Yuan, do you have any progress here? Ivan
On Thursday, May 28, 2020 at 6:47:41 AM UTC+3 Yuan Ting wrote: > Thanks for your correction, overriding CMAKE_C_FLAGS and CMAKE_ASM_FLAGS > with "-fcf-protection=none" works for me. > I also try to build gollvm from a clean slate (by recreating a build area > and rerun cmake without overriding CMAKE_C_FLAGS and CMAKE_ASM_FLAGS). > However, I still failed to build gollvm. I can even see the notice during > the configuration: > > -- trying -fcf-protection=none workaround > > -- Performing Test SPLIT_STACK_WORKAROUND > > -- Performing Test SPLIT_STACK_WORKAROUND - Success > > -- applying -fcf-protection=none workaround > > -- starting libgo configuration. > > > but during building, the same error happened: > > > [2427/2600] Linking C shared library tools/gollvm/libgo/libgo.so.11git > > FAILED: tools/gollvm/libgo/libgo.so.11git > > : && /usr/bin/cc -fPIC -fPIC -Werror=date-time -Wall -Wextra > -Wno-unused-parameter -Wwrite-strings -Wno-missing-field-initializers > -pedantic -Wno-long-long -Wimplicit-fallthrough -Wno-comment > -fdiagnostics-color -ffunction-sections -fdata-sections -O3 > -Wl,-z,nodelete -fuse-ld=gold -Wl,-O3 -Wl,--gc-sections -shared > -Wl,-soname,libgo.so.11git -o tools/gollvm/libgo/libgo.so.11git ...*(omits > all object files)...* -Wl,-rpath,"\$ORIGIN/../lib" -lpthread -lm > -fsplit-stack && : > > ... ... > > /usr/bin/ld.gold: error: > tools/gollvm/libgo/CMakeFiles/libffi_piclib.dir/libffi/src/x86/ffi64.c.o: > failed to match split-stack sequence at section 22 offset 0 > > /usr/bin/ld.gold: error: > tools/gollvm/libgo/CMakeFiles/libffi_piclib.dir/libffi/src/x86/ffi64.c.o: > failed to match split-stack sequence at section 26 offset 0 > > /usr/bin/ld.gold: error: > tools/gollvm/libgo/CMakeFiles/libffi_piclib.dir/libffi/src/x86/ffiw64.c.o: > failed to match split-stack sequence at section 16 offset 0 > > collect2: error: ld returned 1 exit status > > [2429/2600] Linking C static library tools/gollvm/libgo/libgo.a > > ninja: build stopped: subcommand failed. > > On Wednesday, May 27, 2020 at 10:36:23 PM UTC+8, Than McIntosh wrote: > >> >> I can successfully compile and link gollvm by setting >> CMAKE_C_FLAGS="-fcf-protection=none". >> >> The manual override of CMAKE_C_FLAGS shouldn't be needed, if you rerun >> cmake from a clean slate (removing the CMakeCache.txt, etc) it should >> detect that "-fcf-protection=branch" is hard-wired on, and will add the >> right flags under the hood to work around it. >> >> >>root@a78010e8da5d:~# go version >> >>go: symbol lookup error: >> /root/llvm-project/install-release/lib64/libgo.so.11git: undefined symbol: >> __go_getcontext >> >> This symbol is defined in an assembly source file in libgo. What's >> happening is that by setting CMAKE_C_FLAGS="-fcf-protection=none" but not >> also setting CMAKE_ASM_FLAGS="-fcf-protection=none" you've introduced a >> discrepancy between the C and Assembly compiles in libgo. If you >> set CMAKE_C_FLAGS, you should also set CMAKE_ASM_FLAGS. >> >> Thanks, Than >> >> >> On Wed, May 27, 2020 at 10:02 AM Yuan Ting <yuan...