IF YOU WOULD LIKE TO GET AN ACCOUNT, please write an
email to Administrator. User accounts are meant only to access repo
and report issues and/or generate pull requests.
This is a purpose-specific Git hosting for
BaseALT
projects. Thank you for your understanding!
Только зарегистрированные пользователи имеют доступ к сервису!
Для получения аккаунта, обратитесь к администратору.
tabto is used in many lines of strace output.
On glibc, tprintf("%*s", col - curcol, "") is noticeably slow
compared to tprintf(" "). Use the latter.
Observed ~15% reduction of time spent in userspace.
* defs.h: Drop extern declaration of acolumn. Make tabto()
take no parameters.
* process.c (sys_exit): Call tabto() with no parameters.
* syscall.c (trace_syscall_exiting): Call tabto() with no parameters.
* strace.c: Make acolumn static, add static char *acolumn_spaces.
(main): Allocate acolumn_spaces as a string of spaces.
(printleader): Call tabto() with no parameters.
(tabto): Use simpler method to print lots of spaces.
Signed-off-by: Denys Vlasenko <dvlasenk@redhat.com>
Currently, get_scno does *much* more than "get syscall no".
It checks for post-execve SIGTRAP. It checks for changes
in personality. It retrieves params on entry and registers on exit.
Worse still, it is different in different architectures: for example,
for AVR32 regs are fetched in get_scno(), while for e.g. I386
it is done in syscall_enter().
Another problem is that get_scno() is called on both syscall entry and
syscall exit, which is stupid: we don't need to know scno on syscall
exit, it is already known from last syscall entry and stored in
tcp->scno! In essence, get_scno() does two completely different things
on syscall entry and on exit, they are just mixed into one bottle, like
shampoo and conditioner.
The following patches will try to improve this situation.
This change duplicates get_scno into identical get_scno_on_sysenter,
get_scno_on_sysexit functions. Call them in syscall enter and syscall
exit, correspondingly.
* defs.h: Rename get_scno to get_scno_on_sysenter; declare it only
if USE_PROCFS.
* strace.c (proc_open): Call get_scno_on_sysenter instead of get_scno.
* syscall.c (get_scno): Split into two (so far identical) functions
get_scno_on_sysenter and get_scno_on_sysexit.
(trace_syscall_entering): Call get_scno_on_sysenter instead of get_scno.
(trace_syscall_exiting): Call get_scno_on_sysexit instead of get_scno.
Signed-off-by: Denys Vlasenko <dvlasenk@redhat.com>
* defs.h: Explain TCB_INSYSCALL and TCB_WAITEXECVE bits in detail.
* strace.c (choose_pfd): Use entering/exiting macros instead of direct check
for TCB_INSYSCALL.
* syscall.c (get_scno): Use entering/exiting macros instead of direct check
for TCB_INSYSCALL. Fix comments about post-execve SIGTRAP.
(syscall_fixup): Use entering/exiting instead of direct check
for TCB_INSYSCALL. Add a comment what "not a syscall entry" message
usually means. Change wrong "stray syscall exit" messages into
"not a syscall entry" ones.
Signed-off-by: Denys Vlasenko <dvlasenk@redhat.com>
* strace.c (trace): Optimize out dummy PC printing on signal delivery.
While at it, tweak comments.
Signed-off-by: Denys Vlasenko <dvlasenk@redhat.com>
* defs.h: Make struct tcb::pfd fields conditional on USE_PROCFS.
* strace.c (alloc_tcb): Use tcp->pfd only if USE_PROCFS.
(droptcb): Likewise.
Signed-off-by: Denys Vlasenko <dvlasenk@redhat.com>
* strace.c [LINUX] (kill_save_errno): New function.
(test_ptrace_setoptions_followfork): Change return type to void.
Fix and harden error handling. Use kill_save_errno() to avoid errno
clobbering. Treat EIO from ptrace() the same way as EINVAL.
(test_ptrace_setoptions_for_all): Use kill_save_errno() to avoid errno
clobbering. Treat EIO from ptrace() the same way as EINVAL.
(main): Update use of test_ptrace_setoptions_followfork().
tcp->parent is used for only two things:
(1) to send signal on detach via tgkill (need to know tgid).
Solution: use tkill, it needs only tid.
(2) to optimize out ptrace options setting for new tracees.
Not a big deal if we drop this optimization: "set options" op is fast,
doing it just one extra time once per each tracee is hardly measurable.
TCB_CLONE_THREAD is a misnomer. It used only to flag sibling we attached to
in startup_attach. This is used to prevent infinite recursive rescanning
of /proc/PID/task.
Despite the name, there is no guarantee it is set only on non-leader:
if one would run "strace -f -p THREAD_ID" and THREAD_ID is *not*
a thread leader, strace will happily attach to it and all siblings
and will think that THREAD_ID is the leader! Which is a bug, but
since we no longer detach when we think tracee is going to die,
this bug no longer matters, because we do not use the knowledge
about thread group leaders for anything. (We used it to delay
leader's exit).
IOW: after this patch strace has no need to know about threads, parents
and children, and so on. Therefore it does not track that information.
It treats all tracees as independent entities. Overall,
this simplifies code a lot.
* defs.h: Add TCB_ATTACH_DONE flag, remove TCB_CLONE_THREAD flag
and struct tcb::parent field.
* process.c (internal_fork): Don't set tcpchild->parent.
* strace.c (startup_attach): Use TCB_ATTACH_DONE flag instead of
TCB_CLONE_THREAD to avoid attach attempts on already-attached threads.
Unlike TCB_CLONE_THREAD, TCB_ATTACH_DONE bit is used only temporarily,
and only in this function. We clear it on every tcb before we return.
