pax_global_header00006660000000000000000000000064112636770360014525gustar00rootroot0000000000000052 comment=093bc44e821b92405faf8090a56bcf2506532ba7 libaio-0.3.109/000077500000000000000000000000001126367703600131365ustar00rootroot00000000000000libaio-0.3.109/.version000066400000000000000000000000101126367703600146130ustar00rootroot000000000000000.3.109 libaio-0.3.109/COPYING000066400000000000000000000636441126367703600142060ustar00rootroot00000000000000 GNU LESSER GENERAL PUBLIC LICENSE Version 2.1, February 1999 Copyright (C) 1991, 1999 Free Software Foundation, Inc. 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA Everyone is permitted to copy and distribute verbatim copies of this license document, but changing it is not allowed. [This is the first released version of the Lesser GPL. It also counts as the successor of the GNU Library Public License, version 2, hence the version number 2.1.] Preamble The licenses for most software are designed to take away your freedom to share and change it. By contrast, the GNU General Public Licenses are intended to guarantee your freedom to share and change free software--to make sure the software is free for all its users. This license, the Lesser General Public License, applies to some specially designated software packages--typically libraries--of the Free Software Foundation and other authors who decide to use it. You can use it too, but we suggest you first think carefully about whether this license or the ordinary General Public License is the better strategy to use in any particular case, based on the explanations below. When we speak of free software, we are referring to freedom of use, not price. Our General Public Licenses are designed to make sure that you have the freedom to distribute copies of free software (and charge for this service if you wish); that you receive source code or can get it if you want it; that you can change the software and use pieces of it in new free programs; and that you are informed that you can do these things. To protect your rights, we need to make restrictions that forbid distributors to deny you these rights or to ask you to surrender these rights. These restrictions translate to certain responsibilities for you if you distribute copies of the library or if you modify it. For example, if you distribute copies of the library, whether gratis or for a fee, you must give the recipients all the rights that we gave you. You must make sure that they, too, receive or can get the source code. If you link other code with the library, you must provide complete object files to the recipients, so that they can relink them with the library after making changes to the library and recompiling it. And you must show them these terms so they know their rights. We protect your rights with a two-step method: (1) we copyright the library, and (2) we offer you this license, which gives you legal permission to copy, distribute and/or modify the library. To protect each distributor, we want to make it very clear that there is no warranty for the free library. Also, if the library is modified by someone else and passed on, the recipients should know that what they have is not the original version, so that the original author's reputation will not be affected by problems that might be introduced by others. ^L Finally, software patents pose a constant threat to the existence of any free program. We wish to make sure that a company cannot effectively restrict the users of a free program by obtaining a restrictive license from a patent holder. Therefore, we insist that any patent license obtained for a version of the library must be consistent with the full freedom of use specified in this license. Most GNU software, including some libraries, is covered by the ordinary GNU General Public License. This license, the GNU Lesser General Public License, applies to certain designated libraries, and is quite different from the ordinary General Public License. We use this license for certain libraries in order to permit linking those libraries into non-free programs. When a program is linked with a library, whether statically or using a shared library, the combination of the two is legally speaking a combined work, a derivative of the original library. The ordinary General Public License therefore permits such linking only if the entire combination fits its criteria of freedom. The Lesser General Public License permits more lax criteria for linking other code with the library. We call this license the "Lesser" General Public License because it does Less to protect the user's freedom than the ordinary General Public License. It also provides other free software developers Less of an advantage over competing non-free programs. These disadvantages are the reason we use the ordinary General Public License for many libraries. However, the Lesser license provides advantages in certain special circumstances. For example, on rare occasions, there may be a special need to encourage the widest possible use of a certain library, so that it becomes a de-facto standard. To achieve this, non-free programs must be allowed to use the library. A more frequent case is that a free library does the same job as widely used non-free libraries. In this case, there is little to gain by limiting the free library to free software only, so we use the Lesser General Public License. In other cases, permission to use a particular library in non-free programs enables a greater number of people to use a large body of free software. For example, permission to use the GNU C Library in non-free programs enables many more people to use the whole GNU operating system, as well as its variant, the GNU/Linux operating system. Although the Lesser General Public License is Less protective of the users' freedom, it does ensure that the user of a program that is linked with the Library has the freedom and the wherewithal to run that program using a modified version of the Library. The precise terms and conditions for copying, distribution and modification follow. Pay close attention to the difference between a "work based on the library" and a "work that uses the library". The former contains code derived from the library, whereas the latter must be combined with the library in order to run. ^L GNU LESSER GENERAL PUBLIC LICENSE TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION 0. This License Agreement applies to any software library or other program which contains a notice placed by the copyright holder or other authorized party saying it may be distributed under the terms of this Lesser General Public License (also called "this License"). Each licensee is addressed as "you". A "library" means a collection of software functions and/or data prepared so as to be conveniently linked with application programs (which use some of those functions and data) to form executables. The "Library", below, refers to any such software library or work which has been distributed under these terms. A "work based on the Library" means either the Library or any derivative work under copyright law: that is to say, a work containing the Library or a portion of it, either verbatim or with modifications and/or translated straightforwardly into another language. (Hereinafter, translation is included without limitation in the term "modification".) "Source code" for a work means the preferred form of the work for making modifications to it. For a library, complete source code means all the source code for all modules it contains, plus any associated interface definition files, plus the scripts used to control compilation and installation of the library. Activities other than copying, distribution and modification are not covered by this License; they are outside its scope. The act of running a program using the Library is not restricted, and output from such a program is covered only if its contents constitute a work based on the Library (independent of the use of the Library in a tool for writing it). Whether that is true depends on what the Library does and what the program that uses the Library does. 1. You may copy and distribute verbatim copies of the Library's complete source code as you receive it, in any medium, provided that you conspicuously and appropriately publish on each copy an appropriate copyright notice and disclaimer of warranty; keep intact all the notices that refer to this License and to the absence of any warranty; and distribute a copy of this License along with the Library. You may charge a fee for the physical act of transferring a copy, and you may at your option offer warranty protection in exchange for a fee. 2. You may modify your copy or copies of the Library or any portion of it, thus forming a work based on the Library, and copy and distribute such modifications or work under the terms of Section 1 above, provided that you also meet all of these conditions: a) The modified work must itself be a software library. b) You must cause the files modified to carry prominent notices stating that you changed the files and the date of any change. c) You must cause the whole of the work to be licensed at no charge to all third parties under the terms of this License. d) If a facility in the modified Library refers to a function or a table of data to be supplied by an application program that uses the facility, other than as an argument passed when the facility is invoked, then you must make a good faith effort to ensure that, in the event an application does not supply such function or table, the facility still operates, and performs whatever part of its purpose remains meaningful. (For example, a function in a library to compute square roots has a purpose that is entirely well-defined independent of the application. Therefore, Subsection 2d requires that any application-supplied function or table used by this function must be optional: if the application does not supply it, the square root function must still compute square roots.) These requirements apply to the modified work as a whole. If identifiable sections of that work are not derived from the Library, and can be reasonably considered independent and separate works in themselves, then this License, and its terms, do not apply to those sections when you distribute them as separate works. But when you distribute the same sections as part of a whole which is a work based on the Library, the distribution of the whole must be on the terms of this License, whose permissions for other licensees extend to the entire whole, and thus to each and every part regardless of who wrote it. Thus, it is not the intent of this section to claim rights or contest your rights to work written entirely by you; rather, the intent is to exercise the right to control the distribution of derivative or collective works based on the Library. In addition, mere aggregation of another work not based on the Library with the Library (or with a work based on the Library) on a volume of a storage or distribution medium does not bring the other work under the scope of this License. 3. You may opt to apply the terms of the ordinary GNU General Public License instead of this License to a given copy of the Library. To do this, you must alter all the notices that refer to this License, so that they refer to the ordinary GNU General Public License, version 2, instead of to this License. (If a newer version than version 2 of the ordinary GNU General Public License has appeared, then you can specify that version instead if you wish.) Do not make any other change in these notices. ^L Once this change is made in a given copy, it is irreversible for that copy, so the ordinary GNU General Public License applies to all subsequent copies and derivative works made from that copy. This option is useful when you wish to copy part of the code of the Library into a program that is not a library. 4. You may copy and distribute the Library (or a portion or derivative of it, under Section 2) in object code or executable form under the terms of Sections 1 and 2 above provided that you accompany it with the complete corresponding machine-readable source code, which must be distributed under the terms of Sections 1 and 2 above on a medium customarily used for software interchange. If distribution of object code is made by offering access to copy from a designated place, then offering equivalent access to copy the source code from the same place satisfies the requirement to distribute the source code, even though third parties are not compelled to copy the source along with the object code. 5. A program that contains no derivative of any portion of the Library, but is designed to work with the Library by being compiled or linked with it, is called a "work that uses the Library". Such a work, in isolation, is not a derivative work of the Library, and therefore falls outside the scope of this License. However, linking a "work that uses the Library" with the Library creates an executable that is a derivative of the Library (because it contains portions of the Library), rather than a "work that uses the library". The executable is therefore covered by this License. Section 6 states terms for distribution of such executables. When a "work that uses the Library" uses material from a header file that is part of the Library, the object code for the work may be a derivative work of the Library even though the source code is not. Whether this is true is especially significant if the work can be linked without the Library, or if the work is itself a library. The threshold for this to be true is not precisely defined by law. If such an object file uses only numerical parameters, data structure layouts and accessors, and small macros and small inline functions (ten lines or less in length), then the use of the object file is unrestricted, regardless of whether it is legally a derivative work. (Executables containing this object code plus portions of the Library will still fall under Section 6.) Otherwise, if the work is a derivative of the Library, you may distribute the object code for the work under the terms of Section 6. Any executables containing that work also fall under Section 6, whether or not they are linked directly with the Library itself. ^L 6. As an exception to the Sections above, you may also combine or link a "work that uses the Library" with the Library to produce a work containing portions of the Library, and distribute that work under terms of your choice, provided that the terms permit modification of the work for the customer's own use and reverse engineering for debugging such modifications. You must give prominent notice with each copy of the work that the Library is used in it and that the Library and its use are covered by this License. You must supply a copy of this License. If the work during execution displays copyright notices, you must include the copyright notice for the Library among them, as well as a reference directing the user to the copy of this License. Also, you must do one of these things: a) Accompany the work with the complete corresponding machine-readable source code for the Library including whatever changes were used in the work (which must be distributed under Sections 1 and 2 above); and, if the work is an executable linked with the Library, with the complete machine-readable "work that uses the Library", as object code and/or source code, so that the user can modify the Library and then relink to produce a modified executable containing the modified Library. (It is understood that the user who changes the contents of definitions files in the Library will not necessarily be able to recompile the application to use the modified definitions.) b) Use a suitable shared library mechanism for linking with the Library. A suitable mechanism is one that (1) uses at run time a copy of the library already present on the user's computer system, rather than copying library functions into the executable, and (2) will operate properly with a modified version of the library, if the user installs one, as long as the modified version is interface-compatible with the version that the work was made with. c) Accompany the work with a written offer, valid for at least three years, to give the same user the materials specified in Subsection 6a, above, for a charge no more than the cost of performing this distribution. d) If distribution of the work is made by offering access to copy from a designated place, offer equivalent access to copy the above specified materials from the same place. e) Verify that the user has already received a copy of these materials or that you have already sent this user a copy. For an executable, the required form of the "work that uses the Library" must include any data and utility programs needed for reproducing the executable from it. However, as a special exception, the materials to be distributed need not include anything that is normally distributed (in either source or binary form) with the major components (compiler, kernel, and so on) of the operating system on which the executable runs, unless that component itself accompanies the executable. It may happen that this requirement contradicts the license restrictions of other proprietary libraries that do not normally accompany the operating system. Such a contradiction means you cannot use both them and the Library together in an executable that you distribute. ^L 7. You may place library facilities that are a work based on the Library side-by-side in a single library together with other library facilities not covered by this License, and distribute such a combined library, provided that the separate distribution of the work based on the Library and of the other library facilities is otherwise permitted, and provided that you do these two things: a) Accompany the combined library with a copy of the same work based on the Library, uncombined with any other library facilities. This must be distributed under the terms of the Sections above. b) Give prominent notice with the combined library of the fact that part of it is a work based on the Library, and explaining where to find the accompanying uncombined form of the same work. 8. You may not copy, modify, sublicense, link with, or distribute the Library except as expressly provided under this License. Any attempt otherwise to copy, modify, sublicense, link with, or distribute the Library is void, and will automatically terminate your rights under this License. However, parties who have received copies, or rights, from you under this License will not have their licenses terminated so long as such parties remain in full compliance. 9. You are not required to accept this License, since you have not signed it. However, nothing else grants you permission to modify or distribute the Library or its derivative works. These actions are prohibited by law if you do not accept this License. Therefore, by modifying or distributing the Library (or any work based on the Library), you indicate your acceptance of this License to do so, and all its terms and conditions for copying, distributing or modifying the Library or works based on it. 10. Each time you redistribute the Library (or any work based on the Library), the recipient automatically receives a license from the original licensor to copy, distribute, link with or modify the Library subject to these terms and conditions. You may not impose any further restrictions on the recipients' exercise of the rights granted herein. You are not responsible for enforcing compliance by third parties with this License. ^L 11. If, as a consequence of a court judgment or allegation of patent infringement or for any other reason (not limited to patent issues), conditions are imposed on you (whether by court order, agreement or otherwise) that contradict the conditions of this License, they do not excuse you from the conditions of this License. If you cannot distribute so as to satisfy simultaneously your obligations under this License and any other pertinent obligations, then as a consequence you may not distribute the Library at all. For example, if a patent license would not permit royalty-free redistribution of the Library by all those who receive copies directly or indirectly through you, then the only way you could satisfy both it and this License would be to refrain entirely from distribution of the Library. If any portion of this section is held invalid or unenforceable under any particular circumstance, the balance of the section is intended to apply, and the section as a whole is intended to apply in other circumstances. It is not the purpose of this section to induce you to infringe any patents or other property right claims or to contest validity of any such claims; this section has the sole purpose of protecting the integrity of the free software distribution system which is implemented by public license practices. Many people have made generous contributions to the wide range of software distributed through that system in reliance on consistent application of that system; it is up to the author/donor to decide if he or she is willing to distribute software through any other system and a licensee cannot impose that choice. This section is intended to make thoroughly clear what is believed to be a consequence of the rest of this License. 12. If the distribution and/or use of the Library is restricted in certain countries either by patents or by copyrighted interfaces, the original copyright holder who places the Library under this License may add an explicit geographical distribution limitation excluding those countries, so that distribution is permitted only in or among countries not thus excluded. In such case, this License incorporates the limitation as if written in the body of this License. 13. The Free Software Foundation may publish revised and/or new versions of the Lesser General Public License from time to time. Such new versions will be similar in spirit to the present version, but may differ in detail to address new problems or concerns. Each version is given a distinguishing version number. If the Library specifies a version number of this License which applies to it and "any later version", you have the option of following the terms and conditions either of that version or of any later version published by the Free Software Foundation. If the Library does not specify a license version number, you may choose any version ever published by the Free Software Foundation. ^L 14. If you wish to incorporate parts of the Library into other free programs whose distribution conditions are incompatible with these, write to the author to ask for permission. For software which is copyrighted by the Free Software Foundation, write to the Free Software Foundation; we sometimes make exceptions for this. Our decision will be guided by the two goals of preserving the free status of all derivatives of our free software and of promoting the sharing and reuse of software generally. NO WARRANTY 15. BECAUSE THE LIBRARY IS LICENSED FREE OF CHARGE, THERE IS NO WARRANTY FOR THE LIBRARY, TO THE EXTENT PERMITTED BY APPLICABLE LAW. EXCEPT WHEN OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR OTHER PARTIES PROVIDE THE LIBRARY "AS IS" WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESSED OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE ENTIRE RISK AS TO THE QUALITY AND PERFORMANCE OF THE LIBRARY IS WITH YOU. SHOULD THE LIBRARY PROVE DEFECTIVE, YOU ASSUME THE COST OF ALL NECESSARY SERVICING, REPAIR OR CORRECTION. 16. IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MAY MODIFY AND/OR REDISTRIBUTE THE LIBRARY AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES, INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING OUT OF THE USE OR INABILITY TO USE THE LIBRARY (INCLUDING BUT NOT LIMITED TO LOSS OF DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY YOU OR THIRD PARTIES OR A FAILURE OF THE LIBRARY TO OPERATE WITH ANY OTHER SOFTWARE), EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGES. END OF TERMS AND CONDITIONS ^L How to Apply These Terms to Your New Libraries If you develop a new library, and you want it to be of the greatest possible use to the public, we recommend making it free software that everyone can redistribute and change. You can do so by permitting redistribution under these terms (or, alternatively, under the terms of the ordinary General Public License). To apply these terms, attach the following notices to the library. It is safest to attach them to the start of each source file to most effectively convey the exclusion of warranty; and each file should have at least the "copyright" line and a pointer to where the full notice is found. Copyright (C) This library is free software; you can redistribute it and/or modify it under the terms of the GNU Lesser General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version. This library is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public License for more details. You should have received a copy of the GNU Lesser General Public License along with this library; if not, write to the Free Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA Also add information on how to contact you by electronic and paper mail. You should also get your employer (if you work as a programmer) or your school, if any, to sign a "copyright disclaimer" for the library, if necessary. Here is a sample; alter the names: Yoyodyne, Inc., hereby disclaims all copyright interest in the library `Frob' (a library for tweaking knobs) written by James Random Hacker. , 1 April 1990 Ty Coon, President of Vice That's all there is to it! libaio-0.3.109/ChangeLog000066400000000000000000000033041126367703600147100ustar00rootroot000000000000000.3.107 - Make tests compile again on modern systems (warnings + -Werror) - Add 'make partcheck' and don't require manual setup for testing. - Change test harness to compile against this dir, not global install - Fix 5.t for archs where PROT_WRITE mappings are readable. - Allow sending of SIGXFSZ on aio over limits - Explicitly specify bash for runtests.sh - Put deprecating comments on never-merged io_prep_poll - Add io_prep_preadv and io_prep_pwritev - Add eventfd support (io_set_eventfd). 0.4.0 - remove libredhat-kernel - add rough outline for man pages - make the compiled io_getevents() add the extra parameter and pass the timeout for updating as per 2.5 - fixes for ia64, now works - fixes for x86-64 - powerpc support from Gianni Tedesco - disable the NULL check in harness/cases/4.t on ia64: ia64 maps the 0 page and causes this check to fail. 0.3.15 - use real syscall interface, but don't break source compatibility yet (that will happen with 0.4.0) 0.3.13 - add test cases 0.3.11 - use library versioning of libredhat-kernel to always provide a fallback 0.3.9 - add io_queue_release function 0.3.8 - make clean deletes libredhat-kernel.so.1 - const struct timespec * - add make srpm target 0.3.7 - fix assembly function .types - export io_getevents - fix io_submit function prototype to match the kernel - provide /usr/lib/libredhat-kernel.so link for compilation (do NOT link against libredhat-kernel.so directly) - fix soname to libaio.so.1 - fix dummy libredhat-kernel's soname - work around nfs bug - provide and install libredhat-kernel.so.1 stub - Makefile improvements - make sure dummy libredhat-kernel.so only returns -ENOSYS libaio-0.3.109/INSTALL000066400000000000000000000014521126367703600141710ustar00rootroot00000000000000To install the library, execute the command: make prefix=`pwd`/usr install which will install the binaries and header files into the directory usr. Set prefix=/usr to get them installed into the main system. Please note: Do not attempt to install on the system the "libredhat-kernel.so" file. It is a dummy shared library provided only for the purpose of being able to bootstrap this facility while running on systems without the correct libredhat-kernel.so built. The contents of the included libredhat-kernel.so are only stubs; this library is NOT functional for anything except the internal purpose of linking libaio.so against the provided stubs. At runtime, libaio.so requires a real libredhat-kernel.so library; this is provided by the Red Hat kernel RPM packages with async I/O functionality. libaio-0.3.109/Makefile000066400000000000000000000036021126367703600145770ustar00rootroot00000000000000NAME=libaio SPECFILE=$(NAME).spec VERSION=$(shell awk '/Version:/ { print $$2 }' $(SPECFILE)) RELEASE=$(shell awk '/Release:/ { print $$2 }' $(SPECFILE)) CVSTAG = $(NAME)_$(subst .,-,$(VERSION))_$(subst .,-,$(RELEASE)) RPMBUILD=$(shell `which rpmbuild >&/dev/null` && echo "rpmbuild" || echo "rpm") prefix=/usr includedir=$(prefix)/include libdir=$(prefix)/lib pkgname := libaio ver := $(shell cat .version) version_dirty := $(strip $(shell git diff | lsdiff | grep ".version")) dirty := $(strip $(shell git whatchanged ${pkgname}-${ver}... 2>/dev/null)) new_changes := $(strip $(shell git diff 2>/dev/null)) tag_valid := $(strip $(shell git tag -l ${pkgname}-${ver})) default: all all: @$(MAKE) -C src install: @$(MAKE) -C src install prefix=$(prefix) includedir=$(includedir) libdir=$(libdir) check: @$(MAKE) -C harness check partcheck: all @$(MAKE) -C harness partcheck clean: @$(MAKE) -C src clean @$(MAKE) -C harness clean tag: tag-checks git tag ${pkgname}-${ver} archive: uptodate @echo "Creating archive from version ${pkgname}-${ver}:" git archive --format=tar --prefix=${pkgname}-${ver}/ \ ${pkgname}-${ver} | gzip > ${pkgname}-${ver}.tar.gz tag-checks: ifneq (${version_dirty},) @echo "Error: .version is not committed to the repository." @exit 1 endif ifneq (${new_changes},) @echo "Error: there are changes in this checkout that have not been" @echo "committed to the repository." @exit 1 endif uptodate: tag-checks ifneq (${dirty},) @echo -n "Error: this branch contains changes that are not part of tag " @echo "${pkgname}-${ver}." @echo -n "Try running git-whatchanged ${pkgname}-${ver}... to see " @echo "the changes." @exit 1 endif ifeq (${tag_valid},) @echo "Error: tag ${pkgname}-${ver} does not exist. Run 'make tag'" @echo "and try again." @exit 1 endif srpm: archive $(RPMBUILD) --define "_sourcedir `pwd`" --define "_srcrpmdir `pwd`" --nodeps -bs $(SPECFILE) libaio-0.3.109/TODO000066400000000000000000000001721126367703600136260ustar00rootroot00000000000000- Write man pages. - Make -static links against libaio work. - Fallback on userspace if the kernel calls return -ENOSYS. libaio-0.3.109/harness/000077500000000000000000000000001126367703600146015ustar00rootroot00000000000000libaio-0.3.109/harness/Makefile000066400000000000000000000025101126367703600162370ustar00rootroot00000000000000# foo. TEST_SRCS:=$(shell find cases/ -name \*.t | sort -n -t/ -k2) EXTRAPROGS:=cases/8.p cases/10.p PARTPROGS:=$(filter-out $(EXTRAPROGS), $(patsubst %.t,%.p,$(TEST_SRCS))) PROGS:=$(PARTPROGS) $(EXTRAPROGS) HARNESS_SRCS:=main.c # io_queue.c CFLAGS=-Wall -Werror -I../src -g -O #-lpthread -lrt all: $(PROGS) $(PROGS): %.p: %.t $(HARNESS_SRCS) $(CC) $(CFLAGS) -DTEST_NAME=\"$<\" -o $@ main.c ../src/libaio.a clean: rm -f $(PROGS) *.o runtests.out rofile wofile rwfile .PHONY: testdir/rofile: testdir .PHONY rm -f $@ echo "test" >$@ chmod 400 $@ testdir/wofile: testdir .PHONY rm -f $@ echo "test" >$@ chmod 200 $@ testdir/rwfile: testdir .PHONY rm -f $@ echo "test" >$@ chmod 600 $@ testdir testdir.enospc testdir.ext2: mkdir $@ root: .PHONY @if [ `id -u` -ne 0 ]; then echo Need root for check, try partcheck >&2; exit 1; fi partcheck: $(PARTPROGS) testdir/rofile testdir/rwfile testdir/wofile ./runtests.sh $(PARTPROGS) ext2.img: dd if=/dev/zero bs=1M count=10 of=$@ mke2fs -F -b 4096 $@ extracheck: $(EXTRAPROGS) root testdir.ext2 testdir.enospc ext2.img mount -o loop -t ext2 ext2-enospc.img testdir.enospc ./runtests.sh cases/10.p; ret=$$?; umount testdir.enospc; exit $$ret mount -o loop -t ext2 ext2.img testdir.ext2 ./runtests.sh cases/8.p; ret=$$?; umount testdir.ext2; exit $$ret check: partcheck extracheck libaio-0.3.109/harness/README000066400000000000000000000012431126367703600154610ustar00rootroot00000000000000Notes on running this test suite: To run the test suite, run "make check". All test cases should pass and there should be 0 fails. Several of the test cases require a directory on the filesystem under test for the creation of test files, as well as the generation of error conditions. The test cases assume the directories (or symlinks to directories) are as follows: testdir/ - used for general read/write test cases. Must have at least as much free space as the machine has RAM (up to 768MB). testdir.enospc/ - a filesystem that has space for writing 8KB out, but fails with -ENOSPC beyond 8KB. testdir.ext2/ - must be an ext2 filesystem. libaio-0.3.109/harness/attic/000077500000000000000000000000001126367703600157055ustar00rootroot00000000000000libaio-0.3.109/harness/attic/0.t000066400000000000000000000001471126367703600162330ustar00rootroot00000000000000/* 0.t Test harness check: okay. */ int test_main(void) { printf("test_main: okay\n"); return 0; } libaio-0.3.109/harness/attic/1.t000066400000000000000000000001471126367703600162340ustar00rootroot00000000000000/* 1.t Test harness check: fail. */ int test_main(void) { printf("test_main: fail\n"); return 1; } libaio-0.3.109/harness/cases/000077500000000000000000000000001126367703600156775ustar00rootroot00000000000000libaio-0.3.109/harness/cases/10.t000066400000000000000000000034361126367703600163120ustar00rootroot00000000000000/* 10.t - uses testdir.enospc/rwfile - Check results on out-of-space and out-of-quota. (10.t) - write that fills filesystem but does not go over should succeed - write that fills filesystem and goes over should be partial - write to full filesystem should return -ENOSPC - read beyond end of file after ENOSPC should return 0 */ #include "aio_setup.h" #include #include #include int test_main(void) { /* Note: changing either of these requires updating the ext2-enospc.img * filesystem image. Also, if SIZE is less than PAGE_SIZE, problems * crop up due to ext2's preallocation. */ #define LIMIT 65536 #define SIZE 65536 char *buf; int rwfd; int status = 0, res; rwfd = open("testdir.enospc/rwfile", O_RDWR|O_CREAT|O_TRUNC, 0600); assert(rwfd != -1); res = ftruncate(rwfd, 0); assert(res == 0); buf = malloc(SIZE); assert(buf != NULL); memset(buf, 0, SIZE); status |= attempt_rw(rwfd, buf, SIZE, LIMIT-SIZE, WRITE, SIZE); status |= attempt_rw(rwfd, buf, SIZE, LIMIT-SIZE, READ, SIZE); status |= attempt_rw(rwfd, buf, SIZE, LIMIT, WRITE, -ENOSPC); status |= attempt_rw(rwfd, buf, SIZE, LIMIT, READ, 0); res = ftruncate(rwfd, 0); assert(res == 0); status |= attempt_rw(rwfd, buf, SIZE, 1+LIMIT-SIZE, WRITE, SIZE-1); status |= attempt_rw(rwfd, buf, SIZE, 1+LIMIT-SIZE, READ, SIZE-1); status |= attempt_rw(rwfd, buf, SIZE, LIMIT, READ, 0); status |= attempt_rw(rwfd, buf, SIZE, LIMIT, WRITE, -ENOSPC); status |= attempt_rw(rwfd, buf, SIZE, LIMIT, READ, 0); status |= attempt_rw(rwfd, buf, 0, LIMIT, WRITE, 0); res = close(rwfd); assert(res == 0); res = unlink("testdir.enospc/rwfile"); assert(res == 0); return status; } libaio-0.3.109/harness/cases/11.t000066400000000000000000000014661126367703600163140ustar00rootroot00000000000000/* 11.t - uses testdir/rwfile - repeated read / write of same page (to check accounting) (11.t) */ #include "aio_setup.h" #include #include #include int test_main(void) { #define COUNT 1000000 #define SIZE 256 char *buf; int rwfd; int status = 0; int i; rwfd = open("testdir/rwfile", O_RDWR|O_CREAT|O_TRUNC, 0600); assert(rwfd != -1); buf = malloc(SIZE); assert(buf != NULL); memset(buf, 0, SIZE); for (i=0; i #include #include #include #include "aio_setup.h" void test_child(void) { int res; res = attempt_io_submit(io_ctx, 0, NULL, -EINVAL); fflush(stdout); _exit(res); } int test_main(void) { int res, status; pid_t pid; sigset_t set; if (attempt_io_submit(io_ctx, 0, NULL, 0)) return 1; sigemptyset(&set); sigaddset(&set, SIGCHLD); sigprocmask(SIG_BLOCK, &set, NULL); fflush(NULL); pid = fork(); assert(pid != -1); if (pid == 0) test_child(); res = waitpid(pid, &status, 0); if (WIFEXITED(status)) { int failed = (WEXITSTATUS(status) != 0); printf("child exited with status %d%s\n", WEXITSTATUS(status), failed ? " -- FAILED" : ""); return failed; } /* anything else: failed */ if (WIFSIGNALED(status)) printf("child killed by signal %d -- FAILED.\n", WTERMSIG(status)); return 1; } libaio-0.3.109/harness/cases/13.t000066400000000000000000000027351126367703600163160ustar00rootroot00000000000000/* 13.t - uses testdir/rwfile - Submit multiple writes larger than aio-max-size (deadlocks on older aio code) */ #include "aio_setup.h" #include #include #include int test_main(void) { #define SIZE (1024 * 1024) #define IOS 8 struct iocb iocbs[IOS]; struct iocb *iocb_list[IOS]; char *bufs[IOS]; int rwfd; int status = 0, res; int i; rwfd = open("testdir/rwfile", O_RDWR|O_CREAT|O_TRUNC, 0600); assert(rwfd != -1); res = ftruncate(rwfd, 0); assert(res == 0); for (i=0; i #include #include #include #include "aio_setup.h" #include #define SIZE 768*1024*1024 //just submit an I/O int test_child(void) { char *buf; int rwfd; int res; long size; struct iocb iocb; struct iocb *iocbs[] = { &iocb }; int loop = 10; int i; aio_setup(1024); size = SIZE; printf("size = %ld\n", size); rwfd = open("testdir/rwfile", O_RDWR); assert(rwfd != -1); res = ftruncate(rwfd, 0); assert(res == 0); buf = malloc(size); assert(buf != NULL); for(i=0;i #include #include int test_main(void) { #define SIZE 512 #define NUM_IOV 10 char buf[SIZE*NUM_IOV]; struct iovec iov[NUM_IOV]; int rwfd; int status = 0, res, i; rwfd = open("testdir/rwfile", O_RDWR); assert(rwfd != -1); res = ftruncate(rwfd, sizeof(buf)); assert(res == 0); for (i = 0; i < NUM_IOV; i++) { iov[i].iov_base = buf + i*SIZE; iov[i].iov_len = SIZE; memset(iov[i].iov_base, i, SIZE); } status |= attempt_rw(rwfd, iov, NUM_IOV, 0, WRITEV, SIZE*NUM_IOV); res = pread(rwfd, buf, sizeof(buf), 0); assert(res == sizeof(buf)); for (i = 0; i < NUM_IOV; i++) { unsigned int j; for (j = 0; j < SIZE; j++) { if (buf[i*SIZE + j] != i) { printf("Unexpected value after writev at %i\n", i*SIZE + j); status |= 1; break; } } } if (!status) printf("Checking memory: [Success]\n"); memset(buf, 0, sizeof(buf)); status |= attempt_rw(rwfd, iov, NUM_IOV, 0, READV, SIZE*NUM_IOV); for (i = 0; i < NUM_IOV; i++) { unsigned int j; for (j = 0; j < SIZE; j++) { if (buf[i*SIZE + j] != i) { printf("Unexpected value after readv at %i\n", i*SIZE + j); status |= 1; break; } } } /* Check that offset works. */ status |= attempt_rw(rwfd, iov+1, NUM_IOV-1, SIZE, WRITEV, SIZE*(NUM_IOV-1)); memset(buf, 0, sizeof(buf)); res = pread(rwfd, buf, sizeof(buf), 0); assert(res == sizeof(buf)); for (i = 1; i < NUM_IOV; i++) { unsigned int j; for (j = 0; j < SIZE; j++) { if (buf[i*SIZE + j] != i) { printf("Unexpected value after offset writev at %i\n", i*SIZE + j); status |= 1; break; } } } if (!status) printf("Checking memory: [Success]\n"); memset(buf, 0, sizeof(buf)); status |= attempt_rw(rwfd, iov+1, NUM_IOV-1, SIZE, READV, SIZE*(NUM_IOV-1)); for (i = 1; i < NUM_IOV; i++) { unsigned int j; for (j = 0; j < SIZE; j++) { if (buf[i*SIZE + j] != i) { printf("Unexpected value after offset readv at %i\n", i*SIZE + j); status |= 1; break; } } } if (!status) printf("Checking memory: [Success]\n"); return status; } libaio-0.3.109/harness/cases/16.t000066400000000000000000000044351126367703600163200ustar00rootroot00000000000000/* 16.t - eventfd tests. */ #include #include #include /* For SYS_xxx definitions */ #ifndef SYS_eventfd #if defined(__i386__) #define SYS_eventfd 323 #elif defined(__x86_64__) #define SYS_eventfd 284 #elif defined(__ia64__) #define SYS_eventfd 1309 #elif defined(__PPC__) #define SYS_eventfd 307 #elif defined(__s390__) #define SYS_eventfd 318 #elif defined(__alpha__) #define SYS_eventfd 478 #else #error define SYS_eventfd for your arch! #endif #endif int test_main(void) { /* 10 MB takes long enough that we would fail if eventfd * returned immediately. */ #define SIZE 10000000 char *buf; struct io_event io_event; struct iocb iocb; struct iocb *iocbs[] = { &iocb }; int rwfd, efd; int res; io_context_t io_ctx; uint64_t event; struct timespec notime = { .tv_sec = 0, .tv_nsec = 0 }; buf = malloc(SIZE); assert(buf); efd = syscall(SYS_eventfd, 0); if (efd < 0) { if (errno == ENOSYS) { printf("No eventfd support. [SKIPPING]\n"); exit(0); } err(1, "Failed to get eventfd"); } rwfd = open("testdir/rwfile", O_RDWR); assert(rwfd != -1); res = ftruncate(rwfd, 0); assert(res == 0); memset(buf, 0x42, SIZE); /* Write test. */ res = io_queue_init(1024, &io_ctx); assert(res == 0); io_prep_pwrite(&iocb, rwfd, buf, SIZE, 0); io_set_eventfd(&iocb, efd); res = io_submit(io_ctx, 1, iocbs); assert(res == 1); alarm(30); res = read(efd, &event, sizeof(event)); assert(res == sizeof(event)); assert(event == 1); /* This should now be ready. */ res = io_getevents(io_ctx, 0, 1, &io_event, ¬ime); if (res != 1) err(1, "io_getevents did not return 1 event after eventfd"); assert(io_event.res == SIZE); printf("eventfd write test [SUCCESS]\n"); /* Read test. */ memset(buf, 0, SIZE); io_prep_pread(&iocb, rwfd, buf, SIZE, 0); io_set_eventfd(&iocb, efd); res = io_submit(io_ctx, 1, iocbs); assert(res == 1); alarm(30); res = read(efd, &event, sizeof(event)); assert(res == sizeof(event)); assert(event == 1); /* This should now be ready. */ res = io_getevents(io_ctx, 0, 1, &io_event, ¬ime); if (res != 1) err(1, "io_getevents did not return 1 event after eventfd"); assert(io_event.res == SIZE); for (res = 0; res < SIZE; res++) assert(buf[res] == 0x42); printf("eventfd read test [SUCCESS]\n"); return 0; } libaio-0.3.109/harness/cases/2.t000066400000000000000000000016341126367703600162310ustar00rootroot00000000000000/* 2.t - io_setup (#2) - with invalid context pointer - with maxevents <= 0 - with an already initialized ctxp */ int attempt(int n, io_context_t *ctxp, int expect) { int res; printf("expect %3d: io_setup(%5d, %p) = ", expect, n, ctxp); fflush(stdout); res = io_setup(n, ctxp); printf("%3d [%s]%s\n", res, strerror(-res), (res != expect) ? " -- FAILED" : ""); if (res != expect) return 1; return 0; } int test_main(void) { io_context_t ctx; int status = 0; ctx = NULL; status |= attempt(-1000, KERNEL_RW_POINTER, -EFAULT); status |= attempt( 1000, KERNEL_RW_POINTER, -EFAULT); status |= attempt( 0, KERNEL_RW_POINTER, -EFAULT); status |= attempt(-1000, &ctx, -EINVAL); status |= attempt( -1, &ctx, -EINVAL); status |= attempt( 0, &ctx, -EINVAL); assert(ctx == NULL); status |= attempt( 1, &ctx, 0); status |= attempt( 1, &ctx, -EINVAL); return status; } libaio-0.3.109/harness/cases/3.t000066400000000000000000000013471126367703600162330ustar00rootroot00000000000000/* 3.t - io_submit/io_getevents with invalid addresses (3.t) */ #include "aio_setup.h" int test_main(void) { struct iocb a, b; struct iocb *good_ios[] = { &a, &b }; struct iocb *bad1_ios[] = { NULL, &b }; struct iocb *bad2_ios[] = { KERNEL_RW_POINTER, &a }; int status = 0; status |= attempt_io_submit(BAD_CTX, 1, good_ios, -EINVAL); status |= attempt_io_submit( io_ctx, 0, good_ios, 0); status |= attempt_io_submit( io_ctx, 1, NULL, -EFAULT); status |= attempt_io_submit( io_ctx, 1, (void *)-1, -EFAULT); status |= attempt_io_submit( io_ctx, 2, bad1_ios, -EFAULT); status |= attempt_io_submit( io_ctx, 2, bad2_ios, -EFAULT); status |= attempt_io_submit( io_ctx, -1, good_ios, -EINVAL); return status; } libaio-0.3.109/harness/cases/4.t000066400000000000000000000041041126367703600162260ustar00rootroot00000000000000/* 4.t - read of descriptor without read permission (4.t) - write to descriptor without write permission (4.t) - check that O_APPEND writes actually append */ #include "aio_setup.h" #define SIZE 512 #define READ 'r' #define WRITE 'w' int attempt(int fd, void *buf, int count, long long pos, int rw, int expect) { struct iocb iocb; int res; switch(rw) { case READ: io_prep_pread (&iocb, fd, buf, count, pos); break; case WRITE: io_prep_pwrite(&iocb, fd, buf, count, pos); break; } printf("expect %3d: (%c), res = ", expect, rw); fflush(stdout); res = sync_submit(&iocb); printf("%3d [%s]%s\n", res, (res <= 0) ? strerror(-res) : "Success", (res != expect) ? " -- FAILED" : ""); if (res != expect) return 1; return 0; } int test_main(void) { char buf[SIZE]; int rofd, wofd, rwfd; int status = 0, res; memset(buf, 0, SIZE); rofd = open("testdir/rofile", O_RDONLY); assert(rofd != -1); wofd = open("testdir/wofile", O_WRONLY); assert(wofd != -1); rwfd = open("testdir/rwfile", O_RDWR); assert(rwfd != -1); status |= attempt(rofd, buf, SIZE, 0, WRITE, -EBADF); status |= attempt(wofd, buf, SIZE, 0, READ, -EBADF); status |= attempt(rwfd, buf, SIZE, 0, WRITE, SIZE); status |= attempt(rwfd, buf, SIZE, 0, READ, SIZE); status |= attempt(rwfd, buf, SIZE, -1, READ, -EINVAL); status |= attempt(rwfd, buf, SIZE, -1, WRITE, -EINVAL); rwfd = open("testdir/rwfile", O_RDWR|O_APPEND); assert(rwfd != -1); res = ftruncate(rwfd, 0); assert(res == 0); status |= attempt(rwfd, buf, SIZE, 0, READ, 0); status |= attempt(rwfd, "1234", 4, 0, WRITE, 4); status |= attempt(rwfd, "5678", 4, 0, WRITE, 4); memset(buf, 0, SIZE); status |= attempt(rwfd, buf, SIZE, 0, READ, 8); printf("read after append: [%s]\n", buf); assert(memcmp(buf, "12345678", 8) == 0); status |= attempt(rwfd, KERNEL_RW_POINTER, SIZE, 0, READ, -EFAULT); status |= attempt(rwfd, KERNEL_RW_POINTER, SIZE, 0, WRITE, -EFAULT); /* Some architectures map the 0 page. Ugh. */ #if !defined(__ia64__) status |= attempt(rwfd, NULL, SIZE, 0, WRITE, -EFAULT); #endif return status; } libaio-0.3.109/harness/cases/5.t000066400000000000000000000027511126367703600162350ustar00rootroot00000000000000/* 5.t - Write from a mmap() of the same file. (5.t) */ #include "aio_setup.h" #include #include int test_main(void) { int page_size = getpagesize(); #define SIZE 512 char *buf; int rwfd; int status = 0, res; rwfd = open("testdir/rwfile", O_RDWR); assert(rwfd != -1); res = ftruncate(rwfd, 512); assert(res == 0); buf = mmap(0, page_size, PROT_READ|PROT_WRITE, MAP_SHARED, rwfd, 0); assert(buf != (char *)-1); status |= attempt_rw(rwfd, buf, SIZE, 0, WRITE, SIZE); status |= attempt_rw(rwfd, buf, SIZE, 0, READ, SIZE); res = munmap(buf, page_size); assert(res == 0); buf = mmap(0, page_size, PROT_READ|PROT_WRITE, MAP_SHARED, rwfd, 0); assert(buf != (char *)-1); status |= attempt_rw(rwfd, buf, SIZE, 0, READ, SIZE); status |= attempt_rw(rwfd, buf, SIZE, 0, WRITE, SIZE); res = munmap(buf, page_size); assert(res == 0); buf = mmap(0, page_size, PROT_READ, MAP_SHARED, rwfd, 0); assert(buf != (char *)-1); status |= attempt_rw(rwfd, buf, SIZE, 0, WRITE, SIZE); status |= attempt_rw(rwfd, buf, SIZE, 