -f{,no-}{pic,PIC,pie,PIE} as a single set, accepting the last one on the commandline. Previously there were lots of ordering bugs that could creep in due to the nature of the parsing. Let me know if folks would like weird things such as "-fPIE -fno-pic" to turn on PIE, but disable full PIC. This doesn't make any sense to me, but we could in theory support it. Options that seem to have intentional "trump" status (-static, -mkernel, etc) continue to do so and are commented as such. Next, a -pie-level flag is threaded into the frontend, rigged to a language option, and handled preprocessor, setting up the appropriate defines. We'll now have the correct defines when compiling with -fpie. The one place outside of the preprocessor that was inspecting the PIC level (as opposed to the relocation model, which is set and handled separately, yay!) is in the GNU ObjC runtime. I changed it to exactly preserve existing behavior. If folks want to change its behavior in the face of PIE, they can do that in a separate patch. Essentially the only functionality changed here is the preprocessor defines and bug-fixes to the argument management. Tests have been updated and extended to test all of this a bit more thoroughly. llvm-svn: 154291
2012-04-09 00:40:35 +08:00
// RUN: %clang -c %s -target i386-unknown-unknown -static -fPIC -### 2>&1 \
Stop treating -static as overriding -fPIC: they are distinct. For some reason, clang had been treating a command like: clang -static -fPIC foo.c as if it should be compiled without the PIC relocation model. This was incorrect: -static should be affecting only the linking model, and -fPIC only the compilation. This new behavior also matches GCC. This is a follow-up from a review comment on r245447. Differential Revision: http://reviews.llvm.org/D12208 llvm-svn: 245667
2015-08-21 12:14:55 +08:00
// RUN: | FileCheck %s --check-prefix=CHECK-PIC2
// RUN: %clang %s -target i386-linux-gnu -static -fPIC -### \
// RUN: --gcc-toolchain="" \
// RUN: --sysroot=%S/Inputs/basic_linux_tree 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-STATIC
Completely re-work how the Clang driver interprets PIC and PIE options. There were numerous issues here that were all entangled, and so I've tried to do a general simplification of the logic. 1) The logic was mimicing actual GCC bugs, rather than "features". These have been fixed in trunk GCC, and this fixes Clang as well. Notably, the logic was always intended to be last-match-wins like any other flag. 2) The logic for handling '-mdynamic-no-pic' was preposterously unclear. It also allowed the use of this flag on non-Darwin platforms where it has no actual meaning. Now this option is handled directly based on tests of how llvm-gcc behaves, and it is only supported on Darwin. 3) The APIs for the Driver's ToolChains had the implementation ugliness of dynamic-no-pic leaking through them. They also had the implementation details of the LLVM relocation model flag names leaking through. 4) The actual results of passing these flags was incorrect on Darwin in many cases. For example, Darwin *always* uses PIC level 2 if it uses in PIC level, and Darwin *always* uses PIC on 64-bit regardless of the flags specified, including -fPIE. Darwin never compiles in PIE mode, but it can *link* in PIE mode. 5) Also, PIC was not always being enabled even when PIE was. This isn't a supported mode at all and may have caused some fallout in builds with complex PIC and PIE interactions. The result is (I hope) cleaner and clearer for readers. I've also left comments and tests about some of the truly strage behavior that is observed on Darwin platforms. We have no real testing of Windows platforms and PIC, but I don't have the tools handy to figure that out. Hopefully others can beef up our testing here. Unfortunately, I can't test this for every platform. =/ If folks have dependencies on these flags that aren't covered by tests, they may break. I've audited and ensured that all the changes in behavior of the existing tests are intentional and good. In particular I've tried to make sure the Darwin behavior (which is more suprising than the Linux behavior) also matches that of 'gcc' on my mac. llvm-svn: 168297
2012-11-19 11:52:03 +08:00
//
Do not pass -pie flag to linker if -shared specified. This matches the gcc driver and makes it possible to add -pie to $CC or similar and have it apply in the right places. llvm-svn: 172753
2013-01-18 04:17:16 +08:00
// On Linux, disregard -pie if we have -shared.
