| 1 | Revision history for IO::AIO |
1 | Revision history for IO::AIO |
| 2 | |
2 | |
| 3 | TODO: better autoconf.pm that can return LIBS etc. |
3 | TODO: scandir - some dirs mostly contain subdirs - invert logic? |
| 4 | TODO: aio_cptree/mvtree |
4 | TODO: aio_cptree/mvtree |
| 5 | TODO: reduce condvar fairness: schedule hot-cache-threads first? |
5 | TODO: reduce condvar fairness: schedule hot-cache-threads first? |
| 6 | TODO: splice/tee/vmsplice? (http://kerneltrap.org/node/6505 http://lwn.net/Articles/178199/) |
6 | TODO: vmsplice? (http://kerneltrap.org/node/6505 http://lwn.net/Articles/178199/) |
| 7 | TODO: aio_fcntl, at least for file-locking |
7 | TODO: aio_fcntl, at least for file-locking |
|
|
8 | TODO: aio_mincore? |
|
|
9 | TODO: getxattr etc.? |
|
|
10 | TODO: F_DUPFD_CLOEXEC |
|
|
11 | TODO: emulation for splice? |
|
|
12 | TODO: eio_mmap|mlock|munmap|splice... |
|
|
13 | TODO: syncfs/sync windows: |
|
|
14 | TODO: F_SETPIPE_SZ, F_GETPIPE_SZ |
|
|
15 | http://stackoverflow.com/questions/65170/how-to-get-name-associated-with-open-handle/5286888#5286888 |
|
|
16 | http://blogs.msdn.com/b/adioltean/archive/2005/04/16/408947.aspx |
|
|
17 | http://msdn.microsoft.com/en-us/library/aa366789%28v=vs.85%29.aspx |
|
|
18 | http://msdn.microsoft.com/en-us/library/windows/desktop/aa366789%28v=vs.85%29.aspx |
|
|
19 | http://msdn.microsoft.com/en-us/library/windows/desktop/aa364425%28v=vs.85%29.aspx |
|
|
20 | http://msdn.microsoft.com/en-us/library/windows/desktop/aa364963%28v=vs.85%29.aspx |
|
|
21 | http://msdn.microsoft.com/en-us/library/windows/desktop/aa364996%28v=vs.85%29.aspx |
|
|
22 | http://msdn.microsoft.com/en-us/library/windows/desktop/aa364994%28v=vs.85%29.aspx |
|
|
23 | TODO: extra socket/tcp constants &c? |
| 8 | |
24 | |
|
|
25 | 4.3 Fri Apr 11 06:22:38 CEST 2014 |
|
|
26 | - perl5porters broke Async::Interrupt, BDB, EV, IO::AIO, OpenCL |
|
|
27 | without warning by switching the meaning of USE_SOCKETS_AS_HANDLES |
|
|
28 | in 5.18. What's so attractive about giving a shit about backwards |
|
|
29 | compatibility - I will never understand. |
|
|
30 | |
|
|
31 | 4.2 Sat Jan 25 01:13:14 CET 2014 |
|
|
32 | - aio_group could corrupt memory because it didn't restore |
|
|
33 | the stack after req_submit. |
|
|
34 | - be more careful on (e.g. permission) errors in bin/treescan. |
|
|
35 | - work around changes in ExtUtils::MakeMaker. |
|
|
36 | - (libeio) implement aio_realpath for win32. |
|
|
37 | - (xthread) work around compile time bugs in ptw32. |
|
|
38 | - added IO::AIO::pipesize. |
|
|
39 | - (libecb) insignificant update. |
|
|
40 | |
|
|
41 | 4.19 Sun Jan 6 12:47:26 CET 2013 |
|
|
42 | - avoid endless loop in fiemap with some XFS files. |
|
|
43 | - in aio_rename and aio_rmdir, specialcase the case of [$wd, "."] |
|
|
44 | and call rename/rmdir instead of renameat/unlinkat. |
|
|
45 | |
|
|
46 | 4.18 Thu Oct 11 07:01:26 CEST 2012 |
|
|
47 | - fix unintended xthread_create by intentionalising it :) |
|
|
48 | |
|
|
49 | 4.17 Thu Oct 11 05:19:47 CEST 2012 |
|
|
50 | - rename aio_fallocate to aio_allocate, to match documentation. |
|
