Re: [COMMITTED] Prepare for 0.176
On 2/14/19 9:37 PM, Mark Wielaard wrote: > On Thu, 2019-02-14 at 20:51 +0100, Martin Liška wrote: >> I see 2 tests failing for s390x: >> > https://build.opensuse.org/package/live_build_log/home:marxin:branches:Base:System/elfutils/openSUSE_Factory_zSystems/s390x >> > > So those are basically the same issue: > >> [ 269s] elflint /home/abuild/rpmbuild/BUILD/elfutils-0.176/tests/elfstrmerge >> [ 269s] section [34] '.symtab': _GLOBAL_OFFSET_TABLE_ symbol value 0x5fb8 >> does not match .got.plt section address 0x6000 > > First, is that correct? > Could you provide that binary? Hi. Attached. > > Secondly, when did this start happening? > Did you change linker/version? I can confirm it's older, I see in on elfutils-0.175 with binutils 2.31. > > It might be this binutils commit: > https://sourceware.org/ml/binutils/2018-07/msg00200.html > > Which isn't in binutils-2.29 which is used on the s390x fedora builder. > On which this test passes. > > If so, you might have to add a check_special_symbol hook like aarch64 > has (see backends/aarch64_symbol.c). Please let me know and we can eventually create a PR for it. Thanks, Martin > > Cheers, > > Mark > elfstrmerge Description: Binary data
Re: [COMMITTED] Prepare for 0.176
On Fri, 2019-02-15 at 09:42 +0100, Martin Liška wrote:
> > It might be this binutils commit:
> > https://sourceware.org/ml/binutils/2018-07/msg00200.html
> >
> > Which isn't in binutils-2.29 which is used on the s390x fedora
> > builder.
> > On which this test passes.
> >
> > If so, you might have to add a check_special_symbol hook like
> > aarch64
> > has (see backends/aarch64_symbol.c).
>
> Please let me know and we can eventually create a PR for it.
It looks like that was it. Could you try the attached patch?
I'll run it on some older s390x setups.
Thanks,
Mark
From 1e52d4ce3aa2093d12901d32fe07aae70211fe2a Mon Sep 17 00:00:00 2001
From: Mark Wielaard
Date: Fri, 15 Feb 2019 14:39:57 +0100
Subject: [PATCH] s390: elflint should check if _GLOBAL_OFFSET_TABLE_ points to
.got.
The _GLOBAL_OFFSET_TABLE_ symbol might point to the DT_PLTGOT,
which is in the .got section, even if the symbol itself is
associated with the .got.plt section.
See https://sourceware.org/ml/binutils/2018-07/msg00200.html
Signed-off-by: Mark Wielaard
---
backends/ChangeLog | 5 +
backends/s390_init.c | 1 +
backends/s390_symbol.c | 38 ++
3 files changed, 44 insertions(+)
diff --git a/backends/ChangeLog b/backends/ChangeLog
index 58a1b77..0c61a0b 100644
--- a/backends/ChangeLog
+++ b/backends/ChangeLog
@@ -1,3 +1,8 @@
+2019-02-15 Mark Wielaard
+
+ * s390_init.c (s390_init): Hook check_special_symbol.
+ * s390_symbol.c (s390_check_sepcial_symbol): New function.
+
2018-12-27 Jim Wilson
* Makefile.am (riscv_SRCS): Add riscv64_corenote.c.
diff --git a/backends/s390_init.c b/backends/s390_init.c
index ba8df45..0004aee 100644
--- a/backends/s390_init.c
+++ b/backends/s390_init.c
@@ -54,6 +54,7 @@ s390_init (Elf *elf __attribute__ ((unused)),
eh->name = "IBM S/390";
s390_init_reloc (eh);
HOOK (eh, reloc_simple_type);
+ HOOK (eh, check_special_symbol);
HOOK (eh, register_info);
HOOK (eh, return_value_location);
if (eh->class == ELFCLASS64)
diff --git a/backends/s390_symbol.c b/backends/s390_symbol.c
index f91e137..9e80eca 100644
--- a/backends/s390_symbol.c
+++ b/backends/s390_symbol.c
@@ -32,6 +32,7 @@
#include
#include
+#include
#define BACKEND s390_
#include "libebl_CPU.h"
@@ -55,3 +56,40 @@ s390_reloc_simple_type (Ebl *ebl __attribute__ ((unused)), int type,
return ELF_T_NUM;
}
}
+
+/* The _GLOBAL_OFFSET_TABLE_ symbol might point to the DT_PLTGOT,
+ which is in the .got section, even if the symbol itself is
+ associated with the is a .got.plt section.