@ict.ac.cn> wrote: >> > Thank you for your guidance, now I can successfully compile and link >>> gollvm by setting CMAKE_C_FLAGS="-fcf-protection=none". But after >>> installing gollvm, there is still a weird problem: >>> >>> root@a78010e8da5d:~# go version >>> go: symbol lookup error: >>> /root/llvm-project/install-release/lib64/libgo.so.11git: undefined symbol: >>> __go_getcontext >>> >>> On Wednesday, May 27, 2020 at 12:01:06 AM UTC+8, Than McIntosh wrote: >>>> >>>> >>>> This looks exactly like the bug in >>>> https://github.com/golang/go/issues/38728. The prolog of the function >>>> contains >>>> >>>> 0: f3 0f 1e fa endbr64 >>>> >>>> which is confusing the gold linker. This instruction doesn't get added >>>> unless the host compiler has "-fcf-protection=branch" turned on. I would >>>> suggest deleting and recreating your build area or removing your >>>> CMakeCache.txt file to insure that you have a rebuild from a clean start. >>>> >>>> Thanks, Than >>>> >>>> >>>> On Tue, May 26, 2020 at 11:00 AM Yuan Ting <yuan...@ict.ac.cn> wrote: >>>> >>>>> OK, as follows >>>>> >>>>> ./tools/gollvm/libgo/CMakeFiles/libgo_c_piclib.dir/__/gofrontend/libgo/go/syscall/errno.c.o: >>>>> >>>>> file format elf64-x86-64 >>>>> >>>>> >>>>> Disassembly of section .text.syscall.GetErrno: >>>>> >>>>> 0000000000000000 <syscall.GetErrno>: >>>>> 0: f3 0f 1e fa endbr64 >>>>> 4: 64 48 3b 24 25 70 00 cmp %fs:0x70,%rsp >>>>> b: 00 00 >>>>> d: 72 11 jb 20 <syscall.GetErrno+0x20> >>>>> f: 48 83 ec 08 sub $0x8,%rsp >>>>> 13: e8 00 00 00 00 callq 18 <syscall.GetErrno+0x18> >>>>> 14: R_X86_64_PLT32 __errno_location-0x4 >>>>> 18: 48 63 00 movslq (%rax),%rax >>>>> 1b: 48 83 c4 08 add $0x8,%rsp >>>>> 1f: c3 retq >>>>> 20: 41 ba 08 00 00 00 mov $0x8,%r10d >>>>> 26: 45 31 db xor %r11d,%r11d >>>>> 29: e8 00 00 00 00 callq 2e <syscall.GetErrno+0x2e> >>>>> 2a: R_X86_64_PLT32 __morestack-0x4 >>>>> 2e: c3 retq >>>>> 2f: f3 0f 1e fa endbr64 >>>>> 33: eb da jmp f <syscall.GetErrno+0xf> >>>>> >>>>> On Tuesday, May 26, 2020 at 10:44:03 PM UTC+8, Than McIntosh wrote: >>>>>> >>>>>> OK. Just to confirm, what do you see from >>>>>> >>>>>> objdump -dr >>>>>> ./tools/gollvm/libgo/CMakeFiles/libgo_c_piclib.dir/__/gofrontend/libgo/go/syscall/errno.c.o >>>>>> >>>>>> for the prolog of syscall.Geterrno. Should be something like >>>>>> >>>>>> Disassembly of section .text.syscall.GetErrno: >>>>>> >>>>>> 0000000000000000 <syscall.GetErrno>: >>>>>> 0: 64 48 3b 24 25 70 00 cmp %fs:0x70,%rsp >>>>>> 7: 00 00 >>>>>> 9: 72 11 jb 1c <syscall.GetErrno+0x1c> >>>>>> b: 48 83 ec 08 sub $0x8,%rsp >>>>>> f: e8 00 00 00 00 callq 14 <syscall.GetErrno+0x14> >>>>>> 10: R_X86_64_PLT32 __errno_location-0x4 >>>>>> >>>>>> Thanks, Than >>>>>> >>>>>> ? >>>>>> >>>>>> On Tue, May 26, 2020 at 10:05 AM Yuan Ting <yuan...@ict.ac.cn> wrote: >>>>>> >>>>>>> Fine, that is >>>>>>> >>>>>>> [1/3] /usr/bin/cc -D_DEBUG -D_GNU_SOURCE -D__STDC_CONSTANT_MACROS >>>>>>> -D__STDC_FORMAT_MACROS -D__STDC_LIMIT_MACROS -Itools/gollvm/libgo >>>>>>> -I/root/llvm-project/llvm/tools/gollvm/libgo -Iinclude >>>>>>> -I/root/llvm-project/llvm/include >>>>>>> -I/root/llvm-project/llvm/tools/gollvm/gofrontend/libgo/runtime >>>>>>> -Itools/gollvm/libgo/runtime >>>>>>> -I/root/llvm-project/llvm/tools/gollvm/libgo/libbacktrace -fPIC >>>>>>> -Werror=date-time -Wall -Wextra -Wno-unused-parameter -Wwrite-strings >>>>>>> -Wno-missing-field-initializers -pedantic -Wno-long-long >>>>>>> -Wimplicit-fallthrough -Wno-comment -fdiagnostics-color >>>>>>> -ffunction-sections >>>>>>> -fdata-sections -O3 -fPIC -g -Wno-zero-length-array -fsplit-stack >>>>>>> -D_GNU_SOURCE -D_LARGEFILE_SOURCE -D_FILE_OFFSET_BITS=64 -UNDEBUG -MD >>>>>>> -MT >>>>>>> tools/gollvm/libgo/CMakeFiles/libgo_c_piclib.dir/__/gofrontend/libgo/go/syscall/errno.c.o >>>>>>> >>>>>>> -MF >>>>>>> tools/gollvm/libgo/CMakeFiles/libgo_c_piclib.dir/__/gofrontend/libgo/go/syscall/errno.c.o.d >>>>>>> >>>>>>> -o >>>>>>> tools/gollvm/libgo/CMakeFiles/libgo_c_piclib.dir/__/gofrontend/libgo/go/syscall/errno.c.o >>>>>>> >>>>>>> -c >>>>>>> /root/llvm-project/llvm/tools/gollvm/gofrontend/libgo/go/syscall/errno.c >>>>>>> >>>>>>> On Tuesday, May 26, 2020 at 8:41:18 PM UTC+8, Than McIntosh wrote: >>>>>>>> >>>>>>>> OK. >>>>>>>> >>>>>>>> I'm wondering if for some reason the fix I put in is not effective. >>>>>>>> >>>>>>>> Could you please try this in your build area: >>>>>>>> >>>>>>>> rm >>>>>>>> ./tools/gollvm/libgo/CMakeFiles/libgo_c_piclib.dir/__/gofrontend/libgo/go/syscall/errno.c.o >>>>>>>> ninja -v libgo_all >>>>>>>> >>>>>>>> and post the compile line for errno.c? >>>>>>>> >>>>>>>> Thanks, Than >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> On Tue, May 26, 2020 at 7:21 AM Yuan Ting <yuan...@ict.ac.cn> >>>>>>>> wrote: >>>>>>>> >>>>>>>>> Yes, I'm sure I fetched the last version of gollvm, >>>>>>>>> i.e. 0edc44a02852dd0d56cac35163a45634b497d6cd >>>>>>>>> >>>>>>>>> On Tuesday, May 26, 2020 at 6:55:08 PM UTC+8, Than McIntosh wrote: >>>>>>>>>> >>>>>>>>>> Hello, >>>>>>>>>> >>>>>>>>>> The ld.gold error message looks a lot like one reported in a >>>>>>>>>> previous bug, https://github.com/golang/go/issues/38728. >>>>>>>>>> >>>>>>>>>> Can you please check to make sure that your client is sync'd past >>>>>>>>>> the CLs that fix this issue, e.g. >>>>>>>>>> 0edc44a02852dd0d56cac35163a45634b497d6cd. >>>>>>>>>> >>>>>>>>>> Thanks, Than >>>>>>>>>> >>>>>>>>>> >>>>>>>>>> On Mon, May 25, 2020 at 10:15 PM Yuan Ting <yuan...@ict.ac.cn> >>>>>>>>>> wrote: >>>>>>>>>> >>>>>>>>>>> I tried to build gollvm in a docker container (the image is >>>>>>>>>>> based on ubuntu 20.04, and the host OS is MacOS catalina). I >>>>>>>>>>> configured >>>>>>>>>>> llvm by >>>>>>>>>>> >>>>>>>>>>> SHELL=/bin/sh cmake -DCMAKE_BUILD_TYPE=Release >>>>>>>>>>> -DLLVM_TARGETS_TO_BUILD=X86 -DLLVM_ENABLE_ASSERTIONS=On >>>>>>>>>>> -DLLVM_ENABLE_RTTI=On -DLLVM_USE_LINKER=gold -G Ninja ../llvm >>>>>>>>>>> >>>>>>>>>>> and then build and install gollvm by >>>>>>>>>>> >>>>>>>>>>> ninja gollvm && ninja install-gollvm >>>>>>>>>>> >>>>>>>>>>> after a while, I encountered >>>>>>>>>>> >>>>>>>>>>> ... >>>>>>>>>>> >>>>>>>>>>> error: At most two relocations per offset are supported >>>>>>>>>>> >>>>>>>>>>> error: At most two relocations per offset are supported >>>>>>>>>>> >>>>>>>>>>> error: At most two relocations per offset are supported >>>>>>>>>>> >>>>>>>>>>> [212/1279] Creating >>>>>>>>>>> /root/llvm-project/build-debug/tools/gollvm/libgo/tmp-sigtab.go >>>>>>>>>>> >>>>>>>>>>> FAILED: tools/gollvm/libgo/tmp-sigtab.go >>>>>>>>>>> >>>>>>>>>>> cd /root/llvm-project/build-debug/tools/gollvm/libgo && >>>>>>>>>>> GOARCH=amd64 GOOS=linux /bin/sh >>>>>>>>>>> /root/llvm-project/llvm/tools/gollvm/libgo/capturescript.sh >>>>>>>>>>> /root/llvm-project/llvm/tools/gollvm/gofrontend/libgo/mksigtab.sh >>>>>>>>>>> /root/llvm-project/build-debug/tools/gollvm/libgo/tmp-sigtab.go >>>>>>>>>>> >>>>>>>>>>> error: no SHELL setting >>>>>>>>>>> >>>>>>>>>>> [214/1279] Creating >>>>>>>>>>> /root/llvm-project/build-debug/tools/gollvm/libgo/tmp-sysinfo.go >>>>>>>>>>> >>>>>>>>>>> ninja: build stopped: subcommand failed. >>>>>>>>>>> >>>>>>>>>>> >>>>>>>>>>> By add SHELL=/bin/sh, the errors above seems to be skipped but >>>>>>>>>>> another period of time: >>>>>>>>>>> >>>>>>>>>>> >>>>>>>>>>> ... >>>>>>>>>>> >>>>>>>>>>> /usr/bin/ld.gold: error: >>>>>>>>>>> tools/gollvm/libgo/CMakeFiles/libgo_c_piclib.dir/__/gofrontend/libgo/go/syscall/errno.c.o: >>>>>>>>>>> >>>>>>>>>>> failed to match split-stack sequence at section 4 offset 0 >>>>>>>>>>> >>>>>>>>>>> /usr/bin/ld.gold: error: >>>>>>>>>>> tools/gollvm/libgo/CMakeFiles/libgo_c_piclib.dir/__/gofrontend/libgo/go/syscall/errno.c.o: >>>>>>>>>>> >>>>>>>>>>> failed to match split-stack sequence at section 6 offset 0 >>>>>>>>>>> >>>>>>>>>>> /usr/bin/ld.gold: error: >>>>>>>>>>> tools/gollvm/libgo/CMakeFiles/libgo_c_piclib.dir/__/gofrontend/libgo/go/syscall/signame.c.o: >>>>>>>>>>> >>>>>>>>>>> failed to match split-stack sequence at section 5 offset 0 >>>>>>>>>>> >>>>>>>>>>> /usr/bin/ld.gold: error: >>>>>>>>>>> tools/gollvm/libgo/CMakeFiles/libbacktrace_piclib.dir/libbacktrace/backtrace.c.o: >>>>>>>>>>> >>>>>>>>>>> failed to match split-stack sequence at section 4 offset 0 >>>>>>>>>>> >>>>>>>>>>> /usr/bin/ld.gold: error: >>>>>>>>>>> tools/gollvm/libgo/CMakeFiles/libbacktrace_piclib.dir/libbacktrace/backtrace.c.o: >>>>>>>>>>> >>>>>>>>>>> failed to match split-stack sequence at section 6 offset 0 >>>>>>>>>>> >>>>>>>>>>> /usr/bin/ld.gold: error: >>>>>>>>>>> tools/gollvm/libgo/CMakeFiles/libbacktrace_piclib.dir/libbacktrace/dwarf.c.o: >>>>>>>>>>> >>>>>>>>>>> failed to match split-stack sequence at section 18 offset 0 >>>>>>>>>>> >>>>>>>>>>> /usr/bin/ld.gold: error: >>>>>>>>>>> tools/gollvm/libgo/CMakeFiles/libbacktrace_piclib.dir/libbacktrace/dwarf.c.o: >>>>>>>>>>> >>>>>>>>>>> failed to match split-stack sequence at section 79 offset 0 >>>>>>>>>>> >>>>>>>>>>> ... >>>>>>>>>>> >>>>>>>>>>> /usr/bin/ld.gold: error: >>>>>>>>>>> tools/gollvm/libgo/CMakeFiles/libffi_piclib.dir/libffi/src/x86/ffi64.c.o: >>>>>>>>>>> >>>>>>>>>>> failed to match split-stack sequence at section 22 offset 0 >>>>>>>>>>> >>>>>>>>>>> /usr/bin/ld.gold: error: >>>>>>>>>>> tools/gollvm/libgo/CMakeFiles/libffi_piclib.dir/libffi/src/x86/ffi64.c.o: >>>>>>>>>>> >>>>>>>>>>> failed to match split-stack sequence at section 26 offset 0 >>>>>>>>>>> >>>>>>>>>>> /usr/bin/ld.gold: error: >>>>>>>>>>> tools/gollvm/libgo/CMakeFiles/libffi_piclib.dir/libffi/src/x86/ffiw64.c.o: >>>>>>>>>>> >>>>>>>>>>> failed to match split-stack sequence at section 16 offset 0 >>>>>>>>>>> >>>>>>>>>>> collect2: error: ld returned 1 exit status >>>>>>>>>>> >>>>>>>>>>> [901/1066] Linking C static library tools/gollvm/libgo/libgo.a >>>>>>>>>>> >>>>>>>>>>> ninja: build stopped: subcommand failed. >>>>>>>>>>> >>>>>>>>>>> >>>>>>>>>>> I'm not sure the error is caused by the container environment or >>>>>>>>>>> my configurations/other prerequisites. >>>>>>>>>>> >>>>>>>>>>> -- >>>>>>>>>>> You received this message because you are subscribed to the >>>>>>>>>>> Google Groups "golang-nuts" group. >>>>>>>>>>> To unsubscribe from this group and stop receiving emails from >>>>>>>>>>> it, send an email to golan...