// RUN: %clang %s -target i386-unknown-linux -shared -pie -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-NO-PIE
//
Completely re-work how the Clang driver interprets PIC and PIE options. There were numerous issues here that were all entangled, and so I've tried to do a general simplification of the logic. 1) The logic was mimicing actual GCC bugs, rather than "features". These have been fixed in trunk GCC, and this fixes Clang as well. Notably, the logic was always intended to be last-match-wins like any other flag. 2) The logic for handling '-mdynamic-no-pic' was preposterously unclear. It also allowed the use of this flag on non-Darwin platforms where it has no actual meaning. Now this option is handled directly based on tests of how llvm-gcc behaves, and it is only supported on Darwin. 3) The APIs for the Driver's ToolChains had the implementation ugliness of dynamic-no-pic leaking through them. They also had the implementation details of the LLVM relocation model flag names leaking through. 4) The actual results of passing these flags was incorrect on Darwin in many cases. For example, Darwin *always* uses PIC level 2 if it uses in PIC level, and Darwin *always* uses PIC on 64-bit regardless of the flags specified, including -fPIE. Darwin never compiles in PIE mode, but it can *link* in PIE mode. 5) Also, PIC was not always being enabled even when PIE was. This isn't a supported mode at all and may have caused some fallout in builds with complex PIC and PIE interactions. The result is (I hope) cleaner and clearer for readers. I've also left comments and tests about some of the truly strage behavior that is observed on Darwin platforms. We have no real testing of Windows platforms and PIC, but I don't have the tools handy to figure that out. Hopefully others can beef up our testing here. Unfortunately, I can't test this for every platform. =/ If folks have dependencies on these flags that aren't covered by tests, they may break. I've audited and ensured that all the changes in behavior of the existing tests are intentional and good. In particular I've tried to make sure the Darwin behavior (which is more suprising than the Linux behavior) also matches that of 'gcc' on my mac. llvm-svn: 168297
2012-11-19 11:52:03 +08:00
// Darwin is a beautiful and unique snowflake when it comes to these flags.
Fix a tranche of comment, test and doc typos llvm-svn: 196510
2013-12-06 00:25:25 +08:00
// When targeting a 32-bit darwin system, the -fno-* flag variants work and
Completely re-work how the Clang driver interprets PIC and PIE options. There were numerous issues here that were all entangled, and so I've tried to do a general simplification of the logic. 1) The logic was mimicing actual GCC bugs, rather than "features". These have been fixed in trunk GCC, and this fixes Clang as well. Notably, the logic was always intended to be last-match-wins like any other flag. 2) The logic for handling '-mdynamic-no-pic' was preposterously unclear. It also allowed the use of this flag on non-Darwin platforms where it has no actual meaning. Now this option is handled directly based on tests of how llvm-gcc behaves, and it is only supported on Darwin. 3) The APIs for the Driver's ToolChains had the implementation ugliness of dynamic-no-pic leaking through them. They also had the implementation details of the LLVM relocation model flag names leaking through. 4) The actual results of passing these flags was incorrect on Darwin in many cases. For example, Darwin *always* uses PIC level 2 if it uses in PIC level, and Darwin *always* uses PIC on 64-bit regardless of the flags specified, including -fPIE. Darwin never compiles in PIE mode, but it can *link* in PIE mode. 5) Also, PIC was not always being enabled even when PIE was. This isn't a supported mode at all and may have caused some fallout in builds with complex PIC and PIE interactions. The result is (I hope) cleaner and clearer for readers. I've also left comments and tests about some of the truly strage behavior that is observed on Darwin platforms. We have no real testing of Windows platforms and PIC, but I don't have the tools handy to figure that out. Hopefully others can beef up our testing here. Unfortunately, I can't test this for every platform. =/ If folks have dependencies on these flags that aren't covered by tests, they may break. I've audited and ensured that all the changes in behavior of the existing tests are intentional and good. In particular I've tried to make sure the Darwin behavior (which is more suprising than the Linux behavior) also matches that of 'gcc' on my mac. llvm-svn: 168297
2012-11-19 11:52:03 +08:00
// disable PIC, but any other flag enables PIC (*not* PIE) even if the flag
// specifies PIE. On 64-bit targets, there is simply nothing you can do, there
// is no PIE, there is only PIC when it comes to compilation.
