Denys Vlasenko [Wed, 8 Jun 2011 23:43:22 +0000 (01:43 +0200)]
Don't display bogus parameter for sigreturn syscall
* linux/*/syscallent.h: For those arches which use sys_sigreturn,
not printargs, to show [rt_]sigreturn syscall, change number of arguments
from 1 to 0: sys_sigreturn function doesn't use syscall parameters.
(I guess kernel doesn't actually _have_ any parameters for this syscall,
at least on these architectures). Do the same change for I386 and x86-64
even though they use printargs: I looked at kernel code and syscall
definitely doesn't have any parameters on these arches.
(I hesitate to change 1 to 0 params for arches I don't know -
it is remotely possible some of them do have a parameter for this syscall).
Denys Vlasenko [Wed, 8 Jun 2011 23:36:29 +0000 (01:36 +0200)]
Optimize sigreturn handling
* signal.c (sys_sigreturn): move stack pointer variables,
and for SPARC and MIPS, stack pointer and sigmask reading code
into "if (entering) ..." block, because it is only needed
in this branch; load tcp->u_arg[1] into sigmask for display
_after_ we know for sure u_arg[1] does contain valid sigmask
(IOW: perform operation only when we know we will need the result)
Denys Vlasenko [Wed, 8 Jun 2011 23:28:11 +0000 (01:28 +0200)]
"Modernize" four old-style function parameter declarations
* signal.c (signame, long_to_sigset, printsigmask, printsignal):
Convert old-style C function definitions to a "modern" form.
This does not change any actual code.
Denys Vlasenko [Wed, 8 Jun 2011 23:22:10 +0000 (01:22 +0200)]
Fix sigreturn decoding on MIPS
The "return 0" line was accidentally deleted circa 2007,
which made sigreturn on MIPS always display "= 0" return
instead of more informative " = ? (mask now [MASK])".
* strace.c (sys_sigreturn): Add wrongly deleted "return 0" line
Denys Vlasenko [Wed, 8 Jun 2011 14:15:04 +0000 (16:15 +0200)]
Print at least one space between SYSCALL(ARGS) and = RESULT if tracee is killed
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>)"
Denys Vlasenko [Wed, 8 Jun 2011 12:08:59 +0000 (14:08 +0200)]
Update test/* directory, it seem to be a bit bit-rotted
Added README; modified sigkill_rain.c to be more understandable,
made clone.c compile; added wait_must_be_interruptible.c test;
updated Makefile and .gitignore.
Dmitry V. Levin [Sat, 28 May 2011 20:47:43 +0000 (20:47 +0000)]
ARM EABI: fix 64-bit syscall's arguments decoding
ARM OABI and ARM EABI have different function parameters passing rules.
With EABI, 64-bit function parameters passed in registers are aligned to
an even-numbered register instead of using the next available pair, see
http://lkml.org/lkml/2006/1/12/175
This rule also applies to syscall's arguments.
* linux/arm/syscallent.h (pread, pwrite, truncate64, ftruncate64,
readahead, preadv, pwritev): Fix number of arguments.
* util.c (printllval): Align 64bit argument to 64bit boundary on
__ARM_EABI__.
Reported-by: Damir Shayhutdinov <damir@altlinux.org> Signed-off-by: Kirill A. Shutemov <kirill@shutemov.name> Signed-off-by: Dmitry V. Levin <ldv@altlinux.org>
Denys Vlasenko [Fri, 27 May 2011 12:36:01 +0000 (14:36 +0200)]
Cleanups on top of "handle SIGTRAP properly" change,
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.
Denys Vlasenko [Tue, 24 May 2011 18:30:24 +0000 (20:30 +0200)]
Don't perform TCB_WAITEXECVE wait if not needed.
* 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.
Denys Vlasenko [Mon, 23 May 2011 19:29:03 +0000 (21:29 +0200)]
Properly handle real SIGTRAPs.
* 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.
Grant Edwards [Thu, 7 Apr 2011 20:25:40 +0000 (20:25 +0000)]
Add ability to print file descriptor paths and filter by those paths
* 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().
Mike Frysinger [Mon, 21 Feb 2011 04:24:22 +0000 (23:24 -0500)]
linux/ioctlent: unify them all
This unifies all the ioctlent.h's in the linux subdir while still
allowing each arch to maintain its own minor list.
