From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 28153 invoked by alias); 6 Jun 2010 19:51:25 -0000 Received: (qmail 28140 invoked by uid 22791); 6 Jun 2010 19:51:24 -0000 X-SWARE-Spam-Status: No, hits=-4.1 required=5.0 tests=AWL,BAYES_40,FAKE_REPLY_C,RCVD_IN_DNSWL_HI,SPF_HELO_PASS,T_RP_MATCHES_RCVD X-Spam-Check-By: sourceware.org Received: from mx1.redhat.com (HELO mx1.redhat.com) (209.132.183.28) by sourceware.org (qpsmtpd/0.43rc1) with ESMTP; Sun, 06 Jun 2010 19:51:11 +0000 Received: from int-mx05.intmail.prod.int.phx2.redhat.com (int-mx05.intmail.prod.int.phx2.redhat.com [10.5.11.18]) by mx1.redhat.com (8.13.8/8.13.8) with ESMTP id o56JodTx002662 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=OK); Sun, 6 Jun 2010 15:50:40 -0400 Received: from host0.dyn.jankratochvil.net (ovpn01.gateway.prod.ext.phx2.redhat.com [10.5.9.1]) by int-mx05.intmail.prod.int.phx2.redhat.com (8.13.8/8.13.8) with ESMTP id o56JoaHK013879 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=NO); Sun, 6 Jun 2010 15:50:39 -0400 Received: from host0.dyn.jankratochvil.net (localhost [127.0.0.1]) by host0.dyn.jankratochvil.net (8.14.4/8.14.4) with ESMTP id o56Joa0D025916; Sun, 6 Jun 2010 21:50:36 +0200 Received: (from jkratoch@localhost) by host0.dyn.jankratochvil.net (8.14.4/8.14.4/Submit) id o56JoYla025915; Sun, 6 Jun 2010 21:50:34 +0200 Date: Sun, 06 Jun 2010 19:51:00 -0000 From: Jan Kratochvil To: Eli Zaretskii , Pedro Alves Cc: mark.kettenis@xs4all.nl, brobecker@adacore.com, gdb-patches@sourceware.org Subject: Re: [patch] Forbid run with a core file loaded Message-ID: <20100606195033.GA9710@host0.dyn.jankratochvil.net> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <201005232208.44968.pedro@codesourcery.com> <83mxvqnzny.fsf@gnu.org> User-Agent: Mutt/1.5.20 (2009-12-10) 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: 2010-06/txt/msg00161.txt.bz2 On Sun, 23 May 2010 22:14:41 +0200, Eli Zaretskii wrote: > > +There are multiple classes of targets such as: processes, executable files or > > +recording sessions. Core files belong to the process class, there can be > > +active only one of a core or a running process. > > Something is wrong with the last sentence. Used now: Core files belong to the process class making core file and process mutually exclusive. On Sun, 23 May 2010 23:08:44 +0200, Pedro Alves wrote: > On Sunday 23 May 2010 19:54:40, Jan Kratochvil wrote: > > @@ -194,6 +195,10 @@ inf_ptrace_attach (struct target_ops *ops, char *args, int from_tty) > > + pop_all_targets_above (ops->to_stratum - 1, 0); > > Unfortunately, this would break multiprocess. For example, you may > be attaching to your second process (e.g., "start", "add-inferior", > "inferior 2", "attach $pid"), and the thread stratum needs to remain > pushed for the first inferior/process. OK, fixed it by unpush_target only. I hope you agree the target stack should be made specific for each inferior. Like I proposed to fix target_gdbarch making it specific for each inferior. [01/03] Update target_gdbarch for current inferior http://sourceware.org/ml/gdb-patches/2010-01/msg00228.html > > + /* Clear possible core file with its process_stratum. */ > > + push_target (ops); > > + > > pid = fork_inferior (exec_file, allargs, env, inf_ptrace_me, NULL, > > NULL, NULL); > > > > - push_target (ops); > > This makes sense, as indeed fork_inferior touches the current > target, even today, but, can you check what happens if fork_inferior > throws an error? OK, I have placed there a cleanup. No regressions on {x86_64,x86_64-m32,i686}-fedora13-linux-gnu. OK to check-in? Thanks, Jan gdb/ 2010-05-23 Jan Kratochvil Make core files the process_stratum. * corefile.c (core_target): New variable. (core_file_command): Remove variable t, use core_target. * corelow.c (core_ops): Make it static. (init_core_ops): Change to process_stratum. Initialize CORE_TARGET. * defs.h (make_cleanup_unpush_target): New prototype. * gdbarch.sh (core_pid_to_str): Remove core_stratum from its comment. * gdbarch.h: Regenerate. * gdbcore.h (core_target): New declaration. * inf-ptrace.c (inf_ptrace_create_inferior): New variable back_to. Move push_target before fork_inferior, comment it. (inf_ptrace_attach): Call unpush_target for core_target. * record.c (record_open): Replace core_stratum by a core_bfd check. * target.c (find_core_target): Remove. * target.h (enum strata) : Remove. (find_core_target): Remove declaration. * tracepoint.c (init_tfile_ops) : Remove comment. * utils.c (do_unpush_target, make_cleanup_unpush_target): New functions. gdb/doc/ 2010-05-23 Jan Kratochvil Make core files the process_stratum. * gdb.texinfo (Active Targets): Remove core_stratum. Include record_stratum example. gdb/testsuite/ 2010-05-23 Jan Kratochvil Make core files the process_stratum. * gdb.base/corefile.exp (run: load core again) (run: sanity check we see the core file, run: with core) (run: core file is cleared, attach: load core again) (attach: sanity check we see the core file, attach: with core) (attach: core file is cleared): New tests. * gdb.base/coremaker.c (main): New parameters. Implement "sleep" argv. --- a/gdb/corefile.c +++ b/gdb/corefile.c @@ -59,6 +59,10 @@ static int exec_file_hook_count = 0; /* size of array */ /* Binary file diddling handle for the core file. */ bfd *core_bfd = NULL; + +/* corelow.c target (if included for this gdb target). */ + +struct target_ops *core_target; /* Backward compatability with old way of specifying core files. */ @@ -66,18 +70,15 @@ bfd *core_bfd = NULL; void core_file_command (char *filename, int from_tty) { - struct target_ops *t; - dont_repeat (); /* Either way, seems bogus. */ - t = find_core_target (); - if (t == NULL) + if (core_target == NULL) error (_("GDB can't read core files on this machine.")); if (!filename) - (t->to_detach) (t, filename, from_tty); + (core_target->to_detach) (core_target, filename, from_tty); else - (t->to_open) (filename, from_tty); + (core_target->to_open) (filename, from_tty); } --- a/gdb/corelow.c +++ b/gdb/corelow.c @@ -100,7 +100,7 @@ static void init_core_ops (void); void _initialize_corelow (void); -struct target_ops core_ops; +static struct target_ops core_ops; /* An arbitrary identifier for the core inferior. */ #define CORELOW_PID 1 @@ -911,11 +911,17 @@ init_core_ops (void) core_ops.to_thread_alive = core_thread_alive; core_ops.to_read_description = core_read_description; core_ops.to_pid_to_str = core_pid_to_str; - core_ops.to_stratum = core_stratum; + core_ops.to_stratum = process_stratum; core_ops.to_has_memory = core_has_memory; core_ops.to_has_stack = core_has_stack; core_ops.to_has_registers = core_has_registers; core_ops.to_magic = OPS_MAGIC; + + if (core_target) + internal_error (__FILE__, __LINE__, + _("init_core_ops: core target already exists (\"%s\")."), + core_target->to_longname); + core_target = &core_ops; } void --- a/gdb/defs.h +++ b/gdb/defs.h @@ -348,6 +348,9 @@ extern struct cleanup *make_cleanup_obstack_free (struct obstack *obstack); extern struct cleanup *make_cleanup_restore_integer (int *variable); +struct target_ops; +extern struct cleanup *make_cleanup_unpush_target (struct target_ops *ops); + extern struct cleanup *make_final_cleanup (make_cleanup_ftype *, void *); extern struct cleanup *make_my_cleanup (struct cleanup **, --- a/gdb/gdbarch.sh +++ b/gdb/gdbarch.sh @@ -609,8 +609,7 @@ v:struct core_regset_section *:core_regset_sections:const char *name, int len::: # core file into buffer READBUF with length LEN. M:LONGEST:core_xfer_shared_libraries:gdb_byte *readbuf, ULONGEST offset, LONGEST len:readbuf, offset, len -# How the core_stratum layer converts a PTID from a core file to a -# string. +# How the core target converts a PTID from a core file to a string. M:char *:core_pid_to_str:ptid_t ptid:ptid # BFD target to use when generating a core file. --- a/gdb/gdbarch.h +++ b/gdb/gdbarch.h @@ -666,8 +666,7 @@ typedef LONGEST (gdbarch_core_xfer_shared_libraries_ftype) (struct gdbarch *gdba extern LONGEST gdbarch_core_xfer_shared_libraries (struct gdbarch *gdbarch, gdb_byte *readbuf, ULONGEST offset, LONGEST len); extern void set_gdbarch_core_xfer_shared_libraries (struct gdbarch *gdbarch, gdbarch_core_xfer_shared_libraries_ftype *core_xfer_shared_libraries); -/* How the core_stratum layer converts a PTID from a core file to a - string. */ +/* How the core target converts a PTID from a core file to a string. */ extern int gdbarch_core_pid_to_str_p (struct gdbarch *gdbarch); --- a/gdb/gdbcore.h +++ b/gdb/gdbcore.h @@ -108,6 +108,8 @@ extern void specify_exec_file_hook (void (*hook) (char *filename)); extern bfd *core_bfd; +extern struct target_ops *core_target; + /* Whether to open exec and core files read-only or read-write. */ extern int write_files; --- a/gdb/inf-ptrace.c +++ b/gdb/inf-ptrace.c @@ -119,12 +119,17 @@ inf_ptrace_create_inferior (struct target_ops *ops, char *exec_file, char *allargs, char **env, int from_tty) { + struct cleanup *back_to; int pid; + /* Clear possible core file with its process_stratum. */ + push_target (ops); + back_to = make_cleanup_unpush_target (ops); + pid = fork_inferior (exec_file, allargs, env, inf_ptrace_me, NULL, NULL, NULL); - push_target (ops); + discard_cleanups (back_to); /* On some targets, there must be some explicit synchronization between the parent and child processes after the debugger @@ -194,6 +199,12 @@ inf_ptrace_attach (struct target_ops *ops, char *args, int from_tty) if (pid == getpid ()) /* Trying to masturbate? */ error (_("I refuse to debug myself!")); + /* target_pid_to_str already uses the target. Clear possible core file with + its process_stratum. Targets above must remain in place as the target + stack is shared across multiple inferiors. */ + if (core_target) + unpush_target (core_target); + if (from_tty) { exec_file = get_exec_file (0); --- a/gdb/record.c +++ b/gdb/record.c @@ -958,7 +958,7 @@ record_open (char *name, int from_tty) record_beneath_to_stopped_by_watchpoint = tmp_to_stopped_by_watchpoint; record_beneath_to_stopped_data_address = tmp_to_stopped_data_address; - if (current_target.to_stratum == core_stratum) + if (core_bfd) record_core_open_1 (name, from_tty); else record_open_1 (name, from_tty); --- a/gdb/target.c +++ b/gdb/target.c @@ -2699,31 +2699,6 @@ find_run_target (void) return (count == 1 ? runable : NULL); } -/* Find a single core_stratum target in the list of targets and return it. - If for some reason there is more than one, return NULL. */ - -struct target_ops * -find_core_target (void) -{ - struct target_ops **t; - struct target_ops *runable = NULL; - int count; - - count = 0; - - for (t = target_structs; t < target_structs + target_struct_size; - ++t) - { - if ((*t)->to_stratum == core_stratum) - { - runable = *t; - ++count; - } - } - - return (count == 1 ? runable : NULL); -} - /* * Find the next target down the stack from the specified target. */ --- a/gdb/target.h +++ b/gdb/target.h @@ -65,8 +65,7 @@ enum strata { dummy_stratum, /* The lowest of the low */ file_stratum, /* Executable files, etc */ - core_stratum, /* Core dump files */ - process_stratum, /* Executing processes */ + process_stratum, /* Executing processes or core dump files */ thread_stratum, /* Executing threads */ record_stratum, /* Support record debugging */ arch_stratum /* Architecture overrides */ @@ -1496,8 +1495,6 @@ extern void find_default_create_inferior (struct target_ops *, extern struct target_ops *find_run_target (void); -extern struct target_ops *find_core_target (void); - extern struct target_ops *find_target_beneath (struct target_ops *); /* Read OS data object of type TYPE from the target, and return it in --- a/gdb/tracepoint.c +++ b/gdb/tracepoint.c @@ -4049,8 +4049,6 @@ init_tfile_ops (void) tfile_ops.to_get_trace_status = tfile_get_trace_status; tfile_ops.to_trace_find = tfile_trace_find; tfile_ops.to_get_trace_state_variable_value = tfile_get_trace_state_variable_value; - /* core_stratum might seem more logical, but GDB doesn't like having - more than one core_stratum vector. */ tfile_ops.to_stratum = process_stratum; tfile_ops.to_has_all_memory = tfile_has_all_memory; tfile_ops.to_has_memory = tfile_has_memory; --- a/gdb/utils.c +++ b/gdb/utils.c @@ -352,6 +352,24 @@ make_cleanup_restore_integer (int *variable) xfree); } +/* Helper for make_cleanup_unpush_target. */ + +static void +do_unpush_target (void *arg) +{ + struct target_ops *ops = arg; + + unpush_target (ops); +} + +/* Return a new cleanup that unpushes OPS. */ + +struct cleanup * +make_cleanup_unpush_target (struct target_ops *ops) +{ + return make_my_cleanup (&cleanup_chain, do_unpush_target, ops); +} + struct cleanup * make_my_cleanup2 (struct cleanup **pmy_chain, make_cleanup_ftype *function, void *arg, void (*free_arg) (void *)) --- a/gdb/doc/gdb.texinfo +++ b/gdb/doc/gdb.texinfo @@ -15027,33 +15027,20 @@ and @code{show architecture}. @cindex active targets @cindex multiple targets -There are three classes of targets: processes, core files, and -executable files. @value{GDBN} can work concurrently on up to three -active targets, one in each