arnold@mathcs.emory.edu (Arnold D. Robbins {EUCC}) (07/11/89)
Has anyone given any thought to what would be involved in making GCC (and GAS?) produce position-independant code? That would be real useful for producing Sun shared libraries with GCC, and also necessary if the GNU kernel will support shared libraries. Thanks, -- Arnold Robbins -- Emory University Computing Center | Unix is a Registered DOMAIN: arnold@unix.cc.emory.edu | Bell of AT&T Trademark UUCP: gatech!emoryu1!arnold PHONE: +1 404 727-7636 | Laboratories. BITNET: arnold@emoryu1 FAX: +1 404 727-2599 | -- Donn Seeley
donn@wasatch.utah.edu (Donn Seeley) (07/12/89)
Arnold Robbins (arnold@unix.cc.emory.edu) writes: Has anyone given any thought to what would be involved in making GCC (and GAS?) produce position-independant code? That would be real useful for producing Sun shared libraries with GCC, and also necessary if the GNU kernel will support shared libraries. As our local Kool Unix Dude, Mike Hibler, has pointed out, you need look no further than your own .signature :-): | Unix is a Registered | Bell of AT&T Trademark | Laboratories. | -- Donn Seeley For several months a project has been in the works here at Utah to produce an implementation of shared libraries for Berkeley Unix and (potentially) Mach. We're aiming at having something useful in time for 4.4 BSD, although whether the implementation will be included in 4.4 is up to Berkeley CSRG, of course. Our objective is to steal as many ideas as possible from existing shared library implementations while avoiding well-known lossages. The basis for the implementation will be position-independent code produced by a modified GCC. We've already done some successful prototyping with PC-relative code on the 68020, and we expect to test some other strategies as well before we settle on one or more specific techniques. The implementation will be 'free' under the usual GNU and Berkeley licenses. Note that we are not specifically striving to duplicate Sun shared libraries, although we aren't afraid to use published information about their implementation when it seems profitable to do so, and it probably wouldn't be difficult to retarget our GCC changes toward the Sun environment. Currently the project is in low gear because we haven't gotten around to applying for funding; I hope this will be remedied soon :-). If you have useful suggestions about shared libraries, position- independent code, object formats, address space configuration or other topics, let me know... Donn Seeley University of Utah CS Dept donn@cs.utah.edu 40 46' 6"N 111 50' 34"W (801) 581-5668 utah-cs!donn