|
51 | - add list of linux fsid values to aio_statvfs docs. |
|
|
52 | - work around a bug in btrfs' FIEMAP ioctl implementation. |
|
|
53 | - work around AIX bug: statvfs.f_fsid is a struct, not unsigned long |
|
|
54 | as per unix spec. |
|
|
55 | |
|
|
56 | 4.16 Tue Aug 14 05:39:03 CEST 2012 |
|
|
57 | - aio_statvfs was wromgly marked for wd emulation, causing |
|
|
58 | it to malfunction for paths. |
|
|
59 | - fix a crash in aio_fiemap, when used on ranges without any |
|
|
60 | extents. |
|
|
61 | - work around linux kernel bug (at least in 3.2): kernel might |
|
|
62 | trash fiemap input arguments. |
|
|
63 | - work around linux kernel bug (at least in 3.2): kernel does not |
|
|
64 | set FIEMAP_EXTENT_LAST on last segment. |
|
|
65 | - work around linux kernel bug (at least in 3.2): kernel silently |
|
|
66 | truncates segment count if a file. |
|
|
67 | - make fh argument of IO::AIO::mmap optional, also add mmap/munmap |
|
|
68 | to quick overview. |
|
|
69 | - splice can corrupt data. |
|
|
70 | - (libeio) remove pread/pwrite emulation. |
|
|
71 | - do not grow mmaped scalars in aio_read. |
|
|
72 | - add FALLOC_FL_PUNCH_HOLE and document fallocate. |
|
|
73 | - provide pread/pwrite for win32, making it link again on native win32. |
|
|
74 | |
|
|
75 | 4.15 Tue Apr 10 06:59:00 CEST 2012 |
|
|
76 | - always include linux/types.h for fiemap, for compatibility to |
|
|
77 | ancient systems (Paul Howarth). |
|
|
78 | - experimental support for IO::AIO::splice and ::tee (no aio_...). |
|
|
79 | - provide SEEK_HOLE and SEEK_DATA, if available. |
|
|
80 | - work around (again!) an immensely stupid bug in RHEL, defining |
|
|
81 | autoconf macros in linux system headers (analysed by Paul Howarth). |
|
|
82 | |
|
|
83 | 4.14 Sat Apr 7 02:45:18 CEST 2012 |
|
|
84 | - fix stat structure usage on windows, which caused bogus stat results. |
|
|
85 | - (libeio) make readahead emulation behave more like actual readahead by never failing. |
|
|
86 | - new request aio_seek. |
|
|
87 | - new request aio_fiemap. |
|
|
88 | - autogenerate the #ifdef/#define 0 blocks for symbols we export. |
|
|
89 | |
|
|
90 | 4.12 Fri Dec 30 08:51:25 CET 2011 |
|
|
91 | - realpath would return a random value if the name was too long, |
|
|
92 | instead of -1. |
|
|
93 | - port to c89. |
|
|
94 | |
|
|
95 | 4.11 Mon Oct 10 00:24:11 CEST 2011 |
|
|
96 | - libeio didn't compile on !linux (Paul Howarth). |
|
|
97 | |
|
|
98 | 4.1 Sun Oct 9 10:24:11 CEST 2011 |
|
|
99 | - IO::AIO did access uninitialised memory on unsuccessful stats. |
|
|
100 | - (libeio) added syncfs syscall wrapper. |
|
|
101 | - (libeio) set thread name on linux (ps -L/Hcx, top, gdb). |
|
|
102 | - (libeio) support multiple indepenent working directories. |
|
|
103 | - applied speling corrections by Nicholas Bamber. |
|
|
104 | - tune treescan #threads and #outstanding requests a bit. |
|
|
105 | - reduce the number of file descriptors in use at any one time |
|
|
106 | in treescan, to avoid running out of them (and using too much ram). |
|
|
107 | - take advantage of working directoriy abstractions |
|
|