class. This allows you to (for example) -start a process and inspect its activity without abandoning your work on -a core file. - -For example, if you execute @samp{gdb a.out}, then the executable file -@code{a.out} is the only active target. If you designate a core file as -well---presumably from a prior run that crashed and coredumped---then -@value{GDBN} has two active targets and uses them in tandem, looking -first in the corefile target, then in the executable file, to satisfy -requests for memory addresses. (Typically, these two classes of target -are complementary, since core files contain only a program's -read-write memory---variables and so on---plus machine status, while -executable files contain only the program text and initialized data.) - -When you type @code{run}, your executable file becomes an active process -target as well. When a process target is active, all @value{GDBN} -commands requesting memory addresses refer to that target; addresses in -an active core file or executable file target are obscured while the -process target is active. - -Use the @code{core-file} and @code{exec-file} commands to select a new -core file or executable target (@pxref{Files, ,Commands to Specify -Files}). To specify as a target a process that is already running, use -the @code{attach} command (@pxref{Attach, ,Debugging an Already-running -Process}). +There are multiple classes of targets such as: processes, executable files or +recording sessions. Core files belong to the process class making core file +and process mutually exclusive. Otherwise @value{GDBN} can work concurrently +on multiple active targets, one in each class. This allows you to (for +example) start a process and inspect its activity while still having access to +the executable file after the process finishes. Or if you start process +recording (@pxref{Reverse Execution}) and @code{reverse-step} there you are +presented a virtual layer of the recording target while the process target +remains stopped at the chronologically last point of the process execution. + +Use the @code{core-file} and @code{exec-file} commands to select a new core +file or executable target (@pxref{Files, ,Commands to Specify Files}). To +specify as a target a process that is already running, use the @code{attach} +command (@pxref{Attach, ,Debugging an Already-running Process}). @node Target Commands @section Commands for Managing Targets --- a/gdb/testsuite/gdb.base/corefile.exp +++ b/gdb/testsuite/gdb.base/corefile.exp @@ -177,3 +177,62 @@ gdb_load ${binfile} gdb_test "up" "#\[0-9\]* *\[0-9xa-fH'\]* in .* \\(.*\\).*" "up in corefile.exp (reinit)" gdb_test "core" "No core file now." + + +# Test a run (start) command will clear any loaded core file. + +gdb_test "core-file $corefile" "Core was generated by .*" "run: load core again" +gdb_test "info files" "\r\nLocal core dump file:\r\n.*" "run: sanity check we see the core file" + +set test "run: with core" +if [runto_main] { + pass $test +} else { + fail $test +} + +set test "run: core file is cleared" +gdb_test_multiple "info files" $test { + "\r\nLocal core dump file:\r\n.*\r\n$gdb_prompt $" { + fail $test + } + "\r\n$gdb_prompt $" { + pass $test + } +} + +gdb_exit + + +# Test an attach command will clear any loaded core file. + +if ![is_remote target] { + set test "attach: spawn sleep" + set res [remote_spawn host "$binfile sleep"]; + if { $res < 0 || $res == "" } { + perror "$test failed." + fail $test + return + } + set pid [exp_pid -i $res] + # We do not care of the startup phase where it will be caught. + + gdb_start + + gdb_test "core-file $corefile" "Core was generated by .*" "attach: load core again" + gdb_test "info files" "\r\nLocal core dump file:\r\n.*" "attach: sanity check we see the core file" + + gdb_test "attach $pid" "Attaching to process $pid\r\n.*" "attach: with core" + + set test "attach: core file is cleared" + gdb_test_multiple "info files" $test { + "\r\nLocal core dump file:\r\n.*\r\n$gdb_prompt $" { + fail $test + } + "\r\n$gdb_prompt $" { + pass $test + } + } + + gdb_exit +} --- a/gdb/testsuite/gdb.base/coremaker.c +++ b/gdb/testsuite/gdb.base/coremaker.c @@ -133,8 +133,14 @@ func1 () func2 (); } -int main () +int +main (int argc, char **argv) { + if (argc == 2 && strcmp (argv[1], "sleep") == 0) + { + sleep (60); + return 0; + } mmapdata (); func1 (); return 0;