(detach): Use tkill instead of tgkill.
(trace): Set ptrace options on new tracees unconditionally,
not only when tcp->parent == NULL.
Signed-off-by: Denys Vlasenko <dvlasenk@redhat.com>
Since we no longer suspend waitpid'ing tracees, we have only one case when
we suspend tracee: when we pick up a new tracee created by clone/fork/vfork.
Background: on some other OSes, attach to child is done this way:
get fork's result (pid), loop ptrace(PTRACE_ATTACH) until you hook up
new process/thread. This is ugly and not safe, but what matters for us
is that it doesn't require suspending. Suspending is required
on Linux only, because on Linux attach to child is done differently.
On Linux, we use two methods of catching new tracee:
adding CLONE_THREAD bit to syscall (if needed, we change
[v]fork into clone before that), or using ptrace options.
In both cases, it may be so that new tracee appears before one which
created it returns from syscall. In this case, current code
suspends new tracee until its creator returns. Only then
strace can determine who is its parent (it needs child's pid for this,
which is visible in parent's [v]fork/clone result).
This is inherently racy. For example, what if SIGKILL kills
creator after it succeeded creating child, but before it returns?
Looks like we will have child suspended forever.
But after previous commit, we DO NOT NEED parent<->child link for anything.
Therefore we do not need suspending too. Bingo!
This patch removes suspending code. Now new tracees will be continued
right away. Next patch will remove tcp->parent member.
* defs.h: Remove TCB_SUSPENDED constant
* process.c (handle_new_child): Delete this function.
(internal_fork): Do not call handle_new_child on syscall exit.
* strace.c (handle_ptrace_event): Delete this function.
(trace): Do not suspend new child; remove all handling
of now impossible TCB_SUSPENDED condition.
Signed-off-by: Denys Vlasenko <dvlasenk@redhat.com>
Current code plays some ungodly tricks, trying to not detach
thread group leader until all threads exit.
Also, it detaches from a tracee when signal delivery is detected
which will cause tracee to exit.
This operation is racy (not to mention the determination
whether signal is set to SIG_DFL is a horrible hack):
after we determined that this signal is indeed fatal
but before we detach and let process die,
*other thread* may set a handler to this signal, and
we will leak the process, falsely displaying it as killed!
I need to look in the past to figure out why we even do it.
First guess is that it's a workaround for old kernel bugs:
kernel used to deliver exit notifications to the tracer,
not to real parent. These workarounds are ancient
(internal_exit is from 1995).
The patch deletes the hacks. We no longer need tcp->nclone_threads,
TCB_EXITING and TCB_GROUP_EXITING. We also lose a few rather
ugly functions.
I also added a new message: "+++ exited with EXITCODE +++"
which shows exact moment strace got exit notification.
It is analogous to existing "+++ killed by SIG +++" message.
* defs.h: Delete struct tcb::nclone_threads field,
TCB_EXITING and TCB_GROUP_EXITING constants,
declarations of sigishandled() and internal_exit().
* process.c (internal_exit): Delete this function.
(handle_new_child): Don't ++tcp->nclone_threads.
* signal.c (parse_sigset_t): Delete this function.
(sigishandled): Delete this function.
* strace.c (startup_attach): Don't tcbtab[tcbi]->nclone_threads++.
(droptcb): Don't delay dropping if tcp->nclone_threads > 0,
don't drop parent if its nclone_threads reached 0:
just drop (only) this tcb unconditionally.
(detach): don't drop parent.
(handle_group_exit): Delete this function.
(handle_ptrace_event): Instead of handle_group_exit, just drop tcb;
do not panic if we see WIFEXITED from an attached pid;
print "+++ exited with EXITCODE +++" for every WIFEXITED pid.
* syscall.c (internal_syscall): Do not treat sys_exit specially -
don't call internal_exit on it.
Signed-off-by: Denys Vlasenko <dvlasenk@redhat.com>
* strace.c (trace): Do not recalculate "cflag ? &ru : NULL"
again and again. Do not clear errno unnecessarily.
Consistently check wait errors as pid < 0, not pid == -1.
Indent ifdefs for better readability.
Remove comments after endif if ifdef/endif block is really tiny.
Signed-off-by: Denys Vlasenko <dvlasenk@redhat.com>
* strace.c (trace): Change ifdef LINUX to make a bit more sense,
remove wrong comment at its endif. Slightly optimize
"+++ killed by SIG +++" message for systems without WCOREDUMP macro.
Signed-off-by: Denys Vlasenko <dvlasenk@redhat.com>
* strace.c (alloc_tcb): Print number of allocated tcb's if -d.
(droptcb): Likewise.
(handle_ptrace_event): Remove PTRACE_EVENT_EXEC debug message.
(trace): Improve logging of waitpid: show WIFxxx, exitcode/signal,
ptrace event name, WCOREDUMP - all on one line.
Signed-off-by: Denys Vlasenko <dvlasenk@redhat.com>
TCB_FOLLOWFORK flag seems to be unnecessary, because we either follow
all [v]forks/clones or don't follow any, therefore global variable
followfork is an already existing indicator of what we want to do.
This patch drops all setting/clearing of TCB_FOLLOWFORK bit,
and replaces checks for this bit by checks of followfork value.
In internal_fork, check is moved to in front of if(), since
the check is needed on both "entering" and "exiting" branch.
* defs.h: Remove TCB_FOLLOWFORK define.
* process.c (internal_fork): Do not set/clear TCB_FOLLOWFORK,
test followfork instead of tcp->flags & TCB_FOLLOWFORK.
(handle_new_child): Likewise.
* strace.c (startup_attach): Likewise.
