d/freebsd-dev
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
19fe43f796
freebsd-dev/usr.bin/truss/syscalls.c

2710 lines
80 KiB
C
Raw Normal View History

spdx: initial adoption of licensing ID tags. The Software Package Data Exchange (SPDX) group provides a specification to make it easier for automated tools to detect and summarize well known opensource licenses. We are gradually adopting the specification, noting that the tags are considered only advisory and do not, in any way, superceed or replace the license texts. Special thanks to Wind River for providing access to "The Duke of Highlander" tool: an older (2014) run over FreeBSD tree was useful as a starting point. Initially, only tag files that use BSD 4-Clause "Original" license. RelNotes: yes Differential Revision: https://reviews.freebsd.org/D13133
2017-11-18 14:26:50 +00:00
/*-
* SPDX-License-Identifier: BSD-4-Clause
*
Fix copyright spelling. PR: 139825 Submitted by: Ruslan Mahmatkhanov
2010-03-10 20:31:30 +00:00
* Copyright 1997 Sean Eric Fagan
Add some copyright and license terms, at Jordan's request. Note that syscall.h just refers to another file for the copyright notice.
1997-12-20 18:40:43 +00:00
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. All advertising materials mentioning features or use of this software
* must display the following acknowledgement:
* This product includes software developed by Sean Eric Fagan
* 4. Neither the name of the author may be used to endorse or promote
* products derived from this software without specific prior written
* permission.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*/
Several changes to truss. - Refactor the interface between the ABI-independent code and the ABI-specific backends. The backends now provide smaller hooks to fetch system call arguments and return values. The rest of the system call entry and exit handling that was previously duplicated among all the backends has been moved to one place. - Merge the loop when waiting for an event with the loop for handling stops. This also means not emulating a procfs-like interface on top of ptrace(). Instead, use a single event loop that fetches process events via waitid(). Among other things this allows us to report the full 32-bit exit value. - Use PT_FOLLOW_FORK to follow new child processes instead of forking a new truss process for each new child. This allows one truss process to monitor a tree of processes and truss -c should now display one total for the entire tree instead of separate summaries per process. - Use the recently added fields to ptrace_lwpinfo to determine the current system call number and argument count. The latter is especially useful and fixes a regression since the conversion from procfs. truss now generally prints the correct number of arguments for most system calls rather than printing extra arguments for any call not listed in the table in syscalls.c. - Actually check the new ABI when processes call exec. The comments claimed that this happened but it was not being done (perhaps this was another regression in the conversion to ptrace()). If the new ABI after exec is not supported, truss detaches from the process. If truss does not support the ABI for a newly executed process the process is killed before it returns from exec. - Along with the refactor, teach the various ABI-specific backends to fetch both return values, not just the first. Use this to properly report the full 64-bit return value from lseek(). In addition, the handler for "pipe" now pulls the pair of descriptors out of the return values (which is the true kernel system call interface) but displays them as an argument (which matches the interface exported by libc). - Each ABI handler adds entries to a linker set rather than requiring a statically defined table of handlers in main.c. - The arm and mips system call fetching code was changed to follow the same pattern as amd64 (and the in-kernel handler) of fetching register arguments first and then reading any remaining arguments from the stack. This should fix indirect system call arguments on at least arm. - The mipsn32 and n64 ABIs will now look for arguments in A4 through A7. - Use register %ebp for the 6th system call argument for Linux/i386 ABIs to match the in-kernel argument fetch code. - For powerpc binaries on a powerpc64 system, fetch the extra arguments on the stack as 32-bit values that are then copied into the 64-bit argument array instead of reading the 32-bit values directly into the 64-bit array. Reviewed by: kib (earlier version) Tested on: amd64 (FreeBSD/amd64 & i386), i386, arm (earlier version) Tested on: powerpc64 (FreeBSD/powerpc64 & powerpc) MFC after: 1 month Differential Revision: https://reviews.freebsd.org/D3575
2015-09-30 19:13:32 +00:00
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
Sort Xrefs. Use err(3). Remove uneeded #include. Correct usage: one of {-p pid, command} is required. Open output file when command line is fully analyzed: incorrect `truss -o f' command does not create an empty file anymore.
1998-01-05 07:30:26 +00:00
Truss program. Requires procfs.
1997-12-06 05:23:12 +00:00
/*
* This file has routines used to print out system calls and their
* arguments.
*/
Decode the arguments passed to __cap_rights_get() and cap_rights_limit(). Submitted by: tobik
2017-06-02 13:33:50 +00:00
#include <sys/capsicum.h>
Teach truss(1) to display sockaddrs. It currently knows about AF_INET, AF_INET6 and AF_UNIX sockaddrs, and will recognize accept(), bind(), connect(), getpeername() and getsockname() as syscalls taking sockaddr arguments. Some enterprising soul might want to add (and test) support for the send() / recv() family of syscalls as well. MFC after: 1 week
2001-10-21 21:57:10 +00:00
#include <sys/types.h>
Decode kevent structures logged via ktrace(2) in kdump. - Add a new KTR_STRUCT_ARRAY ktrace record type which dumps an array of structures. The structure name in the record payload is preceded by a size_t containing the size of the individual structures. Use this to replace the previous code that dumped the kevent arrays dumped for kevent(). kdump is now able to decode the kevent structures rather than dumping their contents via a hexdump. One change from before is that the 'changes' and 'events' arrays are not marked with separate 'read' and 'write' annotations in kdump output. Instead, the first array is the 'changes' array, and the second array (only present if kevent doesn't fail with an error) is the 'events' array. For kevent(), empty arrays are denoted by an entry with an array containing zero entries rather than no record. - Move kevent decoding tables from truss to libsysdecode. This adds three new functions to decode members of struct kevent: sysdecode_kevent_filter, sysdecode_kevent_flags, and sysdecode_kevent_fflags. kdump uses these helper functions to pretty-print kevent fields. - Move structure definitions for freebsd11 and freebsd32 kevent structures to <sys/event.h> so that they can be shared with userland. The 32-bit structures are only exposed if _WANT_KEVENT32 is defined. The freebsd11 structures are only exposed if _WANT_FREEBSD11_KEVENT is defined. The 32-bit freebsd11 structure requires both. - Decode freebsd11 kevent structures in truss for the compat11.kevent() system call. - Log 32-bit kevent structures via ktrace for 32-bit compat kevent() system calls. - While here, constify the 'void *data' argument to ktrstruct(). Reviewed by: kib (earlier version) MFC after: 1 month Differential Revision: https://reviews.freebsd.org/D12470
2017-11-25 04:49:12 +00:00
#define _WANT_FREEBSD11_KEVENT
Several changes to truss. - Refactor the interface between the ABI-independent code and the ABI-specific backends. The backends now provide smaller hooks to fetch system call arguments and return values. The rest of the system call entry and exit handling that was previously duplicated among all the backends has been moved to one place. - Merge the loop when waiting for an event with the loop for handling stops. This also means not emulating a procfs-like interface on top of ptrace(). Instead, use a single event loop that fetches process events via waitid(). Among other things this allows us to report the full 32-bit exit value. - Use PT_FOLLOW_FORK to follow new child processes instead of forking a new truss process for each new child. This allows one truss process to monitor a tree of processes and truss -c should now display one total for the entire tree instead of separate summaries per process. - Use the recently added fields to ptrace_lwpinfo to determine the current system call number and argument count. The latter is especially useful and fixes a regression since the conversion from procfs. truss now generally prints the correct number of arguments for most system calls rather than printing extra arguments for any call not listed in the table in syscalls.c. - Actually check the new ABI when processes call exec. The comments claimed that this happened but it was not being done (perhaps this was another regression in the conversion to ptrace()). If the new ABI after exec is not supported, truss detaches from the process. If truss does not support the ABI for a newly executed process the process is killed before it returns from exec. - Along with the refactor, teach the various ABI-specific backends to fetch both return values, not just the first. Use this to properly report the full 64-bit return value from lseek(). In addition, the handler for "pipe" now pulls the pair of descriptors out of the return values (which is the true kernel system call interface) but displays them as an argument (which matches the interface exported by libc). - Each ABI handler adds entries to a linker set rather than requiring a statically defined table of handlers in main.c. - The arm and mips system call fetching code was changed to follow the same pattern as amd64 (and the in-kernel handler) of fetching register arguments first and then reading any remaining arguments from the stack. This should fix indirect system call arguments on at least arm. - The mipsn32 and n64 ABIs will now look for arguments in A4 through A7. - Use register %ebp for the 6th system call argument for Linux/i386 ABIs to match the in-kernel argument fetch code. - For powerpc binaries on a powerpc64 system, fetch the extra arguments on the stack as 32-bit values that are then copied into the 64-bit argument array instead of reading the 32-bit values directly into the 64-bit array. Reviewed by: kib (earlier version) Tested on: amd64 (FreeBSD/amd64 & i386), i386, arm (earlier version) Tested on: powerpc64 (FreeBSD/powerpc64 & powerpc) MFC after: 1 month Differential Revision: https://reviews.freebsd.org/D3575
2015-09-30 19:13:32 +00:00
#include <sys/event.h>
#include <sys/ioccom.h>
Add decoding for struct statfs. Reviewed by: jhb (briefly)
2015-10-03 18:57:15 +00:00
#include <sys/mount.h>
Make use of ptrace(2) instead of procfs in truss(1), eliminating yet another need of an available /proc/ mount. Tested with: make universe Submitted by: howardsu Reviewed by: alfred
2007-04-10 04:03:34 +00:00
#include <sys/ptrace.h>
Several changes to truss. - Refactor the interface between the ABI-independent code and the ABI-specific backends. The backends now provide smaller hooks to fetch system call arguments and return values. The rest of the system call entry and exit handling that was previously duplicated among all the backends has been moved to one place. - Merge the loop when waiting for an event with the loop for handling stops. This also means not emulating a procfs-like interface on top of ptrace(). Instead, use a single event loop that fetches process events via waitid(). Among other things this allows us to report the full 32-bit exit value. - Use PT_FOLLOW_FORK to follow new child processes instead of forking a new truss process for each new child. This allows one truss process to monitor a tree of processes and truss -c should now display one total for the entire tree instead of separate summaries per process. - Use the recently added fields to ptrace_lwpinfo to determine the current system call number and argument count. The latter is especially useful and fixes a regression since the conversion from procfs. truss now generally prints the correct number of arguments for most system calls rather than printing extra arguments for any call not listed in the table in syscalls.c. - Actually check the new ABI when processes call exec. The comments claimed that this happened but it was not being done (perhaps this was another regression in the conversion to ptrace()). If the new ABI after exec is not supported, truss detaches from the process. If truss does not support the ABI for a newly executed process the process is killed before it returns from exec. - Along with the refactor, teach the various ABI-specific backends to fetch both return values, not just the first. Use this to properly report the full 64-bit return value from lseek(). In addition, the handler for "pipe" now pulls the pair of descriptors out of the return values (which is the true kernel system call interface) but displays them as an argument (which matches the interface exported by libc). - Each ABI handler adds entries to a linker set rather than requiring a statically defined table of handlers in main.c. - The arm and mips system call fetching code was changed to follow the same pattern as amd64 (and the in-kernel handler) of fetching register arguments first and then reading any remaining arguments from the stack. This should fix indirect system call arguments on at least arm. - The mipsn32 and n64 ABIs will now look for arguments in A4 through A7. - Use register %ebp for the 6th system call argument for Linux/i386 ABIs to match the in-kernel argument fetch code. - For powerpc binaries on a powerpc64 system, fetch the extra arguments on the stack as 32-bit values that are then copied into the 64-bit argument array instead of reading the 32-bit values directly into the 64-bit array. Reviewed by: kib (earlier version) Tested on: amd64 (FreeBSD/amd64 & i386), i386, arm (earlier version) Tested on: powerpc64 (FreeBSD/powerpc64 & powerpc) MFC after: 1 month Differential Revision: https://reviews.freebsd.org/D3575
2015-09-30 19:13:32 +00:00
#include <sys/resource.h>
Teach truss(1) to display sockaddrs. It currently knows about AF_INET, AF_INET6 and AF_UNIX sockaddrs, and will recognize accept(), bind(), connect(), getpeername() and getsockname() as syscalls taking sockaddr arguments. Some enterprising soul might want to add (and test) support for the send() / recv() family of syscalls as well. MFC after: 1 week
2001-10-21 21:57:10 +00:00
#include <sys/socket.h>
Decode FreeBSD 11 compat stat, fstat and lstat calls.
2017-06-23 18:06:46 +00:00
#define _WANT_FREEBSD11_STAT
Several changes to truss. - Refactor the interface between the ABI-independent code and the ABI-specific backends. The backends now provide smaller hooks to fetch system call arguments and return values. The rest of the system call entry and exit handling that was previously duplicated among all the backends has been moved to one place. - Merge the loop when waiting for an event with the loop for handling stops. This also means not emulating a procfs-like interface on top of ptrace(). Instead, use a single event loop that fetches process events via waitid(). Among other things this allows us to report the full 32-bit exit value. - Use PT_FOLLOW_FORK to follow new child processes instead of forking a new truss process for each new child. This allows one truss process to monitor a tree of processes and truss -c should now display one total for the entire tree instead of separate summaries per process. - Use the recently added fields to ptrace_lwpinfo to determine the current system call number and argument count. The latter is especially useful and fixes a regression since the conversion from procfs. truss now generally prints the correct number of arguments for most system calls rather than printing extra arguments for any call not listed in the table in syscalls.c. - Actually check the new ABI when processes call exec. The comments claimed that this happened but it was not being done (perhaps this was another regression in the conversion to ptrace()). If the new ABI after exec is not supported, truss detaches from the process. If truss does not support the ABI for a newly executed process the process is killed before it returns from exec. - Along with the refactor, teach the various ABI-specific backends to fetch both return values, not just the first. Use this to properly report the full 64-bit return value from lseek(). In addition, the handler for "pipe" now pulls the pair of descriptors out of the return values (which is the true kernel system call interface) but displays them as an argument (which matches the interface exported by libc). - Each ABI handler adds entries to a linker set rather than requiring a statically defined table of handlers in main.c. - The arm and mips system call fetching code was changed to follow the same pattern as amd64 (and the in-kernel handler) of fetching register arguments first and then reading any remaining arguments from the stack. This should fix indirect system call arguments on at least arm. - The mipsn32 and n64 ABIs will now look for arguments in A4 through A7. - Use register %ebp for the 6th system call argument for Linux/i386 ABIs to match the in-kernel argument fetch code. - For powerpc binaries on a powerpc64 system, fetch the extra arguments on the stack as 32-bit values that are then copied into the 64-bit argument array instead of reading the 32-bit values directly into the 64-bit array. Reviewed by: kib (earlier version) Tested on: amd64 (FreeBSD/amd64 & i386), i386, arm (earlier version) Tested on: powerpc64 (FreeBSD/powerpc64 & powerpc) MFC after: 1 month Differential Revision: https://reviews.freebsd.org/D3575
2015-09-30 19:13:32 +00:00
#include <sys/stat.h>
Teach truss(1) to display sockaddrs. It currently knows about AF_INET, AF_INET6 and AF_UNIX sockaddrs, and will recognize accept(), bind(), connect(), getpeername() and getsockname() as syscalls taking sockaddr arguments. Some enterprising soul might want to add (and test) support for the send() / recv() family of syscalls as well. MFC after: 1 week
2001-10-21 21:57:10 +00:00
#include <sys/un.h>
- Decode the idtype argument passed to wait6() in kdump and truss. - Don't treat an options argument of 0 to wait4() as an error in kdump. - Decode the wait options passed to wait4() and wait6() in truss and decode the returned rusage and exit status. Approved by: re (kib) MFC after: 1 week
2013-09-12 18:08:25 +00:00
#include <sys/wait.h>
Teach truss(1) to display sockaddrs. It currently knows about AF_INET, AF_INET6 and AF_UNIX sockaddrs, and will recognize accept(), bind(), connect(), getpeername() and getsockname() as syscalls taking sockaddr arguments. Some enterprising soul might want to add (and test) support for the send() / recv() family of syscalls as well. MFC after: 1 week
2001-10-21 21:57:10 +00:00
#include <netinet/in.h>
Improve support for sctp_generic_recvmsg() and sctp_generic_sendmsg() and add support for sctp_generic_sendmsg_iov(). Handle the struct iovec argument and the struct sctp_sndrcvinfo arguments.
2018-01-14 14:33:22 +00:00
#include <netinet/sctp.h>
Teach truss(1) to display sockaddrs. It currently knows about AF_INET, AF_INET6 and AF_UNIX sockaddrs, and will recognize accept(), bind(), connect(), getpeername() and getsockname() as syscalls taking sockaddr arguments. Some enterprising soul might want to add (and test) support for the send() / recv() family of syscalls as well. MFC after: 1 week
2001-10-21 21:57:10 +00:00
#include <arpa/inet.h>
Rework syscall structure lookups. Avoid always using an O(n^2) loop over known syscall structures with strcmp() on each system call. Instead, use a per-ABI cache indexed by the system call number. The first 1024 system calls (which should cover all of the normal system calls in currently-supported ABIs) use a flat array indexed by the system call number to find system call structure. For other system calls, a linked list of structures storing an integer to structure mapping is stored in the ABI. The linked list isn't very smart, but it should only be used by buggy applications invoking unknown system calls. This also fixes handling of unknown system calls which currently trigger a NULL pointer dereference. Reviewed by: kib MFC after: 2 weeks
2016-12-06 00:39:00 +00:00
#include <assert.h>
Add missing include for <ctype.h> Make sockaddr printing code unbad. Always copy in sizeof(sockaddr_un) bytes for an AF_UNIX sockaddr, despite what the length may be.
2001-11-06 19:26:51 +00:00
#include <ctype.h>
Sort Xrefs. Use err(3). Remove uneeded #include. Correct usage: one of {-p pid, command} is required. Open output file when command line is fully analyzed: incorrect `truss -o f' command does not create an empty file anymore.
1998-01-05 07:30:26 +00:00
#include <err.h>
decode fcntl and mmap arguments.
2004-03-23 09:04:06 +00:00
#include <fcntl.h>
Add support for decoding Timespec, Timeval, Itimerval, Pollfd, Fd_set and Sigaction structures. Use these for printing the arguments to sigaction(), nanosleep(), select(), poll(), gettimeofday(), clock_gettime(), recvfrom(), getitimer() and setitimer(). This is based on Dan's patch from the PR but I've hacked it for style and some other issues. While Dan has checked this patch, any goofs are probably my fault. (The PR also contains support for the dual return values of pipe(). These will follow once I've ported that support to platforms other than i386.) PR: 52190 Submitted by: Dan Nelson <dnelson@allantgroup.com>
2004-03-23 12:37:02 +00:00
#include <poll.h>
Decode arguments to sched_* family of system calls. This includes decoding both scheduler policy constants and the sched_param structure for sched_get_priority_max(), sched_get_priority_min(), sched_getparam(), sched_getscheduler(), sched_rr_get_interval(), sched_setparam(), and sched_setscheduler().
2017-06-16 20:03:09 +00:00
#include <sched.h>
Teach truss(1) to display sockaddrs. It currently knows about AF_INET, AF_INET6 and AF_UNIX sockaddrs, and will recognize accept(), bind(), connect(), getpeername() and getsockname() as syscalls taking sockaddr arguments. Some enterprising soul might want to add (and test) support for the send() / recv() family of syscalls as well. MFC after: 1 week
2001-10-21 21:57:10 +00:00
#include <signal.h>
Properly format pointer size independent CloudABI system calls. CloudABI has approximately 50 system calls that do not depend on the pointer size of the system. As the ABI is pretty compact, it takes little effort to each truss(8) the formatting rules for these system calls. Start off by formatting pointer size independent system calls. Changes: - Make it possible to include the CloudABI system call definitions in FreeBSD userspace builds. Add ${root}/sys to the truss(8) Makefile so we can pull in <compat/cloudabi/cloudabi_syscalldefs.h>. - Refactoring: patch up amd64-cloudabi64.c to use the CLOUDABI_* constants instead of rolling our own table. - Add table entries for all of the system calls. - Add new generic formatting types (UInt, IntArray) that we'll be using to format unsigned integers and arrays of integers. - Add CloudABI specific formatting types. Approved by: jhb Differential Revision: https://reviews.freebsd.org/D3836
2015-10-08 05:27:45 +00:00
#include <stdbool.h>
Truss program. Requires procfs.
1997-12-06 05:23:12 +00:00
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
Start on a new library (libsysdecode) that provides routines for decoding system call information such as system call arguments. Initially this will consist of pulling duplicated code out of truss and kdump though it may prove useful for other utilities in the future. This commit moves the shared utrace(2) record parser out of kdump into the library and updates kdump and truss to use it. One difference from the previous version is that the library version treats unknown events that start with the "RTLD" signature as unknown events. This simplifies the interface and allows the consumer to decide how to handle all non-recognized events. Instead, this function only generates a string description for known malloc() and RTLD records. Reviewed by: bdrewery Differential Revision: https://reviews.freebsd.org/D4537
2015-12-15 00:05:07 +00:00
#include <sysdecode.h>
Truss program. Requires procfs.
1997-12-06 05:23:12 +00:00
#include <unistd.h>
- Add decoding of kse_release, kevent, sigprocmask, unmount, socket, getrusage, rename, __getcwd, shutdown, getrlimit, setrlimit, _umtx_lock, _umtx_unlock, pathconf, truncate, ftruncate, kill - Decode more arguments of open, mprot, *stat, and fcntl. - Convert all constant-macro and bitfield decoding to lookup tables; much cleaner than previous code. - Print the timestamp of process exit and signal reception when -d or -D are in use - Try six times with 1/2 second delay to debug the child PR: bin/52190 (updated) Submitted by: Dan Nelson <dnelson@allantgroup.com> Approved by: alfred
2006-05-15 21:18:28 +00:00
#include <vis.h>
Teach truss(1) to display sockaddrs. It currently knows about AF_INET, AF_INET6 and AF_UNIX sockaddrs, and will recognize accept(), bind(), connect(), getpeername() and getsockname() as syscalls taking sockaddr arguments. Some enterprising soul might want to add (and test) support for the send() / recv() family of syscalls as well. MFC after: 1 week
2001-10-21 21:57:10 +00:00
Replace the CloudABI system call table by a machine generated version. The type definitions and constants that were used by COMPAT_CLOUDABI64 are a literal copy of some headers stored inside of CloudABI's C library, cloudlibc. What is annoying is that we can't make use of cloudlibc's system call list, as the format is completely different and doesn't provide enough information. It had to be synced in manually. We recently decided to solve this (and some other problems) by moving the ABI definitions into a separate file: https://github.com/NuxiNL/cloudabi/blob/master/cloudabi.txt This file is processed by a pile of Python scripts to generate the header files like before, documentation (markdown), but in our case more importantly: a FreeBSD system call table. This change discards the old files in sys/contrib/cloudabi and replaces them by the latest copies, which requires some minor changes here and there. Because cloudabi.txt also enforces consistent names of the system call arguments, we have to patch up a small number of system call implementations to use the new argument names. The new header files can also be included directly in FreeBSD kernel space without needing any includes/defines, so we can now remove cloudabi_syscalldefs.h and cloudabi64_syscalldefs.h. Patch up the sources to include the definitions directly from sys/contrib/cloudabi instead.
2016-03-24 21:47:15 +00:00
#include <contrib/cloudabi/cloudabi_types_common.h>
Properly format pointer size independent CloudABI system calls. CloudABI has approximately 50 system calls that do not depend on the pointer size of the system. As the ABI is pretty compact, it takes little effort to each truss(8) the formatting rules for these system calls. Start off by formatting pointer size independent system calls. Changes: - Make it possible to include the CloudABI system call definitions in FreeBSD userspace builds. Add ${root}/sys to the truss(8) Makefile so we can pull in <compat/cloudabi/cloudabi_syscalldefs.h>. - Refactoring: patch up amd64-cloudabi64.c to use the CLOUDABI_* constants instead of rolling our own table. - Add table entries for all of the system calls. - Add new generic formatting types (UInt, IntArray) that we'll be using to format unsigned integers and arrays of integers. - Add CloudABI specific formatting types. Approved by: jhb Differential Revision: https://reviews.freebsd.org/D3836
2015-10-08 05:27:45 +00:00
Parameterize globals. PR: bin/25587 (in part) MFC after: 3 weeks
2002-08-04 00:46:48 +00:00
#include "truss.h"
Partial WARNS=1 fizes with NO_WERROR set to prevent world breakage. Use __FBSDID().
2001-12-11 23:34:02 +00:00
#include "extern.h"
Truss program. Requires procfs.
1997-12-06 05:23:12 +00:00
#include "syscall.h"
/*
- Add decoding of kse_release, kevent, sigprocmask, unmount, socket, getrusage, rename, __getcwd, shutdown, getrlimit, setrlimit, _umtx_lock, _umtx_unlock, pathconf, truncate, ftruncate, kill - Decode more arguments of open, mprot, *stat, and fcntl. - Convert all constant-macro and bitfield decoding to lookup tables; much cleaner than previous code. - Print the timestamp of process exit and signal reception when -d or -D are in use - Try six times with 1/2 second delay to debug the child PR: bin/52190 (updated) Submitted by: Dan Nelson <dnelson@allantgroup.com> Approved by: alfred
2006-05-15 21:18:28 +00:00
* This should probably be in its own file, sorted alphabetically.
Truss program. Requires procfs.
1997-12-06 05:23:12 +00:00
*/
Fix tracking of unknown syscalls for 'truss -c'. This is done by changing get_syscall() to either lookup the known syscall or add it into the list with the default handlers for printing. This also simplifies some code to not have to check if the syscall variable is set or NULL. Reviewed by: jhb Relnotes: yes Differential Revision: https://reviews.freebsd.org/D3792
2015-10-05 18:08:35 +00:00
static struct syscall decoded_syscalls[] = {
Group the decoded system calls by ABI and sort the calls within each ABI. Reviewed by: bdrewery Glanced at by: kib Differential Revision: https://reviews.freebsd.org/D3823
2015-10-06 19:31:07 +00:00
/* Native ABI */
Decode arguments to ACL related system calls. This only decodes the raw arguments but not the contents of the struct acl objects.
2017-06-08 03:51:17 +00:00
{ .name = "__acl_aclcheck_fd", .ret_type = 1, .nargs = 3,
.args = { { Int, 0 }, { Acltype, 1 }, { Ptr, 2 } } },
{ .name = "__acl_aclcheck_file", .ret_type = 1, .nargs = 3,
.args = { { Name, 0 }, { Acltype, 1 }, { Ptr, 2 } } },
{ .name = "__acl_aclcheck_link", .ret_type = 1, .nargs = 3,
.args = { { Name, 0 }, { Acltype, 1 }, { Ptr, 2 } } },
{ .name = "__acl_delete_fd", .ret_type = 1, .nargs = 2,
.args = { { Int, 0 }, { Acltype, 1 } } },
{ .name = "__acl_delete_file", .ret_type = 1, .nargs = 2,
.args = { { Name, 0 }, { Acltype, 1 } } },
{ .name = "__acl_delete_link", .ret_type = 1, .nargs = 2,
.args = { { Name, 0 }, { Acltype, 1 } } },
{ .name = "__acl_get_fd", .ret_type = 1, .nargs = 3,
.args = { { Int, 0 }, { Acltype, 1 }, { Ptr, 2 } } },
{ .name = "__acl_get_file", .ret_type = 1, .nargs = 3,
.args = { { Name, 0 }, { Acltype, 1 }, { Ptr, 2 } } },
{ .name = "__acl_get_link", .ret_type = 1, .nargs = 3,
.args = { { Name, 0 }, { Acltype, 1 }, { Ptr, 2 } } },
{ .name = "__acl_set_fd", .ret_type = 1, .nargs = 3,
.args = { { Int, 0 }, { Acltype, 1 }, { Ptr, 2 } } },
{ .name = "__acl_set_file", .ret_type = 1, .nargs = 3,
.args = { { Name, 0 }, { Acltype, 1 }, { Ptr, 2 } } },
{ .name = "__acl_set_link", .ret_type = 1, .nargs = 3,
.args = { { Name, 0 }, { Acltype, 1 }, { Ptr, 2 } } },
Decode the arguments passed to __cap_rights_get() and cap_rights_limit(). Submitted by: tobik
2017-06-02 13:33:50 +00:00
{ .name = "__cap_rights_get", .ret_type = 1, .nargs = 3,
.args = { { Int, 0 }, { Int, 1 }, { CapRights | OUT, 2 } } },
Group the decoded system calls by ABI and sort the calls within each ABI. Reviewed by: bdrewery Glanced at by: kib Differential Revision: https://reviews.freebsd.org/D3823
2015-10-06 19:31:07 +00:00
{ .name = "__getcwd", .ret_type = 1, .nargs = 2,
.args = { { Name | OUT, 0 }, { Int, 1 } } },
{ .name = "_umtx_op", .ret_type = 1, .nargs = 5,
.args = { { Ptr, 0 }, { Umtxop, 1 }, { LongHex, 2 }, { Ptr, 3 },
{ Ptr, 4 } } },
{ .name = "accept", .ret_type = 1, .nargs = 3,
.args = { { Int, 0 }, { Sockaddr | OUT, 1 }, { Ptr | OUT, 2 } } },
{ .name = "access", .ret_type = 1, .nargs = 2,
.args = { { Name | IN, 0 }, { Accessmode, 1 } } },
{ .name = "bind", .ret_type = 1, .nargs = 3,
Add Socklent for handling args of type socklen_t.
2017-05-03 09:20:36 +00:00
.args = { { Int, 0 }, { Sockaddr | IN, 1 }, { Socklent, 2 } } },
Group the decoded system calls by ABI and sort the calls within each ABI. Reviewed by: bdrewery Glanced at by: kib Differential Revision: https://reviews.freebsd.org/D3823
2015-10-06 19:31:07 +00:00
{ .name = "bindat", .ret_type = 1, .nargs = 4,
.args = { { Atfd, 0 }, { Int, 1 }, { Sockaddr | IN, 2 },
Decode the arguments passed to the *at() family of system calls. This is especially useful now that libc's open() always calls openat(). While here, fix a few other things: - Decode the mode argument passed to access(), eaccess(), and faccessat(). - Decode the atfd paramete to pretty-print AT_FDCWD. - Decode the special AT_* flags used with some of the *at() system calls. - Decode arguments for fchmod(), lchmod(), fchown(), lchown(), eaccess(), and futimens(). - Decode both of the timeval structures passed to futimes() instead of just the first one.
2015-08-06 19:08:33 +00:00
{ Int, 3 } } },
Group the decoded system calls by ABI and sort the calls within each ABI. Reviewed by: bdrewery Glanced at by: kib Differential Revision: https://reviews.freebsd.org/D3823
2015-10-06 19:31:07 +00:00
{ .name = "break", .ret_type = 1, .nargs = 1,
.args = { { Ptr, 0 } } },
Decode the arguments passed to cap_fcntls_get() and cap_fcntls_limit().
2017-03-18 18:10:02 +00:00
{ .name = "cap_fcntls_get", .ret_type = 1, .nargs = 2,
.args = { { Int, 0 }, { CapFcntlRights | OUT, 1 } } },
{ .name = "cap_fcntls_limit", .ret_type = 1, .nargs = 2,
.args = { { Int, 0 }, { CapFcntlRights, 1 } } },
Decode the argument passed to cap_getmode(). The returned integer value is output.
2017-06-02 22:35:18 +00:00
{ .name = "cap_getmode", .ret_type = 1, .nargs = 1,
.args = { { PUInt | OUT, 0 } } },
Decode the arguments passed to __cap_rights_get() and cap_rights_limit(). Submitted by: tobik
2017-06-02 13:33:50 +00:00
{ .name = "cap_rights_limit", .ret_type = 1, .nargs = 2,
.args = { { Int, 0 }, { CapRights, 1 } } },
Several changes to truss. - Refactor the interface between the ABI-independent code and the ABI-specific backends. The backends now provide smaller hooks to fetch system call arguments and return values. The rest of the system call entry and exit handling that was previously duplicated among all the backends has been moved to one place. - Merge the loop when waiting for an event with the loop for handling stops. This also means not emulating a procfs-like interface on top of ptrace(). Instead, use a single event loop that fetches process events via waitid(). Among other things this allows us to report the full 32-bit exit value. - Use PT_FOLLOW_FORK to follow new child processes instead of forking a new truss process for each new child. This allows one truss process to monitor a tree of processes and truss -c should now display one total for the entire tree instead of separate summaries per process. - Use the recently added fields to ptrace_lwpinfo to determine the current system call number and argument count. The latter is especially useful and fixes a regression since the conversion from procfs. truss now generally prints the correct number of arguments for most system calls rather than printing extra arguments for any call not listed in the table in syscalls.c. - Actually check the new ABI when processes call exec. The comments claimed that this happened but it was not being done (perhaps this was another regression in the conversion to ptrace()). If the new ABI after exec is not supported, truss detaches from the process. If truss does not support the ABI for a newly executed process the process is killed before it returns from exec. - Along with the refactor, teach the various ABI-specific backends to fetch both return values, not just the first. Use this to properly report the full 64-bit return value from lseek(). In addition, the handler for "pipe" now pulls the pair of descriptors out of the return values (which is the true kernel system call interface) but displays them as an argument (which matches the interface exported by libc). - Each ABI handler adds entries to a linker set rather than requiring a statically defined table of handlers in main.c. - The arm and mips system call fetching code was changed to follow the same pattern as amd64 (and the in-kernel handler) of fetching register arguments first and then reading any remaining arguments from the stack. This should fix indirect system call arguments on at least arm. - The mipsn32 and n64 ABIs will now look for arguments in A4 through A7. - Use register %ebp for the 6th system call argument for Linux/i386 ABIs to match the in-kernel argument fetch code. - For powerpc binaries on a powerpc64 system, fetch the extra arguments on the stack as 32-bit values that are then copied into the 64-bit argument array instead of reading the 32-bit values directly into the 64-bit array. Reviewed by: kib (earlier version) Tested on: amd64 (FreeBSD/amd64 & i386), i386, arm (earlier version) Tested on: powerpc64 (FreeBSD/powerpc64 & powerpc) MFC after: 1 month Differential Revision: https://reviews.freebsd.org/D3575
2015-09-30 19:13:32 +00:00
{ .name = "chdir", .ret_type = 1, .nargs = 1,
Add -c option to summarize number of calls, errors, and system time. Reviewed by: alfred
2009-05-12 20:42:12 +00:00
.args = { { Name, 0 } } },
Group the decoded system calls by ABI and sort the calls within each ABI. Reviewed by: bdrewery Glanced at by: kib Differential Revision: https://reviews.freebsd.org/D3823
2015-10-06 19:31:07 +00:00
{ .name = "chflags", .ret_type = 1, .nargs = 2,
Decode file flags passed to *chflags*(). While here, decode arguments passed to fchflags() and chflagsat().
2017-03-18 18:21:41 +00:00
.args = { { Name | IN, 0 }, { FileFlags, 1 } } },
{ .name = "chflagsat", .ret_type = 1, .nargs = 4,
.args = { { Atfd, 0 }, { Name | IN, 1 }, { FileFlags, 2 },
{ Atflags, 3 } } },
Several changes to truss. - Refactor the interface between the ABI-independent code and the ABI-specific backends. The backends now provide smaller hooks to fetch system call arguments and return values. The rest of the system call entry and exit handling that was previously duplicated among all the backends has been moved to one place. - Merge the loop when waiting for an event with the loop for handling stops. This also means not emulating a procfs-like interface on top of ptrace(). Instead, use a single event loop that fetches process events via waitid(). Among other things this allows us to report the full 32-bit exit value. - Use PT_FOLLOW_FORK to follow new child processes instead of forking a new truss process for each new child. This allows one truss process to monitor a tree of processes and truss -c should now display one total for the entire tree instead of separate summaries per process. - Use the recently added fields to ptrace_lwpinfo to determine the current system call number and argument count. The latter is especially useful and fixes a regression since the conversion from procfs. truss now generally prints the correct number of arguments for most system calls rather than printing extra arguments for any call not listed in the table in syscalls.c. - Actually check the new ABI when processes call exec. The comments claimed that this happened but it was not being done (perhaps this was another regression in the conversion to ptrace()). If the new ABI after exec is not supported, truss detaches from the process. If truss does not support the ABI for a newly executed process the process is killed before it returns from exec. - Along with the refactor, teach the various ABI-specific backends to fetch both return values, not just the first. Use this to properly report the full 64-bit return value from lseek(). In addition, the handler for "pipe" now pulls the pair of descriptors out of the return values (which is the true kernel system call interface) but displays them as an argument (which matches the interface exported by libc). - Each ABI handler adds entries to a linker set rather than requiring a statically defined table of handlers in main.c. - The arm and mips system call fetching code was changed to follow the same pattern as amd64 (and the in-kernel handler) of fetching register arguments first and then reading any remaining arguments from the stack. This should fix indirect system call arguments on at least arm. - The mipsn32 and n64 ABIs will now look for arguments in A4 through A7. - Use register %ebp for the 6th system call argument for Linux/i386 ABIs to match the in-kernel argument fetch code. - For powerpc binaries on a powerpc64 system, fetch the extra arguments on the stack as 32-bit values that are then copied into the 64-bit argument array instead of reading the 32-bit values directly into the 64-bit array. Reviewed by: kib (earlier version) Tested on: amd64 (FreeBSD/amd64 & i386), i386, arm (earlier version) Tested on: powerpc64 (FreeBSD/powerpc64 & powerpc) MFC after: 1 month Differential Revision: https://reviews.freebsd.org/D3575
2015-09-30 19:13:32 +00:00
{ .name = "chmod", .ret_type = 1, .nargs = 2,
Add -c option to summarize number of calls, errors, and system time. Reviewed by: alfred
2009-05-12 20:42:12 +00:00
.args = { { Name, 0 }, { Octal, 1 } } },
Group the decoded system calls by ABI and sort the calls within each ABI. Reviewed by: bdrewery Glanced at by: kib Differential Revision: https://reviews.freebsd.org/D3823
2015-10-06 19:31:07 +00:00
{ .name = "chown", .ret_type = 1, .nargs = 3,
.args = { { Name, 0 }, { Int, 1 }, { Int, 2 } } },
{ .name = "chroot", .ret_type = 1, .nargs = 1,
.args = { { Name, 0 } } },
{ .name = "clock_gettime", .ret_type = 1, .nargs = 2,
.args = { { Int, 0 }, { Timespec | OUT, 1 } } },
{ .name = "close", .ret_type = 1, .nargs = 1,
.args = { { Int, 0 } } },
Sort the compat11.* syscalls I added. Remove duplicate compat11.stat. Submitted by: jhb@
2017-06-26 22:48:04 +00:00
{ .name = "compat11.fstat", .ret_type = 1, .nargs = 2,
.args = { { Int, 0 }, { Stat11 | OUT, 1 } } },
Decode FreeBSD11 fstatat calls.
2017-07-19 23:34:28 +00:00
{ .name = "compat11.fstatat", .ret_type = 1, .nargs = 4,
.args = { { Atfd, 0 }, { Name | IN, 1 }, { Stat11 | OUT, 2 },
{ Atflags, 3 } } },
Decode kevent structures logged via ktrace(2) in kdump. - Add a new KTR_STRUCT_ARRAY ktrace record type which dumps an array of structures. The structure name in the record payload is preceded by a size_t containing the size of the individual structures. Use this to replace the previous code that dumped the kevent arrays dumped for kevent(). kdump is now able to decode the kevent structures rather than dumping their contents via a hexdump. One change from before is that the 'changes' and 'events' arrays are not marked with separate 'read' and 'write' annotations in kdump output. Instead, the first array is the 'changes' array, and the second array (only present if kevent doesn't fail with an error) is the 'events' array. For kevent(), empty arrays are denoted by an entry with an array containing zero entries rather than no record. - Move kevent decoding tables from truss to libsysdecode. This adds three new functions to decode members of struct kevent: sysdecode_kevent_filter, sysdecode_kevent_flags, and sysdecode_kevent_fflags. kdump uses these helper functions to pretty-print kevent fields. - Move structure definitions for freebsd11 and freebsd32 kevent structures to <sys/event.h> so that they can be shared with userland. The 32-bit structures are only exposed if _WANT_KEVENT32 is defined. The freebsd11 structures are only exposed if _WANT_FREEBSD11_KEVENT is defined. The 32-bit freebsd11 structure requires both. - Decode freebsd11 kevent structures in truss for the compat11.kevent() system call. - Log 32-bit kevent structures via ktrace for 32-bit compat kevent() system calls. - While here, constify the 'void *data' argument to ktrstruct(). Reviewed by: kib (earlier version) MFC after: 1 month Differential Revision: https://reviews.freebsd.org/D12470
2017-11-25 04:49:12 +00:00
{ .name = "compat11.kevent", .ret_type = 1, .nargs = 6,
.args = { { Int, 0 }, { Kevent11, 1 }, { Int, 2 },
{ Kevent11 | OUT, 3 }, { Int, 4 }, { Timespec, 5 } } },
Sort the compat11.* syscalls I added. Remove duplicate compat11.stat. Submitted by: jhb@
2017-06-26 22:48:04 +00:00
{ .name = "compat11.lstat", .ret_type = 1, .nargs = 2,
.args = { { Name | IN, 0 }, { Stat11 | OUT, 1 } } },
{ .name = "compat11.stat", .ret_type = 1, .nargs = 2,
.args = { { Name | IN, 0 }, { Stat11 | OUT, 1 } } },
Group the decoded system calls by ABI and sort the calls within each ABI. Reviewed by: bdrewery Glanced at by: kib Differential Revision: https://reviews.freebsd.org/D3823
2015-10-06 19:31:07 +00:00
{ .name = "connect", .ret_type = 1, .nargs = 3,
Add Socklent for handling args of type socklen_t.
2017-05-03 09:20:36 +00:00
.args = { { Int, 0 }, { Sockaddr | IN, 1 }, { Socklent, 2 } } },
Group the decoded system calls by ABI and sort the calls within each ABI. Reviewed by: bdrewery Glanced at by: kib Differential Revision: https://reviews.freebsd.org/D3823
2015-10-06 19:31:07 +00:00
{ .name = "connectat", .ret_type = 1, .nargs = 4,
.args = { { Atfd, 0 }, { Int, 1 }, { Sockaddr | IN, 2 },
{ Int, 3 } } },
Decode arguments to dup, dup2, getdirentries, pread, and pwrite. - dup and dup2 print fd arguments in decimal. - pread and pwrite are similar to read and write with the addition of the file offset. - getdirentries displays the output entries as a string for now and also prints the value returned in *basep. Eventually the buffer for getdirentries should perhaps be decoded as an array of dirent structures. PR: 214885 Submitted by: Jonathan de Boyne Pollard <J.deBoynePollard-newsgroups@NTLWorld.COM>
2017-06-05 05:25:50 +00:00
{ .name = "dup", .ret_type = 1, .nargs = 1,
.args = { { Int, 0 } } },
{ .name = "dup2", .ret_type = 1, .nargs = 2,
.args = { { Int, 0 }, { Int, 1 } } },
Group the decoded system calls by ABI and sort the calls within each ABI. Reviewed by: bdrewery Glanced at by: kib Differential Revision: https://reviews.freebsd.org/D3823
2015-10-06 19:31:07 +00:00
{ .name = "eaccess", .ret_type = 1, .nargs = 2,
.args = { { Name | IN, 0 }, { Accessmode, 1 } } },
{ .name = "execve", .ret_type = 1, .nargs = 3,
.args = { { Name | IN, 0 }, { ExecArgs | IN, 1 },
{ ExecEnv | IN, 2 } } },
{ .name = "exit", .ret_type = 0, .nargs = 1,
.args = { { Hex, 0 } } },
Decode arguments passed to extended attribute related system calls. The cmd argument passed to extattrctl() is not decoded as a string constant but is just printed in hex. The value is filesystem-specific but in practice is only used with UFS1 filesystems.
2017-06-08 04:31:15 +00:00
{ .name = "extattr_delete_fd", .ret_type = 1, .nargs = 3,
.args = { { Int, 0 }, { Extattrnamespace, 1 }, { Name, 2 } } },
{ .name = "extattr_delete_file", .ret_type = 1, .nargs = 3,
.args = { { Name, 0 }, { Extattrnamespace, 1 }, { Name, 2 } } },
{ .name = "extattr_delete_link", .ret_type = 1, .nargs = 3,
.args = { { Name, 0 }, { Extattrnamespace, 1 }, { Name, 2 } } },
{ .name = "extattr_get_fd", .ret_type = 1, .nargs = 5,
.args = { { Int, 0 }, { Extattrnamespace, 1 }, { Name, 2 },
{ BinString | OUT, 3 }, { Sizet, 4 } } },
{ .name = "extattr_get_file", .ret_type = 1, .nargs = 5,
.args = { { Name, 0 }, { Extattrnamespace, 1 }, { Name, 2 },
{ BinString | OUT, 3 }, { Sizet, 4 } } },
{ .name = "extattr_get_link", .ret_type = 1, .nargs = 5,
.args = { { Name, 0 }, { Extattrnamespace, 1 }, { Name, 2 },
{ BinString | OUT, 3 }, { Sizet, 4 } } },
{ .name = "extattr_list_fd", .ret_type = 1, .nargs = 4,
.args = { { Int, 0 }, { Extattrnamespace, 1 }, { BinString | OUT, 2 },
{ Sizet, 3 } } },
{ .name = "extattr_list_file", .ret_type = 1, .nargs = 4,
.args = { { Name, 0 }, { Extattrnamespace, 1 }, { BinString | OUT, 2 },
{ Sizet, 3 } } },
{ .name = "extattr_list_link", .ret_type = 1, .nargs = 4,
.args = { { Name, 0 }, { Extattrnamespace, 1 }, { BinString | OUT, 2 },
{ Sizet, 3 } } },
{ .name = "extattr_set_fd", .ret_type = 1, .nargs = 5,
.args = { { Int, 0 }, { Extattrnamespace, 1 }, { Name, 2 },
{ BinString | IN, 3 }, { Sizet, 4 } } },
{ .name = "extattr_set_file", .ret_type = 1, .nargs = 5,
.args = { { Name, 0 }, { Extattrnamespace, 1 }, { Name, 2 },
{ BinString | IN, 3 }, { Sizet, 4 } } },
{ .name = "extattr_set_link", .ret_type = 1, .nargs = 5,
.args = { { Name, 0 }, { Extattrnamespace, 1 }, { Name, 2 },
{ BinString | IN, 3 }, { Sizet, 4 } } },
{ .name = "extattrctl", .ret_type = 1, .nargs = 5,
.args = { { Name, 0 }, { Hex, 1 }, { Name, 2 },
{ Extattrnamespace, 3 }, { Name, 4 } } },
Group the decoded system calls by ABI and sort the calls within each ABI. Reviewed by: bdrewery Glanced at by: kib Differential Revision: https://reviews.freebsd.org/D3823
2015-10-06 19:31:07 +00:00
{ .name = "faccessat", .ret_type = 1, .nargs = 4,
.args = { { Atfd, 0 }, { Name | IN, 1 }, { Accessmode, 2 },
{ Atflags, 3 } } },
Decode file flags passed to *chflags*(). While here, decode arguments passed to fchflags() and chflagsat().
2017-03-18 18:21:41 +00:00
{ .name = "fchflags", .ret_type = 1, .nargs = 2,
.args = { { Int, 0 }, { FileFlags, 1 } } },
Several changes to truss. - Refactor the interface between the ABI-independent code and the ABI-specific backends. The backends now provide smaller hooks to fetch system call arguments and return values. The rest of the system call entry and exit handling that was previously duplicated among all the backends has been moved to one place. - Merge the loop when waiting for an event with the loop for handling stops. This also means not emulating a procfs-like interface on top of ptrace(). Instead, use a single event loop that fetches process events via waitid(). Among other things this allows us to report the full 32-bit exit value. - Use PT_FOLLOW_FORK to follow new child processes instead of forking a new truss process for each new child. This allows one truss process to monitor a tree of processes and truss -c should now display one total for the entire tree instead of separate summaries per process. - Use the recently added fields to ptrace_lwpinfo to determine the current system call number and argument count. The latter is especially useful and fixes a regression since the conversion from procfs. truss now generally prints the correct number of arguments for most system calls rather than printing extra arguments for any call not listed in the table in syscalls.c. - Actually check the new ABI when processes call exec. The comments claimed that this happened but it was not being done (perhaps this was another regression in the conversion to ptrace()). If the new ABI after exec is not supported, truss detaches from the process. If truss does not support the ABI for a newly executed process the process is killed before it returns from exec. - Along with the refactor, teach the various ABI-specific backends to fetch both return values, not just the first. Use this to properly report the full 64-bit return value from lseek(). In addition, the handler for "pipe" now pulls the pair of descriptors out of the return values (which is the true kernel system call interface) but displays them as an argument (which matches the interface exported by libc). - Each ABI handler adds entries to a linker set rather than requiring a statically defined table of handlers in main.c. - The arm and mips system call fetching code was changed to follow the same pattern as amd64 (and the in-kernel handler) of fetching register arguments first and then reading any remaining arguments from the stack. This should fix indirect system call arguments on at least arm. - The mipsn32 and n64 ABIs will now look for arguments in A4 through A7. - Use register %ebp for the 6th system call argument for Linux/i386 ABIs to match the in-kernel argument fetch code. - For powerpc binaries on a powerpc64 system, fetch the extra arguments on the stack as 32-bit values that are then copied into the 64-bit argument array instead of reading the 32-bit values directly into the 64-bit array. Reviewed by: kib (earlier version) Tested on: amd64 (FreeBSD/amd64 & i386), i386, arm (earlier version) Tested on: powerpc64 (FreeBSD/powerpc64 & powerpc) MFC after: 1 month Differential Revision: https://reviews.freebsd.org/D3575
2015-09-30 19:13:32 +00:00
{ .name = "fchmod", .ret_type = 1, .nargs = 2,
Decode the arguments passed to the *at() family of system calls. This is especially useful now that libc's open() always calls openat(). While here, fix a few other things: - Decode the mode argument passed to access(), eaccess(), and faccessat(). - Decode the atfd paramete to pretty-print AT_FDCWD. - Decode the special AT_* flags used with some of the *at() system calls. - Decode arguments for fchmod(), lchmod(), fchown(), lchown(), eaccess(), and futimens(). - Decode both of the timeval structures passed to futimes() instead of just the first one.
