4 * On Linux using NPTL threads, -p will now attach to all threads in a process.
5 * Handle new mq_* system calls in Linux 2.6.6 and later.
10 * Report some new VM_* bit values on Linux.
11 * Better output for Linux sched_* system calls.
16 * Display multiple ioctl name matches on Linux.
20 * New port to AMD's x86-64 architecture. One strace binary can
21 handle both new x86-64 and old i386 processes.
22 * Fixed support for LFS64 calls.
23 * New switch -E to add/remove environment variables for the command.
24 * Merged s390/s390x port.
25 * Trace an unbounded number of processes.
26 * Handle numerous new system calls in Linux 2.5, and new threads semantics.
27 * Fixed bugs with attach/detach leaving things stopped.
28 * Fixed traced process seeing ECHILD despite live, traced children
29 in waitpid calls with WNOHANG.
30 * Stuart Menefy contributed a port to Linux/SH.
31 * Stephen Thomas contributed a port to Linux/SH64.
32 * Many other bug fixes.
36 * Fix Linux/ia64 support, looks like someone renamed a few things on us
37 * Fix the ioctl setup for Linux, turned out it did not really work.
38 Improve the ioctl extracter as well so we decode some more ones.
42 * compile fixes for Linux/mips
46 * Linux ia64 and hppa ports added
47 * The usual Linux syscall updates (includes 32bit uid/gid support),
48 * Linux ioctl list updated
49 * Support IPv6 scope ids
50 * FreeBSD/i386 port added
51 * UnixWare and Solaris updates
52 * Better support for tracing multithreaded processes in Linux
56 * Compiles on glibc2.0 systems again
57 * Linux/S390 port added
58 * The usual Linux syscall updates
59 * we can follow fork on arm now
63 * Linux/MIPS port added
64 * Lots of Linux updates again
65 * Improved IPv6 support
71 * Compiles on glibc2.0 systems again
75 * Get stat structure properly on Linux 64bit archs
76 * Personalities work again
77 * Compile correctly on SunOS again
83 * Linux (ultra)sparc fixes
90 * add support for more Linux architectures (powerpc, sparc, arm)
91 * support lots more Linux syscalls
93 * add IPX and IPIP support
94 * check stray syscall after execv
95 * fix hanging children
97 Changes in version 3.1
98 ======================
101 * Linux 68k is supported
102 * Linux alpha is supported
103 * configure is upgraded to autoconf 2.x
104 * using -f in combination with -e now works correctly
105 * output can be piped to a program
106 * tracing setuid programs works better
107 * it is now reasonable to install strace setuid to root in some circumstances
108 * new useful tracing names like file and process to trace whole
109 classes of system calls, e.g. -efile traces all system calls that
110 take a file name as an argument
111 * IPC calls on SunOS 4.1.x are decoded
112 * Linux program memory is reliably dereferenced
113 * Linux decodes at least the name of all syscalls as of pre2.0.4
114 * various cosmetic changes and bug fixes
116 Changes from versions 2.x to version 3.0
117 ========================================
119 * filename arguments are neither abbreviated nor stringified
120 * string arguments are now true C strings using octal instead of hex by default
121 * preprocessor constants are never shortened (e.g. was RDONLY => now O_RDONLY)
122 * by default the output for multiple processes now goes into one file
123 * all structures, vectors, bitsets, etc. use consistent output formats
124 * the -c option now means count calls, -i does what the old -c used to do
126 New Features in version 3.0
127 ===========================
129 * non-ascii strings can be optionally printed entirely in hex
130 * the output format is readable when mutiple processes are generating output
131 * exit values are printed in an alignment column
132 * is is possible to suppress messages about attaching and detaching
133 * various tracing features can be enabled on a per syscall/signal/desc basis
134 * selective tracing of syscalls
135 * selective printing of syscall structures
136 * selective abbreviation of long structures on a per syscall basis
137 * selective printing of raw syscall arguments and results
138 * selective tracing of signals
139 * selective dumping of all I/O read from file descriptors
140 * selective dumping of all I/O written to file descriptors
141 * optional counting of time, calls, and errors for each syscall