Signed-off-by: Denys Vlasenko <dvlasenk@redhat.com>
Error from strace_fopen in main results in call to exit(1).
Error from strace_fopen in newoutf is propagated to newoutf
callers: startup_attach (where it results in exit(1))
and alloc_tcb (where error is ignored). In second case,
the behavior doesn't seem to be right: it means with -ff
on open error for new LOGFILE.PID the output will continue
to go into *the same file as the previous process* - which
would be confusing. Moreover, on droptcb outf may be closed
and the output of other, still running process outputting
to the same outf will be lost. I don't think this is sane.
IOW: in all cases, error in strace_fopen should be fatal.
* strace.c (strace_fopen): Abort on error instead of returning NULL.
(newoutf): Change return type to void.
(startup_attach): Remove error check on newoutf return value.
(main): Remove error check on strace_fopen return value.
Signed-off-by: Denys Vlasenko <dvlasenk@redhat.com>
set_cloexec_flag() may fail only if we pass it a bad fd,
such as -1 or non-opened one. If we do, we have a bug
in the caller. It makes no sense to try to continue
running when we detect such a blatant bug in our own code.
* strace (set_cloexec_flag): Abort instead of returning error
indicator. Change function to return void.
(strace_fopen): Remove error check on set_cloexec_flag return value.
(proc_open): Likewise.
(proc_poll_open): Likewise.
Signed-off-by: Denys Vlasenko <dvlasenk@redhat.com>
It makes no sense to postpone abort on strace_popen error
unti it returns. Moreover, out-of-memory error was exiting
without any message.
While at it, use 0 as "none" for popen_pid, as optimization.
* strace: Initialize popen_pid to 0 - this puts it in bss.
(trace): Reset popen_pid to 0 instead of -1.
(strace_popen): Never return NULL as error indicator,
abort with good error message instead.
(main): Remove NULL check of strace_popen result.
Signed-off-by: Denys Vlasenko <dvlasenk@redhat.com>
Get rid of fork_tcb() function. It used to do what the comment
above it says, but now it doesn't do much:
it only sets tcp->flags |= TCB_FOLLOWFORK and maybe calls
expand_tcbtab(). The second operation is not necessary, since
alloc_tcp() will do it itself when needed.
This patch deletes fork_tcb(), open-coding tcp->flags |= TCB_FOLLOWFORK
where it was formerly called. It also makes nprocs, tcbtabsize and
expand_tcbtab() static. (While at it, I nuked redundant
extern char **environ declaration: strace.c had *two* of them...)
* defs.h: Remove declarations of nprocs, tcbtabsize and
expand_tcbtab.
* process.c (fork_tcb): Remove this function.
(internal_fork): Open-code fork_tcb.
(handle_new_child): Likewise.
* strace.c: Remove redundant "extern char **environ". Declare
nprocs and tcbtabsize static.
(expand_tcbtab): Make it static.
Signed-off-by: Denys Vlasenko <dvlasenk@redhat.com>
Get rid of a few intermediate variables, simplifies a few expressions,
and uses error_msg_and_die instead of more verbose
fprintf+cleanup+exit sequence.
In alloc_tcp, I use memset to clear entire new tcp.
This not only saves a few bytes of code, but lowers the chances
of future bugs where some data "leaks out" into new tcb's
from old ones because we forgot to re-initialize it.
* strace.c (expand_tcbtab): Simplify this function. No logic changes.
(alloc_tcb): Likewise.
Signed-off-by: Denys Vlasenko <dvlasenk@redhat.com>
Commit 3454e4b463
introduced a bug: sometimes, TRACECLONE/TRACE[V]FORK opts were not set.
The check (tcp->parent == NULL) in old code was meant to check
"if we are not a child created by auto-attach" - in this case,
options need to be set on the child; otherwise they are inherited
and do not need to be set.
I misunderstood the check and if tcp->parent is not NULL, I was
setting only ptrace_setoptions_for_all bits.
This change fixes the problem. Since the fixed logic makes it
unnecessary to keep two sets of options in separate variables,
I merge them back into one variable, ptrace_setoptions.
* defs.h: Merge ptrace_setoptions_followfork and ptrace_setoptions_for_all
into one variable, ptrace_setoptions.
* strace.c: Likewise.
(test_ptrace_setoptions_followfork): Use ptrace_setoptions variable.
(test_ptrace_setoptions_for_all): Likewise.
(main): Likewise.
* process.c (internal_fork): Likewise.
(internal_exec): Likewise.
* strace.c (trace): Fix the bug where different options were set
depending on "tcp->parent == NULL" condition. Add a comment
which makes it more clear why this condition is checked.
Signed-off-by: Denys Vlasenko <dvlasenk@redhat.com>
strace used to suspend waitpid until there is a child
for waitpid'ing process to collect status from.
Apparently, it was done because in some very old kernels
(circa 2002 or even earlier) there were ptrace bugs which
were making waitpid in real parent to not see children.
This kernel bug is fixed long ago. This change removes the workaround.
test/wait_must_be_interruptible.c is a test program which
illustrates why without this change strace changes
programs's behavior.
* defs.h: Delete waitpid and nclone_waiting members from from struct tcb.
Remove declaration of internal_wait().
* process.c (internal_wait): Remove this function.
* strace.c (alloc_tcb): Do not set tcp->nclone_waiting.
(resume): Remove this function.
(resume_from_tcp): Remove this function.
(detach): Do not call resume_from_tcp().
(handle_group_exit): Do not call resume_from_tcp().
* syscall.c (internal_syscall): Do not call internal_wait().