0, READ, -EFAULT); res = munmap(buf, page_size); assert(res == 0); buf = mmap(0, page_size, PROT_WRITE, MAP_SHARED, rwfd, 0); assert(buf != (char *)-1); status |= attempt_rw(rwfd, buf, SIZE, 0, READ, SIZE); /* Whether PROT_WRITE is readable is arch-dependent. So compare * against read result. */ res = read(rwfd, buf, SIZE); if (res < 0) res = -errno; status |= attempt_rw(rwfd, buf, SIZE, 0, WRITE, res); return status; } libaio-0.3.109/harness/cases/6.t000066400000000000000000000020421126367703600162270ustar00rootroot00000000000000/* 6.t - huge reads (pinned pages) (6.t) - huge writes (6.t) */ #include "aio_setup.h" #include long getmemsize(void) { FILE *f = fopen("/proc/meminfo", "r"); long size; int gotit = 0; char str[256]; assert(f != NULL); while (NULL != fgets(str, 255, f)) { str[255] = 0; if (0 == memcmp(str, "MemTotal:", 9)) { if (1 == sscanf(str + 9, "%ld", &size)) { gotit = 1; break; } } } fclose(f); assert(gotit != 0); return size; } int test_main(void) { char *buf; int rwfd; int status = 0, res; long size; size = getmemsize(); printf("size = %ld\n", size); assert(size >= (16 * 1024)); if (size > (768 * 1024)) size = 768 * 1024; size *= 1024; rwfd = open("testdir/rwfile", O_RDWR); assert(rwfd != -1); res = ftruncate(rwfd, 0); assert(res == 0); buf = malloc(size); assert(buf != NULL); //memset(buf, 0, size); status |= attempt_rw(rwfd, buf, size, 0, WRITE, size); status |= attempt_rw(rwfd, buf, size, 0, READ, size); //res = ftruncate(rwfd, 0); assert(res == 0); return status; } libaio-0.3.109/harness/cases/7.t000066400000000000000000000015071126367703600162350ustar00rootroot00000000000000/* 7.t - Write overlapping the file size rlimit boundary: should be a short write. (7.t) - Write at the file size rlimit boundary: should give EFBIG. (I think the spec requires that you do NOT deliver SIGXFSZ in this case, where you would do so for sync IO.) (7.t) - Special case: a write of zero bytes at or beyond the file size rlimit boundary must return success. (7.t) */ #include #include void SET_RLIMIT(long long limit) { struct rlimit rlim; int res; /* Seems that we do send SIGXFSZ, but hard to fix... */ signal(SIGXFSZ, SIG_IGN); rlim.rlim_cur = limit; assert(rlim.rlim_cur == limit); rlim.rlim_max = limit; assert(rlim.rlim_max == limit); res = setrlimit(RLIMIT_FSIZE, &rlim); assert(res == 0); } #define LIMIT 8192 #define FILENAME "testdir/rwfile" #include "common-7-8.h" libaio-0.3.109/harness/cases/8.t000066400000000000000000000010671126367703600162370ustar00rootroot00000000000000/* 8.t - Ditto for the above three tests at the offset maximum (largest possible ext2/3 file size.) (8.t) */ #include #include long long get_fs_limit(int fd) { long long min = 0, max = 9223372036854775807LL; char c = 0; while (max - min > 1) { if (pwrite64(fd, &c, 1, (min + max) / 2) == -1) max = (min + max) / 2; else { ftruncate(fd, 0); min = (min + max) / 2; } } return max; } #define SET_RLIMIT(x) do ; while (0) #define LIMIT get_fs_limit(rwfd) #define FILENAME "testdir.ext2/rwfile" #include "common-7-8.h" libaio-0.3.109/harness/cases/aio_setup.h000066400000000000000000000041661126367703600200470ustar00rootroot00000000000000#include io_context_t io_ctx; #define BAD_CTX ((io_context_t)-1) void aio_setup(int n) { int res = io_queue_init(n, &io_ctx); if (res != 0) { printf("io_queue_setup(%d) returned %d (%s)\n", n, res, strerror(-res)); exit(3); } } int attempt_io_submit(io_context_t ctx, long nr, struct iocb *ios[], int expect) { int res; printf("expect %3d: io_submit(%10p, %3ld, %10p) = ", expect, ctx, nr, ios); fflush(stdout); res = io_submit(ctx, nr, ios); printf("%3d [%s]%s\n", res, (res <= 0) ? strerror(-res) : "", (res != expect) ? " -- FAILED" : ""); if (res != expect) return 1; return 0; } int sync_submit(struct iocb *iocb) { struct io_event event; struct iocb *iocbs[] = { iocb }; int res; /* 30 second timeout should be enough */ struct timespec ts; ts.tv_sec = 30; ts.tv_nsec = 0; res = io_submit(io_ctx, 1, iocbs); if (res != 1) { printf("sync_submit: io_submit res=%d [%s]\n", res, strerror(-res)); return res; } res = io_getevents(io_ctx, 0, 1, &event, &ts); if (res != 1) { printf("sync_submit: io_getevents res=%d [%s]\n", res, strerror(-res)); return res; } return event.res; } #define SETUP aio_setup(1024) #define READ 'r' #define WRITE 'w' #define READ_SILENT 'R' #define WRITE_SILENT 'W' #define READV '<' #define WRITEV '>' int attempt_rw(int fd, void *buf, int count, long long pos, int rw, int expect) { struct iocb iocb; int res; int silent = 0; switch(rw) { case READ_SILENT: silent = 1; case READ: io_prep_pread (&iocb, fd, buf, count, pos); break; case WRITE_SILENT: silent = 1; case WRITE: io_prep_pwrite(&iocb, fd, buf, count, pos); break; case WRITEV: io_prep_pwritev(&iocb, fd, buf, count, pos); break; case READV: io_prep_preadv(&iocb, fd, buf, count, pos); break; } if (!silent) { printf("expect %5d: (%c), res = ", expect, rw); fflush(stdout); } res = sync_submit(&iocb); if (!silent || res != expect) { if (silent) printf("expect %5d: (%c), res = ", expect, rw); printf("%5d [%s]%s\n", res, (res <= 0) ? strerror(-res) : "Success", (res != expect) ? " -- FAILED" : ""); } if (res != expect) return 1; return 0; } libaio-0.3.109/harness/cases/common-7-8.h000066400000000000000000000016141126367703600176530ustar00rootroot00000000000000/* common-7-8.h */ #include "aio_setup.h" #include #include #define SIZE 512 int test_main(void) { char *buf; int rwfd; int status = 0, res; long long limit; rwfd = open(FILENAME, O_RDWR|O_CREAT, 0600); assert(rwfd != -1); res = ftruncate(rwfd, 0); assert(res == 0); buf = malloc(SIZE); assert(buf != NULL); memset(buf, 0, SIZE); limit = LIMIT; SET_RLIMIT(limit); status |= attempt_rw(rwfd, buf, SIZE, limit-SIZE, WRITE, SIZE); status |= attempt_rw(rwfd, buf, SIZE, limit-SIZE, READ, SIZE); status |= attempt_rw(rwfd, buf, SIZE, 1+limit-SIZE, WRITE, SIZE-1); status |= attempt_rw(rwfd, buf, SIZE, 1+limit-SIZE, READ, SIZE-1); status |= attempt_rw(rwfd, buf, SIZE, limit, WRITE, -EFBIG); status |= attempt_rw(rwfd, buf, SIZE, limit, READ, 0); status |= attempt_rw(rwfd, buf, 0, limit, WRITE, 0); return status; } libaio-0.3.109/harness/ext2-enospc.img000066400000000000000000014400001126367703600174450ustar00rootroot000000000000008A, 8nZ‡=2\‡= Sï6[‡=Ní €(ä8Cú†KŽ]@?ƒpA,ÿÿÿþÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ~ƒÀ<~ƒÀ<~ƒÀ<íAm}[‡=Á[‡=Á[‡= íA0~ƒÀ<~ƒÀ<~ƒÀ< ¤m¨[‡=Á[‡=Á[‡=Á[‡=ê‚ !"#$%Á@¤m̱[‡=±[‡=±[‡=êlZ[\]^_`abcdefgÛ@ . .. $ lost+foundrwfile Äfiller¸rwfile .ô..&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYghijklmnopqrstuvwxyz{|}~€‚ƒ„…†‡ˆ‰Š‹ŒŽ‘’“”•–—˜™š›œžŸ ¡¢£¤¥¦§¨©ª«¬­®¯°±²³´µ¶·¸¹º»¼½¾¿ÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖ×ØÙÚÛÜÝÞßàáâãäåæçèéêëìíîïðñòóôõö÷øùúûüýþÿ      !"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefhijklmnopqrstuvwxyz{|}~€‚ƒ„…†‡ˆ‰Š‹ŒŽlibaio-0.3.109/harness/main.c000066400000000000000000000012201126367703600156640ustar00rootroot00000000000000#define _GNU_SOURCE #include #include #include #include #include #include #include #include #include #if __LP64__ == 0 #define KERNEL_RW_POINTER ((void *)0xc0010000) #else //#warning Not really sure where kernel memory is. Guessing. #define KERNEL_RW_POINTER ((void *)0xffffffff81000000) #endif char test_name[] = TEST_NAME; #include TEST_NAME int main(void) { int res; #if defined(SETUP) SETUP; #endif res = test_main(); printf("test %s completed %s.\n", test_name, res ? "FAILED" : "PASSED" ); fflush(stdout); return res ? 1 : 0; } libaio-0.3.109/harness/runtests.sh000077500000000000000000000006071126367703600170320ustar00rootroot00000000000000#!/bin/bash passes=0 fails=0 echo "Test run starting at" `date` while [ $# -ge 1 ] ; do this_test=$1 shift echo "Starting $this_test" $this_test 2>&1 res=$? if [ $res -eq 0 ] ; then str="" ; passes=$[passes + 1] ; else str=" -- FAILED" ; fails=$[fails + 1] ; fi echo "Completed $this_test with $res$str". done echo "Pass: $passes Fail: $fails" echo "Test run complete at" `date` libaio-0.3.109/libaio.spec000066400000000000000000000155041126367703600152560ustar00rootroot00000000000000Name: libaio Version: 0.3.109 Release: 1 Summary: Linux-native asynchronous I/O access library License: LGPL Group: System Environment/Libraries Source: %{name}-%{version}.tar.gz BuildRoot: %{_tmppath}/%{name}-root # Fix ExclusiveArch as we implement this functionality on more architectures ExclusiveArch: i386 x86_64 ia64 s390 s390x ppc ppc64 ppc64pseries ppc64iseries alpha alphaev6 %{arm} %description The Linux-native asynchronous I/O facility ("async I/O", or "aio") has a richer API and capability set than the simple POSIX async I/O facility. This library, libaio, provides the Linux-native API for async I/O. The POSIX async I/O facility requires this library in order to provide kernel-accelerated async I/O capabilities, as do applications which require the Linux-native async I/O API. %package devel Summary: Development files for Linux-native asynchronous I/O access Group: Development/System Requires: libaio Provides: libaio.so.1 %description devel This package provides header files to include and libraries to link with for the Linux-native asynchronous I/O facility ("async I/O", or "aio"). %prep %setup %build make %install [ "$RPM_BUILD_ROOT" != "/" ] && rm -rf $RPM_BUILD_ROOT make install prefix=$RPM_BUILD_ROOT/usr \ libdir=$RPM_BUILD_ROOT/%{_libdir} \ root=$RPM_BUILD_ROOT %clean [ "$RPM_BUILD_ROOT" != "/" ] && rm -rf $RPM_BUILD_ROOT %post -p /sbin/ldconfig %postun -p /sbin/ldconfig %files %defattr(-,root,root) %attr(0755,root,root) %{_libdir}/libaio.so.* %doc COPYING TODO %files devel %defattr(-,root,root) %attr(0644,root,root) %{_includedir}/* %attr(0755,root,root) %{_libdir}/libaio.so %attr(0644,root,root) %{_libdir}/libaio.a %changelog * Tue Jun 9 2009 Jeff Moyer - 0.3.108-1 - add ARM architecture support (grabbed from Debian arches tree) - replace check of __i386__ with __LP64__ in test harness (Jeff Moyer) - change the .spec's copyright tag to license * Wed Jan 9 2008 Jeff Moyer - 0.3.107-1 - Fix the test harness (Rusty Russell) - Add eventfd support (Rusty Russell) * Tue Jan 3 2006 Jeff Moyer - 0.3.106-1 - Add a .proc directive for the ia64_aio_raw_syscall macro. This sounds a lot like the previous entry, but that one fixed the __ia64_raw_syscall macro, located in syscall-ia64.h. This macro is in raw_syscall.c, which pretty much only exists for ia64. This bug prevented the package from building with newer version of gcc. * Mon Aug 1 2005 Jeff Moyer - 0.3.105-1 - Add a .proc directive for the ia64 raw syscall macro. * Fri Apr 1 2005 Jeff Moyer - 0.3.104-1 - Add Alpha architecture support. (Sergey Tikhonov ) * Tue Jan 25 2005 Jeff Moyer - 0.3.103-1 - Fix SONAME breakage. In changing file names around, I also changed the SONAME, which is a no no. * Thu Oct 14 2004 Jeff Moyer - 0.3.102-1 - S390 asm had a bug; I forgot to update the clobber list. Lucky for me, newer compilers complain about such things. - Also update the s390 asm to look more like the new kernel variants. * Wed Oct 13 2004 Jeff Moyer - 0.3.101-1 - Revert syscall return values to be -ERRNO. This was an inadvertant bug introduced when clobber lists changed. - add ppc64pseries and ppc64iseries to exclusivearch * Tue Sep 14 2004 Jeff Moyer - 0.3.100-1 - Switch around the tests for _PPC_ and _powerpc64_ so that the ppc64 platforms get the right padding. * Wed Jul 14 2004 Jeff Moyer - 0.3.99-4 - Ok, there was a race in moving the cvs module. Someone rebuild from the old cvs into fc3. *sigh* bumping rev. * Wed Jul 14 2004 Jeff Moyer - 0.3.99-3 - Actually provide libaio.so.1. * Tue Mar 30 2004 Jeff Moyer - 0.3.99-2 - Apparently the 0.3.93 patch was not meant for 0.3.96. Backed it out. * Tue Mar 30 2004 Jeff Moyer - 0.3.99-1 - Fix compat calls. - make library .so.1.0.0 and make symlinks properly. - Fix header file for inclusion in c++ code. * Thu Feb 26 2004 Jeff Moyer 0.3.98-2 - bah. fix version nr in changelog. * Thu Feb 26 2004 Jeff Moyer 0.3.98-1 - fix compiler warnings. * Thu Feb 26 2004 Jeff Moyer 0.3.97-2 - make srpm was using rpm to do a build. changed that to use rpmbuild if it exists, and fallback to rpm if it doesn't. * Tue Feb 24 2004 Jeff Moyer 0.3.97-1 - Use libc syscall(2) instead of rolling our own calling mechanism. This change is inspired due to a failure to build with newer gcc, since clobber lists were wrong. - Add -fpic to the CFLAGS for all architectures. Should address bz #109457. - change a #include from to . Fixes a build issue on s390. * Wed Jul 7 2003 Bill Nottingham 0.3.96-3 - fix paths on lib64 arches * Wed Jun 18 2003 Michael K. Johnson 0.3.96-2 - optimization in io_getevents from Arjan van de Ven in 0.3.96-1 - deal with ia64 in 0.3.96-2 * Wed May 28 2003 Michael K. Johnson 0.3.95-1 - ppc bugfix from Julie DeWandel * Tue May 20 2003 Michael K. Johnson 0.3.94-1 - symbol versioning fix from Ulrich Drepper * Mon Jan 27 2003 Benjamin LaHaise - bump to 0.3.93-3 for rebuild. * Mon Dec 16 2002 Benjamin LaHaise - libaio 0.3.93 test release - add powerpc support from Gianni Tedesco - add s/390 support from Arnd Bergmann * Fri Sep 12 2002 Benjamin LaHaise - libaio 0.3.92 test release - build on x86-64 * Thu Sep 12 2002 Benjamin LaHaise - libaio 0.3.91 test release - build on ia64 - remove libredhat-kernel from the .spec file * Thu Sep 5 2002 Benjamin LaHaise - libaio 0.3.90 test release * Mon Apr 29 2002 Benjamin LaHaise - add requires initscripts >= 6.47-1 to get boot time libredhat-kernel linkage correct. - typo fix * Thu Apr 25 2002 Benjamin LaHaise - make /usr/lib/libredhat-kernel.so point to /lib/libredhat-kernel.so.1.0.0 * Mon Apr 15 2002 Tim Powers - make the post scriptlet not use /bin/sh * Sat Apr 12 2002 Benjamin LaHaise - add /lib/libredhat-kernel* to %files. * Fri Apr 12 2002 Benjamin LaHaise - make the dummy install as /lib/libredhat-kernel.so.1.0.0 so that ldconfig will link against it if no other is installed. * Tue Jan 22 2002 Benjamin LaHaise - add io_getevents * Tue Jan 22 2002 Michael K. Johnson - Make linker happy with /usr/lib symlink for libredhat-kernel.so * Mon Jan 21 2002 Michael K. Johnson - Added stub library * Sun Jan 20 2002 Michael K. Johnson - Initial packaging libaio-0.3.109/man/000077500000000000000000000000001126367703600137115ustar00rootroot00000000000000libaio-0.3.109/man/aio.3000066400000000000000000000212341126367703600145470ustar00rootroot00000000000000.TH aio 3 2002-09-12 "Linux 2.4" Linux AIO" .SH NAME aio \- Asynchronous IO .SH SYNOPSIS .nf .B #include .sp .br .B #include .sp .fi .SH DESCRIPTION The POSIX.1b standard defines a new set of I/O operations which can significantly reduce the time an application spends waiting at I/O. The new functions allow a program to initiate one or more I/O operations and then immediately resume normal work while the I/O operations are executed in parallel. This functionality is available if the .IR "unistd.h" file defines the symbol .B "_POSIX_ASYNCHRONOUS_IO" . These functions are part of the library with realtime functions named .IR "librt" . They are not actually part of the .IR "libc" binary. The implementation of these functions can be done using support in the kernel (if available) or using an implementation based on threads at userlevel. In the latter case it might be necessary to link applications with the thread library .IR "libpthread" in addition to .IR "librt" and .IR "libaio" . All AIO operations operate on files which were opened previously. There might be arbitrarily many operations running for one file. The asynchronous I/O operations are controlled using a data structure named .IR "struct aiocb" It is defined in .IR "aio.h" as follows. .nf struct aiocb { int aio_fildes; /* File desriptor. */ int aio_lio_opcode; /* Operation to be performed. */ int aio_reqprio; /* Request priority offset. */ volatile void *aio_buf; /* Location of buffer. */ size_t aio_nbytes; /* Length of transfer. */ struct sigevent aio_sigevent; /* Signal number and value. */ /* Internal members. */ struct aiocb *__next_prio; int __abs_prio; int __policy; int __error_code; __ssize_t __return_value; #ifndef __USE_FILE_OFFSET64 __off_t aio_offset; /* File offset. */ char __pad[sizeof (__off64_t) - sizeof (__off_t)]; #else __off64_t aio_offset; /* File offset. */ #endif char __unused[32]; }; .fi The POSIX.1b standard mandates that the .IR "struct aiocb" structure contains at least the members described in the following table. There might be more elements which are used by the implementation, but depending upon these elements is not portable and is highly deprecated. .TP .IR "int aio_fildes" This element specifies the file descriptor to be used for the operation. It must be a legal descriptor, otherwise the operation will fail. The device on which the file is opened must allow the seek operation. I.e., it is not possible to use any of the AIO operations on devices like terminals where an .IR "lseek" call would lead to an error. .TP .IR "off_t aio_offset" This element specifies the offset in the file at which the operation (input or output) is performed. Since the operations are carried out in arbitrary order and more than one operation for one file descriptor can be started, one cannot expect a current read/write position of the file descriptor. .TP .IR "volatile void *aio_buf" This is a pointer to the buffer with the data to be written or the place where the read data is stored. .TP .IR "size_t aio_nbytes" This element specifies the length of the buffer pointed to by .IR "aio_buf" . .TP .IR "int aio_reqprio" If the platform has defined .B "_POSIX_PRIORITIZED_IO" and .B "_POSIX_PRIORITY_SCHEDULING" , the AIO requests are processed based on the current scheduling priority. The .IR "aio_reqprio" element can then be used to lower the priority of the AIO operation. .TP .IR "struct sigevent aio_sigevent" This element specifies how the calling process is notified once the operation terminates. If the .IR "sigev_notify" element is .B "SIGEV_NONE" , no notification is sent. If it is .B "SIGEV_SIGNAL" , the signal determined by .IR "sigev_signo" is sent. Otherwise, .IR "sigev_notify" must be .B "SIGEV_THREAD" . In this case, a thread is created which starts executing the function pointed to by .IR "sigev_notify_function" . .TP .IR "int aio_lio_opcode" This element is only used by the .IR "lio_listio" and .IR "lio_listio64" functions. Since these functions allow an arbitrary number of operations to start at once, and each operation can be input or output (or nothing), the information must be stored in the control block. The possible values are: .TP .B "LIO_READ" Start a read operation. Read from the file at position .IR "aio_offset" and store the next .IR "aio_nbytes" bytes in the buffer pointed to by .IR "aio_buf" . .TP .B "LIO_WRITE" Start a write operation. Write .IR "aio_nbytes" bytes starting at .IR "aio_buf" into the file starting at position .IR "aio_offset" . .TP .B "LIO_NOP" Do nothing for this control block. This value is useful sometimes when an array of .IR "struct aiocb" values contains holes, i.e., some of the values must not be handled although the whole array is presented to the .IR "lio_listio" function. When the sources are compiled using .B "_FILE_OFFSET_BITS == 64" on a 32 bit machine, this type is in fact .IR "struct aiocb64" , since the LFS interface transparently replaces the .IR "struct aiocb" definition. .PP For use with the AIO functions defined in the LFS, there is a similar type defined which replaces the types of the appropriate members with larger types but otherwise is equivalent to .IR "struct aiocb" . Particularly, all member names are the same. .nf /* The same for the 64bit offsets. Please note that the members aio_fildes to __return_value have to be the same in aiocb and aiocb64. */ #ifdef __USE_LARGEFILE64 struct aiocb64 { int aio_fildes; /* File desriptor. */ int aio_lio_opcode; /* Operation to be performed. */ int aio_reqprio; /* Request priority offset. */ volatile void *aio_buf; /* Location of buffer. */ size_t aio_nbytes; /* Length of transfer. */ struct sigevent aio_sigevent; /* Signal number and value. */ /* Internal members. */ struct aiocb *__next_prio; int __abs_prio; int __policy; int __error_code; __ssize_t __return_value; __off64_t aio_offset; /* File offset. */ char __unused[32]; }; .fi .TP .IR "int aio_fildes" This element specifies the file descriptor which is used for the operation. It must be a legal descriptor since otherwise the operation fails for obvious reasons. The device on which the file is opened must allow the seek operation. I.e., it is not possible to use any of the AIO operations on devices like terminals where an .IR "lseek" call would lead to an error. .TP .IR "off64_t aio_offset" This element specifies at which offset in the file the operation (input or output) is performed. Since the operation are carried in arbitrary order and more than one operation for one file descriptor can be started, one cannot expect a current read/write position of the file descriptor. .TP .IR "volatile void *aio_buf" This is a pointer to the buffer with the data to be written or the place where the read data is stored. .TP .IR "size_t aio_nbytes" This element specifies the length of the buffer pointed to by .IR "aio_buf" . .TP .IR "int aio_reqprio" If for the platform .B "_POSIX_PRIORITIZED_IO" and .B "_POSIX_PRIORITY_SCHEDULING" are defined the AIO requests are processed based on the current scheduling priority. The .IR "aio_reqprio" element can then be used to lower the priority of the AIO operation. .TP .IR "struct sigevent aio_sigevent" This element specifies how the calling process is notified once the operation terminates. If the .IR "sigev_notify" , element is .B "SIGEV_NONE" no notification is sent. If it is .B "SIGEV_SIGNAL" , the signal determined by .IR "sigev_signo" is sent. Otherwise, .IR "sigev_notify" must be .B "SIGEV_THREAD" in which case a thread which starts executing the function pointed to by .IR "sigev_notify_function" . .TP .IR "int aio_lio_opcode" This element is only used by the .IR "lio_listio" and .IR "lio_listio64" functions. Since these functions allow an arbitrary number of operations to start at once, and since each operation can be input or output (or nothing), the information must be stored in the control block. See the description of .IR "struct aiocb" for a description of the possible values. .PP When the sources are compiled using .B "_FILE_OFFSET_BITS == 64" on a 32 bit machine, this type is available under the name .IR "struct aiocb64" , since the LFS transparently replaces the old interface. .SH "RETURN VALUES" .SH ERRORS .SH "SEE ALSO" .BR aio_cancel(3), .BR aio_cancel64(3), .BR aio_error(3), .BR aio_error64(3), .BR aio_fsync(3), .BR aio_fsync64(3), .BR aio_init(3), .BR aio_read(3), .BR aio_read64(3), .BR aio_return(3), .BR aio_return64(3), .BR aio_suspend(3), .BR aio_suspend64(3), .BR aio_write(3), .BR aio_write64(3), .BR errno(3), libaio-0.3.109/man/aio_cancel.3000066400000000000000000000064661126367703600160660ustar00rootroot00000000000000.TH aio_cancel 3 2002-09-12 "Linux 2.4" Linux AIO" .SH NAME aio_cancel - Cancel asynchronous I/O requests .SH SYNOPSYS .nf .B #include .sp .br .B #include .sp .br .BI "int aio_cancel (int fildes " , struct aiocb *aiocbp " )" .fi .SH DESCRIPTION When one or more requests are asynchronously processed, it might be useful in some situations to cancel a selected operation, e.g., if it becomes obvious that the written data is no