CVS commit: src/libexec/ld.elf_so/arch/powerpc
Module Name:src Committed By: chs Date: Sun Apr 9 17:24:48 UTC 2023 Modified Files: src/libexec/ld.elf_so/arch/powerpc: Makefile.inc Removed Files: src/libexec/ld.elf_so/arch/powerpc: ld.so.script Log Message: ld.elf_so powerpc: remove bogus ldscript This ldscript is not needed and actually makes things worse by putting everything in one LOAD section, which then needs to have rwx permission. Remove it so that we get two LOAD sections with better permissions. Fixes PR 57323. To generate a diff of this commit: cvs rdiff -u -r1.16 -r1.17 src/libexec/ld.elf_so/arch/powerpc/Makefile.inc cvs rdiff -u -r1.4 -r0 src/libexec/ld.elf_so/arch/powerpc/ld.so.script Please note that diffs are not public domain; they are subject to the copyright notices on the relevant files. Modified files: Index: src/libexec/ld.elf_so/arch/powerpc/Makefile.inc diff -u src/libexec/ld.elf_so/arch/powerpc/Makefile.inc:1.16 src/libexec/ld.elf_so/arch/powerpc/Makefile.inc:1.17 --- src/libexec/ld.elf_so/arch/powerpc/Makefile.inc:1.16 Tue Apr 3 21:10:27 2018 +++ src/libexec/ld.elf_so/arch/powerpc/Makefile.inc Sun Apr 9 17:24:48 2023 @@ -1,4 +1,4 @@ -# $NetBSD: Makefile.inc,v 1.16 2018/04/03 21:10:27 joerg Exp $ +# $NetBSD: Makefile.inc,v 1.17 2023/04/09 17:24:48 chs Exp $ SRCS+= ppc_reloc.c LDFLAGS+= -Wl,-e,_rtld_start @@ -12,5 +12,4 @@ CPPFLAGS+= -DELFSIZE=64 .else SRCS+= rtld_start.S CPPFLAGS+= -DELFSIZE=32 -LDFLAGS+= -Wl,--script,${.CURDIR}/arch/powerpc/ld.so.script .endif
CVS commit: src/libexec/ld.elf_so/arch/powerpc
Module Name:src Committed By: chs Date: Sun Apr 9 17:24:48 UTC 2023 Modified Files: src/libexec/ld.elf_so/arch/powerpc: Makefile.inc Removed Files: src/libexec/ld.elf_so/arch/powerpc: ld.so.script Log Message: ld.elf_so powerpc: remove bogus ldscript This ldscript is not needed and actually makes things worse by putting everything in one LOAD section, which then needs to have rwx permission. Remove it so that we get two LOAD sections with better permissions. Fixes PR 57323. To generate a diff of this commit: cvs rdiff -u -r1.16 -r1.17 src/libexec/ld.elf_so/arch/powerpc/Makefile.inc cvs rdiff -u -r1.4 -r0 src/libexec/ld.elf_so/arch/powerpc/ld.so.script Please note that diffs are not public domain; they are subject to the copyright notices on the relevant files.