The basic method is:
- each arch has linux/<arch>/ioctlent.h.in which defines only the
arch-specific ioctls;
- linux/ioctlent.h.in which defines only the common ioctls;
- at build time, these two headers are combined and sorted to produce
the linux/ioctlent.h file.
This also requires a little tweaking of the include files since the
common ioctlent.h is a built file.
* linux/ioctlent.h: Split into linux/ioctlent.h.in and
linux/i386/ioctlent.h.in, remove asm entries from the former, remove
non-asm entries from the latter.
* linux/alpha/ioctlent.h: Rename to linux/alpha/ioctlent.h.in, remove
non-asm entries.
* linux/bfin/ioctlent.h: Rename to linux/bfin/ioctlent.h.in, remove
non-asm entries.
* linux/hppa/ioctlent.h: Rename to linux/hppa/ioctlent.h.in, remove
non-asm entries.
* linux/ia64/ioctlent.h: Rename to linux/ia64/ioctlent.h.in, remove
non-asm entries.
* linux/mips/ioctlent.h: Rename to linux/mips/ioctlent.h.in, remove
non-asm entries.
* linux/powerpc/ioctlent.h: Rename to linux/powerpc/ioctlent.h.in,
remove non-asm entries.
* linux/s390/ioctlent.h: Rename to linux/s390/ioctlent.h.in, remove
non-asm entries.
* linux/sh/ioctlent.h: Rename to linux/sh/ioctlent.h.in, remove
non-asm entries.
* linux/sparc/ioctlent.h: Rename to linux/sparc/ioctlent.h.in, remove
non-asm entries.
* linux/arm/ioctlent.h.in: New file.
* linux/avr32/ioctlent.h.in: Likewise.
* linux/i386/ioctlent.h.in: Likewise.
* linux/m68k/ioctlent.h.in: Likewise.
* linux/microblaze/ioctlent.h.in: Likewise.
* linux/tile/ioctlent.h.in: Likewise.
* linux/x86_64/ioctlent.h.in: Likewise.
* linux/s390x/ioctlent.h.in: Include ioctlent.h.in instead of
ioctlent.h.
* linux/sh64/ioctlent.h.in: Likewise.
* linux/sparc64/ioctlent.h.in: Likewise.
* linux/arm/ioctlent1.h: Update ioctlent.h include.
* linux/powerpc/ioctlent1.h: Likewise.
* linux/sparc/ioctlent1.h: Likewise.
* linux/sparc64/ioctlent1.h: Likewise.
* linux/x86_64/ioctlent1.h: Likewise.
* Makefile.am (AM_CPPFLAGS): Add -I$(builddir)/$(OS).
(EXTRA_DIST): Update.
[MAINTAINER_MODE && LINUX]: Convert from ioctlent_h to ioctlent_h_in.
[LINUX]: Add $(builddir)/$(OS)/ioctlent.h generation rules.
* .gitignore: Add linux/ioctlent.h.
Dmitry V. Levin [Thu, 10 Mar 2011 21:20:35 +0000 (21:20 +0000)]
Show more details about signals received by traced processess
* 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.
Dmitry V. Levin [Thu, 10 Mar 2011 14:44:45 +0000 (14:44 +0000)]
Get rid of PT_GETSIGINFO
* strace.c [!USE_PROCFS] (trace): Assume that PTRACE_GETSIGINFO is
available. Replace PT_GETSIGINFO with PTRACE_GETSIGINFO. Use
PTRACE_GETSIGINFO for all signals.
Dmitry V. Levin [Thu, 3 Mar 2011 01:02:41 +0000 (01:02 +0000)]
Fix struct xlat initialization bugs
* file.c (inotify_modes): Terminate with NULL entry.
* net.c (sock_type_flags): Make this array static.
(socketlayers): Add a comment that this array should remain not
NULL-terminated.
Dmitry V. Levin [Thu, 3 Mar 2011 00:10:20 +0000 (00:10 +0000)]
tests: avoid SIGPIPE
* tests/ptrace_setoptions: Replace "grep -q" with "grep > /dev/null".
The former may result to strace being killed by SIGPIPE, which in
certain configuratons may lead to generation of a core file.
Suggested by Mike Frysinger.