Signed-off-by: Denys Vlasenko <dvlasenk@redhat.com>
We already do it in the normal case, but in rare code path where
tracee is gone (SIGKILLed?) sometimes we were printing this:
"SYSCALL(ARGS <unavailable>)= ? <unavailable>" - note jammed together ")=".
test/sigkill_rain.c can be used to verify the fix.
* strace.c (printleader): add a space after ")" in " <unavailable>)"
Signed-off-by: Denys Vlasenko <dvlasenk@redhat.com>
based on Dmitry's comments.
* defs.h ([p]error_msg[_and_die]): Declare new functions.
* strace.c (SYSCALLTRAP): Rename to syscall_trap_sig.
([p]error_msg[_and_die]): Define new functions.
(strace_tracer_pid): New variable, it controls which pid will
do cleanup on exit via [p]error_msg_and_die.
(main): Set strace_tracer_pid to our initial pid.
(startup_attach): Change strace_tracer_pid if we are in -D mode.
(test_ptrace_setoptions_for_all): Minor changes to logic,
such as better diagnostic messages.
* defs.h (ptrace_setoptions_for_all): Expose this variable.
* strace.c (ptrace_setoptions_for_all): Remove "static".
* process.c (internal_exec): Don't set TCB_WAITEXECVE bit
if we know that post-execve SIGTRAP is not going to happen.
* defs.h (ptrace_setoptions): Variable renamed to ptrace_setoptions_followfork.
* process.c (internal_fork): Ditto.
* strace.c (ptrace_setoptions_for_all): New variable.
(SYSCALLTRAP): New variable.
(error_msg_and_die): New function.
(test_ptrace_setoptions_for_all): New function.
(main): Call test_ptrace_setoptions_for_all() at init.
(handle_ptrace_event): Handle PTRACE_EVENT_EXEC (by ignoring it).
(trace): Check events and set ptrace options without -f too.
Check WSTOPSIG(status) not for SIGTRAP, but for SYSCALLTRAP.
* pathtrace.c: New file, implements matching syscall arguments to
user-specified file paths.
* Makefile.am (strace_SOURCES): Add pathtrace.c.
* defs.h (TCB_FILTERED, filtered): New defines.
(getfdpath, pathtrace_select, pathtrace_match, show_fd_path,
tracing_paths): New declarations.
* strace.c (show_fd_path, tracing_paths): New global variables.
(usage, main): Implement handling of -y and -P options.
* strace.1: Add descriptions of -y and -P options.
* syscall.c (trace_syscall_entering): Add path matching logic to the
print/noprint decision and set the TCB_FILTERED bit appropriately.
(trace_syscall_exiting): Use filtered() macro that checks the
TCB_FILTERED bit to determine print/noprint status.
* util.c (printfd): Use getfdpath().
* strace.c [!USE_PROCFS] (trace): Differentiate output format depending
on PTRACE_GETSIGINFO success or failure. In the former case, use
printsiginfo() to show more details about received signal.
* strace.c [!USE_PROCFS] (trace): Assume that PTRACE_GETSIGINFO is
available. Replace PT_GETSIGINFO with PTRACE_GETSIGINFO. Use
PTRACE_GETSIGINFO for all signals.
* strace.c (test_ptrace_setoptions): Add PTRACE_O_TRACECLONE and
PTRACE_O_TRACEVFORK to PTRACE_SETOPTIONS call, to test exactly
the same set of options that is going to be used later in trace().
* strace.c (main): Call test_ptrace_setoptions() before parsing
-o option, otherwise a forked command will cause a hangup inside
test_ptrace_setoptions().
If PTRACE_O_TRACECLONE et al options are supported by kernel,
use them to do followfork rather than the original setbpt
method that changes registers ourselves.
* defs.h [LINUX] (handle_new_child): New function prototype.
* process.c [LINUX] (handle_new_child): New function based on the
code from internal_fork(), with a trivial change: do reparent only
for sys_clone.
[LINUX] (internal_fork): Use handle_new_child(). Do nothing if
ptrace_setoptions is in effect.
* strace.c [LINUX] (handle_ptrace_event): New function.
[LINUX] (trace): If ptrace_setoptions is in effect, then
call the new function to handle PTRACE_EVENT_* status, and
set PTRACE_SETOPTIONS when we see the initial stop of tracee.
Signed-off-by: Wang Chao <wang.chao@cn.fujitsu.com>
Currently test fork related options only. Fork a child that uses
PTRACE_TRACEME at startup and then does a fork so strace can test
how the PTRACE_SETOPTIONS support works before it handles any real
tracee. Since PTRACE_O_TRACECLONE/*FORK were introduced to kernel
at the same time, this test seems to be enough for these 3 options.
* defs.h [LINUX]: Define PTRACE_O_TRACECLONE et al macros here.
(ptrace_setoptions): New variable declaration.
* strace.c [LINUX] (test_ptrace_setoptions): New function, tests
whether kernel supports PTRACE_O_CLONE/*FORK, the result is stored
in the new variable ptrace_setoptions for later use.
(main): Call test_ptrace_setoptions() if followfork option is set.
Signed-off-by: Wang Chao <wang.chao@cn.fujitsu.com>
* strace.c (pid2tcb): Always match pid. Fail for argument <= 0.
[USE_PROCFS] (first_used_tcb): New function.
[USE_PROCFS] (trace): Use that instead of pid2tcb(0).
* defs.h (struct xlat): Add const qualifier to the field of
type "char *".
(set_sortby, qualify, printnum, printnum_int): Add const qualifier to
arguments of type "char *".
* count.c (set_sortby): Add const qualifier to the argument and
automatic variable of type "char *".
* desc.c (decode_select): Add const qualifier to automatic variables of
type "char *".
* ioctlsort.c (struct ioctlent): Add const qualifier to fields of
type "char *".
(main): Add const qualifier to argv.