2015-08-06 19:08:33 +00:00
.args = { { Int, 0 }, { Octal, 1 } } },
Several changes to truss. - Refactor the interface between the ABI-independent code and the ABI-specific backends. The backends now provide smaller hooks to fetch system call arguments and return values. The rest of the system call entry and exit handling that was previously duplicated among all the backends has been moved to one place. - Merge the loop when waiting for an event with the loop for handling stops. This also means not emulating a procfs-like interface on top of ptrace(). Instead, use a single event loop that fetches process events via waitid(). Among other things this allows us to report the full 32-bit exit value. - Use PT_FOLLOW_FORK to follow new child processes instead of forking a new truss process for each new child. This allows one truss process to monitor a tree of processes and truss -c should now display one total for the entire tree instead of separate summaries per process. - Use the recently added fields to ptrace_lwpinfo to determine the current system call number and argument count. The latter is especially useful and fixes a regression since the conversion from procfs. truss now generally prints the correct number of arguments for most system calls rather than printing extra arguments for any call not listed in the table in syscalls.c. - Actually check the new ABI when processes call exec. The comments claimed that this happened but it was not being done (perhaps this was another regression in the conversion to ptrace()). If the new ABI after exec is not supported, truss detaches from the process. If truss does not support the ABI for a newly executed process the process is killed before it returns from exec. - Along with the refactor, teach the various ABI-specific backends to fetch both return values, not just the first. Use this to properly report the full 64-bit return value from lseek(). In addition, the handler for "pipe" now pulls the pair of descriptors out of the return values (which is the true kernel system call interface) but displays them as an argument (which matches the interface exported by libc). - Each ABI handler adds entries to a linker set rather than requiring a statically defined table of handlers in main.c. - The arm and mips system call fetching code was changed to follow the same pattern as amd64 (and the in-kernel handler) of fetching register arguments first and then reading any remaining arguments from the stack. This should fix indirect system call arguments on at least arm. - The mipsn32 and n64 ABIs will now look for arguments in A4 through A7. - Use register %ebp for the 6th system call argument for Linux/i386 ABIs to match the in-kernel argument fetch code. - For powerpc binaries on a powerpc64 system, fetch the extra arguments on the stack as 32-bit values that are then copied into the 64-bit argument array instead of reading the 32-bit values directly into the 64-bit array. Reviewed by: kib (earlier version) Tested on: amd64 (FreeBSD/amd64 & i386), i386, arm (earlier version) Tested on: powerpc64 (FreeBSD/powerpc64 & powerpc) MFC after: 1 month Differential Revision: https://reviews.freebsd.org/D3575
2015-09-30 19:13:32 +00:00
{ .name = "fchmodat", .ret_type = 1, .nargs = 4,
Decode the arguments passed to the *at() family of system calls. This is especially useful now that libc's open() always calls openat(). While here, fix a few other things: - Decode the mode argument passed to access(), eaccess(), and faccessat(). - Decode the atfd paramete to pretty-print AT_FDCWD. - Decode the special AT_* flags used with some of the *at() system calls. - Decode arguments for fchmod(), lchmod(), fchown(), lchown(), eaccess(), and futimens(). - Decode both of the timeval structures passed to futimes() instead of just the first one.
2015-08-06 19:08:33 +00:00
.args = { { Atfd, 0 }, { Name, 1 }, { Octal, 2 }, { Atflags, 3 } } },
Several changes to truss. - Refactor the interface between the ABI-independent code and the ABI-specific backends. The backends now provide smaller hooks to fetch system call arguments and return values. The rest of the system call entry and exit handling that was previously duplicated among all the backends has been moved to one place. - Merge the loop when waiting for an event with the loop for handling stops. This also means not emulating a procfs-like interface on top of ptrace(). Instead, use a single event loop that fetches process events via waitid(). Among other things this allows us to report the full 32-bit exit value. - Use PT_FOLLOW_FORK to follow new child processes instead of forking a new truss process for each new child. This allows one truss process to monitor a tree of processes and truss -c should now display one total for the entire tree instead of separate summaries per process. - Use the recently added fields to ptrace_lwpinfo to determine the current system call number and argument count. The latter is especially useful and fixes a regression since the conversion from procfs. truss now generally prints the correct number of arguments for most system calls rather than printing extra arguments for any call not listed in the table in syscalls.c. - Actually check the new ABI when processes call exec. The comments claimed that this happened but it was not being done (perhaps this was another regression in the conversion to ptrace()). If the new ABI after exec is not supported, truss detaches from the process. If truss does not support the ABI for a newly executed process the process is killed before it returns from exec. - Along with the refactor, teach the various ABI-specific backends to fetch both return values, not just the first. Use this to properly report the full 64-bit return value from lseek(). In addition, the handler for "pipe" now pulls the pair of descriptors out of the return values (which is the true kernel system call interface) but displays them as an argument (which matches the interface exported by libc). - Each ABI handler adds entries to a linker set rather than requiring a statically defined table of handlers in main.c. - The arm and mips system call fetching code was changed to follow the same pattern as amd64 (and the in-kernel handler) of fetching register arguments first and then reading any remaining arguments from the stack. This should fix indirect system call arguments on at least arm. - The mipsn32 and n64 ABIs will now look for arguments in A4 through A7. - Use register %ebp for the 6th system call argument for Linux/i386 ABIs to match the in-kernel argument fetch code. - For powerpc binaries on a powerpc64 system, fetch the extra arguments on the stack as 32-bit values that are then copied into the 64-bit argument array instead of reading the 32-bit values directly into the 64-bit array. Reviewed by: kib (earlier version) Tested on: amd64 (FreeBSD/amd64 & i386), i386, arm (earlier version) Tested on: powerpc64 (FreeBSD/powerpc64 & powerpc) MFC after: 1 month Differential Revision: https://reviews.freebsd.org/D3575
2015-09-30 19:13:32 +00:00
{ .name = "fchown", .ret_type = 1, .nargs = 3,
Decode the arguments passed to the *at() family of system calls. This is especially useful now that libc's open() always calls openat(). While here, fix a few other things: - Decode the mode argument passed to access(), eaccess(), and faccessat(). - Decode the atfd paramete to pretty-print AT_FDCWD. - Decode the special AT_* flags used with some of the *at() system calls. - Decode arguments for fchmod(), lchmod(), fchown(), lchown(), eaccess(), and futimens(). - Decode both of the timeval structures passed to futimes() instead of just the first one.
2015-08-06 19:08:33 +00:00
.args = { { Int, 0 }, { Int, 1 }, { Int, 2 } } },
Several changes to truss. - Refactor the interface between the ABI-independent code and the ABI-specific backends. The backends now provide smaller hooks to fetch system call arguments and return values. The rest of the system call entry and exit handling that was previously duplicated among all the backends has been moved to one place. - Merge the loop when waiting for an event with the loop for handling stops. This also means not emulating a procfs-like interface on top of ptrace(). Instead, use a single event loop that fetches process events via waitid(). Among other things this allows us to report the full 32-bit exit value. - Use PT_FOLLOW_FORK to follow new child processes instead of forking a new truss process for each new child. This allows one truss process to monitor a tree of processes and truss -c should now display one total for the entire tree instead of separate summaries per process. - Use the recently added fields to ptrace_lwpinfo to determine the current system call number and argument count. The latter is especially useful and fixes a regression since the conversion from procfs. truss now generally prints the correct number of arguments for most system calls rather than printing extra arguments for any call not listed in the table in syscalls.c. - Actually check the new ABI when processes call exec. The comments claimed that this happened but it was not being done (perhaps this was another regression in the conversion to ptrace()). If the new ABI after exec is not supported, truss detaches from the process. If truss does not support the ABI for a newly executed process the process is killed before it returns from exec. - Along with the refactor, teach the various ABI-specific backends to fetch both return values, not just the first. Use this to properly report the full 64-bit return value from lseek(). In addition, the handler for "pipe" now pulls the pair of descriptors out of the return values (which is the true kernel system call interface) but displays them as an argument (which matches the interface exported by libc). - Each ABI handler adds entries to a linker set rather than requiring a statically defined table of handlers in main.c. - The arm and mips system call fetching code was changed to follow the same pattern as amd64 (and the in-kernel handler) of fetching register arguments first and then reading any remaining arguments from the stack. This should fix indirect system call arguments on at least arm. - The mipsn32 and n64 ABIs will now look for arguments in A4 through A7. - Use register %ebp for the 6th system call argument for Linux/i386 ABIs to match the in-kernel argument fetch code. - For powerpc binaries on a powerpc64 system, fetch the extra arguments on the stack as 32-bit values that are then copied into the 64-bit argument array instead of reading the 32-bit values directly into the 64-bit array. Reviewed by: kib (earlier version) Tested on: amd64 (FreeBSD/amd64 & i386), i386, arm (earlier version) Tested on: powerpc64 (FreeBSD/powerpc64 & powerpc) MFC after: 1 month Differential Revision: https://reviews.freebsd.org/D3575
2015-09-30 19:13:32 +00:00
{ .name = "fchownat", .ret_type = 1, .nargs = 5,
Decode the arguments passed to the *at() family of system calls. This is especially useful now that libc's open() always calls openat(). While here, fix a few other things: - Decode the mode argument passed to access(), eaccess(), and faccessat(). - Decode the atfd paramete to pretty-print AT_FDCWD. - Decode the special AT_* flags used with some of the *at() system calls. - Decode arguments for fchmod(), lchmod(), fchown(), lchown(), eaccess(), and futimens(). - Decode both of the timeval structures passed to futimes() instead of just the first one.
2015-08-06 19:08:33 +00:00
.args = { { Atfd, 0 }, { Name, 1 }, { Int, 2 }, { Int, 3 },
{ Atflags, 4 } } },
Group the decoded system calls by ABI and sort the calls within each ABI. Reviewed by: bdrewery Glanced at by: kib Differential Revision: https://reviews.freebsd.org/D3823
2015-10-06 19:31:07 +00:00
{ .name = "fcntl", .ret_type = 1, .nargs = 3,
.args = { { Int, 0 }, { Fcntl, 1 }, { Fcntlflag, 2 } } },
Decode flock() operation.
2017-03-18 18:26:56 +00:00
{ .name = "flock", .ret_type = 1, .nargs = 2,
.args = { { Int, 0 }, { Flockop, 1 } } },
Add -c option to summarize number of calls, errors, and system time. Reviewed by: alfred
2009-05-12 20:42:12 +00:00
{ .name = "fstat", .ret_type = 1, .nargs = 2,
Whitespace fix: remove some spurious spaces before commas.
2015-08-05 18:10:46 +00:00
.args = { { Int, 0 }, { Stat | OUT, 1 } } },
Decode the arguments passed to the *at() family of system calls. This is especially useful now that libc's open() always calls openat(). While here, fix a few other things: - Decode the mode argument passed to access(), eaccess(), and faccessat(). - Decode the atfd paramete to pretty-print AT_FDCWD. - Decode the special AT_* flags used with some of the *at() system calls. - Decode arguments for fchmod(), lchmod(), fchown(), lchown(), eaccess(), and futimens(). - Decode both of the timeval structures passed to futimes() instead of just the first one.
2015-08-06 19:08:33 +00:00
{ .name = "fstatat", .ret_type = 1, .nargs = 4,
.args = { { Atfd, 0 }, { Name | IN, 1 }, { Stat | OUT, 2 },
{ Atflags, 3 } } },
Add decoding for struct statfs. Reviewed by: jhb (briefly)
2015-10-03 18:57:15 +00:00
{ .name = "fstatfs", .ret_type = 1, .nargs = 2,
.args = { { Int, 0 }, { StatFs | OUT, 1 } } },
Group the decoded system calls by ABI and sort the calls within each ABI. Reviewed by: bdrewery Glanced at by: kib Differential Revision: https://reviews.freebsd.org/D3823
2015-10-06 19:31:07 +00:00
{ .name = "ftruncate", .ret_type = 1, .nargs = 2,
Automate the handling of QUAD_ALIGN and QUAD_SLOTS. Previously, the offset in a system call description specified the array index of the start of a system call argument. For most system call arguments this was the same as the index of the argument in the function signature. 64-bit arguments (off_t and id_t values) passed on 32-bit platforms use two slots in the array however. This was handled by adding (QUAD_SLOTS - 1) to the slot indicies of any subsequent arguments after a 64-bit argument (though written as ("{ Quad, 1 }, { Int, 1 + QUAD_SLOTS }" rather than "{ Quad, 1 }, { Int, 2 + QUAD_SLOTS - 1 }"). If a system call contained multiple 64-bit arguments (such as posix_fadvise()), then additional arguments would need to use 'QUAD_SLOTS * 2' but remember to subtract 2 from the initial number, etc. In addition, 32-bit powerpc requires 64-bit arguments to be 64-bit aligned, so if the effective index in the array of a 64-bit argument is odd, it needs QUAD_ALIGN added to the current and any subsequent slots. However, if the effective index in the array of a 64-bit argument was even, QUAD_ALIGN was omitted. This approach was messy and error prone. This commit replaces it with automated pre-processing of the system call table to do fixups for 64-bit argument offsets. The offset in a system call description now indicates the index of an argument in the associated function call's signature. A fixup function is run against each decoded system call description during startup on 32-bit platforms. The fixup function maintains an 'offset' value which holds an offset to be added to each remaining system call argument's index. Initially offset is 0. When a 64-bit system call argument is encountered, the offset is first aligned to a 64-bit boundary (only on powerpc) and then incremented to account for the second argument slot used by the argument. This modified 'offset' is then applied to any remaining arguments. This approach does require a few things that were not previously required: 1) Each system call description must now list arguments in ascending order (existing ones all do) without using duplicate slots in the register array. A new assert() should catch any future descriptions which violate this rule. 2) A system call description is still permitted to omit arguments (though none currently do), but if the call accepts 64-bit arguments those cannot be omitted or incorrect results will be displated on 32-bit systems. Tested on: amd64 and i386
2017-03-15 23:08:11 +00:00
.args = { { Int | IN, 0 }, { QuadHex | IN, 1 } } },
Group the decoded system calls by ABI and sort the calls within each ABI. Reviewed by: bdrewery Glanced at by: kib Differential Revision: https://reviews.freebsd.org/D3823
2015-10-06 19:31:07 +00:00
{ .name = "futimens", .ret_type = 1, .nargs = 2,
.args = { { Int, 0 }, { Timespec2 | IN, 1 } } },
{ .name = "futimes", .ret_type = 1, .nargs = 2,
.args = { { Int, 0 }, { Timeval2 | IN, 1 } } },
{ .name = "futimesat", .ret_type = 1, .nargs = 3,
.args = { { Atfd, 0 }, { Name | IN, 1 }, { Timeval2 | IN, 2 } } },
Decode arguments to dup, dup2, getdirentries, pread, and pwrite. - dup and dup2 print fd arguments in decimal. - pread and pwrite are similar to read and write with the addition of the file offset. - getdirentries displays the output entries as a string for now and also prints the value returned in *basep. Eventually the buffer for getdirentries should perhaps be decoded as an array of dirent structures. PR: 214885 Submitted by: Jonathan de Boyne Pollard <J.deBoynePollard-newsgroups@NTLWorld.COM>
2017-06-05 05:25:50 +00:00
{ .name = "getdirentries", .ret_type = 1, .nargs = 4,
.args = { { Int, 0 }, { BinString | OUT, 1 }, { Int, 2 },
{ PQuadHex | OUT, 3 } } },
Decode arguments passed to getfsstat(). Note that this does not yet decode the statfs structures returned by getfsstat().
2017-03-18 18:31:45 +00:00
{ .name = "getfsstat", .ret_type = 1, .nargs = 3,
.args = { { Ptr, 0 }, { Long, 1 }, { Getfsstatmode, 2 } } },
Group the decoded system calls by ABI and sort the calls within each ABI. Reviewed by: bdrewery Glanced at by: kib Differential Revision: https://reviews.freebsd.org/D3823
2015-10-06 19:31:07 +00:00
{ .name = "getitimer", .ret_type = 1, .nargs = 2,
.args = { { Int, 0 }, { Itimerval | OUT, 2 } } },
Add -c option to summarize number of calls, errors, and system time. Reviewed by: alfred
2009-05-12 20:42:12 +00:00
{ .name = "getpeername", .ret_type = 1, .nargs = 3,
.args = { { Int, 0 }, { Sockaddr | OUT, 1 }, { Ptr | OUT, 2 } } },
Group the decoded system calls by ABI and sort the calls within each ABI. Reviewed by: bdrewery Glanced at by: kib Differential Revision: https://reviews.freebsd.org/D3823
2015-10-06 19:31:07 +00:00
{ .name = "getpgid", .ret_type = 1, .nargs = 1,
.args = { { Int, 0 } } },
Decode arguments to getpriority() and setpriority().
2017-06-10 00:37:02 +00:00
{ .name = "getpriority", .ret_type = 1, .nargs = 2,
.args = { { Priowhich, 0 }, { Int, 1 } } },
Implement getrandom(2) and getentropy(3) The general idea here is to provide userspace programs with well-defined sources of entropy, in a fashion that doesn't require opening a new file descriptor (ulimits) or accessing paths (/dev/urandom may be restricted by chroot or capsicum). getrandom(2) is the more general API, and comes from the Linux world. Since our urandom and random devices are identical, the GRND_RANDOM flag is ignored. getentropy(3) is added as a compatibility shim for the OpenBSD API. truss(1) support is included. Tests for both system calls are provided. Coverage is believed to be at least as comprehensive as LTP getrandom(2) test coverage. Additionally, instructions for running the LTP tests directly against FreeBSD are provided in the "Test Plan" section of the Differential revision linked below. (They pass, of course.) PR: 194204 Reported by: David CARLIER <david.carlier AT hardenedbsd.org> Discussed with: cperciva, delphij, jhb, markj Relnotes: maybe Differential Revision: https://reviews.freebsd.org/D14500
2018-03-21 01:15:45 +00:00
{ .name = "getrandom", .ret_type = 1, .nargs = 3,
.args = { { BinString | OUT, 0 }, { Sizet, 1 }, { UInt, 2 } } },
Group the decoded system calls by ABI and sort the calls within each ABI. Reviewed by: bdrewery Glanced at by: kib Differential Revision: https://reviews.freebsd.org/D3823
2015-10-06 19:31:07 +00:00
{ .name = "getrlimit", .ret_type = 1, .nargs = 2,
.args = { { Resource, 0 }, { Rlimit | OUT, 1 } } },
{ .name = "getrusage", .ret_type = 1, .nargs = 2,
Decode the 'who' argument passed to getrusage(). Add a new sysdecode_getrusage_who() which decodes the RUSAGE_* constant passed as the first argument to getrusage(). Use this function in both kdump and truss to decode the first argument to getrusage(). PR: 215448 Submitted by: Anton Yuzhaninov <citrin+pr@citrin.ru> MFC after: 1 month
2017-06-03 14:22:15 +00:00
.args = { { RusageWho, 0 }, { Rusage | OUT, 1 } } },
Group the decoded system calls by ABI and sort the calls within each ABI. Reviewed by: bdrewery Glanced at by: kib Differential Revision: https://reviews.freebsd.org/D3823
2015-10-06 19:31:07 +00:00
{ .name = "getsid", .ret_type = 1, .nargs = 1,
.args = { { Int, 0 } } },
Add -c option to summarize number of calls, errors, and system time. Reviewed by: alfred
2009-05-12 20:42:12 +00:00
{ .name = "getsockname", .ret_type = 1, .nargs = 3,
.args = { { Int, 0 }, { Sockaddr | OUT, 1 }, { Ptr | OUT, 2 } } },
Add support for [gs]etsockopt().
2017-05-03 12:18:09 +00:00
{ .name = "getsockopt", .ret_type = 1, .nargs = 5,
.args = { { Int, 0 }, { Sockoptlevel, 1 }, { Sockoptname, 2 },
{ Ptr | OUT, 3 }, { Ptr | OUT, 4 } } },
Group the decoded system calls by ABI and sort the calls within each ABI. Reviewed by: bdrewery Glanced at by: kib Differential Revision: https://reviews.freebsd.org/D3823
2015-10-06 19:31:07 +00:00
{ .name = "gettimeofday", .ret_type = 1, .nargs = 2,
.args = { { Timeval | OUT, 0 }, { Ptr, 1 } } },
{ .name = "ioctl", .ret_type = 1, .nargs = 3,
Improve decoding of last arguments to ioctl() and sendto(). Decode the last argument to ioctl() as a pointer rather than an int. Eventually this could use 'int' for the _IOWINT() case and pointers for all others. The last argument to sendto() is a socklen_t value, not a pointer.
2017-03-19 00:36:29 +00:00
.args = { { Int, 0 }, { Ioctl, 1 }, { Ptr, 2 } } },
Group the decoded system calls by ABI and sort the calls within each ABI. Reviewed by: bdrewery Glanced at by: kib Differential Revision: https://reviews.freebsd.org/D3823
2015-10-06 19:31:07 +00:00
{ .name = "kevent", .ret_type = 1, .nargs = 6,
.args = { { Int, 0 }, { Kevent, 1 }, { Int, 2 }, { Kevent | OUT, 3 },
{ Int, 4 }, { Timespec, 5 } } },
{ .name = "kill", .ret_type = 1, .nargs = 2,
.args = { { Int | IN, 0 }, { Signal | IN, 1 } } },
{ .name = "kldfind", .ret_type = 1, .nargs = 1,
Add -c option to summarize number of calls, errors, and system time. Reviewed by: alfred
2009-05-12 20:42:12 +00:00
.args = { { Name | IN, 0 } } },
Group the decoded system calls by ABI and sort the calls within each ABI. Reviewed by: bdrewery Glanced at by: kib Differential Revision: https://reviews.freebsd.org/D3823
2015-10-06 19:31:07 +00:00
{ .name = "kldfirstmod", .ret_type = 1, .nargs = 1,
Add -c option to summarize number of calls, errors, and system time. Reviewed by: alfred
2009-05-12 20:42:12 +00:00
.args = { { Int, 0 } } },
Group the decoded system calls by ABI and sort the calls within each ABI. Reviewed by: bdrewery Glanced at by: kib Differential Revision: https://reviews.freebsd.org/D3823
2015-10-06 19:31:07 +00:00
{ .name = "kldload", .ret_type = 1, .nargs = 1,
Add -c option to summarize number of calls, errors, and system time. Reviewed by: alfred
2009-05-12 20:42:12 +00:00
.args = { { Name | IN, 0 } } },
Several changes to truss. - Refactor the interface between the ABI-independent code and the ABI-specific backends. The backends now provide smaller hooks to fetch system call arguments and return values. The rest of the system call entry and exit handling that was previously duplicated among all the backends has been moved to one place. - Merge the loop when waiting for an event with the loop for handling stops. This also means not emulating a procfs-like interface on top of ptrace(). Instead, use a single event loop that fetches process events via waitid(). Among other things this allows us to report the full 32-bit exit value. - Use PT_FOLLOW_FORK to follow new child processes instead of forking a new truss process for each new child. This allows one truss process to monitor a tree of processes and truss -c should now display one total for the entire tree instead of separate summaries per process. - Use the recently added fields to ptrace_lwpinfo to determine the current system call number and argument count. The latter is especially useful and fixes a regression since the conversion from procfs. truss now generally prints the correct number of arguments for most system calls rather than printing extra arguments for any call not listed in the table in syscalls.c. - Actually check the new ABI when processes call exec. The comments claimed that this happened but it was not being done (perhaps this was another regression in the conversion to ptrace()). If the new ABI after exec is not supported, truss detaches from the process. If truss does not support the ABI for a newly executed process the process is killed before it returns from exec. - Along with the refactor, teach the various ABI-specific backends to fetch both return values, not just the first. Use this to properly report the full 64-bit return value from lseek(). In addition, the handler for "pipe" now pulls the pair of descriptors out of the return values (which is the true kernel system call interface) but displays them as an argument (which matches the interface exported by libc). - Each ABI handler adds entries to a linker set rather than requiring a statically defined table of handlers in main.c. - The arm and mips system call fetching code was changed to follow the same pattern as amd64 (and the in-kernel handler) of fetching register arguments first and then reading any remaining arguments from the stack. This should fix indirect system call arguments on at least arm. - The mipsn32 and n64 ABIs will now look for arguments in A4 through A7. - Use register %ebp for the 6th system call argument for Linux/i386 ABIs to match the in-kernel argument fetch code. - For powerpc binaries on a powerpc64 system, fetch the extra arguments on the stack as 32-bit values that are then copied into the 64-bit argument array instead of reading the 32-bit values directly into the 64-bit array. Reviewed by: kib (earlier version) Tested on: amd64 (FreeBSD/amd64 & i386), i386, arm (earlier version) Tested on: powerpc64 (FreeBSD/powerpc64 & powerpc) MFC after: 1 month Differential Revision: https://reviews.freebsd.org/D3575
2015-09-30 19:13:32 +00:00
{ .name = "kldnext", .ret_type = 1, .nargs = 1,
Add -c option to summarize number of calls, errors, and system time. Reviewed by: alfred
2009-05-12 20:42:12 +00:00
.args = { { Int, 0 } } },
Several changes to truss. - Refactor the interface between the ABI-independent code and the ABI-specific backends. The backends now provide smaller hooks to fetch system call arguments and return values. The rest of the system call entry and exit handling that was previously duplicated among all the backends has been moved to one place. - Merge the loop when waiting for an event with the loop for handling stops. This also means not emulating a procfs-like interface on top of ptrace(). Instead, use a single event loop that fetches process events via waitid(). Among other things this allows us to report the full 32-bit exit value. - Use PT_FOLLOW_FORK to follow new child processes instead of forking a new truss process for each new child. This allows one truss process to monitor a tree of processes and truss -c should now display one total for the entire tree instead of separate summaries per process. - Use the recently added fields to ptrace_lwpinfo to determine the current system call number and argument count. The latter is especially useful and fixes a regression since the conversion from procfs. truss now generally prints the correct number of arguments for most system calls rather than printing extra arguments for any call not listed in the table in syscalls.c. - Actually check the new ABI when processes call exec. The comments claimed that this happened but it was not being done (perhaps this was another regression in the conversion to ptrace()). If the new ABI after exec is not supported, truss detaches from the process. If truss does not support the ABI for a newly executed process the process is killed before it returns from exec. - Along with the refactor, teach the various ABI-specific backends to fetch both return values, not just the first. Use this to properly report the full 64-bit return value from lseek(). In addition, the handler for "pipe" now pulls the pair of descriptors out of the return values (which is the true kernel system call interface) but displays them as an argument (which matches the interface exported by libc). - Each ABI handler adds entries to a linker set rather than requiring a statically defined table of handlers in main.c. - The arm and mips system call fetching code was changed to follow the same pattern as amd64 (and the in-kernel handler) of fetching register arguments first and then reading any remaining arguments from the stack. This should fix indirect system call arguments on at least arm. - The mipsn32 and n64 ABIs will now look for arguments in A4 through A7. - Use register %ebp for the 6th system call argument for Linux/i386 ABIs to match the in-kernel argument fetch code. - For powerpc binaries on a powerpc64 system, fetch the extra arguments on the stack as 32-bit values that are then copied into the 64-bit argument array instead of reading the 32-bit values directly into the 64-bit array. Reviewed by: kib (earlier version) Tested on: amd64 (FreeBSD/amd64 & i386), i386, arm (earlier version) Tested on: powerpc64 (FreeBSD/powerpc64 & powerpc) MFC after: 1 month Differential Revision: https://reviews.freebsd.org/D3575
2015-09-30 19:13:32 +00:00
{ .name = "kldstat", .ret_type = 1, .nargs = 2,
Add -c option to summarize number of calls, errors, and system time. Reviewed by: alfred
2009-05-12 20:42:12 +00:00
.args = { { Int, 0 }, { Ptr, 1 } } },
Decode arguments passed to kldsym() and kldunloadf(). This does not currently decode the kld_sym_lookup structure passed to kldsym().
2017-03-18 19:59:21 +00:00
{ .name = "kldsym", .ret_type = 1, .nargs = 3,
.args = { { Int, 0 }, { Kldsymcmd, 1 }, { Ptr, 2 } } },
Group the decoded system calls by ABI and sort the calls within each ABI. Reviewed by: bdrewery Glanced at by: kib Differential Revision: https://reviews.freebsd.org/D3823
2015-10-06 19:31:07 +00:00
{ .name = "kldunload", .ret_type = 1, .nargs = 1,
Add -c option to summarize number of calls, errors, and system time. Reviewed by: alfred
2009-05-12 20:42:12 +00:00
.args = { { Int, 0 } } },
Decode arguments passed to kldsym() and kldunloadf(). This does not currently decode the kld_sym_lookup structure passed to kldsym().
2017-03-18 19:59:21 +00:00
{ .name = "kldunloadf", .ret_type = 1, .nargs = 2,
.args = { { Int, 0 }, { Kldunloadflags, 1 } } },
Group the decoded system calls by ABI and sort the calls within each ABI. Reviewed by: bdrewery Glanced at by: kib Differential Revision: https://reviews.freebsd.org/D3823
2015-10-06 19:31:07 +00:00
{ .name = "kse_release", .ret_type = 0, .nargs = 1,
.args = { { Timespec, 0 } } },
{ .name = "lchflags", .ret_type = 1, .nargs = 2,
Decode file flags passed to *chflags*(). While here, decode arguments passed to fchflags() and chflagsat().
2017-03-18 18:21:41 +00:00
.args = { { Name | IN, 0 }, { FileFlags, 1 } } },
Group the decoded system calls by ABI and sort the calls within each ABI. Reviewed by: bdrewery Glanced at by: kib Differential Revision: https://reviews.freebsd.org/D3823
2015-10-06 19:31:07 +00:00
{ .name = "lchmod", .ret_type = 1, .nargs = 2,
.args = { { Name, 0 }, { Octal, 1 } } },
{ .name = "lchown", .ret_type = 1, .nargs = 3,
.args = { { Name, 0 }, { Int, 1 }, { Int, 2 } } },
{ .name = "link", .ret_type = 1, .nargs = 2,
.args = { { Name, 0 }, { Name, 1 } } },
{ .name = "linkat", .ret_type = 1, .nargs = 5,
.args = { { Atfd, 0 }, { Name, 1 }, { Atfd, 2 }, { Name, 3 },
{ Atflags, 4 } } },
Add support for listen() call.
2017-05-03 09:09:34 +00:00
{ .name = "listen", .ret_type = 1, .nargs = 2,
.args = { { Int, 0 }, { Int, 1 } } },
{ .name = "lseek", .ret_type = 2, .nargs = 3,
Automate the handling of QUAD_ALIGN and QUAD_SLOTS. Previously, the offset in a system call description specified the array index of the start of a system call argument. For most system call arguments this was the same as the index of the argument in the function signature. 64-bit arguments (off_t and id_t values) passed on 32-bit platforms use two slots in the array however. This was handled by adding (QUAD_SLOTS - 1) to the slot indicies of any subsequent arguments after a 64-bit argument (though written as ("{ Quad, 1 }, { Int, 1 + QUAD_SLOTS }" rather than "{ Quad, 1 }, { Int, 2 + QUAD_SLOTS - 1 }"). If a system call contained multiple 64-bit arguments (such as posix_fadvise()), then additional arguments would need to use 'QUAD_SLOTS * 2' but remember to subtract 2 from the initial number, etc. In addition, 32-bit powerpc requires 64-bit arguments to be 64-bit aligned, so if the effective index in the array of a 64-bit argument is odd, it needs QUAD_ALIGN added to the current and any subsequent slots. However, if the effective index in the array of a 64-bit argument was even, QUAD_ALIGN was omitted. This approach was messy and error prone. This commit replaces it with automated pre-processing of the system call table to do fixups for 64-bit argument offsets. The offset in a system call description now indicates the index of an argument in the associated function call's signature. A fixup function is run against each decoded system call description during startup on 32-bit platforms. The fixup function maintains an 'offset' value which holds an offset to be added to each remaining system call argument's index. Initially offset is 0. When a 64-bit system call argument is encountered, the offset is first aligned to a 64-bit boundary (only on powerpc) and then incremented to account for the second argument slot used by the argument. This modified 'offset' is then applied to any remaining arguments. This approach does require a few things that were not previously required: 1) Each system call description must now list arguments in ascending order (existing ones all do) without using duplicate slots in the register array. A new assert() should catch any future descriptions which violate this rule. 2) A system call description is still permitted to omit arguments (though none currently do), but if the call accepts 64-bit arguments those cannot be omitted or incorrect results will be displated on 32-bit systems. Tested on: amd64 and i386
2017-03-15 23:08:11 +00:00
.args = { { Int, 0 }, { QuadHex, 1 }, { Whence, 2 } } },
Group the decoded system calls by ABI and sort the calls within each ABI. Reviewed by: bdrewery Glanced at by: kib Differential Revision: https://reviews.freebsd.org/D3823
2015-10-06 19:31:07 +00:00
{ .name = "lstat", .ret_type = 1, .nargs = 2,
.args = { { Name | IN, 0 }, { Stat | OUT, 1 } } },
{ .name = "lutimes", .ret_type = 1, .nargs = 2,
.args = { { Name | IN, 0 }, { Timeval2 | IN, 1 } } },
Decode arguments to madvise().
2017-03-19 00:31:21 +00:00
{ .name = "madvise", .ret_type = 1, .nargs = 3,
.args = { { Ptr, 0 }, { Sizet, 1 }, { Madvice, 2 } } },
Decode arguments to minherit().
2017-06-08 04:45:13 +00:00
{ .name = "minherit", .ret_type = 1, .nargs = 3,
.args = { { Ptr, 0 }, { Sizet, 1 }, { Minherit, 2 } } },
Group the decoded system calls by ABI and sort the calls within each ABI. Reviewed by: bdrewery Glanced at by: kib Differential Revision: https://reviews.freebsd.org/D3823
2015-10-06 19:31:07 +00:00
{ .name = "mkdir", .ret_type = 1, .nargs = 2,
.args = { { Name, 0 }, { Octal, 1 } } },
{ .name = "mkdirat", .ret_type = 1, .nargs = 3,
.args = { { Atfd, 0 }, { Name, 1 }, { Octal, 2 } } },
{ .name = "mkfifo", .ret_type = 1, .nargs = 2,
.args = { { Name, 0 }, { Octal, 1 } } },
{ .name = "mkfifoat", .ret_type = 1, .nargs = 3,
.args = { { Atfd, 0 }, { Name, 1 }, { Octal, 2 } } },
{ .name = "mknod", .ret_type = 1, .nargs = 3,
.args = { { Name, 0 }, { Octal, 1 }, { Int, 2 } } },
{ .name = "mknodat", .ret_type = 1, .nargs = 4,
.args = { { Atfd, 0 }, { Name, 1 }, { Octal, 2 }, { Int, 3 } } },
Decode arguments to mlock(), mlockall(), and munlock().
2017-06-08 04:50:50 +00:00
{ .name = "mlock", .ret_type = 1, .nargs = 2,
.args = { { Ptr, 0 }, { Sizet, 1 } } },
{ .name = "mlockall", .ret_type = 1, .nargs = 1,
.args = { { Mlockall, 0 } } },
Group the decoded system calls by ABI and sort the calls within each ABI. Reviewed by: bdrewery Glanced at by: kib Differential Revision: https://reviews.freebsd.org/D3823
2015-10-06 19:31:07 +00:00
{ .name = "mmap", .ret_type = 1, .nargs = 6,
Add a Sizet type for 'size_t' values and use it instead of Int. Various size_t arguments were previously decoded as Int values instead which would have truncated values above 2^31 on 64-bit systems.
2017-03-19 00:27:07 +00:00
.args = { { Ptr, 0 }, { Sizet, 1 }, { Mprot, 2 }, { Mmapflags, 3 },
Automate the handling of QUAD_ALIGN and QUAD_SLOTS. Previously, the offset in a system call description specified the array index of the start of a system call argument. For most system call arguments this was the same as the index of the argument in the function signature. 64-bit arguments (off_t and id_t values) passed on 32-bit platforms use two slots in the array however. This was handled by adding (QUAD_SLOTS - 1) to the slot indicies of any subsequent arguments after a 64-bit argument (though written as ("{ Quad, 1 }, { Int, 1 + QUAD_SLOTS }" rather than "{ Quad, 1 }, { Int, 2 + QUAD_SLOTS - 1 }"). If a system call contained multiple 64-bit arguments (such as posix_fadvise()), then additional arguments would need to use 'QUAD_SLOTS * 2' but remember to subtract 2 from the initial number, etc. In addition, 32-bit powerpc requires 64-bit arguments to be 64-bit aligned, so if the effective index in the array of a 64-bit argument is odd, it needs QUAD_ALIGN added to the current and any subsequent slots. However, if the effective index in the array of a 64-bit argument was even, QUAD_ALIGN was omitted. This approach was messy and error prone. This commit replaces it with automated pre-processing of the system call table to do fixups for 64-bit argument offsets. The offset in a system call description now indicates the index of an argument in the associated function call's signature. A fixup function is run against each decoded system call description during startup on 32-bit platforms. The fixup function maintains an 'offset' value which holds an offset to be added to each remaining system call argument's index. Initially offset is 0. When a 64-bit system call argument is encountered, the offset is first aligned to a 64-bit boundary (only on powerpc) and then incremented to account for the second argument slot used by the argument. This modified 'offset' is then applied to any remaining arguments. This approach does require a few things that were not previously required: 1) Each system call description must now list arguments in ascending order (existing ones all do) without using duplicate slots in the register array. A new assert() should catch any future descriptions which violate this rule. 2) A system call description is still permitted to omit arguments (though none currently do), but if the call accepts 64-bit arguments those cannot be omitted or incorrect results will be displated on 32-bit systems. Tested on: amd64 and i386
2017-03-15 23:08:11 +00:00
{ Int, 4 }, { QuadHex, 5 } } },
Add decoding for modfind(2)
2015-10-05 18:11:30 +00:00
{ .name = "modfind", .ret_type = 1, .nargs = 1,
.args = { { Name | IN, 0 } } },
Group the decoded system calls by ABI and sort the calls within each ABI. Reviewed by: bdrewery Glanced at by: kib Differential Revision: https://reviews.freebsd.org/D3823
2015-10-06 19:31:07 +00:00
{ .name = "mount", .ret_type = 1, .nargs = 4,
Decode flags passed to mount(), nmount(), and unmount().
2017-06-08 08:07:51 +00:00
.args = { { Name, 0 }, { Name, 1 }, { Mountflags, 2 }, { Ptr, 3 } } },
Group the decoded system calls by ABI and sort the calls within each ABI. Reviewed by: bdrewery Glanced at by: kib Differential Revision: https://reviews.freebsd.org/D3823
2015-10-06 19:31:07 +00:00
{ .name = "mprotect", .ret_type = 1, .nargs = 3,
Add a Sizet type for 'size_t' values and use it instead of Int. Various size_t arguments were previously decoded as Int values instead which would have truncated values above 2^31 on 64-bit systems.
2017-03-19 00:27:07 +00:00
.args = { { Ptr, 0 }, { Sizet, 1 }, { Mprot, 2 } } },
Decode arguments passed to msync().
2017-06-08 08:10:57 +00:00
{ .name = "msync", .ret_type = 1, .nargs = 3,
.args = { { Ptr, 0 }, { Sizet, 1 }, { Msync, 2 } } },
Decode arguments to mlock(), mlockall(), and munlock().
2017-06-08 04:50:50 +00:00
{ .name = "munlock", .ret_type = 1, .nargs = 2,
.args = { { Ptr, 0 }, { Sizet, 1 } } },
Group the decoded system calls by ABI and sort the calls within each ABI. Reviewed by: bdrewery Glanced at by: kib Differential Revision: https://reviews.freebsd.org/D3823
2015-10-06 19:31:07 +00:00
{ .name = "munmap", .ret_type = 1, .nargs = 2,
Add a Sizet type for 'size_t' values and use it instead of Int. Various size_t arguments were previously decoded as Int values instead which would have truncated values above 2^31 on 64-bit systems.
2017-03-19 00:27:07 +00:00
.args = { { Ptr, 0 }, { Sizet, 1 } } },
Several changes to truss. - Refactor the interface between the ABI-independent code and the ABI-specific backends. The backends now provide smaller hooks to fetch system call arguments and return values. The rest of the system call entry and exit handling that was previously duplicated among all the backends has been moved to one place. - Merge the loop when waiting for an event with the loop for handling stops. This also means not emulating a procfs-like interface on top of ptrace(). Instead, use a single event loop that fetches process events via waitid(). Among other things this allows us to report the full 32-bit exit value. - Use PT_FOLLOW_FORK to follow new child processes instead of forking a new truss process for each new child. This allows one truss process to monitor a tree of processes and truss -c should now display one total for the entire tree instead of separate summaries per process. - Use the recently added fields to ptrace_lwpinfo to determine the current system call number and argument count. The latter is especially useful and fixes a regression since the conversion from procfs. truss now generally prints the correct number of arguments for most system calls rather than printing extra arguments for any call not listed in the table in syscalls.c. - Actually check the new ABI when processes call exec. The comments claimed that this happened but it was not being done (perhaps this was another regression in the conversion to ptrace()). If the new ABI after exec is not supported, truss detaches from the process. If truss does not support the ABI for a newly executed process the process is killed before it returns from exec. - Along with the refactor, teach the various ABI-specific backends to fetch both return values, not just the first. Use this to properly report the full 64-bit return value from lseek(). In addition, the handler for "pipe" now pulls the pair of descriptors out of the return values (which is the true kernel system call interface) but displays them as an argument (which matches the interface exported by libc). - Each ABI handler adds entries to a linker set rather than requiring a statically defined table of handlers in main.c. - The arm and mips system call fetching code was changed to follow the same pattern as amd64 (and the in-kernel handler) of fetching register arguments first and then reading any remaining arguments from the stack. This should fix indirect system call arguments on at least arm. - The mipsn32 and n64 ABIs will now look for arguments in A4 through A7. - Use register %ebp for the 6th system call argument for Linux/i386 ABIs to match the in-kernel argument fetch code. - For powerpc binaries on a powerpc64 system, fetch the extra arguments on the stack as 32-bit values that are then copied into the 64-bit argument array instead of reading the 32-bit values directly into the 64-bit array. Reviewed by: kib (earlier version) Tested on: amd64 (FreeBSD/amd64 & i386), i386, arm (earlier version) Tested on: powerpc64 (FreeBSD/powerpc64 & powerpc) MFC after: 1 month Differential Revision: https://reviews.freebsd.org/D3575
2015-09-30 19:13:32 +00:00
{ .name = "nanosleep", .ret_type = 1, .nargs = 1,
Add -c option to summarize number of calls, errors, and system time. Reviewed by: alfred
2009-05-12 20:42:12 +00:00
.args = { { Timespec, 0 } } },
Decode flags passed to mount(), nmount(), and unmount().
2017-06-08 08:07:51 +00:00
{ .name = "nmount", .ret_type = 1, .nargs = 3,
.args = { { Ptr, 0 }, { UInt, 1 }, { Mountflags, 2 } } },
Group the decoded system calls by ABI and sort the calls within each ABI. Reviewed by: bdrewery Glanced at by: kib Differential Revision: https://reviews.freebsd.org/D3823
2015-10-06 19:31:07 +00:00
{ .name = "open", .ret_type = 1, .nargs = 3,
.args = { { Name | IN, 0 }, { Open, 1 }, { Octal, 2 } } },
{ .name = "openat", .ret_type = 1, .nargs = 4,
.args = { { Atfd, 0 }, { Name | IN, 1 }, { Open, 2 },
{ Octal, 3 } } },
{ .name = "pathconf", .ret_type = 1, .nargs = 2,
.args = { { Name | IN, 0 }, { Pathconf, 1 } } },
{ .name = "pipe", .ret_type = 1, .nargs = 1,
.args = { { PipeFds | OUT, 0 } } },
{ .name = "pipe2", .ret_type = 1, .nargs = 2,
Move mksubr from kdump into libsysdecode. Restructure this script so that it generates a header of tables instead of a source file. The tables are included in a flags.c source file which provides functions to decode various system call arguments. For functions that decode an enumeration, the function returns a pointer to a string for known values and NULL for unknown values. For functions that do more complex decoding (typically of a bitmask), the function accepts a pointer to a FILE object (open_memstream() can be used as a string builder) to which decoded values are written. If the function operates on a bitmask, the function returns true if any bits were decoded or false if the entire value was valid. Additionally, the third argument accepts a pointer to a value to which any undecoded bits are stored. This pointer can be NULL if the caller doesn't care about remaining bits. Convert kdump over to using decoder functions from libsysdecode instead of mksubr. truss also uses decoders from libsysdecode instead of private lookup tables, though lookup tables for objects not decoded by kdump remain in truss for now. Eventually most of these tables should move into libsysdecode as the automated table generation approach from mksubr is less stale than the static tables in truss. Some changes have been made to truss and kdump output: - The flags passed to open() are now properly decoded in that one of O_RDONLY, O_RDWR, O_WRONLY, or O_EXEC is always included in a decoded mask. - Optional arguments to open(), openat(), and fcntl() are only printed in kdump if they exist (e.g. the mode is only printed for open() if O_CREAT is set in the flags). - Print argument to F_GETLK/SETLK/SETLKW in kdump as a pointer, not int. - Include all procctl() commands. - Correctly decode pipe2() flags in truss by not assuming full open()-like flags with O_RDONLY, etc. - Decode file flags passed to *chflags() as file flags (UF_* and SF_*) rather than as a file mode. - Fix decoding of quotactl() commands by splitting out the two command components instead of assuming the raw command value matches the primary command component. In addition, truss and kdump now build without triggering any warnings. All of the sysdecode manpages now include the required headers in the synopsis. Reviewed by: kib (several older versions), wblock (manpages) MFC after: 2 months Differential Revision: https://reviews.freebsd.org/D7847
2016-10-17 22:37:07 +00:00
.args = { { Ptr, 0 }, { Pipe2, 1 } } },
Group the decoded system calls by ABI and sort the calls within each ABI. Reviewed by: bdrewery Glanced at by: kib Differential Revision: https://reviews.freebsd.org/D3823
2015-10-06 19:31:07 +00:00
{ .name = "poll", .ret_type = 1, .nargs = 3,
.args = { { Pollfd, 0 }, { Int, 1 }, { Int, 2 } } },
Decode arguments passed to posix_fadvise().
2017-03-18 18:12:09 +00:00
{ .name = "posix_fadvise", .ret_type = 1, .nargs = 4,
.args = { { Int, 0 }, { QuadHex, 1 }, { QuadHex, 2 },
{ Fadvice, 3 } } },
Group the decoded system calls by ABI and sort the calls within each ABI. Reviewed by: bdrewery Glanced at by: kib Differential Revision: https://reviews.freebsd.org/D3823
2015-10-06 19:31:07 +00:00
{ .name = "posix_openpt", .ret_type = 1, .nargs = 1,
.args = { { Open, 0 } } },
Decode arguments to dup, dup2, getdirentries, pread, and pwrite. - dup and dup2 print fd arguments in decimal. - pread and pwrite are similar to read and write with the addition of the file offset. - getdirentries displays the output entries as a string for now and also prints the value returned in *basep. Eventually the buffer for getdirentries should perhaps be decoded as an array of dirent structures. PR: 214885 Submitted by: Jonathan de Boyne Pollard <J.deBoynePollard-newsgroups@NTLWorld.COM>
2017-06-05 05:25:50 +00:00
{ .name = "pread", .ret_type = 1, .nargs = 4,
.args = { { Int, 0 }, { BinString | OUT, 1 }, { Sizet, 2 },
{ QuadHex, 3 } } },
Group the decoded system calls by ABI and sort the calls within each ABI. Reviewed by: bdrewery Glanced at by: kib Differential Revision: https://reviews.freebsd.org/D3823
2015-10-06 19:31:07 +00:00
{ .name = "procctl", .ret_type = 1, .nargs = 4,
Automate the handling of QUAD_ALIGN and QUAD_SLOTS. Previously, the offset in a system call description specified the array index of the start of a system call argument. For most system call arguments this was the same as the index of the argument in the function signature. 64-bit arguments (off_t and id_t values) passed on 32-bit platforms use two slots in the array however. This was handled by adding (QUAD_SLOTS - 1) to the slot indicies of any subsequent arguments after a 64-bit argument (though written as ("{ Quad, 1 }, { Int, 1 + QUAD_SLOTS }" rather than "{ Quad, 1 }, { Int, 2 + QUAD_SLOTS - 1 }"). If a system call contained multiple 64-bit arguments (such as posix_fadvise()), then additional arguments would need to use 'QUAD_SLOTS * 2' but remember to subtract 2 from the initial number, etc. In addition, 32-bit powerpc requires 64-bit arguments to be 64-bit aligned, so if the effective index in the array of a 64-bit argument is odd, it needs QUAD_ALIGN added to the current and any subsequent slots. However, if the effective index in the array of a 64-bit argument was even, QUAD_ALIGN was omitted. This approach was messy and error prone. This commit replaces it with automated pre-processing of the system call table to do fixups for 64-bit argument offsets. The offset in a system call description now indicates the index of an argument in the associated function call's signature. A fixup function is run against each decoded system call description during startup on 32-bit platforms. The fixup function maintains an 'offset' value which holds an offset to be added to each remaining system call argument's index. Initially offset is 0. When a 64-bit system call argument is encountered, the offset is first aligned to a 64-bit boundary (only on powerpc) and then incremented to account for the second argument slot used by the argument. This modified 'offset' is then applied to any remaining arguments. This approach does require a few things that were not previously required: 1) Each system call description must now list arguments in ascending order (existing ones all do) without using duplicate slots in the register array. A new assert() should catch any future descriptions which violate this rule. 2) A system call description is still permitted to omit arguments (though none currently do), but if the call accepts 64-bit arguments those cannot be omitted or incorrect results will be displated on 32-bit systems. Tested on: amd64 and i386
2017-03-15 23:08:11 +00:00
.args = { { Idtype, 0 }, { Quad, 1 }, { Procctl, 2 }, { Ptr, 3 } } },
Decode the arguments to ptrace(). This does not decode structures returned by ptrace().
2017-06-10 00:45:07 +00:00
{ .name = "ptrace", .ret_type = 1, .nargs = 4,
.args = { { Ptraceop, 0 }, { Int, 1 }, { Ptr, 2 }, { Int, 3 } } },
Decode arguments to dup, dup2, getdirentries, pread, and pwrite. - dup and dup2 print fd arguments in decimal. - pread and pwrite are similar to read and write with the addition of the file offset. - getdirentries displays the output entries as a string for now and also prints the value returned in *basep. Eventually the buffer for getdirentries should perhaps be decoded as an array of dirent structures. PR: 214885 Submitted by: Jonathan de Boyne Pollard <J.deBoynePollard-newsgroups@NTLWorld.COM>
2017-06-05 05:25:50 +00:00
{ .name = "pwrite", .ret_type = 1, .nargs = 4,
.args = { { Int, 0 }, { BinString | IN, 1 }, { Sizet, 2 },
{ QuadHex, 3 } } },
Decode the arguments to quotactl().
2017-06-10 00:53:00 +00:00
{ .name = "quotactl", .ret_type = 1, .nargs = 4,
.args = { { Name, 0 }, { Quotactlcmd, 1 }, { Int, 2 }, { Ptr, 3 } } },
Group the decoded system calls by ABI and sort the calls within each ABI. Reviewed by: bdrewery Glanced at by: kib Differential Revision: https://reviews.freebsd.org/D3823
2015-10-06 19:31:07 +00:00
{ .name = "read", .ret_type = 1, .nargs = 3,
Add a Sizet type for 'size_t' values and use it instead of Int. Various size_t arguments were previously decoded as Int values instead which would have truncated values above 2^31 on 64-bit systems.