108 | in aio_scandir and treescan. |
|
|
109 | - reduce compiled size by ~10% by not inlining some key functionality. |
|
|
110 | - added --progress switch to treescan. |
|
|
111 | |
|
|
112 | 4.0 Mon Jul 18 05:01:10 CEST 2011 |
|
|
113 | - INCOMPATIBLE CHANGE: fork is no longer supported (indeed, it never was), |
|
|
114 | see FORK BEHAVIOUR in manpage for details. |
|
|
115 | - passes testsuite on win32 now and may actually work |
|
|
116 | (activestate 5.10.1 / MSVC6, YMMV). |
|
|
117 | - (libeio) fix a deadlock where a wakeup signal could be missed when |
|
|
118 | a timeout occured at the same time. |
|
|
119 | - (libeio) added realpath. |
|
|
120 | - (libeio) added fallocate. |
|
|
121 | - (libeio) disabling sendfile on darwin unfortunately broke the emulation. |
|
|
122 | - (libeio) do not acquire any locks when forking. |
|
|
123 | - (libeio) use fewer time() syscalls when waiting for new requests. |
|
|
124 | |
|
|
125 | 3.93 Wed Jun 29 23:44:18 CEST 2011 |
|
|
126 | - ECB.H WAS MISSING. |
|
|
127 | |
|
|
128 | 3.92 Wed Jun 29 14:45:41 CEST 2011 |
|
|
129 | - ecb.h was missing. |
|
|
130 | |
|
|
131 | 3.91 Wed Jun 29 13:24:42 CEST 2011 |
|
|
132 | - (libeio) work around a Linux (and likely FreeBSD and other |
|
|
133 | kernels) bug where sendfile would not transfer all the requested |
|
|
134 | bytes on large transfers, using a heuristic. |
|
|
135 | - buggy sendfile caused aio_move/copy to sometimes fail for big |
|
|
136 | files (fortunately it checks that the whole file has been |
|
|
137 | transferred...) |
|
|
138 | - use libecb for higher performance and higher portability. |
|
|
139 | - (libeio) disable sendfile on darwin, it's too broken. |
|
|
140 | - disable fork tests on !linux, as only linux supports mixing |
|
|
141 | pthread and fork in perl. |
|
|
142 | - document the fact that fork doesn't work anymore when using |
|
|
143 | this module, on anything !GNU/Linux. |
|
|
144 | - increase timeout in t/04_fork.t, as too many CPAN-tester setups |
|
|
145 | run on a 0.1MHz cpu. Or so. |
|
|
146 | |
|
|
147 | 3.9 Fri May 27 02:43:47 CEST 2011 |
|
|
148 | - (libeio) fix memory corruption in aio_readdirx for the flags |
|
|
149 | combination READDIR_STAT_ORDER | READDIR_DIRS_FIRST. |
|
|
150 | - add lots of newer POSIX and GNU/Linux-specific open |
|
|
151 | flags. |
|
|
152 | |
|
|
153 | 3.8 Sun Mar 27 12:25:33 CEST 2011 |
|
|
154 | - use nonstandard but maybe-working-on-bsd fork technique. |
|
|
155 | - support a max_idle value of 0. |
|
|
156 | - support setting of idle timeout value (IO::AIO::idle_timeout). |
|
|
157 | |
|
|
158 | 3.72 Fri Feb 11 04:25:38 CET 2011 |
|
|
159 | - use _POSIX_MEMLOCK_RANGE to detect mlock/munlock. |
|
|
160 | - aio_mknod always used a dev_t value of 0. |
|
|
161 | - new treescan option: --grep. |
|
|
162 | - add more S_IF macros, and major/minor/makedev "macros". |
|
|
163 | |
|
|
164 | 3.71 Thu Dec 30 08:18:46 CET 2010 |
|
|
165 | - the numerical result value passed to callbacks did not stringify |
|
|
166 | correctly, due to internal reuse and failure to reset the sv flags. |
|
|
167 | - actually test for posix_[mf]advise, as at least uClibc defines |