longer accurate and would have to be overwritten soon. As an example, assume an application, which writes data in files in a situation where new incoming data would have to be written in a file which will be updated by an enqueued request. The POSIX AIO implementation provides such a function, but this function is not capable of forcing the cancellation of the request. It is up to the implementation to decide whether it is possible to cancel the operation or not. Therefore using this function is merely a hint. .B "The libaio implementation does not implement the cancel operation in the" .B "POSIX libraries". .PP The .IR aio_cancel function can be used to cancel one or more outstanding requests. If the .IR aiocbp parameter is .IR NULL , the function tries to cancel all of the outstanding requests which would process the file descriptor .IR fildes (i.e., whose .IR aio_fildes member is .IR fildes ). If .IR aiocbp is not .IR NULL , .IR aio_cancel attempts to cancel the specific request pointed to by .IR aiocbp. For requests which were successfully canceled, the normal notification about the termination of the request should take place. I.e., depending on the .IR "struct sigevent" object which controls this, nothing happens, a signal is sent or a thread is started. If the request cannot be canceled, it terminates the usual way after performing the operation. After a request is successfully canceled, a call to .IR aio_error with a reference to this request as the parameter will return .B ECANCELED and a call to .IR aio_return will return .IR -1. If the request wasn't canceled and is still running the error status is still .B EINPROGRESS. When the sources are compiled with .IR "_FILE_OFFSET_BITS == 64" , this function is in fact .IR aio_cancel64 since the LFS interface transparently replaces the normal implementation. .SH "RETURN VALUES" .TP .B AIO_CANCELED If there were requests which haven't terminated and which were successfully canceled. .TP .B AIO_NOTCANCELED If there is one or more requests left which couldn't be canceled, . In this case .IR aio_error must be used to find out which of the, perhaps multiple, requests (in .IR aiocbp is .IR NULL ) weren't successfully canceled. .TP .B AIO_ALLDONE If all requests already terminated at the time .IR aio_cancel is called the return value is . .SH ERRORS If an error occurred during the execution of .IR aio_cancel the function returns .IR -1 and sets .IR errno to one of the following values. .TP .B EBADF The file descriptor .IR fildes is not valid. .TP .B ENOSYS .IR aio_cancel is not implemented. .SH "SEE ALSO" .BR aio(3), .BR aio_cancel64(3), .BR aio_error(3), .BR aio_error64(3), .BR aio_fsync(3), .BR aio_fsync64(3), .BR aio_init(3), .BR aio_read(3), .BR aio_read64(3), .BR aio_return(3), .BR aio_return64(3), .BR aio_suspend(3), .BR aio_suspend64(3), .BR aio_write(3), .BR aio_write64(3), .BR errno(3), libaio-0.3.109/man/aio_cancel64.3000066400000000000000000000017441126367703600162320ustar00rootroot00000000000000.TH aio_cancel64 3 2002-09-12 "Linux 2.4" Linux AIO" .SH NAME aio_cancel64 \- Cancel asynchronous I/O requests .SH SYNOPSYS .nf .B #include .sp .br .B #include .sp .br .BI "int aio_cancel64 (int fildes, struct aiocb64 *aiocbp)" .fi .SH DESCRIPTION This function is similar to .IR aio_cancel with the only difference that the argument is a reference to a variable of type .IR struct aiocb64 . When the sources are compiled with .IR _FILE_OFFSET_BITS == 64 , this function is available under the name .IR aio_cancel and so transparently replaces the interface for small files on 32 bit machines. .SH "RETURN VALUES" See aio_cancel(3). .SH ERRORS See aio_cancel(3). .SH "SEE ALSO" .BR aio(3), .BR aio_cancel(3), .BR aio_error(3), .BR aio_error64(3), .BR aio_fsync(3), .BR aio_fsync64(3), .BR aio_init(3), .BR aio_read(3), .BR aio_read64(3), .BR aio_return(3), .BR aio_return64(3), .BR aio_suspend(3), .BR aio_suspend64(3), .BR aio_write(3), .BR aio_write64(3), .BR errno(3), libaio-0.3.109/man/aio_error.3000066400000000000000000000033341126367703600157610ustar00rootroot00000000000000.TH aio_error 3 2002-09-12 "Linux 2.4" Linux AIO" .SH NAME aio_error \- Getting the Status of AIO Operations .SH SYNOPSYS .nf .B #include .sp .br .B #include .sp .br .BI "int aio_error (const struct aiocb *aiocbp)" .fi .SH DESCRIPTION The function .IR aio_error determines the error state of the request described by the .IR "struct aiocb" variable pointed to by .I aiocbp . When the operation is performed truly asynchronously (as with .IR "aio_read" and .IR "aio_write" and with .IR "lio_listio" when the mode is .IR "LIO_NOWAIT" ), one sometimes needs to know whether a specific request already terminated and if so, what the result was. When the sources are compiled with .IR "_FILE_OFFSET_BITS == 64" this function is in fact .IR "aio_error64" since the LFS interface transparently replaces the normal implementation. .SH "RETURN VALUES" If the request has not yet terminated the value returned is always .IR "EINPROGRESS" . Once the request has terminated the value .IR "aio_error" returns is either .I 0 if the request completed successfully or it returns the value which would be stored in the .IR "errno" variable if the request would have been done using .IR "read" , .IR "write" , or .IR "fsync" . .SH ERRORS .TP .IR "ENOSYS" if it is not implemented. It could also return .TP .IR "EINVAL" if the .I aiocbp parameter does not refer to an asynchronous operation whose return status is not yet known. .SH "SEE ALSO" .BR aio(3), .BR aio_cancel(3), .BR aio_cancel64(3), .BR aio_error64(3), .BR aio_fsync(3), .BR aio_fsync64(3), .BR aio_init(3), .BR aio_read(3), .BR aio_read64(3), .BR aio_return(3), .BR aio_return64(3), .BR aio_suspend(3), .BR aio_suspend64(3), .BR aio_write(3), .BR aio_write64(3), .BR errno(3), libaio-0.3.109/man/aio_error64.3000066400000000000000000000024561126367703600161370ustar00rootroot00000000000000.TH aio_error64 3 2002-09-12 "Linux 2.4" Linux AIO" .SH NAME aio_error64 \- Return errors .SH SYNOPSYS .nf .B #include .sp .br .B #include .sp .br .BI "int aio_error64 (const struct aiocb64 *aiocbp)" .fi .SH DESCRIPTION This function is similar to .IR aio_error with the only difference that the argument is a reference to a variable of type .IR "struct aiocb64". .PP When the sources are compiled with .IR "_FILE_OFFSET_BITS == 64" this function is available under the name .IR aio_error and so transparently replaces the interface for small files on 32 bit machines. .SH "RETURN VALUES" If the request has not yet terminated the value returned is always .IR "EINPROGRESS" . Once the request has terminated the value .IR "aio_error" returns is either .I 0 if the request completed successfully or it returns the value which would be stored in the .IR "errno" variable if the request would have been done using .IR "read" , .IR "write" , or .IR "fsync" . .SH ERRORS See .IR aio_error(3). .SH "SEE ALSO" .BR aio(3), .BR aio_cancel(3), .BR aio_cancel64(3), .BR aio_error(3), .BR aio_fsync(3), .BR aio_fsync64(3), .BR aio_init(3), .BR aio_read(3), .BR aio_read64(3), .BR aio_return(3), .BR aio_return64(3), .BR aio_suspend(3), .BR aio_suspend64(3), .BR aio_write(3), .BR aio_write64(3), .BR errno(3), libaio-0.3.109/man/aio_fsync.3000066400000000000000000000062571126367703600157610ustar00rootroot00000000000000.TH aio_fsync 3 2002-09-12 "Linux 2.4" Linux AIO" .SH NAME aio_fsync \- Synchronize a file's complete in-core state with that on disk .SH SYNOPSYS .nf .B #include .sp .br .B #include .sp .br .BI "int aio_fsync (int op, struct aiocb aiocbp)" .fi .SH DESCRIPTION .PP When dealing with asynchronous operations it is sometimes necessary to get into a consistent state. This would mean for AIO that one wants to know whether a certain request or a group of request were processed. This could be done by waiting for the notification sent by the system after the operation terminated, but this sometimes would mean wasting resources (mainly computation time). Instead POSIX.1b defines two functions which will help with most kinds of consistency. .PP The .IR aio_fsync and .IR "aio_fsync64" functions are only available if the symbol .IR "_POSIX_SYNCHRONIZED_IO" is defined in .I unistd.h . Calling this function forces all I/O operations operating queued at the time of the function call operating on the file descriptor .IR "aiocbp->aio_fildes" into the synchronized I/O completion state . The .IR "aio_fsync" function returns immediately but the notification through the method described in .IR "aiocbp->aio_sigevent" will happen only after all requests for this file descriptor have terminated and the file is synchronized. This also means that requests for this very same file descriptor which are queued after the synchronization request are not affected. If .IR "op" is .IR "O_DSYNC" the synchronization happens as with a call to .IR "fdatasync" . Otherwise .IR "op" should be .IR "O_SYNC" and the synchronization happens as with .IR "fsync" . As long as the synchronization has not happened, a call to .IR "aio_error" with the reference to the object pointed to by .IR "aiocbp" returns .IR "EINPROGRESS" . Once the synchronization is done .IR "aio_error" return .IR 0 if the synchronization was not successful. Otherwise the value returned is the value to which the .IR "fsync" or .IR "fdatasync" function would have set the .IR "errno" variable. In this case nothing can be assumed about the consistency for the data written to this file descriptor. .SH "RETURN VALUES" The return value of this function is .IR 0 if the request was successfully enqueued. Otherwise the return value is .IR -1 and .IR "errno". .SH ERRORS .TP .B EAGAIN The request could not be enqueued due to temporary lack of resources. .TP .B EBADF The file descriptor .IR "aiocbp->aio_fildes" is not valid or not open for writing. .TP .B EINVAL The implementation does not support I/O synchronization or the .IR "op" parameter is other than .IR "O_DSYNC" and .IR "O_SYNC" . .TP .B ENOSYS This function is not implemented. .PP When the sources are compiled with .IR "_FILE_OFFSET_BITS == 64" this function is in fact .IR "aio_return64" since the LFS interface transparently replaces the normal implementation. .SH "SEE ALSO" .BR aio(3), .BR aio_cancel(3), .BR aio_cancel64(3), .BR aio_error(3), .BR aio_error64(3), .BR aio_fsync64(3), .BR aio_init(3), .BR aio_read(3), .BR aio_read64(3), .BR aio_return(3), .BR aio_return64(3), .BR aio_suspend(3), .BR aio_suspend64(3), .BR aio_write(3), .BR aio_write64(3), .BR errno(3), libaio-0.3.109/man/aio_fsync64.3000066400000000000000000000017741126367703600161320ustar00rootroot00000000000000.TH aio_fsync64 3 2002-09-12 "Linux 2.4" Linux AIO" .SH NAME aio_fsync64 \- Synchronize a file's complete in-core state with that on disk .SH SYNOPSYS .nf .B #include .sp .br .B #include .sp .br .BI "int aio_fsync64 (int op, struct aiocb64 *aiocbp)" .fi .SH DESCRIPTION This function is similar to .IR aio_fsync with the only difference that the argument is a reference to a variable of type .IR "struct aiocb64". When the sources are compiled with .IR "_FILE_OFFSET_BITS == 64" this function is available under the name .IR aio_fsync and so transparently replaces the interface for small files on 32 bit machines. .SH "RETURN VALUES" See .IR aio_fsync. .SH ERRORS See .IR aio_fsync. .SH "SEE ALSO" .BR aio(3), .BR aio_cancel(3), .BR aio_cancel64(3), .BR aio_error(3), .BR aio_error64(3), .BR aio_fsync(3), .BR aio_init(3), .BR aio_read(3), .BR aio_read64(3), .BR aio_return(3), .BR aio_return64(3), .BR aio_suspend(3), .BR aio_suspend64(3), .BR aio_write(3), .BR aio_write64(3), .BR errno(3), libaio-0.3.109/man/aio_init.3000066400000000000000000000046031126367703600155730ustar00rootroot00000000000000.TH aio_init 3 2002-09-12 "Linux 2.4" Linux AIO" .SH NAME aio_init \- How to optimize the AIO implementation .SH SYNOPSYS .nf .B #include .sp .br .B #include .sp .br .BI "void aio_init (const struct aioinit *init)" .fi .SH DESCRIPTION The POSIX standard does not specify how the AIO functions are implemented. They could be system calls, but it is also possible to emulate them at userlevel. At the point of this writing, the available implementation is a userlevel implementation which uses threads for handling the enqueued requests. While this implementation requires making some decisions about limitations, hard limitations are something which is best avoided in the GNU C library. Therefore, the GNU C library provides a means for tuning the AIO implementation according to the individual use. .BI "struct aioinit" .PP This data type is used to pass the configuration or tunable parameters to the implementation. The program has to initialize the members of this struct and pass it to the implementation using the .IR aio_init function. .TP .B "int aio_threads" This member specifies the maximal number of threads which may be used at any one time. .TP .B "int aio_num" This number provides an estimate on the maximal number of simultaneously enqueued requests. .TP .B "int aio_locks" Unused. .TP .B "int aio_usedba" Unused. .TP .B "int aio_debug" Unused. .TP .B "int aio_numusers" Unused. .TP .B "int aio_reserved[2]" Unused. .PP This function must be called before any other AIO function. Calling it is completely voluntary, as it is only meant to help the AIO implementation perform better. Before calling the .IR aio_init , function the members of a variable of type .IR "struct aioinit" must be initialized. Then a reference to this variable is passed as the parameter to .IR aio_init which itself may or may not pay attention to the hints. It is a extension which follows a proposal from the SGI implementation in .IR Irix 6 . It is not covered by POSIX.1b or Unix98. .SH "RETURN VALUES" The function has no return value. .SH ERRORS The function has no error cases defined. .SH "SEE ALSO" .BR aio(3), .BR aio_cancel(3), .BR aio_cancel64(3), .BR aio_error(3), .BR aio_error64(3), .BR aio_fsync(3), .BR aio_fsync64(3), .BR aio_read(3), .BR aio_read64(3), .BR aio_return(3), .BR aio_return64(3), .BR aio_suspend(3), .BR aio_suspend64(3), .BR aio_write(3), .BR aio_write64(3), .BR errno(3), libaio-0.3.109/man/aio_read.3000066400000000000000000000061361126367703600155460ustar00rootroot00000000000000.TH aio_read 3 2002-09-12 "Linux 2.4" Linux AIO" .SH NAME aio_read \- Initiate an asynchronous read operation .SH SYNOPSYS .nf .B #include .sp .br .B #include .sp .br .BI "int aio_read (struct aiocb *aiocbp)" .fi .SH DESCRIPTION This function initiates an asynchronous read operation. It immediately returns after the operation was enqueued or when an error was encountered. The first .IR "aiocbp->aio_nbytes" bytes of the file for which .IR "aiocbp->aio_fildes" is a descriptor are written to the buffer starting at .IR "aiocbp->aio_buf" . Reading starts at the absolute position .IR "aiocbp->aio_offset" in the file. If prioritized I/O is supported by the platform the .IR "aiocbp->aio_reqprio" value is used to adjust the priority before the request is actually enqueued. The calling process is notified about the termination of the read request according to the .IR "aiocbp->aio_sigevent" value. .SH "RETURN VALUES" When .IR "aio_read" returns, the return value is zero if no error occurred that can be found before the process is enqueued. If such an early error is found, the function returns .IR -1 and sets .IR "errno". .PP If .IR "aio_read" returns zero, the current status of the request can be queried using .IR "aio_error" and .IR "aio_return" functions. As long as the value returned by .IR "aio_error" is .IR "EINPROGRESS" the operation has not yet completed. If .IR "aio_error" returns zero, the operation successfully terminated, otherwise the value is to be interpreted as an error code. If the function terminated, the result of the operation can be obtained using a call to .IR "aio_return" . The returned value is the same as an equivalent call to .IR "read" would have returned. When the sources are compiled with .IR "_FILE_OFFSET_BITS == 64" this function is in fact .IR "aio_read64" since the LFS interface transparently replaces the normal implementation. .SH ERRORS In the case of an early error: .TP .B EAGAIN The request was not enqueued due to (temporarily) exceeded resource limitations. .TP .B ENOSYS The .IR "aio_read" function is not implemented. .TP .B EBADF The .IR "aiocbp->aio_fildes" descriptor is not valid. This condition need not be recognized before enqueueing the request and so this error might also be signaled asynchronously. .TP .B EINVAL The .IR "aiocbp->aio_offset" or .IR "aiocbp->aio_reqpiro" value is invalid. This condition need not be recognized before enqueueing the request and so this error might also be signaled asynchronously. .PP In the case of a normal return, possible error codes returned by .IR "aio_error" are: .TP .B EBADF The .IR "aiocbp->aio_fildes" descriptor is not valid. .TP .B ECANCELED The operation was canceled before the operation was finished .TP .B EINVAL The .IR "aiocbp->aio_offset" value is invalid. .PP .SH "SEE ALSO" .BR aio(3), .BR aio_cancel(3), .BR aio_cancel64(3), .BR aio_error(3), .BR aio_error64(3), .BR aio_fsync(3), .BR aio_fsync64(3), .BR aio_init(3), .BR aio_read64(3), .BR aio_return(3), .BR aio_return64(3), .BR aio_suspend(3), .BR aio_suspend64(3), .BR aio_write(3), .BR aio_write64(3), .BR errno(3), libaio-0.3.109/man/aio_read64.3000066400000000000000000000022761126367703600157210ustar00rootroot00000000000000.TH aio_read64 3 2002-09-12 "Linux 2.4" Linux AIO" .SH NAME aio_read64 \- Initiate an asynchronous read operation .SH SYNOPSYS .nf .B #include .br .B #include .sp .br .BI "int aio_read64 (struct aiocb *aiocbp)" .fi .SH DESCRIPTION This function is similar to the .IR "aio_read" function. The only difference is that on .IR "32 bit" machines, the file descriptor should be opened in the large file mode. Internally, .IR "aio_read64" uses functionality equivalent to .IR "lseek64" to position the file descriptor correctly for the reading, as opposed to .IR "lseek" functionality used in .IR "aio_read". When the sources are compiled with .IR "_FILE_OFFSET_BITS == 64" , this function is available under the name .IR "aio_read" and so transparently replaces the interface for small files on 32 bit machines. .SH "RETURN VALUES" See .IR aio_read. .SH ERRORS See .IR aio_read. .SH "SEE ALSO" .BR aio(3), .BR aio_cancel(3), .BR aio_cancel64(3), .BR aio_error(3), .BR aio_error64(3), .BR aio_fsync(3), .BR aio_fsync64(3), .BR aio_init(3), .BR aio_read(3), .BR aio_return(3), .BR aio_return64(3), .BR aio_suspend(3), .BR aio_suspend64(3), .BR aio_write(3), .BR aio_write64(3), .BR errno(3), libaio-0.3.109/man/aio_return.3000066400000000000000000000030131126367703600161410ustar00rootroot00000000000000.TH aio_return 3 2002-09-12 "Linux 2.4" Linux AIO" .SH NAME aio_return \- Retrieve status of asynchronous I/O operation .SH SYNOPSYS .nf .B #include .sp .br .B #include .sp .br .BI "ssize_t aio_return (const struct aiocb *aiocbp)" .fi .SH DESCRIPTION This function can be used to retrieve the return status of the operation carried out by the request described in the variable pointed to by .IR aiocbp . As long as the error status of this request as returned by .IR aio_error is .IR EINPROGRESS the return of this function is undefined. Once the request is finished this function can be used exactly once to retrieve the return value. Following calls might lead to undefined behavior. When the sources are compiled with .B "_FILE_OFFSET_BITS == 64" this function is in fact .IR aio_return64 since the LFS interface transparently replaces the normal implementation. .SH "RETURN VALUES" The return value itself is the value which would have been returned by the .IR read , .IR write , or .IR fsync call. .SH ERRORS The function can return .TP .B ENOSYS if it is not implemented. .TP .B EINVAL if the .IR aiocbp parameter does not refer to an asynchronous operation whose return status is not yet known. .SH "SEE ALSO" .BR aio(3), .BR aio_cancel(3), .BR aio_cancel64(3), .BR aio_error(3), .BR aio_error64(3), .BR aio_fsync(3), .BR aio_fsync64(3), .BR aio_init(3), .BR aio_read(3), .BR aio_read64(3), .BR aio_return64(3), .BR aio_suspend(3), .BR aio_suspend64(3), .BR aio_write(3), .BR aio_write64(3), .BR errno(3), libaio-0.3.109/man/aio_return64.3000066400000000000000000000017571126367703600163300ustar00rootroot00000000000000.TH aio_read64 3 2002-09-12 "Linux 2.4" Linux AIO" .SH NAME aio_read64 \- Retrieve status of asynchronous I/O operation .SH SYNOPSYS .nf .B #include .sp .br .B #include .sp .br .BI "int aio_return64 (const struct aiocb64 *aiocbp)" .fi .SH DESCRIPTION This function is similar to .IR "aio_return" with the only difference that the argument is a reference to a variable of type .IR "struct aiocb64". When the sources are compiled with .IR "_FILE_OFFSET_BITS == 64" this function is available under the name .IR "aio_return" and so transparently replaces the interface for small files on 32 bit machines. .SH "RETURN VALUES" See .IR aio_return. .SH ERRORS See .IR aio_return. .SH "SEE ALSO" .BR aio(3), .BR aio_cancel(3), .BR aio_cancel64(3), .BR aio_error(3), .BR aio_error64(3), .BR aio_fsync(3), .BR aio_fsync64(3), .BR aio_init(3), .BR aio_read(3), .BR aio_read64(3), .BR aio_return(3), .BR aio_suspend(3), .BR aio_suspend64(3), .BR aio_write(3), .BR aio_write64(3), .BR errno(3), libaio-0.3.109/man/aio_suspend.3000066400000000000000000000057201126367703600163120ustar00rootroot00000000000000.TH aio_suspend 3 2002-09-12 "Linux 2.4" Linux AIO" .SH NAME aio_suspend \- Wait until one or more requests of a specific set terminates. .SH SYNOPSYS .nf .B "#include " .sp .br .B "#include " .sp .br .BI "int aio_suspend (const struct aiocb *const list[], int nent, const struct timespec *timeout)" .fi .SH DESCRIPTION Another method of synchronization is to wait until one or more requests of a specific set terminated. This could be achieved by the .IR "aio_*" functions to notify the initiating process about the termination but in some situations this is not the ideal solution. In a program which constantly updates clients somehow connected to the server it is not always the best solution to go round robin since some connections might be slow. On the other hand letting the .IR "aio_*" function notify the caller might also be not the best solution since whenever the process works on preparing data for on client it makes no sense to be interrupted by a notification since the new client will not be handled before the current client is served. For situations like this .IR "aio_suspend" should be used. .PP When calling this function, the calling thread is suspended until at least one of the requests pointed to by the .IR "nent" elements of the array .IR "list" has completed. If any of the requests has already completed at the time .IR "aio_suspend" is called, the function returns immediately. Whether a request has terminated or not is determined by comparing the error status of the request with .IR "EINPROGRESS" . If an element of .IR "list" is .IR "NULL" , the entry is simply ignored. If no request has finished, the calling process is suspended. If .IR "timeout" is .IR "NULL" , the process is not woken until a request has finished. If .IR "timeout" is not .IR "NULL" , the process remains suspended at least as long as specified in .IR "timeout" . In this case, .IR "aio_suspend" returns with an error. .PP When the sources are compiled