2017-03-19 00:27:07 +00:00
.args = { { Int, 0 }, { BinString | OUT, 1 }, { Sizet, 2 } } },
Group the decoded system calls by ABI and sort the calls within each ABI. Reviewed by: bdrewery Glanced at by: kib Differential Revision: https://reviews.freebsd.org/D3823
2015-10-06 19:31:07 +00:00
{ .name = "readlink", .ret_type = 1, .nargs = 3,
Add a Sizet type for 'size_t' values and use it instead of Int. Various size_t arguments were previously decoded as Int values instead which would have truncated values above 2^31 on 64-bit systems.
2017-03-19 00:27:07 +00:00
.args = { { Name, 0 }, { Readlinkres | OUT, 1 }, { Sizet, 2 } } },
Group the decoded system calls by ABI and sort the calls within each ABI. Reviewed by: bdrewery Glanced at by: kib Differential Revision: https://reviews.freebsd.org/D3823
2015-10-06 19:31:07 +00:00
{ .name = "readlinkat", .ret_type = 1, .nargs = 4,
.args = { { Atfd, 0 }, { Name, 1 }, { Readlinkres | OUT, 2 },
Add a Sizet type for 'size_t' values and use it instead of Int. Various size_t arguments were previously decoded as Int values instead which would have truncated values above 2^31 on 64-bit systems.
2017-03-19 00:27:07 +00:00
{ Sizet, 3 } } },
Keep list sorted. Thanks to jhb@ to pointing out that I put it at the wrong place in r327919.
2018-01-15 23:17:12 +00:00
{ .name = "readv", .ret_type = 1, .nargs = 3,
.args = { { Int, 0 }, { Iovec | OUT, 1 }, { Int, 2 } } },
Decode the 'howto' argument to reboot().
2017-06-10 01:22:40 +00:00
{ .name = "reboot", .ret_type = 1, .nargs = 1,
.args = { { Reboothowto, 0 } } },
Group the decoded system calls by ABI and sort the calls within each ABI. Reviewed by: bdrewery Glanced at by: kib Differential Revision: https://reviews.freebsd.org/D3823
2015-10-06 19:31:07 +00:00
{ .name = "recvfrom", .ret_type = 1, .nargs = 6,
Decode the fourth argument of sendto and recvfrom call.
2017-05-03 12:23:58 +00:00
.args = { { Int, 0 }, { BinString | OUT, 1 }, { Sizet, 2 },
{ Msgflags, 3 }, { Sockaddr | OUT, 4 },
{ Ptr | OUT, 5 } } },
Add support for sendmsg() and recvmsg().
2017-05-03 12:26:16 +00:00
{ .name = "recvmsg", .ret_type = 1, .nargs = 3,
Decode msghdr argument of sendmsg() and recvmsg(). Sponsored by: Netflix, Inc.
2018-01-15 20:39:42 +00:00
.args = { { Int, 0 }, { Msghdr | OUT, 1 }, { Msgflags, 2 } } },
Group the decoded system calls by ABI and sort the calls within each ABI. Reviewed by: bdrewery Glanced at by: kib Differential Revision: https://reviews.freebsd.org/D3823
2015-10-06 19:31:07 +00:00
{ .name = "rename", .ret_type = 1, .nargs = 2,
.args = { { Name, 0 }, { Name, 1 } } },
{ .name = "renameat", .ret_type = 1, .nargs = 4,
.args = { { Atfd, 0 }, { Name, 1 }, { Atfd, 2 }, { Name, 3 } } },
{ .name = "rfork", .ret_type = 1, .nargs = 1,
.args = { { Rforkflags, 0 } } },
Fix sorting in r303934. Reported by: jhb X-MFC-With: r303934 MFC after: 3 days Sponsored by: EMC / Isilon Storage Division
2016-08-10 21:59:59 +00:00
{ .name = "rmdir", .ret_type = 1, .nargs = 1,
.args = { { Name, 0 } } },
Decode arguments to rtprio() and rtprio_thread().
2017-06-10 01:32:35 +00:00
{ .name = "rtprio", .ret_type = 1, .nargs = 3,
.args = { { Rtpriofunc, 0 }, { Int, 1 }, { Ptr, 2 } } },
{ .name = "rtprio_thread", .ret_type = 1, .nargs = 3,
.args = { { Rtpriofunc, 0 }, { Int, 1 }, { Ptr, 2 } } },
Decode arguments to sched_* family of system calls. This includes decoding both scheduler policy constants and the sched_param structure for sched_get_priority_max(), sched_get_priority_min(), sched_getparam(), sched_getscheduler(), sched_rr_get_interval(), sched_setparam(), and sched_setscheduler().
2017-06-16 20:03:09 +00:00
{ .name = "sched_get_priority_max", .ret_type = 1, .nargs = 1,
.args = { { Schedpolicy, 0 } } },
{ .name = "sched_get_priority_min", .ret_type = 1, .nargs = 1,
.args = { { Schedpolicy, 0 } } },
{ .name = "sched_getparam", .ret_type = 1, .nargs = 2,
.args = { { Int, 0 }, { Schedparam | OUT, 1 } } },
{ .name = "sched_getscheduler", .ret_type = 1, .nargs = 1,
.args = { { Int, 0 } } },
{ .name = "sched_rr_get_interval", .ret_type = 1, .nargs = 2,
.args = { { Int, 0 }, { Timespec | OUT, 1 } } },
{ .name = "sched_setparam", .ret_type = 1, .nargs = 2,
.args = { { Int, 0 }, { Schedparam, 1 } } },
{ .name = "sched_setscheduler", .ret_type = 1, .nargs = 3,
.args = { { Int, 0 }, { Schedpolicy, 1 }, { Schedparam, 2 } } },
Add support for sctp_generic_sendmsg() and sctp_generic_recvmsg().
2017-05-03 16:03:33 +00:00
{ .name = "sctp_generic_recvmsg", .ret_type = 1, .nargs = 7,
Improve support for sctp_generic_recvmsg() and sctp_generic_sendmsg() and add support for sctp_generic_sendmsg_iov(). Handle the struct iovec argument and the struct sctp_sndrcvinfo arguments.
2018-01-14 14:33:22 +00:00
.args = { { Int, 0 }, { Iovec | OUT, 1 }, { Int, 2 },
{ Sockaddr | OUT, 3 }, { Ptr | OUT, 4 },
{ Sctpsndrcvinfo | OUT, 5 }, { Ptr | OUT, 6 } } },
Add support for sctp_generic_sendmsg() and sctp_generic_recvmsg().
2017-05-03 16:03:33 +00:00
{ .name = "sctp_generic_sendmsg", .ret_type = 1, .nargs = 7,
.args = { { Int, 0 }, { BinString | IN, 1 }, { Int, 2 },
Improve support for sctp_generic_recvmsg() and sctp_generic_sendmsg() and add support for sctp_generic_sendmsg_iov(). Handle the struct iovec argument and the struct sctp_sndrcvinfo arguments.
2018-01-14 14:33:22 +00:00
{ Sockaddr | IN, 3 }, { Socklent, 4 },
{ Sctpsndrcvinfo | IN, 5 }, { Msgflags, 6 } } },
{ .name = "sctp_generic_sendmsg_iov", .ret_type = 1, .nargs = 7,
.args = { { Int, 0 }, { Iovec | IN, 1 }, { Int, 2 },
{ Sockaddr | IN, 3 }, { Socklent, 4 },
{ Sctpsndrcvinfo | IN, 5 }, { Msgflags, 6 } } },
Add -c option to summarize number of calls, errors, and system time. Reviewed by: alfred
2009-05-12 20:42:12 +00:00
{ .name = "select", .ret_type = 1, .nargs = 5,
Whitespace fixes to consistently use spaces before }'s and wrap long lines.
2015-08-06 19:36:47 +00:00
.args = { { Int, 0 }, { Fd_set, 1 }, { Fd_set, 2 }, { Fd_set, 3 },
{ Timeval, 4 } } },
Add support for sendmsg() and recvmsg().
2017-05-03 12:26:16 +00:00
{ .name = "sendmsg", .ret_type = 1, .nargs = 3,
Decode msghdr argument of sendmsg() and recvmsg(). Sponsored by: Netflix, Inc.
2018-01-15 20:39:42 +00:00
.args = { { Int, 0 }, { Msghdr | IN, 1 }, { Msgflags, 2 } } },
Group the decoded system calls by ABI and sort the calls within each ABI. Reviewed by: bdrewery Glanced at by: kib Differential Revision: https://reviews.freebsd.org/D3823
2015-10-06 19:31:07 +00:00
{ .name = "sendto", .ret_type = 1, .nargs = 6,
Decode the fourth argument of sendto and recvfrom call.
2017-05-03 12:23:58 +00:00
.args = { { Int, 0 }, { BinString | IN, 1 }, { Sizet, 2 },
{ Msgflags, 3 }, { Sockaddr | IN, 4 },
{ Socklent | IN, 5 } } },
Add -c option to summarize number of calls, errors, and system time. Reviewed by: alfred
2009-05-12 20:42:12 +00:00
{ .name = "setitimer", .ret_type = 1, .nargs = 3,
Whitespace fix: remove some spurious spaces before commas.
2015-08-05 18:10:46 +00:00
.args = { { Int, 0 }, { Itimerval, 1 }, { Itimerval | OUT, 2 } } },
Decode arguments to getpriority() and setpriority().
2017-06-10 00:37:02 +00:00
{ .name = "setpriority", .ret_type = 1, .nargs = 3,
.args = { { Priowhich, 0 }, { Int, 1 }, { Int, 2 } } },
Group the decoded system calls by ABI and sort the calls within each ABI. Reviewed by: bdrewery Glanced at by: kib Differential Revision: https://reviews.freebsd.org/D3823
2015-10-06 19:31:07 +00:00
{ .name = "setrlimit", .ret_type = 1, .nargs = 2,
.args = { { Resource, 0 }, { Rlimit | IN, 1 } } },
Add support for [gs]etsockopt().
2017-05-03 12:18:09 +00:00
{ .name = "setsockopt", .ret_type = 1, .nargs = 5,
.args = { { Int, 0 }, { Sockoptlevel, 1 }, { Sockoptname, 2 },
{ Ptr | IN, 3 }, { Socklent, 4 } } },
Group the decoded system calls by ABI and sort the calls within each ABI. Reviewed by: bdrewery Glanced at by: kib Differential Revision: https://reviews.freebsd.org/D3823
2015-10-06 19:31:07 +00:00
{ .name = "shutdown", .ret_type = 1, .nargs = 2,
.args = { { Int, 0 }, { Shutdown, 1 } } },
{ .name = "sigaction", .ret_type = 1, .nargs = 3,
.args = { { Signal, 0 }, { Sigaction | IN, 1 },
{ Sigaction | OUT, 2 } } },
Several changes to truss. - Refactor the interface between the ABI-independent code and the ABI-specific backends. The backends now provide smaller hooks to fetch system call arguments and return values. The rest of the system call entry and exit handling that was previously duplicated among all the backends has been moved to one place. - Merge the loop when waiting for an event with the loop for handling stops. This also means not emulating a procfs-like interface on top of ptrace(). Instead, use a single event loop that fetches process events via waitid(). Among other things this allows us to report the full 32-bit exit value. - Use PT_FOLLOW_FORK to follow new child processes instead of forking a new truss process for each new child. This allows one truss process to monitor a tree of processes and truss -c should now display one total for the entire tree instead of separate summaries per process. - Use the recently added fields to ptrace_lwpinfo to determine the current system call number and argument count. The latter is especially useful and fixes a regression since the conversion from procfs. truss now generally prints the correct number of arguments for most system calls rather than printing extra arguments for any call not listed in the table in syscalls.c. - Actually check the new ABI when processes call exec. The comments claimed that this happened but it was not being done (perhaps this was another regression in the conversion to ptrace()). If the new ABI after exec is not supported, truss detaches from the process. If truss does not support the ABI for a newly executed process the process is killed before it returns from exec. - Along with the refactor, teach the various ABI-specific backends to fetch both return values, not just the first. Use this to properly report the full 64-bit return value from lseek(). In addition, the handler for "pipe" now pulls the pair of descriptors out of the return values (which is the true kernel system call interface) but displays them as an argument (which matches the interface exported by libc). - Each ABI handler adds entries to a linker set rather than requiring a statically defined table of handlers in main.c. - The arm and mips system call fetching code was changed to follow the same pattern as amd64 (and the in-kernel handler) of fetching register arguments first and then reading any remaining arguments from the stack. This should fix indirect system call arguments on at least arm. - The mipsn32 and n64 ABIs will now look for arguments in A4 through A7. - Use register %ebp for the 6th system call argument for Linux/i386 ABIs to match the in-kernel argument fetch code. - For powerpc binaries on a powerpc64 system, fetch the extra arguments on the stack as 32-bit values that are then copied into the 64-bit argument array instead of reading the 32-bit values directly into the 64-bit array. Reviewed by: kib (earlier version) Tested on: amd64 (FreeBSD/amd64 & i386), i386, arm (earlier version) Tested on: powerpc64 (FreeBSD/powerpc64 & powerpc) MFC after: 1 month Differential Revision: https://reviews.freebsd.org/D3575
2015-09-30 19:13:32 +00:00
{ .name = "sigpending", .ret_type = 1, .nargs = 1,
- Decode the arguments for several signal-related system calls: sigpending, sigqueue, sigreturn, sigsuspend, sigtimedwait, sigwait, sigwaitinfo, and thr_kill. - Print signal sets as a structure (with {}'s) and in particular use this to differentiate empty sets from a NULL pointer. - Decode arguments for some other system calls: issetugid, pipe2, sysarch (operations are only decoded for amd64 and i386), and thr_self.
2015-08-17 17:52:28 +00:00
.args = { { Sigset | OUT, 0 } } },
Several changes to truss. - Refactor the interface between the ABI-independent code and the ABI-specific backends. The backends now provide smaller hooks to fetch system call arguments and return values. The rest of the system call entry and exit handling that was previously duplicated among all the backends has been moved to one place. - Merge the loop when waiting for an event with the loop for handling stops. This also means not emulating a procfs-like interface on top of ptrace(). Instead, use a single event loop that fetches process events via waitid(). Among other things this allows us to report the full 32-bit exit value. - Use PT_FOLLOW_FORK to follow new child processes instead of forking a new truss process for each new child. This allows one truss process to monitor a tree of processes and truss -c should now display one total for the entire tree instead of separate summaries per process. - Use the recently added fields to ptrace_lwpinfo to determine the current system call number and argument count. The latter is especially useful and fixes a regression since the conversion from procfs. truss now generally prints the correct number of arguments for most system calls rather than printing extra arguments for any call not listed in the table in syscalls.c. - Actually check the new ABI when processes call exec. The comments claimed that this happened but it was not being done (perhaps this was another regression in the conversion to ptrace()). If the new ABI after exec is not supported, truss detaches from the process. If truss does not support the ABI for a newly executed process the process is killed before it returns from exec. - Along with the refactor, teach the various ABI-specific backends to fetch both return values, not just the first. Use this to properly report the full 64-bit return value from lseek(). In addition, the handler for "pipe" now pulls the pair of descriptors out of the return values (which is the true kernel system call interface) but displays them as an argument (which matches the interface exported by libc). - Each ABI handler adds entries to a linker set rather than requiring a statically defined table of handlers in main.c. - The arm and mips system call fetching code was changed to follow the same pattern as amd64 (and the in-kernel handler) of fetching register arguments first and then reading any remaining arguments from the stack. This should fix indirect system call arguments on at least arm. - The mipsn32 and n64 ABIs will now look for arguments in A4 through A7. - Use register %ebp for the 6th system call argument for Linux/i386 ABIs to match the in-kernel argument fetch code. - For powerpc binaries on a powerpc64 system, fetch the extra arguments on the stack as 32-bit values that are then copied into the 64-bit argument array instead of reading the 32-bit values directly into the 64-bit array. Reviewed by: kib (earlier version) Tested on: amd64 (FreeBSD/amd64 & i386), i386, arm (earlier version) Tested on: powerpc64 (FreeBSD/powerpc64 & powerpc) MFC after: 1 month Differential Revision: https://reviews.freebsd.org/D3575
2015-09-30 19:13:32 +00:00
{ .name = "sigprocmask", .ret_type = 1, .nargs = 3,
Add -c option to summarize number of calls, errors, and system time. Reviewed by: alfred
2009-05-12 20:42:12 +00:00
.args = { { Sigprocmask, 0 }, { Sigset, 1 }, { Sigset | OUT, 2 } } },
Several changes to truss. - Refactor the interface between the ABI-independent code and the ABI-specific backends. The backends now provide smaller hooks to fetch system call arguments and return values. The rest of the system call entry and exit handling that was previously duplicated among all the backends has been moved to one place. - Merge the loop when waiting for an event with the loop for handling stops. This also means not emulating a procfs-like interface on top of ptrace(). Instead, use a single event loop that fetches process events via waitid(). Among other things this allows us to report the full 32-bit exit value. - Use PT_FOLLOW_FORK to follow new child processes instead of forking a new truss process for each new child. This allows one truss process to monitor a tree of processes and truss -c should now display one total for the entire tree instead of separate summaries per process. - Use the recently added fields to ptrace_lwpinfo to determine the current system call number and argument count. The latter is especially useful and fixes a regression since the conversion from procfs. truss now generally prints the correct number of arguments for most system calls rather than printing extra arguments for any call not listed in the table in syscalls.c. - Actually check the new ABI when processes call exec. The comments claimed that this happened but it was not being done (perhaps this was another regression in the conversion to ptrace()). If the new ABI after exec is not supported, truss detaches from the process. If truss does not support the ABI for a newly executed process the process is killed before it returns from exec. - Along with the refactor, teach the various ABI-specific backends to fetch both return values, not just the first. Use this to properly report the full 64-bit return value from lseek(). In addition, the handler for "pipe" now pulls the pair of descriptors out of the return values (which is the true kernel system call interface) but displays them as an argument (which matches the interface exported by libc). - Each ABI handler adds entries to a linker set rather than requiring a statically defined table of handlers in main.c. - The arm and mips system call fetching code was changed to follow the same pattern as amd64 (and the in-kernel handler) of fetching register arguments first and then reading any remaining arguments from the stack. This should fix indirect system call arguments on at least arm. - The mipsn32 and n64 ABIs will now look for arguments in A4 through A7. - Use register %ebp for the 6th system call argument for Linux/i386 ABIs to match the in-kernel argument fetch code. - For powerpc binaries on a powerpc64 system, fetch the extra arguments on the stack as 32-bit values that are then copied into the 64-bit argument array instead of reading the 32-bit values directly into the 64-bit array. Reviewed by: kib (earlier version) Tested on: amd64 (FreeBSD/amd64 & i386), i386, arm (earlier version) Tested on: powerpc64 (FreeBSD/powerpc64 & powerpc) MFC after: 1 month Differential Revision: https://reviews.freebsd.org/D3575
2015-09-30 19:13:32 +00:00
{ .name = "sigqueue", .ret_type = 1, .nargs = 3,
- Decode the arguments for several signal-related system calls: sigpending, sigqueue, sigreturn, sigsuspend, sigtimedwait, sigwait, sigwaitinfo, and thr_kill. - Print signal sets as a structure (with {}'s) and in particular use this to differentiate empty sets from a NULL pointer. - Decode arguments for some other system calls: issetugid, pipe2, sysarch (operations are only decoded for amd64 and i386), and thr_self.
2015-08-17 17:52:28 +00:00
.args = { { Int, 0 }, { Signal, 1 }, { LongHex, 2 } } },
Several changes to truss. - Refactor the interface between the ABI-independent code and the ABI-specific backends. The backends now provide smaller hooks to fetch system call arguments and return values. The rest of the system call entry and exit handling that was previously duplicated among all the backends has been moved to one place. - Merge the loop when waiting for an event with the loop for handling stops. This also means not emulating a procfs-like interface on top of ptrace(). Instead, use a single event loop that fetches process events via waitid(). Among other things this allows us to report the full 32-bit exit value. - Use PT_FOLLOW_FORK to follow new child processes instead of forking a new truss process for each new child. This allows one truss process to monitor a tree of processes and truss -c should now display one total for the entire tree instead of separate summaries per process. - Use the recently added fields to ptrace_lwpinfo to determine the current system call number and argument count. The latter is especially useful and fixes a regression since the conversion from procfs. truss now generally prints the correct number of arguments for most system calls rather than printing extra arguments for any call not listed in the table in syscalls.c. - Actually check the new ABI when processes call exec. The comments claimed that this happened but it was not being done (perhaps this was another regression in the conversion to ptrace()). If the new ABI after exec is not supported, truss detaches from the process. If truss does not support the ABI for a newly executed process the process is killed before it returns from exec. - Along with the refactor, teach the various ABI-specific backends to fetch both return values, not just the first. Use this to properly report the full 64-bit return value from lseek(). In addition, the handler for "pipe" now pulls the pair of descriptors out of the return values (which is the true kernel system call interface) but displays them as an argument (which matches the interface exported by libc). - Each ABI handler adds entries to a linker set rather than requiring a statically defined table of handlers in main.c. - The arm and mips system call fetching code was changed to follow the same pattern as amd64 (and the in-kernel handler) of fetching register arguments first and then reading any remaining arguments from the stack. This should fix indirect system call arguments on at least arm. - The mipsn32 and n64 ABIs will now look for arguments in A4 through A7. - Use register %ebp for the 6th system call argument for Linux/i386 ABIs to match the in-kernel argument fetch code. - For powerpc binaries on a powerpc64 system, fetch the extra arguments on the stack as 32-bit values that are then copied into the 64-bit argument array instead of reading the 32-bit values directly into the 64-bit array. Reviewed by: kib (earlier version) Tested on: amd64 (FreeBSD/amd64 & i386), i386, arm (earlier version) Tested on: powerpc64 (FreeBSD/powerpc64 & powerpc) MFC after: 1 month Differential Revision: https://reviews.freebsd.org/D3575
2015-09-30 19:13:32 +00:00
{ .name = "sigreturn", .ret_type = 1, .nargs = 1,
- Decode the arguments for several signal-related system calls: sigpending, sigqueue, sigreturn, sigsuspend, sigtimedwait, sigwait, sigwaitinfo, and thr_kill. - Print signal sets as a structure (with {}'s) and in particular use this to differentiate empty sets from a NULL pointer. - Decode arguments for some other system calls: issetugid, pipe2, sysarch (operations are only decoded for amd64 and i386), and thr_self.
2015-08-17 17:52:28 +00:00
.args = { { Ptr, 0 } } },
Several changes to truss. - Refactor the interface between the ABI-independent code and the ABI-specific backends. The backends now provide smaller hooks to fetch system call arguments and return values. The rest of the system call entry and exit handling that was previously duplicated among all the backends has been moved to one place. - Merge the loop when waiting for an event with the loop for handling stops. This also means not emulating a procfs-like interface on top of ptrace(). Instead, use a single event loop that fetches process events via waitid(). Among other things this allows us to report the full 32-bit exit value. - Use PT_FOLLOW_FORK to follow new child processes instead of forking a new truss process for each new child. This allows one truss process to monitor a tree of processes and truss -c should now display one total for the entire tree instead of separate summaries per process. - Use the recently added fields to ptrace_lwpinfo to determine the current system call number and argument count. The latter is especially useful and fixes a regression since the conversion from procfs. truss now generally prints the correct number of arguments for most system calls rather than printing extra arguments for any call not listed in the table in syscalls.c. - Actually check the new ABI when processes call exec. The comments claimed that this happened but it was not being done (perhaps this was another regression in the conversion to ptrace()). If the new ABI after exec is not supported, truss detaches from the process. If truss does not support the ABI for a newly executed process the process is killed before it returns from exec. - Along with the refactor, teach the various ABI-specific backends to fetch both return values, not just the first. Use this to properly report the full 64-bit return value from lseek(). In addition, the handler for "pipe" now pulls the pair of descriptors out of the return values (which is the true kernel system call interface) but displays them as an argument (which matches the interface exported by libc). - Each ABI handler adds entries to a linker set rather than requiring a statically defined table of handlers in main.c. - The arm and mips system call fetching code was changed to follow the same pattern as amd64 (and the in-kernel handler) of fetching register arguments first and then reading any remaining arguments from the stack. This should fix indirect system call arguments on at least arm. - The mipsn32 and n64 ABIs will now look for arguments in A4 through A7. - Use register %ebp for the 6th system call argument for Linux/i386 ABIs to match the in-kernel argument fetch code. - For powerpc binaries on a powerpc64 system, fetch the extra arguments on the stack as 32-bit values that are then copied into the 64-bit argument array instead of reading the 32-bit values directly into the 64-bit array. Reviewed by: kib (earlier version) Tested on: amd64 (FreeBSD/amd64 & i386), i386, arm (earlier version) Tested on: powerpc64 (FreeBSD/powerpc64 & powerpc) MFC after: 1 month Differential Revision: https://reviews.freebsd.org/D3575
2015-09-30 19:13:32 +00:00
{ .name = "sigsuspend", .ret_type = 1, .nargs = 1,
- Decode the arguments for several signal-related system calls: sigpending, sigqueue, sigreturn, sigsuspend, sigtimedwait, sigwait, sigwaitinfo, and thr_kill. - Print signal sets as a structure (with {}'s) and in particular use this to differentiate empty sets from a NULL pointer. - Decode arguments for some other system calls: issetugid, pipe2, sysarch (operations are only decoded for amd64 and i386), and thr_self.
2015-08-17 17:52:28 +00:00
.args = { { Sigset | IN, 0 } } },
{ .name = "sigtimedwait", .ret_type = 1, .nargs = 3,
Decode signal information returned by system calls. Specifically, decode the siginfo structure returned by sigtimedwait(), sigwaitinfo(), and wait6(). While here, also decode the signal number returned in the second argument to sigwait().
2017-08-30 15:45:23 +00:00
.args = { { Sigset | IN, 0 }, { Siginfo | OUT, 1 },
{ Timespec | IN, 2 } } },
- Decode the arguments for several signal-related system calls: sigpending, sigqueue, sigreturn, sigsuspend, sigtimedwait, sigwait, sigwaitinfo, and thr_kill. - Print signal sets as a structure (with {}'s) and in particular use this to differentiate empty sets from a NULL pointer. - Decode arguments for some other system calls: issetugid, pipe2, sysarch (operations are only decoded for amd64 and i386), and thr_self.
2015-08-17 17:52:28 +00:00
{ .name = "sigwait", .ret_type = 1, .nargs = 2,
Decode signal information returned by system calls. Specifically, decode the siginfo structure returned by sigtimedwait(), sigwaitinfo(), and wait6(). While here, also decode the signal number returned in the second argument to sigwait().
2017-08-30 15:45:23 +00:00
.args = { { Sigset | IN, 0 }, { PSig | OUT, 1 } } },
- Decode the arguments for several signal-related system calls: sigpending, sigqueue, sigreturn, sigsuspend, sigtimedwait, sigwait, sigwaitinfo, and thr_kill. - Print signal sets as a structure (with {}'s) and in particular use this to differentiate empty sets from a NULL pointer. - Decode arguments for some other system calls: issetugid, pipe2, sysarch (operations are only decoded for amd64 and i386), and thr_self.
2015-08-17 17:52:28 +00:00
{ .name = "sigwaitinfo", .ret_type = 1, .nargs = 2,
Decode signal information returned by system calls. Specifically, decode the siginfo structure returned by sigtimedwait(), sigwaitinfo(), and wait6(). While here, also decode the signal number returned in the second argument to sigwait().
2017-08-30 15:45:23 +00:00
.args = { { Sigset | IN, 0 }, { Siginfo | OUT, 1 } } },
Add -c option to summarize number of calls, errors, and system time. Reviewed by: alfred
2009-05-12 20:42:12 +00:00
{ .name = "socket", .ret_type = 1, .nargs = 3,
Decode the third argument of socket().
2017-05-03 09:23:13 +00:00
.args = { { Sockdomain, 0 }, { Socktype, 1 }, { Sockprotocol, 2 } } },
Group the decoded system calls by ABI and sort the calls within each ABI. Reviewed by: bdrewery Glanced at by: kib Differential Revision: https://reviews.freebsd.org/D3823
2015-10-06 19:31:07 +00:00
{ .name = "stat", .ret_type = 1, .nargs = 2,
.args = { { Name | IN, 0 }, { Stat | OUT, 1 } } },
{ .name = "statfs", .ret_type = 1, .nargs = 2,
.args = { { Name | IN, 0 }, { StatFs | OUT, 1 } } },
Add -c option to summarize number of calls, errors, and system time. Reviewed by: alfred
2009-05-12 20:42:12 +00:00
{ .name = "symlink", .ret_type = 1, .nargs = 2,
Whitespace fix: remove some spurious spaces before commas.
2015-08-05 18:10:46 +00:00
.args = { { Name, 0 }, { Name, 1 } } },
Decode the arguments passed to the *at() family of system calls. This is especially useful now that libc's open() always calls openat(). While here, fix a few other things: - Decode the mode argument passed to access(), eaccess(), and faccessat(). - Decode the atfd paramete to pretty-print AT_FDCWD. - Decode the special AT_* flags used with some of the *at() system calls. - Decode arguments for fchmod(), lchmod(), fchown(), lchown(), eaccess(), and futimens(). - Decode both of the timeval structures passed to futimes() instead of just the first one.
2015-08-06 19:08:33 +00:00
{ .name = "symlinkat", .ret_type = 1, .nargs = 3,
.args = { { Name, 0 }, { Atfd, 1 }, { Name, 2 } } },
Group the decoded system calls by ABI and sort the calls within each ABI. Reviewed by: bdrewery Glanced at by: kib Differential Revision: https://reviews.freebsd.org/D3823
2015-10-06 19:31:07 +00:00
{ .name = "sysarch", .ret_type = 1, .nargs = 2,
.args = { { Sysarch, 0 }, { Ptr, 1 } } },
{ .name = "thr_kill", .ret_type = 1, .nargs = 2,
.args = { { Long, 0 }, { Signal, 1 } } },
{ .name = "thr_self", .ret_type = 1, .nargs = 1,
.args = { { Ptr, 0 } } },
Decode arguments passed to thr_set_name(). MFC after: 1 month
2017-08-25 19:06:36 +00:00
{ .name = "thr_set_name", .ret_type = 1, .nargs = 2,
.args = { { Long, 0 }, { Name, 1 } } },
Group the decoded system calls by ABI and sort the calls within each ABI. Reviewed by: bdrewery Glanced at by: kib Differential Revision: https://reviews.freebsd.org/D3823
2015-10-06 19:31:07 +00:00
{ .name = "truncate", .ret_type = 1, .nargs = 2,
Automate the handling of QUAD_ALIGN and QUAD_SLOTS. Previously, the offset in a system call description specified the array index of the start of a system call argument. For most system call arguments this was the same as the index of the argument in the function signature. 64-bit arguments (off_t and id_t values) passed on 32-bit platforms use two slots in the array however. This was handled by adding (QUAD_SLOTS - 1) to the slot indicies of any subsequent arguments after a 64-bit argument (though written as ("{ Quad, 1 }, { Int, 1 + QUAD_SLOTS }" rather than "{ Quad, 1 }, { Int, 2 + QUAD_SLOTS - 1 }"). If a system call contained multiple 64-bit arguments (such as posix_fadvise()), then additional arguments would need to use 'QUAD_SLOTS * 2' but remember to subtract 2 from the initial number, etc. In addition, 32-bit powerpc requires 64-bit arguments to be 64-bit aligned, so if the effective index in the array of a 64-bit argument is odd, it needs QUAD_ALIGN added to the current and any subsequent slots. However, if the effective index in the array of a 64-bit argument was even, QUAD_ALIGN was omitted. This approach was messy and error prone. This commit replaces it with automated pre-processing of the system call table to do fixups for 64-bit argument offsets. The offset in a system call description now indicates the index of an argument in the associated function call's signature. A fixup function is run against each decoded system call description during startup on 32-bit platforms. The fixup function maintains an 'offset' value which holds an offset to be added to each remaining system call argument's index. Initially offset is 0. When a 64-bit system call argument is encountered, the offset is first aligned to a 64-bit boundary (only on powerpc) and then incremented to account for the second argument slot used by the argument. This modified 'offset' is then applied to any remaining arguments. This approach does require a few things that were not previously required: 1) Each system call description must now list arguments in ascending order (existing ones all do) without using duplicate slots in the register array. A new assert() should catch any future descriptions which violate this rule. 2) A system call description is still permitted to omit arguments (though none currently do), but if the call accepts 64-bit arguments those cannot be omitted or incorrect results will be displated on 32-bit systems. Tested on: amd64 and i386
2017-03-15 23:08:11 +00:00
.args = { { Name | IN, 0 }, { QuadHex | IN, 1 } } },
Group the decoded system calls by ABI and sort the calls within each ABI. Reviewed by: bdrewery Glanced at by: kib Differential Revision: https://reviews.freebsd.org/D3823
2015-10-06 19:31:07 +00:00
#if 0
/* Does not exist */
{ .name = "umount", .ret_type = 1, .nargs = 2,
.args = { { Name, 0 }, { Int, 2 } } },
#endif
{ .name = "unlink", .ret_type = 1, .nargs = 1,
.args = { { Name, 0 } } },
{ .name = "unlinkat", .ret_type = 1, .nargs = 3,
.args = { { Atfd, 0 }, { Name, 1 }, { Atflags, 2 } } },
{ .name = "unmount", .ret_type = 1, .nargs = 2,
Decode flags passed to mount(), nmount(), and unmount().
2017-06-08 08:07:51 +00:00
.args = { { Name, 0 }, { Mountflags, 1 } } },
Group the decoded system calls by ABI and sort the calls within each ABI. Reviewed by: bdrewery Glanced at by: kib Differential Revision: https://reviews.freebsd.org/D3823
2015-10-06 19:31:07 +00:00
{ .name = "utimensat", .ret_type = 1, .nargs = 4,
.args = { { Atfd, 0 }, { Name | IN, 1 }, { Timespec2 | IN, 2 },
{ Atflags, 3 } } },
{ .name = "utimes", .ret_type = 1, .nargs = 2,
.args = { { Name | IN, 0 }, { Timeval2 | IN, 1 } } },
truss: Add support for utrace(2). This uses the kdump(1) utrace support code directly until a common library is created. This allows malloc(3) tracing with MALLOC_CONF=utrace:true and rtld tracing with LD_UTRACE=1. Unknown utrace(2) data is just printed as hex. PR: 43819 [inspired by] Reviewed by: jhb MFC after: 2 weeks Relnotes: yes Differential Revision: https://reviews.freebsd.org/D3819
2015-10-06 21:58:38 +00:00
{ .name = "utrace", .ret_type = 1, .nargs = 1,
.args = { { Utrace, 0 } } },
- Decode the idtype argument passed to wait6() in kdump and truss. - Don't treat an options argument of 0 to wait4() as an error in kdump. - Decode the wait options passed to wait4() and wait6() in truss and decode the returned rusage and exit status. Approved by: re (kib) MFC after: 1 week
2013-09-12 18:08:25 +00:00
{ .name = "wait4", .ret_type = 1, .nargs = 4,
.args = { { Int, 0 }, { ExitStatus | OUT, 1 }, { Waitoptions, 2 },
{ Rusage | OUT, 3 } } },
{ .name = "wait6", .ret_type = 1, .nargs = 6,
Automate the handling of QUAD_ALIGN and QUAD_SLOTS. Previously, the offset in a system call description specified the array index of the start of a system call argument. For most system call arguments this was the same as the index of the argument in the function signature. 64-bit arguments (off_t and id_t values) passed on 32-bit platforms use two slots in the array however. This was handled by adding (QUAD_SLOTS - 1) to the slot indicies of any subsequent arguments after a 64-bit argument (though written as ("{ Quad, 1 }, { Int, 1 + QUAD_SLOTS }" rather than "{ Quad, 1 }, { Int, 2 + QUAD_SLOTS - 1 }"). If a system call contained multiple 64-bit arguments (such as posix_fadvise()), then additional arguments would need to use 'QUAD_SLOTS * 2' but remember to subtract 2 from the initial number, etc. In addition, 32-bit powerpc requires 64-bit arguments to be 64-bit aligned, so if the effective index in the array of a 64-bit argument is odd, it needs QUAD_ALIGN added to the current and any subsequent slots. However, if the effective index in the array of a 64-bit argument was even, QUAD_ALIGN was omitted. This approach was messy and error prone. This commit replaces it with automated pre-processing of the system call table to do fixups for 64-bit argument offsets. The offset in a system call description now indicates the index of an argument in the associated function call's signature. A fixup function is run against each decoded system call description during startup on 32-bit platforms. The fixup function maintains an 'offset' value which holds an offset to be added to each remaining system call argument's index. Initially offset is 0. When a 64-bit system call argument is encountered, the offset is first aligned to a 64-bit boundary (only on powerpc) and then incremented to account for the second argument slot used by the argument. This modified 'offset' is then applied to any remaining arguments. This approach does require a few things that were not previously required: 1) Each system call description must now list arguments in ascending order (existing ones all do) without using duplicate slots in the register array. A new assert() should catch any future descriptions which violate this rule. 2) A system call description is still permitted to omit arguments (though none currently do), but if the call accepts 64-bit arguments those cannot be omitted or incorrect results will be displated on 32-bit systems. Tested on: amd64 and i386
2017-03-15 23:08:11 +00:00
.args = { { Idtype, 0 }, { Quad, 1 }, { ExitStatus | OUT, 2 },
Decode signal information returned by system calls. Specifically, decode the siginfo structure returned by sigtimedwait(), sigwaitinfo(), and wait6(). While here, also decode the signal number returned in the second argument to sigwait().
2017-08-30 15:45:23 +00:00
{ Waitoptions, 3 }, { Rusage | OUT, 4 },
{ Siginfo | OUT, 5 } } },
Group the decoded system calls by ABI and sort the calls within each ABI. Reviewed by: bdrewery Glanced at by: kib Differential Revision: https://reviews.freebsd.org/D3823
2015-10-06 19:31:07 +00:00
{ .name = "write", .ret_type = 1, .nargs = 3,
Add a Sizet type for 'size_t' values and use it instead of Int. Various size_t arguments were previously decoded as Int values instead which would have truncated values above 2^31 on 64-bit systems.
2017-03-19 00:27:07 +00:00
.args = { { Int, 0 }, { BinString | IN, 1 }, { Sizet, 2 } } },
Add support for readv() and writev() to truss. Sponsored by: Netflix, Inc.
2018-01-13 13:59:35 +00:00
{ .name = "writev", .ret_type = 1, .nargs = 3,
Mark the iovec parameters of writev() and readv() as IN and OUT. This makes truss work on readv() as expected.
2018-01-14 09:51:25 +00:00
.args = { { Int, 0 }, { Iovec | IN, 1 }, { Int, 2 } } },
Group the decoded system calls by ABI and sort the calls within each ABI. Reviewed by: bdrewery Glanced at by: kib Differential Revision: https://reviews.freebsd.org/D3823
2015-10-06 19:31:07 +00:00
/* Linux ABI */
{ .name = "linux_access", .ret_type = 1, .nargs = 2,
.args = { { Name, 0 }, { Accessmode, 1 } } },
{ .name = "linux_execve", .ret_type = 1, .nargs = 3,
.args = { { Name | IN, 0 }, { ExecArgs | IN, 1 },
{ ExecEnv | IN, 2 } } },
{ .name = "linux_lseek", .ret_type = 2, .nargs = 3,
.args = { { Int, 0 }, { Int, 1 }, { Whence, 2 } } },
{ .name = "linux_mkdir", .ret_type = 1, .nargs = 2,
.args = { { Name | IN, 0 }, { Int, 1 } } },
{ .name = "linux_newfstat", .ret_type = 1, .nargs = 2,
.args = { { Int, 0 }, { Ptr | OUT, 1 } } },
{ .name = "linux_newstat", .ret_type = 1, .nargs = 2,
.args = { { Name | IN, 0 }, { Ptr | OUT, 1 } } },
{ .name = "linux_open", .ret_type = 1, .nargs = 3,
.args = { { Name, 0 }, { Hex, 1 }, { Octal, 2 } } },
{ .name = "linux_readlink", .ret_type = 1, .nargs = 3,
Add a Sizet type for 'size_t' values and use it instead of Int. Various size_t arguments were previously decoded as Int values instead which would have truncated values above 2^31 on 64-bit systems.
2017-03-19 00:27:07 +00:00
.args = { { Name, 0 }, { Name | OUT, 1 }, { Sizet, 2 } } },
Group the decoded system calls by ABI and sort the calls within each ABI. Reviewed by: bdrewery Glanced at by: kib Differential Revision: https://reviews.freebsd.org/D3823
2015-10-06 19:31:07 +00:00
{ .name = "linux_socketcall", .ret_type = 1, .nargs = 2,
.args = { { Int, 0 }, { LinuxSockArgs, 1 } } },
Remove duplicate argument from linux_stat64() decoding.
2017-03-15 22:39:15 +00:00
{ .name = "linux_stat64", .ret_type = 1, .nargs = 2,
.args = { { Name | IN, 0 }, { Ptr | OUT, 1 } } },
Group the decoded system calls by ABI and sort the calls within each ABI. Reviewed by: bdrewery Glanced at by: kib Differential Revision: https://reviews.freebsd.org/D3823
2015-10-06 19:31:07 +00:00
Properly format pointer size independent CloudABI system calls. CloudABI has approximately 50 system calls that do not depend on the pointer size of the system. As the ABI is pretty compact, it takes little effort to each truss(8) the formatting rules for these system calls. Start off by formatting pointer size independent system calls. Changes: - Make it possible to include the CloudABI system call definitions in FreeBSD userspace builds. Add ${root}/sys to the truss(8) Makefile so we can pull in <compat/cloudabi/cloudabi_syscalldefs.h>. - Refactoring: patch up amd64-cloudabi64.c to use the CLOUDABI_* constants instead of rolling our own table. - Add table entries for all of the system calls. - Add new generic formatting types (UInt, IntArray) that we'll be using to format unsigned integers and arrays of integers. - Add CloudABI specific formatting types. Approved by: jhb Differential Revision: https://reviews.freebsd.org/D3836
2015-10-08 05:27:45 +00:00
/* CloudABI system calls. */
{ .name = "cloudabi_sys_clock_res_get", .ret_type = 1, .nargs = 1,
.args = { { CloudABIClockID, 0 } } },
{ .name = "cloudabi_sys_clock_time_get", .ret_type = 1, .nargs = 2,
.args = { { CloudABIClockID, 0 }, { CloudABITimestamp, 1 } } },
{ .name = "cloudabi_sys_condvar_signal", .ret_type = 1, .nargs = 3,
.args = { { Ptr, 0 }, { CloudABIMFlags, 1 }, { UInt, 2 } } },
{ .name = "cloudabi_sys_fd_close", .ret_type = 1, .nargs = 1,
.args = { { Int, 0 } } },
{ .name = "cloudabi_sys_fd_create1", .ret_type = 1, .nargs = 1,
.args = { { CloudABIFileType, 0 } } },
{ .name = "cloudabi_sys_fd_create2", .ret_type = 1, .nargs = 2,
.args = { { CloudABIFileType, 0 }, { PipeFds | OUT, 0 } } },
{ .name = "cloudabi_sys_fd_datasync", .ret_type = 1, .nargs = 1,
.args = { { Int, 0 } } },
{ .name = "cloudabi_sys_fd_dup", .ret_type = 1, .nargs = 1,
.args = { { Int, 0 } } },
{ .name = "cloudabi_sys_fd_replace", .ret_type = 1, .nargs = 2,
.args = { { Int, 0 }, { Int, 1 } } },
{ .name = "cloudabi_sys_fd_seek", .ret_type = 1, .nargs = 3,
.args = { { Int, 0 }, { Int, 1 }, { CloudABIWhence, 2 } } },
{ .name = "cloudabi_sys_fd_stat_get", .ret_type = 1, .nargs = 2,
.args = { { Int, 0 }, { CloudABIFDStat | OUT, 1 } } },
{ .name = "cloudabi_sys_fd_stat_put", .ret_type = 1, .nargs = 3,
.args = { { Int, 0 }, { CloudABIFDStat | IN, 1 },
Rename ClouduABIFDSFlags to CloudABIFDSFlags.
2018-03-29 00:04:07 +00:00
{ CloudABIFDSFlags, 2 } } },
Properly format pointer size independent CloudABI system calls. CloudABI has approximately 50 system calls that do not depend on the pointer size of the system. As the ABI is pretty compact, it takes little effort to each truss(8) the formatting rules for these system calls. Start off by formatting pointer size independent system calls. Changes: - Make it possible to include the CloudABI system call definitions in FreeBSD userspace builds. Add ${root}/sys to the truss(8) Makefile so we can pull in <compat/cloudabi/cloudabi_syscalldefs.h>. - Refactoring: patch up amd64-cloudabi64.c to use the CLOUDABI_* constants instead of rolling our own table. - Add table entries for all of the system calls. - Add new generic formatting types (UInt, IntArray) that we'll be using to format unsigned integers and arrays of integers. - Add CloudABI specific formatting types. Approved by: jhb Differential Revision: https://reviews.freebsd.org/D3836
2015-10-08 05:27:45 +00:00
{ .name = "cloudabi_sys_fd_sync", .ret_type = 1, .nargs = 1,
.args = { { Int, 0 } } },
{ .name = "cloudabi_sys_file_advise", .ret_type = 1, .nargs = 4,
.args = { { Int, 0 }, { Int, 1 }, { Int, 2 },
{ CloudABIAdvice, 3 } } },
{ .name = "cloudabi_sys_file_allocate", .ret_type = 1, .nargs = 3,
.args = { { Int, 0 }, { Int, 1 }, { Int, 2 } } },
{ .name = "cloudabi_sys_file_create", .ret_type = 1, .nargs = 3,
.args = { { Int, 0 }, { BinString | IN, 1 },
{ CloudABIFileType, 3 } } },
{ .name = "cloudabi_sys_file_link", .ret_type = 1, .nargs = 4,
.args = { { CloudABILookup, 0 }, { BinString | IN, 1 },
{ Int, 3 }, { BinString | IN, 4 } } },
{ .name = "cloudabi_sys_file_open", .ret_type = 1, .nargs = 4,
.args = { { Int, 0 }, { BinString | IN, 1 },
{ CloudABIOFlags, 3 }, { CloudABIFDStat | IN, 4 } } },
{ .name = "cloudabi_sys_file_readdir", .ret_type = 1, .nargs = 4,
.args = { { Int, 0 }, { BinString | OUT, 1 }, { Int, 2 },
{ Int, 3 } } },
{ .name = "cloudabi_sys_file_readlink", .ret_type = 1, .nargs = 4,
.args = { { Int, 0 }, { BinString | IN, 1 },
{ BinString | OUT, 3 }, { Int, 4 } } },
{ .name = "cloudabi_sys_file_rename", .ret_type = 1, .nargs = 4,
.args = { { Int, 0 }, { BinString | IN, 1 },
{ Int, 3 }, { BinString | IN, 4 } } },
{ .name = "cloudabi_sys_file_stat_fget", .ret_type = 1, .nargs = 2,
.args = { { Int, 0 }, { CloudABIFileStat | OUT, 1 } } },
{ .name = "cloudabi_sys_file_stat_fput", .ret_type = 1, .nargs = 3,
.args = { { Int, 0 }, { CloudABIFileStat | IN, 1 },
{ CloudABIFSFlags, 2 } } },
{ .name = "cloudabi_sys_file_stat_get", .ret_type = 1, .nargs = 3,
.args = { { CloudABILookup, 0 }, { BinString | IN, 1 },
{ CloudABIFileStat | OUT, 3 } } },
{ .name = "cloudabi_sys_file_stat_put", .ret_type = 1, .nargs = 4,
.args = { { CloudABILookup, 0 }, { BinString | IN, 1 },
{ CloudABIFileStat | IN, 3 }, { CloudABIFSFlags, 4 } } },
{ .name = "cloudabi_sys_file_symlink", .ret_type = 1, .nargs = 3,
.args = { { BinString | IN, 0 },
{ Int, 2 }, { BinString | IN, 3 } } },
{ .name = "cloudabi_sys_file_unlink", .ret_type = 1, .nargs = 3,
.args = { { Int, 0 }, { BinString | IN, 1 },
{ CloudABIULFlags, 3 } } },
{ .name = "cloudabi_sys_lock_unlock", .ret_type = 1, .nargs = 2,
.args = { { Ptr, 0 }, { CloudABIMFlags, 1 } } },
{ .name = "cloudabi_sys_mem_advise", .ret_type = 1, .nargs = 3,
.args = { { Ptr, 0 }, { Int, 1 }, { CloudABIAdvice, 2 } } },
{ .name = "cloudabi_sys_mem_map", .ret_type = 1, .nargs = 6,
.args = { { Ptr, 0 }, { Int, 1 }, { CloudABIMProt, 2 },
{ CloudABIMFlags, 3 }, { Int, 4 }, { Int, 5 } } },
{ .name = "cloudabi_sys_mem_protect", .ret_type = 1, .nargs = 3,
.args = { { Ptr, 0 }, { Int, 1 }, { CloudABIMProt, 2 } } },
{ .name = "cloudabi_sys_mem_sync", .ret_type = 1, .nargs = 3,
.args = { { Ptr, 0 }, { Int, 1 }, { CloudABIMSFlags, 2 } } },
{ .name = "cloudabi_sys_mem_unmap", .ret_type = 1, .nargs = 2,
.args = { { Ptr, 0 }, { Int, 1 } } },
{ .name = "cloudabi_sys_proc_exec", .ret_type = 1, .nargs = 5,
.args = { { Int, 0 }, { BinString | IN, 1 }, { Int, 2 },
{ IntArray, 3 }, { Int, 4 } } },
{ .name = "cloudabi_sys_proc_exit", .ret_type = 1, .nargs = 1,
.args = { { Int, 0 } } },
{ .name = "cloudabi_sys_proc_fork", .ret_type = 1, .nargs = 0 },
{ .name = "cloudabi_sys_proc_raise", .ret_type = 1, .nargs = 1,
.args = { { CloudABISignal, 0 } } },
{ .name = "cloudabi_sys_random_get", .ret_type = 1, .nargs = 2,
.args = { { BinString | OUT, 0 }, { Int, 1 } } },
{ .name = "cloudabi_sys_sock_shutdown", .ret_type = 1, .nargs = 2,
.args = { { Int, 0 }, { CloudABISDFlags, 1 } } },
{ .name = "cloudabi_sys_thread_exit", .ret_type = 1, .nargs = 2,
.args = { { Ptr, 0 }, { CloudABIMFlags, 1 } } },
{ .name = "cloudabi_sys_thread_yield", .ret_type = 1, .nargs = 0 },
Add -c option to summarize number of calls, errors, and system time. Reviewed by: alfred
2009-05-12 20:42:12 +00:00
{ .name = 0 },
Truss program. Requires procfs.
1997-12-06 05:23:12 +00:00
};
Fix tracking of unknown syscalls for 'truss -c'. This is done by changing get_syscall() to either lookup the known syscall or add it into the list with the default handlers for printing. This also simplifies some code to not have to check if the syscall variable is set or NULL. Reviewed by: jhb Relnotes: yes Differential Revision: https://reviews.freebsd.org/D3792
2015-10-05 18:08:35 +00:00
static STAILQ_HEAD(, syscall) syscalls;
Truss program. Requires procfs.