* process.c (printargv): Add const qualifier to the argument and
automatic variable of type "char *".
(printargc) Add const qualifier to argument of type "char *".
* signal.c (sprintsigmask, parse_sigset_t): Add const qualifier to
arguments of type "char *".
* strace.c (progname): Add const qualifier.
(detach): Add const qualifier to automatic variable of type "char *".
* stream.c (struct strbuf): Add const qualifier to the field of
type "char *".
* syscall.c (struct qual_options): Add const qualifier to fields of
type "char *".
(qual_syscall, qual_fault, qual_desc, lookup_class): Add const qualifier
to arguments of type "char *".
(qual_signal): Add const qualifier to the argument of type "char *",
avoid modification of constant argument.
(qualify): Likewise.
* util.c (printflags): Add const qualifier to automatic variable of
type "char *".
(printnum, printnum_int): Add const qualifier to arguments of
type "char *".
Remove nclone_detached since CLONE_DETACHED flag was no-op for a very
long time in kernel.
* defs.h (struct tcb): Remove nclone_detached field.
Remove TCB_CLONE_DETACHED flag.
* process.c: Remove CLONE_DETACHED flag.
(clone_flags): Remove CLONE_DETACHED entry.
(internal_fork, internal_wait): Remove code dealing with CLONE_DETACHED
flag and nclone_detached.
* strace.c (startup_attach, alloc_tcb, droptcb, handle_group_exit):
Likewise.
Signed-off-by: Wang Chao <wang.chao@cn.fujitsu.com>
If we use -D and -p option together to trace a multi-thread program, in
addition to the main thread, other threads could not be traced even if we
present -f option. Moreover, when executing 'strace -D -p <non-exist pid>',
strace could not terminate normally.
* strace.c (main): Check it.
Signed-off-by: Wang Chao <wang.chao@cn.fujitsu.com>
* defs.h (cflag_t): New enum.
* strace.1: Document -C option.
* strace.c (cflag): Update type.
(main): Handle -C option.
(trace): Update use of cflag.
* count.c (count_syscall): Move clearing of TCB_INSYSCALL to ...
* syscall.c (trace_syscall): ... here. Update use of cflag.
Based on patch by Adrien Kunysz.
* defs.h (struct tcp): Add curcol.
* strace.c: (alloc_tcb): Initialize it.
(trace): Use curcol from current process and save it before
continuing.
(tprintf): Don't modify curcol on output error.
* configure.ac (AC_CHECK_FUNCS): Add fork.
* strace.c (strace_vforked): Define.
(startup_child): Do not raise SIGSTOP if vforked.
(trace): Skip first exec when starting up after vforked.
* syscall.c [BFIN] (get_scno): Drop waitexec checks.
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
in "collect tasks" loop and in enclosing "collect/process" loop.
Before this patch, "strace cat" + "killall -INT strace" wasn't
making strace stop and exit. This patch restores 4.5.18 behavior
in this regard.
which do proper error-checking and set tcp->ptrace_errno.
In some cases, missing error checking is added.
Error handling for trace_syscall() failures and other cases
where tcp->ptrace_errno is nonzero is cleaned up a bit
and made more verbose if we see error other than ESRC.
Some comments are added or expanded.
* defs.h: Declare ptrace_cmds[]. Modify do_ptrace
declaration (last parameter is long, not void *).
* process.c: Make ptrace_cmds[] non-static.
(change_syscall): Use do_ptrace() instead of bare ptrace().
* signal.c: Use do_ptrace() instead of bare ptrace().
* strace.c: Update trace_syscall() failure handling.
* syscall.c: Use do_ptrace() instead of bare ptrace().
* util.c: Use do_ptrace() instead of bare ptrace().
Update do_ptrace() wrapper.
(str_PTRACE_xxx): New function - helper returning "PTRACE_xxx".
(do_ptrace_peekdata): New function - wrapper for PTRACE_PEEKDATA
(do_ptrace5): New function - wrapper for 5-argument ptrace calls.
There was code which was trying to continue tracing
even if table expansion fails. Now we treat it as fatal
failure, so this code is removed by this change.
* defs.h: Delete TCB_FOLLOWFORK constant.
* process.c: Delete fork_tcb() and all calls of it.
* strace.c (startup_attach): Remove usage of TCB_FOLLOWFORK.
* syscall.c: Indent preprocessor directives.
* strace.c: Fix misplaced #endif.
* util.c: Indent preprocessor directives, mark code parts
which can never be reached by compilation because of
the combination of #if directives. These are likely dead code,
I want to remove them later.
(startup_attach): Fix wrong pid in "Process <PID> attached".
(handle_group_exit): Do not consider exit to be spurious if
tcb has TCB_STARTUP bit set - we can attach to the task
right before its death, it can legitimately happen.
(handle_stopped_tcbs): Ditto.
looks insane.
* strace.c (alloc_tcb): Clear *all* fields in reused tcb.
(main): Query and remember uname() info on startup.
(handle_stopped_tcbs): Do not use PTRACE_SETOPTIONS on Linux < 2.6.29.
(printleader): Correct printing of "<unavailable>" markers.
event, we do not have any viable way to continue. No wonder most
places where that is detected have FIXMEs.
It's way simpler to treat as fatal failure, and handle it inside
tcb table expansion finctions.
Second cleanup: tidy up haphazard locations of a few externs.
* defs.h: Change return type of expand_tcbtab() to void.
Declare change_syscall().
* process.c: Change all callsites of alloctcb(), alloc_tcb() and
fork_tcb(), removing now-redundant error checks.
(fork_tcb): Change return type to void - it can't fail now.
* strace.c: Move extern declarations out of function bodies.