// RUN: %clang -c %s -target i386-apple-darwin -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-PIC2
// RUN: %clang -c %s -target i386-apple-darwin -fpic -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-PIC2
// RUN: %clang -c %s -target i386-apple-darwin -fPIC -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-PIC2
// RUN: %clang -c %s -target i386-apple-darwin -fpie -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-PIC2
// RUN: %clang -c %s -target i386-apple-darwin -fPIE -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-PIC2
// RUN: %clang -c %s -target i386-apple-darwin -fno-PIC -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-NO-PIC
// RUN: %clang -c %s -target i386-apple-darwin -fno-PIE -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-NO-PIC
// RUN: %clang -c %s -target i386-apple-darwin -fno-PIC -fpic -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-PIC2
// RUN: %clang -c %s -target i386-apple-darwin -fno-PIC -fPIE -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-PIC2
// RUN: %clang -c %s -target x86_64-apple-darwin -fno-PIC -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-PIC2
// RUN: %clang -c %s -target x86_64-apple-darwin -fno-PIE -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-PIC2
// RUN: %clang -c %s -target x86_64-apple-darwin -fpic -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-PIC2
// RUN: %clang -c %s -target x86_64-apple-darwin -fPIE -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-PIC2
One more follow-up to r179082 - parse PIC/PIE arguments even on platfroms that force default PIC (like Darwin x86-64), otherwise specifying -fPIC will produce bogus unused argument warning llvm-svn: 179092
2013-04-09 20:28:19 +08:00
// RUN: %clang -c %s -target x86_64-apple-darwin -fPIC -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-NO-UNUSED-ARG
Completely re-work how the Clang driver interprets PIC and PIE options. There were numerous issues here that were all entangled, and so I've tried to do a general simplification of the logic. 1) The logic was mimicing actual GCC bugs, rather than "features". These have been fixed in trunk GCC, and this fixes Clang as well. Notably, the logic was always intended to be last-match-wins like any other flag. 2) The logic for handling '-mdynamic-no-pic' was preposterously unclear. It also allowed the use of this flag on non-Darwin platforms where it has no actual meaning. Now this option is handled directly based on tests of how llvm-gcc behaves, and it is only supported on Darwin. 3) The APIs for the Driver's ToolChains had the implementation ugliness of dynamic-no-pic leaking through them. They also had the implementation details of the LLVM relocation model flag names leaking through. 4) The actual results of passing these flags was incorrect on Darwin in many cases. For example, Darwin *always* uses PIC level 2 if it uses in PIC level, and Darwin *always* uses PIC on 64-bit regardless of the flags specified, including -fPIE. Darwin never compiles in PIE mode, but it can *link* in PIE mode. 5) Also, PIC was not always being enabled even when PIE was. This isn't a supported mode at all and may have caused some fallout in builds with complex PIC and PIE interactions. The result is (I hope) cleaner and clearer for readers. I've also left comments and tests about some of the truly strage behavior that is observed on Darwin platforms. We have no real testing of Windows platforms and PIC, but I don't have the tools handy to figure that out. Hopefully others can beef up our testing here. Unfortunately, I can't test this for every platform. =/ If folks have dependencies on these flags that aren't covered by tests, they may break. I've audited and ensured that all the changes in behavior of the existing tests are intentional and good. In particular I've tried to make sure the Darwin behavior (which is more suprising than the Linux behavior) also matches that of 'gcc' on my mac. llvm-svn: 168297
2012-11-19 11:52:03 +08:00
//
// Darwin gets even more special with '-mdynamic-no-pic'. This flag is only
// valid on Darwin, and it's behavior is very strange but needs to remain
// consistent for compatibility.