1997-12-06 05:23:12 +00:00
- Add decoding of kse_release, kevent, sigprocmask, unmount, socket, getrusage, rename, __getcwd, shutdown, getrlimit, setrlimit, _umtx_lock, _umtx_unlock, pathconf, truncate, ftruncate, kill - Decode more arguments of open, mprot, *stat, and fcntl. - Convert all constant-macro and bitfield decoding to lookup tables; much cleaner than previous code. - Print the timestamp of process exit and signal reception when -d or -D are in use - Try six times with 1/2 second delay to debug the child PR: bin/52190 (updated) Submitted by: Dan Nelson <dnelson@allantgroup.com> Approved by: alfred
2006-05-15 21:18:28 +00:00
/* Xlat idea taken from strace */
struct xlat {
int val;
Make use of ptrace(2) instead of procfs in truss(1), eliminating yet another need of an available /proc/ mount. Tested with: make universe Submitted by: howardsu Reviewed by: alfred
2007-04-10 04:03:34 +00:00
const char *str;
- Add decoding of kse_release, kevent, sigprocmask, unmount, socket, getrusage, rename, __getcwd, shutdown, getrlimit, setrlimit, _umtx_lock, _umtx_unlock, pathconf, truncate, ftruncate, kill - Decode more arguments of open, mprot, *stat, and fcntl. - Convert all constant-macro and bitfield decoding to lookup tables; much cleaner than previous code. - Print the timestamp of process exit and signal reception when -d or -D are in use - Try six times with 1/2 second delay to debug the child PR: bin/52190 (updated) Submitted by: Dan Nelson <dnelson@allantgroup.com> Approved by: alfred
2006-05-15 21:18:28 +00:00
};
- Style(9) cleanup. Approved by: kib (mentor)
2012-09-02 11:03:18 +00:00
#define X(a) { a, #a },
#define XEND { 0, NULL }
- Add decoding of kse_release, kevent, sigprocmask, unmount, socket, getrusage, rename, __getcwd, shutdown, getrlimit, setrlimit, _umtx_lock, _umtx_unlock, pathconf, truncate, ftruncate, kill - Decode more arguments of open, mprot, *stat, and fcntl. - Convert all constant-macro and bitfield decoding to lookup tables; much cleaner than previous code. - Print the timestamp of process exit and signal reception when -d or -D are in use - Try six times with 1/2 second delay to debug the child PR: bin/52190 (updated) Submitted by: Dan Nelson <dnelson@allantgroup.com> Approved by: alfred
2006-05-15 21:18:28 +00:00
Add more static keywords to truss(1) source code. There are some tables in the source code that are only used by the individual source files themselves. Therefore there is no need to export them.
2011-12-10 18:27:55 +00:00
static struct xlat poll_flags[] = {
- Add decoding of kse_release, kevent, sigprocmask, unmount, socket, getrusage, rename, __getcwd, shutdown, getrlimit, setrlimit, _umtx_lock, _umtx_unlock, pathconf, truncate, ftruncate, kill - Decode more arguments of open, mprot, *stat, and fcntl. - Convert all constant-macro and bitfield decoding to lookup tables; much cleaner than previous code. - Print the timestamp of process exit and signal reception when -d or -D are in use - Try six times with 1/2 second delay to debug the child PR: bin/52190 (updated) Submitted by: Dan Nelson <dnelson@allantgroup.com> Approved by: alfred
2006-05-15 21:18:28 +00:00
X(POLLSTANDARD) X(POLLIN) X(POLLPRI) X(POLLOUT) X(POLLERR)
X(POLLHUP) X(POLLNVAL) X(POLLRDNORM) X(POLLRDBAND)
X(POLLWRBAND) X(POLLINIGNEOF) XEND
};
static struct xlat sigaction_flags[] = {
X(SA_ONSTACK) X(SA_RESTART) X(SA_RESETHAND) X(SA_NOCLDSTOP)
X(SA_NODEFER) X(SA_NOCLDWAIT) X(SA_SIGINFO) XEND
};
Tidy the linux_socketcall decoding: - Don't exit if get_struct() fails, instead print the raw pointer value to match all other argument decoding cases. - Use an xlat table instead of a home-rolled switch for the operation name. - Display the nested socketcall args structure as a structure instead of as two inline arguments.
2015-08-17 18:47:39 +00:00
static struct xlat linux_socketcall_ops[] = {
X(LINUX_SOCKET) X(LINUX_BIND) X(LINUX_CONNECT) X(LINUX_LISTEN)
X(LINUX_ACCEPT) X(LINUX_GETSOCKNAME) X(LINUX_GETPEERNAME)
X(LINUX_SOCKETPAIR) X(LINUX_SEND) X(LINUX_RECV) X(LINUX_SENDTO)
X(LINUX_RECVFROM) X(LINUX_SHUTDOWN) X(LINUX_SETSOCKOPT)
X(LINUX_GETSOCKOPT) X(LINUX_SENDMSG) X(LINUX_RECVMSG)
XEND
};
Properly format pointer size independent CloudABI system calls. CloudABI has approximately 50 system calls that do not depend on the pointer size of the system. As the ABI is pretty compact, it takes little effort to each truss(8) the formatting rules for these system calls. Start off by formatting pointer size independent system calls. Changes: - Make it possible to include the CloudABI system call definitions in FreeBSD userspace builds. Add ${root}/sys to the truss(8) Makefile so we can pull in <compat/cloudabi/cloudabi_syscalldefs.h>. - Refactoring: patch up amd64-cloudabi64.c to use the CLOUDABI_* constants instead of rolling our own table. - Add table entries for all of the system calls. - Add new generic formatting types (UInt, IntArray) that we'll be using to format unsigned integers and arrays of integers. - Add CloudABI specific formatting types. Approved by: jhb Differential Revision: https://reviews.freebsd.org/D3836
2015-10-08 05:27:45 +00:00
#undef X
#define X(a) { CLOUDABI_##a, #a },
static struct xlat cloudabi_advice[] = {
X(ADVICE_DONTNEED) X(ADVICE_NOREUSE) X(ADVICE_NORMAL)
X(ADVICE_RANDOM) X(ADVICE_SEQUENTIAL) X(ADVICE_WILLNEED)
XEND
};
static struct xlat cloudabi_clockid[] = {
X(CLOCK_MONOTONIC) X(CLOCK_PROCESS_CPUTIME_ID)
X(CLOCK_REALTIME) X(CLOCK_THREAD_CPUTIME_ID)
XEND
};
static struct xlat cloudabi_fdflags[] = {
X(FDFLAG_APPEND) X(FDFLAG_DSYNC) X(FDFLAG_NONBLOCK)
X(FDFLAG_RSYNC) X(FDFLAG_SYNC)
XEND
};
static struct xlat cloudabi_fdsflags[] = {
X(FDSTAT_FLAGS) X(FDSTAT_RIGHTS)
XEND
};
static struct xlat cloudabi_filetype[] = {
X(FILETYPE_UNKNOWN) X(FILETYPE_BLOCK_DEVICE)
X(FILETYPE_CHARACTER_DEVICE) X(FILETYPE_DIRECTORY)
Import the latest CloudABI definitions, version 0.16. The most important change in this release is the removal of the poll_fd() system call; CloudABI's equivalent of kevent(). Though I think that kqueue is a lot saner than many of its alternatives, our experience is that emulating this system call on other systems accurately isn't easy. It has become a complex API, even though I'm not convinced this complexity is needed. This is why we've decided to take a different approach, by looking one layer up. We're currently adding an event loop to CloudABI's C library that is API compatible with libuv (except when incompatible with Capsicum). Initially, this event loop will be built on top of plain inefficient poll() calls. Only after this is finished, we'll work our way backwards and design a new set of system calls to optimize it. Interesting challenges will include integrating asynchronous I/O into such a system call API. libuv currently doesn't aio(4) on Linux/BSD, due to it being unreliable and having undesired semantics. Obtained from: https://github.com/NuxiNL/cloudabi
2017-10-18 19:22:53 +00:00
X(FILETYPE_PROCESS) X(FILETYPE_REGULAR_FILE)
Merge pipes and socket pairs. Now that CloudABI's sockets API has been changed to be addressless and only connected socket instances are used (e.g., socket pairs), they have become fairly similar to pipes. The only differences on CloudABI is that socket pairs additionally support shutdown(), send() and recv(). To simplify the ABI, we've therefore decided to remove pipes as a separate file descriptor type and just let pipe() return a socket pair of type SOCK_STREAM. S_ISFIFO() and S_ISSOCK() are now defined identically.
2017-09-05 07:46:45 +00:00
X(FILETYPE_SHARED_MEMORY) X(FILETYPE_SOCKET_DGRAM)
X(FILETYPE_SOCKET_STREAM) X(FILETYPE_SYMBOLIC_LINK)
Properly format pointer size independent CloudABI system calls. CloudABI has approximately 50 system calls that do not depend on the pointer size of the system. As the ABI is pretty compact, it takes little effort to each truss(8) the formatting rules for these system calls. Start off by formatting pointer size independent system calls. Changes: - Make it possible to include the CloudABI system call definitions in FreeBSD userspace builds. Add ${root}/sys to the truss(8) Makefile so we can pull in <compat/cloudabi/cloudabi_syscalldefs.h>. - Refactoring: patch up amd64-cloudabi64.c to use the CLOUDABI_* constants instead of rolling our own table. - Add table entries for all of the system calls. - Add new generic formatting types (UInt, IntArray) that we'll be using to format unsigned integers and arrays of integers. - Add CloudABI specific formatting types. Approved by: jhb Differential Revision: https://reviews.freebsd.org/D3836
2015-10-08 05:27:45 +00:00
XEND
};
static struct xlat cloudabi_fsflags[] = {
X(FILESTAT_ATIM) X(FILESTAT_ATIM_NOW) X(FILESTAT_MTIM)
X(FILESTAT_MTIM_NOW) X(FILESTAT_SIZE)
XEND
};
static struct xlat cloudabi_mflags[] = {
X(MAP_ANON) X(MAP_FIXED) X(MAP_PRIVATE) X(MAP_SHARED)
XEND
};
static struct xlat cloudabi_mprot[] = {
X(PROT_EXEC) X(PROT_WRITE) X(PROT_READ)
XEND
};
static struct xlat cloudabi_msflags[] = {
X(MS_ASYNC) X(MS_INVALIDATE) X(MS_SYNC)
XEND
};
static struct xlat cloudabi_oflags[] = {
X(O_CREAT) X(O_DIRECTORY) X(O_EXCL) X(O_TRUNC)
XEND
};
static struct xlat cloudabi_sdflags[] = {
X(SHUT_RD) X(SHUT_WR)
XEND
};
static struct xlat cloudabi_signal[] = {
X(SIGABRT) X(SIGALRM) X(SIGBUS) X(SIGCHLD) X(SIGCONT) X(SIGFPE)
X(SIGHUP) X(SIGILL) X(SIGINT) X(SIGKILL) X(SIGPIPE) X(SIGQUIT)
X(SIGSEGV) X(SIGSTOP) X(SIGSYS) X(SIGTERM) X(SIGTRAP) X(SIGTSTP)
X(SIGTTIN) X(SIGTTOU) X(SIGURG) X(SIGUSR1) X(SIGUSR2)
X(SIGVTALRM) X(SIGXCPU) X(SIGXFSZ)
XEND
};
static struct xlat cloudabi_ulflags[] = {
X(UNLINK_REMOVEDIR)
XEND
};
static struct xlat cloudabi_whence[] = {
X(WHENCE_CUR) X(WHENCE_END) X(WHENCE_SET)
XEND
};
- Add decoding of kse_release, kevent, sigprocmask, unmount, socket, getrusage, rename, __getcwd, shutdown, getrlimit, setrlimit, _umtx_lock, _umtx_unlock, pathconf, truncate, ftruncate, kill - Decode more arguments of open, mprot, *stat, and fcntl. - Convert all constant-macro and bitfield decoding to lookup tables; much cleaner than previous code. - Print the timestamp of process exit and signal reception when -d or -D are in use - Try six times with 1/2 second delay to debug the child PR: bin/52190 (updated) Submitted by: Dan Nelson <dnelson@allantgroup.com> Approved by: alfred
2006-05-15 21:18:28 +00:00
#undef X
#undef XEND
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
/*
* Searches an xlat array for a value, and returns it if found. Otherwise
* return a string representation.
*/
static const char *
lookup(struct xlat *xlat, int val, int base)
- Add decoding of kse_release, kevent, sigprocmask, unmount, socket, getrusage, rename, __getcwd, shutdown, getrlimit, setrlimit, _umtx_lock, _umtx_unlock, pathconf, truncate, ftruncate, kill - Decode more arguments of open, mprot, *stat, and fcntl. - Convert all constant-macro and bitfield decoding to lookup tables; much cleaner than previous code. - Print the timestamp of process exit and signal reception when -d or -D are in use - Try six times with 1/2 second delay to debug the child PR: bin/52190 (updated) Submitted by: Dan Nelson <dnelson@allantgroup.com> Approved by: alfred
2006-05-15 21:18:28 +00:00
{
static char tmp[16];
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
- Add decoding of kse_release, kevent, sigprocmask, unmount, socket, getrusage, rename, __getcwd, shutdown, getrlimit, setrlimit, _umtx_lock, _umtx_unlock, pathconf, truncate, ftruncate, kill - Decode more arguments of open, mprot, *stat, and fcntl. - Convert all constant-macro and bitfield decoding to lookup tables; much cleaner than previous code. - Print the timestamp of process exit and signal reception when -d or -D are in use - Try six times with 1/2 second delay to debug the child PR: bin/52190 (updated) Submitted by: Dan Nelson <dnelson@allantgroup.com> Approved by: alfred
2006-05-15 21:18:28 +00:00
for (; xlat->str != NULL; xlat++)
if (xlat->val == val)
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
return (xlat->str);
- Add decoding of kse_release, kevent, sigprocmask, unmount, socket, getrusage, rename, __getcwd, shutdown, getrlimit, setrlimit, _umtx_lock, _umtx_unlock, pathconf, truncate, ftruncate, kill - Decode more arguments of open, mprot, *stat, and fcntl. - Convert all constant-macro and bitfield decoding to lookup tables; much cleaner than previous code. - Print the timestamp of process exit and signal reception when -d or -D are in use - Try six times with 1/2 second delay to debug the child PR: bin/52190 (updated) Submitted by: Dan Nelson <dnelson@allantgroup.com> Approved by: alfred
2006-05-15 21:18:28 +00:00
switch (base) {
case 8:
sprintf(tmp, "0%o", val);
break;
case 16:
sprintf(tmp, "0x%x", val);
break;
case 10:
sprintf(tmp, "%u", val);
break;
default:
errx(1,"Unknown lookup base");
break;
}
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
return (tmp);
- Add decoding of kse_release, kevent, sigprocmask, unmount, socket, getrusage, rename, __getcwd, shutdown, getrlimit, setrlimit, _umtx_lock, _umtx_unlock, pathconf, truncate, ftruncate, kill - Decode more arguments of open, mprot, *stat, and fcntl. - Convert all constant-macro and bitfield decoding to lookup tables; much cleaner than previous code. - Print the timestamp of process exit and signal reception when -d or -D are in use - Try six times with 1/2 second delay to debug the child PR: bin/52190 (updated) Submitted by: Dan Nelson <dnelson@allantgroup.com> Approved by: alfred
2006-05-15 21:18:28 +00:00
}
Make use of ptrace(2) instead of procfs in truss(1), eliminating yet another need of an available /proc/ mount. Tested with: make universe Submitted by: howardsu Reviewed by: alfred
2007-04-10 04:03:34 +00:00
static const char *
xlookup(struct xlat *xlat, int val)
- Add decoding of kse_release, kevent, sigprocmask, unmount, socket, getrusage, rename, __getcwd, shutdown, getrlimit, setrlimit, _umtx_lock, _umtx_unlock, pathconf, truncate, ftruncate, kill - Decode more arguments of open, mprot, *stat, and fcntl. - Convert all constant-macro and bitfield decoding to lookup tables; much cleaner than previous code. - Print the timestamp of process exit and signal reception when -d or -D are in use - Try six times with 1/2 second delay to debug the child PR: bin/52190 (updated) Submitted by: Dan Nelson <dnelson@allantgroup.com> Approved by: alfred
2006-05-15 21:18:28 +00:00
{
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
return (lookup(xlat, val, 16));
- Add decoding of kse_release, kevent, sigprocmask, unmount, socket, getrusage, rename, __getcwd, shutdown, getrlimit, setrlimit, _umtx_lock, _umtx_unlock, pathconf, truncate, ftruncate, kill - Decode more arguments of open, mprot, *stat, and fcntl. - Convert all constant-macro and bitfield decoding to lookup tables; much cleaner than previous code. - Print the timestamp of process exit and signal reception when -d or -D are in use - Try six times with 1/2 second delay to debug the child PR: bin/52190 (updated) Submitted by: Dan Nelson <dnelson@allantgroup.com> Approved by: alfred
2006-05-15 21:18:28 +00:00
}
Various style and whitespace fixes.
2015-08-19 20:02:03 +00:00
/*
* Searches an xlat array containing bitfield values. Remaining bits
* set after removing the known ones are printed at the end:
* IN|0x400.
*/
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
static char *
xlookup_bits(struct xlat *xlat, int val)
- Add decoding of kse_release, kevent, sigprocmask, unmount, socket, getrusage, rename, __getcwd, shutdown, getrlimit, setrlimit, _umtx_lock, _umtx_unlock, pathconf, truncate, ftruncate, kill - Decode more arguments of open, mprot, *stat, and fcntl. - Convert all constant-macro and bitfield decoding to lookup tables; much cleaner than previous code. - Print the timestamp of process exit and signal reception when -d or -D are in use - Try six times with 1/2 second delay to debug the child PR: bin/52190 (updated) Submitted by: Dan Nelson <dnelson@allantgroup.com> Approved by: alfred
2006-05-15 21:18:28 +00:00
{
- Style(9) cleanup. Approved by: kib (mentor)
2012-09-02 11:03:18 +00:00
int len, rem;
- Add decoding of kse_release, kevent, sigprocmask, unmount, socket, getrusage, rename, __getcwd, shutdown, getrlimit, setrlimit, _umtx_lock, _umtx_unlock, pathconf, truncate, ftruncate, kill - Decode more arguments of open, mprot, *stat, and fcntl. - Convert all constant-macro and bitfield decoding to lookup tables; much cleaner than previous code. - Print the timestamp of process exit and signal reception when -d or -D are in use - Try six times with 1/2 second delay to debug the child PR: bin/52190 (updated) Submitted by: Dan Nelson <dnelson@allantgroup.com> Approved by: alfred
2006-05-15 21:18:28 +00:00
static char str[512];
- Style(9) cleanup. Approved by: kib (mentor)
2012-09-02 11:03:18 +00:00
len = 0;
rem = val;
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
for (; xlat->str != NULL; xlat++) {
if ((xlat->val & rem) == xlat->val) {
Various style and whitespace fixes.
2015-08-19 20:02:03 +00:00
/*
* Don't print the "all-bits-zero" string unless all
* bits are really zero.
*/
- Add decoding of kse_release, kevent, sigprocmask, unmount, socket, getrusage, rename, __getcwd, shutdown, getrlimit, setrlimit, _umtx_lock, _umtx_unlock, pathconf, truncate, ftruncate, kill - Decode more arguments of open, mprot, *stat, and fcntl. - Convert all constant-macro and bitfield decoding to lookup tables; much cleaner than previous code. - Print the timestamp of process exit and signal reception when -d or -D are in use - Try six times with 1/2 second delay to debug the child PR: bin/52190 (updated) Submitted by: Dan Nelson <dnelson@allantgroup.com> Approved by: alfred
2006-05-15 21:18:28 +00:00
if (xlat->val == 0 && val != 0)
continue;
len += sprintf(str + len, "%s|", xlat->str);
rem &= ~(xlat->val);
}
}
Various style and whitespace fixes.
2015-08-19 20:02:03 +00:00
/*
* If we have leftover bits or didn't match anything, print
* the remainder.
*/
- Add decoding of kse_release, kevent, sigprocmask, unmount, socket, getrusage, rename, __getcwd, shutdown, getrlimit, setrlimit, _umtx_lock, _umtx_unlock, pathconf, truncate, ftruncate, kill - Decode more arguments of open, mprot, *stat, and fcntl. - Convert all constant-macro and bitfield decoding to lookup tables; much cleaner than previous code. - Print the timestamp of process exit and signal reception when -d or -D are in use - Try six times with 1/2 second delay to debug the child PR: bin/52190 (updated) Submitted by: Dan Nelson <dnelson@allantgroup.com> Approved by: alfred
2006-05-15 21:18:28 +00:00
if (rem || len == 0)
len += sprintf(str + len, "0x%x", rem);
if (len && str[len - 1] == '|')
len--;
str[len] = 0;
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
return (str);
- Add decoding of kse_release, kevent, sigprocmask, unmount, socket, getrusage, rename, __getcwd, shutdown, getrlimit, setrlimit, _umtx_lock, _umtx_unlock, pathconf, truncate, ftruncate, kill - Decode more arguments of open, mprot, *stat, and fcntl. - Convert all constant-macro and bitfield decoding to lookup tables; much cleaner than previous code. - Print the timestamp of process exit and signal reception when -d or -D are in use - Try six times with 1/2 second delay to debug the child PR: bin/52190 (updated) Submitted by: Dan Nelson <dnelson@allantgroup.com> Approved by: alfred
2006-05-15 21:18:28 +00:00
}
Move mksubr from kdump into libsysdecode. Restructure this script so that it generates a header of tables instead of a source file. The tables are included in a flags.c source file which provides functions to decode various system call arguments. For functions that decode an enumeration, the function returns a pointer to a string for known values and NULL for unknown values. For functions that do more complex decoding (typically of a bitmask), the function accepts a pointer to a FILE object (open_memstream() can be used as a string builder) to which decoded values are written. If the function operates on a bitmask, the function returns true if any bits were decoded or false if the entire value was valid. Additionally, the third argument accepts a pointer to a value to which any undecoded bits are stored. This pointer can be NULL if the caller doesn't care about remaining bits. Convert kdump over to using decoder functions from libsysdecode instead of mksubr. truss also uses decoders from libsysdecode instead of private lookup tables, though lookup tables for objects not decoded by kdump remain in truss for now. Eventually most of these tables should move into libsysdecode as the automated table generation approach from mksubr is less stale than the static tables in truss. Some changes have been made to truss and kdump output: - The flags passed to open() are now properly decoded in that one of O_RDONLY, O_RDWR, O_WRONLY, or O_EXEC is always included in a decoded mask. - Optional arguments to open(), openat(), and fcntl() are only printed in kdump if they exist (e.g. the mode is only printed for open() if O_CREAT is set in the flags). - Print argument to F_GETLK/SETLK/SETLKW in kdump as a pointer, not int. - Include all procctl() commands. - Correctly decode pipe2() flags in truss by not assuming full open()-like flags with O_RDONLY, etc. - Decode file flags passed to *chflags() as file flags (UF_* and SF_*) rather than as a file mode. - Fix decoding of quotactl() commands by splitting out the two command components instead of assuming the raw command value matches the primary command component. In addition, truss and kdump now build without triggering any warnings. All of the sysdecode manpages now include the required headers in the synopsis. Reviewed by: kib (several older versions), wblock (manpages) MFC after: 2 months Differential Revision: https://reviews.freebsd.org/D7847
2016-10-17 22:37:07 +00:00
static void
print_integer_arg(const char *(*decoder)(int), FILE *fp, int value)
{
const char *str;
str = decoder(value);
if (str != NULL)
fputs(str, fp);
else
fprintf(fp, "%d", value);
}
static void
print_mask_arg(bool (*decoder)(FILE *, int, int *), FILE *fp, int value)
{
int rem;
if (!decoder(fp, value, &rem))
fprintf(fp, "0x%x", rem);
else if (rem != 0)
fprintf(fp, "|0x%x", rem);
}
Decode the arguments passed to cap_fcntls_get() and cap_fcntls_limit().
2017-03-18 18:10:02 +00:00
static void
print_mask_arg32(bool (*decoder)(FILE *, uint32_t, uint32_t *), FILE *fp,
uint32_t value)
{
uint32_t rem;
if (!decoder(fp, value, &rem))
fprintf(fp, "0x%x", rem);
else if (rem != 0)
fprintf(fp, "|0x%x", rem);
}
Automate the handling of QUAD_ALIGN and QUAD_SLOTS. Previously, the offset in a system call description specified the array index of the start of a system call argument. For most system call arguments this was the same as the index of the argument in the function signature. 64-bit arguments (off_t and id_t values) passed on 32-bit platforms use two slots in the array however. This was handled by adding (QUAD_SLOTS - 1) to the slot indicies of any subsequent arguments after a 64-bit argument (though written as ("{ Quad, 1 }, { Int, 1 + QUAD_SLOTS }" rather than "{ Quad, 1 }, { Int, 2 + QUAD_SLOTS - 1 }"). If a system call contained multiple 64-bit arguments (such as posix_fadvise()), then additional arguments would need to use 'QUAD_SLOTS * 2' but remember to subtract 2 from the initial number, etc. In addition, 32-bit powerpc requires 64-bit arguments to be 64-bit aligned, so if the effective index in the array of a 64-bit argument is odd, it needs QUAD_ALIGN added to the current and any subsequent slots. However, if the effective index in the array of a 64-bit argument was even, QUAD_ALIGN was omitted. This approach was messy and error prone. This commit replaces it with automated pre-processing of the system call table to do fixups for 64-bit argument offsets. The offset in a system call description now indicates the index of an argument in the associated function call's signature. A fixup function is run against each decoded system call description during startup on 32-bit platforms. The fixup function maintains an 'offset' value which holds an offset to be added to each remaining system call argument's index. Initially offset is 0. When a 64-bit system call argument is encountered, the offset is first aligned to a 64-bit boundary (only on powerpc) and then incremented to account for the second argument slot used by the argument. This modified 'offset' is then applied to any remaining arguments. This approach does require a few things that were not previously required: 1) Each system call description must now list arguments in ascending order (existing ones all do) without using duplicate slots in the register array. A new assert() should catch any future descriptions which violate this rule. 2) A system call description is still permitted to omit arguments (though none currently do), but if the call accepts 64-bit arguments those cannot be omitted or incorrect results will be displated on 32-bit systems. Tested on: amd64 and i386
2017-03-15 23:08:11 +00:00
#ifndef __LP64__
/*
* Add argument padding to subsequent system calls afater a Quad
* syscall arguments as needed. This used to be done by hand in the
* decoded_syscalls table which was ugly and error prone. It is
* simpler to do the fixup of offsets at initalization time than when
* decoding arguments.
*/
static void
quad_fixup(struct syscall *sc)
{
int offset, prev;
u_int i;
offset = 0;
prev = -1;
for (i = 0; i < sc->nargs; i++) {
/* This arg type is a dummy that doesn't use offset. */
if ((sc->args[i].type & ARG_MASK) == PipeFds)
continue;
assert(prev < sc->args[i].offset);
prev = sc->args[i].offset;
sc->args[i].offset += offset;
switch (sc->args[i].type & ARG_MASK) {
case Quad:
case QuadHex:
#ifdef __powerpc__
/*
* 64-bit arguments on 32-bit powerpc must be
* 64-bit aligned. If the current offset is
* not aligned, the calling convention inserts
* a 32-bit pad argument that should be skipped.
*/
if (sc->args[i].offset % 2 == 1) {
sc->args[i].offset++;
offset++;
}
#endif
offset++;
default:
break;
}
}
}
#endif
Fix tracking of unknown syscalls for 'truss -c'. This is done by changing get_syscall() to either lookup the known syscall or add it into the list with the default handlers for printing. This also simplifies some code to not have to check if the syscall variable is set or NULL. Reviewed by: jhb Relnotes: yes Differential Revision: https://reviews.freebsd.org/D3792
2015-10-05 18:08:35 +00:00
void
init_syscalls(void)
{
struct syscall *sc;
STAILQ_INIT(&syscalls);
Automate the handling of QUAD_ALIGN and QUAD_SLOTS. Previously, the offset in a system call description specified the array index of the start of a system call argument. For most system call arguments this was the same as the index of the argument in the function signature. 64-bit arguments (off_t and id_t values) passed on 32-bit platforms use two slots in the array however. This was handled by adding (QUAD_SLOTS - 1) to the slot indicies of any subsequent arguments after a 64-bit argument (though written as ("{ Quad, 1 }, { Int, 1 + QUAD_SLOTS }" rather than "{ Quad, 1 }, { Int, 2 + QUAD_SLOTS - 1 }"). If a system call contained multiple 64-bit arguments (such as posix_fadvise()), then additional arguments would need to use 'QUAD_SLOTS * 2' but remember to subtract 2 from the initial number, etc. In addition, 32-bit powerpc requires 64-bit arguments to be 64-bit aligned, so if the effective index in the array of a 64-bit argument is odd, it needs QUAD_ALIGN added to the current and any subsequent slots. However, if the effective index in the array of a 64-bit argument was even, QUAD_ALIGN was omitted. This approach was messy and error prone. This commit replaces it with automated pre-processing of the system call table to do fixups for 64-bit argument offsets. The offset in a system call description now indicates the index of an argument in the associated function call's signature. A fixup function is run against each decoded system call description during startup on 32-bit platforms. The fixup function maintains an 'offset' value which holds an offset to be added to each remaining system call argument's index. Initially offset is 0. When a 64-bit system call argument is encountered, the offset is first aligned to a 64-bit boundary (only on powerpc) and then incremented to account for the second argument slot used by the argument. This modified 'offset' is then applied to any remaining arguments. This approach does require a few things that were not previously required: 1) Each system call description must now list arguments in ascending order (existing ones all do) without using duplicate slots in the register array. A new assert() should catch any future descriptions which violate this rule. 2) A system call description is still permitted to omit arguments (though none currently do), but if the call accepts 64-bit arguments those cannot be omitted or incorrect results will be displated on 32-bit systems. Tested on: amd64 and i386
2017-03-15 23:08:11 +00:00
for (sc = decoded_syscalls; sc->name != NULL; sc++) {
#ifndef __LP64__
quad_fixup(sc);
#endif
Fix tracking of unknown syscalls for 'truss -c'. This is done by changing get_syscall() to either lookup the known syscall or add it into the list with the default handlers for printing. This also simplifies some code to not have to check if the syscall variable is set or NULL. Reviewed by: jhb Relnotes: yes Differential Revision: https://reviews.freebsd.org/D3792
2015-10-05 18:08:35 +00:00
STAILQ_INSERT_HEAD(&syscalls, sc, entries);
Automate the handling of QUAD_ALIGN and QUAD_SLOTS. Previously, the offset in a system call description specified the array index of the start of a system call argument. For most system call arguments this was the same as the index of the argument in the function signature. 64-bit arguments (off_t and id_t values) passed on 32-bit platforms use two slots in the array however. This was handled by adding (QUAD_SLOTS - 1) to the slot indicies of any subsequent arguments after a 64-bit argument (though written as ("{ Quad, 1 }, { Int, 1 + QUAD_SLOTS }" rather than "{ Quad, 1 }, { Int, 2 + QUAD_SLOTS - 1 }"). If a system call contained multiple 64-bit arguments (such as posix_fadvise()), then additional arguments would need to use 'QUAD_SLOTS * 2' but remember to subtract 2 from the initial number, etc. In addition, 32-bit powerpc requires 64-bit arguments to be 64-bit aligned, so if the effective index in the array of a 64-bit argument is odd, it needs QUAD_ALIGN added to the current and any subsequent slots. However, if the effective index in the array of a 64-bit argument was even, QUAD_ALIGN was omitted. This approach was messy and error prone. This commit replaces it with automated pre-processing of the system call table to do fixups for 64-bit argument offsets. The offset in a system call description now indicates the index of an argument in the associated function call's signature. A fixup function is run against each decoded system call description during startup on 32-bit platforms. The fixup function maintains an 'offset' value which holds an offset to be added to each remaining system call argument's index. Initially offset is 0. When a 64-bit system call argument is encountered, the offset is first aligned to a 64-bit boundary (only on powerpc) and then incremented to account for the second argument slot used by the argument. This modified 'offset' is then applied to any remaining arguments. This approach does require a few things that were not previously required: 1) Each system call description must now list arguments in ascending order (existing ones all do) without using duplicate slots in the register array. A new assert() should catch any future descriptions which violate this rule. 2) A system call description is still permitted to omit arguments (though none currently do), but if the call accepts 64-bit arguments those cannot be omitted or incorrect results will be displated on 32-bit systems. Tested on: amd64 and i386
2017-03-15 23:08:11 +00:00
}
Fix tracking of unknown syscalls for 'truss -c'. This is done by changing get_syscall() to either lookup the known syscall or add it into the list with the default handlers for printing. This also simplifies some code to not have to check if the syscall variable is set or NULL. Reviewed by: jhb Relnotes: yes Differential Revision: https://reviews.freebsd.org/D3792
2015-10-05 18:08:35 +00:00
}
Rework syscall structure lookups. Avoid always using an O(n^2) loop over known syscall structures with strcmp() on each system call. Instead, use a per-ABI cache indexed by the system call number. The first 1024 system calls (which should cover all of the normal system calls in currently-supported ABIs) use a flat array indexed by the system call number to find system call structure. For other system calls, a linked list of structures storing an integer to structure mapping is stored in the ABI. The linked list isn't very smart, but it should only be used by buggy applications invoking unknown system calls. This also fixes handling of unknown system calls which currently trigger a NULL pointer dereference. Reviewed by: kib MFC after: 2 weeks
2016-12-06 00:39:00 +00:00
static struct syscall *
find_syscall(struct procabi *abi, u_int number)
{
struct extra_syscall *es;
if (number < nitems(abi->syscalls))
return (abi->syscalls[number]);
STAILQ_FOREACH(es, &abi->extra_syscalls, entries) {
if (es->number == number)
return (es->sc);
}
return (NULL);
}
static void
add_syscall(struct procabi *abi, u_int number, struct syscall *sc)
{
struct extra_syscall *es;
if (number < nitems(abi->syscalls)) {
assert(abi->syscalls[number] == NULL);
abi->syscalls[number] = sc;
} else {
es = malloc(sizeof(*es));
es->sc = sc;
es->number = number;
STAILQ_INSERT_TAIL(&abi->extra_syscalls, es, entries);
}
}
Truss program. Requires procfs.
1997-12-06 05:23:12 +00:00
/*
* If/when the list gets big, it might be desirable to do it
* as a hash table or binary search.
*/
struct syscall *
Rework syscall structure lookups. Avoid always using an O(n^2) loop over known syscall structures with strcmp() on each system call. Instead, use a per-ABI cache indexed by the system call number. The first 1024 system calls (which should cover all of the normal system calls in currently-supported ABIs) use a flat array indexed by the system call number to find system call structure. For other system calls, a linked list of structures storing an integer to structure mapping is stored in the ABI. The linked list isn't very smart, but it should only be used by buggy applications invoking unknown system calls. This also fixes handling of unknown system calls which currently trigger a NULL pointer dereference. Reviewed by: kib MFC after: 2 weeks
2016-12-06 00:39:00 +00:00
get_syscall(struct threadinfo *t, u_int number, u_int nargs)
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
{
- Style(9) cleanup. Approved by: kib (mentor)
2012-09-02 11:03:18 +00:00
struct syscall *sc;
Rework syscall structure lookups. Avoid always using an O(n^2) loop over known syscall structures with strcmp() on each system call. Instead, use a per-ABI cache indexed by the system call number. The first 1024 system calls (which should cover all of the normal system calls in currently-supported ABIs) use a flat array indexed by the system call number to find system call structure. For other system calls, a linked list of structures storing an integer to structure mapping is stored in the ABI. The linked list isn't very smart, but it should only be used by buggy applications invoking unknown system calls. This also fixes handling of unknown system calls which currently trigger a NULL pointer dereference. Reviewed by: kib MFC after: 2 weeks
2016-12-06 00:39:00 +00:00
const char *name;
char *new_name;
u_int i;
sc = find_syscall(t->proc->abi, number);
if (sc != NULL)
return (sc);
name = sysdecode_syscallname(t->proc->abi->abi, number);
if (name == NULL) {
asprintf(&new_name, "#%d", number);
name = new_name;
} else
new_name = NULL;
STAILQ_FOREACH(sc, &syscalls, entries) {
if (strcmp(name, sc->name) == 0) {
add_syscall(t->proc->abi, number, sc);
free(new_name);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
return (sc);
Rework syscall structure lookups. Avoid always using an O(n^2) loop over known syscall structures with strcmp() on each system call. Instead, use a per-ABI cache indexed by the system call number. The first 1024 system calls (which should cover all of the normal system calls in currently-supported ABIs) use a flat array indexed by the system call number to find system call structure. For other system calls, a linked list of structures storing an integer to structure mapping is stored in the ABI. The linked list isn't very smart, but it should only be used by buggy applications invoking unknown system calls. This also fixes handling of unknown system calls which currently trigger a NULL pointer dereference. Reviewed by: kib MFC after: 2 weeks
2016-12-06 00:39:00 +00:00
}
}
Fix tracking of unknown syscalls for 'truss -c'. This is done by changing get_syscall() to either lookup the known syscall or add it into the list with the default handlers for printing. This also simplifies some code to not have to check if the syscall variable is set or NULL. Reviewed by: jhb Relnotes: yes Differential Revision: https://reviews.freebsd.org/D3792
2015-10-05 18:08:35 +00:00
/* It is unknown. Add it into the list. */
#if DEBUG
fprintf(stderr, "unknown syscall %s -- setting args to %d\n", name,
nargs);
#endif
sc = calloc(1, sizeof(struct syscall));
Rework syscall structure lookups. Avoid always using an O(n^2) loop over known syscall structures with strcmp() on each system call. Instead, use a per-ABI cache indexed by the system call number. The first 1024 system calls (which should cover all of the normal system calls in currently-supported ABIs) use a flat array indexed by the system call number to find system call structure. For other system calls, a linked list of structures storing an integer to structure mapping is stored in the ABI. The linked list isn't very smart, but it should only be used by buggy applications invoking unknown system calls. This also fixes handling of unknown system calls which currently trigger a NULL pointer dereference. Reviewed by: kib MFC after: 2 weeks
2016-12-06 00:39:00 +00:00
sc->name = name;
if (new_name != NULL)
sc->unknown = true;
Fix tracking of unknown syscalls for 'truss -c'. This is done by changing get_syscall() to either lookup the known syscall or add it into the list with the default handlers for printing. This also simplifies some code to not have to check if the syscall variable is set or NULL. Reviewed by: jhb Relnotes: yes Differential Revision: https://reviews.freebsd.org/D3792
2015-10-05 18:08:35 +00:00
sc->ret_type = 1;
sc->nargs = nargs;
for (i = 0; i < nargs; i++) {
sc->args[i].offset = i;
/* Treat all unknown arguments as LongHex. */
sc->args[i].type = LongHex;
Truss program. Requires procfs.
1997-12-06 05:23:12 +00:00
}
Fix tracking of unknown syscalls for 'truss -c'. This is done by changing get_syscall() to either lookup the known syscall or add it into the list with the default handlers for printing. This also simplifies some code to not have to check if the syscall variable is set or NULL. Reviewed by: jhb Relnotes: yes Differential Revision: https://reviews.freebsd.org/D3792
2015-10-05 18:08:35 +00:00
STAILQ_INSERT_HEAD(&syscalls, sc, entries);
Rework syscall structure lookups. Avoid always using an O(n^2) loop over known syscall structures with strcmp() on each system call. Instead, use a per-ABI cache indexed by the system call number. The first 1024 system calls (which should cover all of the normal system calls in currently-supported ABIs) use a flat array indexed by the system call number to find system call structure. For other system calls, a linked list of structures storing an integer to structure mapping is stored in the ABI. The linked list isn't very smart, but it should only be used by buggy applications invoking unknown system calls. This also fixes handling of unknown system calls which currently trigger a NULL pointer dereference. Reviewed by: kib MFC after: 2 weeks
2016-12-06 00:39:00 +00:00
add_syscall(t->proc->abi, number, sc);
Fix tracking of unknown syscalls for 'truss -c'. This is done by changing get_syscall() to either lookup the known syscall or add it into the list with the default handlers for printing. This also simplifies some code to not have to check if the syscall variable is set or NULL. Reviewed by: jhb Relnotes: yes Differential Revision: https://reviews.freebsd.org/D3792
2015-10-05 18:08:35 +00:00
return (sc);
Truss program. Requires procfs.
1997-12-06 05:23:12 +00:00
}
Teach truss(1) to display sockaddrs. It currently knows about AF_INET, AF_INET6 and AF_UNIX sockaddrs, and will recognize accept(), bind(), connect(), getpeername() and getsockname() as syscalls taking sockaddr arguments. Some enterprising soul might want to add (and test) support for the send() / recv() family of syscalls as well. MFC after: 1 week
2001-10-21 21:57:10 +00:00
/*
* Copy a fixed amount of bytes from the process.
*/
Partial WARNS=1 fizes with NO_WERROR set to prevent world breakage. Use __FBSDID().
2001-12-11 23:34:02 +00:00
static int
- Use pid_t type instead of just int. Approved by: kib (mentor)
2012-08-21 14:58:51 +00:00
get_struct(pid_t pid, void *offset, void *buf, int len)
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
{
Make use of ptrace(2) instead of procfs in truss(1), eliminating yet another need of an available /proc/ mount. Tested with: make universe Submitted by: howardsu Reviewed by: alfred
2007-04-10 04:03:34 +00:00
struct ptrace_io_desc iorequest;
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
Make use of ptrace(2) instead of procfs in truss(1), eliminating yet another need of an available /proc/ mount. Tested with: make universe Submitted by: howardsu Reviewed by: alfred
2007-04-10 04:03:34 +00:00
iorequest.piod_op = PIOD_READ_D;
iorequest.piod_offs = offset;
iorequest.piod_addr = buf;
iorequest.piod_len = len;
if (ptrace(PT_IO, pid, (caddr_t)&iorequest, 0) < 0)
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
return (-1);
return (0);
Teach truss(1) to display sockaddrs. It currently knows about AF_INET, AF_INET6 and AF_UNIX sockaddrs, and will recognize accept(), bind(), connect(), getpeername() and getsockname() as syscalls taking sockaddr arguments. Some enterprising soul might want to add (and test) support for the send() / recv() family of syscalls as well. MFC after: 1 week
2001-10-21 21:57:10 +00:00
}
- Style(9) cleanup. Approved by: kib (mentor)
2012-09-02 11:03:18 +00:00
#define MAXSIZE 4096
Rework get_string() to make it more robust when fetching strings of unknown length. In particular, instead of blinding fetching 1k blocks, do an initial fetch up to the end of the current page followed by page-sized fetches up to the maximum size. Previously if the 1k buffer crossed a page boundary and the second page was not valid, the entire operation would fail.
2015-08-05 18:14:01 +00:00
Truss program. Requires procfs.
1997-12-06 05:23:12 +00:00
/*
* Copy a string from the process. Note that it is
* expected to be a C string, but if max is set, it will
* only get that much.
*/
Make use of ptrace(2) instead of procfs in truss(1), eliminating yet another need of an available /proc/ mount. Tested with: make universe Submitted by: howardsu Reviewed by: alfred
2007-04-10 04:03:34 +00:00
static char *
Rework get_string() to make it more robust when fetching strings of unknown length. In particular, instead of blinding fetching 1k blocks, do an initial fetch up to the end of the current page followed by page-sized fetches up to the maximum size. Previously if the 1k buffer crossed a page boundary and the second page was not valid, the entire operation would fail.
2015-08-05 18:14:01 +00:00
get_string(pid_t pid, void *addr, int max)
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
{
Make use of ptrace(2) instead of procfs in truss(1), eliminating yet another need of an available /proc/ mount. Tested with: make universe Submitted by: howardsu Reviewed by: alfred
2007-04-10 04:03:34 +00:00
struct ptrace_io_desc iorequest;
Rework get_string() to make it more robust when fetching strings of unknown length. In particular, instead of blinding fetching 1k blocks, do an initial fetch up to the end of the current page followed by page-sized fetches up to the maximum size. Previously if the 1k buffer crossed a page boundary and the second page was not valid, the entire operation would fail.
2015-08-05 18:14:01 +00:00
char *buf, *nbuf;
size_t offset, size, totalsize;
offset = 0;
if (max)
size = max + 1;
else {
/* Read up to the end of the current page. */
size = PAGE_SIZE - ((uintptr_t)addr % PAGE_SIZE);
if (size > MAXSIZE)
size = MAXSIZE;
}
totalsize = size;
Make use of ptrace(2) instead of procfs in truss(1), eliminating yet another need of an available /proc/ mount. Tested with: make universe Submitted by: howardsu Reviewed by: alfred
2007-04-10 04:03:34 +00:00
buf = malloc(totalsize);
if (buf == NULL)
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
return (NULL);
for (;;) {
Make use of ptrace(2) instead of procfs in truss(1), eliminating yet another need of an available /proc/ mount. Tested with: make universe Submitted by: howardsu Reviewed by: alfred
2007-04-10 04:03:34 +00:00
iorequest.piod_op = PIOD_READ_D;
Rework get_string() to make it more robust when fetching strings of unknown length. In particular, instead of blinding fetching 1k blocks, do an initial fetch up to the end of the current page followed by page-sized fetches up to the maximum size. Previously if the 1k buffer crossed a page boundary and the second page was not valid, the entire operation would fail.
2015-08-05 18:14:01 +00:00
iorequest.piod_offs = (char *)addr + offset;
iorequest.piod_addr = buf + offset;
Make use of ptrace(2) instead of procfs in truss(1), eliminating yet another need of an available /proc/ mount. Tested with: make universe Submitted by: howardsu Reviewed by: alfred
2007-04-10 04:03:34 +00:00
iorequest.piod_len = size;
if (ptrace(PT_IO, pid, (caddr_t)&iorequest, 0) < 0) {
free(buf);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
return (NULL);
Make use of ptrace(2) instead of procfs in truss(1), eliminating yet another need of an available /proc/ mount. Tested with: make universe Submitted by: howardsu Reviewed by: alfred
2007-04-10 04:03:34 +00:00
}
Rework get_string() to make it more robust when fetching strings of unknown length. In particular, instead of blinding fetching 1k blocks, do an initial fetch up to the end of the current page followed by page-sized fetches up to the maximum size. Previously if the 1k buffer crossed a page boundary and the second page was not valid, the entire operation would fail.
2015-08-05 18:14:01 +00:00
if (memchr(buf + offset, '\0', size) != NULL)
return (buf);
offset += size;
if (totalsize < MAXSIZE && max == 0) {
size = MAXSIZE - totalsize;
if (size > PAGE_SIZE)
size = PAGE_SIZE;
nbuf = realloc(buf, totalsize + size);
if (nbuf == NULL) {
buf[totalsize - 1] = '\0';
Make use of ptrace(2) instead of procfs in truss(1), eliminating yet another need of an available /proc/ mount. Tested with: make universe Submitted by: howardsu Reviewed by: alfred
2007-04-10 04:03:34 +00:00
return (buf);
Rework get_string() to make it more robust when fetching strings of unknown length. In particular, instead of blinding fetching 1k blocks, do an initial fetch up to the end of the current page followed by page-sized fetches up to the maximum size. Previously if the 1k buffer crossed a page boundary and the second page was not valid, the entire operation would fail.
2015-08-05 18:14:01 +00:00
}
buf = nbuf;
totalsize += size;
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
} else {
Don't write the terminating NUL past end of buffer. PR: bin/152345 Submitted by: Mateusz Guzik
2010-12-06 09:18:11 +00:00
buf[totalsize - 1] = '\0';
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
return (buf);
Truss program. Requires procfs.