Dmitry V. Levin [Sat, 26 Feb 2011 14:32:12 +0000 (14:32 +0000)]
Remove redundant ioctlent.h files
* linux/s390x/ioctlent.h: Replace old contents with include of
s390/ioctlent.h file.
* linux/sparc64/ioctlent.h: Replace old contents with include of
sparc/ioctlent.h file.
Dmitry V. Levin [Thu, 24 Feb 2011 01:51:15 +0000 (01:51 +0000)]
Generate much of the CREDITS file from git log
* CREDITS.in: New file, derived from CREDITS, without names of
those who are listed as git log 'Author:'s.
* CREDITS: Remove file.
* Makefile.am [MAINTAINER_MODE] (CREDITS): New rule.
* .gitignore: Add CREDITS.
* .mailmap: New file, required to map git author names and email
addresses to canonical/preferred form.
Dmitry V. Levin [Tue, 18 Jan 2011 17:36:20 +0000 (17:36 +0000)]
Fix decoding of get[ug]id, gete[ug]id and setfs[ug]id return values
* defs.h (SYSCALL_NEVER_FAILS): New syscall flag.
* linux/dummy.h: Change redirection for sys_get[ug]id, sys_gete[ug]id
and setfs[ug]id.
* linux/*/syscallent.h: Set SYSCALL_NEVER_FAILS flag for get[ug]id,
gete[ug]id and setfs[ug]id syscalls.
* process.c [LINUX] (sys_getuid, sys_setfsuid): New functions.
* syscall.c (NF): New shorthand macro for use in syscallent.h files.
(get_error): Check SYSCALL_NEVER_FAILS flag.
Reported by Марк Коренберг <socketpair@gmail.com>.
Dmitry V. Levin [Sun, 16 Jan 2011 23:07:51 +0000 (23:07 +0000)]
block.c: cleanup
* block.c: Include <inttypes.h>.
(print_blkpg_req): Always decode struct blkpg_ioctl_arg.
Robustify decoding of strings.
(block_ioctl): Do not decode return values passed by pointers on exit
from failed syscalls.
Use format macros from inttypes.h to print values of type uint64_t.
Dmitry V. Levin [Sat, 15 Jan 2011 20:15:31 +0000 (20:15 +0000)]
Add block ioctl support
* block.c: New file.
* Makefile.am (strace_SOURCES): Add it.
* defs.h [LINUX] (block_ioctl): New function.
* ioctl.c (ioctl_decode) [LINUX]: Use it to decode HDIO_* and BLK*
ioctls.
Patch by Jeff Mahoney <jeffm@suse.com>
Dmitry V. Levin [Thu, 13 Jan 2011 14:49:39 +0000 (14:49 +0000)]
Test PTRACE_O_TRACECLONE and PTRACE_O_TRACEVFORK along with PTRACE_O_TRACEFORK
* 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().
Carmelo AMOROSO [Tue, 4 Jan 2011 15:30:40 +0000 (16:30 +0100)]
sh: Add entry for not-multiplexed accept4
* linux/sh/syscallent.h: Add specific entry for not-multiplexed accept4
available in kernel mainline since v2.6.37-rc6, see
http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commit;h=21b6e4c7106b2d68a6710506d8706608272fd78b
Carmelo AMOROSO [Mon, 13 Dec 2010 07:48:28 +0000 (08:48 +0100)]
sh: Fix compilation warning in do_pipe due to missing prototype
* defs.h [SH]: Make getrval2 prototype visible to do_pipe
and fix the following compiler warning:
.../net.c: In function 'do_pipe':
.../net.c:1632: warning: implicit declaration of function 'getrval2'
.../net.c:1632: warning: format '%lu' expects type 'long unsigned int',
but argument 3 has type 'int'
Dmitry V. Levin [Tue, 7 Dec 2010 12:50:49 +0000 (12:50 +0000)]
Fix strace -f -o '|command' hangup
* strace.c (main): Call test_ptrace_setoptions() before parsing
-o option, otherwise a forked command will cause a hangup inside
test_ptrace_setoptions().
Wang Chao [Fri, 12 Nov 2010 09:26:08 +0000 (17:26 +0800)]
Handle followfork using ptrace_setoptions if available
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>
Wang Chao [Fri, 12 Nov 2010 09:25:19 +0000 (17:25 +0800)]
Test how PTRACE_SETOPTIONS support works
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>