From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 4560 invoked by alias); 16 Dec 2006 23:46:13 -0000 Received: (qmail 4548 invoked by uid 22791); 16 Dec 2006 23:46:11 -0000 X-Spam-Check-By: sourceware.org Received: from 195.22.55.53.adsl.nextra.cz (HELO host0.dyn.jankratochvil.net) (195.22.55.53) by sourceware.org (qpsmtpd/0.31) with ESMTP; Sat, 16 Dec 2006 23:46:07 +0000 Received: from host0.dyn.jankratochvil.net (localhost [127.0.0.1]) by host0.dyn.jankratochvil.net (8.13.8/8.13.8) with ESMTP id kBGNjs2Z007117; Sun, 17 Dec 2006 00:45:54 +0100 Received: (from jkratoch@localhost) by host0.dyn.jankratochvil.net (8.13.8/8.13.8/Submit) id kBGNjr7M007116; Sun, 17 Dec 2006 00:45:53 +0100 Date: Sat, 16 Dec 2006 23:46:00 -0000 From: Jan Kratochvil To: Mark Kettenis Cc: gdb-patches@sourceware.org, Daniel Jacobowitz , Jakub Jelinek Subject: Re: [PATCH] Re: Unwinding CFI gcc practice of assumed `same value' regs Message-ID: <20061216234551.GA6766@host0.dyn.jankratochvil.net> References: <20061211190300.GA4372@host0.dyn.jankratochvil.net> <20061211224022.AD76E1800E7@magilla.sf.frob.com> <20061212155233.GH29911@devserv.devel.redhat.com> <20061213204603.GA11741@host0.dyn.jankratochvil.net> <20061213210357.GA27039@nevyn.them.org> <23342.82.92.89.47.1166046052.squirrel@webmail.xs4all.nl> <20061213214540.GA28965@nevyn.them.org> <5950.82.92.89.47.1166046772.squirrel@webmail.xs4all.nl> <20061213221111.GA26406@host0.dyn.jankratochvil.net> Mime-Version: 1.0 Content-Type: multipart/mixed; boundary="MGYHOYXEY6WxJCY8" Content-Disposition: inline In-Reply-To: <20061213221111.GA26406@host0.dyn.jankratochvil.net> User-Agent: Mutt/1.4.2.2i X-IsSubscribed: yes Mailing-List: contact gdb-patches-help@sourceware.org; run by ezmlm Precedence: bulk List-Id: List-Subscribe: List-Archive: List-Post: List-Help: , Sender: gdb-patches-owner@sourceware.org X-SW-Source: 2006-12/txt/msg00217.txt.bz2 --MGYHOYXEY6WxJCY8 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline Content-length: 1451 On Wed, 13 Dec 2006 23:11:11 +0100, Jan Kratochvil wrote: > On Wed, 13 Dec 2006 22:52:52 +0100, Mark Kettenis wrote: > ... > > No, sorry, it is me who is confused. I didn't actually look at Jan's patch > > and assumed it did the the right thing of marking the return address as > > undefined. > > Therefore what should the patch do? Currently `.cfi_undefined' looks as too > radical. Read all the past threads and I hope this patch complies with all the parties. I would like if the `.cfi_undefined' patch http://sourceware.org/ml/libc-alpha/2006-11/msg00082.html could stay in the development glibc with the unwinder patched by Jakub Jelinek http://sources.redhat.com/ml/gdb/2006-12/msg00100.html as with such CFI patch gdb does not show any bogus 0x0 unwind of `clone'. On the other hand as a legacy compatibility I wrote the attached patch to fix the GDB excessive non-CFI amd64 `clone' unwind as 0x0. Other platforms (and functions?) could be carbon-copied from this one. I consider it the same approach as the way GDB currently on amd64 detects signal frames by checking the instructions - the CFI for signal frames has been also recently checked to the development glibc as was the `clone' CFI. Tested for regressions by the gdb testsuite, it is the most conservative way so it should not have any sideeffects. The 0x0 unwinding of any non-`clone' functions remains the same so Mark should have no objections, I hope. Regards, Jan --MGYHOYXEY6WxJCY8 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline; filename="gdb-6.5-clone-is-outermost.patch" Content-length: 8151 2006-12-16 Jan Kratochvil * gdb/amd64-linux-tdep.c (linux_clone_code, LINUX_CLONE_LEN, amd64_linux_clone_running, amd64_linux_outermost_frame): Detect the caller of the thread `start_routing' as the outermost frame to unwind. (amd64_linux_init_abi): New hooking for `amd64_linux_outermost_frame'. * gdb/i386-tdep.h (gdbarch_tdep): New hook `outermost_frame_p'. * gdb/i386-tdep.c (i386_gdbarch_init): Likewise. * gdb/amd64-tdep.c (amd64_frame_this_id): Call `outermost_frame_p'. 