with .IR "_FILE_OFFSET_BITS == 64" this function is in fact .IR "aio_suspend64" since the LFS interface transparently replaces the normal implementation. .SH "RETURN VALUES" The return value of the function is .IR 0 if one or more requests from the .IR "list" have terminated. Otherwise the function returns .IR -1 and .IR "errno" is set. .SH ERRORS .TP .B EAGAIN None of the requests from the .IR "list" completed in the time specified by .IR "timeout" . .TP .B EINTR A signal interrupted the .IR "aio_suspend" function. This signal might also be sent by the AIO implementation while signalling the termination of one of the requests. .TP .B ENOSYS The .IR "aio_suspend" function is not implemented. .SH "SEE ALSO" .BR aio(3), .BR aio_cancel(3), .BR aio_cancel64(3), .BR aio_error(3), .BR aio_error64(3), .BR aio_fsync(3), .BR aio_fsync64(3), .BR aio_init(3), .BR aio_read(3), .BR aio_read64(3), .BR aio_return(3), .BR aio_return64(3), .BR aio_suspend64(3), .BR aio_write(3), .BR aio_write64(3), .BR errno(3), libaio-0.3.109/man/aio_suspend64.3000066400000000000000000000020671126367703600164650ustar00rootroot00000000000000.TH aio_suspend64 3 2002-09-12 "Linux 2.4" Linux AIO" .SH NAME aio_suspend64 \- Wait until one or more requests of a specific set terminates .SH SYNOPSYS .nf .B #include .sp .br .B #include .sp .br .BI "int aio_suspend64 (const struct aiocb64 *const list[], int nent, const struct timespec *timeout)" .fi .SH DESCRIPTION This function is similar to .IR "aio_suspend" with the only difference that the argument is a reference to a variable of type .IR "struct aiocb64". When the sources are compiled with .IR "_FILE_OFFSET_BITS == 64" this function is available under the name .IR "aio_suspend" and so transparently replaces the interface for small files on 32 bit machines. .SH "RETURN VALUES" See .IR aio_suspend. .SH ERRORS See .IR aio_suspend. .SH "SEE ALSO" .BR aio(3), .BR aio_cancel(3), .BR aio_cancel64(3), .BR aio_error(3), .BR aio_error64(3), .BR aio_fsync(3), .BR aio_fsync64(3), .BR aio_init(3), .BR aio_read(3), .BR aio_read64(3), .BR aio_return(3), .BR aio_return64(3), .BR aio_suspend(3), .BR aio_write(3), .BR aio_write64(3), .BR errno(3), libaio-0.3.109/man/aio_write.3000066400000000000000000000076111126367703600157640ustar00rootroot00000000000000.TH aio_write 3 2002-09-12 "Linux 2.4" Linux AIO" .SH NAME aio_write \- Initiate an asynchronous write operation .SH SYNOPSYS .nf .B #include .sp .br .B #include .sp .br .BI "int aio_write (struct aiocb * aiocbp);" .fi .SH DESCRIPTION This function initiates an asynchronous write operation. The function call immediately returns after the operation was enqueued or if before this happens an error was encountered. The first .IR "aiocbp->aio_nbytes" bytes from the buffer starting at .IR "aiocbp->aio_buf" are written to the file for which .IR "aiocbp->aio_fildes" is an descriptor, starting at the absolute position .IR "aiocbp->aio_offset" in the file. If prioritized I/O is supported by the platform, the .IR "aiocbp->aio_reqprio " value is used to adjust the priority before the request is actually enqueued. The calling process is notified about the termination of the read request according to the .IR "aiocbp->aio_sigevent" value. When .IR "aio_write" returns, the return value is zero if no error occurred that can be found before the process is enqueued. If such an early error is found the function returns .IR -1 and sets .IR "errno" to one of the following values. .TP .B EAGAIN The request was not enqueued due to (temporarily) exceeded resource limitations. .TP .B ENOSYS The .IR "aio_write" function is not implemented. .TP .B EBADF The .IR "aiocbp->aio_fildes" descriptor is not valid. This condition may not be recognized before enqueueing the request, and so this error might also be signaled asynchronously. .TP .B EINVAL The .IR "aiocbp->aio_offset" or .IR "aiocbp->aio_reqprio" value is invalid. This condition may not be recognized before enqueueing the request and so this error might also be signaled asynchronously. .PP In the case .IR "aio_write" returns zero, the current status of the request can be queried using .IR "aio_error" and .IR "aio_return" functions. As long as the value returned by .IR "aio_error" is .IR "EINPROGRESS" the operation has not yet completed. If .IR "aio_error" returns zero, the operation successfully terminated, otherwise the value is to be interpreted as an error code. If the function terminated, the result of the operation can be get using a call to .IR "aio_return" . The returned value is the same as an equivalent call to .IR "read" would have returned. Possible error codes returned by .IR "aio_error" are: .TP .B EBADF The .IR "aiocbp->aio_fildes" descriptor is not valid. .TP .B ECANCELED The operation was canceled before the operation was finished. .TP .B EINVAL The .IR "aiocbp->aio_offset" value is invalid. .PP When the sources are compiled with .IR "_FILE_OFFSET_BITS == 64" , this function is in fact .IR "aio_write64" since the LFS interface transparently replaces the normal implementation. .SH "RETURN VALUES" When .IR "aio_write" returns, the return value is zero if no error occurred that can be found before the process is enqueued. If such an early error is found the function returns .IR -1 and sets .IR "errno" to one of the following values. .SH ERRORS .TP .B EAGAIN The request was not enqueued due to (temporarily) exceeded resource limitations. .TP .B ENOSYS The .IR "aio_write" function is not implemented. .TP .B EBADF The .IR "aiocbp->aio_fildes" descriptor is not valid. This condition may not be recognized before enqueueing the request, and so this error might also be signaled asynchronously. .TP .B EINVAL The .IR "aiocbp->aio_offset" or .IR "aiocbp->aio_reqprio" value is invalid. This condition may not be recognized before enqueueing the request and so this error might also be signaled asynchronously. .SH "SEE ALSO" .BR aio(3), .BR aio_cancel(3), .BR aio_cancel64(3), .BR aio_error(3), .BR aio_error64(3), .BR aio_fsync(3), .BR aio_fsync64(3), .BR aio_init(3), .BR aio_read(3), .BR aio_read64(3), .BR aio_return(3), .BR aio_return64(3), .BR aio_suspend(3), .BR aio_suspend64(3), .BR aio_write64(3), .BR errno(3), libaio-0.3.109/man/aio_write64.3000066400000000000000000000023121126367703600161270ustar00rootroot00000000000000.TH aio_write64 3 2002-09-12 "Linux 2.4" Linux AIO" .SH NAME aio_write64 \- Initiate an asynchronous write operation .SH SYNOPSYS .nf .B #include .sp .br .B #include .sp .br .BI "int aio_write64 (struct aiocb *aiocbp)" .fi .SH DESCRIPTION This function is similar to the .IR "aio_write" function. The only difference is that on .IR "32 bit" machines the file descriptor should be opened in the large file mode. Internally .IR "aio_write64" uses functionality equivalent to .IR "lseek64" to position the file descriptor correctly for the writing, as opposed to .IR "lseek" functionality used in .IR "aio_write". When the sources are compiled with .IR "_FILE_OFFSET_BITS == 64" , this function is available under the name .IR "aio_write" and so transparently replaces the interface for small files on 32 bit machines. .SH "RETURN VALUES" See .IR aio_write. .SH ERRORS See .IR aio_write. .SH "SEE ALSO" .BR aio(3), .BR aio_cancel(3), .BR aio_cancel64(3), .BR aio_error(3), .BR aio_error64(3), .BR aio_fsync(3), .BR aio_fsync64(3), .BR aio_init(3), .BR aio_read(3), .BR aio_read64(3), .BR aio_return(3), .BR aio_return64(3), .BR aio_suspend(3), .BR aio_suspend64(3), .BR aio_write(3), .BR errno(3), libaio-0.3.109/man/io.3000066400000000000000000000212611126367703600144060ustar00rootroot00000000000000.TH io 3 2002-09-12 "Linux 2.4" Linux IO" .SH NAME io \- Asynchronous IO .SH SYNOPSYS .nf .B #include .sp .br .B #include .sp .fi .SH DESCRIPTION The libaio library defines a new set of I/O operations which can significantly reduce the time an application spends waiting at I/O. The new functions allow a program to initiate one or more I/O operations and then immediately resume normal work while the I/O operations are executed in parallel. These functions are part of the library with realtime functions named .IR "libaio" . They are not actually part of the .IR "libc" binary. The implementation of these functions can be done using support in the kernel. All IO operations operate on files which were opened previously. There might be arbitrarily many operations running for one file. The asynchronous I/O operations are controlled using a data structure named .IR "struct iocb" It is defined in .IR "libio.h" as follows. .nf typedef struct io_context *io_context_t; typedef enum io_iocb_cmd { IO_CMD_PREAD = 0, IO_CMD_PWRITE = 1, IO_CMD_FSYNC = 2, IO_CMD_FDSYNC = 3, IO_CMD_POLL = 5, IO_CMD_NOOP = 6, } io_iocb_cmd_t; struct io_iocb_common { void *buf; unsigned __pad1; long nbytes; unsigned __pad2; long long offset; long long __pad3, __pad4; }; /* result code is the amount read or -'ve errno */ struct iocb { void *data; unsigned key; short aio_lio_opcode; short aio_reqprio; int aio_fildes; union { struct io_iocb_common c; struct io_iocb_vector v; struct io_iocb_poll poll; struct io_iocb_sockaddr saddr; } u; }; .fi .TP .IR "int aio_fildes" This element specifies the file descriptor to be used for the operation. It must be a legal descriptor, otherwise the operation will fail. The device on which the file is opened must allow the seek operation. I.e., it is not possible to use any of the IO operations on devices like terminals where an .IR "lseek" call would lead to an error. .TP .IR "long u.c.offset" This element specifies the offset in the file at which the operation (input or output) is performed. Since the operations are carried out in arbitrary order and more than one operation for one file descriptor can be started, one cannot expect a current read/write position of the file descriptor. .TP .IR "void *buf" This is a pointer to the buffer with the data to be written or the place where the read data is stored. .TP .IR "long u.c.nbytes" This element specifies the length of the buffer pointed to by .IR "io_buf" . .TP .IR "int aio_reqprio" Is not currently used. .TP .B "IO_CMD_PREAD" Start a read operation. Read from the file at position .IR "u.c.offset" and store the next .IR "u.c.nbytes" bytes in the buffer pointed to by .IR "buf" . .TP .B "IO_CMD_PWRITE" Start a write operation. Write .IR "u.c.nbytes" bytes starting at .IR "buf" into the file starting at position .IR "u.c.offset" . .TP .B "IO_CMD_NOP" Do nothing for this control block. This value is useful sometimes when an array of .IR "struct iocb" values contains holes, i.e., some of the values must not be handled although the whole array is presented to the .IR "io_submit" function. .TP .B "IO_CMD_FSYNC" .TP .B "IO_CMD_POLL" This is experimental. .SH EXAMPLE .nf /* * Simplistic version of copy command using async i/o * * From: Stephen Hemminger * Copy file by using a async I/O state machine. * 1. Start read request * 2. When read completes turn it into a write request * 3. When write completes decrement counter and free resources * * * Usage: aiocp file(s) desination */ #include #include #include #include #include #include #include #include #define AIO_BLKSIZE (64*1024) #define AIO_MAXIO 32 static int busy = 0; // # of I/O's in flight static int tocopy = 0; // # of blocks left to copy static int dstfd = -1; // destination file descriptor static const char *dstname = NULL; static const char *srcname = NULL; /* Fatal error handler */ static void io_error(const char *func, int rc) { if (rc == -ENOSYS) fprintf(stderr, "AIO not in this kernel\n"); else if (rc < 0 && -rc < sys_nerr) fprintf(stderr, "%s: %s\n", func, sys_errlist[-rc]); else fprintf(stderr, "%s: error %d\n", func, rc); if (dstfd > 0) close(dstfd); if (dstname) unlink(dstname); exit(1); } /* * Write complete callback. * Adjust counts and free resources */ static void wr_done(io_context_t ctx, struct iocb *iocb, long res, long res2) { if (res2 != 0) { io_error("aio write", res2); } if (res != iocb->u.c.nbytes) { fprintf(stderr, "write missed bytes expect %d got %d\n", iocb->u.c.nbytes, res2); exit(1); } --tocopy; --busy; free(iocb->u.c.buf); memset(iocb, 0xff, sizeof(iocb)); // paranoia free(iocb); write(2, "w", 1); } /* * Read complete callback. * Change read iocb into a write iocb and start it. */ static void rd_done(io_context_t ctx, struct iocb *iocb, long res, long res2) { /* library needs accessors to look at iocb? */ int iosize = iocb->u.c.nbytes; char *buf = iocb->u.c.buf; off_t offset = iocb->u.c.offset; if (res2 != 0) io_error("aio read", res2); if (res != iosize) { fprintf(stderr, "read missing bytes expect %d got %d\n", iocb->u.c.nbytes, res); exit(1); } /* turn read into write */ io_prep_pwrite(iocb, dstfd, buf, iosize, offset); io_set_callback(iocb, wr_done); if (1 != (res = io_submit(ctx, 1, &iocb))) io_error("io_submit write", res); write(2, "r", 1); } int main(int argc, char *const *argv) { int srcfd; struct stat st; off_t length = 0, offset = 0; io_context_t myctx; if (argc != 3 || argv[1][0] == '-') { fprintf(stderr, "Usage: aiocp SOURCE DEST"); exit(1); } if ((srcfd = open(srcname = argv[1], O_RDONLY)) < 0) { perror(srcname); exit(1); } if (fstat(srcfd, &st) < 0) { perror("fstat"); exit(1); } length = st.st_size; if ((dstfd = open(dstname = argv[2], O_WRONLY | O_CREAT, 0666)) < 0) { close(srcfd); perror(dstname); exit(1); } /* initialize state machine */ memset(&myctx, 0, sizeof(myctx)); io_queue_init(AIO_MAXIO, &myctx); tocopy = howmany(length, AIO_BLKSIZE); while (tocopy > 0) { int i, rc; /* Submit as many reads as once as possible upto AIO_MAXIO */ int n = MIN(MIN(AIO_MAXIO - busy, AIO_MAXIO / 2), howmany(length - offset, AIO_BLKSIZE)); if (n > 0) { struct iocb *ioq[n]; for (i = 0; i < n; i++) { struct iocb *io = (struct iocb *) malloc(sizeof(struct iocb)); int iosize = MIN(length - offset, AIO_BLKSIZE); char *buf = (char *) malloc(iosize); if (NULL == buf || NULL == io) { fprintf(stderr, "out of memory\n"); exit(1); } io_prep_pread(io, srcfd, buf, iosize, offset); io_set_callback(io, rd_done); ioq[i] = io; offset += iosize; } rc = io_submit(myctx, n, ioq); if (rc < 0) io_error("io_submit", rc); busy += n; } // Handle IO's that have completed rc = io_queue_run(myctx); if (rc < 0) io_error("io_queue_run", rc); // if we have maximum number of i/o's in flight // then wait for one to complete if (busy == AIO_MAXIO) { rc = io_queue_wait(myctx, NULL); if (rc < 0) io_error("io_queue_wait", rc); } } close(srcfd); close(dstfd); exit(0); } /* * Results look like: * [alanm@toolbox ~/MOT3]$ ../taio kernel-source-2.4.8-0.4g.ppc.rpm abc * rrrrrrrrrrrrrrrwwwrwrrwwrrwrwwrrwrwrwwrrwrwrrrrwwrwwwrrwrrrwwwwwwwwwwwwwwwww * rrrrrrrrrrrrrrwwwrrwrwrwrwrrwwwwwwwwwwwwwwrrrrrrrrrrrrrrrrrrwwwwrwrwwrwrwrwr * wrrrrrrrwwwwwwwwwwwwwrrrwrrrwrrwrwwwwwwwwwwrrrrwwrwrrrrrrrrrrrwwwwwwwwwwwrww * wwwrrrrrrrrwwrrrwwrwrwrwwwrrrrrrrwwwrrwwwrrwrwwwwwwwwrrrrrrrwwwrrrrrrrwwwwww * wwwwwwwrwrrrrrrrrwrrwrrwrrwrwrrrwrrrwrrrwrwwwwwwwwwwwwwwwwwwrrrwwwrrrrrrrrrr * rrwrrrrrrwrrwwwwwwwwwwwwwwwwrwwwrrwrwwrrrrrrrrrrrrrrrrrrrwwwwwwwwwwwwwwwwwww * rrrrrwrrwrwrwrrwrrrwwwwwwwwrrrrwrrrwrwwrwrrrwrrwrrrrwwwwwwwrwrwwwwrwwrrrwrrr * rrrwwwwwwwrrrrwwrrrrrrrrrrrrwrwrrrrwwwwwwwwwwwwwwrwrrrrwwwwrwrrrrwrwwwrrrwww * rwwrrrrrrrwrrrrrrrrrrrrwwwwrrrwwwrwrrwwwwwwwwwwwwwwwwwwwwwrrrrrrrwwwwwwwrw */ .fi .SH "SEE ALSO" .BR io_cancel(3), .BR io_fsync(3), .BR io_getevents(3), .BR io_prep_fsync(3), .BR io_prep_pread(3), .BR io_prep_pwrite(3), .BR io_queue_init(3), .BR io_queue_release(3), .BR io_queue_run(3), .BR io_queue_wait(3), .BR io_set_callback(3), .BR io_submit(3), .BR errno(3) libaio-0.3.109/man/io_cancel.1000066400000000000000000000015411126367703600157100ustar00rootroot00000000000000.\"/* sys_io_cancel: .\" * Attempts to cancel an iocb previously passed to io_submit. If .\" * the operation is successfully cancelled, the resulting event is .\" * copied into the memory pointed to by result without being placed .\" * into the completion queue and 0 is returned. May fail with .\" * -EFAULT if any of the data structures pointed to are invalid. .\" * May fail with -EINVAL if aio_context specified by ctx_id is .\" * invalid. May fail with -EAGAIN if the iocb specified was not .\" * cancelled. Will fail with -ENOSYS if not implemented. .\" */ .\" .TH io_cancel 2 2002-09-03 "Linux 2.4" "Linux AIO" .SH NAME io_cancel \- cancel io requests .SH SYNOPSIS .B #include .br .B #include .LP .BI "int io_submit(io_context_t " ctx ", struct iocb *" iocb ", struct io_event *" result ");" libaio-0.3.109/man/io_cancel.3000066400000000000000000000032331126367703600157120ustar00rootroot00000000000000.TH io_cancel 2 2002-09-03 "Linux 2.4" "Linux AIO" .SH NAME io_cancel \- Cancel io requests .SH SYNOPSIS .nf .B #include .sp .br .B #include .sp .br .BI "int io_cancel(io_context_t ctx, struct iocb *iocb)" .br .sp struct iocb { void *data; /* Return in the io completion event */ unsigned key; /* For use in identifying io requests */ short aio_lio_opcode; short aio_reqprio; /* Not used */ int aio_fildes; }; .fi .SH DESCRIPTION Attempts to cancel an iocb previously passed to io_submit. If the operation is successfully cancelled, the resulting event is copied into the memory pointed to by result without being placed into the completion queue. .PP When one or more requests are asynchronously processed, it might be useful in some situations to cancel a selected operation, e.g., if it becomes obvious that the written data is no longer accurate and would have to be overwritten soon. As an example, assume an application, which writes data in files in a situation where new incoming data would have to be written in a file which will be updated by an enqueued request. .SH "RETURN VALUES" 0 is returned on success , otherwise returns Errno. .SH ERRORS .TP .B EFAULT If any of the data structures pointed to are invalid. .TP .B EINVAL If aio_context specified by ctx_id is invalid. .TP .B EAGAIN If the iocb specified was not cancelled. .TP .B ENOSYS if not implemented. .SH "SEE ALSO" .BR io(3), .BR io_fsync(3), .BR io_getevents(3), .BR io_prep_fsync(3), .BR io_prep_pread(3), .BR io_prep_pwrite(3), .BR io_queue_init(3), .BR io_queue_release(3), .BR io_queue_run(3), .BR io_queue_wait(3), .BR io_set_callback(3), .BR io_submit(3), .BR errno(3) libaio-0.3.109/man/io_destroy.1000066400000000000000000000010161126367703600161510ustar00rootroot00000000000000.\"/* sys_io_destroy: .\" * Destroy the aio_context specified. May cancel any outstanding .\" * AIOs and block on completion. Will fail with -ENOSYS if not .\" * implemented. May fail with -EFAULT if the context pointed to .\" * is invalid. .\" */ .\" libaio provides this as io_queue_release. .TH io_destroy 2 2002-09-03 "Linux 2.4" "Linux AIO" .SH NAME io_destroy \- destroy an io context .SH SYNOPSIS .B #include .br .B #include .LP .BI "int io_destroy(io_context_t " ctx ");" libaio-0.3.109/man/io_fsync.3000066400000000000000000000045211126367703600156100ustar00rootroot00000000000000./" static inline int io_fsync(io_context_t ctx, struct iocb *iocb, io_callback_t cb, int fd) ./" { ./" io_prep_fsync(iocb, fd); ./" io_set_callback(iocb, cb); ./" return io_submit(ctx, 1, &iocb); ./" } .TH io_fsync 3 2002-09-12 "Linux 2.4" Linux AIO" .SH NAME io_fsync \- Synchronize a file's complete in-core state with that on disk .SH SYNOPSYS .nf .B #include .sp .br .B #include .sp .br .BI "int io_fsync(io_context_t ctx, struct iocb *iocb, io_callback_t cb, int fd)" .sp struct iocb { void *data; unsigned key; short aio_lio_opcode; short aio_reqprio; int aio_fildes; }; .sp typedef void (*io_callback_t)(io_context_t ctx, struct iocb *iocb, long res, long res2); .sp .fi .SH DESCRIPTION When dealing with asynchronous operations it is sometimes necessary to get into a consistent state. This would mean for AIO that one wants to know whether a certain request or a group of request were processed. This could be done by waiting for the notification sent by the system after the operation terminated, but this sometimes would mean wasting resources (mainly computation time). .PP Calling this function forces all I/O operations operating queued at the time of the function call operating on the file descriptor .IR "iocb->io_fildes" into the synchronized I/O completion state . The .IR "io_fsync" function returns immediately but the notification through the method described in .IR "io_callback" will happen only after all requests for this file descriptor have terminated and the file is synchronized. This also means that requests for this very same file descriptor which are queued after the synchronization request are not affected. .SH "RETURN VALUES" Returns 0, otherwise returns errno. .SH ERRORS .TP .B EFAULT .I iocbs referenced data outside of the program's accessible address space. .TP .B EINVAL .I ctx refers to an unitialized aio context, the iocb pointed to by .I iocbs contains an improperly initialized iocb, .TP .B EBADF The iocb contains a file descriptor that does not exist. .TP .B EINVAL The file specified in the iocb does not support the given io operation. .SH "SEE ALSO" .BR io(3), .BR io_cancel(3), .BR io_getevents(3), .BR io_prep_pread(3), .BR io_prep_pwrite(3), .BR io_queue_init(3), .BR io_queue_release(3), .BR io_queue_run(3), .BR io_queue_wait(3), .BR io_set_callback(3), .BR io_submit(3), .BR errno(3) libaio-0.3.109/man/io_getevents.1000066400000000000000000000025211126367703600164660ustar00rootroot00000000000000./"/* io_getevents: ./" * Attempts to read at least min_nr events and up to nr events from ./" * the completion queue for the aio_context specified by ctx_id. May ./" * fail with -EINVAL if ctx_id is invalid, if min_nr is out of range, ./" * if nr is out of range, if when is out of range. May fail with ./" * -EFAULT if any of the memory specified to is invalid. May return ./" * 0 or < min_nr if no events are available and the timeout specified ./" * by when has elapsed, where when == NULL specifies an infinite ./" * timeout. Note that the timeout pointed to by when is relative and ./" * will be updated if not NULL and the operation blocks. Will fail ./" * with -ENOSYS if not implemented. ./" */ ./"asmlinkage long sys_io_getevents(io_context_t ctx_id, ./" long min_nr, ./" long nr, ./" struct io_event *events, ./" struct timespec *timeout) ./" .TH io_getevents 2 2002-09-03 "Linux 2.4" "Linux AIO" .SH NAME io_getevents \- read resulting events from io requests .SH SYNOPSIS .B #include .br .B #include .sp .BI "int io_getevents(io_context_t " ctx ", long " min_nr ", long " nr ", struct io_events *" events "[], struct timespec *" timeout ");" libaio-0.3.109/man/io_getevents.3000066400000000000000000000050121126367703600164660ustar00rootroot00000000000000./"/* io_getevents: ./" * Attempts to read at least min_nr events and up to nr events from ./" * the completion queue for the aio_context specified by ctx_id. May ./" * fail with -EINVAL if ctx_id is invalid, if min_nr is out of range, ./" * if nr is out of range, if when is out of range. May fail with ./" * -EFAULT if any of the memory specified to is invalid. May return ./" * 0 or < min_nr if no events are available and the timeout specified ./" * by when has elapsed, where when == NULL specifies an infinite ./" * timeout. Note that the timeout pointed to by when is relative and ./" * will be updated if not NULL and the operation blocks. Will fail ./" * with -ENOSYS if not implemented. ./" */ ./"asmlinkage long sys_io_getevents(io_context_t