[Bug binutils/27078] New: c++filt: demangling for float128 type is confusing on PowerPC
https://sourceware.org/bugzilla/show_bug.cgi?id=27078
Bug ID: 27078
Summary: c++filt: demangling for float128 type is confusing on
PowerPC
Product: binutils
Version: 2.35
Status: UNCONFIRMED
Severity: normal
Priority: P2
Component: binutils
Assignee: unassigned at sourceware dot org
Reporter: qiu.chaofan at outlook dot com
Target Milestone: ---
Simple program:
__ibm128 foo(__ibm128 a, __ibm128 b) {
return a + b;
}
__float128 foo2(__float128 a, __float128 b) {
return a + b;
}
GCC will mangle the names as `_Z3foogg` and `_Z4foo2u9__ieee128u9__ieee128`.
However, use `c++filt` to demangle them, it outputs `foo(__float128,
__float128)` and `foo2(__ieee128, __ieee128)`. This is strange and confusing
since we use `u9__ieee128` for `__float128` type.
--
You are receiving this mail because:
You are on the CC list for the bug.
[Bug binutils/27081] New: Unexpected output on nm version 2.35.50.0.1.20201215
https://sourceware.org/bugzilla/show_bug.cgi?id=27081 Bug ID: 27081 Summary: Unexpected output on nm version 2.35.50.0.1.20201215 Product: binutils Version: unspecified Status: UNCONFIRMED Severity: normal Priority: P2 Component: binutils Assignee: unassigned at sourceware dot org Reporter: joao.moreira at intel dot com Target Milestone: --- GNU nm version 2.35.50.0.1.20201215 (which is part of the latest Fedora 33 package) has unexpected outputs. The bug was noticed while configuring upstream glibc with the respective package on Fedora 33. Invoking nm over a manually compiled test, similarly as done in the glibc configure file, will result in the following output: bfd plugin: No register_all_symbols_read hook And this causes the following error in the glibc configure: checking for __builtin_trap with no external dependencies... configure: error: confusing output from nm -u FTR, applying the below diff to the glibc configure works as a temporary workaround for the bug (but is hardly a solution). -libc_undefs=`$NM -u conftest.o | +libc_undefs=`$NM -u conftest.o | grep -v bfd | -- You are receiving this mail because: You are on the CC list for the bug.
[Bug ld/27082] New: Only one libdep.so should installed
https://sourceware.org/bugzilla/show_bug.cgi?id=27082 Bug ID: 27082 Summary: Only one libdep.so should installed Product: binutils Version: 2.36 (HEAD) Status: NEW Severity: normal Priority: P2 Component: ld Assignee: unassigned at sourceware dot org Reporter: hjl.tools at gmail dot com Target Milestone: --- [hjl@gnu-cfl-2 gcc]$ ls /usr/lib64/bfd-plugins libdep.so libdep.so.0 libdep.so.0.0.0 liblto_plugin.so LLVMgold.so [hjl@gnu-cfl-2 gcc]$ libdep.so.0 and libdep.so.0.0.0 are unnecessary. -- You are receiving this mail because: You are on the CC list for the bug.
[Bug binutils/27078] c++filt: demangling for float128 type is confusing on PowerPC
https://sourceware.org/bugzilla/show_bug.cgi?id=27078 Ian Lance Taylor changed: What|Removed |Added CC||ian at airs dot com --- Comment #1 from Ian Lance Taylor --- You filed this against the demangler, but it seems to me that the demangler is doing the right thing given the strings that it is given. If anything should change here, it seems that it should be the mangling. If you disagree, how do you think the demangler should handle this symbol name? -- You are receiving this mail because: You are on the CC list for the bug.
[Bug ld/27082] Only one libdep.so should installed
https://sourceware.org/bugzilla/show_bug.cgi?id=27082 H.J. Lu changed: What|Removed |Added Target Milestone|--- |2.36 -- You are receiving this mail because: You are on the CC list for the bug.