2006-12-13 Jan Kratochvil * gdb.threads/bt-clone-stop.exp, gdb.threads/bt-clone-stop.c: Backtraced `clone' must not have `PC == 0' as its previous frame. --- ./gdb/amd64-linux-tdep.c 19 Aug 2006 15:15:18 -0000 1.12 +++ ./gdb/amd64-linux-tdep.c 16 Dec 2006 20:15:16 -0000 @@ -235,6 +235,72 @@ amd64_linux_register_reggroup_p (struct /* Set the program counter for process PTID to PC. */ +/* Detect the outermost frame as of: + #5 0x000000305cec68c3 in clone () from /lib64/tls/libc.so.6 + We compare the `linux_clone_code' block to be _before_ the unwound PC. */ + +static const unsigned char linux_clone_code[] = +{ +/* libc/sysdeps/unix/sysv/linux/x86_64/clone.S */ +/* #ifdef RESET_PID */ +/* ... */ +/* movl $SYS_ify(getpid), %eax */ + 0x48, 0xc7, 0xc0, 0x27, 0x00, 0x00, 0x00, +/* syscall */ + 0x0f, 0x05, +/* movl %eax, %fs:PID */ + 0x64, 0x89, 0x04, 0x25, 0x94, 0x00, 0x00, 0x00, +/* movl %eax, %fs:TID */ + 0x64, 0x89, 0x04, 0x25, 0x90, 0x00, 0x00, 0x00, +/* #endif */ +/* |* Set up arguments for the function call. *| */ +/* popq %rax |* Function to call. *| */ + 0x58, +/* popq %rdi |* Argument. *| */ + 0x5f, +/* call *%rax$ */ + 0xff, 0xd0 +}; + +#define LINUX_CLONE_LEN (sizeof linux_clone_code) + +static int +amd64_linux_clone_running (struct frame_info *next_frame) +{ + CORE_ADDR pc = frame_pc_unwind (next_frame); + unsigned char buf[LINUX_CLONE_LEN]; + + if (!safe_frame_unwind_memory (next_frame, pc - LINUX_CLONE_LEN, buf, + LINUX_CLONE_LEN)) + return 0; + + if (memcmp (buf, linux_clone_code, LINUX_CLONE_LEN) != 0) + return 0; + + return 1; +} + +static int +amd64_linux_outermost_frame (struct frame_info *next_frame) +{ + CORE_ADDR pc = frame_pc_unwind (next_frame); + char *name; + + find_pc_partial_function (pc, &name, NULL, NULL); + + /* If we have NAME, we can optimize the search. + Check the code even in the case of the unambigious `__clone' + as before the `clone' syscall we still have a valid stack there. + FIXME: The whole part after the successful syscall there no longer has the + valid stack but we still try to unwind it, fortunately it applies only for + the case of tracing through `__clone'. */ + if (name == NULL || strcmp (name, "clone") == 0 + || strcmp ("__clone", name) == 0) + return (amd64_linux_clone_running (next_frame) != 0); + + return 0; +} + static void amd64_linux_write_pc (CORE_ADDR pc, ptid_t ptid) { @@ -273,6 +339,8 @@ amd64_linux_init_abi (struct gdbarch_inf tdep->sc_reg_offset = amd64_linux_sc_reg_offset; tdep->sc_num_regs = ARRAY_SIZE (amd64_linux_sc_reg_offset); + tdep->outermost_frame_p = amd64_linux_outermost_frame; + /* GNU/Linux uses SVR4-style shared libraries. */ set_solib_svr4_fetch_link_map_offsets (gdbarch, svr4_lp64_fetch_link_map_offsets); --- ./gdb/amd64-tdep.c 19 Aug 2006 15:15:18 -0000 1.32 +++ ./gdb/amd64-tdep.c 16 Dec 2006 20:15:16 -0000 @@ -879,11 +879,16 @@ amd64_frame_this_id (struct frame_info * { struct amd64_frame_cache *cache = amd64_frame_cache (next_frame, this_cache); + struct gdbarch_tdep *tdep = gdbarch_tdep (current_gdbarch); /* This marks the outermost frame. */ if (cache->base == 0) return; + /* Detect OS dependent outermost frames; such as `clone'. */ + if (tdep->outermost_frame_p && tdep->outermost_frame_p (next_frame)) + return; + (*this_id) = frame_id_build (cache->base + 16, cache->pc); } --- ./gdb/i386-tdep.h 21 Jan 2006 20:59:50 -0000 1.46 +++ ./gdb/i386-tdep.h 16 Dec 2006 20:15:16 -0000 @@ -104,6 +104,9 @@ struct gdbarch_tdep is deprecated, please use `sc_reg_offset' instead. */ int sc_pc_offset; int sc_sp_offset; + + /* Detect OS dependent outermost frames; such as `clone'. */ + int (*outermost_frame_p) (struct frame_info *next_frame); }; /* Floating-point registers. */ --- ./gdb/i386-tdep.c 8 Aug 2006 21:36:46 -0000 1.225 +++ ./gdb/i386-tdep.c 16 Dec 2006 21:13:19 -0000 @@ -2313,6 +2313,9 @@ i386_gdbarch_init (struct gdbarch_info i tdep->sc_pc_offset = -1; tdep->sc_sp_offset = -1; + /* Unwinding stops on i386 automatically. */ + tdep->outermost_frame_p = NULL; + /* The format used for `long double' on almost all i386 targets is the i387 extended floating-point format. In fact, of all targets in the GCC 2.95 tree, only OSF/1 does it different, and insists --- /dev/null 1 Jan 1970 00:00:00 -0000 +++ ./gdb/testsuite/gdb.threads/bt-clone-stop.c 13 Dec 2006 19:06:19 -0000 @@ -0,0 +1,39 @@ +/* This testcase is part of GDB, the GNU debugger. + + Copyright 2006 Free Software Foundation, Inc. + + This program is free software; you can redistribute it and/or modify + it under the terms of the GNU General Public License as published by + the Free Software Foundation; either version 2 of the License, or + (at your option) any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU General Public License for more details. + + You should have received a copy of the GNU General Public License + along with this program; if not, write to the Free Software + Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, + MA 02110-1301, USA. */ + + +#include +#include +#include + + +void *threader(void *arg) +{ + assert(0); + return NULL; +} + +int main() +{ + pthread_t t1; + + pthread_create(&t1,NULL,threader,(void *)NULL); + for (;;) + pause(); +} --- /dev/null 1 Jan 1970 00:00:00 -0000 +++ ./gdb/testsuite/gdb.threads/bt-clone-stop.exp 13 Dec 2006 19:06:19 -0000 @@ -0,0 +1,56 @@ +# Copyright 2006 Free Software Foundation, Inc. + +# This program is free software; you can redistribute it and/or modify +# it under the terms of the GNU General Public License as published by +# the Free Software Foundation; either version 2 of the License, or +# (at your option) any later version. +# +# This program is distributed in the hope that it will be useful, +# but WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +# GNU General Public License for more details. +# +# You should have received a copy of the GNU General Public License +# along with this program; if not, write to the Free Software +# Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. + +if $tracelevel then { + strace $tracelevel +} + +set testfile bt-clone-stop +set srcfile ${testfile}.c +set binfile ${objdir}/${subdir}/${testfile} +if { [gdb_compile_pthreads "${srcdir}/${subdir}/${srcfile}" "${binfile}" executable {debug}] != "" } { + untested "Couldn't compile test program" + return -1 +} + +# Get things started. + +gdb_exit +gdb_start +gdb_reinitialize_dir $srcdir/$subdir +gdb_load ${binfile} + +# threader: threader.c:8: threader: Assertion `0' failed. +# Program received signal SIGABRT, Aborted. + +gdb_test "run" \ + "Program received signal SIGABRT.*" \ + "run" + +# #5 0x0000003421ecd62d in ?? () from /lib64/libc.so.6 +# #6 0x0000000000000000 in ?? () +# (gdb) +# Line `#6' must not be present +# Keep those two cases in this order! +# Fallback is some invalid output (FAIL). +gdb_test_multiple "bt" "0x0 entry output invalid" { + -re "in threader \\(.*\n#\[0-9\]* *0x0* in .*$gdb_prompt $" { + fail "0x0 entry found there" + } + -re "in threader \\(.*$gdb_prompt $" { + pass "0x0 entry not found there" + } +} --MGYHOYXEY6WxJCY8--