home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.sys.hp
- Path: sparky!uunet!wupost!uwm.edu!spool.mu.edu!nigel.msen.com!sdd.hp.com!hp-col!fc.hp.com!jsm
- From: jsm@fc.hp.com
- (John Marvin)
- Subject: Re: mmap on HP-UX 9.0 on s700
- Sender: news@fc.hp.com (news daemon)
- Message-ID: <BxDsMo.GMF@fc.hp.com>
- Date: Sun, 8 Nov 1992 05:01:36 GMT
- References: <FRANL.92Nov7210901@draco.centerline.com>
- Organization: Hewlett-Packard Workstation Kernel, Ft. Collins, CO
- X-Newsreader: TIN [version 1.1.1 PL6]
- Lines: 30
-
- Fran Litterio (franl@centerline.com) wrote:
- : jsm@fc.hp.com (John Marvin) writes:
- :
- : > Fran Litterio (franl@centerline.com) wrote:
- : > : If this is so, then how is dld.sl guaranteed to be able to map the
- : > : text/data of mylib.sl at the same VM address for all programs which
- : > : were linked against it?
- : >
- : > It doesn't. The data is mapped into the second quadrant, which is private
- : > (not shared with other processes). The text is mapped into the 3rd or 4th
- : > quadrant, which is shared with all processes. But when it is mapped by
- : > the first process that uses it, it can be placed anywhere within the 3rd
- : > or 4th quadrant.
- :
- : Don't you mean "anywhere within the 3rd or 4th quadrant that isn't
- : already allocated to a shared library that another process has
- : mapped"?
- :
- : So if libfoo.sl's text pages are mapped for the very first time by any
- : process, they will not be assigned virtual addresses that are used by
- : libbar.sl's text pages in another process that was started a minute
- : earlier? Even if the program that maps libfoo.sl does not map
- : libbar.sl?
-
- Yes, exactly. But the point I was trying to make was that depending on the
- order of invocation of various shared libraries, a shared library can be
- mapped anywhere within the 3rd or 4th quadrant.
-
- John Marvin
- jsm@fc
-