+ https://sourceware.org/ml/binutils/2018-07/msg00200.html */
+bool
+s390_check_special_symbol (Elf *elf, const GElf_Sym *sym,
+ const char *name, const GElf_Shdr *destshdr)
+{
+ if (name != NULL
+ && strcmp (name, "_GLOBAL_OFFSET_TABLE_") == 0)
+{
+ size_t shstrndx;
+ if (elf_getshdrstrndx (elf, &shstrndx) != 0)
+ return false;
+ const char *sname = elf_strptr (elf, shstrndx, destshdr->sh_name);
+ if (sname != NULL
+ && (strcmp (sname, ".got") == 0 || strcmp (sname, ".got.plt") == 0))
+ {
+ Elf_Scn *scn = NULL;
+ while ((scn = elf_nextscn (elf, scn)) != NULL)
+ {
+ GElf_Shdr shdr_mem;
+ GElf_Shdr *shdr = gelf_getshdr (scn, &shdr_mem);
+ if (shdr != NULL)
+ {
+ sname = elf_strptr (elf, shstrndx, shdr->sh_name);
+ if (sname != NULL && strcmp (sname, ".got") == 0)
+ return (sym->st_value >= shdr->sh_addr
+ && sym->st_value < shdr->sh_addr + shdr->sh_size);
+ }
+ }
+ }
+}
+
+ return false;
+}
--
1.8.3.1
Re: [COMMITTED] Prepare for 0.176
On Fri, 2019-02-15 at 14:43 +0100, Mark Wielaard wrote: > On Fri, 2019-02-15 at 09:42 +0100, Martin Liška wrote: > > > It might be this binutils commit: > > > https://sourceware.org/ml/binutils/2018-07/msg00200.html > > > > > > Which isn't in binutils-2.29 which is used on the s390x fedora > > > builder. > > > On which this test passes. > > > > > > If so, you might have to add a check_special_symbol hook like > > > aarch64 > > > has (see backends/aarch64_symbol.c). > > > > Please let me know and we can eventually create a PR for it. > > It looks like that was it. Could you try the attached patch? > I'll run it on some older s390x setups. This seemed to work out nicely. Both on an older binutils setup and a newer one. So I have pushed the commit. If the buildbot keeps being green I'll do a release soon. Thanks, Mark
elfutils 0.176 released
ELFUTILS 0.176 - http://elfutils.org/ A new release of elfutils is available at: ftp://sourceware.org/pub/elfutils/0.176/ or https://sourceware.org/elfutils/ftp/0.176/ * NEWS * build: Add new --enable-install-elfh option. Do NOT use this for system installs (it overrides glibc elf.h). backends: riscv improved core file and return value location support. Fixes CVE-2019-7146, CVE-2019-7148, CVE-2019-7149, CVE-2019-7150, CVE-2019-7664, CVE-2019-7665 * GIT SHORTLOG * Jim Wilson (3): RISC-V: Improve riscv64 core file support. RISC-V: Add initial return value location support. RISC-V: Add untested 32-bit core file support. Mark Wielaard (23): config/upload-release.sh: Need to make before make dist. Add -Wtrampolines to CFLAGS. libelf: Get alignment correct when calling conversion functions. tests: Call test_cleanup in backtrace-subr.sh check_unsupported. libdw: Enable building with -Og. libdwfl: Fix relocation overlap sanity check. tests: Improve backtrace-data SKIP message. libebl: Fix reading GNU_PROPERTY_STACK_SIZE reading from 32bit notes. libebl: Check GNU property note pr_datasz fits inside note description. libelf: Correct overflow check in note_xlate. libebl: Check NT_PLATFORM core notes contain a zero terminated string. libdwfl: Sanity check partial core file dyn data read. libdw: Check terminating NUL byte in dwarf_getsrclines for dir/file table. readelf: Don't go past end of line data reading unknown opcode parameters. strip: Fix check test for SHN_XINDEX symbol. libebl: Check GNU property note data padding fits inside note. tests: Remove assert (errno == 0) from tests. configure: Add new --enable-install-elfh option. readelf: Check there is enough data to read DWARF line opcodes arguments. libdw: Check there is enough space for CU 64bit length, version and type. libelf: Make sure ar_size is terminated when reading ar long names. Prepare for 0.176 s390: elflint should check if _GLOBAL_OFFSET_TABLE_ points to .got. Ulf Hermann (2): Skip run-readelf-compressed.sh test if built without bzip2 tests: Use separate files for strip outputs Yonghong Song (2): libdwfl: parse inode in /proc/pid/maps correctly tests: parse inode in /proc/pid/maps/correctly in run-backtrace-data.sh signature.asc Description: PGP signature