Change all callsites of alloctcb(), alloc_tcb() and
fork_tcb(), removing now-redundant error checks.
(expand_tcbtab): Change return type to void - it can't fail now.
On failure to expand, print a message, clean up, and exit.
(alloc_tcb): On failure to expand, print a message, clean up, and exit.
* util.c (setbpt): Remove extern declaration from function body.
* strace.c (handle_stopped_tcbs): Discard all execve stops
and clear TCB_WAITEXECVE bit.
* syscall.c (get_scno): Add the code to not mistakenly
treat ptrace stop as execve stop (execve stops can be blocked
by traced program).
Fixes RH#477775 "strace hangs if the target process blocks SIGTRAP".
* strace.c (collect_stopped_tcbs): Stop reversing list of stopped
tcp's. I'm not totally convinced it is crucial, but this is surely
fits the concept of "least surprise".
Do not collect TCB_SUSPENDED tcp's (this is closer to how
it was before).
(handle_stopped_tcbs): Remove the code to reject TCB_SUSPENDED tcp's,
it's done earlier now. In an unobvious way, this was causing
SIGSTOPs from freshly attached children to be misinterpreted.
make flags field wider (ints are easier to work with on many CPUs).
* strace.c (trace): Split this function into two:
collect_stopped_tcbs() and handle_stopped_tcbs().
Now we collect *all* waitable tasks, then handle them all,
then repeat.
Fixes RH#478419 "Some threads stop when strace with -f option
is executed on a multi-thread process"
* test/many_looping_threads.c: example program which cna't be straced
successfully without this fix.
change which added better handling of processes suddenly
disappearing. Now we often do not finish last line
before exiting in those cases.
The only change affecting something other than output
is change in umovestr where we were calling
abort() on ptrace error.
* strace.c (trace): If trace_syscall() failed with ESRCH,
finish current output line with " <unfinished ...>".
(mp_ioctl): While we are at it, fix gross style mismatch
in this function definition. No code chages.
* syscall.c (trace_syscall): If decode fails on syscall exit,
finish current output line with "= ? <unavailable>".
* util.c (umoven): Do not complain if error is ESRCH.
(umovestr): Do not complain and do not abort() if error is ESRCH.
* defs.h: Remove unused tcp parameter from printtrailer().
* process.c: Adjust printtrailer() calls accordingly.
* strace.c: Adjust printtrailer() calls accordingly.
* syscall.c: Adjust printtrailer() calls accordingly.
Unlike normal case, with -D *grandparent* process exec's,
becoming a traced process. Child exits (this prevents traced process
from having children it doesn't expect to have), and grandchild
attaches to grandparent similarly to strace -p PID.
This allows for more transparent interaction in cases
when process and its parent are communicating via signals,
wait() etc. Without -D, strace process gets lodged in between,
disrupting parent<->child link.
* strace.c: Add global flag variable daemonized_tracer for -D option.
(startup_attach): If -D, fork and block parent in pause().
In this case we are already a child, we in fact created a grandchild.
After attaching to grandparent, grandchild SIGKILLs parent.
(startup_child): If -D, parent blocks in wait(), then
execs the program to strace. Normally (w/o -D), it is child
who execs the program.
(main): Detect -D option, call startup_attach() if it is given.
* strace.c: Add ptrace_stop_sig static variable
and use it in place of tcp->sigtrap80.
Add ptrace_opts_set static flag variable.
(trace): Set ptrace options once, not per-process.
If unexpected SIGTRAP is later received,
revert back to using SIGTRAP
(assume old, broken kernel).
kill(2) and by trapping instruction.
* defs.h: Add sigtrap80 field to struct tcb.
* strace.c (alloc_tcb): Initialize it to SIGTRAP.
(detach): Use tcp->sigtrap80 instead of SIGTRAP constant.
(trace): Attempt to set PTRACE_O_TRACESYSGOOD and
PTRACE_O_TRACEEXEC options on each newly attached process,
distinquish between SIGTRAP and (SIGTRAP | 0x80) stops.
Fixes RH#162774 "strace ignores int3 SIGTRAP".
It merely passes tcp pointer to upeek instead of pid.
This is needed if one wants to check or change
some tcp fields.
I have patches which require this. I can imagine someone else
eventually needing to look at tcp for completely orthogonal reasons.
* strace.c (startup_child): Save child pid for future use.
(main): Exit/kill ourself with straced child's exitcode/signal.
(trace): If signalled process pid matches the saved child pid,
save the signal number. If terminated process pid matches the
saved child pid, save its exit status.
Patch from Denys Vlasenko <dvlasenk@redhat.com>
Fix compiler warnings.
* signal.c (sys_signal): Cast to SIG_* to the matching type LONG.
* strace.c (trace): Variables PSR and PC are now signed.
* syscall.c (syscall_enter): Variable RBS_END is now signed long.
Remove/add the RBS_END casts appropriately.
* util.c [IA64] (arg_setup): Variable BSP is now signed long.
Remove/add the BSP casts appropriately.
<ia32>: Initialize *STATE.
Trace even VFORK on -f on all the platforms.
* strace.1 <-f>: Describe Linux catches new children immediately.
<-F>: Make the option obsolete. Move the SunOS VFORK comment to the
`-f' option description.
* defs.h (followvfork): Declaration removed.
* strace.c (followvfork): Variable removed.
(main) <-F>: Merge with the `-f' option.
(trace): Make !FOLLOWVFORK unconditional.
* process.c (internal_fork): Make !FOLLOWVFORK unconditional.
* strace.c [LINUX] (droptcb): Recurse on TCP->parent if it is a
TCB_EXITING zombie group leader.
* test/childthread.c: New file.