1997-12-06 05:23:12 +00:00
}
}
}
Move mksubr from kdump into libsysdecode. Restructure this script so that it generates a header of tables instead of a source file. The tables are included in a flags.c source file which provides functions to decode various system call arguments. For functions that decode an enumeration, the function returns a pointer to a string for known values and NULL for unknown values. For functions that do more complex decoding (typically of a bitmask), the function accepts a pointer to a FILE object (open_memstream() can be used as a string builder) to which decoded values are written. If the function operates on a bitmask, the function returns true if any bits were decoded or false if the entire value was valid. Additionally, the third argument accepts a pointer to a value to which any undecoded bits are stored. This pointer can be NULL if the caller doesn't care about remaining bits. Convert kdump over to using decoder functions from libsysdecode instead of mksubr. truss also uses decoders from libsysdecode instead of private lookup tables, though lookup tables for objects not decoded by kdump remain in truss for now. Eventually most of these tables should move into libsysdecode as the automated table generation approach from mksubr is less stale than the static tables in truss. Some changes have been made to truss and kdump output: - The flags passed to open() are now properly decoded in that one of O_RDONLY, O_RDWR, O_WRONLY, or O_EXEC is always included in a decoded mask. - Optional arguments to open(), openat(), and fcntl() are only printed in kdump if they exist (e.g. the mode is only printed for open() if O_CREAT is set in the flags). - Print argument to F_GETLK/SETLK/SETLKW in kdump as a pointer, not int. - Include all procctl() commands. - Correctly decode pipe2() flags in truss by not assuming full open()-like flags with O_RDONLY, etc. - Decode file flags passed to *chflags() as file flags (UF_* and SF_*) rather than as a file mode. - Fix decoding of quotactl() commands by splitting out the two command components instead of assuming the raw command value matches the primary command component. In addition, truss and kdump now build without triggering any warnings. All of the sysdecode manpages now include the required headers in the synopsis. Reviewed by: kib (several older versions), wblock (manpages) MFC after: 2 months Differential Revision: https://reviews.freebsd.org/D7847
2016-10-17 22:37:07 +00:00
static const char *
- Decode the idtype argument passed to wait6() in kdump and truss. - Don't treat an options argument of 0 to wait4() as an error in kdump. - Decode the wait options passed to wait4() and wait6() in truss and decode the returned rusage and exit status. Approved by: re (kib) MFC after: 1 week
2013-09-12 18:08:25 +00:00
strsig2(int sig)
{
Move mksubr from kdump into libsysdecode. Restructure this script so that it generates a header of tables instead of a source file. The tables are included in a flags.c source file which provides functions to decode various system call arguments. For functions that decode an enumeration, the function returns a pointer to a string for known values and NULL for unknown values. For functions that do more complex decoding (typically of a bitmask), the function accepts a pointer to a FILE object (open_memstream() can be used as a string builder) to which decoded values are written. If the function operates on a bitmask, the function returns true if any bits were decoded or false if the entire value was valid. Additionally, the third argument accepts a pointer to a value to which any undecoded bits are stored. This pointer can be NULL if the caller doesn't care about remaining bits. Convert kdump over to using decoder functions from libsysdecode instead of mksubr. truss also uses decoders from libsysdecode instead of private lookup tables, though lookup tables for objects not decoded by kdump remain in truss for now. Eventually most of these tables should move into libsysdecode as the automated table generation approach from mksubr is less stale than the static tables in truss. Some changes have been made to truss and kdump output: - The flags passed to open() are now properly decoded in that one of O_RDONLY, O_RDWR, O_WRONLY, or O_EXEC is always included in a decoded mask. - Optional arguments to open(), openat(), and fcntl() are only printed in kdump if they exist (e.g. the mode is only printed for open() if O_CREAT is set in the flags). - Print argument to F_GETLK/SETLK/SETLKW in kdump as a pointer, not int. - Include all procctl() commands. - Correctly decode pipe2() flags in truss by not assuming full open()-like flags with O_RDONLY, etc. - Decode file flags passed to *chflags() as file flags (UF_* and SF_*) rather than as a file mode. - Fix decoding of quotactl() commands by splitting out the two command components instead of assuming the raw command value matches the primary command component. In addition, truss and kdump now build without triggering any warnings. All of the sysdecode manpages now include the required headers in the synopsis. Reviewed by: kib (several older versions), wblock (manpages) MFC after: 2 months Differential Revision: https://reviews.freebsd.org/D7847
2016-10-17 22:37:07 +00:00
static char tmp[32];
const char *signame;
- Decode the idtype argument passed to wait6() in kdump and truss. - Don't treat an options argument of 0 to wait4() as an error in kdump. - Decode the wait options passed to wait4() and wait6() in truss and decode the returned rusage and exit status. Approved by: re (kib) MFC after: 1 week
2013-09-12 18:08:25 +00:00
Move mksubr from kdump into libsysdecode. Restructure this script so that it generates a header of tables instead of a source file. The tables are included in a flags.c source file which provides functions to decode various system call arguments. For functions that decode an enumeration, the function returns a pointer to a string for known values and NULL for unknown values. For functions that do more complex decoding (typically of a bitmask), the function accepts a pointer to a FILE object (open_memstream() can be used as a string builder) to which decoded values are written. If the function operates on a bitmask, the function returns true if any bits were decoded or false if the entire value was valid. Additionally, the third argument accepts a pointer to a value to which any undecoded bits are stored. This pointer can be NULL if the caller doesn't care about remaining bits. Convert kdump over to using decoder functions from libsysdecode instead of mksubr. truss also uses decoders from libsysdecode instead of private lookup tables, though lookup tables for objects not decoded by kdump remain in truss for now. Eventually most of these tables should move into libsysdecode as the automated table generation approach from mksubr is less stale than the static tables in truss. Some changes have been made to truss and kdump output: - The flags passed to open() are now properly decoded in that one of O_RDONLY, O_RDWR, O_WRONLY, or O_EXEC is always included in a decoded mask. - Optional arguments to open(), openat(), and fcntl() are only printed in kdump if they exist (e.g. the mode is only printed for open() if O_CREAT is set in the flags). - Print argument to F_GETLK/SETLK/SETLKW in kdump as a pointer, not int. - Include all procctl() commands. - Correctly decode pipe2() flags in truss by not assuming full open()-like flags with O_RDONLY, etc. - Decode file flags passed to *chflags() as file flags (UF_* and SF_*) rather than as a file mode. - Fix decoding of quotactl() commands by splitting out the two command components instead of assuming the raw command value matches the primary command component. In addition, truss and kdump now build without triggering any warnings. All of the sysdecode manpages now include the required headers in the synopsis. Reviewed by: kib (several older versions), wblock (manpages) MFC after: 2 months Differential Revision: https://reviews.freebsd.org/D7847
2016-10-17 22:37:07 +00:00
signame = sysdecode_signal(sig);
if (signame == NULL) {
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
snprintf(tmp, sizeof(tmp), "%d", sig);
Move mksubr from kdump into libsysdecode. Restructure this script so that it generates a header of tables instead of a source file. The tables are included in a flags.c source file which provides functions to decode various system call arguments. For functions that decode an enumeration, the function returns a pointer to a string for known values and NULL for unknown values. For functions that do more complex decoding (typically of a bitmask), the function accepts a pointer to a FILE object (open_memstream() can be used as a string builder) to which decoded values are written. If the function operates on a bitmask, the function returns true if any bits were decoded or false if the entire value was valid. Additionally, the third argument accepts a pointer to a value to which any undecoded bits are stored. This pointer can be NULL if the caller doesn't care about remaining bits. Convert kdump over to using decoder functions from libsysdecode instead of mksubr. truss also uses decoders from libsysdecode instead of private lookup tables, though lookup tables for objects not decoded by kdump remain in truss for now. Eventually most of these tables should move into libsysdecode as the automated table generation approach from mksubr is less stale than the static tables in truss. Some changes have been made to truss and kdump output: - The flags passed to open() are now properly decoded in that one of O_RDONLY, O_RDWR, O_WRONLY, or O_EXEC is always included in a decoded mask. - Optional arguments to open(), openat(), and fcntl() are only printed in kdump if they exist (e.g. the mode is only printed for open() if O_CREAT is set in the flags). - Print argument to F_GETLK/SETLK/SETLKW in kdump as a pointer, not int. - Include all procctl() commands. - Correctly decode pipe2() flags in truss by not assuming full open()-like flags with O_RDONLY, etc. - Decode file flags passed to *chflags() as file flags (UF_* and SF_*) rather than as a file mode. - Fix decoding of quotactl() commands by splitting out the two command components instead of assuming the raw command value matches the primary command component. In addition, truss and kdump now build without triggering any warnings. All of the sysdecode manpages now include the required headers in the synopsis. Reviewed by: kib (several older versions), wblock (manpages) MFC after: 2 months Differential Revision: https://reviews.freebsd.org/D7847
2016-10-17 22:37:07 +00:00
signame = tmp;
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
}
Move mksubr from kdump into libsysdecode. Restructure this script so that it generates a header of tables instead of a source file. The tables are included in a flags.c source file which provides functions to decode various system call arguments. For functions that decode an enumeration, the function returns a pointer to a string for known values and NULL for unknown values. For functions that do more complex decoding (typically of a bitmask), the function accepts a pointer to a FILE object (open_memstream() can be used as a string builder) to which decoded values are written. If the function operates on a bitmask, the function returns true if any bits were decoded or false if the entire value was valid. Additionally, the third argument accepts a pointer to a value to which any undecoded bits are stored. This pointer can be NULL if the caller doesn't care about remaining bits. Convert kdump over to using decoder functions from libsysdecode instead of mksubr. truss also uses decoders from libsysdecode instead of private lookup tables, though lookup tables for objects not decoded by kdump remain in truss for now. Eventually most of these tables should move into libsysdecode as the automated table generation approach from mksubr is less stale than the static tables in truss. Some changes have been made to truss and kdump output: - The flags passed to open() are now properly decoded in that one of O_RDONLY, O_RDWR, O_WRONLY, or O_EXEC is always included in a decoded mask. - Optional arguments to open(), openat(), and fcntl() are only printed in kdump if they exist (e.g. the mode is only printed for open() if O_CREAT is set in the flags). - Print argument to F_GETLK/SETLK/SETLKW in kdump as a pointer, not int. - Include all procctl() commands. - Correctly decode pipe2() flags in truss by not assuming full open()-like flags with O_RDONLY, etc. - Decode file flags passed to *chflags() as file flags (UF_* and SF_*) rather than as a file mode. - Fix decoding of quotactl() commands by splitting out the two command components instead of assuming the raw command value matches the primary command component. In addition, truss and kdump now build without triggering any warnings. All of the sysdecode manpages now include the required headers in the synopsis. Reviewed by: kib (several older versions), wblock (manpages) MFC after: 2 months Differential Revision: https://reviews.freebsd.org/D7847
2016-10-17 22:37:07 +00:00
return (signame);
- Decode the idtype argument passed to wait6() in kdump and truss. - Don't treat an options argument of 0 to wait4() as an error in kdump. - Decode the wait options passed to wait4() and wait6() in truss and decode the returned rusage and exit status. Approved by: re (kib) MFC after: 1 week
2013-09-12 18:08:25 +00:00
}
Truss program. Requires procfs.
1997-12-06 05:23:12 +00:00
Expand the decoding of kevent structures. - Print the ident value as decimal instead of hexadecimal for filter types that use "small" values such as file descriptors and PIDs. - Decode NOTE_* flags in the fflags field of kevents for several system filter types.
2015-08-19 01:44:56 +00:00
static void
Decode kevent structures logged via ktrace(2) in kdump. - Add a new KTR_STRUCT_ARRAY ktrace record type which dumps an array of structures. The structure name in the record payload is preceded by a size_t containing the size of the individual structures. Use this to replace the previous code that dumped the kevent arrays dumped for kevent(). kdump is now able to decode the kevent structures rather than dumping their contents via a hexdump. One change from before is that the 'changes' and 'events' arrays are not marked with separate 'read' and 'write' annotations in kdump output. Instead, the first array is the 'changes' array, and the second array (only present if kevent doesn't fail with an error) is the 'events' array. For kevent(), empty arrays are denoted by an entry with an array containing zero entries rather than no record. - Move kevent decoding tables from truss to libsysdecode. This adds three new functions to decode members of struct kevent: sysdecode_kevent_filter, sysdecode_kevent_flags, and sysdecode_kevent_fflags. kdump uses these helper functions to pretty-print kevent fields. - Move structure definitions for freebsd11 and freebsd32 kevent structures to <sys/event.h> so that they can be shared with userland. The 32-bit structures are only exposed if _WANT_KEVENT32 is defined. The freebsd11 structures are only exposed if _WANT_FREEBSD11_KEVENT is defined. The 32-bit freebsd11 structure requires both. - Decode freebsd11 kevent structures in truss for the compat11.kevent() system call. - Log 32-bit kevent structures via ktrace for 32-bit compat kevent() system calls. - While here, constify the 'void *data' argument to ktrstruct(). Reviewed by: kib (earlier version) MFC after: 1 month Differential Revision: https://reviews.freebsd.org/D12470
2017-11-25 04:49:12 +00:00
print_kevent(FILE *fp, struct kevent *ke)
Expand the decoding of kevent structures. - Print the ident value as decimal instead of hexadecimal for filter types that use "small" values such as file descriptors and PIDs. - Decode NOTE_* flags in the fflags field of kevents for several system filter types.
2015-08-19 01:44:56 +00:00
{
switch (ke->filter) {
case EVFILT_READ:
case EVFILT_WRITE:
case EVFILT_VNODE:
case EVFILT_PROC:
case EVFILT_TIMER:
case EVFILT_PROCDESC:
Decode kevent structures logged via ktrace(2) in kdump. - Add a new KTR_STRUCT_ARRAY ktrace record type which dumps an array of structures. The structure name in the record payload is preceded by a size_t containing the size of the individual structures. Use this to replace the previous code that dumped the kevent arrays dumped for kevent(). kdump is now able to decode the kevent structures rather than dumping their contents via a hexdump. One change from before is that the 'changes' and 'events' arrays are not marked with separate 'read' and 'write' annotations in kdump output. Instead, the first array is the 'changes' array, and the second array (only present if kevent doesn't fail with an error) is the 'events' array. For kevent(), empty arrays are denoted by an entry with an array containing zero entries rather than no record. - Move kevent decoding tables from truss to libsysdecode. This adds three new functions to decode members of struct kevent: sysdecode_kevent_filter, sysdecode_kevent_flags, and sysdecode_kevent_fflags. kdump uses these helper functions to pretty-print kevent fields. - Move structure definitions for freebsd11 and freebsd32 kevent structures to <sys/event.h> so that they can be shared with userland. The 32-bit structures are only exposed if _WANT_KEVENT32 is defined. The freebsd11 structures are only exposed if _WANT_FREEBSD11_KEVENT is defined. The 32-bit freebsd11 structure requires both. - Decode freebsd11 kevent structures in truss for the compat11.kevent() system call. - Log 32-bit kevent structures via ktrace for 32-bit compat kevent() system calls. - While here, constify the 'void *data' argument to ktrstruct(). Reviewed by: kib (earlier version) MFC after: 1 month Differential Revision: https://reviews.freebsd.org/D12470
2017-11-25 04:49:12 +00:00
case EVFILT_EMPTY:
Expand the decoding of kevent structures. - Print the ident value as decimal instead of hexadecimal for filter types that use "small" values such as file descriptors and PIDs. - Decode NOTE_* flags in the fflags field of kevents for several system filter types.
2015-08-19 01:44:56 +00:00
fprintf(fp, "%ju", (uintmax_t)ke->ident);
break;
case EVFILT_SIGNAL:
fputs(strsig2(ke->ident), fp);
break;
default:
fprintf(fp, "%p", (void *)ke->ident);
}
Decode kevent structures logged via ktrace(2) in kdump. - Add a new KTR_STRUCT_ARRAY ktrace record type which dumps an array of structures. The structure name in the record payload is preceded by a size_t containing the size of the individual structures. Use this to replace the previous code that dumped the kevent arrays dumped for kevent(). kdump is now able to decode the kevent structures rather than dumping their contents via a hexdump. One change from before is that the 'changes' and 'events' arrays are not marked with separate 'read' and 'write' annotations in kdump output. Instead, the first array is the 'changes' array, and the second array (only present if kevent doesn't fail with an error) is the 'events' array. For kevent(), empty arrays are denoted by an entry with an array containing zero entries rather than no record. - Move kevent decoding tables from truss to libsysdecode. This adds three new functions to decode members of struct kevent: sysdecode_kevent_filter, sysdecode_kevent_flags, and sysdecode_kevent_fflags. kdump uses these helper functions to pretty-print kevent fields. - Move structure definitions for freebsd11 and freebsd32 kevent structures to <sys/event.h> so that they can be shared with userland. The 32-bit structures are only exposed if _WANT_KEVENT32 is defined. The freebsd11 structures are only exposed if _WANT_FREEBSD11_KEVENT is defined. The 32-bit freebsd11 structure requires both. - Decode freebsd11 kevent structures in truss for the compat11.kevent() system call. - Log 32-bit kevent structures via ktrace for 32-bit compat kevent() system calls. - While here, constify the 'void *data' argument to ktrstruct(). Reviewed by: kib (earlier version) MFC after: 1 month Differential Revision: https://reviews.freebsd.org/D12470
2017-11-25 04:49:12 +00:00
fprintf(fp, ",");
print_integer_arg(sysdecode_kevent_filter, fp, ke->filter);
fprintf(fp, ",");
print_mask_arg(sysdecode_kevent_flags, fp, ke->flags);
fprintf(fp, ",");
sysdecode_kevent_fflags(fp, ke->filter, ke->fflags, 16);
Add abstime kqueue(2) timers and expand struct kevent members. This change implements NOTE_ABSTIME flag for EVFILT_TIMER, which specifies that the data field contains absolute time to fire the event. To make this useful, data member of the struct kevent must be extended to 64bit. Using the opportunity, I also added ext members. This changes struct kevent almost to Apple struct kevent64, except I did not changed type of ident and udata, the later would cause serious API incompatibilities. The type of ident was kept uintptr_t since EVFILT_AIO returns a pointer in this field, and e.g. CHERI is sensitive to the type (discussed with brooks, jhb). Unlike Apple kevent64, symbol versioning allows us to claim ABI compatibility and still name the new syscall kevent(2). Compat shims are provided for both host native and compat32. Requested by: bapt Reviewed by: bapt, brooks, ngie (previous version) Sponsored by: The FreeBSD Foundation Differential revision: https://reviews.freebsd.org/D11025
2017-06-17 00:57:26 +00:00
fprintf(fp, ",%#jx,%p", (uintmax_t)ke->data, ke->udata);
Expand the decoding of kevent structures. - Print the ident value as decimal instead of hexadecimal for filter types that use "small" values such as file descriptors and PIDs. - Decode NOTE_* flags in the fflags field of kevents for several system filter types.
2015-08-19 01:44:56 +00:00
}
truss: Add support for utrace(2). This uses the kdump(1) utrace support code directly until a common library is created. This allows malloc(3) tracing with MALLOC_CONF=utrace:true and rtld tracing with LD_UTRACE=1. Unknown utrace(2) data is just printed as hex. PR: 43819 [inspired by] Reviewed by: jhb MFC after: 2 weeks Relnotes: yes Differential Revision: https://reviews.freebsd.org/D3819
2015-10-06 21:58:38 +00:00
static void
print_utrace(FILE *fp, void *utrace_addr, size_t len)
{
unsigned char *utrace_buffer;
fprintf(fp, "{ ");
Start on a new library (libsysdecode) that provides routines for decoding system call information such as system call arguments. Initially this will consist of pulling duplicated code out of truss and kdump though it may prove useful for other utilities in the future. This commit moves the shared utrace(2) record parser out of kdump into the library and updates kdump and truss to use it. One difference from the previous version is that the library version treats unknown events that start with the "RTLD" signature as unknown events. This simplifies the interface and allows the consumer to decide how to handle all non-recognized events. Instead, this function only generates a string description for known malloc() and RTLD records. Reviewed by: bdrewery Differential Revision: https://reviews.freebsd.org/D4537
2015-12-15 00:05:07 +00:00
if (sysdecode_utrace(fp, utrace_addr, len)) {
truss: Add support for utrace(2). This uses the kdump(1) utrace support code directly until a common library is created. This allows malloc(3) tracing with MALLOC_CONF=utrace:true and rtld tracing with LD_UTRACE=1. Unknown utrace(2) data is just printed as hex. PR: 43819 [inspired by] Reviewed by: jhb MFC after: 2 weeks Relnotes: yes Differential Revision: https://reviews.freebsd.org/D3819
2015-10-06 21:58:38 +00:00
fprintf(fp, " }");
return;
}
utrace_buffer = utrace_addr;
fprintf(fp, "%zu:", len);
while (len--)
fprintf(fp, " %02x", *utrace_buffer++);
fprintf(fp, " }");
}
Decode msghdr argument of sendmsg() and recvmsg(). Sponsored by: Netflix, Inc.
2018-01-15 20:39:42 +00:00
static void
print_sockaddr(FILE *fp, struct trussinfo *trussinfo, void *arg, socklen_t len)
{
char addr[64];
struct sockaddr_in *lsin;
struct sockaddr_in6 *lsin6;
struct sockaddr_un *sun;
struct sockaddr *sa;
u_char *q;
pid_t pid = trussinfo->curthread->proc->pid;
if (arg == NULL) {
fputs("NULL", fp);
return;
}
/* If the length is too small, just bail. */
if (len < sizeof(*sa)) {
Using %p already prints "0x", so don't do it explicitly.
2018-01-16 19:57:30 +00:00
fprintf(fp, "%p", arg);
Decode msghdr argument of sendmsg() and recvmsg(). Sponsored by: Netflix, Inc.
2018-01-15 20:39:42 +00:00
return;
}
sa = calloc(1, len);
if (get_struct(pid, arg, sa, len) == -1) {
free(sa);
Using %p already prints "0x", so don't do it explicitly.
2018-01-16 19:57:30 +00:00
fprintf(fp, "%p", arg);
Decode msghdr argument of sendmsg() and recvmsg(). Sponsored by: Netflix, Inc.
2018-01-15 20:39:42 +00:00
return;
}
switch (sa->sa_family) {
case AF_INET:
if (len < sizeof(*lsin))
goto sockaddr_short;
lsin = (struct sockaddr_in *)(void *)sa;
inet_ntop(AF_INET, &lsin->sin_addr, addr, sizeof(addr));
fprintf(fp, "{ AF_INET %s:%d }", addr,
htons(lsin->sin_port));
break;
case AF_INET6:
if (len < sizeof(*lsin6))
goto sockaddr_short;
lsin6 = (struct sockaddr_in6 *)(void *)sa;
inet_ntop(AF_INET6, &lsin6->sin6_addr, addr,
sizeof(addr));
fprintf(fp, "{ AF_INET6 [%s]:%d }", addr,
htons(lsin6->sin6_port));
break;
case AF_UNIX:
sun = (struct sockaddr_un *)sa;
fprintf(fp, "{ AF_UNIX \"%.*s\" }",
(int)(len - offsetof(struct sockaddr_un, sun_path)),
sun->sun_path);
break;
default:
sockaddr_short:
fprintf(fp,
"{ sa_len = %d, sa_family = %d, sa_data = {",
(int)sa->sa_len, (int)sa->sa_family);
for (q = (u_char *)sa->sa_data;
q < (u_char *)sa + len; q++)
fprintf(fp, "%s 0x%02x",
q == (u_char *)sa->sa_data ? "" : ",",
*q);
fputs(" } }", fp);
}
free(sa);
}
#define IOV_LIMIT 16
static void
print_iovec(FILE *fp, struct trussinfo *trussinfo, void *arg, int iovcnt)
{
struct iovec iov[IOV_LIMIT];
size_t max_string = trussinfo->strsize;
char tmp2[max_string + 1], *tmp3;
size_t len;
pid_t pid = trussinfo->curthread->proc->pid;
int i;
bool buf_truncated, iov_truncated;
if (iovcnt <= 0) {
Using %p already prints "0x", so don't do it explicitly.
2018-01-16 19:57:30 +00:00
fprintf(fp, "%p", arg);
Decode msghdr argument of sendmsg() and recvmsg(). Sponsored by: Netflix, Inc.
2018-01-15 20:39:42 +00:00
return;
}
if (iovcnt > IOV_LIMIT) {
iovcnt = IOV_LIMIT;
iov_truncated = true;
} else {
iov_truncated = false;
}
if (get_struct(pid, arg, &iov, iovcnt * sizeof(struct iovec)) == -1) {
Using %p already prints "0x", so don't do it explicitly.
2018-01-16 19:57:30 +00:00
fprintf(fp, "%p", arg);
Decode msghdr argument of sendmsg() and recvmsg(). Sponsored by: Netflix, Inc.
2018-01-15 20:39:42 +00:00
return;
}
fputs("[", fp);
for (i = 0; i < iovcnt; i++) {
len = iov[i].iov_len;
if (len > max_string) {
len = max_string;
buf_truncated = true;
} else {
buf_truncated = false;
}
fprintf(fp, "%s{", (i > 0) ? "," : "");
if (len && get_struct(pid, iov[i].iov_base, &tmp2, len) != -1) {
tmp3 = malloc(len * 4 + 1);
while (len) {
if (strvisx(tmp3, tmp2, len,
VIS_CSTYLE|VIS_TAB|VIS_NL) <=
(int)max_string)
break;
len--;
buf_truncated = true;
}
fprintf(fp, "\"%s\"%s", tmp3,
buf_truncated ? "..." : "");
free(tmp3);
} else {
Using %p already prints "0x", so don't do it explicitly.
2018-01-16 19:57:30 +00:00
fprintf(fp, "%p", iov[i].iov_base);
Decode msghdr argument of sendmsg() and recvmsg(). Sponsored by: Netflix, Inc.
2018-01-15 20:39:42 +00:00
}
fprintf(fp, ",%zu}", iov[i].iov_len);
}
fprintf(fp, "%s%s", iov_truncated ? ",..." : "", "]");
}
static void
print_gen_cmsg(FILE *fp, struct cmsghdr *cmsghdr)
{
u_char *q;
fputs("{", fp);
for (q = CMSG_DATA(cmsghdr);
q < (u_char *)cmsghdr + cmsghdr->cmsg_len; q++) {
fprintf(fp, "%s0x%02x", q == CMSG_DATA(cmsghdr) ? "" : ",", *q);
}
fputs("}", fp);
}
static void
print_sctp_initmsg(FILE *fp, struct sctp_initmsg *init)
{
fprintf(fp, "{out=%u,", init->sinit_num_ostreams);
fprintf(fp, "in=%u,", init->sinit_max_instreams);
fprintf(fp, "max_rtx=%u,", init->sinit_max_attempts);
fprintf(fp, "max_rto=%u}", init->sinit_max_init_timeo);
}
static void
Rename 'recv' to 'receive' to appease shadow warnings from GCC.
2018-01-16 01:21:07 +00:00
print_sctp_sndrcvinfo(FILE *fp, bool receive, struct sctp_sndrcvinfo *info)
Decode msghdr argument of sendmsg() and recvmsg(). Sponsored by: Netflix, Inc.
2018-01-15 20:39:42 +00:00
{
fprintf(fp, "{sid=%u,", info->sinfo_stream);
Rename 'recv' to 'receive' to appease shadow warnings from GCC.
2018-01-16 01:21:07 +00:00
if (receive) {
Decode msghdr argument of sendmsg() and recvmsg(). Sponsored by: Netflix, Inc.
2018-01-15 20:39:42 +00:00
fprintf(fp, "ssn=%u,", info->sinfo_ssn);
}
fputs("flgs=", fp);
sysdecode_sctp_sinfo_flags(fp, info->sinfo_flags);
fprintf(fp, ",ppid=%u,", ntohl(info->sinfo_ppid));
Rename 'recv' to 'receive' to appease shadow warnings from GCC.
2018-01-16 01:21:07 +00:00
if (!receive) {
Decode msghdr argument of sendmsg() and recvmsg(). Sponsored by: Netflix, Inc.
2018-01-15 20:39:42 +00:00
fprintf(fp, "ctx=%u,", info->sinfo_context);
fprintf(fp, "ttl=%u,", info->sinfo_timetolive);
}
Rename 'recv' to 'receive' to appease shadow warnings from GCC.
2018-01-16 01:21:07 +00:00
if (receive) {
Decode msghdr argument of sendmsg() and recvmsg(). Sponsored by: Netflix, Inc.
2018-01-15 20:39:42 +00:00
fprintf(fp, "tsn=%u,", info->sinfo_tsn);
fprintf(fp, "cumtsn=%u,", info->sinfo_cumtsn);
}
fprintf(fp, "id=%u}", info->sinfo_assoc_id);
}
static void
print_sctp_sndinfo(FILE *fp, struct sctp_sndinfo *info)
{
fprintf(fp, "{sid=%u,", info->snd_sid);
fputs("flgs=", fp);
print_mask_arg(sysdecode_sctp_snd_flags, fp, info->snd_flags);
fprintf(fp, ",ppid=%u,", ntohl(info->snd_ppid));
fprintf(fp, "ctx=%u,", info->snd_context);
fprintf(fp, "id=%u}", info->snd_assoc_id);
}
static void
print_sctp_rcvinfo(FILE *fp, struct sctp_rcvinfo *info)
{
fprintf(fp, "{sid=%u,", info->rcv_sid);
fprintf(fp, "ssn=%u,", info->rcv_ssn);
fputs("flgs=", fp);
print_mask_arg(sysdecode_sctp_rcv_flags, fp, info->rcv_flags);
fprintf(fp, ",ppid=%u,", ntohl(info->rcv_ppid));
fprintf(fp, "tsn=%u,", info->rcv_tsn);
fprintf(fp, "cumtsn=%u,", info->rcv_cumtsn);
fprintf(fp, "ctx=%u,", info->rcv_context);
fprintf(fp, "id=%u}", info->rcv_assoc_id);
}
static void
print_sctp_nxtinfo(FILE *fp, struct sctp_nxtinfo *info)
{
fprintf(fp, "{sid=%u,", info->nxt_sid);
fputs("flgs=", fp);
print_mask_arg(sysdecode_sctp_nxt_flags, fp, info->nxt_flags);
fprintf(fp, ",ppid=%u,", ntohl(info->nxt_ppid));
fprintf(fp, "len=%u,", info->nxt_length);
fprintf(fp, "id=%u}", info->nxt_assoc_id);
}
static void
print_sctp_prinfo(FILE *fp, struct sctp_prinfo *info)
{
fputs("{pol=", fp);
print_integer_arg(sysdecode_sctp_pr_policy, fp, info->pr_policy);
fprintf(fp, ",val=%u}", info->pr_value);
}
static void
print_sctp_authinfo(FILE *fp, struct sctp_authinfo *info)
{
fprintf(fp, "{num=%u}", info->auth_keynumber);
}
static void
print_sctp_ipv4_addr(FILE *fp, struct in_addr *addr)
{
char buf[INET_ADDRSTRLEN];
const char *s;
s = inet_ntop(AF_INET, addr, buf, INET_ADDRSTRLEN);
if (s != NULL)
fprintf(fp, "{addr=%s}", s);
else
fputs("{addr=???}", fp);
}
static void
print_sctp_ipv6_addr(FILE *fp, struct in6_addr *addr)
{
char buf[INET6_ADDRSTRLEN];
const char *s;
s = inet_ntop(AF_INET6, addr, buf, INET6_ADDRSTRLEN);
if (s != NULL)
fprintf(fp, "{addr=%s}", s);
else
fputs("{addr=???}", fp);
}
static void
Rename 'recv' to 'receive' to appease shadow warnings from GCC.
2018-01-16 01:21:07 +00:00
print_sctp_cmsg(FILE *fp, bool receive, struct cmsghdr *cmsghdr)
Decode msghdr argument of sendmsg() and recvmsg(). Sponsored by: Netflix, Inc.
2018-01-15 20:39:42 +00:00
{
void *data;
socklen_t len;
len = cmsghdr->cmsg_len;
data = CMSG_DATA(cmsghdr);
switch (cmsghdr->cmsg_type) {
case SCTP_INIT:
if (len == CMSG_LEN(sizeof(struct sctp_initmsg)))
print_sctp_initmsg(fp, (struct sctp_initmsg *)data);
else
print_gen_cmsg(fp, cmsghdr);
break;
case SCTP_SNDRCV:
if (len == CMSG_LEN(sizeof(struct sctp_sndrcvinfo)))
Rename 'recv' to 'receive' to appease shadow warnings from GCC.
2018-01-16 01:21:07 +00:00
print_sctp_sndrcvinfo(fp, receive,
Decode msghdr argument of sendmsg() and recvmsg(). Sponsored by: Netflix, Inc.
2018-01-15 20:39:42 +00:00
(struct sctp_sndrcvinfo *)data);
else
print_gen_cmsg(fp, cmsghdr);
break;
#if 0
case SCTP_EXTRCV:
if (len == CMSG_LEN(sizeof(struct sctp_extrcvinfo)))
print_sctp_extrcvinfo(fp,
(struct sctp_extrcvinfo *)data);
else
print_gen_cmsg(fp, cmsghdr);
break;
#endif
case SCTP_SNDINFO:
if (len == CMSG_LEN(sizeof(struct sctp_sndinfo)))
print_sctp_sndinfo(fp, (struct sctp_sndinfo *)data);
else
print_gen_cmsg(fp, cmsghdr);
break;
case SCTP_RCVINFO:
if (len == CMSG_LEN(sizeof(struct sctp_rcvinfo)))
print_sctp_rcvinfo(fp, (struct sctp_rcvinfo *)data);
else
print_gen_cmsg(fp, cmsghdr);
break;
case SCTP_NXTINFO:
if (len == CMSG_LEN(sizeof(struct sctp_nxtinfo)))
print_sctp_nxtinfo(fp, (struct sctp_nxtinfo *)data);
else
print_gen_cmsg(fp, cmsghdr);
break;
case SCTP_PRINFO:
if (len == CMSG_LEN(sizeof(struct sctp_prinfo)))
print_sctp_prinfo(fp, (struct sctp_prinfo *)data);
else
print_gen_cmsg(fp, cmsghdr);
break;
case SCTP_AUTHINFO:
if (len == CMSG_LEN(sizeof(struct sctp_authinfo)))
print_sctp_authinfo(fp, (struct sctp_authinfo *)data);
else
print_gen_cmsg(fp, cmsghdr);
break;
case SCTP_DSTADDRV4:
if (len == CMSG_LEN(sizeof(struct in_addr)))
print_sctp_ipv4_addr(fp, (struct in_addr *)data);
else
print_gen_cmsg(fp, cmsghdr);
break;
case SCTP_DSTADDRV6:
if (len == CMSG_LEN(sizeof(struct in6_addr)))
print_sctp_ipv6_addr(fp, (struct in6_addr *)data);
else
print_gen_cmsg(fp, cmsghdr);
break;
default:
print_gen_cmsg(fp, cmsghdr);
}
}
static void
Rename 'recv' to 'receive' to appease shadow warnings from GCC.
2018-01-16 01:21:07 +00:00
print_cmsgs(FILE *fp, pid_t pid, bool receive, struct msghdr *msghdr)
Decode msghdr argument of sendmsg() and recvmsg(). Sponsored by: Netflix, Inc.
2018-01-15 20:39:42 +00:00
{
struct cmsghdr *cmsghdr;
char *cmsgbuf;
const char *temp;
socklen_t len;
int level, type;
bool first;
len = msghdr->msg_controllen;
Improve the printing of cmgs when the length is 0. Fix error handling.
2018-01-16 20:02:07 +00:00
if (len == 0) {
fputs("{}", fp);
return;
}
Decode msghdr argument of sendmsg() and recvmsg(). Sponsored by: Netflix, Inc.
2018-01-15 20:39:42 +00:00
cmsgbuf = calloc(1, len);
if (get_struct(pid, msghdr->msg_control, cmsgbuf, len) == -1) {
Using %p already prints "0x", so don't do it explicitly.
2018-01-16 19:57:30 +00:00
fprintf(fp, "%p", msghdr->msg_control);
Decode msghdr argument of sendmsg() and recvmsg(). Sponsored by: Netflix, Inc.
2018-01-15 20:39:42 +00:00
free(cmsgbuf);
Improve the printing of cmgs when the length is 0. Fix error handling.
2018-01-16 20:02:07 +00:00
return;
Decode msghdr argument of sendmsg() and recvmsg(). Sponsored by: Netflix, Inc.
2018-01-15 20:39:42 +00:00
}
msghdr->msg_control = cmsgbuf;
first = true;
fputs("{", fp);
for (cmsghdr = CMSG_FIRSTHDR(msghdr);
cmsghdr != NULL;
cmsghdr = CMSG_NXTHDR(msghdr, cmsghdr)) {
level = cmsghdr->cmsg_level;
type = cmsghdr->cmsg_type;
len = cmsghdr->cmsg_len;
fprintf(fp, "%s{level=", first ? "" : ",");
print_integer_arg(sysdecode_sockopt_level, fp, level);
fputs(",type=", fp);
temp = sysdecode_cmsg_type(level, type);
if (temp) {
fputs(temp, fp);
} else {
fprintf(fp, "%d", type);
}
fputs(",data=", fp);
switch (level) {
case IPPROTO_SCTP:
Rename 'recv' to 'receive' to appease shadow warnings from GCC.
2018-01-16 01:21:07 +00:00
print_sctp_cmsg(fp, receive, cmsghdr);
Decode msghdr argument of sendmsg() and recvmsg(). Sponsored by: Netflix, Inc.
2018-01-15 20:39:42 +00:00
break;
default:
print_gen_cmsg(fp, cmsghdr);
break;
}
fputs("}", fp);
Add missing assignment to make sure non-first cmsgs are handled as such.
2018-01-17 10:30:49 +00:00
first = false;
Decode msghdr argument of sendmsg() and recvmsg(). Sponsored by: Netflix, Inc.
2018-01-15 20:39:42 +00:00
}
fputs("}", fp);
free(cmsgbuf);
}
Truss program. Requires procfs.
1997-12-06 05:23:12 +00:00
/*
* Converts a syscall argument into a string. Said string is
Various style and whitespace fixes.
2015-08-19 20:02:03 +00:00
* allocated via malloc(), so needs to be free()'d. sc is
Truss program. Requires procfs.
1997-12-06 05:23:12 +00:00
* a pointer to the syscall description (see above); args is
* an array of all of the system call arguments.
*/
char *
Several changes to truss. - Refactor the interface between the ABI-independent code and the ABI-specific backends. The backends now provide smaller hooks to fetch system call arguments and return values. The rest of the system call entry and exit handling that was previously duplicated among all the backends has been moved to one place. - Merge the loop when waiting for an event with the loop for handling stops. This also means not emulating a procfs-like interface on top of ptrace(). Instead, use a single event loop that fetches process events via waitid(). Among other things this allows us to report the full 32-bit exit value. - Use PT_FOLLOW_FORK to follow new child processes instead of forking a new truss process for each new child. This allows one truss process to monitor a tree of processes and truss -c should now display one total for the entire tree instead of separate summaries per process. - Use the recently added fields to ptrace_lwpinfo to determine the current system call number and argument count. The latter is especially useful and fixes a regression since the conversion from procfs. truss now generally prints the correct number of arguments for most system calls rather than printing extra arguments for any call not listed in the table in syscalls.c. - Actually check the new ABI when processes call exec. The comments claimed that this happened but it was not being done (perhaps this was another regression in the conversion to ptrace()). If the new ABI after exec is not supported, truss detaches from the process. If truss does not support the ABI for a newly executed process the process is killed before it returns from exec. - Along with the refactor, teach the various ABI-specific backends to fetch both return values, not just the first. Use this to properly report the full 64-bit return value from lseek(). In addition, the handler for "pipe" now pulls the pair of descriptors out of the return values (which is the true kernel system call interface) but displays them as an argument (which matches the interface exported by libc). - Each ABI handler adds entries to a linker set rather than requiring a statically defined table of handlers in main.c. - The arm and mips system call fetching code was changed to follow the same pattern as amd64 (and the in-kernel handler) of fetching register arguments first and then reading any remaining arguments from the stack. This should fix indirect system call arguments on at least arm. - The mipsn32 and n64 ABIs will now look for arguments in A4 through A7. - Use register %ebp for the 6th system call argument for Linux/i386 ABIs to match the in-kernel argument fetch code. - For powerpc binaries on a powerpc64 system, fetch the extra arguments on the stack as 32-bit values that are then copied into the 64-bit argument array instead of reading the 32-bit values directly into the 64-bit array. Reviewed by: kib (earlier version) Tested on: amd64 (FreeBSD/amd64 & i386), i386, arm (earlier version) Tested on: powerpc64 (FreeBSD/powerpc64 & powerpc) MFC after: 1 month Differential Revision: https://reviews.freebsd.org/D3575
2015-09-30 19:13:32 +00:00
print_arg(struct syscall_args *sc, unsigned long *args, long *retval,
- Style(9) cleanup. Approved by: kib (mentor)
2012-09-02 11:03:18 +00:00
struct trussinfo *trussinfo)
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
{
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
FILE *fp;
- Style(9) cleanup. Approved by: kib (mentor)
2012-09-02 11:03:18 +00:00
char *tmp;
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
size_t tmplen;
- Style(9) cleanup. Approved by: kib (mentor)
2012-09-02 11:03:18 +00:00
pid_t pid;
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
fp = open_memstream(&tmp, &tmplen);
Several changes to truss. - Refactor the interface between the ABI-independent code and the ABI-specific backends. The backends now provide smaller hooks to fetch system call arguments and return values. The rest of the system call entry and exit handling that was previously duplicated among all the backends has been moved to one place. - Merge the loop when waiting for an event with the loop for handling stops. This also means not emulating a procfs-like interface on top of ptrace(). Instead, use a single event loop that fetches process events via waitid(). Among other things this allows us to report the full 32-bit exit value. - Use PT_FOLLOW_FORK to follow new child processes instead of forking a new truss process for each new child. This allows one truss process to monitor a tree of processes and truss -c should now display one total for the entire tree instead of separate summaries per process. - Use the recently added fields to ptrace_lwpinfo to determine the current system call number and argument count. The latter is especially useful and fixes a regression since the conversion from procfs. truss now generally prints the correct number of arguments for most system calls rather than printing extra arguments for any call not listed in the table in syscalls.c. - Actually check the new ABI when processes call exec. The comments claimed that this happened but it was not being done (perhaps this was another regression in the conversion to ptrace()). If the new ABI after exec is not supported, truss detaches from the process. If truss does not support the ABI for a newly executed process the process is killed before it returns from exec. - Along with the refactor, teach the various ABI-specific backends to fetch both return values, not just the first. Use this to properly report the full 64-bit return value from lseek(). In addition, the handler for "pipe" now pulls the pair of descriptors out of the return values (which is the true kernel system call interface) but displays them as an argument (which matches the interface exported by libc). - Each ABI handler adds entries to a linker set rather than requiring a statically defined table of handlers in main.c. - The arm and mips system call fetching code was changed to follow the same pattern as amd64 (and the in-kernel handler) of fetching register arguments first and then reading any remaining arguments from the stack. This should fix indirect system call arguments on at least arm. - The mipsn32 and n64 ABIs will now look for arguments in A4 through A7. - Use register %ebp for the 6th system call argument for Linux/i386 ABIs to match the in-kernel argument fetch code. - For powerpc binaries on a powerpc64 system, fetch the extra arguments on the stack as 32-bit values that are then copied into the 64-bit argument array instead of reading the 32-bit values directly into the 64-bit array. Reviewed by: kib (earlier version) Tested on: amd64 (FreeBSD/amd64 & i386), i386, arm (earlier version) Tested on: powerpc64 (FreeBSD/powerpc64 & powerpc) MFC after: 1 month Differential Revision: https://reviews.freebsd.org/D3575
2015-09-30 19:13:32 +00:00
pid = trussinfo->curthread->proc->pid;
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
switch (sc->type & ARG_MASK) {
case Hex:
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
fprintf(fp, "0x%x", (int)args[sc->offset]);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
break;
case Octal:
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
fprintf(fp, "0%o", (int)args[sc->offset]);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
break;
case Int:
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
fprintf(fp, "%d", (int)args[sc->offset]);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
break;
Properly format pointer size independent CloudABI system calls. CloudABI has approximately 50 system calls that do not depend on the pointer size of the system. As the ABI is pretty compact, it takes little effort to each truss(8) the formatting rules for these system calls. Start off by formatting pointer size independent system calls. Changes: - Make it possible to include the CloudABI system call definitions in FreeBSD userspace builds. Add ${root}/sys to the truss(8) Makefile so we can pull in <compat/cloudabi/cloudabi_syscalldefs.h>. - Refactoring: patch up amd64-cloudabi64.c to use the CLOUDABI_* constants instead of rolling our own table. - Add table entries for all of the system calls. - Add new generic formatting types (UInt, IntArray) that we'll be using to format unsigned integers and arrays of integers. - Add CloudABI specific formatting types. Approved by: jhb Differential Revision: https://reviews.freebsd.org/D3836
2015-10-08 05:27:45 +00:00
case UInt:
fprintf(fp, "%u", (unsigned int)args[sc->offset]);
break;
Decode the argument passed to cap_getmode(). The returned integer value is output.
2017-06-02 22:35:18 +00:00
case PUInt: {
unsigned int val;
if (get_struct(pid, (void *)args[sc->offset], &val,
sizeof(val)) == 0)
fprintf(fp, "{ %u }", val);
else
fprintf(fp, "0x%lx", args[sc->offset]);
break;
}
Decode the arguments passed to _umtx_op(). In particular, decode the opcode. MFC after: 1 week Sponsored by: Norse
2014-10-13 16:37:06 +00:00
case LongHex:
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
fprintf(fp, "0x%lx", args[sc->offset]);
- Decode the arguments for several signal-related system calls: sigpending, sigqueue, sigreturn, sigsuspend, sigtimedwait, sigwait, sigwaitinfo, and thr_kill. - Print signal sets as a structure (with {}'s) and in particular use this to differentiate empty sets from a NULL pointer. - Decode arguments for some other system calls: issetugid, pipe2, sysarch (operations are only decoded for amd64 and i386), and thr_self.
2015-08-17 17:52:28 +00:00
break;
case Long:
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
fprintf(fp, "%ld", args[sc->offset]);
- Decode the arguments for several signal-related system calls: sigpending, sigqueue, sigreturn, sigsuspend, sigtimedwait, sigwait, sigwaitinfo, and thr_kill. - Print signal sets as a structure (with {}'s) and in particular use this to differentiate empty sets from a NULL pointer. - Decode arguments for some other system calls: issetugid, pipe2, sysarch (operations are only decoded for amd64 and i386), and thr_self.
2015-08-17 17:52:28 +00:00
break;
Add a Sizet type for 'size_t' values and use it instead of Int. Various size_t arguments were previously decoded as Int values instead which would have truncated values above 2^31 on 64-bit systems.
2017-03-19 00:27:07 +00:00
case Sizet:
fprintf(fp, "%zu", (size_t)args[sc->offset]);
break;
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
case Name: {
/* NULL-terminated string. */
char *tmp2;
Various style and whitespace fixes.
2015-08-19 20:02:03 +00:00
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
tmp2 = get_string(pid, (void*)args[sc->offset], 0);
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
fprintf(fp, "\"%s\"", tmp2);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
free(tmp2);
break;
}
case BinString: {
Various style and whitespace fixes.
2015-08-19 20:02:03 +00:00
/*
* Binary block of data that might have printable characters.
* XXX If type|OUT, assume that the length is the syscall's
* return value. Otherwise, assume that the length of the block
* is in the next syscall argument.
*/
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
int max_string = trussinfo->strsize;
Various style and whitespace fixes.
2015-08-19 20:02:03 +00:00
char tmp2[max_string + 1], *tmp3;
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
int len;
int truncated = 0;
if (sc->type & OUT)
Several changes to truss. - Refactor the interface between the ABI-independent code and the ABI-specific backends. The backends now provide smaller hooks to fetch system call arguments and return values. The rest of the system call entry and exit handling that was previously duplicated among all the backends has been moved to one place. - Merge the loop when waiting for an event with the loop for handling stops. This also means not emulating a procfs-like interface on top of ptrace(). Instead, use a single event loop that fetches process events via waitid(). Among other things this allows us to report the full 32-bit exit value. - Use PT_FOLLOW_FORK to follow new child processes instead of forking a new truss process for each new child. This allows one truss process to monitor a tree of processes and truss -c should now display one total for the entire tree instead of separate summaries per process. - Use the recently added fields to ptrace_lwpinfo to determine the current system call number and argument count. The latter is especially useful and fixes a regression since the conversion from procfs. truss now generally prints the correct number of arguments for most system calls rather than printing extra arguments for any call not listed in the table in syscalls.c. - Actually check the new ABI when processes call exec. The comments claimed that this happened but it was not being done (perhaps this was another regression in the conversion to ptrace()). If the new ABI after exec is not supported, truss detaches from the process. If truss does not support the ABI for a newly executed process the process is killed before it returns from exec. - Along with the refactor, teach the various ABI-specific backends to fetch both return values, not just the first. Use this to properly report the full 64-bit return value from lseek(). In addition, the handler for "pipe" now pulls the pair of descriptors out of the return values (which is the true kernel system call interface) but displays them as an argument (which matches the interface exported by libc). - Each ABI handler adds entries to a linker set rather than requiring a statically defined table of handlers in main.c. - The arm and mips system call fetching code was changed to follow the same pattern as amd64 (and the in-kernel handler) of fetching register arguments first and then reading any remaining arguments from the stack. This should fix indirect system call arguments on at least arm. - The mipsn32 and n64 ABIs will now look for arguments in A4 through A7. - Use register %ebp for the 6th system call argument for Linux/i386 ABIs to match the in-kernel argument fetch code. - For powerpc binaries on a powerpc64 system, fetch the extra arguments on the stack as 32-bit values that are then copied into the 64-bit argument array instead of reading the 32-bit values directly into the 64-bit array. Reviewed by: kib (earlier version) Tested on: amd64 (FreeBSD/amd64 & i386), i386, arm (earlier version) Tested on: powerpc64 (FreeBSD/powerpc64 & powerpc) MFC after: 1 month Differential Revision: https://reviews.freebsd.org/D3575
2015-09-30 19:13:32 +00:00
len = retval[0];
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
else
len = args[sc->offset + 1];
Various style and whitespace fixes.
2015-08-19 20:02:03 +00:00
/*
* Don't print more than max_string characters, to avoid word
* wrap. If we have to truncate put some ... after the string.
*/
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
if (len > max_string) {
len = max_string;
truncated = 1;
}
- Style(9) cleanup. Approved by: kib (mentor)
2012-09-02 11:03:18 +00:00
if (len && get_struct(pid, (void*)args[sc->offset], &tmp2, len)
!= -1) {
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
tmp3 = malloc(len * 4 + 1);
while (len) {
- Style(9) cleanup. Approved by: kib (mentor)
2012-09-02 11:03:18 +00:00
if (strvisx(tmp3, tmp2, len,
VIS_CSTYLE|VIS_TAB|VIS_NL) <= max_string)
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
break;
len--;
truncated = 1;
Cleanup unnecessary semicolons from utilities we all love.
2016-04-15 22:31:22 +00:00
}
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
fprintf(fp, "\"%s\"%s", tmp3, truncated ?
- Style(9) cleanup. Approved by: kib (mentor)
2012-09-02 11:03:18 +00:00
"..." : "");
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
free(tmp3);
} else {
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
fprintf(fp, "0x%lx", args[sc->offset]);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
}
break;
}
Handle the conditional decoding of execve() argument and environment arrays generically rather than duplicating a hack in all of the backends. - Add two new system call argument types and use them instead of StringArray for the argument and environment arguments execve and linux_execve. - Honor the -a/-e flags in the handling of these new types. - Instead of printing "<missing argument>" when the decoding is disabled, print the raw pointer value.
2015-08-20 14:51:11 +00:00
case ExecArgs:
case ExecEnv:
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
case StringArray: {
Rework the argv and env string fetching for execve to be more robust. Before truss would fetch 100 string pointers and happily walk off the end of the array if it never found a NULL. This also means for a short argv list it could fail entirely if the 100 string pointers spanned into an unmapped page. Instead, fetch page-aligned blocks of string pointers in a loop fetching each string until a NULL is found. While here, make use of the open memstream file descriptor instead of allocating a temporary array. This allows us to fetch each string once instead of twice.
2015-08-20 14:33:30 +00:00
uintptr_t addr;
union {
char *strarray[0];
char buf[PAGE_SIZE];
} u;
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
char *string;
Rework the argv and env string fetching for execve to be more robust. Before truss would fetch 100 string pointers and happily walk off the end of the array if it never found a NULL. This also means for a short argv list it could fail entirely if the 100 string pointers spanned into an unmapped page. Instead, fetch page-aligned blocks of string pointers in a loop fetching each string until a NULL is found. While here, make use of the open memstream file descriptor instead of allocating a temporary array. This allows us to fetch each string once instead of twice.
2015-08-20 14:33:30 +00:00
size_t len;
Several changes to truss. - Refactor the interface between the ABI-independent code and the ABI-specific backends. The backends now provide smaller hooks to fetch system call arguments and return values. The rest of the system call entry and exit handling that was previously duplicated among all the backends has been moved to one place. - Merge the loop when waiting for an event with the loop for handling stops. This also means not emulating a procfs-like interface on top of ptrace(). Instead, use a single event loop that fetches process events via waitid(). Among other things this allows us to report the full 32-bit exit value. - Use PT_FOLLOW_FORK to follow new child processes instead of forking a new truss process for each new child. This allows one truss process to monitor a tree of processes and truss -c should now display one total for the entire tree instead of separate summaries per process. - Use the recently added fields to ptrace_lwpinfo to determine the current system call number and argument count. The latter is especially useful and fixes a regression since the conversion from procfs. truss now generally prints the correct number of arguments for most system calls rather than printing extra arguments for any call not listed in the table in syscalls.c. - Actually check the new ABI when processes call exec. The comments claimed that this happened but it was not being done (perhaps this was another regression in the conversion to ptrace()). If the new ABI after exec is not supported, truss detaches from the process. If truss does not support the ABI for a newly executed process the process is killed before it returns from exec. - Along with the refactor, teach the various ABI-specific backends to fetch both return values, not just the first. Use this to properly report the full 64-bit return value from lseek(). In addition, the handler for "pipe" now pulls the pair of descriptors out of the return values (which is the true kernel system call interface) but displays them as an argument (which matches the interface exported by libc). - Each ABI handler adds entries to a linker set rather than requiring a statically defined table of handlers in main.c. - The arm and mips system call fetching code was changed to follow the same pattern as amd64 (and the in-kernel handler) of fetching register arguments first and then reading any remaining arguments from the stack. This should fix indirect system call arguments on at least arm. - The mipsn32 and n64 ABIs will now look for arguments in A4 through A7. - Use register %ebp for the 6th system call argument for Linux/i386 ABIs to match the in-kernel argument fetch code. - For powerpc binaries on a powerpc64 system, fetch the extra arguments on the stack as 32-bit values that are then copied into the 64-bit argument array instead of reading the 32-bit values directly into the 64-bit array. Reviewed by: kib (earlier version) Tested on: amd64 (FreeBSD/amd64 & i386), i386, arm (earlier version) Tested on: powerpc64 (FreeBSD/powerpc64 & powerpc) MFC after: 1 month Differential Revision: https://reviews.freebsd.org/D3575
2015-09-30 19:13:32 +00:00
u_int first, i;
Rework the argv and env string fetching for execve to be more robust. Before truss would fetch 100 string pointers and happily walk off the end of the array if it never found a NULL. This also means for a short argv list it could fail entirely if the 100 string pointers spanned into an unmapped page. Instead, fetch page-aligned blocks of string pointers in a loop fetching each string until a NULL is found. While here, make use of the open memstream file descriptor instead of allocating a temporary array. This allows us to fetch each string once instead of twice.