// RUN: %clang -c %s -target i386-unknown-unknown -mdynamic-no-pic -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-NON-DARWIN-DYNAMIC-NO-PIC
// RUN: %clang -c %s -target i386-apple-darwin -mdynamic-no-pic -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-DYNAMIC-NO-PIC-32
// RUN: %clang -c %s -target i386-apple-darwin -mdynamic-no-pic -fno-pic -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-DYNAMIC-NO-PIC-32
// RUN: %clang -c %s -target i386-apple-darwin -mdynamic-no-pic -fpie -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-DYNAMIC-NO-PIC-32
// RUN: %clang -c %s -target x86_64-apple-darwin -mdynamic-no-pic -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-DYNAMIC-NO-PIC-64
// RUN: %clang -c %s -target x86_64-apple-darwin -mdynamic-no-pic -fno-pic -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-DYNAMIC-NO-PIC-64
// RUN: %clang -c %s -target x86_64-apple-darwin -mdynamic-no-pic -fpie -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-DYNAMIC-NO-PIC-64
//
// Checks for ARM+Apple+IOS including -fapple-kext, -mkernel, and iphoneos
// version boundaries.
Un-revert r164907 and r164902 (+ follow-ups), 10.6 build fix to follow. llvm-svn: 165988
2012-10-16 06:23:53 +08:00
// RUN: %clang -c %s -target armv7-apple-ios -fapple-kext -miphoneos-version-min=6.0.0 -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-PIC2
// RUN: %clang -c %s -target armv7-apple-ios -mkernel -miphoneos-version-min=6.0.0 -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-PIC2
Remove special case for aarch64 static vs. PIC code in iOS kernel code. I added this check a while back but then made a note to myself that it should be completely unnecessary since iOS always uses PIC code-gen for aarch64. Since I could never come up with any reason why it would be necessary, I'm just going to remove it and we'll see if anything breaks. rdar://problem/13627985 llvm-svn: 223097
2014-12-02 08:27:35 +08:00
// RUN: %clang -c %s -target arm64-apple-ios -mkernel -miphoneos-version-min=7.0.0 -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-PIC2
Do not force "-static" for aarch64 iOS kernel/kext assembly code. Somehow a check for aarch64 was added to the Darwin toolchain's isKernelStatic function as part of the initial commit for Apple's arm64 target (r205100). That check was not in any of Apple's internal code and no one here knows where it came from. It has been harmless because "-static" does not change much, if anything, for arm64 iOS code, but it makes no sense to keep this check. llvm-svn: 228673
2015-02-10 14:33:04 +08:00
// RUN: %clang -x assembler -c %s -target arm64-apple-ios -mkernel -miphoneos-version-min=7.0.0 -no-integrated-as -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-NO-STATIC
Support watchOS and tvOS driver options This patch should add support for almost all command-line options and driver tinkering necessary to produce a correct "clang -cc1" invocation for watchOS and tvOS. llvm-svn: 251706
2015-10-31 00:30:27 +08:00
// RUN: %clang -c %s -target armv7k-apple-watchos -fapple-kext -mwatchos-version-min=1.0.0 -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-PIC2
Un-revert r164907 and r164902 (+ follow-ups), 10.6 build fix to follow. llvm-svn: 165988
2012-10-16 06:23:53 +08:00
// RUN: %clang -c %s -target armv7-apple-ios -fapple-kext -miphoneos-version-min=5.0.0 -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-NO-PIC
// RUN: %clang -c %s -target armv7-apple-ios -fapple-kext -miphoneos-version-min=6.0.0 -static -### 2>&1 \
PIC should not be enabled by default on Darwin with -static. r245667 changed -static so that it doesn't override an explicit -fPIC option, but -static should still change the default for Darwin for -fno-PIC. This matches longstanding GCC and Clang behavior on Darwin and changing it would be disruptive, with no significant benefit. http://reviews.llvm.org/D15455 rdar://problem/23811045 llvm-svn: 256026
2015-12-19 04:37:54 +08:00
// RUN: | FileCheck %s --check-prefix=CHECK-NO-PIC
Generalize r256026 to apply to all MachO targets, not just Darwin targets. The PIC default is set for the MachO toolchain, not just the Darwin toolchain, so this treats those the same. The behavior with -static should be the same for all MachO targets. rdar://24152327 llvm-svn: 257556
2016-01-13 09:19:02 +08:00
// RUN: %clang -c %s -target armv7-apple-unknown-macho -static -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-NO-PIC
Default to PIE and using stack protection on OpenBSD. Patch by Matthew Dempsky. llvm-svn: 183295
2013-06-05 12:28:55 +08:00
//
// On OpenBSD, PIE is enabled by default, but can be disabled.