|
|
168 | _POSIX_ADVISORY_INFO without actually having any of the required |
|
|
169 | funcitonality. ugh. |
|
|
170 | |
|
|
171 | 3.7 Mon Nov 1 23:00:34 CET 2010 |
|
|
172 | - implement/add madvise, mmap, munmap, aio_mlockall, munlockall, |
|
|
173 | aio_mlock, munlock, msync and mtouch to @EXPORT_OK. |
|
|
174 | - document the sad state of affairs w.r.t. pthread on many bsds. |
|
|
175 | - do not enable mmap on systems without _POSIX_MAPPED_FILES |
|
|
176 | (openbsd 4.8). |
|
|
177 | - do not leak memory in IO::AIO::mmap when the scalar already |
|
|
178 | had string-data. |
|
|
179 | - add O_RDWR, O_APPEND and O_EXCL symbols and semi-document |
|
|
180 | them. |
|
|
181 | - cache the result IV, for a minor speedup in the common case. |
|
|
182 | - croak when an mmapped-scalar changes location, to detect |
|
|
183 | user-errors better. |
|
|
184 | - fix aio_readlink prototype. |
|
|
185 | |
|
|
186 | 3.65 Wed Mar 31 02:45:05 CEST 2010 |
|
|
187 | - actually use PATH_MAX instead of NAME_MAX for readlink, as to not |
|
|
188 | cut off long pathnames in aio_readlink (based on patch by |
|
|
189 | Rasmus Andersson). |
|
|
190 | - a double fork partially killed the event pipe (great testcase |
|
|
191 | by dormando). affects IO::AIO, BDB and Async::Interrupt. |
|
|
192 | - suck steve's dick and rename our symbols to avoid colliding |
|
|
193 | with nonstandard identifier spam on the broken os x pseudo-os. |
|
|
194 | affects IO::AIO and BDB. |
|
|
195 | |
|
|
196 | 3.6 Mon Jan 11 00:43:39 CET 2010 |
|
|
197 | - (libeio) more fixes for the freebsd/apple sendfile - broken by |
|
|
198 | manpage standards, but apparently correct accoridng to actual |
|
|
199 | kernel sources. |
|
|
200 | - add IO::AIO::mmap/munmap support. |
|
|
201 | - add IO::AIO::m(un)lockall support. |
|
|
202 | - clean up manpage. |
|
|
203 | |
|
|
204 | 3.5 Thu Jan 7 21:25:04 CET 2010 |
|
|
205 | - (liebio) fix freebsd sendfile (Vanilla Hsu). |
|
|
206 | - (libeio) also fix darwin sendfile that suffered from a similar bug. |
|
|
207 | - add aio_statvfs as interface to statvfs/fstatvfs. |
|
|
208 | - work around buggy symlink() on cygwin in the testsuite. |
|
|
209 | - wtf. freebsd-8.0 actually passes the testsuite! I knew |
|
|
210 | they could implement fork and semaphores one day! cheers! |
|
|
211 | |
|
|
212 | 3.4 Sat Jan 2 15:13:04 CET 2010 |
|
|
213 | - (libeio) max_poll_time was not properly converted to ticks. |
|
|
214 | - clarify cancel_subs description. |
|
|
215 | - IO::AIO::sendfile did not actually return the return value |
|
|
216 | from the sendfile call. |
|
|
217 | - implement aio_msync, aio_mtouch. |
|
|
218 | - (libeio) tentatively support darwin in sendfile. |
|
|
219 | |
|
|
220 | 3.31 Thu Nov 12 02:14:29 CET 2009 |
|
|
221 | - fix result status documentation of aio_copy, aio_move. |
|
|
222 | - speed up object creation considerably by avoiding |
|
|
223 | hash lookups. |
|
|
224 | |
|
|
225 | 3.3 Wed Aug 5 13:52:58 CEST 2009 |
|
|
226 | - use common::sense. |
|
|
227 | - use common schmorp.h header. |
|
|