* test/.cvsignore, test/Makefile: Add it.
Code advisory: Roland McGrath
Fixes RH#354261.
* strace.c (detach): Moved the resume notification code to ...
(resume_from_tcp): ... a new function here.
(handle_group_exit): No longer detach also the thread group leader.
(trace): Fixed panic on exit of the TCB_GROUP_EXITING leader itself.
* strace.c (detach): New prototype. Extended the function comment.
[LINUX] (detach): Call droptcb() instead of the wrongly parametrized
detach() call.
(handle_group_exit): Call droptcb() instead of the wrongly parametrized
detach() call. Always call detach() only once from the group leader.
Comment the leader killing known bug tested by `test/leaderkill.c'.
Code advisory: Roland McGrath
Fixes RH#240961.
Never interrupt when the attached traced process would be left stopped.
* strace.c (main): `-p' attaching moved to ...
(startup_attach): ... a new function, renamed a variable C to TCBI.
Block interrupting signals since the first tracee has been attached.
New comment about INTERRUPTED in the nonthreaded case.
[LINUX] (startup_attach): Check INTERRUPTED after each attached thread.
(main): Command spawning moved to ...
(startup_child): ... a new function, replaced RETURN with EXIT.
[LINUX] (detach): New variable CATCH_SIGSTOP, do not signal
new SIGSTOP for processes still in TCB_STARTUP.
(main): Move signals and BLOCKED_SET init before the tracees attaching,
[SUNOS4] (trace): Removed fixvfork () call as a dead code, SIGSTOP must
have been already caught before clearing TCB_STARTUP.
(trace): Removed the `!WIFSTOPPED(status)' dead code.
Clear TCB_STARTUP only in the case the received signal was SIGSTOP.
New comment when `TCB_BPTSET && TCB_STARTUP' combination can be set.
Code advisory: Roland McGrath
Fixes RH#240986.
Make several global variables static.
#ifdef definitions of rarely unused functions.
* defs.h (rflag, tflag, outfname): Remove.
* strace.c (iflag, interactive, pflag_seen, rflag, tflag,
outfname, username): Make static.
* desc.c (sys_getdtablesize): Define only for
ALPHA || FREEBSD || SUNOS4.
* file.c (sys_fchroot): Define only for SUNOS4 || SVR4.
(sys_mkfifo): Define only for FREEBSD.
* mem.c (sys_sbrk): Define only for FREEBSD || SUNOS4.
(sys_getpagesize): Define only for
ALPHA || FREEBSD || IA64 || SUNOS4 || SVR4.
* net.c (sys_so_socket): Define only for SVR4.
* process.c (sys_gethostid): Define only for
FREEBSD || SUNOS4 || SVR4.
(sys_gethostname): Define only for
ALPHA || FREEBSD || SUNOS4 || SVR4.
(sys_setpgrp): Define only for ALPHA || SUNOS4 || SVR4.
(sys_execv): Define only for SPARC || SPARC64 || SUNOS4.
* signal.c (sys_sigblock): Define only for FREEBSD || SUNOS4.
(sys_sighold, sys_sigwait): Define only for SVR4.
(sys_killpg): Define only for FREEBSD || SUNOS4.
* stream.c (sys_getmsg): Define only for
SPARC || SPARC64 || SUNOS4 || SVR4.
* syscall.c (sys_indir): Define only for SUNOS4.
* strace.c [!USE_PROCFS] (trace): Presence of PT_GETSIGINFO
macro does not mean that PT_CR_IPSR and PT_CR_IIP macros are
also defined, so check them along with PT_GETSIGINFO.
Fixes RH#209856.
Fix race conditions in tcb allocation.
* process.c (fork_tcb): Return error code as documented. Do not
print "tcb table full" error message.
[USE_PROCFS] (internal_fork): Do not print "tcb table full"
error message.
[SYS_clone || SYS_clone2] (internal_clone, internal_fork): Call
fork_tcb() before alloctcb(). Do not print "tcb table full"
error message.
* strace.c (main): Do not print "tcb table full" error message.
(expand_tcbtab): Print error message in case of memory allocation
failure.
(alloctcb): Print error message when tcb table is full.
(trace): Expand tcb table if necessary prior to allocating
entry there. Do not print "tcb table full" error message.
Fixes RH#180293.
Implement qual_flags support for each personality.
* strace.c (main): Move qualify calls after set_personality call.
* syscall.c (qual_flags0): New variable..
[SUPPORTED_PERSONALITIES >= 2] (qual_flags1): New variable.
[SUPPORTED_PERSONALITIES >= 3] (qual_flags2): New variable.
(qual_flags): Change variable definition from array to pointer.
(set_personality): Initialize qual_flags variable.
(qualify_one): Add "pers" argument to specify personality.
[SUPPORTED_PERSONALITIES >= 2]: Set qual_flags1 if requested.
[SUPPORTED_PERSONALITIES >= 3]: Set qual_flags2 if requested.
(qual_syscall): Pass personality to qualify_one.
[SUPPORTED_PERSONALITIES >= 2]: Look for syscall also in sysent1
table.
[SUPPORTED_PERSONALITIES >= 3]: Look for syscall also in sysent2
table.
(qual_signal): Pass personality to qualify_one.
(qual_desc): Likewise.
(qualify): Use qualify_one instead of manual qual_flags
manipulations.
[SUPPORTED_PERSONALITIES >= 2]: Look for syscall also in sysent1
table.
[SUPPORTED_PERSONALITIES >= 3]: Look for syscall also in sysent2
table.
* defs.h: Update qual_flags declaration.
Fixes RH#173986.
* strace.c (detach): If detaching the last live thread in a group with
a zombie leader, then detach the leader too.