Use appropriate default PIE settings for OpenBSD. llvm-svn: 211624
2014-06-25 03:51:29 +08:00
// RUN: %clang -c %s -target amd64-pc-openbsd -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-PIE1
Default to PIE and using stack protection on OpenBSD. Patch by Matthew Dempsky. llvm-svn: 183295
2013-06-05 12:28:55 +08:00
// RUN: %clang -c %s -target i386-pc-openbsd -### 2>&1 \
Use appropriate default PIE settings for OpenBSD. llvm-svn: 211624
2014-06-25 03:51:29 +08:00
// RUN: | FileCheck %s --check-prefix=CHECK-PIE1
// RUN: %clang -c %s -target mips64-unknown-openbsd -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-PIE1
// RUN: %clang -c %s -target mips64el-unknown-openbsd -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-PIE1
// RUN: %clang -c %s -target powerpc-unknown-openbsd -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-PIE2
Use the appropriate PIE level for OpenBSD/sparc. llvm-svn: 239028
2015-06-04 16:45:23 +08:00
// RUN: %clang -c %s -target sparc-unknown-openbsd -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-PIE2
Use appropriate default PIE settings for OpenBSD. llvm-svn: 211624
2014-06-25 03:51:29 +08:00
// RUN: %clang -c %s -target sparc64-unknown-openbsd -### 2>&1 \
Default to PIE and using stack protection on OpenBSD. Patch by Matthew Dempsky. llvm-svn: 183295
2013-06-05 12:28:55 +08:00
// RUN: | FileCheck %s --check-prefix=CHECK-PIE2
// RUN: %clang -c %s -target i386-pc-openbsd -fno-pie -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-NO-PIC
//
// On OpenBSD, -nopie needs to be passed through to the linker.
// RUN: %clang %s -target i386-pc-openbsd -nopie -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-NOPIE-LD
Turn on PIC by default on Android (http://llvm-reviews.chandlerc.com/D2603) llvm-svn: 200290
2014-01-28 14:30:35 +08:00
//
// On Android PIC is enabled by default
// RUN: %clang -c %s -target i686-linux-android -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-PIC2
// RUN: %clang -c %s -target arm-linux-androideabi -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-PIC1
// RUN: %clang -c %s -target mipsel-linux-android -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-PIC1
Android uses -fpic for aarch64. llvm-svn: 207567
2014-04-30 02:18:58 +08:00
// RUN: %clang -c %s -target aarch64-linux-android -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-PIC1
Android uses -fpic for arm64 as well. llvm-svn: 207642
2014-04-30 20:18:12 +08:00
// RUN: %clang -c %s -target arm64-linux-android -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-PIC1
[Driver] x86-64 Windows is always PIC This fixes PR23963. llvm-svn: 240902
2015-06-28 12:23:33 +08:00
//
// On Windows-X64 PIC is enabled by default
// RUN: %clang -c %s -target x86_64-pc-windows-msvc18.0.0 -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-PIC2
// RUN: %clang -c %s -target x86_64-pc-windows-gnu -### 2>&1 \
// RUN: | FileCheck %s --check-prefix=CHECK-PIC2