2015-08-20 14:33:30 +00:00
Handle the conditional decoding of execve() argument and environment arrays generically rather than duplicating a hack in all of the backends. - Add two new system call argument types and use them instead of StringArray for the argument and environment arguments execve and linux_execve. - Honor the -a/-e flags in the handling of these new types. - Instead of printing "<missing argument>" when the decoding is disabled, print the raw pointer value.
2015-08-20 14:51:11 +00:00
/*
* Only parse argv[] and environment arrays from exec calls
* if requested.
*/
if (((sc->type & ARG_MASK) == ExecArgs &&
(trussinfo->flags & EXECVEARGS) == 0) ||
((sc->type & ARG_MASK) == ExecEnv &&
(trussinfo->flags & EXECVEENVS) == 0)) {
fprintf(fp, "0x%lx", args[sc->offset]);
break;
}
Trim trailing whitespace.
2015-09-30 00:08:24 +00:00
Rework the argv and env string fetching for execve to be more robust. Before truss would fetch 100 string pointers and happily walk off the end of the array if it never found a NULL. This also means for a short argv list it could fail entirely if the 100 string pointers spanned into an unmapped page. Instead, fetch page-aligned blocks of string pointers in a loop fetching each string until a NULL is found. While here, make use of the open memstream file descriptor instead of allocating a temporary array. This allows us to fetch each string once instead of twice.
2015-08-20 14:33:30 +00:00
/*
* Read a page of pointers at a time. Punt if the top-level
* pointer is not aligned. Note that the first read is of
* a partial page.
*/
addr = args[sc->offset];
if (addr % sizeof(char *) != 0) {
fprintf(fp, "0x%lx", args[sc->offset]);
break;
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
}
Rework the argv and env string fetching for execve to be more robust. Before truss would fetch 100 string pointers and happily walk off the end of the array if it never found a NULL. This also means for a short argv list it could fail entirely if the 100 string pointers spanned into an unmapped page. Instead, fetch page-aligned blocks of string pointers in a loop fetching each string until a NULL is found. While here, make use of the open memstream file descriptor instead of allocating a temporary array. This allows us to fetch each string once instead of twice.
2015-08-20 14:33:30 +00:00
len = PAGE_SIZE - (addr & PAGE_MASK);
if (get_struct(pid, (void *)addr, u.buf, len) == -1) {
fprintf(fp, "0x%lx", args[sc->offset]);
break;
}
fputc('[', fp);
first = 1;
i = 0;
while (u.strarray[i] != NULL) {
string = get_string(pid, u.strarray[i], 0);
fprintf(fp, "%s \"%s\"", first ? "" : ",", string);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
free(string);
Rework the argv and env string fetching for execve to be more robust. Before truss would fetch 100 string pointers and happily walk off the end of the array if it never found a NULL. This also means for a short argv list it could fail entirely if the 100 string pointers spanned into an unmapped page. Instead, fetch page-aligned blocks of string pointers in a loop fetching each string until a NULL is found. While here, make use of the open memstream file descriptor instead of allocating a temporary array. This allows us to fetch each string once instead of twice.
2015-08-20 14:33:30 +00:00
first = 0;
i++;
if (i == len / sizeof(char *)) {
addr += len;
len = PAGE_SIZE;
if (get_struct(pid, (void *)addr, u.buf, len) ==
-1) {
fprintf(fp, ", <inval>");
break;
}
i = 0;
}
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
}
Rework the argv and env string fetching for execve to be more robust. Before truss would fetch 100 string pointers and happily walk off the end of the array if it never found a NULL. This also means for a short argv list it could fail entirely if the 100 string pointers spanned into an unmapped page. Instead, fetch page-aligned blocks of string pointers in a loop fetching each string until a NULL is found. While here, make use of the open memstream file descriptor instead of allocating a temporary array. This allows us to fetch each string once instead of twice.
2015-08-20 14:33:30 +00:00
fputs(" ]", fp);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
break;
}
Fix Quad handling on 64-bit architectures. On 64-bit machines, a Quad also occupies a single slot. There's no need for any special handling of Quads. While here, remove the silly make_quad() function. We have the 2 longs on 32-bit machines already lined up in the argument array, so we can fetch the Quad with a simple cast. Before: lseek(1,0x123456789,0xd0d0d0d0d0d0d0d0) = 4886718345 (0x123456789) After: lseek(1,0x123456789,SEEK_SET) = 4886718345 (0x123456789)
2004-09-05 05:27:30 +00:00
#ifdef __LP64__
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
case Quad:
The id_t type used to pass IDs to wait6(2) and procctl(6) is a 64-bit integer. Fix the argument decoding to treat this as a quad instead of an int. This includes using QUAD_ALIGN and QUAD_SLOTS as necessary. To continue printing IDs in decimal, add a new QuadHex argument type that prints a 64-bit integer in hex, use QuadHex for the existing off_t arguments, repurpose Quad to print a 64-bit integer in decimal, and use Quad for id_t arguments. This fixes the decoding of wait6(2) and procctl(2) on 32-bit platforms.
2015-10-01 17:50:41 +00:00
fprintf(fp, "%ld", args[sc->offset]);
break;
case QuadHex:
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
fprintf(fp, "0x%lx", args[sc->offset]);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
break;
Fix Quad handling on 64-bit architectures. On 64-bit machines, a Quad also occupies a single slot. There's no need for any special handling of Quads. While here, remove the silly make_quad() function. We have the 2 longs on 32-bit machines already lined up in the argument array, so we can fetch the Quad with a simple cast. Before: lseek(1,0x123456789,0xd0d0d0d0d0d0d0d0) = 4886718345 (0x123456789) After: lseek(1,0x123456789,SEEK_SET) = 4886718345 (0x123456789)
2004-09-05 05:27:30 +00:00
#else
The id_t type used to pass IDs to wait6(2) and procctl(6) is a 64-bit integer. Fix the argument decoding to treat this as a quad instead of an int. This includes using QUAD_ALIGN and QUAD_SLOTS as necessary. To continue printing IDs in decimal, add a new QuadHex argument type that prints a 64-bit integer in hex, use QuadHex for the existing off_t arguments, repurpose Quad to print a 64-bit integer in decimal, and use Quad for id_t arguments. This fixes the decoding of wait6(2) and procctl(2) on 32-bit platforms.
2015-10-01 17:50:41 +00:00
case Quad:
case QuadHex: {
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
unsigned long long ll;
Various style and whitespace fixes.
2015-08-19 20:02:03 +00:00
Several changes to truss. - Refactor the interface between the ABI-independent code and the ABI-specific backends. The backends now provide smaller hooks to fetch system call arguments and return values. The rest of the system call entry and exit handling that was previously duplicated among all the backends has been moved to one place. - Merge the loop when waiting for an event with the loop for handling stops. This also means not emulating a procfs-like interface on top of ptrace(). Instead, use a single event loop that fetches process events via waitid(). Among other things this allows us to report the full 32-bit exit value. - Use PT_FOLLOW_FORK to follow new child processes instead of forking a new truss process for each new child. This allows one truss process to monitor a tree of processes and truss -c should now display one total for the entire tree instead of separate summaries per process. - Use the recently added fields to ptrace_lwpinfo to determine the current system call number and argument count. The latter is especially useful and fixes a regression since the conversion from procfs. truss now generally prints the correct number of arguments for most system calls rather than printing extra arguments for any call not listed in the table in syscalls.c. - Actually check the new ABI when processes call exec. The comments claimed that this happened but it was not being done (perhaps this was another regression in the conversion to ptrace()). If the new ABI after exec is not supported, truss detaches from the process. If truss does not support the ABI for a newly executed process the process is killed before it returns from exec. - Along with the refactor, teach the various ABI-specific backends to fetch both return values, not just the first. Use this to properly report the full 64-bit return value from lseek(). In addition, the handler for "pipe" now pulls the pair of descriptors out of the return values (which is the true kernel system call interface) but displays them as an argument (which matches the interface exported by libc). - Each ABI handler adds entries to a linker set rather than requiring a statically defined table of handlers in main.c. - The arm and mips system call fetching code was changed to follow the same pattern as amd64 (and the in-kernel handler) of fetching register arguments first and then reading any remaining arguments from the stack. This should fix indirect system call arguments on at least arm. - The mipsn32 and n64 ABIs will now look for arguments in A4 through A7. - Use register %ebp for the 6th system call argument for Linux/i386 ABIs to match the in-kernel argument fetch code. - For powerpc binaries on a powerpc64 system, fetch the extra arguments on the stack as 32-bit values that are then copied into the 64-bit argument array instead of reading the 32-bit values directly into the 64-bit array. Reviewed by: kib (earlier version) Tested on: amd64 (FreeBSD/amd64 & i386), i386, arm (earlier version) Tested on: powerpc64 (FreeBSD/powerpc64 & powerpc) MFC after: 1 month Differential Revision: https://reviews.freebsd.org/D3575
2015-09-30 19:13:32 +00:00
#if _BYTE_ORDER == _LITTLE_ENDIAN
ll = (unsigned long long)args[sc->offset + 1] << 32 |
args[sc->offset];
#else
ll = (unsigned long long)args[sc->offset] << 32 |
args[sc->offset + 1];
#endif
The id_t type used to pass IDs to wait6(2) and procctl(6) is a 64-bit integer. Fix the argument decoding to treat this as a quad instead of an int. This includes using QUAD_ALIGN and QUAD_SLOTS as necessary. To continue printing IDs in decimal, add a new QuadHex argument type that prints a 64-bit integer in hex, use QuadHex for the existing off_t arguments, repurpose Quad to print a 64-bit integer in decimal, and use Quad for id_t arguments. This fixes the decoding of wait6(2) and procctl(2) on 32-bit platforms.
2015-10-01 17:50:41 +00:00
if ((sc->type & ARG_MASK) == Quad)
fprintf(fp, "%lld", ll);
else
fprintf(fp, "0x%llx", ll);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
break;
}
Fix Quad handling on 64-bit architectures. On 64-bit machines, a Quad also occupies a single slot. There's no need for any special handling of Quads. While here, remove the silly make_quad() function. We have the 2 longs on 32-bit machines already lined up in the argument array, so we can fetch the Quad with a simple cast. Before: lseek(1,0x123456789,0xd0d0d0d0d0d0d0d0) = 4886718345 (0x123456789) After: lseek(1,0x123456789,SEEK_SET) = 4886718345 (0x123456789)
2004-09-05 05:27:30 +00:00
#endif
Decode arguments to dup, dup2, getdirentries, pread, and pwrite. - dup and dup2 print fd arguments in decimal. - pread and pwrite are similar to read and write with the addition of the file offset. - getdirentries displays the output entries as a string for now and also prints the value returned in *basep. Eventually the buffer for getdirentries should perhaps be decoded as an array of dirent structures. PR: 214885 Submitted by: Jonathan de Boyne Pollard <J.deBoynePollard-newsgroups@NTLWorld.COM>
2017-06-05 05:25:50 +00:00
case PQuadHex: {
uint64_t val;
if (get_struct(pid, (void *)args[sc->offset], &val,
sizeof(val)) == 0)
fprintf(fp, "{ 0x%jx }", (uintmax_t)val);
else
fprintf(fp, "0x%lx", args[sc->offset]);
break;
}
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
case Ptr:
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
fprintf(fp, "0x%lx", args[sc->offset]);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
break;
case Readlinkres: {
char *tmp2;
Various style and whitespace fixes.
2015-08-19 20:02:03 +00:00
Several changes to truss. - Refactor the interface between the ABI-independent code and the ABI-specific backends. The backends now provide smaller hooks to fetch system call arguments and return values. The rest of the system call entry and exit handling that was previously duplicated among all the backends has been moved to one place. - Merge the loop when waiting for an event with the loop for handling stops. This also means not emulating a procfs-like interface on top of ptrace(). Instead, use a single event loop that fetches process events via waitid(). Among other things this allows us to report the full 32-bit exit value. - Use PT_FOLLOW_FORK to follow new child processes instead of forking a new truss process for each new child. This allows one truss process to monitor a tree of processes and truss -c should now display one total for the entire tree instead of separate summaries per process. - Use the recently added fields to ptrace_lwpinfo to determine the current system call number and argument count. The latter is especially useful and fixes a regression since the conversion from procfs. truss now generally prints the correct number of arguments for most system calls rather than printing extra arguments for any call not listed in the table in syscalls.c. - Actually check the new ABI when processes call exec. The comments claimed that this happened but it was not being done (perhaps this was another regression in the conversion to ptrace()). If the new ABI after exec is not supported, truss detaches from the process. If truss does not support the ABI for a newly executed process the process is killed before it returns from exec. - Along with the refactor, teach the various ABI-specific backends to fetch both return values, not just the first. Use this to properly report the full 64-bit return value from lseek(). In addition, the handler for "pipe" now pulls the pair of descriptors out of the return values (which is the true kernel system call interface) but displays them as an argument (which matches the interface exported by libc). - Each ABI handler adds entries to a linker set rather than requiring a statically defined table of handlers in main.c. - The arm and mips system call fetching code was changed to follow the same pattern as amd64 (and the in-kernel handler) of fetching register arguments first and then reading any remaining arguments from the stack. This should fix indirect system call arguments on at least arm. - The mipsn32 and n64 ABIs will now look for arguments in A4 through A7. - Use register %ebp for the 6th system call argument for Linux/i386 ABIs to match the in-kernel argument fetch code. - For powerpc binaries on a powerpc64 system, fetch the extra arguments on the stack as 32-bit values that are then copied into the 64-bit argument array instead of reading the 32-bit values directly into the 64-bit array. Reviewed by: kib (earlier version) Tested on: amd64 (FreeBSD/amd64 & i386), i386, arm (earlier version) Tested on: powerpc64 (FreeBSD/powerpc64 & powerpc) MFC after: 1 month Differential Revision: https://reviews.freebsd.org/D3575
2015-09-30 19:13:32 +00:00
if (retval[0] == -1)
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
break;
Several changes to truss. - Refactor the interface between the ABI-independent code and the ABI-specific backends. The backends now provide smaller hooks to fetch system call arguments and return values. The rest of the system call entry and exit handling that was previously duplicated among all the backends has been moved to one place. - Merge the loop when waiting for an event with the loop for handling stops. This also means not emulating a procfs-like interface on top of ptrace(). Instead, use a single event loop that fetches process events via waitid(). Among other things this allows us to report the full 32-bit exit value. - Use PT_FOLLOW_FORK to follow new child processes instead of forking a new truss process for each new child. This allows one truss process to monitor a tree of processes and truss -c should now display one total for the entire tree instead of separate summaries per process. - Use the recently added fields to ptrace_lwpinfo to determine the current system call number and argument count. The latter is especially useful and fixes a regression since the conversion from procfs. truss now generally prints the correct number of arguments for most system calls rather than printing extra arguments for any call not listed in the table in syscalls.c. - Actually check the new ABI when processes call exec. The comments claimed that this happened but it was not being done (perhaps this was another regression in the conversion to ptrace()). If the new ABI after exec is not supported, truss detaches from the process. If truss does not support the ABI for a newly executed process the process is killed before it returns from exec. - Along with the refactor, teach the various ABI-specific backends to fetch both return values, not just the first. Use this to properly report the full 64-bit return value from lseek(). In addition, the handler for "pipe" now pulls the pair of descriptors out of the return values (which is the true kernel system call interface) but displays them as an argument (which matches the interface exported by libc). - Each ABI handler adds entries to a linker set rather than requiring a statically defined table of handlers in main.c. - The arm and mips system call fetching code was changed to follow the same pattern as amd64 (and the in-kernel handler) of fetching register arguments first and then reading any remaining arguments from the stack. This should fix indirect system call arguments on at least arm. - The mipsn32 and n64 ABIs will now look for arguments in A4 through A7. - Use register %ebp for the 6th system call argument for Linux/i386 ABIs to match the in-kernel argument fetch code. - For powerpc binaries on a powerpc64 system, fetch the extra arguments on the stack as 32-bit values that are then copied into the 64-bit argument array instead of reading the 32-bit values directly into the 64-bit array. Reviewed by: kib (earlier version) Tested on: amd64 (FreeBSD/amd64 & i386), i386, arm (earlier version) Tested on: powerpc64 (FreeBSD/powerpc64 & powerpc) MFC after: 1 month Differential Revision: https://reviews.freebsd.org/D3575
2015-09-30 19:13:32 +00:00
tmp2 = get_string(pid, (void*)args[sc->offset], retval[0]);
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
fprintf(fp, "\"%s\"", tmp2);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
free(tmp2);
break;
}
case Ioctl: {
Various style and whitespace fixes.
2015-08-19 20:02:03 +00:00
const char *temp;
unsigned long cmd;
cmd = args[sc->offset];
Move the mkioctls script to libsysdecode and use it to generate a sysdecode_ioctlname() function. This function matches the behavior of the truss variant in that it returns a pointer to a string description for known ioctls. The caller is responsible for displaying unknown ioctl requests. For kdump this meant moving the logic to handle unknown ioctl requests out of the generated function and into an ioctlname() function in kdump.c instead. Differential Revision: https://reviews.freebsd.org/D4610
2015-12-22 20:33:49 +00:00
temp = sysdecode_ioctlname(cmd);
- Style(9) cleanup. Approved by: kib (mentor)
2012-09-02 11:03:18 +00:00
if (temp)
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
fputs(temp, fp);
- Style(9) cleanup. Approved by: kib (mentor)
2012-09-02 11:03:18 +00:00
else {
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
fprintf(fp, "0x%lx { IO%s%s 0x%lx('%c'), %lu, %lu }",
Various style and whitespace fixes.
2015-08-19 20:02:03 +00:00
cmd, cmd & IOC_OUT ? "R" : "",
cmd & IOC_IN ? "W" : "", IOCGROUP(cmd),
isprint(IOCGROUP(cmd)) ? (char)IOCGROUP(cmd) : '?',
cmd & 0xFF, IOCPARM_LEN(cmd));
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
}
break;
Add support for decoding Timespec, Timeval, Itimerval, Pollfd, Fd_set and Sigaction structures. Use these for printing the arguments to sigaction(), nanosleep(), select(), poll(), gettimeofday(), clock_gettime(), recvfrom(), getitimer() and setitimer(). This is based on Dan's patch from the PR but I've hacked it for style and some other issues. While Dan has checked this patch, any goofs are probably my fault. (The PR also contains support for the dual return values of pipe(). These will follow once I've ported that support to platforms other than i386.) PR: 52190 Submitted by: Dan Nelson <dnelson@allantgroup.com>
2004-03-23 12:37:02 +00:00
}
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
case Timespec: {
struct timespec ts;
Various style and whitespace fixes.
2015-08-19 20:02:03 +00:00
- Style(9) cleanup. Approved by: kib (mentor)
2012-09-02 11:03:18 +00:00
if (get_struct(pid, (void *)args[sc->offset], &ts,
sizeof(ts)) != -1)
Always use %j with an intmax_t cast to print time_t values. time_t is longer than long on 32-bit platforms with a 64-bit time_t. Inspired by: mail from bde
2015-08-19 20:09:14 +00:00
fprintf(fp, "{ %jd.%09ld }", (intmax_t)ts.tv_sec,
- Style(9) cleanup. Approved by: kib (mentor)
2012-09-02 11:03:18 +00:00
ts.tv_nsec);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
else
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
fprintf(fp, "0x%lx", args[sc->offset]);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
break;
}
Decode the arguments passed to the *at() family of system calls. This is especially useful now that libc's open() always calls openat(). While here, fix a few other things: - Decode the mode argument passed to access(), eaccess(), and faccessat(). - Decode the atfd paramete to pretty-print AT_FDCWD. - Decode the special AT_* flags used with some of the *at() system calls. - Decode arguments for fchmod(), lchmod(), fchown(), lchown(), eaccess(), and futimens(). - Decode both of the timeval structures passed to futimes() instead of just the first one.
2015-08-06 19:08:33 +00:00
case Timespec2: {
struct timespec ts[2];
const char *sep;
unsigned int i;
if (get_struct(pid, (void *)args[sc->offset], &ts, sizeof(ts))
!= -1) {
Consistently use both leading and trailing spaces inside of the {}'s when pretty-printing structures. Most structures used both spaces, but some only used a trailing space and some used neither.
2015-08-06 20:05:40 +00:00
fputs("{ ", fp);
Decode the arguments passed to the *at() family of system calls. This is especially useful now that libc's open() always calls openat(). While here, fix a few other things: - Decode the mode argument passed to access(), eaccess(), and faccessat(). - Decode the atfd paramete to pretty-print AT_FDCWD. - Decode the special AT_* flags used with some of the *at() system calls. - Decode arguments for fchmod(), lchmod(), fchown(), lchown(), eaccess(), and futimens(). - Decode both of the timeval structures passed to futimes() instead of just the first one.
2015-08-06 19:08:33 +00:00
sep = "";
for (i = 0; i < nitems(ts); i++) {
fputs(sep, fp);
sep = ", ";
switch (ts[i].tv_nsec) {
case UTIME_NOW:
fprintf(fp, "UTIME_NOW");
break;
case UTIME_OMIT:
fprintf(fp, "UTIME_OMIT");
break;
default:
Always use %j with an intmax_t cast to print time_t values. time_t is longer than long on 32-bit platforms with a 64-bit time_t. Inspired by: mail from bde
2015-08-19 20:09:14 +00:00
fprintf(fp, "%jd.%09ld",
(intmax_t)ts[i].tv_sec,
ts[i].tv_nsec);
Decode the arguments passed to the *at() family of system calls. This is especially useful now that libc's open() always calls openat(). While here, fix a few other things: - Decode the mode argument passed to access(), eaccess(), and faccessat(). - Decode the atfd paramete to pretty-print AT_FDCWD. - Decode the special AT_* flags used with some of the *at() system calls. - Decode arguments for fchmod(), lchmod(), fchown(), lchown(), eaccess(), and futimens(). - Decode both of the timeval structures passed to futimes() instead of just the first one.
2015-08-06 19:08:33 +00:00
break;
}
}
Consistently use both leading and trailing spaces inside of the {}'s when pretty-printing structures. Most structures used both spaces, but some only used a trailing space and some used neither.
2015-08-06 20:05:40 +00:00
fputs(" }", fp);
Decode the arguments passed to the *at() family of system calls. This is especially useful now that libc's open() always calls openat(). While here, fix a few other things: - Decode the mode argument passed to access(), eaccess(), and faccessat(). - Decode the atfd paramete to pretty-print AT_FDCWD. - Decode the special AT_* flags used with some of the *at() system calls. - Decode arguments for fchmod(), lchmod(), fchown(), lchown(), eaccess(), and futimens(). - Decode both of the timeval structures passed to futimes() instead of just the first one.
2015-08-06 19:08:33 +00:00
} else
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
fprintf(fp, "0x%lx", args[sc->offset]);
Decode the arguments passed to the *at() family of system calls. This is especially useful now that libc's open() always calls openat(). While here, fix a few other things: - Decode the mode argument passed to access(), eaccess(), and faccessat(). - Decode the atfd paramete to pretty-print AT_FDCWD. - Decode the special AT_* flags used with some of the *at() system calls. - Decode arguments for fchmod(), lchmod(), fchown(), lchown(), eaccess(), and futimens(). - Decode both of the timeval structures passed to futimes() instead of just the first one.
2015-08-06 19:08:33 +00:00
break;
}
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
case Timeval: {
struct timeval tv;
Various style and whitespace fixes.
2015-08-19 20:02:03 +00:00
- Style(9) cleanup. Approved by: kib (mentor)
2012-09-02 11:03:18 +00:00
if (get_struct(pid, (void *)args[sc->offset], &tv, sizeof(tv))
!= -1)
Always use %j with an intmax_t cast to print time_t values. time_t is longer than long on 32-bit platforms with a 64-bit time_t. Inspired by: mail from bde
2015-08-19 20:09:14 +00:00
fprintf(fp, "{ %jd.%06ld }", (intmax_t)tv.tv_sec,
- Style(9) cleanup. Approved by: kib (mentor)
2012-09-02 11:03:18 +00:00
tv.tv_usec);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
else
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
fprintf(fp, "0x%lx", args[sc->offset]);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
break;
}
case Timeval2: {
struct timeval tv[2];
Various style and whitespace fixes.
2015-08-19 20:02:03 +00:00
- Style(9) cleanup. Approved by: kib (mentor)
2012-09-02 11:03:18 +00:00
if (get_struct(pid, (void *)args[sc->offset], &tv, sizeof(tv))
!= -1)
Always use %j with an intmax_t cast to print time_t values. time_t is longer than long on 32-bit platforms with a 64-bit time_t. Inspired by: mail from bde
2015-08-19 20:09:14 +00:00
fprintf(fp, "{ %jd.%06ld, %jd.%06ld }",
(intmax_t)tv[0].tv_sec, tv[0].tv_usec,
(intmax_t)tv[1].tv_sec, tv[1].tv_usec);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
else
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
fprintf(fp, "0x%lx", args[sc->offset]);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
break;
}
case Itimerval: {
struct itimerval itv;
Various style and whitespace fixes.
2015-08-19 20:02:03 +00:00
- Style(9) cleanup. Approved by: kib (mentor)
2012-09-02 11:03:18 +00:00
if (get_struct(pid, (void *)args[sc->offset], &itv,
sizeof(itv)) != -1)
Always use %j with an intmax_t cast to print time_t values. time_t is longer than long on 32-bit platforms with a 64-bit time_t. Inspired by: mail from bde
2015-08-19 20:09:14 +00:00
fprintf(fp, "{ %jd.%06ld, %jd.%06ld }",
(intmax_t)itv.it_interval.tv_sec,
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
itv.it_interval.tv_usec,
Always use %j with an intmax_t cast to print time_t values. time_t is longer than long on 32-bit platforms with a 64-bit time_t. Inspired by: mail from bde
2015-08-19 20:09:14 +00:00
(intmax_t)itv.it_value.tv_sec,
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
itv.it_value.tv_usec);
else
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
fprintf(fp, "0x%lx", args[sc->offset]);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
break;
}
Add Linux socket call decoding to truss MFC after: 1 month
2014-04-24 14:12:35 +00:00
case LinuxSockArgs:
{
struct linux_socketcall_args largs;
Various style and whitespace fixes.
2015-08-19 20:02:03 +00:00
Add Linux socket call decoding to truss MFC after: 1 month
2014-04-24 14:12:35 +00:00
if (get_struct(pid, (void *)args[sc->offset], (void *)&largs,
Tidy the linux_socketcall decoding: - Don't exit if get_struct() fails, instead print the raw pointer value to match all other argument decoding cases. - Use an xlat table instead of a home-rolled switch for the operation name. - Display the nested socketcall args structure as a structure instead of as two inline arguments.
2015-08-17 18:47:39 +00:00
sizeof(largs)) != -1)
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
fprintf(fp, "{ %s, 0x%lx }",
Tidy the linux_socketcall decoding: - Don't exit if get_struct() fails, instead print the raw pointer value to match all other argument decoding cases. - Use an xlat table instead of a home-rolled switch for the operation name. - Display the nested socketcall args structure as a structure instead of as two inline arguments.
2015-08-17 18:47:39 +00:00
lookup(linux_socketcall_ops, largs.what, 10),
(long unsigned int)largs.args);
else
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
fprintf(fp, "0x%lx", args[sc->offset]);
Add Linux socket call decoding to truss MFC after: 1 month
2014-04-24 14:12:35 +00:00
break;
}
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
case Pollfd: {
/*
- Style(9) cleanup. Approved by: kib (mentor)
2012-09-02 11:03:18 +00:00
* XXX: A Pollfd argument expects the /next/ syscall argument
* to be the number of fds in the array. This matches the poll
* syscall.
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
*/
struct pollfd *pfd;
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
int numfds = args[sc->offset + 1];
size_t bytes = sizeof(struct pollfd) * numfds;
int i;
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
if ((pfd = malloc(bytes)) == NULL)
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
err(1, "Cannot malloc %zu bytes for pollfd array",
- Style(9) cleanup. Approved by: kib (mentor)
2012-09-02 11:03:18 +00:00
bytes);
if (get_struct(pid, (void *)args[sc->offset], pfd, bytes)
!= -1) {
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
fputs("{", fp);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
for (i = 0; i < numfds; i++) {
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
fprintf(fp, " %d/%s", pfd[i].fd,
- Style(9) cleanup. Approved by: kib (mentor)
2012-09-02 11:03:18 +00:00
xlookup_bits(poll_flags, pfd[i].events));
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
}
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
fputs(" }", fp);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
} else {
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
fprintf(fp, "0x%lx", args[sc->offset]);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
}
free(pfd);
break;
}
case Fd_set: {
/*
- Style(9) cleanup. Approved by: kib (mentor)
2012-09-02 11:03:18 +00:00
* XXX: A Fd_set argument expects the /first/ syscall argument
* to be the number of fds in the array. This matches the
* select syscall.
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
*/
fd_set *fds;
int numfds = args[0];
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
size_t bytes = _howmany(numfds, _NFDBITS) * _NFDBITS;
int i;
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
if ((fds = malloc(bytes)) == NULL)
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
err(1, "Cannot malloc %zu bytes for fd_set array",
- Style(9) cleanup. Approved by: kib (mentor)
2012-09-02 11:03:18 +00:00
bytes);
if (get_struct(pid, (void *)args[sc->offset], fds, bytes)
!= -1) {
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
fputs("{", fp);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
for (i = 0; i < numfds; i++) {
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
if (FD_ISSET(i, fds))
fprintf(fp, " %d", i);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
}
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
fputs(" }", fp);
- Style(9) cleanup. Approved by: kib (mentor)
2012-09-02 11:03:18 +00:00
} else
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
fprintf(fp, "0x%lx", args[sc->offset]);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
free(fds);
break;
}
- Decode the idtype argument passed to wait6() in kdump and truss. - Don't treat an options argument of 0 to wait4() as an error in kdump. - Decode the wait options passed to wait4() and wait6() in truss and decode the returned rusage and exit status. Approved by: re (kib) MFC after: 1 week
2013-09-12 18:08:25 +00:00
case Signal:
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
fputs(strsig2(args[sc->offset]), fp);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
break;
case Sigset: {
long sig;
sigset_t ss;
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
int i, first;
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
sig = args[sc->offset];
- Style(9) cleanup. Approved by: kib (mentor)
2012-09-02 11:03:18 +00:00
if (get_struct(pid, (void *)args[sc->offset], (void *)&ss,
sizeof(ss)) == -1) {
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
fprintf(fp, "0x%lx", args[sc->offset]);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
break;
}
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
fputs("{ ", fp);
first = 1;
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
for (i = 1; i < sys_nsig; i++) {
- Decode the idtype argument passed to wait6() in kdump and truss. - Don't treat an options argument of 0 to wait4() as an error in kdump. - Decode the wait options passed to wait4() and wait6() in truss and decode the returned rusage and exit status. Approved by: re (kib) MFC after: 1 week
2013-09-12 18:08:25 +00:00
if (sigismember(&ss, i)) {
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
fprintf(fp, "%s%s", !first ? "|" : "",
Move mksubr from kdump into libsysdecode. Restructure this script so that it generates a header of tables instead of a source file. The tables are included in a flags.c source file which provides functions to decode various system call arguments. For functions that decode an enumeration, the function returns a pointer to a string for known values and NULL for unknown values. For functions that do more complex decoding (typically of a bitmask), the function accepts a pointer to a FILE object (open_memstream() can be used as a string builder) to which decoded values are written. If the function operates on a bitmask, the function returns true if any bits were decoded or false if the entire value was valid. Additionally, the third argument accepts a pointer to a value to which any undecoded bits are stored. This pointer can be NULL if the caller doesn't care about remaining bits. Convert kdump over to using decoder functions from libsysdecode instead of mksubr. truss also uses decoders from libsysdecode instead of private lookup tables, though lookup tables for objects not decoded by kdump remain in truss for now. Eventually most of these tables should move into libsysdecode as the automated table generation approach from mksubr is less stale than the static tables in truss. Some changes have been made to truss and kdump output: - The flags passed to open() are now properly decoded in that one of O_RDONLY, O_RDWR, O_WRONLY, or O_EXEC is always included in a decoded mask. - Optional arguments to open(), openat(), and fcntl() are only printed in kdump if they exist (e.g. the mode is only printed for open() if O_CREAT is set in the flags). - Print argument to F_GETLK/SETLK/SETLKW in kdump as a pointer, not int. - Include all procctl() commands. - Correctly decode pipe2() flags in truss by not assuming full open()-like flags with O_RDONLY, etc. - Decode file flags passed to *chflags() as file flags (UF_* and SF_*) rather than as a file mode. - Fix decoding of quotactl() commands by splitting out the two command components instead of assuming the raw command value matches the primary command component. In addition, truss and kdump now build without triggering any warnings. All of the sysdecode manpages now include the required headers in the synopsis. Reviewed by: kib (several older versions), wblock (manpages) MFC after: 2 months Differential Revision: https://reviews.freebsd.org/D7847
2016-10-17 22:37:07 +00:00
strsig2(i));
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
first = 0;
- Decode the idtype argument passed to wait6() in kdump and truss. - Don't treat an options argument of 0 to wait4() as an error in kdump. - Decode the wait options passed to wait4() and wait6() in truss and decode the returned rusage and exit status. Approved by: re (kib) MFC after: 1 week
2013-09-12 18:08:25 +00:00
}
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
}
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
if (!first)
fputc(' ', fp);
fputc('}', fp);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
break;
}
Move mksubr from kdump into libsysdecode. Restructure this script so that it generates a header of tables instead of a source file. The tables are included in a flags.c source file which provides functions to decode various system call arguments. For functions that decode an enumeration, the function returns a pointer to a string for known values and NULL for unknown values. For functions that do more complex decoding (typically of a bitmask), the function accepts a pointer to a FILE object (open_memstream() can be used as a string builder) to which decoded values are written. If the function operates on a bitmask, the function returns true if any bits were decoded or false if the entire value was valid. Additionally, the third argument accepts a pointer to a value to which any undecoded bits are stored. This pointer can be NULL if the caller doesn't care about remaining bits. Convert kdump over to using decoder functions from libsysdecode instead of mksubr. truss also uses decoders from libsysdecode instead of private lookup tables, though lookup tables for objects not decoded by kdump remain in truss for now. Eventually most of these tables should move into libsysdecode as the automated table generation approach from mksubr is less stale than the static tables in truss. Some changes have been made to truss and kdump output: - The flags passed to open() are now properly decoded in that one of O_RDONLY, O_RDWR, O_WRONLY, or O_EXEC is always included in a decoded mask. - Optional arguments to open(), openat(), and fcntl() are only printed in kdump if they exist (e.g. the mode is only printed for open() if O_CREAT is set in the flags). - Print argument to F_GETLK/SETLK/SETLKW in kdump as a pointer, not int. - Include all procctl() commands. - Correctly decode pipe2() flags in truss by not assuming full open()-like flags with O_RDONLY, etc. - Decode file flags passed to *chflags() as file flags (UF_* and SF_*) rather than as a file mode. - Fix decoding of quotactl() commands by splitting out the two command components instead of assuming the raw command value matches the primary command component. In addition, truss and kdump now build without triggering any warnings. All of the sysdecode manpages now include the required headers in the synopsis. Reviewed by: kib (several older versions), wblock (manpages) MFC after: 2 months Differential Revision: https://reviews.freebsd.org/D7847
2016-10-17 22:37:07 +00:00
case Sigprocmask:
print_integer_arg(sysdecode_sigprocmask_how, fp,
args[sc->offset]);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
break;
Move mksubr from kdump into libsysdecode. Restructure this script so that it generates a header of tables instead of a source file. The tables are included in a flags.c source file which provides functions to decode various system call arguments. For functions that decode an enumeration, the function returns a pointer to a string for known values and NULL for unknown values. For functions that do more complex decoding (typically of a bitmask), the function accepts a pointer to a FILE object (open_memstream() can be used as a string builder) to which decoded values are written. If the function operates on a bitmask, the function returns true if any bits were decoded or false if the entire value was valid. Additionally, the third argument accepts a pointer to a value to which any undecoded bits are stored. This pointer can be NULL if the caller doesn't care about remaining bits. Convert kdump over to using decoder functions from libsysdecode instead of mksubr. truss also uses decoders from libsysdecode instead of private lookup tables, though lookup tables for objects not decoded by kdump remain in truss for now. Eventually most of these tables should move into libsysdecode as the automated table generation approach from mksubr is less stale than the static tables in truss. Some changes have been made to truss and kdump output: - The flags passed to open() are now properly decoded in that one of O_RDONLY, O_RDWR, O_WRONLY, or O_EXEC is always included in a decoded mask. - Optional arguments to open(), openat(), and fcntl() are only printed in kdump if they exist (e.g. the mode is only printed for open() if O_CREAT is set in the flags). - Print argument to F_GETLK/SETLK/SETLKW in kdump as a pointer, not int. - Include all procctl() commands. - Correctly decode pipe2() flags in truss by not assuming full open()-like flags with O_RDONLY, etc. - Decode file flags passed to *chflags() as file flags (UF_* and SF_*) rather than as a file mode. - Fix decoding of quotactl() commands by splitting out the two command components instead of assuming the raw command value matches the primary command component. In addition, truss and kdump now build without triggering any warnings. All of the sysdecode manpages now include the required headers in the synopsis. Reviewed by: kib (several older versions), wblock (manpages) MFC after: 2 months Differential Revision: https://reviews.freebsd.org/D7847
2016-10-17 22:37:07 +00:00
case Fcntlflag:
Various style and whitespace fixes.
2015-08-19 20:02:03 +00:00
/* XXX: Output depends on the value of the previous argument. */
Move mksubr from kdump into libsysdecode. Restructure this script so that it generates a header of tables instead of a source file. The tables are included in a flags.c source file which provides functions to decode various system call arguments. For functions that decode an enumeration, the function returns a pointer to a string for known values and NULL for unknown values. For functions that do more complex decoding (typically of a bitmask), the function accepts a pointer to a FILE object (open_memstream() can be used as a string builder) to which decoded values are written. If the function operates on a bitmask, the function returns true if any bits were decoded or false if the entire value was valid. Additionally, the third argument accepts a pointer to a value to which any undecoded bits are stored. This pointer can be NULL if the caller doesn't care about remaining bits. Convert kdump over to using decoder functions from libsysdecode instead of mksubr. truss also uses decoders from libsysdecode instead of private lookup tables, though lookup tables for objects not decoded by kdump remain in truss for now. Eventually most of these tables should move into libsysdecode as the automated table generation approach from mksubr is less stale than the static tables in truss. Some changes have been made to truss and kdump output: - The flags passed to open() are now properly decoded in that one of O_RDONLY, O_RDWR, O_WRONLY, or O_EXEC is always included in a decoded mask. - Optional arguments to open(), openat(), and fcntl() are only printed in kdump if they exist (e.g. the mode is only printed for open() if O_CREAT is set in the flags). - Print argument to F_GETLK/SETLK/SETLKW in kdump as a pointer, not int. - Include all procctl() commands. - Correctly decode pipe2() flags in truss by not assuming full open()-like flags with O_RDONLY, etc. - Decode file flags passed to *chflags() as file flags (UF_* and SF_*) rather than as a file mode. - Fix decoding of quotactl() commands by splitting out the two command components instead of assuming the raw command value matches the primary command component. In addition, truss and kdump now build without triggering any warnings. All of the sysdecode manpages now include the required headers in the synopsis. Reviewed by: kib (several older versions), wblock (manpages) MFC after: 2 months Differential Revision: https://reviews.freebsd.org/D7847
2016-10-17 22:37:07 +00:00
if (sysdecode_fcntl_arg_p(args[sc->offset - 1]))
sysdecode_fcntl_arg(fp, args[sc->offset - 1],
args[sc->offset], 16);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
break;
case Open:
Move mksubr from kdump into libsysdecode. Restructure this script so that it generates a header of tables instead of a source file. The tables are included in a flags.c source file which provides functions to decode various system call arguments. For functions that decode an enumeration, the function returns a pointer to a string for known values and NULL for unknown values. For functions that do more complex decoding (typically of a bitmask), the function accepts a pointer to a FILE object (open_memstream() can be used as a string builder) to which decoded values are written. If the function operates on a bitmask, the function returns true if any bits were decoded or false if the entire value was valid. Additionally, the third argument accepts a pointer to a value to which any undecoded bits are stored. This pointer can be NULL if the caller doesn't care about remaining bits. Convert kdump over to using decoder functions from libsysdecode instead of mksubr. truss also uses decoders from libsysdecode instead of private lookup tables, though lookup tables for objects not decoded by kdump remain in truss for now. Eventually most of these tables should move into libsysdecode as the automated table generation approach from mksubr is less stale than the static tables in truss. Some changes have been made to truss and kdump output: - The flags passed to open() are now properly decoded in that one of O_RDONLY, O_RDWR, O_WRONLY, or O_EXEC is always included in a decoded mask. - Optional arguments to open(), openat(), and fcntl() are only printed in kdump if they exist (e.g. the mode is only printed for open() if O_CREAT is set in the flags). - Print argument to F_GETLK/SETLK/SETLKW in kdump as a pointer, not int. - Include all procctl() commands. - Correctly decode pipe2() flags in truss by not assuming full open()-like flags with O_RDONLY, etc. - Decode file flags passed to *chflags() as file flags (UF_* and SF_*) rather than as a file mode. - Fix decoding of quotactl() commands by splitting out the two command components instead of assuming the raw command value matches the primary command component. In addition, truss and kdump now build without triggering any warnings. All of the sysdecode manpages now include the required headers in the synopsis. Reviewed by: kib (several older versions), wblock (manpages) MFC after: 2 months Differential Revision: https://reviews.freebsd.org/D7847
2016-10-17 22:37:07 +00:00
print_mask_arg(sysdecode_open_flags, fp, args[sc->offset]);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
break;
case Fcntl:
Move mksubr from kdump into libsysdecode. Restructure this script so that it generates a header of tables instead of a source file. The tables are included in a flags.c source file which provides functions to decode various system call arguments. For functions that decode an enumeration, the function returns a pointer to a string for known values and NULL for unknown values. For functions that do more complex decoding (typically of a bitmask), the function accepts a pointer to a FILE object (open_memstream() can be used as a string builder) to which decoded values are written. If the function operates on a bitmask, the function returns true if any bits were decoded or false if the entire value was valid. Additionally, the third argument accepts a pointer to a value to which any undecoded bits are stored. This pointer can be NULL if the caller doesn't care about remaining bits. Convert kdump over to using decoder functions from libsysdecode instead of mksubr. truss also uses decoders from libsysdecode instead of private lookup tables, though lookup tables for objects not decoded by kdump remain in truss for now. Eventually most of these tables should move into libsysdecode as the automated table generation approach from mksubr is less stale than the static tables in truss. Some changes have been made to truss and kdump output: - The flags passed to open() are now properly decoded in that one of O_RDONLY, O_RDWR, O_WRONLY, or O_EXEC is always included in a decoded mask. - Optional arguments to open(), openat(), and fcntl() are only printed in kdump if they exist (e.g. the mode is only printed for open() if O_CREAT is set in the flags). - Print argument to F_GETLK/SETLK/SETLKW in kdump as a pointer, not int. - Include all procctl() commands. - Correctly decode pipe2() flags in truss by not assuming full open()-like flags with O_RDONLY, etc. - Decode file flags passed to *chflags() as file flags (UF_* and SF_*) rather than as a file mode. - Fix decoding of quotactl() commands by splitting out the two command components instead of assuming the raw command value matches the primary command component. In addition, truss and kdump now build without triggering any warnings. All of the sysdecode manpages now include the required headers in the synopsis. Reviewed by: kib (several older versions), wblock (manpages) MFC after: 2 months Differential Revision: https://reviews.freebsd.org/D7847
2016-10-17 22:37:07 +00:00
print_integer_arg(sysdecode_fcntl_cmd, fp, args[sc->offset]);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
break;
case Mprot:
Move mksubr from kdump into libsysdecode. Restructure this script so that it generates a header of tables instead of a source file. The tables are included in a flags.c source file which provides functions to decode various system call arguments. For functions that decode an enumeration, the function returns a pointer to a string for known values and NULL for unknown values. For functions that do more complex decoding (typically of a bitmask), the function accepts a pointer to a FILE object (open_memstream() can be used as a string builder) to which decoded values are written. If the function operates on a bitmask, the function returns true if any bits were decoded or false if the entire value was valid. Additionally, the third argument accepts a pointer to a value to which any undecoded bits are stored. This pointer can be NULL if the caller doesn't care about remaining bits. Convert kdump over to using decoder functions from libsysdecode instead of mksubr. truss also uses decoders from libsysdecode instead of private lookup tables, though lookup tables for objects not decoded by kdump remain in truss for now. Eventually most of these tables should move into libsysdecode as the automated table generation approach from mksubr is less stale than the static tables in truss. Some changes have been made to truss and kdump output: - The flags passed to open() are now properly decoded in that one of O_RDONLY, O_RDWR, O_WRONLY, or O_EXEC is always included in a decoded mask. - Optional arguments to open(), openat(), and fcntl() are only printed in kdump if they exist (e.g. the mode is only printed for open() if O_CREAT is set in the flags). - Print argument to F_GETLK/SETLK/SETLKW in kdump as a pointer, not int. - Include all procctl() commands. - Correctly decode pipe2() flags in truss by not assuming full open()-like flags with O_RDONLY, etc. - Decode file flags passed to *chflags() as file flags (UF_* and SF_*) rather than as a file mode. - Fix decoding of quotactl() commands by splitting out the two command components instead of assuming the raw command value matches the primary command component. In addition, truss and kdump now build without triggering any warnings. All of the sysdecode manpages now include the required headers in the synopsis. Reviewed by: kib (several older versions), wblock (manpages) MFC after: 2 months Differential Revision: https://reviews.freebsd.org/D7847
2016-10-17 22:37:07 +00:00
print_mask_arg(sysdecode_mmap_prot, fp, args[sc->offset]);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
break;
Move mksubr from kdump into libsysdecode. Restructure this script so that it generates a header of tables instead of a source file. The tables are included in a flags.c source file which provides functions to decode various system call arguments. For functions that decode an enumeration, the function returns a pointer to a string for known values and NULL for unknown values. For functions that do more complex decoding (typically of a bitmask), the function accepts a pointer to a FILE object (open_memstream() can be used as a string builder) to which decoded values are written. If the function operates on a bitmask, the function returns true if any bits were decoded or false if the entire value was valid. Additionally, the third argument accepts a pointer to a value to which any undecoded bits are stored. This pointer can be NULL if the caller doesn't care about remaining bits. Convert kdump over to using decoder functions from libsysdecode instead of mksubr. truss also uses decoders from libsysdecode instead of private lookup tables, though lookup tables for objects not decoded by kdump remain in truss for now. Eventually most of these tables should move into libsysdecode as the automated table generation approach from mksubr is less stale than the static tables in truss. Some changes have been made to truss and kdump output: - The flags passed to open() are now properly decoded in that one of O_RDONLY, O_RDWR, O_WRONLY, or O_EXEC is always included in a decoded mask. - Optional arguments to open(), openat(), and fcntl() are only printed in kdump if they exist (e.g. the mode is only printed for open() if O_CREAT is set in the flags). - Print argument to F_GETLK/SETLK/SETLKW in kdump as a pointer, not int. - Include all procctl() commands. - Correctly decode pipe2() flags in truss by not assuming full open()-like flags with O_RDONLY, etc. - Decode file flags passed to *chflags() as file flags (UF_* and SF_*) rather than as a file mode. - Fix decoding of quotactl() commands by splitting out the two command components instead of assuming the raw command value matches the primary command component. In addition, truss and kdump now build without triggering any warnings. All of the sysdecode manpages now include the required headers in the synopsis. Reviewed by: kib (several older versions), wblock (manpages) MFC after: 2 months Differential Revision: https://reviews.freebsd.org/D7847
2016-10-17 22:37:07 +00:00
case Mmapflags:
print_mask_arg(sysdecode_mmap_flags, fp, args[sc->offset]);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
break;
case Whence:
Move mksubr from kdump into libsysdecode. Restructure this script so that it generates a header of tables instead of a source file. The tables are included in a flags.c source file which provides functions to decode various system call arguments. For functions that decode an enumeration, the function returns a pointer to a string for known values and NULL for unknown values. For functions that do more complex decoding (typically of a bitmask), the function accepts a pointer to a FILE object (open_memstream() can be used as a string builder) to which decoded values are written. If the function operates on a bitmask, the function returns true if any bits were decoded or false if the entire value was valid. Additionally, the third argument accepts a pointer to a value to which any undecoded bits are stored. This pointer can be NULL if the caller doesn't care about remaining bits. Convert kdump over to using decoder functions from libsysdecode instead of mksubr. truss also uses decoders from libsysdecode instead of private lookup tables, though lookup tables for objects not decoded by kdump remain in truss for now. Eventually most of these tables should move into libsysdecode as the automated table generation approach from mksubr is less stale than the static tables in truss. Some changes have been made to truss and kdump output: - The flags passed to open() are now properly decoded in that one of O_RDONLY, O_RDWR, O_WRONLY, or O_EXEC is always included in a decoded mask. - Optional arguments to open(), openat(), and fcntl() are only printed in kdump if they exist (e.g. the mode is only printed for open() if O_CREAT is set in the flags). - Print argument to F_GETLK/SETLK/SETLKW in kdump as a pointer, not int. - Include all procctl() commands. - Correctly decode pipe2() flags in truss by not assuming full open()-like flags with O_RDONLY, etc. - Decode file flags passed to *chflags() as file flags (UF_* and SF_*) rather than as a file mode. - Fix decoding of quotactl() commands by splitting out the two command components instead of assuming the raw command value matches the primary command component. In addition, truss and kdump now build without triggering any warnings. All of the sysdecode manpages now include the required headers in the synopsis. Reviewed by: kib (several older versions), wblock (manpages) MFC after: 2 months Differential Revision: https://reviews.freebsd.org/D7847
2016-10-17 22:37:07 +00:00
print_integer_arg(sysdecode_whence, fp, args[sc->offset]);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
break;
case Sockdomain:
Move mksubr from kdump into libsysdecode. Restructure this script so that it generates a header of tables instead of a source file. The tables are included in a flags.c source file which provides functions to decode various system call arguments. For functions that decode an enumeration, the function returns a pointer to a string for known values and NULL for unknown values. For functions that do more complex decoding (typically of a bitmask), the function accepts a pointer to a FILE object (open_memstream() can be used as a string builder) to which decoded values are written. If the function operates on a bitmask, the function returns true if any bits were decoded or false if the entire value was valid. Additionally, the third argument accepts a pointer to a value to which any undecoded bits are stored. This pointer can be NULL if the caller doesn't care about remaining bits. Convert kdump over to using decoder functions from libsysdecode instead of mksubr. truss also uses decoders from libsysdecode instead of private lookup tables, though lookup tables for objects not decoded by kdump remain in truss for now. Eventually most of these tables should move into libsysdecode as the automated table generation approach from mksubr is less stale than the static tables in truss. Some changes have been made to truss and kdump output: - The flags passed to open() are now properly decoded in that one of O_RDONLY, O_RDWR, O_WRONLY, or O_EXEC is always included in a decoded mask. - Optional arguments to open(), openat(), and fcntl() are only printed in kdump if they exist (e.g. the mode is only printed for open() if O_CREAT is set in the flags). - Print argument to F_GETLK/SETLK/SETLKW in kdump as a pointer, not int. - Include all procctl() commands. - Correctly decode pipe2() flags in truss by not assuming full open()-like flags with O_RDONLY, etc. - Decode file flags passed to *chflags() as file flags (UF_* and SF_*) rather than as a file mode. - Fix decoding of quotactl() commands by splitting out the two command components instead of assuming the raw command value matches the primary command component. In addition, truss and kdump now build without triggering any warnings. All of the sysdecode manpages now include the required headers in the synopsis. Reviewed by: kib (several older versions), wblock (manpages) MFC after: 2 months Differential Revision: https://reviews.freebsd.org/D7847
2016-10-17 22:37:07 +00:00
print_integer_arg(sysdecode_socketdomain, fp, args[sc->offset]);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
break;
Move mksubr from kdump into libsysdecode. Restructure this script so that it generates a header of tables instead of a source file. The tables are included in a flags.c source file which provides functions to decode various system call arguments. For functions that decode an enumeration, the function returns a pointer to a string for known values and NULL for unknown values. For functions that do more complex decoding (typically of a bitmask), the function accepts a pointer to a FILE object (open_memstream() can be used as a string builder) to which decoded values are written. If the function operates on a bitmask, the function returns true if any bits were decoded or false if the entire value was valid. Additionally, the third argument accepts a pointer to a value to which any undecoded bits are stored. This pointer can be NULL if the caller doesn't care about remaining bits. Convert kdump over to using decoder functions from libsysdecode instead of mksubr. truss also uses decoders from libsysdecode instead of private lookup tables, though lookup tables for objects not decoded by kdump remain in truss for now. Eventually most of these tables should move into libsysdecode as the automated table generation approach from mksubr is less stale than the static tables in truss. Some changes have been made to truss and kdump output: - The flags passed to open() are now properly decoded in that one of O_RDONLY, O_RDWR, O_WRONLY, or O_EXEC is always included in a decoded mask. - Optional arguments to open(), openat(), and fcntl() are only printed in kdump if they exist (e.g. the mode is only printed for open() if O_CREAT is set in the flags). - Print argument to F_GETLK/SETLK/SETLKW in kdump as a pointer, not int. - Include all procctl() commands. - Correctly decode pipe2() flags in truss by not assuming full open()-like flags with O_RDONLY, etc. - Decode file flags passed to *chflags() as file flags (UF_* and SF_*) rather than as a file mode. - Fix decoding of quotactl() commands by splitting out the two command components instead of assuming the raw command value matches the primary command component. In addition, truss and kdump now build without triggering any warnings. All of the sysdecode manpages now include the required headers in the synopsis. Reviewed by: kib (several older versions), wblock (manpages) MFC after: 2 months Differential Revision: https://reviews.freebsd.org/D7847
2016-10-17 22:37:07 +00:00
case Socktype:
print_mask_arg(sysdecode_socket_type, fp, args[sc->offset]);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
break;
case Shutdown:
Move mksubr from kdump into libsysdecode. Restructure this script so that it generates a header of tables instead of a source file. The tables are included in a flags.c source file which provides functions to decode various system call arguments. For functions that decode an enumeration, the function returns a pointer to a string for known values and NULL for unknown values. For functions that do more complex decoding (typically of a bitmask), the function accepts a pointer to a FILE object (open_memstream() can be used as a string builder) to which decoded values are written. If the function operates on a bitmask, the function returns true if any bits were decoded or false if the entire value was valid. Additionally, the third argument accepts a pointer to a value to which any undecoded bits are stored. This pointer can be NULL if the caller doesn't care about remaining bits. Convert kdump over to using decoder functions from libsysdecode instead of mksubr. truss also uses decoders from libsysdecode instead of private lookup tables, though lookup tables for objects not decoded by kdump remain in truss for now. Eventually most of these tables should move into libsysdecode as the automated table generation approach from mksubr is less stale than the static tables in truss. Some changes have been made to truss and kdump output: - The flags passed to open() are now properly decoded in that one of O_RDONLY, O_RDWR, O_WRONLY, or O_EXEC is always included in a decoded mask. - Optional arguments to open(), openat(), and fcntl() are only printed in kdump if they exist (e.g. the mode is only printed for open() if O_CREAT is set in the flags). - Print argument to F_GETLK/SETLK/SETLKW in kdump as a pointer, not int. - Include all procctl() commands. - Correctly decode pipe2() flags in truss by not assuming full open()-like flags with O_RDONLY, etc. - Decode file flags passed to *chflags() as file flags (UF_* and SF_*) rather than as a file mode. - Fix decoding of quotactl() commands by splitting out the two command components instead of assuming the raw command value matches the primary command component. In addition, truss and kdump now build without triggering any warnings. All of the sysdecode manpages now include the required headers in the synopsis. Reviewed by: kib (several older versions), wblock (manpages) MFC after: 2 months Differential Revision: https://reviews.freebsd.org/D7847
2016-10-17 22:37:07 +00:00
print_integer_arg(sysdecode_shutdown_how, fp, args[sc->offset]);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
break;
case Resource:
Move mksubr from kdump into libsysdecode. Restructure this script so that it generates a header of tables instead of a source file. The tables are included in a flags.c source file which provides functions to decode various system call arguments. For functions that decode an enumeration, the function returns a pointer to a string for known values and NULL for unknown values. For functions that do more complex decoding (typically of a bitmask), the function accepts a pointer to a FILE object (open_memstream() can be used as a string builder) to which decoded values are written. If the function operates on a bitmask, the function returns true if any bits were decoded or false if the entire value was valid. Additionally, the third argument accepts a pointer to a value to which any undecoded bits are stored. This pointer can be NULL if the caller doesn't care about remaining bits. Convert kdump over to using decoder functions from libsysdecode instead of mksubr. truss also uses decoders from libsysdecode instead of private lookup tables, though lookup tables for objects not decoded by kdump remain in truss for now. Eventually most of these tables should move into libsysdecode as the automated table generation approach from mksubr is less stale than the static tables in truss. Some changes have been made to truss and kdump output: - The flags passed to open() are now properly decoded in that one of O_RDONLY, O_RDWR, O_WRONLY, or O_EXEC is always included in a decoded mask. - Optional arguments to open(), openat(), and fcntl() are only printed in kdump if they exist (e.g. the mode is only printed for open() if O_CREAT is set in the flags). - Print argument to F_GETLK/SETLK/SETLKW in kdump as a pointer, not int. - Include all procctl() commands. - Correctly decode pipe2() flags in truss by not assuming full open()-like flags with O_RDONLY, etc. - Decode file flags passed to *chflags() as file flags (UF_* and SF_*) rather than as a file mode. - Fix decoding of quotactl() commands by splitting out the two command components instead of assuming the raw command value matches the primary command component. In addition, truss and kdump now build without triggering any warnings. All of the sysdecode manpages now include the required headers in the synopsis. Reviewed by: kib (several older versions), wblock (manpages) MFC after: 2 months Differential Revision: https://reviews.freebsd.org/D7847
2016-10-17 22:37:07 +00:00
print_integer_arg(sysdecode_rlimit, fp, args[sc->offset]);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
break;
Decode the 'who' argument passed to getrusage(). Add a new sysdecode_getrusage_who() which decodes the RUSAGE_* constant passed as the first argument to getrusage(). Use this function in both kdump and truss to decode the first argument to getrusage(). PR: 215448 Submitted by: Anton Yuzhaninov <citrin+pr@citrin.ru> MFC after: 1 month
2017-06-03 14:22:15 +00:00
case RusageWho:
print_integer_arg(sysdecode_getrusage_who, fp, args[sc->offset]);
break;
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
case Pathconf:
Decode pathconf() names, *at() flags, and sysarch() numbers in libsysdecode. Move tables that were previously in truss over to libsysdecode. truss output is unchanged, but kdump has been updated to decode these fields. In addition, sysdecode_sysarch_number() should support all platforms whereas the old table in truss only supported x86.