228 | - allow integers (file descriptors) in addition to file handles. |
|
|
229 | - take advantage of linux' eventfd (but it seems slower on smp?). |
|
|
230 | - use poll not select on posix platforms. |
|
|
231 | |
|
|
232 | 3.261 Wed Jul 1 10:11:51 CEST 2009 |
|
|
233 | - more 0S X workarounds (patch by Tokuhiro Matsuno). |
|
|
234 | |
|
|
235 | 3.26 Tue Jun 30 09:33:26 CEST 2009 |
|
|
236 | - 0S X of course claims to be posix 2008 but lacks posix_fadvise. |
|
|
237 | try to work around this horribly broken OS in a somewhat hackish |
|
|
238 | way. might help other os's too. |
|
|
239 | |
|
|
240 | 3.25 Sat Jun 27 05:18:26 CEST 2009 |
|
|
241 | - added IO::AIO::fadvise and IO::AIO::sendfile. |
|
|
242 | - (libeio) replaced quicksort+insertion sort by a tuned radix |
|
|
243 | sort + insertion sort, resulting in comparable runtime (usually |
|
|
244 | faster) to the old version, but without any ill side effects on |
|
|
245 | degenerated (for quicksort) data. |
|
|
246 | - (libeio) correctly sort dirs by inodes if we have perfect type |
|
|
247 | knowledge and DIRS_FIRST is used. |
|
|
248 | - (libeio) internally the wrong DT_* constants were used, but no |
|
|
249 | known system has differing EIO_DT_*/DT_* values, so not an issue. |
|
|
250 | - removed a (harmless) assert that was left over on the code |
|
|
251 | but should not have been. |
|
|
252 | - use more correct types (change IO length from IV to VAL64, |
|
|
253 | which makes a difference on systems without 64 bit perls). |
|
|
254 | |
|
|
255 | 3.23 Sat Jun 13 16:57:58 CEST 2009 |
|
|
256 | - fix off-by-one bug in aio_readdir that was introduced in the |
|
|
257 | rushed 3.22 release. |
|
|
258 | |
|
|
259 | 3.22 Sat Jun 13 15:32:40 CEST 2009 |
|
|
260 | - speed up readdirx sort algorithm slightly. |
|
|
261 | - bin/treescan was missing from distro tarball. |
|
|
262 | |
|
|
263 | 3.21 Fri Jun 12 18:45:53 CEST 2009 |
| 9 | - new options --dirs and --files for treescan. |
264 | - new options --dirs and --files for treescan. |
| 10 | - install bin/treescan by default. |
265 | - install bin/treescan by default. |
| 11 | - (libeio) aio_readdir can now be cancelled while executing. |
266 | - (libeio) aio_readdir can now be cancelled while executing. |
|
|
267 | - fix a printf format string for 64 bit systems (could lead |
|
|
268 | to problems on big endian 64 bit systems). |
|
|
269 | - do not use qsort() but our own algorithm: glibc initialises |
|
|
270 | the whole locale and I/O subsystem inside qsort, causing |
|
|
271 | a stack overflow on 32 bit machines. The new sort uses much less |
|
|
272 | stack and is more than twice as fast in typical situations. |
| 12 | |
273 | |
| 13 | 3.2 Sun Jun 7 20:30:05 CEST 2009 |
274 | 3.2 Sun Jun 7 20:30:05 CEST 2009 |
| 14 | - (libeio) pwrite emulation was even more flawed than intended and did |
275 | - (libeio) pwrite emulation was even more flawed than intended and did |
| 15 | not restore the file offset. |
276 | not restore the file offset. |
| 16 | - add aio_readdirx, which can return inode and filetype and sort |
277 | - add aio_readdirx, which can return inode and filetype and sort |