@googlegroups.com. >>>>>>>>>>> To view this discussion on the web visit >>>>>>>>>>> https://groups.google.com/d/msgid/golang-nuts/23309451-cb0a-4f7b-93a1-27038187302a%40googlegroups.com >>>>>>>>>>> >>>>>>>>>>> <https://groups.google.com/d/msgid/golang-nuts/23309451-cb0a-4f7b-93a1-27038187302a%40googlegroups.com?utm_medium=email&utm_source=footer> >>>>>>>>>>> . >>>>>>>>>>> >>>>>>>>>> -- >>>>>>>>> You received this message because you are subscribed to the Google >>>>>>>>> Groups "golang-nuts" group. >>>>>>>>> To unsubscribe from this group and stop receiving emails from it, >>>>>>>>> send an email to golan...@googlegroups.com. >>>>>>>>> To view this discussion on the web visit >>>>>>>>> https://groups.google.com/d/msgid/golang-nuts/1ffd3fb4-e739-4f13-9168-565e4c64bc70%40googlegroups.com >>>>>>>>> >>>>>>>>> <https://groups.google.com/d/msgid/golang-nuts/1ffd3fb4-e739-4f13-9168-565e4c64bc70%40googlegroups.com?utm_medium=email&utm_source=footer> >>>>>>>>> . >>>>>>>>> >>>>>>>> -- >>>>>>> You received this message because you are subscribed to the Google >>>>>>> Groups "golang-nuts" group. >>>>>>> To unsubscribe from this group and stop receiving emails from it, >>>>>>> send an email to golan...@googlegroups.com. >>>>>>> To view this discussion on the web visit >>>>>>> https://groups.google.com/d/msgid/golang-nuts/2e9d2af9-7d78-443b-9cda-34d7b9c41623%40googlegroups.com >>>>>>> >>>>>>> <https://groups.google.com/d/msgid/golang-nuts/2e9d2af9-7d78-443b-9cda-34d7b9c41623%40googlegroups.com?utm_medium=email&utm_source=footer> >>>>>>> . >>>>>>> >>>>>> -- >>>>> You received this message because you are subscribed to the Google >>>>> Groups "golang-nuts" group. >>>>> To unsubscribe from this group and stop receiving emails from it, send >>>>> an email to golan...@googlegroups.com. >>>>> To view this discussion on the web visit >>>>> https://groups.google.com/d/msgid/golang-nuts/975b8157-5795-4c09-9aed-91278aa5f0e6%40googlegroups.com >>>>> >>>>> <https://groups.google.com/d/msgid/golang-nuts/975b8157-5795-4c09-9aed-91278aa5f0e6%40googlegroups.com?utm_medium=email&utm_source=footer> >>>>> . >>>>> >>>> -- >>> You received this message because you are subscribed to the Google >>> Groups "golang-nuts" group. >>> To unsubscribe from this group and stop receiving emails from it, send >>> an email to golan...@googlegroups.com. >>> >> To view this discussion on the web visit >>> https://groups.google.com/d/msgid/golang-nuts/51910be4-ce4a-49a0-91f4-1e1cc63902f6%40googlegroups.com >>> >>> <https://groups.google.com/d/msgid/golang-nuts/51910be4-ce4a-49a0-91f4-1e1cc63902f6%40googlegroups.com?utm_medium=email&utm_source=footer> >>> . >>> >> -- You received this message because you are subscribed to the Google Groups "golang-nuts" group. To unsubscribe from this group and stop receiving emails from it, send an email to golang-nuts+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/golang-nuts/90a13fea-7ee9-42ae-b2b6-146c6835adf1n%40googlegroups.com.