[Bug binutils/27081] Unexpected output on nm version 2.35.50.0.1.20201215
https://sourceware.org/bugzilla/show_bug.cgi?id=27081 --- Comment #1 from cvs-commit at gcc dot gnu.org --- The master branch has been updated by H.J. Lu : https://sourceware.org/git/gitweb.cgi?p=binutils-gdb.git;h=bcac599f73874fd63a3901f0a9f7d24e50c2ad24 commit bcac599f73874fd63a3901f0a9f7d24e50c2ad24 Author: H.J. Lu Date: Wed Dec 16 13:40:42 2020 -0800 ld: Skip libdep plugin if not all plugin hooks are available Skip plugin if not all required plugin hooks are available. 2020-12-16 Howard Chu H.J. Lu PR ld/27081 * libdep_plugin.c (onload): Skip if not all required plugin hooks are available. * testsuite/config/default.exp (dep_plug_opt): New. * testsuite/ld-elf/elf.exp: Pass $dep_plug_opt to nm. * testsuite/ld-elf/pr26391.fd: New file. -- You are receiving this mail because: You are on the CC list for the bug.
[Bug ld/27082] Only one libdep.so should installed
https://sourceware.org/bugzilla/show_bug.cgi?id=27082 --- Comment #1 from H.J. Lu --- [hjl@gnu-cfl-2 release]$ ls usr/local/lib/bfd-plugins libdep.a libdep.la libdep.so libdep.so.0 libdep.so.0.0.0 [hjl@gnu-cfl-2 release]$ Only one of .so file is needed. -- You are receiving this mail because: You are on the CC list for the bug.
[Bug ld/27082] Only one libdep.so should installed
https://sourceware.org/bugzilla/show_bug.cgi?id=27082 H.J. Lu changed: What|Removed |Added URL||https://sourceware.org/pipe ||rmail/binutils/2020-Decembe ||r/114553.html --- Comment #2 from H.J. Lu --- A patch is posted at https://sourceware.org/pipermail/binutils/2020-December/114553.html -- You are receiving this mail because: You are on the CC list for the bug.
[Bug binutils/27081] Unexpected output on nm version 2.35.50.0.1.20201215
https://sourceware.org/bugzilla/show_bug.cgi?id=27081 H.J. Lu changed: What|Removed |Added Resolution|--- |NOTABUG Target Milestone|--- |2.36 Status|UNCONFIRMED |RESOLVED Version|unspecified |2.36 (HEAD) --- Comment #2 from H.J. Lu --- Fixed for 2.36. -- You are receiving this mail because: You are on the CC list for the bug.
[Bug ld/27082] Only one libdep.so should installed
https://sourceware.org/bugzilla/show_bug.cgi?id=27082 Howard Chu changed: What|Removed |Added CC||hyc at symas dot com --- Comment #3 from Howard Chu --- (In reply to H.J. Lu from comment #2) > A patch is posted at > > https://sourceware.org/pipermail/binutils/2020-December/114553.html Looking at the other two plugins there, shouldn't it instead keep the *.so file, and delete the .so.0 and .so.0.0.0 ? (I recognize that the .so is just a symlink to the versioned filename, so this is a bit awkward.) -- You are receiving this mail because: You are on the CC list for the bug.
[Bug ld/27082] Only one libdep.so should installed
https://sourceware.org/bugzilla/show_bug.cgi?id=27082 --- Comment #4 from H.J. Lu --- (In reply to Howard Chu from comment #3) > (In reply to H.J. Lu from comment #2) > > A patch is posted at > > > > https://sourceware.org/pipermail/binutils/2020-December/114553.html > > Looking at the other two plugins there, shouldn't it instead keep the *.so > file, and delete the .so.0 and .so.0.0.0 ? (I recognize that the .so is > just a > symlink to the versioned filename, so this is a bit awkward.) Any .so file is ok. -- You are receiving this mail because: You are on the CC list for the bug.
[Bug ld/27082] Only one libdep.so should installed
https://sourceware.org/bugzilla/show_bug.cgi?id=27082 --- Comment #5 from H.J. Lu --- We can just build libdep and install it manually. -- You are receiving this mail because: You are on the CC list for the bug.