ctx_id, ./" long min_nr, ./" long nr, ./" struct io_event *events, ./" struct timespec *timeout) ./" .TH io_getevents 2 2002-09-03 "Linux 2.4" "Linux AIO" .SH NAME io_getevents \- Read resulting events from io requests .SH SYNOPSIS .nf .B #include .sp .br .B #include .br .sp struct iocb { void *data; unsigned key; short aio_lio_opcode; short aio_reqprio; int aio_fildes; }; .sp struct io_event { unsigned PADDED(data, __pad1); unsigned PADDED(obj, __pad2); unsigned PADDED(res, __pad3); unsigned PADDED(res2, __pad4); }; .sp .BI "int io_getevents(io_context_t " ctx ", long " nr ", struct io_event *" events "[], struct timespec *" timeout ");" .fi .SH DESCRIPTION Attempts to read up to nr events from the completion queue for the aio_context specified by ctx. .SH "RETURN VALUES" May return 0 if no events are available and the timeout specified by when has elapsed, where when == NULL specifies an infinite timeout. Note that the timeout pointed to by when is relative and will be updated if not NULL and the operation blocks. Will fail with ENOSYS if not implemented. .SH ERRORS .TP .B EINVAL if ctx_id is invalid, if min_nr is out of range, if nr is out of range, if when is out of range. .TP .B EFAULT if any of the memory specified to is invalid. .SH "SEE ALSO" .BR io(3), .BR io_cancel(3), .BR io_fsync(3), .BR io_prep_fsync(3), .BR io_prep_pread(3), .BR io_prep_pwrite(3), .BR io_queue_init(3), .BR io_queue_release(3), .BR io_queue_run(3), .BR io_queue_wait(3), .BR io_set_callback(3), .BR io_submit(3), .BR errno(3) libaio-0.3.109/man/io_prep_fsync.3000066400000000000000000000046511126367703600166420ustar00rootroot00000000000000./" static inline void io_prep_fsync(struct iocb *iocb, int fd) ./" { ./" memset(iocb, 0, sizeof(*iocb)); ./" iocb->aio_fildes = fd; ./" iocb->aio_lio_opcode = IO_CMD_FSYNC; ./" iocb->aio_reqprio = 0; ./" } .TH io_prep_fsync 3 2002-09-12 "Linux 2.4" Linux AIO" .SH NAME io_prep_fsync \- Synchronize a file's complete in-core state with that on disk .SH SYNOPSYS .nf .B #include .br .sp .B #include .br .sp .BI "static inline void io_prep_fsync(struct iocb *iocb, int fd)" .sp struct iocb { void *data; unsigned key; short aio_lio_opcode; short aio_reqprio; int aio_fildes; }; .sp .fi .SH DESCRIPTION This is an inline convenience function for setting up an iocbv for a FSYNC request. .br The file for which .TP .IR "iocb->aio_fildes = fd" is a descriptor is set up with the command .TP .IR "iocb->aio_lio_opcode = IO_CMD_FSYNC: . .PP The io_prep_fsync() function shall set up an IO_CMD_FSYNC operation to asynchronously force all I/O operations associated with the file indicated by the file descriptor aio_fildes member of the iocb structure referenced by the iocb argument and queued at the time of the call to io_submit() to the synchronized I/O completion state. The function call shall return when the synchronization request has been initiated or queued to the file or device (even when the data cannot be synchronized immediately). All currently queued I/O operations shall be completed as if by a call to fsync(); that is, as defined for synchronized I/O file integrity completion. If the operation queued by io_prep_fsync() fails, then, as for fsync(), outstanding I/O operations are not guaranteed to have been completed. If io_prep_fsync() succeeds, then it is only the I/O that was queued at the time of the call to io_submit() that is guaranteed to be forced to the relevant completion state. The completion of subsequent I/O on the file descriptor is not guaranteed to be completed in a synchronized fashion. .PP This function returns immediately . To schedule the operation, the function .IR io_submit must be called. .PP Simultaneous asynchronous operations using the same iocb produce undefined results. .SH "RETURN VALUES" None .SH ERRORS None .SH "SEE ALSO" .BR io(3), .BR io_cancel(3), .BR io_fsync(3), .BR io_getevents(3), .BR io_prep_pread(3), .BR io_prep_pwrite(3), .BR io_queue_init(3), .BR io_queue_release(3), .BR io_queue_run(3), .BR io_queue_wait(3), .BR io_set_callback(3), .BR io_submit(3), .BR errno(3) libaio-0.3.109/man/io_prep_pread.3000066400000000000000000000032451126367703600166110ustar00rootroot00000000000000./" static inline void io_prep_pread(struct iocb *iocb, int fd, void *buf, size_t count, long long offset) ./" { ./" memset(iocb, 0, sizeof(*iocb)); ./" iocb->aio_fildes = fd; ./" iocb->aio_lio_opcode = IO_CMD_PREAD; ./" iocb->aio_reqprio = 0; ./" iocb->u.c.buf = buf; ./" iocb->u.c.nbytes = count; ./" iocb->u.c.offset = offset; ./" } .TH io_prep_pread 3 2002-09-12 "Linux 2.4" Linux AIO" .SH NAME io_prep_pread \- Set up asynchronous read .SH SYNOPSYS .nf .B #include .sp .br .B #include .br .sp .BI "inline void io_prep_pread(struct iocb *iocb, int fd, void *buf, size_t count, long long offset) " .sp struct iocb { void *data; unsigned key; short aio_lio_opcode; short aio_reqprio; int aio_fildes; }; .fi .SH DESCRIPTION .IR io_prep_pread is an inline convenience function designed to facilitate the initialization of the iocb for an asynchronous read operation. The first .TP .IR "iocb->u.c.nbytes = count" bytes of the file for which .TP .IR "iocb->aio_fildes = fd" is a descriptor are written to the buffer starting at .TP .IR "iocb->u.c.buf = buf" . .br Reading starts at the absolute position .TP .IR "ioc->u.c.offset = offset" in the file. .PP This function returns immediately . To schedule the operation, the function .IR io_submit must be called. .PP Simultaneous asynchronous operations using the same iocb produce undefined results. .SH "RETURN VALUES" None .SH ERRORS None .SH "SEE ALSO" .BR io(3), .BR io_cancel(3), .BR io_fsync(3), .BR io_getevents(3), .BR io_prep_fsync(3), .BR io_prep_pwrite(3), .BR io_queue_init(3), .BR io_queue_release(3), .BR io_queue_run(3), .BR io_queue_wait(3), .BR io_set_callback(3), .BR io_submit(3), .BR errno(3) libaio-0.3.109/man/io_prep_pwrite.3000066400000000000000000000031571126367703600170320ustar00rootroot00000000000000./" static inline void io_prep_pwrite(struct iocb *iocb, int fd, void *buf, size_t count, long long offset) ./" { ./" memset(iocb, 0, sizeof(*iocb)); ./" iocb->aio_fildes = fd; ./" iocb->aio_lio_opcode = IO_CMD_PWRITE; ./" iocb->aio_reqprio = 0; ./" iocb->u.c.buf = buf; ./" iocb->u.c.nbytes = count; ./" iocb->u.c.offset = offset; ./" } .TH io_prep_pwrite 3 2002-09-12 "Linux 2.4" Linux AIO" .SH NAME io_prep_pwrite \- Set up iocb for asynchronous writes .SH SYNOPSYS .nf .B #include .br .sp .B #include .br .sp .BI "inline void io_prep_pwrite(struct iocb *iocb, int fd, void *buf, size_t count, long long offset) " .sp struct iocb { void *data; unsigned key; short aio_lio_opcode; short aio_reqprio; int aio_fildes; }; .fi .SH DESCRIPTION io_prep_write is a convenicence function for setting up parallel writes. The first .TP .IR "iocb->u.c.nbytes = count" bytes of the file for which .TP .IR "iocb->aio_fildes = fd" is a descriptor are written from the buffer starting at .TP .IR "iocb->u.c.buf = buf" . .br Writing starts at the absolute position .TP .IR "ioc->u.c.offset = offset" in the file. .PP This function returns immediately . To schedule the operation, the function .IR io_submit must be called. .PP Simultaneous asynchronous operations using the same iocb produce undefined results. .SH "RETURN VALUES" None .SH ERRORS None .SH "SEE ALSO" .BR io(3), .BR io_cancel(3), .BR io_fsync(3), .BR io_getevents(3), .BR io_prep_fsync(3), .BR io_prep_pread(3), .BR io_queue_init(3), .BR io_queue_release(3), .BR io_queue_run(3), .BR io_queue_wait(3), .BR io_set_callback(3), .BR io_submit(3), .BR errno(3) libaio-0.3.109/man/io_queue_init.3000066400000000000000000000024241126367703600166350ustar00rootroot00000000000000.TH io_queue_init 2 2002-09-03 "Linux 2.4" "Linux AIO" .SH NAME io_queue_init \- Initialize asynchronous io state machine .SH SYNOPSIS .nf .B #include .br .sp .B #include .br .sp .BI "int io_queue_init(int maxevents, io_context_t *ctx );" .sp .fi .SH DESCRIPTION .B io_queue_init Attempts to create an aio context capable of receiving at least .IR maxevents events. .IR ctx must point to an aio context that already exists and must be initialized to .IR 0 before the call. If the operation is successful, *cxtp is filled with the resulting handle. .SH "RETURN VALUES" On success, .B io_queue_init returns 0. Otherwise, -error is return, where error is one of the Exxx values defined in the Errors section. .SH ERRORS .TP .B EFAULT .I iocbs referenced data outside of the program's accessible address space. .TP .B EINVAL .I maxevents is <= 0 or .IR ctx is an invalid memory locattion. .TP .B ENOSYS Not implemented .TP .B EAGAIN .IR "maxevents > max_aio_reqs" where max_aio_reqs is a tunable value. .SH "SEE ALSO" .BR io(3), .BR io_cancel(3), .BR io_fsync(3), .BR io_getevents(3), .BR io_prep_fsync(3), .BR io_prep_pread(3), .BR io_prep_pwrite(3), .BR io_queue_release(3), .BR io_queue_run(3), .BR io_queue_wait(3), .BR io_set_callback(3), .BR io_submit(3), .BR errno(3) libaio-0.3.109/man/io_queue_release.3000066400000000000000000000020171126367703600173100ustar00rootroot00000000000000.TH io_queue_release 2 2002-09-03 "Linux 2.4" "Linux AIO" .SH NAME io_queue_release \- Release the context associated with the userspace handle .SH SYNOPSIS .nf .B #include .br .B #include .br .sp .BI "int io_queue_release(io_context_t ctx)" .sp .SH DESCRIPTION .B io_queue_release destroys the context associated with the userspace handle. May cancel any outstanding AIOs and block on completion. .B cts. .SH "RETURN VALUES" On success, .B io_queue_release returns 0. Otherwise, -error is return, where error is one of the Exxx values defined in the Errors section. .SH ERRORS .TP .B EINVAL .I ctx refers to an unitialized aio context, the iocb pointed to by .I iocbs contains an improperly initialized iocb, .TP .B ENOSYS Not implemented .SH "SEE ALSO" .BR io(3), .BR io_cancel(3), .BR io_fsync(3), .BR io_getevents(3), .BR io_prep_fsync(3), .BR io_prep_pread(3), .BR io_prep_pwrite(3), .BR io_queue_init(3), .BR io_queue_run(3), .BR io_queue_wait(3), .BR io_set_callback(3), .BR io_submit(3), .BR errno(3) libaio-0.3.109/man/io_queue_run.3000066400000000000000000000017741126367703600165050ustar00rootroot00000000000000.TH io_queue_run 2 2002-09-03 "Linux 2.4" "Linux AIO" .SH NAME io_queue_run \- Handle completed io requests .SH SYNOPSIS .nf .B #include .br .sp .B #include .br .sp .BI "int io_queue_run(io_context_t ctx );" .sp .fi .SH DESCRIPTION .B io_queue_run Attempts to read all the events events from the completion queue for the aio_context specified by ctx_id. .SH "RETURN VALUES" May return 0 if no events are available. Will fail with -ENOSYS if not implemented. .SH ERRORS .TP .B EFAULT .I iocbs referenced data outside of the program's accessible address space. .TP .B EINVAL .I ctx refers to an unitialized aio context, the iocb pointed to by .I iocbs contains an improperly initialized iocb, .TP .B ENOSYS Not implemented .SH "SEE ALSO" .BR io(3), .BR io_cancel(3), .BR io_fsync(3), .BR io_getevents(3), .BR io_prep_fsync(3), .BR io_prep_pread(3), .BR io_prep_pwrite(3), .BR io_queue_init(3), .BR io_queue_release(3), .BR io_queue_wait(3), .BR io_set_callback(3), .BR io_submit(3), .BR errno(3) libaio-0.3.109/man/io_queue_wait.3000066400000000000000000000025571126367703600166450ustar00rootroot00000000000000.TH io_queue_wait 2 2002-09-03 "Linux 2.4" "Linux AIO" .SH NAME io_queue_wait \- Wait for io requests to complete .SH SYNOPSIS .nf .B #include .br .sp .B #include .br .sp .BI "int io_queue_wait(io_context_t ctx, const struct timespec *timeout);" .fi .SH DESCRIPTION Attempts to read an event from the completion queue for the aio_context specified by ctx_id. .SH "RETURN VALUES" May return 0 if no events are available and the timeout specified by when has elapsed, where when == NULL specifies an infinite timeout. Note that the timeout pointed to by when is relative and will be updated if not NULL and the operation blocks. Will fail with -ENOSYS if not implemented. .SH "RETURN VALUES" On success, .B io_queue_wait returns 0. Otherwise, -error is return, where error is one of the Exxx values defined in the Errors section. .SH ERRORS .TP .B EFAULT .I iocbs referenced data outside of the program's accessible address space. .TP .B EINVAL .I ctx refers to an unitialized aio context, the iocb pointed to by .I iocbs contains an improperly initialized iocb, .TP .B ENOSYS Not implemented .SH "SEE ALSO" .BR io(3), .BR io_cancel(3), .BR io_fsync(3), .BR io_getevents(3), .BR io_prep_fsync(3), .BR io_prep_pread(3), .BR io_prep_pwrite(3), .BR io_queue_init(3), .BR io_queue_release(3), .BR io_queue_run(3), .BR io_set_callback(3), .BR io_submit(3), .BR errno(3) libaio-0.3.109/man/io_set_callback.3000066400000000000000000000017371126367703600171030ustar00rootroot00000000000000./"static inline void io_set_callback(struct iocb *iocb, io_callback_t cb) .TH io_set_callback 3 2002-09-12 "Linux 2.4" Linux AIO" .SH NAME io_set_callback \- Set up io completion callback function .SH SYNOPSYS .nf .B #include .br .sp .B #include .br .sp .BI "static inline void io_set_callback(struct iocb *iocb, io_callback_t cb)" .sp struct iocb { void *data; unsigned key; short aio_lio_opcode; short aio_reqprio; int aio_fildes; }; .sp typedef void (*io_callback_t)(io_context_t ctx, struct iocb *iocb, long res, long res2); .sp .fi .SH DESCRIPTION The callback is not done if the caller uses raw events from io_getevents, only with the library helpers .SH "RETURN VALUES" .SH ERRORS .SH "SEE ALSO" .BR io(3), .BR io_cancel(3), .BR io_fsync(3), .BR io_getevents(3), .BR io_prep_fsync(3), .BR io_prep_pread(3), .BR io_prep_pwrite(3), .BR io_queue_init(3), .BR io_queue_release(3), .BR io_queue_run(3), .BR io_queue_wait(3), .BR io_submit(3), .BR errno(3) libaio-0.3.109/man/io_setup.1000066400000000000000000000016541126367703600156300ustar00rootroot00000000000000./"/* sys_io_setup: ./" * Create an aio_context capable of receiving at least nr_events. ./" * ctxp must not point to an aio_context that already exists, and ./" * must be initialized to 0 prior to the call. On successful ./" * creation of the aio_context, *ctxp is filled in with the resulting ./" * handle. May fail with -EINVAL if *ctxp is not initialized, ./" * if the specified nr_events exceeds internal limits. May fail ./" * with -EAGAIN if the specified nr_events exceeds the user's limit ./" * of available events. May fail with -ENOMEM if insufficient kernel ./" * resources are available. May fail with -EFAULT if an invalid ./" * pointer is passed for ctxp. Will fail with -ENOSYS if not ./" * implemented. ./" */ ./" -- note: libaio is actually providing io_queue_init and io_queue_grow ./" as separate functions. For now io_setup is the same as io_queue_grow. libaio-0.3.109/man/io_submit.1000066400000000000000000000047071126367703600157750ustar00rootroot00000000000000.TH io_submit 2 2002-09-02 "Linux 2.4" "Linux AIO" .SH NAME io_submit \- submit io requests .SH SYNOPSIS .B #include .br .B #include .LP .BI "int io_submit(io_context_t " ctx ", long " nr ", struct iocb *" iocbs "[]);" .SH DESCRIPTION .B io_submit submits to the io_context .I ctx up to .I nr I/O requests pointed to by the vector .IR iocbs . The .B iocb structure is defined as something like .sp .RS .nf struct iocb { void *data; .\" unsigned key; short aio_lio_opcode; short aio_reqprio; int aio_fildes; }; .fi .RE .sp .I data is a an opaque pointer which will upon completion be returned in the .B io_event structure by .BR io_getevents (2). .\" and io_wait(2) Callers will typically use this to point directly or indirectly to a callback function. .sp .I aio_lio_opcode is the I/O operation requested. Callers will typically set this and the arguments to the I/O operation calling the .BR io_prep_ (3) function corresponding to the operation. .sp .I aio_reqprio is the priority of the request. Higher values have more priority; the normal priority is 0. .sp .I aio_fildes is the file descriptor for the I/O operation. Callers will typically set this and the arguments to the I/O operation calling the .BR io_prep_ *(3) function corresponding to the operation. .sp The caller may not modify the contents or resubmit a submitted .B iocb structure until after the operation completes or is canceled. The implementation of .BR io_submit (2) is permitted to modify reserved fields of the .B iocb structure. .SH "RETURN VALUES" If able to submit at least one iocb, .B io_submit returns the number of iocbs submitted successfully. Otherwise, .RI - error is returned, where .I error is one of the Exxx values defined in the Errors section. .SH ERRORS .TP .B EFAULT .I iocbs referenced data outside of the program's accessible address space. .TP .B EINVAL .I nr is negative, .I ctx refers to an uninitialized aio context, the iocb pointed to by .IR iocbs [0] is improperly initialized or specifies an unsupported operation. .TP .B EBADF The iocb pointed to by .IR iocbs [0] contains a file descriptor that does not exist. .TP .B EAGAIN Insufficient resources were available to queue any operations. .SH "SEE ALSO" .BR io_setup (2), .BR io_destroy (2), .BR io_getevents (2), .\".BR io_wait (2), .BR io_prep_pread (3), .BR io_prep_pwrite (3), .BR io_prep_fsync (3), .BR io_prep_fdsync (3), .BR io_prep_noop (3), .BR io_cancel (2), .BR errno (3) libaio-0.3.109/man/io_submit.3000066400000000000000000000071231126367703600157720ustar00rootroot00000000000000./"/* sys_io_submit: ./" * Queue the nr iocbs pointed to by iocbpp for processing. Returns ./" * the number of iocbs queued. May return -EINVAL if the aio_context ./" * specified by ctx_id is invalid, if nr is < 0, if the iocb at ./" * *iocbpp[0] is not properly initialized, if the operation specified ./" * is invalid for the file descriptor in the iocb. May fail with ./" * -EFAULT if any of the data structures point to invalid data. May ./" * fail with -EBADF if the file descriptor specified in the first ./" * iocb is invalid. May fail with -EAGAIN if insufficient resources ./" * are available to queue any iocbs. Will return 0 if nr is 0. Will ./" * fail with -ENOSYS if not implemented. ./" */ .TH io_submit 2 2002-09-02 "Linux 2.4" "Linux AIO" .SH NAME io_submit \- Submit io requests .SH SYNOPSIS .nf .B #include .br .sp .B #include .br .sp .BI "int io_submit(io_context_t " ctx ", long " nr ", struct iocb *" iocbs "[]);" .sp struct iocb { void *data; unsigned key; short aio_lio_opcode; short aio_reqprio; int aio_fildes; }; .fi .SH DESCRIPTION .B io_submit submits .I nr iocbs for processing for a given io context ctx. The .IR "io_submit" function can be used to enqueue an arbitrary number of read and write requests at one time. The requests can all be meant for the same file, all for different files or every solution in between. .IR "io_submit" gets the .IR "nr" requests from the array pointed to by .IR "iocbs" . The operation to be performed is determined by the .IR "aio_lio_opcode" member in each element of .IR "iocbs" . If this field is .B "IO_CMD_PREAD" a read operation is enqueued, similar to a call of .IR "io_prep_pread" for this element of the array (except that the way the termination is signalled is different, as we will see below). If the .IR "aio_lio_opcode" member is .B "IO_CMD_PWRITE" a write operation is enqueued. Otherwise the .IR "aio_lio_opcode" must be .B "IO_CMD_NOP" in which case this element of .IR "iocbs" is simply ignored. This ``operation'' is useful in situations where one has a fixed array of .IR "struct iocb" elements from which only a few need to be handled at a time. Another situation is where the .IR "io_submit" call was canceled before all requests are processed and the remaining requests have to be reissued. The other members of each element of the array pointed to by .IR "iocbs" must have values suitable for the operation as described in the documentation for .IR "io_prep_pread" and .IR "io_prep_pwrite" above. The function returns immediately after having enqueued all the requests. On success, .B io_submit returns the number of iocbs submitted successfully. Otherwise, -error is return, where error is one of the Exxx values defined in the Errors section. .PP If an error is detected, then the behavior is undefined. .PP Simultaneous asynchronous operations using the same iocb produce undefined results. .SH ERRORS .TP .B EFAULT .I iocbs referenced data outside of the program's accessible address space. .TP .B EINVAL .I ctx refers to an unitialized aio context, the iocb pointed to by .I iocbs contains an improperly initialized iocb, .TP .B EBADF The iocb contains a file descriptor that does not exist. .TP .B EINVAL The file specified in the iocb does not support the given io operation. .SH "SEE ALSO" .BR io(3), .BR io_cancel(3), .BR io_fsync(3), .BR io_getevents(3), .BR io_prep_fsync(3), .BR io_prep_pread(3), .BR io_prep_pwrite(3), .BR io_queue_init(3), .BR io_queue_release(3), .BR io_queue_run(3), .BR io_queue_wait(3), .BR io_set_callback(3), .BR errno(3) libaio-0.3.109/man/lio_listio.3000066400000000000000000000114421126367703600161450ustar00rootroot00000000000000.TH lio_listio 3 2002-09-12 "Linux 2.4" Linux AIO" .SH NAME lio_listio - List directed I/O .SH SYNOPSYS .B #include .br .B #include .LP .BI "int lio_listio (int mode, struct aiocb *const list[], int nent, struct sigevent *sig)" .nf .SH DESCRIPTION Besides these functions with the more or less traditional interface, POSIX.1b also defines a function which can initiate more than one operation at a time, and which can handle freely mixed read and write operations. It is therefore similar to a combination of .IR readv and .IR "writev" . The .IR "lio_listio" function can be used to enqueue an arbitrary number of read and write requests at one time. The requests can all be meant for the same file, all for different files or every solution in between. .IR "lio_listio" gets the .IR "nent" requests from the array pointed to by .IR "list" . The operation to be performed is determined by the .IR "aio_lio_opcode" member in each element of .IR "list" . If this field is .B "LIO_READ" a read operation is enqueued, similar to a call of .IR "aio_read" for this element of the array (except that the way the termination is signalled is different, as we will see below). If the .IR "aio_lio_opcode" member is .B "LIO_WRITE" a write operation is enqueued. Otherwise the .IR "aio_lio_opcode" must be .B "LIO_NOP" in which case this element of .IR "list" is simply ignored. This ``operation'' is useful in situations where one has a fixed array of .IR "struct aiocb" elements from which only a few need to be handled at a time. Another situation is where the .IR "lio_listio" call was canceled before all requests are processed and the remaining requests have to be reissued. The other members of each element of the array pointed to by .IR "list" must have values suitable for the operation as described in the documentation for .IR "aio_read" and .IR "aio_write" above. The .IR "mode" argument determines how .IR "lio_listio" behaves after having enqueued all the requests. If .IR "mode" is .B "LIO_WAIT" it waits until all requests terminated. Otherwise .IR "mode" must be .B "LIO_NOWAIT" and in this case the function returns immediately after having enqueued all the requests. In this case the caller gets a notification of the termination of all requests according to the .IR "sig" parameter. If .IR "sig" is .B "NULL" no notification is send. Otherwise a signal is sent or a thread