[Bug dyninst/22004] New: dyninst gets a SIGSEGV with glibc 2.25.90

classic Classic list List threaded Threaded
5 messages Options
Reply | Threaded
Open this post in threaded view
|

[Bug dyninst/22004] New: dyninst gets a SIGSEGV with glibc 2.25.90

schwab@linux-m68k.org
https://sourceware.org/bugzilla/show_bug.cgi?id=22004

            Bug ID: 22004
           Summary: dyninst gets a SIGSEGV with glibc 2.25.90
           Product: systemtap
           Version: unspecified
            Status: NEW
          Severity: normal
          Priority: P2
         Component: dyninst
          Assignee: systemtap at sourceware dot org
          Reporter: scox at redhat dot com
  Target Milestone: ---

--
You are receiving this mail because:
You are the assignee for the bug.
Reply | Threaded
Open this post in threaded view
|

[Bug dyninst/22004] dyninst gets a SIGSEGV with glibc 2.25.90

schwab@linux-m68k.org
https://sourceware.org/bugzilla/show_bug.cgi?id=22004

--- Comment #1 from Stan Cox <scox at redhat dot com> ---
The problem seems related to the expanded use of ifunc in libc 2.25.90. This
could occur with any use of ifunc most probably, it just so happens recent libc
has expanded use of it so we will probably see expanded use of it:
ifunc description

Going through the last few stack frame entries where things go awry:

frame 5 0x00007fca65776c99 in Dyninst::ParseAPI::CFGFactory::_mkfunc
(this=this@entry=0x9aa2c0, addr=addr@entry=132480,
src=src@entry=Dyninst::ParseAPI::RT, name="targ20580", obj=obj
@Entry=0x1076590, reg=reg@entry=0x1d058f0, isrc=0x21744c0)
at
/root/brew-rpms/dyninst-9.3.2-1/BUILD/dyninst-9.3.2/dyninst-9.3.2/parseAPI/src/CFGFactory.C:91
91 Function * ret = mkfunc(addr,src,name,obj,reg,isrc);

where addr = 0x20580 is
Disassembly of section .plt:
0000000000020580 <ABS+0x98e00 at plt>:
20580: ff 25 fa 0a 3c 00 jmpq *0x3c0afa(%rip) # 3e1080
<GLOBAL_OFFSET_TABLE+0x80>
20586: 68 19 00 00 00 pushq $0x19
2058b: e9 10 ff ff ff jmpq 204a0 <catch_hook+0x20418>

where 3e1080 is
Relocation section '.rela.plt' at offset 0x200f0 contains 39 entries:
Offset Info Type Sym. Value Sym. Name + Addend
...
0000003e1080 000000000025 R_X86_64_IRELATIV 98e00

where 98e00 is
Symbol table '.dynsym' contains 2294 entries:
793: 0000000000098e00 42 IFUNC GLOBAL DEFAULT 13 strlen @@ GLIBC_2.2.5

frame 4 0x00007fca65d98813 in DynCFGFactory::mkfunc (this=this @
entry=0x9aa2c0, addr=addr@entry=132480,
src=src@entry=Dyninst::ParseAPI::RT, name="", obj=obj @ entry=0x1076590,
reg=reg@entry=0x1d058f0, isrc=)
at
/root/brew-rpms/dyninst-9.3.2-1/BUILD/dyninst-9.3.2/dyninst-9.3.2/dyninstAPI/src/Parsing.C:164
164 ret = new parse_func(stf, pdmod,_img,obj,reg,isrc,src);

where stf is stf = new PLTFunction(*i);
(if I understand things correctly a function object is being created for the
plt entry)
where *i is the relocation for this plt entry. For ifunc this is the
R_*IRELATIV relocation.
this results in stf.firstSymbol = NULL

frame 3 0x00007fca65d7b5c0 in parse_func::parse_func (this=0x2613a10,
func=0x2611c60, m=0x8c5770, i=0x816e50, obj=0x1076590,
reg=0x1d058f0, isrc=0x21744c0, src=Dyninst::ParseAPI::RT)
at
/root/brew-rpms/dyninst-9.3.2-1/BUILD/dyninst-9.3.2/dyninst-9.3.2/dyninstAPI/src/parse-cfg.C:91
91
Function(func->getOffset(),func->getFirstSymbol()->getMangledName(),obj,reg,isrc),

where func->getFirstSymbol() is NULL so we get a SIGSEGV

--
You are receiving this mail because:
You are the assignee for the bug.
Reply | Threaded
Open this post in threaded view
|

[Bug dyninst/22004] dyninst does not handle R_*_IRELATIV in .rela.plt

schwab@linux-m68k.org
In reply to this post by schwab@linux-m68k.org
https://sourceware.org/bugzilla/show_bug.cgi?id=22004

Stan Cox <scox at redhat dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|dyninst gets a SIGSEGV with |dyninst does not handle
                   |glibc 2.25.90               |R_*_IRELATIV in .rela.plt

--
You are receiving this mail because:
You are the assignee for the bug.
Reply | Threaded
Open this post in threaded view
|

[Bug dyninst/22004] dyninst does not handle R_*_IRELATIV in .rela.plt

schwab@linux-m68k.org
In reply to this post by schwab@linux-m68k.org
https://sourceware.org/bugzilla/show_bug.cgi?id=22004

Stan Cox <scox at redhat dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |ASSIGNED

--- Comment #2 from Stan Cox <scox at redhat dot com> ---
(https://github.com/dyninst/dyninst/issues/396)

--
You are receiving this mail because:
You are the assignee for the bug.
Reply | Threaded
Open this post in threaded view
|

[Bug dyninst/22004] dyninst does not handle R_*_IRELATIV in .rela.plt

schwab@linux-m68k.org
In reply to this post by schwab@linux-m68k.org
https://sourceware.org/bugzilla/show_bug.cgi?id=22004

Stan Cox <scox at redhat dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |RESOLVED
         Resolution|---                         |FIXED

--- Comment #3 from Stan Cox <scox at redhat dot com> ---
Fixed in dyninst-9.3.2-7
Add linux-vdso64.so.1 to blackout list in addrtranslate-sysv.C

--
You are receiving this mail because:
You are the assignee for the bug.