[Bug ld/27082] Only one libdep.so should installed
https://sourceware.org/bugzilla/show_bug.cgi?id=27082 --- Comment #6 from Howard Chu --- (In reply to H.J. Lu from comment #5) > We can just build libdep and install it manually. I took a look at how liblto-plugin is built https://gcc.gnu.org/git/?p=gcc.git;a=tree;f=lto-plugin;h=ab65b7bf798a806198824c7fc578b9f3f4d4c62f;hb=64a54505ec8249178b9767d1420354f8eb55de50 We can get rid of the version and static libraries using this patch diff --git a/ld/Makefile.am b/ld/Makefile.am index 2f59ee6169..929c017a85 100644 --- a/ld/Makefile.am +++ b/ld/Makefile.am @@ -983,6 +983,7 @@ enablings.exp: # # Build a dummy plugin using libtool. # +AM_LIBTOOLFLAGS = --tag=disable-static noinst_LTLIBRARIES = libldtestplug.la libldtestplug2.la \ libldtestplug3.la libldtestplug4.la libldtestplug_la_SOURCES = testplug.c @@ -1001,7 +1002,7 @@ libldtestplug4_la_LDFLAGS = -no-undefined -rpath /nowhere bfdplugindir = $(libdir)/bfd-plugins bfdplugin_LTLIBRARIES = libdep.la libdep_la_SOURCES = libdep_plugin.c -libdep_la_LDFLAGS = -no-undefined -rpath /nowhere +libdep_la_LDFLAGS = -module -avoid-version # DOCUMENTATION TARGETS # Manual configuration file; not usually attached to normal configuration, It will still install the .la file though. -- You are receiving this mail because: You are on the CC list for the bug.
[Bug ld/27082] Only one libdep.so should installed
https://sourceware.org/bugzilla/show_bug.cgi?id=27082 --- Comment #7 from H.J. Lu --- (In reply to Howard Chu from comment #6) > (In reply to H.J. Lu from comment #5) > > We can just build libdep and install it manually. > > I took a look at how liblto-plugin is built > https://gcc.gnu.org/git/?p=gcc.git;a=tree;f=lto-plugin; > h=ab65b7bf798a806198824c7fc578b9f3f4d4c62f; > hb=64a54505ec8249178b9767d1420354f8eb55de50 > > We can get rid of the version and static libraries using this patch > > diff --git a/ld/Makefile.am b/ld/Makefile.am > index 2f59ee6169..929c017a85 100644 > --- a/ld/Makefile.am > +++ b/ld/Makefile.am > @@ -983,6 +983,7 @@ enablings.exp: > # > # Build a dummy plugin using libtool. > # > +AM_LIBTOOLFLAGS = --tag=disable-static > noinst_LTLIBRARIES = libldtestplug.la libldtestplug2.la \ > libldtestplug3.la libldtestplug4.la > libldtestplug_la_SOURCES = testplug.c > @@ -1001,7 +1002,7 @@ libldtestplug4_la_LDFLAGS = -no-undefined -rpath > /nowhere > bfdplugindir = $(libdir)/bfd-plugins > bfdplugin_LTLIBRARIES = libdep.la > libdep_la_SOURCES = libdep_plugin.c > -libdep_la_LDFLAGS = -no-undefined -rpath /nowhere > +libdep_la_LDFLAGS = -module -avoid-version > > # DOCUMENTATION TARGETS > # Manual configuration file; not usually attached to normal configuration, > > It will still install the .la file though. 1. Just remove the .la file after it is installed during "make install". Or 2. Add libdep.la to noinst_LTLIBRARIES and install libdep.so from "make install-data-local". -- You are receiving this mail because: You are on the CC list for the bug.
[Bug ld/26256] Support mixed SHF_LINK_ORDER & non-SHF_LINK_ORDER components in an output section
https://sourceware.org/bugzilla/show_bug.cgi?id=26256 --- Comment #3 from Fangrui Song --- You can use the tests https://github.com/llvm/llvm-project/blob/main/lld/test/ELF/linkerscript/linkorder.s https://github.com/llvm/llvm-project/blob/main/lld/test/ELF/linkorder-mixed.s -- You are receiving this mail because: You are on the CC list for the bug.
[Bug binutils/27078] c++filt: demangling for float128 type is confusing on PowerPC
https://sourceware.org/bugzilla/show_bug.cgi?id=27078 --- Comment #2 from Qiu Chaofan --- Thanks for reply. I mean `__ibm128` is mangled as `g`, but `g` is demangled into `__float128`, which is NOT mangled into `g`. On PPC, the mangling looks correct to me. If in the perspective of demangler, `__float128` is alias to `__ibm128`, that looks reasonable. But this still looks like some type names inconsistency (between compiler and demangler). -- You are receiving this mail because: You are on the CC list for the bug.