2017-09-04 05:34:36 +00:00
print_integer_arg(sysdecode_pathconf_name, fp, args[sc->offset]);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
break;
Properly print arguments to vfork(2) and rfork(2). PR: 180976 Submitted by: Yuri <yuri@tsoft.com> (original patch) MFC after: 1 week
2013-08-01 02:57:04 +00:00
case Rforkflags:
Move mksubr from kdump into libsysdecode. Restructure this script so that it generates a header of tables instead of a source file. The tables are included in a flags.c source file which provides functions to decode various system call arguments. For functions that decode an enumeration, the function returns a pointer to a string for known values and NULL for unknown values. For functions that do more complex decoding (typically of a bitmask), the function accepts a pointer to a FILE object (open_memstream() can be used as a string builder) to which decoded values are written. If the function operates on a bitmask, the function returns true if any bits were decoded or false if the entire value was valid. Additionally, the third argument accepts a pointer to a value to which any undecoded bits are stored. This pointer can be NULL if the caller doesn't care about remaining bits. Convert kdump over to using decoder functions from libsysdecode instead of mksubr. truss also uses decoders from libsysdecode instead of private lookup tables, though lookup tables for objects not decoded by kdump remain in truss for now. Eventually most of these tables should move into libsysdecode as the automated table generation approach from mksubr is less stale than the static tables in truss. Some changes have been made to truss and kdump output: - The flags passed to open() are now properly decoded in that one of O_RDONLY, O_RDWR, O_WRONLY, or O_EXEC is always included in a decoded mask. - Optional arguments to open(), openat(), and fcntl() are only printed in kdump if they exist (e.g. the mode is only printed for open() if O_CREAT is set in the flags). - Print argument to F_GETLK/SETLK/SETLKW in kdump as a pointer, not int. - Include all procctl() commands. - Correctly decode pipe2() flags in truss by not assuming full open()-like flags with O_RDONLY, etc. - Decode file flags passed to *chflags() as file flags (UF_* and SF_*) rather than as a file mode. - Fix decoding of quotactl() commands by splitting out the two command components instead of assuming the raw command value matches the primary command component. In addition, truss and kdump now build without triggering any warnings. All of the sysdecode manpages now include the required headers in the synopsis. Reviewed by: kib (several older versions), wblock (manpages) MFC after: 2 months Differential Revision: https://reviews.freebsd.org/D7847
2016-10-17 22:37:07 +00:00
print_mask_arg(sysdecode_rfork_flags, fp, args[sc->offset]);
Properly print arguments to vfork(2) and rfork(2). PR: 180976 Submitted by: Yuri <yuri@tsoft.com> (original patch) MFC after: 1 week
2013-08-01 02:57:04 +00:00
break;
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
case Sockaddr: {
Rather than groveling around in a socket address structure for a socket address's length (and then overriding it if it "looks wrong"), use the next argument to the system call to determine the length. This is more reliable since this is what the kernel depends on anyway and is also simpler.
2015-10-01 18:18:58 +00:00
socklen_t len;
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
if (args[sc->offset] == 0) {
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
fputs("NULL", fp);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
break;
}
/*
Rather than groveling around in a socket address structure for a socket address's length (and then overriding it if it "looks wrong"), use the next argument to the system call to determine the length. This is more reliable since this is what the kernel depends on anyway and is also simpler.
2015-10-01 18:18:58 +00:00
* Extract the address length from the next argument. If
* this is an output sockaddr (OUT is set), then the
* next argument is a pointer to a socklen_t. Otherwise
* the next argument contains a socklen_t by value.
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
*/
Rather than groveling around in a socket address structure for a socket address's length (and then overriding it if it "looks wrong"), use the next argument to the system call to determine the length. This is more reliable since this is what the kernel depends on anyway and is also simpler.
2015-10-01 18:18:58 +00:00
if (sc->type & OUT) {
if (get_struct(pid, (void *)args[sc->offset + 1],
&len, sizeof(len)) == -1) {
fprintf(fp, "0x%lx", args[sc->offset]);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
break;
}
Rather than groveling around in a socket address structure for a socket address's length (and then overriding it if it "looks wrong"), use the next argument to the system call to determine the length. This is more reliable since this is what the kernel depends on anyway and is also simpler.
2015-10-01 18:18:58 +00:00
} else
len = args[sc->offset + 1];
Decode msghdr argument of sendmsg() and recvmsg(). Sponsored by: Netflix, Inc.
2018-01-15 20:39:42 +00:00
print_sockaddr(fp, trussinfo, (void *)args[sc->offset], len);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
break;
decode fcntl and mmap arguments.
2004-03-23 09:04:06 +00:00
}
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
case Sigaction: {
struct sigaction sa;
- Style(9) cleanup. Approved by: kib (mentor)
2012-09-02 11:03:18 +00:00
if (get_struct(pid, (void *)args[sc->offset], &sa, sizeof(sa))
!= -1) {
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
fputs("{ ", fp);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
if (sa.sa_handler == SIG_DFL)
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
fputs("SIG_DFL", fp);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
else if (sa.sa_handler == SIG_IGN)
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
fputs("SIG_IGN", fp);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
else
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
fprintf(fp, "%p", sa.sa_handler);
fprintf(fp, " %s ss_t }",
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
xlookup_bits(sigaction_flags, sa.sa_flags));
- Style(9) cleanup. Approved by: kib (mentor)
2012-09-02 11:03:18 +00:00
} else
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
fprintf(fp, "0x%lx", args[sc->offset]);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
break;
Add missing include for <ctype.h> Make sockaddr printing code unbad. Always copy in sizeof(sockaddr_un) bytes for an AF_UNIX sockaddr, despite what the length may be.
2001-11-06 19:26:51 +00:00
}
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
case Kevent: {
/*
Various style and whitespace fixes.
2015-08-19 20:02:03 +00:00
* XXX XXX: The size of the array is determined by either the
* next syscall argument, or by the syscall return value,
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
* depending on which argument number we are. This matches the
* kevent syscall, but luckily that's the only syscall that uses
* them.
*/
struct kevent *ke;
int numevents = -1;
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
size_t bytes;
int i;
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
if (sc->offset == 1)
numevents = args[sc->offset+1];
Several changes to truss. - Refactor the interface between the ABI-independent code and the ABI-specific backends. The backends now provide smaller hooks to fetch system call arguments and return values. The rest of the system call entry and exit handling that was previously duplicated among all the backends has been moved to one place. - Merge the loop when waiting for an event with the loop for handling stops. This also means not emulating a procfs-like interface on top of ptrace(). Instead, use a single event loop that fetches process events via waitid(). Among other things this allows us to report the full 32-bit exit value. - Use PT_FOLLOW_FORK to follow new child processes instead of forking a new truss process for each new child. This allows one truss process to monitor a tree of processes and truss -c should now display one total for the entire tree instead of separate summaries per process. - Use the recently added fields to ptrace_lwpinfo to determine the current system call number and argument count. The latter is especially useful and fixes a regression since the conversion from procfs. truss now generally prints the correct number of arguments for most system calls rather than printing extra arguments for any call not listed in the table in syscalls.c. - Actually check the new ABI when processes call exec. The comments claimed that this happened but it was not being done (perhaps this was another regression in the conversion to ptrace()). If the new ABI after exec is not supported, truss detaches from the process. If truss does not support the ABI for a newly executed process the process is killed before it returns from exec. - Along with the refactor, teach the various ABI-specific backends to fetch both return values, not just the first. Use this to properly report the full 64-bit return value from lseek(). In addition, the handler for "pipe" now pulls the pair of descriptors out of the return values (which is the true kernel system call interface) but displays them as an argument (which matches the interface exported by libc). - Each ABI handler adds entries to a linker set rather than requiring a statically defined table of handlers in main.c. - The arm and mips system call fetching code was changed to follow the same pattern as amd64 (and the in-kernel handler) of fetching register arguments first and then reading any remaining arguments from the stack. This should fix indirect system call arguments on at least arm. - The mipsn32 and n64 ABIs will now look for arguments in A4 through A7. - Use register %ebp for the 6th system call argument for Linux/i386 ABIs to match the in-kernel argument fetch code. - For powerpc binaries on a powerpc64 system, fetch the extra arguments on the stack as 32-bit values that are then copied into the 64-bit argument array instead of reading the 32-bit values directly into the 64-bit array. Reviewed by: kib (earlier version) Tested on: amd64 (FreeBSD/amd64 & i386), i386, arm (earlier version) Tested on: powerpc64 (FreeBSD/powerpc64 & powerpc) MFC after: 1 month Differential Revision: https://reviews.freebsd.org/D3575
2015-09-30 19:13:32 +00:00
else if (sc->offset == 3 && retval[0] != -1)
numevents = retval[0];
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
if (numevents >= 0) {
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
bytes = sizeof(struct kevent) * numevents;
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
if ((ke = malloc(bytes)) == NULL)
err(1,
"Cannot malloc %zu bytes for kevent array",
bytes);
} else
ke = NULL;
- Style(9) cleanup. Approved by: kib (mentor)
2012-09-02 11:03:18 +00:00
if (numevents >= 0 && get_struct(pid, (void *)args[sc->offset],
ke, bytes) != -1) {
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
fputc('{', fp);
Expand the decoding of kevent structures. - Print the ident value as decimal instead of hexadecimal for filter types that use "small" values such as file descriptors and PIDs. - Decode NOTE_* flags in the fflags field of kevents for several system filter types.
2015-08-19 01:44:56 +00:00
for (i = 0; i < numevents; i++) {
fputc(' ', fp);
Decode kevent structures logged via ktrace(2) in kdump. - Add a new KTR_STRUCT_ARRAY ktrace record type which dumps an array of structures. The structure name in the record payload is preceded by a size_t containing the size of the individual structures. Use this to replace the previous code that dumped the kevent arrays dumped for kevent(). kdump is now able to decode the kevent structures rather than dumping their contents via a hexdump. One change from before is that the 'changes' and 'events' arrays are not marked with separate 'read' and 'write' annotations in kdump output. Instead, the first array is the 'changes' array, and the second array (only present if kevent doesn't fail with an error) is the 'events' array. For kevent(), empty arrays are denoted by an entry with an array containing zero entries rather than no record. - Move kevent decoding tables from truss to libsysdecode. This adds three new functions to decode members of struct kevent: sysdecode_kevent_filter, sysdecode_kevent_flags, and sysdecode_kevent_fflags. kdump uses these helper functions to pretty-print kevent fields. - Move structure definitions for freebsd11 and freebsd32 kevent structures to <sys/event.h> so that they can be shared with userland. The 32-bit structures are only exposed if _WANT_KEVENT32 is defined. The freebsd11 structures are only exposed if _WANT_FREEBSD11_KEVENT is defined. The 32-bit freebsd11 structure requires both. - Decode freebsd11 kevent structures in truss for the compat11.kevent() system call. - Log 32-bit kevent structures via ktrace for 32-bit compat kevent() system calls. - While here, constify the 'void *data' argument to ktrstruct(). Reviewed by: kib (earlier version) MFC after: 1 month Differential Revision: https://reviews.freebsd.org/D12470
2017-11-25 04:49:12 +00:00
print_kevent(fp, &ke[i]);
Expand the decoding of kevent structures. - Print the ident value as decimal instead of hexadecimal for filter types that use "small" values such as file descriptors and PIDs. - Decode NOTE_* flags in the fflags field of kevents for several system filter types.
2015-08-19 01:44:56 +00:00
}
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
fputs(" }", fp);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
} else {
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
fprintf(fp, "0x%lx", args[sc->offset]);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
}
free(ke);
break;
}
Decode kevent structures logged via ktrace(2) in kdump. - Add a new KTR_STRUCT_ARRAY ktrace record type which dumps an array of structures. The structure name in the record payload is preceded by a size_t containing the size of the individual structures. Use this to replace the previous code that dumped the kevent arrays dumped for kevent(). kdump is now able to decode the kevent structures rather than dumping their contents via a hexdump. One change from before is that the 'changes' and 'events' arrays are not marked with separate 'read' and 'write' annotations in kdump output. Instead, the first array is the 'changes' array, and the second array (only present if kevent doesn't fail with an error) is the 'events' array. For kevent(), empty arrays are denoted by an entry with an array containing zero entries rather than no record. - Move kevent decoding tables from truss to libsysdecode. This adds three new functions to decode members of struct kevent: sysdecode_kevent_filter, sysdecode_kevent_flags, and sysdecode_kevent_fflags. kdump uses these helper functions to pretty-print kevent fields. - Move structure definitions for freebsd11 and freebsd32 kevent structures to <sys/event.h> so that they can be shared with userland. The 32-bit structures are only exposed if _WANT_KEVENT32 is defined. The freebsd11 structures are only exposed if _WANT_FREEBSD11_KEVENT is defined. The 32-bit freebsd11 structure requires both. - Decode freebsd11 kevent structures in truss for the compat11.kevent() system call. - Log 32-bit kevent structures via ktrace for 32-bit compat kevent() system calls. - While here, constify the 'void *data' argument to ktrstruct(). Reviewed by: kib (earlier version) MFC after: 1 month Differential Revision: https://reviews.freebsd.org/D12470
2017-11-25 04:49:12 +00:00
case Kevent11: {
struct kevent_freebsd11 *ke11;
struct kevent ke;
int numevents = -1;
size_t bytes;
int i;
if (sc->offset == 1)
numevents = args[sc->offset+1];
else if (sc->offset == 3 && retval[0] != -1)
numevents = retval[0];
if (numevents >= 0) {
bytes = sizeof(struct kevent_freebsd11) * numevents;
if ((ke11 = malloc(bytes)) == NULL)
err(1,
"Cannot malloc %zu bytes for kevent array",
bytes);
} else
ke11 = NULL;
memset(&ke, 0, sizeof(ke));
if (numevents >= 0 && get_struct(pid, (void *)args[sc->offset],
ke11, bytes) != -1) {
fputc('{', fp);
for (i = 0; i < numevents; i++) {
fputc(' ', fp);
ke.ident = ke11[i].ident;
ke.filter = ke11[i].filter;
ke.flags = ke11[i].flags;
ke.fflags = ke11[i].fflags;
ke.data = ke11[i].data;
ke.udata = ke11[i].udata;
print_kevent(fp, &ke);
}
fputs(" }", fp);
} else {
fprintf(fp, "0x%lx", args[sc->offset]);
}
free(ke11);
break;
}
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
case Stat: {
struct stat st;
Various style and whitespace fixes.
2015-08-19 20:02:03 +00:00
- Style(9) cleanup. Approved by: kib (mentor)
2012-09-02 11:03:18 +00:00
if (get_struct(pid, (void *)args[sc->offset], &st, sizeof(st))
!= -1) {
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
char mode[12];
Various style and whitespace fixes.
2015-08-19 20:02:03 +00:00
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
strmode(st.st_mode, mode);
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
fprintf(fp,
ino_t is unsigned, so use uintmax_t instead of intmax_t when printing it. Submitted by: bde (sort of)
2015-08-19 20:10:58 +00:00
"{ mode=%s,inode=%ju,size=%jd,blksize=%ld }", mode,
(uintmax_t)st.st_ino, (intmax_t)st.st_size,
- Style(9) cleanup. Approved by: kib (mentor)
2012-09-02 11:03:18 +00:00
(long)st.st_blksize);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
} else {
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
fprintf(fp, "0x%lx", args[sc->offset]);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
}
break;
- Add decoding of kse_release, kevent, sigprocmask, unmount, socket, getrusage, rename, __getcwd, shutdown, getrlimit, setrlimit, _umtx_lock, _umtx_unlock, pathconf, truncate, ftruncate, kill - Decode more arguments of open, mprot, *stat, and fcntl. - Convert all constant-macro and bitfield decoding to lookup tables; much cleaner than previous code. - Print the timestamp of process exit and signal reception when -d or -D are in use - Try six times with 1/2 second delay to debug the child PR: bin/52190 (updated) Submitted by: Dan Nelson <dnelson@allantgroup.com> Approved by: alfred
2006-05-15 21:18:28 +00:00
}
Decode FreeBSD 11 compat stat, fstat and lstat calls.
2017-06-23 18:06:46 +00:00
case Stat11: {
struct freebsd11_stat st;
if (get_struct(pid, (void *)args[sc->offset], &st, sizeof(st))
!= -1) {
char mode[12];
strmode(st.st_mode, mode);
fprintf(fp,
"{ mode=%s,inode=%ju,size=%jd,blksize=%ld }", mode,
(uintmax_t)st.st_ino, (intmax_t)st.st_size,
(long)st.st_blksize);
} else {
fprintf(fp, "0x%lx", args[sc->offset]);
}
break;
}
Add decoding for struct statfs. Reviewed by: jhb (briefly)
2015-10-03 18:57:15 +00:00
case StatFs: {
unsigned int i;
struct statfs buf;
Style fix.
2015-10-03 19:08:36 +00:00
Add decoding for struct statfs. Reviewed by: jhb (briefly)
2015-10-03 18:57:15 +00:00
if (get_struct(pid, (void *)args[sc->offset], &buf,
sizeof(buf)) != -1) {
char fsid[17];
bzero(fsid, sizeof(fsid));
if (buf.f_fsid.val[0] != 0 || buf.f_fsid.val[1] != 0) {
for (i = 0; i < sizeof(buf.f_fsid); i++)
snprintf(&fsid[i*2],
sizeof(fsid) - (i*2), "%02x",
((u_char *)&buf.f_fsid)[i]);
}
fprintf(fp,
"{ fstypename=%s,mntonname=%s,mntfromname=%s,"
"fsid=%s }", buf.f_fstypename, buf.f_mntonname,
buf.f_mntfromname, fsid);
} else
fprintf(fp, "0x%lx", args[sc->offset]);
break;
}
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
case Rusage: {
struct rusage ru;
Various style and whitespace fixes.
2015-08-19 20:02:03 +00:00
- Style(9) cleanup. Approved by: kib (mentor)
2012-09-02 11:03:18 +00:00
if (get_struct(pid, (void *)args[sc->offset], &ru, sizeof(ru))
!= -1) {
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
fprintf(fp,
Always use %j with an intmax_t cast to print time_t values. time_t is longer than long on 32-bit platforms with a 64-bit time_t. Inspired by: mail from bde
2015-08-19 20:09:14 +00:00
"{ u=%jd.%06ld,s=%jd.%06ld,in=%ld,out=%ld }",
(intmax_t)ru.ru_utime.tv_sec, ru.ru_utime.tv_usec,
(intmax_t)ru.ru_stime.tv_sec, ru.ru_stime.tv_usec,
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
ru.ru_inblock, ru.ru_oublock);
- Style(9) cleanup. Approved by: kib (mentor)
2012-09-02 11:03:18 +00:00
} else
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
fprintf(fp, "0x%lx", args[sc->offset]);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
break;
}
case Rlimit: {
struct rlimit rl;
Various style and whitespace fixes.
2015-08-19 20:02:03 +00:00
- Style(9) cleanup. Approved by: kib (mentor)
2012-09-02 11:03:18 +00:00
if (get_struct(pid, (void *)args[sc->offset], &rl, sizeof(rl))
!= -1) {
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
fprintf(fp, "{ cur=%ju,max=%ju }",
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
rl.rlim_cur, rl.rlim_max);
- Style(9) cleanup. Approved by: kib (mentor)
2012-09-02 11:03:18 +00:00
} else
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
fprintf(fp, "0x%lx", args[sc->offset]);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
break;
}
- Decode the idtype argument passed to wait6() in kdump and truss. - Don't treat an options argument of 0 to wait4() as an error in kdump. - Decode the wait options passed to wait4() and wait6() in truss and decode the returned rusage and exit status. Approved by: re (kib) MFC after: 1 week
2013-09-12 18:08:25 +00:00
case ExitStatus: {
int status;
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
- Decode the idtype argument passed to wait6() in kdump and truss. - Don't treat an options argument of 0 to wait4() as an error in kdump. - Decode the wait options passed to wait4() and wait6() in truss and decode the returned rusage and exit status. Approved by: re (kib) MFC after: 1 week
2013-09-12 18:08:25 +00:00
if (get_struct(pid, (void *)args[sc->offset], &status,
sizeof(status)) != -1) {
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
fputs("{ ", fp);
- Decode the idtype argument passed to wait6() in kdump and truss. - Don't treat an options argument of 0 to wait4() as an error in kdump. - Decode the wait options passed to wait4() and wait6() in truss and decode the returned rusage and exit status. Approved by: re (kib) MFC after: 1 week
2013-09-12 18:08:25 +00:00
if (WIFCONTINUED(status))
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
fputs("CONTINUED", fp);
- Decode the idtype argument passed to wait6() in kdump and truss. - Don't treat an options argument of 0 to wait4() as an error in kdump. - Decode the wait options passed to wait4() and wait6() in truss and decode the returned rusage and exit status. Approved by: re (kib) MFC after: 1 week
2013-09-12 18:08:25 +00:00
else if (WIFEXITED(status))
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
fprintf(fp, "EXITED,val=%d",
- Decode the idtype argument passed to wait6() in kdump and truss. - Don't treat an options argument of 0 to wait4() as an error in kdump. - Decode the wait options passed to wait4() and wait6() in truss and decode the returned rusage and exit status. Approved by: re (kib) MFC after: 1 week
2013-09-12 18:08:25 +00:00
WEXITSTATUS(status));
else if (WIFSIGNALED(status))
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
fprintf(fp, "SIGNALED,sig=%s%s",
strsig2(WTERMSIG(status)),
- Decode the idtype argument passed to wait6() in kdump and truss. - Don't treat an options argument of 0 to wait4() as an error in kdump. - Decode the wait options passed to wait4() and wait6() in truss and decode the returned rusage and exit status. Approved by: re (kib) MFC after: 1 week
2013-09-12 18:08:25 +00:00
WCOREDUMP(status) ? ",cored" : "");
else
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
fprintf(fp, "STOPPED,sig=%s",
strsig2(WTERMSIG(status)));
fputs(" }", fp);
- Decode the idtype argument passed to wait6() in kdump and truss. - Don't treat an options argument of 0 to wait4() as an error in kdump. - Decode the wait options passed to wait4() and wait6() in truss and decode the returned rusage and exit status. Approved by: re (kib) MFC after: 1 week
2013-09-12 18:08:25 +00:00
} else
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
fprintf(fp, "0x%lx", args[sc->offset]);
- Decode the idtype argument passed to wait6() in kdump and truss. - Don't treat an options argument of 0 to wait4() as an error in kdump. - Decode the wait options passed to wait4() and wait6() in truss and decode the returned rusage and exit status. Approved by: re (kib) MFC after: 1 week
2013-09-12 18:08:25 +00:00
break;
}
case Waitoptions:
Move mksubr from kdump into libsysdecode. Restructure this script so that it generates a header of tables instead of a source file. The tables are included in a flags.c source file which provides functions to decode various system call arguments. For functions that decode an enumeration, the function returns a pointer to a string for known values and NULL for unknown values. For functions that do more complex decoding (typically of a bitmask), the function accepts a pointer to a FILE object (open_memstream() can be used as a string builder) to which decoded values are written. If the function operates on a bitmask, the function returns true if any bits were decoded or false if the entire value was valid. Additionally, the third argument accepts a pointer to a value to which any undecoded bits are stored. This pointer can be NULL if the caller doesn't care about remaining bits. Convert kdump over to using decoder functions from libsysdecode instead of mksubr. truss also uses decoders from libsysdecode instead of private lookup tables, though lookup tables for objects not decoded by kdump remain in truss for now. Eventually most of these tables should move into libsysdecode as the automated table generation approach from mksubr is less stale than the static tables in truss. Some changes have been made to truss and kdump output: - The flags passed to open() are now properly decoded in that one of O_RDONLY, O_RDWR, O_WRONLY, or O_EXEC is always included in a decoded mask. - Optional arguments to open(), openat(), and fcntl() are only printed in kdump if they exist (e.g. the mode is only printed for open() if O_CREAT is set in the flags). - Print argument to F_GETLK/SETLK/SETLKW in kdump as a pointer, not int. - Include all procctl() commands. - Correctly decode pipe2() flags in truss by not assuming full open()-like flags with O_RDONLY, etc. - Decode file flags passed to *chflags() as file flags (UF_* and SF_*) rather than as a file mode. - Fix decoding of quotactl() commands by splitting out the two command components instead of assuming the raw command value matches the primary command component. In addition, truss and kdump now build without triggering any warnings. All of the sysdecode manpages now include the required headers in the synopsis. Reviewed by: kib (several older versions), wblock (manpages) MFC after: 2 months Differential Revision: https://reviews.freebsd.org/D7847
2016-10-17 22:37:07 +00:00
print_mask_arg(sysdecode_wait6_options, fp, args[sc->offset]);
- Decode the idtype argument passed to wait6() in kdump and truss. - Don't treat an options argument of 0 to wait4() as an error in kdump. - Decode the wait options passed to wait4() and wait6() in truss and decode the returned rusage and exit status. Approved by: re (kib) MFC after: 1 week
2013-09-12 18:08:25 +00:00
break;
case Idtype:
Move mksubr from kdump into libsysdecode. Restructure this script so that it generates a header of tables instead of a source file. The tables are included in a flags.c source file which provides functions to decode various system call arguments. For functions that decode an enumeration, the function returns a pointer to a string for known values and NULL for unknown values. For functions that do more complex decoding (typically of a bitmask), the function accepts a pointer to a FILE object (open_memstream() can be used as a string builder) to which decoded values are written. If the function operates on a bitmask, the function returns true if any bits were decoded or false if the entire value was valid. Additionally, the third argument accepts a pointer to a value to which any undecoded bits are stored. This pointer can be NULL if the caller doesn't care about remaining bits. Convert kdump over to using decoder functions from libsysdecode instead of mksubr. truss also uses decoders from libsysdecode instead of private lookup tables, though lookup tables for objects not decoded by kdump remain in truss for now. Eventually most of these tables should move into libsysdecode as the automated table generation approach from mksubr is less stale than the static tables in truss. Some changes have been made to truss and kdump output: - The flags passed to open() are now properly decoded in that one of O_RDONLY, O_RDWR, O_WRONLY, or O_EXEC is always included in a decoded mask. - Optional arguments to open(), openat(), and fcntl() are only printed in kdump if they exist (e.g. the mode is only printed for open() if O_CREAT is set in the flags). - Print argument to F_GETLK/SETLK/SETLKW in kdump as a pointer, not int. - Include all procctl() commands. - Correctly decode pipe2() flags in truss by not assuming full open()-like flags with O_RDONLY, etc. - Decode file flags passed to *chflags() as file flags (UF_* and SF_*) rather than as a file mode. - Fix decoding of quotactl() commands by splitting out the two command components instead of assuming the raw command value matches the primary command component. In addition, truss and kdump now build without triggering any warnings. All of the sysdecode manpages now include the required headers in the synopsis. Reviewed by: kib (several older versions), wblock (manpages) MFC after: 2 months Differential Revision: https://reviews.freebsd.org/D7847
2016-10-17 22:37:07 +00:00
print_integer_arg(sysdecode_idtype, fp, args[sc->offset]);
- Decode the idtype argument passed to wait6() in kdump and truss. - Don't treat an options argument of 0 to wait4() as an error in kdump. - Decode the wait options passed to wait4() and wait6() in truss and decode the returned rusage and exit status. Approved by: re (kib) MFC after: 1 week
2013-09-12 18:08:25 +00:00
break;
Extend the support for exempting processes from being killed when swap is exhausted. - Add a new protect(1) command that can be used to set or revoke protection from arbitrary processes. Similar to ktrace it can apply a change to all existing descendants of a process as well as future descendants. - Add a new procctl(2) system call that provides a generic interface for control operations on processes (as opposed to the debugger-specific operations provided by ptrace(2)). procctl(2) uses a combination of idtype_t and an id to identify the set of processes on which to operate similar to wait6(). - Add a PROC_SPROTECT control operation to manage the protection status of a set of processes. MADV_PROTECT still works for backwards compatability. - Add a p_flag2 to struct proc (and a corresponding ki_flag2 to kinfo_proc) the first bit of which is used to track if P_PROTECT should be inherited by new child processes. Reviewed by: kib, jilles (earlier version) Approved by: re (delphij) MFC after: 1 month
2013-09-19 18:53:42 +00:00
case Procctl:
Move mksubr from kdump into libsysdecode. Restructure this script so that it generates a header of tables instead of a source file. The tables are included in a flags.c source file which provides functions to decode various system call arguments. For functions that decode an enumeration, the function returns a pointer to a string for known values and NULL for unknown values. For functions that do more complex decoding (typically of a bitmask), the function accepts a pointer to a FILE object (open_memstream() can be used as a string builder) to which decoded values are written. If the function operates on a bitmask, the function returns true if any bits were decoded or false if the entire value was valid. Additionally, the third argument accepts a pointer to a value to which any undecoded bits are stored. This pointer can be NULL if the caller doesn't care about remaining bits. Convert kdump over to using decoder functions from libsysdecode instead of mksubr. truss also uses decoders from libsysdecode instead of private lookup tables, though lookup tables for objects not decoded by kdump remain in truss for now. Eventually most of these tables should move into libsysdecode as the automated table generation approach from mksubr is less stale than the static tables in truss. Some changes have been made to truss and kdump output: - The flags passed to open() are now properly decoded in that one of O_RDONLY, O_RDWR, O_WRONLY, or O_EXEC is always included in a decoded mask. - Optional arguments to open(), openat(), and fcntl() are only printed in kdump if they exist (e.g. the mode is only printed for open() if O_CREAT is set in the flags). - Print argument to F_GETLK/SETLK/SETLKW in kdump as a pointer, not int. - Include all procctl() commands. - Correctly decode pipe2() flags in truss by not assuming full open()-like flags with O_RDONLY, etc. - Decode file flags passed to *chflags() as file flags (UF_* and SF_*) rather than as a file mode. - Fix decoding of quotactl() commands by splitting out the two command components instead of assuming the raw command value matches the primary command component. In addition, truss and kdump now build without triggering any warnings. All of the sysdecode manpages now include the required headers in the synopsis. Reviewed by: kib (several older versions), wblock (manpages) MFC after: 2 months Differential Revision: https://reviews.freebsd.org/D7847
2016-10-17 22:37:07 +00:00
print_integer_arg(sysdecode_procctl_cmd, fp, args[sc->offset]);
Extend the support for exempting processes from being killed when swap is exhausted. - Add a new protect(1) command that can be used to set or revoke protection from arbitrary processes. Similar to ktrace it can apply a change to all existing descendants of a process as well as future descendants. - Add a new procctl(2) system call that provides a generic interface for control operations on processes (as opposed to the debugger-specific operations provided by ptrace(2)). procctl(2) uses a combination of idtype_t and an id to identify the set of processes on which to operate similar to wait6(). - Add a PROC_SPROTECT control operation to manage the protection status of a set of processes. MADV_PROTECT still works for backwards compatability. - Add a p_flag2 to struct proc (and a corresponding ki_flag2 to kinfo_proc) the first bit of which is used to track if P_PROTECT should be inherited by new child processes. Reviewed by: kib, jilles (earlier version) Approved by: re (delphij) MFC after: 1 month
2013-09-19 18:53:42 +00:00
break;
Decode the arguments passed to _umtx_op(). In particular, decode the opcode. MFC after: 1 week Sponsored by: Norse
2014-10-13 16:37:06 +00:00
case Umtxop:
Move mksubr from kdump into libsysdecode. Restructure this script so that it generates a header of tables instead of a source file. The tables are included in a flags.c source file which provides functions to decode various system call arguments. For functions that decode an enumeration, the function returns a pointer to a string for known values and NULL for unknown values. For functions that do more complex decoding (typically of a bitmask), the function accepts a pointer to a FILE object (open_memstream() can be used as a string builder) to which decoded values are written. If the function operates on a bitmask, the function returns true if any bits were decoded or false if the entire value was valid. Additionally, the third argument accepts a pointer to a value to which any undecoded bits are stored. This pointer can be NULL if the caller doesn't care about remaining bits. Convert kdump over to using decoder functions from libsysdecode instead of mksubr. truss also uses decoders from libsysdecode instead of private lookup tables, though lookup tables for objects not decoded by kdump remain in truss for now. Eventually most of these tables should move into libsysdecode as the automated table generation approach from mksubr is less stale than the static tables in truss. Some changes have been made to truss and kdump output: - The flags passed to open() are now properly decoded in that one of O_RDONLY, O_RDWR, O_WRONLY, or O_EXEC is always included in a decoded mask. - Optional arguments to open(), openat(), and fcntl() are only printed in kdump if they exist (e.g. the mode is only printed for open() if O_CREAT is set in the flags). - Print argument to F_GETLK/SETLK/SETLKW in kdump as a pointer, not int. - Include all procctl() commands. - Correctly decode pipe2() flags in truss by not assuming full open()-like flags with O_RDONLY, etc. - Decode file flags passed to *chflags() as file flags (UF_* and SF_*) rather than as a file mode. - Fix decoding of quotactl() commands by splitting out the two command components instead of assuming the raw command value matches the primary command component. In addition, truss and kdump now build without triggering any warnings. All of the sysdecode manpages now include the required headers in the synopsis. Reviewed by: kib (several older versions), wblock (manpages) MFC after: 2 months Differential Revision: https://reviews.freebsd.org/D7847
2016-10-17 22:37:07 +00:00
print_integer_arg(sysdecode_umtx_op, fp, args[sc->offset]);
Decode the arguments passed to _umtx_op(). In particular, decode the opcode. MFC after: 1 week Sponsored by: Norse
2014-10-13 16:37:06 +00:00
break;
Decode the arguments passed to the *at() family of system calls. This is especially useful now that libc's open() always calls openat(). While here, fix a few other things: - Decode the mode argument passed to access(), eaccess(), and faccessat(). - Decode the atfd paramete to pretty-print AT_FDCWD. - Decode the special AT_* flags used with some of the *at() system calls. - Decode arguments for fchmod(), lchmod(), fchown(), lchown(), eaccess(), and futimens(). - Decode both of the timeval structures passed to futimes() instead of just the first one.
2015-08-06 19:08:33 +00:00
case Atfd:
Move mksubr from kdump into libsysdecode. Restructure this script so that it generates a header of tables instead of a source file. The tables are included in a flags.c source file which provides functions to decode various system call arguments. For functions that decode an enumeration, the function returns a pointer to a string for known values and NULL for unknown values. For functions that do more complex decoding (typically of a bitmask), the function accepts a pointer to a FILE object (open_memstream() can be used as a string builder) to which decoded values are written. If the function operates on a bitmask, the function returns true if any bits were decoded or false if the entire value was valid. Additionally, the third argument accepts a pointer to a value to which any undecoded bits are stored. This pointer can be NULL if the caller doesn't care about remaining bits. Convert kdump over to using decoder functions from libsysdecode instead of mksubr. truss also uses decoders from libsysdecode instead of private lookup tables, though lookup tables for objects not decoded by kdump remain in truss for now. Eventually most of these tables should move into libsysdecode as the automated table generation approach from mksubr is less stale than the static tables in truss. Some changes have been made to truss and kdump output: - The flags passed to open() are now properly decoded in that one of O_RDONLY, O_RDWR, O_WRONLY, or O_EXEC is always included in a decoded mask. - Optional arguments to open(), openat(), and fcntl() are only printed in kdump if they exist (e.g. the mode is only printed for open() if O_CREAT is set in the flags). - Print argument to F_GETLK/SETLK/SETLKW in kdump as a pointer, not int. - Include all procctl() commands. - Correctly decode pipe2() flags in truss by not assuming full open()-like flags with O_RDONLY, etc. - Decode file flags passed to *chflags() as file flags (UF_* and SF_*) rather than as a file mode. - Fix decoding of quotactl() commands by splitting out the two command components instead of assuming the raw command value matches the primary command component. In addition, truss and kdump now build without triggering any warnings. All of the sysdecode manpages now include the required headers in the synopsis. Reviewed by: kib (several older versions), wblock (manpages) MFC after: 2 months Differential Revision: https://reviews.freebsd.org/D7847
2016-10-17 22:37:07 +00:00
print_integer_arg(sysdecode_atfd, fp, args[sc->offset]);
Decode the arguments passed to the *at() family of system calls. This is especially useful now that libc's open() always calls openat(). While here, fix a few other things: - Decode the mode argument passed to access(), eaccess(), and faccessat(). - Decode the atfd paramete to pretty-print AT_FDCWD. - Decode the special AT_* flags used with some of the *at() system calls. - Decode arguments for fchmod(), lchmod(), fchown(), lchown(), eaccess(), and futimens(). - Decode both of the timeval structures passed to futimes() instead of just the first one.
2015-08-06 19:08:33 +00:00
break;
case Atflags:
Decode pathconf() names, *at() flags, and sysarch() numbers in libsysdecode. Move tables that were previously in truss over to libsysdecode. truss output is unchanged, but kdump has been updated to decode these fields. In addition, sysdecode_sysarch_number() should support all platforms whereas the old table in truss only supported x86.
2017-09-04 05:34:36 +00:00
print_mask_arg(sysdecode_atflags, fp, args[sc->offset]);
Decode the arguments passed to the *at() family of system calls. This is especially useful now that libc's open() always calls openat(). While here, fix a few other things: - Decode the mode argument passed to access(), eaccess(), and faccessat(). - Decode the atfd paramete to pretty-print AT_FDCWD. - Decode the special AT_* flags used with some of the *at() system calls. - Decode arguments for fchmod(), lchmod(), fchown(), lchown(), eaccess(), and futimens(). - Decode both of the timeval structures passed to futimes() instead of just the first one.
2015-08-06 19:08:33 +00:00
break;
case Accessmode:
Move mksubr from kdump into libsysdecode. Restructure this script so that it generates a header of tables instead of a source file. The tables are included in a flags.c source file which provides functions to decode various system call arguments. For functions that decode an enumeration, the function returns a pointer to a string for known values and NULL for unknown values. For functions that do more complex decoding (typically of a bitmask), the function accepts a pointer to a FILE object (open_memstream() can be used as a string builder) to which decoded values are written. If the function operates on a bitmask, the function returns true if any bits were decoded or false if the entire value was valid. Additionally, the third argument accepts a pointer to a value to which any undecoded bits are stored. This pointer can be NULL if the caller doesn't care about remaining bits. Convert kdump over to using decoder functions from libsysdecode instead of mksubr. truss also uses decoders from libsysdecode instead of private lookup tables, though lookup tables for objects not decoded by kdump remain in truss for now. Eventually most of these tables should move into libsysdecode as the automated table generation approach from mksubr is less stale than the static tables in truss. Some changes have been made to truss and kdump output: - The flags passed to open() are now properly decoded in that one of O_RDONLY, O_RDWR, O_WRONLY, or O_EXEC is always included in a decoded mask. - Optional arguments to open(), openat(), and fcntl() are only printed in kdump if they exist (e.g. the mode is only printed for open() if O_CREAT is set in the flags). - Print argument to F_GETLK/SETLK/SETLKW in kdump as a pointer, not int. - Include all procctl() commands. - Correctly decode pipe2() flags in truss by not assuming full open()-like flags with O_RDONLY, etc. - Decode file flags passed to *chflags() as file flags (UF_* and SF_*) rather than as a file mode. - Fix decoding of quotactl() commands by splitting out the two command components instead of assuming the raw command value matches the primary command component. In addition, truss and kdump now build without triggering any warnings. All of the sysdecode manpages now include the required headers in the synopsis. Reviewed by: kib (several older versions), wblock (manpages) MFC after: 2 months Differential Revision: https://reviews.freebsd.org/D7847
2016-10-17 22:37:07 +00:00
print_mask_arg(sysdecode_access_mode, fp, args[sc->offset]);
Decode the arguments passed to the *at() family of system calls. This is especially useful now that libc's open() always calls openat(). While here, fix a few other things: - Decode the mode argument passed to access(), eaccess(), and faccessat(). - Decode the atfd paramete to pretty-print AT_FDCWD. - Decode the special AT_* flags used with some of the *at() system calls. - Decode arguments for fchmod(), lchmod(), fchown(), lchown(), eaccess(), and futimens(). - Decode both of the timeval structures passed to futimes() instead of just the first one.
2015-08-06 19:08:33 +00:00
break;
- Decode the arguments for several signal-related system calls: sigpending, sigqueue, sigreturn, sigsuspend, sigtimedwait, sigwait, sigwaitinfo, and thr_kill. - Print signal sets as a structure (with {}'s) and in particular use this to differentiate empty sets from a NULL pointer. - Decode arguments for some other system calls: issetugid, pipe2, sysarch (operations are only decoded for amd64 and i386), and thr_self.
2015-08-17 17:52:28 +00:00
case Sysarch:
Decode pathconf() names, *at() flags, and sysarch() numbers in libsysdecode. Move tables that were previously in truss over to libsysdecode. truss output is unchanged, but kdump has been updated to decode these fields. In addition, sysdecode_sysarch_number() should support all platforms whereas the old table in truss only supported x86.
2017-09-04 05:34:36 +00:00
print_integer_arg(sysdecode_sysarch_number, fp,
args[sc->offset]);
- Decode the arguments for several signal-related system calls: sigpending, sigqueue, sigreturn, sigsuspend, sigtimedwait, sigwait, sigwaitinfo, and thr_kill. - Print signal sets as a structure (with {}'s) and in particular use this to differentiate empty sets from a NULL pointer. - Decode arguments for some other system calls: issetugid, pipe2, sysarch (operations are only decoded for amd64 and i386), and thr_self.