is started, just as described in the description for .IR "aio_read" or .IR "aio_write" . When the sources are compiled with .B "_FILE_OFFSET_BITS == 64" , this function is in fact .IR "lio_listio64" since the LFS interface transparently replaces the normal implementation. .SH "RETURN VALUES" If .IR "mode" is .B "LIO_WAIT" , the return value of .IR "lio_listio" is .IR 0 when all requests completed successfully. Otherwise the function return .IR 1 and .IR "errno" is set accordingly. To find out which request or requests failed one has to use the .IR "aio_error" function on all the elements of the array .IR "list" . In case .IR "mode" is .B "LIO_NOWAIT" , the function returns .IR 0 if all requests were enqueued correctly. The current state of the requests can be found using .IR "aio_error" and .IR "aio_return" as described above. If .IR "lio_listio" returns .IR -1 in this mode, the global variable .IR "errno" is set accordingly. If a request did not yet terminate, a call to .IR "aio_error" returns .B "EINPROGRESS" . If the value is different, the request is finished and the error value (or .IR 0 ) is returned and the result of the operation can be retrieved using .IR "aio_return" . .SH ERRORS Possible values for .IR "errno" are: .TP .B EAGAIN The resources necessary to queue all the requests are not available at the moment. The error status for each element of .IR "list" must be checked to determine which request failed. Another reason could be that the system wide limit of AIO requests is exceeded. This cannot be the case for the implementation on GNU systems since no arbitrary limits exist. .TP .B EINVAL The .IR "mode" parameter is invalid or .IR "nent" is larger than .B "AIO_LISTIO_MAX" . .TP .B EIO One or more of the request's I/O operations failed. The error status of each request should be checked to determine which one failed. .TP .B ENOSYS The .IR "lio_listio" function is not supported. .PP If the .IR "mode" parameter is .B "LIO_NOWAIT" and the caller cancels a request, the error status for this request returned by .IR "aio_error" is .B "ECANCELED" . .SH "SEE ALSO" .BR aio(3), .BR aio_cancel(3), .BR aio_cancel64(3), .BR aio_error(3), .BR aio_error64(3), .BR aio_fsync(3), .BR aio_fsync64(3), .BR aio_init(3), .BR aio_read(3), .BR aio_read64(3), .BR aio_return(3), .BR aio_return64(3), .BR aio_suspend(3), .BR aio_suspend64(3), .BR aio_write(3), .BR aio_write64(3) libaio-0.3.109/man/lio_listio64.3000066400000000000000000000016421126367703600163200ustar00rootroot00000000000000.TH lio_listio64 3 2002-09-12 "Linux 2.4" Linux AIO" .SH NAME lio_listio64 \- List directed I/O .SH SYNOPSYS .B #include .br .B #include .LP .BI "int lio_listio64 (int mode, struct aiocb *const list[], int nent, struct sigevent *sig)" .nf .SH DESCRIPTION This function is similar to the .IR "code{lio_listio" function. The only difference is that on .IR "32 bit" machines, the file descriptor should be opened in the large file mode. Internally, .IR "lio_listio64" uses functionality equivalent to .IR lseek64" to position the file descriptor correctly for the reading or writing, as opposed to .IR "lseek" functionality used in .IR "lio_listio". When the sources are compiled with .IR "_FILE_OFFSET_BITS == 64" , this function is available under the name .IR "lio_listio" and so transparently replaces the interface for small files on 32 bit machines. .SH "RETURN VALUES" .SH ERRORS .SH "SEE ALSO" libaio-0.3.109/src/000077500000000000000000000000001126367703600137255ustar00rootroot00000000000000libaio-0.3.109/src/Makefile000066400000000000000000000030151126367703600153640ustar00rootroot00000000000000prefix=/usr includedir=$(prefix)/include libdir=$(prefix)/lib ARCH := $(shell uname -m | sed -e s/i.86/i386/) CFLAGS := -nostdlib -nostartfiles -Wall -I. -g -fomit-frame-pointer -O2 -fPIC SO_CFLAGS=-shared $(CFLAGS) L_CFLAGS=$(CFLAGS) LINK_FLAGS= soname=libaio.so.1 minor=0 micro=1 libname=$(soname).$(minor).$(micro) all_targets += libaio.a $(libname) all: $(all_targets) # libaio provided functions libaio_srcs := io_queue_init.c io_queue_release.c libaio_srcs += io_queue_wait.c io_queue_run.c # real syscalls libaio_srcs += io_getevents.c io_submit.c io_cancel.c libaio_srcs += io_setup.c io_destroy.c # internal functions libaio_srcs += raw_syscall.c # old symbols libaio_srcs += compat-0_1.c libaio_objs := $(patsubst %.c,%.ol,$(libaio_srcs)) libaio_sobjs := $(patsubst %.c,%.os,$(libaio_srcs)) $(libaio_objs) $(libaio_sobjs): libaio.h vsys_def.h %.os: %.c $(CC) $(SO_CFLAGS) -c -o $@ $< %.ol: %.c $(CC) $(L_CFLAGS) -c -o $@ $< libaio.a: $(libaio_objs) rm -f libaio.a ar r libaio.a $^ ranlib libaio.a $(libname): $(libaio_sobjs) libaio.map $(CC) $(SO_CFLAGS) -Wl,--version-script=libaio.map -Wl,-soname=$(soname) -o $@ $(libaio_sobjs) $(LINK_FLAGS) install: $(all_targets) install -D -m 644 libaio.h $(includedir)/libaio.h install -D -m 644 libaio.a $(libdir)/libaio.a install -D -m 755 $(libname) $(libdir)/$(libname) ln -sf $(libname) $(libdir)/$(soname) ln -sf $(libname) $(libdir)/libaio.so $(libaio_objs): libaio.h clean: rm -f $(all_targets) $(libaio_objs) $(libaio_sobjs) $(soname).new rm -f *.so* *.a *.o libaio-0.3.109/src/compat-0_1.c000066400000000000000000000036431126367703600157370ustar00rootroot00000000000000/* libaio Linux async I/O interface compat-0_1.c : compatibility symbols for libaio 0.1.x-0.3.x Copyright 2002 Red Hat, Inc. This library is free software; you can redistribute it and/or modify it under the terms of the GNU Lesser General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version. This library is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public License for more details. You should have received a copy of the GNU Lesser General Public License along with this library; if not, write to the Free Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA */ #include #include #include "libaio.h" #include "vsys_def.h" #include "syscall.h" /* ABI change. Provide partial compatibility on this one for now. */ SYMVER(compat0_1_io_cancel, io_cancel, 0.1); int compat0_1_io_cancel(io_context_t ctx, struct iocb *iocb) { struct io_event event; /* FIXME: the old ABI would return the event on the completion queue */ return io_cancel(ctx, iocb, &event); } SYMVER(compat0_1_io_queue_wait, io_queue_wait, 0.1); int compat0_1_io_queue_wait(io_context_t ctx, struct timespec *when) { struct timespec timeout; if (when) timeout = *when; return io_getevents(ctx, 0, 0, NULL, when ? &timeout : NULL); } /* ABI change. Provide backwards compatibility for this one. */ SYMVER(compat0_1_io_getevents, io_getevents, 0.1); int compat0_1_io_getevents(io_context_t ctx_id, long nr, struct io_event *events, const struct timespec *const_timeout) { struct timespec timeout; if (const_timeout) timeout = *const_timeout; return io_getevents(ctx_id, 1, nr, events, const_timeout ? &timeout : NULL); } libaio-0.3.109/src/io_cancel.c000066400000000000000000000017611126367703600160120ustar00rootroot00000000000000/* io_cancel.c libaio Linux async I/O interface Copyright 2002 Red Hat, Inc. This library is free software; you can redistribute it and/or modify it under the terms of the GNU Lesser General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version. This library is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public License for more details. You should have received a copy of the GNU Lesser General Public License along with this library; if not, write to the Free Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA */ #include #include "syscall.h" io_syscall3(int, io_cancel_0_4, io_cancel, io_context_t, ctx, struct iocb *, iocb, struct io_event *, event) DEFSYMVER(io_cancel_0_4, io_cancel, 0.4) libaio-0.3.109/src/io_destroy.c000066400000000000000000000016511126367703600162540ustar00rootroot00000000000000/* io_destroy libaio Linux async I/O interface Copyright 2002 Red Hat, Inc. This library is free software; you can redistribute it and/or modify it under the terms of the GNU Lesser General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version. This library is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public License for more details. You should have received a copy of the GNU Lesser General Public License along with this library; if not, write to the Free Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA */ #include #include #include "syscall.h" io_syscall1(int, io_destroy, io_destroy, io_context_t, ctx) libaio-0.3.109/src/io_getevents.c000066400000000000000000000036621126367703600165730ustar00rootroot00000000000000/* io_getevents.c libaio Linux async I/O interface Copyright 2002 Red Hat, Inc. This library is free software; you can redistribute it and/or modify it under the terms of the GNU Lesser General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version. This library is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public License for more details. You should have received a copy of the GNU Lesser General Public License along with this library; if not, write to the Free Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA */ #include #include #include #include #include "syscall.h" io_syscall5(int, __io_getevents_0_4, io_getevents, io_context_t, ctx, long, min_nr, long, nr, struct io_event *, events, struct timespec *, timeout) #define AIO_RING_MAGIC 0xa10a10a1 /* Ben will hate me for this */ struct aio_ring { unsigned id; /* kernel internal index number */ unsigned nr; /* number of io_events */ unsigned head; unsigned tail; unsigned magic; unsigned compat_features; unsigned incompat_features; unsigned header_length; /* size of aio_ring */ }; int io_getevents_0_4(io_context_t ctx, long min_nr, long nr, struct io_event * events, struct timespec * timeout) { struct aio_ring *ring; ring = (struct aio_ring*)ctx; if (ring==NULL || ring->magic != AIO_RING_MAGIC) goto do_syscall; if (timeout!=NULL && timeout->tv_sec == 0 && timeout->tv_nsec == 0) { if (ring->head == ring->tail) return 0; } do_syscall: return __io_getevents_0_4(ctx, min_nr, nr, events, timeout); } DEFSYMVER(io_getevents_0_4, io_getevents, 0.4) libaio-0.3.109/src/io_queue_init.c000066400000000000000000000020671126367703600167340ustar00rootroot00000000000000/* io_queue_init.c libaio Linux async I/O interface Copyright 2002 Red Hat, Inc. This library is free software; you can redistribute it and/or modify it under the terms of the GNU Lesser General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version. This library is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public License for more details. You should have received a copy of the GNU Lesser General Public License along with this library; if not, write to the Free Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA */ #include #include #include #include #include "syscall.h" int io_queue_init(int maxevents, io_context_t *ctxp) { if (maxevents > 0) { *ctxp = NULL; return io_setup(maxevents, ctxp); } return -EINVAL; } libaio-0.3.109/src/io_queue_release.c000066400000000000000000000017211126367703600174050ustar00rootroot00000000000000/* io_queue_release.c libaio Linux async I/O interface Copyright 2002 Red Hat, Inc. This library is free software; you can redistribute it and/or modify it under the terms of the GNU Lesser General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version. This library is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public License for more details. You should have received a copy of the GNU Lesser General Public License along with this library; if not, write to the Free Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA */ #include #include #include #include int io_queue_release(io_context_t ctx) { return io_destroy(ctx); } libaio-0.3.109/src/io_queue_run.c000066400000000000000000000023361126367703600165740ustar00rootroot00000000000000/* io_submit libaio Linux async I/O interface Copyright 2002 Red Hat, Inc. This library is free software; you can redistribute it and/or modify it under the terms of the GNU Lesser General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version. This library is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public License for more details. You should have received a copy of the GNU Lesser General Public License along with this library; if not, write to the Free Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA */ #include #include #include #include int io_queue_run(io_context_t ctx) { static struct timespec timeout = { 0, 0 }; struct io_event event; int ret; /* FIXME: batch requests? */ while (1 == (ret = io_getevents(ctx, 0, 1, &event, &timeout))) { io_callback_t cb = (io_callback_t)event.data; struct iocb *iocb = event.obj; cb(ctx, iocb, event.res, event.res2); } return ret; } libaio-0.3.109/src/io_queue_wait.c000066400000000000000000000021251126367703600167300ustar00rootroot00000000000000/* io_submit libaio Linux async I/O interface Copyright 2002 Red Hat, Inc. This library is free software; you can redistribute it and/or modify it under the terms of the GNU Lesser General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version. This library is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public License for more details. You should have received a copy of the GNU Lesser General Public License along with this library; if not, write to the Free Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA */ #define NO_SYSCALL_ERRNO #include #include #include #include "syscall.h" struct timespec; int io_queue_wait_0_4(io_context_t ctx, struct timespec *timeout) { return io_getevents(ctx, 0, 0, NULL, timeout); } DEFSYMVER(io_queue_wait_0_4, io_queue_wait, 0.4) libaio-0.3.109/src/io_setup.c000066400000000000000000000016661126367703600157310ustar00rootroot00000000000000/* io_setup libaio Linux async I/O interface Copyright 2002 Red Hat, Inc. This library is free software; you can redistribute it and/or modify it under the terms of the GNU Lesser General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version. This library is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public License for more details. You should have received a copy of the GNU Lesser General Public License along with this library; if not, write to the Free Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA */ #include #include #include "syscall.h" io_syscall2(int, io_setup, io_setup, int, maxevents, io_context_t *, ctxp) libaio-0.3.109/src/io_submit.c000066400000000000000000000017071126367703600160700ustar00rootroot00000000000000/* io_submit libaio Linux async I/O interface Copyright 2002 Red Hat, Inc. This library is free software; you can redistribute it and/or modify it under the terms of the GNU Lesser General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version. This library is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public License for more details. You should have received a copy of the GNU Lesser General Public License along with this library; if not, write to the Free Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA */ #include #include #include "syscall.h" io_syscall3(int, io_submit, io_submit, io_context_t, ctx, long, nr, struct iocb **, iocbs) libaio-0.3.109/src/libaio.h000066400000000000000000000167271126367703600153520ustar00rootroot00000000000000/* /usr/include/libaio.h * * Copyright 2000,2001,2002 Red Hat, Inc. * * Written by Benjamin LaHaise * * libaio Linux async I/O interface * * This library is free software; you can redistribute it and/or * modify it under the terms of the GNU Lesser General Public * License as published by the Free Software Foundation; either * version 2 of the License, or (at your option) any later version. * * This library is distributed in the hope that it will be useful, * but WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU * Lesser General Public License for more details. * * You should have received a copy of the GNU Lesser General Public * License along with this library; if not, write to the Free Software * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA */ #ifndef __LIBAIO_H #define __LIBAIO_H #ifdef __cplusplus extern "C" { #endif #include #include struct timespec; struct sockaddr; struct iovec; typedef struct io_context *io_context_t; typedef enum io_iocb_cmd { IO_CMD_PREAD = 0, IO_CMD_PWRITE = 1, IO_CMD_FSYNC = 2, IO_CMD_FDSYNC = 3, IO_CMD_POLL = 5, /* Never implemented in mainline, see io_prep_poll */ IO_CMD_NOOP = 6, IO_CMD_PREADV = 7, IO_CMD_PWRITEV = 8, } io_iocb_cmd_t; #if defined(__i386__) /* little endian, 32 bits */ #define PADDED(x, y) x; unsigned y #define PADDEDptr(x, y) x; unsigned y #define PADDEDul(x, y) unsigned long x; unsigned y #elif defined(__ia64__) || defined(__x86_64__) || defined(__alpha__) #define PADDED(x, y) x, y #define PADDEDptr(x, y) x #define PADDEDul(x, y) unsigned long x #elif defined(__powerpc64__) /* big endian, 64 bits */ #define PADDED(x, y) unsigned y; x #define PADDEDptr(x,y) x #define PADDEDul(x, y) unsigned long x #elif defined(__PPC__) /* big endian, 32 bits */ #define PADDED(x, y) unsigned y; x #define PADDEDptr(x, y) unsigned y; x #define PADDEDul(x, y) unsigned y; unsigned long x #elif defined(__s390x__) /* big endian, 64 bits */ #define PADDED(x, y) unsigned y; x #define PADDEDptr(x,y) x #define PADDEDul(x, y) unsigned long x #elif defined(__s390__) /* big endian, 32 bits */ #define PADDED(x, y) unsigned y; x #define PADDEDptr(x, y) unsigned y; x #define PADDEDul(x, y) unsigned y; unsigned long x #elif defined(__arm__) # if defined (__ARMEB__) /* big endian, 32 bits */ #define PADDED(x, y) unsigned y; x #define PADDEDptr(x, y) unsigned y; x #define PADDEDul(x, y) unsigned y; unsigned long x # else /* little endian, 32 bits */ #define PADDED(x, y) x; unsigned y #define PADDEDptr(x, y) x; unsigned y #define PADDEDul(x, y) unsigned long x; unsigned y # endif #else #error endian? #endif struct io_iocb_poll { PADDED(int events, __pad1); }; /* result code is the set of result flags or -'ve errno */ struct io_iocb_sockaddr { struct sockaddr *addr; int len; }; /* result code is the length of the sockaddr, or -'ve errno */ struct io_iocb_common { PADDEDptr(void *buf, __pad1); PADDEDul(nbytes, __pad2); long long offset; long long __pad3; unsigned flags; unsigned resfd; }; /* result code is the amount read or -'ve errno */ struct io_iocb_vector { const struct iovec *vec; int nr; long long offset; }; /* result code is the amount read or -'ve errno */ struct iocb { PADDEDptr(void *data, __pad1); /* Return in the io completion event */ PADDED(unsigned key, __pad2); /* For use in identifying io requests */ short aio_lio_opcode; short aio_reqprio; int aio_fildes; union { struct io_iocb_common c; struct io_iocb_vector v; struct io_iocb_poll poll; struct io_iocb_sockaddr saddr; } u; }; struct io_event { PADDEDptr(void *data, __pad1); PADDEDptr(struct iocb *obj, __pad2); PADDEDul(res, __pad3); PADDEDul(res2, __pad4); }; #undef PADDED #undef PADDEDptr #undef PADDEDul typedef void (*io_callback_t)(io_context_t ctx, struct iocb *iocb, long res, long res2); /* library wrappers */ extern int io_queue_init(int maxevents, io_context_t *ctxp); /*extern int io_queue_grow(io_context_t ctx, int new_maxevents);*/ extern int io_queue_release(io_context_t ctx); /*extern int io_queue_wait(io_context_t ctx, struct timespec *timeout);*/ extern int io_queue_run(io_context_t ctx); /* Actual syscalls */ extern int io_setup(int maxevents, io_context_t *ctxp); extern int io_destroy(io_context_t ctx); extern int io_submit(io_context_t ctx, long nr, struct iocb *ios[]); extern int io_cancel(io_context_t ctx, struct iocb *iocb, struct io_event *evt); extern int io_getevents(io_context_t ctx_id, long min_nr, long nr, struct io_event *events, struct timespec *timeout); static inline void io_set_callback(struct iocb *iocb, io_callback_t cb) { iocb->data = (void *)cb; } static inline void io_prep_pread(struct iocb *iocb, int fd, void *buf, size_t count, long long offset) { memset(iocb, 0, sizeof(*iocb)); iocb->aio_fildes = fd; iocb->aio_lio_opcode = IO_CMD_PREAD; iocb->aio_reqprio = 0; iocb->u.c.buf = buf; iocb->u.c.nbytes = count; iocb->u.c.offset = offset; } static inline void io_prep_pwrite(struct iocb *iocb, int fd, void *buf, size_t count, long long offset) { memset(iocb, 0, sizeof(*iocb)); iocb->aio_fildes = fd; iocb->aio_lio_opcode = IO_CMD_PWRITE; iocb->aio_reqprio = 0; iocb->u.c.buf = buf; iocb->u.c.nbytes = count; iocb->u.c.offset = offset; } static inline void io_prep_preadv(struct iocb *iocb, int fd, const struct iovec *iov, int iovcnt, long long offset) { memset(iocb, 0, sizeof(*iocb)); iocb->aio_fildes = fd; iocb->aio_lio_opcode = IO_CMD_PREADV; iocb->aio_reqprio = 0; iocb->u.c.buf = (void *)iov; iocb->u.c.nbytes = iovcnt; iocb->u.c.offset = offset; } static inline void io_prep_pwritev(struct iocb *iocb, int fd, const struct iovec *iov, int iovcnt, long long offset) { memset(iocb, 0, sizeof(*iocb)); iocb->aio_fildes = fd; iocb->aio_lio_opcode = IO_CMD_PWRITEV; iocb->aio_reqprio = 0; iocb->u.c.buf = (void *)iov; iocb->u.c.nbytes = iovcnt; iocb->u.c.offset = offset; } /* Jeff Moyer says this was implemented in Red Hat AS2.1 and RHEL3. * AFAICT, it was never in mainline, and should not be used. --RR */ static inline void io_prep_poll(struct iocb *iocb, int fd, int events) { memset(iocb, 0, sizeof(*iocb)); iocb->aio_fildes = fd; iocb->aio_lio_opcode = IO_CMD_POLL; iocb->aio_reqprio = 0; iocb->u.poll.events = events; } static inline int io_poll(io_context_t ctx, struct iocb *iocb, io_callback_t cb, int fd, int events) { io_prep_poll(iocb, fd, events); io_set_callback(iocb, cb); return io_submit(ctx, 1, &iocb); } static inline void io_prep_fsync(struct iocb *iocb, int fd) { memset(iocb, 0, sizeof(*iocb)); iocb->aio_fildes = fd; iocb->aio_lio_opcode = IO_CMD_FSYNC; iocb->aio_reqprio = 0; } static inline int io_fsync(io_context_t ctx, struct iocb *iocb, io_callback_t cb, int fd) { io_prep_fsync(iocb, fd); io_set_callback(iocb, cb); return io_submit(ctx, 1, &iocb); } static inline void io_prep_fdsync(struct iocb *iocb, int fd) { memset(iocb, 0, sizeof(*iocb)); iocb->aio_fildes = fd; iocb->aio_lio_opcode = IO_CMD_FDSYNC; iocb->aio_reqprio = 0; } static inline int io_fdsync(io_context_t ctx, struct iocb *iocb, io_callback_t cb, int fd) { io_prep_fdsync(iocb, fd); io_set_callback(iocb, cb); return io_submit(ctx, 1, &iocb); } static inline void io_set_eventfd(struct iocb *iocb, int eventfd) { iocb->u.c.flags |= (1 << 0) /* IOCB_FLAG_RESFD */; iocb->u.c.resfd = eventfd; } #ifdef __cplusplus } #endif #endif /* __LIBAIO_H */ libaio-0.3.109/src/libaio.map000066400000000000000000000004041126367703600156610ustar00rootroot00000000000000LIBAIO_0.1 { global: io_queue_init; io_queue_run; io_queue_wait; io_queue_release; io_cancel; io_submit; io_getevents; local: *; }; LIBAIO_0.4 { global: io_setup; io_destroy; io_cancel; io_getevents; io_queue_wait; } LIBAIO_0.1; libaio-0.3.109/src/raw_syscall.c000066400000000000000000000006321126367703600164150ustar00rootroot00000000000000#include "syscall.h" #if defined(__ia64__) /* based on code from glibc by Jes Sorensen */ __asm__(".text\n" ".globl __ia64_aio_raw_syscall\n" ".proc __ia64_aio_raw_syscall\n" "__ia64_aio_raw_syscall:\n" "alloc r2=ar.pfs,1,0,8,0\n" "mov r15=r32\n" "break 0x100000\n" ";;" "br.ret.sptk.few b0\n" ".size __ia64_aio_raw_syscall, . - __ia64_aio_raw_syscall\n" ".endp __ia64_aio_raw_syscall" ); #endif ; libaio-0.3.109/src/syscall-alpha.h000066400000000000000000000256321126367703600166430ustar00rootroot00000000000000#define __NR_io_setup 398 #define __NR_io_destroy 399 #define __NR_io_getevents 400 #define __NR_io_submit 401 #define __NR_io_cancel 402 #define inline_syscall_r0_asm #define inline_syscall_r0_out_constraint "=v" #define inline_syscall_clobbers \ "$1", "$2", "$3", "$4", "$5", "$6", "$7", "$8", \ "$22", "$23", "$24", "$25", "$27", "$28", "memory" #define inline_syscall0(name, args...) \ { \ register long _sc_0 inline_syscall_r0_asm; \ register long _sc_19 __asm__("$19"); \ \ _sc_0 = name; \ __asm__ __volatile__ \ ("callsys # %0 %1 <= %2" \ : inline_syscall_r0_out_constraint (_sc_0), \ "=r"(_sc_19) \ : "0"(_sc_0) \ : inline_syscall_clobbers, \ "$16", "$17", "$18", "$20", "$21"); \ _sc_ret = _sc_0, _sc_err = _sc_19; \ } #define inline_syscall1(name,arg1) \ { \ register long _sc_0 inline_syscall_r0_asm; \ register long _sc_16 __asm__("$16"); \ register long _sc_19 __asm__("$19"); \ \ _sc_0 = name; \ _sc_16 = (long) (arg1); \ __asm__ __volatile__ \ ("callsys # %0 %1 <= %2 %3" \ : inline_syscall_r0_out_constraint (_sc_0), \ "=r"(_sc_19), "=r"(_sc_16) \ : "0"(_sc_0), "2"(_sc_16) \ : inline_syscall_clobbers, \ "$17", "$18", "$20", "$21"); \ _sc_ret = _sc_0, _sc_err = _sc_19; \ } #define inline_syscall2(name,arg1,arg2) \ { \ register long _sc_0 inline_syscall_r0_asm; \ register long _sc_16 __asm__("$16"); \ register long _sc_17 __asm__("$17"); \ register long _sc_19 __asm__("$19"); \ \ _sc_0 = name; \ _sc_16 = (long) (arg1); \ _sc_17 = (long) (arg2); \ __asm__ __volatile__ \ ("callsys # %0 %1 <= %2 %3 %4" \ : inline_syscall_r0_out_constraint (_sc_0), \ "=r"(_sc_19), "=r"(_sc_16), "=r"(_sc_17) \ : "0"(_sc_0), "2"(_sc_16), "3"(_sc_17) \ : inline_syscall_clobbers, \ "$18", "$20", "$21"); \ _sc_ret = _sc_0, _sc_err = _sc_19; \ } #define inline_syscall3(name,arg1,arg2,arg3) \ { \ register long _sc_0 inline_syscall_r0_asm; \ register long _sc_16 __asm__("$16"); \ register long _sc_17 __asm__("$17"); \ register long _sc_18 __asm__("$18"); \ register long _sc_19 __asm__("$19"); \ \ _sc_0 = name; \ _sc_16 = (long) (arg1); \ _sc_17 = (long) (arg2); \ _sc_18 = (long) (arg3); \ __asm__ __volatile__ \ ("callsys # %0 %1 <= %2 %3 %4 %5" \ : inline_syscall_r0_out_constraint (_sc_0), \ "=r"(_sc_19), "=r"(_sc_16), "=r"(_sc_17), \ "=r"(_sc_18) \ : "0"(_sc_0), "2"(_sc_16), "3"(_sc_17), \ "4"(_sc_18) \ : inline_syscall_clobbers, "$20", "$21"); \ _sc_ret = _sc_0, _sc_err = _sc_19; \ } #define inline_syscall4(name,arg1,arg2,arg3,arg4) \ { \ register long _sc_0 inline_syscall_r0_asm; \ register long _sc_16 __asm__("$16"); \ register long _sc_17 __asm__("$17"); \ register long _sc_18 __asm__("$18"); \ register long _sc_19 __asm__("$19"); \ \ _sc_0 = name; \ _sc_16 = (long) (arg1); \ _sc_17 = (long) (arg2); \ _sc_18 = (long) (arg3); \ _sc_19 = (long) (arg4); \ __asm__ __volatile__ \ ("callsys # %0 %1 <= %2 %3 %4 %5 %6" \ : inline_syscall_r0_out_constraint (_sc_0), \ "=r"(_sc_19), "=r"(_sc_16), "=r"(_sc_17), \ "=r"(_sc_18) \ : "0"(_sc_0), "2"(_sc_16), "3"(_sc_17), \ "4"(_sc_18), "1"(_sc_19) \ : inline_syscall_clobbers, "$20", "$21"); \ _sc_ret = _sc_0, _sc_err = _sc_19; \ } #define inline_syscall5(name,arg1,arg2,arg3,arg4,arg5) \ { \ register long _sc_0 inline_syscall_r0_asm; \ register long _sc_16 __asm__("$16"); \ register long _sc_17 __asm__("$17"); \ register long _sc_18 __asm__("$18"); \ register long _sc_19 __asm__("$19"); \ register long _sc_20 __asm__("$20"); \ \ _sc_0 = name; \ _sc_16 = (long) (arg1); \ _sc_17 = (long) (arg2); \ _sc_18 = (long) (arg3); \ _sc_19 = (long) (arg4); \ _sc_20 = (long) (arg5); \ __asm__ __volatile__ \ ("callsys # %0 %1 <= %2 %3 %4 %5 %6 %7" \ : inline_syscall_r0_out_constraint (_sc_0), \ "=r"(_sc_19), "=r"(_sc_16), "=r"(_sc_17), \ "=r"(_sc_18), "=r"(_sc_20) \ : "0"(_sc_0), "2"(_sc_16), "3"(_sc_17), \ "4"(_sc_18), "1"(_sc_19), "5"(_sc_20) \ : inline_syscall_clobbers, "$21"); \ _sc_ret = _sc_0, _sc_err = _sc_19; \ } #define inline_syscall6(name,arg1,arg2,arg3,arg4,arg5,arg6) \ { \ register long _sc_0 inline_syscall_r0_asm; \ register long _sc_16 __asm__("$16"); \ register long _sc_17 __asm__("$17"); \ register long _sc_18 __asm__("$18"); \ register long _sc_19 __asm__("$19"); \ register long _sc_20 __asm__("$20"); \ register long _sc_21 __asm__("$21"); \ \ _sc_0 = name; \ _sc_16 = (long) (arg1); \ _sc_17 = (long) (arg2); \ _sc_18 = (long) (arg3); \ _sc_19 = (long) (arg4); \ _sc_20 = (long) (arg5); \ _sc_21 = (long) (arg6); \ __asm__ __volatile__ \ ("callsys # %0 %1 <= %2 %3 %4 %5 %6 %7 %8" \ : inline_syscall_r0_out_constraint (_sc_0), \ "=r"(_sc_19), "=r"(_sc_16), "=r"(_sc_17), \ "=r"(_sc_18), "=r"(_sc_20), "=r"(_sc_21) \ : "0"(_sc_0), "2"(_sc_16), "3"(_sc_17), "4"(_sc_18), \ "1"(_sc_19), "5"(_sc_20), "6"(_sc_21) \ : inline_syscall_clobbers); \ _sc_ret = _sc_0, _sc_err = _sc_19; \ } #define INLINE_SYSCALL1(name, nr, args...) \ ({ \ long _sc_ret, _sc_err; \ inline_syscall##nr(__NR_##name, args); \ if (_sc_err != 0) \ { \ _sc_ret = -(_sc_ret); \ } \ _sc_ret; \ }) #define io_syscall1(type,fname,sname,type1,arg1) \ type fname(type1 arg1) \ { \ return (type)INLINE_SYSCALL1(sname, 1, arg1); \ } #define io_syscall2(type,fname,sname,type1,arg1,type2,arg2) \ type fname(type1 arg1,type2 arg2) \ { \ return (type)INLINE_SYSCALL1(sname, 2, arg1, arg2); \ } #define io_syscall3(type,fname,sname,type1,arg1,type2,arg2,type3,arg3) \ type fname(type1 arg1,type2 arg2,type3 arg3) \ { \ return (type)INLINE_SYSCALL1(sname, 3, arg1, arg2, arg3); \ } #define io_syscall4(type,fname,sname,type1,arg1,type2,arg2,type3,arg3,type4,arg4) \ type fname (type1 arg1, type2 arg2, type3 arg3, type4 arg4) \ { \ return (type)INLINE_SYSCALL1(sname, 4, arg1, arg2, arg3, arg4); \ } #define io_syscall5(type,fname,sname,type1,arg1,type2,arg2,type3,arg3,type4,arg4, \ type5,arg5) \ type fname (type1 arg1,type2 arg2,type3 arg3,type4 arg4,type5 arg5) \ { \ return (type)INLINE_SYSCALL1(sname, 5, arg1, arg2, arg3, arg4, arg5);\ } libaio-0.3.109/src/syscall-arm.h000066400000000000000000000076771126367703600163460ustar00rootroot00000000000000/* * linux/include/asm-arm/unistd.h * * Copyright (C) 2001-2005 Russell King * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License version 2 as * published by the Free Software Foundation. * * Please forward _all_ changes to this file to rmk@arm.linux.org.uk, * no matter what the change is. Thanks! */ #define __NR_OABI_SYSCALL_BASE 0x900000 #if defined(__thumb__) || defined(__ARM_EABI__) #define __NR_SYSCALL_BASE 0 #else #define __NR_SYSCALL_BASE __NR_OABI_SYSCALL_BASE #endif #define __NR_io_setup (__NR_SYSCALL_BASE+243) #define __NR_io_destroy (__NR_SYSCALL_BASE+244) #define __NR_io_getevents (__NR_SYSCALL_BASE+245) #define __NR_io_submit (__NR_SYSCALL_BASE+246) #define __NR_io_cancel (__NR_SYSCALL_BASE+247) #define __sys2(x) #x #define __sys1(x) __sys2(x) #if defined(__thumb__) || defined(__ARM_EABI__) #define __SYS_REG(name) register long __sysreg __asm__("r7") = __NR_##name; #define __SYS_REG_LIST(regs...) "r" (__sysreg) , ##regs #define __syscall(name) "swi\t0" #else #define __SYS_REG(name) #define __SYS_REG_LIST(regs...) regs #define __syscall(name) "swi\t" __sys1(__NR_##name) "" #endif #define io_syscall1(type,fname,sname,type1,arg1) \ type fname(type1 arg1) { \ __SYS_REG(sname) \ register long __r0 __asm__("r0") = (long)arg1; \ register long __res_r0 __asm__("r0"); \ __asm__ __volatile__ ( \ __syscall(sname) \ : "=r" (__res_r0) \ : __SYS_REG_LIST( "0" (__r0) ) \ : "memory" ); \ return (type) __res_r0; \ } #define io_syscall2(type,fname,sname,type1,arg1,type2,arg2) \ type fname(type1 arg1,type2 arg2) { \ __SYS_REG(sname) \ register long __r0 __asm__("r0") = (long)arg1; \ register long __r1 __asm__("r1") = (long)arg2; \ register long __res_r0 __asm__("r0"); \ __asm__ __volatile__ ( \ __syscall(sname) \ : "=r" (__res_r0) \ : __SYS_REG_LIST( "0" (__r0), "r" (__r1) ) \ : "memory" ); \ return (type) __res_r0; \ } #define io_syscall3(type,fname,sname,type1,arg1,type2,arg2,type3,arg3) \ type fname(type1 arg1,type2 arg2,type3 arg3) { \ __SYS_REG(sname) \ register long __r0 __asm__("r0") = (long)arg1; \ register long __r1 __asm__("r1") = (long)arg2; \ register long __r2 __asm__("r2") = (long)arg3; \ register long __res_r0 __asm__("r0"); \ __asm__ __volatile__ ( \ __syscall(sname) \ : "=r" (__res_r0) \ : __SYS_REG_LIST( "0" (__r0), "r" (__r1), "r" (__r2) ) \ : "memory" ); \ return (type) __res_r0; \ } #define io_syscall4(type,fname,sname,type1,arg1,type2,arg2,type3,arg3,type4,arg4)\ type fname(type1 arg1, type2 arg2, type3 arg3, type4 arg4) { \ __SYS_REG(sname) \ register long __r0 __asm__("r0") = (long)arg1; \ register long __r1 __asm__("r1") = (long)arg2; \ register long __r2 __asm__("r2") = (long)arg3; \ register long __r3 __asm__("r3") = (long)arg4; \ register long __res_r0 __asm__("r0"); \ __asm__ __volatile__ ( \ __syscall(sname) \ : "=r" (__res_r0) \ : __SYS_REG_LIST( "0" (__r0), "r" (__r1), "r" (__r2), "r" (__r3) ) \ : "memory" ); \ return (type) __res_r0; \ } #define io_syscall5(type,fname,sname,type1,arg1,type2,arg2,type3,arg3,type4,arg4,type5,arg5) \ type fname(type1 arg1, type2 arg2, type3 arg3, type4 arg4, type5 arg5) {\ __SYS_REG(sname) \ register long __r0 __asm__("r0") = (long)arg1; \ register long __r1 __asm__("r1") = (long)arg2; \ register long __r2 __asm__("r2") = (long)arg3; \ register long __r3 __asm__("r3") = (long)arg4; \ register long __r4 __asm__("r4") = (long)arg5; \ register long __res_r0 __asm__("r0"); \ __asm__ __volatile__ ( \ __syscall(sname) \ : "=r" (__res_r0) \ : __SYS_REG_LIST( "0" (__r0), "r" (__r1), "r" (__r2), \ "r" (__r3), "r" (__r4) ) \ : "memory" ); \ return (type) __res_r0; \ } libaio-0.3.109/src/syscall-i386.h000066400000000000000000000042051126367703600162400ustar00rootroot00000000000000#define __NR_io_setup 245 #define __NR_io_destroy 246 #define __NR_io_getevents 247 #define __NR_io_submit 248 #define __NR_io_cancel 249 #define io_syscall1(type,fname,sname,type1,arg1) \ type fname(type1 arg1) \ { \ long __res; \ __asm__ volatile ("xchgl %%edi,%%ebx\n" \ "int $0x80\n" \ "xchgl %%edi,%%ebx" \ : "=a" (__res) \ : "0" (__NR_##sname),"D" ((long)(arg1))); \ return __res; \ } #define io_syscall2(type,fname,sname,type1,arg1,type2,arg2) \ type fname(type1 arg1,type2 arg2) \ { \ long __res; \ __asm__ volatile ("xchgl %%edi,%%ebx\n" \ "int $0x80\n" \ "xchgl %%edi,%%ebx" \ : "=a" (__res) \ : "0" (__NR_##sname),"D" ((long)(arg1)),"c" ((long)(arg2))); \ return __res; \ } #define io_syscall3(type,fname,sname,type1,arg1,type2,arg2,type3,arg3) \ type fname(type1 arg1,type2 arg2,type3 arg3) \ { \ long __res; \ __asm__ volatile ("xchgl %%edi,%%ebx\n" \ "int $0x80\n" \ "xchgl %%edi,%%ebx" \ : "=a" (__res) \ : "0" (__NR_##sname),"D" ((long)(arg1)),"c" ((long)(arg2)), \ "d" ((long)(arg3))); \ return __res; \ } #define io_syscall4(type,fname,sname,type1,arg1,type2,arg2,type3,arg3,type4,arg4) \ type fname (type1 arg1, type2 arg2, type3 arg3, type4 arg4) \ { \ long __res; \ __asm__ volatile ("xchgl %%edi,%%ebx\n" \ "int $0x80\n" \ "xchgl %%edi,%%ebx" \ : "=a" (__res) \ : "0" (__NR_##sname),"D" ((long)(arg1)),"c" ((long)(arg2)), \ "d" ((long)(arg3)),"S" ((long)(arg4))); \ return __res; \ } #define io_syscall5(type,fname,sname,type1,arg1,type2,arg2,type3,arg3,type4,arg4, \ type5,arg5) \ type fname (type1 arg1,type2 arg2,type3 arg3,type4 arg4,type5 arg5) \ { \ long __res; \ long tmp; \ __asm__ volatile ("movl %%ebx,%7\n" \ "movl %2,%%ebx\n" \ "int $0x80\n" \ "movl %7,%%ebx" \ : "=a" (__res) \ : "0" (__NR_##sname),"rm" ((long)(arg1)),"c" ((long)(arg2)), \ "d" ((long)(arg3)),"S" ((long)(arg4)),"D" ((long)(arg5)), \ "m" (tmp)); \ return __res; \ } libaio-0.3.109/src/syscall-ia64.h000066400000000000000000000030751126367703600163160ustar00rootroot00000000000000#define __NR_io_setup 1238 #define __NR_io_destroy 1239 #define __NR_io_getevents 1240 #define __NR_io_submit 1241 #define __NR_io_cancel 1242 #define __ia64_raw_syscall(fname, sname) \ __asm__ (".text\n" \ ".globl " SYMSTR(fname) "\n" \ ".proc " SYMSTR(fname) "\n" \ SYMSTR(fname) ":\n" \ " mov r15=" SYMSTR( __NR_ ## sname ) "\n" \ " break 0x100000\n" \ " ;;\n" \ " cmp.eq p6,p0=-1,r10\n" \ " ;;\n" \ " (p6) sub r8=0,r8\n" \ " br.ret.sptk.few b0\n" \ ".size " SYMSTR(fname) ", . - " SYMSTR(fname) "\n" \ ".endp " SYMSTR(fname) "\n" \ ); #define io_syscall0(type, name) \ extern type name(void); \ __ia64_raw_syscall(name); #define io_syscall1(type, fname, sname, type1, arg1) \ extern type fname(type1 arg1); \ __ia64_raw_syscall(fname, sname); #define io_syscall2(type, fname, sname, type1, arg1, type2, arg2) \ extern type fname(type1 arg1, type2 arg2); \ __ia64_raw_syscall(fname, sname); #define io_syscall3(type, fname, sname, type1, arg1, type2, arg2, type3, arg3) \ extern type fname(type1 arg1, type2 arg2, type3 arg3); \ __ia64_raw_syscall(fname, sname); #define io_syscall4(type, fname, sname, type1, arg1, type2, arg2, type3, arg3, type4, arg4) \ extern type fname(type1 arg1, type2 arg2, type3 arg3, type4 arg4); \ __ia64_raw_syscall(fname, sname); #define io_syscall5(type, fname, sname, type1, arg1, type2, arg2, type3, arg3, type4, arg4, type5, arg5) \ extern type fname(type1 arg1, type2 arg2, type3 arg3, type4 arg4, type5 arg5); \ __ia64_raw_syscall(fname, sname); libaio-0.3.109/src/syscall-ppc.h000066400000000000000000000064611126367703600163370ustar00rootroot00000000000000#define __NR_io_setup 227 #define __NR_io_destroy 228 #define __NR_io_getevents 229 #define __NR_io_submit 230 #define __NR_io_cancel 231 /* On powerpc a system call basically clobbers the same registers like a * function call, with the exception of LR (which is needed for the * "sc; bnslr" sequence) and CR (where only CR0.SO is clobbered to signal * an error return status). */ #define __syscall_nr(nr, type, name, args...) \ unsigned long __sc_ret, __sc_err; \ { \ register unsigned long __sc_0 __asm__ ("r0"); \ register unsigned long __sc_3 __asm__ ("r3"); \ register unsigned long __sc_4 __asm__ ("r4"); \ register unsigned long __sc_5 __asm__ ("r5"); \ register unsigned long __sc_6 __asm__ ("r6"); \ register unsigned long __sc_7 __asm__ ("r7"); \ register unsigned long __sc_8 __asm__ ("r8"); \ \ __sc_loadargs_##nr(name, args); \ __asm__ __volatile__ \ ("sc \n\t" \ "mfcr %0 " \ : "=&r" (__sc_0), \ "=&r" (__sc_3), "=&r" (__sc_4), \ "=&r" (__sc_5), "=&r" (__sc_6), \ "=&r" (__sc_7), "=&r" (__sc_8) \ : __sc_asm_input_##nr \ : "cr0", "ctr", "memory", \ "r9", "r10","r11", "r12"); \ __sc_ret = __sc_3; \ __sc_err = __sc_0; \ } \ if (__sc_err & 0x10000000) return -((int)__sc_ret); \ return (type) __sc_ret #define __sc_loadargs_0(name, dummy...) \ __sc_0 = __NR_##name #define __sc_loadargs_1(name, arg1) \ __sc_loadargs_0(name); \ __sc_3 = (unsigned long) (arg1) #define __sc_loadargs_2(name, arg1, arg2) \ __sc_loadargs_1(name, arg1); \ __sc_4 = (unsigned long) (arg2) #define __sc_loadargs_3(name, arg1, arg2, arg3) \ __sc_loadargs_2(name, arg1, arg2); \ __sc_5 = (unsigned long) (arg3) #define __sc_loadargs_4(name, arg1, arg2, arg3, arg4) \ __sc_loadargs_3(name, arg1, arg2, arg3); \ __sc_6 = (unsigned long) (arg4) #define __sc_loadargs_5(name, arg1, arg2, arg3, arg4, arg5) \ __sc_loadargs_4(name, arg1, arg2, arg3, arg4); \ __sc_7 = (unsigned long) (arg5) #define __sc_asm_input_0 "0" (__sc_0) #define __sc_asm_input_1 __sc_asm_input_0, "1" (__sc_3) #define __sc_asm_input_2 __sc_asm_input_1, "2" (__sc_4) #define __sc_asm_input_3 __sc_asm_input_2, "3" (__sc_5) #define __sc_asm_input_4 __sc_asm_input_3, "4" (__sc_6) #define __sc_asm_input_5 __sc_asm_input_4, "5" (__sc_7) #define io_syscall1(type,fname,sname,type1,arg1) \ type fname(type1 arg1) \ { \ __syscall_nr(1, type, sname, arg1); \ } #define io_syscall2(type,fname,sname,type1,arg1,type2,arg2) \ type fname(type1 arg1, type2 arg2) \ { \ __syscall_nr(2, type, sname, arg1, arg2); \ } #define io_syscall3(type,fname,sname,type1,arg1,type2,arg2,type3,arg3) \ type fname(type1 arg1, type2 arg2, type3 arg3) \ { \ __syscall_nr(3, type, sname, arg1, arg2, arg3); \ } #define io_syscall4(type,fname,sname,type1,arg1,type2,arg2,type3,arg3,type4,arg4) \ type fname(type1 arg1, type2 arg2, type3 arg3, type4 arg4) \ { \ __syscall_nr(4, type, sname, arg1, arg2, arg3, arg4); \ } #define io_syscall5(type,fname,sname,type1,arg1,type2,arg2,type3,arg3,type4,arg4,type5,arg5) \ type fname(type1 arg1, type2 arg2, type3 arg3, type4 arg4, type5 arg5) \ { \ __syscall_nr(5, type, sname, arg1, arg2, arg3, arg4, arg5); \ } libaio-0.3.109/src/syscall-s390.h000066400000000000000000000072071126367703600162520ustar00rootroot00000000000000#define __NR_io_setup 243 #define __NR_io_destroy 244 #define __NR_io_getevents 245 #define __NR_io_submit 246 #define __NR_io_cancel 247 #define io_svc_clobber "1", "cc", "memory" #define io_syscall1(type,fname,sname,type1,arg1) \ type fname(type1 arg1) { \ register type1 __arg1 asm("2") = arg1; \ register long __svcres asm("2"); \ long __res; \ __asm__ __volatile__ ( \ " .if %1 < 256\n" \ " svc %b1\n" \ " .else\n" \ " la %%r1,%1\n" \ " .svc 0\n" \ " .endif" \ : "=d" (__svcres) \ : "i" (__NR_##sname), \ "0" (__arg1) \ : io_svc_clobber ); \ __res = __svcres; \ return (type) __res; \ } #define io_syscall2(type,fname,sname,type1,arg1,type2,arg2) \ type fname(type1 arg1, type2 arg2) { \ register type1 __arg1 asm("2") = arg1; \ register type2 __arg2 asm("3") = arg2; \ register long __svcres asm("2"); \ long __res; \ __asm__ __volatile__ ( \ " .if %1 < 256\n" \ " svc %b1\n" \ " .else\n" \ " la %%r1,%1\n" \ " svc 0\n" \ " .endif" \ : "=d" (__svcres) \ : "i" (__NR_##sname), \ "0" (__arg1), \ "d" (__arg2) \ : io_svc_clobber ); \ __res = __svcres; \ return (type) __res; \ } #define io_syscall3(type,fname,sname,type1,arg1,type2,arg2, \ type3,arg3) \ type fname(type1 arg1, type2 arg2, type3 arg3) { \ register type1 __arg1 asm("2") = arg1; \ register type2 __arg2 asm("3") = arg2; \ register type3 __arg3 asm("4") = arg3; \ register long __svcres asm("2"); \ long __res; \ __asm__ __volatile__ ( \ " .if %1 < 256\n" \ " svc %b1\n" \ " .else\n" \ " la %%r1,%1\n" \ " svc 0\n" \ " .endif" \ : "=d" (__svcres) \ : "i" (__NR_##sname), \ "0" (__arg1), \ "d" (__arg2), \ "d" (__arg3) \ : io_svc_clobber ); \ __res = __svcres; \ return (type) __res; \ } #define io_syscall4(type,fname,sname,type1,arg1,type2,arg2, \ type3,arg3,type4,arg4) \ type fname(type1 arg1, type2 arg2, type3 arg3, type4 arg4) { \ register type1 __arg1 asm("2") = arg1; \ register type2 __arg2 asm("3") = arg2; \ register type3 __arg3 asm("4") = arg3; \ register type4 __arg4 asm("5") = arg4; \ register long __svcres asm("2"); \ long __res; \ __asm__ __volatile__ ( \ " .if %1 < 256\n" \ " svc %b1\n" \ " .else\n" \ " la %%r1,%1\n" \ " svc 0\n" \ " .endif" \ : "=d" (__svcres) \ : "i" (__NR_##sname), \ "0" (__arg1), \ "d" (__arg2), \ "d" (__arg3), \ "d" (__arg4) \ : io_svc_clobber ); \ __res = __svcres; \ return (type) __res; \ } #define io_syscall5(type,fname,sname,type1,arg1,type2,arg2, \ type3,arg3,type4,arg4,type5,arg5) \ type fname(type1 arg1, type2 arg2, type3 arg3, type4 arg4, \ type5 arg5) { \ register type1 __arg1 asm("2") = arg1; \ register type2 __arg2 asm("3") = arg2; \ register type3 __arg3 asm("4") = arg3; \ register type4 __arg4 asm("5") = arg4; \ register type5 __arg5 asm("6") = arg5; \ register long __svcres asm("2"); \ long __res; \ __asm__ __volatile__ ( \ " .if %1 < 256\n" \ " svc %b1\n" \ " .else\n" \ " la %%r1,%1\n" \ " svc 0\n" \ " .endif" \ : "=d" (__svcres) \ : "i" (__NR_##sname), \ "0" (__arg1), \ "d" (__arg2), \ "d" (__arg3), \ "d" (__arg4), \ "d" (__arg5) \ : io_svc_clobber ); \ __res = __svcres; \ return (type) __res; \ } libaio-0.3.109/src/syscall-x86_64.h000066400000000000000000000040051126367703600165030ustar00rootroot00000000000000#define __NR_io_setup 206 #define __NR_io_destroy 207 #define __NR_io_getevents 208 #define __NR_io_submit 209 #define __NR_io_cancel 210 #define __syscall_clobber "r11","rcx","memory" #define __syscall "syscall" #define io_syscall1(type,fname,sname,type1,arg1) \ type fname(type1 arg1) \ { \ long __res; \ __asm__ volatile (__syscall \ : "=a" (__res) \ : "0" (__NR_##sname),"D" ((long)(arg1)) : __syscall_clobber ); \ return __res; \ } #define io_syscall2(type,fname,sname,type1,arg1,type2,arg2) \ type fname(type1 arg1,type2 arg2) \ { \ long __res; \ __asm__ volatile (__syscall \ : "=a" (__res) \ : "0" (__NR_##sname),"D" ((long)(arg1)),"S" ((long)(arg2)) : __syscall_clobber ); \ return __res; \ } #define io_syscall3(type,fname,sname,type1,arg1,type2,arg2,type3,arg3) \ type fname(type1 arg1,type2 arg2,type3 arg3) \ { \ long __res; \ __asm__ volatile (__syscall \ : "=a" (__res) \ : "0" (__NR_##sname),"D" ((long)(arg1)),"S" ((long)(arg2)), \ "d" ((long)(arg3)) : __syscall_clobber); \ return __res; \ } #define io_syscall4(type,fname,sname,type1,arg1,type2,arg2,type3,arg3,type4,arg4) \ type fname (type1 arg1, type2 arg2, type3 arg3, type4 arg4) \ { \ long __res; \ __asm__ volatile ("movq %5,%%r10 ;" __syscall \ : "=a" (__res) \ : "0" (__NR_##sname),"D" ((long)(arg1)),"S" ((long)(arg2)), \ "d" ((long)(arg3)),"g" ((long)(arg4)) : __syscall_clobber,"r10" ); \ return __res; \ } #define io_syscall5(type,fname,sname,type1,arg1,type2,arg2,type3,arg3,type4,arg4, \ type5,arg5) \ type fname (type1 arg1,type2 arg2,type3 arg3,type4 arg4,type5 arg5) \ { \ long __res; \ __asm__ volatile ("movq %5,%%r10 ; movq %6,%%r8 ; " __syscall \ : "=a" (__res) \ : "0" (__NR_##sname),"D" ((long)(arg1)),"S" ((long)(arg2)), \ "d" ((long)(arg3)),"g" ((long)(arg4)),"g" ((long)(arg5)) : \ __syscall_clobber,"r8","r10" ); \ return __res; \ } libaio-0.3.109/src/syscall.h000066400000000000000000000014111126367703600155450ustar00rootroot00000000000000#include #include #define _SYMSTR(str) #str #define SYMSTR(str) _SYMSTR(str) #define SYMVER(compat_sym, orig_sym, ver_sym) \ __asm__(".symver " SYMSTR(compat_sym) "," SYMSTR(orig_sym) "@LIBAIO_" SYMSTR(ver_sym)); #define DEFSYMVER(compat_sym, orig_sym, ver_sym) \ __asm__(".symver " SYMSTR(compat_sym) "," SYMSTR(orig_sym) "@@LIBAIO_" SYMSTR(ver_sym)); #if defined(__i386__) #include "syscall-i386.h" #elif defined(__x86_64__) #include "syscall-x86_64.h" #elif defined(__ia64__) #include "syscall-ia64.h" #elif defined(__PPC__) #include "syscall-ppc.h" #elif defined(__s390__) #include "syscall-s390.h" #elif defined(__alpha__) #include "syscall-alpha.h" #elif defined(__arm__) #include "syscall-arm.h" #else #error "add syscall-arch.h" #endif libaio-0.3.109/src/vsys_def.h000066400000000000000000000023551126367703600157250ustar00rootroot00000000000000/* libaio Linux async I/O interface Copyright 2002 Red Hat, Inc. This library is free software; you can redistribute it and/or modify it under the terms of the GNU Lesser General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version. This library is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public License for more details. You should have received a copy of the GNU Lesser General Public License along with this library; if not, write to the Free Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA */ extern int vsys_io_setup(unsigned nr_reqs, io_context_t *ctxp); extern int vsys_io_destroy(io_context_t ctx); extern int vsys_io_submit(io_context_t ctx, long nr, struct iocb *iocbs[]); extern int vsys_io_cancel(io_context_t ctx, struct iocb *iocb); extern int vsys_io_wait(io_context_t ctx, struct iocb *iocb, const struct timespec *when); extern int vsys_io_getevents(io_context_t ctx_id, long nr, struct io_event *events, const struct timespec *timeout);