src/sys/kern/kern_exec.c, branch release/2.2.5_cvs FreeBSD source tree https://cgit-dev.freebsd.org/src/atom?h=release%2F2.2.5_cvs 1997-10-21T01:56:05Z This commit was manufactured by cvs2svn to create tag 1997-10-21T01:56:05Z cvs2svn cvs2svn@FreeBSD.org 1997-10-21T01:56:05Z urn:sha1:14d01796b1ab61a57dbd60da66d8dac699a0a2e7 'RELENG_2_2_5_RELEASE'. This commit was manufactured to restore the state of the 2.2.5-RELEASE image. Releases prior to 5.3-RELEASE are omitting the secure/ and crypto/ subdirs. Brought in fix from rev 1.64: fix for rfork security hole. 1997-08-04T05:40:37Z David Greenman dg@FreeBSD.org 1997-08-04T05:40:37Z urn:sha1:40ae4344e11f3c0e8a00531ae601131504c1cf50 Fixed two bugs that came into kern_exec just after the 2.2.1 release: 1997-04-18T02:37:08Z David Greenman dg@FreeBSD.org 1997-04-18T02:37:08Z urn:sha1:ccba9c3f37cf55bb7c0740cf9901c17afd277a77 1) I wasn't assuring that the file size was large enough for the data to be contained in a FS block. Calling bread for a block when the data is in a fragment causes BAD things to happen, including file corruption. 2) I was not passing in a residual pointer to vn_rdwr(), and this caused "slow" mode image header reads that were less than the page size to fail with EIO. I now get the residual and properly zero the tail of the image header buffer, too. From -current (1.56): free cn buffer only once. 1997-04-04T07:30:44Z David Greenman dg@FreeBSD.org 1997-04-04T07:30:44Z urn:sha1:d86aa8b7c9e3b326d8019b6195636b8e8ebfd6cd Merged from -current: Bruce's fixes. 1997-04-04T04:18:20Z David Greenman dg@FreeBSD.org 1997-04-04T04:18:20Z urn:sha1:363b5060430905fc706d12adeb6ddb610e11291b Sort of from -current: lose the vnode lock on a permissions failure. 1997-04-04T01:32:09Z David Greenman dg@FreeBSD.org 1997-04-04T01:32:09Z urn:sha1:d01342629b6ab37e18810315f5c0ef1cc19f2ca5 Merged from -current: Killed exech_map and do image header reads in a FS- 1997-04-03T06:37:49Z David Greenman dg@FreeBSD.org 1997-04-03T06:37:49Z urn:sha1:5397377bf130df03558bc83d80ffb75ba425e31d centric way instead of VM-centric. This appears to fix a hang that Peter was seeing as well as allowing detection of errors during the header read (important for NFS). Brought in change from rev 1.51: preserve P_SUGID if {u,g}gid != r{u,g}id 1997-02-19T03:53:35Z David Greenman dg@FreeBSD.org 1997-02-19T03:53:35Z urn:sha1:a527226685e90caf217b180a052035b99769ead2 Note that this doesn't actually fix any known security holes, it just protects against sloppy programming. Merge the fix from rev 1.48: fix an ordering bug. 1996-11-09T10:42:28Z Joerg Wunsch joerg@FreeBSD.org 1996-11-09T10:42:28Z urn:sha1:6acb8030bb257c46f62ae44cdb87c1c00d87648e Requested by: dyson Performance optimizations. One of which was meant to go in before the 1996-10-12T21:35:25Z John Dyson dyson@FreeBSD.org 1996-10-12T21:35:25Z urn:sha1:9d3fbbb5f4bb727f86ea497435cbc6d70539c33f previous snap. Specifically, kern_exit and kern_exec now makes a call into the pmap module to do a very fast removal of pages from the address space. Additionally, the pmap module now updates the PG_MAPPED and PG_WRITABLE flags. This is an optional optimization, but helpful on the X86.