2015-08-17 17:52:28 +00:00
break;
Several changes to truss. - Refactor the interface between the ABI-independent code and the ABI-specific backends. The backends now provide smaller hooks to fetch system call arguments and return values. The rest of the system call entry and exit handling that was previously duplicated among all the backends has been moved to one place. - Merge the loop when waiting for an event with the loop for handling stops. This also means not emulating a procfs-like interface on top of ptrace(). Instead, use a single event loop that fetches process events via waitid(). Among other things this allows us to report the full 32-bit exit value. - Use PT_FOLLOW_FORK to follow new child processes instead of forking a new truss process for each new child. This allows one truss process to monitor a tree of processes and truss -c should now display one total for the entire tree instead of separate summaries per process. - Use the recently added fields to ptrace_lwpinfo to determine the current system call number and argument count. The latter is especially useful and fixes a regression since the conversion from procfs. truss now generally prints the correct number of arguments for most system calls rather than printing extra arguments for any call not listed in the table in syscalls.c. - Actually check the new ABI when processes call exec. The comments claimed that this happened but it was not being done (perhaps this was another regression in the conversion to ptrace()). If the new ABI after exec is not supported, truss detaches from the process. If truss does not support the ABI for a newly executed process the process is killed before it returns from exec. - Along with the refactor, teach the various ABI-specific backends to fetch both return values, not just the first. Use this to properly report the full 64-bit return value from lseek(). In addition, the handler for "pipe" now pulls the pair of descriptors out of the return values (which is the true kernel system call interface) but displays them as an argument (which matches the interface exported by libc). - Each ABI handler adds entries to a linker set rather than requiring a statically defined table of handlers in main.c. - The arm and mips system call fetching code was changed to follow the same pattern as amd64 (and the in-kernel handler) of fetching register arguments first and then reading any remaining arguments from the stack. This should fix indirect system call arguments on at least arm. - The mipsn32 and n64 ABIs will now look for arguments in A4 through A7. - Use register %ebp for the 6th system call argument for Linux/i386 ABIs to match the in-kernel argument fetch code. - For powerpc binaries on a powerpc64 system, fetch the extra arguments on the stack as 32-bit values that are then copied into the 64-bit argument array instead of reading the 32-bit values directly into the 64-bit array. Reviewed by: kib (earlier version) Tested on: amd64 (FreeBSD/amd64 & i386), i386, arm (earlier version) Tested on: powerpc64 (FreeBSD/powerpc64 & powerpc) MFC after: 1 month Differential Revision: https://reviews.freebsd.org/D3575
2015-09-30 19:13:32 +00:00
case PipeFds:
/*
* The pipe() system call in the kernel returns its
* two file descriptors via return values. However,
* the interface exposed by libc is that pipe()
* accepts a pointer to an array of descriptors.
* Format the output to match the libc API by printing
* the returned file descriptors as a fake argument.
*
* Overwrite the first retval to signal a successful
* return as well.
*/
fprintf(fp, "{ %ld, %ld }", retval[0], retval[1]);
retval[0] = 0;
break;
truss: Add support for utrace(2). This uses the kdump(1) utrace support code directly until a common library is created. This allows malloc(3) tracing with MALLOC_CONF=utrace:true and rtld tracing with LD_UTRACE=1. Unknown utrace(2) data is just printed as hex. PR: 43819 [inspired by] Reviewed by: jhb MFC after: 2 weeks Relnotes: yes Differential Revision: https://reviews.freebsd.org/D3819
2015-10-06 21:58:38 +00:00
case Utrace: {
size_t len;
void *utrace_addr;
len = args[sc->offset + 1];
utrace_addr = calloc(1, len);
if (get_struct(pid, (void *)args[sc->offset],
(void *)utrace_addr, len) != -1)
print_utrace(fp, utrace_addr, len);
else
fprintf(fp, "0x%lx", args[sc->offset]);
free(utrace_addr);
break;
}
Properly format pointer size independent CloudABI system calls. CloudABI has approximately 50 system calls that do not depend on the pointer size of the system. As the ABI is pretty compact, it takes little effort to each truss(8) the formatting rules for these system calls. Start off by formatting pointer size independent system calls. Changes: - Make it possible to include the CloudABI system call definitions in FreeBSD userspace builds. Add ${root}/sys to the truss(8) Makefile so we can pull in <compat/cloudabi/cloudabi_syscalldefs.h>. - Refactoring: patch up amd64-cloudabi64.c to use the CLOUDABI_* constants instead of rolling our own table. - Add table entries for all of the system calls. - Add new generic formatting types (UInt, IntArray) that we'll be using to format unsigned integers and arrays of integers. - Add CloudABI specific formatting types. Approved by: jhb Differential Revision: https://reviews.freebsd.org/D3836
2015-10-08 05:27:45 +00:00
case IntArray: {
int descriptors[16];
unsigned long i, ndescriptors;
bool truncated;
ndescriptors = args[sc->offset + 1];
truncated = false;
if (ndescriptors > nitems(descriptors)) {
ndescriptors = nitems(descriptors);
truncated = true;
}
if (get_struct(pid, (void *)args[sc->offset],
descriptors, ndescriptors * sizeof(descriptors[0])) != -1) {
fprintf(fp, "{");
for (i = 0; i < ndescriptors; i++)
fprintf(fp, i == 0 ? " %d" : ", %d",
descriptors[i]);
fprintf(fp, truncated ? ", ... }" : " }");
} else
fprintf(fp, "0x%lx", args[sc->offset]);
break;
}
Move mksubr from kdump into libsysdecode. Restructure this script so that it generates a header of tables instead of a source file. The tables are included in a flags.c source file which provides functions to decode various system call arguments. For functions that decode an enumeration, the function returns a pointer to a string for known values and NULL for unknown values. For functions that do more complex decoding (typically of a bitmask), the function accepts a pointer to a FILE object (open_memstream() can be used as a string builder) to which decoded values are written. If the function operates on a bitmask, the function returns true if any bits were decoded or false if the entire value was valid. Additionally, the third argument accepts a pointer to a value to which any undecoded bits are stored. This pointer can be NULL if the caller doesn't care about remaining bits. Convert kdump over to using decoder functions from libsysdecode instead of mksubr. truss also uses decoders from libsysdecode instead of private lookup tables, though lookup tables for objects not decoded by kdump remain in truss for now. Eventually most of these tables should move into libsysdecode as the automated table generation approach from mksubr is less stale than the static tables in truss. Some changes have been made to truss and kdump output: - The flags passed to open() are now properly decoded in that one of O_RDONLY, O_RDWR, O_WRONLY, or O_EXEC is always included in a decoded mask. - Optional arguments to open(), openat(), and fcntl() are only printed in kdump if they exist (e.g. the mode is only printed for open() if O_CREAT is set in the flags). - Print argument to F_GETLK/SETLK/SETLKW in kdump as a pointer, not int. - Include all procctl() commands. - Correctly decode pipe2() flags in truss by not assuming full open()-like flags with O_RDONLY, etc. - Decode file flags passed to *chflags() as file flags (UF_* and SF_*) rather than as a file mode. - Fix decoding of quotactl() commands by splitting out the two command components instead of assuming the raw command value matches the primary command component. In addition, truss and kdump now build without triggering any warnings. All of the sysdecode manpages now include the required headers in the synopsis. Reviewed by: kib (several older versions), wblock (manpages) MFC after: 2 months Differential Revision: https://reviews.freebsd.org/D7847
2016-10-17 22:37:07 +00:00
case Pipe2:
print_mask_arg(sysdecode_pipe2_flags, fp, args[sc->offset]);
break;
Decode the arguments passed to cap_fcntls_get() and cap_fcntls_limit().
2017-03-18 18:10:02 +00:00
case CapFcntlRights: {
uint32_t rights;
if (sc->type & OUT) {
if (get_struct(pid, (void *)args[sc->offset], &rights,
sizeof(rights)) == -1) {
fprintf(fp, "0x%lx", args[sc->offset]);
break;
}
} else
rights = args[sc->offset];
print_mask_arg32(sysdecode_cap_fcntlrights, fp, rights);
break;
}
Decode arguments passed to posix_fadvise().
2017-03-18 18:12:09 +00:00
case Fadvice:
print_integer_arg(sysdecode_fadvice, fp, args[sc->offset]);
break;
Decode file flags passed to *chflags*(). While here, decode arguments passed to fchflags() and chflagsat().
2017-03-18 18:21:41 +00:00
case FileFlags: {
fflags_t rem;
if (!sysdecode_fileflags(fp, args[sc->offset], &rem))
fprintf(fp, "0x%x", rem);
else if (rem != 0)
fprintf(fp, "|0x%x", rem);
break;
}
Decode flock() operation.
2017-03-18 18:26:56 +00:00
case Flockop:
print_mask_arg(sysdecode_flock_operation, fp, args[sc->offset]);
break;
Decode arguments passed to getfsstat(). Note that this does not yet decode the statfs structures returned by getfsstat().
2017-03-18 18:31:45 +00:00
case Getfsstatmode:
print_integer_arg(sysdecode_getfsstat_mode, fp,
args[sc->offset]);
break;
Decode arguments passed to kldsym() and kldunloadf(). This does not currently decode the kld_sym_lookup structure passed to kldsym().
2017-03-18 19:59:21 +00:00
case Kldsymcmd:
print_integer_arg(sysdecode_kldsym_cmd, fp, args[sc->offset]);
break;
case Kldunloadflags:
print_integer_arg(sysdecode_kldunload_flags, fp,
args[sc->offset]);
break;
Decode arguments to madvise().
2017-03-19 00:31:21 +00:00
case Madvice:
print_integer_arg(sysdecode_madvice, fp, args[sc->offset]);
break;
Add Socklent for handling args of type socklen_t.
2017-05-03 09:20:36 +00:00
case Socklent:
fprintf(fp, "%u", (socklen_t)args[sc->offset]);
break;
Decode the third argument of socket().
2017-05-03 09:23:13 +00:00
case Sockprotocol: {
Improve the decoding of the third argument of the socket() call. Decoding of the third argument depends on the first one. For doing this, add a corresponding function to libsysdecode. Thanks to jhb@ for suggesting this.
2017-05-25 14:27:54 +00:00
const char *temp;
int domain, protocol;
Decode the third argument of socket().
2017-05-03 09:23:13 +00:00
Improve the decoding of the third argument of the socket() call. Decoding of the third argument depends on the first one. For doing this, add a corresponding function to libsysdecode. Thanks to jhb@ for suggesting this.
2017-05-25 14:27:54 +00:00
domain = args[sc->offset - 2];
Decode the third argument of socket().
2017-05-03 09:23:13 +00:00
protocol = args[sc->offset];
if (protocol == 0) {
fputs("0", fp);
} else {
Improve the decoding of the third argument of the socket() call. Decoding of the third argument depends on the first one. For doing this, add a corresponding function to libsysdecode. Thanks to jhb@ for suggesting this.
2017-05-25 14:27:54 +00:00
temp = sysdecode_socket_protocol(domain, protocol);
if (temp) {
fputs(temp, fp);
} else {
fprintf(fp, "%d", protocol);
}
Decode the third argument of socket().
2017-05-03 09:23:13 +00:00
}
break;
}
Add support for [gs]etsockopt().
2017-05-03 12:18:09 +00:00
case Sockoptlevel:
print_integer_arg(sysdecode_sockopt_level, fp,
args[sc->offset]);
break;
case Sockoptname: {
const char *temp;
int level, name;
level = args[sc->offset - 1];
name = args[sc->offset];
temp = sysdecode_sockopt_name(level, name);
if (temp) {
fputs(temp, fp);
} else {
fprintf(fp, "%d", name);
}
break;
}
Decode the fourth argument of sendto and recvfrom call.
2017-05-03 12:23:58 +00:00
case Msgflags:
print_mask_arg(sysdecode_msg_flags, fp, args[sc->offset]);
break;
Decode the arguments passed to __cap_rights_get() and cap_rights_limit(). Submitted by: tobik
2017-06-02 13:33:50 +00:00
case CapRights: {
cap_rights_t rights;
if (get_struct(pid, (void *)args[sc->offset], &rights,
sizeof(rights)) != -1) {
fputs("{ ", fp);
sysdecode_cap_rights(fp, &rights);
fputs(" }", fp);
} else
fprintf(fp, "0x%lx", args[sc->offset]);
break;
}
Decode arguments to ACL related system calls. This only decodes the raw arguments but not the contents of the struct acl objects.
2017-06-08 03:51:17 +00:00
case Acltype:
print_integer_arg(sysdecode_acltype, fp, args[sc->offset]);
break;
Decode arguments passed to extended attribute related system calls. The cmd argument passed to extattrctl() is not decoded as a string constant but is just printed in hex. The value is filesystem-specific but in practice is only used with UFS1 filesystems.
2017-06-08 04:31:15 +00:00
case Extattrnamespace:
print_integer_arg(sysdecode_extattrnamespace, fp,
args[sc->offset]);
break;
Decode arguments to minherit().
2017-06-08 04:45:13 +00:00
case Minherit:
print_integer_arg(sysdecode_minherit_inherit, fp,
args[sc->offset]);
break;
Decode arguments to mlock(), mlockall(), and munlock().
2017-06-08 04:50:50 +00:00
case Mlockall:
print_mask_arg(sysdecode_mlockall_flags, fp, args[sc->offset]);
break;
Decode flags passed to mount(), nmount(), and unmount().
2017-06-08 08:07:51 +00:00
case Mountflags:
print_mask_arg(sysdecode_mount_flags, fp, args[sc->offset]);
break;
Decode arguments passed to msync().
2017-06-08 08:10:57 +00:00
case Msync:
print_mask_arg(sysdecode_msync_flags, fp, args[sc->offset]);
break;
Decode arguments to getpriority() and setpriority().
2017-06-10 00:37:02 +00:00
case Priowhich:
print_integer_arg(sysdecode_prio_which, fp, args[sc->offset]);
break;
Decode the arguments to ptrace(). This does not decode structures returned by ptrace().
2017-06-10 00:45:07 +00:00
case Ptraceop:
print_integer_arg(sysdecode_ptrace_request, fp,
args[sc->offset]);
break;
Decode the arguments to quotactl().
2017-06-10 00:53:00 +00:00
case Quotactlcmd:
if (!sysdecode_quotactl_cmd(fp, args[sc->offset]))
fprintf(fp, "%#x", (int)args[sc->offset]);
break;
Decode the 'howto' argument to reboot().
2017-06-10 01:22:40 +00:00
case Reboothowto:
print_mask_arg(sysdecode_reboot_howto, fp, args[sc->offset]);
break;
Decode arguments to rtprio() and rtprio_thread().
2017-06-10 01:32:35 +00:00
case Rtpriofunc:
print_integer_arg(sysdecode_rtprio_function, fp,
args[sc->offset]);
break;
Decode arguments to sched_* family of system calls. This includes decoding both scheduler policy constants and the sched_param structure for sched_get_priority_max(), sched_get_priority_min(), sched_getparam(), sched_getscheduler(), sched_rr_get_interval(), sched_setparam(), and sched_setscheduler().
2017-06-16 20:03:09 +00:00
case Schedpolicy:
print_integer_arg(sysdecode_scheduler_policy, fp,
args[sc->offset]);
break;
case Schedparam: {
struct sched_param sp;
if (get_struct(pid, (void *)args[sc->offset], &sp,
sizeof(sp)) != -1)
fprintf(fp, "{ %d }", sp.sched_priority);
else
fprintf(fp, "0x%lx", args[sc->offset]);
break;
}
Decode signal information returned by system calls. Specifically, decode the siginfo structure returned by sigtimedwait(), sigwaitinfo(), and wait6(). While here, also decode the signal number returned in the second argument to sigwait().
2017-08-30 15:45:23 +00:00
case PSig: {
int sig;
if (get_struct(pid, (void *)args[sc->offset], &sig,
sizeof(sig)) == 0)
fprintf(fp, "{ %s }", strsig2(sig));
else
fprintf(fp, "0x%lx", args[sc->offset]);
break;
}
case Siginfo: {
siginfo_t si;
if (get_struct(pid, (void *)args[sc->offset], &si,
sizeof(si)) != -1) {
fprintf(fp, "{ signo=%s", strsig2(si.si_signo));
decode_siginfo(fp, &si);
fprintf(fp, " }");
} else
fprintf(fp, "0x%lx", args[sc->offset]);
break;
}
Decode msghdr argument of sendmsg() and recvmsg(). Sponsored by: Netflix, Inc.
2018-01-15 20:39:42 +00:00
case Iovec:
Add support for readv() and writev() to truss. Sponsored by: Netflix, Inc.
2018-01-13 13:59:35 +00:00
/*
* Print argument as an array of struct iovec, where the next
* syscall argument is the number of elements of the array.
*/
Decode msghdr argument of sendmsg() and recvmsg(). Sponsored by: Netflix, Inc.
2018-01-15 20:39:42 +00:00
print_iovec(fp, trussinfo, (void *)args[sc->offset],
(int)args[sc->offset + 1]);
break;
case Sctpsndrcvinfo: {
struct sctp_sndrcvinfo info;
Add support for readv() and writev() to truss. Sponsored by: Netflix, Inc.
2018-01-13 13:59:35 +00:00
if (get_struct(pid, (void *)args[sc->offset],
Decode msghdr argument of sendmsg() and recvmsg(). Sponsored by: Netflix, Inc.
2018-01-15 20:39:42 +00:00
&info, sizeof(struct sctp_sndrcvinfo)) == -1) {
Add support for readv() and writev() to truss. Sponsored by: Netflix, Inc.
2018-01-13 13:59:35 +00:00
fprintf(fp, "0x%lx", args[sc->offset]);
break;
}
Decode msghdr argument of sendmsg() and recvmsg(). Sponsored by: Netflix, Inc.
2018-01-15 20:39:42 +00:00
print_sctp_sndrcvinfo(fp, sc->type & OUT, &info);
Add support for readv() and writev() to truss. Sponsored by: Netflix, Inc.
2018-01-13 13:59:35 +00:00
break;
}
Decode msghdr argument of sendmsg() and recvmsg(). Sponsored by: Netflix, Inc.
2018-01-15 20:39:42 +00:00
case Msghdr: {
struct msghdr msghdr;
Improve support for sctp_generic_recvmsg() and sctp_generic_sendmsg() and add support for sctp_generic_sendmsg_iov(). Handle the struct iovec argument and the struct sctp_sndrcvinfo arguments.
2018-01-14 14:33:22 +00:00
if (get_struct(pid, (void *)args[sc->offset],
Decode msghdr argument of sendmsg() and recvmsg(). Sponsored by: Netflix, Inc.
2018-01-15 20:39:42 +00:00
&msghdr, sizeof(struct msghdr)) == -1) {
Improve support for sctp_generic_recvmsg() and sctp_generic_sendmsg() and add support for sctp_generic_sendmsg_iov(). Handle the struct iovec argument and the struct sctp_sndrcvinfo arguments.
2018-01-14 14:33:22 +00:00
fprintf(fp, "0x%lx", args[sc->offset]);
break;
}
Decode msghdr argument of sendmsg() and recvmsg(). Sponsored by: Netflix, Inc.
2018-01-15 20:39:42 +00:00
fputs("{", fp);
print_sockaddr(fp, trussinfo, msghdr.msg_name, msghdr.msg_namelen);
fprintf(fp, ",%d,", msghdr.msg_namelen);
print_iovec(fp, trussinfo, msghdr.msg_iov, msghdr.msg_iovlen);
fprintf(fp, ",%d,", msghdr.msg_iovlen);
print_cmsgs(fp, pid, sc->type & OUT, &msghdr);
fprintf(fp, ",%u,", msghdr.msg_controllen);
print_mask_arg(sysdecode_msg_flags, fp, msghdr.msg_flags);
fputs("}", fp);
Improve support for sctp_generic_recvmsg() and sctp_generic_sendmsg() and add support for sctp_generic_sendmsg_iov(). Handle the struct iovec argument and the struct sctp_sndrcvinfo arguments.
2018-01-14 14:33:22 +00:00
break;
}
Properly format pointer size independent CloudABI system calls. CloudABI has approximately 50 system calls that do not depend on the pointer size of the system. As the ABI is pretty compact, it takes little effort to each truss(8) the formatting rules for these system calls. Start off by formatting pointer size independent system calls. Changes: - Make it possible to include the CloudABI system call definitions in FreeBSD userspace builds. Add ${root}/sys to the truss(8) Makefile so we can pull in <compat/cloudabi/cloudabi_syscalldefs.h>. - Refactoring: patch up amd64-cloudabi64.c to use the CLOUDABI_* constants instead of rolling our own table. - Add table entries for all of the system calls. - Add new generic formatting types (UInt, IntArray) that we'll be using to format unsigned integers and arrays of integers. - Add CloudABI specific formatting types. Approved by: jhb Differential Revision: https://reviews.freebsd.org/D3836
2015-10-08 05:27:45 +00:00
case CloudABIAdvice:
fputs(xlookup(cloudabi_advice, args[sc->offset]), fp);
break;
case CloudABIClockID:
fputs(xlookup(cloudabi_clockid, args[sc->offset]), fp);
break;
Rename ClouduABIFDSFlags to CloudABIFDSFlags.
2018-03-29 00:04:07 +00:00
case CloudABIFDSFlags:
Properly format pointer size independent CloudABI system calls. CloudABI has approximately 50 system calls that do not depend on the pointer size of the system. As the ABI is pretty compact, it takes little effort to each truss(8) the formatting rules for these system calls. Start off by formatting pointer size independent system calls. Changes: - Make it possible to include the CloudABI system call definitions in FreeBSD userspace builds. Add ${root}/sys to the truss(8) Makefile so we can pull in <compat/cloudabi/cloudabi_syscalldefs.h>. - Refactoring: patch up amd64-cloudabi64.c to use the CLOUDABI_* constants instead of rolling our own table. - Add table entries for all of the system calls. - Add new generic formatting types (UInt, IntArray) that we'll be using to format unsigned integers and arrays of integers. - Add CloudABI specific formatting types. Approved by: jhb Differential Revision: https://reviews.freebsd.org/D3836
2015-10-08 05:27:45 +00:00
fputs(xlookup_bits(cloudabi_fdsflags, args[sc->offset]), fp);
break;
case CloudABIFDStat: {
cloudabi_fdstat_t fds;
if (get_struct(pid, (void *)args[sc->offset], &fds, sizeof(fds))
!= -1) {
fprintf(fp, "{ %s, ",
xlookup(cloudabi_filetype, fds.fs_filetype));
fprintf(fp, "%s, ... }",
xlookup_bits(cloudabi_fdflags, fds.fs_flags));
} else
fprintf(fp, "0x%lx", args[sc->offset]);
break;
}
case CloudABIFileStat: {
cloudabi_filestat_t fsb;
if (get_struct(pid, (void *)args[sc->offset], &fsb, sizeof(fsb))
!= -1)
Fix printf format warning.
2016-10-11 17:11:17 +00:00
fprintf(fp, "{ %s, %ju }",
Properly format pointer size independent CloudABI system calls. CloudABI has approximately 50 system calls that do not depend on the pointer size of the system. As the ABI is pretty compact, it takes little effort to each truss(8) the formatting rules for these system calls. Start off by formatting pointer size independent system calls. Changes: - Make it possible to include the CloudABI system call definitions in FreeBSD userspace builds. Add ${root}/sys to the truss(8) Makefile so we can pull in <compat/cloudabi/cloudabi_syscalldefs.h>. - Refactoring: patch up amd64-cloudabi64.c to use the CLOUDABI_* constants instead of rolling our own table. - Add table entries for all of the system calls. - Add new generic formatting types (UInt, IntArray) that we'll be using to format unsigned integers and arrays of integers. - Add CloudABI specific formatting types. Approved by: jhb Differential Revision: https://reviews.freebsd.org/D3836
2015-10-08 05:27:45 +00:00
xlookup(cloudabi_filetype, fsb.st_filetype),
Fix printf format warning.
2016-10-11 17:11:17 +00:00
(uintmax_t)fsb.st_size);
Properly format pointer size independent CloudABI system calls. CloudABI has approximately 50 system calls that do not depend on the pointer size of the system. As the ABI is pretty compact, it takes little effort to each truss(8) the formatting rules for these system calls. Start off by formatting pointer size independent system calls. Changes: - Make it possible to include the CloudABI system call definitions in FreeBSD userspace builds. Add ${root}/sys to the truss(8) Makefile so we can pull in <compat/cloudabi/cloudabi_syscalldefs.h>. - Refactoring: patch up amd64-cloudabi64.c to use the CLOUDABI_* constants instead of rolling our own table. - Add table entries for all of the system calls. - Add new generic formatting types (UInt, IntArray) that we'll be using to format unsigned integers and arrays of integers. - Add CloudABI specific formatting types. Approved by: jhb Differential Revision: https://reviews.freebsd.org/D3836
2015-10-08 05:27:45 +00:00
else
fprintf(fp, "0x%lx", args[sc->offset]);
break;
}
case CloudABIFileType:
fputs(xlookup(cloudabi_filetype, args[sc->offset]), fp);
break;
case CloudABIFSFlags:
fputs(xlookup_bits(cloudabi_fsflags, args[sc->offset]), fp);
break;
case CloudABILookup:
if ((args[sc->offset] & CLOUDABI_LOOKUP_SYMLINK_FOLLOW) != 0)
fprintf(fp, "%d|LOOKUP_SYMLINK_FOLLOW",
(int)args[sc->offset]);
else
fprintf(fp, "%d", (int)args[sc->offset]);
break;
case CloudABIMFlags:
fputs(xlookup_bits(cloudabi_mflags, args[sc->offset]), fp);
break;
case CloudABIMProt:
fputs(xlookup_bits(cloudabi_mprot, args[sc->offset]), fp);
break;
case CloudABIMSFlags:
fputs(xlookup_bits(cloudabi_msflags, args[sc->offset]), fp);
break;
case CloudABIOFlags:
fputs(xlookup_bits(cloudabi_oflags, args[sc->offset]), fp);
break;
case CloudABISDFlags:
fputs(xlookup_bits(cloudabi_sdflags, args[sc->offset]), fp);
break;
case CloudABISignal:
fputs(xlookup(cloudabi_signal, args[sc->offset]), fp);
break;
case CloudABITimestamp:
fprintf(fp, "%lu.%09lus", args[sc->offset] / 1000000000,
args[sc->offset] % 1000000000);
break;
case CloudABIULFlags:
fputs(xlookup_bits(cloudabi_ulflags, args[sc->offset]), fp);
break;
case CloudABIWhence:
fputs(xlookup(cloudabi_whence, args[sc->offset]), fp);
break;
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
default:
errx(1, "Invalid argument type %d\n", sc->type & ARG_MASK);
}
Change the argument formatting function to use a stdio FILE object opened with open_memstream() to build the string for each argument. This allows for more complicated argument building without resorting to intermediate malloc's, etc. Related, the strsig*() functions no longer return allocated strings but use a static global buffer instead.
2015-08-19 00:49:50 +00:00
fclose(fp);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
return (tmp);
Truss program. Requires procfs.
1997-12-06 05:23:12 +00:00
}
/*
Fetch the current thread and it's syscall state from the trussinfo object instead of passing some of that state as arguments to print_syscall() and print_syscallret(). This just makes the calls of these functions shorter and easier to read.
2016-02-16 22:00:01 +00:00
* Print (to outfile) the system call and its arguments.
Truss program. Requires procfs.
1997-12-06 05:23:12 +00:00
*/
void
Fetch the current thread and it's syscall state from the trussinfo object instead of passing some of that state as arguments to print_syscall() and print_syscallret(). This just makes the calls of these functions shorter and easier to read.
2016-02-16 22:00:01 +00:00
print_syscall(struct trussinfo *trussinfo)
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
{
Fetch the current thread and it's syscall state from the trussinfo object instead of passing some of that state as arguments to print_syscall() and print_syscallret(). This just makes the calls of these functions shorter and easier to read.
2016-02-16 22:00:01 +00:00
struct threadinfo *t;
const char *name;
char **s_args;
int i, len, nargs;
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
Fetch the current thread and it's syscall state from the trussinfo object instead of passing some of that state as arguments to print_syscall() and print_syscallret(). This just makes the calls of these functions shorter and easier to read.
2016-02-16 22:00:01 +00:00
t = trussinfo->curthread;
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
Rework syscall structure lookups. Avoid always using an O(n^2) loop over known syscall structures with strcmp() on each system call. Instead, use a per-ABI cache indexed by the system call number. The first 1024 system calls (which should cover all of the normal system calls in currently-supported ABIs) use a flat array indexed by the system call number to find system call structure. For other system calls, a linked list of structures storing an integer to structure mapping is stored in the ABI. The linked list isn't very smart, but it should only be used by buggy applications invoking unknown system calls. This also fixes handling of unknown system calls which currently trigger a NULL pointer dereference. Reviewed by: kib MFC after: 2 weeks
2016-12-06 00:39:00 +00:00
name = t->cs.sc->name;
Fetch the current thread and it's syscall state from the trussinfo object instead of passing some of that state as arguments to print_syscall() and print_syscallret(). This just makes the calls of these functions shorter and easier to read.
2016-02-16 22:00:01 +00:00
nargs = t->cs.nargs;
s_args = t->cs.s_args;
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
Add support for displaying thread IDs to truss(1). - Consolidate duplicate code for printing the metadata at the start of each line into a shared function. - Add an -H option which will log the thread ID of the relevant thread for each event. While here, remove some extraneous calls to clock_gettime() in print_syscall() and print_syscall_ret(). The caller of print_syscall_ret() always updates the current thread's "after" time before it is called. Reviewed by: kib MFC after: 1 month Differential Revision: https://reviews.freebsd.org/D5363
2016-02-23 19:56:29 +00:00
len = print_line_prefix(trussinfo);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
len += fprintf(trussinfo->outfile, "%s(", name);
for (i = 0; i < nargs; i++) {
Fetch the current thread and it's syscall state from the trussinfo object instead of passing some of that state as arguments to print_syscall() and print_syscallret(). This just makes the calls of these functions shorter and easier to read.
2016-02-16 22:00:01 +00:00
if (s_args[i] != NULL)
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
len += fprintf(trussinfo->outfile, "%s", s_args[i]);
else
- Style(9) cleanup. Approved by: kib (mentor)
2012-09-02 11:03:18 +00:00
len += fprintf(trussinfo->outfile,
"<missing argument>");
len += fprintf(trussinfo->outfile, "%s", i < (nargs - 1) ?
"," : "");
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
}
len += fprintf(trussinfo->outfile, ")");
for (i = 0; i < 6 - (len / 8); i++)
fprintf(trussinfo->outfile, "\t");
Change the output of truss to more closely resemble SysV's. (Yes, it really is much nicer looking.) Submitted by: "Matthew N. Dodd" <winter@jurai.net>
2000-03-18 08:49:41 +00:00
}
void
Fetch the current thread and it's syscall state from the trussinfo object instead of passing some of that state as arguments to print_syscall() and print_syscallret(). This just makes the calls of these functions shorter and easier to read.
2016-02-16 22:00:01 +00:00
print_syscall_ret(struct trussinfo *trussinfo, int errorp, long *retval)
Port truss(1) to 64-bit architectures: o Syscall return values do not fit in int on 64-bit architectures. Change the type of retval in <arch>_syscall_exit() to long and change the prototype of said function to return a long as well. o Change the prototype of print_syscall_ret() to take a long for the return address and change the format string accordingly. o Replace the code sequence tmp = malloc(X); sprintf(tmp, format, ...); with X by definition too small on 64-bit platforms by asprintf(&tmp, format, ...); With these changes the output makes sense again, although it does mess up the tabulation on ia64. Go widescreen... Not tested on: alpha, sparc64.
2003-11-09 03:48:13 +00:00
{
Add -c option to summarize number of calls, errors, and system time. Reviewed by: alfred
2009-05-12 20:42:12 +00:00
struct timespec timediff;
Fetch the current thread and it's syscall state from the trussinfo object instead of passing some of that state as arguments to print_syscall() and print_syscallret(). This just makes the calls of these functions shorter and easier to read.
2016-02-16 22:00:01 +00:00
struct threadinfo *t;
struct syscall *sc;
Add handling for non-native error values to libsysdecode. Add two new functions, sysdecode_abi_to_freebsd_errno() and sysdecode_freebsd_to_abi_errno(), which convert errno values between the native FreeBSD ABI and other supported ABIs. Note that the mappings are not necessarily perfect meaning in some cases multiple errors in one ABI might map to a single error in another ABI. In that case, the reverse mapping will return one of the errors that maps, but which error is non-deterministic. Change truss to always report the raw error value to the user but use libsysdecode to map it to a native errno value that can be used with strerror() to generate a description. Previously truss reported the "converted" error value. Now the user will always see the exact error value that the application sees. Change kdump to report the truly raw error value to the user. Previously kdump would report the absolute value of the raw error value (so for Linux binaries it didn't output the FreeBSD error value, but the positive value of the Linux error). Now it reports the real (i.e. negative) error value for Linux binaries. Also, use libsysdecode to convert the native FreeBSD error reported in the ktrace record to the raw error used by the ABI. This means that the Linux ABI can now be handled directly in ktrsysret() and removes the need for linux_ktrsysret(). Reviewed by: bdrewery, kib Helpful notes: wblock (manpage) Differential Revision: https://reviews.freebsd.org/D5314
2016-02-23 20:00:55 +00:00
int error;
Add -c option to summarize number of calls, errors, and system time. Reviewed by: alfred
2009-05-12 20:42:12 +00:00
Fetch the current thread and it's syscall state from the trussinfo object instead of passing some of that state as arguments to print_syscall() and print_syscallret(). This just makes the calls of these functions shorter and easier to read.
2016-02-16 22:00:01 +00:00
t = trussinfo->curthread;
sc = t->cs.sc;
Add -c option to summarize number of calls, errors, and system time. Reviewed by: alfred
2009-05-12 20:42:12 +00:00
if (trussinfo->flags & COUNTONLY) {
Fetch the current thread and it's syscall state from the trussinfo object instead of passing some of that state as arguments to print_syscall() and print_syscallret(). This just makes the calls of these functions shorter and easier to read.
2016-02-16 22:00:01 +00:00
timespecsubt(&t->after, &t->before, &timediff);
Revert r247300 for now. I'll post a new changeset for review.
2013-02-26 19:46:59 +00:00
timespecadd(&sc->time, &timediff, &sc->time);
Add -c option to summarize number of calls, errors, and system time. Reviewed by: alfred
2009-05-12 20:42:12 +00:00
sc->ncalls++;
if (errorp)
sc->nerror++;
return;
}
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
Fetch the current thread and it's syscall state from the trussinfo object instead of passing some of that state as arguments to print_syscall() and print_syscallret(). This just makes the calls of these functions shorter and easier to read.
2016-02-16 22:00:01 +00:00
print_syscall(trussinfo);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
fflush(trussinfo->outfile);
Use ptrace(2) LWP events to track threads reliably in truss. - truss can now log the system call invoked by a thread during a voluntary process exit. No return value is logged, but the value passed to exit() is included in the trace output. Arguments passed to thread exit system calls such as thr_exit() are not logged as voluntary thread exits cannot be distinguished from involuntary thread exits during a system call. - New events are now reported for thread births and exits similar to the recently added events for new child processes when following forks. Reviewed by: kib Differential Revision: https://reviews.freebsd.org/D5561
2016-03-09 18:45:41 +00:00
if (retval == NULL) {
/*
* This system call resulted in the current thread's exit,
* so there is no return value or error to display.
*/
fprintf(trussinfo->outfile, "\n");
return;
}
Add handling for non-native error values to libsysdecode. Add two new functions, sysdecode_abi_to_freebsd_errno() and sysdecode_freebsd_to_abi_errno(), which convert errno values between the native FreeBSD ABI and other supported ABIs. Note that the mappings are not necessarily perfect meaning in some cases multiple errors in one ABI might map to a single error in another ABI. In that case, the reverse mapping will return one of the errors that maps, but which error is non-deterministic. Change truss to always report the raw error value to the user but use libsysdecode to map it to a native errno value that can be used with strerror() to generate a description. Previously truss reported the "converted" error value. Now the user will always see the exact error value that the application sees. Change kdump to report the truly raw error value to the user. Previously kdump would report the absolute value of the raw error value (so for Linux binaries it didn't output the FreeBSD error value, but the positive value of the Linux error). Now it reports the real (i.e. negative) error value for Linux binaries. Also, use libsysdecode to convert the native FreeBSD error reported in the ktrace record to the raw error used by the ABI. This means that the Linux ABI can now be handled directly in ktrsysret() and removes the need for linux_ktrsysret(). Reviewed by: bdrewery, kib Helpful notes: wblock (manpage) Differential Revision: https://reviews.freebsd.org/D5314
2016-02-23 20:00:55 +00:00
if (errorp) {
error = sysdecode_abi_to_freebsd_errno(t->proc->abi->abi,
retval[0]);
Several changes to truss. - Refactor the interface between the ABI-independent code and the ABI-specific backends. The backends now provide smaller hooks to fetch system call arguments and return values. The rest of the system call entry and exit handling that was previously duplicated among all the backends has been moved to one place. - Merge the loop when waiting for an event with the loop for handling stops. This also means not emulating a procfs-like interface on top of ptrace(). Instead, use a single event loop that fetches process events via waitid(). Among other things this allows us to report the full 32-bit exit value. - Use PT_FOLLOW_FORK to follow new child processes instead of forking a new truss process for each new child. This allows one truss process to monitor a tree of processes and truss -c should now display one total for the entire tree instead of separate summaries per process. - Use the recently added fields to ptrace_lwpinfo to determine the current system call number and argument count. The latter is especially useful and fixes a regression since the conversion from procfs. truss now generally prints the correct number of arguments for most system calls rather than printing extra arguments for any call not listed in the table in syscalls.c. - Actually check the new ABI when processes call exec. The comments claimed that this happened but it was not being done (perhaps this was another regression in the conversion to ptrace()). If the new ABI after exec is not supported, truss detaches from the process. If truss does not support the ABI for a newly executed process the process is killed before it returns from exec. - Along with the refactor, teach the various ABI-specific backends to fetch both return values, not just the first. Use this to properly report the full 64-bit return value from lseek(). In addition, the handler for "pipe" now pulls the pair of descriptors out of the return values (which is the true kernel system call interface) but displays them as an argument (which matches the interface exported by libc). - Each ABI handler adds entries to a linker set rather than requiring a statically defined table of handlers in main.c. - The arm and mips system call fetching code was changed to follow the same pattern as amd64 (and the in-kernel handler) of fetching register arguments first and then reading any remaining arguments from the stack. This should fix indirect system call arguments on at least arm. - The mipsn32 and n64 ABIs will now look for arguments in A4 through A7. - Use register %ebp for the 6th system call argument for Linux/i386 ABIs to match the in-kernel argument fetch code. - For powerpc binaries on a powerpc64 system, fetch the extra arguments on the stack as 32-bit values that are then copied into the 64-bit argument array instead of reading the 32-bit values directly into the 64-bit array. Reviewed by: kib (earlier version) Tested on: amd64 (FreeBSD/amd64 & i386), i386, arm (earlier version) Tested on: powerpc64 (FreeBSD/powerpc64 & powerpc) MFC after: 1 month Differential Revision: https://reviews.freebsd.org/D3575
2015-09-30 19:13:32 +00:00
fprintf(trussinfo->outfile, " ERR#%ld '%s'\n", retval[0],
Add handling for non-native error values to libsysdecode. Add two new functions, sysdecode_abi_to_freebsd_errno() and sysdecode_freebsd_to_abi_errno(), which convert errno values between the native FreeBSD ABI and other supported ABIs. Note that the mappings are not necessarily perfect meaning in some cases multiple errors in one ABI might map to a single error in another ABI. In that case, the reverse mapping will return one of the errors that maps, but which error is non-deterministic. Change truss to always report the raw error value to the user but use libsysdecode to map it to a native errno value that can be used with strerror() to generate a description. Previously truss reported the "converted" error value. Now the user will always see the exact error value that the application sees. Change kdump to report the truly raw error value to the user. Previously kdump would report the absolute value of the raw error value (so for Linux binaries it didn't output the FreeBSD error value, but the positive value of the Linux error). Now it reports the real (i.e. negative) error value for Linux binaries. Also, use libsysdecode to convert the native FreeBSD error reported in the ktrace record to the raw error used by the ABI. This means that the Linux ABI can now be handled directly in ktrsysret() and removes the need for linux_ktrsysret(). Reviewed by: bdrewery, kib Helpful notes: wblock (manpage) Differential Revision: https://reviews.freebsd.org/D5314
2016-02-23 20:00:55 +00:00
error == INT_MAX ? "Unknown error" : strerror(error));
}
Several changes to truss. - Refactor the interface between the ABI-independent code and the ABI-specific backends. The backends now provide smaller hooks to fetch system call arguments and return values. The rest of the system call entry and exit handling that was previously duplicated among all the backends has been moved to one place. - Merge the loop when waiting for an event with the loop for handling stops. This also means not emulating a procfs-like interface on top of ptrace(). Instead, use a single event loop that fetches process events via waitid(). Among other things this allows us to report the full 32-bit exit value. - Use PT_FOLLOW_FORK to follow new child processes instead of forking a new truss process for each new child. This allows one truss process to monitor a tree of processes and truss -c should now display one total for the entire tree instead of separate summaries per process. - Use the recently added fields to ptrace_lwpinfo to determine the current system call number and argument count. The latter is especially useful and fixes a regression since the conversion from procfs. truss now generally prints the correct number of arguments for most system calls rather than printing extra arguments for any call not listed in the table in syscalls.c. - Actually check the new ABI when processes call exec. The comments claimed that this happened but it was not being done (perhaps this was another regression in the conversion to ptrace()). If the new ABI after exec is not supported, truss detaches from the process. If truss does not support the ABI for a newly executed process the process is killed before it returns from exec. - Along with the refactor, teach the various ABI-specific backends to fetch both return values, not just the first. Use this to properly report the full 64-bit return value from lseek(). In addition, the handler for "pipe" now pulls the pair of descriptors out of the return values (which is the true kernel system call interface) but displays them as an argument (which matches the interface exported by libc). - Each ABI handler adds entries to a linker set rather than requiring a statically defined table of handlers in main.c. - The arm and mips system call fetching code was changed to follow the same pattern as amd64 (and the in-kernel handler) of fetching register arguments first and then reading any remaining arguments from the stack. This should fix indirect system call arguments on at least arm. - The mipsn32 and n64 ABIs will now look for arguments in A4 through A7. - Use register %ebp for the 6th system call argument for Linux/i386 ABIs to match the in-kernel argument fetch code. - For powerpc binaries on a powerpc64 system, fetch the extra arguments on the stack as 32-bit values that are then copied into the 64-bit argument array instead of reading the 32-bit values directly into the 64-bit array. Reviewed by: kib (earlier version) Tested on: amd64 (FreeBSD/amd64 & i386), i386, arm (earlier version) Tested on: powerpc64 (FreeBSD/powerpc64 & powerpc) MFC after: 1 month Differential Revision: https://reviews.freebsd.org/D3575
2015-09-30 19:13:32 +00:00
#ifndef __LP64__
Fix tracking of unknown syscalls for 'truss -c'. This is done by changing get_syscall() to either lookup the known syscall or add it into the list with the default handlers for printing. This also simplifies some code to not have to check if the syscall variable is set or NULL. Reviewed by: jhb Relnotes: yes Differential Revision: https://reviews.freebsd.org/D3792
2015-10-05 18:08:35 +00:00
else if (sc->ret_type == 2) {
Several changes to truss. - Refactor the interface between the ABI-independent code and the ABI-specific backends. The backends now provide smaller hooks to fetch system call arguments and return values. The rest of the system call entry and exit handling that was previously duplicated among all the backends has been moved to one place. - Merge the loop when waiting for an event with the loop for handling stops. This also means not emulating a procfs-like interface on top of ptrace(). Instead, use a single event loop that fetches process events via waitid(). Among other things this allows us to report the full 32-bit exit value. - Use PT_FOLLOW_FORK to follow new child processes instead of forking a new truss process for each new child. This allows one truss process to monitor a tree of processes and truss -c should now display one total for the entire tree instead of separate summaries per process. - Use the recently added fields to ptrace_lwpinfo to determine the current system call number and argument count. The latter is especially useful and fixes a regression since the conversion from procfs. truss now generally prints the correct number of arguments for most system calls rather than printing extra arguments for any call not listed in the table in syscalls.c. - Actually check the new ABI when processes call exec. The comments claimed that this happened but it was not being done (perhaps this was another regression in the conversion to ptrace()). If the new ABI after exec is not supported, truss detaches from the process. If truss does not support the ABI for a newly executed process the process is killed before it returns from exec. - Along with the refactor, teach the various ABI-specific backends to fetch both return values, not just the first. Use this to properly report the full 64-bit return value from lseek(). In addition, the handler for "pipe" now pulls the pair of descriptors out of the return values (which is the true kernel system call interface) but displays them as an argument (which matches the interface exported by libc). - Each ABI handler adds entries to a linker set rather than requiring a statically defined table of handlers in main.c. - The arm and mips system call fetching code was changed to follow the same pattern as amd64 (and the in-kernel handler) of fetching register arguments first and then reading any remaining arguments from the stack. This should fix indirect system call arguments on at least arm. - The mipsn32 and n64 ABIs will now look for arguments in A4 through A7. - Use register %ebp for the 6th system call argument for Linux/i386 ABIs to match the in-kernel argument fetch code. - For powerpc binaries on a powerpc64 system, fetch the extra arguments on the stack as 32-bit values that are then copied into the 64-bit argument array instead of reading the 32-bit values directly into the 64-bit array. Reviewed by: kib (earlier version) Tested on: amd64 (FreeBSD/amd64 & i386), i386, arm (earlier version) Tested on: powerpc64 (FreeBSD/powerpc64 & powerpc) MFC after: 1 month Differential Revision: https://reviews.freebsd.org/D3575
2015-09-30 19:13:32 +00:00
off_t off;
#if _BYTE_ORDER == _LITTLE_ENDIAN
off = (off_t)retval[1] << 32 | retval[0];
#else
off = (off_t)retval[0] << 32 | retval[1];
#endif
fprintf(trussinfo->outfile, " = %jd (0x%jx)\n", (intmax_t)off,
(intmax_t)off);
Try to make this code slightly less painful to read.
2008-07-31 17:15:21 +00:00
}
Several changes to truss. - Refactor the interface between the ABI-independent code and the ABI-specific backends. The backends now provide smaller hooks to fetch system call arguments and return values. The rest of the system call entry and exit handling that was previously duplicated among all the backends has been moved to one place. - Merge the loop when waiting for an event with the loop for handling stops. This also means not emulating a procfs-like interface on top of ptrace(). Instead, use a single event loop that fetches process events via waitid(). Among other things this allows us to report the full 32-bit exit value. - Use PT_FOLLOW_FORK to follow new child processes instead of forking a new truss process for each new child. This allows one truss process to monitor a tree of processes and truss -c should now display one total for the entire tree instead of separate summaries per process. - Use the recently added fields to ptrace_lwpinfo to determine the current system call number and argument count. The latter is especially useful and fixes a regression since the conversion from procfs. truss now generally prints the correct number of arguments for most system calls rather than printing extra arguments for any call not listed in the table in syscalls.c. - Actually check the new ABI when processes call exec. The comments claimed that this happened but it was not being done (perhaps this was another regression in the conversion to ptrace()). If the new ABI after exec is not supported, truss detaches from the process. If truss does not support the ABI for a newly executed process the process is killed before it returns from exec. - Along with the refactor, teach the various ABI-specific backends to fetch both return values, not just the first. Use this to properly report the full 64-bit return value from lseek(). In addition, the handler for "pipe" now pulls the pair of descriptors out of the return values (which is the true kernel system call interface) but displays them as an argument (which matches the interface exported by libc). - Each ABI handler adds entries to a linker set rather than requiring a statically defined table of handlers in main.c. - The arm and mips system call fetching code was changed to follow the same pattern as amd64 (and the in-kernel handler) of fetching register arguments first and then reading any remaining arguments from the stack. This should fix indirect system call arguments on at least arm. - The mipsn32 and n64 ABIs will now look for arguments in A4 through A7. - Use register %ebp for the 6th system call argument for Linux/i386 ABIs to match the in-kernel argument fetch code. - For powerpc binaries on a powerpc64 system, fetch the extra arguments on the stack as 32-bit values that are then copied into the 64-bit argument array instead of reading the 32-bit values directly into the 64-bit array. Reviewed by: kib (earlier version) Tested on: amd64 (FreeBSD/amd64 & i386), i386, arm (earlier version) Tested on: powerpc64 (FreeBSD/powerpc64 & powerpc) MFC after: 1 month Differential Revision: https://reviews.freebsd.org/D3575
2015-09-30 19:13:32 +00:00
#endif
else
fprintf(trussinfo->outfile, " = %ld (0x%lx)\n", retval[0],
retval[0]);
Truss program. Requires procfs.
1997-12-06 05:23:12 +00:00
}
Add -c option to summarize number of calls, errors, and system time. Reviewed by: alfred
2009-05-12 20:42:12 +00:00
void
print_summary(struct trussinfo *trussinfo)
{
struct timespec total = {0, 0};
- Style(9) cleanup. Approved by: kib (mentor)
2012-09-02 11:03:18 +00:00
struct syscall *sc;
Add -c option to summarize number of calls, errors, and system time. Reviewed by: alfred
2009-05-12 20:42:12 +00:00
int ncall, nerror;
fprintf(trussinfo->outfile, "%-20s%15s%8s%8s\n",
- Style(9) cleanup. Approved by: kib (mentor)
2012-09-02 11:03:18 +00:00
"syscall", "seconds", "calls", "errors");
Add -c option to summarize number of calls, errors, and system time. Reviewed by: alfred
2009-05-12 20:42:12 +00:00
ncall = nerror = 0;
Fix tracking of unknown syscalls for 'truss -c'. This is done by changing get_syscall() to either lookup the known syscall or add it into the list with the default handlers for printing. This also simplifies some code to not have to check if the syscall variable is set or NULL. Reviewed by: jhb Relnotes: yes Differential Revision: https://reviews.freebsd.org/D3792
2015-10-05 18:08:35 +00:00
STAILQ_FOREACH(sc, &syscalls, entries)
Add -c option to summarize number of calls, errors, and system time. Reviewed by: alfred
2009-05-12 20:42:12 +00:00
if (sc->ncalls) {
Cast time_t values to intmax_t and use %jd with printf. OK'ed by: delphij Approved by: trasz (mentor)
2009-12-21 17:41:57 +00:00
fprintf(trussinfo->outfile, "%-20s%5jd.%09ld%8d%8d\n",
sc->name, (intmax_t)sc->time.tv_sec,
sc->time.tv_nsec, sc->ncalls, sc->nerror);
Revert r247300 for now. I'll post a new changeset for review.
2013-02-26 19:46:59 +00:00
timespecadd(&total, &sc->time, &total);
Add -c option to summarize number of calls, errors, and system time. Reviewed by: alfred
2009-05-12 20:42:12 +00:00
ncall += sc->ncalls;
nerror += sc->nerror;
}
fprintf(trussinfo->outfile, "%20s%15s%8s%8s\n",
- Style(9) cleanup. Approved by: kib (mentor)
2012-09-02 11:03:18 +00:00
"", "-------------", "-------", "-------");
Cast time_t values to intmax_t and use %jd with printf. OK'ed by: delphij Approved by: trasz (mentor)
2009-12-21 17:41:57 +00:00
fprintf(trussinfo->outfile, "%-20s%5jd.%09ld%8d%8d\n",
- Style(9) cleanup. Approved by: kib (mentor)
2012-09-02 11:03:18 +00:00
"", (intmax_t)total.tv_sec, total.tv_nsec, ncall, nerror);
Add -c option to summarize number of calls, errors, and system time. Reviewed by: alfred
2009-05-12 20:42:12 +00:00
}
Reference in New Issue Copy Permalink