(handle_group_exit): Use detach, not droptcb, for predeceased thread.
Mark process about to take a signal with TCB_GROUP_EXITING flag.
Fixes RH#161919.
* strace.c (handle_group_exit): Don't detach leader that wasn't
TCB_ATTACHED. Instead mark it with TCB_GROUP_EXITING.
Remove droptcb loop at end, no longer required since 2.6 reports each
thread death.
Fixes RH#135254.
* strace.c (trace): Use handle_group_exit for non-TCB_ATTACHED child
taking signal when it has nclone_threads > 0.
* strace.c (handle_group_exit): Don't detach leader that wasn't
TCB_ATTACHED.
* strace.c (handle_group_exit, trace): Mark leader with
TCB_GROUP_EXITING and don't be surprised at child deaths when their
leader has it set.
Fixes RH#132150.
* strace.c (main) [LINUX]: Expand TCBTAB as necessary for threads
attached. Attach threads only under -f. Set TCB_FOLLOWFORK in them.
(expand_tcbtab): New function, broken out of ...
* process.c (fork_tcb): ... here, call that.
* defs.h: Declare expand_tcbtab.
* strace.c (main): Treat piped output more like file output.
Disallow -ff with piped output. Fixes RH#105366.
Reported by Dmitry V. Levin <ldv@altlinux.org>
* strace.c (main): Set -q when given -o and not -p, and not when not
given -o, to match what the man page always said.
Fixes Debian bug #47113, #153678.
* defs.h (struct tcb): New member `nzombies'.
* strace.c (alloctcb): Initialize it.
(droptcb): Increment our parent's zombie count.
* process.c (internal_wait): Don't go into TCB_SUSPENDED if the
process has zombies it can reap. On the way out, if we reaped
an untraced process, decrement the zombie count.
* strace.c (usage): Omit -z, since it has never worked properly.
* NEWS: Likewise.
* strace.c (main): Grok new option `-E var=val' or `-E var' to put
var=val in environ or to remove var, respectively.
(usage): Mention it.
* strace.1, NEWS: Document it.
Support for new Linux 2.5 thread features.
* defs.h [LINUX]: Define __NR_exit_group if not defined.
(struct tcb): New members nclone_threads, nclone_detached,
and nclone_waiting.
(TCB_CLONE_DETACHED, TCB_CLONE_THREAD, TCB_GROUP_EXITING): New macros.
(waiting_parent): Macro removed.
(pid2tcb): Declare it.
* process.c (internal_clone) [TCB_CLONE_THREAD]: Reparent the new
child to our parent if we are a CLONE_THREAD child ourselves.
Maintain TCB_CLONE_THREAD and TCB_CLONE_DETACHED flags and counts.
(internal_wait) [TCB_CLONE_THREAD]: Factor out detached children when
determining if we have any. If TCB_CLONE_THREAD is set, check
parent's children instead of our own, and bump nclone_waiting count.
(internal_exit) [__NR_exit_group]: Set the TCB_GROUP_EXITING flag if
the syscall was exit_group.
* syscall.c (internal_syscall): Use internal_exit for exit_group.
* strace.c (pid2tcb): No longer static.
(alloctcb) [TCB_CLONE_THREAD]: Initialize new fields.
(droptcb) [TCB_CLONE_THREAD]: Maintain new fields.
If we have thread children, set TCB_EXITING and don't clear the TCB.
(resume) [TCB_CLONE_THREAD]: Decrement parent's nclone_waiting.
(detach) [TCB_CLONE_THREAD]: When calling resume, check all thread
children of our parent that might be waiting for us too.
[TCB_GROUP_EXITING] (handle_group_exit): New function.
(trace) [TCB_GROUP_EXITING]: Use that in place of detach or droptcb.
Revamp -f support for Linux.
* util.c [LINUX] (setbpt, clearbpt): New implementations that tweak
the system call to be clone with CLONE_PTRACE set. Various new static
helper functions.
* process.c (internal_clone): Define also #ifdef SYS_clone2.
Initialize TCPCHILD->parent field.
[CLONE_PTRACE]: Don't do PTRACE_ATTACH here, because it's preattached.
Check in case the new child is in the tcb already.
(internal_fork) [LINUX]: Just call internal_clone.
* strace.c (trace) [LINUX]: Under -f/-F, grok an unknown pid
reporting to wait, put it in the TCB with TCB_ATTACHED|TCB_SUSPENDED.
* strace.c (tcbtab): Make this a pointer to pointers, not an array.
(tcbtabsize): New variable.
(main): Initialize them using dynamic allocation.
(alloctcb, main): Use tcbtabsize in place of MAX_PROCS; indirect.
(pid2tcb, cleanup): Likewise.
[USE_PROCFS] (pollv): Make this a pointer, not an array; make static.
(rebuild_pollv): Dynamically allocate the vector.
* defs.h (tcbtab): Update decls.
(MAX_PROCS): Macro removed, no more static limit on this.
* process.c (fork_tcb): New function.
(internal_clone, internal_fork): Use it instead of checking nprocs.
Another bunch of patches from John Hughes merged:
signal.c:
+ SVR4 printcontext(): sigset_t != sigset_t*
+ getcontext returns a value, so print on exit of syscall
+ add UC_FP to ucontext_flags for OS writers that can't spell
+ sys_signal(): special case SIG_{ERR,DFL,IGN}
+ decode_subcall(): only do subcall range checking when needed
bunch of UnixWare updates
aclocal.m4, acconfig.h, configure.in: add test for long long type
test/.cvsignore: new file
defs.h: Up maximum number of traced processed to 64
strace.c: Disable some debugging code from davidm
implement setarg for more architectures
implement change_syscall