CVS commit: src/libexec/ld.elf_so/arch/powerpc
Module Name:src Committed By: martin Date: Wed Jul 6 17:35:20 UTC 2022 Modified Files: src/libexec/ld.elf_so/arch/powerpc: ppc_reloc.c Log Message: Fix copy & pasto: DTPREL relocations do not need to allocate a static TLS index. Patch from joerg@ To generate a diff of this commit: cvs rdiff -u -r1.61 -r1.62 src/libexec/ld.elf_so/arch/powerpc/ppc_reloc.c Please note that diffs are not public domain; they are subject to the copyright notices on the relevant files. Modified files: Index: src/libexec/ld.elf_so/arch/powerpc/ppc_reloc.c diff -u src/libexec/ld.elf_so/arch/powerpc/ppc_reloc.c:1.61 src/libexec/ld.elf_so/arch/powerpc/ppc_reloc.c:1.62 --- src/libexec/ld.elf_so/arch/powerpc/ppc_reloc.c:1.61 Sun May 24 02:33:11 2020 +++ src/libexec/ld.elf_so/arch/powerpc/ppc_reloc.c Wed Jul 6 17:35:20 2022 @@ -1,4 +1,4 @@ -/* $NetBSD: ppc_reloc.c,v 1.61 2020/05/24 02:33:11 macallan Exp $ */ +/* $NetBSD: ppc_reloc.c,v 1.62 2022/07/06 17:35:20 martin Exp $ */ /*- * Copyright (C) 1998 Tsubai Masanari @@ -30,7 +30,7 @@ #include #ifndef lint -__RCSID("$NetBSD: ppc_reloc.c,v 1.61 2020/05/24 02:33:11 macallan Exp $"); +__RCSID("$NetBSD: ppc_reloc.c,v 1.62 2022/07/06 17:35:20 martin Exp $"); #endif /* not lint */ #include @@ -313,9 +313,6 @@ _rtld_relocate_nonplt_objects(Obj_Entry break; case R_TYPE(DTPREL): - if (!defobj->tls_done && _rtld_tls_offset_allocate(obj)) -return -1; - *where = (Elf_Addr)(def->st_value + rela->r_addend - TLS_DTV_OFFSET); rdbg(("DTPREL32 %s in %s --> %p in %s",
CVS commit: src/libexec/ld.elf_so/arch/powerpc
Module Name:src Committed By: martin Date: Wed Jul 6 17:35:20 UTC 2022 Modified Files: src/libexec/ld.elf_so/arch/powerpc: ppc_reloc.c Log Message: Fix copy & pasto: DTPREL relocations do not need to allocate a static TLS index. Patch from joerg@ To generate a diff of this commit: cvs rdiff -u -r1.61 -r1.62 src/libexec/ld.elf_so/arch/powerpc/ppc_reloc.c Please note that diffs are not public domain; they are subject to the copyright notices on the relevant files.
Re: CVS commit: src/libexec/ld.elf_so/arch/powerpc
On Fri, Oct 28, 2011 at 07:53:22AM +0100, David Laight wrote: On Wed, Oct 26, 2011 at 03:56:58PM +, Chuck Silvers wrote: Module Name:src Committed By: chs Date: Wed Oct 26 15:56:58 UTC 2011 Modified Files: src/libexec/ld.elf_so/arch/powerpc: Makefile.inc Log Message: avoid linker warnings with the new binutils by not forcing ld.elf_so to be soft-float. the floating-point usage in ld.elf_so is only because of stdarg functions like printf(), and gcc 4.5 is good about not actually executing the FP instructions in such functions unless FP values are passed to them, which ld.elf_so doesn't do. I thought gcc had a habit of using FP registers for memory copy? (Or some other integer-only action) Or is that only a problem with specific architectures due to the way the instruction set has been defined to gcc. that kind of thing is always platform-specific. the only FP instructions that I see in ld.elf_so on powerpc are for stdarg stuff, so it doesn't look like it's a problem here. -Chuck
Re: CVS commit: src/libexec/ld.elf_so/arch/powerpc
On Wed, Oct 26, 2011 at 03:56:58PM +, Chuck Silvers wrote: Module Name: src Committed By: chs Date: Wed Oct 26 15:56:58 UTC 2011 Modified Files: src/libexec/ld.elf_so/arch/powerpc: Makefile.inc Log Message: avoid linker warnings with the new binutils by not forcing ld.elf_so to be soft-float. the floating-point usage in ld.elf_so is only because of stdarg functions like printf(), and gcc 4.5 is good about not actually executing the FP instructions in such functions unless FP values are passed to them, which ld.elf_so doesn't do. I thought gcc had a habit of using FP registers for memory copy? (Or some other integer-only action) Or is that only a problem with specific architectures due to the way the instruction set has been defined to gcc. David -- David Laight: da...@l8s.co.uk