Date: Sun, 13 Apr 1997 09:06:51 -0400 (EDT) From: Peter Dufault <dufault@hda.com> To: current@freebsd.org, hackers@freebsd.org Subject: POSIX4 patches on freebsd.org Message-ID: <199704131306.JAA07596@hda.hda.com>
next in thread | raw e-mail | index | archive | help
I put POSIX4 patches on freefall in ~ftp/pub/dufault/posix4.tgz for feedback. These are against current as of this morning. I have verified that they apply against current, and I've built world with the last iteration, but I haven't actually verified that you can build world with these patches applied against today's current. I'm interested in feedback at the moment. These add: 1. An infrastructure for plugging in POSIX4 optional pieces at the option level; 2. sysctl and sysconf support for POSIX4; 3. Access control to posix 4 option level functionality through a /dev/posix4, permitting scheduler or memory locking to be restricted and not only at the root level; 4. An implementation of _POSIX_PRIORITY_SCHEDULING; 5. Some regression tests. If you do try to build this you'll need to manually create the new directory /usr/src/sys/posix4, add options "POSIX4" to your config and add -DPOSIX4 to your /etc/make.conf. After that you should be able to build and load the LKM in the "posix4" subdirectory of the distribution, build the library in "libposix4" and build the tests in "regress". As long as you don't include required but as yet non-existent headers unrelated to _POSIX_PRIORITY_SCHEDULING and use only _POSIX_PRIORITY_SCHEDULING functions this should work properly. What I'd like now is: 1. Some discussion of how this should be interfaced to the kernel. At the moment it is through an ioctl against a pseudo device. This is non-traditional and doesn't play well with ktrace. However, if every POSIX4 function gets a system call I need many new system calls. If every POSIX4 option gets a system call I need about 16 new system calls. Is there a limit of 255 in the number of system calls? 2. Review of the scheduler changes. 3. Review of whether it seems POSIX compliant. Where I go next is: 1. Modify to whatever we decide is the appropriate kernel interface. 2. Add all required POSIX4 headers. 3. Add man pages. Peter NB: DO NOT GROUP REPLY to this message as it is on two lists for announcement purposes. Follow up privately or to -hackers only. -- Peter Dufault (dufault@hda.com) Realtime Machine Control and Simulation HD Associates, Inc. Voice: 508 433 6936
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199704131306.JAA07596>